Research Article

Review
Received: 26 October 2009 Revised: 12 May 2010 Accepted: 17 May 2010 Published online in Wiley Interscience: 16 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4041
Genetic evaluation of dairy cattle using
a simple heritable genetic ground
Josef Pribyl, a Vaclav Rehout, b Jindrich Citek b∗ and Jana Pribylova a
Abstract
The evaluation of an animal is based on production records, adjusted for environmental effects, which gives a reliable
estimation of its breeding value. Highly reliable daughter yield deviations are used as inputs for genetic marker evaluation.
Genetic variability is explained by particular loci and background polygenes, both of which are described by the genomic
breeding value selection index. Automated genotyping enables the determination of many single-nucleotide polymorphisms
(SNPs) and can increase the reliability of evaluation of young animals (from 0.30 if only the pedigree value is used to 0.60 when
the genomic breeding value is applied). However, the introduction of SNPs requires a mixed model with a large number of
regressors, in turn requiring new algorithms for the best linear unbiased prediction and BayesB. Here, we discuss a method that
uses a genomic relationship matrix to estimate the genomic breeding value of animals directly, without regressors. A one-step
procedure evaluates both genotyped and ungenotyped animals at the same time, and produces one common ranking of all
animals in a whole population. An augmented pedigree–genomic relationship matrix and the removal of prerequisites produce
more accurate evaluations of all connected animals.
c○ 2010 Society of Chemical Industry
Keywords: genomic breeding value; methods; QTL; SNP; linear model; genomic relationship
INTRODUCTION
Laboratory techniques and mathematical and statistical methods
for the evaluation of animal breeding values (BV) are undergoing
continuous improvement. Molecular genetic data can be analysed
for associations with production traits. 1 However, the relationships
between farm animal production traits and molecular-genetic
information are often measured imprecisely. Many studies of the
relationships between genetic markers and quantitative traits
are methodologically flawed and do not reflect contemporary
breeding practices; sometimes even the basic context of breeding
and farming conditions are not taken into consideration. This
type of research requires very careful experimental design that
considers pedigree structure and generates an adequate quantity
of data using sophisticated mathematical and statistical methods.
The general objective of each evaluation is to explain the
variability of the characteristics studied and determine why
animals or groups of animals differ from one another. Farm
animal productivity is simultaneously influenced by many genetic
and non-genetic factors, and it is practically impossible to
plan a completely balanced experiment. Therefore, sophisticated
statistical procedures must be used.
Recently, the evaluation of animal performance based on
molecular-genetic information has become more widespread.
Dairy cattle populations evaluated for several groups of traits
of moderate and low heritability (production, conformation,
reproduction) using genetic markers have been presented by
several authors 2–8 as well as in Interbull Bulletin No. 39. 9 In pig
and poultry populations, whole-genome scanning and genetic
diversity analysis are quite extensive. 10,11 The methodologies used
may be generalised across species, but several facts influence
methodological advancements in relation to dairy cattle: the cost
of genotyping is favourable relative to the price of each animal;
advanced reproductive methods are routinely applied and the BV
of sires is therefore highly reliable; there is a huge global market
for sperm and breeding animals encompassing many companies
and breeder associations; and worldwide workshops on animal
evaluation are frequently organised through publications such as
the Interbull Bulletin. 9
The aim of this review is to provide a survey of the procedures
used to evaluate animal production traits using simple heritable
genetic markers. Some basic methodological approaches will be
emphasised, particularly those that connect genomic breeding
value (GEBV) to traditional methodologies.
ANIMAL EVALUATION
An overview of the standard procedures used worldwide in
the genetic evaluation (BV prediction) of production traits in
farm animals, as well as new developments, are continuously
published by ICAR, Interbeef, Interbull, Interstallion and other
international organisations. A number of countries cooperate
in the international evaluation of dairy cattle, which invokes
international inspection of the methods used to estimate BV in
domesticcountry. 9 Here,specialattentionispaidtothecontinuous
updating of national and international evaluation procedures.
∗ Correspondence to: Jindrich Citek, Department of Genetics, Faculty of
Agriculture, South Bohemia University, Studentska 13, CZ 370 05 České
Budějovice, Czech Republic. E-mail: citek@zf.jcu.cz
a Institute of Animal Science, CZ 104 01 Praha 10-Uhrineves, Czech Republic
b Department of Genetics, Faculty of Agriculture, South Bohemia University,
Studentska 13, CZ 370 05 Ceske Budejovice, Czech Republic
J Sci Food Agric 2010; 90: 1765–1773 www.soci.org c○ 2010 Society of Chemical Industry
1765

1766
Generally, mixed linear models of best linear unbiased prediction
(BLUP) in an animal model (AM) are used, and a pedigree of three
or more generations of ancestors is taken into account according
to the model equation:
Y = Xb + Zu + e (1)
where Y is the vector of measured performances, X and Z are
known matrices that relate performance to the systematic effects
of the breeding environment and the animals, b and u are the
estimated vectors of fixed systematic environmental effects and
the random effects of an animal (BV) with the additive numerator
relationship matrix (A), and e is the vector of random error.
Using this model equation, a system of normal equations is
constructed in which the unknown constants (b) and(u) are
estimated. These systems of equations are vast, and special
algorithms are required for their solution. 12
Most of the variability in any measured production trait is
caused by systematic environmental effects. The influence of
the herd–year–season, or herd–test-day, which identifies a
contemporary group of animals kept under the same conditions,
is usually the most important factor.
Evaluations are generally oriented to the MT-AM (multi-trait
animal model), RR-TDAM (random regression test-day animal
model), AM-maternal, and nonlinear methodologies for survival
(kit) analysis. 13–23
It is important to find a method of evaluation that minimises
residual error and simultaneously considers all of the effects that
may influence the performance variable being measured. From
a genetic perspective, it is important to ask: What proportion
of variability is explained by the statistical model used? Is this
proportion different in a model that does not account for genetic
effects? Is this model the best (optimal) of all the possibilities
tested? The proportion of variability explained by the statistical
model used (R 2 ) and other information criteria for testing the
suitability of the model, such as Akaike’s information criterion
(AIC), Bayes information criterion (BIC), Bayes factor (BF) and the
likelihood ratio test (LRT), are very important in answering these
questions. 24–28
Molecular-genetic information can be used to improve selection
programmes. 29 Animals are evaluated more accurately when their
entire genetic value is partitioned into causal factors and withinfamily
genetic components are exploited. The use of moleculargenetic
markers in breeding is the inclusion of additional criteria in
the selection indices. These markers increase selection differences
relativetoexistingtraditionalbreedingprogrammesbydecreasing
the correlation among sib individuals, increasing the accuracy of
animal selection, allowing the utilisation of genetic variability
that is usually included in non-utilisable residuum, and allowing
a shortening of the generation interval (because they may be
analysed in young animals). The use of selection markers is
conditional upon the timely laboratory analysis of the entire
subpopulation subjected to pre-selection (e.g., young bulls) and
rapid application before the determined gene linkages change.
This requires frequent updates of selection indices, as shown
below (Eqn (2)). The consistent application of genomic selection
markedly reduces the cost of a selection programme. 3,30
However, data analysis becomes more complicated, the number
of estimated parameters becomes higher, and a modified
information criterion (mBIC) is necessary for the selection of a
suitable model of evaluation. 31,32
In order to select individuals for breeding, marker-assisted
selection (MAS) may be applied if several genetic markers are to
www.soci.org J Pribyl et al.
be used. Alternatively, genomic selection utilises high numbers of
markers that densely cover the whole genome. 3,33 BV is usually
calculated in two steps. In the first step, the regression coefficients
(v) (substitution effects of the alleles of a considered locus) are
determined in a reference population with known performance
and highly reliable BVs. This reference population usually includes
only a part of the population under selection. From the first step,
quantitative trait loci (QTL) effects are estimated. Subsequently,
BV is determined for all of the young animals in the evaluated
(sub)population by means of a selection index, as described in
Eqn (2). 30
The reference population and the evaluated population are
separated by at least one generation. Therefore, the relationships
between markers and QTLs determined in the older generation
may not be fully applicable to the younger evaluated population,
as the QTLs are not fully covered by study markers. Furthermore,
the influences of selection, mutation, immigration of sires used
intensively in artificial insemination, changes in environment, and
the development of the commercial population under selection
can also affect the applicability of QTL data across generations.
Therefore, it is necessary to periodically redetermine u in Eqn (1),
allele frequencies (q) in Eqn (5), their inherence in the genotypes
of individual animals (T), and regression coefficients (v) in(3)so
that the gap between the reference and the evaluated population
will be as small as possible. 3,30,34
The GEBV of a given trait is calculated based on known loci and
remaining polygenes according to the selection index:
GEBVj = k1 DGVj + k2u ∗ j
where GEBVj is the genomic (total) BV for an individual (j)
determined based on the genomic information at the locus (i)
and remaining polygenic effect. DGVj is the direct genetic value,
calculated as the sum of BVs for a particular loci:
DGVj = �jTijvij
where Tij (with regard to Eqn (9)) is the ith element in the jth row
of the known incidence matrix correlating the genetic effects of
particular alleles to the observed individual, vij is the vector of
genetic marker effects, u ∗ j represents the BV calculated based on
the remaining polygenes, and k1 and k2 are the weights of the
information sources in the index. 35
If the GEBV is calculated for young animals without their own
production records, u ∗ j represents only information about their
parents. In cases where a high density of genetic markers is
available, the u ∗ j in Eqn (2) is frequently omitted.
GENETICALLY CONDITIONED VARIABILITY
OF PERFORMANCE
Reliably determined population-genetic parameters are a precondition
for genetic evaluation. We are usually interested in
phenotype variability (σ 2 P), which can be separated into genetic
additive (σ 2 A), genetic dominant (σ 2 D), genetic epistatic (σ 2 I), and
unpredictable residual (σ 2 E) components, plus covariance caused
by genotype/environment interaction (2σGE). 36 It is generally assumed
that the genotype/environment interaction is negligible.
Therefore:
σ 2 P = σ 2 G + σ 2 E = σ 2 A + σ 2 D + σ 2 I + σ 2 E
The additive effects that are accumulated over successive
generations of selection are used in breeding. Nevertheless, other
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1765–1773
(2)
(3)
(4)

Evaluation of animals by simple heritable markers www.soci.org
genetic effects are also reflected in performance, and if they are
omitted the results of the evaluation may be distorted.
One locus
Some genes may have a direct impact on quantitative production
traits, and therefore efforts are made to utilise them directly in
breeding. These candidate genes (or quantitative trait loci (QTL), if
describing markers) explain a portion of the genetic variability in
the trait being considered.
At two alleles in a locus, the portion of the additive genetic
variance conditioned by one gene (i) can be approximated by a
binomial distribution: 36
σ 2 Ai = 2qi(1 − qi)vi 2
where qi is the frequency of the studied allele at locus i and vi is
the additive substitution effect of alleles at locus i.
The portion of the variance caused by a dominant allele at a
given locus is
σ 2 Di = (2qi(1 − qi)di) 2
(6)
where di is the dominance effect in locus i. Often, it is assumed
that di = 0.
In the case of genes with major effects on the trait being
studied, the analysis is slightly easier because animals carrying a
desirable allele frequently exceed the normal variable range for the
measured production trait. This is obvious from the distribution
function of the estimated BV of the evaluated trait (additional
peaks, outliers), which indicates that a special genetic effect is
occurring and should be included as a separate factor in the
model. 37
Several loci
Correlations may exist between the genes in question. Therefore,
thevariabilityofanobservedtraitthatisexplainedbyseveralgenes
depends on the variability caused by each gene and combinations
thereof (λ). 38 The additive covariance between two loci can be
expressed as
(5)
cov(Ai,A i ′) = (1 − 2λ) 2 (σAiσ Ai ′) (7)
The theory of selection indices is used to determine the shares of
several loci in the total genotype. 39,40
Genes interact with one another, and any gene may have
pleiotropic effects. These interactions are mostly unknown and
may be quite extensive. This implies genetic epistatic variability
(σ 2 I) based on two or more interactions among all loci studied.
Itisexpectedthatmulti-generational,similarlyorientedongoing
selection in commercial breeds will lead to the stabilisation of
favourable genetic combinations. The fixation of desirable alleles
couldalsooccuratanumberoflociinanimprovedbreed.However,
breeding conditions change constantly, and combinations of
genes are disturbed by selection, mutation and by the immigration
of sires from other populations. Therefore, inter-gene interactions
within some families may be expressed differently for a certain
period before the gene linkages are again stabilised. This can be
exploited in selection.
When studying the influence of a selected gene on performance,
the effects of nearby (linked) genes will also be included; thus the
result does not correspond only to the studied gene or to the
studied marker–QTL relationship. Therefore, when a low number
of sparsely located markers is analysed, the effects of any given
marker are frequently overestimated. 33 The effects calculated for
each genetic parameter strongly depend on the number and
density of genetic parameters included in a simultaneous analysis.
One locus can also have an epistatic effect on several traits, which
may be either positively or negatively correlated. It is therefore
necessary to distinguish between pleiotropic and closely linked
QTL effects. 41
Polygenic effects: the ‘infinitesimal model’ (pol)
A large number of unknown genes are assumed to affect the
majority of production traits, and their overall influence on
performance and its variability is the object of interest. In general,
the components of variance are currently determined as in Eqn (1),
by REML methods or by applying the Bayesian approach using
the Gibbs sampling method. 12 These methods require the analysis
and adjustment of input datasets so that specific components of
variance (for example, within families, between families, caused
by different effects of genes) can be estimated. 19,21
Joint effects of particular loci with remaining polygenes
The overall influence of genetic effects on the observed production
trait is expressed by the coefficient of heritability (h 2 = σ 2 A/σ 2 P).
The specific roles of the genes which exert these effects generally
remain unknown.
The total additive genetic variability is the sum of the known
loci according to Eqn (5), adjusted for mutual linkages (7) and
‘residual’ additive genetic variability caused by the remaining
polygene σ 2 Apol:
σ 2 A = �jσ 2 Ai + �j�j; cov(Ai, A i ′) + σ 2 Apol
Hence both single-locus effects and the remaining polygenic
effects of the ‘genetic background’ should be considered
simultaneously. 7,42–45
EXPERIMENTAL DESIGN FOR THE EVALUATION
OF GENETIC MARKERS
The objective is to estimate the genetic contribution to specific
productiontraits.However,theexperimentaldesignshouldensure
the reliable estimation of all factors that influence performance.
The power of the evaluation of data depends on the structure
and the size of the experiment, and the minimum number of
observations required to achieve adequate predictive power can
be calculated. 46 Large datasets spanning progeny from many sires
are usually necessary. 2,3,30,43 Generally, thousands of animals are
included in any one experiment.
Laboratory analyses are expensive, and therefore the decision
of which animals from which generation should be genotyped
should be made carefully, to achieve the highest possible reliability
with the lowest possible cost. Several methods based on the
relationship matrices between animals have been developed for
this purpose. 49
Both the screening of allele frequencies and the evaluation of
their relationship to production traits require a pedigree analysis.
Sires,especiallythoseimportedfromotherpopulationsforartificial
insemination, can dramatically change the frequencies of alleles
in a herd or an entire population in a short period of time.
There are differences in the methodologies used to evaluate
F1/F2 generation-designed experiments in which extremely
J Sci Food Agric 2010; 90: 1765–1773 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
(8)
1767

1768
different breeds are crossed 38,50 and studies involving stably
selected commercial populations, where alleles are expected to
be in favourable interactions. The second case, which is connected
with the continuous improvement of already productive breeds,
is generally of greater interest to breeders.
DD + GDD
Sib animals, belonging to the same families, generally have similar
performance capabilities. They share both the observed genes
and background polygenes. It is crucial to determine whether
performance is influenced by the studied locus or by other
polygenes.
Study designs that incorporate data from multiple generations
have been developed for the analysis of small numbers of markers.
These daughter design (DD) and granddaughter design (GDD)
analyses allow estimation of the effects of the studied loci within
families, i.e., within groups of sib animals with a similar genetic
background. 38,51 In this type of analysis, the initial generations of
sires (i.e., parents or grandparents) must be heterozygous at the
studied locus. In this way, each initial animal gives rise to two
genetically different groups of progeny with respect to the alleles
studied.
In the proposed GDD experiment, only generations of ancestors
without their own performance measurements can be genotyped.
Their performance scores are assigned by means of progeny
testing from a large set of non-genotyped progeny. This
considerably decreases the number of individuals that must be
genotyped despite achieving a high reliability of evaluation. The
total number of animals required for the experiment is relative
to the proportion of genetic variability influenced by the locus,
allele frequencies, and the level of recombination between QTL
and the marker. However, only the additive effects of genes can
be estimated in this design. The GDD and general pedigree design
analysis of QTL in dairy cattle have been compared in simulation
studies. 52
Design with a large number of markers (SNP)
An increased number of markers introduces more complexity. The
size of the reference population of sires with highly reliable BV
estimates is particularly important. 2,3,43 Larger numbers are better,
and several thousands of sires are desirable.
A MODEL FOR ESTIMATING GENETIC EFFECTS
The principle of evaluation consists in the separation of the
effects of purely heritable loci from the effects of other genetic
background. 43,47,48,53 The BLUP method and similar approaches
are the best procedures that can be used to adjust measured
BVvalues.ConsistentwithEqns(1)and(2),theevaluationcanbe
formally expressed by a modified mixed linear model:
Y = Xb + Z[u ∗ + Tv] + e (9)
where T is the known matrix of the experiment design that links
an animal to the genetic effects of particular alleles. Each row
may include columns according to particular loci and several
genetic effects of each locus with values for additive effects
(tAi¤ ), dominance effects (tDi¤ ), twolocus
epistatic interactions between loci (t Iii ′ = tAit Ai ′); 32 u ∗ is the
estimated vector of the random ‘residual’ polygenic effects for
each animal (i.e., the partial BV after the effects of the studied loci
www.soci.org J Pribyl et al.
are excluded) with the additive relationship matrix; and v is the
estimated vector of the effects of genetic markers. This may also
comprise several loci and encompass additive, dominance and
epistatic effects.
Genetic markers (v) can be considered to have fixed 32,53 or
random effects. In the latter case, either the diagonal genetic
matrix alone, Iσ 2 Ai, is considered 54 for each random effect (i), or
the complete covariance structure and its relationship with the
identity by descent matrix (IBD), IBDσ 2 Ai, is taken into account.
IBD describes the probable positional relationships between each
marker/QTL pair and the probability of inheriting the paternal
or maternal QTL allele. The construction of IBD depends on
whether linkage analysis (LA), linkage disequilibrium (LD) or a
combination of both methods (LDLA) was used to determine
linkage status. 8 In this context, several teams have developed
algorithms for the construction of an IBD matrix. 41,44,55 They have
also derived genotype values for non-genotyped sib animals
whose performance data may be then used to identify candidate
genes.
DATA FOR EVALUATION
Several types of data describing performance can be used for
these evaluations. Either direct performance records or adjusted
values may be used. For the second approach, data are adjusted
for non-genetic noise as precisely as possible, when BV with high
reliability is estimated in large populations. This yields adjusted
(pseudo) values for BV, yield deviation (YD) or daughter yield
deviation (DYD) that may be used in further analyses.
Direct individual performance
If genetic parameters (markers) are determined directly in animals
from their performance records, the evaluated trait (Y) according
to Eqn (9) is their recorded performance.
Given the pedigree structure and design of the experiment, it
is possible to estimate additive, dominance and epistatic genetic
effects, all of which could be included in v. However, in practice
relatively few performance values are known for each animal.
Therefore the values of vectors u ∗ , v and other effects b in
Eqn (9) can be estimated only with considerable error. 56,57
Breeding value
Animals with highly reliable BV are used for evaluation (usually
sires whose value has been proven by progeny testing). In BV
analysis, the effects of selected loci on major traits associated with
milk performance are determined58–60 according to the following
model:
û = Tv + u ∗
(10)
where û is the vector of BV determined by a routine method
BLUP-AM according to Eqn (1) based on all polygenes.
The BV of an animal summarises the data on performance
deviations of the contemporaries of all sib animals. The expected
BV of the progeny (uO) is related to the BV of sires (uS) and mothers
(uM) and to the random Mendelian sampling of parental gametes
(MS).
uO = 0.5[uS + uM] + MS (11)
One half of the additive genetic variability of (uO) iscausedby
MS. Therefore the result of Eqn (10) significantly depends on the
volume and sources of information that contributed to the BV. A
reliable input BV, which can be achieved only for animals with a
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1765–1773

Evaluation of animals by simple heritable markers www.soci.org
large set of progeny, is a condition for a correct evaluation. This,
however, implies that it is possible to evaluate only the additive
genetic component.
We must take into account the fact that BV represents a
random effect (regressed value) and its value directly depends
on the reliability of estimation (r2 ). The variability of BV (σ 2 u)is
therefore higher at higher reliabilities, as shown by the following
relationship:
r 2 = σ 2 u/σ 2 A
(12)
This demonstrates that for BV estimates with low, unbalanced
reliabilities the animal rank may change and the results of markers
analysis are not very reliable.
Daughter yield deviations
DYD computed from Eqn (1) are used in most routine evaluations
of markers. 45 Initially, yield deviations (YD) adjusted for all
non-genetic effects are determined according to the following
equation: 61
YD = Y − Xb (13)
where YD is the vector of yield deviations. The average deviations
of sires’ daughters (DYD) is then determined and adjusted for 0.5
BV of their mothers:
DYD = Z ′ S [YD − 0.5Z ′ MuM]N −1
(14)
where ZS is the known matrix that relates daughter performance to
the sire; ZM is the known matrix that relates daughter performance
to the mother; uM is the vector of mothers’ BV; and N is the diagonal
matrix that describes the number of daughters per sire.
The values of DYD are independent of the reliability of sires BV
estimates, and therefore are more comparable between sires with
different reliabilities of BV estimation. In agreement with Eqn (11),
DYD comprise 0.5 BV of a sire, including MS and random error. The
alternative of DYD is de-regressed BV. 3
Performances adjusted in this way are evaluated by weighted
analysis according to (9), where the vector Y is substituted by the
vector DYD, and vector b may encompass additional fixed effects.
DYD values are the means for n daughters of sires. Taking
into account the number of contemporaries connected to each
daughter in DYD, and the structure of the entire dataset, we
can generate the effective daughter contribution (EDC), which is
determined based on the reliabilities of the estimation of sires’ BVs
(r 2 ):
EDC = (r 2 /(1 − r 2 ))((4 − h 2 )/h 2 ) (15)
The weight (w) for weighted analysis, which is the inverse of DYD
variance, corresponds to the value of EDC. The exact derivation
of the weight factor for special situations has been described
previously. 61,62
DYD has been used in several GDD studies; one evaluated 39
markers in a set of 4993 sires and another evaluated 263 markers
in a set of 872 sires. 7,63
As in the evaluation of BV, only the additive genetic component
can be determined by DYD. If the number of progeny per sire is
large then they prevail in his BV, r 2 is high and balanced, and the
sire’s MS is almost completely contained both in BV and in DYD.
The correlation between BV and DYD is in this case high, and the
results of evaluation for genetic markers on the basis of BV and
DYD are similar. 32
METHODS FOR EVALUATING GENETIC
MARKERS
When only a small number of genes is studied, it is not
possible to evaluate the experiment correctly without splitting
the genotype influence into the part played by singular observed
genes and the part played by the other (residual) polygenic
‘genetic background’. 7,33,64,65 On the other hand, single observed
genes also contribute to the additive effects of all genes, and
the polygenic effect (u) is the sum of these additive effects.
Therefore, it is not easy to distinguish between the influence of
the polygenic ‘genetic background’ and the effects of individual
genes; in these cases the effect of the individual observed genes
is frequently reduced. 66 Therefore careful experimental design,
particularly with respect to the size of the experiment, is necessary
to estimate the effects of genetic markers.
Several connected questions must be asked in any evaluation of
genetic markers: (A) What proportion of the genetic variability of
the evaluated trait is explained by the studied genetic factors?
(B) What is the genetic correlation between the influence of
the factors studied and the influence of the ‘remaining’ genetic
background on the evaluated trait? (C) Do the results from a model
that considers only polygenic effects and a model that includes
both QTL and remaining polygenic effects differ from each other?
(D) What is the influence of each allele? (Note that it does not
make sense to answer (D) without first answering (A).) (E) Do the
studied genetic factors have similar effects in all groups of related
animals?
AsmallnumberofQTLs
When a small number of loci are evaluated, QTLs are often used
to represent fixed effects of the genotype in a linear model,
for example in GLM/SAS. 67–69 The evaluation model also reflects
systematic effects of the breeding environment or of groups of
animals according to their relationships. With this method, it is
not possible to wholly avoid the influence of correlated loci, and
the effects of individual loci are therefore usually significantly
overestimated. 33 The model can be improved by including a
parameter for the random effects of the parents of genotyped
animals. 56,57
The effect of the studied locus depends on the genetic background
of the animal and could differ between populations. 43,65
The BLUP, REML and Bayesian analysis methods incorporate common
fixed effects for particular loci and ‘residual’ random effects of
remaining polygenes to provide more exact results. 7,43,45 Another
approach for obtaining more exact results is also to use particular
loci as random effects with IBD to account for their variability. 8,55
A large number of SNPs
Production traits depend on a large number of mutually linked,
interacting genes that may be distributed across the entire
genome. Currently, it is possible to sequence tens to hundreds of
thousands of single-nucleotide polymorphisms (SNPs) for many
individual animals, densely covering the entire genome. A multiple
regression analysis of all SNP markers describes their relationships
to the production trait in question. Thus this analysis can be used
to find the DGV and GEBV according to Eqns (3) and (2). Because
a large number of SNPs is considered, there is less emphasis on
the quantitative relationships between individual markers and the
relevant QTL; instead, the overall relationship to the production
trait in question is important.
J Sci Food Agric 2010; 90: 1765–1773 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1769

1770
The high density of markers also allows the generalisation of
effects. Relationships no longer need to be calculated individually
within particular families and the effects of alleles are assumed to
be consistent across all families for simplification. 33
While the breeding values of young dairy cattle can be predicted
with a reliability of about 30% by pedigree value on the basis of
polygenes, an increase in reliability (to 50–70%) can be expected
when large number of SNPs are evaluated. 2,3,33
Computational strategies used to evaluate SNP data
Generally, techniques based on the BLUP and BayesB methods
are used to evaluate large numbers of SNPs. 33 Depending on
the total number of SNPs sequenced, it is usually necessary to
calculate many genetic regression relationships between a given
production trait and the studied alleles. 54,70 These relationships
may be formally solved according to Eqn (9). Compared to a
general AM calculated on the basis of polygenes only, the size
of the vector DYD is relatively smaller (thousands of sires) but
thesizeofthevectorv is large (tens of thousands of regression
coefficients). When there is a high density of SNPs across the
entire genome, the term u ∗ is often omitted from the solution
and only SNPs are used (vector v). 43 In practice, however, it is
expected that even when a high density of markers is obtained
some QTLs will not be covered and the polygene effect is therefore
still considered. 3 Only additive effects are evaluated due to the
large number of SNPs; the inclusion of non-additive effects would
increase the number of effects in the model enormously. After
simplification, the computation model can be expressed as
DYD = Xb + Tv + e (16)
where Xb describes the total mean and fixed effects included in
this step.
Often, the majority of SNPs do not have any information content.
If the relationships between the markers and QTLs are already
known, it is possible to reduce the number of regressors in the
model, which will simplify the solution and also reduce the cost of
laboratory analyses. 71
Because of the large number of independent variables, these
systems of equations (16) are poorly conditioned and cannot
always be solved. Therefore, the systems of equations and
algorithms of solutions must be rearranged. For example, ridge
regression may be applied, which means that SNPs are treated as
random effects. At the same time, numerical values are added to
the diagonal of the matrix of the system to ensure the solubility of
the equations. 34 The added values are the inverse of the genetic
variabilities of each SNP. These values are not usually known for
many genetic parameters, so other simplifications must be used
when constant components of variance are required for all SNPs. 33
The sum of components across all loci yields the total additivegenetic
variability of the studied trait σ 2 A.MatrixT in Eqn (16)
has f rows corresponding to the number of evaluated sires with
known daughter performance. If only additive gene effects are
considered, matrix T has m columns corresponding to the number
of SNP markers considered (m > f). Therefore, a system of the
matrix size m × m at least is solved, and tens of thousands of
regression coefficients are estimated. 54
DIRECT ESTIMATION OF GEBV
Based on T, a genomic relationships matrix (G) can be
determined. 72,73 This requires the introduction of the matrix Q,
www.soci.org J Pribyl et al.
which describes deviations of allelic frequencies in the basic ‘nonselected’
population. The ith column of Q contains the deviation
of the frequency of the second allele in locus i from the expected
value (0.5) multiplied by two Qi = 2(qi − 0.5). 72 The dimensions
of matrix Q correspond to those of matrix T.ThematrixGhas the
form
G = ([T − Q][T − Q] ′ )/(2�qi(1 − qi)) (17)
which is analogous to the generally used numerator relationship
matrix (A) in Eqns (1) and (9). Its dimensions are f × f, where
the diagonal indicates the number of homozygous loci in the
evaluated animal and the elements off the diagonal indicate the
numbers of alleles shared by sib animals.
The diagonal residual covariance matrix Rσ 2 E of dimensions
f × f is then constructed. This matrix corresponds to the residual
effect (e) in Eqn (16). Relative to Eqn (15), the elements on diagonal
R are connected with the reliabilities of BV estimates for particular
sires, but only on the basis of their progeny from which DYD were
computed (excluding other sources of information):
Rjj = (1 − r 2 jP)/r 2 jP
(18)
where r 2 jP is the partial reliability of the sire’s BV based on his
progeny.
Based on the theory of selection indices, it is then possible
to determine the direct genetic value of sires with known
daughter performance (DGVS) 72 by adapting 2DYD for the vector
of observations: 42
DGVS = G[G + Rk] −1 2DYD (19)
where k = σ 2 E/σ 2 A.
The genomic covariance matrix between proven sires (S) and
young unevaluated animals (O) is
C = ([TO − QO][TS − QS] ′ )/(2�qi(1 − qi)) (20)
where TO and TS are the known matrices assigning particular loci
to the young animals and proven sires and QO and QS are Q
matrices that have been modified according to the number of
young animals and proven sires included.
The predicted direct genetic value for young animals (DGVO)isa
genomic regression based on proven animals with already known
BV:
DGVO = CG −1 DGVS
(21)
The solution by means of the selection index according to
Eqns (17)–(21) is identical to the preceding solutions in Eqns (16)
and(3) butthedimensionsofthematricesaresubstantiallysmaller,
corresponding to the numbers of genotyped animals (f). 72,73 The
estimation of genetic regression coefficients according to the
particular loci (v) may also be omitted. Hence the solution is
simplified and does not require iterative methods. 72 Therefore,
a direct determination of DGV estimate reliability is feasible. For
sires with known daughter performance (S), reliability estimates
correspond to the diagonal elements of the term:
G[G + Rk] −1 G (22)
For young animals without known performance (O), reliability
estimates correspond to the diagonal elements of the term
C[G + Rk] −1 C ′
(23)
A similar solution can also be obtained by the weighted analysis
of a linear model as in Eqn (1) with DYD substituted for input data
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1765–1773

Evaluation of animals by simple heritable markers www.soci.org
and weighted according to Eqn (18). In this case, A is substituted
by G and Xb covers only the general mean. 72,73
A one-step approach
The process of evaluation described above has several disadvantages,
namely that it is influenced by the input parameters used
in a multi-step procedure. Inaccuracies in these parameters may
bias the evaluation. It is also difficult to compare genotyped and
ungenotyped animals evaluated by different procedures. This may
be overcome by incorporating all parts of the evaluation into a
one-step procedure.
From Eqn (19), it follows that molecular-genetic information
is collected in G. The additive numerator relationship matrix
(A) is probability based and deviates from expected values
due to random Mendelian sampling. 74 The realised genomic
relationship matrix (G) should therefore be more precise and
lead to more precise selection. 73 A single-step evaluation using
original measured performances (Y) as input has been proposed,
in which the pedigree-based numerator relationship matrix (A)
covering all evaluated animals is augmented by a contribution
from (G) with genotyped animals. 75
AmatrixHhas been derived, which is substituted for the
usual matrix (A) in Eqn (1). 76 Further, a computational procedure
has been developed for the solution of animal models directly
from the accumulated measured data of all genotyped and
non-genotyped animals in large commercial populations. 6,75 The
essential component of the system of equations constructed
according to Eqn (1) is the inverse of the relationship matrix, in
this case:
H −1 = A −1 �
0 0
+
0 λ(G−1 − A −1
22 )
�
(24)
where H is the pedigree–genomic relationship matrix, λ is a scaling
factor and A22 is a block of A that corresponds to the genotyped
animals.
This one-step procedure eliminates several assumptions that
must be made for multi-step procedures. It is less biased and allows
the evaluation of large commercial populations even when only
some individuals in the population are genotyped. This improves
evaluationaccuracybothforgenotypedandungenotypedanimals
and generates a single common rank for all animals. This model
further enables the use of multi-trait AM and models with different
complexities, which are now common in animal evaluations. 6
CONCLUSIONS
The majority of traits are conditioned in a complex way; it can
be assumed that a production trait will only rarely be genetically
conditioned in a simple way by a small number of independent
genes. In practice, a large number of markers and a large number
of animals with accurate performance estimates are necessary for
the reliable evaluation of animals. 2,3
Simplifications are often used in evaluations, but at the cost
of lower reliability of results. A description of the variability
of the evaluated data and the validity of the model are
therefore necessary. Considerable attention should be paid to
the development of ‘traditional’ methods of BV estimation on
the basis of polygenes, which enables the correct adjustment of
performance for environmental effects.
Typically, a two-step evaluation of genotyped animals is
performed. The first step is the estimation of ‘traditional’ BV
for recorded traits according to BLUP-AM or a similar method.
Based on these results, DYD are computed for particular sires with
a highly reliable BV. In the second step, GEBV is determined for
young animals without performance records.
Marker-assisted selection is gradually being replaced by
genomic selection, in which a huge number of genetic markers
(SNPs) that densely cover the entire genome are analysed. In
this case, polygenic effects can be eliminated from the solution
without strongly affecting the results.
Molecular-genetic information can be gathered in the genomic
relationship matrix G in order to estimate DGV and GEBV with
computationally simpler procedures.
A one-step procedure that combines the animal model with a
pedigree–genomic relationship matrix can be used to evaluate all
animals in a population. This protocol is useful as it produces more
accurate evaluations than do other methods and also generates a
common rank for all genotyped and ungenotyped animals in the
population.
The methods described here have significant practical importance
in animal breeding.
ACKNOWLEDGEMENTS
This work was supported bythe Ministryof Agriculture of the Czech
Republic (MZe 0002701404) and by the Ministry of Education of
theCzechRepublic(MSM6007665806).Wegratefullyacknowledge
the helpful comments of anonymous reviewers.
REFERENCES
1 Czarnik U, Galinski M, Pareek ChS, Zabolewicz T and Wielgosz-Groth Z,
Study of an association between SNP 775C>T within the bovine
ITBG2 gene and milk performance traits in Black and White cows.
Czech J Anim Sci 52:1–6 (2007).
2 VanRaden PM, Van Tassell CP, Wiggans GR, Sonstegard TS,
Schnabel RD, Taylor JF, et al, Invited review: Reliability of genomic
predictions for North American Holstein bulls. JDairySci92:16–24
(2009).
3 Hayes BJ, Bowman PJ, Chamberlain AJ and Goddard ME, Invited
review: Genomic selection in dairy cattle: progress and challenges.
JDairySci92:433–443 (2009).
4 Seefried F, Liu Z, Thaller G and Reinhadt F, Genomic breeding value of
German Holstein. Zuchtungskunde 82:14–21 (2010).
5 Guillaume F, Fritz S, Boichard D and Druet T, Short communication:
Correlations of marker-assisted breeding values with progeny test
breeding values for eight hundred ninety-nine French Holstein
bulls. JDairySci91:2520–2522 (2008).
6 Aguilar I, Misztal I, Johnson D, Legarra A, Tsuruta S and Lawlor TJ, Hot
topic: An unified approach to utilize phenotypic, full pedigree, and
genomic information for a genetic evaluation of Holstein final score.
JDairySci93:743–752 (2010).
7 Ben Jemaa S, Fritz S, Guillaume F, Druet T, Denis C, Eggen A, et al,
Detection of quantitative trait loci affecting non-return rate in
French dairy cattle. J Anim Breed Genet 125:280–288 (2008).
8 Holmberg M, Sahana G and Andersson-Eklund L, Fine mapping of a
quantitative trait locus on chromosome 9 affecting non-return rate
in Swedish dairy cattle. J Anim Breed Genet 124:257–263 (2007).
9 Interbull, [Online]. Available: http://www.interbull.org [21 October
2009].
10 Amaral AJ, Megens HJ, Kerstens HHD, Heuven HCM, Dibbits B,
Crooijmans RPMA, et al, Application of massive parallel sequencing
to whole genome SNP discovery in the porcine genome. BMC
Genomics 10:374 (2009).
11 MuirWM, WongGK, ZhangY, WangJ, GroenenMAM, Crooijmans
RPMA, et al, Genome-wide assessment of worldwide chicken
SNP genetic diversity indicates significant absence of rare alleles
in commercial breeds. Proc Natl Acad Sci USA 105:17312–17317
(2008).
12 Misztal I, Reliable computing in estimation of variance components.
J Anim Breed Genet 125:363–370 (2008).
J Sci Food Agric 2010; 90: 1765–1773 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1771

1772
13 VeseláZ,Pˇribyl J, ˇSafus P, Vostr´yL,ˇSeba K and ˇStolc L, Breeding value
for type traits in beef cattle in the Czech Republic. Czech J Anim Sci
50:385–393 (2005).
14 KrejčováH, Pˇribyl J, PˇribylováJ, ˇStípková M and Mielenz N, Genetic
evaluation of daily gains of dual-purpose bulls using a random
regression model. Czech J Anim Sci 53:227–237 (2008).
15 OravcováM, Margetín M, PeˇskovičováD, Daňo J, Milerski M,
Hetényi L, et al, Factors affecting ewe’s milk fat and protein content
and relationships between milk yield and milk components. Czech
JAnimSci52:189–198 (2007).
16 Pˇribyl J, KrejčováH, Pˇribylová J, Misztal I, BohmanováJ and
ˇStípková M, Trajectory of body weight of performance tested dualpurpose
bulls. Czech J Anim Sci 52:315–324 (2007).
17 Pˇribyl J, KrejčováH, Pˇribylová J, Misztal I, Tsuruta S and Mielenz N,
Models for evaluation of growth of performance tested bulls. Czech
JAnimSci53:45–54 (2008).
18 Pˇribyl J, PˇribylováJ,Krejčová H and Mielenz N, Comparison of different
traits to evaluate the growth of bulls. Czech J Anim Sci 53:273–283
(2008).
19 Zavadilová L, Jamrozik J and Schaeffer LR, Genetic parameters for testday
model with random regressions for production traits of Czech
Holstein cattle. Czech J Anim Sci 50:142–154 (2005).
20 Vostr´yL, Pˇribyl J, Jakubec V, Veselá Z and Majzlík I, Selection of a
suitable definition of environment for the estimation of genotype
× environment interaction in the weaning weight of beef cattle.
Czech J Anim Sci 53:407–417 (2008).
21 Vostr´yL,Pˇribyl J, Veselá Z and Jakubec V, Selection of a suitable data
set and model for the estimation of genetic parameters of the
weaning weight in beef cattle. Arch Tierz 50:562–574 (2007).
22 Mészáros G, Wolf J and Kadlečík O, Factors affecting the functional
length of productive life in Slovak Pinzgau cows. Czech J Anim Sci
53:9–97 (2008).
23 Páchová E,ZavadilováLandSölkner J,Geneticevaluationofthelength
of productive life in Holstein cattle in the Czech. Czech J Anim Sci
50:493–498 (2005).
24 Bozdogan H, Akaike’s information criterion and recent developments
in information complexity. J Math Psychol 44:62–91 (2000).
25 Guo Z and Schaeffer LR, Random regression submodel comparison,
in 7th World Congress on Genetics Applied to Livestock Production
(WCGALP), Montpellier, France, 19–23 August (2002).
26 Kaas ER and Raftery AE, Bayes factors. J Am Stat Assoc 90:773–795
(1995).
27 Sorensen D and Waagepetersen R, Model selection for prediction of
breeding value, in 7thWorldCongressonGeneticsAppliedtoLivestock
Production (WCGALP), Montpellier, France, 19–23 August (2002).
28 Kaps M and Lamberson W, Biostatistics for Animal Science. CABI
Publishing, Wallingford, UK (2004).
29 Pˇribyl J, [A way of using markers for farm animals selection] (in Czech).
Ziv Vyr 40:375–382 (1995).
30 Schaeffer LR, Strategy for applying genome-wide selection in dairy
cattle. J Anim Breed Genet 123:218–223 (2006).
31 Baierl A, Bogdan M, Frommlet F and Futschik A, On locating multiple
interacting quantitative trait loci in intercross designs. Genetics
173:1693–1703 (2006).
32 Szyda J and Komisarek J,Statistical modeling of candidate gene effects
on milk production traits in dairy cattle. JDairySci90:2971–2979
(2007).
33 Meuwissen THE, Hayes BJ and Goddard ME, Prediction of total
genetic value using genome-wide dense marker maps. Genetics
157:1819–1829 (2001).
34 Muir WM, Comparison of genomic and traditional BLUP-estimated
breeding value accuracy and selection response under alternative
trait and genomic parameters. J Anim Breed Genet 124:342–355
(2007).
35 Dekkers JCM, Prediction of response to marker-assisted and genomic
selection using selection index theory. J Anim Breed Genet
124:331–341 (2007).
36 Falconer DS and Mackay TFC, Introduction to Quantitative Genetics.
Longmans Green, Harlow, UK (1996).
37 Milerski M, The Booroola (FecB) gene in Czech Merinolandschaft
population, in 59th Annual Meeting of EAAP (Book of abstracts
191), Vilnius, Lithuania, 24–27 August (2008).
38 Soller M and Genizi A, The efficiency of experimental designs for the
detection of linkage between a marker locus and a locus affecting a
quantitative trait in segregating populations. Biometrics 34:47–55
(1978).
www.soci.org J Pribyl et al.
39 Dolezal M, Schwarzenbacher H, Soller M, Sölkner J and Visscher PM,
Multiple-marker mapping for selective DNA pooling within large
families. JDairySci91:2864–2873 (2008).
40 Visscher PM, Proportion of the variation in genetic composition in
backcrossing programs explained by genetic markers. J Hered
87:136–138 (1996).
41 Freyer G, Sørensen P, Kühn C and Weikard R, Investigations in the
character of QTL affecting negatively correlated milk traits. JAnim
Breed Genet 121:40–51 (2004).
42 Ansari-Mahyari S, Sørensen AC, Lund MS, Thomsen H and Berg P,
Across-family marker-assisted selection using selective genotyping
strategiesindairycattlebreedingschemes.JDairySci91:1628–1639
(2008).
43 Calus MPL and Veerkamp RF, Accuracy of breeding value when using
and ignoring the polygenic effect in genomic breeding value
estimation with a marker density of one SNP per cM. J Anim Breed
Genet 124:362–368 (2007).
44 Gengler N, Abras S, Verkenne C, Vanderick S, Szydlowski M and
Renaville R, Accuracy of prediction of gene content in large animal
populations and its use for candidate gene detection and genetic
evaluation. JDairySci91:1652–1659 (2008).
45 Szyda J, Ptak E, Komisarek J and Zarnecki A, Practical application of
daughter yield deviations in dairy cattle breeding. J Appl Genet
49:83–191 (2008).
46 Simianer H and Stricker C, Mapping of quantitative trait loci,
in Genomics and Biotechnology in Livestock Breeding, ed. by
Schellander K, Li N, Neeteson AM and Wimmers K. Shaker, Aachen,
pp. 47–58 (2002).
47 Bílek K, Knoll A, Stratil A, SvobodováK, Horák P, BechyňováR, et al,
Analysis of mRNA expression of CNN3, DCN, FBN2, POSTN, SPARC
and YWHAQ genes in porcine foetal and adult skeletal muscles.
Czech J Anim Sci 53:181–186 (2008).
48 Pan CY, Lan XY, Chen H, Yang DY, Hua LS, Yang XB, et al, ADdeI PCR-
RFLP detecting a novel missense mutation of the POU1F1 gene
showednoeffectsongrowthtraitsincattle.Czech J Anim Sci
53:532–527 (2008).
49 Spangler ML, Sapp RL, Bertrand JK, MacNeil MD and Rekaya R,
Different methods of selecting animals for genotyping to maximize
the amount of genetic information known in the population. JAnim
Sci 86:2471–2479 (2008).
50 Daley JW and Shepherd RK, Utilization of F1 information in estimating
QTL effects in F2 crosses between outbred lines. J Anim Breed Genet
125:35–44 (2008).
51 Weller JI, Kashi Y and Soller M, Power of daughter and granddaughter
design for determining linkage between marker loci and
quantitative trait loci in dairy cattle. JDairySci 73:2525–2537
(1990).
52 Freyer G and Vukasinovic N,Comparison of granddaughterdesign and
general pedigree design analysis of QTL in dairy cattle: a simulation
study. Czech J Anim Sci 50:545–552 (2005).
53 Goliáˇsová E and Wolf J, Herd specific effects of the ESR gene on litter
size and production traits in Czech Large White sows. Czech J Anim
Sci 49:373–382 (2004).
54 Legarra A and Misztal I, Technical note: Computing strategies in
genome-wide selection. JDairySci91:360–366 (2008).
55 Meuwissen THE and Goddard ME, Prediction of identity by descent
probabilities from marker-haplotypes. Genet Sel Evol 33:605–634
(2001).
56 Humpolíček P, Urban T, Matouˇsek V and TvrdoňZ,Effectofestrogen
receptor, follicle stimulating hormone and myogenin genes on the
performance of Large White sows. Czech J Anim Sci 52:334–340
(2007).
57 Vykoukalová Z, Knoll A and Cepica S, Porcine perilipin (PLIN) gene:
structure, polymorphism and association study in Large White pigs.
Czech J Anim Sci 54:359–364 (2009).
58 HradeckáE,Čítek J, Panicke L, ˇRehout V and HanusováL,Therelation
of GH1, GHR and DGAT1 polymorphisms with estimated breeding
values for milk production traits of German Holstein sires. Czech J
Anim Sci 53:238–246 (2008).
59 Matějíček A, MatějíčkováJ, NěmcováE, JandurováOM, ˇStípkováM,
Bouˇska J, et al, Joint effects of CSN3 and LGB genotypes and their
relation to breeding values of milk production parameters in Czech
Fleckvieh. Czech J Anim Sci 52:83–87 (2007).
60 Matějíček A, MatějíčkováJ, ˇStípkováM, HanuˇsO, GenčurováV,
Kysel’ováJ, et al, JointeffectsofCSN3 and LGB genes on milk
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1765–1773

Evaluation of animals by simple heritable markers www.soci.org
quality and coagulation properties in Czech Fleckvieh. Czech J Anim
Sci 53:246–252 (2008).
61 VanRaden PM and Wiggans GR, Derivation, calculation, and use of
national animal model information. JDairySci74:2737–2746 (1991).
62 Neuner S, Emmerling R, Thaller G and Götz KU, Strategies for
estimating genetic parameters in Marker-Assisted Best Linear
Unbiased Predictors models in dairy cattle. JDairySci91:4344–4354
(2008).
63 Harder B, Bennewitz J, Reinsch N, Thaller G, Thomsen H, Kühn C, et al,
Mapping of quantitative trait loci for lactation persistency traits in
German Holstein dairy cattle. J Anim Breed Genet 123:89–96 (2006).
64 Uleberg E, Widerøe IS, Grindflek E, Szyda J, Lien S and Meuwissen THE,
FinemappingofaQTLforintramuscularfatonporcinechromosome
6 using combined linkage and linkage disequilibrium mapping.
J Anim Breed Genet 122:1–6 (2005).
65 Goddard ME and Hayes BJ, Genomic selection. J Anim Breed Genet
124:323–330 (2007).
66 Weller JI, Golik M, Seroussi E, Ezra E and Ron M, Population-wide
analysis of a QTL affecting milk-fat production in the Israeli Holstein
population. JDairySci86:2219–2227 (2003).
67 SAS, SAS/STAT User’s Guide, Version 9.1. SAS Institute, Cary, NC (2005).
68 LiFE,MeiSQ,DengCY,JiangSW,ZuoB,ZhengR,et al, Association
of a microsatellite flanking FSHB gene with reproductive traits and
reproductive tract components in pigs. CzechJAnimSci 53:139–144
(2008).
69 MichalcováA and Krupová Z, Influence of composite κ-casein and
β-lactoglobulin genotypes on composition, rennetability and heat
stability of milk of cows of Slovak Pied breed. Czech J Anim Sci
52:292–298 (2007).
70 Tsuruta S and Misztal I, Technical note: Computing options for genetic
evaluation with a large number of genetic markers. JAnimSci
86:1514–1518 (2008).
71 Long N, Gianola D, Rosa GJM, Weigel KA and Avendaño S, Machine
learning classification procedure for selecting SNPs in genomic
selection: application to early mortality in broilers. J Anim Breed
Genet 124:377–389 (2007).
72 VanRaden PM, Efficient methods to compute genomic predictions.
JDairySci91:4414–4423 (2008).
73 Hayes BJ, Visscher PM and Goddard ME, Increased accuracy of artificial
selection by using the realised relationship matrix. Genet Res Camb
91:47–60 (2009).
74 Guo SW, Variation in genetic identity among relatives. Hum Hered
46:61–70 (1996).
75 Misztal I, Legarra A and Aguilar I, Computing procedures for genetic
evaluation including phenotypic, full pedigree, and genomic
information. JDairySci92:4648–4655 (2009).
76 Legarra A, Aguilar I and Misztal I, A relationship matrix including full
pedigreeandgenomicinformation.JDairySci92:4656–4663(2009).
J Sci Food Agric 2010; 90: 1765–1773 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1773

1774
Research Article
Received: 26 May 2009 Revised: 25 March 2010 Accepted: 29 March 2010 Published online in Wiley Interscience: 22 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.3998
Root colonisation by the arbuscular
mycorrhizal fungus Glomus intraradices alters
the quality of strawberry fruits (Fragaria ×
ananassa Duch.) at different nitrogen levels
Vilma Castellanos-Morales, a∗ Javier Villegas, b Silvia Wendelin, c
Horst Vierheilig, d Reinhard Eder c and Raúl Cárdenas-Navarro a
Abstract
BACKGROUND: Arbuscular mycorrhizal fungi (AMF) increase the uptake of minerals from the soil, thus improving the growth
of the host plant. Nitrogen (N) is a main mineral element for plant growth, as it is an essential component of numerous plant
compounds affecting fruit quality. The availability of N to plants also affects the AMF–plant interaction, which suggests that the
quality of fruits could be affected by both factors. The objective of this study was to evaluate the influence of three N treatments
(3, 6 and 18 mmol L −1 ) in combination with inoculation with the AMF Glomus intraradices on the quality of strawberry fruits.
The effects of each factor and their interaction were analysed.
RESULTS: Nitrogen treatment significantly modified the concentrations of minerals and some phenolic compounds, while
mycorrhization significantly affected some colour parameters and the concentrations of most phenolic compounds. Significant
differences between fruits of mycorrhizal and non-mycorrhizal plants were found for the majority of phenolic compounds and
for some minerals in plants treated with 6 mmol L −1 N. The respective values of fruits of mycorrhizal plants were higher.
CONCLUSION: Nitrogen application modified the effect of mycorrhization on strawberry fruit quality.
c○ 2010 Society of Chemical Industry
Keywords: mycorrhizal; nitrogen; strawberry; quality; fruit
INTRODUCTION
Most land plants benefit from their interaction with symbiotic
soil-borne fungi known as arbuscular mycorrhizal fungi (AMF).
In this symbiosis the AMF receives carbon from the plant, while
the fungus takes up nutrients with its extraradical mycelium and
provides them to the host plant. 1
The uptake of nitrogen (N) by the extraradical mycelium has
been shown before and this N is available to the host plant, 2,3
so the AMF improves the N status of the host. 4,5 Nevertheless, it
also has been reported that the N availability in the soil affects the
dynamic of plant–AMF association. 4,6
Nitrogen is an essential element for plant growth. Owing
to its role in the synthesis of proteins, nucleic acids, various
coenzymes and many products of secondary plant metabolism, 7
it is important for strawberry fruit quality. It has been shown
that a leaf N concentration below 19 g kg −1 (deficiency) causes
chlorosis of strawberry leaves, thus decreasing the leaf area, fruit
size and anthocyanin concentration, 8 whereas an excess of foliar N
(∼40 g kg −1 ) promotes vegetative growth, delays fruit maturation
and causes a loss of firmness in fruits, thus reducing quality. 9,10
Strawberry quality and consumer preference for strawberry
fruits are determined by parameters such as size, firmness, levels
of soluble sugars and acid concentration, the last of which affects
the aromatic compounds that impart flavour and aroma. 11,12
Strawberry fruits possess antioxidant activity owing to their high
content of anthocyanins, flavonoids, phenolic acids and other
compounds. 13
Recent data suggest that mycorrhization not only has a positive
effect on various plant growth parameters but can also affect
the quality of crop products. For example, root colonisation by
different AMF enhances the essential oil concentration in a number
of plants from different plant families such as oregano (Origanum
vulgare), 14 basil (Ocimum basilicum L.), 15,16 menthol mint (Mentha
∗ Correspondence to: Vilma Castellanos-Morales, Estación Experimental del
Zaidín, Prof. Albareda 1, Apdo 419, E-18008 Granada, Spain.
E-mail: vilma c 99@yahoo.com
a Instituto de Investigaciones Agropecuarias y Forestales, Universidad
Michoacana de San Nicolás de Hidalgo, Km 9.5, Carretera Morelia-Zinapécuaro,
CP 58880, Tarímbaro, Michoacán, Mexico
b Instituto de Investigaciones Químico-Biológicas, Universidad Michoacana de
San Nicolás de Hidalgo, CP 58000, Ciudad Universitaria, Morelia, Michoacán,
Mexico
c Federal College and Research Institute for Viticulture and Pomology,
Wienerstrasse 74, A-3400 Klosterneuburg, Austria
d Estación Experimental del Zaidín (CSIC), E-18008 Granada, Spain
J Sci Food Agric 2010; 90: 1774–1782 www.soci.org c○ 2010 Society of Chemical Industry

Effects of AMF and N level on quality of strawberry fruits www.soci.org
arvensis) 17 andcoriander(CoriandrumsativumL.). 18 Inothersplants
such as alfalfa (Medicago sativa L.), 19–21 barrel medic (Medicago
truncatula), 22 red clover (Trifolium pratense) 23 and soybean (Glycine
max L.), 24 increases in flavonoid levels after mycorrhization have
been reported.
There are several reports on strawberry plants concerning
inoculation with AMF and its effects on plant growth. It has been
shown that AMF root colonisation stimulates plant growth, 25
modifies the production of runners, 26 enhances photosynthesis 27
and increases the number of fruits. 28 However, to the best
of our knowledge, there are currently no data on how AMF
root colonisation in combination with different N levels affects
strawberry fruit quality parameters such as colour, soluble sugars,
acids, minerals and phenolic compounds.
MATERIALS AND METHODS
The experiment was conducted in a ‘shade’-type greenhouse
with 30% shade at the Instituto de Investigaciones Agropecuarias
y Forestales (IIAF), Universidad Michoacana de San Nicolás de
Hidalgo (UMSNH), Morelia, Michoacán, Mexico. Maximum and
minimum temperatures in the greenhouse varied between 28 and
32 ◦ C and between 8 and 18 ◦ C respectively.
Plants of the strawberry cultivar ‘Aromas’ were used that had
previously been grown in a sterilised (95 ◦ C water/steam, 40 min)
substrate of coconut fibre/perlite (1 : 3 v/v) under greenhouse
conditions. Before the experiment was established, the absence of
AMF in the roots was verified by the ink and vinegar technique, 29
modifying the duration of immersion in KOH and ink/vinegar
solution (7 and 5 min respectively). Before planting, roots were
disinfected by submerging them for 20 s in 20 g L−1 sodium
hypochlorite solution and rinsing them in water.
The inoculum was prepared with spores of Glomus intraradices
cultivated in liquid medium (3.5 × 106 spores L−1 , 90% viability;
Premier Tech Biotechnologies Company, Quebec, Canada), which
was diluted with fitagel (Sigma P-8169, Saint Louis, MO, USA)
solution at 50 g L−1 to obtain a final concentration of about 5×104 spores L−1 . The viability of spores was determined according to
the method of An and Hendrix. 30
Eighteen days after setting up the experiment, each plant
received 2 mL of inoculum applied directly to the recently formed
roots. One month later, after staining, 29 the percentage of root
colonisation was determined by the gridline intersect method. 31
The experiment was organised as a full factorial, completely
randomised design with two factors: inoculation (two levels:
mycorrhizal and non-mycorrhizal plants) and N concentration
in the nutrient solution (three levels: 3, 6 and 18 mmol L−1 ).
The six treatments were replicated four times, producing 24 experimental
units with ten plants each. Every second day, all plants
were irrigated up to substrate saturation. Nitrogen was supplied
as NO − 3
and the cation/anion ratio was kept constant by varying
the concentration of SO 2−
4 . When N was below 18 mmol L−1 ,the
cation concentrations were maintained as follows: K + ,3;Ca 2+ ,3.5;
Mg 2+ , 1.5 mmol L −1 . They were increased in the 18 mmol L −1 N
treatment: K + ,6.5;Ca 2+ ,7.5;Mg 2+ , 3.25 mmol L −1 .Inallnutrient
solutions the concentration of phosphorus (P) was 0.3 mmol L −1 .
The other nutrients in the solutions were: H3BO3, 20; CuSO4 ·5H2O,
0.5; Fe-EDTA (Ethylenediaminetetraacetic acid iron (III) sodium
salt), 15; MnSO4·H2O, 12; (NH4)6Mo7O24 ·4H2O, 0.05; ZnSO4 ·7H2O,
3 µmol L −1 . The pH was adjusted to 5.5 at every application date.
Mature fruits of each experimental unit were collected between
140 and 160 days after setting up the experiment. At sampling
time the fruits were separated into two equal batches. One batch
was used for the determination of fruit fresh weight, diameter,
length and Brix grade (total solids). The last measurement was
done at 25 ◦ C using a refractometer (ATAGO CO., LTD) (N-1α). The
other batch was frozen in liquid nitrogen and stored at −20 ◦ C.
Prior to chemical and colour analyses, these samples were ground
to a fine powder (Retsch MM200 mill, Thomas Scientific, New
Jersey, United States) in liquid N2 and then freeze-dried.
Titratable acidity is a measure of organic acids in a sample and
is determined by adding enough alkali of known molarity to the
sample to neutralise all acids present. For the measurement of
titratable acidity, 0.1 g of freeze-dried fruit was mixed with 5 mL
of distilled water and shaken, then 0.05 mol L −1 NaOH was added
up to a pH of 8.1. The results are expressed as % citric acid.
For macro- and micro-nutrient determination, 10 mL of distilled
water was added to 0.2 g of ground sample. The mixture was
sonicated (FS30H, Fisher Scientific, Pittsburgh, United State)
and then centrifuged (2744 × g ′ , 10 min). The supernatant was
filtered through a 0.45 µm membrane (Millipore, Thebarton, South
Australia). For macronutrient measurement, 9 mL of 0.5 mol L −1
HCl and 200 µL of lanthanum oxide were added to 1 mL of the
filtrate. For micronutrient determination, 200 µL of concentrated
HCl was added to 9 mL of the filtrate. All samples were shaken on
a vortex for 5 min and their mineral contents were quantifed
by atomic absorption (Solar 939, ATI Unicam, Basingstoke,
U.K).
Soluble sugars were extracted by the method of Gomez
et al., 32 with some modifications. All extractions were carried
outat4 ◦ C. Briefly, 4 mL of methanol/water (1 : 1 v/v) and 1 mL
of chloroform were added to 15 mg of lyophilised sample. The
mixture was shaken on a vortex for 2 min and then on a horizontal
agitator (Libline 4638, Melrose Park, Illinois) at medium speed
for 30 min. After centrifugation (1585 × g ′ , 30 min), two liquid
phases separated by the plant powder were obtained. A 2.8 mL
volume of the methanol/water supernatant was recovered and
dried in a vacuum evaporator (Labconco 7810000 Speed-Vac,
Kansas City, Missouri). The resulting pellet was stored at −20 ◦ C
overnight. Next day it was redissolved in 2 mL of distilled water
by shaking on a vortex for 20 min. The aqueous extract was then
poured into a tube with 0.015 g of polyvinylpyrrolidone (Sigma
P6755) to remove residual phenols by crosslinking. After shaking
on a vortex for 20 min, the tube was centrifuged (1585 × g ′ ,
90 min). The supernatant was recovered using a 1 mL insulin
syringe and stored at −20 ◦ C for the direct measurement of
glucose and the indirect measurement of fructose and sucrose
by the enzymatic method 33 with a photometer (Multiskan
Ascent 354, Thermolabsystem, Finlandia imported by Labtech,
Mexico) at 340 nm, using a calibration curve in the range
0–0.2g L −1 glucose (Baker 1916-01, Xalostoc, Edo. México). To
verify the correct measurement of soluble sugars, controls of
fructose (Sigma F0127) and sucrose (Sigma S7903) were used.
Before measurement the extract was diluted with distilled water
(1 : 20 v/v).
Total phenols and the anthocyanins cyanidin-3-glucoside and
pelargonidin-3-glucoside were extracted by the method of
Markakis, 34 with some modifications. Briefly, 5 mL of methanol/HCl
(1 : 5 v/v) was added to 0.1 g of lyophilised sample. The mixture
was sonicated for 300 s and then centrifuged (2744 × g ′ ,10min).
The supernatant was filtered (0.45 µm membrane, Millipore) and
the filtrate obtained was used for the measurement of total
phenol and anthocyanin concentrations. Colour parameters and
the absorbance at 500 nm were also measured in the same filtrate.
J Sci Food Agric 2010; 90: 1774–1782 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1775

1776
Total phenol concentration was quantified by the Folin–
Ciocalteu method, 35 with minor modifications. The volumes of
sample, Folin–Ciocalteu’s phenol reagent and sodium carbonate
were reduced to one-tenth of those used in the original method,
giving a final volume of 20 mL. The measurement was made
at 765 nm (Spectrophotometer – Cintra 10e, GBC, Dandenong,
Victoria, Australia), using a linear calibration curve of caffeic acid
(0–250 mg L −1 ) to calculate the total phenol concentration.
Strawberry fruit colour was determined by measuring the
absorbance at 500 nm 36 with a spectrophotometer (Shanghai,
Analytical Instrument LTD, China) (HP-8452A, Cheadle Heath,
Stockport Cheshire, UK). Additionally, colour was measured using
a photometer (Licor-2000, DR Lange, Dusseldorf, Germany) in
terms of L ∗ , a ∗ and b ∗ values, where L ∗ defines lightness (from
white = 100 to black = 0), a ∗ defines red/greenness (from −60
to +60) and b ∗ defines blue/yellowness (from −60 to +60). From
the a ∗ and b ∗ values the following colour parameters were also
calculated: colour evolution (a ∗ /b ∗ ), shade (tan −1 (b ∗ /a ∗ ), ranging
from 0 ◦ (red) to 90 ◦ (yellow) to 270 ◦ (blue)) and chromaticity
(C ∗ = (a 2 + b 2 ) 1/2 , indicating the vividness of colour and ranging
from 0 (discoloured) to 60 (powerful)).
Phenolic acids and flavonols were extracted by acid hydrolysis. 37
Briefly, 7.5 mL of 5.33 g L −1 ascorbic acid solution, 12.5 mL of
methanol (liquid chromatography/mass spectrometry grade) and
5mLof6molL −1 HCl were added to 0.25 g of sample. The mixture
was sonicated for 2 min, the air in the mixture was replaced with
gaseousN2 (1–1.5 min)andthemixturewasshakenonahorizontal
agitator (35 ◦ C) for 16 h. The cold sample was filtered (0.45 µm
membrane, Millipore), concentrated in a rotavapor (35 ◦ C) and
redissolved in 1 mL of methanol. This solution was filtered (0.45 µm
membrane, Millipore) and 10 µL of the filtrate was used for the
measurement of phenolic acids and flavonols.
Phenolic compounds (anthocyanins, phenolic acids and
flavonols) were quantified by reverse phase high-performance
liquid chromatography (RP-HPLC) 38 using an Agilent 1090 Aminoquant
HPLC system (Waldbrot, Germany). Each 10 µL sample was
injected for separation on two narrow-bore HP-ODS Hypersil RP-
18 columns (Shandon, U.K) (5 µm, 200 mm × 2.1 mm and 5 µm,
100 mm × 2.1 mm) linked in series. A linear gradient of 5 g L −1
formic acid (pH 2.3) and methanol at a flow rate of 0.2 mL min −1
was used. The column temperature was 40 ◦ C and detection was
achieved at 320 nm for all compounds. The standards used and the
concentration ranges of their calibration curves were as follows:
callistephin (Extrasynthese 0907S, Lyon, France), 1–200 mg L −1 ;
kuromanin (Extrasynthese 0915S), 1–200 mg L −1 ; gallic acid
monohydrate (Roth 7300, Karlsruhe, Germany), 10.9–545 mg L −1 ;
p-coumaric acid (Roth 9908), 26.2–1308 mg L −1 p-coumaric acid;
ferulic acid (Roth 9936), 9.8–490 mg L −1 ; ellagic acid (Sigma
E2250), 8.1–81.4 mg L −1 ; quercetin dehydrate (Extrasynthese
1135S), 8.7–436 mg L −1 ; kaempferol (Fluka 60010, Saint Louis,
MO, USA), 8.5–426 mg L −1 ; catechin (Roth 6200), 9.2–460 mg L −1 .
The results are presented as means of four replicates (each
replicate consists of fruits from all plants in one experimental
unit). Statistical analyses were performed using SYSTAT for
Windows, Version 9.01 (Systat Software versión 9.01, Cranes
Software International, LTD). The effects of each single factor
(N concentration and inoculation) and their interaction (N
concentration × inoculation) were evaluated using two-way
analysis of variance (ANOVA). Multiple comparisons were made
using Tukey’s test. Differences at P < 0.05 were considered
significant.
www.soci.org V Castellanos-Morales et al.
Table 1. Fresh weight, diameter and length of fruits of strawberry
plants inoculated with Glomus intraradices and fertilised with different
nitrogen concentrations in irrigation water
Factor Fresh weight (g per fruit) Diameter (cm) Length (cm)
Nitrogen concentration (mmol L −1 )
3 14.19a 2.98a 3.30a
6 13.14a 2.93a 3.22a
18 13.36a 2.93a 3.26a
Inoculation
M 13.39a 2.93a 3.22a
NM 13.73a 2.96a 3.26a
Interaction (nitrogen concentration × inoculation)
3 × M 14.44a 3.01a 3.23a
3 × NM 13.94a 2.96a 3.27a
6 × M 12.47a 2.88a 3.19a
6 × NM 13.80a 2.98a 3.25a
18 × M 13.27a 2.90a 3.27a
18 × NM 13.46a 2.96a 3.25a
Each value represents the mean of four replicates. Two-way ANOVA was
applied for each parameter; when statistical differences were found,
aTukeytest(P < 0.05) was conducted independently for nitrogen
concentration (3, 6 and 18 mmol L −1 ), inoculation (mycorrhizal (M) and
non-mycorrhizal (NM)) and nitrogen concentration × inoculation (3 ×
M, 3 × NM, 6 × M, 6 × NM, 18 × Mand18× NM). For each factor,
means with the same letter in a column do not differ significantly.
RESULTS
Tables 1–5 show the results for the effects of the two factors and
their interaction on the variables evaluated. At the end of the
experiment the extent of AMF colonisation ranged from 65 to
80%. None of the treatments affected the fresh weight, diameter
and length of fruits (Table 1).
In terms of colour, different N concentrations resulted in statistically
significant effects only on fruit lightness and absorbance
at 500 nm (Table 2). Lightness was significantly higher and absorbance
was significantly lower in fruits of plants fertilised with
3 mmol L −1 N than in fruits of plants treated with 6 mmol L −1
N, but both values did not differ from those in fruits of plants
fertilised with 18 mmol L −1 N. Mycorrhization resulted in statistically
significant effects on all colour parameters except colour
evolution and shade. Fruits of mycorrhizal plants showed a 2.0%
increase in lightness and 14.3, 12.9, 13.9 and 21.2% decreases
in red/greenness, blue/yellowness, chromaticity and absorbance
respectively compared with fruits of non-mycorrhizal plants. Increasing
N concentration in the irrigation solution did not lead
to statistically significant differences in colour parameters between
fruits within each mycorrhizal treatment. Nor were there
significant differences between fruits of mycorrhizal and nonmycorrhizal
plants fertilised with the same N concentration
(Table 2).
Titratable acidity, glucose, fructose and Brix grade were lowest
in fruits of plants fertilised with 3 mmol L −1 N (Table 3). Their
titratable acidity, glucose and fructose values were significantly
lower than those in fruits of plants treated with 6 mmol L −1 N,
while their Brix grade was significantly lower than that in fruits of
plants treated with 18 mmol L −1 N. Mycorrhization modified only
fructose concentration, with fruits of mycorrhizal plants containing
8.5% less fructose than those of non-mycorrhizal plants. When the
applied N was increased, a significant difference in titratable acidity
between mycorrhizal and non-mycorrhizal plants treated with the
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1774–1782

Effects of AMF and N level on quality of strawberry fruits www.soci.org
Table 2. Lightness (L ∗ ), red/greenness (a ∗ ), blue/yellowness (b ∗ ), colour evolution (a ∗ /b ∗ ), shade (tan −1 (b ∗ /a ∗ )), chromaticity (C ∗ = (a 2 + b 2 ) 1/2 )
and absorbance at 500 nm of fruits of strawberry plants inoculated with Glomus intraradices and fertilised with different nitrogen concentrations in
irrigation water
Factor Lightness Red/greenness Blue/yellowness Colour evolution Shade Chromaticity Absorbance
Nitrogen concentration (mmol L −1 )
3 84.93a 27.69a 17.16a 1.62a 31.77a 32.59a 0.54b
6 83.31b 31.07a 19.79a 1.57a 32.47a 36.85a 0.64a
18 84.31a 29.50a 18.18a 1.63a 31.67a 34.67a 0.57ab
Inoculation
M 84.99a 27.45b 17.26b 1.59a 32.21a 32.44b 0.52b
NM 83.37b 31.38a 19.49a 1.62a 31.72a 36.95a 0.63a
Interaction (nitrogen concentration × inoculation)
3 × M 85.72a 25.98a 16.52a 1.57a 32.47a 30.80a 0.48b
3 × NM 84.14ab 29.40a 17.80a 1.66a 31.07a 34.38a 0.59ab
6 × M 83.80ab 29.40a 18.68a 1.57a 32.56a 34.85a 0.60ab
6 × NM 82.82b 32.74a 20.90a 1.56a 32.38a 38.85a 0.68a
18 × M 85.46ac 26.98a 16.58a 1.63a 31.59a 31.69a 0.50b
18 × NM 83.16bc 32.02a 19.78a 1.62a 31.72a 37.65a 0.63ab
Each value represents the mean of four replicates. Two-way ANOVA was applied for each parameter; when statistical differences were found, a Tukey
test (P < 0.05) was conducted independently for nitrogen concentration (3, 6 and 18 mmol L −1 ), inoculation (mycorrhizal (M) and non-mycorrhizal
(NM)) and nitrogen concentration × inoculation (3 × M, 3 × NM, 6 × M, 6 × NM, 18 × Mand18× NM). For each factor, means with different letters
in a column differ significantly.
Table 3. Titritable acidity, soluble sugar concentrations and Brix
grade of fruits of strawberry plants inoculated with Glomus intraradices
and fertilised with different nitrogen concentrations in irrigation water
Titratable Soluble sugars (g kg −1 DM)
acidity (%
Factor citric acid) Glucose Fructose Sucrose Brix grade
Nitrogen concentration (mmol L −1 )
3 1.28b 136.63b 148.96b 94.16a 4.93b
6 1.36a 153.21a 167.54a 62.68a 5.39ab
18 1.30b 140.50ab 149.35b 83.07a 6.09a
Inoculation
M 1.30a 139.85a 148.99b 80.45a 5.96a
NM 1.33a 147.04a 161.58a 79.45a 5.99a
Interaction (nitrogen concentration × inoculation)
3 × M 1.21c 132.02c 141.96c 113.10a 5.05bc
3 × NM 1.35ab 141.24bc 155.96bc 75.21a 4.81c
6 × M 1.38a 146.34bc 158.85bc 45.83a 5.30abc
6 × NM 1.34ab 160.07ab 176.24ab 79.54a 5.48abc
18 × M 1.31ab 141.20bc 146.16c 82.43a 6.17a
18 × NM 1.28bc 139.81bc 152.54bc 83.71a 6.00ab
Each value represents the mean of four replicates. Two-way ANOVA was
applied for each parameter; when statistical differences were found,
aTukeytest(P < 0.05) was conducted independently for nitrogen
concentration (3, 6 and 18 mmol L −1 ), inoculation (mycorrhizal (M) and
non-mycorrhizal (NM)) and nitrogen concentration × inoculation (3 ×
M, 3 × NM, 6 × M, 6 × NM, 18 × Mand18× NM). For each factor,
means with different letters in a column differ significantly.
same N concentration was observed only in the treatment with
3 mmol L −1 N.
Some nutrient concentrations were significantly different between
fruits of plants treated with 3 and 18 mmol L −1 N (Table 4).
Fruits from the treatment with 3 mmol L −1 N contained 9.4, 13.3,
61.0 and 48.0% more K, Mg, Fe and Zn respectively and 11.3% less
Ca than fruits of plants fertilised with 18 mmol L −1 N. The Mn concentration
in fruits of plants fertilised with 3 mmol L −1 Nwassignificantly
higher than that in fruits of plants treated with 6 mmol L −1
N. Fruits of mycorrhizal plants had higher K and Cu concentrations
but lower Mn concentration than fruits of non-mycorrhizal plants.
Mycorrhization significantlymodified the Ca, Mg, Fe, Cu, Zn and Mn
concentrations in fruits when the N applied was changed from 3
to 18 mmol L −1 , and the K concentration in fruits when N changed
from 3 to 6 mmol L −1 . With the exception of Ca, the concentrations
of all elements studied were higher in fruits of plants fertilised with
3 mmol L −1 N. Significantdifferences between fruits of mycorrhizal
and non-mycorrhizal plants of the same N treatment were found
for Cu, Zn and Mn concentrations. Fruits of mycorrhizal plants had
38.0 and 39.3% more Cu and Zn respectively in the 6 mmol L −1
N treatment and 39.6% less Mn in the 18 mmol L −1 N treatment
than their non-mycorrhizal counterparts.
Nitrogen treatment significantly affected the concentrations
of total phenols, gallic acid, ferulic acid, ellagic acid, cyanidin-
3-glucoside, quercetin and kaempferol in fruits (Table 5). Fruits
of plants fertilised with 3 mmol L −1 N had 20.5, 31.2 and 11.4%
lower concentrations of total phenols, gallic acid and cyanidin-
3-glucoside respectively and 21.0, 50.0 and 61.5% higher concentrations
of ellagic acid, quercetin and kaempferol respectively
than fruits of plants treated with 18 mmol L −1 N. Fruits of plants
fertilised with 6 mmol L −1 N had a significantly higher concentration
of ferulic acid than fruits of plants treated with 3
and 18 mmol L −1 N. Mycorrhization significantly modified the
concentrations of all phenolic compounds except pelargonidin-
3-glucoside and catechin. Fruits of mycorrhizal plants had 20.0,
15.0, 50.0 and 28.6% higher concentrations of p-coumaric acid,
cyanidin-3-glucoside, quercetin and kaempferol respectively and
29.0, 50.0 and 11.0% lower concentrations of gallic acid, ferulic
acid and ellagic acid respectively than fruits of non-mycorrhizal
plants.
Fruits of mycorrhizal plants fertilised with 6 mmol L −1 Nhad
a lower gallic acid concentration than fruits of mycorrhizal plants
J Sci Food Agric 2010; 90: 1774–1782 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1777

1778
www.soci.org V Castellanos-Morales et al.
Table 4. Macro- and micronutrient concentrations of fruits of strawberry plants inoculated with Glomus intraradices and fertilised with different
nitrogen concentrations in irrigation water
Macronutrients (g kg −1 DM) Micronutrients (mg kg −1 DM)
Factor K Na Ca Mg Cu Fe Zn Mn
Nitrogen concentration (mmol L −1 )
3 190.94a 2.09a 15.09b 15.04a 2.5a 6.6a 11.1a 9.9a
6 174.63b 1.69a 16.25ab 14.03ab 2.5a 4.6b 10.1a 7.1b
18 174.55b 2.24a 16.79a 13.28b 2.7a 4.1b 7.5b 9.3a
Inoculation
M 185.53a 1.93a 15.86a 13.75a 2.8a 4.8a 9.8a 7.7b
NM 177.38b 2.09a 16.24a 14.49a 2.3b 5.4a 9.3a 9.9a
Interaction (nitrogen concentration × inoculation)
3 × M 194.90a 1.20a 14.14b 14.65a 3.1a 6.3ab 10.7abc 9.2b
3 × NM 185.49ab 2.19a 16.05ab 15.43a 1.8b 6.9a 11.5ab 10.6ab
6 × M 178.33b 1.49a 16.98a 14.03ac 2.9ad 5.2abc 11.7a 6.8c
6 × NM 170.40b 1.89a 15.52ab 14.04ac 2.1be 4.0bc 8.4bcd 7.4c
18 × M 183.38ab 2.30a 16.44a 12.56bc 2.5cde 2.9c 7.0d 7.0c
18 × NM 175.72b 2.19a 17.15a 13.99ac 2.9ac 5.3abc 8.0cd 11.6a
Each value represents the mean of four replicates. Two-way ANOVA was applied for each parameter; when statistical differences were found, a Tukey
test (P < 0.05) was conducted independently for nitrogen concentration (3, 6 and 18 mmol L −1 ), inoculation (mycorrhizal (M) and non-mycorrhizal
(NM)) and nitrogen concentration × inoculation (3 × M, 3 × NM, 6 × M, 6 × NM, 18 × Mand18× NM). For each factor, means with different letters
in a column differ significantly.
Table 5. Total phenol and phenolic compound concentrations of fruits of strawberry plants inoculated with Glomus intraradices and fertilised with
different nitrogen concentrations in irrigation water
Phenolic compounds (mg kg −1 DM)
Flavonoids
Phenolic acids Anthocyanins a Flavonols
Factor Total phenols (g kg −1 DM) Gallic p-Coumaric Ferulic Ellagic Cya-3-glu Pel-3-glu Quercetin Kaempferol Catechin
Nitrogen concentration (mmol L −1 )
3 541.63b 11b 99a 1b 753a 248b 3370a 3a 21a 249a
6 503.32b 10b 96a 2a 699ab 275a 3710a 2b 15b 219a
18 682.04a 16a 100a 1b 622b 280a 3545a 2b 13b 368a
Inoculation
M 571.31a 10b 107a 1b 681b 287a 3692a 3a 18a 322a
NM 580.01a 14a 89b 2a 765a 249b 3391a 2b 14b 236a
Interaction (nitrogen concentration × inoculation)
3 × M 590.38ab 11bc 109ab 1b 633bc 258bc 3564a 4a 21a 324a
3 × NM 492.87b 11bc 89ab 1b 873a 238b 3176a 3ab 21a 174a
6 × M 506.74b 6c 113a 1b 595c 314a 4198a 3ab 20a 233a
6 × NM 499.89b 14ab 78b 2a 803ab 236b 3222a 2c 10b 206a
18 × M 616.79ab 13ab 99ab 1b 626bc 288ac 3314a 3ac 13ab 408a
18 × NM 747.29a 18a 101ab 1b 618c 272ab 3776a 2bc 12ab 328a
Each value represents the mean of four replicates. Two-way ANOVA was applied for each parameter; when statistical differences were found, a Tukey
test (P < 0.05) was conducted independently for nitrogen concentration (3, 6 and 18 mmol L −1 ), inoculation (mycorrhizal (M) and non-mycorrhizal
(NM)) and nitrogen concentration × inoculation (3 × M, 3 × NM, 6 × M, 6 × NM, 18 × Mand18× NM). For each factor, means with different letters
in a column differ significantly.
a Cya-3-glu, cyanidin-3-glucoside; Pel-3-glu, pelargonidin-3-glucoside.
treated with 18 mmol L −1 N and a higher cyanidin-3-glucoside
concentration than fruits of mycorrhizal plants treated with
3 mmol L −1 N, the difference being significant in both cases.
Significant differences between fruits of mycorrhizal and nonmycorrhizal
plants of the same N treatment were found for
all phenolic compounds except pelargonidin-3-glucoside and
catechin. Fruits of mycorrhizal plants had higher p-coumaric acid,
cyanidin-3-glucoside, quercetin and kaempferol concentrations
and lower gallic acid, ferulic acid and ellagic acid concentrations
than fruits of non-mycorrhizal plants when fertilised with
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1774–1782

Effects of AMF and N level on quality of strawberry fruits www.soci.org
6 mmol L −1 N, and a lower ellagic acid concentration when
fertilised with 3 mmol L −1 N (Table 5).
DISCUSSION
In fruit production, parameters such as fruit fresh weight, diameter
and length are important for fruit quality. In a study of the effect of
N application on peaches, Crisosto et al. 39 observed that different
N levels did not affect the size of peach fruits, 39 indicating that the
N levels tested were not a determining factor for this parameter.
In our study, neither N treatment nor mycorrhization had an effect
on these parameters of strawberry fruits.
Colour is another important determinant of fruit quality. Shade
and chromaticity are two parameters used to quantify purity,
while red intensity is used for the description of colour. 40,41 The
values of these variables determined in the present study are
within the ranges observed previously in strawberry fruits. 42 In
our experiment, some colour parameters were modified by N
treatment. Fruits of plants fertilised with 6 mmol L −1 Nhadlower
lightness and higher absorbance at 500 nm than fruits of plants
treated with 3 mmol L −1 N.
Changes in chromaticity due to mycorrhization have been reported
previously in Capsicum annuum L. by Mena-Violante et al., 43
who found that fruits of mycorrhizal plants had a lower chromaticity
than fruits of non-mycorrhizal plants. Interestingly, our study
showed a similar effect of mycorrhization on chromaticity, with
a lower chromaticity being found in fruits of mycorrhizal plants
than in fruits of non-mycorrhizal plants. These data suggest that
the effect of mycorrhization on chromaticity is a general one and
not fruit-specific. It has been proposed that the colour of strawberry
fruits is closely linked with the synthesis and/or expression
of pelargonidin-3-glucoside and cyanidin-3-glucoside, two principal
anthocyanins. 44 In our context, this means that the colour
changes we observed as a result of mycorrhization are possibly
due to changes in the levels of these two anthocyanins.
The flavour of strawberry fruits is determined by the balance
of sugars and acids. 12 Glucose, fructose and sucrose are the most
important sugars for the sensory quality of strawberry fruits,
representing 99% of the total carbohydrate content. 45 Moreover,
citric acid and malic acid are the most important acids in strawberry
fruits. 46 Besides their impact on flavour, acids are important
because they affect the gelling properties of pectin. Brix grade
is a composite parameter reflecting sugars, acids, salts and others
compounds soluble in water and is measured as the total soluble
solids present in the fruit.
In our study, titratable acidity and Brix grade varied between
1.21 and 1.38% and between 4.81 and 6.17 respectively in all
treatments. These values are wthin the ranges reported by Perkins-
Veazie and Collins 47 for titratable acidity (0.5–1.87%) and Brix
grade (5–12). Glucose and fructose concentrations were higher
than sucrose concentration in all cases and the fructose/glucose
ratio was about 1 : 1, in agreement with values reported previously
for strawberry fruits. 42
The level of applied N had a significant effect on titratable
acidity, glucose and fructose concentrations and Brix grade. Fruits
of plants fertilised with 6 mmol L −1 N were more acidic and their
glucose and fructose concentrations were higher in comparison
with fruits from other treatments, without significant differences
in Brix grade. These data indicate that fruits from the 6 mmol L −1
N treatment had the best quality according to Mitcham. 48 In our
experiment, foliar area was also measured (data not shown). The
higher production of sugars in the 6 mmol L −1 N treatment could
be explained by the enhanced foliar area of these plants (35.9
and 25.3% higher than that of plants in the 3 and 18 mmol L −1 N
treatments respectively) when fructification started.
Fruits of mycorrhizal plants had a lower fructose concentration
than fruits of non-mycorrhizal plants, indicating that mycorrhization
reduced the accumulation of this carbon compound in the
fruits. This could be explained by the fact that AMF act as carbon
sinks (4–20% of the total carbon fixed by the plant). 49
Around 4% of the dry matter of plants comprises mineral
elements, which, owing to their role in enzymatic reactions
essential for fruit development and its cold conservation, are
importantforfruitquality. 50 Inthisstudyweobservedthatdifferent
N levels modified the concentrations of some minerals in the fruits.
Fruits of plants fertilised with 3 mmol L −1 N showed higher K, Mg,
Fe and Zn levels than fruits of plants treated with 18 mmol L −1
N. These results suggest that the roots of plants fertilised with
3 mmol L −1 N took up higher amounts of these minerals.
It has been shown previously that a low availability of N in the
soil affects root growth. Tolley-Henry and Raper 51 suggested that
under conditions of low N availability the roots have priority to
N compared with other plant organs and therefore root growth
is promoted. Rufty et al. 52 demonstrated that a low availability of
N in the soil increases the amount of photosynthates addressed
to the roots, thus being available for enhanced root growth. In
our experiment, root dry weight and volume were also measured
(data not shown). The root dry weight of plants fertilised with 3
and 18 mmol L −1 N was 2.0 and 1.7 g per plant respectively, while
the root volume of these plants was 22.0 and 14.8 cm 3 per plant
respectively. These data suggest that a higher soil volume was
explored by plants fertilised with 3 mmol L −1 N compared with
plants treated with 18 mmol L −1 N, which could explain the higher
K, Mg, Fe and Zn levels in fruits of plants of the 3 mmol L −1 N
treatment.
To date, no adequate data are available on the effect
of mycorrhization on macro- and micronutrients in fruits.
In our experiment, mycorrhization significantly modified the
concentrations of K, Cu and Mn, with fruits of mycorrhizal plants
having higher K and Cu levels and a lower Mn level. Although
mycorrhizal root colonisation frequently increases macro- and
micronutrient accumulation in the leaves and stalks of plants, 53,54
Liu et al. 55 found lower Cu, Zn, Mn and Fe concentrations in the
shoots of mycorrhizal corn plants. The inconsistent results on
nutrient acquisition by mycorrhizal plants have been attributed to
changes in the rhizosphere due to increased N levels in the soil,
which affect mycorrhizal development. 56
Fruits of mycorrhizal plants fertilised with 3 mmol L −1 Nhad
higher concentrations of those minerals than fruits of mycorrhizal
plants treated with 18 mmol L −1 N. Similar results of lower, equal
or higher acquisition of macro- and/or micronutrients dependent
on the level of mineral fertilisation have been reported in lettuce
inoculated with Glomus mosseae. 4 The lack of a beneficial effect of
mycorrhization in terms of mineral acquisition in our 18 mmol L −1
N treatment could be attributed to a negative effect of this N
concentration on the extraradical mycelium development of the
AMF. The suppressive effect of high N levels on the formation
of extraradical mycelium has been described previously and has
been linked with reduced nutrient acquisition in mycorrhizal
plants. 57 Fruits of mycorrhizal plants fertilised with 6 mmol L −1
N had significantly higher Cu and Zn concentrations than fruits
of non-mycorrhizal plants fertilised with the same N level. This
indicates a positive effect of mycorrhization and N treatment on
J Sci Food Agric 2010; 90: 1774–1782 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1779

1780
the acquisition of Cu and Zn by the roots and their translocation
to the fruits.
Our results on the effect of N fertilisation on phenolic
compounds in strawberry fruits show that from the 3 mmol L −1
N treatment to the 18 mmol L −1 N treatment the concentrations
of ellagic acid, quercetin and kaempferol decreased while the
concentrations of total phenols, cyanidin-3-glucoside and gallic
acid increased.
A decrease in quercetin and kaempferol concentrations at high
N levels has been reported in tomato fruits, 58 while an increase
in ellagic acid concentration at low N levels has been found in
strawberry fruits. 59 In addition, Keller and Hrazdina 60 reported that
the N concentration in the soil had different effects on the total
phenol concentration in grapes. The application of high N levels
led to low accumulation of flavonols, whereas the proportion of
anthocyanins was similar to that at low N levels. Our results could
be explained by the effect that N has on the biosynthetic pathways
of phenolic compounds. Phenylalanine ammonia-lyase (PAL) is the
principal enzyme of the phenylpropanoid pathway. 61 This enzyme
catalyses the transformation of the amino acid L-phenylalanine by
deamination to trans-cinnamic acid, which is the first product
necessary for the synthesis of phenolic compounds. Interestingly,
it has been reported that at low N levels the enzymatic activity of
PAL is increased, liberating N for the amino acid metabolism, and
whereas the carbon products are diverted via 4-coumaroyl-CoA
into the flavonoid biosynthetic pathway. 62 In our study, this could
be an explanation for the increase in concentrations of some
flavonoids (cyanidin-3-glucoside, kaempferol and quercetin) in
fruits of plants fertilised with 3 mmol L −1 N.
Mycorrhization modified the levels of most phenolic compounds.
The cyanidin-3-glucoside, p-coumaric acid, quercetin and
kaempferol concentrations were higher and the gallic acid, ferulic
acid and ellagic acid concentrations were lower in fruits of
mycorrhizal plants than in fruits of non-mycorrhizal plants. To our
knowledge, there are no data on the effect of AMF on phenolic
compound accumulation in fruits. However, there are reports on
changes in the levels of p-coumaric acid and ferulic acid in the roots
of mycorrhizal onion plants, 63 changes in the levels of biochanin A,
formononetin, genistein and daidzein in the roots of mycorrhizal
alfalfa (M. sativa L.) 19–21 and barrel medic (M. truncatula) 22 and
changes in the level of glyceoline in mycorrhizal soybean (G. max
L.). 24 Most recently, it has been shown that through mycorrhization
the levels of phenols can also be altered in plant shoots. 23 Our
results extend these observations, showing that mycorrhization
can induce changes in phenolic compound levels even in fruits.
An increase in applied N modified the concentrations of
some phenolic compounds between fruits of mycorrhizal and
non-mycorrhizal plants. Differences were determined in fruits
of plants fertilised with 6 mmol L −1 N. These results indicate
that N fertilisation modifies the response of the strawberry
plant to the AMF G. intraradices. This could be attributed to
changes in the rhizosphere due to N levels in the soil, which
affect mycorrhizal development 56 and thus the acquisition of
other nutrients necessary for the production of phenols. To our
knowledge, we have provided the first evidence that, depending
on the N level applied, the accumulation of phenolic compounds
is altered in fruits of mycorrhizal strawberry plants.
CONCLUSION
Mycorrhization did not modify the weight, diameter or length
of strawberry fruits but had a negative effect on most colour
www.soci.org V Castellanos-Morales et al.
parameters. Moreover, fruits of mycorrhizal plants had higher K
and Cu concentrations and showed greater accumulation of most
phenolic compounds. The results indicate that the 3 mmol L −1
N treatment had a positive effect on the accumulation of some
minerals in strawberry fruits, and fruits of mycorrhizal plants had
significantlyhigherphenoliccompound,CuandZnconcentrations
than fruits of non-mycorrhizal plants when they were treated with
6 mmol L −1 N. In recent years, much interest has focused on the
intake of phenolic compounds from the human diet and the
health benefits due to their antioxidant nature. It is therefore of
interest to produce crops rich in flavonols without the need for
genetic modification. Although previous studies have identified
a link between nutrient deficiency and phenolic compound
accumulation in plant tissue, the present study provides evidence
that mycorrhization and N application in strawberry plants can be
one strategy for increasing phenolic compound concentrations in
the fruits. In addition, up-regulation of the flavonoid biosynthetic
pathway in strawberry fruits may afford protection against
pathogen attack or light-induced damage. Further studies are
required to test this theory.
ACKNOWLEDGEMENTS
The authors are grateful to ‘Fondos Mixtos CONACyT – Gobierno
del Estado de Michoacán’, Mexico for support of project 12268
‘Optimization of nitrogen and water in the strawberry crop
(Fragaria × ananassa Duch.) by the use of arbuscular mycorrhizal
fungi’ and to CONACYT for provision of a PhD grant to Vilma
Castellanos. Moreover, we thank Dr Philippe Lobit, Sandra and
Silvia Velasco López, Flor Lorena Reyes Sánchez and Alejandrino
López Hernández from UMSNH, Morelia, Michoacán, Mexico and
Veronica Schober, Monika Marek and Karin Korntheuer from the
Chemistry Laboratory of the Federal College and Research Institute
for Viticulture and Pomology, Klosterneuburg, Austria for their
support.
REFERENCES
1 SmithSEandReadDJ,Mycorrhizas Symbiosis. Academic Press, London
(1997).
2 Bago B, Vierheilig H, PichéY and Azcón-Aguilar C, Nitrate depletion
and pH changes induced by the extraradical mycelium of the
arbuscular mycorrhizal fungus Glomus intraradices. New Phytol
133:273–280 (1996).
3 Govindarajulu M, Pfeffer PE, Jin H, Abubaker J, Douds DD, Allen JW,
et al, Nitrogen transfer in the arbuscular mycorrhizal symbiosis.
Nature 1038:1–5 (2005).
4 Azcón R, Ambrosano E and Charest C, Nutrient acquisition in
mycorrhizal lettuce plants under different phosphorus and nitrogen
concentration. Plant Sci 165:1137–1145 (2003).
5 Johansen A, Jakobsen I and Jensen ES, Hyphal transport by a vesiculararbuscular
mycorrhizal fungus of N applied to the soil as ammonium
or nitrate. Biol Fertil Soils 16:66–70 (1993).
6 Corkidi L, Rowland DL, Jonson NC and Allen EJ, Nitrogen fertilization
alters the functioning of arbuscular mycorrhizas at two semiarid
grasslands. Plant Soil 240:299–310 (2002).
7 HopkinsW, Introduction to Plant Physiology. Wiley, Chichester
(1995).
8 Yoshida Y, Goto T, Hirai M and Masuda M, Anthocyanin accumulation
in strawberry fruits as affected by nitrogen nutrition. Acta Hort
567:357–360 (2002).
9 May G and Pritts M, Strawberry nutrition. Adv Strawberry Prod 9:10–23
(1990).
10 Miner GS, Poling EB, Caroll DE, Nelson LA and Campbell CR, Influence
of all nitrogen and spring nitrogen–potassium applications on
yield and fruit quality of ‘Chandler’ strawberry. JAmSocHortSci
122:290–295 (1997).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1774–1782

Effects of AMF and N level on quality of strawberry fruits www.soci.org
11 Wang SY, Bunce JA and Maasl JL, Elevated carbon dioxide increases
concentrations of antioxidant compounds in field grown
strawberries. J Agric Food Chem 51:4315–4320 (2003).
12 Kallio H, Hakala M, Pelkkikangas A and Lapvetelainen A, Sugars and
acids of strawberry varieties. Eur Food Res Technol 212:81–85
(2000).
13 Kahkonen MP, Hopia AI and Heinonen M, Berry phenolics and
their antioxidant activity. J Agric Food Chem 49:4076–4082
(2001).
14 Khaosaad T, Vierheilig H, Nell N, Zitterl-Eglseer K and Novak J,
Arbuscular mycorrhiza alter the concentration of essential oils
in oregano (Origanum sp., Lamiaceae). Mycorrhiza 16:443–446
(2006).
15 Copetta A, Lingua G and Berta G, Effects of three AM fungi on
growth, distribution of glandular hairs, and essential oil production
in Ocimum basilicum L. var. Genovese. Mycorrhiza 16:485–494
(2006).
16 Toussaint JP, Kraml M, Nell M, Smith SE, Smith FA, Steinkellner S, et al,
The arbuscular mycorrhizal fungus Glomus mosseae confers a
bioprotective effect against Fusarium oxysporum f.sp. basilici in
basil, which is not mediated by increases in phytochemicals. Plant
Pathol 57:1109–1116 (2008).
17 Kapoor R, Giri B and Mukerji KG, Effect of the vesicular arbuscular
mycorrhizal (VAM) fungus Glomus fasciculatum on the essential oil
yield related characters and nutrient acquisition in the crops of
different cultivars of menthol mint (Mentha arvensis) underfield
conditions. Bioresour Technol 81:77–79 (2002).
18 Kapoor R, Giri B and Mukerji KG, Mycorrhization of coriander
(Coriandrum sativum L.) to enhance the concentration and quality
of essential oil. J Sci Food Agric 82:339–342 (2002).
19 Volpin H, Elkind Y, Okon Y and Kapulnik Y, A vesicular arbuscular
mycorrhizal fungus Glomus intraradix induces a defense response
in alfalfa roots. Plant Physiol 104:683–689 (1994).
20 Larose G, Chenevert R, Moutoglis P, Gagne S, Piché Y and Vierheilig H,
Flavonoid levels in roots of Medicago sativa are modulated by the
developmental stage of the symbiosis and the root colonizing
arbuscular mycorrhizal fungus. J Plant Physiol 159:1329–1339
(2002).
21 Catford JG,Staehelin C,Larose G,Piché YandVierheilig H,Systemically
suppressed isoflavonoids and their stimulating effects on
nodulation and mycorrhization in alfalfa split-root systems. Plant
Soil 285:257–266 (2006).
22 Harrison M and Dixon RA, Isoflavonoid accumulation and expression
of defense gene transcripts during the establishment of vesiculararbuscular
mycorrhizal associations in roots of Medicago truncatula.
Mol Plant Microbe Interact 6:643–654 (1993).
23 Khaosaad T, Krenn L, Medjakovic S, Ranner A, Lössl A, Nell M, et al,
Effect of mycorrhization on the isoflavone concentration and the
phytoestrogen activity of red clover. J Plant Physiol 165:1161–1167
(2008).
24 Morandi D, Occurrence of phytoalexins and phenolic compounds on
endomycorrhizal interactions, and their potential role in biological
control. Plant Soil 185:241–251 (1996).
25 Niemi M and Vestberg M, Inoculation of commercially grown
strawberry with VA mycorrhizal fungi. Plant Soil 144:133–142
(1992).
26 Alarcón A, Ferrera-Cerrato R, González-Chavez MC and Villegas-
Monter A, Hongos micorricicos arbusculares en la dinámica de
aparición de estolones y nutrición de plantas de fresa Cv. Obtenidas
por cultivo in vitro. Terra 18:211–218 (2000).
27 Borkowska B, Growth and photosynthetic activity of micropropagated
strawberry plants inoculated with endomycorrhizal fungi (AMF)
and growing under drought stress. Acta Physiol Plant 24:365–370
(2002).
28 Vestberg M, Konen S, Neuvonen EL and Uosukainen M, Mycorrhizal
inoculation of micropropagated strawberry – case studies on
mineral soil and a mined peat bog. Acta Hort 530:297–304
(2000).
29 Vierheilig H, Coughlan AP, Wyss U and Piché Y, Ink and vinegar, a
simple staining technique for arbuscular-mycorrhizal fungi. Appl
Environ Microbiol 12:5004–5007 (1998).
30 An ZQ and Hendrix JW, Determining viability of endogonaceous
spores with a vital stain. Mycologia 80:259–261 (1988).
31 Giovannetti MandMosse B,Anevaluationoftechniquesformeasuring
vesicular-arbuscular mycorrhizal infection on roots. New Phytol
84:489–500 (1980).
32 Gomez L, Rubio E and Augé M, A new procedure for extraction and
measurement of soluble sugars in ligneous plants. J Sci Food Agric
82:360–369 (2002).
33 Gomez L, Bancel D, Rubio E and Varcambre G, The microplate reader:
an efficient tool for the separate enzymatic analysis of sugars
in plant tissues – validation of a micro-method. J Sci Food Agric
87:1893–1905 (2007).
34 Markakis P, Anthocyanins as Food Colors. Academic Press, New York,
NY (1982).
35 Folin C and Ciocalteu V, Tyrosine and tryptophan determination in
proteins. JBiolChem73:627–650 (1927).
36 Sanisavljevic M, Sreckovic M and Miltrovic M, Field performance of
some foreign strawberry cultivars grown in Yugoslavia. Acta Hort
439:291–296 (1997).
37 Tsavnova-Savova S and Ribarova F, Free and conjugated myricetin,
quercetin and kaempferol in Bulgarian red wines. J Food Compos
Anal 15:639–645 (2002).
38 Vrhovˇsek U, Wendelin S and Eder R, Quantitative Bestimmung von
Hydroxyzimtsäurederivaten (Hydroxycinnamaten) in Weißweinen
mittels HPLC. Mitteilungen Klosterneuburg 47:164–172
(1997).
39 Crisosto CH, Johnson RS and DeJong T, Orchard factors affecting
postharvest stone fruit quality. HortScience 32:820–823
(1997).
40 Wang S and Camp M, Temperatures after bloom affect plant
growth and fruit quality of strawberry. Sci Hort 85:183–199
(2000).
41 Sacks E and Shaw D, Color change in fresh strawberry fruit of seven
genotypes stored at 0 ◦ C. HortScience 28:209–210 (1993).
42 Pelayo-Saldívar C, Ebeler ES and Kader AA, Cultivar and harvest
date effects on flavor and other quality attributes of California
strawberries. J Food Qual 28:78–97 (2005).
43 Mena-Violante HG, Ocampo-Jimenez O, Dendooven L, Martínez-
Soto G, González-Castañeda J, Davies Jr FT, et al, Arbuscular
mycorrhizal fungi enhance fruit growth and quality of chile ancho
(Capsicum annuum L. cv San Luis) plants exposed to drought.
Mycorrhiza 16:261–267 (2006).
44 Gil MI,Holcroft DM and Kader AA,Changes in strawberry anthocyanins
and other polyphenols in response to carbon dioxide treatments.
J Agric Food Chem 45:1662–1667 (1997).
45 Pérez A, OlíasR,EspadaJ,Olías J and Sanz C, Rapid determination of
sugars, nonvolatile acids, and ascorbic acid in strawberry and other
fruits. J Agric Food Chem 45:3545–3549 (1997).
46 Cordenunsi B, Oliveira J, Genovesse M and Lajolo F, Influence of
cultivar on quality parameters and chemical composition of
strawberry fruits grown in Brazil. J Agric Food Chem 50:2581–2586
(2002).
47 Perkins-Veazie P and Collins JK, Strawberry fruit quality and its
maintenance in postharvest environments. Adv Strawberry Res
14:1–8 (1995).
48 Mitcham B,Quality assurance forstrawberries: a case study. Perishables
Handling Newslett 85:6–8 (1996).
49 Douds DD, Pffefer PE and Shachar-Hill Y, Carbon partitioning, cost and
metabolism of arbuscular mycorrhizas: physiology and function, in
Arbuscular Mycorrhizas: Physiology and Function, ed. by Kapulnick Y
and Douds DDJ. Kluwer Academic, Dordrecht, pp. 107–129
(2000).
50 Fallahi E, Conway W, Hickey K and Sams C, The role of calcium and
nitrogen in postharvest quality and disease resistance of apples.
HortScience 32:831–835 (1997).
51 Tolley-Henry L and Raper Jr CD, Expansion and photosynthetic rate
of leaves of soybean plants during onset of and recovery from
nitrogen stress. Bot Gaz 147:400–406 (1986).
52 Rufty Jr TW, Huber SC and Volk RJ, Alterations in leaf carbohydrate
metabolism in response to nitrogen stress. Plant Physiol 88:725–730
(1988).
53 Kaya C, Higgs D, Kirnak H and Tas I, Mycorrhizal colonization improves
fruit yield and water use efficiency in watermelon (Citrullus lanatus
Thunb.) grown under well-watered and water-stressed conditions.
Plant Soil 253:287–292 (2003).
54 Ghazi N, Al-Karaki-Hammand R and Rusan M, Response of two tomato
cultivars differing in salt tolerance to inoculation with mycorrhizal
fungi under salt stress. Mycorrhiza 11:43–47 (2001).
55 Liu A, Hamel C, Hamilton RI, Ma BL and Smith DL, Acquisition of Cu,
Zn, Mn and Fe by mycorrhizal maize (Zea mays L.) grown in
J Sci Food Agric 2010; 90: 1774–1782 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1781

1782
soil at different P and micronutrient levels. Mycorrhiza 9:331–336
(2000).
56 Clarke RB and Zeto SK, Mineral acquisition by arbuscular mycorrhizal
plants. JPlantNutr23:867–902 (2000).
57 Gryndler M, Hrselova H, Vosatka M, Votruba J and Klir J, Organic
fertilization changes the response of mycelium of arbuscular
mycorrhizal fungi and their sporulation to mineral NPK supply.
Folia Microbiol 46:540–542 (2001).
58 Stewart AJ, Chapman W, Jenkins GI, Graham I, Martin T and Crozier A,
The effect of nitrogen and phosphorus deficiency on flavonol
accumulation in plant tissues. Plant Cell Environ 24:1189–1197
(2001).
59 Anttonen MJ, Hoppuls KI, Nestby MJ and Karjalainen RO, Influence
of fertilization, mulch color, early forcing, fruit order, planning
date, shading, growing environment, and genotype on the
concentrations of selected phenolics in strawberry (Fragaria ×
ananassa Duch.) fruits. J Agric Food Chem 54:2614–2620 (2006).
www.soci.org V Castellanos-Morales et al.
60 Keller M and Hrazdina G, Interaction on nitrogen availability during
bloom and light intensity during variation. II. Effects on anthocyanin
and phenolic development during grape ripening. Am J Enol Vitic
49:341–349 (1998).
61 Hao Z, Charles DJ, Yu L and Simon JE, Purification and characterization
of a phenylalanine ammonia-lyase from Ocimum basilicum.
Phytochemistry 43:735–739 (1996).
62 Margna U, Control at the level of substrate supply – an alternative
in the regulation of phenylpropanoid accumulation in plant cells.
Phytochemistry 16:419–426 (1977).
63 Grandmaison J, Olah GM, Van Calsteren MR and Furlan V,
Characterization and localization of plant phenolics likely involved
in the pathogen resistance expressed by endomycorrhizal roots.
Mycorrhiza 3:155–164 (1993).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1774–1782

Research Article
Received: 28 September 2009 Revised: 6 April 2010 Accepted: 9 April 2010 Published online in Wiley Interscience: 25 May 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4012
Effect of different light transmittance paper
bags on fruit quality and antioxidant capacity
in loquat
Hong-xia Xu, Jun-wei Chen ∗ and Ming Xie
Abstract
BACKGROUND: Bagging has been widely used to improve the commercial value of fruit. The purpose of this study was to
evaluate the effects of different light transmittance paper bags on the quality and antioxidant capacity of loquat fruit. Two
loquat cultivars, Baiyu and Ninghaibai (Eriobotrya japonica Lindl.), were used for materials. One-layered white paper bags
(OWPB) with ∼50% light transmittance and two-layered paper bags with a black inner layer and a grey outer layer (TGDPB)
with ∼0% light transmittance were used as treatments and unbagged fruits were used as the control (CK) in this experiment.
Fruit quality was determined by physicochemical characteristics, the quantity of sugar, total phenolic, flavonoid, carotenoid
and vitamin C. The antioxidant capacities of the methanol extracted from the pulp were tested using three different assays.
RESULTS: The results showed that bagging decreased the weight of fruit but promoted the appearance of loquat fruits. The
total sugar content in the fruit bagged with OWPB was higher than in controls and in fruit bagged with TGDPB. The total
phenolic and flavonoid contents were decreased by both bagging treatments, with the lowest occurring in the fruit bagged
with TGDPB. Bagging also decreased the total antioxidant capacity of the fruit pulp, which was again lower in TGDPB-treated
fruits than in those bagged using OWPB. Correlation analysis showed a linear relationship between total antioxidant capacity
and the content of total phenolic and flavonoid.
CONCLUSION: The results showed that different light transmittance bags had different effects on fruit quality and antioxidant
capacity. In particular, bags with low light transmittance (TGDPB) decreased the inner quality and total antioxidant capacity of
loquat fruit. All results indicated that bagging with OWPB was more suitable for maintaining the quality of the loquat fruit than
bagging with TGDPB.
c○ 2010 Society of Chemical Industry
Keywords: antioxidant capacity; bagging; Eriobotrya japonica Lindl; flavonoid content; loquat; total phenolic content
INTRODUCTION
Loquat (Eriobotrya japonica Lindl.) is widely cultivated in subtropical
regions of Asia and other continents. Ripe loquat fruits are
spherical or oval in shape, orange/yellow or white in colour and
have a soft and juicy flesh. During their growth and maturation,
they are susceptible to insect pests, birds, diseases and mechanical
damage, which reduce their commercial value. Bagging, a physical
protection technique commonly applied to many fruits, not
only improves fruit visual quality, 1–4 by promoting fruit coloration
and reducing the incidences of fruit cracking and russet, but can
also change the microenvironment of fruit development, which
has multiple effects on the inner quality of fruits. Sugars and organic
acids are the major determinants of fruit taste and flavour.
However, varied results have been obtained from experiments on
the effects of bagging on the sugar and organic acid contents of
fruits. Chundawat et al. 5 showed that bagging generally reduces
the sugar content of fruit, whereas Hussein et al. 6 reported that
bagging significantly increased the total sugar content. Huang
et al. 7 reported that bagging treatments did not affect the total
soluble sugar content, but decreased the organic acid contents of
fruit. Kim et al. 8 showed that titratable acids tended to increase
after bagging with yellow paper of low light transmittance.
In addition to sugars and organic acids, loquat fruits also contain
diverse nutrient and non-nutrient molecules, such as phenolics
(especially the flavonoids), vitamin C, and β-carotene, many of
which have antioxidant properties. These compounds exert a
range of biological effects including antibacterial, antiviral, antiinflammatory,
antithrombotic and vasodilatory actions. 9,10 They
also have pronounced antioxidant and free-radical-scavenging
activities. 11–13 However, most of the studies on bagging fruit
focus on the appearance and general qualities of the fruit and
studies of the effects of bagging on antioxidant compounds and
antioxidant capacity of fruits are rare. Therefore, this study was
carried out to examine the effect of different light transmittance
paper bags on fruit quality and antioxidant capacity in two loquat
cultivars.
∗ Correspondence to: Jun-wei Chen, Institute of Horticulture, Zhejiang Academy
of Agricultural Sciences, Hangzhou, Zhejiang, 310021, China.
E-mail: chenjunwe@tom.com
InstituteofHorticulture,ZhejiangAcademy ofAgricultural Sciences,Hangzhou,
Zhejiang, 310021, China
J Sci Food Agric 2010; 90: 1783–1788 www.soci.org c○ 2010 Society of Chemical Industry
1783

1784
MATERIALS AND METHODS
Standards and chemicals
ABTS [2,2-azino-bis (3-ethylbenzthiozoline-6-sulfonic acid)],
DPPH (the 1,1-diphenyl-2- picrylhydrazyl radical), TPTZ
[2,4,6-tri(2-pyridyl)-s-triazine], Trolox (6-hydroxy-2,5,7,8tetramethylchroman-2-carboxylic
acid), DPIP (phenolindo-
2,6- dichlorophenol), rutin, β-carotene, HPLC-grade sucrose,
glucose, fructose and sorbitol were all purchased from Sigma-
Aldrich (Shanghai, China). All reagents were of analytical grade
unless indicated otherwise.
Plant material
Two loquat cultivars, Baiyu and Ninghaibai (Eriobotrya japonica
Lindl.),growninacommercialorchardinQingpuDistrict,Shanghai,
China were used to test the bagging treatment. Approximately
100 fruitlets that were similar in appearance and size and which
received sunlight uniformly were randomly selected for each
bagging treatment. Approximately 100 unbagged fruitlets that
were also similar in appearance and size were tagged as controls
(CK). Bagging treatments were conducted after fruit thinning in
early April. At maturity, 50 fruits of each treatment were used for
fruit quality and antioxidant capacity analyses. Fruit maturity and
ripeness were assessed based on fruit firmness and skin colour.
Two types of bag were used in this study: (1) 25 cm × 36 cm onelayered
white paper bags (OWPB) with ∼50% light transmittance;
(2) 25 cm × 36 cm two-layered paper bags with a black inner layer
and a grey outer layer (TGDPB) with ∼0% light transmittance. All
bags were coated with wax and supplied by Shangyu Jiali Paper
Bag Product Co., Ltd. (Ningbo, China).
Physicochemical characteristics analyses
The fruit mass of 30 loquats from each bagging treatment was
measured by using an electronic balance (0–210 g ± 0.001 g;
model C-600-SX; Cobos, Barcelona, Spain). These fruits were
randomly divided into five groups (replicates) with six fruits in each
group. The fruits were then manually peeled, cut into small pieces
and juiced together. The soluble solids concentration (SSC) was
measured in the filtered juice by using a hand-held refractometer
(Atago, Tokyo, Japan) and calibrated with distilled water. The
juice was also analysed for titratable acidity (TA) by titration
with 0.01 mol L −1 NaOH, using phenolphthalein as an indicator.
Twenty fruits of each treatment were selected for surface colour
determination using a chromameter (ADCI-60-C, Beijing, China)
calibrated with a manufacturer-supplied white calibration plate.
Results were expressed as lightness (L ∗ ), redness (a ∗ ), yellowness
(b ∗ ) and hue angle (hab = tan −1 [(b ∗ )(a ∗ ) −1 ]). The colour reading
was taken fourth atthe equatorial region of each fruitand averaged
to give a value for each fruit. After surface colour determination,
the fruits were manually peeled, cut into small pieces and the
composite fruit samples ranging from 2 to 10 g were weighed and
frozen in liquid nitrogen, then stored at −80 ◦ C until analysis.
Determination of sugar content
The quantity and types of sugar were determined as described
by Chen et al. 14 Soluble sugars were extracted by grinding 5 g
frozen fruit in five volumes (w/v) of methanol : chloroform : water
as 12 : 5:3 (v/v). Extracts were centrifuged at 5000×g for 5 min. The
extraction was performed three times. Water and chloroform were
then added to bring the final methanol : chloroform : water ratio
to 10 : 6:5 and the chloroform layer was removed. The remaining
aqueous–alcohol phase was adjusted to pH 7.0 using 0.1 mol L −1
www.soci.org H-X Xu, J-W Chen, M Xie
NaOH, then dried in a vacuum and redissolved with distilled
water. The sugar in water solution was analysed by using HPLC
(Waters 1525; Waters, Milford, Massachusetts, USA). The column
temperature was 90 ◦ C and 80% acetonitrile was used as an elution
at a flow rate of 1 mL min −1 . Fructose, glucose, sucrose and sorbitol
were identified and quantified by comparing the retention and
integrated peak areas of external standards.
Total carotenoid content analysis
The total carotenoid content was determined as described by
Reyes et al. 15 Carotenoids were extracted from 2 g frozen fruit by
homogenising with 25 mL of acetone : ethanol (1 : 1) containing
200 mg L −1 butylated hydroxytoluene (BHT). The homogenate
was filtered through a Whatman no. 4 filter, washed with the solvent
(∼60 mL) and diluted to 100 mL using the extraction solvent.
Extracts were transferred to a plastic container to which 50 mL
hexane was added. The container was then shaken and allowed
to stand for 15 min after which 25 mL of nanopure water was
added. The container was shaken again and the contents were allowed
to separate for 30 min. The spectrophotometer was blanked
with hexane and absorbance of the samples in 1-cm quartz cuvettes
was measured at 470 nm. Carotenoid was quantified as
β-carotene using a standard curve for this compound
(1–4 µgmL −1 ). Results were expressed as µg β-carotene equivalent
g −1 fresh weight.
Vitamin C content analysis
The vitamin C content of the fruit extracts was determined
by the 2,6-dichloroindophenol titrimetric method. 16 Briefly, the
samples were mixed with 40 mL of buffer (1 g L −1 oxalic acid plus
4gL −1 anhydrous sodium acetate) and were titrated against
the dye solution containing 295 mg L −1 DPIP (phenolindo-
2,6-dichlorophenol) and 100 mg L −1 sodium bicarbonate. The
standard curve was generated with concentrations of 0.2, 0.4,
0.6, 0.8 and 1 mg of standard L-ascorbic acid (AnalaR; BDH,
Buffalo, New York, USA). The ascorbic acid content in the samples
was determined from the standard curve and the results were
expressed as µg ascorbic acid equivalent g −1 fresh weight.
Extracts for phenolic and antioxidant capacity measurement
To analyse the total phenolic and antioxidant activity, fruit extracts
in methanol were prepared using the method of Swain and
Hillis, 17 with some modifications. A 10 g sample of fruit were
homogenised in 25 mL absolute methanol using a Waring blender.
The homogenates were kept at 4 ◦ C for 12 h and then centrifuged
at 15 000 × g for 20 min. The supernatants were collected, and
extraction of the residue was repeated using the same conditions.
The two supernatants of methanol were combined and divided
into two equal aliquots and then stored at −20 ◦ C until analysis.
The first supernatant was used for the quantitative analysis of
phenolic compounds and the second was used to determine the
antioxidant activity.
Total phenolic content analysis
The Folin–Ciocalteu reagent assay 18 was used to determine the
total phenolic content. A 0.1 mL sample aliquot was mixed with
5mL of 0.2molL −1 Folin–Ciocalteu reagent. The solution was
allowed to stand at 25 ◦ C for 5min before adding 4mL of 15%
(w/v) sodium carbonate solution in distilled water. The absorbance
at765 nm was read after the initial mixing and then for up to 90 min
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1783–1788

Effect of bagging on loquat fruit quality and antioxidant capacity www.soci.org
until it reached a plateau. Gallic acid was used as a standard for
the calibration curve. Results were expressed as µg gallic acid
equivalent g −1 fresh weight.
Total flavonoid content analysis
The flavonoid content was measured using a colorimetric assay
developed byJia et al. 19 Plantextract(2.0 mL)or standard solutions
of rutin (Sigma) were added to a 10 mL volumetric flask. Distilled
water was added to make a volume of 5 mL. At zero time, 0.3 mL
of 5% w/v NaNO2 was added to the flask. After 5 min, 0.6 mL of
10% w/v AlCl3 was added and after 6 min, 2 mL of 1 mol L −1 NaOH
was added to the flask, followed by 2.1 mL distilled water. The
absorbance was read at 510 nm against the blank (water) and the
flavonoid content was expressed as µg rutin equivalent g −1 fresh
weight.
Antioxidant capacity determinations
Free radical scavenging activity on DPPH
The free radical scavenging activity of the extracts, based on
the scavenging activity of the stable 1,1-diphenyl-2-picrylhydrazyl
(DPPH) free radical, was determined by the method described by
Braca et al. 20 Plant extract (0.1 mL) was added to 3 mL of a 0.004%
MeOH solution of DPPH. Absorbance at 517 nm was determined
after 30 min, and the percentage inhibition activity was calculated
from [(A0 − A1)/A0] × 100, where A0 is the absorbance of the
control, and A1 is the absorbance of the extract/standard. Results
were expressed as µmol Trolox equivalent g −1 fresh weight.
Antioxidant activity using the ABTS assay
TheABTS • scavengingabilityofextractswasdeterminedaccording
to the method described by Re et al. 21 ABTS • was generated by
reacting an ABTS aqueous solution (7 mmol L −1 )withK2S2O8
(2.45 mmol L −1 , final concentration) in the dark for 16 h and
adjusting the absorbance at 734 nm to 0.700 with ethanol. A
0.2 mL aliquot of appropriate dilution of the extract was added
to 2.0 mL ABTS • solution and the absorbance was measured
at 734 nm after 15 min. Results were expressed as µmol Trolox
equivalent g −1 fresh weight.
Table 1. Effects of bagging on the physicochemical characteristics of loquat fruit
Ferric reducing/antioxidant power assay
The FRAP assay was used as described by Benzie and Strain 22
with some modifications. The stock solutions included 300 mmol
acetate buffer (3.1 g C2H3NaO2·3H2Oand16mLC2H4O2), pH 3.6;
10 mmol TPTZ (2,4,6-tripyridyl-s-triazine) solution in 40 mmol HCl,
and 20 mmol FeCl3·6H2O solution. The fresh working solution was
prepared by mixing 25 mL acetate buffer, 2.5 mL TPTZ solution,
and 2.5 mL FeCl3·6H2O solution and then warmed at 37 ◦ Cbefore
use 150 µL of fruit extracts or methanol (for the reagent blank) was
reacted with 2850 µLoftheFRAPsolutionat37 ◦ Cfor30mininthe
dark (in a water bath). Readings of the coloured product (ferrous
tripyridyltriazine complex) were then taken at 593 nm. Results
were expressed as µmol Trolox equivalent g −1 fresh weight.
Statistical analysis
The significance of the results and statistical differences were
analysed using SYSTAT version 10.0 (SPSS, Chicago, IL, USA).
Analysis of variance (ANOVA) of the data was performed to
compare mean values for each variable under different cultivars.
Theleastsignificantdifferencetest(LSD)wasusedtodeterminethe
differences between means at a 5% significance level. Correlation
coefficients of DPPH, TEAC and FRAP with respect to total phenolic,
total flavonoid, total carotenoid and vitamin C contents were
evaluated.
RESULTS AND DISCUSSION
The effects of bagging on the physicochemical characteristics
of loquat fruit
Bagging is already known to affect the size and weight of
pomegranate, 6,23 apple 24 and banana. 25 In this study, all bagging
treatments decreased the weight of loquat fruit compared with
controls (Table 1). And, fruits treated with TGDPB were smaller than
that treated with OWPB. The total soluble solids remained constant
and titratable acid decreased in Baiyu fruits treated with OWPB,
whereas total soluble solid significantly decreased and titratable
acid markedly increased in Baiyu fruits treated with TGDPB as
compared with Baiyu controls. However, there was little effect on
the total soluble solids in Ninghaibai fruits bagged with either
Surface colour (n = 20)
Cultivar Treatment Fruit mass (g) (n = 30) SSC (%) (n = 5) TA (%) (n = 5) L ∗ a ∗ b ∗ hab
Baiyu CK 28.1 ± 6.3 a 13.1 ± 1.1 a 0.39 ± 0.00 b 65.1 ± 1.1 b 14.9 ± 1.3 a 44.0 ± 1.8 b 71.3 ± 1.6 b
OWPB 24.2 ± 5.0 b 13.3 ± 0.9 a 0.30 ± 0.03 c 65.5 ± 0.9 b 14.7 ± 1.3 a 45.2 ± 1.3 a 71.9 ± 1.6 b
TGDPB 24.1 ± 6.2 b 11.1 ± 1.6 b 0.55 ± 0.88 a 71.0 ± 1.8 a 10.3 ± 1.9 b 44.7 ± 1.8 ab 77.1 ± 2.5 a
F-value 4.35 ∗ 13.07 ∗∗∗ 56.73 ∗∗∗ 150.36 ∗∗∗ 65.79 ∗∗∗ 3.14 ∗ 61.62 ∗∗∗
LSD0.05 3.20 0.69 0.05 0.73 0.87 0.97 1.09
Ninghaibai CK 29.4 ± 4.5a 14.2 ± 1.0a 0.23 ± 0.04c 65.6 ± 1.7b 14.4 ± 1.2a 42.8 ± 1.8c 71.4 ± 1.8c OWPB 27.3 ± 5.4a 14.1 ± 1.5a 0.32 ± 0.02a 68.7 ± 1.6a 10.6 ± 3.4b 45.7 ± 3.0b 77.1 ± 3.8b TGDPB 24.1 ± 4.0b 14.6 ± 1.4a 0.26 ± 0.18b 69.4 ± 2.0a 8.4 ± 1.9c 47.2 ± 2.4a 80.0 ± 2.2a F-value 6.97∗∗ 0.32 29.67∗∗∗ 32.25∗∗∗ 34.73∗∗∗ 18.66∗∗∗ 53.76∗∗∗ LSD0.05 2.70 1.03 0.02 0.97 1.40 1.41 1.62
Values are expressed as means ± SD. Means within the same cultivar followed by the same superscript letter are not significantly different at P = 0.05.
∗ Significant at P = 0.05.
∗∗ significant at P = 0.01.
∗∗∗ significant at P = 0.001.
CK, control (unbagged); OWPB, one-layered white paper bags; TGDPB, two-layered paper bags with a black inner layer and a grey outer layer.
J Sci Food Agric 2010; 90: 1783–1788 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1785

1786
Table 2. Effects of bagging on sugar content in loquat fruit (mg g −1 )
www.soci.org H-X Xu, J-W Chen, M Xie
Cultivar Treatment Sucrose Glucose Fructose Sorbitol Total sugar
Baiyu CK 2.43 ± 0.04 b 29.0 ± 1.47 b 42.1 ± 2.14 b 2.26 ± 0.09 a 75.8 ± 3.71 b
OWPB 3.00 ± 0.04 a 36.7 ± 0.10 a 47.6 ± 0.40 a 2.59 ± 0.24 a 89.9 ± 0.53 a
TGDPB 2.27 ± 0.10 c 25.4 ± 1.34 c 42.3 ± 2.42 b 1.77 ± 0.10 b 71.8 ± 3.90 b
F-value 91.85 ∗∗∗ 76.17 ∗∗∗ 8.19 ∗ 19.77 ∗∗ 27.87 ∗∗∗
LSD0.05 0.14 2.29 3.76 0.32 6.24
Ninghaibai CK 1.59 ± 0.08b 39.7 ± 2.57a 51.7 ± 3.32a 3.42 ± 0.56a 96.4 ± 6.30a OWPB 2.26 ± 0.10a 40.8 ± 0.99a 52.8 ± 1.67a 2.40 ± 0.10b 98.2 ± 2.31a TGDPB 2.57 ± 0.34a 35.1 ± 1.52b 45.6 ± 0.82b 2.15 ± 0.19b 85.5 ± 2.01b F-value 17.03∗ 8.07∗ 9.27∗ 11.50∗∗ 8.70∗ LSD0.05 0.42 3.63 4.39 0.69 8.08
Values are expressed as means ± SD of three replications. Means within the same cultivar followed by the same superscript letters are not significantly
different at P = 0.05.
∗ Significant at P = 0.05.
∗∗ significant at P = 0.01.
∗∗∗ significant at P = 0.001.
CK, control (unbagged); OWPB, one-layered white paper bags; TGDPB, two-layered paper bags with a black inner layer and a grey outer layer.
OWPB or TGDPB, although the titratable acid content increased
significantly compared with Ninghaibai controls.
Surface colour is an important marketable (consumer acceptance)
quality attribute and is a measure of L ∗ , a ∗ , b ∗ and hab.
Table 1 shows that bagging improved fruit surface lightness as
L ∗ was higher in the bagged than in the control fruits of both
Baiyu and Ninghaibai. Fruit treated with TGDPB had the highest
lightness values. Reduced light is known to promote the degradation
of existing chlorophyll and inhibits carotenoid synthesis in
fruit peel, 26 and this resulted in unbagged fruits having a lower
hab and a more red than yellow hue (i.e. were more orange in
colour compared with bagged fruits). In addition, fruits treated
with TGDPB had higher hab than did those treated with OWPB.
The effects of bagging on sugar content
Sugar content is considered to be an important quality characteristic
of fresh fruit. However, bagging with different materials
can exert different effects on the composition of soluble sugars.
For example, Padmavathamma and Hulamani 23 found that total
sugars varied significantly with bag colour, whereas Yang et al. 27
observed that bagging tended to reduce sugar content slightly,
although the sugar content was not significantly affected by bag
type. Table 2 indicates that the effects of bagging type on sugar
content varied between cultivars. OWPB treatment increased the
sucrose, glucose, fructose and sorbitol content and significantly
increased the content of total sugar in Baiyu fruit. However, OWPB
treatment increased the sucrose content, did not affect the glucose,
fructose and total sugar content but decreased the sorbitol
content in Ninghaibai fruit. Total sugar contents in Baiyu and Ninghaibai
after TGDPB treatment were reduced by 5.6% and 11.3%,
respectively, as compared with the controls.
The effects of bagging on antioxidant compounds
and antioxidant capacity
Numerous studies have shown that fruit and vegetables are
sources of diverse nutrient and non-nutrient molecules, many of
which have antioxidant properties. The present study determined
the antioxidant capacities of loquat fruit and analysed fruit extracts
for compounds (total phenolic, flavonoid, carotenoid and vitamin
C) that might contribute to the antioxidant activity. Table 3 shows
that the total phenolic and flavonoid contents decreased after
bagging treatment. Following OWPB and TGDPB treatment, the
total phenolic content of Baiyu fruit was reduced by 9.5% and
45.6%, respectively, and that of Ninghaibai fruit was reduced by
5.0% and 26%, respectively. This indicates that bagging influences
the metabolism of phenolic compounds, of which the flavonoids
are the dominant family. The pattern of variation in flavonoid
content was similar to that observed for total phenolic, with
maximum levels occurring in unbagged Baiyu (28.2 ± 4.4 µgg −1 )
and Ninghaibai (51.0 ± 6.4 µgg −1 ) fruits. The flavonoid content
was also lower in TGDPB-treated fruits than in OWPB-treated fruit.
Carotenoids and vitamin C are also the antioxidant compounds
in loquat. The study found that the carotenoid and vitamin C
contents increased after bagging fruit with OWBP, but decreased
in fruit bagged with TGDBP. Our other experiments have also
observed that the carotenoid and vitamin C content varies
significantly with bag type; however, both decreased markedly
when light was excluded during the maturation of loquat as
compared with that of control fruit (data not shown).
Three independent methods, the DPPH, TEAC and FRAP assays,
were used to compare the antioxidant capacity of fruit extracts.
The results presented in Table 3 show that the antioxidant
potential of loquat fruit extracts was significantly affected by
light transmittance. The highest antioxidant potential in both of
Baiyu and Ninghaibai loquat fruits was under full sunlight and
lowest under bagging with TGDPB.
Table 4 indicates that the total phenolic content and antioxidant
capacity are well correlated (DPPH, r = 0.64; TEAC, r = 0.77;
FRAP, r = 0.90). Fruits with the highest phenolic content
(unbagged fruits of Baiyu and Ninghaibai) had the highest
antioxidant potentials whereas fruit extracts characterised by
low total phenolic levels exhibited a poor antioxidant capacity.
Numerous studies have reported similar linear relationships
between antioxidant activities and phenolic content. 28–30 A
good correlation was also observed between total flavonoid and
antioxidant capacity (DPPH, r = 0.84; TEAC, r = 0.87; FRAP, r =
0.99) (Table 4). Flavonoids are low-molecular-weight polyphenolic
compounds that are widely distributed in fruit and vegetables, 31
and many have been shown to have antioxidant 32 and anticancer
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1783–1788

Effect of bagging on loquat fruit quality and antioxidant capacity www.soci.org
Table 3. Changes in total phenolic, flavonoid, carotenoid and vitamin C content and antioxidant activities as assessed by DPPH, TEAC and FRAP
assays of loquat fruit after bagging
Cultivar Treatment
Total phenolic
(µg gallic acid
g −1 FW)
Flavonoid (µg
rutin g −1 FW)
Carotenoid (µg
β-carotene g −1 FW)
Vitamin C
(µgascorbic
acid g −1
fresh weigh)
DPPH (µmol
Trolox g −1 FW)
TEAC (µmol
Trolox g −1 FW)
FRAP (µmol
Trolox g −1 FW)
Baiyu CK 220.0 ± 26.9 a 28.2 ± 4.4 a 35.7 ± 4.1 b 13.6 ± 0.9 b 1.85 ± 0.10 a 1.61 ± 0.15 a 2.21 ± 0.09 a
OWPB 199.2 ± 30.0 a 21.6 ± 2.7 b 48.3 ± 3.1 a 15.7 ± 0.6 a 1.54 ± 0.10 b 1.53 ± 0.12 a 1.96 ± 0.04 b
TGDPB 119.7 ± 17.5 b 16.1 ± 1.4 c 27.4 ± 0.7 c 11.5 ± 0.7 c 1.22 ± 0.06 c 0.98 ± 0.04 b 1.65 ± 0.09 c
F-value 59.60 ∗ 39.88 ∗ 63.90 ∗ 68.70 ∗ 143.55 ∗ 75.64 ∗ 110.54 ∗
LSD0.05 21.73 2.77 3.62 0.75 0.08 0.12 0.07
Ninghaibai CK 450.7 ± 31.1a 51.0 ± 6.4a 23.4 ± 0.3b 16.8 ± 1.5b 3.23 ± 0.12a 2.46 ± 0.17a 3.6 ± 0.16a OWPB 427.9 ± 24.7a 43.5 ± 3.7b 30.8 ± 0.3a 18.3 ± 0.5a 2.08 ± 0.06b 1.83 ± 0.05b 3.03 ± 0.11b TGDPB 333.5 ± 20.7b 30.3 ± 1.4c 24.5 ± 0.1b 15.4 ± 1.9c 1.59 ± 0.10c 1.62 ± 0.13c 2.32 ± 0.10c F-value 61.79∗ 71.08∗ 15.58∗ 9.78∗ 746.79∗ 111.49∗ 268.37∗ LSD0.05 23.05 3.54 3.03 1.37 0.09 0.12 0.11
Values are expressed as means ± SD of five replications. Means within the same cultivar followed by the same superscript letter are not significantly
different at P = 0.05.
∗ Values are significant at P = 0.001.
CK, control (unbagged); FW, fresh weight; OWPB, one-layered white paper bags; TGDPB, two-layered paper bags with a black inner layer and a grey
outer layer.
Table 4. Correlation coefficients of DPPH, TEAC and FRAP with
respect to total phenolic, flavonoid, carotenoid and vitamin C content
of loquat fruit
Assay Total phenolic Flavonoid Carotenoid Vitamin C
DPPH 0.64 0.84 ∗ 0.14 0.35
TEAC 0.77 ∗ 0.87 ∗ 0.07 0.56
FRAP 0.90 ∗∗ 0.99 ∗∗ 0.20 0.60
∗ Significant at P = 0.05; ∗∗ significant at P = 0.01; (n = 6).
properties. 33 Bagging resulted in lower irradiation, which has
an important role in the synthesis of phenolic compounds.
Bakhshi and Arakawa 34 reported that light irradiation could
induce phenolic compound biosynthesis in the flesh of apples.
Some studies have observed that enhanced light conditions could
activate the expression of the flavonoid biosynthetic genes. 35–38 In
addition, several studies have reported changes in the qualitative
and quantitative composition of flavonoids as a consequence of
high solar radiation. 39–41 Therefore, reduced light irradiation by
bagging is the probable cause of the decreased total phenolic and
flavonoid contents recorded in both loquat cultivars.
However, the total carotenoid level does not contribute
significantly to the antioxidant potential of loquat, as shown
by the poor correlations between the DPPH, TEAC and FRAP
measurements of antioxidant capacity and total carotenoid
content. Similarly, vitamin C also makes a minor contribution to
antioxidant capacity. Wang et al. 42 reported that the contribution
of vitamin C to the total antioxidant activity of a fruit is

1788
6 Hussein AA, Abdel-Rahman AG and Ahmed RB, Effectiveness of fruit
bagging on yield and fruit quality of pomegranate (Punicagranatum
L.). Ann Agric Sci 32:949–957 (1994).
7 HuangCH,YuB,TengYW,SuJ,ShuQ,ChengZQ,et al,Effectsoffruit
bagging on coloring and related physiology, and qualities of red
Chinese sand pears during fruit maturation. Sci Hort 121:149–158
(2009).
8 Kim YH, Kim SK, Lim SC, Lee CH, Youn CK, Kim HH, et al, Effectsof
bagging material on coloration, maturity, and quality of peach
fruits. J Korean Soc Hort Sci 41:395–400 (2000).
9 Duthie GG, Duthie SJ and Kyle JAM, Plant polyphenols in cancer and
heart disease: implications as nutritional antioxidants. Nutr Res Rev
13:79–106 (2000).
10 Middleton E and Kandaswami C, The impact of plant flavonoids on
mammalian biology: implication for immunity, inflammation and
cancer, in The Flavonoids: Advances in Research since 1986, ed.by
Harborne JB. Chapman and Hall, London, pp. 619–952 (1994).
11 Bahorun T, Trotin F, Pommery J, Vasseur J and Pinkas M, Antioxidant
activities of Crataegus monogyna extracts. Planta Med 60:323–328
(1994).
12 Bahorun T, Gressier B, Trotin F, Brunet C, Dine T, Luyckx M, et al,
Oxygen species scavenging activity of phenolic extracts from
hawthorn fresh plant organs and pharmaceutical preparations.
Arzneimittelforschung 46:1086–1089 (1996).
13 Bahorun T, Trotin F and Vasseur J, Polyphenol production in Crataegus
tissue cultures (hawthorn), in Biotechnology in Agriculture and
Forestry: Medicinal and Aromatic Plants XII, ed.byNagataTand
Ebizuka Y. Springer, Berlin, pp. 23–49 (2002).
14 Chen JW, Zhang SL, Zhang LC, Zhao ZZ and Xu JG, Fruit
photosynthesis and assimilate translocation and partitioning: Their
characteristics and role in sugar accumulation in developing Citrus
unshiu fruit. Acta Botan Sin 44:158–163 (2002).
15 Reyes LF, Villarreal JE and Cisneros-Zevallos L, The increase in
antioxidant capacity after wounding depends on the type of fruit
or vegetable tissue. Food Chem 101:1254–1262 (2007).
16 Association of Official Analytical Chemists, Official Methods of Analysis,
16th edn. AOAC International, Arlington, VA, pp. 16–17 (1995).
17 Swain T and Hillis WE, The phenolic constituents of Prunus domestica,
I – the quantitative analysis of phenolic constituents. JSciFoodAgric
10:63–68 (1959).
18 Singleton VL and Rossi JA, Colorimetry of total phenolics with
phosphomolybdic–phosphotungstic acid reagents. Am J Enol Vitic
16:144–158 (1965).
19 Jia ZS, Tang MC and Wu JM, The determination of flavonoid contents
in mulberry and their scavenging effects on superoxide radicals.
Food Chem 64:555–559 (1999).
20 Braca A, Tommasi ND, Bari LD, Pizza C, Politi M and Morelli I,
Antioxidant principles from Bauhinia terapotensis. J Nat Prod
64:892–895 (2001).
21 Re R, Pellegrini N and Anna PA, Antioxidant activity applying an
improved ABTS radical cation decolorization assay. Free Radical
Biol Med 26:1231–1237 (1999).
22 Benzie IFF and Strain JJ, The ferric reducing ability of plasma (FRAP)
as a measure of ‘‘antioxidant power’’: the FRAP assay. Anal Biochem
239:70–76 (1996).
23 Padmavathamma AS and Hulamani NC, Studies on the effect of fruit
bagging on size, quality and colour development of pomegranate
(Punica granatum C.). J Res ANGRAU 24:96–99 (1996).
www.soci.org H-X Xu, J-W Chen, M Xie
24 Arakawa O, Uematsu N and Nakajima H, Effect of bagging on fruit
quality in apples. Bull Faculty Agric Hirosaki Univ 57:25–32 (1994).
25 Hasan MA, Bhattacharjee S and Debnath U, Fruit quality and
microclimate variation inside the bunch cover of Dwarf Cavendish
banana (Musa AAA). OrissaJHort29:46–50 (2001).
26 Lopez M, Candela ME and Sabarer F, Carotenoids from Capsicum
annuum fruits: Influence of spectral quality of radiation. Biol Plant
28:100–104 (1986).
27 Yang WH, Zhu XC, Bu JH, Hu GB, Wang HC and Huang XM, Effects of
baggingonfruitdevelopmentandqualityincross-winteroff-season
longan. Sci Hort 120:194–200 (2009).
28 PriorRL, CaoG, MartinA, SoficE, McewenJ, O’BrienB, et al,
Antioxidant capacity as influenced by total phenolic and
anthocyanin content, maturity and variety of Vaccinium species.
J Agric Food Chem 46:2686–2693 (1998).
29 Ehlenfeldt MK and Prior RL, Oxygen radical absorbance capacity
(ORAC) and phenolic and anthocyanin concentrations in fruit and
leaf tissues of highbush blueberry. J Agric Food Chem 49:2222–2227
(2001).
30 Deighton N, Brennan R, Finn C and Davies HV, Antioxidant properties
of domesticated and wild Rubus species. J Agric Food Chem
80:1307–1313 (2000).
31 Namiki M, Antioxidants/antimutagens in food. Crit Rev Food Sci Nutr
29:273–300 (1990).
32 Birt DF, Hendrich S and Wang W, Dietary agents in cancer prevention:
flavonoids and isoflavonoids. Pharmacol Ther 90:157–177 (2001).
33 Cantuti-Castelvetri I, Shukitt-Hale B and Joseph JA, Neurobehavioural
aspects of antioxidants in aging. Int J Develop Neurosci 18:367–381
(2000).
34 Bakhshi D and Arakawa O, Induction of phenolic compounds
biosynthesis with light irradiation in the flesh of red and yellow
apples. J Appl Hort 8:101–104 (2006).
35 Logemann E, Wu S, Schröder J, Schmelzer E, Somssich I and
Hahlbrock K, Gene activation by UV light, fungal elicitor or fungal
infection in Petroselinum crispum is correlated with repression of
cell cycle-related genes. Plant J 8:865–876 (1995).
36 Logemann E, Tavernaro A, Schulz W, Somssich IE and Hahlbrock K,
UV light selectively coinduces supply pathways from primary
metabolism and flavonoid secondary product formation in parsley.
Proc Natl Acad Sci U S A 97:1903–1907 (2000).
37 Bieza K and Lois R, An Arabidopsis mutant tolerant to lethal ultraviolet-
B levels shows constitutively elevated accumulation of flavonoids
and other phenolics. Plant Physiol 126:1105–1115 (2001).
38 Jaakola L, Määttä-Riihinen K, Kärenlampi S and Hohtola A, Activation
of flavonoid biosynthesis by solar radiation in bilberry (Vaccinium
myrtillus L.) leaves. Planta 218:721–728 (2004).
39 Awad M, Wagenmakers P and de Jager A, Effects of light on flavonoid
and chlorogenic acid levels in the skin of ‘Jonagold’ apples. Sci Hort
88:289–298 (2001).
40 Kolb C, Käser M, Kopec´yJ,ZotzG,RiedererMandPfündel E, Effects of
natural intensities of visible and ultraviolet radiation on epidermal
ultraviolet screening and photosynthesis in grape leaves. Plant
Physiol 127:863–875 (2001).
41 Tattini M, Gravano E, Pinelli P, Mulinacci N and Romani A, Flavonoids
accumulate in leaves and glandular trichomes of Phillyrea latifolia
exposed to excess solar radiation. New Phytol 148:69–77 (2000).
42 Wang H, Cao G and Prior RL, Total antioxidant capacity of fruits. JAgric
Food Chem 44:701–705 (1996).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1783–1788

Research Article
Received: 16 February 2010 Revised: 8 April 2010 Accepted: 15 April 2010 Published online in Wiley Interscience: 25 May 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4014
Preparation of a monoclonal antibody and
development of an indirect competitive ELISA
for the detection of chlorpromazine residue in
chicken and swine liver
Wei Liu, a Weihua Li, a Weiwei Yin, a Meng Meng, b∗ Yuping Wan, c
Caiwei Feng, c Shanliang Wang c and Rimo Xi b∗
Abstract
BACKGROUND: Chlorpromazine is a typical antipsychotic drug used to make food-producing animals calm and promote growth
as feed additives. Accumulation of chlorpromazine in animal bodies would cause side effects in the circulatory and nervous
systems, and have adverse effects on blood cells, the skin and the eye. To detect the chlorpromazine residue in food producing
animals, an indirect competitive enzyme-linked immunosorbent assay (ELISA) was developed based on preparation of an
anti-chlorpromazine monoclonal antibody.
RESULTS: The antibody generated from immunogen of cationic bovine serum albumin (cBSA) coupled with chlorpromazine
showed high sensitivity toward chlorpromazine with an IC50 value of 0.73 ppb. The ELISA method was applied to detect swine
liver and chicken samples spiked by chlorpromazine and satisfactory results were obtained. The recovery rates in chicken and
swine liver were in the range of 88–95% and 86–95%, respectively; the intra-assay coefficients of variation were both

1790
F
F
O
O
N
S
N
www.soci.org W Liu et al.
Acepromazine Chloropromazine
Azaperone
Haloperidol
Figure 1. Chlorpromazine and other structurally related sedatives analysed in this study.
raphy–mass spectrometry14 – 16 have been used. These methods
require extensive sample preparation, sometimes expensive apparatus,
highly trained personnel to operate sophisticated instruments
and interpret complicated results, and can only determine
a limited number of samples at one time. Therefore these methods
are not suitable as screening tests. In contrast, because of the rapidity,
mobility, convenience, high sensitivity, and low detection limit,
enzyme-linked immunosorbent assay (ELISA) methods have been
N
N
N
OH
used for the detection of various drug residues in real systems.
O
N
17 – 22
A key factor for an ELISA test is whether a polyclonal antibody
or monoclonal antibody (MAb) towards the compound detected
was used. Compared with a polyclonal antibody, the application
of a monoclonal antibody is advantageous in terms of better
purity, satisfactory sensitivity and high specificity. 23 – 26 In this
paper, an ELISA test kit based on a monoclonal antibody toward
chlorpromazine was developed and applied to detect chlorpromazine
spiked in swine liver and chicken samples. This study is
the first to prepare a monoclonal antibody of chlorpromazine
and develop an immunoassay based on an antibody to detect
residues of chlorpromazine in swine liver and chicken.
EXPERIMENTAL
Chemicals and materials
Bovine serum albumin (BSA), ovalbumin (OVA) and goat
anti-mouse IgG–horseradish peroxidase (HRP) conjugate
were provided by Beijing Wanger Biotechnology Co., Ltd.
(Beijing, China). o-Phenylenediamine (OPD) and 3,3 ′ ,5,5 ′ -
F
Cl
N
S
OH
N
S
Azaperol
N
Promethazine
N
N
Cl
N
tetramethylbenzidine (TMB), N,N ′ -dicyclohexylcarbodiimide (DCC)
and N-hydroxysuccinimide (NHS) were purchased from Xinjingke
Biotechnology (Beijing, China). Freund’s complete (cFA) and incomplete
adjuvants (iFA) were obtained from Sigma–Aldrich (St
Louis, MO, USA). n-Hexane, hydrochloric acid, dimethylsulfoxide
(DMSO), sulfuric acid, sodium hydroxide, acetonitrile, hydrogen
peroxide (30% H2O2) and other reagents used were provided by
Guangmang Chemical Co. (Jinan, China). Chicken samples were
purchased from commodity exchange and swine liver samples
were from a supermarket in Jinan City, China.
Instrumentation and supplies
ELISA was performed on polystyrene 96-well microtitre plates (Bio
Basic Inc., Ontario, Canada) and spectrophotometrically read with
a GF-M3000 microplate reader (Ruicong Shanghai Technology
Development Co., Ltd, Shanghai, China). Centrifugation was
carried out with a Biofuge Stratos refrigerated centrifuge (Heraeus,
Hanau, Germany). Protein dialyses were performed using dialysis
bags from Aibo Economic & Trade Co., Ltd (Jinan, China). LC-MS
was performed on a LC/MS-2010A instrument from Shimadzu
(Kyoto, Japan).
Buffers
Ultra-pure deionised water was used for the preparation of all
buffers and reagents for the immunoassays, unless especially
indicated. Phosphate-buffered saline (PBS, pH 7.4) consisted of
138 mmol L −1 NaCl, 1.5 mmol L −1 KH2PO4, 7 mmol L −1 Na2HPO4
and 2.7 mmol L −1 KCl. The wash buffer (PBST) was a PBS buffer
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1789–1795
N

Detection of chlorpromazine in swine liver and chicken by ELISA www.soci.org
added 0.05% (v/v) Tween 20. 0.05 M carbonate buffer (15 mmol L −1
Na2CO3 and 35 mmol L −1 NaHCO3, pH9.6)wasusedasacoating
buffer. The blocking buffer was a solution of PBS mixed with
1% of OVA and 0.05% (v/v) Tween 20. The substrate buffer was
0.1 mol L −1 sodium acetate/citrate buffer (pH 5.0). Eighty millilitres
of acetonitrile added to 20 mL of 0.1 mol L −1 HCl was used as the
extractive solution. To prepare the substrate solution, 10 mg of
TMB + 5 mL DMSO was defined as substrate solution A and 5 µLof
H2O2 [30% (w/w)] + 15 mL citrate buffer was defined as solution
B. The stopping solution was 2 mol L −1 H2SO4.
Preparation of immunogen and coating antigen
Modification of hapten
As described in Fig. 2 (Scheme 1), the mixture of acepromazine
(compound I) (10 mg, 0.031 mmol) in 2.5 mL of ethanol and
hydroxylamine hydrochloride (2.5 mg) in 1.0 mL of distill water
was stirring under reflux in water bath for 2.5 h. A solution of NaOH
(1.0 mL, 0.05 mol L −1 ) was added dropwise during the procedure.
An acetate buffer (1.0 mL, pH 4.0) was added dropwise followed
by adding 4 mg of ice until a white precipitate appeared. The
white solid was isolated by centrifugation (10 000 × g)after1day.
The white solid was dissolved in DMF (2.5 mL) and the mixture
was kept reaction at room temperature for 2 h after succinic
anhydride was added. The reaction was continuing for 4 h after
100 mL of triethylamine was added to yield chlorpromazine hapten
(compound III).
Preparation of immunogen
The immunogen of chlorpromazine was prepared via a mixed
acid anhydride reaction (Fig. 2, Scheme I). In this procedure, 15 µL
of isobutyl chloroformate was added into the 1.0 mL of hapten
prepared above at 10 ◦ C to obtain solution A. The solution B
was prepared by dissolving 36 mg of BSA in 2 mL of sodium
carbonate (50 mmol L −1 ). Under stirring, the solution A was added
dropwise into solution B with molar ratio of hapten : BSA =
10 : 1 for 4 h at 10 ◦ C to get the immunogen of cationic BSA
(cBSA)–chlorpromazine (compound V). The mixture was stirred
overnight at 4 ◦ C and dialysed for 3 days against PBS (0.01 mol L −1 ),
exchanging the dialysis solution twice each day. The solution
obtained was stored at −20 ◦ C for future use.
Preparation of coating antigen
The mixture of 1.0 mL of hapten (compound III) obtained in
procedure Scheme I, 20 mg of DCC, and 12.5 mg of NHS in 0.5 mL
of DMF was stirring for 24 h at room temperature to obtain solution
C. The solution D was prepared by dissolving 50 mg OVA in 3.5 mL
of PBS (0.01 mol L −1 , pH 7.2). The solution C was added dropwise
into solution D and the mixture obtained was stirred at room
temperature for 3 h to remove small amounts of impurities. The
precipitation was removed by centrifugation at 13 000 × g for
30 min and the supernatant was collected to obtain the coating of
cationic OVA (cOVA)–chlorpromazine (compound VII), which was
stored at −20 ◦ C for future use.
Immunisation, cell fusion and purification of MAbs
BALB/c mice were initially immunised by an intraperitoneal injection
of 150 µg of cBSA–chlorpromazine in an equal volume of cFA.
Two weeks later, a booster injection was performed using the same
amount of cBSA–chlorpromazine in iFA. More double boosts were
continued with 100 mg cBSA–chlorpromazine given in the tail vein
at an interval of 2 weeks. After the immune response had been validated,
the splenocytes from immunised mice were fused with a
logarithmically growing hypoxanthine–aminopterin–thymidine
(HAT)-sensitive mouse myeloma cells Sp2/0 (7 : 1) by the polyethylene
glycol (PEG) method. The hybridoma cells were screened by
indirect ELISA and cloned by the limited dilution method. The
hybridoma cells were cultured in the medium (pH 7.4) containing
0.2% NaHCO3 and RPMI1640 added 20% newborn calf serum
in 37 ◦ C to obtain MAbs for chlorpromazine. The purification was
performed according to the acid–ammonium sulfate method, and
the purified MAbs obtained was stored at −20 ◦ C for future use.
Antibody titre determination by indirect competitive ELISA
The antibody titre was tested by indirect ELISA. The procedure
was carried out as described below. The microplates were coated
with coating antigen cOVA–chlorpromazine at 1/500, 1/1000 and
1/2000 by overnight incubation at 4 ◦ C. Plates were washed with
wash buffer three times and blocked with 250 µLwell −1 of blocking
buffer, followed by incubation for 1 h at room temperature. Plates
were washed three times again, then the appropriate dilution
of antisera was added, and the plates were incubated for 2 h at
room temperature. After the plates had been washed three times,
goat anti-mouse IgG-HRP (1 : 1000, 100 µL well −1 ) was added,
followed by incubation for 2 h at room temperature. Plates were
washed three times and TMB substrate solution A and B was added
(50 µL well −1 ) in turn. After that, the plates were incubated for
another 15 min at room temperature. The colour development
was inhibited by adding stopping solution (100 µL well −1 ), and
absorbances were measured at 450 nm. Absorbance values were
corrected by blank reading. Pre-immune withdrawal serum (the
serum before immunisation) was used as a negative control, and
the antibody titre was defined as the reciprocal of the dilution that
resulted in an absorbance value of twice the blank value.
Development of indirect competitive ELISA
The checker-board procedure was used to obtain the optimised
coating antigen and the primary antibody concentrations.
To each well of a 96-well plate, 100 µL of10µg mL −1 of
cOVA–chlorpromazine in bicarbonate buffer (0.05 mol L −1 , pH
9.6) was added, and the mixture was incubated overnight at 4 ◦ C.
The plate was washed with wash buffer three times between each
step, and blocked with 250 µL well −1 of blocking buffer, followed
by incubation for 1 h at room temperature. After the blocking
solution was removed, 100 µL of primary antibody was added to
each well followed by the addition of PBST buffer or competitor
in PBST buffer, and the plate was incubated for 2 h. Then, the goat
anti-mouse IgG HRP (1 : 1000, 100 µL well −1 ) was added, followed
by incubation for 2 h at room temperature. Substrate solutions
A and B were added (50 µL well −1 ) in turn, and the plate was
incubated for another 15 min at room temperature. The colour development
was inhibited by adding stopping solution (2 mol L −1
H2SO4, 100 µL well −1 ), and absorbance was measured at 450 nm.
Absorbance was corrected by blank reading. Pre-immune withdrawal
serum was used as a negative control. The result was
expressed in % inhibition as follows: % inhibition = %B/B0, where
B is the absorbance of the well with competitor and B0 is the
absorbance of the well without competitor.
Standard curve generation
The cOVA–chlorpromazine (1/1000) was used as coating antigen,
and indirect cELISA was performed as described above. The
J Sci Food Agric 2010; 90: 1789–1795 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1791

1792
Scheme 1
N
S
(III) +
Scheme 2
N O
NH 2 OH⋅HCl
N
S
www.soci.org W Liu et al.
N
N OH
(I) (II) (III)
(III)
O
Cl O
isobutyl chloroformate
O
+ HO N
O
Et3N DMF
DCC
DMF
N
S
(IV)
N
S
O
O
O
O
N
N O C CH2CH2COCOOCH2CH(CH3 ) 2
BSA-NH 2
(V)
O
N
N O C CH2CH2CO N
S
N
S
N
OVA NH 2
N
O
N
S
NH BSA
N O C (CH2 ) 2COO (VI)
O
O
N
N O C CH2CH2COOH N
O
O
N O C (CH2 ) 2CH2OO (VII)
NH OVA
Figure 2. The synthetic procedure for immunogen of cBSA–chlorpromazine (Scheme 1) and the coating antigen of cOVA–chlorpromazine (Scheme 2).
selected antiserum at 1/8000 dilution was utilised as primary
antibody and co-incubated with chlorpromazine. The standard
calibration curve with final chlorpromazine concentrations of 0.05,
0.25, 0.5, 1.0, 2.0 and 10.0 ng mL −1 was run in PBST.
The pretreatment of the samples (swine liver and chicken)
The sample was first milled for 10–15 min for homogenising. One
gram of the chicken was weighed into a polythene tube, and
then 4 mL of the extraction solution was added. The mixture was
shaken vigorously for 5 min and centrifuged at room temperature
(20–25 ◦ C) for 10 min at 3000×g. Two millilitres of the supernatant
was transferred and mixed with 4 mL of 1 mol L −1 NaOH, and then
10 mL of n-hexane was added. After being shaken vigorously
for 5 min, the mixture was centrifuged at room temperature
(20–25 ◦ C) for 5 min at 3000 × g. Five millilitres of the supernatant
was extracted and placed into a 50 mL of round-bottom glass flask.
Then the solution was evaporated to dryness at low pressure at
55 ◦ C. Then, 1 mL of PBST was added to the flask and the solution
was treated as the blank sample which was stored at 0–4 ◦ Cfor
future use.
Optimisation of the blank sample
Three different chicken and swine liver samples were purchased
from different supermarkets for preparing blank samples. The
absorbance of the blank samples was tested by the indirect
ELISA method, as described below. The microplates were coated
with coating antigen cOVA–pefloxacin at 1/1000 by overnight
incubation at 4 ◦ C. Plates were washed with wash buffer three
times and blocked with 250 µLwell −1 of blocking buffer, followed
by incubation for 1 h at room temperature. Plates were washed
three times again, then the blank samples (50 µLwell −1 ), incubated
with the antibody (1/8000, 50 µL well −1 ), were added. The plates
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1789–1795

Detection of chlorpromazine in swine liver and chicken by ELISA www.soci.org
were incubated for 2 h at room temperature, then washed three
times, and goat anti-mouse IgG-HRP (1 : 1000, 100 µL well −1 )was
added, followed by incubation for 2 h at room temperature. Plates
were washed three times and TMB substrate solutions A and B
were added (50 µL well −1 ) in turn, and the plates were incubated
for another 15 min at room temperature. The colour development
was halted by adding stopping solution (100 µL well −1 ), and
absorbance was measured at 450 nm. Absorbance was compared
with the blank reading of wells with PBST added instead of samples.
The selected blank sample was defined according to the standard
curve.
The blank sample selected was spiked with chlorpromazine
standard solution in PBST. Competitive curves with final chlorpromazine
concentrations of 0.05, 0.25, 0.5, 1.0, 2.0 and 10 ng mL −1
were run in PBST and in blank sample to determine the matrix
effect of swine liver and chicken. IC50 and B0 values from the blank
sample curve were obtained by comparing IC50 and B0 values
generated from the PBST buffer solution.
RESULTS AND DISCUSSION
Preparation of antigen and characterisation of the antibody
As a small molecule, chlorpromazine has to be connected with
carrier protein in order to be immunogenic. The structure of
chlorpromazine (Fig. 1) showed that the side chain composed
of N–(CH2)3 –N(CH3)2 is a very important structural feature for
this molecule. In order to obtain a specific antibody towards
chlorpromazine, it will be important to expose this chain outside
instead of connecting this chain directly to avoid covering the most
important structural feature by carrier protein. There is no suitable
functional group in the chlorpromazine molecule to use for linking
with carrier protein. To synthesise an immunogen for preparation
of anti-chlorpromazine antibody, acepromazine was chosen as the
starting material because there is acetyl group in the acepromazine
molecule (Fig. 2, Scheme 1), which can be used as a linking site
with carrier protein. Considering that the acetyl group is located
in a position that is close to the side chain N–(CH2)3 –N(CH3)2, it
requires a chain to separate hapten from carrier protein to ensure
that the hapten is uncovered by carrier protein. First, acepromazine
reacted with hydroxylamine hydrochloride to form an immediate
(II) (Fig. 2, Scheme 1), which was then treated by succinic anhydride
to form activated immediate (III). This derivatisation introduced a
carboxylic acid moiety into the hapten, which was a convenient
functional group for conjugation with a carrier protein. The space
arm consisting of seven atoms will be helpful to increase the
reorganisation of antibody to chlorpramazine hapten. Starting
from compound (III), both immunogen and coating antigen could
be prepared with different carrier proteins. The immunogen was
prepared through a mixed anhydride method to link hapten with
carrier protein BSA (cBSA) (Fig. 2). The coating antigen using OVA
as carrier protein was prepared by the DCC method in DMF as
shown in Fig. 2 (Scheme 2). The yields of immunogen and coating
antigen obtained from Scheme 1 and Scheme 2 were both more
than 30%.
In order to characterise the antibody produced in this
research, the titre, sensitivity and cross-reactivity were determined
according to the indirect competitive ELISA procedure described
above. The titre of the antibody determined for the ELISA method
was defined as the reciprocal of the dilution which resulted in
an absorbance value that was twice of the background value.
The titre of Mabs developed by the immunisation process
was more than 100 000 for all mice. The prepared antibody
B/B 0 (%)
1.0
0.8
0.6
0.4
0.2
0.0
0.1 1 10
Chlorpromazine concentration (log[ppb])
Figure 3. Inhibition curve of anti-chlorpromazine monoclonal antibody
with chlorpromazine as a competitor in PBST.
Table 1. IC50 and percentage of cross-reactivities of selected
compounds
Compound IC50 a (ppb) Cross-reactivities b (%)
Chlorpromazine 0.73 100
Acepromazine 75.0 1.0
Promethazine >150 750 750 150

1794
B/B 0
0.8
0.7
0.6
0.5
0.4
0.3
0.2
y = 0.42724 – 0.54079x
R 2 = 0.99834
Chlorpromazine concentration (ppb)
Figure 4. Standard calibration curve of indirect competitive ELISA in PBST.
Table 2. Calculation and summary of limit of detection (LOD) of swine
liver and chicken samples
Parameter Swine liver samples Chicken samples
Number of samples 20 20
Added (µgkg−1 ) 0 0
Means observed
(µgkg−1 )
0.10 0.08
Standard deviation
(s) (µgkg−1 )
0.12 0.08
Mean + 3s (µgkg−1 ) 0.46 0.32
Highest observed
blank (µgkg−1 )
0.30 0.20
LOD (µgkg−1 ) 0.45 0.31
Determination of matrix effect on the indirect competitive
ELISA
According to the Council Regulation (EEC) No 2377/90, chlorpromazine
must not to be detectable in ‘veterinary medicinal products
in respect of all the various foodstuffs of animal origin, including
meat, fish, milk, eggs and honey’, so we detected chlorpromazine
in foodstuffs of animal origin. We selected two classic animal tissues
from two classic animals commonly fed by chlorpromazine
for validation. One was chicken, another was swine liver. In addition,
it is reported that pigs are particularly sensitive to the change
of surrounding conditions, so the stress could cause high morality
rates or make pigs produce low-quality meat, pale and soft. These
serious adverse events make farmers utilise more sedatives as
feed additives. Hence, swine liver samples were also selected to
evaluate the ELISA established in our study in biological system
besides chicken samples. Other matrices, such as urine, milk, eggs
and honey, have not been analysed yet.
To determine the limit of detection (LOD), 20 batches of blank
swine liver samples and 20 batches of blank chicken samples were
purchased from the local supermarket. All the samples have been
tested by LC-MS to make sure there is no chlorpromazine residue
in them.
The linear range of the ELISA determined as the concentration
of 20–80% inhibition of maximal absorbance value was
0.60–2.0 ppb, and the B/B0 valueinthisrangewasplottedversus
chlorpromazine concentration to obtain standard curve (Fig. 4).
1
www.soci.org W Liu et al.
2
Table 3. Inter- and intra-assay variations of swine liver and chicken
samples spiked with chlorpromazine
Level (µg
kg −1 ) n
Average
recovery (%)
Inter-assay
variation a (%)
Intra-assay
variation b (%)
Swine liver
0.5 4 95.6 12.6 13.5
1.0 4 86.2 9.7 11.0
2.0 4 90.0 11.5 12.8
4.0 4 89.2 11.0 11.7
Chicken
0.5 4 94.1 10.6 12.2
1.0 4 92.4 13.0 15.3
2.0 4 88.9 12.4 13.9
4.0 4 91.2 10.8 11.3
a Inter-assay variation was determined by four replicates on 15 different
days.
b Intra-assay variation was determined by four replicates on a single
day.)
The curve obtained showed good linearity (R 2 = 0.9983) in this
range. Using this curve, the 20 blank samples were analysed using
the ELISA to demonstrate the range of blank matrix effects
in the assay. As shown in Table 2, results of these 20 known
chlorpromazine-free samples gave a mean of 0.10 µgkg −1 for
swine liver samples and 0.08 µg kg −1 for chicken samples. The
highest observed blank was 0.30 µgkg −1 for swine liver samples
and 0.20 µgkg −1 for chicken samples. As a general rule, 27 the
LOD is defined as the mean observed chlorpromazine concentration
plus three times the standard deviations, or the highest
observed chlorpromazine concentration, whichever the greater.
As was shown in Table 2, in both cases, the LOD was determined
by the mean plus three standard deviations due to their higher
values compared to the highest observed blank value.
The same swine liver and chicken samples spiked by 0.5, 1.0 and
2.0 µg kg −1 , respectively, were measured by the immunoassay to
determine the variations of the ELISA method. As was shown in
Table 3, the variation of coefficients were determined to be in the
range 9.7–13.0% for inter-assay and 11.0–15.3% for intra-assay,
whereas the average recovery rates were in the range 86.2–95.6%,
indicating satisfactory accuracy and precision.
In our study, the animal tissues analysed were spiked with
chlorpromazine in vitro, in order to evaluate the matrix effect on
recovery and variation efficiency of the method. However, in vivo,
chlorpromazine is rapidly metabolised. Based on the evaluation
of chlorpromazine by JECFA, 8 the major metabolic pathways are
hydroxylation, oxidation, demethylation and glucuronidation. The
metabolites varied differently in different species. Further work of
our study is to apply the method in the analysis of tissues from
animals fed by chlorpromazine and specifically demonstrate the
antibody with chlorpromazine metabolites in different animals.
CONCLUSION
In summary, an immunogen of chlorpromazine was designed
and synthesised. The monoclonal antibody for chlorpromazine
based on the immunogen was prepared for the first time. The
antibody showed high sensitivity with an IC50 value of 0.73 ppb,
limit of detection of 0.05 ppb and specificity with almost no
cross-reactivity towards commonly used sedatives. When applied
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1789–1795

Detection of chlorpromazine in swine liver and chicken by ELISA www.soci.org
to detect chlorpromazine spiked in swine liver and chicken,
satisfactory accuracy and precision were obtained. Hopefully,
the ELISA test method is an alternative to chromatography for
regulatory analysis of chlorpromazine residues in foodstuffs of
animal orgin.
ACKNOWLEDGEMENT
This research was supported by the National Natural Science
Foundation of China (No.20675048), the Shandong Natural Science
Foundation (Y2008B31), the National High-Tech Research and the
Development Program of China (863 Program, No. 07AA10Z435,
No. 2007AA06A407).
REFERENCES
1 Amaral L, Viveiros M and Molnar J, Antimicrobial activity of
phenothiazines. In Vivo 18:725–731 (2004).
2 Kojlo A, Karpinska J and Kuzmicka L, Analytical study of the reaction
of phenothiazines with some oxidants, metal ions, and organic
substances. J Trace Microprobe Techniques 19:45–70 (2001).
3 Amaral L, Kristiansen JE and Viveiros M, Activity of phenothiazines
against antibiotic-resistant Mycobacterium tuberculosis: a review
supporting further studies that may elucidate the potential use of
thioridazine as anti-tuberculosis therapy. J Antimicrob Chemother
47:505–511 (2001).
4 Dyer KS and Woolf AD, Use of phenothiazines as sedatives in
children – What are the risks? Drug Saf 21:81–90 (1999).
5 Thornley B, Adams CE and Awad AG, Chlorpromazine versus placebo
for the treatment of schizophrenia: A systematic review. Schizophr
Res 29:155–155 (1998).
6 Rijcken CAW, Monster TBM and Brouwers JRBJ, Chlorpromazine
equivalents versus defined daily doses: How to compare
antipsychotic drug doses? J Clin Psychopharmacol 23:657–659
(2003).
7 Cooper J, Delahaut P and Fodey TL, Development of a rapid screening
test for veterinary sedatives and the beta-blocker carazolol in
porcine kidney by ELISA. Analyst 129:169–174 (2004).
8 Evaluation of certain veterinary drug residues in food, in Thirty-eighth
Report of the Joint FAO/WHO Expert Committee on Food Additives.
World Health Organization, Geneva (1991).
9 Committee for Veterinary Medicinal Products, Chlorpromazine
– summary report, in EMEA/MRL/111/96-FINAL, The European
Agency for the Evaluation of Medicinal Products, Veterinary
Medicines Evaluation Unit (1996).
10 EEC, Council Regulation (EEC) 2377/90 of 26 June 1990. Off J Eur
Commun L224:1–6 (1990).
11 Cruz-Vera M, Lucena R and Cardenas S, Determination of
phenothiazine derivatives in human urine by using ionic liquidbased
dynamic liquid-phase microextraction coupled with liquid
chromatography. J Chromatogr B 877:37–42 (2009).
12 Sobhi HR, Yamini Y and Abadi RHHB, Extraction and determination of
trace amounts of chlorpromazine in biological fluids using hollow
fiber liquid phase microextraction followed by high-performance
liquid chromatography. J Pharm Biomed Anal 45:769–774 (2007).
13 Saracino MA, Amore M and Baioni E, Determination of selected
phenothiazines in human plasma by solid-phase extraction and
liquid chromatography with coulometric detection. Anal Chim Acta
624:308–316 (2008).
14 Mitrowska K, Posyniak A and Zmudzki J, Rapid method for the
determination of tranquillizers and a beta-blocker in porcine
and bovine kidney by liquid chromatography with tandem mass
spectrometry. Anal Chim Acta 637:185–192 (2009).
15 Pujadas M, Pichini S and Civit E, A simple and reliable procedure
for the determination of psychoactive drugs in oral fluid by
gas chromatography–mass spectrometry. J Pharm Biomed Anal
44:594–601 (2007).
16 Wu HQ, Jin YC and Cai MZ, Simultaneous determination of 10 mental
drugs by gas chromatography–mass spectrometry. Chin J Anal
Chem 35:500–504 (2007).
17 Zhang YL, Lu SX and Liu W, Preparation of anti-tetracycline antibodies
and development of an indirect heterologous competitive enzymelinked
immunosorbent assay to detect residues of tetracycline in
milk. J Agric Food Chem 55:211–218 (2007).
18 Lu SX, Zhang YL and Liu JT, Preparation of anti-pefloxacin antibody
and development of an indirect competitive enzyme-linked
immunosorbent assay for detection of pefloxacin residue in chicken
liver. J Agric Food Chem 54:6995–7000 (2006).
19 Liu W, Zhao CB and Zhang YL, Preparation of polyclonal antibodies
to a derivative of 1-aminohydantoin (AHD) and development of
an indirect competitive ELISA for the detection of nitrofurantoin
residue in water. J Agric Food Chem 55:6829–6834 (2007).
20 Zhao CB, Liu W and Ling HL, Preparation of anti-gatifloxacin antibody
and development of an indirect competitive enzyme-linked
immunosorbent assay for the detection of gatifloxacin residue
in milk. J Agric Food Chem 55:6879–6884 (2007).
21 Liu ZQ, Lu SX and Zhao CH, Preparation of anti-danofloxacin antibody
and development of an indirect competitive enzyme-linked
immunosorbent assay for detection of danofloxacin residue in
chicken liver. J Sci Food Agric 89:1115–1121 (2009).
22 Ding K, Zhao CH and Cao ZZ, Chemiluminescent detection of
gatifloxacin residue in milk. Anal Lett 42:505–518 (2009).
23 Gajewski KG and Hsieh YHP, Monoclonal antibody specific to a major
fish allergen: parvalbumin. J Food Prot 72:818–825 (2009).
24 Stanker LH, Merrill P and Scotcher MC, Development and partial
characterization of high-affinity monoclonal antibodies for
botulinum toxin type A and their use in analysis of milk by sandwich
ELISA. J Immunol Methods 336:1–8 (2008).
25 Saravanan P, Sen A and Balamurugan V, Rapid quality control of a live
attenuated Peste des petits ruminants (PPR) vaccine by monoclonal
antibody based sandwich ELISA. Biologicals 36:1–6 (2008).
26 Kim KH, Shim JH and Seo EH, Interleukin-32 monoclonal antibodies
for immunohistochemistry, western blotting, and ELISA. J Immunol
Methods 333:38–50 (2008).
27 Cooper KM, Elliott CT and Kennedy DG, Detection of 3-amino-2oxazolidinone
(AOZ), a tissue-bound metabolite of the nitrofuran
furazolidone, in prawn tissue by enzyme immunoassay. Food Addit
Contam 21:841–848 (2004).
J Sci Food Agric 2010; 90: 1789–1795 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1795

1796
Research Article
Received: 25 January 2010 Revised: 18 March 2010 Accepted: 16 April 2010 Published online in Wiley Interscience: 22 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4015
Comparison of PCR-DGGE and PCR-SSCP
analysis for bacterial flora of Japanese
traditional fermented fish products,
aji-narezushi and iwashi-nukazuke
Choa An, Hajime Takahashi, Bon Kimura and Takashi Kuda ∗
Abstract
BACKGROUND: The bacterial flora of two Japanese traditional fermented fish products, aji-narezushi (salted and long-fermented
horse mackerel (Trachurusjaponicas) with rice) and iwashi-nukazuke (salted and long-fermented sardine (Sardinopsmelanostica)
with rice bran), was analysed using non-culture-based polymerase chain reaction (PCR) denaturing gradient gel electrophoresis
(DGGE) and culture-based PCR single-strand conformation polymorphism (SSCP) methods.
RESULTS: Viable plate counts in aji-narezushi and iwashi-nukazuke were about 6.3–6.6 and 5.7–6.9 log colony-forming units
g −1 respectively. In the PCR-DGGE analysis, Lactobacillus acidipiscis was detected as the predominant bacterium in two of three
aji-narezushi samples, while Lactobacillus versmoldensis was predominant in the third sample. By the PCR-SSCP method, Lb.
acidipiscis and Lactobacillus plantarum were isolated as the predominant bacteria, while Lb. versmoldensis was not detected.
The predominant bacterium in two of three iwashi-nukazuke samples was Tetragenococcus muriaticus, while Tetragenococcus
halophilus was predominant in the third sample.
CONCLUSION: The results suggest that the detection of some predominant lactic acid bacteria species in fermented fish by
cultivation methods is difficult.
c○ 2010 Society of Chemical Industry
Keywords: fermented fish; denaturing gradient gel electrophoresis (DGGE); single-strand conformation polymorphism (SSCP);
Lactobacillus; Tetragenococcus
INTRODUCTION
In modern Japanese cuisine, sushi is made from vinegar-flavoured
rice combined with seafood. It is thought that sushi originates
from the salted and long-fermented fish called narezushi in Japan.
The earliest reference to sushi appears in a code named the Yoro-
Ritsuryo issued in AD 718, and this earliest sushi is postulated to
have been narezushi. 1 Since that time, narezushi products have
been made from various freshwater fish in several areas located
inland rather than in coastal regions. Funazushi, a fermented
crucian carp with cooked rice made near Lake Biwa in central Japan,
is the most famous narezushi, characterised by its strong flavours
and odours. Currently, narezushi is also made from marine fish.
For example, aji-narezushi, made from horse mackerel (Trachurus
japonicas) and rice, has been manufactured in the Noto peninsula
in Ishikawa, Japan since the middle of the last century. 2
Fish nukazuke, salted and long-fermented fish with rice bran
(nuka, by-product of polished rice), is also one of the traditional
and popular fermented fish products in Japan. Puffer fish ovary
nukazuke is a famous nukazuke product, because its deadly poison
is eliminated during long-term salting (2 years) and fermentation
with rice bran (1–2 years). 3 However, iwashi-nukazuke, made
from sardine (Sardinops melanostica), is the most reasonable and
popular fish nukazuke. Recently, some bioactive substances such
as antioxidants have been reported in iwashi-nukazuke. 4
It is well known that lactic acid bacteria (LAB) in fermented
foods affect not only product quality and preservation 5 but also
food functionality, such as improving the intestinal environment
and antihypertensive effects. 6,7 Recently, a high content of
γ -aminobutyric acid (GABA) was detected in aji-narezushi. 2
Production of GABA by LAB, including isolates from traditional
fermented foods, has been reported. 8
Determination of the microflora of aji-narezushi and iwashinukazuke
using culture-based methods has been reported. 2,9
During fermentation, LAB and lactic acid increase and the pH value
decreases (to 4.2 or less). However, the microflora has not yet been
clearly identified at the levels of genus and species. Owing to
the known limitations of cultivation methods, many recent studies
haveusedculture-independent16SrDNA-basedpolymerasechain
reaction (PCR) techniques, including PCR denaturing gradient gel
electrophoresis (DGGE), to determine the bacterial flora of various
traditional fermented foods such as fermented milk and soybean
∗ Correspondence to: Takashi Kuda, Department of Food Science and Technology,
Tokyo University of Marine Science and Technology, Minato-ku, Tokyo
108-8477, Japan. E-mail: kuda@kaiyodai.ac.jp
Department of Food Science and Technology, Tokyo University of Marine
Science and Technology, Minato-ku, Tokyo 108-8477, Japan
J Sci Food Agric 2010; 90: 1796–1801 www.soci.org c○ 2010 Society of Chemical Industry

Analysis of bacterial flora of traditional Japanese fermented fish products www.soci.org
products. 10–12 In this study, to clarify the quality and functional
properties of Japanese traditional products of lactic-fermented
fish with rice, we investigated the bacterial flora using a molecular
approach, combining PCR amplification of the V3 region of the
16S rDNA gene and DGGE (PCR-DGGE). Furthermore, we identified
the isolated strains using the culture-based PCR single-strand
conformation polymorphism (PCR-SSCP) method. 13
MATERIALS AND METHODS
Samples
In order to carry out the study, aji-narezushi samples were obtained
from three fish shops in Noto, Ishikawa, Japan in August 2008.
During aji-narezushi processing, fresh horse mackerel (20–60 g
body weight) were gutted and, after removal of the eyes or whole
head, placed in a barrel with plenty of salt (180–330 g salt kg −1
fish). The salted fish were then desalted in a vat filled with thin
(diluted two to five times) rice vinegar (komesu, approximately
45 g acetic acid L −1 ). The salting (from 3 days to several weeks)
and desalting (from several seconds to 5 h) periods varied with
each manufacturer. The treated fish were stuffed and covered
with cooked rice, scattered with a small amount of Japanese
pepper leaves and red pepper and pickled at ambient temperature
(15–30 ◦ C). The fermentation time was about 2–3 months. During
fermentation the lid of the fermenting barrel was kept closed by
stone weights.
Three iwashi-nukazuke samples were purchased from retail
shops in Hakusan, Ishikawa, Japan in September 2008. During
iwashi-nukazuke processing, fresh sardines (∼100 g body weight)
were gutted and placed in a barrel with plenty of salt for several
days. The salted fish were then washed and pickled in a barrel with
plenty of nuka and moulted rice (koji) for about 1 year at ambient
temperature.
For microbiological analysis, some of the whole samples were
separated and collected aseptically. Then the remaining samples
were stored at −30 ◦ C for chemical analysis.
Chemical analysis
Moisture, salinity, pH and organic acids of the samples were
determined using methods cited in our previous reports. 2,14 Water
activitywas measured with a water activitymeter (Pawkit, Decagon
Devices, Pullman, WA, USA).
Viable plate count
Samples (25 g) were emulsified in 225 mL of sterile phosphatebuffered
saline (PBS; 20 mmol L −1 KH2PO4, 10 mmol L −1 K2HPO4,
pH 7.2) and blended for 60 s (Stomacher 400, Seward, London,
UK). The sample suspensions were diluted in PBS and appropriate
dilutions were spread in duplicate on tryptone soy (TS) agar (Oxoid,
Basingstoke, UK), Gifu anaerobic medium (GAM) agar (Nissui,
Tokyo, Japan), de Man, Rogosa and Sharpe (MRS) agar (Oxoid) and
potato dextrose (PD) agar (Nissui) plates containing 100 mg L −1
chloramphenicol. To detect halophilic and halotolerant bacteria,
agar plates containing 100 g L −1 NaCl (TS-HS, GAM-HS, MRS-HS
and PD-HS) were also used. TS and PD agar plates were incubated
aerobically at 30 ◦ Cfor3days.GAMandMRSagarplateswere
incubated anaerobically at 30 ◦ C for 5 days using an AnaeroPack
system (Mitsubishi Gas Chemical, Tokyo, Japan). All agar plates
containing high salt (HS) were incubated for 7 days. In the case
of aji-narezushi samples, ten colonies each from GAM and MRS
agar plates were selected and restreaked for purification prior to
PCR-SSCP analysis. In the case of iwashi-nukazuke the colonies
were selected from GAM-HS and MRS-HS agar plates.
Direct extraction of DNA and PCR amplification
DNA from 1 mL of homogenate per sample and the isolates was
extracted using a FastPure DNA kit (Takara, Otsu, Japan). Purified
DNA was dissolved in ethylendiamine tetraacetic acid (EDTA)
buffer (TE buffer) and used as the DNA template in PCR.
The following primer pair was chosen for amplification of the V3
region of the 16S rRNA gene: forward primer with GC clamp GC-
339f (5 ′ -CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCC CCG
CCC GCT CCT ACG GGA GGC AGC AG-3 ′ ) and reverse primer V3-53r
(5 ′ -GTA TTA CCG CGG CTG CTG G-3 ′ ). 15 This primer set has been
widely used in DGGE analysis. 16 PCR amplification was performed
in 100 µL reaction mixtures composed of 10 mmol L −1 Tris/HCl
(pH 8.3), 50 mmol L −1 KCl, 1.5 mmol L −1 MgCl2, 50pmolofeach
primer, 0.2 mmol L −1 each of four dNTPs, 2.5 U of Takara Taq DNA
polymerase (Takara Bio, Shiga, Japan) and 50 ng of template DNA.
To minimise amplification of non-specific products and to obtain
largeamountsofPCRproducts,‘touchdown’PCRwasperformed, 17
where the initial annealing temperature was set at 8 ◦ C above the
expected annealing temperature and decreased by 0.8 ◦ C every
second cycle until the expected annealing temperature (62 ◦ C)
was reached (total 20 cycles) and then five additional cycles were
carried out. Amplification was carried out using the following
cycle: denaturation at 94 ◦ C for 30 s, annealing for 30 s and primer
extension at 72 ◦ C for 10 s in a GeneAmp 9700 thermal cycler
(Applied Biosystems, Foster City, CA, USA). Aliquots (5 µL) of
PCR products were analysed first by electrophoresis on 20 g L −1
agarose gels.
DGGE analysis of PCR products
DGGE analysis of PCR amplification products was performed as
described previously 16 using a DCode System apparatus (Bio-Rad
Laboratories, Hercules, CA, USA). Polyacrylamide gels (80 g L −1
acrylamide/bisacrylamide (37.5 : 1 w/w)) in 1× Tris/acetate/EDTA
buffer with a denaturing gradient ranging from 30 to 60%
denaturant (100% denaturation corresponds to 7 mol L −1 urea
and 400 mL L −1 formamide) were prepared with a Bio-Rad 475
gradient delivery system (Bio-Rad Laboratories). Polymerisation
was achieved by adding 9 mL L −1 ammonium persulfate and
0.9 mL L −1 N,N,N,N-tetramethyl ethylene diamine. The gels were
electrophoresed at a constant voltage of 200 V at 60 ◦ Cfor3h.The
DNA fragments were stained with ethidium bromide and washed
with distilled water prior to UV transillumination.
The main DGGE fragments were selected for nucleotide
sequence determination. Each band was excised with a sterile
razor. The DNA of each fragment was eluted in 50 µL ofTE
buffer at 100 ◦ C for 10 min. The extracts were reamplified by
PCR using the same primers and purified with SUPREC 138-PCR
(Takara) according to the manufacturer’s instructions. Purified
DNA fragments were ligated in pT7 blue vectors (Novagen,
Darmstadt, Germany) and transformed into Escherichia coli JM109.
The transformants were grown on Luria-Bertani broth (LB) agar
containing ampicillin and screened by β-galactosidase assay.
Plasmid DNA of selected transformants was isolated using a
Plasmid Miniprep kit (Bio-Rad Laboratories). The inserted DNA
sequence, approximately 200 bp of 16S rDNA (E. coli position
389–530), 18 was determined using an Applied Biosystems 3130
genetic analyser with a Big Dye Terminator V3.1 Cycle Sequencing
kit (Applied Biosystems). To identify the inserted sequences, the
J Sci Food Agric 2010; 90: 1796–1801 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1797

1798
BLAST 2.0 algorithm was used to compare the derived sequence
with 16S rDNA sequences in the DNA Data Bank of Japan (DDBJ)
database. The DGGE analysis was carried out duplicate.
PCR-SSCP analysis of 16S rDNA V3 region
In the PCR-SSCP analysis we used precast polyacrylamide gels
followed by silver staining because of the high sensitivity of silver
staining. This method visualises even a small amount of nonspecific
amplification product; therefore several PCR primers and
thermal profiles were tested for specificity and differences in PCR
efficiency. The primer set, SRV3-1 (5 ′ -CGG YCC AGA CTC CTA CGG
G-3 ′ ) as the forward primer and V3R53 (5 ′ -GTA TTA CCG CGG
CTG CTG GC-3 ′ ), which was designed based on 536R with minor
modifications, as the reverse primer, gave acceptable results. 19,20
PCR amplification was performed in 100 µL reaction mixtures
composed of 10 mmol L −1 Tris/HCl (pH 8.3), 50 mmol L −1 KCl,
1.5 mmol L −1 MgCl2, 50 pmol of each primer, 0.2 mmol L −1 each
of four dNTPs, 2.5 U of Takara Taq DNA polymerase (Takara Bio) and
50 ng of template DNA. In this analysis, ‘touchdown’ PCR was also
performed, where the initial annealing temperature was set at 6 ◦ C
above the target annealing temperature and decreased by 0.6 ◦ C
every second cycle until the target annealing temperature (61 ◦ C)
was reached and then five additional cycles were carried out at
the target annealing temperature. Amplifications were carried out
in a GeneAmp 9700 thermal cycler (Applied Biosystems) using the
following cycle: denaturation at 94 ◦ C for 30 s, annealing in the
temperature regime describedabove for 30 s andprimer extension
at 72 ◦ C for 10 s for ‘touchdown’ cycles and 72 ◦ C for 30 s for the
last five additional cycles.
SSCP analysis of PCR products was performed as described
previously. 21 Briefly, PCR products were mixed 1 : 2 with load-
Table 1. Chemical compounds in lactic-fermented fish products
www.soci.org C An et al.
ing buffer (980 mL L −1 formamide/10 mmol L −1 EDTA/5 mL L −1
bromophenol blue), denatured by heating for 10 min at 100 ◦ C,
cooled on ice, loaded on a precast, ready-to-use gel (GeneGel
Excel 12.5/24 kit, GE Healthcare, UK) and electrophoresed in a
GenePhor electrophoresis unit (GE Healthcare) at 650 V, 25 mA
and 5 ◦ C until the bromophenol blue front reached the anode
buffer strip (∼90 min). The gel was stained with a PlusOne DNA
silver staining kit (GE Healthcare). Scanned photographs of SSCP
gels were stored as TIFF images.
RESULTS AND DISCUSSION
Chemical compounds
Table 1 shows the chemical constituents of the fermented fish
products. The salinities of aji-narezushi and iwashi-nukazuke were
moderately high, about 60 and 100–160 g kg −1 respectively, while
their respective water activities were about 0.89 and 0.77. The
predominant organic acid in all samples was lactic acid, with
other organic acids present only at very low levels. The lactic acid
content was particularly high (>60 g kg −1 )inoneaji-narezushi
sample (As-1). The pH of aji-narezushi was lower than 4.3. These
results agree with our previous reports. 2,9,14,22
Viable plate count
The viable plate counts of the fermented fish products are
summarisedinTable2.Intheaji-narezushi samples, maximum
viable plate counts were 6.5–7.4 log colony-forming units (CFU)
g −1 . The viable plate count was lowered by HS. In one aji-narezushi
sample (As-1), viable cells were not detected on TS and TS-HS agar
plates. It is considered that a richer nutrient condition is required
Organic acids (g kg −1 )
Fermented Moisture Salinity Water
fish product Sample (g kg −1 ) (gkg −1 ) activity pH Lactic acid Acetic acid Total
Aji-narezushi As-1 603 58 0.87 3.89 63.7 2.3 66.1
As-2 601 57 0.89 4.28 23.7 2.5 26.2
As-3 662 62 0.90 4.27 25.7 2.2 27.8
Iwashi-nukazuke In-1 NT 126 0.78 5.08 18.6 2.1 20.7
In-2 NT 157 0.75 4.95 16.0 1.0 17.0
In-3 NT 105 NT 5.07 7.6 2.1 9.7
NT, not tested.
Table 2. Plate counts in lactic-fermented fish products
Plate count (log CFU g−1 )
Fermented
fish product Sample TS TS-HS GAM GAM-HS MRS MRS-HS PD PD-HS
Aji-narezushi As-1 ND ND 6.34 ND 6.52 3.49 3.51 3.36
As-2 7.34 5.00 6.77 5.45 6.64 3.56 3.53 3.11
As-3 6.38 6.08 6.46 5.20 6.28 4.91 4.18 3.86
Iwashi-nukazuke In-1 5.52 5.26 ND 5.62 5.68 5.59 5.64 5.45
In-2 4.60 ND 4.72 ND 6.28 6.60 4.61 6.66
In-3 3.34 5.81 3.80 5.38 ND 6.91 3.15 4.32
TS, tryptone soy agar; GAM, Gifu anaerobic medium agar; MRS, de Man, Rogosa and Sharpe agar; PD, potato dextrose agar; HS, high salt (containing
100 g L −1 NaCl); ND, not detected (

Analysis of bacterial flora of traditional Japanese fermented fish products www.soci.org
1
As-1 As-2 As-3 In-1 In-2 In-3
2
3
4
5
6
7
Figure 1. DGGE analysis of PCR-amplified 16S rDNA fragments from ajinarezushi
(As-1–As-3) and iwashi-nukazuke (In-1–In-3).
for the predominant bacteria. The viable counts for yeasts on PD
and PD-HS agar plates were 3.1–4.2 log CFU g −1 .
Maximum viable counts of the iwashi-nukazuke samples were
5–6 log CFU g −1 . The microbial count of fish nukazuke varies
depending on the manufacturer. 22 In samples In-1 and In-3 the
counts on the agar plates containing HS were similar to or higher
than those on the agar plates with no added salt. Previous reports
indicate that the predominant bacteria in fish nukazuke products
are Tetragenoccocus spp. 9 However, no micro-organisms were
detected in sample In-2 on TS-HS and GAM-HS agar plates, though
the viable counts were high on MRS-HS and PD-HS agar plates.
Bacterial flora analysed by PCR-DGGE method
For DGGE analysis we selected the V3 region of 16S rDNA as the
target region. This region has been widely used in the analysis of
bacterial communities or the identification of isolated bacteria. 11,23
PCR products originating from sample preparations were divided
into one to four main fragments by DGGE analysis (Fig. 1), with the
banding patterns differing by sample. Subsequently, to identify
the main bands, each band was recovered from the DGGE gel
and sequenced. The results obtained from clone sequencing are
shown in Table 3.
Inthecaseofaji-narezushi, there was only one main band for
sample As-1, which differed from the main bands of the other two
samples. Sequencing of the recovered DGGE gels indicated that
the predominant bacteria in sample As-1 and samples As-2 and
As-3 were Lactobacillus versmoldensis and Lactobacillus acidipiscis
respectively. Lactobacillus versmoldensis frequently dominates the
LAB populations of raw fermented sausage products, 24 while Lb.
acidipiscis is isolated from fermented fish products (pla-ra and
pla-chom) made in Thailand. 25 The difference in predominant LAB
species may be correlated with the difference in TS agar plate
counts of the samples (Table 2).
The DGGE patterns of the iwashi-nukazuke samples indicated
that the predominant bacterium in sample In-1 was Tetragenococcus
muriaticus, while the main band of sample In-2 was identified
as the chloroplast of rice (Oryza sativa). Interestingly, sample In-3
showed two main bands corresponding to those of both samples
In-1 and In-2.
8
9
10
11
12
13
14
Table 3. Identities of cloned fragments obtained from DGGE analysis
of aji-narezushi and iwashi-nukazuke
Sample DGGE no. a Identification b
Identity
(%)
Aji-narezushi
As-1 1 Lactobacillus versmoldensis 96
As-2 2 Lactobacillus acidipiscis 97
3 Lactobacillus acidipiscis 98
4 Lactobacillus acidipiscis 98
As-3 5 Lactobacillus acidipiscis 97
6 Lactobacillus acidipiscis 98
7 Lactobacillus acidipiscis 98
Iwashi-nukazuke
In-1 8 Tetragenococcus muriaticus 92
9 Tetragenococcus muriaticus 99
10 Tetragenococcus muriaticus 95
11 Oryza sativa 100
In-2 12 Oryza sativa 100
In-3 13 Tetragenococcus muriaticus 95
14 Oryza sativa 100
a See Fig. 1.
b The main bands are shown in bold type.
Tetragenococcus muriaticus, a moderately halophilic LAB, is
isolated from salted and fermented fish products along with
Tetragenococcus halophilus. 26 Although T. muriaticus is reported
to be a histamine-forming bacterium, 27 Satomi et al. 28 found that
the histidine decarboxylase gene (hdc) is encoded in the plasmid
of T. halophilus and suggested that the hdc could be encoded on
transformable elements among LAB.
In samples In-2 and In-3 a clear band of rice chloroplast was
detected. In a previous DGGE analysis of fermented plant foods
the chloroplast was detected as the main band in the early
fermentation stage. 29 Ward et al. 30 also successfully differentiated
Lactococcus lactis subsp. lactis and Lc. lactis subsp. cremoris based
on 16S rRNA sequencing, but Ercolini et al. 31 could not use the
V3 region from 16S rDNA to identify these two subspecies of
Lc. lactis. Furthermore, Walter et al. 32 also failed to distinguish
Lactobacillus casei and Lactobacillus rhamnosus using DGGE or
BLAST comparisons of V2–V3 sequences. The authors suggested
that differentiation of these species might be possible by using
primers targeting other regions of 16S rRNA. We also think
that further DGGE experiments using other regions or LABspecific
regions to clarify the LAB flora, particularly halophilic
or halotolerant LAB, of fermented fish are necessary.
Bacterial flora analysed by PCR-SSCP method
PCR-SSCP analysis enables DNA fragments of similar sizes to be
separated according to their configuration (secondarystructure). 33
Targeting the 16S rRNA V3 region, which permits phylogenetic
discrimination of microbial species, allows for LAB monitoring in
the fermented food microbial community by one profile of bands,
where each band corresponds to a different sequence of the 16S
rRNA V3 region, i.e. one bacterium. 13
As shown in the PCR-DGGE analysis, the predominant bacteria
in the fermented fish products were LAB. Therefore we isolated
bacterial strains from MRS and GAM agars for aji-narezushi and
from MRS-HS and GAM-HS agars for iwashi-nukazuke for the PCR-
J Sci Food Agric 2010; 90: 1796–1801 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1799

1800
Table 4. Results of PCR-SSCP analysis and identification of 16S rDNA
sequencing of aji-narezushi and iwashi-nukazuke
Sample Medium
No. of
isolates
Identification of 16S
rDNA sequencing
Aji-narezushi
As-1 MRS 7 Not amplified (yeasts)
MRS 2 Lactobacillus acidipiscis
MRS 1 Lactobacillus plantarum
GAM 2 Lactobacillus paralimentarius
GAM 7 Lactobacillus acidipiscis
GAM 1 Lactobacillus plantarum
As-2 MRS 8 Lactobacillus plantarum
MRS 2 Not amplified (yeasts)
GAM 3 Lactobacillus plantarum
GAM 7 Lactobacillus acidipiscis
As-3 MRS 10 Not amplified (yeasts)
GAM 2 Lactobacillus casei
GAM 7 Lactobacillus acidipiscis
GAM 1 Tetragenococcus halophilus
Iwashi-nukazuke
In-1 MRS-HS 10 Not amplified (yeasts)
GAM-HS 10 Tetragenococcus muriaticus
In-2 MRS-HS 10 Not amplified (yeasts)
GAM-HS 10 Not amplified (yeasts)
In-3 MRS-HS 10 Not amplified (yeasts)
GAM-HS 8 Tetragenococcus halophilus
GAM-HS 2 Tetragenococcus muriaticus
MRS, de Man, Rogosa and Sharpe agar; GAM, Gifu anaerobic medium
agar; HS, high salt (containing 100 g L −1 NaCl).
SSCP analysis. These agar plates were incubated under anaerobic
conditions.
As summarised in Table 4, the V3 region of some isolates from
MRS agar plates was not amplified. Although MRS agar is regarded
as a medium for LAB, the cells of the isolates were observed as
oval yeast shapes under a microscope. Particularly in the iwashinukazuke
samples, all isolates from MRS-HS agar were yeasts.
The diversity of bacterial flora was not very high in the ajinarezushi
samples. The predominant LAB isolated from all three
samples using GAM agar plates was Lb. acidipiscis. In the case
of samples As-2 and As-3, Lb. acidipiscis was also detected
as predominant by the non-culture-based PCR-DGGE analysis
(Table 3). On the other hand, Lb. versmoldensis, which was shown
to be predominant in sample As-1 by the PCR-DGGE analysis, was
not detected by the PCR-SSCP method. Kröckel et al. 24 reported
that Lb. versmoldensis grows better in MRS broth than on MRS agar
and that its colonies on MRS agar are small. Furthermore, a lag
phase of up to 4 days could be observed when it was transferred
from MRS agar to MRS broth. 24 It is considered that the growth
rate of Lb. versmoldensis is slower than that of other LAB such as Lb.
acidipiscis. These results suggest that isolation of Lb. versmoldensis
under cultivation method conditions is difficult and the population
of Lb. versmoldensis was not reflected in the viable cell counts in
Table 2.
Lactobacillus plantarum was detected in samples As-1 and As-2,
though this bacterium was not detected by the PCR-DGGE analysis.
The composition of GAM agar, which contains liver extract, may
be suitable for growth of Lb. plantarum. Lactobacillus plantarum
is isolated not only from fermented vegetables but also from
www.soci.org C An et al.
fermented meat and fish. 14 Furthermore, it is well known that
Lb. plantarum has beneficial activities in fermented foods, such
as high lactic acid production, acid tolerance and bacteriocin
production. 14 Other LAB species, Lb. casei and T. halophilus, were
isolated from sample As-3.
In iwashi-nukazuke sample In-1 the predominant bacterium was
identified as T. muriaticus, while no bacterial colony was detected
in sample In-2. These results are in agreement with those of the
PCR-DGGE analysis (Table 3). However, in the case of sample In-3
the predominant isolate was identified as T. halophillus, which was
not detected by the PCR-DGGE analysis.
As reported above, different results were obtained from the
non-culture-based PCR-DGGE method and the culture-based PCR-
SSCP method. It is considered that the PCR-DGGE method is more
useful than the PCR-SSCP method to determine the predominant
bacteria and check the growth of starter strains. However, a greater
variety of microflora was expressed in the PCR-SSCP method than
in the PCR-DGGE method. The non-bacterial (rice chloroplast)
band in the PCR-DGGE analysis may hide the bacterial band.
Therefore further study of the PCR-DGGE analysis using the V3 and
other LAB-specific regions is necessary. Furthermore, biochemical
investigation of the LAB strains isolated from aji-narezushi and
iwashi-nukazuke is now in progress.
CONCLUSIONS
We studied the bacterial flora of traditional fermented fish
products, aji-narezushi and iwashi-nukazuke, using non-culturebased
PCR-DGGE and culture-based PCR-SSCP methods. In the
PCR-DGGE analysis, Lb. acidipiscis and Lb. versmoldenis were
detected as the predominant bacteria in aji-narezushi. However,
Lb. versmoldensis could not be isolated using GAM and MRS agars.
The PCR-DGGE analysis showed that the predominant bacterium
in iwashi-nukazuke was T. muriaticus rather than T. halophilus.
Some of our results differed from those of previous studies using
cultivation methods. Further studies on the detection, isolation
and biochemical and fermentation properties of LAB, particularly
Lb. versmoldensis,inaji-narezushi are necessary.
ACKNOWLEDGEMENT
ThisstudywassupportedbyafundfromtheMinistryofAgriculture,
Forestry and Fisheries for research and development projects
promoting the new policies of Agriculture, Forestry and Fisheries
(No. 2041).
REFERENCES
1 IchishimaE, Invitation to Fermented Foods [Hakkou Shokuhin eno
Shotai]. Shokabo, Tokyo, pp. 36–39 (2004). (in Japanese).
2 KudaT,TanibeR,MoriM,TakeH,MichihataT,YanoT,et al, Microbial
and chemical properties of aji-no-susu, a traditional fermented
fish with rice product in the Noto Peninsula, Japan. Fish Sci
75:1499–1506 (2009).
3 Kobayashi T, Kimura B and Fujii T, Differentiation of Tetragenococcus
populations occurring in products and manufacturing processes
of puffer fish ovaries fermented with rice-bran. Int J Food Microbiol
56:211–218 (2000).
4 Chang CM, Ohshima T and Koizumi C, Changes in composition of
lipids, free amino acids and organic acids in rice-bran-fermented
sardine (Etrumes teres) during processing and subsequent storage.
J Sci Food Agric 59:521–528 (1992).
5 Riebroy S, Benjakul S and Viessanguan W, Properties and acceptability
of Som-fug, a Thai fermented fish mince, inoculated with lactic acid
bacteria starters. LWT – Food Sci Technol 41:569–580 (2008).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1796–1801

Analysis of bacterial flora of traditional Japanese fermented fish products www.soci.org
6 Vinderola G, Capellini B, Villarreal F, Suárez V, Quiberoni A and
Reinheimer J, Usefulness of a set of simple in vitro tests for the
screening and identification of probiotic candidate strains for dairy
use. LWT – Food Sci Technol 41:1678–1688 (2008).
7 Kilpi EER, Kahara MM, Steele JL, Philanto AM and Joutsjoki VV,
Angiotensin I converting enzyme inhibitory activity in milk
fermented by wild-type and peptidase-deletion derivatives of
Lactobacillus helveticus CNRZ32. Int Dairy J 17:976–984 (2007).
8 Komatsuzaki N, Shima J, Kawamoto S, Momose H and Kimura T,
Production of γ -aminobutyric acid (GABA) by Lactobacillus
paracasei isolated from traditional fermented foods. Food Microbiol
22:497–504 (2005).
9 Kuda T, Mihara T and Yano T, Detection of histamine and histaminerelated
bacteria in fish nukazuke, a salted and fermented fish with
rice-bran, by simple colorimetric microplate assay. Food Control
18:677–681 (2007).
10 Singh S, Goswami P, Singh R and Heller KJ, Application of molecular
identification tools for Lactobacillus, with a focus on discrimination
between closely related species. LWT – FoodSciTechnol 42:448–457
(2009).
11 Chen HC, Wang SY and Chen MJ, Microbiological study of lactic
acid bacteria in kefir grains by culture-dependent and cultureindependent
methods. Food Microbiol 25:492–501 (2008).
12 Kim TW, Lee JH, Kim SE, Park MH, Chang HC and Kim HY, Analysis of
microbial communities in doenjang, a Korean fermented soybean
paste, using nested PCR-denaturing gradient gel electrophoresis.
Int J Food Microbiol 131:265–271 (2009).
13 Chamkha M, Sayadi S, Bru V and Godon JJ, Microbial diversity in
Tunisian olive fermentation brine as evaluated by small subunit
rRNA-single strand conformation polymorphism analysis. Int J Food
Microbiol 122:211–215 (2008).
14 Kuda T, Kaneko N, Yano T and Mori M, Induction of superoxide anion
radical scavenging capacity in Japanese white radish juice and
milk by Lactobacillus plantarum isolated from aji-narezushi and
kaburazushi Food Chem 120:517–522 (2010).
15 Sheffield VC, Cox DR, Lerman LS and Myers RM, Attachment of a
40-base-pair G+C-rich sequence (GC-clamp) to genomic DNA
fragments by the polymerase chain reaction results in improved
detectionofsingle-basechanges.ProcNatlAcadSciUSA86:232–236
(1998).
16 Muyzer G, De Waal EC and Uitterlinden AG, Profiling of complex
microbial populations by denaturing gradient gel electrophoresis
analysis of polymerase chain reaction-amplified genes for 16S rRNA.
Appl Environ Microbiol 59:695–700 (1993).
17 Don RH, Cox PT, Wainwright BJ, Baker K and Mattick JS, ‘Touchdown’
PCR to circumvent spurious priming during gene amplification.
Nucleic Acids Res 19:4008 (1991).
18 Neefs JM, Hendriks Van de Peer YL and Wachter RD, Compilation
of small ribosomal subunit RNA sequences. Nucleic Acids Res
18:2237–2317 (1990).
19 Lee DH, Zo YG and Kim SJ, Nonradioactive method to study genetic
profiles of natural bacterial communities by PCR-single-strandconformation
polymorphism. Appl Environ Microbiol 62:697–703
(1996).
20 Weisburg WG, Barns SM, Pelletier DA and Lane DJ, 16S ribosomal
DNA amplification for phylogenetic study. J Bacteriol 173:697–703
(1991).
21 Takahashi H, Kimura B, Yoshikawa M, Gotou S, Watanabe I and Fujii T,
Direct detection and identification of lactic acid bacteria in a food
processing plant and in meat products using denaturing gradient
gel electrophoresis. J Food Protect 67:2515–2520 (2004).
22 Kuda T, Miyamoto H, Sakajiri M, Ando K and Yano T, Microflora of
fish nukazuke made in Ishikawa, Japan. NipponSuisanGakkaishi
67:296–301 (2001).
23 Ercolini D, Moschetti G, Blaiotta G and Coppola S, Behavior of variable
V3 region from 16S rDNA of lactic acid bacteria in denaturing
gradient gel electrophoresis. Curr Microbiol 42:199–202 (2001).
24 Kröckel L, Schillinger U, Franz CMAP, Bantleon A and Ludwig W,
Lactobacillus versmoldensis sp. nov., isolated from raw fermented
sausage. Int J Syst Evol Microbiol 53:513–517 (2003).
25 Tanasupawt S, Shida O, Okada S and Komagata K, Lactobacillus
acidipiscis sp.nov.andWeissella thailandensis sp. nov., isolated from
fermented fish in Thailand. Int J Syst Evol Microbiol 50:1479–1485
(2000).
26 Satomi M, Kimura B, Mizoi M, Sato T and Fujii T, Tetragenococcus
muriaticus sp. nov., a new moderately halophilic lactic acid
bacterium isolated from fermented squid liver sauce. Int J Syst
Bacteriol 47:832–836 (1997).
27 Kimura B, Konagaya Y and Fujii T, Histamine formation by
Tetragenococcus muriaticus, a halophilic lactic acid bacterium
isolated from fish sauce. Int J Food Microbiol 70:71–77 (2001).
28 Satomi M,Furushita M,Oikawa H,Yoshikawa-Takahashi M and Yano Y,
Analysis of a 30 kbp plasmid encoding histidine decarboxylase gene
in Tetragenococcus halophilus isolated from fish sauce. Int J Food
Microbiol 15:202–209 (2008).
29 Nakayama J, Hoshiko H, Fukuda M, Tanaka H, Sakamoto N, Tanaka S,
et al, Molecular monitoring of bacterial community structure
in long-aged nukadoko: pickling bed of fermented rice
bran dominated by slow-growing lactobacilli. J Biosci Bioeng
104:481–489 (2007).
30 Ward LJH, Brown JCS and Davey GP, Two methods for the genetic
differentiation of Lactococcus lactis ssp. lactis and cremoris based
on differences in the 16S rRNA gene sequence. FEMS Microbiol Lett
166:15–20 (1998).
31 Ercolini DP, Hill J and Dodd CER, Bacterial community structure and
location in Stilton cheese. Appl Environ Microbiol 69:3540–3548
(2003).
32 Walter J, Tannock GW, Tilsala-Timisjarvi A, Rodtong S, Loach DM,
Munro K, et al, Detection and identification of gastrointestinal
Lactobacillus species by using denaturing gradient gel
electrophoresis and species-specific PCR primers. Appl Environ
Microbiol 66:2418–2423 (2000).
33 Hebenbrock K, Williams MP and Karger LB, Single-strand conformation
polymorphism using capillary electrophoresis with
two-dye laser-induced fluorescence detection. Electrophoresis
16:1429–1436 (1995).
J Sci Food Agric 2010; 90: 1796–1801 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1801

1802
Research Article
Received: 23 February 2010 Revised: 19 April 2010 Accepted: 23 April 2010 Published online in Wiley Interscience: 14 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4017
Identification of ferulate oligomers from corn
stover
Diane Dobberstein a and Mirko Bunzel b∗
Abstract
BACKGROUND: Cross-links between plant cell wall polymers negatively impact forage digestibility. Hydroxycinnamates and
their oligomers act as cross-links between polysaccharides and/or polysaccharides and lignin. Higher ferulate oligomers such as
dehydrotrimers were identified in cereal grains but not in vegetative organs of grasses. The aim of this study was to characterize
ester-linked hydroxycinnamate oligomers from corn stover with special emphasis on ferulate dehydrotrimers.
RESULTS: With the exception of the 4-O-5-dehydrodiferulic acid all known ferulate dehydrodimers, including the recently
described 8-8(tetrahydrofuran) dimer, were identified in the alkaline hydrolyzate of corn stover after chromatographic
fractionation. Next to dehydrodimers, 18 cyclobutane dimers made up of ferulic acid and/or p-coumaric acid were identified
by GC-MS of the dimeric size exclusion chromatography fraction. Ferulate dehydrotrimers were isolated by using multiple
chromatographic procedures and identified by UV spectroscopy, MS and NMR. Four trimers were unambiguously identified
as 5-5/8-O-4-, 8-O-4/8-O-4-, 8-8(aryltetralin)/8-O-4-, and 8-O-4/8-5-dehydrotriferulic acids, a fifth tentatively as 8-5/5-5dehydrotriferulic
acid.
CONCLUSION: The formation of ferulate dehydrotrimers is not limited to reproductive organs of grasses but also contribute to
network formation in the cell walls of vegetative organs. Although radically coupled hydroxycinnamate dimers and oligomers
were in the focus of researchers over the last decade, the earlier described cyclobutane dimers significantly contribute to cell
wall cross-linking.
c○ 2010 Society of Chemical Industry
Keywords: ferulic acid; p-coumaric acid; dehydrotrimer; dehydrotriferulic acid; cyclobutane dimers; forage digestibility
INTRODUCTION
Forage digestibility and hence quality is widely controlled by
the cell walls in the plant organs. Structural and matrix polysaccharides,
lignin, and proteins are the major constituents of the
walls. Cell wall composition and interactions between cell wall
polymers mediated by, for example, polysaccharide or polysaccharide–lignin
cross-links are important factors limiting their
digestibility. 1–6 In grasses, hydroxycinnamates, particularly ferulate
and p-coumarate, are minor constituents of the cell wall. While
p-coumarates are mostly bound to lignin 7 and only to a lesser degree
to arabinoxylans, 8,9 ferulates are primarily acylating the O-5
position of arabinose side-chains in arabinoxylans. 10 Radical- and
light-induced coupling reactions of esterified ferulates lead to the
formation of dimeric ferulate cross-links between cell wall arabinoxylans.
These dimers are known as dehydrodiferulates (radical
coupling) or cyclobutane ferulate dimers (light-induced coupling).
More recently, dehydrotriferulates and dehydrotetraferulates were
isolated from corn bran. 11 – 15 Theoretically, these compounds can
cross-link up to four polysaccharide chains forming a strong
network in the cell wall. Ferulates can also cross-couple with
monolignols. 16,17 Thus, they are co-polymerized into lignins 18,19
and cross-link arabinoxylans with lignins. Although radicals can
easily be generated from p-coumarate, radically formed dimers
of p-coumarates have not been identified from plant materials.
Radical transfer reactions with other phenolics in the cell wall
are discussed to explain these findings. 20 Cyclobutane dimers
of p-coumarates and mixed cyclobutane dimers of ferulates and
p-coumarates, however, were identified in different grasses. 21 – 23
As these compounds are formed by a photochemical mechanism
the formation of the cyclobutane dimers requires sunlight during
plant growth. The formation of these compounds is therefore
supposed to vary strongly depending on the localization of the
considered organ or tissue in the plant.
While ferulate oligomers such as trimers were isolated form
corn bran they were not yet identified in vegetative organs, e.g.
stems or leaves, widely used as forages. The aim of this study
was to demonstrate that ferulate oligomers, especially ferulate
trimers, are not exclusively involved in cell wall cross-linking of
reproductive organs but also occur in vegetative organs, thus
having a potential influence on forage digestibility.
∗ Correspondence to: Mirko Bunzel, Department of Food Science and Nutrition,
University of Minnesota, 1334 Eckles Avenue, St Paul, MN 55108, USA.
E-mail: mbunzel@umn.edu
a Department of Biochemistry and Food Chemistry, University of Hamburg,
Grindelallee 117, 20146 Hamburg, Germany
b DepartmentofFoodScienceandNutrition,UniversityofMinnesota,1334Eckles
Avenue, St Paul, MN 55108, USA
J Sci Food Agric 2010; 90: 1802–1810 www.soci.org c○ 2010 Society of Chemical Industry

Ferulate oligomers from corn stover www.soci.org
mV
2500
2000
1500
1000
500
B1
B2
B3 B4
0
200 220 240 260 280 300 320 340 360 380
Figure 1. Separation of alkali-extracted hydroxycinnamate derivatives/oligomers from corn stover by size exclusion chromatography (detection at
325 nm).
EXPERIMENTAL
General
Alcalase 2.4 L (EC 3.4.21.62, from Bacillus licheniformis, 2.4 AU
g −1 ) was kindly donated by Novo Nordisk (Bagsvaerd, Denmark).
Bio-Beads S-X3 was from Bio-Rad Laboratories (Munich,
Germany), the solvent resistant BioBeads size exclusion chromatography
(SEC) glass column ECOPLUS from Kronlab (Sinsheim,
Germany). Sephadex LH-20 was from Amersham Pharmacia
Biotech (Freiburg, Germany). SEC and Sephadex LH-20 chromatography
instrumentation (L-6000 pump, L-7400 UV detector)
was from Merck/Hitachi (Darmstadt, Germany). Phenyl-hexyl
HPLC-columns were purchased from phenomenex (Aschaffenburg,
Germany). Phenyl-hexyl-RP-HPLC was carried out using
either of the following instrumentations: L-6200 intelligent pump,
T-6300 column thermostat, L-7400 UV detector or L-7150 intelligent
pump, L-7300 column oven, L-7455 photodiode array
detector (Merck/Hitachi, Darmstadt, Germany). HPLC-MS instrumentation
was from Hewlett-Packard (Waldbronn, Germany): HP
Series 1100: autosampler G1313, pump G 1312A, mass spectrometer
G 1946A, photodiode array detector 1314A. Nuclear
magnetic resonance (NMR) experiments were performed on a
Bruker DRX-500 (Rheinstetten, Germany) instrument. Gas chromatography–mass
spectroscopy (GC-MS) was carried out on a
Trace 2000 GC coupled to a PolarisQ ion trap mass spectrometer
(ThermoQuest, Thermo Scientific, Dreieich, Germany) using a
HP-5-MS fused-silica capillary column (Hewlett-Packard).
Plant material and sample preparation
Seeds (DK 233) were made available by Monsanto Agrar
Deutschland GmbH (Düsseldorf, Germany). The plant material
was provided by Professor F. Schwarz and Dr F. Zeller from the
Technical University of Munich, Department of Animal Nutrition.
The plant material was seeded on 28 April 2004 in a field trial
in Freising, Germany, and harvested on 27 September 2004. The
dry matter content of the grains at the time of harvest was
640 g kg −1 , representing a typical maturity stage used for the
production of silage. The duration of sunshine from sowing to
harvest was calculated as 1226 h. Immediately after harvesting,
the cob was removed, and the corn stover was coarsely chopped,
freeze-dried, and stepwise ground to pass a 0.5 mm screen. Dried
material (280 g) was washed five times with water (3 L each,
stirring for 10 min) followed by centrifugation. The residue was
consecutively extracted in a Soxhlet apparatus for 8 h with ethanol
and 6 h with acetone followed by a drying step. Residual material
(200 g) was suspended in a phosphate buffer pH 7.5, and proteins
min
were partially degraded by using the protease alcalase (30 µL
enzyme g −1 dry material, 30 min at 60 ◦ C, continuous agitation).
The residue was centrifuged, washed with hot water, 95% (v/v)
ethanol, and acetone, and dried.
Alkaline hydrolysis and extraction
Alkaline hydrolysis and extraction was performed according to
a previously described procedure. 12 In brief, extracted and dried
stover (165 g) was saponified (NaOH (2 mol L −1 ); 20 mL NaOH g −1
stover) under nitrogen, protected from light and under continuous
stirring for 18 h. Following acidification of the mixture (pH <
2), liberated phenolic acids were extracted into diethyl ether.
Ether extracts were extracted with NaHCO3 solution (50 g L −1 ).
After acidification (pH < 2) of the combined aqueous layers the
phenolic acids were re-extracted into diethyl ether. Ether extracts
were dried over Na2SO4, evaporated to dryness and re-dissolved
in tetrahydrofuran (10 mL).
Fractionation of phenolic acids
SEC was carried out by using Bio-Beads S-X3 (gel bed: 1.5 cm ×
95 cm) swollen in tetrahydrofuran which was also used as mobile
phase. Hydrolyzate dissolved in tetrahydrofuran (about 200 mg in
500 µL) was applied to the column. The flow rate was maintained
at 0.25 mL min −1 for 360 min, increased to 0.5 mL min −1 between
360 and 395 min and further increased to 0.75 mL min −1 until all
material was eluted. Fractions were collected according to the
chromatogram monitored at 325 nm (Fig. 1). Fractions B2 and B3
from 20 runs were pooled, dried under a stream of nitrogen, redissolved
in methanol (MeOH)/water 50/50 (v/v) (ultrasonic bath,
addition of a few drops acetone to improve solubility), and used
for Sephadex LH-20 chromatography.
Sephadex LH-20 chromatography was performed as described
previously 11 with some minor modifications. Fraction B2 (386 mg)
was separated in two Sephadex runs whereas only a portion
of fraction B3 (875 mg) was separated in a single run. In brief,
the sample was applied to the column (gel bed: 2.5 cm ×
85 cm) pre-conditioned with aqueous trifluoroacetic acid (TFA)
(0.5 mmol L −1 )/MeOH 95/5 (v/v). Elution was carried out as follows:
(1) elution with TFA (0.5 mmol L −1 )/MeOH 95/5 (v/v) for 72 h, flow
rate: 1.5 mL min −1 ; (2) elution with TFA (0.5 mmol L −1 ))/MeOH
50/50 (v/v) for 72 h, flow rate: 1.0 mL min −1 ;(3)elutionwithTFA
(0.5 mmol L −1 )/MeOH 40/60 (v/v) for 65 h, flow rate: 1.0 mL min −1 ;
(4) rinsing step with TFA (0.5 mmol L −1 )/MeOH 10/90 (v/v).
Detection was carried out at 280 and 325 nm. Fractions were
collected over 12-min periods, combined according to the
J Sci Food Agric 2010; 90: 1802–1810 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1803

1804
mV
1400
1200
1000
800
600
400
200
A
S Di 3
S Di 2
S Di 4
S Di 5
0
0 500 1000 1500 2000 2500 3000 3500
mV
300
250
200
150
100
50
C
S Tri 7
S Di 6
0
0 500 1000 1500 2000 2500 3000 3500
www.soci.org D Dobberstein, M Bunzel
mV
700 B
600
500
400
300
200
100
SDi8 SDi9 S Di 11
0
min 0 500 1000 1500 2000 2500 3000 3500 min
min
450 D
mV
400
350
300
250
200
150
100
50
S Tri 9
S Tri 10
S Tri 12
S Tri 11
S Tri 13
0
0 500 1000 1500 2000 2500 3000 3500 min
Figure 2. Separation of the size exclusion fractions B3 (top; A, B) and B2 (bottom; C, D) by using Sephadex LH-20 chromatography. (A) Elution with
0.5 mmol L −1 aqueous trifluoroacetic acid (TFA)/methanol (MeOH) (50/50, v/v), detection at 280 nm. (B) Elution with 0.5 mmol L −1 aqueous TFA/MeOH
(40/60, v/v), detection at 280 nm. (C) Elution with 0.5 mmol L −1 aqueous TFA/MeOH (50/50, v/v), detection at 325 nm. (D) Elution with 0.5 mmol L −1
aqueous TFA/MeOH (40/60, v/v), detection at 325 nm. Elution with 0.5 mmol L −1 aqueous TFA/MeOH (95/5, v/v) is not shown for both size exclusion
fractions. S Di 2–S Di 11 contained the following dehydrodiferulic acids: S Di 2: 8-8(tetrahydrofuran)-; S Di 3: 8-8(aryltetralin)-, S Di 4: 8-8-, S Di 5: 8-5-, S Di 6: 8-O-4-,
8-5(benzofuran)-, S Di 8andS Di 9: 5-5-, S Di 11: 8-5(decarboxylated)-dehydrodiferulic acid. For S Tri 7toS Tri 13 see text.
chromatogram (Fig. 2), and evaporated. Fractions S tri 7 and
S tri 9–S tri 13 were further separated by using semi-preparative
RP-HPLC.
RP-HPLC was carried out by using a semi-preparative phenylhexyl
column (250×10 mm i.d., 5 µm particle size), ternary gradient
systems made up of aqueous TFA (1 mmol L −1 ), acetonitrile (ACN),
and MeOH, and a flow rate of 2.5 mL min −1 .Injectionvolume
was 60 µL and the separation was performed at either 35 or 45 ◦ C.
Chromatograms were monitored at 280 and 325 nm. The following
gradients were used (eluent A: aqueous TFA (1 mmol L −1 ); eluent
B: ACN/aq. TFA (1 mmol L −1 ) 90/10 (v/v); eluent C: MeOH/aq. TFA
(1 mmol L −1 ) 90/10 (v/v)): Fractionation of S tri 7: initially A 60%, B
25%, C 15%, held for 15 min, linear over 5 min to A 45%, B 35%,
C 20%, held for 5 min, linear over 5 min to A 25%, B 35%, C 40%,
held for 5 min, linear over 5 min to A 0%, B 0%, C 100%, held for
5 min, following an equilibration step. Fractionation of S tri 9and
S tri 10: initially A 75%, B 10%, C 15%, held for 10 min, linear over
5 min to A 60%, B 25%, C 15%, held for 5 min, linear over 5 min
to A 50%, B 25%, C 25%, held for 5 min, linear over 5 min to A
20%, B 25%, C 55%, held for 5 min, following an equilibration step.
Fractionation of S tri 11–S tri 13: initially A 65%, B 20%, C 15%, held
for 15 min, linear over 5 min to A 50%, B 35%, C 15%, held for 5 min,
linear over 5 min to A 40%, B 35%, C 25%, linear over 5 min to A
10%, B 35%, C 55%, held for 5 min, following an equilibration step.
Fractions were collected according to the chromatograms, pooled
and evaporated. Some fractions had to be re-chromatographed to
increase purity for structural characterization.
Characterization of the fractions and isolated compounds
The UV spectra and the molecular weights were determined
by RP-HPLC coupled to both a photodiode array detector
and a mass spectrometer (atmospheric pressure–electrospray
ionization, positive and negative mode). The fragmentor voltage
was either 75 V (positive mode) or 90 V (negative mode), the
scan range m/z 100–1000. Elution on an analytical phenyl-hexyl
column (250 × 4.6 mm i.d.) was carried out by using the following
gradient (eluent A: 0.1% (v/v) formic acid; eluent B: ACN): Initially
A 82%, B 18%, linear over 5 min to A 80%, B 20%, linear over
5 min to A 75%, B 25%, linear over 5 min to A 70%, B 30%, linear
over 10 min to A 65%, B 35%, held for 5 min, linear over 5 min
to A 55%, B 45%, followed by rinsing and equilibration steps.
The column temperature was held at 35 or 45 ◦ C, the injection
volume was 20 µL. Structural identification was performed using
1 H- and HMQC-NMR experiments. Samples were dissolved in
0.7 mL acetone-d6. Chemical shifts (d) were referenced to the
central solvent signals (dH 2.04 ppm; dC 29.8 ppm). J-values were
calculated in hertz.
Screening for ester-bound cyclobutane dimers and monolignol–ferulate
cross-products by GC-MS
An aliquot of the BioBeads fraction B3 (about 0.3 mg) was
silylated in pyridine/BSTFA 1/4 (v/v) for 30 min at 60 ◦ C. The
silylated compounds were separated and detected by GC-MS.
He (1 mL min −1 ) was used as carrier gas. GC conditions were as
follows: HP-5-MS fused-silica capillary column (30 m, 0.32 mm i.d.,
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1802–1810

Ferulate oligomers from corn stover www.soci.org
Relative Abundance %
100
90
80
70
60
50
40
30
20
10
0
F-CA 1
CC ht
CC h
CCht CChh
?
CFht CCht CFht 8-8(aryltetralin)-DFA
CC hh
CF/FF ht
FF ht
FFht
FF ht
FFht FFht F-CA 2
39 40 41 42 43 44 45 46 47 48 49 50 51
Figure 3. GC-MS chromatogram of the silylated compounds of size exclusion fraction B3. CC, CF, FF: cyclobutane dimers made up of p-coumaric acid
(C) and/or ferulic acid (F); DFA: dehydrodiferulic acid; F-CA: ferulate-4-O-β-coniferyl alcohol cross-coupled structures; MEL: ferulate-8-β-coniferyl alcohol
cross-product of the monoepoxylignaloide type.
0.25 µm film thickness); initial column temperature, 150 ◦ C, held
for 1 min, ramped at 3 ◦ Cmin −1 to 250 ◦ C, ramped at 30 ◦ Cmin −1
to 300 ◦ C; held for 25 min; injector temperature 300 ◦ C; split 1/10;
ionization energy 70 eV; scan range: m/z 50–700.
RESULTS AND DISCUSSION
Isolation procedure
Liberation of hydroxycinnamate derivatives from their ester
linkages in the cell wall as well as extraction and fractionation of
these compounds according to their molecular weights followed
a procedure formerly applied to corn bran. 11 However, compared
to the SEC of hydroxycinnamate derivatives from corn bran
the SEC separation achieved here was worse, particularly the
separation between the ‘dimer’ fraction B3 (supposed to contain
mostly hydroxycinnamate dimers) from the ‘trimer’ fraction B2
(supposed to contain mostly hydroxycinnamate trimers) (Fig. 1).
As detailed below, the composition of both the dimer and the
trimer fraction is more complex for corn stover than for corn bran.
As corn stover partially contains highly lignified cell walls, crossproducts
between hydroxycinnamic acids and mono-/oligolignols
are expected to add to the complexity of the chromatogram.
Further separation of the dimer and trimer fractions B3 and B2 by
using Sephadex LH-20 chromatography resulted in more complex
chromatograms as compared to those from corn bran also (Fig. 2).
The diversity of the hydroxycinnamate derivatives extracted from
corn stover made unambiguous identification more difficult and
required additional fractionation work.
Identification of hydroxycinnamate dimers:
dehydrodiferulates
An aliquot of SEC fraction B3 was silylated and analyzed by GC-MS.
Figure 3 shows a section of the chromatogram. In contrast to
corn bran, the dimeric fraction of corn stover is highly complex
8-5-DFA
F-CA 3
?
8-O-4-DFA
8-5(benzofuran)-DFA
MEL
5-5-DFA
8-5(decarboxylated)-DFA
min
and dominated by signals other than dehydrodiferulate signals.
Comparison of retention times and mass spectra 24 with those of
authentic standard compounds 25,26 led to the identification of the
following dehydrodiferulates in alkaline corn stover hydrolyzates:
8-8(aryltetralin)-, 8-5(benzofuran)-, 8-5-, 8-5(decarboxylated),
8-O-4-, and 5-5-dehydrodiferulic acids. As is true for all alkaline
hydrolyzates, the 8-5- and 8-5(decarboxylated) dimers are presumably
formed during alkaline hydrolysis from their precursor in
the plant, the 8-5(benzofuran)-dehydrodiferulate. Due to the complexity
of the chromatogram neither the 8-8-dehydrodiferulic acid
nor the recently discovered 8-8(tetrahydrofuran)-dehydrodiferulic
acid 26 could be clearly identified by GC-MS of the SEC fraction B3.
Further fractionation of fraction B3 by Sephadex LH-20 chromatography
(Fig. 2) and analysis of these fractions by HPLC–photodiode
array detection/MS, however, led to an unambiguous identification
of the 8-8- and the 8-8(tetrahydrofuran)-dehydrodiferulic
acids in the alkaline corn stover hydrolyzate. Thus, with the exception
of the 4-O-5-dehydrodiferulic acid, 27 which is usually only
a minor dimer, all known dehydrodimers were identified in the
stover hydrolyzate.
Identification of hydroxycinnamate dimers:
cyclobutane dimers
In the late 1980s and early 1990s cyclobutane dimers were
discussed as major hydroxycinnamate cross-links in grasses.
However, the discovery of dehydrodiferulates other than the
5-5-dehydodiferulate in 1994 28 shifted interest towards the radically
coupled dehydrodimers. As the plant has more control
over the formation of radically formed dehydrodimers than over
the formation of cyclobutane dimers induced by UV light this
shift of interest is logical from a plant physiological point of
view. Also, in plant organs and tissues which are protected
from light, e.g. corn seeds protected by the husk, only low
amounts of cyclobutane dimers were detected. However, as evident
from Fig. 3, cyclobutane dimers are abundant in the analyzed
J Sci Food Agric 2010; 90: 1802–1810 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1805

1806
Table 1. Cyclobutane dimers identified from the alkaline hydrolyzate
of corn stover ∗
Cyclobutane
dimer
Retention
time (min) †
m/z
245
Ratio m/z
219/249
Ratio m/z
249/308
www.soci.org D Dobberstein, M Bunzel
Ratio m/z
293/308
m/z
661
CC ht 39.32 − – – – –
CC hh 40.29 + – – – –
CC hh 40.81 + – – – –
CC ht 41.03 − – – – –
CC ht 41.17 − – – – –
CC hh 41.25 + – – – –
CF ht 41.73 −

Ferulate oligomers from corn stover www.soci.org
mV
2500
2000
1500
1000
500
S tri 7
0
0 5 10 15 20 25 30 35 40 min
mV
250
200
150
100
50
0
mV
200
150
100
50
S tri 10
S tri 12
8-5-DFA
8-O-4/8-5-triFA
8-8(aryltetralin)/8-O-4-triFA
0 5 10 15 20 25 30 35 40 min
X 1
X 2
X 4
8-5/5-5-triFA
120
100
80
60
40
20
0
1600
1200
0
0
0 5 10 15 20 25 30 35 40 min 0 5 10 15 20 25 30 35 40 min
mV
mV
800
400
0
mV
1600
1200
800
400
S tri 9
0 5 10 15 20 25 30 35 40 min
S tri 11
S tri 13
5-5-DFA
0 5 10 15 20 25 30 35 40 min
Figure 4. Semi-preparative RP-HPLC chromatograms (detection at 325 nm) showing the fractionation of the ‘trimeric’ Sephadex LH-20 fractions (Fig. 2).
triFA: dehydrotriferulic acid.
doublets indicate one unsubstituted trans-cinnamate side-chain,
and two doublet-of-doublets indicate that only two of the three
guaiacyl units are unsubstituted whereas the third one is linked
in 5-position (absence of an 8 Hz coupling for the proton in
6-position). Comparison of the assigned proton data shows good
agreement with literature data. 13 The molecular mass of the
thirdcompoundisolatedfromS tri 7 was 578 indicating a ferulate
trimer, too. Interpreting 1 Hand 13 C (from 2D HMQC experiment)
NMR data this compound was unambiguously identified as 8-
8(aryltetralin)/8-O-4-dehydrodiferulic acid. Characteristic signals
are for example two singlets at 7.33 and 7.61 ppm indicating two
8-linked ferulate units and two doublets at 4.58 and 3.93 ppm
(2 Hz coupling) indicating the 8-8(aryltetralin) linkage. Absence of
16 Hz doublets indicated that none of the cinnamate side-chains is
unsubstituted. Comparison of the NMR data with literature data 12
fully supported this assignment. Although two more compounds
were isolated from S tri 7 (X1 and X2, Fig.4) an unambiguous
identification of these compounds failed. Here, as also true for
other fractions, solubility issues led to substance losses, reduced
sensitivity in NMR, etc.
Fraction S tri 9 contained 8-O-4/8-O-4-dehydrotriferulic acid
(Fig. 5, III). The trimeric nature of this compound was confirmed by
its molecular mass (578), and the assigned proton NMR data are in
full agreement with published data. 12 Again, two singlets 7.31 and
7.34 ppm indicate two 8-O-4-linkages, confirmed by the absence of
16 Hz doublets and the fact that three doublet-of-doublets indicate
that neither guaiacyl unit is further substituted. The second
compound (X3) isolated from this fraction was characterized as a
trimer but could not be unambiguously identified.
The only compound which could be identified from the
fractions S tri 10 and S tri 11 was the 5-5-coupled dehydrodiferulic
acid. Identification of the compounds X4 and X5 failed. The
compound isolated from S tri 12 was tentatively identified as 8-5/5-
5-dehydrotriferulic acid (Fig. 5, V). The chromatographic behavior
of this compound during the fractionation as well as the mass
spectrum and UV spectrum of this compound matches 8-5/5-
J Sci Food Agric 2010; 90: 1802–1810 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
8-O-4/8-O-4-triFA
5-5/8-O-4-triFA
X 3
X 5
1807

1808
H 3CO
H 3 CO
H
HO
OH
HO O
OH
IV
8
5
8
5
O
O
5
HO
4
OH
OH
OH
8
O
O
O
OCH 3
4
OCH 3
OCH 3
OCH 3
I
OH
O
O
www.soci.org D Dobberstein, M Bunzel
OH
OH
V
O
HO
OH
8
O
O
OCH 3
5
HO
OCH 3
H
O
H 3 CO
OCH 3
5
OH
HO O OH
OH
OH
Figure 5. Structures of the isolated and identified trimers 8-O-4/8-5-dehydrotriferulic acid (I), 8-8(aryltetralin)/8-O-4-dehydrotriferulic acid (II), 8-O-4/8-
O-4-dehydrotriferulic acid (III), and 5-5/8-O-4-dehydrotriferulic acid (IV). Compound V, 8-5/5-5-dehydrotriferulic acid, was only tentatively identified in
corn stover.
5-dehydrotriferulic acid. 11 However, due to solubility issues and
low sample amounts the final characterization by NMR was not
successful leaving this assignment tentative.
5-5/8-O-4-Dehydrotriferulic acid (Fig. 5, IV) was successfully
isolated from fraction S tri 13. UV spectrum, mass spectrum, proton
and carbon NMR data of the isolated compound perfectly match
those of 5-5/8-O-4-dehydrotriferulic acid. 14 Four doublets with
16 Hz coupling constants in the proton spectrum indicate two
intact trans-cinnamate side chains, a singlet at 7.40 ppm is
indicative for one 8-coupled side chain, and the fact that only one
doublet-of-doublets is present shows that two guaiacyl units are
linked in 5-position (demonstrated by the missing 8 Hz coupling
for the protons in the 6 position).
To date, seven dehydrotriferulates were isolated. 36 As corn
bran is (1) an excellent source for ferulates in general, 24,37 (2) only
slightly lignified, 38 and (3) obtained from the seeds which are
protected from light by the husk suppressing the formation
of cyclobutane dimers, it was chosen as ideal starting material
searching for higher radically formed ferulate oligomers. As
recently reviewed, 36 the 5-5/8-O-4-trimer was also identified
by HPLC-UV and comparison with a standard compound in
other cereal grains (wheat bran, 39 rye bran 40,41 and wild rice 41 ).
However, the identification of ferulate trimers from the vegetative
plant organs of grasses has not been performed yet. As this
material shows such a complex composition of hydroxycinnamate
derivatives, simple identification of these trimers by HPLC-UV or
HPLC coupled to a single-quadrupole MS could not be readily
achieved. By performing the more laborious but unequivocal
4
OCH 3
5
8
8
4
III
OCH 3
HO
O
H 3CO
O
8
OCH 3
O
4
OH
O
O
II
OH
OH
8
8
O
O
OH
OH
OCH 3
method of isolating these compounds throughout a series
of chromatographic steps, four of the seven known trimers
were isolated, and a fifth trimer was tentatively assigned. This
demonstrates that the formation of ferulate trimers is not limited
to the reproductive organs of the grasses but also contributes
to the formation of networks in the cell walls of vegetative
organs. Quantitative aspects are hard to judge from these
studies. The isolation or synthesis of higher milligram quantities
of ferulate trimers and higher oligomers and the development
and full validation of an HPLC/UPLC–triple-quadrupole MS or
HPLC/UPLC-quadrupole-ion trap MS methodology is required to
gain quantitative information about these compounds in different
plant tissues.
CONCLUSION
By isolation of ferulate dehydrotrimers, we demonstrated that
higher ferulates potentially contribute to the formation of
networks in the cell walls of corn stover. Although only four
trimers were unambiguously identified it can be assumed that
more trimers and even higher oligomers contribute to the
structure of the cell walls of corn stover. The complexity of the
alkaline hydrolyzate, however, prevented us from isolating more
compounds in sufficient purity to unambiguously identify their
structures. As quantification of ferulate trimers is yet not possible
the significance of these compounds for the formation of cell
wall cross-links in corn stover and for the reduction of forage
digestibility cannot be judged at this time. In addition to the
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1802–1810

Ferulate oligomers from corn stover www.soci.org
radically formed ferulate dimers and higher oligomers our studies
‘remind’ us that cyclobutane dimers have to be integrated into our
analytical procedures if the degree of polymer cross-linking is to be
correctly determined. Whereas many studies have dealt with the
quantification of radically formed dehydrodiferulates only a few
attempts were made to determine the amounts of cyclobutane
dimers in plant materials. Most studies aimed to identify these
compounds or provided a semi quantitative analysis only. More
work should be invested in determining the concentrations of
higher ferulate oligomers but also cyclobutane dimers in plant
materials by developing and applying well-validated analytical
procedures.
ACKNOWLEDGEMENTS
The authors are grateful to Professor Frieder Schwarz and Dr
Friederike Zeller for providing the plant material and excellent
collaboration. This project was funded by Monsanto Agrar
Deutschland GmbH.
REFERENCES
1 Hatfield RD, Ralph J and Grabber JH, Cell wall structural foundations:
Molecular basis for improving forage digestibilities. Crop Sci
39:27–37 (1999).
2 Grabber JH, Ralph J and Hatfield RD, Severe inhibition of maize wall
degradation by synthetic lignins formed with coniferaldehyde. JSci
Food Agric 78:81–87 (1998).
3 Grabber JH, Ralph J and Hatfield RD, Ferulate cross-links limit the
enzymatic degradation of synthetically lignified primary walls of
maize. J Agric Food Chem 46:2609–2614 (1998).
4 Grabber JH, Hatfield RD and Ralph J, Diferulate cross-links impede the
enzymatic degradation of non-lignified maize walls. J Sci Food Agric
77:193–200 (1998).
5 Jung HG and Casler MD, Maize stem tissues: cell wall concentration and
composition during development. Crop Sci 46:1793–1800 (2006).
6 Jung H, Identification of cell wall traits that can be manipulated to
improve forage digestibility, in Informational Conference with Dairy
and Forage Industries, US Dairy Forage Research Center, Madson,
WI, USA, pp. 9–14 (1996).
7 Ralph J, Hatfield RD, Quideau S, Helm RF, Grabber JH and Jung H-JG,
Pathway of p-coumaric acid incorporation into maize lignin as
revealed by NMR. JAmChemSoc116:9448–9456 (1994).
8 Ishii T and Hiroi T, Linkage of phenolic acids to cell-wall
polysaccharides of bamboo shoot. Carbohydr Res 206:297–310
(1990).
9 Allerdings E, Ralph J, Steinhart H and Bunzel M, Isolation and structural
identification of complex feruloylated heteroxylan side-chains from
maize bran. Phytochemistry 67:1276–1286 (2006).
10 Ishii T, Structure and functions of feruloylated polysaccharides. Plant
Sci 127:111–127 (1997).
11 Bunzel M, Ralph J, Brüning P and Steinhart H, Structural identification
of dehydrotriferulic and dehydrotetraferulic acids from insoluble
maize bran fiber. J Agric Food Chem 54:6409–6418 (2006).
12 Funk C, Ralph J, Steinhart H and Bunzel M, Isolation and structural
characterisation of 8–O–4/8–O–4- and 8–8/8–O–4-coupled
dehydrotriferulic acids from maize bran. Phytochemistry
66:363–371 (2005).
13 Bunzel M, Ralph J, Funk C and Steinhart H, Structural elucidation of
new ferulic acid-containing phenolic dimers and trimers isolated
from maize bran. Tetrahedron Lett 46:5845–5850 (2005).
14 Bunzel M, Ralph J, Funk C and Steinhart H, Isolation and identification
of a ferulic acid dehydrotrimer from saponified maize bran insoluble
fiber. Eur Food Res Technol 217:128–133 (2003).
15 Rouau X, Cheynier V, Surget A, Gloux D, Barron C, Meudec E, et al,
A dehydrotrimer of ferulic acid from maize bran. Phytochemistry
63:899–903 (2003).
16 Jacquet G, Pollet B, Lapierre C, Mhamdi F and Rolando C, New etherlinked
ferulic acid–coniferyl alcohol dimers identified in grass
straws. J Agric Food Chem 43:2746–2751 (1995).
17 Bunzel M, Ralph J, Lu F, Hatfield RD and Steinhart H, Lignins and
ferulate–coniferyl alcohol cross-coupling products in cereal grains.
J Agric Food Chem 52:6496–6502 (2004).
18 Ralph J, Grabber JH and Hatfield RD, Lignin–ferulate cross-links in
grasses: active incorporation of ferulate polysaccharide esters into
ryegrass lignins. Carbohydr Res 275:167–178 (1995).
19 Quideau S and Ralph J, Lignin–ferulate cross-links in grasses. Part
4. Incorporation of 5–5-coupled dehydrodiferulate into synthetic
lignin. J Chem Soc Perkin Trans 1:2351–2358 (1997).
20 Hatfield RD, Ralph J and Grabber JH, A potential role for sinapyl pcoumarate
as a radical transfer mechanism in grass lignin formation.
Planta 228:919–928 (2008).
21 Hartley RD, Morrison WH, Balza F and Towers GHN, Substituted
truxillic and truxinic acids in cell walls of Cynodon dactylon.
Phytochemistry 29:3699–3703 (1990).
22 Hartley RD, Morrison WH, Borneman WS, Rigsby LL, Oneill M,
Hanna WW, et al, Phenolic constituents of cell-wall types of normal
and brown midrib mutants of pearl-millet (Pennisetumglaucum (L) R
Br) in relation to wall biodegradability. J Sci Food Agric 59:211–216
(1992).
23 Ford CW and Hartley RD, Cyclodimers of p-coumaric and ferulic acids
in the cell walls of tropical grasses. J Sci Food Agric 50:29–43 (1990).
24 Bunzel M, Ralph J, Marita JM, Hatfield RD and Steinhart H, Diferulates
as structural components in soluble and insoluble cereal dietary
fibre. J Sci Food Agric 81:653–660 (2001).
25 Bunzel M, Funk C and Steinhart H, Semipreparative isolation of
dehydrodiferulic and dehydrotriferulic acids as standard substances
from maize bran. JSepSci27:1080–1086 (2004).
26 Schatz PF, Ralph J, Lu F, Guzei IA and Bunzel M, Synthesis
and identification of 2,5-bis-(4-hydroxy-3-methoxyphenyl)tetrahydrofuran-3,4-dicarboxylic
acid, an unanticipated ferulate
8-8-coupling product acylating cereal plant cell walls. Org Biomol
Chem 4:2801–2806 (2006).
27 Bunzel M, Ralph J, Marita J and Steinhart H, Identification of 4-O-5 ′ -
coupled diferulic acid from insoluble cereal fiber. J Agric Food Chem
48:3166–3169 (2000).
28 Ralph J, Quideau S, Grabber JH and Hatfield RD, Identification and
synthesis of new ferulic acid dehydrodimers present in grass cellwalls.
J Chem Soc Perkin Trans 1:3485–3498 (1994).
29 Packert M and Steinhart H, Separation and identification of some
monomeric and dimeric phenolic acids by a simple gas
chromatographic method using a capillary column and FID-MSD.
J Chromatogr Sci 33:631–639 (1995).
30 Ford CW and Hartley RD, GC/MS characterisation of cyclodimers from
p-coumaric and ferulic acids by photodimerisation – a possible
factor influencing cell wall biodegradability. J Sci Food Agric
46:301–310 (1989).
31 Stewart D, Robertson GW and Morrison IM, Identification of
cyclobutane-type dimers of substituted cinnamic-acids by
gas-chromatography mass-spectrometry. Rapid Commun Mass
Spectrom 6:46–53 (1992).
32 Morrison IM, Robertson GW, Stewart D and Wightman F, Determination
and characterization of cyclodimers of naturally occuring
phenolic acids. Phytochemistry 30:2007–2011 (1991).
33 Hartley RD, Whatley FR and Harris PJ, 4,4 ′ -Dihydroxytruxillic acid as
a component of cell walls of Lolium multiflorum. Phytochemistry
27:349–351 (1988).
34 Packert M, Analytik und Bedeutung gebundener aromatischer
Carbonsäuren der Nahrungsfaser aus Getreide und anderen
Nutzpflanzen, PhD thesis. Institut für Biochemie und
Lebensmittelchemie, Universität Hamburg, Hamburg (1993).
35 Grabber JH, Ralph J and Hatfield RD, Model studies of
ferulate–coniferyl alcohol cross-product formation in primary
maize walls: Implications for lignification in grasses. J Agric Food
Chem 50:6008–6016 (2002).
36 Bunzel M, Chemistry and occurence of hydroxycinnamate oligomers.
Phytochem Rev 9:47–64 (2010).
37 Hatfield RD, Ralph J and Grabber JH, Cell wall cross-linking by ferulates
and diferulates in grasses. J Sci Food Agric 79:403–407 (1999).
38 Lapierre C, Pollet B, Ralet MC and Saulnier L, The phenolic fraction
of maize bran: evidence for lignin–heteroxylan association.
Phytochemistry 57:765–772 (2001).
39 Antoine C, Peyron S, Lullien-Pellerin V, Abecassis J and Rouau X, Wheat
bran tissue fractionation using biochemical markers. J Cereal Sci
39:387–393 (2004).
J Sci Food Agric 2010; 90: 1802–1810 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1809

1810
40 Cyran MR and Saulnier L, Cell wall fractions isolated from outer layers
of rye grain by sequential treatment with alpha-amylase and
proteinase: Structural investigation of polymers in two ryes with
contrasting breadmaking quality. J Agric Food Chem 53:9213–9224
(2005).
www.soci.org D Dobberstein, M Bunzel
41 Dobberstein D, Steinhart H and Bunzel M, Bestimmung mono-, di- und
trimerer Phenolcarbonsäuren pflanzlicher Herkunft (Determination
of mono-, di- and trimeric phenolic acids from plant origin), Abstract
written in German. Lebensmittelchemie 60:101–102 (2006).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1802–1810

Research Article
Received: 16 December 2009 Revised: 13 April 2010 Accepted: 26 April 2010 Published online in Wiley Interscience: 10 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4019
A modified spectrophotometric assay
to estimate deglycosylation of steroidal
saponin to sapogenin by mixed ruminal
microbes
Yuxi Wang ∗ and Tim A McAllister
Abstract
BACKGROUND: The lack of a method for measuring deglycosylation of saponins in ruminal fluid has limited the ability to
investigate the impact of these compounds on rumen microorganisms. A simple spectrophotometric assay was adapted and a
protocol developed to enable measurement of steroidal saponin and sapogenin in ruminal fluid. The procedure was used for
in vitro determination of deglycosylation activity of rumen bacteria obtained from cattle fed or not fed Yucca schidigera saponin,
and to determine the relative deglycosylase activities of extracellular and cell-associated enzymes from ruminal content.
RESULTS: Modifications to the spectrophotometric assay (i.e. heating time shortened to 10 min and 0.5 mL dH2O added to
the reaction mixture) improved the stability of the optical density (425 nm) of the chromophore for up to 24 h post-reaction.
Centrifugation (12 000 × g, 20 min) enabled differential estimations of steroidal saponin and sapogenin in ruminal fluid.
Steroidal saponin added to defaunated ruminal fluid (dRF) or clarified ruminal fluid (cRF) was recovered completely from the
mixture as saponin + sapogenin (99.1% and 100.6%, respectively), whereas saponin recovery from the supernatant of dRF
was greatly reduced (P < 0.001) compared to that from supernatant of cRF (58.5 vs. 98.7%). Saponin recoveries from the
supernatants of dRF and cRF did not differ between donor cattle fed or not fed Yucca schidigera saponin (59.2 vs. 57.3%
and 98.4 vs. 99.3%, respectively). The majority (89–90%) of saponin added to a ruminal extracellular enzyme preparation
was recoverable in supernatant after 24 h, compared with only 26–32% remaining in supernatant from incubation with a
cell-associated enzymes fraction.
CONCLUSION: Mixed rumen bacteria deglycosylate steroidal saponin to sapogenin, at activity levels unaffected by prior
exposure to saponin, but they were unable to degrade the sapogenin core structure. Deglycosylation activity occurred primarily
in the cell-associated enzyme fraction of ruminal content.
Copyright c○ 2010 Crown in the right of Canada. Published by John Wiley & Sons, Ltd
Keywords: assay; deglycosylation; enzymes; rumen fluid; steroidal saponins
INTRODUCTION
Steroidal saponins are present in a wide variety of monocotyledonous
plants and less commonly in dicots and have been
found to alter rumen fermentation 1–3 through direct effects on
rumen microorganisms, 4–6 leading to responses that may improve
ruminant productivity. 7–9 In response to the total ban on
antibiotic growth promoters imposed by the European Union
and increasing pressure on livestock industries worldwide to restrict
sub-therapeutic use of antibiotics as growth promoters,
researchers have been investigating potential alternative feed
additives, among which are steroidal saponins.
Although steroidal saponins may exert a desirable influence
on ruminal fermentation, there is evidence that these benefits
may be transient, attributable to adaptation by rumen microbial
populations. 10,11 In general, two factors may contribute to rumen
microbes adapting to saponins: development of tolerance or
reduced sensitivity to the saponins’ antimicrobial properties,
or direct microbial deactivation of the saponin via enzymatic
degradation. Defining the relative importance of these two factors
has been hampered by the lack of a suitable method for measuring
degradation of steroidal saponins in ruminal fluid. Baccou et al. 12
described a spectrophotometric assay that accurately measured
steroidal saponin and sapogenin as total steroidal sapogenin, via
production of a chromophore generated from a reaction involving
the aglycone core of saponin/sapogenin, p-anisaldehyde and
sulfuric acid in ethyl acetate. In a preliminary study, however, we
determined that non-specific reactions occurring in the reaction
mixture generated an intense, unstable background color that
gave rise to substantial variation in the estimation of steroidal
saponin and sapogenin in ruminal fluid.
Several studies have shown that saponins are deglycosylated
in the rumen to form free sapogenins. 5,13,14 Segal and Schlösser 15
∗ Correspondence to: Yuxi Wang, AAFC Research Centre, P.O. Box 3000,
Lethbridge, Alberta, T1J 4B1, Canada. E-mail: yuxi.wang@agr.gc.ca
Agriculture and Agri-Food Canada Research Centre, P.O. Box 3000, Lethbridge,
Alberta, T1J 4B1, Canada
J Sci Food Agric 2010; 90: 1811–1818 www.soci.org Copyright c○ 2010 Crown in the right of Canada.
Published by John Wiley & Sons, Ltd
1811

1812
proposed that destabilization of cell membranes by saponins
requires their deglycosylation in the immediate vicinity of the
membrane. In agreement, we observed that purified steroidal
sapogenin had no effect on ruminal cellulolytic bacteria (Wang
et al., unpublished data). Similarly, others have found that once
deglycosylated, saponins no longer exhibit their antifungal
properties. 16,17 Thus, further definition of the mechanisms of
microbial metabolism of saponins in the rumen could provide
valuable information with regard to their potential to favorably
manipulate ruminal fermentation.
This study was conducted with the objective of modifying
the spectrophotometric method of Baccou et al. 12 to enable
measurement, in ruminal fluid, of steroidal saponin and of
sapogenin, its deglycosylated, insoluble form, and to use the
modified procedure to assess the deglycosylase activity of ruminal
microbes by comparing the activities in cell-associated and extracellular
enzyme fractions.
MATERIALS AND METHODS
This study used smilagenin (minimum 98%; Sigma Chemical Co.,
St Louis, MO, USA) and steroidal saponin extracted from Yucca
schidigera (Desert King International, San Diego, CA USA) as
model steroidal sapogenins and steroidal saponins, respectively.
Smilagenin was dissolved in absolute ethanol (50 mg 100 mL −1 )
for use in all assays. Baccou et al. 12 reported that all steroidal
sapogenins and/or saponins tested including smilagenin and
saponins containing smilagenin, produced similar chromophores
(max. absorption at 430 nm) upon reaction with p-anisaldehyde
and sulfuric acid in ethyl acetate. Thus, steroidal saponin and/or
sapogenin measured in this study were expressed as smilagenin
equivalents (SE).
Extraction of steroidal saponins from Yucca schidigera
Powdered Yucca schidigera (YS) plant material was used as a
source of saponins in this study. To remove fat and endogenous
smilagenin from the dry powder, 60 g were combined with 200 mL
of petroleum ether (Sigma Chemical Co.) and mixed continuously
for 2 h in a sealed container at room temperature (21 ◦ C), then
filtered through Waterman No. 1 filter paper. The residue was
washed with 100 mL of petroleum ether, filtered again and ether
in the residue was allowed to volatilize under continuous air flow in
a fume hood. The ether-extracted residue was mixed with 200 mL
of dH2O at room temperature for 60 min prior to being filtered
through Waterman No. 1 filter paper. The residue was washed
with 100 mL of dH2O and filtered two more times and filtrates
were combined. To remove tannins and phenolic compounds, the
combined filtrate was mixed with 5 g of polyvinyl-polypyrolidone
(GAF Materials Corporation, Wayne, NJ, USA). After centrifugation
at 5000 × g (10 min, 4 ◦ C), the supernatant was lyophilized and the
dried residue was dissolved in 100 mL of dH2O and centrifuged
(10 000 × g,20min,4 ◦ C). To capture saponins from the resulting
supernatant (100 mL), it was mixed with 200 mL of n-butanol
(Sigma Chemical Co.) and 0.2 mL of concentrated HCl and held at
room temperature for 30 min, then centrifuged (10 000×g,10min,
4 ◦ C). The butanol fraction was collected and the aqueous fraction
was subjected to a second n-butanol extraction. Butanol fractions
were combined and rotary evaporated at 50 ◦ C to dryness. The
residue was dissolved in 100 mL of dH2O, centrifuged (12 000 × g,
20 min, 4 ◦ C), and the supernatant was freeze-dried. The dried YS
saponin extract was stored in a sealed amber container at 0 ◦ C.
www.soci.org Y Wang, TA McAllister
High-performance thin-layer chromatography (HPTLC) of the
extract prepared as described above confirmed that it produced
no band and that the acid-hydrolysed (deglycosylated) extract
produced a band corresponding to steroidal sapogenin (Wang
et al., unpublished data). Using the modified spectrophotometric
method described below, the steroidal saponin concentration in
the YS extract was determined as 242.5 mg SE 100 g −1 .
Experiment1:Modificationofthespectrophotometricmethod
for application in ruminal fluid
This study was based on the spectrophotometric method as
described by Baccou et al. 12 In that method, a dry sample
containing saponin or sapogenin is dissolved in 2 mL of ethyl
acetate, to which is added 1 mL of 0.5% (v/v) anisaldehyde in
ethyl acetate and then, after mixing, 1 mL of 50% (v/v) H2SO4
in ethyl acetate. Reaction mixtures are then incubated at 60 ◦ C
for 20 min for color development. Saponin present in samples
is deglycosylated via acid hydrolysis, such that chromophore
development arises from total saponin+sapogenin in a sample.
Preliminary studies indicated to us that the Baccou et al. 12 assay
was unsuitable for analysis of steroidal saponin and sapogenin in
ruminal fluid because of interference from an intense background
chromophore. Further investigation showed that reducing the
reaction time at 60 ◦ C from 20 min to 10 min did not affect
the density of chromophores formed, as judged by the optical
density of the solution at 425 nm (OD425 values), and also that
adding 0.5 mL of dH2O to the post-reaction solution (i.e. after a
10minincubationat60 ◦ C)significantlyincreased the constancyof
the reaction mixture OD425. These conditions (10 min incubation;
post-reaction addition of dH2O) were used subsequently in testing
saponin analysis in defined solvent vs. ruminal fluid.
Preparation of samples for analysis from solution in dH2Oorruminal
content
Ruminal fluids collected from two steers fed a 40 : 60 barley
grain : barley silage diet were strained through four layers of
cheesecloth, combined in equal portions and then centrifuged
(10 000 × g,20min,4 ◦ C), and the supernatant (denoted ‘partially
clarified ruminal fluid’, pcRF) was used as a test solvent. The ruminal
fluid donors used in this study were cared for according to the
standards of the Canadian Council on Animal Care. 18
Smilagenin (as a model sapogenin) and YS saponin extracted
as described above were each assayed following suspension in
pcRF to assess the usefulness of the protocol improvements in a
ruminal application. Aqueous solutions of these compounds were
included for comparison. Smilagenin–ethanol solution (50 mg SE
100 mL −1 ) was combined with four volumes of either pcRF or
dH2O. In 10-mL plastic tubes, 5-mL quantities of the suspension
were frozen and lyophilised, and the residues were re-suspended
into 2.0 mL of methanol. Extracted YS saponins were dissolved
(125 µg SEmL −1 )indH2O or pcRF, and 4.0-mL quantities were
frozen, lyophilised and the residues were re-suspended in 2.0 mL of
methanol. Four replicates of each solution were prepared. Samples
were centrifuged (1000 × g; 10 min) to remove particulates prior
to assay.
Assay and test of post-reaction addition of dH2O. Aliquots of the
clear methanol solutions as prepared above were transferred
into 20-mL glass test tubes to produce duplicate series of 0,
5, 10, 20, 30 and 40 µg of steroidal saponin or smilagenin (to
model sapogenin) per tube. Tubes were placed in a 60 ◦ Cwater
www.interscience.wiley.com/jsfa Copyright c○ 2010 Crown in the right of Canada. J Sci Food Agric 2010; 90: 1811–1818
Published by John Wiley & Sons, Ltd

Assay for saponin deglycosylation www.soci.org
bath to evaporate methanol, after which 2.0 mL of ethyl acetate
(OmniSolv; EMD Chemicals Inc., Gibstown, NJ, USA), 1.0 mL of 0.5%
(v/v) p-anisaldehyde (Sigma Chemical Co.) in ethyl acetate, and
1.0 mL of 50% (v/v) sulfuric acid in ethyl acetate were added to
each. After thorough mixing (vortexing for 30 s), each tube was
returned to the 60 ◦ C water bath. After a 10 min incubation, each
tube was placed in a cold tap water bath. An aliquot of 0.5 mL of
dH2O was added to one of each pair of duplicates, with the other
left as it was, to determine the effect of post-reaction addition of
dH2O ontheOD425 reading of the reaction solution. A Stasar III
spectrophotometer (Gilford Instrument Laboratories Inc., Oberlin,
OH, USA) was used to measure OD425 of each reaction solution
immediately after mixing, and again at 0.5, 2, 6 and 24 h after the
first measurement.
Determining the effects of centrifugation to separate steroidal
saponin from sapogenin
Ruminal fluid collected from the same steers as in Experiment
1 was strained through cheesecloth and the filtrate was then
centrifuged (20 000 × g; 20min;4 ◦ C) to yield clarified ruminal
fluid (cRF). Aliquots of smilagenin–ethanol or YS saponin–H2O
solution were added to 15.0-mL tubes containing 7.0 mL of
cRF to yield quaduplicate series of smilagenin at 250, 500,
750 and 1000 µgmL −1 and of saponin at 450 or 500 µg SE
mL −1 . From each of the quadruplicate tubes, four 1-mL subsamples
were transferred into 2.0-mL vials. Two of these were
transferred directly to a freezer set at −20 ◦ C, and the other
two were centrifuged (12 000 × g, 20 min). The pellets were
washed twice with 250-µL quantities of dH2O, collected each
time by re-centrifugation, and then frozen. For each sub-sample,
the supernatants from each wash were combined and frozen. All of
the frozen samples (non-centrifuged sub-samples, and 12 000 × g
pellets and supernatants) were subsequently lyophilised and the
residues were re-suspended in 1.0 mL of methanol immediately
prior to analysis.
All of the methanol solutions were centrifuged briefly (10 min;
1000×g) and aliquots were assayed using the modified procedure
described above, to determine the recoveries of saponin and
sapogenin (smilagenin) from non-centrifuged cRF suspensions
and from supernatant.
Experiment 2: Determination of saponin deglycosylation
by mixed ruminal microbes
Ruminal fluid was obtained from six Angus heifers maintained
on a diet comprising 61 : 39 barley grain : alfalfa silage (w/w, dry
matter basis) and supplemented with a dry powdered preparation
of YS at 0, 20 or 60 g heifer −1 day −1 foratleast14dayspriorto
this ruminal fluid collection. For experimentation, fluid from each
of the two heifers per treatment was strained through four layers
of cheesecloth and combined in equal volumes. Each sample
was centrifuged (500 × g; 10min;25 ◦ C) to remove protozoa.
The supernatant, denoted ‘defaunated ruminal fluid’ (dRF), was
divided. One portion was centrifuged again (20 000 × g; 20min,
4 ◦ C) whilst the other portion was held at 39 ◦ C under anaerobic
conditions. This second supernatant was retained and referred to
as ‘clarified ruminal fluid’ (cRF) and was kept at 39 ◦ C prior to use.
Eight 1.0-mL aliquots of each type (dietary treatment) of dRF
and cRF were dispensed into 2.0-mL vials, followed by 0.5 mL of YS
saponin solution (1150 µg SEmL −1 in mineral buffer 19 and 40 µL
of reducing solution (5.7% (w/v) Na2S·9H2Oin0.1molL −1 NaOH).
All vials and YS saponin solutions were equilibrated to 39 ◦ Cin
a water bath prior to use and all transfers were done under a
streamofCO2. The vials were sealed and incubated anaerobically
at 39 ◦ C for 4 h. Four of the eight vials for each treatment
were transferred directly to −20 ◦ C storage (for subsequent
lyophilisation). The other four vials were centrifuged (12 000 × g,
20 min) after the incubation and 1.0 mL of supernatant from each
vial was frozen for subsequent lyophilisation. As described above,
each of the lyophilised residues was re-suspended in 1.0 mL of
methanol and duplicate aliquots were assayed using the modified
procedure. The difference between the concentration of total
saponin+sapogenin determined in the whole mixture (lyophilized
without centrifugation) and the concentration of soluble saponin
remaining in the supernatant was considered to represent the
portion of the saponin that had been deglycosylated to sapogenin.
Experiment 3: Localizing enzymatic deglycosylation of
steroidal saponin to extracellular or cell-associated enzyme
rumen microbial enzyme fractions
Ruminal fluid collected from the two donor steers in Experiment 1
was processed as described above to produce 2.0 L of dRF, all of
which was then centrifuged further to produce cRF. The cRF was
then divided into two portions. One portion was used directly,
and considered as the extracellular enzyme fraction (ECE). The
second portion of cRF was autoclaved (115 ◦ C; 15 min) to inactivate
enzymes. While this autoclaved cRF was cooling, one half of the
pelleted material remaining from conversion of dRF to cRF was
washed twice with 500 mL of 0.5 mmol L −1 phosphate buffer (pH
7.0), re-harvested each time by centrifugation (20 000 × g;20min;
4 ◦ C). This washed pellet was then re-suspended in the cooled,
autoclaved cRF and sonicated (three 30-s pulses separated by
15-s intervals; output 8, duty cycle 60–70%) using a Vibra-Cell
processor (Sonics & Materials, Inc., Newtown, CT, USA) to disrupt
bacterial cells. The resulting preparation was considered as the
cell-associated enzyme fraction (CAE).
The ECE and CAE fractions were dispensed (19-mL aliquots)
into 45-mL serum vials, followed by 1.0 mL of mineral buffer 19
or stock solutions of YS saponin in mineral buffer, which yielded
final concentrations of 0, 240 or 480 µg SEmL −1 . Quadruplicate
vials of each saponin concentration in each enzyme fraction
were prepared. Sodium azide aqueous solution (0.1 mL) was also
added to each vial to inhibit microbial growth (final concentration
100 µgml −1 ). The serum vials were then sealed and incubated
at 39 ◦ C with shaking (120 rpm). After 24 h, four 1.0-mL subsamples
were withdrawn from each vial. As described above, two
of these were immediately frozen, and the remaining two were
centrifuged (12 000 × g, 20 min). The supernatants were collected
and the pellets washed twice with 200 µL ofdH2O followed by
centrifugation (12 000 × g, 20 min). Supernatants for each subsample
were combined, and these and the pellets from each
sample were frozen for subsequent lyophilisation. Each residue
was re-suspended in 1.0 mL of methanol for analysis.
The saponin remaining in the supernatant (liquid) fraction
of the mixtures that were centrifuged was assumed to be
steroidal saponin (soluble) whereas that detected in the pellet
(solid) fraction after centrifugation was assumed to be the
deglycosylated, insoluble sapogenin. The presence of sapogenin
in the mixture prior to centrifugation and in supernatants and
pellets after centrifugation was assessed by HPTLC, using the
method described by Muzquiz et al. 20
J Sci Food Agric 2010; 90: 1811–1818 Copyright c○ 2010 Crown in the right of Canada. www.interscience.wiley.com/jsfa
Published by John Wiley & Sons, Ltd
1813

1814
Calculations and statistical analyses
In Experiment 1, OD425 readings, which reflected chromophores
formed during reaction among sapogenin, p-anisaldehyde and
sulfuric acid, were compared over a 24-h period between dH2Oand
RF-based suspensions, and with and without dH2O added to
the post-reaction mixture. In Experiments 2 and 3, recoveries of
steroidal saponin were calculated on the basis of the amounts of
saponin/sapogenin detected in various fractions as a proportion
of the amount of saponin present initially. All data were analyzed
statistically by analysis of variance using the MIXED procedure
of SAS, 21 with individual vial or tube as a random factor. The
significance of differences among treatments was tested using
LSMEANS with the PDIFF option. 21
RESULTS
Experiment 1
Addition of dH2O to the post-reaction mixture after a 10-min
incubation in the water bath reduced (P < 0.01) interference from
background color and improved (P < 0.01) the stability of the
OD425 readings over 24-h period (Fig. 1). With no water added,
OD425 readings increased (P < 0.01) between 0.5 and 24 h postreaction,
whereas if 0.5 mL of dH2O had been added, the OD425
readings remained essentially unchanged over this period. This
response was observed both with smilagenin (Fig. 1A) and with
OD 425
OD 425
A 2.4
2.0
1.6
1.2
0.8
0.4
0.0
2.8
2.4
2.0
1.6
1.2
0.8
0.4
Recovered from dH 2 O
Recovered from pcRF
No H 2O added
0.0
0 8 16 24 32 40
Smilagenin added (µg mL-1 )
www.soci.org Y Wang, TA McAllister
YS saponin (Fig. 1B), irrespective of their having been recovered
from suspension in dH2OorinpcRF.
The utility of centrifugation as a means of separating sapogenin
from saponin was confirmed. After saponin or sapogenin
suspensions in cRF were centrifuged, essentially all saponin measured
in the whole (uncentrifuged) mixture remained detectable
in the supernatant, whereas virtually no sapogenin (smilagenin
equivalents) was present in the supernatant (Table 1). Nearly all
(>99%) of the saponin and sapogenin added to cRF was recoverable
from the non-centrifuged mixture. Rates of recovery from
cRF were similar across steroidal sapogenin concentrations of
250–1000 µgmL −1 , and between saponin concentrations of 450
and 500 µgSEmL −1 .
Experiment 2
Patterns of recovery of YS saponins added to dRF or cRF were
unaffected by the ruminal fluid donor heifers having been fed
powdered Y. schidigera (0 vs. 20 or 60 g day −1 ; Table 2). When
the mixtures were assayed without centrifugation, all of the YS
saponin (99.1 to 100.6%) added to dRF or cRF was detected
as steroidal saponin+sapogenin. In contrast, measurement of
steroidal saponin in the supernatant fraction of the incubations
was greatly reduced (P < 0.01) in dRF (58.5% recovered), as
compared with cRF (98.7% recovered).
Recovered from dH 2 O
Recovered from pcRF
H 2 O added post-reaction
0.5 h
2.0 h
6.0 h
24 h
0 8 16 24 32 40
Smilagenin added (µg mL-1 )
Figure 1. Stabilization of chromophore development (OD425) during colorimetric determination of (A) smilagenin or (B) saponin from Y. schidigera that
had been recovered from solution in dH2O or in partially clarified ruminal fluid (pcRF). Values shown are means of quadruplicate determinations. Bars
representing standard error fell within symbols.
www.interscience.wiley.com/jsfa Copyright c○ 2010 Crown in the right of Canada. J Sci Food Agric 2010; 90: 1811–1818
Published by John Wiley & Sons, Ltd

Assay for saponin deglycosylation www.soci.org
B
OD 425
OD 425
2.4
2.0
1.6
1.2
0.8
0.4
0.0
2.4
2.0
1.6
1.2
0.8
0.4
Figure 1. (Continued).
Recovered from dH 2O
Recovered from pcRF
No H 2O added
0.0
0 8 16 24 32 40
Saponin added (µg SE mL-1 )
Table 1. Recoveries (%) of Y. schidigera steroidal saponin or of
smilagenin (as a model sapogenin) from solution/suspension in
clarified ruminal fluid a
Substance
added
Original
concentration
(µgmL−1 )
Fraction assayed b
Whole
mixture Supernatant
Extracted saponins YS 450 99.69 ± 0.83 99.38 ± 0.51
500 100.26 ± 0.42 99.57 ± 0.54
Smilagenin 250 99.81 ± 0.27 ND
500 99.85 ± 0.32 ND
750 101.26 ± 0.30 ND
1000 99.95 ± 0.24 0.01 ± 0.02
a Clarified ruminal fluid (cRF) was the supernatant obtained by
centrifuging cheesecloth-filtered ruminal fluid at 20 000 × g for 20 min
at 4 ◦ C.
b Assays (n = 4) were conducted either on the entire solutions/suspensions
in cRF (whole mixture) or on supernatant produced
by centrifugation of the mixtures at 12 000 × g for 20 min at 4 ◦ C.
ND, not detected.
Experiment 3
Consistent with observations from Experiment 2, virtually all of
the added YS saponin was recovered as total saponin+sapogenin
from incubations in ECE or in CAE, when the 24-h incubation
mixtures were assayed whole (Table 3). When vial contents were
Recovered from dH 2 O
Recovered from pcRF
H 2O added post-reaction
0.5 h
2.0 h
6.0 h
24 h
0 8 16 24 32 40
Saponin added (µg SE mL-1 )
centrifuged, however, recovery of saponin from the supernatant
and pellet fractions of ECE differed substantially (P < 0.001)
from the recoveries in fractionated CAE. After 24 h of incubation,
>88% of the saponin added to ECE remained detectable in the
supernatant fraction (i.e. as saponin) whereas with CAE, ≤32%
of the added saponin was detected in supernatant. Accordingly,
much less (P < 0.001) of the added saponin was detected in
the pellet fraction with ECE than with CAE (average 6.45% vs.
79.5%, respectively). Both in ECE and CAE suspensions, additive
recoveries (i.e. recovery in supernatant + recovery in pellet) were
similar (P > 0.05) to the total saponin+sapogenin analyzed in
whole mixtures. In CAE suspensions, with saponin added at 240
or at 480 µg SEmL −1 ), both calculated and analysed recoveries
exceeded 100%, and they were higher (P < 0.05) than recoveries
from incubations with ECE.
No band corresponding to steroidal sapogenin from the
extracted YS saponin was observed during HPTLC analysis of
ECE incubations at 0 or 24 h. This was also the case with CAE
incubation at 0 h (Fig. 2), but the characteristic yellow band was
clearly evident in the analysis of CAE incubation after 24 h, and it
was detected mainly in the pellet fraction, rather than supernatant
(data not shown).
DISCUSSION
Effect of adding dH2O to the post-reaction mixture
The chromophore measured in this assay develops from the
reaction of steroidal sapogenin, p-anisaldehyde and sulfuric acid
J Sci Food Agric 2010; 90: 1811–1818 Copyright c○ 2010 Crown in the right of Canada. www.interscience.wiley.com/jsfa
Published by John Wiley & Sons, Ltd
1815

1816
Table 2. Recoveries (%) of steroidal saponin suspended in ruminal
fluid fractions and incubated anaerobically for 4 h at 39 ◦ C(n = 4)
Mixture incubated
Donor
animal
dietb Whole
mixture
Material assayed a
www.soci.org Y Wang, TA McAllister
Mixture
supernatant
YS saponins in dRF c YS-0 98.1 ± 1.25 57.3 ± 1.21
YS-20 102.5 ± 1.56 59.6 ± 2.02
YS-60 98.8 ± 1.01 58.7 ± 1.15
Average 99.1 58.5
YS saponins in cRF c YS-0 99.3 ± 1.03 99.3 ± 1.72
YS-20 101.9 ± 0.58 98.7 ± 1.35
YS-60 100.7 ± 1.34 98.2 ± 2.06
Average 100.6 98.7
a Immediately following the 4-h incubation, four of the eight replicate
reaction tubes were assayed directly. The other four were centrifuged
(12 000 × g;20min;4 ◦ C) and supernatant was assayed.
b Ruminal fluid donors were fed daily amounts of 0, 20 or 60 g of
powdered Yuccaschidigera for >14 day prior to ruminal fluid collection.
Fluidsfromthetwoheiferspertreatmentwerepooledinequalvolumes.
c Pooled ruminal fluid was centrifuged (500×g;10min;4 ◦ C) to remove
protozoa. dRF is the defaunated supernatant. A portion of the dRF was
re-centrifuged (20 000 × g;20min;4 ◦ C), yielding clarified supernatant,
cRF.
in ethyl acetate, and its formation is not impeded by sugars,
sterols, fatty acids or vegetable oils. 12 In this study, the similar color
development (OD425) per unit of steroidal sapogenin or saponin
between dH2O- and RF-suspended samples indicates that residual
compounds in cRF also did not interfere with chromophore
formation. However, background absorbance was observed
both in dH2O and in RF samples that contained no steroidal
saponin/sapogenin, and this background color intensified with
time. The origin of this background color is not known, but
its presence in aqueous incubations indicates it was not likely
specific to ruminal fluid. We did notice that mixing concentrated
sulfuric acid with reagents in the absence of sapogenin gave
rise to a red color that intensified with time. Possibly, this
phenomenon is the source of the background interference. Baccou
et al. 12 also observed that a mixture of concentrated sulfuric acid,
p-anisaldehyde and acetic acid rapidly developed a dark red color.
The present study showed that adding an additional 0.5 mL dH2O
1 2 3 4 5 6 7
Std YS cRF
ECE-0 ECE-24 CAE-0 CAE-24
Figure 2. High performance thin layer chromatographic separation of
(lane 1) smilagenin standard (as model sapogenin) in ethanol; (lane 2)
extracted Yucca schidigera saponin (YS) in methanol; (lane 3) clarified
ruminal fluid; and (lanes 4–7) mixtures of YS (480 µgmL −1 )after0hor
24 h of incubation in (lanes 4 and 5) the extracellular enzyme (ECE) fraction
or (lanes 6 and 7) the cell-associated enzyme (CAE) fraction prepared from
ruminal fluid. Arrows near lanes 1 and 7 indicates yellow-pigmented bands
corresponding to sapogenin.
to the mixtures subsequent to the 10-min incubation at 60 ◦ C
stabilized the chromophore, likely through formation of a stable
hydrate, and substantially reduced background interference. The
reason for this reduction in interference is not known, but further
dilution of the sulfuric acid (0.5 mL dH2O added to 0.5 mL of
H2SO4 present in each 4-mL reaction mixture) may have impeded
oxidation of component(s) in the reaction mixture that gave rise
to background interference.
Centrifugation for separating sapogenin from saponin
and its biological implications
The markedly different detection patterns of YS saponin and smilagenin
(as a model sapogenin) in ruminal fluid with and without
centrifugation (Experiment 1), together with virtually complete recovery
across a range of added concentrations, confirmed that
centrifugation would be effective for distinguishing between
soluble (glycosylated) steroidal saponin and its insoluble, deglycosylated
analog, sapogenin. We observed similar partitioning
Table 3. Recoveries (%) of steroidal saponin added to ruminal fluid enzyme fractions a following 24 h of anaerobic incubation at 39 ◦ C(n = 4)
Enzyme fraction and original saponin (SE) concentration
Extracellular enzymes (ECE) Cell-associated enzymes (CAE)
Material assayed b 240 480 240 480
Whole mixture 96.4 ± 1.19 97.3 ± 1.23 109.1 ± 1.21 106.2 ± 0.71
Supernatant (S) 88.5 ± 1.19 89.6 ± 1.46 25.7 ± 0.48 31.8 ± 1.84
Pellets (P) 7.4 ± 1.22 5.5 ± 1.19 82.2 ± 1.61 76.8 ± 1.92
Total (S + P) 95.9 ± 1.23 95.1 ± 1.62 107.9±1.96 108.6 ± 2.25
a Pooled ruminal fluid was centrifuged (20 000 × g; 20min;4 ◦ C), yielding clarified supernatant, half of which was used for assay, considered as
the extracellular enzyme fraction (ECE). Half of the pelleted material was re-suspended in the remaining half of the supernatant, which had been
autoclaved and cooled. This resuspension was sonicated to disrupt cells, and was considered as the cell-associated enzyme fraction (CAE).
b Immediately following the 24 h incubation, sub-samples from each of the quadruplicate reaction tubes were assayed directly (whole mixture), or
after centrifugation (12 000 × g;20min;4 ◦ C) and supernatant and pellets were assayed separately.
www.interscience.wiley.com/jsfa Copyright c○ 2010 Crown in the right of Canada. J Sci Food Agric 2010; 90: 1811–1818
Published by John Wiley & Sons, Ltd

Assay for saponin deglycosylation www.soci.org
when sapogenin and saponin were suspended in water (Wang,
unpublished data). The association of sapogenin with the precipitate
likely reflects its hydrophobic nature, whereas hydrophilic
steroidal saponin (MW ∼1300, depending on the number of
sugars it contains) remained in the supernatant fraction after
centrifugation.
Experiment 2 confirmed that in the absence of ruminal bacteria
(i.e. in cRF), the solubility of YS saponin did not change during
the 4-h incubation, whereas when ruminal bacteria were present
(in dRF) during the incubation, a portion of the YS saponins
were no longer detectable in supernatant after 4 h, i.e. they had
become insoluble. Together with the findings from Experiment 1,
namely, that glycosylated saponins were soluble, whereas the nonglycosylated
smilagenin was not, these observations suggest that
the decrease in recovery of YS saponins from dRF supernatant
represents the portion that were deglycosylated by ruminal
bacteria or otherwise precipitated by ruminal compounds such
as soluble protein. Saponins are known to react with proteins,
and the stability of a saponin–protein complex is influenced by
characteristics of the protein as well as the saponin. 22–25 Soluble
protein contents of the cRF from YS-0, YS-20 and YS-60 (Table 2)
were 1.2, 1.5 and 1.6 mg mL −1 , respectively (measured as bovine
serum albumin equivalents). However, the complete recovery of
added saponin from cRF supernatant suggests that the soluble
proteins in RF used in this experiment had little effect on the
solubility of steroidal saponins. Thus, the difference between total
saponin+sapogenin in the whole (uncentrifuged) mixture and the
saponin that remained in supernatant following centrifugation
was assumed to represent that portion of saponin deglycosylated
to sapogenin. The fact that the pellet fraction contained sapogenin
was confirmed by the results from HPTLC in Experiment 3.
Research has shown that the effects of steroidal saponins
on ruminal microbes are species-specific. 4,5 Steroidal saponins
are known to react with membrane sterols, disrupt membrane
function and consequently arrest cell growth. 15,26,27 Sapogenin
was suggested as the region of the saponin that exerts these
biological effects. 28 Sapogenins, however, are hydrophobic; it is
the hydrophilic sugar moiety that renders the steroidal saponin
molecule soluble in aqueous environments. Segal et al. 29 proposed
that after a saponin contacts a cell membrane, β-glycosidase(s)
in the membrane hydrolyse the glycosidic bond in the saponin
molecule, releasing the sapogenin to combine with membrane
sterols. Saponins, therefore, are the biologically active form of
sapogenins.
The antifungal properties of steroidal saponins are lost upon
their deglycosylation. 16,17 Similarly, in a preliminary study, we also
found that growth of rumen cellulolytic bacteria was unaffected by
purified smilagenin or sarsasapogenin (Wang et al., unpublished
data). It is reasonable to assume, therefore, that saponin present
in a ruminal supernatant fraction (i.e. soluble) would be in a
biologically active form, and that a post-incubation decrease in
recovery of saponin from the supernatant of a ruminal culture
would reflect a loss of solubility attributable to deglycosylation
during exposure to microbial enzymes and presumably, a loss of
biological activity.
Degradation of steroidal saponin by rumen microbes
The complete recovery of steroidal saponin added to dRF or cRF
in Experiment 2, measured as total saponin + sapogenininwhole
mixture, suggests that the steroidal sapogenin aglycone core
structure was not degraded by the activities present in dRF or in
cRF, a result that is consistent with our previous research. 2 This
finding also agrees with other reports that rumen microbes are
capable of deglycosylating steroidal saponin to sapogenin, but
are unable to degrade the sapogenin aglycone structure. 30–32
In this study, protozoa had been removed both from dRF and
from cRF. Further, cRF would also have been nearly devoid of bacteria
as a result of centrifugation at 20 000 × g. Thus the difference
in recoveries of saponin from supernatant of dRF incubations vs.
cRF incubations, i.e. 58% vs. 99%, suggests that ruminal bacteria
were responsible for this insolubilization of saponin. Irreversible
adsorption of saponin to cell membranes occurs rapidly 33 and is
a prerequisite for its deglycosylation to sapogenin by membraneassociated
β-glucosidase. 29 Interestingly, in the present study, the
similar recoveries of total saponin+sapogenin (in whole mixture)
and of saponin (in supernatant) from cRF and dRF incubations
between donor heifers fed and not fed YS saponin suggests that
the deglycosylase activity of rumen bacteria had not been altered
over the course of their prior exposure to YS saponin. The adaptation
of ruminants to saponins observed as a result of prolonged
dietary exposure 10,11,34 is apparently due to mechanisms other
than an increase in the deglycosylase activity of rumen bacteria.
The glycosidic composition or linkages in the saponin molecules
may play a role in this adaptation. Wang et al. 5 also observed no
effect of pre-exposure to YS steroidal saponin on bacterial growth.
In the present study, rumen protozoa were removed by centrifugation,
thus the nature or extent of their role in the adaptation of
microbial populations to saponins is not known.
Deglycosylation of steroidal saponin by cell-associated
and extra-cellular enzymes
Experiment 2 demonstrated that prior exposure to YS steroidal
saponin did not affect deglycosylation of saponins by ruminal
bacteria, thus for Experiment 3 we utilized ruminal content from
the steers with no exposure to YS. The finding that significant
amounts of sapogenin were obtained only with CAE and not with
ECE demonstrates that the deglycosylating activity of rumen
bacteria is mainly cell-associated. This is consistent with the
finding in Experiment 2 that cell-free supernatant (cRF), whether
from pre-exposed or non-exposed heifers, exhibited limited
deglycosylase activity. Previous research has determined that
saponin deglycosylation is accomplished by β-glycosidases. 15,35,36
Segal et al. 29 and Segal and Schlösser 15 observed conversion
of saponins into their corresponding aglycones by various
membrane-associated glycosidases of erythrocytes and fungi, and
glycosidase activity associated with plant cell membranes has
also been observed. 37 Rumen bacterial glycosidase activity was
reportedbyWilliamset al. 38 Thus,thedeglycosylationofsaponinto
sapogenin observed in this study is likely attributable to bacterial
cell-associated glycosidases. This finding, in turn, supports the
conclusion from Experiment 2 that adaptation of ruminal bacteria
to YS saponin does notarise from increased bacterial deglycosylase
activity. Enhancement of that activity, presumably cell-associated,
would result in conversion of more saponin to sapogenin in the
vicinity of the cells, and therefore amplify the negative impact of
sapogenin on bacterial cells.
CONCLUSION
A previously established colorimetric method 12 for saponin determination
was modified to stabilize chromophore development,
enabling OD425 values to be recorded earlier and over longer
periods with no loss of repeatability. A simple procedure was established
to estimate the deglycosylation of steroidal saponin to
J Sci Food Agric 2010; 90: 1811–1818 Copyright c○ 2010 Crown in the right of Canada. www.interscience.wiley.com/jsfa
Published by John Wiley & Sons, Ltd
1817

1818
sapogenin by ruminal microbes. The procedure consisted of measuring
total steroidal saponin+sapogenin in whole culture and the
levels of steroidal saponin that remained in the supernatant after
centrifugation (12 000×g, 20 min) using the modified spectrophotometric
method. This study demonstrated that rumen bacteria
are capable of deglycosylating steroidal saponin to sapogenin,
but do not degrade the sapogenin core structure. Deglycosylase
activity in rumen bacteria is mainly cell-associated. Pre-exposure
to steroidal saponins did not alter the saponin-deglycosylating
capacity of rumen bacteria, suggesting that ruminal adaptation to
saponins occurs via another mechanism.
ACKNOWLEDGEMENTS
This study was conducted with financial support from Agriculture
and Agri-Food Canada. The authors appreciate the assistance
of Katherine Jakober and Sheila Torgunrud. This is Lethbridge
Research Centre contribution number 38709007.
REFERENCES
1 Wang Y, McAllister TA, Newbold CJ, Rode LM, Cheeke PR and
Cheng KJ, Effect of Yucca schidigera extract on fermentation
and degradation of steroidal saponins in the rumen simulation
technique (RUSITEC). Anim Feed Sci Technol 74:143–153 (1998).
2 Wang Y, McAllister TA, Yanke LJ, Xu ZJ, Cheeke PR and Cheng KJ,
In vitro effect of steroidal saponins from Yucca schidigera extract on
ruminal microbial protein synthesis and ruminal fermentation. JSci
Food Agric 80:2114–2122 (2000).
3 Wina E, Muetzel S, Hoffman E, Makkar HPS and Becker K, Saponins
containing methanol extract of Sapindus rarak affect microbial
fermentation, microbial activity and microbial community structure
in vitro. Anim Feed Sci Technol 121:159–174 (2005).
4 Wallace RJ, Arthauo L and Newbold CJ, Influence of Yucca shidigera
extract on ruminal ammonia concentrations and ruminal
microorganisms. Appl Environ Microbiol 60:1762–1767 (1994).
5 Wang Y, McAllister TA, Yanke LJ and Cheeke PR, Effect of steroidal
saponins from Yucca schidigera extract on ruminal microbes. J Appl
Microbiol 88:887–896 (2000).
6 Wina E, Muetzel S and Becker K, The dynamics of major fibrolytic
microbes and enzyme activity in the rumen in response to short
and long-term feeding of Sapindus rarak saponins. J Appl Microbiol
100:114–122 (2006).
7 Wilson RC, Overton TR and Clark JH, Effects of Yucca schidigera extract
and soluble protein on performance of cows and concentrations of
urea nitrogen in plasma and milk. JDairySci81:1022–1027 (1998).
8 Wang Y, Greer D and McAllister TA, Effects of moisture, roller setting
and saponin-based surfactant on barley processing, ruminal
degradation and growth performance of feedlot steers. JAnim
Sci 81:2145–2154 (2003).
9 Gorgulu M, Yurtseven S, Unsal I and Kutle HR, Effect of Yucca powder
on fattening performance male lambs. Appl Anim Res 25:33–36
(2004).
10 Newbold CJ, El-Hassan SM, Wang J, Ortega ME and Wallace RJ,
Influence of foliage from African multipurpose trees on activity
of rumen protozoa and bacteria. Br J Nutr 78:237–249 (1997).
11 Ivan M, Koenig KM, Teferedegne B, Newbold CJ, Entz T, Rode LM, et al,
Effect of the dietary Enterolobium cyclocarpum foliage on the
population dynamics of rumen ciliate protozoa in sheep. Small
Rum Res 52:81–91 (2004).
12 Baccou JC, Lambert F and Sauvaire Y, Spectrophotometric method
for the determination of total steroidal sapogenin. Analyst
102:458–465 (1977).
13 Makkar HPS and Becker K, Degradation of quillaja saponins by mixed
culture of rumen microbes. Lett Appl Microbiol 25:243–245 (1997).
14 Flåøyen A, Wilkins AL, Deng D and Brekke T, Ovine metabolism of
saponins: Evaluation of a method for estimating the ovine uptake
of steroidal saponins from Narthecium ossifragum. Vet Res Commun
25:225–238 (2001).
www.soci.org Y Wang, TA McAllister
15 Segal R and Schlösser E, Role of glycosidases in the membranlytic,
antifungal action of saponins. Arch Microbiol 104:147–150 (1975).
16 Arneson PA and Durbin RD, Studies on the mode of action of tomatine
as a fungitoxic agent. Plant Physiol 43:683–686 (1968).
17 Mahato SB, Ganguly AN and Sahu NP, Steroid saponins. Phytochemistry
21:959–978 (1982).
18 Canadian Council on Animal Care, Guide to the Care and Use
of Experimental Animals, 2nd edn., vol. 1, ed. by Olfert ED,
Cross BM and McWilliam AA. [Online]. CCAC, Ottawa, ON. Available:
http://www.ccac.ca/en/CCAC Programs/Guidelines Policies/
GUIDES/ENGLISH/toc v1.htm [16 December 2009].
19 Menke KH, Raab L, Salewski A, Steingass H, Fritz D and Schneider W.
The estimation of the digestibility and metabolizable energy
content of ruminant feedingstuffs from the gas production when
they are incubated with rumen liquor in vitro. J Agric Sci (Cambr.)
93:217–222 (1979).
20 Muzquiz M, Ridout CL, Price KR and Fenwick GR, The saponin content
and composition of sweet and bitter lupin seed. J Sci Food Agric
63:47–52 (1993).
21 SAS Institute Inc. SAS/STAT User’s Guide. SAS Inst. Inc. Cary, NC (1999).
22 Potter SM, Jimenez-Flores R, Pollack J, Lone TA and Berber-
Jimened MD, Protein–saponin interaction and its influence on
blood lipids. J Agric Food Chem 41:1287–1291 (1993).
23 Tanaka M, Fang-I L, Ishizaki S and Taguchi T, Interaction of saponins
with salt soluble proteins from walleye pollack meat. Fisheries Sci
61:373–374 (1995).
24 Ikedo S, Shimoyamada M and Watanabe K, Interaction between
bovine serum albumin and saponin as studied by heat stability
and protease digestion. J Agric Food Chem 44:792–795 (1996).
25 Shimoyamada M, Ootsubo R, Naruse T and Watanabe K. Effects of
soybean saponin on protease hydrolyses of β-lactoglobulin and
α-lactalbumin. Biosci Biotechnol Biochem 64:891–893 (2000).
26 Bangham AD, Horne RW, Glauert AM, Dingle JT and Lucy JA, Action of
saponin on biological cell membranes. Nature 196:952–955 (1962).
27 Price KR, Johnson IT and Fenwick GR, The chemistry and biological
significance of saponins in foods and feedstuffs. CRC Crit Rev Food
Sci Nutr 26:27–135 (1987).
28 Segal R, Mansour M and Zaitschek DV, Effect of ester groups on the
haemolytic action of some saponins and sapogenins. Biochem
Pharmacol 15:1411–1416 (1966).
29 Segal R, Shatkovsky P and Milo-Goldzweig I, On the mechanism of
saponin hemolysis – I. Hydrolysis of the glycosidic bond. Biochem
Pharmacol 23:973–981 (1974).
30 Flåøyen A and Wilkins AL, Metabolism of saponins from Narthecium
ossifragum – a plant implicated in the aetiology of alveld, a
hepatogenous photosensitization of sheep. Vet Res Commun
21:335–345 (1997).
31 Meagher LP, Smith BL and Wilkins AL, Metabolism of diosgeninderived
saponins: Implication for hepatogenous photosensitization
diseases in ruminants. Anim Feed Sci Technol 91:157–170 (2001).
32 Flåøyen A, Wilkins AL and Sandvik M, Rumen metabolism in sheep
of saponins from Yucca schidigera. Vet Res Commun 26:159–169
(2002).
33 Segal R, Milo-Goldzweig I and Seiffe M, The haemolytic properties of
non-ionic hemolysins. Life Sci 11:61–70 (1972).
34 Teferedegne B, McIntosh F, Osuji PO, Odenyo A, Wallace RJ and
Newbold CJ, Influence of foliage from different accessions of the
subtropical leguminous tree, Sesbania sesban on rumen protozoa
in Ethiopian and Scottish sheep. Anim Feed Sci Technol 78:11–20
(1999).
35 Matsuura H. Saponins in garlic as modifiers of the risk of cardiovascular
disease. JNutr131:1000S–1005S (2001).
36 Yang DJ, Lu TJ and Hwang LS. Effect of endogenous glycosidase
on stability of steroidal saponins in Taiwanese yam (Dioscorea
pseudojaponica yamamoto) during drying processes. Food Chem
113:155–159 (2009).
37 Grünweller S and Kesselmeier J. Characterization of a membrane
bound beta-glucosidase responsible for the activation of oat leaf
saponins. Phytochemistry 24:1941–1943 (1985).
38 Williams AG, Withers SE and Coleman GS. Glycoside hydrolases of
rumen bacteria and protozoa. Curr Microbiol 10:287–294 (1984).
www.interscience.wiley.com/jsfa Copyright c○ 2010 Crown in the right of Canada. J Sci Food Agric 2010; 90: 1811–1818
Published by John Wiley & Sons, Ltd

Research Article
Received: 22 January 2010 Revised: 23 April 2010 Accepted: 26 April 2010 Published online in Wiley Interscience: 3 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4020
Impact of ultrafiltration and nanofiltration
of an industrial fish protein hydrolysate on its
bioactive properties
Laurent Picot, a∗ Rozenn Ravallec, b Martine Fouchereau-Péron, c Laurent
Vandanjon, d,e Pascal Jaouen, d Maryse Chaplain-Derouiniot, d Fabienne
Guérard, f Aurélie Chabeaud, f Yves LeGal, c Oscar Martinez Alvarez, c,g
Jean-Pascal Bergé, h Jean-Marie Piot, a Irineu Batista, i Carla Pires, i Gudjon
Thorkelsson, j,k Charles Delannoy, l Greta Jakobsen, m Inez Johansson m
and Patrick Bourseau d,e
Abstract
BACKGROUND: Numerous studies have demonstrated that in vitro controlled enzymatic hydrolysis of fish and shellfish proteins
leads to bioactive peptides. Ultrafiltration (UF) and/or nanofiltration (NF) can be used to refine hydrolysates and also to
fractionate them in order to obtain a peptide population enriched in selected sizes. This study was designed to highlight the
impact of controlled UF and NF on the stability of biological activities of an industrial fish protein hydrolysate (FPH) and to
understand whether fractionation could improve its content in bioactive peptides.
RESULTS: The starting fish protein hydrolysate exhibited a balanced amino acid composition, a reproducible molecular weight
(MW) profile, and a low sodium chloride content, allowing the study of its biological activity. Successive fractionation on UF and
NF membranes allowed concentration of peptides of selected sizes, without, however, carrying out sharp separations, some MW
classes beingfound in several fractions. Peptides containingPro, Hyp, Aspand Glu were concentrated in the UF and NF retentates
compared to the unfractionated hydrolysate and UF permeate, respectively. Gastrin/cholecystokinin-like peptides were present
in the starting FPH, UF and NF fractions, but fractionation did not increase their concentration. In contrast, quantification
of calcitonin gene-related peptide (CGRP)-like peptides demonstrated an increase in CGRP-like activities in the UF permeate,
relative to the starting FPH. The starting hydrolysate also showed a potent antioxidant and radical scavenging activity, and a
moderate angiotensin-converting enzyme (ACE)-1 inhibitory activity, which were not increased by UF and NF fractionation.
CONCLUSION: Fractionation of an FPH using membrane separation, with a molecular weight cut-off adapted to the peptide
composition, may provide an effective means to concentrate CGRP-like peptides and peptides enriched in selected amino
acids. The peptide size distribution observed after UF and NF fractionation demonstrates that it is misleading to characterize
the fractions obtained by membrane filtration according to the MW cut-off of the membrane only, as is currently done in the
literature.
c○ 2010 Society of Chemical Industry
Keywords: fish protein hydrolysate; ultrafiltration; nanofiltration; membrane separation; fractionation process; bioactive peptide
INTRODUCTION
Peptides resulting from natural digestion or controlled enzymatic
hydrolysisoffoodproteinsexhibitnotonlynutritionalactivitiesbut
also biological activities of dietary or pharmacological interest. 1,2
Following ingestion, proteins are hydrolysed in the stomach by HCl
and pepsins and, to a lesser extent, in the upper partof the intestine
by trypsin and chymotrysin. The resulting enzymatic hydrolysate
mostly consists of a mixture of free amino acids and dipeptides,
able to bind gastric receptors on mucosal cells, and to be resorbed
via a transcellular transport across the intestinal brush border
cells, as demonstrated using synthetic oligopeptides. 3,4 Beyond
this classical paradigm, it was demonstrated that oligopeptides
∗ Correspondence to: Laurent Picot, UMR CNRS 6250 LIENSs, Université de La Rochelle,
La Rochelle, France. E-mail: laurent.picot@univ-lr.fr
a UMR CNRS 6250 LIENSs, Université de La Rochelle, La Rochelle, France
b ProBioGEM, UPRES EA-1026, IUT A-Polytech’Lille, USTL, Lille, France
c UMR BOREA (Biologie des Organismes et Ecosystèmes Aquatiques), MNHN/CNRS
7208/IRD 207/UPMC, Station de Biologie Marine, Concarneau, France
d UMR CNRS 6144 GEPEA, Université deNantes,Saint-Nazaire,France
e LIMATB, Université deBretagneSud,Lorient,France
f ANTiOX, Université de Bretagne Occidentale, Quimper, France
g CSIC, Consejo Superior de Investigaciones Científicas, Instituto del Frío, Madrid,
Spain
h IFREMER, STAM, Nantes, France
i Ipimar, Lisbon, Portugal
j Matis ohf, Reykjavik, Iceland
k University of Iceland, Reykjavik, Iceland
l Copalis, Boulogne sur Mer, France
m Marinova, Højmark, Denmark
J Sci Food Agric 2010; 90: 1819–1826 www.soci.org c○ 2010 Society of Chemical Industry
1819

1820
can also cross the intestinal border following a transepithelial route
or oligopeptides transporters. 5,6 Some factors such as intraluminal
hypertonicity increase this process. 7
In vitro and in vivo studies have shown that enzymatic hydrolysis
of dietary proteins leads to bioactive peptides such as growth
factors, 8 immunomodulators, 9 antimicrobial, 10 antithrombotic, 11
angiotensin-converting enzyme (ACE)-1 inhibitors, 12 opiates or
antiproliferative peptides. 13 Research focused on peptides contained
in fish protein hydrolysates (FPH) demonstrated that they
also contain molecules showing promising health benefits for
nutritional or pharmaceutical applications. 14–16 Only few studies
were performed to confirm the in vivo biological activity of FPH
peptides, but convincing data were obtained on relevant animal
models, e.g. spontaneously hypertensive or diabetic rats. For instance,
in vitro 17–19 and in vivo studies on hypertensive rats 20–22
demonstrated that pollack flesh or tuna muscle peptides inhibit
ACE-1 and decrease blood pressure. The in vivo hypotensive activity
of a short peptide identified in a sardine muscle hydrolysate
was also demonstrated in humans by a dietary study on mild
hypertensive subjects. 23
Fish and shellfish peptides could also contribute to free radical
scavenging and pathology prevention. 24 Peptides obtained from
shrimp by-products, 25 hoki, 26 tuna backbone proteins, 27 jumbo
squid skin gelatine, 28,29 conger eel muscle 30 and yellow stripe
trevally 31 exert a significant in vitro antioxidant activity. Most
antioxidant peptides identified are short (5–16 amino acids),
contain hydrophobic amino acids, Val and Leu, at the N-terminus
and Pro, His or Tyr in their sequence. 14 Tyrosines substantially
contributetothescavengingoffreeradicalsbecausetheirphenolic
lateral chains act as potent electron donors, 28,29,32 thus allowing
the termination of the radical chain reaction. 31,33
FPH have also been identified as a source of
hormonal peptides, 34–39 particularly those able to bind gastrin/cholecystokinin
(CCK) and calcitonin gene-related peptide
(CGRP) receptors. FPH peptides also cross-react with specific
antibodies directed against these hormones. Gastrointestinal secretagogue
peptides such as gastrin/CCK 34–36 exhibit a large
spectrum of activities ranging from the stimulation of protein
synthesis 37 to the control of intestinal mobility and secretion of
digestive enzymes. 40 Involvement of CCK-8 in the satiety mechanisms
controlling food intake in humans is well documented, 41,42
and the gastric localization of gastrin and CCK receptors suggests
that dietary peptides acting as agonists on these receptors could
be of interest as satietogenic ingredients in functional food. CGRP
is a neuropeptide synthesized by alternative splicing of the calcitonin
gene 38 and regulating a large number of physiological
functions such as vasodilatation, gastric acid secretion and cardiac
metabolism. 39,43 Binding of CGRP to its receptors reduces acidic
secretion and the risk of ulceration, suggesting that peptides
acting as CGRP receptor agonists could be of pharmacological
interest. 44,45
Several studies refer to the use of ultrafiltration (UF) in order
to refine hydrolysates and to increase their specific activity in the
perspective of industrial upgrading of by-products to produce
bioactive ingredients for human food or animal feeding. The
specific activity of UF fractions is then compared with that of
the initial hydrolysate, with the aim of identifying the most
active fractions. Several hydrolysates from different substrates
have been fractionated in this way at the Pukyong National
University in Pusan, South Korea: hot washing waters of cod
frame proteins, 46 Alaska pollack frame proteins, 47 jumbo squid
skin gelatine, 28 giant squid muscle, 48 hoki frame proteins 26 and
www.soci.org L Picot et al.
conger eel muscle. 30 Activities tested are generally antioxidant and
antihypertensive, but radical scavenging activities and foaming
or emulsifying properties have also been considered. Similar
studies have been made by other teams. Neves et al. 49 studied
the impact of enzyme source and hydrolysis conditions on the
molecular weight distribution of a hydrolysate of brackish water
minced fish and minced shrimps. Sumaya-Martinez 50 fractionated
a shrimp frame hydrolysate by successive microfiltration (0.45 µm)
and ultrafiltration (30 kDa, 5 kDa). Unfortunately, only few details
are given in these works concerning the operating conditions
of the fractionation process, except the molecular weight
cut-off (MWCO) of the membranes used. For instance, the
volume reduction factor (VRF) is rarely defined, so the peptidic
population whose activity is tested is not precisely known.
It is recognized, however, that short peptides, below 3 or
4 kDa, usually harbour bioactive properties and the peptide
size is a physicochemical parameter – although not a unique
one – controlling peptide activities. Some studies, however,
discuss the impact of ultrafiltration on peptidic populations on
the basis of size exclusion chromatograms. 51–54 The present study
was designed to discover whether UF and nanofiltration (NF) of a
bioactive FPH, performed under controlled conditions, represent
innovative industrial processes for obtaining fractions defined by
a strict range of peptide size and enriched in bioactive peptides.
MATERIALS AND METHODS
Fish protein hydrolysate
The FPH selected for this study was Prolastin, a commercial
product from the French company Copalis. Prolastin is an
elastin hydrolysate obtained by controlled proteolysis of skins
from North Atlantic lean fish (Gadidae: mostly cod and pollack),
followed by purification steps based on sieving, centrifugation
and discoloration. Proteolysis was performed at Copalis, using
an industrial bacterial endopeptidase under optimized conditions
of time, pH and enzyme/substrate ratio. Prolastin is composed
of polypeptides with a low molecular weight (1000–5000 Da),
which makes it soluble and very digestible. Its NaCl content is very
low (0.79%, w/w), which allows a relevant study of the biological
activity of the starting hydrolysate and related fractions. It is sold
as a functional ingredient for dietary complements, helping to
give elasticity to tissues and limit their ageing. It is produced on
an industrial scale and the reproducibility of the process and final
product were confirmed before fractionation.
Ultrafiltration and nanofiltration of FPH
Prolastin was fractionated successively by UF and NF, the UF
permeate being used as the feed solution for the NF step (Fig. 1).
Experiments were carried out on a Microlab40 pilot plant (VMA
Industrie, Versailles, France) with a maximum capacity of 5 L
(launching tank 4.3 L + dead volume 0.7 L). The pilot plant was
equipped with tubular organic membranes (PCI Ltd, Singapore),
diameter 12 mm, surface area 0.033 m 2 : an NF membrane in
polyamide/polyethersulfone (60% retention in CaCl2, ref. AFC40)
and a UF membrane in modified polyethersulfone (MWCO 4 kDa,
ref. ESP04). The ESP04 UF membrane was chosen from the PCI
range according to the molecular weight distribution of the
hydrolysate. It was recently shown that a nominal MWCO between
the molecular weight of the largest and medium-sized peptides is a
good choice for the fractionation of a hydrolysate. 55 The AFC40 NF
membrane was selected as it has the poorest salt retention in the
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1819–1826

Ultrafiltration and nanofiltration of industrial fish protein hydrolysate www.soci.org
Figure 1. The ultrafiltration–nanofiltration sequence.
PCI range, which makes it the most appropriate to fractionate the
UF permeate. An MWCO of about 300 Da has been estimated by
the membrane supplier for the fractionation of organic molecules
onto this NF membrane.
A 10 L volume of solution was prepared by mixing 100 g
of hydrolysate powder in warm pure water. A 1 L aliquot was
reserved for analysis. Thus 9 L of the initial solution was filtered
using the 4 kDa membrane. At the end of the concentration
process, volumes of about 8 L permeate and 1 L retentate were
recovered. All retentate was kept for analysis, as well as a
sample of about 1 L permeate. The remaining permeate (7 L)
constituted the initial solution filtered by NF in the second
step of the fractionation cascade. Six litres of permeate and
1 L retentate were obtained and kept for analysis. At the end
of each filtration step a sample of 10–50 mL ‘instantaneous’
permeate was reserved in order to estimate the retention factor
at the end of this operation. Filtrations were carried out under the
following operating conditions: tangential velocity V = 2.5ms −1 ,
temperature T = 55 ◦ C, transmembrane pressure �P = 30 bar for
UF or 35 bar for NF. The selected values for V and �P were the
maximum allowed by the pilot and the membrane, in order to
minimize filtration time. However, filtration times were also quite
long due to the high-reduction volume factors that were reached.
Thus a fairly high temperature was chosen in order to maximize
permeation fluxes, as well as to limit bacterial growth.
Previously unused UF and NF membranes were conditioned
before use according to the following sequence of washing:
water – Ultrasil 11 – water – nitric acid – water. Recovery rate in
peptides in UF or NF fractions were computed as the nitrogen
content of the fraction considered (retentate or permeate) divided
by the sum of nitrogen content in retentate and permeate.
Nitrogen content of samples was determined by the Kjeldahl
method following the norm NF EN 25 663 ISO 5663 (1994).
Physical and biochemical characterization of FPH, UF and NF
fractions
Colour, dry matter, pH, NaCl content, protein content and
protein recovery rates were measured or calculated following
normalized procedures. NaCl contents were determined using
a coulometric method (chlorimeter 926, Corning, Halstead, UK).
Protein contents were calculated as 6.25 times the total Kjeldahl
nitrogen amount. Aminogram of Prolastin was determined using
10 mg of freeze-dried sample hydrolysed for 24 h at 110 ◦ Cin
200 µL of6molL −1 HCl in vacuum-sealed vials. The hydrolysed
sample was dried under N2, diluted with 2.5 mL deionized water,
and amino acid composition was determined using the EZ : faast
procedure (Phenomenex, Torrance, CA, USA), consisting of a solidphase
extraction step followed by derivatization and liquid–liquid
extraction. The organic phase containing derivatized amino acids
was analysed by gas chromatography–flame ionization detection
(PerkinElmer Autosystem XL, Waltham, MA, USA). Amino acids
were identified according to their retention time and quantified
using a calibration curve and an internal standard (norvaline
200 µmol L −1 added to each sample). The molecular weight
(MW) distributions of the native hydrolysate and UF/NF fractions
were analysed by size exclusion chromatography in fast protein
liquid chromatography mode using a Superdex Peptide® HR
10/300 column (Amersham, Little Chalfont, UK; fractionation
range: 7000–100 Da) according to Guérard. 25 Samples were
diluted so that the mass injected was the same for each sample.
The resulting normalized chromatograms obtained in this way
are representative of the peptide distribution, both in terms
of MW and mass composition, and are more expressive than
raw chromatograms to analyse the impact of a membrane
fractionation. Total areas of the chromatograms were integrated
and separated into five MW ranges expressed as the percentage of
the total surface (>7000, 7000–3000, 3000–1000, 1000–300 and

1822
Table 1. Physical properties of the PROLASTIN hydrolysate and its UF and NF fractions
www.soci.org L Picot et al.
Colour Dry matter (DM) g L −1 pH NaCl/DM (%) Protein/DM (%) Protein recovery rate % a
Unfractionated hydrolysate Yellow-orange 111.50 8 0.79 88.5
Retentate UF 4 kDa Yellow-orange 355.25 7.9 0.06 90.9 31
Permeate UF 4 kDa Yellow 85 7.9 1.06 81.6 69
Retentate NF 300 Da Yellow 325 7.9 0.09 91.3 77
Permeate NF 300 Da Yellow 25.55 8.1 3.60 71.1 23
a Protein recovery rate was determined with reference to the unfractionated hydrolysate for ultrafiltration to the UF permeate for NF fractions.
rats according to the method of Neville 56 until step 11. Proteins
were quantified by the method of Folin–Lowry using bovine
serum albumin (BSA) as standard. 57 Incubations, in a 400 µL
final volume, were performed at 22 ◦ C for 1h. 58 At the end
of the incubation, bound and free ligands were separated by
centrifugation in a solution containing 2% BSA. Each batch was
tested at four increasing protein concentrations, and only the
straight lines presenting slopes similar to that obtained with
the standard hormone (0.01–1 ng per assay) were considered as
positive. The receptor binding ability of each purified fraction was
determined in triplicate independent assays and used to calculate
the quantity (pg) of CGRP-like activity per milligram of protein.
Data were also expressed as the amount of protein (mg) that
induced a 50% inhibition of the initial CGRP binding to rat liver
membranes (ED50).
Antioxidant activity
DPPH radical scavenging activity
Radical scavenging activity was measured according to the
procedure reported by Morales and Jiménez-Pérez. 59 An aliquot
of sample (200 µL) was added to 1 mL of a daily-prepared solution
of 1,1-diphenyl-2-picrylhydrazyl (DPPH ·) at a concentration of
74 mg L −1 in ethanol. The mixture was shaken for 1 h at 25 ◦ C. The
sample was centrifuged at 10 000 × g for 5 min and absorption of
the supernatant was measured at 520 nm. DPPH · concentration
in the reaction medium was calculated from the calibration curve,
determined by linear regression:
[DPPH]t = 0.0241(A520 nm) + 0.022(r 2 = 0.9995)
The radical scavenging activity of the sample was expressed as
percentage disappearance of DPPH ·:
DPPH radical scavenging activity (%) =
(1 − ([DPPH]t/[DPPH]H2O) × 100)
where [DPPH]H2O is the concentration of DPPH in the presence
of water instead of hydrolysate. In vitro DPPH radical scavenging
activity was determined in triplicate independent assays and
expressed as AC50, corresponding to the concentration of
hydrolysate (mg mL −1 protein determined according to the
Kjeldahl method) able to scavenge 50% of DPPH radical.
Quantification of antioxidant activity using the β-carotene/linoleate
model system
Antioxidant activity was determined using the β-carotene–
linoleate model system as described by Marco 60 with some slight
modifications. Five-millilitre aliquots of a β-carotene–linoleate
emulsion were transferred to glass tubes, containing 200 µL
of sample at several protein concentrations in MilliQ water.
All samples were stirred and incubated simultaneously for
2h at 50 ◦ C. Absorbance values at 470 nm of samples and
controls (water and butylhydroxyanisol (BHA)) were recorded
every 30 min with a microplate reader (SpectraMax Plus 384,
Molecular Devices, Sunnyvale, CA, USA). The antioxidant activity
gives an estimate of the relative protection of each sample against
the oxidation of linoleate, observed by the bleaching extent of
the β-carotene–linoleate emulsion. The antioxidant activity was
calculated as follows:
Antioxidant activity = (ODt=60 min/ODt=0 min) × 100
In vitro antioxidant activity was determined in triplicate independent
assays and expressed as AC50, corresponding to the
concentration of hydrolysate (mg mL −1 protein determined according
to the Kjeldahl method) inducing 50% antioxidant activity.
ACE-1 inhibition assay
In vitro inhibition of ACE-1 was assayed following the spectrophotometric
Holmquist method. 61 Inhibition was calculated from
triplicate independent assays and expressed as IC50 and maximal
percentage inhibition. IC50 corresponds to the concentration of
hydrolysate or fraction inducing 50% inhibition of ACE-1 activity.
Captopril 21.7 × 10 −3 µgmL −1 (0.1 µmol L −1 ) was used as a
reference inducing 100% inhibition.
RESULTS AND DISCUSSION
Physical, chemical and biochemical properties of FPH
and fractions
Colour, dry matter, pH, NaCl content, protein content, protein
recovery rates and aminograms of Prolastin and related UF and NF
fractions are given in Tables 1 and 2. Aminograms of Prolastin and
related fractions revealed a balanced composition and confirmed
their high nutritional value (Table 2). It is of interest to note that
peptides containing Pro, Hyp, Asp and Glu were concentrated
in the UF and NF retentates compared to the unfractionated hydrolysate
and UF permeate respectively (Table 2), suggesting that
the presence of specific amino acids may influence the interactions
of peptides with the peptide layer in contact with the membranes.
Peptidic profiles and impact of fractionation on MW
distribution
Prolastin is a highly hydrolysed peptide mix, in which 98.6%
(by mass) of peptides have a molecular weight lower than 4000
Da (Table 3). Ultrafiltration of the crude hydrolysate, as well as
nanofiltration of the ultrafiltrated permeate, constituted very
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1819–1826

Ultrafiltration and nanofiltration of industrial fish protein hydrolysate www.soci.org
Table 2. Amino acid composition of unfractionated Prolastin and related UF and NF fractions
Aminoacid(mmolg −1 dry sample)
Ala Gly Val Leu Ile Thr Ser Pro Asp Met Hyp Glu Phe Lys His Hly Tyr
Unfractionated hydrolysate 0.71 1.83 0.25 0.35 0.18 0.29 0.46 0.56 0.40 0.13 0.22 0.43 0.14 0.12 0.05 0.00 0.04
Retentate UF 4 kDa 0.65 2.16 0.26 0.27 0.20 0.30 0.47 0.78 0.64 0.11 0.40 0.64 0.12 0.14 0.06 0.02 0.04
Permeate UF 4 kDa 0.72 1.62 0.29 0.37 0.22 0.29 0.44 0.49 0.43 0.12 0.20 0.56 0.13 0.15 0.06 0.01 0.05
Retentate NF 300 Da 0.70 1.85 0.31 0.33 0.23 0.31 0.44 0.62 0.52 0.12 0.27 0.68 0.13 0.16 0.07 0.01 0.04
Permeate NF 300 Da 0.88 1.25 0.28 0.51 0.19 0.27 0.47 0.27 0.21 0.11 0.04 0.25 0.16 0.13 0.06 0.01 0.05
Ala, alanine; Gly, glycine; Val, valine; Leu, leucine; Ile, isoleucine; Thr, threonine; Ser, serine; Pro, proline; Asp, aspartic acid; Met, methionine; Hyp,
hydroxyproline; Glu, glutamic acid; Phe, phenylalanine; Lys, lysine; His, histidine; Hly, hydroxylysine; Tyr, tyrosine.
Table 3. Molecular weight distribution of peptides contained in Prolastin and related UF and NF fractions (% repartition)
Molecular weight Da) >7000 7000–4000 4000–1000 1000–300 4000 Da for the
UF retentate, 300–4000 Da for the NF retentate, etc.).
Presence of gastrin/CCK-like peptides in FPH and fractions
Dosage of gastrin/CCK-like peptides in the fractions is presented
in Table 4. Gastrin/CCK-like peptides were identified in the unfractionated
hydrolysate and in all UF and NF fractions. The curve
Flux (L h -1 m -2 )
140
120
100
80
60
40
UF
NF
20
Time (min)
0
0 50 100 150 200
Figure 2. Flux evolution with time for UF and NF steps, Prolastin
hydrolysate (Reprinted with permission from Ref. [55]. Copyright 2009.
Elsevier).
slope measured with the unfractionated hydrolysate was very
close to that of standard 125 I-gastrin, indicating the presence of
peptides binding specifically to gastrin antibodies. The amount
of gastrin/CCK-like peptides present in the unfractionated hydrolysate
was in the range of data previously published for fish and
shellfish hydrolysates, 36 the best activity (about 4 pg of gastrin-like
peptides) being obtained with a shrimp hydrolysate fraction containing
peptides between 4000 and 1000 Da. The ED50 value of the
unfractionated hydrolysate was very high, which is characteristic of
alowbindingaffinity.UFandNFofProlastinresultedinvariationsin
the curve slope and dilution of gastrin/CCK-like peptides, suggesting
that these processes do not allow an enrichment of peptides
of interest. Interestingly, the ED50 measured for UF and NF retentates
was much lower than that of the unfractionated hydrolysate,
suggesting that the gastrin antibody binding affinity of peptides
present in these fractions was increased. However, the main conclusion
of this dosage is that UF or NF fractionations do not allow
concentration of gastrin/CCK-like peptides in selected fractions.
J Sci Food Agric 2010; 90: 1819–1826 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1823

1824
Table 4. Secretagogue activities (gastrin/CCK-like peptides) in Prolastin
hydrolysate and related UF and NF fractions
Fraction Slope
Gastrin/CCK-like
peptides
(pg mg −1 dry
weight)
www.soci.org L Picot et al.
ED50
(mg dry
weight)
125 I-gastrin-17 −41.99 – –
Unfractionated hydrolysate −38.87 1.30 11.03
Retentate UF 4 kDa −58.86 0.98 5.97
Permeate UF 4 kDa −48.83 0.93 9.03
Retentate NF 300 Da −64.61 0.76 5.99
Permeate NF 300 Da ND ND ND
ND, not determined.
Table 5. CGRP-like peptides in the Prolastin hydrolysate and related
UF and NF fractions
Fraction
CGRP-like peptides
(pg mg −1 dry weight)
ED50
(mg dry weight)
Unfractionated hydrolysate 4 19
Retentate UF 4 kDa ND ND
Permeate UF 4 kDa 42 3.6
Retentate NF 300 Da 21.5 6.1
Permeate NF 300 Da 26.6 6.9
ND, not detectable, as the slope was very different from the CGRP
standard slope.
Presence of CGRP-like peptides in FPH and fractions
The dosage of CGRP-like peptides in fractions demonstrated an
important increase (tenfold) of CGRP-like activities in the UF
permeate, relative to the crude extract (Table 5). NF of the UF
permeate did not allow a higher concentration of CGRP-like
peptides. The highest activity found in the UF permeate was
also confirmed by the ED50 values (Table 5). The best interaction
was observed with the UF permeate, with only 3.6 mg dry weight
of sample necessary to obtain 50% inhibition of the maximal
binding of radiolabelled CGRP; this value corresponded to a 5.2fold
affinity increase compared to the unfractionated hydrolysate.
Moreover, this ED50 value could not be increased by a subsequent
NF process. Therefore it appears that UF allows concentration
of CGRP-like peptides in the UF permeate, with an important
increase of affinity for CGRP receptors, as measured using the
binding displacement of 125 I-CGRP.
Antioxidant activities
The in vitro antioxidant activities of Prolastin hydrolysate and
fractions are presented in Table 6. Before fractionation, Prolastin
showed a high DPPH radical scavenging activity, with an AC50
of 24.7 mg mL −1 . This result suggests that Prolastin contains
peptides reacting with free radicals to form more stable products.
Prolastin also showed a potent antioxidant activity, with an AC50
of 0.12 g L −1 . Fractionation of Prolastin induced a weak loss of
antioxidant and DPPH radical scavenging activities. According to
Pihlanto, 62 the structure–activity relationship of the antioxidant
mechanism of protein hydrolysates is not yet fully understood.
The antioxidant activity seems to be inherent to the characteristic
amino acid sequences of the peptides, depending both on the
Table 6. Radical scavenging and antioxidant activities of Prolastin
hydrolysate and related UF and NF fractions
Fraction
Radical scavenging
activity (DPPH · test)
AC50 (mg mL −1 )
Antioxidant activity
(β-carotene test)
AC50 (mg mL −1 )
Unfractionated hydrolysate 24.7 0.12
Retentate UF 4 kDa 39.3 0.24
Permeate UF 4 kDa 40 0.36
Retentate NF 300 Da 32.7 0.24
Permeate NF 300 Da ND

Ultrafiltration and nanofiltration of industrial fish protein hydrolysate www.soci.org
ACE inhibitory activity is mostly dependent on peptide size rather
than on fish species source. Ultrafiltration of Prolastin did not
result in drastic changes in ACE inhibitory activity, contrary to
previous observations obtained on Pacific hake peptides. Several
hypotheses can be drawn to explain why UF had only a weak
impact on ACE inhibitory activities. The presence of a large amount
of peptides of

1826
20 Fujii M, Matsumura N, Mito K, Shimizu T, Kuwahara M, Sugano S, et al,
Antihypertensive effects of peptides in autolysate of bonito bowels
on spontaneously hypertensive rats. Biosci Biotechnol Biochem
57:2186–2188 (1993).
21 Karaki H, Kuwahara M, Sugano S, Doi C, Doi K, Matsumura N, et al,
Oral administration of peptides derived from bonito bowels
decreases blood pressure in spontaneously hypertensive rats by
inhibiting angiotensin converting enzyme. Comp Biochem Physiol C
104:351–353 (1993).
22 Fujita , Yokoyama K, Yasumoto R and Yoshikawa M, Antihypertensive
effect of thermolysin digest of dried bonito in spontaneously
hypertensive rat. Clin Exp Pharmacol Physiol Suppl 22:S304–305
(1995).
23 Kawasaki T, Seki E, Osajima K, Yoshida M, Asada K, Matsui T, et al,
Antihypertensive effect of valyl-tyrosine, a short chain peptide
derived from sardine muscle hydrolyzate, on mild hypertensive
subjects. J Hum Hypertens 14:519–523 (2000).
24 Diplock AT, Charleux JL, Crozier-Willi G, Kok FG, Rice-Evans C,
Roberfroid M, et al, Functional food science and defence against
reactive oxidative species: functional food science in Europe. Br J
Nutr 80:77–112 (1998).
25 Guérard F, Sumaya-Martinez MT, Laroque D, Chabeaud A and
Dufossé L, Optimization of free radical scavenging activity in
the hydrolysis of shrimp processing discards by response surface
methodology. Process Biochem 42:1486–1491 (2007).
26 Kim Y, Je JY and Kim SK, Purification and characterization of
antioxidant peptide from hoki (Johnius belengerii) frameprotein
by gastrointestinal digestion. JNutrBiochem18:31–38 (2007).
27 Je JY, Qian ZJ, Byun HG and Kim SK, Purification and characterization
of an antioxidant peptide obtained from tuna backbone protein by
enzymatic hydrolysis. Process Biochem 42:840–846 (2007).
28 Mendis E, Rajapakse N, Byun HG and Kim SK, Investigation of jumbo
squid (Dosidicus gigas) skin gelatin peptides for their in vitro
antioxidant effects. Life Sci 77:2166–2178 (2005).
29 Mendis E, Rajapakse N and Kim SK, Antioxidant properties of a radicalscavenging
peptide purified from enzymatically prepared fish skin
gelatin hydrolysate. J Agric Food Chem 53:581–587 (2005).
30 Ranathunga S, Rajapakse N and Kim SK, Purification and
characterization of antioxidative peptide derived from muscle of
conger eel (Conger myriaster). Eur Food Res Technol 222:310–315
(2006).
31 Klompong V, Benjakul S, Kantachote D and Shahidi F, Antioxidative
activity and functional properties of protein hydrolysate of yellow
stripe trevally (Selaroides leptolepis) as influenced by the degree of
hydrolysis and enzyme type. Food Chem 102:1317–1327 (2007).
32 Suetsuna K, Antioxidant peptides from the protease digest of prawn
(Penaeus japonicus)muscle.Marine Biotechnol 2:5–10 (2000).
33 Thiansilakul Y, Benjakul S and Shahidi F, Antioxidative activity of
protein hydrolysate from round scad muscle using alcalase and
flavourzyme. J Food Biochem 31:266–287 (2007).
34 Cancre I, Ravallec R, Van Wormhoudt A, Stenberg E, Gildberg A and
le Gal Y, Secretagogues and growth factors in fish and crustacean
protein hydrolysates. Marine Biotechnol 1:489–494 (1999).
35 Ravallec-Plé R, Charlot C, Pires C, Batista I, Van Wormhoudt A, Le Gal Y,
et al, The presence of bioactive peptides in hydrolysates prepared
from processing wastes of sardine (Sardina pilchardus). J Sci Food
Agric 81:1120–1125 (2001).
36 Ravallec-Plé R and Van Wormhoudt A, Secretagogues activities in cod
(Gadus morhua)andshrimp(Penaeus aztecus) extracts and alcalase
hydrolysates determined in AR4-2J pancreatic tumour cells. Comp
Biochem Physiol B: Biochem Mol Biol 134:669–679 (2003).
37 Johnson LRE, Copeland EM and Dudrick SJ, Luminal gastrin stimulates
growth of the distal intestine. Scand J Gastroenterol 13:95 (1978).
38 Amara SG, Jonas V, Rosenfeld MG, Ong ES and Evans RM, Alternative
RNA processing in calcitonin gene expression generates mRNAs
encoding different polypeptide products. Nature 298:240–244
(1982).
39 Brain SD and Grant AD, Vascular actions of calcitonin gene-related
peptide and adrenomedullin. Physiol Rev 84:903–934 (2004).
40 Beinfeld MC, Cholecystokinin/gastrin, in Psychopharmacology: The
Fourth Generation of Progress, ed. by Bloom FE and Kupfer D. Raven
Press, New York, pp. 585–594 (1995).
41 Bray GA, Afferent signals regulating food intake. Proc Nutr Soc
59:373–384 (2000).
42 Thavanathan R and Volkoff H, Effects of amylin on feeding of goldfish:
interactions with CCK. Regul Peptides 133:90–96 (2006).
www.soci.org L Picot et al.
43 LenzHJ,MortrudMT,ValeWW,RivierJEandBrownMR,Calcitonin
gene related peptide acts within the central nervous system to
inhibit gastric acid secretion. Regul Peptides 9:271–277 (1984).
44 Fouchereau-Peron M, Duvail L, Michel C, Gildberg A, Batista I and Le
Gal Y, Isolation of an acid fraction from a fish protein hydrolysate
with a CGRP-like biological activity. Biotechnol Appl Biochem
29:87–92 (1999).
45 Martinez Alvarez O, Delannoy C, Guimas L and Fouchereau-Peron M,
TheoccurrenceofCGRPlikemoleculeinasiki(Centroscymnus
coelolepsis) hydrolysate of industrial origin. J Agric Food Chem
55:5469–5475 (2007).
46 Jeon YJ, Byun HG and Kim SK, Improvement of functional properties
of cod frame protein hydrolysates using ultrafiltration membranes.
Process Biochem 35:471–478 (1999).
47 Je JY, Park PJ and Kim SK, Antioxidant activity of a peptide isolated
from Alaska pollack (Theragra chalcogramma) frame protein
hydrolysate. Food Res Int 38:45–50 (2005).
48 Rajapakse N, Purification and in vitro antioxidative effects of giant
squid muscle peptides on free radical-mediated oxidative systems.
JNutrBiochem16:562–569 (2005).
49 Neves RAM, Mira NVM and Marquez UML, Characterization of
enzymatic fish hydrolysates. Cienc Tecnol Aliment 24:101–108
(2004).
50 Sumaya-Martinez MT, Valorisation d’hydrolysats de co-produits de
crevettes : étude de l’activité antiradicalaire et antioxydante,
fractionnement des substances actives et effet de la glycation. (Upgrading
shrimps by-products hydrolysates : study of the antiradical
and antioxidant activity, fractionation of active molecules and effect
of glycation). Doctoral thesis, Université deBretagneOccidentale,
France (in French) (2004).
51 Vandanjon L, Johannsson R, Derouiniot M, Jaouen P and Bourseau P,
Concentration and purification by ultrafiltration of marine peptides
solutions. J Food Eng 83:581–589 (2007).
52 Chabeaud A, Vandanjon L, Bourseau P, Jaouen P and Guérard F,
Fractionation by ultrafiltration of a saithe protein hydrolysate
(Pollachius virens): effect of material and molecular weight cut-off
onthemembraneperformances.J Food Eng 91:408–414 (2009).
53 Chabeaud A, Vandanjon L, Bourseau P, Jaouen P, Chaplain-
Derouiniot M and Guérard F, Performances of ultrafiltration
membranes for fractionating a fish protein hydrolysate: application
to the refining of bioactive peptidic fractions. Sep Purif Technol
66:463–471 (2009).
54 Vandanjon L, Grignon M, Courois E, Bourseau P and Jaouen P,
Fractionating of fish hydrolysates by a cascade of ultrafiltration
and nanofiltration. J Food Eng 95:36–44 (2009).
55 Bourseau P, Vandanjon L, Jaouen P, Chaplain-Derouiniot M, MasséA,
Guérard F, et al, Fractionation of fish protein hydrolysates by
ultrafiltration and nanofiltration: impact on peptidic populations.
Desalination 244:303–320 (2009).
56 Neville DMJ, Isolation of an organ specific protein antigen from cell
surface membrane rat liver. Biochim Biophys Acta 154:540–552
(1968).
57 Lowry OM, Rosenbrough WJ, Farr AL and Randall RJ, Protein
measurement with folin phenol reagent. JBiolChem193:265–275
(1951).
58 Yamaguchi A, Chiba T, Yamatani T, Inui T, Morishita T, Nakamura A,
et al, Calcitonin gene-related peptide stimulates adenylate cyclase
activation via guanine nucleotide-dependent process in rat liver
plasma membranes. Endocrinology 123:2591–2597 (1988).
59 Morales FJ and Jiménez-Pérez S, Free radical scavenging capacity of
Maillard reaction products as related to colour and fluorescence.
Food Chem 72:119–125 (2001).
60 Marco GJ, A rapid method for evaluation of antioxidants. JAmOil
Chem Soc 45:594–598 (1968).
61 Holmquist B, Bunning P and Riordan J, A continuous spectrophotometric
assay for angiotensin converting enzyme. J Anal Chem
95:540–548 (1979).
62 Pihlanto A, Antioxidative peptides derived from milk proteins. Int Dairy
J 16:1306–314 (2006).
63 Cinq-Mars CD and Li-Chan EC, Optimizing angiotensin I-converting
enzyme inhibitory activity of Pacific hake (Merluccius productus)
fillet hydrolysate using response surface methodology and
ultrafiltration. J Agric Food Chem 55:9380–9388 (2007).
64 Kitts DD and Weiler K, Bioactive proteins and peptides from food
sources: applications of bioprocesses used in isolation and recovery.
Curr Pharm Des 9:1309–1323 (2003).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1819–1826

Research Article
Received: 7 September 2009 Revised: 31 March 2010 Accepted: 26 April 2010 Published online in Wiley Interscience: 10 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4021
Seasonal variation in content, chemical
composition and antimicrobial and cytotoxic
activities of essential oils from four Mentha
species
Abdullah I Hussain, a,b Farooq Anwar, b∗ Poonam S Nigam, c
Muhammad Ashraf d,e and Anwarul H Gilani f
Abstract
BACKGROUND: The aim of the present study was to appraise variation in the chemical composition, and antimicrobial
and cytotoxic activities of essential oils from the leaves of four Mentha species – M. arvensis, M. piperita, M. longifolia and
M. spicata – as affected by harvesting season. Disc diffusion and broth microdilution susceptibility assays were used to evaluate
the antimicrobial activity of Mentha essential oils against a panel of microorganisms. The cytotoxicity of essential oils was
tested on breast cancer (MCF-7) and prostate cancer (LNCaP) cell lines using the MTT assay.
RESULTS: The essential oil contents of M. arvensis, M. piperita, M. longifolia and M. spicata were 17.0, 12.2, 10.8 and 12.0 g kg −1
from the summer and 9.20, 10.5, 7.00 and 9.50 g kg −1 from the winter crops, respectively. Gas chromatographic–mass
spectrometric analysis revealed that mostly quantitative rather than qualitative variation was observed in the oil composition
of each species. The principal chemical constituents determined in M. arvensis, M. piperita, M. longifolia and M. spicata essential
oils from both seasons were menthol, menthone, piperitenone oxide and carvone, respectively. The tested essential oils and
their major components exhibited notable antimicrobial activity against most of the plant and human pathogens tested. The
tested essential oils also exhibited good cytotoxicity potential.
CONCLUSION: Of the Mentha essential oils tested, M. arvensis essential oil showed relatively better antimicrobial and cytotoxic
activities. A significant variation in the content of most of the chemical components and biological activities of seasonally
collected samples was documented.
c○ 2010 Society of Chemical Industry
Keywords: Mentha arvensis; Mentha piperita; Mentha longifolia; Mentha spicata; antimicrobial; antioxidant; menthol; carvone
INTRODUCTION
The genus Mentha (family Lamiaceae), comprising more than
25 species, grows widely throughout the temperate regions
of the world. 1 Mentha arvensis, M. piperita, M. longifolia and
M. spicata, commonly known as menthol mint, peppermint, wild
mint and spearmint, respectively, are frequently cultivated in
many countries of East Asia, Europe, America and Australia for
the production of essential oils. 1,2 The essential oils and extracts
from Mentha species have been in use since ancient times for the
treatment of many digestive tract diseases and in cuisines. 3
The essential oils of some Mentha species, including M. arvensis,
M. piperita, M. longifolia and M. spicata, are potential candidates
for exhibiting antimicrobial, antioxidant, radical-scavenging and
cytotoxic activities. 1,2,4,5 Such multiple biological activities of
Mentha essential oils might be ascribed to the presence of
some chemical components, such as menthone, piperitone oxide,
camphor and linalool. 6–8
The chemical composition of plants is known to be influenced
by several external factors including climate, as some compounds
may be accumulated at a particular period to respond to
environmental changes. 9 The chemical composition of the
essential oils from plants is thus subject to quantitative and
qualitative variations. Biological activity which is dependent on
∗ Correspondence to: Farooq Anwar, Department of Chemistry and Biochemistry,
University of Agriculture, Faisalabad 38040, Pakistan.
E-mail: fqanwar@yahoo.com
a Department of Chemistry, GC University, Faisalabad 38040, Pakistan
b Department of Chemistry and Biochemistry, University of Agriculture,
Faisalabad 38040, Pakistan
c Centre of Molecular Biosciences, Institute of Biomedical Sciences Research,
University of Ulster, Coleraine BT52 1SA, UK
d Department of Botany, University of Agriculture, Faisalabad 38040, Pakistan
e King Saud University, Riyadh, Saudi Arabia
f Department of Biological and Biomedical Sciences, Aga Khan University Medical
College, Karachi 74800, Pakistan
J Sci Food Agric 2010; 90: 1827–1836 www.soci.org c○ 2010 Society of Chemical Industry
1827

1828
the chemical composition is similarly subject to variation. Plant
material collected at different times of the year may contain
different novel compounds with other bioactivities. 10
The effects of seasonal variations on the chemical and biological
characteristics of some essential oils of the family Lamiaceae
have been reported in the literature. 8,11,12 However, no earlier
reports are available on the detailed chemical composition and
biological activities of the essential oils from M. arvensis, M. piperita,
M. longifolia and M. spicata native to the subcontinental region. In
the present study, we investigated for the first time the effects of
two different harvesting seasons on the biosynthesis of essential
oils extracted from the leaves of four Mentha species native
to Pakistan, and subsequently their impact on cytotoxicity and
antimicrobial potentials.
EXPERIMENTAL
Chemicals
Dimethylsulfoxide, 3-(4,5-dimethylthiazol-2,5-diphenylte trazolium
bromide) (MTT), penicillin/streptomycin solution, homologous
series of C9 –C24 n-alkanes and various reference chemicals
used in this study were obtained from Sigma Chemical Co. (St
Louis, MO, USA). All other chemicals (analytical grade) used in this
study were purchased from Merck (Darmstadt, Germany), unless
stated otherwise. All culture media and standard antibiotic discs
were purchased from Oxoid Ltd (Basingstoke, UK).
Plant materials
The leaves of wild M. longifolia (L.) Huds. were assayed from three
wild populations in the periphery of Gilgit Valley of the North-West
Frontier Province of Pakistan (NWFP), whereas the leaf samples
of M. arvensis L., M. piperita L. and M. spicata (L.) Huds. were
collected from cultivated fields of the Horticulture and Botanical
Gardens, University of Agriculture, Faisalabad, Pakistan. Samples
were collected twice: during summer (April–May) and winter
(October–November) 2007. The mean values for maximum and
minimum temperature ( ◦ C) for the months of April–May and
October–November 2007 in the Gilgit Valley were 30.5 ± 2.9,
15.5 ± 2.2 (average 23.0); 28.1 ± 3.3, 8.9 ± 1.6 (average 18.5),
respectively, whereas those in the University of Agriculture,
Faisalabad region were 40.0 ± 4.2, 24.0 ± 3.7 (average 32.0);
30.8 ± 6.1, 17.2 ± 3.8 (average 24.0), respectively. The average
relative humidity and total rainfall in the months of April–May and
October–November 2007 in the Gilgit Valley were 26.5 ± 10.0%
and 14.3 mm; 46.2 ± 11.3% and 8.2 mm, respectively and those in
the University of Agriculture, Faisalabad region were 23.5 ± 8.3%
and 16.6 mm; 48.0 ± 15.2% and 10.2 mm, respectively. Three
different samples for each species during each of the harvesting
season were assayed. The plant specimens were identified
and authenticated by Dr Mansoor Hameed, taxonomist of the
Department of Botany, University of Agriculture, Faisalabad,
Pakistan. Further authentication was made by comparison with
authentic vouchers of M. arvensis (No. 1003), M. piperita (No. 2212),
M. longifolia (No. 8095) and M. spicata (No. 9058) deposited in
the Herbarium of the Botany Department of the University of
Agriculture, Faisalabad, Pakistan. The leaf samples were dried at
30 ◦ C in a hot air-oven (IM-30 m, IRMECO, Geesthacht, Germany)
to constant weight.
Extraction of essential oils
The dried leaves of M. arvensis, M. piperita, M. longifolia and
M. spicata were ground prior to the operation and then 100 g
www.soci.org AI Hussain et al.
of samples were subjected to hydro-distillation for 3 h using a
Clevenger-type apparatus. 8 The distilled essential oils were dried
over anhydrous sodium sulfate, filtered and stored at −4 ◦ Cuntil
analysis. The yields (g kg −1 ) of the oils were calculated on a
moisture-free basis.
Chemical composition of essential oils
Gas chromatography (GC)
The essential oils were analysed using a gas chromatograph
(model 8700, PerkinElmer, Norwalk, CT, USA) equipped with
flame ionization detector (FID) and HP-5 MS capillary column
(30 m × 0.25 mm, film thickness 0.25 µm). Injector and detector
temperatures were set at 220 and 290 ◦ C, respectively. Column
oven temperature was programmed from 80 to 220 ◦ Catarate
of 4 ◦ Cmin −1 ; initial and final temperatures were held for 3 and
10 min, respectively. Helium was used as a carrier gas with a flow
rate of 1.5 mL min −1 .Asampleof1.0 µL was injected using the
split mode (split ratio 1 : 100). All quantification was done using
a built-in data-handling program provided by the manufacturer
of the gas chromatograph (PerkinElmer). The composition was
reported as relative percentage of the total peak area.
Gas chromatography–mass spectrometry (GC-MS)
GC-MS analysis of the essential oils was performed using an
Agilent Technologies (Little Falls, CA, USA) 6890N network GC
system, equipped with an Agilent Technologies 5975 inert XL mass
selective detector and 7683B series auto-injector. Compounds
were separated on an HP-5 MS capillary column (30 m × 0.25 mm,
film thickness 0.25 µm). A 1.0 µL sample was injected in the
split mode, with a split ratio of 1 : 100. For GC-MS detection, an
electron ionization system with ionization energy of 70 eV was
used. Column oven temperature programme was the same as
in GC analysis. Helium was used as a carrier gas at a flow rate
of 1.5 mL min −1 . Mass range was 50–550 m/z, while the injector
and MS transfer line temperatures were set at 220 and 290 ◦ C,
respectively.
Compound identification
Identification of components was based on comparison of their
mass spectra with those of the NIST mass spectral library 13,14 and
those described by Adam, 15 as well as on comparison of their
retention indices either with those of authentic compounds or
with literature values. 7,15,16
Antimicrobial activity of essential oils
Mentha essential oils were individually tested against a panel of
microorganisms, including three strains of bacteria: Staphylococcus
aureus ATCC 25923, Bacillus subtilis ATCC 10707 and Escherichia
coli ATCC 25922; and seven strains of pathogenic fungi: Aspergillus
flavus ATCC 32612, Alternaria solani ATCC 11078, Fusarium solani
ATCC 36031, Rhizopus solani, Alternaria alternata ATCC 20084,
Aspergillus niger ATCC 10575 and Rhizopus spp. The pure bacterial
and fungal strains were obtained from the Biological Division of the
Nuclear Institute for Agriculture and Biology (NIAB), Faisalabad,
Pakistan. Purity and identity were verified by the Department
of Veterinary Microbiology, University of Agriculture, Faisalabad,
Pakistan. Bacterial strains were cultured overnight (16 h) at 37 ◦ C
in nutrient agar (NA, Oxoid), while the fungal strains were cultured
overnight (16 h) at 30 ◦ C using potato dextrose agar (PDA, Oxoid).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1827–1836

Biological activities of Mentha essnetial oils www.soci.org
Disc diffusion method
The antimicrobial activity of the Mentha essential oils and the
principal components menthol, menthone, limonene and carvone
were determined by the disc diffusion method. 17 Briefly, 100 µL
of suspension in NA and PDA containing 10 8 colony-forming
units (CFU) mL −1 of bacteria cells and 10 4 spores mL −1 of
fungi were spread on to NA and PDA medium, respectively.
The paper discs (6 mm in diameter) were separately impregnated
with 15 µL of essential oils or main components and placed on
the agar, which had previously been inoculated with the selected
test microorganism. Amoxycillin (30 µg per disc) and fluconazole
(30 µg per disc) were used as a positive reference for bacteria
and fungi, respectively while the discs without samples were used
as a negative control. Plates, after 1 h at 4 ◦ C, were incubated at
37 ◦ C for 24 h for bacteria and at 30 ◦ C for 48 h for fungal strains.
Antimicrobial activity was assessed by measuring the diameter
of the growth inhibition zone (IZ) in millimetres (including disc
diameter of 6 mm) for the test organisms compared to controls.
Micro-dilution broth method
For minimum inhibitory concentration (MIC), a micro-dilution
broth susceptibility assay was used, as reported in NCCLS. 18 Microdilution
broth test was performed in nutrient broth (NB, Oxoid) for
bacterial strains and in sabouraud dextrose broth (SDB, Oxoid) for
fungal strains. Essential oils were solubilized in dimethylsulfoxide
(DMSO), then diluted in culture media for use. Dilution series
were prepared from 0.01 to 30.0 mg mL −1 of the essential oils or
their chief components in a 96-well microtitre plate, including one
growth control, solvent control and one sterility control. 160 µL
of nutrient broth and sabouraud dextrose broth for bacteria and
fungi, respectively were added to microplates and 20 µL oftest
solution. Then, 20 µL of5× 10 5 cfu mL −1 (confirmed by viable
count) of standard microorganism suspension were inoculated
onto microplates. The plates were incubated at 37 ◦ Cfor24h
for bacteria, and at 30 ◦ C for 48 h for fungi. Amoxicillin was used
as a reference compound for antibacterial and fluconazole for
antifungal activities. Growth was indicated by the presence of a
white ‘pellet’ on the well bottom. MIC was calculated as the highest
dilution showing complete inhibition of the test strains.
Cytotoxicity of Mentha essential oils
The human breast cancer cell line MCF-7 was maintained in
Dulbecco’s Minimum Essential Medium (DMEM), while hormonedependent
prostate carcinoma LNCaP was cultured in RPMI
1640 medium. Both media were supplemented with 10% heatinactivated
fetal calf serum, 1% L-glutamine and 1% penicillin–streptomycin.
Cells of MCF-7 (10 4 per well) and LNCaP
(10 5 per well) were cultivated in 96-well plates for 24 h before the
Mentha essential oils were added. Essential oils were solubilized
in DMSO, then diluted in culture media for use. The essential oil
dilutions (0.01–0.50 mg mL −1 ) were added to triplicate wells and
cells were incubated for a further 24 h. DMSO was tested as solvent
control, while doxorubicin was used as a reference standard. Cell
viability was assessed by MTT assay and the percentage inhibition
of cell viability was calculated using cells treated with DMSO as
control. 19 The IC50 values (concentration at which 50% of cells were
killed) were calculated from inhibition versus concentration graphs.
Statistical analysis
All the experiments were conducted in triplicate and the data
are presented as mean values ± standard deviation of triplicate
determinations. Statistical analysis of the data was performed
by analysis of variance (ANOVA) using Statistica 5.5 (StatSoft Inc.,
Tulsa, OK, USA) software, and a probability value of P ≤ 0.05
was considered to represent a statistically significance difference
among mean values.
RESULTS AND DISCUSSION
Variation in yield and composition of essential oils
The yield of essential oils of all four Mentha species ranged from
7.00 to 17.0 g kg −1 (w/w) (Table 1). The minimum oil contents
(7.00 g kg −1 ) were found in M. longifolia during winter, while
the maximum yield (17.0 g kg −1 ) of essential oil of M. arvensis
was during summer. All the tested Mentha species demonstrated
higher essential oil yield in summer (when the plants were in full
bloom) than in winter (when the plants reached the end of their
growing cycle). Variations in the content of essential oils with
respect to Mentha species and harvesting seasons were significant
(P < 0.05). Our results are in agreement with those of Kofidis
et al., 11 who also reported higher essential oil yield of M. spicata
samples from late summer crops. In view of another report, the
Thymbra spicata also exhibited maximum essential oil yield during
summer, when the plants were in full bloom. 20
The components found in the essential oils of each Mentha
species at different seasons are reported in Table 1: 23 and 21;
48 and 47; 35 and 36; 33 and 30 compounds were found in
the essential oils of M. arvensis, M. piperita, M. longifolia and
M. spicata from summer and winter crops, respectively. The
major constituents (>5%) in the essential oils of M. arvensis
during summer and winter were found to be menthol (78.90%
and 81.30%) and isomenthone (6.35% and 6.19%), respectively.
The main components of M. piperita essential oils collected
during summer and winter were menthone (28.13% and 25.54%),
menthyl acetate (9.51% and 9.68%), limonene (7.58% and 7.73%)
and isomenthone (4.04% and 7.63%), respectively. Piperitenone
oxide (60.10% and 64.60%), piperitenone (6.37% and 1.97%) and
germacrene D (5.13% and 5.97%), were the main components in
the essential oils of wild-growing M. longifolia during summer and
winter, respectively, while the major constituents of M. spicata
essential oils from summer and winter harvests were determined
to be carvone (59.50% and 63.24%), limonene (10.44% and 9.09%)
and 1,8-cineol (6.36% and 4.51%), respectively. In addition, the
tested Mentha essential oils also contained substantial amounts of
various minor constituents, as detailed in Table 1.
The analysed Mentha essential oils consisted mainly of oxygenated
monoterpenes as a major fraction. Mentha arvensis,
M. piperita, M. longifolia and M. spicata essential oils collected
during summer and winter consisted of 95.71% and 97.06%;
65.39% and 68.37%; 75.96% and 75.31%; 81.48% and 78.33%
oxygenated monoterpenes, respectively. Menthol, menthone,
piperitenone oxide and carvone were the main oxygenated
monoterpenes in M. arvensis, M. piperita, M. longifolia and M. spicata
essential oils during both seasons, respectively. Mentha piperita,
M. longifolia and M. spicata also contained considerable quantities
of monoterpene hydrocarbons (13.95–17.62%; 7.03–7.24%;
9.09–10.44%) and sesquiterpene hydrocarbons (14.55–14.57;
14.25–14.94; 6.12–8.30%), whereas M. arvensis had very low
amounts of these compounds.
The variation in the content of most of the essential oils
investigated in the present study, with respect to species and
harvesting season, was quantitatively significant (P < 0.05). Most
fluctuation in oil components of M. arvensis includes menthone
J Sci Food Agric 2010; 90: 1827–1836 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1829

1830
Table 1. Seasonal variation in content and chemical composition of the essential oils from leaves of four Mentha species
Composition (%) c
Mentha arvensis Mentha piperita Mentha longifolia Mentha spicata
Components a RI b Summer Winter Summer Winter Summer Winter Summer Winter Mode of identification d
Monoterpene hydrocarbons
α-Pinene 939 – – 3.53 ± 0.17b 1.31 ± 0.12a – – – – RT, RI,MS
Camphene 954 – – 0.16 ± 0.07 t – – – – RT, RI, MS
β-Pinene 979 t – 5.70 ± 0.31d 4.30 ± 0.13c 2.01 ± 0.19a 2.42 ± 0.14b – – RT, RI, MS
β-Myrecene 991 – – 0.29 ± 0.04b 0.11 ± 0.02a – – – – RT, RI,MS
4-Carene 1022 – – t 0.05 ± 0.04 – – – – RI, MS
p-Cymene 1025 – – 0.36 ± 0.10a 0.45 ± 0.10a – – – – RT, RI,MS
Limonene 1029 1.18 ± 0.14a 0.93 ± 0.14a 7.58 ± 0.25c 7.73 ± 0.20c 3.74 ± 0.11b 3.20 ± 0.13b 10.44 ± 0.31e 9.09 ± 0.27d RT, RI, MS
cis-β-Ocimene 1037 – – – – 1.11 ± 0.10a 1.42 ± 0.14b – – RT, RI, MS
γ -Terpinene 1060 – – – – 0.06 ± 0.04a 0.06 ± 0.01a – – RT, RI, MS
δ-Terpinene 1089 – – – – 0.11 ± 0.03a 0.14 ± 0.07a – – RT, RI, MS
www.soci.org AI Hussain et al.
Oxygenated monoterpenes
1,8-Cineol 1031 – – 0.90 ± 0.09a 1.06 ± 0.12a – – 6.36 ± 0.36c 3.51 ± 0.26b RT, RI, MS
cis-Sabinene hydrate 1070 – – – – 0.08 ± 0.02a t 0.13 ± 0.04a t RT, RI, MS
Linalool oxide 1088 0.05 ± 0.01 t – – – – – – RT, RI,MS
Linalool 1097 0.37 ± 0.15b 0.17 ± 0.03a 0.61 ± 0.10c 1.48 ± 0.13d 0.89 ± 0.15c 0.77 ± 0.10c 0.71 ± 0.11c 3.13 ± 0.10e RT, RI, MS
trans-p-Mentha-2,8-dienol 1123 – – – – – – 0.18 ± 0.09 t RI, MS
trans-Pinocarveol 1139 0.08 ± 0.05 – – – – – – – RT, RI,MS
Isopulegol 1147 0.45 ± 0.14a 0.43 ± 0.12a 0.53 ± 0.15a 0.63 ± 0.14a – – – – RT, RI,MS
Menthone 1152 4.42 ± 0.13c 1.38 ± 0.40b 28.13 ± 0.90e 25.54 ± 1.19d – – 0.61 ± 0.06a t RT, RI, MS
Isomenthone 1162 6.35 ± 0.16b 9.19 ± 0.64d 4.04 ± 0.23a 7.63 ± 0.33c – – – – RT, RI,MS
Borneol 1169 – – – – 13.3 ± 0.60d 4.36 ± 0.15b 1.38 ± 0.09a 5.90 ± 0.13c RT, RI, MS
Menthol 1172 78.9 ± 1.6b 81.3 ± 2.0b 4.83 ± 0.18a 3.31 ± 0.11a – – – – RT, RI,MS
Terpinene-4-ol 1177 – – 2.66 ± 0.21b – 0.19 ± 0.03a 0.24 ± 0.05a 0.24 ± 0.07a 0.32 ± 0.06a RT, RI, MS
p-Cymen-8-ol 1183 – – – – 0.20 ± 0.07a 0.21 ± 0.05a – – RT, RI, MS
Neoisomenthol 1187 0.54 ± 0.06a 0.53 ± 0.09a 6.53 ± 0.41c 2.23 ± 0.17b – – – – RI, MS
α-Terpineol 1189 0.71 ± 0.11b 0.67 ± 0.12b – 6.13 ± 0.24d 0.19 ± 0.05a 0.07 ± 0.05a 1.32 ± 0.17c 0.49 ± 0.14b RT, RI, MS
Myrtenal 1194 – – – – 0.11 ± 0.05a 0.10 ± 0.05a – – RT, RI, MS
Estragole 1195 0.27 ± 0.08a 0.29 ± 0.05a – – – – – – RT, RI,MS
Dihydrocarveol 1197 – – – – – – 2.45 ± 0.17 t RT, RI, MS
γ -Terpineol 1199 – – 2.34 ± 0.11a 2.61 ± 0.10a – – – – RT, RI,MS
cis-Dihydrocarvone 1200 – – – – – – 4.84 ± 0.19b 1.47 ± 0.09a RT, RI, MS
trans-Dihydrocarvone 1203 – – – – – – 0.27 ± 0.05 t RT, RI, MS
trans-Carveol 1217 – – – – – – 0.34 ± 0.05 – RT, RI, MS
cis-Carveol 1229 – – 0.18 ± 0.05a t – – 2.33 ± 0.10b t RT, RI, MS
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1827–1836

Biological activities of Mentha essnetial oils www.soci.org
Table 1. (Continued)
Composition (%) c
Mentha arvensis Mentha piperita Mentha longifolia Mentha spicata
Components a RI b Summer Winter Summer Winter Summer Winter Summer Winter Mode of identification d
Pulegone 1237 0.75 ± 0.05b 0.35 ± 0.05a 4.42 ± 0.19e 6.42 ± 0.13f 3.91 ± 0.07d 2.10 ± 0.05c – – RT, RI, MS
trans-Carane 1240 1.58 ± 0.20a 1.54 ± 0.13a – – – – RI, MS
Carvone 1243 0.06 ± 0.05a 0.06 ± 0.06a 0.20 ± 0.05a 0.90 ± 0.09a – 0.08 ± 0.05a 59.5 ± 1.2b 63.24 ± 1.9c RT, RI, MS
Carvone oxide 1247 – – – – – – 0.26 ± 0.05 – RI, MS
Pipretone 1253 1.01 ± 0.05a 1.00 ± 0.06a – – – – – – RT, RI, MS
Bornyl acetate 1289 – – – – 0.31 ± 0.05a 0.34 ± 0.05a 0.34 ± 0.05a – RI, MS
Thymol 1290 – – – – 0.18 ± 0.06a 0.20 ± 0.05a – – RT, RI, MS
Menthyl acetate 1301 0.17 ± 0.05a 0.15 ± 0.06a 9.51 ± 0.18b 9.68 ± 0.29b – – – – RI, MS
Isopulegyl acetate 1263 – – 0.35 ± 0.10b 0.30 ± 0.05a – – – – RT, RI, MS
Piperitenone 1343 – – 0.15 ± 0.05a 0.45 ± 0.05b 16.4 ± 0.18d 1.97 ± 0.10c t 0.27 ± 0.07a RT, RI, MS
Thymol acetate 1352 – – – – 0.12 ± 0.05a 0.27 ± 0.06b – – RT, RI, MS
cis-Carvyl acetate 1368 – – – – – – 0.22 ± 0.05 t RT, RI, MS
Piperitenone oxide 1370 – – – – 40.1 ± 1.1a 64.6 ± 1.6b – – RI, MS
Sesquiterpene hydrocarbons
α-Ylangene 1372 – – 0.31 ± 0.07a 0.30 ± 0.05a – – – – RT, RI, MS
α-Copaene 1377 – – 0.41 ± 0.05b 0.22 ± 0.10a 0.81 ± 0.05c 0.41 ± 0.07b – – RT, RI, MS
β-Bourbonene 1388 0.23 ± 0.12a 0.17 ± 0.05a 1.82 ± 0.10f 1.12 ± 0.14c 0.61 ± 0.08b 0.31 ± 0.09a 0.93 ± 0.15c 0.56 ± 0.07b RT, RI, MS
β-Elemene 1391 – – 0.84 ± 0.09a 0.99 ± 0.06a – – – – RT, RI, MS
cis-Jasmone 1393 – – 0.15 ± 0.05a 0.95 ± 0.10c 1.11 ± 0.17c 2.16 ± 0.06d t 0.34 ± 0.05b RT, RI, MS
Longifolene 1409 – – 0.45 ± 0.05a 0.49 ± 0.06a – – – – RT, RI, MS
α-Gurjunene 1410 – – – – 0.34 ± 0.05b 0.24 ± 0.05a – – RT, RI, MS
β-Caryophyllene 1421 – – 3.55 ± 0.21c 2.55 ± 0.10b 4.22 ± 0.13d 2.53 ± 0.20b 2.75 ± 0.14b 2.02 ± 0.10a RT, RI, MS
β-Cubebene 1423 – – 0.61 ± 0.06b 0.46 ± 0.05a – – – – RT, RI, MS
Thujopsene 1426 – – 0.36 ± 0.05a 0.33 ± 0.05a – – – – RI, MS
β-Copaene 1434 – – – – 0.17 ± 0.05a 0.13 ± 0.03a – – RT, RI, MS
α-Bergamotene 1439 – – – – – – t 1.35 ± 0.08 RT, RI, MS
Aromadendrene 1441 – – 0.49 ± 0.06a 0.69 ± 0.05b – – – – RT, RI, MS
α-Caryophyllene 1455 – – 0.71 ± 0.05b 0.31 ± 0.07a 0.62 ± 0.07b 0.30 ± 0.06a 0.75 ± 0.05b 0.32 ± 0.05a RT, RI, MS
γ -Muurolene 1480 – – 0.95 ± 0.08c 0.99 ± 0.06c 0.57 ± 0.05b 0.94 ± 0.09c 0.29 ± 0.07a 0.59 ± 0.04b RT, RI, MS
Germacrene D 1485 – – 0.69 ± 0.07a 1.69 ± 0.15b 5.13 ± 0.17c 5.97 ± 0.23f 0.83 ± 0.10a 1.40 ± 0.15b RT, RI, MS
Ledene 1498 – – 0.51 ± 0.07a 0.71 ± 0.06b – – – – RI, MS
α-Muurolene 1500 – – 0.73 ± 0.10a 0.77 ± 0.14a – – – – RT, RI, MS
Bicyclogermacrene 1502 – – – – 0.13 ± 0.05a 0.94 ± 0.09c t 0.30 ± 0.08b RT, RI, MS
J Sci Food Agric 2010; 90: 1827–1836 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1831

1832
Table 1. (Continued)
Composition (%) c
Mentha arvensis Mentha piperita Mentha longifolia Mentha spicata
Components a RI b Summer Winter Summer Winter Summer Winter Summer Winter Mode of identification d
Cuparene 1505 – – 0.12 ± 0.05a 0.22 ± 0.06a – – – – RT, RI,MS
Amorphene 1441 – – 0.45 ± 0.07a 0.41 ± 0.12a – – – – RT, RI,MS
β-Bisabolene 1506 – – – – 0.21 ± 0.05a 0.16 ± 0.05a – – RT, RI, MS
γ -Cadinene 1514 – – – – t 0.16 ± 0.05a 0.21 ± 0.06a 0.75 ± 0.05b RT, RI, MS
δ-Cadinene 1523 – – 1.21 ± 0.13a 1.02 ± 0.11a – – – – RT, RI,MS
α-Cadinene 1539 – – 0.13 ± 0.05a 0.33 ± 0.06b – – – – RT, RI,MS
Calamenene 1540 – – – – 0.33 ± 0.07a 0.69 ± 0.12b 0.36 ± 0.09a 0.67 ± 0.15b RT, RI, MS
α-Calacorene 1546 – – 0.08 ± 0.05 – – – – – RT, RI,MS
www.soci.org AI Hussain et al.
Oxygenated sesquiterpenes
Elemol 1550 – – t 0.06 ± 0.01 – – – – RT, RI,MS
Spathulenol 1578 – – 0.47 ± 0.06b 0.78 ± 0.06c 0.09 ± 0.05a 0.46 ± 0.05b t 0.37 ± 0.08b RT, RI, MS
Caryophyllene oxide 1583 0.19 ± 0.05a t 0.98 ± 0.13c 0.24 ± 0.08a 0.42 ± 0.05b 0.13 ± 0.05a 0.71 ± 0.05e 0.38 ± 0.05b RT, RI, MS
α-Cedrol 1596 0.18 ± 0.05a 0.19 ± 0.06a 0.13 ± 0.05a 0.41 ± 0.09b – – RT, RI, MS
1,10-Di-epi-cubenol 1616 – – – – – – 0.22 ± 0.07a 0.59 ± 0.05b RI, MS
γ -Eudesmol 1632 0.26 ± 0.06a 0.17 ± 0.05a – – – – – – RT, RI,MS
α – Muurolol 1646 – – – – 0.27 ± 0.06a 0.51 ± 0.06b 0.56 ± 0.05b 1.63 ± 0.12c RT, RI, MS
β-Eudesmol 1651 0.32 ± 0.05a 0.22 ± 0.07a 0.13 ± 0.05a 0.17 ± 0.05a – – – – RT, RI,MS
α-Cadinol 1654 0.16 ± 0.05 t – – – – – – RT, RI,MS
α-Eudesmol 1656 0.19 ± 0.05b 0.12 ± 0.05a – – – – – – RI, MS
Total 98.24 98.67 99.34 99.31 98.15 99.0 99.53 98.69
Essential oil content (g kg−1 ) 17.0 ± 1.1e 9.20 ± 0.60b 12.2 ± 1.0d 10.5 ± 0.70c 10.8 ± 1.2c 7.00 ± 1.0a 12.0 ± 1.0d 9.50 ± 0.50b
a Compounds are listed in order of elution from an HP-5 MS column.
b Retention indices relative to C9 –C24 n-alkanes on HP-5 MS column.
c Values are mean ± standard deviation of three different samples of each Mentha species, analysed individually in triplicate. Means followed by different letters (a–f) in the same row represent significant
difference (P < 0.05).
d RT, identification based on retention time; RI, identification based on retention index; MS, identification based on comparison of mass spectra.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1827–1836
t, trace (

Biological activities of Mentha essnetial oils www.soci.org
Table 2. Seasonal variation in antimicrobial activity of essential oils from leaves of four Mentha species a
Mentha arvensis Mentha piperita Mentha longifolia Mentha spicata
Tested
microorganisms Summer Winter Summer Winter Summer Winter Summer Winter
Agar disc diffusion methodb Staphylococcus aureus 32 ± 2c 33 ± 2c 16 ± 1a 15 ± 1a 31 ± 2c 29 ± 2bc 25 ± 2b 24 ± 2b
Bacillus subtilis 33 ± 1c 33 ± 2c 17 ± 1a 20 ± 1a 29 ± 2b 31 ± 2bc 29 ± 2b 29 ± 2b
Escherichia coli 14 ± 1ab 16 ± 1b 20 ± 2c 19 ± 2bc 16 ± 1b 17 ± 1b 14 ± 1ab 12 ± 1a
Aspergillus niger 28 ± 1bc 30 ± 2c 30 ± 1c 25 ± 2b 28 ± 1bc 30 ± 2c 25 ± 2b 22 ± 0a
Aspergillus flavus 21 ± 1a 25 ± 2b 22 ± 1a 20 ± 1a 29 ± 1c 30 ± 2c 26 ± 2b 29 ± 1c
Alternaria solani 17 ± 1b 22±1c 14 ± 1a 15 ± 1ab 25 ± 2cd 23 ± 1c 27 ± 1d 23 ± 1c
Fusarium solani 27 ± 2b 26 ± 2b 20 ± 2a 22 ± 2ab 27 ± 2b 25 ± 1b 24 ± 2ab 23 ± 2ab
Rhizopus solani 29 ± 2ab 29 ± 2ab 32 ± 2b 28 ± 2a 30 ± 2bc 29 ± 2ab 28 ± 2a 27 ± 2a
Alternaria alternata 30 ± 2d 30 ± 1d 18 ± 1b 15 ± 1a 29 ± 1d 23 ± 2c 19 ± 1b 17 ± 1ab
Rhizopus spp. 16 ± 0b 18 ± 1c 12 ± 0a 11 ± 0a 19 ± 1cd 19 ± 1cd 20 ± 1d 16 ± 1b
Minimum inhibitory concentration (MIC, µg mL−1 )
Staphylococcus aureus 30.5 ± 1.5b 20.0 ± 1.1a 120.6 ± 3.1e 120.3 ± 3.7e 32.3 ± 1.6b 40.2 ± 2.0c 78.0 ± 4.1d 80.0 ± 3.0d
Bacillus subtilis 20.3 ± 1.0a 20.5 ± 0.6a 123.4 ± 5.8e 113.4 ± 4.5d 45.9 ± 1.6c 40.5 ± 1.6c 25.4 ± 1.1b 25.3 ± 1.1b
Escherichia coli 330.3 ± 9.3bc 327.0 ± 6.1b 310.4 ± 7.2a 310.3 ± 7.5a 320.3 ± 9.2ab 310.0 ± 8.8a 345.3 ± 10.1c 349.2 ± 8.2c
Aspergillus niger 63.5 ± 3.2b 58.3 ± 2.3ab 49.5 ± 3.0a 59.3 ± 2.7ab 82.5 ± 3.3 82.3 ± 3.5c 89.1 ± 5.1c 107.2 ± 4.3d
Aspergillus flavus 110.7 ± 5.5d 83.3 ± 3.0c 122.0 ± 4.2e 127.4 ± 7.3e 60.3 ± 3.1a 59.3 ± 2.2a 81.4 ± 4.2c 69.3 ± 2.7b
Alternaria solani 129.0 ± 7.7d 111.4 ± 5.2c 127.1 ± 5.0d 129.0 ± 7.2d 86.1 ± 2.2ab 93.7 ± 3.0b 81.3 ± 4.8a 101.5 ± 5.3bc
Fusarium solani 89.8 ± 2.5ab 80.6 ± 2.7a 130.7 ± 6.7d 119.3 ± 4.7c 81.3 ± 2.7a 88.7 ± 4.7ab 89.7 ± 5.3ab 92.4 ± 4.6b
Rhizopus solani 63.9 ± 3.2c 65.1 ± 3.1c 44.1 ± 2.5a 53.7 ± 2.2b 52.9 ± 2.1b 61.5 ± 3.0c 53.2 ± 3.1b 59.0 ± 3.0bc
Alternaria alternata 57.9 ± 3.3a 56.2 ± 2.2a 117.0 ± 7.2d 131.0 ± 4.2e 82.5 ± 3.0b 99.4 ± 4.2c 122.0 ± 7.3d 133.1 ± 6.6e
Rhizopus spp. 137.1 ± 5.5c 139.0 ± 4.2c 149.7 ± 8.2cd 157.8 ± 8.0d 125.0 ± 4.9b 130.1 ± 6.2bc 103.3 ± 4.1a 130.0 ± 7.8bc
a Values are mean ± standard deviation of three different samples of each Mentha species, analysed individually in triplicate. Means followed by
different letters in the same row represent significant difference (P < 0.05).
b Diameter of inhibition zone (mm) including disc diameter of 6 mm.
(4.42–1.38%) and isomenthone (6.35–9.19%) from summer and
winter crops, respectively. The major variation in M. piperita
essential oil includes isomenthone (4.04–7.63%), terpinene-
4-ol (2.66%–trace), neoisomenthol (6.53–2.23%), α-terpineol
(trace–6.13%) and pulegone (4.42–6.42%), respectively. In essential
oil from M. longifolia, the major variation observed
was in borneol (13.3–4.36%), piperitenone (16.4–1.97%) and
piperitenone oxide (40.1–64.6%). Prominent variations in the
essential oil components of M. spicata were observed in 1,8-cineol
(6.36–3.51%), linalool (0.71–3.13%), borneol (4.84–1.47%), and
cis-dihydrocarvone (4.84–1.47%), respectively.
The yield of most of the compounds was higher in summer,
while isomethane was found to be high in winter. The yield of
pulegone was dependent on season, though it did not follow the
same sequence for all plants.
The variations in chemical composition of the essential oils
with respect to season might have been due to the influence of
phenological status, and environmental conditions can influence
the regulation of the biosynthesis of essential oil. 21 Previous
investigations have demonstrated that harvesting season can
alter the chemical composition of the essential oils of M. spicata,
M. pulegium and Ocimum basilicum. 8,11,20 Our results extend this
understanding in the way that harvesting season can also alter
composition of M. arvensis, M. piperita and M. longifolia essential
oil. There are some reports in the literature on the qualitative and
quantitative analyses of some Mentha essential oils from different
countries, 1,2,22,23 but we could not find a single report showing the
seasonal variation on the essential oils composition and biological
activity of these four Mentha species. Therefore the results are
different from others and indicate that this study is by no means a
repetition.
Antimicrobial activity
The antimicrobial activities of Mentha essential oils and main
components were assessed against a panel of plant and
human pathogenic and food-borne microorganisms. As seen
in Tables 2 and 3, the essential oils of M. arvensis, M. piperita,
M. longifolia and M. spicata exhibited excellent antimicrobial
activity against all the microorganisms tested. The results from
the disc diffusion method followed by MIC indicated that M.
arvensis showed maximum antimicrobial activity with larger IZ
(14–33 and 16–30 mm) and smallest MIC values (20.0–330.3
and 56.2–139.0 µgmL −1 ) against selected strains of bacteria and
fungi, respectively. Mentha piperita, M. longifolia and M. spicata
also exhibited good antimicrobial activities with IZ 15–20, 16–31,
12–29 mm and 11–32, 19–30, 16–29 mm against bacterial and
fungal strains, respectively. The MIC values of these oils were
113.4–310.4, 32.3–320.3, 25.3–349.2 µgmL −1 and 44.1–157.8,
52.9–130.1, 53.2–133.1 µgmL −1 , against strains of bacteria and
fungi, respectively. Overall, all the tested essential oils showed
high antibacterial activity against the Gram-positive bacteria
tested, i.e. Staphylococcus aureus and Bacillus subtilis. Among
the plant pathogenic fungi tested, Rhizopus solani, Aspergillus
niger and Alternaria alternata were the most sensitive strains,
while Rhizopus spp. was the most resistant strain tested. The
variation in antimicrobial activities of Mentha essential oils with
respect to species was statistically significant (P < 0.05). Seasonal
J Sci Food Agric 2010; 90: 1827–1836 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1833

1834
Table 3. Antimicrobial activity of pure components a
www.soci.org AI Hussain et al.
Tested microorganisms Menthol Menthone Piperitenone oxide Carvone Standard drugs b
Agar disc diffusion methodc Staphylococcus aureus 32 ± 1c 25 ± 1b 22 ± 1a 23 ± 2ab 31 ± 2c
Bacillus subtilis 32 ± 2b 30 ± 2b 24 ± 1a 25 ± 1a 33 ± 2b
Escherichia coli 29 ± 2c 12 ± 1ab 8 ± 1a 14 ± 1b 30 ± 2c
Aspergillus niger 29 ± 2b 26 ± 1ab 26 ± 1ab 24 ± 1a 30 ± 2b
Aspergillus flavus 19 ± 1a 19 ± 0a 22 ± 1b 20 ± 1a 28 ± 1c
Alternaria solani 21 ± 1c 17 ± 0b 21 ± 0c 16 ± 1ab 15 ± 1a
Fusarium solani 28 ± 2b 24 ± 2ab 21 ± 1a 28 ± 2b 30 ± 2b
Rhizopus solani 33 ± 2c 28 ± 2b 22 ± 1a 28 ± 2b 32 ± 2c
Alternaria alternate 28 ± 2c 18 ± 1a 23 ± 1b 22 ± 1b 23 ± 1b
Rhizopus spp. 24 ± 1c 11 ± 0a 21 ± 0b 23 ± 1c 22 ± 2b
Minimum inhibitory concentration (MIC, µg mL−1 )
Staphylococcus aureus 30.0 ± 1.5a 90.9 ± 4.5c 80.0 ± 7.2b 80.3 ± 3.2b 10.2 ± 1.1a
Bacillus subtilis 30.3 ± 1.2a 40.5 ± 2.1b 72.3 ± 2.9d 60.5 ± 3.5c 10.3 ± 1.5a
Escherichia coli 80.5 ± 3.8a 189.1 ± 7.5c 310.2 ± 9.9d 170.4 ± 7.4b 69.5 ± 3.0a
Aspergillus niger 78.3 ± 3.2b 88.0 ± 3.3cd 79.5 ± 5.5b 96.6 ± 5.3d 50.4 ± 2.7a
Aspergillus flavus 107.7 ± 3.2c 103.2 ± 4.1bc 97.3 ± 3.8b 107.1 ± 3.3c 80.2 ± 2.5a
Alternaria solani 101.2 ± 3.9b 123.3 ± 4.9c 113.0 ± 6.7b 133.5 ± 3.1d 90.0 ± 5.0a
Fusarium solani 69.3 ± 2.7b 97.3 ± 3.8c 103.3 ± 8.1c 79.0 ± 8.0b 40.2 ± 2.4a
Rhizopus solani 30.8 ± 1.2b 60.4 ± 2.2c 99.7 ± 9.3d 59.4 ± 2.1c 10.4 ± 1.6a
Alternaria alternate 72.3 ± 3.0a 110.0 ± 4.8d 81.0 ± 3.5b 94.3 ± 4.6c 90.1 ± 4.2c
Rhizopus spp. 90.0 ± 5.2a 204.1 ± 7.6b 97.2 ± 7.7a 89.0 ± 4.9a 100.0 ± 5.7a
a Values are mean ± standard deviation of three different experiments. Means followed by different letters (a–d) in the same row represent significant
difference (P < 0.05).
b Amoxycillin for bacterial and fluconazole for fungal strains.
c Diameter of inhibition zone (mm) including disc diameter of 6 mm.
variation exercised notable effects on the antimicrobial activity
of Mentha essential oils, but the trends observed were quite
inconsistent.
In our previous study, we investigated the variation in
antimicrobial activity of Ocimum basilicum essential oils with
respect to season. 8 Our results are in good agreement with the
findings of Wannissorn et al., 24 who reported that M. arvensis
essential oil exhibited good antimicrobial activity against a wide
range of microorganisms. Our findings are contrary to those
of Aridogan et al., 25 who recorded no antimicrobial activity of
M. piperita essential oils against E. coli. This difference could
be due to differences in chemical composition of the oils.
In other reports 3,4 an essential oil from M. piperita exhibited
good antimicrobial activity against both S. aureus and E. coli
strains. There are also some reports in the literature on the
antimicrobial activities of essential oils from M. longifolia and
M. spicata. 1,26,27
Menthol, menthone, piperitenone oxide and carvone, the major
constituents of M. arvensis, M. piperita, M. longifolia and
M. spicata, respectively, were also tested for their potential
antimicrobial activity (Table 3). Menthol, the major component
of M. arvensis, exhibited excellent antimicrobial activity
(IZ 19–33 mm; MIC 30.0–107.7 µgmL −1 ) – even stronger than
standard drugs (amoxycillin and fluconazole) – against all the
strains tested. Menthone, piperitenone oxide and carvone also
showed considerable antimicrobial activities, with 11–30, 8–26,
14–28 mm inhibition zones and MIC values of 40.5–204.1,
72.3–310.2, 59.4–170.4 µg mL −1 against the selected microorganisms.
The excellent antimicrobial activity of the essential oil of
M. arvensis could be attributed to the high content of menthol
present in it. Iscan et al. 3 reported that menthol seems
to be the constituent responsible for antimicrobial activity.
Essential oils rich in compounds of known antimicrobial activities
such as menthone, piperitone oxide, carvone and linalool
are widely reported to possess high levels of antimicrobial
activity. 6,7,16,28
Cytotoxicity of Mentha essential oils
The MTT assay is a sensitive, simple and reliable approach used
to evaluate the cytotoxicity of plant-based products. The effect
of increasing amounts of the tested Mentha essential oils on
the cell proliferation of two human cancer cell lines (MCF-7 and
LNCaP) was appraised. The inhibitory effect of Mentha essential
oils on cell viability ranged from 91% to 97% at 0.5 mg mL −1
(data not shown). IC50 values, calculated from the graphs, are
presented in Table 4, indicating that the tested Mentha essential
oils showed prominent cytotoxic activity against both cancer cell
lines. The cytotoxicity of M. longifolia essential oils against MCF-
7(IC50 45.2 and 50.6 µg mL −1 for summer and winter samples,
respectively) and LNCaP (IC50 43.5 and 52.0 µgmL −1 for summer
and winter samples, respectively) were notably stronger than
those of other oils (Table 4). Among the two cancer cell lines
employed, LNCaP was more sensitive than MCF-7. ANOVA showed
significant (P < 0.05) variation in toxicity with respect to species
and season. According to published guidelines, IC50 < 10 µgmL −1
represents potentially ‘very toxic’; IC50 10–100 µgmL −1 represents
‘potentially toxic’; IC50 100–1000 µg mL represents ‘potentially
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1827–1836

Biological activities of Mentha essnetial oils www.soci.org
Table 4. Seasonal variation in cytotoxicity of essential oils from leaves
of four Mentha species a
Cytotoxic activity IC50 (µgmL −1 )
Species Season MCF-7 LN-CaP
Mentha arvensis Summer 55.3 ± 1.9d 50.2 ± 2.7c
Winter 59.7 ± 2.2e 55.7 ± 1.5d
Mentha piperita Summer 75.2 ± 2.9f 90.4 ± 3.7f
Winter 80.8 ± 3.2g 95.7 ± 4.5f
Mentha longifolia Summer 45.2 ± 2.0b 43.5 ± 2.1b
Winter 50.6 ± 2.0c 52.0 ± 3.0cd
Mentha spicata Summer 80.0 ± 2.4g 75.8 ± 2.3e
Winter 80.6 ± 2.0g 90.0 ± 3.0f
Doxorubicin 28.8 ± 1.2a 33.3 ± 1.1a
a Values are mean ± standard deviation of three different samples of
each Mentha species, analysed individually in triplicate. Means followed
by different letters (a–g) in the same column represent significant
difference (P < 0.05).
harmful’ and IC50 > 1000 µgmL −1 represents ‘potentially nontoxic’.
29
CONCLUSION
In general, harvesting season affected the chemical composition as
well as the biological activities of Mentha essential oils. The results
of the present study indicate that Mentha essential oils possess
very good antimicrobial and cytotoxic potential. A further study
under in vivo conditions is recommended to further elaborate the
biological activities of Mentha essential oils for various applications.
The investigated essential oils of Mentha species may be used for
the preservation of processed foods as well as pharmaceutical
and natural therapies for the treatment of infectious diseases in
humans and plants, and the information observed on seasonal
variation may be useful in selecting the best season for optimal
yield.
ACKNOWLEDGEMENTS
We would like to extend our special gratitude to Professor Dr
MI Bhanger, Director, National Center of Excellence in Analytical
Chemistry (NCEAC), University of Sindh, Jamshoro, Pakistan, for
providing the GC–MS instrumental facility. Thanks are due to
the Higher Education Commission for the award of a scholarship
to AI Hussain under the scheme ‘International Research Support
Initiative Program (IRSIP)’ to conduct a part of the project work in
the UK.
REFERENCES
1 GulluceM,ShainF,SokmenM,OzerH,DafereraD,SokmenA, et al,
Antimicrobial and antioxidant properties of the essential oils and
methanol extract from Mentha longifolia L. spp. longifolia. Food
Chem 103:1449–1456 (2007).
2 Pandey AK, Rai MK and Acharya D, Chemical composition and
antimycotic activity of the essential oils of corn mint (Mentha
arvensis) andlemongrass(Cymbopogon flexuosus) againsthuman
pathogenic fungi. Pharm Biol 41:421–425 (2003).
3 Iscan G, Kirimer N, Kurkcuoglu M, Baser KHC and Demirci F,
Antimicrobial screening of Mentha piperita essential oils. J Agric
Food Chem 50:3943–3946 (2002).
4 Yadegarinia D, Gachkar L, Rezaei MB, Taghizadeh M, Astaneh SA and
Rasooli I, Biochemical activities of Iranian Mentha piperita L. and
Mentha communis L. essential oils. Phytochemistry 67:1249–1255
(2006).
5 Kaur C and Kapoor HC, Anti-oxidant activity and total phenolic content
of some Asian vegetables. Int J Food Sci Technol 37:153–161 (2002).
6 Bendaoud H, Bouajila J, Rhouma A, Savagnac A and Romdhane M,
GC/MS analysis and antimicrobial and antioxidant activities of
essential oil of Eucalyptus radiate. J Sci Food Agric 89:1292–1297
(2009).
7 Mimica-Dukic N, Bozin B, Sokovic M, Mihajlovic B and Matavulj M,
Antimicrobial and antioxidant activities of three Mentha species
essential oils. Planta Med 69:413–419 (2003).
8 Hussain AI, Anwar F, Sherazi STH and Przybylski R, Chemical
composition, antioxidant and antimicrobial activities of basil
(Ocimum basilicum) essential oils depends on seasonal variations.
Food Chem 108:986–995 (2008).
9 KoenenEV, Medicinal Poisonous and Edible Plants in Namibia. Klaus
Hess, Berlin (2001).
10 Eloff JN, It is possible to use the herbarium specimens to screen for
antibacterial activity components in some plants. J Ethnopharmacol
67:355–360 (1999).
11 Kofidis G, Bosabalidis A and Kokkini S, Seasonal variation of essential
oils in a linalool-rich chemotype of Mentha spicata grown wild in
Greece. J Essent Oil Res 16:469–472 (2004).
12 Celiktas OY, Kocabas EEH, Bedir E, Sukan FV, Ozek T and Baser KHC,
Antimicrobial activities of methanol extracts and essential oils
of Rosmarinus officinalis, depending on location and seasonal
variations. Food Chem 100:553–559 (2007).
13 Massada Y, Analysis of Essential Oils by Gas Chromatography and Mass
Spectrometry. Wiley, New York (1976).
14 Mass Spectral Library, NIST/EPA/NIH. National Institute of Standards
and Technology, Gaithersburg, MD (2002).
15 Adam RP, Identification of essential oils components by gas
chromatography/quadrupole mass spectroscopy. Allured Publishing,
Carol Stream, IL (2001).
16 Vagionas K, Graikou K, Ngassapa O, Runyoro D and Chinou I,
Composition and antimicrobial activity of the essential oils of three
Satureja species growing in Tanzania. Food Chem 103:319–324
(2007).
17 NCCLS, Performance Standards for Antimicrobial Disc Susceptibility Test
(6th edn). Approved Standard M2-A6, National Committee for
Clinical Laboratory Standards, Wayne, PA (1997).
18 NCCLS, Performance Standards for Antimicrobial Disc Susceptibility
Testing (9th int. suppl.). M100-S9, National Committee for Clinical
Laboratory Standards, Wayne, PA (1999).
19 Mosmann T, Rapid colorimetric assay or cellular growth and survival:
application to proliferation and cytotoxicity assays. J Immunol Meth
65:55–63 (1983).
20 Muller-Riebau F, Berger BM, Yegen O and Cakir C, Seasonal variation
in the chemical composition of essential oils of selected aromatic
plants growing wild in Turkey. J Agric Food Chem 45:4821–4825
(1997).
21 Masotti V, Juteau F, Bessiere JM and Viano J, Seasonal and
phonological variation of the essential oil from the narrow endemic
species Artemisia molinieri and its biological activities. J Agric Food
Chem 51:7115–7121 (2003).
22 Singh AK, Raina VK, Naqvi AA, Patra NK, Kumar B, Ram P, et al, Essential
oil composition and chemoarrays of menthol mint (Mentha arvensis
L.) cultivars. Flavour Frag J 20:302–305 (2005).
23 Viljoen AM, Petkar S, Van-Vuuren SF, Figueiredo AC, Pedro LG and
Barroso JG, The chemo-geographical variation in essential oil
composition and the antimicrobial properties of wild mint, Mentha
longifolia subsp. polyadena (Lamiaceae) in Southern Africa. J Essent
Oil Res 18:60–65 (2006).
24 Wannissorn B, Jarikasem S, Siriwangchai T and Thubthimthed S,
Antibacterial properties of essential oils from Thai medicinal plants.
Fitoterapia 76:233–236 (2005).
25 Aridogan BC, Baydar H, Kaya S, Demirici M, Ozbasar D and Mumcu E,
Antimicrobial activity and chemical composition of some essential
oils. Arch Pharmacol Res 25:860–864 (2002).
26 Sokovic M and Griensven LJLDV, Antimicrobial activity of essential oils
and their components against the three major pathogens of the
cultivated button mushroom, Agaricus bisporus. Eur J Plant Pathol
116:211–224 (2006).
J Sci Food Agric 2010; 90: 1827–1836 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1835

1836
27 Aggarwal KK, Khanuja SPS, Ahmad A, Kumar TRS, Gupta VK and
Kumar S, Antimicrobial activity profile of the two enantiomers
of limonene and carvone isolated from the oils of Mentha spicata
and Anethum sowa. Flavour Frag J 17:59–63 (2002).
28 Sivropoulou A, Kokkini S, Lanaras T and Arsenakis M, Antimicrobial
activity of mint essential oils. J Agric Food Chem 43:2384–2388
(1995).
www.soci.org AI Hussain et al.
29 Gad-Shayne C, Alternatives to in vivo studies in toxicology, in General
and Applied Toxicology, Vol. 6. John Wiley & Sons Inc. (2009).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1827–1836

Research Article
Received: 3 February 2010 Revised: 23 April 2010 Accepted: 25 April 2010 Published online in Wiley Interscience: 14 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4022
Impact of glycated pea proteins on the activity
of free-swimming and immobilised bacteria
Dominika ´Swia¸tecka, a∗ Henryk Kostyra a and Aleksander ´Swia¸tecki b
Abstract
BACKGROUND: Glycation (non-enzymatic glycosylation), a spontaneously occurring process, is responsible for alteration of the
structures and biological activities of proteins, making them highly active. Regrettably, information regarding the impact of
glycated food proteins on intestinal bacteria still remains sparse. Pea seeds are considered to be a biological material of a high
nutritional value, low content of anti-nutritional substances and proven health-promoting action and therefore they were used
in this study. Since glycated pea proteins are proven to display a lowered susceptibility to the enzymatic digestion, their impact
on the activity of both free-swimming and immobilised bacteria was studied.
RESULTS: In vitro model systems were used to prove the stimulatory impact of glycated pea proteins on the proliferation rate
and survival, as well as on the metabolic activity of free-swimming and immobilised bacteria.
CONCLUSIONS: This phenomenon is of great importance because glycated food proteins are not only a source of nutrients and
energy but also display new properties and increased biological activities. Additionally, they are able to modify the bacterial
intestinal ecosystem, thus affecting the general health status of a consumer.
c○ 2010 Society of Chemical Industry
Keywords: free-swimming bacteria; immobilised bacteria; glycated pea proteins; non-enzymatic glycosylation; proliferation rate; survival
rate; metabolic activity
INTRODUCTION
Legumes have always been considered as a substantial element
of the human diet. Although they are a food consumed mostly by
paupers in the past, they have now gained wide recognition and
are consumed by everyone regardless of social status. Legumes
are a good source of protein, polysaccharides and vitamins, and
hence are used worldwide in both human and animal nutrition. 1,2
The pea (Pisum sativum L.) is a long established and significant
crop with a worldwide production exceeded only by soybeans,
peanuts and dry beans. 3 This legume has extraordinary nutritional
potential derived from its high levels of good quality protein,
antioxidants and fibre, as well as complex carbohydrates with a
low glycaemic index. 4–7 Therefore, pea proteins, which contain
high levels of nutritionally valuable lysine, are recommended
as a constituent of the human diet due to their acknowledged
beneficial effect. Nevertheless, lysine may trigger a spontaneous
non-enzymatic glycosylation process (glycation, the Maillard
reaction) which occurs during storage and processing of food
products. 8–10 This glycosylation process influences the structure
of food proteins, leading to modifications in their properties and
biological functions. 11–13 Interestingly, glycated food proteins
often display a lowered susceptibility to enzymatic degradation
and therefore may at least partially escape digestion and reach
the lower parts of the gastrointestinal tract. That is of great
importance for the intestines, especially the colon, which serves as
an ecosystem of microorganisms. Intestinal bacteria are not only
considered essential for the maintenance of homeostasis of the
local, intestinal environment but also for the general health status
of the consumer. Therefore, an extensive understanding of proper
nutrition ought to include functions of the intestinal ecosystem as
well as prolonged biological activities of modified biomolecules,
such as the glycated food proteins. Drastic insufficiencies in the
findings concerning the above-mentioned issue require studies
with simplified experimental models to understand the basic
interactions between glycated proteins and bacteria as well as
their ensuing implications. For that reason, this study aimed to
determine the impact of glycated pea proteins on various bacteria
in different life forms.
EXPERIMENTAL
Material
Raw peas (Pisum sativum L.) of the Polish variety Ramrod were
delivered from the Production and Experimental Research Station,
located in Ba3cyny, Poland, and were subsequently ground to
flour in a W˙Z-1 grinder (ZBPP Sadkiewicz Instruments, Bydgoszcz,
∗ Correspondence to: Dominika ´Swia¸tecka, Food Immunology and Microbiology
Department at theInstituteof Animal Reproduction andFoodResearch
of the Polish Academy of Sciences, ul. J. Tuwima 10, 10-747 Olsztyn, Poland.
E-mail: d.swiatecka@pan.olsztyn.pl
a Food Immunology and Microbiology Department at the Institute of Animal
Reproduction and Food Research of the Polish Academy of Sciences, ul.
J. Tuwima 10, 10-747 Olsztyn, Poland
b Department of Microbiology at the Faculty of Biology of the University of
Warmia and Mazury, ul. M. Oczapowskiego 1A, 10-957 Olsztyn, Poland
J Sci Food Agric 2010; 90: 1837–1845 www.soci.org c○ 2010 Society of Chemical Industry
1837

1838
Poland) and subsequently sieved through a standard 0.20 mm
sieve (NAGEMA, Magdeburg, Germany).
Protein extraction
Samples (35 g) of the ground pea flour were extracted with
140 mL of 50 mmol L −1 Tris-HCl (pH 8.8) for 1 h at 4 ◦ Cwith
constant agitation and subsequently centrifuged (20 000 × g,
20 min). The extraction was repeated twice. The supernatant,
containing albumins and globulins, was dialysed at 4 ◦ Cfor48h
against distilled water and lyophilised later. 14
Bacterial strains and growth conditions
The strains Bacillus subtilis PCM 1903, Enterococcus faecalis PCM
1861, Enterococcus faecium PCM 1859, Enterobacter aerogenes
PCM 532, Escherichia coli PCM 360, Proteus mirabilis PCM 543,
Pseudomonas aeruginosa PCM 1113, Staphylococcus aureus PCM
517 and Staphylococcus saprophyticus PCM 2109 were obtained
from the Polish Collection of Microorganisms, the Institute of
Immunology and Experimental Therapy of the Polish Academy
of Sciences (Wroc3aw, Poland), whereas Escherichia coli 22 and
Lactobacillus acidophilus 15 were obtained from the Faculty of
Food Science of the University of Warmia and Mazury (Olsztyn,
Poland). The strains were isolated from human faeces.
Cultivation of these species was carried out aerobically in 37 ◦ C
on selected media. Bacteria from the genera Bacillus, Escherichia,
Enterobacter, Pseudomonas, Proteus and Staphylococcus were
grown on a nutrient broth medium (Biocorp, Warsaw, Poland),
whereas bacteria from the genus Enterococcus were grown on a
brain heart infusion medium (BHI; BioMerieux, Warsaw, Poland)
and L. acidophilus weregrownonandeMan,RogosaandSharpe
medium (MRS) medium (BTL, Warsaw, Poland).
Preparation of bacterial inoculum
Particular bacteria were transferred from agar slants into liquid
medium (as described in the previous section) and incubated
for 24/48 h at 37 ◦ C. The overnight cultures were refreshed by
transferring 0.5 mL into freshly prepared liquid media and again
incubated under identical conditions with constant absorbance
measurement until reaching Aλ=550 nm = 0.2. The absorbance was
measured with diode spectrophotometer (DU 7500; Beckman,
Warsaw, Poland) by separating an aliquot of the culture medium.
Such prepared bacterial suspensions were used for inoculation in
further microbial analysis.
The dynamics of bacterial proliferation in a liquid medium
Liquid, 100-fold diluted media for bacterial growth (as described
in the section ‘Bacterial strains and growth conditions’) were
supplemented respectively with the non-glycated pea proteins
and the glycated pea proteins to achieve the final concentration
of 1 mg mL −1 . The cultures were subsequently inoculated with
1% of the particular bacterial strain. The cultures without any
pea protein hydrolysates or pea protein extract supplementation
were treated as controls of bacterial growth rate. Cultivation
was conducted at 37 ◦ C with constant, gentle agitation to
ensure the optimal nutrients usage. Four measurements were
conducted during the whole duration of the experiment and
samples were collected at regular 3 h intervals for fluorescence
analysis. Total bacterial numbers, resulting from the exposition
to the tested pea proteins, were estimated with the use of 4 ′ ,6diamidino-phenylindole
(DAPI). 15 After 15 min incubation with
www.soci.org D ´Swia¸tecka, H Kostyra, A ´Swia¸tecki
DAPI in the dark, at room temperature and with constant, gentle
agitation for optimal fluorochrome accessibility, the samples
were filtrated trough 0.2 µm black polycarbonate filters (Millipore,
Warsaw, Poland), air dried and mounted on a microscopic slide
with a drop of Citifluor (Cargille, Stem-Service, Warsaw, Poland)
and subsequently analysed microscopically. The experiment was
repeated three times for statistical analysis.
The survival rate of bacteria in liquid, mineral medium
Sterile, experimentally chosen salt solutions (0.8% NaCl) were
used to create a nutrient-poor growth environment. These media
were supplemented with the non-glycated pea proteins and the
glycated pea proteins to give the final concentration 1.0 mg mL −1 .
The cultures without any supplementations were treated as
controls of bacterial growth rate and dying out. Subsequently,
the cultures were inoculated with 2% of particular bacterial
strain and incubated at 37 ◦ C for 10 days with constant, gentle
agitation for proper nutrient accessibility and avoidance of clumps
formation. The duration of the experiment was estimated in
a preliminary study, which indicated the increase of artefacts
occurrence after the 10th day of the experiment. In experimentally
predetermined time intervals (0, 4, 7 and 10 days of incubation),
the samples were collected and incubated with the Live/Dead
Bacterial Viability Kit (Molecular Probes, Warsaw, Poland) to
determine the percentage of dead bacterial cells. 16,17 After
15 min of incubation with the kit, the samples were filtered
through 0.2 µm black polycarbonate filters (Millipore), air dried
and mounted on microscopic slides with a drop of BacLight
Mounting Oil (Molecular Probes) and subsequently analysed
microscopically.
Metabolic activity and survival of immobilised bacteria
The bacteria were grown on 0.45 µm nitrocellulose filters
(Millipore) placed on the previously prepared solid, sterile
media (as described in the section ‘Bacterial strains and growth
conditions’) supplemented with the non-glycated pea proteins
and the glycated pea proteins to give the final concentration
of 1.5 mg mL −1 at 37 ◦ C. A higher concentration of analysed
substrates was used due to their hindered diffusion in the solid
medium. Small pieces of the mentioned filters were rinsed with
5-cyano-2,3-ditolyl tetrazolium chloride (CTC; Polyscience, Inc.,
Poznan, Poland) at regular 3 h intervals and incubated for 1 h in
ordertoassesstherateofCTCreduction. 18 The Live/Dead Bacterial
Viability Kit (Molecular Probes) was used to evaluate bacterial
membrane integrity. The CTC was used at the concentration of
5 mmol L −1 in order to exclude the toxic effect of tetrazolium
salts on the bacteria. 19 Later, the filters were washed with distilled
water, air dried and mounted on slides and analysed under an
epifluorescence microscope.
Microscopic analysis of bacteria
The bacterial number, survival rate and metabolic activity were
examined with an epifluorescence microscope Olympus U-RFL-T
(Olympus, Olsztyn, Poland) equipped with filters BP 360 nm, BA
420 nm, DM 400 nm (for the TBN), BP 530–550 nm, BA 590 nm,
DM 570 nm (for the estimation of dead and alive bacteria) and BP
546 nm,BA590DM320(forthemetabolicactivity).Themicroscope
was equipped with the automatic MultiScan Program for image
analysis (Computer Scanning Systems II, Warsaw, Poland).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1837–1845

Impact of pea proteins on types of bacteria www.soci.org
TBN [n x 10 4 ]
TBN [n x 10 4 ]
TBN [n x 10 4 ]
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
(A) (B) (C)
E. coli 22 E. aerogenes L. acidophilus
(D) (E) (F)
E. faecalis E. faecium P. aeruginosa
(G)
P. mirabilis B. subtilis S. saprophyticus
TBN [n x 10 4 ]
(J) (K)
100
E. coli 360 S. aureus
80
60
40
20
0
0 3 6 9
Time [hours]
(H) (I)
0 3 6 9
Time [hours]
Figure 1. The impact of the non-glycated and the glycated pea proteins on the proliferation rate of bacteria: (A) E. coli 22, (B) E. aerogenes,(C)L. acidophilus,
(D) E. faecalis,(E)E. faecium,(F)P. aeruginosa,(G)P. mirabilis,(H)B. subtilis,(I)S. saprophyticus,(J)E. coli 360, and (K) S. aureus.( ), The control culture;
(- - - - ), the culture supplemented with pea proteins; ( .... ), the culture supplemented with glycated pea proteins. TBN, total bacterial number.
Statistical analysis
The results obtained were analysed statistically with the Statistica 8
Program (StatSoft Poland, Cracow, Poland). The standard error was
used to demonstrate the results obtained. Each mean value for the
bacteria per cell represents the results of three experiments. The
statistical significance was determined by the variance analysis
using the F distribution by Fisher.
RESULTS
The impact of the non-glycated (N) and the glycated (G) pea
proteins on the proliferation rate of the referential bacterial strains
in liquid cultures is presented in Fig. 1. The experiment confirmed
the stimulatory influence of the glycated pea proteins (G) on the
proliferation activity of all bacteria (Fig. 1). The observed effect
did not depend on the bacterial cell wall type (gram positive and
negative bacteria). A stronger stimulatory impact of the glycated
pea proteins than that of the non-glycated ones C < N < G) was
observedinthecaseof E.aerogenes(Fig. 1B),L.acidophilus(Fig. 1C),
E. faecalis (Fig. 1D), P. aeruginosa (Fig. 1F), P. mirabilis (Fig. 1G),
S. saprophyticus (Fig. 1I) and B. subtilis (Fig. 1H). At the same time,
the non-glycated pea proteins (N) were stronger stimulators than
the glycated ones (G) in the case of proliferation of E. coli 360
(Fig. 1J) and E.faecium (Fig. 1E). Interestingly, the non-glycated pea
J Sci Food Agric 2010; 90: 1837–1845 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1839

1840
proteins caused a constant, statistically significant decrease of the
proliferation rate of E. coli 22 (Fig. 1A) in comparison to the control
culture (C) and the culture supplemented with the glycated pea
proteins. Additionally, the non-glycated pea proteins (N) caused a
decrease of the proliferation rate of L. acidophilus (Fig. 1C) at the
beginning of the cultivation period when compared to the control
culture (C).
The impact of the non-glycated (N) and the glycated pea
proteins (G) on the dying out of bacteria in nutrient-poor cultures
(A)
TBN [n x 10 4 ]
(B)
TBN [n x 10 4 ]
(C)
TBN [n x 10 4 ]
(D)
TBN [n x 10 4 ]
(E)
TBN [n x 10 4 ]
100
80
60
40
20
0
E. coli 22
100 E. aerogenes
80
60
40
20
0
100 L. acidophilus
80
60
40
20
0
200
160
120
80
40
0
400
300
200
100
0
E. faecalis
www.soci.org D ´Swia¸tecka, H Kostyra, A ´Swia¸tecki
C N G
E. coli 22
is demonstrated in Fig. 2. The survival of Gram-positive bacteria
was increased by supplementing the media with non-glycated
peaproteinsaswellasglycatedones(Fig.2C–E,H,IandK).
The glycated pea proteins (G) had a stronger stimulatory effect,
increasing the survival time of the bacteria in comparison to
the non-glycated pea proteins (N) in the case of the cultures of
L. acidophilus (Fig. 2C) and S. saprophyticus (Fig. 2I). The survival
rate of the rest of the Gram-positive bacterial strains (Fig. 2D, E,
H and K) was strengthened by supplementing the cultures with
E. coli 22
E. aerogenes E. aerogenes
L. acidophilus
L. acidophilus
E. faecalis E. faecalis
500 E. faecium E. faecium E. faecium
1 4 7 10 1 4 7 10 1 4 7 10
Time [days]
Time [days]
Time [days]
Figure 2a. The impact of the non-glycated and the glycated pea proteins on the bacterial survival rate. Vertically: (A) E. coli 22, (B) E. aerogenes,
(C) L. acidophilus,(D)E. faecalis,(E)E. faecium,(F)P. aeruginosa,(G)P. mirabilis,(H)B. subtilis,(I)S. saprophyticus,(J)E. coli 360, and (K) S. aureus. Horizontally:
C, cultures without any protein supplementation; N, cultures supplemented with non-glycated pea proteins; and G, cultures supplemented with glycated
pea proteins. ( ), Total bacterial number (TBN); ( .... ), percentage of dead bacterial cells.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1837–1845
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
% of dead bacterial cells
% of dead bacterial cells
% of dead bacterial cells
% of dead bacterial cells
% of dead bacterial cells

Impact of pea proteins on types of bacteria www.soci.org
Figure 2b. (Continued).
(F)
TBN [n x 10 4 ]
(G)
TBN [n x 10 4 ]
(H)
TBN [n x 10 4 ]
(I)
TBN [n x 10 4 ]
(J)
TBN [n x 10 4 ]
(K)
TBN [n x 10 4 ]
200
160
120
80
40
0
100
80
60
40
20
0
200
150
100
50
0
200
160
120
80
40
0
400
320
240
160
80
0
400
320
240
160
80
0
P. aeruginosa P. aeruginosa P. aeruginosa
P. mirabilis
P. mirabilis P. mirabilis
B. subtilis B. subtilis B. subtilis
S. saprophyticus S. saprophyticus S. saprophyticus
E. coli 360
C N G
E. coli 360
E. coli 360
S. aureus S. aureus S. aureus
1 4 7 10 1 4 7 10 1 4 7 10
Time [days]
Time [days]
Time [days]
J Sci Food Agric 2010; 90: 1837–1845 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
% of dead bacterial cells
% of dead bacterial cells
% of dead bacterial cells
% of dead bacterial cells
% of dead cells
% komórek matwych
1841

1842
non-glycated pea proteins (N) rather than with glycated ones (G).
However, in spite of the initial high percentage of dead cells in the
culture of E. faecalis (Fig. 2D) supplemented with the non-glycated
pea proteins, the presence of these proteins eventually increased
the survival time in the culture. The non-glycated (N) and the
glycated pea proteins (G) decreased the percentage of the dead
bacterial cells in the cultures of Gram-negative bacteria from the
genera Enterobacter (Fig. 2B) and Pseudomonas (Fig. 2F). There was
no statistically significant impact of the non-glycated pea proteins
on the survival of bacteria from the genera Escherichia (Fig. 2A and
J) and Proteus (Fig. 2G), whereas the non-glycated pea proteins
CTC+ [n x 10 2 ]
CTC+ [n x 10 2 ]
CTC+ [n x 10 2 ]
240
200
160
120
80
40
0
240
200
160
120
80
40
0
240
200
160
120
80
40
0
E. coli 22
www.soci.org D ´Swia¸tecka, H Kostyra, A ´Swia¸tecki
(A) (B) (C)
(D)
(G) significantly decreased the percentage of the dead cells in the
culture of P. mirabilis (Fig. 2G). Although the glycated pea proteins
increased the percentage of dead bacterial cells in the bacterial
cultures from the genus Escherichia (Fig. 2A and J) at the final stage
of the cultivation, it did not differ significantly from the control
culture (C) and culture supplemented with the non-glycated pea
proteins (N).
Figure 3 presents the metabolic activity of bacteria immobilised
to solid surfaces measured by the content of CTC formazan (CTF).
Metabolic activity of the immobilised bacteria began after an
adaptive period, which lasted 2 h in the case of E. coli 22 (Fig. 3A),
E. aerogenes L. acidophilus
(E) (F)
E. faecalis E. faecium P. aeruginosa
(G)
(H) (I)
P. mirabilis B. subtilis S. saprophyticus
CTC+ [n x 10 2 ]
240
200
160
120
80
40
0
(J)
E. coli 360
(K)
S. aureus
0 2 4 6 8 0 2 4 6 8
Time [hours]
Time [days]
Figure 3. The impact of non-glycated and glycated pea proteins on the metabolic activity of bacteria forming a biofilm. (A) E. coli 22, (B) E. aerogenes,
(C) L. acidophilus, (D)E. faecalis, (E)E. faecium, (F)P. aeruginosa, (G)P. mirabilis, (H)B. subtilis, (I)S. saprophyticus, (J)E. coli 360 and (K) S. aureus. ( ),
Control culture; (- - - - ), the culture supplemented with non-glycated pea proteins; ( .... ), the culture supplemented with glycated pea proteins.
TBN, total bacterial number.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1837–1845

Impact of pea proteins on types of bacteria www.soci.org
E. aerogenes (Fig. 3B) and S. aureus (Fig.3K)and4hinthecase
of the other bacterial strains, except for L. acidophilus (Fig. 3C),
which was metabolically active from the very beginning of the
cultivation period. Both the non-glycated (N) and the glycated
pea proteins (G) functioned as good energetic substrates as they
increased the metabolic activity of many examined bacterial
strains: E. coli 22 (Fig. 3A), E. aerogenes (Fig. 3B), L. acidophilus
(Fig. 3C), P. aeruginosa (Fig. 3F), P. mirabilis (Fig. 3G), B. subtilis
(Fig. 3H), E.coli 360 (Fig. 3J) and S.aureus (Fig. 3H). The glycated pea
proteins were better energetic substrate than the non-glycated
pea proteins, causing an increase of the metabolic activity of the
higher number of immobilised bacteria; apart from bacteria given
above and, additionally, E. faecalis (Fig. 3D) and S. saprophyticus
(Fig. 3I). Moreover, the glycated pea proteins (G) were preferably
used as the energy source rather than the non-glycated ones
(N) (Fig. 3A–D, H and I), displaying statistical differences when
compared to the control culture (C) and the culture supplemented
with the non-glycated pea proteins (N). The metabolic activity of
the Gram-positive cocci E. faecium (Fig. 3E) and S. saprophyticus
(Fig. 3I) was decreased by the supplementation of the cultures
with the non-glycated pea proteins (N). In addition, the nonglycated
pea proteins (N) initially decreased the metabolic activity
of the immobilised E. coli 22 (Fig. 3A) but eventually did not
demonstrate any statistical differences when compared to the
control culture (C) at the final stage of the cultivation period.
Moreover, there were no statistical differences in the metabolic
activity of E. faecalis culture supplemented with the non-glycated
peaproteins(N) whencomparedtothecontrolculture(C) (Fig. 3D).
DISCUSSION
It has already been demonstrated that substances delivered
with food have an influence on intestinal homeostasis, and in
consequence also on the general health status of the whole human
organism. 20–22 Therefore, glycated food proteins, which display
prolonged biological activities, may potentially be important
modulators of the structure and function of intestinal flora as are
other food substances. Limited information on that issue entails
the creation of simplified model systems in order to scrutinise
the scope of interactions that glycated proteins may have with
bacterial representatives of the intestinal ecosystem.
The bacterial intestinal ecosystem is highly complex and uses
various available niches. The intestinal environment imposes
different states of bacterial life, including microorganisms freely
swimming in the intestinal contents as well as microbes
immobilised to the solid surfaces, creating a biofilm. This study
used various experimental models with both free-swimming
bacteria, as well as bacteria immobilised to the solid surfaces, to
establish the impact of non-glycated and glycated pea proteins on
the physiological activity of the referential bacteria. The influence
of the examined substrates on the proliferation rate of the analysed
bacteria was assessed with the DAPI fluorescent marker (Fig. 1).
The bifunctional, stimulatory and inhibitory role of non-glycated
pea proteins was demonstrated in this study. The non-glycated
pea proteins stimulated the proliferation activity of the most of
the bacteria examined (Fig. 1B, D, F–H, J, K), which implies their
use as energy sources as well as potential nutrients. The inhibitory
effect of the non-glycated pea proteins was noticed for E. coli
22 (Fig. 1A), E. faecium (Fig. 1E), S. saprophyticus at the final stage
of the cultivation (Fig. 1I) and L. acidophilus at the initial phase
of the cultivation (Fig. 1C). Since the DAPI fluorescent marker
stains both dead and living bacterial cells, the observed decrease
in the bacterial number must have been caused by lysis of the
microorganisms. This suggests the occurrence of strong inhibitory
sequences in the non-glycated pea proteins, influencing E. coli 22
and E. faecium. The peptides displaying an antibacterial mode of
action, achieved by enzymatic degradation, were a probable cause
of the decrease of the number of S. saprophyticus. Interestingly,
bacteria from the genus Lactobacillus demonstrated adaptive
abilities, probably synthesising enzymes capable of utilising
peptides and/or protein inhibitory structures, thus increasing their
proliferation activities. Glycation of pea proteins with glucose
suppressed their inhibitory properties. The glucose condensed
onto the pea proteins and not only changed their structure,
thus suppressing their inhibitory action, but also constituted an
additional source of energy. In addition, glycation of the pea
increased their bacterial stimulatory effect in comparison to the
non-glycated proteins (Fig. 1B–D, F–H, and I). The results obtained
suggest an increased attractiveness of glycated pea proteins as a
source of nutrients and energy. Similar observations were made in
the case of potato proteins, indicating their preferential utilisation
as nutrients and an energy source by E. faecalis. 23 However,
glycation of pea proteins may also decrease their stimulatory
effect as shown in the case of E. coli 360 (Fig. 1J) and S. aureus
(Fig. 1K). Such results imply that the susceptibility to enzymatic
degradation is lowered due to the glycation process. Although
the general tendency in the bacterial response to glycated pea
proteins indicates their stimulatory impact, the observed effects
were likely to be strongly dependent on the enzymatic apparatus
of particular microorganisms and their adaptive abilities. This is
confirmedbythedifferentreactionstotheglycatedpeaproteinsby
bacteria belonging to the same genera: Escherichia, Enterococcus
and Staphylococcus. The published scientific data mainly indicate
the antibacterial effect of glycated proteins on bacteria. 23–25
The inhibitory impact of amino acids glycated with D-glucose,
tryptophan and lysine, on the proliferation rate of Aeropyrum
pernix was demonstrated by measuring the optical density of the
culture. 26
The condition of the bacterial culture is determined not only by
its proliferation rate but also by the rate of its dying out. Death
of a bacterial cell is indicated by a disturbance of the envelope
structures. 27 Dead bacteria with damaged membranes may be
detected with the complex of the Live/Dead fluorescent markers
until the time of its lysis. 27 So called ‘ghost cells’, which retain
their cellular envelopes but have lost their nucleoids, may also be
present in bacterial cultures. 28 Both non-glycated and glycated
pea proteins decreased dying out of Gram-positive bacteria in
comparison to the control cultures (Fig. 2C–E, H, I and K). Glycated
pea proteins more strongly stimulated the proliferation rate of
the Gram-positive bacteria and their survival, which suggests a
protective function of the modified proteins. These glycoproteins
might have functioned as a potential storage material, thus making
it possible for the culture to be maintained at the phase of
the constant proliferation with a low percentage of dead cells
(Fig. 2C, H and I) or might have covered the surfaces of the
bacterial cells, thus forming thus a specific layer protecting them
from a deleterious environmental impact. The high percentage
of dead bacterial cells at the initial days of the cultivation of
E. faecium with the non-glycated pea proteins (Fig. 2E) confirmed
the occurrence of the inhibitory sequences encrypted in the pea
proteins and/or the release of peptides displaying the antibacterial
impact on the bacterial cellular envelopes. The later increase of
the total number of these bacteria correlated with the decrease
of the percentage of the dead bacterial cells and implies the
J Sci Food Agric 2010; 90: 1837–1845 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1843

1844
synthesis of enzymes utilising the deleterious substances. The
survival rate of the Gram-negative bacteria was not significantly
modified by the non-glycated pea proteins in the final stage of
the cultivation of bacteria from the genera Escherichia (Fig. 2A
and J) and Proteus (Fig. 2G). However, a high percentage of dead
bacterial cells observed during the initial days of the culture of E.
coli 22 (Fig. 2A), when compared to the control culture, confirms
the inhibitory action of the non-glycated pea proteins. The high
proliferation activity and the lowered percentage of dead bacterial
cells, in comparison to the control culture, as well as to the one
supplemented with the non-glycated pea proteins, in the cultures
of E. aerogenes (Fig. 2B), P. mirabilis (Fig. 2G) and P. aeruginosa
(Fig. 2F) indicate the use of the glycated pea proteins as attractive
nutrients during the starvation period. Although an increase of
the percentage of the dead cells in the bacterial culture from the
genus Escherichia was observed (Fig. 2A and J), it was statistically
insignificant when compared to the control culture as well as
to the culture supplemented with non-glycated pea proteins.
This effect observed in the case of E. coli 360 was a probable
result of the dying out of the culture due to a utilisation of
available nutrients and accumulation of deleterious metabolites.
The intensity of the proliferation and the increasing percentage of
the dead cells in the case of E. coli 22 may suggest a short time
of generation, which may result from the destruction of bacterial
cellular envelopes. There is a lack of publications concerning the
effect of glycated proteins on the survival of bacteria, particularly
in the context of the percentage of dead bacterial cells in bacterial
cultures. Adamberg et al. 29 demonstrated the inhibitory effect of
glycated whey proteins and their hydrolysates on the survival
of E. coli, K. pneumoniae and S. aureus. Those authors suggested
a formation of antibacterial peptide sequences displaying the
inhibitory activities due to the glycation process.
Ninety-nine per cent of bacteria living in natural environments
are believed to form biofilms, 30 which equips them with the
properties and metabolic activities that differ from those of freeswimming
bacteria. 31,32 For that reason, simplified experimental
models using bacteria immobilised to the filters were used in this
study in order to determine the effect of the examined substrates
on their activity. Microscopic analysis indicated formations of
agglomerations constituting the proliferating bacteria, which, as
time progressed, changed into multilayered conglomerates, which
implies the formation of a biofilm. These observations remain
consistent with the findings stating that once immobilised bacteria
in the environment rich in the nutrients start to proliferate, they
dynamically create the so-called ‘conditioning layer’ as soon as
15 min after detecting the nutrients. 33 Moreover, rapid formation
of the biofilm was also reported by Dunne, 34 who noticed that
genes responsible for the formation of the polysaccharide matrix
in P. aeruginosa were active 15 min after bacterial immobilisation
to a solid surface.
Bacterial monobiofilms were used to determine the impact of
the non-glycated and glycated pea proteins on the metabolic
activity of particular bacteria (Fig. 3). The glycated pea proteins
stimulated dehydrogenase activity of all the bacterial cultures
analysed, having a stronger stimulatory effect in comparison
to the non-glycated pea proteins (Fig. 3A–D, H and I). The
results obtained are consistent with the tendency observed
in this study with free-swimming bacteria, as well as confirm
the thesis of the attractiveness of glycated pea proteins in
terms of metabolic activity. Since the contents of the CTC
formazan formed is correlated with the number of the bacteria
with active dehydrogenases (CTC+), the increase of metabolic
www.soci.org D ´Swia¸tecka, H Kostyra, A ´Swia¸tecki
activity approximately demonstrates the increase of the number
of bacteria forming biofilms, also. However, the decrease of
metabolic activity noticed in the final stage of the experiment
is not tantamount to a decrease of the number of bacteria forming
the biofilm, but indicates a hindrance to the metabolic processes.
Such phenomena noticed in the case of the biofilm created by
P. aeruginosa (Fig. 3F) may suggest a partial use of glycoproteins
as energy sources. The non-glycated pea proteins were better
sources of energy for E. coli 360 (Fig. 3J) and S. aureus (Fig. 3K),
which once again suggests that the condensation of glucose onto
pea proteins impedes their utilisation by the bacteria mentioned.
The inhibitory effect of non-glycated pea proteins on E. coli 22
(Fig. 3A), E. faecium (Fig. 3E) and S. saprophyticus (Fig. 3I) was also
confirmed. The mentioned inhibitory action on the Gram-positive
cocci might have been initially suppressed by the protective
function of the biofilm. The inhibitory molecules might have been
bounded by elements of the polysaccharide matrix formed or by
the outer layer of bacterial cells forming the biofilm. Therefore,
utilisation of the harmless protein fractions as well as the presence
of substances displaying inhibitory activities might have been the
cause of the later decrease in metabolic activities of the biofilms
mentioned. The low metabolic activity of the immobilised E. coli
22 (Fig. 3A) at the initial time of cultivation might also result from
the action of inhibitory protein fractions. The fractions might have
subsequently been immobilised to the matrix of the biofilm or
deactivated and utilised by enzymes synthesised as a result of
the adaptive bacterial response. The lack of any published data
on that issue makes the explanation of the observed phenomena
very difficult.
CONCLUSIONS
Theresultsobtainedprovedthattheglycationprocesssignificantly
alters not only the structure of pea proteins examined but also
their biological characteristics, making them active modulators of
bacterial physiological activity. Glycated pea proteins influence the
physiological activity of bacteria by stimulating the proliferation
rate and metabolic activity of free-swimming and immobilised
bacteria. They also decrease the percentage of dead bacterial
cells in cultures of Gram-positive cocci. The results obtained are
of great importance in terms of understanding the interaction of
glycated food products and the intestinal ecosystem, which, in
turn, may alter the general nutritional approach. Investigations
into the beneficial effects of glycated pea proteins on the human
heterogeneous bacterial population are currently under way in
order to broadly elucidate their impact on the human intestinal
ecosystem.
REFERENCES
1 Ofuya ZM and Akhidue V, The role of pulses in human nutrition: a
review. J Appl Sci Environ Mgt 9:99–104 (2005).
2 Singh S, Kundu SS, Negi AS and Singh PN, Cowpea (Vigna unguiculata)
legume grains as protein source in the ration of growing sheep.
Small Ruminant Res 64:247–254 (2006).
3 Adsule RN and Kadam SS, Proteins, in Handbook of World
Food Legumes: Nutritional Chemistry, Processing Technology and
Utilization, vol. II, ed. by Salunkhe DK and Kadam SS. CRC Press,
Boca Raton, FL, pp. 75–97 (989).
4 Alonso R, Grant G, Dewey P and Marzo F, Nutritional assessment
in vitro and in vivo of raw and extruded peas (Pisum sativum L.).
J Agric Food Chem 48:2286–2290 (2000).
5 Hedley CL,Carbohydrates,inGrainLegumeSeeds.ImprovingNutritional
Quality and Agronomic Characteristics. CABI Publishing, Wallingford
(2001).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1837–1845

Impact of pea proteins on types of bacteria www.soci.org
6 Schafer G, Schenk U, Ritzel U, Ramadori G and Leonhardt U,
Comparison of the effects of dried peas with those of potatoes
in mixed meals on postprandial glucose and insulin concentrations
in patients with type 2 diabetes. Am J Clin Nutr 78:99–103 (2003).
7 Vidal-Valverde C, Frias J, Hernández A, Martín-Alvarez PJ, Sierra I
and Rodríguez C, Assessment of nutritional compounds and
antinutritional factors in pea (Pisum sativum) seeds. J Sci Food
Agric 83:298–306 (2003).
8 Ames JM, Application of the Maillard reaction in the food industry.
Food Chem 62:431–439 (1998).
9 Chuyen NV, Maillard reaction and food processing. Application
aspects. Adv Exp Med Biol 434:213–235 (1998).
10 Michalska A and Zieliński H, Produkty reakcji Maillarda w ˙zywno´sci.
˙Zywno´sć Nauka Technologia Jako´sć 51:5–16 (2007).
11 Nakamura S, Kobayashi K and Kato A, Role of positive charge of
lysozyme in the excellent emulsifying properties of Maillardtype
lysozyme–polysaccharide conjugate. J Agric Food Chem
42:2688–2691 (1994).
12 Morgan F,MolléD,HenryG,Vénien A,Léonil J,Peltre G,et al,Glycation
of bovine β-lactoglobulin: effect on the protein structure. Int J Food
Sci Technol 34:429–435 (1999).
13 Sato R, Sawabe T, Kishimura H, Hayashi K and Saeki H, Preparation
of neoglycoprotein from carp myofibrillar protein and alginate
oligosaccharide: improved solubility in low ionic strength medium.
J Agric Food Chem 48:17–21 (2000).
14 Weiss W, Vogelmeier C and Görg A, Electrophoretic characterization
of wheat grain allergens from different cultivars in bakers’ asthma.
Electrophoresis 14:805–816 (1993).
15 Porter KG and Feig YS, The use of DAPI for identifying and counting
aquatic microflora. Limnol Oceanogr 25:943–948 (1980).
16 Boulos L, Prevost M, Barbeau B, Coallier J and Desjardins R,
LIVE/DEAD® BacLight : application of a new rapid staining method
for direct enumeration of viable and total bacteria in drinking water.
JMicrobMeth37:77–86 (1999).
17 Berman T, Kaplan B, Chava S, Viner Y, Sherr BF and Sherr EB,
Metabolically active bacteria of Lake Kinneret. Aquat Microb Ecol
23:213–224 (2001).
18 Rodriguez GG, Phipps D, Ishiguro K and Ridgway HF, Use of a
fluorescent redox probe for direct visualization of actively respiring
bacteria. Appl Environ Microb 58:1801–1808 (1992).
19 ´Swia¸teckaD and ´Swia¸tecka A, The use of fluorescent markers to
estimate the metabolic activity of heterotrophic bacteria. Limnol
Papers 113:165–172 (2005).
20 Macfarlane GT and Cummings JH, Probiotics and prebiotics: can
regulating the activities of intestinal bacteria benefit health? BMJ
318:999–1003 (1999).
21 Bird AR, Brown IL and Topping DL, Starches, resistant starches, the gut
microflora and human health. Curr Issues Intest Microbiol 1:25–37
(2000).
22 Macfarlane GT, Steed H and Macfarlane S, Bacterial metabolism
and health-related effects of galacto-oligosaccharides and other
prebiotics. J Appl Microbiol 104:305–344 (2008).
23 ´Swia¸tecka D, Marciniak-Darmochwał K, Kostyra H and ´Swia¸tecka A,
The influence of non-enzymatically glycosylated potato proteins
on the survival and metabolic activity of chosen bacteria. Reprod
Nutr Dev 46:59–60 (2006).
24 Einarsson H, Snygg BG and Eriksson C, Inhibition of bacterial growth
by Maillard reaction products. J Agric Food Chem 31:1043–1047
(1983).
25 Einarsson H, The effect of time, temperature, pH and reactants on
the formation of antibacterial compound in the Maillard reaction.
Lebens Wiss Technol 20:51–55 (1987).
26 Kim KW and Lee SB, Inhibitory effect of Maillard reaction products
on growth of the aerobic marine hyperthermophilic archeon
Aeropyrum pernix. Appl Environ Microbiol 69:4325–4328 (2003).
27 Joux FandLebaron P,Useoffluorescentprobestoassessphysiological
functions of bacteria at single-cell level. Microb Infect 2:1523–1535
(2000).
28 Zweifel UL and Hagström A, Total counts of marine bacteria include
a large fraction of non-nucleoid-containing bacteria (ghosts). Appl
Environ Microbiol 61:2180–2185 (1995).
29 Adamberg K, Kostyra H, Kostyra E, ´Swia¸teckaA and ´Swia¸teckaD,
Influence of non-enzymatic glycosylation of whey proteins and
concentration of glucose on their hydrolysis by trypsin and growth
and survival of selected bacteria. Milchwiss 60:422–426 (2005).
30 Costerton JW, Cheng K-J, Geesey GG, Ladd TI, Nickel NC, Dosgupton
M, et al, Bacterial biofilms in nature and disease. Annu Rev
Microbiol 41:435–441 (1987).
31 Whiteley M, Bangera MG, Bumgarner RE, Parsek MR, Teitzel GM,
Lory S, et al, Gene expression in Pseudomonas aeruginosa biofilms.
Nature 413:860–864 (2001).
32 Stoodley P, Sauer K, Davies DG and Costerton JW, Review. Biofilms
as complex differentiated communities. Annu Rev Microbiol
56:187–209 (2002).
33 Stewart PS and Costerton JW, Antibiotic resistance of bacteria in
biofilms. Lancet 358:135–138 (2001).
34 Dunne Jr WM, Review. Bacterial adhesion: Seen any good biofilms
lately? Clin Microbiol Rev 15:155–166 (2002).
J Sci Food Agric 2010; 90: 1837–1845 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1845

1846
Research Article
Received: 18 October 2009 Revised: 13 March 2010 Accepted: 26 April 2010 Published online in Wiley Interscience: 14 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4023
Suppressive effects of extracts from the aerial
part of Coriandrum sativum L. on LPS-induced
inflammatory responses in murine RAW 264.7
macrophages
Trang-Tiau Wu, a,b† Chia-Wen Tsai, c† Hsien-Tsung Yao, c Chong-Kuei Lii, c
Haw-Wen Chen, c Yu-Ling Wu, d Pei-Yin Chen d and Kai-Li Liu d,e∗
Abstract
BACKGROUND: Coriandrum sativum is used not only as a spice to aid flavour and taste values in food, but also as a folk
medicine in many countries. Since little is known about the anti-inflammatory ability of the aerial parts (stem and leaf) of
C. sativum, the present study investigated the effect of aerial parts of C. sativum on lipopolysaccharide (LPS)-stimulated RAW
264.7 macrophages. We further explored the molecular mechanism underlying these pharmacological properties of C. sativum.
RESULTS: Ethanolic extracts from both stem and leaf of C. sativum (CSEE) significantly decreased LPS-induced nitric oxide and
prostaglandin E2 production as well as inducible nitric oxide synthase, cyclooxygenase-2, and pro-interleukin-1β expression.
Moreover, LPS-induced IκB-α phosphorylation and nuclear p65 protein expression as well as nuclear factor-κB(NF-κB) nuclear
protein–DNA binding affinity and reporter gene activity were dramatically inhibited by aerial parts of CSEE. Exogenous
addition of CSEE stem and leaf significantly reduced LPS-induced expression of phosphorylated mitogen-activated protein
kinases (MAPKs).
CONCLUSION: Our data demonstrated that aerial parts of CSEE have a strong anti-inflammatory property which inhibits
pro-inflammatory mediator expression by suppressing NF-κB activation and MAPK signal transduction pathway in LPS-induced
macrophages.
c○ 2010 Society of Chemical Industry
Keywords: Coriandrum sativum; inflammation; lipopolysaccharides; mouse RAW 264.7 macrophages; NF-κB
INTRODUCTION
Macrophages, the major immune cells in the innate immune
system, are essential for host defence and inflammation against
intracellular parasitic bacteria, pathogenic protozoa and fungi by
producing a variety of cytokines and inflammatory mediators, such
as interleukin-1β (IL-1β), nitric oxide (NO) and prostaglandin E2
(PGE2). 1 IL-1β from cleavage of an inactive pro-IL-1β protein is
yielded by stimulated leukocytes and is a fundamental contributor
to local and systemic inflammatory responses. 2 Large amounts
of NO and PGE2, produced by inducible nitric oxide synthase
(iNOS) and cyclooxygenase-2 (COX-2), are described not only
at inflammatory sites but also in process of carcinogenesis. 3–5
Although inflammation plays an important role in the host
defence system, inappropriate activation of macrophages results
in an overproduction of the inflammatory mediators, which is
involved in the pathologies of many chronic diseases of modern
society, such as rheumatoid arthritis, atherosclerosis, diabetes and
cancer. 5–7 Current clinical approaches to treatments of chronic
inflammation involve the inhibition of pro-inflammatory mediator
production and the suppression of mechanisms responsible for
the initiation of inflammatory responses.
Activation of the nuclear factor-κB (NF-κB) pathway plays a
key role in the transcriptional regulation of pro-inflammatory
response gene expression, including iNOS, COX-2, cytokines,
chemokines, growth factors, cell adhesion molecules, and several
acute phase proteins. 8 The transcription factor NF-κB existsin
∗ Correspondence to: Kai-Li Liu, Department of Nutrition, Chung Shan Medical
University, NO. 110, Sec. 1, Chien-Kuo N. Rd., Taichung 40203, Taiwan.
E-mail: kaililiu@csmu.edu.tw
† Trang-Tiau Wu and Chia-Wen Tsai contributed equally to this work.
a Departments of Pediatric Surgery, Chung Shan Medical University Hospital,
Taichung, Taiwan
b School of Medicine, Chung Shan Medical University, Taichung, Taiwan
c Department of Nutrition, China Medical University, Taichung, Taiwan
d Department of Nutrition, Chung Shan Medical University, NO. 110, Sec. 1,
Chien-Kuo N. Rd., Taichung 40203, Taiwan
e Department of Dietitian, Chung Shan Medical University Hospital, Taichung,
Taiwan
J Sci Food Agric 2010; 90: 1846–1854 www.soci.org c○ 2010 Society of Chemical Industry

Effect of C. sativum on inflammatory responses in macrophages www.soci.org
the cytoplasm of most eukaryotes by forming homodimers or
heterodimers with proteins of the NF-κB family, including p65
(RelA), p50/p105 (NF-κB1), p52/p100 (NF-κB2), RelB, and c-Rel.
In unstimulated cells, NF-κB is sequestered in the cytoplasm by
noncovalentlybinding to an inhibitor protein termed IκB(α, β or ε).
Activation of NF-κB occurs in response to inflammatory cytokines,
oxidative stress, ultraviolet irradiation, or bacterial endotoxins,
which induce phosphorylation, ubiquitination and degradation
of IκBα. ThentheactivatedNF-κB migrates into the nucleus and
induces transcriptional expression of its target genes. Constitutive
activation of the NF-κB pathway leads to persistent increases in the
expression of pro-inflammatory mediators and exerts pathogenic
effects on chronic inflammatory related diseases. Regulation and
control of NF-κB activation may be a key molecular target for
anti-inflammatory therapy. 8,9
Spices are used not only to aid flavour, colour and nutritional
values in food, but also to treat various physical problems in
traditional medicines. 10 Coriandrum sativum, commonly known
as coriander or Chinese parsley, belongs to the family Apiaceae,
which is widely cultivated all over the world. The seeds and aerial
parts (stem and leaf) of C. sativum are commonly used as spices
in Middle Eastern, Mediterranean, Indian, Latin American, African,
Southeast Asian and Taiwanese cuisines. Data from numerous
researchers have shown the therapeutic values of the seeds and
seed oil of C. sativum due to its hypoglycaemic, hypolipidaemic,
hepatoprotective, antimutagenic, antihypertensive, antioxidant,
anxiolytic, antimicrobial and post-coital antifertility activity. 11–19
The aerial parts of C. sativum have antioxidant and free
radical scavenging activities, suppressive activity on lead and
mercury deposition and bactericidal and anti-adhesive effects on
Helicobacter pylori. 20–23 Recent studies reported that C. sativum
seed oil reduced UV-induced erythema test of human skin and
leaves of C. sativum water extract decreased LPS-induced NO
production and had scavenging effects on NO. 24,25 Based on
previous studies of C. sativum presented herein, it is worth
conducting a detailed investigation on the anti-inflammatory
property of aerial parts of C. sativum. The objective of the
present study is to assess the regulatory efficacy of aerial parts
of C. sativum on LPS-induced inflammatory responses in RAW
264.7 macrophages as well as to explore the possible molecular
mechanism behind these actions.
MATERIALS AND METHODS
Materials
The mouse macrophage-like cell line RAW 264.7 was purchased
from the Food Industry Research and Development Institute
(Hsinchu, Taiwan), and fetal bovine serum was from Thermo
Fisher Scientific Inc. (Waltham, MA, USA). RPMI 1640 medium
and media supplements for cell culture were obtained from
Invitrogen Corporation (Carlsbad, CA, USA). LPS, ferulic acid, gallic
acid, 4-hydroxycoumarin, hesperidin, luteolin, dihydroquercetin,
quercetin and Folin–Ciocalteu phenol reagent were from Sigma
Chemical Company (St Louis, MO, USA). The specific antibodies for
iNOS, p65, c-Jun NH2-terminal kinase (JNK), and phosphorylated
JNK were purchased from Santa Cruz Biotechnology (Santa Cruz,
CA,USA).Theantibodiesagainstp38,extracellularsignal-regulated
kinase 1/2 (ERK1/2), phosphorylated p38, ERK1/2 and IκB-α, as
well as p65 were from Cell Signaling Technology Inc. (Beverly,
MA, USA). The specific antibodies for COX-2, pro-IL-1β, andβactin
were obtained form Cayman Chemical Co. (Ann Arbor,
MI, USA), CytoLab Ltd. (Rehovot, Israel), and Sigma Chemical
Company, respectively. Nucleotides and RNase inhibitor were
obtained from Promega Co. (Madison, WI, USA) and M-MMLV
reverse transcriptase was from Gibco BRL (Gaithersburg, MD).
Real-time polymerase chain reaction (PCR) primers and TaqMan®
Universal PCR Master Mix were from Applied Biosystems (Foster
City, CA, USA). The biotin-labelled and unlabelled double-stranded
NF-κB consensus oligonucleotides and a mutant double-stranded
NF-κB oligonucleotide for electrophoretic mobility shift assay
(EMSA) were synthesised by MDBio Inc. (Taipei, Taiwan). The pNFkB-Luc
plasmid was from Stratagene Inc. (La Jolla, CA, USA) and the
Luciferase Assay System, β-Galactosidase Enzyme Assay System
with Reporter Lysis Buffer, and pSV-β-galactosidase control vector
were from Promega. All other chemicals were of the highest quality
available.
Preparation of extractions
C. sativum was purchased from local markets at Taichung, Taiwan,
and was identified by Dr Lee-Yan Sheen (Graduate Institute of
Food Science and Technology, National Taiwan University, Taipei,
Taiwan). A voucher specimen was kept in our laboratory, at
the Department of Nutrition, Chung Shan Medical University,
Taichung, Taiwan, for further reference.
C. sativum was washed and its leaves and stems were separately
cut into small pieces, placed in a −70 ◦ C freezer for 12 h and then
freeze-dried (FD4; Heto Lab Equipment, Birkerød, Denmark) for at
least 12 h. The dried C. sativum leaves and stems were ground to a
fine power using an electric food grinder and were then stored at
−20 ◦ Cforfurtheruse.
The powdered freeze-dried food material was extracted by
95% ethanol (25 mL g −1 powder) for 24 h and then centrifuged at
3200 × g for 10 min. The supernatants were concentrated at 37 ◦ C
under reduced pressure and then dissolved in ether. The ether in
the ethanolic extract was evaporated by nitrogen. The ethanolic
extracts were weighed to measure the extraction yield and then
dissolved in dimethyl sulfoxide (DMSO) for cell treatments.
Determination of total phenolics and flavonoids
The amount of total phenolics in the leaf or stem of C. sativum
ethanolic extract (CSEE) was measured according to a modification
of the Folin–Ciocalteu method. 26 A 300 µL aliquot of each extract
or standard solution was mixed with 300 µL of Folin–Ciocalteu
reagent. The mixture was allowed to stand for 5 min followed by
the addition of 600 µL of 20% Na2CO3. After a 10 min incubation
at room temperature, the absorbance of the reaction mixture was
measured at 730 nm and total phenolics were calibrated by using
a standard curve of gallic acid. The total phenolic content of stem
and leaf of CSEE was expressed as the gallic acid equivalent (GAE)
in mg g −1 dry material.
A 250 µL aliquot of each extract or standard solution was mixed
with 1.25 mL of distilled water and 75 µL of5%NaNO2 solution.
After 6 min, 150 µL of 10% AlCl3-H2O solution was added. After
5 min, 0.5 mL of 1 mol L −1 NaOH solution was added and the
total volume was brought up to 2.5 mL with ddH2O. Following
thorough mixing of the solution, absorbance against the blank was
determined at 510 nm and the total flavonoids were calibrated by
a standard curve of quercetin. The total flavonoid content of the
stem and leaf of CSEE was expressed as the quercetin equivalent
(QE) in mg g −1 dry material. 27
Cell culture
The RAW 264.7 macrophages of passages 10 to 15 were
maintained in RPMI-1640 medium supplemented with 2 mmol L −1
J Sci Food Agric 2010; 90: 1846–1854 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1847

1848
L-glutamine,100UmL −1 penicillin,100 µgmL −1 streptomycin,and
10% heat-inactivated fetal bovine serum at 37 ◦ Cinahumidified
atmosphere of 5% CO2. Cells were plated at a density of 8 × 10 5
per 30 mm culture dish and were incubated until 90% confluence
was reached.
Cells were treated with 25–150 µgmL −1 leaf or stem of CSEE in
the presence of 1 µgmL −1 LPS for 8–24 h as indicated or treated
with the leaf or stem of CSEE for 1 h or 14 h prior to addition of LPS
(1 µgmL −1 ). The final DMSO concentration in the medium was
0.1%.
Cell viability assay
The mitochondrial-dependent reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide(MTT)toformazan
was used to measure cell respiration as an indicator of cell
viability. 28 After incubation with leaf or stem of CSEE with or without
LPS for 24 h, cells were incubated in RPMI medium containing
0.5 mg mL −1 MTT for an additional 3 h. The medium was then removed
and isopropanol was added to dissolve the formazan. After
centrifugation at 5000 × g for 5 min, 100 µL of supernatant from
each sample was transferred to 96-well plates, and the absorbance
was read at 570 nm in a VersaMaxTM Tunable Microplate Reader
(Molecular Devices Corporation, Sunnyvale, CA, USA).
Nitrite and PGE2 determination
The nitrite in the medium was measured by use of the
Griess assay and used as an indicator of NO synthesis in
cells. 29 Briefly, equal volumes of the culture supernatants and
Griess solution [1 : 1 mixture of 1% sulfanilamide and 0.1% N-
(naphthyl)ethylenediamine dihydrochloride in 5% H3PO4] were
added to 96-well plates at room temperature for 10 min.
Absorbance was measured at 550 nm and nitrite concentration
was determined by using a standard curve of sodium nitrite
prepared in the culture medium.
Cells were treated with leaf or stem of CSEE in the presence
of LPS for 12 h. The diluted culture supernatants were used to
quantify PGE2 by use of an enzyme immunoassay kit (Cayman
Chemical Company, Ann Arbor, MI, USA) according to the protocol
provided by the manufacturer.
Western blot analysis
Cells were washed twice with cold phosphate-buffered saline (PBS)
and were harvested in 150 µL lysis buffer containing 10 mmol L −1
Tris-HCl, 5 mmol L −1 EDTA, 0.2 mmol L −1 phenylmethylsulfonyl
fluoride (PMSF), and 20 µgmL −1 aprotinin, pH 7.4. The protein
contentineachsamplewasquantifiedbyuseoftheCoomassiePlus
Protein Assay Reagent Kit (Pierce Chemical Co., Rockford, IL, USA).
Equal amounts of proteins were denatured and separated on SDSpolyacrylamide
gels and were then transferred to polyvinylidene
difluoride membranes (NewTM Life Science Product, Inc., Boston,
MA, USA). Nonspecific binding sites on the membranes were
blockedwith5%nonfatdrymilkinabuffercontaining10 mmol L −1
Tris-HCl and 100 mmol L −1 NaCl, pH 7.5, at 4 ◦ C overnight. The
blots were then incubated sequentially with primary antibody and
horseradish peroxidase-conjugated anti-goat or anti-rabbit IgG
(Bio-Rad, Hercules, CA, USA). Immunoreactive protein bands were
developed by enhanced chemiluminescence kits (Amersham Life
Sciences, Arlington Heights, IL, USA) and then were quantified
through densitometric analysis by Zero-Dscan (Scanalytics Inc.,
Fairfax, VA, USA).
www.soci.org T-T Wu et al.
Isolation of RNA and real-time quantitative reverse
transcriptase-PCR
Total RNA was isolated from cells by using Tri-Reagent TM (Molecular
Research Center Inc., Cincinnati, OH, USA) as described by the
manufacturer and RNA extracts were suspended in nuclease-free
water. Total RNA (0.1–0.25 µg) was reverse transcribed with M-
MMLV reverse transcriptase in a 20 µL final volume of the reaction
buffer consisting of 1 mmol L −1 of each deoxynucleotide triphosphate,
2.5 units RNase inhibitor and 2.5 m mol L −1 oligo(dT)16.
For the synthesis of complementary DNA, reaction mixtures were
incubated for 15 min at 45 ◦ C and stopped by denaturing the
reverse transcriptase at 99 ◦ C for 5 min. Complementary DNA was
amplified with TaqMan® Universal PCR Master Mix primers and
probes and the reactions were measured in ABI 7000 Real Time PCR
System (Applied Biosystems). The primers and probes were obtained
from Applied Biosystems: iNOS (Mm00440502 m1), COX-2
(Mm00478374 m1),IL-1β (Mm01336189 m1)andglyceraldehyde-
3-phosphate dehydrogenase (GAPDH, Mm00484668 m1). Glyceraldehyde
3-phosphate dehydrogenase (GAPDH) was used as an
internal standard gene and the threshold cycles (Ct) of a test
sample to a control sample (��Ct method) was used for relative
quantification of target gene expressions. 30
Preparation of nuclear protein and EMSA
At the time of harvest, cells were scraped with cold PBS
and centrifuged. The pellets were resuspended in the hypotonic
extraction buffer (10 mmol L −1 HEPES, 10 mmol L −1 KCl,
1 mmol L −1 MgCl2, 1 mmol L −1 EDTA, 0.5 mmol L −1 dithiothreitol,
0.2 mmol L −1 PMSF, 4 µgmL −1 leupeptin, 20 µgmL −1 aprotinin,
and 0.5% NP-40) for 15 min on ice and were then centrifuged at
6000×g for 15 min. The pelleted nuclei were resuspended in 50 µL
hypertonic extraction buffer (10 mmol L −1 HEPES, 400 mmol L −1
KCl, 1 mmol L −1 MgCl2, 1 mmol L −1 EDTA, 0.5 mmol L −1 dithiothreitol,
0.2 mmol L −1 PMSF, 4 µgmL −1 leupeptin, 20 µgmL −1
aprotinin, and 10% glycerol), were constantly shaken at 4 ◦ Cfor
30 min, and were then centrifuged at 10 000 × g for 15 min. The
resultant supernatants containing nuclear proteins were collected
and stored at −70 ◦ C until the EMSA was performed.
EMSA was performed according to our previous study. 31 The
LightShiftTM Chemiluminescent EMSA Kit from Pierce Chemical
Co. and synthetic biotin-labelled double-stranded NF-κBconsensus
oligonucleotide (5 ′ -AGTTGAGGGGACTTTCCCAGGC-3 ′ ) were
used to measure the effect of the CSEE leaf or stem on NFκB
nuclear protein-DNA binding activity. Nuclear proteins (2 µg),
poly(dI-dC), and biotin-labelled double-stranded NF-κB oligonucleotide
were mixed with the binding buffer to a final volume
of 20 µL and were incubated at room temperature for 30 min.
In addition, the excess amount (100-fold molar excess) of unlabelled
and a mutant double-stranded NF-κB oligonucleotide
(5 ′ -AGTTGAGGCGACTTTCCCAGGC-3 ′ ) were used for the competition
assay to confirm specificity of binding. The nuclear
protein–DNA complex was separated by electrophoresis on a
6% Tris/boric acid/EDTA–polyacrylamide gel and was then electrotransferred
to a nylon membrane (HybondTM-N+, Amersham
Pharmacia Biotech Inc, Piscataway, NJ, USA). The membrane was
treated with streptavidin–horseradish peroxidase, and the nuclear
protein–DNA bands were developed by using a SuperSignal West
Pico kit (Pierce Chemical Co.).
Reporter gene assay
Reporter enzyme activity was evaluated by a cell-based analysis
method for assaying NF-κB activity. The pNF-κB-Luc reporter
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1846–1854

Effect of C. sativum on inflammatory responses in macrophages www.soci.org
Figure 1. Effects of stem and leaf of CSEE on LPS-induced iNO, COX-2 and
Pro-IL-1β protein expressions in RAW 264.7 macrophages. (A) iNOS protein
expression was measured by RAW 264.7 macrophages treated with or
without LPS (10 ng mL −1 ) plus DMSO vehicle control and 25–150 µgmL −1
stem or leaf of CSEE for 24 h. (B) COX-2 and (C) Pro-IL-1β protein expression
were measured in cells preincubated with 25–150 µgmL −1 stem or leaf
of CSEE for 1 h and then treated with either DMSO vehicle control or
10 ng mL −1 LPSfor6h.Dataarethemean± SD of at least four separate
experiments and are expressed as the percentage of the culture treated
with LPS alone. Values not sharing the same letter are significantly different
(P < 0.05).
plasmid contains five tandem copies of the NF-κB consensus
sequences, and it permits luciferase expression in response to NFκB
activity. When RAW264.7 cells reached confluence, transient
transfection of the reporter plasmid and the pSV-β-galactosidase
control vector was performed using the Lipofectamine transfection
reagent for 6 h. Cells were then placed in fresh culture
media for 18 h before treating with DMSO vehicle control, or
LPS plus stem or leaf of CSEE for 8 h. Supernatants of the cell
lysates were applied to measure the luciferase and β-galactosidase
activities by the Luciferase Assay System and β-Galactosidase Enzyme
Assay System with Reporter Lysis Buffer from Promega Co.,
respectively.
Figure 2. Effects of stem and leaf of CSEE on LPS-induced of expressions
of cytoplasmic phosphorylated IκBα and nuclear p65. RAW 264.7
macrophages were preincubated with 150 µgmL −1 stem or leaf of CSEE for
14 h and then treated with either DMSO vehicle control or 10 ng mL −1 LPS
for 30 min. Western blot analysis was used to measure the protein content
of phosphorylated IκB-α in the cytosolic factions and to measure p65
protein content in the nuclearprotein fractions of RAW 264.7 macrophages.
Data are the mean ± SD of at least four separate experiments and are
expressed as the percentage of the culture treated with LPS alone. Values
not having the same letter are significantly different (P < 0.05).
Liquid chromatography/mass spectrometry analysis
The stem and leaf of CSEE were dissolved in DMSO, diluted with
an appropriate volume of 50% methanol (in deionised water) and
injected into the liquid chromatography–mass spectroscopy (LC-
MS) system. LC-MS was carried out with an Agilent 1100 Series
LC System equipped with a UV detector (Agilent Technologies,
Palo Alto, CA, USA). An Alltima C18 reverse-phase column
(5 µm, 250 × 4.6 mm) was used. The mobile phase consisted
of 10 mmol L −1 ammonium acetate containing 0.5% formic
acid (solvent A) and acetonitrile (solvent B). The flow rate was
0.6 mL min −1 . The total running time was 90 min. The gradient
system used was 5% B (0–10 min), 5% B to 40% B (10–40 min),
40–55% B (40–55 min), 55% B to 80% B (55–65 min), 80–100% B
(65–70 min), 100% B to 5% B (75–80 min), and 5% B (80–90 min).
The column temperature was 25 ◦ C. The effluent was monitored
using a UV detector set at 254 nm and a mass spectrometer
operating in the negative ion mode over the m/z range 100 to
700. Identification of constituents was carried out by LC-UV and
LC-MS analyses by comparing their retention times, UV and mass
spectra of the peaks with those of authentic standards based
on the literature. 32 For quantitative analysis of rutin, calibration
curves were constructed from working standard solutions of rutin
at final concentrations of 1–1000 ng mL −1 and applied the same
way for the stem and leaf of CSEE. Ions representing the [M-H] −
species at m/z 609 were selected and the peak area was measured.
Statistical analysis
Data are expressed as the means ± SD from at least three
independent experiments. Differences among treatments were
analysed by ANOVA and Tukey’s multiple-range test by using the
Statistical Analysis System (Cary, NC, USA). P values less than 0.05
were considered to be significant.
RESULTS
The extraction yields of leaf and stem of CSEE were 11.76%
and 11.97%, respectively. In addition, the phenolic contents,
J Sci Food Agric 2010; 90: 1846–1854 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1849

1850
Table 1. Effects of an ethanolic extract from the stem of Coriandrum
sativum (CSEE) on the MTT assay and LPS-induced NO and PGE2
production in RAW 264.7 macrophages
Treatment ∗ MTT † NO PGE2
Control 102.9 ± 2.0b ND 6.7 ± 0.2d LPS 100.0 ± 0.0b 100.0 ± 0.0a 100.0 ± 0.0a LPS + Stem of CSEE
25 µgmL−1 98.9 ± 4.6b 94.4 ± 4.6b 53.6 ± 8.9b LPS + Stem of CSEE
50 µgmL−1 100.3 ± 3.8b 60.0 ± 7.6c 43.2 ± 2.1b LPS + Stem of CSEE
100 µgmL−1 117.7 ± 7.9a 32.7 ± 7.5d 44.7 ± 9.7b LPS + Stem of CSEE
150 µgmL−1 110.6 ± 5.0a 19.7 ± 6.8e 25.2 ± 2.4c ∗ RAW 264.7 macrophages were treated with or without LPS
(10 ng mL −1 ) plus DMSO vehicle control and 25 to 150 µgmL −1
stem of CSEE for 24 h (MTT assay and NO production) or for 12 h
(PGE2 production).
† Data are the mean ± SD of at least four separate experiments and are
expressed as the percentage of the culture treated with LPS alone.
a,b,c,d Means with different letters within the same column are
significantly different (P < 0.05).
Table 2. EffectsofleafofCSEEonMTTassayandLPS-inducedNO
and PGE2 production in RAW 264.7 macrophages
Treatment ∗ MTT † NO PGE2
Control 102.9 ± 2.0b ND 6.7 ± 0.2d LPS 100.0 ± 0.0b 100.0 ± 0.0a 100.0 ± 0.0a LPS + Leaf of CSEE
25 µgmL−1 90.9 ± 0.7c 82.8 ± 6.1a 96.6 ± 6.2a LPS + Leaf of CSEE
50 µgmL−1 102.9 ± 6.6b 62.0 ± 7.2b 80.7 ± 2.6b LPS + Leaf of CSEE
100 µgmL−1 93.1 ± 1.4c 46.9 ± 9.0c 82.8 ± 3.8b LPS + Leaf of CSEE
150 µgmL−1 116.0 ± 3.2a 24.7 ± 7.8d 52.5 ± 7.7c ∗ RAW 264.7 macrophages were treated with or without LPS
(10 ng mL −1 ) plus DMSO vehicle control and 25 to 150 µgmL −1
leaf of CSEE for 24 h (MTT assay and NO production) or for 12 h
(PGE2 production).
† Data are the mean ± SD of at least four separate experiments and are
expressed as the percentage of the culture treated with LPS alone.
a,b,c,d Means with different letters within the same column are
significantly different (P < 0.05).
expressed as GAE, of leaf and stem extracts were 15.5 ± 1.9 and
17±3.8mgg −1 dryextract,respectively.Theamountofflavonoids,
expressed as QE, of the leaf extract was 16.14 ± 1.17 mg g −1 dry
extract which was 5.7 times higher than that of stem extract.
At the test concentrations, cell viability of the LPS-activated cells
treated with leaf or stem of CSEE was more than 90% of that of cells
treated with LPS alone, as assessed by mitochondrial reduction of
MTT after 18 h challenge (Tables 1 and 2).
As shown in Tables 1 and 2, stimulation of macrophages with
LPS resulted in a strong increase in NO and PGE2 production. A
dose-dependent decrease in NO and PGE2 production was noted
in cells treated with leaf or stem of CSEE in the presence of LPS. At
a concentration of 150 µgmL −1 , 80% and 75% reduction in nitrite
production were noted in cells treated with leaf and stem of CSEE,
www.soci.org T-T Wu et al.
Table 3. EffectsofstemofCSEEonLPS-inducediNO,COX-2,and
IL-1β mRNA expressions in RAW 264.7 macrophages
Treatment ∗ iNOS † COX-2 IL-1β
Control 0.7 ± 0.7e 0.8 ± 0.7d 0.1 ± 0.2d LPS 100.0 ± 0.0a 100.0 ± 0.0a 100.0 ± 0.0a LPS + Stem of CSEE
25 µgmL−1 59.1 ± 6.7b 66.0 ± 4.7b 78.8 ± 8.8ab LPS + Stem of CSEE
50 µgmL−1 59.8 ± 5.1b 63.1 ± 9.0b 71.2 ± 6.4b LPS + Stem of CSEE
100 µgmL−1 42.1 ± 7.2c 54.6 ± 11.6c 51.2 ± 10.2c LPS + Stem of CSEE
150 µgmL−1 28.0 ± 4.4d 45.3 ± 2.7c 47.8 ± 9.0c ∗ iNOS mRNA expression was measured by RAW 264.7 macrophages
treated with or without LPS (10 ng mL −1 ) plus DMSO vehicle control
and 25 to 150 µgmL −1 stem of CSEE for 8 h. COX-2 and IL-1β
mRNA expression were measured in cells preincubated with 25 to
150 µgmL −1 stem or leaf of CSEE for 1 h and then treated with either
DMSO vehicle control or 10 ng mL −1 LPS for 6 h.
† Data are the mean ± SD of at least four separate experiments and are
expressed as the percentage of the culture treated with LPS alone.
a,b,c,d Means with different letters within the same column are
significantly different (P < 0.05).
Table 4. EffectsofleafofCSEEonLPS-inducediNO,COX-2,andIL-1β
mRNA expressions in RAW 264.7 macrophages
Treatment ∗ iNOS † COX-2 IL-1β
Control 0.7 ± 0.7e 0.8 ± 0.7e 0.1 ± 0.2d LPS 100.0 ± 0.0a 100.0 ± 0.0a 100.0 ± 0.0a LPS + Leaf of CSEE
25 µgmL−1 79.5 ± 8.8b 47.8 ± 1.2b 94.5 ± 5.4a LPS + Leaf of CSEE
50 µgmL−1 42.6 ± 6.2c 46.7 ± 4.1bc 71.1 ± 8.7b LPS + leaf of CSEE
100 µgmL−1 23.1 ± 5.8d 26.6 ± 4.2cd 62.6 ± 17.0b LPS + Leaf of CSEE
150 µgmL−1 14.1 ± 3.8d 21.6 ± 2.8d 42.0 ± 6.2c ∗ iNOS mRNA expression was measured by RAW 264.7 macrophages
treated with or without LPS (10 ng mL −1 ) plus DMSO vehicle control
and 25 to 150 µgmL −1 leaf of CSEE for 8 h. COX-2 and IL-1β
mRNA expression were measured in cells preincubated with 25 to
150 µgmL −1 stem or leaf of CSEE for 1 h and then treated with either
DMSO vehicle control or 10 ng mL −1 LPS for 6 h.
† Data are the mean ± SD of at least four separate experiments and are
expressed as the percentage of the culture treated with LPS alone.
a,b,c,d Means with different letters within the same column are
significantly different (P < 0.05).
respectively. The PGE2 levels in cells treated with 150 µgmL −1 of
the leaf and stem of CSEE were 25.2 ± 2.4% and 52.5 ± 7.7%,
respectively, of the level of cells treated with LPS alone.
The immunoblot assay showed that the protein expression
of iNOS, COX-2 and proIL-1β was undetectable in resting RAW
264.7 macrophages and was highly induced in the presence of
LPS. The addition of exogenous leaf or stem of CSEE significantly
reduced LPS-induced protein expression of iNOS, COX-2 and pro-
IL-1β (P < 0.05, Fig. 1). As noted for the changes in protein
expression, real-time RT-PCR further showed that LPS-induced
mRNA expression of iNOS, COX-2 and pro-IL-1β was significantly
decreased by leaf and stem of CSEE (Tables 3 and 4).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1846–1854

Effect of C. sativum on inflammatory responses in macrophages www.soci.org
Figure 3. Effects of stem and leaf of CSEE on LPS-induced of activation of
MAPKs.RAW264.7 macrophages were preincubated with 25–150 µgmL −1
stem or leaf of CSEE for 1 h and then treated with either DMSO vehicle
control or 10 ng mL −1 LPS for 30 min. Cells were lysed and western
blotting was performed with the antibodies for phosphorylated (A) ERK
1/2, (B) p38 and (C) JNK and the cells were then reprobed with antibodies
against the corresponding MAPKs. The ratios of immunointensity
between the MAPKs and the phosphorylated MAPKs are shown and
are expressed as the percentage of the culture treated with LPS
alone. Data are the mean ± SD of at least four separate experiments
and values not sharing the same letter are significantly different
(P < 0.05).
Figure 4. Effects of stem and leaf of CSEE on activation of NF-κB. (A) RAW
264.7 macrophages were preincubated with 150 µgmL −1 stem or leaf
of CSEE for 14 h and then treated with either DMSO vehicle control or
10 ng mL −1 LPS for 30 min. EMSA experiments were carried out by using
the LightShift Chemiluminescent EMSA Kit from Pierce Chemical Co. The
unlabelled double-stranded oligonucleotides of NF-κB and the unlabelled
double-stranded mutant NF-κB oligonucleotide were added for the
competition assay and specificity assay, respectively. Bands were detected
by using streptavidin–horseradish peroxidase and were developed by
using a SuperSignal West Pico kit from Pierce Chemical Co. (B) Cells were
transiently transfected with pSV-β-galactosidase and pNF-κB-Luc reporter
gene for6 h and cells were treated with eithervehicle control or10 ng mL −1
LPS plus 150 µgmL −1 stem or leaf of CSEE for 8 h. Cells were harvested
and the level of luciferase and β-galactosidase activity were measured by
the Luciferase Assay System and β-Galactosidase Enzyme Assay System
with Reporter Lysis Buffer from Promega Co., respectively. Data are the
mean ± SD of at least three separate experiments and are expressed as
the percentage of the culture treated with LPS alone. Values not having
the same letter are significantly different (P < 0.05).
Upon LPS treatment, the amounts of cytoplasmic phosphorylated
IκB-α protein and nuclear p65 protein were greatly
increased compared with those of the control (Fig. 2). Addition of
150 µgmL −1 leaf or stem of CSEE significantly abolished the level
of LPS-induced protein expression of phosphorylated IκB-α and
nuclear p65 (P < 0.05).
To test whether the mitogen-activated protein kinase (MAPK)
signalling pathway was involved in the anti-inflammatory property
of C. sativum, we examined the effect of leaf and stem of CSEE
on LPS-induced MAPK activation. LPS treatment resulted in strong
increases in the amounts of phosphorylated ERK1/2, p38 and JNK-1
expression (P < 0.05, Fig. 3). Addition of stem extracts significantly
reduced LPS-induced phosphorylated JNK and p38. However,
LPS-induced activation of ERK1/2 was diminished only by high
doses of CSEE stem. Addition of high doses of leaf extracts
significantly inhibited LPS-induced activation of MAPKs. The
amount of the unphosphorylated form of MAPKs was not
influenced by the LPS treatment or LPS plus stem or leaf of
CSEE.
EMSA experiments were used to evaluate the effect of C.sativum
on activation of NF-κB. As shown in Fig. 4A, the nuclear extract
from LPS-stimulated macrophages showed a marked increase in
NF-κB nuclear protein DNA-binding activity compared with that
J Sci Food Agric 2010; 90: 1846–1854 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1851

1852
www.soci.org T-T Wu et al.
Figure 5. The HPLC-UV (UV 254 nm) and selectively extracted ion (rutin: [M-H] − , m/z 609.0) chromatograms of CSEE of stem (A, B) and leaf (C, D). 1 and 2:
unknown compounds; 3: rutin.
in unstimulated macrophages. Pretreatment of cells with leaf or
stem of CSEE suppressed the activation of NF-κB binding to its
consensus DNA sequences. The specificity of the NF-κB nuclear
protein–DNA binding was verified by competition assay with a
50-fold excess of unlabelled NF-κB probe and unlabelled mutant
NF-κBprobe.
To investigate the transcriptional activity of NF-κB, the
expression of reporter genes in cells transfected with pNF-κB-
Luc and the internal control pSV-β-galactosidase were analysed.
Consistent with the EMSA assay result, the expression of
LPS-induced NF-κB-Luc activity was significantly inhibited in
cultures treated with 150 µgmL −1 leaf or stem of CSEE (Fig. 4B,
P < 0.05).
LC-MS analysis of the testing samples and several authentic
standards revealed that only rutin is identified in both the stem
andleafofCSEE.Rutinshowedthe[M-H] − ion at m/z 609 and
its retention time was 31.9 min. Based on the peak area of mass
spectral peak, the rutin concentrations in stem and leaf extracts
were 130.5 and 42.0 µgg −1 , respectively. Two unknown large UV
peaks presented at 48.3 min ([M-H] − ion at m/z 253) and 52.6 min
([M-H] − ion at m/z 221) were not identified (Fig. 5).
DISCUSSION
Numerous studies have focused on herbal remedies and botanicals
because they offer much promise in health benefits and disease
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1846–1854

Effect of C. sativum on inflammatory responses in macrophages www.soci.org
treatments without excessive side effects and cytotoxicity. 33 A
wide variety of plant-derived products, especially spices, have
shown an anti-inflammatory effect but only a few have been
examined to determine the molecular mechanism of this inhibitory
action. 34 In this study, we investigated the anti-inflammatory
properties of aerial parts of CSEE and dissected the possible
molecular mechanism of action of C. sativum. The data presented
herein showed that both the leaf and stem of CSEE significantly
decreased NO and PGE2 production. The inhibition of NO and
PGE2 was due to the inhibition of iNOS and COX-2 expression
respectively, at mRNA and protein levels as shown by real-time
RT-PCR and western blot. Moreover, we demonstrated that the
aerial part of CSEE suppressed iNOS, COX-2 and IL-1β expression,
acting at the transcriptional level possibly via inhibition of the LPSinduced
MAPK pathway and transcription factor NF-κB activation.
Considerable interest in immunomodulation therapy is now
focusedonblockingtheactivationofNF-κBinmacrophages,which
resultsinsuppressingarangeofinflammatorymediatorexpression
such as iNOS, COX-2 and IL-1β. 2,35,36 Our data clearly showed
that CSEE effectively inhibited the NF-κB pathway by blocking
LPS-induced IκB-α phosphorylation, nuclear p65 expression and
subsequent DNA binding affinity and transcriptional activation.
These results suggested that stem and leaf of CSEE decreased the
expression of pro-inflammatory mediators via down-regulating
the NF-κB pathway in stimulated macrophages.
MAPKs, one of the most important intracellular signalling
pathways, are a family of serine/threonine protein kinases, which
include JNK, ERK and p38 kinase subgroups at least in mammalian
cells. MAPK pathways are involved in a battery of cellular events,
including cell proliferation and growth, cell differentiation, cell
movement, cellular senescence and apoptosis. 37 Although the
exact signal pathways of MAPKs are still unclear, LPS-induced
phosphorylation and activation of MAPKs in macrophages lead
to the production of pro-inflammatory mediators as a result of
the activation of transcription factors including NF-κB. 38 In the
present study, the aerial part of CSEE significantly decreased
LPS-induced phosphorylation of the three MAPKs, which implies
that the inflammatory signal transduction by the MAPK pathways
could be impeded by C. sativum in LPS-induced macrophages.
Numerous studies have demonstrated that phenolic compounds
in spices contribute to the health benefits of spices. 39 A previous
study indicated that luteolin, vicenin, ferulic acid and arbutin were
the main components in the aerial part of CSEE. 32 However, in this
study, only rutin was identified in the stem and leaf of CSEE and
the rutin concentration in stem extracts (130.5 µgg −1 )washigher
than that in the leaf extracts (42.0 µgg −1 ). It was interesting to
note that the total amount of flavonoids in the stem extracts was
also lower than that in the leaf extracts, although there was no
significant difference in the amount of total phenolics between
stem and leaf of CSEE. Furthermore, the amount of rutin in the
IC50 values of the stem and leaf of CSEE against LPS-induced
NO production in RAW264.7 macrophage was 8.2 µgmL −1 and
2.7 µgmL −1 , much lower than the reported IC50 value of rutin
on LPS-induced NO production (25.3 µgmL −1 ). 40 These results
indicated that it is not only rutin which provides contributions
to the anti-inflammatory properties of C. sativum, but other
yet unknown compounds in the stem and leaf of CSEE may
present further contributions as well. It is noteworthy that the
antimutagenicity of C. sativum was dependent on the chlorophyll
content in C. sativum juice. 14 Moreover, the chlorophyllin, a watersoluble
derivative of chlorophyll, inhibited NO production and
iNOS expression by modulating LPS-induced NF-κB activation in
RAW264.7 macrophages. 41 Therefore, the amount of chlorophyll
in CSEE is likely to play an important in its anti-inflammatory
property.
In conclusion, we have demonstrated that both the leaf and
the stem of CSEE modulate LPS-induced inflammatory events in
RAW 264.7 macrophages. This inhibitory activity of C. sativum, at
least in part, occurs through C. sativum modulating the NF-κB
activation and MAPK pathway. According to these experimental
results supporting the anti-inflammatory property of the leaf and
stem of CSEE, it would be worthwhile to explore the biomedical
importance of the aerial parts of CSEE in the treatment and
prevention of chronic inflammatory related diseases.
ACKNOWLEDGEMENTS
This research was funded by the National Science Council, Republic
of China, under Grant NSC 96-2320-B-040-027-MY3.
REFERENCES
1 Fujiwara N and Kobayashi K, Macrophages in inflammation. Curr Drug
Targets Inflamm Allergy 4:281–286 (2005).
2 Stylianou E and Saklatvala J, Interleukin-1. Int J Biochem Cell Biol
30:1075–1079 (1998).
3 Korhonen R, Lahti A, Kankaanranta H and Moilanen E, Nitric oxide
productionandsignalingininflammation.CurrDrugTargetsInflamm
Allergy 4:471–479 (2005).
4 Park JY,Pillinger MH and Abramson SB,Prostaglandin E2 synthesis and
secretion: the role of PGE2 synthases. Clin Immunol 119:229–240
(2006).
5 Kundu JK and Surh YJ, Inflammation: gearing the journey to cancer.
Mutat Res 659:15–30 (2008).
6 Manzi S and Wasko MC, Inflammation-mediated rheumatic diseases
and atherosclerosis. Ann Rheum Dis 59:321–325 (2000).
7 King GL, The role of inflammatory cytokines in diabetes and its
complications. JPeriodontol79:1527–1534 (2008).
8 Chen F, Castranova V, Shi X and Demers LM, New insights into the role
of nuclear factor-kappaB, a ubiquitous transcription factor in the
initiation of diseases. Clin Chem 45:7–17 (1999).
9 Makarov SS, NF-kappaB as a therapeutic target in chronic
inflammation: recent advances. Mol Med Today 6:441–448 (2000).
10 Tapsell LC, Hemphill I, Cobiac L, Patch CS, Sullivan DR, Fenech M, et al,
Health benefits of herbs and spices: the past, the present, the future.
Med J 185:S4–S24 (2006).
11 Gray AM and Flatt PR, Insulin-releasing and insulin-like activity of the
traditional anti-diabetic plant Coriandrum sativum (coriander). Br
JNutr81:203–209 (1999).
12 Lal AA, Kumar T, Murthy PB and Pillai KS, Hypolipidemic effect of
Coriandrum sativum L. in triton-induced hyperlipidemic rats. Indian
JExpBiol42:909–912 (2004).
13 Sreelatha S, Padma PR and Umadevi M, Protective effects of
Coriandrum sativum extracts on carbon tetrachloride-induced
hepatotoxicity in rats. Food Chem Toxicol 47:702–708 (2009).
14 Cortes-Eslava J, Gomez-Arroyo S, Villalobos-Pietrini R and Espinosa-
Aguirre JJ, Antimutagenicity of coriander (Coriandrum sativum)
juiceonthemutagenesisproducedbyplantmetabolitesofaromatic
amines. Toxicol Lett 153:283–292 (2004).
15 Medhin DG, Hadhazy P, Bakos P and Verzar-Petri G, Hypotensive
effects of Lupinus termis and Coriandrum sativum in anaesthetized
rats. A preliminary study. Acta Pharm Hung 56:59–63 (1986).
16 Bajpai M, Mishra A and Prakash D, Antioxidant and free radical
scavenging activities of some leafy vegetables. Int J Food Sci Nutr
56:473–481 (2005).
17 Emamghoreishi M, Khasaki M and Aazam MF, Coriandrum sativum:
evaluation of its anxiolytic effect in the elevated plus-maze.
J Ethnopharmacol 96:365–370 (2005).
18 KubobI,FujitaK,KuboA,NiheiKandOguraT,Antibacterialactivity
of coriander volatile compounds against Salmonella choleraesuis.
J Agric Food Chem 52:3329–3332 (2004).
19 Al-Said MS, Al-Khamis KI, Islam MW, Parmar NS, Tariq M and Ageel AM,
Post-coital antifertility activity of the seeds of Coriandrum sativum
in rats. J Ethnopharmacol 21:165–173 (1987).
J Sci Food Agric 2010; 90: 1846–1854 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1853

1854
20 Bajpai M, Mishra A and Prakash D, Antioxidant and free radical
scavenging activities of some leafy vegetables. Int J Food Sci Nutr
56:473–481 (2005).
21 AgaM,IwakiK,UedaY,UshioS,MasakiN,FukudaS,et al, Preventive
effect of Coriandrum sativum (Chinese parsley) on localized lead
deposition in ICR mice. J Ethnopharmacol 77:203–208 (2001).
22 Karunasagar D, Krishna MV, Rao SV and Arunachalam J, Removal and
preconcentration of inorganic and methyl mercury from aqueous
mediausingasorbentprepared from the plant Coriandrumsativum.
J Hazard Mater 118:133–139 (2005).
23 O’Mahony R, Al-Khtheeri H, Weerasekera D, Fernando N, Vaira D,
Holton J, et al, Bactericidal and anti-adhesive properties of
culinary and medicinal plants against Helicobacter pylori. World
J Gastroenterol 11:7499–7507 (2005).
24 Reuter J, Huyke C, Casetti F, Theek C, Frank U, Augustin M, et al, Antiinflammatory
potential of a lipolotion containing corianderoil in the
ultraviolet erythema test. J Dtsch Dermatol Ges 6:847–851 (2008).
25 Bor JY, Chen HY and Yen GC, Evaluation of antioxidant activity and
inhibitory effect on nitric oxide production of some common
vegetables. J Agric Food Chem 54:1680–1686 (2006).
26 Nurmi K,Ossipov V,Haukioja EandPihlaja K,Variationoftotalphenolic
content and individual low-molecular-weight phenolics in foliage
of mountain birch trees (Betula pubescens ssp. tortuosa). JChemEcol
22:2023–2040 (1996).
27 Jia Z, Tang M and Wu J, The determination of flavonoid contents in
mulberry and their scavenging effects on superoxide radicals. Food
Chem 64:555–559 (1999).
28 Denizot F and Lang R, Rapid colorimetric assay for cell growth and
survival. Modifications to the tetrazolium dye procedure giving
improved sensitivity and reliability. J Immunol Methods 89:271–277
(1986).
29 Green LC, Wagner DA, Glogowski J, Skipper PL, Wishnok JS and
Tannenbaum SR, Analysis of nitrate, nitrite, and [ 15 N]-nitrate in
biological fluids. Anal Biochem 126:131–138 (1982).
30 Livak KJ and Schmittgen TD, Analysis of relative gene expression data
using real-time quantitative PCR and the 2[−Delta Delta C(T)]
method. Methods 25:402–408 (2001).
www.soci.org T-T Wu et al.
31 Liu KL, Chen WC, Wang RY, Lei YP, Sheen LY and Lii CK, DATS reduces
LPS-induced iNOS expression, NO production, oxidative stress and
NF-κB activation in RAW 264.7 macrophages. J Agric Food Chem
54:3472–3478 (2006).
32 Oganesyan ET, Nersesyan ZM and Parkhomenko AY, Chemical
composition of the above-ground part of Coriandrum sativum.
Pharm Chem J 41:149–153 (2007).
33 Jachak SM and Saklani A, Challenges and opportunities in drug
discovery from plant. Curr Sci 92:1251–1257 (2007).
34 Aggarwal BB and Shishodia S, Suppression of the nuclear factorkappaB
activation pathway by spice-derived phytochemicals:
reasoning for seasoning. Ann N Y Acad Sci 1030:434–441 (2004).
35 Appleby SB, Ristimaki A, Neilson K, Narko K and Hla T, Structure of the
human cyclo-oxygenase-2 gene. Biochem J 302:723–727 (1994).
36 Xie QW, Kashiwabara Y and Nathan C, Role of transcription factor
NF-kappa B/Rel in induction of nitric oxide synthase. Biochem J
269:4705–4708 (1994).
37 Schramek H, MAP kinases: from intracellular signals to physiology and
disease. News Physiol Sci 17:62–67 (2002).
38 KimHG,YoonDH,LeeWH,HanSK,ShresthaB,KimCH,et al, Phellinus
linteusinhibitsinflammatorymediatorsbysuppressingredox-based
NF-kappaB and MAPKs activation in lipopolysaccharide-induced
RAW 264.7 macrophage. J Ethnopharmacol 114:307–315 (2007).
39 Surh Y, Molecular mechanisms of chemopreventive effects of selected
dietary and medicinal phenolic substances. Mutat Res 428:305–327
(1999).
40 Chen YC, Shen SC, Lee WR, Hou WC, Yang LL and Lee TJ, Inhibition
of nitric oxide synthase inhibitors and lipopolysaccharide induced
inducible NOS and cyclooxygenase-2 gene expressions by rutin,
quercetin, and quercetin pentaacetate in RAW 264.7 macrophages.
J Cell Biochem 82:537–48 (2001).
41 Cho KJ, Han SH, Kim BY, Hwang SG, Park KK, Yang KH, et al,
Chlorophyllin suppression of lipopolysaccharide-induced nitric
oxide production in RAW 264.7 cells. Toxicol Appl Pharmacol
166:120–127 (2000).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1846–1854

Research Article
Received: 1 February 2010 Revised: 27 April 2010 Accepted: 28 April 2010 Published online in Wiley Interscience: 3 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4024
Antioxidant activity of peptide fractions
derived from cottonseed protein hydrolysate
Dandan Gao, Yusheng Cao ∗ and Haixing Li
Abstract
BACKGROUND: Cottonseed protein is widely regarded as a potential source of nutrients for humans and animals, but it is mainly
used as forage in China. In the present study, Neutrase was employed to hydrolyse cottonseed protein to produce a hydrolysate
with antioxidant activity suitable for conversion to high-value products. The antioxidant potential of the cottonseed protein
hydrolysate (CPH) and its fractions was investigated using different in vitro methods. Furthermore, the amino acid composition
of the CPH fractions was determined to evaluate the relationship between antioxidant activity and amino acid composition.
RESULTS: The CPH prepared using Neutrase was separated into four fractions (I, II, III and IV) by gel filtration on Sephadex G-25.
All fractions were effective antioxidants, with fraction III (0.8–1.2 kDa) showing the strongest activity. The amino acid analysis
showed that fraction III also had the highest total amino acid content (616.8 g kg −1 protein) and was rich in Phe, His, Pro, Met,
Ile and Cys compared with the other fractions.
CONCLUSION: The results showed that the hydrolysate derived from cottonseed protein, particularly fraction III, could be a
natural antioxidant source suitable for use as a food additive.
c○ 2010 Society of Chemical Industry
Keywords: cottonseed protein hydrolysate; peptide fractions; free radicals; antioxidant activity; amino acid composition
INTRODUCTION
Free radicals are constantly generated in the body tissues as a
result of oxidative metabolism. There is much evidence that free
radicals play a critical role in a variety of pathological conditions,
including the processes of aging, cancer, multiple sclerosis,
inflammation, coronary heart and cardiovascular diseases, senile
dementia, arthritis and atheroscelerosis. 1 Free radicals are also
of great concern in the food industry, since the oxidation of
fats and oils is one of the major causes of the deterioration
of the quality of lipid-containing foods during processing and
storage. In order to provide protection against serious diseases
and to prevent foods from undergoing deterioration, many
chemicals with strong antioxidant activity are used as additives,
such as butylated hydroxyanisole, butylated hydroxytoluene and
n-propyl gallate. However, their use in foodstuffs is restricted
or prohibited in some countries because of the potential risks
of artificial antioxidants in vivo. Therefore there is a growing
interest in identifying antioxidant properties in natural sources,
including some dietary protein compounds. Recently, protein
hydrolysates from different sources such as soybean, silver carp,
wheat gluten, whey and rapeseed have been found to possess
antioxidant activity. 2–6 The peptides from natural proteins with
various activities have become a topic of great interest for the
pharmaceutical, health food and preservation industries.
Cotton is not only the most important fibre crop in the world
but also the second best potential source for plant proteins after
soybean. Cottonseed protein is widely regarded as a potential
source of nutrients for human and animals and has been the
subject of numerous investigations. 7 In China, about 10 million
tons of cottonseed is produced annually. The presence of toxic
gossypol in the cottonseed meal is a limiting factor for human
consumption, and various methods have been proposed for
reduction of the free gossypol content to the allowable limit
of 0.45 g kg −1 (FDA regulations, 1974). The preparation method
and foaming and emulsifying properties of cottonseed protein
isolate (CPI) have been studied. However, to our knowledge, there
have been no reports on the antioxidant activity of cottonseed
protein hydrolysate (CPH). In the present study, CPH was prepared
by enzymatic hydrolysis and separated into four fractions by
gel filtration on Sephadex G-25. The antioxidant potential of
the 4 h CPH and its fractions was investigated by measuring
their inhibitory effect on the autoxidation of linoleic acid and
their scavenging ability on 1,1-diphenyl-2-picrylhydrazyl, hydroxyl
radical and superoxide radical. Furthermore, the amino acid
composition of the CPH fractions was determined to evaluate
the relationship between antioxidant activity and amino acid
composition.
MATERIALS AND METHODS
Materials
Non-toxic cottonseed meal (650 g kg −1 protein) was obtained
from China Cotton-Unis Co. (Beijing, China). Neutrase was obtained
∗ Correspondence to: Yusheng Cao, State Key Laboratory of Food Science
and Technology, Sino-German Joint Research Institute, Nanchang University,
Nanchang 330047, China. E-mail: yyssccc@hotmail.com
State Key Laboratory of Food Science and Technology, Sino-German Joint
Research Institute, Nanchang University, Nanchang 330047, China
J Sci Food Agric 2010; 90: 1855–1860 www.soci.org c○ 2010 Society of Chemical Industry
1855

1856
from Novozymes A/S (Novo Nordisk, Bagsvaerd, Denmark).
Butylated hydroxytoluene (BHT), 1,1-diphenyl-2-picrylhydrazyl
(DPPH), linoleic acid, 2-thiobarbituric acid (TBA), bovine serum
albumin (BSA), lysozyme, insulin from bovine pancreas, vitamin
B12,oxidisedglutathioneandreducedglutathionewerepurchased
from Sigma Chemical Co. (St Louis, MO, USA). All other reagents
used were of analytical grade.
Production of CPI
CPI was prepared according to the method of Tunc and Duman 8
with slight modification. Defatted non-toxic cottonseed flour was
mixed with NaOH solution (pH 10.5) in a ratio of 1 : 15 (w/v)
and incubated at 45 ◦ C for 2 h. The mixture was subsequently
centrifuged at 8000 × g for 15 min and the supernatant was
collected. After adjusting its pH to the isoelectric point (pH 4.3)
with 0.1 mol L −1 HCl, the supernatant was centrifuged at 8000 × g
for 15 min. The precipitate was lyophilised and stored at −20 ◦ C.
Enzymatic hydrolysis and degree of hydrolysis
Neutrase was employed in this study. CPI (10 g) was suspended
in distilled water (200 mL) and heated in a water bath at 90 ◦ C
for 15 min prior to enzymatic hydrolysis. The enzymatic hydrolysis
was carried out at 50 ◦ C and pH 7.5 for 4 h at an enzyme/substrate
ratio of 6000 U g −1 . During hydrolysis the pH of the reaction
mixture was adjusted every 30 min with 1 mol L −1 NaOH. When
the reaction was finished, the enzyme was inactivated by heating
at 95 ◦ C for 15 min. The resulting hydrolysate was rapidly cooled
to ambient temperature in an ice bath and centrifuged at 8000 × g
for 15 min. The supernatant was freeze-dried and stored at −20 ◦ C
until use.
The degree of hydrolysis (DH) was determined at 30 min
intervals using a pH-stat method 9 based on the equation DH
(%) = (h/htot) × 100, where h = B × Nb × (1/a) × (1/MP), B =
base consumption (mL), Nb = concentration of base (1 mol L −1
NaOH), 1/a = calibration factor for pH-stat, MP = mass of
protein (g) andh = hydrolysis equivalents. For cottonseed protein,
htot = 7.21 mmol g −1 .
Size exclusion chromatography
TheCPHsamplewasseparatedonaSephadexG-25column(1.5 cm
× 50 cm) using an AKTA purifier (Amersham pharmacia biotech
In., Piscataway, NJ, USA). Phosphate buffer solution (10 mmol L −1 ,
pH 7.4) was used to equilibrate the column and to elute the
proteins at a flow rate of 0.3 mL min −1 . A fixed amount of sample
(1 mL) at a protein concentration of 100 mg mL −1 was applied to
the column, and the absorbance of the eluate was measured at
210 nm. A molecular weight calibration curve was plotted using
the following standards: BSA (66 kDa), lysozyme (14.4 kDa), insulin
from bovine pancreas (5.7 kDa), vitamin B12 (1.355 kDa), oxidised
glutathione (0.612 kDa) and reduced glutathione (0.307 kDa). The
CPH fractions were lyophilised and used for antioxidant activity
tests.
Measurement of antioxidant activity
Inhibition of linoleic acid autoxidation
The in vitro lipid peroxidation-inhibitory activity of peptides was
determined by assessing their ability to inhibit the oxidation of
linoleic acid in an emulsified model system. Briefly, the sample
(15 mg) was dissolved in 1.5 mL of 50 mmol L −1 phosphate buffer
(pH 7) and added to a mixture of ethanol (2.5 mL) and linoleic
www.soci.org D Gao, Y Cao, H Li
acid (32.5 µL), the final volume being adjusted to 5 mL with
distilled water. The reaction mixture was incubated in a screwcapped
tube at 60 ◦ C in the dark. At 24 h intervals, aliquots of the
reaction mixture were withdrawn for measurement of oxidation
by the TBA method 10 with modification. The reaction mixture
(100 µL) was added to a mixture of 0.8 mL of trichloroacetic acid
(TCA, 100 mg mL −1 ) and 1.5 mL of TBA (8 mg mL −1 )solutionin
water. The mixture was cooled in an ice bath and centrifuged
at 8000 × g for 10 min. The absorbance of the supernatant was
measured at 532 nm using an UltroSpec 4300 Pro UV–visible
spectrophotometer (Pharmacia Co., Peapack, NJ, USA). The
antioxidant activities of 10 mg mL −1 BHT and α-tocopherol were
also assayed under the same conditions for comparison purposes.
Scavenging effect on DPPH radical
DPPH radical-scavenging activity was determined according to
the method of Brand-Williams et al. 11 Briefly, 200 µL of test
sample (10 mg mL −1 ) was added to 200 µL of 0.1 mmol L −1 DPPH
dissolved in 950 mL L −1 ethanol. The mixture was shaken and
incubated for 30 min at room temperature in the dark. The
absorbance of the resulting solution was measured at 517 nm. For
comparison, ascorbate (10 mg mL −1 ) was used. The scavenging
effect was expressed by the following equation:
DPPH radical-scavenging activity (%) =
[(blank absorbance − sample absorbance)/blank absorbance] × 100
Scavenging effect on hydroxyl radical
The scavenging effect on hydroxyl radical was measured by the
deoxyribose method with modification. 12 The reaction mixture
contained 1 mL of 0.2 mol L −1 sodium phosphate buffer (pH 7.4),
0.2mL of 10mmolL −1 2-deoxyribose, 0.2 mL of 10 mmol L −1
FeSO4/ethylene diamine tetraacetic acid (EDTA), 0.2 mL of
10 mmol L −1 hydrogen peroxide, 0.2 mL of distilled water and
0.2 mL of sample solution (10 mg mL −1 ). The reaction was initiated
by the addition of hydrogen peroxide. The reaction solution
was incubated at 37 ◦ C for 1 h, then the reaction was stopped
by the addition of 1 mL of TCA (28 mg mL −1 )and1mLofTBA
(10 mg mL −1 ). The mixture was boiled for 10 min and cooled
in ice, then its absorbance was measured at 532 nm. Ascorbate
(10 mg mL −1 ) was tested under the same conditions for comparison.Theresultswerecalculatedaspercentageinhibitionaccording
to the following equation:
hydroxyl radical-scavenging activity (%) =
[(Ac − As)/(Ac − A0)] × 100
where Ac, As and A0 are the absorbances of the control, sample
and blank sample respectively.
Scavenging effect on superoxide radical
The superoxide radical-scavenging capacity of the 4 h CPH
and its fractions was examined using a pyrogallol autoxidation
system with modification. 13 The reaction mixture contained
70 µL of pyrogallol (10 mmol L −1 ), 4.5 mL of Tris/HCl/EDTA buffer
(50 mmol L −1 ,pH8)and0.5mLofsample(10mgmL −1 ). The
reaction solution was incubated at 25 ◦ Cfor30min,thenits
absorbance was measured at 325 nm by spectrophotometry.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1855–1860

Antioxidant activity of cottonseed protein hydrolysate fractions www.soci.org
Ascorbate (10 mg mL −1 ) was used for comparison. The scavenging
rate was calculated according to the following equation:
superoxide radical-scavenging rate (%) = [1 − (A1 − A2)/A0]×100
where A0 is the absorbance of the control (without sample), A1
is the absorbance in the presence of the sample and A2 was the
absorbance of the sample without pyrogallol.
Amino acid composition
The lyophilised CPH fractions were hydrolysed with 6 mol L −1
HCl at 110 ◦ C for 24 h under nitrogen and their derivatives were
prepared using phenyl isothiocyanate prior to high-performance
liquid chromatography (HPLC) analysis. Total and free amino acids
were analysed by HPLC using a Capcell Pak (Shiseido CO., Tokyo,
Japan) C18 column (4.6 mm × 250 mm) and UV detection at a
flow rate of 1 mL min −1 . Seventeen amino acid standards were
analysed at equal concentration. The amino acid composition was
expressed as g amino acid kg −1 protein.
Statistical analysis
All tests were executed in triplicate and mean values were
calculated. Student’s t test was used to calculate the significance of
differences between mean effects of a given compound compared
with the control. Statistical significance was defined as P < 0.05.
RESULTS AND DISCUSSION
Enzymatic hydrolysis
In quantitative work on protein hydrolysis it is necessary to have
a measurement for the extent of hydrolytic degradation. It should
be evident that the number of peptide bonds cleaved during the
reaction is the parameter that most closely reflects the catalytic
action of proteases. 9 The DH is defined as the percentage ratio of
the number of peptide bonds broken (h) to the total number of
bonds per unit weight (htot). It is generally used as a parameter for
monitoring proteolysis and is the most widely used indicator for
comparison among different protein hydrolysates. In our study the
DH of the reaction solution increased with increasing incubation
time, maintaining a high rate during the initial 90 min and then
slowing down (Fig. 1), which implied that the maximum cleavage
of peptides occurred within 90 min of hydrolysis. Thereafter the
DH reached a plateau at 24.84% after 4 h. This result was similar to
previous reports on the hydrolysis of other proteins such as those
of silver carp and wheat gluten. 2,3
Fractionation of CPH
The CPH sample was fractionated by gel filtration column
chromatography on Sephadex G-25 (Fig. 2). The hydrolysate
was successfully separated into four fractions, I, II, III and IV,
corresponding to molecular weights of >3.8, 1.2–3.8, 0.8–1.2
and 4hCPH> fraction II > fraction IV.
It was reported that the DPPH radical-scavenging activity of food
protein hydrolysates may depend on the size of their constituent
peptides. 16 Wang et al. 5 studied the antioxidant properties of
wheat gluten hydrolysate. The hydrolysate was produced with
papain and separated through an ultrafiltration membrane with a
molecular weight cut-off of 5 kDa. They found that the 5 kDa
permeate fraction had the strongest antioxidant activity and
J Sci Food Agric 2010; 90: 1855–1860 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1857

1858
Figure 2. Pattern of CPH fractions separated by gel filtration on Sephadex G-25.
Figure 3. Inhibitory effects of 4 h CPH and its fractions on lipid peroxidation
measured in linoleic acid model systemover9days.HigherUVabsorbance
at 532 nm represents higher linoleic acid peroxidation. Values are mean
± standard deviation of three experiments. All experimental values were
significant at P < 0.05 compared with the control.
that the efficacy of the peptides depended on their molecular
weight. Peptides with DPPH radical-scavenging activity derived
from various enzymatic food protein hydrolysates have been
reported in recent years. 5,15,16 Our results suggest that the CPH
fractions possibly contained some substrates that were electron
donors and could react with free radicals to convert them to more
stable products, thus terminating the radical chain reaction.
Scavenging effect on hydroxyl radical
Hydroxyl radicals are extremely reactive species and induce severe
damage to adjacent biomolecules, resulting in lipid peroxidation
in biological systems. Therefore removal of hydroxyl radicals is
probably one of the most effective defences of a living body
against various diseases. In this study, hydroxyl radical-scavenging
ability was measured by the 2-deoxyribose oxidation method.
Table 1 shows the hydroxyl radical-scavenging effects of the 4 h
CPH and its fractions. Fraction III exhibited the strongest hydroxyl
radical-scavenging activity (52.67%). In addition, fraction II, the 4 h
CPH, fraction I and fraction IV at the same concentration exhibited
www.soci.org D Gao, Y Cao, H Li
Table 1. Free radical-scavenging effects of CPH and its fractions
Free radical-scavenging effect (%)
Test sample DPPH Hydroxyl Superoxide
4hCPH 50.03 ± 0.91 47.33 ± 1.12 34.59 ± 3.31
Fraction I 50.27 ± 1.19 47.12 ± 0.80 14.81 ± 4.64
Fraction II 48.08 ± 0.59 48.61 ± 0.67 51.06 ± 1.31
Fraction III 72.19 ± 1.55 52.67 ± 0.32 73.23 ± 2.75
Fraction IV 40.60 ± 1.17 46.59 ± 0.63 15.13 ± 0.55
Ascorbate 77.36 ± 1.18 69.19 ± 0.49 79.31 ± 0.34
48.61,47.33,47.12and46.59%hydroxylradical-scavengingactivity
respectively. Thus all tested fractions possessed hydroxyl radicalscavenging
activity. The antioxidant activity of hydrolysates from
many kinds of food proteins has been studied in recent years. Peng
et al. 15 reported that whey protein hydrolysate and its peptide
fractions showed antioxidant properties against hydroxyl radical
similar to the results of this study. Li et al. 17 obtained a peptide
fraction from chickpea protein hydrolysate showing 81.39%
hydroxyl radical-scavenging ability. Kim et al. 18 found that hoki
(Johnius belengerii) frame protein hydrolysate exhibited 84.93%
hydroxyl radical-scavenging activity. It is difficult to compare the
results from related studies, because the antioxidant assay systems
employed are frequently modified, so that the antioxidants most in
use, such as ascorbate, BHT and butylated hydroxyanisole, should
be applied as a comparison at a certain concentration to evaluate
the antioxidant ability of the test sample.
Scavenging effect on superoxide radical
Superoxide radicals are generated by a number of biological
reactions. Although they do not directly initiate lipid oxidation, superoxide
radical anions are potential precursors of highly reactive
species such as hydroxyl radicals and hydrogen peroxide. 19 Not
only superoxide anion radicals but also their derivatives are celldamaging,
which can cause damage to the DNA and membrane
of cells. Therefore it is of great importance to scavenge superoxide
anion radicals.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1855–1860

Antioxidant activity of cottonseed protein hydrolysate fractions www.soci.org
Pyrogallic acid can automatically oxidise under alkaline conditions
to produce superoxide radicals directly, the constant rate of
this autoxidation reaction being dependent on the pyrogallic acid
concentration. The results of our experiment showed that fraction
III possessed the highest superoxide radical-scavenging activity
(73.23%), which was slightly inferior to that of ascorbate (79.31%).
The scavenging effect of fraction III on superoxide radicals was
similar to that of antioxidant peptides isolated from gelatin hydrolysates
obtained from the skin of sole and squid. 20 Fraction
II, the 4 h CPH, fraction IV and fraction I showed 51.06, 34.59,
15.13 and 14.81% superoxide radical-scavenging ability respectively.
Based on the results described above, fraction III had good
free radical-scavenging activity and could be a potential source of
natural antioxidants.
Amino acid composition
Kim et al. 21 reported that several amino acids such as His, Pro,
Ala and Leu contribute to the scavenging of free radicals. For
protein hydrolysates and peptides an increase in hydrophobicity
will increase their solubility in lipid and therefore enhance their
antioxidantactivity. 22 In particular, His-containing peptides exhibit
strong antioxidant activity owing to the decomposition of the
imidazole group of His. 23 In the case of wheat gelatin peptides the
abundance of amino acids such as His, Leu, Val and Ala present
in the sequence of hydrolysate peptides favours their radicalscavenging
properties. 5 The amino acid composition of the CPH
fractions in the present study is shown in Table 2. Fraction III was
rich in Phe, His, Pro, Met, Ile and Cys compared with the other
fractions. In addition, the total amino acid content in fraction III
was higher than that in the other fractions. It was demonstrated
that the antioxidant activity of the CPH fractions was related to the
composition and sequence of amino acids and that the abundance
of the amino acids Phe, His, Pro, Met, Ile and Cys in fraction III may
correlate with its strong antioxidant activity.
Table 2. Amino acid (AA) composition of CPH fractions
Composition (g kg −1 protein)
Amino acid Fraction I Fraction II Fraction III Fraction IV
Aspartic acid 25.18 42.89 40.23 16.19
Glutamic acid 137.60 51.80 13.56 2.24
Serine 11.03 9.18 3.50 1.08
Histidine 7.52 10.22 12.79 10.86
Arginine 62.05 33.10 25.47 –
Glycine 12.33 12.32 6.63 1.55
Threonine 6.18 9.29 2.85 7.26
Proline 12.57 7.74 10.97 –
Alanine 3.11 9.88 3.62 –
Valine 3.54 7.88 2.84 –
Methionine 6.25 8.45 12.39 –
Cysteine 8.36 11.10 14.74 –
Isoleucine 30.20 3.32 37.22 41.93
Leucine 4.74 11.12 4.94 –
Phenylalanine 228.98 330.24 425.62 322.75
Lysine 7.89 4.71 1.01 –
Tyrosine – – – –
Total AA 567.5 563.2 616.8 403.9
CONCLUSIONS
Peptides with strong antioxidant activity were produced by
hydrolysis of cottonseed protein with Neutrase. The CPH sample
was successfully separated into four fractions (I, II, III and IV) by gel
filtration column chromatography on Sephadex G-25. The results
revealed that fraction III (0.8–1.2 kDa) had the highest antioxidant
and free radical-scavenging activities. The amino acid composition
of the fractions was analysed and it was found that fraction III
contained higher levels of Phe, His, Pro, Met, Ile and Cys. This may
correlate with its strong antioxidant activity. The results suggested
that the antioxidant peptide fractions from cottonseed protein
might be useful as food additives and diet nutrients and also
pharmaceutically. However, further research is needed to isolate
the individual peptides responsible for the antioxidant activity of
CPH and to identify their amino acid sequences, which will allow
a better understanding of the peptide structure–functionality
relationship.
REFERENCES
1 Halliwell B and Gutteridge JMC, Free Radicals in Biology and Medicine.
Oxford University Press, Oxford, pp. 143–157 (1999).
2 DongS,ZengM,WangD,LiuZ,ZhaoYandYangH,Antioxidantand
biochemical properties of protein hydrolysates prepared from silver
carp (Hypophthalmichthys molitrix). Food Chem 107:1485–1493
(2008).
3 Kong X, Zhou H and Qian H, Enzymatic hydrolysis of wheat gluten by
proteases and properties of the resulting hydrolysates. Food Chem
102:759–763 (2007).
4 MoureA, Domínguez H and Parajo JC, Antioxidant properties of
ultrafiltration-recovered soy protein fractions from industrial
effluents and their hydrolysates. Process Biochem 41:447–456
(2006).
5 Wang JS, Zhao MM, Zhao QZ and Jiang YM, Antioxidant properties of
papain hydrolysates of wheat gluten in different oxidation systems.
Food Chem 101:1658–1663 (2007).
6 Yoshie-Stark Y, Wada Y and Wäsche A, Chemical composition,
functional properties, and bioactivities of rapeseed protein isolates.
Food Chem 107:32–39 (2008).
7 Ory RL and Flick GJ, Peanut and Cotton Seed Protein for Food Uses,
In new and developing sources of food proteins, ed. by Hudson BJF.
Chapman and Hall, London, pp. 195–240 (1994).
8 Tunc S and Duman O, Thermodynamic properties and moisture
adsorption isotherms of cottonseed protein isolate and different
forms of cottonseed samples. J Food Eng 81:133–143 (2007).
9 Alder-Nissen J, Enzymic Hydrolysis of Food Proteins. Elsevier Applied
Science, New York, NY, pp. 122–124 (1986).
10 Ohkawa H, Ohishi N and Yagi K, Assay for lipid peroxides in animal
tissues by thiobarbituric acid reaction. Anal Biochem 95:351–358
(1979).
11 Brand-Williams W, Cuvelier ME and Berset C, Use of a free radical
method to evaluate antioxidant activity. LWT – Food Sci Technol
28:25–30 (1995).
12 Halliwell B, Gutteridge JMC and Aruoma OI, The deoxyribose method:
a simple test tube assay for determination of rate constants for
hydroxyl radicals. Anal Biochem 165:215–219 (1987).
13 Marklund S and Marklund G, Involvement of the superoxide anion
radical in the autoxidation of pyrogallol and a convenient assay for
superoxide dismutase. Eur J Biochem 47:469–474 (1974).
14 Janero DR, Malondialdehyde and thiobarbituric acid reactivity as
diagnostic indices of lipid peroxidation and peroxidative tissue
injury. Free Radic Biol Med 9:515–540 (1990).
15 Peng X, Xiong YL and Kong B, Antioxidant activity of peptide fractions
from whey protein hydrolysates as measured by electron spin
resonance. Food Chem 113:196–201 (2009).
16 Li XX, Han LJ and Chen LJ, In vitro antioxidant activity of protein
hydrolysates prepared from corn gluten meal. J Sci Food Agric
88:1660–1666 (2008).
17 Li Y, Jiang B, Zhang T, Mu W and Liu J, Antioxidant and free radicalscavenging
activities of chickpea protein hydrolysate (CPH). Food
Chem 106:444–450 (2008).
J Sci Food Agric 2010; 90: 1855–1860 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1859

1860
18 Kim SY, Je JY and Kim SK, Purification and characterization of
antioxidant peptide from hoki (Johnius belengerii) frameprotein
by gastrointestinal digestion. JNutrBiochem18:31–38 (2007).
19 Chawla SP, Chander R and Sharma A, Antioxidant properties of
Maillard reaction products obtained by gamma-irradiation of whey
proteins. Food Chem 116:122–128 (2009).
20 Giménez B, Alemán A, Montero P and Gómez-Guillén MC, Antioxidant
and functional properties of gelatin hydrolysates obtained from
skin of sole and squid. Food Chem 114:976–983 (2009).
21 Kim SK, Kim YT, Byun HG, Nam KS, Joo DS and Shahidi F, Isolation and
characterization of antioxidative peptides from gelatin hydrolysate
of Alaska pollack skin. J Agric Food Chem 49:1984–1989 (2001).
www.soci.org D Gao, Y Cao, H Li
22 RajapakseN,MendisE,ByunHGandKimSK,Purificationandin vitro
antioxidative effects of giant squid muscle peptides on free radicalmediated
oxidative systems. JNutrBiochem16:562–569 (2005).
23 Li B, Chen F, Wang X, Ji B and Wu Y, Isolation and identification
of antioxidative peptides from porcine collagen hydrolysate by
consecutive chromatography and electrospray ionization-mass
spectrometry. Food Chem 102:1135–1143 (2007).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1855–1860

Research Article
Received: 18 February 2010 Revised: 25 March 2010 Accepted: 25 March 2010 Published online in Wiley Interscience: 19 May 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4025
Birds select conventional over organic wheat
when given free choice
Ailsa J McKenzie ∗ and Mark J Whittingham
Abstract
BACKGROUND: Global demand for organic produce is increasing by ¤4 billion annually. One key reason why consumers buy
organic food is because they consider it to be better for human and animal health. Reviews comparing organic and conventional
food have stated that organic food is preferred by birds and mammals in choice tests.
RESULTS: This study shows the opposite result – that captive birds in the laboratory and wild garden birds both consumed
more conventional than organic wheat when given free choice. There was a lag in preference formation during which time
birds learnt to distinguish between the two food types, which is likely to explain why the present results differ from those of
previous studies. A further experiment confirmed that, of 16 potential causal factors, detection by birds of consistently higher
levels of protein in conventional seeds (a common difference between many organic and conventional foodstuffs) is the likely
mechanism behind this pattern.
CONCLUSION: The results of this study suggest that the current dogma that organic food is preferred to conventional food may
not always be true, which is of considerable importance for consumer perceptions of organically grown food.
c○ 2010 Society of Chemical Industry
Supporting information may be found in the online version of this article.
Keywords: organic food; farmland; consumer perceptions; diet selection; protein selection
INTRODUCTION
Sales of organic food have increased exponentially over the last
decade and now account for between 2 and 3% of all food
purchased in Europe and the USA. International sales are now
estimated at ¤28.5 billion per annum, twice the figure reported
in 2000. 1 While perceptions about improved ‘quality’ and ‘safety’
of organic foods are widely held, 2 scientific evidence for the
superiority of organic produce is scarce. 3
One body of evidence that supports these perceptions is the
reported preference of a range of birds and mammals for organic
over conventional foodstuffs. While results of preference tests in
humans have been largely equivocal, several studies have found
both hens and rats to prefer organically grown beetroot and wheat
to that grown conventionally. 4–7
However, we believe the methodologies employed by the
majority of these studies to be unsuitable. No attempt has
typically been made to determine the proximate cause of the
observed preference. Preference for one particular food type over
another must be driven by some quantifiable difference between
the foods presented. Without investigation of such differences,
interpretation of results, and an assessment of their broader
application, is impossible. Determination of causation is also
important in trial design, as it will influence the time that test
animals require to make a correct selection. While differences in
physical properties such as taste or texture will tend to be identified
rapidly by test animals, preferences governed by differences that
require a post-ingestional response (e.g. nutritional differences
not associated with a flavour) will tend to take longer to
establish. Depending on the intensity of the difference detected,
post-ingestional responses can take up to several weeks to
manifest in a consistent choice. Turkeys, for example, were
found to take up to 3 weeks to show a consistent selection from
nutritionally limiting and non-limiting diets. 8 Previous preference
studies, on which the conclusions that organic food is preferred
to conventional food have been based, were carried out for a
maximum of 7 days. 4–7 This may simply not have been sufficient
timeforareliablechoicetobecomeestablished.
We carried out a series of four preference experiments using
a number of conventional and organic wheat samples. Organic
and conventional foodstuffs have been found to differ in many
ways, although not always consistently. They have been shown
to differ in their physical properties (e.g. size, shape, texture,
taste, odour), toxin and residue levels (e.g. fungi and mycotoxins,
plant secondary metabolites, pesticide residues, microbial toxins)
and nutrition levels (e.g. protein and nitrogen levels, amino
acids, certain vitamins and minerals). 9,10 Differences in many
of these properties have been shown previously to instigate
a selection response for one particular food over another (e.g.
mycotoxin deoxynivalenol (DON) level, 11,12 protein, 13,14 pesticide
residue 15,16 ). In accordance with this evidence, test foods were
analysed for the 16 properties deemed most likely (from the
∗ Correspondence to: Ailsa J McKenzie, School of Biology, Newcastle University,
Ridley Building, Claremont Road, Newcastle upon Tyne NE1 7RU, UK.
E-mail: a.j.mckenzie@ncl.ac.uk
School of Biology, Newcastle University, Ridley Building, Claremont Road,
NewcastleuponTyneNE17RU,UK
J Sci Food Agric 2010; 90: 1861–1869 www.soci.org c○ 2010 Society of Chemical Industry
1861

1862
Table 1. Justification for selection of the 16 compounds measured in wheat samples
www.soci.org AJ McKenzie, MJ Whittingham
Compound/chemical Reason for inclusion
Nutritional
Energy Selection for energy forms basis of optimal diet theory59,60 Protein Certain bird species shown to be able to select diets to optimise protein intake. 13,33 Levels shown to
vary between organic and conventional foods4,9,10 Amino acids Birds possess amino acid appetites, especially when foods are particularly deficient. 59,61 Amino acid
levels shown to differ between organic and conventional foods62 Moisture Shown to influence seed selection28 Fat Shown to influence seed selection28 Carbohydrate Shown to influence seed selection28 Contaminants
DON Detection of this mycotoxin in food shown to deter feeding in birds and mammals11,12 Microbes (E. coli, Salmonella spp.,
Unclear if they deter feeding, but organic produce may contain higher levels
Enterobacteriaceae)
3
Heavy metals Some suggestion that cadmium and lead levels vary between organic and conventional foods. 3 Birds
able to detect and avoid high levels in food34 Oxalic acid Used as a proxy for general plant secondary compounds which may be higher in organic foods and
deter feeding. Oxalic acid levels reflect stress in plants and thus likely to correlate with levels of
other stress chemicals39 Pesticide residues Shown to deter feeding by birds. 15,16 Levels typically differ between foods of the two regimes as a
result of input differences (i.e. lower in organic produce) 10,36
Physical
Thousand-seed weight (TSW) Seed size affects handling time, which ultimately influences food selection28 Hardness Texture shown to influence seed selection28 literature) to differ between foods from the two regimes and
to instigate a selection response (Table 1). This enabled us
to investigate mechanisms behind any observed preference
behaviour in the way not achieved previously.
MATERIALS AND METHODS
Test foods
Wheat of the variety ‘Alchemy’ was selected for use in the current
suite of experiments. Wheat is the most important agricultural
crop worldwide, with a production of 585 × 10 6 tons per annum, 17
and is known to be eaten by a wide range of farmland bird
species 18 (e.g. yellowhammer Emberiza citrinella, house sparrow
Passer domesticus,cornbuntingEmberiza calandra). In Europe it is
a particularly important food source during winter when granivorous
birds are known to select stubble fields (of which wheat is the
most common) in preference to other field types. 19–21 Therefore
the preference of organic and conventional wheat is of relevance
to the wider ecology of many declining farmland bird species.
Experiment 1. Laboratory canary experiment
One 50 kg sample each of commercially grown conventional and
organic wheat (variety ‘Alchemy’) was purchased from farmers in
the UK for use in this experiment. Product characteristics therefore
reflect what is produced and available to birds in the UK. Twelve
canaries were housed in individual cages (90 cm × 80 cm × 45 cm)
under a 12/12 h light/dark regime. Room temperature was kept
relatively constant, varying between 12 and 15 ◦ C. Birds were
settled for 10 days prior to the commencement of feeding trials.
They were provided with standard canary food when not taking
part in trials and water ad libitum. Prior to and during the trial
periods, birds were also given small quantities of a mix of the
organic and conventional wheat subsequently used in the trial
to become accustomed to consuming the test foods. During
trial periods the basic canary feed was removed from each cage
1 h prior to lights-off the night before the trial, trials beginning
approximately 1 h after lights-on the following morning. The order
in which trials were carried out was randomised daily between
birds. The trial involved birds being presented with organic and
conventional wheat in identical bowls on the cage floor. The seeds
were ground in a coffee grinder (for approximately equal amounts
oftime)andsieved(using2 mmsieves)toasimilarsize.Thecontent
and position of the bowls were randomised between birds but
remained the same for each bird throughout each set of trials, as
the length of learning period required for associations to be made
between food and position in the laboratory was unknown. Food
was presented for 20 min and the number of successful pecks
(seeds eaten rather than seeds handled) made during this time
was recorded using a click-counter (by an observer sitting quietly
at a standard distance of 2 m from each cage). Peck rate was used
in the experiment rather than weight consumed, as test birds
tended to knock seed from bowls onto the cage floor, making it
impossible to calculate total amounts ingested through weight
measurements. Trials were mostly carried out on consecutive days.
General linear mixed models (GLMM) were constructed for data
from the trials in R (Version 2.9.1) 22 using the ‘lmer’ function of the
package LME4. 23,24 Difference in peck number on the two grain
types was included as the response variable and bird as a random
effect. The key test was the significance of the difference in peck
number between the two grain types. This was determined by
fitting the null model (i.e. no predictors) whilst controlling for
the random effect ‘bird’. The intercept was tested for a difference
from zero, with a zero value indicating the same amount of food
eaten from both samples. T and P values for the significance of
the difference from zero are given in the output of the GLMM.
In addition to this, a maximum likelihood ratio test was run to
determine the effect of trial number on the magnitude of the
difference in peck rate.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1861–1869

Birds select conventional over organic wheat www.soci.org
Experiments 2 and 3. Garden bird experiments
Experiment 2 was carried out using three 50 kg samples of
conventional wheat supplied by different farmers (one of which
was the same as that used in Experiment 1) and one organic
sample (the same as was used in Experiment 1). Wheat samples
supplied by two different farmers were used in Experiment 3 (one
each of conventional and organic). All samples were of the variety
‘Alchemy’. Experiment 2 was carried out over 6 weeks during
January–March2008.Consumptionoforganicversusconventional
grain (of one of the three types) from plastic tube-style bird
feeders was measured every 2 days in 36 gardens in northeast
England. Wheat typically represents a very small proportion of
commercial foods presented to garden birds (CJ Wild Bird Foods,
personal communication), therefore the likelihood of a neophobic
response to one type of wheat over another was unlikely (i.e. birds
are unlikely to be already accustomed to either type of wheat).
Selected gardens were a minimum of 400 m apart to limit the
risk of the same birds visiting feeders in different gardens. Two
feeders (A and B) were placed close together (maximum 40 cm
apart) in each garden in positions matched for distance to cover,
to limit positional effects in feeder selection. Feeders were filled
from coded bags individual to each garden, which allowed all fills
to be carried out blind. The position of the feeders was swapped
at the beginning of week 4 to eliminate any positional effects.
Experiment 3 was carried out in the same way over 8 weeks from
January to March 2009 using 15 gardens.
DatawereanalysedasinExperiment1, withbothyearsmodelled
separately, using difference in consumption (conventional minus
organic) as the response variable and garden as a random
effect. Initial inspection of the data showed a likely interaction
between time (i.e. week) and the magnitude of the difference in
consumption. Therefore the above model was rerun with a time
component added as a predictor, with the difference between the
two models (with and without time) calculated to determine the
overall effect of time (a likelihood ratio test). The time component
used was not ‘week’ but ‘time since switch’ to account for the
impact of switching the feeders midway through the trial period.
Experiment 4. Protein experiment
New samples of ‘Alchemy’ wheat (i.e. not used in any previous
experiment) were obtained from Suffolk and Cambridge Crop
Station (SACCS), Great Wilbraham, UK. This wheat had been grown
in a single field under identical conditions but had received four
different fertiliser levels (calcium ammonium nitrate at 150, 175,
200 and 225 kg ha −1 ). This resulted in four conventionally grown
samples that varied in protein content but were similar in every
other way. The two samples with the lowest and the highest
protein levels were presented to the canaries in the trials (9.21 and
10.23 g per 100 g respectively). These protein levels were similar
to those detected in the wheat used in Experiments 1–3. All grain
was ground as in Experiment 1. Given the significance of time in
the establishment of preference for one grain type over another,
it was decided that a 10 day learning period should be carried out
prior to data collection with the two grain types. Therefore birds
were presented with the two types of grain in the same way as in
Experiment 1 for 10 days, with data collection beginning on day
11 and running for 5 days. It should be noted that only nine birds
were used in this experiment, as one died in the interval between
Experiments 1 and 2 and a further two became unwell. Statistical
analyses were carried out as in Experiments 1–3.
Physical and chemical analyses of test foods
Substantial literature searches were carried out before selection of
the compounds and chemicals for analysis in the wheat samples
(Table 1). As outlined in the ‘Introduction’, those selected are
the ones most likely to vary between organic and conventional
foodstuffs and to instigate a bird selection response (for full
justification see Table 1).
All samples used in Experiments 1–3 (two organic samples
and four conventional samples) were analysed for physical
properties (thousand-seed weight (TSW), hardness), nutritional
information (moisture, protein (N × 6.25), fat, carbohydrate, energy,
amino acids), toxin burden (DON, microbes), secondary
compounds (oxalic acid as a proxy) and pesticide residues.
The wheat used in Experiment 4 was analysed for protein
only. As these wheat samples were grown in the same field
under identical conditions other than nitrogen input, they
are unlikely to differ from one another in any significant
way. Justification of this decision is continued in the ‘Discussion’.
All samples were analysed using samples of equal weight.
Means and standard errors for each of these were calculated from
samples taken (minimum of four). Physical analysis was carried out
by NIAB, Cambridge, nutritional and heavy metal analysis by ILS
and Salamon and Seaber, London, pesticide screening by Eclipse
Scientific Group, Chatteris, mycotoxin analysis by Central Science
Laboratory (CSL), Sand Hutton, amino acid analysis by Protein and
Nucleic Acid Chemistry Facility (PNAC), University of Cambridge
and oxalic acid analysis by Intertek, Germany.
RESULTS
Experiment 1. Laboratory canary experiment
Across all 12 birds over all trials, 34% pecks were carried out on
organic and 66% on conventional grain. Overall, birds significantly
preferred conventional to organic wheat (T = 4.07, d.f. = 180,
P < 0.001). There was a significant temporal trend in the data,
with the significance of treatment increasing with trial number
(T = 11.90, d.f. = 1, P < 0.005) (Fig. 1). Mean preference decreased
during trials 5–7 before increasing at trial 8 and remaining at a
high level for the remainder of the trial period (Fig. 1). For the
first seven trials, birds made on average 69 (±23.4) more pecks
per 20 min trial on conventional than on organic wheat, while for
the remaining eight trials the difference was more than double
that – 155 (±22.8) more pecks made on conventional than on
organic seeds.
Experiments 2 and 3. Garden bird experiments
In Experiment 2, of 36 gardens, no feeding occurred in four, so
only data from the remaining 32 gardens could be included in the
analysis. Total depletion over the entire trial period was 58 954 g,
45% organic and 55% conventional wheat. More conventional
than organic grain was consumed in 24 out of 32 gardens overall.
Results of the GLMM showed a highly significant preference for
conventional over organic wheat across all sites (T = 3.41, d.f.
= 128, P < 0.001) (Fig. 2). The effect of time was not significant
overall (T = 2.68, d.f. = 1, P = 0.10); however, when the first and
last 3 weeks were modelled separately, time significantly affected
the difference during weeks 1–3 (T = 6.14, d.f. = 1, P < 0.05) but
not during weeks 4–6 (T = 0.57, d.f. = 1, P = 0.50) (Fig. 2). The
three different types of conventional wheat were not consumed
differently (relative to the same organic wheat) by wild birds
(χ 2 = 1.55, d.f. = 2, P = 0.46).
J Sci Food Agric 2010; 90: 1861–1869 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1863

1864
www.soci.org AJ McKenzie, MJ Whittingham
Figure 1. Experiment 1. Mean (± standard error) difference in the number of successful pecks made on the two types of wheat (conventional minus
organic) from trials 1–15 across 12 canaries. Successful pecks were those made where a piece of grain was consumed by the bird. Significantly more
conventional than organic wheat was consumed overall (T = 4.07, d.f. = 180, P < 0.001) and there was also a significant effect of trial on the magnitude
of the difference (T = 11.90, d.f. = 1, P < 0.005).
Figure 2. Experiment 2. Mean (± standard error) difference in consumption per week between conventional and organic wheat from feeders in 32
gardens in northeast England in February–March 2008 (all gardens were at least 400 m apart). Feeders were switched at the beginning of week 4.
Significantly more conventional than organic grain was consumed overall (T = 3.41, d.f. = 128, P < 0.001). In both periods (weeks 1–3 and weeks 4–6)
the preference for conventional grain increased with time (i.e. time since trial start and time since feeder position switch), but only in the first 3 weeks
was this significant (T = 6.14, d.f. = 1, P < 0.05).
In Experiment 3, total depletion over the 8 week period
was 41 508 g, 40% organic and 60% conventional wheat. More
conventional than organic grain was consumed in all 15 gardens
in the experiment. The GLMM again found a significant effect of
treatment on consumption (T = 2.25, d.f. = 83, P < 0.05) (Fig. 3).
There was no significant effect of time on preference overall
(T = 1.10, d.f. = 1, P = 0.29) or when the first and second periods
(i.e. before and after position switch) were considered separately
(weeks 1–4: T = 0.94, d.f. = 1, P = 0.33; weeks 5–8: T = 1.11, d.f.
= 1, P = 0.29). However, the parameter estimates do indicate a
tendency to consume more conventional wheat as the experiment
progressed.
Experiment 4. Protein experiment
Two birds were not used in the experiment as they became unwell
midway through the learning period. They later recovered but not
in time to be included in the experiment. The remaining birds
ate more of the higher- than the lower-protein wheat during the
trial days (mean difference 77 ± 17.6 pecks)(T = 2.58, d.f. = 41,
P < 0.05) (Fig. 4). Eight out of nine birds consumed more highthan
low-protein wheat overall, although some birds showed
stronger preferences than others.
Physical and chemical analyses of test foods
A summary of the analysis can be found in Table 2 (for full analysis
results see ‘Supporting information (SI)’). Of the 16 different
properties of the seeds measured (and considered a priori to have
some potential to influence selection), only protein content (g per
100 g) differed consistently between organic and conventional
samples, with levels being always higher in conventional than in
organic samples (Experiment 1: 26% higher (organic 8.96±0.04 vs
conventional 12.10 ± 0.05); Experiment 2: 6–26% higher (organic
8.96 ± 0.04 vs conventional 10.45 ± 0.37 (9.51 ± 0.08, 9.71 ± 0.35,
12.10 ± 0.05)); Experiment 3: 14% higher (organic 7.75 ± 0.11 vs
conventional 8.97 ± 0.07)) (Table 2).
Energy content (kJ per 100 g) was not always significantly
higher in conventional wheat seeds (Experiment 1: conventional
1476 ± 4.23 vs organic 1450 ± 2.72 (F1,6 = 39.29, P < 0.001);
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1861–1869

Birds select conventional over organic wheat www.soci.org
Figure 3. Experiment 3. Mean (± standard error) difference in consumption per week between conventional and organic wheat from feeders in 15
gardens in January–March 2009. Feeders were switched at the beginning of week 5. Significantly more conventional than organic grain was consumed
overall (T = 2.25, d.f. = 83, P < 0.05). There was no significant effect of time on the magnitude of the difference (T = 1.10, d.f. = 1, P = 0.29), although
the trend was in the same direction as in Experiment 2.
Figure 4. Experiment 4. Mean (± standard error) difference in peck number between the low- and high-protein wheat (high minus low) from trials 1–5
across nine canaries. Significantly more high- than low-protein wheat was consumed overall (T = 2.58, d.f. = 41, P < 0.05).
Experiment 2: conventional 1482 ± 3.61 vs organic 1450 ± 2.72
(F1,14 = 24.62, P < 0.001); Experiment 3: conventional 1483±2.58
vs organic 1493 ± 1.22 (F1,6 = 39.29, P < 0.01)) (Table 2).
No consistency was found in levels of the mycotoxin DON
(µgkg −1 ) between treatments across experiments (Experiment
1: conventional 46.0 ± 21.5 vs organic 201 ± 59.9 (F1,6 = 5.98,
P = 0.05); Experiment 2: conventional 323 ± 110 vs organic
201 ± 59.9(F1,14 = 0.20, P = 0.66 n.s.)) (Table 2).
TSW of the organic grain used in Experiments 1 and 2 was
statistically lower than that of any of the conventional grain
samples (Experiment 1: 10.3% less (F1,6 = 319.4, P < 0.001);
Experiment 2: 11.1% less (F1,14 = 345.9, P < 0.001)) (Table 2).
Levels of moisture, fat, carbohydrate, pesticide residues, cadmium,
lead, microbial contamination (Escherichia coli, Salmonella
spp., Enterobacteriaceae), oxalic acid (as a proxy for plant secondary
compounds), hardness (endosperm percentage crude
protein) and amino acid content were not significantly different
between samples, did differ but not consistently between
experiments (see ‘SI’) or, in the case of pesticide residues, were
operating in the opposite direction to that predicted (see ‘SI’).
DISCUSSION
We show that, across a range of experiments in two different
study systems, birds preferred conventionally to organically grown
wheat. This finding is novel and differs from all previous preference
studies with organic and conventional foods. 4–7 As such, it is likely
to be of considerable interest to a range of scientists and shed new
light on consumer perceptions of organic food. 2
We believe our findings may differ from those of previous
studies because of methodological issues – previous studies may
not have been carried out in a way that allowed true preferences
to become fully realised. In the current study, preference for the
conventional food presented was only consistent after a delay
period (around seven trials in Experiment 1 (Fig. 1) and 1 week
in Experiments 2 and 3 (Figs 2 and 3)). This delay is evidence of
a learning interval, something which may have been precluded
by the short trial duration employed in previous studies. If this
interval forms a significant part of the food testing period, then
the results of the test may be compromised. 25 Previous feeding
studies (which have all been carried out in the laboratory) have
been run for a maximum of 7 days, typically 5 days or less, normally
J Sci Food Agric 2010; 90: 1861–1869 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1865

1866
www.soci.org AJ McKenzie, MJ Whittingham
Table 2. Mean values (± standard error; n = 4 for each sample) and test statistics of four of the main potential drivers of selection from the organic
and conventional wheat used in Experiments 1–3
Exp. Wheat Protein (g per 100 g) Energy (kJ per 100 g) DON (µgkg −1 ) TSW(g)
1 Org. 8.96 ± 0.04 F1,6 = 2345 1450 ± 2.72 F1,6 = 39.29 201 ± 59.9 F1,6 = 5.98 40.90 ± 0.14 F1,6 = 319.4
Conv. 12.10±0.05 P < 0.001 1476±4.23 P < 0.001 46.0 ± 21.5 P = 0.05 45.58±0.22 P < 0.001
2 Org. 8.96 ± 0.04 F1,14 = 5.05 1450 ± 2.72 F1,14 = 24.62 201 ± 59.9 F1,14 = 0.20 40.90 ± 0.14 F1,14 = 345.9
Conv. 10.45±0.37 P < 0.05 1482±3.61 P < 0.001 323 ± 110 P = 0.66 46.03±0.15 P < 0.001
3 Org. 7.75±0.11 F1,6 = 85.00 1493±1.22 F1,6 = 39.29 Not tested Not tested
Conv. 8.97 ± 0.07 P < 0.001 1483 ± 2.58 P < 0.01
with no learning period prior to trials. Therefore these studies may
simply not have allowed sufficient time for true preferences to
become established.
In order for learning to occur, the ingestion of a certain food must
instigate some sort of metabolic response that can be monitored
and learned. 13 In birds, this effect must be associated with some
visual cue that acts to differentiate between foodstuffs. 13 In the
current study the visual cue is most likely to be bowl or feeder
position, as switching positions was shown to temporarily disrupt
food selection preferences in Experiments 2 and 3 (positions were
not switched in Experiment 1).
Analysis of the wheat used in Experiments 1–3 (based on
samples of wheat of equal weight) has allowed us to discount a
large number of factors as responsible for the preference observed
(for summary of results see Table 2; for full details see ‘SI’). We
consider each of these in turn below.
Differences in physical properties – size, hardness and taste
Granivorous birds have been found to select seeds in accordance
with a range of variables. 26 Many of these relate to the physical
properties of seeds such as size, texture and hardness, taste and
speed of processing. 27,28 While conventional wheat grains in the
current experiment were typically larger in terms of TSW than
organic grains, the grinding and sieving of particles to equal
size in Experiment 1 eliminates size as a driver of preference for
conventional over organic food in the current experiment. Food
texture (measured as endosperm hardness) was not found to differ
between samples, making differences in this component unlikely
to be driving the preference for conventional grain.
Birds typically possess relatively few taste buds, 29,30 the majority
of which are located on the posterior tongue and pharyngeal floor
rather than in the front part of the mouth as in mammals. 31
Therefore they have not historically been credited with a highly
developed sense of taste. However, more recent studies have
shown certain species of bird to respond to a number of taste
stimuli, particularly bitter tastes in foodstuffs and solutions. 31,32
However, the marked delay in formation of preference observed
in the present study is more indicative of a post-ingestive effect
of food consumption than of a simple taste preference, where
the response would likely be much more rapid. Taste is unlikely,
therefore, to be driving the preference for conventional wheat in
the current experiment.
Toxin and contaminant differences
Aversive toxins and contaminants can deter feeding by interfering
with nutrient utilisation (e.g. proteinase inhibitors, tannins)
or by giving an unpleasant sensation. 33 As discussed in the
‘Introduction’, there is the potential for four main classes of toxins
and/or contaminants to be more abundant in organic produce and
thus be driving the preference for conventional wheat observed
in the current study. Levels of the relevant compounds have been
measured and their likely role can now be assessed. It is important
to note that none of the previous organic versus conventional food
preference studies made any toxin/contaminant assessments.
Environmental contamination
Birds have been shown to be deterred by high levels of
environmental contaminants such as heavy metals in foods. 34
While differences in levels of such contamination have been
reported in organic versus conventional foods, these differences
can largely be explained by variations in environmental factors
such as location, crop type and soil pH rather than by overall
farming regime. 3 Two potential exceptions to this are cadmium
and lead, shown to accumulate in lower and higher levels in
organic produce respectively. 35 Chickens have been shown to
detect these metals in foodstuffs, but only at very high levels and
not at those typical of polluted regions. 34
Cadmium and lead were not found at detectable levels in the
samples used in the current study. Environmental contamination
of this sort can therefore be discounted as a possible driving factor
in conventional preference.
Microbial contamination
Magkos et al. 3 stated that the use of manures rather than
chemical fertilisers contributes to an increased risk of microbial
contamination in organic food. While manures are also used widely
in conventional farming, this regime allows decontamination
of the manure using irradiation and synthetic disinfectants
to reduce microbial contamination, processes prohibited in
organic farming. 3 However, the bulk of available evidence from
comparative studies shows no significant differences in the
bacterial status of organically and conventionally grown cereals
(wheat and rye) and vegetables (carrots, spring mix, Swiss chard,
salad vegetables). 3,36
Neither E. coli nor Salmonella spp. were found in any of the
samples used in the current study. Enterobacteriaceae were
detected in small quantities (

Birds select conventional over organic wheat www.soci.org
or water. One theory suggests that exclusion of the majority of
artificial inputs in organic farming makes it an inherently more
‘stressful’ regime than conventional farming and that organic
foods will contain more of these secondary compounds as a
consequence. While a number of these compounds have been
shown to benefit animal and human health (e.g. polyphenolics
as antioxidants), some (e.g. alkaloids and tannin-based phenolics)
are potentially harmful to herbivores either through direct toxicity
or through interference with nutrient assimilation. 28,37
Increased interest in the antioxidant abilities of natural plant
foods has led to a proliferation of studies concerning the levels
of secondary metabolites in organic versus conventional foods.
While results have been mixed, there appears to be a slight
trend for certain organic foodstuffs to contain higher levels
of these chemicals than conventional foodstuffs. However, this
may not be the case for cereals. Studies of organic versus
conventional wheat, for example, have not tended to detect
increased levels of phenolics or stress-related metabolites in
that grown organically. 17,38,39 Similarly, no difference was found
between the phenolic content of organic and conventional oats. 40
Added to this the fact that commercial cereals tend to be very
low in defence chemicals such as tannins regardless of regime, 28
differences in secondarymetabolite levels are unlikelyto be driving
the preference for conventional food in the current study.
Oxalic levels were used as a proxy for secondary chemical levels
in the current experiment. While it would have been useful to carry
out a series of tests on a range of stress markers and secondary
chemicals (e.g. phytic acid, phenols, total antioxidant capacity),
suitable methodologies could not be found. Oxalic acid levels have
been shown to reflect stress levels in plants 39 and are, as such, likely
to correlate with levels of other stress chemicals. No difference in
oxalate level was detected between the organic and conventional
samples used in the current experiment. Similar results have been
found by other comparison studies, Langenkämper et al., 39 for
example, finding no difference between organic and conventional
wheat samples in terms of free radical-scavenging capacity, oxalic
or phytic acid content.
Pesticide residues
Pesticide residues in foods have been found to deter feeding by
a range of bird species. 15,16 However, as conventional wheat
was preferred to organic wheat in the current study, either
chemical residues in conventional food are too small to induce
post-ingestional effects or the improved nutritional quality (i.e.
increased protein level) of conventional food is such that it
overrides the risk to health from other compounds.
Mycotoxins
Mycotoxins are secondary metabolites produced by fungi in a
similar way to defensive phytochemicals produced by vascular
plants. 41 One of the most commonly detected mycotoxins in
temperate areas, found extensively in cereal crops and animal
feeds, is DON, a mycotoxin produced by Fusarium spp. fungi. 42,43
IngestionofDONcanhaveanegativeaffectonthehealthofboth
mammals and birds and, as a result, is often avoided in foods. Pigs,
for example, reduced intake of food contaminated with naturally
occurring DON even at relatively low levels (Around 1 mg kg −1 ). 11
Similar observations have been made with rats and poultry, 12
although major effects tend only to occur at concentrations
higher than those observed with pigs (>10 mg kg −1 for poultry
and >2mgkg −1 for rats/mice). 12,44,45
There is much debate over whether the organic farming regime
results in increased or decreased levels of DON in cereals. As
organic crops do not receive any chemical antifungal treatments,
it is predicted that they will be more susceptible to fungal attack
and harbour higher levels of DON and other fungus-based toxins
than conventional crops. Indeed, this has been found in a number
of studies. 46,47 However, the converse may also be true. The high
nitrogen inputs associated with conventional farming tend to
weaken plant cell walls, increasing susceptibility to fungal attack
and subsequent infection by DON. 3 Moreover, the increased tillage
carried out on organic farms has been shown to decrease the
incidence fungal attack. 3 Other authors have found no difference
in DON levels between samples from the two regimes. 4,48–50
While levels of DON in the wheat samples used in the current
experiment were extremely varied, they did not vary in a manner
compatible with DON driving preference for conventional wheat.
That is, DON content of organic wheat was not consistently
higher – one conventional sample used in Experiment 2, for
example, had a DON level around four times that of the paired
organic sample and remained preferred in the paired test. It is
therefore unlikely that avoidance of high levels of DON in organic
grain is driving the preference in the current experiment.
Nutritional differences
Birds have been shown to display appetites for both macronutrients
(e.g. protein, fat) and micronutrients (e.g. vitamins, specific
amino acids). Domestic chickens, for example, have been shown
to be highly adept at selecting an optimal diet from a range of
foodstuffs and to have specific appetites for individual dietary
components. 13,25,33,51 It is possible, therefore, that levels of one or
more important nutrients are enhanced in the conventional grain
and are driving its preference in the current study.
The most likely candidate in the current experiment is protein. In
terms of essential nutrients, only protein levels were consistently
different between foodstuffs – the conventional wheat samples
contained significantly and consistently more protein than the
organic wheat samples (Table 2). This is a common finding for
a wide range of organic versus conventional foods 9,10 (including
wheat 39 ) and is a consequence of differences in the levels of
applied nitrogen fertiliser in each regime. 52,53 Applications of
nitrogen on organic crops tend to be around 30% lower than
those on conventional crops, 4 with availability of nitrogen to
organic crops reduced further by the tendency of natural fertilisers
(i.e. manures) to release nutrients more slowly than chemical
fertilisers. 54 Thus, although our experiments are based on one
food type (wheat), they may potentially generalise across to many
other food types. Further work is required to confirm this.
The results of Experiment 4 support the idea that protein
is important in determining the choice of conventional wheat
by birds, with birds selecting the higher-protein wheat from an
otherwise matching pair (Fig. 4). One weakness of Experiment
4 is that (owing to resource constraints) we did not measure
components of the wheat samples other than protein. However,
the results of the analysis of samples used in Experiments 1–3
had allowed us, by Experiment 4, to discount the importance of
the majority of other potential causal factors (e.g. particle size
ruled out in Experiment 1, DON ruled out in Experiment 2, etc.).
Therefore the selection being made by birds in Experiment 4 is
likely to result from the higher protein content of the selected
food rather than any other causal factor.
Protein is an essential nutrient in the diet of all birds and
mammals and is often limiting. This is especially the case for
J Sci Food Agric 2010; 90: 1861–1869 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1867

1868
animals that rely mainly on plant-based foodstuffs such as seeds,
which typically contain very low levels of protein. 55 Perhaps as a
result of this, birds and mammals have been shown to display very
specific protein appetites – that is, be capable of selecting foods
for their protein content alone. 13,14,25,33,56,57 Such preferences
tend to be stronger when animals are subjected to some kind of
stress (temperature, food/water shortage, etc.). 28 In birds, most
protein detection studies have focused on commercial species
such as chickens and quail, leaving this ability in the majority
of small passerines largely understudied. Moreover, studies that
have been carried out with passerines have mainly concerned
detection of extreme protein differences rather than detection of
differences of the magnitude observed in the current study (e.g.
70% difference 58 vs 6–26% in the current study). Therefore the
results of Experiment 4 are particularly noteworthy.
As discussed previously, this preference for the higher-protein
food of a pair is not in itself a novel finding. The novelty of our work
is that differences in protein between conventional and organic
food may explain the observed selection of the former, rather
than the range of other differences between the two food types
(discussed above).
As this work was not carried out in a closed system (i.e. the birds
were free to consume food ad libitum outwith the observation
period), it is impossible to make any definitive statement on
the mechanism behind selection, but our results do suggest a
mechanism based more on nutrient selection rather than toxin
or physical property aversion. Clearly there is a need to carry out
further work in other taxa with similar protocols to those employed
here to confirm the generality of our findings.
Our work provides evidence that the preference reported by
other studies for organic over conventional foods is questionable
and that, at least in our series of experiments, the preference is
actually reversed when other factors (such as length of study and
nutritional content of foods) are considered. Our findings are likely
to be of considerable interest to the wider scientific community
and to the general public in the debate over the relative merits of
consuming organic food.
SUPPORTING INFORMATION
Supporting information may be found in the online version of this
article.
ACKNOWLEDGEMENTS
MJW was funded by a BBSRC David Phillips Fellowship. AJM
was funded by an NERC studentship and by CASE funding from
the British Trust for Ornithology. We would like to thank the
following for their input into this project: Tom Gray, Lowell Mills,
Sarah McElhatton, Claudia Garratt, Nick Burton, Graeme Ruxton,
Kirsten Brandt, Candy Rowe, Juliet Vickery, Anna McKenzie, Jeroen
Minderman, and all the house-owners who allowed access to their
gardens.
REFERENCES
1 Willer H, Yussefi-Menzler M and Sorensen N (eds), The world of organic
agriculture. Statistics and emerging trends 2008. Report for IFOAM,
Bonn, Germany and FiBL, Frick, Switzerland (2008).
2 Makatouni A, What motivates consumers to buy organic food in the
UK? Br Food J 104:345–352 (2002).
3 Magkos F, Arvaniti F and Zampelas A, Organic food: buying more
safety or just peace of mind? A critical review of the literature. Crit
Rev Food Sci 46:23–56 (2006).
www.soci.org AJ McKenzie, MJ Whittingham
4 Mäder P, Hahn D, Dubois D, Gunst L, Alfoldi T, Bergmann H, et al,
Wheat quality in organic and conventional farming: results of a
21 year field experiment. J Sci Food Agric 87:1826–1835 (2007).
5 Mäder P, Pfiffner L, Niggli U, Balzar U, Balzar F, Plochberger K, et al,
Effect of three farming systems (bio-dynamic, bio-organic,
conventional) on yield and quality of beetroot (Beta vulgaris L.
var. esculenta L.)inasevenyearcroprotation.Acta Hort 330:10–31
(1993).
6 Plochberger K and Velmirov A, Are food preference tests with
laboratory rats a proper method for evaluating nutritional quality.
Biol Agric Hort 8:221–233 (1992).
7 Plochberger K, Feeding experiments – a criterion for quality
estimation of biologically and conventionally produced foods. Agric
Ecosyst Environ 27:419–428 (1989).
8 Leeson S and Summers JD, Metabolism and nutrition – dietary selfselection
by turkeys. Poultry Sci 57:1579–1585 (1978).
9 Benbrook C, Zhao X, Yáñez J, Davies N and Andrews P (eds), New
evidence confirms the nutritional superiority of plant-based organic
foods. State of Science Report, The Organic Center (2008).
10 Woese K, Lange D, Boess C and Werner Bögl K, A comparison of
organically and conventionally grown foods – results of a review of
the relevant literature. J Sci Food Agric 74:281–293 (1997).
11 Tiemann U and Danicke S, In vivo and in vitro effects of the mycotoxins
zearalenone and deoxynivalenol on different non-reproductive and
reproductive organs in female pigs: a review. Food Addit Contam
24:306–314 (2007).
12 Awad WA, Bohm J, Razzazi-Fazeli E and Zentek J, Effects of feeding
deoxynivalenol contaminated wheat on growth performance,
organ weights and histological parameters of the intestine of
broiler chickens. J Anim Physiol Anim Nutr 90:32–37 (2006).
13 Forbes JM and Shariatmadari F, Diet selection for protein by poultry.
World Poultry Sci J 50:7–24 (1994).
14 Felton AM, Felton A, Lindenmayer DB and Foley WJ, Nutritional goals
of wild primates. Funct Ecol 23:70–78 (2009).
15 McKay HV, Prosser PJ, Hart ADM, Langton SD, Jones A, McCoy C, et al,
Do wood-pigeons avoid pesticide treated cereal seed? J Appl Ecol
36:283–296 (1999).
16 Bennett RS and Prince HH, Influence of agricultural pesticides on food
preference and consumption by ring-necked pheasants. J Wildl
Manag 45:74–82 (1981).
17 Zörb C, Langenkämper G, Betsche T, Niehaus K and Barsch A,
Metabolite profiling of wheat grains (Tritium aestivum L.)
from organic and conventional agriculture. J Agric Food Chem
54:8301–8306 (2006).
18 Wilson JD, Morris AJ, Arroyo BE, Clark SC and Bradbury RB, A review
of the abundance and diversity of invertebrate and plant foods
of granivorous birds in northern Europe in relation to agricultural
change. Agric Ecosyst Environ 73:13–30 (1999).
19 Gillings S, Newson SE, Noble DG and Vickery JA, Winter availability
of cereal stubbles attracts declining farmland birds and positively
influences breeding population trends. Proc Biol Sci 272:733–739
(2005).
20 Moorcroft D, Whittingham MJ, Bradbury RB and Wilson WD, The
selection of stubble fields by wintering granivorous birds reflects
vegetation cover and food abundance. J Appl Ecol 39:535–547
(2002).
21 Wilson JD, Taylor R and Muirhead LB, Field use by farmland birds
in winter: an analysis of field type preferences using resampling
methods. Bird Study 43:320–332 (1996).
22 R Core Development Team, R: a Language and Environment
for Statistical Computing. [Online]. R Foundation for Statistical
Computing, Vienna (2009). Available:http://www.R-project.org [13
May 2010].
23 Bates D, Maechler M and Dai B, LME4: Linear Mixed-effects Models
Using S4 Classes. R Package Version 0.999375–28. [Online]. (2008).
Available:http://lme4.r-forge.r-project.org/ [13 May 2010].
24 Crawley MJ, The R Book (1st edn). Wiley, Chichester (2007).
25 Rose SP and Kyriazakis I, Diet selection of pigs and poultry. Proc Nutr
Soc 50:87–98 (1991).
26 Kelrick MI, MacMahon JA, Parmenter RR and Sisson DV, Native seed
preferences of shrub-steppe rodents, birds and ants: the
relationships of seed attributes and seed use.Oecologia 68:327–337
(1986).
27 Frazer KI and McWilliams SR, Determinants of dietary preference in
yellow-rumped warblers. Wilson Bull 114:243–248 (2002).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1861–1869

Birds select conventional over organic wheat www.soci.org
28 Diaz M, Food choice by seed-eating birds in relation to seed chemistry.
Comp Biochem Physiol A 113:239–246 (1996).
29 Duncan CJ, Preference tests and the sense of taste in the pigeon
(Columba livia Var. Gmelin). Anim Behav 8:54–60 (1960).
30 Duncan CJ, The sense of taste in birds. Ann Appl Biol 48:409–414
(1960).
31 Werner SJ, Kimball BA and Provenza FD, Food color, flavor, and
conditioned avoidance among red-winged blackbirds. Physiol
Behav 93:110–117 (2008).
32 Rowe C and Skelhorn J, Colour biases are a question of taste. Anim
Behav 69:587–594 (2005).
33 Murphy ME, Dietary complementation by wild birds: considerations
for field studies. JBiosci19:355–368 (1994).
34 Phillips CJC and Strojan ST, The ability of chickens to select nutritive
and avoid toxic concentrations of heavy metals in feeds. JSustain
Agric 30:31–45 (2007).
35 Rossi F, Bertuzzi T, Comizzoli S, Turconi G, Roggi C, Pagani M, et al,
Preliminary survey on composition and quality of conventional and
organic wheat. Ital J Food Sci 4:355–366 (2006).
36 Bourn D and Prescott J, A comparison of the nutritional value,
sensory qualities, and food safety of organically and conventionally
produced foods. Crit Rev Food Sci 42:1–34 (2002).
37 Winter CK and Davis SF, Organic foods. J Food Sci 71:R117–R124
(2006).
38 ZuchowskiJ,KapustaI,SzajwajB,JonczykKandOleszekW,Phenolic
acid content of organic and conventionally grown winter wheat.
Cereal Res Commun 37:189–197 (2009).
39 Langenkämper G, Zörb C, Seifert M, Mäder P, Fretzdorff B and
Betsche T, Nutritional quality of organic and conventional wheat. J
Appl Bot Food Qual 80:150–154 (2006).
40 Dimberg LH, Gissen C and Nilsson J, Phenolic compounds in oat grains
(Avena sativa L.) grown in conventional and organic systems. Ambio
34:331–337 (2005).
41 Yiannikouris A and Jouany JP, Mycotoxins in feeds and their fate in
animals: a review. Anim Res 51:81–99 (2002).
42 D’Mello JPF, Placinta CM and Macdonald AMC, Fusarium mycotoxins:
a review of global implications for animal health, welfare and
productivity. Anim Feed Sci Technol 80:183–205 (1999).
43 Placinta CM, D’Mello JPF and Macdonald AMC, A review of worldwide
contamination of cereal grains and animal feed with Fusarium
mycotoxins. Anim Feed Sci Technol 78:21–37 (1999).
44 Sprando RL, Collins TFX, Black TN, Olejnik N, Rorie JI and Eppley RM,
Characterization of the effect of deoxynivalenol on selected male
reproductive endpoints. Food Chem Toxicol 43:623–635 (2005).
45 Khera KS, Arnold DL, Whalen C, Angers G and Scott PM, Vomitoxin (4deoxynivalenol)
– effects on reproduction of mice and rats. Toxicol
Appl Pharmacol 74:345–356 (1984).
46 Malmauret L, Parent-Massin D, Hardy JL and Verger P, Contaminants
in organic and conventional foodstuffs in France. FoodAdditContam
19:524–532 (2002).
47 Marx H, Gedek B and Kollarczik B, Comparative investigations on
mycotoxicological status of alternatively and conventionally grown
crops. Z Lebensm Unters Forsch 201:83–86 (1995).
48 Edwards SG, Fusarium mycotoxin content of UK organic and
conventional wheat. Food Addit Contam A 26:496–506 (2009).
49 Hoogenboom LAP, Bokhorst JG, Northolt MD, Van de Vijver LPL,
Broex NJG, Mevius DJ, et al, Contaminants and microorganisms in
Dutch organic food products: a comparison with conventional
products. Food Addit Contam A 25:1195–1207 (2008).
50 Vanova M, Klem K, Misa P, Matusinsky P, Hajslova J and Lancova K, The
content of Fusarium mycotoxins, grain yield and quality of winter
wheat cultivars under organic and conventional cropping systems.
Plant Soil Environ 54:395–402 (2008).
51 Newman RK and Sands DC, Dietary selection for lysine by the chick.
Physiol Behav 31:13–19 (1984).
52 Weston LA and Barth MM, Preharvest factors affecting postharvest
quality of vegetables. HortScience 32:812–816 (1997).
53 Shier NW, Kelman J and Dunson JW, A comparison of crude protein,
moisture, ash and crop yield between organic and conventionally
grown wheat. Nutr Rep Int 30:71–76 (1984).
54 Berry PM, Sylvester-Bradley R, Philipps L, Hatch DJ, Cuttle SP,
Rayns FW, et al, Is the productivity of organic farms restricted by the
supply of available nitrogen? Soil Use Manag 18:248–255 (2002).
55 Murphy ME and Olson DB, Control of daily protein-intake by whitecrowned
sparrows (WCS). Am Zool 31:A116–A116 (1991).
56 White TCR, The importance of a relative shortage of food in animal
ecology. Oecologia 33:71–86 (1978).
57 Shariatmadari F and Forbes JM, Growth and food intake responses
to diets of different protein contents and a choice between diets
containing two concentrations of protein in broiler and layer strains
of chickens. Br Poultry Sci 34:959–970 (1993).
58 Musten B, Peace D and Anderson DH, Food-intake regulation in
weanling rat – self-selection of protein and energy. J Nutr
104:563–572 (1974).
59 Schoener TW, Theory of feeding strategies. Annu Rev Ecol Syst
2:369–404 (1971).
60 Emlen JM, The role of time and energy in food preferences. Am Nat
100:611–617 (1966).
61 Murphy ME and Pearcy SD, Dietary amino acid complementation as a
foraging strategy for wild birds. Physiol Behav 53:689–698 (1993).
62 Wolfson JL and Shearer G, Amino acid composition of grain protein of
maize grown with and without pesticides and standard commercial
fertilizers. Agron J 73:611–613 (1981).
J Sci Food Agric 2010; 90: 1861–1869 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1869

1870
Research Article
Received: 17 March 2010 Revised: 29 April 2010 Accepted: 30 April 2010 Published online in Wiley Interscience: 3 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4027
Determination of maize kernel hardness:
comparison of different laboratory tests
to predict dry-milling performance
Massimo Blandino, a∗ Mattia Ciro Mancini, a Alessandro Peila, a Luca Rolle, b
Francesca Vanara a and Amedeo Reyneri a
Abstract
BACKGROUND: Numerous foods are produced from maize, and grain hardness has been described to have an impact on grain
end-use value, and in particular for dry-milling performance.
RESULTS: Thirty-three samples of commercial hybrids have been analysed for test weight (TW), thousand-kernel weight
(TKW), hard : soft endosperm ratio (H/S), milling time (MT) and total milling energy (TME) through the Stenvert hardness test,
coarse : fine material ratio (C/F), break force (HF) and break energy (HWF) through the puncture test, floating test (FLT), kernel
dimensions and sphericity (S), protein (PC), starch (SC), lipid (LC), ash (AC) content and amylose : amylopectin ratio (AS/AP).
Total grit yield (TGY) has been obtained through a micromilling procedure and used to compare the efficiency of the tests to
predict the dry-milling performance. TW, H/S, MT, TME, C/F, FLT, S, PC, SC and AS/AP were significantly correlated with each
other. TW has been confirmed to be a simple estimator of grain hardness. Among the hardness tests, C/F was shown to be the
best descriptor of maize milling ability, followed by FLT. A good correlation with TGY has also been observed with H/S, MT, TME
and PC, while SC, S and AS/AP seem to play a minor role. The puncture test (HF and HWF) did not offer good indications on the
impact of hardness on kernel grinding properties.
CONCLUSION: This study can be considered as a contribution towards determining kernel properties which influence maize
hardness measurement in relation to the end-use processing performance.
c○ 2010 Society of Chemical Industry
Keywords: maize kernel; dry-milling; hardness; endosperm; texture analysis
INTRODUCTION
Grain hardness is an important grain quality attribute that plays a
role in the processing of cereal grains and in the end-use quality
of cereal grain products. 1 Maize (Zea mays L.) is dry-milled to
produce a range of flours and grits which are further processed
for snacks, breakfast cereals and cooked or extruded products. 2,3
Maize hardness has been shown to have a remarkable influence
on the efficiency of the extraction yield and quality of the final
product. 4 Maize for dry-milling and alkaline cooking processes
should be hard, with large kernels and with pericarps and germs
that are easy to remove during the process. 5,6 On the other hand,
wet millers prefer soft maize grain, which usually requires less
steeping and leads to a better starch–protein separation. 7
The physical and biochemical aspects of maize hardness
have been described in numerous publications. As far as the
biochemical contribution to hardness in maize is concerned, both
the protein and starch compositions have been associated with
maize hardness. 8 Although the protein content comprises a lower
proportion of the total kernel composition compared to starch,
it would appear that it plays a significant role in influencing
hardness, 3 and the variation in zein classes has, in particular,
been linked to differences in hardness. 9 On the other hand, some
studies 10 have not shown any link between grain protein content
(PC) and hardness. Fox and Manley 11 suggest that the type of
hardness test adopted may be influenced by protein, and some
tests could be more influenced by the endosperm structure,
thereby giving a stronger correlation with the protein content. At
present very few studies have carried out multiple hardness tests
and linked the result to PC.
Among the physical tests, the ratio of the cross-sectional area
from the hard to the soft (H/S ratio) endosperm, which can be
measured with different techniques, 10 is probably the most direct
way of measuring the fraction of kernel which influences drymilling
processing performance to the greatest extent, although
this method is not practical and is time consuming. 12 The other
methods used to assess whole grain hardness are empirical and
give an indirect measurement of the hardness that is generally
∗ Correspondence to: Massimo Blandino, Department of Agriculture, Forestry
and Land Management, University of Turin, via Leonardo da Vinci 44, 10095
Grugliasco (TO), Italy. E-mail: massimo.blandino@unito.it
a DepartmentofAgriculture,ForestryandLandManagement,UniversityofTurin,
10095 Grugliasco (TO), Italy
b Dipartimento di Valorizzazione e Protezione delle Risorse Agroforestali – Food
Technology sector, University of Turin, 10095 Grugliasco (TO), Italy
J Sci Food Agric 2010; 90: 1870–1878 www.soci.org c○ 2010 Society of Chemical Industry

Hardness methods for testing maize kernel www.soci.org
correlated with H/S. The maize physical characteristics, kernel
size and shape, weight and density, resistance to grinding or
to abrasion and quantification of coarse and fine material after
grinding and sieving have all been linked to hardness and its
subsequent effects on processing. 11 Other indirect available tests
toestimatemaizehardnessarebasedontheviscosityoftheground
material (Rapid ViscoAnalyser) 13 or on near-infrared reflectance
(NIR) and transmittance (NIT), both of which use whole kernels or
kernels after the grinding step. 14 Most of these methods provide
variableinformationontherangeofhardnessfromamaizesample.
Moreover, in spite of the importance of hardness in dry-milling and
the number of studies that have been published on this subject,
there is still no generally accepted standard for the evaluation of
maize kernel hardness and there is a need to evaluate new simple,
rapid and reliable tests that could relate maize quality to product
yields. 15
At present there is little data concerning a single-kernel testing
methodology for maize, whereas for wheat and barley the singlekernel
characterization system (SKCS) has been shown to be
suitable to determine hardness and provide an indication of
quality. 16 One of the few methods that has the potential for
measuring single maize kernels, similar to the SKCS for wheat,
is the compression or puncture test. This test, already used for
wheat 17 and other cereals, 18 relies on a resistance measurement
of single kernels and involves a rod being pressed into the kernels.
Shandera and Jackson 19 have used a single-kernel puncture
texture analysis to discover the components and associative forces
that are responsible for the endosperm structure of maize kernels,
and Gaytán Martínezet al. 20 have studied the hardness of 21 maize
cultivars in relation to texture, floating test, size and arrangement
of starch granules within the endosperm.
The objectives of this study were: (1) to increase the understanding
of maize quality factor correlations and their relationship
with the yield of dry-milling products; (2) to evaluate the range
of variation of maize grain hardness that occurs in commercial
maize hybrids that are normally cultivated in northern Italy; (3) to
compare the parameters obtained from the puncture test with
other standard tests estimating maize grain hardness.
EXPERIMENTAL
Maize sample collection
Thirteen commercial maize hybrids, all of which are normally cultivated
in northern Italy and processed for dry-milling foodstuffs,
were strip-test sown in five sites in 2007. The plot size was 100 m by
eight rows, and the row spacing was 0.75 m. The geographic and
main agronomic information concerning the experimental fields
are reported in Table 1. The experimental fields were cultivated
adopting the normal agronomic technique of each site. All 13
Table 1. Geographic and main agronomic information about the experimental fields
Site Location
Geographic
coordinates Soil a Altitude (m)
hybrids were compared at site D, while five hybrids were selected
at the other sites. At harvest, 100 ears were collected for each
hybrid by hand from each strip at the end of maturity (moisture
content of the grains between 20% and 26%) and shelled using
an electric sheller. The kernels were mixed thoroughly to obtain a
random distribution of the kernels and a 5 kg sample was slowly
dried to ∼14% moisture and stored in a cool room at 7 ◦ Cand
30% relative humidity until required. Storage of the kernels, equilibrated
with the air in the cool room, resulted in a mean moisture
content of 10.2% (range 9.2–10.9%) when tested. Before testing,
all the samples were equilibrated to room temperature (25 ± 1 ◦ C)
in paper bags for 48 h.
The33maizesamples,whichweretestedforseveralphysicaland
chemical properties, are listed in Tables 2 and 3. All the compared
tests were performed only on typical, flat-shaped, whole kernels
of the middle part of the ear, free from defects, which were
selected visually from each sample. The compared tests and their
abbreviations are summarized in Table 4.
Analytical methods
Moisture content and test weight (TW)
The moisture content and TW of the stored and dried maize
samples were determined by means of a grain analysis meter
(Dickey-John GAC2000, Colombes, France) using the supplied
programme. Calibration for moisture was checked using ovendrying
techniques. The test weight was recorded as kg hL −1 and
themoisturecontentasgkg −1 on the wet weight.
Thousand-kernel weight (TKW)
One hundred kernels were randomly collected from each sample
and weighed using an electronic balance to assess the thousandkernel
weight; this process was repeated three times. Thereafter,
the mean value was used to calculated the TKW.
Hard : soft endosperm ratio (H/S)
The H/S endosperm ratio in the grain samples was estimated by
sectioning the kernels and measuring the hard and softendosperm
areas visible at the cut surface. 12 Dried kernels were sectioned just
above the top of the embryo region using secateurs. The H/S ratios
were calculated for 15 kernels from each sample by measuring
the area of the cut surface and the soft endosperm region, using
their scanned images and an image analysis system, with ImageJ
software (version 1.38), which calculates the percentage of hard
and soft endosperm in each kernel.
Stenvert test
This test was based on the method described by Stenvert 21 and
Pomeranz et al. 22 A 20 g sample of kernel was ground using a
Sowing
date Harvest date
A Carignano 44 ◦ 55 ′ N, 07 ◦ 40 ′ E Sandy loam, Typic Udifluvents 236 12 April 2007 12 October 2007
B Chivasso 45 ◦ 14 ′ N, 7 ◦ 51 ′ E Sandy, Typic Hapludalfs 209 10 April2007 11 October 2007
C Feletto 18 ′ N, 7 ◦ 45 ′ E Sandy–medium texture, Mollic Hapluquepts 275 9 April 2007 4 October 2007
D Vigone 44 ◦ 51 ′ N, 07 ◦ 30 ′ E Sandy, Mollic Hapluquepts 256 29 March 2007 1 October 2007
E Villafranca 44 ◦ 47 ′ N, 7 ◦ 33 ′ E Sandy–medium texture, Typic Udifluvents 253 4 April 4 2007 10 October 2007
a USDA soil classification
J Sci Food Agric 2010; 90: 1870–1878 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1871

1872
www.soci.org M Blandino et al.
Table 2. Total grit yield and results of hardness tests for all maize samples, ranked according to total grit yield
Hybrid
Site of
cultivation
Total grit
yield
(g kg −1 )
Test
weight
(kg hL −1 )
Thousand
kernel
weight (g)
Hard/soft
endosperm
ratio
Milling
time a
(s)
Total
energy a
(J)
C/F
ratio a
Break
force b
(N)
Break
energy b
(mJ)
Floating
test
HCP CECINA D 600 81.1 350 3.5 9.8 1366 1.4 253 119 2123
Pioneer 3245 D 591 82.4 404 4.2 9.5 1379 1.3 266 119 2190
Dekalb DKC 6309 C 583 78.6 430 2.0 9.0 1344 1.2 193 75 2530
Dekalb DKC 6309 D 580 79.2 406 1.9 8.3 1352 1.2 236 104 2395
Pioneer 3235 B 578 80.6 380 4.0 8.5 1130 1.4 219 88 2058
Pioneer 3235 E 560 79.0 360 3.5 8.2 1151 1.1 222 96 2475
Pioneer 3235 A 559 79.6 370 4.8 8.8 1235 1.1 225 94 2323
Dekalb DKC 6309 A 557 80.1 435 3.5 9.0 1343 1.2 220 93 2428
Pioneer X1132R D 555 80.5 403 1.9 8.7 1353 1.3 242 101 2158
Pioneer X1132R B 552 79.1 395 1.0 9.1 1117 1.3 214 87 2368
Pioneer 3235 C 550 78.6 405 3.1 8.1 1293 1.3 203 87 2323
Pioneer 3235 D 550 80.8 372 4.4 9.1 1388 1.3 278 135 2088
KWS Kuadro D 548 77.8 325 0.9 9.1 1198 0.9 259 124 2415
Dekalb DKC 6309 B 542 79.8 410 3.7 8.9 1074 1.2 212 87 2343
KWS Kermess D 535 77.2 353 1.6 8.1 1149 1.1 215 91 2425
Syngenta NX6413 D 534 80.2 391 2.3 8.5 1216 1.1 300 155 2528
Pioneer X1132R A 532 78.6 390 3.0 8.5 1226 1.2 198 77 2175
HCP DORIA D 532 80.0 313 3.2 10.2 1520 1.3 233 98 2110
Pioneer X1132R C 526 75.3 395 3.0 8.8 1151 1.0 162 60 2555
Pioneer X1132R E 526 76.7 385 1.3 7.8 1270 1.0 148 56 2605
Dekalb DKC 6309 E 522 78.8 380 2.1 9.1 1211 1.1 198 80 2600
Pioneer PR34G44 D 501 79.4 444 2.5 8.4 1259 1.0 211 86 2425
Syngenta NX7034 D 493 76.1 421 1.4 8.9 1065 1.0 277 145 2653
Dekalb Tevere E 484 73.3 370 1.2 6.7 1037 0.8 180 78 3225
Dekalb Tevere B 478 75.2 385 0.8 7.9 1129 0.8 190 79 2890
Syngenta NX7234 E 475 74.1 359 0.4 5.5 989 0.8 145 52 2775
Dekalb Tevere D 448 73.1 388 0.4 7.8 1185 0.7 232 109 2835
Dekalb Tevere A 444 75.6 385 0.3 7.2 1193 0.8 229 103 2935
Syngenta NX7234 B 441 75.6 355 0.8 6.7 1068 0.8 184 75 2765
Syngenta NX7234 A 416 75.1 380 0.6 6.9 1117 0.7 177 70 2775
Syngenta NX7234 D 410 75.4 358 0.9 7.6 1029 0.7 208 91 3040
Dekalb Tevere C 408 74.3 375 0.3 7.1 1084 0.6 199 81 3195
Syngenta NX7234 C 404 73.5 365 0.2 6.2 939 0.5 178 71 3230
Average 516 77.7 383 2.1 8.2 1199 1.0 215 93 2544
Coefficient of variation (%) 11.1 3.3 7.5 66.5 12.8 11.1 23.6 16.9 25.6 13.2
a Parameters that refer to the Stenvert hardness test.
b Parameters that refer to the puncture test.
Culatti micro hammer mill (Labtech Essa, Belmont, Australia) fitted
with a 2 mm aperture particle screen at a speed of 2500 rpm when
empty. The laboratory mill was equipped with a computerized
data-logging system to log the instantaneous electric power
consumption during the milling test, as reported by Mestres
et al. 23 and Li et al. 12 Total milling energy (TME) and the milling
time (MT) taken to completely mill the 20 g kernel sample, were
determined from these data. These parameters were determined
three times for each maize sample.
Particle size index
A 20 g kernel sample was ground using a Culatti micro-hammer
mill fitted with a 2 mm aperture particle screen and was sieved into
two fractions using a Ro-Tap testing sieve shaker (WS Tyler Co.,
Cleveland, OH, USA) with 8 in. diameter brass sieves. Sieve meshes
of 500 and 700 µm were chosen to represent the most common
product obtained in the milling industry: prime or large grits
(700–2000 µm) and fine meal (

Hardness methods for testing maize kernel www.soci.org
Table 3. Chemical and physical characteristics for all maize samples, ranked according to total grit yield
Hybrid
Site of
cultivation
Moisture
content
(g kg −1 )
Protein
content
(g kg −1 )
Starch
content
(g kg −1 )
Lipid
content
(g kg −1 )
Ash
content
(g kg −1 )
Amylose/
amylopectin
rate
Kernel
length
(mm)
Kernel
width
(mm)
Kernel
depth
(mm) Sphericity
HCP CECINA D 103 105 613 56 16 0.27 12.9 8.2 4.5 0.60
Pioneer 3245 D 102 102 638 48 15 0.28 13.0 8.8 4.7 0.62
Dekalb DKC 6309 C 104 104 629 58 17 0.34 13.4 9.6 4.5 0.62
Dekalb DKC 6309 D 98 102 623 49 16 0.25 13.5 9.1 4.5 0.61
Pioneer 3235 B 109 109 641 47 17 0.32 12.8 8.9 4.3 0.62
Pioneer 3235 E 104 101 670 40 16 0.32 12.5 8.8 4.4 0.63
Pioneer 3235 A 105 116 610 49 17 0.33 12.7 9.0 4.3 0.62
Dekalb DKC 6309 A 101 109 602 55 17 0.34 12.9 9.1 4.4 0.62
Pioneer X1132R D 92 107 615 52 17 0.29 13.5 8.7 4.6 0.60
Pioneer X1132R B 109 110 635 52 18 0.32 13.4 8.8 4.5 0.60
Pioneer 3235 C 105 114 612 57 19 0.31 12.9 8.9 4.5 0.62
Pioneer 3235 D 99 113 618 47 16 0.35 12.8 9.0 4.6 0.63
KWS Kuadro D 104 97 638 41 15 0.30 13.3 7.5 4.3 0.57
Dekalb DKC 6309 B 104 99 635 49 16 0.30 12.9 9.1 4.6 0.63
KWS Kermess D 92 93 656 31 14 0.28 13.6 8.1 4.3 0.57
Syngenta NX6413 D 104 97 646 47 16 0.33 12.4 8.8 5.2 0.67
Pioneer X1132R A 99 102 629 55 17 0.35 13.3 8.6 4.4 0.60
HCP DORIA D 103 115 608 51 17 0.31 13.2 8.8 4.1 0.59
Pioneer X1132R C 104 106 644 53 18 0.35 14.0 8.5 4.5 0.58
Pioneer X1132R E 98 113 642 55 19 0.35 13.3 8.6 4.6 0.61
Dekalb DKC 6309 E 103 110 639 46 17 0.34 13.5 9.0 4.4 0.60
Pioneer PR34G44 D 95 97 640 45 16 0.26 13.8 9.1 4.7 0.61
Syngenta NX7034 D 99 97 639 48 16 0.32 13.1 9.3 5.1 0.65
Dekalb Tevere E 108 88 664 47 16 0.28 13.4 8.6 4.3 0.59
Dekalb Tevere B 104 92 649 49 16 0.27 13.2 9.1 4.2 0.60
Syngenta NX7234 E 102 83 672 43 15 0.28 13.6 8.6 4.1 0.57
Dekalb Tevere D 96 91 639 55 16 0.24 13.8 9.0 4.4 0.59
Dekalb Tevere A 99 94 641 50 16 0.26 13.1 9.0 4.2 0.60
Syngenta NX7234 B 103 81 693 41 15 0.26 13.7 9.0 4.0 0.57
Syngenta NX7234 A 108 94 648 46 15 0.29 14.1 8.8 4.1 0.57
Syngenta NX7234 D 99 86 657 47 14 0.24 14.1 8.9 4.0 0.56
Dekalb Tevere C 104 94 644 52 17 0.27 13.4 8.9 4.3 0.60
Syngenta NX7234 C 106 91 672 52 16 0.25 14.1 8.9 4.1 0.57
Average 102 100 639 49 16 0.30 13.3 8.8 4.4 0.6
Coefficient of variation (%) 4.3 9.5 3.2 11.7 7.3 11.9 3.4 4.2 6.2 4.2
small cylinder into the tissue and measuring the evolution of
stress–strain. It is assumed to measure a mix of compression
(under the plunger) and shearing. 25 Hardness was expressed as
the break force (HF) evaluated in newtons and the break energy
(HWF) evaluated in megajoules. HF corresponds to the resistance
of kernels to the penetration of the probe, while HWF measures
the area beneath the deformation curve between force values 0
and HF. 26 The data were all acquired at 400 Hz and using Texture
ExpertExceed software (Texture Technologies, Scarsdale, NY, USA).
Figure 1 shows a typical force–time deformation curve of texture
analysis, obtained from the kernel puncture test.
Floating test (FLT)
This test was used to assess the density of the maize grain; the
number of floating kernels (floaters) in a variable density solution
was recorded. The method that was adopted is a modification of
that proposed by Wichser. 27 100 mL tetrachloroethylene (density
1.62 g mL −1 ) and 40 mL petroleum ether (density 0.653 g mL −1 )
were added to an Erlenmeyer flask, and the solution density
obtained was 1.34 g mL −1 . A sample of 50 kernels was put
into Erlenmeyer flasks; 5 mL petroleum ether was gradually
added to the solution and the density of the solution was
decreased until there were no kernels left floating. The number
of kernels floating at each addition of petroleum ether to the
solution was recorded and a precipitation curve was obtained.
FLT measures the area beneath the precipitation curve and this
parameter is adversely correlated to the density of the kernels.
These parameters were determined three times for each maize
sample.
Kernel dimensions and sphericity (S)
The spatial dimensions of 50 kernels of each hybrid were calculated
by measuring the average length (L), width (W) anddepth(D) of
the whole kernels using a 0.1 mm precise gauge. These data were
used to calculate S by means of the following formula: 22
�
Volume of solid
S =
Volume of circumscribed sphere
� 1/3
�
LWD
=
L
J Sci Food Agric 2010; 90: 1870–1878 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
� 1/3
1873

1874
Table 4. List of abbreviations of parameters analysed on maize kernel
Abbreviation Parameter
TGY Total grit yield
TW Test weight
TKW Thousand-kernel weight
H/S Hard/soft endosperm ratio
MT Milling time
TME Total milling energy
C/F Coarse/fine ratio
HF Break force
HWF Break energy
FLT Floating test
L Kernel length
W Kernelwidth
D Kernel depth
S Sphericity
PC Protein content
SC Starch content
OC Oil content
AC Ash content
AS/AP Amylose/amylopectin rate
MT, TME: parameters that refer to Stenvert hardness test.
HF, HWF: parameters that refer to puncture test.
Figure 1. A typical force–time deformation curve, obtained from the
texture analysis puncture test of maize kernels.
The sphericity values range from 0 (no three-dimensional object)
to 1 (perfect sphere). The closer the sphericity is to unity, the more
spherical the kernel; conversely, the lower the sphericity, the flatter
the kernel.
Kernel composition
A grab sample of approximately 300 g of maize was ground
to a fine flour using a Foss Tecator Cyclotec 1093 sample
mill fitted with a 1 mm screen. The protein (PC), starch (SC),
oil (OC) and ash (AC) contents were estimated by nearinfrared
reflectance spectroscopy, using a NIRsystems 6500
monochromator instrument (Foss-NIRsystems, Silver Spring, MD,
www.soci.org M Blandino et al.
USA) that was calibrated for wet chemical methods. The protein,
starch, oil and ash contents were adjusted to 15% moisture content
using the NIR-predicted moisture content of the ground grain.
The amylose : amylopectin ratio (AS/AP)of the maize kernels was
estimated using a Megazyme commercial assay kit (Megazyme
International Ireland Ltd, Wicklow, Ireland), based on the concanavalin
A precipitation procedure. 28
Micromilling procedure
According to Yuan and Flores, 29 a micromilling procedure was
used to process the maize grain sample and provide an index
of the efficiency of the quality tests for dry-milling processing.
Twenty intact, whole kernels were soaked in distilled water for
1 h at room temperature (25 ± 1 ◦ C) and the bran and germ
were removed manually with a scalpel. The procedure was always
performedbythesametrainedresearchertoensureastandardized
determination and avoid subjective determination. The obtained
endosperms were conditioned in an oven at 40 ◦ Cfor48h,and
were then ground and sieved using the same procedure as the
particle size index test. The total grit yield (TGY) corresponded
to the percentage of the fraction from 2.000 to 700 µm, which
was chosen to represent the main products obtained in the
conventional dry-milling industry. 6
Thetotalgrityieldwasexpressedasapercentageofthetotaldrymilled
fractions (g kg −1 ). Considering that this procedure achieved
a good separation of the bran, germ and endosperm, 29 and
that the grounding operations were conducted under standard
conditions for all the compared maize samples, micromilling
can be considered to provide a good index of dry-milling
performance.
Statistical analyses
When present, replicates were averaged. The coefficient of variation
(CV) was calculated for each parameter. Simple correlation
coefficients were then obtained for all the quality factors relative
to one another (SPSS, Version 16.0, SPSS Inc., Chicago, IL,
USA).
RESULTS AND DISCUSSION
In Tables 2 and 3 the data for each analysed parameter are shown
for each maize sample, ranked for grit yield. Table 5 reports
the correlation coefficients and their significance between the
parameters of maize kernels analysed.
Although the compared maize hybrids are ordinary commercial
hybrids that are normally cultivated in northern Italy,
their composition varied in starch (602–693 g kg −1 ), protein
(81–116 g kg −1 ), lipid (31–58 g kg −1 ) and ash (14–19gkg −1 )
content and the relative amounts of amylose and amylopectin
(0.24–0.35). As expected, the starch (SC) and protein (PC) contents
were negatively correlated with each other (r =−0.79), while
PC was positive correlated with lipid (0.43) and ash (0.71). The
amylose : amylopectin ratio (AS/AP) was positively correlated with
PC (0.68) and AC (0.56), while a negative relationship was observed
for SC (−0.49).
TGY was between 404 and 600 g kg −1 . Although the evaluated
hybrids are normally used in food processes, they showed remarkable
differences in their aptitude to dry-milling transformation.
Among the compared hybrids, HCP Cecina, Pioneer 3245, Dekalb
DKC 6309, Pioneer X1132R and Pioneer 3235 showed the highest
grit yield.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1870–1878

Hardness methods for testing maize kernel www.soci.org
Table 5. Correlation matrix between the physical and chemical parameters tested on the maize kernels a
Parameter TGY TW TKW H/S MT TME C/F HF HWF FLT L W D S PC SC OC AC
TW 0.84∗∗ TKW 0.19 0.19
H/S 0.75∗∗ 0.80∗∗ 0.15
MT 0.74∗∗ 0.79∗∗ 0.11 0.68∗∗ TME 0.66∗∗ 0.72∗∗ −0.35 0.56∗∗ 0.72∗∗ C/F 0.92∗∗ 0.91∗∗ 0.20 0.78∗∗ 0.79∗∗ 0.72∗∗ HF 0.40∗ 0.57∗∗ 0.03 0.37∗ 0.59∗∗ 0.44∗∗ 0.43∗ HWF 0.30 0.44∗∗ 0.01 0.27 0.50∗∗ 0.35∗ 0.32 0.98∗∗ FLT −0.85∗∗ −0.89∗∗ −0.08 −0.78∗∗ −0.77∗∗ −0.70∗∗ −0.93∗∗ −0.45∗∗ −0.33
L −0.60∗∗ −0.60∗∗ −0.06 −0.61∗∗ −0.44∗ −0.36∗ 0.55∗∗ −0.55∗∗ −0.53∗∗ 0.48∗∗ W −0.13 0.01 0.61∗∗ 0.07 −0.05 0.02 −0.30 −0.04 −0.07 0.13 −0.05
D 0.46∗∗ 0.46∗∗ 0.57∗∗ 0.35∗ 0.47∗∗ 0.30 0.43∗ 0.57∗∗ 0.60∗∗ −0.35∗ −0.48∗∗ 0.16
S 0.51∗∗ 0.56∗∗ 0.52 0.54∗∗ 0.45∗∗ 0.34 0.50∗∗ 0.57∗∗ 0.56∗∗ −0.39∗ −0.80 0.45∗∗ 0.82∗∗ PC 0.70∗∗ 0.69∗∗ 0.17 0.68∗∗ 0.73∗∗ 0.71∗∗ 0.76∗∗ 0.24 0.14 −0.72∗∗ −0.46∗∗ 0.05 0.35∗ 0.44∗∗ SC −0.58∗∗ −0.63∗∗ −0.27 −0.56∗∗ −0.73∗∗ −0.78∗∗ −0.67∗∗ −0.43∗ −0.34 0.63∗∗ 0.39∗ −0.08 −0.34 −0.40∗ −0.79∗∗ OC 0.12 0.09 0.40∗ 0.12 0.22 0.34 0.18 −0.08 −0.09 −0.07 −0.07 0.35∗ 0.20 0.25 0.43∗ −0.55∗∗ AC 0.34 0.22 0.38∗ 0.29 0.27 0.31 −0.37∗ −0.23 −0.28 −0.27 −0.24 0.27 0.24 0.34 0.71∗∗ −0.49∗∗ 0.71∗∗ AS/AP 0.51∗∗ 0.42∗ 0.17 0.52∗∗ 0.47∗∗ 0.31 0.50∗∗ 0.05 0.03 −0.48∗∗ −0.48 0.06 0.40∗ 0.49∗∗ 0.68∗∗ −0.37∗ 0.22 0.56∗∗ J Sci Food Agric 2010; 90: 1870–1878 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
a Abbreviations: see Table 4.
The data reported in the table are Pearson product–moment correlation coefficients. ∗ Correlation significant at P ≤ 0.05; ∗∗ correlation significant at P ≤ 0.01.
1875

1876
Considering the compared tests and parameters recorded,
TGY was shown to be significantly correlated with C/F (0.92),
FLT (−0.85), TW (0.84), H/S (0.74), MT (0.74), PC (0.70), TME (0.66),
L(−0.60), SC (−0.58), S (0.51), AL/AP (0.51), D (0.46) and HF (0.40),
as reported in Table 5.
Particle size index
The correlation matrix shows that the percentage repartition into
coarse (C) and fine (F) material after standard milling (C/F) is the
best descriptor of maize milling ability. The data obtained from
our study showed that the C/F ratio was closely correlated with the
H/S ratio (0.78). Although the H/S ratio ranged from 0.2 to 4.8 and
exhibited a CVof 67, while the C/F ratio ranged from 0.5 to 1.4 with a
CVof26,thelatterisaconsiderablylesstime-consuminglaboratory
test compared to quantification of the hard and soft fractions in
the endosperm. Moreover, considering that the differences in
the germ and pericarp content among commercial maize hybrids
are generally low, the C/F ratio is also an obvious descriptor of
maize milling performance. 30,31 This test is a good estimator of
grain hardness, since it can measure objectively and accurately
and give an indirect but clear evaluation of the hard (H) and soft
(S) fractions. In fact, coarse material is mostly obtained by milling
the hard endosperm fraction. 32
Floating test
The FLT was significantly related to the physical kernel characteristics
obtained from the other tests (C/F, TW, H/S, MT, TME, HF), thus
confirming the data reported by Pomeranz et al. 24 Furthermore,
this test was less affected by the shape of the kernels than the other
tests, e.g. TW and H/S, which showed a higher correlation with
kernel sphericity. FLT was highly and negatively correlated with
PC and AS/AP, while a direct relationship was observed with SC.
The observed CV of the floating test was 13. Considering the
capacity of the FLT to discriminate maize samples according to
hardness, quantification of the precipitation curve in a variabledensity
solution, applied in our experiment, seems to be more
efficient than the simple percentage of floaters observed in a
standard solution. 27,33 This study confirms that kernel density,
measured by means of the FLT, is a good descriptor of kernel
hardness and of maize milling performance. 10,33
Test weight
Of all the compared tests, TW showed the third strongest correlation
with TGY (0.84). It also demonstrated a close relationship
with all the other tests used to estimated grain hardness (C/F, FLT,
H/S, MT, TME, HF, HWF). On the other hand, no correlation was
found between these parameters and TKW. This result confirms
that kernel hardness does not depend on kernel weight alone but
also on its shape. TW is correlated negatively with kernel length
(−0.60) and positively with kernel depth (0.46) and increases with
the sphericity of grain (0.56).
The collected data confirm that TW is a simple estimator of grain
hardness 12 and, since it is widely used, it is the first parameter that
needs to be considered to evaluate the dry-milling performance
of maize hybrids. On the other hand, TKW was not able to provide
effective information on grain hardness, thus confirming data
reported by Pomeranz et al. 24 and Mestres et al. 10
H/S ratio
Compared to C/F, FLT and TW, the ratio of the cross-sectional
area of hard and soft endosperm (H/S) has demonstrated a lower
www.soci.org M Blandino et al.
correlation coefficient with TGY (0.74). This parameter instead
showed the highest CV of all the methods (66.5), thus confirming
that H/S ratio is an efficient kernel hardness test, which is also able
to discriminate the kernels from ordinary commercial hybrids. 29
The correlation matrix shows a close correlation between H/S
and other physical kernel characteristics, such as test weight (0.80),
FLT (−0.78) and MT (0.68). Considering the chemical composition
of the kernel, H/S was shown to be significantly related to PC (0.68)
and AS/AP (0.52). This parameter was also directly related to kernel
sphericity (0.54): the thickness of the hard endosperm region is
higher in rounder kernels than in flat ones. The H/S ratio increases
with an increase in kernel depth and a reduction in kernel length,
confirming data reported by Mestres et al. 10 and Pomeranz et al. 22
The digital image processing software used in this experiment
for the estimation of H/S ratio allows simpler and more accurate
measurements of the hard and soft areas of the cross-section to be
obtainedthanwiththeothermethods,suchasquantificationusing
camera lucida drawings, or enlarged photographs, 34,35 which are
subjective and require sophisticated equipment and particular
expertise. However, this method has been confirmed to be time
consuming and not practical for very large numbers of samples,
although it is the most direct measurement of the proportion
of hard endosperm available in maize kernels, since the internal
endosperm tissue is viewed directly.
Stenvert test
As reported by Li et al., 12 the parameters obtained using the
Stenvert hardness test – milling time (MT) and total milling energy
(TME) – were highly correlated with H/S. Of these two parameters,
MT appears to be a better descriptor of maize grain hardness
than TME, since the correlation with H/S is closer (0.68 and 0.56
for MT and TME, respectively) and the CV is moderately higher
(12.8 and 11.1 respectively). In the same way, MT showed a higher
correlation with TGY than TME. However, TME could provide more
objective information, since the data are not recorded by an
operator, but through a computerized data-logging system which
logs the instantaneous electric power consumption during the
milling test.
Kernel composition
In this experiment, PC was significantly correlated with several
parameters obtained from tests used to estimate grain hardness,
thus confirming the data reported by Dorsey-Redding et al. 36 and
Lee et al., 15 but was also correlated with the shape of the kernel.
Round kernels were higher in protein content than flat kernels,
as reported by Pomeranz et al. 22 Moreover, kernels with a high
protein content have been confirmed to have a higher grain
density (FLT) and produce more total grits. 29
The dry-milling yield and maize kernel hardness-associated
properties measured in this study were negatively related to SC
but positively associated with AS/AP. These data are consistent
with previous studies. 3,37
Dombrink-Kurtzam and Knutson 38 suggest that increased
amounts of amylose may result in increased compressibility of
the starch granules in the hard endosperm, which leads to a
compactedstateandapolygonalshape,whileagreaterproportion
of amylopectin, potentially more crystalline, could lead to a less
compressible and softer endosperm.
No significant correlations were observed between TGY and
OC or AS. Significant correlations between the dry-milling
performance of maize hybrids and ash or lipid content have
been reported by Mestres et al. 10 and Dorsey-Redding et al. 36
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1870–1878

Hardness methods for testing maize kernel www.soci.org
Sphericity (S)
Although all the tests in our study have only been performed on
the typical flat-shaped whole kernels of the middle part of the
maize ear to reduce variability among maize samples, a positive
correlation between S and TGY was observed. Mestres et al. 10
reported that the dry-milling yield of maize hybrids could also be
predicted on the basis of sphericity, although this parameter was
not significantly correlated with kernel vitreousness. Moreover, in
this dataset a significant correlation coefficient was established
between S and the various hardness parameters. These data
confirm that kernels with higher sphericity are generally higher
in flint-like characteristics than flatter kernels 22 ,althoughstudies
exist that report a lower hardness ratio for round kernels than for
flat ones. 39
Texture analysis
As far as the puncture test is concerned, only break force (HF) was
correlated, though weakly, to TGY (0.40), while the relationship
between TGY and break energy (HWF) was not significant. Both
these parameters – HF and HWF – were correlated with TW, MT
and TME, and HF showed a higher correlation coefficient than
HWF for all the relationships with these parameters. Only HF was
significantly correlated with H/S (0.37), C/F (0.43) and FLT (−0.45).
In our experiment, HF and HWF were not significantly correlated
with PC or AS/AP, while a negative correlation was observed
between HF and SC. Moreover, both HF and HWF were highly
correlated with S (0.57 and 0.56, respectively) and in particular with
D (0.57 and 0.60 respectively), while they were not significantly
related to PC. These observations suggest that HF and HWF
are probably affected more by the kernel shape than by their
endosperm composition: kernels with a higher sphericity, as a
consequence of a greater depth, present a higher physical and
structural resistance to penetration to the probe than do flatter
kernels. The moderate correlation observed between HF and HWF
with other kernel hardness parameters, such as TW, MT, TME, H/S,
C/F and FLT, and between HF and GY, is probably an indirect effect
of the relationship that exists between these parameters and the
sphericity of kernels. In the study conducted by Gaytán Martínez
et al., 20 the resistance obtained from a puncture test through
texture analysis showed a better correlation with the floating test
(−0.74), compared to our experiment. In the same study, since a
positive correlation (0.59) was also observed between the grain
resistance to a puncture test and starch granule size, the authors
suggested an important effect of the structural arrangement of
the starch granules within the endosperm structure: soft maize
starch granules are predominantly spherical and loosely packed
within a protein matrix, while hard maize has mostly polygonal and
densely packed starch granules. A similar result has been obtained
on wheat by Greffeuille et al., 17 who suggested that greater
starch–protein matrix adhesion in hard grains compared to soft
ones could explain the greater energy necessary to break the grain.
In their study, Shandera and Jackson 19 compared different
solvents and heat treatments on kernels accurately selected,
according to dimension, from two maize hybrids. The authors
underlined that kernel shape, surface area and thickness could
have an important effect on textural testing.
Since the results of the puncture test on single kernels have
shown a very weak correlation with TGY, this test appears to
be an inadequate predictor of kernel hardness, when this kernel
characteristic is used to distinguish maize genotypes according
to their dry-milling behaviour and performance. On the other
hand, if hardness is only considered for the mechanical resistance
of the whole grain to applied deformation, 40 this test could
provide information on the power required to break kernels during
processing although, regarding dry-milling, correlation with the
energy required to mill grain (TME) was moderate.
This method could be used to clearly show the difference in
maize endosperm texture when applied to a product which is
able to resist compression forces without completely breaking.
For example, the application of this method to maize grits or meal
extrudates has shown a clear relationship between the obtained
force–deformation curve and the porosity and composition of the
product tested. 41,42
CONCLUSION
This research confirms that the dry-milling behaviour of maize
is clearly influenced by the various physical and chemical
characteristics of the kernel, which combine to determine its
hardness. First, the results have indicated the significance of test
weight as an index of maize hardness. The test weight has been
confirmed to be a simple estimator of grain hardness and is the
first parameter to consider in evaluating the grain hardness of
a maize hybrid. Among the hardness tests, C/F has been shown
to be the best descriptor of maize milling ability, followed by
FLT. Since both these indicators also show a close relationship
with the proportion of hard endosperm in the kernel and a good
capacity to discriminate among commercial maize hybrids, these
tests could be the most interesting to characterize maize lots for
their dry-milling performance. A good correlation with total grit
yield after dry milling has also been observed for H/S, MT, TME
and PC, while SC, S and AS/AP seem to play a minor role. Puncture
test did not provide good indications concerning the impact of
hardness on kernel grinding properties.
At present no quality criteria are universally recognized by maize
kernel end users. In fact the hybrids in this study, which were
chosen from the varieties normally cultivated and processed in
dry-milling processes, exhibited remarkable differences in total grit
yield. A forthcoming objective will be to adapt commercial maize
varieties according to their end uses and to breed new hybrids,
not only as far as their agronomic performance is concerned,
but also for their technological properties. In order to obtain
a widely accepted hardness evaluation and milling behaviour
standard for maize kernels, it will be necessary to establish a
limited list of simple, rapid and reliable tests, which could improve
the explanation of maize hardness measurements in relation to
end-use value, using a multivariate approach that simultaneously
takes into account the hardness associated with both physical
and chemical properties. The data of this study offer preliminary
information that could help provide the main grain traits that
determine end-use processing performance and reduce the risk of
misclassification of maize hardness.
ACKNOWLEDGEMENTS
The funds for this research were provided thanks to grants from
the Provincia di Torino, Servizio Agricoltura.
REFERENCES
1 Bettge AD and Morris CF, Relationships among grain hardness,
pentosan fractions and end-use quality of wheat. Cereal Chem
77:241–247 (2000).
J Sci Food Agric 2010; 90: 1870–1878 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1877

1878
2 Watson SA, Measurement and maintenance of quality, in Corn:
Chemistry and Technology, ed. by Watson SA and Ramstad PE.
American Association of Cereal Chemists, St Paul, MN, pp. 125–183
(1987).
3 Lee KM, Bean SR, Alavi S, Herrman TJ and Waniska RD, Physical and
biochemical properties of maize hardness and extrudates of
selected hybrids. J Agric Food Chem 54:4260–4269 (2006).
4 Shandera DL, Jackson DS and Johnson BE, Quality factor impacting
processing of maize dent hybrids. Maydica 42:281–289 (1997).
5 Serna-Saldivar SO, Gomez MH and Rooney LW, Food uses of regular
and specialty corns and their dry-milled fraction, in Specialty Corns
(2nd edn), ed. by Hallauer AR. CRC Press, Boca Raton, pp. 303–333
(1991).
6 Wu YV, Corn hardness as related to yield and particle size of fractions
from micro hammer-cutter mill. Cereal Chem 69:343–347 (1992).
7 Chandrashekar AandMazhar H,Thebiochemicalbasisandimplication
of grain strength in sorghum and maize. J Cereal Sci 30:193–207
(1999).
8 Mestres C and Matencio F, Biochemical basis of kernel milling
characteristics and endosperm vitreousness of maize. J Cereal Sci
24:283–290 (1996).
9 Robutti JL, Borras FS, and Eyherabide GH, Zein composition of
mechanically separated coarse and fine portions of maize kernels.
Cereal Chem 74:75–78 (1997).
10 Mestres C, Louis-Alexander A, Matencio F and Lahlou A, Dry milling
properties of maize. Cereal Chem 68:51–56 (1991).
11 Fox G and Manley M, Hardness methods for testing maize kernels. J
Agric Food Chem 57:5647–5657 (2009).
12 Li PX-P, Hardacre AK, Campanella OH and Kirkpatrick KJ, Determination
of endosperm characteristics of 38 corn hybrids using the
Stenvert hardness test. Cereal Chem 73:466–471 (1996).
13 Almeida-Dominguez HD, Suhendro EL and Rooney LW, Factors
affecting rapid visco analyser curves for the determination of maize
kernel hardness. J Cereal Sci 25:93–102 (1997).
14 Williams P, Geladi P, Fox G and Manley M, Maize kernel hardness
classification by near infrared (NIR) hyperspectral imaging and
multivariate data analysis. Anal Chim Acta 653:121–130 (2009).
15 Lee KM, Herrman TJ, Lingenfelser J and Jackson DS, Classification
and prediction of maize hardness-associated properties using
multivariate statistical analyses. J Cereal Sci 41:85–93 (2005).
16 Fox GP, Osborne B, Bowman J, Kelly A, CaKir M, Poulsen D, et al,
Predicting a hardness measurement using the single-kernel
characterization system. Cereal Chem 73:278–283 (1996).
17 Greffeuille V, Abecassis J, Rousset M, Oury F-X, Faye A and Bar
L’Helgouac’h C, Grain characterization and milling behaviour of
near-isogeniclinesdifferingbyhardness.TheorApplGenet 114:1–12
(2006).
18 Haddad Y, Benet JC and Abecassis J, A rapid general method for
appraising the rheological properties of the starchy endosperm of
cereal grains. Cereal Chem 75:673–676 (1998).
19 Shandera DL and Jackson DS, Corn kernel structural integrity: analysis
using solvent and heat treatments. Cereal Chem 79:308–316 (2002).
20 Gaytán Martínez M, Figueroa Cardenas JD, Reyes Vega ML, Rincon
Sanchez F and Morales Sanchez E, Microstructure of starch granules
related to kernel hardness in corn. Rev Fitotec Mex 29:135–139
(2006).
21 Stenvert NL, Grinding resistance, a simple measure of wheat hardness.
Flour Anim Feed Milling 12:24–26 (1974).
22 Pomeranz Y, Czuchjowska Z, Martin CR and Lai F, Determination of
corn hardness by Stenvert hardness tester. CerealChem 62:108–110
(1985).
www.soci.org M Blandino et al.
23 Mestres C, Matencio F and Louis-Alexandre A, Mechanical behaviour
of corn kernels: development of a laboratory friability test that can
predict milling behaviour. Cereal Chem 72:652–657 (1995).
24 Pomeranz Y, Hall GE, Czuchjowska Z and Lai F, Test weight, hardness,
and breakage susceptibility of yellow dent corn hybrids. Cereal
Chem 63:349–351 (1986).
25 Roudot AC,Someconsiderationsforatheoryofplanttissuemechanics.
Sci Aliments 26:409–426 (2006).
26 Letaief H, Rolle L and Gerbi V, Mechanical behavior of wine grapes
under compression tests. Am J Enol Vitic 59:323–329 (2008).
27 Wichser WR, The world of corn processing. Am Miller Process 89:29–31
(1961).
28 Gibson TS, Solah VA and McCleary BV, A procedure to measure
amylose in cereal starches and flour with concanavalin A. J Cereal
Sci 25:111–119 (1997).
29 Yuan J and Flores RA, Laboratory dry-milling performance of white
corn: effect of physical and chemical corn characteristics. Cereal
Chem 73:574–578 (1996).
30 Kirleis AW and Stroshine RL, Effect of hardness and drying
air temperature on breakage susceptibility and dry-milling
characteristics of yellow dent corn. Cereal Chem 67:523–528 (1990).
31 Pomeranz Y, Czuchjowska Z and Lai F, Gross composition of coarse
and fine fractions of small corn samples ground on the Stenvert
hardness tester. Cereal Chem 63:22–26 (1986).
32 Le Deschault de Monredon F, Devaux MF and Chaurand M, Particle
size characterization of ground fraction of dent and flint maize. Sci
Aliments 16:117–132 (1996).
33 Wu YV and Bergquist RR, Relation of corn grain density to yields of
dry-milling products. Cereal Chem 68:542–544 (1991).
34 Kirleis AW, Crosby KD and Housley TL, A method for quantitatively
measuring vitreous endosperm area in sectioned sorghum grain.
Cereal Chem 61:556–558 (1984).
35 Louis-Alexandre A and Mestres C, Measurement of endosperm
vitreousness of corn: a quantitative method and its application
to African cultivars. Cereal Chem 68:614–617 (1991).
36 Dorsey-Redding C, Hurburgh CR Jr, Johnson LA and Fox SR,
Relationship among maize quality factors. Cereal Chem 68:602–605
(1991).
37 Robutti J, Borras F, Ferrer M, Percibaldi M and Knutson CA, Evaluation
of quality factors in Argentine maize races. Cereal Chem 77:24–26
(2000).
38 Dombrink-Kurtzam MA and Knutson CA, A study of maize endosperm
hardness in relation to amylose content and susceptibility to
damage. Cereal Chem 74:776–780 (1997).
39 Kim TH, Hampton JG, Opara LU, Hardacre AK and Mackay BR, Effects
of maize grain size, shape and hardness on drying rate and the
occurrence of stress cracks. J Sci Food Agric 82:1232–1239 (2002).
40 Greenaway WT, A wheat hardness index. Cereal Sci Today 14:4–7
(1969).
41 Martìnex-Bustos F, Chang YK, Bannwart AC, Rodrìguez M, Guedes PA
and Gaiotti ER, Effect of calcium hydroxide and processing
conditions on corn meal extrudates. Cereal Chem 75:796–801
(1998).
42 Gonzales RJ, De Greef DM, Torres RL, Borras FS and Robutti J, Effect of
endosperm hardness and extrusion temperature on properties of
product obtained with grits from two commercial maize cultivars.
Lebensm Wiss Technol 37:193–198 (2004).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1870–1878

Research Article
Received: 26 February 2010 Revised: 22 April 2010 Accepted: 4 May 2010 Published online in Wiley Interscience: 3 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4028
Use of earthworms (Eisenia fetida)toreduce
phytotoxicity and promote humification of
pre-composted olive oil mill wastewater
Grazia Masciandaro, Cristina Macci, ∗ Serena Doni and Brunello Ceccanti
Abstract
BACKGROUND: Olive mill wastewaters (OMWW) contain a high recalcitrant organic load and an associated toxicity that make
their treatment necessary before environmental application. The aim of this study was to evaluate the feasibility of promoting
the valorization and reducing the phytotoxicity of OMWW through a pre-composting process together with straw-chip bulking
materials followed by the application of earthworms (Eisenia fetida) in the presence of oat seedlings (Avena sativa L.) seedlings.
RESULTS: After 3 months, the pre-composted material showed properties similar to a partially digested compost with
some significant amelioration of chemical–physical and biochemical properties. The application of earthworms permitted
a significant decrease in chemical (total organic carbon, water-extractable organic carbon, total nitrogen) and biological
parameters (dehydrogenase enzyme activity), and an increase in humic substances and available nitrogen forms. In the
presence of plants a higher C/N ratio and a lower content of nitrates were observed. In addition, the reduction in phenolic
compounds observed in treatments with earthworms caused a decrease in phytotoxicity by about 50% with respect to the
pre-composted material, which results in an increase in germination index.
CONCLUSION: The utilization of earthworms, in particular in the presence of plants, may be an ecologically sound and
economically feasible technology to obtain a non-toxic, high-value product useful for agricultural purposes.
c○ 2010 Society of Chemical Industry
Keywords: phenolic compounds; organic matter; olive oil mill wastewater; vermicompost; earthworms
INTRODUCTION
Olive oil mill wastewaters (OMWW) are produced during the
olive oil extraction process. OMWW are characterized by low pH
(4.0–6.5) and by a high pollution load due to (1) high content of
mineral salts and (2) large amounts of organic compounds and
polyphenol substances. 1 In particular, these last compounds are
mainly responsible for the high toxicity of the OMWW because they
are resistant to biological degradation and also have antimicrobial
properties.
In Italy the disposal of OMWW into soil is only possible
according to strict national technical regulations (law 574, 11
November 1996). These criteria are very rigorous in reducing
possible environmental damage and take into account mainly
soil texture, effluent volume/soil surface ratio and organic
load of the effluents. As an alternative to spreading on the
land, in recent years physicochemical methods and biological
treatments have been employed in order to recover or recycle
this residue. The physicochemical methods include dilution,
evaporation, sedimentation, filtration and centrifugation. 2 All
these treatments are expensive and do not solve the problem
completely because the product is a residue which needs
to be discharged. Biological methods, on the other hand,
are generally low-cost processes and have the advantage of
producing a stabilized material with fertilizing properties. They
are mostly based on anaerobic degradation, 3 composting, 4,5 and
the utilization of fungi that are able to metabolize lignin-related
compounds. 6
In recent years also vermicomposting – a special composting
process that involves the addition of certain species of earthworms
to enhance the conversion and detoxification of different organic
wastes, 7–9 including olive oil mill wastes (OMW), such as dry olive
cake and OMWW, 10,11 has been established. The worms are able to
decompose the organic compounds by fragmenting the substrate
material and by providing the more microbiologically active casts
(fecal material). This leads to humification through which the
unstable organic matter is oxidized and stabilized. 12
Vermicompost is characterized by a higher nutrient content,
homogeneity and freedom from pathogens compared to traditional
compost. 13,14 Vermicompost is, moreover, believed to have
additional attributes of providing enzymes and hormones which
stimulate plant growth. 15 – 17 On the other hand, the presence of
plants could affect earthworm development since they are considered
to feed on roots 18 and are found to concentrate in the
rhizosphere zone. 19,20
∗ Correspondence to: Cristina Macci, Institute of Ecosystem Study – National
Research Council, 56124 Pisa, Italy. E-mail: cristina.macci@ise.cnr.it
Institute of Ecosystem Study – National Research Council (CNR), 56124 Pisa,
Italy
J Sci Food Agric 2010; 90: 1879–1885 www.soci.org c○ 2010 Society of Chemical Industry
1879

1880
It has been hypothesized that plants, at the rhizosphere level,
provide a complex and dynamic microenvironment in which
microorganisms and earthworms, in association with roots, form
unique communities that have a high metabolic activity and
considerable potential for the detoxification of OMWW. 21,22
In view of this, the aim of the present study was to evaluate
the feasibility of promoting humification and reducing the
phytotoxicity of OMWW through: (1) OMWW adsorbed onto strawchip
bulking materials in order to obtain a pre-composted material
(CM), thus affecting positively the biochemical characteristics and
making it suitable for living organisms; and (2) CM treatment with
earthworms (Eisenia fetida) and/or oat seedlings (Avena sativa L.)
in order to drive the process forward and production of a nontoxic,
high-quality, biologically active organic fertilizer (E and EP
vermicomposts, respectively).
MATERIALS AND METHODS
Materials
OMWW was obtained from a local olive oil production plant (San
Giuliano Terme, Pisa, Italy), which uses a traditional discontinuous
process for the extraction of olive oil. The cellulose material (chips
and straw) was obtained from municipal prunings. The plants and
earthworms used in the vermicomposting process were Avena
sativa L. and Eisenia fetida, respectively. The characteristics of
the starting materials (chips and straw and OMWW), mixture and
pre-composting material (CM) are reported in Table 1.
Experimental layout
The OMWW were absorbed onto cellulose material (chips and
straw) (OMWW : cellulose material 1 : 1, v/v). The mixture was
incubated in a 100 dm 3 container for 3 months, with the aim of
obtaining a partially stabilized composted material (CM). Then,
500 g of CM was placed in plastic pots (25×15×5 cm) and treated
as follows: (1) CM + 40 adult earthworms (Eisenia fetida)(E);(2)CM
+ 40 adult earthworms (Eisenia fetida) + 100 seeds of Avena sativa
L. (EP); (3) CM alone, used as a control (C).
The experiment was carried out in triplicate under controlled
laboratory conditions (temperature and humidity) for 3 months.
The sampling was done every month; samples were sieved at
www.soci.org G Masciandaro et al.
2 mm and stored at room temperature for chemical and biological
analyses.
Methods
Chemical parameters
Electrical conductivity (EC) and pH were measured in 1 : 10 (w/v)
aqueous solution. Total organic carbon (TOC), total extractable carbon
(TEC) (1 : 10 w/v sodium pyrophosphate solution 0.1 mol L −1
pH 11, 4 h at 60 ◦ C) and water-extractable organic carbon (WSC)
(1 : 10 w/v aqueous solution, 1 h at 60 ◦ C) were determined using
dichromate oxidation. 23 Total Kjeldahl nitrogen (TKN) was
determined by the Kjeldahl method. 24
Nitrate and NH3 were measured in a 1 : 10 (w/v) water extract
by ionic chromatography using a DIONEX chromatograph and by
an ammonia-selective electrode, respectively. Total phenols were
determined using the spectrophotometric method associated
with Folin–Ciocalteu phenol reagent. 25 Specifically, a 0.5 mL
water-extractable phenol or phenol standard (Sigma-Aldrich n.
24–1520 phenol stock solution) was mixed with 3.5 mL sodium
carbonate solution (3.7%), 0.5 mL copper sulfate (0.06%) and
0.5 mL Folin–Ciocalteu reagent. The final solution was left in
the dark for 15 min at 37 ◦ C and afterwards the absorbance of
the solution was measured at 725 nm. A similar procedure was
followed using a distilled-water reagent blank. For instrumental
calibration, standard phenol solutions at different concentrations
were used.
The analysis of monomeric phenolic fractions was done
using gas chromatography–mass spectrometry to evaluate the
concentration of some of the most common phenolic compounds
found in OMWW, 26,27 such as tyrosol, p-hydroxybenzoic acid,
p-hydroxyphenylacetic acid, vanillic acid, protocatechuic acid,
p-coumaric acid and caffeic acid. Samples were mixed in distilled
water (1 : 10 w/v) and shaken for 1 h at 60 ◦ C. After centrifugation,
an aliquot (5 mL) was acidified to pH < 2with1molL −1 HCl.
The solution was washed three times with 5 mL n-hexane in an
ultrasonic bath for 10 min, in order to eliminate small amounts
of triglycerides. 28 The hexane extract (phase) was discharged
every time. The solutions were then extracted three times with
5 mL ethyl acetate in an ultrasonic bath for 15 min. The pooled
organic phase was treated with anhydrous sodium sulfate to
Table 1. Chemical parameters of olive oil mill wastewaters (OMWW), cellulose material, mixture and composted material after 3 months (CM). Mean
of three replicates ± standard deviation
Parameters OMWW a Cellulose Mixture CM
pH 4.8 ± 0.1 6.97 ± 0.22 7.8 ± 0.1 7.2 ± 0.1
EC 8.20 ± 0.33 2.41 ± 0.15 1.23 ± 0.11 2.23 ± 0.20
COD 180 ± 14 – – –
TOC 67.5 ± 5.3 219 ± 7 380 ± 11 360 ± 6
WSC – 18.4 ± 0.1 27.9 ± 1.3 8.74 ± 0.52
TEC – – 60.2 ± 2.3 33.3 ± 2.6
TN 5.0 ± 0.2 7.32 ± 0.37 13.1 ± 0.2 8.45 ± 0.30
NH3 44.2 ± 1.2 51.1 ± 2.3 91.6 ± 2.2 44.3 ± 3.5
C/N 13.5 30 30.4 42.6
Total phenols 10.1 ± 0.7 – 3.20 ± 0.20 1.18 ± 0.08
DH-ase 1.47 ± 0.09 – 120 ± 1 16.8 ± 0.3
EC, electrical conductivity (dS m −1 ); COD, chemical oxygen demand ( a gL −1 ); TOC, total organic carbon ( a gCL −1 and g C kg −1 ); WSC, water-soluble
carbon (g C kg −1 ); TEC, total extractable carbon (g C kg −1 ); TN, total nitrogen ( a gNL −1 and g N kg −1 ); NH3, ammonia ( a mg NH3 L −1 and mg NH3 kg −1 );
total phenols ( a gL −1 and g kg −1 ); DH-ase, dehydrogenase activity ( a mg INTF L −1 h −1 and mg INTF kg −1 h −1 ).
a units for OMWW.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1879–1885

Earthworm treatment to reduce phytotoxicity of pre-composted olive mill wastewater www.soci.org
Table 2. Chemical parameters of treatments (E: only earthworms; EP: earthworms and plants) and control (C) after 30 days (30d), 60 days (60d) and
90 days (90d)
C E EP
30 days 60 days 90 days 30 days 60 days 90 days 30 days 60 days 90 days
pH 7.9 ± 0.3a 7.4 ± 0.2a 7.8 ± 0.2a 8.0 ± 0.1a 7.7 ± 0.2a 8.0 ± 0.1a 7.9 ± 0.4a 7.8 ± 0.2a 8.0 ± 0.1a
EC (dS m−1 ) 2.04 ± 0.08c 2.64 ± 0.09b 1.95 ± 0.10b 2.71 ± 0.11a 2.93 ± 0.09a 2.47 ± 0.10a 2.46 ± 0.07b 2.27 ± 0.11c 2.01 ± 0.12b
TOC
(g kg−1 )
340 ± 2a 364 ± 9a 360 ± 13a 243 ± 20b 261 ± 6b 151 ± 3c 264 ± 21b 250 ± 16b 252 ± 0b
WSC
(g kg−1 )
10.3 ± 0.2a 11.5 ± 0.4a 10.2 ± 0.1a 9.06 ± 0.15b 9.16 ± 0.75b 7.63 ± 0.52b 9.02 ± 0.11b 8.85 ± 0.23c 8.33 ± 0.22b
TN (g kg−1 ) 8.0 ± 0.2b 7.90 ± 0.18b 7.60 ± 0.07a 10.1 ± 0.5a 9.54 ± 0.07a 6.95 ± 0.07b 10.2 ± 0.1a 9.30 ± 0.20a 6.77 ± 0.12b
NH3
(mg kg−1 )
50.3 ± 2.6b 45.6 ± 0.5c 49.1 ± 1.1b 65.1 ± 2.9a 62.0 ± 0.1a 70.1 ± 1.4a 60.5 ± 4.0a 60.0 ± 1.2b 67.5 ± 1.4a
NO3 −
(mg kg−1 )
2.06 ± 0.15a 4.91 ± 0.42c 8.40 ± 0.81c 1.62 ± 0.17b 51.5 ± 3.3a 105 ± 8a 1.55 ± 0.10b 35.6 ± 1.7b 64.7 ± 5.0b
Total
phenols
(g kg−1 )
1.04 ± 0.06c 1.26 ± 0.05a 1.15 ± 0.03a 1.15 ± 0.04b 1.10 ± 0.02b 0.99 ± 0.03b 1.20 ± 0.03a 1.09 ± 0.04b 0.92 ± 0.05c
EC, electrical conductivity; TOC, total organic carbon; WSC, water-soluble carbon; TN, total nitrogen.
Mean of three replicates ± standard deviation. For each time period different letters ( a , b , c ) indicate statistically significant different values among
treatments (P < 0.05).
remove the residue water and evaporated to dryness under a
stream of nitrogen; the residue was derivatized with 50 µL N,Obis(trimethylsilyl)trifluoroacetamide
at 60 ◦ Cfor30min. 29,30 Then,
10 µL hexadecane and 100 µL isooctane as solvent were added.
A2µL volume of the derivatized solution was injected into the
gas chromatograph–mass spectrometer. Linear calibration curves
were obtained in the range of 1–20 µg mL −1 . Chromatographic
analyses were performed using a gas chromatograph model
1800A (Hewlett Packard, Palo Alto, CA, USA), equipped with
a split–splitless injection port and coupled with a quadrupole
mass spectrometric detector with an electron impact of 70 eV.
Chromatographic separation of the analytes was performed on a
chemically bonded fused-silica capillary column HP-5 MS (Hewlett
Packard) (5% phenyl - 95% methyl-polysiloxane, 30 m × 0.25 mm
× 0.25 µm), connected to a 2 m × 0.32 mm deactivated fusedsilica
capillary pre-column. Injector temperature was 250 ◦ Cand
detector temperature was 280 ◦ C. The carrier gas was helium,
(purity: 99.995%), at a constant flow of 1 mL min −1 . The oven
temperature program was as follows: from 80 to 135 ◦ C at
8 ◦ Cmin −1 ; 5 min at 135 ◦ C; from 135 to 140 ◦ Cat1 ◦ Cmin −1 ;
5 min at 140 ◦ C; from 140 to 300 ◦ Cat5 ◦ Cmin −1 . The mass range
was from 45 to 450. The selected ion monitoring (SIM) mode was
used for the quantitative determination of phenols, acquiring the
following ion fragments (m/z): 57 for hexadecane; 179 for tyrosol;
267 for p-hydroxybenzoic acid; 179 for p-hydroxyphenylacetic
acid; 297 for vanillic acid; 193 for protocatechuic acid; 219 for
p-coumaric acid and caffeic acid.
Biological parameters
Dehydrogenase (DH-ase) activity was determined by the method
of Masciandaro et al., 31 using iodonitrotetrazolium chloride (INT)
as substrate and determining spectrometrically (490 nm) the
iodonitrotetrazolium formazan (INTF) product of the reaction. A
germination test was carried out using seeds of Lepidiumsativum 32
and Lolium perenne, 33 two plants that are extremely sensitive to
toxic substances. Ten seeds were tested in Petri dishes using
3 mL of 1 : 10 (w/v) water extract of material from each treatment
and OMWW (controls were tested with distilled water). The Petri
dishes were kept at 20 ◦ C in the dark until germination of seeds.
The germination index (GI) was calculated after 72 h. The index
is calculated as follows: GI% = P(T/C), where P is the mean
percentage of seed germination, and T and C are mean lengths of
the root in the treatment and in the control, respectively.
Statistical analysis
All results are the means of three replicates. One-way analysis
of variance (ANOVA) was performed to test the effect of the
treatments on chemical and biochemical parameters. The means
were compared by using least significant differences calculated at
P < 0.05 (Newman–Keuls test). A statistical correlation between
the data was also calculated for determining the relationship
between chemical and biological parameters (data not reported).
RESULTS
The pH was higher in the mixture with respect to OMWW and
cellulose material. A decrease in carbon and nitrogen compounds,
and in their total (TOC and TN) and soluble (WSC and NH3) forms,
was observed in the composted material (CM) after 3 months
from the mixture preparation; this was a clear indicator that the
more labile fractions of organic matter were being mineralized
by microorganisms, which used these compounds as an energy
source (Table 1). The higher dehydrogenase activity observed
in the mixture with respect to CM confirmed the stimulation of
microbial activity by the huge amount of labile organic compounds
present in it (Table 1). Also, in the vermicomposting processes
(Table 2; E and EP), particularly in the E treatment, TOC and
WSC amounts decreased significantly with respect to the control
treatment (C). On the other hand, N-soluble compounds, NH3 and
NO3 − , increased in the E and EP treatments (Table 2). The NO3 −
content was significantly higher in the E treatment compared to
EP at the end of the experimental period.
The values of TEC (humic matter) increased significantly during
the early period (first 30 days) of the vermicomposting process,
J Sci Food Agric 2010; 90: 1879–1885 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1881

1882
Figure 1. Total extractable carbon (a) and dehydrogenase activity
(b) during the experimental period. (�) Earthworms;(�) earthworms+
plants; (•) control. Vertical bars indicate standard error.
while in C it started to increase after 1 month (Fig. 1(a)). The DH-ase
activity, which decreased over the 3 months of pre-composting
(Table 1; CM), increased rapidly, in particular in the E treatment, in
the first 30 days of vermicomposting experiments. Subsequently,
DH-ase activity decreased until stabilization at the end of the
experiment (Fig. 1(b)).
In order to evaluate the detoxification process of OMWW,
total phenols and monomeric phenolic fractions were detected.
The highest decrease in total phenols (60%) was observed
during the 3 months of the composting process (Table 1). In
the vermicomposting experiments a significant decrease in total
phenols was observed in the two treatments (E and EP), especially
in EP, while in C no significant variation during the same time was
detected (Table 2).
Analysis of monomeric phenolic fractions was carried out on
the water extracts of the samples, so that only soluble phenolic
substances which could have an ecological toxic action
towards microorganisms, plants and groundwater were determined.
A gas chromatographic analysis of the raw OMWW
used in the experiment was performed to obtain a characterization
of monomeric phenolic fractions. All of the expected
phenolic marker compounds 26,27 were found in OMWW (tyrosol,
p-hydroxyphenylacetic acid, protocatechuic acid, caffeic acid, phydroxybenzoic
acid, vanillic acid and p-coumaric acid), but also
other phenolic substances, such as p-hydroxybenzaldehyde, 4hydroxy-3
ymethoxybenzaldehyde and ferulic acid were detected.
p-Coumaric acid was the major compound detected in OMWW,
at a concentration of 20 µgg −1 , while the other phenolic com-
www.soci.org G Masciandaro et al.
pounds had concentrations in the range 1–5 µg g −1 .InCMthe
amount of phenolic monomers decreased, generally, by about
50% if compared to the raw OMWW; in particular, p-coumaric acid
decreased by about 90% (data not reported). In addition, specific
monomers found in OMWW such as p-hydroxybenzaldehyde and
4-hydroxy-3-ymethoxybenzaldehyde, were not present in CM. In
CM, tyrosol had the lowest concentration, even though it is known
as one of the most representative compounds of OMWW. 26,27 At
the end of the vermicomposting experiments, the chromatograms
of E and EP treatments and control samples showed a very low
concentration (

Earthworm treatment to reduce phytotoxicity of pre-composted olive mill wastewater www.soci.org
Figure 2. Reduction percentage of phenolic compounds during vermicomposting experiment in C (control), E (earthworms) and EP (earthworms +
plants) treatments with respect to CM. The data are means of triplicate analysis (coefficient of variation of the three samples ranged from 2% to 10%).
Figure 3. Germination test carried out withLepidiumsativum (a) andLolium
perenne (b) on olive oil mill wastewaters (OMW), mixture, compost material
(CM) and treatments at the end of laboratory experiment. C (control), E
(earthworms) and EP (earthworms + plants. Vertical bars indicate standard
error.
the increase of TEC, especially in the E treatment. In addition,
the negative correlation between DH-ase and TEC (r =−0.87;
P < 0.05) suggested that mineralization and humification are
consecutive processes carried out by microbial activity. The
positive effect of earthworms on the N cycle was shown by the
reduction in total ntrogen and high increase in nitrate. This was
probably due to the capacity of earthworms and microorganisms
to favor aerobic nitrification processes by their movement in the
material, as already found in other studies. 37 – 39
At the end of the experimental period, the lower amount in
nitrates in EP with respect to E treatment was probably due to
plant absorption, thus reducing the risk of NO3 nitrogen leaching
from this organic amendment. Moreover, the presence of plants
positively affected the TOC content through the release of root
exudates and plant remains. 40 As a consequence, vermicompost
with plants showed, after 90 days, a higher C/N ratio (about 30)
with respect to that found in the E treatment (about 20), making
it more suitable for agricultural purposes. 41
A germination test has been carried out in order to evaluate
the feasibility of the final product as an organic amendment.
The results of the germination test showed a large reduction in
toxicity of OMWW at the end of the vermicomposting process.
In fact, the GI of both vermicomposts ranged from 90% to
110%, compared with the 0% shown for OMWW as such.
Therefore, after a vermicomposting process, the OMWW lose
toxicity, thus stimulating seed germination and plant growth.
The stimulation of plant germination and growth could also
be due to the reduction in the total (in particular in EP
treatment) and monomeric phenolic compounds. This result is
in accordance with that of Hachicha et al., 42 who demonstrated a
direct relationship between the polyphenol content and toxicity.
On the other hand, in the control a large increase in concentration
of coumaric and protocatechuic acids was observed; these
compounds represent the last degradation products of phenol
monomers before cleavage of the aromatic ring, 43 so that their
accumulation in the control material could be the result of the
partial degradation of phenolic compounds but without reaching
complete degradation. Also, vanillic acid is one of the main lignin
degradation products 44 and its increase would agree with the
above.
In the E and EP treatments the decrease in phenolic compounds,
suggested the positive effect of earthworms in the degradation
of these recalcitrant compounds through the aeration of material
and their catabolic activity. On the other hand, the mineralization
of lignin–cellulose substrates 45 and the metabolic processes of
plants 46 can induce release of phenolic monomers, thus explaining
the lower reduction in these compounds in EP with respect to E
treatment.
J Sci Food Agric 2010; 90: 1879–1885 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1883

1884
CONCLUSION
Our study found that earthworms, in the presence (EP) or absence
of plants (E), were efficient in the mineralization and stabilization
of OMWW, as shown by the decrease in total organic compounds,
the trend of dehydrogenase activity and the increase in humic
matter. The reduction in OMWW phenolic fractions was promoted
in both E and EP experiments, suggesting the effectiveness of
earthworms in the decrease of OMWW toxicity, as confirmed also
by the increase in germination index. Even if the presence of
plants did not affect phenolic monomer removal, the different
evolutioninC/NratioandNO3 − between E and EP treatments
suggested that the vermicompost obtained with plants could be
more suitable for agricultural purposes owing to a higher C/N ratio
(about 30) and a lower content of nitrates, which are considered
potential contaminants for surface groundwater.
REFERENCES
1 D’Annibale A, Bozzoli V, Crognale S, Gallo AM, Federici F and
Petruccioli M, Optimisation by response surface methodology of
fungal lipase production on olive mill wastewater. J Chem Technol
Biotechnol 81:1586–1593 (2006).
2 Paraskeva P and Diamadopoulos E, Technologies for olive mill
wastewater (OMW) treatment: a review. J Chem Technol Biotechnol
81:1475–1485 (2006).
3 Niaounakis M and Halvadakis CP, Olive Processing Waste Management:
Literature Review and Patent Survey (2nd edn). Elsevier, Amsterdam
(2006).
4 Hachicha S, Sallemi F, Medhioub K, Hachicha R and Ammar E, Quality
assessment of composts prepared with olive mill wastewater and
agricultural wastes. Waste Manage 28:2593–2603 (2008).
5 Baeta-Hall L, Ceu Saagua M, Lourdes Bartolomev M, Anselmo AM and
Fernanda Rosa M, Bio-degradation of olive oil husks in composting
aerated piles. Bioresour Technol 96:69–78 (2005).
6 Casa R, D’Annibale A, Pieruccetti F, Stazi SR, Sermanni GG and Lo
Cascio B, Reduction of the phenolic components in olive-mill
wastewater by enzymatic treatment and its impact on durum wheat
(Triticum durum Desf.) germinability. Chemosphere 50:959–966
(2003).
7 Nogales R,Elvira C,Benítez E, Thompson R and Gómez M, Feasibility of
vermicomposting dairy biosolids using a modified system to avoid
earthworm mortality. J Environ Sci Health B 34:151–169 (1999).
8 Edwards CA, Earthworm Ecology. CRC Press, Boca Raton, FL (2004).
9 Ceccanti B, Garcia C, Masciandaro G, Macci C and Doni S, Soil
bioremediation: combination of earthworms and compost for the
ecological remediation of a hydrocarbon polluted soil. Water Air Soil
Pollut 177:383–397 (2006).
10 Benitez E, Sainz H, Melgar R and Nogales R, Vermicomposting of a
lignocellulosic by-product from olive oil industry: a pilot scale
study. Waste Manage Res 20:134–142 (2002).
11 Mari I, Ehaliotis C, Kotsou M, Balis C and Georgakakis D, Respiration
profiles in monitoring the composting of by-products from the
olive oil agro-industry. Bioresour Technol 87:331–336 (2003).
12 Suthar S, Potential of Allolobophora parva (Oligochaeta) in
vermicomposting. Bioresour Technol 100:6422–6427 (2009).
13 Szczech MM, Suppressiveness of vermicompost against Fusarium wilt
of tomato. J Phytopathol 155–161 (1999).
14 Slocum K, Pathogen reduction in vermicomposting. Worm Digest
23:27–28 (2002).
15 Atiyeh RM, Edwards CA, Subler S and Metzger JD, Pig manure
vermicompost as a component of a horticultural bedding plant
medium: effects on physicochemical properties and plant growth.
Bioresour Technol 78:11–20 (2001).
16 Atiyeh RM, Lees S, Edwards CA, Arancon NQ and Metzger JD, The
influence of humic acids derived from earthworm processed organic
wastes on plant growth. Bioresour Technol 84:7–14 (2002).
17 Chaoui HI, Zibilske LM and Ohno T, Effects of earthworm casts and
compost on soil microbial activity and plant nutrient availability.
Soil Biol Biochem 35:295–302 (2003).
18 Cortez J and Bouclè MB, Do earthworms eat living roots? Soil Biol
Biochem 9:913–915 (1992).
www.soci.org G Masciandaro et al.
19 Babel U, Ehrmann O and Krebs M, Relationships between earthworms
and some plant species in a meadow. Soil Biol Biochem
12:1477–1481 (1992).
20 Milleret R, Le Bayon RC and Gobat JM, Root, mycorrhiza and
earthworm interactions: their effects on soil structuring processes,
plant and soil nutrient concentration and plant biomass. Plant Soil
316:1–12 (2009).
21 Alphei J, Bonkowski M and Scheu S, Protozoa, Nematoda and
Lumbricidae in the rhizosphere of Hordelymus europaeus (Poaceae):
faunal interactions, response of microorganisms and effects on
plant growth. Oecologia 106:111–126 (1996).
22 Atiyeh RM, Lee S, Edwards CA, Arancon NQ and Metzger JD, The
influence of humic acids derived from earthworm-processed
organic wastes on plant growth. Bioresour Technol 84:7–14 (2002).
23 Yeomans JC and Bremner JM, A rapid and precise method for routine
determination of organic carbon in soil. Commun Soil Sci Plant Anal
19:1467–1476 (1988).
24 Jackson ML, Soil Chemical Analysis. Prentice-Hall, Englewood Cliffs, NJ
(1960).
25 Folin O and Ciocalteu V, On tyrosine and tryptophane determinations
in proteins. JBiolChem73:627–650 (1927).
26 Knupp G, Rucker G, Ramos-Cormenzana A, Hoyos GS, Neugebauer M
and Ossenkop T, Problems of identifying phenolic compounds
during the microbial degradation of olive mill wastewater. Int
Biodeterior Biodegradation 38:277–282 (1996).
27 Miranda MA, Galindo F, Amat AM and Arques A, Pyrylium saltphotosensitised
degradation of phenolic contaminants present in
olive oil wastewater with solar light part II. Appl Catal B 30:437–444
(2001).
28 Saitta M, Lo Curto S, Salvo F, Di Bella G and Dugo G, Gas
chromatographic–tandem mass spectrometric identification of
phenolic compounds in Sicilian olive oils. Anal Chim Acta
466:335–344 (2002).
29 Angerosa F, D’Alessandro N, Corana F and Mellerio G, Characterization
of phenolic and secoiridoid algycons present in virgin olive oil
by gas chromatography–chemical ionization mass spectrometry.
J Chromatogr A 736:195–203 (1996).
30 Helaleh MJH, Takabayashi Y, Fujii S and Korenaga T, Gas
chromatographic–mass spectrometric method for separation and
detection of endocrine disruptors from environmental water
samples. Anal Chim Acta 428:227–234 (2001).
31 Masciandaro G, Ceccanti B, Ronchi V and Bauer C, Kinetic parameters
of dehydrogenase in the assessment of the response of soil to
treatments by vermicompost and inorganic fertilisers. Biol Fertil
Soils 32:479–483 (2000).
32 Zucconi F, Pera A, Forte M and De Bertoldi M, Biological evaluation of
compost maturity. Biocycle 22:27–29 (1981).
33 Cheng CY and Chu LM, Phytotoxicity data safeguard the performance
of the recipient plants in leachate irrigation. Environ Pollut
145:195–202 (2007).
34 Tomati U, Galli E, Pasetti L and Volterra E, Bioremediation of olive-mill
wastewaters by composting. Waste Manage Res 13:509–518 (1995).
35 Benitez E, Nogales R, Elvira C, Masciandaro G and Ceccanti B, Enzyme
activities as indicator of the stabilization of sewage sludges
composting of Eisenia foetida. Bioresour Technol 67:297–303 (1999).
36 Wang P, Qu Y and Zhou J, Biodegradation of mixed phenolic
compounds under high salt conditions and salinity fluctuations by
Arthrobacter sp. W1. Appl Biochem Biotechnol 159:623–633 (2009).
37 Parkin TB and Berry EC, Microbial nitrogen transformation in
earthworm burrows. Soil Biol Biochem 31:1765–1771 (1999).
38 Masciandaro G, Ceccanti B and Garcia C, Soil agro-ecological
management: fertirrigation and vermicompost treatments.
Bioresour Technol 59:199–206 (1997).
39 Macci C, Masciandaro G and Ceccanti B, Vermicomposting of olive
oil mill wastewater. Waste Manage Res in press (2010). DOI:
10.1177/0734242X09345278.
40 Pascual JA, Garcia C, Hernandez T,MorenoJLandRosM,Soilmicrobial
activity as a biomarker of degradation and remediation processes.
Soil Biol Biochem 32:1877–1883 (2000).
41 Avnimelech Y and Cohen A, Use of organic manures for amendment of
compacted clay soils. II. Effect of carbon to nitrogen ratio. Commun
Soil Sci Plant Anal 20:1635–1644 (1989).
42 Hachicha S, Cegarra J, Sellami F, Hachicha R, Drira N, Medhioub K, et al,
Elimination of polyphenols toxicity from olive mill wastewater
sludge by its co-composting with sesame bark. J Hazard Mater
161:1131–1139 (2009).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1879–1885

Earthworm treatment to reduce phytotoxicity of pre-composted olive mill wastewater www.soci.org
43 Gallet C and Keller C, Phenolic composition of soil solution:
comparative study of lysimeter and centrifuge waters. Soil Biol
Biochem 31:1151–1160 (1999).
44 Gallet C and Lebreton P, Evolution of phenolic patterns in plants and
associated litters and humus of a mountain forest ecosystem. Soil
Biol Biochem 27:157–165 (1995).
45 Masselter S, Zemann A and Bobleter O, Analysis of lignin degradation
products by capillary electrophoresis. Chromatographia 40:51–57
(1995).
46 Ryan D, Robards K, Prenzier P and Antolovich M, Applications of mass
spectrometry to plant phenols. Trends Anal Chem 18:362–372
(1999).
J Sci Food Agric 2010; 90: 1879–1885 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1885

1886
Research Article
Received: 28 January 2010 Revised: 30 April 2010 Accepted: 2 May 2010 Published online in Wiley Interscience: 22 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4029
Isolation and physicochemical characterisation
of starch from cocoyam (Colocasia esculenta)
grown in Malawi
Davies E Mweta, a∗ Maryke T Labuschagne, a Susanna Bonnet, b
Jannie Swarts b and John D K Saka c
Abstract
BACKGROUND: The aim of this study was to determine the physicochemical properties of starches isolated from Malawian
cocoyams and compare them with those of cassava and corn starches.
RESULTS: The purity of the isolated starches varied from 851 to 947 g kg −1 and pH from 4.93 to 6.95. Moisture, ash, protein, fat
and amylose contents ranged from 104 to 132, 0.3 to 1.5, 3.5 to 8.4, 0.9 to 1.6, and 111 to 237 g kg −1 , respectively. Cocoyam
starches gave higher potassium and phosphorus but lower calcium levels than the other starches. The shape of starch granules
varied from spherical to polygonal with cocoyam starches displaying smaller-sized granules than cassava and corn starches.
Cocoyam starches gave a higher wavelength of maximum iodine absorption and blue value but lower reducing capacity
values than cassava and corn starches. The extent of acid hydrolysis of the starches also differed. Cocoyam starches exhibited
amylopectin molecules of higher molecular weights but amylose molecules of lower molecular weights than cassava and corn
starches. Cocoyam starches exhibited lower water absorption capacity and swelling power, paste clarity and viscosity but
higher solubility, gelatinisation temperatures and retrogradation tendencies than cassava and corn starches.
CONCLUSIONS: The physicochemical properties of native Malawian cocoyam starches vary among the different accessions and
differ from those of cassava and corn starches.
c○ 2010 Society of Chemical Industry
Keywords: cocoyam; cassava; corn; starch; physicochemical properties
INTRODUCTION
Cocoyam (Colocasia esculenta L. Schott), a member of the Araceae
family, is an important tuber crop worldwide mostly grown in
tropical and subtropical countries for its edible corms and leaves.
It is an important food crop in many Pacific Island countries, parts
of Africa, Asia and the Caribbean. Ranking as the fourteenth most
consumed vegetable worldwide, cocoyam is largely produced in
Africa accounting for 60% of the world cocoyam production and
Asia and the Pacific the remaining 40%. 1 In Malawi, cocoyam ranks
third after cassava and sweetpotato and is mainly grown as a food
crop. 2
Cocoyam has a great potential as a source of starch that could
replace commercial starches in various industrial applications.
Its corms are known to have a high content of tiny, easily
digestible, starch grains ranging between 22% and 40% ideal
for use in food and cosmetic formulations, and pharmaceutical
products. 3–5 However corn, potato and wheat still remain the most
predominant sources for the commercial starches. 6 In Malawi, the
industry predominantly depends on imported maize starch to
meet its ever-increasing demand for starch. Therefore, there exists
an opportunity for local Malawians and the country at large to
benefit from the commercialisation of locally grown crops such as
cocoyam through starch production.
Several investigations into the characteristic properties of
cocoyam starches have been conducted and the results of such
investigations have shown that physicochemical properties of
cocoyam starches vary with cultivar. 4,7,8 Comparison studies have
also revealed that cocoyam starches exhibit properties different
from those of starches from other crops. 3,9,10 However, none of
these studies document the characteristic properties of Malawian
cocoyam starches hence limiting their utilisation in the industry.
Since starches from different botanical sources have their own
characteristics, there is a need for determining the characteristic
properties of starches isolated from cocoyam grown in Malawi so
as to unravel their potential and increase their competitiveness on
∗ Correspondence to: Davies E Mweta, Plant Sciences Department, University of
the Free State, P.O. Box 339, Bloemfontein 9300, South Africa.
E-mail: MwetaDE@ufs.ac.za
a Plant Sciences Department, University of the Free State, P.O. Box 339,
Bloemfontein 9300, South Africa
b Chemistry Department, University of the Free State, P.O. Box 339, Bloemfontein
9300, South Africa
c Chemistry Department, Chancellor College, University of Malawi, P.O. Box 280,
Zomba, Malawi
J Sci Food Agric 2010; 90: 1886–1896 www.soci.org c○ 2010 Society of Chemical Industry

Properties of Malawian cocoyam starches www.soci.org
the commercial market. This paper therefore reports the results of
a study undertaken to determine the characteristic properties of
native starch isolated from cocoyam grown in Malawi.
EXPERIMENTAL
Materials
Seven starches from different cocoyam accessions were used in
this study. Cocoyam tubers were harvested from fields of local
farmers in seven different cocoyam growing districts of Malawi:
Chitipa, Mzimba (Mzuzu), Nkhotakota, Machinga, Mulanje, Thyolo
and Zomba in May 2008. Starch was isolated from the fresh tubers
as described by Benesi et al. 11 Starches from five different cassava
varieties (Gomani, Maunjili, Mbundumali, Mkondezi and Sauti)
and commercial corn starch from earlier starch studies in Malawi 11
were used for comparison.
Proximate composition
Moisture content, pH and protein content of the starches
were determined as described by Benesi et al. 11 Fat content
was determined by Soxhlet extraction of 3 g of dried starch
using hexane for 3 h. Ash content was determined by weight
difference after ashing 2–3 g of dried starch samples in a muffle
furnace at 525 ◦ C for 5 h. Amylose content was determined
using an Amylose/Amylopectin Assay Kit (Megazyme International
Ireland Ltd, Bray, Ireland). Purity of the isolated starches was
determined by determining total starch using the Megazyme
enzymatic/colorimetric method. 12
Mineral content
Starch samples for mineral content determination were prepared
following the method of Njoku and Ohia. 13 Phosphorus content
in the samples was determined by the ascorbic acid colorimetric
method. The metal ions sodium (Na), potassium (K), iron (Fe), zinc
(Zn), magnesium (Mg), manganese (Mn) and calcium (Ca) were
analysed on a Varian SpectrAA 300 spectrometer (Varian Techtron
Pty Limited, Mulgrave, Victoria, Australia) using an air–acetylene
flame.
Granular morphology
The granular morphology was studied using a Jeol Scanning
Electron Microscope (JSM-6400, Tokyo, Japan). Starch granules
were mounted on circular aluminium stubs using adhesive, coated
with a thin layer of gold using a Bio-Rad sputter coating system and
then examined at several magnifications and photographed. The
range of the granule size was determined by measuring the length
and width of 150 granules from the pictures. Size distributions of
the starch granules were estimated by classifying the size of starch
granules into four groups: large (>25 µm), medium (10–25 µm),
small (5–10 µm) and very small (

1888
at room temperature for moisture equilibration. The sealed pans
were heated from 20 ◦ Cto95 ◦ C under nitrogen gas at a heating
rate of 10 ◦ Cmin −1 to gelatinise the starch samples. From the DSC
thermograms, the onset temperature (T0), peak temperature (Tp),
conclusion temperature (Tc) and enthalpy of gelatinisation (�HG)
were determined using STARe SW 9.00 software. Temperature
range and peak height index (PHI) were also calculated as Tc − T0
and as the ratio �HG/(Tp − T0), respectively. The gelatinised
samples were stored at 4 ◦ C for 7 days for retrogradation studies.
The pans were then equilibrated at room temperature for 2 h,
and then rescanned in the DSC from 20 to 95 ◦ Cat10 ◦ Cmin −1 to
measure the retrogradation transition temperatures and enthalpy.
The degree of retrogradation was determined as the ratio of
enthalpy change of retrograded starch to enthalpy change of
gelatinised starch. 26
Data analysis
The data was subjected to analysis of variance (ANOVA) using
Statistix 8 for Windows (Analytical Software, Tallahassee, USA).
Pearson correlation coefficients for relationships between various
starch properties were also calculated.
RESULTS AND DISCUSSION
Chemical composition
The purity of the isolated starches determined as total starch
content varied from 851–947 g kg −1 (Table 1). These levels are
lower than those reported for cocoyam (989 g kg −1 ), makal
(970 g kg −1 ) and sorghum starches (933–941 g kg −1 ) in the
literature but comparable with those of maca root starch
(878 g kg −1 ). 10,27–29 On average, corn starch displayed the highest
purity (947 g kg −1 ) than cassava (892 g kg −1 ) and cocoyam
(876 g kg −1 ) starches. The amylose content was the lowest in
Nkhotakota cocoyam starch (106 g kg −1 ) and the highest in
Gomani cassava starch (237 g kg −1 ). Generally, cocoyam starches
displayed lower amylose content than those of cassava and corn
starches. On the contrary, other researchers have reported higher
levels of amylose in cocoyam starch than in cassava starch. 3,9
The moisture content of the starches fell within the prescribed
www.soci.org DE Mweta et al.
industrial specification required for safe storage of starches to
prevent deterioration in starch quality 30 ranging from 104 to
132 g kg −1 for cocoyam starches, 124 to 130 g kg −1 for cassava
starches and 113 g kg −1 for corn starch. The pH of the starches
ranged from 5.18 to 6.95 for cocoyam starches, 4.93 to 5.70 for
cassava starches and 5.98 for corn starch which is within the
acceptable range for low acid food starches. 31
The starches exhibited low levels of ash, protein and fat. The
ash content ranged from 0.30 to 1.57 g kg −1 and was generally
higher in cocoyam starches (1.44 g kg −1 ) than in cassava starches
(1.09 g kg −1 ). Similar trend was reported by Pérez et al.; 10 however,
Nwokocha et al. 9 reported the opposite. Fat content of the starches
ranged from 1.07 to 1.60 g kg −1 within the cocoyam accessions,
0.87 to 1.40 g kg −1 within the cassava cultivar and was 1.30 g kg −1
for corn starch. Cocoyam starches displayed higher protein levels
(4.4–8.4 g kg −1 ) than cassava (3.5–3.9 g kg −1 ) but similar to that
of corn starch (7.2 g kg −1 ).
Phosphorus content of the starches varied from 92.7 to
110.1 mg kg −1 within the cocoyam accessions and 67.3 to
121.3 mg kg −1 among the cassava cultivars (Table 2). These results
fall within the reported phosphorus contents of cocoyam and
cassava starches. 30 Generally, the phosphorus contents of the cocoyam
and cassava starches were comparable but higher than that
of corn starch. Higher levels of potassium were found in cocoyam
than in cassava and corn starches. Calcium content was comparativelyhigherincassavastarchesthanincocoyamandcornstarches.
Magnesium and sodium were present in intermediate levels while
iron, zinc and manganese were present in very low amounts with
manganese being the lowest (Table 2). Levels of minerals obtained
in this study are lower than those reported for canna and arrowroot
starches except for potassium levels that were higher. 32
Iodine absorption spectra and reducing capacity
There was significant variation (P < 0.001) in the iodine absorption
spectra and reducing capacity of the starches (Table 3). Cocoyam
starches displayed higher wavelength of maximum absorption
and blue values than cassava starches. Corn starch exhibited
intermediate values. These results suggest that cocoyam starches
in this study contain longer chain starch molecules than cassava
Table 1. Total starch, pH, moisture (MC), ash, fat, protein, and amylose contents of cocoyam, cassava and corn starches ∗
Botanical
source Accession/genotype
Total starch
(g kg −1 )
Amylose
(g kg −1 )
MC
(g kg −1 ) pH
Ash
(g kg −1 )
Fat
(g kg −1 )
Cocoyam Chitipa 909 bc 161 e 132 a 6.38 de 1.37 abc 1.50 b 5.5 d
Machinga 860 fg 136 f 111 gh 6.32 e 1.53 a 1.33 de 7.1 bc
Mulanje 875 ef 111 g 109 h 6.54 c 1.33 abc 1.60 a 4.5 f
Mzuzu 869 ef 155 e 126 cd 6.46 cd 1.50 ab 1.37 de 4.4 f
Nkhotakota 861 fg 106 g 121 f 6.68 b 1.33 abc 1.07 f 7.0 c
Thyolo 865 fg 161 e 104 i 5.18 j 1.57 a 1.47 bc 8.4 a
Zomba 894 cd 210 b 122 ef 6.95 a 1.43 abc 1.30 e 4.9 e
Cassava Gomani 921 b 237 a 124 de 5.52 hi 1.33 abc 0.87 g 3.7 hi
Maunjili 883 de 177 d 129 ab 4.93 k 1.20 cd 1.40 cd 3.7 h
Mbundumali 903 c 196 c 127 bc 5.43 i 1.27 bcd 1.10 f 3.5 j
Mkondezi 851 g 194 c 128 bc 5.60 gh 0.30 e 1.10 f 3.9 g
Sauti 900 cd 133 f 130 ab 5.70 g 1.33 abc 0.90 g 3.5 ij
Corn – 947 a 202 bc 113 g 5.98 f 1.07 d 1.30 e 7.2 b
Means followed by the same letter (a–k) within the same column are not significantly different from each other (P ≤ 0.05).
∗ Values are means of three determinations.
Protein
(g kg −1 )
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1886–1896

Properties of Malawian cocoyam starches www.soci.org
Table 2. Mineral content (in mg kg −1 ) of the cocoyam, cassava and corn starches ∗
Botanical
source Accession/genotype P Ca K Mg Na Fe Zn Mn
Cocoyam Chitipa 106.9abcd 48.0ef 368.8ab 30.0d 27.0de 9.3cd 2.09bcd 0.41cde Machinga 110.1abc 42.6efg 381.9a 40.3c 21.6e 8.5cde 2.35ab 0.43cd Mulanje 99.1cdef 29.0g 293.7c 30.9d 31.5bcd 6.2ef 1.29g 0.39de Mzuzu 116.2ab 26.0g 346.3b 20.9f 29.1cd 10.7bc 1.78def 0.43cd Nkhotakota 102.5bcde 28.7g 376.9ab 27.2de 26.0de 7.5de 1.76ef 0.39de Thyolo 92.7def 37.2fg 357.4ab 57.1a 51.3a 4.4f 2.54a 0.39de Zomba 105.8bcde 58.7de 288.3c 31.2d 51.9a 7.8de 1.56fg 0.79s Cassava Gomani 108.9abc 124.0b 64.1e 55.0a 57.2a 17.8a 2.11bc 0.75a Maunjili 86.9fg 85.9c 38.4e 47.1b 36.9b 12.2b 1.56fg 0.37de Mbundumali 92.2ef 141.7a 40.5e 4.3 32.3bcd 10.0bcd 2.06bcde 0.39de Mkondezi 67.3h 50.8def 32.0e 23.9ef 29.9bcd 7.9de 1.32g 0.35e Sauti 121.3a 132.3ab 42.6e 47.3b 31.4bcd 16.7a 1.42g 0.52b Corn – 75.9gh 67.1d 219.2d 37.8c 35.5bc 9.9bcd 1.82cdef 0.47bc Means followed by the same letter (a–h) within the same column are not significantly different from each other (P ≤ 0.05).
∗ Values are means of three determinations.
Table 3. Reducing capacity (RC), wavelength of maximum iodine absorption (λmax), blue values (BV) and the extent of acid hydrolysis of the
cocoyam, cassava and corn starches ∗
Extent of acid hydrolysis (%)
Botanical source Accession/genotype λmax (nm) BV Reducing capacity Day 1 Day 2 Day 4 Day 6 Day 8 Day 12 Day 16
Cocoyam Chitipa 605 a 0.365 bc 5.2 fg 2.2 c 3.5 cd 5.0 cd 7.8 a 9.2 cd 22.5 ab 24.8 cd
Machinga 596 c 0.356 c 7.0 e 2.7 a 4.7 a 6.1 b 6.5 bc 10.9 a 23.6 a 25.8 bcd
Mulanje 602 b 0.373 b 6.5 efg 2.2 c 4.2 b 4.9 cde 6.0 cde 9.8 bc 21.3 abc 24.0 de
Mzuzu 594 c 0.333 d 5.0 g 1.6 e 4.0 b 5.5 c 8.2 a 10.1 abc 20.9 bcd 21.5 f
Nkhotakota 588 d 0.334 d 7.4 e 1.9 d 4.0 b 5.0 cd 7.0 b 10.4 ab 20.2 bcde 21.9 ef
Thyolo 603 b 0.372 bc 13.5 b 2.4 b 3.9 b 4.3 efg 7.9 a 10.9 a 19.3 cdef 25.0 bcd
Zomba 607 a 0.408 a 5.6 fg 1.1 f 3.9 b 4.5 def 5.8 de 8.5 de 17.9 efg 20.4 f
Cassava Gomani 574 g 0.250 f 10.8 c 1.0 g 2.6 f 3.9 fgh 6.2 cde 7.9 ef 17.3 fg 24.4 cd
Maunjili 577 fg 0.265 f 6.6 ef 2.2 c 3.1 e 3.5 h 5.7 e 7.6 ef 15.7 g 28.0 a
Mbundumali 579 ef 0.290 e 17.0 a 0.9 h 3.3 de 4.0 fgh 6.5 bc 7.5 f 17.3 fg 27.0 ab
Mkondezi 576 fg 0.255 f 9.3 d 0.9 h 3.0 e 3.7 gh 6.3 cde 8.2 ef 18.6 def 26.5 abc
Sauti 570 h 0.226 g 16.9 a 2.7 a 4.7 a 7.1 a 8.3 a 9.3 cd 21.5 abc 28.1 a
Corn – 582 e 0.296 e 9.5 cd 0.6 i 3.2 e 5.4 c 7.8 a 9.2 cd 17.7 fg 25.1 bcd
Means followed by the same letter (a–h) within the same column are not significantly different from each other (P ≤ 0.05) by LSD test.
∗ All values are means of three replicates of each sample.
and corn starches. 33 Mbundumali cassava starch exhibited the
highest reducing capacity value while Mzuzu cocoyam starch gave
the lowest. Cocoyam starches on average gave lower reducing
capacity values than cassava and corn starches. This indicates the
presence of starch molecules of higher molecular weights than
cassava and corn starches as the reducing number is inversely
related to the molecular weight. These results are consistent
with the differences in iodine binding capacity of the starches
and confirm the existence of structural differences between the
cocoyam, cassava and corn starches. 17
Acid hydrolysis
The results of acid solubilisation revealed a steady increase in
acid hydrolysis in the first 8 days with a large increase observed
between 8 and 12 days of acid hydrolysis (Table 3). This pattern
is different from one reported in literature. 34 The extent of
acid hydrolysis differed significantly (P < 0.001) among the
starches suggesting differences in the packing and orientation of
starch chains in the amorphous regions of the different starches,
however no consistent trends were observed. After 16 days of acid
solubilisation,theextentofhydrolysisvariedfrom20.4%forZomba
cocoyam starch to 28.1% for Sauti cassava starch. This range of
values is lower than those reported for other root and tuber crops
indicating much stronger chain interactions in the amorphous and
crystalline regions of the Malawian cocoyam and cassava starches
compared to other root and tuber crop starches reported. 35
Molecular weight distribution of the starches
The elution profiles of the standards showed a decrease in
retention time with increasing molecular weight and the
J Sci Food Agric 2010; 90: 1886–1896 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1889

1890
(a) Chitipa
(b) Machinga
(c) Mzuzu
(d) Nkhotakota
www.soci.org DE Mweta et al.
(e) Thyolo
(f) Zomba
(g) Gomani
(h) Maunjili
(i) Mbundumali
(j) Mkondezi
(k) Sauti
(l) Corn
Figure 1. Scanning electron micrographs of cocoyam (a–f), cassava (g–k) and corn (l) starch granules.
associated calibration curve was strongly linear (R 2 = 0.9972).
The chromatograms showed that the starches consisted of mainly
two fractions: a higher molecular weight fraction (amylopectin)
with elution time between 12 and 16 min, and a lower molecular
weight fraction (amylose) eluted between 16 and 20 min (Fig. 2).
The molecular weight averages, average molecular weight (Mw)
and number-average molecular weight (Mn), and polydispersity
indexes (PDIs) of the amylopectin and amylose fractions calculated
using the calibration curve varied significantly with source of the
starch (Table 5). Chitipa cocoyam starches exhibited the highest
Mw and Mn values for the amylopectin molecules and Gomani
cassava starch the lowest. In the amylose fraction, Gomani cassava
starch gave the highest Mw and Mn values and Mzuzu cocoyam
starch the lowest. Generally, cocoyam starches exhibited amylose
molecules of lower molecular weight but amylopectin molecules
of higher molecular weight than cassava and corn starches.
The PDI values of amylose molecules were higher for cocoyam
starches than corn and cassava starches; however, the amylopectin
molecules of the starches displayed similar PDI values.
This suggests differences in molecular weight distribution in the
amylose molecules but a similar range of molecular weight distributions
for the high molecular fraction. 21 The differences in the
amylose and amylopectin molecular weight account for observed
differences in reducing capacity values of the different starches.
Morphological properties
Cocoyam, cassava and corn starch granules displayed various
shapes when viewed under scanning electron microscope (Fig. 1).
Cocoyam starches exhibited round/spherical as well as polyhedral
(polygonal) and truncated granules similar to those reported in
literature. 7,10,36,37 Cassava starch granules were mostly rounded
(spherical) and their surfaces were smooth with some portions
being irregular, indicating fissures. Irregular with oval and
truncated, ellipsoidal granules were also observed within the
cassava starch samples. Corn starch granules were polygonal in
shape.
Cocoyam starches displayed smaller sized granules
(9.4–10.4 µm) than cassava (11.4–12.9 µm) and corn starches
(11.8 µm) (Table 4). The size range of cocoyam starches obtained
is higher than those reported (0.05–0.08 µm, 2.96–5.19 µm
and 0.5–5.0 µm) 4,10,37 while that of cassava starches is within
the reported range (3–43 µm). 30 Cassava starches had mostly
medium-sized granules (>60%) with Mbundumali having a larger
fraction of its granules in this size range. Cocoyam starch had
almost equal size distribution of small-sized and medium-sized
granules except for Nkhotakota starch.
Water absorption capacity and swelling power
Water absorption capacity and swelling power were temperature
dependent, increasing with rising temperature (Fig. 3a and b). The
highest water absorption capacity was observed in Sauti, Maunjili
and Mkondezi cassava starches at 50, 70 and 90 ◦ C, respectively,
while the lowest was observed in Mzuzu and Nkhotakota cocoyam
starches at 50 ◦ Cand70 ◦ C, respectively. Corn starch displayed the
lowest water absorption capacity at 90 ◦ C. The same pattern
was observed for swelling power of the starches. Cocoyam
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1886–1896

Properties of Malawian cocoyam starches www.soci.org
Table 4. Granule size and size distribution the starches as determined by scanning electron microscopy
Botanical source Accession/genotype Range Diameter
Granule size (µm) Granule size distribution (%)
Very small
(0–5 µm)
Small
(5–10 µm)
Medium
(10–25 µm)
Cocoyam Chitipa 4.0–18 10.3c 3 45 52 0
Machinga 4.0–16 10.1cd 1 51 48 0
Mulanje 5.3–16 10.4c 0 43 57 0
Mzuzu 5.3–16 9.6de 0 56 44 0
Nkhotakota 5.3–17 9.4e 0 65 35 0
Thyolo 5.3–17 10.1cd 0 51 49 0
Zomba 5.3–18 10.3c 0 51 49 0
Cassava Gomani 6.7–20 11.8b 0 27 73 0
Maunjili 5.3–21 12.0b 0 27 73 0
Mbundumali 6.7–20 12.9a 0 17 83 0
Mkondezi 6.7–22 12.7a 0 23 77 0
Sauti 5.3–20 11.4b 0 33 67 0
Corn 5.3–22 11.8b 0 33 67 0
Means followed by the same letter (a–e) within the same column are not significantly different from each other (P ≤ 0.05)
Large
(>25 µm)
Table 5. The average values and mean separation of average molecular weight (Mw), number-average molecular weight (Mn), and polydispersity
index (PDI) of the amylopectin of the cocoyam, cassava starches and corn starches
Botanical
source
Genotype/
accession
Mw × 10 6
(g mol −1 )
Fraction I (amylopectin) Fraction II (amylose)
Mn × 10 6
(g mol −1 ) PDI
Mw × 10 5
(g mol −1 )
Mn × 10 5
(g mol −1 ) PDI
Cocoyam Chitipa 1.70 a 1.64 a 1.04 a 3.90 de 3.33 def 1.17 abc
Machinga 1.56 abc 1.53 ab 1.02 bc 3.83 def 3.23 def 1.19 ab
Mulanje 1.56 abc 1.51 ab 1.03 ab 3.71 ef 3.15 ef 1.18 abc
Mzuzu 1.52 abc 1.47 ab 1.03 ab 3.62 f 3.04 f 1.19 ab
Nkhotakota 1.66 ab 1.60 a 1.04 a 3.68 ef 3.09 f 2.00 a
Thyolo 1.59 abc 1.56 ab 1.02 bc 4.02 cd 3.50 bcd 1.15 de
Zomba 1.62 abc 1.59 ab 1.02 bc 4.01 cd 3.43 cde 1.17 abc
Cassava Gomani 1.44 c 1.41 b 1.02 bc 4.38 a 3.83 ab 1.14 de
Maunjili 1.57 abc 1.46 ab 1.02 bc 4.24 abc 3.71 abc 1.14 de
Mbundumali 1.49 bc 1.47 ab 1.01 c 4.20 abc 3.74 abc 1.16 bcd
Mkondezi 1.58 abc 1.51 ab 1.04 a 4.36 a 3.87 a 1.16 bcd
Sauti 1.52 abc 1.49 ab 1.02 bc 4.30 ab 3.71 abc 1.13 e
Corn – 1.57 abc 1.52 ab 1.03 ab 4.02 bcd 3.47 cde 1.16 bcd
Means followed by the same letter (a–f) in the same column are not significantly different at P ≤ 0.05.
Figure 2. Typical HPSEC chromatograms of the starches.
starches generally displayed lower water absorption capacity
and swelling power than cassava starches. Corn starch displayed
water absorption and swelling capacity similar to that of cocoyam
starches. The differences in the water absorption capacity could be
due to differences in starch structure giving rise to varying internal
associative forces maintaining granule structure and, degree of
engagement to form hydrogen and covalent bonds between
starch chains and hence the degree of availability of water binding
sites. 38 Differences in granule size could also have contributed to
the observed differences as starches with large granule size swell
rapidly, increasing water retention. 39
Solubility of the starches also increased with rising temperature
(Fig. 3c). Cocoyam starches generally displayed higher solubility
than cassava starches. This pattern disagrees with the water
absorption and swelling behaviour of the cocoyam and cassava
J Sci Food Agric 2010; 90: 1886–1896 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1891

1892
Figure 3. Water absorption capacity, swelling power and solubility of the
cocoyam, cassava and corn starch at 50, 70 and 90 ◦ C. Bars within the
same series labelled with different letters are significantly different from
each other (P = 0.05). (a) Water absorption capacity (g H2O g −1 starch).
(b) Swelling power (g g −1 starch). (c) Solubility (g kg −1 ).
starches.Itisexpectedthathigherswellingshouldresultinincrease
in solubility as higher granule swelling permits exudation of
amylose. The lower solubility of cocoyam starches compared to
cassava starches could therefore be attributed to smaller granule
sizes of the cocoyam starches. 40
Paste clarity and viscosity
Paste clarity and viscosity were the lowest in corn starch and
highest in Mkondezi cassava starch (Fig. 4). Cocoyam starches
exhibited lower paste clarity and viscosity than cassava starches.
Thisisconsistentwithlowerswellingcapacityforcocoyamstarches
than cassava starches observed earlier on. 41 Similar trends in paste
clarity have been reported for cocoyam and cassava starches. 9
It is reported that starches with low amylose content usually
exhibit high paste clarity due to ease of dispersion of amylose
molecules; 42 however, in this study, cocoyam starches displayed
lower amylose contents but lower paste clarity than cassava
www.soci.org DE Mweta et al.
Paste clarity (%T)
Viscosity (cP)
45
40
35
30
25
20
15
10
5
0
16000
14000
12000
10000
8000
6000
4000
2000
0
(a)
f
Chitipa
h
Machinga
(b)
gh
Chitipa
g
Mulanje
gh
Machinga
gh
Mulanje
f
Mzuzu
e
Mzuzu
h
Nkhotakota
f
Nkhotakota
i
Thyolo
h
Thyolo
g
Zomba
gh
Zomba
e
Gomani
b
Gomani
a
Maunjli
d
Maunjli
d
Mbundumali
bc
Mbundumali
b
Mkondezi
a
c
Sauti
c
j
Corn
i
Mkondezi
Sauti
Corn
Figure 4. Paste clarity and viscosity of the cocoyam, cassava and corn
starch pastes at 50 ◦ C. Bars labelled with different letters are significantly
different from each other (P = 0.05). (a) Clarity (%T) ofstarchpastes.
(b) Viscosity (cP) of starch pastes.
starches. The low clarity of cocoyam starch pastes could therefore
be due to the presence of amylose molecules of high susceptibility
to retrogradation. 43
Thermal properties of native and retrograded starches
The thermal properties of the starches varied significantly (P <
0.001) with source (Table 6). Cocoyam starches gave higher values
of gelatinisation temperatures (61.8–83.7 ◦ C) and enthalpies
(8.2–14.7 J g −1 ) than those reported in literature. 44 Cassava
starches had gelatinisation temperatures (58.2–77.5 ◦ C) similar
to those reported for cassava starch (61.55–72.94 ◦ C); however,
the transition enthalpies (13.1–15.1 J g −1 ) were higher than the
reported values of 10.4 J g −1 . 10 Cocoyam starches generally
exhibited higher onset, peak and conclusion temperatures
than cassava and corn starches indicating the presence of
strong bonding forces within the granule interior of cocoyam
starches. 45 These results also confirm structural differences among
the starches under study as low gelatinisation temperatures
are characteristic of starches with larger proportions of short
amylopectin branch chains. 33,44 Cassava starches displayed higher
values of gelatinisation temperature range than cocoyam and
corn starches. This reflects on the differences in the distribution of
starchgranules:themoreheterogeneousthegranules,thebroader
the temperature range. 46 Higher amylopectin content can also
lead to the narrowing of temperature range of gelatinisation. 47
The differences in amylose/amylopectin ratio could also explain
the differences in gelatinisation temperature range as cocoyam
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1886–1896

Properties of Malawian cocoyam starches www.soci.org
Table 6. Thermal properties of the native and retrograded starches: onset (T0), peak (Tp) and conclusion (Tc) temperatures, temperature range (R),
and transition energy (�HG) ∗
Native starches Retrograded starches
Botanical
source Accession/genotype T0 ( ◦ C) Tp ( ◦ C) Tc ( ◦ C) R ( ◦ C) �HG (J g −1 ) T0 ( ◦ C) Tp ( ◦ C) Tc ( ◦ C) �HR (J g −1 )
Degree of
retrogradation (%)
Cocoyam Chitipa 61.8 f 69.4 f 78.6 d 16.8 bc 14.2 abc 45.9 f 56.1 g 62.5 f 3.5 d 24.6 e
Machinga 71.4 b 76.7 b 82.4 b 11.1 f 12.1 e 49.7 e 58.6 f 65.2 bc 5.0 b 41.5 b
Mulanje 70.2 c 75.7 c 82.6 b 12.4 e 14.0 abc 51.2 d 59.3 cde 66.2 a 4.4 c 31.2 d
Mzuzu 68.2 d 73.1 d 79.2 d 11.0 f 8.2 f 52.0 c 59.0 e 65.3 abc 3.8 d 45.7 a
Nkhotakota 75.6 a 78.7 a 83.7 a 8.0 g 14.7 ab 51.8 c 59.7 bc 66.0 ab 5.4 a 37.0 c
Thyolo 67.8 d 73.3 d 80.9 c 13.1 e 14.6 ab 52.5 bc 59.3 de 65.8 ab 4.2 c 28.8 d
Zomba 71.2 b 76.4 b 82.5 b 11.3 f 13.8 bcd 52.9 b 60.4 a 66.1 ab 4.8 b 35.0 c
Cassava Gomani 60.8 g 65.6 i 76.9 efg 16.1 c 14.0 abc 54.3 a 59.3 cde 63.4 ef 1.1 g 7.9 g
Maunjili 58.2 j 67.4 h 77.5 e 19.3 a 15.1 a 54.6 a 59.7 bc 63.9 de 1.6 f 10.6 fg
Mbundumali 62.1 f 68.6 g 76.2 g 14.1 d 13.4 cd 54.4 a 59.7 bc 64.0 de 1.6 f 11.9 f
Mkondezi 60.2 h 65.7 i 76.8 efg 16.6 bc 13.1 cde 54.7 a 59.7 bc 63.7 de 1.4 fg 10.6 fg
Sauti 59.1 i 65.4 i 76.4 fg 17.3 bc 14.1 abc 54.6 a 59.5 cd 63.8 de 1.1 g 7.9 g
Corn – 66.2 e 70.9 e 76.9 ef 10.7 f 12.8 de 54.1 a 60.1 ab 64.6 cd 2.9 e 22.7 e
Means followed by the same letter (a–j) within the same column are not significantly different from each other (P ≤ 0.05).
∗ values are means of three replicates of each sample.
starches generally displayed lower amylose content than cassava
starches, hence higher amylopectin content.
The retrograded starches gave lower gelatinisation temperatures
and enthalpy indicating weaker starch crystallinity (Table 6).
Cocoyam starches displayed higher enthalpies of retrogradation,
consistent with higher levels of retrogradation than cassava and
corn starches. The degree of retrogradation ranged from 24.6 to
45.7%, and 7.9 to 11.9% within the cocoyam and cassava starches,
respectively. Corn starch displayed the degree of retrogradation
of 22.7%. The differences in the retrogradation tendencies of the
starch confirm structural differences among the starches as lower
degree of retrogradation is attributed to higher content of short
branches of amylopectin chains and long amylose molecules. 48
Pearson correlation coefficients between different properties
of starches
The Pearson’s correlation coefficient for the relationship between
various properties of starches from cocoyam, cassava and corn
are presented in Table 7. Granule size exhibited a significant
positive correlation with Ca (r = 0.71, P < 0.01), AM (r = 0.61,
P < 0.05), WAC (r = 0.74, P < 0.01), SP (r = 0.71, P < 0.01),
PV (r = 0.63, P < 0.05), and PC (r = 0.65, P < 0.05), and a
negative correlation with ash (r =−0.65, P < 0.05), K (r =−0.91,
P < 0.01), gelatinisation parameters, T0 and Tp (r =−0.76, −0.78,
P < 0.01), �HR (r =−0.86, P < 0.01), Retro (r =−0.85, P < 0.01)
and BV (r =−0.72, P < 0.01). This agrees with Amani et al. 49
who reported positive correlations of granule size with amylose
content, paste clarity and swelling power for yams grown in Ivory
Coast. Amylose content is one important characteristic that affects
starch functionality. Positive significant correlation of amylose
content with hot-paste viscosity, and negative correlation with
swelling power, solubility and paste clarity have been reported
for potato and amaranthus starches. 50,51 However, in this study,
no significant correlation was observed between amylose content
and the functional properties of the starches. Similar observations
were reported by Wickramasinghe et al. 3 who found no significant
correlation between amylose content and thermal and pasting
characteristics of starches isolated from root and tuber crops
grown in Sri Lanka. Ash content showed a positive correlation
with P (r = 0.78, P < 0.01) and K contents (r = 0.56, P < 0.05).
Ca had a significant positive correlation with WAC, SP, PV and PC
but a negative correlation with gelatinisation temperatures (T0
and Tp) and degree of retrogradation whereas potassium showed
positive correlation with gelatinisation temperatures and degree
of retrogradation but a negative correlation with WAC, SP, PV
and PC. Contrarily, Zaidul et al. 52 reported a negative correlation
between peak viscosity determined by RVA and calcium content
but positive correlation of the same with potassium content for
potato starches.
There were significant inter-relationships between gelatinisation
and retrogradation parameters. T0 and Tp were positively
correlated (r = 0.97, P < 0.01) and both were also positively
correlated with Tc (r = 0.90, 0.94, P < 0.01). Similar positive
relationships between T0, Tp and Tc have been reported for
amaranthus 51 and corn starches. 53 T0, Tp and Tc were all positively
correlated with retrogradation parameters, �HR (r = 0.94,
0.97, 0.94, P < 0.01) and Retro (r = 0.88, 0.90, 0.80, P < 0.01);
however, they were not significantly correlated with �HG. Weak
relationship between �HG with T0, Tp and Tc or lack of it is attributed
to the fact that enthalpy of gelatinisation is essentially due
to hydrogen bonding not crystallinity. 54 T0, Tp and Tc were negatively
correlated with WAC (r =−0.98, −0.95, −0.90, P < 0.01),
SP (r =−0.97, −0.95, −0.88, P < 0.01), PV (r =−0.65, −0.72,
−0.61, P < 0.05; r =−0.72, P < 0.01) and PC (r =−0.77, −0.73,
−0.61, P < 0.01). SP, WAC, PC and PV were all positively correlated
with each other.
CONCLUSIONS
The study has revealed differences in physicochemical properties
of starches from different cocoyam accessions and with those of
cassava and corn starches. Cocoyam starches generally exhibited
higher pH, ash, fat, protein, potassium and phosphorus levels but
lower amylose and calcium levels than cassava starches. Cocoyam
J Sci Food Agric 2010; 90: 1886–1896 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1893

1894
Table 7. Pearson correlation coefficients for physicochemical properties of the cocoyam, cassava and corn starches
GS Ash P Ca K Am T0 Tp Tc �HG �HR Retro WAC SP PV PC
www.soci.org DE Mweta et al.
Ash −0.65∗ – – – – – – – – – – – – – –
P −0.66∗ 0.78∗∗ – – – – – – – – – – – – – –
Ca 0.71∗∗ −0.07 −0.18 – – – – – – – – – – – – –
K −0.91∗∗ 0.56∗ 0.58∗ −0.80∗∗ – – – – – – – – – – – –
Am 0.61∗ −0.29 −0.05 0.48 −0.50 – – – – – – – – – – –
T0 −0.76∗∗ 0.42 0.41 −0.67∗ 0.81∗∗ −0.47 – – – – – – – – – –
Tp −0.78∗∗ 0.50 0.45 −0.70∗∗ 0.84∗∗ −0.52 0.97∗∗ – – – – – – – – –
Tc −0.81∗∗ 0.45 0.52 −0.71∗∗ 0.79∗∗ −0.56∗ 0.90∗∗ 0.94∗∗ – – – – – – – –
�HG 0.22 −0.07 −0.07 0.27 −0.24 0.01 −0.18 −0.13 0.05 – – – – – – –
�HR −0.86∗∗ 0.50 0.51 −0.78∗∗ 0.93∗∗ −0.52 0.94∗∗ 0.97∗∗ 0.94∗∗ −0.13 – – – – –
Retro −0.85∗∗ −0.50 0.50 −0.78∗∗ 0.90∗∗ −0.48 0.88∗∗ 0.90∗∗ 0.80∗∗ −0.50 0.92∗∗ – – – – –
WAC 0.74∗∗ −0.38 −0.38 0.67∗ −0.80∗∗ 0.47 −0.98∗∗ −0.95∗∗ −0.90∗∗ 0.09 −0.93∗∗ −0.83∗∗ – – – –
SP 0.71∗∗ −0.36 −0.35 0.58∗ −0.77∗∗ 0.46 −0.97∗∗ −0.95∗∗ −0.88∗∗ 0.09 −0.91∗∗ −0.82∗∗ 1.00∗∗ – – –
PV 0.63∗ −0.48 −0.31 0.62∗ −0.81∗∗ 0.34 −0.65∗ −0.72∗∗ −0.61∗ 0.02 −0.78∗∗ −0.68∗ 0.72∗∗ 0.71∗∗ – –
PC 0.65∗ −0.46 −0.39 0.58∗ 0.84∗∗ 0.20 −0.77∗∗ −0.73∗∗ −0.61∗ 0.16 −0.77∗∗ −0.70∗∗ 0.80∗∗ 0.88∗∗ 0.84∗∗ –
BV −0.72∗∗ 0.52 0.62∗ −0.69∗∗ −0.85∗∗ −0.28 0.75∗∗ 0.83∗∗ 0.80∗∗ −0.08 0.88∗∗ 0.79∗∗ −0.73 −0.70∗∗ −0.76∗∗ −0.67∗ www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1886–1896
GS, granule size; P, phosphorus; Ca, calcium; K, potassium; Am, amylose; T0, onset gelatinisation temperature; Tp, peak gelatinisation temperature; �HG, enthalpy of gelatinisation; �HR, enthalpy of
retrogradation; Retro, degree of retrogradation; WAC, water absorption capacity at 70 ◦ C; SP, swelling power at 70 ◦ C, PV, apparent viscosity; PC, paste clarity; BV, blue value.
∗ Significant at P < 0.05, and ∗∗ significant at P < 0.01.

Properties of Malawian cocoyam starches www.soci.org
starches displayed smaller-sized granules than cassava and corn
starches. Cocoyam starches have different structures from cassava
and corn starches as evidenced by differences in the wavelength
of maximum iodine absorption, blue value, reducing capacity,
extent of acid hydrolysis and the carbohydrate profiles. Cocoyam
starches had higher gelatinisation temperatures and solubility
but lower water absorption capacity and swelling power, paste
clarity and viscosity but higher solubility than cassava and corn
starches. Cocoyam starches also displayed higher retrogradation
tendencies than cassava starches.
ACKNOWLEDEGMENTS
The authors would like to thank International Programme in
Chemical Sciences (IPICS), Uppsala University in Sweden for
financing this research through the Malawi 01 ‘Genetics and
Chemistry of Root and Tuber Crops’ project.
REFERENCES
1 FAO Statistics, Food and Agriculture Organization, Data base
results, (2003).
2 Sandifolo VS, Paper 15: Estimation of crop losses due to
different causes in root and tuber crops: The case of
Malawi, Proceedings of the Expert Consultation on root crop
statistics, Harare, Zimbabwe, 3–6 December (2002). Available:
http://www.fao.org/docrep/005/Y9422E/y9422e0f.htm [12 June
2007].
3 Wickramasinghe HAM, Takigawa S, Matsuura-Endo C, Yamauchi H
and Noda T, Comparative analysis of starch properties of different
root and tuber crops of Sri Lanka. Food Chem 112:98–103 (2009).
4 Moorthy SN, Pillai PKT and Unnikrishnan M, Variability in starch
extracted from taro. Carbohydr Polym 20:169–173 (1993).
5 Subhashirasakul S, Yuenyoungsawad S, Ketjinda W, Phadoongsombut
N and Faroongsang D, Study on tablet binding and disintegrating
properties of alternative starches prepared from taro and
sweetpotato tubers. Drug Dev Ind Pharm 27:81–87 (2001).
6 Ellis RP, Cochrane MP, Dale MFB, Duffus CM, Lynn A, Morrison IM, et al,
Starch production and industrial use. J Sci Food Agric 77:289–311
(1998).
7 Jane J, Shen L, Chen J, Lim S, Kasemsuwan T and Nip WK, Physical
and chemical studies of taro starches and flours. Cereal Chem
69:528–535 (1992).
8 Aboubakar • Njitang YN, Scher J and Mbofung CMF, Physicochemical,
thermal properties and microstructure of six varieties of taro
(Colocasia esculenta L. Schott) flours and starches, J Food Eng
86:294–305 (2008).
9 Nwokocha LM, Aviara NA, Senan C and Williams PA, A comparative
study of some properties of cassava (Manihot esculenta, Crantz)
and cocoyam (Colocasia esculenta, Linn) starches. Carbohydr Polym
76:362–367 (2009).
10 Pérez E, Schultz FS and de Delahaye EP, Characterisation of some
properties of starches isolated from Xanthosoma sagittifolium
(tannia)and Colocasiaesculenta(taro). CarbohydrPolym60:139–145
(2005).
11 Benesi IRM, Labuschagne MT, Dixon AGO and Mahungu NM, Stability
of native starch quality parameters, starch extraction and root dry
matter of cassava genotypes in different environments. J Sci Food
Agric 84:1381–1388 (2004).
12 McCleary BV, Solah V and Gibson TS, Quantitative measurement of
total starch in cereal flours and products. J Cereal Sci 20:51–58
(1994).
13 Njoku PC and Ohia CC, Spectrophotometric estimation studies of
mineral nutrient in three cocoyam cultivars. Pak J Nutr 6:616–619
(2007).
14 Lindeboom N, Chang PR and Tyler RT, Analytical, biochemical and
physicochemical aspects of starch granule size, with emphasis on
small granule starches: A review. Starch/Stärke 56:89–99 (2004).
15 Singh N, Inouchi N and Nishinari K, Morphological, structural, thermal,
and rheological characteristics of starches separated from apples
of different cultivars. J Agric Food Chem 53:10193–10199 (2005).
16 Hizukuri S, Takeda Y, Yasuda M and Suzuki A, Multibranched nature
of amylose and the action of debranching enzymes. Carbohydr Res
94:205–213 (1981).
17 Van Hung P and Morita N, Physicochemical properties and enzymatic
digestibility of starch from edible canna (Canna edulis) grownin
Vietnam. Carbohydr Polym 61:314–321 (2005).
18 Jayakody L, Hoover R, Liu Q and Donner E, Studies on tuber starches II.
Molecular structure, composition and physicochemical properties
of yam (Dioscorea sp.) starches grown in Sri Lanka. Carbohydr Polym
69:148–163 (2007).
19 Brook JR, Griffin VK and Kattan W, A modified method for total
carbohydrate analysis of glucose syrups, maltodextrins and other
starch hydrolysis products. Cereal Chem 63:465–467 (1986).
20 Chung Y-L and Lai H-M, Molecular and granular characteristics of
corn starch modified by HCl–methanol method at different
temperatures. Carbohydr Polym 63:527–534 (2006).
21 Lehtonen P, Effect of method of solution on observed molecular
weight distribution of barley starch and starch derivatives in size
exclusion chromatography. Chromatographia 26:157–159 (1988).
22 Mishra S and Rai T, Morphology and functional properties of corn,
potato and tapioca starches. Food Hydrocolloids 20:557–566 (2006).
23 Kojima M, Shimizu H, Ohashi M and Ohba K, Physico-chemical
properties and digestibility of pulse starch after four different
treatments. J Appl Glycosci 53:105–110 (2006).
24 Craig SAS, Maningat CC, Seib PA and Hoseney RC, Starch paste clarity.
Cereal Chem 66:173–182 (1989).
25 International Starch Institute, ISI 17-1e: Determination of
viscosity of starch by Brookfield Viscometer (2002). Available:
http://www.starch.dk/isi/methods/17brookfield.htm [17 January
2008].
26 Peroni FH, Rocha TS and Franco CM, Some structural and
physicochemical characteristics of tuber and root starches. Food
Sci Technol Int 12:505–512 (2006).
27 Rondán-Sanabria GG and Finardi-Filho F, Physical–chemical and
functional of maca root starch (Lepidium meyenii Walpers). Food
Chem 114:492–498 (2009).
28 Boudries N, Belhaneche N, Nadjemi B, Deroanne C, Mathlouthi M,
Roger B, et al, Physicochemical and functional properties of starches
from sorghum cultivated in the Sahara of Algeria. Carbohydr Polym
78:475–480 (2009).
29 Torruco-Uco JG and Betancur-Ancona D, Physicochemical and
functional properties of Makal (Xanthosoma yucatanensis) starch.
Food Chem 101:1319–1326 (2007).
30 Moorthy SN, Physicochemical and functional properties of tropical
tuber starches: A review. Starch/Stärke 54:559–593 (2002).
31 Moore CO, Tuschhoff JV, Hastings CW and Schanefelt RV, Applications
of starches in foods, in Starch: Chemistry and Technology, ed.
by Whistler RL, Bemiller JN and Pascall EF. Academic Press Inc.,
Orlando, FL, pp. 575–591 (1984).
32 Pérez E and Lares M, Chemical composition, mineral profile, and
functional properties of canna (Canna edulis) and arrowroot
(Maranta spp)starches.Plant Food Hum Nutr 60:113–116 (2005).
33 Tetchi FA, Rolland-Sabaté A, Amani NG and Colona P, Molecular and
physicochemical characterization of starches from yam, cocoyam,
cassava, sweetpotato and ginger produced in Ivory Coast. J Sci Food
Agric 87:1906–1916 (2007).
34 Hoover R, Acid-treated starches, Food Rev Int 16:369–392 (2000).
35 Jayakody L, Hoover R, Liu Q and Weber E, Studies on tuber and root
starches. I. Structure and physicochemical properties of innala
(Solenostemon rotundifolius) starches grown in Sri Lanka. Food Res
Int 38:615–629 (2005).
36 Huang C-C, Physicochemical, pasting and thermal properties of tuber
starches as modified by guar gum and locust bean gum. Int J Food
Sci Technol 44:50–58 (2008).
37 Sefa-Dedeh S and Kofi-Agyir Sackey E, Starch structure and some
properties of cocoyam (Xanthosoma sagittifolium and Colocasia
esculenta) starch and raphides. Food Chem 79:435–444 (2002).
38 Hoover R and Sosulski F, Effect of cross linking on functional properties
of legume starches. Starch/Stärke 38:149–155 (1986).
39 Akalu G, Tufvesson F, Jonsson C and Nair BM, Physicochemical
characteristics and functional properties of starch and dietary fibre
in grass pea seeds. Starch/Stärke 50:374–382 (1998).
40 Tian SJ, Rickard JE and Blanshard JM, Physicochemical properties of
sweet potato starch. J Sci Food Agric 57:451–491 (1991).
41 Hoover R, Sailaja Y and Solsuski F, Characterization of starches from
wild and long grain brown rice. Food Res Int 29:99–107 (1996).
J Sci Food Agric 2010; 90: 1886–1896 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1895

1896
42 Wang LZ and White PJ, Structure and physicochemical properties
of starches from oats with different lipid content. Cereal Chem
71:443–450 (1994).
43 Mélo EA, Stamford TLM, Silva MPC, Krieger N and Stamford NP,
Functional properties of yam bean (Pachyrizus erosus) starch.
Bioresource Technol 89:103–106 (2003).
44 Jane J, Chen YY, Lee LF, McPherson AE, Wong KS and Radosavljevic M,
Effects of amylopectin branch chain length and amylose content
on the gelatinization and pasting properties of starch. Cereal Chem
76:629–637 (1999).
45 Tester RF and Morrison WR, Swelling and gelatinization of cereal
starches. I. Effects of amylopectin, amylose and lipids. Cereal Chem
60:551–557 (1990).
46 Sandhu KS, Singh N and Malhi NS, Physicochemical and thermal
properties of starches separated from corn produced from crosses
of two germ pools. Food Chem 89:541–548 (2005).
47 Krueger BR, Knutson CA, Inglett GE and Walker CE, A differential
scanning calorimetry study on the effect of annealing on
gelatinization behaviour of corn starch. J Food Sci 52:715–718
(1987).
48 Spence KE and Jane J, Chemical and physical properties of ginkgo
(Ginkgo biloba)starch.Carbohydr Polym 40:261–269 (1999).
www.soci.org DE Mweta et al.
49 Amani N’GG, Buleon A, Kamenan A and Collona P, Variability in starch
physicochemical and functional properties of yam (Dioscorea sp.)
cultivated in Ivory Coast. J Sci Food Agric 84:2085–2096 (2004).
50 Kaur A, Singh N, Ezekiel R and Guraya HS, Physicochemical, thermal
and pasting properties of starches separated from different potato
cultivars grown at different locations. Food Chem 101:643–645
(2007).
51 Kong X, Bao J and Corke H, Physical properties of Amaranthus starch.
Food Chem 113:371–376 (2009).
52 Zaidul ISM, Norulaini N, Omar MOK, Yamauchi H and Noda T,
Correlations of the composition, minerals, and RVA pasting
properties of various potato starches. Starch/Stärke 59:269–276
(2007).
53 Sandhu KS and Singh N, Some properties of corn starches II:
Physicochemical, gelatinisation, retrogradation, pasting and gel
textural properties. Food Chem 101:1499–1507 (2007).
54 Singh J, McCarthy OJ and Singh H, Physico-chemical and
morphological characteristics of New Zealand Taewa (Maori potato)
starches. Carbohydr Polym 64:569–581 (2006).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1886–1896

Research Article
Received: 24 December 2009 Revised: 30 April 2010 Accepted: 04 May 2010 Published online in Wiley Interscience: 14 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4030
A 75% reduction in herbicide use through
integration with sorghum+sunflower extracts
for weed management in wheat
Muhammad Naeem Mushtaq, ∗,† Zahid Ata Cheema, Abdul Khaliq
and M Rashid Naveed
Abstract
BACKGROUND: To reduce herbicide use by 75%, integrated use of sorghum and sunflower extracts each at 18 L ha −1 combined
with 1/4 th (75% less) of label rates of four herbicides (mesosulfuron+idosulfuron, metribuzin, phenoxaprop-p-ethyl and
isoproturon) were investigated for the management of wild oat and canary grass, the two pernicious weeds in wheat fields
worldwide.
RESULTS: The results revealed that sorghum+sunflower extracts combined with 1/4 th (75% less) of label rates of herbicides
inhibited dry matter production of wild oat by up to 89% and canary grass by up to 92%. The wild oat and canary grass
persistence index in sorghum+sunflower extracts combined with 1/4 th (75% less) of label rates of herbicides was either lower or
equal to respective label rates of herbicides, except sorghum+sunflower extract+1/4 th phenoxaprop-p-ethyl. Lower herbicide
rates+water extracts also produced wheat grain yield statistically equal with label rates of respective herbicides. Two treatments
having water extracts+lower herbicides rates were economical and sorghum+sunflower+1/4 th mesosulfuron+idosulfuron
produced the highest (4404%) marginal rate of return.
CONCLUSION: Herbicides use can be reduced by 75% through integration with sorghum+sunflower extracts without
compromising yield and net benefits for cost-effective and eco-friendly management of wild oat and canary grass in wheat.
c○ 2010 Society of Chemical Industry
Keywords: wheat; reduce herbicide use; integrated weed management; allelopathy; Avena fatua; Phalaris minor
INTRODUCTION
Crop production has been facing a menace from weeds in terms
of growth and yield reduction since the start of agriculture. Weeds
reduce the yield of all crops by competing for growth factors
and inhibiting crop growth through release of phytotoxins. 1
With the advent of the green revolution in the 1960s and
increased demand for food, herbicide use has been increased
many fold in the last few decades. Of the total pesticides
used worldwide, herbicides accounts for 37%, while 24%, 9%
and 29% are insecticides, fungicides and others, respectively. 2
Although herbicides are efficient for weed control, continuous
use has caused the development of resistance in weeds against
several herbicides. Presently, 332 herbicide resistant biotypes
including 189 species (113 broad-leaved and 76 grasses) have
been reported in more than 0.30 million fields worldwide. 3
Furthermore, herbicides also pollute the soil, water and aerial
environments and herbicide residues in food have deteriorated
food quality and enhanced the risk of disease. 4 Therefore, there
is growing public as well as scientific concern about the use
of herbicides. Although herbicide use cannot be completely
eliminated due to dire need of food for an ever increasing
world population, the use of herbicides may be reduced
through using integrated weed management approaches in
field crops. 5
Allelopathy, a phenomenon in which one plant inhibits growth
of other plants through release of allelochemicals, has been
accepted as a viable option in recent years, for reducing herbicide
use to obtain eco-friendly and cost-effective weed control. 6,7
Combined use of allelopathic extracts (sorghum, sunflower or
brassica) and lower rates of herbicides has successfully been
reduced the herbicides use by up to 50% in field crops, mung bean
(Vigna radiata) and cotton (Gossypium hirsutum). 8,9
Sorghum and sunflower are well studied allelopathic crops. 10
Many allelochemicals responsible for inhibitory action of these
two crops have been identified. Sorghum contains gallic acid,
protocateuic acid, syringic acid, vanillic acid, p-hydroxybenzoic
acid, p-coumaric acid, benzoic acid, ferulic acid, m-coumaric
acid, caffeic acids, p-hydroxybenzaldehyde and sorgoleone. 11,12
Sunflower possesses chlorogenic acid, isochlorogenic acid, αnaphthol,
scopolin and annuionones. 13–15
∗ Correspondence to: Muhammad Naeem Mushtaq, Department of Agronomy,
University of Agriculture, Faisalabad 38040, Pakistan.
E-mail: mnmushtaq@gmail.com
† Presentaddress:GraduateSchoolofLifeandEnvironmentalSciences,University
of Tsukuba, Tsukuba 305-8572, Ibaraki, Japan.
DepartmentofAgronomy,UniversityofAgriculture,Faisalabad38040,Pakistan
J Sci Food Agric 2010; 90: 1897–1904 www.soci.org c○ 2010 Society of Chemical Industry
1897

1898
A significant quantity of total herbicides is used for weed
control in wheat crops worldwide, while in Pakistan, 63% of the
total herbicides are used for weed control in wheat (M. Ashiq,
personal communication). Wild oat and canary grass are the two
pernicious weeds of wheat worldwide, each causing a reduction
in yield of, on average, 30%. 16–18 Herbicide use can be reduced
by up to 50% through combined use of herbicide and 250%
sorghum extract at 12 L ha −1 for weed control in wheat. 19 There
is a need to further reduce herbicide use for sustainable and
economical weed control. Herbicide use may be further decreased
by following two approaches: (1) increasing rate of allelopathic
extracts, and (2) using a mixture of two allelopathic crop extracts.
In this research, these approaches were combined to explore the
possibility of reducing herbicide use by 75%. Recently, sorghum
and sunflower allelopathic extracts have shown inhibitory effects
against grassy weeds. 7 The authors are not aware of any report
on integrated use of sorghum+sunflower extracts and 1/4 th (75%
less) of label rates of herbicides for control of wild oat and canary
grass in wheat. The present study was therefore conducted to
investigate the possibility of reducing herbicides use by 75%,
when employed in combination with sorghum (Sorghum bicolor
L. Moench) and sunflower (Helianthus annuus L.) extracts for wild
oat and canary grass management in wheat (Triticum aestivum
L.) fields.
EXPERIMENTAL
Site and crop husbandry
Two years field experiment was conducted during 2006–2007 and
2007–2008 at Agronomic research Area, University of Agriculture,
Faisalabad, Pakistan (31.25 ◦ N, 73.09 ◦ E, and 184 m) where a history
of the field showed infestation of wild oat and canary grass. Wheat
(Triticum aestivum L.) cultivar Auqab-2000 was sown in winter on
30 November 2006 and 5 December 2007 during the first and
second years, respectively. A soaking irrigation was applied before
preparing the seed bed. The crop was planted in 22 cm between
rows with a single row hand-drill using a seed rate of 125 kg
ha −1 . The plot size was 2.2 m × 5 m. A basal dose of nitrogen
and phosphorus (100 kg N and 115 kg P2O5 ha −1 ) was applied
at the time of sowing in the form of urea and diammonium
phosphate, respectively. The whole quantity of phosphorus and
half of nitrogen was broadcast at the time of sowing, while the
remaining half of the nitrogen was top dressed 50 days after
sowing (DAS). The first irrigation was applied 25 DAS, while
subsequent irrigations were managed as and when needed by
the crop. The crop was harvested manually 135 and 130 DAS
during the first and second years, respectively.
Preparation of extracts
Sorghum (Sorghum bicolor L. Moench) and sunflower (Helianthus
annuus L.)above-groundbiomass were harvestedfrom Agronomic
Research Area, University of Agriculture, Faisalabad, Pakistan. The
herbage of both crops were dried and stored under shade to
avoid possible leaching by rain water. Both plant materials were
chopped with an electric fodder cutter into pieces of 2–3 cm
each. The chopped material was soaked in water (100 g L −1 )at
room temperature (21 ◦ C ± 2 ◦ C) for 24 h. The soaked material
was filtered through 10 and 60 mesh sieve to obtain the extract.
These extracts were boiled at 100 ◦ C to reduce the extract volume
by 95% for easy handling and field application. After boiling, the
concentration of water extracts was 250%. The boiling did not
affect nature, concentration and efficacy of allelochemicals. 7,20,21
www.soci.org MN Mushtaq et al.
Application of treatments
Sorghum and sunflower water extracts each at 18 L ha −1 were
combined with 1/4 th of label rates of four herbicides, viz.
mesosulfuron+idosulfuron (Atlantis 3.6-WG) at 3.6 g a.i. ha −1 ,
metribuzin (Sencor 70-WP) at 43.75 g a.i. ha −1 , phenoxaprop-pethyl
(Puma Super 7.5-EW) at 23.44 g a.i. ha −1 and isoproturon
(Partner 50-WP) at 250 g a.i. ha −1 at the time of application.
Full rates of herbicides as mesosulfuron+idosulfuron at 14.4 g
a.i. ha −1 , metribuzin at 175 g a.i. ha −1 , phenoxaprop-p-ethyl at
93.75 g a.i. ha −1 and isoproturon at 1000 g a.i. ha −1 were used as
standard. A control (weedy check) was maintained for comparison.
All treatments including water extracts+1/4 th rates of herbicides
and label rates of herbicides were sprayed in the respective plots
35 DAS. The volume of the spray (330 L ha −1 ) was determined by
calibration. 22 Spraying was done with a Knapsack hand-sprayer
fitted with a T-Jet nozzle maintaining a pressure of 207 kPa. All
treatments were applied during both years of experimentation.
Data collection
Data on wild oat and canary grass density and dry matter
production were recorded 70 DAS. A quadrate measuring
0.50 × 0.50 m was placed randomly at two places in each
experimental unit. The above-ground biomass of both weeds
inside the quadrates was harvested. After counting, weeds were
placed in an oven at 70 ◦ C for 72 h to obtain the dry weight. The
data of two quadrates were averaged. Weed persistence index
(WI) was calculated by following equation:
WI =
Weed density of control Weed dry matter of treatment
×
Weed density of treatment Weed dry matter of control
The number of fertile tillers of wheat was counted at the time of
harvesting. A quadrate measuring 1 m × 1mwasplacedrandomly
in each plot. The fertile tillers inside the quadrate were counted. A
spike-bearing tiller was considered as fertile. For number of grains
per spike, ten plants were selected at random from each plot. Each
spike was threshed individually to count the number of grains. The
data of ten plants were averaged. After manual harvesting, the
grain yield of each plot was weighed in kilograms and expressed as
mega gram per hectare (Mg ha −1 ), then 1000 grains were selected
randomly from the yield of each plot and weighed.
Statistical and economic analyses
The experiment was conducted in a randomised complete block
design with four replications. ANOVA was performed using Fisher’s
analysis of variance technique while multiple comparison among
treatment means was made using least significance difference test
at P < 0.05. 23 Economic and marginal analyses were carried out
using the method devised by CIMMYT. 24
RESULTS
Effect of various treatments on density and dry matter
production of wild oat and canary grass
All treatments significantly (P < 0.05) reduced the density of
wild oat and canary grass as compared with control (Table 1).
Sorghum+sunflower water extracts each at 18 L ha −1 combined
with 1/4 th (75% less) of label rates of herbicides inhibited wild oat
density by 25–75% and 70–74% during the first and second years
of experimentation, respectively. The label (full) rates of herbicides
reduced density of wild oat by 38–75% and 79–81% during the
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1897–1904

Integrated use of allelopathic extracts reduce herbicide use www.soci.org
Table 1. Effect of sorghum+sunflower extracts combined with 1/4 th rates of herbicides on density of wild oat and canary grass
Density (m 2 )
Wild oat Canary grass
Extract/herbicide Rate (ha −1 ) 2006–2007 2007–2008 2006–2007 2007–2008
Control (weedy check) – 2.00 a 10.75 a 11.7 a 9.75 a
Sorghum+sunflower+mesosulfuron+idosulfuron (Atlantis 3.6-WG) Each at 18 L+3.6 g a.i. 0.50 c (75) 2.75 bc (74) 2.75 e (76) 5.25 b (46)
Sorghum+sunflower+metribuzin (Sencor 70-WP) Each at 18 L+43.75 g a.i 1.50 b (25) 2.75 bc (74) 8.5 b (27) 3.25 d (67)
Sorghum+sunflower+phenoxaprop-p-ethyl (Puma Super 7.5-EW) Each at 18 L+93.75 g a.i. 0.50 c (75) 3.25 b (70) 3.25de (72) 4.25 c (56)
Sorghum+sunflower+Isoproturon (Partner 50-WP) Each at 18 L+250 g a.i. 1.50 b (25) 2.75 bc (74) 4.75 c (59) 2.75 de (72)
Mesosulfuron+idosulfuron (Atlantis 3.6-WG) Each at 18 L+14.4 g a.i. 1.25 b (38) 2.25 c (79) 1.00 f (91) 2.00 fg (79)
Metribuzin (Sencor 70-WP) Each at 18 L+175 g a.i. 0.50 c (75) 2.00 c (81) 1.00 f (91) 2.00 fg (79)
Phenoxaprop-p-ethyl (Puma Super 7.5-EW) Each at 18 L+315 g a.i. 0.50 c (75) 2.25 c (79) 0.00 g (100) 1.50 g (85)
Isoproturon (Partner 50-WP) Each at 18 L+1000 g a.i. 0.50 c (75) 2.00 c (81) 3.50 d (70) 2.50 df (74)
Values in parenthesis show % inhibition in density as compared with control.
Values having a different letter in a column differ significantly according to least significant difference test (LSD) at P < 0.05.
a.i. = active ingredient.
Table 2. Effect of sorghum+sunflower extracts combined with 1/4 th rates of herbicides on dry matter production of wild oat and canary grass
Dry matter (g m −2 )
Wild oat Canary grass
Extract/herbicide Rate ha −1 2006–2007 2007–2008 2006–2007 2007–2008
Control (weedy check) – 0.71 a 11.69 a 8.34 a 6.00 a
Sorghum+sunflower+mesosulfuron+idosulfuron (Atlantis 3.6-WG) Each at 18 L+3.6 g a.i. 0.08 d (89) 3.29 c (72) 0.69 cd (92) 2.96 b (51)
Sorghum+sunflower+metribuzin (Sencor 70-WP) Each at 18 L+43.75 g a.i 0.24 b (65) 2.27 e (81) 2.13 b (75) 1.86 de (69)
Sorghum+sunflower+phenoxaprop-p-ethyl (Puma Super 7.5-EW) Each at 18 L+93.75 g a.i. 0.08 d (89) 3.94 b (66) 0.80 c (90) 2.36 c (61)
Sorghum+sunflower+Isoproturon (Partner 50-WP) Each at 18 L+250 g a.i. 0.25 b (65) 3.32 c (72) 1.20 c (86) 1.43 ef (76)
Mesosulfuron+idosulfuron (Atlantis 3.6-WG) Each at 18 L+14.4 g a.i. 0.20 bc (72) 2.30 de (80) 0.26 de (97) 1.19 fg (80)
Metribuzin (Sencor 70-WP) Each at 18 L+175 g a.i. 0.12 cd (83) 1.68 f (86) 0.26 de (97) 0.88 g (85)
Phenoxaprop-p-ethyl (Puma Super 7.5-EW) Each at 18 L+315 g a.i. 0.08 d (88) 2.88 cd (75) 0.00 e (100) 1.17 fg (81)
Isoproturon (Partner 50-WP) Each at 18 L+1000 g a.i. 0.09 d (87) 2.27 e (81) 0.89 c (89) 1.43 ef (76)
Values in parenthesis show % inhibition in dry matter production as compared with control.
Values having a different letter in a column differ significantly according to least significant difference test (LSD) at P < 0.05.
a.i. = active ingredient.
first and second years, respectively. Sorghum+sunflower water
extracts combined with 1/4 th (75% less) of label rates of herbicides
suppressed densityof canarygrass by27–76%during firstyear and
46–72% in the second year. Label rates of herbicides decreased
density of canary grass by 70–100% and 74–85% during the first
and second years, respectively.
Sorghum+sunflower extracts combined with 1/4 th (75% less) of
label rates of herbicides and alone label rates of herbicides significantly
(P < 0.05) inhibited dry matter production of both weeds,
wild oat and canary grass as compared with control (Table 2).
Sorghum+sunflower extracts in combination with 1/4 th of label
rates of herbicides reduced dry matter of wild oat by 65–89%
and 66–81% during the first and second years, respectively. The
label rates of herbicides inhibited wild oat dry matter production
by 72–88% and 75–86% during 2 years of experimentation.
Sorghum+sunflower extracts combined with 1/4 th of label rates
of herbicides controlled canary grass by 86–92% and 51–76%
during the first and second years, respectively. The label rates of
herbicides decreased wild oat dry matter production by 89–100%
during the first year and 76–80% during the second year.
Effect of various treatments on persistence index of wild oat
and canary grass
Weed persistence index was calculated to know the perseverance
of wild oat and canary grass (Table 3). A lower persistence
index indicates higher efficiency of any weed control practice.
Persistence index of wild oat in sorghum+sunflower extracts
combined with 1/4 th of label rates of herbicides was either
lower or statistically at par with respective label rates of
herbicides during both years. Likewise, sorghum+sunflower
extracts combined with 1/4 th of label rates of herbicides gave
either lower or statistically equal canary grass persistence index
with respective label rates of herbicides during both years,
except sorghum+sunflower extract+1/4th phenoxaprop-p-ethyl
during first year. The persistence index was calculated considering
together weed density and dry matter production. The lower
persistence index reflects that water extracts+lower rates of
J Sci Food Agric 2010; 90: 1897–1904 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1899

1900
www.soci.org MN Mushtaq et al.
Table 3. Effect of sorghum+sunflower extracts combined with 1/4 th rates of herbicides on persistence index of wild oat and canary grass
Weed persistence index
Wild oat Canary grass
Extract/herbicide Rate (ha −1 ) 2006–2007 2007–2008 2006–2007 2007–2008
Sorghum+sunflower.+ mesosulfuron+idosulfuron (Atlantis 3.6-WG) Each at 18 L+3.6 g a.i. 0.48 b 1.13 a 0.35 a 0.93 b
Sorghum+sunflower+metribuzin (Sencor 70-WP) Each at 18 L+43.75 g a.i 0.49 b 0.80 b 0.36 a 0.94 b
Sorghum+sunflower+phenoxaprop-p-ethyl (Puma Super 7.5-EW) Each at 18 L+93.75 g a.i. 0.48 b 1.13 a 0.35 a 0.91 b
Sorghum+sunflower+Isoproturon (Partner 50-WP) Each at 18 L+250 g a.i. 0.47 b 1.16 a 0.37 a 0.87 b
Mesosulfuron+idosulfuron (Atlantis 3.6-WG) Each at 18 L+14.4 g a.i. 0.47 b 0.96 ab 0.36 a 0.97 b
Metribuzin (Sencor 70-WP) Each at 18 L+175 g a.i. 0.73 a 0.78 b 0.36 a 0.72 b
Phenoxaprop-p-ethyl (Puma Super 7.5-EW) Each at 18 L+315 g a.i. 0.49 b 1.22 a 0.00 b 1.40 a
Isoproturon (Partner 50-WP) Each at 18 L+1000 g a.i. 0.52 b 1.05 ab 0.36 a 1.01 b
Values having a different letter in a column differ significantly according to least significant difference test (LSD) at P < 0.05.
a.i. = active ingredient.
herbicides treatments not only effectively controlled the density
of weeds but also dry matter production of the two weeds. The
lower or statistically equal weed persistence index in water extracts
combined with 1/4 th rates of herbicides indicates efficacy of this
technique compared to full rates of herbicides.
Effect of various treatments on growth and yield of wheat
Data on the number of fertile tillers, number of grains per
spike, thousand grain weight and grain yield of wheat were
recorded to assess the impact of water extracts and herbicides on
growth and yield of wheat. The number of fertile tillers was not
significantly different during 2 years of experimentation (Table 4).
Sorghum+sunflower extracts combined with 1/4 th of label rates
of herbicides produced 47–55 and 53–56 grains per spike during
the first and second years, respectively (Table 4). The label rates of
herbicides gave 47–56 grains during the first year and about 48
grains per spike during the second year. Thousand grain weight
was higher in all treatments as compared with control during
both years (Table 4). Sorghum+sunflower extracts combined
with 1/4 th of label rates of herbicides gave 36.88–44.95 g and
38.15–39.25 g thousand grain weight during the first and second
years, respectively. The label rates of herbicides resulted in
36.55–47.55 g and 37.28–38.14 g thousand grain weight during
the first and second years, respectively.
All treatments produced significantly (P< 0.05) higher
wheat grain yield as compared with control (Table 4).
Sorghum+sunflower extracts combined with 1/4 th of label rates
of herbicides increased wheat grain yield by 23–36% and 31–54%
during the first and second years, respectively. The label rates of
herbicides increased wheat grain yield by 27–41% and 37–50%
during the first and second years, respectively. However, the wheat
grain yield was statistically equal in sorghum+sunflower extracts
combined with 1/4 th of label rates of herbicides with respective
label rates of herbicides.
Effect of various treatments on economicand marginal returns
All treatments produced higher net benefits as compared with
control (Table 5). The treatments containing sorghum+sunflower
water extracts combined with 1/4 th (75%) of label rates
of herbicides gave equal or higher net benefits than their
correspondinglabelratesexceptphenoxaprop-p-ethylandisoproturon.
Sorghum+sunflower extracts each at 18 L ha −1 combined
with 1/4 th (3.4 g a.i. ha −1 ) rate of mesosulfuron+idosulfuron
produced the highest net benefits (40%).The marginal analysis
revealed that sorghum+sunflower extracts combined with 1/4 th
(3.4 g a.i. ha −1 ) rate of mesosulfuron+idosulfuron gave maximum
(4404%) marginal rate of return (MRR) (Table 6). It was followed by
sorghum+sunflower extracts combined with 1/4 th rate (43.75 g
a.i. ha −1 ) of metribuzin with 2175% MRR. All label rates of herbicides
were dominated due to higher total cost that vary and/or
lower net benefits except metribuzin at 175 g a.i. ha −1 (1289%
MRR).
DISCUSSION
Integrated use of sorghum+sunflower extracts and 1/4 th (75%
less) of label rates of herbicides significantly inhibited density and
dry matter production of wild oat and canary grass (Tables 1 and 2).
The main finding of this study is that sorghum+sunflower extracts
combined with 1/4 th rates of herbicides controlled wild oat by
65–89% and canary grass by 51–92%, which is considered good
under field conditions. Furthermore, the equal persistence index
in water extracts+reduced rates of herbicides and label rates of
herbicides showed that 75% less use of herbicides in combination
with sorghum+sunflower water extracts is as effective as label
rates of herbicides for wild oat and canary grass management in
wheat (Table 3). It indicated that herbicide use can be reduced
by 75% through combined application of sorghum+sunflower
extracts and 1/4 th rates of herbicides. Previously, 50% reduction in
herbicide use was reported using sorghum extracts at 12 L ha −1
for weed control in wheat. 25 In the present study, 75% decrease
in herbicide use may be due to two reasons. Firstly, combined
use of two extracts (sorghum+sunflower) might have increased
their efficacy. A mixture of two compounds may be more effective
because they can replace each other on the basis of their biological
exchange rate and may increase efficacy of each other. 26 Jamil
et al. 7 found that combined use of sorghum+sunflower water
extracts was 52% more effective than sorghum extracts alone,
for weed control in wheat. Secondly, a higher rate (18 L ha −1 )of
sorghum and sunflower extracts was used. The inhibitory effects
of allelochemicals against weeds are concentration dependent. 27
Higher concentration of allelopathic extracts may add more
allelochemicals that resulted in more inhibition of weeds. 12 The
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1897–1904

Integrated use of allelopathic extracts reduce herbicide use www.soci.org
Table 4. Effect of sorghum+sunflower extracts combined with 1/4 th rates of herbicides on growth and yield of wheat
Fertile tillers m −2 Number of grains per spike 1000 grain weight (g) Grain yield (Mg ha −1 )
Extract/herbicide Rate (ha −1 ) 2006–2007 2007–2008 2006–2007 2007–2008 2006–2007 2007–2008 2006–2007 2007–2008
Control (weedy check) – 384NS 312NS 42.73e 39.83f 34.72e 27.85c 2.50d 1.45b
Sorghum+sunflower+ Each at 18 L+3.6 g a.i. 370 408 55.20
mesosulfuron+
idosulfuron (Atlantis
3.6-WG)
a 53.20ab 44.95a 39.03ab 3.41ab (36) 2.23a (54)
Sorghum+sunflower+ Each at 18 L+43.75 g a.i 401 432 52.05
metribuzin (Sencor
70-WP)
bc 54.97a 41.00b 38.15ab 3.25bc (30) 2.07a (43)
Sorghum+sunflower+ Each at 18 L+93.75 g a.i. 382 400 51.63
phenoxaprop-p-ethyl
(Puma Super 7.5-EW)
c 56.23a 38.35bcd 39.25a 3.18bc (27) 1.92a (32)
Sorghum+sunflower+ Each at 18 L+250 g a.i. 395 374 47.10
isoproturon (Partner
50-WP)
d 53.37ab 36.88cde 38.83ab 3.07c (23) 1.90a (31)
Mesosulfuron+
Each at 18 L+14.4 g a.i. 401 402 55.80
idosulfuron (Atlantis
3.6-WG)
a 47.30de 47.53a 37.28b 3.52a (41) 2.18a (50)
Metribuzin (Sencor
Each at 18 L+175 g a.i. 403 427 54.25
70-WP)
ab 48.73cd 46.05a 37.33b 3.41ab (36) 2.05a (41)
Phenoxaprop-p-ethyl Each at 18 L+315 g a.i. 415 396 47.43
(Puma Super 7.5-EW)
d 48.2cd 36.55de 38.13ab 3.07c (23) 2.16a (49)
Isoproturon (Partner Each at 18 L+1000 g a.i. 376 374 50.83
50-WP)
c 47.8de 39.43bc 37.6ab 3.18bc (27) 1.98a (37)
Values in parenthesis show % increase in grain yield as compared with control.
Values having a different letter in a column differ significantly according to least significant difference test (LSD) at P < 0.05.
a.i. = active ingredient; NS = non significant.
J Sci Food Agric 2010; 90: 1897–1904 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1901

1902
Table 5. Economic analysis
Treatments
Parameter A B C D E F G H I Remarks
www.soci.org MN Mushtaq et al.
Grain yield∗ 1.98 2.82 2.66 2.55 2.48 2.85 2.73 2.61 2.58 Mg ha−1 Adjusted yield 1.78 2.54 2.39 2.29 2.24 2.57 2.46 2.35 2.32 Mg ha−1 (10% reduction to
bring at farmers’ level)
Gross income 26 663 38 063 35 876 34 370 33 536 38 494 36 848 35 292 34 844 Wheat price at PKR 15
000 Mg−1 Cost of herbicides 0 360 70 278 175 1440 280 1112 700 Mesosulfuron+
idosufuron (PKR
9982/100 g a.i.),
phenoxaprop-p-ethyl (PKR
1186/100 g a.i.), isoproturon
(PKR 70/100 g a.i.),
metribuzin (PKR 160/100 g
a.i.)
Costofextracts 0 140 140 140 140 0 0 0 0 Expenditureonpreparation
of extracts (PKR 70/18 L)
Sprayer rent 0 65 65 65 65 65 65 65 65 PKR 65 spray−1 Spray application cost 0 130 130 130 130 130 130 130 130 PKR 130 man-day−1 (one
man-day ha−1 )
Total cost that vary 0 695 405 613 510 1635 475 1307 895 PKR
Net benefits 26 663 37 368 35 471 33 757 33 026 36 859 36 373 33 985 33 949 PKR ha−1 Increase in net benefits (%) – 40 33 27 24 38 36 27 27 Compared with control
∗ Mean of 2 years.
A = Control (weedy check).
B = Sorghum+sunflower extracts each at 18 L+ mesosulfuron+idosulfuron at 3.6 g a.i. ha−1 .
C = Sorghum+sunflower extracts each at 18 L+ metribuzin at 43.75 g a.i ha−1 .
D = Sorghum+sunflower extracts each at 18 L+ phenoxaprop-p-ethyl at 93.75 g a.i.
E = Sorghum+sunflower extracts each at 18 L+ isoproturon at 250 g a.i.
F = mesosulfuron+idosulfuron at 14.4 g a.i. ha-1;
G = metribuzin at 175g a.i.
H = phenoxaprop-p-ethyl at 315 g a.i.
I = isoproturon at 1000 g a.i.
PKR = Pakistan rupees (1 US dollar = 83 PKR); a.i. = active ingredient.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1897–1904

Integrated use of allelopathic extracts reduce herbicide use www.soci.org
Table 6. Marginal analysis
Extract/herbicide Rate (ha −1 )
Total cost that
vary (PKR ha −1 )
Net benefits
(PKR ha −1 ) Marginalcost
Marginal
net benefits
Marginal rate
of return (%)
A Control (weedy check) – 0 26 63 0 0 0
C Sorghum+sunflower
+metribuzin (Sencor 70-WP)
Each at 18 L+43.75 g a.i 405 35 71 405 8808 2175
G Metribuzin (Sencor 70-WP) Each at 18 L+175 g a.i. 475 36 73 70 902 1289
E Sorghum+sunflower
+isoproturon (Partner 50-WP)
Each at 18 L+250 g a.i. 510 33 26 0 0 D∗ D Sorghum+sunflower
+phenoxaprop-p-ethyl (Puma
Super 7.5-EW)
Each at 18 L+93.75 g a.i. 613 33 57 0 0 D
B Sorghum+sunflower
+mesosulfuron+idosulfuron
(Atlantis 3.6-WG)
Each at 18 L+3.6 g a.i. 695 37 68 82 3611 4404
I Isoproturon (Partner 50-WP) Each at 18 L+1000 g a.i. 895 33 49 0 0 D
H Phenoxaprop-p-ethyl (Puma Super
7.5-EW)
Each at 18 L+315 g a.i. 1307 33 85 0 0 D
F Mesosulfuron+idosulfuron (Atlantis
3.6-WG)
Each at 18 L+14.4 g a.i. 1635 36 59 0 0 D
PKR = Pakistan rupees (1 US dollar=83 PKR);
∗ D=Dominated due to less benefits or higher total cost that vary than the preceding treatment; Mean of 2 years.
weed control by allelopathic extracts+lower rates of herbicide
treatments varied because of different weed suppression by label
rates of respective herbicides.
Growth and yield of wheat was improved in all treatments
as compared with control (Table 4). Sorghum+sunflower extracts
combined with 1/4 th of label rates of herbicides produced wheat
grain yield statistically at par with alone label rates of respective
herbicides during both years of the study. Cheema et al. 25 reported
thatsorghum extracts combined with half (50%)rates of herbicides
gave wheat yield comparable with label rates of herbicides. It
indicates that combined use of sorghum+sunflower extracts at
higher concentration (18 L ha −1 ) had no adverse effects on the
yield of wheat and shifting from label rates of herbicides to
combination of sorghum+sunflower extracts and 75% less rates
of herbicides is viable option for weed management in wheat.
The economic evaluation of any technique is essential for its
adoption at farmers’ level. The economic and marginal analyses
showed that combined use of sorghum+sunflower extracts and
1/4 th of label rates of herbicides is economically viable approach
(Tables 5 and 6). Sorghum+sunflower extracts combined with
1/4 th rate of mesosulfuron+idosulfuron resulted in highest (PKR
37 368, 1 US dollar = 83 PKR) net benefits. The marginal rate of return
(MMR) revealed that out of four sorghum+sunflower extracts
plus 1/4 th rates of herbicides treatments, sorghum+sunflower
plus 1/4 th rates of mesosulfuron+idosulfuron and metribuzin
were economical with 4404% and 2175% MRR, respectively
(Table 6). All label rates of herbicides were uneconomical due to
higher costs that vary and/or lower net benefits except metribuzin
that gave 1289% MMR. It is evident that combined use of
sorghum+sunflower extracts and 1/4 th (75% less) of label rates of
herbicides is not only effective for the control of notorious weeds,
wild oat and canary grass but also cost effective.
CONCLUSION
Integrated use of sorghum+sunflower extracts and 1/4 th (75%
less) of label rates of herbicides (35 DAS) controlled wild oat
and canary grass comparable to label rates of herbicides. They
produced wheat grain yield statistically equal with label rates
of herbicides and were also economical. Hence, this approach
reduces the herbicide use by 75% without compromising wheat
yield and net benefits that will minimise reliance upon synthetic
herbicides, conserve the environment, improve food quality and
help in sustainable weed management. Sorghum and sunflower
are widely grown crops throughout the world. Their herbage is
easilyavailableandfarmerscanprepareextractsattheirown farms.
ACKNOWLEDGEMENTS
The authors are grateful to the Pakistan Agricultural Research
Council for providing financial support for this study under the
Agricultural Linkages Programme (ALP).
REFERENCES
1 GuptaOP, Modern Weed Management, 2nd edn. Agrobios, Jodhpur,
pp. 18–23 (2004).
2 FishelFM, Pesticide Use Trends in the U.S: Global Comparison. Florida
Cooperative Extension Service, Institute of Food and Agricultural
Sciences, University of Florida (2007).
3 HeapI,The International Survey of Herbicide Resistant Weeds. Available:
http://www.weedscience.com [7 October 2009].
4 RonaldE, Hand Book of Chemical Risk Assessment: Health Hazards to
Humans, Plants and Animals, vol. II. Lewis Publishers, Washington
DC (2000).
5 Anderson RL,Managingweeds with a dualistic approach of prevention
and control: a review. Agron Sustain Dev 27:13–18 (2007).
6 Macias FA, Molinillo JM, Varela RM and Galindo JC, Allelopathy – a
natural alternative for weed control. Pest Manage Sci 63:
327–348 (2007).
7 Jamil M, Cheema ZA, Mushtaq MN, Farooq M and Cheema MA,
Alternative control of wild oat and canary grass in wheat fields
by allelopathic plant water extracts. Agron Sustain Dev 29:475–482
(2009).
8 Khaliq A, Aslam Z and Cheema ZA, Efficacy of different weed
management strategies in mungbean (Vigna Radiata L.). Int J Agric
Biol 4:237–239 (2002).
9 Iqbal J, Cheema ZA and Mushtaq MN, Allelopathic crop water extracts
reduced the herbicide dose for weed control in cotton (Gossypium
hirsutum L.). Int J Agric Biol 11:360–366 (2009).
J Sci Food Agric 2010; 90: 1897–1904 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1903

1904
10 Weston LA and Duke SO, Weed and crop allelopathy. Crit Rev Plant Sci
22:367–389 (2003).
11 Netzly DH and Butler LG, Roots of sorghum exude hydrophobic
droplets containing biologically active components. Crop Sci
26:775–780 (1986).
12 Cheema ZA, Mushtaq MN, Farooq M, Hussain A and Din IU, Purple
nutsedge management with allelopathic sorghum. Allelopathy J
23:305–312 (2009).
13 Wilson RE and Rice EL, Allelopathy as expressed by Helianthus annuus
anditsroleinoldfieldsuccession.BullTBotClub 95:432–448 (1968).
14 Macias FA, Varela RM, Torres A, Oliva RM and Molinillo JMG, Bioactive
norsesquiterpenes from Helianthus annuus with potential
allelopathic activity. Phytochemistry 48:631–636 (1998).
15 Macias FA, Torres A, Galindo JLG, Varela RM, Alvarez JA and
Molinillo JMG, Bioactive terpenoids from sunflower leaves cv.
Peredovick. Phytochemistry 61:687–692 (2002).
16 Bell AR and Nalewaja JD, Competition of wild oat in wheat and barley.
Weed Sci 16:505–508 (1968).
17 Malik RK and Singh S,Littleseed canarygrass (Phalarisminor)resistance
to isoproturon in India. Weed Technol 9:419–425 (1995).
18 Hobbs PR, Sayre KD and Monasterio Jio, Increasing Wheat Yields
Sustainably Through Agronomic Means. NRG Paper 98–01, Mexico,
D.F. (1998).
19 Sharif MM, Cheema ZA and Khaliq A, Reducing herbicide dose in
combination with sorghum water extract for weed control in wheat
(Triticum aestivum L.). Int J Agric Biol 7:560–563 (2005).
www.soci.org MN Mushtaq et al.
20 Parveen Z, Identification of allelochemicals in sorghum (Sorghum
bicolor L.) and their effect on germination and seedling growth
of wheat (Triticum aestivum L.). MSc dissertation, Department of
Chemistry, University of Agriculture, Faisalabad, Pakistan (2000).
21 Mushtaq MN, Cheema ZA and Khaliq A, Effects of mixture of
allelopathic plant aqueous extracts on Trianthema portulacastrum
L. weed. Allelopathy J 25:205–212 (2010).
22 Ross MA and Lembi CA, Applied Weed Science. Burgess Publishing Co.,
Minnesota, pp. 133–139 (1985).
23 Steel RGD,Torrie JHandDickey D,PrinciplesandProceduresofStatistics:
A Biometrical Approach, 3rd edition. McGraw Hill, New York (1997).
24 CIMMYT, From Agronomic Data to Farmer Recommendations: An
Economics Training Manual, Mexico, D.F. (1988).
25 Cheema ZA, Jaffer I and Khaliq A, Reducing isoproturon dose in
combination with sorgaab for weed control in wheat. Pak J Weed
Sci Res 9:153–160 (2003).
26 Gerig TM and Blum U, Effects of mixtures of four phenolic acids on leaf
area expansion of cucumber seedlings grown in Portsmouth B1 soil
materials. JChemEcol17:29–39 (1991).
27 Hachinohe M, Sunohara Y and Matsumoto H, Absorption,
translocation and metabolism of L-DOPA in barnyardgrass and
lettuce: their involvement in species-selective phytotoxic action.
Plant Growth Regul 43:237–243 (2004).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1897–1904

Research Article
Received: 26 November 2009 Revised: 29 March 2010 Accepted: 29 April 2010 Published online in Wiley Interscience: 14 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4031
Supercritical CO2 oil extraction from Chinese
star anise seed and simultaneous
compositional analysis using HPLC
by fluorescence detection and online
atmospheric CI-MS identification
Guoliang Li, a,b Zhiwei Sun, a,b Lian Xia, a,b Junyou Shi, a,b Yongjun Liu, a
Yourui Suo a and Jinmao You a∗
Abstract
BACKGROUND: Supercritical CO2 was utilised to extract Chinese star anise seed oil (CSASO), and a three-level Box–Behnken
factorial design from response surface methodology was applied to optimise the extraction conditions, including pressure,
temperature and amount of modifier (ethanol). The compositional analysis of fatty acids in CSASO was performed by HPLC with
fluorescence detection using 2-(11H-benzo[a]carbazol-11-yl)-ethyl-4-methylbenzenesulfonate (BCETS) as labelling reagent.
Identification was carried out by online atmospheric chemical ionisation–mass spectrometry.
RESULTS: The optimum extraction conditions were as follows: extraction pressure, 27.72 MPa, extraction temperature, 46.22 ◦ C,
and amount of modifier, 8.58 vol.%. The experimental result showed that the maximum extraction yield was 25.31 ± 0.22%
(w/w) under the conditions proposed. The compositional analysis indicated that CSASO mainly contained C18: 2, C18 : 1, C18 : 3,
C20 : 4, C16, C18 and C20 fatty acids.
CONCLUSION: In this study, a fast, simple and high-efficiency supercritical technique for extracting oil from Chinese star anise
seed was developed. Simultaneous determination of fatty acids in CSASO using BCETS as the labelling reagent with HPLC
fluorescence detection and online mass spectroscopy identification has been successfully achieved.
c○ 2010 Society of Chemical Industry
Keywords: Chinese star anise seed oil; supercritical CO2 fluid extraction; response surface methodology; fatty acids; HPLC/APCI/MS
INTRODUCTION
Chinese star anise (Illicium verum Hook. f.), the seed pod of an
evergreen tree grown in eastern Asia, has a pungent, licoricelike
flavour. It is used as a spice in cooking and for treatment
of dyspeptic complaints and catarrhs of the respiratory tract
in traditional Chinese medicine. 1,2 However, in the last year,
intoxications after consumption of Chinese star anise have been
reported due to adulterations of I. verum with the morphologically
similar Japanese star anise (I. anisatum L.) containing anisatin
and other related toxic sesquiterpene lactones. 2,3 Most previous
studies were focused on the essential oils in the fruit peel of
Chinese star anise, 1,4,5 but there is little study on the Chinese star
anise seed oil (CSASO).
The most valuable part in Chinese star anise is the essential
oil, which have a wide range of commercial applications in the
production of perfumes, cosmetics, soaps, food and beverage
flavourings. 6 The seeds are a by-product of the process, and
account for 20.4% of the total fruit. It has an estimated annual
production potential of one million metric tons in China. Usually
they are treated as waste disposal.
Supercritical CO2 fluid extraction (SFE) offers numerous
potential advantages over conventional extraction processes,
including the facts that it is non-toxic, non-explosive, environmentally
friendly, cost-effective, has a lower consumption of organic
solvent, is time-saving and has high selectivity. 7 SFE has been
widely used for seed oil extractions. 8–11 However, to the best of
our knowledge, the effect of SFE parameters on the CSASO yield
and the optimum operation conditions for CSASO remain poorly
investigated. For a possible industrial application, the optimisation
and assessment of the extraction process with mathematical
modelling seem to be essential. In classical methods, process parameters
are optimised by conducting experiments concentrating
∗ Correspondenceto:Jinmao You,NorthwestPlateauInstituteofBiology,Chinese
Academy of Sciences, Xining, 810001, P.R. China.
E-mail: jmyou6304@163.com
a Northwest Plateau Institute of Biology, Chinese Academy of Sciences, Xining,
810001, P.R. China
b Graduate School of the Chinese Academy of Sciences, Beijing 100039, P.R. China
J Sci Food Agric 2010; 90: 1905–1913 www.soci.org c○ 2010 Society of Chemical Industry
1905

1906
on one factor at a time. This method is also troublesome and timeconsuming
as well as ignoring the interaction effect of parameters.
Compared to the classical methods, response surface methodology
(RSM) is more efficient, requires fewer data and provides
interaction effects on the response besides factor effects. It has
also been extensively applied to optimise processing parameters
in the production of food, drugs and other commodities. 12–15
Quantitative determination of the fatty acids composition is
helpful to control the quality of CSASO, which has not been
reported yet. Most fatty acids show neither natural absorption in
the visible or ultra-violet (UV) regions nor fluorescence; thus,
their detection at trace levels using absorptiometry is fairly
difficult. Therefore, derivatisation of these analytes with labelling
reagents has been widely adopted, since high-performance liquid
chromatography (HPLC) with UV or fluorescence detection has
a higher sensitivity. 16–18 In this study, 2-(11H-benzo[a]carbazol-
11-yl) ethyl 4-methylbenzenesulfonate (BCETS), which is a new
fluorescent labelling regent, is utilised to analyse the fatty acids
composition in CSASO.
The aims of the present work were to (1) optimise the SFE
conditions including extraction pressure, extraction temperature,
the added amount of modifier, and study on the interaction
effect of the parameters using RSM; (2) to develop a sensitive
method using BCETS as labelling reagent for the simultaneous
determination of saturated and unsaturated fatty acids; and (3) to
determine and quantify the total fatty acids (TFA) and free fatty
acids (FFA) in CSASO.
MATERIALS AND METHODS
Materials
The dried Chinese star anise seeds were purchased in Guangxi
(China), and were ground into powder with a cyclone mill
and passed through a 60 mesh sieve. All fatty acids were
purchased from Sigma Reagent Co. (St Louis, MO, USA). Water
was purified on a Milli-Q system (Millipore, Bedford, MA, USA).
2-(11H-benzo[a]carbazol-11-yl) ethyl 4-methylbenzenesulfonate
(BCETS) was synthesised in our laboratory (the synthesis of BCETS
is not shown). All other chemicals and solvents used were of
analytical grade.
Cell wall breakage pretreatment and supercritical CO2
extraction of CSASO
Cell breakage pretreatment and extraction measurements were
carried out by a semi-bath flow extraction apparatus (Hua’an
SupercriticalFluidExtractionCorp.,Nantong,China).Theschematic
flow diagram was described in detail in a previous study. 19
The cell wall breakage pretreatment was carried out by the
method described by Xu et al. 20 with minor modifications. The
prepared samples (400 g) were placed into a steel cylinder
equipped with mesh filters on both ends to prevent the particles
being flushed out. The loaded cylinder was then introduced into
the extraction vessel and liquefied CO2 was pumped into the
vessel by a high pressure pump. Extraction pressure, extraction
temperature and CO2 flow rate were controlled by adjusting
the valves on the front panel. The pressure and temperature
were controlled to an accuracy of 45 ± 0.5MPaand40± 0.5 ◦ C,
respectively, and kept for 10 min. At the end of each treatment,
the pressure was quickly released to atmospheric pressure within
1min.
The next step was consecutive oil extraction, which was
conducted at the specified extraction conditions (see Table 1).
www.soci.org G Li et al.
Table 1. Experimental and predicted data for the oil yield obtained
from the Box–Behnken design (n = 3)
Run
number
X1,
pressure
(MPa)
Independent variables Oil recovery (%, w/w)
X2,
temperature
( ◦ C)
X3,
modifier
(vol.%)
Experimental
Predicted
1 22.5(0) 40.0 (0) 7.5 (0) 22.11 22.02
2 22.5(0) 40.0 (0) 7.5 (0) 22.60 22.02
3 15.0 (−1) 30.0 (−1) 7.5 (0) 10.80 10.21
4 30.0 (+1) 40.0 (0) 0.0 (−1) 21.90 21.00
5 15.0 (−1) 40.0 (0) 15.0 (+1) 14.70 15.60
6 22.5 (0) 30.0 (−1) 0.0 (−1) 12.10 12.45
7 22.5 (0) 40.0 (0) 7.5 (0) 21.78 22.02
8 30.0 (+1) 40.0 (0) 15.0 (+1) 23.70 23.45
9 15.0 (−1) 50.0 (+1) 7.50 (0) 10.40 9.85
10 22.5 (0) 40.0 (0) 7.50 (0) 21.40 22.02
11 22.5 (0) 40.0 (0) 7.50 (0) 22.20 22.02
12 30.0 (+1) 50.0 (+1) 7.50 (0) 23.72 24.32
13 15.0 (−1) 40.0 (0) 0.00 (−1) 6.80 7.05
14 22.5 (0) 50.0 (+1) 15.0 (+1) 21.50 21.15
15 22.5 (0) 50.0 (+1) 0.00 (−1) 12.60 12.90
16 30.0 (+1) 30.0 (−1) 7.50 (0) 17.00 17.55
17 22.5(0) 30.0 (−1) 15.0 (+1) 15.53 15.20
With some extraction procedures, ethanol, which was used as
modifier, was pumped into the system from the modifier bottle
after the selected pressure and temperature had been achieved.
Each extraction lasted for 90 min, since longer extraction times
did not significantly increase the yield of oil. At the end of the
extraction, supercritical CO2 was depressurised by a flow control
valve to atmospheric pressure, and the oil was collected in a
collection vial. The oil yield was calculated by the weight increased.
Experimental design and statistical analysis
The single-factor experimental design (extracting pressure, extracting
temperature, extracting time and the added amount
of modifier) were carried out before RSM experiments (data not
shown).Threefactors(extractingpressure,extractingtemperature,
and the added amount of modifier) were chosen for further optimisation
by employing a three-level, three-variable Box–Behnken
factorial design (BBD) from RSM. 21 The coded and uncoded independent
variables used in the RSM design and their respective
levels were listed in Table 1. A total of 17 experiments were designed
(Table 1). Each experiment was performed in triplicate and
the average oil yield (%, w/w) was taken as the response, Y.Based
on the experimental data, regression analysis was performed and
was fitted into an empirical second-order polynomial model:
Y = β0 + β1X1 + β2X2 + β3X3 + β11X1 2 + β22X2 2
+ β33X3 2 + β12X1X2 + β13X1X3 + β23X2X3
where Y represents the response variable, β0 is a constant term,
β1, β2 and β3, are linear coefficients, β11, β22 and β33 are quadratic
coefficients, β12, β13 and β23 are interaction coefficients.
A software Design-Expert 7.1.3 Trial (State-Ease, Inc., Minneapolis,
MN, USA) was used to obtain the coefficients of the quadratic
polynomial model. The quality of the fitted model was expressed
by the determined coefficient (R 2 ), and its statistical significance
was checked by an F test.
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1905–1913

Analysis of Chinese star anise by oil extraction www.soci.org
Figure 1. Derivatisation and identification of fatty acids. (A) Derivatisation scheme of fatty acids with BCETS; (B) the MS/MS cleavage mode of a
BCETS–fatty acid derivative; (C) typical LC/MS profile of the C18 acid derivative (BCETS-C18) from full scanning range from 100 to 600 amu with APCI in
positive-ion detection mode and typical APCI-MS/MS profile of C18 acid derivative (BCETS-C18) from full scanning range from 100 to 600 amu with APCI
in positive-ion detection mode.
Fatty acids analysis
Instrumentation
Experiments were performed with an LC-MSD-Trap-SL liquid
chromatograph mass spectrometer (1100 Series LC-MSD Trap,
a complete LC-MS-MS instrument; Waldbronn, Germany). All the
HPLC system devices were from the HP 1100 series (Waldbronn,
Germany) and consisted of a vacuum degasser (model G1322A),
a quaternary pump (model G1311A), an autosampler (model
G1329A), a thermostatic column compartment (model G1316A),
a fluorescence detector (FLD; model G1321A), and a diode array
detector (DAD; model G1315A). The mass spectrometer, from
Bruker Daltonik (Bremen, Germany), was equipped with an APCI
ion-source. The mobile phase was filtered through a 0.2 µmnylon
membrane filter (Alltech, Deerfield, IL, USA).
Saponification of seed oil
To a 10 mL test tube, 0.1 g seed oil and 2.0 mL potassium
hydroxide/methanol solution (2 mol L −1 ) were added. After being
sealed, the test tube was immersed in a water bath at 90 ◦ Cfor2h.
After cooling, the contents were transferred to a centrifugal test
tube, to which 2 mL water was added, and the pH was adjusted
to 7.0 with 2 mol L −1 hydrochloric acid solution. This solution was
extracted with chloroform three times (3 mL × 3). The combined
chloroformwasfilteredandevaporatedunderastreamofnitrogen.
The residue was re-dissolved in 50 mL N, N-dimethylformamide
(DMF), filtered through a 0.2 mm nylon membrane filter, and
stored at −10 ◦ C until HPLC analysis.
Pre-column derivatisation of fatty acids
To a 1 mL vial, 50 µLBCETS,10mgK2CO3, 100 µL fatty acid mixture
and 200 µL DMF was successively added. The vial was sealed and
allowed to react in a water bath at 90 ◦ C with shaking in 5 min
intervals for 30 min. After the reaction was completed, the mixture
was cooled to room temperature. The derivatisation procedure is
showninFig.1A.
Separation fatty acid derivatives with HPLC
HPLC separation of BCETS–fatty acid derivatives was carried out
on a reversed-phase Eclipse XDB-C8 column (150 mm × 4.6 mm,
5 µm; Agilent, Agilent Technologies, USA) with a gradient elution.
Eluent A was water, B was a mixed solvent of acetonitrile (ACN)
and DMF (1 : 1.v/v), and C was acetonitrile (100%). The flow rate
was constant at 1.0 mL min −1 and the column temperature was
set at 30 ◦ C. The injection volume was 10 µL. The fluorescence
excitation and emission wavelengths were set at λex 279 nm and
λem 380 nm, respectively. The gradient elution program was as
follows: initial = 45% A, 50% B; 30 min = 10% A, 80% B; 40 min =
3% A, 87% B; 50 min = 2% A, 88% B; 70 min = 0% A, 85% B.
Quantitative analysis
Quantitative conversion of fatty acids from CSASO to their BCETS
derivatives was ensured by using an excess of BCETS. All fatty
acids were quantified using the external standard method. The
calibration curves for each fatty-acid derivative were obtained
by linear regression plotting peak area versus concentration (see
Table 2).
RESULTS AND DISCUSSION
Model fitting and statistical analysis
The conditions for supercritical CO2 oil extraction from Chinese
star anise seed were optimised using different parameters by the
combinations of the Box–Behnken design (3 3 factorial). Table 1
shows the experimental and predicted oil yields. The regression
coefficients of the intercept, linear, quadratic and interaction terms
of the model were calculated using the least square technique,
J Sci Food Agric 2010; 90: 1905–1913 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1907

1908
Table 2. Linear regression equations, correlation coefficients, detection limits, and repeatability
Fatty acid
Regression
equation
Correlation
coefficient
www.soci.org G Li et al.
Detection
limit (fmol)
Retention time
(RSD%, n = 6)
Peak area
(RSD%, n = 6)
C5 Y = 1.70X + 2.10 0.9994 20.71 0.21 0.47
C6 Y = 1.18X − 1.07 0.9997 23.67 0.18 0.70
C7 Y = 1.21X + 5.32 0.9994 17.07 0.13 0.59
C8 Y = 1.05X + 7.02 0.9999 32.06 0.10 0.58
C9 Y = 0.84X + 5.16 0.9997 20.25 0.09 0.49
C10 Y = 0.90X + 4.23 0.9995 19.76 0.07 0.40
C11 Y = 0.89X − 1.93 0.9994 20.25 0.08 0.50
C12 Y = 0.93X + 0.38 0.9998 18.92 0.06 0.47
C20 : 5 Y = 1.19X + 2.90 0.9996 16.32 0.06 0.38
C13 Y = 0.60X + 3.47 0.9994 34.19 0.06 0.22
C18 : 3 Y = 1.23X + 3.60 0.9996 15.65 0.06 0.25
C22 : 6 Y = 1.09X + 4.20 0.9998 16.19 0.06 0.30
C14 Y = 0.74X + 4.49 0.9994 20.79 0.04 0.27
C20 : 4 Y = 1.08X + 2.65 0.9999 15.18 0.07 0.33
C18 : 2 Y = 1.42X − 5.38 0.9996 14.26 0.04 0.27
C15 Y = 0.87X − 1.81 0.9994 12.06 0.04 0.19
C16 Y = 1.26X − 3.36 0.9994 10.79 0.04 0.11
C18 : 1 Y = 1.94X − 3.47 0.9994 12.32 0.03 0.14
C17 Y = 0.86X − 3.63 0.9998 12.06 0.03 0.19
C18 Y = 0.92X − 2.24 0.9999 12.06 0.02 0.18
C20 : 1 Y = 1.14X − 2.76 0.9995 15.34 0.04 0.20
C19 Y = 0.82X − 4.19 0.9996 10.79 0.02 0.19
C20 Y = 0.74X + 0.01 0.9997 14.65 0.05 0.40
C22 : 1 Y = 0.76X − 1.56 0.9998 18.74 0.09 0.52
C21 Y = 0.97X − 2.09 0.9999 15.78 0.07 0.88
C22 Y = 0.67X − 2.45 0.9997 16.34 0.10 1.25
C24 : 1 Y = 0.54X + 0.87 0.9996 20.48 0.18 1.14
C23 Y = 1.03X − 5.89 0.9994 28.05 0.12 1.38
C24 Y = 1.01X − 2.48 0.9999 24.75 0.19 1.92
C25 Y = 1.08X − 11.64 0.9996 25.62 0.24 2.30
C26 Y = 0.91X − 4.33 0.9997 26.79 0.23 2.96
Y is the peak area, and X is the amount injected (pmol).
and are presented in Table 3. The independent and dependent
variables were analysed to obtain a regression equation that
could predict the response within the given range. The predicted
second-order polynomial model was:
Y = 22.02 + 5.45X1 + 1.62X2 + 2.75X3 − 2.59X1 2 − 3.94X2 2
− 2.65X3 2 + 1.78X1X2 − 1.53X1X3 + 1.38X2X3.
The analysis of variance (ANOVA) for the experimental results
of BBD is shown in Table 3. Obviously, all of the linear parameters,
interaction parameters and quadratic parameters were found to be
significant at the level of P < 0.05 or P < 0.01. In this experiment,
the value of R 2 (0.9912) revealed that the experimental data was
in good agreement with the predicted values of the oil yield. The
value of adj-R 2 (0.9800) suggested that the total variation of 98%
for the yield of seed oil was attributed to the independent variables
and only about 2% of the total variation could not be explained
by the model.
The lack of fit was an indication of the failure for a model, if
there is a significant lack of fit which could be indicated by a
low probability value, the response predictor is discarded. 22 The F
value for the lack of fit was insignificant (P > 0.05), meaning that
this model was sufficiently accurate for predicting the relevant
responses. The coefficient of variation (CV%) of less than 4.42%
indicated that the model was reproducible. 23
Response surface analysis and optimisation of SFE conditions
The three-dimensional (3D) response surface and 2D contour
plots can provide a method to visualise the relationship between
responses and experimental levels of each variable and the type
of interactions between two test variables. Figure 2A shows the
effect of the extraction pressure and temperature on the oil yield
at a fixed modifier amount of 7.50 vol.%. With a definite extraction
temperature, pressure had a positive linear effect on the oil yield,
the oil yield increased significantly with the increasing extraction
pressure (Fig. 2A), most likely due to the increase of solvent density
resulting in the improvement of oil solubility. 24 However, the
extraction temperature showed the different results compared to
extraction pressure. Oil yields increased with increasing extraction
temperature and reached a maximum value, followed by a decline
with its further increase (Fig. 2A). That was probably due to the
fact that the density of CO2 decreased at further high temperature.
The temperature showed a negative quadratic effect, while the
complex interaction between temperature and pressure had a
positive effect on the oil yield (see Table 3).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1905–1913

Analysis of Chinese star anise by oil extraction www.soci.org
(A)
(B)
(C)
Figure 2. The 3D response surface and 2D contour plots of the oil recovery affected by extraction pressure, extraction temperature, and the amount of
modifier.
J Sci Food Agric 2010; 90: 1905–1913 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1909

1910
www.soci.org G Li et al.
Table 3. Estimated regression coefficients for the quadratic polynomial model and ANOVA for the experimental results
Regression
coefficient
Estimated
coefficients Standard error
Degrees of
freedom Sum of squares F value Prob > F
β0 22.02 0.35 1 485.48 88.02

Analysis of Chinese star anise by oil extraction www.soci.org
Figure 3. Comparison of the oil yields of Chinese star anise seed powder with cell wall breakage pretreatment and without (the squares represent seed
powder with pretreatment; and the triangles represent seed powder without pretreatment).
Table 4. MS data of BCETS-fatty acid derivatives and content of fatty acids from CSASO (n = 3)
Fatty
acid
Molecular
weight [M+H] +
Specific
MS/MS data TFA 1 %(µgmL −1 ) a TFA 2 %(µgmL −1 ) a FFA 3 %(µgmL −1 ) a
C9 401 402.3 261.4, 185.4 0.16 (15.13) 0.12 (12.03) 0.31 (1.46)
C10 415 416.3 261.4, 199.5 0.09 (8.66) 0.08 (8.06) 0.05 (0.22)
C18 : 3 521 521.9 261.4, 304.9, 503.9 0.46 (44.15) 0.46 (45.78) 0.34 (1.61)
C20 : 4 547 547.9 261.4, 330.9, 529.9 0.19 (17.80) 0.17 (16.52) 0.00 (0.00)
C18 : 2 523 523.9 261.4, 306.7, 516.0 47.15 (4518.20) 45.10 (4473.24) 46.94 (223.45)
C16 499 500.2 261.4, 283.5 20.38 (1953.09) 21.08 (2091.09) 24.99 (118.95)
C18 : 1 525 525.8 261.4, 309.0, 507.7 26.33 (2522.81) 25.77 (2555.61) 22.67 (107.90)
C18 527 528.3 261.4, 311.3 4.81 (461.00) 6.65 (659.55) 4.30 (20.48)
C20 555 556.3 261.4, 339.4 0.44 (42.30) 0.57 (56.67) 0.42 (1.99)
Total SFA – – – 26 (2480.17) 29 (2827.39) 31 (143.10)
Total UFA – – – 74 (7102.97) 71 (7091.15) 69 (332.96)
1 Total fatty acids in CSASO extracted under the optimum conditions without modifier.
2 Total fatty acids in CSASO extracted under the optimum conditions.
3 Free fatty acids in CSASO extracted under the optimum conditions.
a a: mass percent (%, ratio of the mass of a fatty acid with that of all fatty acids), absolute content (µg fatty acid mL −1 oil).
graph was established with the peak area (y axis) versus the fatty
acid concentration (x axis: pmol,), and all of the fatty acids provided
excellent linear responses, with correlation coefficients >0.9994.
For the 1.0 pmol injections, the calculated detection limits (at a
signal-to-noise of 3 : 1) of all of the derivatised fatty acids ranged
from 10.79 to 34.19 fmol.
Analysis of fatty acids in CSASO
The total fatty acids (TFA) and free fatty acids (FFA) from CSASO extractedundertheoptimumconditionsweredeterminedandquantified
by HPLC/APCI/MS. The representative chromatograms of
fatty acid standards and total fatty acids from CSASO are presented
in Fig. 4 (chromatograms of other samples not shown). The result
in Table 4 indicated that the seed oil extracted mainly contained
C16, C18, C18 : 1, C18 : 2, C18 : 3 and C20 : 4 fatty acid. Unsaturated
fatty acids (UFA) accounted for 71% at a concentration order of
C18 : 2 > C18 : 1 > C18 : 3 > C20 : 4. In order to investigate the effect
of modifier on the content of fatty acids, the composition of fatty
acids in CSASO obtained under the optimum conditions without
modifier was determined and the resultis also presented in Table 4.
As is shown in Table 4, SFE without modifier yielded 9583 µgmL −1
of TFA; when ethanol was added as modifier, 9918 µgmL −1 was
obtained. The result indicated that modifier in SFE has great impact
on the content of fatty acids. FFA was also determined and
the composition was basically consistent with the total fatty acids
(Table 4), but the C20 : 4 fatty acid was not detected in FFA.
Inthisstudy,thefattyacidcontentinthesamplewasdetermined
bypre-columnderivatisation(usingBCETSasthelabellingreagent)
using HPLC/APCI/MS by fluorescence detection. This method
provedtobesensitiveandquantitativeforanalysingthefattyacids.
CONCLUSION
In this work, the SFE parameters for CSASO were optimised
using BBD from RSM, and under the optimum conditions, the
experimental yield (25.31 ± 0.22%, w/w) was well matched with
the predicted yield (25.00%, w/w). The effect of cell wall breakage
pretreatment by supercritical CO2 rapid depressurisation was
significant, which could greatly reduce the extraction time and
increase the oil yield. Simultaneous determination of fatty acids in
CSASO using BCETS as labelling reagent with HPLC fluorescence
detection and online MS identification has been successfully
achieved. The established method can be applied to the extraction
J Sci Food Agric 2010; 90: 1905–1913 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1911

1912
www.soci.org G Li et al.
Figure 4. Chromatogram of fatty acids standards derivative (A) and fatty acids derivative in extracted oil (B). Chromatographic conditions: column
temperature at 30 ◦ C; excitation wavelength, λex, 279 nm, emission wavelength, λem, 380 nm; Eclipse XDB-C8 column (4.6 × 150 mm, 5 µm); flow
rate, 1.0 mL min −1 . Peak labels: C6 (hexanoic acid); C7 (heptoic acid); C8 (octanoic acid); C9 (nonanoic acid); C10 (decoic acid); C11 (undecanoic acid);
C12 (dodecanoic acid); C20 : 5 (5,8,11,14,17-eicosapentaenoic acid); C13 (tridecanoic acid); C18 : 3 (8,11,14-octadecatrienoic acid); C22 : 6 (2,5,8,11,14,17docosahexaenoic
acid); C14 (myristic acid); C20 : 4 (6,9,12,15-arachidonic acid); C18 : 2 (9,12-octadecadienoic acid); C15 (pentadecanoic acid); C16
(hexadecanoic acid);C18: 1 (12-octadecenoic acid); C17 (heptadecanoic acid); C18 (stearic acid); C20 : 1 (11-eicosenoic acid); C19 (nonadecanoic acid);
C20(eicosoic acid); C22 : 1 (12-docosenoic acid); C21 (heneicosanoic acid); C22 (docosanoic acid); C24 : 1 (20-tetracosenoic acid); C23 (tricosanoic acid);
C24 (tetracosanoic acid); C25 (pentacosanoic acid); C26 (hexacosanoic acid).
and determination of fatty acids from various food, drugs, plants
and biochemistry samples.
ACKNOWLEDGEMENT
This work was supported by the National Science Foundation
of China (No. 20075016) and supported by the 100 Talents
Programme of The Chinese Academy of Sciences (No. 328).
REFERENCES
1 WangZ, WangL, LiT, ZhouX, DingL, YuY, et al., Rapid analysis of
the essential oils from dried Illicium verum Hook. f. and Zingiber
officinale Rosc. by improved solvent-free microwave extraction
with three types of microwave-absorption medium. Anal Bioanal
Chem 386:1863–1868 (2006).
2 Lederer I, Schulzki G, Gross J and Steffen J, Combination of TLC and
HPLC-MS/MS methods. Approach to a rational quality control of
Chinese star anise. J Agric Food Chem 54:1970–1974 (2006).
3 Ize-LudlowD,RagoneS,BruckI,BernsteinJ,DuchownyMandPenaB,
Neurotoxicities in infants seen with the consumption of star anise
tea. Am Acad Pediatr 114:653–656 (2004).
4 Yan JH, X. Xiao XX, and Huang KL, Component analysis of volatile oil
from Illicium verum Hook. F. J Cent South Univ of Technol 9:173–176
(2002).
5 Tuan DQ and Ilangantileket SG, Liquid CO2 extraction of essential oil
from star anise fruits (Illicium verum H.). J Food Eng 31:47–57 (1997).
6 Al-MasriH,Lainé D, Al-Jourani M, Abu-Taha M and Salem K, Detection
of vapour emissions from star anise seeds and mint leaves using
the CO. Int J Nat Eng Sci 1:105–113 (2007).
7 Gouveia L, Nobre B, Marcelo F, Mrejen S, Cardoso M, Palavra A, et al.,
Functional food oil coloured by pigments extracted from
microalgae with supercritical CO2. Food Chem 101:717–723 (2007).
8 Han X, Cheng L, Zhang R and Bi J, Extraction of safflower seed oil by
supercritical CO2. J Food Eng 92:370–376 (2009).
9 Machmudah S, Sulaswatty A, Sasaki M, Goto M and Hirose T,
Supercritical CO2 extraction of nutmeg oil: Experiments and
modeling. J Supercrit Fluid 39:30–39 (2006).
10 LiuW,FuY,ZuY,TongM,WuN,LiuX,et al., Supercritical carbon
dioxide extraction of seed oil from Opuntia dillenii Haw. and its
antioxidant activity. Food Chem 114:334–339 (2008).
11 WestermanD,SantosR,BosleyJ,Rogers J and Al-Duri B, Extraction of
amaranth seed oil by supercritical carbon dioxide. J Supercrit Fluid
37:38–52 (2006).
12 Liyana-Pathirana C and Shahidi F, Optimization of extraction of
phenolic compounds from wheat using response surface
methodology. Food Chem 93:47–56 (2005).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1905–1913

Analysis of Chinese star anise by oil extraction www.soci.org
13 Li K, Lai P, Lu S, Fang Y and Chen H, Optimization of acid hydrolysis
conditions for feruloylated oligosaccharides from rice bran through
response surface methodology. J Agric Food Chem 56:8975–8978
(2008).
14 Lafronza L, McAlpine A, Keane A and Astley R, Response surface
method optimization of uniform and axially segmented duct
acoustics liners. JAircraft43:1089–1102 (2006).
15 Desai K, Akolkar S, Badhe Y, Tambe S and Lele S, Optimization
of fermentation media for exopolysaccharide production
from Lactobacillus plantarum using artificial intelligence-based
techniques. Process Biochem 41:1842–1848 (2006).
16 You JM, Zhu F, Zhao WC, Zhao XE and Suo YR, Analysis of saturated
free fatty acids from pollen by HPLC with fluorescence detection.
Eur J Lipid Sci Technol 109:225–236 (2007).
17 You JM, Shi YW, Zhao XE, Zhang H, Suo YR, Li YL, et al., Enhancement
of atmospheric pressure chemical ionization for the determination
of free and glycine-conjugated bile acids in human serum. JSepSci
29:2837–2846 (2006).
18 You JM, Zhao XE, Suo YR, Li YL and Wang HL, Determination of
long-chain fatty acids in bryophyte plant extracts by HPLC with
fluorescence detection and identification with MS. J Chromatogr B
848:283–291 (2007).
19 Koga Y, Iwai Y, Hata Y, Yamamoto M and Arai Y, Influence of cosolvent
on solubilities of fatty acids and higher alcohols in supercritical
carbon dioxide. Fluid Phase Equilib 125:115–128 (1996).
20 Xu X, Sun L, Dong J and Zhang H, Breaking the cells of rape bee
pollen and consecutive extraction of functional oil with supercritical
carbon dioxide. Innov Food Sci Emerg Technol 10:42–46 (2009).
21 Box G and Wilson K, On the experimental attainment of optimum
conditions. J R Statist Soc 8:1–45 (1951).
22 Montgomery D, Design and Analysis of Experiments, Wiley, Chichester,
pp. 427–500 (2001).
23 Daniel W, Biostatistics: A Foundation for Analysis in the Health Sciences,
Wiley, Chichester, pp. 591–598 (1991).
24 Abbasi H, Rezaei K and Rashidi L, Extraction of essential oils from the
seeds of pomegranate using organic solvents and supercritical CO2.
JAmOilChemSoc85:83–89 (2008).
25 Koga Y, Iwai Y, Hata Y, Yamamoto M and Arai Y, Influence of cosolvent
on solubilities of fatty acids and higher alcohols in supercritical
carbon dioxide. Fluid Phase Equilibr 125:115–128 (1996).
26 Sánchez-Vicente Y, Cabañas A, Renuncio J and Pando C, Supercritical
fluid extraction of peach (Prunus persica) seed oil using carbon
dioxide and ethanol. J Supercrit Fluid 49:167–173 (2009).
27 Salgn U, Extraction of jojoba seed oil using supercritical CO2+ethanol
mixture in green and high-tech separation process. J Supercrit Fluid
39:330–337 (2007).
28 Cocero M and Calvo L, Supercritical fluid extraction of sunflower seed
oil with CO2−ethanol mixtures. JAmOilChemSoc73:1573–1578
(1996).
J Sci Food Agric 2010; 90: 1905–1913 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1913

1914
Research Article
Received: 14 January 2010 Revised: 15 April 2010 Accepted: 7 May 2010 Published online in Wiley Interscience: 10 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4032
Response of nitrogen metabolism in lettuce
plants subjected to different doses and forms
of selenium
Juan J Rios, ∗ Begoña Blasco, Miguel A Rosales, Eva Sanchez-Rodriguez,
Rocio Leyva, Luis M Cervilla, Luis Romero and Juan M Ruiz
Abstract
BACKGROUND: Currently, biofortification programmes are being carried out with selenium (Se), since it is an essential element
for humans and its ingestion depends partly on a vegetable diet, this not being so for plants. In this sense, few studies have
tested the effect that Se has on some of the main plant metabolisms, such as nitrogen (N) metabolism. Thus the aim of this study
was to establish the effect of the application of different doses (5, 10, 20, 40, 60, 80 and 120 µmol L −1 ) and forms (selenate and
selenite) of Se on the reduction of nitrate (NO − 3
) and subsequent assimilation of ammonium (NH+
4 ).
RESULTS: The results showed an increase in all enzyme activities analysed (nitrate reductase (NR), nitrite reductase (NiR),
glutamine synthetase (GS) and glutamate synthase (GOGAT)), especially with application of the selenite form, in addition to a
decline in foliar NO − 3 concentration.
CONCLUSION: Se applied in both forms increased N metabolism, with selenite inducing this physiological process more strongly,
since it prompted a stronger activation of NR, GS and GOGAT as well as a greater concentration of total reduced N.
c○ 2010 Society of Chemical Industry
Keywords: Lactuca sativa L.; nitrate; nitrogen assimilation enzymes; selenium
INTRODUCTION
Selenium (Se) has been deemed essential to animal nutrition since
1957. It is a component of the enzymes glutathione peroxidase,
selenoprotein P and tetraiodothyroine 5 ′ -deiodinase. 1,2 In this
sense, Se is important in mammalian nutrition, since Se deficiency
may promote cancer. 3 Also, Se has low bioavailability in most
crop soils, 4,5 especially in China, the UK, Eastern Europe and
Australia. 6 The concentration of this element is minimal in plant
foods, leading to dietary Se deficiency in humans. 7 Therefore,
given this low bioavailability and the fact that plants are the main
dietary source of this element, different studies have appeared in
recent years concerning ways to increase the Se content in plants
consumed by humans. 5,8,9
Biofortification has been defined as the process of increasing the
bioavailable concentration of essential elements in edible portions
of plants through agricultural intervention or genetic selection. 10
Current reports on wheat varieties, radish and lettuce show that
fertilisation with Se raises the concentration of this trace element
in plants, and they also analyse the contribution of this increase
to daily Se ingestion in humans. 5,9,11 Although diverse studies
have shown a directly proportional relationship between the Se
increase in the culture medium and the rise in foliar concentration
of Se, this relationship is more significant with application in the
form of selenate. 5,9
However, none of these studies has thoroughly analysed the
possible effects of Se application on nutrients essential for plant
growth and development, such as nitrogen (N). This nutrient is
necessary for plants, between 2 and 5% dry weight being required
for optimal development and growth. 12 On the other hand, nitrate
(NO − 3 ) is the main form of N taken up by the roots from the soil,13
and, once taken up, this anion is translocated to the leaves, where
it is stored in the vacuoles and/or transformed into assimilation
products such as amino acids and proteins required for biomass
formation. 14 Also, an excess of foliar NO − 3 can compromise crop
quality, since this anion represents risks for human health, 15
because after ingestion it is rapidly transformed into nitrites and
nitrogenous compounds, which are toxic and can cause serious
pathological disorders such as methaemoglobinaemia and blue
baby syndrome and increase the risk of cancer. 16,17
There are few reports on the influence of Se application on
N metabolism. Aslam et al. 18 reported a decline in the uptake
and reduction of NO − 3 in barley plants after Se application as
either selenate or selenite. Munshi et al., 19 by applying selenite to
potato tubers, found an increase in the content of amino acids
and proteins in the plant organs. More recent studies by Nowak
et al. 20,21 have demonstrated that Se influences, to a greater or
lesser degree, the enzymes nitrate reductase (NR) and nitrite
∗ Correspondence to: Juan J Rios, Departamento de Fisiología Vegetal, Facultad
de Ciencias, Universidad de Granada, E-18071 Granada, Spain.
E-mail: jjrios@ugr.es
Department of Plant Physiology, Faculty of Science, University of Granada,
E-18071 Granada, Spain
J Sci Food Agric 2010; 90: 1914–1919 www.soci.org c○ 2010 Society of Chemical Industry

Response of nitrogen metabolism to selenium in lettuce www.soci.org
reductase (NiR), diminishing their activity and therefore negatively
affecting N metabolism in general. In those reports a decline in
provoked by Se application implies a lower activity of the
NO − 3
enzymes of the glutamine synthetase (GS)/glutamate synthase
(GOGAT) cycle and consequently a decline in the quantity of endproducts
such as amino acids and proteins, thus depressing plant
growth.
In short, the present study seeks to analyse the way in which
Se, under different forms and application rates, influences the
reduction of NO − 3 , the assimilation of ammonium (NH+ 4 )andthe
concentration of various nitrogenous organic compounds.
MATERIAL AND METHODS
Plant material and growing conditions
Seeds of Lactuca sativa L. cv. Philipus were germinated (∼250)
and grown for 35 days in perlite-filled cell flats (cell size 3 cm
× 3cm × 10 cm) placed on benches in an experimental
greenhouse (Saliplant SL, Motril, Granada, Spain). The 35-dayold
seedlings were transferred to a cultivation chamber under
controlled environmental conditions with a relative humidity of
60–80%, a day/night temperature of 25/15 ◦ C and a 12/12 h
photoperiod at a photosynthetic photon flux density (PPFD) of
350 µmol m −2 s −1 , measured at the top of the plants with a 190
SB quantum sensor (LI-COR Inc., Lincoln, NE, USA). The plants
were grown in individual pots (25 cm upper diameter, 17 cm lower
diameter, 25 cm height, 8 L volume) filled with vermiculite. Until
the end of the experiment the plants received a growth solution
composed of 4 mmol L −1 Ca(NO3)2, 6 mmol L −1 KNO3, 2 mmol L −1
MgSO4 · 7H2O, 1 mmol L −1 NaH2PO4 · 2H2O, 50 µmol L −1 H3BO3,
2 µmol L −1 MnCl2 · 4H2O, 1 µmol L −1 ZnSO4 · 7H2O, 0.1 µmol L −1
Na2MoO4 · 2H2O, 0.25 µmol L −1 CuSO4 · 5H2O and10µmol L −1
iron ethylenediamine-N,N ′ -bis(2-hydroxyphenylacetic acid) (Fe-
EDDHA). The nutrient solution (pH 5.5–6) was renewed every
3 days and the vermiculite partly rinsed with Millipore-filtered
water in order to avoid nutrient accumulation.
At 45 days after germination the different treatments (Se doses
and forms) were applied together with the nutrient solution
described above and maintained for 21 days. The treatments
consisted of the application of Se at rates of 5, 10, 20, 40, 60,
80 and 120 µmol L −1 as Na2SeO4 or Na2SeO3. The different Se
application rates were chosen to cover a broad range, from
the recommended rate for biofortification programmes to rates
causing phytotoxicity. 9 In addition to these treatments, a control
treatment consisted of applying the complete growth solution
without Se supplementation. The experimental design was a
randomised complete block with 15 treatments, arranged in
individualpotswithsixplantspertreatment,andthreereplications.
The experiment was repeated three times under the same
conditions (n = 27).
Plant sampling
Lettuce leaves were sampled on day 66 after sowing. Leaf samples
were standardised by using only fully expanded leaves from the
middle part of plants in each replicate, as these reflect most clearly,
from the nutritional and metabolic standpoint, the effect of the
treatment applied. 9 The material was rinsed three times in distilled
water after disinfection with 10 mL L −1 non-ionic detergent and
then blotted on filter paper. At each sampling, fresh matter was
used for the analysis of enzyme (NR, NiR, GOGAT and GS) activities.
The rest of the plant material (edible leaves) was lyophilised and
used to determine the biomass and the NO − 3 ,NH+ 4
reduced N concentrations.
and total
Analysis of nitrogen forms
The total reduced N concentration was analysed after digestion
of 0.15 g of dry and milled leaf material with sulfuric acid (5 mL
at 980 mL L −1 ) and hydrogen peroxide (300 mL L −1 ). To measure
organic N, after digestion and dilution with deionised water, an
aliquot of 1 mL was added to a reaction medium containing buffer
(50 mL L −1 potassium sodium tartrate, 100 mmol L −1 sodium
phosphate and 54 g L −1 sodium hydroxide), 150 g L −1 sodium
salicylate/0.3 g L −1 sodium nitroprusside and 53.5 mL L −1 sodium
hypochlorite. Samples were incubated at 37 ◦ Cfor15min,and
organic N was measured by spectrophotometry. 22
NO − 3 and NH+ 4 were analysed from an aqueous extract of 0.2 g of
dried and ground leaf material in 10 mL of Millipore-filtered water.
determination and added to
A 100 µL aliquot was taken for NO − 3
100 g L−1 salicylic acid in sulfuric acid at 960 mL L−1 ,andtheNO − 3
concentration was measured by spectrophotometry according to
Cataldo et al. 23 NH + 4 was determined using the method described
by Krom. 24
Enzyme extractions and assays
Leaves were ground in a mortar at 0 ◦ C in 50 mmol L −1 KH2PO4
buffer(pH7.5)containing2mmolL −1 ethylenediamine tetraacetic
acid (EDTA), 15 g L −1 soluble casein, 2 mmol L −1 dithiothreitol
(DTT) and 10 g L −1 insoluble polyvinylpolypyrrolidone. The homogenate
was filtered and then centrifuged at 30 000 × g for
20 min. The resulting extract (cytosol and organelle fractions) was
used to measure enzyme activities. The extraction medium was
optimised for the enzyme activities so that they could be extracted
jointly according to the same method. 25–28
The NR (EC 1.6.6.1) assay followed the methodology of Kaiser
and Lewis. 27 In a final volume of 2 mL the reaction mixture
contained 100 mmol L −1 KH2PO4 buffer (pH 7.5), 100 mmol L −1
KNO3, 10 mmol L −1 cysteine, 2 mmol L −1 nicotinamide adenine
dinucleotide (NADH) and enzyme extract. Incubation was carried
out at 30 ◦ C for 30 min and stopped by the addition of
1molL −1 zinc acetate. The nitrite formed was determined
colorimetrically at 540 nm after coupling with sulfanilamide
and naphthylethylenediamine dihydrochloride according to the
method of Hageman and Hucklesby. 29
NiR (EC 1.7.7.1) activity was determined by the disappearance
of NO − 2 from the reaction medium. 26 The reaction mixture
contained 50 mmol L −1 KH2PO4 buffer (pH 7.5), 20 mmol L −1
KNO2, 5 mmol L −1 methyl viologen, 300 mmol L −1 NaHCO3 and
0.2 mL of enzyme extract. Following incubation at 30 ◦ Cfor30min,
the nitrite content was determined colorimetrically as above. 29
GS (EC 6.3.1.2) activity was determined by the hydroxamate
synthetase assay adapted from Kaiser and Lewis. 27
The reaction mixture was composed of 100 mmol L −1 KH2PO4
buffer (pH 7.5), 4 mmol L −1 EDTA, 1 mol L −1 L-sodium glutamate,
450 mmol L −1 MgSO4 ·7H2O, 300 mmol L −1 hydroxylamine,
100 mmol L −1 adenosine triphosphate (ATP) and enzyme extract.
Two controls were prepared, one without glutamine and the
other without hydroxylamine. Following incubation at 28 ◦ Cfor
30 min, the formation of glutamylhydroxamate was determined
colorimetrically at 540 nm after complexing with acidified ferric
chloride. 30
GOGAT (EC 1.4.1.14) activity was assayed spectrophotometrically
at 30 ◦ C by monitoring the oxidation of NADH at 340 nm,
J Sci Food Agric 2010; 90: 1914–1919 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1915

1916
g DW/ shoot
3
2.5
2
1.5
1
0.5
0
Selenite LSD 0.05 0.21 Selenate LSD 0.05 0.12
0 5 10 20 40 60 80 120
µmol L-1 Figure 1. Shoot biomass in lettuce plants subjected to different forms and
doses of Se. Bars represent mean ± standard error (n = 27).
essentially as indicated by Groat and Vance 25 and Singh and
Srivastava, 28 always within 2 h of extraction. The reaction mixture
consisted of 50 mmol L −1 KH2PO4 buffer (pH 7.5), 1 mL L −1 mercaptoethanol,
1 mmol L −1 EDTA, 18.75 mmol L −1 2-oxoglutarate,
75 mmol L −1 L-glutamine and enzyme extract. Two controls, one
without ketoglutarate and the other without glutamine, were
used to correct for endogenous NADH oxidation. The decrease in
absorbance (linear for at least 10 min) was recorded for 5 min.
Statistical analysis
Data were subjected to one-way analysis of variance (ANOVA)
at 95% confidence using Statgraphics 6.1 (Statistical Graphics
Corporation, Warrenton, USA). Two-way ANOVA was applied to
ascertain whether the Se doses and forms applied significantly
affected the results, and means were compared by Fisher’s least
significant difference (LSD) test. The significance levels for both
analyses were expressed as follows: ∗ P < 0.05; ∗∗ P < 0.01;
∗∗∗ P < 0.001; NS, not significant.
RESULTS
Shoot biomass showed significant differences depending on the
form and rate of Se application, higher shoot yield generally
occurring when selenate was applied rather than selenite
(P < 0.001) (Fig. 1). The highest values of shoot biomass were
registered at 20 µmol L−1 for selenate and 5 µmol L−1 for selenite.
These concentrations resulted in greater plant growth than in
the control. On the other hand, it should be highlighted that
the application of selenate at a dose of 80 µmol L−1 or higher
reduced the shoot biomass compared with control plants, while
this biomass reduction was triggered at a dose of only 10 µmol L−1 for selenite. For both forms of Se the minimum production of shoot
biomass was found after the application of 120 µmol L−1 .
Foliar NO − 3 concentration declined as the Se application rate
increased, the drop being more drastic for selenite (Fig. 2). The
lowest NO − 3
concentration was found with the highest Se rate
applied for both forms, while the highest concentration was
found in control plants (P < 0.001). Furthermore, the NO − 3
concentration was higher when the Se form applied was selenate
(P < 0.001). NR activity was also influenced by application of this
trace element. Both Se forms elevated NR activity, with significant
differences between different rates and forms applied (Table 1).
When selenite was applied, the highest NR activity was registered
in the 20 µmol L −1 treatment, while in the case of selenate the
highest NR activity appeared in the 10 µmol L −1 treatment. It
www.soci.org JJ Rios et al.
mg NO 3 - kg -1 FW
5000
4500
4000
3500
3000
2500
2000
1500
1000
500
0
Selenite LSD 0.05 125 Selenate LSD 0.05 138
0 5 10 20 40 60 80 120
µmol L -1
Figure 2. Concentration of NO − 3 in lettuce plants subjected to different
forms and doses of Se. Bars represent mean ± standard error (n = 27).
is noteworthy that in both cases, from these rates upwards, Se
application diminished NR activity. NR activity was higher on
applying selenite with respect to selenate. With regard to NiR
activity, Se had no effect irrespective of its rate and form (Table 1).
Foliar NH + 4 concentration showed no significant differences
either between the different forms of Se used or between
application rates (P > 0.05) (Fig. 3). In terms of the enzymes
involved in NH + 4
assimilation, GS activity increased as the Se rate
rose, with the highest value being registered at 120 µmol L −1 for
both Se forms, while selenite prompted the greatest boost in
GS activity (Table 2). In addition, GOGAT activity also proved to
be affected by this trace element, increasing with increasing Se
application rate (Table 2). Again, GOGAT activity was highest with
the application of selenite in this experiment.
With regard to total reduced N, the data reflected significant
differences between the various Se application rates (P < 0.001)
but not between the two forms applied (P > 0.05) (Fig. 4).
The highest total reduced N concentration was registered at
60 µmol L −1 for selenite and 80 µmol L −1 for selenate, the
assimilated quantity in all cases being higher than in control
plants.
DISCUSSION
In plants, stress depresses growth and development, provoking
mainly the generation of reactive oxygen species (ROS), which
damage numerous macromolecules and the cell structure.
Consequently, under adverse conditions, biomass is one of the
main parameters used to determine the stress undergone by
plants. 31,32 In general, the results indicate that selenate was less
toxic than selenite, since the plants tolerated selenate up to a level
of 60 µmol L −1 , while this effect for selenite occurred only up to a
level of 10 µmol L −1 (Fig. 1). Our data agree with those of Cartes
et al., 8 who reported higher phytotoxicity of Se in selenite form
with respect to selenate form. The phytotoxicity of selenite in our
work did not appear to be due to excessive foliar accumulation of
Se, since all selenate treatments led to higher Se concentrations
in the lettuce leaves than did selenite treatment. 9 However, Rios
et al. 33 investigated the influence of Se on oxidative metabolism,
finding that the selenite form led to a greater concentration of
hydrogen peroxide and lower activity of antioxidant enzymes, this
explaining the greater phytotoxicity and therefore, according to
our results, lower biomass production (Fig. 1).
The main form of N taken up by plants from the soil via the roots
isNO − 3 ,13 and,oncetranslocatedtotheleaves,thisanionisstoredin
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1914–1919

Response of nitrogen metabolism to selenium in lettuce www.soci.org
Table 1. Enzymatic activities of nitrate reductase (NR) and nitrite reductase (NiR) in lettuce plants subjected to different forms and doses of Se
NR activity NiR activity
Dose (µmol L −1 ) Selenite Selenate Selenite Selenate
0 0.036 ± 0.01 0.036 ± 0.01 19.48 ± 1.35 19.48 ± 1.35
5 0.089 ± 0.03 0.096 ± 0.05 19.51 ± 1.42 19.51 ± 1.08
10 0.075 ± 0.03 0.104 ± 0.06 19.50 ± 1.27 19.52 ± 1.18
20 0.110 ± 0.04 0.078 ± 0.04 19.57 ± 1.30 19.55 ± 1.37
40 0.070 ± 0.02 0.054 ± 0.03 19.53 ± 1.92 19.53 ± 1.18
60 0.097 ± 0.03 0.067 ± 0.04 19.52 ± 1.60 19.51 ± 1.25
80 0.098 ± 0.04 0.051 ± 0.03 19.52 ± 1.27 19.50 ± 1.23
120 0.083 ± 0.04 0.042 ± 0.02 19.51 ± 1.60 19.83 ± 1.30
P value ∗∗∗ ∗∗∗ NS NS
LSD 0.02 0.02 2.12 2.63
Statistical analysis
Form of Se (F) ∗∗∗ NS
Dose of Se (D) ∗∗∗ NS
F × D ∗∗∗ NS
LSD 0.01 2.15
NR activity expressed as µmol NO − 2 formed g−1 FW min −1 . NiR activity expressed as µmol NO − 2 reduced g−1 FW min −1 . Levels of significance:
∗∗∗ P < 0.001; NS, not significant. Values represent mean ± standard error (n = 27).
mg NH 4 + g -1 DW
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
Selenite LSD 0.05 0.15
Selenate LSD 0.05 0.13
0 5 10 20 40 60 80 120
µmol L-1 Figure 3. Concentration of NH + 4 in lettuce plants subjected to different
forms and doses of Se. Bars represent mean ± standard error (n = 27).
the vacuoles and/or transformed into assimilation products such
as amino acids and proteins required for biomass formation. 14
In addition, excess NO − 3 in the leaves reduces crop quality,
because, after ingestion, NO − 3 is transformed into nitrites and
nitrogenous compounds, which are toxic and can trigger a series
of pathological disorders. 15–17 Furthermore, it should be taken
into account that lettuce is a hyperaccumulator of NO − 3 ,34 and in
this sense the European Commission has imposed a maximum
limit on the concentration of this anion in lettuce plants for human
consumption of 4500 mg NO − 3 kg−1 fresh weight (FW). As a result,
the influence of Se application on the uptake and concentration
of this element in these plants is very important. Thus the data in
Fig. 2 show a decline in NO − 3 concentration after the application of
Se, primarily in the form of selenite, as reported in a previous study
by Aslam et al. 18 in barley plants. The lower foliar concentration of
NO − 3 afterSeapplicationmaybefortworeasons.First,moresharply
in the case of selenite application than of selenate, there may be
a negative effect on the membrane transporters. 33 The reduction
in foliar NO − 3 concentration prompted by selenate could be due
to an antagonistic effect between the two anions. The second
reason could be the induction of NO − 3 assimilation by NR activity
stimulated by the application of both forms of Se.
The first step in NO − 3 assimilation is its reduction to nitrite
(NO − 2 ) by the enzyme NR, the activity of which is regulated
concentration in the leaves. This stage of N
mainly by the NO − 3
assimilation is defined as the most important and limiting step in
this physiological process. 35 In this sense, as reflected in Table 1,
Se application induced NR activity, although this stimulation
was not the same for both Se forms, selenite being superior
to selenate in this respect (Table 1). Our data reflect a reduction in
NR activity at the highest application rates of Se, as observed in
other studies, where high selenite and/or selenate concentrations
have been applied to different plants such as barley, 18 wheat20 and sunflower. 36 Those studies indicate an inhibition of NR when
high concentrations were applied. Also, later studies hold that
selenite, when applied at low concentrations (similar to those in
our work), induces NR. 21 In short, these results indicate a possible
induction by Se, mainly in the form of selenite, with respect to NR
when the application rates are not high.
The second step in NO − 3 assimilation is the reduction of NO−2 to NH + 4 by the enzyme NiR.35 Despite it being considered a
constitutive enzyme, 13 other research indicates a toxic effect on
NiR caused by Se applied at high concentrations, showing a
decline in its activity. 18,36 However, our results do not reflect a
lowering of NiR activity at higher concentrations of applied Se
(Table 1). In short, according to our results, the decline in the foliar
concentration of NO − 3 could be explained in part by the NR activity
induced by application of the different rates and forms of Se.
The reduction of NO − 3 produces NH+ 4 , which is subsequently
incorporated into organic forms. 37 Our results with respect to foliar
NH + 4 concentration suggest that there is a constant flow between
the formation and the removal of this cation, as the data do
not indicate significant differences in foliar concentrations (Fig. 3).
Also, NH + 4 is transformed into organic nitrogenous compounds by
the GS/GOGAT enzymatic cycle. 35,37 Thefirstenzymeinthiscycle,
GS, controls the association of NH + 4 with a glutamate molecule,
forming a molecule of glutamine. 13 This glutamine can either be
J Sci Food Agric 2010; 90: 1914–1919 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1917

1918
www.soci.org JJ Rios et al.
Table 2. Enzymatic activities of glutamine synthetase (GS) and glutamate synthase (GOGAT) in lettuce plants subjected to different forms and doses
of Se
GS activity GOGAT activity
Dose (µmol L −1 ) Selenite Selenate Selenite Selenate
0 1.16 ± 0.11 1.16 ± 0.11 11.58 ± 0.92 11.58 ± 0.86
5 1.24 ± 0.13 1.24 ± 0.12 13.71 ± 0.95 11.21 ± 0.82
10 1.41 ± 0.13 1.29 ± 0.11 12.87 ± 1.01 13.04 ± 0.95
20 1.75 ± 0.16 1.44 ± 0.14 12.81 ± 0.87 11.43 ± 0.90
40 1.85 ± 0.17 1.74 ± 0.17 12.89 ± 0.96 12.50 ± 0.92
60 1.46 ± 0.17 1.69 ± 0.16 13.22 ± 0.90 12.36 ± 0.92
80 1.85 ± 0.21 1.51 ± 0.16 15.55 ± 1.12 13.02 ± 0.95
120 2.37 ± 0.25 1.79 ± 0.18 14.97 ± 1.07 12.52 ± 0.91
P value ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗
LSD 0.18 0.16 0.83 0.69
Statistical analysis
Form of Se (F) ∗∗ ∗∗∗
Dose of Se (D) ∗∗∗ ∗∗∗
F × D ∗∗∗ ∗∗∗
LSD 0.14 0.45
GS activity expressed as µmol γ -glutamylhydroxamate formed g −1 FW min −1 . GOGAT activity expressed as µmol NADH oxidised g −1 FW min −1 .
Levels of significance:
∗∗ P < 0.01; ∗∗∗ P < 0.001; NS, not significant. Values represent mean ± standard error (n = 27).
exported from the cell or used in the formation of other amino
acids and proteins, or else serve as a substrate for α-ketoglutarate,
forming two glutamates. 37,38
Previously, Ruiz et al. 36 reported that the application of Se in
the form of selenate diminished the activities of GS and GOGAT
in sunflower plants. In the present work we find an induction of
the activity of both enzymes with Se application. The difference
between the data reported by Ruiz et al. 36 andthosefoundinthe
present work may be due to the application rate used, since Ruiz
et al. used higher rates. Also, they worked with sunflower plants,
which may be more sensitive than lettuce. These results, together
with the induction of NR activity, imply that there was an increase
in the requirements of nitrogenous compounds on the part of
the plant when this trace element was applied. In fact, it could
be explained by the increase in sulfur (S) metabolism caused in
these plants when Se is applied, mainly in the form of selenite. 39
The increase in S assimilation under these conditions would have
repercussions on the need for nitrogenous organic compounds
by the plant, 40 which would explain the induction of the enzymes
of N assimilation in our work.
Finally, the result of N assimilation can be noted by analysing
the concentrations of amino acids, proteins and total reduced N.
Both amino acids and proteins 39 as well as total reduced N (Fig. 4)
showed an increase with the application of both forms of Se,
confirming the effect of Se in inducing N metabolism in our work.
CONCLUSION
Se applied in both forms increased N metabolism, selenite being
the form that more strongly induced this physiological process,
since it prompted greater activation of the enzymes NR, GS and
GOGAT as well as a greater concentration of total reduced N.
These results could be explained by the coordination between N
and S metabolisms. Thus the application of Se, especially selenite,
induces the assimilation of S and Se, 39 stimulating the metabolism
mg total reduced
N g-1 DW
25
20
15
10
5
0
Selenite LSD 0.05 0.88
Selenate LSD0 0.05 0.70
0 5 10 20 40 60 80 120
µmol L-1 Figure 4. Concentration of total reduced N in shoots of lettuce plants
subjected to different forms and doses of Se. Bars represent mean ±
standard error (n = 27).
of N with the aim of maintaining the proper equilibrium between
nitrogenous organic compounds and sulfates.
Finally, the application of selenite, by inducing a greater
incorporation of Se, causes greater phytotoxicity than does
selenate, a result that would explain the biomass reduction despite
the increase in N assimilation.
ACKNOWLEDGEMENT
This work was supported by PAI (Plan Andaluz de Investigación,
AGR-161).
REFERENCES
1 Papp LV, Lu J, Holmgren A and Khanna KK, From selenium to
selenoproteins: synthesis, identity, and their role in human health.
Antiox Redox Signal 9:775–806 (2007).
2 Messaoudi I, El Heni J, Hammouda F, Said K and Kerkeni A, Protective
effects of selenium, zinc or their combination on cadmium induced
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1914–1919

Response of nitrogen metabolism to selenium in lettuce www.soci.org
oxidative stress in rat kidney. Biol Trace Elem Res 130:152–161
(2009).
3 Diwadkar-Navsariwala V, Prins GS, Swanson SM, Birch LA, Ray VH,
Hedayat S, et al, Selenoprotein deficiency accelerates prostate
carcinogenesis in a transgenic model. Proc Natl Acad Sci USA
103:8179–8184 (2006).
4 Pedrero Z, Madrid Y and Camara C, Selenium species bioaccessibility
in enriched radish (Raphanus sativus): a potential dietary source of
selenium. J Agric Food Chem 54:2412–2417 (2006).
5 Smokolji P, Pograje L, Hlaston-Ribic C and Stibilj V, Selenium content
in selected Slovenian foodstuffs and estimated daily intakes of
selenium. Food Chem 90:691–697 (2005).
6 Lyons G, Ortiz-Monasterio I, Stangoulis J and Graham R, Selenium
concentration in wheat grain. Is there sufficient genotypic variation
to use in breeding? Plant Soil 269:369–380 (2005).
7 Hartikainen H, Xue T and Piironen V, Selenium as an antioxidant and
pro-oxidant in ryegrass. Plant Soil 225:193–200 (2000).
8 Cartes P, Gianfreda L and Mora LM, Uptake of selenium and its
antioxidant activity in ryegrass when applied as selenate and
selenite form. Plant Soil 256:359–367 (2005).
9 Rios JJ, Rosales MA, Blasco B, Cervilla LM, Romero L and Ruiz JM,
Biofortification of Se and induction of the antioxidant capacity
in lettuce plants. Sci Hort 106:248–255 (2008).
10 White PJ and Broadley MR, Biofortifying crops with essential mineral
elements. Trends Plant Sci 10:586–593 (2005).
11 Chen L, Yang F, Xu J, Yun H, Hu Q, Zhang Y, et al, Determination of
selenium concentration of rice in China and effect of fertilization
of selenite and selenate on Se content of rice. J Agric Food Chem
50:5128–5130 (2002).
12 Miller AJ and Cramer MD, Root nitrogen acquisition and assimilation.
Plant Soil 274:1–36 (2005).
13 Barneix AJ, Physiology and biochemistry of source-regulated protein
accumulation in the wheat grain. J Plant Physiol 164:581–590
(2007).
14 Ruiz JM and Romero L, Cucumber yield and nitrogen metabolism in
response to nitrogen supply. Sci Hort 82:309–316 (1999).
15 Prasad S and Chetty AA, Nitrate-N determination in leafy vegetables:
study of the effects of cooking and freezing. Food Chem
106:772–780 (2008).
16 Mensinga TT, Speijers JGA and Meulenbelt J, Health implications of
exposure to environmental nitrogenous compounds. Toxicol Rev
14:576–584 (2003).
17 Santamaria P, Nitrate in vegetables: toxicity, content, intake and EC
regulation. J Sci Food Agric 86:10–17 (2006).
18 Aslam M, Harbit KB and Huffaker RC, Comparative effects of selenite
and selenate on nitrate assimilation in barley seeding. Plant Cell
Environ 13:773–782 (1990).
19 Munshi CB, Combs GF and Mondy NI, Effect of selenium on the
nitrogenous constituents of the potato. J Agric Food Chem
38:2000–2002 (1990).
20 Nowak J, Kaklewski K and Klódka D, Influence of various concentration
of selenic acid (IV) on the activity of soil enzymes. Sci Total Environ
29:105–110 (2002).
21 Nowak J, Kaklewski K and Ligocki M, Influence of selenium on
oxidoreductive enzymes activity in soil and in plants. Soil Biol
Biochem 36:1553–1558 (2004).
22 Baethgen WE and Alley MM, A manual colorimetric procedure for
measuring ammonium nitrogen in soil and plant. Commun Soil Sci
Plant Anal 20:961–969 (1989).
23 Cataldo DA, Haroon M, Schrader LE and Young VL, Rapid colorimetric
determination of nitrate in plant tissue by nitration of salicylic acid.
Commun Soil Sci Plant Anal 6:71–80 (1975).
24 Krom MD, Spectrophotometric determination of ammonia: study
of a modified Berthelot reaction using salicylate and
dichloroisocyanurate. Analyst 105:305–316 (1980).
25 Groat RG and Vance CP, Root nodule enzymes of ammonia
assimilation in alfalfa (Medicago sativa L.). Plant Physiol
67:1198–1203 (1981).
26 Lillo C, Diurnal variations of nitrite reductase, glutamine synthetase,
glutamate synthase, alanine aminotransferase and aspartate
aminotransferase in barley leaves. Physiol Plant 61:214–218 (1984).
27 Kaiser JJ and Lewis OAH, Nitrate reductase and glutamine synthetase
activity in leaves and roots of nitrate fed Helianthus annuus L. Plant
Soil 70:127–130 (1984).
28 Singh RP and Srivastava HS, Increase in glutamate synthase activity in
maize seedlings in response to nitrate and ammonium nitrogen.
Physiol Plant 66:413–416 (1986).
29 Hageman RH and Hucklesby DP, Nitrate reductase. Meth Enzymol
23:497–503 (1971).
30 Wallsgrove RM, Lea PJ and Miflin BJ, Distribution of enzymes of
nitrogen assimilation within the pea leaf cell. Plant Physiol
63:232–236 (1978).
31 Moran JF, Becana M, Iturbe-Ormaetxe I, Frechillas S, Klucas RV and
Aparicio-Tejo P, Drought induces oxidative stress in pea plants.
Planta 194:346–352 (1994).
32 Blasco B, Rios JJ, Cervilla LM, Sanchez-Rodriguez E, Ruiz JM and
Romero L,Iodinebiofortificationandantioxidantcapacityoflettuce:
potential benefits for cultivation and human health. Ann Appl Biol
152:289–299 (2008).
33 Rios JJ, Blasco B, Cervilla LM, Rosales MA, Sanchez-Rodriguez E,
Romero L, et al, Production and detoxification of H2O2 in lettuce
exposed to selenium. Ann Appl Biol 154:107–116 (2009).
34 MAFF, 1997/8 UK monitoring programme of nitrate in lettuce and
spinach. Food Surveillance Information Sheet No. 154, Ministry of
Agriculture, Fisheries and Food, London (1998).
35 Ruiz JM, Rivero RM, Garcia PC, Baghour M and Romero L, Role of
CaCl2 in nitrate assimilation in leaves and roots of tobacco plants
(Nicotiana tabacum L.). Plant Sci 141:107–115 (1999).
36 Ruiz JM, Rivero RM and Romero L, Comparative effect of Al, Se and
Mo toxicity on NO3 assimilation in sunflower (Helianthus annuus L.)
plants. J Environ Manag 83:207–212 (2007).
37 Temple SJ, Vance CP and Grant JS, Glutamate synthase and nitrogen
assimilation. Trends Plant Sci 3:51–56 (1998).
38 Lea PJ and Azevedo RA, Nitrogen use efficiency. 1. Uptake of nitrogen
from the soil. Ann Appl Biol 149:243–247 (2006).
39 Rios JJ, Blasco B, Cervilla LM, Rubio-Wilhelmi MM, Ruiz JM and
Romero L, Regulation of sulphur assimilation in lettuce plants in the
presence of selenium. Plant Growth Regul 56:43–51 (2008).
40 Gerendás J, Sailer M, Fendrich ML, Stahl T, Mersh-Sundermann V and
Mühling KM, Influence of sulfur and nitrogen supply on growth
nutrient status and concentration of benzyl-isothiocyanate in cress
(Lepidium sativum L.). J Sci Food Agric 88:2576–2580 (2008).
J Sci Food Agric 2010; 90: 1914–1919 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1919

1920
Research Article
Received: 13 October 2009 Revised: 4 May 2010 Accepted: 9 May 2010 Published online in Wiley Interscience: 14 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4033
A thermostable lectin from the rhizomes
of Kaempferia parviflora
Wichchulada Konkumnerd, a Aphichart Karnchanatat b
and Polkit Sangvanich c∗
Abstract
BACKGROUND: Kaempferia parviflora, or black galingale (Kra-Chai-Dam), belongs to the Zingiberaceae family and is used as
both a food ingredient and a medicinal plant. There are diverse reports on the biological activities of compounds extracted
from the plant, such as antimalarial, antifungal and an effective sexual-enhancing role, but not on the lectins.
RESULTS: A lectin was isolated from the rhizomes of Kaempferia parviflora using affinity chromatography on Concanavalin A
followed by gel filtration chromatography on Sephacryl S-100. The molecular weight of the purified lectin was about 41.7 kDa.
This lectin showed haemagglutinating activity against erythrocytes from several sources, with the highest level being against
those from rabbits. Moreover, the lectin was thermostable, with significant haemagglutinating activity detectable up to 75 ◦ C.
The results of trypsin digestion and liquid chromatography/tandem mass spectrometry analysis suggested that this protein
could be a member of the lectin/endochitnase1 family.
CONCLUSION: A lectin that showed thermotolerant haemagglutinating activity against erythrocytes from several sources
was successfully purified from K. paviflora rhizomes. Peptide sequence analysis indicated that this lectin is similar to
lectin/endochitinase 1 (Urtica dioica) or Hevein-like protein (Hevea brasiliensis).
c○ 2010 Society of Chemical Industry
Keywords: Kaempferia parviflora; thermostable; lectin; LC/MS/MS
INTRODUCTION
Lectins are polyvalent sugar-binding proteins or glycoproteins
of non-immune origin that have an apparently ubiquitous
distribution in nature, show a specificityfor terminal or subterminal
carbohydrate residues and can agglutinate cells and/or precipitate
glycoconjugates. 1 The main characteristic of this class of proteins is
their abilityto interactwith carbohydrates and so combine with the
glycocomponents of the cell surface as well as with cytoplasmic
and nuclear structures and the extracellular matrix of cells and
tissues from throughout the five kingdoms. 2 The availability of
a large number of lectins with distinct carbohydrate specificities
has resulted in the use of these proteins as tools in medical and
biological research. 3 Moreover, they have attracted considerable
interest because of their remarkable effects in a wide range of
biological systems, including the purification and characterisation
of glycoconjugates and the study of cell surface architecture. 4
The family Zingiberaceae is a large, important and well-known
monocot family (ginger plants) that is conspicuous throughout the
tropics. It comprises approximately 52 genera and 1400 species,
with the centre of diversity being in South/Southeast Asia. In
Thailand there are 26 genera and more than 300 species of
these plants, 5 many of which are used in food and traditional
medicine. 6 Black galingale (Kaempferia parviflora Wall ex. Baker),
known locally in Thai as Kra-Chai-Dam, is a herbaceous plant within
the Zingiberaceae family, and its black to purple rhizomes have
traditionally been employed in local food as a flavouring agent and
as a folklore medicinal plant for the treatment of a wide spectrum
of illnesses. Indeed, since ancient times it has traditionally been
used as a health-promoting and vitalising agent. 7 For example,
its rhizomes or their ethanolic extracts are applied as a general
health-promoting agent, as an anti-inflammatory agent and for
the treatment of gastrointestinal disorders. 7,8 Their purported
therapeutic activities have attracted a great deal of interest
in recent years. A preliminary study on K. parviflora plantlets
revealed the presence of haemagglutinating activity against rabbit
erythrocytes in rhizome extracts. 9 In this paper we report the
purification and physicochemical characterisation of a lectin from
K. parviflora rhizomes as a prerequisite to any further study of this
lectin with respect to its physiological role in the plant.
∗ Correspondence to: Polkit Sangvanich, Research Center for Bioorganic
Chemistry, Department of Chemistry, Faculty of Science, Chulalongkorn
University, Bangkok 10330, Thailand. E-mail: polkit@gmail.com
a Program of Biotechnology, Faculty of Science, Chulalongkorn University,
Bangkok 10330, Thailand
b Institute of Biotechnology and Genetic Engineering, Chulalongkorn University,
Bangkok 10330, Thailand
c Research Center for Bioorganic Chemistry, Department of Chemistry, Faculty of
Science, Chulalongkorn University, Bangkok 10330, Thailand
J Sci Food Agric 2010; 90: 1920–1925 www.soci.org c○ 2010 Society of Chemical Industry

Thermostable lectin from Kaempferia parviflora www.soci.org
MATERIALS AND METHODS
Biological materials
Fresh rhizomes of K. parviflora were purchased from a local market
in Bangkok, Thailand. A voucher specimen (BK51772) is deposited
at the Bangkok Herbarium of the Plant Variety Protection Division,
Department of Agriculture, Bangkok, Thailand. Human blood
samples (ABO system) were obtained from healthy donors at the
Thai Red Cross Society, Bangkok, Thailand. All other non-human
animal blood was supplied by the Division of Production and
Supply, National Laboratory Animal Center, Mahidol University,
Nakhon Pathom, Thailand.
Chemicals and reagents
Concanavalin A Sepharose (ConA Sepharose), bovine serum albumen
(BSA) and α-glucosidase from Saccharomyces cerevisiae were
purchased from Sigma Chemicals Co. (St. Louis, Missouri, USA).
The Sephacryl S-100 size exclusion column was obtained from
Amersham Pharmacia Biotech (Uppsala, Sweden). Methyl-α-Dglucopyranoside
was purchased from Fluka (Steinheim, Germany).
The reagents used in sodium dodecyl sulfate polyacrylamide gel
electrophoresis (SDS-PAGE) were obtained from Plusone Pharmacia
Biotech (Uppsala, Sweden), except for the low-molecularweight
marker protein calibration kit, which was purchased from
Amersham Pharmacia Biotech (Amersham, UK). All other biochemicals
and chemicals used in the investigation were of analytical
grade.
Purification of lectin from rhizomes of K. parviflora
To purify the K. parviflora lectin, fresh rhizomes (1 kg wet weight)
were washed, cut into small pieces and homogenised in a
blender in TBS (20 mmol L −1 Tris-HCl buffer, pH 7.4, containing
0.15 mol L −1 NaCl) at 1.5 kg L −1 , then left to extract overnight
at 4 ◦ C with stirring. The insoluble material was subsequently
removed by filtration through two layers of cheesecloth, and the
filtrate was clarified by centrifugation at 15 000 × g for 30 min
at 4 ◦ C. To the clear filtrate (the crude TBS-soluble rhizome
extract) was then added ammonium sulfate to 80% saturation,
and the mixture was stirred overnight at 4 ◦ C to precipitate the
proteins. The precipitate, mostly as a suspension, was harvested
by centrifugation at 15 000 × g for 30 min at 4 ◦ C. The supernatant
was discarded and the pelleted proteins were dissolved in TBS and
dialysed against excess water. The dialysate was then adjusted
to ∼2mgmL −1 , and 10 mL aliquots were fractionated by affinity
chromatography on a 1.6 cm × 20 cm ConA Sepharose column
equilibrated and eluted with TBS at a flow rate of 60 mL h −1 .After
elution of the unbound proteins in the equilibrium buffer (i.e.
when the eluate A280 fell to

1922
A 280 (mAU)
300
200
100
0
www.soci.org W Konkumnerd, A Karnchanatat, P Sangvanich
20 mmol L -1 Tris-HCl (pH 7.4)
+ 150 mmol L -1 NaCl
+ 0.5 mol L -1 methyl a-Dglucopyranoside
(A)
0 50 100 150 200 250
Elution volume (mL)
A 280 (mAU)
15
10
5
0
0 10 20 30 40 50
Elution volume (mL)
Figure 1. (A) Affinity chromatogram of Kaempferia parviflora rhizome lectin on ConA Sepharose. The column was equilibrated and then washed with
TBS. The lectin was then eluted with TBS containing 0.5 mol L −1 methyl-α-D-glucopyranoside (arrow indicates addition of methyl-α-D-glucopyranoside)
asdescribedinthetext.(B)ElutionprofileofpurifiedK. parviflora lectin from Sephacryl S-100 gel filtration chromatography column. The elution profiles
shown are representative of at least three independent trials.
Internal amino acid sequence of lectin by liquid
chromatography/tandem mass spectrometry (LC/MS/MS)
The amino acid sequence of an internal fragment of the
purified lectin from K. parviflora rhizomes was determined
by in-gel trypsin digestion of the selected SDS-PAGE-resolved
protein band, as reported previously, 14 and sequencing of the
different tryptic peptides by LC/MS/MS. The selected band was
excised from the SDS-PAGE gel and washed with 30 mL L −1
hydrogen peroxide. The protein was in-gel reduced, alkylated
and digested with trypsin. After digestion the peptides were
subjected to two sequential extractions from the gel with
500 mL L −1 acetonitrile/1 g L −1 trifluoroacetic acid at 200 µLg −1
gel (wet weight), the two extracts subsequently being pooled and
air dried. The extracted tryptic peptides were then subjected
to LC/nano-electrospray ionisation (ESI)-MS/MS. All collected
LC/MS/MS data were processed and submitted to a MASCOT
search of an in-house NCBI database. The following criteria were
used in the MASCOT search: trypsin cleavage specificity with
up to three missed cleavage sites, cysteine carbamidomethyl
fixed modification, methionine oxidation variable modifications,
±0.2 Da peptide tolerance and MS/MS tolerance and ESI-TRAP
fragmentation scoring.
RESULTS AND DISCUSSION
Purification of lectin from rhizomes of K. parviflora
The crude TBS-soluble protein extract from the black rhizomes of
K. parviflora was first preciptated with 80% saturation ammonium
sulfate, which reduced the total protein level by 10% and resulted
in a slight (1.04-fold) increase in specific haemagglutinating
activity with a 6% yield loss. Indeed, haemagglutinating activity
(total and specific) was used as a surrogate lectin marker to monitor
all purification procedures. Following dialysis, ConA Sepharose
affinity chromatography was employed. All haemagglutinating activity
remained in the bound fraction, with no detectable activity
in the void elutant. In the presence of the competitor, 0.5 mol L −1
methyl-α-D-glucopyranoside, 80% of the haemagglutinating activity
was eluted (Fig. 1(A)), resulting in a further decrease in
the total protein of ∼99% and the effective loss of the purple
pigments, and thus a further 72-fold purification (Table 1). The
recovered bound fraction with haemagglutinating activity was
Table 1. SummaryofpurificationofTBS-soluble Kaempferiaparviflora
rhizome lectin
Purification step
Total
protein
(mg) a
Total
activity
(HU) b
Specific
activity
(HU mg −1 ) c
(B)
Yield
(%)
Purification
(fold) d
Crude extract 1958 25 600 13.1 100 1
80% sat.
(NH4)2SO4
precipitation
1758 24 000 13.7 94 1.04
Con A Sepharose
(bound
fraction)
19.5 19 200 985 75 75.3
Gel filtration
(Sephacryl
S-100)
8.72 10 240 1174 40 89.8
a Crude protein extract from 334 g wet weight of rhizomes.
b Haemagglutinating activity (HU) as the minimal concentration
of protein able to cause visible agglutination of a 20–40 mL L −1
suspension of rabbit erythrocytes.
c Specific activity is defined as the haemagglutination unit (HU) divided
by the protein concentration (mg mL −1 ) of the assay solution. Rabbit
erythrocytes were used for the assay.
d Thepurificationindexwascalculatedastheratiobetweentheminimal
concentration of the crude extract able to cause visible agglutination
of rabbit erythrocytes and that of the protein fraction obtained at each
purification step.
dialysed and concentrated and then resolved by Sephacryl S-100
gel filtration chromatography, giving the protein elution profile
shown in Fig. 1(B). Only one peak of protein eluted from the column,
and this final (potentially homogeneous) lectin preparation
was purified 89.8-fold with a 40% yield relative to the initial crude
extract (Table 1).
The protein from each step of the purification procedure was
analysed for purity by reducing SDS-PAGE (data not shown). After
ConA Sepharose affinity chromatography a single main protein
band on the SDS-PAGE gel was observed (Fig. 2), indicating that
the enriched lectin fraction obtained from the ConA Sepharose
column was a relatively pure lectin protein. Therefore, assuming
both purification to homogeneity and a constant relationship
between haemagglutinating activity and total lectin mass, the
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1920–1925

Thermostable lectin from Kaempferia parviflora www.soci.org
Figure 2. Reducing SDS-PAGE analysis of enriched Kaempferia parviflora
rhizome lectin purification: lane 1, molecular weight standards; lane 2,
35 µg of crude extract (homogenate); lane 3, 50 µg of 80% saturation
ammonium sulfate-precipitated fraction; lane 4, 20 µg of discarded
unbound fraction from ConA Sepharose; lane 5, 15 µg ofretainedConA
Sepharose-bound fraction; lane 6, 10 µg of peak with haemagglutinating
activity obtained from gel filtration on Sephacryl S-100.
total yield of this TBS-soluble K. parviflora rhizome lectin was
26.1 mg kg −1 (wet weight) rhizomes, which is about 0.45% (w/w)
of the total protein in the crude rhizome extract. However, allowing
for purification losses, the lectin itself likely accounts for ∼1.1%
(w/w) of the total rhizome protein.
From the total neutral carbohydrate analysis, this K. parviflora
rhizome lectin was found to contain 14.7% (w/w) sugar, which
is significantly higher than the levels reported previously for the
lectins from the Chinese evergreen chinkapin (5.8%), Helianthus
tuberosus L. tubers (5.3%) and Arundu donex (2.1%). 15–17 It remains
plausible that, during the enrichment procedures prior to and
during ConA Sepharose chromatography, residual endoglycanase
activity, in conjunction with the preferential binding of the natural
glycoprotein-rich lectin isoforms to the ConAresin, would selectfor
purified lectins of a lower carbohydrate content than the real level.
Conversely, we may have enriched for high-carbohydrate isoforms
by the use of the ConA Sepharose affinity chromatography. Note,
however, that against this latter point no haemagglutinating
activity was detected in the void or early elution fractions,
suggesting that any low-carbohydrate isoforms present are either
inactive or relatively rare.
Molecular weight determination
Discontinuous reducing SDS-PAGE revealed a single strong band
for the enriched K. parviflora lectin corresponding to an apparent
molecular weight of 41.7 kDa after Coomassie blue R-250 staining,
although another weaker band was potentially present at ∼39 kDa
(Fig. 2). A single band of the same apparent size was seen
under non-reducing conditions (not shown), suggesting that
the enriched lectin fraction could be a monomeric protein and
capable of haemagglutinating activity alone. The apparent size of
∼41.7 kDa is slightly larger than the previously published sizes of
most other plant lectins, which range from 30 to 35 kDa, 18–20 and is
more in agreement with that of the bacterial lectin from Escherichia
coli. 21 Regardless, this higher apparent molecular weight may
reflect the greater degree of observed glycosylation (see above).
Table 2. Haemagglutinating activity of purified Kaempferia parviflora
rhizome lectin against human and other animal erythrocytes
Blood type Total activity (HU)
Human A 640
Human B 640
Human AB 640
Human O 1280
Rabbit 19 200
Mouse 1200
Rat 4800
Guinea pig 1200
Goose 4800
Sheep 9600
The concentration of K. parviflora lectin used in these assays was
1mgmL −1 and was serially 1 : 1 (v/v) diluted. Data shown are mean ±
standard deviation and are derived from three repeats.
Assay for haemagglutinating activity
Kaempferia parviflora lectin powerfully agglutinated rabbit erythrocytes,
with, in order of haemagglutinating activity, rabbit ≫
sheep > rat = goose > mouse = guinea pig (Table 2). In addition,
it showed a weak but essentially non-specific agglutination
of human erythrocytes, where the A, B and AB groups showed
an equal agglutination activity and group O was only marginally
stronger than the other three human groups, but still weak and
on a par with that seen against mouse and guinea pig erythrocytes.
The low haemagglutinating activity of this lectin towards
mouse and guinea pig erythrocytes is similar to that seen with
the lectins from Pisum sativum and Bauhinia monandra seeds, 22,23
while the high haemagglutinating activity seen with rabbit erythrocytes
is broadly similar to that observed with the lectins from
Hevea brasiliensis and Canavalia cathartica. 24,25 However, the combined
species–activity profile of the lectin from K. parviflora is
somewhat unique. In terms of species specificity only, the lectin
from Trichosanthes anguina exhibited haemagglutinating activity
against all four human group erythrocytes as well as mouse,
rabbit and goat erythrocytes. 26 Examples of lectins with higher
and different species specificities include the seed lectins from
Lonchocarpus capassa and Galactia lindenii, which are specific for
human group O as well as rabbit erythrocytes, 27,28 but the two
differ in activities and the latter also agglutinates rat erythrocytes,
while the mannose/glucose-specific lectin from Chinese evergreen
chinkapin (Castanopsischinensis) seeds exhibited haemagglutinating
activity against mouse and rabbit erythrocytes. 17
Effect of temperature on lectin haemagglutinating activity
and stability
At 70 and 80 ◦ C the haemagglutinating activity of the enriched
K. parviflora rhizome lectin decreased to 40 and 60% respectively
of the corresponding activity at room temperature (Fig. 3(A)),
suggesting different conformational changes of the lectin under
these conditions. This phenomenon, although not the exact
temperature shift values, is quite similar to that seen for the
lectin from Phaseolus vulgaris, 29 which is extremely thermostable
at temperatures around 82 ◦ C, for the lectin from Astragalus
mongholicus, which is still active at 65 ◦ C, 30 and for the lectin
from C. cathartica, which is active at 60 ◦ Cbutnotat70 ◦ C. 25
Contrastingly, examples of thermolabile lectins, with respect
to haemagglutinating activity, include that from Psophocarpus
J Sci Food Agric 2010; 90: 1920–1925 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1923

1924
log 2 of Hemagglutination titre
10 (A)
8
6
4
2
0
www.soci.org W Konkumnerd, A Karnchanatat, P Sangvanich
30 40 50 60 70 80 90 100
Temperature ( ο C)
log 2 of Hemagglutination titre
10
8
6
4
2
0
0 50 100 150 200
Pre-incubation time (min)
Figure 3. (A) Effect of temperature on haemagglutinating activity of purified Kaempferia parviflora rhizome lectin towards rabbit erythrocyte suspension
in TBS. (B) Thermostability of same purified lectin towards rabbit erythrocyte suspension in TBS at 60 ◦ C(◦), 70 ◦ C(•), 80 ◦ C(�) and90 ◦ C(�). In both
parts, data are shown as mean ± standard deviation and are derived from three repeats. Full activity (100%) corresponds to a titre of 2 5 .
5 10
15 Accession Number
Kaempferia parviflora lectin G P I Q L S Y N Y N Y G P A G K
Lectin/endochitinase 1 (Urtica dioica) 240 G P I Q L T H N F N Y G L A G Q 255 P11218
Hevein-like protein (Sambucus nigra) 200 G P I Q L T H N F N Y G L A G E 215 Q9ZT61
Figure 4. Amino acid sequence from resolved tryptic fragments of enriched Kaempferia parviflora lectin, in comparison with other members of the lectin
family that showed the highest sequence homology in BLASTp searches of NCBI and SwissProt databases. Shaded regions represent regions of identity.
palustric, whose haemagglutinating activity declined rapidly when
heated above 50 ◦ C, being reduced to 50 and 0% at 60 and 70 ◦ C
respectively. 31 The lectins from the rhizomes of Smilax glabra,
Aspidistra elatior Blume and Arundo donax were very stable at
temperaturesupto50 ◦ Cbutshowedverylittlehaemagglutinating
activity above 80–85 ◦ C and no enhanced activity with increased
temperatures. 16,32 Together,thesedatasupportthenotionthatthe
haemagglutinating activity depends on the native conformation
of the protein. 33 Indeed, the activity of the lectin is related to
cations, just like the metal ions in Concanavalin A, which protect it
from proteolytic and temperature degradation. 34
In terms of haemagglutinating activity, the thermostability
of the purified K. parviflora rhizome lectin varied with both
incubation time and temperature (Fig. 3(B)). Pre-incubation at
70 and 80 ◦ C and to a lesser extent at 90 ◦ C for 30–60min
increased the subsequent haemagglutinating activity, with longer
pre-incubation times leading to a stable higher (70 and 80 ◦ C)
or broadly the same (90 ◦ C) final haemagglutinating activity level
as that seen with no pre-incubation. The markedly lower final
activity seen at 90 ◦ C compared with that seen at the other two
temperatures is presumably due to a lower enhanced activation
level in the early pre-incubation period. In contrast, pre-incubation
at 60 ◦ C rapidly decreased the haemagglutinating activity with
increasing pre-incubation time, until the lectin was fully inactive
after 90 min. The implication here is that the conformational
structure adopted by the lectin at 60 ◦ C is not suitable for
haemagglutination.
The Gibb’s free energy of thermostability at all incubation
temperatures and times was calculated and found to vary
significantly, with both positive and negative values. For example,
the �G values at 60 and 70 ◦ C pre-incubation for 30 min were 58.3
and −60.6calmol −1 respectively. Since �G = �H − T�S, ata
constant temperature �T is zero and �G depends on the �H term.
If �G is positive, energy is absorbed in a non-spontaneous process,
whereas, if �G is negative, energy is evolved in a spontaneous
process. Considering that the high haemagglutinating activity
seen at 70 ◦ C for 20 min corresponds to a �G of 59.8 cal mol −1 ,
the stability of the K. parviflora rhizome lectin is maximal at the
temperature where the entropies of the native and denatured
states are equal. Because many lectins exert antinutritional effects
on mammals, including humans, 35 the free energy of activation of
the lectin denaturation process is an important physicochemical
parameter to determine, especially considering the increasing
applications of lectins in agriculture, medicine and related areas
as well as human and animal nutrition.
Internal amino acid sequence of lectin by LC/MS/MS
The internal amino acid sequence of the purified lectin from
K. parviflora rhizomes was obtained by in-gel digestion with
trypsin and sequence analysis with LC/MS/MS. The main resolved
peptide sequence was GPIQL SYNYN YGPAG K (m/z 1742.91).
Comparisons were then made with all available protein sequences
in the SwissProt database using BLASTp searches. This sequence
is part of the conserved lysozyme-like superfamily domain that is
conserved in a fairly diverse array of proteins, including chitinases.
However, coupled with its haemagglutinating activity, the high
degree of internal amino acid sequence identity between this
peptide sequence and other lectins suggests that this protein
could be a member of the lectin/endochitnase 1 family, 36 as
showninFig.4.
CONCLUSION
The TBS-soluble lectin extracted from K. paviflora rhizomes was
successfully purified in two steps by affinity chromatography on
ConcanavalinAandgelfiltrationonSephacrylS-100,resultinginan
increase in the specific activity to 1174 HU mg −1 proteinwitha40%
yield. Reducing SDS-PAGE analysis yielded an estimated molecular
weight for this lectin of about 41.7 kDa. The lectin showed nonspecific
weak haemagglutinating activity on human (A, B, AB and
O) erythrocytes as well as stronger activity on non-human (rabbit
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1920–1925
(B)

Thermostable lectin from Kaempferia parviflora www.soci.org
≫ sheep > rat = goose > mouse = guinea pig) erythrocytes. The
optimal temperature for haemagglutinating activitywas 75–85 ◦ C,
suggesting that this K. paviflora lectin is thermally stable.
ACKNOWLEDGEMENTS
The authors thank Chulalongkorn University (graduate school
thesis grant), the Thai Government Stimulus Package 2 (TKK2555),
under the Project for Establishment of Comprehensive Center for
Innovative Food, Health Products and Agrigulture, Ratchadapisek
Somphot Endowment Fund (AG001B) for financial support of
this research, and the Institute of Biotechnology and Genetic
Engineering, Faculty of Science, Chulalongkorn University for
supportandfacilities.WealsothankDr.RobertButcher(Publication
Counselling Unit, Chulalongkorn University) for his constructive
comments in preparing this manuscript.
REFERENCES
1 Goldstein IJ, Hughes RC, Monsigny M, Osawa T and Sharon N, What
should be called a lectin? Nature 285:66 (1980).
2 Brooks SA and Leathem AJ, Expression of N-acetyl galactosaminylated
and sialylated glycans by metastases arising from primary breast
cancer. Invasion and Metastasis 18:115–121 (1998).
3 Singh RS, Tiwary AK and Kennedy JF, Lectins: sources, activities and
applications. Crit Rev Biotechnol 19:145–178 (1999).
4 LisH and SharonN, Biochemistry of Plants: a Comprehensive Treatise.
Academic Press, New York, NY, pp. 371–447 (1980).
5 Larsen K and Larsen SS, Gingers of Thailand. Queen Sirikit Botanic
Garden, Chiang Mai, pp. 1–130 (2006).
6 Sirirugsa P, Thai Zingiberaceae: species diversity and their uses. Pure
Appl Chem 70:2111–2118 (1998).
7 Yenjai C, Prasanphen K, Daodee S, Wongpanich V and Kittakoop P,
Bioactive flavonoids from Kaempferia parviflora. Fitoterapia
75:89–92 (2004).
8 Rujjanawate C, Kanjanapothi D, Amornlerdpison D and Pojanagaroon
S,Anti-gastric ulcereffect of Kaempferiaparviflora.JEthnopharmacol
102:120–122 (2005).
9 Sangvanich P, Kaeothip S, Srisomsap C, Thiptara P, Petsom A,
Boonmee A, et al, Hemagglutinating activity of Curcuma plants.
Fitoterapia 78:29–31 (2007).
10 Bradford MM, A rapid and sensitive method for the quantitation
of microgram quantities of protein utilizing the principle of
protein–dye binding. Anal Biochem 72:248–254 (1976).
11 Dubois M, Giles K, Hamilton JK, Rebers PA and Smith F, Colorimetric
method for determination of sugars and related substances. Anal
Chem 28:350–356 (1956).
12 Laemmli UK, Cleavage of structural proteins during the assembly of
the head of bacteriophage T4. Nature 227:680–685 (1970).
13 Dawes C, Circadian rhythms in human salivary flow rate and
composition. JPhysiol220:529–545 (1972).
14 Mortz E, Vorm O, Mann M and Roepstorff P, Identification of proteins
in polyacrylamide gels by mass spectrometric peptide mapping
combined with database search. Biol Mass Spectrom 23:249–261
(1994).
15 Suseelan KN, Mitra R, Pandey R, Sainis KB and Krishna TG, Purification
and characterization of a lectin from wild sunflower (Helianthus
tuberosus L.) tubers. Arch Biochem Biophys 407:241–247 (2002).
16 Kaur A, Singh J, Kamboj SS, Sexana AK, Pandita RM and
Shamnugavel M, Isolation of an N-acetyl-D-glucosamine specific
lectin from the rhizomes of Arundo donax with antiproliferative
activity. Phytochemistry 66:1933–1940 (2005).
17 Wong JH, Chan HYE and Ng TB, A mannose/glucose-specific lectin
from Chinese evergreen chinkapin (Castanopsis chinensis). Biochim
Biophys Acta 1780:1017–1022 (2008).
18 Franz H, Ziska P and Kindt A, Isolation and properties of three lectins
from mistletoe (Viscum album L.). Biochem J 195:481–484 (1981).
19 Van Damme EJM, Rougé BP, Van Leuven F and Peumans WJ, The
seed lectins of black locust (Robinia pseudoacacia) are encoded by
two genes which differ from the bark lectin genes. Plant Mol Biol
29:1197–1210 (1995).
20 Kamemura K, Furuichi Y, Umekawa H and Takahashi T, Purification
and characterization of a pod lectin from Great Northern bean.
Phaseolus vulgaris L. Biochim Biophys Acta 1289:87–94 (1996).
21 Basu S, Ghosh S, Ganguly NK and Majumdar S, A biologically active
lectin of enteroaggregative Escherichia coli. Biochimie 86:657–666
(2004).
22 Coelho LCBB and Silva MBR, Simple method to purify milligram
quantitiesofthegalactose-specificlectinfromtheleavesofBauhinia
monandra. Phytochem Anal 11:295–300 (2000).
23 Sitohy M, Doheim M and Badr H, Isolation and characterization of a
lectin with antifungal activity from Egyptian Pisum sativum seeds.
Food Chem 104:971–979 (2007).
24 Wittsuwannakul R, Wititsuwannakul D and Sakulborirug C, A lectin
from the bark of the rubber tree (Hevea brasiliensis). Phytochemistry
47:183–187 (1998).
25 Suseelan KN, Bhagwath A, Pandey R and Gopalakrishna T,
Characterization of Con C,alectin from Canavaliacathartica Thouars
seeds. Food Chem 104:528–535 (2009).
26 Shanavas K, Elyas K and Vasudevan D, Purification and some
properties of a lectin from the seeds of Trichosanthes anguina.
Biol Plant 37:417–422 (1995).
27 Joubert FJ, Sharon N and Merrifield EH, Purification and properties of a
lectin from Lonchocarpus capassa (apple-leaf) seed. Phytochemistry
25:323–327 (1986).
28 Almanza M, Vega N and Perez G, Isolating and characterising a
lectin from Galactia lindenii seeds that recognises blood group
H determinants. Arch Biochem Biophys 429:180–190 (2009).
29 Biswas S and Kayastha AM, Thermal stability of Phaseolus vulgaris
leucoagglutinin: a differential scanning calorimetry study. JBiochem
Mol Biol 35:472–475 (2002).
30 Yan Q, Jiang Z, Yang S, Deng W and Han L, A novel homodimeric lectin
from Astragalus mongholicus with antifungal activity. Arch Biochem
Biophys 422:72–81 (2005).
31 Adenike K,Kayode FAandAkintola A,Purificationandcharacterization
of a lectin from the seeds of Psophocarpus palustris. Pak J Biol Sci
8:1667–1678 (2005).
32 Xu X, Wu C, Liu C, Luo Y, Li J, Zhao X, et al, Purification and
characterization of a mannose-binding lectin from the rhizomes of
Aspidistra elatior Blume with antiproliferative activity. Acta Biochim
Biophys Sin 39:507–519 (2007).
33 Ng TB and Yu YL, Isolation of a novel heterodimeric agglutinin from
rhizomes of Smilax glabra, the Chinese medicinal material tufuling.
Int J Biochem Cell Biol 33:269–277 (2001).
34 Doyle RJ and Keller KF, Lectins in the clinical microbiology laboratory.
Clin Microbiol Newslett 8:157–159 (1986).
35 Vasconcelos IM and Oliveira JTA, Antinutritional properties of plant
lectins. Toxicon 44:385–403 (2004).
36 Lerner DR and Raikhel NV, The gene for stinging nettle lectin (Urtica
dioica agglutinin) encodes both a lectin and a chitinase. JBiolChem
267:11085–11091 (1992).
J Sci Food Agric 2010; 90: 1920–1925 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1925

1926
Research Article
Received: 23 December 2009 Revised: 16 April 2010 Accepted: 12 May 2010 Published online in Wiley Interscience: 22 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4036
Metabolite profiling of doenjang, fermented
soybean paste, during fermentation
Hye-Jung Namgung, a Hye-Jung Park, a In Hee Cho, a Hyung-Kyoon Choi, b
Dae-Young Kwon, c Soon-Mi Shim a∗ and Young-Suk Kim a∗
Abstract
BACKGROUND: A fermented soybean paste known as doenjang is a traditional fermented food that is widely consumed in
Korea. The quality of doenjang varies considerably by its basic ingredients, species of microflora, and fermentation process.
The classification of predefined metabolites (e.g. amino acids, organic acids, sugars and sugar derivatives, and fatty acids) in
doenjang samples according to fermentation was performed by using GC-FID and GC-MS data sets with the application of a
multivariate statistical method.
RESULTS: The predominantly produced amino acids included alanine, valine, leucine, isoleucine, proline, glutamine,
phenylalanine and lysine, showing remarkable increases in amounts during the later stages of fermentation. Carbonic
acid, citric acid, lactic acid and pyrogultamic acid were identified as the major organic acids. Significant amounts of erythrose,
xylitol, inositol and mannitol were detected during fermentation. Regarding fatty acids, relatively higher amounts of palmitic
acid, stearic acid, oleic acid, linoleic acid and linolenic acid were found in the doenjang at each fermentation time point. Principal
component analysis (PCA) successfully demonstrated changes in composition patterns as well as differences in non-volatile
metabolites according to fermentation period.
CONCLUSION: A set of metabolites could be determined representing the quality of doenjang during fermentation, and which
might also be correlated with taste ingredients, flavour, nutrition, and physiology activities that are claimed to be dependent
on the quality control of commercial doenjang.
c○ 2010 Society of Chemical Industry
Keywords: fermented soybean paste; metabolites; GC-MS; PCA; amino acids; organic acids; sugars and sugar derivatives; fatty acids
INTRODUCTION
The consumption of fermented foods has been gaining popularity
due to their healthful aspects. 1 A fermented soybean paste
known as doenjang is a traditional fermented food that is widely
consumed in Korea, and it has historical worth from a food-culture
point of view. 2 Doenjang has been an important source of protein
and flavouring agents for Koreans, and remains popular even
today. Doenjang is recognised as a nutritious food that provides
essential amino acids that are lacking in cereal protein diets, as
well as fatty acids, organic acids, minerals and vitamins, 3 which are
highly related to its quality and diverse nutritional benefits. The
quality of doenjang varies considerably by its basic ingredients,
species of microflora, and fermentation process. 4,5,6
As a traditional Korean fermented soybean paste, doenjang has
unique qualities in terms of its flavour and taste as compared
to other similar products from Asian countries such as Japan
and China. 7 This is due to the fact that the fermentation process
and natural environment, such as temperature and humidity, are
unique in Korea. 8 Doenjang is traditionally produced through the
fermentation of cooked soybeans (meju), by naturally occurring
bacteria and fungi such as Bacillus subtilis, Rizopus, Mucor and
Aspergillus species. 1,5 More recently, doenjang has been prepared
commercially by local manufactures using a slightly different
procedure. 4 Commercial products are prepared from cultivated
soybeans mixed with cooked rice or barely and partly fermented
with Aspergillus oryzae and Aspergillus sojae to form koji. 5,7,9
The multivariate analysis approach provides a potential means
for studying global changes in metabolite concentrations that
are related to variations in physiological or pathological status. 10
In particular, metabolite profiling, which aims to quantify several
pre-defined targets, has been used to understand changes in predefined
metabolites of diverse foods and biological systems. 11 One
of the most commonly used analytical approaches for metabolite
profiling is gas chromatography–mass spectrometry (GC-MS) after
derivatisation procedures. 10 It is less expensive compared to
other analytical approaches (e.g. CE-MS, LC-MS and NMR) and
has high reproducibility, high resolution and few matrix effects. 12
∗ Correspondence to: Soon-Mi Shim and Young-Suk Kim, Department of Food
Science and Technology, Ewha Woman’s University, 11-1 Daehyun-dong,
Seodaemun-gu, Seoul 120-750, South Korea. E-mail: yskim10@ewha.ac.kr;
soonmishim@gmail.com
a Department of Food ScienceandTechnology,EwhaWoman’s University, 11-1
Daehyun-dong, Seodaemun-gu, Seoul 120-750, South Korea
b College of Pharmacy, Chung-Ang University, Seoul 156-756, South Korea
c Korea FoodResearch Institute,516Baekhyun-Dong,Bundang-Ku,Sungnam-Si,
Gyeonggi-Do, 463-746, South Korea
J Sci Food Agric 2010; 90: 1926–1935 www.soci.org c○ 2010 Society of Chemical Industry

Metabolic profiling of doenjang www.soci.org
Since the datasets obtained by metabolite profiling are generally
large and complex, principal component analysis (PCA), which
is an unsupervised clustering method, is often employed. 13 The
overall goal of PCA is to describe datasets of numerical variables
with smaller sets of new and synthetic variables, called principal
components.
In this study, the profiling of pre-defined metabolites in
doenjang during fermentation was investigated by applying a
multivariate statistical method to GC-MS datasets. This approach
may be a useful tool to elucidate changes in diverse metabolites
suchasaminoacids,organicacids,fattyacids,andsugarsaccording
to the doenjang fermentation process.
MATERIALS AND METHODS
Preparation of doenjang
Soybeans (Korean Bactae, Glycine max L.) used in this study
were cultivated in Sunchang, Jeollanamdo, South Korea, in 2005.
The soybeans were immersed in water at 15 ◦ Cfor15hand
then steamed under steam pressure at 1.7 kgf cm −2 for 30 min
using a commercial steamer (Nippon Kikkoman type, Eunkwang
Machinery, Seoul, Korea). The soybeans were mashed and cooled
to 40 ◦ C before being formed into blocks, called meju. An aliquot
of meju was freeze-dried for preparation 0 day of sample. Meju
were then exposed to sunlight for 3 months for drying, as well
as for natural fermentation by microorganisms such as Bacillus
subtillis occurring in the air, by hanging them with a straw. The
meju was next washed by flowing water to remove mould on the
surface, put into large opaque pottery jars containing 26% brine,
and left for 2 months until the salinity dropped to 18%. Samples
of the meju immersed in brine solution were collected at 20-day
intervals during the 160 days of fermentation. The solid parts that
had separated from the liquid were mashed before being used as
specimens.
Analysis of non-volatile metabolites in doenjang
Amino acids
Two grams of specimen were mixed with 10 mL of methanol and
then heated at 70 ◦ C for 25 min. The mixture was subjected to
vortexing for 1 min, followed by additions of 10 mL of distilled
water and 5 mL of chloroform. The final mixture was centrifuged
twice at 1910 × g for 10 min. A 100 µL of norvaline as an internal
standardcompoundwasaddedinto100 µLoftheaqueousfraction
and then derivatised using an EZ : fastTM amino acid analysis kit
(Phenomenex, Inc., Phenomenex, Torrance, CA, USA). A Zebron
column (ZB-AAA, 10 m length × 0.25 mm film thickness i.d.) which
was connected to an HP 6890 series gas chromatograph with
a built-in FID for application (Hewlett-Packard, Palo Alto, CA,
USA) was used. The temperatures of the injector and detector
were 250 and 320 ◦ C, respectively. The oven temperature was
maintained at 110 ◦ C for 10 min, increased to 320 ◦ Catarate
of 30 ◦ Cmin −1 , and then maintained at 320 ◦ Cfor1min.The
split ratio was 1 : 15 at 250 ◦ C and the injection volume was
2.0 µL. Helium was used as a carrier gas at a constant flow
rate of 1.1 mL min −1 . The amino acids in the doenjang were
identified by comparing their retention times to those of standard
derivatives. The amounts of amino acids in the doenjang were
calculated by comparing their peak areas to those of the internal
standard compounds. The quantitative data were the mean values
of triplicate measurements.
Organic acids
Two grams of specimen, 10 mL of methanol, and 1 mL of salicylic
acid [1% (w/v) in methanol] as an internal standard, were
placed in a 60 mL vial prior to heating at 70 ◦ C for 25min.
After cooling at ambient temperature for 30 min, the mixture
was subjected to vortexing for 1 min along with additions
of 10 mL of distilled water and 5 mL of chloroform. It was
then centrifuged twice at 1910 × g for 10 min. The aqueous
fraction was extracted using the method described above and
then condensed to a final volume of 1 mL in a rotary vacuum
evaporator (EYELSA; Rikakai Co., Ltd, Tokyo, Japan). The moisture
was completely removed by using a vacuum drying oven (EYELA
NDO-600SD; Rikakikai). Trimethylsilyl (TMS) derivatisation was
performed in order to increase the volatility of the non-volatile
organic acids before GC-MS analysis. Three hundred microlitres of
bis-(trimethylsilyl) trifluoroacetamide (BSTFA; Supelco, Bellefonte,
PA, USA) containing 1% trimethylchlorosilane (TMCS) and 300 µL
of acetonitrile (Fisher Scientific Ltd., Ottawa, Ontario, Canada) were
added into the dehydrated specimen and then heated at 70 ◦ C
for 40 min followed by cooling to ambient temperature. A DB-5
column (30 m length × 0.25 mm i.d. × 0.25 µm film thickness;
J&W Scientific, Folsom, CA, USA) attached to an HP 6890 series
gas chromatograph that was connected to an HP 5975A mass
selective detector was employed. The temperatures of injector and
detector transfer line were 200 ◦ C and 250 ◦ C, respectively. The
oven temperature was maintained at 80 ◦ C for 10 min, increased
to 230 ◦ Catarateof3 ◦ Cmin −1 , and then maintained at 230 ◦ C
for 1 min. Helium was used as a carrier gas at a constant flow rate
of 0.8 mL min −1 , and the mass spectra were obtained at 70 eV
through the electron ionisation (EI) method. The injection volume
was 1.0 µL. The organic acids in the doenjang were identified by
comparing their retention times and mass spectra with those of
authentic standards. The amounts of organic acids in the doenjang
were calculated by comparing their peak areas with those of the
internal standard compounds. The quantitative data were the
mean values of triplicate measurements.
Sugars and sugar derivatives
The same procedure of sample preparation carried out for
determining organic acids was used for sugars and sugar
derivatives. One millilitre of salicylic acid [1% (w/v) in methanol]
wasusedasaninternalstandard.Trimethylsilyl(TMS)derivatisation
was performed in order to increase the volatility of the
sugars and sugar derivatives. Three hundred microlitres of
hexamethyldisilazane, 300 µL of TMCS and 400 µL of pyridine were
added to a completely dehydrated specimen, and then heated
at 85 ◦ C for 50 min. The sample was then cooled at ambient
temperature prior to being analysed by GC-MS.
GC-MS analytical method for quantification and identification of
sugars and sugar derivatives were same as in organic acid analysis.
Fatty acids
The fatty acids in the doenjang were analysed using method
described by Metcalfe et al. 14 Two grams of freeze-dried doenjang
specimen were mixed with 30 mL of chloroform and 15 mL
of methanol, followed by heating for 25 min at 75 ◦ Cusinga
reflux condenser. The mixture was cooled for 30 min at ambient
temperature and then washed twice with 20 mL of distilled water.
The solution was allowed to stand for 30 min for separation. Then,
the solution in the lower layer was retrieved for complete removal
of the solvent in a rotary evaporator (EYELSA; Rikakai). Fifteen
J Sci Food Agric 2010; 90: 1926–1935 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1927

1928
millilitres of 500 mol/L NaOH/MeOH, 5 mL of 14% BF3/methanol,
and 10 mL of n-heptane were added and heated for 5 min. The
fatty acid methyl esters (FAMEs) were extracted from the saturated
sodium chloride mixture (5 mL) with hexane (10 mL). The mixture
was then separated by placing it in a fractional funnel to retrieve
the organic portion in the upper phase. The solution was then
concentrated to a final volume of 1 mL with a gentle stream
of nitrogen gas after dehydration by anhydrous sodium sulfate.
One millilitre of isopentanoic acid [1% (w/v) in heptane] was
injected into the specimen as an internal standard compound
prior to carrying out GC-MS analysis. An HP 5890 series linked
to an HP 5972 series with a DB-FFAP column (30 m length ×
0.25 mm i.d. × 0.25 µm film thickness; J & W Scientific) was used
to carry out the analysis of fatty acids. The oven temperature
was maintained at 90 ◦ C for 2 min, increased to 240 ◦ Catarate
of 8 ◦ Cmin −1 , and then maintained at 240 ◦ Cfor10min.The
temperatures of injector and detector transfer line were 250
and 280 ◦ C, respectively. Helium was used as a carrier gas at a
constant flow rate of 0.8 mL min −1 , and the mass spectra were
obtained at 70 eV through the EI method. The injection volume
was 1.0 µL. The fatty acids in the doenjang were identified by
comparing their retention times and mass spectra with those of
authentic standards. The amounts of fatty acids in the doenjang
were calculated by comparing their peak areas with those of the
internal standard compounds. The quantitative data were the
mean values of triplicate measurements.
Statistical analysis
Analysis of variance (ANOVA) was performed using the general
linear model procedure in SPSS (version 12.0; SPSS, Chicago,
IL, USA) to evaluate significant differences in the non-volatile
metabolites including amino acids, organic acids, sugars and sugar
derivatives, and fatty acids occurring in the doenjang. Duncan’s
multi-rangetestwasusedwhenthesamplesexhibitedsignificantly
different peak areas of non-volatile metabolites, with the level of
significance set at P < 0.05. PCA was applied to raw values
(n = 3) of the relative peak areas of metabolites to clarify the
relationships between doenjang samples and the metabolites
present according to fermentation times.
RESULTS AND DISCUSSION
Profiling and analysis of non-volatile metabolites
The diverse non-volatile metabolites in the doenjang, such as
amino acids, organic acids, sugars, sugar derivatives, and fatty
acids, were elucidated by GC-FID or GC-MS after derivatisation.
Amino acids
The profiling of amino acids was performed throughout 160 days
of fermentation (Table 1). The main amino acids identified in the
doenjang were leucine, isoleucine, valine, phenylalanine, lysine,
glutamic acid, proalanine, alanine, tyrosin and histidine. After both
40 and 100 days of fermentation, the relative amounts of leucine,
phenylalanine, lysine and alanine were three times greater than
the initial fermentation period. These levels did not show further
increases but remained constant until 160 days of fermentation.
A similar pattern was shown for tyrosine and histidine, but their
levels decreased after 120 days of fermentation. Park et al. 8 found
that between 90 and 120 days of fermentation, glutamic acid,
which is related to the unique unami taste of doenjang, was a
major occurring amino acid followed by proline, leucine, alanin,
www.soci.org H-J Namgung et al.
and lysine. Other studies have also detected glutamic acid as the
most abundant amino acid in soybean paste during ripening and
storage, providing a savoury taste. 8,15 Lysine and alanine, which
provide a sweet taste, were predominantly found in the doenjang
after 160 days of fermentation. The amino acids responsible for
both sweet taste (glycine, alanine, serine and threonin) and unami
flavour (glutamic acid and asparagine) increased between 140
and 160 days of fermentation. Bitter taste components, namely
leucine and isoleucine, were also principal amino acids detected in
the fermented soybean paste. These amino acids exhibited similar
patterns, showing no significant increases or decreases in amounts
after 100 days of fermentation (Table 1).
According to a study by An et al. 16 the contents of amino acids
showed maximum levels at 60 days of fermentation, in which
glutamic acid content was the highest and cysteine content was
the lowest among them. In the current study, aminobutyric acid
(ABA) was detected after 40 days of fermentation and showed
its highest amount at 60 days of fermentation (Table 1). It was
not detected, however, in the later stages of fermentation.
Shizuka et al. confirmed this finding, reporting that ABA, abundant
in soybeans, decreased as fermentation processed. 17 ABA has
been recognised as a non-protein amino acid, biosynthesised
by the decarboxylation of L-glutamic acid through the reaction of
decarboxylase. 18 ABA, which is predominantly found in nature and
in soybeans, is produced in negligible amounts in doenjang during
fermentation. 17 ABAplays an importantrole in regulating neuronal
excitability throughout the nervous system. In addition, it has
shown biological activities for the treatment of depression, manicdepressive
disorder, seizures, premenstrual dysphoric disorder,
and anxiety. 19
Organic acids
Composition of organic acids in doenjang depends on the
microflora used in manufacturing the meju. These microflora
show variations according to inoculated microorganisms, their
total counts, and growing conditions that can affect organic
acid metabolism and energy efficiency by the microorganisms. 16
Organic acid analysis is a quality indicator because certain
identified compounds can greatly affect the quality of doenjang as
a flavouring agent through increases in acidity and sweet aroma
during fermentation. Although previous studies have analysed
several kinds of organic acids, 8,20,21 the current study identified
a more diverse range of organic acids including carbonic acid,
glucaric acid, glycolic acid, 2-ketoglutaric acid and mandelic acid.
The profile of 19 organic acids in doenjang during fermentation
is shown in Table 2. Citric acid, lactic acid and pyroglutamic acid
were the primary organic acids at the initial stage of fermentation
(0–40 days). Citric acid gradually decreased according to the
fermentation time. The results from our study suggest that citric
acid occurring in the soybeans was metabolised and converted
to acetic acid during fermentation, thus causing a significant
amount of acetic acid. The formation of acetic acid is considered
to provide an unpleasant flavour in fermented soy foods. 22 The
current study detected an insignificant amount of acetic acid in
the latter stage of fermentation (120–160 days), implying that this
doenjang would not possess an unpleasant flavour due to acetic
acid. For lactic acid and succinic acid, which are related to sourness,
significant increases were found at 60 days of fermentation along
with increasing patterns. Previous studies demonstrated similar
results stating that lactic acid was highly formed in doenjang
followed by acetic acid and citric acid. 8
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1926–1935

Metabolic profiling of doenjang www.soci.org
Table 1. Changes in the contents of amino acids in doenjang during fermentation
Relative peak area (mean ± SD) ∗
Compound 0 day 20 days 40 days 60 days 80 days 100 days 120 days 140 days 160 days ID †
ALA 3.631 ± 0.385a† 3.005 ± 0.317a 8.774 ± 0.470b 9.487 ± 1.036b 2.514 ± 5.028a 14.476 ± 1.282c 13.919 ± 0.423c 14.067 ± 1.428c 16.969 ± 2.828c A
SAR 0.067 ± 0.386a 0.005 ± 0.009a 0.048 ± 0.035a 0.078 ± 0.018a 0.051 ± 0.038a 0.048 ± 0.057a 1.505 ± 2.846a 0.032 ± 0.037a 7.351 ± 1.310b A
GLY 1.505 ± 0.387a 1.543 ± 0.197a 4.789 ± 0.192b 4.995 ± 0.669b 5.305 ± 0.347b 6.239 ± 0.405b 4.618 ± 3.041b 6.079 ± 0.550b 0.127 ± 0.019a A
ABA – ‡ – 0.036 ± 0.042ab 0.122 ± 0.050c 0.069 ± 0.053bc 0.080 ± 0.054bc – 0.095 ± 0.064bc – A
VAL 2.713 ± 0.296a 2.644 ± 0.187a 7.367 ± 0.528b 7.433 ± 0.572b 8.261 ± 0.339b 9.731 ± 0.555c 9.634 ± 0.516c 9.396 ± 0.408c 11.189 ± 1.275d A
LEU 6.060 ± 0.559a 6.664 ± 1.012a 16.380 ± 1.348b 16.971 ± 2.324b 18.416 ± 2.071bc 23.305 ± 1.146d 24.109 ± 3.591d 21.388 ± 2.440cd 24.092 ± 2.511d A
ILE 2.852 ± 0.348a 3.482 ± 0.892a 8.360 ± 0.587b 8.821 ± 1.053b 9.618 ± 0.691b 12.182 ± 0.518c 12.583 ± 1.656c 11.565 ± 1.078c 12.827 ± 0.946c A
THR 0.602 ± 0.245a 0.440 ± 0.153a 0.713 ± 0.866a – 2.563 ± 1.558b 3.043 ± 0.409b 4.534 ± 0.769c 4.885 ± 1.363c 3.614 ± 0.834bc A
SER 1.783 ± 0.190ab 1.221 ± 0.717a 2.849 ± 1.008b 2.618 ± 0.468b 1.723 ± 1.789ab 1.260 ± 0.177a 1.792 ± 0.308ab 2.888 ± 0.873b 2.067 ± 0.341ab A
PRO 3.715 ± 0.278a 2.241 ± 0.714a 7.111 ± 0.873b 6.460 ± 1.437b 7.381 ± 1.003a 6.454 ± 1.129b 6.513 ± 0.689b 7.281 ± 1.352b 9.006 ± 1.672c A
ASN 0.791 ± 0.097b 0.505 ± 0.110a 1.285 ± 0.064de 1.412 ± 0.132de 0.973 ± 0.193bc 1.155 ± 0.235cd 1.172 ± 0.186cd 1.205 ± 0.211cd 1.522 ± 0.318e A
TPR – 0.052 ± 0.020ab – 0.094 ± 0.016bc – 0.105 ± 0.014bc 0.106 ± 0.090bc 0.129 ± 0.026c 0.061 ± 0.071ab A
ASP 2.857 ± 0.142a 2.237 ± 0.597a 7.588 ± 1.271f 5.726 ± 1.408de 7.634 ± 1.181f 4.565 ± 0.879cd 4.050 ± 0.608bc 4.850 ± 0.978cde 6.294 ± 0.950ef A
MET 0.558 ± 0.163a 0.895 ± 0.095a 1.912 ± 0.277b 2.021 ± 0.228bc 2.347 ± 0.114cd 2.539 ± 0.292de 2.857 ± 0.300ef 2.534 ± 0.309de 3.072 ± 0.524f A
GLU 2.804 ± 0.136a 2.915 ± 0.129a 7.417 ± 1.152bc 4.857 ± 3.278a 7.600 ± 0.792bc 9.372 ± 1.329c 6.928 ± 0.708b 7.451 ± 0.989bc 9.420 ± 0.664e A
PHE 5.252 ± 0.666a 6.245 ± 1.384a 14.599 ± 1.678bc 13.651 ± 4.655b 16.325 ± 2.982bcd 21.528 ± 0.518e 21.955 ± 4.121e 18.676 ± 4.420cde 20.184 ± 1.417de A
AAA – 0.097 ± 0.025ab 0.223 ± 0.067bc 0.228 ± 0.063bc 0.233 ± 0.115bc 0.357 ± 0.071cd 0.515 ± 0.246d 0.462 ± 0.175d 0.365 ± 0.075cd A
ORN 0.561 ± 0.058a – – – – – – 7.714 ± 4.273b – A
LYS 4.883 ± 0.722a 6.098 ± 2.227a 16.452 ± 3.076ab 19.401 ± 6.871ab 22.762 ± 9.607bc 37.489 ± 2.943cd 40.723 ± 17.457d 36.506 ± 22.211cd 36.915 ± 4.379cd A
HIS 0.482 ± 0.114a 1.563 ± 0.725a 4.898 ± 0.761b 4.936 ± 1.266b 5.523 ± 2.348b 9.898 ± 0.739c 11.514 ± 3.929c 5.876 ± 3.416b 4.610 ± 0.402b A
TYR 2.219 ± 0.398a 3.463 ± 0.632a 8.580 ± 1.028cd 7.422 ± 1.202c 8.228 ± 1.319c 11.802 ± 1.015e 9.962 ± 1.701d 5.604 ± 1.544b 7.575 ± 0.382c A
TRP 0.555 ± 0.193a 1.436 ± 0.338a 3.599 ± 0.435b 3.291 ± 0.795b 3.404 ± 1.033b 6.328 ± 0.413c 6.312 ± 1.748c 5.398 ± 2.540c 4.851 ± 0.495bc A
ALA, alanine; SAR, sarcosine; GLY, glycine; ABA, aminobutyric acid; VAL, valine; LEU, leucine; ILE, isoleucine; THR, threonine; SER, serine; PRO, proline;ASN,asparagine;TRP,trytophan;ASP,asparticacid;
MET, methionine; GLU, glutamic acid; PHE, phenylalanine; AAA, aminoaldipic acid; ORN, ornithine; LYS, lysine; HIS, histidine; TYR, tyrosine; TPR, thioproline.
∗ Each value is the mean of the relative peak areas (n = 3) to that of an internal standard with the standard deviation. Values with different letters are significantly different at α = 0.05.
† Metabolites were identified on the basis of the following criteria: A, mass spectrum and retention time were consistent with those of an authentic standard).
J Sci Food Agric 2010; 90: 1926–1935 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
‡ Below the limit of detection.
1929

1930
Table 2. Changes in the contents of organic acids in doenjang during fermentation
Relative peak area (mean ± SD) ∗
Compound 0 day 20 days 40 days 60 days 80 days 100 days 120 days 140 days 160 days ID †
www.soci.org H-J Namgung et al.
Acetic acid – ‡ – – – – – 0.017 ± 0.005b 0.019 ± 0.003bc 0.025 ± 0.009c B
Carbonic acid 0.053 ± 0.006a† 0.166 ± 0.129abc 0.121 ± 0.027ab 0.270 ± 0.067bcd 0.340 ± 0.072cde 0.385 ± 0.209de 0.362 ± 0.099de 0.403 ± 0.105de 0.486 ± 0.085e B
Citric acid 1.015 ± 0.103d 1.120 ± 0.050d 1.009 ± 0.020d 0.511 ± 0.257ab 0.806 ± 0.027c 0.676 ± 0.027bc 0.580 ± 0.001ab 0.475 ± 0.029a 0.677 ± 0.030bc A
Formic acid – 0.028 ± 0.009b – 0.024 ± 0.005ab 0.022 ± 0.009ab 0.054 ± 0.035cd 0.038 ± 0.010bc 0.039 ± 0.013bc 0.065 ± 0.001d B
Fumaric acid – 0.008 ± 0.001a 0.037 ± 0.030b 0.034 ± 0.001b 0.034 ± 0.003b 0.039 ± 0.001b 0.053 ± 0.006bc 0.067 ± 0.004c 0.058 ± 0.003c A
Glucaric acid 0.075 ± 0.010a 0.125 ± 0.017cd 0.156 ± 0.002e 0.133 ± 0.013d 0.122 ± 0.007bcd 0.108 ± 0.007bc 0.106 ± 0.003b 0.120 ± 0.008bcd 0.130 ± 0.007d B
Glycolic acid 0.028 ± 0.002b 0.025 ± 0.001a 0.028 ± 0.001b 0.025 ± 0.001a 0.029 ± 0.001bc 0.028 ± 0.001b 0.028 ± 0.001b 0.029 ± 0.001bc 0.030 ± 0.001c A
2-Ketoglutaric acid 0.032 ± 0.001a 0.041 ± 0.001ab 0.037 ± 0.002a 0.041 ± 0.004ab 0.038 ± 0.006ab 0.048 ± 0.014b 0.061 ± 0.001c 0.035 ± 0.004a 0.037 ± 0.002a B
Lactic acid 0.806 ± 0.0630a 1.097 ± 0.014a 1.466 ± 0.037b 3.198 ± 0.209d 2.310 ± 0.074c 3.074 ± 0.100d 3.362 ± 0.251d 4.143 ± 0.450e 3.920 ± 0.256e A
Maleic acid 0.011 ± 0.001ab 0.021 ± 0.002c 0.021 ± 0.002c 0.018 ± 0.001bc 0.020 ± 0.001bc 0.007 ± 0.011a 0.017 ± 0.001bc 0.016 ± 0.002bc 0.010 ± 0.009ab A
Malic acid 0.096 ± 0.010d 0.113 ± 0.004e 0.101 ± 0.001d 0.046 ± 0.001b 0.078 ± 0.002c 0.049 ± 0.011b 0.028 ± 0.002a 0.021 ± 0.004a 0.023 ± 0.001a A
Malonic acid 0.022 ± 0.002a 0.023 ± 0.001a 0.024 ± 0.002ab 0.027 ± 0.001b 0.027 ± 0.001b 0.028 ± 0.002b 0.033 ± 0.004c 0.036 ± 0.002c 0.035 ± 0.001c A
Manelic acid 0.042 ± 0.007a 0.081 ± 0.049a 0.060 ± 0.004a 0.231 ± 0.013b 0.298 ± 0.023c 0.360 ± 0.041d 0.296 ± 0.007c 0.185 ± 0.025b 0.227 ± 0.028bc B
Oxalic acid 0.015 ± 0.002bc 0.022 ± 0.004c 0.012 ± 0.010abc 0.006 ± 0.010ab 0.010 ± 0.008abc 0.005 ± 0.008ab – – 0.005 ± 0.008ab A
Pipecolic acid 0.011 ± 0.002abc – 0.007 ± 0.012ab 0.025 ± 0.007bc 0.024 ± 0.005bc 0.017 ± 0.030abc 0.034 ± 0.005cd 0.027 ± 0.006bcd 0.049 ± 0.015d B
Propionic acid 0.015 ± 0.005a 0.047 ± 0.032a 0.036 ± 0.007a 0.073 ± 0.020ab 0.085 ± 0.027abc 0.217 ± 0.136d 0.187 ± 0.066cd 0.171 ± 0.072bcd 0.266 ± 0.037d A
Pyroglutamic acid 0.257 ± 0.025a 0.451 ± 0.008b 0.552 ± 0.021b 0.768 ± 0.022c 0.713 ± 0.041c 0.678 ± 0.133c 0.804 ± 0.053c 0.935 ± 0.073d 1.024 ± 0.118d A
Succinic acid 0.026 ± 0.002a 0.043 ± 0.003bc 0.046 ± 0.002cd 0.047 ± 0.003cd 0.047 ± 0.004bc 0.044 ± 0.001bc 0.045 ± 0.003b 0.051 ± 0.003d 0.046 ± 0.001bc A
Vanillic acid 0.018 ± 0.002a 0.025 ± 0.002b 0.017 ± 0.001a 0.031 ± 0.002b 0.029 ± 0.002b 0.032 ± 0.005bc 0.037 ± 0.006cd 0.042 ± 0.006de 0.048 ± 0.004e A
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1926–1935
∗ Each value is the mean of the relative peak areas (n = 3) to that of an internal standard with the standard deviation. Values with different letters are significantly different at α = 0.05.
† Metabolites were identified on the basis of the following criteria: A, mass spectrum and retention time were consistent with those of an authentic standard; B, mass spectrum was consistent with that of
Wiley 275 Imass spectral or by manual interpretation (tentative identification).
‡ Below the limit of detection.

Metabolic profiling of doenjang www.soci.org
Table 3. Changes in the contents of sugars and sugar derivatives in doenjang during fermentation
Relative peak area (mean ± SD) ∗
Compound 0 day 20 days 40 days 60 days 80 days 100 days 120 days 140 days 160 days ID †
Erythrose 0.468 ± 0.052a† 0.628 ± 0.062ab 0.665 ± 0.027ab 2.307 ± 2.804b 0.713 ± 0.011ab 0.758 ± 0.020ab 0.593 ± 0.013ab 0.721 ± 0.026ab 0.736 ± 0.026ab B
Xylitol 0.373 ± 0.035a 0.470 ± 0.003ab 0.445 ± 0.007ab 0.451 ± 0.270ab 0.568 ± 0.057abc 0.720 ± 0.043bcd 0.484 ± 0.349ab 0.771 ± 0.084cd 0.876 ± 0.050d A
Ribitol 0.010 ± 0.001a 0.022 ± 0.001ab 0.031 ± 0.001ab 0.079 ± 0.033cd 0.039 ± 0.0339ab 0.106 ± 0.007d 0.053 ± 0.033bc 0.100 ± 0.011d 0.090 ± 0.006d A
Ribonic acid 0.060 ± 0.004a 0.156 ± 0.014ab 0.176 ± 0.016ab 0.746 ± 0.982b 0.192 ± 0.015ab 0.230 ± 0.028ab 0.167 ± 0.025ab 0.194 ± 0.007ab 0.190 ± 0.018ab B
Inositol 1.062 ± 0.296a 2.030 ± 0.033bc 2.188 ± 0.021bcd 0.894 ± 0.100a 2.657 ± 0.358cd 2.343 ± 1.176bcd 1.428 ± 0.825ab 3.005 ± 0.345cd 3.045 ± 0.197d A
Glucosamine 0.069 ± 0.004cd 0.089 ± 0.037d 0.080 ± 0.003d 0.036 ± 0.021ab 0.046 ± 0.005bc 0.034 ± 0.004ab 0.018 ± 0.003a 0.020 ± 0.003ab 0.020 ± 0.004ab B
Mannitol 1.064 ± 0.108a 1.546 ± 0.024a 1.530 ± 0.019a 1.072 ± 0.930a 1.581 ± 0.213a 1.802 ± 0.097a 1.060 ± 0.867a 1.752 ± 0.204a 1.893 ± 0.140a A
Glucitol 0.122 ± 0.018abc 0.123 ± 0.003abc 0.109 ± 0.004ab 0.083 ± 0.072a 0.169 ± 0.025cd 0.160 ± 0.007bcd 0.159 ± 0.030bcd 0.208 ± 0.012d 0.436 ± 0.028e A
Galactonic acid 0.113 ± 0.004bc 0.193 ± 0.006d 0.197 ± 0.003d 0.195 ± 0.123d 0.169 ± 0.024cd 0.104 ± 0.016bc 0.079 ± 0.003b 0.074 ± 0.017b – ‡ B
Fructose 0.009 ± 0.002b 0.022 ± 0.007d 0.025 ± 0.001d – 0.016 ± 0.004c – – – – A
Galactose 0.039 ± 0.005a 0.068 ± 0.010ab 0.070 ± 0.003ab 0.104 ± 0.068b 0.071 ± 0.012ab 0.057 ± 0.025ab 0.055 ± 0.002a 0.042 ± 0.015a 0.042 ± 0.012a B
Glucose 0.185 ± 0.008d 0.270 ± 0.051e 0.113 ± 0.006c 0.055 ± 0.018ab 0.078 ± 0.011bc 0.077 ± 0.019bc 0.039 ± 0.011a 0.044 ± 0.007ab 0.027 ± 0.006a A
J Sci Food Agric 2010; 90: 1926–1935 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
∗ Each value is the mean of the relative peak areas (n = 3) to that of an internal standard with the standard deviation. Values with different letters are significantly different at α = 0.05.
† Metabolites were identified on the basis of the following criteria: A, mass spectrum and retention time were consistent with those of an authentic standard; B, mass spectrum was consistent with that of
Wiley 275 Imass spectral or by manual interpretation (tentative identification).
‡ Below the limit of detection.
1931

1932
Table 4. Changes in the contents of fatty acids in doenjang during fermentation
Relative peak area (mean ± SD) ∗
Compound 0 day 20 days 40 days 60 days 80 days 100 days 120 days 140 days 160 days ID †
www.soci.org H-J Namgung et al.
C10 : 0 – ‡ 0.417 ± 0.124c† 0.140 ± 0.008b – – – – – – A
C12 : 0 – 0.353 ± 0.147c 0.066 ± 0.005b – – – – – – A
C14 : 0 0.839 ± 0.105b 0.587 ± 0.067a 0.620 ± 0.091a 0.600 ± 0.158a 0.576 ± 0.039a 0.886 ± 0.012b 1.753 ± 0.362d 1.406 ± 0.155c 1.420 ± 0.288c A
C15 : 0 – 0.145 ± 0.051c 0.134 ± 0.020bc 0.117 ± 0.012b 0.123 ± 0.033bc 0.213 ± 0.011d 0.536 ± 0.167g 0.335 ± 0.056e 0.414 ± 0.232f A
C16 : 0 18.386 ± 3.745ab 62.596 ± 12.530c 90.986 ± 18.028e 89.365 ± 6.913e 15.083 ± 2.642a 23.647 ± 5.320b 112.573 ± 19.653f 84.262 ± 15.490de 81.845 ± 5.040d A
C16 : 1 0.757 ± 0.157a 0.879 ± 0.051a 1.022 ± 0.126ab 1.444 ± 0.271c 0.891 ± 0.048a 1.364 ± 0.016bc 3.247 ± 0.992e 2.977 ± 0.867de 2.688 ± 1.360d A
C17 : 0 0.162 ± 0.023a 0.469 ± 0.154b 0.414 ± 0.088b 0.154 ± 0.043a 0.346 ± 0.185ab 0.855 ± 0.038c 2.194 ± 1.255e 1.031 ± 0.434cd 1.184 ± 1.149d A
C18 : 0 5.198 ± 1.127a 68.421 ± 16.709d 14.674 ± 1.147b 25.073 ± 10.202c 6.213 ± 2.676a 24.996 ± 4.808c 24.624 ± 6.255c 27.993 ± 15.899e 25.218 ± 18.461c A
C18 : 1 77.164 ± 15.143b 102.781 ± 16.254c 160.985 ± 1.116d 32.147 ± 6.754a 114.099 ± 11.791c 149.206 ± 14.362d 217.597 ± 16.246f 188.409 ± 12.415e 189.553 ± 19.037e A
C18 : 2 107.104 ± 16.203b 105.653 ± 9.936b 100.674 ± 29.289b 50.205 ± 2.795a 35.816 ± 10.946a 212.220 ± 10.038c 349.758 ± 17.651f 257.254 ± 8.719d 306.127 ± 14.751e A
C18 : 3 21.769 ± 3.188a 79.002 ± 11.248d 74.376 ± 9.259d 49.647 ± 5.043bc 21.637 ± 8.199a 40.518 ± 3.678b 52.567 ± 13.683c 53.805 ± 10.844c 49.343 ± 17.736bc A
C20 : 0 1.053 ± 0.970a 2.281 ± 0.325bcd 1.143 ± 0.114ab 1.755 ± 0.402abc 1.436 ± 0.112abc 2.447 ± 0.189cd 6.580 ± 5.711f 3.437 ± 0.446de 4.525 ± 1.918e A
C20 : 1 0.712 ± 0.064a 1.552 ± 0.424c 1.074 ± 0.084ab 1.556 ± 0.279c 1.203 ± 0.129bc 2.029 ± 0.201d 2.909 ± 0.611e 2.781 ± 0.454e 3.579 ± 1.604f A
C20:2 – – – 0.302 ± 0.058b 0.282 ± 0.032b 0.447 ± 0.122c – – – A
C22 : 0 3.893 ± 5.893d 1.634 ± 0.368b 1.202 ± 0.104b 2.231 ± 0.235c 1.537 ± 0.110b 2.649 ± 0.261c – 3.694 ± 0.569d 11.271 ± 0.118e A
C23 : 0 – 0.321 ± 0.014cd 0.340 ± 0.018cd 0.211 ± 0.054b 0.284 ± 0.258bcd 0.271 ± 0.067bc 1.523 ± 0.347e 0.388 ± 0.061d 0.238 ± 0.412bc A
C10 : 0, capric acid; C12 : 0, lauric acid; C14 : 0, myristic acid; C15 : 0, pentadecylic acid; C16 : 0, palmitic acid; C16 : 1, palmitoleic acid; C17 : 0, margaric acid; C18 : 0, stearic acid; C18 : 1, oleic acid; C18 : 2,
linoleic acid; C18 : 3, linolenic acid; C20 : 0, arachidic acid; C20 : 1, eicosanic acid; C20 : 2, eicosadienoic acid; C22:0,behenicacid;C23:0,tricosanoic acid.
∗ Each value is the mean of the relative peak areas (n = 3) to that of an internal standard with the standard deviation. Values with different letters are significantly different at α = 0.05.
† Metabolites were identified on the basis of the following criteria: A, mass spectrum and retention time were consistent with those of an authentic standard).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1926–1935
‡ Below the limit of detection.

Metabolic profiling of doenjang www.soci.org
Mandelic acid, propionic acid and glucaric acid were detected
in negligible amounts in the doenjang at the beginning of
fermentation, but rapidly increased after 60, 100 and 120 days
fermentation, respectively. A previous study reported a similar
finding where an insignificant amount of glutaric acid was initially
found but then increased up to 60 days of fermentation. 20
Sugars and sugar derivatives
Sugars such as glucose, fructose and galactose were identified
during doenjang fermentation (Table 3). Glucose was identified as
a primary sugar in doenjang at the initial stage of fermentation and
its amount was reduced after 20 days of fermentation. Fructose,
the sweetest sugar, was found in an insignificant amount, and was
not detected after 100 days of fermentation. This finding is quite
different from an earlier study that reported measuring 1937 mg of
fructose in 100 g of commercial soybean paste. 5 Previous studies
suggested that the composition of free sugars in doenjang can
differ according to soybean cultivar or the ingredient mixture for
meju manufacture. 23 Yoo found a difference in the composition
of major sugars between a traditionally manufactured doenjang
and a commercial doenjang. 21 Furthermore, glucose and total
sugars were higher in doenjang that was inoculated with Bacillus
natto or Aspergillus oryzae compared to a doenjang manufactured
by traditional methods. 16 Overall, in this study, levels of total
sugars decreased as fermentation proceeded. A major reason for
reductions in total sugars during fermentation is that a portion
of the sugars is utilised as substrate for microbial growth and
organic acid fermentation. 24 Interestingly, in the current study,
glucosamine is an amino sugar that was formed in the doenjang.
It is plausible that glucosamine was synthesised through the
donation of an amine from glutamine to glucose contained in
the soybeans. 25 Since glucosamine is a normal constituent of
glycosaminoglycans in joint cartilage, it is commonly used for the
treatment of osteoarthritis. 26 Erythrose, a tetrose carbohydrate,
was also detected in a significant amount in the doenjang during
fermentation (Table 3).
To date, limited information has been reported on the
identification of sugar derivatives in fermented soybean paste. The
current study identified eight sugar derivatives, and significant
amounts of xylitol, inositol and mannitol were detected in the
doenjang during fermentation (Table 3). Xylitol, a low-calorie
alternative to table sugar, showed an increasing pattern during
fermentation. After 140 days of fermentation, the amount of xylitol
detected in the doenjang was twice that at the beginning of
fermentation. Our results also showed a remarkable increase in
inositol during fermentation. Inositol is a chemical compound with
a formula containing a six-fold alcohol of cyclohexane hexol, and is
involved in a number of biological processes as the basis for several
signalling and secondary messenger molecules. 27 Inthecaseof
mannitol, a significant amount was formed during fermentation
(Table 3). Mannitol is assumed to have several beneficial effects
as an antioxidant as well as a non-metabolisable sweetener. 28
Insignificant amounts of ribitol and ribonic acid were also found.
Finally, results from current study showed that the amount of
glucitol, known as sorbitol, steadily increased after 80 days of
fermentation and the highest level of glucitol was found at
160 days of fermentation. This result implies that glucitol may
be derived from either glucose or fructose
Fatty acids
A total of 16 fatty acids, including saturated and unsaturated
fatty acids, were identified using GC-MS. Overall, relatively higher
contents of unsaturated fatty acids than saturated fatty acids
were observed during fermentation (Table 4). Most unsaturated
fatty acids showed increasing patterns between 120 and 140 days
of fermentation, and then gradually decreased. A previous study
showed similar results in which unsaturated fatty acids consisted of
81.97% of the total fatty acids in doenjang prepared by traditional
methods. 8 At the initial stage of fermentation (20–40 days), oleic
acid (C18 : 1), linoleic acid (C18 : 2), palmitic acid (C16 : 2) and
linolenic acid (C18 : 3) were found as major fatty acids in the
doenjang (Table 4). The relative amounts of both oleic acid and
linoleic acid increased as fermentation progressed. In particular,
a greater amount of linoleic acid was detected after 160 days
of fermentation, approximately three times that occurring at
the beginning of fermentation. Similar to our results, previous
studies have reported that fatty acids amounts were found in the
following order: linoleic acid>oleic acid>linolenic acid > palmitic
acid. 8,29 Park et al. 9 also reported thatmystric acid (C14 : 0), palmitic
acid (C16 : 0), stearic acid (C18 : 0), oleic acid (C18 : 1), linoleic acid
(C18 : 2) and linolenic acid (C18 : 3) were identified, and linoleic acid
was the most abundant fatty acid, consisting of 38.56–51.86% of
the total fatty acids. Both pentadecylic acid (C15 : 0) and margaric
acid (C17 : 0) were identified, but insignificant amounts were
observed in the doenjang during fermentation.
Principal components analysis
Figure 1 provides the score plot that illustrates the outcomes
of doenjang metabolite analysis according to PCA. As shown in
Fig. 1, significant changes in metabolites are clearly distinguished
according to fermentation time in the score plot generated by
combining PC1 (52% of the total variance) with PC2 (19% of
the total variance). Overall, the plots are aggregated for each
fermentation time, and similarities in doenjang metabolites during
fermentation can be inferred from the relative variables of these
plots. Significant differences in metabolites during the initial
and interim stage of fermentation (positive PC1 dimension) can
be separated from those of the latter stage of fermentation
(negative PC1 dimension) mainly in the score of PC1. However,
there were no significant changes in metabolites between 60 and
80 days or 100 and 120 days of fermentation. Figure 2 illustrate
the loading plots of metabolites contributing to changes in
the plots for each fermentation stage in principal components
analysis. In the analysis of the primary principal components,
malic acid (no. 50), glucosamine (no. 64), oxalic acid (no. 53),
fructose (no. 68), galactronic acid (no. 67), citric acid (no. 42),
lysine (no. 36), aminoaldipic acid (no. 33), lactic acid (no. 48),
fumaric acid (no. 44), pyrogultamic acid (no. 56), carbonic acid
(no. 41), alanine (no. 17), methionine (no. 30), leucine (no. 23),
isoleucine (no. 22), phenylalanine (no. 32), tryptophan (no. 39),
malonic acid (no. 51), propionic acid (no. 55) and eicosanic
acid (no. 13) were found to be the metabolites contributing to
changes (Fig. 2). Among them, malic acid, glucosamine, oxalic
acid, fructose, galactonic acid and citric acid, with positive
absolute values, were found in significant amounts in the
early stage of fermentation (0–40 days). Furthermore, fructose,
D-glucosamine, malic acid, oxalic acid and citric acid were found
to be the metabolites mainly contributing to the differentiation
of the doenjang in the initial stage of fermentation. Meanwhile,
lysine, aminoaldipic acid, lactic acid, fumaric acid, pyrogultamic
acid, carbonic acid, alanine, methionine, leucine, isoleucine,
phenylalanine, tryptophan, malonic acid, propionic acid and
eicosanic acid were significantly responsible for changes during
the latter stage of doenjang fermentation. The metabolites that
J Sci Food Agric 2010; 90: 1926–1935 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1933

1934
www.soci.org H-J Namgung et al.
Figure 1. PCA score plot of metabolites in the doenjang according to fermentation period by combining PC1 and PC2.
Figure 2. PCA loading plot showing the variables contributing to the PC1 score plots (top) and PC2 score plots (bottom).
determined the characteristics of the doenjang after 60 days
of fermentation can be explained by the secondary principal
component (Fig. 2). Myristic acid (no. 3), linoleic acid (no. 10),
eicosadienoic acid (no. 14), behenic acid (no. 15), sacrosine (no.
18), glycine (no. 19), aminobutyric acid (no. 20), glucaric acid (no.
45), glycolic acid (no. 46), erythrose (no. 59), ribonic acid (no.
62) and glucitol (no. 66) showed decreasing tendencies while
behenic acid, SAR, glycolic acid and glucitol increased during the
fermentation period.
Overall, it was possible to differentiate the doenjang samples
according to fermentation time by assessing non-volatile metabolites
such as amino acids, organic acids, fatty acids, sugars and
sugar derivatives. The sugar derivatives were considered as main
contributors to discriminate the doenjang samples in the early
stage of fermentation, while the amino acids were indicators for
the latter stages of fermentation. Organic acids were also involved
in determining changes in the doenjang by presenting increasing
patterns as fermentation time progressed. The major metabolites
contributing to the differentiation of the doenjang during fermentation
were leucine (no. 22), isoleucine (no. 23), aminoaldipic acid
(no. 33), lysine (no. 36), malic acid (no. 50), oxalic acid (no. 53) and
glucosamine (no. 64).
CONCLUSION
The GC-MS technique followed by PCA, used for non-volatile
metabolite profiling of doenjang, successfully demonstrated
changes in composition patterns as well as difference in
metabolites according to fermentation period. Malic acid, oxalic
acid, citric acid, glucosamine and fructose were found to be the
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1926–1935

Metabolic profiling of doenjang www.soci.org
metabolites that differentiated the doenjang samples at the initial
stage of fermentation, while pipecolic acid, malonic acid, formic
acid, propionic acid, mandelic acid and xylitol were the metabolites
that differentiated the samples after 100 days of fermentation. A
set of metabolites was able to be determined representing the
quality of doenjang during fermentation, and which might also be
correlated to taste ingredients, flavour, nutrition and physiology
activities that are claimed to be dependent on the quality control
of commercial doenjang.
ACKNOWLEDGEMENT
This study was supported by research grants from the Korea
Science and Engineering Foundation (KOSEF) for Biofoods Research,
Ministry of Science and Technology, South Korea (no.
M1051012000306N101200310).
REFERENCES
1 Jung KO, Park SY and Park KY, Longer aging time increases the
anticancer and antimetastatic properties of doenjang. Nutrition
22:539–545 (2006).
2 Kim SH, Kim SJ, Kim BH, Kang SG and Jung ST, Fermentation of
doenjang prepared with sea salts. Korean J Food Sci Technol
32:1365–1370 (2000).
3 Lim SY, Park KY and Rhee SH, Anticancer Effect of doenjang in in vitro
dulforhodamineb(SRB)assay.JKoreanSocFoodSciNutr 28:240–245
(1999).
4 Kim TW, Lee JH, Kim SE, Park MH, Chang HC and Kim HY, Analysis of
microbial communities in doenjang, a Korean fermented soybean
paste, using nested PCR-denaturing gradient gel electrophoresis.
Int J Food Microbiol 131:265–271 (2009).
5 Park HK, Gil B and Kim JK, Characteristics of taste components of
commercial soybean paste. Food Sci Biotechnol 11:376–379 (2002).
6 Yoo SK, Kang SM and Noh YS, Quality properties on soy bean pastes
made with microorganisms isolated from traditional soy bean
pastes. Korean J Food Sci Technol 32:1266–1270 (2000).
7 Kim JK, Biologically active components of soy-fermented foods. Korea
Soybean Digest 19:1–7 (2002).
8 Park SK, Seo KI, Choi SH, Moon JS and Lee YH, Quality assessment of
commercial doenjang prepared by traditional method. J Korean Soc
Food Sci Nutr 29:211–217 (2000).
9 Park JS, Lee MY and Lee TS, Compositions of sugars and fatty acids
in soybean paste (doenjang) prepared with different microbial
sources. J Korean Soc Food Sci Nutr 24:917–924 (1995).
10 Gullberg J, Jonsson P, Nordstrom A, Sjostrom M and Moritz T,
Design of experiments: an efficient strategy to identify factors
influencing extraction and derivatization of Arabidopsis thaliana
samples in metabolomic studies with gas chromatography/mass
spectrometry. Anal Biochem 331:283–295 (2004).
11 Fiehn O, Metabolomics – the link between genotypes and
phenotypes. Plant Mol Biol 48:155–171 (2002).
12 Kopka J, Current challenges and developments in GC-MS based
metabolite profiling technology. J Biotechnol 124:312–322 (2006).
13 Jansen JJ,Hoefsloot HCJ,Boelens HFM,Greef JandSmilde AK,Analysis
of longitudinal metabolomics data. Bioinformatics 20:2438–2446
(2004).
14 Metcalfe LD, Schmitz AA and Pelka JR, The rapid preparation of
fatty acid esters for gas chromatographic analysis. Anal Chem
38:514–515 (1966).
15 Jung BM and Roh SB, Physicochemical quality comparison of
commercial doenjang and traditional green tea doenjang. JKorean
Soc Food Sci Nutr 33:132–139 (2004).
16 An HS, Bae JS and Lee TS, Comparison of free amino acids, sugars, and
organic acids in soybean paste prepared with various organisms.
J Korean Agric Chem Soc 30:345–350 (1987).
17 Shizuka F, Kido Y, Nakazawa T, Kitajima H, Aizawa C, Kayamura H, et al,
Antihypertensive effect of gamma-amino butyric acid enriched soy
productsinspontaneouslyhypertensiverats.Biofactors22:165–167
(2004).
18 Navayan VS and Nair PM, Metabolism zymology and possible roles of
4-aminobutyrate higher plants. Phytochemistry 29:367–375 (1990).
19 Sustková-FiserováM, Vávrová J and Krsiak M, Brain levels of GABA,
glutamate and aspartate in sociable, aggressive and timid mice: an
in vivo microdialysis study. Neuro Endocrinol Lett 30:79–84 (2009).
20 Kim MJ and Rhee HS, Studies on the changes of taste compounds
during soy paste fermentation (II). KoreanJFoodCookSci 9:257–260
(1993).
21 Yoo SM, Quality characteristics of doenjang prepared using selective
starters. PhD thesis. Department of Food Science and Technology,
Graduate School, Chonbuk National University, Chonbuk, Korea
(2003).
22 Sulisyo J, Natioshi T, Kazaumi F and Kan K, Production of Natto starter.
J Food Technol 35:280–281 (1988).
23 Hwang ES, Quality of 4 precommended soybean cultivars for meju and
doenjang.PhDthesis.Department of Food Science and Technology,
Graduate School, Chungnam National University, Taejeon, Korea
(2002).
24 Joo HK, Kim DH and Oh KT, Chemical composition changes in
fermented doenjang depend on doenjang koji and its mixture.
J Korean Soc Agric Chem Biotechnol 35:351–360 (1992).
25 Sayeski PP and Kudlow JE, Glucose metabolism to glucosamine
is necessary for glucose stimulation of transforming growth
factor–gene transcription. JBiolChem271:15237–15243 (1996).
26 Reginster JY, Deroisy R, Rovati LC, Lee RL, Lejeune E, Bruyere O, et al,
Long-term effects of glucosamine sulphate on osteoarthritis
progression: a randomized, placebo-controlled clinical trial. Lancet
357:251–256 (2001).
27 Clements RS and Darnell B, Myo-inositol content of common foods:
development of a high-myo-inositol diet. Am J Clin Nutr
33:1954–1967 (1980).
28 Wisselink HW, Weusthuisa RA, Eggink G, Hugenholtz J and
Grobben GJ, Mannitol production by lactic acid bacteria: a review.
Int Dairy J 12:151–161 (2002).
29 Choi KS and Rhee HS, Characteristics of doenjang made from different
material and ratio of koji. Korean J Food Cook Sci 10:39–44 (1994).
J Sci Food Agric 2010; 90: 1926–1935 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1935

1936
Research Article
Received: 8 October 2009 Revised: 23 April 2010 Accepted: 5 May 2010 Published online in Wiley Interscience: 16 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4038
Timing of the inhibitory effect of fruit on return
bloom of ‘Valencia’ sweet orange (Citrus
sinensis (L.) Osbeck)
Amparo Martínez-Fuentes, Carlos Mesejo, Carmina Reig
and Manuel Agustí ∗
Abstract
BACKGROUND: In Citrus the inhibitory effect of fruit on flower formation is the main cause of alternate bearing. Although there
are some studies reporting the effect on flowering of the time of fruit removal in a well-defined stage of fruit development, few
have investigated the effect throughout the entire fruit growth stage from early fruitlet growth to fruit maturity. The objective
of this study was to determine the phenological fruit developmental stage at which the fruit begins its inhibitory effect on
flowering in sweet orange by manual removal of fruits, and the role of carbohydrates and nitrogen in the process.
RESULTS: Fruit exerted its inhibitory effect from the time it was close to reaching its maximum weight, namely 90% of its final
size (November) in the present experiments, to bud sprouting (April). The reduction in flowering paralleled the reduction in bud
sprouting. This reduction was due to a decrease in the number of generative sprouted buds, whereas mixed-typed shoots were
largely independent of the time of fruit removal, and vegetative shoots increased in frequency. The number of leaves and/or
flowers per sprouted shoot was not significantly modified by fruit load.
CONCLUSION: In ‘Valencia’ sweet orange, fruit inhibits flowering from the time it completes its growth. Neither soluble sugar
content nor starch accumulation in leaves due to fruit removal was related to flowering intensity, but some kind of imbalance
in nitrogen metabolism was observed in trees tending to flower scarcely.
c○ 2010 Society of Chemical Industry
Keywords: carbohydrates; Citrus; flowering; nitrogen fractions; sprouting
INTRODUCTION
The inhibitory effect of fruit on flower formation is the main
cause of alternate bearing in ‘Valencia’ sweet orange, 1 an effect
depending on the number of developed fruits and on the date of
harvest.
It has been suggested that fruit either reduces the sensitivity
of buds to floral inductive conditions or else increases bud
requirements to be induced to flower. 2 In either event, since
the existence of endogenous flowering promoters has not yet
been proven, it has been argued that, from the completion of
juvenility, citrus buds are continually induced to flower, but their
development is under the negative control of an inhibitor, 3,4
which maintains the meristems in a vegetative condition. 5 The
presence of endogenous root tissue-, seed- and fruit tissueproduced
gibberellins regulates (negatively) flowering in Citrus,as
proposed by Goldschmidt and Monselise. 3 The effect of gibberellic
acid inhibiting flowering from late summer until the winter rest
period 6–8 reinforces this hypothesis.
Although there are some studies reporting the effect on
flowering of the time of fruit removal by hand, 9,10 they are always
restricted to a well-defined stage of fruit development, e.g. early
fruit development stage (fruitlet drop) or maturation stage, and
few have investigated the effect throughout the entire fruit growth
stage, 11 i.e. from early fruitlet growth to fruit maturity.
In this study we examine the effect of defruiting on both the
carbohydrate concentration and the total and soluble nitrogen
fraction (NH4 + -N, NO3 − -N and NO2 − -N) contents in leaves
and their relationships with flowering intensity in ‘Valencia’
sweet orange. It has been reported that control of flowering
is mediated by carbohydrate content 12 and that the accumulation
of leaf ammonia content caused by water-deficit stress or lowtemperature
stress is an early stress-linked event influencing floral
initiation in Citrus. 13 Further, a disturbance in the nitrate reduction
mechanism was found in trees that tended to flower less, both in
citrus 9 and in peach. 14
The main aim of the present study was to determine the
phenological fruit developmental stage at which the fruit begins
its inhibitory effect on flowering in sweet orange by manual
removal of fruits from fruitlet drop to late in the maturation stage
up to harvest.
∗ Correspondence to: Manuel Agustí, Instituto Agroforestal Mediterráneo,
Universidad Politécnica de Valencia, Camino de Vera s/n, E-46022 Valencia,
Spain. E-mail: magusti@prv.upv.es
Instituto Agroforestal Mediterráneo, Universidad Politécnica de Valencia,
Camino de Vera s/n, E-46022 Valencia, Spain
J Sci Food Agric 2010; 90: 1936–1943 www.soci.org c○ 2010 Society of Chemical Industry

Flowering and time of fruit remaining on tree in Citrus www.soci.org
MATERIAL AND METHODS
Plant material and treatments
The study was conducted in a commercial orchard located in Lliria,
Spain containing mature 20-year-old ‘Valencia’ sweet orange
(Citrus sinensis (L.) Osbeck) trees grafted onto ‘Carrizo’ citrange
(C. sinensis (L.) Osbeck × Poncirus trifoliate (L.) Raf.) rootstock,
grown in a loamy-clay soil, planted 5 m × 6 m apart, with drip
irrigation.
In the late spring, trees similar in size, vigour and crop load
were selected. Four branches per tree with similar size (4–6 cm
base diameter) and similar number of developed fruits (50 ± 4
fruits), located all around the tree at 1.5–2 m above soil level
and totalling 2000 nodes, were labelled for flowering evaluation
in the following spring. Branches were defruited by hand at
phenological fruit growth stages 72, 74, 78, 79, 80, 81 and 89
of the BBCH scale, 15 i.e. from the onset of physiological fruitlet
abscission(BBCHstage72)tofruitmatureforconsumption/harvest
date (BBCH stage 89). The original BBCH scale was developed
by the Biologische Bundesanstalt für Land- und Forstwirtschaft
by considering numerical scales suitable for describing the
development of different species. The Citrus BBCH scale consists
of two-digit stages, e.g. 74, the first digit expressing the major
stage, i.e. 7 refers to the development of fruit, and the second digit
expressing the secondary stage within the course of the specific
major stage, i.e. 4 refers to fruits about 40% of final size. 15
Fruits from selected branches were counted and their diameters
measured at the time of fruit removal.
Sprouting and flowering evaluation
Bud sprouting and flowering were measured in the spring.
The shoots initiated were counted, and also their numbers of
flowers and leaves, and classified according to Guardiola et al. 16
Unsprouted nodes were also counted. From the number of flowers
per shoot and the number of shoots developed per branch, the
total number of flowers per branch could be calculated. The
results are expressed in flowers per 100 nodes to compensate for
the differences in size of the branches selected for counting. Only
buds younger than 24 months of age were considered for the
countings, since older buds hardly contribute to the spring flush.
Carbohydrate analysis
Approximately 20 leaves from spring terminal flushes of defruited
branches at BBCH stage 72 and fruited branches (defruited at
BBCH stage 89) were collected at BBCH stages 78 (during cell
enlargement stage), 81 (colour break) and 89 (mature fruit).
Samples were frozen immediately in liquid nitrogen, lyophilised
and stored as powders at −28 ◦ C. The procedure for carbohydrate
determination was as described previously by Mehouachi et al. 17
In brief, 100 mg powdered samples were extracted with 1 mL of
800 mL L −1 ethanol and purified sequentially using cation and
anion exchange columns. The eluates were then passed through
a C18 Sep-Pak cartridge (Waters-Millipore, Barcelona, Spain) and
analysed in a Spectra HPLC System ® (Spectra, San Jose, CA, USA)
equipped with a vacuum pump (Spectra P2000) and a differential
refractometer (Spectra R150). Sucrose, glucose and fructose were
identified according to their retention times.
Starch levels were determined in the pellets that remained
after the extraction of soluble sugars. The residue was incubated
by shaking for 2 h at 55 ◦ C with 0.2 mL of 60 mg mL −1 fucose
(internal standard), 0.5 mL of sodium acetate (pH 4.5) and 1 mL
of 1218 U amyloglucosidase from Rhizopus (Sigma Chemical
Co. Inc. Sigma-Aldrich Chemie Gmbh, Steinheim, Germany). The
glucose released was determined by high-performance liquid
chromatography (HPLC) as above. Results were expressed as g
glucose released g −1 dry weight (DW).
Nitrogen fraction analysis
The same leaves sampled for carbohydrate analysis were used
for total and soluble nitrogen fraction analysis. Proteinaceous
nitrogen (N-Prot), ammonium nitrogen (NH4 + -N) and nitrate
nitrogen (NO3 − -N) were determined according to Raigón et al. 18
and Beljaars et al. 19 In brief, 500 mg powdered samples were
homogenisedin10 mLof50 g L −1 trichloroaceticacid(TCA)at4 ◦ C,
rinsed with 30 mL of cold 50 g L −1 TCA, filtered, rinsed three times
with 10 mL of cold 50 g L −1 TCA, made up to 100 mL with Milli-Q
water and injected into a FIAstar 5000 Analyser ® (Foss Tecator,
Höganäs, Sweden) for NH4 + -N determination. Results were
expressed as µgNH4 + -N g −1 DW. The residue, containing N-Prot,
was then digested by the micro-Kjeldahl method and distilled in a
Kjeltec 2200 Autodistillation ® apparatus (Foss, Höganäs, Sweden).
Results were expressed as mg N-Prot g −1 DW. For the NO3 − -N
fraction analysis, 500 mg powdered samples were homogenised
in 50 mL of Milli-Q water, filtered and injected into the FIAstar 5000
Analyser ® . Results were expressed as µgNO3 − -N g −1 DW.
Statistical design
The experiment was laid out in randomised blocks, with single-tree
plots and ten replicates per date of fruit removal. Variance and
regression analyses were performed on the data, and means were
separated using Duncan’s new multiple range test.
RESULTS
In our experiments with ‘Valencia’ sweet orange, flowering
depended on the time of fruit remaining on the tree (Fig. 1). Our
results show two periods of fruit influence on flowering defined
by the date of fruit removal: (1) from the onset of physiological
fruitlet drop (June drop) (BBCH stage 72) to full fruit development
(BBCH stage 78) and (2) from the date on which the fruit reached
its final size (BBCH stage 79) to full maturity (BBCH stage 89),
coinciding with the harvest date. For the first period the following
spring flowering averaged 41.8 flowers per 100 nodes, while
for the second period the following spring flowering averaged
only 6.9 flowers per 100 nodes (Fig. 1). Flowering of branches
that maintained all developed fruits up to final size was 90%
lower on average than that of branches defruited during the fruit
developmental stage. An additional but non-significant reduction
in flowering intensity was found from the early stages of peel
colouration (BBCH stage 81) up to full fruit ripeness (BBCH stage
89). It is important to note that this latter developmental stage
coincided with bud sprouting. The number of summer/autumn
vegetative shoots was also significantly increased (P ≤ 0.05) by
removing fruits at BBCH stages 72–78 (3.4 shoots per 100 nodes
on average) compared with fruit removal at BBCH stages 79–89
(0.49 shoots per 100 nodes on average).
The reduction in flowering paralleled the reduction in bud
sprouting (Fig. 2), and there was a significant relationship between
the percentage of sprouted buds and the number of flowers per
100 nodes (r = 0.976, P ≤ 0.01).
Buds producing generative shoots were the most sensitive to
inhibition by fruit. Comparing the two periods differing in bud
sensitivity to fruit inhibition, the frequency of this type of shoot
J Sci Food Agric 2010; 90: 1936–1943 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1937

1938
Flowers / 100 nodes
50
45
40
35
30
25
20
15
10
5
a a
a
www.soci.org A Martínez-Fuentes et al.
b b
Flowers
Fruit diameter
0
0
Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
72 74 78 79 80 81 89
BBCH-phenological fruit growth stage at fruit removal
Figure 1. Time course of fruit growth and influence of time of fruit removal on flowering in ‘Valencia’ sweet orange. Each value is the average of ten trees.
Standard errors are given as vertical bars. Different letters indicate significant differences (P ≤ 0.05). Phenological growth stages of the BBCH scale at the
time of fruit removal are given.
sprouting (%)
25
20
15
10
5
0
a
a
a
Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr
72 74 78 79 80 81 89
BBCH-phenological fruit growth stage atfruit removal
Figure 2. Influence of time of fruit removal on pattern of sprouting in ‘Valencia’ sweet orange. Each value is the average of ten trees. Standard errors
are given as vertical bars. Different letters indicate significant differences (P ≤ 0.05). Phenological growth stages of the BBCH scale at the time of fruit
removal are given.
was reduced by 80% on average by maintaining the fruit from
colour break onwards (Fig. 3). On the contrary, vegetative shoots
increased by 60% in frequency. The proportion of mixed-typed
shoots was largely independent of the time of fruit removal. In
absolute values, multiflowered leafless shoots were reduced from
7.8 to 1.2 shoots per 100 nodes on average when comparing the
two periods of fruit removal, and single-flowered leafless shoots
b
b
b
bc
b
were reduced from 3.0 to 0.4 shoots per 100 nodes; leafy shoots
were reduced only slightly by fruit load, multiflowered ones from
1.2 to 0.8 shoots per 100 nodes and single-flowered ones from 0.5
to 0.3 shoots per 100 nodes. Shoot characteristics, i.e. number of
flowers and/or leaves per shoot, were not significantly modified
by fruit removal, regardless of the shoot type and date of removal
(data not shown).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1936–1943
c
80
70
60
50
40
30
20
10
Fruit diameter (mm)

Flowering and time of fruit remaining on tree in Citrus www.soci.org
shoots /100 nodes
25
20
15
10
5
0
72
Jun
74
Jul
78
Sep
79
Nov
80
Dec
Mixed shoots
Generative leaflessshoots
Vegetative shoots
81
Jan
BBCH-phenological fruit growth stage at fruit removal
89 Apr
Significance
Mixed sh n.s.
Gen.sh a a a b b b b
Veg.sh. a a a b b ab ab
Figure 3. Influence of time of fruit removal on pattern of shoot sprouting in ‘Valencia’ sweet orange. Results are expressed as the number of shoots per
100 nodes. Each value is the average of ten trees. Different letters indicate significant differences (P ≤ 0.05). Phenological growth stages of the BBCH
scale at the time of fruit removal are given.
Significant differences in carbohydrate concentration in leaves
due to fruit load were found, the intensity of the response
depending on the date of analysis (Table 1). Thus, whereas
glucose concentration was significantly lower in leaves from
defruited branches both during fruit growth (BBCH stage 78)
and at fruit colour break (BBCH stage 81) (Table 1), fructose
concentration was significantly lower only at fruit colour break,
and sucrose concentration only during fruit growth. Leaves of
branches that maintained all their developed fruits up to maturity
had the lowest starch content at fruit ripeness (BBCH stage
89), which coincided with bud sprouting. Except for starch and
for glucose in leaves of branches defruited at BBCH stage 72,
all sugars showed higher concentrations at fruit colour break
stage, regardless of the presence of fruit. For starch a higher
concentration was found at bud sprouting, regardless of the date
of fruit removal.
NO3 − -N levels accumulated in leaves increased during fruit
development, irrespective of fruit removal, but those accumulated
in leaves of branches that maintained all fruits up to maturity
differed significantly from those accumulated in leaves of defruited
branches; the former remained almost constant for the two
last dates of analysis, whereas the latter decreased from 68 to
29 µgg −1 DW (Table 2). Thus, at fruit colour break (BBCH stage
81), leaves of branches maintaining all fruits contained 27 µg
NO3 − -N g −1 DW less than leaves of defruited branches, whereas
at the mature fruit stage (BBCH stage 89), which coincided with
bud sprouting in ‘Valencia’ sweet orange, they contained 17 µg
NO3 − -N g −1 DW more than leaves of defruited branches. NH4 + -
N content in leaves also increased during fruit development,
irrespective of fruit removal, and the content in leaves of branches
maintaining all fruits also remained constant during BBCH stages
81–89 (28–29 µgg −1 DW); no significant differences due to
fruit load were found during fruit growth (BBCH stage 78) and
at fruit colour break (BBCH stage 81), but, at bud sprouting
(coinciding with BBCH stage 89), NH4 + -N content in leaves of
branches maintaining all fruits (29 µgg −1 DW) was significantly
lower than that in leaves of defruited branches (36 µgg −1
DW), differences becoming statistically significant. Interestingly,
whereas, in defruited branches, leaf NO3 − -N content decreased
significantly and leaf NH4 + -N content increased significantly from
fruit colour break to mature fruit growth stage, no significant
changes were found in leaves of branches that maintained all
their developed fruits between these phenological fruit growth
stages (Table 2). Non-significant differences in N-Prot levels due
to fruit load were found (Table 2); nevertheless, at the mature
fruit stage (BBCH stage 89), leaves from branches maintaining all
fruits and from defruited branches were found to contain 11.6 and
10.4 mg g −1 DW respectively. These levels were lower than those
found at the colour break stage (BBCH stage 81).
In our experiments the reduction in flowering paralleled a
reduction in the number of fruits cropped per tree in the
next season (data not shown). Further, a significant relationship
(r = 0.937, P ≤ 0.01) was found between the number of flowers
per 100 nodes and the number of fruits cropped per tree. However,
this reduction in flowering intensity resulted in an increase in both
fruit set and average fruit weight, which compensated for the
reduced flowering; thus an average flowering inhibition of more
than 75% due to fruit load (Fig. 1) gave rise to only a 50% reduction
in the crop (data not shown).
DISCUSSION
The effect of fruit inhibiting flowering has been shown in most
polycarpic (perennial) fruit trees, both evergreen and deciduous
species, such as apple and pear, 20 peach 14 and citrus, 1 but none of
those studies reported the exact moment when the fruit initiates
its inhibitory effect. Recently, Verreynne and Lovatt 11 provided
evidence that removal of all fruit from ‘Pixie’ mandarin on-crop
trees early in summer increased total flower number in spring by
increasing summer/autumn shoot number.
J Sci Food Agric 2010; 90: 1936–1943 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1939

1940
Table 1. Effect of fruit load on time course of glucose, fructose, sucrose and starch concentrations in leaves of ‘Valencia’ sweet orange. Dates of fruit removal and analyses are given as phenological
growth stages of the BBCH scale. Each value is the average ± standard error of ten trees
Glucose (mg g −1 DW) Fructose (mg g −1 DW) Sucrose (mg g −1 DW) Starch (mg glucose g −1 DW)
www.soci.org A Martínez-Fuentes et al.
BBCH stage 78 81 89 78 81 89 78 81 89 81 89
72 4.7 ± 1.9a 8.5 ± 0.3b 8.1 ± 0.2b 5.2 ± 0.9a 8.4 ± 0.3b 6.3 ± 0.1a 3.8 ± 0.9a 26.1 ± 1.9c 12.1 ± 0.1b 76.3 ± 3.1a 152.5 ± 3.6b
89 7.4 ± 0.8a 10.0 ± 0.5b 8.0 ± 0.4a 5.3 ± 0.4a 10.8 ± 0.3b 6.6 ± 0.1a 13.7 ± 1.4a 26.9 ± 1.1b 11.5 ± 0.7a 72.2 ± 8.3a 138.1 ± 0.5b
∗ ∗ NS NS ∗ NS ∗ NS NS NS ∗
Fa P = 0.0001 P = 0.0001 P = 0.0001 P = 0.2267
Db P = 0.0001 P = 0.0001 P = 0.0001 P = 0.0001
F × Dc P = 0.0006 P = 0.0001 P = 0.0001 P = 0.3071
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1936–1943
∗ Significant at P ≤ 0.01; NS, not significant.
a Fruit removal effect.
b Date effect. Different letters in the same row for a given sugar indicate significant differences.
c Fruit removal × date effect.

Flowering and time of fruit remaining on tree in Citrus www.soci.org
Table 2. Effect of fruit load on nitrate nitrogen (NO3 − -N), ammonium nitrogen (NH4 + -N) and proteinaceous nitrogen (N-Prot) contents at fruit
colour break and at fruit full maturity in leaves of ‘Valencia’ sweet orange. Dates of fruit removal and analyses are given as phenological growth stages
of the BBCH scale. BBCH stage 89 coincided with bud sprouting in spring
NO3 − -N (µgg −1 DW) NH4 + -N (µgg −1 DW) N-Prot (mg g −1 DW)
BBCH stage 78 81 89 78 81 89 78 81 89
72 36.0 ± 1.0b 68.0 ± 1.0c 29.0 ± 1.0a 52.1 ± 1.2c 24.0 ± 1.0a 36.0 ± 1.0b 15.7 ± 0.1b 14.5 ± 0.8b 10.4 ± 0.2a
89 17.0 ± 4.9a 41.0 ± 1.0b 46.0 ± 1.0b 51.3 ± 1.4b 28.0 ± 1.0a 29.0 ± 1.0a 18.1 ± 0.4b 16.1 ± 0.2b 11.6 ± 0.1a
∗ ∗ ∗ NS NS NS NS NS NS
F a P ≤ 0.05 NS P ≤ 0.05
D b P ≤ 0.05 P ≤ 0.05 P ≤ 0.01
F × D c P ≤ 0.05 NS NS
∗ Significant at P ≤ 0.05; NS, not significant.
a Fruit removal effect.
b Date effect. Different letters in the same row for a given nitrogen fraction indicate significant differences.
c Fruit removal × date effect.
In ‘Valencia’ sweet orange we found that removal of all fruit
from individual branches on on-crop trees during BBCH stages
72–78 resulted in eightfold more flowers in spring compared
with non-defruited branches on the same tree. Accordingly, the
fruit exerts its inhibitory effect on flowering from the time it is
close to reaching its maximum weight, namely 90% of its final
size (November) in our experiments, similarly to ‘Marsh’ grapefruit
under tropical conditions. 21 Fruit had no effect on flowering from
BBCH stage 72 (43 flowers per 100 nodes) to BBCH stage 78, but,
at BBCH stage 79 (November), flowering was reduced by 80% (8.6
flowers per 100 nodes). This particular time coincides with that of
bud sensitiveness to gibberellic acid (GA3) inhibiting flowering. 22
Later on, during fruit maturation, i.e. from the moment when
the fruit has reached its maximum size onwards, no additional
reduction in flowering was observed up to bud sprouting. García-
Luis et al. 10 suggested, for ‘Owari’ Satsuma mandarin, that the
inhibitory effect of the fruit on flowering is more closely related to
peel maturity than to fruit growth, but their study was carried out
only during the fruit maturation stage. In our experiments with
‘Valencia’ sweet orange the fruit changed colour 45 days after it
attained its maximum size (BBCH stages 81 and 80 respectively)
(Fig. 1). On the other hand, ‘Owari’ Satsuma mandarin matures
earlier than ‘Valencia’ sweet orange and its sensitivity to inhibition
of flowering by exogenous GA3 takes place 1 month later than
it does for sweet orange. 7 These reasons explain the differences
between species.
Soluble sugar contents in leaves were altered by removing
the fruit. In defruited branches, carbohydrates were reduced
immediately in the leaves, since the main carbohydrate sink had
been removed; thereafter there was an increase both in leaves
of branches maintaining all developed fruits and in leaves of
defruited branches, just as García-Luis and Guardiola 23 reported,
followed by a late reduction at the time of fruit maturation, as
was shown for peach. 14 However, at the time of bud sprouting,
non-significant differences in leaf soluble sugar contents due to
crop load were observed, indicating that flowering is unrelated to
carbohydrate contents in leaves, as shown in other citrus species
andcultivars 24,25 andinpeach. 14 Theaccumulationofstarchduring
the winter has been observed by several authors 26 and quantified
by Yelenosky and Guy, 27 who related this accumulation to low
temperatures. The significantly lower starch content in leaves
of branches maintaining all developed fruits seemingly provides
a good reason to correlate starch levels with flowering; however,
differences appear only when fruit load scarcely inhibits flowering.
Moreover, although a minimal amount of carbohydrates is
required for bud sprouting and flower initiation, data from several
studies suggest that, when separate experiments (shading,
girdling, thinning, etc.) are compared, the occurrence of other
controlling factors may obscure the existence of this correlation,
indicating that the flowering–starch relationship remains
unproved. 23,24,28 Similarly, no differences in starch content in the
bark tissue from leaf abscission to bud bursting were observed
in peach trees differing significantly in flowering intensity owing
to crop load. 14 Accordingly, neither soluble sugar content nor
the accumulation of reserve carbohydrates seems to fulfil an inductive
function, but our results show evidence that sprouting
and/or flower formation may require a threshold level of carbohydrates
as an energy source, as reported by Goldschmidt et al. 28
and indicated by the reduction in sucrose, glucose and fructose
concentrations from colour break (BBCH stage 81) to full maturity
and sprouting (BBCH stage 89), regardless of crop load.
In Citrus, certain forms of stress, such as low-temperature stress
and water-deficit stress, result in the accumulation of ammonia
in leaves, and it has been questioned whether this compound is
involved in flowering control. 13 On the other hand, an imbalance
in nitrogen metabolism has been detected in leaves of on-trees
of ‘Wilking’ mandarin compared with off-trees 9 and in phloem
sap of peach trees that kept all fruits up to senescence compared
with trees fully thinned at bloom. 14 This imbalance consists of an
oppositetrendoftotalandNO3 − -N,sothattheoff-/on-treeratiofor
NO3 − -N in leaves, twigs and roots is much less than one. Our results
agree with this disturbance in the nitrate reduction mechanism,
since no reduction in leaf NO3 − -Ncontentwas observed during the
fruit maturation stage in branches maintaining all fruits up to bud
sprouting, whereas branches defruited at BBCH stage 72 showed
significantly reduced leaf NO3 − -N content at bud sprouting (BBCH
stage 89) compared with fruit colour break (BBCH stage 81). The
increased NH4 + -N fraction in these branches reinforces this result.
Accordingly, in Citrus the nitrate-reducing mechanism in leaves
is impaired by the fruit load. Despite this, as NO3 − -N is only a
small fraction of the total nitrogen, ranging from 0.4 to 0.6% (w/w)
(Table 2), this relatively high nitrate amount in leaves of early
defruited trees conveys a metabolic message rather than one of
direct nutritional meaning. 9
Fruit maintained on the tree up to BBCH stage 79 or later
significantly reduced the number of summer/autumn vegetative
J Sci Food Agric 2010; 90: 1936–1943 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1941

1942
shoots that developed, thus reducing the number of branches on
which to bear inflorescences the following spring, as has been
shown in ‘Pixie’ 11 and ‘Nour’ 29 mandarin. Further, fruit removal
also affects the number of nodes on these shoots, which in turn
affects the number of flowers produced during spring bloom. 11,30
The reduction in flowering brought about by fruit load is also
due in part to a decrease in the number of sprouted nodes. This
effectcoincides with thatreported for ‘Owari’ Satsuma mandarin. 10
In that study, as well as in our study with ‘Valencia’ sweet orange,
buds producing generative buds were the most sensitive to fruit
load, whereas vegetative shoots were significantly increased by
leaving the fruit on the tree up to the moment of its maximum
size. Similarly, Okuda et al. 31,32 reported that, during spring bloom
of Satsuma mandarin trees, fruit-bearing trees had a lower
number of leafless and leafy floral shoots but a higher number
of vegetative shoots compared with non-bearing trees. Further,
scoring performed in summer increased flowering of ‘Salustiana’
sweet orange and ‘Owari’ Satsuma mandarin by stimulating bud
sprouting in spring. 33
It is noteworthy that shoot characteristics, i.e. number of flowers
and/or leaves, were not altered by the date of fruit removal. These
results indicate a direct effect of fruit upon the bud, which is
prevented from sprouting. Only buds that elude the effect of fruit
are able to sprout and develop into flowers, 5 thus maintaining the
average number of flowers and leaves. Therefore it seems that fruit
does not affect the number of flowers per sprouted bud but rather
the number of buds that eventually sprout. These results reinforce
the hypothesis thatall buds of adulttrees are induced to flower, but
their development is under the control of an inhibitor and, when
the inhibitor acts, the bud does not sprout or reverses its capacity
to flower and sprouts as a vegetative bud. 4,34 In Citrus, terminal
buds flower mainly on the outside of the canopy, 35 whereas buds
inside the canopy scarcely sprout, but, when trees are cut back,
some of them start flushing and appear as flowers, indicating that
they had undergone floral induction but were kept dormant. 36
In conclusion, in ‘Valencia’ sweet orange, fruit inhibits flowering
from the time it reaches its maximum weight by reducing the
number of nodes to sprout and the number of sprouted nodes, but
it does not affect the number of leaves and/or flowers per sprouted
shoot. Neither soluble sugar content nor starch accumulation in
leaves caused by fruit removal was related to flowering intensity,
but some kind of imbalance in nitrogen metabolism was observed
in trees tending to flower scarcely.
ACKNOWLEDGEMENTS
The authors wish to thank Luisa Moneva for her technical
assistance, Agrimarba, SA for its collaboration and Debra Westall
for editing the manuscript.
REFERENCES
1 Monselise SP and Goldschmidt EE, Alternate bearing in fruit trees. Hort
Rev 4:128–173 (1982).
2 AlbrigoGandGalán V, Flower bud induction, flowering and fruit-set of
some tropical and subtropical fruit tree crops with special reference
to Citrus. Acta Hort 632:81–90 (2004).
3 Goldschmidt EE and Monselise SP, Hormonal control of flowering in
citrus trees and other woody perennials, in Plant Growth Substances,
ed. by Carr DJ. Springer, Berlin, pp. 758–766 (1972).
4 Martínez-Fuentes A, Mesejo C, Juan M, Almela V and AgustíM,
Restrictions on the exogenous control of flowering in Citrus. Acta
Hort 632:91–98 (2004).
www.soci.org A Martínez-Fuentes et al.
5 Lang A, Physiology of flower initiation, in Handbunch der
Pflanzenphysiologie, Vol. XV/1, ed. by Ruhland W. Springer, Berlin,
pp. 1380–1586 (1965).
6 Monselise SP and Halevy AH, Chemical inhibition and promotion of
citrus flower bud induction. JAmSocHortSci84:141–146 (1964).
7 Guardiola JL, Monerri C and Agustí M, The inhibitory effect of
gibberellic acid on flowering in Citrus. Physiol Plant 55:136–142
(1982).
8 Lord EM and Eckard MJ, Shoot development in Citrus sinensis
(L.) Osbeck, cv. ‘Washington’ navel. II. Alteration of development
rate of flowering shoots after GA3 treatment. Bot Gaz 148:17–22
(1987).
9 Monselise SP, Goldschmidt EE and Golomb A, Alternate bearing in
citrus and ways of control. Proc Int Soc Citricult 1:239–242 (1981).
10 García-Luís A, Almela V, Monerri C, Agusti M and Guardiola JL,
Inhibition of flowering in vivo by existing fruits and applied growth
regulators in Citrus unshiu. Physiol Plant 66:515–520 (1986).
11 Verreynne JS and Lovatt CJ, The effect of crop load on budbreak
influences return bloom in alternate bearing ‘Pixie’ mandarin. JAm
Soc Hort Sci 134:299–307 (2009).
12 Goldschmidt EEandGolomb A,Thecarbohydratebalanceofalternatebearing
Citrus trees and the significance of reserves for flowering
and fruiting. JAmSocHortSci107:206–208 (1982).
13 Lovatt CJ, Zheng Y and Hake KD, Demonstration of a change in
nitrogen metabolism influencing flower initiation in Citrus. Isr J Bot
37:181–188 (1988).
14 Reig C, González-Rossia D, Juan M and Agustí M, Effects of fruit load
on flower bud initiation and development in peach. J Hort Sci
Biotechnol 81:1079–1085 (2006).
15 Agustí M, Zaragoza S, Bleiholder H, Buhr L, Hack H, Klose R, et al,
Adaptation of the BBCH scale for the description of Citrus fruits’
phenological stages. Fruits 52:287–295 (1997).
16 Guardiola JL, Agustí M and García-Mari F, Gibberellic acid and flower
bud development in sweet orange. Proc Int Soc Citricult 2:696–699
(1977).
17 Mehouachi J, Serna D, Zaragoza S, Agustí M, Talon M and Primo-
Millo E, Defoliation increases fruit abscission and reduces
carbohydrate levels in developing fruits and woody tissues of
Citrus unshiu. Plant Sci 107:189–197 (1995).
18 Raigón MD, Pérez-García M, Maquieira A and Puchades R, Determination
of available nitrogen (nitric and ammoniacal) in soils
by flow injection analysis. Analysis 20:483–487 (1992).
19 Beljaars PR,Van Dijk Rand Van Der Horst GMJ,Determination of nitrate
in vegetables by continuous flow: interlaboratory study. JAOACInt
77:1522–1529 (1994).
20 Jonkers H, Biennial bearing in apple and pear: a literature survey. Sci
Hort 11:303–317 (1979).
21 Betancourt M, Martinez-Fuentes A, Garcia ME, Mesejo C, Sanchez CD,
Nuñez M, et al, The influence of fruit load on flowering intensity of
grapefruit (Citrus paradisi Macf.) under tropical climatic conditions.
Proc Int Soc Citricult Abstract:203 (2008).
22 El-Otmani M, Coggins CHW, Agusti M and Lovatt CG, Plant growth
regulators in citriculture. World current uses. Crit Rev Plant Sci
19:395–447 (2000).
23 García-Luis A and Guardiola JL, Influence of citrus internal factors and
climatic effects on flowering. Proc Int Soc Citricult 1:292–295 (2000).
24 García-Luís A, Fornes F, Sanz A and Guardiola JL, The regulation of
flowering and fruit set in Citrus: relationship with carbohydrate
levels. Isr J Bot 37:189–201 (1988).
25 Lovatt CJ, Zheng Y and Hake KD, A new look at the Kraus–Kraybill
hypothesis and flowering in Citrus. Proc. 6th Int. Citrus Congr.,Vol.1,
pp. 475–483 (1988).
26 Davenport L, Citrus flowering. Hort Rev 12:349–408 (1990).
27 Yelenosky G and Guy CI, Carbohydrate accumulation in leaves and
stems of ‘Valencia’ orange at progressively colder temperatures.
Bot Gaz 138:13–17 (1977).
28 Goldschmidt EE, Aschkenazi N, Herzano Y, Schaffer AA and
Monselise SP, A role for carbohydrate levels in the control of
flowering in Citrus. Sci Hort 26:159–166 (1985).
29 El-Otmani M, Goumari M, Srairi I, Lbrek A, Charif L and Lovatt CJ, Heavy
fruit load and late harvest inhibit flowering of the ‘Nour’ Clementine
mandarin. Proc Int Soc Citricult 2:525–527 (2004).
30 Agustí M, Biología y control de la floración en el género Citrus. PhD
Thesis, Universidad Politécnica de Valencia (1980).
31 Okuda H, Noda K, Kihara T and Hirabayashi T, Fine root volume,
flowering, sprouting and carbohydrate content in the leaves and
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1936–1943

Flowering and time of fruit remaining on tree in Citrus www.soci.org
roots of bearing and non-bearing Satsuma mandarin trees in a
systemized alternate-bearing orchard. JJpnSocHortSci71:588–590
(2002).
32 OkudaH,KiharaT,NodaKandHirabayashiT,Systemizedalternate
bearing method for mature Satsuma mandarin trees. Bull Natl Inst
Fruit Tree Sci 1:61–69 (2002).
33 Agustí M, Almela V and Pons J, Effects of girdling on alternate bearing
in Citrus. JHortSci67:203–210 (1992).
34 Goldschmidt EE and Samach A, Aspects of flowering in fruit trees. Acta
Hort 653:23–27 (2004).
35 Valiente JL and Albrigo LG, Flower bud induction of sweet orange
trees (Citrus sinensis (L.) Osbeck): effect of low temperatures, crop
load and bud age. JAmSocHortSci129:158–164 (2004).
36 Bangerth KF, Floral induction in mature, perennial angiosperm fruit
trees: similarities and discrepancies with annual/biennial plants
and the involvement of plant hormones. Sci Hort 122:153–163
(2009).
J Sci Food Agric 2010; 90: 1936–1943 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1943

1944
Research Article
Received: 8 March 2010 Revised: 11 May 2010 Accepted: 17 May 2010 Published online in Wiley Interscience: 16 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4039
Improvement of L(+)-lactic acid production
from cassava wastewater by Lactobacillus
rhamnosus B 103
Luciana Fontes Coelho, a Cristian J Bolner de Lima, a
Marcela Piassi Bernardo, a Georgina Michelena Alvarez b
and Jonas Contiero a∗
Abstract
BACKGROUND: L(+)-Lactic acid is used in the pharmaceutical, textile and food industries as well as in the synthesis of
biodegradable plastics. The aim of this study was to investigate the effects of different medium components added in cassava
wastewater for the production of L(+)-lactic acid by Lactobacillus rhamnosus B 103.
RESULTS: The use of cassava wastewater (50 g L −1 of reducing sugar) with Tween 80 and corn steep liquor, at concentrations
(v/v) of 1.27 mL L −1 and 65.4 mL L −1 respectively led to a lactic acid concentration of 41.65 g L −1 after 48 h of fermentation. The
maximum lactic acid concentration produced in the reactor after 36 h of fermentation was 39.00 g L −1 using the same medium,
but the pH was controlled by addition of 10 mol L −1 NaOH.
CONCLUSION: The use of cassava wastewater for cultivation of L. rhamnosus is feasible, with a considerable production of
lactic acid. Furthermore, it is an innovative proposal, as no references were found in the scientific literature on the use of this
substrate for lactic acid production.
c○ 2010 Society of Chemical Industry
Keywords: cassava wastewater; corn steep liquor; response surface methodology
INTRODUCTION
Lactic acid has a wide range of applications in the pharmaceutical,
food, leather, textile and cosmetic industries. 1 One of the
most important applications of lactic acid is in biodegradable
markets, such as polylactic acid, which can be used to improve
physical properties in the production of garbage bags, agricultural
plastic sheeting and computer parts. 2 It can also be applied in
sutures and surgical implants owing to its biocompatible and
bioabsorbable characteristics. 3 L(+)-Lactic acid with high optical
purity provides polylactic acid with a high melting point and high
crystallinity. 4,5 Lactic acid is industrially produced either through
chemical synthesis or microbial fermentation. The advantage of
the fermentation method resides in the fact that an optically pure
lactic acid can be obtained by choosing a strain of lactic acid
bacteria, whereas chemical synthesis always results in a mixture of
L(+)- and D(−)-lactic acid. 6
Refined sugars such as glucose or sucrose have been more
frequently used as the carbon source and yeast extract as a
nitrogen source for lactic acid production, but this is economically
unfavorable.
In order to lower the cost of the production process, a number
of agro-industrial by-products or wastes have been evaluated as
substrates for the production of lactic acid, such as sugarcane, 7
molasses 8 and whey 9 as carbon sources and corn steep liquor
(CSL), 10 a by-product of the corn wet milling industry, as a nitrogen
source.
Lactic acid bacteria have complex nutrient requirements due to
their limited ability to biosynthesize B vitamins and amino acids. 11
Therefore CSL is an excellent source of nitrogen for lactic acid
bacteria because it has a high concentration of amino acids and
polypeptides,withconsiderableamountsofB-complexvitamins. 12
Cassava wastewater (CW) is a residue, generated in large
amounts during the production of cassava flour, composed
of carbohydrates, nitrogen, minerals, and trace elements, and
therefore has potential as a substrate for biotechnological
processes. 13
Cassava wastewater has been used to produce a surfactin by
Bacillussubtilis, 14 citricacidbyAspergillusniger, 15 Polyhydroxyalkanoates
(PHAs) and rhamnolipids by Pseudomonas aeruginosa. 16
Some studies in the literature report the use of cassava bagasse
in the production of lactic acid, 17–21 but no papers were found
on lactic acid production using cassava wastewater as a substrate.
∗ Correspondence to: Jonas Contiero, Department of Biochemistry and Microbiology,
Universidade Estadual Paulista – UNESP, 13506-900 Rio Claro, SP, Brazil.
E-mail: jconti@rc.unesp.br
a UNESP – Department of Biochemistry and Microbiology, Biological Sciences
Institute, Universidade Estadual Paulista, Rio Claro, SP, Brazil
b Instituto Cubano de Investigaciones de los Derivados de la Caña de Azúcar,
(ICIDCA), 488243 Havana, Cuba
J Sci Food Agric 2010; 90: 1944–1950 www.soci.org c○ 2010 Society of Chemical Industry

Improvement of L(+)-lactic acid production from cassava wastewater www.soci.org
Therefore, the use of CW for lactic acid production is an innovative
proposal.
Lactobacillus rhamnosus B 103 has an advantage over other
lactic acid bacteria such as homofermentative metabolism, highest
productivity and production of L(+)-lactic acid that is optically
pure. Therefore the aim of the present study was to investigate the
effects of different medium components on cassava wastewater
for the production of L(+)-lactic acid by Lactobacillus rhamnosus
B 103.
MATERIALS AND METHODS
Microorganism
Lactobacillus rhamnosus B 103 was obtained from the Instituto
Cubano de Investigaciones de los Derivados de la Caña de Azúgar
(ICIDCA). The strain was stored in Man, Rogosa and Sharpe (MRS)
medium with 20% (v/v) glycerol at −20 ◦ C.
Substrates
CSL was obtained from Corn Products Co. (São Paulo state, Brazil)
and CW was collected from cassava flour manufacturer Plaza SA
(São Paulo state, Brazil) and stored at −20 ◦ C until needed. Solids
as well as cyanide were removed from the CW by boiling for 5 min,
followed by cooling and centrifugation at 5000 × g for 10 min. 14,22
The supernatant was used as a production medium.
The same CW was used for all experiments. CW composition:
fructose (24.5 g L −1 ), glucose (30.1 g L −1 ), maltose
(1.8 g L −1 ), nitrate (0.7 g L −1 ), phosphorus (0.9 g L −1 ), potassium
(3.9 g L −1 ), magnesium (0.5 g L −1 ), nitrite (0.05 mg L −1 ), sodium
(23.1 mg L −1 ), iron (6.1 mg L −1 ), zinc (11.1 mg L −1 ), manganese
(4.1 mg L −1 ), copper (14.1 mg L −1 )andprotein(9gL −1 ). The composition
of the CW was determined using the methodology
described by Nitschke and Pastore. 14
Cultivation
The inocula were prepared through the transference of 1 mL of
stock culture to Erlenmeyer flasks containing 100 mL of growth
medium(MRS).MRSgrowthmediumcomposition(g L −1 ):peptone
(10.0), yeastextract(5.0), meatextract(10.0), glucose (20.0), sodium
acetate (5.0), ammonium citrate (2.0), K2HPO4 (5.0), MgSO4.7H2O
Table 1. Plackett–Burman design (real and coded values) with respective resulting lactic acid production
(0.1) and MnSO4.4H2O (0.05). Initial pH of the medium was
adjusted to 6.2. The inocula were incubated at 37 ◦ C, 200 rpm
for 18 h. 10% (v/v) of inoculum was added in all experiments.
Calcium carbonate (50 g L −1 ) was added to experimental medium
(Plackett–Burman experimental design and central composite
design), the pH was adjusted to 6.2 with 2 mol L −1 NaOH. After
that, 45 mL of experimental medium was transferred to a 250 mL
Erlenmeyer flask and incubated in orbital shakers at 37 ◦ C, 200 rpm
for 48 h. In all experiments, the concentration of initial reducing
sugar was 50 g L −1 .
Plackett–Burman experimental design
The purpose of this first step of the optimization was to identify
the medium components with a significant effect on lactic acid
production. Twelve experiments were generated from seven
factors: CSL, sodium acetate, magnesium sulfate, manganese
sulfate, ammonium citrate, potassium phosphate and Tween
80. Variables with a confidence level greater than 95% were
consideredtohaveasignificantinfluenceonlacticacidproduction.
The Plackett–Burman experimental design was based on the
first-order model, with no interaction among the factors. The
concentrations used for each variable are displayed in Table 1.
A central composite design (CCD) was performed with the
variables that significantly increased the production of lactic acid.
CCD and optimization by response surface methodology
A CCD for two independent variables – each at five levels with
four star points (α = 1.41) and four replicates at the center
points – was used to develop a second-order polynomial model,
which determined the optimal values of variables for lactic acid
production. Screened through previous work, CSL and Tween 80
were taken as the variables for investigation.
The variables of the experiments were coded according to the
following equation:
xi = (Xi − Xcp)/Xi i = 1, 2, ..., K (1)
in which xi is the coded value of an independent variable; Xi is
the real value of an independent variable; Xcp is the real value of
Independent variables a Result
Run X1 X2 X3 X4 X5 X6 X7 Lactic acid (g L −1 )
1 2(1) b 0(−1) 2 (1) 0 (−1) 0 (−1) 0 (−1) 30 (1) 29.53
2 2 (1) 5 (1) 0 (−1) 0.2 (1) 0 (−1) 0 (−1) 0 (−1) 19.685
3 0(−1) 5 (1) 2 (1) 0 (−1) 0.05 (1) 0 (−1) 0 (−1) 16.215
4 2(1) 0(−1) 2 (1) 0.2 (1) 0 (−1) 1 (1) 0 (−1) 22.795
5 2 (1) 5 (1) 0 (−1) 0.2 (1) 0.05 (1) 0 (−1) 30 (1) 34.05
6 2 (1) 5 (1) 2 (1) 0 (−1) 0.05 (1) 1 (1) 0 (−1) 19.07
7 0(−1) 5 (1) 2 (1) 0.2 (1) 0 (−1) 1 (1) 30 (1) 30.66
8 0(−1) 0 (−1) 2 (1) 0.2 (1) 0.05 (1) 0 (−1) 30 (1) 29.945
9 0(−1) 0 (−1) 0 (−1) 0.2 (1) 0.05 (1) 1 (1) 0 (−1) 28.1
10 2 (1) 0 (−1) 0 (−1) 0 (−1) 0.05 (1) 1 (1) 30 (1) 39.54
11 0 (−1) 5 (1) 0 (−1) 0 (−1) 0 (−1) 1 (1) 30 (1) 34.38
12 0 (−1) 0 (−1) 0 (−1) 0 (−1) 0 (−1) 0 (−1) 0 (−1) 16.585
a X1,citrate;X2, acetate; X3,K2HPO4; X4,MgSO4; X5,MnSO4; X6, Tween 80; X7,CSL.
b (−1) and (1) are coded levels.
J Sci Food Agric 2010; 90: 1944–1950 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1945

1946
Table 2. Central composite design for optimization of two variables
(each at five levels) and production of lactic acid
Independent variables (mL L −1 ) Results(gL −1 )
Run CSL (X1) Tween 80 (X2) Lacticacid
1 20(−1) a 0.4 (−1) 33.42
2 20(−1) 1.6 (1) 36.21
3 60 (1) 0.4 (−1) 36.41
4 60 (1) 1.6 (1) 41.82
5 11.8 (−1.41) 1 (0) 31.71
6 68.2 (1.41) 1 (0) 41.61
7 40 (0) 0.154 (−1.41) 35.61
8 40 (0) 1.846 (1.41) 38.52
9 40 (0) 1 (0) 38.48
10 40 (0) 1 (0) 38.90
11 40 (0) 1(0) 38.79
12 40 (0) 1 (0) 38.60
a (−1.41), (−1), (0), (1) and (1.41) are coded levels.
an independent variable at the center point; and �Xi is the step
change value.
The behavior of the system is explained by the following
quadratic equation:
Y = b0 + � bixi + � biixi 2 + � bijxixj
in which Y is the predicted response, i.e. lactic acid concentration;
b0 is the offset term; bi is the linear effect; bii is the squared effect;
bij is the interaction effect; and xi is the independent variable.
The Statistica 7.0 (StatSoft, Tulsa, OK, USA) software package
was used for the experimental design and regression analysis of
the experimental data. The response surface was generated to
understand the interactions among the variables. The optimal
points for the variables were obtained from Maple 9.5 (Waterloo
Maple Inc., Waterloo, Ontario, Canada).
Using the CCD method, a total of 12 experiments with various
combinations of CSL and Tween 80 were conducted. Table 2
displays the range and levels of the variables investigated.
In order to validate the optimization of the medium composition,
tests were carried out using the optimized condition in order
to confirm the results of the response surface analysis.
Scale-up fermentation of lactic acid with the optimized
medium
A scale-up fermentation of lactic acid with the optimized medium
was then carried out in a 1.5 L glass vase bioreactor, with an
initial culture volume of 500 mL. Agitation speed and culture
temperature were controlled at 200 rpm and 37 ◦ C, respectively.
The pH was controlled at 6.2 by the automatic addition of
10 mol L −1 NaOH. Samples of 1 mL were withdrawn from the
fermentation broth every 3 h for 48 h and centrifuged at 7800 × g
for 10 min.
Analysis
Lactic acid concentrations were determined using a highperformance
liquid chromatography system equipped with a UV
detector at 210 nm. A Rezex ROA (300 × 7.8 mm, Phenomenex,
Torrance, CA, USA) column was eluted with 5 mmol L −1 H2SO4
www.soci.org LF Coelho et al.
(2)
as a mobile phase at a flow rate of 0.4 mL min −1 and the
column temperature was maintained at 60 ◦ C. Reducing sugars
were measured using the 3.5-dinitrosalicylic acid method. 23 Cell
growth was determined using a spectrophotometer at 650 nm
(OD650) after centrifugation and washing of cells. The dry mass
was determined through a standard curve of optical density versus
dry mass.
RESULTS AND DISCUSSION
Plackett–Burman experimental design
Table 1 displays the Plackett–Burman design matrix (real and
coded values) in 12 experiments with seven variables added to
CW (X1 = citrate, X2 = acetate, X3 = K2HPO4, X4 = MgSO4,
X5 = MnSO4, X6 = Tween 80, X7 = CSL) and respective lactic acid
production.
Figure 1 (Pareto chart) illustrates the effects of the different
variables. CSL was the most influential variable in the production of
lactic acid, followed by Tween 80 and K2HPO4. Only CSL and Tween
80 had a significant positive effect on lactic acid production, with a
95% confidence level, and were therefore used in the optimization
of lactic acid production.
Although Büyükkileci and Harsa 24 reportthatMnSO4 is essential
for L.casei to produce lactic acid because of Mn 2+ , which stimulates
lactate dehydrogenase activity, 25,26 MnSO4 did not significantly
increase the production of lactic acid in the present study. Using
the Plackett–Burman design to study solid-state fermentation
by Lactobacillus amylophilus GV6, Naveena et al. 27 report that
MnSO4.H2O had negative coefficients; MgSO4, sodium acetate
and CSL were found to be insignificant; and ammonium citrate
and Tween 80 improved the production of lactic acid.
According to Honorato et al., 28 the addition of phosphate to the
culture medium increases microorganism growth and enhances
lactic acid production; in this case the component maintains
the pH near the optimal growth value, thereby allowing the
conduction of fermentation for a longer time. In the present
study, however, K2HPO4 had a negative effect on lactic acid
production. This may be explained by the excessive amount of this
componentinthemedium.CWisrichinmanganese(4.1mgL −1 ),
potassium (3.9 g L −1 ) and phosphorus (0.9 g L −1 ), which fulfills the
requirements of the organism. In shake flask fermentation the
pH was maintained by CaCO3 added to the medium. Thus the
addition of K2HPO4 and MnSO4 in CW is not necessary and should
be avoided.
Response surface method
CSL and Tween 80 were further optimized by using the response
surface method. Table 2 displays the design matrix of the variables
in coded units and real values with the respective results. The
application of multiple regression analysis methods yielded the
following regression equation for the experimental data:
Y = 38.69 + 2.82X1 + 1.54X2 − 0.99X1X1 + 0.65X1X2 − 0.788X2X2
(3)
in which Y is the predicted response (lactic acid concentration)
and X1 and X2 are the coded values of the test variables CSL and
Tween 80, respectively.
The highest production of lactic acid was 41.82 g L −1 , obtained
from 60 mL L −1 of CSL and 1.6 mL L −1 of Tween 80 (Table 2).
The response surface quadratic model was performed in
the form of analysis of variance (ANOVA) and the results are
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1944–1950

Improvement of L(+)-lactic acid production from cassava wastewater www.soci.org
Figure 1. Pareto chart for lactic acid production.
Table 3. Analysis of variance for the quadratic model
Source
Sum of
squares
Degrees
of freedom
Mean
square F-value P > F
Model 93.1299 5 18.6259 19.1045 0.001266
Error 5.84974 6 0.9749
Lack of fit 5.74347 3
Pure error 0.10627 3
Total 98.97964 11
Adjusted R 2 = 0.891.
summarizedinTable 3.Fisher’stestwasusedtocheckthestatistical
significance of Eqn (3). ANOVA of the quadratic regression model
demonstrates that the model is highly significant, as is evident
from Fisher’s test (Fcalc(5, 6) = 19.1045 > Ft(5, 6) = 4.387), and
has a very low probability value ((Pmodel > F) = 0.001266).
The value of the adjusted coefficient of determination (adjusted
R 2 = 0.891) is high, which indicates the high significance of the
model. The high R value (0.969) demonstrates strong agreement
between the experimental observations and predicted values. This
correlation was also demonstrated by the plot of predicted versus
observed lactic acid values, since all points were clustered around
the diagonal line, which means that no significant violations of the
model were found. A plot of residuals versus predicted response
displays no pattern or trend, suggesting that the variance of the
original observation is constant.
The independent variables (X1, X2 and X1 2 ) had a significant
effect (observed from the P-values) and the variables X1, X2 had
a positive effect (Table 4). Thus an increase in the concentration
of these variables led to an increase in response (lactic acid
production).
The 3D response surface is the graphic representation of the
regression equation and is plotted to determine the interaction
of the variables and locate the optimal level of each variable for
maximal response.
Table 4. Least-squares fit and parameter estimates
Term Estimate Standard error t P>|t|
Intercept 38.69300 0.493698 78.37381 0.000000
X1 2.82559 0.349097 8.09399 0.000191
X2 1.53892 0.349097 4.40828 0.004528
X1 2 −0.99063 0.390303 −2.53809 0.044194
X1X2 0.65400 0.493698 1.32470 0.233489
X2 2 −0.78813 0.390303 −2.01927 0.089992
A strong interaction between CSL and Tween 80 in lactic
acid production was found (Fig. 2). The area of greatest lactic
acid production is located between 56 and 66 g L −1 of CSL with
1.0–1.5 mL L −1 of Tween 80.
Once lactic acid bacteria are nutritionally fastidious and require
various amino acids and vitamins for growth, it is very important
to choose the right nitrogen source. Nitrogen is necessary for
the synthesis of amino acids, lipids, enzyme cofactors, some
carbohydrates and other substances. The nitrogen source is a
major factor of influence on the growth of Lactobacillus. 29 As
the synthesis of lactic acid by fermentation is associated with cell
growth, there is no product formation if the medium does not have
an adequate concentration of nitrogen for promoting growth. 30
On the other hand, high concentrations of nitrogen can lead to
cell death. 31
The use of a cheap nitrogen source for the complete
replacement of yeast extract has been widely discussed. 32 Yu
et al. 33 found that CSL not only replaces yeast extract as the
sole nitrogen source in an optimized medium, but also helps
to enhance lactic acid production when associated with other
beneficial medium components.
In the present study, the addition of Tween 80 to the
fermentation medium significantly increased the production
of lactic acid (Fig. 2). Authors have reported that Tween 80
is responsible for increasing the growth of Lactobacillus and
J Sci Food Agric 2010; 90: 1944–1950 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1947

1948
www.soci.org LF Coelho et al.
Figure 2. Response surface of lactic acid production by L. rhamnosus B 103 showing the interaction between CSL and Tween 80.
Table 5. Stationary point for lactic acid production and coded values
of the variables X1 and X2 on the optimization point
P0 Lactic acid Coordinates Lactic acid
λ1 −1.229 X1 1.276
λ2 −0.548 X2 0.450
lactic acid production, unsaturated fatty acids such as Tween
80 being essential growth factors. 34,35 On the other hand, higher
concentrations of Tween 80 (1.6%, w/v) decreased lactic acid
production. An explanation for this may be that Tween 80 as a
surfactant could dissolve the lipid in the cell membrane, destroy
the membrane structure and then cause the death of the cell. 36,37
The point of maximal lactic acid production was determined
through canonical analysis of the adjusted model. A study was
carried out to identify the nature of the stationary point (maximal
point, low response or saddle point). An algorithm carried out
on the Maple 9.5 program (Waterloo Maple, Inc.) was used to
calculate the stationary point (P0) for the synthesis of lactic acid.
These values are displayed in Table 5.
λ values referring to CSL and Tween 80 indicate that these
responses have a maximal point, as they have equal, negative signs
(Table 5). The analysis determined that the maximal predicted
lactic acid concentration was 41.58 g L −1 with the corresponding
optimal values of the test variables in uncoded units at 65.4 mL L −1
CSL and 1.27 mL L −1 Tween 80. All optimal points were located
within the experimental range and varied around their center
points to different extents. To confirm the adequacy of the model
for predicting maximal lactic acid production, three additional
experiments in a shaker were performed with this optimal medium
composition. The mean value of lactic acid concentration was
41.65 g L −1 , which is in excellent agreement with the predicted
value of 41.58 g L −1 . Thus the model proved adequate.
The scale-up fermentation of lactic acid in the optimal medium
was carried out in the bioreactor. The time courses are displayed in
Fig. 3. During fermentation, the concentration of reducing sugar
decreased from 50 to 0 g L −1 at the end of cultivation and the
growth of L.rhamnosus B 103 kept increasing quickly and appeared
to reach stationary phase at 36 h. The highest productivity
(1.59 g L −1 h −1 ) was obtained after 12 h of fermentation and the
greatest production (39 g L −1 ) occurred after 36 h. Moreover, the
yield and the average volumetric productivity of lactic acid were
as high as 96% and 4.58 g L −1 h −1 , respectively.
The high yield of lactic acid from CW can be attributed to the
high nutritional value found in the production medium. CW is
a nutritious product containing natural sugars, proteins, amino
acids and vitamins that are suitable for the growth of lactic acid
bacteria. Furthermore, the fermentation medium was carried out
after hydrolysis with heating, which removed the inhibitory and
toxic compounds (hydrogen cyanide) and favored the production
of lactic acid. 22
CONCLUSIONS
The use of CW for cultivation of L. rhamnosus is an innovative
proposal, as no references were found in the scientific literature on
the use of this substrate for lactic acid production. Optimization
of the responses revealed that the best result for lactic acid
production (41.58 g L −1 ) was obtained with 65.4 mL L −1 CSL
and 1.27 mL L −1 Tween 80. Thus the results of the present
study demonstrate that the use of CW for the production of
lactic acid from fermentation by L. rhamnosus B 103 is feasible,
with a considerable production of biomass and lactic acid,
requiringonlysupplementationwithacheapnitrogensource(CSL)
and Tween 80. The Plackett–Burman design, central composite
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1944–1950

Improvement of L(+)-lactic acid production from cassava wastewater www.soci.org
Figure 3. Concentrations of substrate, product and biomass as a function of fermentation time under optimal conditions in a bioreactor (g L −1 ); (�)
reducing sugar; (�)lacticacid,(◦) biomass.
design, response surface method, regression analysis and model
generation were effective methods for the medium optimization
of lactic acid production.
ACKNOWLEDGEMENTS
The authors thank Plaza SA and Corn Products ® for kindly supplying
the cassava wastewater and corn steep liquor, respectively,
and the Brazilian fostering agency Fundação de Amparo a Pesquisa
do Estado de São Paulo (FAPESP) for the fellowships and financial
support.
REFERENCES
1 Datta R, Tsai SP, Bonsignor P, Moon S and Frank J, Technological and
economical potential of polylactic acid and lactic acid derivatives.
FEMS Microbiol Rev 16:221–231 (1995).
2 Ohara H, Biorefinery. Appl Microbiol Biotechnol 62:474–477 (2003).
3 Di Lorenzo ML, Crystallization behavior of poly(L-lactic acid). Eur Polym
J 41:569–575 (2005).
4 RyuHW,YunJSandWeeYJ,Lacticacid,inConcise Encyclopedia of
Bioresource Technology, ed. by Pandey A. Haworth Press, New York
(2003).
5 Yun JS and Ryu HW, Lactic acid production and carbon catabolite
repression from single and mixed sugars using Enterococcus faecalis
RKY1. Proc Biochem 37:235–240 (2001).
6 Lunt J, Large-scale production, properties and commercial
applications of polylactic acid polymers. Polym Degrad Stabil
59:145–152 (1998).
7 Calabia BP and Tokiwa Y, Production of D-lactic acid from sugarcane
molasses, sugarcane juice and sugar beet juice by Lactobacillus
delbrueckii. Biotechnol Lett 29:1329–1332 (2007).
8 Dumbrepatil A, Adsul M, Chaudhari S, Khire J and Gokhale D,
Utilization of molasses sugar for lactic acid production by
Lactobacillus delbrueckii subsp. delbrueckii mutant Uc-3 in batch
fermentation. Appl Environ Microbiol 74:333–335 (2008).
9 Buyukkileci AO and Harsa S, Batch production of L(+)lacticacidfrom
whey by Lactobacillus casei (NRRL B-441). J Chem Technol Biotechnol
79:1036–1040 (2004).
10 Yu L, Lei T, Ren X, Pei X and Feng X, Response surface optimization of
L-(+)-lactic acid production using corn steep liquor as an alternative
nitrogen source by Lactobacillus rhamnosus CGMCC 1466. Biochem
Eng J 39:496–502 (2008).
11 Fitzpatrick JJ and Keeffe UO, Influence of whey protein hydrolyzate
addition to whey permeate batch fermentations for producing
lactic acid. Proc Biochem 37:183–186 (2001).
12 Cardinal EV and Hedrick LR, Microbiological assay of corn steep liquor
for amino acid content. JBiolChem172:609–612 (1948).
13 Nitschke M and Pastore GM, Cassava flour wastewater as a substrate
for biosurfactant production. ApplBiochemBiotechnol 106:295–301
(2003).
14 Nitschke M and Pastore GM, Production and properties of a surfactant
obtained from Bacillus subtilis grown on cassava wastewater.
Bioresour Technol 97:336–341 (2006).
15 Leonel M and Cereda MP, Citric acid production by Aspergillus niger
from ‘Manipueira’, a manioc liquid residue. Sci Agric 52:299–304
(1995).
16 Costa SGVAO, Lépine F, Milot S, Déziel E, Nitschke M and Contiero J,
Cassava wastewater as a substrate for the simultaneous production
of rhamnolipids and polyhydroxyalkanoates by Pseudomonas
aeruginosa. J Ind Microbiol Biotechnol 36:1063–1072 (2009).
17 Xiaodong W, Xuan G and Rakshit SK, Direct fermentative production
of lactic acid on cassava and other starch substrates. Biotechnol Lett
19:841–843 (1997).
18 John RP,Nampoothiri KM and Pandey A, L(+) Lactic acid recovery from
cassava bagasse based fermented medium using anion exchange
resins. Braz Arch Biol Technol 51:1241–1248 (2008).
19 John RP, Nampoothiri KM and Pandey A, Simultaneous saccharification
and fermentation of cassava bagasse for L(+) lacticacidproduction
using Lactobacilli. Appl Biochem Biotechnol 134:263–272
(2006).
20 John RP, Sukumaran RK, Nampoothiri KM and Pandey A, Statistical
optimization of simultaneous saccharification and L(+) lactic
acid fermentation from cassava bagasse using mixed culture of
Lactobacilli by response surface methodology. Biochem Eng J
36:262–267 (2007).
21 Roble ND, Ogbonna JC and Tanaka H, l-Lactic acid production from
raw cassava starch in a circulating loop bioreactor with cells
immobilized in loofa (Luffacylindrica). BiotechnolLett 25:1093–1098
(2003).
22 Westby A, Cassava utilization, storage and small-scale processing,
in Cassava: Biology, Production and Utilization, ed. by Hillocks RJ,
Tresh JM and Bellotti AC. CAB International, Wallingford, UK,
pp. 281–300 (2002).
23 Miller GL, Use of dinitrosalicylic acid reagent for determination of
reducing sugar. Anal Chem 31:426–428 (1959).
24 Büyükkileci AO and Harsa S, Batch production of L(+)lacticacidfrom
whey by Lactobacillus casei (NRRL B-441). J Chem Technol Biotechnol
79:1036–1040 (2004).
J Sci Food Agric 2010; 90: 1944–1950 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1949

1950
25 Fitzpatrick JJ, Ahrens M and Smith S, Effect of manganese on
Lactobacillus casei fermentation to produce lactic acid from whey
permeate. Proc Biochem 36:671–675 (2001).
26 Krischke W, Schroder M and Trosch W, Continuous production of
L-lactic acid from whey permeate by immobilized Lactobacillus
casei subsp. casei. Appl Microbiol Biotechnol 34:573–578 (1991).
27 Naveena BJ, Altaf MD, Bhadriah K and Reddy G, Selection of medium
components by Plackett–Burman design for production of L (+)
lactic acid by Lactobacillus amylophilus GV6 in SSF using wheat
bran. Bioresour Technol 96:485–490 (2005).
28 Honorato TL, Rabelo MC, Pinto GAS and Rodrigues S, Produção de
ácido lático e dextrana utilizando suco de caju como substrato.
Cienc Technol Aliment 27:254–258 (2007).
29 Wood BJB and Holzapfel WH, The Genera of Lactic Acid Bacteria (1st
edn). Blackie Academic and Professional, London (1995).
30 Pritchard G and Coolbear T, The physiology and biochemistry of
the proteolytic system in lactic acid bacteria. FEMS Microbiol
12:179–206 (1993).
31 De Lima CJB, Coelho LF, Blanco KC and Contiero J, Response surface
optimization of D(−) lactic acid production by Lactobacillus SMI8
www.soci.org LF Coelho et al.
using corn steep liquor and yeast autolysate as an alternative
nitrogen source. Afr J Biotechnol 8:5842–5846 (2009).
32 Liggett RW and Koffler H, Corn steep liquor in microbiology. Microbiol
Mol Biol 12:297–311 (1948).
33 Yu L, Lei T, Ren X, Pei X and Feng X, Response surface optimization of
l-(+)-lactic acid production using corn steep liquor as an alternative
nitrogen source by Lactobacillus rhamnosus CGMCC 1466. Biochem
Eng J 39:496–502 (2008).
34 Oh S, Rheem S, Sim J, Kim S and Baek Y, Optimizing conditions for the
growth of Lactobacillus casei YIT 9018 in tryptone-yeast extractglucose
medium by using response surface methodology. Appl
Environ Microbiol 61:3809–3814 (1995).
35 Belhocine D, Investigations on lactose valorization by lactic acid
fermentation. PhD dissertation, University of Rennes I, France (1987).
36 Ben-Kun Q, Ri-Sheng Y, Min L and Sheng-Song D, Effect of Tween 80
on production of lactic acid by Lactobacillus casei. Songklanakarin J
Sci Technol 31:85–89 (2009).
37 Ben-kun Q and Ri-sheng Y, L-Lactic acid production from Lactobacillus
casei by solid state fermentation using rice straw. BioResources
2:419–429 (2007).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1944–1950

Research Article
Received: 7 March 2010 Revised: 25 April 2010 Accepted: 12 May 2010 Published online in Wiley Interscience: 16 June 2010
(www.interscience.wiley.com) DOI 10.1002/jsfa.4040
Antioxidant activities of aged oat vinegar
in vitro and in mouse serum and liver
Ju Qiu, a Changzhong Ren, b,c Junfeng Fan d and Zaigui Li a∗
Abstract
BACKGROUND: The present study focused on the antioxidant activities of aged oat (Avena sativa L.) vinegar. The antioxidant
activities of oat and vinegar have been proved by many previous research studies. It should be noted that oat vinegar, as a
novel seasoning, has antioxidant activity.
RESULTS: Oat vinegar showed stronger radical scavenging activities, reducing power, and inhibition of lipid peroxidation than
rice vinegar. The concentrations of polyphenols and flavonoids in oat vinegar were higher than those in rice vinegar. Ethyl
acetate extract of oat vinegar possessed the most varieties of phenolic acids and showed the strongest antioxidant activity
compared with ethanol and water extracts. At suitable doses of oat vinegar, the malondialdehyde value was decreased, activities
of superoxide dismutase and glutathione peroxidase were promoted, and hepatic damage induced by 60 Co γ -irradiation was
ameliorated in aging mice.
CONCLUSION: Oat vinegar manifested antioxidant activity which was stronger than that of rice vinegar in vitro and the same as
that of vitamin E in vivo.
c○ 2010 Society of Chemical Industry
Keywords: oat (Avena sativa L.) vinegar; antioxidant; in vitro; in vivo
INTRODUCTION
Oxidative free radicals will induce oxidative stress if the
unnecessary free radicals are not eradicated efficiently. Overproduction
of free radicals is toxic to hepatocytes and initiates
reactive oxygen species (ROS). 1 Previous studies indicated that
oxidative stress might play a key role in the pathogenesis of
liver diseases, including drug-induced hepatic damage, alcoholic
hepatitis, and viral hepatitis or ischemic liver injury. 2 Therefore,
antioxidative treatment has been proposed as a potential means
to prevent or attenuate oxidative damage in vivo.
Vinegar, as a widely used acidic seasoning, has medicinal uses by
virtue of its physiological effects, such as antioxidant, antibacterial
activity, promoting recovery from exhaustion, and regulating
blood pressure and blood glucose. 3–6 There have been many
reports concerning antioxidant activities of different vinegars
made from various materials, such as rice vinegar, wine vinegar,
balsamic vinegar and aromatic rice vinegar. 7–10
Oat vinegar is a new kind of aged vinegar, which is produced
mainly from oats. Oats (AvenasativaL.) contain several families
of phytochemicals that display antioxidant properties, such as
tocotrienols, phenolic compounds, flavonoids, sterols, phytic acids
and avenanthramides. 11 Oats contain 8.7 mg kg −1 free phenolic
acids, 20.6 mg kg −1 soluble phenolic acids and 57 mg kg −1
insoluble phenolic acids. The total phenolic acid content (mainly
avenanthramides and ferulic and vanillic acids) was significantly
correlated with antioxidant activity in vitro. 12 However, the
antioxidant activity of oat vinegar and its antioxidant compounds
still need to be researched.
This study focused on the antioxidant activities of oat vinegar
in vitro and in vivo. The indexes of antioxidant activity in vitro
were analyzed by 2,2 ′ -azino-bis-(3-ethylbenzthiazoline-6-sulfonic
acid) (ABTS) radical scavenging activity, reducing power and
inhibition of lipid peroxidation. Aging model mice exposed to 60 Co
γ -irradiation once were orally administered with different doses of
oat vinegar or vitamin E (VE) (as positive control). Enzyme activities
of superoxide dismutase (SOD) and glutathione peroxidase (GSH-
PX), and the values of malondialdehyde (MDA) in liver and blood
serum of mice, as well as liver histopathological section, were
detected to demonstrate the antioxidant activity in mice.
MATERIALS AND METHODS
Materials
Oat vinegar, produced in 2007, was obtained from Shanxi Ziyuan
Microorganism R&D Co., Ltd (Shanxi, China). The vinegar was
produced from oats (about 700 g kg −1 of all raw materials) mixed
∗ Correspondence to: Zaigui Li, Box 40, China Agricultural University, No. 17
Qinghua Dong Lu, Haidian District, Beijing 100083, China.
E-mail: lizg@cau.edu.cn
a College of Food Science and Nutritional Engineering, China Agricultural
University, Haidian, Beijing 100083, China
b Department of Agronomy, China Agricultural University, Haidian, Beijing
100094, China
c Oat Engineering and Technique Research Center of Jilin Province, Baicheng,
Jilin 137000, China
d College of Bioscience and Biotechnology, Beijing Forestry University, Haidian,
Beijing 100083, China
J Sci Food Agric 2010; 90: 1951–1958 www.soci.org c○ 2010 Society of Chemical Industry
1951

1952
Table 1. Compositions of oat vinegar
Content Method
Protein (g L −1 ) 64.8 ± 0.82 AACC 46-11A, 2000
Aminoacid(g L −1 ) 15.2 ± 0.00 AACC 07-01, 2000
Fat (g L −1 ) 9.8 ± 0.02 AACC 30-20, 2000
Reducing sugar (g L −1 ) 129.8 ± 2.09 AACC 80-68, 2000
Moisture (g L −1 ) 889.4 ± 0.11 AACC 44-19, 2000
pH value 3.95 ± 0.01
with rice hull, wheat bran and Koji (oats and pea yeast). It was
processed through saccharification, alcohol fermentation, static
surface acetic acid fermentation, roasting, leaching and aging.
Rice vinegar was purchased from Beijing Liubiju Co. (Beijing,
China). The composition of the oat vinegar is shown in Table 1.
Polyphenolic and flavonoid content of vinegar
Total polyphenolic content in oat vinegar was determined using
the Folin–Ciocalteu method. 13 Oat vinegar (200 µL) at an appropriate
concentration (diluted 20, 40 or 80 times with distilled water)
was added sequentially with 1.0 mL Folin–Ciocalteu reagent,
1.0 mL of 0.707 mol L −1 Na2CO3 solution and 2.8 mL distilled
water. The mixture was analyzed at 765 nm by spectrophotometer
(UVmini-1240, Shimadzu, Kyoto, Japan) 30 min later. Results were
expressed as mg gallic acid equivalents mL −1 vinegar. Rice vinegar
was used as a control and compared with oat vinegar.
Total flavonoid content in oat vinegar was determined as
described by Zhishen et al. 14 Briefly, 1.0 mL of appropriately
diluted vinegar (diluted four, six or eight times with distilled water)
was added to 4.0 mL distilled water. NaNO2 (0.3 mL, 0.725 mol L −1 )
was added immediately, and AlCl3 (0.3 mL, 0.758 mol L −1 )was
added 5 min later. After 6 min, 2.0 mL of 1 mol L −1 NaOH was
added and the solution was made up to 10.0 mL with distilled
water and mixed for 5 s by vibrator. Absorbance was determined
at 510 nm. The total flavonoid content was expressed in mg rutin
equivalent mL −1 vinegar.
Antioxidant activity of vinegar in vitro
The effect of the vinegar on ABTS radical was detected using a
modified ABTS decolorization assay, 15 which is applicable to both
hydrophilic and lipophilic compounds. The ABTS radical cation was
produced by reacting a 7.0 mmol L −1 stock solution of ABTS with
2.45 mmol L −1 potassium persulfate. The mixture was stood in the
dark for at least 12 h at room temperature before use. The ABTS
radical solution was diluted to obtain an absorbance of 0.75±0.05
at 734 nm. Diluted ABTS solution, prepared as described above,
was mixed with vinegar and measured at 405 nm after 6 min.
The parameter IC50, reflecting the concentration of scavenging
free radical to 50%, was calculated using the standard curve,
plotting the percentage for inhibition of absorbance versus the
concentration of sample. The IC50 of the sample was compared
with that of Trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2carboxylic
acid), a reference standard. Finally, the ABTS radical
scavenging activities of samples were expressed as mg Trolox
equivalents mL −1 .
The reducing power of oat vinegar was determined according
to the method of Amarowicz et al. 16 Oat vinegar (0.5 mL),
200 mmol L −1 sodium phosphate buffer (0.5 mL, pH 6.6) and
0.5 mL potassium ferricyanide (0.03 mol L −1 ) were mixed and
www.soci.org J Qiu et al.
incubated in a water bath at 50 ◦ C. After 20 min, 0.5 mL
trichloroacetic acid (0.612 mol L −1 ) was added to the mixture
and centrifuged at 240 × g for 10 min. The supernatant (1.0 mL)
was then mixed with 1 mL distilled water and 1 mL ferric chloride
solution (6.16 mmol L −1 ). The intensity of blue-green color was
measured at 700 nm using the spectrophotometer. The reducing
powers of rice vinegar and VE solution (4.5 mg mL −1 ) were
measured in the same way.
The efficacy of inhibiting lipid peroxidation was determined
according to the method of Zin et al. 17 Vinegar sample (diluted
five times, 4.0 mL), 0.089 mol L −1 linoleicacidinethanol(4.1mL),
0.05 mol L −1 phosphate buffer (pH 7.0, 8.0 mL) and distilled water
(3.9 mL) were mixed and then kept at 40 ◦ C in the dark. Every
24 h, 0.1 mL of this reaction mixture was drawn and mixed with
9.7 mL of 75.0% (v/v) ethanol, 0.1 mL of 3.94 mol L −1 ammonium
thiocyanate and 0.1 mL of 0.02 mol L −1 ferrous chloride in
0.96 mol L −1 hydrochloric acid. After 3 min, the intensity of
red color was measured at 500 nm. These procedures were
repeateduntiltheblankcontrolwithoutsamplereachedmaximum
absorbance. Rice vinegar and VE solution (4.5 mg mL −1 ) were
measured in the same way. The antioxidant activities of all samples
detected by three methods were analyzed in triplicate.
Phenolic compounds and antioxidant activities of oat vinegar
in solvents with different polarities
Ethyl acetate, ethanol and water with different polarities were
used to extract phenolic compounds of oat vinegar consecutively.
Oat vinegar (100 mL) was concentrated to a final volume of 50 mL
at 37 ◦ C in a vacuum rotary evaporator (Laborota 4000, Heidolph,
Schwabach, Germany). Concentrated oat vinegar was extracted
with 150 mL ethyl acetate with sufficient shaking for 30 s and
standing in a water bath at 40 ◦ C for 10 min. The ethyl acetate layer
and residue were separated absolutely. These procedures were
repeated three times and ethyl acetate layers were mixed. The
residue was further extracted successively with ethanol (150 mL)
and water (150 mL) using the same steps as that of ethyl acetate
extracts. All three solutions of extracts were evaporated to dryness
at 37 ◦ C under vacuum. The dried extracts were redissolved in
methanol to a final volume of 50 mL for further analysis by
high-performance liquid chromatography–diode array detection
(HPLC-DAD). Aliquots of dissolved extracts were diluted with
Tris-HCl buffer (pH 7.4, 0.1 mmol L −1 ) to measure ABTS radical
scavenging activity and total polyphenols. The solutions of ethyl
acetate, ethanol and water extracts were analyzed in triplicate.
Ethyl acetate, ethanol and water extracts were analyzed by
HPLC-DAD (Shimadzu, Kyoto, Japan), including a (model LC-
10ATvp instrument with two pumps and DGU-12A degasser)
and a diode array detector (model SPD-M10Avp). Separation
was performed on a Shim-Pack VP-ODS column (150 × 4.6mm
i.d., particle size 5 µm) with a guard column (Shim-pack G VP-
ODS, 10 × 4.6mm,particlesize5µm) (Shimadzu). Mobile phases
consisted of acetonitrile (solvent B) and purified water with 0.1%
trifluoroacetic acid (solvent A) at a flow rate of 1.0 mL min −1 .
Gradient elution was performed as described by Tian et al. 18
Phenolic compounds in the samples were detected at 280 nm by
an external standard method and identified by comparing their
relative retention times and UV spectra with authentic compounds
(Sigma-Aldrich, Steinheim, Germany).
Animals and diets
Sixty Kun Ming male mice (body weight (b.w.) 18–22 g) were
purchased from the Laboratory Animal Center of the Academy
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1951–1958

Antioxidant activities of oat vinegar in vitro and in vivo www.soci.org
of Military Medical Sciences of China, Beijing, China. The mice, in
stainless wire netting cages, were acclimatized under laboratory
conditions (temperature 21–23 ◦ C; relative humidity 55–60%) for
1 week. These mice were then divided randomly into six groups of
10 animals each, referred to as blank, aging model, VE, and three
oat vinegar groups (low, medium and high). All mice were fed with
basic diet ad libitum. Mice in the VE group were administered VE
at a dose of 25.0 mg kg −1 b.w. by gavage every day. Mice in the
oat vinegar groups of low, medium and high were administered
oat vinegar at doses of 1.5, 3.0 and 6.0 mL kg −1 b.w., respectively,
by gavage every day. After 26 days of gastric perfusion, all mice
except the blank group were exposed to 60 Co γ -irradiation at a
dose of 1 Gy min −1 for 4 min. After 30 days of gastric perfusion, all
mice were fasted for 12 h before operation.
The basic dietwas purchased from the LaboratoryAnimal Center
of the Academy of Military Medical Sciences of China, Beijing,
China. The pH value of oat vinegar was adjusted to 6.0 ± 0.1 by
NaOH.
Preparation of blood serum and tissue samples
All mice were cared for according to the Guiding Principles in
the Care and Use of Animals. The experiments were approved by
Peking University Council on Animal Care Committee. At the end
of the trial, blood samples of mice in each group were collected
by extirpating their eyes, and then the livers were immediately
excised. The blood samples collected into tubes were centrifuged
at 3000 × g for 10 min for the separation of serum, which was
stored at −80 ◦ C until analysis.
The livers, washed with ice-cold physiological saline solution
(0.155 mol L −1 ), were stored at −80 ◦ C in liquid nitrogen. Part
of the liver tissue, dipped in 0.1 mol L −1 of ice-cold phosphatebuffered
saline (PBS, pH 7.4, 0.158 mol L −1 KCl), was cut into pieces
and milled to prepare a 100 g L −1 solution of tissue homogenate.
This tissue homogenate was centrifuged at 10 000 × g for 15 min
and the supernatant was kept for analysis. The tissue samples of
livers and blood serum of all mice were used to measure enzyme
activities of SOD and GSH-PX, and the value of MDA.
Antioxidant activity of oat vinegar in vivo
The protein contents of different blood serum and tissue samples
were measured using a bovine serum albumin (BSA) protein assay
kit using BSA as standard. Malondialdehyde (MDA) level was
determined using an MDA assay kit A003 based on the method of
Esterbauer and Cheeseman. 19 SOD activity was determined with
SOD assay kit A001 according to Oyanagui’s method. 20 GSH-PX
activity was determined using a GSH-PX assay kit A005. 21 All kits
were purchased from the Institute of Biological Engineering of
Nanjing Jianchen, Nanjing, China.
Liver tissues were removed immediately after sacrifice and fixed
in 10.0% (v/v) buffered formalin solution for at least 24 h, then
embedded in paraffin wax and sectioned (5.0 µm thickness) for
histopathological evaluation. Liver sections were stained with
hematoxylin and eosin using a standard protocol, and then
analyzed by light microscopy.
Statistical analysis
Statistical analyzes were run using SPSS 13.0 software (SPSS Inc.,
Chicago, IL, USA). Data were expressed as means and standard
deviations. Data were subjected to one-way analysis of variance
(ANOVA). Duncan’s multiple range test was used to determine
differences among means. Results were considered statistically
significant at P < 0.05.
Antioxidant compounds mg mL -1
7
6
5
4
3
2
1
0
*
Polyphenols Flavonoids
*
Oat vinegar
Rice vinegar
7
6
5
4
3
2
1
0
ABTS radical scavenging activity
(mg trolox eq mL-1 )
Figure 1. Concentration of polyphenolic and flavonoid and activity of ABTS
radical scavenging of vinegars. Values represent the mean ± standard
deviation (n = 3). Data were analyzed by ANOVA and within each column
an asterisk indicates statistically different values according to Duncan’s
multiple range test at P < 0.01.
RESULTS AND DISCUSSION
Antioxidant activities of oat vinegar in vitro
Polyphenols and flavonoids were found to be the most effective
antioxidant constituents in oats and vinegars. 9–11,22 As shown in
Fig. 1, the contents of polyphenols and flavonoids in oat vinegar
were 5.29 and 2.04 mg mL −1 , respectively. They were not only
significantly higher than rice vinegar (P < 0.01) but also higher
than traditional balsamic vinegar (3.72 and 0.58 mg mL −1 ) 9 and
Zhenjiang aromatic vinegar (4.18 and 1.10 mg mL −1 ). 10
ABTS radical scavenging activity was determined as a function
of concentration and calculated with the reactivity of Trolox as
standard. 15 The IC50 value of Trolox measured in this study was
4.65 µgmL −1 , which was the same as in a previous report. 23 The
ABTS radical scavenging activity of oat vinegar was 4.42 mg mL −1 ,
which was significantly higher (P < 0.05) than that of rice vinegar
(0.37 mg mL −1 )(Fig.1).
Reducing power was measured by the potassium ferricyanide
reductionmethod.Antioxidantsreducedtheferricion/ferricyanide
complex to the ferrous form: the Perls Prussian blue complex. 24
Antioxidant activity was expressed as the increase in absorbance.
As shown in Fig. 2, reducing power was increased with the
increase in concentrations of vinegars and VE. Oat vinegar showed
significantly higher reducing power than rice vinegar (P < 0.05),
but was not significantly different from VE (P > 0.05).
Peroxidation of fatty acids causes deleterious effects in foods by
forming complex mixtures of secondary breakdown products of
lipidperoxides.Therefore,thevinegarwasfurthercharacterizedfor
antioxidant activities by assessing the ability to protect linoleic acid
against oxidation. The oxidation of linoleic acid without vinegar
was accompanied by a rapid increase in peroxide value (Fig. 3).
Conversely, the peroxide value of oat or rice vinegar increased
slowly, indicating that vinegars had an inhibitory activity on lipid
peroxidation. VE exhibited the most obvious effect and oat vinegar
was the second.
Different systems focused on different mechanisms of the
oxidant defense system were used to measure the antioxidant
activity of oat vinegar in vitro. ABTS assay is based on the
activation of metmyoglobin with hydrogen peroxide in the
presence of ABTS to produce the radical cation, in the presence
or absence of antioxidants. 15 Reducing power based on the
J Sci Food Agric 2010; 90: 1951–1958 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
*
1953

1954
Figure 2. Reducing power of vinegars and vitamin E (VE). VE was used as a
positive control. Values represent the mean ± standard deviation (n = 3).
Figure 3. Inhibition of lipid peroxidation of vinegars and vitamin E (VE).
VE was used as a positive control. Values represent the mean ± standard
deviation (n = 3).
FeCl3/K3Fe(CN)6 system offers a sensitive method for detecting
the electron-donating ability of antioxidants. Antioxidants such as
polyphenols should be able to terminate radical chain reactions
by converting free radicals to more stable products. 16 Besides, the
inhibition of lipid peroxidation was measured by ferric thiocyanate
(FTC) methods. This method measures the ability of antioxidants
to scavenge peroxyl radicals, which react with polyunsaturated
fatty acids, through hydrogen donation. 25 Therefore, reducing
power and inhibition of lipid peroxidation can explain the radical
scavenging ability.
In the present study, the antioxidant activity of oat vinegar
was first evaluated by ABTS radical scavenging activity. This
demonstrated that the antioxidant activity of 1 mL oat vinegar
was equal to that of 4.42 mg Trolox. According to this relationship,
4.5 mg mL −1 VE solution was used as a positive control in
the following measurements of reducing power and lipid
peroxidation. VE showed the same reducing power as oat vinegar
www.soci.org J Qiu et al.
but higher inhibition of lipid peroxidation. That is to say, the ability
of oat vinegar to supply electron and eliminate ABTS radical was
the same as that of VE, but the capacity of scavenging peroxyl
radicals was lower than that of VE.
The results also showed that oat vinegar possessed stronger
antioxidant activity than rice vinegar in all systems. It is considered
that high contents of polyphenols and flavonoids play an
important role in antioxidant activity of oat vinegar. It has been
reported that the antioxidant activities of vinegars are result
mainly from antioxidant compounds such as polyphenols and
flavonoids. 7–9 Polyphenolspotentiallyhaveantioxidantproperties
due to the presence of an aromatic phenolic ring that can stabilize
and delocalize the unpaired electron within its aromatic ring. 12
Verzelloni et al. 9 demonstrated that polyphenols and flavonoids
were highly correlated (r = 0.975 and r = 0.914, respectively)
with the ABTS radical scavenging activity of traditional balsamic
vinegar, as well as red wine vinegar. They also showed that
the concentrations of polyphenols and flavonoids were positive
correlated with reducing power. Alonso et al. 8 showed that the
content of phenolic acids in sherry wine vinegar was highly
correlated with ABTS radical scavenging activities, especially gallic
acid, protocatechuic acid, vanillin, p-coumaric acid, ferulic acid and
vanillic acid. In this study, excellent antioxidant activities of oat
vinegar with high concentrations of polyphenols and flavonoids
also confirmed the results.
Phenolic compounds of oat vinegar extracts
Liquid–liquid extraction is commonly used to analyze polyphenols
and simple phenolics in natural plants for its efficiency and wideranging
applicability. 26 Ethyl acetate, ethanol and water were used
as solvents in the present study. The compositions of polyphenols
and antioxidant activities of their extracts from oat vinegar are
shown in Table 2. Concentrations of gallic acid and (+)-catechin
in ethanol extract were the highest, but the other phenolic acids
in ethyl acetate extract were higher than in ethanol or water
extract. Water extract had less phenolic acid, and only gallic
and protocatechuic acids were detected. Protocatechuic acid
in extracts of oat vinegar was highest, with gallic acid second.
Total polyphenols in ethanol and water extracts of oat vinegar
were significantly (P < 0.05) lower than that in ethyl acetate
extracts, which showed the strongest ABTS radical scavenging
activity. However, the content of total polyphenols and ABTS
radical scavenging activity of water extracts were higher than
those of ethanol extract, which may be ascribed to the aqueous
antioxidant compounds in vinegar. In addition, the summation
of Trolox equivalents of three extracts accounted for about 87%
of that of oat vinegar. That is, three extracts contained primary
antioxidant compounds of oat vinegar, which played a central role
as antioxidant.
Early work aiming at identifying the compounds that were
responsible for the antioxidant properties of oat or vinegar showed
a good correlation between antioxidant capacity and content of
phenolic compounds. 8,27 In oat, ferulic acid was the dominant
phenolic acid, while p-coumaric, caffeic and vanillic acids could
be detected in small quantities. 11 In vinegar, the contents of
gallic, protocatechuic, p-coumaric, ferulic, vanillic and caffeic acids
were very closely correlated with antioxidant activities, according
to research on sherry wine vinegar. 8 Moreover, some studies
showed that (+)-catechin and (−)-epicatechin were present in
red wine vinegar. 28 The concentration of protocatechuic acid in
oat vinegar extracts was higher than that of red wine vinegar, 28
sherry wine vinegar, 8 common white vinegar and rose vinegar. 29
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1951–1958

Antioxidant activities of oat vinegar in vitro and in vivo www.soci.org
Table 2. Compositions of polyphenols and antioxidant activities in different extracts of oat vinegar
Ethyl acetate extract Ethanol extract Water extract
Gallic acid (mg L−1 ) 10.12 ± 0.78Aa 50.38 ± 2.48Ba 5.48 ± 0.12Ca
Protocatechuic acid (mg L−1 ) 185.26 ± 1.48Ab 91.94 ± 1.73Bb 20.65 ± 1.02Cb
(+)-Catechin (mg L−1 ) – 7.26 ± 0.32c –
Vanillic acid (mg L−1 ) 37.39 ± 0.50c – –
Caffeic acid (mg L−1 ) 39.99 ± 2.52c – –
(−)-Epicatechin (mg L−1 ) 16.98 ± 2.83Ad 9.28 ± 2.49Bc –
p-Coumaric acid (mg L−1 ) 10.82 ± 0.79Aa 4.17 ± 0.31Bd –
Ferulic acid (mg L−1 ) 9.31 ± 1.40a – –
Total polyphenols (mg g−1 ) 39.94 ± 3.50A 9.48 ± 1.71B 15.75 ± 2.30C
ABTS radical scavenging activity (mg Trolox
eq. mL −1 ) a
2.13 ± 0.11A 0.70 ± 0.10B 1.04 ± 0.08C
mg L −1 : mean of triplicate determinations ± SD expressed as milligrams of phenolic acid per liter of oat vinegar.
mg g −1 : mean of triplicate determinations ± SD expressed as milligrams of gallic acid equivalents per gram of oat vinegar extracts.
a Mean of triplicate determinations ± SD expressed as milligrams of Trolox equivalents of extracts per milliliter of oat vinegar.
Data were analyzed by ANOVA and statistically significant differences were analyzed by Duncan’s multiple range test at P < 0.05. Different capital
letters (A, B, C) in a row indicate significant difference among extracts. Different lower-case letters (a, b, c) in a column indicate significant difference
among phenolic acids.
Table 3. Values of MDA, SOD and GSH-PX activities in blood serum of mice
Groups MDA (nmol mL −1 ) SOD(UmL −1 ) GSH-PX(UmL −1 )
Blank 5.07 ± 0.55ab 212.81 ± 19.97a 872.18 ± 82.14a
Aging model 5.98 ± 0.38a 174.95 ± 37.92b 734.83 ± 87.63b
VE 5.10 ± 0.53ab 214.62 ± 24.82a 943.35 ± 95.46a
Oat vinegar (low) 4.83 ± 0.36b 229.30 ± 19.96a 863.70 ± 83.79a
Oat vinegar (medium) 4.70 ± 0.412b 236.49 ± 25.67a 875.35 ± 84.44a
Oat vinegar (high) 4.95 ± 0.32b 176.71 ± 29.54b 853.82 ± 80.23a
Data were analyzed by ANOVA and the groups in the same column with different letters indicate statistically significant differences according to
Duncan’s multiple range test at P < 0.05. Values represent the mean ± standard deviation of duplicate assays in 10 animals in each group.
Blank: fed with basic diet; Aging model: fed with basic diet; VE: fed with basic diet and administered VE at a dose of 25 mg kg −1 b.w. by gavage every
day; Oat vinegar low, medium and high: fed with basic diet and oat vinegar administered at a dose of 1.5, 3.0 and 6.0 mL kg −1 b.w. respectively. With
the exception of Blank, other groups were exposed to 60 Co γ -irradiation at a dose of 1 Gy min −1 for 4 min.
However, the content of gallic acid was less than that of red wine
vinegar 28 or sherry wine vinegar. 8 Ferulic acid, the most abundant
phenolic acid in oat, was not so high in oat vinegar, indicating
that the contents of phenolic compounds were changed during
processing. Cerezo et al. have shown the effects of aging and wood
on the phenolic profile of wine vinegar. 30 The effect of processing
such as fermentation, roasting and aging on the antioxidant
activity of oat vinegar should be further studied.
Antioxidant activities of oat vinegar in vivo
Many studies have shown that 60 Co γ -irradiation induces hepatic
damage and reduces SOD and GSH-PX activities in liver and blood
serum. 31 This study used 60 Co γ -irradiation to set up the aging
model. It was shown that SOD and GSH-PX activities of the aging
model group were decreased both in blood serum (Table 3) and
in liver (Table 4).
VE, as a peroxyl radical scavenger, is one of the most popular
natural phenolic type antioxidants. 23 In this study MDA values in
liver were lower, while activities of SOD and GSH-PX in liver and
blood serum of mice in the VE group were higher than those in
the aging model group. That is, VE showed obvious antioxidant
activities in blood serum (Table 3) and liver (Table 4). There was no
significant difference of MDA value and enzyme activities among
VE and blank group in blood serum and liver, indicating that the
antioxidant activity of VE was strong enough to make aging mice
become normal, as the report of Wang et al. 23
Compared with the aging model group, the MDA value of
oat vinegar groups was obviously decreased (P < 0.05) in blood
serum (Table 3). There was no significant difference in MDA values
(P > 0.05) among VE and oat vinegar groups. That is, oat vinegar
had a strong ability to reduce MDA content, similar to VE. The
SOD activities of low and medium oat vinegar groups were higher
(P < 0.05) than that of the aging model group, and not obviously
different (P > 0.05) from that of VE. Hence oat vinegar at a dose of
medium or low could strengthen SOD activity. The GSH-PX activity
of the aging model group was evidently lower (P < 0.05) than that
of other groups, suggesting that oat vinegar could promote GSH-
PX activity in the blood serum of mice. There was no significant
difference in MDA, SOD and GSH-PX values among oat vinegar
(low and medium) and blank groups (P > 0.05). It was concluded
that the antioxidant activity of oat vinegar was high enough to
restore the blood serum of aging mice to normal.
Table 4 demonstrates that the MDA content of oat vinegar
groups was lower (P < 0.05) than that of the aging model in
liver of mice. In the case of low and high doses, there was no
significant difference in MDA value among oat vinegar groups
J Sci Food Agric 2010; 90: 1951–1958 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1955

1956
Table 4. Values of MDA and SOD and GSH-PX activities in liver of mice
www.soci.org J Qiu et al.
Groups MDA (nmol mg −1 protein) SOD (U mg −1 protein) GSH-PX (U mg −1 protein)
Blank 2.21 ± 0.29ab 164.83 ± 26.65a 120.30 ± 26.92a
Aging model 2.63 ± 0.23b 121.05 ± 28.77b 97.61 ± 7.17b
VE 2.12 ± 0.31a 165.24 ± 7.10a 223.73 ± 18.52c
Oat vinegar (low) 1.80 ± 0.20ac 150.63 ± 20.85a 120.16 ± 21.81a
Oat vinegar (medium) 1.60 ± 0.24c 160.38 ± 22.16a 180.99 ± 34.89ac
Oat vinegar (high) 1.74 ± 0.29ac 127.79 ± 14.98b 110.51 ± 20.81a
Data were analyzed by ANOVA and the groups in the same column with different letters indicate statistically significant differences according to
Duncan’s multiple range test at P < 0.05. Values represent the mean ± standard deviation of duplicate assays in 10 animals in each group.
For explanation of groups see note to Table 3.
Figure 4. Effect of oat vinegar on acute liver damage induced by 60 Co γ -irradiation in mice. (A) Normal liver from blank group (magnification 200×).
(B) Liver from aging model (magnification 200×). (C) Liver from aging model (magnification 400×). (D) Liver from VE group (magnification 200×). (E) Liver
from low oat vinegar group (magnification 200×). (F) Liver from medium oat vinegar group (magnification 200×). 1: hepatocyte necrosis or apoptosis; 2:
inflammatory cell infiltration; 3: vacuolar degeneration.
and VE or blank group. For medium dose, however, the MDA
value of oat vinegar was higher than that of VE or blank group.
SOD and GSH-PX activities of medium oat vinegar group were
significantly higher (P < 0.05) than those of the aging model
group, but not evidently different (P > 0.05) from VE and blank
groups. These results suggested that oat vinegar at a medium dose
(3.0 mL kg −1 b.w.) showed the most efficient effects on decreasing
MDA content, while increasing SOD and GSH-PX activities in
mouse liver. The relatively weaker effect of oat vinegar at low
dose (1.5 mL kg −1 b.w.) may be attributed to insufficient intake
of antioxidant compounds such as polyphenols and flavonoids in
the diet, because of the high correlation between concentrations
of polyphenols and flavonoids and the antioxidant activity. 9 As for
high dose (6.0 mL kg −1 b.w.), the antioxidant activities were not
obvious, which may be due to an excessively high concentration
of acetic acid in vinegar, which can cause gastric ulcer 32 or colitis, 33
thus affecting the healthy growth of mice.
MDA content is a good indicator of lipid peroxidation. In
blood serum and liver of mice, the values for oat vinegar groups
were decreased significantly (Tables 3 and 4). This indicated that
the protective role of oat vinegar against oxidative damage
in vivo might be due to the decrease in lipid oxidation. Another
antioxidant mechanism of oat vinegar may be related to
antioxidant enzymes. Antioxidant enzymes such as SOD and GSH-
PX are capable of eliminating active oxygen species and inhibiting
lipid peroxidation to protect tissues from oxidative damage. In
fact, the normal body possesses enzymatic systems to protect
tissues and organs. 34 However, 60 Co γ -irradiation can lead to
the oxidation and the decrease of antioxidant enzyme in vivo. 35
Compared with the aging model group, low and medium dose
of oat vinegar promoted the activities of SOD and GSH-PX both
in blood serum and in liver. This suggests that oat vinegar could
protect tissues and organs from oxidative damage by promoting
the antioxidant enzyme. Interestingly, a high dose of oat vinegar
can promote GSH-PX activity but not SOD activity, which might
be attributed to the different effects of oat vinegar on different
antioxidant enzymes.
Moreover, the change in enzyme activities may be related to
the absorption and metabolism of antioxidant components in oat
vinegar. Phenolic type antioxidant influences the polyethylene
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1951–1958

Antioxidant activities of oat vinegar in vitro and in vivo www.soci.org
network formulation and radical yield during the process of
irradiation. 35 Phenolic acids are believed to act mainly as free
radical scavengers or chelators of transition metals. However, their
mechanisms of action in vivo are not fully elucidated.
Histopathological analyses
Histopathological changes of mice from blank, aging model, VE
and oat vinegar (low and medium) groups are compared in Fig. 4.
There was no abnormal appearance or histological change in the
liver of normal mice (Fig. 4(A)). Irradiation with 60 Co γ -rays caused
damage in the mouse liver, demonstrated by severe hepatocyte
necrosis or apoptosis, inflammatory cell infiltration (Fig. 4(B)) and
vacuolar degeneration in the aging model group (Fig. 4(C)). It was
reported that 60 Co γ -irradiation induced physical and chemical
damage to tissues, leading to organelle injuries, cell death or
neoplastic transformation. 31
Conversely, in the VE group, much vacuolar degeneration was
observed but hepatocyte necrosis, apoptosis and inflammatory
cell infiltration were not obvious (Fig. 4(D)). This means that
irradiation damage in the VE group was mitigated somewhat
compared with that of the aging model group. VE acts as a free
radical scavenger, more specifically within cell membranes via
preventing the oxidation of polyunsaturated lipids by free radicals
such as hydroxyl radicals (OH). Thus VE can improve various
parameters of oxidative stress in animals. 12 The same result was
detected in vivo in this study.
Compared with the blank group, some vacuolar degeneration
and hepatocyte necrosis or apoptosis could still be observed
in liver from the low oat vinegar group (Fig. 4(E)). However, in
the medium oat vinegar group, not only were inflammatory cell
infiltration and vacuolar not obvious, but also little hepatocyte
necrosis or apoptosis was found (Fig. 4(F)), suggesting that oat
vinegar of the medium group significantly ameliorated the liver
injuries induced by 60 Co γ -irradiation. Exposure of cells to ionizing
radiation could lead to an increase in ROS, including hydroxyl
radicals (OH ·), superoxide anions (O2 − ), singlet oxygen ( 1 O2)
and hydrogen peroxide (H2O2). 34 Our experiment showed that
oat vinegar has obvious radical scavenging activity in vitro and
the ability to promote antioxidant activity of SOD or GSH-PX
in vivo. Compared with the result of the blank group, oat vinegar
also restored antioxidant enzymes like SOD and GSH-PX to
near-normal levels. Thus oat vinegar can attenuate oxidative
damage in liver and exhibit antioxidant activities in aging mice.
CONCLUSIONS
In summary, the evidentantioxidantactivities of oatvinegar in vitro
were confirmed by high ABTS radical scavenging activity, stronger
reducing power and the inhibitory capacity of lipid peroxidation.
Ethyl acetate extract of oat vinegar contained more varieties
of phenolic acids and showed stronger antioxidant activity
than ethanol and water extracts. Three kinds of extracts of oat
vinegar included major antioxidant compounds. Protocatechuic
acid in extracts of oat vinegar was 2–20 times higher than
other phenolic acids. Oat vinegar increased SOD and GSH-PX
activities, while decreasing the MDA content in aging mice. Results
suggested that oat vinegar at a medium dose of 3.0 mL kg −1 b.w.
had the strongest antioxidant activity. Also, histopathological
assessment suggested that oat vinegar significantly ameliorated
the liver injuries induced by 60 Co γ -irradiation because less
cell infiltration and vacuolar degeneration were observed, and
hepatocyte necrosis or apoptosis was rarely found when aging
mice were administered suitable oat vinegar. Further studies on
the identification and purification of other components, besides
phenolic acid, responsible for the antioxidant activities in oat
vinegar are now in progress.
ACKNOWLEDGEMENTS
This study was supported by nyhyzx07-009 (public industry project
of the Ministry of Agriculture) and nycytx07-14 (the earmarked
fund for Modern Agro-industry Technology Research System).
REFERENCES
1 Bissel DM, Gores GJ, Laskin DL and Hoorhagle JH, Drug-induced liver
injury: mechanisms and test systems. Hepatology 33:1009–1013
(2001).
2 Albano E, Free radical mechanisms in immune reactions associated
with alcoholic liver disease. Free Radic Biol Med 32:110–114 (2002).
3 Tsujihata S, Entani E, Asai M, Tsukamoto Y and Ohta M, Mathematical
modeling to predict the bactericidal effect of processed vinegar on
Escherichia coli O157:H7. Int J Food Microbiol 43:135–138 (1998).
4 Fushimi T, Tayama K, Fukaya M, Kitakoshi K, Nakai N, et al, Acetic acid
feeding enhances glycogen repletion in liver and skeletal muscle of
rats. JNutr131:1973–1977 (2001).
5 Liljeberg H and Bjorck I, Delayed gastric emptying rate may explain
improved glycaemia in healthy subjects to a starchy meal with
added vinegar. Eur J Nutr 52:368–371 (1998).
6 Kondo S, Tayama K, Tsukamoto Y, Ikeda K and Yamori Y,
Antihypertensive effects of acetic acid and vinegar on
spontaneously hypertensive rats. Biosci Biotechnol Biochem
65:2690–2694 (2001).
7 Shimoji Y, Tamura Y, Nakamura Y, Nanda K, Nishidai S, Nishikawa Y,
et al, Isolation and identification of DPPH radical scavenging
compounds in Kurosu (Japanese unpolished rice vinegar). JAgric
Food Chem 50:6501–6503 (2002).
8 Alonso AM, Castro R, Rodriguez MC, Guillen DA and Barroso CG, Study
of the antioxidant power of brandies and vinegars derived from
sherry wines and correlation with their content in polyphenols.
Food Res Int 37:715–721 (2004).
9 Verzelloni E, Tagliazucchib D and Conte A, Relationship between the
antioxidant properties and the phenolic and flavonoid content in
traditional balsamic vinegar. Food Chem 105:564–571 (2007).
10 Xu QP, Tao W and Ao Z, Analysis of the source of antioxidant
compounds in Zhenjiang vinegar. Food Ferment Ind 31:33–36
(2005).
11 Peterson DM, Oat antioxidants. J Cereal Sci 33:115–129 (2001).
12 Is Fardet A, Rock E and Remesy C, The in vitro antioxidant potential
of whole-grain cereals and cereal products well reflected in vivo?
J Cereal Sci 48:258–276 (2008).
13 Peterson DM, Emmons CL and Hibbs AH, Phenolic antioxidants and
antioxidant activity in pearling fractions of oat groats. J Cereal Sci
33:97–103 (2001).
14 Zhishen J, Mengeheng T and Jianming W, The determination of
flavonoid contents in mulberry and their scavenging effects on
superoxide radicals. Food Chem 64:555–559 (1999).
15 Re R, Pellegrini N, Proteggente A, Pannala A, Yang M and Rice-Evans C,
Antioxidant activity applying an improved ABTS radical cation
decolorization assay. Free Radic Biol Med 26:1231–1237 (1999).
16 Amarowicz R, Pegg RB, Rahimi-Moghaddam P, Barl B and Weil JA,
Free-radical scavenging capacity and antioxidant activity of
selected plant species from the Canadian prairies. Food Chem
84:551–562 (2004).
17 Zin ZM,Abdul-Hamid AandOsman A,Antioxidativeactivityofextracts
from Mengkudu (Morinda citrifolia L.) root, fruit and leaf. Food Chem
78:227–231 (2002).
18 Tian S, Nakamura K and Kayahara H, Analysis of phenolic compounds
in white rice, brown rice and germinated brown rice. J Agric Food
Chem 52:4808–4813 (2004).
19 Esterbauer H and Cheeseman KH, Determination of aldehydic lipid
peroxidation products: malonaldehyde and 4-hydroxynonenal.
Meth Enzymol 186:407–421 (1990).
J Sci Food Agric 2010; 90: 1951–1958 c○ 2010 Society of Chemical Industry www.interscience.wiley.com/jsfa
1957

1958
20 Oyanagui Y, Reevaluation of assay methods and establishment of
kit for superoxide dismutase activity. Anal Biochem 142:290–296
(1984).
21 Paglia DE and Valentine WNJ, Studies on the quantitative and
qualitative characterization of erythrocyte glutathione peroxidase.
J Lab Clin Med 70:158–169 (1967).
22 Peterson DM, Hahn MJ and Emmons CL, Oat avenanthramides exhibit
antioxidant activities in vitro. Food Chem 79:473–478 (2002).
23 Wang D, Wang LJ, Zhu FX, Zhu JY, Chen XD, Zou L, et al, In vitro and
in vivo studies on the antioxidant activities of the aqueous extracts
of Douchi (a traditional Chinese salt-fermented soybean food). Food
Chem 107:1421–1428 (2008).
24 Chou ST, Chao WW and Chung YC, Antioxidative activity and safety of
50% ethanolic red bean extract (Phaseolus radiatus L. var. Aurea).
J Food Sci 68:21–25 (2003).
25 Rumbaoa RGO, Cornago DF and Geronimo IM, Phenolic content and
antioxidant capacity of Philippine sweet potato (Ipomoea batatas)
varieties. Food Chem 113:1133–1138 (2009).
26 Stalikas CD,Extraction,separation and detection methods forphenolic
acids and flavonoids. JSepSci30:3268–3295 (2007).
27 Mikulajova A, Takacsova M, Rapta P, Brindzova L, Zalibera M and
Nemeth K, Total phenolic contents and antioxidant capacities of
cereal and pseudocereal genotypes. J Food Nutr Res 46:50–157
(2007).
28 Cerezo AB, Tesfaye W, Torija MJ, Mateo E, Garcia-Parrilla MC and
Troncoso AM, The phenolic composition of red wine vinegar
produced in barrels made from different woods. Food Chem
109:606–615 (2008).
www.soci.org J Qiu et al.
29 Lopez F, Pescador P, Guell C, Morales ML, Garcia-parrilla MC and
Troncoso AM, Industrial vinegar clarification by cross-flow
microfiltration effect on colour and polyphenol content. J Food
Eng 68:133–136 (2005).
30 Cerezo AB, Tesfaye W, Soria-Diaz ME, Torija MJ, Mateo E, Garcia-
Parrilla MC, et al, Effect of wood on the phenolic profile and sensory
properties of wine vinegars during ageing. J Food Compos Anal
23:175–184 (2010).
31 Chung HC, Kim SE, Lee MG, Cho CK, Kim TH, Lee DH, et al,
Mitochondrial dysfunction by γ -irradiation accompanies the
induction of cytochrome P450 2E1 (CYP2E1) in rat liver. Toxicology
161:79–91 (2001).
32 WenCY,ItoM,MatsuuM,FukudaE,ShichijoK,NakashimaM,et al,
Mechanism of the antiulcerogenic effect of IL-11 on acetic acidinduced
gastric ulcer in rats. Life Sci 70:2997–3005 (2002).
33 Ashizuka S, Ishikawa N, Kato J, Yamaga J, Inatsu H, Eto T, et al, Effect
of adrenomedullin administration on acetic acid-induced colitis in
rats. Peptides 26:2610–2615 (2005).
34 Li CR, Zhou Z, Zhu D, Sun YN, Dai JM and Wang SQ, Protective effect
of paeoniflorin on irradiation-induced cell damage involved in
modulation of reactive oxygen species and the mitogen-activated
protein kinases. Int J Biochem Cell B 39:426–438 (2007).
35 Andreucetti NA, Sarmoria C, Curzio OA and Valles EM, Effect of the
phenolic antioxidants on the structure of gamma-irradiated model
polyethylene. Radiat Phys Chem 52:177–182 (1998).
www.interscience.wiley.com/jsfa c○ 2010 Society of Chemical Industry J Sci Food Agric 2010; 90: 1951–1958