v - ISMAR

BULLETIN OF MAGNETIC RESONANCE
The Quarterly Review Journal of the
International Sodiety of Magnetic Resonance
VOLUME 14 October 1992 NUMBERS 1-4
,\ Proceedings
*' International
Society of ;
; Magnetic
Resonance
Xlth,
- .Meeting
•>• 4
1- a
VF:
]\n\ 19 :i
\ \NC()U\ TR
BC,C\N\O\

BULLETIN OF MAGNETIC RESONANCE
The Quarterly Review Journal of the
International Society of Magnetic Resonance
Editor:
DAVID G. GORENSTEIN
Department of Chemistry
Purdue University
West Lafayette, IN 47907 USA.
Fax: 317-494-0230
INTERNET:david@chem.purdue.edu
Editorial Board:
E.R.ANDREW LAWRENCE BERLINER ROBERT BLINC
University of Florida Ohio State University E. Kardelj University of Ljubljana
Gainesville, Florida, U.S.A. Columbus, Ohio, U.S.A. Ljubljana, Yugoslavia
H.CHIHARA GARETH R. EATON DANIEL FIAT
Osaka University University of Denver University of Illinois at Chicago
Toyonaka, Japan Denver, Colorado, U.S.A. Chicago, Illinois, U.S.A.
SHIZUO FUJIWARA DAVID GRANT ALEXANDER PINES
University of Tokyo University of Utah University of California
Bunkyo-Ku, Tokyo, Japan Salt Lake City, Utah, U.S.A. Berkeley, California, U.S.A.
MIK PINTAR CHARLES P. POOLE, JR. BRIAN SYKES
University of Waterloo University of South Carolina University of Alberta
Waterloo, Ontario, Canada Columbia, South Carolina, U.S.A. Edmonton, Alberta, Canada
The Bulletin of Magnetic Resonance is a quarterly review journal by the International Society of
Magnetic Resonance. Reviews cover all parts of the broad field of magnetic resonance, viz.. the
theory and practice of nuclear magnetic resonance, electron paramagnetic resonance, and nuclear
quadrupole resonance spectroscopy including applications in physics, chemistry, biology, and
medicine. The BULLETIN also acts as a house journal for the International Society of Magnetic
Resonance.
CODEN: BUMRDT ISSN: 0163-559X
Bulletin of Magnetic Resonance, The Quarterly Journal of International Society of Magnetic
Resonance. 1992 copyright by.the International Society of Magnetic Resonance. Rates: Libraries
and non-ISMAR members $80.00, members of JSMAR, $25.00. All subscriptions are for a volume
year. All rights reserved. No part of this journal may be reproduced in any form for any purpose or by
any means, abstracted, or entered into any data base, electronic or otherwise, without specific
permission in writing from the publisher.

J. ANGLISTER
Israel
R. BLINC
Yugoslavia
P.T. CALLAGHAN
New Zealand
L.G.CONTI
Italy
E.L.HAHN
U.S.A.
M.J.R. HOCH
5. Africa
P.C. LAUTERBUR
U.S.A.
M.MEHRING
Germany
C.P: POOLE
U.S.A.
G.C.K. ROBERTS
England
N.V. VUGMAN
Brazil
C.S. YANNONI
U.S.A.
Council of the International Society of Magnetic Resonance
President: R. FREEMAN, England
Vice-President: A. PINES, U.S.A.
Founding Chairman: D. FIAT, U.S.A.
Secretary-General: R.K. HARRIS, England
Treasurer: R.R. VOLD, U.S.A.
Past President: C.P. SLICHTER, U.S.A.
E.D. BECKER
U.S.A.
G. BODENHAUSEN
Switzerland
H.CHIHARA
Japan
R. DESLAURIERS
Canada
KM. HAUSSER
Germany
C.L. KHETRAPAL
India
E. LIPPMAA
Estonia
H. PFEIFER
Germany
M. PUNKINEN
Finland
P. SERVOZ-GAVESf
France
J.S. WAUGH
U.S.A.
MR. BENDALL
Australia
W.S. BREY
U.S.A.
S. CLOUGH
England
R.R. ERNST
Switzerland
J.W. HENNEL
Poland
VJ. KOWALEWSKI
Argentina
B. MARAVIGLIA
Italy
M.M. PINTAR
Canada
L.W. REEVES
Canada
J. STANKOWSKI
Poland
K. WUTHRICH
Switzerland
The aims of the International Society of Magnetic Resonance are to advance and diffuse knowledge
of magnetic resonance and its applications in physics, chemistry, biology, and medicine, and to
encourage and develop international contacts between scientists.
The Society sponsors international meetings and schools in magnetic resonance and its applications
and publishes the quarterly review journal. The Bulletin of Magnetic Resonance, the house journal of
ISMAR.
The annual fee for ISMAR membership is $20 plus $25 for a member subscription to the Bulletin of
Magnetic Resonance.
Send subscription to: International Society of Magnetic Resonance
Professor Regitze R. Void, Treasurer
Department of Chemistry, 0342
University of California, San Diego
9500 Gilman Drive
La Jolla, CA 92093-0342
(619) 534-0200; FAX (619) 534-7042
Bitnet: rrvold@ucsd.bitnet

Vol. 14, No. 1-4
ORGANIZING COMMITTEE
for
ISMAR 92
C. Fyfe, Chairman
Department of Chemistry
University of British Columbia
Vancouver, British Columbia
CANADA
R. Andersen
Department of Chemistry
University of British Columbia
Vancouver, British Columbia
CANADA
G. S. Bates, Treasurer
Department of Chemistry
University of British Columbia
Vancouver, British Columbia
CANADA
M. Bloom
Department of Physics
University of British Columbia
Vancouver, British Columbia
CANADA
E. E. Burnell
Department of Chemistry
University of British Columbia
Vancouver, British Columbia
CANADA
G. Drobny
Department of Chemistry
University of Washington
Seattle, Washington
USA
I. Gay
Department of Chemistry
Simon Fraser University
Burnaby, British Columbia
CANADA
F. G. Herring
Department of Chemistry
University of British Columbia
Vancouver, British Columbia
CANADA
CO-EDITORS OF THE PROCEEDINGS
D. G. Gorenstein
Department of Chemistry
Purdue University
West Lafayette, Indiana
USA
C. Fyfe
Department of Chemistry
University of British Columbia
Vancouver, British Columbia
CANADA

4 Bulletin of Magnetic Resonance
ACKNOWLEDGMENTS
The Organizing Committee of the Xlth conference of the International Society of Magnetic Resonance
gratefully acknowledges the financial support of the following organizations:
B.P. Research, pic
Bio-Mega, Inc.
Bio Rad - Sadtler Division
Bruker Spectrospin Canada Ltd.
Bulletin of Magnetic Resonance
Chemagnetics/Otsuka Electronics
Dean of Graduate Studies, U.B.C.
Dean of Science, U.B.C.
Doty Scientific, Inc.
Dow Chemical Co.
ICON (Isotope) Services Inc.
JEOL USA, Inc.
John Wiley & Sons
Molecular Simulations
Oxford Instruments
Pfizer
Quadra Logic Technologies, Inc.
Varian NMR Instruments
Special thanks are extended to the National Sciences and Engineering Research Council of Canada (NSERC)
for its support through the award of a Conference Grant to ISMAR 92. We are also grateful to the Department
of Chemistry at the University of British Columbia for its support.

Vol. 14, No. 1-4 5
Contents
Proceedings of the Xlth Meeting of the
International Society of Magnetic Resonance
July 19 - 24, 1992
Vancouver, B. C. Canada
Solid-State Polarization-Transfer Experiments Involving Quadrupolar Nuclei,
K. T. Mueller, C. A. Fyfe, H. Grondey, K. C. Wong-Moon and T. Markus 9
Magnetic Resonance Evidence for Superconductivity in a Semimetal, I. P. Goudemond,
G. J. Hill and M. J. R. Hoch 12
NMR Spectroscopy in Cardiac Surgery, R. Deslauriers, S. Lareau, R. S. Labow,
W. J. Keon, G-H. Tian, A. L. Panos, C. A. M. Barrozo, I. S. Ali, O. Al-Nowaiser and T. A. Salerno 15
Topology and Spin Alignment in Organic High-Spin Molecules, Y. Teki, K. Sato,
M. Okamoto, A. Yamashita, Y. Yamaguchi, T. Takui, T. Kinoshita and K. Itoh 24
New Developments in Pulsed Electron Paramagnetic Resonance: Relaxation Mechanisms
of Nitroxide Spin Labels, C. Mailer, B. H. Robinson and D. A. Haas 30
New Developments in Pulsed Electron Paramagnetic Resonance: Direct Measurement
of Rotational Correlation Times from Decay Curves, D. A. Haas, C. Mailer,
T. Sugano and B. H. Robinson 35
Non-Linear Effects in Standard 2D NOE Experiments in Coupled Spin Systems,
R. C. R. Grace and A. Kumar 42
Deriving Structures from 2D NMR. A Method for Denning the Conformation of
a Protein Adsorbed to Surfaces, D. A. Keire and D. G. Gorenstein 57
3D-Structure Determination of Flavoridin in Solution: New Computational Strategy
for Disulfide-Bridge Mapping, H. Senn, W. Klaus and P. Gerber 64
NMR Approaches to Large Proteins: trp Repressor and Chloramphenicol Acetyltransferase,
L.-Y. Lian, J. P. Derrick, V. Ramesh, R. O. Frederick, S. E. H. Syed and
G. C. K. Roberts 65
2D NMR Study of Drug-Protein Interactions: Ethidium Bromide - Neocarzinostatin
Complex, S. Mohanty, L. C. Sieker and.G. P. Drobny 68
Quantitative Analysis in Multi-Dimensional Transferred NOE Experiments: Improved
Spectral Acquisition and Processing, F. Ni 75
Time-Resolved Solid-State NMR: Small Molecules and Enzymes in Rapidly Frozen
Solution, J. N. S. Evans, R. J. Appleyard and W. Shuttleworth 81
Coupled Methyl Groups in Dimethyl Sulphide, M. R. Johnson, S. Clough, A. J. Horsewill
and I. B. I. Tomsah 86

Bulletin of Magnetic Resonance
NMR Relaxation Studies of Microdynamics in Chloroaluminate Melts, P. A. Shaw,
W. R. Carper, C. E. Keller and J. S. Wilkes .92
Structure and Dynamics of a Membrane Bound Polypeptide, T. A. Cross,
R. R. Ketchem, W. Hu, K.-C. Lee, N. D. Lazo and C. L. North 96
The Role of Metal Ions in Processes of Conformational Selection during Ligand-
Macromolecule Interactions, E. Gaggelli, N. Gaggelli, G. Valensin and A. Maccotta .... 102
Detection and Characterization of CFC, HCFC and HFC Gases in Foamed Insulation
by High Field NMR Imaging, L. H. Randall, C. A. Fyfe, Z. Mei and S. Whitworth 108
Mysterious Negative Peaks in the 1 H{ 1 H}NOE Difference Spectra of Some Thiopyran
Compounds, C. Szantay, Jr 112
H-l and C-13 NMR Spectra of the Carbanions Produced from Phenylpropene
Derivatives, A. Yoshino, K. Aoki, M. Ushio and K. Takahashi 116
Intracellular pH and Inorganic Phosphate Effects on Skeletal Muscle Force,
E. R. Barton-Davis, R. W. Wiseman and M. J. Kushmerick 122
A Simple Model for the Influence of Motion on the NMR Line Shape, M. Goldman,
T. Tabti, C. Fermon, J. F. Jacquinot and G. Saux 126
The Effect on Ti of Correlated Water Motions in the Polar Phase of Colemanite,
J. Sun and A. Watton 131
Measurement of Deuteron Spin Relaxation Times in Liquid Crystals by a Broadband
Excitation Sequence, R. Y. Dong 134
Carbon-13 Relaxation Mechanisms and Motional Studies in Selected Halomethane
Molecules, A. A. Rodriguez, T. Davis and L. E. Nance 139
An Efficient Large Sample Volume System for Solid State NMR, R. J. Pugmire,
Y. J. Jiang, M. S. Solum and D. M. Grant . . . . 144
Magnetic Resonance Spectroscopic Investigations of Poly(p-Phenylene Sulfide/Disulfide),
PPS/DS, D. W. Lowman and D. R. Fagerburg 148
Application of 2-D HETCOR NMR to Investigate Polymer Blend Heterogeneity,
S. Kaplan 153
New High Resolution NMR Studies in Polycrystalline Tetracyanoquinodimethane,
M. T. Nunes, A. Vainrub, M. Ribet, F. Rachdi, P. Bernier, M. Almeida and G. Feio 156
Use of NMR Relaxation Measurements to Derive the Binding Site of Plastocyanin
in Complexes With Cytochrome-F and C, S. Modi, E. McLaughlin, D. S. Bendall, S. He
and J. C. Gray 159
Metal-peptide Interaction: Influence of the Aminoacid Sequence on the Binding
of Co(II) to Glycyltryptophan and Tryptophylglycine Studied by X H NMR and
Fluorescence, A. Spisni, G. Sartor, L. Franzoni, A. Orsolini, P. Cavatorta and M. Tabak . . . 165
Assignments of the X H NMR Spectrum of a Consensus DN A-Binding Peptide from
the HMG-I Protein, J. N. S. Evans, M. S. Nissen and R. Reeves 171
Solution Structure of the DNA-binding Domain of GAL4 from Saccharomyces cerevisiae,
J. D. Baleja, V. Thanabal, T. Mau and G. Wagner . 175

Vol. 14, No. 1-4 7
Structural Investigation of Folic Acid by NMR Proton Relaxation and Molecular
Mechanics Analysis, C. Rossi, A. Donati, S. Ulgiati, and M. R. Sansoni 181
Characterization of Water-in-Bitumen Emulsions in Model Porous Media by NMR
Microscopic Imaging Techniques, L. H. Randall and G. E. Sedgwick and C. A. Fyfe .... 186
Computer Graphics for Pulse Sequence Analysis, J. Callahan, D. Mattiello and
G. P. Drobny 191
NMR Investigation of the Simultaneous Fermentation of Xylose and Glucose by
a Selected Strain of Klebsiella Planticola (Gil), C. Rossi, A. Lepri, M. P. Picchi, S. Bastianoni,
D. Medaglini, M. Vanassina and E. Cresta 197
Interleukin-1 Receptor Antagonist Protein: Solution Secondary Structure from
NOE's and l Ha and 13 Ca Chemical Shifts, B. J. Stockman, T. A. Scahill, A. Euvrard,
N. A. Strakalaitis, D. P. Brunner, A. W. Yem and M. R. Deibel, Jr. 202
Green's Function Calculation of Effective Nuclear Relaxation Times in Metals and
Disordered Metals, M. Martin-Landrove and J. A. Moreno 208
Stochastic Averaging Revisited, D. H. Jones, N. D. Kurur and D. P. Weitekamp 214
Magnetic Resonance of Trapped Ions by Spin-Dependent Cyclotron Acceleration,
P. J. Pizarro and D. P. Weitekamp . . 220
Coordination Modes of Histidine Moiety in Copper (II) Dipeptide Complexes Detected
by Multifrequency ESR, R. Basosi, R. Pogni, and G. D. Lunga 224
An EPR and ab initio Study of a Phosphaalkene Radical Anion, and Comparison
with other Phosphorus-Containing Radical Ions, M. Geoffroy, G. Terron, A. Jouaiti,
P. Tordo and Y. Ellinger 229
Conformational Substate Distribution in Myoglobin as Studied by EPR Spectroscopy,
A. R. Bizzarri and S. Cannistraro . 234
Effect of Paramagnetic Ions in Aqueous Solution for Precision Measurement of the
Proton Gyromagnetic Ratio, A. R. Lim, C. S. Kim and S. H. Choh 240
Magnetic Resonance of 23 Na and 14 N Nuclei in Single and Multi-Domain Crystals
of Ferroelectric NaNO2, S. H. Choh and K. T. Han 246
Knight Shifts and Spin Dynamics in Disordered Systems, M. J. R. Hoch and
S. T. Stoddart .252
Numerical Design and Evaluation of Broadband Pulse Sequences for 1=1 Spin
Systems, D. L. Mattiello, J. Callahan, T. Alam and G. P. Drobny 255
A BASIC Program to Calculate the Evolution of Cartesian Product Operators,
S. Mammi 259
Selective Long-Range Polarization Transfer via DEPT, T. Parella, F. Sanchez-Ferrando
and A. Virgili . . 263
Computer Simulations of High Resolution NMR Spectra, S. A. Smith, W, E. Palke
and J.T. Gerig 267
Variation of 13 C NMR Linewidths of Metallocenes as a Function of Magic Angle
Sample Spinning Frequency, I. J. Shannon, K. D. M. Harris and S. Arumugam 273

Bulletin of Magnetic Resonance
The Structural Role of Water in Silicate Glasses: X H and 29 Si NMR Evidence,
J. Kiimmerlen, T. Schaller, A. Sebald and H. Keppler 278
High-Resolution Solid-State NMR Study of Microstructures in Layered Aluminosilicate,
S. Hayashi, T. Ueda, K. Hayamizu and E. Akiba 282
Broadline NMR of Structural Ceramics, C. Connor 285
Permeability of Liposomal Membranes to Molecules of Environmental Interest:
Results from NMR Experiments Employing Shift Agents, F. G. Herring, W. R. Cullen,
J. C. Nelson and P. S. Phillips 289
Nuclear Magnetic Resonance Partitioning Studies of Solute Action in Lipid Membranes,
L. Ma, T. F. Taraschi and N. Janes 293
Weak Molecular Interactions: NMR Spectroscopy of Oriented Molecules,
C. L. Khetrapal 299
Structural Studies of Collagen by Solid State NMR, R. J. Wittebort and A. M. Clark . 303
Calendar of Forthcoming Conferences 307
Recent Magnetic Resonance Books 308
Instructions for Authors 314

Vol. 14, No. 1-4
Solid-State Polarization-Transfer Experiments
Involving Quadrupolar Nuclei
K. T. Mueller, C. A. Fyfe, H. Grondey, K. C. Wong-Moon and T. Markus
1 Introduction
Department of Chemistry, University of British Columbia
2036 Main Mall, Vancouver, BC Canada V6T 1Z1
Historically, cross-polarization experiments
[1,2] have been used to transfer spin
coherence from abundant spins to a dilute spin
system. Protons (*H) have been used almost
exclusively as the source of strong nuclear
polarization for cross-polarization experiments,
although there have been some examples where
other abundant nuclei have been used. Coupled
with magic-angle spinning (MAS) NMR [3],
cross-polarization techniques have proven
extremely powerful for the study of organic
solids.
Inorganic systems such as zeolites, gels,
and ceramics are of great technological
importance and interest and contain many
quadrupolar nuclei but very few protons. For
quadrupolar nuclei with non-integral spins such
as n B, 17 O, and 27 A1, the second-order
quadrupolar broadening of the readily observed
central (+1/2 -1/2) transition is not
completely averaged by MAS, and the NMR
lines from quadrupolar spins are shifted and
distorted in single-axis spinning experiments
[4,5]. Very few examples exist of crosspolarization
experiments involving quadrupolar
nuclei, and they all involve magnetization
transfer from protons to quadrupolar nuclei. We
have undertaken a study to determine the
feasibility of polarization transfer and dipolar
dephasing experiments between spin pairs in
these systems, particularly between 3l P (I = 1/2)
and 27 A1 (I = 5/2). Our preliminary results show
that these experiments are indeed possible [6],
The attainment of cross-polarization from
quadrupolar spin systems is particularly
important in materials chemistry as these nuclei
usually have very short Ti relaxation times.
Spin-1/2 nuclei in dense inorganic systems may
have Ti values ranging from many seconds to
hours, effectively precluding their observation in
many instances. By using cross-polarization
from the quickly relaxing quadrupolar spins,
spectra of the spin-1/2 nuclei could be obtained
in a relatively short time. Additional
information regarding the local structure and
bonding in these systems might also be obtained
through the distance dependence of the crosspolarization
process.
Similarly, dipolar-dephasing NMR
experiments such as rotational-echo doubleresonance
(REDOR) and transferred-echo doubleresonance
(TEDOR) have been demonstrated to
be useful for demonstrating connectivities and
determining internuclear distances [7,8] in
heteronuclear spin systems with dipolar
couplings. Experimental verification of these
experiments with the same heteronuclear spin
pair ( 31 P and 27 A1) demonstrates the feasability
of applying these techniques to systems
containing quadrupolar nuclei.
2 Experimental
The sample chosen for study was the very
large pore molecular sieve VPI-5, an
aluminophosphate dihydrate containing 18membered-rings
[9]. NMR experiments were
performed under MAS conditions in a 9.4 T
superconducting magnet where the resonance
frequencies for 31 P and 27 A1 are 161.98 MHz
and 104.26 MHz respectively. The rotational
frequencies in all experiments were
approximately 3.1 kHz, and 90° pulse times for
the nuclei studied ranged from 9 to 12 (xsec.
3 Results
The spectra in Figure 1 demonstrate the
transfer of magnetization using cross-polarization
in both directions between the 27 A1 and 31 P
spins in the Al-O-P bonding units in VPI-5.
The cross-polarization is accomplished with an
appropriate spin-locking pulse sequence [2] after
a preparation pulse creates spin coherence for the
nuclei used as the polarization source. With
MAS the 31 P chemical shift anisotropies are
averaged to their isotropic values for the three
crystallographically inequivalent 31 P sites in the
unit cell. For the 27 A1 nuclei, MAS partially
averages the second-order quadrupolar interaction
and two resonances are seen: One from the

10
(b)
0 -SO '
ppm from 85% H3PO4
(d)
» 0 -JO
ppm from A1(NO3>3
Figure 1. Demonstration of 27 A1 —> 31 P and
31 P —> 27 A1 cross-polarization in VPI-5
(projection of unit cell shown at top). Spinning
sidebands are marked with an (s).
tetrahedrally coordinated aluminum sites (41
ppm) and a second from the octahedrally
coordinated aluminum (approximately -18 ppm).
The observed signals are solely due to crosspolarization
and not caused by direct irradiation
during the spin-lock as proven by a series of
cross-check experiments, of which spectra (b) and
(d) of Figure 1 are representative.
A two-dimensional heteronuclear
correlation experiment [10] using crosspolarization
can be performed by preparing the
aluminum spins with a 90° pulse, and then
encoding their evolution frequencies in an initial
time period. The aluminum polarization is
subsequently transferred to the phosphorus spins
with a spin-lock, and a phosphorus free induction
decay is accumulated after each of a set of
aluminum evolution times. Two-dimensional
Fourier transformation provides the correlation
spectrum of Figure 2. From the twodimensional
spectrum it is evident that each of
the three 31 P resonances is connected to both
tetrahedral and octahedral 27 A1 resonances, in
agreement with the proposed crystal structure of
VPI-5 [11].
The REDOR experiment [7] was carried
out in both directions between 27 A1 and 31 P
nuclei in VPI-5, and the results are shown in
Figure 3. In both directions, negative crosscheck
experiments (not shown) were undertaken
to ensure that the observed signal came from the
dipolar dephasing phenomenon and not from
timing missets or experimental artifacts.
00
3
100 JO 0 - 5 0
ppm from A1(NO3)3
Bulletin of Magnetic Resonance
Figure 2. Two-dimensional heteronuclear
correlation spectrum of - 7 A1 and 31 P in VPI-5.
The TEDOR experiment [8] was also
accomplished with initial evolution of 27 A1
spins and subsequent transfer to the 31 P after two
rotor periods of preparative dephasing. The
signal in Figure 3(c) was obtained after one
additional period of dipolar evolution in order to
create observable spin coherence from the
antiphase signal which was transferred at the time
of the 27 A1 spin echo. As before, negative
cross-check experiments were performed and gave
null signals.
A two-dimensional TEDOR experiment.
[12] was performed with 27 A1 spin frequency
encoding before the initial dipolar dephasing
period. After transfer of the coherence to the 31 P
spins, an FID was accumulated and the ti value
incremented. The two-dimensional correlation
spectrum is shown in Figure 4, revealing crosspeaks
between all three 31 P resonances and both
the resonances from the tetrahedrally coordinated
and octahedrally coordinated 27 A1 sites, in
agreement with the proposed crystal structure and
the results of the two-dimensional CP
experiment discussed above. These results are
taken as demonstrating the general success of the
experiments. A more detailed interpretation of
the results in terms of the individual T-sites from
the present data alone is not attempted.

Vol. 14, No. 1-4 11
oc
"P REDOR
IS C -B -SO -73
ppm from 85% H3PO4
Wrahedral
r AI REDOR
IX 75 X 25 0 -23 -SO
ppm from A1(NO})3
P TEDOR
0 .J5 -JO -75
ppm from 85% H3PO4
Figure 3. One-dimensional REDOR and
TEDOR spectra of 27 A1 and 3I P in VPI-5.
SO 25 0 -25
ppm from A1(NO3>3
Figure 4. Two-dimensional TEDOR spectrum of
27 Al and 31 P in VPI-5.
4 Conclusions
In summary, cross-polarization to and
from quadrupolar nuclei has been experimentally
verified using the 31 P and 27 A1 spin systems in
an aluminophosphate molecular sieve. This
bodes well for the use of heteronuclear
correlations for further investigation of local
micrdstructure in solids. Dipolar-dephasing
experiments have also been accomplished, with
both REDOR and TEDOR results confirming the
connectivities detected by the cross-polarization
studies. A two-dimensional TEDOR experiment
has also been demonstrated that separates
connectivities between distinct resonances.
References
1 S. R. Hartmann and E. Hahn Phys. Rev.,
128, 2042, 1962.
2 A. Pines, M. G. Gibby, and J. S. Waugh /.
Chem. Phys., 59, 569, 1973.
3 J. Schaefer and E. 0. Stejskal J. Am. Chem.
Soc.,9%, 1031, 1976.
4 H. J. Behrens and B. Schnabel Physic a,
114B, 185. 1982.
->A. Samoson. E. Kundla, and A. Lippmaa J.
Magn. Reson.. 49, 350, 1982.
6 C. A. Fyfe, H. Grondey, K. T. Mueller, K. C.
Wong-Moon, and T. Markus /. Am. Chem.
Soc, 114, 5876, 1992.
7 T. Gullion and J. Schaefer J. Magn. Reson.,
81, 196, 1989.
8 Y. Pan and J. Schaefer J. Magn. Reson., 90,
341, 1990.
9 M. E. Davis, C. Sadarriaga, C. Montes, J.
Garces, and C. Crowder Nature (London), 331,
698, 1988.
l0 P. Caravatti, G. Bodenhausen, and R. R.
Ernst Chem. Phys. Lett., 89, 363, 1982.
n L. B. McCusker, Ch. Baerlocher, E. Jahn,
and M. Bulow Zeolites. 11, 308, 1991.
12 C. A. Fyfe, K. T. Mueller. H. Grondey, and
K. C. Wong-Moon, submitted to Chem. Phys.
Lett.

12
1. Introduction
Magnetic Resonance
Evidence for Superconductivity
in a Semimetal
I.P. Goudemond, G. J. HiU and M.J.R. Hoch
Department of Physics and
Condensed Matter Physics Research Unit,
University of the Witwatersrand, Johannesburg
The group V semimetals As, Sb and Bi have
low carrier densities and may be viewed
as rather poor metals. Their electronic
properties have been studied using a variety
of methods, including NQR [1], [2]. At
temperatures below the Debye temperature
the nuclear spin—lattice relaxation rate in As
and Sb has been found to obey the Korringa
relation [1], [2]. For As, this relation has
been found to hold down to 150 mK [3].
In the present work Ti measurements
on As have been extended to still lower
temperatures. Motivation has come from
the interesting electrical conductivity
behaviour found by Uher [4] for a single
crystal sample in the vicinity of 100 mK.
These results may be interpreted as evidence
for a superconducting transition, although no
further experiments appear to have been
carried out to confirm this. Probing of the
superconducting state in zero magnetic field
using NQR methods offers interesting
challenges and opportunities.
2. Experimental Details
The experiments were carried out in an
Oxford dilution refrigerator using procedures
that have been described previously [3].
Pulsed NQR spin echo methods with signal
averaging were used at 23.5 MHz on a
powdered As sample. ( 75 As has I = 3/2 and
100% abundance). The powdered material
was prepared by crushing and sieving (25 \i
mesh) high purity (99.9995%) arsenic,
Bulletin of Magnetic Resonance
followed by annealing in vacuo and further
careful sieving. An oxide layer on the
surface of the grains prevented metallic
contact between neighbouring particles.
Refrigerator
Mixing Chamber
Sintered Silver
Heat Exchanger
Vacuum Space
Coaxial Cable
Helium Fill
Capillary
Copper Block
Cermanium
Thermometer
R F Coil
Stycast Sample
Holder
Figure 1
Sample holder and rf coil assembly for NQR
measurements in the dilution refrigerator. The
sample is immersed in liquid *Be, which is in
contact with the sintered silver heat
exchanger.
Figure 1 depicts the sample
arrangement used to ensure good thermal
contact to the refrigerator mixing chamber.
Liquid 4 He surrounds the sample and is
in contact with a sintered silver heat
exchanger. Further details may, be found
in reference [3]. Fractional 'rf pulses
were used to minimize heating effects.
Temperatures were measured using a
calibrated germanium thermometer mounted
on the mixing chamber.

Vol. 14, No. 1-4 13
3. Results and Discussion
In an effort to establish that the sample
was in good thermal contact with it's
surroundings, careful measurements of the
echo amplitude were made as a function
of temperature down to the lowest
temperatures reached in these experiments,
40 mK. Down to 150 rnK Curie law type
behaviour is observed. At lower
temperatures the data depart from linear
behaviour. We do not believe that this is
due to heating effects or the loss of thermal
contact. Changes in the pulse sequence
repetition rate did not change the amplitude
of the echo signal. It is likely that some
mechanism, characteristic of the sample, is
responsible for the departure from Curie law
behaviour.
At temperatures of 4 K and below, the
skin depth 6 at 23.5 MHz is comparable to,
or less than, the mean particle radius r. We
estimate that 5 ~ .3.0 [im at 150 rnK, while
r < 10 fim. It is clearly desirable that
smaller particles should be used, although
this is difficult because of rapid surface
oxidation which occurs in air and the
tendency of the arsenic particles to sinter
during annealing.
Taking into account the attenuation of
rf pulses and the presence of a core of
undisturbed spins in the particles, we have
examined the situation in some detail. In
order to see whether spin diffusion can
operate between spins near the surface of a
particle and those in the interior, we have
calculated the spin diffusion coefficient
D = V30 V^d 2 , where M2 is the second
moment and d the spin spacing, and obtain
D ~ 2 x 10" 13 cm 2 s"'. On the time—scale of
our experiments (10 2 — 10 3 s) spin diffusion
operates over a distance v/2Dt ~10~ l fim.
It may be concluded that this mechanism
should have negligible effects on our results.
The change in the skin depth with
temperatures below 1 K is of the order of 1
to 2%, which our calculations show will not
lead to detectable changes in the echo
amplitude beyond the Curie law changes.
We do not believe that the departure from
Curie law behaviour is due to effects of this
kind.
The measured spin lattice relaxation
rates are shown as a function of temperature
in Figure 2. Note that, for magnetic
relaxation in an I = 3 /2 system, a unique
relaxation rate may be defined using
1/T] = 6 Wm, where Wn, is the transition
rate between the ±'/2 and ± 3 /2 spin states.
At temperatures down to roughly
120 mK the data obey the Korringa relation.
Below this temperature the relaxation rate
0.09
8.64
0.01 8.62
30 50 70 90 110 130 150 170
T (mK)
Figure 2
Plots of the ?!>As Spin lattice relaxation rate
and of the resistivity of a single crystal
arsenic sample [4] as a function of temperature
down to 50 mK. The straight line which joins
with the relaxation rate plot represents an
extra— polation of the Korringa relation for
arsenic found at higher temperatures.
decreases less rapidly with T than expected
from the Korringa relation. Figure 2 also
shows the electrical conductivity data of
Uher [4] in the same temperature range as
the present measurements. It can be seen
that the conductivity starts to decrease quite
rapidly at ~100 mK. Inspection of the two
plots in Figure 2 suggests a common
underlying physical mechanism for the
changes in behaviour observed in this
temperature range.
It appears likely that a fairly broad
superconducting transition occurs with a Tc
around 100 fiK. The T"i data suggest a
slightly higher Tc value than the
conductivity data.
Cohen [5] has pointed out that the
semimetals may be candidates for
superconductivity through a BCS pairing
mechanism. This is largely because of the
multivalley character of these materials. On
the basis of calculations given by Cohen for
systems of this kind, a Tc of 100 mK is not
unreasonable for As. Doped Bi has been
found by Uher and Opsal [6] to have
a Tc < 100 mK, depending on the
concentration of the Sn or Te dopant.
We have attempted to fit the observed
relaxation data for As using the Hebel—
Slichter expression based on BCS theory.
Anisotropy of the gap is allowed for by
introducing a parameter r = Ao(0)/A, where
Ao is the BCS gap at 0 K and A is a
measure of the gap anisotropy. However the
theoretical curve does not fit the
experimental data plotted in reduced form.
Details will be given elsewhere. It is quite
possible that the suggested transition in As
is not of the standard BCS type. It appears,

14
however, that other effects could be
important.
Cohen [5] suggests that the semimetals
will become type II superconductors with
rather low upper critical fields. In the NQR
experiment rf fields of 30 or 40 G are used
and it is possible that some remnant field
effects may be produced in the sample which
contribute to relaxation.
Further experiments should be carried
out to confirm the onset of
superconductivity in As around 100 mK.
Clearly, Meissner effect measurements
should be attempted. Further NQR
experiments involving CW methods with
low rf fields may prove useful and complementary
to the present measurements.
4. Conclusion
Evidence has been obtained of marked
departures from Korringa relaxation
behaviour in As below 120 mK. Taken
together with previous electrical
conductivity results, it appears likely that
the anomalous behaviour is due to the onset
of superconductivity.
Bulletin of Magnetic Resonance
The relaxation rate measurements are
not in agreement with the Hebel—Slichter
theory predictions. Further work is required
to determine whether this is because of the
non—BCS nature of the transition or some
other cause, such as the magnitude of the rf
fields used in our pulsed NQR. experiments.
5. References
1. J.M. Keartland, G.C.K. Folscher and
M.J.R. Hoch, Phys. Rev. B 43, 8362
(1991).
2. J.M. Keartland, G.C.K. Folscher and
M.J.R. Hoch, Phys. Rev. B. 45, 7882
(1992).
3. IP Goudemond, J M Keartland, and
M J R Hoch, J. Low Temp. Physics 82,
369 (1991).
4. C Uher, J. de Physique, C6, 39, 1054
(1978).
5. ML Cohen, in "Superconductivity",
edited by R D Parks (Marcel Dekker,
New York, 1969), Vol.1.
6. C Uher, J L Opsal, Phys. Rev. Letters
40,1518(1978).

Vol. 14, No. 1-4 15
NMR Spectroscopy in Cardiac Surgery
Roxanne Deslauriers, Sylvain Lareau" 1 ", Rosalind S. Labow+, Wilbert J. Keon + , Gang-Hong Tian # , Anthony L. Panos*,
Carlos A.M. Barrozo*, Imtiaz S. Ali *, Owayed Al-Nowaiser*, and Tomas A. Salerno*
1 Introduction
Institute for Biodiagnostics, National Research Council of Canada, Ottawa
+ University of Ottawa Heart Institute, Ottawa
# Department of Physiology, University of Ottawa, Ottawa, Ont.
* Division of Cardiovascular Surgery, University of Toronto, Toronto, Ont., Canada
Applications of magnetic resonance spectroscopy in
medicine have been restricted mostly to the research
laboratory. The technique is now entering the field of
medical diagnosis and therapy. In the heart, levels of
phosphorus metabolites are often correlated with function.
Nuclear magnetic resonance spectroscopy has been used to
monitor high energy phosphorus metabolite levels in the
heart to evaluate the effect of work and ischemic stress. Our
applications of magnetic resonance to the practice of cardiac
surgical have been in three areas: a) preservation of tissue
for transplantation b) optimization of myocardial protective
techniques (cardioplegia) and c) monitoring of the heart
during the aortic clamping period.
2 Heart Preservation for Transplantation
With increasing demand for a limited number of donor
hearts, organ preservation during procurement is critically
important. Controversy still exists over issues such as the
optimal temperature, optimal solution, and maximum time
limit for donor heart preservation. Numerous studies have
been, and continue to be, conducted on various animal
tissues. Not much data have been obtained in human
myocardium primarily because of the difficulty in obtaining
adequate quantities of viable tissue for laboratory
investigation.
a) Human Atrial Tissue
Portions of human atrial appendages normally discarded
during cannulation in the course of surgery requiring
cardiopulmonary bypass have been used, with informed
patient consent, for studies of heart preservation. We have
used 31 P and ^H NMR spectroscopy to define the optimal
temperature for long-term (up to 24 hours) preservation of
high energy metabolite levels and contractile function, and
to gain a fundamental understanding of the energy
generating pathways in preserved human cardiac tissue [1,
2]. The studies were carried out on a Bruker AM-360
spectrometer. 31 P spectra were obtained using a 60° pulse
and a 1 s recycle time. 1 H spectra were acquired using a
spin-echo sequence based on the water-suppressing 1331
pulse sequence. The acquisition of 31 P and *H NMR
spectra were interleaved (Figure 1). 31 P spectra were used
to measure ATP levels on a continuous basis, as an index of
net energy preservation. *H spectra of lactate provided
information on generation of ATP through the glycolytic
pathway.
10 0 -10 -20 1
PPM
PPM
Lactate
FIGURE 1 31 P (left) and l H (right) NMR spectra of an
atrial appendage (ca. 0.5 g) preserved at 20°C in saline, as a
function of time [1] {ref, reference capillary; PME,
phosphomonoester; Pi, inorganic phosphate).
Studies of atrial appendages preserved at 1°, 4°, 12° and
20°C in physiological saline (0.9% NaCl) for up to 20 hours
demonstrate that preservation of ATP is better at 1° and 4°
than at 12° or 20°C. Based on measurements of lactate
production, glycolysis is active at all the temperatures, its
rate correlating positively with increasing temperature.
However, the ATP generated by glycolysis falls short of

16
demand at all temperatures, but the difference is small at 1°
and 4°C (Table 1).
TABLE 1
Energy balance in human cardiac tissue preserved in NaCl
0.9% [1].
Temp.
(°C)
1
4
12
20
ATP* #
loss
7
8
12
20
Lactate*
production
43
52
106
212
ATP*+
generated
65
78
159
318
ATP* &
utilization
72
86
171
338
* nmoleg" 1 (wet myocyte mass) min' 1 .
* From the rate of change of NMR-visible ATP.
+ Assuming 1.5 mole of ATP produced per mole of lactate
from glycolysis.
& Calculated from (rate of generation of ATP by
glycolysis) + (2 x rate of ATP loss). This takes into
account the ATP generated by adenylate kinase.
In a separate series of studies, we tested the possibility
of improving the maintenance of high energy phosphates at
12°C, one of the higher temperatures currently used in some
institutions for the preservation of heart grafts. Our
hypothesis was that the poor maintenance of high energy
phosphates at 12° and 20°C results from the increased
intracellular acidosis that occurs at higher temperatures [1].
Ultimately, acidosis partially inhibits ATP production by
glycolysis, the only metabolic pathway for generation of
ATP in the anoxic heart. We postulated that the addition of
a buffer to the solution used for heart preservation would
increase the rate of transport of protons and lactate to the
extracellular space, thereby maintaining better intracellular
pH homeostasis. Our studies showed that at 12°C, the halftime
for loss of ATP increased from 300 minutes in saline to
over 900 minutes in a modified Krebs-Henseleit solution
containing 100 mM buffer [2, 3]. This observation was
confirmed independently using biochemical measurements
of high energy phosphates [3]. The beneficial effects of
high buffer concentration observed at 12°C did not occur at
4°C (figure 2 [3]), which lead us to postulate that at that
temperature, glycolysis was rate limited by the temperature
rather than by the acidosis. These studies show the
continued need for testing all the conditions to which a graft
may be subjected, and the importance of avoiding broad
Bulletin of Magnetic Resonance
generalizations when dealing with complex multi-enzyme
systems.
4°C
3-j
Q 0.9% saline (3)
3 • 20 mM PIPES (6)
Ito
o 60 mM PIPES (6)
• 100 mM PIPES (6)
2- —
c/5
•d
1-
0-
0 5
2-
1 -
12°C
EP
dm
10 15
Time
>
20 25
• 0.9% saline (3)
• 20 mM PIPES (6)
o 60 mM PIPES (6)
• 100 mM PIPES (6)
10 15 20 25
Time (h)
FIGURE 2. Effect of the concentration of PIPES buffer on
the preservation of ATP in isolated atrial appendages
preserved at 4° or 12 °C in modified Krebs-Henseleit
solution [3].
The importance of defining the relationship between
high energy phosphate levels and contractile function in
human cardiac tissue led to the development of a
temperature-controlled NMR microprobe incorporating a
perifusion system and a non-magnetic strain gauge (figure
3). The system has been used to study human atrial
trabeculae, which are small functional muscle fibers
weighing 8-25 mg that can be isolated from atrial
appendages. The perifusion system provides the tissue with
the oxygen and nutrients required for its function. The strain
gauge allows for measurement of developed force in

Vol. 14, No. 1-4
electrically stimulated muscle fibers while- NMR
simultaneously assesses the high energy phosphate
compound levels [4]. In addition, the perifusion system can
mimic preservation conditions by allowing muscles to be
perifused at low temperature (down to 1°C) during
acquisition of NMR data (figure 4).
Stimulation wire
I
Muscle in the
NMR tube
Flexible tubing
Reflector
Optical fiber
Fiber optic
strain gauge Perfusate line
Stimulation wire
4
Adjustable
hook mount
FIGURE 3. Schematic drawing of the microperifusion
system used for NMR studies of human atrial trabeculae.
The total length of the system is 6.2 cm.
By studying atrial trabeculae in the presence of
metabolic inhibitors under conditions simulating
preservation, it is possible to assess the contribution of
various cellular mechanisms of energy production to the
total energy balance of the tissue. Our 31 P NMR studies of
isolated human atrial trabeculae [5] preserved at 4° and 12°C
in oxygenated St-Thomas II solution showed that the high
energy phosphates (ATP and phosphocreatine (PCr)) are
well maintained during 18 hours of preservation.
Contractile studies performed under similar conditions
showed high recovery of developed force. Glycolysis, the
only pathway available for energy generation under the
anoxic conditions existing in large preserved organs, is
capable of maintaining ATP levels in hypothermically
preserved tissue. Under anoxia, ATP levels are stable for 6 -
10 hours at 12°C, and for a longer period at 4°C. In a
resting heart, the major energy source is provide by the lipid
catabolism. To test whether this pathway is active at low
• temperatures, we have measured the ability of the oxydative
pathway to maintains ATP levels. At 12°C, when glycolysis
is inhibited by iodoacetate, the oxidative pathway can
maintained ATP levels, but only if an external source of
substrate (10 mM acetate) is present in the perfusate. Thus,
the oxidative pathway is functional but depends on both,
oxygen and glycolysis.
In tissue preserved ischemically (no flow), metabolic
waste products such as lactate cannot be eliminated. This
results in considerable extracellular and intracellular acidosis
[1] which has profound effects on energy production
because glycolysis is inhibited by low pH. Using the atrial
trabecula model, we simulated the conditions that exist in
the ischemically preserved human heart. Such a large organ
(300 - 450 gm) cannot obtain sufficient oxygen to maintain
oxidative phosphorylation simply through diffusion from the
surrounding medium as in the case with trabeculae.
Trabeculae were subjected to acidosis by perifusing the
trabeculae with modified St-Thomas preservation solution
containing 10 mM lactate at pH 6.0. Anoxia was
simultaneously induced with 1 mM cyanide, a potent
inhibitor of oxidative phosphorylation, and nitrogen. In the
trabeculae, these conditions reproduced those previously
observed in the larger atrial appendages where an anoxic
core probably existed [1]. Under acidosis and anoxia at
12°C, ATP decreased linearly by 40 to 100% over a 12 h
period. At 4°C, ATP decreased less over the same time
period.
40
DMMP
1 I •
30
1 I '
20
10
a -ATP Y.ATP
1 I '
-10
-20
PPM
FIGURE 4. 31 P NMR spectrum (147 MHz) of a 10.5 mg
trabecula perifused at 12°C in modified St-Thomas II
solution (60° pulse, 1 sec recycling delay, 2400 scans).
These observations can be reconciled to the following
model of energetics: the maintenance of cellular ATP
depends on matching of supply and demand. At 4°C,
glycolysis appears to be limited by temperature. ATP
regeneration cannot be driven at an adequate rate until the
feedback drive (ADP + Pj) is increased considerably above
the normal level. ATP is then maintained at a low
phosphorylation potential. At 12°C glycolysis is not limited
by temperature but is limited by low intracellular pH.
17

18
Other than gaining a better understanding of energetics
in atrial trabeculae, we can ask whether these studies
provided the transplant surgeon with any information of
practical value? In answer to our initial question on the
optimal temperature for preservation of grafts, we have
provided evidence that, for preservation times of 5 hours or
less, ATP levels are better maintained at 12°C [3]. For
longer preservation times, ATP levels are better preserved at
the lower temperatures. In addition, increasing the buffer
capacity of preservation solutions used at 12°C has a major
impact on maintenance of high energy phosphates.
b) Intact Hearts
Most published NMR studies on heart preservation have
used rodent hearts, with a particularly large number of
studies being performed on the rat heart. We have
developed the isolated perfused pig heart for preservation
studies [6] because it is architecturally, biochemically, and
in size most similar to the human heart. As we discussed
above, provision of oxygen and removal of metabolic waste
products are critically important for long term heart
preservation. Perfusion preservation, which can enhance
oxygen delivery and waste removal from the heart, has not
yet achieved much clinical application and remains mostly
in the realm of the research laboratory. Some of the reasons
for this are related to the implementation difficulties in
situations in which the heart must be transported over long
distances under sterile conditions. In addition, hearts
preserved ischemically for less than 5-6 hours generally
show good recovery of mechanical function after
transplantation.
We have consequently focused our efforts on methods
of improving long-term ischemic preservation of hearts.
The goal is to optimize the conditions currently in use and to
extend the safe preservation time between harvest and
implantation of the donor organ. This should allow harvest
of donor hearts to occur over a wider geographical range and
provide for better immunological organ matching.
The studies of isolated, Langendorff perfused pig hearts
[6] are performed using a Bruker Biospec instrument
equipped with a 4.7 T / 30 cm horizontal bore magnet. The
heart is arrested and isolated using techniques similar to
those used for human hearts. The isolated heart is then
placed in an NMR probe to observe high energy phosphate
levels and pH on a continuous basis, with a two minute time
Bulletin of Magnetic Resonance
resolution, during a hypothermic preservation period that
usually lasts 8 hours. Following preservation, the heart is
rewarmed to 37°C without removing it from the magnet and
NMR spectra are then recorded in the beating heart. A
balloon placed in the left ventricle measures the developed
pressure and serves as an index of functional recovery of the
heart [6]. In this manner, energy levels during and after
preservation can be correlated with functional performance
of the heart following preservation.
A number of technical difficulties arise in trying to
obtain quantitative results from large, isolated, perfused
hearts because they change shape when beating and often
swell when perfused with solutions other than whole blood.
To alleviate the problems caused by the sample moving in a
heterogeneous Bi field and the consequent uncertainty in
received signal strength, a high Bi homogeneity probe was
designed [7]. The prototype probe comprised four separate
tuned rings on a spherical surface (Figure 5) giving a Bi
field homogeneity of ± 5% over 60% of the radius of a 14
cm sphere (Figure 6).
tobalun
coupling ring
FIGURE 5. Geometry of the 4 coil system [7].
The received signal was rendered less sensitive to the
dielectric constant of the sample by distributing the
capacitance around the rings. Inter-ring coupling and to a
fifth ring used for matching was by induction. The coupling
loop was tuned with its own capacitor to Larmor frequency,
thereby ensuring that the probe was always on resonance,
and rendering the tuning and matching independent. In
addition, the use of a low input impedance preamplifier

Vol. 14, No. 1-4 19
virtually eliminated the dependence of signal strength on
coil loading [8].
o V* °
its
/ £
/ -
0/ :g
/ -2-
/ *
7 1
/ I il
0.8-
0.6
0.4'
B1
i
\° \V\°\\V
, . 1 1—-,—*-y
-10 Distance in cm
FIGURE 6 Plot down the Y axis of the Bi field of the
probe (open circles, f = 7 cm at 81 MHz) and the form of the
plot predicted by the theory.
Strategies for improving hypothermic preservation have
ranged from improving the buffer capacity of the
preservation solution [6] to the use of secondary
cardioplegia [6, 9] to maintain the heart in an arrested state
during the entire rewarming phase prior to reperfusion. The
purpose of secondary cardioplegia is to allow the energy of
the heart to be directed towards the re-establishment of ionic
balances that become disrupted by hypothermia and
ischemia, rather than to expend energy in mechanical
function. We have found that the use of secondary
cardioplegia prior to reperfusion does not affect the net
energy levels of the heart but rather eliminates ventricular
fibrillation that is normally observed upon rewarming of
hypotfaermically preserved hearts [6].
As a result of the increasing requirement for donor
organs, a number of organs are frequently harvested from a
single donor. This has led to the need for a single
preservation solution suitable for all thoracic and abdominal
organs. The University of Wisconsin Cold Storage Solution
(UW-CSS, DuPont Pharmaceuticals) is currently in use for
the preservation of liver, kidney and pancreas. Its utility for
heart preservation remains to be determined. We compared
the efficacy of UW-CSS to St-Thomas II solution, which is
in widespread use for heart preservation [10]. Pig hearts
were preserved for 8 hours at 4° or 12°C and then tested
functionally after rewarming to 37°C. These temperatures
were chosen because they are both in use clinically. At 4°C,
UW-CSS and St-Thomas II were equally effective for
preservation of heart function. Figure 7 and 8 shows the
results obtained with UW-CSS and St-Thomas at 12°C. The
lack of functional recovery observed with UW-CSS shows
that this solution is unsuitable for use at 12°C. The results
also demonstrate the necessity for precise temperature
regulation when the solution is used at 4°C. This is not
necessary with St-Thomas II solution because recovery is
not severely compromised by use at either 4° or 12°C.
Ref.
PCr
20 0 -20
PPM
P-ATP
20 0 -20
PPM
20 0 -20
PPM
St-Thomas UW-CSS
4°C
Reperfusion
8 h ischemia
4 h ischemia
2 h ischemia
Reperfusion
8 h ischemia
4 h ischemia
2 h ischemia
FIGURE 7. Typical time course of the 3 l P NMR spectra of
4 hearts, two preserved at 4°C (top panels), two preserved at
12°C (bottom panels) [10]. Spectra on the left were
obtained from hearts stored with St-Thomas II and spectra
on the right were from hearts preserved with UW-CSS. The
ATP and PCr disappeared upon reperfusion in the hearts
stored with UW-CSS at 12°C.
One of the reasons for the failure of UW-CSS in heart
preservation at 12°C could be the calcium paradox. This
phenomenon occurs when a heart has been subjected to a
calcium-free medium (UW-CSS contains no calcium) and
then is reperfused with a solution containing calcium. The
calcium paradox results in massive irreversible damage to
cell membranes, as calcium from the reperfusion medium
overloads the cells. This phenomenon is temperaturedependent
and does not occur readily at low temperatures.
In order to test the "calcium paradox" hypothesis, we added
0.5 mM calcium (0.08 mM free calcium) to UW-CSS and

I
III
20 Bulletin of Magnetic Resonance
repeated our studies at 12°C [11]. Figures 9 and 10 show
the improvement observed in the high energy phosphates
during reperfusion of a heart preserved with UW-CSS
containing calcium.
6000-
•a 5000-
4000-
3000-
Cu_ 2000-
1000-
0
0
6OOO-1
^ 5000-
.9
-M 4000-
1 3000-
2000-
looo H
o-
4°C A u f *
10
12 P C
20 30 40
Time (min)
St-Thomas
UW-CSS
50 60
* *
UW-CSS
0 10 20 30 40 50 60
Tune (min)
FIGURE 8 Time course of the rate pressure product (RPP:
heart rate x developed pressure, a measure of heart function)
during reperfusion (n = 7 in each group) [10]. The
functional recovery was extremely poor in the hearts stored
with UW-CSS at 12°C (* : p < 0.05)
These studies show that NMR spectroscopy can be a
valuable tool in the design and modification of solutions for
protecting the myocardium prior to transplantation. Cardiac
surgery is another area in which NMR spectroscopy is being
used.
3 Cardiac Surgery
Cardiac surgery is being offered to higher and higher
risk patients. This has led to the need for improved methods
of myocardial protection during surgery. Traditionally the
heart is arrested and kept cold (4°C) with one or more
infusions of a crystalloid solution, such as the St-Thomas II
solution described above. The hypothermia and ischemia
associated with the use of cold crystalloid solutions can
impose additional stress on the damaged heart. One of the
most recent modifications to cardiac surgical practice has
been the use of continuous normothennic blood cardioplegia
(CNBC) [12]. The purpose of CNBC is to avoid ischemia.
With CNBC, the heart is maintained at 37°C in an arrested
state by increasing the potassium concentration of a blood
solution that continuously flows through the coronary
vessels. Many questions remain unanswered regarding
CNBC, such as the route of administration (retrograde
and/or antegrade); b) the optimal volume of cardioplegia; c)
the flow distribution of cardioplegia, and d) metabolic
monitoring of the heart during cardioplegic arrest. For these
purposes, a blood-perfused porcine heart model was
developed for NMR studies of CNBC. In this model, the
heart is continuously perfused with blood while being
isolated from the animal. The heart can then be placed in
the NMR magnet and its initial function assessed before it is
arrested in the magnet. NMR surveillance of the high
energy phosphates during CNBC may allow optimization of
flow rates and the route of delivery (antegrade and/or
retrograde) for maintenance of the energy status of the
myocardium. In this context, localized NMR spectroscopy
using either spectroscopic imaging [13] or surface gradient
coils [14], allows assessment of protective techniques in
selected regions of the heart [13, 15] or at various depths
across the heart wall [14], respectively.
40 20 0 -20
PPM
-Calcium
2 hour ischemia
ji. Reperfusioa
40 20 0 -20
PPM
+ Calcium
FIGURE 9 Typical 3 *P NMR spectra of hearts preserved in
unmodified UW-CSS (left panel), or with UW-CSS
containing Ca 2+ (right panel) [11]. The peaks of ATP and
PCr disappeared upon reperfusion in the heart stored with
unmodified UW-CSS.
The difference between the two curves is statistically
significant (p < 0.001) [ll]The NMR technique

Vol. 14, No. 1-4
may determine whether there is a safe time limit during
which blood cardioplegia can be interrupted for surgical
visualization. Initially, we evaluated the effect on cardiac
energetics and function by interrupting the flow of CNBC,
as occurs for instance during aorto-coronary bypass surgery.
Our data show that the high energy phosphate profile
deteriorates and PCr becomes unobservable within 14 ± 2
minutes when flow is interrupted for 20 minutes in the
middle of a 1 hour period of CNBC. This is associated with
decreased left ventricular function when the heart is tested
after reperfusion, in spite of the fact that both PCr and pH
returned to normal within 3 minutes of resuming CNBC
[16].
600O,
5000-1
4000-
3000-
2000-
Ca 2+
1000- Control
10 20 30 40 50 60
Reperfusion Time
(min)
FIGURE 10. Time course of the rate pressure product of
hearts reperfused with UW-CSS containing Ca 2+ (0.5 mM)
and without Ca 2+ (mean ± SD, n=7 per group).
It has been proposed that CNBC could resuscitate the
damaged heart during surgery by providing continuous
delivery of oxygen and nutrients to the heart. In order to test
one aspect of resuscitation, we subjected isolated hearts that
had been previously stressed by a 20 minute period of
normothermic ischemia to two types of cardioplegia and
measured functional recovery following reperfusion [17].
Twenty minutes of normothermic ischemia reduced the ATP
levels, measured by 31 P NMR, in CONTROL hearts (n=6)
to 70 ± 7%. These hearts recovered 86 ± 18 % of theninitial
function (systolic elastance) when reperfused with
normal blood perfusate. The experimental hearts were
subjected to either intermittent cold blood cardioplegia
(ICBC, n=6) for 5 min at 14°C, every 20 minutes, and
repeated 3 times, or CNBC (n=6) for 60 minutes at a flow
rate of 0.5 mL mur 1 g" 1 heart wet weight. Both ICBC and
CNBC prevented exacerbation of the initial ischemic injury.
PCr recovered to initial levels following reperfusion in all
three groups (CONTROL, ICBC and CNBC) indicating that
the mitochondria still possessed sufficient phosphorylating
capacity to maintain appropriate physiological activity. ATP
levels did not recover to initial levels in any of the groups.
This could be related to the loss of nucleotide precursors
from the cells during the initial ischemic period. Functional
recovery with CNBC was 115 ± 30% compared to ICBC
which was 88 ± 9% but there were no significant difference
among the three groups by ANOVA (p>0.05).
Is there a safe period of normothermic ischemia? From
a biochemical perspective, there may be a partial answer.
Using 31 P NMR spectroscopy, we have seen that during an
ischemic episode, PCr decreases before ATP. Theoretical
calculations using the enzymatic equilibria of the creatine
kinase and adenylate kinase reactions support this
observation and have shown that there are two phases of
energy depletion [18]: the buffering phase and the depletion
phase. During the buffering phase, energy is derived from
PCr and the adenine pool is stable. During the depletion
phase, energy is primarily derived from adenine nucleotides.
As ATP is consumed, AMP is produced which subsequently
acts as a substrate for deamination and dephosphorylation
reactions, whose products are lost from the cell. Upon
reperfusion, although the PCr levels may return to normal,
adenine nucleotides may not reach normal levels for a
number of days. To avoid imposing a metabolic stress on
the myocytes, PCr levels should not be allowed to become
depleted to the point where adenine nucleotides will begin to
be depleted. Although the role and critical level of ATP
necessary for recovery of function in the ischemic heart are
controversial, it seems logical to avoid any type of
preventable metabolic stress to the heart during surgery.
NMR spectroscopy is useful for monitoring the energy
depletion and repletion processes in model systems such as
the pig heart. This information can subsequently be used to
verify the predictions of the theoretical calculations.
NMR can monitor PCr and ATP levels directly and
continuously in the isolated perfused heart and in vivo [19].
In our studies, PCr levels reflect the balance of energy
supply and demand; in the arrested heart they decrease and
increase in concert with the availability of oxygen. However
NMR techniques are not compatible with direct use in the
surgical theatre. Recently developed fiber optic pO2 probes
based on oxygen quenching of fluorescence have been used
21

22 Bulletin of Magnetic Resonance
to monitor the arrested heart during surgery. We have
performed NMR measurements on isolated blood perfused
pig hearts and correlated the data with simultaneously
measured levels of pO2 using custom built (Innerspace,
Irvine, Calif.) NMR-compatible probes. We have found a
good correlation between tissue p(>2 (measured in mmHg)
and PCr level in the normal heart. The data are illustrated in
Figure 11. Although the data are preliminary, we see that
PCr levels drop precipitously when the tissue pO2 decreases
below 30 mmHg. By establishing similar relationships in
metabolically damaged or physically abnormal
(hypertrophied for instance) hearts it may be possible to
provide the surgeon with insight into a) the acceptable limits
of oxygen deprivation should delivery of cardioplegia be
discontinued during surgery; b ) the optimal flow rates of
cardioplegia; c) assessment of retrograde versus antegrade
delivery; and d) cardioplegic flow distribution across the
heart.
150, -r. 120
120
0 20 40 60 80 100 120 140
Time (min)
20 40 60 80 100 120 140
pO2 (mmHg)
FIGURE 11. Correlation between the pC>2 and PCr levels
in the normal heart. Top: pC

Vol. 14, No. 1-4
7. D.I. Hoult and R. Deslauriers. A High-Sensitivity, High
Bi Homogeneity Probe for Quantification of
Metabolites, Magn. Reson. Med., 16,411-417 (1990).
D.I. Hoult and R. Deslauriers. Elimination of Signal
Strength Dependency upon Coil Loading - An Aid to
Metabolite Quantitation when Sample Volume
Changes, Magn. Reson. Med., 16,418-424 (1990).
G.H. Tian, G.P. Biro, B. Xiang, K.W. Butler and R.
Deslauriers. The Effect of Magnesium Added to
Secondary Cardioplegia on Postischemic Myocardial
Metabolism and Contractile Function - A
10.
31 P NMR
Spectroscopy and Functional Study in the Isolated Pig
Heart, Basic Res. Cardiol. (in press).
G. H. Tian, K.E. Smith, G.P. Biro, K.W. Butler, N.
Haas, J. Scott, R. Anderson and R. Deslauriers. A
Comparison of University of Wisconsin Cold Storage
Solution and St-Thomas Solution II for Hypothennic
Cardiac Preservation: A 31 P NMR and Functional Study
of Isolated Porcine Hearts, /. Heart Lung Transplant.,
10, 975-985 (1991).
11 G.H. Tian, G.P. Biro, K.W. Butler, B. Xiang, C. Vu and
R. Deslauriers. The Effects of Ca
13
14.
++ Ion on the
Preservation of Myocardial Energy and Function with
UW Solution. A 31 P NMR Study of Isolated Pig Hearts,
/. Mol. Cell. Cardiol., 24, S. 190 (1992).
12. A. Panos, S.J. Kingsley, A.P. Hong, T.A. Salerno and S.
Lichtenstein. Continuous Warm Blood Cardioplegia,
Surg. Forum, 61, 233-235 (1990).
D. Bourgeois and R. Deslauriers. Phasing Spin-Echo
Acquired 31 P Spectroscopic Images Using Complex
Conjugate Data Reversal, Magn. Reson. Med. (in press).
A. Jasinski, P. Kozlowski, A. Urbanski and J.K.
Saunders. Hexagonal Surface Gradient Coil for
Localized MRS of the Heart, Magn. Reson. Med., 21,
296-301 (1991).
15. R. Deslauriers, S. Lareau, G.H. Tian, A.L. Panos,
C.A.M. Barrozo and T.A. Salerno. Surgical
Technology - Applications of Magnetic Resonance
Spectroscopy to Cardiac and Transplant Surgery,
Current Surg., 49, 95-101 (1992).
16. R. Deslauriers, A.L. Panos, C.A.M. Barrozo, O. Al-
Nowaiser, K.W. Butler, N. Haas, K.H. Teoh and T.A.
Salerno. Myocardial Protection: Energy Profile During
Continuous Normothermic Blood Cardioplegia,
Abstracts of the Works-in-Progress, Tenth Annual
Meeting of the Society of Magnetic Resonance in
Medicine, San Francisco, Aug. 10-16,1991, p. 1098.
17. R. Deslauriers, K.W. Butler, N. Haas, C.A.M. Barrozo,
A.L. Panos, I.S. Ali, O. Al-Nowaiser and T.A. Salerno.
The Effects of Intermittent Cold, and Continuous
Warm, Blood Cardioplegia on Isolated Pig Hearts: 31 P
NMR and Functional Studies, Abstacts of the Eleventh
Annual Meeting of the Society of Magnetic Resonance
in Medicine, Berlin, Aug. 8-14,1992.
18.
RJ. Connett. Analysis of Metabolic Control: New
Insights Using Scaled Creatine Kinase Model, Am. J.
Physiol, 254, R949-R959 (1988).
19. P.A. Bottomley, CJ. Hardy and P.B. Roemer.
Phosphate Metabolite Imaging and Concentration
Measurements in Human Heart by Nuclear Magnetic
Resonance, Magn. Reson. Med., 14,425-434 (1990).
23

24
1. Introduction
Topology and Spin Alignment in Organic
High-Spin Molecules
Yoshio Teki, Kazunobu Sato, Masayuki Okamoto, Atsuya Yamashita,
Yoji Yamaguchi, Takeji Takui, Takamasa Kinoshita, and Koichi Itoh
Department of Chemistry, Faculty of Science, Osaka City University,
Sugimoto 3-3-138, Sumiyoshi-ku, Osaka 558, Japan
Organic high-spin molecules are ideal model compounds
for organic magnetic materials such as organic superparaand
ferro-magnets which have recently been attracting
increasing interest [1]. The related experimental [2] and
theoretical works [3] have been done to serve their
molecular design for the last two decades. High-spin
polycarbenes as organic high-spin systems, in spite of their
highly chemical reactivity, are very important from the
view point .of organic magnetism as well as of spin
ordering/spin control in chemistry from the following
reasons, (i) One of the most prominent features of the highspin
polycarbenes is multi-electron open shell systems in
the ground or low-lying excited states which arise from
degenerate delocalized 7t orbitals and from o dangling
orbitals localized at divalent carbon atoms, the latter
orbitals being nearly degenerate with the highest half-filled
K orbitals. (ii) The degeneracy of their n orbitals is
governed by a particular connectivity of n electron network,
i.e. by the topology of the rc electron network.
We have been studying a series of high-spin polycarbenes
as well as their n topological isomers [2a,2c,4-19] which
are designed by exploiting their % electron networks. We
define a Ji-topological isomer as a molecule which differs
from others only in the topology of its TC electron network,
i.e. in the linking positions of its TC bonds. These high-spin
carbenes have been successfully detected up to a tridecet
polycarbene having twelve parallel spins (S=6) by means of
ESR spectroscopy 118].
In this work, we have studied the spin density distributions
of high-spin polycarbenes and their topological
isomers, biphenyl-n,n'-bis(phenylmethylene) (I: n,n'=3,3';
II: n,n'=3,4') and m- and p-phenylenebis(phenylmethylene),
in order to clarify the mechanism of the spin correlation
and its role in the intramolecular spin alignment of organic
systems.
2. Topological Isomers and Their
Spin States
Figure 1 shows the ground states and the low-lying
excited states of the high-spin polycarbenes and their
topological isomers studied in this work. They were determined
from our previous ESR experiments [2a,5,15,
17]. Molecule I has a unique spin alignment in that it has
low-lying high-spin states (S=l and S=2) above the low-
Bulletin of Magnetic Resonance
Figure 1. Topological Isomers and their spin states
S=l
S=0
spin ground state (S=0) as shown in this figure. In
contrast, its topological isomer II has the high-spin ground
state. Both molecules have four unpaired electron spins in
their nearly degenerate nonbonding n and o orbitals. On the
other hand, molecule III is the first organic high-spin
system with the quintet ground state (S=2) [2a], while its
topological isomer IV has the low-spin ground state and the
low-lying triplet excited state located ca. 200 cm"' above
the ground state.
3. Experimental
The syntheses of the diazo precursors of III and IV were
carried out according to the literatures [20,21]. The
synthetic work of the carbon 13 labeled compounds of
biphenyl-3,3'-bis(phenylmethylene) and other precursors
will be published elsewhere. Each diazo precursor was
diluted in a host single crystal of benzophenone-di()- Mixed
single crystals were grown in the dark by slowly cooling an
ethanol or an ether solution containing the diazo precursor.
The polycabenes were generated at 2 - 4 K by the
photolysis of the corresponding diazo precursors. The
photolysis was carried out with an XBO 500W high pressure
mercury lamp using a quartz rod which guided the 405
nm light into an X-band TMon ENDOR cavity. All the
EPR and ENDOR spectra were recorded with a Bruker ESP
300/350 spectrometer equipped with an Oxford variable temperature
controller ESR910. The spectra of the low-lying
excited states were observed under the thermal excitation of
these levels. Other spectra were measured at ca. 2 - 4 K.

Vol. 14, No. 1-4 25
4. Theoretical
In addition to the ENDOR work, we have calculated the
spin density distributions of molecules I - VI using two
model Hamiltonian approaches. The following unrestricted
Hartree-Fock calculation based on a generalized Hubbard
model [8] as well as the exact numerical solution of a
valence-bond Heisenberg Hamiltonian [9] have provided
rather satisfactory and complementary descriptions for spin
structures of organic high-spin polycarbenes.
The generalized Hubbard model Hamiltonian is given by
[22]
ft - -l Zj

26
(a) 34 K
QAt
(b) 15K
24.0
(c) I5K
—I
0.2
QB+
T: Triplet
Q: Quinlet
QB-
0.4 0.6
MAGNETIC FIELD/T
26.0 28.0 30.0
FREQUENCY / MHz
52.0 56.0 60.0 80.0
FREQUENCY / MHz
84.0 88.0
Figure 2. Typical EPR, l H- and 13 C-ENDOR spectra of
molecule I. (a) EPR spectrum, (b) ^-ENDOR spectrum
(c) l^C-ENDOR spectrum. The external magnetic field is
along the a axis. The microwave frequency v is 9413.0
MHz for the spectra (a) and (b), and 9456.0MHz for the
spectrum (c) of the carbon 13 labeled compound,
respectively.
the two end phenyl groups were deuterated, in order to
facilitate the assignment of 1H-ENDOR transitions by
the reduction of the spectral density. By a comparison of
the ^H-ENDOR spectra of the deuterium labeled compound
with those of the normal compound, the eight transitions
labeled by the asterisk in figure 2(b) were assigned to the
protons of the central biphenyl group and the remaining
unlabeled transitions to those of the end phenyl groups.
Thus, we have achieved reliable assignment for all the *H-
ENDOR signals observed.
On the basis of the assignment above, we have
determined the K spin densities on the carbon sites
adjacent to the hydrogen atoms and the n and a spin
densities on each divalent carbon atom from the analysis of
the angular dependence of the ENDOR frequencies as
described in section 5. The experimentally obtained spin
densities are given in figure 3(a). The spin densities on the
six carbon atoms without circles could not be obtained,
since they have no adjacent protons.
We have also calculated the spin density distribution
theoretically by two model Hamiltonian approaches
[8,9,19] as described in section 4. The calculated spin
Bulletin of Magnetic Resonance
density distribution based on the generalized Hubbard model
is shown in figure 3(b). In this calculation, we used the
weakly interacting model in which molecule I, biphenyl-
3,3'-bis(phenylmethyIene), is regarded as composed of two
diphenylmethylene moieties, unit A and unit B, weakly
interacting with each other. This model has shown to
interpret well the observed particular relationship Dj = -
3DQt as described in our previous work [5J. We have
applied this model to account for the spin density
distribution of molecule I. The spin densities p(S,Ms) in
molecule I can be derived from p0(SA=l,Ms=l) of the
isolated triplet diphenylmethylene moieties using the
equation [15,19]
p(S=l,Ms=l)i = (l/2)pO(SA=l,Ms=l)i (6)
for the triplet state, and
p(S=2,Ms=2)j = p°(SA=l,Ms=l)i (7)
for the quintet state. Similar expressions hold also for unit
B since A and B are equivalent. The spin densities of the
isolated diphenylmethylene moieties were obtained from the
UHF calculation based on the generalized Hubbard model.
This calculation well interprets the ENDOR results as
shown in figure 3. The observed and calculated spin density
distributions show that the sign of the n spin density is
alternatively distributed on the carbon sites within the
diphenylmethylene moiety, thus forming the up-and-down
(a) O : positive
0.093
0.105
-0.019 n -086 0.092 -0.072
0.096 0.098
: negative
0126 -0.075
0.095
0.105
0.711 0.711
Figure 3 Spin density distribution of the thermally excited
triplet state of molecule I. (a) Experimental values, (b)
Theoretical values obtained from the UHF generalized
Hubbard calculation based on the weakly interacting model.
(c) Theoretical values obtained by the valence-bond
Heisenberg Hamiltonian approach.

Vol. 14, No. 1-4
network of the K spin. We define the up-and-down network
of the % spin as the pseudo spin density wave (pseudo-
SDW), since in infinite systems the up-and-down spin
network forms a spin density wave. This network resulting
from spin correlation is most favorable in view of the total
spin energy (the sum of the spin exchange-correlation
energy). The observed spin density distribution is, in
principle, symmetrical with respect to the center of the
molecule as predicted by the weakly interacting model
above. Thus, the pseudo 7t-SDW is formed within each
diphenylmethylene moiety. As a result of these, the central
two carbon sites of the biphenyl group (the bridged carbons)
should have the same (negative) sign. This violates the upand-down
pseudo JI-SDW in the whole molecule, resulting
in a node of spin density distribution at the central bridged
carbons. The existence of this node, which unstabilizes the
spin-correlation energy, makes the observed triplet state to
be an excited state above the spin-less ground state. A
similar spin distribution is also expected for the quintet
state from the comparison of eqs. (6) and (7).
Figure 3(c) shows the spin density distribution calculated
by the Heisenberg model Hamiltonian approach where we
replaced the end phenyl groups of molecule I with the
hydrogen atoms to reduce the dimensionality of the
Hamiltonian matrix. In this calculation, we have obtained
the results similar to figure 3(b) without using the weakly
interacting model. In this approach, the spin correlation
is exactly taken into account within the Heisenberg model,
leading to the correct ordering of the ground state and the
low-lying excited states [9,11].
(B) ESR. ENDOR and Spin Density Distribution
of Molecule III
As mentioned above, the lowest spin ground state, i.e.
the singlet ground state, is realized in molecule I, since the
high-spin states are unstabilized because of the existence of
the node at the central bridged carbons of the biphenyl
group. However, if the position of the bridge is shifted by
one carbon site as shown in figure 5, it is expected that the
nodeless pseudo rc-SDW is formed in the whole molecule,
leading to the high-spin ground state as a result of the
stabilization of spin correlation energy. In order to confirm
this, we have observed the ^H-ENDOR spectra of biphenyl-
3,4'-bis(phenylmethylene) (molecule II). It was shown by
our previous ESR experiments that the ground state of
molecule II was a high-spin (S=2) state without low-lying
excited states [17]. The fine structure parameters and the g
value of the quintet ground state were determined as
D=+0.1250 cm" 1 , E=-0.0065 cm" 1 , and g=2.003
(isotropic). Figures 4(a) and (b) show a typical ESR and
H-ENDOR spectra of the quintet ground state of molecule
II observed at 2.7 K. The ENDOR spectrum was obtained
by monitoring the B+ ESR transition (Ms=0*^+l). Three
transitions by o in figure 4(b) correspond to the *H-
ENDOR signals arising from the Ms = 0 spin sublevel and
the remaining unlabeled signals to those from the Ms=+1
sublevel. The eight transitions labeled by the asterisk
are due to the protons of the central biphenyl group.
(a) 2.7K
A-
B-
0.20 0.30 0.40
MAGNETIC FIELD/T
(b)
* *
12.0 16.0 20.0 24.0
FREQUENCY/MHz
Figure 4. Typical ESR and 'H-ENDOR spectra of molecule
II. (a) ESR spectrum, (b) I H-ENDOR spectrum. The
external magnetic field is along the b axis. The microwave
frequency vis 9378.7 MHz.
(a)
Q : positive
0 : negative .o.os8 °- 224
-0.064 0.218
-0.138 0.198
0.206
(b) "205
-0.154 0.230
-0.180 "-I 54
HIT = 2.0. JIT = 0.25
-0.076
0.151
Figure 5 Spin density distribution of the quintet ground
state of molecule II. (a) Experimental values, (b)
Theoretical values obtained from the UHF generalized
Hubbard calculation.
The spin densities obtained from McConnell's equation
- Qpi n /2S and the theoretical values calculated on the basis
of the Hubbard model are given in figures 5(a) and 5(b),
respectively. The observed and calculated spin density
27

28 Bulletin of Magnetic Resonance
distributions indicate that the nodeless pseudo rc-SDW is
formed in the whole molecule as expected from the
topology of the n electron network of molecule II.
These findings give the following physical picture for
the intramolecular spin alignment of molecule II: The unpaired
n electrons are distributed over the carbon skeleton
with alternating the sign of the spin density from carbon to
carbon, thus forming the pseudo SDW in the K electron
network. On the other hand, the two unpaired a spins in
the localized o dangling orbitals become parallel to each
other as a results of the ferromagnetic coupling to the
unpaired n spins at each divalent carbon site, since the onecenter
exchange integral / in eq. (3) between the nearly
degenerate a and n orbitals on the same carbene site is
usually ferromagnetic. Thus, this picture shows that the
spin alignment in molecule II is predominantly determined
by the formation of the pseudo rc-SDW.
(O ESR. ENDOR and Spin Density Distribution
of Molecule III and IV
To demonstrate the role of spin correlation as determined
by the topology of n electron networks, we have also
investigated the spin density distribution in the quintet
ground state (S=2) of the first organic, high-spin molecule,
m-phenylenebis(phenylmethylene) III, and that in the lowlying
excited triplet state (S=l) of its topological isomer, pphenylenebis(phenylmethylene)
(molecule IV). In the
previous work, we reported ESR studies for III and IV
[2a,5] and ^-ENDOR experiments for III [2c]. Figures
6(a)and6(b) show typical ESR and 'H-ENDOR spectra
of III observed at 2 K. The signals labeled by the circle (o)
in the ESR spectrum are those due to a minor byproducts.
The primed pairs A'+ and B'±, and the unprimed pairs A+
and B+ arise from the two magnetically nonequivalent sites
occupied by the guest molecule III in the host single crystal.
A typical ^C-ENDOR spectrum is shown in figure
6(c). We have roughly estimated the n and a spin
densities at each divalent carbon atom from the angular
dependence of the * ^C-ENDOR frequencies in figure 6(c);
the complete analysis by the numerical diagonalization of
the spin Hamiltonian (4) is in progress. The anisotropy of
the l^C hyperfine tensor is about one half of that of
diphenylmethylene reported by Hutchison et al. 126]. This
is due to the projection factor 1/(2S) in eq. (5). Our
preliminary results indicate that the spin densities on both
divalent carbons have values similar to those of
diphenylmethylene. Thus, the sign of the spin densities on
each divalent carbon site is determined to be positive as
expected from the topology of the n electron network of
III, and their n spin densities were estimated to be similar
in magnitude. The K spin densities on the other carbon
sites having adjacent protons were also determined from the
isotropic term of each proton hyperfine tensor using
McConnell's relation. The spin-density distribution
experimentally determined showed that the pseudo 7t-SDW
is formed in the n electron network in a manner similar
to the ground state of II.
A question arises as for the low-lying triplet excited state
of IV whether there exists a node in the n spin distribution,
since II and IV have similar spin structures, i.e. low-lying
high-spin excited states above the low-spin (singlet) ground
state (figure 1). To answer this question, we measured the
^H-ENDOR spectra of the low-lying triplet excited state of
p-phenylenebis(phenylmethylene) (molecule IV). Its typical
ESR and *H-ENDOR spectra are shown in figures 7(a) and
7(b), respectively. The spin density distribution obtained
from the analysis of the angular dependence of the ENDOR
signals is given in figure 7(c). This shows that the four
carbon sites in the central phenyl ring have the same
negative sign in the K spin density, leading to a node in the
7t spin density distribution in the central ring. This finding
confirms the physical picture for the formation of low-lying
high-spin states above a low-spin/spin-less ground state in
topological isomers of high-spin polycarbebes, as described
in section 6(A): The node of the spin distribution
violates the formation of the pseudo rc-SDW in the
whole molecule. This unstabilizes the spin correlation
energy, leading to the observed triplet state above the
singlet ground state.
0.1 0.2 0.3 0.4
8.0 12.0 16.0
FREQUENCY/MHz
(c)
6ab = 5 Temp. 3K
20.0
40.0 44.0 48.0 60.0 64.0 68.0
FREQUENCY / MHz
Figure 6. Typical ESR, 1 H- and l3 C-ENDOR spectra of
molecule III. (a) ESR spectrum, (b) JH-ENDOR
spectrum, (c) 13 C-ENDOR spectrum. The external
magnetic field is along ©ab= 5° from the a axis. The
microwave frequency v is 9430.4 MHz.

Vol. 14, No. 1-4 29
1.2.0
(c)
0.28 0.32 036 ' O40
MAGNETIC FIELD/T
16.0 20.0 24.0
FREQUENCY / MHz
O Positive Spin (P = O.l9i - 0.293)
9 Negative Spin (p = -o.oio~-o.08i)
IF.NDOR measurement
28.0
Figure 7. Typical ESR and ' H- ENDOR spectra and spin
density distribution of the low-lying triplet excited state of
molecule IV. (a) ESR spectrum, (b) JH-ENDOR spectrum.
The external magnetic field is along 0ab= 12° from the a
axis. The microwave frequency v is 9457.0 MHz. (c)
Experimentally determined spin densities.
7. Conclusions
We have investigated the spin alignment in the ground
states and the low-lying excited states of the high-spin
polycarbenes and their topological isomers shown in figure
1. Their spin density distributions were determined by
single-crystal 1 H- and ^G-ENDOR as well as by the
theoretical calculations using the two model Hamiltonians
(the generalized Hubbard model and the valence-bond
Heisenberg model). The results obtained in this work can
be summarized as follows: (1) It is shown that the spin
alignment is highly dependent on the topological nature in
the K electron network. (2) The pseudo JC-SDW governed
by the topological nature plays an important role in the
stabilization of the high-spin ground state. (3) The spin
correlation also plays essential part for the formation of the
low-lying spin states of molecule I and IV, as well as of
the high-spin ground states of molecule II and HI. (4) The
physical picture of the intramolecular spin alignment in
polycarbenes has been clarified in view of spin correlation.
8. References
1. J.S. Miller and D.A. Dougherty, eds.. Proceedings of the
symposium on Ferromagnetic and High-Spin Molecular
Based Materials, 197th ACS Meeting, Dallas, USA (April
9-12, 1989); Mol. Cryst. Liquid Cryst. 176, 1-562 (1989).
2. (a) K. Itoh, Chem. Phys. Lett., 1, 235 (1967). (b) E.
Wasserman et al., J. Am. Chem. Soc, 89, 5076 (1967).
(c) T. Takui, S. Kita, S. Ichikawa, Y. Teki, T. Kinoshita,
and K. Itoh, Mol. Cryst. Liquid Cryst., 176, 67 (1989),
and references cited therein.
3. (a) K. Itoh, Bussei, 12, 635 (1971). (b) N. Mataga,
Theoret. Chim. Acta (Berl.), 10, 372 (1968). (c) N.
Tyutyulkov, G. Olbvich, H. Brenzen, O, Polansky, Theoret.
Chim. Acta, 73, 27 (1988), and references cited therein.
4. T.Takui and K. Itoh, Chem. Phys. Lett.,19, 120 (1973).
5. K. Itoh, Pure & Appl. Chem., 50, 1251 (1978).
6. Y. Teki, T. Takui, K. Itoh, and H. Iwamura, J. Chem.
Phys., 83, 539 (1985).
7. Y. Teki, T. Takui, K. Itoh, H. Iwamura, and K.
Kobayashi, J. Am. Chem. Soc, 108, 2147 (1986).
8. Y. Teki, T. Takui, T. Kinoshita, S. Ichikawa, H. Yagi,
and K. Itoh, Chem. Phys. Lett., 141, 201 (1987).
9. Y. Teki, T. Takui, M. Kitano, and K. Itoh, Chem. Phys.
Lett., 142, 181 (1987).
10. Y. Teki, T. Takui, and K. Itoh, J. Chem. Phys., 88,
6134(1988).
11. K. Itoh, T. Takui, Y. Teki, and T. Kinoshita, J. Mol.
Electronics, 4, 181 (1988).
12. K. Itoh, T. Takui, Y. Teki, and T. Kinoshita, Mol.
Cryst. Liquid Cryst, 176, 49 (1989).
13. I. Fujita, Y. Teki, T. Takui, T. Kinoshita, K. Itoh, F.
Miko, Y. Sawaki, H. Iwamura, A. Izuoka, and T.
Sugawara, J. Am. Chem. Soc, 112, 4047 (1990).
14. M. Matsushita, T. Momose, T. Sida, Y. Teki, T.
Takui, and K. Itoh, J. Am. Chem..Soc, 112, 4701 (1990).
15. M. Okamoto, Y. Teki, T. Takui, T. Kinoshita, and K.
Itoh, Chem. Phys. Lett., 173, 265 (1990).
16. M. Matsushita, T. Nakamura, T. Momose, T. Sida, Y.
Teki, T. Takui, T. Kinoshita, and K. Itoh, J. Am. Chem.
Soc, in press (1992).
17. Y. Teki, I. Fujita, T. Takui, T. Kinoshita, and K. Itoh,
J. Am. Chem. Soc, submitted (1992).
18. K. Furukawa, T. Matsumura, Y. Teki, T. Kinoshita, T.
Takui, and K. Itoh, J. Am. Chem. Soc, submitted (1992).
19. Y. Teki, M. Okamoto, T. Takui, T. Kinoshita, and K.
Itoh, J. Am. Chem. Soc, submitted (1992).
20.R. W. Murray and A. M. Trozzolo, J. Org. Chem., 26,
3109 (1961).
21. S. I. Murahashi, Y. Yoshimura, Y. Yamamoto, and I.
Moritani, Tetrahedron, 28, 1485 (1972).
22. (a)R. M. White, Quantum Theory of Magnetism;
Springer, p. 139 (1983). (b) K. Nasu, Phys. Rev., B33,
330 (1986).
23. S. A. Alexander and D. J. Klein, J. Am. Chem. Soc,
110, 3401 (1988).
24. N. Hirota, C. A. Hutchison Jr. and P. Palmer, J.
Chem. Phys., 40, 3717 (1964).
25. M. E. Rose, Elementary Theory of Angular
Momentum, John Wiley & Sons: New York (1957).
26. C. A. Hutchison Jr., B. E. Kohler, J. Chem. Phys.,
51, 3327 (1979).

30 Bulletin of Magnetic Resonance
New Developments in
Pulsed Electron Paramagnetic Resonance:
Relaxation Mechanisms of
Nitroxide Spin Labels
Colin Mailer, Bruce H. Robinson, and Duncan A. Haas
Department of Chemistry
University of Washington
Seattle WA 98195
Recent technical developments in this laboratory in the areas of pulsed Saturation Recovery Electron
Paramagnetic Resonance (SR-EPR) and Saturation Recovery Electron-Electron Double Resonance (SR-ELDOR)
have enabled experiments with these two techniques to be done with high sensitivity. We have studied the mechanisms
of the relaxation rates of spin labels. The electron spin-lattice relaxation rate Tic' 1 and the nitrogen spinlattice
relaxation rate (Tin" 1 ) of per-deuterated 15 N TEMPOL in glycerol-water solutions have been measured by
SR-EPR and SR-ELDOR. The motional range covered is from a few picoseconds to hundreds of nanoseconds and
the motion is characterized by simple Brownian dynamics. The dependence of Tie 1 upon the rotational
correlation time is explained by a combination of spin rotation and electron -nuclear dipolar coupling mechanisms
plus a proton spin diffusion process. Tin-1 is explained by the electron-nuclear dipolar mechanism and proton
spin diffusion.
1 Introduction
Given the utility of nitroxide spin labels as a probe
of molecular motion in the biological and physical
sciences for the past 25 years it is somewhat surprising
that a clear understanding of their relaxation
mechanisms is still unknown. The quantities
desired are the spin-lattice relaxation rates both
electronic (Tie-i) and nuclear (TV 1 - 14 N or 15 N)
of the nitroxide label. Pulsed EPR can be used to
monitor such relaxation processes directly regardless
of T/2e or line inhomogeneity. In a pulsed
Saturation Recovery (SR) experiment a high rf
field pump pulse is applied for a short time and the
recovery of the magnetization is measured with a
low power observer at the same frequency. The
pulsed Electron-Electron Double Resonance (SR-
ELDOR) experiment has the pump and observer
frequencies quite different (commonly pump and
observer are set to resonate on different spin manifolds).
Huisjen and Hyde [1] pioneered the use
of the SR-EPR technique in liquids and applied it
to a number of systems [2], [3], [4], [5], [6]. The
1 MHuisjen and J.S.Hyde. Rev. Sci. lost., 45, 669-675,
1974.
2 P.W. Percival and J.S. Hyde. /. Mag. Res., 23,249-
257, 1976.
3 T. Sarna and J.S. Hyde. J. Chem. Phys., 69, 1945-
1948, 1978.
Hyde group has used SR-EPR in conjunction with
Continuous Wave (CW) - ELDOR to measure lateral
diffusion of 14 N labelled lipids in bilayers [7],
[8], [9]. Hyde et al. [10] have observed the transfer
of energy from one line to another line in a
CTPO 14 N spin label with pulsed SR-ELDOR.
They clearly saw that cross relaxation took place,
and measured the cross-relaxation rate. We discuss
this experiment below.
Freed and co-workers [11], [12], [13], [14] analyzed
the EPR spectra of peroxylamine disulphon-
10
11
C. Altenbach, W. Froncisz, J.S. Hyde, and W.L.
Hubbell. Biophys. J., 56, 1183-1191, 1989.
J-J. Yin and J.S. Hyde. Zeitschrift fur Physikalische
Chemie. 153, 57-65, 1987.
A. Kusumi, W.K. Subczynski and J.S. Hyde. Proc.
Nat. Acad. Sci. USA, 79, 1854-1858, 1982.
C.A. Popp and J.S. Hyde Proc. Nat. Acad. Sci. USA,
79, 2559-2563, 1982.
J.B. Feix, C.A. Popp, S.D. Venkataaramu, A.H. Beth,
J.H. Park and J.S. Hyde. Biochemistry, 23, 2293-
2299, 1984.
J.-J. Yin, M. Pasenkiewicz-Gierula, and J. S. Hyde
Proc. Nat. Acad. Sci. USA, 84, 964-968, 1987.
.S. Hyde, W. Froncisz and C. Mottley Chem. Phys.
Lett., 110, 621-625, 1984.
J.H. Freed, in Spin Labelling: Theory and
Applications vol I ed. LJ.Berliner. Academic Press
1976. Chapter 3. pp.53-132.

Vol. 14, No. 1-4 31
ate ( 14 N-PADS) and deuterated 2,2,6,6-tetramethyl-4-piperidinone-l-oxyl
(pd-TEMPONE) in
glycerol-water mixture solvents. The linewidths
as a function of temperature were obtained,
corrected for inhomogeneous broadening, and
analyzed according to: l/T2e(M)= A+ B • M +
C-M 2 where l/T2e(M) is the homogeneous
line width of the Mth line (M= -1, 0, +1) in the
three line spectrum. Spectral simulations and
calculation of the A, B and C parameters were
done using Redfield theory [15]. The best
simulated EPR spectral fits to the experimental data
for correlation times (TR) slower than 10" 8 sec
used a non-Brownian spectral density function
This extremely detailed and careful study took the
linewidth analysis method to its limits, but this
alone was not sufficient to completely determine
the motion. The molecular dynamics was inferred
from plots of the A, B, and C parameters versus
each other or viscosity and temperature. Subtle
changes in slopes indicated deviations from
isotropic motion. The simulations to test motional
models were not able to unequivocally determine
the type of motion either. This work presents an
independent test of the relaxation rates and the correlation
times, as estimated by the CW-EPR lineshape
analysis.
2 Experimental
Spectrometer The 9.3 GHz (X-band) pulsed
EPR spectrometer for these studies follows published
designs [16], [17]. There are three arms -
pump, observer and detector bias. The observer
and bias arms act as a conventional high sensitivity
spectrometer for both linear EPR and ST-EPR experiments.
The pump arm. klystron is phase
locked to the observe klystron. For Free Induction
decay (FID) measurements the frequency difference
is zero. SR-EPR and SR-ELDOR experiments
have the pump-observer frequency differ-
12
13
S.A. Goldman.G.V. Bruno.CF. Polnaszek, and J.H.
Freed, J. Chem. Phys., 56, 716-735, 1972.
S.A. Goldman,G.V. Bruno, and J.H. Freed,. /. Chem.
Phys., 59, 3071-3091, 1973
J.S. Hwang,R. Mason,L.P. Hwang, and J.H. Freed, J.
Phys. Chem., 79, 489-511, 1975.
Spin labelling 1 chapter 2 Freed
C Mailer, J.D.S. Danielson and B.H. Robinson. Rev.
Sci. Inst., 56, 1917-1925, 1985.
C Mailer, D.A. Haas, EJ. Hustedt, J.G. Gladden, and
B.H. Robinson, J. Mag. Res., 91, 475-496, 1991.
ence phase locked to a low frequency (MHz) oscillator.
SR-ELDOR operation was simplified by the
use of a Loop Gap Resonator (LGR) [18] instead
of a bimodal EPR cavity. The Q of the LGR is
approximately 300 which gives a 3 dB resonator
bandwidth of 30 MHz. Figure 1 shows the measured
value of rf field versus offset from the 9.3
GHz resonant frequency of the LGR. It is clear
that at the 60 - 70 MHz offset needed for SR-
ELDOR the rf field in the resonator is about 50%
of maximum
20 40 60 80 100
OFFSET FREQUENCY MHz
Figure 1 Plot of 1 mm Loopgap Resonator Bandwidth.
The data are the relative heights of the FID produced by
a short pulse of rf power on the resonator at the frequency
offset from LGR resonance as indicated on the
abscissa. The solid line is the relative rf field amplitude
produced by a resonator with a Q of 300.
The small size of the LGR produces a high power
density leading to high rf fields for moderate powers
(less than 1 Watt).Typical experimental conditions
were: pump power of 200 mW, observer
power 100 nWatt, dead time 50 nanoseconds, acquisition
time 2 nanoseconds/point (or longer) for
1024 points. The pulse repetition rate was about 3
kiloHertz. The decay curves took about 50
seconds to obtain.
CW-EPR We have carried out a CW-EPR and
pulsed EPR study of per-deuterated 15 N TEMPOL
in glycerol-water mixtures. The EPR linewidths in
the fast motion region were measured and corrected
18 Medical Advances, Milwaukee, Wis Loopgap
Resonator Model #XP-0201

32 Bulletin of Magnetic Resonance
Figure 2A. Pulsed ELDOR response of l5 N TEMPOL in 40 % glycerol- 60% water at a TR of 0.015 nanoseconds High field
line pumped with a 250 mW 100 nanosecond duration pulse and the low field line observed with 100 microwatts in the LGR.
Figure 2B. Pulsed. Saturation Recovery. Same experimental conditions as 2A except that pump and observer are both set to
the low field line.
for inhomogeneous broadening using Bales'
method [19] to obtain the motionally dependent
Lorentzian linewidths. These line widths when
subtracted gave the B term of the A(M) = A + B
M + C • M^ expression. (Taking the difference
removed residual Lorentzian effects common to
both lines - such as Heisenberg exchange [20]).
To obtain longer correlation times the high glycerol
percentage samples (> 85%) were cooled to subzero
temperatures and the rotational correlation
time (xR) was estimated from the Stokes-Einstein
equation: tR = V • TJ/T where r\ and T are the
19 B. Bales in Spin Labeling: Theory and Applications,
Biological Magnetic Resonance vol 8 eds. L.J. Berliner
and J. Reuben. Plenum Press NY 1989 Chapter 3.
20 S. Lee and A. Shetty. J. Chem. Phys., 83, 499-505,
1985.
viscosity and temperature of the solvent, and V is
the hydrodynamic volume of the spin probe. This
information calibrates the motion and connects TR
with the temperature and percent glycerol.
Time Domain EPR An example of a pulsed
ELDOR experiment is shown in Figure 2A. The
magnetization from the pumped manifold arrives at
the observer field position at rate Tin-l and then
decays slowly to thermal equilibrium at rate Tie' 1 -
A Saturation Recovery decay is shown in Figure
2B. Some of the magnetization leaves for the
other manifold (at the same rate as it arrived in the
pulsed ELDOR experiment) and the remainder returns
to thermal equilibrium by the Tie process.
The opposite sign of the faster rate in the two experiments
shows that it arises from cross-relax-

Vol. 14, No. 1-4 33
ation between the nuclear manifolds. Superimposed
on the data are the least-squares, best fits
which assume two independent exponential
relaxation components and a baseline. The relaxation
rates of each component, the amplitudes
and the baseline were adjusted using a Marquardt
non-linear least squares algorithm [21].
The electron and the nitrogen spin-lattice relaxation
rates of per-deuterated 15 N TEMPOL in glycerolwater
have been measured with pulsed ELDOR
and SR. The TR range covered is from a few picoseconds
to tens of nanoseconds as indicated on
Figure 3. Also plotted are electron spin-lattice relaxation
rates of *% TEMPOL obtained in the
early pulsed experiments of Percival and Hyde [2].
The Tje" 1 are similar for similar correlation times,
despite the fact that glycerol-water mixtures between
room temperature and -20°C were used in
our experiments, and the Percival and Hyde results
were obtained between -20°C and -90°C with the
label in sec-buty\ benzene. This similarity suggests
;
_
Id U)
to z<
I oUJ
I
UJ
<
,10 -12
10*
10r
2
10
10" 10" 10" 10-8 10"
"""I e is the spectrometer frequency.
Spin diffusion represents the coupling of a paramagnetic
center to the relaxation of distant spins in
the solvent [24], most often protons, and in liquids
the expression has the weak dependence on the
rotational correlation time [25]:
11 jSD = fi / x^ (4)
where P is an adjustable constant The nuclear
spin-relaxation rate is explained with just the electron-nuclear
dipolar mechanism plus proton spin
diffusion:
1/TTOTAL=1/TEND l
ln
with the END term:
((l/xR) 2 +(Ye.A/2) 2 )
(5)
(6)
22 P.W. Atkins and D. Kivelson. J.Chem.Phys., 44,
169, 1966.
23 A. Abragam. The Principles of Nuclear Magnetism,
Oxford University Press. London 1961
24 A.Abragam. loc.cit. p379.
25 B.I. Hunt and J.G. Powles. Proc. Phys. Soc, 88, 513-
528, 1966.

34 Bulletin of Magnetic Resonance
where A is the average of the A tensor. The spin
diffusion term is identical in form to Tie" 1 w i*h a
larger value for p.
We believe that the present interpretation
of the spin lattice relaxation rates, in
large measure, solves the the long-standing
mystery of the nature of Tie and Tin
in spin labels. All the theoretical curves in
Figure 3 assumed simple Brownian motion; there
was no need for modified spectral density functions.
It is important to note that virtually all the
measured values of Tie-i and Tin 4 lie above the
theoretical predictions. It is possible that residual
oxygen in the glycerol-water solutions and spinlabel
concentration effects may have slightly increased
the observed rates.
The maximum rate of T^" 1 (upper curve ) occurs
when^ the motional frequency TR* 1 is equal to
Ye • A / 2. Any deviation from Brownian motion
would have reduced this maximum rate from the
value found. The maximum value thereby becomes
an extremely sensitive indicator of the type
of motional process. The shape of the nuclear relaxation
curve is quite different from that found by
Hyde et al. [10] in their pulsed ELDOR work.
The value of Tin" 1 found by them was about one
MegaRadian/sec and was approximately independent
of correlation time. Unfortunately these
workers used a high Q EPR cavity which put an
upper limit of a few MHz on the response bandwidth
of their spectrometer, rendering the fast
Tin 1 unobservable. Our pulsed ELDOR results
show that the actual Tin' 1 is about a factor of ten
shorter and clearly varies dramatically with the
motion.
In the study of Yin et al. [9] the lipids were
moving sufficiently fast (TR about 50 picoseconds)
that the Tin- 1 (~ 1.3 usec 1 ) was competitive with
Tie 1 and with the 0.1-1 MHz exchange rate
typical of lateral diffusion. The three rates differ
by about factors of two and the data contains all
three decays with the same sign. However,
slower motion of the lipids (such as due to a phase
transition) would lead to a rapid increase in Tin 1
and make the faster exponentials difficult to
distinguish. The only way to overcome this
difficulty would be to perform pulsed ELDOR
experiments, thereby changing the signs of one of
the faster exponentials.
A comment on using CW progressive saturation
techniques for obtaining Tie 1 : CW-EPR assumes
slow passage through the resonance line. The
typical 100 kHz Zeeman modulation frequency
used for detection must be much slower than
Tie" 1 . The long Tie's found for motions slower
than 10 nanoseconds violate this condition. This
is one of the reasons why CW-EPR saturation experiments
are difficult to perform: the modulation
acts as a relaxation mechanism. Knowledge of the
correct spin-lattice relaxation rate is important in
order to avoid operating in saturation.
4 Conclusions
We have determined the electron and nuclear spinlattice
relaxation rates in a nitroxide spin label.
The rates found agree well with those predicted assuming
isotropic Brownian motion. The mechanisms
of spin rotation, proton spin diffusion and
electron-nuclear dipolar coupling appear to explain
the spin label spin-lattice relaxation mechanisms
quite satisfactorily. Instrumental development is
clearly at a stage where the full range of Tie" 1 and
Tin" 1 rates can be studied with relative ease. We
note that in this fast motional regime both spin lattice
relaxation rates are approximately simply proportional
to the correlation time but have opposite
functional dependencies: T," 1 «= TR and T^" 1 «=
TR" 1 . Pulsed SR-ELDOR is much more directly
able (than CW-EPR) to connect the experimentally
determinable parameters (spin lattice relaxation
rates) to the general and powerful methods of
interpretation in terms of motional correlation
functions and the associated spectral density
functions.

Vol. 14, No. 1-4 35
New Developments in
Pulsed Electron Paramagnetic Resonance:
Direct Measurement of Rotational
Correlation Times from
Decay Curves
Duncan A. Haas, Colin Mailer, Tetsukuni Sugano and Bruce H. Robinson
Department of Chemistry
University of Washington
Seattle WA 98195
Perdeuterated ^N TEMPOL in glycerol-water mixtures and spin-labeled hemoglobin are studied using the time
domain technique of saturation recovery electron double resonance (SR-ELDOR), in the ultra-slow motional time
regime. The electron spin lattice relation rate (Tig" 1 ) and the nitrogen spin-lattice relaxation rate (Tin* 1 ) are determined.
Moreover, the characteristic rotational correlation times (TR) are measured directly. It is not possible,
in general, to separate uniquely these rates from one another with Saturation Recovery Electron Paramagnetic
Resonance (SR-EPR) experiments alone. The SR-ELDOR technique of pumping one spin manifold or orientation
and observing at another changes the sign of the amplitudes of selected components contributing to the overall
signal. Pooling the SR-EPR and various SR-ELDOR spectra enables all rates to be uniquely determined.
1 Review of Isotropic motion
Perdeuterated 15 N-Tanol [1] and spin-labeled
hemoglobin (sl-Hb) [2] have been used as model
systems undergoing Brownian isotropic rotational
motion in the micro to millisecond time range as
described by the characteristic rotational correlation
time, (TR). The continuous wave technique of
Saturation Transfer EPR (ST-EPR), which is very
sensitive to motion on this time scale [3], relies on
the competition among the different magnetization
transfer mechanism of Tie, Tin, TR and the
Zeeman modulation frequency, com. There are
many applications of ST-EPR to slow motion,
primarily using 14 N spin labels. However, the
relaxation and rotational rates can only be indirectly
inferred from ST-EPR [4]. We will show
that these rates may be directly obtained from SR-
ELDOR experiments.
L.R. Dalton, B.H. Robinson, L.A. Dalton and P.
Coffey, in Advances in Magnetic Resonance VIII,
(J.S. Waugh, ed.) Academic Press, NY, 1976.
D.D. Thomas, L.R. Dalton, J.S. Hyde, J. Chem.
Phys., 65, 3006-3024, 1976.
J.S. Hyde and L.R. Dalton, Chem. Phys. Lett., 16,
568-572, 1972.
A.H. Beth, and B.H. Robinson, in Biological Magnetic
Resonance VIII Spin Labeling: Theory and
Application (L J. Berliner and J. Reuben, eds.),
Plenum Press, NY, 1989.
Hyde and co-workers invented the technique of
Continuous Wave Electron-Electron Double Resonance
(CW-ELDOR) [5]. They demonstrated that,
by pumping one point in a spectrum and observing
at another point, energy transfer could be detected.
However, CW-ELDOR is an extremely complex
technique and the results can be influenced by instrumental
artefacts [6]. The technique has been
applied by Stetter et al. [7] to nitroxides moving
with sub-nanosecond times and by Smigel et al.
[8] to nitroxides with microsecond correlation
times. From CW-ELDOR, one cannot estimate
either Tie or Tin independently, only their ratio.
The Stetter [7] work used different isotopes of nitrogen
to separate exchange due to lateral diffusion
from nitrogen relaxation. (There can be no nuclear
spin lattice relaxation between different isotopes so
any interaction between them must come from
collisions.)
J.S. Hyde, J.C.W. Chien and J.H. Freed, J. Chem.
Phys.,4S, 4211-4226, 1968.
L.A. Dalton and L.R. Dalton, in Multiple Electron
Resonance Spectroscopy eds. M.M. Dorio and J.H.
Freed. Plenum Press 1979 Chapter 5.
E. Stetter, H.-M. Vieth, and K.H. Hauser, J. Mag.
Res., 23, 493-504, 1976.
M.D. Smigel, L.R. Dalton, J.S. Hyde, and L.A. Dalton,
Proc. Nat.Acad. Sci. USA, 71, 1925-1929, 1974.

36 Bulletin o[ Magnetic Resonance
Time Domain EPR: In a time domain EPR experiment,
the spins are subjected to a short pulse
of microwave power; the time dependence of the
transient relaxation after the pulsing is monitored.
The rates of relaxation are functions of the motional
process. Two types of time domain experiments
exist: one monitors spin echoes (a coherent
detection of the x- and y-components of the magnetization)
and the other monitors Saturation
Recovery (as a polarization detection of the z-magnetization).
Freed [9] has applied spin-echo methodology to
study motion in liquids. A (90°-T-180°-i-Observe)
sequence measures echo height as a function of the
delay time t to give the phase memory decay time
TM- Other workers have studied spin-labeled
membranes and vesicles - showing changes in TM
with temperature, but do not relate results to any
theory [10]. Freed [9] has also developed the
technique of Fourier Transform EPR using
multiple pulse techniques, originally developed for
NMR. The variation of TM across the spectrum
can be directly obtained and related to motional
models. Cross-relaxation from one manifold to
another can also be measured. Rates are measured
from the volumes of the main and cross-peaks in
the 2-D display, and not from actual decay curves.
Huisjen and Hyde [11] pioneered the use of the
SR-EPR technique in liquids and applied it to a
number of systems [12], [13]. The Hyde group
[14] has used SR-EPR, in conjunction with CW-
ELDOR, to measure lateral diffusion of 14 N labelled
lipids in bilayers. These important papers on
translational motion set the stage foT our work in
rotational motion.
Measurement of rotational diffusion directly in the
time domain with SR-EPR alone has proved im-
9 J. Gorcester, G.L. Millhauser, and J.H. Freed, in
Modern Pulsed and Continuous-Wave Electron Spin
Resonance, L. Kevan and M.K. Bowman(eds.) John
Wiley 1990 Chapter 3.
10 K. Madden, L. Kevan, P.D. Morse and R.N. Schwartz,
/. Am. Chem. Soc, 104, 10, 1982.
11 M. Huisjen and J.S. Hyde, Rev. Set Inst., 45, 669-
675, 1974.
12 P.W. Percival and J.S. Hyde, /. Mag. Res., 23, 249-
257, 1976.
13 A. Kusumi,W.K. Subczynski and J.S. Hyde, Proc.
Nat. Acad. Sci. USA, 79, 1854-1858, 1982.
14 J.-J. Yin, M. Pasenkiewicz-Gierula, and J.S. Hyde,
Proc. Nat. Acad. Sci. USA, 84, 964-968, 1987.
possible. Typical of an SR-EPR experiment is that
of Fajer et al. on sl-Hb [15] which produced biexponential
decays. The slower rate was
(correctly) identified with Tie" 1 and the faster decay
rate was associated with the rotational motion
of the hemoglobin. The faster decay rate remained
approximately constant despite a two order of
magnitude change of correlation time. These authors
then used a qualitative theory of spin diffusion
to extract an effective correlation time. The
data agreed only approximately with the known 1R
values. The problem with this type of experiment
is the strong influence of the nitrogen nuclear spinlattice
relaxation, on the decays. As explained below,
the observed decays were a mixture of all
three relaxation processes. This makes the protocol
for data collection and for analysis of the various
decays much more complex than originally
expected. Attempts in our laboratory using
Saturation Recovery alone were similarly unsuccessful
(unpublished experiments with Dr. P.
Fajer). We believe we have now solved these
fundamental problems using pulsed SR-ELDOR
and SR-EPR together.
We will demonstrate that Pulsed EPR techniques,
such as SR-EPR and SR-ELDOR, can be used to
obtain both the nuclear and electronic spin-lattice
relaxation rates directly and monitor changes in exchange
rates. Moreover, the correlation time connecting
two spectral positions can be directly measured
as well. Relaxation time measurements, especially
nuclear spin-lattice relaxation (Tin). can
also characterize the motion in the ultra-slow motional
time range.
2 Theory
Figure 1 shows an absorption CW-EPR signal
when the correlation time is longer than 0.1 microseconds.
The figure illustrates the dependence of
the signal on molecular orientation. The reorientation,
characterized by a rotational correlation time,
TR, moves the magnetization around within a given
manifold. Nuclear spin flips, occurring at rate
T^" 1 , move the magnetization from one manifold
to the other without causing molecular reorientation.
Regardless of orientation or manifold, the
magnetization relaxes to the lattice at rate Tie'
15 P. Fajer, D.D. Thomas, J.B. Feix, and J.S. Hyde,
Biophys. J., 50, 1195-1202, 1986.

Vol. 14, No. 1-4 37
Theory [4], [16] predicts for the SR-ELDOR experiment
that the TR of a molecule should appear as
an exponential decay with a rate (TR" 1 + T^" 1 ) and
that Tin should appear as a decay of rate (T\a' 1 +
T^" 1 ). These two processes have very similar
rates, which is why motional rates have been so
difficult to determine. However, suitable positions
of pump and observer can be chosen so that the
signs and the amplitudes of both Tin and the motional
decay curves can be individually changed,
enabling analysis to disentangle the rates. The SR-
ELDOR experiment measures the decay of the
transient component of the z magnetization,
(Mz(t;v,8)), where v is the nuclear manifold
(which is ±1/2 for 15 N), and 6 is the orientation,
as shown in Figure 1. The evolution of the magnetization
can be understood, qualitatively at least,
by a simple population analysis treatment
Lattice!
^Figure 1: Linear Absorption EPR spectrum of sl-Hb
illustrating the positions where SR-ELDOR experiments
are performed and the dependence of the resonance
'positions of the magnetization on spin-label orienta-
°" The figure further illustrates how TR, and Tin
! spectral diffusion, whereas Tie induces true spin-
"J relation.
^•Sugano, C. Mailer, and B.H. Robinson, /. Chem.
zfkys. 87, 2478-2488, 1987.
Therefore the master equation is:
{(M2(t;v,e))-(Mz(t;-v,0))}
-DV2 (Mz(t;v,e))
(1)
where D is the Einstein Rotational Diffusion
Coefficient (D = 1/6 tR) and V 2 , is the angular
Laplacian operator. The system is pumped with a
weak selective pulse so that the spin system is
excited in manifold vp and at orientation 8p. The
the observer frequency is set to manifold v0 and at
orientation 80. The recovery signal then has the
approximate form:
(Mz(t;vo,eo vp,ep)) =
* (l + P2(cos(9o)) • P2(cos(8p))e- t/x R)
where is the magnetization at
the observer position subject to the condition that
all of the magnetization was at the pump position at
time zero; and (Mz(O;vp,0p)) is the initial magnetization
at time zero (when the pumping is completed).
P2 (cos(8)) = (3 cos 2 (8) - l)/2 is the I =
2 component of the Legendre polynomials P;(x),
and fvy = +1 if v = v' and fvv- = -1 if v = -v'
(which is the case of going from one manifold to
the other). This result assumes a short duration
pump time, a low observer amplitude and that the
higher rotational functions are not significant [17].
Equation 2 predicts that one will observe 4 exponentially
decaying components, wherein the rate of
each component is a linear combination of T^" 1 ,
Ti,," 1 , and TR" 1 . The amplitude of the components
containing Tin" 1 , may be switched in sign by setting
the pump and observer positions to different
spin manifolds. The amplitude of the components
containing TR" 1 may be switched in sign by setting
the pump and the observer to different ends of the
same manifold.
17 T. Sugano, "A Study of Very Slow Rotational
Diffusion by SR-EPR", Ph.D. Thesis, University of
Washington, 1987.

38 Bulletin of Magnetic Resonance
3 Experiment
The details of the spectrometer and its performance
are discussed more fully in another paper in these
Proceedings [18]. SR-EPR and SR-ELDOR experiments
are both polarization experiments which
measure the recovery of the system to equilibrium.
A selective pulse is applied to the spins which alters
the polarization and burns a partial hole in the
line. Under conditions of SR-EPR the frequency
of observation is the same as the pump and therefore
the signal is that of a pure recovery as the polarization
spreads throughout the spin system and
out to the lattice. Under conditions of SR-ELDOR,
the frequency of the observer is different from that
of the pump and the arrival of magnetization, as
well as relaxation to equilibrium, are both detected.
It is neither necessary, nor desirable, for the pump
to be coherent with the observer. The experiments
obtain decay curves taken with various pump times
and pump-observer frequency differences; the rotational
correlation time is determined by the solvent
viscosity and temperature. The pump time
controls the relative amplitudes of the various components
which contribute to the overall relaxation
of the magnetization; and the pump-observer frequency
difference controls the signs of the amplitudes
of the exponentially decaying components in
ways predicted by equation 2.
The experiment proceeds as follows: for one set of
experimental conditions, a number of decay curves
over different time scales are obtained. These decays
are linked by the fact that the exponentials
which comprise them are all identical and only the
time range of data collection is altered. For a different
set of conditions e.g. in pulse length or in
observer field position, the relative amplitudes
and/or signs of the individual decays are altered
and another set of linked spectra is collected Both
sets of linked data are then pooled with the constraint
that the rates of the exponentials which
make up the spectra are common to all and the
amplitudes of each component within a linked set
are the same. This technique [19], as an
18 C. Mailer, B.H. Robinson and D.A. Haas, "New
Developments in Pulsed EPR: Relaxation Mechanisms
of Nitroxide Spin Labels", [these proceedings] 1992.
19 J.M. Beecham, E. Gratton, M. Ameloot, J.R.
Knutson, and L. Brand, in Fluorescence
Spectroscopy: Principles J.R. Lakowicz (ed.) Plenum
Press NY Volume 2 Chapter 5,1991.
application of Global Analysis, is well recognized
in optical spectroscopy as being extremely effective
in determining relaxation rates.
Once one has obtained a unique set of decay rates
that minimizes the global x 2 one must identify the
processes which gives rise to each individual decay
rate i.e.whether it is due to Tie" 1 , Tin" 1 o f rotation.
Suitable choice of experimental conditions
for acquisition of the decays enable the rates to be
distinguished. The strategies we have found useful
are:
(i) Pumping and observing at the magic angle
will reduce the amplitude of the motional rate,
leaving just the Tie and Tin" 1 .
(ii) Pumping in one spin manifold and observing
in the other will always invert the amplitude of
a Tin containing term.
(iii) Pumping at one extreme turning point of a
spin manifold and observing the other will invert
the amplitudes of the motionally dependent components.
(iv) Suppression of free induction decay (FID)
is always important. In the nanosecond range of
motion the FED has approximately the same time as
Tie and Tin.
(v) A pump time that is long compared to the
relaxation time of a particular component will tend
to suppress the amplitude of that component.
(vi) Pooling the Saturation Recovery EPR and
pulsed ELDOR decay curves for analysis is essential
for all rates to be found - no single experiment
alone is sufficient.
4 Results
We have obtained correlation times from perdeuterated
15 N TEMPOL in glycerol-water mixtures
over the TR range from 0.1 to 100 microsec- .
onds: Bi-expOnential recovery curves were ob-1
tained when using the SR-ELDOR protocol de- ^
scribed above, when the "magic angle" (8 = 54°);l
was chosen for the pumping and observing posi-Jj
tions (as shown in Figure 1). In Figure 2 are
plotted the Tie' 1 and Tin" 1 rates versus correlation
time in the slow motion range. The very weak;
dependence of T^' 1 on correlation time suggest|
that the more traditional relaxation mechanisms"

Vol. 14, No. 1-4
such as spin rotation and Electron-Nuclear-Dipolar
coupling (END) are not the dominant ones. The
power law dependence of Tie' 1 is on the order of
1/TR 1/8 . The NMR literature suggests that a model
of spin diffusion in liquids will have the same
power law dependence as that experimentally
observed [20]. This same power law dependence
was also observed in similar measurements using
SR-EPR [15]. The solid line superimposed on the
Tie" 1 data is defined as:
(coexR)
(3)
where fie is an adjustable parameter, and the second
term (r.h.s.) is the END term and the third term
u
q>

40
SR-EPR: «B
Bulletin of Magnetic Resonance
Figure 3: Three different SR-ELDOR spectra, taken at different pump and observation positions (see Figure 1). The top
curve is the SR-EPR spectrum at position B (in Figure 1), the middle curve is the SR-ELDOR spectrum pumping at position
A and observing at position B, and the bottom spectrum is the SR-ELDOR spectrum pumping at position B and observing
at position C. Superimposed on each spectra is the least-squares best fit consisting of an adjustable baseline and three
components each of which is a single exponential decay (see equation 2), the rates are given in Figure 2.
Figure 3 shows an example of the SR-EPR and
SR-ELDOR data for sl-Hb tumbling with xR in the
1 to 5 microsecond time range. The top curve,
(SR-EPR at B) is the SR-EPR data acquired at
point B (defined in Figure 1). This curve is a
simple recovery and all components have the same
sign. The bottom curve (SR-ELDOR from B to C)
is the data when the pump is set on B and the observer
frequency is set to C. This represents a
jump from one manifold to the other and the amplitude
of the components, containing a Tin" 1 in the
rate, change sign. The middle trace of Figure 3
(SR-ELDOR from A to B) is the data for the pump
on A and the observer on B. This corresponds to
the pump-observe case within a manifold.
According to equation 2, the amplitude of the
components containing TR in the rate will change
sign. Clearly there is a peculiar dip in the shape of
this SR-ELDOR curve. The dip is characteristic of
the component that depends directly on the correlation
time. Superimposed on each of the data sets
is a least-squares best fit simulation composed of
three exponentials and a baseline. The three rates
of the exponentials are the same in all three data
sets. The rates arc interpreted, according to equation
2, as Tie* 1 , Tin" 1 and TR' 1 as 0.0937, 1.983
and 0.329 MRad/sec respectively with around a|
10% error. Notice that the motional rate is in be-.|
tween the spin-lattice relaxation rate and the nu-|
clear relaxation rate. These data are also shown inf
Figure 2. The nominal correlation time, d cte f1
mined from solvent viscosity and hydrodynami

Vol. 14, No. 1-4 41
radius of sl-Hb, is 1.5 |j.sec. From the L"/L ratio
calibration curve of the ST-EPR [4] one may estimate
the correlation time to be 4.0 p.sec and the
rotational correlation time measured by SR-
ELDOR is 3.3 jisec, which is in excellent agreement
with the ST-EPR calibration data. This
work, demonstrates that the different components
of the experimental curve can be uniquely identified
and their rates quantitatively measure, and
hence Tie, Tin and TR may be directly measured.
This is the first time that xR has been directly
measured in EPR and represents a
fundamental advance in time domain
methodology.
5 Conclusions
We have measured the values of the electron and
nuclear spin lattice relaxation times in the ultraslow
motion region using SR-ELDOR and SR-
EPR and find that the measured values of Tie' 1
depend on 1/TR 1 / 8 , or a power law dependence,
which is consistent with a model of spin diffusion
in liquids. The measured values of Tin are well
described by the electron-nuclear dipolar relaxation,
(which has no adjustable parameters in it)
and only the rates at motional times longer than
100 fisec suggest the need for a spin diffusion
mechanism for Tin as well as Tie. Notice that for
the case of spin labels the nuclear relaxation is
much faster than the electron relaxation (an unusual
situation). This is primarily a result of the
ability of the electron, at this particular rotational
rate, to efficiently relax the nucleus and the fact
that the electron has very few other spins to relax
it. By using SR-ELDOR and by pooling the data
at different orientations the rotational correlation
time can be measured directly, and is seen experimentally
as a single exponential relaxation. This
is, we believe, the first time rotational motion has
been quantitatively measured as a single exponential
decay in this type of experiment. These exper-
- imental results suggest that it may be possible to
. directly measure rotational reorientation by SR-
, ELDOR even for systems characterized by aniso-
> tropic motion.
M-
|r. These type of time domain experiments are a necunderpinning
and improvement on tradi-
CW techniques. Quantitative simulation of
f-EPR spectra requires knowledge of the relax-
ation times of the spin system. When performing
progressive saturation studies using CW-EPR, the
experiments can only be used to detect relative
changes in relaxation times when motion is longer
than a nanosecond [22]. (Absolute values are difficult
to obtain accurately because the effective relaxation
time for the CW experiment is a complex
mixture of competing relaxation processes.) Direct
measurement of Tie" 1 an d Tin w ^h pulse techniques
is clearly superior.
Prior to the time domain experiments it was found
in this laboratory that the Tin* 1 value in calculations
of ST-EPR spectra in the microsecond motional
range had to be set artificially high for good
agreement between simulation and experiment
[23]. The simulation assumes that the electronnuclear
dipolar (END) interaction is the sole mechanism
for nuclear spin-lattice relaxation. The data
in Figure 2 clearly show that other mechanisms
add to the END rate (e.g. from proton spin diffusion)
justifying the ad hoc addition of an extra rate
into the calculations.
22 B.H. Robinson, C. Mailer and D.A. Haas, Biophys. J.
61, A167, Abstract # 960, 1992.
23 D. Haas, R. St Denis, C. Mailer, B.H. Robinson, 13th
Intl. EPR Conf., Denver, CO, 1990.

42 Bulletin of Magnetic Resonance
Non-Linear effects in Standard 2D NOE
experiments in Coupled spin systems
R.Christy Rani Grace* and Anil Kumar*t
''Department of Physics and ^Sophisticated Instruments Facility
Indian Institute of Science, Bangalore - 560 012, INDIA.
INTRODUCTION
The nuclear Overhauser effect (NOE)
which monitors the transfer of magneti-
zation from one spin to another, is criti-
cally dependent on the internuclear dis-
tance and has therefore become a pow-
erful tool for elucidation of the struc-
tures of Biomolecules. Experimental
methods for monitoring these effects of-
ten use radio frequency pulses which si-
multaneously excite and/or detect sev-
eral spins at a time. If the spins are
not coherently coupled (no J coupling),
there are no non-linear effects of the
pulses, except for a scaling factor. The
non-linear effects in the presence of J-
coupling for one-dimensional NOE ex-
periments are well known(l,2). In this
paper the non-linear effects in the 2D
NOE (NOESY) experiment are anal-
ysed in detail.
The standard NOESY experiment
uses the sequence 90° — ii — 90° — rm —
90° —12, in which relaxation takes place
during the mixing interval rm . The
rate equations governing relaxation are
exactly identical to the transient NOE
experiment(3-5). It has been known
that for uncoupled spins each cross-
section in the NOESY experiment is
equivalent to a ID transient NOE exper-
iment in which the peak corresponding
to the diagonal peak is selectively in-
verted^). When there are J-couplings
present in the spin system, selective in-
version has to be carefully defined. Re-
cently, it has been shown that for small
values of the second pulse (90° — t\ —
a — Tm — 90° — t2 at frequency u>i = ua is
equivalent to a ID difference transient
NOE experiment in which the transition
at frequency ua is selectively inverted.
This is true irrespective of the strength
of the coupling(6,7). It has also been
shown for weakly coupled spins that in

Vol. 14, No. 1-4 43
the standard NOESY experiment, any
cross-section parallel to CJ2 at u\ = u>a, is
equivalent to a ID transient experiment
in which, the whole multiplet of which
ua is a part is non-selectively inverted.
When the spins are strongly coupled the
90° pulse distributes the perturbation
over all the transitions of the strongly
coupled network and the 2D NOE ex-
periment is not equivalent to any stan-
dard transient ID experiment. In ad-
dition, the third pulse in the NOESY
experiment (the measuring pulse) mea-
sures the state of the spin system in a
non-linear manner for finite angles. As
a result it is shown here that in strongly
coupled spin systems one can obtain
'cross-peaks' in the standard NOESY
experiment without relaxation. The ori-
gin of these cross-peaks in terms of the
non-linearity of the second and/or the
third pulse is also discussed with the
help of an ABX spin system.
Cross-correlations between pairwise
dipolar relaxation and between dipolar
and other mechanism of relaxation such
as chemical shift anisotropy(CSA) are
known to yield a multiplet effect in J-
coupled spectra(7-ll). A measurement
of this effect in one and two dimensional
spectra is carried out using small angle
pulses. Recently Osckinat et al. have
used small angles for the second and
the third pulses in the NOESY experi-
ment and have shown that in the initial
rate approximation the effect of cross-
correlations is present in all the mul-
tiplets of an AMX spin system(7). In
their experiment the direct pumping ef-
fects and cross-correlation effects both
give rise to multiplet effects. We pro-
pose here simple modifications which al-
lows the direct pumping effects to be
absent, with the cross-correlations ex-
clusively exhibiting multiplet effects in
weakly coupled spins.
A. STRONG COUPLING
INDUCED CROSS-PEAKS IN
NOESY
The signal in a NOESY experiment
utilizing 90° — ti — a — rm — j3 — t2 se-
quence in which only longitudinal mag-
netization is retained during rm period
can be expressed as,
S(h,t2) =
Tr{{Fx)exp(-iHt2)exp(-if5Fx)
[exp(—iaFx)exp(—iHti)exp(—i'^Fy)
exp(WTm)exp(i(3Fx)exp(iHt2)}
(1)
where the prime indicates retention of
only the diagonal elements of the den-
sity matrix after the a pulse, W is the
matrix governing relaxation during rTO
period and cr0 is the initial density ma-
trix. If cr0 is an equilibrium density ma-

44 Bulletin of Magnetic Resonance
trix, then only single quantum coher-
ences are created during ii period and
since during period t2 only single quan-
tum coherences are detected, the above
equation can be written as(7)
exp(WNxNrm)
(2)
X) represents a matrix which
transforms the N populations into M
single quantum coherences by a pulse
of angle 7X . The N populations are
arranged in descending order of energy
while the M coherences represented by
vectors

Vol. 14, No. 1-4 45
PN*M{I) =
-s 2
-C 2 +
S 2 (l - v 2 )
v 2 S 2
c 2
-C 2
-u' 2 S 2
C 2 +
S 2 (u 2 - 1)
s 2
—S 2
u 2 S 2
—C 2 —
S 2 (u 2 - 1)
c 2
-C 2
c 2 -
S 2 (l - v 2 )
-v' z S 2
s 2
u 0 0 0
0 v 0 0
0 0 v 0
0 0 0 u
(6) '
where S = Sin(7 / 2) ; C = Cosfr / 2); u = Cos 9 + Sin 9 ;v = Cos 9 - Sin 9 and
tan(29) — JAB/($A — &B) defines the strength ofthe coupling.
Frequencies
U)2
(1) Diagonal peaks
1-3
3-4
1-3
3-4
1-2 1-2
2-4 2-4
(2) Auto-peaks
1-3 2-4
3-4 1-2
2-4 1-3
1-2 3-4
(3) Cross-peaks
1-3 3-4
3-4 1-3
2-4
1-2
1-3
2-4
1-2
3-4
1-2
2-4
1-2
3-4
1-3
2-4
V
Table. 1.
u\4GIC}
+ 4S 2 aS 2 {v 2 + (1 - ^ 2 ) 2 } — (1 — C2aC2/3)(l — V 2 )]
u 2 v '[-4^(1-«V) +2(1
-c2ac
U 2 V
Intensities *(—5*2
%CIC} + 4S 2 aS 2 p{u 2 + {v 2
2 M^(i-^) + 2(i
v 4 [-2C 2 aC 2 -2S 2 aS 2 {uU
- m
— {u 2 \o — zu
2 — v)(C2a
-
- (u 2 - 1)
2 } + (1 - C2aC2P)v 2 }
- (1 - v'
A)
— (1 — ^2cc^- / 2/3)(l — u )\
iff) + (1 - V 2 )(C2c - CV)]
w) - (1 - v 2 )(C2a - C2P)}
\2\ i^ /i /"i /"i \ 2*1
/ J ' v ^- / 2a^- / 2/5/^ J
u r — '^\ -*- — ^- J '2f"y^-^2/?/
Cos(i) ; ^i = Sin(i) ; d = Cos(^) ; Si = Sin(^) where i = a, f3 and
u = Cos 9 + Sin 9 ; v = Cos 9 - Sin 9.

46
The origin of these cross-peaks lies in
the creation of a initial state in which
the initial perturbation is distributed
over all the transitions of a strongly cou-
pled spin system as well as due to the
non-linear measurement of the strongly
coupled spin system by the third 90°
pulse. The initial state in this experi-
ment can also be described using mag-
netization modes(12,13). For an AB
system the initial state in terms of the
magnetization modes at various cross-
sections parallel to UJ^ is given in Ta-
ble.2. From these it is seen that the
single spin modes of both the spins are
created in each cross-section. This is the
origin of the cross-peaks in strongly cou-
pled spins. In the limit of weak coupling
(u = v = 1) each cross-section contains
only one single spin mode belonging to
the inverted spin and the cross-peaks
are absent.
ABX spin system
If there are two groups of spins which
are strongly coupled among themselves
but weakly coupled to others then it
is not a priory clear that there will
be cross-peaks between the two groups.
This is investigated here with the help
of an ABX spin system. Fig.2 shows
the standard NOESY (a = j8 = 90° )
spectrum calculated using eqn.[5] for an
Bulletin of Magnetic Resonance
ABX spin system with zero mixing time.
Actual intensities of the peaks in the
2D spectrum is obtained by multiply-
ing the expressions given in Table.3 with
the corresponding ID intensities in both
u>i and ui2 dimensions of that particu-
lar peak. From this it is seen that ev-
ery peak has a cross-peak to every other
peak. The cross-peaks in this spectrum
including those between A and B spins
and between AB spins and X spin arise
due to the strong coupling among the A
and B spins, and disappear under weak
coupling approximation. The appear-
ance of these cross-peaks needs further
investigation in terms of whether they
are due to the non-linearity of the sec-
ond or the third pulse. To investigate
this, calculations have been carried out
for the cases when the excitation pulse is
small(in the linear regime) or the detec-
tion pulse is small(in the linear regime).
The following results were noted from
these experiments:
(i) 90° - a - 90° Experiment
The ABX spin system has eight AB-
transitions and six X-transitions two of
which are between states which are un-
perturbed by strong coupling (the so >
called pure states 1,2,7 and 8)(14). In
this system in the 90° — a — 90° experi-
ment there are no cross-peaks from the

Vol. 14, No. 1-4 47
K.-
At ui
< AA2 >Tm=0
< ABZ >Tm=0
< AAZBZ >Tm=0
u 2
u 2
V 2
V2
-u 2 (l - v 2 )
-u 2 (l+v 2 )
0
Table. 2.
~V 2 (1 + U 2 )
v 2 (l - v 2 )
0
2 u 2
• • • 1i
• • • 4
• • • *
to,
I
•
v 2
-u\l + v 2 )
-u 2 (l - v 2 )
0
V 2 {l-V 2 )
-v 2 (l + u 2 )
0
= -(1+v 4 )
• - (1-v 4 )
- (1-u 4 )
- (1-U 2 V 2 )
Figure 1. Schematic spectrum of an AB spin system calculated for the 90°—90°—90°
2D NOESY experiment with zero mixing time. The symbols represent -PMxiv(90°) x
-fWxM(90°), the |FX| 2 are given along the ID spectra and the final intensities are
obtained using eq [5].

48
B
A
C_
2
V_
U
V
u
V.
AiB3
A1B1
A,
An
A12
37 13
A2X2
A2B3
A2B2
A3X4
^3X3
A3X2
A3X1
A3B3
A3B1
-43
u 2
26 47
A4X2
AtB3
B1X4
fijA'3
BiX2
Bu
B13
B12
B,
u 2
B2X4
4"
- Bt for i = 1 io 4
B2X3
B2X2
B2X1
B24
B23
B2
B3X6
B3X5
B3X4
B3X3
B3X2
BsXj
B34
B3
68
B4X4
B4X3
B4X2
B4X1
B4
36,
A'l4
X13
X12
X!
12
Bulletin of Magnetic Resonance
A'26
X25
x24
x2
46 35
X35
X34
X3
X4
x4
78
45 1
Xe
.1 SI 1 Cl Cl I Si
o>2
* V — V — V
A2 — A5 — A25
D{j A12 — A15
AjBi for i ^ j and j > i A"23 = X35
— B{Xj for i = 1 to 4 (md A24 = A45
A{X2 for j = 1 io 6 A26 = A56
5.A-.2 " AXX, = A4A6
A3JB4 A^X'3 = A3X4
A34 A2A4 = A4A3
A3X1 = A2A6
Figure 2.

Vol. 14, No. 1-4
a3
a4
Peaks
Al
A2
A3
A4
Au
Tl^ —
A13
A14
A23
A24
Xi
x2
x3
x4
X6
XX2
Xi3
X\4
Xm
x23
X24
X26
^34
-^36
X4Q
Strong
Coupling
-(1 + 5/6^)
-(1 + 56?,)
-(1 + *%)
-(1 + s/b 2 )
-(1 +5)
-(1 +564/63)
-(1 + 5/6x62)
-(1 + 56x62)
-(1 + 363/64)
-(a 2 4 + ka 2 3)
-(l + k)
-(al + kaf)
—(al + ka\)
-(a 2 + kal)
sa3 — a4
ka2a4 — axa.3
kaxa4 - a2a3
-a3a4(l + k)
—(ax + 502)
—(a2 + 5ax)
504 — a3
-axa2(l + k)
Table 3. Intensities of the peaks
Weak
Coupling
-2
-2
-2
-2
-2
-2
-2
-2
_2
0
-2
-2
-2
0
0
0
0
0
-2
_2
0
0
0
0
61 = V+/U+
62 = v_/u_
h =
b4
=
— ,,2
U +~ V +
u_v+v_/u+;
u+u_v+/v_;
m2
k
= u
s
-
=
f,,2 ,,2
Peaks
AlBl
A2B2
A3B3
A4B4
AiB2
AiB3
AXB4
A2B3
A2B4
AxXx
AiX2
AiX3
A1X4
AiX6
A2XX
A2X2
A2X3
A2X4
A2X&
A3X2
A3X3
A3X6
A4X1
A4X2
A4X4
Strong
Coupling
s/b 2 -l
562 -1
sb 2 -l
s/bl - 1
5-1
564/63 - 1
563/64 - 1
56162 - 1
5/6i62-l
2 2
a4miif+ — a3m2U_
[n3mx — ra2ui]
a2m2u 2 _ - axmiv\
ax[n3mx — rn2u 2 _]
-a4[n3mx — m2u 2 _]
—a3[nxmx — m2f 2 _]
—[nxmx — m2v 2 _]
r 2 1
—a2[ni?Tix ~ Tn2v_\
axm2v 2 _ - a2mxu%
a3mxv\ — a4m2v 2 __
—[n2m2 — mx^]
a2JVl2 m 2 ~~ ^•1^4.]
— a4\jl2V(l2 — T^lU-iJ
—a3[?i47TT.2 — mxu , ]
-[n4m2 - mxu\]
ax\n4m2 — mxU^.1
u%v\ ulv 2 _)/2
J- •= Cos(6+-9_); u+ - Cos(0+) + Sin(6+); U_
>- = Sin(8+-6_); v+ = Cos(0+) - Sin(0+); v_
Weak
Coupling
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
= Cos{BJ) + Sin{0J)
= Cos{dJ) - Sin(9_)
49

50
X transitions between pure states to
all AB transitions, while there are cross-
peaks between the other X transitions
to all AB transitions and also cross-
peaks between all AB transitions to all
X transitions. The selective inversion of
X^ or X^ by a small angle a pulse
does not cause any perturbation of the
strongly coupled states and hence there
are no cross-peaks from these transi-
tions to all AB transitions. The 90°
third pulse mixes the X-magnetisation
unequally between all the X transitions
giving rise to the auto-peaks. On the
other hand, the selective inversion of
an AB transition perturbs the strongly
coupled states leading to cross-peaks
to X transitions between mixed states.
The non-linear detection pulse in turn
mixes the intensities of all the X transi-
tions giving rise to cross-peaks to even
the X transitions between pure states.
The spectrum is not symmetrical 15).
(ii) 90° - 90° - a Experiment
In this experiment there are no cross-
peaks between all AB transitions to X
transitions between pure states, while
there are cross-peaks to all X transitions
between the mixed states and also cross-
peaks between all X transitions to all
AB transitions. This is due to the fact
Bulletin of Magnetic Resonance
that the second 90° pulse perturbs un-
equally all the transitions of AB as well
as the X spin. The mixed states of the
AB spins do not give directly any cross-
peak to X-pure transitions. Since the
detection pulse is a small angle pulse
it does not mix the X transitions be-
tween pure and mixed states and there-
fore there are no cross-peaks from AB
to X-pure transitions. The appearance
of cross-peaks between X transitions be-
tween pure states and the AB transi-
tions is due to the mixing produced be-
tween the various transitions of the X
spin by the second 90° pulse. This state
of the system is faithfully measured by
the detection pulse. Here also the spec-
trum is not symmetrical 15).
The results of 90° - a - 90° and
90° — 90° — a experiments are trans-
pose of each other. This is due to the
fact that PJVXMIOJX) = -PMXJVC 0 ^)-
The conversion of populations into co-
herences and vice versa are described by
mirror operations(15).
Experimental
Experimental observation of these •
cross-peaks was carried out in acetone
oriented in liquid crystal ZLI 1167.
Acetone oriented in liquid crystal is a
strongly coupled spin system of the type
(A3A3) with C3u C3v symmetry. The

Vol. 14, No. 1-4 51
spectra is shown in Fig.3. From this
spectrum it is clear that there are cross-
peaks from every peak to all others
within the same irreducible representa-
tion. Theoretical simulations of these
cross-peaks show a very good match
with the experimental results(6), con-
firming the existence of strong coupling
induced cross-peaks in the 2D NOE ex-
periments even in the absence of relax-
ation.
B. CROSS-CORRELATIONS IN
2D NOE
If a spin has more than one pathway
for relaxation, then there can be cross-
terms between these pathways that may
contribute to the relaxation of the spin.
For example, if there is another spin
nearby, and the mutual dipolar interac-
tion contributes to the relaxation of the
spin and if in addition the first spin has
a partial relaxation by CSA, there can
be cross-terms between the dipolar re-
laxation and CSA(16). If on the other
hand there is a third spin contributing
to the relaxation of the first two through
dipolar relaxation then there can be
cross-terms between various dipolar in-
teractions and between the dipolar and
CSA interactions contributing to the re-
laxation of the various spins. These
cross-terms known as cross-correlations
are often neglected in the relaxation
analysis such as those using generalized
Solomons equations(17). It turns out
that while the cross-terms may be sig-
nificant in magnitude their manifesta-
tion in a particular experiment may be
small. For example the dominant ef-
fect of the cross-terms is to make the re-
laxation of various transitions of a spin
unequal. In a given spin system or in
an experiment if these transitions are
not resolved then this dominant effect
of cross-terms is absent. This can hap-
pen for example when the spins are not
J-coupled or if one uses a 90° pulse for
measuring the intensities of the multi-
plet. In the later case the non-linearity
of the pulse yields an average intensity
over all the transitions of a spin oblit-
erating the multiplet effect and largely
the cross-correlation effects. The use
of a small flip angle for the measuring
pulse is a necessary requirement for the
observation of the multiplet effect and
in turn the cross-correlation effects in
the ID and the standard NOESY ex-
periments^).
In two-dimensional NOE experiment
the most significant attempts to ob-
serve the effect of cross-correlations
have been made by Bodenhausen and
his group(7,16,18-20). One of the ex-

52
periments they have used is a small
flip angle NOESY experiment namely
90°—1\—a—rm—a—£2 , where a is small.
Each cross-section of the small flip angle
NOESY (NOESY 90° - a - a) is then
equivalent to a ID difference transient
NOE experiment in which the peak cor-
responding to the diagonal peak is se-
lectively inverted. This experiment has
both the direct pumping effects and the
cross-correlation induced multiplet ef-
fects present which are measured by the
small angle third pulse. For example the
intensities of the X diagonal and the AX
cross-peak multiplet in a weakly coupled
three spin (AMX) system, in the initial
rate approximation are given by(7),
Xi
x2
x3
x4
Ax
A2
A3
A4
dn
l[A
KM
_ hAM
r rw
r3 0)
pf
X2 X3 X4
1\A
'lM
r2 x)
(o)
P3
r4 0)
«A
r(0)
P2 0)
r3 1}
P ( 4 1}
^2AM
L
dTT
_
pf 1
r(o)
r 2
P3 X)
r 4
(7)
where X1; X2, X3, X4 are the four
X transitions and Ax, A2, A3, A4 are
the four A transitions. The expressions
for the various intensities of the peaks
are given in ref (7) except that when
the cross-correlations due to CSA and
Bulletin of Magnetic Resonance
dipole-dipole interaction are included
W11 ^ wu ? WU £ w^ and l\i ^ l[i
where i = A, M or X. The r and p
terms signify regressive and progressive
peaks respectively. From eqn[7] it is
seen that while the cross-correlation in-
formation is contained in the small flip
angle NOESY experiment, it is coupled
with the direct pumping effects.
We propose here simple modifications
to the small flip angle NOESY (NOESY
90° — a — a). If the second or the third
pulse is made 90° then the intensities
in the initial rate approximation are ob-
tained as averages of the multiplets in
either u>x or o>2 direction respectively.
This removes the direct pumping effects
from the 2D spectra. The following re-
sults are obtained.
90° - a - 90° NOESY
The intensities of the various peaks in
the initial rate approximation are given
by
x2
x3
x4
Ax
A2
A3
A4
Xx X2 X3
Rz
2 ^2
C3 Cz
Ri
R2
Rz R2
Rz
Cz
x4
R2
Rz
Ci
C2
(8)

Vol. 14, No. 1-4
where
i2i =
R2
=
Rs =
R4 —
Co =
c4 =
l0AM
l\M
XM +
rx ^AX-
Sx)]rm
+ hAM + l\M +
- (Px -
A A
(!) , (1) 1 (0) , (0)
r x + Px + r x + Pi
MX
(9)
(10)
Here p^ is the rate of self relaxation of
spin X, a AX 1S the cross-relaxation rate
between spins A and X, 6X = SAXMXI
which gives the cross-correlation rate
between the dipolar vectors AX and
MX and A^ gives the cross-correlation
raten between the dipolar vector AX
and the CSA of spin X. The expres-
sions for the spectral density functions
for the various relaxation rates (p, cr,
A, 6) are given in ref(ll). The inten-
sities of the various peaks in each mul-
tiplet are identical in u>2 dimension and
differ in o?j dimension, the differences
directly yielding the cross-correlations.
If the multiplet is resolved in the wi
dimension the difference in the intensi-
ties of the inner or the outer lines gives
the dipole-CSA cross-correlations(A s)
and the difference between the inner and
outer lines gives the dipole-dipole cross-
correlations (6 s). The diagonal multi-
plet result is identical to the differences
in the initial rates of recovery of the
outer and inner multiplets in inversion-
recovery Ti measurements (21). How-
ever many analyses of inversion recov-
ery measurements including (21) ignore
CSA-dipole cross-correlations, while re-
taining dipole-dipole cross-correlations.
90° - 90° - a NOESY
The intensities of the diagonal and
the cross-peak multiplets in the initial
rate approximation in this case are given
by
x2
X,
At
where
c2 =
c3 =
c: =
X4
R\ R2 Rs R4
R\ R2 Rs R4
R\ R2 R3 R4
r> r> r? r>
ti\ Ii2 Kz it.4
' ri' ri' p' ri'
1 2 o 4
ri' ri' rt' ri 1
^1 ^2 U 3 W
ri' r" r>' r>'
O-i uo Lyo Ly A
X Jr O 4
ri' r" si' ri'
Oi Wo L/o Ly^
2{aAX + A; lx + 6A)Tm
2(VAX - A; AX ~~ "A) T m
2{

54
a b cdefg h
* •
4 %
* »
500
i j k
#
• *
: • ' .
I I
pmm'l
\> til,
0
CO2
1e' d c b o
,, d c b
Ml./
* f • »•- * •
• « #
•t: : :::.
* •
0 9 • • *.
• • •
• • • • • •
Bulletin of Magnetic Resonance
Figure 3. 2D NOESY spectrum of oriented acetone recorded at 400 MHz with
rm = 20 /isec. The cross-peaks are mainly due to strong coupling. Zero- quantum
interference during rm was shifted out in another experiment and the residual strong
coupling peaks showed satisfactory correlation with the calculated intensities(6).
(a)
(b)
-500
_Jj\A
Figure 4. Cross-sections taken from (a) 90° - 90° - 15° (b) 90° - 90° - 90° 2D
NOESY experiment with rm = r0 + k using a 400 MHz spectrometer. r0 was 400
msec and k was randomly varied between 10 and 1000 msec.
-500
500

Vol. 14, No. 1-4
From these expressions it is seen that
the intensities differ in u>2 and an aver-
aging takes place along u>i . The sec-
ond 90° pulse excites the multiplet as
a whole, the correct state being moni-
tored by the small angle a pulse. The
differences in the intensities again yield
the cross-correlations except that in this
case the AX multiplet yields 8% and
A^. The diagonal multiplet has in-
tensities identical to 90° — a — 90° ex-
periment. Since in the 90° — 90° — a
experiment the intensities differ in UJ2
domain which is easier to resolve, this
experiment may be preferred over the
90° - a - 90° experiment. In addi-
tion, since all the lines of a multiplet
along uj\ have equal intensities, a u)\ -
decoupled (90° - (A + h)/2 - 180° -
(A - *i)/2 - 90° -rm-a-t2) NOESY
experiment can replace the undecoupled
(90° - tx - 90° - rm - a - t2) NOESY
experiment without loss of information.
Experimental
Two-dimensional NOESY experiment
was carried out in 2,3-dibromo propi-
onic acid using the 90° - 90° - a se-
quence with small a(~ 15°) and a mix-
ing time of 400msec plus a random vari-
ation from 10 to 1000msec. Some of
the cross-sections are shown in Fig.4.
( The differences between the intensities
of various transitions of a multiplet in-
dicate the presence of cross-correlations.
Lack of any particular symmetry in
these multiplets indicates the presence
of both dipole-dipole and CSA-dipole
cross-correlations.
CONCLUSIONS
The use of 90° angle for the ex-
citation or detection pulses allows an
easier method for studying the cross-
correlations in 2D NOE experiment.
One can use either the second or the
third pulse as small angle pulse to high-
light the cross-correlation effects. In
strongly coupled spins the non-linearity
of the pulses can give rise to cross-peaks
even in the absence of relaxation. The
origin of these cross-peaks arising due
to the non-linearity of the second or the
third pulse are discussed with the help
of an ABX spin system.
Acknowledgement
AK wishes to acknowledge discussion
with Ms. Irene Burghardt of University
of Lausanne regarding Fig.l.
References
*R. R. Ernst, G. Bodenhausen and
A. Wokaun,'Principles of Nuclear Ma g-
netic Resonance in One and Two Di-
mensions', Clarendon, Oxford, 1987.
55

56
2 J. Keeler, D. Neuhaus and M. P.
Williamson, J. Magn. Reson. 73, 45-
68 (1987).
3 I. Solomon, Phys. Rev. 99, 559-565
(1955).
4 I. Solomon and N. Bloembergen, J.
Chem. Phys. 25, 261-266 (1956).
5 S. Macura and R. R. Ernst, Mol.
Phys. 41, 95-117 (1980).
6 R. C. R. Grace and Anil Kumar, J.
Magn. Reson. 97, 184-191 (1992).
7 H. Oschkinat, D. Limat, L. Emsley
and G. Bodenhausen, J. Magn. Reson.
81, 13-42 (1989).
8 J. Keeler and F. S. Fernando, J.
Magn. Reson. 75, 96-109 (1987).
9 T. E. Bull, J. Magn. Reson. 72,
397-413 (1987).
10 V. V. Krishnan and Anil Kumar, J.
Magn. Reson. 92, 293-311 (1991).
11 C. Dalvit and G. Bodenhausen,
Adv. Magn. Reson. 14, 1-32 (1990).
12 L. G. Werbelow and D. M. Grant,
Adv. Magn. Reson. 9, 189-299 (1977).
13 D. Canet, Prog. NMR. Spectrosc.
21, 237-291 (1989).
14 J. A. Pople, W. G. Schneider
and H. J. Bernstein, 'High Resolu-
tion Nuclear Magnetic Resonance Spec-
troscopy', McGraw-Hill, New York,
1959.
15 R. C. R. Grace and Anil Kumar,
(unpublished results)
Bulletin of Magnetic Resonance
16 I. Burghardt, R. Konrat and G. Bo-
denhausen, Mol. Phys. 75, 467-486
(1992).
17 J. H. Noggle and R. E. Schirmer,
'The Nuclear Overhauser Ef-
fect : Chemical applications', Academic
Press, New York, 1971.
18 H. Oschkinat, A. Pas-tore and G.
Bodenhausen, J. Am. Chem. Soc. 109,
4110-4111 (1987).
19 C. Dalvit and G. Bodenhausen, J.
Am. Chem. Soc. 110, 7924-7926
(1988).
20 J. M. Bohlen, S. Wimperis and G.
Bodenhausen, J. Magn. Reson. 77,
599-605 (1988).
21 E. Ilyina and Daragan (in press).

Vol. 14, No. 1-4 57
Deriving Structures from 2D NMR. A Method for Defining the
Conformation of a Protein Adsorbed to Surfaces
With the development of multi-dimensional
NMR methods for the specific assignment of many of
the *H NMR signals of small proteins, it is now possible
to determine three-dimensional solution structures
and dynamics for these proteins. Unfortunately
these methods fail to provide detailed structural
information on proteins bound to macroscopic
surfaces because of the very slow overall rotational
correlation time of the particle-bound protein. Indeed
other spectroscopic and structural methods
have provided few details of the structure of proteins
when adsorbed to surfaces (1). Commercially,
surface-bound, immobilized enzymes provide an important
method for efficient utilization of these catalysts
in bioreactors. The development of biosensors
and novel biomaterials such as self-assembled monolayers
(2) requires a better understanding of the
structure of proteins bound to surfaces (1). Identification
of specific residues of proteins involved in
interactions with stationary phase surfaces is critical
in understanding their chromatographic behavior
(3). Finally, the association of proteins with
hydrophobic or glass surfaces has been shown to
cause denaturation or partial unfolding of the pro-
. tein at the surface, with often irreversible loss of
the strongly adsorbed protein. It is believed that
;.these irreversibly bound proteins undergo a confor-
|mational change to expose a portion of the interior
sidues to the surface for effective adsorption.
J this report we describe an NMR methodology
[it for the first time allows us to probe the detailed
^formation of a protein bound to a macroscopic
This was accomplished by 2D amide hy-
«i exchange NMR spectroscopy (4, 5, 6). In a
?i, the NH exchange rates for different residues
over a factor of greater than 10 8 (4, 5).
fate of amide hydrogen/deuterium exchange
"I on the pH (being both acid and base cat-
! the accessibility of the hydrogen to solvent,
Mlization by hydrogen-bonded secondary as
tertiary structure and finally, local fluctu-
•jf the protein (7). Importantly amide hy-
David A. Keire and David G. Gorenstein*
Department of Chemistry
Purdue University
West Lafayette, Indiana 47907
drogen exchange can be studied under conditions
where high resolution resolvable proton NMR signals
would otherwise not be observed - this would
be true for a protein immobilized at a polymeric or
glass surface. However, analysis of the degree of NH
exchange is feasible if the protein can be desorbed
from the surface and the 2D NMR study of the free
protein carried out in solution.
As demonstration of the feasibility of the
method, we have used amide hydrogen exchange
NMR spectroscopy of lysozyme bound to a hydrophobic
chromatographic stationary phase support.
Hen egg white lysozyme (E.C.3.2.1.17) was
chosen for this initial study because the ^-NMR
assignments have been made (8), the crystal structure
was known to 2 A resolution (9) and much was
known about its adsorption onto hydrophobic substrates
(10). Lysozyme is one of the major constituents
of protein deposits on contact lenses (11)
and is commonly used as a standard in the chromatographic
separation of proteins on reverse-phase
columns.
In our protocol we first adsorb lysozyme to the
hydrophobic surface (solid 5 \i diameter polystyrene
divinylbenzene chromatographic stationary phase
support; Polymer Labs Inc., Amherst, Mass.). This
forms a tightly bound single monolayer on the
surface (10). We then expose the surface bound
lysozyme to D2O under fast amide hydrogen exchange
conditions (high pH), desorb the protein under
slow NH exchange conditions (low pH) using
a detergent, and after removal of the detergent, run
the high resolution 2D spectra in D2O solution. The
intensities of the 2D NHCHa TOCSY crosspeaks
then reflect the degree of exchange of the specific
residue amide hydrogens when adsorbed to the hydrophobic
surface.
The stock solution of lysozyme (Sigma) was purified
using a home built hollow fiber bundle dialysis
device versus 10 mM NaH2PC>4 pH = 7.4 buffer. An
aliquot of the stock solution containing 125 mg of

58
lysozyme was diluted to 10 mL in buffer and mixed
with 10 g of hydrated stationary phase support for
30 min. with stirring. The solution was then removed
by the use of a filter funnel with a medium
grain glass frit.
Eighty milliliters of buffer were used to wash the
stationary phase support in 10 to 20 mL aliquots.
For each aliquot a sample of the filtrate was taken
for UV determination of protein concentration (e280
= 2.313 O.D. mL/mg) to ensure that only an "irreversibly"
bound monolayer coverage remained.
Typically, out of 125 mg of lysozyme initially added
~25 mg would remain on the support after extensive
washing. The stationary phase support has a
surface area of ~3 A/gm and a rough calculation
using the molecular dimensions of lysozyme (4.5 x 3
x 3 nra 3 ) shows that ~50 mg of protein could adsorb
to 10 gm of support with monolayer coverage. The
25 mg of lysozyme adsorbed represents 50% coverage
which is in good agreement with the 65% coverage
by the irreversibly bound layer determined by
Schmidt et al. (10) to a similar hydrophobic surface.
At this point a 10 mM NaH^PCU D2O solution
pH* = 7.0 (uncorrected pH meter reading) was prepared
and mixed with the lysozyme covered support.
The adsorbed protein was then stirred for 45
min. during which time the amide hydrogens that
are exposed to solvent can exchange with deuterons.
After 45 min. the solution was aspirated away
and another 10 mL of the buffered D2O solution at
pH*=2.5 was added to quench the amide exchange.
Two aliquots of 0.1% Triton X-114-RS (Sigma)
detergent in 10 mL D2O, pH* = 2.5 were used to
remove the tightly adsorbed protein. The reduced
form of Triton was used to allow determination of
desorbed protein concentration by UV. The ~30
mL of desorbed lysozyme/Triton solution was then
concentrated by pressure dialysis (Amicon, Beverly,
Mass.) to ~4 mL using a 5000 molecular weight cutoff
membrane. The 4 mL was then passed through
an Extracti-Gel column (Pierce) equilibrated with
pH* = 2.5 D2O buffer to remove the Triton. The
column eluent was further concentrated to ~750 fib
and placed in an NMR tube. Approximately 5 mg
(0.5 mM) of desorbed lysozyme was obtained by this
procedure for the NMR study. Much of the protein
loss occurred during the repurification scheme since
the adsorbed protein was nearly quantitatively removed
from the surface by the detergent treatment.
Bulletin of Magnetic Resonance
In the control experiment the identical procedure
was used except that only 5 mg of lysozyme was
used and no stationary phase support was present.
An additional control was run without stationary
phase support and without Triton to ensure that
any residual Triton not detectable in the ID NMR
spectrum did not interfer with the amide exchange
results. This control gave essentially the same integrated
area for the TOCSY cross peaks of the NH-
CaH region as the control involving Triton.
The 600 MHz X H-NMR NH-CaH region of the
30 ms TOCSY spectra of the control and surface
adsorbed/desorbed lysozyme is shown in Figure 1.
Assignment of the signals was based upon complete
analysis of the TOCSY spectra, taking advantage
of the reported 1 H assignments of lysozyme at 500
MHz (8). Normally, a series of TOCSY spectra at
various time points are taken to measure the rate
of NH exchange. Since this has been shown to be
first order, comparing the spectra at a fixed time
of exchange under conditions where the protein is
either surface-bound or not will also provide a measure
of the relative degree of NH protection. We
define protons as being "exposed" when their crosspeaks
present in the control spectrum are decreased
in intensity by >40% relative to the surface exposed
protein spectrum. Protons are "protected" if the
crosspeaks are either absent in the control spectra or
increase in intensity by >40% in the surface exposed
sample spectra. Three control experiments (without
the support present) were used to calculate the percent
deviation from the mean of the integrated areas
of 39 of the slow exchanging amide proton NH-CaH
TOCSY cross peaks. The mean of the 39 deviations
from the means of the integrated areas was 20±13%.
Thus, only those changes in integrated area >40%;
were considered exposed or protected after averag|
ing the cross peak areas from two separate adsorbaj
protein and control experiments. These results ar|
tabulated for the slow exchanging amides [as definw
by Redfield & Dobson (8) at pH = 3.8, 35°C, n/2.
the half life for exchange > 1-5 h] in Table I. Also
included in Table I are the integrated intensities oj
2 intermediate exchange (1 h < r1/2 < 5 h) and on«
fast exchange amide hydrogens (T1/2 < 1-5 h) whicg
are protected relative to the control.
All of the amides which are protected (Table.*
with the exception of the 2 arginines are neutr
or hydrophobic residues. A CPK model {Figuiej

Vol. 14, No. 1-4 59
B
Fl (ppm)
5.5 3.5 3.0 2.5
Fl (ppm)
1: Identical 600 MHz ^-NMR TOCSY experiments at 35°C and pH*=2.5 were run on both the
>1 and adsorbed lysozyme samples. The data was collected in phase sensitive mode with a mixing time
using a MLEV17 spin lock and 2 ms trim pulses. 2K points were collected in F2 and 312 in Fl
^sweep width of 7500 Hz and a repetition time of 3 s. 32 transients were coadded at each ti increment.
£JD signal was suppressed through low power decoupling at the HOD frequency. The spectra were
with zero filling to 2K points in the Fl dimension and a 45° phase shifted sine bell apodization in
tensions.

60 Bulletin of Magnetic Resonance
based upon the crystal structure of lysozyme (9)]
shows that all of the protected amide residues (labeled
by shading) are found on the surface. Four
of the protected amide residues (T40, Q41, A82
and L84) are located close in space at the hinge region
between the a-helical and /3-sheet domains of
lysozyme opposite the active site cleft.
All of the exposed amides - those that show enhanced
amide exchange upon binding to the surface
- with the exception of W63 are involved in ahelices
or /3-sheet structures and are mostly buried
in the interior of lysozyme (Table I). Of the exposed
amides only D52 and W63 show significantly exposed
side chains as shown in the CPK model (Figure
2; exposed residues shown by stripes).
The protein amide hydrogens that are observable
by 2D amide exchange spectroscopy are mostly the
slow exchanging amides (58/129) which are either
involved in secondary or tertiary structure hydrogen
bonding or buried in the interior of the protein
and inaccessible to solvent (12). Surprisingly, one
fast exchanging amide hydrogen (III 14), and two intermediately
exchanging hydrogens (T40 and Q41)
CaH-NH cross peaks were observable in the spectrum
of lysozyme exposed to the hydrophobic surface
which were not present in the control spectrum
(Figure 1).
Figure 2A shows that the protected (stippled)
amide hydrogen side chains lie on one face of the
globular protein in a narrow ridge from R125 to
L17. In contrast, in the back-side view (Figure 2B),
only a few of the largely buried side chains of the
exposed residues (striped) are visible. None of the
protected residues are visible in this view. These results
suggest that lysozyme is not randomly oriented
with respect to the surface, but that it is oriented
with a relatively hydrophobic ridge facing towards
the hydrophobic surface. This is consistent with the
observed retardation of amide hydrogen exchange in
binding an amphiphilic helix to a micelle (13), believed
to result from burial of the hydrophobic face
of an amphiphilic helix into the hydrophobic interior
of a detergent micelle.
In addition a number of residues exchange more
rapidly when the protein is surface bound than when
it is in solution. All of these are not exposed to
the solvent in the native structure but are located
on elements of structure that are likely involved in
segmental motion of the two domains that form the
active site cleft (14). Binding of the hydrophobic
ridge of the protein to the surface thus appears to
induce a conformational change, exposing the active
site residues and residues adjacent to the active site
to solvent.
These perturbations in the amide exchange rates
do not simply reflect proximity of the protein to
the surface because enhancement and protection towards
exchange are observed. In addition most
of the amide hydrogen exchange rates which are
not exposed or protected are the same for surface
bound or free lysozyme. In contrast only decreases
in amide exchange rates are observed in binding an
amphiphilic helix to a micelle (13) and in crystalline
lysozyme (15).
Taken together these data indicate that
lysozyme adsorbs to the surface on the side opposite
the active site cleft. This protects this narrow ridge
of amides from exchange by either blocking solvent
access to these residues at the hydrophobic surface
or by reducing the rate of local fluctuations in the
surface oriented residues. In addition upon binding
to the hydrophobic surface a significant disruption
of several of the buried a-helices and the active site
cleft occurs as the protein partially unfolds at the
surface presumably by opening of the "hinge" at
the active site (16), exposing a number of buried
residues that now show enhanced rate of hydrogen
amide exchange. This conformational change alters
the amide solvent exposure and/or local fluctuations
that allow access of solvent to these interior residues.
This partial unfolding exposes the catalytically
important D52 residue and other important residues
in the active site cleft. This model is supported by
the fact that the enzyme is inactive on adsorption
to the hydrophobic surface of alkylated silica (10).
These conclusions are also in agreement with the
results of Fausnaugh and Regnier (3) based upon
analysis of chromatographic behavior of various bird
lysozymes that also suggests the protein adsorbs on
a side opposite the active site with a contact surface j
that extends from residues 41 to 102 and 75 to 89.1
Computer modeling has also revealed a relatively^
hydrophobic patch identified as a possible binding
site in this region and total internal reflection i j
sic fluorescence (TIRIF) shows a decreased quantun^
yield (and hence altered conformation) of the
sorbed hen lysozyme (17).
Importantly, the surface desorbed lysozj

Vol. 14, No. 1-4
A
B
wire 2: Three dimensional structure of hen lysozyme. A) Front-side view of a CPK model [MIDAS modeling
", UCSF (21)] showing amide NH's which are relatively protected (shaded and stippled) or exposed
and striped) when the protein is bound to the polystyrene surface. The active site cleft is oriented
ne upper left. B) No residues that are protected from exchange (shaded and stippled) can be seen
^.ackside view of the protein. Several of the side chains of residues that are more exposed (shaded and
,

62
Lysozyme CH-NH TOCSY Cross Peaks
A9
All
L17
W28
C30
A31
N37
T40
Q41
D52
W63
A82
L84
C94
198
R114
W123
R125
1.19
4.26
4.53
1.06
0.68
0.27
4.71
4.36
0.99
1.15
2.55
2.40
2.73
0.59
0.32
1.26
0.71
1.30
2.84
2.47
1.03
1.49
1.28
0.24
0.87
1.91
1.45
Slow Exposed
Slow Exposed
Bulletin of Magnetic Resonance
Slow Protected
Slow Exposed
Slow Exposed
Slow Exposed
Slow Protected
Inter. Protected
Inter. Protected
Slow Exposed
Slow Exposed
Slow Protected
Slow Protected
Slow Protected
Slow Exposed
Fast Protected
Slow Protected
Slow Protected
a.) Blanks indicate no observable cross peak.
b.) The integrated area of the adsorbed protein CH-NH TOCSY cross peaks were normalized via the integraM
area of the non-exchangeable Trp 108 H4-H5 cross peak in the control and adsorbed protein spectra. Wt
integrated areas in the table are the average of two control and two adsorbed protein experiments. Jjj
c.) Slow, intermediate and fast exchange amide protons are classified as per Redfield and Dobson (4). |ip
d.) Exposed amides are present in the control and diminished by >40% in integrated area or absent inj|
adsorbed protein spectra. Protected amides are present in the adsorbed protein spectra and diminish^
integrated area by >40% or absent in the control spectra. |i

Vol. 14, No. 1-4 63
spectra have identical chemical shifts as the native
form spectra. This suggests that while the protein
partially unfolds at the surface, it refolds by
the same kinetically or thermodynamically favorable
pathway upon desorption. Thus, the surface
unfolded state may represent a protein folding intermediate
of native hen lysozyme. Only a transient
folding intermediate has been previously identified
in lysozyme (12). By binding the protein to a surface
it may be possible to trap a partially unfolded
state of lysozyme that bears some resemblance to
the transient folding intermediate observed by Miranker
et al. (12). This would be complementary to
the rapid quench NMR amide hydrogen exchange
spectroscopy methods (12, 18, 19, 20) which provides
structural details on the initial refolding kinetic
intermediates. Our method would presumably
provide information on the initial unfolding intermediate.
Surface 2D amide exchange spectroscopy offers
a new method by which protein adsorption can be
monitored at the level of individual residues. This
work demonstrates for the first time the feasibility
of the method which should be applicable to a number
of other enzymes. The method offers a more detailed
picture of protein adsorption than provided by
currently used techniques (e.g. TIRF, ATR-FTIR
and Raman spectroscopy (1)). Future work will
include the extension of the methodology to other
surfaces and proteins. Indeed at the XVth International
Conference on Magnetic Resonance in Bi-
. ological Systems (Jerusalem, Israel, August 16-21,
,., 1992, abstracts) K. Kawano et al., described simii
-lar application of the NH exchange experiment to
\j the binding of lysozyme to hydroxyapatite. They
'also find that certain residues are protected from
nange when the protein is adsorbed to the sur-
However unlike our results no enhancement of
•exchange is observed.
!' Acknowledgments
rSupported by the Office of Naval Research
jp00l4-91-J-1686) and the Purdue University Bio-
4cal Magnetic Resonance Laboratory which is
ported by the NSF Biological Facilities Center on
Secular NMR, Structure and Design at Pur-
[grants BBS 8614177 and DIR-9000360 from
jjyision of Biological Instrumentation) and NIH
•ferences
1. J. D. Andrade and V. Hlady, Adv. Polym.
Sci. 79, 1-63 (1986).
2. K. L. Prime and G. M. Whitesides, Science
252, 1164-1166 (1991).
3. J. L. Fausnaugh and F. E. Regnier, J. Chromatogr.
359, 131- 146 (1986).
4. H. Roder, "Methods in Enzymology" (N. J.
Oppenheimer and T. L. James, eds. ), 446-473, Academic
Press, Inc., 1989.
5. S. W. Englander and N. R. Kallenbach, Q.
Rev. Biophys. 19, 521-655 (1984).
6. H. Roder, G. Wagner, and Wuthrich, Biochemistry
24, 7396-7407 (1985).
7. Y. Paterson, S. W. Englander, and H. Roder,
Science 249, 755-759 (1990).
8. C. Redfield and C. M. Dobson, Biochemistry
21, 122-136 (1988).
9. C. C. F. Blake, G. A. Koenig, A. C. Mair, C.
T. North, D. C. Philips, and V. R. Sarma, Nature
206, 757-761 (1965).
10. C. F. Schmidt, R. M. Zimmermann, and H.
E. Gaub, Biophys. J. 57, 577-588 (1990).
11. E. J. Castillo, J. L. Koenig, and J. M. Anderson,
Biomaterials 6, 338-344 (1985).
12. A. Miranker, S. E. Radford, M. Karplus, and
C. M. Dobson, Nature 349, 633-636 (1991).
13. C. Karslake, M. E. Piotto, Y. M. Pak, H.
Weiner, and D. G. Gorenstein, Biochemistry 29,
9872-9878 (1990).
14. C. C. F. Blake, G. A. Mair, A. C. T. North,
D. C. Phillips, and V. R. Sarma, Proc. R. Soc.
Lond. Ser. B 167, 365-377 (1967).
15. T. G. Pedersen, B. W. Sigurskjold, K. V.
Andersen, M. Kjaer, F. M. Poulsen, C. M. Dobson,
and C. Redfield, J. Mol. Biol. 218, -413-426
(1991).
16. J. A. McCammon, B. R. Gelin, M. Karplus,
and P. G. Wolynes, Nature 262, 325-326 (1976).
17. D. Horsley, J. Herron, V. Hlady, and J. D.
Andrade, "Proteins at Interfaces", 290-305, 1987.
18. H. Roder, G. A. Elove, and S. W. Englander,
Nature 335, 700 (1988).
19. J. B. Udgaonkar and R. L. Baldwin, Nature
335, 694-699 (1988).
20. C. M. Dobson and P. A. Evans, Nature 335,
666 (1988).
21. T. E. Ferrin, C. C. Huang, L. C. Jarvis, and
R. Langridge, J. Mol. Graphics 6, 13-27 (1988).

64
Flavoridin, a protein with 70
amino acids from the venom of
Trimeresurus gramineus, is a very
potent inhibitor of blood platelet
aggregation. The protein contains
a local Arg-Gly-Asp (RGD)
sequence at position 49 to 51. This
primary sequence element is
known to inhibit fibrinogen
binding by a specific interaction
with the integrin-type platelet
receptor GPIIb/IIIa. By now, a
rather large family of homologous
RGD containing snake toxins have
been sequenced. Most of these
proteins contain n=12 cysteines
which are all linked by disulfide
bridges. Previous biochemical
studies, however, have so far not
revealed the native pattern of the
individual cysteine pairings.
^ l , Werner Klaus and Pau, Gerber
The ! H NMR spectrum of
Flavoridin was almost fully
assigned in aqueous solution by
conventional 2D ^H NMR methods
(2QF-COSY, clean TOCSY, 2Qspectroscopy,
NOESY). The 3Dstructure
calculation with
distance geometry methods
(DIANA) proceeded in several
rounds:
(1) The global fold of Flavoridin
was calculated from the collected
set of NOE distance constraints
(#666) and dihedral angle
constraints obtained from vicinal
coupling constants (#88) but
without the use of any disulfide
bridge constraints.
(2) The interatomic cP-C-P distances
between all possible pairs [n(nl)/2]
of cysteines were measured
in a set of 20 converged distance
geometry structures. A
probability weight wjj (0 - wij - 1)
i« assigned to each individual
Bulletin of Magnetic Resonance
cysteine pair according to a
gaussian shaped distribution
function. When the average
crystallographic CP-CP distance of
a cystine disulfide bridge is
matched, the weight wy = 1.
(3) All combinatorial patterns of 6
disulfide-bridges involving the 12
cysteines in Flavoridin were
calculated and the 6 individual
weights, w^j, for each pattern
were multiplied. Only patterns
with individual wjj — 0.3 were
considered. On this objective basis,
a single Cys-Cys pairing pattern
could unambiguously be
determined. The method was
validated with known protein
crystal structures of Cys-rich
proteins. Also the experimentally
observed NOE's between Ci«H/CjPH
and/or CiP/CjP of Cys residues,
which commonly are taken as
direct evidence for Cys-Cys
disulfide links, do agree' with the
computationally evaluated Cys-Cys
pattern.
(4) In a second step of the
structure calculation, the Cys-Cys
pairing was used as additional
input for the distance geometry
program. Finally, the 50 best
structures were refined by
energy minimization.
Our structural results show that
the polypeptide backbone is folded
in two domain-like structures -
composed of 8 turns and stabilised
by 6 cystine cross-links. The
conformation of the Arg-Gly-Asp
(RGD) sequence is located in an
extended loop structure exposed at
the tip of a so called hairpin,
which is rather flexible.

Vol. 14, No. 1-4 65
1. Introduction
NMR Approaches to Large Proteins:
trp Repressor and Chloramphenicol Acetyltransferase
L.-Y. Lian, J. P. Derrick, V. Ramesh, R. O. Frederick,
The current upper limit for protein structure
determination by *H nmr [1] is in the region
of Mr 12-15,000. The use of stable isotope
labelling, with 2 H, 13 C or 15 N, can
substantially extend this limit, perhaps to Mr
30,000 (for methodological reviews, see [2]-
[4]). However, there remain many interesting
and important proteins which are much larger
than this, and we have been trying to
establish what information nmr can provide in
such cases. Two systems we have been
studying in this context are the E. coli trp
repressor and the enzyme chloramphenicol
acetyltransferase.
The trp repressor is a dimer of total Mr
•; 25,000, and is a member of the "helix-turnpielix"
family of DNA binding proteins. A
umber of high resolution crystal structures
the protein are available, but the
hanisms of its activation and DNAinding
specificity remain poorly understood,
"lain priority is therefore to determine
solution structure of the proteinion
ucleotide complex, and the effects on
of changes in protein and oligonucleotide
S ure; our work to date has lar s el y been
* to developing the necessary methods
S. E. H. Syed and G. C. K. Roberts
Biological NMR Centre, University of Leicester,
PO Box 13 8, Medical Sciences Bldg.,
University Road, Leicester LEI 9HN, U.K.
for this. We have, in collaboration with the
group of Jardetzky at Stanford, achieved a
virtually complete sequence-specific assignment
of the *H nmr spectrum of the protein
[5]-[7], and Jardetzky's group have used this
data to determine the solution structure of
the repressor-tryptophan complex [8]. Nmr
studies of the binding of corepressors such as
L-tryptophan and the inducer, indole-3propionic
acid, show that the environments of
the two classes of ligand in the protein differ,
and strongly suggest that this arises from a
difference in the orientation of their indole
rings [9]. Even the repressor alone is of such
a size that these studies required the
extensive use of isotope labelling, and we
have developed appropriate expression
systems for efficient biosynthetic labelling of
the protein. The overall molecular mass of
the repressor-tryptophan-operator oligonucleotide
complex is ca. 38,000, and for this
complex rather little useful information can
be obtained by nmr without labelling.
Chloramphenicol acetyltransferase is
responsible for resistance to the antibiotic
chloramphenicol in bacteria. Its crystal
structure is known to high resolution, and it
is the subject of a substantial programme of
protein engineering in the laboratory of Prof.

66 Bulletin of Magnetic Resonance
W.V. Shaw in Leicester [10]. Since it is a
trimer of total molecular mass 75,000, it is
very large for study by nmr, and we are using
it as a test system for the development of nmr
methods suitable for large proteins, focussing
particularly on complexes, such as that with
the product, diacetyl-chloramphenicol, which
have not proved amenable to crystallographic
study.
2. Deuteration
Selective deuteration has been used to
simplify the *H nmr spectra of proteins for
many years [11], and more recently has been
combined with 2D nmr spectroscopy (e.g.,
[12]). Selective deuteration of carefully
chosen combinations of residues has been
combined with the sequential assignment
method of Wiithrich (see [1]) to yield a
considerable number of resonance
assignments in the trp repressor, both alone
and in its complex with an operator
oligonucleotide [6], [7]. With the improved
resolution of the 3D spectra, we have
recently assigned all the intermolecular NOEs
between protons of the protein and those of
bound tryptophan, permitting the kind of
'ligand-docking' experiments which we have
already carried out on phospholipase A2 [13].
In a species as large as the repressor-operator
complex, the substitution of most of the
protons in the protein by deuterons leads to a
notable decrease in the resonance linewidth
of the remaning protons; similarly, we have
employed random fractional deuteration [14]
to good effect in these large systems. In
particular, deuteration of chloramphenicol
acetyl-transferase to the level of about 85%
had three beneficial effects: (a) it allowed the
unambiguous observation and assignment of
resonances of the bound substrate, (b) it
allowed the observation of intra-molecular
NOEs in the bound substrate, thus defining
its conformation, and (c) it allowed the
observation of inter-molecular NOEs
between protons of the (isotopically normal)
substrate and nearby residues of the protein
[15].
3. 13 C and 15 N labelling
Notwithstanding the usefulness of selective
deuteration, complete assignments of the
backbone resonances of the repressor have
required the use of 15 N-labelled protein, in
combination with 3D nmr spectroscopy [7].
The same approach has led to a substantial
number of resonance assignments in the
repressor-operator complex, and to the clear
observation of inter-molecular NOEs [7],
[16]. In addition to the standard NOESY-
HMQC and TOCSY-HMQC experiments,
we have found the HMQC-NOESY-HMQC
experiment [17] valuable in this a-helical
protein in which there is significant overlap of
both *H and 15 N chemical shifts of the
backbone amides. In some cases, there is a
considerable benefit in selective labelling; in
the trp repressor we have used this
successfully for 15 N-leucine and for [amide-
15 N]-asparagine and glutamine [16], and in
chloramphenicol acetyltransferase for
[imidazole 2- 13 C]-histidine [18]. The latter
experiment permitted the detection of the
histidine C2- 1 H signals, and the use of 2D
*H- 13 C correlation spectra allowed
overlapping signals to be resolved, even in a
protein of Mr 75,000. We are also exploring
the usefulness of 50-65% perdeuteration in
combination with 15 N-labelling as a means of
improving the quality of the 3D spectra.
In chloramphenicol acetyltransferase we
have used nmr to study the binding of
diacetyl[ 13 C]-chloramphenicol by means of
13 C-edited l U- l B. NOESY experiments. A
number of clear intermolecular NOEs were
observed, in particular to aromatic protons of
the protein. Candidate aromatic residues
were identified by model-building on the
basis of the crystal structure, and were
replaced in turn by isoleucine residues. 13 Credited
NOESY spectra of the complexes with
these mutants allowed the two aromatic
residues with which the acetyl groups of the
ligand made contact to be identified
unambiguously, thus allowing the orientation
of the product in the binding site to be

Vol. 14, No. 1-4 67
defined [15], and assisting in the modelling of
the transition-state complex.
4. References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
Wiithrich, K., Science 243 (1989) 45.
Bax, A. (1989) Ann. Rev. Biochem. 58,
223.
Oppenheimer, N.J., and James, T.L.
(1989) Methods in Enzymology, 176,
Academic Press, New York.
Oppenheimer, N.J., and James, T.L.
(1989) Methods in Enzymology, 177,
Academic Press, New York.
Hyde, E.I., Ramesh. V., Roberts.
G.C.K., Arrowsmith, C.H., Treat-
Clemons, L., Klaic, B., and Jardetzky,
O. (1989) Eur. J. Biochem. 183,545.
Arrowsmith, C.H., Pachter, R.,
Altman, R.B., Iyer, S.B., and
Jardetzky, O. (1990) Biochemistry 29,
6332.
Ramesh, V., Frederick, R.O., Syed,
S.E.H., and Roberts, G.C.K., unpublished
work; Arrowsmith, C.H., and
Jardetzky, O., unpublished work.
Arrowsmith, C.H., Pachter, R.,
Altman, R.B., and Jardetzky, O. (1991)
Eur. J. Biochem., 202, 53.
Hyde, E.I., Ramesh, V., Frederick, R.,
and Roberts, G.C.K. (1991) Eur. J.
Biochem., 201, 569.
Shaw, W.V., and Leslie, A.G.W.
(1991) Ann. Rev. Biophys. Biophys.
Chem., 20, 363.
[11] Jardetzky, O. & Roberts, G.C.K.
(1981) NMR in Molecular Biology,
Academic Press, New York.
[12] Feeney, J., Birdsall, B., Akiboye, J.,
Tendler, S.J.B., Barbero, J.J., Ostler,
G., Arnold, J.R.P., Roberts, G.C.K.,
Kuhn, A., & Roth, K. (1989) FEBS
Lett., 248, 57.
Bennion, C, Connolly, S., Gensmantel,
N.P., Hallam, C, Jackson, C.G.,
Primrose, W.U., Roberts, G.C.K.,
Robinson, D.H., and Slaich, P.K.
0992) J. Med. Chem., in press.
[14] LeMaster, D.M. (1990) Quart. Rev.
Biophys., 23, 133.
[15] Derrick, J.P., Lian, L.-Y., Roberts,
G.C.K., and Shaw, W.V. (1992)
Biochemistry, in press.
[16] Ramesh, V., Frederick, R.O., Syed,
S.E.H., and Roberts, G.C.K. (1992)
manuscript in preparation.
[17] Frenkiel, T., Bauer, C, Carr, M.D.,
Birdsall, B., and Feeney, J. (1990) J.
Magn. Reson., 90, 420.
[18] Derrick, IP., Lian, L.-Y., Roberts,
G.C.K., and Shaw, W.V. (1991) FEBS
Lett., 280, 125.

68
Bulletin of Magnetic Resonance
2D NMR STUDY OF DRUG-PROTEIN INTERACTIONS : ETHIDIUM BROMIDE -
NEOCARZINOSTATIN COMPLEX
Introduction
Smita Mohanty*> Larry C. Sieker"*" and Gary P. Drobny*
Department of Chemistry* and Biological Structure 1 "
University of Washington
Seattle, WA 98195, USA
N eocarzinostatin (NCS) is a small
acidic holo-protein isolated from the
culture broth of Streptomyces
Carzinostaticus [1]. It has a protein
component (apo-NCS) of 113 amino acid
residues and a non-covalently bound
heat and light sensitive chromophore
(NCS-chr) (Fig.l). This protein possesses
antibiotic activity against organisms
such as Sarcina Lutea and antitumor
activity against the experimental tumors
Ascitic Sarcoma 180, Ascitic
Leukemia SN-36, Leukemia L-210
[1-3]. It is known that the chromophore
is responsible for all the biological
activities and the apo-protein stabilizes
and acts as a carrier for this UV
sensitive component of the antitumor
drug [4].
Though the secondary and tertiary
structure of the apo-protein is well
understood from X-ray and NMR studies,
little is known about the binding region
and the amino acid residues involved in
the drug - protein interactions in the
holo-protein.
Both the crystal structure at 2.8 A
[5] and the 2-D NMR work done on apo-
NCS [6-8] indicate that a major part of
the protein is composed of a seven
strand antiparallel B-sandwich formed
by a three strand B-sheet and a four
strand B-sheet. The rest of the protein is
composed of two loops oriented
somewhat perpendicularly to the B
sandwich, thus forming a distinct Ushaped
cleft between the four strand
face of the sandwich and one of the
loops of the molecule (Fig. 2). The
crystal structure of holo-NCS at 2.0 A
resolution indicates that the
chromophore is located in this cleft. But
a detailed knowledge in the region of
the chromophore could not be obtained
from the 2.0A map. Preliminary NMR
studies also indicate that the
chromophore binds within the cleft and
interacts with amino acid residues in the
region C37-D48 [9, 10]. Although NCS
does not bind the chromophore of other
streptomyces derived anti-tumoi
proteins (e.g. Auromomycin) [11] it is
known from X-ray studies [12] tc
strongly bind a number of drugs
including ethidium bromide (Fig. 3) anc
daunomycin. In order to furthe
elucidate the nature of drug-NC!
interactions, we have initiated a study o
both holo-NCS and the complex betwee
ethidium bromide and NCS.
Materials and Methods
Apo-NCS solution used in o 1
experiments was prepared from hoi
NCS. The chromophore was extracted
i

Vol. 14, No. 1-4
Fig. 1: NCS-Chromophore
60
Fig. 2: Ribbon picture of NCS by X-ray
Fig. 3: Eihidium Bromide
69

I:
I
70
the procedure of Napier et.al.[13].
Purified lyophilized holo-NCS was a gift
from Kayaku Co. Ltd. The 1:1 Ethidium
Bromide-NCS complex was prepared by
adding the protein solution in to vials
containing the drug. The final solution
was purified by passing through
sephadex G-25 column followed by
lyophilization. The lyophilized complex
was brought up in lOmM acetate buffer
(pH 5, 90% H2O/ 10% D2O for non
exchanged protein sample and 99.98%
D2O for exchanged protein sample) and
lOmM EDTA. The concentration was
adjusted between 2.0 mM to 2.5 mM for
500|il sample.
All NMR experiments were
performed on a Bruker AM-500
Spectrometer at 313 K. DQF- and TQF-
COSY [14], RELAY [15], TOCSY [16] and
NOESY [17] were acquired in TPPI mode
with standard phase cycling schemes.
The water resonance was presaturated
by selective irradiation between 1.5 s to
2 s. RELAY spectra were performed with
mixing times of 30 ms (90%H2O) and 25
ms (for 99.98%D2O). TOCSY spectra were
performed with a variety of mixing
times ranging between 40 to 80 ms.
NOESY spectra were recorded with 150
ms mixing time, randomly varied by
10%. The data were processed with
FTNMR software of Dr. Dennis Hare [18].
Results and Discussion
Our H-NMR assignments of
ethidium bromide - NCS complex in
solution indicates the drug to be located
in the cleft region. There are two lines of
evidence which support this conclusion.
First, the chemical shifts of numerous
Bulletin of Magnetic Resonance
SI nn — horn ethidium
to the ring ^ aromatic system,
bromide's extensive aromatic y
£> tp"t e -45 Gln Cy;-47 y , Asp-**
Cy's-9 Gln-94, Leu-97 in the
fingerprint region (Fig. 5 and F.g. 6).
0.0
ppm
Fig 4: DQF-COSY spectrum in D2O
' snowing uprtcldshid or Uu_4
chemical shifts (a): apo-NCS. (b).
P ,

Vol. 14, No. 1-4
O36
3 '•
Q94" 5
C47 'C37
XL -
L97
8.8 8.0
ppm
9
7!2
Fig. 5: DQF-COSY fingerprint region of apo-NCS recorded at 40'C .
so
10 03
C93
DO
OK)
00
i< 00
00

72
8.0
ppm
7. 2
Bulletin of Magnetic Resonance
Fig.7: DQF-COSY and NOESY spectra of the complex in D2O
showing some of the intermolecular NOE observed between
the aromatic protons of ethidium bromide (shown by dotted
lines) and B-protons of certain residues within the cleft.

Vol. 14, No. 1-4 73
While residues Cys-93, Gln-94,
Leu-97 are in the four strand face of the
6-sandwich, which forms one side of the
cleft, residues Gln-36, Cys-37, Ala-38,
Trp-39, Leu-45, Cys-47, Asp-48 are in
one of the loops that forms the other
side of the cleft. Second, a number of
NOEs have been observed to occur
between protons on ethidium bromide
and residue protons within the cleft. The
Leu-45 methyl group is ring current
shifted and shows NOEs to the
methylene protons and aromatic protons
of ethidium bromide. Intermolecular
NOEs are also observed between the
aromatic protons of ethidium bromide to
the aromatic proton of Trp-39 and to the
8 protons of Ser-98, Cys-37 and Gly-96
(Fig. 7). Additional intermolecular NOEs
are observed but have not been
unambiguously assigned.
Conclusion
NMR assignment based on
i coherence transfer experiments and
y

74
[11] L. S. Kappen, M. A. Napier, I. H.
Goldberg, and T. S. A. Samy,
Biochemistry., 19, 4780.(1980).
[12] L. C. Sieker, Personal communication(
1991).
[13] M. A. Napier et.al., Biochem.
Biophys. Res. Commun., 89,
635. (1979).
[14] U. Piantini, O. W. S0rensen, and
R. R. Ernst, J. Am. Chem. Soc,
104, 6800. (1982).
[15] G. Eich, G. Bodenhausen, and R.
R. Ernst, J. Am. Chem. Soc, 104,
3732. (1982).
[16] L. Braunschweiler and R.R.
Ernst, /. Magn. Reson., 53, 521.
(1983).
[17] Anil Kumar, R. R. Ernst, and K.
Wuthrich, Biochem. Biophys.
Res. Commun., 95, 1. (1980).
[18] D. Hare, Hare Research:
Woodinville, WA.
BulJetin of Magnetic Resonance
'•A

Vol. 14, No. 1-4 75
1 Introduction
Transferred NOE experiments are being utilized to
determine the structure of small peptide ligands bound to
proteins of molecular weight up to 10 6 daltons (1-4). For
larger peptides, 2D phase-sensitive NOESY is the optimal
method for obtaining transferred NOE's between all the
protons in the free ligands (1-2). In case of further
spectral overlap, transferred NOE's between protons of
selected amino acid residues can be obtained by use of the
homonuclear TOCSY-editted 2D NOESY technique (5).
Ultimately, NOE's between all the spin systems may be
resolved by use of the 3D NOESY-TOCSY experiment
(6). The resolved NOE's can be input into a procedure for
the refinement of the dynamic structures of bound ligands
(7).
In practice, the interpretation of NOESY data is
complicated by the presence of the large solvent resonance
and baseplane problems, especially for samples with very
low concentration of the material under study. In addition,
magnetization transfers from the binding protein often
result in further spectral distortions. We have been
, developing methods to improve the quantitative accuracy
|of transferred NOE's (7-9). In this paper, we summarize
|Procedures developed for the optimized acquisition and
ssing of multi-dimensional transferred NOE spectra
• that more accurate quantitative results can be obtained
i experimental data.
Quantitative Analysis
in Multi-Dimensional Transferred NOE Experiments:
Improved Spectral Acquisition and Processing
jlimination of Baseline Distortions in
volution Dimensions
evel of tj ridges in 2D phase-sensitive NOESY
~ can be minimized if the FID matrix is recorded
proportional phase incrementation (TPPI) with
lulation along n (10). Since the tj interferograms
zero at the zero time for a sine-modulated
to our implementation is to start the ex-
Feng Ni
Biotechnology Research Institute
National Research Council Canada
6100 Royalmount Avenue
Montreal, Quebec, Canada H4P 2R2
periment at the second FID with a compensated initial
delay of,
where IN = At/2 is the increment time between successive
FID's and x9o is the width of the 90° pulse. This
implementation of the sine modulated NOESY experiment
completely removes Fj baseline distortions (Figure 1A)
with no need for any further baseline correction along this
dimension (8). Furthermore, the resulting NOESY matrix
can be phased to absorption along the F i direction with a
Oth order phase of exactly 90° (in effect, a sine transform).
This eliminates possible baseline distortions associated
with phase correction after a real Fourier transformation
(11). The procedure of delay-compensated sine modulation
has also been applied successfully along the evolution
dimensions of both homonuclear and heteronuclear 3D
experiements (9,12).
3 Ridge Suppression Along Detection
Dimension
The cause for ridges (baseline offsets and curvatures) in the
detection dimension is more complicated than that for
ridges along the evolution dimensions. There is often
need for further correction in the frequency domain. Figure
2a shows the baseline points recognized based on the first
derivative (13) of a row slice of the NOESY spectrum in
Figure 1A. The regions with sharp peaks are filled with
interpolations (linear or polynomial) from adjacent
baseline points. It is seen that the recognized baseline
includes all the broad signals in the original spectrum.
One can then construct a smooth curve through the
available baseline points with some sort of curve fitting
(Figures 2b and 2c). We adopted a simple and fast method
[1]

76
Bulletin of Magnetic Resonance
Figure 1: Enhanced processing of 2D transferred NOE spectra. The FID matrix was acquired using sine-modulation along the
ti direction. All spectra were processed by use of cosine-square windows in both directions and were plotted with the same
parameters except that post-acquisition water suppression (8) was applied in (A)-(D); linear-prediction baseline correction was
applied in (B); polynomial baseline correction was applied in (C); and baseline Fourier filtering was applied in (D).
0.0

Vol. 14, No. 1-4
10.0 8.0 0.0
ppm
Figure 2: Baseline fitting from incomplete data, (a) baseline points from one row of Figure 1 A. The missing points were
filled with linear interpolations, (b) one possible baseline reconstructed by Fourier smoothing, (c) an approximate baseline
calculated by a fit to a polynomial of fifth degree.
for data smoothing based on Fourier filtering (8). Figure
2b is the result of a 30-point Fourier smoothing of
Figure 2a. Figure ID shows the NOESY spectrum
(Figure 1A) after subtraction of the Fourier-filterred
baseline points along the F2 direction. In comparison to
-••• other methods (Figures IB and 1C), there is a dramatic
... improvement in the clarity of the spectrum with complete
^ elimination of broad signals from the original spectrum.
Optimized Spectral Processing
frequency domain spectrum is usually generated via a
i Fourier transform of the interferogram if the data are
"pled using the TPPI procedure (14). The computed
"jtrum contains both real and imaginary parts which are
|;cosine or sine transforms of the original real data,
ctively. Theoretically, an absorption spectrum is
ly the cosine transform of a cosine modulated FID or
i;ne transform of a sine modulated FID (10). In
however, both the real and imaginary parts have
Iculated so that they can be suitably combined
»g) to correct for possible phase distortions. For
|«mensional NMR experiments, the evolution
not ~ns can usually be sampled using the procedure of
[jpensated sine (or cosine) modulation (see section
2) to obtain well-phased interferograms (8,9,15). Thus,
the processing procedure can be greatly simplified if only
the sine or the cosine transform of the data is retained. We
thus implemented the procedures for fast sine (or cosine)
transformation (16). Compared to a real Fourier
transform, there is in principle a factor of two increase in
computational efficiency (16).
To minimize truncation artifacts, special attention
must be paid to the selection of window functions and/or
to data extension by use of linear prediction. Attenuation
of truncation is usually accompanied by broadening of
spectral peaks (14). Linear prediction data extension, on
the other hand, tends to increase the noise level of the
spectrum as a result of errors accumulated with predicted
data points (17, 18). We found that linear prediction
followed by data windowing usually gives a good compromise
between spectral resolution and noise suppression
(9). In this case, truncation is removed by linear
prediction while linear prediction errors are attenuated by
window functions (18). In multi-dimensional NMR,
linear prediction should be used only with the last spectral
dimension after Fourier transformations along all other
directions. Otherwise, linear prediction parameters must
be carefully optimized to minimize error accumulation in
the intermediate stages of multi-dimensional Fourier
transformation.
77

78
5 Sensitivity Enhanced 3D NOESY-
TOCSY
In the usual pulse sequence for 3D NOESY-TOCSY,
spin-locked coherence transfers are achieved by use of the
trimmed MLEV-17 sequence (Figure 3A). We replaced
the MLEV-17 sequence by a z-filtered WALTZ-16 pulse
sequence (Figure 3B) to obtain sensitivity enhancement.
This method involves post-acquisition combinations of
the FIDs acquired from sub-groups of the full phase cycle
utilized for the pulse sequence (9). The two detected FIDs
B
MLEV-17
WALTZ-16 ($,,
acq(t3)
Figure 3: Pulse sequences for 3D NOESY-TOCSY. (A).
The original 3D sequence incorporates a MLEV-17 pulse
sandwiched by two short cw spin-lock pulses (trimmed
MLEV-17). (B). The new 3D NOESY-TOCSY features
a z-filtered WALTZ-16 sequence for spin-locking. The rf
phases are cycled as 4>i=x, -x, y, -y; 2=-x, -x, -y -y;
fa=-x, -x, -y, -y; 4=x, x, y, y; sl=y, y, -x, -x; 5=x, x,
y, y; acq=x, -x, y, -y. For the same values of l\ and t2,
5 is incremented by 180 degrees and another FID is
acquired and stored in a different memory location. These
seperated FIDs can be suitably combined to obtain a
sensitivity-enhanced 3D NOESY-TOCSY spectrum
(Figure 4).
are stored in separate memory blocks for subsequent
processing. Adequate combinations of these FIDs would
restore the two orthogonal components, Iiy and l[x. If
both the Iiy and 1^ FIDs are processed to yield absorptive
spectra, they can be combined to produce a spectra with
peaks doubled in size compared to each individual
spectrum. The key to sensitivity enhancement lies in the
fact that noise components in the two spectra are
statistically independent and the post-processing combinations
are then equivalent to signal averaging, reducing
noise in the process (19).
Bulletin of Magnetic Resonance
With the sensitivity-enhanced NOESY-TOCSY, we
utilized delay-compensated sine modulation along both ti
and t2 to eliminate baseline distortions. Baseline
adjustments along F3 were achieved in the frequency
domain by use of Fourier baseline reconstruction (8). An
absorptive spectrum can be obtained from Ijy if a sine
transform is applied along both tj and t2- With the Iix
components, an absorptive spectrum can be obtained only
if a sine transform is used along t\ and a cosine transform
used along t2. However, the points of the 1^ matrix for t2
= 0 can not be sampled due to the finite widths of the 90
degree pulses. Therefore, the missing first point was
estimated via a linear prediction algorithm (17).
Alternatively, the first data point can be left as zero during
cosine transformation. Baseline offsets as a result can be
corrected afterwards in the frequency domain.
Sensitivity-enhanced NOESY-TOCSY was applied to
an anticoagulant pep tide (2). The concentration of the
peptide was 6 mM and the thrombin concentration was 0.5
mM in an aqueous solvent of 90% H2O and 10% D2O at
pH 5.5. The experiment composed of four scans for each
of the two FIDs with each pair of fixed values for ti and
t2- The data were acquired on a Briiker AMX-500 MHz
NMR spectrometer and the 3D FID matrix (I(y or 1^ components)
was of the sizes 512(t3) x 160(ti) x 62(t2).
Residual solvent signals were suppressed by linear
prediction time-domain convolution (8). The data along ti
was extended to 200 and along t2 to 80 by linear prediction
(17). Kaiser windows were used in all dimensions to
reduce the effects of error propagation in linear prediction.
Frequency-domain baseline adjustments were applied only
along the F3 spectral dimension. The final sizes of the
spectra were 512(F3) x 256(Fj) x 128(F2). Figure 4A is
the F1-F2 plane sliced through the frequency of one of the
well-resolved 6CH2 protons of Pro along the F3
dimension of the Iiy 3D spectral matrix. The corresponding
IiX spectrum (Figure 4B) also contains similar
information but with somewhat reduced intensities for
some of the crosspeaks. This is probably due to the fact
that the 1^ components travel through different transfer
pathways and are much more sensitive to prolonged delays
and/or phase offsets both before and after the WALTZ-16
spin-lock pulse. Nonetheless, combination of the two
spectra still produced a sensitivity-enhanced spectrum
(Figure 4C). This is evident if one compares the NOE
crosspeaks between the aCH and PCH protons of ProfiO
and the NH proton of Glufi! (60A/61N and 60B/61N of
Figure 4) and if one inspects the selected slices (Figure 5)
through the sensitivity-enhanced spectrum compared to the.,
original spectrum. |

Vol. 14, No. 1-4
A
t
608/6 IN
.' i |,
608/63E
600/630 1
.6 ft* fom
\ ft
60A/6IN AX
0 /
6.0 4.0
F1( ppm )
a /^ 59C/600
608/6 IN
2.0 8.0 6.0 4.0
F"K ppm)
2.0
Figure 4. 3D transferred NOES Y-TOCSY spectra of an anticoagulant
peptide, DSS-F-E-E-I-P-E-E-Y-L-QGS. (A) the
plane was extracted from the conventional 3D NOESY--
TOCSY Iiy spectrum. Only positive levels above 0.008
are plotted. (B) the same plane as in (4A), but extracted
from the 1^ spectrum. Only negative levels below -0.008
are plotted. (C) the same plane as in (4A) and (4B), but
from the combined spectra of both the and the Ijy components.
The contour levels are above 0.012.
79

80
10.0
o.o
Figure 5. Selected slices from the sensitivity-enhanced 3D
spectrum (top trace in each box) compared to those from
the conventional spectrum (bottom trace in each box).
Top trace of Box A, ID spectrum sliced through the PCH2
chemical shift of Pro GO (labelled as 60B in Figure 4C);
bottom trace of Box A, the corresponding slice from
Figure 4A. Box B, slices corresponding to the aCH
chemical shift of Pro^o (Labelled as 60A in Figure 4).
6 Summary
We have optimized procedures for both acquisition and
processing of homonuclear 2D and 3D spectra in aqueous
solutions. These include a new 3D NOESY-TOCSY
pulse sequence that can be used with sine modulation to
simplify spectral processing and to improve spectral
baselines. It is also demostrated that orthogonal
components of the spin-locked magnetizations can be
suitably combined during processing to achieve sensitivity
enhancements for 3D NOESY-TOCSY. These improved
schemes are not limited to transferred NOE experiments.
They should be of general applicability for resonance
assignments and structure determination of dilute proteins.
7 References
Bulletin of Magnetic Resonance
!R Ni, Y. Konishi, R. B. Frazier, H. A. Scheraga, and S.
T. Lord, Biochemistry 28,3082 (1989).
2 F. Ni, Y. Konishi, and H. A. Scheraga, Biochemistry
29, 4479 (1990).
3A. P. Campbell, and B. D. Sykes, J. Mol. Biol. 222,
405 (1991).
4 S. J. Landry, R. Jordan, R. McMacken, and L. M.
Gierasch, Nature 355,455 (1992).
5 V. Sklenar, and J. Feigon, J. Am. Chem. Soc. 112,
5644 (1990).
6G. W. Vuister, R. Boelens, and R. Kaptein, /. Magn.
Reson. 80,176 (1988).
TF. Ni, J. Magn. Reson. 96,651 (1992).
8F. Ni, /. Magn. Reson. (1992a), in press.
9F. Ni,7. Magn. Reson. (1992b), in press.
10
G. Otting, H. Widmer, G. Wagner, and K.
Wuthrich, J. Magn. Reson. 66,187 (1986).
11D. Marion, and A. Bax, 7. Magn. Reson. 79, 352
(1988).
12K. A. Carpenter, and F. Ni, J. Magn. Reson.
(1992), in press.
13 W. Dietrich, C. H. Riidel, and M. Neumann, J.
Magn. Reson. 97,1(1991).
14
R. R. Ernst, G. Bodenhausen, and A. Wokaun, "
Principles of Nuclear Magnetic Resonance in One and
Two Dimensions", Clarendon Press, Oxford, 1987.
15
D. Marion, and A. Bax, J. Magn. Reson. 83, 205
(1989).
i&W. H. Press, B. P. Flannery, S. A. Teukolsky, and
W. T. Vetterling, "Numerical Recipes'. The Art of
Scientific Computing," Chap.13 and 14, Cambridge Univ.
Press, Cambridge, 1986.
«F. Ni, and H. A. Scheraga, /. Magn. Reson. 70,
506 (1986).
l«E, T. Olejniczak, and H. L. Eaton, /. Magn. Reson.
87,628 (1990).
19 J. Canavagh, and M. Ranee, J. Magn. Reson. 88,
72 (1990).

Vol. 14, No. 1-4 81
TIME-RESOLVED SOLID-STATE NMR: SMALL MOLECULES AND ENZYMES
IN RAPIDLY FROZEN SOLUTION
INTRODUCTION
Jeremy N. S. Evans§**, Richard J. Appleyard§ and Wendy Shuttleworth§
Departments of Biochemistry/Biophysics§ and Chemistry^,
Washington State University, Pullman, WA 99164-4660. U. S. A.
An important aspect of understanding enzymatic reaction
mechanisms is the determination of the molecular structures
of enzyme-bound substrates and products. NMR spectroscopy
is in a unique position in that it is the only relatively
high resolution structural technique which can focus
on specific parts of the molecule, such as enzyme-bound
intermediates. It has been widely applied to the study of enzyme
mechanisms in both the solution and solid states [1-3]
at ambient temperatures and at sub-zero temperatures [4].
However, the current molecular weight limit in solution is
around 50 kDa [2]. In contrast, with solid-state NMR spectroscopy
[5] anisotropic and dipolar broadening can be
reduced by cross-polarization magic angle sample spinning
(CP-MASS), and there is no known molecular weight limit.
We have therefore sought to develop a new technique called
time-resolved solid-state NMR spectroscopy [6] which involves
applying CP-MASS NMR to studying rapidly
freeze-quenched enzyme-substrate mixtures.
The rapid freeze-quench method involves rapidly mixing
enzyme and substrate together and freezing by spraying the
mixture directly into a secondary cryogen such as liquid
propane cooled to ~ 85 K. CP-MASS NMR of the enzymesubstrate
mixture is carried out as a function of mixing
time, and the transient enzyme-bound species may be
detected. Since it is largely unknown how the NMR spectrum
of solutes are affected by the structure of water in
frozen solution, we have studied this with a small molecule,
glycine [7], and report the results in preliminary form here.
We have subsequently applied these methods to the direct
observation of an enzyme-intermediate complex by timeresolved
solid-state NMR spectroscopy [6] and report our
preliminary results here.
MATERIALS AND METHODS
NMR spectroscopy was carried out using a wide-bore,
T Chemagnetics CMX-400 spectrometer, operating at
iPO.l MHz for *H and 100.6 MHz for 13 C. The small
nolecule studies were carried out using zirconia rotors (7
m) in a double resonance, variable temperature (VT),
ic angle sample spinning (MASS) Chemagnetics proto-
Pencil-rotor probe. The enzyme studies were carried out
*? Dotv °SI-368 double resonance probe using a 5mm
jpnire rotor. Stable spinning was achieved with a
"lagnetics spinning speed controller, which uses a
'"-cessor controlled valve on the drive gas line to
the speed ±5 Hz. The MAS probe was of a triple
mel (dnve, bearing and VT) design. Boil-off N2 gas
was cooled by an exchange dewar filled with liquid nitrogen
(280 kPa, -110K). The sample cooling was performed by
cooling the VT line, with the drive and bearing lines at approximately
room temperature (-293K). Stable VT operation
could be achieved for over 24h with this apparatus.
The nuclear relaxation constants were determined by
established techniques [8-10]. The proton Zeeman spinlattice
relaxation, Ti H , was determined using the inversion
recovery technique [11], with 13 C detection via cross
polarization. The carbon Zeeman spin-lattice relaxation,
Ti c , was determined using a modified cross polarization and
inversion recovery technique [12]. The rotating-frame spinlattice
relaxations, Tip H and Tjp C , were determined using
an appropriate spin lock (50kHz) of varied length, with 13 C
detection via cross polarization [9,13]. Unlabelled glycine
was purchased from Sigma (St Louis, MO). 99% [1-
13 C] Glycine was purchased from Cambridge Isotope
Laboratories (Cambridge, MA), and 99% [2- 13 C]Glycine
was purchased from MSD Isotopes (Canada). Propane gas
was purchased from Bemzomatic (Medina, NY). The crystalline
glycine sample was re-crystallized from a saturated
aqueous solution, using ethanol as the triturant. The dimensions
of the crystals were consistent with the a-form. A 1M
1:1 solution (pH 7.5) of [l- 13 Ci]glycine and [2-
13 Ci]glycine was used for the frozen samples. The slow
frozen samples were prepared in situ. A 200 pL sample was
loaded into the rotor and spun slowly (

82 Bulletin of Magnetic Resonance
TABLE 1 Activation Energies for motions in different states of glycine from Ti and T\p data.
Ea (kj mol* 1 )
Crystalline
Slow Frozen
Fast Frozen
(10"
Crystalline
Slow Frozen
Fast Frozen
1 Ks" 1 )
(10 5 Ks 4 )
(10" 1 Ks" 1 )
(10 5 Ks" 1 from Ti
)
H data
/romTj
23.5
21.1
19.4
from Tip** data
28.2
24.6
23.1
c Cl
data
23.5
16.5
19.5
from Tip c data
Cl
13.6
4.4
5.2
BL21(XDE3)(pLysS)(pWS230) and purified by literature
methods [14].
RESULTS
The structures of frozen water have been extensively studied
in electron microscopy [15] and materials science [16],
and has been shown to be greatly affected by the freezing
rate used. The main problem with water is that due to its
strong propensity for hydrogen bonding, it is incredibly hard
not to form crystalline hexagonal ice (I^). Two other forms
of frozen water known are cubic ice (Ic) and vitreous water
(Iv), also referred to as amorphous solid water (ASW). ^ ice
is a metastable form of ice (with respect to 1^ ice) produced
when Iv is heated. In Iv, the water molecules are randomly
distributed throughout the solid phase having been unable to
form an Ih lattice during the freezing process.
The freeze-quench apparatus used in these studies is a
modified version of that used in rapid freeze quench ESR
spectroscopy studies [17,18]. It involves firing a solution
at a pre-determined flow-rate through a fine nozzle into a
receptacle containing a secondary cryogen (e.g. liquid
propane) cooled by immersion in a primary cryogen (usually
liquid nitrogen). The cooling rate is dependent on the size of
the droplets created by the exit nozzle and can be determined
by the study of a reaction with a known rate. ESR studies
[17,18] have estimated the freezing time by this technique to
be 2-6 ms and the cooling rate to be -10 5 Ks' 1 .
We have investigated how nuclear relaxation rates of a
solute in frozen solution are affected by the freezing rate,
since they have been shown to be a sensitive probe of
molecular motion in the solid state [19,20]. We have chosen
to study glycine because of its simple mode of motion that
has been characterized previously using NMR relaxation
techniques [21]. Since the relaxation properties of pure ice
have also been studied [22,23], this provides a good model
for characterization of the molecular motions of solute
molecules in frozen solution.
Previous studies [21,24] of polycrystalline glycine have
shown that the main source of Tj H relaxation is the random
modulation of the proton magnetic dipolar interaction by the
C2
30.2
17.6
19.4
C2
0.26
3.3
1.6
reorienting ammonium group (-NH3+). The activation
energy of the -NH3 + rotation in glycine was determined to be
28.6 kJ mol" 1 which agrees well with our T^ and Ti c Ea
values for crystalline glycine shown in Table 1. The Ea values
from the Tip C data (Table 1) display a deviation from
this value, which is even more evident for the slow and fast
frozen states. However, since the latter fits are less welldefined,
these Ea values should be regarded as less reliable.
The crystal structures of a-glycine has been solved to a
high degree of accuracy by X-ray [25] and neutron diffraction
[26] techniques. The glycine molecule is in the zwitterion
form. The dipolar glycine molecules are linked by two short
-N-H--O- hydrogen bonds to form layers connected in an
antiparallel manner by weaker -N-H •••0- hydrogen bonds.
The close packing nature of the crystalline lattice, coupled
with the hydrogen bond interactions, will produce large barriers
to rotation for both the -NH3"*" and -CO2' groups as
observed.
The relaxation process in the frozen solution is more
complicated due to the presence of an additional relaxation
pathway provided by the ice lattice. Previous studies [22,23]
have shown the measured activation energy for the Tj and
Tip processes in pure crystalline ice to be 59.8 kJ mor 1 .
This gives a xc (-10°C) = 7.5 (is, and it has been concluded
[23] that the dominant magnetic relaxation mechanism must
be the diffusion of Schottky defects through the ice lattice.
Although there are insufficient data points for the number of
variables to fit two correlation times, it is clear that the diffusion
due to Schottky defects is not detected since no activation
energy of ~60 kJ mol" 1 is evident from the Ti dau
for the frozen solutions given in Table 1. The field strengu
used is much higher than those in previous studies
Therefore the (Ti)min occurs at a temperature above tin
melting point of the ice and cannot be detected.
Table 1 clearly shows a drop in Ea for -NH3 group rot*
tion of glycine calculated from the Ti H and Tip H data in tr
order, crystalline; slow-frozen solution; fast-frozen soluua
The structure of the frozen solution samples is less cle|
Many studies [16,27] have been made of water frozen unfl

Vol. 14, No. 1-4 83
a variety of different freezing rates. Much effort has been
made to develop techniques of freezing with rates that
approach 10 5 - 10 10 Ks- 1 , the estimated [28-31] rates
necessary to form vitreous ice, Iv (amorphous solid water).
It is now generally accepted that freezing rates >10 5 Ks" 1
are required before Iv will form, however there are other
factors to consider, such as sample size, freezing medium
and so on. For solutions frozen at rates well below the 10 5 -
10 10 Ks -1 range, the frozen sample is thought to be made
up of bulk hexagonal ice and aggregated solute. For a slowfrozen
1M glycine solution, aggregates of crystalline
glycine will form along with the Ih, i.e. the freezing rate is
slow enough that the crystalline states have time to form.
The crystalline glycine is distributed throughout the Ih
lattice, resulting in a large surface to volume ratio. The
glycine molecules exposed to the ice front have fewer lattice
interactions, giving the -NH3 + group a lower energy barrier
to rotation. This explains the drop in Ea observed in Table I
between crystalline and slow frozen glycine.
Even freezing rates of -10 5 Ks' 1 have been shown [32]
not to prevent completely the segregation of solute from
solvent, nor to form Iv, in both pure water and dilute aqueous
solutions. However, the inclusion of a solute increases
the propensity for vitrification [16], and is the basis for the
use of cryoprotectants. In sufficiently large quantities, a
solute will depress the free energy of the liquid water relative
to the Ih lattice and reduce the driving force for crystallization.
It is therefore unclear what the Iv / Ih ratio is under the
conditions used here. However, once the secondary cryogen
(liquid propane) is removed under vacuum at 223 K, any Iv
that might have been formed at 85 K would readily undergo
a phase transition to Ih at -160 K [33]. Therefore, at the
temperatures used in this NMR study, it is unlikely that Iv
is present. Rapid cooling has been shown [34] to cause a
highly dispersed Ih phase regardless of the degree of vitrification.
A rapidly frozen 1M glycine solution will consist of
.' highly dispersed ice and glycine phases and the very high
^ surface: volume ratio achieved renders the system metastable.
'Under these conditions, maturation can take place with the
^crystal size distribution broadening and shifting to larger
fSfy?^ 1 dimensions. However, the rate is dependent on the
'" ' temperature and is not significant within the timeof
the relaxation studies. It can be rationalized that
very high surface: volume ratio contributes to the
r drop in Ea observed between the slow frozen and fast
solutions. It is also possible that the glycine is
*nt predominantly in the amorphous form in the rapidly
*en sample, since the freezing rate is fast enough to
Jtnt the glycine from forming a microcrystalline
" ""The barriers to the -NH3+ group rotation in the
"i state will be lower since there will not be such a
» of ordered hydrogen bonding, and this will
'- further to the drop in Ea.
applied these methods to a well-characterized
g-€nolpyravylshikimate-3-phosphate (EPSP) syn-
thase (EC 2.5.1.19), which catalyzes the penultimate step in
the aromatic amino acid biosynthetic pathway in higher
plants and bacteria. EPSP (4) is formed from shikimate-3phosphate
(S3P, 1) and phosphoenolpyruvate (PEP, 2) (see
Scheme 1). The enzyme is a monomer with molecular
weight Mx = 46,000 and the cloned E.coli gene has been
used to generate a hyperexpressing strain [14], so that the
bacterial enzyme is available in gram quantities.
Furthermore, EPSP synthase is the primary site of action of
the herbicide glyphosate [35], or N-phosphonomethylglycine.
This is a broad spectrum post-emergence herbicide
with worldwide applications in agriculture and horticulture.
This enzyme has been extensively studied by kinetic and
biophysical methods in the last 5 years. The direct observation
of the enzyme-intermediate (E«I) complex was first reported
by our laboratory [1,2], later confirmed by another
laboratory [36]. There are only a handful of enzymes for
which the full kinetic and thermodynamic profile has been
determined, and EPSP synthase is one of this select group
[37].
Concerns about the fate of the protein under these conditions
of freezing have been addressed [38], and at the protein
concentrations and freezing rates (10 5 K s 1 , or millisecond
time regime) employed here, there is significant dispersal of
the solute in the frozen water [7]. Furthermore the frozen
water is probably largely amorphous [39], with the protein
itself acting in a similar manner to a cryoprotectant [39]
thereby reducing the formation of hexagonal ice that is
detrimental to the protein. We have found that the specific
activity of EPSP synthase employed in the experiments
reported here was found to be unchanged before and after
rapid freezing at these high protein concentrations.
Figure 1 shows 13 C CP-MASS solid-state NMR spectra
of EPSP synthase«S3P mixed with [2- 13 C]PEP under
steady-state conditions (in the presence of the product, inorganic
phosphate) and under pre-steady state conditions where
/ indicates the time elapsed from the start of the reaction.
The intermediate (E«I) is clearly visible at 104 ppm [1,2]
under steady-state conditions and its build-up demonstrated as
the reaction proceeds under pre-steady state conditions. It is
worth noting that the intensities of the E«I resonance correlate
well with the concentrations observed by chemical
quench methods [37]. On allowing the pre-steady state reaction
to proceed for a few minutes, the turnover of intermediate
(E«I) to product (E«EPSP) is evident. In addition to the
resonance due to the E«I complex, the resonance due to the
E'EPSP product complex builds up at 155 ppm, and one
tentatively assigned to the E-PEP substrate complex appears
transiently at 151 ppm. Note that under the conditions
which these spectra were obtained, the free small molecules
(substrate and product) are not detected due to their relative
isotropic motion in frozen solution. Although the spin
locking fields used provide excellent cross-polarization for
rapidly-frozen solutions of EPSP synthase and enzymebound
species also provide very poor cross-polarization for
PEP and EPSP.

84 Bulletin of Magnetic Resonance
240.0 160.0 80.0 0.0
ppm
Figure 1 The 9.4 T 13 C CP-MASS solid-state NMR
spectra of EPSP synthase at 233 K under conditions indicated:
steady-state (EPSP synthase (4 mM) in 20 mM phosphate
buffer, pH 7.8,15 % D2O in the presence of S3P (9.1
mM) and [2- 13 C]PEP (7.6 mM) slow frozen in the NMR
rotor over 90 s; rapidly mixed and freeze-quenched after time
1 (EPSP synthase (4 mM) in 50 mM tris buffer containing
5 mM P-mercaptoethanol, pH 7.8, in the presence of S3P
(40 mM) and rapidly mixed with [2- 13 C]PEP (40 mM) and
sprayed into liquid propane at -85 K.
DISCUSSION
These relaxation studies have provided evidence that the
distribution of the Ih and the solute becomes more
dispersive as the freezing rate is increased. This is
manifested as a lowering of the activation energy for the
-NH3 + group rotation as the amount of glycine free from
crystalline packing forces increases. This is rationalized by
two factors: the increase in surface: volume ratio in the
dispersed phase; and the increase in the percentage of
amorphous glycine present This demonstrates the difference
between fast and slow freezing and suggests that at 10 5 Ks"
1 , either some Iv or highly dispersed Ih is forming. This is
encouraging from the point of view of studying rapidlyfrozen
proteins, since this suggests minimal possible
damage through Ih formation. The characterization of the
freezing rate also indicates that it will be sufficient for
trapping transient intermediates in an enzymatic reaction.
Time-resolved solid-state NMR spectroscopy provides a
major technological advance in the study of enzymatic reaction
mechanisms. One important consideration in any
attempt to detect transient intermediates in addition to their
lifetimes, is their pre-steady state concentrations. This is
dependent upon the kinetics and thermodynamic stability of
intermediates of each individual enzyme. Some enzymes,
like EPSP synthase, have unusually stable intermediates.
However, we would expect that the majority of enzymes are
evolving towards "perfection" [40], and stabilize highly unstable
intermediates. Furthermore, when coupled with the
elegant solid-state NMR distance measurements that have
been introduced recently [41] this technique will be uniquely
able to "map out" molecular conformations of intermediates
and enzyme active site-intermediate distances as an enzymatic
reaction proceeds. This will provide the crucial missing
structural details which Laue X-ray diffraction and allied
techniques cannot provide, and enable the complete definition
of the molecular events of enzyme catalysis.
ACKNOWLEDGEMENTS
We should like to thank Prof. R.C.Bray (University of
Sussex) for helpful discussions in the design of the freezequench
apparatus, and Dr. Yves Dupont of Biologic Co. for
implementing some of these designs. We are also grateful to
Drs. Allan Palmer and Jim Frye of Chemagnetics/Otsuka
Electronics Ltd. for help with the custom design of the lowtemperature
equipment for the NMR spectrometer, member!
of WSU technical services instrument shop and Fret
Schuetze of WSU electronics shop for numerous customiza
tions, and Dr. Dave Cleary for help with the ESR spec
troscopy. This work was supported in the early stages by th;
donors of the Petroleum Research Fund (American ChemicJ
Society) and National Institutes of Health grant RR O60Q|
and mainly by National Institutes of Health graf
GM43215. The WSU NMR Center is supported by NJ
grant RR 0631401, NSF grant CHE-9115282 and Battd
Pacific Northwest Laboratories Contract No.12- 097718-/
L2.

Vol. 14, No. 1-4 85
REFERENCES
11] P.N.Barlow,R.J.Appleyard, BJ.O.Wilson and
J.N.S .Evans, Biochemistry, 28, 7985 and p. 10093
[2]
[ 3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
U6]
[17]
B US]
[19]
:[20]
(1989).
J. N. S. Evans, "NMR and Enzymes", in "Pulsed
Magnetic Resonance, Optics and Imaging (Honoring
E. L. Hahn)", (Ed. D. Bagguley), Oxford University
Press, in press (1992).
J.N.S.Evans, CBurton, P.E.Fagerness,
N.E.Mackenzie & A.I.Scott, Biochemistry 25, 905
(1986); JP.G.Malthouse, Prog, in NMR Spectroscopy
18, pp 1-59 (1985); PHosch, Ibid. 18, 123 (1986);
K.Kanamori & J.D.Roberts, Acc.ChemJRes. 152, 35
(1983); W.W.Bachovchin, Biochemistry 25, 7751
(1986).
J.P.G.Malthouse, M.P.Gamcsik, A.S.F.Boyd,
KE.Mackenzie, and Ai.Scott, JAmer.Chem.Soc.
104, 6811 (1982); N.E.Mackenzie, J.P.G.Malthouse
and A.I.Scott, BiochemJ. 219, 437 (1984).
S.O.Smith, LPalings, V.Copie, D.P.Raleigh,
J.Courtin, J.A.Pardoen, Jlugtenburg, R.A.Mathies
and R.G.Griffin, Biochemistry 26, 1606 (1987).
J.N.S.Evans, RJ.Appleyard & W.A,Shuttleworth,
submitted (1992).
RXAppleyard and JJsf.S.Evans, submitted (1992).
J. S. Frye, Concepts Magn. Res. 1, 27 (1989).
C. A. Fyfe, "Solid State NMR for Chemists," p. 48,
CFC Press, Canada, 1983.
D. W. McCall, Ace. Chem. Res. 4, 223 (1971).
R. L. Void, J. S. Waugh, M. P. Klein and D. E.
Phelps, J. Chem. Phys. 48, 3831 (1968).
D. A. Torchia, /. Magn. Res. 30, 613 (1978).
E. 0. Stejskal, J. Schaefer, M. D. Sefcik and R. A.
McKay, Macromolecules 14,275 (1981).
W.A.Shuttleworth, CD.Hough, KP.Bertrand and
J.N.S .Evans, Protein Engineering 5, in press (1992).
H. Plattner and L. Bachmann, Int. Rev. Cytol. 79,
237 (1982).
C. A. Angell and Y. Choi, /. Microsc. 141, 251
(1986).
R. C. Bray, D. J. Lowe, C. Capeillere-Blandin and E.
M. Fielden, Biochem. Soc. Trans. 1, 1067 (1973).
D. P. Ballou and G. A. Palmer, Anal. Chem. 46,
1248 (1974).
A. M. P. Goncalves, Prog. Solid St. Chem. 13, 1
(1980).
A. Weiss, Angew. Chem. Internal. Edit. 11, 607
(1972).
E. R. Andrew, W. S. Hinshaw, M. G. Hutehins and
R. O. I. Sjoblom, Mol. Phys. 31, 1479 (1976a).
[22] D. E. Barnaal and I. J. Lowe, /. Chem. Phys. 48,
4614 (1968).
[23] M. Weithase, F. Noack and J. von Schiitz, Z. Phys.
246, 91 (1971).
[24] E. R. Andrew, W. S. Hinshaw, M. G. Hutehins and
R. O. I. Sjoblom, Mol. Phys. 34, 1695 (1977).
[25] R. E. Marsh and J. Donohue, Advanc. Protein Chem.
22, 235 (1967).
[26] P.-G. Jonsson and A. Kvick, Ada Cryst. B28, 1827
(1972).
[27] E. Mayer, /. Appl. Phys. 58, 663 (1985).
[28] P. T. Sargeant and R. Roy, Mater. Res. Bull. 333,
265 (1968).
[29] D. Tumbull, Contemp. Phys. 10, 473 (1969).
[30] N. H. Fletcher, Rep. Prog. Phys. 34, 913 (1971).
[31] D. Uhlmann, /. Non-Cryst. Solids 7, 337 (1972).
[32] P. Briiggeller and E. Mayer, Nature 288, 569 (1980).
[33] D. R. MacFarlane and C. A. Angell, J. Phys. Chem.
88, 759 (1984).
[34] A. Calvelo, "Development in Meat Science," p. 125,
Applied Science Publishers, London, 1981.
[35] H.C.Steinrucken and KAmrhein, EurJ.Biochem.
143, 351 (1984).
[36] K.S.Anderson et a!., Biochemistry 29, 1460 (1990).
[37] K.S.Anderson, J.A.Sikorski & K.A.Johnson,
Biochemistry 27, 7395 (1988).
[38] See for example, Proteins at Low Temperatures (Ed.
O. Fennema), Am. Chem. Soc. Adv. in Chem. Series
No. 180 (1979); RFranks, Biophysics and
Biochemistry at Low Temperatures, Cambridge
University Press (1985).
[39] L. Bachmann & E. Mayer, in Cryotechniques in
Biological Electron Microscopy (Ed. R.A. Steinbrecht
and K. Zierold), Springer-Verlag, Berlin p.3. (1987);
P.Douzou, Cryobiochemistry: An Introduction,
Academic Press (1977).
[40] WJ.Albery and JJl.Knowles, Biochemistry 15, 5631
(1976).
[41] D.P.Raleigh, M.H.Levitt & R.G.Griffin,
Chem.PhysJLett. 146, 71 (1988); D.P.Raleigh,
F.Creuzet, S.K.Das Gupta, M.H.Levitt &
R.G.Griffin, JAm.Chem.Soc. Ill, 4502 (1989);
T.Gullion & J.Schaefer, JMagnReson. 81, 1%
(1989); G.R.Marshall et aL.JAm.Chem.Soc. 112,
963 (1990); S. M. Holl, R. A. McKay, T. Gullion, J.
Schaefer, J. Magn. Reson. 89, 620 (1990); V.Copi6 et
al.. Biochemistry 29, 9176 (1990).

86
Abstract
Coupled methyl groups in dimethyl sulphide
M.R. Johnson, S. Clough, A.J. Horsewill and I.B.I. Tomsah
Bulletin of Magnetic Resonance
Department of Physics, University of Nottingham, Nottingham. NG7 2RD.
Low field methyl tunnelling spectra of dimethyl sulphide
have been measured using field cycling NMR and indicate
the existence of two distinct methyl groups with tunnel.
frequencies of 100kHz and 750kHz. Spectra show a significant
broadening of the Larmor peak at those fields at
which the Larmor and tunnel frequencies are equal. These
are resonances between the rotational and spin dynamics
of the methyl groups. An associated" change in the intensity
and position of the 100kHz sidebands suggests that
methyl group coupling may be responsible for these effects.
Narrow 100kHz sidebands are restored by irradiating
with an external RF field of this frequency which we
suggest is due to decoupling of the rotor pairs. Measurements
on partially deuterated dimethyl suphide show that
the coupling cannot be intra-molecular.
1 Introduction
The methyl group tunnel frequency wt depends on the
height of the potential barrier hindering rotation due to
the molecular environment. Three stationary states are
distributed equally in three potential wells and form an
energy singlet and doublet split by ftwt. Excitations consist
of a superposition of a pair of eigenfunctions to form
a partially localised wavepacket which rotates in a definite
sense at a frequency given by the energy splitting,
ie ±wt or 0. With the aid of external fields the rotation
frequency of a wavepacket can be changed by exerting a
rotational impulse her. The rotation frequency is then determined
by the difference between two of the three values
(2w where i, y and z are the spin states, a or /?, at ,,
proton sites which are defined by the hindering P° ten .*|
[1]. The protons themselves are each equally ditbufr
of

Vol. 14, No. 1-4
between the three sites satisfying proton indistinguishabilitj.
In this represenoation the matrix elements cf the
tunnelling interaction reflect the balance of clockwise and
anti-clockwise rotation of an undriven methyl group.
< xyz\HR\zxy >=< zxy\HR\xyz >= A (2)
A is the overlap integral between states localised in neighbouring
wells of the three-fold hindering potential.
Away from the anti-crossings the methyl group eigenstates
are linear combinations of the localised states and
are the delocalised representations of the Cz symmetry
group,
1 [\xyz> + exp(t2irn/3)\zxy (3)
where n = 0 for an A-state and n = ±1 for Ea and £&
states. Temporarily ignoring the dipole-dipole interaction,
these symmetrised states have energies
(4)
E{\, m) = u>Lm - 2A cos(2;rn/3) (5)
where OIL is the Larmor frequency and m is the spin component
of \xyz > in the direction of the magnetic field.
Mixtures of these eigenstates are partially localised, rotating
wavepackets which have energies resulting from the
differences between the rotational and magnetic energies
of the component states E(Xi,rm) — E(Xj ,mj). Magnetic
and rotational wavepacket energies add when the Larmor
precession and the rotation have the same sense and they
subtract when these senses are opposite.
The exact eigenstates and eigenvalues of the methyl
' group in the magnetic field are calculated numerically by
;:"'diagonalising H. The field dependence of the energy levels
Mis shown in figure 1. The level anti-crossings can be seen
^ . 1.2 and 2.4mT for the 100kHz methl group and 8.8 and
|17.6mT for the .750kHz rotor, where the A and E symmetry
states reflect off each other and exchange symmedue
to the finite non-secular parts of the intra-methyl
1 dipole-dipole interaction.
a energy levels diagram of figure 1 can be transformed
1 diagram of energy differences, or wavepacket ener-
|..as shown in figure 2 from which frequency spectra at
magnetic fields can be predicted by taking horsections.
In this way the level anti-crossings are
manifest themselves as resonant motional broad-
M the Larmor peak. Symmetry mixing at the
sings results in stationary wavepackets with en-
Pjroportional to u>£ evolving into rotating wavepackets
~" T gy Proportional to wfc (x to y in figure 2). This
he ability of the non-secular parts of the dipoleraction
for converting spin angular momentum
•tional angular momentum of the methyl group.
18
A(3/2)
^(-3/2)
A(-3/2)
B(mT)
Figure 1: Energy levels of dimethyl sulphide in a magnetic
field showing resonances between tunnelling motion and
spin dynamics as level anti-crossings at 1.2, 2.4 ,8.8 and
17.6mT
15-; '1:-.
10-
Reld(mT)
5 - • .?• .:•?" ,v*
0 S~=r
200 400 600 800
Frequency(kHz)
1000
Figure 2: Field/frequency plane of dimethyl sulphide
showing resonant broadening of the Larmor peak at the
level anti-crossings, notably at 9mT and 18mT
87

88
3 Low field, field cycling experiments
Measuring magnetisation at low field using a Faraday law
detector results in poor signal to noise. Thus in order to
measure a low field frequency spectrum either a different
type of detector [5] or a different technique such as field
cycling must be employed. Field cycling has been used
in this investigation. Each cycle of the experiment begins
with the preparation of a standard magnetisation by
destroying all magnetisation with a train of 90° pulses followed
by the preparation period of 20 seconds in a field
of 0.6T, The field is then switched rapidly in about 2 seconds
to the chosen magnetic field where the initial magnetisation
is destroyed by spin-lattice relaxation by an
amount proportional to the relaxation, rate T{~ and the
time at low field, 15 seconds. Anomalously rapid relaxation
can be stimulated by an external RF field of scanning
frequency / if this frequency matches the characteristic
wavepacket frequencies in figure 2 and can therefore
excite these wavepackets. The remnant magnetisation is
measured by a single 90° pulse after a rapid field switch
to 0.6T. The cycle is repeated increasing / each time, producing
a frequency spectrum which is a flat plateau in
which holes are drilled where f is equal to the frequencies
predicted in figure 2. These spectra are inverted to give
peaks for presentation.
A slightly more complicated experiment entails the application
of a second RF field of fixed frequency fsU at
low field , which is applied on the same coils as the scanning
field in an alternating sequence of short bursts of 0.5
seconds of each field. This is a stirring experiment which
may affect the intensity of spectral peaks by depopulating
energy levels and enhancing transitions which would otherwise
become suppressed as the populations of the initial
and final states become equal [6].
4 Sample preparation
Dimethyl sulphide was obtained from the Aldrich Chemical
Company and was used in an unsealed sample tube.
The deuterated sample was prepared in the Chemistry Department
by mixing aqueous solutions of CHzS~ No. and
CD2I as outlined in [7]. It was also used in an unsealed
sample tube.
5 Low field frequency spectra of
dimethyl sulphide
A typical low field spectrum of dimethyl sulphide, measured
at 3.5rnT, is shown in figure 3 and it is in excellent
agreement with the corresponding slice through figure 2,
indicated by the broken line. The Larmor peak occurs at
175kHz indicating a true magnetic field of 4.1mT and the
Am = 2 version of this peak occurs at 350kHz. These are
Bulletin of Magnetic Resonance
250 BOO 750 1000
Frequency [kHz]
1250
Figure 3: Frequency spectrum of dimethyl sulphide
mesaured at 3.5mT
labelled 1 and 2 respectively. They are each flanked by a
pair of 100kHz sidebands (labelled 1± and 2±) and also
by a pair of 750kHz sidebands (labelled 1± and 2±). Since
the Larmor frequency is much smaller than the larger tunnel
frequency, the reflected low frequency sidebands give
rise to a 750kHz spectrum which appears as a Am = 0
peak split into five peaks, each separated by the Larmor
frequency.
Frequency spectra of dimethyl sulphide have been measured
at many magnetic fields up to 35mT and a selection
of these are shown in figure 4. The spectra are aligned by
their Larmor peaks and only the frequency range incorporating
the 100kHz sidebands is covered. Away from the
level anti-crossings of the 750kHz rotor, that is below 7mT,
between 10 and 14mT, and above 20mT, the Larmor peak
and sidebands are well defined and reasonably symmetrical.
However in the vicinity of the level anti-crossings,
between 7 and lOmT and between 14 and 20mT, this discrete
structure is lost as the Larmor peak broadens, the
low frequency sideband virtually disappears and the high
frequency sideband is poorly defined. Where a frequency
separation of the Larmor peak and a sideband can be determined
it is apparent that rotation frequency of these
wavepackets is less 100kHz and is sometimes as small a?
70kHz. . 1
Although these effects occur at the level anti-crossing
of the 750kHz rotor, the field range over which they occ^
and the extent of the frequency broadening are too gre^
for them to be explained by dipolar broadening alone. Ful
thermore the the dramatic change in the sideband mtei
sity and frequency is not predicted by the dipole-dipole"
teraction. That it is the 100kHz spectrum which is so pr'
foundly affected by the level anti-crossings of the 750K
rotor suggests that these spectra may be evidence of sp

Vol. 14, No. 1-4 89
Z
o
m zo<
:
111
DC
Q
LU
h-
DC
LU
• \
:
v"V
• ".}'
• ' : I'.
I
I
• I
i '."
I
I
I
I V
coz-1OO co'z ©2+100
22.5
20
• 4'~.~*~ 15
• • !•
'.. ' i.
• r-^ '12.5
Si W9
A 5
• •"?• y '
E
Q
UJ
U_
o
LU
z
CD
<
FREQUENCY/kHz
4: Field dependence of the Larmor peak and the
coupling between the 100kHz and 750kHz rotors.
The 100kHz sidebands can be restored in a stirring experiment
with an external RF field with a stirring frequency
of 100kHz, as shown in the spectra in figure 5. It
appears that stirring at 100kHz drives rotation of methyl
wavepackets at this frequency and therefore prevents coupling
which has been seen to modify this frequency, as
shown in figure 4.
In order to probe the methyl group coupling a partially
deuterated sample of dimethyl sulphide, in which
one methyl group per molecule is fully deuterated, was
investigated. Figure 6 shows a low field spectrum of the
deuterated sample measured at 4mT which is very similar
to the spectrum of the fully protonated sample shown in
figure 3, one difference being the increase in the large tunnel
frequency from 750kHz to 780kHz. Field dependent
spectra of the deuterated sample covering the frequency
range which incorporates the Larmor peak and the 100kHz
sidebands are shown in figure 7 and they too are almost
identical to the spectra from the fully protonated sample
of figure 4. It therefore appears that the coupling of
protonated rotors persists in the deuterated sample, suggesting
that the packing of the molecules in the solid regarding
the relative positions of protonated and deuterated
methyl groups is random. Deuteration reduces the
number of coupled protonated rotors and slightly distorts
the packing of the molecules, decreasing the magnitude
of the hindering potential of the 750kHz rotor which becomes
780kHz rotor. These spectra indicate clearly that
the anti-crossing effects cannot arise from intra-molecular
methyl group coupling since each molecule has a protonated
and a deuterated methyl group. One new feature of
these spectra which may arise from coupling of protonated
and deuterated rotors is a small, sharp, field independent
peak at 200kHz which is indicated by a V in figure 7.
6 Discussion - Coupled rotating
wavepackets
Tunnelling of small molecular rotors like CH3, CH4 and
NHf is generally single particle in character. Few examples
exist,of coupled tunnelling of rotors (see [8] and references
therein) perhaps the most notable example being of
coupled methyl groups in lithium acetate [9]. In both of
these papers the coupling of methyl groups is mechanical,
being propagated by the modulation of the hindering potential
which has the three-fold symmetry of the individual
methyl groups. The coupled states of the system are a
direct product of the symmetrised eigenstates of each rotor.
Lithium acetate has weakly hindered methyl groups,
the tunnelling spectrum of the coupled system has been
observed using neutron scattering, and it is thought that
such a coupling is unlikely to be observable in strongly hindered
methyl groups, although the computational method
in [8] has been extended to consider such systems.
This kind of coupling is in stark contrast to that seen

90
G)z-100
"FREQUENCY/kHz
Figure 5: Stirred spectra of dimethyl sulphide showing '
restoration of the 100kHz sidebands
Bulletin of Magnetic Resonance
O 250 BOO 7S0 1000 1250
Frequency (kHz]
Figure 6: Frequency spectrum of partially deuterated
dimethyl sulphide mesaured at 4mT
in dimethyl sulphide. An analysis in terms of uncoupled
rotors was pursued in previous sections because first, the
coupling only occurs at those magnetic fields corresponding
to level anti-crossings and secondly, the form of the
spectra indicate that the coupled states are not simple
products, with three-fold symmetry, of the eigenstates of
the individual rotors. Furthermore, these methyl groups
are very strongly hindered in comparison with the methyl
groups in lithium acetate.
It is noted from figure 2 that at the level anti-crossings
there is a matching of the frequencies associated with
the rotational and magnetic evolution of the states of
the 100kHz and 750kHz rotors. If the stable states of
the methyl groups at these magnetic fields are rotating
wavepackets then the frequency of evolution of the spins
on each proton site is modulated by the rotation in a way
that depends upon the composition bf the wavepacket.
For example a coherent mixture of A and Ea states with
the same spin quantum number is a partially localised
wavepacket, which precesses clockwise at the tunnel frequency
and consequently modulates the longitudinal component
of the spins at this frequency. A similar mixed
symmetry state, but composed of states with spin quanne
turn numbers differing by unity modulates the transverse
component of the spin states at each proton site at a frequency
equal to the Larmor frequency plus or minus the
tunnel frequency, depending on whether the rotation and
Larmor precession have the same or opposite senses. Thus,
providing the proton sites of the two methyl groups arft
close enough together, the rotational motions may be
pled by the spin dynamics and angular momentum may
transferred between the groups. This results in an imb
ance of clockwise and anti-clockwise rotation, the unic l u

Vol. 14, No. 1-4 91
ide
O
<
CO
LU
<
Hi
DC
Q
LU
DC
LU
,v
J •
J
v/AV \
V
00
C0z-100
/ ; •
i
ll
V /
\ '
1/
v \
v 21
\ 20
v
\
19
18
16
u
10
8
h-
E
Q
_1
UJ
LL
g
FREQUENCY/kHz
7: Field dependence of the Larmor peak and the
z sidebands in partially deuterated dimethyl sul-
tunnel frequency being replaced by a pair of discrete rotation
frequencies. A reduced rotation frequency is clearly
observed in figure 4.
Taken as evidence not only of coupled methyl groups
but also for rotating wavepackets at 4K, these results are
very significant. The wavepackets do not have three-fold
symmetry which throughout the history of methyl dynamics
has wrongly been regarded as an essential prerequisite
for satisfying the indistinguishabilty of the methyl protons
(e.g. [10]). Each basis function \xyz > accomodates proton
indistinguishability since each proton has an equal amplitude
at each proton site and therefore any combinations
of these functions, ranging from delocalised states which
ressemble spin symmetry species to completely localised
functions, satisfy proton indistinguishability requirements.
7 Acknowledgements
The authors are grateful to the B.P. Venture Research
Unit for supporting this work. I.B.I.T. would like to express
his gratitude to the Sudanese government for his
fellowship. We would also like to thank Dr D.K. Knight
of the Chemistry Department for preparing the deuterated
sample.
References
[1] S. Clough, A.J. Horsewill, M.R.Johnson and
I.B.I.Tomsah (1992) submitted to Molec. Phys.
[2] P.J. McDonald, G.J. Barker, S. Clough, R.M. Green
and A.J. Horsewill (1986) Molec. Phys. 57,901
[3] S. Clough, A. Heidemann, A.J. Horsewill, A.J. Lewis
and M.N.J. Paley (1982)J. Phys. C 15,2495
[4] S. Clough, A.J. Horsewill, P.J. McDonald and F.O.
Zelaya (1985)Phys.Rev. Lett. 55,1794
[5] C. Connor, A. Chang and A. Pines (1986)Rev. Sci.
Instrum. 61,1059
[6] M.J. Barlow, S. Clough, P.A. Debenham and A.J.
Horsewill (1992)J. Phys. C 4,4165
[7] L.Pierce and M. Hayashi (1960)J. Chem. Phys. 35,479
[8] W. Hausler and A. Huller (1985) Z. Phys. B 59,177
[9] S. Clough, A. Heidemann, A.J. Horsewill and M.N.J.
Paley (1984) Z. Phys. B 55,1
[10] J.H. Freed (1965)J. Chem. Phys. 43,1710

92 Bulletin of Magnetic Resonance
INTRODUCTION
NMR RELAXATION STUDIES OF MICRODYNAMICS IN
CHLOROALUMINATE MELTS
Pamela A. Shaw, W. Robert Carper, Charles E. Keller
Room temperature molten salts consisting of mixtures of
A1C13 and l-ethyl-3-methylimidazolium chloride (MEIC1),
are of interest as aprotic solvents for studying a wide range
of both organic and inorganic compounds [1-7]. These
chloroaluminate molten salts possess considerable potential
as battery electrolytes and various types of electrochemical
agents [8-10].
The composition of a chloroaluminate melt has a
considerable effect on its physical properties. The
variations in physical properties of the melt are due to a
combination of factors including ion-ion interactions [4],
and Lewis acid-base properties. Chloroaluminate melts
with A1C13 present in excess (mole fraction, N, of A1C13 >
0.5) are termed acidic with A1C14' and A12C17" the
predominant anions.
The use of NMR relaxation methods provides useful
information about the dynamics and structure of various
chemical systems and chloroaluminate systems in
particular. In a previous work[ll], 13 C NMR relaxation
measurements were used to investigate the motion and
interactions of the MEI cation. The results indicate that
A1C14' in a Na + 0.22MEI + 0,7gAlCl4" melt forms a complex by
interacting with the C-2, C-4 and C-5 hydrogens on the
MEI + ring. This investigation was followed by studies
[12,13] in which the Dual Spin Probe method [14]
supported the existence of MEI(AlCl4)n

Vol. 14, No. 1-4 93
The quadrupole coupling constant, QCC, is given by:
QCC = [e 2 Qq/h] (4)
The DSP method has been applied to chloroaluminate
melts[12,13] and has provided evidence that the ring
hydrogens of MEI + interact with the tetrachloroaluminate
anion. The existence of these complexes has been
supported by linear plots of 13 C dipolar relaxation
rates(R, dd ) vs. quadrupolar 27 A1 relaxation rates(R,) that
pass through the origin as predicted by equation (5):
where a = [3TT 2 /10][(2I + 3)/I 2 (2I -
QCC = x-
= R,( 27 Al)/aX 2
(5)
+ (z 2 /3)], and
In this study, the DSP method is applied to melts
containing MEIC1, A1C13 and EtAlCl2. The inclusion of
EtAlCl2 provides a "baseline" as there is a covalent bond
between the ethyl group and aluminum in EtAlCl2. The
existence of covalent bonding(or complexation) between
quadrupolar and dipolar nuclei in a molecule results in a
linear plot of eqn. (5) that passes through the origin. In
the MEICl-EtAlCl2 melts reported herein, we observe a
linear plot of eqn (5) that passes through the origin when
applied to the terminal CH3 carbon in EtAlCl2 and one of
the peaks in the 27 A1 NMR of the melts.
EXPERIMENTAL
Materials
The l-ethyl-3-methylimidazolium chloride (MEIC1) and
. chloroaluminate molten salts were prepared as described
. previously [1]. Ethylaluminum dichloride (EtAlCy was
•^.obtained from Aldrich. All materials were stored under
ft^hydrous helium gas atmosphere in a dry box. All
molten salt preparations and manipulations were performed
• the dry box. Samples were loaded into 5 mm sample
s > capped in the dry box, removed, and sealed
Jiately with a torch.
Measurements
and 27 A1 NMR spectra were recorded on a Varian
300- spectrometer at 75.43 or 78.15 MHz.
measurements were calibrated against
or ethylene glycol and are accurate to within
§£Pulse widths(90°) were typically 8.6 (75.43 MHz)
1^(78.15 MHz) jts. Longitudinal relaxation times
ured by the the inversion-recovery method
(180°-r-90°-T) with T> 10T,. At least 12 delay times(r)
were used and the results fitted to a three parameter
exponential. NOE measurements were made using the
gated decoupler method[18]. It is likely that the error in
the NOE measurements is in the 5-10% range[18].
RESULTS AND DISCUSSION
The ability of both A1C13 and EtAlCl2 to form C2H
dimers[19,20] led us to examine the 27 A1 spectra of: (1)
neat EtAlCl2, (2) mixtures of MEICl-EtAlCl2 and (3)
ternary melts (N = AlCl3/MEICl/EtAlCl2)[21]. The neat
EtAlCl2 27 A1 NMR spectrum contains two peaks [21].
Peak 1 is a broad downfield peak that domi-nates the
spectrum. The second peak (upfield) overlaps peak 1 and
is only a fraction of peak 1 in total peak area. Peak 2
collapses into peak 1 as the temperature is lowered from
60 to 25°C. These two aluminum sites are consistent with
the extent of monomer-dimer formation in liquid
EtAlCl2[21].
The MEICI-EtAlCl2 (N = 0.5/0.5) melt 27 A1 NMR
spectrum also has two peaks. In this case, peak
1 (downfield) is very broad while peak 2 is very sharp, and
has a low peak area. Peak 2 increases slightly in area and
peak 1 broadens as the temperature is lowered from 70 to
0°C. We have previously[21] made the tentative assignments
of EtAlCl3- for peak l(downfield) and EtjAljClj" for
peak 2.
0.25
0.20
0.15
O
a.
O o.io
O CO
0.05
0.00
EtAia
10 20 30 40
27AI R1
Fig. 1. °C Dipolar Rl's vs "AI Rl's(25 to 70°C) for AI
peak 1 (127-131 ppm from A1(H2O)6 3+ ).

94
In this study, we first apply the DSP method to the CH3
carbon in EtAlCl2 and 27 A1 NMR peaks 1 and 2 from
several melt combinations and neat EtAlCl2. Fig. 1
contains the results for 27 A1 peak 1 (downfield) and Fig. 2
contains the results for 27 A1 peak 2. The fact that both
plots are linear and pass through the origin, indicate that:
(1) the DSP method is appropriate for these systems and
(2) the species associated with each peak contains EtAlCl2.
Furthermore, the slopes of these lines can be used to
cn
DIPOL
13C
. u
0 .20
0 .15
0 .10
0 .05
n nn
| EtAlClj /
r
I J .25/.40/.35 /
/
y
• I * /
t 1 A35/.40/.25
i ; /
i/
0 240 480 720 960 1200
27 Al R1
Fig. 2. "C Dipolar Rl's vs 27 A1 Rl's(25 to 70°C) for Al
peak 2 (102.5-103.0 ppm from Al(H2O)6 3+ ).
calculate the relative quadrupole coupling constants for the
EtAlCl2 -containing species in solution. The QCC values
obtained from Fig. l(Al peak 1) are 171, 119, 106 and 93
MHz for the (.5/.5), (.35/.40/.25), (.25/.40/.35) melts and
neat EtAlCl2, respectively. The QCC values obtained from
Fig. 2(A1 peak 2) are 6.9, 20, 11 and 93 MHz for the
(.5/.5), (.35/.40/.25), (.25/.40/.35) melts and neat
EtAlCl2(repeated).
Results of the Dual Spin Probe method (eqn. [5]) applied
to the (.5/.5), (.35/.40/.25) and (.25/.40/.35) melts
indicate interactions between the Al-containing species in
peak 2(102.5-103.0 ppm relative to A1(H2O)6 3+ ) and both
the NCH3 and ethyl terminal CH3 groups of MEI + . Fig.
3 contains the plots for the NCH3 group in each melt and
Fig. 4 contains data for the terminal CH3 on the MEI ethyl
group.
cc
0.50
0.40
o.oo
.5/.5
Bulletin of Magnetic Resonance
• / .35/.40/.
40/.25
0 64 128 192 255 320
27AI R1
Fig. 3. 13 C Dipolar Rl's vs. 27 A1 Rl's(25 - 70 C) for
NCH3 carbon vs Al peak 2(25 - 70°C).
0.55
0.44
T .5/.5
» •
I / .35/.40/.: 40/.25
64 128 192 256 320
27AI R1
Fig. 4. l3 C Dipolar Rl's for ethyl CH3 carbon vs 27 A1
Rl's(25 - 70°C) for Al peak 2.
The QCC's obtained from the slopes in Fig. 3(MEI
NCH3) are 1.7, 2.3 and 4.4 MHz for the (.5/.5),
(.35/.40/.25) and (.25/.40/.35) melts. The QCC's froffl|

Vol. 14, No. 1-4 95
Fig. 4(terminal CH3 on the MEI ethyl group) are 1.6, 6.9
and 1.3 MHz for the (.5/.5), (.35/.40/.25) and
(.25/.40/.35) melts.
Finally, there is no correlation between the ring hydrogen
dipolar Rl's and any of the 27 A1 peak Rl's. This result is
directly opposite to that found in MEIC1-A1C13 systems
[11,12].
CONCLUSIONS
Application of the DSP probe method to these mixed
melt systems indicates a lack of complexation between the
ring hydrogens of MEI + and any of these aluminum
containing species. These and previous results[21] suggest
that the formation of various charged dimers containing
EtAlCl2 takes precedence over the formation of complexes
between EtAlCl3" and the MEI + ring hydrogens. However,
there is evidence of interaction between the various Alcontaining
species and the CH3 groups(NCH3 and terminal
CH3 in the ethyl group) of MEI + in these melts.
ACKNOWLEDGMENTS
This work was partially supported by a National
Research Council and Summer Faculty Research
Associateship to W. R. C. and a summer Graduate
Fellowship to P. A. S.
REFERENCES
[1] J. S. Wilkes, J. A. Levisky, R. A. Wilson and C. L.
, Hussey, Inorg. Chem., 21 1263 (1982).
(2| J. S. Wilkes, J. S. Frye and G. F. Reynolds, Inorg.
.•Chem., 22(1983)3870.
jfI3| A. A. Fannin, L. A. King, J. A. Levisky and J. S.
'"Vilkes.J. Phys. Chem., 55(1984)2609.
| 4 1 A. A. Fannin, D. A. Floreani, L. A. King, J. S.
toders, B. J. Piersma, D. J. Stech, R. L. Vaughn, J. S.
jfllkes and J. L. Williams, J. Phys. Chem., 88 (1984)
514.
M. Dieter, C. J. Dymek, N. E. Heimer, J. W.
and J. s. Wilkes, J. Amer. Chem. Soc, 110
2722.
J - D ymek and J. J. P. Stewart, Inorg. Chem., 28
,1472.
[7] J. A. Boon, J. A. Levisky, J. L. Pflug and J. S.
Wilkes, J. Org. Chem., 51 (1986) 480.
[8] C. J. Dymek, J. L. Williams, D. J. Groeger and J. J.
Auborn, J. Electrochem. Soc, 131 (1989) 2887.
[9] C. J. Dymek and L. A. King, J. Electrochem. Soc.,
132 (1985) 1375.
[10] C. L. Hussey, T. B. Scheffler, J. S. Wilkes and A.
A. Fannin, J. Electrochem. Soc, 133 (1986) 1389.
[11] W. R. Carper, J. L. Pflug, A. M. Elias and J. S.
Wilkes, J. Phys. Chem. 96 (1992) 3828.
[12] W. R. Carper, J. L. Pflug and J. S. Wilkes,
Inorganica Chimica Acta 193 (1992) 201.
[13] W. R. Carper, J. L. Pflug and J. S. Wilkes,
Inorganica Chimica Acta (in press).
[14] J. J. Dechter and U. Henriksson, J. Magn. Res., 48
(1982) 503.
[15] A. Abragam, "Principles of Nuclear Magnetism",
Oxford University Press, Oxford (1961).
[16] K. F. Kuhlmann and D. M. Grant, J. Amer. Chem.
Soc, 90 (1968) 7355.
[17] B. Lindman and S. Forsen, in "NMR Basic Principles
and Progress," P. Diehl, E. Fluck and R. Kosfeld,
Editors, Vol. 12, p. 22, Springer-Verlag, New York
(1976).
[18] D. Neuhaus and M. Williamson, "The Nuclear
Overhauser Effect in Structural and Conformational
Analysis", VCH Publishers, New York (1989).
[19] J. Weidlein, J. Organomet. Chem., 27(1969)213.
[20] B. Gilbert, Y. Chauvin and I. Guibard, Vib.
Spectros., 1 (1991)299.
[21] W. R. Carper, C. E. Keller, P. A. Shaw, M. P. and
J. S. Wilkes, in "Eighth International Symposium on
Molten Salts", Electrochem. Soc, New York (in press).

96
Structure and Dynamics of a
Membrane Bound
Polypeptide
T. A. Cross, R.R.Ketchem, W. Hu, K.-C. Lee,
N.D.Lazo & C.L. North
Department of Chemistry &
Institute of Molecular Biophysics
Florida State University, Tallahassee, FL 32306-3006
INTRODUCTION:
Orientational constraints can be used to build-up
three dimensional structures of biological macromolecules
in much the same way that distance constraints are used
today. In an anisotropic environment where molecular
motions do not average NMR signals to their isotropic
average the resonant frequencies become orientation
dependent. Consequently, chemical shift frequencies, dipolar
couplings, and quadrupolar interactions are all dependent on
the specific orientation of the interaction tensor with respect
to the magnetic field. In unoriented samples this results in
broad spectral lineshapes, which arc useful in their own
right, but if the samples are aligned so that all molecules
have the same orientation with respect to the magnetic field,
then sharp line spectra can be obtained that reflect the
orientation dependence of the spin interactions.
to interpret these frequencies for orientational
constraints two pieces of information are critical. The first
is a refined knowledge of the static tensor element
magnitudes and orientation with respect to the molecular
frame. The tensor element magnitudes represent the unique
frequencies of the orthogonal axes of the interaction
ellipsoid. In other words, the magnitudes represent the
resonant frequency when the individual axes are aligned with
the magnetic field. These values can, in many cases, be
readily obtained from spectra of unoriented samples. The
second critical characterization is a knowledge of the
molecular motions that result in averaging both the tensor
element magnitudes and orientation. Consequently, a
detailed picture of dynamics is needed to characterize the axis
about which motions are occurring, the amplitude of the
motions and whether the motions are continuous or
discontinuous, such as a flip of an aromatic ring. In fact, it
Bulletin of Magnetic Resonance
is this separation of dynamics and structure that represents
the most difficult challenge for the biological solid state
NMR spectroscopist. By using low temperature (120 K)
experiments, torsional motions are essentially eliminated
except in specific cases such as methyl group three site
jumps. Once both the static tensors and motional averaging
of the tensors are characterized it is possible to achieve very
high resolution quantitative constraints. Here we present
both an overview of the structural constraints and the
dynamic characterizations achieved for the channel forming
polypeptide, gramicidin A in a fully hydrated lipid bilayer.
Gramicidin A is a hydrophobic polypeptide of 15
amino acid residues with both end groups blocked so that
there are no formal charges. As a dimer it forms a helical
channel with a 4A pore that accommodates a single file of
water molecules and cations that are stripped of all but two
waters in the primary hydration sphere. The peptide linkages
that line the channel are thought to help solvate the cations
during transport by rotating the carbonyl oxygens toward
the channel axis. While crystal structures of gramicidin A
have been achieved in organic solvents (Langs, 1988;
Wallace & Ravikumar, 1988; Langs et al., 1991) no crystal
structure of the channel conformation has been achieved.
However, much is known about the channel conformation.
The backbone folding motif is a (5-sheet type of structure
that has been wound into a helix (Urry, 1971). This is
possible because the amino acids of gramicidin alternate in
stereochemistry between D and L configurations. The helix
sense has been determined from the orientational constraints
of the 15 N amide sites (Nicholson & Cross, 1989).
Recently, the first backbone torsion angles for the channel
conformation have been determined from orientational
constraints (Teng et al., 1991). Complementing this work

Vol. 14, No. 1-4 97
f
13 C Chemical Shift
15 N Chemical Shift
15 N- 13 C Dipolar
^NMH Dipolar
14 N- 13 C Dipolar
15 N- 15 N Dipolar
13 C- 13 C Dipolar
2 H Quadrupolar
Figure 1: A model of the gramicidin channel dimer showing the interaction tensors that have been studied. The model
structure is that of Lomize et al., 1992. Each tensor is represented by a orthogonal set of three unit vectors that have
been oriented correctly for the specific site of interest. All backbone and tryptophan indole I5 N chemical shift and I5 N-
! H dipolar tensors have been studied with the exception of the ethanolamine blocking group. About half of the 15 N-
13 Cj dipolar interactions in the backbone have been studied, four of the backbone carbonyl l ^C chemical shift tensors
and in so doing the 14 N electric field gradient tensor has been characterized. Four sidechains have been deuterated and the
quadrupole splittings obtained. 15 N spin diffusion has been observed between selectively labeled sites. To achieve this
data approximately 50 separate syntheses of gramicidin A have been performed (Fields et al., 1989).

98
are solution NMR studies of gramicidin in SDS micelles
that have also shown the same backbone folding motif
(Arseniev & Barsukov, 1986; Lomize et al., 1992).
RESULTS & DISCUSSION:
Fig. 1 illustrates the large number of nuclear spin
interactions that have been studied in gramicidin. For each
interaction the tensor represented by an orthogonal set of
unit vectors is placed on each atomic site that has been
studied. More than 40 different isotopically labeled (Fields
et al., 1989) gramicidins have been synthesized and more
than 100 different structural constraints have been
developed. The structural task is one of determining the
torsion angles. If it is assumed that the peptide linkages are
planar then there are two backbone torsion angles for each
amino acid residue, one for valine sidechains and two each
for leucine and tryptophan sidechains. There are two
additional torsion angles in the ethanolamine blocking
Val
Ala
Leu,
End View
Side View
Figure 2: A partial structure for the gramicidin channel
dimer, experimentally derived. The N-terminal four peptide
planes have been oriented with respect to the bilayer normal
through a combination of ^N-'H, 15 N- 13 Ci dipolar and
15 N chemical shift interactions in uniformly aligned bilayer
samples. The experimentally verified symmetry of the two
monomers has permitted the docking of these two partial
structures to form a turn of the helix that supports previous
models with 6.3 residues per turn of the channel helix. The
amino acid residues have been identified for one of the
monomers and the amide protons and oxygens labeled.
Bulletin of Magnetic Resonance
group of the carboxy terminus. Therefore, the structural
problem comes down to finding the solution for 52 torsion
angles.
Fig. 2 shows the N-terminal / N-terminal
backbone junction for the channel as determined by
orientational constraints. For each peptide linkage plane the
* S N - !Hand 15 N - 13 Ci dipolar interactions, which have
their unique static tensor elements directed along the
internuclear vectors, were obtained. These two vectors define
the orientation of the plane with some ambiguity that is
minimized by an interpretation of the 15 N chemical shift for
each plane. The orientation of the chemical shift tensor
elements with respect to the molecular frame has been
determined for each of these i 5 N sites (Teng & Cross, 1989;
W. Mai, W. Hu, C. Wang and T.A. Cross- unpublished
results).
This structure clearly shows the right-handed
helical sense and the ^-helical class of torsion angles, in
that the orientation of adjacent peptide planes alternates
between parallel and antiparallel with respect to the channel
axis. Furthermore, like many of the computationally refined
structures (e.g. Roux and Karplus, 1988; Chiu et al., 1991)
and unlike the original structural model (Urry, 1971) many
of the carbonyl oxygens are rotated in toward the channel
axis. The prime exception to this observation is the formyl
oxygen of the amino terminal blocking group. Here we
suspect that this is an artifact of the assumption that this
peptide linkage is planar. Preliminary evidence suggests that
the © torsion angle for this plane is far from 180°, the
value for normal trans-planar linkages.
Fig. 3 shows that we have also studied the
sidechains of gramicidin A. In (A) the chemical shift powder
pattern of a dry powder sample of 15 N indole labeled
tryptophan is shown. The chemical shift tensor elements
agree fairly well with those from the dry powder sample of
gramicidin shown in (B). However the tensor elements for
the fast frozen hydrated lipid bilayer preparation of
gramicidin is very different. Both o"22 and a33 tensor
elements differ by 10 ppm from the dry samples. This
sample was frozen in liquid propane to avoid distortions in
the lipid bilayer which occur when a bilayer preparation is
slowly cooled through its phase transition temperature. The
sample was then transferred to liquid nitrogen and from there
into our low temperature NMR probe. It is unlikely that
the former two samples have the nitrogen bound hydrogen
involved in a hydrogen bond. However, there is some
electrophysiological evidence that these hydrogens for each
of the tryptophan rings in the gramicidin channel state are
hydrogen bonded to the carbonyl oxygens of the ester
linked lipids. Consequently, it may be that the substantial
changes seen in the chemical shift tensor element
magnitudes reflect hydrogen bonding of the indole group.
Further indirect evidence, given below, for hydrogen
bonding comes from the orientation of the indole N-H with

Vol. 14, No. 1-4
respect to the bilayer surface.
200 100
ppm
Figure 3: Powder pattern spectra of 1 5 N labeled indole. The
spectra were obtained at 40 MHz for 15 Nona spectrometer
that has been home built around a Ghemagnetics data
acquisition system and an Oxford 400/89 superconducting
magnet. A] Dry powder sample of the amino acid,
tryptophan obtained at room temperature. Spectral
simulation yields tensor elements: (?i i = 35, o~22 = 104, and
C33 = 158 ppm. B] Dry powder of 15 N-Trp9 gramicidin A:
On = 36, 022 = 106, and a33 = 161 ppm. C] Fast frozen
sample of 1S N-Trp9 gramicidin A in fully hydrated lipid
bilayers: a^ = 36, G22 = 116, and a33 = 171 ppm.
The 2 H quadrupolar spectrum (Fig. 4) of d5-Trpj r
gA provides an example of the spectroscopic data that has
been used for structural constraints. The arrows point to a
linewidth of 1.4 kHz that represents an uncertainty for the
0
orientation of ±0.2°. Not only does this spectrum describe a
single conformation for this tryptophan ring, but it also
dictates that the orientation for all of the gramicidin
molecules in the sample is remarkably uniform (Moll &
Cross, 1990). For this sample there are five deuterons on
the tryptophan ring and five quadrupole splittings are
observed. The analysis of this data combined with the 15 N
chemical shift and 15 N-iH dipolar interaction for the indole
nitrogen (W. Hu, K.-C. Lee and T.A. Cross - unpublished
results) yields two possible orientations for this ring. Each
of these structures has the same orientation with respect to
the channel axis and the orientation is similar to that shown
in Fig. 1 for Trpn, where the N-H points towards the
bilayer surface. In recent computational (Meulendijks et al.,
1989) and electrophysiological (O'Connell et al., 1990)
studies it has been suggested that the indoles of gramicidin
A are hydrogen bonded to the carbonyl oxygens of the esterlinked
lipids. These results indicate that such hydrogen
bonding may be present and this may be one of the prime
reasons why this conformation is present in lipid bilayers
rather than the double-helical structures that dominate
organic solutions (Veatch and Blout, 1974; Zhang et al.,
1992). Furthermore, the tryptophan dipole moment is
oriented primarily along the channel axis, rather than radial
to this axis. This has significant implications for cation
transport by gramicidin A.
Figure 4: 2 H NMR spectrum of ds-Trpj ] gramicidin A in an
oriented lipid bilayer preparation. The arrows indicate a
resonance linewidth of 1.4 kHz. The assignment of two of
these quadrupole splittings is clear from a knowledge of
other spin interactions in this ring system: £3 = 19 2; and
T|2 = 9 9 kHz.
Detailed dynamics have come from a combination
of lineshape simulation and relaxation studies. Below the
gel to liquid crystalline phase transition temperature of the
bilayers all large amplitude motions cease. Both the global
motion about the bilayer normal and the local backbone
-80
99

100
motions become slower than the chemical shift frequency
scale (Nicholson et al., 1989; 1991). This is because the
backbone motions are dependent upon motion of the C a-Cp
axis and hence, if the sidechain conformations are frozen in
the gel phase the backbone motions will be greatly
impeded. When oriented samples are lowered through the
phase transition temperature the broadened resonance
represents a considerable orientational dispersion. Instead of
the backbone having a single conformation in this static
environment a range of conformational substates has been
trapped (Frauenfelder et al., 1988; Nicholson et al., 1989).
This set of substates represents the range of orientations
over which local motions occur above the phase transition
temperature. From the observed lineshape it has been
possible to determine the orientation of the axis about
which the local backbone motions occur with respect to the
magnetic field (Nicholson et al., 1991). This axis has been
shown to be coincident with the Ca-Ca axis for each
peptide linkage. Since the number of conformational
substates is greater than three, the local motion can be
modeled as a diffusion within a gaussian well. Furthermore,
the amplitude of the motion could be estimated.
Once such a motional model has been
experimentally developed for a specific backbone site in the
gramicidin channel conformation it is possible to interpret
relaxation data for these same i*N sites. Tj relaxation times
were obtained from oriented samples of fully hydrated lipid
bilayers. To fix the frequency of the local motions it has
been necessary to measure the relaxation times at two
different field strengths. In Fig. 5 an analysis of such data is
shown. The global correlation time (tp) and the local
motion correlation time (tj) are variables as well as the
amplitude of the local motion. However, we have a
previous estimate of the motional amplitude (±15°,
Nicholson et al., 1991) as well as the global correlation
time (200 ns, Seelig and Macdonald, 1987). Consequently,
it is possible to achieve a unique solution for the local
motion correlation time (10 ns). A similar correlation time
has been reported for an even numbered site (Leu4) in the
gramicidin channel (North and Cross, 1992)
This is a remarkably slow correlation time for a
motion of a molecular group with nominally such a small
molecular weight. Roux and Karplus (1988) have analyzed
the normal modes in the gramicidin channel and concluded
that fluctuations occur with frequencies of 4.6 to 20 cm'
corresponding to harmonic oscillator periods of 2 to 8 ps,
approximately, a factor of lO^-lO 4 slower than the
determination reported here. Similarly, estimates from
molecular dynamics indicate frequencies for these local
motions that are in a similar range to those of the normal
mode analysis.
The experimental evidence presented here is not
A.
-7.0
Log %. -9.0
B.
Log
c.
-11.0
-7.U
T i " 9 -°
-11.0
-7.0
LogTj -9.0
-11.0
Bulletin of Magnetic Resonance
i!! 1
j||
iii
•>t
ii!
•IJJX,
-7.0 -6.0
.f |||1|lW
iii
HI
Hi
Hi i
IP' i
lit} I
a
° 4.7 T
• 9.4 T
' lll|ltI| ! ll "3J l lJJ;;
° 4.7 T
• 9.4 T
• 1 ' • • r— 1 1 1 , . 1
-7.0 -6.0
iii*
ill
f iii
111
Hi
ill
-7.0 -6.0
i
° 4.7 T
• 9.4 T
Figure 5: T, relaxation times that were obtained from 15 N-
Gly2 gramicidin A in oriented lipid bilayers at 20 and 40
MHz have been analyzed to achieve a solution for local and
global correlation times as well as the local motional
amplitude. A] calculated for 12° rms deviation; B] 15°; and
C] 18°. Because of previous estimates for the amplitude and
global correlation time, it is possible to determine the local
correlation time as 10 ns with an rms amplitude of 15°.
unique for slow motions in polypeptide backbones, in fact
most of the experimental evidence suggests much slower
motions than the computational techniques. Analysis of Tj
and NOEs from the backbone of the filamentous virus fd
suggested correlation times of 1 ns (Cross and Opella,
1982). A similar timescale has been reported for collagen

Vol. 14, No. 1-4 101
(Sarkar et al., 1985). More recently, Cole and Torchia
(1991) have reported local motional frequencies in the
backbone of crystalline staphyloccal nuclease in the ns or
near-ns timescale.
For the gramicidin channel there appear to be two
primary reasons for the discrepancy between the
computational and experimental frequencies. First, it maybe
as suggested by Venkatachalam and Urry (1984) that the
local motions are correlated along the polypeptide backbone.
Computational studies have argued that the extent of such
correlations are limited to nearest neighbors. The second
reason is that the lipid environment may damp the backbone
motions severely. Evidence has already been presented that
the lipid environment can reduce the motional frequencies
below the kHz range when the lipids are in the gel phase.
Above the phase transition temperature it is well known
that the lipid environment damps global motional
frequencies. For instance, the 200 ns global correlation time
for gramicidin (1880 daltons) would be consistent with a
protein of 50,000 daltons or greater in aqueous solution.
Furthermore, evidence of specific gramicidin - lipid
interactions described here suggests an additional mechanism
for damping by the lipid environment.
The transit time for cations to move between
dipeptide carbonyl sites in the polypeptide backbone can be
estimated from kinetic measurements (Anderson, 1983) and
analyses of the energetics (Roux and Karplus, 1991) to be
in the range of 10 ns. Consequently, it is now likely that
there is a correlation between kinetics and the local
dynamics of the polypeptide backbone. This unique
correlation between structure, dynamics and function
emphasizes the importance of pursuing such detailed studies
of polypeptides and proteins in functional native-like
environments.
ACKNOWLEDGEMENTS:
We are indebted to the staff of the FSU NMR
facility, Joseph Vaughn, Richard Rosanske, and Thomas
Gedris for their skillful maintenance, modification and
service of the NMR spectrometers. This effort has been
supported in large part through grant #AI-23007 from the
National Institutes of Health. TAC also gratefully
acknowledges support of the Alfred P. Sloan Foundation for
a Research Fellowship.
REFERENCES:
Anderson, O. S. (1983) Biophys. J. 41:119.
Arseniev, A.S. and Barsukov, V.F. (1986) in Chemistry of
Peptides and Proteins, Vol. 3; W. Voelter.E. Bayer,
Y.A. Ovchinnikov, V.T. Ivanov, Eds. Walter de Gryter &
Co. Berlin, pgs. 127-158.
Chiu, S.W., Nicholson, L.K., Brenneman, M.T., Teng, Q.,
Subramaniam, S., McCammon, J.A., Cross, T.A.,
Jakobsson, E. (1991) Biophys. J. 60:974-978.
Cole, H.B.R. and Torchia, D.A. (1991) Chem. Phys.
158:271.
Cross, T.A. and Opella, S.J. (1982) J. Mol. Biol. 159:543-
549.
Fields, C.G., Fields, G.B., Noble, R.L. and Cross, T.A.
(1989) Int. J. Peptide Protein Res. 33:298-303.
Frauenfelder, H., Parak, F. and Young, R.D. (1988) Annu.
Rev. Biophys. Biophys. Chem. 17:451-479.
Langs, D.A. (1988) Science 241:188-191.
Langs, D.A., Smith, G.D., Courseille, C, Precigoux, G.
and Hospital, M. (1991) Proc. Natl. Acad. Sci. USA
88:5345-5349.
Lomize, A.L., Orechov, V. Yu. and Arseniev, A.S. (1992)
Bioorgan. Khimia 18:182-200.
Meulendijks, G.H.W.M., Sonderkamp, T, Dubois, J.E.,
Nielen, R.J., Kremers, J.A. and Buck, H.M. (1989)
Biochim. Biophys. Acta 979:321-330.
Moll, F. Ill, and Cross, T.A. (1990) Biophys. J. 57:351-
362.
Nicholson, L.K. and Cross, T.A. (1989) Biochemistry
28:9379-9385.
Nicholson, L.K. LoGrasso, P.V. and Cross, T.A. (1989) J.
Am. Chem. Soc. 111:400-401.
Nicholson, L.K., Teng, Q. and Cross, T.A. (1991) J. Mol.
Biol. 218:621-637.
North, C.,L. and Cross, T.A. (1992) J. Magn. Res. - in
press.
O'Connell, A. M., Koeppe, R. E. II and Anderson, O. S.
(1990) Science 250:1256-1258.
Roux, B. and Karplus, M. (1988) Biophys. J. 53:297-309.
Roux, B. and Karplus, M. (1991) J. Phys. Chem. 95:4856.
Sarkar, S.K., Sullivan, C.E. and Torchia, D.A. (1985)
Biochemistry 24:2348-2354.
Seelig, J. and Macdonald, P.M. (1987) Ace. Chem. Res.
20:221-228.
Teng, Q. and Cross, T.A. (1989) J. Magn. Res. 85:439-
447.
Teng, Q., Nicholson, L.K. and Cross, T.A. (1991) J. Mol.
Biol. 218:607-619.
Urry, D.W. (1971) Proc. Natl. Acad. Sci. USA 68:672-676.
Veatch, W.R. and Blout, E.R. (1974) Biochemistry
13:5257-5264.
Venkatachalam, CM. and Urry, D.W. (1984) J. Comp.
Chem. 5:64.
Wallace, B.A. and Ravikumar, K. (1988) Science 241:182-
187.
Zhang, Z., Pascal, S. and Cross, T.A. (1992) Biochemistry
- in press.

102 Bulletin of Magnetic Resonance
The Role of Metal Ions in
Processes of Conformational
Selection during Ligand-
Macromolecule Interactions
E.Gaggelli, N.Gaggelli, G.Valensin
Department of Chemistry, University of Siena
Pian dei Mantellini 44
Siena 53100, Italy
and
A.Maccotta
Department of Chemistry, University of Basilicata
Via N.Sauro 85
Potenza 85100, Italy
1 Introduction
The interaction with macromolecules
plays a major role in
eliciting the biochemical activity
of relatively small flexible
ligands. Several processes are
involved in such interaction
such that the key and hole
assumption may often fail. One
of the processes not carefully
considered so far is that of
conformational selection, in
which the macromolecule
stabilizes the conformation of
the ligand at the bound site
after a selection among several
conformational arrangements.
The conformation at the bound
state may or may be not connected
to some low-energy
conformation or to the conformation
stabilized at the solid
state. The comprehension of
this process is expected to
provide a valuable aid in
rationale drug design where
synthesized molecules are
sought yielding the same or
even more specific responses
than the natural ligands.
Investigation of this pro-

Vol. 14, No. 1-4 103
cess requires to delineate the
change in conformation when
going from the free to the
bound state and, from this
point of view, NMR is the
technique of choice.
Here we present evidence
that NMR allows to
detect and delineate the
change in conformation experienced
by a flexible ligand,
the dipeptide carnosine (p -
alanyl-L-histidine), when it
binds to the serum protein
albumin. We show also that
the process of conformational
selection is favoured by the
presence of divalent metal
ions, such as Ca(II) and Cu(II),
that stabilize, in the metal
complex, a conformation of the
ligand very close to that assumed
in the bound state.
2 NMR Parameters
The preferred conformation
assumed by the ligand in its
free state in solution can be
easily delineated by measuring
dipolar interaction energies
between pairs of homo- or
hetero-nuclear spin. Since such
interaction terms are functions
of tc/r 6 , distances can be calculated
if the reorientational
dynamics can be characterized,
even in some approximate
way. This last purpose can be
accomplished by measuring
and interpreting the 13 C-NMR
spin-lattice relaxation rates,
that are, in general, determined
by the one bond (r = 1.09
A [1]) 13C-1H dipole-dipole interaction.
Once the motional correlation
time(s) is (are) determined,
relevant geometric
features can be obtained by
one or more of the following
experiments:
a) evaluation of the iH-pH}
n.O.e. if spectral resolution
is not limited and crosscorrelation
effects can be
neglected;
b) measurement of singleand
double-selective l H-
NMR spin-lattice relaxation
rates of selected proton
pairs; these provide a
means of calculating absolute
values of pairwise
dipolar cross-relaxation
terms, Ojj [2-4];
c) measurement of the * 3 C -
pH} n.O.e. upon selective
presaturation of resolved
proton resonances [5];
d) evaluation of relative
values of cross-relaxation
terms from intensities of
cross peaks in 2D NOESY
maps [6].
All these methods are
very efficient in providing the
desired information on the
preferred conformation in
solution of any 'NMR visible'
ligand and, eventually, the
change in conformation caused
by the presence of metal ions.
In this last case, if the metal is
paramagnetic, a great piece of
structural information is
gained by investigating the
paramagnetic effects on nu-

11 ;
104 Bulletin of Magnetic Resonance
clear relaxation rates and chemical
shifts [7,8] or on 2D
spectra [9,10].
In order to investigate
the eventual change in the
conformation of the ligand
when it binds to a macromolecule
either in the presence or
in the absence of metal ions,
the previously outlined NMR
methods, at least not all of
them, are not as efficient as in
the free state. The concentration
of the macromolecule is a
limiting factor, since it must be
kept quite small if spectral
distortion is to be avoided. As
a consequence exchange of the
ligand between the free (bulk)
and the bound state must be
taken into consideration, yielding:
Pobs = xfPf + xbPb
where P is any observed NMR
parameter, f and b refer to the
free and bound states and the
x's are molar fractions. It
follows that the change in P
from the free to the bound
state is expressed by:
AP = xbPb
where Xb is usually of the
order 0.01-0.1. The consequence
is that spin-lattice
relaxation rates and chemical
shifts are no longer suitable
parameters for delineation of
geometric features of the
bound state, unless a paramagnetic
centre, either intrinsic
or extrinsic, is present. One is
then left with measurements
of ID or 2D transferred n.O.e.
or, which we prefer, of singleand
double-selective proton
spin-lattice relaxation rates. In
fact, in absence of spectral
distortions, the same measurements
can be easily accomplished
for the free ligand as
well as in the system where
the ligand is exchanging
between the two states, and
absolute values of the crossrelaxation
rate can be separately
obtained for the free and
the bound ligand.
3 Free Carnosine
The relevant features of
carnosine in water solution can
be summarized as follows:
a) reorientational dynamics
can be interpreted in
terms of a principal correlation
time (Tc = 58 ps)
describing reorientation
around a molecular axis
passing through the imidazole
ring, coupled with
segmental motion of the
amino-terminal moiety
and librational motion of
the ring (Tg = 10 ps);
b) predominance of the g~
rotamer around the C6-C7
bond (CH2-CH segment of
the histidyl residue);
c) folding of the fi-alanyl
moiety towards the imidazole
ring.

Vol. 14, No. 1-4 105
4 Calcium Complex
Calcium forms two complex
speies with carnosine in solution:
a 1:1 complex where the
carbonyl and carboxyl oxygens
are the metal binding atoms
and a dimeric complex where
the two carbonyl and one carboxyl
oxygens and the imidazole
nitrogen are the four
coordinated atoms. The overall
dissociaton constant of the
complexes is Kd = 0.04 mol
dm"3. In the monomer complex
the dipeptide retains the
conformation detected by NMR
as the 'preferred' one in the
free state in solution. In the
dimer species extensive intermolecular
interactions are
favoured and the conformation
of the dipeptide is less folded
than in the free state or in the
1:1 complex.
It is concluded that calcium
ions stabilize a particular
geometric arrangement that is
itself representing the 'preferred'
conformation in solution,
as it raises from motional averaging
of all the particular
conformations assumed by the
flexible peptide.
5 Interaction with HSA
The interaction of carnosine
with human serum albumin
(HSA) can be detected and
delineated by measuring
selective and double-selective
proton spin-lattice relaxation
rates of carnosine protons in
the presence of low molar
fractions of the protein [11,12].
Detection of binding is allowed
by appreciable relaxation rate
enhancements of selective
relaxation rates of the imidazole
protons, as well as of the
His Ha:
AR sel = pbRb sel
Even at very low fractions of
bound carnosine the selective
relaxation rate at the bound
site is so fast that the observed
rate undergoes enhancements
as high as 50-100 %. The effect
allows also a titration of the
binding process, yielding an
apparent dissociation constant
of 2.5xlO- 4 mol dm- 3 .
The observed enhancements
are consistent not only
with a very tight binding of
the whole peptide molecule to
the protein but also with
occurrence of dipolar interaction
between ligand* and protein
protons, although there is
no possibility of obtaining
quantitative estimations of
such interactions.
More information can be
obtained by measuring the
double-selective proton spinlattice
relaxation rates within
the His Ha-Hpi-H(32 moiety. As
in the free state in solution,
such measurements yield the
dipolar cross-relaxation rate
between the excited protons,
e.g.:

106
a,Bl sel
aa.pl = R« - Ra
where the first term on the
right hand defines the doubleselective
spin-lattice relaxation
rate measured on Ha when
both Ha and Hpi are excited.
The cross-relaxation rate
in the bound ligand is obtained
by:
Gobs - Gf
Ob =
As a consequence of binding,
the cross-relaxation rate changes
from positive to negative
values and allows to gain geometric
information on the
bound molecule.
In fact, substitution of
the motional correlation time
of the protein in the equation:
ob = -0.1
provides a means of evaluating
proton-proton distances in the
bound peptide. It comes out
that at least the investigated
moiety retains the conformation
that was shown to be
stabilized by calcium ions with
exclusive occurrence of the g~
rotamer.
6 Effect of Calcium
The same experiments used to
detect and delineate binding of
carnosine to HSA can be repea-
h
Bulletin of Magnetic Resonance
ted in the presence of Ca(II) at
equimolar ratios with the
ligand. No substantial change is
observed, as far as the selective
relaxation rate enhancement
and the change in the
double-selective relaxation
rates are concerned. An
appreciable change can be
however observed in the
ligand-protein dissociation
constant that is now measured
at 2.0xl0- 5 mol dm" 3 .
It is therefore possible to
conclude that the effect of the
metal ion is to stabilize the
conformation that occurs at the
bound site. In absence of the
ion, such conformation has to
be selected among all the
several conformations that are
possibly assumed by the
flexible ligand in solution. This
process leads to an appreciable
reduction of the binding constant.
7 Effect of Copper
It is important to underline
that the same experiments
cannot be carried on when
using a paramagnetic ion such
as copper. It is still possible to
delineate the geometry of the
metal complex, but, only of
that having the maximum
number of ligands in the
coordination sphere. One is in
fact forced to work at very low
[metal]/[ligand] ratios.
By the same way, it is
possible to detect and delineate
the ternary complexes

Vol. 14, No. 1-4 107
formed in the presence of the
protein but there is no way of
shedding light on the geometrical
and conformational features
of the ligand bound to
the macromolecule.
References
[1] Dill,K. and Allerhand,A.
J.Am.Chem.Soc.lOl, 4376
(1979).
[2] Hall,L.D. and Hill, H.D.W.
J.Am.Chem.Soc. 98, 1269
(1976).
[3] Gaggelli, E., Kushnir, T.,
Navon, G. and Valensin, G.
Magn.Reson.Chem., in the
press.
[4] Marchettini, N. and
Valensin, G. J.Phys.Chem.
94, 4508 (1990)
[5] Niccolai, N., Rossi, C,
Mascagni, P., Neri, P. and
Gibbons, W.A. Biochem.
Biophys. Res. Commun.
124, 739 (1984).
[6] Jeener, J., Meier, B.H.,
Bachmann, P. and Ernst,
R.R. J.Chem.Phys. 71,
4546 (1979).
[7] Niccolai, N., Tiezzi, E. and
Valensin,G. Chem.Rev. 82,
359 (1982)
[8] Bertini, I. and Luchinat, C.
"NMR of paramagnetic
species in biological
systems", Benjamin Cummings,
Menlo Park, 1986.
[9] Gaggelli, E., Tiezzi, E. and
Valensin, G. J.Chem.Soc,
Faraday Trans. II 84,
141 (1988).
[10] Gaggelli, E., Gaggelli, N.,
Maccotta,A. and Valensin,
G. Inorg.Chem., submitted.
[11] Valensin, G., Valensin, P.E.
and Gaggelli, E. in "NMR
spectroscopy in drug research"
(Jaroszewski, J.W.,
Schaumburg,K. and Kofod,
H. eds.), Munksgaard,
Copenhagen, 1988, p.409.
[12] Gaggelli, E., Di Perri, T.,
Orrico, A., Capecchi, P.L.,
Laghi Pasini, F. and
Valensin, G. Biophys.
Chem. 36, 209 (1990).

108 Bulletin of Magnetic Resonance
1. Introduction
Detection and Characterization Of CFC, HCFC AND HFC
Gases in Foamed Insulation by High Field NMR Imaging
Leslie H. Randall
Alberta Research Council, PO Box 8330, Station F,
Edmonton, Alberta, T6H 5X2.
In recent years, there has been considerable concern
regarding the environmental impact of
chlorofluorocarbons (CFC's). CFC's have been
widely used as blowing agents for both cellular
polyurethane and polystyrene insulation applications.
[1] In many non-critical applications, GFC's are
being replaced by non-fluorochemical blowing agents,
but in insulation based applications where the final
performance of the product is dependent on the
superior insulating characteristics of a closed cell
network which retains the fluorochemical blowing
agent, the use of similar blowing agents will probably
need to continue. Hydrochlorofluorocarbons
(HCFC's) have been identified as intermediate
replacements for the CFC's, but hydrofluorocarbons
(HFC's) are likely to become the blowing agents of
choice in insulation based products. The eventual
environmental fate of the blowing agent combined
with the dependency of product performance on the
fluorocarbon distribution within the cellular structure
requires an accurate knowledge of its spatial
distribution over a period of time in order to optimize
performance. Currendy, there is no reliable and
readily accessible technique with which the
Colin A. Fyfe, Zhiming Mei
University of British Columbia,
Dept. of Chemistry and Pathology,
Vancouver, B.C., V6T 1Y6.
and
Steve Whitworth
Du Pont Canada Research Centre,
Kingston, Ontario, K7L 5A5.
distribution of these gases can be detected or monitored.
Microscopic imaging has recently emerged as an excellent
technique by which the distribution of mobile fluids in
polymeric materials can be monitored. [2] - [8] In principal, it
should be possible to perform similar experiments on samples
which contain gaseous materials, the limiting factor being the
signal to noise. In the present study, we demonstrate that 19 F
NMR microscopic imaging is ideally suited for measuring the
changes that occur in the spatial distribution as a function of
time and yields quantitatively reliable information which will
be critical to the fabrication of optimized insulating materials.
2. Experimental
Aged foam samples were provided by Dupont Canada and
contained either a single fluorinated gas or a mixture of gases.
NMR measurements were made on a Bruker MSL 400
spectrometer equipped with a microimaging system. All
experiments were performed using the microimaging probe
supplied except that the probehead was modified by replacing
the vertical saddle proton rf coil by a 16 mm horizontal
solenoid coil that was tuned to fluorine (376.13 MHz). The
nonselective 90° and 180° rf pulses were 12.5 us and 25 ps
respectively. Quadrature phase cycling was used in all
spectroscopic measurements.

Vol. 14, No. 1-4 109
One dimensional 'H NMR spectra were obtained
by the standard one pulse method and by the Carr-
Purcell spin-echo experiment The Carr-Purcell NMR
sequence was used to determine the T2 spin-spin
relaxation times. [9] The inversion-recovery pulse
sequence was used to determine the Tj spin-lattice
times. [10] ID quantitative spectroscopic data was
obtained by using a short 1 us ring-down delay.
The spin-echo imaging sequence [11] was used
for all samples. For samples which contained a single
blowing agent, non-selective 90° and 180° rf pulses
were used. This reduced the echo time to 2 ms which
was advantageous, since spin-spin relaxation time
constant, T2 for the fluorinated gases absorbed into
the foam matrix was less than 5 ms. Images were
composed of 128 phase encoding steps. The number
of transients per experiment was typically 160. The
in-plane resolution was typically 270 pm using a
frequency encode gradient on the order of 7 G/cm.
Due to the short Tj relaxation time of the fluorinated
gases, the recycle delay was 100 ms which allowed
the acquisition of an image in under 30 minutes.
Concentration profiles were obtained using a
frequency selective gradient in a time period on the
order of 20 seconds.
Samples which contained a mixture of gases were
examined by a spin-echo imaging sequence in which
the initial excitation pulse was frequency selective
(Gaussian shaped, 300 ps duration). The echo time
for this experiment was typically 3 ms.
3. Results and Discussion
To compare the aging characteristics, rectangular
samples 1 cm x 1 cm x 2 cm in size were cut from
foam boards which had been aged for several months.
A variety of fluorinated gases were examined for both
polyurethane and polystyrene (high and low density)
foams. To devise the most appropriate imaging
protocol, it is important to know the relaxation
parameters of the 19 F nuclei. Tj was typically 10 on
the order of 10 ms while T2 was on the order of 4
ms. (Table 1). Thus, 19 F imaging will be very
efficient in that the experiment can be repeated very
quickly due to the short Tj values. The data must also
be acquired with a short time between excitation and
data acquisition due to the short T2 value. Using a
spin-echo sequence comprised of hard 90° and 180°
pulses, the echo time was reduced to 2 ms.
Figure 1 shows the 19 F image of a sample of
polystyrene foam of rectangular cross-section which
has been cut from the outside edge of a sheet of
foamed insulation. The sheet has been aged at room
temperature for 14 months. The concentration of gas
(CH3CF2C1) decreases from the outside edge inwards. The
quantitative distribution of cell gas is also obtained from an
examination of the image projection (Figure IB). This shows
the actual distribution as a function of distance. Although
there is only minute quantity of fluorinated gas present in the
sample, an image is easily obtained at high magnetic fields due
to the high sensitivity of the 19 F nucleus and its short spinlattice
relaxation time (10 ms).
Figure 1. Concentration of CH3CF2C1 in a sample of insulating
foam which has been aged 14 months,
(a) Projection, (b) Spin echo image.
Edge of foam
Table 1: I9 F NMR Relaxation Behaviour of
CFC, HFC and HCFC gases in Insulating Foams
Foam Cell Gas
Polyurethane CFC13
T2
5
3.6 5.4 -5.0
Polyurethane CF3CHC12 13.3 3.7 -85.9
Polystyrene CF2C12 4.4 4.5 -13.5
Polystyrene CH3CF2C1 4.9 4.2 -53.1
Polystyrene CF3CH2F 7.4 3.0 -85.0
-246.4
Tj and T2 are in ms.

110 Bulletin of Magnetic Resonance
Polystyrene samples which had been formed
using a mixture of blowing agents (CF2C12 and
CH3CF2C1) was also examined. The distribution of
the two gases was compared in samples which had
been aged two months and 14 months (Figure 2).
The concentration and distribution of CF2C12 gas in
the two samples is similar. However, the
concentration of CH3CF2C1 as a function of distance
from the foam edge has substantially decreased. This
implies that the loss of CF2C12 is somehow slowed by
the presence of the CH3CF2C1. These observations
have been verified by one- dimensional spectroscopic
measurements in which a calibrated standard has been
used.
Figure 2. (a) Concentration of CF2C12 gas
(i) after 2 months and (ii) after 14 months
Figure 2. (b) Concentration of CH3CF2C1 gas
(i) after 2 months and (ii) after 14 months
In a separate study, a cylindrical piece of insulating foam
which had been formed using CH3CF2C1 was then treated to
CF3CH2F in an oven at 80° C. Chemical shift selective
imaging was performed on the sample and revealed that the
post-treatment gas has penetrated the outside edge of the foam
and has formed a ring, approximately 2-3 mm in thickness.
(Figure 3a) The initial blowing gas, CH3CF2C1 has an almost
uniform distribution in the centre of the foam, which decreases
smoothly to the outside edge and includes the volume occupied
by the CF3CH2F gas (Figure 3b). The loss of CH3CF2C1 is
substantially less in these samples as compared with samples
that had been placed in the oven (in air). Thus, in the
presence of a fluorinated gas, the blowing agent is lost at a
much slower rate.
Figure 3. (a) Concentration of CH3CF2C1 gas.
Figure 3. (b) Concentration of CF3CH2F gas.

Vol. 14, No. 1-4 11!
4. Conclusions
These data are typical of those we have obtained on
a variety of fluorocarbon gas/foam matrices and
clearly indicate that 19 F NMR microscopic imaging is
ideally suited for measuring the distribution of
fluorinated hydrocarbons in polystyrene and
polyurethane foams. Presently, a protocol is under
development which will ensure that quantitatively
reliable information is being obtained. This will
allow us to monitor the changes that occur in the
spatial distribution as a function of time (accelerated
aging tests) or as a function of blowing agent. This
information can then be used to optimize the
fabrication of these insulating materials.
5. References
1. Proceedings of the Polyurethane World Congress,
1987.
2. Rothwell, W. P., Holeck P. R. & Kershaw, J. A.
/. Polym. Sri. Polym. Lett. Ed., 22, 241 (1984).
3. Blackband, S. & Mansfield, P. J. Phys. C: Solid
State Phys., 19, L49 (1986).
4. Marcei, T. H., Donstrup, S. & Rigamonti, A. J.
Mol. Liquids, 38, 185, (1988).
5. Weisenberger, L. A. & Koenig, J. L. Appl.
Spectrosc, 42, 1117(1989).
6. Weisenberger, L. A. & Koenig, J. L.
Macromolecules, 23, 2445 (1990).
7. Webb A. G. & Hall, L. D. Polym. Commun. 11,
422 (1990).
8. Webb A. G. & Hall, L. D. Polym. Commun. 11,
425 (1990).
9. H. Y. Carr and E. M. Purcell, Phys. Rev., 94, 630
(1954).
10. R. L. Void, J. S. Waugh, M. P. Klein and D. E.
Phelps, J. Chem. Phys., 48, 383 (1968).
11. Edelstein, W.A.; Hutchinson, J.M.S.; Johnson G.;
Redpath, T. Phys. Med. BioL, 25, 751 (1980).

112
MYSTERIOUS NEGATIVE PEAKS
IN THE 1 H{ 1 H} NOE DIFFERENCE SPECTRA
OF SOME THIOPYRAN COMPOUNDS
Thioyrans 1-4 were formed in the course of
regiospecific and stereoselective Diels-Alder
cycloadditions involving 2-(N-Acylamino)-lthia-l,3-dienes.
The constitution and the main
conformational features of thiopyrans 1-4 were
established using classical high field NMR
methods: 2D ^ H and 13 C- X H shift correlation,
selective 13 C{ l U} and ^H} NOE difference
experiments [1], [2].
Conformational aspects:
a) ring pseudorotation:
In the case of compounds 3 and 4 the
thiopyran ring exhibits a slow ring flip between
the relevant half-chair forms. This gives rise to
saturation transfer effects in the 1 H{ 1 H} NOE
difference experiments, as illustrated for 4 in
Fig. 1.
Csaba Szantay, Jr.
Chemical Works of Gedeon Richter Ltd.
H-1475, Budapest, POB 27, Hungary.
Bulletin of Magnetic Resonance
In 1 and 2 the ring conformational equilibrium
is completely shifted towards the form
in which H-2 and H-3 are antiperiplanarly arranged
(Fig. 2).
b) The C(3) sidechain:
The sidechain shows free rotameric mobility
at room temperature. The conformational features
were characterized in terms of the
measured NOEs and MM calculations [1]. The
relative contributions of the major sidechain
conformations are depicted in Fig. 2.
Ac2N-
AcN-
anti syn
Ph
H
s cis—trans
H P \ h f
. trans—cis
\h
Ph
H
0-
i %
1
A-
~100%
2
Ac2N—(/
AcN—V
Et S
Figure 2. The main rotameric forms of the
C(3) sidechain in compounds 1-4.
Q

Vol. 14, No. 1-4 113
minor
NHMe
major
H-5
U.I
NH
HMe
H-3
NMe
NHMe
•» ifr
COMe
Figure 1. One of the several NOE difference spectra of compound 4 showing the presence of
saturation transfer as a result of a slow ring interconversion.
H-5
JL
G.00 5.S0
Ph
-14X
H-2
PPH S.00
4.SO
+9X
H-4
JL
Figure 3. A segment of the NOE difference
spectra showing negative peaks in
the case of 2. (CDC13,25 °C, 400 MHz).
Ph
2 PPH
-30°C
+7X
' I ' • ' ' I ' ' • • I ' • « " T " t ' • • ! • ! •
6.0 5.S 5.0 4.5 4.0
PPH
Figure 4. A segment of the NOE difference
spectra showing negative peaks in
the case of 1. (CDC13, -30 °C, 400 MHz).

t
,!
11
114
The mystery:
During these invetigations a highly unusual
phenomenon was observed in the ^^H} NOE
difference spectra of compounds 1 and 2, in that
the H-3 and H-2 protons are connected by massive
negative peaks while all other NOEs are
positive as illustrated in Figs. 3 and 4.
The facts:
a) The observed effect involves H-2 and H-3
only.
b) The phenomenon appears to be specifically
linked to the presence of the C = O unit in
the sidechain, and is compeletely absent in 3
and 4.
c) As measured at 400 and 300 MHz, respectively,
the negative peaks showed no field dependence
beyond that attributable to
experimental errors.
d) The intensity of the negative enhancements
increases markedly with decreasing
temperature and viscosity, and tends towards
zero at elevated temperatures in accordance
with the antiperiplanar arrangement of H-2 and
H-3:
Negative enhancements at 400 MHz,
measured at various temperatures using
otherwise identical experimental
parameters:
1
-30°C
-20%
CDCb
+ 25°C
-3%
+25°C
-14%
+50°C
0%
+ 50°C
-4%
DMSO
+25°C
-14%
e) The nonselective Tx relaxation times of all
ring protons in 1 (CDC13, +25°C, 400 MHz)
were measured to be ca. 1 s. This accords with
the size and expectedly fast tumbling of the
molecule.
Bulletin of Magnetic Resonance
Conceivable (but rejectable) explanations:
The effect seems to be inexplicable in terms
of the well understood cross-relaxation or conceivable
saturation transfer processes:
a) The trivial case of irradiation spillover can
be discounted because: 1) This would be incompatible
with the observed temperature dependence
of the negative peaks; 2) Fig. 4. illustrates
a situation in which the signal due to H-3 is
"halfway" in between those of H-5 and H-2, of
which only the latter shows the negative effect.
(The irradiation power levels used in all experiments
were: 54L (Bruker AM 400) and
DLP = 30 (Varian VXR-300).
b) A three-spin effect is geometrically unjustified,
and is incommensurate with the magnitude
of these negative enhancements.
c) The possibility of saturation transfer may
be considered as being the result of some form
of chemical exchange in 1 and 2. Possible options
are: 1) direct proton transfer between H-2
and H-3. However, *H and 2 H NMR studies on
the C(3)- 2 H labeled isotopomer of 1 have conclusively
shown that any possibility of a
stereoselective H-2—H-3 proton transfer can be
discounted. 2) A possible chemical exchange
may also involve restricted ring pseudorotation
or restricted C(3) sidechain mobility, both
being slow on the chemical shift time-scale. Assuming
that in such a hypothetical situation the
H-3 signal of the minor conformer lies directly
underneath the H-2 signal of the major conformer
(and vice versa !), this could lead to
negative peaks that are (vaguely) similar to
those observed in the NOE difference spectra.
However, such a situation can be ruled out on
account of several quite obvious considerations,
the most trivial being that 1) the multiplet patterns
of the observed negative peaks do not
conform to those expected for either of the
above possibilities for slowly interconverting
conformational species; 2) the phenomenon is
observed only in relation to H-2 and H-3.

Vol. 14, No. 1-4
Moreover, the observed temperature and
viscosity dependence of the effect is contrary to
that normally expected for saturation transfer:
1) The longer correlation times associated with
higher solution viscosity (in our case in DMSO)
or lower temperatures are, in the extreme narrowing
limit, related to more efficient Ti relaxation
which works against saturation transfer. 2)
Decreased temperatures usually diminish the
efficiency of the exchange process.
d) A major contribution from scalar relaxation
[3] which could be perceived as being
brought about by the somewhat labile character
of H-3 can be ruled out since the J(H-2,H-3)
couplings show no sign of the rapid modulation
that would result in the collapse of the relevant
rmiltiplet patterns.
e) Strong coupling effects [4] are not present,
and would not give negative peaks of this size.
f) Anisotropic reorientation in the intermediate
region between fast and slow tumbling
can be considered. This carries the possibility of
exhibiting small positive and small negative enhancements
simultaneously. However, all the
115
positive enhancements measured between spatially
analogously related protons in 1-4 were of
similar magnitude, and in line with that expected
for a fastly tumbling molecule.
The phenomenon therefore still awaits adequate
rationalization. The question of how the
effect may be related to the geometrical and
constitutional specifics of these molecules is
still being pursued.
1. I.T. Barnish, C.W.G. Fishwick, D.R. Hill,
Cs. Szantay Jr., Tetrahedron, 45 (1989) 6771.
2. Cs. Szantay Jr., I. Moldvai, C.W.G. Fishwick,
D.R. Hill, Tetrahedron Lett, 32 (1991)
2529.
3. D. Neuhaus, M.P. Williamson, The
Nuclear Overhauser Effect in Structural and
Conformational Analysis, VCH Publishers, New
York, 1989, pp. 203-207.
4. J. Keeler, D. Neuhaus, M.P. Williamson, /.
Magn. Reson., 73 (1987) 45.

116 Bulletin of Magnetic Resonance
H-1 and C-13 NMR Spectra of the Carbanions
Produced from Phenylpropene Derivatives
Akihiro Yoshino, Kensuke Aoki, Masahiro Ushio,
and Kensuke Takahashi
Department of Applied Chemistry, Nagoya Institute of Technology,
Gokiso-cho, Showa-ku, Nagoya 466, Japan
Abstract l,l-Diphenyl-2-methyl-l-propene produces an equimolar mixture
of two carbanions in contact with excess potassium-sodium alloy in
tetrahydrofuran. The carbanions can be interpreted as the products of disproportionation
reaction between two radical anions produced by one-electron
reduction of the phenylpropene with alkali metal.
1 Introduction
Various carbanions can be prepared
from substituted ethylenic
compounds (1) in contact with alkali
metal in tetrahydrofuran(THF).
In these reactions a radical anion
(2) is formed as shown in Scheme 1
Scheme 1
and its reactivity or stability will
depend on the nature of the substituents.
The three routes of reactions
can be considered as follows.
(1) If the ethylene has bulky
substituents on both Ci and C2, the
corresponding ethylene dianion is
produced (Scheme l).[l-4] (2) If
the ethylene has no or sterically
small substituent such as only one
methyl group on either Ci or C2, the
corresponding dimer dianion is produced
(Scheme 2).[5-7] (3) In the
present paper, three title phenyl
R, © • _R
2 X >C,-C2 '>C,-CH< 3 + >C,-C2< '
^ R4THF R/ V R4 R/ 2K R4
2 5 6
Scheme 3
carbanions, 5 and 6 respectively,
whose structures are confirmed by
!H and 13 C NMR spectra. One of
these anions(5) is a phenylalkyl
carbanion and the other is a phenylallyl
carbanion (6). Thus the result
may be interpreted as a dispro-

Vol. 14, No. 1-4 117
portionation reaction between two
anion radicals generated first by
one-electron reduction of the
starting phenylpropene with alkali
metal. As far as we know, this type
of simultaneous formation of two
different carbanions has not yet
been reported. The conditions of
these reactions will be discussed in
terms of steric bulkiness of the
substituents.
2 Experimental
The starting materials were prepared
from dehydration of the corresponding
phenylpropanols which
were prepared by Grignard reaction
and followed by dehydration with
anhydrous acetic acid. The starting
materials dissolved in THF or THFd8
were kept in contact with
potassium-sodium alloy in vacuum
at room temperature for about 24 h.
The resulting dark red solutions
were filtered, concentrated if necessary,
and then sealed into a 5-mm
NMR sample tube. Their concentrations
were about 1 M. The *H and
13 C NMR measurements were carried
out at 22°C using Varian XL-
200 or Unity-400 spectrometer.
The chemical shifts were evaluated
from the upfield peak of THF or
THF-dg, used as an internal reference.
This peak was taken as 1.79
or 1.75 for l H and 26.4 or 26.0 ppm
for 13 C resonances, respectively,
from TMS.
3 Results and Discussion
1 H NMR Spectra of the Carbanions.
A typical l H NMR spectrum and
chemical shifts are shown in Fig. 1
and Table 1, respectively. In Fig. 1,
there are three signals in the region
from 0 to 3 ppm, which are assigned
to one methyl and one isopropyl
groups. There are two
ethylenic signals in the region from
3 to 4 ppm. One problem in the
spectrum is the origin of hydrogen
in the isopropyl group. To clarify
the origin, the experiments using
two different solvents, THF and
THF-d8 were carried out. Since
these two spectra are similar each
Table 1. *H Chemical Shifts of Related Carbanions and Their Precursors.
No.
1 a
1 b
1 c
5 a
5 •b
5
6
6
6
c
a
b
c
R?
Ph
Ph
CH3
Ph
Ph
CH3
Ph
Ph
CH3
R 4
CH3
C2H5
CH3
CH3
C2H5
CH3
CH3
C2H5
CH3
H2
2 .904
2 .591
2 .540
t rans ci s
1
H3
1.862
1.107
.580 a) 1.
1.115
1.114
0.875
780
4 .035
4
3
b> 4.
.375 b) 4.
.141 b> 695
811
3. 343
a )
b)
b)
b)
Ho
( H o )
1
1
1
6.741
6.756
5.084
5.368
7.119
7.091
6.584
H m
(H»- )
7.24
7.25
7.14
6.593
6.563
6.076
6.044
6.711
6.661
6.656
Hp
1
1
5.760
5.726
4.364
6.121
5.966
5.735
a) Trans(R4) and cis(Ra) configurations are defined for
b) They are defined for Ci.
Hothe
CH3
1.842
1.950
0.872
1.287
1.870
1.068
1.858
1.971
r s
CH2
2.802
1.467
1.628
2.252

118
6Ho
5Ho
l6Hm 6HMe 5H3
Ph CH3 Ph CHs
\ - / \ - /
C-C H + C-C
/ 1 2 \ / 1 2 ^
P h 3 C H i P h 3 C-H
- /
H
5a 6a
8 4
6 I ppm
Fig.l !H NMR spectrum of an equimolar mixture
of 5a and 6a in THF-d8 at 200MHz.
5Cm
6C2
6Ci 5Ci
1 f
160
6Cm
6Co P h e n , P h C H ,
\- / \- /
C-CH + C-C
5CO / 1 2 \ /12V
Ph 3CHj Ph H 3 C-H
5a 6a
6Cp
Bulletin of Magnetic Resonance
5Cp 6CMe
5C3
5C2
6C3
5C1
Jt
l l l 1 1 i i
80
Fig .2. 13 C NMR spectrum of a mixture of 5a
and 6a in THF-d8 at 50.3 MHz.
l i i i i i i
61 ppm
I
0
0

Vol. 14, No. 1-4
other, it must be concluded that the
CH hydrogen in the isopropyl group
of 5a is coming from another radical
anion (2), but not from the solvent.
This is also supported by an
experimental fact that the product
ratio of 5 and 6 is 1 to 1. Another
point of interest is that two
methylene hydrogens of 5b give
different chemical shifts, 1.467
and 1.628 ppm, respectively. This
is ascribed to the presence of an
asymmetric center on C2.
13 C NMR Spectra of the Carbanions.
A typical 13 C NMR spectrum and
chemical shifts are shown in Fig. 2
and Table 2, respectively. In Fig. 2,
there are 15 signals for a mixture
of 5a and 6a except for two solvent
peaks. One methyl signal is overlapped
with a more shielded solvent
peak. The carbon species can be
differentiated by the DEPT technique.
The assignment of each peak
is shown in Fig. 2.
IH and 13 C NMR Chemical Shifts.
All the *H and 13 C except for C2 and
Cj of the carbanions (5 and 6) show
upfield shifts as compared with
those of the starting materials.(1)
These upfield shifts are mainly
Table 2. 13 C Chemical Shifts of Re
No. R2 R.4 C
(C
la Ph CH3 138.48 131.24
1 b Ph C2H5 138 .47 136.74
1 c CH3 CH3 146.11 127.47
5 a
5 b
5 c
6 a Ph
6 b Ph
6 c CH3
Ph CH3
Ph C2H5
CH3 CH3
90.26
89.56
78.19
30.30
38.24
29.10
CH3 89.61 147.54
C2H5 88.65 156.93
CH3 80.24 144.30
ascribed to the extra negative
charges on carbons of the carbanions.
Since the carbanions have their
stability due to partial delocalization
of the extra charge into the
phenyl rings, phenyl rings are
necessary for formation of a stable
carbanion. In fact, tetramethylethylene
does not react with a
potassium-sodium alloy in a
similar condition used for the
present study.
The extent of charge delocalization
can be thus estimated by comparison
of the *H or 13 C NMR chemical
shifts. The *H and 13 C chemical
shifts of Hp and Cp of 6a are less
shielded than those of 5a, by about
0.361 and 5.49 ppm, respectively.
These shift differences divided
by 10.7 and 160 ppm, respectively,
gave about 0.034 unit of extra
charge difference on Cp. [8]
Therefore, the chemical shift
difference between Hp or Cp of 6a
and 5a is explained in terms of
their localized excess charges. The
same is true for 6b and 6c in
comparison with 5b and 5c, respectively.
The especially large
lated Carbanions and Their Precursors.
C
) (c
V-» o t h e r s
CH3
22.71 144.22 130.58 128.61 126.85 ---
19.26 144.25 130.17 128.64 126.83 13.64
20
21
22
19
21
64
07
11
46
76
130.40 128.71 126.89
131.14 129.08 128.71 126.55 22.33
CH2
119
29.20
144.28 118.75 128.90 107.91 —
145.02 118.77 128.81 107.80 14.19 29.45
135.54 105.30 130.75 88.50 12.09
105.47 130.25
93.16 148.38 124.86 128.61 113.40 24.76
97.66 148.25 123.21 128.61 112.11 15.87 29.94
76.36 146.24 118.64 128.59 107.46 19.39
27.07

120
difference between 6c and 5c, is
ascribed to the delocalized capacity
for excess charge consisting of only
one phenyl ring. Therefore, Ci of 5c
is more shielded by about 10 ppm
than that of 5a or 5b. The same is
true for Ci and C3 in 6c; they are
more shielded than those in 6a or
6b. Thus the p-orbital of Ci can
conjugate with those in both phenyl
and allyl groups in 6., while that of
5 can only conjugate with phenyl
rings.
For the allyl anions the average
shift of the outer carbons (Ci and
C3) is shielded by about 60 ppm
than that of the center carbon (C2),
as shown in Table 2. This is a characteristic
nature of the allyl anions.[12]
Dunkelbium and Brenner reported
on the same carbanion (6a) with
lithium as a counterion.[13]
Although their chemical shifts of
6a were deshielded by about 0.1
ppm, their *H NMR data are consistent
with ours in consideration of
different counterion.
Internal Rotations. Three internal
rotations can be considered around
the bonds of Ci to Ci, Ci to C2, and
C2 to C3 for 5 and 6 (these will
hereafter be referred to as rotations
A, B, and C). In the *H and 13 C
NMR spectra of 5c, two signals for
Ho, Co, Cm, Hm were observed and
their chemical shifts are given in
Tables 1 and 2, respectively. This
is interpreted as a restricted rotation
A for the phenyl ring of 5c
even at room temperature. A similar
observation was presented earlier
in a methylbenzyl anion.[14] A
chemical shift difference of 0.284
ppm for Ho of 5c is smaller than
that of 0.43 ppm for Ho of methyl-
Bulletin of Magnetic Resonance
benzyl anion. The cause of the
shifts cannot be clarified yet; the
shift differences for Ho, Co, Cm,
and Hm changed to 0.284, 0.17,
0.50, and 0.032 ppm. These four
sites are adjacent to each other.
But the values change alternatively.
Therefore the origin of the shift
may not be explained simply. On the
other hand, the internal rotations A
of the phenyl groups of 5a and 5b
were not restricted at room temperature.
At lower temperature,
however, the rotation B was inhibited
at -60°C for 5b. For 5a the
aromatic carbon signals were significantly
broadened at -80°C but
were not split. In comparison with
this behavior, however, those aromatic
carbon signals of 6a and 6b
were observed sharply even at
-80°C. Further study is necessary.
Two Ho signals of 5c at 5.084 and
5.368 ppm were correlated with
two Co signals at 105.30 and
105.47 ppm, respectively, in a 2D
CH COSY spectrum for their assignment
purpose. [15] Similarly, two
Hm signals (6.044 and 6.076 ppm)
are correlated with two Cm signals
(130.25 and 130.75 ppm, respectively).
Which one of the two Ho's
locates near to methyl or isopropyl
group? NOE of the methyl signal
(1.287 ppm) was observed on the
more shielded Ho signal (ca. 20 %).
Therefore the methyl group on Ci is
nearer to Ho than Ho'. NOE of another
methyl signal (0.875 ppm)
was observed on the less shielded
Ho" signal (ca. 10 %).
Condition of Disproportionation
Reaction. The condition can be
discussed in terms of bulkiness of
the substituents. Four groups are

Vol. 14, No. 1-4 121
used as stubstituents on Ci and C2;
namely H, CH3, C2H5, and CeHs. They
are differentiated to have approximately
1, 2, 3, and 5 A in their diameters
as spherical models. For
simplified discussion, we consider
that 2 has bulkier substituents on
Ci than on C2. Thus the C2 site will
be more reactive than the Ci site
for dimerization. In this occasion
discussion will be rather limited to
the bulkiness of substituents on C2.
In cases where substituents are
two C6H5 or one CgHs and one H,
monomer dianions are formed
(Scheme 1). Dimerization of the
radical anion occurs in cases where
the substituents are two hydrogens
or one H and one CH3 (Scheme 2). In
cases where the substituents are
two CH3, disproportionate occurs
(Scheme 3). Therefore bulkiness of
the substituents may control the
progress of the reaction. Two alkyl
groups are large in size for dimerization,
and small for dianion formation,
and may be suitable for
disproportionation. For disproportionation
reaction, the substituent
must have a hydrogen to be abstracted.
One CH3 on C2 of la is
substituted by C2H5 in order to investigate
which hydrogen in two
substituents, either CH3 or C2H5, is
easily abstracted. From analyses of
a mixture of the products, it is
concluded that the hydrogen-leaving
power is about ten times stronger
in CH3 than in C2H5.
Acknowledgement
The authors wish to thank Mr. K.
Kushida and A. Ono (Varian
Instruments, Ltd.) for their kindness
in measuring several 2D CH
COSY spectra at 100.6 MHz.
4 References
1) K.Takahashi,Y.Inoue, and R.Asami,
Org. Magn. Reson., 3, 349 (1971).
2) M.Morikawa, H.Matsui,A.Yoshino,
and K.Takahashi,£«//. Chem.Soc.Jpn.,
57,3327(1984).
3) H.Fujiwara,A.Yoshino,
Y.Yokoyama, and K.Takahashi,
Bull.Chem.Soc.Jpn., 65, No.8, in
press (1992).
4) Y. Yokoyama, T.Koizumi, and
O.Kikuchi, Chem. Lett., 1991, 2205.
5) K.Takahashi and R.Asami,
Bull.Chem.Soc.Jpn., 41, 231 (1968).
6) K.Takahashi,M.Takaki, and
R.Asami, J.Phys.Chem.,
75,1062(1971).
7) Y.Yokoyama, et.,
Bull.Chem.Soc.Jpn., 61, 1557(1988).
8) The charge calculation was followed
by using an empirical equation
proposed by Fraenkel et al.[9]
and proportional factors presented
by Schaefer et al.[10,ll]
9) G.Fraenkel, et al.,
J.Am.Chem.Soc.,82, 5846(1960).
10) T.Schaefer and W.G.Schneider,
Can. J. Chem., 41, 966 (1963).
11) H.Spiesecke and W.G.Schneider,
Tetrahedron Letters, 1961, 468.
12) D.H.O'Brien/'Comprehensive
Carbanion Chemistry," ed by
E.Buncel and T.Durst, Elsvier, New
York (1980), Vol 5A, p.273.
13) E.Dunkelblum and S.Brenner,
Tetrahedron Letters, 1973, 669.
14) K.Takahashi, et al.,
Org.Magn.Reson., 3, 539 (1971).
15) The 2D CH COSY spectra were
measured at 75.4 MHz with a
Hitachi R-3000 spectrometer. The
authors wish to thank Mr. M. Tamura
(Instrument Division Hitachi, Ltd.)
for his kindness in measuring the
spectra.

122 Bulletin of Magnetic Resonance
Intracellular pH and Inorganic Phosphate
Effects on
Skeletal Muscle Force
E. R. Barton-Davis 1 , R. W. Wiseman 2 ,
and
M. J. Kushmerick 1 ' 2
Depts. of Physiology and Biophysics* and of Radiology 2 ,
University of Washington,
Seattle, WA 98195, USA
INTRODUCTION
The molecular mechanisms of skeletal muscle
fatigue are not fully understood. Intracellular pH
(pHi) and inorganic phosphate (Pi) have been
implicated as causes of peripheral fatigue. Both a
rise in [H + ] (decreased pHi) and elevated Pi
result in the inhibition of force generation in
skinned fibers (2) and sometimes in whole
muscle preparations (1, 3). During exercise, an
accumulation of H + and Pi occur simultaneously
over time, which confounds the interpretation of
the relative role of each with regard to skeletal
muscle fatigue. The specific protonation state of
Pi might also inhibit force generation, because
the relative amount of Pi species (H2PO4" and
HPO4 2 ") is dependent on pHi, and these amounts
change over the physiologic range of pHi (pKa =
6.79) (3). In order to evaluate the role these
metabolites have in muscle force, the effects of
pHi and Pi on force have been dissociated by
independently manipulating each.Changes in the
amount of CO2 in the perfusate (PCO2)
manipulates pHi (1), and exogenous pyruvate
lowers Pi (6). Performing simultaneous NMR
spectroscopic and mechanical measurements
can assess the effects of these treatments. We
have compared the relationship between force
and pHi in the presence and absence of inorganic
phosphate and interpreted the relative
contribution of Pi and pHi on muscle force
generation.
METHODS
Isolated soleus from male Swiss Webster mice
were attached to a Harvard Apparatus isometric
force transducer and placed in a bath containing
MOPS Ringers (llmM glucose, pH=7.4, 95%
O2, 5% CO2) at 23°C. Tetanic force, as well as
rise and relaxation time, were measured at
optimum length every 10 minutes. Stimulation
was at supramaximal voltage and fusion
frequency for 1 second. The collateral muscle
was placed in a custom designed NMR probe (5)
and superfused from the same source of Ringers
as the mechanics bath for 31 P-NMR spectra were
acquired on a General Electric 7 Tesla GN300 at
121 MHz. Intracellular pH measurements were
made using the chemical shift of Pi relative to
PCr and the equation
pHi = 6.79 + log{(.89 - 8)/(8 - 3.19)}.
Mechanical and metabolic measurements were
repeated with control (11 mM glucose) and
substrate (20mM pyruvate) solution (PYR),
varying pHi by equilibration with 5%, 25%, and
50% CO2, which resulted in PCO2 values of 22,
174, and 266 torr respectively. All isometric
contractions were bracketted by llmM Glucose
5% CO2 treatment to normalize data. Statistical
analysis was performed using 2-way ANOVAS
with relative force as the dependent variable, and
pHi and Pi as independent variables. Paired
comparisons of means were performed between
each PCO2 l eve * aa ^ substrate treatment. Values
of p

Vol. 14, No. 1-4
RESULTS
PYR lowered Pi levels from ~5 mM to

124 Bulletin of Magnetic Resonance
TABLE II: Total, Diprotonated, and
Monoprotonated Phosphate Levels in
Soleus
Glue
5%
CO2
Glue
25%
CO2
Glue
50%
CO2
Pyr
5%
CO2
Pyr
25%
CO2
Pyr
50%
CO2
pHi
7.19
+.015
6.81
+.018
6.54
+.013
7.22
+.04
6.74
+.04
6.48
+0.0
DISCUSSION
Total
Pi
(fi«/g)
6.76
+.73
(n=22)
7.74
+2.10
(n=4)
6.74
+.05
(n=2)
2.64
+1.01
(n=2)
1.69
(n=l)
N.D.
H2PO4
(M«/g)
1.91
+.22
3.68
+.95
4.25
+.02
0.69
+.22
0.88
N.D.
HPO4-
(uste)
4.85
+.52
4.06
+1.16
2.48
+.07
1.96
+.78
0.81
N.D.
Data presented in mean ± SE
Decreased [Pi] increased force at all PCO2
levels, and did enhance the dependence of force
generation on pHi. The effect of pHi on force
concurs with skinned fiber data (2), but does not
agree with results from hypercapnic
manipulation in perfused muscle preparations
(1). There are several possible explanations for
this discrepency. One possibility is that there is
inadequate O2 delivery to the isolated muscle
preparation. However, since the mouse soleus is
small (~1 mm in diameter), there is little chance
of O2 limitation during hypercapnia because the
diffusional distances are less than 500p.m. It is
also possible that at low pHi, a 1 second tetanic
stimulation was not sufficiently long enough for
the muscle to attain maximum force . However,
longer stimulations did not appreciably increase
force (data not shown). Force measured at 1
second tetani was not significantly different from
that of longer tetani. Hence, the results observed
give an accurate presentation of the effects pHi
has on tetanic force in isolated whole muscle.
It has been suggested that a specific protonation
state of phosphate is correlated to muscle force
(3). The diprotonated species (H2PO4") is
thought to lower force by binding more tightly to
actomyosin crossbridges. Because force
production is correlated with the release of Pi
from the crossbridge, the higher binding affinity
of H2PO4" would result in a decrease in force. As
pHi decreases, there is a greater proportion of
H2PO4" present, indicating that this
phenomenon is an indirect effect of pHi on
skeletal muscle force. We examined the
dependence of force on pHi, HPO4 2 ", and
H2PO/. The slopes of each regression line show
a higher correlation between force and pHi
(R^.966 in normal [Pi]) (Figure 1) than between
force and either phosphate species (H2PO4",
R^.138; HPO4 2 ", R 2 =.O35) alone (Table II). In
these experiments, the primary effect of pHi on
force does not appear to be mediated by a
specific species of phosphate.
Table I shows that pHi and Pi influence other
properties during tetanic stimulation. Proton
accumulation could alter kinetics of force
development (rise and relaxation times) in
several ways. Enzymes, including Ca 2+ -ATPase,
actomyosin ATPase, and those in glycolysis,
such as phosphofructokinase, are inhibited when
pHi falls. The decline in enzymatic activity
inhibits the speed at which a muscle can develop
force or relax after stimulation. Low Pi causes
faster rise yet slower relaxation times by
different mechanisms which are not fully
understood. Faster kinetics are expected since the
phosphate release step in the crossbridge cycle
would be more favorable due to lower Pi.
However, this hypothesis can not explain why
relaxation rates are significantly slower than in
control conditions. It would seem plausible that
Ca 2+ uptake at the sarcoplasmic Ca 2+ -ATPase
would be enhanced by the increase in AG for
ATP breakdown caused by low free Pi. Yet this
is in direct opposition to our observations. If
phosphorylation of the Ca 2+ -ATPase is
necessary to stimulate Ca 2+ transport, this type
of upregulation is not as effective as in normal
conditions because free Pi has been depleted.

Vol. 14, No. 1-4 125
Ca 2+ would remain in the sarcoplasm for a
longer period of time, and relaxation would
slow. It seems that there are two independent
mechanisms responsible for the effect of low [Pi]
on rise and relaxation.
By maintaining muscles in a resting state, our
experimental design avoids introducing factors
arising from constant stimulation, specifically, a
concomitant rise of proton and phosphate levels.
The independent diminution of Pi in this
preparation allows the examination of the relative
contribution of pHi on force generation. In so
doing, we conclude that the relation between
skeletal muscle force and pHi is affected by the
depletion of Pi.
REFERENCES
(1) Adams et al. 1991. AJP, 260 (29), C805 -
C812.
(2) Chase & Kushmerick. 1988. Biophysical
J., 53,935-946.
(3) Dawson et al. 1986. Biophysical J., 49,
286a.
(4) Kushmerick et al. (in press) PNAS.
(5) Wiseman et al. 1990. Biophysical J., 59(2),
517a.
(6) Wiseman et al. 1991. SMRM Abstracts,
10th Annual Mtg., 284.

126 Bulletin of Magnetic Resonance
Introduction
A simple model for the influence of motion
on the NMR line shape.
M. Goldman, T. Tabti, C. Fermon, J.F. Jacquinot and G. Saux.
A problem as old as NMR is that of the influence
of motion on the shape of the NMR absorption
signal, or equivalently the shape of the FID.
Although it is known that motion means
narrowing, only two limiting cases are well
defined.
In the rigid solid, one has a clean theoretical
expression for the FID function, in terms of the
secular part 3TSS of the spin-spin interactions, most
of the time essentially dipolar (1). This is
insufficient to know the FID shape, nor that of its
Fourier transform, the absorption f signal, but one
can compute the first few moments of the latter
and obtain reasonably good value for its frequency
width Ad)0 (1-3).
The other extreme is that of a fast motion whereby
the spin-spin interactions have a vanishing average
value and are modulated at a rate x' 1 which is fast
on the time scale of the solid-state FID decay rate
AGV.
ACOOTC [1]
Service de Physique de FEtat Condense,
C.E.N. SACLAY, 91191 GIF SUR YVETTE
FRANCE.
In that case, relaxation theory predicts that the
transverse magnetization decay is exponential,
with a relaxation time T2 related in a precise way
to the magnitude of the spin-spin interactions and
the nature of the motion (1,4). The absorption line
is then Lorentzian, and its half-width at half
intensity T^ x is of the order of:
2
IC
[2]
Things are much more confuse in the intermediate
case, that is for motions rates for which
Ao)oxc«l. In that domain tjiere exists only
qualitative models of the Anderson-Weiss (5) or of
the Kubo-Tomita type (6), which are admittedly
very crude and whose sole ambition is to provide a
general physical feeling of how the signals are
evolving from slow to fast motion.
In this article, we attempt to correlate the FID
shape under intermediate rate motion to that of the
rigid solid. We use for that purpose a simpleminded
model that is not new: it is the so-called
Strong Collision Model (7). It has been used in
particular for (i.SR studies (8) but in conjunction
with other approximation. This is at variance with
our approach, which treats the rigid lattice FID as
an empirical information. This model is not

Vol. 14, No. 1-4
rigorous either, but it fits remarkably well the
experimental results, as will be shown later.
The theoretical model
The scenario of the Strong Collision Model is the
following. Let us consider a spin system whose
FID signal is Gr(t) when it is rigid, with the usual
normalization:
G,(0) = l [3]
This system undergoes sudden motions at random
time intervals with probability X per unit time.
After each motion, the FID starts anew with the
same shape as initially, but with an initial value
equal to Gr immediately before the jump. In other
words all correlations developed in the system by
its previous evolution are lost, and the only
"memory" left is that of the transverse polarization
at the time of the sudden motion. This results in a
FID signal Gm(t) (where m stands for motion)
which is the weighted average of all possible
random successions of initial rigid-state FBD's.
A pulse being applied at time 0, the possibilities at
time t are:
i) No jump between 0 and t. This corresponds to a
FED signal shape Gr(t) and its probability to occur
is equal to exp(-Ai). Its contribution to Gm(t) is
then:
Gr(f)«p(-X0
ii) One jump only at a time between f, and /, +dtv
The corresponding probability is:
x exp[-X(t - tx)] = X exp(-Xt)dt1
and the signal at time t is then:
mt egrating over the time f, we obtain the
contribution to Gm(t):
127
iii) Two jumps between 0 and /. By an extension of
the preceding argument, the contribution to Gm(t)
is:
JT Gr{t2)Gr(t-tx-t2)dt2
etc...
By taking the Laplace transforms:
we obtain:
that is:
e
= [G(t)exp(-zt)dt
n=0
[4]
[5]
[6]
which is identical with Eq.(18) of ref.(8).
The same result can be obtained much more simply
by noting that after a jump the subsequent FID is
that of the mobile system normalized to the
transverse polarization at the time of the jump.
The two possibilities being no jump between 0 and
/, or at least one jump, we obtain :
[7]
whence the following relations between Laplace
transforms:
[8]
from which Eq.[6] follows.
The advantage of the first treatment is to make
explicit use of the assumption that each partial FID
between jumps has the same shape. The simplest
way of interpreting Eq.[6] is through the
consideration of memory functions (2,3,9). The
memory function K of a function G(t) is defined
through:

128 Bulletin of Magnetic Resonance
[9]
This form yields remarkably simple expressions for
the Laplace transforms. That of the left-hand side
is equal to: z0(z)-G(O) = z8(z)-l, where we
have used Eq.[3], and that of the right-hand side is
equal to -cp(z)9(z), where cp(z) is the Laplace
transform of K(t):
Then, Eq.[9] yields:
z8(z)-l = -(p(z)8(z)
Now, Eq.[6] yields:
1 1
or else:
1
1
[10]
[11]
"•— A* [12]
-z-X
whence, according to Eq.[ll]:
[13]
[14]
This corresponds to the following relations
between memory functions:
[15]
We obtain the remarkably simple result that in the
Strong Collision Model, the rate of decay of the
"memory" shows up simply by an extra
exponential decay of the memory function.
Constraints and limitations to the model
The most questionable assumption underlying the
model is that following a motion, everything is lost
but the transverse magnetization. This will be
discussed in a forthcoming article.
A key assumption in the formulation of the model
is that the shape of the rigid-state FID is not
modified by a motion, which implies that the
motions do not change the form of the spin-spin
interactions. This is only possible if the nuclear
environment of each individual spin is the same
after as before a motion, the only change being
that it is not the same nuclei that occupy the same
relative sites. This case corresponds to atomic or
molecular motion induced by the diffusion of
vacancies at low concentrations in a single crystal.
One must make the distinction, first introduced by
Eisenstadt and Redfield (10), between a jump and
an encounter. A given portion of the crystal
experiences the sudden arrival of a vacancy, which
performs many jumps before disappearing for
away (This is the standard expression. It is evident
that it is the atoms, or molecules, that jump into
vacant sites.). The whole process is sudden, in the
sense that it takes place in a time too short for the
spin system to undergo any significant evolution.
That portion of the crystal was vacancy-free at the
end of the encounter, so that the form of the spinspin
interactions is indeed the same. The motions
referred to in describing the model are in fact
encounters, and not individual jumps.
In a powder, each crystallite obeys the relation
[15], but with different rigid-lattice memory
functions Kr(t). The motion rate X being
independent of crystal orientation, we have on the
average:
and:
[151
[14 1 ]
These relations being linear, it seems that we might
use powder samples, and not solely single crystals
to test the model. Unfortunately, this is not the ,
case because all one can observe in a powder is the
average FID (or average absorption signal). We
have, in place of Eq.[8]:

Vol. 14, No. 1-4
This is a non-linear relation, and we have in
general:
[16]
It is therefore impossible to test the model with
powders.
Experimental study
We have tested the model with a single crystal of
Hexamethylethane (HME) which is the most
symmetrical octane molecule.
Single crystals of HME are obtained from the
liquid state by the Stockbarger method. Due to the
high vapour pressure of the solid, the crystal is
enclosed in a sealed tube.
This molecular crystal experiences a first-order
phase transition at 152.5 K whereby the molecules
undergo rapid reorientation up to the melting point
at 374 K. As a consequence, the inter-molecular
dipolar interactions average to zero, whereas the
average dipolar interactions between protons of
.1
129
different molecules are the same as if each proton
was located at the centre of gravity of its
molecule. These centres form a body centred cubic
structure with a unit cell of size 7.69 A.
NMR relaxation measurements reveal the
existence in this phase of a thermally activated
translational diffusion of the molecules through the
motion of vacancies (11). Analysis of these
measurements yields the value of the average time
x between successive motions, as a function of
temperature (12).
The proton FID's were observed at 91 MHz with a
home-made pulse spectro-meter. We have used
0.5 (is pulses with a repetition time of 0.3 sec. The
FID signal were sampled in 2048 channels with a
time of O.3(xs per channel. The FID signals were
extrapolated back to the origin through a fourthorder
Taylor expansion approxi-mation:
4!
The main test consists in checking whether the
model is actually able to account for the
variation of the FID shape with temperature. This
is done with the help of Eq.[6] for z = i(0. The
results of the fits are given Figure 1.
100 200 300 400 500 600*
time (jxsec)
figure 1: Free Induction Decay Signals at different temperatures: open circles: 283 K, open squares
292.5 K, full circles: 303 K, full squares : 311.3 K. Solid curves are calculated with our model.

130 Bulletin of Magnetic Resonance
The only fit parameter is X,. The values of X are
plotted as a function of 1/T in figure 2, together
with values of 1/T taken from ref.(12) . It is seen
that both X and 1/x obey an Arrhenius law with the
same activation energy.
The ratio (Xi)' 1 ~ 1.6 is compatible with X being
equal to the decay rate of the secular dipolar autocorrelation
function.
2
§
I
1E+06 T
1E+05 •:
1E+04 -:
1E+03 -:
1E+02
3.1 32 3.3 3.4 35 3.6 3.7 3.8
1000 ,„_..
figure 2 : Memory decay rates A. (white circles, present
work) and average jump rates f* (black circles, deduced
fromref. (12)).
Conclusion
The naive strong collision model is remarkably
successful in accounting for the variation of the
FID with motion of intermediate rate. There is
however a surprising and unexplained fact : the
rate X entering eq.[7] is smaller than the average
rate between encounters. This rate X is found
approximately equal to the rate of decay of the
dipolar auto-correlation function, which is well
known from theory to involve several encounters
(13).
References
(1) A. ABRAGAM, The principles of nuclear magnetism,
Oxford University Press, Oxford (1961).
(2) M. MEHRING, High resolution NMR in solids, 2 nd Ed.
Springer-Verlag, Berlin (1983). Appendix G.
(3) A. ABRAGAM and M. GOLDMAN, Nuclear
magnetism: order and disorder, Oxford University Press,
Oxford (1982) chap 1.
(4) N. BLOEMBERGEN, E.M. PURCELL and R.V.
POUND, Phys. Rev. 73 679 (1948).
(5) P. W. ANDERSON and P.R. WEISS, Rev. Mod. Phys.
25, 269 (1953).
(6) R. KUBO and K. TOMTTA, J. Phys. Soc. Japan 9, 888
(1954).
(7) R. KUBO, J. Phys. Soc. Japan 9, 935 (1954).
(8) R. S. HAYANO, Y. J. UEMURA, J. IMAZATO, N.
NISHJDA, T. YAMAZAKI and R. KUBO, Phys. Rev. B20,
850 (1979).
(9) H. MORI, Prog. Theor. Phys. 33,423 (1965).
(10) M. EISENSTADT and A.G. REDFTELD Phys. Rev.
132, 635 (1963).
(11) J.M. CHEZEAU, J. DUFOURQ and J.H. STRANGE,
Molec. Phys. 20, 305 (1971).
(12) A.R. BRICHTER and J.H. STRANGE, Molec. Phys.
37, 181 (1979).
(13) D. WOLF, Phys.Rev. B10,2710 (1974).

Vol. 14, No. 1-4 131
1. Introduction
THE EFFECT ON Ti OF CORRELATED WATER
MOTIONS IN THE POLAR PHASE OF COLEMANITE
Proton magnetic resonance in the ferroelectric
colemanite, CaB3O4(OH)3.H2O, is dominated by
the dynamical motion of the water molecules.
Absorption lineshape [1] and spin lattice
relaxation studies [2,3] have shown that in the non
polar phase (above 270 K) the water molecules
undergo two kinds of reorientational motion; a
180° flipping motion about the H-O-H bisectrix
which merely exchanges the hydrogen positions,
together with a jumping motion in which one of
the water hydrogens takes up one of two possible
sites. This latter motion is also accompanied by a
jump of some of the hydroxyl hydrogens between
two possible sites.
H',
H'84(O —
^N 2.16 A
J. Sun and A. Watton
Dept. of Physics and Astronomy,
University of Victoria
Victoria, BC
Canada V8W 3P6
He*
l g- 1: Projection along the b axis of colemanite
mowing the relative configuration of a pair of
[ater molecules and neighbouring hydroxyl
jrorogens. The smaller water hydrogen -
f?S? yl hydrogen distances, which make a
111 d Dt contribution t0 the relaxation, are
In the polar phase, as illustrated by the
hydrogen positions in Fig. 1, the jumping motion
freezes out accompanied by a slight
rearrangement of the heavy atom network [4]-
The predominant lineshape and relaxation
mechanism in this phase is the remaining 180°
flipping motion of the water groups. This
accounts very well for most of the observed
features of the absorption lineshape and BPP type
temperature dependence of the spin lattice
relaxation time (Fig. 2). However there still
remains a small discrepancy between the
0.13±0.01 s value observed for the Ti minimum
at a Larmor frequency of 30 MHz and the 0.09 s
value predicted by this model.
T (K)
500 300 200 150 125 100
100
i r
0.1
Fig. 2: Temperature dependence of the proton
spin lattice relaxation time in powdered
colemanite at a Larmor frequency of 30 MHz.

132
Although experimental values for the Ti minimum
smaller than model predictions can usually
be accounted for by additional mechanisms not
included in the model, this is clearly not the
situation here and the interpretation is not so
straightforward. It appears that the flipping
mechanism in the theoretical model, assuming this
motion is an appropriate one, is not as effective as
it should be. One possible reason for this reduced
efficiency is that since the water molecules occur
in adjacent pairs their flips are correlated with
each other. That is to say that a flip of one water
molecule is accompanied by a simultaneous flip
of the other one in the pair. Such a correlation
would result in the number of intermolecular
configurations available to the proton pairs being
reduced over those available in uncorrelated
motion. It would seem reasonable then that the
corresponding reduction in the fluctuations of the
intermolecular dipole interaction could lead to a
reduction in relaxation time. We have therefore
extended the theoretical model for water flipping
to include correlations within adjacent water pairs.
The goal was to see whether such correlations
would lead to an increase of the theoretical Ti and
if so whether this could, by itself, account for the
larger value observed.
2. Model Calculation
The spin lattice relaxation time resulting from
the reorientations of the water molecules is given
by [5,6]
£ = f y 4 * 2 X
where the J(co)'s represent the spectral density of
the fluctuating dipolar interaction between protons
i and j, and coo is th e Larmor frequency.
In the formalism which we will adopt the water
dynamics are described by a dimensionless
transition matrix [7]
Y--A-
" 0
where the matrix A has elements Xjj, i^ j, which
are the probabilities/unit time of the interproton
vector ?j- making a transition to r • among the n
possible position vectors (k[[ is chosen so that
Bulletin of Magnetic Resonance
In a similar way the proton dipolar interaction is
described by a geometrical matrix A with
elements given by
l-3cos 2 a;,-
3 3
where r^, r • =1 r^ I, I r • I and ay is the angle between
r- andr: .
X has eigenvalues XJ and is diagonalized with a
matrix T.i.e.
jj
If A is then transformed by the same matrix T,
AT = TAT~ l
the spin lattice relaxation time can be written as
where
M(©0)= X -^
In a 180° flip of a water molecule the
intramolecular proton-proton vector merely
changes sign leaving its contribution to the dipolar
interaction invariant and hence contributing
nothing to the relaxation. As a result only the
intermolecular couplings need be considered, of
which there are two dominant kinds; the
interaction between water pairs and the waterhydroxyl
interaction.
In uncorrelated water motions each water-water
interproton vector can occupy the four positions
illustrated schematically in Fig. 3 and labelled,
arbitrarily 1,2,3 and 4. The transition matrix is
then
-2
0
1
0
-2
1
1
1
-2
1 1 0 - 2
where X is the probability/unit time of any water a
molecule, all assumed dynamically equivalent, jf
making a 180° flip. However, if the water -
motions are correlated in the sense that the water
molecules in each pair reorient in unison then the
transition matrix for the same inter-water proton
vectors becomes
1
1
0

Vol. 14, No. 1-4
"-1 1 0 0"
1 - 1 0 0
0 0 - 1 1
0 0 1-1
Fig. 3: Schematic representation of a pair of
water molecules showing the four positions
occupied by each inter proton vector between
water molecules during uncorrelated motions.
The choice of label numbers is arbitrary.
In either case the water-hydroxyl interproton
vectors have only two values for each waterhydroxyl
group and their corresponding transition
matrix is given by
In fact, for correlated water motions the waterwater
interproton vectors can be partitioned into
two sets {1,2} and {3,4} each of which has a
transition matrix of the simple 2x2 form above.
Applying the diagonalization and transformation
^procedure described previously to these matrices
^produces the following expression for Ti resulting
ifrpm correlated 180° flips of the water molecules,
= 7.0
4coQr
where x=l/\ is the correlation time for the 180°
up motion.
his gives a Ti minimum of 0.10s for a Larmor
fluency of 30 MHz which is greater than that
^"jcorrelated motion by about 10% but still
nilicantly short of the 0.13s observed. It
133
appears that at least the simple correlated model
considered here is inadequate to fully account for
the minimum in Ti and that some other
mechanism is presumably in effect.
In colemanite, as shown by Fig. 1, some
hydroxyl hydrogens are quite close to the water
molecules and there are only two water hydrogens
for every three hydroxyl hydrogens in each
molecular unit. As a result the hydroxyl-water
coupling and the water-water coupling make
comparable contributions to the proton relaxation.
Since the hydroxyl-water contribution is
independent of any correlations in the water
dynamics it seems reasonable that there is not a
large difference in the overall Ti between the
correlated and uncorrelated motions, but it is
gratifying that the change is in the right direction
and is consistent with the reduction in the dipolar
fluctuation discussed earlier.
References
1. A. Watton, H.E. Petch and M.M. Pintar,
Can. J. Phys. 48, 1081 (1970)
2. R. Blinc, M. Brenman, S.R. Miller and J.S.
Waugh, J. Phys. Chem. Solids 23_, 156
(1962)
3. A. Watton, H.E. Petch and M.M. Pintar,
Can. J. Phys. 51, 1005 (1973)
4. F.N. Hainsworth and H.E. Petch, Can. J.
Phys. 44, 2083, (1966)
5. N. Bloembergen, E.M. Purcell and R.V.
Pound; Phys Rev. 73, 679 (1948)
6. A. Abragam, The Principles of Nuclear
Magnetisim (Oxford University Press, New
York, 1961)
7. A. Watton, Phys. Rev. BI7, 945 (1978)

134 Bulletin of Magnetic Resonance
MEASUREMENT OF DEUTERON SPIN RELAXATION
TIMES IN LIQUID CRYSTALS by a BROADBAND
EXCITATION SEQUENCE
I. Introduction
Ronald Y. Dong
Department of Physics and Astronomy, Brandon University
Brandon, Manitoba R7A 6A9
Liquid crystals are composed of flexible organic
molecules and capable of forming different
ordered structures in their mesophases. Nuclear
spin relaxation [1], [2] is a powerful technique
that provides useful information on the
molecular dynamics of liquid crystals. There
are collective director fluctuations, molecular
reorientation and internal rotations in flexible
end chains. Recently internal dynamics
of mesogenic molecules has attracted much attention.
Both theoretical [3], [4] and experimental
[5], [6] studies have been carried out.
Experimentally carbon-13 and deuteron may
be used to probe internal dynamics of flexible
mesogens. In aligned samples of deuterated
liquid crystals, deuterium NMR spectroscopy
yields well-resolved spectral lines having different
quadrupolar splittings for various atomic
sites. These quadrupolar splittings result from
incomplete averaging by anisotropic reorientation
of molecules in the mesophases. For a single
deuteron spin (1=1), there are five independent
spin relaxation times [7]. These are two
spin-lattice relaxation times and three independent
spin-spin relaxation times. Since the deuterium
Zeeman (T\z) and quadrupolar (T\Q)
spin-lattice relaxation times are given by [7],
[8]
rp -l
~3Ji(u;0),
4J2(2w0)
they can be used to separate the two spectral
densities of motion Ji(wo) and J2(2u>o)- Accurate
determination of these spectral parameters
as a function of temperature and the Larmor
frequency (u>o) is necessary for testing various
motional models.
Both T\z and T\Q can be simultaneously measured
by the Jeener-Broekaert (J-B) method [9]
or the selective-inversion method [5]. However a
separate experiment has to be performed for the
deuterons on each labelled site in order to maximize
their quadrupolar order for better signalto-noise
considerations. The J-B sequence has
been used to determine T\z and T\Q in several
nematogens [10]. The pulse sequence was modified
using an additional 45° pulse to produce
the net effect of subtracting the equilibrium Moc
signal from the J-B signal. Here we examine the
modification of the J-B sequence (Table 1) to
produce [11] a broadband excitation sequence
(Table 2) in order to minimize the number of
separate experiments required to give T\z an< l
T\Q for various deuterons on an alkyl chain. Recently
this broadband J-B excitation sequence
was vised to create quadrupolar order with same
efficiency on all the labelled sites in a liquid -j
crystal [12].

Vol. 14,i No. 1-4
\
X
-y
X
-y
X
-y
X
-y
TABLE 1
J-B Sequence with Phase-cycling
Receiver
7 3 Aqu T 2 4>s Aqu T Aqu T Phase
-y y y X +•
0
-X X X y
90
-y y y -X
0
-X X X -y +
y +
90
90
X
0
-y
90
-x +
0
-x +
0
-y
90
X
0
y + 90
y -y -y
90
X -X -X
0
y -y -y
90
)X -X -X
0
-y +
-X
y
x +
2. Experimental
; The deuterium Txz and T\Q were measured on
a home-built superheterodyne coherent pulsed
|INMR spectrometer operating at 15.3 and 46.05
^MHz for deuteron with a Varian 15 in electromagnet
and a 7.1 Tesla Oxford superconductmagnet.
The TT/2 pulse width of ca. 4.5 //s
^produced by an Amplifier Research Model
power amplifier. Pulse control, signal coltiou,
Fourier transformation and data pro-
«mg were performed using a General Electric
1280 computer [10]. Both the J-B sequence and
the broadband J-B excitation sequence (Figure
1) were used with the appropriate phase-cycling
[9] of RF and receiver phases to rid of the unwanted
double quantum coherence (see Tables 1
and 2). Several nematogens (5CB, MBBA and
60 CB) were employed to check the spin relaxation
times obtained by the two different multipulse
sequences.
135
3. Results and Discussion
In figure 2 we show a comparison of the
J-B sequence (2(a)) and the broadband J-B
excitation sequence (2(b)) for a set of partially
relaxed spectra at 15.3 MHz in the nematic
phase of 4-n-pentyl-dn-4'-cyano-2,3,5,6d4-biphenyl
(5CB-di5). Minimal phase correction
was required and the baseline of each spectrum
has been corrected. In figure 2( a) the J-B
sequence was set to maximize the quadrupolar
order of the C4 methylene deuterons. In comparison
with the J-B sequence, we found that
maximum quadrupolar order was created for all
the chain deuterons with r = 5/j.s (or an excitation
bandwith of ca. 75 kHz). In table 3 we
summarize the T\z and T\Q measured at 33.2°C
in 5CB-di5 by the two different J-B sequences.
As seen in this table, their values agree with
each other for all the labelled sites within experimental
errors.
a)
b)
90, 67.5, 45, 45,
9 ^ 4>

136 Bulletin of Magnetic Resonance
a)
1 32 4 5R
R>-c5oll
A 70m ^ h Hijft A, A M A 70m
.50 ft W iiii A A I il 60
A 45
Figure 2 Plots of partially relaxed spectra at 33.2°C and 15.3 MHz. (a) Using the J-B sequence, (b) Using the
broadband J-B excitation sequence.
b)
a)
12 34 5
0 0 DO
6R
R
_ ! , , 1 1 , 1 1 1 r
40000 20000
= 36.2°c
, . 1 1 . r—j-
0 -20000 -40000 Hz
Figure 3 (a) A typical deuterium NMR spectrum of 60CB-d2i showing the peak assignments; (b) A schematic
diagram of a 60CB-d2i molecule.

Vol. 14, No. 1-4 137
>N
CO
c
0)
D
o
20
15 -
10 -
0
o J1
o
oJ2
o
•o
AA J
AA
J 2
AA AA
0
O #)
A Ai
A j^\
t
o
A A A
A A A
40 50 60 70
T(°C)
.We 4 Plots of spectral densities versus temperature in the nematic phase of 60CB-d2i. Open symbols are obtained
^ e broadband J-B excitation sequence, while closed symbols by the J-B sequence. O and A denote data for C4
[a '^6, respectively.
1

138
TABLE 3
Comparison of T\ z and Tiq in ms measured
at 15.3 MHz and 33.2°C
for 5CB-d^
12.6
(13.5)
10.4
(10.1)
c2
26.9
(25.8)
21.5
(21.8)
c3
30.0
(30.7)
22.8
(26.4)
ct
55.4
(48.9)
51.9
(46.1)
159
(137)
107
(104)
Ring
8.5
(9.1)
10.9
(11.4)
* Ti values in parentheses were obtained by the J-B sequence,
while those without parentheses were obtained by the broadband
J-B excitation sequence.
Figure 3 shows the molecular structure of
60CB-d2i and a typical deuterium NMR spectrum
for this mesogen at 15.3 MHz. We have
used both pulse sequences to measure spectral
densities JI(JJJQ) and J2(2u;o) for all the labelled
sites except the ring R!, because of excessive
overlapping of its signal with that from the
methyl (C6) at high temperatures. The agreement
between the two methods are extremely
good. As an example, we show in figure 4
plots of spectral densities versus temperature
for C\ and C&. Since the relaxation times of the
methyl deuterons are much longer than those of
the ring R! deuterons, the overlapped doublet
signals can still be used to determine T\z and
T\Q for the methyl deuterons as long as t is chosen
larger than 40 ms in the pulse sequence.
Finally MBBA-di3 has been studied at 15.3
MHz using the J-B sequence [10]. For comparison
we summarize in Table 4 the results obtained
using the broadband J-B excitation sequence
at 26°C and at 15.3 and 46 MHz. Thus
both Ji(wo) and J2(2u>o) show frequency dependence.
The frequency dependence of J2(2w0) is
weaker; it is negligible for the methine deuteron
(Co). Currently we are analyzing the temperature
and frequency dependences of the measured
spectral parameters in MBBA using models
[3], [4] proposed for flexible mesogens.
In conclusion, the relaxation data can be effectively
obtained in liquid crystals by using the
broadband J-B excitation sequence.
Bulletin of Magnetic Resonance
TABLE 4
Measurements of Ji(u>o) an d ^2(2wo) ins ' for
MBBA-di3 at 26°C (u>o = Larmor frequency
in MHz) using broadband excitation
15.3 46
Co 49.8
40.7
14.6
13.7
1.73
36.2
21.0
9.0
7.5
1.58
15.3 46
Acknowledgments
24.9 27.7
15.1 10.2
6.7 5.4
5.1 4.0
1.05 0.98
The financial support of the Natural Sciences
and Engineering Council of Canada is gratefully
acknowledged. We thank Ms. L. Friesen for her
assistance in carrying out some experiments.
References
1. C.G. Wade, Annu. Rev. Phys. Chem. 28,
47 (1977) and references therein.
2. R.R. Void, in "Nuclear Magnetic Resonance
of Liqud Crystals" J.W. Emsley,
Ed., Reidel, Dordrecht (1985).
3. A. Ferrarini, G.J. Moro and P.L. Nordio,
Liq. Cryst. 8, 593(1990).
4. R.Y. Dong, Phys. Rev. A 43, 4310 (1991).
5. P.A. Beckmann, J.W. Emsley, G.R. Luckhurst
and D.L. Turner, Mol. Phys. 50, 699
(1983); 59, 97 (1986).
6. R.Y. Dong and G.M. Richards, J. Chem.
Soc. Faraday Trans. 88, ni the press.
7. J.P. Jacobsen, H.K. Bildsor and K. Schumburg,
J. Magn. Reson. 23, 153 (1976).
8. S.B. Ahmad, K.J. Packer and J.M. Ramsden,
Mol. Phys. 33, 857 (1977); R.R. Void
and R.L. Void, J. Chem. Phys. 66, 4018
(1977).
9. R.L. Void, W.H. Dickerson and R.R. Void,
J. Magn. Reson. 43, 213 (1981).
10. R.Y. Dong and G.M. Richards, J. Chem.
Soc. Faraday Trans. 84, 1053 (1988) and
references therein.
11. S. Wimperis, J. Magn. Reson. 86, 46
(1990).
12. G.L. Hoatson, J. Magn. Reson. 94, 152
(1991).

Vol. 14, No. 1-4 139
Introduction
CARBON-13 RELAXATION MECHANISMS AND MOTIONAL STUDIES
The carbon-13 relaxation rates of several symmetric
top halomethane species, CH3I [1], CClBr3
[2], CHBr3 [3] and three others that have been
shown to be quasi-symmetric top, CH2Br2 [4,5],
CH2C12 [6], CH2I2[7] were determined. These
results were evaluated at two field strengths,
2.1 and4.7Tesla, at UNC-Wilmington and East
Carolina University, respectively, and at an identical
set of temperatures at each site. With
these data and several theoretical models we
were able to determine, or calculate, the contributions
for all plausible relaxation modes.
IN SELECTED HALOMETHANE MOLECULES
Art A. Rodriguez , Tim Davis
Stokes-Einstein-Debye (SED) [8,9] theory of
rotational diffusion and several variants to characterize
the anisotropic reorientation of spheroids
were also used to investigate for goodness of fit
for hydrodynamically controlled rotational motion.
In the hydrodynamic model, terms called
"stick " and "slip" that attempt to describe
the involvement of probe molecules and the
solvent are exploited. The stick limit is normally
encountered where the solute molecular radius is
'•much larger than that of the solvent, while the
^slip condition is approached as solvent radius
lears or exceeds that of the solute molecule. The
'-extended diffusion model is used to deterinertial
properties of rotational diffusion
Department of Chemistry, East Carolina University
Greenville, NC 27858, USA
and
Lewis E. Nance.
Department of Chemistry, UNC-Wilmington
Wilmington, NC 28403, USA
[4] from J-diffusion to free rotation of the molecule.
Separation of R, 510 from R/ 01 allowed cal-
culation of Tin, Q for Brand 'Jr.Rr for CClBr,,
IBr
C-Br
CH2Br2, and CHBr3 by use of plots of derived
sc 2
T^vs. -s.A(O .
Experimental Section
Separation of Relaxation Mechanisms
Tj values were obtained by the inversion recovery
method using (Mo,cos6, T\), a three parameter
fit for magnetization, M(x), as shown
below.
M{x) = Mo [ 1 - (1 - cos G) exp(-T/ri)] (1)
Partitioning of relaxation rates into contributions
by specific mechanisms was generally
made with the following set of relationships in
mind: When present, contribution to the relaxation
rate by the scalar mechanism of the "second
kind" (SC) is greater at 2.1 T than at 4.7
T while the reverse will be true for chemical
shift anisotropy (CSA). Relaxation is more efficient
at higher temperatures for SC while it is
less so for CSA. Dipole-dipole (DD) relaxation
is more rapid at lower temperatures while that of
spin-rotation (SR) is less. These contributions

140
can be summed as follows where R, tot is the experimental
rate of relaxation:
JL
tot
CSA SC
jCSA 4.
(2)
Integration of the C 13 signals with and without
decoupling (pulse delay set to a full 10 Tt
[10]) for the protonated species, CH3I, CH2I2,
CHBr3, CH2Br2, and CH2C12 permitted determination
of NOE values, r\, and allowed direct
calculation of the H contribution to R, DD [10] by
the following relationship.
The evaluation of %C{C-H) [11] follows immediately
from equation (4).
(3)
XC(C-H) = (y 2 cfHh 2 /2ii 2 r 6 CH)R° D {C-H) (4)
The dipole-dipole contribution to C fromBr
can then be calculated from the relationship below
[12].
•)DD, •>DD<
R" U (C-H)IR\ )U {C-Bf) =
2 (rCH) 6
(5)
A value for the quadrupole coupling constant,
(2ne 2 Qq/h), orT, Q for Br allows a calculation
of xc(C-Br) in equation (6) [13] or (7) [14] respectively.
Rr~-
3 2/+3
40
If
(6)
(7)
R, DD subtracted from R, tot leaves R, 0 *" to be
partitioned among the other contributions. The
correlation time, xc, that was acquired from
R, DD (C-H)canbe used in calculating Ri CSA by
Bulletin of Magnetic Resonance
equation (8) [12] if aper and cpar, the perpendicular
and parallel components of the sheilding
tensor, are known.
(8)
A classic case for relaxation by the CSA
mode was that of CH3I [1]. We found that the
CSA component to relaxation was significant
since Rj tot was much greater at 4.7 T than at 2.1
T and that this value decreased with an increase
in temperature. Using the relationship
CSA _ CSA
(9)
we utilized the fact that the variance in Rj tot
going from 2.1 to 4.7 T must be due to the CSA
mechanism since the SR mechanism is independent
of field strength. A plot of R, tot vs. v 2 at
303K has at the intercept R, tot = R, 1 and at this
point the contribution of R, CSA to the value of
R, tot vanishes. These results show the R, CSA
contributions at 303K of 7.52 x 10" 3 (T, CSA =
133s) at 4.7 T and 1.52 xlO" 3 (T, CSA =659s) at
2.1 T.
The Tc(C- H) values were also employed in
conjunction with the J-extended diffusion theory
computer program by McClung to calculate Xj,
and its reduced form, x}, for symmetric tops
[15,16,17 ]. The equation for R, is as
follows [12]:
Rf t = (2KlkT/h 2 )C 2 effxJ
(10)
A second program written by McClung relates
R, SR to xj. This program requires values
for Ix,Iz,Xj and the spin-rotation tensor components
, Cx and Cz, which are not generally available.
These spin-rotation components can
be approximated by the method of Flygare [18],
which uses the facts that op( n CO) = -259.5 |
ppm and 8( 13 CO) - -182.2 ppm, and that the |

Vol. 14, No. 1-4 141
difference between Adp and A5 for n C0 and
the particular halomethane of interest, for
instance, (Aap = ap(halomethane) - ap(CO),
etc.) are the same . In general one must assume
that Ac is zero. In this manner the relationship
IZCZ ~ IXCX holds if Aa is small compared to
aP.
Scalar relaxation of the "second kind" is expected
to be a prevalent relaxation mechanism in
the Br bearing halomethanes. It is predominantly
the 79 Br isotope instead of 81 Br that makes
the greatest contribution to relaxation of C-13
[19,20]. This is due to the fact that the
(G)/-G)S) term in equation (7) is smaller for
79 Br. Scalar coupling is greater at 2.1 T than at
4.7 T since the Aw term is smaller in the lower
magnetic field. R, sc also increases with temperature.
A plot of Aco 2 vs. T, sc gives
rrQ _
IBr = Jm/b
and
J= \)m)
(11)
(12)
where m and b are the slope and intercept, and
N is the number of Br atoms attached to the
halomethane carbon atom.
Rotational Motion
In the limit of extreme narrowing , the small-step
diffusion theory [13,21,22] predicts the relationship
between xc and the diffusion constants
Dx and Dz in a symmetric top environment by
the following equation [4]
X - °- 3sin 2 6cos 2 e 0.7Ssin 4 6
5DX+D2 2
where 8 is the angle between the reorienting
vector and the unique axis of rotation. This is
the axis with the lowest moment of inertia. For
example, this is the C3 axis in CH3I, but not in
CHBr3 where the C-H vector is at 90° to this
unique axis.
( We have applied the J-extended diffusion
.theory of Gordon [23], as expanded to symmet-
ric tops byMcClung [15, 16], an inertial model,
to generate Xj values using our previously
determined xc(C-H), or xc, parameters. In this
model molecules undergo a period of free rotation
generating angular momentum, then upon
hard molecular collision randomize both magnitude
and direction of this vector. In the smallstep
diffusion limit, Xj « xc • Here the reduced
correlation time, x}, is found to obey the
following equation:
t} = Ty(^-)T«l (14)
In addition, under these conditions, the rotational
diffusion constants, DzandDx of equation (13)
are related to Xy by
Dz = Z-Xj and Dx = 4H (15)
In the inertial model Xy = xc. The period of
time for rotation through 1 radian, X/, as determined
by the equipartition principle is given as
[24]:
y=(~)^ (16)
Thus one is able to follow with this program
by McClung the limits from small-step diffusion
to free rotation where the rotations are inertially
controlled.
At the other extreme, molecular shape instead
of inertial effects may dictate the mode of rotational
diffusion of a molecular species. Extension
of the SED theory to prolate and oblate
spheroids has produced many equations as
boundary conditions for stick and slip rotational
motion are considered for Dx and Dz. For
instance, Perrin [25] was able to show a relationship
be tween the Stokes diffusion constant,
Ds, by solving the Navier-Stokes equation
assuming a stick boundary condition.
Dt = A-Ds = A. kT (17)
The factors fPiX and fPtZ are functions of p = bla
(1 for oblate spheroid). The

142 Bulletin of Magnetic Resonance
average molecular radius is used for r and the
bulk viscosity for T\.
Hu and Zwanzig [26] tackled the rotational diffusion
problem by assessing the fact that the
Perrin values , using the stick boundary condition,
did not fit the experimental work well.
They instead assumed a slip boundary condition
in solving the Navier-Stokes equation. A
separate value for a friction coefficient, £*,
was obtained and tabulated for each axis ratio
for prolate and oblate spheroids. Under this
treatment, motion parallel to the top axis would
experience no tangential stress and would have
the motion of a free rotor.
180
(18)
There would be some solvent displacement possible
about the x-axis for perpendicular rotational
motion.. The expression used here is
Dx = ±LDs
(19)
where in our work /#z = ^*/8.
Gillen [27] and Griffiths [28] note that the
Dx motion in some molecules is diffusionally
controlled and as such may be treated by
the Gierer- Wirtz micro viscosity model [29]. The
parallel motion, Dz, is treated in the slip boundary
conditions, as essentially frictionless rotation.
(20)
The Gierer-Wirtz factor, fGw, is 0.1633 for neat
liquids.
Tanabe [30, 31] has extended the Hynes, Kapral,
Weinberg (HKW) model [32] to include
nonspherical molecules in solution and in neat
form.
(21)
The a factor is zero for Dz but is fitted to a Hu-
Zwanzig coefficient for Dx since even in the
case of slip ((3=0) there is a finite friction coefficient.
Results
In CH2Br2 [4, 5] and CHBr3 [3 ], DD and SC of
the second kind were shown to be the important
relaxation modes. The dipolar contribution to R,
fell off as the temperature was raised, while the
scalar coupling rate increased. There was a
larger value forRj at any particular temperature
at 2.1 T than found at 4.7 T. R, SR and R, CSA were
calculated and found to be negligible . The dipole-
dipole relaxation contribution from Br to
C-13 was calculated by equation (5) and was
much less than the experimental error. In CQBr3
[3 ] the relaxation results were clearly scalar.
Average values of l JcBr and T\Br for CQBr3,
CH2Br2, and CHBr3 are respectively: 75 Hz and
4.3 x 10 6 ; 53 Hz and 4.3 x 10' 6 ; 49.7 Hz and
8.03 x 10" 7 .
The rotational diffusional motion of CH2Br2
andCHBr3 as treated by the several models
above, matches very closely that predicted by
the J-extended diffusion model.
The trend towards faster relaxation rates in
CH3I [1] at 4.7 T compared to 2.1 furnishes positive
evidence for the importance of CSA rather
than SC in this molecular species. SRandDD
are also found to significant.
The approach taken here was that of Gillen [27]
and Griffiths [28]. Motion about the top axis in
CH3I fit closely the Gierer-Wirfz microviscosity
model with an average difference of less than
1% compared to experimental results. The excellent
correlation indicates that the tumbling motion
is hindered by viscous drag while the motion
about the top axis is dominated by inertial
effects.
The modes of relaxation in CH2I2 [7] were as
follows: DD; the dominant mechanism, SR; a
very small contribution, and SC; about 18% at all
temperatures.
The J-extended diffusion model was a good
predictor of rotational motion for both CH2I2
andCH2Cl2[4,6].

Vol. 14, No. 1-4 143
References
1. Manuscript in preparation.
2. Submitted to /. ofMolec. Spectroscopy.
3. Manuscript in preparation.
4. D. N. Dixon and A. A. Rodriquez, J. of
Molec. Liq. 44, 79 (1990).
5. P. B. Simcox, A. A. Rodriguez, L. E, Nance,
The J. of Physical Chem. 96, (1992).
6. A. A. Rodriguez, S. J. H. Chen and M.
Schwartz,/. Magn. Reson. 74, 114 (1987).
7. L.E. Nance, M. R. Nealey, and A. A.
Rodriguez, Magn. Reson. Chem. 28, 11
(1990).
8. R. T. Boere 1 and R. G. Kidd, Annu. Rep.
NMR
Spectrosc, edited by G. A. Webb, Academic
Press, New York, 13, 319 (1982).
9. P. Debye, Polar Molecules, Dover, New
York 1929.
10. M. L. Martin, J.-J. Delpuech and G. L.
Martin, Practical NMR Spectroscopy,
Heyden, London (1980).
11. T. C. Farrar,and E. D. Becker, Pulse and
Fourier Transform NMR, Chap. 4,
Academic Press, New York, (1971).
12. E. D. Becker, High Resolution NMR:
Theory and Applications, 2nd ed., Chap. 8.
Academic Press, New York (1980).
13. W. T. Huntress, J. Chem. Phys. 48, 3524
(1968).
14. A. Abragam, Principles of Nuclear
Magnetism, Chap. 8. Oxford University
Press, Oxford (1961).
15. R. E. D. McClung, /. Chem. Phys. 57, 5478
(1972).
16. R. E. D. McClung, Adv. Mol. Relaxation
Interact. Processes 10, 83 (1977).
17. R. E. D. McClung, J. Chem. Phys. 51, 3842
(1969).
18. W. H. Flygare, J. Chem. Phys. 41,7903
(1964).
19. C. R. Lassigne and E. J. Wells, J. Magn.
Reson. 27, 215 (1977).
.20. T. C. Farrar, S. J. Druck, R. R. Shoup and E.
D. Becker, J. Am. Chem. Soc. 94, 669
(1972).
1- H. Shmizu, J. Chem. Phys. 40, 754 (1964).
22. D. E. Woessner, J. Chem. Phys. 37, 647
(1962).
23. R. G. Gordon,/. Chem. Phys. 44, 1830
(1966).
24. George C. Levy, Topics in Carbon-13 NMR
Spectroscopy, Vol 1, Chapter 3,
Wiley-Interscience, New York, (1974).
25. E. Perrin, /. Phys. Radium, 5,497 (1934).
26. C. M. Hu and R. Zwanzig, J. Chem. Phys.,
60 4354 (1974).
27. K. T. Gillen and J. E. Griffiths, Chem. Phys.
Lett., 17 359 (1972).
28. J. E. Griffiths, Chem. Phys. Lett., 21 354
(1973).
29. A. Gierer and K. Wirtz, Naturforsch, A8
532 (1953).
30. K. Tanabe, Chem. Phys., 31 319 (1978).
31. K. Tanabe and J. Hiraishi, Molec. Phys., 39
493 (1980).
32. J. T. Hynes, R. Kapral and M. Weinberg, /.
Chem. Phys., 69 2725 (1978).

144
An Efficient Large Sample Volume System
for Solid State NMR
Ronald J. Pugmire, Yi Jin Jiang, Mark S. Solum and David M. Grant
Department of Chemistry and Fuels Engineering
University of Utah
Salt Lake City, Utah 84112
ABSTRACT
U.S.A.
An efficient large sample volume
system has been developed to carry out
MAS solid state NMR experiments. The
system components are primarily
zirconia and macor and no background
1 3 C is observed. The stator design
employs separate air bearing and drive
systems and is run using dry air. At a
bearing pressure of about 32 psi, the
rotor can be spun in a stable manner
from less than one hundred Hz (with a
driving pressure of 5 psi) to 4.3 KHz (24
psi driving pressure). This low gas
pressure feature makes the system easy
to operate. The volume of the rotor is
1.8 cm 3 and it can hold 1.1 g of HMB.
The S/N ratio obtained is a factor of 4.6
better than the rotor previously
designed and used in our laboratory
(volume 0.6 cm 3 : 0.28g HMB). This
increased sample size allows us to obtain
the same S/N ratio in a MAS spectrum
with a factor of 21 saving in
spectrometer time. The time saving
achieved with this rotor system is
extremely useful in obtaining data on
biological samples and polymers, and is
especially useful when experiments on
fossil fuels require the use of the Bloch
delay technique. Examples of relevant
applications will be discussed.
Bulletin of Magnetic Resonance
INTRODUCTION
A large MAS spinner system has been
designed to achieve spinning speeds
from less than a hundred Hz to
approximately 4.3 KHz with a relatively
large sample volume of about 1.8 cm^.
The reason for employing a large
sample volume was to improve the S/N
ratio in l3 c NMR experiments. This *is
extremely important for obtaining
quantitative 13 C Bloch decay (BD)
spectra on coal samples with an
acceptable amount of spectrometer
time, due to their reasonably long T\
relaxation times.
In order to avoid a * 3 C background
signal the spinner and stator system
are constructed from zirconia and
macor respectively both of which
contain no carbon.[l] Attention has
also been focused on increasing the S/N
ratio by using a probe electronic circuit
which optimizes the sensitivity of the
l3 C observe channel.[2,3] The result is
a system which gives a S/N ratio for a
* 3 C CP/MAS spectrum of l.lOg sample of
hexamethylbenzene (HMB) that has a
factor of 4.6 improvement over the S/N
ratio of a spectrum taken on our
traditional spinner system, using 0.28g
of HMB and the same spectral
parameters.

t.!-.
* * • •
K
Vol. 14, No. 1-4 145
DESIGN AND CONSTRUCTION OF THE
SPINNER
A cross-sectional drawing of the
spinner system is shown in Figure 1.
There are eighteen flutes in the Up of
the rotor. They are driven by a set of
eight air jets (with a diameter of 0.025
in.) that are mounted on the stator,
along with a set of eight bearing holes
of the same diameter located in the
middle of the stator. These air holes are
spaced at 45' intervals around the
circumference of the stator. The
clearance between the rotor and stator
is approximately 0.002 in. The stator
housing is machined from Kel-F which
has a stepped surface, as shown in
Figure 1, in order to separate the
driving and bearing gases. This design
leads to ease in assembling the system,
and also increases operational safety.
The spinning speed of the rotor as a
function of the driving gas pressure,
with the rotor containing 1.1 Og of HMB
and a bearing gas pressure of 32 psi is
shown in Figure 2. Figures 3 through 5
are a series of photographs of this
system. Due to the high rotational
energy of the spinning system, the
probe must either be in the bore of the
magnet or behind an explosion screen
when in operation.
EXPERIMENTAL RESULTS
Some typical results of data obtained
with the large sample spinning system
are shown below. Figure 7 shows the
CP/MAS 13 C spectrum of 1.1 Og of HMB
taken with only 12 scans and with 21 Hz
line broadening applied. The S/N ratio
of this spectrum is approximately 517.
Figure 8 is a comparison of the spectral
results using this new large rotor and
the standard rotor system previously
used in our laboratory. The size of the
standard rotor (it holds about 0.28g of
HMB) is fairly typical of those
commercially available. In Figure 8a,
CP/MAS spectrum is shown of l.lOg of
HMB consisting of 20 scans with no line
broadening. A S/N ratio of 130 is
obtained. Figure 8b shows a 20 scan
spectrum taken with the smaller
standard rotor system which exhibits a
S/N ratio of 28, again, with no line
broadening. The comparison of Figures
8a and 8b demonstrate that an
improvement in the S/N ratio of about
4.6 can be achieved which provides a
factor of 21 in terms of spectrometer
time. Figure 9 demonstrates that there
are no l^C background signals in BD
and cross polarization (CP) experiments.
This is an important feature in the
quantitative application of CP NMR
experiments. This rotor/stator system
has been used to carefully compare the
BD and CP spectra of the eight coals in
the Argonne Premium Coal Safhple
Bank and these data are presented
elsewhere.[4]
ACKNOWLEDGMENT
This work was supported through the
Advanced Combustion Engineering
Research Center at the University of
Utah and Brigham Young University
which is supported by the NSF, 23
industrial firms and DOE/PETC.
Additional support was provided
through the Consortium for Fossil Fuel
Liquefaction Science at the Univeristy
of Kentucky, Auburn University,
University of Pittsburgh, University of
West Virginia, and the University of
Utah.

1
i
146
REFERENCES
1. Ming Zhang and Gary E. Maciel; J.
Magn. Reson., 85, 156 (1989).
2. Yi Jin Jiang, Ronald J. Pugmire,
and David M. Grant; /. Magn.
Reson.,11, 485 (1987).
3 Yi Jin Jiang, Warner R.
Woolfenden, Anita M. Orendt,
Karen L. Anderson, Ronald J.
Pugmire, and David M. Grant;
Poster MB, 30th ENC, Asilomar,
California, April 1989.
4. Mark S. S,olum, Yi Jin Jiang,
Ronald J. Pugmire, and David M.
Grant, submitted for publication.
Figure 1: Cross-sectional view of the spinner system:
I. Stator, 2. Rotor. 3. Turbine driving jet, 4. Housing
- upper pan, 5. NMR coil terminal plug, 6. Bearing gas
inlet plug, 7. Bearing gas orifice, 8. Light fiber mount.
9. Screw for fixing light fiber. 10. Light fiber.
II. Housing - lower pan, 12. Driving gas inlet plug.
13. Bearing gas exit holes, 14. Mounting nng.
5000
Bulletin of Magnetic Resonance
Driving Gas Pressure (PSD
Figure 2" Plot of the spinning speed of :he
arge volume rotor system as a function ot
he driving pressure used. Bearing pressure
vas 32 psi.
Figure 3: Photograph of the large sample rotor (a) as
compared to the typical rotor (b) and the /mm
rotor supplied by Doty Scientific (c).

if
&
Figure 4: Photograph of the large volume rotor and the
stator.
Figure 5: Photograph of the spinner system.
£' Figure 6: Photograph of the probe containing this
spinning system.
A
B
Figure 7: CP/MAS spectrum of l.lOg of HMB. taken with 12 sc
5ms contact time. Is delay time. 21 Hz line broadening
Spectrum at 25.12 MHz on a Bruker CXP-100.
(b)
CP
2280 scans
lms contact time
1 sec. recycle time
BD
944 scans
180 sec. recycle
delay
CP/MAS spectra of HMB taken at 25.12 MHz:
(t)l«r|e volume (1.8 cm 3 ) system, l.lOg HMB.
20 scans, 3m* contact time. Is delay, no line
broadening, S/N ratio 130; (b) small (0.63 cm 3 )
rotor system. 0.28| HMB, 20 scan*. 3ms contact
time. Is delay, no line broadening. S/N ratio 28.
330 230 130 30
PPM from TMS
-70
Figure »• Comparison of the spectral
Iineshapes obtained for BD and CP
experiments on Pittsburgh #8 Argonne
Premium Coal. A is the spectrum and B is the
background obtained by taking the spectrum
with an empty rotor under identical
conditions to A.
147

148
Bulletin of Magnetic Resonance
Magnetic Resonance Spectroscopic Investigations
of Poly(p-Phenylene Sulfide/Disulfide), PPS/DS
Douglas W. Lowman and David R. Fagerburg
Research Laboratories, Eastman Chemical Company
P. O. Box 1972, Kingsport, TN 37662-5150 USA
Introduction
Recently a new process for the
synthesis of the commercially important
semi-crystalline, engineering
thermoplastic poly(p-phenylene
sulfide/disulfide), PPS/DS, was presented
[1]. This process, based on the reaction of
p-diiodobenzene (DIB) and sulfur at
elevated temperatures (Figure 1),
generates a polymeric material containing
para-substituted aromatic groups
connected by sulfide and disulfide
linkages.
+ S
230 to 300°C
I + S »•
Melt, air bleed
230 to 250°C
• —
Melt, air bleed
Pro-Polymer + '2
PPS/DS
Solid-Stats
Build-Up
(240°C, Nj)
Figure 1. Synthesis of PPS/DS
PPS/DS is not easily characterized by
conventional magnetic resonance
techniques operating under normal
conditions due to its high thermal stability
and solvent resistance. Wade and
coworkers [2] recently reported on their
application of high-temperature solution-
state carbon-13 NMR to commercially
available poly(phenylene sulfide), PPS We
have extended this study by examining
PPS/DS with epr spectroscopy as well as
high-temperature solution-state carbon-13
NMR.
The PPS/DS chemistry is accomplished
under conditions that are in certain
respects considerably less stringent than
those previously employed for other PPS
synthetic routes [3,4]. Free radicals have
been observed in PPS previously by
several workers [5-8] but not under the
conditions of synthesis. Since it was
anticipated that radicals would play an
important role in our chemistry, we
examined this reaction process directly
employing high-temperature epr
spectroscopy under conditions of the
reaction. In this report we present
evidence for sulfur and carbon radical
formation during PPS/DS synthesis.
Experimental
EPR spectra were collected on a Bruker
ER 200D SRC EPR spectrometer operating
on X-band (9.65 GHz) in the general
temperature range of 230 to 300°C and
using 3-mm O.D. glass epr tubes with a 200
gauss sweep width, 100 sec sweep time,
modulation frequency of 100 KHz and
modulation of 40 milligauss. Measurement
of g values was accomplished in a dual
cavity with the resonance of
diphenylpicrylhydrazyl (g value = 2.0037)
as the reference.
PPS/DS synthesis in the epr tube was
accomplished in a manner similar to that

Vol. 14, No. 1-4 149
20 G
Figure 2. EPR spectra of perdeuterated PPS/DS
at (A) 235°C, g(iso) = 2.0073, and (B) 290°C, g =
2.0042.
Figure 3. EPR spectra of melting
perdeuterated PPS/DS
Table 1. Linkage and end group
species found in PPS/DS. Letters
and numbers are used in Figure 4
for resonance assignments.
B
D
G PPS
Table 2. Chemical shift assignments
in ppm for iodo-terminated PPS
(structure shown in Figure 5)
Cl
C2
C3
C4
C5
C6
92.3
138.4
132.7
135.7
135.1
131.8

150
A1
140 138 136 134 132
Chemical Shift, ppm
130 128
Figure 4. Solution-state carbon-13 NMR spectrum of PPS/DS
previously reported [1]. Perdeuterated
PPS/DS (PPS-d4) (degree of polymerization,
DP, of 25) was synthesized in a manner
similar to that previously reported [1].
Iodo-terminated PPS (DP of 9) was prepared
according to the previously reported
method [1] but employing a 53 mole %
excess of DIB.
High-temperature solution-state
carbon-13 NMR employed a hightemperature
10-mm probe system from
Doty Scientific, Inc. (Columbia, SC) on a
JEOL Model GX-400 NMR spectrometer.
Samples were dissolved in N-cyclohexylpyrrolidinone
(CHP) under a blanket of
argon and examined at either 230 or 260°C.
Field/frequency stabilization was
accomplished with glyme-d6 ne ld
concentrically in a 5-mm capillary.
Solid-state NMR spectra were collected
on a Varian XL-300 NMR spectrometer.
Discussion
EPR Spectroscopy. During the
synthesis of PPS/DS in the epr
spectrometer, two distinctly different
resonances are observed (Figure 2). At
235°C a triplet resonance is observed, g
Values for the components of the triplet
resonance are 2.0033 ± 0.0002 (gi), 2.0072 ±
0.0001 (g2) and 2.0113 ± 0.0001 (53). The
Bulletin of Magnetic Resonance
126 124
isotropic g value is 2.0073 This resonance
is assigned to a sulfur-centered radical
cation. The isotropic g value is in excellent
agreement with g values of 2.0079 and
2.0076 reported by Murray and coworkers
[9] for a sulfur-centered radical cation in
two ASF5-doped PPS samples.
Above 280°C, the triplet resonance is
replaced by a singlet resonance with a g
value of 2.0042. This resonance is assigned
to an aryl radical.
It is interesting that the melting
process for PPS/DS can be easily monitored
by epr (Figure 3). Using PPS-d4, two
overlapping triplet resonances are
observed at 230°C. The triplet resonance
arises from hyperfine coupling between
the carbon-centered radical and the
deuterium attached to the carbon. The two
triplet resonances arise from free radicals
being in two different environments. At
230°C, the larger triplet resonance arises
from radicals in the solid phase while the
smaller triplet resonance arises from
radicals in a melted phase. As the
temperature is increased, the solid-phase
component decreases in intensity until it
disappears just above 280°C. The DSC
melting point for this polymer is 279°C.
It is unlikely that the carbon radical
observed here is mechanistically involved
in the synthesis of linear para-substituted
PPS/DS since these radicals would produce

Vol. 14, No. 1-4
J8 3 2
180 160 140 120 100
CHP
Carbonyl
160 140
Chemical Shift, ppm
151
Solid-State NMR Spectrum
Solution-State NMR Spectrum
120 100
Figure 5. Solid-State and Solution-State Carbon-13 NMR
spectra of an Iodo-Terminated PPS
1

152
aromatic substitution other than parasubstitution.
Our PPS/DS chemistry has
been shown to produce exclusively parasubstituted
polymer [10].
NMR Spectroscopv. Solution-state
carbon-13 NMR at 260°C in CHP provides
high-resolution NMR spectra of PPS/DS
that enable an analysis for various linkage
groups, such as sulfide and disulfide, as
well as numerous end groups of the type p-
X-phenyl, where X = H, SH or I. Resonances
for the species in Table 1 are assigned in
the spectrum shown in Figure 4.
Assignments are based on comparison with
spectra for model compounds and analysis
of several PPS/DS spectra. Species A, B and
C were assigned by comparison of spectra
from PPS or PPS/DS with and without .
excess sulfur as well as model compounds
such as diphenyl sulfide and diphenyl
disulfide. Based on the thermal history of
the synthesis, excess sulfur is expected to
be present only as disulfide linkages.
Species D was assigned by comparison with
the spectrum of thianthrene. Species E
assignments are discussed below. Species F
and G were assigned based on previous
assignments [2]. Several resonance
assignments in Figure 4, labelled with "?",
have not been made conclusively.
Complete assignments await synthesis of
appropriate model compounds.
Solution-state carbon-13 NMR of an
iodo-terminated PPS polymer (Species E,
Table 1) at 230°C in CHP clearly shows
sufficient detail to confirm the degree of
polymerization of 9 and to assign all
resonances in the terminal iodophenyl
group (Figure 5, Bottom). The assignments
for these carbons are shown in Table 2. By
comparison, the room temperature solidstate
NMR spectrum of this polymer is not
very informative (Figure 5, top).
Conclusions
PPS/DS does not lend itself to study by
conventional magnetic resonance
techniques operating under normal
conditions. Through the use of several
magnetic resonance techniques, we have
begun to understand some of the
limitations to studying PPS/DS by these
techniques and to enhance our
understanding of the chemistry of this
Bulletin of Magnetic Resonance
unusual polymer. We have presented here
data on the first direct observation of free
radicals generated under conditions of
synthesis of PPS/DS and have expanded the
applicability of the high-temperature
solution-state carbon-13 NMR technique.
Solution-state NMR has provided evidence
for several linkage groups and end groups
in PPS/DS enabling us to better appreciate
the chemistry of this polymer. Room
temperature solid-state NMR has not
proven to be useful in these studies.
References
1. Rule, M.; Fagerburg, D. R.; Watkins, J. J.;
Lawrence, P. B.; Makromol. Chem.,
Rapid Commun. 1991. 12, 221
2. Wade, B.; Abhiraman, A. S.; Wharry, S.;
and Sutherlin, D.; J. Poly. Sci: Pt B:
Polym. Phys., 28 1233 (1990)
3. Fahey, D. R.; Ash, C. E. Macromolecules
1991,24, 4242
4. Lopez, L. C; Wilkes, G. L. J. Macromol.
Sci., Rev. Macromol. Chem. Phys. 1989,
C29, 83
5. Ma, C.-C. M.; Hsiue, L.-T.; Liu, W.-L. J.
Appl. Polm. Sci. 1990,39, 1399
6. Hill, D. J. T.; Hunter, D. S.; Lewis, D. A.;
O'Donnell, J. H.; Pomery, P. J. Radiat.
Phys. Chem. 1990, 36, 559
7. Kreja, L.; Rozploch, F.; Warszawski, A.
Angew. Makromol. Chem. 1988,160, 163
8. Kispert, L. D.; Files, L. A.; Frommer, J. E.;
Shacklette, L. W.; Chance, R. R. J. Chem.
Phys. 1983,57, 4858
9. Murray, D. P.; Kispert, L. D.; Frommer, J.
E. J. Chem. Phys. 1985, 83, 3681
10. Rule, M.; Fagerburg, D. R.; Watkins, J.
J.; Lawrence, P. B.; Zimmerman, R. L.;
Cloyd, J. D.; Makromol. Chem.,
Macromol. Symp. 1992,54/55, 233

Vol. 14, No. 1-4 153
Application of 2-D HETCOR NMR to Investigate Polymer Blend Heterogeneity
Introduction:
Most NMR techniques for determining
domain structures in polymeric systems are
based upon the classic Goldman-Shen
experiment [1]. Domain sizes are calculated
from the time for spin diffusion to transfer
magnetization from one region of the sample
to another. These experiments generally
require resolution of the individual
components in the proton spectrum,
although recent experiments [2] demonstrate
that, in some cases where there is
inadequate proton chemical shift resolution,
it is possible to follow the evolution of spin
diffusion when the constituents have significant
differences in proton lineshapes.
An alternative approach to analyzing
domain structures entails application of
heteronuclear ^C- 1 !! NMR. Of the several
techniques introduced to measure heteronuclear
correlated spectra in the solid state,
a particularly effective method has been proposed
by Burum and Bielecki [3]. In the
basic 2-D experiment, crosspeaks occur primarily
for carbon-proton distances of less
than ~3 A. By incorporation of a spin
diffusion evolution time, crosspeak intensities
reflect longer range couplings, thus
I &.. 1- Goldman, M.; Shen, L. Phys. Rev. 1966,
^ 144,^21. J
U. 2 - C^P^ll, G.C.; VanderHart, D.L. J.
Magn. Reson. 1992,96,69-93.
\ 3 - Burum, D.P.; Bielecki, A. J. Magn.
Reson. 1991,94,645-652.
S. Kaplan
Xerox "Webster Research Center
800 Phillips Road 1004-39D
Webster, NY 14580, USA
enabling conformational and domain size
analyses [4]. This latter experiment consists
(Fig. 1) of an evolution period during which
homo- and heteronuclear interactions are
suppressed by simultaneous application of
BLEW-24 ( J H) and BB-24 ( 13 C) pulse
sequences, a waiting period without rf
during which exchange of magnetization
(spin diffusion) takes place, a mixing period
for isotropic cross polarization transfer of
proton magnetization to carbons utilizing
WIM-24 sequences on both nuclei, and
finally, carbon acquisition with proton
decoupling. In this paper we examine the
90° 63° 90° 90°
BLEW-24 WIM-24
13C BB-24 WIM-24
Figure 1. The HETCOR Experiment
cw
Decoupling
applicability of the HETCOR technique to
investigate domain sizes in a two component
blend.
Results and Discussion:
We have chosen for the present study a blend
of an aromatic diamine, N,N'-diphenyl-N-
4. Simpson, J.H.; Ruggeri, G.; Rice, D.M.;
Karasz, F.E., submitted.

154
N'-bis(3-methylphenyl)-[l,l'-biphenyl]-4,4'diamine
(TPD) in a bisphenol A polycarbonate
matrix. Pure TPD is highly crystal-
TPD
line, with a melting point of Tm = 167 °C and
a glass transition temperature of Tg=63 °C
[5]. A well mixed 50/50 blend with polycarbonate
(Tg=137 °C) cast from methylene
chloride is amorphous and shows a single Tg
of ~89 °C. Since the aromatic region of the
13 C spectrum of the blend has severe overlap,
we have focused our analysis on the aliphatic
region, where the individual component
methyl resonances are resolved.
Figure 2 shows contour plots of the
carbon aliphatic region in the two dimensional
HETCOR spectra as a function of the
spin diffusion mixing time. The 13 C peaks at
22 ppm and 32 ppm are from TPD and polycarbonate
methyl carbons, respectively.
Crosspeaks in the 20 us mixing time plot are
due primarily to short range directly bonded
carbon-proton couplings. However, with
increasing mixing times, longer range correlations,
e.g., between the methyl carbons
and aromatic protons, intensify. Ultimately,
for very long spin diffusion times, the
relative signal intensities for the contours
associated with a particular carbon will
reflect quantitatively the chemical shift distribution
of protons that are within an
effective range of spin diffusion from the protons
directly bonded to that carbon. Figure 3
shows the volume integral fraction of
aliphatic protons for the polycarbonate and
TPD methyl carbons as a function of spin
diffusion time. Both curves asymptote to an
5. Prest, W.M., unpublished data.
Bulletin of Magnetic Resonance
40 30 20 40 30 20
ppm (13Q
ppm OH)
Figure 2. HETCOR contour plots of the aliphatic
carbon region of a 50/50 (wt/wt) blend
as a function of spin diffusion time.
aliphatic volume fraction of 0.3 (point c),
which corresponds exactly to the fraction of
protons in the entire sample that are from
methyl groups. The results are very
different from measurements on a physical
mixture of the same components, where
separate asymptotes, corresponding to the
individual aliphatic proton fractions, are
observed for each component (polycarbonate
at point a and TPD at point b). Also shown
in Figure 2 is the difference response
between these two curves. The short term
initial rise can be attributed to
intramolecular and the long term decay to
intermolecular proton spin diffusion. From
the ratio of these rates (-10) and the known
intramolecular proton-proton distances (0.3
nm) the intermolecular distances can be

Vol. 14, No. 1-4 155
Aliphatic Fraction
1.0
0.8'
0.6
0.4 -
: •fW*t
0.2 r.
•* * •
—O— PC Icompatible 1
-••-TPD J Blend i
D PC 1 Physical z .
• TPD J Blend j
• 4i
m :
(a)
(c)
.. Difference (X5) =
(b)
0
0 0.5 1.0 1.5 2.0
Spin Diffusion Time (ms)
"I I 1 1 1 1 1 I • 1 1 i • 1 > • i i f 1 I I i t 1 i i i i 1 i i I i 1 f i i i 1 11 i r
Figure 3. TPD and polycarbonate methyl
carbon correlations with protons.
estimated to be about 1 nm (0.3XV10),
indicative of intimate molecular level
mixing. This treatment of the data permits
estimation of interdomain separation
without knowledge of the spin diffusion
constant, which is only assumed to be equal
in both domains.
An alternative approach to analyzing the
results of this experiment is shown in Figure
4. Plotted here is a measure of the similarity
(mean square difference) of the proton slices
for each of the methyl carbon resonances as a
function of spin diffusion time. This plot is a
direct measure of the diffusion of spin order
between the two blend constituents. In spite
of scatter, the best fit for the data is a single
exponential with a decay rate of 350 ms.
Employing the standard equation for
determining domain sizes from spin diffusion
i- times, r = (nDt)- 1/2 , where n represents the
domain dimensionality, t is the measured
if', spin diffusion time, and D is the spin diffusion
constant, typically 5X10" 12 cm2 sec" 1 ,
• domain sizes in the range of 0.8-1.4 nm are
• estimated.
0.2 r
0 -
0 0.5 1.0 1.5 2.0
Spin Diffusion Time (ms)
Figure 4. Intermolecular correlations between
TPD & polycarbonate methyl protons.
Heating the blend overnight at 110 °C induced
some phase separation, as evidenced
by a change in appearance from clear to
opaque. HETCOR diffusion spectra of this
sample showed negligible differences from
the data of the clear films. We conclude that
a very small amount of phase separation
occurred, e.g., near impurities or on the
surface only.
In summary, it is shown that the two
dimensional heteronuclear correlated spin
diffusion experiment is capable of extending
the range of applicability of domain size
measurements to blends comprised of
components that are structurally very
similar.
Acknowledgements:
The author wishes to thank J. Yanus for
supplying the blend sample used for the
current study and D. Rice (Univ. of
Massachusetts) and D. Burum (Bruker
Instruments, Inc.) for helpful discussions on
the HETCOR experiment.

156 Bulletin of Magnetic Resonance
NEW HIGH RESOLUTION NMR STUDIES
IN
POLYCRYSTALLINE TETRACYANOQUINODIMETHANE
M.T. Nunes
ICTPOL/CFMC-INIC, Av.Prof. Gama Pinto 2, 1699 Lisboa Codex, Portugal;
Dep. Quimica, ICEN, LNETI 2686 Sacavem Codex Portugal
A. Vainrub
Institute of Chemical and Biophysics, Estonian Academy of Sciences,
200001 Tallinn, Estonia
M. Ribet, F. Rachdi, P. Bernier
Groupe de Dynamique des Phases Condensees, U.S.T.L., 34095 Montpellier
Cedex 05, France
M. Almeida
Dep. Quimica, ICEN, LNETI 2686 Sacavem Codex, Portugal
and
G.Feio
ICTPOL/CFMC-INIC,Av.Prof. Gama Pinto 2,1699 Lisboa Codex,Portugal
Introduction
Tetracyanoquinodimethane (TCNQ) is
a well-known electron acceptor
that has been widely used to
prepare charge transfer complexes,
some of them exhibiting very high
electrical conductivity; locally
resolved "C Knight shifts [1] were
measured, accordingly [2].
NC
NC
CH==CH
CH=CH
TCNQ
\c==c.
CN
CN
In a static magnetic field of
4.7T, high resolution NMR studies
were performed on neutral TCNQ
using different polycrystalline
samples: natural isotopic
abundant, selectively enriched on
13 15
C isotope content and N enriched
[3]. Consequently, a complete
assignment of the 13 C resonances
was reported; the principal
components of the chemical
shielding tensor, the shielding
anisotropy and the shielding
asymmetry factor were also
obtained for the 13 C nuclei in the
fragments C-(CN)2, by graphical
analysis and by a computer fit of
the line intensities of CP/MAS
spectra [4]. However, large
uncertainties were reported on CN
groups data. 14 N quadrupole effects
were observed on 13 C spectra of CN
groups and the "N quadrupole
coupling constant was obtained (*"
3.84±0.12MHz). Using different
experimental conditions (like a
static magnetic field of 7.0T) we
highlight here
cyano groups.
new
3
C data on

Vol. 14, No. 1-4 157
RESULTS and DISCUSSION
When applying the numerical method
[4] to the spectra obtained at
4.7T, two different data sets were
found for the principal components
of the 13 C shielding tensor in TCNQ
cyano groups [ 3 ]; only one of the
corresponding asymmetry factors
(t]= 0 and 0.4) was characteristic
of an axially symmetric system.
Aiming to a clarification of this
problem, the improvement of the
experimental data for the
application of the numerical
method seems to be the next step;
in particular, the acquisition of
13 C spectra displaying an higher
number of spinning sidebands and a
better signal/noise is required.
These conditions are fulfilled by
running the spectra at higher
static magnetic field but
selecting the same MAS rate and a
longer recovering delay (600s);
Figure 1 shows typical 13 C CP/MAS
high resolution NMR spectra of
selectively enriched TCNQ to 99%
13 C isotope content on CN groups.
Again, two data sets are obtained
by a computer fit of the
intensities of the lines, one of
them corresponding to an axial
symmetry for the cyano groups, in
close agreement with that obtained
at 4.7T [3 ] :
ffll
ppm
228
210
a22
ppm
190
208
ppm
-80
-80
Aa
ppm
-289
-289
0.2
0.0
However, regarding the other set,
a much lower deviation from axial
symmetry is now obtained (axx-o22
equal to 38ppm instead of 69ppm).
Indeed, Clayden and co-workers
already pointed out that similar
simulated MAS spectra
Too 1«5 us ito iio M o -So
FIG.l. 13 C CP/MAS high resolution
NMR spectra of TCNQ cyano groups
obtained at 75.47 (on the top) and
50.13 MHz (on the bottom) with the
same spinning rate: 3.01 kHz.
are obtained when the chemical
shift tensor is axial or nearaxial
(ii0.2. On the
light of these conclusions the
present results are definitely
acceptable.
For the acquisition of the
spectra previously reported [ 3 ],
too short relaxation delays were
used possibly inducing a distorted
lineshape for the sidebands
envelope; anisotropic molecular
motion would influence distinctly
the relaxation times of magnetic
equivalent 13 C nuclei on TCNQ
molecules perpendicular and

158
parallel to the static magnetic
field. In fact, if 1.5s is now
used instead of 600s for that
duration period, at least three
fits are obtained (depending on
the sidebands used) with much
higher residual values; comparing
the data sets in this case, large
uncertainties are observed for all
the principal components of the
shielding tensor.
The study of 13 C TCNQ relaxation
in a wide temperature range are
now in progress and will certainly
contribute to the understanding of
these results; also, additional
insight on the role of TCNQ on
charge transfer complexes (like
DMTM(TCNQ)2 [2]) will be obtained.
To obtain the 14 N quadrupole
coupling constant from X3 C CP/MAS
spectra, run at 7.0T, the
procedure previously described was
used [3]; a C-N dipolar splitting
close to 230 Hz should be measured
now, which is in fact the case.
References
1 M.Mehring and J.Spengler,
Phys.Rev.Lett. ,53., 2441 (1984)
2 for a recent nmr study on these
materials see: F.Rachdi, T.Nunes,
M.Ribet, P.Bernier, M.Helmle,
M.Mehring and M.Almeida,
Phys.Rev.B, 45(14), 8134 (1992)
3 T.Nunes, A.Vainrub, M.Ribet,
F.Rachdi, P.Bernier and M.Almeida,
J.Chem.Phys.,96(11) , 8021 (1992)
4 J.Herzfeld and A.E.Berger,
J.Chem.Phys., 73_, 6021 (1980)
5 N.J.Clayden, C.M.Dobson, L.Lian
and D.J.Smith,
J.Magn.Reson.,69/476 (1986)
Bulletin of Magnetic Resonance

Vol. 14, No. 1-4 159
USE OF NMR RELAXATION
MEASUREMENTS TO DERIVE THE
BINDING SITE OF PLASTOCYANIN
IN COMPLEXES WITH
CYTOCHROME-F AND C
Sandeep Modi 1 , Ewen McLaughlin 1 , Derek S. Bendall 1 ,
S. He 2 and J.C. Gray 2
Departments of Biochemistry 1 and Plant Sciences 2
University of Cambridge, Cambridge, CB2 1QW
England (U.K.)
1 Introduction
Plastocyanin (PC) is a small (Mr 10
500), 'blue' copper protein which
transfers electrons from cytochrome
/ to the primary electron
donor of photosystem I (P700) in
the photosynthetic electron transport
chain. It reacts rapidly with
cytochrome / in vitro and also
with several other cytochromes,
although somewhat more slowly
[1].. The crystal structures of both
oxidized and reduced poplar
plastocyanin have been determined
and show that redox changes
cause only small changes at the Cu
site, leaving the structure of the
rest of the molecule essentially
unchanged [2,3]. The Cu atom and
its ligands (His-37, Cys-84, His-87
and Met-92) are located in a
hydrophobic pocket near one end
of the molecule (the "northern"
end) such that only the imidazole
ring of His-87 (the northern
histidine) is accessible to solvent
(Fig 1).
Studies with small molecules,
such as [Fe(CN)6] 3 " and
[Co(phen)3]3+, have identified two
reaction sites on plastocyanin; one
close to the copper ligand His-87
at the northern hydrophobic patch
and the other close to the more
remote Tyr-83 at the eastern
acidic patch [4-8]. Chemical modification
of the acidic amino acid
residues, inhibitory effects of
small molecules and ionic strength

160 Bulletin of Magnetic Resonance
effects on electron transfer
strongly suggest that cytochrome /
binds at the remote eastern site
[9,10]. However the pathway of
electron transfer from cytochrome
/ to the copper site in plastocyanin
C
L12 H87
E60
Fig. 1 : Structure of Plastocyanin. The Cu
ligands and the side-chains of Tyr83 and the
residues of the eastern acidic patch are shown
here with the coordinates of poplar
plastocyanin from the Brookhaven Database,
except for substitution of Glu45 (as in pea
plastocyanin) for Ser45. ^-Strands are
shaded.
is not clear. One possibility is that
cytochrome / binds to the negative
charges of the eastern acidic patch
and donates electrons via a
tunneling pathway starting at
Tyr-83 [5,11]. However, the
distance from Tyr-83 to the
copper ion is approximately 12A,
whereas at the northern site the
copper ion is only 6A from the
surface of the molecule [11].
The recent development of
expression systems for the small
blue copper protein, plastocyanin
[12-15], has provided a valuable
tool for study of the molecular
details of its interaction with its
native reaction partners (cytochrome
/ and photosystem I in the
photosynthetic electron transport
chain). To examine the pathway of
electron transfer from cytochrome
/ to plastocyanin we have altered
Tyr-83 to Phe-83 and Leu-83 by
site-directed mutagenesis of the
pea plastocyanin gene [13-14].
Measurements of binding constants
and electron transfer rates
indicate, that Tyr-83 not only
forms part of the main route of
electron transfer from cytochrome
/ to plastocyanin but is also involved
in binding to cytochrome /.
Nuclear magnetic resonance
spectroscopy has been established
as a very convenient and effective
technique for structural studies of
proteins. For haemproteins, the
characteristic hyperfine-shifted
NMR spectrum of a paramagnetic
haemprotein carries the signature
of the electronic and structural
properties of the haem group.
Measurements of spin - lattice

Vol. 14, No. 1-4 161
relaxation (Tj) and spin-spin
relaxation (T2) times provides
useful methods for the
determination of dissociation
constants and distances of various
nuclei from the paramagnetic
centre in a protein-protein
complex. Relaxation measurements
were carried out to get
information about the relative
dispositions of the two proteins
(PC and cytochromes) in their
complexes.
2 Assignment of
Proton NMR resonances
for pea
Plastocyanin
Proton NMR measurements were
carried out on a Bruker AM 500-
MHz FT-NMR spectrometer at
300K in 50 raM phosphate buffer
(pH 6.0). Proton chemical shifts
were referred to a proton signal of
dioxan as a reference at 3.74 ppm.
Proton NMR resonances for pea PC
were assigned using 2D-NMR
spectroscopy at 300K (pH = 6.0).
The conformation of the Phe-83
plastocyanin was examined by
^-NMR. A ID spectrum in H2O,
compared with that of the wildtype
protein, clearly demonstrated
the disappearance of the amide
resonance of Tyr-83 at 9.40 ppm,
and the appearance of a new
resonance at 9.37 ppm which is
likely to be that of Phe-83 [13],
although positive identification
must await further 2D
experiments. Overall the close
similarity between the spectra of
the mutant and wild-type proteins
indicated that the replacement of
Tyr-83 with Phe-83 had not led to
major conformational changes
throughout the protein. Insufficient
protein of the Leu-83
mutant was available for NMR
analysis, so its conformation was
examined by CD between 190 and
260 nm. The spectrum obtained
for the oxidized protein was
essentially identical to that of the
wild-type. The Phe-83 mutant
protein also gave a closely similar
spectrum. These results confirm
that the three proteins had
identical gross conformations.
3 Kinetics
surements
Mea-
Electron transfer from reduced
cytochrome to oxidized plastocyanin
was monitored at 422 nm
with an Applied Photophysics
stopped-flow spectrophotometer
(SF.17MV). The rate of binding of
plastocyanin and cytochrome was
measured by following the

162 Bulletin of Magnetic Resonance
increase in absorbance of oxidized
cytochrome at 410 nm in the
stopped-flow spectrophotometer.
K& was determined by taking
advantage of the increased absorbance
of the Soret band of
cytochrome on binding to
plastocyanin
The results reported in Table I
[13,14] demonstrate convincingly
that Tyr-83 of plastocyanin is part
of the main tunneling pathway for
the electron between the haem
rings of both cytochrome c and
cytochrome / and the Cu atom of
plastocyanin. A leucine residue in
this position is much less effective
and it seems likely that the
facilitation of electron transfer by
tyrosine or phenylalanine is due
to the aromatic nature of the ring,
as has been proposed in other
proteins. A striking difference
between the kinetics of reduction
of plastocyanin by the two
cytochromes is that the wild-type
protein and the Phe-83 mutant
behave identically towards
cytochrome c, but not towards
cytochrome /. In the latter case
the rate of reduction of the
mutant protein is about seven
times slower, a difference that can
be ascribed entirely to weaker
Table I
Kinetic parameters for reduction of pea plastocyanin by cytochromes c and/
Plastocyanin
Cytochrome
Wild type
Phe-83
Leu-83
k2 (x 10-6)
(M-V 1 )
c as donor
3.26
3.28
0.421
Cytochrome/as donor
Wild type
Phe-83
Leu-83
40.6
5.43
0.955
*A
(M-l)
1253
1295
1260
9890
1270
968
•*a(xlO-6)
(MrV 1 )
20.0
22.7
21.7
43.5
5.86
1.27
*.a(xl0< *) k( (x 10-3)
(s- 1 )
16.0
17.5
17.2
4.40
4.61
1.31
(s- 1 )
3.11
2.96
0.340
Values are given as mean + standard deviation. Data for cytochrome/and c are from
[13].and [14] respectively.
62
58
4.0

Vol. 14, No. 1-4 163
binding. We therefore predict that
when the structure of cytochrome
/ becomes known it will reveal a
surface residue in the region of
the exposed haem edge which is
capable of hydrogen bonding to
the -OH of Tyr-83.
4 NMR relaxation
(Ti) measurements
Protein concentrations were
determined from the following
absorption coefficients: reduced
horse heart cytochrome c, £550nm
= 2.76 x 10 4 M-icm" 1 ; reduced
oil-seed rape cytochrome /,
£554nm = 2.6 x 10 4 M-icm-l;
oxidised plastocyanin, £597nm =
4.7 x 103 M-lcm-1. Proton NMR
relaxation measurements were
carried out on a Bruker AM 500-
MHz FT NMR spectrometer at
300K. The samples were in 0.01 M
phosphate buffer (containing 90
raM NaCl) at pH 6.0 (volume, 0.4
ml). Proton NMR spectra of Cd-PC
were obtained by accumulation of
about 160 transients at 16K data
points in quadrature mode. To
facilitate relaxation measurements,
a redox-inactive form of
plastocyanin was prepared, in
which Cu was replaced by Cd. 2D
NMR spectra of Cd-PC showed that
the conformations of the two
proteins are essentially the same.
For the relaxation time
measurements, samples were
treated with Chelex 100 (Bio-Rad)
to remove any traces of free metal
ions. To obtain the longitudinal
relaxation time (Tiobs), the
inversion recovery method with
180^-T -90° pulse sequence was
used.
5 Determination of
the Apparent Dissociation
Constant of
Cd-PC Binding to
Cytochromes using
1H-NMR Tl Measurements
Longitudinal proton relaxation
times of Cd-PC were measured in
the presence of various
concentrations of cytochrome (cor
f) to find the binding constants
and the distances from various
protons of Cd-PC to the
cytochrome iron atom. Observed
longitudinal relaxation time
( T lobs) of Cd-PC proton
resonances can be considered as
the sum of the relaxation rates of
the bound and free Cd-PC
fractions and is related to Kj), T\ 5
and Tif, where Kj) is the apparent
dissociation constant of the
Cyt/Cd-PC complex, Tjb is the Ti

164
of the Cyt/Cd-PC complex, and Tif
is the Ti of the Cd-PC in the
absence of the cytochrome. KTJ and
Tib for Cd-PC binding to
cytochrome was obtained from the
above data. Kj) obtained from NMR
relaxation measurements agreed
very well with the value obtained
from optical spectroscopy (Table
I).
6 Determination of
Distance using
Measurements
The Solomon and Bloembergen
equations were used to determine the
distance (r) of individual protons of
bound Cd-PC from the ferric centre
of cytochromes. The distances for
these protons were used to get the
relative position and conformation of
PC with respect to the ferric ion of
cytochromes. Our initial results show
that the ferric ion of cytochrome is
very near to the Tyr-83 residue of
PC, which is consistent with our
kinetics studies. This work is still in
progress.
7 References
1. P.M. Wood, Biochim. Biophys.
Acta 357, 370 (1974).
2. J.M. Guss, and H.C. Freeman, J.
Mol. Biol. 169, 521 (1983).
Bulletin of Magnetic Resonance
3. J.M. Guss, P.R. Harrowell, M.
Murata, V.A. Norris and H.C. Freeman,
J. Mol. Biol. 192, 361 (1986).
4. DJ. Cookson, M.T. Hayes and P.E.
Wright, Biochim. Biophys. Acta
591, 162 (1980).
5. A.G. Sykes, Chem. Soc. Rev. 14,
283 (1985).
6. A.G. Sykes, Struct. Bonding, 75,
175 (1990).
7. F.A. Armstrong, H.A.O. Hill and
C. Redfield, J. Inorg. Biochem. 28,
171 (1986).
8. J.D. Sinclair-Day and A.G. Sykes
J. Chem. Soc. Dalton Trans. 2069
(1986).
9. G.P. Anderson, D.G. Sanderson
and E.L. Gross, Biochim. Biophys.
Acta, 894, 386 (1987).
10. C. Beoku-Betts, S.K. Chapman
and A.G. Sykes Inorg. Chem. 24,
1677 (1985).
11. P.M. Colman, H.C. Freeman,
J.M. Guss, M. Murata, V.A. Norris,
J.A.M. Ramshaw and M.P. Venkatappa,
Nature, 272, 319 (1978).
12. M. Nordling, T. Olausson and
L.G. Lundberg, FEBS Lett., 276, 98
(1990).
13. S. He, S. Modi, D.S. Bendall and
J.C. Gray, EMBO J, 10, 4011 (1991).
14. S. Modi, S. He, D.S. Bendall and
J.C. Gray, Biochim. Biophys. Acta
(in press).
15. S. Modi, M. Nordling, L.G. Lundberg,
O. Hansson and D.S. Bendall,
Biochim. Biophys. Acta (in press).

Vol. 14, No. 1-4
1 Introduction
Metal-peptide Interaction:
Influence of the Aminoacid Sequence on the Binding of Co(II)
to Glycyltryptophan and Tryptophylglycine
Studied by ^NMR and Fluorescence
A. Spisni, G. Sartor, L. Franzoni
Institute of Biological Chemistry, University of Parma
Via Gramsci, 14,
43100 Parma, Italy
A. Orsolini, P. Cavatorta
Department of Physics, Section of Biophysics, University of Parma
Viale delle Scienze,
43100 Parma, Italy
and
M. Tabak
Instituto de Fisica e Quimica de Sao Carlos, University of Sao Paulo
Av. Dr. Carlos Botelho, 1465,
13560 Sao Carlos (SP), Brazil
Aim of this work is to investigate the nature of the
interactions between transition metals and peptides.
Peptides bind metal ions in various manner
depending on their aminoacid composition, on the pH of
the solution as well as on the metal/peptide ratio. Among
the various aminoacids those with a charged side-chain
are the most efficient for metal binding, though, ions can
also be coordinated by the peptidic nitrogen as well as by
the terminal arnino group. It is well known that the
binding of transition metals to peptides decreases the
apparent pKa of both the terminal amino group and of the
peptidic NH. The fluorescence of the aromatic
aminoacids (Trp, Tyr and Phe) can be used to monitor
those changes. It is well recognized that fluorescence
spectroscopy is suitable for the study of peptides and
proteins, and that Trp is the best internal probe due to its
' Hgh quantum efficiency and molar absorption as
compared to Tyr and Phe. Stemming from these
considerations, conformational changes of proteins or
Peptides can be monitored following the modifications of
.the Trp fluorescence. Similarly, the interaction of Trp
•*ith transition metal ions can be easily detected by
measuring the fluorescence quenching induced by the
metal ions themselves.
Recently [1], [2] we have been studying the
interaction of Cu(II) and Ni(II) with Trp and Gly-Trp
and we found not only that the quenching of Trp
fluorescence is mainly due to a ground state interaction
but also that, for the two metals, the formation of the
metal complex with the dipeptide and the AA is
different, both in terms of binding constants and
stoichiometry.
Another technique that can be used for the study of
metal-peptide complexes is high resolution 'HNMR.
The possibility of a transition metal to act as a line
broadening or as a shift reagent is strictly associated to
its electronic relaxation time. Paramagnetic ions with
a short relaxation time are responsible for changes in
chemical shift without line broadening while, if the
relaxation time is long enough, the effect is a strong
line broadening with no changes of the chemical shift.
Therefore the study of the chemical shift variation as
a function of the metal concentration can lead to
interesting results relevant for a better understanding of
the complexe's stoichiometry. At the same time
relaxation studies will provide valuable information on
their geometry.
165

166
2 Materials and Methods
Tryptophylglycine (Trp-Gly) and glycyltryptophan (Gly-
Trp) were obtained from Sigma Co. S. Louis, MO, their
purity was checked by gas-chromatography. CoCl2-6H2O
was obtained from Merk, Darmstad, Germany and used
without further purification.
*HNMR experiments were carried out using a Bruker
AMX 400 spectrometer, operating at 9.41 T, static
fluorescence experiments were carried out using a Perkin
Elmer MPF 44A spectrofluorimeter, time resolved
fluorescence experiments were carried out with a time
correlated single photon counter equipped with an
Edinburgh F199 nanosecond flash lamp operating in a N2
flux of 1 2/min, a Philips XP2020Q fast photomultiplier
and an EG&G Ortec fast NIM electronics. Time resolved
fluorescence data analysis was carried out using the
Global Analysis [3].
The temperature for all the experiments was 25°C.
Peptide's solutions for *HNMR were freshly prepared
in double distilled water, with 10% D2O, at a final
concentration of 10 mM. Solutions for fluorescence
experiments were obtained by appropriate dilution of 1
mM stock solution in double distilled water. No buffers
were used. CoCl2 solutions were 3 M for 'HNMR and
1 M for fluorescence experiments.
Distinct protocols were used to obtain the pH
titration of the two peptides in : HNMR and fluorescence
experiments. In 'HNMR experiments small aliquots (y.£)
of 0.1 M HC1 or 0.1 M NaOH were added to 0.5 m£ of
the sample in order to obtain a given pH. As for the
fluorescence measurements, additions of 1 M HCl or 1 M
NaOH were made on 50 m£ in order to avoid dilution
artifacts, the same pHmeter was used. The data from pH
titrations, for both *HNMR and fluorescence, were fitted
using the Henderson-Hasselbach equation in order to
obtain the pK, values .
In the case of fluorescence experiments correction for
the inner filter effect was made using Parker's equation
[4]-
3 Results and Discussion
Fluorescence quenching experiments demonstrate that the
complexes fonned by Gly-Trp and Trp-Gly with Co(II)
have distinct properties. The binding of the metal ion is
strictly influenced by the aminoacid sequence and by pH.
Moreover Co(II)like other paramagnetic transition metals
such as Cu(II) and Ni(II), is known to influence the pK,
of the ionizable groups in aminoacids and peptides [1],
[2].
Bulletin of Magnetic Resonance
In the case of the binding of metal ions to
fluorescent aminoacids a non fluorescent ground state
complex is formed, thus, it is possible to calculate the
binding constant from the variation of the fluorescence
intensity. Co(II) binding to Trp-Gly and Gly-Trp
produces biphasic quenching curves [5], indicating that
at least two different complexes are formed in each
case. In table I the binding constants of Co(II) for the
two peptides are reported together with the fraction of
the two complexes present at the given pHs.
Unfortunately, fluorescence spectroscopy is not
able to give the required informations for the direct
determination of the stoichiometry of the complexes
and of their geometry. To overcome these limitations
J HNMR has been used.
The pK, values of the ionizable groups
(carboxylate and amino group) in the absence and in
the presence of Co(II) have been obtained from
fluorescence data. These values turn out to be quite
different with respect to those obtained using the same
technique and reported in a recent publication [6].
To verify these values, the determination of the
two pK,s was carried out by means of 'HNMR. We
studied the pH dependence of the proton chemical
shifts for the two dipeptides in the absence of the metal
ion. Figure 1A reports the plot of the data for the two
a protons of glycine and figure IB for the NH proton
and for the indolic one. From these data the pK,s of
the carboxylate and of the amino group (Table II) were
calculated. The pK,s values of the amino group
obtained from pH dependence of the chemical shift oi
the various protons were in good agreement with those
obtained from fluorescence experiments. In the case oi
the indolic and the NH proton resonances, as they
disappear above pH 7.5, probably because of their fasi
exchange with H2O, the fitting of the data was obtained
imposing, as pK^ values, the average values obtained
from all the other protons. As can be seen the fitting is
quite satisfactory.
When Co(II) is present, while with fluorescence
spectroscopy it is possible to carry out a pH titration oi
the dipeptide-metal complex up to pH 12, NMR i*
limited to pH 7. In fact, due to its high sensitivity
fluorescence spectroscopy allows to operate at (iw
concentration for the dipeptides and for Co(II), thus
avoiding the precipitation of the complex at high pn
NMR, being less sensitive, requires concentrations ii
the range of 10 mM, therefore, above pH 7.5 w<
observe the formation of a mixed complex with OH
that precipitates, with a consequent disappearing of th<
signal.

Vol. 14, No. 1-4 167
•>"• .
TABLE I: Binding constants (K} and K^) obtained from fluorescence quenching experiments, /, and f2 represent the
fraction of complex formed
Dipeptides
Trp-Gly
pH 3.2
Gly-Trp
pH 3.2
Trp-Gly
pH 8.2
Gly-Trp
pH 8.2
3.9 -
fi
1
0.36
0.22
0.55
0
2810.6
896.6
900.6
0 2 4 8 8 10 12 14 0 2 4
f2
—
0.64
0.78
0.45
K, (M- 1 )
Figure 1 A Glycine a protons chemical shifts of glycyltryptophan (•,•) and tryptophylglycine (O,E).
B Indolic proton (», v) and NH(*, A) chemical shifts of glycyltryptophan ( T , A ) and tryptophylglycine (V,A).
—
33.5
131.3
12.1

168
TABLE II: pK,,s obtained from fluorescence and from 1 H NMR experiments
pKal
P*^
P*^
pK,,
P*^
4.0
3.8
3.7
s 3 - 8
a, ' 3.5
3.3
3.Z
Gly-Trp Gly-Trp:Co ++
2.74
8.27
—
3.13
8.30
Fluorescence intensities
1.85
7.46
9.66
a H NMR Chemical Shifts
2.89
3.1
0 1 2 3 4 5 6 7
pH
—
Bulletin of Magnetic Resonance
Trp-Gly Trp-Gly:Co ++
2.59
7.73
—
2.86
7.74
1 ; I I I
2 3 4
PH
B 10.2
Figure 2 A Glycine a protons chemical shifts of glycyltryptophan (•,•) and tryptophylglycine (O,Q) in presence q
Co(II) (1:5 peptide metal ratio). >
B Indolic proton (^.v) and NH (^,A)chemical shifts of glycyltryptophan (•,*) and tryptophylglyctoe (v,i
in presence of Co(II) (1:5 peptide metal ratio).
10.0
8.8
K>
8.0
7.8
7.6
7.4
1Z
2.51
7.05
9.15
2.88
—

Vol. 14, No. 1-4 169
Figure 3 Top
Bottom
""I"'
-SO
WG ilOmM) CO • 1; 5 pH-7.15
SW-30

170
In figure 2A and 2B the 'HNMR titration of the glycine
a protons and of the indolic and NH protons in the
presence of Co(II) (1:5 ratio) are reported. It can be seen
that above pH 7.5 no signal was detectable due to the
formation of a bluish precipitate and that, as a
consequence only the carboxylate pK, has been calculated
and reported in Table II.
Despite these limitations, we have been able to detect
the proton NMR spectrum for the two dipeptides
complexes (Figure 3). A group of resonances is shifted
between -50 ppm to -100 ppm and two or three peaks
appear between 80 ppm and 130 ppm. Interestingly,
though the assignment of the peaks is still to be
completed, it can be seen that the NMR profile for the
two complexes is quite different suggesting that Trp-Gly
and Gly-Trp, indeed, are forming two distinct complexes.
The half line width is approximatively 200 Hz for Gly-
Trp and 1000 Hz for Trp-Gly indicating that Co(II) is
either closest or more tightly bound to Trp-Gly as
compared to the Gly-Trp. Indeed the Trp-Gly binding
constants suggests an average high affinit of this peptide
respect to Gly-Trp. The NMR spectra of the two
peptide-metal complexes present one single peak with
the same chemica shift.A that peak tend to disappear
upon D2O addition we believe it is a proton bound to a
nitrogen atom. The NMR spectra of the two complexes
posses other peculiar characteristics. The chemical shifts
of their resonance lines are insensitive to the pH variation
suggesting a high stability of the complexes. Moreover,
the NMR spectra disappear below pH 5.2 for Trp-Gly
and below pH 4.15 for Gly-Trp. We do believe these
evidences are an indication both of the need for the
carboxylic group to be ionized as well as of the
requirement for a small fraction (1/1000) of NH2 (the pK,
for the amino group are 8.27 and 7.73 respectively) in
order to have metal binding.
In conclusion, these preliminary results indicate that,
because of the complementarity of NMR and
fluorescence spectroscopy, it is possible to better evaluate
the pK,s of the dissociable groups and the metal binding
properties of small peptides. We believe that such an
integrated approach can be relevant for the study of more
complex macromolecules such as polypeptides and
proteins.
4. References
Bulletin of Magnetic Resonance
1. Tabak M., Sartor G. and Cavatorta P., /. of
Luminescence, 43., 355 (1989).
2. Tabak M., Sartor G., Neyroz P. and Cavatorta P.,
J. of Luminescence, 46, 291 (1990).
3. Knutson J.R., Beechem J.M. and Brand L., Chem.
Phys. Lett. 102, 501 (1983).
4. Parker C.A. Photoluminescence of solutions,
Elsevier, (1968).
5. Sartor G. Franzoni L., Cavatorta P., Tabak M. and
Spisni A., manuscrip in preparation.
6. Chen F., Knutson J.R., Ziffer H. and Porter D.,
Biochemistry, 2Q, 5184 (1991).
Acknowledgments This work was supported
by a CNR Grants #92.00752.CT04 and #92.02243.ctl4,
by MURST 60% (SG) and MURST 40% (SA)

Vol. 14, No. 1-4 171
INTRODUCTION
ASSIGNMENTS OF THE *H NMR SPECTRUM OF A CONSENSUS
DNA-BINDING PEPTIDE FROM THE HMG-I PROTEIN
Jeremy N. S. Evans§$*, Mark S. Nissen§ and Raymond Reeves§
Departments of Biochemistry/Biophysics§ and Chemistry*,
Washington State University, Pullman, WA 99164-4660. U. S. A.
The HMG-I subfamily [1-3] of high mobility group
(HMG 1 ) chromatin proteins [4] consists of DNA-binding
proteins that preferentially bind to stretches of A'T-rich sequence
both in vitro [5-8] and in vivo [9]. Recently, members
of the HMG-I family have been suggested to bind in
vitro to the narrow minor groove of A*T-DNA by means of
an 11 amino acid peptide binding domain (BD) which, because
of its predicted structure is called the "A*T-hook motif
[10]. The HMG-I proteins are specific substrates for the
ii i i • / \ -, Acdc2/cdc28 , . , ,
cell cycle regulating enzyme(s) p34 kmase (also
known as histone HI kinase) both in vivo and in vitro [11-
13]. The sites of phosphorylation by cdc2 kinase are the
threonine residues at the amino terminal ends of the A*Thook
motifs and such modifications have been demonstrated
to reduce markedly the affinity of binding of the HMG-I proteins
to DNA [12,13].
HMG-I proteins are also of considerable biological interest
because they are expressed at elevated levels in actively
proliferating cells and have been observed to be a
characteristic feature of undifferentiated [1,7] or neoplastically
transformed cellular phenotypes [14,15]. High HMG-I
levels have been found to be a consistent feature of rat and
mouse malignant cells [14-17] and have been suggested to
be protein markers for both neoplastic transformation [15]
and metastatic potential [18]. The HMG-I proteins have also
been implicated in control of DNA replication [19,20] and
the regulation of gene transcription [8,21,22]. It is known
from their primary sequences that the HMG-I proteins have
the overall structure typical of Ptashne-type transcriptional
activator proteins possessing both a DNA binding domain(s)
and a highly acidic COOH terminus [23]. In vitro HMG-I-
Ijke proteins have been demonstrated to increase transcription
of isolated ribosomal genes [21] and to alter the conformation
and stability of A-T-rich regions of DNA [24]
properties often associated with DNA-binding gene regulatory
proteins.
As an initial attempt to determine in molecular detail the
interaction of the BD peptide with A»T rich DNA, we have
examined a synthetic 13 residue BD peptide by NMR
spectroscopy. In this paper we report the assignments of the
j
HMG - Wfi 11 mobility group; BD,
domain; DQF, double quantum filtered;
C0Trelated
spectroscopy; NOESY, nuclear
! effCt spectroscopy; ROESY, rotatingei
VT 3USer effect spectroscopy; TOCSY, total
elated spectroscopy.
resonances for the peptide at 295 K and pH 3.4, and provide
preliminary evidence on the peptide structure.
MATERIALS AND METHODS
Peptide Synthesis. The 13mer BD peptide
(VPTPKRPRGRPKK) was synthesized by solid-phase synthesis
(on a Departmental Applied Biosystems model 431A
peptide synthesizer), and purified by reverse-phase HPLC on
a Vydak C4 column (1 x 25 cm) using, a water (containing
0.1% trifluoroacetic acid)-acetonitrile gradient under standard
conditions. The 13mer BD peptide eluted at 15% acetonitrile
(Rt = 13 mins @ 1.5 mL min" 1 ).
NMR Spectroscopy. High field Fourier transform (FT)
NMR studies were performed on a Varian VXR-500S (11.75
T, 500MHz J H) NMR spectrometer. Deuterium was used
for locking the field. *H NMR chemical shifts were referenced
externally to samples of similar dielectric constant
containing sodium 3-(trimethylsilyl) propanoate-2,2,3,3-d4
(TSP) in D2O buffer (5H = 0.00 ppm). Sample temperature
was maintained with a Varian variable temperature control
unit, using gaseous nitrogen (from boil-off liquid nitrogen)
cooled using an FTS XR-85 cryo-cooler. The majority of
samples of peptide were maintained at 295 K, except where
the amide exchange rates were being measured, when the
sample was maintained at 277 K. Data was downloaded to
either a Silicon Graphics 4D25TG or a-4D70GT workstation,
and converted from Varian format to FELIX format using
the VNMR2FELIX conversion program (a gift from
Darrell Davies, University of Utah). The output from this
was processed using FELIX (Hare Research Ltd.). All 2D
data was obtained using the hypercomplex phase sensitive
method [25] and processed as 2K x 2K complex data sets
with baseline correction and sine-bell squared weighting
functions in both dimensions.
DQF-COSY were recorded with the pulse sequence tQ-
90°-f7-90 o -S-90 o -f2, where tj is the evolution time, t2 is
the acquisition time, and 8 is a fixed delay of 3 \is [26].
TOCSY spectra were recorded with the pulse sequence /#-
90°-fi-SLx-(MLEV-17)-SLx-/2, where SLX was a 4 ms
trim pulse along the x axis [27]. The MLEV-17 spin-locking
pulse sequence was repeated to give a mixing time of
40 ms. ROESY spectra were recorded with the pulse sequence
/o-90 0 -r;-90 0 -SLx(30°)-90°-t2, where SLx(30°) is a
small spin-lock pulse repeated to give a mixing time of 200
ms [28]. NOESY spectra were recorded with the pulse

172 Bulletin of Magnetic Resonance
TABLE 1 Sequential Assignments of HMG-I 13mer Binding Domain Peptide, pH 3.4,295K
Residue
VI
P2
T3
P4
K5
R6
P7
R8
G9
RIO
Pll
K12
K13
NH
_
_ 8.44
• _
8.10
8.35
8.54
8.41
8.28
— 8.44
7.24
otCH
4.23
4.58
4.60
4.33
4.23
4.68
4.46
4.34
3.92, 4.04
4.68
4.45
4.32
3.27
sequence ta-90°-tj-90°-Tm-90 o -t2 [29,30] with mixing
times, TW, of 100, 300, 400, and 600 ms. Solvent suppression
was achieved by presaturation of the H2O resonance for
all the 2D experiments.
Sample Preparation. The BD peptide was dried with successive
cycles of lyophilization and re-hydration with either
H2O or D2O and then dissolved in one of the following:
Buffer (i) 25 mM potassium phosphate, 0.01% (w/v) NaN3,
in 10% v/v D2O/H2O, pH 3.4; or Buffer (ii) 25 mM potassium
phosphate, 0.01% (w/v) NaN3, in 99% v/v
D2O/H2O, pH 3.4. pH titrations were carried out by careful
4.0 3.0 2.0
(ppm)
tPx
m 6*
f\j
I
&CH
2.36
1.91, 2.37
4.16
1.91, 2.35
1.76, 1.87
1.77, 1.87
1.93, 2.34
1.73, 1.87
1.76, 1.88
1.94, 2.34
1.79, 1.84
1.74, 1.89
Others
1.03, 1.14
2.08
1.31
2.04
1.43
1.71
2.05
1.44, 1.50
1.73
2.05
1.51
1.49
3.67, 3.80
3.73, 3.90
3.03, 3.25
3.05, 3.25, 7.55
3.68, 3.87
3.05, 3.25
3.25
3.66, 3.84
3.05
3.05
addition of small quantities of HC1 or NaOH and measurement
with a 4 mm pH electrode (Ingold Co.).
RESULTS
The 13mer peptide was studied by NMR at 277 and 295 K
and at a variety of pH values. The linewidths of the ID 500
MHz NMR spectra were not sensitive to concentration in
the range 0.1-20 mM, indicating negligible aggregation of
the peptide [31]. The use of 2-dimensional double-quantum
filtered correlated spectroscopy (DQF-COSY) and total
correlated spectroscopy (TOCSY) has enabled us to assign
resonances to amino acid spin types, although distinctions
1.0 4.0 3.0 2.0
(ppm)
Figure 1 500 MHz *H DQF-COSY (A) and TOCSY (B) NMR spectra of the aliphatic resonances of the 13mer BD
(20 mM) in 90%H2O/10%D2O phosphate (25 mM) buffer, pH 3.4,295K. Cross-peak assignments are indicated.

Vol. 14, No. 1-4 173
G9NH
•°KI2NHa
R8nHa
• K-JNHa
»T3NHO
* »Rl 6 hours.
Interestingly, the slowly exchanging amide resonances also
do not titrate over the pH range 2.2 to 7.1, with the exception
of the K5 NH, and all appear to be located along one
side of the peptide molecule.
DISCUSSION
A model for how the consensus binding domain peptide
from the HMG-I protein binds to the minor groove of A«T
rich DNA has been proposed by this laboratory [10] on the
basis of molecular modelling. In order to test this model, we
have initiated NMR studies of a 13mer BD peptide in
solution. While no

HI
i ill,
I 1
174
the 13mer peptide). These amides might be expected to be
less accessible to solvent, and potentially may participate in
hydrogen-bonds with the carbonyl groups from (i, i + 2)
residues. At this stage, there is no additional corroborating
evidence from medium-range or long-range NOEs either to
support or refute these speculations. However, in the
presence of A«T rich DNA, the conformational lability of
the peptide backbone and side-chains would be expected to be
reduced significantly, and longer range NOEs detectable.
ACKNOWLEDGEMENTS
We should like to acknowledge Gerhard Munske for the
synthesis of the 13mer peptide, Wendy Shuttleworth for
help with some sample preparations, Darrell Davies
(University of Utah) for the VNMR2FELIX conversion
software. Supported in part by an American Cancer Society
Institutional Grant IGR-119L (JNSE), NIH Grant R01
GM46352 (RR), and the WSU NMR Center is supported by
NIH grant RR 0631401, NSF grant CHE 9115282 and
Battelle Pacific Northwest Laboratories Contract No. 12-
097718-A-L2.
REFERENCES
[ 1] Lund, T., Holtlund, J., Fredriksen, M. & Laland, S.
(1983) FEBS Lett. 152, 163-167.
t 2] Johnson, K., Lehn, D., Elton, T., Barr, P. & Reeves,
R. (1988) J. Biol. Chem. 263, 18338-42.
[ 3] Johnson, K., Lehn, D. & Reeves, R. (1989) Mol.
Cell. Biol. 9,2114-23.
[ 4] Goodwin, G. & Bustin, M. (1988) in Kahl, G.
Architecture of Eukaryotic Genes, 187-205, VCH
Weinheim, Germany.
[ 5] Strauss, F. & Varshavsky, A. (1984) Cell 37, 889-
901.
[ 6] Solomon, M., Strauss, F. & Varshavsky, A. (1986)
Proc. Nail. Acad. Sci. U.SA. 83, 1276-1280.
[ 7] Elton, T., Nissen, M. & Reeves, R. (1987)
Biochem. Biophys. Res. Commun. 143, 260-265.
[ 8] Reeves, R., Elton, T., Nissen, M., Lehn, D. &
Johnson, K. (1987) Proc. Natl. Acad. Sci. U.SA.
84, 6531-6535.
[ 9] Disney, J., Johnson, K., Magnuson, N., Sylvester,
S. & Reeves, R. (1989) / Cell Biol. 109,1975-82.
[10] Reeves, R. & Nissen, M. (1990) J. Biol. Chem.
265, 8573-8582.
[11] Lund, T. & Laland, S. (1990) Biochem. Biophys.
Res. Commun. 171, 342-7.
[12] Reeves, R., Langan, T. & Nissen, M. (1991) Proc.
Natl. Acad. Sci. USA88, 1671-5.
[13] Nissen, M., Langan, T. & Reeves, R. (1991) /.
Biol. Chem. 266, 19945-19952.
Bulletin of Magnetic Resonance
[14] Giancotti, V., Pani-D'Andrea, P., Berlingieri, M. T.,
Di Fiore, P. P., Fusco, A., Veccio, G., Philip, R.,
Crane-Robinson, C, Nicolas, R. H., Wright, C. A., &
Goodwin, G. H. (1987) EMBO J. 6,1981-1987.
[15] Giancotti, V., Buratti, E., Perissin, L., Zorzet, S.,
Balmain, A., Portella, G., Fusco, A., & Goodwin,
G. H. (1989) Exp. Cell Res. 184, 538-45.
[16] Elton, T. & Reeves, R. (1986) Anal. Biochem.157, 53-
62.
[17] Vartiainen, E., Palvimo, J., Mahonen, A., Linnala-
Kankkunen, A. & Maenpaa, P. (1988) FEBS Lett
228, 45-48.
[18] Bussemakers, M., van de Ven, W., Debruyne, F. &
Schalken, J. (1991) Cancer Res. 51,606-611.
[19] Grummt, F., Hoist, A., Muller, F., Wegner, F.,
Schwender, S., Luksza, H., Zastow, G., & Klavinius,
A. (1988) Cancer Cells 6,463-466.
[20] Wegner, M., Zastrow, G., Klavinius, A., Schwender,
S., Muller, F., Luksza, J., Hoppe, J., Wienberg, J. &
Grummt, F. (1989) Nucleic Acids Res. 17, 9909-9932
[21] Yang-Yen, H. & Rothblum, L. (1988) Mol. Cell. Bio
8, 3406-3414.
[22] Eckner, R. & Bimstiel, M. (1989) Nucleic Acids Res.
17, 5947-59.
[23] Ptashne, M. (1988) Nature 335,683-689.
[24] Lehn, D., Elton, T., Johnson, K. & Reeves, R. (198*
Biochem. Int. 16, 963-971.
[25] States, D. J., Haberkorn, R. A., & Ruben, D. J. (198:
/. Magn. Reson. 48, 286-293.
[26] Ranee, M., S0rensen, O.W., Bodenhausen, G., Wagne
G., Ernst, R. R., & Wuthrich, K. (1983) Biochem.
Biophys. Res. Commun. 117, 479-485.
[27] Bax, A., & Davis, D. G. (1985) J. Magn. Reson. 65,
355-360.
[28] Kessler, H., Griesinger, C, Kerssebaum, R., Wagner,
R., & Ernst, R. R. (1987) J. Am. Chem. Soc. 109,
607-609.
[29] Jeener, J., Meier, B. H., Bachmann, P., & Ernst, R. 1
(1979) /. Chem. Phys. 71, 4546-4553.
[30] Bodenhausen, G., Kogler, H., & Ernst, R. R. (1984;
/. Magn. Reson. 58, 370.
[31] Dyson, H. J., & Wright, P. E. (1991) Annu. Rev.
Biophys. Biophys. Chem. 20, 519-538.
[32] Wuthrich, K., Billeter, M., & Braun, W. (1984) /.
Mol. Biol. 180, 715-740.
[33] Mayo, K. M., Parra-Diaz, D., McCarthy, J. B., &
Chelberg, M. (1991) Biochemistry 30, 8251-8267.
[34] Grathwohl, C. & Wuthrich, K. (1976) Biopolymers
15, 2025-2041.
[35] Grathwohl, C. & Wuthrich, K. (1981) Biopolymers
20,2623-2633.

Vol. 14, No. 1-4 175
Solution Structure of the
DNA-binding Domain of GAL4
from Saccharomyces cerevisiae
James D. Baleja, V. Thanabal, Ted Mau, and Gerhard Wagner
Department of Biological Chemistry and Molecular Pharmacology
1 Introduction
The GAL4 transcriptional
activator protein has long been a
favorite for the study of
transcription in eukaryotic biology.
Genetic studies reveal a modular
architecture for the protein with
different functions associated with
each module [1]. A DNA-binding
domain of the protein recognizes and
binds to a sequence of DNA termed
the Upstream Activating Sequence
(UASQ). Other parts of the protein are
relevant for activation. They interact
with the transcriptional machinery
including RNA polymerase to activate
transcription. The UASQ is near the
genes that encode the proteins
required for galactose utilization.
Upon presentation of galactose to the
yeast cell, this DNA site is specifically
bound by GAL4, the transcription
function of RNA polymerase is
activated, and enzymes required for
galactose utilization are produced [2].
As a dimer of 881 amino acids,
GAL4 is too large for the
determination of a high-resolution
NMR structure and we have instead
studied a fragment containing the Nterminal
65 amino acid residues
Harvard Medical School
Boston, Massachusetts 02115
including its DNA-binding domain
[3]. In the absence of DNA, GAL4(65)
is monomeric in solution,
presumably because it does not have
the amino acid residues necessary for
dimerization [4]. GAL4(65) is dimeric
when bound to any of four DNA sites
present in the UAS [5]. Each of these
binding sites is approximately twofold
palindromic, and like many
other dimeric DNA-binding proteins,
each monomer of the GAL4 dimer
interacts with a half-site of DNA.
The structure and dynamics of the
monomeric DNA-binding domain of
GAL4 (residues 1-65; Figure 1), an
investigation of the binding to a DNA
half-site (Figure 2), and the structure
of the resultant protein-DNA complex
are presented in this paper.
1 11 21
MKLLSSIEQA CDICRLKKLK CSKEKPKCAK
31 41 51
CLKNNWECRY SPKTKRSPLT RAHLTEVESR
61
LERLE
Figure 1. Primary amino acid
sequence for the DNA-binding
domain of GAL4.

176
1 2 3 4 5 6 7 3 9 10
C C G G A G G A C T
G G C C T C C T G A
20 19 18 17 16 15 14 13 12 11
Figure 2. Nucleotide sequence for the
one-half DNA-binding site of GAL4.
2 1H, 15N, and n^Cd NMR
resonance assignments
Assignment of specific nuclei to
the observed NMR resonance
frequencies is the first step in
determining the structure of a
protein using NMR techniques [6].
Assignments for the 1 H, ^$N, and
113 Cd NMR resonances of GAL4(65)
were made using homonuclear and
hetero-nuclear NMR experiments
and by following standard protocol
([6], Baleja and Wagner, unpublished
results).
Zinc is required for the DNAbinding
activity of GAL4 [4]. It can be
replaced by NMR-active 113 Cadmium
without loss of DNA-binding. The
113 Cd NMR spectrum (Figure 3)
shows that two metal ions are
coordinated by a Cys - (X)2 - Cys - (X>6
- Cys - (X)6 - Cys - (X)2 - Cys - (X)6 -
Cys motif using six cysteines and
forming a bimetal-thiolate cluster
[7]. Hetero-nuclear correlation
experiments between the l* 3 Cd and
*H define the liganding of the two
central metal ions (Figure 3).
Having assigned resonance frequencies
to specific nuclei, the
correspondence of cross peaks to the
protons can be made and local
structural information was can then
be determined. A section of the twodimensional
NOESY spectrum is
shown in Figure 4. The region shows
the cross peaks among the amide
protons of the protein backbone and
protons of the aromatic sidechains.
BuJietin of Magnetic Resonance
Figure 3. Heteronuclear ^ C d H
correlation experiments. The standard
reverse INEPT pulse sequence
[8] was followed by a short MLEV-17
TOCSY transfer [9] and proton
detection./
Indicating the presence of two ahelices,
there are a series of close
approaches between amide proton
resonance frequencies for several
sequential residues. Other NOEs arise
between amino acid residues more
distant in sequence and define the
unique topological features for the
protein. NOE intensities at 56, 150, and
250, and 500 millisecond mixing times
were converted to the corresponding
interproton distances using the usual
inverse r 6 relationship [3]. In case of
overlap in the homonuclear NOESY,
spectrum analysis was also made
using a three-dimensional ^JSJ.IJJ.IH
NOESY-HMQC spectrum recorded on a
uniformly ^N-labeled protein
(Figure 5, [10]). From the NOESY data,

Vol. 14, No. 1-4
residues 9-40 are observed to form a
well-defined, compact globular cluster.
The terminal residues 1-8 and 41-
66 show little persistent structure in
solution. There are many interresidue
NOE crosspeaks for the
central core (30 per residue, on
average), but only a few weak NOE
cross peaks for protons in the
disordered region. In addition, the
amide protons of the flexible trails
are in rapid exchange with solvent
[10] and have narrow resonance
lines, indicating that these residues
have considerable conformational
mobility in the absence of DNA. This
unstructured character for parts of
the GAL4 protein may be vital to its
biological function. On the other
hand, at least some aspects of this
mobility may merely be a
consequence of the truncation used
for GAL4 in this study. Nonetheless,
our picture of GAL4 is one in which
two N-terminal recognition modules
are connected by flexible linkers to a
dimeric core [4],
-©V
o
K23
o
J28
R15
D1i! L1
O
E37
OC11
S41
DS22
>C21
C38,
9.0 8.5 8.0 7.5 7.0
6)2 (ppm)
6.5 6.0
Figure 4. NOESY spectrum of amide
and aromatic protons of GAL4(65).
Sample concentration was 1.5 mM in
0.2 M NaCl, 20 mM sodium phosphate,
pH 7.0 at 25°C. Numbers indicate the
residues for which sequential strong
amide to amide cross peaks are
observed which are indicative of ahelices.
K&O T»K27
,©113
C31

178
3 Structure of the GAL4
DNA-binding domain
Sets of interproton distances and
torsion angles formed the basis for
structure determination. Distances
were determined from the NOESY data
[3]. § torsion angles were interpreted
from measured NH-oc coupling constants
and x 1 angles were derived
from ^N-j} and a-(5 coupling constants
[3]. Sulfur-Cd liganding distances
were imposed to be 2.35 to 2.45 A, Cys
Cp-Cd distances to be less than 3.4 A,
and sulphurs liganding the same Cd
to be between 3.3 and 4.2 A. 614
distance and 41 torsion angle
measurements were used with the
distance geometry package DG-II to
generate a set of structures (0.6 A
backbone atom rmsd) in agreement
with the NMR data [3]. A schematic of
the structure (Figure 6) shows the
two central metal ions coordinated by
the six cysteines. The DNA-binding
domain consists of an a-helix and an
extended structure, then a sharp turn
that contains a cis-proline bond, and
then a 2nd a helix followed by an
extended structure. If the central
metal binding subdomain of GAL4 is
split, and the two halves
superimposed, a striking correspondence
between each part is revealed.
The conformation of the 13 residue
segment from residues 10 to 22 is
almost identical to that of residues 27
to 39, with an rmsd of 0.8 A for the
backbone atoms. Although the
structural integrity of the protein
would be provided mainly by the way
in which the two metal ions are
liganded, some hydrophobic packing
is observed with the side chains of
W36 and Y40 [3]. Slowly exchanging
amides of GAL4, in both the free and
DNA-bound forms [10], can be
attributed, in part, to hydrogen
bonding to carbonyl oxygens within
the a-helices, and to sulfurs of the
cysteinyl side-chains [12].
Bulletin of Magnetic Resonance
Figure 6. Structure of the central
core for the DNA-binding domain of
GAL4. Cysteines that ligand the two
central metal ions are numbered.
4 Formation of a GAL4-DNA
complex
Imino protons of one half of the
consensus dimeric GAL4 DNAbinding
were monitored in forming
the complex between the protein and
DNA (Figure 7). One imino proton is
present for each base-pair of a 10
base-pair DNA duplex.
_JV
13.8 13.6 13.4 13.2 13.0 12.8 12.6 ppm
Figure 7. Titration of DNA with GAL4.
Experimentally, eight imino proton
resonances are observed, since the
imino protons on the base-pairs at
each end of the duplex exchange
rapidly with the bulk solvent at room
temperature. Each imino resonance
has been assigned to a specific base-

Vol. 14, No. 1-4
pair. Resonances shift as the
environment around each proton
changes upon addition of GAL4. All
resonances broaden as the more
slowly protein-DNA complex is
formed. The equilibrium dissociation
constant is 169± 13 |J.M. Resonances
are in fast exchange between the
protein-DNA complex and the free
components indicating rapid equilibrium
between free and bound
forms.
NOESY and TOCSY two-dimensional
data for the protein-DNA complex
14.0 12.0 10.0 8.0
F2
6.0
(ppm )
have been collected (Figure 8). These
spectra are promising for full
analysis since the resonance lines
are not too broad for resonance
assignment, chemical shift dispersion
is good, and solubility of the
protein-DNA complex is adequate. The
DNA resonances have been completely
re-assigned and the protein
resonances have been partially reassigned
within the protein-DNA
complex.
4.0 2.0
Figure 8. NOESY spectrum of the GAL4:DNA complex. The concentration of the
complex was approximately 0.5 mM. Buffer conditions were 0.15 M NaCl, 20 mM
PO4, pH 7, 25°C.
5 NMR structure of
GAL4-DNA complex
the
Several NOE contacts have been
observed between protein and DNA
(dashed lines, Figure 9) yielding a
preliminary stucture for this GAL4-
DNA complex. The recognition helix
for the free form of the protein was
docked onto a B-DNA. The amino acid
residues responsible for recognizing
DNA are part of the metal-binding
cluster and interact with edges of
'base-pairs exposed in the
major
groove of the DNA.
The structure of the DNA-binding
domain of GAL4 (residues 1-65) bound
to full site DNA containing two GAL4
binding sites has been determined
crystallographically. In agreement
with our observed intermolecular
NOE cross peaks (Figure 8), the Cterminal
end of the first a-helix of
the metal binding cluster (residues 9-
40) provides for sequence-specific
re-cognition of DNA by GAL4. When
bound to full site DNA, parts of
GAL4(65), before unstructured in
solution, adopt a regular conformation.
Residues 41 to 49 form a
linker region which interacts with
179

180
the backbone of the DNA. In addition,
residues 50-64 form a small dimerization
domain using a coiled-coil type
packing arrangement [4]. Our preliminary
results on an intact
dimerization element for GAL4
(residues 50-106) indicate that the ahelical
character for the coiled-coil
of the protein extends beyond residue
64 (Baleja, Marmorstein, and Wagner,
unpublished results). The same
conformation (1.1 A rmsd) is observed
for the recognition module of
GAL4(65) in solution using NMR
techniques as for the central metalbinding
cluster of GAL4(65) bound to
DNA using crystallographic techniques.
Thus the core of the DNAbinding
domain changes little upon
binding DNA.
Figure 9. Preliminary structure for a
GAL4-DNA complex.
6 Summary
The DNA-binding domain of GAL4
has many interesting features. It is a
novel DNA-binding motif with a
globular two-metal cluster that has
hydrogen-bonds to sulfur and twofold
internal symmetry. GAL4 reads
the sequence of DNA through the
amino acids present at the C-terminal
Bulletin of Magnetic Resonance
end of the first a-helix. The two DNAreading
modules of intact GAL4 are
tethered by flexible linkers to the
central body of the protein. Once
bound to DNA through the Nterminal
recognition domains, the Cterminal
portion of GAL4 is in the
correct position to interact with the
components of the transcriptional
machinery to bring about transcriptional
activation.
7 References
1. Keegan, L., Gill, G., and Ptashne,
M., Science 231, 699 (1986).
2. Johnston, M., Micro. Rev. 51, 458
(1987).
3. Baleja, J. D., Marmorstein, R.,
Harrison, S. C, and Wagner G.,
Nature 356, 448 (1992).
4. Marmorstein, R., Carey, M.,
Ptashne, M., and Harrison, S. C,
Nature 356, 408 (1992).
5. Carey, M., Kakidani, H.,
Leatherwood, J., Mostashari, F.,
and Ptashne, M., J. Mol. Biol. 209,
423 (1989).
6. Wuthrich, K. NMR of Proteins
and Nucleic Acids (Wiley, New
York, 1986).
7. Pan, T., and Coleman, J. E.,
Biochemistry 209, 3023 (1990).
8. Briihwiler, D., and Wagner, G., J.
Magn. Reson. 69, 546 (1986).
9. Bax, A., and Davis, D. G., J. Magn.
Reson. 65, 355 (1985).
10. Mau, T., Baleja, J. D., and Wagner,
G., Protein Science (submitted).
11. Bodenhausen, G., and Ruben, D. J.,
Chem. Phys. Lett. 69, 185 (1980).
12. Kraulis, P., Raine, A. R. C
Gadhavi, P. L., and Laue, E. D.,
Nature 356, 448 (1992).

Vol. 14, No. 1-4
STRUCTURAL INVESTIGATION OF FOLIC ACID BY
NMR PROTON RELAXATION AND MOLECULAR
MECHANICS ANALYSIS
1 Introduction
Claudio Rossi, Alessandro Donati, Sergio Ulgiati*
and Maria Rosaria Sansoni
Department of Chemistry, University of Siena, Pian dei Mantellini, 44
53100 Siena ITALY
•Department of Chemistry, University of Sassari, Via Vienna, 2
07100 Sassari ITALY
Folic acid N-[4-(2-Amino-4-hydroxypteridinyl-(6)-methylamino)-benzoyl]-Laminoglutaric
acid (Figure 1) is a
fundamental coenzyme involved in onecarbon
unit transfer processes 1 . The solid
state conformation of folic acid has been
defined but there have been few
investigations on the structure of this
coenzyme in solution^.
12
by Nuclear Magnetic Resonance (NMR).
Information on dynamical motion,
magnetic dipolar connectivities and
energy minimization calculations are
combined in order to define the solution
structure of folic acid.
2 Experimental
Two-dimensional COSY, NOESY and Hetcor
Figure 1: Structure and numbering of folic acid.
f ole is also fundamental in reductive
zymatic processes in which
fchydrofolate (the reduced form of folic
) is oxidized to dihydrofolate and folate.
•aihydrofolate reductase NADPH-
•ndent enzyme controls the biological
of tetrahydrofolate 3 .
present paper the conformational
of this molecule were analyzed
experiments were obtained by (3C/2-ti -7C/2-
AT)n, 4 (rc/2-tl-TC/2-tm-JU/2AT)n 5 a n d
pulse sequences respectively. Spin-lattice
relaxation rates were measured using the
(180°-T-90°-t)n pulse sequence. The NMR
measurements were performed using a 0.12
mol.dm' 3 DMSO-d6 solution at 27°C. NMR
spectra were recorded on a Varian XL-200
181

182
and a Bruker AMX-600 spectrometers
operating at 200 and 600 MHz respectively.
Molecular mechanics calculations were
computed by the MacroModel program,
version 2.5^ implemented on a Vax 11/750
computer. The force field used was that
reported by Weiner et al^.
3 Results and Discussion
In the present paper the NMR properties of
folic acid were investigated in depth in
order to define the molecular structure in
DMSO-d6 solution.
As the proton and carbon assignments of
folic acid refer only to water solution at
basic pH^^O, both proton and carbon
chemical shifts in DMSO-d6 were
determined by a method based on
conventional two-dimensional COSY and
Hetcor experiments. Figures 2 and 3 show
the COSY and Hetcor spectra of folic acid.
The complete proton assignments and
chemical shifts of protonated carbons can
be determined from these sets of data. The
quaternary carbons were assigned by a
frequency-dependent selective protoncarbon
NOE experiment 11 > 1 ^.
The results obtained are reported in Table 1.
The strategy for structural analysis was
based on a combined approach. First we
studied the dynamical properties of folic
acid in solution which led to two possible
scenarios:
i) the molecular motion is subject to
different degrees of freedom, in which case
each molecular moiety behaves
independently as a consequence of the lack
of "ordered" elements;
ii) the molecule is subject to overall
dynamical reorientation, characterized by
a single rotational correlation time, %c.
These conditions, for molecules of the size
of folic acid, are verified whenever noncovalent
interactions stabilize specific
conformations.
The appropriate method for dynamical
investigation is based on analysis of the
carbon spin-lattice relaxation rate, Ric.
The experimental Ric> calculated for
protonated carbons, are reported in Table 1.
From these data and using the Allerhand's
approach 1 3 a unique correlation time
value of 3x10'^ s was calculated. In Table 1
the selective and non-selective proton
Bulletin of Magnetic Resonance
relaxation rates are also reported. These
experimental values confirm the dynamical
region of the isotropic molecular motion of
folic acid in solution. A second set of
structural information can be derived from
the study of the extent of the dipolar
magnetization transfer in the protonic
environment.
In this case the NOESY spectrum can
provide the complete network of the
proton-proton dipolar interactions, which
is related to internuclear distances. The
analysis of the NOESY spectrum enables us
to identify proton pairs in which the
cross-relaxation contribution is
significant. These include the NH(ifj)-
H(12/16). NH(io)-H(9), H(i2/16)-H(9),
H(12/16)-H(13/15) and H( i 8)-H(2 lb).
Different cross-peaks due to exchange
contributions between two different sites
can also be detected in the NOESY spectrum.
These cross-peaks are related to the
NH(1O)/NH2(2) — HOD exchange.
Information on the extent of dipolar
interactions can be used as experimental
"constraints" in theoretical energy
minimization calculations. The presence of
an exchange process selectively restricted
to the NH(iO)/NH2(2) — HOD protons
suggests the involvement of other
exchangeable nuclei such as NH(i8) in non
covalent interactions (e.g. hydrogenbonds),
important for the stabilization of
the conformation of folic acid in solution.
Further evidence of the slow chemical
exchange process of NH(J8) with respect to
NH (10)/NH2(2) can be obtained from
saturation transfer experiments.
By irradiating the HOD resonance for a
sufficient period of time, with a selective
frequency, a strong reduction in signal
intensity is observed on protons involved
in exchange phenomena. The experimental
findings show that the NH(1O)/NH2(2) .
signal is drastically affected by HOD ;
saturation whereas NH(18) does not show ]
significant intensity variation. This is
further evidence of the importance of |
NH(i8) in the stabilization of selective ;
conformation by hydrogen bond with the J
C(23) carboxyl group of glutamic acid. Thi
hypothesis is confirmed by the selectivity j
in NH(i8)-H(2l) dipolar connectivities,
observed in the NOESY spectrum. In fact t
specific NH(i8)-H(21b) cross-peak
observed, suggesting the stabilization of ^
unique conformation in solution.

Vol. 14, No. 1-4
F2 (PPM)
9 8 7 6 5 4 3 2 }
Figure 2: COSY spectrum of 0.12 mol.dm-3 folic acid DMSO-d6 solution at 27°C.
12S 1M
Figure 3: 1H-13C Hetero correlation
' solution at 27°C.
I I '—I—I—|—i—t-
spectrum of folic acid in DMSO-d6
183

184 Bulletin of Magnetic Resonance
Table 1
Proton and carbon NMR parameters of 0.12 mol.dm'3 folic acid solution at 27°C.
Nuclei
2
4
4a
6
7
8a
9
10
11
12
13
14
15
16
17
18
19
20
21A
21B
22
23
8 ppm
8.75
. .
4.59
7.02
6.74
7.75
7.75
6.74
8.22
4.44
2.15
2.01
2.42
—
8 ppm
156.160
161.274
127.945
148.610
148.610
153.823
45.922
150.793
111.216
128.998
121.321
128.998
111.216
166.438
51.762
173.744
26.045
26.045
30.439
173.932
Further evidence of the structure assumed
by folic acid can be obtained from
molecular mechanics calculations using
the Macro Model program and experimental
NMR constraints, a low energy
conformation of -135.5 KJ/mol was
computed after several iterative and
minimizzation cycles.
Figure 4: Solution structure of folic acid
as determined by NMR experimental
constraints and subsequent energy
minimization calculations.
Ric
s-1
0.11
0.21
0.12
0.55
11.9
0.73
6.12
5.70
0.43
5.70
6.12
0.38
6.07
0.37
11.75
11.75
11.60
0.30
RlNS
s-1
0.77
5.40
2.80
2.00
1.87
1.87
2.00
4.71
1.62
5.50
5.55
5.00
—
RlSE
s-1
0.75
--
5.30
1.46
1.34
1.34
1.46
1.17
5.40
5.40
4.90
The minimized structure calculated (Figure
4), shows a NH(18)-C(i9)-C(21)-H(21b)
torsion angle of 315°. This is in agreement
with the experimental NOESY data and
confirms the conformation of folic acid
stabilized in solution by the NH(i8)-C(23)
hydrogen bond.
4 References
1) R.L. Blakley; "The Biochemistry of Folic
Acid and Related Pteridines"; North-
Holland publishers, Amsterdam, (1969).
2) D. Mastropaolo, A. Camerman and H.
Camerman; Science 210, 334 (1980).
3) J.M. Blaney, C. Hansch, C. Silipo and A.
Vittoria; J. Am. Chem. Soc. £4, 303 (1984).
4) A.D. Bax, R. Freeman and G. Morris; J-
Magn. Reson. 42, 169 (1981).

Vol. 14, No. 1-4
5) D.J. States, R.A. Haberkorn and D.J.
Ruben; J. Magn. Reson. 4JL 286 (1982).
6) A.D. Bax and G. Morris; J. Magn. Reson.
42, 501 (1981).
7) C. Still; Macromodel, Columbia Universty
Molecular Modelling System (1987).
8) S. Weiner, P.A. Kollman, D.A. Case, U.C.
Singh, C.Ghio, G. Alagona and P.K. Weiner;
J. Am. Chem. Soc. 106, 765 (1984).
185
9) W. Frick, R. Weber and M. Viscontini;
Helv. Chim. Acta 52, 2658 (1974).
10) M. Poe; Method in Enzymology 6JL 483
(1980).
11) N. Niccolai, C. Rossi, V. Brizzi and W.A.
Gibbons; J. Am. Chem. Soc. 106, 5732 (1984).
12) C. Rossi, N. Niccolai and F. Laschi; J.
Phys. Chem. 9_i, 3903 (1987).
13) A. Allerhand, R.A. Komoroski; J. Am.
Chem. Soc. 9_5_, 8828 (1973).

186
Bulletin of Magnetic Resonance
Characterization of Water-in-Bitumen Emulsions
in Model Porous Media by NMR Microscopic Imaging Techniques
1. Introduction
Leslie H. Randall and George E. Sedgwick
Alberta Research Council,
Oil Sands and Hydrocarbon Recovery Division
PO Box 8330, Station F, Edmonton, Alberta, T6H 5X2.
Production from thermal (steam enhanced) oil
recovery processes is-complicated by the presence of
water-in-oil emulsions. [1], [2] Critical to monitoring
the in-situ formation and flow of these water-in-oil
emulsions is the ability to distinguish between the
various components (bitumen, water/steam and
emulsified water) present in a core flood experiment.
In principle, NMR imaging is ideally suited to
monitor the spatial distribution of absorbed fluids, and
in this regard, the viability of the NMR imaging
technique to examine the distribution of fluids in
reservoir rock samples has recently been
demonstrated. [3] - [15]. In general, several fluids or
phases may be present and the ability to distinguish
these components is of prime interest.
In the majority of these studies, the NMR
imaging technique has been applied to samples which
contain a low viscosity crude oil and water/brine. In
the case of heavy oil production, the fluids have a
high and varying viscosity, (ie: bitumen, water and
emulsions of bitumen and water). Under these
circumstances, it is possible that these differences in
viscosity will lead to the ability to discriminate
between phases via differences in NMR relaxation
behaviour. In addition, the influence that the solid
matrix has on the relaxation behaviour of water
dispersed as an emulsion may be quite different from
water absorbed into the porous medium. In an effort
to evaluate the feasibility of characterizing water-inbitumen
emulsions by NMR microimaging techniques,
and
Colin A. Fyfe
University of British Columbia,
Dept. of Chemistry and Pathology,
Vancouver, B.C., V6T 1Y6.
the one-dimensional NMR spectra, the relaxation time
constants and the spin-echo images for a series of samples
consisting of water, bitumen and water-in-bitumen emulsions
absorbed into glass beads were examined and are presented
herein.
2. Experimental
All water-in-oil (bitumen) emulsions were prepared from Cold
Lake bitumen which has been ultracentrifuged to remove solid
particles. The emulsion samples were created by passing a
heated mixture of bitumen and water through an auxiliary sand
column at a suitable flow rate. The emulsions were checked
under an optical microscope to ensure that the water phase was
well dispersed prior to packing. Eight samples were prepared
each consisting of a different fluid/porous medium mixture.
Sample 1 contained 10 mL of Cold Lake Bitumen. Sample 2
contained 10 mL of a 20 % (w/w) water-in-bitumen. The
remaining samples contained fluid absorbed into a glass bead
matrix. Two different sizes of glass beads were used. The
small glass beads were determined to be 88 -104 pm in
diameter which represents fine sand. The large glass beads
were 0.8 - 1.0 mm in diameter. Sample 3 contained
approximately 10 g of the large glass beads with 5 mL ol
distilled water. Sample 4 had a similar composition to sampl<
3 except that the smaller glass beads were used. Sample -
contained 10 g of the large glass beads with 5 mL of a water
in-oil emulsion (20% w/w water). Sample 6 had a simila
composition except that the smaller glass beads were used. T<
ensure that the composition of the water-in-bitumen emulsio)
was maintained in the glass bead matrix, samples 5 and 6 wep
prepared by mixing the emulsion with the glass beads by haflj

Vol. 14, No. 1-4 187
and then placing the appropriate amount of the
mixture at the bottom of a 10 mm NMR tube. The
samples were then spun at low speeds on a bench-top
centrifuge for 10 minutes to obtain a uniform packing.
In an effort to compare the NMR behaviour of the
two types of fluids directly, Sample 7 was composed
of two regions, the bottom layer contained emulsion
in large glass beads and the top layer contained
distilled water in large glass beads. Sample 8 was
identical in composition to sample 7 except that the
smaller glass beads were used. The contents of the
samples are summarized in Table 1.
Table 1. Composition of Samples
Sample Fluid Matrix
1 Bitumen
2 Emulsion
3 Distilled Water
4 Distilled Water
5 Emulsion
6 Emulsion
7 Emulsion/Water
8 Emulsion/Water
None
None
88 - 104 urn
0.8 - 1.0 mm
0.8 - 1.0 mm
88 - 104 urn
0.8 - 1.0 mm
88 - 104 um
All NMR measurements were made on a Bruker
MSL 400 spectrometer equipped with a microimaging
system using the proton microimaging probe
equipped with a vertical 12 mm saddle coil. The
nonselective 90° rf pulse length was 14.5 ps.
Quadrature phase cycling was used in all the
spectroscopic measurements. ID J H NMR spectra and
Carr-Purcell spin-echo (90-tau-180) NMR spectra [16]
were obtained to characterize the samples. The spinecho
sequence was also used to determine the average
spin-spin relaxation times (T2). The inversionrecovery
sequence [17] was used to determine the
average Tj spin-lattice relaxation times.
After evaluating the relaxation time constants
and the NMR lineshapes for several of the samples,
was determined that the spin-echo imaging
luence [18] was an appropriate choice. The spinio
imaging pulse sequence employs a 90-tau-180 rf
sequence in which a hard 180° pulse sequence
* 110 refocus ^ eff ects of field inhomogeneity.
selection was performed by using a selective
Pulse with an appropriate Gz gradient. Echo
fes for the spin-echo imaging experiments varied
from 4.5 ms to over 100 ms and the actual echo times are
indicated in the text. The slice thickness was typically 2.1
mm. The phase encoding gradient was incremented through
256 experiments. The frequency encode gradient was 5.8
G/cm resulting in an in-plane resolution of 95 um. The
samples used in this study had a porosity of approximately 30
-35 % and the imaging experiments required 1-4 hours to
acquire. Images presented in this paper follow the convention
in which an inverse gray scale is used to indicate relative
intensity. The darker the region on the image, the higher the
concentration of water.
3. Results and Discussion
The relaxation behaviour and linewidths were investigated
(Table 2) to determine the appropriate imaging sequence for
the bitumen and water-in-bitumen emulsions. A ID *H NMR
spectrum of Cold Lake Bitumen consisted of a large peak (Vj^
= 690 Hz) which is assigned to the heavy oil component
(bitumen) and a small shoulder which is attributed to trace
amounts of connate water. The spin-spin relaxation time (T2)
of the oil component was determined to be 1.3 ms which
indicates that quantitative spin-echo imaging of the bitumen
component in the samples will not be possible.
The linewidth determined for distilled water placed in the
small glass beads (Sample 3) is approximately 1000 Hz, an
increase of more than 2 orders of magnitude over that observed
for a distilled water phantom. Similar line broadening was
observed for the other samples (Table 2). The line broadening
is a result of the magnetic susceptibility differences between
the solid matrix and the absorbed fluids. These line-widths
indicate that the application of the gradient echo sequence [18]
would be difficult due to the short T2* values (T2* = 300-800
ps). The spin-spin relaxation parameter for the distilled water
absorbed into either glass bead matrix (Table 2) is markedly
reduced from that observed for bulk water. An NMR image
of Sample 4 with an echo time of 4.5 ms demonstrates that the
distribution of water in these types of samples are possible.
(Figure la) The image with an echo time of 104 ms (Figure
lb) is more interesting. It shows a large decrease in signal
intensity, particularly where the distilled water is in contact
with the glass beads. Due to the non-uniform packing of the
large glass beads, small pockets of water on the order of 100 -
400 um in diameter can be observed.
An examination of the NMR relaxation parameters of the
20% (w/w) water-in-bitumen emulsion (Sample 5) absorbed
into the large glass bead matrix reveals that the T2 relaxation
time constant of the water phase is unaffected by placing the
emulsion into the glass bead matrix. The image obtained with
a 104 ms echo time (Figure 2) displays a strong uniform signal
intensity. The bitumen component of the sample has a T2
relaxation time constant on the order of a millisecond, and thus
does not contribute to the intensity of the image.

i ii
'I,' , ',
il i I
i i
188 Bulletin of Magnetic Resonance
Figure 1 (a) Spin echo images of sample 4.
a) Echo time = 4.5 ms. (b) Echo time = 104 ms
During a core-flood experiment it is likely that both
water and emulsified water phases will be present.
The ability to distinguish between bulk absorbed
water and emulsified water is considered essential to
analyzing the formation and flow of the emulsions in
sand packs. To this end, a sample which contains
both distilled water and water-in-bitumen emulsion in
a glass bead pack was examined (Samples 7 and 8).
The conditions of the imaging experiments were
chosen such that only the water component of the
samples were observed. Using an echo time of 4.5
ms the distribution of water in sample 7 (Figure 3a)
can be obtained. The distilled water plus large glass
beads are in the top half of the NMR tube and this
results in a much higher signal intensity due to the
higher water concentration. To discriminate between
the emulsified water phase and the absorbed bulk
water, an echo time of 104 ms was used. (Figure 3b)
Under these conditions, only water which has a high
mobility or low surface contact will be observed (ie:
water which is emulsified will be favoured). This
results in an image in which small pockets of water
are observed in the distilled water region, while the
emulsion containing region appears nearly uniform.
The effect of the smaller grain size of the
relaxation parameters was examined using sample 8.
The smaller glass beads have a more uniform pore
size distribution and are more representative of the
matrix used in core flood experiments. In such
samples, the pore size is typically on the order of 30
pm which means that all of the water will be in
intimate contact with the solid matrix.
Figure 2. Spin-echo image of sample 5.
Echo time = 104 ms.
Figure 3. Spin echo images of sample 7.
a) Echo time = 4.5 ms. (b) Echo time = 104 ms
The T2 of the distilled water component is reduced to 4.9 ms,
on the same order of magnitude as the echo time (4.5 ms).
This means that the intensity of the water will be strongly T2
weighted and the region which contains the distilled water will
no longer have an intensity which is much higher than the
emulsion. (Figure 4a) When the echo time is increased to 34
ms, (Figure 4b) the region which contains the distilled water
disappears in a uniform manner. The smaller grain size of the
solid matrix has enlarged the relaxation time differences
between the bulk absorbed water and the emulsified water and
thus the contrast between the two physical states are enlarged.

Vol. 14, No. 1-4 189
Figure 4. Spin echo images of sample 8.
a) Echo time = 4.5 ms. (b) Echo time = 34.5 ms
3. Summary
In the present study, the feasibility of examining
heavy oil emulsion samples has been demonstrated.
Bitumen, water and emulsified water placed into glass
beads are easily distinguished in an NMR imaging
experiment on the basis of their relaxation times. The
spin-spin relaxation time constant, T2 observed for the
water component of the emulsion samples placed in
contact with glass beads was dramatically different
than that found for distilled water. The NMR
experiment is therefore sensitive to the physical state
of the water, ie: whether the water is emulsified and
therefore 'protected' from the solid matrix. This
difference in relaxation times was exploited to provide
water-selective images of the emulsified water phase
in the presence of oil and non-emulsified water
phases. Good S/N images can be obtained with high
in-plane resolution (50-100 pm) can be obtained on a
microimaging system in reasonable time periods.
Saturation profiles can be obtained in a manner of
seconds which would allow for the continuous
monitoring of the formation of an emulsion under
flow conditions. The ability to monitor the formation
and flow of water emulsions will be important to
understanding enhanced oil recovery processes.
Table 2: Proton Relaxation Times and Linewidths of Water
and Bitumen
Sample
1
2 water
bitumen
3
4
5 water
bitumen
6 water
bitumen
Ti(s)
0.71
2.2
0.62
2.4
2.3
1.6
0.58
1.5
0.45
T2 (ms)i
1.3
200*
1.8
29
4.9
495
1.3
484
1.4
v l/2
690
150
650
350
1000
The water component of emulsion samples show a variation
in T2. The reasons for this variation are under investigation
but are believed to be due to the size distribution of the water
droplets.
Acknowledgements
The authors wish to acknowledge G. A. Kissel and D. Vu of
the Alberta Research Council for the invaluable technical
assistance in preparing the samples. The financial assistance
of NSERC is acknowledged (CAF).
4. References
1. Thermal Recovery of Oil and Bitumen, Butler, R. B.
Prentice Hall, New Jersey, 1991.
2. Emulsions; Fundamentals and Applications in the Petroleum
Industry, Schram, L. L., Ed. American Chemical Society, 1992.
3. Rothwell, R. P.; Vinegar, H. J. Applied Optics, 24, 1985,
3969-3972.
4. Blackband, S.; Mansfield, P.; Barnes, J. R.; Clague, A. D.
H.; Rice, S. A. Soc. Pet. Eng. Form. Evaln.l, 1986, 31-34.
5. Hall, L. D.; Rajanayzgmi, V.; Hall, C. /. Magn. Res. 68,
1986, 185-188.
6. Baldwin, B. A.; Yamanashi, W. S. Mag. Res. Imaging, 6,
1988, 493-500.

190 Bulletin of Magnetic Resonance
7. Hall, L. D.; Rajanayagmi, V. /. Magn. Res. 74,
1987, 139-147.
8. Chen, J. D.; Dias, M. M.; Patz, S.; Schwartz, L. M.
Phys. Rev. Lett. 61, 1988, 1489-1492.
9. Edelstein, W. A.; Vinegar, H. J.; Tutunjian, P. N.;
Roemer, P. B.; Mueller, O. M. SPE preprint 18272,
63rd Annual Technical Conference, Houston, 1988,
101-112.
10. Mandava, S. S.; Watson, A. T.; Edwards, C. M.
Amer. Inst. Chem. Eng. 36, 1990, 1680-1686
11. Majors, P. D.; Smith, J. L.; Kovarik, F. S.;
Fukushima, E. /. Magn. Res. 89, 1990, 470-478.
12. Woessner, D. E.; Gleeson, J. W.; Jordan, C. F.
SPE preprint 20493, 65th Annual Technical
Conference, New Orleans, 1990, 247-253.
13. Osment, P. A.; Packer, K. J.; Taylor, M. J.;
Attard, J. J.; Carpenter, T. A.; Hall, L. D.; Herrod, N.
J.; Doran, S. J. Phil. Trans. R. Soc. Lond. A. 333,
1990, 441-452
14. Dereppe, J. M.; Moreaux, C; Schenker, K. J.
Magn. Res. 91, 1991, 596-603.
15. Dechter, J. J.; Komoroski, R. A.; Ramaprasad, S.
/. Magn Res. 93, 1991, 142-150.
16. H. Y. Carr and E. M. Purcell, Phys. Rev., 94,630
(1954).
17. R. L. Void, J. S. Waugh, M. P. Klein and D. E.
Phelps, J. Chem. Phys., 48, 383 (1968).
18. Edelstein, W.A.; Hutchinson, J.M.S.; Johnson G.;
Redpath, T. Phys. Med. Bioi, 25, 751 (1980).
19. A. Haase, J. Frahm, D. Matthaei, W. Hanicke and
K. D. Merboldt, /. Magn. Reson., 67, 258 (1986).

Vol. 14, No. 1-4 191
COMPUTER GRAPHICS FOR PULSE SEQUENCE
ANALYSIS
Introduction
Jonathan Callahan, Debbie Mattiello and Gary P. Drobny
The development of software for
the simulation of NMR experiments
has increased at an enormous pace in
the last decade. Its usefulness has
spread far beyond the analysis of
lineshapes and spectra. Today,
search-and-optimize strategies are
used to develop new pulse sequences
while other programs measure the
performance of pulse sequences on
spin systems of interest [1, 2]. Up to
now, most of the output from these
programs has been displayed as twodimensional
hard-copy output. With
the arrival of relatively inexpensive
graphics workstations, the possibility
of visualizing the time development of
the density operator has spurred us to
develop graphics software in
conjunction with our ongoing
development of simulation software.
Current projects which benefit
from graphical analysis include the
development of "time-suspension"
sequences for use with solids imaging,
development of improved sequences
for the creation of Zeeman or
quadrupolar order in deuterium NMR,
and analysis of artefacts seen in
imaging experiments in the presence
of flow.
Chemistry Dept. University of Washington
Seattle, Washington 98195 USA
Time Suspension
"Time suspension" sequences are
multi-pulse sequences which remove
the effects of both the dipolar coupling
and the chemical shift Hamiltonians
[3-5]. They are useful in pulsedgradient
imaging experiments where
imaging gradients are applied only
during the multi-pulse windows [6-8].
Currently implemented pulse
sequences apply average Hamiltonian
theory and use carefully cycled
"wahuha" subcycles to achieve the
suppression of internal Hamiltonians
[3]. Like other multi-pulse
experiments these sequences perform
best at or near resonance and show
decreased line-nawowing as one
moves off resonance. A new "timesuspension"
sequence developed with
computer search-and-optimization
techniques (CDIS-4) [9] achieves
comparable reduction of the internal
Hamiltonians but shows
complementary behavior: poor line
narrowing on resonance but improved
performance off resonance. In an
effort to understand this behavior we
calculate the evolution of the density
operator for a two-spin system under
the influence of effective Hamiltonians
defined by the multi-pulse

192 Bulletin of Magnetic Resonance
propagators associated with each
sequence.
With perfect wahuha type
sequences the trajectory of the net
magnetization should be around the
base of a cone whose axis is along the
cube diagonal in spin space (ie. in the
nodal plane of the coupling ternsor).
When chemical shift offsets are small
(

Vol. 14, No. 1-4 193
greater than 10KHz the conventional
sequence shows modulated rotation
(Fig. 1C) whereas the new sequence
traces out a somewhat bent circle
lying in the nodal plane of the
chemical shift tensor (Fig.ID). From
such pictures we hope to gain a better
understanding of the effect of error
terms which are not easily amenable
to analytical treatment.
Deuterium
The deuterium quadrupole is an
excellent probe of dynamics and as
such is synthetically incorporated into
DNA and other biologically important
molecules [10]. The same quadrupole
which allows one to use deuterium as
a probe also presents formidable
experimental difficulties when the
strength of the quadrupole approaches
the strength of the rf field. This
situation is realized in some of our
labeled oligonucleotides. In this
regime, the effective axis about which
the magnetization is rotated (the sum
of rf and quadrupolar terms) is
substantially different from the rf
axis.
Current pulse sequences which
convert Iz magnetization into -Iz or
into quadrupolar order are found to
be inefficient at higher values of the
quadrupole coupling. For an inversion
pulse this means incomplete inversion
at the shoulders of the powder
pattern. This is unacceptable in
experiments where one attempts to
measure the orientation dependence
of spin-lattice relaxation time. Such
measurements are necessary to prove
or disprove particular dynamical
models.
An example of non-uniform
excitation is given in Fig. 2) which
shows the development of spin
coherences during a 180 degree pulse
at five different values of the
quadrupole coupling (OKHz, +/- 62.5
KHz and +/- 125KHz). The operator
basis of Vega and Luz [11] is used
because of the simple form of the
quadrupolar operator in that basis:
Ix
Jx=IyIZ+IzIy
Kx=Iy 2 -Iz 2 Ky=Iz IVy=lZ —1X
x-lx -1 ^y-Iy -1
2 -Ix 2 K2=Ix2-Iy 2
QZ=Iz 2 -I 2
Rotation about Ix during the
pulse is modified by (0q dependent
rotation about the Qz axis. This causes
a buildup of "antiphase" Jy and zeroquantum
Jz. By the end of the pulse it
is clear that a simple n pulse is
ineffective at creating -Iz over the
entire range of couplings (+/-125 KHz).
We are again using search and
optimize strategies to find pulse
sequences which create -Iz or Qz
evenly over a broad range of
quadrupole couplings [12]. With its
small operator basis, deuterium NMR
provides us with an excellent system
on which to develop interactive
computer aided pulse sequence
design. With user control of pulse
amplitude and phase, rapid calculation
of the time evolution of the spin 1
density operator and a graphical view
of Hilbert space we will soon have the
opportunity to design pulse
sequences and shaped pulses
intelligently rather than relinquishing
our insight to the cpu.

194
Qz
\lx
Jx
Iz
iy
Bulletin of Magnetic Resonance
Figure 2) Evolution of spin coherences for deuterium with ()(•), +/- 62.5(AY) and +/-
KHz quadrupole coupling. The evolution is depicted in the operator basis of Vega and Lus
Rotation about Ix due to the rf pulse is seen but rotation about quadrupole operator Qz is als
evident. For +/- 125 KHz coupling the simple K pulse is very inefficient at creating -Iz.
Flow Imaging
Another simulation program
which benefits from graphical display
calculates the response of flowing
spins in an NMR imaging experiment.
Our interest in this area focuses on the
artefacts that arise when spins move
from a region with one gradient
strength to a region with anothe
during an imaging experiment. Ou
simulation incorporates the effects c
flow and of arbitrarily comple
imaging sequences on an array c
spins 1/2. When following an
particular spin through time we kee
track of its three-dimensional positio
and its magnetization vector. Thus w

Vol. 14, No. 1-4 195
have for each spin seven parameters
(3 space, 3 spin and time) which
describe its state.
In order to understand how
artefacts develop during the
experiment we display the system as
an array of vectors whose position
corresponds to spatial position and
whose orientation corresponds to
magnetization state. From the z-axis
we observe spatial flow of spins in the
spatial x-y plane and also the
magnitude and phase of transverse
magnetization in the superposed spin
x-y plane.
A
B
c
/*-
In Fig. 3) we see how the sliceselect
portion of an imaging
experiment can be perturbed by flow
in the direction of the slice-select
gradient. As the rate of flow
increases, the width of the slice
remains fairly constant but the phase
order of the spins in that slice
deteriorates. Such phase disorder will
lead to artefacts along the phaseencode
dimension at the edges of the
slice. As it now exists our simulation
will allow us to evaluate imaging
sequences on flow geometries
Flow / Imaging Gradient
Figure 3) Excitation profiles for the slice-select portion of an imaging experiment. In this
simulation the direction of flow is along the slice-select gradient causing spins to change
their frequency during the sine excitation pulse. The flow velocities are in arbitrary units
but three regimes are displayed: A) static spins; B) slow flow; C) moderate flow.

iii.:
196
of interest. A better understanding of
the formation of such errors will aid
us in the development of improved
imaging sequences.
Conclusion
We have extended the computer
techniques available to the NMR
spectroscopist by presenting
simulated data in multi-dimensional
animations. With these animations it
is much easier to see the development
of spin coherences which are the
result of experimental imperfection or
which are intended by design. Our
original goal with computer graphics
was to enhance our own
understanding of the experiments we
perform and to aid us in experimental
development In the process we have
found animated simulations to be a
generally useful pedagogical tool for
explaining all types of NMR
experiments. With computer designed
pulse sequences containing unusual
phases and non-analytical shapes
becoming more and more common we
hope to bring some of the intuition
back to experimental design.
References
[1] S. J. Glaser and G. P. Drobny,
.Adv. Mag. Res., 14, 35 (1990)
[2] H. Liu, S. J. Glaser, and G. P.
Drobny, J. Chem. Phys., 93(111),
7543 (1990)
[3] D. G. Cory, J. B. Miller, and A. N.
Garroway, /. Mag. Res.,9Q, 205
(1990)
Bulletin of Magnetic Resonance
[4] P. Caravatti, L. Braunschweiler,
and R. R. Ernst, Chem. Phys. Lett.,
100(4), 305 (1983)
[5] P. Mansfield and P. K. Grannell,
Phys. Rev. B., 12(9), 3618
(1975)
[6] D. G. Cory and W. S. Veeman, J.
Mag. Res., 84, 392 (1989)
[7] J. B. Miller, D. G. Cory, and A. N.
Garroway, Chem. Phys. Lett.,
164(1), 1 (1989)
[8] J. B. Miller, D. G. Cory, and A. N.
Garroway, Philos. Trans. R. Soc.
London, Ser. A., 333(1632), 413
(1990)
[9] J. Iwamiya, S. Sinton, J. Callahan,
and G. P. Drobny, in abstracts of
the 33 rd ENC, Asilomar, CA USA,
221 (1991)
[10] T. M. Alam and G. P. Drobny,
Chem. Rev., 91, 1545 (1991)
[11] A. J. Vega and Z. Luz, /. Chem.
Phys., $6, 1803 (1987)
[12] D. Mattiello, J. Callahan, T. M
Alam, and G. P. Drobny, in
Proceedings of the 13 th ISMAR
Meeting, Vancouver, B.C Canada
(1992)

Vol. 14, No. 1-4 197
NMR INVESTIGATION OF THE SIMULTANEOUS
FERMENTATION OF XYLOSE AND GLUCOSE BY A
SELECTED STRAIN OF KLEBSIELLA PLANTICOLA (Gil).
C.Rossi*, A.Lepri*, M.P.Picchi*, S.Bastianoni*, D.Medaglini 0 ,
M.Vanassina 0 and E.Cresta*.
•Department of Chemistry, University of Siena, Pian dei Mantellini
44, 53100 Siena, ITALY.
°Department of Molecular Biology, University of Siena, 53100
Siena, ITALY
1. INTRODUCTION
The hydrolysis of hemicellulose
yields a mixture of sugars of which
D-xylose and D-glucose are the
major constituents.O) This sugar
mixture is an excellent substrate for
growing microorganisms and yields
high energy products such as
ethanoH 2 ' 3 ). In developing this
project two main problems have to
be analyzed in detail: i) the isolation
and identification of microorganisms
whose metabolism can be sustained
by the hemicellulose-derived sugar
mixture, ii) the characterization of
the metabolism, and the selection of
specific metabolic pathways of
microorganisms growing on sugar
mixtures. It has already been shown
that the use of selectively carbon-13
enriched substrates enables the "in
vivo" metabolization process of
microorganisms and tissues^ 4 ' 5 ) to be
studied by NMR. This technique was
applied to the study of the
simultaneous fermentation of xylose
and glucose by a newly isolated
Klebsiella planticola (G 11) strain.
In the present investigation [2- 13 C]glucose
and [l- 13 C]-xylose were
used as carbon-13 enriched
substrates. Isotopic enrichment in
different positions of the sugar chain
enabled us to: i) separate the xylose
from the glucose signals in the
carbon spectrum and ii) calculate
the contribution of each sugar to
end-product yield.
2. EXPERIMENTAL
Klebsiella planticola Gil was
isolated from the soil of a corn field
and selected for its capacity of
growing on a mixed sugar
substrate^ 6 ). Identification of the
bacterium was performed on the
basis of taxonomic characters and
biochemical behaviour. The
microorganism was cultivated in a
nitrogen atmosphere at pH 7.5 and
35°C on a mineral medium
containing 0.2 g/1 of yeast extract.
The metabolism of the bacterium
was investigated by "in vivo" NMR
spectroscopy, using selectively

198
carbon-13 enriched sugar
substrates. The cell culture used for
microbatch NMR experiments had an
initial Optical Density (O D) of 0.5.
The microorganism was grown on a
NMR coaxial tube containing D2O as
lock signal on the external section
and positioned permanently in the
magnetic cavity.during fermentation
[l-^CJ-xylose and R-^CJ-glucose
obtained from Cambridge Isotope
Laboratories were used as enriched
substrates.
c)
Bulletin of Magnetic Resonance
The sugar metabolic process was
followed by recording carbon
spectra at 30 minute intervals until
the end fermentation.
3. RESULTS AND DISCUSSION
The NMR spectra of the enriched
sugar and the end-products of sugar
fermentation are shown in Figure 1.
The metabolization of xylose and
glucose followed two different and
iBU|iiii|Dii|iuuiiiijiiii|uiijuii|au|iiiiiniijiiiiiau|iiiiiiiii|iiiijuii|in inii|iui IIIII|INI IIIII|IIII jaiijiiimiii|iiiiiuii|iiii iu
180 160 140 120 100 80 60 40 20 PPM 180 160 140 120 100 80 60 40 20 PPM
Figure 1 - 1 3 C-NMR spectra obtained at the begining (left) and at the end
(right) of fermentation by K. planticola Gil. Total sugar lOg/1; pH
7.5; Temperature 35 °C. a) [2- 13 C]-glucose fermentation; b) [1- 13 C]xylose
fermentation and c) simultaneous [l-l 3 C]-xylose and [2-
* 3 C]-glucose fermentation. End-product signals detected by * 3 C-
NMR were:l) [2- 13 C]-lactic acid, 2) [2- 13 C]-ethanol, 3) formic acid,
4) [2- 13 C]-succinic acid 5) [l- 13 C]-lactic acid 6) [I- 13 C]-acetic acid
and 7) [l- 13 C]-ethanol.
The end-products of fermentation
were identified on the basis of the
NMR chemical shifts. Analysis of the
dependence of chemical shift on pH
enabled the preliminary
identification of carboxylic and non
carboxylic end-products.
independent pathways: "ethanol and
mixed acids" for xylose( ? ) and
"Emboden-Meyerhof" for glucose.
Diagrams of the sugar pathways are
shown in Figure 2. Glucose
metabolism was also analyzed using
[1- 13 C] enrichment. The same end-

Vol. 14, No. 1-4
products as
enrichment
detected.
a)
for [2- 13 C]-glucose
(Figure 1A) are
i
Xylulose
5, phosphate
Pentose phosphate pathway
biomass have been detected. The
increase in uptake rate with
increasing xylose concentration in
b)
Fructose
1,6 diphosphote
Dihydroxyaceton
phosphate
Figure 2 - Metabolic pathways of xylose (a) and glucose (b) fermentation
by Klebsiella planticola Gil.
The metabolization of xylose by
Klebsiella planticola Gil was
interesting from the view points of
the metabolization rate and the endproducts
of the process. Following
the xylose metabolization process by
NMR in a range of sugar
concentration from 5 to 100 g/1, the
uptake rate can be calculated. This
parameter it is very important
because is correlated with the
efficiency of the process of sugar
transport through the cell
membrane. Table 1 shows the xylose
uptake rate during the first three
hours of fermentation. Sugar
consuption rates for Klebsiella
planticola ATCC 33531, in the range
of 0.8-1.6 g.l^.h" 1 per gram of
AoeBc
Acid
our case is indicative of a "low
affinity" mechanism (not previously
detected in Klebsiella species)
similar to the "facilitated diffusion"
transport process described in
Candida sheabetaeS 9 ^
Figure 3 A shows the effect of
glucose on the metabolization of
xylose. The NMR spectra show that
glucose does not interfere with
xylose metabolism and its uptake
rate. Figure 3B reports the results of
a similar experiment, in which
glucose was added after two hours
of xylose fermentation. The addition
of glucose did not affect xylose
fermentation. It is very unusual for
two substrates to be metabolized at
199

200 Bulletin of Magnetic Resonance
rm 111T111111 nr [ 1111 n 11111 rn iTTT I I I I M I I I I I I I I I I I I I M I I
170 165 160 155 150PPM45 95 90 85 BO 75 PPH7C
Figure 3)- ^C NMR spectra recorded at different stages of sugar
fermentation, a) Simultaneous fermentation of xylose
and glucose b) The effect of addition of glucose after 2
hours of xylose metabolization. Spectra were recorded
every 30 minutes.
TABLE 1
Xylose uptake rate calculated from NMR spectra during the
first three hours of fermentation.
Xylose
concentration
g/1
5
10
20
40
50
80
100
a) per gram of dry weight biomass
Uptake rate a
g.l-i.h-l
1
2.1
3.5
4.6
5.1
6.2
7.0

Vol. 14, No. 1-4
the same time. The phenomenon has
been hypothesized before but has
only ever been demonstrated in
Candida sheabetae. The non diauxic
growth of Klebsiella planticola Gil is
very important because it could be
used to ferment sugar mixtures like
that obtained from the hydrolysis of
hemicellulose.
If the property of good xylose
uptake can be combined with good
end-product yields by selection or
genetic engineering^ 10 ' 11 ), it will
constitute a further advance in the
development of bioethanol
production.
4. REFERENCES
1) T.E.Timell; Adv. Carbohydr. Chem.
19, 247 (1964).
2) K.Skoog and B.Hahn-Hgerdal; Enz.
Microb. Technol., 10, 66 (1988).
3) H.Shneider; Critical Review in
Biotecnology, 9, 18 (1989).
4) K.Ugurbil, T.R.Brown, J.A.Den
Hollander, P.Glynn and R.Shulman;
Proc. Acad.
(1978).
Sci. USA, 75., 3742
5) J.A.Den Hollander, T.R.Brown,
K.Ugurbil and R.Shulman; Proc. Acad.
Sci. USA, 76, 6096 (1979).
6) E.Cresta, SJez, A.Lepri A.Pisani,
C.Rossi and G.Sabatini; "NMR
investigation of xylose bioconversion
by Klebsiella sp. strain isolated from
soil". In "Biomass for Energy and
Industry" G.Grassi, G.Gosse and G.dos
Santos eds., Elsevier Applied
Sciences, 2, 253 (1990).
7) C.S.Gong, L.F.Chen, M.C.Flikinger
and G.T.Tsao; Adv. Biochem. Eng.,
20,93 (1981).
8) J.Tolan and R.K. Finn; Appl.
Environ. Microbiol., 53_, 2039 (1987).
9) C.Lucas and N.van Uden; Appl.
Microbiol. Botechnol., 2J3_, 491
(1986).
10) F.Alterthun and L.O.Ingram;
Appl. Environ. Microbiol., 55_ 1943
(1989).
11) J.S.Tolan and R.K.Finn; Appl.
Environ. Microbiol., 51 2039 (1987).
201

202 Bulletin of Magnetic Resonance
Interleukin-1 Receptor Antagonist Protein:
Solution Secondary Structure from NOE's and
1H« and 13C« Chemical Shifts
Brian J. Stockman, Terrence A. Scahill, Annica Euvrard, Nancy A. Strakalaitis,
David P. Brunner, Anthony W. Yem, and Martin R. Deibel, Jr.
1 Introduction
The Upjohn Company, 301 Henrietta St., Kalamazoo, MI 49007
Interleukin-la and interleukin-ip are two
polypeptides which share a significant
number of inflammatory, immunological
and pathological properties (for a review
see [1]). Importantly, these dissimilar 17
kDa proteins bind to two classes of interleukin-1
receptors, resulting in the
mediation of several immune and inflammatory
responses and in the induction of a
variety of biological changes in neurologic,
metabolic, hematologic, and endocrinologic
systems [1]. In addition to IL-loc and IL-1(3,
an interleukin-1 receptor antagonist
protein (termed either IRAP or IL-lra) has
been isolated, characterized, cloned and
expressed in E. coli [2-4]. This newer
member of the IL-1 gene family is a
naturally occurring inhibitor of the
interleukin-1 receptor [2,4], and represents
the first described naturally occurring
cytokine that functions entirely as a
specific receptor antagonist.
Site-directed mutagenesis [5-7] and
protein modification [6] studies have
identified three regions of IL-1 that are
involved in either receptor binding or
transmission of the biological response
upon binding. For IRAP, it can be hypothesized
that the regions of structure important
for receptor binding are maintained, but
that the region or regions responsible for
eliciting the response are somehow
different. To this end, we have begun an
intensive program to determine the solution
structure of IRAP using NMR spectroscopy.
Since the solution [8-12] and crystalline
[13,14] structures of IL-lp have been
determined, direct comparisons can be made
between IRAP and IL-lp. This may lead to a
correlation between structural and biological
differences.
2 Methods
Expression of IRAP was carried out using E.
coli K-12 strain DU379. Fermentation media
were supplemented with (15NH4)2SO4, [ 13 Ci]and/or
[i5N]-L-methionine, 1 5 NH4C1, and
[ 13 C]-D-glucose (stable isotopes were
obtained from Cambridge Isotope Laboratories,
Isotec, and/or MSD Isotopes) as
required to produce either 15 N- or doubly
13 C/ 15 N-enriched IRAP. Analysis of
resolved *H resonances indicated that both
13 C and 15 N were incorporated at an
enrichment level greater than 95%.
Samples for NMR spectroscopy contained 2
mM IRAP, 50 mM 2H4-ethanolamine and 300
mM NaCl at pH 6.4. Trace amounts of PMSF
and NaN3 were added to prevent any
protease digestion or bacterial growth in
the sample.
All NMR spectra were recorded at
27 °C on a Bruker AMX-600 spectrometer
equipped with a triple-resonance probe and
a multi-channel interface. Threedimensional
1H-15N NOESY-HMQC and TOCSY-
HMQC experiments were recorded with
slight modification of the methods of
Zuiderweg and Fesik [15] and Marion et al.
[16]. Three-dimensional 1H-15N-13C HNCA
and HN(CO)CA triple resonance experiments
were recorded with constant-time 15 N
evolution as described by Grzesiek and Bax
[17]. Detailed acquisition parameters have
been described elsewhere [18]. Three-

Vol. 14, No. 1-4 203
6.0
Figure 1. Region of the slice corresponding the 1H-13H-13C-1H TOCSY spectrum of IRAP.
to 13 C frequencies of 18.7 and 51.8 ppm in Several assigned correlations are labeled.
dimensional 1H-13C-13C-1H COSY [19], m
13C-1H TOCSY [20], and 1H-13C NOESY-HMQC
[21] experiments were also recorded.
3 Results
Assignment of the majority of the backbone
!H, 13 c, and 15 N resonances of IRAP was
accomplished by analysis of four threedimensional
data sets. First, 1H-15N NOESY-
HMQC and TOCSY-HMQC experiments were
recorded on uniformly 15 N-enriched IRAP.
Then, two *H- 15 N- 1 3 C triple resonance
experiments were recorded, the so-called
HNCA and HN(CO)CA experiments [22,23].
Redundant sequential connectivities
obtained from the heteronuclear data sets
simplified and increased the reliability of
the assignments. During the assignment
process, NOE's indicative of secondary
structure were identified.
For many residues, magnetization
transfer in the 1H-15N TOCSY-HMQC spectrum
extended resonance assignments to at
least one X HP resonance and sometimes even
further down the side chain. In cases of
favorable resolution, such as for high-field
shifted resonances, two-dimensional DQF-
COSY and TOCSY spectra confirmed and/or
extended these side-chain assignments.
Extensive side-chain assignments, however,
will require ^C-directed strategies [19,20]
and are currently in progress. A represen-
tative slice from the 1H-13C-«C-»H TOCSY
spectrum of IRAP is shown in Figure 1.
Once the majority of correlations
were assigned in the !H- 15 N HSQC spectrum
recorded in J H2O, an identical spectrum was
recorded after exchanging the protein into
2 H2O. Only 50 1H- 15 N correlations remained
after six hours in 2 H2O solvent. As discussed
below, each of these residues was found to
participate in the p-sheet framework of
IRAP.
4 Discussion
During analysis of the 1H-15N NOESY-HMQC
data set, NOE's indicative of the solution
secondary structure [24] of IRAP were
identified. The majority of these- were
classified as cross-strand NOE's between
residues involved in ($ -sheet structure. They
are manifested in the NOESY-HMQC spectrum
as a third 1 H a NOE to an amide proton (the
others being the interresidue and intraresidue
1 H« *s), or as weak, non-sequential !H N -
!H N NOE's. Stretches of residues giving rise
to these types of NOE's also had other
characteristics associated with p-sheet
residues: low field 1 H a and !H N chemical
shifts, strong ^N- 1 !!" coupling (manifested
by intense DQF-COSY and TOCSY correlations),
and reduced } H N exchange rates. In
addition, 22 intense iHa-iH« NOE's, characteristic
of antiparallel p-sheet [24], were

204 Bulletin of Magnetic Resonance
-iY^k XII 146-152
119-122
X 120-125
III30_34
II23-27
VI.5-73
VH76.84
Figure 2. Schematic diagram of the topology strand NOE's. Dashed lines indicate interof
the p-sheet framework of IRAP. Double- strand hydrogen bonds inferred from
arrowhead lines identify assigned inter- analysis of *H N exchange rates.
11-19

Vol. 14, No. 1-4 205
1.20
1.00
0.80
0.60
0.40
A5 0.20
0.00
-0.20
-0.40
-0.60
-0.80 -L i i....?...J I !.........! n i [in! i.J
-0.50 --
-1.00 -•;•
-1.50 --
-2.00 --
10 20 30
Alpha proton
iv I | v | [ vi i | vii
40 50 60 70 80 90
Residue
Alpha carbon
: IXj
100 110 120 130 140 150
-2.50 --
rrni_n mm —yyj |_yj L^L
-3. 1
DQ DL J US — IEJ CO OD
10 20 30 40 50 60 70 80 90 140 150
Residue
Figure 3. Comparison of 'H" (top) and 13 C a shown at the bottom of each plot, p-strands
(bottom) A5 chemical shifts for IRAP and IL- are boxed, while helical regions are denoted
ip. Locations of NOE-defined secondary by lines,
structure elements in both proteins are

206 Bulletin of Magnetic Resonance
identified in the two-dimensional NOESY
spectrum recorded in 2 H2O and confirmed in
the 1H-13C NOESY-HMQC spectrum. Analysis
of this pattern of NOE's results in alignment
of the 12 p-sheet strands as shown in Figure
2. Arrows indicate observed cross-strand
NOE's, while dashed lines indicate hydrogen
bonds inferred from !H N exchange rates.
The p-sheet strands have been
presented in Figure 2 in a manner that
allows easy comparison to the P-sheet
framework elucidated for IL-lp in solution
by Driscoll et al. ([9], Figure 5). Comparison
of the two figures illustrates how the
overall topology of the two proteins is
identical, but in several regions is composed
of different stretches of the primary
sequence. Strands II and III, which are
adjacent strands connected by a fiveresidue
turn in IL-lp, are shifted by six
residues in the primary sequence and are
connected by a four-residue turn in IRAP.
Similarly, strands I and IV in IRAP are
shifted by six arid five residues, respectively.
The consequence of shifting the
residues that comprise these portions of the
P-sheet is that the N-terminal six residues of
IRAP have no structural counterpart in the
IL-ip structure. Structurally significant
shifts of one residue are seen for strands VI,
VII, and XII.
As expected for predominantly Psheet
proteins, large positive (downfield)
deviations from random coil chemical shifts
are observed for the l H a resonances [25],
and large negative (upfield) deviations
from random coil chemical shifts are
observed for the !3C a resonances [25,26].
Comparison of the secondary 'H a and 13 C a
chemical shifts of IRAP and IL-ip also
illustrates the differences and similarities
in location of secondary structure elements,
as shown in Figure 3. Note the excellent
agreement of the out-of-phase appearance
of the plots over the first 50 residues with
the five- or six-residue offset in location of
the P-strands. Also note how the in-phase
sections of the plots correspond to P-strands
at identical positions in both proteins.
While the solution secondary
structure of IRAP is dominated by antiparallel
p-sheet, short stretches of strong !H N -
!H N NOE's between adjacent residues,
indicative of a helical conformation, were
also observed. As shown in Figure 2, these
regions involve residues 40-45 and 86-89.
In addition, NOE's from 1HN of L89 to 1H« of
186 and from IH* of E44 to 1H« of V41, both
medium-range NOE's characteristic of a
helical conformation [24], were observed.
These were the only NOE's of this type
unambiguously assigned. In addition, the
i 3 C a chemical shifts in these two stretches
are shifted downfield slightly compared to
their random coil values, as would be
expected for a helical conformation [26].
These residues correspond to residues 35-40
and 87-90 in IL-ip, the former of which is a
3 io helix in solution [9]. Isolated strong !H N -
!H N NOE's between two or three sequential
residues were also observed, and locate turn
conformations at positions: 18-22, 27-30, 62-
65, and 116-118.
5 Acknowledgements
We thank Dr. Paul E. Fagerness, Dr. Eldon L.
Ulrich, Dr. Melinda Roy, Kathleen A. Farley,
Ron L. VanZanten, and Cindy A. Granatir for
their continuing contributions. We thank
Dr. Ad Bax (NIH) for advice and for supplying
preprints of the constant-time HNCA
and HN(CO)CA pulse sequences.
6 References
1. Dinarello, C. A. (1989) Adv. Immunology
44, 153-205.
2. Hannum, C. H., et al. (1990) Nature 343,
336-340.
3. Eisenberg, S. P., et al. (1990) Nature 343,
341-346.
4. Carter, D. B., et al. (1990) Nature 344, 633-
638.
5. MacDonald, H. R., et al. (1986) FEBS Lett.
209, 295-298.
6. Wingfield, P., et al. (1989) Eur. J. Biochem.
179, 565-571.
7. Gehrke, L., et al. (1990) J. Biol. Chem. 265,
5922-5925.
8. Driscoll, P.C., et al. (1990a) Biochemistry
29, 3542-3556.
9. Driscoll, P. C, et al. (1990b) Biochemistry
29, 4668-4682.
10. Clore, G. M., Wingfield, P. T., &
Gronenborn, A. M. (1991) Biochemistry 30,

Vol. 14, No. 1-4 207
2315-2323.
11 Clore, G. M, &. Gronenborn, A. M. (1991)
/. Mol. Biol. 221, 47-53.
12. Tate, S., et al. (1992) Biochemistry 31,
2435-2442.
13. Finzel, B. C, et al. (1989) J. Mol. Biol. 209,
779-791.
14. Priestle, J. P., Schar, H. P., & Griitter, M.
G. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 9667-
9671.
15. Zuiderweg, E. R. P., & Fesik, S. W. (1989)
Biochemistry 28, 2387-2391.
16. Marion, D., et al. (1989) Biochemistry 28,
6150-6156.
17. Grzesiek, S., & Bax, A. (1992)7. Magn.
Reson. 96, 432-440.
18. Stockman, B. J., et al. (1992)
Biochemistry 31, 5237-5245.
19. Ikura, M., Kay, L. E., & Bax, A. (1991) J.
Biomol. NMR 1, 299-304.
20. Clore, G. M, et al. (1990) Biochemistry
29, 8172-8184.
21. Ikura, M., et al. (1990) J. Magn. Reson.
86, 204-209.
22. Ikura, M., Kay, L. E., & Bax, A. (1990)
Biochemistry 29, 4659-4667.
23. Bax, A., & Ikura, M. (1991) J. Biomol.
NMR 1, 99-104.
24. Wiithrich, K. (1986) NMR of Proteins and
Nucleic Acids, Wiley, New York.
25. Wishart, D. S., Sykes, B. D., & Richards, F.
M. (1991) /. Mol. Biol. 222, 311-333.
26. Spera, S., & Bax, A. (1991) J. Am. Chem.
Soc. 113, 5490-5492.

208
1 Introduction
Green's Function Calculation of
Effective Nuclear Relaxation Times in
Metals and Disordered Metals
M. Martin-Landrovc and J A. Moreno
Departamento de Fisica and Centro de Resonancia Magnetica,
Facultad de Cicncias, Universidad Central de Venezuela,
Apartado 47586, Caracas 1041-A, Venezuela.
The theoretical derivation of expressions for die
magnetic nuclear relaxation times at very low
temperatures and the description of the behaviour
with temperature for such relaxation times,
has been of major interest among the researchers
in the field, specially because of the recent experimental
possibility to obtain measurements of
nuclear magnetic properties at such low temperatures.
There has been a considerable amount of
work in the area of nuclear magnetism [1], but a
comprehensive theoretical interpretation of
NMR relaxation times, at arbitrary temperatures,
is still lacking. Recently, Sbibata et al. has published
a series of papers [2], [3] concerning the
theoretical determination of the nuclear spin lattice
relaxation time for a system of nuclear spins
interacting with conduction electrons ina a metal.
Using a theory of nonlinear spin relaxation [4], [5]
they predicted a multicxponential spin-lattice relaxation
behaviour.
In the case of disordered metals and high temperatures,
where a Korringa law is aplicable, Warren
[6] predicted an enhancement of the relaxation
rate, which in some cases [7] could be as large
as 6,500. More recently [8], Gdtze and Ketterle
derived expressions for die Warren enhancement
factor by means of normalized Kubo response
functions [9].
ED the present work, we make use of the two-times
Green's function formalism in the regime of
the Linear Response Theory to derive the temperature
behaviour of nuclear relaxation times [10]
for nuclei in metals and disordered metals. The
Bulletin of Magnetic Resonance
results obtained are in complete agreement with
those derived by Sbibata in the assumption of an
effective unique relaxation time [2], [3] and with
experimental evidence [11], [12]. Also, there is
agreement between our results and those derived
by Gdtze and Ketterle [8] in the high temperature
regime, where Korringa law is valid but additionally
we obtained expressions for the enhanced
relaxation rate which are valid in the whole temperature
range. The organization of this paper is
as follows, in section 2, the general formalism is
derived, in section 3, we work out the Hamiltonian
of die system from which the equation for the
Green's function < < I°/I°> > to, which contains
all the information relevant to the spin-lattice
relaxation, is derived. This equation is then
solved including terms up to second order in the
electron nucleus interaction and the disorder parameters.
Finally in section 4 we discuss the relaxation
times formulas.
2 Green's Functions and the Relaxation
Rate
We will consider a system which can be modelled
by the total Hamiltonian:
H = Hs + HSL + HL (1)
where Hs represents the nuclear spin Hamiltonian,
HL is the Hamiltonian for the heat bath and
HSL represents the couplig between both
systems, usually under the condition
HL > Hs > HSL which occurs commonly in NMR
experiments. In order to consider the evolution of

Vol. 14, No. 1-4 209
the system toward thermal equilibrium, it is necessary
to prepare the system in an initial nonequilibrium
state. This can be achieved by the
adiabatic switching on of a pertrubation, which in
our case is a magnetic field along a particular
direction, and suddenly at time t—0, this perturbation
is switched off letting the system to evolve
freely according to die Hamiltonian H. The perturbation
can be written as:
where:
Hi' = - M .
H\* = Q(-t)e Et Hi
(2)
(3)
Within the linear response theory, the magnetization
for t>0 is given by the following expression
[13]:
H\ r dt
which is written in terms of the two times retarded
Green's function. By using the causality property
[13] of the retarded Green's function, it is
posible to write the equation for the displacement
of magnetization from the equilibrium situation
as:
1 [+OO > lT ) H\ _ lm ,
where < 8Af(/> = - o.Itcan
be shown that:
(4)
. H\ (6)
so that taking the Laplace transform of eq. (5), it
can be written as:
/(z) . (7)
(5)
where:
with GM ( >
- iz)
We are interested in the asymptotic behaviour
of < 8 M(f > since in that regime is that the relaxation
rates are experimentally measured. According
to Tauber's theorem [14], [15] an asymptotic
espanskm of f(t) can be written as:
*« 0
where Z\ represents the poles of the function
f(z), n* ( v) the order of the pole and Ck (y) the
coefficient in a Laurent expansion of f(z) around
the pole. In the particular case of f(z) having only
first order poles, eq. (9) can be written:
(8)
(10)
Let us suppose that the Green's function can be
written in the general form:
)
o> - acoo -
(11)
where Qxand Waare complexfunctions. According
to eq. (8) the poles of the function f(z) are
related to the poles of the Oieen's function so that
only those poles that are located at the upper
complex plane must be considered [16]. In general
it is necessary to make a complete analysis of
the pole structure of the Green's function at the
upper complex plane in order to describe the total
aymptotic behaviour of the function f(t). As a simplifying
assumption, we will consider the following
zeroth-order approximation to the pole
structure. We will first assume that the function
Ox(ci>) does not contribute with any pole. This
(V)

210 Bulletin of Magnetic Resonance
analyzed with some care. Secondly that the function
Wo. can be approximated by its value at die
frequency aoo. This second assumption ussually
is a very strong one since the whole pole structure
is sometimes collapsed to a single first order pole,
but nevertheless in many cases, this approximation
gives the right relaxation behaviour correlating
quite well with the experimental results. The
relaxation rates obtained under this assumptions
can be written in terms of the imaginary part of
the function Wa evaluated at the frequency aoao,
that is:
and:
T2
j - ImWo(O) (12)
too) - (- too)
(13)
3 Hamiltonian and Equation for the
Green's Function Go (coX
For tile system considered in this work, the Hamiltonian
can be written as in equation (1) where
the different terms are:
vv'tt
X Vz «*v*+ ~ «v'f-
with:
r
(14)
(15)
The lattice Hamiltonian can be written as the
sum of two terms [8]:
HL = 2- £v (ks) at k s av k s
\ks
where:
p + (q)V(q) (17)
X at- §.*«* + (18)
and represents the electron density fluctuations
for wavevector q. The disorder in the lattice is
represented by the Fourier transform U(q) of the
random potential. This coefficient satisfies the
symmetry property:
U*(q)= V(-q) (19)
The equation for the Green function Go(

Vol. 14, No. 1-4 211
and:
with:
W(to) = § **'
Att - A** A** -
At* -
Gf-+ (0 JU
Ajt* -
oi
(22)
(23)
X {»(* *-)(!- *(*+)) + «(lt+)(i- «(*:*-))}
At* -
2
- OJO)+
at-- at+ ,
i
(24)
o
(25)
(26)
(27)
(28)
21 1 + I
\k-g At •+,.*]
(29)
As it was discussed previously [17], the longitudinal
relaxation rate is proportional to the imaginary
part of the function W ( 0
ImAt*(w)= * q> 0
V(q)\ Z ,
&t',k-
(30)
(31)
(32)
Equation (31) represents the dynamical shift in
electronic energy due to the disorder and equation
(32) corresponds to the electronic relaxation
rate function, which in the limit of a>-»O,gives the
electronic relaxation rate 1/J> . The imaginary
part of W (to ) now become s:
1ml
x r
kk
XI-
(At*-ReAt*) 2 +(/mAt*) 2
(A**- RcAt*) 2 + (
I
(33)
In the case of a perfect metal, that is for U(q) -
0, we get from equation (33), the following expression
for the relaxation rate [17]:

212
k* '
[n(*+)(!-
+ n(*'-)(!-«(*+))]5( At*•) (34)
which exhibits a Korringa behaviour at high
temperatures and attains a maximum for the relaxation
time at very low temperatures, comparable
with nuclear spin energies [17]. in the case of
disorder we get for the relaxation rate:
njk - XI- n(k+
(At*- ReAt*) 2 + (ImA**) 2
I (A*i- ReAi"*) 2 -
in the limit goes to 0.
4 Conclusions
(35)
In die whole temperature range, equation (34)
can be calculated numerically, and the result for
the case of a perfect metal was obtained in reference
17, where it can be appreciated that T i
shows a maximum at a temperature that is approximately
half of the nuclear spin temperature,
which is consistent with the result obtained by
Shibata and co-workers in the supposition that
the relaxation process can be described by a unique
effective relaxation time T l as the observed
experimental behaviour is. Also there is a correspondence
between both results in the limit of low
temperature, leading us to think that our calculation,
even in its simpler approximation, is quantitatively
correct From the experimental point of
view there are not enough data to decide whether
a single or a multiexponential relaxation takes
place, but the general tendency is to believe that
even though the process seems actually to be multiexponential,
it could be described by an effective
relaxation time T 'i, which is die time that characterizes
the evolution of observables, in particu-
Bulletin of Magnetic Resonance
lar, the longitudinal magnetization [11], [12]. The
temperature dependence shown experimentally
by this time T 'i, agrees completely with the behaviour
calculated in reference 17. Also, equation
(35) takes into account the enhancement of the
relaxation rate for the whole temperature range,
showing at high temperatures a departure from
Korringa's law proportional to the electronic relaxation
time as Warren proposed. The result obtained
for the enhanced relaxation rate shows
that this enhancement will be present even at low
temperatures, as equation (35) is valid in the whole
temperature range. The approximation assumed
in this work, besides its simplicity, takes into
account the main features present in die temperature
behaviour of relaxation times, witiiin the
limits of Linear Response Theory, and it can be
extended to consider more realistic models or
systems.
Acknowledgments
This work was partially supported by CONICTT
and die Consejo de Desarrollo Cientffico y Humanistico
of die Universidad Central de Venezuela.
We also dutnk die collaboration of Fundacion
Polar for die necessary funding to present
diis work.
References
[1] Abragam A. and Goldman M., in Nuclear
Magnetism: Order and Disorder (Oxford, Clarendon,
1982).
[2] Shibata, F. and Hamano, Y., Solid St. Cotnmun.,
44,921 (1982).
[3] Shibata, F. and Hamano, Y., J. Phys. Soc.
Japan, 52,1410(1983).
[4] Shibata, F., J. Phys. Soc. Japan, 49,15 (1980).
[5] Shibata, F. and Asou, M., J. Phys. Soc. Japan,
49,1234(1980).
[6] Warren, W.W., Phys. Rev. B, 3,3708 (1971).
[7] Warren, W.W. and Brennerrt, G.F., Amorphous
and Liquid Semiconductors, (London, Taylor
and Francis, 1974), 2, p. 1074.
[8] Gdtze, W. and Ketterle, W., Z. Phys. B, 54,49
(1983).
[9] Kubo, R., J. Phys. Soc. Japan, 12,570 (1957).

Vol. 14, No. 1-4 213
[10] Mardn Landrove, M. and Moreno, J.A., Acta.
Oent Venez., 37,387 (1986).
[11] Bacon, F., Barclay, A., Brewer, W.D., Shirley,
D.A. and Templeton, JJE., Phys. Rev. B, 5,
2397 (1972).
[12] Brewer, WJD., Shirley, D.A. and Tcmplcton,
J.E., Phys. Lett. A, 27,81 (1968).
[13] Zubarev, D.N., Nonequilibrium Statistical
Thermodynamics (Consultants Bureau, New
York, 1974).
[14] Berg, h., Introduction to the Operational
Calculus (John WHey, New York, 1967).
[15] Krasnov, M.L., Kiselev, A.I. and Makarenko,
G.I., Functions of Complex Variable, Operational
Calculus and Stability Theory (Mir Publishers,
Moskow, 1983).
[16] Paley, R.E.A.C. and Wiener, W., Fourier
Transforms in the Complex Domain (A.M.S., New
York, 1934).
[17] Martin Landrove, M. and Moreno, J.A.,
Phil. Mag., 58,103 (1988).

214
1 Introduction
The extraction of molecular
potentials and rates by modeling the
dependence of spectra on thermodynamic
state is one of the major contributions of
magnetic resonance to molecular physics.
We argue here that this entire endeavor is
conceptually suspect due to the implicit
factorization of spin and spatial degrees of
freedom in calculating stochastic averages.
All such averages have heretofore neglected
spin—dependent energies in the potential for
nuclear motion and are therefore not results
of equilibrium statistical mechanics. We
develop an averaging procedure from
equilibrium statistical mechanics and find
that it predicts large, previously
unrecognized contributions to motionally
averaged spin Hamiltonians which are
directly proportional to spatial terms in the
energy. We discuss the present state of the
experimental evidence for the accepted
average, with particular emphasis on
indirect scalar couplings averaged by
conformer equilibria, and find that the
quality of presently available data cannot
resolve the issue conclusively. The weakness
of the theoretical basis of the accepted
theory is also outlined.
2 The Traditional Stochastic Average
Stochastic Averaging Revisited
D.H. Jones, N.D. Kurur, and D.P. Weitekamp
Arthur Amos Noyes Laboratory of Chemical Physics
California Institute of Technology
Pasadena, CA 91125, USA
For over forty years it has been
accepted 1 " 36 that spin states are transported
between spatial states with spinindependent
rates. This unexamined
assumption was clearly stated in the seminal
work of Bloembergen, Purcell and Pound: 1
"The atom or molecule is simply a vehicle
Contribution No. 8724
Bulletin of Magnetic Resonance
by which the nucleus is conveyed from point
to point. We thus neglect the reaction of
magnetic moments of the nuclei upon the
motion." This notion is the" basis for all
existing formalisms 1 " 36 for calculating the
magnetic resonance lineshapes of spin
systems undergoing spatial rate processes —
most importantly, chemical exchange. A
well—known consequence is that the average
value of a spin—Hamiltonian parameter
(chemical shift, scalar coupling, dipolar
coupling, etc.) in the fast—exchange limit is
given by
= SpnX:
n n, (1)
where Xn is the value of this parameter in
the spin Hamiltonian of the nth spatial
manifold and pn is viewed as the probability
of the system being in that manifold,
irrespective of spin state. The sum may be
over molecular eigenstates or over' large
groups of them, as when n indexes molecular
conformers. The traditional prescription,
which neglects spin energies, is to express
the molecular partition function q as a sum
of parts qn associated with each indexed
manifold. The probability for each manifold
is pn = qn/ and the ratio of two such
probabilities is (in the absence of work
terms)
Pn/pn' = exp[-AAnn' /RT] (2a)
= exp[(-AUnn'+TASnn')/RT], (2b)
where AAnn'= -RTln(qn/qn') is the
difference in the molar Helmholtz free
energies of the manifolds. In Eq. 2b, the free
energy differences have been divided into

Vol. 14, No. 1-4 215
differences in energy AUnn' and entropy
ASnn'- The connection to molecular
energies is AUnn' = N^(En-En'), where En
is the common spatial contribution to free
energy of the n th manifold of spin states and
N 4 is Avogadro's number.
3 An Alternative Stochastic Average
The actual molecular energies may be
written as E? = En+E (n), where 7 indexes
a spin eigenstate within the n th manifold.
Since spin Hamiltonians are constructed as
traceless, the spin—dependent contributions
E (n) sum to zero in each manifold.
Nevertheless, the absence of the spin energy
terms in the pn indicates unambiguously
that Eq. 1 is not derivable without
approximation from equilibrium statistical
mechanics.
We address the following questions.
What is the exact equilibrium expression for
the stochastically averaged parameters? For
which systems will it differ measurably from
Eq. 1? Do existing experiments decide the
issue conclusively?
To proceed, we specialize to the case
that the spin Hamiltonians in all
significantly occupied spatial manifolds are
mutually commuting. Then the spin
eigenbasis {I7)} is independent of spatial
state and is the basis needed to describe the
stochastically averaged spectrum. The
spatially-averaged energy of a particular
such spin eigenstate 17) is
E =SE2exp(-A2/RT)/Sexp(-A2/RT) (3a)
I n n
= SpM (3b)
n
Note that Eq. 3 uses the complete
manifold free energy, A 2 = NAEZ-TSn,
including spin terms, according to the
prescription of equilibrium statistical
mechanics. Each such summation is
restricted to a particular spin state and the
distribution among spatial states for each
spin state is assumed to be the equilibrium
distribution at the lattice temperature.
Thus p2 is the conditional probability of
being in the n th spatial manifold, given that
the spin state is [7). As in the traditional
formulation, no specification of the
distribution of population among spin states
is needed. Our hypothesis is that the
spectral line positions for sufficiently fast
exchange between spatial manifolds are the
Bohr frequencies, ^^(E^-E^/h
corresponding to differences between the
average energies of Eq. 3. The
corresponding motionally averaged
spin—Hamiltonian parameters are those
which generate this spectrum.
As a simple illustration, consider a
spin Hamiltonian with only two distinct
eigenvalues, ±Xn/2 (in Hz), for each n.
Then the proposed alternative to Eq. (1) is
(with 7 = ±)
(X) =
(
= S [(p£ -
n
(4a)
(4b)
(4c)
For the usual high—temperature case where
hXn

i "i
t! I
m
216
depends on spin, spatial energy differences
between manifolds contribute to the
statistically averaged frequencies for spin
transitions. The product of a small
spin—dependent probability difference
multiplied by a large spatial energy
difference gives a readily measured
contribution to the magnetic resonance line
position.
4 Test Systems
In both formulations it is possible
ideally to predict the fast exchange
observations without adjustable parameters.
This would constitute a purely experimental
test, the quality of which depends only on
the experimental uncertainties. In practice
there seems to be no case where NMR
spectra of individual molecular eigenstates
have been obtained separately and also as a
thermal average.
Thus, it seems necessary to look to
those situations where n indexes conformers.
The experimental concept is simple and
well-known. At low temperature, the spin
Hamiltonian for each conformer can be
determined, since, in the limit of negligible
chemical exchange between them, separate
spectra are seen for each. The relative areas
of these spectra at each slow—exchange
temperature provide the relative populations
and thus the free—energy difference between
conformers. A linear fit to the temperature
dependence of this free—energy difference
allows it to be decomposed into two terms,
which can be viewed as an energy difference
and an entropy difference if one additionally
assumes that the temperature dependence of
these is negligible over the experimental
range.
Since a conformer is a set of
molecular eigenstates, additional dependence
on thermodynamic state (eg. temperature
dependence) is possible due to averaging
within this set. Theoretically, one has
precisely the same problem in deciding how
to take this average as for the averaging over
conformers. However, if one can measure any
such temperature dependence within the
slow—exchange regime and extrapolate to
fast—exchange, then the stochastic theory
need not enter at this level.
Thus, we arrive at a set of criteria
which need to be met for a compelling, fully
experimental test of any theory relating
Bulletin of Magnetic Resonance
slow-exchange and fast—exchange spectra:
i) The system must have
state—dependent rates such that
measurements in both the slow— and
fast-exchange regimes are possible. For
fluids this typically requires barriers between
conformers on the order of 10 kcal/mole.
ii) The state dependence of the
conformer spin Hamiltonians and the
free-energy differences must be measured in
the slow—exchange region to allow
extrapolation through the fast—exchange
region. This is often the major source of
uncertainty because of the small
temperature range corresponding to
slow-exchange. The difference in
thermodynamic state between these regimes
ideally is small or even zero, so as to
minimize the propagation of errors due to
phenomenological extrapolation. Using
different NMR transitions or field conditions
to measure the same spin Hamiltonian
parameter can help in this regard; since the
criterion for motional collapse varies with
the transition observed, there is no minimum
8.
8 1 4.
o
•a o
GO
6. -
1 1
- 'V^
1 __l
—
0.
f1 1 1 1
0. 200. 400. 600. 800. 1000.
Temperature (K)
Figure 1. Traditional (dashed) and
alternative (solid) stochastic averages for a
simple two—conformer, two—spin system.
Simulations have the general features of
some substituted ethanes: a gauche coupling
of 2.0 Hz, a trans coupling of 20.0 Hz, the
trans conformer with a free energy 200
cal/mol greater than the doubly—degenerate
gauche conformer.

Vol. 14, No. 1-4 217
difference in thermodynamic state between
fast and slow exchange.
iii) Some or all of the fast-exchange
data should fall outside the error bars on the
predictions of one of the theories, thereby
disproving it. The accepted and alternative
theories embodied in Eqs. 1 and 3,
respectively, have identical predictions for
mutual exchange and whenever the occupied
conformers are degenerate in spatial energy
or in spin Hamiltonians. For two—site
problems, the theories will typically differ
measurably when the conformer free energies
differ by > 102 cal/mole. When this
difference exceeds 10 3 cal/mole, sensitivity
will usually preclude observing the
slow-exchange spectrum of the minor
conformer. Figure 1 is a numerical
comparison of the two theories for the simple
case of two conformers and a two—spin
system. It demonstrates that the predictions
of the theories are different by a magnitude
that should be measurable. There is also a
qualitative difference: the sign of the
temperature dependence of the
fast—exchange spin Hamiltonian parameter
is opposite to that of the accepted theory
over part of the temperature range. This
behavior is inconsistent with Eq. 1 regardless
of how the pn are calculated.
The above conditions are not
extremely restrictive; a substantial fraction
of the molecules whose conformer equilibria
have been studied by solution-state NMR
fall into this range of free—energy differences.
Since the accepted theory has been in
increasing use for four decades, it might be
expected that it would have substantial and
diverse experimental support. While it is
difficult to have confidence in the
completeness of a search through such a
large literature, we are as yet unaware of
any data set that meets the criteria above.
Thus, no theory has presently been
evaluated by this seemingly reasonable
standard.
The only theories ever considered
previously are of the form of Eq. 1.
Numerous authors have noted failures in its
application, 19 " 21 but these have usually been
plausibly attributed to inadequacies in the
data, most commonly uncertainties of
conformer assignment or unmeasured
temperature dependence of a conformer spin
Hamiltonian.
5 Substituted Ethanes
Substituted ethanes in solution 14 " 24
are the most studied systems and include
cases which nearly meet the criteria of the
previous section. Figure 2 illustrates, with
the example of 1-fluoro-l,1,2,2—
tetrachloroethane, the trans and gauche
rotational isomers which interconvert at
convenient rates. Since the two gauche
conformers are mirror—images, there are
only two magnetically distinct conformers,
one with the H and F atoms trans to one
another and the other a degenerate pair of
gauche rotamers at the other two staggered
positions of the dihedral angle of rotation
about the carbon—carbon single bond.
Increasing temperature carries the system
from the slow—exchange limit to the
fast—exchange limit without a change in
composition or phase. The fast—exchange,
three—bond vicinal coupling (Jrrp) is known
to be temperature dependent and this has
been attributed to the averaging, according
to Eqs. 1 and 2, between distinct values Jt
and Jg in the trans and gauche conformers,
respectively. 15 " 16 Unlike chemical shifts,
scalar couplings do not require a nominally
temperature—independent reference
resonance to compensate for the usual
uncontrolled shifts of internal field with
temperature. Also, scalar couplings are
generally believed to be less sensitive to
intermolecular interactions which could
provide a confounding mechanism of
temperature dependence.
For the particular case of
1—fluoro—1,1,2,2—tetrachloroethane, a test as
E(

218 Bulletin of Magnetic Resonance
described of the stochastic averaging theories
is prevented by the failure to resolve the
coupling Jg. 37 This quantity has an
experimental upper bound of 2 Hz. If it is
assumed that Jg is independent of
temperature, then the data are consistent
with the accepted theory. 16 However a
temperature dependence of 0.005-O.01 Hz/K
between 150 K, where slow-exchange
observations have been made on the 19 F
resonances, and 300 K, where the *H
spectrum is motionally averaged, would
allow our alternative stochastic average in
the form
(5)
to fit the data as well or better.
Reason to suspect such a temperature
dependence can be found from a close
reading of the literature on the related
system 1—fluoro—1,1,2,2—tetrabromo—
ethane. 22 A value of Jg = 2.4 ± 0.3 Hz in
dimethylether at 188 K can be measured
from published data, 39 but has been reported
as 1.7 Hz at 180 K in the same solvent. 23 In
CFCI3 this coupling has been tabulated as
1.15 Hz from 171 to 178 K, 22 but our recent
fits of the spectrum (at 171 K only) in
reference 23 indicate Jg = 1.5 ± 0.3 Hz.
Thus the reported absence of temperature
dependence to three significant figures is
dubious and further experimental work is
needed.
6 Discussion
The present conformer model is by no
means the most refined version of Eq. 3
possible, but has the advantage of employing
only quantities that are experimentally
measured on the same sample. Continuous
classical—mechanical forms of the present
theory are readily written down and the
differences from the accepted theory persist.
Alternatively, the quantum partition
functions could be modeled. Such
modifications would introduce unmeasured
parameters. Such extensions might be
warranted after improvements in the
experimental data base.
It is of interest to note that the same
theoretical prediction for (Jjrp) obtained by
use of Eqs. 3 and 5 also results if the same
conformer model is evaluated using Eq. 4
with Xn = Jn and + and — indicating,
respectively, triplet and singlet zero-field
eigenstates. Thus, no measurable field
dependence of (Jjjp) is predicted by the
alternative theory, which is not obvious
because of the entanglement of different spin
Hamiltonian parameters in the spin energies.
The theoretical justification for Eq. 1
and related propositions is also weaker than
has been appreciated. Such an average
follows from dynamic models 1 " 36 based on
the assumption that spin states in
superposition are transported between
different spatial states in perfect concert.
Any such model exists in a truncated
Liouville space that excludes superpositions
of states that differ in both their spin and
spatial factors. Whether such a truncated
space suffices to describe magnetic resonance
lineshapes is an open question. What is
clear is that such a space cannot describe the
approach to equilibrium of the total system,
since this requires spin—dependent rates
between spatial manifolds. Thus a full
dynamic solution is needed in this complete
Liouville space. One result of such a full
solution will be the equilibrium average spin
energies of Eq. 3. Less clear is under what
dynamic assumptions either these energies or
those that follow from Eq. 1 will describe the
fast—exchange spectrum. In any case the
problem is richer than has been appreciated.
The accepted idea of how to calculate
a stochastically averaged spectrum is
universal in the literature of magnetic
resonance, underlying the interpretation of
average chemical shifts, dipolar couplings,
tunnel splittings, quadrupole couplings, and
hyper-fine interactions. In most situations
the number of unknowns is such that it is
not possible to verify the form of the
stochastic average, but valuable information
could be obtained if the correct form were
known. If the traditional ideas are generally
incorrect, many thousands of experiments
need to be reinterpreted with forms of the
present theory to in fact obtain the
quantitative information on molecular
structure that they were designed to yield.
Ultimately, the choice of theory will be
decided by a preponderance of data. The
present work indicates clearly that the issue
must be reopened and is a first step in the
reexamination of the experimental basis of
Eq. 1. Accurate experimental measurements

Vol. 14, No. 1-4
of systems such as those discussed here will
be needed in order to discriminate between
the stochastic theories, as opposed to fitting
free parameters according to one or the other
theory. Precisely the same issue of how to
calculate a stochastic average also arises in
the (ab initio) theoretical calculation of a
measurable spin Hamiltonian from
expectation values of the underlying
molecular eigenstates, which are almost
never sufficiently long—lived to measure
individually by magnetic resonance.
Application of the correct statistical
prescription will often be needed to in fact
test experimentally whether the quantummechanical
part of the calculation is
adequate.
This work was supported by the
National Science Foundation
(CHE-9005964). DHJ holds a Department of
Education Graduate Fellowship. DPW is a
Camille and Henry Dreyfus
Teacher—Scholar.
*N. Bloembergen, E.M. Purcell, and R.V.
Pound, Phys. Rev. 73, 679 (1948).
2
E.R. Andrew and R. Bersohn, J. Chem.
Phys. 18, 159 (1950).
3
E.L. Hahn and D.E. Maxwell, Phys. Rev.
88, 1070 (1952).

220
1 Introduction
Magnetic Resonance of Trapped Ions
by Spin-Dependent
Cyclotron Acceleration
Pedro J. Pizarro and Daniel P. Weitekamp
Arthur Amos Noyes Laboratory of Chemical Physics
California Institute of Technology, 127-72
Pasadena, CA 91125 USA
The frequencies of motion of ions trapped by static
magnetic and electric fields may be detected through the
charge induced on the trap plates. In a homogeneous
magnetic field, these frequencies have long been used for
high resolution measurements of the charge-to-mass ratio.
The most important chemical application of this is mass
spectroscopy via Fourier transform ion cyclotron resonance
(ICR). 1 ' 2 ' 3 Single-ion sensitivity in the detection of the
axial trapping frequency has also been demonstrated
recently for masses of chemical interest by D. Pritchard. 4 - 5
We have recently proposed methods for transferring this
exquisite sensitivity to the detection of the internal
spectroscopy of ions, in particular the magnetic resonance
spectra. 6 ' 7 This presentation focuses on design issues for
the most promising such method, in which spin-dependent
cyclotron acceleration is imposed and the resulting change
in ion orbit is detected as a change in the axial trapping
frequency.
As in the electron g-factor measurement of H.
Dehmelt, 8 the shift in axial trapping frequency is
proportional to the strength of a static magnetic bottle field
gradient. However, that direct effect of a spin flip is
impractically small for nuclear spin flips of ions. In the
present case, the transverse magnetic moment is coupled to
a radiofrequency gradient to provide an accelerating force.
A precedent is M. Bloom's deflection of neutral molecular
beams by radiofrequency field gradients (the "transverse
Stern-Gerlach effect"). 9 ' 10
We have derived both semiclassically and quantummechanically
the conditions under which a magnetic field
gradient modulated at both the Larmor and cyclotron
Contribution No. 8714
Bulletin of Magnetic Resonance
frequencies will lead to cyclotron acceleration proportional
to the transverse magnetic moment of a coherent state of
the particle and radiation field. In the presence of a
magnetic bottle, the corresponding shift in the axial
trapping frequency due to this spin-dependent work can be
made much larger than the shift due directly to a spin flip.
This effect has been incorporated into a proposed
experimental procedure in which the spin-flip probability,
resulting from a period of high-resolution magnetic
resonance, controls the presence or absence of a net axial
frequency shift between two detection periods. A data
reduction algorithm based on the fast Fourier transform
allows rapid conversion of the "before" and "after" signals
from one or many trapped ions into a point of the magnetic
resonance spectrum or interferogram. Simulated signals,
including the anticipated noise from both the detection
circuit and intrinsic quantum fluctuations in the number of
spins flipped, indicate the method is practical.
2 Ion Confinement in a Penning Trap and a
Magnetic Bottle
An ion in a homogeneous static magnetic field Bo,
whose direction defines the z-axis, will circle in the
transverse (x-y) plane at frequency ©c=qB0/m. Axial
(z-axis) trapping can be obtained by adding a static electric
field E = (vo/d 2 )(xx/2+yy/2-zz). Here Vo is a DC
trapping potential and d is a characteristic linear
dimension dependent on the details of the electrode
geometry (e.g., hyperbolic, cubic, cylindrical). An ion so
trapped undergoes three different types of harmonic
translational motion: 11 axial oscillation at frequency

Vol. 14, No. 1-4 221
a, = ^qV0/md 2 , rapid cyclotron motion in the
transverse plane at frequency ©+, and slower magnetron
motion of frequency co_, with ©± = y{ ©c + J© 2 ~2© 2 I.
In ICR, one monitors the radiofrequency voltage at © +
induced on a capacitor by the cyclotron orbit of a group of
ions coherently excited by an oscillating electric field
resonant with the cyclotron motion. Single ion sensitivity
has been achieved for detection of both ©z 4 and, indirectly,
©+, 5 using axial detection. The charge-to-mass ratio is the
only structural quantity measured on trapped ions by ICR
to date.
In order to describe how other quantities, in particular
the NMR spectrum, may be encoded into trapped ion
signals, it is necessary to analyze the case of ion motion in
the presence of a magnetic field gradient. In particular we
consider the introduction of a magnetic bottle field of the
form (in cylindrical coordinates, f = zz+pp+)
AB = B2f(z 2 -p 2 /2)z-zpp]. n > 12 The presence of AB
couples the axial, cyclotron and magnetron motions to the
spin. A classical analysis of the ion motion in a known
spin state is adequate though the principal results can be
confirmed with the trapping frequencies from first-order
quantum perturbation theory. 11 To a very good
approximation the axial amplitude is a simple onedimensional
sinusoid even in the presence of the magnetic
bottle. The potential energy for the axial motion is
The spin magnetic moment operator has been replaced by
its two high-field eigenvalues (upper and lower signs)
assuming a single spin 1/2 nucleus with gyromagnetic ratio
y. The only coupling to the transverse motion is through
the mechanical magnetic moment \xm = (q/2)[x(dy/dt)y(dx/dt)].
This quantity is however a constant of the
motion, 13 so that the axial motion is separable with a
parametric dependence on the ion's transverse orbit. The
ellipsis denotes terms quadratic or higher in B2/Bo. We
have performed exact three-dimensional trajectory
calculations to confirm that Eq. 1 suffices for the times and
orbits of the numerical examples discussed later. The axial
frequency for each ion, including corrections to ©z due to
the magnetic bottle, can be written from Eq. 1 as
mco,
A key illustration of this coupling was the use of the spindependent
shift of the axial frequency to measure the g-
(i)
factor of the electron 8 ' 11 cooled at 4 K to the ground state
of its cyclotron motion. The straightforward generalization
of this to ions is not practical, since the shift is inversely
proportional to mass at fixed observation frequency (oz.
Extending this shift to a 100 amu ion with a proton
magnetic moment under conditions similar to those used in
the single electron experiments yields an unpractically
small 4 nHz shift. A more difficult problem is that the
three trapping frequencies and the Larmor frequency
become inhomogeneously broadened due to the wide range
of nm values present in a thermal ensemble. Minimizing
this by ion cooling methods would be time-consuming and
reducing this range to be less than the spin magnetic
moment requires lower temperatures as mass increases,
since this quantum decreases inversely with mass.
3 Spin-Dependent Cyclotron Acceleration:
IRICE
Rather than attempt to measure the small spindependent
term in the axial motion, we derive a form of
spin-dependent cyclotron acceleration analogous to
cyclotron excitation via ICR: this is mternally resonant /'on
cyclotron excitation (IRICE). The resonant electric field of
ICR is replaced by an oscillating magnetic gradient with
components at the cyclotron and Larmor (©0) frequencies.
This field is constructed by arranging two orthogonal
quadrupole coils: one is parallel to the x-axis with current
proportional to cos(©0t+n/2)cos(©+t), and the other is
directed along the y-axis with current proportional to
cos(©0t-Hi)cos(©+t-Hc/2). 14 With gradient field strength G
for each, the total magnetic field is
t = {B0 +Gzsin[(©0+© +
-Gcos(oooO sm ( to +
A quantum-mechanical description of a spin-1/2 magnetic
moment in this field shows that the eigenstates of spin lie
in the transverse plane, aligned such that the spin-
dependent force Fs=(fi»v)B resonant with the ion
cyclotron motion is
Like the force in ICR excitation, this is resonant with the
cyclotron motion, but with explicit spin dependence due to
a gradient dipole force. 9 ' 10 * 15
Neglecting the magnetron mode, the transverse ion
motion is described as a cyclotron oscillation of radius p+
and phase + (the sense of rotation used here is appropriate
for a positively charged ion):
(3)
(4)
(5)

222
—- = -ra+p+fsin((a+t+

pa*:'
Vol. 14, No. 1-4 223
' 4 R.M. Weisskoff, G.P. Lafyatis, K.R. Boyce, E.A. Cornell,
R.W. Flanagan, Jr., and D.E. Pritchard, J. Appl. Phys. 63,
4599 (1988).
5 E.A. Cornell, R.M. Weisskoff, K.R. Boyce, R.W.
Flanagan, Jr., G.P. Lafyatis, and D.E. Pritchard, Phys. Rev.
Lett. 63, 1674 (1989).
6 D.P. Weitekamp and P.J. Pizarro, U.S. Patent No.
4,982,088 (1991).
7 C.R. Bowers, S.K. Buratto, P.J. Carson, H.M. Cho, J.Y.
Hwang, L. J. Mueller, P.J. Pizarro, D.N. Shykind, and D.P.
Weitekamp, SPIE Proc. 1435,36 (1991).
8 R.S. Van Dyck, Jr., P.B. Schwinberg, and H.G. Dehmelt,
in New Frontiers in High Energy Physics, edited by B.
Kursunoglu, A. Perlmutter, and L. Scott (Plenum, New
York, 1978); in Atomic Physics 9, edited by R.S. Van
Dyck, Jr. and E.N. Fortson (World Scientific, Singapore,
1984).
9 M. Bloom and K. Erdman, Can. J. Phys. 40, 179 (1962).
10 M. Bloom, E. Enga, and H. Lew, Can. J. Phys. 45, 1481
(1967).
n L.S. Brown and G. Gabrielse, Rev. Mod. Phys. 58, 233
(1986).
12 R.S. Van Dyck, Jr., F.L. Moore, D.L. Farnham, and P.B.
Schwinberg, Rev. Sci. Instrum. 57, 593 (1986).
13 G. Schmidt, Physics of High Temperature Plasmas. 2nd.
ed. (Academic Press, New York, 1979).
14 Note that this field configuration has different symmetry
from that suggested in Ref. 9; the "Transverse Stern-
Gerlach" experiment on ions suggested there uses a field
whose symmetry does not in fact lead to a linear change in
the cyclotron radius over an extended period of time.
15 J.P. Gordon and A. Ashkin, Phys. Rev. A 21, 1606
(1980).

224
Bulletin of Magnetic Resonance
Coordination Modes of Histidine Moiety in Copper (II) Dipeptide
Complexes Detected by Multifrequency ESR
1 Introduction
R.Basosi, R.Pogni, G.Delia Lunga
Department of Chemistry - University of Siena
Pian dei Mantellini 44,53100 Siena, Italy
The copper complexes of histidinecontaining
peptides naturally occurring in
blood plasma [1,2] can be used as a model
for metal/protein interactions.In fact a low
molecular weight peptide containing
histidine or even histidine itself may
compete with ceruloplasmin for copper in
the blood, significantly altering free ligand
levels. For this reason, the coordination
behaviour of the histidine moiety in copper
(II) dipeptides has been the subject of
much investigation [3-7]. From this point
of view the study under physiological
conditions, namely room temperature and
neutral pH, is crucial for maintaining the
biological significance of the model.
Unfortunately a definitive speciation has so
far been hindered by the strong
dependence on the ligand-to-metal ion ratio
and the multiplicity of possible equilibria,
function of pH, in solution.
ESR may be used to study
paramagnetic metal complexes, and
multifrequency ESR in combination with
computer simulation is a decisive tool for
determining the chemical structure of
copper biosystems under physiological
conditions. For ESR spectra the
requirement of a good fit at different
frequencies put constraints on the precision
of magnetic parameters and allows
definitive assignments, despite the lack of
resolution typical of room temperature
spectra. In this paper the above method
was used in the study of complexes of
copper(II) and Glycylhistidine (GlyHis),
Histidylglycine (HisGly) and p-Alanyl-L-
Histidine or Carnosine (Cam) formed in
solution in a large range of pH,
concentration and metal/ligand ratios. The
aim was to show how a computer-aided
multifrequency ESR approach can provide
valuable information on the systems under
investigation, avoiding undesirable effects
due to changes in the physical state of
samples.
2 Material and methods
Glycyl-L-Histidine, L-Histidylglycine and
p-Alanyl-Histidine (Carnosine) from
Sigma Chemical Co. were used without
further purification.
Solutions were made with distilled
water and the uncorrected pH was adjusted
with HC1 or NaOH and determined with
an "LCD" Model pH meter. Isotopically
pure 63 CuO (from Oak Ridge National
Laboratory, Oak Ridge , TN) was used for
the EPR experiments. Stock solutions
with 1:2 molar ratio were prepared, [Cu]
= 10- 2 M, [peptide] = 2 xlO" 2 M.
X-band EPR spectra were obtained
with a Bruker 200D SRC X-band
spectrometer, and S-band spectra were
obtained with a microwave bridge by
Medical Advances Inc., Milwaukee, USA.
All the bridges were equipped with loopgap
resonators (Jagmar, Krakow, Poland)
operating at v = 9.5 GHz for the X-band
bridge and v = 4.04 GHz for the S-band
bridge. Microwave frequencies were
measured with an XL Microwave Model
3120 counter. The spectrometer was
interfaced with a Compaq Deskpro
486/50L computer with an 8-megabyte
memory and a 50 MHz clock. The data
were acquired using the EPR data system
CS-EPR produced by Stelar Inc., Mede,
Italy.
The CUKOS program for the
simulation of EPR spectra of fast tumbling
copper complexes in an isotropic
environment, written in QuickBasic, is
based on Kivelson's theory of linewidth
[8] and includes the second order shift
equation of Bruno et al. [9] and the further
assumption of Lorentzian lineshapes. A
Monte Carlo calculation method was added

Vol. 14, No. 1-4 225
to the program. The Monte Carlo method
randomly varied selected spectral
parameters within defined limits in order to
fit experimental data.
3 Results and Discussion
In recent years, data for copper(II)
complexes of histidine and histidine
containing peptides has been generated by
a variety of techniques such as infraredspectroscopy
[10], x-ray diffraction [11],
nuclear magnetic resonance [12], circular
dichroism [13,14], thermodynamic [3] and
potentiometric studies [15,16] and ESR at
liquid nitrogen temperature [17,18].
Table I
between two species. Aggregation can
occur [21] and often tedious empirical
approaches for the formulation of good
glasses must be employed. In water,
freezing can change the local pH [20,22],
and Wilson and Kivelson [8] found that
the isotropic g- and A-values of copper
acetylacetonate are actually temperature
dependent. All of these complications can
be avoided by the use of multifrequency
ESR of copper complexes above 0 °C.
ESR spectra obtained for Cu-Histidine
complexes with a 1:1,1:2 and 1:100 molar
ratios, in the pH range 5-9 are reported in
Refs. [17,18]. At pH= 7.4 and high molar
ratios the ESR pattern shows a
Parameters Used to Simulate ESR Spectra for Cupric Ion complex with Dipeptides a
system giso Ag AisO( 63 Cu) AA A N isO xc(ps)b Ref.
63 Cu-His
3N
4N
63 Cu-GlyHis
63 Cu-(GlyHis)2
63 Cu-(HisGly)2
63 Cu-B-Ala-His
2.098
2.104
2.1134
2.107
2.114
2.1468
0.162
0.159
0.171
0.166
0.235
0.221
61.50
61.60
69.32
80.36
54.4
62.91
137.4
143.0
174.95
195.87
193.79
190.00
a Values for the hyperfine and superhyperfine splittings arc given in G.
11.1
11.0
12.4
13.5
11.03
12.5
b Correlation times in ps (10~ 12 sec.). 63 Cu magnetogyric ratio 0.70904 s^
A conventional ESR spectroscopic
approach is to freeze the sample and extract
the magnetic parameters by analysis of the
frozen solution powder pattern. There are
many reasons why this is undesirable.
Vanng&rd [19] called attention to changes
in coordination. Falk et al. [20] observed
the temperature dependence of equilibrium
85.00
85.00
50.64
88.23
75.00
45.00
18
18
this work
this work
this work
this work
septet 1:3:6:7:6:3:1 consistent with mixed
glycine-like and histamine-like
coordination. At higher pH, this species is
in equilibrium with a complex in which
both the histidines are coordinated in the
histamine way (4N). The magnetic
parameters for the two species are
reported in Table 1.

226 Bulletin of Magnetic Resonance
Fig. 1 shows ESR spectra at X-band for
the complex Cu-GlyHis in a fast tumbling
regime. All the experimental spectra are
paired with the simulations for different
species.
100 G
Fig. 1. Exp. ( ) and simulated (- • - ) ESR
spectra for Cu-GlyHis at pH = 7.3. a) monomeric
species, b) 62 % monomeric species + 38 % biscomplex,
c) bis-complex.
Fig. la represents the monomeric species
obtained for a 1:1 molar ratio and pH=7.3.
This species is readly obtained for a molar
ratio of 1:2 at lower pH. Fig. lc shows the
ESR spectrum of the bis-complex. This
species is obtained when there is a large
excess of the ligand with respect to the
metal. Magnetic parameters for the above
two species are reported in Table 1. With a
molar ratio of 1:2 and physiological
conditions, the two species occur
simultaneously present in solution and the
overall ESR spectrum is reported in Fig.
lb paired with the simulated spectra. The
best fit in Fig. lb is obtained for a pattern
in which 62% of the monomeric species is
considered together with 38% of the biscomplex.
As the fit is reasonable, the
concomitant presence of the two species
can be supposed. A similar best fit
procedure was performed for S-band. Fig.
2 shows the expansion of the second
derivative of the mi = +3/2 component for
the monomeric species of Cu-GlyHis in
the X and S-bands.
A
Fig. 2. Expanded 2nd derivative experimental mj =
+3/2 component ( ~) paired with
simulation (-••-) for the monomeric species Cu-
GlyHis at: a) X-band, b) S-band.
The procedure proposed [23] is based on
the relative intensities of the patterns of the
three center lines in order to discriminate
between three and four nitrogen
coordination. The simulations are adjusted
until the best fit is obtained. In this case a

Vol. 14, No. 1-4 227
three nitrogens coordination (amino,
peptide and imidazole) from the ligand
molecule accounts for a very stable
monomeric species. The process of
simulation was initiated by reading the
approximate values of all parameters from
the spectra. The g-tensor values were
adjusted by fitting the spectra at X-band
with the highest sensitivity to these
changes. The simulations for liquid phase
EPR spectra started with a Monte Carlo
calculation method. This method randomly
varied selected spectral parameters within
defined limits in order to fit experimental
data. When a good fit was obtained at one
frequency the magnetic parameters were
tested at the other frequency and the
parameters were varied by an iterative
process until the best fit was obtained in
the X- and S-bands for the two physical
states of the samples. The use of second
derivative displays was a crucial step in
obtaining a good set of parameters. In
Fig. 2, second derivative, (or second
harmonic), display emphasizes sharp
features and discriminates against broad
features. This display is particularly useful
for analyzing superhyperfine patterns.
Shoulders in the spectrum become peaks
with precisely defined turning points that
are useful for accurate measurements of
coupling constants. The correlation times
(xc) reported in Table 1 are consistent with
the proposed speciation at pH=7.3.
Despite the fact that HisGly is chemically
similar to GlyHis, ESR spectra obtained
under similar conditions are strikingly
different. The relative parameters are
reported in Table 1. A big difference in
Aiso is evident: 54.4 G for Cu-(HisGly)2
and 80.4 G for Cu-(GlyHis)2. Data
obtained at low temperature confirmed this
assignment. In the case of the Cu-HisGly
complex, if excess HisGly is present, there
may be bis-complex formation in which
deprotonation of the peptide-NH linkage
is suppressed [24]. Voelter et al.
confirmed this hypothesis with 13 C NMR
experiments at pH 7, at which the glycine
moiety is not involved in copper ligation
[25]. The ligation is pure histamine-like
coordination with two imidazole nitrogens
and two amine nitrogens in the first
coordination sphere of copper. This
arrangement is in agreement with the EPR
results previously obtained for histidine
[17,18]. In fact in the case of Cu-HisGly,
the carboxyl group is blocked by peptide
bonding in favour of a total histamine-like
coordination whereas in the histidine
complex a mixture of histamine-like
glycine-like structures has been proposed
on the basis of experimental findings
[17,18,25].
Fig. 3 shows the ESR spectra obtained for
Cu-Carn with a 1:2 molar ratio at pH=5.6
paired with its simulation.
Fig. 3. Experimental ESR spectrum for Cu-Carn at pH
= 5.6 (——) paired with simulation (-•-).
Low pH was explored because at pH=7 a
very distorted ESR pattern, previously
attributed to a predominant dimeric
species, is obtained. If we increase the
ligand concentration up to 1:100 molar
ratio, a very different spectrum arises
which is shown in Fig. 4 with its
simulation.
Fig. 4. Experimental Cu-Carn ESR spectrum ( )
for 1:100 molar ratio paired with
simulation (-•-).
This spectrum is attributed to the Cu-
(Carn)4 complex and has very different

228 Bulletin of Magnetic Resonance
magnetic parameters. In this case, our data
confirms results reported in the literature
[26] for experiments at low temperature.
4 Conclusions
The ESR features of the above copperdipeptide
complexes show very different
coordination behaviour for homologue
Species under similar solution conditions.
At physiological pH, the predominant
histidine species present in solution are a
histamine-like and a mixed histamine-like
glycine-like complex. Under the same
conditions HisGly only shows the
histamine-like coordination whereas for
GtyHis, two species (monomeric and the
bis-complex) characterized by different
magnetic parameters occur simultaneously.
For Carnosine, a dimeric species is
dominant and is possibly in equilibrium
with a monomeric form.
A good characterization of the real species
present in solution can be achieved only by
exploiting the sensitivity of a computer
aided multifrequency ESR approach. A
crucial role is played by a procedure based
on the high selectivity of the second
derivative display.
5 References
1) CJ. Gubler, M.E. Lahey, G.E.
Cartwright, M.M. Wintrobe,
J.ClinJnvest., 32, 405, 1987
2) D.R. Williams, C. Furnival, P.M. May
in "Inflammatory Diseases and Copper",
J.RJ. Sorenson Ed., Humana Press,
Clifton, New Jersey, 45,1982
3) G. Brookes, L.D.Pettit, J.C.S.Dalton
1975,2112
4) R.P.Agarwal, D.D.Perrin, J.C.S.
Dalton 1975,268
5) RJ. Sundberg, R.B. Martin,
ChemRev. 1974, 74(4), 472
6) H. Sigel, "Metal ions in biological
systems", 1973, 63
7) D.B. McPhail, B.A. Goodman,
J.Chem.Soc.Faraday Trans. 1, 1987,
83(12), 3683
8) R. Wilson, D. Kivelson, /. Chem.
Phys., 1966, 44, 4445; 1966, 44, 154
9) G.V. Bruno, J.K. Harrington, M.P.
Eastman, J.Phys.Chem., 1977, 81, 1111
10) R.H.Carlson, T.L.Brown, Inorg.
Chem., 1966, 5, 268
11) N.Camerman, J.K.Fawcett, T.P.A.
Kruck, et al., J.A.C.S., 1978, 100:9,
2690
12) H.Sigel, B.McCormick, J.A.C.S.,
1971,93:8,2041
13) L.Casella, M.Gullotti, J.Inorg.
Biochem., 1983, 18,19
14) E.W.Wilson, M.H.Kasperian,
R.B.Martin, J.A.C.S., 1970, 92:18, 5365
15) D.D.Perrin, V.S.Sharma, J.Chem.
Soc. (A), 1967, 724
16) T.P.A. Kruck, B.Sarkar, Can.J.
Chem., 1973, 51, 3563
17) R.Basosi, G.Valensin, E.Gaggelli et
al., Inorg.Chem., 1986, 25, 3006
18) M.Pasenkiewicz-Gierula, R.Basosi et
al., Inorg. Chem., 1987, 26(6), 801
19) T.Vanngard, in: Biological
Applications of Electron Spin Resonance
(H.M. Swartz, J.R.Bolton and D.C.Borg,
eds) p. 411, Wiley-Interscience, New
York (1972
20) K.E.Falk, E.Ivanova, B.Roos,
T.Vanngard, Inorg.Chem., 1970, 9, 556
21) A.Saryan, K.Mailer, C.Krishnaruti, et
al., Biochem. Pharmacol., 1981, 30, 1595
22) Y.Orii, M.Morita, /. Biochem.
(Tokyo), 1977, 81, 163
23) J.S. Hyde, W.E. Antholine, W.
Froncisz, R. Basosi, In Advanced
Magnetic Resonance Techniques in
Systems of High Molecular Complexity;
N. Niccolai, G. Valensin, Eds.;
Birkhausen: Boston, 1986; p 363
24) I. Sdvagd, E. Farkas, A. Gergely
J.Chem.Soc.Dalton Trans., 1982, 2159
25) W. Voelter, G. Sokolowski, U.
Weber, U. Weser, Eur.J.Biochem.,
1975, 58, 159
26) C.E. Brown, W.E. Antholine, W.
Froncisz, J.C.S.Dalton, 1980, 590

Vol. 14, No. 1-4
An EPR and ab initio Study of a Phosphaalkene Radical Anion, and Comparison
with other Phosphorus-Containing Radical Ions
M. Geoffroy*, G. Terron, A. Jouaiti
Departement de Chimie Physique, Universite de Geneve, 30 Quai E. Ansermet, 1211, Geneve
(Switzerland)
P. Tordo
Universite de Provence, CNRS-URA 1412,13397 Marseille Cedex 13 (France)
and
Y. Ellinger
Ecole Normale Superieure et Observatoire de Paris, 24 rue Lhomond, 75005 (France).
1 Introduction
Electron Paramagnetic Resonance (EPR) is an
efficient method to obtain information about the
spin densities in organic radicals containing a
heteroatom and, when used in conjonction with
ab initio calculations, this spectroscopy can
yield a precise description of the structure of
these species. In the present study, our purpose
is to compare the spin delocalization on the fol-
lowing three paramagnetic moieties (-P= 13 C

230
a P=P bond and to record the EPR spectrum of
the corresponding frozen solution. Diphosphines
are well known stable molecules and we have
measured the 31 P anisotropic hyperfine
interaction after freezing a solution of
tetra,2,4,6,trimethylphenyldiphosphine [3] HI
previously oxidized in an electrolytic cell.
2 Experimental
The various compounds have been obtained by
adapting already published syntheses:
ArP=C(H)Ar' [1], ArP=PAr [2] and Ar"2P-PAr"2
[3] (where Ar: trit-t-butyl phenyl., An phenyl,
Ar": 2,4,6, trimethylphenyl ). The electrolyses
were performed in the EPR cavity of an X-band
Bruker spectrometer by using solutions of the
organophosphorus compound in presence of
tetra-n-butyl amonium hexafluor phosphate. The
31 P, 13 C and *H isotropic and anisotropic
coupling constants were obtained after
simulation of the experimental spectrum with a
program using second order perturbation.
The ab initio calculations were carried out on
a Silicon Graphics (Iris 4D) and a Vax 6700
Bulletin of Magnetic Resonance
computer by using G-82 and G-90 Gaussian
programs [4]. The 6-31G* basis set was
generally used and annihilation of the spin
contamination was performed for the UHF
calculations. The coupling constants were
obtained by calculating the expectation values
of the Fermi contact interaction and of the
hyperfine dipolar interaction. ROHF calculations
were also carried out, in particular when the
UHF method led to final values which
were not equal to 0.75.
3 Results and Discussion
EPR spectra..
(ArP=C(H)Ar')~. The liquid solution spectrum
is characterized by a large splitting of 152 MHz
and a rather broad linewidth exhibiting a poorly
resolved additional structure [5]. Full deuteration
of the C(H)C6H5 fragment only affects the
linewidth which is then equal to 8 MHz . The
spectrum obtained with ArP=C(H)C6D5 clearly
shows a coupling of 11 MHz with the ethylenic
proton while the spectrum obtained with
ArP= 13 C(H)C6H5 indicates a 13 C coupling equal
to 16 MHz. From these couplings and from the
observed linewidth one can conclude that an
appreciable spin delocalization occurs onto the
phenyl ring and the simulated spectra are
consistent with the following three additional
proton couplings: 11MHz, 7MHz and 7MHz.
The frozen solution spectrum obtained with
(ArP=C(H)Ar')" is characterized by an axial
hyperfine coupling with 31 P: T//=455MHz and
Tx =1 MHz; the full deuteration of the
fragment only decreases the

Vol. 14, No. 1-4 231
linewidth whereas the use of the 13 C(H)C6H5
fragment leads to an additional hyperfine
interaction: 13 C-T// =47 MHz, 13 C-T±=1 MHz.
(ArP=PAr)~ : As already mentioned [6,7,8], the
liquid solution spectrum exhibits a 31 P isotropic
coupling constant equal to 158 MHz. The
frozen solution spectrum has still not been
reported; it is characterized (Fig. 1) by an axial
coupling tensor : 31 P: T//=458 MHz and T± =10
MHz. No additional J H coupling is observed
with this compound.
100G
Fig 1. EPR spectrum obtained with a frozen solution of
(ArP=PAr)"
(Ar"2P-PAr"2) + .Only the liquid solution spectrum
has been previously reported [9], the
splittings observed with the frozen solution
spectrum (Fig. 2) show that the two phosphorus
nuclei are equivalent and that their hyperfine
tensors exhibit an axial symmetry: T/y= 796
MHz and T± =356 MHz.
100 G
Fig. 2. EPR spectrum obtained with a frozen solution of
(Ar"2P-PAr"2) + .
A'
Assuming a positive sign for all the hyperfine
eigenvalues leads to the isotropic and
anisotropic coupling constants shown in Table 1
together with the spin densities calculated by
using the atomic parameters obtained from [10].
Table 1. Experimental coupling constants (MHz) and
spin densities
radical
T
il 7± Ps V
ArP=CHR- 31p 152 303 -152 0.01 0.41
ArP=PAT
Ar'2PPAr' 2
13 C
31p
31p
31p
31p
16
158
158
502
502
ab initio calculations.
31
300
300
293
293
-15
-148
-148
-146
146
0.00
0.01
0.01
0.04
0.04
0.18
0.41
0.41
0.39
0.39
Phosphaalkene radical onion: In the Cs
symmetry, the UHF optimized structure for
(HP=CH2)- is characterized by HPC=97.5°, P-C
=1.791 A, P-H= 1.428 A; using the ROHF
method does not significantly affect these
parameters. The structure of (HP=C(H)Ar')-
was optimized (ROHF calculations) by fixing
the geometry of the phenyl group and by
assuming a planar structure: HPC=96.6°, H-P=
1.420 A and P-C =1.761 A.
Diphosphene onion and diphosphine cation.
The optimized structures of these two species
are known [11,12,13] and agree with our UHF
results : for (HP=PH)": HPH=95.8°, P-P=
2.133A (Cs symmetry) and for (H2P-PH2)+:
HPH= 102.2°, PPH=104.09° and P-P= 2.164 A

232
=0.750
symmetry). For these two ions
The various hyperfine coupling constants
resulting from UHF calculations are given in
Table 2 . For both the phosphaalkene and the
diphosphene anions, the "parallel" 31 P coupling
eigenvectors are oriented perpendicular to the
molecular plane (x structure), whereas for the
diphosphine cation these eigenvectors make an
angle of 45° with the P-P bond direction in
accordance with the orientation of the n" orbital.
Table 2. Calculated 31 P hyperfine coupling constants
(MHz).
radical
(HP=CH2)-
(HP=PH)-
(H2PPH2)+
^iso
25
70
445
T ll
130
279
328
T il.
-60
-136
-153
T ±2
-70
-142
-174
For the radical ions containing no phenyl ring,
the various spin densities calculated by using
the UHF method are very similar to those
obtained from ROHF calculations. These spin
densities are shown in Table 3.
Table 3. Calculated spin densities.
radical
(HP=CH2)" a
(HP=CHAiO" a
(HP=PH)' a
+ b
(H2P-PH2)
phosphorus
Ps Pp
0.00 0.22
0.00 0.44
0.00 0.49
0.06 0.42
adjacent carbon or
phosphorus
Ps Pp
0.00 0.77
0.00 0.33
0.00 0.49
0.06 0.42
a ) ROHF calculations, b ) UHF calculations.
phenyl
0.22
Bulletin of Magnetic Resonance
These data show that the substitution, in the
phosphaalkene anion, of an ethylenic hydrogen
atom by a phenyl ring drastically decreases the
spin density on the carbon atom and increases
the spin density on the phosphorus atom. The
spin delocalisation onto the phenyl ring is in
good accordance with the variation of the
linewidth observed after deuteration of the
benzene ring.
The calculated hyperfine tensors for (HP=PH)"
and (H2P-PH2) + reasonably agree with the
values measured for (ArP=PAr)~ and
(Ar"2PPAr"2) + . For (HP=CH2 )", the isotropic
and anisotropic coupling constants are quite dif-
ferent from those measured for (ArP=CF£Ar')-,
but, as indicated by the spin densities calculated
for (HP=CHAr')-> this difference is due to the
spin delocalisation induced by the phenyl ring.
0
0 / 0
In summary, the frozen solution EPR spectra
confirm the x* structure for phosphaalkene and
diphosphene radical anions ( negligible contri-
bution of the s orbitals of the phosphorus or
carbon atom to the SOMO) and the non-
bonding character of the SOMO for the
diphosphine cation.

Vol. 14, No. 1-4 233
References
* R. Appel, J. Menzel, F. Knoch and P.Volz, Z. anorg.
allg. chem. 534,100 (1986).
2 M. Yoshifuji, I. Shima, N. Inamoto, K.Hirotsu and T.
Higuchi, J. Am. Chem. Soc. 103, 4587 (1981).
3 B. I. Stepanov, E. N. Karpova and A. I. Bokanov, Zh.
Obshch. Khim., 39, 1544 (1969).
4 F.M.J. Frisch, M. Head-Gordon, G.W. Trucks, J.B.
Foresman, H.B. Schlegel, K. Raghavachari, M. Robb,
J.S. Binkley, C. Gonzalez, D.J. Defrees, D.J. Fox, R.A.
Whiteside, R. Seeger, C.F. Melius, J. Baker, R.L.Martin,
L.R. Kahn,J.J.P. Stewart, S. Topiol and J.A. Pople.
GAUSSIAN 90, Gaussian Inc., Pittsburg, PA (1990).
5 M. Geoffroy, A. Jouaiti, G. Terron, M. Cattani-
Lorente, and Y. Ellinger. J. Phys. Chem. (in press).
6 B. Cetinkaya,A. Hudson, M. F. Lappert and H.
Goldwhite, J. Chem. Soc. Chem. Commun. 609 (1982).
7 A.J. Bard, A.H Cowley, J.E. Kilduff, J. K. Leland,
N.C. Norman and M. Palkulski, J. Chem. Soc.., Dalton
Trans., 249 (1987).
8 M. Calcasi, G. Grouchi, J. Escudie, C. Couret, L. Pujol
and P. Tordo, J. Am. Chem. Soc., 108, 3130 (19886).
9 M. Culcasi, G. Grouchi,and P. Tordo, J. Am. Chem.
Soc.,107, 7191 (1985).
10 J.R. Morton and K.F. Preston. JMagn. Reson., 30,
577 (1978).
11 M.T. Nguyen, J. Chem. Phys., 91, 2679 (1987).
12 T. Clark, J.Am.Chem.Soc, 107,2598,(1985).
13 D. Feller,E.Davidson and W.T.Borden, J. Am.Chem.
Soc, 107, 2596 (1985)

234 Bulletin of Magnetic Resonance
Conformational Substate Distribution in
Myoglobin as studied by EPR Spectroscopy
Anna Rita Bizzarri 1 ^ and Salvatore Cannistraro 1 ' 2 '
' INFM-CNR, Dipartimento di Fisica dell'Universita', Perugia, Italy
2 ) Dipartimento di Scienze Ambientali, Sezione Chimica e Fisica, Universita'
della Tuscia, Viterbo, Italy
Introduction
It has recently been shown that EPR, in
connection with the aid of a computer
simulation approach, can be successfully
applied to investigate the structural
heterogeneity displayed by metallo-proteins [1-
6]. The g-strain effect characterizing the low
temperature EPR spectra of metallo-proteins
can be interpreted by taking into account for the
presence of an ensemble of molecules frozen in
many slightly different structures [6-8].
Different experimental and theoretical
approaches point out that a protein molecule
can assume a very large number of different
substates, called conformational substates (CS)
[9,10] whose sampling is important for the
biological functionality of the protein [11]. At
physiological temperature, proteins fluctuate
among CS; such a behaviour affecting the
kinetic response of the molecules. By
decreasing the temperature, the protein
solution undergoes a glass-like transition and
the fluctuations among CS are progressively
suppressed [12]. Below the glass-temperature,
T „, the molecules are frozen in many different
CS whose distribution may be modulated by
external agents such as pressure, pH, solvent
composition [13-15]. However, the role of the
solvent on the dynamical coupling between the
protein and the CS distribution is still open.
To get further information on this aspect, we
have analyzed the high and low spin ferric
myoglobin (Mb) samples in different
conditions. The EPR spectra of high spin ferric
Mb samples have been interpreted in terms of a
distribution of the crystal field parameters A ,,
A 2 connected with the energy differences of
the low-lying electronic states of the ferric ion;
whereas, the EPR spectra of the low spin Mb
samples have been analyzed in terms of a
distribution of the tetragonal and rhombic
splitting parameters, A and V. An accurate
computer simulation of the spectra has allowed
us to extract the parameters characterizing
these distributions which, in turn, have been put
into a relationship to the CS distribution. The
effects on these distributions as induced by
different solvent compositions and by different
cooling rates are analyzed.
Materials and experimental
methods
Mb EPR samples were prepared by dissolving
commercial (Sigma Chem. Co.) lyophilized
horse skeletal muscle Mb in 0.2 M phosphate
buffer. The highest concentration of Mb in the
solutions was about 5 mM. Final pH for Mb
solutions was about 6.8. Ferricyanide was used
to oxidize the heme iron to the ferric valence
state and solutions were dialysed several times
against buffers to remove the oxidant.
Approximately a twofold molar excess of
sodium azide was used to convert metMb to the
low spin form. Samples in mixed waterglycerol
solvent were prepared by adding
aliquots of glycerol to Mb solutio until the
required concentration was reached. A fast
cooling rate (Fast) has been obtained by
dipping the the samples into liquid nitrogen at
77K; while in the slow cooling rate ( Slow), the
system was cooled with a rate of 0.5 deg/min
from300Kto WOK
All the EPR spectra were recorded at 77 K by an
X-band Varian E109 spectrometer equipped with
a variable temperature control which was also
used to cool the samples in a controlled way. To
calculate the experimental g-values, a
magnetic field calibration was performed with

Vol. 14, No. 1-4
a Magnion Precision . NMR gaussmeter
Mod.G-542; the microwave frequency being
measured with a Marconi 2440 counter.
The acquisition of EPR data was carried out on
a HP 86A personal computer through a home
made interface connected to a IEEE 488 bus [16].
To run both simulations and bestfit programs,
the same microcomputer was switched to an
intelligent terminal of the main frame
computer (VAX 8350), through a serial interface
and an HP terminal emulator.
Analysis of the EPR spectra
It is well-known that the ferric ion, in Mb heme
complexes, is placed in a crystalline electric
field of cubic symmetry in which four ligands
are provided by the four nitrogen atoms of the
porphyrin ring, the fifth ligand is the nitrogen
of the proximal histidine, finally, in the sixth
coordination site different ligands can be
bound. In general, the presence of a weak
ligand causes the ferric ion to assume a high
spin state, S=5/2; while a strong ligand
determines a low spin state, S=l/2. In this paper
we consider metMb in which the weak ligand
HgO + is present, and azide Mb samples with the
strong sixth ligand N% [17].
The EPR spectrum, at 77 K, of metMb is
characterized by two resonances, one at g ~ 6
and a weaker one at g ~2 (spectrum not shown).
This system can be described by the spin
hamiltonian
Hs=ge(l H-S+D[SZ 2 -S (S + 1)/3]
where ge is the value for the free electron; D
and E are the tetragonal and the rhombic zerofield
splittings, respectively. For heme
proteins, the condition of large zero field
splitting is satisfied ( D -10 cm "*) [18] and
only transitions within the lowest Kramers
doublet occur; a fictitious spin S=l/2 can then be
used to fully represent the spin Hamiltonian of
the system, which for axial symmetry (g = g
= gx and gz = g | |) can be expressed by
H S = 9| | P H z SZ+QJLPI H X S x +H y s y) ( 2 )
where g| |~ 2 and g ~ 6 are the g-values
which are observed in the experimental
spectra. Splitting of the in-plane value into two
values, gx and gv, may result in a broadening
(as in our case) or even in a splitting of the g-
235
6 line [19,20]. High order corrections, arising
from spin-orbit mixing of the excited quartet
states into the lowest Kramers doublet lead,
under the assumption of a four-state model
[21,22] (Fig. 1) to the following expression for
gx and gy
gxy=3ge±24E/D-18.7 (E/D) 2 - 12 n 2 (3)
where the tetragonal zero-field splitting D is
given by
D = J L
and the rhombic zero-field splitting E
-2
(4)
2 (5)
the spin-orbit mixing of excited quartet states
into the lowest Kramers doublet is
Jl
~ 2 _ ^ / 1 . 1 ,1\
\ \
(6)
i is the effective spin-orbit coupling constant
(i = 300 cm ** ) which is reduced from the freeion
valued = 420 cm ~* ; A |, A 2 and y are the
energy differences between the low-lying
electronic states of high spin ferric heme (see
Fig. 1). •
V///////////////A
z
V///////777/7//I/
Z '/////// '///////A
777777777777/7////
Figure 1 Energy level diagram of the low-lying electronic
states of high-spin heme. It has been assumed A-p 2000 cm"
1 ,A2 ~ 6000 cm' 1 ' Y -60 cm" 1 . The shaded regions
indicate the variability of the energy levels (not in scale).
The low temperature (77 K) EPR spectra of
azide Mb samples are characterized by three
absorption lines to which three principal
different g- values (about gx = 2.8, gy = 2.2 and

236
gz = 1.7 ) correspond.
Within the Griffith's model [23], the ground
state electronic configuration is a ^T2 state
that can be described by one hole in the shell
made by the iron dxz, dyZ) dxy orbitals.
Owing to the presence of a rhombic distortion,
the orbitals are split into three Kramers
doublets with energies respectively of -V/2,
V/2 and A ( see Fig.2).
7/77/7/7/////////A
7/ 7777777777//////
7/7777/7/777777/
Figure 2 Energy hole levels of the low-lying d-orbitals for
the low spin ferric ion. The shaded regions indicate the
variability of the energy levels (not in scale).
Accordingly, the EPR spectra of azide Mb
samples can be described by the spin
Hamiltonian associated to S=l/2
H S=P ( 9x H x V9y Hy Sy+QzHzSz) ( 7 )
The three principal g-values are given by the
expressions
gx=2[2 AC - B 2 + k B(C- A) (2 1/2 )]
gy= 2[2 AC + B 2 + k B(C+ A) (2 1/2 )]
gz=2[A 2 -B 2 +C 2 +k ( A 2 - C 2 )] (8)
where k is the orbital reduction factor and A,
B and C are the coefficient characterizing the
Kramers doublet of the lowest energy
where 11* >, !, <
wavefunctions within t^ 2 |-1±> are the
T values
for V and A have been assigned, A, B and C
can be calculated by solving for ^/ + and \j/~ ,the
secular equations associated with the matrix
which takes into account for both the spin-orbit
Bulletin of Magnetic Resonance
coupling and the distortion field; then,
assigned a value for k, the g-values can be
determined from eq.(8).
In a general way, once the g values are
known, the EPR spectra can be generated by
computer simulation with the aid of a suitable
model. The derivative field-swept EPR
absorption spectrum, related to randomly
oriented paramagnetic centers with S=l/2, can
be reproduced by the expression
dS(vc,H) vc h p(e,
(10)
where the 1 / g(e.) is the Aasa-Vanngard [24]
correction, C is a constant that encompasses all
the instrumental parameters, P(8, ) is the
orientation dependent transition probability
which, for an S =1/2 system, can be exactly
expressed by [25]
cos 2 + g 4 y sin 2 8 cos 2 + g 4 z cos :
[g 4 x sin 2
(11)
finally f(H ) is the lineshape function
(residual linewidth [26] centered at the
resonance field HQ and with a linewidth
parameter a^ measured in magnetic field
units.
The integration over 8,4> in eq.(10) takes into
account for the random orientation of the
molecular axes with respect to the magnetic
field.
Use of eq.(10) is not, however, sufficient to
reliably reproduce the EPR spectra of metalloproteins.
It is known in fact that EPR spectra
of metallo-proteins are characterized by a
large inhomogeneous broadening resulting
into a spread of the g-tensor values (g-strain)
[1,4,6]. Such an effect can be interpreted by
taking into account the presence of the CS
distribution [8]; the heterogeneity
corresponding to the presence of a frozen
ensemble of molecules in different CS could
entail a spread of the low-lying electronic state
energies of the metal ion and, in turn, a
modulation of the g-values [7,22]. On such a
ground, and accordingly to previous works
[7,8], it has been assumed that the low-lying
electronic state energies of the ferric iron are
distributed.
In definitive, our spectra of metMb samples
have been simulated by introducing two
independent gaussian distributions for the

Vol. 14, No. 1-4
crystal field parameters A j, A2; on the other
hand, the azide Mb samples have been
simulated by considering two independent
gaussian distributions for the energy
differences A and V.
In both cases, the resulting simulated spectra
can be visualized as a superposition, weighed
in a proper way, of different spectra each one of
them corresponds to a different g-tensor: the
final expression of eq.(lO) convoluted with two
gaussians is
dS(vc,
dH r 2%ar dH
r 2- r ;
(12)
where T, and ?2 refer to the gaussian
distributions.
Computer-synthesized spectra have been used to
fit the experimental EPR spectra; a
minimization procedure of the x^-function,
based on a simulated annealing approach [27],
has been followed to extract the parameters A 0 ,,
and and
A2
characterizing the two gaussian distributions
for high and low spin Mb samples, respectively;
the chi-square function being
i = 1 a. (13)
where I^PCHj) is the derivative of the
experimental EPR absorption spectrum
sampled at 500 discrete points of the magnetic
field, I 8im (Hj,p) is the simulated spectrum,
finally cTj is the standard deviation calculated
for the i-th experimental point of the EPR
spectrum by repeated runs.
Results and Discussion
Fig.3 shows two examples of the experimental
and the corresponding simulated spectra for the
analyzed metMb and azide Mb samples. The
parameters A 0 ,, A°2,

238 Bulletin of Magnetic Resonance
TABLE 1 Central values and variances of the gaussian distributions of the crystal field parameters Ax A2
for high spin ferric Mb samples, and A V for low spin ferric Mb samples obtained by simulations (through
eq.(12) ) of the experimental 77 K EPR spectra of Mb frozen solutions. Gly means that the sample has
been prepared in 1:1 (by volume) water-glycerol mixture. Fast means that the sample has been submitted
to a fast cooling rate ( about 50 deg/min). Slow means that the sample has been submitted to a slow
cooling rate ( about 0.4 deg/min ).
SAMPLE
High spin Mb (Fast)
High spin Mb (Slow)
High spin Mb+Gly (Fast)
High spin Mb+Gly (Slow)
Low spin Mb (Fast)
Low spin Mb (Slow)
Low spin Mb+Gly (Fast)
Low spin Mb+Gly (Slow)
A? cm- 1
2266
2250
2194
2248
Ao
3.03
3.01
3.08
3.05
parameters distributions in both the high and
low spin case and in fast and in slow cooled
samples; such an effect can be interpreted in
terms of a decrease in the structural
heterogeneity of the protein as induced by
glycerol [7,8]. Different molecular
mechanisms could be invoked to interpret such
an effect; it is possible that addition of glycerol
could result into a viscosity-induced damping
of the protein motion; on the other hand,
changes in the dielectric properties of the
solvent could result into a different shielding of
the electrical charges of the amino acid
residues [30,31] and then into a modification of
the protein dynamics; moreover, glycerol, by
decreasing the ice-crystal dimensions, could
minimize the freezing strains [32].
In the high spin Mb samples, the slow cooling
rate induces, in presence and in absence of
glycerol, a narrowing of the crystal field
parameters distributions A , and A2- Such an
effect, which has been observed also in high
spin ferric Hb samples [7] can be interpreted in
different ways. First of all, a sort of
"condensation" could take place in the
molecules populating the frozen CS distribution
[7]; moreover, the cooling rate could induce
some modifications in the state of the hydration
water, as it has been observed by calorimetric
259
239
280
232
CTA
0.13
0.14
0.09
0.10
A£ cm- 1
5759
5750
5500
5417
Vo
2.00
2.01
2.03
2.02
aA, cm- 1
936
919
549
516
*v
0.06
0.08
0.05
0.07
measurements [33], and consequently affect the
CS distribution; finally, it cannot be ruled out
the possibility that cooling rate, acting on the
crystal growth, modifies the freezing straininduced
effects that might be present in low
temperature heme-proteins [34,35]. It is aspected
that all these mechanisms should be operative
also in low spin Mb samples, in which,
however, it has been observed that the slow
cooling rate induces an increase of the
variances a ^ and 0y.
The different behaviour registered in this case
requires a deeper investigation of the role
played by the strong sixth ligand in connection
with the freezing procedure.
In particular, we can speculate about the
possibility that the freezing procedure might
affect, in some way, the average position and
also the spread of the N'g ligand. In this
context, it should be noted that the different
number of ligand orientation, as induced by
different cooling rates, have been observed in
oxycobalt Mb [34].Therefore, different cooling
rates could result into different arrangements
of the ligand with respect to the metal ion.

Vol. 14, No. 1-4
Conclusions
EPR results to be a suitable spectroscopy to
study the structural heterogeneity displayed
by the ferric Mb samples in both the high
and the low spin configuration as derived
from an ensemble of different structures. In
particular, it can fruitfully be applied to
investigate the dynamical coupling between
the solvent and the protein CS distribution.
References
[I] D. O. Hearshen, W. R. Hagen, R. H.
Sands, H. J. Grande, H. L. Crespi, I. C.
Gunsalus, W. R. Dunham, J. Magn.
Resonance 69 (1986)440.
[2] C. More, P. Bertrand and J.P. Gayda, J.
Magn. Resonance 73 (1987) 13.
[3] A. S. Yang and B. J. Gaffney, Biophys. J. 51
(1987)55.
[4] J.C. Salerno: Biochem. Soc. Trans. 13
(1985) 611.
[5] A. S. Brill, F. G. Fiamingo, D. A. Hampton,
J. Inorg. Biochem. 28 (1986) 137.
[6] S. Cannistraro, J. Phys. France 51 (1990)
131.
[7] A.R. Bizzarri and S. Cannistraro, Appl.
Magn. Res. 2(1991)627.
[8] A. R. Bizzarri and S. Cannistraro, Biophys.
Chem. 42(1991)79.
[9] H. Frauenfelder, F. Parak, R. D. Young
Ann. Rev. Biophys. Biophys. Chem. 17 (1988)
451.
[10] V.I.Goldanskii and Y. F. Krupyanskii,
Quart. Rev. Biophys. 22 (1989) 39.
[II] A. Ansari, J. Berendzen, S. F. Bowne, H.
Frauenfelder, I. E. T.Iben, T. B. Sauke, E.
Shyamsunder, R. D. Young, PNAS USA 82
(1985) 5000.
[12] H. Frauenfelder, in Proteins and glasses
(E. Luscher, G. Fritsch and G. Jacucci eds.)
Amorphous and Liquid Materials, Nato series,
Martinus Nijhoff Publishers, Dordrecht (1987).
[13] M. K. Hong, D. Braunstein, B. R. Cowen,
H . Frauenfelder, I. E. T. Iben, J. R. Mourant,
P. Ormos, R. Scholl, A. Schulte, P. J. Steinbach,
A. H. Xie, R. D. Young, Biophys. J. 58 (1990)
429.
[14] H. Frauenfelder, N. A. Alberding, A.
Ansari, D. Braunstein, B. R. Cowen, M. K.
Hong, I. E. T. Iben, J. B. Johnson, S. Luck, M.
C. Marden, J. R. Mourant, P. Ormos, L.
Reinisch, R. Scholl, A. Schulte, E.
Shyamsunder, L. B. Sorensen, P. J. Steinbach,
A. H. Xie, R. D. Young, K. T. Yue, J. Phys.
Chem. 94(1990) 1024.
[15]E.E. Di Iorio, U.R. Hiltpold, D. Filipovic,
K.H. Winterhalter, E. Gratton, E. Vitrano, A.
Cupane, M. Leone and L. Cordone, Biophys. J.
59(1991)742.
[16] G. Giugliarelli, P. Tancini and S.
Cannistraro, J. Phys. E (Sci. Instrum) 22
(1989)702.
[17] M. Kotani Adv. Quantum Chem. 4 (1968)
227.
[18] J.F. Gibson, in ESR and NMR of
Paramagnetic Species in Biological and
related Systems, , eds. I. Bertini and R.S.
Drago, Reider Publishing Company (1979).
[19] J. Peisach, W.E. Blumberg, S. Ogawa,
E.A. Rachmilewitz and R. Oltzik, J. Biol.
Chem. 25(1971)3342.
[20] S. Cannistraro, Chem. Phys. Lett. 122
(1985)165.
[21]C.P. Scholes, J. Chem.Phys.52(1970)4890.
[22] F.G. Fiamingo, A. S. Brill, D. A.
Hampton, R. Thorkildsen, Biophys. J. 55
(1989)67.
123] J.H.E. Griffith, Theory of Transition of
Metal Ions, Cambridge Univ. Press, London
and New York (1961).
[24] R. Aasa and T. Vanngard, J. Magn.
Resonance 19(1975) 308.
[25] A. Isomoto, H. Watari, M. Kotani, J.
Phys. Soc. Japan 29 (1970) 1571.
[26] S. Cannistraro and G. Giugliarelli, Mol.
Phys. 58(1986) 173.
[27] S. Kirkpatrick, C. D. Gelatt, M. P. Vecchi,
Science 220(1983)671. .
[28] C. T. Migita, K. Migita and M. Iwaizumi,
Biochim. Biophys. Acta 743(1983)290.
[29] C. More, J.P. Gayda and P. Bertrand J.
Magn. Resonance 90(1990)486.
[30] G.P. Singh, H. J. Schink, H. V.
Lohneysen, F. Parak, S. Hunklinger, Z. Phys.
B55(1984)23.
[31] G. P. Singh, F. Parak, S. Hunklinger, K.
Dransfert, Phys. Rev. Lett. 47(1981) 685.
[32]W.R. Hagen, J. Magn. Resonance44(1981)
447.
[33] W. Doster, A. Bachleitner, R. Dunau, M.
Hiebl, E. Luscher, Biophys. J. 50 (1986) 213.
[34] H. Hori, M. Iketa-DSaito and T.
Yoketani, Nature 288 (1980) 501.
[35]M.R. Ondrias and D.L. Rousseau, Science
213(1981)2066.
239

240 Bulletin of Magnetic Resonance
1 Introduction
Effect of Paramagnetic Ions in
Aqueous Solution
for Precision Measurement
of Proton Gyromagnetic Ratio
Ae Ran Lim, Chang Suk Kim
Korea Research Institute of Standards and Science,
Taejon 305-606, Korea
and
Sung Ho Choh
The measurement of proton gyromagnetic ratio }'p'
has been the object of an intensive experimental
program for several decades [l]-[3]. The
gyromagnetic ratio of the proton is defined as a
resonance frequency «p divided by a magnetic field
Bo [4], when a spherical water sample at 25°C is
applied by a magnetic field. The ?P' for a defined
pure water sample is somewhat difficult to measure
because of the weak absorption due to the long
relaxation time of the proton [5]. In order to
reduce the relaxation time, paramagnetic ions
are added to the water sample [6]. However,
as the concentration of paramagnetic ions
increases, the resonance point of *H shifts and
differs according to the sample shapes.
The purpose of present work is to investigate the
effect of paramagnetic ions(Fe 3 *, Mn 2 *, Co 2 *, and
Cu 2 *) on the l H NMR in paramagnetic aqueous
solutions [Fe(NO3)3- 9H2O, FeCl3, MnCl2-4H2O,
CoCh-2H2O, and CuCl2-2H2O]. !H NMR in
aqueous solution containing paramagnetic ions was
measured as a function of concentration for the
fixed spherical and cylindrical sample shapes, and
the spinning cylindrical sample shape. The
magnetic susceptibility per unit volume of the
paramagnetic solution has also been measured as a
function of concentration at room temperature.
The interaction between the ' H nucleus and
Department of Physics, Korea University,
Seoul 136-701, Korea
paramagnetic ion is discussed in terms of the
shift of >H resonance point measured with two
sample shapes and the magnetic susceptibility of
the solution. From these experimental results,
we discuss the paramagnetic solution having a
short relaxation time and nearly zero shift of
resonance point to implement the precision
measurement of proton gyromagnetic ratio in a
low magnetic field.
2 Experimental Method
*H NMR experiment was performed by a
Brucker model MSL 200 FT pulse spectrometer.
l H NMR of paramagnetic aqueous solutions in
high magnetic field of 4.7 and 11.74 T, and
frequency of 200.13 and 500 MHz was
measured with two sample shapes at room
temperature. The magnetic susceptibility has
been measured using the Gouy magnetic balance.
2.1 *H NMR and Relaxation Time in
Paramagnetic Aqueous Solutions
The paramagnetic aqueous solutions [Fe(NO3)3-
9H2O, FeCb, MnCl2-4H2O, CoCl2-2H2O,
CuCl2-2H2O] were prepared by dissolving
paramagnetic ions of various concentration in

Vol. 14, No. 1-4 241
distilled water. The linewidth and the shift of 'H
resonance point were measured at room
temperature according to the shape of the
sample tube and the paramagnetic ion
concentration of aqueous solution. The
linewidth and the resonance point of *H NMR
in pure water were also measured.
The spin-lattice relaxation time (7"i) and
spin-spin relaxation time (T2*) were determined
from the signal of l H NMR at room
temperature by the inversion recovery method
and the inverse of linewidth, respectively.
2.2 Magnetic Susceptibility of Paramagnetic
Aqueous Solution
The density of paramagnetic aqueous solution
was measured with the mass and volume, and
the magnetic susceptibility per unit volume was
obtained with the susceptibility per unit mass.
3 Experimental Results
3.1 Shift of !H Resonance Point and Linewidth
In case of cylindrical sample tube, the
resonance point of r H NMR was shifted to the
negative direction with respect to that of pure
water according to the concentration of
paramagnetic ion as shown in Figure 1. It was
nearly unchanged with the variation of
concentration of paramagnetic ion in
Fe(NO3>3-9H2O solution. However, the identical
Fe 3+ ion in FeCb shows the largest shift of the
resonance point. Since the distribution of
valence electrons is influenced by the chemical
bqnding of an atom, it could be explained that
the displacement of nuclear magnetic resonance
frequency depends upon the chemical
environment [7].
Figure 2 shows the linewidth of } H NMR
according to the concentration of paramagnetic
ions. Here we have used the Lorentzian
absorption lineshape, and the linewidth
corresponds to the full width at the half
maximum. The Co 2+ and Cu 2+ ions are
almost ineffective to the linewidth. Whereas the
paramagnetic aqueous solutions containing Fe 3+
ion influence the linewidth as a function of
concentration of paramagnetic ions. The trend
in aqueous solution containing the Mn 2+ ion
differs from those in other paramagnetic ions.
For the case of spherical sample shape, the
shift of resonance point as a function of
concentration of paramagnetic ions is shown in
OO
o
5
-10-
> -15-1
B = 4.7 T
gy-;wHyt-w3
Fe(NO3)3-9H2O
FeCb
-20-
0 To" 20 30
Ion Concen.(10 20 ions/cc)
Figure 1. The frequency shift of 'H NMR signal as
a function of concentration of paramagnetic ions in
aqueous solution contained in a fixed cylinder (* is
the resonance point of 'H in pure H2O).
e(NO3) 9H2O
0
Ion Concen.(10 20 ions/cc)
Figure 2. The linewidth of 'H NMR signal as a
function of concentration of paramagnetic ions in
aqueous solution contained in a fixed cylinder (* is
the linewidth of 'H in pure H2O, 0.15 kHz).
Figure 3. The resonance point of 'H NMR was
shifted to the positive direction compared with
that of pure water. The frequency shift was
nearly unchanged with the variation of the
concentration of paramagnetic Cu 2+ ion. The
linewidth of *H NMR in the spherical shape is
the same as that in the cylindrical one as shown
in Figure 2, i.e. the linewidth has no difference
between the cylindrical and spherical samples
within the experimental error.

242 Bulletin of Magnetic Resonance
00
O
o
20-
15-
B = 4.7 T
B"
MnCl2-4H2O
•CuCU-2H;O
0 10 20 30
Ion Concen.(10 20 ions/cc)
Figure 3. The frequency shift of 'H NMR signal as
a function of concentration of paramagnetic ions in
aqueous solution contained in a sphere (* is the
resonance point of 'H in pure H2O).
However, the shift of the resonance point of
l H in case of spinning cylindrical sample
shows the similar trend as that of the fixed
spherical sample, but with the larger shift than
the fixed sphere as shown in Figure 4. Each
100-
B = 11.74 T /A^MnClr4H,O
FeClj
/
Fe(NOj)j-9HzO
CoCl;-2H2O
CuCl2-2H2O
0 10 20 30
Ion Conceti.(10 20 ions/cc)
Figure 4. The frequency shift of *H NMR signal as
a function of concentration of paramagnetic ions in
aqueous solution contained in a spinning cylinder (*
is the resonance point of 'H in pure H2O).
line in these figures was determined by the least
square fit with the experimental data.
3.2 l H Relaxation Time
Figure 5 shows the *H spin-lattice relaxation
time (Ti) for the cylindrical and spherical
samples measured by the inversion recovery
method. As the concentration of paramagnetic
ions increased, the relaxation time was
shortened. The 'H relaxation time of
paramagnetic ions containing Co 2 * or Cu 2+ was
longer than that of Mn 2 + or Fe 3 + . The 'H
spin-lattice relaxation time of 2.51 s measured in
pure water was consistent with the previously
reported value of 2.3 s at 20°C and 29 MHz
[8]. Figure 6 shows the spin-spin relaxation
time (7*2 *) for the cylindrical and spherical
samples obtained with the inverse linewidth of
the resonance line. This result shows the
similar trend as that of Ti. However, the J H
relaxation time 7z* is shorter than Ti in
aqueous solutions.
10-5. 10 20
10 21
! T c(N03)-9HzO
1022
Ion Concentration(ions/cc)
Figure 5. Spin-lattice relaxation time Ti of 'H due
to the paramagnetic ions in aqueous solution
contained in the fixed cylindrical and spherical
shapes. Both shapes have the same values within
experimental error.
3.3 Magnetic Susceptibility
The magnetic susceptibility per unit volume
of the paramagnetic aqueous solution was
obtained by the Gouy magnetic balance as a
function of concentration of paramagnetic ions

Vol. 14, No. 1-4 243
o
GO
"B
H
o
•X
i
'E,
10" 4 -
10 -5-
B = 4.7 T
ESS 3SS1
10 20 10 21
Ion Concentration(ions/cc)
Figure 6. Spin-spin relaxation time r2* of >H due
to the paramagnetic ions in aqueous solution
contained in the cylindrical and spherical shapes.
Both shapes have the same values within
experimental error.
at room temperature as shown in Figure 7.
The susceptibility was proportional to the
concentration of paramagnetic ions.
4 Analysis and Discussion
10-4-
MnCh-4H2O
FeClj
«CoCl2-2H2O
Fe(NO3)3-9H2O
CuCU- 2HjO
10-6-
1020 1021 1022
Ion Concentration(ions/cc)
Figure 7. The magnetic susceptibility of aqueous
solution as a function of concentration of
paramagnetic ions.
The l H resonance point in pure water differs
from that in the paramagnetic aqueous
solution. The paramagnetic aqueous solution
induced the shift of resonance point due to the
presence of paramagnetic ions. In this study,
we have tried to search a suitable paramagnetic
solution, having the short relaxation time and
nearly zero shift of resonance point in order to
obtain the correct proton resonance frequency in
a low magnetic field.
4.1 The Shift of 'H Resonance Point and
Interaction Factor
For a liquid, the time averaged field at a
nucleus may be divided into three significant
components
Bav = Bo B' + B" (1)
where Bo is the external magnetic field, which is
the main component in Bav. B' is the magnetic
shielding field at the nucleus due to the induced
motion of the electrons in the atom or molecule.
B" is the magnetization field due to the
paramagnetic ions to shorten the spin-lattice
relaxation time T\ of the nuclear spin system.
The dipole interaction between the *H nucleus
and paramagnetic ion is given by [9]
(2)
The field B\ is ascribed to the induced magnetic
dipoles on the surface of a small hypothetical
sphere with its center at the nucleus. This is
the so-called Lorentz or cavity field and has the
value (4H/3)M, where W is the magnetization.
The field Bz is the familiar demagnetizing field,
defined by #2 = -aW, where a is the
demagnetizing factor. The value of a is 47?/3 and
2n for the spherical and infinite cylindrical
sample perpendicular to the field, respectively. It
might be expected that the remaining field #3
due to those paramagnetic ions inside the
hypothetical sphere would be exactly zero.
However, it is found experimentally that #3 may
differ significantly from zero. Therefore, we
define an "interaction factor" q=B3/M. The
expression for B" hence becomes [10]
B" = [(4B/3) - a + q]M (3)
The magnetization H was obtained from the
susceptibility per unit volume according to the
concentration of paramagnetic ions. Also, the

V '
244 Bulletin of Magnetic Resonance
shift of resonance point(B") for paramagnetic
ions with respect to resonance point of proton in
pure water was measured from *H NMR
experiment in various paramagnetic solutions.
Using the magnetic field induced to the 'H
nucleus and the value of magnetization, we
calculated the interaction factor q from eq.(3)
for the fixed spherical and cylindrical samples.
A summary for the various paramagnetic ions is
given in Table 1. The consistency of the
Table 1. Experimental values of the interaction
factor q for the fixed spherical and cylindrical
samples, obtained with eq.(3).
paramagnetic
ions
Fe 3 *
Fe 3 '
Mn 2i
Co 2t
Cu 2 '
chemical
compound
Fe(NO3)3-9HjO
FeCl3
MnCI2-4H2O
CoCl2-2H2O
CuCl2-2H2O
Q
cylinder
2.19
0.85
1.70
1.28
0.39
value
sphere
1.77
0.86
1.21
1.06
0.85
experimental data is indicated by the agreement
between the interaction factors for the
corresponding cylindrical and spherical cases.
The amount of disagreement can be attributed
partly to the experimental error and partly to
the meniscus effect and the lack of perfect
sphericity of the spherical sample.
4.2 Relaxation Time
The spin-lattice relaxation time measured by the
inversion recovery method with a pulse sequence
of 180°(2 /is) - t - 90°( 1 us) - 5 fis(Td) - free
induction decay. The ringing down delay-time Ta
was used to remove the effect of the pulse and
the free induction decay was measured with
time t.
The spin-lattice relaxation time obtained with
the inversion recovery method decreases as the
concentration of paramagnetic ions increases.
The relaxation time measured with the
spherical sample is similar to that with the
cylindrical sample. The 'H relaxation time of
paramagnetic solution containing Co 2 + or Cu 2 *
shows longer than that containing Mn 2 * or
Fe 3+ . The spin-lattice relaxation time of J H in
various paramagnetic solutions is shorter than
that in pure water because of the interaction
between the nuclear spin and paramagnetic
ions. When the number of paramagnetic ions
was increased, the shortening mechanism of the
relaxation time could be understood as follows.
If the number of paramagnetic ions is increased,
the nuclear spin is coupled more with the
magnetic field produced by the paramagnetic
ions. This magnetic interaction between the
nuclear spin and magnetic field of the
paramagnetic ions can contribute to the decrease
in the spin-lattice relaxation time [11].
In case of the short spin-lattice relaxation
time Ti, the following relation generally holds
[12] :
(V)" 1 =
(4)
where Tz is the "natural" spin-spin relaxation
time, and Tz' is the time due to the field
inhomogeneity. The value of Tz" is measured
from the full width at half maximum of the
NMR lineshape.
The linewidth of 'H NMR was brodened when
the concentration of paramagnetic ions was
increased. In case of the aqueous solution
containing Fe 3+ ion, the linewidth was
remarkably increased according to the
concentration of paramagnetic ions. The
linewidth could be broadened by the magnetic
dipole field produced by the paramagnetic ions
at the site of *H nucleus. Normally the dipole
field of the paramagnetic ions has the field
strengths of several thousands times greater than
that due to the magnetic monents of the nucleus,
but it is averaged out at the site of ] H nucleus.
Consequently only a small effect, the linewidth
broadening is occured in the magnetic resonance
[13].
5 Conclusion
For the spherical and cylindrical samples, B"
would be always zero and positive,
respectively, if q were zero. The deviation of
the shift of resonance point between the
experimental results and the theoretical
prediction (q = 0) could be understood as an
effect due to an additional interaction between
the paramagnetic ions and the 'H nucleus.
The spin-lattice and spin-spin relaxation times of
l H NMR in paramagnetic aqueous solution were
shortened as the concentration of paramagnetic ions
was increased.
From these experimental results, we found
that the paramagnetic solution having the
short relaxation time and nearly zero shift of
resonance point is CUO22H2O aqueous
solution. Therefore, the aqueous solution
containing Cu 2+ would be the best candidate to
implement the precise determination of the

Vol. 14, No. 1-4 245
proton gyromagnetic ratio.
Acknowledgement
This work was supported by the Ministry of
Science of Technology and in part the KOSEF
through the SRC of Excellence Program
(1991-94).
References
[1] E.R.Williams and P.T.Olsen, Phys. Rev.
Lett. 42, 1575 (1979).
[2] E.R.Williams, G.R.Jones, J.S. Song,
W. D. Phillips, and P. T. Olsen, IEEE
Trans. Instrum. Meas. IM-38(2), 233
(1989).
[3] H. Nakamura, N. Kasai and H. Sasaki,
IEEE Trans. Instrum. Meas. IM-36, 196
(1987).
[4] N. Bloembergen, Nuclear Magnetic
Relaxation (W. A. Benjamin, New York,
1961), Chap. 4.
[5] J. H. Simpson and H. Y. Carr, Phys. Rev.
Ill, 1201 (1958).
[6] N. Bloembergen, E. M. Purcell, and R. V.
[7]
Pound, Phys. Rev. 73, 679 (1948).
J. T. Arnold, S. S. Dharmatti and M. E.
Packard, J. Chem. Phys. 19, 509 (1951).
[8] N. Bloembergen and W. C. Dickinson,
Phys. Rev. 79, 179 (1950).
[9] W. C. Dickinson, Phys. Rev. 77, 736
[10]
(1950).
A. R. Lim, S. H. Choh, Saemulli 26, 381
(1986).
[11] A. Abragam, The Principles of Nuclear
Magnetism(Oxford Univ. Press, Oxford,
1961), Chap. 3.
[12] D. Pines and C. P. Slichter, Phys. Rev.
100, 1014 (1955).
[13] W. C. Dickinson, Phys. Rev. 81, 717
(1951).

246 Bulletin of Magnetic Resonance
1. Introduction
Magnetic Resonances of 23 Na and 14 N Nuclei
in Single and Multi-Domain
.Crystals of Ferroelectric NaN02
Sung Ho Choh and Kee Tae Han
Department of Physics, Korea University, Seoul
136-701, Korea.
Betsuyaku [1] and Kanashiro et al [2] reported
that the two satellite lines of 23 Na NMR
in a ferroelectric NaNCh crystal were asymmetric,
and that even the central line for Bo//a
deviates appreciably from the first derivative of
the Gaussian line shape. Kanashiro et al [2] tried
to explain this effect in terms of the nuclear
dipole coupling between the Na atoms. Hughes
and Pandey [3] also studied the same effect, and
atempted to explain it by means of the magnetic
dipolar third moments. They concluded that the
asymmetry arises from the perturbation of the
magnetic dipolar coupling, and this is the
intrinsic property of NaNO2 crystal from the
similarity of experimental results obtained with
two different crystals.
In connection with these reports [1-3], the
present work is focused on the 23 Na NMR line
shape as well as the electric field gradient
(e.f.g.) at 23 Na and 14 N sites in NaNO2 crystal
by employing the NMR and NQR technique. Two
NaNO2 crystals prepared here have different
domain states each other: one has a multi-domain
and the other has a single domain state,
respectively.
2. Experimental
A. Procedure
The NaNO2 is body centered orthorombic and
its space group is C2v 20 -Im2m in the ferro-
electric phase at room temperature [4],. The used
sample crystals were grown from the melt. Virgin
crystals of NaNC>2 turned out to be in the
multi-domain state (herein named by Sm). The
crystal of single domin (Ss sample) was prepared
by applying an external electric field of 3 kV/cm
along the ferroelectric axis of a virgin crystal for
8 hrs near its Tc of 163.5°C. Their domain states
were confirmed by empolying optical polarizing
microscopy and etching technique [5].
The NMR signals for 23 Na nucleus were observed
with an rf-frequency of 6 MHz by using
the cw-NMR (Varian WL-112). They were recorded
by setting the magnetic field modulation at
35 Hz with an approximate amplitude of less than
one-third of the resonance line width. The
pulse-NMR (Bruker MSL 200) system was also
employed to investigate the exact line shape of
23 Na with an inversion recovery sequence, and
NQR measurements were made by using the
Robinson type spectrometer [6]. All the
measurements were made at room temperature.
B. Results
Since the 23 Na nucleus has a spin of 1 = 3/2,
three resonance lines are typically observed: one
central and two satellite lines. The line shape of
23 Na NMR in two single crystals of NaNO2 had
been measured [7] by using the cw-NMR
spectrometer. In the Ss sample, as shown in Fig.
1, the asymmetric satellite lines were observed
for Bo//a and Bo//c, consistent with the previous
reports [1-3]. However, in the Sm, the observed
satellite lines was nearly symmetric (see Fig. 2).

Vol. 14, No. 1-4
(a) Bo//c-axis
(b) Bo//a-axis
-4— -+-
510.0 511.0 531.0 531.5 552.5 553.5 (mT)
505.0 510.0 530.5 531.5 558.0 559.0 (mT)
Fig. 1. In the Ss sample, asymmetric satellite lines were observed for (a) Bo//c and
(b) Bo //a with the cw-NMR spectrometer.
(a) Bo//c-axis
(b) Bo//a-axis
510.0 511.0 531.0 552.5 553.5 (mT)
505.0 510.0 530.5 531.5 558.0 559.0 (mT)
Fig. 2. In the Sm sample, symmetric satellite lines were observed for (a) Bo//c and
(b) Bo//a with the cw-NMR spectrometer.
247

248 Bulletin of Magnetic Resonance
The line width of the satellite lines (ABS) in Ss
was found to he broader than that in Sm. whereas
the central line in these two samples is symmetric
and its line shape is nearly the same; it was 0.29
mT (0.21 mT) for Bo//a (B0//c). Meanwhile, the
quadrupole coupling constant (Qcc) and asymmetry
parameter (7?) for 23 Na of these two samples are
nearly the same within the experimental accuracy
(AQcc : + 0.003 MHz, AT? : ± 0.003) [7].
For the 14 N NQR measurements in the two
samples, there were no detectable differences in
their line widths (At 1 * and AP~ ) and NQR parameters.
The quadrupole parameters at the 23 Na
and 14 N nuclei in the Sm and Ss samples are
summarized in Table 1 together with the line
width. The only difference between the Sm and Ss
samples is in the line shape of 23 Na NMR and
its line width.
The line shape of 23 Na NMR in the two
samples obtained with the pulse-NMR is shown in
Fig. 3 and 4, respectively. In both of two
samples, the central line has one peak and its
line shape is a Gaussian as well known [1-3]. So
the central line is not shown in Fig. 3 and 4,
where PL S(III) (i4i s(m) ) is the satellite line of low
(high) frequency, and superscript s(m) stands for
the single domain (multi-domain) state. In Fig. 3,
(a) shows that each satellite line obtained with Ss
has two peaks (PL S ' and VL S , m s and v\\ s ' ),
and (b) displays their first derivative curves.
However, in the Sm sample, the satellite line (vi m
or PH" 1 ) has only one peak and it is well fitted
with a Gaussian line shape as shown in Fig. 4,
which displays VL m representatively since the two
satellite lines all are symmetric. The inner set
(vi s and vn s ) of the two couples obtained with
Ss [Fig. 3 (a)] corresponds to those (^L m and fH m )
with the Sm sample (Fig. 4). The outer set (vi s '
and m s ' ) in Ss is newly observed additionally.
3. Discussion
Betsuyaku and Kanashiro et al [1-2] had
measured the NMR line shapes for Bo //a and
Bo//c. Their data for Bo//a was distinctively
asymmetric but those for Bo//c was hardly asymmetric.
They, by considering the dipolar interaction,
argued that this effect arised from the
geometrical arrangement of Na atoms in NaNO2
crystal having two fold symmetry along the
b-axis. Namely, the dipole coupling between the
two Na nuclei along the a-axis contribute considerably
to the asymmetry of the satellite lines
because the interatomic distance along the a-axis
is about two-thirds of that along the c-axis.
Kanashiro et al [2] determined the sign of the
Qcc at Na in NaNCh by comparing the line
shape of the satellites for Bo //a with the
calculated line spectrum of two interacting I =
3/2 spins. However, Hughes et al indicated that
there was some uncertainty in the determination
by Kanashiro et al since the resonances showed a
substantial inhomogeneous quadrupole broadening,
and no steps were taken to confirm that the
asymmetry was indeed associated with the dipolar
interaction and not with the quadrupole
broadening. Hughes et al confirmed the sign of
Qcc determined by Kanashiro et al by comparing
the magnetic dipolar third moment with the
observed asymmetry, and reported that the
asymmetry due to the quadrupole broadening was
dominant at some crystal orientations by measuring
quantitatively the asymmetry of inhomo-
Table 1. The quadrupole parameters and the line widths at room temperature.
samples
Sm
Ss
23 Na NMR
Qcc(MHz) .»?(%) ABs(Bo//a) ABs(Bo//c)
1.094 11.0 0.30 mT 0.23 mT
1.094 11.0 0.38 mT 0.25 mT
14 N NQR
Qcc (MHz) 7?(%) At> + (kHz) Ai""(kHz
5.490 35.7 0.66 0.53
5.490 35.7 0.67 0.53

Vol. 14, No. 1-4
(b)
240 230
(KHz)
-250 (KHz)
240 230 (KHz) -240 -250 (KHz)
Fig. 3. The satellite lines of 23 Na NMR in the Ss sample observed for B0//c with the
pulse-NMR spectrometer, (a) Each satellite line has two peaks, (b) the first derivative
curve of (a)
m
I
240 230 KHz
Fig. 4. The satellite line of 23 Na NMR in the Sm sample observed for Bo//c with the
pulse-NMR spectrometer. The satellite line has only one peak and is well fitted
with a Gaussian line shape, where + denotes signal trace and the dashed line is
a best fitted Gaussian line shape.
\
249

250 Bulletin of Magnetic Resonance
geneous quadrupole broadening. They also suggested
that the asymmetry was the intrinsic property
of this crystal by observing the similar quadrupolar
broadening for two different crystals.
As can be seen in Fig. 1, our NMR results
show that the satellite line shape in Ss is
asymmetric, which is in accordance with the
previous reports [1-3]. However, those in Sm is
nearly symmetric (Fig. 2). The exact line shapes
of Fig. 3 and 4 obtained with the pulse-NMR
system may provide a decisive clue for the origin
of the observed asymmetry. Fig. 3(a) shows that
each satellite line in Ss has two peaks, and the
line intensity of the inner set is stronger than
that of the outer one. Fig. 3(b) displays their
first derivative curves in Fig. 3(a). It can be
expected to obtain the asymmetric satellite line of
Fig. 1 by drawing a curve as an envelope of the
satellite line made of two peaks of Fig. 3(b).
Whereas, in the Sm sample, the satellite line has
only one peak as shown in Fig. 4 and it is found
to be well fitted with a Gaussian line shape.
Our experimental results shown in Fig. 4
suggest that the asymmetry due to the dipole
coupling is quite small or very weak based on the
fact that the satellite line in the Sm is well fitted
with a Gaussian. Two peaks in the satellite line
of Fig. 3 may imply that there are possibly two
different absorption centers in the Ss sample.:
one is a normal set of the satellite lines
corresponding to those in Sm and the other is a
new one (vi s ' and PH S ' ). Such a new set might
have orignated from the strained part of the
crystal during the process of the polarization
reversal to make the virgin crystal into a single
domain state. Meanwhile, as listed in Table 1,
the Qcc and r? for 23 Na in these two samples
with the cw-NMR were the same within the experimental
uncerntainty [7]. NQR results such as
line widths and quadrupole parameters of 14 N in
Sm and Ss are also similar to each other. The
appreciable difference between the Sm and Ss
samples is only in the line shape of 23 Na NMR.
These facts on the e.f.g. at 23 Na and 14 N sites
supports the proposal of the existence of two
absorption regions in the Ss sample. The asymmetry
and broad line width of the satellites in
the Ss sample can be understood by means of the
formation of a new region in the crystal due to
the strain induced by the polarization reversal.
From our experimental results, one can
propose that the asymmetric line shape of the
satellite line of 23 Na is due to an imperfection
caused by the applied electric field rather than
the magnetic dipole coupling in the crystal. In
general, crystal imperfection may be produced
during the crystal growth [8] or by the external
stress [9], and due to the external electric field
as the present case [10].
4. Conclusion
We observed the symmetric line shape as well
as the asymmetric one in the 23 Na NMR in the
ferroelectric NaNO2 crystal by employing the
cw-NMR and pulse NMR technique. The pulse
NMR spectrometer provided a better resolved resonance
line shape than the cw-NMR. Each satellite
line obtained with Ss has two peaks, while
only one peak in the Sm sample. Thus the
observable asymmetry might have originated from
the existence of two absorption regions in the
crystal such as a normal region and a new one
abnormally induced in NaNO2 crystal. Such a
new region seems to be due to the internal strain
induced during the process of polarization reversal
by the external electric field. These results imply
that the asymmetry is not the intrinsic property of
an NaNO2 crystal but due to a kind of crystal
imperfection caused by strain.
Acknowledgements
This work is supported by the Korea Science
and Engineering Foundation through the SRC of
Excellence Program (1991-94). Authors are
grateful to Professor R. Blinc of University of
Ljublijana for his helpful discussion during his
visit to Korea in February 1992.

Vol. 14, No. 1-4
References
[1] H. Betsuyaku, J. Phys. Soc. Japan. 27,
1485 (1969)
[2] T. Kanashiro, T. Ohno, and M. Satoh, J.
Phys. Soc. Japan. 54, 2720 (1985).
[3] D. G. Hughes and L. Pandey, J. Mag.
Reson. 75, 272 (1987)
[4] M. I. Kay and B. C. Frazer, Acta cryst.
J4, 56 (1961)
[5] K. T. Han, Ph. D. thesis, Korea Univ.
(1992)
[6] J. Lee and S. H. Choh, Rev. Sci. Instr.
53, 232 (1982)
[7] K. T. Han, H. W. Shin, I. W. Park and
S. H. Choh, J. Korean. Phys. Soc. 25,
67 (1992)
[8] S. H. Choh, J. Lee and K. H. Kang,
Ferroelectrics 36, 297 (1981)
[9] K. T. Han, T. H. Yeom and S. H. Choh,
Ferroelectrics 107, 349 (1990)
[10} Private communication with Prof. R. Blinc
of University of Ljublijana
251

252
1. Introduction
Knight Shifts and
Spin Dynamics in Disordered
Systems
M.J.R. Hoch and S.T. Stoddart
Department of Physics and
Condensed Matter Physics Research Unit,
University of the Witwatersrand, Johannesburg
The metal insulator (MI) transition is a
problem that has received much attention in
solid state physics. Heavily doped
semiconductors have featured prominently in
this work with Si:P the archetypal system.
In order to explain the observed low
temperature properties, such as the magnetic
susceptibility, of MI systems in the vicinity
of the critical concentration, n(;, of dopant
atoms, a phenomenological two—fluid model
has been proposed [I]. The present work is
concerned with interpreting 29 Si NMR
relaxation time measurements and Knight
shifts for Si:P and Si:(P,B) in the vicinity of
nc. The results are analyzed in terms of
available theory in the context of the twofluid
model.
2. The Two-Fluid Model and the Bhatt-
Lee Theory
For the just metallic or just insulating
phases of MI systems, the two—fluid model
distinguishes between two types of electron
spins. As the transition is traversed, the
proportions of the two fluids change. The
fluids are comprised of localized moments
associated with isolated dopant atoms, or
small clusters of dopant atoms, on the one
hand, and delocalized moments on the other.
In broad terms, the localized moments
dominate in determining the magnetic
properties in the vicinity of nc, while the
delocalized moments determine the electrical
properties, such as the conductivity.
Bulletin of Magnetic Resonance
For n < nc, the localized moments
constitute a disordered antiferromagnetic
system. The exchange Hamiltonian may be
written in the usual way as
H = Hi - 1 ' -J '
where Jij is the exchange coupling between
spins i and j and Si and Sj are the spin
operators.
In order to explain the behaviour of
the magnetic susceptibility x with
temperature for n < nc,Bhatt and Lee [2]
have developed a theory in which the
exchange coupling Jy between nearest
neighbour pairs of localized moments is used
to separate the moments into two groups.
In simple terms, spin pairs with Jjj >> kT
are tightly coupled or frozen in the singlet
state and effectively do not contribute to X-
The remaining spins do contribute and the
susceptibility may be written in terms of the
Curie law susceptibility as
n Curie (1)
Numerical procedures were used to
determine ne(T), the effective number of
spins at temperature T. Good quantitative
agreement was obtained with available
experimental susceptibility data. Bhatt and
Lee used the following form for the J
distribution in their calculations :
P(J) « J-tt, with 0.6 < a < 0.8. This led to
X « T-«, as observed.

Vol. 14, No. 1-4
3.
29 Si Spin Relaxation and Localized
Electron Spin Dynamics (n < nc)
Previous work [3] has provided strong
evidence that localized moments dominate in
determining the 29 Si spin lattice relaxation
times at low temperatures. These moments
constitute an exchange coupled reservoir to
which the nuclear spin system is coupled via
the dipolar interaction. Hoch and Holcomb
[3] have analyzed available Ti results using
a model which allows for spin diffusion
to the localized moments, which exhibit
fluctuations in orientation with a frequency
related to the strength of the exchange
coupling to neighbouring spins. Frozen spin
pairs have been excluded by introducing an
effective number of spins in the spirit of the
Bhatt—Lee approach to the susceptibility.
Available T( data, measured at various
fields B, at 1.5 K and 13.5 mK were fitted
reasonably well by choosing the spectral
function for the spin fluctuations to have the
form f(w) « »/w and using calculated values
for other quantities, such as the spin
diffusion coefficient D.
The best fits were obtained by keeping
the diffusion barrier radius, b, constant
independent of the field used. Calculations,
however, suggest that b should vary as in B.
We now propose that the form of the
spectral function can be obtained from the
form of the J distribution used by Bhatt and
Lee. An outline of the treatment is given
below. Converting the J distribution into a
r distribution and integrating over all
correlation times for the unfrozen spins gives
J(u/) = P(T) dr
with P(T) « V 7 " 2 "" • The form of the J(w,r)
may be chosen in various ways corresponding
to different possible forms for the
correlation function. Our calculations
suggest that the form of J(w) is not very
sensitive to this, and for simplicity, we use
an exponential correlation function, leading
to a Debye form for J(w,r). If we put a - 1,
this leads to J(w) « 70;, as used in the work
referred to above. Numerical integration is,
in general, necessary for other values of a.
For values of a < 1, we obtain a
spectral function which varies as l /uP- and
therefore more slowly with frequency than
the l /w form. This permits b to vary when
fitting the experimental data. Further
details will be published elsewhere.
The approach to the spectral function
for an amorphous antiferromagnet, outlined
above, is consistent with the ideas of the
253
Bhatt-Lee theory. Susceptibility measurements
as a function of n and T and nuclear
relaxation measurements as a function of n,
B and T can be explained in terms of the
properties of the localized fluid component.
4. 29Si Knight Shifts (n > nc)
Knight shifts have previously been measured
as a function of dopant concentration n at
4.3 K and 1.5 K [4,5,6] in various magnetic
fields. At high concentrations the Knight
shifts tend towards the Pauli susceptibility
behaviour (xP « n^*), as expected for a
metal. At lower concentrations the values
fall below the Pauli susceptibility
predictions.
Using a tight binding approximation
based on the approach given by Kaveh
and Liebert [7], we have given a semiquantitative
explanation [8] for the observed
Knight shift behaviour for both Si:P and
Si:(P,B). The delocalized moment fluid
determines the Knight shift.
In order to see whether there is any
temperature dependence of the Knight shift,
we made measurements on two just metallic
samples at temperatures down to 50 mK in
an Oxford dilution refrigerator. The samples
were in the form of a stack of wafers, which
were well anchored thermally to the high
purity copper tail used in the refrigerator.
The field of 1 T was supplied by a high
homogeneity superconducting solenoid.
The results are shown in Figure 1.
10
10
10 -6
10 -2
,Si:P
oSi:P
(n/nc = 1.6)
(n/nc = 1.6) Ref. 5
• Si:(P,8)(n/nc = 1.1)
°Si:(P.BHn/ne a 1.1) Ref. 6
10 1
Temperature (K)
Figure 1
The mean Knight shift for Si:P (n/nc = 1.6)
andSi:(P,B) (n/nc = 1.1) as a function of
temperature down to 50 mK.
Within experimental uncertainty it can be
seen that there is no temperature
dependence of the mean Knight shift
over the temperature range 4 K — 50 mK for
either Si:P or Si:(P,B).
In terms of the two-fluid model this
aro irarxr \irpaV
10

254 Bulletin of Magnetic Resonance
interactions between the localized and
delocalized spins. The local susceptibility of
the delocalized electrons does not change
with temperature.
5. Conclusion
Measurements of the 29 Si relaxation rates
and Knight shifts as a function of donor
concentration, magnetic field and
temperature have provided evidence which
supports the two—fluid model for the MI
transition in Si:P. The Ti measurements
provide information on the spin dynamics of
the localized moments, while the Knight
shifts probe the properties of the delocalized
moments.
The relaxation rate
interpreted using ideas
Bhatt—Lee theory for
susceptibility. The form
results may be
based on the
the magnetic
of the spectral
function for spin fluctuations in the
amorphous antiferromagnet system has been
deduced using an accepted form for the
distribution of exchange couplings.
Knight shifts may be explained using a
tight binding model for the delocalized fluid.
No temperature dependence of has
been found over the range 4 K — 50 mK.
This may be interpreted to mean that
interactions between the two fluids, which
occupy spatially distinct regions in the
sample, are very weak.
6. References
1. H. Alloul and P. Dellouve, Phys. Rev.
Lett. 59, 578 (1987).
S. Sachdev, R.N. Bhatt and
M.A. Paalanen, J. Appl. Phys. 63,
4285 (1988).
2. R.N. Bhatt and P.A. Lee, Phys. Rev.
Lett. 48, 344 (1982).
3. M.J.R. Hoch and D.F. Holcomb, Phys.
Rev. B 38, 10550 (1988).
4. S. Kobayashi, Y. Fukagawa, S. Ikehata
and W. Sasaki, J. Phys. Soc. Jpn. 45,
1276 (1978).
5. M.J. Hirsch and D.F. Holcomb, Phys.
Rev. B 33, 25201 (1986).
6. M.J.R. Hoch, U. Thomanschefsky and
D.F. Holcomb, Physica B 165/166, 305
(1990).
7. M. Kaveh and A. Liebert, Phil. Mag.
Lett. 58, 247 (1988).
8. S.T. Stoddart and M.J.R. Hoch - to be
published in Phys. Rev. B.

Vol. 14, No. 1-4 255
Numerical Design and Evaluation of Broadband
Pulse Sequences for 1=1 spin systems
Debra Lynne Mattiello, Jonathan Callahan, Todd Alam^ and Gary Drobny
Chemistry Department, University of Washington,
Seattle WA USA 98195
INTRODUCTION
Quadrupolar coupling constants of
170 kHz result from reduced motional
averaging in biological polymers. The
decay rates of Zeeman and quadrupolar
order are direct measures of the spectral
densities of motion. Information on the
orientation-dependence of relaxation rates
of Zeeman and quadrupolar order, Tlz and
Tlq respectively, assists in understanding
the dynamics of molecules. 1 Sensitivity and
dynamic range considerations mandate
optimum efficiency and uniformity over
large spectral widths. The design of
composite pulses to create broadband
excitation and inversion is well established
for deuterium NMR. 2 " 11 The numerical
optimization of existing composite inversion
pulse sequences was performed in order to
increase the uniformity of excitation over
spectral widths of 250 kHz. 2 " 7
Pulse sequences to create
quadrupolar order over moderately broad
spectral widths have recently been
proposed. 12 " 14 The pulse sequence design
f Present Address: University of New Mexico,
Albuquerque, NM
of Wimperis is a good starting place from
which to numerically optimize for the
creation of quadrupolar order. 13 Broadband
quadrupolar order is transferred to
detectable magnetization with a 45 degree
pulse and an additional refocusing pulse
eliminates large phase corrections. 15
Numerical optimization of the conversion
from quadrupolar order to well-behaved
transverse magnetization is worthy of
investigation.
A program developed in our
laboratory for the numerical optimization of
pulse sequences has been modified for 1=1
spin systems. 15 " 18 This research is part of an
ongoing investigation into the local and
global dynamics of oligonucleotides utilizing
solid state deuterium NMR. At the present
time, the dynamics of the sugar rings of
DNA are under study. The lower levels of
hydration of DNA exhibit rigid-lattice
lineshapes. Information on orientation
dependence and the substantiation of
motional models requires increased
sensitivity and large spectral windows.

256
METHODS
Solid state deuterium NMR spectra
were obtained at 76.72 MHz on a homebuilt
spectrometer controlled by a DEC
micro VAX II. Both the inversion
sequences and the broadband Jeener-
Broekaert experiments were phase cycled to
eliminate double quantum coherence. 19 ' 20
Phase shifting was accomplished with a
homebuilt digital phase shifter. The sample,
a labeled nucleoside, 2"-deutero-2'deoxyguanosine,
was prepared by Jerome
Shiels at the University of Washington.
Either 2K or 4K scans were taken with a
recycle delay of 5.0 seconds. The field
strength was 100 kHz and the dwell time
was 200 nanoseconds. Each spectrum was
acquired with 4096 points.
Calculations were accomplished on a
DEC UNIX 3100 workstation. The strategy
thus far has been to parameterize the pulse
sequence and generate random trial pulse
sequences. The basis set proposed by Vega
and Luz was used. The coherences of
interest were single basis elements rather
than linear combinations of basis
elements. 19 The expectation value of -Iz
Qz quadrupolar order, or Iy was compared
with the target function over a specific
spectral width. These quality factors
quantify the performance of the pulse
sequence as a function of the quadrupolar
frequency. Excitation profiles of the
expectation value as a function of reduced
frequency illustrate overall smoothness and
breadth. Evolution profiles of the elements
of the density operator as a function of time
demonstrate the effects of rf pulses and
evolution under a strong quadrupolar
Hamiltonian. Additional three-dimensional
graphics developed in our laboratory assist
in visualizing the transfer of coherences. 22 "
RESULTS
Bulletin of Magnetic Resonance
Measurement of spin-lattice
relaxation times and investigation of the
orientation-dependence of Tt in solids are
important tools for understanding
dynamics. 1 Inversion pulse lengths of more
than 4 microseconds compromise the
inversion breadth and uniformity across
wide line deuterium powder patterns- 7 A
variety of composite pulse schemes have
been proposed for spectra with widths
approaching the Rabi frequency of the
radiofrequency pulse. The composite
excitation triplet designed by Levitt, Suter
and Ernst and supercycled in the method
proposed by Levitt in order to create
broadband inversion of deuterium
lineshapes was numerically optimized. •
The sequence consists of the composite
excitation pulse, 45o9O18O135o, supercycled
in the triplet form, O0 O9Q 3>O Only the
flip angles of the excitation triplet were
parametrized. The sequence has already
been shown to invert rigid-lattice deuterium
spectra with an rf field of 139 kHz while a
weaker field led to a small loss in
performance. 9
Higher order pulses would lead to
longer total pulse lengths. The total
duration of the composite pulse should be
short to avoid irreversible loss of
magnetization during the excitation. 9
Experimental spectra acquired with the
optimized composite excitation pulse of
43010018Q142Q, supercycled as above (B)
and the Levitt triplet (A) are displayed in
figure 1.
Recent application of Tycko's use of
the Magnus expansion for the design of
composite pulses has led to the
development of new excitation schemes. 8 ' 11

Vol. 14, No. 1-4 257
B
P
460 n6
kHz
-46a
fig-1
The modified Jeener-Broekaert pulse
sequence was founded on the model of a
composite excitation pulse broadband with
respect to rf field strength 13 ' 14
The multipulse sequence eliminated
the frequency selection of the original
Jeener-Broekaert experiment. The
broadband sequence has been utilized to
measure the spectral densities of liquid
crystals. 21 Increased signal sensitivity
becomes highly desired as one goes to
broader linewidths, lower Larmor
frequencies and smaller biological samples
with dilute nuclei.
Computer search routines to create
broadband quadrupolar order were
performed with 10, 8 and 7 parameters.
The 10 parameter search consisted of 4
pulses, 3 phases, and 3 seperate delays. The
8 parameter search had a single delay
parameter. An assortment of the resulting
sequences were tested experimentally. The
"91" sequence was a 10 parameter search
optimized over a spectral width of 300 kHz.
It proved to perform well in breadth and
sensitivity. Experimental spectra obtained
with the Wimperis sequences A and B and
the "91" pulse sequence, C, found in this
study, are shown in figure 2.
400 -400
A: 900 -4.0p. -67.5270 -4.0n -4590 -2.0p. -4590
B: 900 -4.0(i -75270- 4.0M. -52.590 -2.0M -4590
C: 1130 -5.5n -72 284 -3.2n -52142 -3.9M -59112
CONCLUSION
fig-2
The use of solid state deuterium
NMR as a probe of dynamics in DNA offers
the luxury of selective labeling. The same
luxury introduces some of the limitations of
site-selective wide line spectroscopy. The
technique enables one to focus on the
dynamics of single positions within large
molecules. The small, precious samples are
dilute in the observe nuclei yet one can
monitor the onset of motion in both a local
and global fashion as water is added to the
spaces in DNA. The technique requires high
power, short pulses for broad lineshapes,
resistant samples, and extensive phase
cycling and signal averaging. The dry DNA
has longitudinal relaxation times of several
seconds making signal intensity even more
elusive. For these reasons, this investigation
aims to numerically optimize existing pulse
sequences for the creation of selected
coherences over static deuterium powder
linewidths of 250 kHz. Analysis of the

258
evolution of the system provides clues for
producing the very best tailored excitation.
REFERENCES
1 R.R. Void and R.L. Void, "Advances in Magnetic
and Optical Resonance". Vol. 16, Academic Press,
San Diego (1991)
2 R. Tycko. Phys. Rev. Lett. 51. 775, (1983)
-* M.H. Levitt. D. Suter. and R.R. Ernst, J. Chem.
Phys.,S0, 3064 (1984)
4 R. Tycko, E. Schneider. A. Pines. J. Chem. Phys.,
81, 680 (1984)
5 R. Tycko, H.M. Cho, E. Schneider, A. Pines, J.
Mag. Res., 61, 90 (1985)
6 M.H. Levitt. Prog, in NMR Spec. 18, 61 (1986)
7 D J Simonivitch. DP Raleigh. E.T. Olejniczak.
and R.G. Griffin../. Chem. Phys.,84, 2556 (1986)
1 S. Wimperis and G. Bodenhausen. J. Mag. Res.,
69.264(1986)
9 N.J. Healon. R.R. Void, and R.L.Vold, J. Mag.
Res..17. 572(1988)
10
DP. Raleigh. E.T. Olejniczak and R.G. Griffin,
J. Mag. Res.,$l, 455 (1989)
11 S. Wimperis../. Mag. Res..83.509 (1989)
12 J. Jeener and P. Broekaert. Phys. Rev. , 157, 232
(1967)
13 S. WimperisJ. Mag. Res.. 86, 46 (1990)
S. Wimperis and G. Bodenhausen. Chem. Phys.
Lett. .132. 194(1986)
15
G. Hoatson.J. Mag. Res.. 94. 152-159 (1991)
16
S.J. Glaser and G.P. Drobny. "Advances in
Magnetic Resonance" (W.S. Warren. ED) Vol 14.
Academic Press. San Diego. 1990
17 H. Liu. S.J. Glaser. G.P. Drobny. J. Chem. Phys..
93.7543.(1990)
18 B. Ewing. S.J. Glaser and G.P. Drobny, Chem.
Phys. Lett. J47. 121 (1990)
19 A.J Vega and Z. Luz. ./. Chem. Phys.. 86. 1803-
1813(1987)
Bulletin of Magnetic Resonance
20 R.R. Void and G. Bodenhausen. J. Mag. Res..
39, 363 (1980)
21 G. Hoatson. T. Tse, R.L. Void. J. Mag. Res., 98.
342 (1992)
zl J. Callahan. D. Mattiello and Gary Drobny,
ISMAR conference, Vancouver , B.C., July 1992,
Poster 54

Vol. 14, No. 1-4 259
1 Introduction
A BASIC Program to Calculate
the Evolution of Cartesian Product Operators
Stefano Mammi
Biopolymer Research Center, National Research Council
Via Marzolo 1,35131 Padova, Italy
The introduction of the product operator
formalism [1] has greatly improved the
description of multiple-pulse NMR
experiments allowing the understanding of the
fate of the magnetization in a direct manner.
The use of this formalism has become
widespread [2] because of its advantages ower
both the complete density matrix approach
and the method of vector diagrams. It is
much simpler than the former and it permits a
clear description of the results while
remaining rigorous. With respect to the
latter, it allows one to visualize all the states
of the magnetization and to follow their
evolution over very complicated pulse
sequences.
The rules that govern the evolution of
product operators are very simple and lend
themselves to automation by means of
computer programs. Automation is especially
desirable considering that the description of
any two-dimensional experiment leads quickly
to very long expressions which can be affected
by trivial mistakes.
Recently, computer programs that perform
such calculations have been reported in the
literature. Among these are the program by
Nakashima and McClung [3] and the more
recent one by Shriver [4]. The first was
written in FORTRAN 77 and describes the
evolution of product operators in the
spherical basis [3]. While this approach is
extremely useful in following coherence
pathways and thus deriving phase cycling
schemes, many times the use of the cartesian
basis is preferable as in the development of
new pulse sequences.
The second program [4] was written in the
new computer language Mathematica whose
major advantage is the easy simplification of
algebraic expressions. This program was
written for Macintosh systems and is not yet
available for IBM personal computers. The
input seems to be stepwise and rather
cumbersome.
The program presented here runs within
MS-DOS and describes the evolution of
cartesian product operators. The program,
named "EVOLVE", was written using the
QuickBasic (C) 4.50 language. The input is
from a file which contains all the information
pertaining to the spin system and the pulse
sequence, and the output is filed separately.
2 Features of the Program
A sample input file is presented in Fig. 1. In
this example, an HMQC experiment {5] is
applied to a system composed of an X
nucleus and two protons, only one of which is
coupled to the heteronucleus.
The first information in the input file is the
spin system which can be made of up to four
spins, denoted by capital letters. The initial
magnetization is entered next.

260 Bulletin of Magnetic Resonance
mi Spin System
HHX
### Initial Operators CAxis(Sp#)Axis(Sp#) : coeffj
### Terms in sine or cosine must nave exactly 6 characters in parenthesis
z(2): *1
### Coupling Constants (Spin 1,Spin 2,J)
2,3.90
### Sequence (1 Line for Name, N Lines for Pulse Sequence)
HMQC - 2 spins + 1 spin
P(90)H PH1
D1
P(90)X PH2
DO
P(180)H PH3
DO
P(90)X PH4
D1
A0 PH5 DEC(X)
### Phase Cycles (as Brutcer, separated by commas)
PH1,0
PH2,0,1,2,3
PH3.0
PH4.0
PH5,0,3,2,1
### Delays: D#=Num/Oen*J(Sp#,Sp#) (#, Numerator, Denominator, Spin 1, Spin 2)
1,1,2,2,3
### Do you want to skip the Evolution under Chemical Shift?
N
### Print out only the Observable Operators?
Y
Figure 1. Sample input file for an HMQC sequence applied to a (H + HX) spin system.
In listing the operators, the spins are
numbered from one to four to prevent
ambiguities among like spins. Any number of
operators can be listed with appropriate
coefficients as required. It is not necessary to
start with equilibrium magnetization: for
example, the magnetization of any spin can be
neglected. Moreover, it is possible to follow
the evolution of a specific operator, e.g.,
2x(2)z(3), generated after the first Dl in the
sequence of Fig 1, by restricting the initial
input to just that operator with its own
coefficients, e.g., +cos(Q 2*D1), utilizing an
appropriately shortened pulse sequence.
Next, the scalar coupling network is described
in terms of the spins which are coupled and
the relevant coupling constant.
The format for the pulse sequence is very
similar to the Bruker one. This allows for
simple transcription of sequences already in
use and for easy modification of any part of
the phase cycling scheme. Each pulse is
written as a "P" followed by the flip angle in
parenthesis, by the spin(s) to which it is
applied and by the phase cycle to be used.
Only 90° and 180° pulses are currently
accepted. Each delay is written as a "D"
followed by a single digit. The acquisition is
referred to as "AQ" followed by its own phase
cycle. Up to ten different phase cycles can be
utilized each containing up to 128 steps. Each
step is recorded as a number according to
the usual notation: 0= +x; 1 = +y; 2 =
-x; 3 s -y.
Many sequences require decoupling during
specific delays, including the acquisition.

Vol. 14, No. 1-4 261
z(2):
-- 90 (H)+x -->
y(2): -1
- D1 -->
2x(2)z

262 Bulletin of Magnetic Resonance
EVOLVE will "decouple" any spin if an
appropriate statement is added on the same
line of any delay.
EVOLVE was written specifically for
sequences in which some delays are inversely
proportional to certain scalar coupling
constants as in heteronuclear correlation
experiments. Proper space is provided in the
input file for defining such cases.
The user is then asked to indicate if
evolution under both coupling and chemical
shift should be taken into account or if the
latter should be neglected. Finally, the user
chooses whether all the resulting operators or
just the observable ones should be printed out
after the acquisition step.
The program runs through the pulse
sequence as many times as required by the
phase cycle. If only the observables are
chosen as output, the final intensities of the
signal are written in two separate files, one for
the real part and one for the imaginary part.
At the end of the phase cycle, these files are
read and final simplifications are carried out.
In Fig. 2, a portion of the output obtained
with the input file of Fig. 1 is reported, Le., the
result from the first of the four steps of the
phase cycle and the signal obtained at the end
of the four step cycle. It can be seen that the
term y(2) generated by the first 90° pulse gives
rise only to antiphase terms at the end of the
first Dl, because Dl = 1/2J23.
The signal from the proton not coupled to
the X-nucleus is present in the first
acquisition, but is canceled out at the end of
the four step cycle while the single quantum
terms from the proton coupled to the Xnucleus
add up to generate the final signal.
The BASIC language does not have built-in
routines for the simplification of algebraic
expressions. A sizable portion of the program
is devoted to such routines. Beside the more
trivial expressions containing n/2, EVOLVE
is able to handle terms containing sin(^/4) or
cos(n/4), most commonly encountered in
heteronuclear sequences. The program does
not evaluate these expressions; rather, it
simplifies them according to the rules it
knows. This entails a longer computation
time but with the advantage of eliminating all
numerical coefficients. This compromise was
found satisfactory.
The only terms that would be useful to have
simplified and that the program is unable to
handle at this stage are those including sin(20)
or cos(20) terms, encountered for example
when there is a 180° pulse in the middle of a
delay. This is not a serious limitation of the
BASIC language, especially considering that
even in Mathematica this simplification must
be explicitly requested by the user.
3 Conclusions
EVOLVE has been applied to numerous
complicated pulse sequences avoiding lengthy
and monotonous calculations and providing
the results in a way suitable for
straightforward analysis.
A copy of the program, including the source
code, can be obtained by sending a 5.25" or
3.5" diskette and return postage to the author.
4 References
[1] O. W. Stfrensen, G. W. Eich, M. H.
Levitt, G. Bodenhausen, and R. R. Ernst,
Prog. NMR Spectrosc 16,163 (1983).
[2] See for example H. Kessler, M. Gehrke,
and C. Griesinger, Angew. Chem. Int. Ed.
Engl 27,490 (1988).
[3] T. T. Nakashima and R. E. D. McClung, /.
Magn. Reson. 70,187 (1986).
[4] J. W. Shriver, /. Magn. Reson. 94, 612
(1991).
[5] A. Bax, R. H. Griffey, and B. L. Hawkins
/. Magn. Reson. 55,301 (1983).

Vol. 14, No. 1-4 263
SELECTIVE LONG-RANGE POLARI
ZATION TRANSFER via DEPT.
Introduction
T. Parella, F. Sanchez-Ferrando* and A. Virgili.
Departament de Quimica. Universitat Autonoma de Barcelona,
08193 Bellaterra, Barcelona, Spain.
The structural assignment of organic compounds
containing quaternary carbons and/or heteroatoms is
often greatly helped by measurements of long-range
proton-carbon coupling constants, "J^, with particular
emphasis on two-bond and three-bond couplings which
usually show values in the range 3-10 Hz [1,2], depending
on carbon hybridization, torsion angles, substttuent
electronegativity and orientation, etc.
Selective ID NMR methods, such as the selective
INEPT method proposed by Bax [3], have long been
used to reveal connectivities with a given proton.
Thus, application of a selective INEPT (or, more
frequently, refocussed INEPT) pulse sequence on a
well resolved proton, with delays optimized for a longrange
heteronuclear coupling (usually around 5-7
Hz), results in a ID carbon spectrum displaying large
intensity enhancements at the carbons coupled (at
long range) with the perturbed proton. Furthermore,
the intensity of these carbons shows a maximum when
the true "iCH value is used for the delay optimization
of the selective INEPT sequence. We have recently
shown the use of this dependence to obtain a quick
estimate of "J^, in several polycyclic derivatives [4].
Selective 2D methods, however, are to be
preferred because they can easily yield accurate values
for the desired long-range coupling constants, provided
the perturbed proton appears as a well resolved multiplet
in the proton spectrum. Thus, the selective spin flip
method first proposed by Bax and Freeman [5] has
been widely used for the determination of "J^ values
from a given proton.
In recent years a number of methods have been
suggested which combine the intensity enhancements
obtained from selective polarization transfer with the
easy measurement of "J^ characteristic of the selective
spin flip method. Thus, Jippo et al. proposed a 2D
selective INEPT method [6] which consists in a
conventional 2D INEPT pulse train containing a
selective spin echo sequence. We have modified this
method by delivering all decoupler pulses in the
selective mode [4], and in this way we have suppressed
the artefact peaks otherwise observed with this method.
Another combination of polarization transfer
and selective spin flip was proposed by Uhrin et at. [7].
In this method, a standard non-selective DEPT
preparation period (optimized for I JCH) is followed by
a conventional selective (or semiselective) spin flip
sequence, yielding a J-resolved 2D spectrum displaying
the desired long range couplings. More recently,
Poppe and van Halbeek [8] have introduced inverse
detection, either in ID or 2D methods, by combining
selective polarization transfer with long range
heteronuclear coupling measurements in lH-detected
sequences.
We now report two new u C-detected, 2D Jresolved
sequences, based on a DEPT pulse train,
which allow fast and accurate measurements of longrange
heteronuclear coupling constants from well
resolved protons. Both sequences are particularly
useful for the determination of couplings to quaternary
carbons, and their main feature is that all proton
pulses are delivered as soft, selective pulses (20-30
ms).

264 .
Parametrization
Starting from the 1D-SDEPT sequence (Scheme
1), we have derived the new sequences 2D-SDEPT1
(Scheme 2) and 2D-SDEPT2 (Scheme 3), in which
SDEPT stands for Selective DEPT. Since ail proton
pulsesareselective (yBjln=20-30 Hz), both sequences
achieve a selective polarization transfer from the
pulsed proton to long-range coupled carbons, provided
that the fixed interpulse delay A is optimized for longrange
couplings. Both sequences yield comparable
results in similar experiment times.
scheme 1
scheme 2
90° 180
scheme 3
180° 4>° 180°
i i i i
i A i A < Evolution •
90" 180°
180° 90° 180°
In both sequences the incrementable evolution
period t,, which brackets a selective spin echo moiety,
allows only the evolution of heteronuclear long range
couplings which will therefore be detected in Fl
dimension in the final 2D J-resolved spectrum.
Bulletin of Magnetic Resonance
We have studied the dependence of carbon
signal intensities on several experimental parameters,
such as the delay A or the pulse angle of the last
selective proton pulse in the DEPT part of the sequence,
using camphor (Fig. 1) as a readily available, rigid
model compound. These determinations have been
most conveniently carried out using the ID selective
DEPT sequence 1D-SDEPT.
Fig.l
The extent of polarization transfer is a function
of both, the interpulse delay A and the pulse angle $.
Fig. 2 shows the intensity of the quaternary C-l carbon
signal of camphor when pulsing the H-4 methine
proton, for 4>=45° and for =90°, as a function of
interpulse delay A, using sequence 1D-SDEPT. In this
case, the heteronuclear coupling constant involved is
3 JH4^,= 4.4 Hz. As expected, the quaternary carbon C-
1 intensity follows a typical sinus function, and maximum
signal is obtained for (j>=90° and A=50-90 ms, a range
which does not contain the theoretical value, (2* 3 JH4^
C1)'—115 ms. Instead, this range of maximum signal is
centered around the practical value (3* 3 JH4^1) l =75
ms.
1000-
800-
~ 600
f
£ 400
200
0
0.02
-i 1 « 1 1 1—r
0.04 0.06 0.08 0.1
Fig. 2
Delay (s)

Vol. 14, No. 1-4
However, when pulsing on a methyl proton the
optimum delay has to be modified. Fig. 3 shows the
variation of the quaternary O7 carbon signal of camphor
when pulsing the methyl H-10 protons as a function of
the pulse angle, for three A values, using also sequence
1D-SDEPT. In this case, the heteronuclear coupling
constant involved is 2 JH1(>C7=4.1 Hz (sign not determined).
If the optimization is carried out as in a conventional
DEPT, using A=(2J)S the intensities follow the typical
sin*cos 2

266
The results (Fig. 7) easily allow the assignment
of the four quaternary carbons coupled to the H-5
proton. The particular example shown was obtained
using sequence 2D-SDEPT1, but the alternative sequence
2D-SDEPT2 gave the same results. We also performed
a comparison between 2D-SDEPT1 and our previous
[4] sequence 2D-SINEPT. Both methods can yield the
desired long-range heteronuclear coupling constants
in less than one hour of accumulation (using a 400
M Hz spectrometer), with excellent sensitivity even for
quaternary carbons. However, the SDEPT method is
to be preferred, because it is less sensitive to mismatch
between the delay A and the long range coupling "J^.
b)
d)
6'7'
5'6 4' 87a' 7 4a 3a' 8a 5
6 8 7
'.58 128 1!!
Fig. 7
a) Decoupled "C spectrum; b) 1D-SDEPT spectrum after pulsing on
H-5. Only long-range coupled carbons with it are presents. Then, The
2D- J spectra (d) and its internal projection (c) obtained with the pulse
sequence of the scheme 2.
Acknowledgements.
Financial support from DGICYT through project
number PB89-0304 is gratefully acknowledged. We
also thank the Servei de Ressonancia Magnetica Nuclear,
UAB, for allocating instrument time to this project. A
grant (to T.P.) from Universitat Autonoma de Barcelona
is gratefully acknowledged.
References
Bulletin of Magnetic Resonance
1.- P.E. Hansen, Progress in NMR Spectroscopy, 1980,
14, 175.
2.- J.L. Marshall, "Carbon-Carbon and Carbon-Proton
NMR Couplings. Application to Organic
Stereochemistry and Conformational Analysis", VCH
Publishers, Deerfield Beach, Florida, 1982.
3.- A. Bax, J. Magn. Reson., 1984, 57,314.
4.- T. Parella, F. Sanchez-Ferrando and A. Virgili,
Magn. Reson. Chem., 1992, in press.
5.- A. Bax and R. Freeman, J. Am. Chem. Soc, 1982,
104, 1099.
6.- T. Jippo, O. Kamo and K. Nagayama, J. Magn.
Reson., 1986,66, 344.
7.- D. Uhrin, T. Liptaj, M. Hricovini and P. Capek, J.
Magn. Reson., 1989,85,137.
8.- L. Poppe and H. van Halbeek, J. Magn. Reson.,
1991,92,636.

Vol. 14, No. 1-4 267
Computer Simulations of High Resolution NMR Spectra
1 Introduction
Scott A. Smith, William E. Palke, and J. T. Gerig
Department of Chemistry, University of California
The evolution of a spin system during the application
of an RF pulse is a central aspect of high
resolution NMR spectroscopy. An appropriately
applied field can lead to saturation or decoupling
effects. When a field is used to maintain a spinlocked
condition, both scalar and dipolar interactions
come into play, leading to coherence transfers
and nuclear Overhauser effects that form the
basis of TOCSY, ROESY and related experiments.
These experiments have important roles
in structural studies of biological macromolecules
in solution . The accompanying bad
news is that the interpretation of these experiments
is tricky; computer simulations can be
very helpful in assisting the analysis of these
spectra.
The theoretical formalism of relaxation in
the presence of an RF field can be written elegantly
in superoperator notation , or in a more
conventional notation laden with superscripts
and subscripts. The latter approach displays
more details and shows how the presence of the
RF field introduces new combinations of frequencies
into the relaxation expressions. Of
course, it can be shown that both methods generate
identical results. Details of the theoretical
developments formulated in this lab are given in
reference 7 and in a manuscript in preparation.
We have used our theoretical results to
extend the program GAMMA 8 to include the
effects of relaxation in the presence of an RF
field and present here the results of several calculations
of double resonance or rotating frame
experiments that illustrate the capabilities
Santa Barbara, CA 93106, USA
thereby developed. For the examples presented,
only dipole-dipole relaxation was included, and
isotropic diffusional tumbling was assumed with
a correlation time of 1.0 nsec. In all cases, the
spectrometer proton frequency was 500 MHz.
2 Applications
Presaturation - The first example is a presaturation
experiment on a three spin system whose
parameters were chosen to mimic three protons
in the trans conformation of a glycine residue.
Details are given in Table 1.
Table 1: Spin System 1 Parameters
R12 = 1.75A
R13 = 3.00A
R23 = 2.53A
J12 = -16.3Hz
J13=4.0Hz
J23 = 12.0 Hz
v = -700 Hz
t> = -950Hz
v = 950 Hz
The pulse sequence consisted of a pre-irradiation
period followed by an analyzing 90° pulse, as
shown in Figure 1.
Pre-saturation
Figure 1. Pre-saturation Pulse Sequence
The initial pulse length was chosen to be long
enough to establish a steady state, and the RF frequency
chosen to match the chemical shift of
proton 2. The resulting spectra for several RF
field strengths are shown in Figure 2. Notice the

268 Bulletin of Magnetic Resonance
750
960 950 940
V3
-690-700-710 -940-950-960
V2
Figure 2. Simulation of Presaturation. Shading is
used to highlight NOE enhanced intensities.
Units on both axes are Hz.
non-symmetrical effects on spins 1 and 3.
Decoupling - The second example illustrates the
decoupling of a two spin proton system with a
chemical shift difference of 1000 Hz and a spin
coupling constant of 10 Hz. The protons were
2A apart. After an initial 90° pulse, an RF field
of a specified magnitude was applied during the
collection of the FID. The pulse sequence is
shown in Figure 3. Some spectra are shown in
Figure 4. Notice the slight oscillation in the peak
height as a function of the strength of the decoupling
field.
Figure 3. Decoupling Pulse Sequence
ROESY/TOCSY Simulations - Two-dimensional
rotating frame experiments may produce coherence
transfer (TOCSY) and spin-spin cross
relaxation (ROESY) effects simultaneously, and
1125 1100
v (Hertz)
1075
(Hertz)
Figure 4. Simulation of Decoupling. Spectra at the
first and last field strengths are shown to
the upper left.
ambiguities of interpretation arise when bothkinds
of effects are present. Griesinger et al.
have described a strategy for suppressing
ROESY effects when doing TOCSY experiments
9 . The interpretation of ROESY experiments
is complicated because coherence transfer
(COSY) and indirect interactions (HOHAHA)
can lead to spectral effects that interfere with or
imitate rotating frame Overhauser features 10 .
Methods for suppressing these unwanted effects
have been proposed , and it has been suggested
that use of a train of short RF pulses to generate
an effective spin-locking field leads to diminution
of COSY-type cross peaks .
Here we simulate several variations of
ROESY experiments. First, consider the pulse
sequence shown in Figure 5 in which a continuous
spin-locking pulse is applied.
JC/2 Spin Lock
Figure 5. CW ROESY Pulse Sequence
We apply this to an equilateral triangle of protons
as described in Table 2. The 2D spectra that

Vol. 14, No. 1-4 269
1
1
C-l. 0 sec
M
• 1 11
150 100 50 0 -50 -100 -150 150 100 50 0 -50 -100 -150
150 100 0 -50 -100 -150
Figure 6. 2D ROESY spectra for differing spin-lock
times.
result from several different spin-lock times are
shown in Figure 6. A point to notice here is that
the ROESY peaks grow more slowly than the
Table 2: Spin System 2 Parameters
Rl2=V3A
Ri3=V3A
R23=V3A
J12 = 0
Jl3 = 0
J23 = 4.0 Hz
v = 140 Hz
v = -90 Hz
?? = -140Hz
D -1.1 sec
150 100 50 0 -50 -100 -150
©
.©
-o
o
• o
©
1 V)
-o
COSY peaks, reaching their maximum intensity
well after the latter have begun to subside.The-
ROESY peaks also persist much longer and
decay more smoothly than both the COSY and
diagonal peaks for protons 2 and 3. The plots for
1.0 and 1.1 sec. spin-lock times show oscillation
in both the latter features while the ROESY
peaks decay smoothly.
Interesting contrasts to these results arise in
changing to an isosceles triangle configuration.
The spin system is given in Table 3. In this
geometry, the ratio of the 1-3 distance to the 1-2
©

270 Bulletin of Magnetic Resonance
distance is \2. Thus, the direct 1-2 dipole-dipole
interaction is 8 times stronger than the corresponding
1-3 interaction. Keeping a fixed
Table 3: Spin System 3 Parameters
Ri2=V3A
Ri3=V6A
R23=V3A
J12 = 0
J13 = 0
J23 = 0 -10 Hz
v = 140 Hz
v = -90 Hz
v = -U0Uz
geometry and a spin-lock time of 0.1 sec, the 2-
3 spin-spin coupling is varied from 0 to 10 Hz as
we repeat the pulse sequence of Figure 5. Slices
through the resulting 2D spectra at 140 Hz are
shown in Figure 7.
J = 10 Hz
J = 8Hz
= 6Hz
J = 4Hz
J = 2Hz
= 0Hz
J)LJL_
150 100 50 0 -50 -100 -150
Figure 7. The effect of J23 variation on ROESY/
TOCSY experiment.
Indeed, the simulation with J23 = 0 shows an 8:1
ratio of NOE peak intensities. However, variation
of J23 generates a ROESY-like peak in the
location of the 1-3 interaction that becomes even
larger than the true 1-2 ROESY peak for some
values of J23- This situation provides an excellent
example of a spectrum that cannot directly
yield reliable information about internuclear distances.
In the next example, the frequency of the
spin-locking RF field is varied. In this simulation
spin system of Table 3 is used with J23 = 4 Hz. It
is subjected to a 1.0 sec. spin-locking pulse as
depicted in the ROESY pulse sequence in Figure
5. The frequency of the RF field is indicated by
the arrow on each of the stacked plots in Figure
8. While the 1-2 ROESY peak is relatively unaffected
by the frequency of the RF field, the 1-3
peaks which are generated by indirect
HOHAHA effects show extreme sensitivity both
in magnitude and sign. As has been suggested in
the literature, several experiments should be run
with different RF frequencies to discriminate
between these interactions.
\
1
150 100 50 0 -50 -100 -150
Figure 8. The effect of varying the applied RF frequency
on ROESY/TOCSY experiment.
i
1 1
As a final demonstration, we investigate the
suggestion that dividing the spin-locking period
into a sequence of numerous pulse-delay steps
can emphasize the ROESY interactions. Again,
the three spin proton system in Table 3 is used
with J23 = 4 Hz. It was subjected to the ROESY
pulse sequence utilizing a pulse train as shown
in Figure 9. The length of the pulse and delay
steps were varied keeping the average yBj con-
I

Vol. 14, No. 1-4 271
Figure 9. ROESY Pulse Sequence with pulse train
150 100 50 0 -50 -100 -150
Figure 10. Simulation of ROESY/TOCSY experiment
using a spin-locking pulse train.
o
.©
o
-o
150 100 50 0 -50 -100 -150
Figure 11. Simulation of ROESY/TOCSY using continuous
irradiation for spin-locking.
o
•o
stant at 2000 Hz by adjusting the number of
pulses per second. This simulation has been performed
with pulse lengths that correspond to
individual rotations of 180°, 30°, and 10°, and
also with CW irradiation. The contour plot using
30° pulses is presented in Figure 10. For comparison,
the plot which was simulated using continuous
irradiation during the spin lock is shown
in Figure 11. Cross sections of these contour
plots taken at -90 Hz are shown in Figure 12.
= 10deg
= 180deg
150 100 50 0 -50 -100 -150
Figure 12. Cross sections of ROESY/TOCSY simulations
using a spin-locking pulse train.
Shortening the pulse length indeed suppress the
COSY peaks, but they remain even for the shortest
pulses. Also, the limit of the sequence of
shorter and more frequent pulses gives the same
result as continuous irradiation. These two processes
appear to become the same once the delay
between pulses is sufficiently short that a negli-

272 Bulletin of Magnetic Resonance
gible precession occurs before the next pulse.
3 Conclusions
It is clear that full computer simulation may be
essential for the correct interpretation of ROESY
experiments. Our initial results indicate that
there may be additional information about structure
and dynamics that can be extracted from
classical one-dimensional multiple irradiation
experiments with the aid of simulations. An
appreciable expansion of this program in terms
of the size of spin systems that can be handled is
probably needed to make such applications practical;
this along with other expansions of the program's
capabilities is planned.
4 References
1 - L. Braunschweiler and R. R. Ernst, J. Magn.
Reson. 53, 521 (1983).
2 - A. Bax and D. G. Davis, /. Magn. Reson.
63,207 (1985).
3 - A. A. Bothner-By and R. Shukla, /. Magn.
Reson. 77, 524 (1988).
4 - M. Ranee and J. Cavanaugh, J. Magn.
Reson. 87, 363 (1990).
5 - S. J. Glaser and G. P. Drobny, Advan. Magn.
Reson. 14, 35 (1990).
6 - R. R. Ernst, G. Bodenhausen, and A. Wokaun,
"Principles of Nuclear Magnetic Resonance
in One and Two Dimensions",
Clarendon Press, Oxford (1987); J. Jeneer,
Advan. Magn. Reson. 10, 2 (1982); M.
Ravikumar, R. Shukla, and A. A. Bothner-
By, J. Chem, Phys. 95, 3092 (1991).
7 - S. A. Smith, W. E. Palke, and J. T. Gerig, /.
Magn. Reson. in press (1992).
8- S.A. Smith, T. Levante, B.H. Meier, and
R.R. Ernst, manuscript in preparation.
9 - C. Griesinger, G. Otting, K. Wiithrich, and
R.R. Ernst, J. Am. Chem Soc. 110, 7870
(1988).
10-D. Neuhaus and M. P. Williamson, "The
Nuclear Overhauser Effect in Structural and
Conformational Analysis". VCH, New York
(1989) p. 312-327.
11 - H. Kessler, C. Griesinger, R. Kessebaum, K.
Wagner, and R.R. Ernst, J. Am. Chem Soc.
109,607-609(1987).

Vol. 14, No. 1-4 273
1. Introduction
Variation of 13 C NMR Linewidths of
Metallocenes as a Function of Magic
Angle Sample Spinning Frequency
In this paper we report high-resolution solid state 13 C
NMR investigations of various metallocenes
[(Ti5-C5H5)M(Ti5-C5H5); M = Fe, Ru, Ni], carried out as
a function of magic angle sample spinning (MAS)
frequency. Specifically, we focus on how the linewidth of
the isotropic peak varies with MAS frequency at fixed
temperature. Unexpectedly, it has been found that the
linewidth increases as the MAS frequency is increased, and
it is demonstrated that the underlying reason is that the *H
decoupling becomes less efficient as the MAS frequency is
increased. It is suggested that molecular motion within
these solids is an important factor underlying this
phenomenon.
It is well known that, at room temperature, there is
substantial molecular motion in crystalline metallocenes
[1-3]. In ferrocene, for example, it has been shown [4-6]
that there is rapid reorientation of the cyclopentadienyl
(C5H5) rings via a five-fold jump mechanism, with
correlation time xc = 5 X 10~ 12 s at 293 K. The
correlation time for ring reorientation in nickelocene at
room temperature is essentially the same as that for
ferrocene, whereas the ring reorientation in ruthenocene is
associated with a significantly longer correlation time (tc ~
5 X 10- 10 s at 293 K) [4,5]. It has been suggested [2]
that there may be some additional slower molecular
motions in crystalline ferrocene at ambient temperature.
All experiments reported in this paper were carried out at
constant temperature, and hence the correlation time for
molecular motion for a given metallocene can be assumed
to be constant for the series of experiments discussed here.
Before presenting our results, we discuss briefly relevant
aspects of magic angle sample spinning (MAS) and high
power *H decoupling in relation to the measurement of
high-resolution 13 C NMR spectra for organic solids, with
Ian J. Shannon, Kenneth D.M. Harris*, s* S. Arumugam
Department of Chemistry
University of St. Andrews
St. Andrews
Fife KY16 9ST
Scotland
particular emphasis on the application of the technique to
systems in which there is substantial molecular motion.
2. Theory
* Author to whom all correspondence should be addressed.
In solid state 13 C NMR of organic materials, the two
major sources of line-broadening are chemical shift
anisotropy (CSA) and direct ^C- 1 !! dipole-dipole
interaction. Magic angle sample spinning (MAS) [7,8]
will transform an NMR line broadened by these effects into
a set of comparatively narrow, equally-spaced lines
comprising an isotropic peak and spinning sidebands. The
spacing between adjacent lines in this set is equal to the
MAS frequency vr. The chemical shift of the isotropic
peak is independent of vr, and only this line remains when
the anisotropic interactions have been averaged completely
(i.e. at sufficiently large vr).
As discussed fully elsewhere [7-9], there is an important
difference concerning the way in which CSA and direct
dipole-dipole interaction are affected by MAS. For an
NMR line broadened by CSA, relatively slow MAS is
generally sufficient to transform this line into the set of
narrow, equally-spaced lines discussed above, whereas a
line broadened by direct dipole-dipole interaction will be
narrowed significantly only if vr is in the region of (or
greater than) the magnitude of the dipole-dipole interaction.
The underlying reason for this difference [9] is that CSA
gives rise to inhomogeneous broadening of the spectral
line, whereas dipole-dipole interaction is a source of
homogeneous broadening; the value of vr required to
achieve effective line-narrowing is larger (relative to the
linewidth of the static sample) in the case of homogeneous
broadening. For most organic solids with natural isotopic
abundance, the only appreciable dipole-dipole interaction
directly affecting the 13 C NMR spectrum is that between

274
13 C and 1 H. Since the magnitude of this interaction
(typically ca. 30 kHz for rigid organic solids) is generally
much larger than the MAS frequencies that can be obtained
on conventional instruments, substantial averaging of this
interaction cannot be achieved using MAS. For this
reason, high power *H decoupling is generally applied (in
addition to MAS) during acquisition of the 13 C spectrum
in order to eliminate line-broadening due to direct ^C-^H
dipole-dipole interaction.
In 13 C NMR spectroscopy of many crystalline
metallocenes (such as ferrocene and ruthenocene), GSA and
direct ^C- l U dipole-dipole interaction are the important
sources of line-broadening.
A detailed publication [10] has considered, from both
theoretical and experimental standpoints, the linewidth of
the isotropic peak, recorded under MAS conditions, for a
system that is subject to line-broadening by CSA. It was
also shown that, at fixed temperature, the linewidth of the
isotropic peak decreases as the' MAS frequency (vr) is
increased, approaching a limiting value at sufficiently large
vr. Another paper [11] has considered the linewidth of a
spin system S, dipolar coupled to an unlike spin system /,
under conditions of isotropic molecular motion and
decoupling of the / spins. Using coi to denote the
decoupler field strength, it was shown that, in the limit of
long correlation time (i.e. (0itc » 1) the linewidth is
proportional to (o>i)~ 2 , whereas in the limit of short
correlation time (i.e. coixc « 1) the linewidth is
independent of ca i. Thus, for a sample at fixed
temperature (and hence fixed xc), the linewidth should
either decrease as the decoupler field strength is increased
(long correlation limit) or remain independent of the
decoupler field strength (short correlation limit). The
effects of anisotropic molecular motion on the measured
spectrum were also discussed briefly in ref. 11.
In the studies discussed in this paper, both MAS
frequency and *H decoupler field strength are important in
controlling the linewidth of the isotropic peak in the 13 C
NMR spectrum. If the separate effects discussed above can
be combined in a simple way, then it should be expected
that: (a) at fixed temperature and fixed decoupler field
strength, the linewidth should decrease with increasing
MAS frequency (up to a limiting value, beyond which the
linewidth should be effectively independent of the MAS
frequency); and (b) at fixed temperature and fixed MAS
frequency, the linewidth should either decrease or remain
constant as the decoupler field strength is increased,
depending on the motional regime (i.e. long or short
correlation limit) of the sample at the temperature of
interest.
It is shown here that, for ferrocene and ruthenocene at
room temperature, the effects of MAS frequency and *H
decoupler field strength on the 13 C NMR linewidth cannot
be combined in this simple way, since the effective
decoupler field strength is modulated by altering the MAS
frequency. Nickelocene is paramagnetic, and for this
Bulletin of Magnetic Resonance
reason it is not valid to consider nickelocene in the same
way as ferrocene and ruthenocene in relation to the NMR
properties discussed here.
3. Experimental
13 C NMR spectra were recorded at 125.758 MHz on a
Bruker MSL500 spectrometer using a Bruker doublebearing
magic angle spinning probe capable of MAS
frequencies between ca. 1 kHz and 12 kHz with stability
better than ca. ± 10 Hz. All spectra were recorded at room
temperature (293 ± 2 K) with the samples contained in
zirconia rotors (4 mm external diameter).
The 13 C "single pulse" sequence was used to record the
spectra, with high power *H decoupling applied during
acquisition. Typical parameters were: 13 C 90° pulse
length = 3.5 \xs; recycle delay = 10 s for ferrocene and 20 s
for ruthenocene.
The *H decoupler field was set on resonance for the *H
of each metallocene. An accurate assessment of the
decoupler field strength was made by measuring (for
adamantane) the length of the *H 90° pulse (t9o( 1 H)) for
the *H r.f. power level used in the experiments involving
*H decoupling. For the discussion of results, tgo^H) has
been converted to a decoupling frequency vi via:
1
vi is related to the decoupler field strength Hi and to the
parameter coi used in ref. 11 by the equations:
_ Y( 1 H) Hi _ ©!_
L 71 Z %
The linewidth of the isotropic peak in the 13 C NMR
spectrum was measured as the full width at half maximum
height, and the experimental error in the measured
linewidth is estimated to be less than ca. ± 5 Hz.
4. Results and Discussion
Initially we present the main results for ferrocene [12], and
then consider the other metallocenes studied. Highresolution
13 C NMR spectra of ferrocene were recorded
initially at two different decoupler field strengths (vi =
64.9 kHz and 26.6 kHz), and at several MAS frequencies
(vr) ranging from ca. 1 kHz to 11 kHz. The spectrum at
vr

Vol. 14, No. 1-4
increases as vr is increased. The relationship between A
and vr is approximately linear, particularly at the higher
decoupler field strength, and the gradient is greater at the
lower decoupler field strength. At fixed vr, the linewidth
A is smaller at higher decoupler field strength, as shown in
Fig. 1.
A/Hz
400
300
a n
vr / kHz
Fig. 1 Linewidth (A) versus MAS frequency (vr) for the
isotropic peak in the 13 C NMR spectrum of ferrocene.
The spectra were recorded at 'H decoupler field
strengths corresponding to Vi = 64.9 kHz (•) and Vj =
26.6 kHz (A).
Our observation that the linewidth of the isotropic peak
increases as the MAS frequency is increased conflicts with
the discussion in Section 2 that, under conventional
conditions, A should decrease, or remain constant, as vr is
increased. We propose that the observed increase in A
with increasing vr for ferrocene arises as a result of MAS
indirectly modulating the efficiency of the *H decoupling
in such a way that the effective decoupler field strength is
decreased, leading to line-broadening, as vr is increased.
This conclusion is supported by the results of three further
experiments.
(1) l^C NMR spectra of ferrocene were recorded for a
series of different decoupler field strengths (with vi in the
range 20 kHz to 80 kHz) at fixed vr. As expected, A
decreases as vi is increased at fixed vr (Fig. 2).
Furthermore, in the limit of sufficiently high decoupler
field strength, A becomes essentially independent of both
vr and vi, and converges to a limiting value of ca. 100 Hz.
From this it can be concluded that both values of vr (5.06
kHz and 9.05 kHz) used to record the data shown in Fig. 2
are sufficiently rapid to remove any Vr-dependent sources
of line-broadening due to CSA; i.e. at sufficiently high
decoupler field strength, A is essentially independent of vr
at these values of vr. This is consistent with the view
that, under the conditions of the experiments shown in Fig.
1, the property that does depend on vr is the efficiency of
the *H decoupling (and not the ability of MAS to remove
275
the line-broadening effects due to CSA). As discussed
fully elsewhere [12], there is a linear relationship between
A and (vi)~ 2 .
5001
400
A/Hz 300-
200
o o
20 SO 70 80
/ kHz
Fig. 2 Linewidth (A) versus Vj for the isotropic peak in the
13 C NMR spectrum of ferrocene. The spectra were
recorded at MAS frequencies vr = 5.06 kHz (O) and vr =
9.05 kHz (•).
(2)
13 C NMR spectra of ferrocene were recorded as a
function of vr, but with no *H decoupling field applied;
the variation of A with vr in these experiments is shown
in Fig. 3. Under these conditions, A decreases as vr is
increased, as predicted from the discussion in Section 1 and
from ref. 10. This relationship between A and vr reflects
the direct effect of MAS on the linewidth of the isotropic
peak for a system that is subject to line-broadening by
CSA and by direct 13 C-*H dipole-dipole interaction. At
the lowest value of vr studied (ca. 3 kHz), A is ca. 592 Hz,
and it is clear that slow MAS alone can substantially
average the dipole-dipole interaction (as well as CSA) in
A/Hz
6001
400"
300
vr / kHz
Fig. 3 Linewidth (A) versus MAS frequency (vr) for the
isotropic peak in the 13 C NMR spectrum of ferrocene,
with no 'H decoupling applied.
12

276
this system. Undoubtedly, this is a consequence of the
fact that the direct ^C- 1 !! dipole-dipole interaction is
already extensively averaged by molecular motion.
(3)
13 C NMR spectra were recorded (using the 13 C
"single pulse" method with no *H decoupling field applied)
for perdeuterated ferrocene (denoted ferrocene-dio) as a
function of MAS frequency. Over the range vr« 1 kHz to
12 kHz, the linewidth of the isotropic peak is independent
of vr (Fig. 4), and the value (A = 117 Hz) is close to the
limiting linewidth obtained in our experiments for
undeuterated ferrocene. In 13 C NMR spectroscopy of
ferrocene-dio, the principal source of line-broadening is
CSA and, furthermore, the CSA should be substantially
the same in ferrocene-dio and in undeuterated ferrocene.
The fact that A is essentially independent of vr for
ferrocene-dio thus strongly supports the view that the
increase of A with vr for undeuterated ferrocene is not
arising from CSA being modulated by MAS.
A / Hz
200l
ISO-
100
50
vr / kHz
10 12
Fig. 4 Linewidth (A) versus MAS frequency (vr) for the
isotropic peak in the 13 C NMR spectrum of
ferrocene-djo, with no J H decoupling applied.
High-resolution 13 C NMR spectra of ruthenocene and
nickelocene were also recorded as a function of MAS
frequency in order to establish whether the behaviour
observed for ferrocene is also exhibited by other
structurally-related systems. In Fig. 5, the relationships
between A and vr for ruthenocene and ferrocene are
compared, with all spectra recorded at the same *H
decoupler field strength (vi = 64.9 kHz).
It is clear that ruthenocene exhibits the same general
trend as ferrocene, with A increasing as vr is increased,
although the relationship for ruthenocene is apparently less
linear than that for ferrocene. At the higher values of vr
studied, the gradient 3A/3vr is larger for ruthenocene.
These small differences in behaviour between ferrocene and
ruthenocene presumably reflect small differences in the
Bulletin of Magnetic Resonance
dynamic properties of these solids at 293 K (as suggested
in ref. 5).
2501
200"
A /Hz lso-
50-
c ! 1
vr / kHz
10 12 14
Fig. 5 Linewidth (A) versus MAS frequency (yr) for the
isotropic peak in the 13 C NMR spectra of ferrocene (•)
and ruthenocene (A), at fixed *H decoupler field
strength corresponding to Vj = 64.9 kHz.
The linewidth in the 13 C NMR spectrum of nickelocene
is substantially greater (A = 15.7 kHz at vr = 5 kHz) than
that in spectra recorded for ferrocene and ruthenocene under
the same conditions. The linewidth for nickelocene is not
significantly affected by increasing the MAS frequency,
although there is a measurable decrease to A = 15.4 kHz at
vr = 11 kHz. The very large 13 C NMR linewidth
observed for nickelocene, even under conditions of MAS
and high power *H decoupling, is due to the paramagnetic
properties of this molecule (which contains two unpaired
electrons). For this reason, it is not expected that the
NMR properties of nickelocene will be comparable, in any
way, to those of ferrocene and ruthenocene.
5. Conclusions
The increase in linewidth of the isotropic peak in the 13 C
NMR spectra of ferrocene and ruthenocene as the MAS
frequency is increased (at fixed temperature and fixed *H
decoupler field strength) is due to an indirect effect in which
the effective *H decoupler field strength is decreased as vr
is increased. Considering the results of high-resolution
13 C NMR experiments for several crystalline organic
solids carried out in our laboratory, it is clear that, for
systems in which there is no appreciable molecular
motion, A is essentially independent of vr. This fact
tends to suggest that the molecular motions present within
crystalline ferrocene and ruthenocene are important in
relation to the observed increase in the linewidth of the
isotropic peak with increasing MAS frequency.

Vol. 14, No. 1-4
A similar anomalous relationship between isotropic 13 C
NMR linewidth and temperature has been observed recently
by Muller [13] in studies of thiourea inclusion compounds;
in this case, linewidths for 13 C environments in the guest
molecules (which undergo substantial molecular motion)
have been observed to increase with increasing temperature,
and again this effect has been attributed to an "interference"
between the molecular motion and the efficiency of the *H
decoupling.
In view of the fact (see Section 1) that cyclopentadienyl
ring reorientation occurs on a timescale of the order of ca.
10~ 12 s for ferrocene and ca. 10~ 10 s for ruthenocene at
room temperature, it is not clear whether ring reorientation
is indeed the dynamic process responsible for influencing
the value of A in our experiments; it might be expected
that a motion occurring at a frequency comparable to vi
and/or vr (and thus in the approximate frequency range 10 3
Hz to 10 6 Hz) would be required. At present, we make no
attempt to assign the dynamic process that is important in
giving rise to the observed NMR phenomena for ferrocene
and ruthenocene at room temperature. However, it may be
that the slower motion in ferrocene alluded to in ref. 2 is
influential in this regard. Future experiments, at different
temperatures, will investigate the variation of A with vr at
different values of the correlation time for this motion, and
other experimental investigations of the dynamic properties
of crystalline metallocenes are in progress. We are also
currently investigating various theoretical implications of
the results reported here.
Finally, it is relevant to strike a cautionary note in regard
to recording high-resolution solid state 13 C NMR spectra
for systems that are subject to line-broadening by CSA and
direct ^C-*H dipole-dipole interactions. In view of the
results reported here, it is clear that there are circumstances
under which optimum resolution can be obtained by
recording the spectrum at high *H decoupler field strength
and low MAS frequency, rather than at high *H decoupler
field strength and high MAS frequency.
Acknowledgements
277
We are grateful to the S.E.R.C. (studentship to US) and
the Nuffield Foundation (research grant to KDMH) for
financial support.
References
[I] D. Braga, Chem. Rev., 92 (1992) 633.
[2] C.H. Holm and J.A. Ibers, /. Chem. Phys., 30
(1959) 885.
[3] B.T.M. Willis, Ada Cryst., 13 (1960) 1088.
[4] A.B. Gardner, J. Howard, T.C.Waddington, R.M.
Richardson and J. Tomkinson, Chem. Phys., 57
(1981) 453.
[5] A.J. Campbell, CA. Fyfe, D. Harold-Smith and
K.R. Jeffrey, Mol. Cryst. Liq. Cryst., 36 (1976) 1.
[6] A. Kubo, R. Ikeda and D. Nakamura, /. Chem. Soc,
Faraday Trans. 2, 82 (1986) 1543.
[7] E.R. Andrew, Int. Rev. Phys. Chem., 1 (1981) 195.
[8] E.R. Andrew, Phil. Trans. Roy. Soc. (Lond.) A, 299
(1981)505.
[9] M.M. Maricq and J.S. Waugh, /. Chem. Phys., 70
(1979) 3300.
[10] D. Suwelack, W.P. Rothwell and J.S. Waugh,
/. Chem. Phys., 73 (1980) 2559.
[II] W.P. Rothwell and J.S. Waugh, /. Chem. Phys., 74
(1981)2721.
[12] I.J. Shannon, K.D.M. Harris and S. Arumugam,
Chem. Phys. Lett., in press.
[13] K. Muller,/. Phys. Chem., 96(1992) 5733.

278
1 Introduction
The structural role of water in silicate glasses:
1H and 29si NMR evidence
J. Kiimmerlen, T. Schaller, A. Sebald and H. Keppler
Bulletin of Magnetic Resonance
Bayerisches Geoinstitut, Universitat Bayreuth, Postfach 101251
W-8580 Bayreuth, Germany
Over the last years various spectroscopic
methods [1-6] have been used to
investigate hydrous silicate and aluminosilicate
glasses and to clarify the structural
role of water in such systems. Especially,
the existence of molecular water and/or
Si-OH/Al-OH species, i.e. the H2O-induced
depolymerisation of the Si/Al or Si
network in such glasses has been discussed
controversially in the literature.
We have chosen two hydrous Na2Si4O9
glasses with different H2O-contents as an
Al-free model system.
The Al-free Na2Si4O9 system is particularly
well suited for high-resolution
solid-state NMR investigations as the
various Q-species can easily be resolved in
the respective 29 Si MAS and CP/MAS
spectra.
Various high resolution solid state NMR
techniques were used: 2!? Si singie pulse
MAS, *H -> 29 Si-CP/MAS and *H-CRAMPS.
Only the combined use of all these methods
provides an insight into the interactions
between the Si-network and the proton
system. Additional modifications of these
standard techniques were then used to
confirm and to refine the picture of the
Na2Si4O

Vol. 14, No. 1-4 279
selectively and to obtain separate
information about the dipolar protoninteractions
("dipolar dephasing").
29 Si dipolar dephasing
10% H2O
I • .I . . . I . . . I . . .I...I
-70 -90 -110 ppm
dephasing time
1 ms
5 ms
10 ms
Fig. 1: 29 Si-CP/MAS dipolar dephasing
spectra of Na2Si4O9 glass with
9.1% H2O. The glass with the
lower water content shows
similar decays of the signal
components.
3 Results and Discussion
The 29 Si-MAS and 29 Si-CP/MAS spectra
of the hydrous glasses show the presence
°f Q4» Q3 and Q2 silicon species. The higher
water content corresponds to a higher
relative intensity of the Q2 signal and to a
lower relative intensity of the Q4 signal,
respectively. Furthermore, the 29 Si-
CP/MAS experiments show significant
differences in the time dependence of the
29 Si magnetization of the two glasses. For
the glass with the higher water content
the signal intensity as a function of
contact time is describable by a
biexponential curve according to the usual
thermodynamic model used to describe I-S
CP-dynamics. The cross polarization times
Tis and the relaxation times Tip show a
trend for (Q4) > (Q3) >(Q2). Both Tis and Tip
are significantly longer for the glass with
the lower water content. The 29 Si CP-data
of the Q3 and Q4 species for the lower H2Ocontent
material can only be described by
double-biexponential magnetization
curves. This leads to the following
conclusions: i) in the glass with the
higher water content a uniform *H spin
lattice is established, ii) due to the dilution
of the protons in the other glass at least
two ^H spin baths are present and iii) in
the lower H2O-content glass the average
interatomic distances between 29 Si and *H
for the two different reservoirs must be
significantly different. In consequence,
one can assume the presence of both Si-OH
groups and molecular water in these
hydrous glasses. In fact, quantitative
determination of the Q2, Q3 and Q4 species
in both hydrous glasses from 29 Si MAS
spectra and comparison of these results
with the respective stoichiometric
requirements if fully in accord with the
depolymerisation of the silicate network.
Results of the 29 Si-CP/MAS experiment
with interrupted decoupling ("dipolar
dephasing") are illustrated in Fig. 1. The
decay of the 29 Si magnetization is mainly
determined by the strength of the dipolar
interaction between the 29 Si nuclei and
the nearest surrounding proton system.
Obviously, for the loss of magnetization
characterized by a decay time T2 the
relation T2(Q2)

280
'H dipolar dtphasing
dtphasing tinw
»%H20
20 10 0 ppm 20 10 0 ppm
Fig. 2: ^-dipolar dephasing CRAMPS
spectra of both glasses (450
scans). (The spectra on the left
hand side show an experimental
artefact at ca. 18ppm)
To confirm these CP results various J H
multiple pulse experiments were used. The
^H-CRAMPS spectra clearly show two well
resolved signals at ca. 4.7 ppm and 12.2
ppm (with respect to TMS = 0.0 ppm). The
chemical shifts and the results of the CP
experiments allow tentative assignment of
these signals as molecular water (4.7 ppm)
and protons of Si-OH groups (12.2 ppm).
This assignment was confirmed by the *H
"dipolar dephasing" CRAMPS experiment
(jc/2-T-rc-T-MREV8). As illustrated in Fig. 2,
in both glasses the signal at 4.7 ppm decays
much faster than the less shielded signal,
i.e., the dipolar * H - 1 H interaction is
significantly stronger. This experimental
fact corroborates the model proposed
above. Comparing the decay rates found
for the two different glasses, differences
in these rates were found (Fig. 3). Again,
these differences are in good agreement
with the results of the standard 29 s i-
CP/MAS experiments.
. »0-
3
C
v
100
>*, 90
V)
C
Fig. 3:
Bulletin of Magnetic Resonance
a water protons
A Si-OH protons
a D
5 10 15 20 25
dephasing time (/is)
Signal intensities in the * H
dipolar dephasing CRAMPS
experiment for the
glass with
a) 4.9 % H2O (top)
b) 9.1 % H2O (below)
Additionally, the longitudinal relaxation
times Ti of both *H signal components in
both glasses were measured. The use of a
JC-T-JC/2-T-MREV8 pulse sequence provides
a uniform Ti of 0.70 ± 0.05 s for the glass
with the higher water content. In contrast,
the signal components of the other
glass decay with Ti(SiOH) = 0.85 ± 0.05 s and
Ti(H2O) = 0.97 ± 0.05 s, respectively. These
data also confirm the interpretation of the
CP experiments and provide an additional
support for the following conclusions:

Vol. 14, No. 1-4 " 281
(1)H2O does depolymerize the silicate
network,
(2) both Si-OH and molecular water are
present, and
(3) in the lower H2O-content glass the
more dilute proton system, forming
these components is best described as
"separate" sub-systems.
To answer further questions concerning
the interactions between Si-OH and
H2O protons and to address questions
concerning the structural role of cations
like e.g. Na + in such silicate networks, ID
and 2D-spin diffusion experiments are in
preparation.
Acknowledgment: Support of this work
by the Deutsche Forschungsgemeinschaft
and the Alexander von Humboldt-
Foundation is gratefully acknowledged.
Literatur:
[I] Bartholomew, R.F., Butler, B.L., Hoover,
H.L., Wu, C.K.J., J. Am. Ceram. Soc. 1980,
63, 481
[2] Kohn, S.C., Dupree, R., Geochim.
Cosmochim. Acta 1987, 53, 2925
[3] McMillan, P., Holloway, J.R., Contrib.
Mineral. Petrol. 1987, 97, 320
[4] Mysen, B.O. Virgo, D., Chem. Geol. 1986,
64, 2623
[5] Stolper, E.M., Am. Mineral. 1989, 74,
1247
[6] Eckert, H., Yesinowski, J.P., Silver,
L.A., Stolper, E.M., J. Phys. Chem. 1988,
92, 2055
[7] Kiimmerlen, J., Merwin, L.H., Sebald,
A., Keppler, H., J. Phys. Chem. (in
press)
[8] Oppella, S.J., Frey, M.H., J. Am. Chem.
Soc. 1979, 101, 5854
[9] Bodenhausen, G., Stark, R.E., Ruben,
D.J., Griffin, R.G., Phys. Lett. 1979, 67,
424
[10] Rhim, W.K., Elleman, D.D., Vaughan,
R.W., j. Chem. Phys. 1973, 59, 3740
[II] Bronnimann, C.E., Zeigler, R.C.,
Maciel, G.E., J. Am. Chem. Soc. 1988,
110, 2023

282
High-Resolution Solid-State NMR Study
of Microstructures in Layered Aluminosilicate
Bulletin of Magnetic Resonance
Shigenobu Hayashi, Takahiro Ueda, Kikuko Hayamizu, and Etsuo Akiba
1. INTRODUCTION
National Chemical Laboratory for Industry,
Tsukuba, Ibaraki 305, Japan
Kaolin ite, Al4Si4Qo(OH)8> is a layered
aluminosilicate with a dioctah edr al 1:1 layer
structure consisting of an octahedral aluminum
hydroxide sheet and a tetrahedral silica sheet.
Figure 1 shows the structure of kaolin ite. Ihe crystal
structure is not fully understood because of the
absence of a large single crystal.
In the present paper, we have traced 2 ^Si, 27 A1,
and !H NMR spectra of various kaolinites, using
high-resolution solid-state techniques. Analyzing
the spectra theoretically, correlations between the
NMR data and the local structures are discussed
quantitatively.
2. EXPERIMENTAL
Totally eight samples were used. Six samples
were natural, which were Kanpaku kaolin (called
Nl; Hinckley index 1.4), API No.9 standard kaolin ite
specimen (N2; 1.4), Georgia kaolin (N3; 0.7), Hakone
Cbwakudani kaolin (N4; 0.4), Kibushi clay (N5; 0.2),
and Gairome clay (N6; 0.2). Two synthetic samples
were used,wh ich were synth esized at 290°C(51; 0.9)
and220°C(52;0.8).
NMR spectra were traced at room temperature by
aBruker MSL400 (a static m agn etic field of 9.4 T) and
a JBXGSH200 (4.71). Th e lin e sh apes of th e spectra
were analyzed using computer programs written by
ourselves.
3. RESULTS AND DISCUSSION
3.1. 2 ^Si spectra
Figur e 2 sh ows 29 Si O7 MAS NMR spectr a of th e
sample Nl, which has the highest crystallinity and
the lowest content of paramagnetic impurities
Fig. 1. Projection of the structure of kaolin ite from
the(100) direction.
among the eight samples studied. The spectra have
two signals at -90.8 and -91.4 ppm from
tetramethylsilane, being ascribed to Q'(GAl). The
line shapes do not depend on the contact time, and
the two peaks have the same cross relaxation time
between ^H and 29 Si, which is 2.0 HIS. Maximum
intensities are obtained at the contact time of 8 ms.
The field dependence experiments demonstrate
clearly th at two in equivalent Si sites are present.
The linewidth in the 29 Si CP/MAS spectra is
originated from the dipole-dipole interaction with
27 Al,the chemical shift dispersion due to structural
disorders, and the anisotropic magnetic
susceptibility due to paramagnetic impurities.
For Kanpaku kaolin,the chemical shift dispersion
is 0.39 ppm, while the contributions of the dipolar
interaction are0.08and033ppm for the fields of 9.4
and4.7T,respectively,being estimatedfrom th e field

Vol. 14, No. 1-4 283
B
-85 -90 -95
ppm
79.496MHz 104.263MHz
39.683MHz
Fig. 2. 29 Si O7 MAS NMR spectra of th e sample JV1,
measured at (A) 79.496 MHz and (B) 39.683
MHz. Chemical shifts are expressed with
r esp ect to tetr am eth y lsilan e.
dependence experiments. The effect of
paramagnetic impurities is negligible. The
contribution of the dipole-dipole interaction with
27 Al spin sis calculated theoretically from the crystal
structure. The estimated linewidths at 9.4Tare 0.14
and 0.09 ppm for Si(l) and Si(2), respectively,
whereas those at 4.7 Tare 055 and 0.35 ppm. These
values are in excellent agreement with the values
estimated experimentally.
Qher kaolinites, with lower crystallinities and/or
higher contents of paramagnetic impurities, have
broader linewidths due to the structural disorders
an d th e par am agn etic impurities.
3.2. 27 Al spectra
Figur e 3 sh ows 27 A1 DE^ MAS NMR spectr a of th e
sampleiv"l,measuredatdifferent fields. The Al atom
in the kaolinite structure is coordinated by six
oxygen atoms,and they give a signal around 0 ppm
with respect to 1M A1(NQ)3 aqueous solution. The
spectrum at the lower field gives the broader signal
atthelower frequency position,which suggests that
the signal is the central transition, being broadened
by the second-order quadrupole interaction. The
B
52.051MHz
100 50 0 -50 -100
ppm
Fig. 3. 27 A1 Hy MAS NMR spectra of the sample M,
measured at (A) 104.263 MHz and (B) 52.051
MHz. Chemical shifts are expressed with
r espect to 1 M A1(NC$)3 aqueous solution.
line shapes are simulated by our computer
programs. The observed spectra cannot be
simulated by one component, but can be simulated
much better by two components with equal
intensities. The obtained parameters for Al(l) are a
chemical shift of 7.8 ppm, a quadrupole coupling
constant of 3.36 MHz, an dan asymmetry factor of the
quadrupole interaction of 055, while they are 7.8
ppm, 2.88 MHz, and 1.00 for Al(2). The crystal
structure indicates the presence of two in equivalent
Al sites with a population ratio of 1:1.
The 27 A1 spectra are also recorded for .the other
kaolin ites. The same quadrupole coupling
parameters as those in the sample JV1 can well
explain th e lin e sh apes of th e oth er kaolin ites.
A small fraction of tetrahedral Al is observed at
about 70 ppm for several samples, which might be
ascribed to impurities.
3.3. 1 H static spectra
Figure 4Ashows an 1 H static NMR spectrum of
the sample JV1. The spectrum consists of two
components with different line shapes; a narrow
Lorentzian line with a width of 1.4 kHz and a broad
Gaussian with a28.9kHzwidth.

284 Bulletin of Magnetic Resonance
40 -40
20 10 -10
ppa
Fig.4. *H NMR spectra of the sample Nl, measured at
400.136 MHz. (A) The ordinary single-pulse
sequen ce is used for th e static sample. (B) Th e
CRAMPS spectrum measured with the BR24
pulse sequence in the quadrature phase
detection mode. Chemical shifts are expressed
with respect to tetramethylsilane.
The narrow component can be ascribed to water
molecules adsorbed on the outer surface. This
component is easily diminished by evacuation, and
they grow up gradually in the air atmosphere.
Qi the other hand,the broad component can be
ascribed to the hydroxyl groups in the kaolinite
structure, whose second moment is 105 kHz 2 . The
second moment estimated from the crystal structure
is 92 kHz 2 , in which iH-iH and 1 H- 27 A1
contributions are 67 and 25 kHz 2 , respectively. The
calculated second moment agrees with the
experimental value. These results demonstrate that
the hydrogen atoms in the CH group is in a rigid
lattice state at room temperature.
The *H static NMR spectra of other kaolin ites also
con sist of th e two compon en ts.
3.4. 1 E CRAMPS
The CRAMPS technique is successfully applied to
the kaolinite samples. Figure 4Bshows l H CRAMPS
spectra of the sample JV1. The BR24 pulse sequence is
used in the quadrature phase detection mode.
Considerably large spinning sidebands are observed
on both sides of the central peak, which are caused
by the strong dipole-dipole interaction between *H
and 27 A1 spins. The linewidth in the static state, 29
kHz, is reduced to about 600 Hz by the use of the
CRAMPS technique, where the reduction factor is
about 50. The chemical shift is 2.8 ppm from
tetramethylsilane.
The chemical sh ift of the hydroxyl groups in the
kaolin ites changes slightly depending on the
sample, which might reflect the strength of the
hydrogen bonding or the acidity of the hydrogen .
Acknowledgement
The authors wish to express their thanks to Dr. R.
Miyawaki at Government Industrial Research
Institute, Nagoya and Dr. K. Kuroda at Waseda
University for kindly supplying the kaolin ite
samples.

Vol. 14, No. 1-4 285
1 Introduction
Broadline NMR of Structural Ceramics
Magic angle spinning (MAS) has become the most widely
used NMR technique for the study of inorganic solids.
This popularity has come about because MAS removes
the effects of chemical shift anisotropy, permitting acquisition
of chemical shift spectra in these materials.
However, there are two weaknesses to the MAS
technique, which are accentuated in the study of structural
ceramics.
The first difficulty with MAS is that rotation at the
magic angle does not remove the effects of second order
quadrupolar broadening of the central (+1/2 «—» -1/2)
transition [1]. This has limited the applicability of NMR
for half-integral quadrupolar nuclei like 27 A1, U B, 17 O,
and 91 Zr, which are important constituents of ceramics.
While recent work has shown that the effects of second
order quadrupolar broadening are reduced by working in
larger magnetic fields [2], or by employing more sophisticated
spinning techniques [3], these solutions can be
difficult and expensive to implement.
The other important limitation of MAS, also true of
other sample spinning techniques, is that the physical
form of the sample is restricted to fine powders, homogeneous
cylinders, or chunks of material packed in an NMR
inert powder of similar density. All of these forms present
problems when working with ceramics. Many ceramics
are challenging to machine or grind due to their extreme
hardness. Even when grinding is possible, other
requirements may limit the use of powders. For example,
in determining the effect of long term or repeated heating
of a ceramic, heating a powdered sample may provide
unreliable data because of surface oxidation. Packing
large chunks of sample in a powder with similar density
becomes time consuming when a number of samples have
to be studied, or the sample is being subjected to heating
which changes its chemical or phase composition.
We have used broadline NMR of static samples as an
alternative to MAS for samples with large quadrupolar
splittings. In favorable cases, the resulting powder
pattern, produced by the first order quadrupolar coupling
Chuck Connor
Defence Research Establishment Pacific
FMO Victoria, B.C., VOS 1B0, Canada
[1], directly yields information about the electronic environment
of the nuclei under investigation. Although the
broadline technique for quadrupolar nuclei is not new,
having been first applied over forty years ago [4], it is a
simple, useful technique which is often overlooked. It is
generally applicable only to nuclei with large quadrupolar
splittings, which provide the most difficulty for MAS;
accordingly the broadline technique can serve as a complement
to MAS. The work reported here demonstrates
that, for many ceramics, broadline spectra can provide
useful structural information.
2 Experimental
All spectra were recorded on a Bruker MSL-300
spectrometer, equipped with the BC-131 5 MHz 9 bit
digitizer. The spectrometer operates at 96 MHz for U B,
and 78 MHz for 27 A1. A standard Bruker multinuclear
solenoid probe was used for the broadline spectra. Free
induction decays were collected after a single 1 \is pulse.
Although the excitation profile of this pulse drops to zero
at 1 MHz, reasonable sensitivity is still maintained for
satellites 600 to 700 kHz off resonance. The delay time
before the start of data collection was typically 5 u.s.
Usually the first 2 or 4 points of the FID were discarded
because of pulse breakthrough. The relatively long
deadtime usually prohibits observation of the broad
pedestal portion of the powder pattern, but the cusps,
which by themselves are sufficient to characterize the
quadrupolar splitting, do not appear greatly affected. The
width of the spectral window was 2.5 MHz (+1.25 MHz)
for U B and 1.67 MHz (±0.833 MHz) for 27 A1. A filter
bandwidth of 1 MHz, the largest available on the MSL-
300 in quadrature mode, was used. Additional data collection
parameters used for individual spectra are noted in
the figure captions.
Quadrupolar splittings were measured from a point on
the outside edge of one satellite, at about 80% of the
satellite peak height, to the corresponding point on the
other satellite of the pair. The value 80% was obtained by

286 Bulletin of Magnetic Resonance
comparing simulated powder patterns with and without
Gaussian broadening. Measurements between the maxima
of the satellites give values that are smaller than the true
value, because dipolar broadening shifts the position of
the maxima toward the central transition.
3 Results and Discussion
The 27 A1 (/ = 5/2) resonance in corundum (a-Al2O3) has
been observed by a variety of NMR techniques [4], [5].
For comparison, we show an 27 A1 powder pattern in
Figure 1. For this sample, excellent sensitivity is obtained
500000 0
HERTZ
-500000
Fig. 1. 27 A1 powder pattern of a-Al2O3. 32 scans were coadded,
using a recycle delay of eight seconds. The upper trace is
a magnification of the lower trace by ten.
with about four minutes of signal averaging. From the
splitting of each pair of satellites, we find
e 2 qQ/h = 2.42+0.02 MHz and r| = 0.0, in fair agreement
with the values determined by Pound [4]. The slight
reduction in precision of these numbers, as compared with
Pound's work, is offset by the speed and ease with which
the results can be obtained on a standard solids NMR
spectrometer.
One of the many applications for NMR of ceramics is
to follow high temperature phase changes and reactions.
We have studied the calcination of gibbsite (A1(OH)3) to
form a-Al2O3, which is a complex and poorly understood
process [6]. Results of preliminary work to determine the
applicability of broadline NMR to this problem are shown
in Figure 2. The ^Al spectrum of the unheated material
(determined by X-ray diffraction to be at least 90%
gibbsite) shows five pairs of satellite transitions, with
splittings of 96 kHz, 397 kHz, 517 kHz, 608 kHz, and
960 kHz. The peaks near ±700 kHz, which appear in
many 27 A1 spectra, are probably artifacts. Assuming three
sites are present, the splittings can be paired as follows:
397 and 608 kHz, from a site with e 2 qQ/h = 2.1 MHz and
500000
0
HERTZ
-500000
Fig. 2. 27 A1 powder patterns of gibbsite (A1(OH)3) before
heating (lower trace), after heating at 700 C for 35 minutes
(middle trace), and after further heating at 1070 C for 16 hours
(upper trace). Each trace represents 1 to 2 hours of signal
averaging, with a one second recycle delay.
T] = 0.5; 517 and 960 kHz, from a site with
e^qQ/h = 3.2 MHz and Y| = 0.2. The other possible pairing
gives one site with e*qQlh = 1.8 MHz (r\ = 0.7) and one
with e 2 qQlh = 3.3 MHz (r\ = 0.5). Only one pair of
satellites from the remaining site is visible, with a splitting
of 96 kHz. Two possibilities may explain this absence of
a second pair. If e 2 qQ/h = 0.36 MHz and r\ ~ 1, the
second pair of satellites will overlap the observed pair. If
e*qQ/h = 0.32 MHz and r\ - 0, the second pair will have a
splitting of about 48 kHz, and will not be resolved from
the central transition. An accurate quantification of the
relative population of the three sites from this data is
difficult, but one can say the three sites are roughly equally
populated. The nuclei with the larger couplings are
probably in octahedral environments, as expected from the
reported structure, in which all the aluminum nuclei are in
octahedral environments [7]. The relatively large values
of r| can be attributed to hydrogen bonding, which distorts
the octahedral symmetry of the aluminum sites in gibbsite.
The smaller quadrupolar coupling is about an order of
magnitude less than that typically observed for octahedral
sites (cf. a-Al2O3), so is probably due to a slightly
distorted tetrahedral site. From a simulation of MAS
results at 6.35 T [2], it appears that only two sites are
observed by MAS, both of which are octahedral.
Conversion of A1(OH)3 to CC-A12O3 involves several
intermediate phases such as boehmite, X-A12O3, y-Al2O3,
K-A12O3, 6-Al2O3, and 8-Al2O3, with formation of oc-
A12O3 reportedly occurring at 1140 C [6]. The middle

Vol. 14, No. 1-4
trace in Figure 2 shows the effect of heating gibbsite at
700 C for 35 minutes. The satellites attributed to gibbsite
have disappeared, and the central transition is flanked by
broad features indicative of amorphous material. The
width of the central transition, which may be attributed to
the second order quadrupolar interaction, corresponds to
values of e 2 qQ/h ranging up to 4.4 MHz. After further
heating for 16 hours at 1070 C, the upper spectrum in
Figure 2 was obtained. The sharp features indicate a
substantial amount of amorphous material has converted
to a-Al2O3. This contradicts the notion that conversion to
a-Al2O3 occurs only at temperatures above 1140 C.
However, the conversion is quite sluggish at 1070 C.
These broadline spectra demonstrate conversion of
gibbsite to a-Al2O3, but unfortunately information on the
intermediate phases cannot be obtained from these spectra.
Chemical shift spectra obtained with sample spinning will
probably not be useful in determining which phases are
present in the amorphous intermediate stage, considering
the large quadrupolar couplings which appear to be
present.
Another problem of interest is the oxidation of
ceramics at high temperatures. When boron nitride (BN)
is heated in an oxidizing environment, conversion to
boron oxide gradually takes place. Since both BN and
B2O3 have large quadrupole couplings [8], [9], one can
expect difficulty in resolving the signals from the two
materials in MAS spectra. In fact, the second order
quadrupolar broadening is on the order of 50 ppm, based
on a Larmor frequency of 96 MHz, which is a significant
fraction of the expected range of chemical shift for n B.
To demonstrate that broadline NMR may prove more
useful in monitoring the oxidation of BN, we prepared a
mixture of roughly equal amounts of BN and B(OH)3.
Boric acid was used instead of boron oxide because the
fine powder required for MAS work is not hygroscopic, as
is the B2O3 powder, and the boric acid was readily available.
Boric acid has a quadrupolar splitting of 1282 kHz,
similar to that of B2O3 (1308 kHz).
U B (/ = 3/2) MAS and broadline spectra of the boron
nitride/boric acid mixture were recorded, and are shown in
Figures 3 and 4 respectively. From the MAS spectrum,
one can see that the single peak observed is a composite of
several resonance lines. However, one would have
difficulty determining which boron species are present in
the sample, or even how many species there are. This
spectrum should be contrasted with the broadline
spectrum shown in Figure 4. One immediately sees that
two boron sites are present, with quadrupolar splittings of
1462±10 kHz and 1256±10 kHz. By comparison with
previous work [8], [9] the species are identified as boron
nitride and boric acid respectively. Based on this model
mixture, we expect to be able to monitor the oxidation of
boron nitride.
A variety of approaches, including double-rotation
(DOR), dynamic-angle-spinning (DAS) [3], and MAS in
larger static fields [2], have been used to improve the
50000 0
HERTZ
-50000
Fig. 3. MAS spectrum of n B in a mixture of boron nitride and
boric acid, with a 5 kHz spinning rate. 2296 scans were
collected, with a recycle delay of one second.
resolution of chemical shift spectra beyond that attainable
by conventional MAS. Broadline NMR may compare less
favorably with these new techniques than it does with
MAS. However, these techniques have not been widely
exploited in materials research, and, like MAS, they suffer
from the physical limitations imposed on the sample as
discussed in the Introduction.
Titanium boride (TiB^ has been presented as a candidate
for lightweight armor [10]. The structure consists of
alternating planar layers of boron and titanium, with each
boron nucleus trigonally bound to three other boron nuclei
[11]. To yield the 2:1 stoichiometry, each boron layer
contains twice as many nuclei as a titanium layer. The
crystal structure suggests only one type of boron site is
present, and previous workers have reported only one site,
with a quadrupolar splitting of 180*10 kHz [9]. However,
500000 0
HERTZ
-500000
AJL.
Fig. 4. U B powder pattern of the mixture examined in Figure
3. This spectrum was obtained after about 16 hours of signal
averaging (56989 scans, one second recycle delay).
287

288
we see evidence for at least two sites in breadline U B
NMR spectra (Figure 5), with quadrupolar splittings of
500000
0
HERTZ
-500000
Fig. 5. n B powder pattern of titanium boride (TiBj), from 324
co-added scans with a recycle delay of one second.
177 kHz and 355 kHz. In addition, a weak, poorly
resolved pair of satellites may be present with a splitting
of 195 kHz. It appears that roughly four fifths of the
boron nuclei are located in the 177 kHz site, one fifth in
the 355 kHz sites, with a much smaller fraction in the
195 kHz site.
X-ray diffraction confirmed the sample was greater
than 90% TiB2, with most of the remainder consisting of
titanium oxides. The general features of the pattern
matched that of A1B2, which also has planar layers. The
diffraction pattern showed no similarity to patterns from
compounds containing puckered boron layers, such as
RhB2, TcB2, and Ru2B3. Thus it appears the two sites are
in equivalent geometrical positions within the boron
plane. The larger quadrupolar splitting could be due to
decreased donation of electrons from the titanium atoms
to the boron it orbitals in about one fifth of the boron
nuclei [9]. This heterogeneity within the boron layers
may explain why titanium boride does not fracture along
the interface between layers, as does graphite and many
other layered materials.
4 Summary
For many ceramics, the second order quadrupolar broadening
in MAS spectra exceeds the dispersion provided by
the chemical shift. The far greater dispersion from the
first order quadrupolar interaction can be used to obtain
resolved spectra from these samples. We have provided
examples of several structural ceramics for which
broadline spectra, obtained with a standard solids
spectrometer, yield useful structural information. Thus
the technique shows the potential to complement MAS in
the study of inorganic materials.
5 References
Bulletin of Magnetic Resonance
1. A. Abragam, Principles of Nuclear Magnetism
(Clarendon Press, Oxford, 1961), Chapter VII.
2. e.g. D.E. Woessner, Am. Mineral. 74, 203 (1989).
3. B.F. Chmelka, K.T. Mueller, A. Pines, J. Stebbins,
Y. Wu, and J.W. Zwanziger, Nature, 339, 42
(1989).
4. R.V. Pound, Phys. Rev. 79,685 (1950).
5.a) D. Lee and P.J. Bray, /. Magn. Reson. 94, 51
(1991).
b) J. Haase and H. Pfeifer, /. Magn. Reson. 86, 217
(1990).
c) HJ. Jakobsen, J. Skibsted, H. Bilds0e, and N.C.
Nielsen,/. Magn. Reson. 85,173 (1989).
6. Engineered Materials Handbook, Vol. 4, Ceramics
and Glasses (ASM International, 1991), p.112.
7. A.F. Wells, Structural Inorganic Chemistry
(Clarendon Press, Oxford, 1962), p. 552.
8.a) A.H. Silver, J.Chem.Phys. 32, 959 (1960).
b) C. Connor, J. Chang and A. Pines, J. Chem. Phys.
93, 7639 (1990).
9. PJ. Bray, AIP Conf. Proc. 140,142 (1986).
10. D.J. Viechnicki, M.J. Slavin, and M.I. Kliman,
Ceramic Bulletin, 70,1035 (1991).
11. T. Lundstrom in Boron and Refractory Borides,
edited by V.I. Matkovich (Springer-Verlag, Berlin,
1977), p. 351.

Vol. 14, No. 1-4 289
PERMEABILITY OF LIPOSOMAL MEMBRANES TO MOLECULES OF
ENVIRONMENTAL INTEREST: RESULTS FROM NMR EXPERIMENTS
EMPLOYING SHIFT AGENTS
INTRODUCTION:
by
F.G. Herring*, W.R. Cullen, J.C. Nelson and P.S. Phillips,
Department of Chemistry, University of British Columbia,
Vancouver, B.C., Canada V6T 1Z1
Dimethylarsinic acid (DMA) is a
widely used pesticide and is an important
intermediate in the marine bio-cycling of
arsenic. Studies into the uptake mechanism
of this organo-arsenical have shown that it
enters cells by slow passive diffusion (1).
The work presented in this article describes a
NMR technique that has been developed to
measure the rate of diffusion of compounds
through a phospholipid bilayer. A similar
method has been described by Prestegard
et.al. (2) for the measurement of maleic acid
diffusion constants.
The diffusion rates of molecules
across the membrane of liposomes have been
measured using a variety of techniques, the
most common of which is radio-labelling.
The technique demonstrated in this study has
many advantages over radio-labelling some
of which are as follows; the cost and
difficulties associated with working with
radio-labelled compounds are eliminated, the
method is readily automated, and sampling is
eliminated. The NMR method is applicable
to any water soluble compound which has a
*H resonance signal which does not overlap
*The abbreviation (DMA) does not
distinguish between the protonated (DMAH)
and the unprotonated (DMA") forms of the
acid.
with any other peaks or which can be shifted
either upfield or downfield agent. It can be
used on molelecules which have a
permeability through the phospholipid bilayer
of 10'^cm/s or less.
We present here an investigation of
DMA transport in a model membrane system
by using this NMR method as applied to
diffusion across the membranes of extruded
large unilamella vesicles (LUVs) (3). The
study contributes to our understanding of
diffusive transport across bilayer membranes.
It also illustrates the potential of NMR
spectroscopy for membrane permeability
studies in large unilamella vesicles.
MATERIALS AND METHODS:
Dry egg phosphatidyl-choline was
hydrated with a buffered solution (in D2O)
of DMA at the appropriate pH. The solution
was then subjected to several freeze-thaw
cycles using liquid nitrogen to enhance
entrapment of DMA and to increase the
degree of unilamellarity of the phospholipid
bilayer. The multi-lamella suspension was
extruded through polycarbonate filters with a
200 nm pore diameter under high pressures
to ensure that the vesicles used were all
approximately the same size (3,4). The
resultant solution of LUVs was then passed
down a Sephadex column and eluted with
buffer (in H2O) to remove most of the DMA

290
that was not encapsulated; this procedure
establishes the desired concentration gradient
for diffusion studies. The eluted LUVs were
added to a NMR tube which already
contained Mn 2+ , HEPES, TSP, and glucose.
NMR spectra were obtained at appropriate
time intervals by using a Bruker AM
4OO.The water signal was suppressed by presaturation.
The FIDs were processed with a
line-broadening of 10 Hz. The DMA and
HEPES peaks were integrated and the ratio
of the integrals taken as a measure of the
efflux of DMA, this procedure was adopted
to account for instrument variability.
THEORY:
-dnin/dt = dnout/dt = k (nin/Vin-n0Ut/V0Ut)
Inside peak:
nt. 11 =neq. +(no. 1 .neq. II \e-(l+f)kt/V.
e
in " in v in m> in
Outside peak:
P = k/A
°out
(1)
(4)
Equation (1) is the basic rate
equation for the flux of particles across the
bilayer assuming that the rate of appearance
of the particles on the outside is equal to the
rate of disappearance of the particle on the
inside. Equations (2) and (3) are the
equations which relate the number of
particles on either side of the membrane as a
function of time. Equation (4) relates the
internal volume to the external volume.
Equation (5) relates the rate constant to the
permeability coefficient which is the standard
measure of the rate of diffusion of a
compound through a membrane.
For the integral of the composite
methyl resonance (DMAH and DMA") we
have:
(5)
Bulletin of Magnetic Resonance
where kobs = a ICDMAH Vjn, is the
observed rate constant of diffusion for
DMAH and a = Ka / (Ka + [H+]).
Equation (6) is valid if it is assumed that
^DMA"

Vol. 14, No. 1-4 291
0.25
both sides of the membrane. The membrane
is impermeable to these three additives over
the time scale of an experiment. It should be
noted that DMA is a weak acid which is
about 80% dissociated at pH = 7 (pKa =
6.28).
The decrease of the NMR peak for
those molecules effusing and a corresponding
increase in the peak for DMA outside is
displayed in Fig 1. At the equilibrium position
(the last spectrum shown) the ratio of
the integrals of these peaks correspond to
the ratio of the volume inside to outside (eq
(4))-
The amplitude ratios (total integral of
DMAH and DMA" to the integral of the
HEPES buffer) of both the inside and outside
peaks for the 25 spectra which make up a
single experimental run are shown as a function
of time in Fig. 2. An iterative fit of
these curves using a spreadsheet program
(QPRO) permits the estimation of the rate
constant.
10 15
TIME(SECS.)
(Thousands)
Fig. 2
20 25
In order to demonstrate that the
transport is dominated by the neutral species
(DMAH) as suggested above, we performed
experiments at different pH. The rate of
change of the integral ratio will depend upon
the concentration of DMAH. The integral
ratio will decrease faster due to the increased
fraction (a) of DMAH contributing to the
single methyl resonance observed for both
DMA" and DMAH. The table illustrates the
results we obtained.
Table
The pH dependence of the
observed rate constant
pH
7.00
7.15
7.40
7.73
7.97
(/cm 3 s-l*10" 5 )
14.50
8.89
5.65
2.39
1.56

292
Equation (6) gives the true rate
constant for DMAH from the measured
observed rate constant. The true rate
constant and permeability coefficient are
1.08.x 10" 3 cm 3 /s and 3 x 10' 8 cm/s
respectively. Similar studies for monomethyl
arsonic acid (MMAH) show that the
permeability is 2 x 10"*" cm/s. This
difference is consistent with the general rule
that replacing a hydroxyl group with a
methyl group will increase the permeability
of the molecule by approximately two to
three orders of magnitude(6).
ACKNOWLEDGMENTS:
This work is supported by operating
grants from the NSERC of Canada. The
exceptional technical assistance of Alice Ho
and Anna Mason is appreciated. Alice and
Bulletin of Magnetic Resonance
Anna also received Summer '92 Studentships
from NSERC.
REFERENCES:
(1) W.R. Cullen, B.C. McBride, A.W.
Pickett, Appl. Organomet. Chem. 4,
119,(1990).
(2) J.H. Prestegard, J.A. Cramer and
D.B. Viscio, Biophys. J. (Biophysical
Society) 26, 575, (1979).
(3) M.J. Hope, M.B. Bally, G. Webb and
P.R. Cullis, Biochim. Biophys. Acta
872. 55-65, (1985).
(4) L.D. Mayer, M.J. Hope and
P.R.Cullis, Biochim. Biophys. Acta
858, 181,(19861
(5) W.C. Stein, Channels Carriers and
Pumps, Academic Press, New York,
(1990).

Vol. 14, No. 1-4
NUCLEAR MAGNETIC RESONANCE PARTITIONING STUDIES OF
SOLUTE ACTION IN LIPID MEMBRANES
Lan Ma, Theodore F. Taraschi, and Nathan Janes
Department of Pathology and Cell Biology,
Thomas Jefferson University, 1020 Locust St., Philadelphia, PA 19107
INTRODUCTION: Lipid theories of
anesthesia implicate perturbation of membrane
lipids as the locus for acute anesthetic
action. [1] Chronic exposure to alcohols and
anesthetics induces an adaptive response in
membrane phospholipids that confers resistance
to many of the acute actions of alcohols
and anesthetics. [2]
We have proposed a colligative thermodynamic
reformulation of the Meyer-Overton
hypothesis for anesthetic action. [3,4] This
reformulation implicates configurational
entropy (Scf), the entropy imparted by a solute
upon a membrane structure in the partitioning
process, as the driving force of solute action
on cooperative membrane equilibria. Solute
potency is determined by the competing contributions
of configurational and thermal
entropy (ASt). Equilibria most susceptible to
solute action (where dilute concentrations of
solutes induce a perturbation equivalent to a
large change in temperature) involve large
changes in configurational entropy and small
changes in thermal entropy according to the
following relation. [3]
AT/Tm = AScf/ASt (1)
AT is the perturbation of the midpoint temperature,
Tm, from its value in the absence of
solute. The thermal entropy of an equilibrium
is deduced from calorimetry and is approximately
constant for solute levels of biological
relevance. The remaining unknowns are the
configurational entropy, which is determined
from the partitioning of the solute, and the
perturbation of the equilibrium midpoint.
The colligative thermodynamic framework
implicates solute partitioning as the energetic
force that drives perturbations of cooperative
membrane equilibria by altering the relative
free energies of membrane states. Tests of
the framework require simultaneous measures
of solute partitiomng and membrane structure
over a range of solute concentrations and
temperatures.
293
Spin label* partitioning protocols have often
been used in ESR studies of membrane structure.
[5] Such studies are designed so that the
spin label partitioning probe is a
nonperturbing reporter of membrane structure.
To study solute action on membranes,
however, a partitioning probe should serve the
multifarious role of membrane perturbant,
reporter of perturbations, and reporter of
solute partitioning. Since NMR methods are
not limited to dilute solute levels, such flexibility
is offered. Furthermore, complementary
structural information is available from
simultaneous wideline X H [6], 2 H [7], or 31 P
[8] studies.
PARTITIONING APPROACH TO SOLUTE
ACTION: In this abstract, we describe a H
NMR partitioning approach based on the
uncharged local anesthetic alcohol, benzyl
alcohol. Benzyl alcohol is a clinically used
topical bacteriostatic agent. A variety of
commercial pharmaceutical agents prepared
for injection contain benzyl alcohol for its
preservative properties and for pain relief.
The partitioning approach is based on (/) the
sensitivity of the ring proton chemical shift to
the polarity of its environment and («) the
sensitivity of the ring proton linewidth to
membrane binding. The chemical shift of the
ring resonances in a hydrophobic environment
are shielded and resolved from the ring resonances
of the aqueous alcohol. The sensitivity
of the ring proton resonance to its environment
provides a means of discriminating the
aqueous alcohol resonance from the partitioned
alcohol resonance. The dependence of
the three chemically distinct ring proton
chemical shifts on their environment is shown
in Table 1 for benzyl alcohol (5 mole fraction
%) in a variety of bulk solvents at 22°C. The
resonance exhibits a diamagnetic shift in
hydrophobic solvents. A modest correlation
between the chemical shift and Hildebrandt's
solubility parameter ($*) for the solvent is
evident.

294
SOLVENT
Water
Methanol
1-Propanol
1-Butanol
1-Octanol
1-Decanol
Acetone
Methylene Chloride
Chloroform-dj
Carbon Tetrachloride
Hexane
I
7.8
I
7.6
I
7.4
7.2
PPM
5*
23.4
14.5
11.9
11.4
10.3
9.9
9.8
9.2
8.6
7.3
7.0 6.8
7.41
7.32
7.29
7.29
7.27
7.27
7.34
7.29
7.32
7.19
7.20
Figure 1: The ring proton resonances of benzyl alcohol
are shifted upfield and broadened upon binding to
lecithin membranes.
Further discrimination stems from the
motional restrictions imparted by the membrane
environment that is reflected in the
spin-spin relaxation. The ring resonances
corresponding to the free and bound drug are
shown in Figure 1 for a lecithin model membrane
in the L_ state. The resonance of the
bound agent is oroadened due to immobilization
in the membrane. The T2 of the bound
agent is approximately 6 msec, while the T2 of
the free agent is more than three orders of
magnitude greater (11 sec). The different
TABLE 1
(ppm from TMS)
7.41
7.32
7.25
7.24
7.21
7.21
7.30
7.29
7.32
7.16
7.20
7.41
7.23
7.17
7.16
7.12
7.12
7.21
7.29
7.32
7.16
7.13
Bulletin of Magnetic Resonance
S (ppm) avg
7.41
7.29
7.24
7.23
7.20
7.20
7.28
7.29
7.32
7.17
7.18
relaxation properties allow for spectral editing
based on T2 using spin echoes.
The partitioning of benzyl alcohol into
membranes is modest. Consequently, the
lipid to water ratios of the sample must be
large to obtain accurate simulations of the
broad bound resonance, while a sample size
and geometry consistent with high Zeeman
field homogeneity must be maintained. In
practice, to reduce sample demands, an internal
acetate standard was used to determine
the aqueous alcohol concentration in a dilute
membrane suspension. Since the lipid concentration
is known, the intramembrane concentration
is obtained by difference to yield
the partition coefficient. To ensure that the
integrated aqueous resonance is not contaminated
by the broad bound resonance, a
CPMG sequence is used to delay acquisition
by 25 msec in order to filter the broad bounddrug
resonance. This filtering method also
removes the dipolar broadened lipid resonance
to improve baseline definition.
Since the colligative thermodynamic framework
equates the action of solute and temperature
through entropy, precise temperature
regulation is required. The aqueous benzyl
alcohol resonance exhibits a temperature
dependent chemical shift. In order to maintain
a consistently reproducible temperature,
we have taken advantage of this temperature
dependence. The chemical shift differences
between the HOD and free benzyl alcohol
resonances as a function of temperature is
shown in Figure 2.

Vol. 14, No. 1-4
Q.
0.
UI
o
cc
UI
u.
u.
a
u.
X
in
2.9
2.8
2.7
2.6
2.5
2.4
2 ^
-
-
-
c
i 1
1 , 1
) 10 20
Figure 2: The chemical shift difference between t
benzyl alcohol ring proton resonance and the HOD resonance
shows a temperature dependence.
1
1
30
TEMPERATURE (C)
-
_
-
_
1
40 5
ANALYSIS OF PARTITIONING: The
degree of anesthetic partitioning into a membrane
system is sensitive to and characteristic
of the state of the lipid assembly. The equilibrium
constant, K , is deduced from the
partitioning changes characteristic of the
interchange between membrane states. The
temperature dependence of the partitioning
exhibits the following functional form for a
state change between two membrane structures.
[3]
KP =
C =
expC(T-Tm)
AHvH
„
RTT m
(2)
(3)
The partition coefficient for the membrane
states a and ft are Kp« and Kp", respectively.
These partition coefficients are not necessarily
constant and may be altered to include a
temperature dependence. The total partition
coefficient is Kp. The midpoint temperature
is Tm. A fit of the experimental data to this
function yields partition coefficients for each
phase, the midpoint temperature, and the
van't Hoff enthalpy (AHvH), a measure of the
cooperativity of the equilibrium.
295
TEMPERATURE (K)
Figure 3: The molal partition coefficient of benzyl alcohol
into multilamellar DPPC membranes is shown as a
function of temperature for two concentrations of benzyl
alcohol. The fit corresponds to the theoretical multiparameter
least-squares analysis described in the text. The
derivative of the fit to the data is shown offset below.
The percent mole fraction intramembrane benzyl alcohol
concentrations at the L
p,
p p equilibrium
midpoint are as follows: Panel A: Lp , = 0.23%, Pp ' =
1.1%; Panel B: 2.2%, 16.8%; The mole fraction benzyl
alcohol concentrations at the P^/ -* La equilibrium
midpoint are as follows: Panel A: Pp, = 0.84%, La=
2.1%; Panel B: 11.7%, 24.2%; For comparative purposes,
general anesthetic intramembrane concentrations
are considered less than 5 mole fraction percent.
The analytical framework presented is not
specific to the partitioning analysis. It is
broadly applicable to any technique in which
the observable is characteristic of each state.
ALCOHOL ACTION IN MODEL
MEMBRANES: The lecithin membrane,
DPPC (l,2-dipalmitoyl-.stt-glycero-3-phosphocholine),
adopts three well-studied structures
or phases, a gel-structure (L«,), a ripplestructure
(Pp,), and a fluid bilayer-structure
(La). [9] Since the interchange between these
three membrane structures is driven by

296
entropy, changes in solute, temperature and
pressure alter the energetic balance to favor a
given structure. The Lp, •* P^/ equilibrium
(pretransition) exhibits an equilibrium midpoint
temperature determined by calorimetry
as 34.8°C. [9] This change in state is accompanied
by a small change in thermal entropy
(12.5 J mol" 1 K- 1 ). The P^, -* La (main transition)
exhibits an equilibrium midpoint temperature
determined by calorimetry as 41.0°C.
This change in state is accompanied by a relatively
large change in thermal entropy (85.6 J
mol' 1 K*i). [9] The large difference between
the thermal entropy changes associated with
these two equilibria provides a simple system
in which to test the predictions of the colligative
thermodynamic framework, that solute
action occurs through entropy and that equilibria
characterized by a small thermal
entropy change should be most susceptible to
perturbation.
The temperature dependence of benzyl
alcohol partitioning at two substantially different
alcohol concentrations is shown in
Figure 3. Figure 3A corresponds to benzyl
alcohol concentrations below that required for
general anesthesia; whereas, the concentration
in Figure 3B is near that required for
local anesthesia. The partition coefficients
obtained by the NMR method are in excellent
agreement with direct radiolabel measures.
[10] Two discontinuities correspond to the
two membrane equilibria. It is these changes
in partitioning that provide the configurational
entropy by which solutes perturb equilibria.
The low entropy Le, -+ Ppr equilibrium
exhibits greater sensitivity to the alcohol
than the high entropy P^, -*• La, as qualitatively
predicted by the thermodynamic model.
The quantitative test for the model is
shown in Figure 4. The partitioning method
provides intramembrane solute concentrations,
which, in turn, provide the magnitude of
the configurational entropy imparted to each
membrane structure. This contribution
lowers the free energy of each state according
to the magnitude of the partitioning, and
thereby alters the difference in free energy
and shifts the equilibrium. The experimental
points are in good agreement with the theoretical
predictions (represented by the lines)
at dilute alcohol concentrations for which the
thermodynamic treatment is derived and
which corresponds to pharmacological levels
Bulletin of Magnetic Resonance
1 10
INTRAMEMBRANE HOLE FRACTION U) DIFFERENCE
Figure 4: The dependence of the equilibrium midpoint
temperature of DPPC on the presence of benzyl alcohol.
The benzyl alcohol intramembrane concentration difference
between the initial and final states at the equilibrium
midpoint is shown. Data are presented for the low
entropy Lp, -+ Fg, (pretransition; filled circles) and
the high entropy Fp, -*• La (main transition; open circles)
equilibria. The coUigative thermodynamic predictions
(eq. 1) are represented by the lines. The solid
portions of the lines designate the average
intramembrane concentrations at the midpoint which
correspond to the range of pharmacological relevance
for general anesthesia.
Particularly striking is the dramatic contrast
in benzyl alcohol sensitivity exhibited by these
two equilibria. At average intramembrane
concentrations of 5 m.f.%, the low entropy
equilibrmm is perturbed by approximately
12°C, whereas the high entropy equilibrium is
perturbed by approximately 1°C. Not only
does this observation support the predictions
of the thermodynamic model, but it demonstrates
that remarkably low intramembrane
concentrations of nonspecific solutes can precipitate
quite substantial effects upon membrane
structure.
ALCOHOL ACTION IN LIPOSOMES
MADE FROM RATS CHRONICALLY
EXPOSED TO ANESTHETICS: Rat liver
microsomes obtained from rats exposed to
nitrous oxide or fed ethanol were isolated and
liposcmes formed from the extracted phospholipids.
Shown in Figure 5 are representative
benzyl alcohol partitioning traces for the
liposomes prepared from the ethanol-fed and
control animals. The partition coefficient into

Vol. 14, No. 1-4
UJ
FFIC]
UJ
o
t i
TION (
IRTI
«*.
Q.
40LAL
40
35
30
25
• •! 1
• •
-i—' 1 ' r~ —i 1 1 r—
\mr •
^ T
1 , 1 , 1
i . \
20 30 40 50 60 70
TEMPERATURE
Figure 5: Benzyl alcohol partitioning traces are shown
for liposomes made from rat-liver microsomal-phospholipids.
The samples represented in the lower trace
(filled triangles) were prepared from chronically
ethanol-fed rats. The samples represented in the upper
trace (filled circles) were prepared from their pair-fed
littermates.
the control samples is larger than the treated
samples. This difference in partitioning is
characteristic of 'membrane tolerance'. [2] A
structural equilibrium is apparent in the control
samples near 37°C that is lacking in the
samples obtained from treated animals. Similar
results are obtained from the chronic
nitrous oxide paradigm. These results evidence
an adaptive response to the chronic
presence of anesthetic agents that results in
altered domain structure in the reconstituted
system. Similarly, structural lipid domains are
predicted at the anesthetic locus in our thermodynamic
reformulation of the Meyer-
Overton hypothesis.
CONCLUSIONS: Alcohols and anesthetics
act through the entropy imparted by partitioning
to modify membrane architecture.
Analysis of anesthetic action requires simultaneous
measures of solute partitioning and
membrane structure over a wide range of
-
-
297
solute concentrations. NMR partitioning
methods offer unique advantages in such
inquiry since the solute can serve the multifaceted
role of perturbant, reporter of membrane
perturbations, and reporter of solute
partitioning.
METHODS: Spectra were obtained on a Bruker 8.5T
AM spectrometer operating at 360 MHz with deuterium
lock. Lipids were dried under N2, evacuated (

7. Smith, R.L. and Oldfield, E. Science 225,
280-288 (1984).
8. Taraschi, T.F., Lee, Y.-C, Janes, N., and
Rubin, E. Annals New York Acad. Sci.
625,698-706(1991).
9. Chen, S.C. and Sturtevant, J.M.
Biochemistry 20. 713-718 (1981V
10. Colley, CM. and Metcalfe, J.C. FEBS
Letters 24,241-246 (1972).
11. Ellingson, J.S., Janes, R, Taraschi, T.F.,
and Rubin, E. Biochim. Biophys. Acta
1062,199-205 (1991).
Bulletin of Magnetic Resonance

Vol. 14, No. 1-4
ABSTRACT
WEAK MOLECULAR INTERACTIONS: NMR SPECTROSCOPY
OF ORIENTED MOLECULES
C.L. Khetrapal
National Institutes of Health, Bethesda, MI)-20892, USA
"Indian Institute of Science, Bangalore 560 012, India
NMR spectroscopy of oriented
molecules is employed to study
weak molecular interactions. The
information is obtained from the
changes in the molecular order
and the structure as a result of
the complexation.
Specific results on the pi
and charge transfer complexes
formed by iodine, chloroform, and
silver nitrate with nitrogen
heterocycles, aromatic systems
and acetonitrile are reported. A
method for the determination of
the order parameter and the
structure of the 'complexed'
species is presented and its
utility demonstrated. The use of
mixture of liquid crystals of
opposite diamagnetic anisotropies
to investigate the extraordinary
symmetry of Buckminster
Fullerene , Cgo> 1S illustrated.
INTRODUCTION
Application of NMR to
weak molecular interactions
study
from
changes in chemical shifts,
indirect spin-spin coupling
constants and line width is quite
well known and has been employed
since the early days of NMR.
However, the use of direct
dipolar couplings/order parameters
derived from the spectra
of molecules oriented in thermotropic
liquid crystals is relatively
new (1) and provides more
quantitative results since
changes in molecular structure if
any as a result of complexation
* Address for Correspondence
can also be determined directly
and used for such a purpose.
Results on the pi and the
charge transfer complexes formed
by chloroform, iodine, silver
nitrate with aromatic systems,
nitrogen heterocycles and acetonitrile
are described in this
communication. A study of
Buckminster Fullerene (C, )
oriented in mixture of liquid
crystals of opposite diamagnetic
anisotropies is also reported in
order to establish the extraordinary
symmetry of the molecule
indicating, thereby, negligible
solvent-solute interactions.
2. METHODOLODY
Information on weak molecular
complexes has been obtained
from the changes in the degree of
order as well as the molecular
structure produced as a result of
the complex formation.
The degree of order of a
molecule dissolved in a "hematic
liquid crystal usually decreases
with the increase of temperature
or concentration. Any abnormal
change in the degree of order is
attributed to the formation of
the complex(es). Recently, we
have investigated metal ion
ligand interactions between monovalent
metal ions such as Li+ and
Ag+ in LiCIO , LiBF and AgNC
with ligands like aoetonitri1e?
dimethyl sulphoxide, pyridine and
acetone in thermotropic liquid
crystals (2-4). Interactions of
iodine (and bromine) with
299

300 Bulletin of Magnetic Resonance
pyridine (4-6), pyrimidine (7)
and quinazoline (8) have also
been investigated. Work on chloroform-benzene
pi complexes has
also been undertaken. We have
also employed the use of mixture
of liquid crystals of opposite
diamagnetic anisotropies in order
to find out if any detectable
distortions in the spherical symmetry
are present in Buckminster
Fullerene , C 59, as result of
solvent-solute interactions (9).
Some such results are reported
below briefly.
LiCIO •-! igand
1igands used in interactions:
(a)
The 1 igands "'used in such studies
are dimethyIsulphoxide, acetonitrile
and pyridine. The liquid
crystals employed are Schiff
bases such as N-(p-methoxy (or
ethoxy) benzylidene)-p-n-butyl
aniline (MBBA or EBBA) and their
deuterated (-d2) analogue with
deuterium substituted at position
ortho to the -N= group in the
butyl aniline ring. The results
indicate ' the formation of two
types of complexes in these cases
with one having "isotropic-1ike"
structure. The "isotropic-1ike"
complex may be of the type ML4
where M is the metal ion and L is
the 1igand and the other could be
the one containing different
ratios
gands.
of the metal ions to li-
(b) LiBF -Ligand solutions: The
1igand used in such studies was
acetonitrile. The liquid crystals
employed were trans, trans-4-npropyl(1,1'-bicyclohexyl)-4'carbonitrile
(ZLI-1184) and
trans-4-pentyl-1-(4-cyanophenyl)
cyclohexane (S-1114). The results
indicate the formation of tetrahedral
1ithium-acetonitrile complexes
and the coordination of
lithium from LiBF^ and LiClC^
with the cyano group of ZLI-1184.
No evidence was found to support
either lithium ion complexation
with S-1114 or the tetrafluorobo-
rate moiety in these systems.
(c) AgJNO3-ligand complexes: In
this case AgNC^-CI^CN (CD3CN)
complexes have been investigated
in two different liquid crystals,
namely, ZLI-1167 (a ternary eutectic
mixture of propyl, pentyl
and heptyl bicyclohexyl carbonitrile)
and S-1114 from both the
proton and the deuteron NMR
studies. The results indicate the
formation of different types of
AgNO -CHoCN complexes, i.e., ML4,
ML 3.?..etc.
(d) Benzene-chloroform complex:
It has been studied from the
reduction of molecular order of
benzene upon addition of iodine.
From the shape considerations,
the order parameter of such a
complex should be either opposite
in sign to or smaller in magnitude
than that of benzene under
comparable conditions. The observed
spectrum of benzene in
ZLI-1167 containing iodine which
arises from the average orientation
of the "complexed" and the
"free" benzene, therefore, has an
order parameter which is smaller
in magnitude than that of the'
"uncomplexed" benzene. The change
of the chemical shift of the
benzene carbon is also consistent
with this observation.
(e) Iodine(bromine)-pvridine
charge transfer complex: Such
complexes using NMR spectroscopy
of oriented systems were first
studied in 1973 and 1983 (5, 6)
from the drastic change of the
order parameter of the C2~axis of
symmetry of pyridine. We have
however, detected the formation
of the "inner" complexes
[(PYlfl"] as well as those of
the type PY.I2 unambiguously. The
order parameters of the
"complexed" species have also
been determined (4).

Vol. 14, No. 1-4
(f) Extraordi nary symmetry
Buckmi nster Fullerene. C,_
of
in
REFERENCES
nematic solutions: The "^ C-NMR
spectrum of C6o was studied in a
mixture of nematic liquid
crystals of opposite diamagnetic 1.
anisotropies. The method makes
use of the change in the anisotropic
parameters resulting from
the switching of the order parameters
at the critical point in
the mixture; the change is by a 2.
factor of 2 or -1/2 depending
upon the direction of approach of
the critical point (10). By an
appropriate adjustment of the
concentration and temperature, it
is possible to observe both the 3.
types of orientations to coexist.
Even in tetrahedral molecules
such as methane and tetramethysilane,
the coexistence of 4.
the two spectra has been observed
(11). On the other hand, the
observation of a single line at
the same position as in the isotropic
phase at the critical 5.
point where the coexistence of
the two spectra in methane or
tetramethylsi 1ane is observed,
indicates the absence of any 6.
detectable distortions in the
molecule. This has actually been
observed in a solution of C6Q in
a 1:1 mixture of liquid crystals
ZLI-1167 and S-1114 containing 7.
tetramethylsilane, at 332.4 K.
The spectrum clearly shows the
coexistence of two spectra for
tetramethylsi 1ane whereas for the 8.
C60 only a single line at its
isotropic position (140.37 ppm
with respect to TMS) is observed.
The- results, therefore, establish 9.
that there are no detectable
distortions from spherical symmetry
in C in the nematic solvents.
TcP® our knowledge this is
the first molecule where no detectable
distortation from spherical
symmetry have been observed.
It should, therefore, serve
as an ideal reference for the 10
study of the chemical shift
sotropy.ani-
301
P. Diehl, H.R. Wasser, G.A.
Nagana Gowda, N. Suryaprakash
and C.L. Khetrapal,
Chem. Phys. Lett. 159. 199
(1989)
G.A. Nagana Gowda, N. Suryaprakash,
R.G. Weiss, C.L.
Khetrapal and P. Diehl,
Magn. Reson. Chem. .28, 642
(1990)
G.A. Nagana Gowda, R.G.
Weiss and C.L. Khetrapal,
Liq. Cryst. 10, 659 (1991)
C.L. Khetrapal, G.A. Nagana
Gowda and N. Suryaprakash,
Spect. Chim. Acta. (in the
press)
C.A. Veracini, M. Longeri
and P.L. Barili, Chem. Phys.
Lett. 19., 592 (1973)
D. Catalano, C.A. Veracini,
P.L. Barili and M. Longeri,
J. Chem. Soc. Perk Trans II,
171, 1983
N. Suryaprakash, R. Ugolini
and P. Diehl, Magn. Reson.
Chem. 29., 1024 (1991 )
N. Suryaprakash, C.L.
Khetrapal and P. Diehl (To
be published)
C.N.R. Rao, T. Pradeep, R.
Seshadri, R. Nagarajan, V.N.
Murthy, G.N. Subbanna, F.
D'Souza, V. Krishnan, G.A.
Nagana Gowda, N. Suryaparakash,
C.L. Khetrapal and
S.V. Bhat, Ind. J. Chem. 31.
F5 (1992).
C.L. Khetrapal and A.C.
Kunwar, Chem. Phys. Lett.
.82, 170 ( 1981 ) .

302 Bulletin of Magnetic Resonance
11. C.L. Khetrapal, A.C. Kunwar
and M.R. Lakshminarayana,
Mol. Cryst. Liq. Cryst. 111.
189 (1984).

Vol. 14, No. 1-4 303
Introduction:
Structural Studies of Collagen by Solid State NMR.
Richard J. Wittebort and Anne Marie Clark
University of Louisville, Department of Chemistry,
Louisville, KY 40292 USA
Collagen is one of the most abundant
proteins nature and has the same structure and
virtually the same composition for diverse
species. It provides the organic matrix for teeth
and bones, and gives tendons and blood
vessels their strength. Collagen has three
polypeptide chains, two al chains and one a2
chain, coiled together in a 3-1 helix. Each
chain is formed from a repeating (Gly-X-Y)
triad with Gly always in the first position. The
other two positions can have any amino acid
but frequently (about 30%) have proline (Pro)
and hydroxyproline (Hyp). Hyp always
occurs in position 3 while Pro usually occurs in
position 2 and only rarely in position 3 in adult
animals. The principle difference between the
types of chains is their direction of orientation.
Several models have been developed to
explain the three dimensional structure of
collagen. From steric considerations, it is
thought that the collagen triple helix has Gly
alpha hydrogens pointing inside. Despite
numerous attempts to solve the X-ray structure,
collagen's lack of long range order has made it
impossible to determine a unique structure from
diffraction methods. 1 " 5
Collagen is a uniaxially ordered fiber
with helical symmetry about the fiber axis. In
order to do structural studies by solid state
NMR, the fiber axis must be aligned with the
magnetic field. Opella used solid state
techniques to determine the three dimensional
structure of bactreiophage fd coat protein. 6
This experiment requires at least one direction
of orientation and the molecular sites of interest
must be immobilized and uniformly oriented
with respect to the magnetic field. In single
crystals, any arbitrary sample orientation can be
studied; however, for uniaxially oriented -
samples the direction of orientation must be
along the applied field to obtain single crystal-
like spectra. The bacteriophages are well suited
for this type of experiment. They
spontaneously align with the magnetic field and
their large size and rod-like shape immobilize

304
the protein subunits. The alpha helical
structure determined agrees well with X-ray
structures of alpha helical proteins.
The three dimensional structure of
polypeptides can be described as a series of
connected peptide planes. By selectively
labelling specific sites and determining their
orientation, structural constraints will be added
to the three dimensional structure of collagen.
The frequency of resonance lines in
solid state NMR spectra depend on the
orientation of the local molecular environment
relative to the magnetic field Ho. The observed
splitting can be used to determine the angle a
bond makes with the magnetic field by the
following equation, assuming an axially
symmetric interaction:
Av = vn(3cos 2 6-1) Eqn. 1
Av is the observed splitting, 0 is the angle
between the bond and the applied field and vn
is the quadrupolar coupling constant. It has
been found that in order to determine the
orientation of a peptide plane at least two labels
per plane must be examined.
Experimental:
The solenoidal coil geometry typically
used in solid state NMR probes is
Bulletin of Magnetic Resonance
unsatisfactory since a bundle of fibers is only
conveniently placed in the coil in the
perpendicular orientation. Attempts were made
to wrap tendons around several small slides
and stack them in the sample holder but
satisfactory alignment was not obtained. A
different probe design is called for by this
experiment. One probe design is based on a
small flat coil geometry suggested by Opella. 7
Small, flat frames were made for us upon
which to wind the coil. A series of rat-tail
tendons were tied together and wrapped around
a 1 x 1 cm polycarbonate card. The card was
then placed in the coil frame and the coil wound
around it. This arrangement consistently gives
a 90 degree pulse width of 2.7 JIS at 1 kW
power. This design has the advantage of
allowing the fiber axis to be placed at any angle
relative to Ho. Proper turn spacing for good
RF field homogeneity and consistent
inductance is insured by using a frame with set
holes. This also allows the sample to be kept
in the center of the coil, away from the edges
where the field homogeneity is poor. We
obtained satisfactory results with this probe.
We were, however, limited to a small sample

Vol. 14, No. 1-4 305
size due to the polycarbonate cards and
resultant poor filling factor. We had to retain
the cards to keep the fibers aligned; otherwise,
they shrink during changes in humidity and
temperature.
We are principally interested in
structural information obtainable from samples
with the fiber axis along Ho, that is, the parallel
orientation. From that premise, we designed a
probe using a Helmhotz coil similar to liquid
NMR probes. The probe was constructed
based on our circuit for a double resonance
design with a 0.5 x 1 cm saddle. The probe
has a 90 pulse width of 3.2 microseconds with
400 watts power at the 2 H frequency
(38.8 MHz). This probe is capable of holding
ten times the sample as the flat probe and has
provisions for stretching the sample.
We have labelled selected sites in the
collagen molecule by incorporating deuterated
amino acid into the rat tail tendon. Torchia
incorporated labelled glycine into 1/3 of the Gly
positions in rat tail tendon by injecting it into
rats. 8 We slightly modified his injection
scheme in order to label selected amino acids in
rat tail tendon. Given the low natural
abundance of deuterium, we were confident
that the only signal we would see would be
from our covalently bound label.
The first covalently labelled position we
tried was alpha-d-Pro. The solutions for
injection were 1.3 M in the amino acid and
were neutralized with NaOH. Ten rat pups
were injected once a day for 21 days and then
sacrificed. The rat tail tendon was extracted for
use in our experiments. The degree of
incorporation of d-Pro in the tendon was
determined by GC/MS and found to be 5.4%
which agrees well with our NMR results. No
incorporation of deuterium in any amino acids
other than Pro and Hyp was seen. The
oriented fibers were run on the Helmholtz coil
probe. A splitting of Av=l 17 kHz was
observed and the angle of the C-D bond relative
to the magnetic field determined.
The glycine amides were labelled by
exchange. Samples were heated at 40°C and at
constant humidity to completely exchange the
labile H's. At 78% relative humidity, exchange
reaches a constant level after 24 hours, at 38%
relative humidity three days are required to
reach a constant level of exchange. The
samples were cooled to room temperature at
constant humidity for an hour and then back

306
exchanged in liquid H2O. Samples that were
not heated to label the amide positions showed
an extremely rapid loss of the solid echo peak.
The heated sample, after back-exchange, had
10% of the solid echo remaining, indicating the
exchanged sample has a significant amount of
bound water. The doublet splitting of 155 kHz
shows that the hydrogen bonded amides are in
fact nearly perpendicular to the fiber axis in
agreement with the predictions from the X-ray
structure. Recent packing studies predict the
collagen molecule is tilted 4 to 5 degrees off the
fiber axis.
Results and Conclusions:
Bulletin of Magnetic Resonance
The asymmetric lineshape of the back
exchanged sample indicates a distribution of
orientation about 6=89 degrees. The spectral
simulation for a Gaussian distribution of
orientations P(6) a exp(-6 2 /2a) with 6=90 and
o=17 matches our experimental data quite well.
We have successfully labelled selected
positions and determined their orientation
relative to the fiber axis. By carefully choosing
our labels, we will be able to determine the
orientation of various peptide planes in collagen
and gradually build the three-dimensional
structure.
Table I. Angle between X-D bond and the fiber axis comparison of solid state NMR results and
various models derived from X-ray crystallography data.
Deuteron Experiment Fraser's 4 Ramachandran's 1 Ramachandran's 2 Yonath's 5
Rich-Crick bridging water two bonded
ProCa-D 90
GlyN-D 89
References:
81.3
89.4
1 Ramachandran, G. N. and Kartha, G.,
Nature, 1955, 776, 593.
2 Ramachandran, G. N. and
Chandrasekaran, R., Biopolymers, 1968,
6,1649.
3 Rich, A. and Crick, F. H. C, Nature,
1955,276,915.
4 Fraser, R. D. B., /. Mol Biol., 1979,
129, 463.
73.8
78.7
73.1
78.4 83.1
5 Yonath, A. and Traub, W., J.Mol. Biol,
1969,43,461.
6 Opella, S. J., Quart. Rev. Biophys.,
1987,79,7.
7 Bechinger, B. and Opella, S. J., /. Magn.
Reson., 1991, 95, 585.
8 Jelinski, L. W. and Torchia, D. A., /.
Mol. Biol, 1979,133, 45.

Vol. 14, No. 1-4 307
Calender of Forthcoming
Conferences in Magnetic
Resonance
March 8-14, 1993
1993 Keystone Symposia on Molecular & Cellular
Biology, Taos, New Mexico, USA
Frontiers of NMR in Molecular Biology - III;
Organizers: Thomas L. James, Stephen W. Fesik
and Peter E. Wright.
Information:
Keystone Symposia
Drawer 1630
Silverthorne, CO 80498
Phone: 303-262-1230
March 14-18, 1993
Experimental Nuclear Magnetic Resonance Conference,
The Adam's Mark Hotel, St. Louis, Missouri,
USA
There will be sessions covering high resolution
and multi-dimensional NMR in liquids, materials
imaging, biological imaging, hardware, NMR
in solids, calculations and data processing, and that
old standby, miscellaneous. The deadline for submitting
poster abstracts will be December 30,
1992.
Contact:
ENC
815 Don Gaspar
Santa Fe, New Mexico 87501
Phone: 505-989-4573
FAX: 505-989-5073
April 1993
High Resolution NMR Spectroscopy (a residential
school), University of Sheffield, England
For information contact:
Ms. L. Hart
The Royal Society of Chemistry
Burlington House
Piccadilly, London W1V 0BN
England
Tel: 071-437-8656
September 6-10, 1993
Second International Conference on Magnetic
Resonance Microscopy, Heidelberg, Germany
The program includes plenary lectures, as well
as oral and poster contributions selected from the
submitted papers. The papers will be judged solely
on the basis of an abstract. Preregistration deadline:
April 30, 1993.
Organization and Program: Winfried Kuhn
(St. Ingbert), Bernhard Blumich (Mainz).
International Advisory Board: J. L. Ackerman
(Charlestown), L. J. Berliner (Columbus),
P. T. Callaghan (Palmerston), W. Edelstein
(Schenectady), A. N. Garroway (Washington),
A. Haase (Wiirzburg), W. E. Hull (Heidelberg),
L. W. Jelinski (Ithaca), J. L. Koenig (Cleveland),
G. A. Johnson (Durham), P. Jonson (London),
P. C. Lauterbur (Urbana), P. Mansfield (Nottingham),
B. Maraviglia (Rome), G. D. Mateescu
(Cleveland), J. M. Pope (Kensington), V. Sarafis
(Richmond), S. Sarkar (King of Prussia)
For further information please write to:
Dr. Winfried Kuhn
Fraunhofer Institute
Ensheimer Str. 48
DW-6670 St. Ingbert, Germany
phone: +49-6894-89738
FAX: +49-6894-89750
or
Dr. Bernhard Blumich
Max-Planck Institute for Polymer Research
Postfach 3148
D-6500 Mainz, Germany
phone: +49-6131-379125
FAX: +49-6131-379100
The editor would be pleased to receive
notices of future meetings in the field of
magnetic resonance so that they could be
recorded in this column.

308
Recent Magnetic Resonance Books
1 Magnetic Resonance Spectroscopy in Biology
and Medicine (1992). Edited by J. D. De Certaines,
W. M. M. J. Bovee and F. Podo. Contents: Presents
the experimental and basic aspects of functional and
pathological tissue characterization of MRS. A balance
is drawn between the basic science, practical
technologies and biomedical applications. Covers
recent developments in the field: localization, 2D
NMR, spectroscopic imaging, data quantification
and quality assessment, as well as the basic principles
of magnetic resonance spectroscopy. Pergamon
Press, ISBN 0-08-0410170 (flexicover) $70.00; ISBN
0-08-0410189 (hardcover) $170.00.
1 In Vivo Magnetic Resonance Spectroscopy I.
Probeheads and Radiofrequency Pulses, Spectrum
Analysis (1992). Edited by M. Rudin, Springer, 345
pp. ISBN 0-387-54547-6 (hardcover) $119.00.
l In Vivo Magnetic Resonance Spectroscopy II.
Localization and Spectral Editing (1992). Edited by
M. Rudin and J. Seelig, Springer, 368 pp. ISBN
0-387-55022-4 (hardcover) $119.00.
^In Vivo Magnetic Resonance Spectroscopy III.
In Vivo MR Spectroscopy: Potential and Limitations
(1992). Edited by M. Rudin and J. Seelig,
Springer, 293 pp. ISBN 0-387-55029-1 (hardcover)
$98.00.
1 Annual Reports on NMR Spectroscopy. Volume
24 (1992). Contents: Developments in solid state
NMR. Solid state NMR imaging. NMR studies of
interfacial phenomena. NMR measurements of intracellular
ions in living systems. 199Hg NMR parameters.
Applications of NMR methods in coal
research.
1 Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 24 No. 3(1992). Contents: Structural
characterization of noncrystalline solids and
glasses using solid state NMR.
1 Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 24 No. 2 (1992). Contents: 129 Xe
J New additions to the list.
Bulletin of Magnetic Resonance
NMR as a probe for the study of microporous solids:
A critical review. Simulation of 2D NMR spectra for
determination of solution conformations of nucleic
acids.
Progress in Biophysics & Molecular Biology.
Volume 57 No. 1 (1992). Contents: ENDOR and
EPR of metalloproteins. Free energy transduction
in polypeptides and proteins based on inverse temperature
transitions.
Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 24 No. 1 (1992). Contents: 13 C
NMR spectroscopy of oleanane triterpenoids.
NMR at Very High Field (1991). Guest editor:
J. B. Robert, Springer, 168 pp. ISBN 0-387-52946-2
(hardcover) $79.00.
1 Transition Metal Nuclear Magnetic Resonance
(1991). Edited by P. S. Pregosin. Contents: The
book contains a collection of review articles concerned
with measuring, understanding and using the
nuclear magnetic resonance spectra of the metals of
Groups 3-12. The reader is provided with a view
on how these nuclei are currently being approached,
and what information can be obtained. The authors
have liberally reproduced spectra as well as correlations
relating metal-NMR data to different physical
characteristics of their molecules. 364 pp. ISBN
0-444-88176-X $169.00.
Chemical Reviews. Volume 91 No. 7 (1991).
Contents: Low-temperature solid-state NMR of proteins.
Structure and dynamics of solid polymers
from 2D- and 3D-NMR. NMR under high gas pressure.
Nuclear magnetic resonance at high temperature.
Gas-phase NMR spectroscopy. Selective
excitation in high-resolution NMR. Application of
the linear prediction method to NMR spectroscopy.
High-resolution fluorine-19 magnetic resonance of
solids. NMR determination of enantiomeric purity.
Solid-state NMR studies of molecular sieve
catalysis. Pulsed electron-nuclear double resonance
methodology. Multidimensional NMR and data processing.
One- and two-dimensional high-resolution
solid-state NMR studies of zeolite lattice structures.
Solid-state NMR studies of DNA structure and dynamics.
Spin-lattice relaxation of coupled nuclear

Vol. 14, No. 1-4 309
spins with applications to molecular motion in liquids.
Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 23 No. 2 (1991). Contents: Solvent
signal suppression in NMR.
Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 23 No. 3 (1991). Contents: Modern
methods of NMR data processing and data evaluation.
X H NMR magic angle spinning (MAS) studies
of heterogenous catalysis.
NMR - Basic Principles and Progress. Volume
23: Deuterium and Shift Calculation (1991).
Eds.: P. Diehl, E. Fluck, H. Giinther, R. Kosfeld,
J. Seelig. Contents: M.L. Martin, G.J. Martin,
Nantes, France: Deuterium NMR in the Study of
Site-Specific Natural Isotope Fractionation (SNIF-
NMR); H.-H. Limbach, Freiburg, FRG: Dynamic
NMR Spectroscopy in the Presence of Kinetic Hydrogen/Deuterium
Isotope Effects; W. Kutzelnigg,
U. Fleischer, M. Schindler, Bochum, FRG: The
IGLO-Method: Ab-initio Calculation and Interpretation
of NMR Chemical Shifts and Magnetic Susceptibilities.
Approx. 270 pp. 92 figs. 45 tabs.
ISBN 3-540-52949-7.
NMR - Basic Principles and Progress. Volume
24: High Pressure NMR (1991). Eds.: P. Diehl,
E. Fluck, H. Giinther, R. Kosfeld, J. Seelig, J.
Jonas, University of Illinois, Urbana, IL (Guest-
Ed.). Contents: D. Brinkmann, Zurich, Switzerland:
Solid-State NMR Studies at High Pressure;
K.O. Prins, Amsterdam, The Netherlands: High
Pressure NMR Investigations of Motion and Phase
Transitions in Molecular Systems; J. Jonas, Urbana,
IL: High Pressure NMR Studies of the Dynamics
in Liquids and Complex Systems; E.W. Lang, H.-
D. Liidemann, Regensburg, FRG: High Pressure
NMR Studies on Water and Aqueous Solutions;
J.W. Akitt, A.E. Merbach, Lausanne, Switzerland:
High Resolution Variable Pressure NMR for Chemical
Kinetics; H. Yamada, Kobe, Japan: Glass Cell
Method for High-Pressure, High-Resolution NMR
Measurements. Applications to the Studies of Pressure
Effects on Molecular Conformation and Structure.
Approx. 270 pp. 148 figs. 28 tabs. ISBN
3-540-52938-1.
NMR - Basic Principles and Progress. Volume
25: NMR at Very High Field (1991). Eds.: P. Diehl,
E. Fluck, H. Giinther, R. Kosfeld, J. Seelig, J.B.
Robert, CNRS, Grenoble,-France (Guest-Ed.). Contents:
R. Freeman, Cambridge, UK, J.B. Robert,
Grenoble, France: A Brief History of High Resolution
NMR; E.W. Bastiaan, C. MacLean, Amsterdam,
The Netherlands: Molecular Orientation
in High-Field High-Resolution NMR; D. Canet,
Vandoeuvre-les-Nancy, France, J.B. Robert, Grenoble,
France: Behaviour of the NMR Relaxation Parameters
at High Fields; D. Marion, Orl ans, France:
Structural Studies of Biomolecules at High Field; U.
Haeberlen, Heidelberg, FRG: Solid State NMR in
High and Very High Magnetic Fields. Approx. 175
pp. 44 figs. 10 tabs. ISBN 3-540-52946-2.
Modern NMR Techniques and Their Application
in Chemistry (Practical Spectroscopy Series Volume
11). Edited by Alexander I. Popov and Klaas Hallenga,
Marcel Dekker, Inc. (1991). ISBN 0-8247-
8332-8
Annual Reports of NMR Spectroscopy. Volume
23 (1991). Contents: NMR studies of isolated spin
pairs in the solid state. The oxidation-state dependence
of transition-metal shieldings. The Cinderella
nuclei. Permutation symmetry in NMR relaxation
and exchange. Nuclear spin relaxation in organic
systems and solutions of macromolecules and aggregations.
NMR of coals and coal products.
Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 23 No. 1 (1991). Contents:
Nuclear magnetic resonance imaging in the solid
state. Applications of three-and four-dimensional
heteronuclear NMR spectroscopy to protein structure
determination. Angiography and perfusion
measurements by NMR.
EPR Imaging and in vivo EPR (1991). Edited
by Gareth R. Eaton, Sandra S. Eaton, and Keiichi
Ohno, CRC Press, Boca Raton, FL. 320 pages,
$89.95, ISBN: 0-8493-4923-0.
Basic One-and Two-dimensional NMR Spectroscopy
by Horst Friebolin (1991). VCH, New York.
344 pages.

310
Advances in Magnetic and Optical Resonance
Volume 16 (1991). Contents: Laser excitation and
detection of magnetic resonance. Deuterium relaxation
in molecular solids. On the growth of multiple
spin coherences in NMR of solids.
1 Progress in Biophysics & Molecular Biology
Volume 56 No. 1 (1991). Contents: An evaluation
of computational strategies for use in the determination
of protein structure from distance constraints
obtained by nuclear magnetic resonance.
1 Radiospectroscopy of Natural Substances by B.
F. Alekseev, Y. V. Bogachev, V. Z. Drapkin, A. S.
Serdjuk, N. B. Strakhov and S. G. Fedin, Engl. Tr.
Norell Pr., New Jersey, 1991.
1 Electron Paramagnetic Resonance of Exchange
Coupled Systems by A. Bencini and D. Gatteschi,
Springer Verlag, Berlin, 1990.
Modern Pulsed and Continuous Wave Electron
Spin Resonance by L. Kevan and M. K. Bowman
(1990). Wiley, New York.
1 Transition Ion Electron Paramagnetic Resonance
by J. R. Pilbrow, Clarendon Press, Oxford,
1990.
1 Electron Paramagnetic Resonance of Exchange
Coupled Systems by A. Bencini and D. Gattechi
(1990). Springer, 287 pp. ISBN 0-387-50944-5
(hardcover) $83.00.
1 Isotope Effects in NMR Spectroscopy by S.
Berger, J. M. Risley, N. M. Sergeyev and
R. L. Van Etten (1990). Springer, 173 pp. ISBN
0-387-51286-1 (hardcover) $83.00.
17 O NMR Spectroscopy in Organic Chemistry
(1990). Edited by David W. Boykin. This book provides
a comprehensive review of the application of
17 O NMR spectroscopy to organic chemistry. Topics
include the theoretical aspects of chemical shift,
quadrupolar and J coupling; 17 O enrichment; the
effect of steric interactions on 17 O chemical shifts of
functional groups in flexible and rigid systems; the
additions to the list.
Bulletin of Magnetic Resonance
application of 17 O NMR spectroscopy to hydrogen
bonding investigations; mechanistic problems in organic
and bioorganic chemistry; and 17 O NMR spectroscopy
of oxygen monocoordinated to carbon in
alcohols, ethers, and derivatives. CRC Press, Inc.,
Florida. ISBN: 0-8493-4867-6.
Advances in Magnetic and Optical Resonance.
Volume 15 (1990). Contents: Iterative methods in
the design of pulse sequences for NMR excitation.
Electron-nuclear polarization transfer in the nuclear
rotating frame. Multipole NMR. Solid state and solution
NMR of nonclassical transition metal polyhydrides.
Low-frequency magnetic resonance with
a dc SQUID.
Advances in Biophysical Chemistry. Volume
1 (1990). Contents: Stable-isotope-assisted protein
NMR spectroscopy in solution.
31 P and
1
H two-dimensional NMR and NOESY-distance
restrained molecular dynamics methodologies for
defining sequence-specific variations in duplex
oligonucleotides: A comparison of NOESY two-spin
approximation and the relaxation matrix analyses.
NMR study of B- and Z-DNA hairpins of d[(CG)3]
in solution. Molecular dynamics simulations of carbohydrate
molecules. Diversity in the structure of
hemes.
Biological Magnetic Resonance. Volume 9
(1990). Contents: Phosphorus NMR of membranes.
Investigation of ribosomal 5S ribonucleic acid solution
structure and dynamics by means of highresolution
nuclear magnetic resonance spectroscopy.
Structure determination via complete relaxation
matrix analysis (CORMA) of two-dimensional nuclear
overhauser effect spectra: DNA fragments.
Methods of proton resonance assignment for proteins.
Solid-state NMR spectroscopy of proteins.
Methods for suppression of the H2O signal in proton
FT/NMR spectroscopy: A review.
Progress in Nuclear Magnetic Resonance Spectroscopy.
Vol. 25 pt. 5 (1990). Contents: Solid
state NMR techniques for the study of surface phenomena.
A primer on isotopic labeling in NMR investigations
of biopolymers. Vanadium-51 NMR.

Vol. 14, No. 1-4 311
One-dimensional and Two-dimensional NMR
Spectra by Modern Pulse Techniques. Koji Nakanishi.
(1990). University Science Books, Mill Valley,
CA. 234 p.
Annual Reports on NMR Spectroscopy. Volume
22 (1990). Contents: Metal-ion NMR studies of ion
binding. NMR studies of ligand-macromolecule interactions.
Applications of NMR in the analysis of
agrochemicals and pesticides. NMR nuclear shielding
and the electronic structures of macromolecules.
207 Pb-NMR parameters. Nuclear spin relaxation in
diamagnetic fluids part 1. General aspects and inorganic
applications.
Fourier Transforms in NMR, Optical, and Mass
Spectrometry: A User's Handbook. By A. G. Marshall
and F. R. Verdun (Ohio State University). Elsevier:
Amsterdam and New York. 1990. xvi + 450
pp. $107.25. ISBN 0-444-87360-0.
Progress in Nuclear Magnetic Resonance Spectroscopy.
Volume 22, pt. 1 (1990). Contents: Scaling
in one and two dimensional NMR spectroscopy
in liquids. Oligosaccharide conformations: Application
of NMR and energy calculations. Relaxation
matrix analysis of 2D NMR data.
Progress in Magnetic Resonance Spectroscopy.
Volume 22, Part 3 (1990). Contents: NMR parameters
of alkynes. Improved methods for quantitative
spectral analysis of NMR data.
Advances in Magnetic Resonance. Volume 14
(1990). Contents: Measurement of dipole-dipole
cross correlation by triple-quantum filtered twodimensional
exchange spectroscopy. Assessment
and optimization of pulse sequences for homonuclear
isotropic mixing. Spin-1/2 description of spins 3/2.
Optical pumping measurements of nuclear cross relaxation
and electrix doublets.
Quarterly Review of Biophysics. Volume 23
(Number 1) February 1990. Contents: Biosynthetic
incorporation of 15 N and 13 C for assignment and interpretation
of nuclear magnetic resonance spectra
of proteins. Heteronuclear niters in two-dimensional
[1H, 1H]- NMR spectroscopy: combined use with
isotope labelling for studies of macromolecular con-
formation and intermolecular interactions.
Quarterly Reviews of Biophysics. Volume 23
(Number 2) May 1990. Contents: Heteronuclear
three-dimensional NMR spectroscopy of isotopically
labelled bioilogical macromolecules. Deuterium labelling
in NMR structural analysis of larger proteins.
Use of deuterium labelling in NMR studies of
antibody combining site structure.
Principles of Nuclear Magnetic Resonance in
One and Two Dimensions. Richard R. Ernst and
Geoffrey Bodenhausen. Oxford University Press.
1990. 640 pp. paper $39.95
A Dictionary of Concepts in NMR. S.W.
Homans. Oxford University Press. 1990. 352 pp."
$80.00
Nuclear Magnetic Resonance: Principles and
Theory. Ryozo Kitamaru. Elsevier, New York,
1990.
Quantum Description of High-Resolution NMR
in Liquids. Maurice Goldman. Oxford University
Press. 1990. 288 pp. $65.00
Modern Pulsed and Continuous-wave Electron
Spin Resonance. Edited by Larry Kevan and
Michael K. Bowman. Wiley, New York, 1990. 440
p.
Principles of Magnetic Resonance, Second Ed.
by C. P. Slichter, Springer, New York, 1990. 655 p.
Soviet Scientific Reviews Section B: Chemistry
Reviews. Volume 14, Part 2 (1990). Contents:
Pulsed NMR study of molecular motion in solids.
Progress in Nuclear Magnetic Resonance Spectroscopy,
Volume 22 No. 6 1990. Contents: Fieldcycling
relaxometry of protein solutions and tissue.
Implications for MRI. Solid state NMR studies of
local motions in polymers.
Nuclear Magnetic Resonance, Volume 20 1989/
1990. Contents: NMR books and reviews. Theoretical
and physical aspects of nuclear shielding. Applications
of nuclear shielding. Theoretical aspects

312
of spin-spin couplings. Applications of spin-spin
couplings. Nuclear spin relaxation in liquids and
gases. Solid state NMR Multiple pulse NMR Natural
macromolecules. Synthetic macromolecules.
Conformational analysis. Nuclear magnetic resonance
spectroscopy of living systems. Nuclear magnetic
resonance imaging of living systems. NMR of
paramagnetic species. NMR of liquid crystals and
micellar solutions.
1 Spin Labeling: Theory and Applications.
Edited by L. J. Berliner and J. Reuben, Academic
Press, New York, Vol. 3, 1989.
1 Advanced EPR: Applications in Biology and
Biochemistry. Edited by A. J. Hoff, Elsevier, Amsterdam,
1989.
1 Pulsed EPR: A New Field of Applications.
Edited by C. P. Keijzers, E. J. Reijerse and J.
Schmidt, North Holland, Amsterdam, 1989.
1 Electron Spin Resonance, Specialist Periodical
Report, Vol. 1 IB, Royal Chemical Society, London,
1989.
Nuclear Magnetic Resonance: Structure and
Mechanism. Edited by Norman J. Oppenheimer and
Thomas L. James, Academic Press, New York, 1989.
507 p. (Methods in Enzymology).
NMR Spectroscopy and Polymer Micro structure.
The Conformation Connection. Alan E. Tonelli.
VCH, New York, 1989. x 252 pp., illus. $69.50.
Methods in Stereochemical Analysis.
Annual Reports on NMR Spectroscopy, Vol. 21.
Edited by G. A. Webb, Academic Press, London,
1989. ISBN: 0-12-505321-5.
EPR of Exchange-Coupled Systems. Alessandro
Bencini and Dante Gatteschi. Springer-Verlag,
Berlin, 1989. 287 pages. ISBN: 0-387-50944-5.
Nuclear Magnetic Resonance, Vol. 18, Specialist
Periodical Reports, G. A. Webb, Senior Reporter,
Royal Society of Chemistry, London, 1989. 511
x New additions to the list.
pages. ISBN: 0-85186-412-0.
Bulletin of Magnetic Resonance
Advances in Magnetic Resonance Imaging.
Edited by Ephraim Feig, IBM Research Division,
Thomas J. Watson Research Center. Ablex Publishing
Corporation. 1989. 272 pp. $55.00
Advances in Magnetic Resonance. Volume 13
(1989). Contents: Single crystal nuclear magnetic
resonance studies of high temperature superconductors.
Deuterium nuclear magnetic resonance and
molecular dynamics in alkane/urea inclusion compounds.
1 H nuclear magnetic resonance imaging of
solids with magic-angle spinning. Two-dimensional
nuclear magnetic resonance experiments for studying
molecular order and dynamic in static and rotating
solids. Electrophoretic nuclear magnetic resonance
experiments for studying molecular order
and dynamics in static and rotating solids. Electrophoretic
nuclear magnetic resonance. Ultraslow
atomic motion by site-selective excitation of highly
resolved nuclear magnetic resonance lines in dilute
spin systems. Two-dimensional hybrid experiments
for the measurement of small anisotropies in magicangle
spinning nuclear magnetic resonance.
Analytical NMR. Edited by L. D. Field and S.
Sternhell. Wiley, New York, 1989, 250 p.
Modern NMR spectroscopy. A workbook of
chemical problems by Jeremy K.M. Sanders, Edwin
C. Constable and Brian K. Hunter. Oxford University
Press, New York, 1989, 1.18 p.
Introduction to Pulse NMR Spectroscopy, Second
Edition., Thomas C. Farrar. The Farragut
Press, Madison, Wisconsin 53705, 1989. 211 pages.
$39.95 (hard cover); $24.95 (paperback).
Nuclear Magnetic Resonance. Volume 19
(1988/1989) Contents: Theoretical and physical aspects
of nuclear shielding. Applications of nuclear
shielding. Theoretical aspects of spin-spin couplings.
Applications of spin-spin coupling. Nuclear
spin relaxation in liquids. Solid state NMR Multiple
pulse NMR Conformational analysis. Nuclear
magnetic resonance of living systems. Oriented molecules.
Heterogeneous systems.

Vol. 14, No. 1-4 313
1 Electron Nuclear Double Resonance Spectroscopy
of Radicals in Solution, by H. Kurreck, B.
Kirste and W. Lubitz. VCH Publishers, New York,
1988.
Quantum Description of High-Resolution NMR
in Liquids. Maurice Goldman. Clarendon Press,
Oxford University Press, New York, 1988.
Electron Spin Resonance. Volume 11B (1988).
Contents: In vivo detection of free radical metabolites
by spin trapping. Theoretical aspects of ESR
Transition metal ions. Recent developments of EN-
DOR spectroscopy in the study of defects in solids.
Inorganic and organometallic radicals and clusters
prepared in a rotating cryostat by metal vapour
techniques. Inorganic and organometallic radicals.
Metalloproteins. Complexes of paramagnetic metals
with paramagnetic ligands.
Nuclear Magnetic Resonance
Spectroscopy. Frank A. Bovey, Lynn Jelinski and
Peter A. Mirau. Academic, New York, 1988. 653 p.
Coherence and NMR, Michael Munowitz. Wiley,
New York, 1988. 289 pages. $39.95.
Biomedical Magnetic Resonance Imaging, Principles,
Methodology, and Applications. Edited by
Felix W. Wehrli, Derek Shaw, and J. Bruce Kneeland.
VCH Publishers, New York, 1988. 601 pages.
$95.00.
Interpretation of Carbon-13 NMR Spectra, F. W.
Wehrli, A. P. Marchand, and S. Wehrli. Wiley, New
York, 1988. 484 pages. $89.95. ISBN 0-471-91742-
7.
Introduction to NMR Spectroscopy. By R. J.
Abraham (University of Liverpool) et al. John Wiley
and Sons: Chicester and New York. 1988. xiii
+ 271 pages. $44.95. ISBN 0-471-91893-8.
Nuclear Magnetic Resonance. Volume 18
(1987/1988) Contents: Theoretical and physical
aspects of nuclear shielding. Theoretical aspects
of spin-spin couplings. Applications of spin-spin
1 New additions to the list.
coupling. Nuclear spin relaxation in liquids and
gases. Solid state NMR Multiple pulse NMR Natural
macromolecules. Synthesis macromolecules.
Conformational analysis. Nuclear magnetic resonance
of living systems. NMR of paramagnetic
species. NMR of liquid crystals and micellar solutions.
The Nuclear Overhauser Effect in Stereochemical
and Conformational Analysis. David Neuhaus
and Michael Williamson. Contents: Integrated account
of the theory, experimental practice and applications
of the nuclear Overhauser effect (NOE). Describes
such recent developments as NOE difference
spectroscopy, NOESY, heteronuclear NOE, and rotating
frame NOE. Covers experimental design in
depth and the underlying theory of NOE. Assumes
a graduate level knowledge of NMR spectroscopy.
496 pp. 1987. $95.00. .
Modern NMR Spectroscopy: A Guide for
Chemists. Jeremy K.M. Sanders and Brian K.
Hunter. Oxford University Press. 1987.
Stereochemical Analysis of Alicyclic Compounds
by C-13 NMR Spectroscopy, James K. Whitesell and
M. A. Minton. Chapman & Hall, London, 1987. 231
pages. $60.00.
Frequency Synthesizers, Theory and Design, 3rd
Edition, Vadim Manassewitsch. Wiley, New York,
1987. 608 pages. $52.95.'
Carbon-13 NMR Spectroscopy. High-Resolution
Methods and Applications in Organic Chemistry
and Biochemistry, 3rd Edition, Eberhard Breitmaier
and Wolfang Voelter. VCH Publishers,
New York/Weinheim, Germany, 1987. 515 pages.
$135.00. ISBN 0-89573-493-1 (3-527-26466-3 at
Weinheim).
Two-dimensional NMR Spectroscopy: Applications
for Chemists and Biochemists. Edited by
William R. Croasmun and Robert M. K. Carlson.
VCH Publishers: New York and Weinheim. 1987
xx + 511 pages. $95.00 ISBN 0-89573-308-0.

314
Instructions for Authors
Because of the ever increasing difficulty of keeping
up with the literature there is a growing need for
critical, balanced reviews covering well-defined areas
of magnetic resonance. To be useful these must
be written at a level that can be comprehended by
workers in related fields, although it is not the intention
thereby to restrict the depth of the review.
In order to reduce the amount of time authors must
spend in writing we will encourage short, concise
reviews, the main object of which is to inform nonexperts
about recent developments in interesting aspects
of magnetic resonance.
The editor and members of the editorial board
invite reviews from authorities on subjects of current
interest. Unsolicited reviews may also be accepted,
but prospective authors are requested to
contact the editor prior to writing in order to avoid
duplication of effort. Reviews will be subject to critical
scrutiny by experts in the field and must be
submitted in English. Manuscripts should be sent
to the editor, Dr. David G. Gorenstein, Chemistry
Department, Purdue University, West Lafayette, Indiana
47906, USA/(317) 494 7851. Fax No. 317 494
0239.
MANUSCRIPTS must be submitted in triplicate
(one copy should be the original), on approximately
22x28 cm paper, type-written on one side
of the paper, and double spaced throughout. If the
manuscript cannot be submitted on computer tapes,
floppy disks, or electronically (see below), please
type with a carbon ribbon using either courier 10
or 12, gothic 12, or prestige elite type face with 10
or 12 pitch. All pages are to be numbered consecutively,
including references, tables, and captions to
figures, which are to be placed at the end of the
review.
ARRANGEMENT: Considerable thought
should be given to a logical ordering of the subject
matter and the review should be divided into
appropriate major sections, and subsections, using
Roman numerals, capital letters, and Arabic numerals
respectively. A table of contents should be included.
TABLES: These are to be numbered consecutively
in the text with Arabic numerals. Their place
of insertion should be mentioned in the text, but
they are to be placed in order at the end of the
Bulletin of Magnetic Resonance
paper, each typed on a separate sheet. Each table
should be supplied with a title. Footnotes to tables
should be placed consecutively, using lower case letters
as superscripts.
FIGURES are also to be numbered consecutively
using Arabic numerals and the place of insertion
mentioned in the manuscript. The figures are
to be grouped in order at the end of the text and
should be clearly marked along the edge or on the
back with figure number and authors' names. Each
figure should bear a caption, and these should be
arranged in order and placed at the end of the text.
Figures should be carefully prepared in black ink
to draftsman's standards with proper care to lettering
(typewritten or freehand lettering is not acceptable).
Graphs should include numerical scales
and units on both axes, and all figures and lettering
should be large enough to be legible after reduction
by 50-60%. Figures should be generally placed on
sheets of the same size as the typescript and larger
originals may be handled by supplying high-contrast
photographic reductions. One set of original figures
must be supplied; reproduction cannot be made
from photocopies. Two additional copies of each figure
are required. Complex molecular formula should
be supplied as ink drawings.
. REFERENCES to the literature should be
cited in order of appearance in the text by numbers
on the line, in parentheses. The reference list is to be
placed on separate sheets in numerical order at the
end of the paper. References to journals should follow
the order: author's (or authors') initials, name,
name of journal, volume number, page, and year of
publication. The abbreviation of the journal follows
that used in the Chemical Abstracts Service Source
Index. References to books should include in order:
author's (or authors') initials, name, title of book,
volume, edition if other than the first, publisher,
address, date of publication, and pages.
FOOTNOTES should be used sparingly and
only in those cases in which the insertion of the
information in the text would break the train of
thought. Their position in the text should be
marked with a superscript Arabic numeral and the
footnotes should be typed at the bottom of the relevant
page in the text, separated from the latter by
a line.
SYMBOLS AND ABBREVIATIONS:
Mathematical symbols should be typewritten wher-

Vol. 14, No. 1-4 315
ever possible. Greek letters should be identified in
pencil in the margin. In reviews containing a number
of mathematical equations and symbols, the author
is urged to supply a list of these on a separate
sheet for the assistance of the printer; this will not
appear in print. Standard abbreviations will follow
the American Chemical Society's HANDBOOK
FOR AUTHORS names and symbols for units.
PERMISSIONS: It is the responsibility of the
author to obtain all permissions concerned with the
reproduction of figures, tables, etc, from copyrighted
publications. Written permission must be obtained
from the publisher (not the author or editor) of the
journal or book. The publication from which the
figure or table is taken must be referred to in the
reference list and due acknowledgement made, e.g.
reprinted by permission from ref. (00).
REPRINTS: Thirty reprints of a review will
be supplied free to its senior author and additional
reprints may be purchased in lots of 100.
INSTRUCTIONS FOR SUBMITTING
MANUSCRIPTS ON FLOPPY DISKS
OR ELECTRONICALLY: If you have used a
word processor to type your manuscript, please forward
your manuscript after review and revision, in a
computer readable form. Tape should be unlabeled,
in a standard IBM format, 1600 BPI, 80x80 blocks
and ASCII image. Floppy disks or tapes readable on
IBM, DEC, or Macintosh computers are also acceptable.
For direct submission over the phone lines use
300 - 2400 baud ASCII transmission. Submission
via INTERNET (address David@chem.purdue.edu)
is also recommended. Additional details will be provided
for the appropriate hand-shake requirements.
Please supply us with the code for interpreting superscripts,
greeks, etc. on your word processor.

INTERNATIONAL SOCIETY OF MAGNETIC RESONANCE
(ISMAR)
ISMAR MEMBERSHIP APPLICATION
Name and Address (Telephone & Fax numbers, e-mail) of ISMAR Member
Membership Fee for 1991, $20.00 $.
Membership Fee for 1992, $20.00 $
Membership Fee for 1993, $20.00 $
Membership Fee for 1994, $20.00 $
If this is a new membership application, please check Q
If you wish to be registered as a member of the Division of Biology and Medicine,
in addition to ISMAR itself, at no extra charge, please check box. D
BULLETIN OF MAGNETIC RESONANCE
The Quarterly Review Journal of the
International Society of Magnetic Resonance
Telephone: (317)494-7851 DAVID G. GORENSTEIN, Editor
FAX: (317)494-0239 Department of Chemistry
Internet: david@adam.chem. Purdue University
purdue.edu W. Lafayette, IN, U.SA.47907
Subscription to Bulletin of Magnetic Resonance: ISMAR Member Rate
Volume 13 (1991), $25.00 $
Volume 14 (1992), $25.00 $
Volume 15 (1993), $25.00 $
Volume 16 (1994), $25.00 $
Subscription to Bulletin of Magnetic Resonance: Library & NON-ISMAR Member Rate
Volume 13 (1991), $80.00 $
Volume 14 (1992), $80.00 $
Volume 15 (1993), $80.00 $
Volume 16 (1994), $80.00 $
TOTAL $
Send this completed form with your check made payable to:
International Society of Magnetic Resonance
Professor Regitze R. Void, Treasurer
Department of Chemistry, 0342
University of California, San Diego
9500 Gilman Drive
La Jolla,CA 92093-0342
(619) 534-0200; Fax (619) 534-7042
Bitnet: rrvold@ucsd.bitnet