Posterhall Metabolomics 2012 eBook - Bruker

Metabolomics @ Bruker 

All the pieces to solve the Metabolomics Puzzle 

Poster Hall Edition 2 

Innovation with Integrity 

Mass Spectrometry, NMR

-2- 

Table of content 

Food Metabolomics 

Urinary excretion of phenolic compounds from Hibiscus sabdariffa 

in Humans by advanced analytical techniques 4 

RP-HPLC-DAD-ESI-TOF for the pharmacokinetic study of quercetin 

in adipocytes 5 

Metabolomic consequences of aronia juice intake by human 

volunteers 6 

Bacterial Metabolomics 

Comprehensive workflow for Metabolite profiling and identification 

of secondary metabolites from myxobacteria 7 

The impact of high-resolution MS and NMR techniques on discovery 

of novel products from myxobacteria 8 

Discovery of novel myxobacterial secondary metabolites by 

mass spectrometry – based metabolome mining 9 

Metabolic profiling of a Corynebacterium glutamicum ∆prpD2 by 

GC-APCI high-resolution Q-TOF analysis 10 

GC-APCI-MS/MS based identification of metabolites in bacterial 

cell extracts 11 

Yeast Metabolomics 

Untargeted metabolic profiling of yeast extracts by UHR-Q-TOF 

analysis to study arginine biosynthesis mutants 12 

Clinical Metabolomics 

A workflow for metabolic profiling and determination of the 

elemental compositions using MALDI-FT-ICR MS 13 

NMR-based screening of neonate urines 14 

NMR-based metabolomics in clinical studies of human population: 

Quantification 15 

Plant Metabolomics 

Structure elucidation and confirmation for plant metabolomics: 

Novel approaches 16 

MS-based dereplication and classification of diterpene glycoside 

profiles of 23 Solanaceous plant species 17 

GC-APCI-QTOF-MS: An innovative technique for metabolite 

profiling studies 18 

Improved method for characterization of phenolic compounds from 

rosemary extracts using ESI-Qq-TOF MS 19 

Sulfur-containing metabolite-targeted analysis using liquid 

chromatography – mass spectrometry 20 

Invertebrate Metabolomics – 

Drosophila melanogaster and Caenorhabditis elegans 

Metabolic characterization of Sod1 null mutant flies using liquid 

chromatography – mass spectrometry 21 

A mass spectrometry based metabolomic platform for the analysis 

of Caenorhabditis elegans 22 

Complementary GC-EI-MS and GC-APCI-TOF-MS analysis of a 

short-lived mitochondrial knockdown of C. elegans to study metabolic 

changes during aging 23 

Structure Elucidation 

The Fastest Way to a Molecular Formula: How fast can it be? 

TLC-UHPLC-ESI-QTOF MS Coupling for Molecular Formula 

Determination of New Natural Products 24 

MALDI Imaging 

Automated tissue state assignment for high-resolution 

MALDI-FTMS Imaging data 25 

References – Further Reading 26-27

„Putting Together the Pieces to Solve 

the Metabolomics Puzzle...“ 

Metabolomics has emerged as an exciting new area of 

biological research and gained substantial attention in recent 

years. Previously, significant efforts have been made to probe 

and understand the genomes, transcriptomes and proteomes 

of a variety of organisms. Despite massive investments to 

sequence and identify cellular DNA, RNA and proteins, our 

understanding of many key biological functions remains 

rudimentary and incomplete. Many researchers believe that 

perhaps more complete answers can be gleaned from studying 

small molecules that are utilized for a variety of metabolic and 

cellular functions. Because of this belief, in many laboratories, 

Metabolomics is catching up with traditional Proteomics 

and Transcriptomics efforts and has been christened 

“Biochemistry’s new look” [1]. 

Typically, during a Metabolomics study, a large number of 

cell or patient samples are analyzed and compared to identify 

potential biomarker compounds which correlate to a disease, 

drug toxicity, or genetic or environmental variation. Validated 

biomarkers can form the basis for new diagnostic assays, and 

lead the way to help establish truly personalized medicine, 

since changes in small molecule concentrations are closely 

related to the observed phenotype. 

Answering the Challenges of Metabolomics 

Metabolomics researchers are regularly confronted with the 

daunting tasks of acquiring large sets of data and then analyzing 

them in minute detail. Due to the chemical diversity of small 

molecule metabolites, it is impossible to study the entire 

metabolome using a single analytical technique or technology. 

Bruker’s comprehensive solution for metabolomics is based on 

putting together “the pieces of the puzzle” to give a complete 

picture: The Metabolome’s chemical space can be fully 

explored by utilizing Bruker’s high performance LC-MS, GC-MS, 

CE-MS and NMR systems in conjunction with dedicated 

software for data evaluation. 

One of the most challenging tasks in the Metabolomics studies 

is the identification of unknown compounds. Exact mass 

and isotopic pattern information in MS and MS-MS spectra 

acquired by the solariX (FT-MS) or micrOTOF and maXis series 

(Q)-TOF-MS instruments can be used to definitively identify 

unknown compounds. The intrinsic capabilities of these 

instruments in conjunction with the SmartFormula3D software 

enables accurate and reliable molecular formula generation for 

a wide variety of unknowns. Subsequent database queries 

identify likely compound structures, which can be confirmed 

by complementary information derived from NMR spectra. 

In order to enhance separation and increase throughput, 

many laboratories have turned to utilization of UHPLC for 

Metabolomics studies. Since the mass accuracy, isotopic 

fidelity and resolution of maXis and micrOTOF-series ESI-(Q)- 

TOF instruments are independent of the acquisition rate, they 

are perfectly suited for a coupling to UHPLC separations. 

By utilizing a Bruker solution, their high-resolution and accurate 

mass LC-MS capabilities enable both targeted and untargeted 

Metabolomics workflows. In addition, the broad dynamic 

range characteristics of these systems allows both, high- and 

low-abundance compounds to be studied within the same 

dataset – even within the same mass spectrum. 

Finally, Bruker offers the capability to integrate both LC-MS 

and NMR techniques into a single analytical system for 

maximum metabolite coverage and molecular identification. 

The MetabolicProfilerTM is a powerful combination of 

techniques and is exclusively designed to address all analytical 

needs for metabolic profiling and structure elucidation. 

This poster hall is a “snapshot” of some of the current work 

utilizing a variety of Bruker solutions for Metabolomics studies. 

It nicely illustrates Bruker’s aim of providing and putting 

togehter the “puzzle pieces” to give a complete picture of 

the metabolome. We would like to thank all of our partners 

and customers for their constant input and feedback. This 

information has served as an invaluable guide for us in our 

solution development efforts. A special thanks are directed 

to everyone who has worked to generate the outstanding 

posters which we have assembled in this poster hall for your 

benefit. 

Sincerely yours, 

Dr. Aiko Barsch 

Market Manager Metabolomics 

-3-

-4- 

Metabolomics Posterbook 2012 


I. Borrás‐Linares (1, 2) , R. Beltrán‐Debón (3) , Gabriela Zurek (4) , Jorge Joven (3) , D. Arráez‐Román (1, 2) , 

A. Segura‐Carretero (1, 2) , A. Fernández‐Gutiérrez (1, 2) 

(1) Department of Analytical Chemistry, Faculty of Sciences, University of Granada, Granada, Spain. 

(2) Functional Food Research and Development Center (CIDAF), Health Science Technological Park, Granada, Spain. 

(3) Centre de Recerca Biomèdica, Hospital Universitari de Sant Joan, IISPV, Universitat Rovira i Virgili, Reus, Spain. 

(4) Bruker Daltonik GmbH, Bremen, Germany. 

Hibiscus sabdariffa (HS) L. (family: Malvaceae), usually named bissap, karkade or roselle is a tropical plant commonly used as local soft drink. This plant contains organic 

and phenolic acids, anthocyanins, and flavonols as main compounds [1,2]. Traditionally the aqueous extracts of HS have been used in folk medicine to treat 

hypertension, fever, inflammation, liver disorders, and obesity. 

In recent years, this plant has been studied in depth because it exerts a great number of pharmacological activities. Previous studies have shown that HS possesses anti‐ 

tumor and antioxidant properties [3, 4]. Recent data also indicate that aqueous extracts of HS might ameliorate metabolic disturbances [5, 6]. The antioxidant and 

antihyperlipemic properties of HS aqueous extract have been previously reported using a mouse model [7]. Furthermore, the hypolipemic properties of PEHS have been 

recently reported in a cell model for adipogenesis and insulin resistance [8]. However, to establish conclusive evidence for the effectiveness of phenolic compounds 

from this plant on health, it is very important to define the bioavailability and the excretion of these compounds in humans. The objective of this study was to identify 

the urinary excretion of phenolic compounds in humans after the ingestion of a phenolic‐enriched Hibiscus sabdariffa extract (PEHS). 

1 

Hibiscus sabdariffa 

flowers 

PEHS pills 

In this study a phenolic‐enriched Hibiscus sabdariffa extract 

(PEHS) has been obtained and encapsulated. Urine samples 

from a healthy volunteer were collected before and 2, 4 and 6 

hours after intake of 500 mg of the PEHS for the evaluation of 

the urinary excretion of the phenolic compounds present in this 

plant extract. 

2 

PEHS 

1000 μl AcOEt 

10 μl HCl 1M 

500 μl Sample 

Vortex 

Supernatant Dryness 

Centrifuge 

18800 g 

5 min 

250 μl 

Mobil phase A UHPLC‐ESI‐UHR‐Q‐TOF 

analysis 

UHPLC 

BPC OBTAINED BY 

UHPLC‐ESI‐UHR‐Q‐TOF 

[1] Salah, A. M., Gathaumbi, J., Vierling, W. (2002) Phytother. Res. 16:283‐285. 

[2] Tseng T. H., Hsu J., Dong L., Ming H., Chu C. Y. (1998) J. Agric. Food Chem. 46: 1101‐1105. 

[3] Chang Y. C., Huang H. P., Hsu J. D. (2005) Toxicol Appl Pharmacol. 205: 201‐207. 

[4] Ochani P. C., D´Mello P. (2009). Indian J. Exp. Biol. 47: 276‐282. 

[5] Carvajal‐Zarrabal O., Waliszewski S. M., Barradas‐Dermitz D. M. A. (2005) Plant Food Hum. Nutr. 60: 153‐159. 

[6] Alarcon‐Aguilar F. J., Zamilpa A., Pérez‐García M.D. (2007) J. Ethnopharmacol. 114: 66‐72. 

[7] Fernández‐Arroyo S., Rodríguez‐Medina I.C., Beltrán‐Debón R. (2011) Food Res. Int. 44: 1490‐1495. 

[8] Herranz‐López M., Fernández‐Arroyo S., Pérez‐Sánchez A. (2012) Phytomed. 19: 253‐261. 

3 

•Instrument: Dionex Ultimate 3000 

UHPLC 

•Column: Zorbax Eclipse Plus C18, 

3x250 mm, 5 μm 

•Flow: 0.3 mL/min 

•Temperature: 20 °C 

•Injection volume: 5 μl 

•Phase A: H 2O/AcN 90:10 + 0.5% 

Formic acid 

•Phase B: AcN 

•Gradient: : 

•from 0 to 20 min, from 5% B to 20% 

B, 

•from 20 to 25 min, from 20%B to 

40%B, 

•from 25 to 30 min, from 40% B to 

5% B, 

•from 30 to 35 min, to hold 5% B. 

UHR‐Q‐TOF 

•Instrument: Bruker Daltonics 

MaXis 4G 

UHR‐Q‐TOF 

•Mode: Negative 

•Mass range: 50‐1000 m/z 

•Nebulizing gas pressure: 2 bar 

•Drying gas flow: 8 L/min 

•Drying gas temp: 200 ºC 

•Calibrant: sodium formate 5mM 

•End Plate offset: ‐500 V 

•Capillary: ‐ 4500V 

•Funnel RF: 300,0 Vpp 

•Ion Energy: 4,0 eV 

•Collision Energy: 8,0 eV 

•Collision RF: 400,0 Vpp 

•Transfer time: 65 μs 

•PrePulseStorage time: 6 μs 

Compound Retention time (min) [M‐H ] ‐ 

This powerful analytical technique 

revealed the presence of several 

phenolic compounds from the PEHS 

in the urine samples. The main 

compounds detected previously 

described in the extract were 

hibiscus acid, hibiscus acid 

dimethylesther, hydroxycitric acid, 

chlorogenic acid isomers, methyl 

digallate, cumaroylquinic acid and 

N‐feruloyltyramine. The maximum 

excretions of these compounds 

were found at 2 and 4 hours after 

the intake of PEHS. 

Molecular Formula 

Hydroxycitric acid 3,40 207,0146 C6H8O8 

Hibiscus acid 3,60 189,0040 C6H6O7 

Chlorogenic acid Isom I 5,30 353,0878 C16H18O9 

Hibiscus acid dimethylesther 6,00 217,0353 C9H10O7 

Chlorogenic acid 6,90 353,0878 C16H18O9 

Chlorogenic acid Isom II 7,10 353,0878 C16H18O9 

Methyldigallate 7,60 335,0408 C15H12O9 

Coumaroylquinic acid 9,60 337,0928 C16H18O8 

N‐Feruloyltyramine 25,60 312,1241 C18H20NO4 

The authors are grateful to the Spanish Ministry of Science and Innovation for the 

project AGL2011‐29857‐C03‐02, Andalusian Regional Government Council of 

Innovation and Science for the excellence project P10‐FQM‐6563 and P11‐CTS‐7625 and 

to GREIB.PT.2011.18 project. IBL acknowledges financial support from the Spanish 

Ministry of Education and Science (FPI grant, BES‐2009‐028128).


1 

I. Borrás‐Linares (1, 2) , M. Herranz‐López (3) , E.Barrajón‐Catalán (3) , D. Arráez‐Román (1, 2) , V. Micol (3) , 

A. Segura‐Carretero (1, 2) , A. Fernández‐Gutiérrez (1, 2) 

(1) Department of Analytical Chemistry, Faculty of Sciences, University of Granada, Granada, Spain. 

(2) Research and Development of Functional Food Centre (CIDAF), Health Science Technological Park, Granada, Spain. 

(3) Institute of Molecular & Cell Biology (IMCB), Miguel Hernández University, Elche, Spain. 

Hibiscus sabdariffa (HS) L. (family: Malvaceae), usually named bissap, karkade or roselle is a tropical plant commonly used as local 

soft drink. The aqueous extracts of HS have been used in folk medicine to treat hypertension, fever, inflammation, liver disorders, 

and obesity. This plant contains organic and phenolic acids, anthocyanins, and flavonols as main compounds. One of these major 

compounds found in this plant is quercetin, a flavonol commonly present in fruits and vegetables. 

In recent years, this plant has been studied in depth because it exerts a great number of pharmacological activities. Previous studies 

showed that HS possesses anti‐tumor and antioxidant properties [1]. In humans, HS elicits a significant decrease in MCP‐1 plasma 

concentration, suggesting that HS may be a valuable traditional herbal medicine for the treatment of chronic inflammatory diseases 

[2]. Recent data also indicate that HS aqueous extracts might ameliorate metabolic disturbances [3]. Furthermore, the antioxidant 

and antihyperlipemic properties of HS aqueous extract have been previously reported by our research group using a mouse model 

[4]. Moreover, the strong antioxidative and hypolipidemic properties of a HS polyphenolic extract have been recently reported in a 

cell model for adipogenesis and insulin resistance [5]. In this studies, we suggested that flavonol conjugated forms (such as 

quercetin) could be the responsible compounds for the observed antioxidant effects and also contribute to the salutary effects of 

HS. In this sense, a study of the molecular and cellular targets of these compounds in different cellular models deserves further 

attention. For this reason, a pharmacokinetic study of quercetin in a model of adipogenesis (3T3‐L1 cells) has been carried out. 

100 μl Cytoplasm 

2 

500 μl MeOH‐EtOH 

(50:50) 

2 hours 

Supernatant Dryness 

In this study 3T3‐L1 pre‐ 

adipocytes were propagated 

and differentiated according 

to conventional procedures. 

Quercetin was added at 100 

µM to the media at the 

beginning of the study. 

Subsequently, samples of 

cytoplasm were taken at 

different time (3, 6, 12 and 

18 hours) after quercetin 

addition. 

Centrifuge 

18800 g 

5 min 

50 μl MeOH 

HPLC‐DAD‐ESI‐TOF 

analysis 

3 

HPLC 

BPC OBTAINED 

BY HPLC‐ESI‐TOF 

• Instrument: Agilent 1200 RRLC 

• Column: Zorbax Eclipse Plus C18, 

4.6x150 mm, 1.8 μm 

• Flow: 0.3 mL/min 

• Temperature: 25 °C 

• Injection volume: 10 μl 

• Phase A: H 2O/AcN 90:10 + 0.5% Formic acid 

• Phase B: AcN 

• Gradient: : 

• from 0 to 20 min, from 5% B to 20% B, 

• from 20 to 25 min, from 20%B to 40%B, 

• from 25 to 40 min, from 40% B to 5% B, 

• from 40 to 45 min, to hold 5% B. 

[1] Farombi EA, Fakoya A (2005) Free radical scavenging and antigenotoxic activities of natural phenolic compounds in dried flowers of Hibiscus sabdariffa L. Molecular Nutrition & Food Research 49:1120 

[2] Beltran‐Debon R, Alonso‐Villaverde C, Aragones G et al (2010) The aqueous extract of Hibiscus sabdariffa calices modulates the production of monocyte chemoattractant protein‐1 in humans. Phytomedicine 

17:186 

[3] Carvajal‐Zarrabal O, Waliszewski SM, Barradas‐Dermitz DMA et al (2005) The consumption of Hibiscus sabdariffa dried calyx ethanolic extract reduced lipid profile in rats. Plant Foods for Human Nutrition 60:153 

[4] Fernández‐Arroyo S, Rodríguez‐Medina IC, Beltrán‐Debón R et al (2011) Quantification of the polyphenolic fraction and in vitro antioxidant and in vivo anti‐hyperlipemic activities of Hibiscus sabdariffa aqueous 

extract. Food Res Int 44:1490 

[5] Herranz‐López M, Fernández‐Arroyo S, Pérez‐Sánchez A et al. Synergism of plant‐derived polyphenols in adipogenesis: perspectives and implications. Phytomedicine 

ESI‐TOF 

Quercetin‐3‐O‐Glucuronide 

• Instrument: Bruker Daltonics 

micrOTOF 

• Mode: Negative 

• Mass range: 50‐1000 m/z 

• Nebulizing gas pressure: 2 bar 

• Drying gas flow: 9 L/min 

• Drying gas temp: 200 ºC 

• Calibrant: sodium formate 5mM 

• Capillary Exit: ‐120 V 

• Skimmer 1: ‐40 V 

• Hexapole 1: ‐23 V 

• Hexapole RF: 100 Vpp 

• Skimmer 2: ‐22,5 V 

• Transfer time: 50 μs 

• PrePulseStorage time: 3 μs 

Quercetin 

Concentration (ppm) 

45 

40 

35 

30 

25 

20 

15 

10 

5 

0 

0 0,5 31 1,5 62 2,5 123 3,5 184 

4,5 

Time after incubation (h) 

Quercetin 

Quercetin was detected in all analyzed 

samples, except in the control. From the 

obtained data, the maximum concentration 

found in cytoplasm was at 3 hour after the 

quercetin addition. Nevertheless, it is 

important to note that at 6 hours time point 

the concentration of quercetin is still high. At 

the end of the assay, 18 hours after the 

addition, the level of quercetin is minimum. In 

all samples except in the control, quercetin‐3‐ 

O‐glucuronide has been detected. 

The authors are grateful to the Spanish Ministry of 

Science and Innovation for the project AGL2011‐ 

29857‐C03‐02, Andalusian Regional Government 

Council of Innovation and Science for the excellence 

project P10‐FQM‐6563 and P11‐CTS‐7625 and to 

GREIB.PT.2011.18 project. IBL acknowledges financial 

support from the Spanish Ministry of Education and 

Science (FPI grant, BES‐2009‐028128). 

-5-

-6- 



Bruker Daltonics 

Metabolomic consequences of aronia juice intake by 

human volunteers 

Medina, S. 1 ; García-Viguera, C. 1 ; Ferreres, F. 1 ; Savirón, M. 2 ; Orduna, 

J. 2 ; Zurek, G. 3 ; Gil-Izquierdo, A. 1* 

1Research Group on Quality, Safety and Bioactivity of Plant Foods, CEBAS- 

CSIC, Murcia, Spain 

2Instituto de Ciencias de Materiales de Aragón. CSIC- Universidad de 

Zaragoza, Zaragoza, Spain 

3Bruker Daltonik GmbH, 28359 Bremen, Germany. 

*corresponding author: angelgil@cebas.csic.es 

Introduction 

Metabolomics is a powerful technique which focuses in highthroughput characterization of 

metabolome (the complete set of low-MW metabolites in biological samples) in order to obtain a 

molecular profile. This approach has shown a considerable potential in the field of nutrition, so, there 

is an increasing interest in the use of metabolomics as a tool for the understanding of the interaction 

between diet and metabolome and to be a promising technique for biomarker discovery related to 

food intake [1]. 

Our study was conducted with the aronia fruit (Aronia melanocarpa) common name black 

chokeberry which belongs to the Rosaceae family. This berry stands out in high amount of flavanols, 

quercetin (71.13 mg Kg-1; 93.07%) and kaempferol (5.3 mg Kg-1; 6.9%). Aronia berries are one of 

richest plant sources of anthocyanins, mainly containing cyanidin glycosides [2]. 

In a recent study, anthocyanins showed significant potency of antiobesity and ameliorate adipocyte 

function in vitro and in vivo systems, as well as, antioxidant, hypoglycemic and hypolipidomic 

activities and important implications for preventing metabolic syndrome of a chokeberry juice, indicate 

their contribution to the potential health benefits [3,4]. 

The aim of this study was to observe the single and five days changes in the metabolic profile of 

healthy volunteers following the consumption of aronia juice and to identify the key metabolites. 

Urine samples 

(treatment and control) 

Assay design 

Number of patients: 8 (4 men and 4 women) 

Age: 28-47 

Race: Caucasian 

BMI means: 24.5 Kg/m 2 

Wash out 

2 days 

Day 1 Day 2 Day 3 Day 4 

200 mL 8:00 a.m 

200 mL 18:00 p.m 

Diet absent of fruit-vegetables and food derived products 

200 mL 8:00 a.m 

200 mL 18:00 p.m 

• Workflow for metabolomic analysis: 

Data adquisition 

LC-MS-q-TOF (Bruker Daltonics, 

Bremen, Germany) 

Column ACE 3 C18: 150 x 0.075 

mm 

Fase A: H2O/0.1% HCOOH 

Fase B: ACN/0.1% HCOOH 

Flow rate: 312.50 nL/min 

Injection volume: 6.25 nL 

Positive and negative mode 

Full scan: 50-900 m/z 

200 mL 8:00 a.m 

200 mL 18:00 p.m 

24 h urine collection and frozen at -80º C 

Data preprocessing 

Profileanalysis 2.0. (Bruker 

Daltonics, Bremen, Germany) 

Software for profiling and 

statistical evaluation of LC/MS 

data set. 

200 mL 8:00 a.m 

200 mL 18:00 p.m 

Multivariate statistical 

analysis 

PCA (Principal Component 

Analysis) with Pareto scaling 

algorith and Student´s t-test (P-value 


COMPREHENSIVE WORKFLOW FOR METABOLIC 

PROFILING AND IDENTIFICATION OF SECONDARY 

METABOLITES FROM MYXOBACTERIA 

Thomas Hoffmann 1,2 , Daniel Krug 1, 2 , Aiko 

Barsch 3 , Gabriela Zurek 3 , Rolf Müller 1,2 

1 Pharmaceutical Biotechnology, Saarland 

University, Saarbruecken, Germany 

2 Helmholtz-Institute for 

Pharmaceutical Research Saarland (HIPS), 

Saarbruecken, Germany 

3 Bruker Daltonik GmbH, Bremen, Germany 

Introduction 

Extracting the relevant information from complex 

data sets is an important bottleneck in (microbial) 

metabolomics research. Compared to a focused 

targeted approach, this objective becomes even 

more challening in untargeted metabolomics, aiming 

for the identification of all compounds produced by a 

particular bacterial strain. Besides the identification 

of all known compounds it is important to assign the 

subset of interesting compounds that are “worth” 

further examination. 

Here, we present the combination of targeted and 

untargeted metabolomics utilizing statistical methods 

of data mining. ESI-UHR-Q-TOF based analysis of 

myxobacterial extracts allows for fast generation of 

high resolution MS as well as MS² spectra which 

support our screening workflow for the discovery of 

novel secondary metabolites. 

6 x extract 

6 x blank 

Fig. 1: Overall 24 runs were analyzed. The feature 

finding algorithm assigned ~ 3500 features per run. 

Sample Set 

• The sample set consisted of 6 biological replicates 

of a Sorangium cellulosum strain and 6 replicates of 

growth medium blank. Extracts were prepared by 

methanol extraction of an adsorber resin which was 

added to the cultivation broth (Amberlite XAD-16). 

Samples were diluted 5x with methanol prior to 

injection. For each sample two technical replicates 

were measured. 

Methods 

• Sorangium cellulosum extracts were separated using a Waters 

BEH C18 column (100 x 2.1 mm, 1.7 µm particle size) on an 

UltiMate 3000TM RSLC system (Dionex). Gradient elution was 

performed at 600 µl/min and 45 °C using H2O + 0.1 % FA (A) and 

ACN + 0.1 % FA (B). The gradient started at 5 % B for 0.5 min 

before increasing to 95 % B in another 19 min. 

• ESI-MS measurements were performed using positive ionization 

on a maXis UHR-Q-TOF-MS (Bruker Daltonik) m/z range: 150- 

1500, repetition rate: 2 Hz for MS, 3 Hz for MS². 

• Targeted screening was carried out using precise extracted ion 

chromatograms (EICs), retention time and isotope pattern 

evaluation via SigmaFitTM with TargetAnalysisTM (Bruker Daltonik). 

• Pre-processing of data and statistical interpretation using t-test 

was performed with ProfileAnalysisTM 2.0 (Bruker Daltonik). 

Scheduled precursor lists (SPL) for targeted MS/MS experiments 

were created by the same software tool. SPL derived MS² data 

was processed with MZmine 2.2 [1] . Features were extracted from 

raw data using the FindMolecularFeatures (FMF) algorithm in 

DataAnalysisTM 4.1. 

Feature Finding 

• Myxobacterial extracts represent a complex 

mixture of small molecules derived from the medium 

and bacterial secondary metabolites covering several 

magnitudes in dynamic range. The feature finding 

algorithm (FMF) assigned approximately 3500 

features for each measurement (Fig. 1). 

• Overall 24 samples gave ~80,000 features. 

• The FMF algorithm is able to automatically detect 

and combine adduct ions and common neutral losses 

like [M+NH4] + , [M+Na] + , or [2M+H] + to one feature. 

This approach lowered the total number of features 

by around 10 % to ~70,000 features. 

For research use only. Not for use in diagnostic procedures. 

80,000 features 

(24 x 3500) 

Untargeted Analysis 

• The result was evaluated by a t-test comparing 

extracts and blanks and value-count filtering for 

features which were present in at least 11 out of 12 

measurements in the extract. 

• A resulting subset of 176 features was exported as 

a scheduled precursor list (SPL) for directed 

precursor selection in a subsequent MS/MS 

experiment (Fig. 2). 

Fig. 2: Features that are only present within bacterial 

extracts are assigned by means of a t-test. An SPL of 

176 features was exported automatically for further 

CID analysis. 

A 

# rt [min] m/z Extract Blank 

1 1.1 383.2 12 0 

2 2.2 298.1 12 0 

3 2.3 293.1 11 0 

4 2.4 180.1 12 0 

… … … … … 

176 17.9 253.25 11 0 

Name Formula [M+X] + m/z theo. m/z meas. Error [ppm] mSigma 

Ambruticin A 

[M+H] + 

B 

C28H40O6 473.28977 473.28964 0.3 8.3 

Ambruticin A 

[M+NH + 

4] 

C28H43N1O6 490.31632 490.31572 1.2 4.5 

Ambruticin A 

[M-H + 

2O+H] 

C28H38O5 455.27920 455.27909 0.2 23.6 

Fig. 3: A: Result of target compound identification as 

presented in TargetAnalysis. B: Example: Assignment 

of several adducts for the same compound gives more 

reliable hits. 

References 

t-test result 

Targeted Analysis 

Important features 

assigned by t-test analysis 

Targeted search against 

in-house database 

• An in-house database (Myxobase) covering ~700 

myxobacterial secondary metabolites representing 

more than 100 distinct compound classes with 

retention time and sum formula information was 

used to identify the known compounds (Fig. 3A). 

• The 176 filtered features obtained by the 

untargeted workflow were then compared to the 

TargetAnalysis result to assign known metabolites. 

22 features are related to TA hits or fragments 

thereof. 

• To increase the confidence in compound 

identification by targeted analysis common adduct 

information was additionally taken into account. The 

Myxobase system enables the export of a “screening 

file” for TargetAnalysis which automatically creates 

common adduct ions for a given sum formula 

(Fig. 3B). 

[1] T. Pluskal, S. Castillo, A. Villar-Briones, M. Orešič, 

BMC Bioinformatics 11:395 (2010). 

Distribution of 176 SPL features 

Funding from BMB+F, DFG and DAAD is gratefully acknowledged. 

176 features 

(22 assigned) 

MS/MS 

Unknown identification 

Conclusions 

Bruker Daltonics & HIPS 

Parent ion selection for MS² 

• Automatic precursor selection during acquisition 

routinely allows for the fragmentation of dominant species. 

However, compounds of interest may be low abundant and 

coeluting with matrix compounds from cultivation media. 

In such cases, the preselection of parent ions can increase 

the number of MS² spectra for compounds of interest. 

• Manual definition of hundreds of precursors is time 

consuming. Therefore, the subset of 176 features was used 

for directed precursor selection via an SPL (Fig. 2). 

• 37 % of the precursors of interest were not 

fragmented previously during an auto-MS² run. Using SPL 

lists generated based on t-test results could improve the 

MS² coverage to 100 % for the target compounds. 

Putative ambruticin derivatives: 

Fragmentation proposal: 

Intens. 

2000 

1500 

1000 

500 

0 

1+ 

224.1272 

1+ 

250.1434 

O 

OH 

O N 

H 

High accurate m/z of CID 

peaks allow for structural 

identification 

271.2439 288.1587 

1+ 

Peak 10 

C18H26NO3, [I+H], -2.1 ppm 

1+ 

304.1914 

1+ 

333.2231 352.2979 

• The number of features representing potentially novel 

metabolites was reduced further by matching specific 

neutral losses within coeluting features. 

Features in one MS file 3500 

Features after t-test 176 

- fragmented with SPL method 176 

- fragmented with auto-MS² 110 

Related to known compounds - 22 

Assigned by online database search - 0 

Identified as neutral loss - 26 

Remaining unknown features 128 

Assignment using MS/MS data 

Peptides - 4 

found due to similar fragmentation - 11 

- related to 3 Thuggacins 

- related to 8 Ambruticins 

New features for further investigation 113 

• High resolution accurate mass LC-MS data sets allow 

for both targeted and untargeted analysis. 

• Complexity is reduced by statistical data evaluation. 

• Parent ion selection is significantly improved by a 

scheduled precursor list (SPL) 

• 63 out of 176 filtered features from myxobacterial 

extracts could be identified based on MS 2 data, thereby 

effectively supporting de-replication during screening 

NH2 

C25H40NO, [I+H], +0.7 ppm 

1+ 

370.3102 

O 

C22H42N3O3, [I+H], -3.9 ppm 

1+ 

396.3236 

1+ 

423.3188 

381.2747 

+MS2(Ambruticin new01), 29.4eV, 11.87min #1401 

O 

O 

OH 

NH2 

1+ 

440.3134 

1+ 

458.3263 

1+ 

476.3357 

250 300 350 400 450 m/z 

Fig. 4: Unknown identification: Calculation of sum 

formulae suggests the presence of new ambruticin 

derivatives. Subsequent MS² spectra evaluation using 

Fragment Explorer allows for the annotation of highly 

accurate MS² peaks, supporting the identification. 

• Guided MS/MS 

acquisition allows for 

the assignment of 

interesting features 

and facilitates their 

identification (Fig. 4). 

• Searching in 

publicly accessible 

databases did not 

yield any hits. 

• 113 features remained 

for further 

evaluation by MS and 

NMR experiments. 

ESI-UHR-QTOF-MS 

C24H38N7O, [I+H], -0.4ppm 

C28H46NO5, [I+H], +2.8 ppm 

O 

-7-

-8- 



The impact of high-resolution MS and NMR 

techniques on the discovery of novel natural 

products from myxobacteria 

Daniel Krug 1,2 , Thomas Hoffmann 1 , Kirsten 

Harmrolfs 2 , Alberto Plaza 1 , Aiko Barsch 3 , 

Gabriela Zurek 3 , Rolf Müller 1,2 

1 Helmholtz-Institute for Pharmaceutical Research 

Saarland (HIPS), 66123 Saarbrücken/Germany 

2 Saarland University, 66123 Saarbrücken/Germany 

3 Bruker Daltonik GmbH, 28359 Bremen, Germany 

Introduction 

Microbial natural products represent a rich and still 

largely untapped resource for novel chemical 

scaffolds. Their discovery enables characterization of 

the underlying biosynthetic pathways and at the 

same time chances are good to find compounds with 

potent biological activity. UHPLC-coupled UHR-TOF 

mass spectrometry and LC-NMR represent most 

valuable analytical tools for the discovery of novel 

natural products, efficient dereplication, comprehensive 

mining of metabolomes and rapid 

structure elucidation of novel metabolites. 

Chondromyces crocatus Sorangium cellulosum 

A 

B 

Name Formula [M+X] + m/z theo. m/z meas. Error [ppm] mSigma 

Ambruticin A 

[M+H] + 

Ambruticin A 

[M+NH + 

4] 

Ambruticin A 

[M-H + 

2O+H] 

C 28H43O6 

Dm/z = 0.9 ppm 

14.8 mSigma 

C 28H41O5 


6.8 mSigma 

C 28H39O4 


8.9 mSigma 

1+ 

439.2847 

1+ 

457.2953 

1+ 

475.3059 

493.3160 

510.3428 

360 380 400 420 440 460 480 500 520 m/z 

+MS, 11.20min 

Fig. 1 Chromatogram: 

separation of a crude extract 

from a Sorangium strain. Right 

inset: comparison of measured 

and calculated spectra for 

Ambruticin, a myxobacterial 

compound that exhibits potent 

antifungal activity. 

475.3054 

5 475.3059 

x10 

1.0 

0.8 

Ambruticin S 

C28H42O6 [M+H] 

0.6 

476.3088 

0.4 

476.3091 

0.2 

477.3117 

calculated 

477.3111 

measured 

0.0 

475 476 477 478 

+ = 475.3054 


14.8 mSigma 

C 28H40O6 473.28977 473.28964 0.3 8.3 

C 28H43N1O6 490.31632 490.31572 1.2 4.5 

C 28H38O5 455.27920 455.27909 0.2 23.6 

Fig. 2: A: Result of target compound identification as presented in 

TargetAnalysis TM . B: Example: Assignment of several adducts for 

the same compound gives more reliable hits. 

Methods 

• Myxobacteria were cultivated in complex media and 

extracts were separated by RP chromatography 

using an U-HPLC system. MS measurements were 

performed in ESI positive mode (scan range 100- 

2000 m/z) using an ultra high resolution Q-TOF 

mass spectrometer (Bruker maXis 4G). 

• A specialized database system – MYXOBASE - was 

developed to support targeted screening. Identification 

of known compounds was accomplished 

using precise EICs and considering exact mass, 

retention time and isotope pattern for increased 

confidence (Bruker TargetAnalysis 1.3 software). 

• In addition, feature-extracted HR-MS data were 

used for untargeted profiling by PCA (Principle 

Component Analysis) using the Bruker Profile 

Analysis 2.1 software. 

• LC-SPE-NMR analysis was performed using a VWR 

2400 series solvent delivery system, with a 

Phenomenex Luna C-18 150 x 4.6 mm (2.5 μm) 

column, SPE unit and Bruker Biospin Avance 700 

NMR platform with cryoprobe. 


LCMS profiling 

& dereplication 

M. xanthus 

wildtype 

5 10 15 

Scores & Loadings plot 

wt mutant 

1 

3 

2 

4 

BPC 

(500-600 m/z) 

506.2708 m/z 

2.97 min 

4 x 

Samples 

2 3 4 5 min 

#3 

#3 

1 H-NMR 

p3919gp, TD: 360, NS: 16 

5 

x10 

2.0 

1.5 

1.0 

Targeted Profiling 

• Full scan spectra enable the targeted screening 

for known myxobacterial compounds based on 

highly selective EIC traces, using MYXOBASE as 

analyte database (Fig. 1, 2). 

•Profiling reports are parsed into MYXOBASE with all 

relevant analytical performance qualifiers, enabling 

collaborative evaluation of results and crosslinking 

with bioactivity assays and strain-related data. 

• Our screening campaign enables the in-depth 

comparison of high numbers of complex myxobacterial 

metaboilte profiles and the identification 

of alternative producers across all myxobacterial 

strains from our world-wide collection. 

Genome- 

& metabolomemining 

2 

4 

3 

1 

M. xanthus 

mutant 

PCA 

Natural product 

discovery 

5 10 15 

4 x 

506.271 m/z 

2.97 min 

Molecular Features 

2+ 

506.2708 

Identification 

& structure 

elucidation 

Myxochelin A New Myxochelin derivative 

405.167 

+MS, 7.7min 

#457 

0.5 

269.2 

212.1 

432.2 

374.3 

387.2 

200 300 400 500 

x10 

5 

1.50 

1.25 

1.00 

0.75 

0.50 

1 H-NMR 

p3919gp, TD: 360, NS: 16 

1+ 

652.4018 

1+ 

652.4039 

+MS, 2.97min #483 

500 600 700 m/z 

419.182 

0.25 

283.1 

334.7 

402.2 472.2 

205.1 223.1 267.2 343.7 

0.00 200 300 400 500 

+MS, 8.6min 

#512 

UV, l= 220nm 

Myxobase 

MS-guided collection of 

LC-SPE fractions 

+MS2(506.271), 25-30 eV 

Intens. 

8000 

6000 

4000 

2000 

Myxochelin A 

C 20H24N2O7 

[M+H] + 405.1656 m/z 

0 

Intens. 

x10 4 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 

1+ 

215.1403 

N-MeSer-Leu + H + 

D = 1.1 ppm 

2+ 

506.2708 

2+ 

506.7725 

+MS, 2.97min #483 

2+ 

507.2738 

2+ 

507.7743 

2+ 

506.2711 

506 507 508 m/z 

References 

Bruker Daltonics & HIPS 

Fig. 3 Statistical treatment of extracts derived from wild-type and geneknockout 

strains of Mycococcus xanthus DK1622. Principle component 

analysis (PCA, plot shows PC1 vs. PC2) facilitates the discovery of a novel 

compound class, the myxoprincomides [2], by using a combined genome- 

and metabolome-mining strategy (left). Full scan and MS2 spectra for 

myxoprincomide-c506 (right). 

- Myxobacteria strain collection 

- Chemical compounds database 

- Analytical observations library 

- Screening results archive 

- Metabolome data repository 

- MS2 reference spectra collection 

M+H + - N-MeSer-Leu 

D = 0.7 ppm 

Myxoprincomide c506 

C 45H 74O 16N 10 

[M+2H] 2+ 506.2714 m/z 

Summary 

1+ 

797.4045 

M+H + - N-MeSer 

D = 1.2 ppm 

1+ 

910.4891 

200 300 400 500 600 700 800 900 m/z 

HSQC 

hsqcetgpprsisp2.2 , TD: 128, NS: 16 

Fig. 4: Rapid structure elucidation of metabolites from myxobacterial crude extracts enabled by LC- 

SPE-NMR, following semi-preparative liquid chromatography coupled to MS- and UV/Vis detection. 

The NMR spectrum (right path) corresponds to a new derivative of Myxochelin. 

Metabolome Mining 

• The number of known compounds identified to 

date from many myxobacterial strains is often 

significantly lower than expected from genome 

sequence information [1, 2]. 

• The comparison of feature-extracted secondary 

metabolomes using statistical tools thus has the 

potential to uncover previously "hidden" natural 

products (Fig. 3). 

• This comprehensive genome- and metabolomemining 

approach is exemplified by the recent 

discovery of myxoprincomides, an intriguing family 

of novel secondary metabolites from Myxococcus 

xanthus [2, 3]. 

Myxochelin 419 

C 20H22N2O8 

[M+H] + 419.1454 m/z 


• Accurate-mass MS 2 fragmentation 

is used for the identification of 

compounds differentiating bacterial 

strains, for evaluation of chemical 

novelty and as starting point for 

structure elucidation. 

• LC-SPE-NMR is an important key 

for the rapid identification of known 

and novel compound classes due to 

its ability to generate a partial or 

even complete structure based on 

NMR information early in the course 

of the discovery workflow (Fig. 4). 

[1] “The biosynthetic potential of myxobacteria and their impact 

on drug discovery”, Curr. Opin Drug Disc Dev., 2009, 12(2), 

p. 220-230. 

[2] “Myxoprincomide; a natural product from Myxococcus xanthus 

discovered by comprehensive analysis of the secondary meta 

bolome”, Angew. Chem. Int. Ed., 2012, 51(3), p. 811-816. 

[3] “Discovering the hidden secondary metabolome of Myxococcus 

xanthus: a study of intraspecific diversity”, Appl. Environ. 

Microbiol., 2008, 74, p. 3058-3068. 

• LC-UHR-QTOF-MS and databaseassisted 

strategies enhance the natural 

products identification workflow 

• Metabolome-mining using statistical 

tools enable novel metabolite discovery 

• LC-SPE-NMR/MS for rapid dereplication 

and structure elucidation 

ESI-UHR-QTOF-MS 

& LC-SPE-NMR


Daniel Krug 1,2 , Niña S. Cortina 1,2 , Ole Revermann 2 , Rolf Müller 1,2 

1 Helmholtz Institute for Pharmaceutical Research Saarland, Saarbrücken, and Helmholtz Center 

for Infection Research, Braunschweig, Germany 

2 Saarland University, Dept. of Pharmaceutical Biotechnology, Saarbrücken, Germany 

Discovery of Novel Myxobacterial Secondary Metabolites 

by Mass Spetrometry-based Metabolome Mining 

Discovery of myxobacterial natural products supported by Myxobase 

Myxoprincomide biosynthesis: a molecular assembly line “under construction” 

INTRODUCTION 

M. xanthus DK1622 M. xanthus DK897 

3 mDa 

mxp 1622 gene (MXAN_3779, 42.8 kbp) mxp 897 gene (46.3 kbp) 

Myxobacteria represent an important source of biologically active 

natural products with considerable promise for human therapy. 

While the enormous genetic potential of many myxobacterial species 

for secondary metabolite biosynthesis is well recognized, [1] 

resolution mass spectrometry and statistical data evaluation, can 

significantly facilitate the process of uncovering these "hidden" 

secondary metabolites. 

the number of compound classes reported from individual strains 

clearly falls short of the genetic capabilities (A). Improved analytical 

methods, based on the combined use of LC-coupled high- 

[2-4] This metabolome-mining approach 

is exemplified here by the assignment of three novel compound 

classes produced in low quantities to NRPS/PKS gene clusters in 

the genome of Myxococcus xanthus DK1622 (B), including the intriguing 

myxoprincomides. [2] 

m/z 

C 

D 

A B 

497 s 

c506 

A B 

3.0 

506.2712 

506.7727 

507.2742 

1.5 506 508 

E 

MXAN_3779 

I x 10 6 

m/z 

DKxanthene 

[M+H] + = 416.316 m/z 

@ Ret. = 497 s 

Ret. 

HZI collection: 

~9000 myxobacteria 

c382 

c534 

c649 

c683 

c798 

O 

O 

H 

H 

H 

N 

N 

N 

N 

N 

Myxoprincomide 


H 

H 

O 

HO 

c506 

c580 

OH 

(1010 Da) 

(1158 Da) 

HO 

1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 12 

predicted: Ser Phe Val Ser Cys mCoA Ser Phe β-Lys Ile Ser predicted: Ser Orn Val Val Ser Cys mCoA Ser Phe β-Lys Ile Ser 

observed: Ser Leu β-OHVal Ser Val mCoA Ser Tyr β-Lys - - - Ala observed: Ser β-OHVal Phe β-OHVal Ser Val mCoA Ser Tyr β-Lys - - - Ala 

or Met 

or Leu 

or Leu 

ACL ACP C A N-MTA PCP C A PCP C A PCP C A PCP C A PCP KS AT ACP C Ox C A PCP C A PCP C A PCP C A PCP C A PCP TE 

ACL ACP C A N-MTA PCP C A PCP C A PCP A PCP C A PCP C A PCP KS AT ACP 

C A PCP C A PCP C A PCP C A PCP C A PCP C 

C Ox 

TE 

S S S S 

S 

S 

S 

S 

S 

S 

S S S 

S S 

S 

S 

O O 

O O 

O 

S 

S 

S 

S 

O 

O O H2N O 

O 

O O 

O O 

HO O O 

O 

O 

O O NH O 

2 

HO 

OH 

OH 

HO NH 

HN 

NH 

HN OH 

NH 

O 

? 

HO 

HO NH 

HN 

NH 

HO NH 

HN 

NH 

HN 

NH 

HN 

O 

O 

O 

O 

O 

? 

? HN OH 

NH 

? 

HN 

O 

O 

HO 

OH NH 

H2N O 

O 

O 

O 

O 

HO O 

NH 

O HO O 

HN 

O NH2 O 

O 

OH 

OH 

HN 

HO NH 

HN 

NH 

HN OH 

O 

O 

HO 

O 

HO NH 

HO NH 

HN 

NH 

HN 

NH 

O O 

O 

HO O 

HN OH 

NH 

O 

HN 

NH 

O O 

HO 

O O 

HN OH 

HN 

O 

NH 

HN 

O 

O 

O 

O 

HO NH 

HN 

NH 

O 

OH 

OH 

OH NH 

HO O 

O 

O 

O 

HO NH 

HN 

O 

O 

NH 

HN 

O 

HO 

HN OH 

NH HO NH 

HN 

O 

NH 

O 

O 

OH 

NH 

HN 

HN OH 

NH 

O O 

O 

O 

HO NH 

HN 

OH 

O 

OH 

HN 

O 

HO O 

O 

O 

HO 

O 

O 

HO NH 

HN 

NH 

O 

OH 

O 

NH 

HN OH 

HO NH 

HN 

O O 

HO NH 

O 

O 

NH 

O 

HN 

NH 

HN OH 

OH 

O 

NH 

HO 

O 

O 

O 

O 

HN 

O 

HO NH 

HN 

O 

OH HO O 

O 

HN 

NH 

O 

HO NH 

O O 

HN OH 

NH 

OH 

NH 

HN 

O 

O 

HN 

O 

O 

NH 

HN 

HO NH 

HO NH 

HN 

O 

HO 

OH 

O 

HO O 

O 

NH 

HN OH 

HN 

NH 

OH 

O 

O 

HO NH 

O 

O 

HN 

NH 

OH 

HO NH 

HO 

O 

O 

HN 

NH 

HO NH 

HN 

OH 

O 

O 

O 

NH 

HN 

O 

O 

HO NH 

HN 

OH 

O 

HN 

NH 

O 

O 

HO NH 

OH 

HO NH 

HN 

O 

HO NH 

OH 

NH2 OH 

OH 

HO 

O O 

O 

O 

O 

O O 

O O 

H 

H H 

H 

H 

H 

H 

N 

N N 

N 

N 

N 

N 

OH 

N 

OH 

N 

N 

N N 

N N 

H 

H 

H 

H H 

H H 

O O 

O O 

O 

O 

O O 

O 

OH 

HO 

OH 

NH OH 

2 

0.0 

c844 

3.0 

O 

O 

N 

N 

O 

H 

HN 

O O 

NH2 

HN 

N 

O O 

H 

O 

N 

? 

O 

? 

Myxochromide 

O 

O 

O 

? ? 

HN 

O 

Myxovirescin 

OMe 

OH 

? OH 

? 

? 

OH 

Myxalamid 

HO 

NH 

? 

O 

c844 

? 

OH 

Myxochelin 

HN NH 

HO 

O O 

OH HO 

OH 

HO 

M. xanthus DK1622 

Genome ~ 9 Mbp 

A B 

844.3742 

845.3770 

846.3810 

MXAN_1608 - MXAN_1598 

5.0 

4.0 

3.0 

BPC 

EIC 535.26 m/z 

EIC 506.27 m/z 

2.0 

WT 

MXAN_3779- 

EIC 844.37 m/z 

WT 

1.0 

MXAN_1607- 

EIC 329.19 m/z 

WT 

0.0 

MXAN_3631- 

0 2 4 6 8 10 

t / min 

I x 10 8 

Myxalamid A 

E 

844 846 

1.5 

I x 10 

0.0 

6 

Strains and bioactivity 

profiles 

C 

LCMS library 

dereplication 

3.0 

D 

c329 

329.1861 

330.1892 c329 

331.1893 

MXAN_3633 - MXAN_3628 

329 331 

1.5 

I x 10 

0.0 

6 

c506 

O 

N 


COOH 

NH2 

? 

DKxanthene 

H 

N 

O 

HO 

H 

N 

0 2 4 6 

t / min 

O 

Myxobase 

506.271 m/z @ 2.9 min 

Genomemining 

1 2 3 4 

5 6 

Metabolome 

mining 

? 

? 

? 

? 

? 

.CGCGATCGAGATC 

CGCTAGATCCGTGG 

TGCGCGATAGCGA. 

Discovery of myxoprincomide by comprehensive analysis of the Myxococcus 

xanthus DK1622 secondary metabolome 

Starter unit 

ACP C A N-MTA PCP C A PCP C A PCP C A PCP KS AT ACP 

C A PCP C A PCP C A PCP C A PCP C A PCP C A PCP C Ox 

TE 

CL A 

Module 2 & 3 Module 6 Module 8 & 9 Module 10 Termination 

? ? 

Scores Loadings 

Mxp 1622 protein (~1.6 MDa) 

1. 

Mutant 

506.271 

[M+2H] 

Ret. = 2.97 min 

2+ 506.773 

507.274 

m/z 

= 506.271 min 

WT 

ACL ACP C A N-MTA PCP C A PCP C A PCP C A 

PCP C A PCP C 

8 9 

4 

• modules 8 and 9 in Mxp in- 1622 

corporate Tyr and β-Lys, respectively 

PCP C 

A N-MTA PCP C A PCP C A PCP C A 

ACL ACP C 

1 

• no myxoprincomides with 

acyl starter unit observed 

WT 

4 x 

M. xanthus 

DK1622 

O 

H 

N 

O 

N 

H 

HO 

H 

N 

O 

H 

N 

PCA 

N 

H 

506.2712 

? 

O 

O 

OH 

HO 

PCP C A PCP C A 

S 

O O NH2 HO 

HN 

O 

NH 

βLys 

Tyr 

Ser 

Val* 

Ser 

OHVal 

Leu 

S 

NH 

βLys 

Tyr 

Ser 

Val* 

Ser 

OHVal 

Leu 

NMe 

-Ser 

C A 

8 9 

C A PCP C A PCP C A 

S 

S 

O O NH2 HO 

NH 

HO 

HN 

Ser 

O 

NH 

Val* 

Ser 

Ser 

Val* 

OHVal 

Ser 

Leu 

OHVal 

Leu 

PC P 

SH 

A 

HO 

tn5 

• however, promotor insertion 

upstream of module 1 results 

in loss of production 

506.7727 

507.2742 

1.2 

0.8 

Mutant 

NMe 

-Ser 

38 x 

• these modules are sometimes 

used iteratively as a „mixed 

double“, a to date undescribed 

feature in NRPS biosynthesis 

A N-MTA PCP C A PCP C A PCP C A 

506 508 

m/z 

Molecular feature 

506.271 m/z at 3.0 min 

WT 

NMe 

-Ser 

PCP C 

0.4 

I x 10 6 

4 x 

? 

Mutant 

0.0 

600 

NMe 

-Ser 

Reference: DK1622 

strain A2 

500 

% 

400 

300 

2 

c382 (381 Da) NMeSer – OHVal – Phe 

• insertion of OHVal-activating 

A-domain 2 in Mxp and c683 (682 Da) NMeSer – OHVal – Phe – OHVal – Ser – Val 

897 

specificity change to Phe in 

O 

module 3 

c812 NMeSer – OHVal – Phe – OHVal – Ser – Leu 

Ser 

(811 Da) 

O 

• A-domain promiscuity in 

O 

Mxp module 2 (Leu, Met 

1622 c798 NMeSer – OHVal – Phe – OHVal – Ser – Val 

Ser 

& Mxp module 6 (Val, Leu) 

897 (797 Da) 

O 

O 

c515 (1028 Da) NMeSer – Met – OHVal – Ser – Val 

Ser –Tyr – βLys – Ala 

O 

2.6 2.8 3.0 3.2 3.4 

t / min 

H2N O 

OH 

• High-resolution LC-MS measurements of 

samples from knockout mutant strains (replicates) 

and Principal Component Analysis (PCA). 

O 

H 

N 

O 

H 

N 

H 

N 

O 

H 

N 

O 

H 

N 

O 

N 

H 

HO 

H 

N 

O 

H 

N 

c708 (1414 Da) 

OH 

N 

H 

N 

H 

N 

H 

N 

H 

O 

O 

O 

O 

O 

O 

OH 

OH 

HO 

MIC (M. hiemalis): 60 µg 

OH 

H 2 N 

• Relative production of myxoprincomide “c506” as determined 

from the secondary metabolomes of 98 M. xanthus 

strains, using data from HR LC-MS measurements. 

2. 

DK1622 wildtype 

Intens. 

5 

x10 

200 

5 

• A-domain 10 in Mxp occas- 

1622 

sionally incorporates Leu, but is 

also skipped frequently 

3 

• Identification of a lowabundance 

candidate 

compound (506.2712+ m/z) 

in LC-MS datasets. 

• Identification of alternative producers represents an important 

key for the isolation and structure elucidation of 

novel myxobacterial metabolites which are initially produced 

in low yields only. 

0.06 x 

2 

100 

10 11 

A PCP C A PCP C A PCP TE 

S S S 

βLys Leu Ala 

Tyr βLys Leu 

Ser Tyr βLys 

Val* Ser Tyr 

Ser Val* Ser 

Ser Val* 

Ser 

OHVal 

Leu 

BPC 100-2000 m/z 

1 

OHVal 

Leu 

strain no. 

OHVal 

Leu 

NMe 

-Ser 

10 11 

A PCP C A PCP C A PCP TE 

S S S 

βLys βLys Ala 

Tyr Tyr βLys 

Ser Ser Tyr 

Val* Val* Ser 

Ser Ser Val* 

OHVal OHVal Ser 

Leu Leu OHVal 

NMe NMe Leu 

-Ser -Ser 

NMe 

-Ser 

• skipping apparently always 

occurs in the corresponding 

module 11 of Mxp897 3 

EIC 506.2713 m/z 

Ret. (min) 

0.0 

NMe 

-Ser 

O 

2 4 

Wildtype 

NMe 

-Ser 

Ser 

– OHVal – Ser – Val 

NMeSer – Leu 

c649 

(648 Da) 

• module 6 is involved in formation 

of the central α-oxo 

moiety from Val and mCoA 

DK1622 mutant 

Intens. 

5 

x10 

O 

O 

SUMMARY REFERENCES & ACKNOWLEDGEMENT 

– Tyr – βLys – Leu – Ala 

Ser 

c562 (1122 Da) 

3 

NMeSer – Leu – OHVal – Ser – Val 

OH 

O 

Ser 

NMeSer – Leu – OHVal – Ser – Val 

• 13 c651 

C acetate feeding revealed 

(650 Da) 

that only C2 is maintained in 

myxoprincomides 

2 

[1] Wenzel SC, Müller R (2009). Curr.Opin.Drug Disc.Dev. 12(2), 220-230 

[2] Cortina NS, Krug D, Plaza A, Revermann O, Müller R (2012). Angew. 

Chem.Int.Ed. 51, 811-816. 

• Metabolome-mining led to the disovery of myxprincomides 

from M. xanthus DK1622 and subsequently 

enabled characterization of a new compound family 

widely distributed in Myxococcus strains. 

O 

O 

– Tyr – βLys – Ala 

Ser 

– OHVal – Ser – Leu 

NMeSer – OHVal – Phe 

c587 (1172 Da) 

BPC 100-2000 m/z 

EIC 506.2713 m/z 

Ret. (min) 

1 

Mutant 

O 

0 

2 4 

[3] Krug D, Zurek G, Schneider B, Garcia R, Müller R (2008). Anal.Chim. 

Acta 624, 97-106 

8 9 10 11 

C A PCP C A PCP C A PCP C A PCP TE 

OH 

3. 

[4] Krug D, Zurek G, Revermann O, Vos M, Velicer GJ, Müller R (2008). 

Appl.Environ.Microbiol. 74, 3058-3068 

6 

• biosynthetic intermediates 

are released from the assembly 

line (LC-MS and NMR data) 

C A PCP 

S 

O 

HO NH 

O 

OH 

NH 

O 

C A PCP 

S 

O 

H2N OH 

ACP 

S 

O 

O 

OH 

NH 

O 

• Analysis of the giant NRPS/PKS mxp genes from 

strains DK1622 and DK897 sheds light on the intriguing 

details of myxoprincomide biosynthesis. 

6 7 

KS AT ACP C Ox C A PCP C 

S 

S 

O 

O 

O 

[Ox] 

H2N OH 

NH 

O 

OH 

T7A1 

∗ 

HO 

O 

H 

N 

H 

N 

O 

H 

N 

O 

H 

N 

O 

H 

N 

O 

H 

N 

We wish to thank Bruker Daltonik for their continued support and for providing their 

LC-MS software tools. Funding from DAAD is gratefully acknowledged (NSC). 

• active-site mutation of the TE 

domain does not alter the observed 

product profile 

OH 

N 

H 

N 

H 

N 

H 

M. xanthus A2 

MXAN_3779 

∗ 

C A PCP 

S 

O 

H2N OH 

O 

O 

O 

O 

O 

OH 

OH 

HO 

• A number of myxoprincomides which could be isolated 

and structurally elucidated are actually biosynthetic intermediates, 

drawing the picture of a NRPS/PKS 

assembly line “under evolutionary construction”. 

C A PCP C A PCP C A PCP C A PCP TE 

C A PCP 

S 

O 

HO NH 

O 

O 

[Ox] 

NH 

O 

ACP 

S 

O 

O 

OH 

NH 

O 

ACP 

S 

OH 

O 

O 

NH 

O 

ACP 

S 

O 

HO 

O 

NH 

O 

ACP 

S 

O 

O 

OH 

NH 

O 

NH 2 

Ser 14090 

∗ 

Myxoprincomide c506 

C H N O 45 74 10 16 

Ala 

• Selection of an alternative producer and yield improvement 

by promotor insertion prior to upscaling for fermentation, compound 

isolation and full structure elucidation. 

-9-

-10- 



Metabolic profiling of a Corynebacterium 

glutamicum ΔprpD2 by GC-APCI high 

resolution Q-TOF analysis 

Aiko Barsch 1 , Marcus Persicke 2 , Jens 

Plassmeier 2 , Karsten Niehaus 2 , Gabriela 

Zurek 1 


2 Centrum für Biotechnologie, Universität 

Bielefeld, Germany 

Introduction 

Metabolomics studies based on Gas 

chromatography – Mass spectrometry 

(GC-MS) are well established and 

typically employ electron impact (EI) 

ionisation. Currently, many possible 

biomarkers remain “unknowns” in 

these studies. This is due to the lack of 

reference spectra for a majority of 

biologically relevant compounds. 

Hyphenating GC with ESI-(Q)-TOF 

technology by atmospheric pressure 

chemical ionisation (APCI) permits the 

characterization of compounds which 

can not easily be identified by classical 

GC-EI-MS. 

Corynebacterium glutamicum, a gram 

positive bacterium, is used in the 

industrial production of amino acids 

and can be grown on different carbon 

sources. Glucose is metabolized via 

glycolysis and tricarboxylic acid cycle 

(TCA) whereas propionate is 

catabolized through the methylcitric 

acid pathway (see Fig. 1). An 

involvement of the prpD2 gene, 

encoding 2-methylcitrate dehydratase, 

in the degradation of propionate has 

previously been shown by Plassmeier 

et. al [1]. 

Here, we used a GC-APCI-MS based 

metabolic profiling approach to analyse 

metabolite extracts of a prpD2 mutant 

C. glutamicum strain grown on glucose, 

with a propionate pulse in the 

exponential growth phase. The benefits 

of high resolution MS data in 

combination with GC separations to 

facilitate structure elucidation will be 

demonstrated. 

References 

[1] J. Plassmeier, A. Barsch, M. Persicke, K. 

Niehaus, J. Kalinowski, J. Biotechnol. 130 

(2007) 354-363 

Propionate 

Propionyl-P 

Propionyl-CoA 

2-Methylcitrate 

PrpD2 

2-Methyl-cis-aconitate 

2-Methylisocitrate 

Pyruvate 

Oxaloacetate 

Malate 

Fumarate 

Succinate 

Glucose 

PEP 

Pyruvate 

Acetyl-CoA 

TCA cycle 

Succinyl-CoA 

Citrate 

cis-Aconitate 

Isocitrate 

2-Oxo-glutarate 

Fig. 1: Metabolic pathways of the glycolysis, 

tricarboxylic acid (TCA) and methylcitric acid cycles 

highlighting the step catalyzed by PrpD2, which is 

not present in the C. glutamicum ΔprpD2 deletion 

mutant - adapted from [1]. 


Methods 

C. glutamicum ΔprpD2 strains were 

cultivated and harvested as described in 

[1]. 

• two biological replicates were grown on 

glucose, with a propionate pulse in the 

exponential growth phase. 

• Two technical replicates of each culture 

were harvested by centrifugation before 

and 1 h after the propionate pulse. 

• Dried methanolic metabolite extracts 

and reference standards were derivatized 

by methoxymation and trimethylsilylation. 

• Gas chromatography conditions: 

Injection volume: 1µl; Column: HP-5MS 

(30 m x 0.25 mm i.d.; 0.25 µm); Injector 

temperature: 250°C; Helium carrier gas: 

1 ml / min; Temp. Gradient: 3 min @ 

80°C, 5°C / min to 325°C. 

• MS: Bruker Daltonics micrOTOF-Q IITM interfaced via a Bruker GC-APCI source. 

Data was acquired in positive ionization 

mode from 85 – 750m/z at 4 spectra per 

second. 

• Relevant features were extracted using 

the Find Molecular Features algorithm. 

Feature intensities were normalized to the 

intensity of the internal standard ribitol. 

Principal Component Analysis (PCA) as 

well as sum formula generation by 

SmartFormulaTM were performed using 

ProfileAnalysisTM 2.0. MS/MS data was 

correlated with structural hypothesis using 

the FragmentExplorerTM in DataAnalysis 

4.1TM . 

Results 

• High resolution MS data acquisition 

by coupling of gas chromatography 

via an APCI source to a 

micrOTOF-Q II instrument enabled to 

use the same data (pre-)processing 

methods established for LC-TOF-MS 

data. 

• A PCA based on all extracted features 

revealed a clear separation of 

samples according to the carbon 

sources used for cultivating the C. 

glutamicum ΔprpD2 cells (Fig.2 A 

scores plot) . 

• Peak intensities for all samples of two 

selected compounds are displayed as 

bucket statistics in Fig. 2 B. Both 

show a higher abundance in cells 

metabolizing propionate. 

A 

B 

PC 2 

Scores 

1.0 

0.5 

0.0 

-0.5 

Intensity x 10 5 

1.5 

1.0 

0.5 

0.0 

-Prop 

+Prop 

-1.0 -0.5 0.0 0.5 PC 1 

Bucket: 27.2min : 495.21m/z 

2 4 6 Analysis 

• Utilizing exact mass and isotopic 

pattern information SmartFormula 

generated 18 possible elemental 

formulae for compound 27.2min : 

495.21m/z (Fig.3 A). C 19H 43O 7Si 4 

was ranked as first sum formulae 

according to the goodness of the 

isotopic fit - the SigmaRank. 

• Combining accurate mass and 

isotopic pattern information of MS 

and MS/MS spectra, sum formulae 

suggestions for this compound 

could be reduced to a single hit 

(Fig.3 B). The formula C 19H 43O 7Si 4, 

is in accordance to derivatized 2methylcitric 

acid (Fig. 3B inset) and 

could be confirmed in comparison 

to the reference standard. 

• An accumulation of 2-methylcitrate 

in C. glutamicum cells metabolizing 

propionate could be expected in 

cells lacking 2-methylcitrate 

dehydratase (see Fig.1). A higher 

concentration for this compound in 

prpD2 mutant C. glutamicum 

grown on propionate was also 

observed by Plassmeier et al. 

based on “classical low resolution” 

GC-EI-(Iontrap)-MS measurements 

[1]. 

• The molecular formula of 

Compound 9.3min : 234.13m/z 

(Fig. 2 B) could be limited to 

C 9H 24NO 2Si 2 by SmartFormula 3D 

(not shown). A Sum formula query 

in public databases using the 

Compound Crawler TM returned TMS 

derivatized alanine as a likely 

structure (Fig.4 A). The top part of 

Fig. 4 B displays the MS/MS 

spectrum of derivatized alanine 

with assigned sum formulae for 

neutral losses. The automatically 

calculated SmartFormula 3D 

spectrum (Fig.4 B bottom) includes 

annotations for the observed 

fragment ions. 

• Correlating accurate mass MS/MS 

data with possible fragments of the 

likely precursor structure using the 

FragmentExplorer TM further 

substantiated this hypothesis (Fig. 

4 C), which could be confirmed by 

comparison to the reference 

standard. 

PC 2 

Loadings 

1.0 

-0.5 

-1.0 -0.5 0.0 0.5 PC 1 

Fig. 2: A: PCA scores and loading plot (Explained Variance PC1 vs. 

PC2: 44.4% + 29.9%) of C. glutamicum ΔprpD2 mutant strains grown 

on glucose before and after a propionate pulse. The scores plot 

revealed a clustering of samples according to the carbon sources used 

for cultivation. B: Bucket statistics plots for two selected loadings 

mainly contributing to the grouping of samples observed in A. 

0.5 

0.0 


5 

4 

3 

2 

1 

Bucket: 9.3min : 234.13m/z 

2 4 6 Analysis 

A 

B 

TMS-O 

O 

O 

TMS-O 

H3C O-TMS 

TMSO - 

O 

Elemental formulae 

of precursor ion 

combining MS and 

MS/MS information 

Fragment and neutral loss 

elemental formulae 

Fig. 3: Identification of compound 27.2min: 

495.21m/z (see Fig.2 B): Molecular Formula 

generation by SmartFormula (A) and 

SmartFormula3D (B). Combining accurate mass 

and isotopic pattern information in MS and MS/MS 

spectra one candidate formula was generated 

which is in accordance to derivatized 2methycirate 

(inset). 

A 

B 



C 

+MS2(234.1327) 

C4 H10 2 Si1 + 0.83 mDa 

6 

C1 H4 + 0.30 mDa 

C3 H10 O1 Si1 - 0.22 mDa 

234.1336 

4 

C4 H14 O1 Si1 + 0.09 mDa 

C1 H4 - 0.32 mDa 

2 

144.0833 

116.0894 

128.0523 

0 

2.5 

218.1019 

SmartFormula3D spectrum: C 9 H 24 N 1 O 2 Si 2 

C9 H24 N1 O2 Si2, 

2.0 

C6 H14 N1 O1 Si1, 

0.66 mDa, 

0.38 mDa, 

234.1340 

1.5 C5 H14 N1 Si1, 

-0.39 mDa 

144.0839 

1.0 C5 H10 N1 O1 Si1, 

0.36 mDa, 

C 8 H 20 N 1 O 2 Si 2, 

0.5 116.0890 

0.76 mDa, 

0.0 

120 140 160 180 200 220 240 m/z 

Fig. 4: Compound 9.3min:234.13m/z (see Fig.2 

B) – Identification as alanine - 2TMS. A: 

Compound Crawler database query. B: 

SmartFormula 3D spectrum with annotated sum 

formulae for fragment ions and neutral losses. C: 

FragmentExplorer TM - correlation of fragments of 

likely precursor structure with MS/MS data. 

Conclusions 

• PCA analysis of GC-APCI-TOF-MS 

data revealed several compounds 

elevated in C. glutamicum extracts 

following an addition of propionate 

• 2-methylcitric acid and alanine 

were identified using accurate 

mass and isotopic pattern 

information in MS and MS/MS 

spectra - proof of concept for the 

identification of target compounds 

using high resolution MS 

technology 

• TOF instrument performance 

(mass accuracy, resolution and 

isotopic fidelity) are independent 

of the acquisition rate. Therefore 

these systems are perfectly suited 

for coupling to gas 

chromatography. 

GC-APCI-QTOF-MS


GC-APCI-MS/MS based identification of metabolites 

in bacterial cell extracts 

Christian Jäger, Maike Sest & Dietmar Schomburg 

Department of Bioinformatics & Biochemistry, TU Braunschweig, Germany 

ch.jaeger@tu-braunschweig.de 

Introduction 

In order to analyze and understand the metabolism of microorganisms, it is 

important to know all the components and their interaction. The goal of 

metabolomic research is to measure concentrations of as many metabolites as 

possible. It is estimated that about 800 different chemical entities, covering a wide 

range of polarities, reactive behavior and sizes, are involved in a microbial 

metabolism. 

Conventionally used ionization techniques in gas chromatography – mass 

spectrometry, like Electron Ionization, generate reliable and unique fragment 

patterns. Commercial and in-house libraries can be used to identify the separated 

compounds. For identification of unknown compounds additional information is 

needed. 

cell 

cultivation 

Conclusions 

harvesting 

washing 

cell lysis 

metabolite 

extraction 

Metabolic profiling with GC-APCI-MS 

Metabolome analysis 

derivatization 

(MeOX & MSTFA) 

Normally, 250-300 targets were detected in GC-EI-MS chromatograms of 

P. putida. Only 100 derivatives could be identified. With GC-APCI-MS 

measurements it is possible to detect up to 350 chemical species with accurate 

mass. We used the same chromatographic methods on every GC-MS system for 

a fast compound recovery. 

GC-APCI-MS chromatogram 

(deconvoluted) of a polar cell 

extract (P. putida) 

The presence/absence of a mass shift between the 12 C- and the co-eluting 13 C- 

peak indicates the biological / non-biological origin and the number of carbon 

atoms of the measured compound. 

C-source: 

labeled / unlabeled 

glucose 

12 C 

Almost all of the identified derivatives from EI - measurements could be confirmed 

by sum formula comparison. In addition, many compounds below matching factor 

(0,75) were identified as the first suggested possible hit. 

● deconvoluted targets GC-EI-MS : 265 → 96 identified derivatives (36%) 

● deconvoluted targets GC-APCI-MS : 343 

●biological origin: 220 (64%) 

●non-biological origin: 43 (13%) 

●no assignment: 80 

● GC-APCI-MS/MS data provides information of the molecular ion and enables the 

assignment of the elemental composition to the TMS derivatives and its fragments. 

Additionally, functional groups can be derived from specific neutral losses. 

● The developed identification strategy is a valuable supplementing method in 

combination with GC-EI-MS. It can be used for a fast, reliable and comprehensive 

evaluation of metabolic profiles on following terms: 

► sum formulae of the metabolite ► biological / non-biological origin 

► confirmation of ambiguous EI–spectra ► improvement of deconvolution 

13 C 

Identification strategy 

We screened GC-MS chromatograms of bacteria especially for unknown 

substances with biological relevance. The microorganisms were cultivated on 

glucose and stable isotope-labeled [U- 13 C]-glucose to allow unambiguous 

identification of biologically related compounds. Sum formulae were calculated 

with high resolution - mass spectrometry and structural information were gained 

by MS/MS fragmentation. 

We present the evaluation of GC-APCI-MS chromatograms using the example of 

Pseudomonas putida and the structural analysis of a compound from the polar 

cell extract of Yersinia pseudotuberculosis. This compound does not match to 

already known derivatives from our in-house library, the Golm Metabolome 

Database 1 (GMD) and the NIST08 database. 

GC-MS 

measurement 

References 

untargeted 

compound analysis 2 

identified derivates 

selected unknowns 

identification 

pipeline 

Mass spectrometry based structure elucidation 

The analysis of 70 eV fragmentation pattern of (derivatized) unknown compounds 

is difficult without specific information. For instance, the molecular ion is of low 

abundance or not detectable. 

TMS+H 

TMS-OH+H 

C6H4O3 C7H5O4 - 2 TMS-OH 

- CO 

[M-15] + 

- 2 TMS-OH 

C13H23O5Si2 MS/MS 

- TMS-OH 

[M+H] + 

C 16 H 33 O 6 Si 3 

Top GC-APCI-MS(/MS) – spectra. One metastable ion (m/z 315.1075) appears. 

Bottom MS/MS - fragment pattern of m/z 405.1579 plus possible neutral losses. 

● molecular ion [M+H] + → m/z 405.1579 

● molecular ion [M+H] + → m/z nominal 412 ( 13 C) 

● sum formula derivative → C 16 H 33 O 6 Si 3 

● sum formula metabolite → C 7 H 8 O 6 (- 3TMS groups) 

● rings and double bonds → 4 

Electron Ionization – mass spectrum of 

'unknown YPY / RI 1846'. Loss of one 

methyl group is a common reaction of 

derivatized metabolites with MSTFA. 

With GC-APCI-MS/MS measurements we could determine the molecular weight, 

sum formulae of almost all measured ions/fragments and the presence of 

functional groups. 

- TMS-OH 

- CO 

1 Golm Metabolome Database (GMD), http://gmd.mpimp-golm.mpg.de (download, 2011) 

Schauer, N., et al. (2005) 'GC-MS libraries for the rapid identification of metabolites in complex biological 

samples', FEBS letters, 579, pp 1332-1337 

2 Hiller, K., et al. (2009), 'MetaboliteDetector: Comprehensive Analysis Tool for Targeted and 

Nontargeted GC/MS Based Metabolome Analysis', Anal. Chem., 81, pp 3429–3439 

GC-APCI-MS system 

Agilent 7890A gas chromatograph equipped with a ZB-5MS column (Phenomenex), 29 min run 

Bruker micrOTOF-QII with APCI, in full scan mode m/z 50...1550, 0.4 Hz repetition rate 

- CH 4 

All library entries can gradually be 

updated with new information 

concerning the chemical nature of 

the unknown compounds. 

-11-

-12- 


Yeast Metabolomics 

Untargeted metabolic profiling of yeast 

extracts by UHR-Q-TOF analysis to study 

arginine biosynthesis mutants 

Sandy Yates 1 ; Manhong Wu 2 , Aiko Barsch 3 ; 

Gabriela Zurek 3 ; Bob St. Onge²; Sundari Suresh², 

Ron Davis², Gary Peltz² 

1 Bruker Daltonics, Fremont, CA 

2 Stanford University, CA 


Introduction 

Many primary polar metabolites, such as 

amino acids, are poorly retained by reversed 

phase LC column chemistries which are 

widely used for untargeted metabolic 

profiling studies. Dansylation of primary 

amines, secondary amines and phenolic 

hydroxyl functional groups in these 

compounds alters the chromatographic 

behaviour, thus enabling a standard RPLC 

separation with simultaneously 

enhancement of the signal intensities in 

electrospray ionization [1]. 

We analysed dansylated metabolite extracts 

from yeast wild type and deletion mutants 

in the arginine biosynthetic pathway using a 

novel UHR-Q-TOF instrument. To prove the 

hypothesis that upstream or downstream 

metabolites are altered in abundance in 

mutants we applied an untargeted 

Metabolomics workflow. Target compounds 

could be identified based on accurate mass 

and isotopic pattern information as well as 

based on fragment ions of the dansylated 

target compounds. 

Fig. 1. Dansylation reaction schema 

Methods 

HPLC: U3000 RSLC, (Dionex). 

Column: Kinetex C18 2.1x100mm; 2.6µm 

(Phenomenex). Column temp. 30 °C. 

Flow rate: 0.5 mL/min. Injection volume: 5 

µL. Mobile phase: A = H 2O, B =AcN (each 

containing 0.1% HCOOH). Gradient: 0.5min 5% 

B; linear gradient 5 – 60% B in 20 min, linear 

gradient 60 – 95% B in 4.5 min hold 5 min. MS: 

maXis Impact (UHR-Q-TOF MS, Bruker Daltonik 

GmbH). Ionization: ESI(+). Scan range: m/z 

50-1000. Acquisition rate: 2.5 Hz 

Data processing: SmartFormula3D and the 

FragmentExplorer used for sum formula 

generation and assignment of fragment structures 

are part of DataAnalysis 4.1. ProfileAnalysis 2.1 

was used for statistical data evaluation (Bruker 

Daltonik GmbH). 

Results 

150 200 250 300 350 400 m/z 

Sample: Yeast wild type as well as arg4 and arg7 

gene deletion mutant strains were grown on a 

synthetic growth medium. Six biological 

replicates were harvested by centrifugation at the 

same growth phase and washed with PBS buffer. 

Metabolites were extracted using the Folch 

method and dansylated as described in [1]. 

21 amino acid standards (2.5 µmol/ml each 

peptide) were mixed and dansylated yielding a 

final concentration of 1 µmol amino acid / ml. 

C 

Intens. Fig. 3. Example - Dansylmethionine A: MS 

5 

x10 

2.0 

1.5 

1.0 

0.5 

Dns-Asparagine 

Dns-Arginine 

Dns-Glutamine 

Dns-Serine 

Dns-Aspartate 

Dns-Glutamate 

Dns-Glycine 

Dns-Threonine 

Dns-Alanine 

Dns-GABA 

Dns-Proline 

Dns-Valine 

Dns-Methionine 

Dns-Tryptophan 

Dns-Phenylalanine 

Dns-Isoleucine 

Dns-Leucine 

Dns-Norleucine 

0.0 

4 6 8 10 12 14 16 18 Time [min] 

Fig. 2. 2mDa trace width high resolution 

EICs for dansylated standards reveal a 

significant retention for derivatized amino 

acids on a reversed phase column. 


Dns-Lysine 

Dns-Histidine 

Dns-Tyrosine 

Dansylation of purified amino acids 

– method development 

A mix of pure amino acids was derivatized 

by dansylation (see Fig. 1) and separated by 

C18 RP chromatography. The high mass 

accuracy of the acquired full scan MS data 

enabled to create 2mDa trace width high 

resolution Extracted Ion Chromatograms for 

the target compounds, providing a very high 

selectivity (Fig. 2). 

As shown for the example of Dansylmethionine 

(Fig. 3) the correct sum formula 

could be generated with a mass error of 

1ppm and a very good mSigma value which 

provided a measure for the goodness of the 

measured and theoretical isotopic fit. The 

MS/MS spectrum of dansylmethionine 

contained a characteristic fragment ion near 

m/z 171. The SmartFormula3D tool 

correctly assigned the sum formulae for the 

observed fragment ions. This information 

was linked to structural information using 

the FragmentExplorer, which also enabled a 

quick annotation of the MS/MS spectrum. 

Untargeted Metabolomics of Yeast 

Mutants in the Arginine Synthesis 

Pathway 

The established method was applied to 

study yeast mutants blocked in the 

arginine biosynthetic pathway. Fig. 4 

shows the biochemical pathway of 

arginine with the reactions catalysed by 

the proteins encoded by arg4 and arg7 

highlighted. 

A 

B 

Intens. 

4 

x10 

6 

4 

2 

0 

Intens. 

x10 4 

6 

4 

2 

0 

1+ 

383.1090 

Dansyl-Methionine 

[M+H] + C 17H 23N 2O 4S 2 

Error 1.0 ppm 

mSigma 6.7 

Res. 23921 

382.5 383.0 383.5 384.0 384.5 385.0 385.5 386.0 386.5 387.0 m/z 

+MS2(383.1090) 

170.0966 

N 

234.0587 

MS/MS 

MS 

383.1096 

spectrum measured on a maXis Impact; B: 

MS/MS spectrum with annotated fragment 

structures, including the 170.0997 m/z 

significant reporter ion indicating a 

danylated compound; C: SmartFormula3D: 

sum formula generation based on precursor 

and fragment ion information. D: 

FragmentExplorer links molecular formulae 

from SmartFormula3D result with structural 

suggestions for fragment ions and enables 

a quick annotation of the MS/MS spectrum. 

References 

[1] Guo K., Li L. (2009). Anal. Chem. 81, 3919-3932 

D 

All features were extracted from the 

raw data using “Find Molecular 

Features”. This algorithm can combine 

isotopic peaks, charge states and 

adducts which belong to the same 

compound into one feature. Based on 

these extracted features an 

unsupervised clustering could clearly 

separate the derivatized metabolite 

extracts from wild type and both 

mutant strains (Fig. 5). 

Fig. 4. Arginine biosynthesis 

highlighting steps blocked in yeast arg7 

and arg4 deletion mutants. 

Arg4 

arg4 

arg7 

WT 

Fig. 5. Unsupervised clustering: 

Dendrogram reveals a class separation 

according to the analysed yeast wild type 

and mutants. 

Comparison of arg7 vs. Wild Type 

In order to determine statistically significant 

differences between sample groups, a 

student’s t-test was performed comparing 

wild type vs. arg 4 and wild type vs. arg7 

samples. Fig. 6 presents the Volcano plot of 

the t-test result, plotting the log2 of fold 

changes vs. -log10 of the p-Value, for the 

arg 7 vs. wild type comparison. The 

compound with the highest fold change of 

60 fold increase in arg7 samples and a p- 

Value of 0.0000026 had an accurate mass of 

408.1588 m/z. Fig. 6 B shows the overlaid 

measured spectrum and the simulated 

spectrum for C19H26N3O5S, which was 

ranked #1 with the best isotopic fit by 

SmartFormula. This sum formula fits to 

Dansylated N-acetylornithine which is the 

metabolite upstream of the reaction 

catalysed by the arg7 gene product. EIC 

traces for the target compound were plotted 

in ProfileAnalysis which could validate the 

accumulation of Dansyl N-acetylornithine in 

arg7 mutant samples (Fig. 6 B inset). 

Comparison of arg4 vs. Wild Type 

Table 1 presents several selected 

compounds which were significantly 

different (p-Value


A workflow for metabolic profiling and 

determination of the elemental 

compositions using MALDI-FT-ICR MS 

Kazunori Saito 1 ; Tatsuhiko Nagao 2 ; 

Daisuke Miura 2 ; Takashi Nirasawa 1 ; 

Hiroyuki Wariishi 2 

1 Bruker Daltonics K.K., Yokohama, Japan; 

2 Kyushu University, Fukuoka, Japan; 

Introduction 

Metabolomic studies can lead to the 

enhanced understanding of 

mechanisms for drug or xenobiotic 

effects and an increased ability to 

predict individual variations in drug 

response phenotypes. Tandem mass 

spectrometry (MS) has been 

extensively used for this purpose, 

however, an identification of 

metabolites in this approach is 

dependent on spectral databases or 

comparisons with authentic reference 

standard spectra. This is hampered by 

the fact that only about 20% of all 

expected metabolites are commercially 

available. Since Fourier transform ion 

cyclotron resonance (FT-ICR) MS 

provides an ultrahigh mass resolution 

and accuracy which enables isotopic 

fine structure analysis it allows for 

highly confident chemical annotations. 

In addition, due to this high resolving 

power a separation of complex samples 

by liquid chromatography is often not 

essential. Here we present a metabolic 

profiling workflow based on matrix 

assisted laser desorption ionization 

(MALDI)-FT-ICR MS which offers rapid 

profiling capabilities for metabolomic 

analyses. 

Methods 

HepG2 cells, a human liver carcinoma cell, 

were treated for 24 h with the anticancer 

drug 5-fluorouracil (5FU) at IC50 (10 µM; 

5FU-Low) and excess (50 µM; 5FU-High) 

concentrations. Intracellular metabolites 

of five biological replicates (n=5) were 

extracted by a biphasic extraction method 

using methanol/water/chloroform=2/2/1. 

A solariX FT-ICR MS (Bruker Daltonik 

GmbH, Bremen, Germany) has been used 

for MS measurements. The MS system 

was equipped with a 9.4 tesla 

superconductive magnet and an 

ESI/MALDI dual ion source. 

t[2] 

7 

6 

5 

4 

3 

2 

1 

0 

-1 

-2 

-3 

-4 

-5 

-6 

-7 

(A) (B) 

5FU-Low 

5FU-High 

Control 

-8 -7 -6 -5 -4 -3 -2 -1 0 

t[1] 

1 2 3 4 5 6 7 8 


Measurements were carried out in MALDI 

mode. 9-aminoacridine (9AA) was applied 

as matrix in negative ion mode. Data 

were acquired within 30 seconds per 

spectrum and external calibration was 

applied. Acquired data were evaluated 

using multivariate statistical analysis. 

Principal component analysis (PCA) and 

orthogonal partial least-squares 

discriminant analysis (OPLS-DA) were 

carried out using SIMCA-P+ ver. 12.0.1. 

software (Umetrics AB, Umeå, Sweden). 

Elemental compositions were calculated 

by SmartFormula software (Bruker 

Daltonik GmbH). The derived formula 

candidates were queried in the public 

database Chemspider using the 

CompoundCrawler software tool (Bruker 

Daltonik GmbH). 

Results and Discussion 

5FU acts as a pyrimidine analogue and 

is transformed inside the cell into 

different cytotoxic metabolites which 

can be incorporated into DNA and RNA, 

finally inducing cell cycle arrest and 

apoptosis by inhibition of cellular DNA 

synthesis. Although these drugs are 

well-known and well-established 

anticancer drugs that inhibit nucleotide 

biosynthesis, there is little information 

concerning the metabolic response of 

cancer cells against 5FU. Metabolic 

responses of HepG2 cells to this 

anticancer drug were studied by 

comparing extracts of treated and 

untreated cells by metabolic profiling. 

For comprehensive identification of 

metabolites, FT-ICR MS was selected 

as analytical platform. 

MALDI-FT-ICR MS analysis revealed 

thousands of peaks in the spectra. We 

evaluated the complex FT-ICR MS data 

by PCA and OPLS-DA to differentiate 

metabolic states of HepG2 cell treated 

with different concentrations of 5FU. As 

shown in Fig. 1A, metabolic profiles of 

5FU-high and control groups could be 

separated. The OPLS S-plot comparing 

5FU-High and control samples is shown 

in Fig. 1B. Several loadings were 

responsible for this separation. 

Elemental compositions for these 

compounds were calculated and 

queried in public databases to retrieve 

possible structures. As shown in Table 

1, those could be nucleotides and 

amino acids. Figure 2 reveals that the 

peak intensities of 328.045 and 

540.054Da were reduced in a dosedependent 

manner. 

Fig. 1. PCA scores plot between 5FU-High, 5FU-Low, and control groups (A) and OPLS S-plot of 5FU-High versus control (B). 

p(corr)[1] 

1.0 

1.0 

0.9 

0.8 

0.7 

0.6 

0.5 

0.5 

0.4 

0.3 

0.2 

0.1 

0.0 

0.0 

-0.1 

-0.2 

-0.3 

-0.4 

-0.5 

-0.5 

-0.6 

-0.7 

-0.8 

-0.9 

-1.0 

-1.0 

p(corr)[1] 

4 

9 

2 

11 6 

10 

7 

8 

1 3 

110513_OPLS2.M1 (OPLS/O2PLS-DA) 

p[Comp. 1]/p(corr)[Comp. 1] 

Colored according to Var ID (Primary) 

▲ 

12 

5 

-0.22 -0.20 -0.20 -0.18 -0.16 -0.14 -0.12 -0.10 -0.10 -0.08 -0.06 -0.04 -0.02 0.00 -0.00 0.02 0.04 0.06 0.08 0.10 0.10 0.12 0.14 0.16 0.18 0.20 0.20 0.22 0.24 0.26 

p[1] 

p[1] 

R2X[1] = 0.40124 SIMCA-P+ 12.0.1 - 2011-05-18 11:44:06 (UTC+9) 

328.05860 

328.05860 Measured Spectrum 

329.06205 

(A) (B) (C) 

328.04524 

328.05847 

Δ13mDa 

328.04541 

328.040 328.060 m/z 

329.06205 330.05448 

329.04860 

400,000 FWHM 

Theoretical Spectrum 

for (C10H11N5O6P) - 






for (C10H15N3O6SNa) - 





Fig. 3. Measured FT-ICR mass spectrum and theoretical spectra of [C 10H 11N 5O 6P] - and 

[C 10H 15N 3O 6SNa] - . Resolving power was 400,000 FWHM. m/z 328-330 (A), 329 (B), 330 

(C) are indicated. Isotope fine structures are in accordance to the theoretical ones. 

To enhance the confidence of the 

formula assignments, isotopic fine 

structures were compared with the 

theoretical ones. Figure 3 shows 

that the ultrahigh resolution 

spectrum for two compounds is in 

accordance to the theoretical 

spectra of the corresponding sum 

formulae which were suggested by 

elemental composition analyses. 

The high spectral resolution of 

>400,000 FWHM was required not 

only for separation of both 

monoisotopic peaks close each 

other but also for the isotopic fine 

structure separation. Since 

metabolomics samples are typically 

highly complex mixtures, this level 

of ‘ultrahigh resolution’ is one of 

the key features for non-targeted 

metabolomics and biomarker 

discovery. 

No m/z Formula Compound 

1 328.045 C10H11N5O6P cAMP 

2 294.071 C10H13N3O6Na 

3 540.054 C15H20N5O13P2 

cyclic-ADPribose 

4 328.059 C10H15N3O6SNa Glutathione 

5 228.099 C9H14N3O4 

6 606.074 C17H26N3O17P2 UDP-GlcNAc 

7 448.004 C10H13N5O10P2Na ADP 

8 306.077 C10H16N3O6S Glutathione 

9 429.994 C10H11N5O9P2Na ADP-H2O 

10 408.012 C10H12N5O9P2 ADP-H2O 

11 272.089 C10H14N3O6 

12 146.046 C5H8NO4 

Glutamic 

acid 

329.04888 

329.05574 

13 C 

329.04860 

15 N 

(A) 

(B) 

13 C 

329.06183 

329.06183 

330.05427 

15 N 33S 

329.05551 

328.0 328.5 329.0 329.5 330.0 m/z 329.040 329.050 329.060 m/z 

Table 1. Summary of elemental 

composition analysis and database 

search. 

Summary 

330.04763 

No.1. 328.045 

5FU-High 5FU-Low Control 

No.3. 540.054 

330.05448 

18 O 13 13C2 C2 C2 

330.04949 

34 S 

330.05427 

5FU-High 5FU-Low Control 

MALDI-FT-ICR MS analysis was 

successfully employed for metabolic 

profiling of HepG2 cells treated with 

the drug 5FU. PCA and OPLS-DA of 

ultrahigh resolution MS data could 

differentiate sample groups and find 

several responsible loadings. 

Elemental composition analysis 

provided candidates which could be 

confirmed by the ultrahigh resolution 

data permitting a comparison of 

measured and theoretical isotopic 

fine structures. The presented 

method enables rapid and nontargeted 

metabolic profiling. 

Conclusions 

• High throughput metabolic 

profiling was achieved by MALDI- 

FT-ICR MS 

• Ultrahigh resolution mass spectra 

can provide more reliable 

elemental composition analysis by 

resolving isotopic fine structures 

• Candidate compounds were found 

through database search 

• Metabolic profiling results indicate 

that 5FU seems to inhibit the 

biosynthesis of the nucleotides and 

amino acids in human liver 

carcinoma cell. 

FT-ICR MS 

330.06283 

18 O 

330.06272 

13 C 15 N 13 

C2 

330.05886 

330.050 330.060 m/z 

Fig. 2. Peak intensities of the metabolites 

no. 1 and 3. Bars show standard deviation 

(n=5). 

-13-

-14- 



Bruker BioSpin 

NMR-based screening of neonate urines 

C.Cannet 1 , D.Krings 1 , M. Spraul 1 ,H. Schäfer 1 , B. Schütz 1 , F.Fang 1 ,S.Aygen 2 ,U.Dürr 2 ,P.Tremmel 2 

1 Bruker BioSpin GmbH, Rheinstetten, Germany 

2 INFAI Institute for Biomedical Analysis and NMR Imaging GmbH, Köln, Germany 

Introduction 

NMR is a quantitative technique which can be used for identification of 

clinically relevant metabolites after a simple preparation of body fluids. A 

routine 1D proton spectrum contains hundreds to thousands of 

spectroscopic lines. (Fig. 1: typical 1H-NMR neonate urine spectrum). 

Hence from a single unspecific measurement it is possible to obtain 

information on a plethora of metabolites in a multi-marker approach. 

Figure 1: 1 H-NMR spectrum (~4 minutes) of a turkish neonate urine and two 

enlargements (magnification of factor 10 and 50) with some signal assignments. 

NMR can deliver targeted and nontargeted results from one measurement 

of neonate urine in short time, thus combining quantification of a large set 

of compounds indicative of inborn errors with the statistical analysis of 

deviations from normality. 

Study Design 

700 neonate urine samples from 8 centers in Turkey were collected and 

Spectra were recorded in full automation in two different laboratories 

(INFAI GmbH and Bruker BioSpin GmbH). 

Targeted Analysis: Quantification 

Typically, analysis of bodyfluids relies on direct quantification of preselected 

diagnostic biomarkers. The NMR signals of such biomarkers can be 

described in knowledgebases and quantified in urine spectra. Two 

examples of quantification are illustrated in Fig. 2. These 

2-Hydroxyisovaleric Acid 

Phenylalanine 

Figure 2: Quantification of 2-Hydroxyisovaleric Acid (top) and Phenylalanine 

(bottom). Left panel: Comparison of quantified values with real spiked 

concentration values. Right panel: Illustration of quantification of 

2-Hydroxyisovalecic Acid (quantified value: 52 mmol/mol creatinine, spiked value: 

50 mmol/mol creatinine) and Phenylanlanine (quantified value: 588 mmol/mol 

creatinine, spiked value: 500 mmol/mol creatinine) in NMR spectra. 

examples based on a real spiking study and, as a result, we can compare 

the quantified values with the real concentration values. 

Figure 3: Comparison of Glycine and Alanine concentrations by NMR to 

textbook ranges [1]. 

The ability to quantify a large number of compounds in every sample 

allows valid concentration distributions to be accumulated. Some were 

found to be different from the typical value sets published in textbooks 

and literature about inborn error analysis, where the number of samples 

previously used has been typically limited to small cohorts. Fig. 3 shows 

distributions of two examples with the textbook ranges drawn in. 

Untargeted Analysis 

The idea of getting references ranges for relevant metabolites can be 

extended to the complete spectroscopic NMR-pattern of a urine. This leads 

to models which can even detect deviations of unknown or not yet 

recognized compounds, a so called “untargeted analysis”. Figure 4 

illustrates this methodology on the example of 4-Hydroxyphenyllactic Acid 

(marker for Tyrosinemia I). The coloured model shows the natural 

variance of more than 700 urine samples. The spectrum with the 

Figure 4: Automated detection of deviating spectroscopical regions. The model is 

plotted as a quantile-plot, showing the natural variance of each chemical shift. 

The overlaid black spectrum is detected as non-compliant since the signals of 

4-Hydroxyphenyllactic Acid (1052 mmol/mol creatinine) do not fit in the model. 

4-Hydroxyphenyllactic Acid does obviously not fit into this cohort. This 

approach allows an automated screening procedure for the detection of 

diseases without any previous knowledge. 

References 

[1] N. Blau, M. Duran, M. E. Blaskovics, K. M. Gibson, Physican’s Guide to the Laboratory Diagnosis of Metabolic 

Diseases, Springer 2005, p. 13. 

NMR


Bruker BioSpin 

NMR Based Metabolomics in Clinical Studies of 

Human Population : Quantification 

D.Krings, C. Cannet, B. Schütz, H. Schäfer, M. Spraul , F. Fang 

Bruker BioSpin GmbH, Rheinstetten, Germany 

Introduction 

NMR is a quantitative technique which can be used for identification of 

clinically relevant metabolites after a simple preparation of body fluids. 

With the resulting NMR spectra it is possible to detect and quantify 

selected metabolites and thus creating a snapshot of the current metabolic 

phenotype. 

Preliminary Studies 

Taking into account the natural variability in urine NMR-profiles, we 

perform a virtual spiking study on the basis of 700 urine spectra of Turkish 

neonates. Consequently, we obtain ranges for reliable quantification of the 

respective metabolites. In Fig. 1 we show results of the detection of 

Methylmalonic Acid and 3-Phenyllactic Acid produced in 50000 spiking 

experiments for each metabolite. 

Figure 1: Detection rate of a spiking experiment in a cohort of newborns. The 

plot shows results of Methylmalonic Acid and 3-Phenyllactic Acid. 

Detection Limits 

In each spiking experiment we store the true and false detections. The 

curves in Fig. 1 are functions that represent the true and false detection 

rate in dependency on the spiked concentration value. The courses of such 

curves are determined by several characteristics of the searched signal 

like the chemical background of the search region and the complexity 

Methylmalonic Acid 

3-Phenyllactic Acid 

Figure 2: Two different metabolites spiked with 50 mmol/mol Creatinine virtually 

in three spectra. In addition to the concentration, the area of the signals depends 

on the number of protons (Methylmalonic Acid:3 protons, 3-Phenyllactic Acid:1 

proton). 

of the signal pattern. Fig. 2 shows a comparison of three spectra after 

spike experiments with Methylmalonic Acid and 3-Phenyllactic Acid. 

Quantification Accuracy 

Quantification accuracy of Methylmalonic Acid 

Figure 3: Measured concentration value and distribution of the true 

spiked concentration values. 

After successful detection of a metabolite we start a quantification 

procedure. Again the variability of the chemical background in different 

spectra can influence the results. A statistical analysis of such experiments 

delivers distributions of possible true concentrations values when 

measured a certain concentration value. Such distributions are exemplarily 

shown in Fig. 3 for Methylmalonic Acid. 

Detection and Quantification Example 

Finally, real spiking experiments can be used to verify the implementation 

that based on the virtual spiking studies. A process of a complete 

quantification procedure is illustrated in Fig. 4. After a successful detection 

we fit the curve of the concentration and obtain a quantification result. If 

the result reaches or exceeds the detection limit a distribution with 

Detection 

Quantification 

Result 

Distribution of 

possible Values 

527 

mmol/mol 

creatinine 

Detection 

Limit 

reached 

Figure 4: After successful detection of the desired pattern the 

quantification procedure starts and produces a distribution of possible 

concentration values (Example of L-Isoleucine) 

possible real concentration values can be generated. The distribution 

based again on the results of a virtual spiking study in 700 urine spectra. 

NMR 

-15-

-16- 



OR2 

Structure elucidation and confirmation 

for plant metabolomics: novel approaches 

Sven Heiling 1 , Gabriela Zurek 2 Emmanuel 

Gaquerel 1 , Matthias Schöttner 1 , Bernd 

Schneider 1 , Marina Behrens 2 , Aiko 

Barsch 2 , Ian Baldwin 1 

1 Max-Planck-Inst. for Chemical Ecology, Jena, 

Germany 


Introduction 

Presently, the key bottleneck of plant 

metabolomics is structural confirmation 

and elucidation of secondary 

metabolites. 

Nicotiana attenuata is a well-established 

model system to monitor plantherbivore 

interactions with 

metabolomics being a novel approach to 

investigate the underlying biology [1]. 

17-Hydroxygeranyllinallool diterpene 

glycosides (HGL-DTGs) are abundant 

direct defense compounds with their 

mode of action being largely unknown 

[1-3]. New acyclic HGL-DTGs 

(attenoside, nicotianoside I, II, III) 

were characterized using MS and NMR 

(1H, 13C) after extraction of several 

hundred grams of raw plant material 

[2,3]. Such scale is not compatible to 

the analytical scope of metabolomics. 

Here, we present novel tools facilitating 

the identification process of natural 

products when mass spectral libraries 

are not yet available and the sample 

amount is limited. 

[1] Gaquerel, E., J. Agric. Food Chem. 

58 (2010), 9418-9427. 

[2] Jassbi, A., Z. Naturforsch. 61b 

(2006), 1138-1142. 

[3] Heiling, S., Plant Cell 22 (2010), 

273-292. 

Methods 

Especially enriched samples in HGL-DTGs 

from Nicotiana leaves were generated by 

extraction with 80% methanol, subsequent 

washing and preparative chromatography 

[2,3]. Samples were subjected to LC-MS 

analysis and detailed fragmentation studies 

by means of direct infusion measurements 

(3µL/min; dilution 1:50). Standard extracts 

of various Nicotiana species were prepared 

as described in [1]. 

Chromatographic separation was carried 

out using an RSLC system (Dionex) with a 

100x2mm Acclaim RSLC 120 C18 column, 

flow rate 0.3mL/min, Solvent A: water + 

0.1% HCOOH, Solvent B: acetonitrile + 

0.1% HCOOH, linear gradient from 5%B to 

80% B in 10min. 

R1 R1 R2 R2 

Formula Mass 

Malonyl 

milli 

Sugar Sugar Sugar Sugar 

detected Accuracy 

-groups 

Sigma 

1 2 1 2 

NH4-Adduct [ppm] 

Lyciumoside I Glc Glc C32H58NO12 -0.10 3.8 

Lyciumoside II Glc Glc Glc C38H68NO17 -0.05 7.9 

Lyciumoside IV Glc Rh Glc C38H68NO16 -0.16 10.5 

Nicotianoside I Glc Rh Glc 1 C41H70NO19 0.27 3.4 

Nicotianoside III Glc Rh Glc Rh C44H78NO20 0.21 5.8 

Nicotianoside IV Glc Rh Glc Rh 1 C47H80NO23 0.13 6.3 

Attenoside Glc Rh Glc Glc C44H78NO21 -0.37 6.4 

Nicotianoside VI Glc Rh Glc Glc 1 C47H80NO24 -0.28 4.8 

Fig. 1 Core structure of HGL-DTGs (Glc= Glucose, 

Rh= Rhamnose) with the respective sugar and 

malonylgroups and identification from one 

selected Nicotiana extract. 


OR1 

MS detection was performed using a 

maXis UHR-Qq-TOF mass spectrometer 

(Bruker Daltonik) operated in ESI 

positive and negative mode in a scan 

range from 50-1500m/z (MS fullscan & 

AutoMS/MS). Mass calibration was 

achieved with sodium formate clusters. 

Data was evaluated with DataAnalysis 

4.1 software (Bruker Daltonik) using the 

FindMolecularFeatures (FMF) and Dissect 

algorithms for deconvolution. Molecular 

formula determination was carried out by 

combined evaluation of mass accuracy, 

isotopic patterns, adduct and fragment 

information using SmartFormula3D and 

the FragmentExplorer. Library spectra 

were stored in the LibraryEditor. 

Results 

Deconvolution- FMF 

Intensity x10 6 

FMF: Positive Mode 

1.5 

M+H 

1.0 

M-H2O+H 

0.5 921.4691 

939.4798 

+MS, MolFeature, 6.40-6.65min 

FMF: Negative Mode 

M+NH4 

956.5065 

M+Na 

937.4654 

983.4713 

Meas. m/z Ion Formula Sum Formula Adduct m/z 

err 

[ppm] mSigma 

0.0 

0 

920 925 930 935 940 945 950 955 960 m/z 930 940 950 960 970 980 990 m/z 

921.4691 C44H73O20 C44 H74 O21 M-H2O+H 921.469 -0.14 6.7 

939.4798 C44H75O21 M+H 939.4795 -0.33 15.4 

956.5065 C44H78NO21 M+NH4 956.5061 -0.42 6.8 

961.4616 C44H74NaO21 M+Na 961.4615 -0.10 5.9 

937.4654 C44H73O21 C44 H74 O21 M-H 937.465 -0.39 10.6 

983.4713 C45H75O23 M+HCOOH-H 983.4705 -0.87 9.9 

4 

3 

2 

1 

Fig. 2 LC-MS Basepeak chromatogram of plant 

extract enriched in HGL-DTGs; FMF spectra & 

formula assignment of Attenoside (�) in 

positive and negative mode. 

M-H 

-MS, MolFeature, Line, 6.36-6.65min 

M+HCOOH-H 

HGL-DTGs often show a very distinct 

pattern of adduct formation in ESI 

positive mode. In many cases the 

ammonium adduct is the most prominent 

ion in a series of the [M+H] + , [M+NH4] + 

and [M+Na] + adducts, however the 

adduct ion ratios are easily changed by 

different salt concentrations. In ESI 

negative mode, only the [M-H] - and the 

formic acid adduct [M+HCOO] - are 

observed. Adduct information can be 

combined with isotopes and charge states 

by the FindMolecularFeatures (FMF) peak 

finding algorithm. The corroboration of a 

sum formula by evaluating multiple 

adducts is achieved in SmartFormula. 

The FMF spectra of Attenoside are 

presented in Fig. 2 in combination with 

the SmartFormula results. Peaks assigned 

with a formula are highlighted by a yellow 

circle. In metabolomics, the FMF peak 

finder is the first step of data preprocessing 

prior to statistical analysis. 

Deconvolution- Dissect 

HGL-DTGs measured in MS positive mode 

exhibit many in-source fragments due to 

cleavages of the glycosidic bonds even at 

very soft ion transfer conditions. In 

complex samples such as the Nicotiana 

extracts, a peak deconvolution algorithm 

combining the adduct and fragments to 

individual compounds is one possible 

processing strategy in the identification 

process. 

The Dissect algorithm for deconvolution 

combines all ions with a similar 

chromatographic elution profile to one 

compound. 

A set of Dissect chromatogram traces is 

shown in Fig. 3 with an example 

spectrum of Attenoside. Combining 

precursor ion information with in-source 

fragments enabled sum formula 

annotation for neutral losses. The 

neutral losses of glucose and rhamnose 

are observed. In general MS spectra 

with in-source fragments and MS/MS 

spectra from CID fragments (Fig. 4) 

have a very similar fingerprint. 



1.0 

0.8 

0.6 

0.4 

0.2 

0.0 

1.0 

0.8 

0.6 

0.2 

0.0 

6.48 min 

6.2 6.4 6.6 6.8 7.0 7.2 7.4 Time [min] 

H2O - 0.0001 

0.4 

271.2420 

Gluc + 0.0001 

417.3000 

Rha 

471.1710 

597.3633 

Gluc + 0.0003 

+MS, Dissect, 6.33-6.65min #377-396 

956.5064 

939.4798 

759.4164 

300 400 500 600 700 800 900 m/z 

Gluc - 0.0002 

921.4691 

Fig. 3 Dissect chromatogram traces after 

deconvolution and Dissect MS spectrum 

grouping the in-source fragments of 

Attenoside. 

MS/MS Interpretation 

A 

B 


O 

HO 

O 

O 

H O 

H O 

HO 

O 

O H 

H O H O 

1.25 

H O H O 

O 

H 

O 

+ 

HO 

O 

O H 

O 

H O 

O H 

Attenoside 

H O 

1.00 

H O 

C 20 H 31 , -0.21 mDa 

271.2420 

C 18 H 31 O 14 , -0.27 mDa, 

0.75 

471.1708 

0.50 

0.25 

0.00 

SmartFormula3Dspectrum:C44H75O21, 939.48, -0.44 mDa, -0.47 ppm, mSigma: 12.0 

C 20 H 33 O 1 , -0.25 mDa 

C 12 H 21 O 9 , -0.2 mDa 

C 12 H 21 O 10 , -0.25 mDa 

C 26 H 41 O 4 , -0.09 mDa 

417.2999 

C 18 H 33 O 15 , -0.29 mDa 

C 32 H 53 O 10 , -0.11 mDa 

C 24 H 41 O 19 , -0.47 mDa 

597.3633 

633.2237 

759.4161 

939.4795 

300 400 500 600 700 800 900 m/z 

H + 

O 

O 

H O 

O H 

O 

H O 

O 

Fig. 4 (A) SmartFormula 3D result for 

Attenoside showing the proposed sum 

formulae for precursor (left) and fragment 

(right) ions; (B) SmartFormula 3D MS/MS 

spectrum of Attenoside with formula 

assignments within the spectrum; (C) 

Fragment Editor for MS/MS interpretation 

linking molecular formulae, structural 

suggestions for fragments & the MS/MS 

spectrum. 

In a direct infusion experiment, the 

[M+H] + ion of Attenoside at 

939.4795m/z was submitted to an 

MS/MS experiment. As the glycosidic 

bonds fragment easily, very low collision 

energies of only 10-12eV are applied. 

The combined evaluation of mass 

accuracy and isotopic pattern in MS and 

MS/MS in SmartFormula 3D provides a 

fast way of safely assigning molecular 

formulae to the precursor and 

fragments as shown in Fig. 4. The 

SmartFormula 3D results can be 

combined with structural information 

using the FragmentExplorer. The 

structure of Attenoside and possible 

resulting fragment structures are linked 

to the MS/MS spectrum. 

O H 

C 38 H 63 O 15 , -0.47 mDa 

C 38 H 65 O 16 , -0.44 mDa 

C 44 H 73 O 20 , -0.74 mDa 

H + 

C 44 H 75 O 21 , -0.44 mDa 

H2O + 0.0002 

C 

Characteristic for the fragmentation of 

the HGL-DTGs are successive neutral 

losses of the sugar moieties glucose 

(Dm= 162.0528 =C6H10O6) and 

rhamnose (Dm= 146.0579 =C6H10O5). 

The fragment at 271.242 m/z with the 

formula C20H31 represents the 

hydroxygeranylinalool backbone after 

losing the hydroxyl groups. 

This fragment in combination with the 

neutral losses of sugars and malonyl 

groups is a very good indicator for 

presence of a HGL-DTG. 

The table in Fig. 1 lists all HGL-DTGs that 

were successfully identified in a selected 

enriched Nicotiana extract by means of 

accurate mass and isotopic pattern only. 



0.8 

0.6 

0.4 

0.2 

BPC of pooled Flower Extract 

EIC 271.2420±0.001 of pooled Flower Extract 

0.0 

3 4 5 6 7 8 9 Time [min] 

Library Search Nicotiana Nicotiana Nicotiana Nicotiana x 

MS/MS 

attenuata attenuata obtusifolia sanderae 

Effective Purity leaf flower leaf leaf 

Lyciumoside II 725(MS) - - 

Lyciumoside IV 984 977 - - 

Nicotianoside I 965 926 - 551 

Nicotianoside II 983 976 963 - 

Nicotianoside III 652 - - - 

Nicotianoside IV 802 823 - 

Nicotianoside V - 934(MS) 873 - 

Attenoside 807 899 - - 

Nicotianoside VI 863 - 848 - 

Nicotianoside VII 914 968 867 - 

Fig. 5 Chromatogram of pooled flower & 

reproductive organ extracts from Nicotiana 

attenuata in ESI positive mode, BPC and 

EIC for m/z 271; Table with effective purity 

scores from MS/MS library search for 

different Nicotiana samples. 

Based on the identified HGL-DTGs an 

MS/MS library was then created. The 

identification results of different HGL- 

DTGs in extracts (no special enrichment) 

from Nicotiana using the library search 

are summarized in Fig. 5. 

Conclusions 

HGL-DTGs in complex samples can 

be recognized by 

• distinct adduct pattern of [M+H] + , 

[M+NH4] + and [M+Na] + in ESI 

positive mode 

• rich in-source fragmentation with 

neutral losses of rhamnose & 

glucose 

• backbone fragment 271.242m/z 

Deconvolution algorithms (FMF, 

Dissect) have been successfully 

applied in the early identifcation 

steps. 

Dedicated MS/MS interpretation 

incorporating structural 

assignments was performed 

leading to a well characterized 

MS/MS library. 

All presented tools for identification 

of HGL-DTGs will be applied now to 

other Solanum species. 

UHR-Qq-TOF-MS


MS-based dereplication and classification of diterpene 

glycoside profiles of 23 Solanaceous plant species 

Sven Heiling1 , Emmanuel Gaquerel1 , Aiko Barsch2 , Gabriela Zurek2 , Matthias Schöttner1 and Ian T. Baldwin1 1Max-Planck Institute for Chemical Ecology, Jena, Germany 

2Bruker Daltonics, Bremen, Germany 

HGL-DTG profile of 23 solanaceous species 

4. Results 

5. Conclusion 

3. Workflow 

1. Introduction 

HGL-DTGs found 

with malonyl- 

Species Varieties all together 

groups 

Capsicum annuum 12 11 6 

Capsicum baccatum 3 5 3 

Capsicum chinese 2 9 4 

Capsicum frutescens 2 9 5 

Datura wrightii 1 - - 

Lycium barbarum 1 15 15 

Lycium sp. 1 - - 

Lycium torreyi 1 - - 

Nicandra physaloides 1 - - 

Nicotiana alata 1 3 1 

Nicotiana attenuata 1 17 12 

Nicotiana glauca 1 - - 

Nicotiana obtusifolia 1 21 15 

Nicotiana sylvestris 1 - - 

Nicotiana tabacum 1 - - 

Nicotiana x sanderae 1 3 2 

Petunia exserta 1 - - 

Petunia graniflora 1 - - 

Petunia violacea 1 - - 

Physalis 1 - - 

Solanum burchelii 1 - - 

Solanum lycopersicum 1 - - 

Solanum tuberosum 1 - - 

+MS, Dissect, 20.2-20.6min #2404-2452 

A 

LC/MS-Chromatogram 

271.2417 

743.4218 

430.1561 

889.4797 

2.5 

1.5 

0.5 

Intens. ×10 5 

1172.5709 

597.3631 

5 10 15 20 25 30 35 40 [min] 

127.0411 

1035.5395 

1.2 

1.0 

0.8 

0.6 

0.4 

0.2 

0 

�We developed an efficient pipeline for the dereplication 

and class prediction of HGL-DTGs produced by different 

solanaceous species 

Dissect cluster 

[M+NH 4] + 

Intens.×10 4 

EIC trace 271.24 

17-hydroxygeranyllinalool diterpene glycosides 

(HGL-DTGs) are abundant secondary metabolites 

in aboveground tissue of the coyote tobacco, 

Nicotiana attenuata. These HGL-DTGs act as potent 

anti-herbivore compounds, whose biosynthetic 

steps are regulated by the defense phytohormones 

jasmonic acid. Although the strongest effect can be 

seen in malonylation of these compounds, nearly no 

malonylated HGL-DTGs are described so far. 

200 400 600 800 1000 1200 m/z 

�This pipeline is based on the extraction of high 

resolution IS-CID spectra containing the characteristic 

fragments for HGL-DTG aglycone and their classification 

based on diagnostic fragment analysis and alignment 

with known HGL-DTG tandem MS. 

4.0 

2.0 

0 

18 20 22 24 26 [min] 


+MS2(1155.5429 [M+H] + ), 12.0eV, 20.4min #1207 

DHex 

MS 

DHex 

Hex 

Ma 

DHex 

DHex Ma 

DHex Hex 

DHex Hex+Ma DHex 

2 Spectrum 

[M+H] + 

1155.5429 

3 

7 

11 

15 

16 

1 

H 2O 

B 

H2O H2O 10 

Dissect clustering 

14 

13 

18 

20 

1 2 3 5 7 8 

3.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 


R 

R1 R2 Malonyl- 

Class* Sugar 1 Sugar 2 Sugar 1 Sugar 2 groups 

Lyciumoside I 1 Glc Glc 

Lyciumoside II 1 Glc Glc Glc 

Lyciumoside IV 1 Glc Rh Glc 

Nicotianoside I 2 Glc Rh Glc 1 

Nicotianoside II 3 Glc Rh Glc 2 

Nicotianoside III 1 Glc Rh Glc Rh 

Nicotianoside IV 2 Glc Rh Glc Rh 1 

Nicotianoside V 3 Glc Rh Glc Rh 2 

Attenoside 1 Glc Rh Glc Glc 

Nicotianoside VI 2 Glc Rh Glc Glc 1 

Nicotianoside VII 3 Glc Rh Glc Glc 2 

* 1 Core 2 Malonylated 3 Dimalonylated 

2 OR 

O 

1 

17 18 19 20 

H 2O 

H 2O 

H 2O 

H 2O 

4.0 

Dissect 

Deconvolution 

H 2O 

3.0 

H 2O 

DHex 

Hex 

H 2O 

DHex 

2.0 

H 2O 

Intens.×10 4 

� Applying this annotation pipeline to the analysis of the 

metabolomics profile of 23 solanaceae identifies 

i. non-uniform HGL-DTG biosynthetic capacities in the 

four main genus analyzed and 

ii. malonylation as a highly conserved reaction. 

1.0 

0 

300 400 500 600 700 800 900 1000 1100 m/z 

� Deep annotation of tandem MS fragments supports 

that malonylation takes place on glucose moieties and 

future work will examine the biological relevance of this 

reaction. 

A) Dissect and B) MS/MS-Spectrum of Nicotiana alata. Compound 

1172 [M+NH4] + is a malonylated HGL-DTG. Structure prediction: 

2 hexose (Hex), 3 desoxyhexose (DHex) , 1 malonylgroup (Ma) 

18 20 22 24 26 [min] 

First Annotation 

We defined a set of rules deduced from typical 

neutral losses Dm (18.0106 = H2O, A) Ma = m/z 

86.0004 = C3H2O3, B) Rha = m/z 146.0579 = 

C6H10O4, C) Glc = 162.0528 = C6H10O5, D) 

266.0638 = C9H14O9) and characteristic adduct 

formation ([M+NH4] + and [M+Na] + ). 

Direct injection MS 2 of Nicotianoside II 

C H 2 

2. Methods 

m/z: 417.2999 

Accurate mass: 417.2999 

ppm: 0.00 

Formula: C + 

26H41O4 O + 

+MS2(863.4300), 10.0eV, 3.2-3.6min #(92-100) 

+MS, Dissect, 22.0-22.4min #2610-2659 

O H 

C 

H 3 

H 3C 

O 

H 

O 

H 

CH 2 

863.4278 

H 3C 

CH 2 CH3 CH3 

845.4169 

597.3635 

395.1184 

417.3000 

271.2419 

O 

B 

683.3640 

O H 

C 

H 3 

O 

H 

O 

O 

D 

O 

O 

O 

H 

827.4065 

6 

5 

4 

3 

2 

1 

0 

Dissect cluster 

(14) 

880.45 

3.0 


Glc+H 2O Glc+Ma+NH 3 

Rha 

Fragmenter 

Diagnostic fragments 

MS2 library constructed with 11 

purified and semi-purified HGL- 

DTGs at different CID voltages (10 

– 30 eV) in positive electrospray 

ion mode. We annotated 988 

fragments and found 176 nonredundant 

fragments. 

2.0 

H + 

1.0 

H O H O 

O H 

Nicotiana attenuata samples 

were prepared as described 

previously [1]. Chromatographic 

separation of extracts was 

carried out using a UHPLC 

system combined with ultrahigh-resolution 

QTOF MS 

O 

A 

O 

0 

300 400 500 600 700 800 m/z 

271.24 417.31 

451.31 

395.13 597.37 

249.06 

557.18 

845.43 

200 300 400 500 600 700 800 m/z 


O 

MS2 library / Library editor Editor 

O 

detection. 

O 

H 

C 

H O H O 

m/z: 451.3050 


ppm: 0.89 

Formula: C + 

26H43O6 O 

O 

H 

O H 

O H 

O 

H O H O 

Glc Rha 

H O H O 

MS 2 spectra 

H + 

13 

37 

20 

Annotation Smart Formula 3D 

O 

C H 2 

O 

O 

H 

O H 

60 

13 0 

Due to the presence of multiple labile glycosidic bonds, 

HGL-DTGs exhibit intensive fragmentation which 

provides information valuable for structure elucidation 

[2]. Deconvolution of IS-CID clusters was performed 

using the Dissect algorithm, molecular formulae were 

determined by SmartFormula 3D (Bruker Daltonics). 

m/z: 597.3631 


ppm: 0.33 

Formula: C + 

32H53O10 Sugar fragmentation 

O 

H O 

+MS2(863.4300), 10.0eV, 3.2-3.6min #(92-100) 

0 

O 

O 

O + 

O 

O 

H 

Glc 

Glc 

Ma 

863.4278 

Rha 

Ma 

Ma 

O H 

H 2O 

H 2O 

H 2O 

Fragmentation with aglycone 

+MS2(863.4300), 10.0eV, 3.2-3.6min #(92-100) 

Rha Glc 

Glc 

Glc Glc 

Glc 

Rha Ma 

Glc 

Ma 

C 

H 3 

O 

O 

H 

C H 2 

Z17,0 O H Y17,0 O 

O 

H 

Acknowlegdment and Reference 

O 

H O 

H 2O 

H 2O 

H 2O 

H 2O 

O 

O 

H 

Rha 

H 2O 

Thanks to Thomas Hahn for the technical support and 

the MPI and DAAD for funding. 

O 

H 

6 

5 

4 

3 

2 

1 

0 

H 2O 

H 2O 

H 2O 

H 2O 

6 

5 

4 

3 

2 

1 

0 

Z 3,0 

O 

H C 

C H 3 

3 

O C3,0 V3,6 O B3,0 O 

W0 H O 

H O 

O H 

O H 

H 2O 



Y 3,0 

C H 3 

H O 

O 

C3,1 

B3,1 O 

O 

C17,0 B17,0 H3C H O 

H O 

O 

H O 

W0 = random 

loss of water 

VX,6 = loss of 

malonyl-group 

O H 

[1] Heiling et al. (2010). The Plant Cell, 22, 273-292. 

[2] Gaquerel et al. (2010) Journal of Agricultural and 

Food Chemistry, 58, 9418-9427. 

O H 

300 400 500 600 700 800 m/z 

300 400 500 600 700 800 m/z 

Z 3,1 

Y 3,1 

Nicotianoside II 

Selected plant samples were 

fractionated by HPLC. 

Enriched fractions of individual 

HGL-DTGs were subjected to 

detailed fragmentation studies 

by means of direct infusion 

measurements on micrOTOF- 

Q II and Maxis UHR-Q-TOF- 

MS systems in electrospray 

positive ion mode. In-source 

collision induced dissociation 

(IS-CID) accurate mass full 

scan spectra as well as 

precursor and product ion 

spectra were recorded for 

structural assignment and 

confirmation. 

O H 

Venn diagram of all sugar 

containing fragments. Comparison 

shows that malonyl-moieties are 

just present, when glucose is part 

of the fragment. 

-17-

-18- 



1. Motivation 

2. Concept 

GC-APCI-QTOF-MS: An innovative Technique 

for Metabolite Profiling Studies 

Strehmel N, Boettcher C, Bernstein C, Scheel D 

Leibniz Institute of Plant Biochemistry, Department of Stress and Developmental Biology, Weinberg 3, 06120 Halle (Saale) 

GC-MS-based metabolite profiling of crude plant extracts reveals more than 100 compounds. Thereof, more than half can be identified using comprehensive mass spectral libraries [1,2] , the rest remains 

unknown. In order to identify the unknown compounds we applied GC-APCI-QTOF-MS in succession to our metabolite profiling studies since this technique allows for the determination of elemental 

compositions of quasi-molecular and fragment ions. For the mapping of spectral features of both techniques we used the retention index (based on Fatty Acid Methyl Ester) and the molecular ion as long as it 

could be derived from metabolite profiles. As proof of concept, we cocultivated Arabidopsis thaliana for two weeks with Piriformospra indica with the objective to study the metabolic phenotype of the growth 

promoting effect. 

1. Metabolite Profiling by GC-EI-QUAD-MS 2. Multivariate Statistics (ICA, ANOVA, t-test) 

Samples 

RI, m/z 

3. MST Annotation by Library Search 

3. Materials and Methods 

4. MST Annotation 

5. Mapping of unknown MSTs 

4. Structure Elucidation of unknown MSTs 

Electron Impact 

Ionization 

- Molecular Fingerprint 

Sugar Breakdown and Glycolysis Amino Acid Biosynthesis 

TCA 

Miscellaneous Metabolites 

Mass 

Spectral 

Tag 

(m/z; RI) 

Atmospheric Pressure 

Chemical Ionization 

- Elemental Composition 

- CID Mass Spectra 

GC-EI-QUAD-MS-based metabolite profiling of roots (At and AtPi) revealed 29 differential compounds (t-test: 

control vs. co-cultivated, p < 0.05), of which 15 could be identified. Thereof, 13 could be classified into 

metabolic pathways. 

Retention Index GC-APCI-QTOF-MS 

EXPERIMENTAL SPECTRUM 

LIBRARY SPECTRUM 

1. Plant Cultivation 2. Extraction of Roots 3. Derivatization 4. Measurement 

AtPi 

At 

Two week old 

hydroponically grown 

plantlets were 

inoculated with fungal 

mycelium (Piriformospora 

indica; Pi). After 

two weeks of cocultivation 

roots were 

harvested (control: At; 

cocultivated: AtPi) and 

analyzed for metabolic 

differences. 

- 30 mg fresh weight 

- 300 µL MeOH 

- 15 min at 70°C 

- 200 µL CHCl3 - 400 µL H2O - 40 µL MeOX 

- 60 µL BSTFA 

- 10 µL FAMES 

- 10 µL Alkanes 

1. Metabolite Profiling: 

GC-EI-QUAD-MS 

RI Transfer 

Molecular Ion 

2. Structure Elucidation: 

GC-APCI-QTOF-MS 

Retention Index GC-EI-QUAD-MS 

Roots of Arabidopsis thaliana (At) 

Piriformospora indica co-cultivated roots of Arabidopsis thaliana (AtPi) 

In order to identify the remaining MSTs we used the 

retention index, as this is a useful measure to map peaks 

from one chromatographic system to another [3] . Often 

alkanes are used as retention index markers. However, 

they do not ionize under atmospheric pressure chemical 

ionization conditions. Consequently, we had to switch to 

fatty acid methyl esters and estimated the retention 

index according to Fiehn [4] . 

Besides the retention index we consulted the molecular 

ion for the MST mapping as long as it was available. 

MST MAPPING 

Retention Index + Molecular Ion 

6. Elemental Composition 

Intens. 1. +MS2(594.0000), 20-20eV, 25.2-25.3min #(4469-4480) 

x105 

361.1755 APCI (+)-CID-MS (594 @20 eV) 

8 

6 

4 

2 

0 

Retention 

Index 

[FAMES] 

169.0732 

126.0424 191.0956 

144.0553 

217.1099 

150 200 250 300 350 400 450 500 550 m/z 

EI 

6. GC-APCI-QTOF-MS vs. GC-EI-QUAD-MS 

IAA (2TMS) 

ABA (1MeOX, 1TMS) 

JA (1MeOX, 1TMS) 

GA1 (3TMS) 

Fumaric acid (2TMS) 

Isocitric acid (4TMS) 

Tartaric acid (4TMS) 

7. Conclusions 

GC-APCI-QTOF-MS can complement GC-EI-MS-based metabolite profiling. Its soft 

ionisation manner, high mass accuracy and resolution make it feasible. In addition we 

could show that GC-APCI-QTOF-MS is much more sensitive for most substance 

classes relevant for metabolite profiling studies than GC-EI-MS and can not only be 

used for structure elucidation purposes of so-far unidentified compounds, but also for 

the profiling of low-abundant compounds. 

8. References 

Molecular 

Ion 

[m/z] 

243.1273 

[1] Hummel J, Strehmel N, Bölling C, Schmidt S, Walther D, Kopka J: Mass Spectral Search and 

Analysis using the Golm Metabolome Database (2012) Metabolomics and More, in press. 

[2] Horai H et al.: MassBank: a public repository for sharing mass spectral data for life sciences (2010) 

Journal of Mass Spectrometry 45(7) 703-714. 

[3] Strehmel N, Hummel J, Erban A, Strassburg K, Kopka J: Retention index thresholds for compound 

matching in GC-MS metabolite profiling (2008) Journal of Chromatography B 871(2) 182-90. 

[4] Kind T, Wohlgemuth G, Lee DY, Lu Y, Palazoglu M, Shahbaz S, Fiehn O: FiehnLib: Mass Spectral 

and Retention Index Libraries for Metabolomics Based on Quadrupole and Time-of-Flight Gas 

Chromatography/Mass Spectrometry (2009) Analytical Chemistry 81 (24) 10038-10048. 

9. Acknowledgement 

C 11H23O2Si2 + C 13H31O3Si3 + C 14H31O3Si3 + 

271.1234 

Differential 

Analytes 

319.1632 

331.1636 

Putrescine (4TMS) 

Spermidine (5TMS) 

Glucose (1MeOX, 5TMS) 

Ribitol (5TMS) 

Maltose (1MeOX, 8TMS) 

myo Inositol (6TMS) 

Analyte 

Composition 

Metabolite 

Composition 

1016361 530 At > AtPi C 28H 42O 6Si 2 C 22H 26O 6 

726948 374 At > AtPi C 13H 30N 4O 3Si 3 C 4H 6N 4O 3 

644769 324 At > AtPi C 15H 24O 4Si 2 C 9H 8O 4 

501849 329 At > AtPi C 13H 27N 3OSSi 2 C 7H 11N 3OS 

905598 593 At < AtPi C 24H 51NO 6SSi 4 C 12H 19NO 6S 

476838 232 At < AtPi C 9H 20O 3Si 2 C 3H 4O 3 

Thymine (2TMS) 

Kinetin (2TMS) 

100 µL standard stock solution 

(39 nM – 20 µM) were dried 

down and derivatized 

according to the routine 

protocol. GC-EI-QUAD-MS 

resulted in lower limits of 

detection (10 µM – 100 µM) 

than GC-APCI-QTOF-MS, i.e. 

GC-APCI providing the better 

sensitivity. 

The authors thank Jessica Thomas for the preparation of extracts and acknowledge 

Aiko Barsch as well as Thomas Arthen-Engeland from Bruker Daltonics for technical 

support during the setup of the system.


Improved method for characterization of 

phenolic compounds from rosemary extracts 

using ESI-Qq-TOF MS 

Isabel Borrás-Linares 1 , Gabriela Zurek 2 , David 

Arráez-Román 1 , Antonio Segura-Carretero 1 , 

Alberto Fernández-Gutiérrez 1 

1 CIDAF, Granada, Spain 

2 Bruker Daltonik, Bremen, Germany 

Introduction 

Rosemary (Rosmarinus officinalis L.) is a shrub 

traditionally used as culinary spice, as well as in folk 

medicine, being a greatly valued medicinal herb. In 

fact, this plant exerts a great number of 

pharmacological activities: hepatoprotective, 

antibacterial, antithrombotic, antiulcerogenic, diuretic, 

antidiabetic, antinociceptive, anti-inflammatory, 

antitumor and antioxidant activities. Most of these 

observed effects are linked to its phenolic content. The 

characterization of phenolic compounds (Fig. 1) in the 

Rosemary extracts is an important step towards a 

better understanding of the pharmacological action. 

Recently there has been a growing interest in the use 

of natural antioxidants in the food industry as a 

preservation method. Due to its high antioxidant 

capacity, rosemary turned out being one of the most 

used natural antioxidant because of its value for 

preservation and its beneficial effects mentioned 

above. 

Carnosic 

acid 

Rosmanol 

Fig. 1 Structures of selected phenolic diterpenes from 

Rosemary. 

Methods 

Two rosemary extracts obtained using supercritical 

fluid extraction (SFE, CO2 at 150 bar) and pressurized 

liquid extraction (PLE, H2O at 200°C) were dissolved in 

ethanol at 10 mg/mL. Chromatographic separation was 

performed on an Ultimate 3000 UHPLC with a Kinetex 

C18 column (2x100mm, 2.6µm particles) using as 

mobile phase A 0.1% aqueous formic acid and as 

mobile phase B acetonitrile. A 30 min linear gradient 

from 5 to 100 % solvent B at a flow rate of 0.5 ml/min 

was used. Extracts were analyzed in ESI positive and 

negative mode using a maXis impact ESI-Qq-TOF 

system (Bruker Daltonik) in the mass range 50- 

1000m/z operated at 3Hz, in MS full scan and MS/MS 

mode. 

Compounds were characterized by taking into account 

retention time, elemental formulae derived from 

accurate mass and isotopic pattern information 

(calculated in the DataAnalysis software) as well as 

both in-house and public databases (SciFinder, 

ChemSpider, MassBank) and literature. 

Results 

Carnosol 

This work aims for the comprehensive characterization 

of bioactive secondary metabolites present in two 

rosemary extracts obtained by SFE and PLE. Both 

extracts were analyzed in positive and negative 

ionization modes in order to obtain a complete picture. 

Compared to a previous study [1], the 

chromatographic runtime was shortened to 30 min 

without significant loss of chromatographic resolution. 

The four resulting base peak chromatograms (BPCs) 

after spectral background subtraction are depicted in 

Fig. 2. The elution profiles indicate that the PLE 

extraction method yields compounds of higher polarity 

compared to the SFE extraction. 

Target compounds were tentatively identified taking 

into account retention time, accurate mass and 

isotopic pattern (yielding the molecular formula) and 

subsequent database and literature searches. In this 

particular work, the data was compared to the 

previous characterization of rosemary metabolite 

extracts by HPLC-DAD-ESI-TOF-MS [1]. 


Extract Polarity 

The presented methodology has proven to be a useful 

tool for the separation of a wide range of phenolic 

compounds in rosemary extracts: phenolic diterpenes 

(carnosol, carnosic acid, rosmanol and its isomers 

epirosmanol /epiisorosmanol), flavonoids (apigenin, 

genkwanin, ladanein, scutellarein, cirsiliol), phenolic 

acids (artepillin C), among others classes of phenolic 

compounds such as rosmarinic acid or 

glycoside/glucuronide conjugates of flavonoids 

(nepitrin, hesperidin, luteolin-7-glucuronide). Due to 

the higher sensitivity of the ESI-Qq-TOF system, we 

were able to find new compounds which were not 

detected previously in these extracts such as cirsiliol, 5methoxy-6,7-methylenedioxyflavone,12-Omethylcarnosol, 

12-O-methylcarnosic acid. Table 1 

presents a summary how many compounds have been 

detected in the different extract types as well as the 

mass accuracy and mSigma values obtained. mSigma 

values provide a measure for the goodness of the 

isotopic fit (ranging from 0-1000, the lower the value 

the better the fit), Fig. 3 shows the mass spectrum of 

scutellarein with its simulated spectrum as an example. 

In total, 51 phenolic compounds were tentatively 

identified. Table 2 presents the identification results in 

ESI negative mode as a representative excerpt for all 

characterized metabolites. 

Tab. 1 Summary of identification results. 

Number of identified 

phenolic compounds 

Average mass 

accuracy [ppm] 

Average 

mSigma 

SFE positive 13 -0.6 7.7 

SFE negative 15 -0.86 13.9 

PLE positive 15 -0.11 10.3 

PLE negative 34 -0.52 9.7 

Tab. 2 Excerpt of identification results from 

ESI negative mode. 

RT 

Ion Formula 

Meas. m/z 

[min] [M-H] - 

SFE negative mode 

SFE positive mode 

Fig. 2 Base peak chromatograms (BPC) of SFE and PLE 

Rosemary extracts in electrospray positive and negative 

mode. 

m/z 

err 

mSigma Extract Proposed compound 

[ppm] 

... 

8,1 269,046 C15H9O5 269,0455 -1,8 3,6 SFE /PLE Apigenin 

8,2 299,056 C16H11O6 299,0561 0,4 15,9 PLE Diosmetin / Hispidulin 

8,4 329,0671 C17H13O7 329,0667 -1,3 4,3 SFE Cirsiliol 

9,4 345,1711 C20H25O5 345,1707 -1 4,3 SFE/PLE Rosmanol 

9,5 299,0561 C16H11O6 299,0561 0,2 10,7 PLE Diosmetin / Hispidulin 

9,6 313,0715 C17H13O6 313,0718 0,7 8,2 SFE/PLE Cirsimaritin 

9,6 313,0721 C17H13O6 313,0718 -1,1 12,3 SFE/PLE Ladanein 

9,9 345,1705 C20H25O5 345,1707 0,7 9,5 SFE/PLE Epiisorosmanol 

10,4 345,171 C20H25O5 345,1707 -0,8 31,5 SFE/PLE Epirosmanol 

10,8 283,0619 C16H11O5 283,0612 -1,4 4,7 SFE/PLE Genkwanin 

12,7 319,192 C19H27O4 319,1915 -1,5 10,6 SFE Ubiquinol 10 

13 299,1654 C19H23O3 299,1653 -0,4 14,5 SFE Artepillin C 

13,3 329,1761 C20H25O4 329,1758 -0,7 2,9 SFE/PLE Carnosol 

14 343,1557 C20H23O5 343,1551 -1,8 10,8 SFE Rosmadial 

14,3 373,2015 C22H29O5 373,202 1,6 8,9 SFE 7-Ethoxyrosmanol 

14,9 331,1918 C20H27O4 331,1915 -1,1 22 SFE/PLE Carnosic Acid 

PLE negative mode 

PLE positive mode 

maXis 

Impact 

References: 

[1] Borrás-Linares et al, J. 

Chromatography A, 1218 (2011) 7682- 

7690. 

[2] http://msbi.ipb-halle.de/MetFrag/ 

Intens. 

21. -MS, 7.0-7.3min #1242-1294, -Peak Bkgrnd 

5 

x10 

285.0407 

1.5 

Mass Accuracy -0.7ppm 

Isotopic Pattern Fit 7.1mSigma 

1.0 

Resolution 21926 

0.5 

286.0439 

0.0 

5 

x10 

21. 

1- 

C15H9O6, -285.0405 

285.0405 

1.5 

Scutellarein 

1.0 

0.5 

0.0 

286.0438 

284 285 286 287 288 289 290 m/z 

Fig. 3 Measured and simulated mass spectrum of 

Scutellarein in ESI negative mode. 

For further characterization MS/MS experiments were 

carried out with a special focus on the negative 

ionization mode. Some general observations could be 

made: at medium collision energies of approx. 20eV 

most fragmentation occurs in the periphery of the 

phenolic compounds. The phenolic diterpenes, such as 

rosmanol, primarily exhibit a neutral loss of CO 2 from 

the carboxylic acid group and H 2O. SmartFormula 3D 

allows for the combined evaluation of MS and MS/MS 

spectra and supports the direct link to the in-silico 

fragmentation in MetFrag [2]. Out of 815 hits in 

PubChem for the elemental composition C 20H 26O 5, 

rosmanol was returned as the 3. hit (Fig. 4) . 

Intens. 

5 

x10 

0.8 

0.6 

0.4 

7. 

0.2 

182.9888 

0.0 7. 

5 

x10 

0.8 

0.6 

0.4 

0.2 

0.0 

3. hit 

Rosmanol 

221.1191 248.9601 

283.1711 

283.1707 

Conclusions 

• The novelty of this work is the 

development of a faster and more 

sensitive method for characterization 

of rosemary phenolic extracts 

compared to previous studies. 

• The presented methodology has 

proven to be a useful tool for the 

separation of a wide range of 

phenolic compounds in rosemary 

extracts: phenolic diterpenes, 

flavonoids, phenolic acids, among 

other classes of polyphenols. 

• Due to the higher sensitivity of the 

novel ESI-Qq-TOF system used, we 

were able to find and tentatively 

identify new compounds which were 

not detected previously in these 

extracts. 

UHR-QqTOF 

313.0725 

-MS, 9.4-9.5min #1117-1135 

345.1712 

-MS2(345.1712), 22.4eV, 9.4-9.5min #1118-1137 

301.1814 

345.1708 

180 200 220 240 260 280 300 320 340 m/z 

815hits in PubChem 

Fig. 4 MS and MS/MS spectrum of Rosmanol in ESI 

negative mode with SmartFormula 3D and MetFrag 

result. 

-19-

-20- 



Sulfur-containing metabolite-targeted analysis using 

liquid chromatography-mass spectrometry 

Ryo Nakabayashi 1 , Yuji Sawada 1 , Makoto Suzuki 1 , Masami Yokota Hirai 1 , 

Noriya Masamura 2 , Masayoshi Shigyo 3 and Kazuki Saito 1,4 

1 RIKEN Plant Science Center, 2 Somatech Center, House Foods Corporation, 3 Faculty of Agriculture, 

Yamaguchi University, 4 Graduate School of Pharmaceutical Sciences, Chiba University 

P 

lants accumulate human beneficial metabolites such as amino acids, 

flavonoids and sulfur-containing metabolites (S-metabolites) with a wide 

range of biological activities related to antioxidation, platelet aggregation 

inhibition, and anticancer benefit. Profiling methods of S-metabolites with 

high accuracy and throughput are being required for efficient 

phytochemical functional genomics and crop breeding. Here, we describe 

the S-metabolite-targeted analysis using liquid chromatography (LC)-fourier 

transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) with U- 

13C-labeled onion bulb. Natural abundance of 32S (31.972072 Da) and its 

isotope 34S (33.967868 Da) is 95.02 % and 4.21 %, respectively. These facts 

are obviously reflected on molecular ion peaks of S-metabolites. Thus, the 

use of such information allows us to efficiently and accurately extract the 

peaks related to S-metabolites. By using this method, finally, Salk(en)ylcysteine 

sulfoxides together with their intermediates were 

chemically annotated in onion bulb. It means to realize S-omics using LC- 

FT-MS for screening of biomarkers in any other organisms. 

Signal intensity 

Monoisotopic ion M�1 ion M�2 ion 

12 Cv 1 Hw 14 Nx 16 Oy 32 Sz 

1 

2 

Theoretical ion peak pattern 

0.99703 Da 

15 N-substituted 

1.99579 Da 

m/z 

0.0063 Da 

13 C- 

0.0084 Da 

0.0024 Da 

34 S- 18O- 13C�2- 

Signal intensity 

BPI 

Extraction of S-peaks 

min 

Summary 

Health-promoting crops 

Amino acids 

Vitamines 

Lipids 

Terpenoids 

Phenylpropanoids 

Flavonoids 

Sulfur-metabolites 

Alkaloids 

Genomics 

Transcriptomics 

Next generation sequencer etc 

ATGC 

Integration 

Metabolomics 

LC-MS etc 

High throughput analysis 

Highly accurate analysis 

FT-ICR-MS (Solarix 7.0 T) 

Dual ion funnel 

Phytochemical genomics 

Crop breeding 

Diverse population of 

plants/chromosomes 

Profiling of S-metabolites using LC-FT-MS in non-labeled and 13 C-labeled onion bulb 

EICs of 6 representative S-metabolites 

3 

0 2 4 6 8 10 12 14 

min 

S-1-propenyl mercaptoglutathione (5) 

m/z 380.09438; resolution 330,000 

S-peak picking 

C0-50H0-100N0-5O0-50S0-5 < 1 mDa 

5 elemental compositions 

13 C-labeled 

1 elemental composition 

C13H22N3O6S2 

Comparison of 

peak pattern 

Determination of 

elemental composition 

C13H21N3O6S2 

4 

5 

6 

Characterization 

MS/MS analysis 

S-trans-1-prenylcysteine sulfoxide 

(PRENCSO) 

1 

�-glutamyl-S-1-propenylcysteine 

4 

13 C-labeled Non-labeled 

Non-labeled 

Theoretical 

Image of picking S-metabolite peaks 

10.20 

10.20 

min 

380.09438 

380.09445 

380.09438 

381.09773 

382.09018 

M M+1 

Peak picking of S-metabolite 

EIC MS 

381.09132 

381.09149 

381.09773 

m/z 

C13 

m/z 

�-glutamyl-S-1-carboxy 

propylcysteine sulfoxide 

2 

S-1-propenyl mercaptoglutathione 

5 

392.13473 

392.13056 

M+2 

393.13805 

382.09018 

382.09849 

382.10100 

381.09781 382.09025 

13C 382.09870 

34S 382.10116 

15N 13 

18 C�2 

O 

13 Simulated Non-labeled 

C-labeled Non-labeled 

125.52647 

C4 

129.53986 

125.52647 

C4H2N1O2S1 

127.98008 

C5 

177.03274 

188.01982 

C7 

182.04983 

195.04333 

177.03274 

188.01982 

217.06474 

MS/MS 

234.02540 

251.05195 

C8 

C8 

242.05225 

259.07880 

m/z 

MS/MS 

234.02540 

251.05195 

Q 

S-2-carboxypropylglutathione 

3 

S-1-propyl mercaptoglutathione 

6 

287.05208 

305.06266 

C11 C11 

298.08899 

287.05208 

316.09941 

C5H9N2O3S1 

177.03282 

C7H10N1O1S2 

188.01983 

C8H13N2O3S1 

217.06414 

C8H12N1O3S2 

234.02531 

C8H15N2O3S2 

251.05186 

C11H15N2O3S2 

287.05186 

C11H17N2O4S2 

305.06242 

m/z 

380.09481 

C13 

C5H7N1O3 

C5H10N2O3 

C6H12N2O5 

C8H13N1O3S1 

C9H20N2O4S1 

C5H9N1O3S1 

C2H7N1O3 

C2H5N1O2 

305.06266 

393.13853 

380.09481 

C13H22N3O6S2 

380.09445 

Q 

mQTL 

Selection of potential inbred 

parents of higher-yielding hybrids 

Crossing of inbreds to make 

hybrids 

Selection and evaluation of 

heteroic hybrids for commerce 

Hexapol ion transfer 

Magnet 

Magnet 

ICR cell 

13C-labeled onion bulb was purchased from 

SI Science (http://www.si-science.co.jp/) 

Conclusion 

�LC-FT-MS is a strong tool for extraction 

and characterization of S-metabolites. 

�Onion bulb includes quite rare Smetabolites 

which have strong antiinflammatory 

activity. 

�Strategy of biomarker screening will be 

changed by this method with approaches 

of natural products chemistry 

Acknowledgements 

�Strategic International Collaborative 

Research Program (SICORP) for JP-NZ 

and JP-US, JST 

�Japan Advanced Plant Science Network

Invertebrate Metabolomics: 


Metabolic Characterization of Sod1 null mutant flies 

using Liquid Chromatography/Mass Spectrometry 

Jose M. Knee 1 , Teresa 1 Z. 

Rzezniczak 1 , Kevin Kun Guo 2 , Aiko 

Barsch 3 & Thomas J. S. Merritt 1, 

1 Department of Chemistry & Biochemistry 

Laurentian University Sudbury, Ontario, Canada 

tmerritt@laurentian.ca 

2 Bruker Daltonics Inc., Billerica, MA 

3 Bruker Daltonics Inc., Bremen, Germany 

Introduction 

We are characterizing the metabolomic impact of knocking out 

(KO) cytosolic Superoxide dismutase (Sod1) in Drosophila 

melanogaster using the Sod1 KO allele cSOD n108 . Sod1 is 

responsible for scavenging the O -. ion, a highly reactive chemical 

that contributes to oxidative stress (1) . cSOD n108 mutants have a 

variety of phenotypes including an accumulation of oxidative 

damage, decreased lifespan, male infertility and an overall 

enfeebled phenotype (2) . Here we describe our initial steps to 

use a liquid chromatography coupled mass spectrometry (LC- 

MS) based approach to characterize the broad metabolomic 

impact of this KO. 

Methods 

Adult Sod1 null (cSOD n108 ) and transgenic rescue control 

Drosophila melanogaster were homogenized in cold methanol 

and protein and debris were removed by centrifugation. 

Supernatant was brought to approximately 80% water before 

injection into a micrOTOF QII mass spectrometer (Bruker 

Daltonics) with an ESI source coupled to a UltiMate 3000 rapid 

separation liquid chromatography system (Dionex). The LC was 

run with diamylammonium acetate at pH 4.95 using a watermethanol 

multi-step gradient on a Kinetex C18 RP 100x2.1mm, 

1.7µm column (Phenomenex). The MS was run under negative 

ionization mode. Chromatograms were analyzed using 

DataAnalysis software (Bruker Daltonics). Principal component 

analysis (PCA) and t-tests were performed with ProfileAnalysis 

software (Bruker Daltonics). 

Results 

• Polar molecules are notoriously difficult to separate with C-18 

reversed phase columns and an alternative, hydrophilic 

interaction chromatography (HILIC), generates less than 

optimal peak shapes and reproducibility (3) . We have recently 

focused on establishing a C-18 reversed phase column based 

separation with a basic ion pairing reagent to improve 

resolution. Fig 1 and Table 1 show a series of polar standards 

we are currently able to resolve. Polar basic compounds are 

still poorly retained. 

• LC-MS raw data from fly samples was subject to peak finding 

using the FindMolecularFeatures (FMF) algorithm which can 

efficiently differentiate real signals and background noise. 

FMF features from all samples were assigned to buckets, a 

combination of time and m/z, in a bucket table as basis for 

PCA or t-test analysis. 

• PCA loadings plot (Fig 2 A) and bucket statistics (Fig 2 D) 

reveal how the FMF compounds contribute to the variation 

between groups. Each point in the loadings plot represents a 

bucket and the bucket statistics displays the levels at which 

each sample expresses the bucket. 

• PCA scores plot (Fig 2 B: PC1 vs PC2) demonstrates separate 

grouping between the two fly sample types. As expected, 

genetic differences between the groups results in 

downstream alterations of the metabolic profile. 

A 

B 

Figure 1: LC-MS chromatograms of A) 0.05mM prepared standards B) Fly 

homogenate with extracted chromatograms of targeted compounds 

A 

B 

C 

ATP 

Control 

ADP 

Prepared Standards 

Fly Homogenate 

Sod1-null 

Control 

1.28 fold 

Change 

p=0.002 

Sod-null 

[M-H] - = C6H9N3O2 Histidine 

Figure 2: PCA analysis of cSODn108 flies and controls (A) 

Loadings plot (B) Scores plot (C) Chromatograms and spectrum 

of differentially expressed metabolite, Histidine (D) Bucket 

statistics of histidine across all samples (E) Histidine structure 

and elemental formula 

• Fig 2 C and D represent information of the loading 

highlighted in green in Fig 2 A. Fig 2 C shows the extracted 

ion chromatogram (EIC) of bucket (0.8min; 154.06m/z) 

and the mass spectrum. On average there is a 1.28 fold 

decrease of this compound in Sod1-null flies. Fig 2 E 

presents the identity of the bucket at 0.8min; 154.06m/z: 

histidine as revealed in comparison to the standard. 

• t-test results generated 30 FMF compounds differentially 

expressed with p≤0.05 between the cSODn108 flies and the 

controls and 15 differentially expressed with p≤0.01. There 

were a total of 336 detected compounds. The first 11 most 

varying features are represented in Table 2. 

• Analysis of an unknown bucket is demonstrated in Figure 3. 

An EIC and spectrum of bucket 0.8 min; 376.10 m/z is 

depicted in Fig 3 A and Fig 3 B, respectively. The 

SmartFormula feature uses accurate mass measurement, 

isotopic pattern, and, potentially, fragmentation pattern to 

generate a candidate formula. Fig 3 C reveals the 

SmartFormula results for the selected bucket ranked 

according to the best isotopic fit. Unknown metabolites can 

then be identified using the CompoundCrawler software, 

which searches online metabolite databases such as 

KEGG (4,5) Time (min) 

for possible structures. 

(m/z) 

D 

E 

D 

Control 

Loadings 

Scores 

Sod-null 

Table 1: Targeted Standards. ( * ) denotes standards not 

detected in our fly samples 

Bucket P-Value "Sodnull/Control” 

Fold Change 

“Sodnull /control” 

0.8min : 376.10m/z 0.00008 0.53 -1.88 

5.3min : 421.07m/z 0.00026 1.62 1.62 

23.4min : 354.22m/z 0.00047 2.18 2.18 

23.7min : 356.24m/z 0.00077 2.46 2.46 

0.8min : 152.08m/z 0.00085 0.64 -1.57 

0.8min : 718.21m/z 0.00123 0.58 -1.74 

1.2min : 168.04m/z 0.00127 0.71 -1.4 

23.6min : 362.24m/z 0.00158 1.81 1.81 

0.8min : 154.06m/z 0.00232 0.78 -1.28 

23.1min : 460.92m/z 0.00236 0.74 -1.35 

7.5min : 393.04m/z 0.00312 0.72 -1.39 

Table 2: MS t-Test results of Sod-null flies vs. controls of the 

first 11 significant metabolites. Blue highlighted metabolite 

was identified as Histidine. Red highlighted metabolite is 

shown as example in figure 3. 

A 

1 Arginine 13 valine 25 Fructose 6-phosphate 35 α-ketoglutaric acid 

2 Ornithine 14 methionine 25 Mannose 6-phosphate 36 Fumarate 

3 Lysine 15 tyrosine 25 Glucose 1-phosphate 37 6-phosphogluconoic acid * 

4 Dopamine* 16 leucine 26 Pyroglutamic acid 38 Fructose 1,6-biphosphate 

5 Trehalose 17 isoleucine 27 Glycerol 1-phosphate 39 Citric acid 

6 Asparagine 18 phenylalanine 28 Glutathione (red) 39 

B 

(m/z) 

C 

Figure 3: Analysis of Bucket 0.8 min; 376.10 m/z a) Extracted 

ion chromatogram B) Mass spectra of unknown compound C) 

SmartFormula result 

Conclusions 

Isocitric acid 

7 Glutamine 19 Glutamic acid 29 β-glycerophosphate * 40 Phosphoenolpyruvate * 

8 Histidine 20 Aspartic acid 30 NAD + 41 NADP + * 

9 Threonine 21 Uric acid 31 AMP 42 ADP 

10 Serine 22 

Tryptophan 

32 Succinic Acid 43 NADH 

11 Proline 23 Ascorbic Acid 33 Malic acid 44 NADPH 

12 Cysteine 24 Glucose 6-phosphate 34 Glutathione (ox) 45 ATP 

Control 

Sod-null 

These are preliminary results from LC-MS analysis aimed 

at characterizing the metabolic impact of Sod-null 

genotypes and oxidative stress in D. melanogaster . 

This poster also demonstrates the applicability of LC-MS 

analysis for D. melanogaster metabolomics. 

The future directions, outlined below, will allow us to 

more completely characterize the SOD1 metabolic 

fingerprint and to move into a broader suite of metabolites 

and genes of interest. 

• Acquire MS/MS on all known standards to confirm 

compound identity in fly samples 

• MS/MS on unknown compounds to identify structure 

and compare to metabolite databases 

• Refining metabolite extraction procedures to detect 

currently undetected compounds 

• Optimize resolution and analysis of basic analytes 

• Analysis of relatively non-polar compounds 

• Pathway reconstruction and biological significance 

References 

(1) Zelko, Mariani, & Folz. (2002). Free Radic Biol Med 33 (3), 337-349. 

(2) Phillips, Campbell, Michaud, et al. (1989). Proc. Natl. Acad. Sci. 86, 

2761-2765. 

(3) Luo, Groenke, Takors, et al. (2007). J. Chromatogr. A. 1147, 153- 164. 

(4) Kanehisa, Sato, Furumichi, & Tanabe. (2012). Nucleic Acids Res. 40, D109- 

D114. 

(5) Kanehisa & Goto. (2000). Nucleic Acids Res. 28, 27-30. 

-21-

-22- 




introduction 

workflow instrumentation 

conclusion 

Analytical BioGeoChemistry 

A mass spectrometry based metabolomic platform 

for the analysis of Caenorhabditis elegans 

Witting Michael 1,2 , Ruf Alexander 1 , Garvis Steve 3 , Wägele Brigitte 2,4 , Voulhoux Romé 3 , 

Schmitt-Kopplin Philippe 1,5 

michael.witting@helmholtz-muenchen.de 

schmitt-kopplin@helmholtz-muenchen.de 

Caenorhabditis elegans is a widely used model organism, that was introduced in the 1960s by 

Sydney Brenner. Its sequenced genome, the easy cultivation, the short reproduction cycle and 

transparency makes it an ideal model organism. 

It is routinely used for studying host-pathogen interactions, neurobiology, physiology, ecology 

and many more. Despite its popularity, surprisingly few metabolomics studies, mostly using NMR 

have been carried out using C. elegans [1]. Here we present the setup of a MS based 

metabolomic platform for C. elegans studies. 

UPLC-MS: 

Waters Acquity UPLC 

Bruker maXis UHR-ToF-MS 

Resolution > 50,000 

error < 2 ppm 

Advion NanoMate for Chip-ESI 

and fraction collection 

Sample preparation: 

Metabolite extraction from worm pellets (~1000 worms) 

with 50% MeOH at -80°C 

Non-targeted metabolomics using UPLC-MS: 

RP method for non- and mid-polar metabolites: 

Acquity BEH C8, 1.7 µm particles, 150 x 2.1 mm 

A: 10% MeOH + 0.1% formic acid 

B: 100 % MeOH + 0.1% formic acid 

HILIC method for polar metabolites: 

Acquity BEH Amide, 1.7 µm particles, 150 x 2.1 mm 

A: 95 % ACN + 10 mM ammonium acetate + 0.1 % formic acid 

B: 50 % ACN + 10 mM ammonium acetate + 0.1 % formic acid 

Measurements in positive and negative mode 

Calibration of individual UPLC-MS runs through injection of 

standard mix via 6-port valve at the beginning of each run and 

lock mass calibration 

Data pre-processing I: 

Automated internal recalibration using standard mix and 

lock mass and export to net-CDF format using Bruker 

Data Analysis 4.0 and VB scripting 

ICR-FT-MS: 

Bruker solariX with 12 T magnet 

Resolution > 300,000 

error < 100 ppb 

Gilson 223 sample changer, 

6-port valve and Agilent 1100 

quad pump for automated 

sample infusion 

Non-targeted metabolomics using 

ICR-FT-MS: 

Dilution of sample, measurements in 

positive and negative mode 

1 Department of BioGeoChemistry and Analytics, Helmholtz Zentrum München, German Research Center for Environmental Health, Ingolstaedter Landstrasse 1, 85764 Neuherberg, Germany 

2 Department of Genome Oriented Bioinformatics, Life and Food Science Center Weihenstephan, Technische Universität München, 85354 Freising Weihenstephan, Germany 

3 Centre National de la Recherche Scientifique, 31 Chemin Joseph Aiguier, 13402 Marseille Cedex 20, France 

4 Institute of Bioinformatics and Systems biology, Helmholtz Zentrum München, German Research Center for Environmental Health, Ingolstaedter Landstrasse 1, 85764 Neuherberg, Germany 

5 Chair of Analytical Food Chemistry, Life and Food Science Center Weihenstephan, Technische Universität München, 85354 Freising Weihenstephan, Germany 

Intens. 

x108 

8 

6 

4 

2 

172.09437 

212.56868 

261.13090 

300.35351 

393.29757 

500.27451 589.47997 

442.30089 

630.87865 

674.40266 

745.46711 

Size: ~ 1 mm x 65 µm 

Cells: 959 (hermaphrodite) 

1031 (male) 

Lifespan: 2-3 weeks 

Food: bacteria 

803.09675 

834.54253 

960.78217 

0 

200 300 400 500 600 700 800 900 m/z 

Cel Ecoli OP50 fed 2M 300s pos 001_000001.d: +MS 

Data pre-processing II: 

Raw data import into MZmine 2.1 [4], 

chromatogram deconvolution, peak picking, 

identification of isotopes and adducts, alignment 

and database searching for metabolite annotation 

(All steps can be automated by the MZmine batch 

function) 

Further development: 

• Development of metabolite extraction method using 96-well plate format and “fast” 

UPLC-MS for high throughput screening and deep phenotyping 

• Identification of novel compounds using a LC-SPE-NMR-MS system with a 800 MHz 

NMR (Bruker Metabolic Profiler) 

Data analysis / Bioinformatics: 

Internal spectra recalibration, masslist 

export and automated metabolite 

annotation using MassTRIX server 

[2,3] 

Multivariate statistic / Data 

analysis 

Unsupervised methods (PCA and 

HCA) and supervised methods 

(PLS) for pattern recognition and 

identification of masses 

contributing to groups separation 

Comparison and Combination of 

ICR-FT-MS and UPLC-MS 

results 

Collection and identification of 

unknown peaks using NanoMate 

based fraction collection, exact 

mass from ICR-FT-MS, MS/MS 

experiments and NMR 

• Mass difference networking [5] 

Our presented MS based metabolomic platform allows the measurement of both, polar and non-polar, metabolites in one platform. The developed 

methods are currently applied to different projects using C. elegans as model organism in the fields of host-pathogen interactions, nutritional studies, 

probiotic research and the analysis of knock-out mutants. 

Ongoing developments in sample preparation and chromatographic methods, in combination with the implemented automated data pre-processing will 

allow us to use this platform for high throughput metabolome analysis of bigger sample amounts and studies. 

Literature: 

[1] Blaise BJ, Giacomotto J, Elena Bnd, Dumas M-E, Toulhoat P, et al. (2007) Metabotyping of Caenorhabditis elegans reveals latent phenotypes. Proceedings of the National Academy of Sciences 104: 19808-19812. 

[2] Suhre K, Schmitt-Kopplin P (2008) MassTRIX: mass translator into pathways. Nucleic Acids Research 36: W481-484. 

[3] Wägele B, Witting M, Schitt-Kopplin P, Suhre K (2012) MassTRIX Reloaded: Combinded Analysis and Visualization of Transcriptome and Metabolome Data. PLoS ONE 7(7). doi:10.1371/journal.pone.0039860 

[4] Pluskal T, Castillo S, Villar-Briones A, Oresic M (2010) MZmine 2: Modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinformatics 11: 395. 

[5] Tziotis D, Hertkorn N, Schmitt-Kopplin P (2011) Kendrick-analogous network visualisation of ion cyclotron resonance Fourier transform mass spectra: improved options for the assignment of elemental compositions and the classification of 

organic molecular complexity. European Journal of Mass Spectrometry 17(4):415-421



Complementary GC-EI-MS and GC-APCI-TOF-MS 

analysis of a short-lived mitochondrial knockdown 

of C. elegans to study Metabolic changes during aging 

Verena Tellström 1 , Aiko Barsch 1 , Gabriela Zurek 1 , 

Carsten Jaeger 1 , Bernd Kammerer 1 

1 Bruker Daltonik GmbH, Bremen, GERMANY 

2 Albert-Ludwigs-Universität Freiburg, Freiburg, 

GERMANY 

Introduction 

Non-targeted GC-EI-MS profiling is a core 

metabolomics technique for studying central 

metabolism in a wide range of organisms. However, 

many detected metabolites cannot be identified and 

unknown identification is mostly impossible due to 

lack of molecular ion information. By contrast, GC- 

APCI-TOF-MS readily yields pseudomolecular ions, 

facilitating generation of molecular formulae and 

thereby partial or complete compound identification. 

As a test case, we studied primary metabolism in the 

nematode worm Caenorhabditis elegans, a popular 

model organism for aging processes and mitochondrial 

dysfunction. Profiling with GC-EI-MS revealed 

biochemical differences between control and 

mitochondrially deficient animals, but many 

metabolites including important ones for phenotype 

differentiation could not be identified. We therefore 

compared GC-APCI-TOF-MS to GC-EI-MS profiles of 

C. elegans with respect to unknown identification and 

also phenotype differentiation. 

Methods 

For metabolite profiling, worms were homogenized in cold 

methanol/water by bead beating and extracted metabolites were 

derivatized with methoxyamine/pyridine and MSTFA (Fig. 1). 1-µL 

aliquots of the reaction mixture were injected into GC-EI-MS and 

GC-APCI-TOF-MS instruments (Fig. 4), respectively. Identical 

column type and gas chromatographic conditions were used to 

ensure the same compound elution order. 

Peak identification of EI data was achieved by mass spectral and 

retention index matching using AMDIS 2.68 and NIST05/Golm 

Metabolite Library (GMD: http://gmd.mpimpgolm.mpg.de/Default.aspx) 

as databases. High resolution TOF-MS 

data from micrOTOF-Q II (Bruker Daltonik) was processed with 

Bruker DataAnalysis 4.0 software. Molecular formulae were 

derived from accurate mass and isotopic pattern using the 

SmartFormula algorithm. Corresponding structures were searched 

in public databases (PubChem, KEGG, Chemspider) with Bruker 

CompoundCrawler. 

C. elegans worms were maintained on NGM plates seeded with E. 

coli OP50 as food source. To study the mitochondrial phenotype, 

animals were subjected to RNAi-mediated knockdown of 

prohibitin, a 1 MDa protein complex acting as a molecular 

chaperone at the respiratory chain. Depletion of this protein 

affects mitochondrial function and reduces lifespan. Changes in 

metabolite abundances after treatment were compared against 

non-treated control animals. 

Results 

Initial profiling with GC-EI-MS yielded 212 

reconstructed mass spectra, of which 134 could be 

assigned structures at a match factor of ≥800. To 

evaluate GC-APCI-TOF-MS as identification platform, 

we tested if EI matched metabolites could also be 

found in TOF-MS chromatograms. Searching for the 

respective mass traces, successful molecular formula 

confirmation was possible for 57 out of 60 test 

candidates, with measured m/z values deviating in 

average 1.52 ppm from theoretical values (Fig. 1). 

GC-APCI-TOF-MS also allowed identification of 

unknown metabolites. By comparing peak elution 

order, EI-MS peaks without database match were 

located in the APCI-MS chromatogram. Molecular 

formulae were generated, in part allowing structural 

assignment (Fig. 2) after searching public databases. 

Differentiation of metabolic phenotypes was tested by 

principal component analysis (PCA) carried out on both 

EI-MS and APCI-MS datasets. Both technologies 

indicated different metabolic patterns between 

knockdown and control animals (Fig. 3). 

Summary 

In a test set of 60 metabolites identified by GC-EI-MS, 

95% of the compounds were confirmed via their 

molecular formula by GC-APCI-TOF-MS. Identification 

of unknown compounds from GC-EI-MS measurements 

is enabled by GC-APCI-TOF-MS. A GC-APCI interface 

preserves the precursor ion information and enables 

molecular formula generation based on accurate mass 

and isotopic pattern information. 

Both techniques yield similar clustering of treated and 

control animals in principal component analysis. 


Fig. 1 Metabolite profiling by GC-EI-MS and GC-APCI-TOF-MS in C. elegans. (A) Metabolite extraction and derivatization. 

(B) EI-MS and APCI-TOF-MS profiles. (C) Identified metabolites with measured TOF-MS m/z values. 

Fig. 2 Unknown identification by GC-APCI-TOF-MS. (A) Peak #3 could not be identified by EI-MS but its location in the APCI- 

TOF-MS chromatogram could be derived from known neighboring peaks #1, 2 and 4. (B) Exact mass and isotopic pattern 

allowed for molecular formula generation and structure proposal. 

Fig. 3 Discrimination of biochemical phenotypes by GC-EI-MS and GC-APCI- 

TOF-MS. Metabolic features (GC-EI-MS: 212, GC-APCI-TOF-MS: 455) were 

extracted from profiling data and submitted to principal component analysis 

(PCA). Two treatment groups (P1, P2) cluster separately from the control (C) 

in the PCA scores plot. 

Fig. 4 micrOTOF QII 

(QTOF) coupled to 450- 

GC via APCI source 

Conclusions 

• GC-EI-MS and GC-APCI-TOF-MS 

were successfully applied as 

complementary platforms for 

metabolic profiling and 

identification in C. elegans. 

• The potential of GC-APCI-TOF- 

MS for the identification of 

unknown metabolites has been 

demonstrated with an 

unidentified compound from EI- 

MS measurements. 

• Further investigations with GC- 

APCI acquiring high resolution Q- 

TOF MS/MS data will extend the 

capabilities for structural 

elucidation. 

GC-APCI-TOF-MS 

-23-

-24- 


Structure Elucidation 

The Fastest Way to a Molecular Formula: How fast can it 

be? TLC-UHPLC-ESI-QTOF MS Coupling for Molecular 

Formula Determination of New Natural Products 

O 

OH 

O 

•Semi-automation with TLC-MS 

OH 

O 

O 

•Mass accuracy of TLC-MS as good 

as with (U)HPLC-MS: ~ 1 ppm 

O 

H 

O 

Emetine 

O 

H 

O 

O 

H 

COOH 

•Sum Formula determination as 

reliable as with (U)HPLC-MS 

Methods 

Mass Spectrometer: Bruker 

Daltonik maXis 4G and maXis 

IMPACT 

HPLC: Dionex RSLC 

Column: Waters, BEHC18, 1.7 µm, 

2.1 x 50 mm 

A: H2O, B: ACN, both 0.1% HCOOH 

5 minutes gradient from 5 to 90%B 

TLC-MS: CAMAG TLC-MS Interface 

Plates: Merck, TLC Silica 60 F254 Software SmartFormula3D; 

Compound Crawler and 

Fragmentation Explorer, Bruker 

Daltonik. 

Dirk Wunderlich1 , Sven Meyer1 , 

Stephanie Grond2 , Torben Hofeditz2 , and 

Tilmann Weber 2 

1Bruker Daltonik GmbH, Fahrenheitstraße 4, 

28359 Bremen, Germany 2Faculty of Science, 

Eberhard-Karls-Universität Tübingen, Auf der 

Morgenstelle 18, 72076 Tübingen, Germany 

OH 

H 

Collinolacton 

Hexacyclinic Acid 

O 

•TLC-MS analysis requires high 

sample amounts (5-50 µg on plate) 

Introduction 

O 

H 

HO 

OH 

OH 

7 

OH 

1 

O 

15 

21 

O 

•High background from TLC plates: 

a) challenging to find compound 

peak 

9 O 

Bafilomycin A1 

b) compounds with no or only one 

nitrogen atom ionize as sodium 

adducts (most MS/MS spec not 

useful) 

The coupling of (U)HPLC with high 

resolution QTOF mass spectrometers 

allows fast determination of 

sum formulae for both known and 

unknown compounds in a Natural 

Product Screening. 

Fig. 1 Structural formulae of several analyzed metabolites for this study 

and instrumental setup of TLC-UHR MS 

+MS2(825.4396), 3.9-3.9min #886-902 

Intens. 

x105 

284.2221 

c) only main compound ionizes, 

difficulties in detection of side 

products 

Results 

Arylomycine A 2 

3 

•TLC-MS analysis well suited for 

pure compounds and samples of low 

complexity. 

This approach is often applied to 

crude extracts or enriched 

fractions. The combination of 

UHPLC systems with fast scanning 

QTOF mass spectrometers has the 

potential to perform the analysis in 

a few minutes. 

2 

1 

•Only few compounds found by TLC- 

MS of iromycine extract, but many 

by (U)HPLC-MS (separation, less ion 

supression) 

383.1238 

Several Natural Products were 

analyzed by a combination of 

(U)HPLC-MS and TLC-MS: 

Reserpine, curcurmine, emetine, 

bafilomycin, several iromycines, 

collinoketon, collinolacton, 

hexacyclinic acid, arylomycin, PKS 

products of LLp-series and three 

Streptomyces crude extracts. 

Here, we investigated if the 

analysis time could be further 

reduced by coupling TLC-MS with 

direct sum formula determination 

from TLC spots by high resolution 

MS analysis. 

160.1079 

0 

100 200 300 400 500 600 700 800 m/z 

•Software FragmentationExplorer 

helpful for MS/MS spectrum 

interpretation 

•TLC-MS analysis only 30 sec 

Fig. 2 Interpretation of arylomycine 

MS/MS spectrum by 

Fragmentation Explorer 

Iromycin 

+MS, 0.41-0.49min #24-29 

342.2040 

Intens. 

x105 

4 

b) 

a) 

SmartFormula₃Dspectrum:C₂ ₆H₁ ₅ O₁ ₀ , 487.07, -0.20 mDa, -0.42 ppm, mSigma: 16.3 

Intens. 

x105 

b) 

a) 

TLC-MS of spot 043 

C19H29NO3Na 0.1 ppm 

0.8 

2 

C 25 H 15 O 8 , -0.4 mDa, 

Outlook 

617.3047 

164.9207 

1 

443.0772 

1- 

0 

•Structure elucidation of 

compounds with similar sum 

formula as iromycine. Are they 

all derivatives of iromycins? 

200 400 600 m/z 

Iro Spot 4 pos MS_1-C,6_01_2502.d: TIC +All MS 

Intens. 

x107 

c) 

C 26 H 15 O 10 , -0.2 mDa, 

C 26 H 13 O 9 , 0.19 mDa, 

C 23 H 13 O 7 , -0.24 mDa, 

0.6 

0.4 

m/z 403 

C19H32NO5 C20H32NO4 C20H32NO3 C20H30NO4 C19H30NO3 C18H30NO4 C19H28NO3 1.25 

487.0671 

1- 

401.0667 

1- 

0.2 

1.00 

LC-MS of spot 04 

469.0565 

1- 

•Structure elucidation of 

additional arylomycines found by 

HR-MS/MS. 

0.75 

C25H18O9 M = 462 

C19H30NO8 C19H30NO6 0.0 

0.50 

400 420 440 460 480 500 m/z 

0.25 

0.00 

Engineered PKS product 

LLp-WR1 

•Addition of a short column 

between TLC and MS for salt 

removal. 

2.0 2.5 3.0 3.5 Time [min] 

Fig. 3 a) TLC of crude extract b) MS spectrum of TLC spot 04 (new oxidized iromycine) 

indicates only 1 compound as sodium adduct 

c) LC-MS Total Ion Chromatogram of manually extracted TLC spot 04 reveals a complex 

mixture measured as protonated species 

Fig. 4 a) Structure proposal for the new LLp-WR1 natural product of the engineered PKSII; 

gives only 1 fragment in MS/MS (m/z 403, C22H11O8, [M-H] - ) 

b) MSMS spectrum for the second, purple colored compound LLp-WL1 in the same extract 

UHR-QqTOF 

For research use only. Not for use in diagnostic procedures.

MALDI Imaging 

Automated Tissue State Assignment for 

High Resolution MALDI-FTMS Imaging Data 

ASMS 2012, #WP396 

Results 

Spleen 

Brain 

Kidney 

A 

Lung 

Stomach 

Heart 

Liver 

Intestines 

J. Paul Speir1 ; Soeren-Oliver 

Deininger2 ; Jens Fuchser2 ; Katherine A. 

Kellersberger1 ; D. Shannon Cornett1 ; 

Cristine Quiason3 ; Sheerin K. Shahidi- 

Latham3 ; Michael Becker2 ; Rainer 

Paape2 1Bruker Daltonics, Inc., Billerica, MA 

2Bruker Daltonik GmbH, Bremen, Germany 

3Genentech, San Francisco, CA 

Fig. 4) Segmentation map of normalized and spatially smoothed dataset (hierarchical clustering 

with euclidean distance and ward linkage of first 7 PCs. A total of 21 clusters is shown. 

Introduction 

Fig. 1 Optical image of the whole rat 

section 

The calculation of a PCA on the whole rat 

dataset with almost 20k pixels typically took 

around 20 minutes, the calculation of the 

clustering around 3 hours on a well equipped 

state-of-the-art workstation. Since these 

calculations can run unattended, these timeframes 

are compatible with a routine workflow. 

Multivariate feature extraction methods such as 

principal component analysis (PCA) and 

segmentation methods are routinely used in 

MALDI-TOF imaging. These methods provide a 

concise overview of the main structure in 

MALDI imaging datasets. 

Summary 

To assess the parameters for the PCA and the 

clustering, calculations were performed on a 

subset of data (each 64 spectra from selected 

organs). Figure 2 shows the scores plots for 

three parameter settings in the PCA. It is seen 

that PCA on the raw data leads to scattered 

and overlapping clusters (Fig. 2a) . After 

normalization, the clusters are better 

separated, but still overlap to some extent (Fig. 

2b). Spatial smoothing (all peak intensities are 

replaced with the average of a 3x3 pixel 

window) leads to well separated clusters (Fig. 

2c). 

B 

A 

High mass resolution imaging datasets 

generated from MALDI-FTMS imaging typically 

are much more complex than MALDI-TOF data, 

containing thousands of mass signals. Although 

FTMS imaging datasets should, therefore, 

benefit highly from concise representations, not 

much work has been done in this field so far. 

Principal component analysis and clustering 

enable efficient and concise interpretation of 

small molecule MALDI-FTMS data. Best results 

are obtained when a spatial smoothing step is 

incorporated. 

B 

This effect is also visible in the whole dataset. 

The scores of the first PC align well with the 

anatomy of the sample (Fig. 3a) as is expected 

when the main variance in the data comes from 

real features in the sample. However, a slight 

granularity (salt-and-pepper-noise) can be 

observed. 

C 

Here we investigated the use of PCA and 

hierarchical clustering for FTMS imaging data. 

Methods 

Conclusions 

C 

• Hierarchical clustering and PCA of 

FTMS-imaging data 

• Automated assignment of tissue 

states 

• Spatial smoothing clustering 

results 

• Detailed annotation of anatomical 

features 

This noise leads to the fact that the clusters of 

spectra are not perfectly separated. The 

clustering shows main anatomical features 

(such as liver, brain, cecum) but with a lot of 

granularity that prevents a perfect 

segmentation (Fig. 3b). 

Cecum Heart Kidney Liver 

Muscle Spleen Stomach 

Spatial smoothing can effectively remove the 

noise. The PC scores show a much smoother 

distribution after spatial smoothing of the 

dataset (Fig. 3c). 

Fig. 3 

A) 1st PC scores (on normalized data) 

B)) Segmentation map based on 

hierarchical clustering (ward linkage, 

euclidean distance on normalized data) 

C) 1st PC score on normalized and 

spatially smoothed data 

MALDI-FTMS-Imaging 

Hierarchical clustering of the dataset after 

normalization and smoothing leads to a very 

detailed and good segmentation, and all 

clusters can be attributed to actual anatomical 

features (Fig. 4). Various clusters were 

assigned a color to aid in visualization, and in 

general one cluster per organ or region of an 

organ (such as medulla and cortex of the 

kidney) is observed. 

Fig. 2 PCA-scores of a subset of spectra 

from 7 regions in the whole body dataset 

A) raw data, B) after normalization to 

RMS, C) after normalization to RMS and 

spatial smoothing (average of 3x3 pixel 

window) 

The MALDI imaging dataset of a whole body rat 

section (19275 pixels, 500µm spatial 

resolution) was acquired on a Solarix 9.4T 

mass spectrometer (Bruker). The sample was 

sectioned at 20 um thickness and mounted on 

a polished steel target. Matrix (DHB, 40 mg/mL 

in 50% methanol) was applied using the HTX 

Sprayer (HTX). 

The peak lists (mass range 400 Da to 900 Da) 

of the individual spectra were binned into one 

bucket table (bin width=0.02 Da). In all 

calculations Pareto scaling on the peak 

intensities was used. Data were normalized to 

RMS intensity. All calculations were done in R 

(www.r-project.orf). The results of the PCA and 

hierarchical clustering were saved in the 

ClinProTools format and imported into 

flexImaging 3.0 (Bruker) for visualization. 


-25-

-26- 

References – Further Reading 

Original papers 


Development and validation of a liquid chromatography-electrospray ionizationtime-of-flight 

mass spectrometry method for induced changes in Nicotiana 

attenuata leaves during simulated herbivory. 

E. Gaquerel, S. Heiling, M. Schoettner, G. Zurek, I.T. Baldwin; J Agric Food Chem. 

2010 Sep 8;58(17):9418-27. 

17-Hydroxygeranyllinalool glycosides are major resistance traits of Nicotiana obtusifolia 

against attack from tobacco hornworm larvae. 

A.R. Jassbi, S. Zamanizadehnajari, I.T. Baldwin; J Chem Ecol. 2010; 36(12):1398-407. 

Jasmonate and ppHsystemin Regulate Key Malonylation Steps in the Biosynthesis 

of 17-Hydroxygeranyllinalool Diterpene Glycosides, an Abundant and Effective 

Direct Defense against Herbivores in Nicotiana attenuata 

S. Heiling, M. C. Schuman, M. Schoettner, P. Mukerjee, B. Berger, B. Schneider, 

A. R. Jassbi, I. T. Baldwin; The Plant Cell 2010; 22: 273–292 

The multifunctional enzyme CYP71B15 (PHYTOALEXIN DEFICIENT3) converts 

cysteine-indole-3-acetonitrile to camalexin in the indole-3-acetonitrile metabolic 

network of Arabidopsis thaliana. 

C. Böttcher, L. Westphal, C. Schmotz, E. Prade, D. Scheel and E. Glawischnig; 

The Plant Cell 2009 Jun 30; 21: 1830-1845 

LC-MSMS Profiling of Flavonoid Conjugates in Wild Mexican Lupine, Lupinus 

reflexus. 

M. Stobiecki, A. Staszkow, A. Piasecka, P. M. Garcia-Lopez, F. Zamora-Natera 

and P. Kachlicki; J. Nat. Prod. 2010; 73, 1254–1260 

Fragmentation Pathways of Acylated Flavonoid Diglucuronides from Leaves of 

Medicago truncatula. 

L. Marczak, M. Stobiecki, M. Jasin´ski, W. Oleszekb and P. Kachlickic; 

Phytochem. Anal. 2010; 21, 224–233 

Evaluation of glycosylation and malonylation patterns in flavonoid glycosides 

during LC/MS/MS metabolite profiling 

P. Kachlicki, J. Einhorn, D. Muth, L. Kerhoas, M. Stobiecki; J. Mass Spectrom. 

2008; 43: 572–586 

Differential metabolic response of narrow leafed lupine (Lupinus angustifolius) 

leaves to infection with Colletotrichum lupini 

D. Muth, P. Kachlicki, P. Krajewski, M. Przystalski, M. Stobiecki; 

Metabolomics 2009; 5:354–362 

LC/MS profiling of flavonoid glycoconjugates isolated from hairy roots, suspension 

root cell cultures and seedling roots of Medicago truncatula 

A. Staszków, B. Swarcewicz, J. Banasiak, D. Muth, M. Jasinski, M. Stobiecki; 

Metabolomics 2011; 7:604–613 

Combined NMR and LC-MS analysis reveals the metabonomic changes in Salvia 

miltiorrhiza Bunge induced by water depletion. 

H. Dai, C. Xiao, H. Liu, H. Tang; J Proteome Res. 2010 Mar 5; 9(3):1460-75. 

Overexpression of Sinapine Esterase BnSCE3 in Oilseed Rape Seeds Triggers 

Global Changes in Seed Metabolism 

K. Clauß, E. von Roepenack-Lahaye, C. Boettcher, M. R. Roth, R. Welti, A. Erban, 

J. Kopka, D. Scheel, C. Milkowski, D. Strack; Plant Phys.2011 March; 155: 1127–1145 

Combined NMR and flow injection ESI-MS for Brassicaceae metabolomics 

J.M. Baker, J.L. Ward, M.H. Beale; Methods Mol Biol. 2012; 860:177-91. 

Chemical identification strategies using liquid chromatography-photodiode 

array-solid-phase extraction-nuclear magnetic resonance/mass spectrometry. 

S. Moco, J. Vervoort; Methods Mol Biol. 2012; 860:287-316 

Bacterial / Sponge Metabolomics 

Efficient mining of myxobacterial metabolite profiles enabled by liquid chromatography-electrospray 

ionisation-time-of-flight mass spectrometry and compound-based 

principal component analysis. 

D. Krug, G. Zurek, B. Schneider, R. Garcia, R. Müller; Anal Chim Acta. 2008 Aug 22; 

624(1):97-106. 

Discovering the Hidden Secondary Metabolome of Myxococcus xanthus: a Study 

of Intraspecific Diversity 

D. Krug, G. Zurek, O. Revermann, M. Vos, G.J. Velicer, R. Müller; Appl Environ 

Microbiol. 2008 May; 74(10):3058-68. 

Myxoprincomide, a novel natural product from Myxococcus xanthus discovered by 

a comprehensive secondary metabolome mining approach 

N.S. Cortina, D. Krug, A. Plaza, O. Revermann, R. Müller; Angewandte Chemie 

International Edition 2012; 51(3): 811–816. 

Methods to optimize myxobacterial fermentations using off-gas analysis 

S. Hüttel, R. Müller; Microbial Cell Factories 2012; 11:59 

Microbial Strain Prioritization Using Metabolomics Tools for the Discovery of 

Natural Products 

Y. Hou, D.R. Braun, C.R. Michel, J.L. Klassen, N. Adnani, T.P. Wyche, T.S. Bugni; 

Anal Chem. 2012 May 15; 84(10): 4277–4283. 

Metabolic fingerprinting as an indicator of biodiversity: towards understanding 

inter-specific relationships among Homoscleromorpha sponges 

J. Ivaniševi, O. P. Thomas, C. Lejeusne, P. Chevaldonné, T. Pérez; 

Metabolomics 2011; 7(2): 289-304 

Synthesis of 7-(15)N-Oroidin and evaluation of utility for biosynthetic studies 

of pyrrole-imidazole alkaloids by microscale (1)H-(15)N HSQC and FTMS. 

Y.G. Wang, B.I. Morinaka, J.C. Reyes, J.J. Wolff, D. Romo, T.F. Molinski.; 

J Nat Prod. 2010 Mar 26; 73(3):428-34. 

GC-APCI based Metabolomics 

Gas chromatography/atmospheric pressure chemical ionization-time of flight 

mass spectrometry: analytical validation and applicability to metabolic profiling. 

A. Carrasco-Pancorbo , E. Nevedomskaya , T. Arthen-Engeland, T. Zey, G. Zurek, 

C. Baessmann, A.M. Deelder, O. Mayboroda; Anal Chem. 2009 Dec 15; 

81(24):10071-9. 

Evaluation of GC-APCI/MS and GC-FID As a Complementary Platform. 

T. Pacchiarotta, E. Nevedomskaya, A. Carrasco-Pancorbo, A.M. Deelder, and 

O.A. Mayboroda; J Biomol Tech. 2010 Dec; 21(4): 205–213. 

Performance Evaluation of Gas Chromatography Atmosphericc Pressure 

Chemical Ionization Time-of-Flight Mass Spectrometry for Metabolic 

Fingerprinting and Profiling 

C. J. Wachsmuth M. F. Almstetter, M. C. Waldhier, M. A. Gruber, N. Nuernberger, 

P. J. Oefner, K. Dettmer; Anal. Chem., 2011; 83 (19): 7514–7522 

Lipidomics 

Combined Reversed Phase HPLC, Mass Spectrometry, and NMR Spectroscopy 

for a Fast Separation and Efficient Identification of Phosphatidylcholines. 

J. Willmann, H. Thiele, and D. Leibfritz; J Biomed Biotechnol. 2011; 2011: 385786. 

Time-Dependent Oxidation during Nano-Assisted Laser Desorption Ionization 

Mass Spectrometry: A Useful Tool for Structure Determination or a Source of 

Possible Confusion? 

K. Pavlásková, M. Strnadová, M. Strohalm, V. Havlícek, M. Šulc, M. Volný; 

Anal. Chem., 2011, 83 (14): 5661–5665 


Unraveling different chemical fingerprints between a champagne wine and 

its aerosols. 

G. Liger-Belair, C. Cilindre, R. D. Gougeon, M. Lucio, I. Gebefügi, P. Jeandet, 

P. Schmitt-Kopplin; PNAS. 2009 Sept. 29; 106(39):16545–16549 

Expressing forest origins in the chemical composition of cooperage oak woods 

and corresponding wines by using FTICR-MS. 

R.D. Gougeon, M. Lucio, A. De Boel, M. Frommberger, N. Hertkorn, D. Peyron, 

D. Chassagne, F. Feuillat, P. Cayot, A. Voilley, I. Gebefügi, P. Schmitt-Kopplin; 

Chemistry. 2009; 15(3):600-11. 

Exploratory analysis of human urine by LC-ESI-TOF MS after high intake of olive oil: 

understanding the metabolism of polyphenols. 

R. García-Villalba, A. Carrasco-Pancorbo, E. Nevedomskaya, O. Mayboroda, 

A. Deelder, A. Segura-Carretero, and A. Fernández-Gutiérrez; 

Analytical and bioanalytical chemistry. 2010 Sep; 398(1):463-75 

Automated identification of phenolics in plant-derived foods by using library 

search approach. 

M. Gómez-Romero, G. Zurek, B. Schneider, C. Baessmann, A. Segura-Carretero, 

and A. Fernández-Gutiérrez; Food Chemistry. 2011 Jan; 124(1): 379-386 

Scope and limitations of principal component analysis of high resolution LC-TOF-MS 

data: the analysis of the chlorogenic acid fraction in green coffee beans as a case 

study 

N. Kuhnert, R. Jaiswal, P. Eravuchira, R. M. El-Abassy, B. von der Kammer; 

Anal. Methods 2011; 3, 144-155 

Apple Peels, from Seven Cultivars, Have Lipase-Inhibitory Activity and Contain 

Numerous Ursenoic Acids As Identified by LC-ESI-QTOFHRMS 

T. K. McGhie, S. Hudault, R. C. M. Lunken, J. T. Christeller; J. Agric.Food Chem. 

2012; 60: 482−491

Ultra high performance liquid chromatography-time of flight mass spectrometry 

for analysis of avocado fruit metabolites: Method evaluation and applicability to the 

analysis of ripening degrees 

E. Hurtado-Fernández, T. Pacchiarotta, M. Gómez-Romero, B. Schoenmaker, 

R. Derks, A. M. Deelder, O. A. Mayboroda, A. Carrasco-Pancorbo, A. Fernández- 

Gutiérrez; Journal of Chromatography A 211; 1218(42): 7723–7738 

Human / Clinical Metabolomics 

Amino acid profiling in urine by capillary zone electrophoresis - mass spectrometry. 

O.A. Mayboroda, C. Neusüss, M. Pelzing, G. Zurek, R. Derks, I. Meulenbelt, 

M. Kloppenburg, E. Slagboom, and A. Deelder; Journal of chromatography A. 2007 

Aug; 1159(1-2):149-53 

CE-MS for metabolic profiling of volume-limited urine samples: application to 

accelerated aging TTD mice. 

E. Nevedomskaya, R. Ramautar, R. Derks, I. Westbroek, G. Zondag, I. van der 

Pluijm, A. M. Deelder, O. A. Mayboroda; J. Proteome Res. 2010; 9 (9): 4869–4874 

Cross-platform analysis of longitudinal data in metabolomics 

E. Nevedomskaya, O. A. Mayboroda, A. M. Deelder; Mol. BioSyst., 2011; 7: 3214– 

3222 

Explorative Analysis of Urine by Capillary Electrophoresis-Mass Spectrometry in 

Chronic Patients with Complex Regional Pain Syndrome 

R. Ramautar, A. A. van der Plas, E. Nevedomskaya, R. J. E. Derks, G. W. Somsen, 

G. J. de Jong, J. J. van Hilten, A. M. Deelder, O. A. Mayboroda; 

Journal of Proteome Research 2009; 8: 5559–5567 

Metabolic profiling of human urine by CE-MS using a positively charged capillary 

coating and comparison with UPLC-MS 

R. Ramautar, E. Nevedomskaya, O. A. Mayboroda, A. M. Deelder, I. D. Wilson, 

H. G. Gika, G. A. Theodoridis, G. W. Somsena, G. J. de Jonga; 

Mol. BioSyst., 2011; 7: 194–199 

Metabolite profiling of blood plasma of patients with prostate cancer 

P. G. Lokhov, M. I. Dashtiev, S. A. Moshkovskii. A. I. Archakov; 

Metabolomics 2012; 6(1): 156-163 

Potential Effect of Diaper and Cotton Ball Contamination on NMR- and LC/MS- 

Based Metabonomics Studies of Urine from Newborn Babies 

A. M. Goodpaster, E. H. Ramadas, M. A. Kennedy; Anal. Chem. 2011; 83: 896–902 

Metabolomics Reveals Metabolic Biomarkers of Crohn’s Disease 

J. Jansson, B. Willing, M. Lucio, A. Fekete, J. Dicksved, J. Halfvarson, C. Tysk, 

P. Schmitt-Kopplin; PLoS ONE 4(7): e6386 

Holistic metabonomic profiling of urine affords potential early diagnosis for bladder 

and kidney cancers 

Z. Huang, Y. Chen, W. Hang, Y. Gao, L. Lin, D. Y. Li, J. Xing, X. Yan; 

Metabolomics 2012, DOI: 10.1007/s11306-012-0433-5 

Single Cell Metabolomics 

Capillary electrophoresis with electrospray ionization mass spectrometric 

detection for single-cell metabolomics. 

T. Lapainis, S. S. Rubakhin, J. V. Sweedler; Anal Chem. 2009; 81(14): 5858-64. 

Metabolic Differentiation of Neuronal Phenotypes by Single-cell Capillary 

Electrophoresis Electrospray Ionization-Mass Spectrometry 

P. Nemes, A. M. Knolhoff, S. S. Rubakhin, J. V. Sweedler; Anal. Chem. 2011; 83: 

6810–6817 

Environmental Metabolomics 

Metabolic evidence for biogeographic isolation of the extremophilic bacterium 

Salinibacter ruber 

R. Rossello´-Mora, M. Lucio, A. Pena, J. Brito-Echeverrıa, A. Lopez-Lopez, 

M. Valens-Vadell, M. Frommberger, J. Anto´n, P. Schmitt-Kopplin; 

The ISME Journal 2008; 2: 242–253 

High molecular diversity of extraterrestrial organic matter in Murchison meteorite 

revealed 40 years after its fall 

P. Schmitt-Kopplina, Z. Gabelicab, R. D. Gougeonc, A. Feketea, B. Kanawatia, 

M. Harira, I. Gebefuegia, G. Eckeld, N. Hertkorn; PNAS 2010; 107(7): 2763–2768 

Localization of secondary metabolites in marine invertebrates: contribution of 

MALDI MSI for the study of saponins in Cuvierian tubules of H. forskali. 

S. Van Dyck, P. Flammang, C. Meriaux, D. Bonnel, M. Salzet, . Fournier, 

M. Wisztorski; PLoS One. 2010 Nov 10; 5(11):e13923. 

Unknown Identification 

A strategy for the determination of the elemental composition by fourier 

transform ion cyclotron resonance mass spectrometry based on isotopic 

peak ratios. 

D. Miura, Y. Tsuji, K. Takahashi, H. Wariishi, K. Saito; Anal Chem. 2010 Jul 1; 8 

2(13): 5887-91 

Bruker Daltonics – Application Notes 

ET-19 

Metabolic profiling of tea extracts by high-resolution LC in combination 

with maXis UHR-TOF MS analysis 

ET-21 

Challenges in metabolomics addressed by targeted and untargeted 

UHR-Q-TOF analysis 

ET-22 

Metabolic profiling of a Corynebacterium glutamicum ∆prpD2 by GC-APCI 

high-resolution Q-TOF analysis 

ET-23 

Accurate molecular formula determination and identification of molecules 

with > 1100 m/z with UHR-TOF 

ET-26 

Novel approaches for small molecule identification in metabolomics research 

ET-29 

Screening for novel natural products from myxobacteria using LC-MS and 

LC-NMR 

ET-30 

Metabolic profiling of Arabidopsis thaliana secondary metabolites using 

a maXis impact 

TN-23 

Certainty in small molecule identification by applying SmartFormula3D on 

a UHR-TOF mass spectrometer 

CA #271518 

Using gas chromatography – tandem mass spectrometry to improve the 

reliability and accuracy of organic acid analyses in urine 

-27-


Bruker Daltonik GmbH 

Bremen · Germany 

Phone +49 (0)421-2205-0 

Fax +49 (0)421-2205-103 

sales@bdal.de 

www.bruker.com/metabolomics 

Bruker Daltonics Inc. 

Billerica, MA · USA 

Phone +1 (978) 663-3660 

Fax +1 (978) 667-5993 

ms-sales@bdal.com 

Fremont, CA · USA 

Phone +1 (510) 683-4300 

Fax +1 (510) 490-6586 

ms-sales@bdal.com 

Bruker Daltonics is continually improving its products and reserves the right 

to change specifications without notice. © BDAL 09-2012, #703376

Posterhall Metabolomics 2012 eBook - Bruker

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?