You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
BULLETIN OF MAGNETIC RESONANCE<br />
The Quarterly Review Journal of the<br />
International Sodiety of Magnetic Resonance<br />
VOLUME 14 October 1992 NUMBERS 1-4<br />
,\ Proceedings<br />
*' International<br />
Society of ;<br />
; Magnetic<br />
Resonance<br />
Xlth,<br />
- .Meeting<br />
•>• 4<br />
1- a<br />
VF:<br />
]\n\ 19 :i<br />
\ \NC()U\ TR<br />
BC,C\N\O\
BULLETIN OF MAGNETIC RESONANCE<br />
The Quarterly Review Journal of the<br />
International Society of Magnetic Resonance<br />
Editor:<br />
DAVID G. GORENSTEIN<br />
Department of Chemistry<br />
Purdue University<br />
West Lafayette, IN 47907 USA.<br />
Fax: 317-494-0230<br />
INTERNET:david@chem.purdue.edu<br />
Editorial Board:<br />
E.R.ANDREW LAWRENCE BERLINER ROBERT BLINC<br />
University of Florida Ohio State University E. Kardelj University of Ljubljana<br />
Gainesville, Florida, U.S.A. Columbus, Ohio, U.S.A. Ljubljana, Yugoslavia<br />
H.CHIHARA GARETH R. EATON DANIEL FIAT<br />
Osaka University University of Denver University of Illinois at Chicago<br />
Toyonaka, Japan Denver, Colorado, U.S.A. Chicago, Illinois, U.S.A.<br />
SHIZUO FUJIWARA DAVID GRANT ALEXANDER PINES<br />
University of Tokyo University of Utah University of California<br />
Bunkyo-Ku, Tokyo, Japan Salt Lake City, Utah, U.S.A. Berkeley, California, U.S.A.<br />
MIK PINTAR CHARLES P. POOLE, JR. BRIAN SYKES<br />
University of Waterloo University of South Carolina University of Alberta<br />
Waterloo, Ontario, Canada Columbia, South Carolina, U.S.A. Edmonton, Alberta, Canada<br />
The Bulletin of Magnetic Resonance is a quarterly review journal by the International Society of<br />
Magnetic Resonance. Reviews cover all parts of the broad field of magnetic resonance, viz.. the<br />
theory and practice of nuclear magnetic resonance, electron paramagnetic resonance, and nuclear<br />
quadrupole resonance spectroscopy including applications in physics, chemistry, biology, and<br />
medicine. The BULLETIN also acts as a house journal for the International Society of Magnetic<br />
Resonance.<br />
CODEN: BUMRDT ISSN: 0163-559X<br />
Bulletin of Magnetic Resonance, The Quarterly Journal of International Society of Magnetic<br />
Resonance. 1992 copyright by.the International Society of Magnetic Resonance. Rates: Libraries<br />
and non-<strong>ISMAR</strong> members $80.00, members of JSMAR, $25.00. All subscriptions are for a volume<br />
year. All rights reserved. No part of this journal may be reproduced in any form for any purpose or by<br />
any means, abstracted, or entered into any data base, electronic or otherwise, without specific<br />
permission in writing from the publisher.
J. ANGLISTER<br />
Israel<br />
R. BLINC<br />
Yugoslavia<br />
P.T. CALLAGHAN<br />
New Zealand<br />
L.G.CONTI<br />
Italy<br />
E.L.HAHN<br />
U.S.A.<br />
M.J.R. HOCH<br />
5. Africa<br />
P.C. LAUTERBUR<br />
U.S.A.<br />
M.MEHRING<br />
Germany<br />
C.P: POOLE<br />
U.S.A.<br />
G.C.K. ROBERTS<br />
England<br />
N.V. VUGMAN<br />
Brazil<br />
C.S. YANNONI<br />
U.S.A.<br />
Council of the International Society of Magnetic Resonance<br />
President: R. FREEMAN, England<br />
Vice-President: A. PINES, U.S.A.<br />
Founding Chairman: D. FIAT, U.S.A.<br />
Secretary-General: R.K. HARRIS, England<br />
Treasurer: R.R. VOLD, U.S.A.<br />
Past President: C.P. SLICHTER, U.S.A.<br />
E.D. BECKER<br />
U.S.A.<br />
G. BODENHAUSEN<br />
Switzerland<br />
H.CHIHARA<br />
Japan<br />
R. DESLAURIERS<br />
Canada<br />
KM. HAUSSER<br />
Germany<br />
C.L. KHETRAPAL<br />
India<br />
E. LIPPMAA<br />
Estonia<br />
H. PFEIFER<br />
Germany<br />
M. PUNKINEN<br />
Finland<br />
P. SERVOZ-GAVESf<br />
France<br />
J.S. WAUGH<br />
U.S.A.<br />
MR. BENDALL<br />
Australia<br />
W.S. BREY<br />
U.S.A.<br />
S. CLOUGH<br />
England<br />
R.R. ERNST<br />
Switzerland<br />
J.W. HENNEL<br />
Poland<br />
VJ. KOWALEWSKI<br />
Argentina<br />
B. MARAVIGLIA<br />
Italy<br />
M.M. PINTAR<br />
Canada<br />
L.W. REEVES<br />
Canada<br />
J. STANKOWSKI<br />
Poland<br />
K. WUTHRICH<br />
Switzerland<br />
The aims of the International Society of Magnetic Resonance are to advance and diffuse knowledge<br />
of magnetic resonance and its applications in physics, chemistry, biology, and medicine, and to<br />
encourage and develop international contacts between scientists.<br />
The Society sponsors international meetings and schools in magnetic resonance and its applications<br />
and publishes the quarterly review journal. The Bulletin of Magnetic Resonance, the house journal of<br />
<strong>ISMAR</strong>.<br />
The annual fee for <strong>ISMAR</strong> membership is $20 plus $25 for a member subscription to the Bulletin of<br />
Magnetic Resonance.<br />
Send subscription to: International Society of Magnetic Resonance<br />
Professor Regitze R. Void, Treasurer<br />
Department of Chemistry, 0342<br />
University of California, San Diego<br />
9500 Gilman Drive<br />
La Jolla, CA 92093-0342<br />
(619) 534-0200; FAX (619) 534-7042<br />
Bitnet: rrvold@ucsd.bitnet
Vol. 14, No. 1-4<br />
ORGANIZING COMMITTEE<br />
for<br />
<strong>ISMAR</strong> 92<br />
C. Fyfe, Chairman<br />
Department of Chemistry<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA<br />
R. Andersen<br />
Department of Chemistry<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA<br />
G. S. Bates, Treasurer<br />
Department of Chemistry<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA<br />
M. Bloom<br />
Department of Physics<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA<br />
E. E. Burnell<br />
Department of Chemistry<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA<br />
G. Drobny<br />
Department of Chemistry<br />
University of Washington<br />
Seattle, Washington<br />
USA<br />
I. Gay<br />
Department of Chemistry<br />
Simon Fraser University<br />
Burnaby, British Columbia<br />
CANADA<br />
F. G. Herring<br />
Department of Chemistry<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA<br />
CO-EDITORS OF THE PROCEEDINGS<br />
D. G. Gorenstein<br />
Department of Chemistry<br />
Purdue University<br />
West Lafayette, Indiana<br />
USA<br />
C. Fyfe<br />
Department of Chemistry<br />
University of British Columbia<br />
Vancouver, British Columbia<br />
CANADA
4 Bulletin of Magnetic Resonance<br />
ACKNOWLEDGMENTS<br />
The Organizing Committee of the Xlth conference of the International Society of Magnetic Resonance<br />
gratefully acknowledges the financial support of the following organizations:<br />
B.P. Research, pic<br />
Bio-Mega, Inc.<br />
Bio Rad - Sadtler Division<br />
Bruker Spectrospin Canada Ltd.<br />
Bulletin of Magnetic Resonance<br />
Chemagnetics/Otsuka Electronics<br />
Dean of Graduate Studies, U.B.C.<br />
Dean of Science, U.B.C.<br />
Doty Scientific, Inc.<br />
Dow Chemical Co.<br />
ICON (Isotope) Services Inc.<br />
JEOL USA, Inc.<br />
John Wiley & Sons<br />
Molecular Simulations<br />
Oxford Instruments<br />
Pfizer<br />
Quadra Logic Technologies, Inc.<br />
Varian NMR Instruments<br />
Special thanks are extended to the National Sciences and Engineering Research Council of Canada (NSERC)<br />
for its support through the award of a Conference Grant to <strong>ISMAR</strong> 92. We are also grateful to the Department<br />
of Chemistry at the University of British Columbia for its support.
Vol. 14, No. 1-4 5<br />
Contents<br />
Proceedings of the Xlth Meeting of the<br />
International Society of Magnetic Resonance<br />
July 19 - 24, 1992<br />
Vancouver, B. C. Canada<br />
Solid-State Polarization-Transfer Experiments Involving Quadrupolar Nuclei,<br />
K. T. Mueller, C. A. Fyfe, H. Grondey, K. C. Wong-Moon and T. Markus 9<br />
Magnetic Resonance Evidence for Superconductivity in a Semimetal, I. P. Goudemond,<br />
G. J. Hill and M. J. R. Hoch 12<br />
NMR Spectroscopy in Cardiac Surgery, R. Deslauriers, S. Lareau, R. S. Labow,<br />
W. J. Keon, G-H. Tian, A. L. Panos, C. A. M. Barrozo, I. S. Ali, O. Al-Nowaiser and T. A. Salerno 15<br />
Topology and Spin Alignment in Organic High-Spin Molecules, Y. Teki, K. Sato,<br />
M. Okamoto, A. Yamashita, Y. Yamaguchi, T. Takui, T. Kinoshita and K. Itoh 24<br />
New Developments in Pulsed Electron Paramagnetic Resonance: Relaxation Mechanisms<br />
of Nitroxide Spin Labels, C. Mailer, B. H. Robinson and D. A. Haas 30<br />
New Developments in Pulsed Electron Paramagnetic Resonance: Direct Measurement<br />
of Rotational Correlation Times from Decay Curves, D. A. Haas, C. Mailer,<br />
T. Sugano and B. H. Robinson 35<br />
Non-Linear Effects in Standard 2D NOE Experiments in Coupled Spin Systems,<br />
R. C. R. Grace and A. Kumar 42<br />
Deriving Structures from 2D NMR. A Method for Denning the Conformation of<br />
a Protein Adsorbed to Surfaces, D. A. Keire and D. G. Gorenstein 57<br />
3D-Structure Determination of Flavoridin in Solution: New Computational Strategy<br />
for Disulfide-Bridge Mapping, H. Senn, W. Klaus and P. Gerber 64<br />
NMR Approaches to Large Proteins: trp Repressor and Chloramphenicol Acetyltransferase,<br />
L.-Y. Lian, J. P. Derrick, V. Ramesh, R. O. Frederick, S. E. H. Syed and<br />
G. C. K. Roberts 65<br />
2D NMR Study of Drug-Protein Interactions: Ethidium Bromide - Neocarzinostatin<br />
Complex, S. Mohanty, L. C. Sieker and.G. P. Drobny 68<br />
Quantitative Analysis in Multi-Dimensional Transferred NOE Experiments: Improved<br />
Spectral Acquisition and Processing, F. Ni 75<br />
Time-Resolved Solid-State NMR: Small Molecules and Enzymes in Rapidly Frozen<br />
Solution, J. N. S. Evans, R. J. Appleyard and W. Shuttleworth 81<br />
Coupled Methyl Groups in Dimethyl Sulphide, M. R. Johnson, S. Clough, A. J. Horsewill<br />
and I. B. I. Tomsah 86
Bulletin of Magnetic Resonance<br />
NMR Relaxation Studies of Microdynamics in Chloroaluminate Melts, P. A. Shaw,<br />
W. R. Carper, C. E. Keller and J. S. Wilkes .92<br />
Structure and Dynamics of a Membrane Bound Polypeptide, T. A. Cross,<br />
R. R. Ketchem, W. Hu, K.-C. Lee, N. D. Lazo and C. L. North 96<br />
The Role of Metal Ions in Processes of Conformational Selection during Ligand-<br />
Macromolecule Interactions, E. Gaggelli, N. Gaggelli, G. Valensin and A. Maccotta .... 102<br />
Detection and Characterization of CFC, HCFC and HFC Gases in Foamed Insulation<br />
by High Field NMR Imaging, L. H. Randall, C. A. Fyfe, Z. Mei and S. Whitworth 108<br />
Mysterious Negative Peaks in the 1 H{ 1 H}NOE Difference Spectra of Some Thiopyran<br />
Compounds, C. Szantay, Jr 112<br />
H-l and C-13 NMR Spectra of the Carbanions Produced from Phenylpropene<br />
Derivatives, A. Yoshino, K. Aoki, M. Ushio and K. Takahashi 116<br />
Intracellular pH and Inorganic Phosphate Effects on Skeletal Muscle Force,<br />
E. R. Barton-Davis, R. W. Wiseman and M. J. Kushmerick 122<br />
A Simple Model for the Influence of Motion on the NMR Line Shape, M. Goldman,<br />
T. Tabti, C. Fermon, J. F. Jacquinot and G. Saux 126<br />
The Effect on Ti of Correlated Water Motions in the Polar Phase of Colemanite,<br />
J. Sun and A. Watton 131<br />
Measurement of Deuteron Spin Relaxation Times in Liquid Crystals by a Broadband<br />
Excitation Sequence, R. Y. Dong 134<br />
Carbon-13 Relaxation Mechanisms and Motional Studies in Selected Halomethane<br />
Molecules, A. A. Rodriguez, T. Davis and L. E. Nance 139<br />
An Efficient Large Sample Volume System for Solid State NMR, R. J. Pugmire,<br />
Y. J. Jiang, M. S. Solum and D. M. Grant . . . . 144<br />
Magnetic Resonance Spectroscopic Investigations of Poly(p-Phenylene Sulfide/Disulfide),<br />
PPS/DS, D. W. Lowman and D. R. Fagerburg 148<br />
Application of 2-D HETCOR NMR to Investigate Polymer Blend Heterogeneity,<br />
S. Kaplan 153<br />
New High Resolution NMR Studies in Polycrystalline Tetracyanoquinodimethane,<br />
M. T. Nunes, A. Vainrub, M. Ribet, F. Rachdi, P. Bernier, M. Almeida and G. Feio 156<br />
Use of NMR Relaxation Measurements to Derive the Binding Site of Plastocyanin<br />
in Complexes With Cytochrome-F and C, S. Modi, E. McLaughlin, D. S. Bendall, S. He<br />
and J. C. Gray 159<br />
Metal-peptide Interaction: Influence of the Aminoacid Sequence on the Binding<br />
of Co(II) to Glycyltryptophan and Tryptophylglycine Studied by X H NMR and<br />
Fluorescence, A. Spisni, G. Sartor, L. Franzoni, A. Orsolini, P. Cavatorta and M. Tabak . . . 165<br />
Assignments of the X H NMR Spectrum of a Consensus DN A-Binding Peptide from<br />
the HMG-I Protein, J. N. S. Evans, M. S. Nissen and R. Reeves 171<br />
Solution Structure of the DNA-binding Domain of GAL4 from Saccharomyces cerevisiae,<br />
J. D. Baleja, V. Thanabal, T. Mau and G. Wagner . 175
Vol. 14, No. 1-4 7<br />
Structural Investigation of Folic Acid by NMR Proton Relaxation and Molecular<br />
Mechanics Analysis, C. Rossi, A. Donati, S. Ulgiati, and M. R. Sansoni 181<br />
Characterization of Water-in-Bitumen Emulsions in Model Porous Media by NMR<br />
Microscopic Imaging Techniques, L. H. Randall and G. E. Sedgwick and C. A. Fyfe .... 186<br />
Computer Graphics for Pulse Sequence Analysis, J. Callahan, D. Mattiello and<br />
G. P. Drobny 191<br />
NMR Investigation of the Simultaneous Fermentation of Xylose and Glucose by<br />
a Selected Strain of Klebsiella Planticola (Gil), C. Rossi, A. Lepri, M. P. Picchi, S. Bastianoni,<br />
D. Medaglini, M. Vanassina and E. Cresta 197<br />
Interleukin-1 Receptor Antagonist Protein: Solution Secondary Structure from<br />
NOE's and l Ha and 13 Ca Chemical Shifts, B. J. Stockman, T. A. Scahill, A. Euvrard,<br />
N. A. Strakalaitis, D. P. Brunner, A. W. Yem and M. R. Deibel, Jr. 202<br />
Green's Function Calculation of Effective Nuclear Relaxation Times in Metals and<br />
Disordered Metals, M. Martin-Landrove and J. A. Moreno 208<br />
Stochastic Averaging Revisited, D. H. Jones, N. D. Kurur and D. P. Weitekamp 214<br />
Magnetic Resonance of Trapped Ions by Spin-Dependent Cyclotron Acceleration,<br />
P. J. Pizarro and D. P. Weitekamp . . 220<br />
Coordination Modes of Histidine Moiety in Copper (II) Dipeptide Complexes Detected<br />
by Multifrequency ESR, R. Basosi, R. Pogni, and G. D. Lunga 224<br />
An EPR and ab initio Study of a Phosphaalkene Radical Anion, and Comparison<br />
with other Phosphorus-Containing Radical Ions, M. Geoffroy, G. Terron, A. Jouaiti,<br />
P. Tordo and Y. Ellinger 229<br />
Conformational Substate Distribution in Myoglobin as Studied by EPR Spectroscopy,<br />
A. R. Bizzarri and S. Cannistraro . 234<br />
Effect of Paramagnetic Ions in Aqueous Solution for Precision Measurement of the<br />
Proton Gyromagnetic Ratio, A. R. Lim, C. S. Kim and S. H. Choh 240<br />
Magnetic Resonance of 23 Na and 14 N Nuclei in Single and Multi-Domain Crystals<br />
of Ferroelectric NaNO2, S. H. Choh and K. T. Han 246<br />
Knight Shifts and Spin Dynamics in Disordered Systems, M. J. R. Hoch and<br />
S. T. Stoddart .252<br />
Numerical Design and Evaluation of Broadband Pulse Sequences for 1=1 Spin<br />
Systems, D. L. Mattiello, J. Callahan, T. Alam and G. P. Drobny 255<br />
A BASIC Program to Calculate the Evolution of Cartesian Product Operators,<br />
S. Mammi 259<br />
Selective Long-Range Polarization Transfer via DEPT, T. Parella, F. Sanchez-Ferrando<br />
and A. Virgili . . 263<br />
Computer Simulations of High Resolution NMR Spectra, S. A. Smith, W, E. Palke<br />
and J.T. Gerig 267<br />
Variation of 13 C NMR Linewidths of Metallocenes as a Function of Magic Angle<br />
Sample Spinning Frequency, I. J. Shannon, K. D. M. Harris and S. Arumugam 273
Bulletin of Magnetic Resonance<br />
The Structural Role of Water in Silicate Glasses: X H and 29 Si NMR Evidence,<br />
J. Kiimmerlen, T. Schaller, A. Sebald and H. Keppler 278<br />
High-Resolution Solid-State NMR Study of Microstructures in Layered Aluminosilicate,<br />
S. Hayashi, T. Ueda, K. Hayamizu and E. Akiba 282<br />
Broadline NMR of Structural Ceramics, C. Connor 285<br />
Permeability of Liposomal Membranes to Molecules of Environmental Interest:<br />
Results from NMR Experiments Employing Shift Agents, F. G. Herring, W. R. Cullen,<br />
J. C. Nelson and P. S. Phillips 289<br />
Nuclear Magnetic Resonance Partitioning Studies of Solute Action in Lipid Membranes,<br />
L. Ma, T. F. Taraschi and N. Janes 293<br />
Weak Molecular Interactions: NMR Spectroscopy of Oriented Molecules,<br />
C. L. Khetrapal 299<br />
Structural Studies of Collagen by Solid State NMR, R. J. Wittebort and A. M. Clark . 303<br />
Calendar of Forthcoming Conferences 307<br />
Recent Magnetic Resonance Books 308<br />
Instructions for Authors 314
Vol. 14, No. 1-4<br />
Solid-State Polarization-Transfer Experiments<br />
Involving Quadrupolar Nuclei<br />
K. T. Mueller, C. A. Fyfe, H. Grondey, K. C. Wong-Moon and T. Markus<br />
1 Introduction<br />
Department of Chemistry, University of British Columbia<br />
2036 Main Mall, Vancouver, BC Canada V6T 1Z1<br />
Historically, cross-polarization experiments<br />
[1,2] have been used to transfer spin<br />
coherence from abundant spins to a dilute spin<br />
system. Protons (*H) have been used almost<br />
exclusively as the source of strong nuclear<br />
polarization for cross-polarization experiments,<br />
although there have been some examples where<br />
other abundant nuclei have been used. Coupled<br />
with magic-angle spinning (MAS) NMR [3],<br />
cross-polarization techniques have proven<br />
extremely powerful for the study of organic<br />
solids.<br />
Inorganic systems such as zeolites, gels,<br />
and ceramics are of great technological<br />
importance and interest and contain many<br />
quadrupolar nuclei but very few protons. For<br />
quadrupolar nuclei with non-integral spins such<br />
as n B, 17 O, and 27 A1, the second-order<br />
quadrupolar broadening of the readily observed<br />
central (+1/2 -1/2) transition is not<br />
completely averaged by MAS, and the NMR<br />
lines from quadrupolar spins are shifted and<br />
distorted in single-axis spinning experiments<br />
[4,5]. Very few examples exist of crosspolarization<br />
experiments involving quadrupolar<br />
nuclei, and they all involve magnetization<br />
transfer from protons to quadrupolar nuclei. We<br />
have undertaken a study to determine the<br />
feasibility of polarization transfer and dipolar<br />
dephasing experiments between spin pairs in<br />
these systems, particularly between 3l P (I = 1/2)<br />
and 27 A1 (I = 5/2). Our preliminary results show<br />
that these experiments are indeed possible [6],<br />
The attainment of cross-polarization from<br />
quadrupolar spin systems is particularly<br />
important in materials chemistry as these nuclei<br />
usually have very short Ti relaxation times.<br />
Spin-1/2 nuclei in dense inorganic systems may<br />
have Ti values ranging from many seconds to<br />
hours, effectively precluding their observation in<br />
many instances. By using cross-polarization<br />
from the quickly relaxing quadrupolar spins,<br />
spectra of the spin-1/2 nuclei could be obtained<br />
in a relatively short time. Additional<br />
information regarding the local structure and<br />
bonding in these systems might also be obtained<br />
through the distance dependence of the crosspolarization<br />
process.<br />
Similarly, dipolar-dephasing NMR<br />
experiments such as rotational-echo doubleresonance<br />
(REDOR) and transferred-echo doubleresonance<br />
(TEDOR) have been demonstrated to<br />
be useful for demonstrating connectivities and<br />
determining internuclear distances [7,8] in<br />
heteronuclear spin systems with dipolar<br />
couplings. Experimental verification of these<br />
experiments with the same heteronuclear spin<br />
pair ( 31 P and 27 A1) demonstrates the feasability<br />
of applying these techniques to systems<br />
containing quadrupolar nuclei.<br />
2 Experimental<br />
The sample chosen for study was the very<br />
large pore molecular sieve VPI-5, an<br />
aluminophosphate dihydrate containing 18membered-rings<br />
[9]. NMR experiments were<br />
performed under MAS conditions in a 9.4 T<br />
superconducting magnet where the resonance<br />
frequencies for 31 P and 27 A1 are 161.98 MHz<br />
and 104.26 MHz respectively. The rotational<br />
frequencies in all experiments were<br />
approximately 3.1 kHz, and 90° pulse times for<br />
the nuclei studied ranged from 9 to 12 (xsec.<br />
3 Results<br />
The spectra in Figure 1 demonstrate the<br />
transfer of magnetization using cross-polarization<br />
in both directions between the 27 A1 and 31 P<br />
spins in the Al-O-P bonding units in VPI-5.<br />
The cross-polarization is accomplished with an<br />
appropriate spin-locking pulse sequence [2] after<br />
a preparation pulse creates spin coherence for the<br />
nuclei used as the polarization source. With<br />
MAS the 31 P chemical shift anisotropies are<br />
averaged to their isotropic values for the three<br />
crystallographically inequivalent 31 P sites in the<br />
unit cell. For the 27 A1 nuclei, MAS partially<br />
averages the second-order quadrupolar interaction<br />
and two resonances are seen: One from the
10<br />
(b)<br />
0 -SO '<br />
ppm from 85% H3PO4<br />
(d)<br />
» 0 -JO<br />
ppm from A1(NO3>3<br />
Figure 1. Demonstration of 27 A1 —> 31 P and<br />
31 P —> 27 A1 cross-polarization in VPI-5<br />
(projection of unit cell shown at top). Spinning<br />
sidebands are marked with an (s).<br />
tetrahedrally coordinated aluminum sites (41<br />
ppm) and a second from the octahedrally<br />
coordinated aluminum (approximately -18 ppm).<br />
The observed signals are solely due to crosspolarization<br />
and not caused by direct irradiation<br />
during the spin-lock as proven by a series of<br />
cross-check experiments, of which spectra (b) and<br />
(d) of Figure 1 are representative.<br />
A two-dimensional heteronuclear<br />
correlation experiment [10] using crosspolarization<br />
can be performed by preparing the<br />
aluminum spins with a 90° pulse, and then<br />
encoding their evolution frequencies in an initial<br />
time period. The aluminum polarization is<br />
subsequently transferred to the phosphorus spins<br />
with a spin-lock, and a phosphorus free induction<br />
decay is accumulated after each of a set of<br />
aluminum evolution times. Two-dimensional<br />
Fourier transformation provides the correlation<br />
spectrum of Figure 2. From the twodimensional<br />
spectrum it is evident that each of<br />
the three 31 P resonances is connected to both<br />
tetrahedral and octahedral 27 A1 resonances, in<br />
agreement with the proposed crystal structure of<br />
VPI-5 [11].<br />
The REDOR experiment [7] was carried<br />
out in both directions between 27 A1 and 31 P<br />
nuclei in VPI-5, and the results are shown in<br />
Figure 3. In both directions, negative crosscheck<br />
experiments (not shown) were undertaken<br />
to ensure that the observed signal came from the<br />
dipolar dephasing phenomenon and not from<br />
timing missets or experimental artifacts.<br />
00<br />
3<br />
100 JO 0 - 5 0<br />
ppm from A1(NO3)3<br />
Bulletin of Magnetic Resonance<br />
Figure 2. Two-dimensional heteronuclear<br />
correlation spectrum of - 7 A1 and 31 P in VPI-5.<br />
The TEDOR experiment [8] was also<br />
accomplished with initial evolution of 27 A1<br />
spins and subsequent transfer to the 31 P after two<br />
rotor periods of preparative dephasing. The<br />
signal in Figure 3(c) was obtained after one<br />
additional period of dipolar evolution in order to<br />
create observable spin coherence from the<br />
antiphase signal which was transferred at the time<br />
of the 27 A1 spin echo. As before, negative<br />
cross-check experiments were performed and gave<br />
null signals.<br />
A two-dimensional TEDOR experiment.<br />
[12] was performed with 27 A1 spin frequency<br />
encoding before the initial dipolar dephasing<br />
period. After transfer of the coherence to the 31 P<br />
spins, an FID was accumulated and the ti value<br />
incremented. The two-dimensional correlation<br />
spectrum is shown in Figure 4, revealing crosspeaks<br />
between all three 31 P resonances and both<br />
the resonances from the tetrahedrally coordinated<br />
and octahedrally coordinated 27 A1 sites, in<br />
agreement with the proposed crystal structure and<br />
the results of the two-dimensional CP<br />
experiment discussed above. These results are<br />
taken as demonstrating the general success of the<br />
experiments. A more detailed interpretation of<br />
the results in terms of the individual T-sites from<br />
the present data alone is not attempted.
Vol. 14, No. 1-4 11<br />
oc<br />
"P REDOR<br />
IS C -B -SO -73<br />
ppm from 85% H3PO4<br />
Wrahedral<br />
r AI REDOR<br />
IX 75 X 25 0 -23 -SO<br />
ppm from A1(NO})3<br />
P TEDOR<br />
0 .J5 -JO -75<br />
ppm from 85% H3PO4<br />
Figure 3. One-dimensional REDOR and<br />
TEDOR spectra of 27 A1 and 3I P in VPI-5.<br />
SO 25 0 -25<br />
ppm from A1(NO3>3<br />
Figure 4. Two-dimensional TEDOR spectrum of<br />
27 Al and 31 P in VPI-5.<br />
4 Conclusions<br />
In summary, cross-polarization to and<br />
from quadrupolar nuclei has been experimentally<br />
verified using the 31 P and 27 A1 spin systems in<br />
an aluminophosphate molecular sieve. This<br />
bodes well for the use of heteronuclear<br />
correlations for further investigation of local<br />
micrdstructure in solids. Dipolar-dephasing<br />
experiments have also been accomplished, with<br />
both REDOR and TEDOR results confirming the<br />
connectivities detected by the cross-polarization<br />
studies. A two-dimensional TEDOR experiment<br />
has also been demonstrated that separates<br />
connectivities between distinct resonances.<br />
References<br />
1 S. R. Hartmann and E. Hahn Phys. Rev.,<br />
128, 2042, 1962.<br />
2 A. Pines, M. G. Gibby, and J. S. Waugh /.<br />
Chem. Phys., 59, 569, 1973.<br />
3 J. Schaefer and E. 0. Stejskal J. Am. Chem.<br />
Soc.,9%, 1031, 1976.<br />
4 H. J. Behrens and B. Schnabel Physic a,<br />
114B, 185. 1982.<br />
->A. Samoson. E. Kundla, and A. Lippmaa J.<br />
Magn. Reson.. 49, 350, 1982.<br />
6 C. A. Fyfe, H. Grondey, K. T. Mueller, K. C.<br />
Wong-Moon, and T. Markus /. Am. Chem.<br />
Soc, 114, 5876, 1992.<br />
7 T. Gullion and J. Schaefer J. Magn. Reson.,<br />
81, 196, 1989.<br />
8 Y. Pan and J. Schaefer J. Magn. Reson., 90,<br />
341, 1990.<br />
9 M. E. Davis, C. Sadarriaga, C. Montes, J.<br />
Garces, and C. Crowder Nature (London), 331,<br />
698, 1988.<br />
l0 P. Caravatti, G. Bodenhausen, and R. R.<br />
Ernst Chem. Phys. Lett., 89, 363, 1982.<br />
n L. B. McCusker, Ch. Baerlocher, E. Jahn,<br />
and M. Bulow Zeolites. 11, 308, 1991.<br />
12 C. A. Fyfe, K. T. Mueller. H. Grondey, and<br />
K. C. Wong-Moon, submitted to Chem. Phys.<br />
Lett.
12<br />
1. Introduction<br />
Magnetic Resonance<br />
Evidence for Superconductivity<br />
in a Semimetal<br />
I.P. Goudemond, G. J. HiU and M.J.R. Hoch<br />
Department of Physics and<br />
Condensed Matter Physics Research Unit,<br />
University of the Witwatersrand, Johannesburg<br />
The group V semimetals As, Sb and Bi have<br />
low carrier densities and may be viewed<br />
as rather poor metals. Their electronic<br />
properties have been studied using a variety<br />
of methods, including NQR [1], [2]. At<br />
temperatures below the Debye temperature<br />
the nuclear spin—lattice relaxation rate in As<br />
and Sb has been found to obey the Korringa<br />
relation [1], [2]. For As, this relation has<br />
been found to hold down to 150 mK [3].<br />
In the present work Ti measurements<br />
on As have been extended to still lower<br />
temperatures. Motivation has come from<br />
the interesting electrical conductivity<br />
behaviour found by Uher [4] for a single<br />
crystal sample in the vicinity of 100 mK.<br />
These results may be interpreted as evidence<br />
for a superconducting transition, although no<br />
further experiments appear to have been<br />
carried out to confirm this. Probing of the<br />
superconducting state in zero magnetic field<br />
using NQR methods offers interesting<br />
challenges and opportunities.<br />
2. Experimental Details<br />
The experiments were carried out in an<br />
Oxford dilution refrigerator using procedures<br />
that have been described previously [3].<br />
Pulsed NQR spin echo methods with signal<br />
averaging were used at 23.5 MHz on a<br />
powdered As sample. ( 75 As has I = 3/2 and<br />
100% abundance). The powdered material<br />
was prepared by crushing and sieving (25 \i<br />
mesh) high purity (99.9995%) arsenic,<br />
Bulletin of Magnetic Resonance<br />
followed by annealing in vacuo and further<br />
careful sieving. An oxide layer on the<br />
surface of the grains prevented metallic<br />
contact between neighbouring particles.<br />
Refrigerator<br />
Mixing Chamber<br />
Sintered Silver<br />
Heat Exchanger<br />
Vacuum Space<br />
Coaxial Cable<br />
Helium Fill<br />
Capillary<br />
Copper Block<br />
Cermanium<br />
Thermometer<br />
R F Coil<br />
Stycast Sample<br />
Holder<br />
Figure 1<br />
Sample holder and rf coil assembly for NQR<br />
measurements in the dilution refrigerator. The<br />
sample is immersed in liquid *Be, which is in<br />
contact with the sintered silver heat<br />
exchanger.<br />
Figure 1 depicts the sample<br />
arrangement used to ensure good thermal<br />
contact to the refrigerator mixing chamber.<br />
Liquid 4 He surrounds the sample and is<br />
in contact with a sintered silver heat<br />
exchanger. Further details may, be found<br />
in reference [3]. Fractional 'rf pulses<br />
were used to minimize heating effects.<br />
Temperatures were measured using a<br />
calibrated germanium thermometer mounted<br />
on the mixing chamber.
Vol. 14, No. 1-4 13<br />
3. Results and Discussion<br />
In an effort to establish that the sample<br />
was in good thermal contact with it's<br />
surroundings, careful measurements of the<br />
echo amplitude were made as a function<br />
of temperature down to the lowest<br />
temperatures reached in these experiments,<br />
40 mK. Down to 150 rnK Curie law type<br />
behaviour is observed. At lower<br />
temperatures the data depart from linear<br />
behaviour. We do not believe that this is<br />
due to heating effects or the loss of thermal<br />
contact. Changes in the pulse sequence<br />
repetition rate did not change the amplitude<br />
of the echo signal. It is likely that some<br />
mechanism, characteristic of the sample, is<br />
responsible for the departure from Curie law<br />
behaviour.<br />
At temperatures of 4 K and below, the<br />
skin depth 6 at 23.5 MHz is comparable to,<br />
or less than, the mean particle radius r. We<br />
estimate that 5 ~ .3.0 [im at 150 rnK, while<br />
r < 10 fim. It is clearly desirable that<br />
smaller particles should be used, although<br />
this is difficult because of rapid surface<br />
oxidation which occurs in air and the<br />
tendency of the arsenic particles to sinter<br />
during annealing.<br />
Taking into account the attenuation of<br />
rf pulses and the presence of a core of<br />
undisturbed spins in the particles, we have<br />
examined the situation in some detail. In<br />
order to see whether spin diffusion can<br />
operate between spins near the surface of a<br />
particle and those in the interior, we have<br />
calculated the spin diffusion coefficient<br />
D = V30 V^d 2 , where M2 is the second<br />
moment and d the spin spacing, and obtain<br />
D ~ 2 x 10" 13 cm 2 s"'. On the time—scale of<br />
our experiments (10 2 — 10 3 s) spin diffusion<br />
operates over a distance v/2Dt ~10~ l fim.<br />
It may be concluded that this mechanism<br />
should have negligible effects on our results.<br />
The change in the skin depth with<br />
temperatures below 1 K is of the order of 1<br />
to 2%, which our calculations show will not<br />
lead to detectable changes in the echo<br />
amplitude beyond the Curie law changes.<br />
We do not believe that the departure from<br />
Curie law behaviour is due to effects of this<br />
kind.<br />
The measured spin lattice relaxation<br />
rates are shown as a function of temperature<br />
in Figure 2. Note that, for magnetic<br />
relaxation in an I = 3 /2 system, a unique<br />
relaxation rate may be defined using<br />
1/T] = 6 Wm, where Wn, is the transition<br />
rate between the ±'/2 and ± 3 /2 spin states.<br />
At temperatures down to roughly<br />
120 mK the data obey the Korringa relation.<br />
Below this temperature the relaxation rate<br />
0.09<br />
8.64<br />
0.01 8.62<br />
30 50 70 90 110 130 150 170<br />
T (mK)<br />
Figure 2<br />
Plots of the ?!>As Spin lattice relaxation rate<br />
and of the resistivity of a single crystal<br />
arsenic sample [4] as a function of temperature<br />
down to 50 mK. The straight line which joins<br />
with the relaxation rate plot represents an<br />
extra— polation of the Korringa relation for<br />
arsenic found at higher temperatures.<br />
decreases less rapidly with T than expected<br />
from the Korringa relation. Figure 2 also<br />
shows the electrical conductivity data of<br />
Uher [4] in the same temperature range as<br />
the present measurements. It can be seen<br />
that the conductivity starts to decrease quite<br />
rapidly at ~100 mK. Inspection of the two<br />
plots in Figure 2 suggests a common<br />
underlying physical mechanism for the<br />
changes in behaviour observed in this<br />
temperature range.<br />
It appears likely that a fairly broad<br />
superconducting transition occurs with a Tc<br />
around 100 fiK. The T"i data suggest a<br />
slightly higher Tc value than the<br />
conductivity data.<br />
Cohen [5] has pointed out that the<br />
semimetals may be candidates for<br />
superconductivity through a BCS pairing<br />
mechanism. This is largely because of the<br />
multivalley character of these materials. On<br />
the basis of calculations given by Cohen for<br />
systems of this kind, a Tc of 100 mK is not<br />
unreasonable for As. Doped Bi has been<br />
found by Uher and Opsal [6] to have<br />
a Tc < 100 mK, depending on the<br />
concentration of the Sn or Te dopant.<br />
We have attempted to fit the observed<br />
relaxation data for As using the Hebel—<br />
Slichter expression based on BCS theory.<br />
Anisotropy of the gap is allowed for by<br />
introducing a parameter r = Ao(0)/A, where<br />
Ao is the BCS gap at 0 K and A is a<br />
measure of the gap anisotropy. However the<br />
theoretical curve does not fit the<br />
experimental data plotted in reduced form.<br />
Details will be given elsewhere. It is quite<br />
possible that the suggested transition in As<br />
is not of the standard BCS type. It appears,
14<br />
however, that other effects could be<br />
important.<br />
Cohen [5] suggests that the semimetals<br />
will become type II superconductors with<br />
rather low upper critical fields. In the NQR<br />
experiment rf fields of 30 or 40 G are used<br />
and it is possible that some remnant field<br />
effects may be produced in the sample which<br />
contribute to relaxation.<br />
Further experiments should be carried<br />
out to confirm the onset of<br />
superconductivity in As around 100 mK.<br />
Clearly, Meissner effect measurements<br />
should be attempted. Further NQR<br />
experiments involving CW methods with<br />
low rf fields may prove useful and complementary<br />
to the present measurements.<br />
4. Conclusion<br />
Evidence has been obtained of marked<br />
departures from Korringa relaxation<br />
behaviour in As below 120 mK. Taken<br />
together with previous electrical<br />
conductivity results, it appears likely that<br />
the anomalous behaviour is due to the onset<br />
of superconductivity.<br />
Bulletin of Magnetic Resonance<br />
The relaxation rate measurements are<br />
not in agreement with the Hebel—Slichter<br />
theory predictions. Further work is required<br />
to determine whether this is because of the<br />
non—BCS nature of the transition or some<br />
other cause, such as the magnitude of the rf<br />
fields used in our pulsed NQR. experiments.<br />
5. References<br />
1. J.M. Keartland, G.C.K. Folscher and<br />
M.J.R. Hoch, Phys. Rev. B 43, 8362<br />
(1991).<br />
2. J.M. Keartland, G.C.K. Folscher and<br />
M.J.R. Hoch, Phys. Rev. B. 45, 7882<br />
(1992).<br />
3. IP Goudemond, J M Keartland, and<br />
M J R Hoch, J. Low Temp. Physics 82,<br />
369 (1991).<br />
4. C Uher, J. de Physique, C6, 39, 1054<br />
(1978).<br />
5. ML Cohen, in "Superconductivity",<br />
edited by R D Parks (Marcel Dekker,<br />
New York, 1969), Vol.1.<br />
6. C Uher, J L Opsal, Phys. Rev. Letters<br />
40,1518(1978).
Vol. 14, No. 1-4 15<br />
NMR Spectroscopy in Cardiac Surgery<br />
Roxanne Deslauriers, Sylvain Lareau" 1 ", Rosalind S. Labow+, Wilbert J. Keon + , Gang-Hong Tian # , Anthony L. Panos*,<br />
Carlos A.M. Barrozo*, Imtiaz S. Ali *, Owayed Al-Nowaiser*, and Tomas A. Salerno*<br />
1 Introduction<br />
Institute for Biodiagnostics, National Research Council of Canada, Ottawa<br />
+ University of Ottawa Heart Institute, Ottawa<br />
# Department of Physiology, University of Ottawa, Ottawa, Ont.<br />
* Division of Cardiovascular Surgery, University of Toronto, Toronto, Ont., Canada<br />
Applications of magnetic resonance spectroscopy in<br />
medicine have been restricted mostly to the research<br />
laboratory. The technique is now entering the field of<br />
medical diagnosis and therapy. In the heart, levels of<br />
phosphorus metabolites are often correlated with function.<br />
Nuclear magnetic resonance spectroscopy has been used to<br />
monitor high energy phosphorus metabolite levels in the<br />
heart to evaluate the effect of work and ischemic stress. Our<br />
applications of magnetic resonance to the practice of cardiac<br />
surgical have been in three areas: a) preservation of tissue<br />
for transplantation b) optimization of myocardial protective<br />
techniques (cardioplegia) and c) monitoring of the heart<br />
during the aortic clamping period.<br />
2 Heart Preservation for Transplantation<br />
With increasing demand for a limited number of donor<br />
hearts, organ preservation during procurement is critically<br />
important. Controversy still exists over issues such as the<br />
optimal temperature, optimal solution, and maximum time<br />
limit for donor heart preservation. Numerous studies have<br />
been, and continue to be, conducted on various animal<br />
tissues. Not much data have been obtained in human<br />
myocardium primarily because of the difficulty in obtaining<br />
adequate quantities of viable tissue for laboratory<br />
investigation.<br />
a) Human Atrial Tissue<br />
Portions of human atrial appendages normally discarded<br />
during cannulation in the course of surgery requiring<br />
cardiopulmonary bypass have been used, with informed<br />
patient consent, for studies of heart preservation. We have<br />
used 31 P and ^H NMR spectroscopy to define the optimal<br />
temperature for long-term (up to 24 hours) preservation of<br />
high energy metabolite levels and contractile function, and<br />
to gain a fundamental understanding of the energy<br />
generating pathways in preserved human cardiac tissue [1,<br />
2]. The studies were carried out on a Bruker AM-360<br />
spectrometer. 31 P spectra were obtained using a 60° pulse<br />
and a 1 s recycle time. 1 H spectra were acquired using a<br />
spin-echo sequence based on the water-suppressing 1331<br />
pulse sequence. The acquisition of 31 P and *H NMR<br />
spectra were interleaved (Figure 1). 31 P spectra were used<br />
to measure ATP levels on a continuous basis, as an index of<br />
net energy preservation. *H spectra of lactate provided<br />
information on generation of ATP through the glycolytic<br />
pathway.<br />
10 0 -10 -20 1<br />
PPM<br />
PPM<br />
Lactate<br />
FIGURE 1 31 P (left) and l H (right) NMR spectra of an<br />
atrial appendage (ca. 0.5 g) preserved at 20°C in saline, as a<br />
function of time [1] {ref, reference capillary; PME,<br />
phosphomonoester; Pi, inorganic phosphate).<br />
Studies of atrial appendages preserved at 1°, 4°, 12° and<br />
20°C in physiological saline (0.9% NaCl) for up to 20 hours<br />
demonstrate that preservation of ATP is better at 1° and 4°<br />
than at 12° or 20°C. Based on measurements of lactate<br />
production, glycolysis is active at all the temperatures, its<br />
rate correlating positively with increasing temperature.<br />
However, the ATP generated by glycolysis falls short of
16<br />
demand at all temperatures, but the difference is small at 1°<br />
and 4°C (Table 1).<br />
TABLE 1<br />
Energy balance in human cardiac tissue preserved in NaCl<br />
0.9% [1].<br />
Temp.<br />
(°C)<br />
1<br />
4<br />
12<br />
20<br />
ATP* #<br />
loss<br />
7<br />
8<br />
12<br />
20<br />
Lactate*<br />
production<br />
43<br />
52<br />
106<br />
212<br />
ATP*+<br />
generated<br />
65<br />
78<br />
159<br />
318<br />
ATP* &<br />
utilization<br />
72<br />
86<br />
171<br />
338<br />
* nmoleg" 1 (wet myocyte mass) min' 1 .<br />
* From the rate of change of NMR-visible ATP.<br />
+ Assuming 1.5 mole of ATP produced per mole of lactate<br />
from glycolysis.<br />
& Calculated from (rate of generation of ATP by<br />
glycolysis) + (2 x rate of ATP loss). This takes into<br />
account the ATP generated by adenylate kinase.<br />
In a separate series of studies, we tested the possibility<br />
of improving the maintenance of high energy phosphates at<br />
12°C, one of the higher temperatures currently used in some<br />
institutions for the preservation of heart grafts. Our<br />
hypothesis was that the poor maintenance of high energy<br />
phosphates at 12° and 20°C results from the increased<br />
intracellular acidosis that occurs at higher temperatures [1].<br />
Ultimately, acidosis partially inhibits ATP production by<br />
glycolysis, the only metabolic pathway for generation of<br />
ATP in the anoxic heart. We postulated that the addition of<br />
a buffer to the solution used for heart preservation would<br />
increase the rate of transport of protons and lactate to the<br />
extracellular space, thereby maintaining better intracellular<br />
pH homeostasis. Our studies showed that at 12°C, the halftime<br />
for loss of ATP increased from 300 minutes in saline to<br />
over 900 minutes in a modified Krebs-Henseleit solution<br />
containing 100 mM buffer [2, 3]. This observation was<br />
confirmed independently using biochemical measurements<br />
of high energy phosphates [3]. The beneficial effects of<br />
high buffer concentration observed at 12°C did not occur at<br />
4°C (figure 2 [3]), which lead us to postulate that at that<br />
temperature, glycolysis was rate limited by the temperature<br />
rather than by the acidosis. These studies show the<br />
continued need for testing all the conditions to which a graft<br />
may be subjected, and the importance of avoiding broad<br />
Bulletin of Magnetic Resonance<br />
generalizations when dealing with complex multi-enzyme<br />
systems.<br />
4°C<br />
3-j<br />
Q 0.9% saline (3)<br />
3 • 20 mM PIPES (6)<br />
Ito<br />
o 60 mM PIPES (6)<br />
• 100 mM PIPES (6)<br />
2- —<br />
c/5<br />
•d<br />
1-<br />
0-<br />
0 5<br />
2-<br />
1 -<br />
12°C<br />
EP<br />
dm<br />
10 15<br />
Time<br />
><br />
20 25<br />
• 0.9% saline (3)<br />
• 20 mM PIPES (6)<br />
o 60 mM PIPES (6)<br />
• 100 mM PIPES (6)<br />
10 15 20 25<br />
Time (h)<br />
FIGURE 2. Effect of the concentration of PIPES buffer on<br />
the preservation of ATP in isolated atrial appendages<br />
preserved at 4° or 12 °C in modified Krebs-Henseleit<br />
solution [3].<br />
The importance of defining the relationship between<br />
high energy phosphate levels and contractile function in<br />
human cardiac tissue led to the development of a<br />
temperature-controlled NMR microprobe incorporating a<br />
perifusion system and a non-magnetic strain gauge (figure<br />
3). The system has been used to study human atrial<br />
trabeculae, which are small functional muscle fibers<br />
weighing 8-25 mg that can be isolated from atrial<br />
appendages. The perifusion system provides the tissue with<br />
the oxygen and nutrients required for its function. The strain<br />
gauge allows for measurement of developed force in
Vol. 14, No. 1-4<br />
electrically stimulated muscle fibers while- NMR<br />
simultaneously assesses the high energy phosphate<br />
compound levels [4]. In addition, the perifusion system can<br />
mimic preservation conditions by allowing muscles to be<br />
perifused at low temperature (down to 1°C) during<br />
acquisition of NMR data (figure 4).<br />
Stimulation wire<br />
I<br />
Muscle in the<br />
NMR tube<br />
Flexible tubing<br />
Reflector<br />
Optical fiber<br />
Fiber optic<br />
strain gauge Perfusate line<br />
Stimulation wire<br />
4<br />
Adjustable<br />
hook mount<br />
FIGURE 3. Schematic drawing of the microperifusion<br />
system used for NMR studies of human atrial trabeculae.<br />
The total length of the system is 6.2 cm.<br />
By studying atrial trabeculae in the presence of<br />
metabolic inhibitors under conditions simulating<br />
preservation, it is possible to assess the contribution of<br />
various cellular mechanisms of energy production to the<br />
total energy balance of the tissue. Our 31 P NMR studies of<br />
isolated human atrial trabeculae [5] preserved at 4° and 12°C<br />
in oxygenated St-Thomas II solution showed that the high<br />
energy phosphates (ATP and phosphocreatine (PCr)) are<br />
well maintained during 18 hours of preservation.<br />
Contractile studies performed under similar conditions<br />
showed high recovery of developed force. Glycolysis, the<br />
only pathway available for energy generation under the<br />
anoxic conditions existing in large preserved organs, is<br />
capable of maintaining ATP levels in hypothermically<br />
preserved tissue. Under anoxia, ATP levels are stable for 6 -<br />
10 hours at 12°C, and for a longer period at 4°C. In a<br />
resting heart, the major energy source is provide by the lipid<br />
catabolism. To test whether this pathway is active at low<br />
• temperatures, we have measured the ability of the oxydative<br />
pathway to maintains ATP levels. At 12°C, when glycolysis<br />
is inhibited by iodoacetate, the oxidative pathway can<br />
maintained ATP levels, but only if an external source of<br />
substrate (10 mM acetate) is present in the perfusate. Thus,<br />
the oxidative pathway is functional but depends on both,<br />
oxygen and glycolysis.<br />
In tissue preserved ischemically (no flow), metabolic<br />
waste products such as lactate cannot be eliminated. This<br />
results in considerable extracellular and intracellular acidosis<br />
[1] which has profound effects on energy production<br />
because glycolysis is inhibited by low pH. Using the atrial<br />
trabecula model, we simulated the conditions that exist in<br />
the ischemically preserved human heart. Such a large organ<br />
(300 - 450 gm) cannot obtain sufficient oxygen to maintain<br />
oxidative phosphorylation simply through diffusion from the<br />
surrounding medium as in the case with trabeculae.<br />
Trabeculae were subjected to acidosis by perifusing the<br />
trabeculae with modified St-Thomas preservation solution<br />
containing 10 mM lactate at pH 6.0. Anoxia was<br />
simultaneously induced with 1 mM cyanide, a potent<br />
inhibitor of oxidative phosphorylation, and nitrogen. In the<br />
trabeculae, these conditions reproduced those previously<br />
observed in the larger atrial appendages where an anoxic<br />
core probably existed [1]. Under acidosis and anoxia at<br />
12°C, ATP decreased linearly by 40 to 100% over a 12 h<br />
period. At 4°C, ATP decreased less over the same time<br />
period.<br />
40<br />
DMMP<br />
1 I •<br />
30<br />
1 I '<br />
20<br />
10<br />
a -ATP Y.ATP<br />
1 I '<br />
-10<br />
-20<br />
PPM<br />
FIGURE 4. 31 P NMR spectrum (147 MHz) of a 10.5 mg<br />
trabecula perifused at 12°C in modified St-Thomas II<br />
solution (60° pulse, 1 sec recycling delay, 2400 scans).<br />
These observations can be reconciled to the following<br />
model of energetics: the maintenance of cellular ATP<br />
depends on matching of supply and demand. At 4°C,<br />
glycolysis appears to be limited by temperature. ATP<br />
regeneration cannot be driven at an adequate rate until the<br />
feedback drive (ADP + Pj) is increased considerably above<br />
the normal level. ATP is then maintained at a low<br />
phosphorylation potential. At 12°C glycolysis is not limited<br />
by temperature but is limited by low intracellular pH.<br />
17
18<br />
Other than gaining a better understanding of energetics<br />
in atrial trabeculae, we can ask whether these studies<br />
provided the transplant surgeon with any information of<br />
practical value? In answer to our initial question on the<br />
optimal temperature for preservation of grafts, we have<br />
provided evidence that, for preservation times of 5 hours or<br />
less, ATP levels are better maintained at 12°C [3]. For<br />
longer preservation times, ATP levels are better preserved at<br />
the lower temperatures. In addition, increasing the buffer<br />
capacity of preservation solutions used at 12°C has a major<br />
impact on maintenance of high energy phosphates.<br />
b) Intact Hearts<br />
Most published NMR studies on heart preservation have<br />
used rodent hearts, with a particularly large number of<br />
studies being performed on the rat heart. We have<br />
developed the isolated perfused pig heart for preservation<br />
studies [6] because it is architecturally, biochemically, and<br />
in size most similar to the human heart. As we discussed<br />
above, provision of oxygen and removal of metabolic waste<br />
products are critically important for long term heart<br />
preservation. Perfusion preservation, which can enhance<br />
oxygen delivery and waste removal from the heart, has not<br />
yet achieved much clinical application and remains mostly<br />
in the realm of the research laboratory. Some of the reasons<br />
for this are related to the implementation difficulties in<br />
situations in which the heart must be transported over long<br />
distances under sterile conditions. In addition, hearts<br />
preserved ischemically for less than 5-6 hours generally<br />
show good recovery of mechanical function after<br />
transplantation.<br />
We have consequently focused our efforts on methods<br />
of improving long-term ischemic preservation of hearts.<br />
The goal is to optimize the conditions currently in use and to<br />
extend the safe preservation time between harvest and<br />
implantation of the donor organ. This should allow harvest<br />
of donor hearts to occur over a wider geographical range and<br />
provide for better immunological organ matching.<br />
The studies of isolated, Langendorff perfused pig hearts<br />
[6] are performed using a Bruker Biospec instrument<br />
equipped with a 4.7 T / 30 cm horizontal bore magnet. The<br />
heart is arrested and isolated using techniques similar to<br />
those used for human hearts. The isolated heart is then<br />
placed in an NMR probe to observe high energy phosphate<br />
levels and pH on a continuous basis, with a two minute time<br />
Bulletin of Magnetic Resonance<br />
resolution, during a hypothermic preservation period that<br />
usually lasts 8 hours. Following preservation, the heart is<br />
rewarmed to 37°C without removing it from the magnet and<br />
NMR spectra are then recorded in the beating heart. A<br />
balloon placed in the left ventricle measures the developed<br />
pressure and serves as an index of functional recovery of the<br />
heart [6]. In this manner, energy levels during and after<br />
preservation can be correlated with functional performance<br />
of the heart following preservation.<br />
A number of technical difficulties arise in trying to<br />
obtain quantitative results from large, isolated, perfused<br />
hearts because they change shape when beating and often<br />
swell when perfused with solutions other than whole blood.<br />
To alleviate the problems caused by the sample moving in a<br />
heterogeneous Bi field and the consequent uncertainty in<br />
received signal strength, a high Bi homogeneity probe was<br />
designed [7]. The prototype probe comprised four separate<br />
tuned rings on a spherical surface (Figure 5) giving a Bi<br />
field homogeneity of ± 5% over 60% of the radius of a 14<br />
cm sphere (Figure 6).<br />
tobalun<br />
coupling ring<br />
FIGURE 5. Geometry of the 4 coil system [7].<br />
The received signal was rendered less sensitive to the<br />
dielectric constant of the sample by distributing the<br />
capacitance around the rings. Inter-ring coupling and to a<br />
fifth ring used for matching was by induction. The coupling<br />
loop was tuned with its own capacitor to Larmor frequency,<br />
thereby ensuring that the probe was always on resonance,<br />
and rendering the tuning and matching independent. In<br />
addition, the use of a low input impedance preamplifier
Vol. 14, No. 1-4 19<br />
virtually eliminated the dependence of signal strength on<br />
coil loading [8].<br />
o V* °<br />
its<br />
/ £<br />
/ -<br />
0/ :g<br />
/ -2-<br />
/ *<br />
7 1<br />
/ I il<br />
0.8-<br />
0.6<br />
0.4'<br />
B1<br />
i<br />
\° \V\°\\V<br />
, . 1 1—-,—*-y<br />
-10 Distance in cm<br />
FIGURE 6 Plot down the Y axis of the Bi field of the<br />
probe (open circles, f = 7 cm at 81 MHz) and the form of the<br />
plot predicted by the theory.<br />
Strategies for improving hypothermic preservation have<br />
ranged from improving the buffer capacity of the<br />
preservation solution [6] to the use of secondary<br />
cardioplegia [6, 9] to maintain the heart in an arrested state<br />
during the entire rewarming phase prior to reperfusion. The<br />
purpose of secondary cardioplegia is to allow the energy of<br />
the heart to be directed towards the re-establishment of ionic<br />
balances that become disrupted by hypothermia and<br />
ischemia, rather than to expend energy in mechanical<br />
function. We have found that the use of secondary<br />
cardioplegia prior to reperfusion does not affect the net<br />
energy levels of the heart but rather eliminates ventricular<br />
fibrillation that is normally observed upon rewarming of<br />
hypotfaermically preserved hearts [6].<br />
As a result of the increasing requirement for donor<br />
organs, a number of organs are frequently harvested from a<br />
single donor. This has led to the need for a single<br />
preservation solution suitable for all thoracic and abdominal<br />
organs. The University of Wisconsin Cold Storage Solution<br />
(UW-CSS, DuPont Pharmaceuticals) is currently in use for<br />
the preservation of liver, kidney and pancreas. Its utility for<br />
heart preservation remains to be determined. We compared<br />
the efficacy of UW-CSS to St-Thomas II solution, which is<br />
in widespread use for heart preservation [10]. Pig hearts<br />
were preserved for 8 hours at 4° or 12°C and then tested<br />
functionally after rewarming to 37°C. These temperatures<br />
were chosen because they are both in use clinically. At 4°C,<br />
UW-CSS and St-Thomas II were equally effective for<br />
preservation of heart function. Figure 7 and 8 shows the<br />
results obtained with UW-CSS and St-Thomas at 12°C. The<br />
lack of functional recovery observed with UW-CSS shows<br />
that this solution is unsuitable for use at 12°C. The results<br />
also demonstrate the necessity for precise temperature<br />
regulation when the solution is used at 4°C. This is not<br />
necessary with St-Thomas II solution because recovery is<br />
not severely compromised by use at either 4° or 12°C.<br />
Ref.<br />
PCr<br />
20 0 -20<br />
PPM<br />
P-ATP<br />
20 0 -20<br />
PPM<br />
20 0 -20<br />
PPM<br />
St-Thomas UW-CSS<br />
4°C<br />
Reperfusion<br />
8 h ischemia<br />
4 h ischemia<br />
2 h ischemia<br />
Reperfusion<br />
8 h ischemia<br />
4 h ischemia<br />
2 h ischemia<br />
FIGURE 7. Typical time course of the 3 l P NMR spectra of<br />
4 hearts, two preserved at 4°C (top panels), two preserved at<br />
12°C (bottom panels) [10]. Spectra on the left were<br />
obtained from hearts stored with St-Thomas II and spectra<br />
on the right were from hearts preserved with UW-CSS. The<br />
ATP and PCr disappeared upon reperfusion in the hearts<br />
stored with UW-CSS at 12°C.<br />
One of the reasons for the failure of UW-CSS in heart<br />
preservation at 12°C could be the calcium paradox. This<br />
phenomenon occurs when a heart has been subjected to a<br />
calcium-free medium (UW-CSS contains no calcium) and<br />
then is reperfused with a solution containing calcium. The<br />
calcium paradox results in massive irreversible damage to<br />
cell membranes, as calcium from the reperfusion medium<br />
overloads the cells. This phenomenon is temperaturedependent<br />
and does not occur readily at low temperatures.<br />
In order to test the "calcium paradox" hypothesis, we added<br />
0.5 mM calcium (0.08 mM free calcium) to UW-CSS and
I<br />
III<br />
20 Bulletin of Magnetic Resonance<br />
repeated our studies at 12°C [11]. Figures 9 and 10 show<br />
the improvement observed in the high energy phosphates<br />
during reperfusion of a heart preserved with UW-CSS<br />
containing calcium.<br />
6000-<br />
•a 5000-<br />
4000-<br />
3000-<br />
Cu_ 2000-<br />
1000-<br />
0<br />
0<br />
6OOO-1<br />
^ 5000-<br />
.9<br />
-M 4000-<br />
1 3000-<br />
2000-<br />
looo H<br />
o-<br />
4°C A u f *<br />
10<br />
12 P C<br />
20 30 40<br />
Time (min)<br />
St-Thomas<br />
UW-CSS<br />
50 60<br />
* *<br />
UW-CSS<br />
0 10 20 30 40 50 60<br />
Tune (min)<br />
FIGURE 8 Time course of the rate pressure product (RPP:<br />
heart rate x developed pressure, a measure of heart function)<br />
during reperfusion (n = 7 in each group) [10]. The<br />
functional recovery was extremely poor in the hearts stored<br />
with UW-CSS at 12°C (* : p < 0.05)<br />
These studies show that NMR spectroscopy can be a<br />
valuable tool in the design and modification of solutions for<br />
protecting the myocardium prior to transplantation. Cardiac<br />
surgery is another area in which NMR spectroscopy is being<br />
used.<br />
3 Cardiac Surgery<br />
Cardiac surgery is being offered to higher and higher<br />
risk patients. This has led to the need for improved methods<br />
of myocardial protection during surgery. Traditionally the<br />
heart is arrested and kept cold (4°C) with one or more<br />
infusions of a crystalloid solution, such as the St-Thomas II<br />
solution described above. The hypothermia and ischemia<br />
associated with the use of cold crystalloid solutions can<br />
impose additional stress on the damaged heart. One of the<br />
most recent modifications to cardiac surgical practice has<br />
been the use of continuous normothennic blood cardioplegia<br />
(CNBC) [12]. The purpose of CNBC is to avoid ischemia.<br />
With CNBC, the heart is maintained at 37°C in an arrested<br />
state by increasing the potassium concentration of a blood<br />
solution that continuously flows through the coronary<br />
vessels. Many questions remain unanswered regarding<br />
CNBC, such as the route of administration (retrograde<br />
and/or antegrade); b) the optimal volume of cardioplegia; c)<br />
the flow distribution of cardioplegia, and d) metabolic<br />
monitoring of the heart during cardioplegic arrest. For these<br />
purposes, a blood-perfused porcine heart model was<br />
developed for NMR studies of CNBC. In this model, the<br />
heart is continuously perfused with blood while being<br />
isolated from the animal. The heart can then be placed in<br />
the NMR magnet and its initial function assessed before it is<br />
arrested in the magnet. NMR surveillance of the high<br />
energy phosphates during CNBC may allow optimization of<br />
flow rates and the route of delivery (antegrade and/or<br />
retrograde) for maintenance of the energy status of the<br />
myocardium. In this context, localized NMR spectroscopy<br />
using either spectroscopic imaging [13] or surface gradient<br />
coils [14], allows assessment of protective techniques in<br />
selected regions of the heart [13, 15] or at various depths<br />
across the heart wall [14], respectively.<br />
40 20 0 -20<br />
PPM<br />
-Calcium<br />
2 hour ischemia<br />
ji. Reperfusioa<br />
40 20 0 -20<br />
PPM<br />
+ Calcium<br />
FIGURE 9 Typical 3 *P NMR spectra of hearts preserved in<br />
unmodified UW-CSS (left panel), or with UW-CSS<br />
containing Ca 2+ (right panel) [11]. The peaks of ATP and<br />
PCr disappeared upon reperfusion in the heart stored with<br />
unmodified UW-CSS.<br />
The difference between the two curves is statistically<br />
significant (p < 0.001) [ll]The NMR technique
Vol. 14, No. 1-4<br />
may determine whether there is a safe time limit during<br />
which blood cardioplegia can be interrupted for surgical<br />
visualization. Initially, we evaluated the effect on cardiac<br />
energetics and function by interrupting the flow of CNBC,<br />
as occurs for instance during aorto-coronary bypass surgery.<br />
Our data show that the high energy phosphate profile<br />
deteriorates and PCr becomes unobservable within 14 ± 2<br />
minutes when flow is interrupted for 20 minutes in the<br />
middle of a 1 hour period of CNBC. This is associated with<br />
decreased left ventricular function when the heart is tested<br />
after reperfusion, in spite of the fact that both PCr and pH<br />
returned to normal within 3 minutes of resuming CNBC<br />
[16].<br />
600O,<br />
5000-1<br />
4000-<br />
3000-<br />
2000-<br />
Ca 2+<br />
1000- Control<br />
10 20 30 40 50 60<br />
Reperfusion Time<br />
(min)<br />
FIGURE 10. Time course of the rate pressure product of<br />
hearts reperfused with UW-CSS containing Ca 2+ (0.5 mM)<br />
and without Ca 2+ (mean ± SD, n=7 per group).<br />
It has been proposed that CNBC could resuscitate the<br />
damaged heart during surgery by providing continuous<br />
delivery of oxygen and nutrients to the heart. In order to test<br />
one aspect of resuscitation, we subjected isolated hearts that<br />
had been previously stressed by a 20 minute period of<br />
normothermic ischemia to two types of cardioplegia and<br />
measured functional recovery following reperfusion [17].<br />
Twenty minutes of normothermic ischemia reduced the ATP<br />
levels, measured by 31 P NMR, in CONTROL hearts (n=6)<br />
to 70 ± 7%. These hearts recovered 86 ± 18 % of theninitial<br />
function (systolic elastance) when reperfused with<br />
normal blood perfusate. The experimental hearts were<br />
subjected to either intermittent cold blood cardioplegia<br />
(ICBC, n=6) for 5 min at 14°C, every 20 minutes, and<br />
repeated 3 times, or CNBC (n=6) for 60 minutes at a flow<br />
rate of 0.5 mL mur 1 g" 1 heart wet weight. Both ICBC and<br />
CNBC prevented exacerbation of the initial ischemic injury.<br />
PCr recovered to initial levels following reperfusion in all<br />
three groups (CONTROL, ICBC and CNBC) indicating that<br />
the mitochondria still possessed sufficient phosphorylating<br />
capacity to maintain appropriate physiological activity. ATP<br />
levels did not recover to initial levels in any of the groups.<br />
This could be related to the loss of nucleotide precursors<br />
from the cells during the initial ischemic period. Functional<br />
recovery with CNBC was 115 ± 30% compared to ICBC<br />
which was 88 ± 9% but there were no significant difference<br />
among the three groups by ANOVA (p>0.05).<br />
Is there a safe period of normothermic ischemia? From<br />
a biochemical perspective, there may be a partial answer.<br />
Using 31 P NMR spectroscopy, we have seen that during an<br />
ischemic episode, PCr decreases before ATP. Theoretical<br />
calculations using the enzymatic equilibria of the creatine<br />
kinase and adenylate kinase reactions support this<br />
observation and have shown that there are two phases of<br />
energy depletion [18]: the buffering phase and the depletion<br />
phase. During the buffering phase, energy is derived from<br />
PCr and the adenine pool is stable. During the depletion<br />
phase, energy is primarily derived from adenine nucleotides.<br />
As ATP is consumed, AMP is produced which subsequently<br />
acts as a substrate for deamination and dephosphorylation<br />
reactions, whose products are lost from the cell. Upon<br />
reperfusion, although the PCr levels may return to normal,<br />
adenine nucleotides may not reach normal levels for a<br />
number of days. To avoid imposing a metabolic stress on<br />
the myocytes, PCr levels should not be allowed to become<br />
depleted to the point where adenine nucleotides will begin to<br />
be depleted. Although the role and critical level of ATP<br />
necessary for recovery of function in the ischemic heart are<br />
controversial, it seems logical to avoid any type of<br />
preventable metabolic stress to the heart during surgery.<br />
NMR spectroscopy is useful for monitoring the energy<br />
depletion and repletion processes in model systems such as<br />
the pig heart. This information can subsequently be used to<br />
verify the predictions of the theoretical calculations.<br />
NMR can monitor PCr and ATP levels directly and<br />
continuously in the isolated perfused heart and in vivo [19].<br />
In our studies, PCr levels reflect the balance of energy<br />
supply and demand; in the arrested heart they decrease and<br />
increase in concert with the availability of oxygen. However<br />
NMR techniques are not compatible with direct use in the<br />
surgical theatre. Recently developed fiber optic pO2 probes<br />
based on oxygen quenching of fluorescence have been used<br />
21
22 Bulletin of Magnetic Resonance<br />
to monitor the arrested heart during surgery. We have<br />
performed NMR measurements on isolated blood perfused<br />
pig hearts and correlated the data with simultaneously<br />
measured levels of pO2 using custom built (Innerspace,<br />
Irvine, Calif.) NMR-compatible probes. We have found a<br />
good correlation between tissue p(>2 (measured in mmHg)<br />
and PCr level in the normal heart. The data are illustrated in<br />
Figure 11. Although the data are preliminary, we see that<br />
PCr levels drop precipitously when the tissue pO2 decreases<br />
below 30 mmHg. By establishing similar relationships in<br />
metabolically damaged or physically abnormal<br />
(hypertrophied for instance) hearts it may be possible to<br />
provide the surgeon with insight into a) the acceptable limits<br />
of oxygen deprivation should delivery of cardioplegia be<br />
discontinued during surgery; b ) the optimal flow rates of<br />
cardioplegia; c) assessment of retrograde versus antegrade<br />
delivery; and d) cardioplegic flow distribution across the<br />
heart.<br />
150, -r. 120<br />
120<br />
0 20 40 60 80 100 120 140<br />
Time (min)<br />
20 40 60 80 100 120 140<br />
pO2 (mmHg)<br />
FIGURE 11. Correlation between the pC>2 and PCr levels<br />
in the normal heart. Top: pC
Vol. 14, No. 1-4<br />
7. D.I. Hoult and R. Deslauriers. A High-Sensitivity, High<br />
Bi Homogeneity Probe for Quantification of<br />
Metabolites, Magn. Reson. Med., 16,411-417 (1990).<br />
D.I. Hoult and R. Deslauriers. Elimination of Signal<br />
Strength Dependency upon Coil Loading - An Aid to<br />
Metabolite Quantitation when Sample Volume<br />
Changes, Magn. Reson. Med., 16,418-424 (1990).<br />
G.H. Tian, G.P. Biro, B. Xiang, K.W. Butler and R.<br />
Deslauriers. The Effect of Magnesium Added to<br />
Secondary Cardioplegia on Postischemic Myocardial<br />
Metabolism and Contractile Function - A<br />
10.<br />
31 P NMR<br />
Spectroscopy and Functional Study in the Isolated Pig<br />
Heart, Basic Res. Cardiol. (in press).<br />
G. H. Tian, K.E. Smith, G.P. Biro, K.W. Butler, N.<br />
Haas, J. Scott, R. Anderson and R. Deslauriers. A<br />
Comparison of University of Wisconsin Cold Storage<br />
Solution and St-Thomas Solution II for Hypothennic<br />
Cardiac Preservation: A 31 P NMR and Functional Study<br />
of Isolated Porcine Hearts, /. Heart Lung Transplant.,<br />
10, 975-985 (1991).<br />
11 G.H. Tian, G.P. Biro, K.W. Butler, B. Xiang, C. Vu and<br />
R. Deslauriers. The Effects of Ca<br />
13<br />
14.<br />
++ Ion on the<br />
Preservation of Myocardial Energy and Function with<br />
UW Solution. A 31 P NMR Study of Isolated Pig Hearts,<br />
/. Mol. Cell. Cardiol., 24, S. 190 (1992).<br />
12. A. Panos, S.J. Kingsley, A.P. Hong, T.A. Salerno and S.<br />
Lichtenstein. Continuous Warm Blood Cardioplegia,<br />
Surg. Forum, 61, 233-235 (1990).<br />
D. Bourgeois and R. Deslauriers. Phasing Spin-Echo<br />
Acquired 31 P Spectroscopic Images Using Complex<br />
Conjugate Data Reversal, Magn. Reson. Med. (in press).<br />
A. Jasinski, P. Kozlowski, A. Urbanski and J.K.<br />
Saunders. Hexagonal Surface Gradient Coil for<br />
Localized MRS of the Heart, Magn. Reson. Med., 21,<br />
296-301 (1991).<br />
15. R. Deslauriers, S. Lareau, G.H. Tian, A.L. Panos,<br />
C.A.M. Barrozo and T.A. Salerno. Surgical<br />
Technology - Applications of Magnetic Resonance<br />
Spectroscopy to Cardiac and Transplant Surgery,<br />
Current Surg., 49, 95-101 (1992).<br />
16. R. Deslauriers, A.L. Panos, C.A.M. Barrozo, O. Al-<br />
Nowaiser, K.W. Butler, N. Haas, K.H. Teoh and T.A.<br />
Salerno. Myocardial Protection: Energy Profile During<br />
Continuous Normothermic Blood Cardioplegia,<br />
Abstracts of the Works-in-Progress, Tenth Annual<br />
Meeting of the Society of Magnetic Resonance in<br />
Medicine, San Francisco, Aug. 10-16,1991, p. 1098.<br />
17. R. Deslauriers, K.W. Butler, N. Haas, C.A.M. Barrozo,<br />
A.L. Panos, I.S. Ali, O. Al-Nowaiser and T.A. Salerno.<br />
The Effects of Intermittent Cold, and Continuous<br />
Warm, Blood Cardioplegia on Isolated Pig Hearts: 31 P<br />
NMR and Functional Studies, Abstacts of the Eleventh<br />
Annual Meeting of the Society of Magnetic Resonance<br />
in Medicine, Berlin, Aug. 8-14,1992.<br />
18.<br />
RJ. Connett. Analysis of Metabolic Control: New<br />
Insights Using Scaled Creatine Kinase Model, Am. J.<br />
Physiol, 254, R949-R959 (1988).<br />
19. P.A. Bottomley, CJ. Hardy and P.B. Roemer.<br />
Phosphate Metabolite Imaging and Concentration<br />
Measurements in Human Heart by Nuclear Magnetic<br />
Resonance, Magn. Reson. Med., 14,425-434 (1990).<br />
23
24<br />
1. Introduction<br />
Topology and Spin Alignment in Organic<br />
High-Spin Molecules<br />
Yoshio Teki, Kazunobu Sato, Masayuki Okamoto, Atsuya Yamashita,<br />
Yoji Yamaguchi, Takeji Takui, Takamasa Kinoshita, and Koichi Itoh<br />
Department of Chemistry, Faculty of Science, Osaka City University,<br />
Sugimoto 3-3-138, Sumiyoshi-ku, Osaka 558, Japan<br />
Organic high-spin molecules are ideal model compounds<br />
for organic magnetic materials such as organic superparaand<br />
ferro-magnets which have recently been attracting<br />
increasing interest [1]. The related experimental [2] and<br />
theoretical works [3] have been done to serve their<br />
molecular design for the last two decades. High-spin<br />
polycarbenes as organic high-spin systems, in spite of their<br />
highly chemical reactivity, are very important from the<br />
view point .of organic magnetism as well as of spin<br />
ordering/spin control in chemistry from the following<br />
reasons, (i) One of the most prominent features of the highspin<br />
polycarbenes is multi-electron open shell systems in<br />
the ground or low-lying excited states which arise from<br />
degenerate delocalized 7t orbitals and from o dangling<br />
orbitals localized at divalent carbon atoms, the latter<br />
orbitals being nearly degenerate with the highest half-filled<br />
K orbitals. (ii) The degeneracy of their n orbitals is<br />
governed by a particular connectivity of n electron network,<br />
i.e. by the topology of the rc electron network.<br />
We have been studying a series of high-spin polycarbenes<br />
as well as their n topological isomers [2a,2c,4-19] which<br />
are designed by exploiting their % electron networks. We<br />
define a Ji-topological isomer as a molecule which differs<br />
from others only in the topology of its TC electron network,<br />
i.e. in the linking positions of its TC bonds. These high-spin<br />
carbenes have been successfully detected up to a tridecet<br />
polycarbene having twelve parallel spins (S=6) by means of<br />
ESR spectroscopy 118].<br />
In this work, we have studied the spin density distributions<br />
of high-spin polycarbenes and their topological<br />
isomers, biphenyl-n,n'-bis(phenylmethylene) (I: n,n'=3,3';<br />
II: n,n'=3,4') and m- and p-phenylenebis(phenylmethylene),<br />
in order to clarify the mechanism of the spin correlation<br />
and its role in the intramolecular spin alignment of organic<br />
systems.<br />
2. Topological Isomers and Their<br />
Spin States<br />
Figure 1 shows the ground states and the low-lying<br />
excited states of the high-spin polycarbenes and their<br />
topological isomers studied in this work. They were determined<br />
from our previous ESR experiments [2a,5,15,<br />
17]. Molecule I has a unique spin alignment in that it has<br />
low-lying high-spin states (S=l and S=2) above the low-<br />
Bulletin of Magnetic Resonance<br />
Figure 1. Topological Isomers and their spin states<br />
S=l<br />
S=0<br />
spin ground state (S=0) as shown in this figure. In<br />
contrast, its topological isomer II has the high-spin ground<br />
state. Both molecules have four unpaired electron spins in<br />
their nearly degenerate nonbonding n and o orbitals. On the<br />
other hand, molecule III is the first organic high-spin<br />
system with the quintet ground state (S=2) [2a], while its<br />
topological isomer IV has the low-spin ground state and the<br />
low-lying triplet excited state located ca. 200 cm"' above<br />
the ground state.<br />
3. Experimental<br />
The syntheses of the diazo precursors of III and IV were<br />
carried out according to the literatures [20,21]. The<br />
synthetic work of the carbon 13 labeled compounds of<br />
biphenyl-3,3'-bis(phenylmethylene) and other precursors<br />
will be published elsewhere. Each diazo precursor was<br />
diluted in a host single crystal of benzophenone-di()- Mixed<br />
single crystals were grown in the dark by slowly cooling an<br />
ethanol or an ether solution containing the diazo precursor.<br />
The polycabenes were generated at 2 - 4 K by the<br />
photolysis of the corresponding diazo precursors. The<br />
photolysis was carried out with an XBO 500W high pressure<br />
mercury lamp using a quartz rod which guided the 405<br />
nm light into an X-band TMon ENDOR cavity. All the<br />
EPR and ENDOR spectra were recorded with a Bruker ESP<br />
300/350 spectrometer equipped with an Oxford variable temperature<br />
controller ESR910. The spectra of the low-lying<br />
excited states were observed under the thermal excitation of<br />
these levels. Other spectra were measured at ca. 2 - 4 K.
Vol. 14, No. 1-4 25<br />
4. Theoretical<br />
In addition to the ENDOR work, we have calculated the<br />
spin density distributions of molecules I - VI using two<br />
model Hamiltonian approaches. The following unrestricted<br />
Hartree-Fock calculation based on a generalized Hubbard<br />
model [8] as well as the exact numerical solution of a<br />
valence-bond Heisenberg Hamiltonian [9] have provided<br />
rather satisfactory and complementary descriptions for spin<br />
structures of organic high-spin polycarbenes.<br />
The generalized Hubbard model Hamiltonian is given by<br />
[22]<br />
ft - -l Zj
26<br />
(a) 34 K<br />
QAt<br />
(b) 15K<br />
24.0<br />
(c) I5K<br />
—I<br />
0.2<br />
QB+<br />
T: Triplet<br />
Q: Quinlet<br />
QB-<br />
0.4 0.6<br />
MAGNETIC FIELD/T<br />
26.0 28.0 30.0<br />
FREQUENCY / MHz<br />
52.0 56.0 60.0 80.0<br />
FREQUENCY / MHz<br />
84.0 88.0<br />
Figure 2. Typical EPR, l H- and 13 C-ENDOR spectra of<br />
molecule I. (a) EPR spectrum, (b) ^-ENDOR spectrum<br />
(c) l^C-ENDOR spectrum. The external magnetic field is<br />
along the a axis. The microwave frequency v is 9413.0<br />
MHz for the spectra (a) and (b), and 9456.0MHz for the<br />
spectrum (c) of the carbon 13 labeled compound,<br />
respectively.<br />
the two end phenyl groups were deuterated, in order to<br />
facilitate the assignment of 1H-ENDOR transitions by<br />
the reduction of the spectral density. By a comparison of<br />
the ^H-ENDOR spectra of the deuterium labeled compound<br />
with those of the normal compound, the eight transitions<br />
labeled by the asterisk in figure 2(b) were assigned to the<br />
protons of the central biphenyl group and the remaining<br />
unlabeled transitions to those of the end phenyl groups.<br />
Thus, we have achieved reliable assignment for all the *H-<br />
ENDOR signals observed.<br />
On the basis of the assignment above, we have<br />
determined the K spin densities on the carbon sites<br />
adjacent to the hydrogen atoms and the n and a spin<br />
densities on each divalent carbon atom from the analysis of<br />
the angular dependence of the ENDOR frequencies as<br />
described in section 5. The experimentally obtained spin<br />
densities are given in figure 3(a). The spin densities on the<br />
six carbon atoms without circles could not be obtained,<br />
since they have no adjacent protons.<br />
We have also calculated the spin density distribution<br />
theoretically by two model Hamiltonian approaches<br />
[8,9,19] as described in section 4. The calculated spin<br />
Bulletin of Magnetic Resonance<br />
density distribution based on the generalized Hubbard model<br />
is shown in figure 3(b). In this calculation, we used the<br />
weakly interacting model in which molecule I, biphenyl-<br />
3,3'-bis(phenylmethyIene), is regarded as composed of two<br />
diphenylmethylene moieties, unit A and unit B, weakly<br />
interacting with each other. This model has shown to<br />
interpret well the observed particular relationship Dj = -<br />
3DQt as described in our previous work [5J. We have<br />
applied this model to account for the spin density<br />
distribution of molecule I. The spin densities p(S,Ms) in<br />
molecule I can be derived from p0(SA=l,Ms=l) of the<br />
isolated triplet diphenylmethylene moieties using the<br />
equation [15,19]<br />
p(S=l,Ms=l)i = (l/2)pO(SA=l,Ms=l)i (6)<br />
for the triplet state, and<br />
p(S=2,Ms=2)j = p°(SA=l,Ms=l)i (7)<br />
for the quintet state. Similar expressions hold also for unit<br />
B since A and B are equivalent. The spin densities of the<br />
isolated diphenylmethylene moieties were obtained from the<br />
UHF calculation based on the generalized Hubbard model.<br />
This calculation well interprets the ENDOR results as<br />
shown in figure 3. The observed and calculated spin density<br />
distributions show that the sign of the n spin density is<br />
alternatively distributed on the carbon sites within the<br />
diphenylmethylene moiety, thus forming the up-and-down<br />
(a) O : positive<br />
0.093<br />
0.105<br />
-0.019 n -086 0.092 -0.072<br />
0.096 0.098<br />
: negative<br />
0126 -0.075<br />
0.095<br />
0.105<br />
0.711 0.711<br />
Figure 3 Spin density distribution of the thermally excited<br />
triplet state of molecule I. (a) Experimental values, (b)<br />
Theoretical values obtained from the UHF generalized<br />
Hubbard calculation based on the weakly interacting model.<br />
(c) Theoretical values obtained by the valence-bond<br />
Heisenberg Hamiltonian approach.
Vol. 14, No. 1-4<br />
network of the K spin. We define the up-and-down network<br />
of the % spin as the pseudo spin density wave (pseudo-<br />
SDW), since in infinite systems the up-and-down spin<br />
network forms a spin density wave. This network resulting<br />
from spin correlation is most favorable in view of the total<br />
spin energy (the sum of the spin exchange-correlation<br />
energy). The observed spin density distribution is, in<br />
principle, symmetrical with respect to the center of the<br />
molecule as predicted by the weakly interacting model<br />
above. Thus, the pseudo 7t-SDW is formed within each<br />
diphenylmethylene moiety. As a result of these, the central<br />
two carbon sites of the biphenyl group (the bridged carbons)<br />
should have the same (negative) sign. This violates the upand-down<br />
pseudo JI-SDW in the whole molecule, resulting<br />
in a node of spin density distribution at the central bridged<br />
carbons. The existence of this node, which unstabilizes the<br />
spin-correlation energy, makes the observed triplet state to<br />
be an excited state above the spin-less ground state. A<br />
similar spin distribution is also expected for the quintet<br />
state from the comparison of eqs. (6) and (7).<br />
Figure 3(c) shows the spin density distribution calculated<br />
by the Heisenberg model Hamiltonian approach where we<br />
replaced the end phenyl groups of molecule I with the<br />
hydrogen atoms to reduce the dimensionality of the<br />
Hamiltonian matrix. In this calculation, we have obtained<br />
the results similar to figure 3(b) without using the weakly<br />
interacting model. In this approach, the spin correlation<br />
is exactly taken into account within the Heisenberg model,<br />
leading to the correct ordering of the ground state and the<br />
low-lying excited states [9,11].<br />
(B) ESR. ENDOR and Spin Density Distribution<br />
of Molecule III<br />
As mentioned above, the lowest spin ground state, i.e.<br />
the singlet ground state, is realized in molecule I, since the<br />
high-spin states are unstabilized because of the existence of<br />
the node at the central bridged carbons of the biphenyl<br />
group. However, if the position of the bridge is shifted by<br />
one carbon site as shown in figure 5, it is expected that the<br />
nodeless pseudo rc-SDW is formed in the whole molecule,<br />
leading to the high-spin ground state as a result of the<br />
stabilization of spin correlation energy. In order to confirm<br />
this, we have observed the ^H-ENDOR spectra of biphenyl-<br />
3,4'-bis(phenylmethylene) (molecule II). It was shown by<br />
our previous ESR experiments that the ground state of<br />
molecule II was a high-spin (S=2) state without low-lying<br />
excited states [17]. The fine structure parameters and the g<br />
value of the quintet ground state were determined as<br />
D=+0.1250 cm" 1 , E=-0.0065 cm" 1 , and g=2.003<br />
(isotropic). Figures 4(a) and (b) show a typical ESR and<br />
H-ENDOR spectra of the quintet ground state of molecule<br />
II observed at 2.7 K. The ENDOR spectrum was obtained<br />
by monitoring the B+ ESR transition (Ms=0*^+l). Three<br />
transitions by o in figure 4(b) correspond to the *H-<br />
ENDOR signals arising from the Ms = 0 spin sublevel and<br />
the remaining unlabeled signals to those from the Ms=+1<br />
sublevel. The eight transitions labeled by the asterisk<br />
are due to the protons of the central biphenyl group.<br />
(a) 2.7K<br />
A-<br />
B-<br />
0.20 0.30 0.40<br />
MAGNETIC FIELD/T<br />
(b)<br />
* *<br />
12.0 16.0 20.0 24.0<br />
FREQUENCY/MHz<br />
Figure 4. Typical ESR and 'H-ENDOR spectra of molecule<br />
II. (a) ESR spectrum, (b) I H-ENDOR spectrum. The<br />
external magnetic field is along the b axis. The microwave<br />
frequency vis 9378.7 MHz.<br />
(a)<br />
Q : positive<br />
0 : negative .o.os8 °- 224<br />
-0.064 0.218<br />
-0.138 0.198<br />
0.206<br />
(b) "205<br />
-0.154 0.230<br />
-0.180 "-I 54<br />
HIT = 2.0. JIT = 0.25<br />
-0.076<br />
0.151<br />
Figure 5 Spin density distribution of the quintet ground<br />
state of molecule II. (a) Experimental values, (b)<br />
Theoretical values obtained from the UHF generalized<br />
Hubbard calculation.<br />
The spin densities obtained from McConnell's equation<br />
- Qpi n /2S and the theoretical values calculated on the basis<br />
of the Hubbard model are given in figures 5(a) and 5(b),<br />
respectively. The observed and calculated spin density<br />
27
28 Bulletin of Magnetic Resonance<br />
distributions indicate that the nodeless pseudo rc-SDW is<br />
formed in the whole molecule as expected from the<br />
topology of the n electron network of molecule II.<br />
These findings give the following physical picture for<br />
the intramolecular spin alignment of molecule II: The unpaired<br />
n electrons are distributed over the carbon skeleton<br />
with alternating the sign of the spin density from carbon to<br />
carbon, thus forming the pseudo SDW in the K electron<br />
network. On the other hand, the two unpaired a spins in<br />
the localized o dangling orbitals become parallel to each<br />
other as a results of the ferromagnetic coupling to the<br />
unpaired n spins at each divalent carbon site, since the onecenter<br />
exchange integral / in eq. (3) between the nearly<br />
degenerate a and n orbitals on the same carbene site is<br />
usually ferromagnetic. Thus, this picture shows that the<br />
spin alignment in molecule II is predominantly determined<br />
by the formation of the pseudo rc-SDW.<br />
(O ESR. ENDOR and Spin Density Distribution<br />
of Molecule III and IV<br />
To demonstrate the role of spin correlation as determined<br />
by the topology of n electron networks, we have also<br />
investigated the spin density distribution in the quintet<br />
ground state (S=2) of the first organic, high-spin molecule,<br />
m-phenylenebis(phenylmethylene) III, and that in the lowlying<br />
excited triplet state (S=l) of its topological isomer, pphenylenebis(phenylmethylene)<br />
(molecule IV). In the<br />
previous work, we reported ESR studies for III and IV<br />
[2a,5] and ^-ENDOR experiments for III [2c]. Figures<br />
6(a)and6(b) show typical ESR and 'H-ENDOR spectra<br />
of III observed at 2 K. The signals labeled by the circle (o)<br />
in the ESR spectrum are those due to a minor byproducts.<br />
The primed pairs A'+ and B'±, and the unprimed pairs A+<br />
and B+ arise from the two magnetically nonequivalent sites<br />
occupied by the guest molecule III in the host single crystal.<br />
A typical ^C-ENDOR spectrum is shown in figure<br />
6(c). We have roughly estimated the n and a spin<br />
densities at each divalent carbon atom from the angular<br />
dependence of the * ^C-ENDOR frequencies in figure 6(c);<br />
the complete analysis by the numerical diagonalization of<br />
the spin Hamiltonian (4) is in progress. The anisotropy of<br />
the l^C hyperfine tensor is about one half of that of<br />
diphenylmethylene reported by Hutchison et al. 126]. This<br />
is due to the projection factor 1/(2S) in eq. (5). Our<br />
preliminary results indicate that the spin densities on both<br />
divalent carbons have values similar to those of<br />
diphenylmethylene. Thus, the sign of the spin densities on<br />
each divalent carbon site is determined to be positive as<br />
expected from the topology of the n electron network of<br />
III, and their n spin densities were estimated to be similar<br />
in magnitude. The K spin densities on the other carbon<br />
sites having adjacent protons were also determined from the<br />
isotropic term of each proton hyperfine tensor using<br />
McConnell's relation. The spin-density distribution<br />
experimentally determined showed that the pseudo 7t-SDW<br />
is formed in the n electron network in a manner similar<br />
to the ground state of II.<br />
A question arises as for the low-lying triplet excited state<br />
of IV whether there exists a node in the n spin distribution,<br />
since II and IV have similar spin structures, i.e. low-lying<br />
high-spin excited states above the low-spin (singlet) ground<br />
state (figure 1). To answer this question, we measured the<br />
^H-ENDOR spectra of the low-lying triplet excited state of<br />
p-phenylenebis(phenylmethylene) (molecule IV). Its typical<br />
ESR and *H-ENDOR spectra are shown in figures 7(a) and<br />
7(b), respectively. The spin density distribution obtained<br />
from the analysis of the angular dependence of the ENDOR<br />
signals is given in figure 7(c). This shows that the four<br />
carbon sites in the central phenyl ring have the same<br />
negative sign in the K spin density, leading to a node in the<br />
7t spin density distribution in the central ring. This finding<br />
confirms the physical picture for the formation of low-lying<br />
high-spin states above a low-spin/spin-less ground state in<br />
topological isomers of high-spin polycarbebes, as described<br />
in section 6(A): The node of the spin distribution<br />
violates the formation of the pseudo rc-SDW in the<br />
whole molecule. This unstabilizes the spin correlation<br />
energy, leading to the observed triplet state above the<br />
singlet ground state.<br />
0.1 0.2 0.3 0.4<br />
8.0 12.0 16.0<br />
FREQUENCY/MHz<br />
(c)<br />
6ab = 5 Temp. 3K<br />
20.0<br />
40.0 44.0 48.0 60.0 64.0 68.0<br />
FREQUENCY / MHz<br />
Figure 6. Typical ESR, 1 H- and l3 C-ENDOR spectra of<br />
molecule III. (a) ESR spectrum, (b) JH-ENDOR<br />
spectrum, (c) 13 C-ENDOR spectrum. The external<br />
magnetic field is along ©ab= 5° from the a axis. The<br />
microwave frequency v is 9430.4 MHz.
Vol. 14, No. 1-4 29<br />
1.2.0<br />
(c)<br />
0.28 0.32 036 ' O40<br />
MAGNETIC FIELD/T<br />
16.0 20.0 24.0<br />
FREQUENCY / MHz<br />
O Positive Spin (P = O.l9i - 0.293)<br />
9 Negative Spin (p = -o.oio~-o.08i)<br />
IF.NDOR measurement<br />
28.0<br />
Figure 7. Typical ESR and ' H- ENDOR spectra and spin<br />
density distribution of the low-lying triplet excited state of<br />
molecule IV. (a) ESR spectrum, (b) JH-ENDOR spectrum.<br />
The external magnetic field is along 0ab= 12° from the a<br />
axis. The microwave frequency v is 9457.0 MHz. (c)<br />
Experimentally determined spin densities.<br />
7. Conclusions<br />
We have investigated the spin alignment in the ground<br />
states and the low-lying excited states of the high-spin<br />
polycarbenes and their topological isomers shown in figure<br />
1. Their spin density distributions were determined by<br />
single-crystal 1 H- and ^G-ENDOR as well as by the<br />
theoretical calculations using the two model Hamiltonians<br />
(the generalized Hubbard model and the valence-bond<br />
Heisenberg model). The results obtained in this work can<br />
be summarized as follows: (1) It is shown that the spin<br />
alignment is highly dependent on the topological nature in<br />
the K electron network. (2) The pseudo JC-SDW governed<br />
by the topological nature plays an important role in the<br />
stabilization of the high-spin ground state. (3) The spin<br />
correlation also plays essential part for the formation of the<br />
low-lying spin states of molecule I and IV, as well as of<br />
the high-spin ground states of molecule II and HI. (4) The<br />
physical picture of the intramolecular spin alignment in<br />
polycarbenes has been clarified in view of spin correlation.<br />
8. References<br />
1. J.S. Miller and D.A. Dougherty, eds.. Proceedings of the<br />
symposium on Ferromagnetic and High-Spin Molecular<br />
Based Materials, 197th ACS Meeting, Dallas, USA (April<br />
9-12, 1989); Mol. Cryst. Liquid Cryst. 176, 1-562 (1989).<br />
2. (a) K. Itoh, Chem. Phys. Lett., 1, 235 (1967). (b) E.<br />
Wasserman et al., J. Am. Chem. Soc, 89, 5076 (1967).<br />
(c) T. Takui, S. Kita, S. Ichikawa, Y. Teki, T. Kinoshita,<br />
and K. Itoh, Mol. Cryst. Liquid Cryst., 176, 67 (1989),<br />
and references cited therein.<br />
3. (a) K. Itoh, Bussei, 12, 635 (1971). (b) N. Mataga,<br />
Theoret. Chim. Acta (Berl.), 10, 372 (1968). (c) N.<br />
Tyutyulkov, G. Olbvich, H. Brenzen, O, Polansky, Theoret.<br />
Chim. Acta, 73, 27 (1988), and references cited therein.<br />
4. T.Takui and K. Itoh, Chem. Phys. Lett.,19, 120 (1973).<br />
5. K. Itoh, Pure & Appl. Chem., 50, 1251 (1978).<br />
6. Y. Teki, T. Takui, K. Itoh, and H. Iwamura, J. Chem.<br />
Phys., 83, 539 (1985).<br />
7. Y. Teki, T. Takui, K. Itoh, H. Iwamura, and K.<br />
Kobayashi, J. Am. Chem. Soc, 108, 2147 (1986).<br />
8. Y. Teki, T. Takui, T. Kinoshita, S. Ichikawa, H. Yagi,<br />
and K. Itoh, Chem. Phys. Lett., 141, 201 (1987).<br />
9. Y. Teki, T. Takui, M. Kitano, and K. Itoh, Chem. Phys.<br />
Lett., 142, 181 (1987).<br />
10. Y. Teki, T. Takui, and K. Itoh, J. Chem. Phys., 88,<br />
6134(1988).<br />
11. K. Itoh, T. Takui, Y. Teki, and T. Kinoshita, J. Mol.<br />
Electronics, 4, 181 (1988).<br />
12. K. Itoh, T. Takui, Y. Teki, and T. Kinoshita, Mol.<br />
Cryst. Liquid Cryst, 176, 49 (1989).<br />
13. I. Fujita, Y. Teki, T. Takui, T. Kinoshita, K. Itoh, F.<br />
Miko, Y. Sawaki, H. Iwamura, A. Izuoka, and T.<br />
Sugawara, J. Am. Chem. Soc, 112, 4047 (1990).<br />
14. M. Matsushita, T. Momose, T. Sida, Y. Teki, T.<br />
Takui, and K. Itoh, J. Am. Chem..Soc, 112, 4701 (1990).<br />
15. M. Okamoto, Y. Teki, T. Takui, T. Kinoshita, and K.<br />
Itoh, Chem. Phys. Lett., 173, 265 (1990).<br />
16. M. Matsushita, T. Nakamura, T. Momose, T. Sida, Y.<br />
Teki, T. Takui, T. Kinoshita, and K. Itoh, J. Am. Chem.<br />
Soc, in press (1992).<br />
17. Y. Teki, I. Fujita, T. Takui, T. Kinoshita, and K. Itoh,<br />
J. Am. Chem. Soc, submitted (1992).<br />
18. K. Furukawa, T. Matsumura, Y. Teki, T. Kinoshita, T.<br />
Takui, and K. Itoh, J. Am. Chem. Soc, submitted (1992).<br />
19. Y. Teki, M. Okamoto, T. Takui, T. Kinoshita, and K.<br />
Itoh, J. Am. Chem. Soc, submitted (1992).<br />
20.R. W. Murray and A. M. Trozzolo, J. Org. Chem., 26,<br />
3109 (1961).<br />
21. S. I. Murahashi, Y. Yoshimura, Y. Yamamoto, and I.<br />
Moritani, Tetrahedron, 28, 1485 (1972).<br />
22. (a)R. M. White, Quantum Theory of Magnetism;<br />
Springer, p. 139 (1983). (b) K. Nasu, Phys. Rev., B33,<br />
330 (1986).<br />
23. S. A. Alexander and D. J. Klein, J. Am. Chem. Soc,<br />
110, 3401 (1988).<br />
24. N. Hirota, C. A. Hutchison Jr. and P. Palmer, J.<br />
Chem. Phys., 40, 3717 (1964).<br />
25. M. E. Rose, Elementary Theory of Angular<br />
Momentum, John Wiley & Sons: New York (1957).<br />
26. C. A. Hutchison Jr., B. E. Kohler, J. Chem. Phys.,<br />
51, 3327 (1979).
30 Bulletin of Magnetic Resonance<br />
New Developments in<br />
Pulsed Electron Paramagnetic Resonance:<br />
Relaxation Mechanisms of<br />
Nitroxide Spin Labels<br />
Colin Mailer, Bruce H. Robinson, and Duncan A. Haas<br />
Department of Chemistry<br />
University of Washington<br />
Seattle WA 98195<br />
Recent technical developments in this laboratory in the areas of pulsed Saturation Recovery Electron<br />
Paramagnetic Resonance (SR-EPR) and Saturation Recovery Electron-Electron Double Resonance (SR-ELDOR)<br />
have enabled experiments with these two techniques to be done with high sensitivity. We have studied the mechanisms<br />
of the relaxation rates of spin labels. The electron spin-lattice relaxation rate Tic' 1 and the nitrogen spinlattice<br />
relaxation rate (Tin" 1 ) of per-deuterated 15 N TEMPOL in glycerol-water solutions have been measured by<br />
SR-EPR and SR-ELDOR. The motional range covered is from a few picoseconds to hundreds of nanoseconds and<br />
the motion is characterized by simple Brownian dynamics. The dependence of Tie 1 upon the rotational<br />
correlation time is explained by a combination of spin rotation and electron -nuclear dipolar coupling mechanisms<br />
plus a proton spin diffusion process. Tin-1 is explained by the electron-nuclear dipolar mechanism and proton<br />
spin diffusion.<br />
1 Introduction<br />
Given the utility of nitroxide spin labels as a probe<br />
of molecular motion in the biological and physical<br />
sciences for the past 25 years it is somewhat surprising<br />
that a clear understanding of their relaxation<br />
mechanisms is still unknown. The quantities<br />
desired are the spin-lattice relaxation rates both<br />
electronic (Tie-i) and nuclear (TV 1 - 14 N or 15 N)<br />
of the nitroxide label. Pulsed EPR can be used to<br />
monitor such relaxation processes directly regardless<br />
of T/2e or line inhomogeneity. In a pulsed<br />
Saturation Recovery (SR) experiment a high rf<br />
field pump pulse is applied for a short time and the<br />
recovery of the magnetization is measured with a<br />
low power observer at the same frequency. The<br />
pulsed Electron-Electron Double Resonance (SR-<br />
ELDOR) experiment has the pump and observer<br />
frequencies quite different (commonly pump and<br />
observer are set to resonate on different spin manifolds).<br />
Huisjen and Hyde [1] pioneered the use<br />
of the SR-EPR technique in liquids and applied it<br />
to a number of systems [2], [3], [4], [5], [6]. The<br />
1 MHuisjen and J.S.Hyde. Rev. Sci. lost., 45, 669-675,<br />
1974.<br />
2 P.W. Percival and J.S. Hyde. /. Mag. Res., 23,249-<br />
257, 1976.<br />
3 T. Sarna and J.S. Hyde. J. Chem. Phys., 69, 1945-<br />
1948, 1978.<br />
Hyde group has used SR-EPR in conjunction with<br />
Continuous Wave (CW) - ELDOR to measure lateral<br />
diffusion of 14 N labelled lipids in bilayers [7],<br />
[8], [9]. Hyde et al. [10] have observed the transfer<br />
of energy from one line to another line in a<br />
CTPO 14 N spin label with pulsed SR-ELDOR.<br />
They clearly saw that cross relaxation took place,<br />
and measured the cross-relaxation rate. We discuss<br />
this experiment below.<br />
Freed and co-workers [11], [12], [13], [14] analyzed<br />
the EPR spectra of peroxylamine disulphon-<br />
10<br />
11<br />
C. Altenbach, W. Froncisz, J.S. Hyde, and W.L.<br />
Hubbell. Biophys. J., 56, 1183-1191, 1989.<br />
J-J. Yin and J.S. Hyde. Zeitschrift fur Physikalische<br />
Chemie. 153, 57-65, 1987.<br />
A. Kusumi, W.K. Subczynski and J.S. Hyde. Proc.<br />
Nat. Acad. Sci. USA, 79, 1854-1858, 1982.<br />
C.A. Popp and J.S. Hyde Proc. Nat. Acad. Sci. USA,<br />
79, 2559-2563, 1982.<br />
J.B. Feix, C.A. Popp, S.D. Venkataaramu, A.H. Beth,<br />
J.H. Park and J.S. Hyde. Biochemistry, 23, 2293-<br />
2299, 1984.<br />
J.-J. Yin, M. Pasenkiewicz-Gierula, and J. S. Hyde<br />
Proc. Nat. Acad. Sci. USA, 84, 964-968, 1987.<br />
.S. Hyde, W. Froncisz and C. Mottley Chem. Phys.<br />
Lett., 110, 621-625, 1984.<br />
J.H. Freed, in Spin Labelling: Theory and<br />
Applications vol I ed. LJ.Berliner. Academic Press<br />
1976. Chapter 3. pp.53-132.
Vol. 14, No. 1-4 31<br />
ate ( 14 N-PADS) and deuterated 2,2,6,6-tetramethyl-4-piperidinone-l-oxyl<br />
(pd-TEMPONE) in<br />
glycerol-water mixture solvents. The linewidths<br />
as a function of temperature were obtained,<br />
corrected for inhomogeneous broadening, and<br />
analyzed according to: l/T2e(M)= A+ B • M +<br />
C-M 2 where l/T2e(M) is the homogeneous<br />
line width of the Mth line (M= -1, 0, +1) in the<br />
three line spectrum. Spectral simulations and<br />
calculation of the A, B and C parameters were<br />
done using Redfield theory [15]. The best<br />
simulated EPR spectral fits to the experimental data<br />
for correlation times (TR) slower than 10" 8 sec<br />
used a non-Brownian spectral density function<br />
This extremely detailed and careful study took the<br />
linewidth analysis method to its limits, but this<br />
alone was not sufficient to completely determine<br />
the motion. The molecular dynamics was inferred<br />
from plots of the A, B, and C parameters versus<br />
each other or viscosity and temperature. Subtle<br />
changes in slopes indicated deviations from<br />
isotropic motion. The simulations to test motional<br />
models were not able to unequivocally determine<br />
the type of motion either. This work presents an<br />
independent test of the relaxation rates and the correlation<br />
times, as estimated by the CW-EPR lineshape<br />
analysis.<br />
2 Experimental<br />
Spectrometer The 9.3 GHz (X-band) pulsed<br />
EPR spectrometer for these studies follows published<br />
designs [16], [17]. There are three arms -<br />
pump, observer and detector bias. The observer<br />
and bias arms act as a conventional high sensitivity<br />
spectrometer for both linear EPR and ST-EPR experiments.<br />
The pump arm. klystron is phase<br />
locked to the observe klystron. For Free Induction<br />
decay (FID) measurements the frequency difference<br />
is zero. SR-EPR and SR-ELDOR experiments<br />
have the pump-observer frequency differ-<br />
12<br />
13<br />
S.A. Goldman.G.V. Bruno.CF. Polnaszek, and J.H.<br />
Freed, J. Chem. Phys., 56, 716-735, 1972.<br />
S.A. Goldman,G.V. Bruno, and J.H. Freed,. /. Chem.<br />
Phys., 59, 3071-3091, 1973<br />
J.S. Hwang,R. Mason,L.P. Hwang, and J.H. Freed, J.<br />
Phys. Chem., 79, 489-511, 1975.<br />
Spin labelling 1 chapter 2 Freed<br />
C Mailer, J.D.S. Danielson and B.H. Robinson. Rev.<br />
Sci. Inst., 56, 1917-1925, 1985.<br />
C Mailer, D.A. Haas, EJ. Hustedt, J.G. Gladden, and<br />
B.H. Robinson, J. Mag. Res., 91, 475-496, 1991.<br />
ence phase locked to a low frequency (MHz) oscillator.<br />
SR-ELDOR operation was simplified by the<br />
use of a Loop Gap Resonator (LGR) [18] instead<br />
of a bimodal EPR cavity. The Q of the LGR is<br />
approximately 300 which gives a 3 dB resonator<br />
bandwidth of 30 MHz. Figure 1 shows the measured<br />
value of rf field versus offset from the 9.3<br />
GHz resonant frequency of the LGR. It is clear<br />
that at the 60 - 70 MHz offset needed for SR-<br />
ELDOR the rf field in the resonator is about 50%<br />
of maximum<br />
20 40 60 80 100<br />
OFFSET FREQUENCY MHz<br />
Figure 1 Plot of 1 mm Loopgap Resonator Bandwidth.<br />
The data are the relative heights of the FID produced by<br />
a short pulse of rf power on the resonator at the frequency<br />
offset from LGR resonance as indicated on the<br />
abscissa. The solid line is the relative rf field amplitude<br />
produced by a resonator with a Q of 300.<br />
The small size of the LGR produces a high power<br />
density leading to high rf fields for moderate powers<br />
(less than 1 Watt).Typical experimental conditions<br />
were: pump power of 200 mW, observer<br />
power 100 nWatt, dead time 50 nanoseconds, acquisition<br />
time 2 nanoseconds/point (or longer) for<br />
1024 points. The pulse repetition rate was about 3<br />
kiloHertz. The decay curves took about 50<br />
seconds to obtain.<br />
CW-EPR We have carried out a CW-EPR and<br />
pulsed EPR study of per-deuterated 15 N TEMPOL<br />
in glycerol-water mixtures. The EPR linewidths in<br />
the fast motion region were measured and corrected<br />
18 Medical Advances, Milwaukee, Wis Loopgap<br />
Resonator Model #XP-0201
32 Bulletin of Magnetic Resonance<br />
Figure 2A. Pulsed ELDOR response of l5 N TEMPOL in 40 % glycerol- 60% water at a TR of 0.015 nanoseconds High field<br />
line pumped with a 250 mW 100 nanosecond duration pulse and the low field line observed with 100 microwatts in the LGR.<br />
Figure 2B. Pulsed. Saturation Recovery. Same experimental conditions as 2A except that pump and observer are both set to<br />
the low field line.<br />
for inhomogeneous broadening using Bales'<br />
method [19] to obtain the motionally dependent<br />
Lorentzian linewidths. These line widths when<br />
subtracted gave the B term of the A(M) = A + B<br />
M + C • M^ expression. (Taking the difference<br />
removed residual Lorentzian effects common to<br />
both lines - such as Heisenberg exchange [20]).<br />
To obtain longer correlation times the high glycerol<br />
percentage samples (> 85%) were cooled to subzero<br />
temperatures and the rotational correlation<br />
time (xR) was estimated from the Stokes-Einstein<br />
equation: tR = V • TJ/T where r\ and T are the<br />
19 B. Bales in Spin Labeling: Theory and Applications,<br />
Biological Magnetic Resonance vol 8 eds. L.J. Berliner<br />
and J. Reuben. Plenum Press NY 1989 Chapter 3.<br />
20 S. Lee and A. Shetty. J. Chem. Phys., 83, 499-505,<br />
1985.<br />
viscosity and temperature of the solvent, and V is<br />
the hydrodynamic volume of the spin probe. This<br />
information calibrates the motion and connects TR<br />
with the temperature and percent glycerol.<br />
Time Domain EPR An example of a pulsed<br />
ELDOR experiment is shown in Figure 2A. The<br />
magnetization from the pumped manifold arrives at<br />
the observer field position at rate Tin-l and then<br />
decays slowly to thermal equilibrium at rate Tie' 1 -<br />
A Saturation Recovery decay is shown in Figure<br />
2B. Some of the magnetization leaves for the<br />
other manifold (at the same rate as it arrived in the<br />
pulsed ELDOR experiment) and the remainder returns<br />
to thermal equilibrium by the Tie process.<br />
The opposite sign of the faster rate in the two experiments<br />
shows that it arises from cross-relax-
Vol. 14, No. 1-4 33<br />
ation between the nuclear manifolds. Superimposed<br />
on the data are the least-squares, best fits<br />
which assume two independent exponential<br />
relaxation components and a baseline. The relaxation<br />
rates of each component, the amplitudes<br />
and the baseline were adjusted using a Marquardt<br />
non-linear least squares algorithm [21].<br />
The electron and the nitrogen spin-lattice relaxation<br />
rates of per-deuterated 15 N TEMPOL in glycerolwater<br />
have been measured with pulsed ELDOR<br />
and SR. The TR range covered is from a few picoseconds<br />
to tens of nanoseconds as indicated on<br />
Figure 3. Also plotted are electron spin-lattice relaxation<br />
rates of *% TEMPOL obtained in the<br />
early pulsed experiments of Percival and Hyde [2].<br />
The Tje" 1 are similar for similar correlation times,<br />
despite the fact that glycerol-water mixtures between<br />
room temperature and -20°C were used in<br />
our experiments, and the Percival and Hyde results<br />
were obtained between -20°C and -90°C with the<br />
label in sec-buty\ benzene. This similarity suggests<br />
;<br />
_<br />
Id U)<br />
to z<<br />
I oUJ<br />
I<br />
UJ<br />
<<br />
,10 -12<br />
10*<br />
10r<br />
2<br />
10<br />
10" 10" 10" 10-8 10"<br />
"""I e is the spectrometer frequency.<br />
Spin diffusion represents the coupling of a paramagnetic<br />
center to the relaxation of distant spins in<br />
the solvent [24], most often protons, and in liquids<br />
the expression has the weak dependence on the<br />
rotational correlation time [25]:<br />
11 jSD = fi / x^ (4)<br />
where P is an adjustable constant The nuclear<br />
spin-relaxation rate is explained with just the electron-nuclear<br />
dipolar mechanism plus proton spin<br />
diffusion:<br />
1/TTOTAL=1/TEND l<br />
ln<br />
with the END term:<br />
((l/xR) 2 +(Ye.A/2) 2 )<br />
(5)<br />
(6)<br />
22 P.W. Atkins and D. Kivelson. J.Chem.Phys., 44,<br />
169, 1966.<br />
23 A. Abragam. The Principles of Nuclear Magnetism,<br />
Oxford University Press. London 1961<br />
24 A.Abragam. loc.cit. p379.<br />
25 B.I. Hunt and J.G. Powles. Proc. Phys. Soc, 88, 513-<br />
528, 1966.
34 Bulletin of Magnetic Resonance<br />
where A is the average of the A tensor. The spin<br />
diffusion term is identical in form to Tie" 1 w i*h a<br />
larger value for p.<br />
We believe that the present interpretation<br />
of the spin lattice relaxation rates, in<br />
large measure, solves the the long-standing<br />
mystery of the nature of Tie and Tin<br />
in spin labels. All the theoretical curves in<br />
Figure 3 assumed simple Brownian motion; there<br />
was no need for modified spectral density functions.<br />
It is important to note that virtually all the<br />
measured values of Tie-i and Tin 4 lie above the<br />
theoretical predictions. It is possible that residual<br />
oxygen in the glycerol-water solutions and spinlabel<br />
concentration effects may have slightly increased<br />
the observed rates.<br />
The maximum rate of T^" 1 (upper curve ) occurs<br />
when^ the motional frequency TR* 1 is equal to<br />
Ye • A / 2. Any deviation from Brownian motion<br />
would have reduced this maximum rate from the<br />
value found. The maximum value thereby becomes<br />
an extremely sensitive indicator of the type<br />
of motional process. The shape of the nuclear relaxation<br />
curve is quite different from that found by<br />
Hyde et al. [10] in their pulsed ELDOR work.<br />
The value of Tin" 1 found by them was about one<br />
MegaRadian/sec and was approximately independent<br />
of correlation time. Unfortunately these<br />
workers used a high Q EPR cavity which put an<br />
upper limit of a few MHz on the response bandwidth<br />
of their spectrometer, rendering the fast<br />
Tin 1 unobservable. Our pulsed ELDOR results<br />
show that the actual Tin' 1 is about a factor of ten<br />
shorter and clearly varies dramatically with the<br />
motion.<br />
In the study of Yin et al. [9] the lipids were<br />
moving sufficiently fast (TR about 50 picoseconds)<br />
that the Tin- 1 (~ 1.3 usec 1 ) was competitive with<br />
Tie 1 and with the 0.1-1 MHz exchange rate<br />
typical of lateral diffusion. The three rates differ<br />
by about factors of two and the data contains all<br />
three decays with the same sign. However,<br />
slower motion of the lipids (such as due to a phase<br />
transition) would lead to a rapid increase in Tin 1<br />
and make the faster exponentials difficult to<br />
distinguish. The only way to overcome this<br />
difficulty would be to perform pulsed ELDOR<br />
experiments, thereby changing the signs of one of<br />
the faster exponentials.<br />
A comment on using CW progressive saturation<br />
techniques for obtaining Tie 1 : CW-EPR assumes<br />
slow passage through the resonance line. The<br />
typical 100 kHz Zeeman modulation frequency<br />
used for detection must be much slower than<br />
Tie" 1 . The long Tie's found for motions slower<br />
than 10 nanoseconds violate this condition. This<br />
is one of the reasons why CW-EPR saturation experiments<br />
are difficult to perform: the modulation<br />
acts as a relaxation mechanism. Knowledge of the<br />
correct spin-lattice relaxation rate is important in<br />
order to avoid operating in saturation.<br />
4 Conclusions<br />
We have determined the electron and nuclear spinlattice<br />
relaxation rates in a nitroxide spin label.<br />
The rates found agree well with those predicted assuming<br />
isotropic Brownian motion. The mechanisms<br />
of spin rotation, proton spin diffusion and<br />
electron-nuclear dipolar coupling appear to explain<br />
the spin label spin-lattice relaxation mechanisms<br />
quite satisfactorily. Instrumental development is<br />
clearly at a stage where the full range of Tie" 1 and<br />
Tin" 1 rates can be studied with relative ease. We<br />
note that in this fast motional regime both spin lattice<br />
relaxation rates are approximately simply proportional<br />
to the correlation time but have opposite<br />
functional dependencies: T," 1 «= TR and T^" 1 «=<br />
TR" 1 . Pulsed SR-ELDOR is much more directly<br />
able (than CW-EPR) to connect the experimentally<br />
determinable parameters (spin lattice relaxation<br />
rates) to the general and powerful methods of<br />
interpretation in terms of motional correlation<br />
functions and the associated spectral density<br />
functions.
Vol. 14, No. 1-4 35<br />
New Developments in<br />
Pulsed Electron Paramagnetic Resonance:<br />
Direct Measurement of Rotational<br />
Correlation Times from<br />
Decay Curves<br />
Duncan A. Haas, Colin Mailer, Tetsukuni Sugano and Bruce H. Robinson<br />
Department of Chemistry<br />
University of Washington<br />
Seattle WA 98195<br />
Perdeuterated ^N TEMPOL in glycerol-water mixtures and spin-labeled hemoglobin are studied using the time<br />
domain technique of saturation recovery electron double resonance (SR-ELDOR), in the ultra-slow motional time<br />
regime. The electron spin lattice relation rate (Tig" 1 ) and the nitrogen spin-lattice relaxation rate (Tin* 1 ) are determined.<br />
Moreover, the characteristic rotational correlation times (TR) are measured directly. It is not possible,<br />
in general, to separate uniquely these rates from one another with Saturation Recovery Electron Paramagnetic<br />
Resonance (SR-EPR) experiments alone. The SR-ELDOR technique of pumping one spin manifold or orientation<br />
and observing at another changes the sign of the amplitudes of selected components contributing to the overall<br />
signal. Pooling the SR-EPR and various SR-ELDOR spectra enables all rates to be uniquely determined.<br />
1 Review of Isotropic motion<br />
Perdeuterated 15 N-Tanol [1] and spin-labeled<br />
hemoglobin (sl-Hb) [2] have been used as model<br />
systems undergoing Brownian isotropic rotational<br />
motion in the micro to millisecond time range as<br />
described by the characteristic rotational correlation<br />
time, (TR). The continuous wave technique of<br />
Saturation Transfer EPR (ST-EPR), which is very<br />
sensitive to motion on this time scale [3], relies on<br />
the competition among the different magnetization<br />
transfer mechanism of Tie, Tin, TR and the<br />
Zeeman modulation frequency, com. There are<br />
many applications of ST-EPR to slow motion,<br />
primarily using 14 N spin labels. However, the<br />
relaxation and rotational rates can only be indirectly<br />
inferred from ST-EPR [4]. We will show<br />
that these rates may be directly obtained from SR-<br />
ELDOR experiments.<br />
L.R. Dalton, B.H. Robinson, L.A. Dalton and P.<br />
Coffey, in Advances in Magnetic Resonance VIII,<br />
(J.S. Waugh, ed.) Academic Press, NY, 1976.<br />
D.D. Thomas, L.R. Dalton, J.S. Hyde, J. Chem.<br />
Phys., 65, 3006-3024, 1976.<br />
J.S. Hyde and L.R. Dalton, Chem. Phys. Lett., 16,<br />
568-572, 1972.<br />
A.H. Beth, and B.H. Robinson, in Biological Magnetic<br />
Resonance VIII Spin Labeling: Theory and<br />
Application (L J. Berliner and J. Reuben, eds.),<br />
Plenum Press, NY, 1989.<br />
Hyde and co-workers invented the technique of<br />
Continuous Wave Electron-Electron Double Resonance<br />
(CW-ELDOR) [5]. They demonstrated that,<br />
by pumping one point in a spectrum and observing<br />
at another point, energy transfer could be detected.<br />
However, CW-ELDOR is an extremely complex<br />
technique and the results can be influenced by instrumental<br />
artefacts [6]. The technique has been<br />
applied by Stetter et al. [7] to nitroxides moving<br />
with sub-nanosecond times and by Smigel et al.<br />
[8] to nitroxides with microsecond correlation<br />
times. From CW-ELDOR, one cannot estimate<br />
either Tie or Tin independently, only their ratio.<br />
The Stetter [7] work used different isotopes of nitrogen<br />
to separate exchange due to lateral diffusion<br />
from nitrogen relaxation. (There can be no nuclear<br />
spin lattice relaxation between different isotopes so<br />
any interaction between them must come from<br />
collisions.)<br />
J.S. Hyde, J.C.W. Chien and J.H. Freed, J. Chem.<br />
Phys.,4S, 4211-4226, 1968.<br />
L.A. Dalton and L.R. Dalton, in Multiple Electron<br />
Resonance Spectroscopy eds. M.M. Dorio and J.H.<br />
Freed. Plenum Press 1979 Chapter 5.<br />
E. Stetter, H.-M. Vieth, and K.H. Hauser, J. Mag.<br />
Res., 23, 493-504, 1976.<br />
M.D. Smigel, L.R. Dalton, J.S. Hyde, and L.A. Dalton,<br />
Proc. Nat.Acad. Sci. USA, 71, 1925-1929, 1974.
36 Bulletin o[ Magnetic Resonance<br />
Time Domain EPR: In a time domain EPR experiment,<br />
the spins are subjected to a short pulse<br />
of microwave power; the time dependence of the<br />
transient relaxation after the pulsing is monitored.<br />
The rates of relaxation are functions of the motional<br />
process. Two types of time domain experiments<br />
exist: one monitors spin echoes (a coherent<br />
detection of the x- and y-components of the magnetization)<br />
and the other monitors Saturation<br />
Recovery (as a polarization detection of the z-magnetization).<br />
Freed [9] has applied spin-echo methodology to<br />
study motion in liquids. A (90°-T-180°-i-Observe)<br />
sequence measures echo height as a function of the<br />
delay time t to give the phase memory decay time<br />
TM- Other workers have studied spin-labeled<br />
membranes and vesicles - showing changes in TM<br />
with temperature, but do not relate results to any<br />
theory [10]. Freed [9] has also developed the<br />
technique of Fourier Transform EPR using<br />
multiple pulse techniques, originally developed for<br />
NMR. The variation of TM across the spectrum<br />
can be directly obtained and related to motional<br />
models. Cross-relaxation from one manifold to<br />
another can also be measured. Rates are measured<br />
from the volumes of the main and cross-peaks in<br />
the 2-D display, and not from actual decay curves.<br />
Huisjen and Hyde [11] pioneered the use of the<br />
SR-EPR technique in liquids and applied it to a<br />
number of systems [12], [13]. The Hyde group<br />
[14] has used SR-EPR, in conjunction with CW-<br />
ELDOR, to measure lateral diffusion of 14 N labelled<br />
lipids in bilayers. These important papers on<br />
translational motion set the stage foT our work in<br />
rotational motion.<br />
Measurement of rotational diffusion directly in the<br />
time domain with SR-EPR alone has proved im-<br />
9 J. Gorcester, G.L. Millhauser, and J.H. Freed, in<br />
Modern Pulsed and Continuous-Wave Electron Spin<br />
Resonance, L. Kevan and M.K. Bowman(eds.) John<br />
Wiley 1990 Chapter 3.<br />
10 K. Madden, L. Kevan, P.D. Morse and R.N. Schwartz,<br />
/. Am. Chem. Soc, 104, 10, 1982.<br />
11 M. Huisjen and J.S. Hyde, Rev. Set Inst., 45, 669-<br />
675, 1974.<br />
12 P.W. Percival and J.S. Hyde, /. Mag. Res., 23, 249-<br />
257, 1976.<br />
13 A. Kusumi,W.K. Subczynski and J.S. Hyde, Proc.<br />
Nat. Acad. Sci. USA, 79, 1854-1858, 1982.<br />
14 J.-J. Yin, M. Pasenkiewicz-Gierula, and J.S. Hyde,<br />
Proc. Nat. Acad. Sci. USA, 84, 964-968, 1987.<br />
possible. Typical of an SR-EPR experiment is that<br />
of Fajer et al. on sl-Hb [15] which produced biexponential<br />
decays. The slower rate was<br />
(correctly) identified with Tie" 1 and the faster decay<br />
rate was associated with the rotational motion<br />
of the hemoglobin. The faster decay rate remained<br />
approximately constant despite a two order of<br />
magnitude change of correlation time. These authors<br />
then used a qualitative theory of spin diffusion<br />
to extract an effective correlation time. The<br />
data agreed only approximately with the known 1R<br />
values. The problem with this type of experiment<br />
is the strong influence of the nitrogen nuclear spinlattice<br />
relaxation, on the decays. As explained below,<br />
the observed decays were a mixture of all<br />
three relaxation processes. This makes the protocol<br />
for data collection and for analysis of the various<br />
decays much more complex than originally<br />
expected. Attempts in our laboratory using<br />
Saturation Recovery alone were similarly unsuccessful<br />
(unpublished experiments with Dr. P.<br />
Fajer). We believe we have now solved these<br />
fundamental problems using pulsed SR-ELDOR<br />
and SR-EPR together.<br />
We will demonstrate that Pulsed EPR techniques,<br />
such as SR-EPR and SR-ELDOR, can be used to<br />
obtain both the nuclear and electronic spin-lattice<br />
relaxation rates directly and monitor changes in exchange<br />
rates. Moreover, the correlation time connecting<br />
two spectral positions can be directly measured<br />
as well. Relaxation time measurements, especially<br />
nuclear spin-lattice relaxation (Tin). can<br />
also characterize the motion in the ultra-slow motional<br />
time range.<br />
2 Theory<br />
Figure 1 shows an absorption CW-EPR signal<br />
when the correlation time is longer than 0.1 microseconds.<br />
The figure illustrates the dependence of<br />
the signal on molecular orientation. The reorientation,<br />
characterized by a rotational correlation time,<br />
TR, moves the magnetization around within a given<br />
manifold. Nuclear spin flips, occurring at rate<br />
T^" 1 , move the magnetization from one manifold<br />
to the other without causing molecular reorientation.<br />
Regardless of orientation or manifold, the<br />
magnetization relaxes to the lattice at rate Tie'<br />
15 P. Fajer, D.D. Thomas, J.B. Feix, and J.S. Hyde,<br />
Biophys. J., 50, 1195-1202, 1986.
Vol. 14, No. 1-4 37<br />
Theory [4], [16] predicts for the SR-ELDOR experiment<br />
that the TR of a molecule should appear as<br />
an exponential decay with a rate (TR" 1 + T^" 1 ) and<br />
that Tin should appear as a decay of rate (T\a' 1 +<br />
T^" 1 ). These two processes have very similar<br />
rates, which is why motional rates have been so<br />
difficult to determine. However, suitable positions<br />
of pump and observer can be chosen so that the<br />
signs and the amplitudes of both Tin and the motional<br />
decay curves can be individually changed,<br />
enabling analysis to disentangle the rates. The SR-<br />
ELDOR experiment measures the decay of the<br />
transient component of the z magnetization,<br />
(Mz(t;v,8)), where v is the nuclear manifold<br />
(which is ±1/2 for 15 N), and 6 is the orientation,<br />
as shown in Figure 1. The evolution of the magnetization<br />
can be understood, qualitatively at least,<br />
by a simple population analysis treatment<br />
Lattice!<br />
^Figure 1: Linear Absorption EPR spectrum of sl-Hb<br />
illustrating the positions where SR-ELDOR experiments<br />
are performed and the dependence of the resonance<br />
'positions of the magnetization on spin-label orienta-<br />
°" The figure further illustrates how TR, and Tin<br />
! spectral diffusion, whereas Tie induces true spin-<br />
"J relation.<br />
^•Sugano, C. Mailer, and B.H. Robinson, /. Chem.<br />
zfkys. 87, 2478-2488, 1987.<br />
Therefore the master equation is:<br />
{(M2(t;v,e))-(Mz(t;-v,0))}<br />
-DV2 (Mz(t;v,e))<br />
(1)<br />
where D is the Einstein Rotational Diffusion<br />
Coefficient (D = 1/6 tR) and V 2 , is the angular<br />
Laplacian operator. The system is pumped with a<br />
weak selective pulse so that the spin system is<br />
excited in manifold vp and at orientation 8p. The<br />
the observer frequency is set to manifold v0 and at<br />
orientation 80. The recovery signal then has the<br />
approximate form:<br />
(Mz(t;vo,eo vp,ep)) =<br />
* (l + P2(cos(9o)) • P2(cos(8p))e- t/x R)<br />
where is the magnetization at<br />
the observer position subject to the condition that<br />
all of the magnetization was at the pump position at<br />
time zero; and (Mz(O;vp,0p)) is the initial magnetization<br />
at time zero (when the pumping is completed).<br />
P2 (cos(8)) = (3 cos 2 (8) - l)/2 is the I =<br />
2 component of the Legendre polynomials P;(x),<br />
and fvy = +1 if v = v' and fvv- = -1 if v = -v'<br />
(which is the case of going from one manifold to<br />
the other). This result assumes a short duration<br />
pump time, a low observer amplitude and that the<br />
higher rotational functions are not significant [17].<br />
Equation 2 predicts that one will observe 4 exponentially<br />
decaying components, wherein the rate of<br />
each component is a linear combination of T^" 1 ,<br />
Ti,," 1 , and TR" 1 . The amplitude of the components<br />
containing Tin" 1 , may be switched in sign by setting<br />
the pump and observer positions to different<br />
spin manifolds. The amplitude of the components<br />
containing TR" 1 may be switched in sign by setting<br />
the pump and the observer to different ends of the<br />
same manifold.<br />
17 T. Sugano, "A Study of Very Slow Rotational<br />
Diffusion by SR-EPR", Ph.D. Thesis, University of<br />
Washington, 1987.
38 Bulletin of Magnetic Resonance<br />
3 Experiment<br />
The details of the spectrometer and its performance<br />
are discussed more fully in another paper in these<br />
Proceedings [18]. SR-EPR and SR-ELDOR experiments<br />
are both polarization experiments which<br />
measure the recovery of the system to equilibrium.<br />
A selective pulse is applied to the spins which alters<br />
the polarization and burns a partial hole in the<br />
line. Under conditions of SR-EPR the frequency<br />
of observation is the same as the pump and therefore<br />
the signal is that of a pure recovery as the polarization<br />
spreads throughout the spin system and<br />
out to the lattice. Under conditions of SR-ELDOR,<br />
the frequency of the observer is different from that<br />
of the pump and the arrival of magnetization, as<br />
well as relaxation to equilibrium, are both detected.<br />
It is neither necessary, nor desirable, for the pump<br />
to be coherent with the observer. The experiments<br />
obtain decay curves taken with various pump times<br />
and pump-observer frequency differences; the rotational<br />
correlation time is determined by the solvent<br />
viscosity and temperature. The pump time<br />
controls the relative amplitudes of the various components<br />
which contribute to the overall relaxation<br />
of the magnetization; and the pump-observer frequency<br />
difference controls the signs of the amplitudes<br />
of the exponentially decaying components in<br />
ways predicted by equation 2.<br />
The experiment proceeds as follows: for one set of<br />
experimental conditions, a number of decay curves<br />
over different time scales are obtained. These decays<br />
are linked by the fact that the exponentials<br />
which comprise them are all identical and only the<br />
time range of data collection is altered. For a different<br />
set of conditions e.g. in pulse length or in<br />
observer field position, the relative amplitudes<br />
and/or signs of the individual decays are altered<br />
and another set of linked spectra is collected Both<br />
sets of linked data are then pooled with the constraint<br />
that the rates of the exponentials which<br />
make up the spectra are common to all and the<br />
amplitudes of each component within a linked set<br />
are the same. This technique [19], as an<br />
18 C. Mailer, B.H. Robinson and D.A. Haas, "New<br />
Developments in Pulsed EPR: Relaxation Mechanisms<br />
of Nitroxide Spin Labels", [these proceedings] 1992.<br />
19 J.M. Beecham, E. Gratton, M. Ameloot, J.R.<br />
Knutson, and L. Brand, in Fluorescence<br />
Spectroscopy: Principles J.R. Lakowicz (ed.) Plenum<br />
Press NY Volume 2 Chapter 5,1991.<br />
application of Global Analysis, is well recognized<br />
in optical spectroscopy as being extremely effective<br />
in determining relaxation rates.<br />
Once one has obtained a unique set of decay rates<br />
that minimizes the global x 2 one must identify the<br />
processes which gives rise to each individual decay<br />
rate i.e.whether it is due to Tie" 1 , Tin" 1 o f rotation.<br />
Suitable choice of experimental conditions<br />
for acquisition of the decays enable the rates to be<br />
distinguished. The strategies we have found useful<br />
are:<br />
(i) Pumping and observing at the magic angle<br />
will reduce the amplitude of the motional rate,<br />
leaving just the Tie and Tin" 1 .<br />
(ii) Pumping in one spin manifold and observing<br />
in the other will always invert the amplitude of<br />
a Tin containing term.<br />
(iii) Pumping at one extreme turning point of a<br />
spin manifold and observing the other will invert<br />
the amplitudes of the motionally dependent components.<br />
(iv) Suppression of free induction decay (FID)<br />
is always important. In the nanosecond range of<br />
motion the FED has approximately the same time as<br />
Tie and Tin.<br />
(v) A pump time that is long compared to the<br />
relaxation time of a particular component will tend<br />
to suppress the amplitude of that component.<br />
(vi) Pooling the Saturation Recovery EPR and<br />
pulsed ELDOR decay curves for analysis is essential<br />
for all rates to be found - no single experiment<br />
alone is sufficient.<br />
4 Results<br />
We have obtained correlation times from perdeuterated<br />
15 N TEMPOL in glycerol-water mixtures<br />
over the TR range from 0.1 to 100 microsec- .<br />
onds: Bi-expOnential recovery curves were ob-1<br />
tained when using the SR-ELDOR protocol de- ^<br />
scribed above, when the "magic angle" (8 = 54°);l<br />
was chosen for the pumping and observing posi-Jj<br />
tions (as shown in Figure 1). In Figure 2 are<br />
plotted the Tie' 1 and Tin" 1 rates versus correlation<br />
time in the slow motion range. The very weak;<br />
dependence of T^' 1 on correlation time suggest|<br />
that the more traditional relaxation mechanisms"
Vol. 14, No. 1-4<br />
such as spin rotation and Electron-Nuclear-Dipolar<br />
coupling (END) are not the dominant ones. The<br />
power law dependence of Tie' 1 is on the order of<br />
1/TR 1/8 . The NMR literature suggests that a model<br />
of spin diffusion in liquids will have the same<br />
power law dependence as that experimentally<br />
observed [20]. This same power law dependence<br />
was also observed in similar measurements using<br />
SR-EPR [15]. The solid line superimposed on the<br />
Tie" 1 data is defined as:<br />
(coexR)<br />
(3)<br />
where fie is an adjustable parameter, and the second<br />
term (r.h.s.) is the END term and the third term<br />
u<br />
q><br />
40<br />
SR-EPR: «B<br />
Bulletin of Magnetic Resonance<br />
Figure 3: Three different SR-ELDOR spectra, taken at different pump and observation positions (see Figure 1). The top<br />
curve is the SR-EPR spectrum at position B (in Figure 1), the middle curve is the SR-ELDOR spectrum pumping at position<br />
A and observing at position B, and the bottom spectrum is the SR-ELDOR spectrum pumping at position B and observing<br />
at position C. Superimposed on each spectra is the least-squares best fit consisting of an adjustable baseline and three<br />
components each of which is a single exponential decay (see equation 2), the rates are given in Figure 2.<br />
Figure 3 shows an example of the SR-EPR and<br />
SR-ELDOR data for sl-Hb tumbling with xR in the<br />
1 to 5 microsecond time range. The top curve,<br />
(SR-EPR at B) is the SR-EPR data acquired at<br />
point B (defined in Figure 1). This curve is a<br />
simple recovery and all components have the same<br />
sign. The bottom curve (SR-ELDOR from B to C)<br />
is the data when the pump is set on B and the observer<br />
frequency is set to C. This represents a<br />
jump from one manifold to the other and the amplitude<br />
of the components, containing a Tin" 1 in the<br />
rate, change sign. The middle trace of Figure 3<br />
(SR-ELDOR from A to B) is the data for the pump<br />
on A and the observer on B. This corresponds to<br />
the pump-observe case within a manifold.<br />
According to equation 2, the amplitude of the<br />
components containing TR in the rate will change<br />
sign. Clearly there is a peculiar dip in the shape of<br />
this SR-ELDOR curve. The dip is characteristic of<br />
the component that depends directly on the correlation<br />
time. Superimposed on each of the data sets<br />
is a least-squares best fit simulation composed of<br />
three exponentials and a baseline. The three rates<br />
of the exponentials are the same in all three data<br />
sets. The rates arc interpreted, according to equation<br />
2, as Tie* 1 , Tin" 1 and TR' 1 as 0.0937, 1.983<br />
and 0.329 MRad/sec respectively with around a|<br />
10% error. Notice that the motional rate is in be-.|<br />
tween the spin-lattice relaxation rate and the nu-|<br />
clear relaxation rate. These data are also shown inf<br />
Figure 2. The nominal correlation time, d cte f1<br />
mined from solvent viscosity and hydrodynami
Vol. 14, No. 1-4 41<br />
radius of sl-Hb, is 1.5 |j.sec. From the L"/L ratio<br />
calibration curve of the ST-EPR [4] one may estimate<br />
the correlation time to be 4.0 p.sec and the<br />
rotational correlation time measured by SR-<br />
ELDOR is 3.3 jisec, which is in excellent agreement<br />
with the ST-EPR calibration data. This<br />
work, demonstrates that the different components<br />
of the experimental curve can be uniquely identified<br />
and their rates quantitatively measure, and<br />
hence Tie, Tin and TR may be directly measured.<br />
This is the first time that xR has been directly<br />
measured in EPR and represents a<br />
fundamental advance in time domain<br />
methodology.<br />
5 Conclusions<br />
We have measured the values of the electron and<br />
nuclear spin lattice relaxation times in the ultraslow<br />
motion region using SR-ELDOR and SR-<br />
EPR and find that the measured values of Tie' 1<br />
depend on 1/TR 1 / 8 , or a power law dependence,<br />
which is consistent with a model of spin diffusion<br />
in liquids. The measured values of Tin are well<br />
described by the electron-nuclear dipolar relaxation,<br />
(which has no adjustable parameters in it)<br />
and only the rates at motional times longer than<br />
100 fisec suggest the need for a spin diffusion<br />
mechanism for Tin as well as Tie. Notice that for<br />
the case of spin labels the nuclear relaxation is<br />
much faster than the electron relaxation (an unusual<br />
situation). This is primarily a result of the<br />
ability of the electron, at this particular rotational<br />
rate, to efficiently relax the nucleus and the fact<br />
that the electron has very few other spins to relax<br />
it. By using SR-ELDOR and by pooling the data<br />
at different orientations the rotational correlation<br />
time can be measured directly, and is seen experimentally<br />
as a single exponential relaxation. This<br />
is, we believe, the first time rotational motion has<br />
been quantitatively measured as a single exponential<br />
decay in this type of experiment. These exper-<br />
- imental results suggest that it may be possible to<br />
. directly measure rotational reorientation by SR-<br />
, ELDOR even for systems characterized by aniso-<br />
> tropic motion.<br />
M-<br />
|r. These type of time domain experiments are a necunderpinning<br />
and improvement on tradi-<br />
CW techniques. Quantitative simulation of<br />
f-EPR spectra requires knowledge of the relax-<br />
ation times of the spin system. When performing<br />
progressive saturation studies using CW-EPR, the<br />
experiments can only be used to detect relative<br />
changes in relaxation times when motion is longer<br />
than a nanosecond [22]. (Absolute values are difficult<br />
to obtain accurately because the effective relaxation<br />
time for the CW experiment is a complex<br />
mixture of competing relaxation processes.) Direct<br />
measurement of Tie" 1 an d Tin w ^h pulse techniques<br />
is clearly superior.<br />
Prior to the time domain experiments it was found<br />
in this laboratory that the Tin* 1 value in calculations<br />
of ST-EPR spectra in the microsecond motional<br />
range had to be set artificially high for good<br />
agreement between simulation and experiment<br />
[23]. The simulation assumes that the electronnuclear<br />
dipolar (END) interaction is the sole mechanism<br />
for nuclear spin-lattice relaxation. The data<br />
in Figure 2 clearly show that other mechanisms<br />
add to the END rate (e.g. from proton spin diffusion)<br />
justifying the ad hoc addition of an extra rate<br />
into the calculations.<br />
22 B.H. Robinson, C. Mailer and D.A. Haas, Biophys. J.<br />
61, A167, Abstract # 960, 1992.<br />
23 D. Haas, R. St Denis, C. Mailer, B.H. Robinson, 13th<br />
Intl. EPR Conf., Denver, CO, 1990.
42 Bulletin of Magnetic Resonance<br />
Non-Linear effects in Standard 2D NOE<br />
experiments in Coupled spin systems<br />
R.Christy Rani Grace* and Anil Kumar*t<br />
''Department of Physics and ^Sophisticated Instruments Facility<br />
Indian Institute of Science, Bangalore - 560 012, INDIA.<br />
INTRODUCTION<br />
The nuclear Overhauser effect (NOE)<br />
which monitors the transfer of magneti-<br />
zation from one spin to another, is criti-<br />
cally dependent on the internuclear dis-<br />
tance and has therefore become a pow-<br />
erful tool for elucidation of the struc-<br />
tures of Biomolecules. Experimental<br />
methods for monitoring these effects of-<br />
ten use radio frequency pulses which si-<br />
multaneously excite and/or detect sev-<br />
eral spins at a time. If the spins are<br />
not coherently coupled (no J coupling),<br />
there are no non-linear effects of the<br />
pulses, except for a scaling factor. The<br />
non-linear effects in the presence of J-<br />
coupling for one-dimensional NOE ex-<br />
periments are well known(l,2). In this<br />
paper the non-linear effects in the 2D<br />
NOE (NOESY) experiment are anal-<br />
ysed in detail.<br />
The standard NOESY experiment<br />
uses the sequence 90° — ii — 90° — rm —<br />
90° —12, in which relaxation takes place<br />
during the mixing interval rm . The<br />
rate equations governing relaxation are<br />
exactly identical to the transient NOE<br />
experiment(3-5). It has been known<br />
that for uncoupled spins each cross-<br />
section in the NOESY experiment is<br />
equivalent to a ID transient NOE exper-<br />
iment in which the peak corresponding<br />
to the diagonal peak is selectively in-<br />
verted^). When there are J-couplings<br />
present in the spin system, selective in-<br />
version has to be carefully defined. Re-<br />
cently, it has been shown that for small<br />
values of the second pulse (90° — t\ —<br />
a — Tm — 90° — t2 at frequency u>i = ua is<br />
equivalent to a ID difference transient<br />
NOE experiment in which the transition<br />
at frequency ua is selectively inverted.<br />
This is true irrespective of the strength<br />
of the coupling(6,7). It has also been<br />
shown for weakly coupled spins that in
Vol. 14, No. 1-4 43<br />
the standard NOESY experiment, any<br />
cross-section parallel to CJ2 at u\ = u>a, is<br />
equivalent to a ID transient experiment<br />
in which, the whole multiplet of which<br />
ua is a part is non-selectively inverted.<br />
When the spins are strongly coupled the<br />
90° pulse distributes the perturbation<br />
over all the transitions of the strongly<br />
coupled network and the 2D NOE ex-<br />
periment is not equivalent to any stan-<br />
dard transient ID experiment. In ad-<br />
dition, the third pulse in the NOESY<br />
experiment (the measuring pulse) mea-<br />
sures the state of the spin system in a<br />
non-linear manner for finite angles. As<br />
a result it is shown here that in strongly<br />
coupled spin systems one can obtain<br />
'cross-peaks' in the standard NOESY<br />
experiment without relaxation. The ori-<br />
gin of these cross-peaks in terms of the<br />
non-linearity of the second and/or the<br />
third pulse is also discussed with the<br />
help of an ABX spin system.<br />
Cross-correlations between pairwise<br />
dipolar relaxation and between dipolar<br />
and other mechanism of relaxation such<br />
as chemical shift anisotropy(CSA) are<br />
known to yield a multiplet effect in J-<br />
coupled spectra(7-ll). A measurement<br />
of this effect in one and two dimensional<br />
spectra is carried out using small angle<br />
pulses. Recently Osckinat et al. have<br />
used small angles for the second and<br />
the third pulses in the NOESY experi-<br />
ment and have shown that in the initial<br />
rate approximation the effect of cross-<br />
correlations is present in all the mul-<br />
tiplets of an AMX spin system(7). In<br />
their experiment the direct pumping ef-<br />
fects and cross-correlation effects both<br />
give rise to multiplet effects. We pro-<br />
pose here simple modifications which al-<br />
lows the direct pumping effects to be<br />
absent, with the cross-correlations ex-<br />
clusively exhibiting multiplet effects in<br />
weakly coupled spins.<br />
A. STRONG COUPLING<br />
INDUCED CROSS-PEAKS IN<br />
NOESY<br />
The signal in a NOESY experiment<br />
utilizing 90° — ti — a — rm — j3 — t2 se-<br />
quence in which only longitudinal mag-<br />
netization is retained during rm period<br />
can be expressed as,<br />
S(h,t2) =<br />
Tr{{Fx)exp(-iHt2)exp(-if5Fx)<br />
[exp(—iaFx)exp(—iHti)exp(—i'^Fy)<br />
exp(WTm)exp(i(3Fx)exp(iHt2)}<br />
(1)<br />
where the prime indicates retention of<br />
only the diagonal elements of the den-<br />
sity matrix after the a pulse, W is the<br />
matrix governing relaxation during rTO<br />
period and cr0 is the initial density ma-<br />
trix. If cr0 is an equilibrium density ma-
44 Bulletin of Magnetic Resonance<br />
trix, then only single quantum coher-<br />
ences are created during ii period and<br />
since during period t2 only single quan-<br />
tum coherences are detected, the above<br />
equation can be written as(7)<br />
exp(WNxNrm)<br />
(2)<br />
X) represents a matrix which<br />
transforms the N populations into M<br />
single quantum coherences by a pulse<br />
of angle 7X . The N populations are<br />
arranged in descending order of energy<br />
while the M coherences represented by<br />
vectors
Vol. 14, No. 1-4 45<br />
PN*M{I) =<br />
-s 2<br />
-C 2 +<br />
S 2 (l - v 2 )<br />
v 2 S 2<br />
c 2<br />
-C 2<br />
-u' 2 S 2<br />
C 2 +<br />
S 2 (u 2 - 1)<br />
s 2<br />
—S 2<br />
u 2 S 2<br />
—C 2 —<br />
S 2 (u 2 - 1)<br />
c 2<br />
-C 2<br />
c 2 -<br />
S 2 (l - v 2 )<br />
-v' z S 2<br />
s 2<br />
u 0 0 0<br />
0 v 0 0<br />
0 0 v 0<br />
0 0 0 u<br />
(6) '<br />
where S = Sin(7 / 2) ; C = Cosfr / 2); u = Cos 9 + Sin 9 ;v = Cos 9 - Sin 9 and<br />
tan(29) — JAB/($A — &B) defines the strength ofthe coupling.<br />
Frequencies<br />
U)2<br />
(1) Diagonal peaks<br />
1-3<br />
3-4<br />
1-3<br />
3-4<br />
1-2 1-2<br />
2-4 2-4<br />
(2) Auto-peaks<br />
1-3 2-4<br />
3-4 1-2<br />
2-4 1-3<br />
1-2 3-4<br />
(3) Cross-peaks<br />
1-3 3-4<br />
3-4 1-3<br />
2-4<br />
1-2<br />
1-3<br />
2-4<br />
1-2<br />
3-4<br />
1-2<br />
2-4<br />
1-2<br />
3-4<br />
1-3<br />
2-4<br />
V<br />
Table. 1.<br />
u\4GIC}<br />
+ 4S 2 aS 2 {v 2 + (1 - ^ 2 ) 2 } — (1 — C2aC2/3)(l — V 2 )]<br />
u 2 v '[-4^(1-«V) +2(1<br />
-c2ac<br />
U 2 V<br />
Intensities *(—5*2<br />
%CIC} + 4S 2 aS 2 p{u 2 + {v 2<br />
2 M^(i-^) + 2(i<br />
v 4 [-2C 2 aC 2 -2S 2 aS 2 {uU<br />
- m<br />
— {u 2 \o — zu<br />
2 — v)(C2a<br />
-<br />
- (u 2 - 1)<br />
2 } + (1 - C2aC2P)v 2 }<br />
- (1 - v'<br />
A)<br />
— (1 — ^2cc^- / 2/3)(l — u )\<br />
iff) + (1 - V 2 )(C2c - CV)]<br />
w) - (1 - v 2 )(C2a - C2P)}<br />
\2\ i^ /i /"i /"i \ 2*1<br />
/ J ' v ^- / 2a^- / 2/5/^ J<br />
u r — '^\ -*- — ^- J '2f"y^-^2/?/<br />
Cos(i) ; ^i = Sin(i) ; d = Cos(^) ; Si = Sin(^) where i = a, f3 and<br />
u = Cos 9 + Sin 9 ; v = Cos 9 - Sin 9.
46<br />
The origin of these cross-peaks lies in<br />
the creation of a initial state in which<br />
the initial perturbation is distributed<br />
over all the transitions of a strongly cou-<br />
pled spin system as well as due to the<br />
non-linear measurement of the strongly<br />
coupled spin system by the third 90°<br />
pulse. The initial state in this experi-<br />
ment can also be described using mag-<br />
netization modes(12,13). For an AB<br />
system the initial state in terms of the<br />
magnetization modes at various cross-<br />
sections parallel to UJ^ is given in Ta-<br />
ble.2. From these it is seen that the<br />
single spin modes of both the spins are<br />
created in each cross-section. This is the<br />
origin of the cross-peaks in strongly cou-<br />
pled spins. In the limit of weak coupling<br />
(u = v = 1) each cross-section contains<br />
only one single spin mode belonging to<br />
the inverted spin and the cross-peaks<br />
are absent.<br />
ABX spin system<br />
If there are two groups of spins which<br />
are strongly coupled among themselves<br />
but weakly coupled to others then it<br />
is not a priory clear that there will<br />
be cross-peaks between the two groups.<br />
This is investigated here with the help<br />
of an ABX spin system. Fig.2 shows<br />
the standard NOESY (a = j8 = 90° )<br />
spectrum calculated using eqn.[5] for an<br />
Bulletin of Magnetic Resonance<br />
ABX spin system with zero mixing time.<br />
Actual intensities of the peaks in the<br />
2D spectrum is obtained by multiply-<br />
ing the expressions given in Table.3 with<br />
the corresponding ID intensities in both<br />
u>i and ui2 dimensions of that particu-<br />
lar peak. From this it is seen that ev-<br />
ery peak has a cross-peak to every other<br />
peak. The cross-peaks in this spectrum<br />
including those between A and B spins<br />
and between AB spins and X spin arise<br />
due to the strong coupling among the A<br />
and B spins, and disappear under weak<br />
coupling approximation. The appear-<br />
ance of these cross-peaks needs further<br />
investigation in terms of whether they<br />
are due to the non-linearity of the sec-<br />
ond or the third pulse. To investigate<br />
this, calculations have been carried out<br />
for the cases when the excitation pulse is<br />
small(in the linear regime) or the detec-<br />
tion pulse is small(in the linear regime).<br />
The following results were noted from<br />
these experiments:<br />
(i) 90° - a - 90° Experiment<br />
The ABX spin system has eight AB-<br />
transitions and six X-transitions two of<br />
which are between states which are un-<br />
perturbed by strong coupling (the so ><br />
called pure states 1,2,7 and 8)(14). In<br />
this system in the 90° — a — 90° experi-<br />
ment there are no cross-peaks from the
Vol. 14, No. 1-4 47<br />
K.-<br />
At ui<br />
< AA2 >Tm=0<br />
< ABZ >Tm=0<br />
< AAZBZ >Tm=0<br />
u 2<br />
u 2<br />
V 2<br />
V2<br />
-u 2 (l - v 2 )<br />
-u 2 (l+v 2 )<br />
0<br />
Table. 2.<br />
~V 2 (1 + U 2 )<br />
v 2 (l - v 2 )<br />
0<br />
2 u 2<br />
• • • 1i<br />
• • • 4<br />
• • • *<br />
to,<br />
I<br />
•<br />
v 2<br />
-u\l + v 2 )<br />
-u 2 (l - v 2 )<br />
0<br />
V 2 {l-V 2 )<br />
-v 2 (l + u 2 )<br />
0<br />
= -(1+v 4 )<br />
• - (1-v 4 )<br />
- (1-u 4 )<br />
- (1-U 2 V 2 )<br />
Figure 1. Schematic spectrum of an AB spin system calculated for the 90°—90°—90°<br />
2D NOESY experiment with zero mixing time. The symbols represent -PMxiv(90°) x<br />
-fWxM(90°), the |FX| 2 are given along the ID spectra and the final intensities are<br />
obtained using eq [5].
48<br />
B<br />
A<br />
C_<br />
2<br />
V_<br />
U<br />
V<br />
u<br />
V.<br />
AiB3<br />
A1B1<br />
A,<br />
An<br />
A12<br />
37 13<br />
A2X2<br />
A2B3<br />
A2B2<br />
A3X4<br />
^3X3<br />
A3X2<br />
A3X1<br />
A3B3<br />
A3B1<br />
-43<br />
u 2<br />
26 47<br />
A4X2<br />
AtB3<br />
B1X4<br />
fijA'3<br />
BiX2<br />
Bu<br />
B13<br />
B12<br />
B,<br />
u 2<br />
B2X4<br />
4"<br />
- Bt for i = 1 io 4<br />
B2X3<br />
B2X2<br />
B2X1<br />
B24<br />
B23<br />
B2<br />
B3X6<br />
B3X5<br />
B3X4<br />
B3X3<br />
B3X2<br />
BsXj<br />
B34<br />
B3<br />
68<br />
B4X4<br />
B4X3<br />
B4X2<br />
B4X1<br />
B4<br />
36,<br />
A'l4<br />
X13<br />
X12<br />
X!<br />
12<br />
Bulletin of Magnetic Resonance<br />
A'26<br />
X25<br />
x24<br />
x2<br />
46 35<br />
X35<br />
X34<br />
X3<br />
X4<br />
x4<br />
78<br />
45 1<br />
Xe<br />
.1 SI 1 Cl Cl I Si<br />
o>2<br />
* V — V — V<br />
A2 — A5 — A25<br />
D{j A12 — A15<br />
AjBi for i ^ j and j > i A"23 = X35<br />
— B{Xj for i = 1 to 4 (md A24 = A45<br />
A{X2 for j = 1 io 6 A26 = A56<br />
5.A-.2 " AXX, = A4A6<br />
A3JB4 A^X'3 = A3X4<br />
A34 A2A4 = A4A3<br />
A3X1 = A2A6<br />
Figure 2.
Vol. 14, No. 1-4<br />
a3<br />
a4<br />
Peaks<br />
Al<br />
A2<br />
A3<br />
A4<br />
Au<br />
Tl^ —<br />
A13<br />
A14<br />
A23<br />
A24<br />
Xi<br />
x2<br />
x3<br />
x4<br />
X6<br />
XX2<br />
Xi3<br />
X\4<br />
Xm<br />
x23<br />
X24<br />
X26<br />
^34<br />
-^36<br />
X4Q<br />
Strong<br />
Coupling<br />
-(1 + 5/6^)<br />
-(1 + 56?,)<br />
-(1 + *%)<br />
-(1 + s/b 2 )<br />
-(1 +5)<br />
-(1 +564/63)<br />
-(1 + 5/6x62)<br />
-(1 + 56x62)<br />
-(1 + 363/64)<br />
-(a 2 4 + ka 2 3)<br />
-(l + k)<br />
-(al + kaf)<br />
—(al + ka\)<br />
-(a 2 + kal)<br />
sa3 — a4<br />
ka2a4 — axa.3<br />
kaxa4 - a2a3<br />
-a3a4(l + k)<br />
—(ax + 502)<br />
—(a2 + 5ax)<br />
504 — a3<br />
-axa2(l + k)<br />
Table 3. Intensities of the peaks<br />
Weak<br />
Coupling<br />
-2<br />
-2<br />
-2<br />
-2<br />
-2<br />
-2<br />
-2<br />
-2<br />
_2<br />
0<br />
-2<br />
-2<br />
-2<br />
0<br />
0<br />
0<br />
0<br />
0<br />
-2<br />
_2<br />
0<br />
0<br />
0<br />
0<br />
61 = V+/U+<br />
62 = v_/u_<br />
h =<br />
b4<br />
=<br />
— ,,2<br />
U +~ V +<br />
u_v+v_/u+;<br />
u+u_v+/v_;<br />
m2<br />
k<br />
= u<br />
s<br />
-<br />
=<br />
f,,2 ,,2<br />
Peaks<br />
AlBl<br />
A2B2<br />
A3B3<br />
A4B4<br />
AiB2<br />
AiB3<br />
AXB4<br />
A2B3<br />
A2B4<br />
AxXx<br />
AiX2<br />
AiX3<br />
A1X4<br />
AiX6<br />
A2XX<br />
A2X2<br />
A2X3<br />
A2X4<br />
A2X&<br />
A3X2<br />
A3X3<br />
A3X6<br />
A4X1<br />
A4X2<br />
A4X4<br />
Strong<br />
Coupling<br />
s/b 2 -l<br />
562 -1<br />
sb 2 -l<br />
s/bl - 1<br />
5-1<br />
564/63 - 1<br />
563/64 - 1<br />
56162 - 1<br />
5/6i62-l<br />
2 2<br />
a4miif+ — a3m2U_<br />
[n3mx — ra2ui]<br />
a2m2u 2 _ - axmiv\<br />
ax[n3mx — rn2u 2 _]<br />
-a4[n3mx — m2u 2 _]<br />
—a3[nxmx — m2f 2 _]<br />
—[nxmx — m2v 2 _]<br />
r 2 1<br />
—a2[ni?Tix ~ Tn2v_\<br />
axm2v 2 _ - a2mxu%<br />
a3mxv\ — a4m2v 2 __<br />
—[n2m2 — mx^]<br />
a2JVl2 m 2 ~~ ^•1^4.]<br />
— a4\jl2V(l2 — T^lU-iJ<br />
—a3[?i47TT.2 — mxu , ]<br />
-[n4m2 - mxu\]<br />
ax\n4m2 — mxU^.1<br />
u%v\ ulv 2 _)/2<br />
J- •= Cos(6+-9_); u+ - Cos(0+) + Sin(6+); U_<br />
>- = Sin(8+-6_); v+ = Cos(0+) - Sin(0+); v_<br />
Weak<br />
Coupling<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
0<br />
= Cos{BJ) + Sin{0J)<br />
= Cos{dJ) - Sin(9_)<br />
49
50<br />
X transitions between pure states to<br />
all AB transitions, while there are cross-<br />
peaks between the other X transitions<br />
to all AB transitions and also cross-<br />
peaks between all AB transitions to all<br />
X transitions. The selective inversion of<br />
X^ or X^ by a small angle a pulse<br />
does not cause any perturbation of the<br />
strongly coupled states and hence there<br />
are no cross-peaks from these transi-<br />
tions to all AB transitions. The 90°<br />
third pulse mixes the X-magnetisation<br />
unequally between all the X transitions<br />
giving rise to the auto-peaks. On the<br />
other hand, the selective inversion of<br />
an AB transition perturbs the strongly<br />
coupled states leading to cross-peaks<br />
to X transitions between mixed states.<br />
The non-linear detection pulse in turn<br />
mixes the intensities of all the X transi-<br />
tions giving rise to cross-peaks to even<br />
the X transitions between pure states.<br />
The spectrum is not symmetrical 15).<br />
(ii) 90° - 90° - a Experiment<br />
In this experiment there are no cross-<br />
peaks between all AB transitions to X<br />
transitions between pure states, while<br />
there are cross-peaks to all X transitions<br />
between the mixed states and also cross-<br />
peaks between all X transitions to all<br />
AB transitions. This is due to the fact<br />
Bulletin of Magnetic Resonance<br />
that the second 90° pulse perturbs un-<br />
equally all the transitions of AB as well<br />
as the X spin. The mixed states of the<br />
AB spins do not give directly any cross-<br />
peak to X-pure transitions. Since the<br />
detection pulse is a small angle pulse<br />
it does not mix the X transitions be-<br />
tween pure and mixed states and there-<br />
fore there are no cross-peaks from AB<br />
to X-pure transitions. The appearance<br />
of cross-peaks between X transitions be-<br />
tween pure states and the AB transi-<br />
tions is due to the mixing produced be-<br />
tween the various transitions of the X<br />
spin by the second 90° pulse. This state<br />
of the system is faithfully measured by<br />
the detection pulse. Here also the spec-<br />
trum is not symmetrical 15).<br />
The results of 90° - a - 90° and<br />
90° — 90° — a experiments are trans-<br />
pose of each other. This is due to the<br />
fact that PJVXMIOJX) = -PMXJVC 0 ^)-<br />
The conversion of populations into co-<br />
herences and vice versa are described by<br />
mirror operations(15).<br />
Experimental<br />
Experimental observation of these •<br />
cross-peaks was carried out in acetone<br />
oriented in liquid crystal ZLI 1167.<br />
Acetone oriented in liquid crystal is a<br />
strongly coupled spin system of the type<br />
(A3A3) with C3u C3v symmetry. The
Vol. 14, No. 1-4 51<br />
spectra is shown in Fig.3. From this<br />
spectrum it is clear that there are cross-<br />
peaks from every peak to all others<br />
within the same irreducible representa-<br />
tion. Theoretical simulations of these<br />
cross-peaks show a very good match<br />
with the experimental results(6), con-<br />
firming the existence of strong coupling<br />
induced cross-peaks in the 2D NOE ex-<br />
periments even in the absence of relax-<br />
ation.<br />
B. CROSS-CORRELATIONS IN<br />
2D NOE<br />
If a spin has more than one pathway<br />
for relaxation, then there can be cross-<br />
terms between these pathways that may<br />
contribute to the relaxation of the spin.<br />
For example, if there is another spin<br />
nearby, and the mutual dipolar interac-<br />
tion contributes to the relaxation of the<br />
spin and if in addition the first spin has<br />
a partial relaxation by CSA, there can<br />
be cross-terms between the dipolar re-<br />
laxation and CSA(16). If on the other<br />
hand there is a third spin contributing<br />
to the relaxation of the first two through<br />
dipolar relaxation then there can be<br />
cross-terms between various dipolar in-<br />
teractions and between the dipolar and<br />
CSA interactions contributing to the re-<br />
laxation of the various spins. These<br />
cross-terms known as cross-correlations<br />
are often neglected in the relaxation<br />
analysis such as those using generalized<br />
Solomons equations(17). It turns out<br />
that while the cross-terms may be sig-<br />
nificant in magnitude their manifesta-<br />
tion in a particular experiment may be<br />
small. For example the dominant ef-<br />
fect of the cross-terms is to make the re-<br />
laxation of various transitions of a spin<br />
unequal. In a given spin system or in<br />
an experiment if these transitions are<br />
not resolved then this dominant effect<br />
of cross-terms is absent. This can hap-<br />
pen for example when the spins are not<br />
J-coupled or if one uses a 90° pulse for<br />
measuring the intensities of the multi-<br />
plet. In the later case the non-linearity<br />
of the pulse yields an average intensity<br />
over all the transitions of a spin oblit-<br />
erating the multiplet effect and largely<br />
the cross-correlation effects. The use<br />
of a small flip angle for the measuring<br />
pulse is a necessary requirement for the<br />
observation of the multiplet effect and<br />
in turn the cross-correlation effects in<br />
the ID and the standard NOESY ex-<br />
periments^).<br />
In two-dimensional NOE experiment<br />
the most significant attempts to ob-<br />
serve the effect of cross-correlations<br />
have been made by Bodenhausen and<br />
his group(7,16,18-20). One of the ex-
52<br />
periments they have used is a small<br />
flip angle NOESY experiment namely<br />
90°—1\—a—rm—a—£2 , where a is small.<br />
Each cross-section of the small flip angle<br />
NOESY (NOESY 90° - a - a) is then<br />
equivalent to a ID difference transient<br />
NOE experiment in which the peak cor-<br />
responding to the diagonal peak is se-<br />
lectively inverted. This experiment has<br />
both the direct pumping effects and the<br />
cross-correlation induced multiplet ef-<br />
fects present which are measured by the<br />
small angle third pulse. For example the<br />
intensities of the X diagonal and the AX<br />
cross-peak multiplet in a weakly coupled<br />
three spin (AMX) system, in the initial<br />
rate approximation are given by(7),<br />
Xi<br />
x2<br />
x3<br />
x4<br />
Ax<br />
A2<br />
A3<br />
A4<br />
dn<br />
l[A<br />
KM<br />
_ hAM<br />
r rw<br />
r3 0)<br />
pf<br />
X2 X3 X4<br />
1\A<br />
'lM<br />
r2 x)<br />
(o)<br />
P3<br />
r4 0)<br />
«A<br />
r(0)<br />
P2 0)<br />
r3 1}<br />
P ( 4 1}<br />
^2AM<br />
L<br />
dTT<br />
_<br />
pf 1<br />
r(o)<br />
r 2<br />
P3 X)<br />
r 4<br />
(7)<br />
where X1; X2, X3, X4 are the four<br />
X transitions and Ax, A2, A3, A4 are<br />
the four A transitions. The expressions<br />
for the various intensities of the peaks<br />
are given in ref (7) except that when<br />
the cross-correlations due to CSA and<br />
Bulletin of Magnetic Resonance<br />
dipole-dipole interaction are included<br />
W11 ^ wu ? WU £ w^ and l\i ^ l[i<br />
where i = A, M or X. The r and p<br />
terms signify regressive and progressive<br />
peaks respectively. From eqn[7] it is<br />
seen that while the cross-correlation in-<br />
formation is contained in the small flip<br />
angle NOESY experiment, it is coupled<br />
with the direct pumping effects.<br />
We propose here simple modifications<br />
to the small flip angle NOESY (NOESY<br />
90° — a — a). If the second or the third<br />
pulse is made 90° then the intensities<br />
in the initial rate approximation are ob-<br />
tained as averages of the multiplets in<br />
either u>x or o>2 direction respectively.<br />
This removes the direct pumping effects<br />
from the 2D spectra. The following re-<br />
sults are obtained.<br />
90° - a - 90° NOESY<br />
The intensities of the various peaks in<br />
the initial rate approximation are given<br />
by<br />
x2<br />
x3<br />
x4<br />
Ax<br />
A2<br />
A3<br />
A4<br />
Xx X2 X3<br />
Rz<br />
2 ^2<br />
C3 Cz<br />
Ri<br />
R2<br />
Rz R2<br />
Rz<br />
Cz<br />
x4<br />
R2<br />
Rz<br />
Ci<br />
C2<br />
(8)
Vol. 14, No. 1-4<br />
where<br />
i2i =<br />
R2<br />
=<br />
Rs =<br />
R4 —<br />
Co =<br />
c4 =<br />
l0AM<br />
l\M<br />
XM +<br />
rx ^AX-<br />
Sx)]rm<br />
+ hAM + l\M +<br />
- (Px -<br />
A A<br />
(!) , (1) 1 (0) , (0)<br />
r x + Px + r x + Pi<br />
MX<br />
(9)<br />
(10)<br />
Here p^ is the rate of self relaxation of<br />
spin X, a AX 1S the cross-relaxation rate<br />
between spins A and X, 6X = SAXMXI<br />
which gives the cross-correlation rate<br />
between the dipolar vectors AX and<br />
MX and A^ gives the cross-correlation<br />
raten between the dipolar vector AX<br />
and the CSA of spin X. The expres-<br />
sions for the spectral density functions<br />
for the various relaxation rates (p, cr,<br />
A, 6) are given in ref(ll). The inten-<br />
sities of the various peaks in each mul-<br />
tiplet are identical in u>2 dimension and<br />
differ in o?j dimension, the differences<br />
directly yielding the cross-correlations.<br />
If the multiplet is resolved in the wi<br />
dimension the difference in the intensi-<br />
ties of the inner or the outer lines gives<br />
the dipole-CSA cross-correlations(A s)<br />
and the difference between the inner and<br />
outer lines gives the dipole-dipole cross-<br />
correlations (6 s). The diagonal multi-<br />
plet result is identical to the differences<br />
in the initial rates of recovery of the<br />
outer and inner multiplets in inversion-<br />
recovery Ti measurements (21). How-<br />
ever many analyses of inversion recov-<br />
ery measurements including (21) ignore<br />
CSA-dipole cross-correlations, while re-<br />
taining dipole-dipole cross-correlations.<br />
90° - 90° - a NOESY<br />
The intensities of the diagonal and<br />
the cross-peak multiplets in the initial<br />
rate approximation in this case are given<br />
by<br />
x2<br />
X,<br />
At<br />
where<br />
c2 =<br />
c3 =<br />
c: =<br />
X4<br />
R\ R2 Rs R4<br />
R\ R2 Rs R4<br />
R\ R2 R3 R4<br />
r> r> r? r><br />
ti\ Ii2 Kz it.4<br />
' ri' ri' p' ri'<br />
1 2 o 4<br />
ri' ri' rt' ri 1<br />
^1 ^2 U 3 W<br />
ri' r" r>' r>'<br />
O-i uo Lyo Ly A<br />
X Jr O 4<br />
ri' r" si' ri'<br />
Oi Wo L/o Ly^<br />
2{aAX + A; lx + 6A)Tm<br />
2(VAX - A; AX ~~ "A) T m<br />
2{
54<br />
a b cdefg h<br />
* •<br />
4 %<br />
* »<br />
500<br />
i j k<br />
#<br />
• *<br />
: • ' .<br />
I I<br />
pmm'l<br />
\> til,<br />
0<br />
CO2<br />
1e' d c b o<br />
,, d c b<br />
Ml./<br />
* f • »•- * •<br />
• « #<br />
•t: : :::.<br />
* •<br />
0 9 • • *.<br />
• • •<br />
• • • • • •<br />
Bulletin of Magnetic Resonance<br />
Figure 3. 2D NOESY spectrum of oriented acetone recorded at 400 MHz with<br />
rm = 20 /isec. The cross-peaks are mainly due to strong coupling. Zero- quantum<br />
interference during rm was shifted out in another experiment and the residual strong<br />
coupling peaks showed satisfactory correlation with the calculated intensities(6).<br />
(a)<br />
(b)<br />
-500<br />
_Jj\A<br />
Figure 4. Cross-sections taken from (a) 90° - 90° - 15° (b) 90° - 90° - 90° 2D<br />
NOESY experiment with rm = r0 + k using a 400 MHz spectrometer. r0 was 400<br />
msec and k was randomly varied between 10 and 1000 msec.<br />
-500<br />
500
Vol. 14, No. 1-4<br />
From these expressions it is seen that<br />
the intensities differ in u>2 and an aver-<br />
aging takes place along u>i . The sec-<br />
ond 90° pulse excites the multiplet as<br />
a whole, the correct state being moni-<br />
tored by the small angle a pulse. The<br />
differences in the intensities again yield<br />
the cross-correlations except that in this<br />
case the AX multiplet yields 8% and<br />
A^. The diagonal multiplet has in-<br />
tensities identical to 90° — a — 90° ex-<br />
periment. Since in the 90° — 90° — a<br />
experiment the intensities differ in UJ2<br />
domain which is easier to resolve, this<br />
experiment may be preferred over the<br />
90° - a - 90° experiment. In addi-<br />
tion, since all the lines of a multiplet<br />
along uj\ have equal intensities, a u)\ -<br />
decoupled (90° - (A + h)/2 - 180° -<br />
(A - *i)/2 - 90° -rm-a-t2) NOESY<br />
experiment can replace the undecoupled<br />
(90° - tx - 90° - rm - a - t2) NOESY<br />
experiment without loss of information.<br />
Experimental<br />
Two-dimensional NOESY experiment<br />
was carried out in 2,3-dibromo propi-<br />
onic acid using the 90° - 90° - a se-<br />
quence with small a(~ 15°) and a mix-<br />
ing time of 400msec plus a random vari-<br />
ation from 10 to 1000msec. Some of<br />
the cross-sections are shown in Fig.4.<br />
( The differences between the intensities<br />
of various transitions of a multiplet in-<br />
dicate the presence of cross-correlations.<br />
Lack of any particular symmetry in<br />
these multiplets indicates the presence<br />
of both dipole-dipole and CSA-dipole<br />
cross-correlations.<br />
CONCLUSIONS<br />
The use of 90° angle for the ex-<br />
citation or detection pulses allows an<br />
easier method for studying the cross-<br />
correlations in 2D NOE experiment.<br />
One can use either the second or the<br />
third pulse as small angle pulse to high-<br />
light the cross-correlation effects. In<br />
strongly coupled spins the non-linearity<br />
of the pulses can give rise to cross-peaks<br />
even in the absence of relaxation. The<br />
origin of these cross-peaks arising due<br />
to the non-linearity of the second or the<br />
third pulse are discussed with the help<br />
of an ABX spin system.<br />
Acknowledgement<br />
AK wishes to acknowledge discussion<br />
with Ms. Irene Burghardt of University<br />
of Lausanne regarding Fig.l.<br />
References<br />
*R. R. Ernst, G. Bodenhausen and<br />
A. Wokaun,'Principles of Nuclear Ma g-<br />
netic Resonance in One and Two Di-<br />
mensions', Clarendon, Oxford, 1987.<br />
55
56<br />
2 J. Keeler, D. Neuhaus and M. P.<br />
Williamson, J. Magn. Reson. 73, 45-<br />
68 (1987).<br />
3 I. Solomon, Phys. Rev. 99, 559-565<br />
(1955).<br />
4 I. Solomon and N. Bloembergen, J.<br />
Chem. Phys. 25, 261-266 (1956).<br />
5 S. Macura and R. R. Ernst, Mol.<br />
Phys. 41, 95-117 (1980).<br />
6 R. C. R. Grace and Anil Kumar, J.<br />
Magn. Reson. 97, 184-191 (1992).<br />
7 H. Oschkinat, D. Limat, L. Emsley<br />
and G. Bodenhausen, J. Magn. Reson.<br />
81, 13-42 (1989).<br />
8 J. Keeler and F. S. Fernando, J.<br />
Magn. Reson. 75, 96-109 (1987).<br />
9 T. E. Bull, J. Magn. Reson. 72,<br />
397-413 (1987).<br />
10 V. V. Krishnan and Anil Kumar, J.<br />
Magn. Reson. 92, 293-311 (1991).<br />
11 C. Dalvit and G. Bodenhausen,<br />
Adv. Magn. Reson. 14, 1-32 (1990).<br />
12 L. G. Werbelow and D. M. Grant,<br />
Adv. Magn. Reson. 9, 189-299 (1977).<br />
13 D. Canet, Prog. NMR. Spectrosc.<br />
21, 237-291 (1989).<br />
14 J. A. Pople, W. G. Schneider<br />
and H. J. Bernstein, 'High Resolu-<br />
tion Nuclear Magnetic Resonance Spec-<br />
troscopy', McGraw-Hill, New York,<br />
1959.<br />
15 R. C. R. Grace and Anil Kumar,<br />
(unpublished results)<br />
Bulletin of Magnetic Resonance<br />
16 I. Burghardt, R. Konrat and G. Bo-<br />
denhausen, Mol. Phys. 75, 467-486<br />
(1992).<br />
17 J. H. Noggle and R. E. Schirmer,<br />
'The Nuclear Overhauser Ef-<br />
fect : Chemical applications', Academic<br />
Press, New York, 1971.<br />
18 H. Oschkinat, A. Pas-tore and G.<br />
Bodenhausen, J. Am. Chem. Soc. 109,<br />
4110-4111 (1987).<br />
19 C. Dalvit and G. Bodenhausen, J.<br />
Am. Chem. Soc. 110, 7924-7926<br />
(1988).<br />
20 J. M. Bohlen, S. Wimperis and G.<br />
Bodenhausen, J. Magn. Reson. 77,<br />
599-605 (1988).<br />
21 E. Ilyina and Daragan (in press).
Vol. 14, No. 1-4 57<br />
Deriving Structures from 2D NMR. A Method for Defining the<br />
Conformation of a Protein Adsorbed to Surfaces<br />
With the development of multi-dimensional<br />
NMR methods for the specific assignment of many of<br />
the *H NMR signals of small proteins, it is now possible<br />
to determine three-dimensional solution structures<br />
and dynamics for these proteins. Unfortunately<br />
these methods fail to provide detailed structural<br />
information on proteins bound to macroscopic<br />
surfaces because of the very slow overall rotational<br />
correlation time of the particle-bound protein. Indeed<br />
other spectroscopic and structural methods<br />
have provided few details of the structure of proteins<br />
when adsorbed to surfaces (1). Commercially,<br />
surface-bound, immobilized enzymes provide an important<br />
method for efficient utilization of these catalysts<br />
in bioreactors. The development of biosensors<br />
and novel biomaterials such as self-assembled monolayers<br />
(2) requires a better understanding of the<br />
structure of proteins bound to surfaces (1). Identification<br />
of specific residues of proteins involved in<br />
interactions with stationary phase surfaces is critical<br />
in understanding their chromatographic behavior<br />
(3). Finally, the association of proteins with<br />
hydrophobic or glass surfaces has been shown to<br />
cause denaturation or partial unfolding of the pro-<br />
. tein at the surface, with often irreversible loss of<br />
the strongly adsorbed protein. It is believed that<br />
;.these irreversibly bound proteins undergo a confor-<br />
|mational change to expose a portion of the interior<br />
sidues to the surface for effective adsorption.<br />
J this report we describe an NMR methodology<br />
[it for the first time allows us to probe the detailed<br />
^formation of a protein bound to a macroscopic<br />
This was accomplished by 2D amide hy-<br />
«i exchange NMR spectroscopy (4, 5, 6). In a<br />
?i, the NH exchange rates for different residues<br />
over a factor of greater than 10 8 (4, 5).<br />
fate of amide hydrogen/deuterium exchange<br />
"I on the pH (being both acid and base cat-<br />
! the accessibility of the hydrogen to solvent,<br />
Mlization by hydrogen-bonded secondary as<br />
tertiary structure and finally, local fluctu-<br />
•jf the protein (7). Importantly amide hy-<br />
David A. Keire and David G. Gorenstein*<br />
Department of Chemistry<br />
Purdue University<br />
West Lafayette, Indiana 47907<br />
drogen exchange can be studied under conditions<br />
where high resolution resolvable proton NMR signals<br />
would otherwise not be observed - this would<br />
be true for a protein immobilized at a polymeric or<br />
glass surface. However, analysis of the degree of NH<br />
exchange is feasible if the protein can be desorbed<br />
from the surface and the 2D NMR study of the free<br />
protein carried out in solution.<br />
As demonstration of the feasibility of the<br />
method, we have used amide hydrogen exchange<br />
NMR spectroscopy of lysozyme bound to a hydrophobic<br />
chromatographic stationary phase support.<br />
Hen egg white lysozyme (E.C.3.2.1.17) was<br />
chosen for this initial study because the ^-NMR<br />
assignments have been made (8), the crystal structure<br />
was known to 2 A resolution (9) and much was<br />
known about its adsorption onto hydrophobic substrates<br />
(10). Lysozyme is one of the major constituents<br />
of protein deposits on contact lenses (11)<br />
and is commonly used as a standard in the chromatographic<br />
separation of proteins on reverse-phase<br />
columns.<br />
In our protocol we first adsorb lysozyme to the<br />
hydrophobic surface (solid 5 \i diameter polystyrene<br />
divinylbenzene chromatographic stationary phase<br />
support; Polymer Labs Inc., Amherst, Mass.). This<br />
forms a tightly bound single monolayer on the<br />
surface (10). We then expose the surface bound<br />
lysozyme to D2O under fast amide hydrogen exchange<br />
conditions (high pH), desorb the protein under<br />
slow NH exchange conditions (low pH) using<br />
a detergent, and after removal of the detergent, run<br />
the high resolution 2D spectra in D2O solution. The<br />
intensities of the 2D NHCHa TOCSY crosspeaks<br />
then reflect the degree of exchange of the specific<br />
residue amide hydrogens when adsorbed to the hydrophobic<br />
surface.<br />
The stock solution of lysozyme (Sigma) was purified<br />
using a home built hollow fiber bundle dialysis<br />
device versus 10 mM NaH2PC>4 pH = 7.4 buffer. An<br />
aliquot of the stock solution containing 125 mg of
58<br />
lysozyme was diluted to 10 mL in buffer and mixed<br />
with 10 g of hydrated stationary phase support for<br />
30 min. with stirring. The solution was then removed<br />
by the use of a filter funnel with a medium<br />
grain glass frit.<br />
Eighty milliliters of buffer were used to wash the<br />
stationary phase support in 10 to 20 mL aliquots.<br />
For each aliquot a sample of the filtrate was taken<br />
for UV determination of protein concentration (e280<br />
= 2.313 O.D. mL/mg) to ensure that only an "irreversibly"<br />
bound monolayer coverage remained.<br />
Typically, out of 125 mg of lysozyme initially added<br />
~25 mg would remain on the support after extensive<br />
washing. The stationary phase support has a<br />
surface area of ~3 A/gm and a rough calculation<br />
using the molecular dimensions of lysozyme (4.5 x 3<br />
x 3 nra 3 ) shows that ~50 mg of protein could adsorb<br />
to 10 gm of support with monolayer coverage. The<br />
25 mg of lysozyme adsorbed represents 50% coverage<br />
which is in good agreement with the 65% coverage<br />
by the irreversibly bound layer determined by<br />
Schmidt et al. (10) to a similar hydrophobic surface.<br />
At this point a 10 mM NaH^PCU D2O solution<br />
pH* = 7.0 (uncorrected pH meter reading) was prepared<br />
and mixed with the lysozyme covered support.<br />
The adsorbed protein was then stirred for 45<br />
min. during which time the amide hydrogens that<br />
are exposed to solvent can exchange with deuterons.<br />
After 45 min. the solution was aspirated away<br />
and another 10 mL of the buffered D2O solution at<br />
pH*=2.5 was added to quench the amide exchange.<br />
Two aliquots of 0.1% Triton X-114-RS (Sigma)<br />
detergent in 10 mL D2O, pH* = 2.5 were used to<br />
remove the tightly adsorbed protein. The reduced<br />
form of Triton was used to allow determination of<br />
desorbed protein concentration by UV. The ~30<br />
mL of desorbed lysozyme/Triton solution was then<br />
concentrated by pressure dialysis (Amicon, Beverly,<br />
Mass.) to ~4 mL using a 5000 molecular weight cutoff<br />
membrane. The 4 mL was then passed through<br />
an Extracti-Gel column (Pierce) equilibrated with<br />
pH* = 2.5 D2O buffer to remove the Triton. The<br />
column eluent was further concentrated to ~750 fib<br />
and placed in an NMR tube. Approximately 5 mg<br />
(0.5 mM) of desorbed lysozyme was obtained by this<br />
procedure for the NMR study. Much of the protein<br />
loss occurred during the repurification scheme since<br />
the adsorbed protein was nearly quantitatively removed<br />
from the surface by the detergent treatment.<br />
Bulletin of Magnetic Resonance<br />
In the control experiment the identical procedure<br />
was used except that only 5 mg of lysozyme was<br />
used and no stationary phase support was present.<br />
An additional control was run without stationary<br />
phase support and without Triton to ensure that<br />
any residual Triton not detectable in the ID NMR<br />
spectrum did not interfer with the amide exchange<br />
results. This control gave essentially the same integrated<br />
area for the TOCSY cross peaks of the NH-<br />
CaH region as the control involving Triton.<br />
The 600 MHz X H-NMR NH-CaH region of the<br />
30 ms TOCSY spectra of the control and surface<br />
adsorbed/desorbed lysozyme is shown in Figure 1.<br />
Assignment of the signals was based upon complete<br />
analysis of the TOCSY spectra, taking advantage<br />
of the reported 1 H assignments of lysozyme at 500<br />
MHz (8). Normally, a series of TOCSY spectra at<br />
various time points are taken to measure the rate<br />
of NH exchange. Since this has been shown to be<br />
first order, comparing the spectra at a fixed time<br />
of exchange under conditions where the protein is<br />
either surface-bound or not will also provide a measure<br />
of the relative degree of NH protection. We<br />
define protons as being "exposed" when their crosspeaks<br />
present in the control spectrum are decreased<br />
in intensity by >40% relative to the surface exposed<br />
protein spectrum. Protons are "protected" if the<br />
crosspeaks are either absent in the control spectra or<br />
increase in intensity by >40% in the surface exposed<br />
sample spectra. Three control experiments (without<br />
the support present) were used to calculate the percent<br />
deviation from the mean of the integrated areas<br />
of 39 of the slow exchanging amide proton NH-CaH<br />
TOCSY cross peaks. The mean of the 39 deviations<br />
from the means of the integrated areas was 20±13%.<br />
Thus, only those changes in integrated area >40%;<br />
were considered exposed or protected after averag|<br />
ing the cross peak areas from two separate adsorbaj<br />
protein and control experiments. These results ar|<br />
tabulated for the slow exchanging amides [as definw<br />
by Redfield & Dobson (8) at pH = 3.8, 35°C, n/2.<br />
the half life for exchange > 1-5 h] in Table I. Also<br />
included in Table I are the integrated intensities oj<br />
2 intermediate exchange (1 h < r1/2 < 5 h) and on«<br />
fast exchange amide hydrogens (T1/2 < 1-5 h) whicg<br />
are protected relative to the control.<br />
All of the amides which are protected (Table.*<br />
with the exception of the 2 arginines are neutr<br />
or hydrophobic residues. A CPK model {Figuiej
Vol. 14, No. 1-4 59<br />
B<br />
Fl (ppm)<br />
5.5 3.5 3.0 2.5<br />
Fl (ppm)<br />
1: Identical 600 MHz ^-NMR TOCSY experiments at 35°C and pH*=2.5 were run on both the<br />
>1 and adsorbed lysozyme samples. The data was collected in phase sensitive mode with a mixing time<br />
using a MLEV17 spin lock and 2 ms trim pulses. 2K points were collected in F2 and 312 in Fl<br />
^sweep width of 7500 Hz and a repetition time of 3 s. 32 transients were coadded at each ti increment.<br />
£JD signal was suppressed through low power decoupling at the HOD frequency. The spectra were<br />
with zero filling to 2K points in the Fl dimension and a 45° phase shifted sine bell apodization in<br />
tensions.
60 Bulletin of Magnetic Resonance<br />
based upon the crystal structure of lysozyme (9)]<br />
shows that all of the protected amide residues (labeled<br />
by shading) are found on the surface. Four<br />
of the protected amide residues (T40, Q41, A82<br />
and L84) are located close in space at the hinge region<br />
between the a-helical and /3-sheet domains of<br />
lysozyme opposite the active site cleft.<br />
All of the exposed amides - those that show enhanced<br />
amide exchange upon binding to the surface<br />
- with the exception of W63 are involved in ahelices<br />
or /3-sheet structures and are mostly buried<br />
in the interior of lysozyme (Table I). Of the exposed<br />
amides only D52 and W63 show significantly exposed<br />
side chains as shown in the CPK model (Figure<br />
2; exposed residues shown by stripes).<br />
The protein amide hydrogens that are observable<br />
by 2D amide exchange spectroscopy are mostly the<br />
slow exchanging amides (58/129) which are either<br />
involved in secondary or tertiary structure hydrogen<br />
bonding or buried in the interior of the protein<br />
and inaccessible to solvent (12). Surprisingly, one<br />
fast exchanging amide hydrogen (III 14), and two intermediately<br />
exchanging hydrogens (T40 and Q41)<br />
CaH-NH cross peaks were observable in the spectrum<br />
of lysozyme exposed to the hydrophobic surface<br />
which were not present in the control spectrum<br />
(Figure 1).<br />
Figure 2A shows that the protected (stippled)<br />
amide hydrogen side chains lie on one face of the<br />
globular protein in a narrow ridge from R125 to<br />
L17. In contrast, in the back-side view (Figure 2B),<br />
only a few of the largely buried side chains of the<br />
exposed residues (striped) are visible. None of the<br />
protected residues are visible in this view. These results<br />
suggest that lysozyme is not randomly oriented<br />
with respect to the surface, but that it is oriented<br />
with a relatively hydrophobic ridge facing towards<br />
the hydrophobic surface. This is consistent with the<br />
observed retardation of amide hydrogen exchange in<br />
binding an amphiphilic helix to a micelle (13), believed<br />
to result from burial of the hydrophobic face<br />
of an amphiphilic helix into the hydrophobic interior<br />
of a detergent micelle.<br />
In addition a number of residues exchange more<br />
rapidly when the protein is surface bound than when<br />
it is in solution. All of these are not exposed to<br />
the solvent in the native structure but are located<br />
on elements of structure that are likely involved in<br />
segmental motion of the two domains that form the<br />
active site cleft (14). Binding of the hydrophobic<br />
ridge of the protein to the surface thus appears to<br />
induce a conformational change, exposing the active<br />
site residues and residues adjacent to the active site<br />
to solvent.<br />
These perturbations in the amide exchange rates<br />
do not simply reflect proximity of the protein to<br />
the surface because enhancement and protection towards<br />
exchange are observed. In addition most<br />
of the amide hydrogen exchange rates which are<br />
not exposed or protected are the same for surface<br />
bound or free lysozyme. In contrast only decreases<br />
in amide exchange rates are observed in binding an<br />
amphiphilic helix to a micelle (13) and in crystalline<br />
lysozyme (15).<br />
Taken together these data indicate that<br />
lysozyme adsorbs to the surface on the side opposite<br />
the active site cleft. This protects this narrow ridge<br />
of amides from exchange by either blocking solvent<br />
access to these residues at the hydrophobic surface<br />
or by reducing the rate of local fluctuations in the<br />
surface oriented residues. In addition upon binding<br />
to the hydrophobic surface a significant disruption<br />
of several of the buried a-helices and the active site<br />
cleft occurs as the protein partially unfolds at the<br />
surface presumably by opening of the "hinge" at<br />
the active site (16), exposing a number of buried<br />
residues that now show enhanced rate of hydrogen<br />
amide exchange. This conformational change alters<br />
the amide solvent exposure and/or local fluctuations<br />
that allow access of solvent to these interior residues.<br />
This partial unfolding exposes the catalytically<br />
important D52 residue and other important residues<br />
in the active site cleft. This model is supported by<br />
the fact that the enzyme is inactive on adsorption<br />
to the hydrophobic surface of alkylated silica (10).<br />
These conclusions are also in agreement with the<br />
results of Fausnaugh and Regnier (3) based upon<br />
analysis of chromatographic behavior of various bird<br />
lysozymes that also suggests the protein adsorbs on<br />
a side opposite the active site with a contact surface j<br />
that extends from residues 41 to 102 and 75 to 89.1<br />
Computer modeling has also revealed a relatively^<br />
hydrophobic patch identified as a possible binding<br />
site in this region and total internal reflection i j<br />
sic fluorescence (TIRIF) shows a decreased quantun^<br />
yield (and hence altered conformation) of the<br />
sorbed hen lysozyme (17).<br />
Importantly, the surface desorbed lysozj
Vol. 14, No. 1-4<br />
A<br />
B<br />
wire 2: Three dimensional structure of hen lysozyme. A) Front-side view of a CPK model [MIDAS modeling<br />
", UCSF (21)] showing amide NH's which are relatively protected (shaded and stippled) or exposed<br />
and striped) when the protein is bound to the polystyrene surface. The active site cleft is oriented<br />
ne upper left. B) No residues that are protected from exchange (shaded and stippled) can be seen<br />
^.ackside view of the protein. Several of the side chains of residues that are more exposed (shaded and<br />
,
62<br />
Lysozyme CH-NH TOCSY Cross Peaks<br />
A9<br />
All<br />
L17<br />
W28<br />
C30<br />
A31<br />
N37<br />
T40<br />
Q41<br />
D52<br />
W63<br />
A82<br />
L84<br />
C94<br />
198<br />
R114<br />
W123<br />
R125<br />
1.19<br />
4.26<br />
4.53<br />
1.06<br />
0.68<br />
0.27<br />
4.71<br />
4.36<br />
0.99<br />
1.15<br />
2.55<br />
2.40<br />
2.73<br />
0.59<br />
0.32<br />
1.26<br />
0.71<br />
1.30<br />
2.84<br />
2.47<br />
1.03<br />
1.49<br />
1.28<br />
0.24<br />
0.87<br />
1.91<br />
1.45<br />
Slow Exposed<br />
Slow Exposed<br />
Bulletin of Magnetic Resonance<br />
Slow Protected<br />
Slow Exposed<br />
Slow Exposed<br />
Slow Exposed<br />
Slow Protected<br />
Inter. Protected<br />
Inter. Protected<br />
Slow Exposed<br />
Slow Exposed<br />
Slow Protected<br />
Slow Protected<br />
Slow Protected<br />
Slow Exposed<br />
Fast Protected<br />
Slow Protected<br />
Slow Protected<br />
a.) Blanks indicate no observable cross peak.<br />
b.) The integrated area of the adsorbed protein CH-NH TOCSY cross peaks were normalized via the integraM<br />
area of the non-exchangeable Trp 108 H4-H5 cross peak in the control and adsorbed protein spectra. Wt<br />
integrated areas in the table are the average of two control and two adsorbed protein experiments. Jjj<br />
c.) Slow, intermediate and fast exchange amide protons are classified as per Redfield and Dobson (4). |ip<br />
d.) Exposed amides are present in the control and diminished by >40% in integrated area or absent inj|<br />
adsorbed protein spectra. Protected amides are present in the adsorbed protein spectra and diminish^<br />
integrated area by >40% or absent in the control spectra. |i
Vol. 14, No. 1-4 63<br />
spectra have identical chemical shifts as the native<br />
form spectra. This suggests that while the protein<br />
partially unfolds at the surface, it refolds by<br />
the same kinetically or thermodynamically favorable<br />
pathway upon desorption. Thus, the surface<br />
unfolded state may represent a protein folding intermediate<br />
of native hen lysozyme. Only a transient<br />
folding intermediate has been previously identified<br />
in lysozyme (12). By binding the protein to a surface<br />
it may be possible to trap a partially unfolded<br />
state of lysozyme that bears some resemblance to<br />
the transient folding intermediate observed by Miranker<br />
et al. (12). This would be complementary to<br />
the rapid quench NMR amide hydrogen exchange<br />
spectroscopy methods (12, 18, 19, 20) which provides<br />
structural details on the initial refolding kinetic<br />
intermediates. Our method would presumably<br />
provide information on the initial unfolding intermediate.<br />
Surface 2D amide exchange spectroscopy offers<br />
a new method by which protein adsorption can be<br />
monitored at the level of individual residues. This<br />
work demonstrates for the first time the feasibility<br />
of the method which should be applicable to a number<br />
of other enzymes. The method offers a more detailed<br />
picture of protein adsorption than provided by<br />
currently used techniques (e.g. TIRF, ATR-FTIR<br />
and Raman spectroscopy (1)). Future work will<br />
include the extension of the methodology to other<br />
surfaces and proteins. Indeed at the XVth International<br />
Conference on Magnetic Resonance in Bi-<br />
. ological Systems (Jerusalem, Israel, August 16-21,<br />
,., 1992, abstracts) K. Kawano et al., described simii<br />
-lar application of the NH exchange experiment to<br />
\j the binding of lysozyme to hydroxyapatite. They<br />
'also find that certain residues are protected from<br />
nange when the protein is adsorbed to the sur-<br />
However unlike our results no enhancement of<br />
•exchange is observed.<br />
!' Acknowledgments<br />
rSupported by the Office of Naval Research<br />
jp00l4-91-J-1686) and the Purdue University Bio-<br />
4cal Magnetic Resonance Laboratory which is<br />
ported by the NSF Biological Facilities Center on<br />
Secular NMR, Structure and Design at Pur-<br />
[grants BBS 8614177 and DIR-9000360 from<br />
jjyision of Biological Instrumentation) and NIH<br />
•ferences<br />
1. J. D. Andrade and V. Hlady, Adv. Polym.<br />
Sci. 79, 1-63 (1986).<br />
2. K. L. Prime and G. M. Whitesides, Science<br />
252, 1164-1166 (1991).<br />
3. J. L. Fausnaugh and F. E. Regnier, J. Chromatogr.<br />
359, 131- 146 (1986).<br />
4. H. Roder, "Methods in Enzymology" (N. J.<br />
Oppenheimer and T. L. James, eds. ), 446-473, Academic<br />
Press, Inc., 1989.<br />
5. S. W. Englander and N. R. Kallenbach, Q.<br />
Rev. Biophys. 19, 521-655 (1984).<br />
6. H. Roder, G. Wagner, and Wuthrich, Biochemistry<br />
24, 7396-7407 (1985).<br />
7. Y. Paterson, S. W. Englander, and H. Roder,<br />
Science 249, 755-759 (1990).<br />
8. C. Redfield and C. M. Dobson, Biochemistry<br />
21, 122-136 (1988).<br />
9. C. C. F. Blake, G. A. Koenig, A. C. Mair, C.<br />
T. North, D. C. Philips, and V. R. Sarma, Nature<br />
206, 757-761 (1965).<br />
10. C. F. Schmidt, R. M. Zimmermann, and H.<br />
E. Gaub, Biophys. J. 57, 577-588 (1990).<br />
11. E. J. Castillo, J. L. Koenig, and J. M. Anderson,<br />
Biomaterials 6, 338-344 (1985).<br />
12. A. Miranker, S. E. Radford, M. Karplus, and<br />
C. M. Dobson, Nature 349, 633-636 (1991).<br />
13. C. Karslake, M. E. Piotto, Y. M. Pak, H.<br />
Weiner, and D. G. Gorenstein, Biochemistry 29,<br />
9872-9878 (1990).<br />
14. C. C. F. Blake, G. A. Mair, A. C. T. North,<br />
D. C. Phillips, and V. R. Sarma, Proc. R. Soc.<br />
Lond. Ser. B 167, 365-377 (1967).<br />
15. T. G. Pedersen, B. W. Sigurskjold, K. V.<br />
Andersen, M. Kjaer, F. M. Poulsen, C. M. Dobson,<br />
and C. Redfield, J. Mol. Biol. 218, -413-426<br />
(1991).<br />
16. J. A. McCammon, B. R. Gelin, M. Karplus,<br />
and P. G. Wolynes, Nature 262, 325-326 (1976).<br />
17. D. Horsley, J. Herron, V. Hlady, and J. D.<br />
Andrade, "Proteins at Interfaces", 290-305, 1987.<br />
18. H. Roder, G. A. Elove, and S. W. Englander,<br />
Nature 335, 700 (1988).<br />
19. J. B. Udgaonkar and R. L. Baldwin, Nature<br />
335, 694-699 (1988).<br />
20. C. M. Dobson and P. A. Evans, Nature 335,<br />
666 (1988).<br />
21. T. E. Ferrin, C. C. Huang, L. C. Jarvis, and<br />
R. Langridge, J. Mol. Graphics 6, 13-27 (1988).
64<br />
Flavoridin, a protein with 70<br />
amino acids from the venom of<br />
Trimeresurus gramineus, is a very<br />
potent inhibitor of blood platelet<br />
aggregation. The protein contains<br />
a local Arg-Gly-Asp (RGD)<br />
sequence at position 49 to 51. This<br />
primary sequence element is<br />
known to inhibit fibrinogen<br />
binding by a specific interaction<br />
with the integrin-type platelet<br />
receptor GPIIb/IIIa. By now, a<br />
rather large family of homologous<br />
RGD containing snake toxins have<br />
been sequenced. Most of these<br />
proteins contain n=12 cysteines<br />
which are all linked by disulfide<br />
bridges. Previous biochemical<br />
studies, however, have so far not<br />
revealed the native pattern of the<br />
individual cysteine pairings.<br />
^ l , Werner Klaus and Pau, Gerber<br />
The ! H NMR spectrum of<br />
Flavoridin was almost fully<br />
assigned in aqueous solution by<br />
conventional 2D ^H NMR methods<br />
(2QF-COSY, clean TOCSY, 2Qspectroscopy,<br />
NOESY). The 3Dstructure<br />
calculation with<br />
distance geometry methods<br />
(DIANA) proceeded in several<br />
rounds:<br />
(1) The global fold of Flavoridin<br />
was calculated from the collected<br />
set of NOE distance constraints<br />
(#666) and dihedral angle<br />
constraints obtained from vicinal<br />
coupling constants (#88) but<br />
without the use of any disulfide<br />
bridge constraints.<br />
(2) The interatomic cP-C-P distances<br />
between all possible pairs [n(nl)/2]<br />
of cysteines were measured<br />
in a set of 20 converged distance<br />
geometry structures. A<br />
probability weight wjj (0 - wij - 1)<br />
i« assigned to each individual<br />
Bulletin of Magnetic Resonance<br />
cysteine pair according to a<br />
gaussian shaped distribution<br />
function. When the average<br />
crystallographic CP-CP distance of<br />
a cystine disulfide bridge is<br />
matched, the weight wy = 1.<br />
(3) All combinatorial patterns of 6<br />
disulfide-bridges involving the 12<br />
cysteines in Flavoridin were<br />
calculated and the 6 individual<br />
weights, w^j, for each pattern<br />
were multiplied. Only patterns<br />
with individual wjj — 0.3 were<br />
considered. On this objective basis,<br />
a single Cys-Cys pairing pattern<br />
could unambiguously be<br />
determined. The method was<br />
validated with known protein<br />
crystal structures of Cys-rich<br />
proteins. Also the experimentally<br />
observed NOE's between Ci«H/CjPH<br />
and/or CiP/CjP of Cys residues,<br />
which commonly are taken as<br />
direct evidence for Cys-Cys<br />
disulfide links, do agree' with the<br />
computationally evaluated Cys-Cys<br />
pattern.<br />
(4) In a second step of the<br />
structure calculation, the Cys-Cys<br />
pairing was used as additional<br />
input for the distance geometry<br />
program. Finally, the 50 best<br />
structures were refined by<br />
energy minimization.<br />
Our structural results show that<br />
the polypeptide backbone is folded<br />
in two domain-like structures -<br />
composed of 8 turns and stabilised<br />
by 6 cystine cross-links. The<br />
conformation of the Arg-Gly-Asp<br />
(RGD) sequence is located in an<br />
extended loop structure exposed at<br />
the tip of a so called hairpin,<br />
which is rather flexible.
Vol. 14, No. 1-4 65<br />
1. Introduction<br />
NMR Approaches to Large Proteins:<br />
trp Repressor and Chloramphenicol Acetyltransferase<br />
L.-Y. Lian, J. P. Derrick, V. Ramesh, R. O. Frederick,<br />
The current upper limit for protein structure<br />
determination by *H nmr [1] is in the region<br />
of Mr 12-15,000. The use of stable isotope<br />
labelling, with 2 H, 13 C or 15 N, can<br />
substantially extend this limit, perhaps to Mr<br />
30,000 (for methodological reviews, see [2]-<br />
[4]). However, there remain many interesting<br />
and important proteins which are much larger<br />
than this, and we have been trying to<br />
establish what information nmr can provide in<br />
such cases. Two systems we have been<br />
studying in this context are the E. coli trp<br />
repressor and the enzyme chloramphenicol<br />
acetyltransferase.<br />
The trp repressor is a dimer of total Mr<br />
•; 25,000, and is a member of the "helix-turnpielix"<br />
family of DNA binding proteins. A<br />
umber of high resolution crystal structures<br />
the protein are available, but the<br />
hanisms of its activation and DNAinding<br />
specificity remain poorly understood,<br />
"lain priority is therefore to determine<br />
solution structure of the proteinion<br />
ucleotide complex, and the effects on<br />
of changes in protein and oligonucleotide<br />
S ure; our work to date has lar s el y been<br />
* to developing the necessary methods<br />
S. E. H. Syed and G. C. K. Roberts<br />
Biological NMR Centre, University of Leicester,<br />
PO Box 13 8, Medical Sciences Bldg.,<br />
University Road, Leicester LEI 9HN, U.K.<br />
for this. We have, in collaboration with the<br />
group of Jardetzky at Stanford, achieved a<br />
virtually complete sequence-specific assignment<br />
of the *H nmr spectrum of the protein<br />
[5]-[7], and Jardetzky's group have used this<br />
data to determine the solution structure of<br />
the repressor-tryptophan complex [8]. Nmr<br />
studies of the binding of corepressors such as<br />
L-tryptophan and the inducer, indole-3propionic<br />
acid, show that the environments of<br />
the two classes of ligand in the protein differ,<br />
and strongly suggest that this arises from a<br />
difference in the orientation of their indole<br />
rings [9]. Even the repressor alone is of such<br />
a size that these studies required the<br />
extensive use of isotope labelling, and we<br />
have developed appropriate expression<br />
systems for efficient biosynthetic labelling of<br />
the protein. The overall molecular mass of<br />
the repressor-tryptophan-operator oligonucleotide<br />
complex is ca. 38,000, and for this<br />
complex rather little useful information can<br />
be obtained by nmr without labelling.<br />
Chloramphenicol acetyltransferase is<br />
responsible for resistance to the antibiotic<br />
chloramphenicol in bacteria. Its crystal<br />
structure is known to high resolution, and it<br />
is the subject of a substantial programme of<br />
protein engineering in the laboratory of Prof.
66 Bulletin of Magnetic Resonance<br />
W.V. Shaw in Leicester [10]. Since it is a<br />
trimer of total molecular mass 75,000, it is<br />
very large for study by nmr, and we are using<br />
it as a test system for the development of nmr<br />
methods suitable for large proteins, focussing<br />
particularly on complexes, such as that with<br />
the product, diacetyl-chloramphenicol, which<br />
have not proved amenable to crystallographic<br />
study.<br />
2. Deuteration<br />
Selective deuteration has been used to<br />
simplify the *H nmr spectra of proteins for<br />
many years [11], and more recently has been<br />
combined with 2D nmr spectroscopy (e.g.,<br />
[12]). Selective deuteration of carefully<br />
chosen combinations of residues has been<br />
combined with the sequential assignment<br />
method of Wiithrich (see [1]) to yield a<br />
considerable number of resonance<br />
assignments in the trp repressor, both alone<br />
and in its complex with an operator<br />
oligonucleotide [6], [7]. With the improved<br />
resolution of the 3D spectra, we have<br />
recently assigned all the intermolecular NOEs<br />
between protons of the protein and those of<br />
bound tryptophan, permitting the kind of<br />
'ligand-docking' experiments which we have<br />
already carried out on phospholipase A2 [13].<br />
In a species as large as the repressor-operator<br />
complex, the substitution of most of the<br />
protons in the protein by deuterons leads to a<br />
notable decrease in the resonance linewidth<br />
of the remaning protons; similarly, we have<br />
employed random fractional deuteration [14]<br />
to good effect in these large systems. In<br />
particular, deuteration of chloramphenicol<br />
acetyl-transferase to the level of about 85%<br />
had three beneficial effects: (a) it allowed the<br />
unambiguous observation and assignment of<br />
resonances of the bound substrate, (b) it<br />
allowed the observation of intra-molecular<br />
NOEs in the bound substrate, thus defining<br />
its conformation, and (c) it allowed the<br />
observation of inter-molecular NOEs<br />
between protons of the (isotopically normal)<br />
substrate and nearby residues of the protein<br />
[15].<br />
3. 13 C and 15 N labelling<br />
Notwithstanding the usefulness of selective<br />
deuteration, complete assignments of the<br />
backbone resonances of the repressor have<br />
required the use of 15 N-labelled protein, in<br />
combination with 3D nmr spectroscopy [7].<br />
The same approach has led to a substantial<br />
number of resonance assignments in the<br />
repressor-operator complex, and to the clear<br />
observation of inter-molecular NOEs [7],<br />
[16]. In addition to the standard NOESY-<br />
HMQC and TOCSY-HMQC experiments,<br />
we have found the HMQC-NOESY-HMQC<br />
experiment [17] valuable in this a-helical<br />
protein in which there is significant overlap of<br />
both *H and 15 N chemical shifts of the<br />
backbone amides. In some cases, there is a<br />
considerable benefit in selective labelling; in<br />
the trp repressor we have used this<br />
successfully for 15 N-leucine and for [amide-<br />
15 N]-asparagine and glutamine [16], and in<br />
chloramphenicol acetyltransferase for<br />
[imidazole 2- 13 C]-histidine [18]. The latter<br />
experiment permitted the detection of the<br />
histidine C2- 1 H signals, and the use of 2D<br />
*H- 13 C correlation spectra allowed<br />
overlapping signals to be resolved, even in a<br />
protein of Mr 75,000. We are also exploring<br />
the usefulness of 50-65% perdeuteration in<br />
combination with 15 N-labelling as a means of<br />
improving the quality of the 3D spectra.<br />
In chloramphenicol acetyltransferase we<br />
have used nmr to study the binding of<br />
diacetyl[ 13 C]-chloramphenicol by means of<br />
13 C-edited l U- l B. NOESY experiments. A<br />
number of clear intermolecular NOEs were<br />
observed, in particular to aromatic protons of<br />
the protein. Candidate aromatic residues<br />
were identified by model-building on the<br />
basis of the crystal structure, and were<br />
replaced in turn by isoleucine residues. 13 Credited<br />
NOESY spectra of the complexes with<br />
these mutants allowed the two aromatic<br />
residues with which the acetyl groups of the<br />
ligand made contact to be identified<br />
unambiguously, thus allowing the orientation<br />
of the product in the binding site to be
Vol. 14, No. 1-4 67<br />
defined [15], and assisting in the modelling of<br />
the transition-state complex.<br />
4. References<br />
[1]<br />
[2]<br />
[3]<br />
[4]<br />
[5]<br />
[6]<br />
[7]<br />
[8]<br />
[9]<br />
[10]<br />
Wiithrich, K., Science 243 (1989) 45.<br />
Bax, A. (1989) Ann. Rev. Biochem. 58,<br />
223.<br />
Oppenheimer, N.J., and James, T.L.<br />
(1989) Methods in Enzymology, 176,<br />
Academic Press, New York.<br />
Oppenheimer, N.J., and James, T.L.<br />
(1989) Methods in Enzymology, 177,<br />
Academic Press, New York.<br />
Hyde, E.I., Ramesh. V., Roberts.<br />
G.C.K., Arrowsmith, C.H., Treat-<br />
Clemons, L., Klaic, B., and Jardetzky,<br />
O. (1989) Eur. J. Biochem. 183,545.<br />
Arrowsmith, C.H., Pachter, R.,<br />
Altman, R.B., Iyer, S.B., and<br />
Jardetzky, O. (1990) Biochemistry 29,<br />
6332.<br />
Ramesh, V., Frederick, R.O., Syed,<br />
S.E.H., and Roberts, G.C.K., unpublished<br />
work; Arrowsmith, C.H., and<br />
Jardetzky, O., unpublished work.<br />
Arrowsmith, C.H., Pachter, R.,<br />
Altman, R.B., and Jardetzky, O. (1991)<br />
Eur. J. Biochem., 202, 53.<br />
Hyde, E.I., Ramesh, V., Frederick, R.,<br />
and Roberts, G.C.K. (1991) Eur. J.<br />
Biochem., 201, 569.<br />
Shaw, W.V., and Leslie, A.G.W.<br />
(1991) Ann. Rev. Biophys. Biophys.<br />
Chem., 20, 363.<br />
[11] Jardetzky, O. & Roberts, G.C.K.<br />
(1981) NMR in Molecular Biology,<br />
Academic Press, New York.<br />
[12] Feeney, J., Birdsall, B., Akiboye, J.,<br />
Tendler, S.J.B., Barbero, J.J., Ostler,<br />
G., Arnold, J.R.P., Roberts, G.C.K.,<br />
Kuhn, A., & Roth, K. (1989) FEBS<br />
Lett., 248, 57.<br />
Bennion, C, Connolly, S., Gensmantel,<br />
N.P., Hallam, C, Jackson, C.G.,<br />
Primrose, W.U., Roberts, G.C.K.,<br />
Robinson, D.H., and Slaich, P.K.<br />
0992) J. Med. Chem., in press.<br />
[14] LeMaster, D.M. (1990) Quart. Rev.<br />
Biophys., 23, 133.<br />
[15] Derrick, J.P., Lian, L.-Y., Roberts,<br />
G.C.K., and Shaw, W.V. (1992)<br />
Biochemistry, in press.<br />
[16] Ramesh, V., Frederick, R.O., Syed,<br />
S.E.H., and Roberts, G.C.K. (1992)<br />
manuscript in preparation.<br />
[17] Frenkiel, T., Bauer, C, Carr, M.D.,<br />
Birdsall, B., and Feeney, J. (1990) J.<br />
Magn. Reson., 90, 420.<br />
[18] Derrick, IP., Lian, L.-Y., Roberts,<br />
G.C.K., and Shaw, W.V. (1991) FEBS<br />
Lett., 280, 125.
68<br />
Bulletin of Magnetic Resonance<br />
2D NMR STUDY OF DRUG-PROTEIN INTERACTIONS : ETHIDIUM BROMIDE -<br />
NEOCARZINOSTATIN COMPLEX<br />
Introduction<br />
Smita Mohanty*> Larry C. Sieker"*" and Gary P. Drobny*<br />
Department of Chemistry* and Biological Structure 1 "<br />
University of Washington<br />
Seattle, WA 98195, USA<br />
N eocarzinostatin (NCS) is a small<br />
acidic holo-protein isolated from the<br />
culture broth of Streptomyces<br />
Carzinostaticus [1]. It has a protein<br />
component (apo-NCS) of 113 amino acid<br />
residues and a non-covalently bound<br />
heat and light sensitive chromophore<br />
(NCS-chr) (Fig.l). This protein possesses<br />
antibiotic activity against organisms<br />
such as Sarcina Lutea and antitumor<br />
activity against the experimental tumors<br />
Ascitic Sarcoma 180, Ascitic<br />
Leukemia SN-36, Leukemia L-210<br />
[1-3]. It is known that the chromophore<br />
is responsible for all the biological<br />
activities and the apo-protein stabilizes<br />
and acts as a carrier for this UV<br />
sensitive component of the antitumor<br />
drug [4].<br />
Though the secondary and tertiary<br />
structure of the apo-protein is well<br />
understood from X-ray and NMR studies,<br />
little is known about the binding region<br />
and the amino acid residues involved in<br />
the drug - protein interactions in the<br />
holo-protein.<br />
Both the crystal structure at 2.8 A<br />
[5] and the 2-D NMR work done on apo-<br />
NCS [6-8] indicate that a major part of<br />
the protein is composed of a seven<br />
strand antiparallel B-sandwich formed<br />
by a three strand B-sheet and a four<br />
strand B-sheet. The rest of the protein is<br />
composed of two loops oriented<br />
somewhat perpendicularly to the B<br />
sandwich, thus forming a distinct Ushaped<br />
cleft between the four strand<br />
face of the sandwich and one of the<br />
loops of the molecule (Fig. 2). The<br />
crystal structure of holo-NCS at 2.0 A<br />
resolution indicates that the<br />
chromophore is located in this cleft. But<br />
a detailed knowledge in the region of<br />
the chromophore could not be obtained<br />
from the 2.0A map. Preliminary NMR<br />
studies also indicate that the<br />
chromophore binds within the cleft and<br />
interacts with amino acid residues in the<br />
region C37-D48 [9, 10]. Although NCS<br />
does not bind the chromophore of other<br />
streptomyces derived anti-tumoi<br />
proteins (e.g. Auromomycin) [11] it is<br />
known from X-ray studies [12] tc<br />
strongly bind a number of drugs<br />
including ethidium bromide (Fig. 3) anc<br />
daunomycin. In order to furthe<br />
elucidate the nature of drug-NC!<br />
interactions, we have initiated a study o<br />
both holo-NCS and the complex betwee<br />
ethidium bromide and NCS.<br />
Materials and Methods<br />
Apo-NCS solution used in o 1<br />
experiments was prepared from hoi<br />
NCS. The chromophore was extracted<br />
i
Vol. 14, No. 1-4<br />
Fig. 1: NCS-Chromophore<br />
60<br />
Fig. 2: Ribbon picture of NCS by X-ray<br />
Fig. 3: Eihidium Bromide<br />
69
I:<br />
I<br />
70<br />
the procedure of Napier et.al.[13].<br />
Purified lyophilized holo-NCS was a gift<br />
from Kayaku Co. Ltd. The 1:1 Ethidium<br />
Bromide-NCS complex was prepared by<br />
adding the protein solution in to vials<br />
containing the drug. The final solution<br />
was purified by passing through<br />
sephadex G-25 column followed by<br />
lyophilization. The lyophilized complex<br />
was brought up in lOmM acetate buffer<br />
(pH 5, 90% H2O/ 10% D2O for non<br />
exchanged protein sample and 99.98%<br />
D2O for exchanged protein sample) and<br />
lOmM EDTA. The concentration was<br />
adjusted between 2.0 mM to 2.5 mM for<br />
500|il sample.<br />
All NMR experiments were<br />
performed on a Bruker AM-500<br />
Spectrometer at 313 K. DQF- and TQF-<br />
COSY [14], RELAY [15], TOCSY [16] and<br />
NOESY [17] were acquired in TPPI mode<br />
with standard phase cycling schemes.<br />
The water resonance was presaturated<br />
by selective irradiation between 1.5 s to<br />
2 s. RELAY spectra were performed with<br />
mixing times of 30 ms (90%H2O) and 25<br />
ms (for 99.98%D2O). TOCSY spectra were<br />
performed with a variety of mixing<br />
times ranging between 40 to 80 ms.<br />
NOESY spectra were recorded with 150<br />
ms mixing time, randomly varied by<br />
10%. The data were processed with<br />
FTNMR software of Dr. Dennis Hare [18].<br />
Results and Discussion<br />
Our H-NMR assignments of<br />
ethidium bromide - NCS complex in<br />
solution indicates the drug to be located<br />
in the cleft region. There are two lines of<br />
evidence which support this conclusion.<br />
First, the chemical shifts of numerous<br />
Bulletin of Magnetic Resonance<br />
SI nn — horn ethidium<br />
to the ring ^ aromatic system,<br />
bromide's extensive aromatic y<br />
£> tp"t e -45 Gln Cy;-47 y , Asp-**<br />
Cy's-9 Gln-94, Leu-97 in the<br />
fingerprint region (Fig. 5 and F.g. 6).<br />
0.0<br />
ppm<br />
Fig 4: DQF-COSY spectrum in D2O<br />
' snowing uprtcldshid or Uu_4<br />
chemical shifts (a): apo-NCS. (b).<br />
P ,
Vol. 14, No. 1-4<br />
O36<br />
3 '•<br />
Q94" 5<br />
C47 'C37<br />
XL -<br />
L97<br />
8.8 8.0<br />
ppm<br />
9<br />
7!2<br />
Fig. 5: DQF-COSY fingerprint region of apo-NCS recorded at 40'C .<br />
so<br />
10 03<br />
C93<br />
DO<br />
OK)<br />
00<br />
i< 00<br />
00<br />
72<br />
8.0<br />
ppm<br />
7. 2<br />
Bulletin of Magnetic Resonance<br />
Fig.7: DQF-COSY and NOESY spectra of the complex in D2O<br />
showing some of the intermolecular NOE observed between<br />
the aromatic protons of ethidium bromide (shown by dotted<br />
lines) and B-protons of certain residues within the cleft.
Vol. 14, No. 1-4 73<br />
While residues Cys-93, Gln-94,<br />
Leu-97 are in the four strand face of the<br />
6-sandwich, which forms one side of the<br />
cleft, residues Gln-36, Cys-37, Ala-38,<br />
Trp-39, Leu-45, Cys-47, Asp-48 are in<br />
one of the loops that forms the other<br />
side of the cleft. Second, a number of<br />
NOEs have been observed to occur<br />
between protons on ethidium bromide<br />
and residue protons within the cleft. The<br />
Leu-45 methyl group is ring current<br />
shifted and shows NOEs to the<br />
methylene protons and aromatic protons<br />
of ethidium bromide. Intermolecular<br />
NOEs are also observed between the<br />
aromatic protons of ethidium bromide to<br />
the aromatic proton of Trp-39 and to the<br />
8 protons of Ser-98, Cys-37 and Gly-96<br />
(Fig. 7). Additional intermolecular NOEs<br />
are observed but have not been<br />
unambiguously assigned.<br />
Conclusion<br />
NMR assignment based on<br />
i coherence transfer experiments and<br />
y
74<br />
[11] L. S. Kappen, M. A. Napier, I. H.<br />
Goldberg, and T. S. A. Samy,<br />
Biochemistry., 19, 4780.(1980).<br />
[12] L. C. Sieker, Personal communication(<br />
1991).<br />
[13] M. A. Napier et.al., Biochem.<br />
Biophys. Res. Commun., 89,<br />
635. (1979).<br />
[14] U. Piantini, O. W. S0rensen, and<br />
R. R. Ernst, J. Am. Chem. Soc,<br />
104, 6800. (1982).<br />
[15] G. Eich, G. Bodenhausen, and R.<br />
R. Ernst, J. Am. Chem. Soc, 104,<br />
3732. (1982).<br />
[16] L. Braunschweiler and R.R.<br />
Ernst, /. Magn. Reson., 53, 521.<br />
(1983).<br />
[17] Anil Kumar, R. R. Ernst, and K.<br />
Wuthrich, Biochem. Biophys.<br />
Res. Commun., 95, 1. (1980).<br />
[18] D. Hare, Hare Research:<br />
Woodinville, WA.<br />
BulJetin of Magnetic Resonance<br />
'•A
Vol. 14, No. 1-4 75<br />
1 Introduction<br />
Transferred NOE experiments are being utilized to<br />
determine the structure of small peptide ligands bound to<br />
proteins of molecular weight up to 10 6 daltons (1-4). For<br />
larger peptides, 2D phase-sensitive NOESY is the optimal<br />
method for obtaining transferred NOE's between all the<br />
protons in the free ligands (1-2). In case of further<br />
spectral overlap, transferred NOE's between protons of<br />
selected amino acid residues can be obtained by use of the<br />
homonuclear TOCSY-editted 2D NOESY technique (5).<br />
Ultimately, NOE's between all the spin systems may be<br />
resolved by use of the 3D NOESY-TOCSY experiment<br />
(6). The resolved NOE's can be input into a procedure for<br />
the refinement of the dynamic structures of bound ligands<br />
(7).<br />
In practice, the interpretation of NOESY data is<br />
complicated by the presence of the large solvent resonance<br />
and baseplane problems, especially for samples with very<br />
low concentration of the material under study. In addition,<br />
magnetization transfers from the binding protein often<br />
result in further spectral distortions. We have been<br />
, developing methods to improve the quantitative accuracy<br />
|of transferred NOE's (7-9). In this paper, we summarize<br />
|Procedures developed for the optimized acquisition and<br />
ssing of multi-dimensional transferred NOE spectra<br />
• that more accurate quantitative results can be obtained<br />
i experimental data.<br />
Quantitative Analysis<br />
in Multi-Dimensional Transferred NOE Experiments:<br />
Improved Spectral Acquisition and Processing<br />
jlimination of Baseline Distortions in<br />
volution Dimensions<br />
evel of tj ridges in 2D phase-sensitive NOESY<br />
~ can be minimized if the FID matrix is recorded<br />
proportional phase incrementation (TPPI) with<br />
lulation along n (10). Since the tj interferograms<br />
zero at the zero time for a sine-modulated<br />
to our implementation is to start the ex-<br />
Feng Ni<br />
Biotechnology Research Institute<br />
National Research Council Canada<br />
6100 Royalmount Avenue<br />
Montreal, Quebec, Canada H4P 2R2<br />
periment at the second FID with a compensated initial<br />
delay of,<br />
where IN = At/2 is the increment time between successive<br />
FID's and x9o is the width of the 90° pulse. This<br />
implementation of the sine modulated NOESY experiment<br />
completely removes Fj baseline distortions (Figure 1A)<br />
with no need for any further baseline correction along this<br />
dimension (8). Furthermore, the resulting NOESY matrix<br />
can be phased to absorption along the F i direction with a<br />
Oth order phase of exactly 90° (in effect, a sine transform).<br />
This eliminates possible baseline distortions associated<br />
with phase correction after a real Fourier transformation<br />
(11). The procedure of delay-compensated sine modulation<br />
has also been applied successfully along the evolution<br />
dimensions of both homonuclear and heteronuclear 3D<br />
experiements (9,12).<br />
3 Ridge Suppression Along Detection<br />
Dimension<br />
The cause for ridges (baseline offsets and curvatures) in the<br />
detection dimension is more complicated than that for<br />
ridges along the evolution dimensions. There is often<br />
need for further correction in the frequency domain. Figure<br />
2a shows the baseline points recognized based on the first<br />
derivative (13) of a row slice of the NOESY spectrum in<br />
Figure 1A. The regions with sharp peaks are filled with<br />
interpolations (linear or polynomial) from adjacent<br />
baseline points. It is seen that the recognized baseline<br />
includes all the broad signals in the original spectrum.<br />
One can then construct a smooth curve through the<br />
available baseline points with some sort of curve fitting<br />
(Figures 2b and 2c). We adopted a simple and fast method<br />
[1]
76<br />
Bulletin of Magnetic Resonance<br />
Figure 1: Enhanced processing of 2D transferred NOE spectra. The FID matrix was acquired using sine-modulation along the<br />
ti direction. All spectra were processed by use of cosine-square windows in both directions and were plotted with the same<br />
parameters except that post-acquisition water suppression (8) was applied in (A)-(D); linear-prediction baseline correction was<br />
applied in (B); polynomial baseline correction was applied in (C); and baseline Fourier filtering was applied in (D).<br />
0.0
Vol. 14, No. 1-4<br />
10.0 8.0 0.0<br />
ppm<br />
Figure 2: Baseline fitting from incomplete data, (a) baseline points from one row of Figure 1 A. The missing points were<br />
filled with linear interpolations, (b) one possible baseline reconstructed by Fourier smoothing, (c) an approximate baseline<br />
calculated by a fit to a polynomial of fifth degree.<br />
for data smoothing based on Fourier filtering (8). Figure<br />
2b is the result of a 30-point Fourier smoothing of<br />
Figure 2a. Figure ID shows the NOESY spectrum<br />
(Figure 1A) after subtraction of the Fourier-filterred<br />
baseline points along the F2 direction. In comparison to<br />
-••• other methods (Figures IB and 1C), there is a dramatic<br />
... improvement in the clarity of the spectrum with complete<br />
^ elimination of broad signals from the original spectrum.<br />
Optimized Spectral Processing<br />
frequency domain spectrum is usually generated via a<br />
i Fourier transform of the interferogram if the data are<br />
"pled using the TPPI procedure (14). The computed<br />
"jtrum contains both real and imaginary parts which are<br />
|;cosine or sine transforms of the original real data,<br />
ctively. Theoretically, an absorption spectrum is<br />
ly the cosine transform of a cosine modulated FID or<br />
i;ne transform of a sine modulated FID (10). In<br />
however, both the real and imaginary parts have<br />
Iculated so that they can be suitably combined<br />
»g) to correct for possible phase distortions. For<br />
|«mensional NMR experiments, the evolution<br />
not ~ns can usually be sampled using the procedure of<br />
[jpensated sine (or cosine) modulation (see section<br />
2) to obtain well-phased interferograms (8,9,15). Thus,<br />
the processing procedure can be greatly simplified if only<br />
the sine or the cosine transform of the data is retained. We<br />
thus implemented the procedures for fast sine (or cosine)<br />
transformation (16). Compared to a real Fourier<br />
transform, there is in principle a factor of two increase in<br />
computational efficiency (16).<br />
To minimize truncation artifacts, special attention<br />
must be paid to the selection of window functions and/or<br />
to data extension by use of linear prediction. Attenuation<br />
of truncation is usually accompanied by broadening of<br />
spectral peaks (14). Linear prediction data extension, on<br />
the other hand, tends to increase the noise level of the<br />
spectrum as a result of errors accumulated with predicted<br />
data points (17, 18). We found that linear prediction<br />
followed by data windowing usually gives a good compromise<br />
between spectral resolution and noise suppression<br />
(9). In this case, truncation is removed by linear<br />
prediction while linear prediction errors are attenuated by<br />
window functions (18). In multi-dimensional NMR,<br />
linear prediction should be used only with the last spectral<br />
dimension after Fourier transformations along all other<br />
directions. Otherwise, linear prediction parameters must<br />
be carefully optimized to minimize error accumulation in<br />
the intermediate stages of multi-dimensional Fourier<br />
transformation.<br />
77
78<br />
5 Sensitivity Enhanced 3D NOESY-<br />
TOCSY<br />
In the usual pulse sequence for 3D NOESY-TOCSY,<br />
spin-locked coherence transfers are achieved by use of the<br />
trimmed MLEV-17 sequence (Figure 3A). We replaced<br />
the MLEV-17 sequence by a z-filtered WALTZ-16 pulse<br />
sequence (Figure 3B) to obtain sensitivity enhancement.<br />
This method involves post-acquisition combinations of<br />
the FIDs acquired from sub-groups of the full phase cycle<br />
utilized for the pulse sequence (9). The two detected FIDs<br />
B<br />
MLEV-17<br />
WALTZ-16 ($,,<br />
acq(t3)<br />
Figure 3: Pulse sequences for 3D NOESY-TOCSY. (A).<br />
The original 3D sequence incorporates a MLEV-17 pulse<br />
sandwiched by two short cw spin-lock pulses (trimmed<br />
MLEV-17). (B). The new 3D NOESY-TOCSY features<br />
a z-filtered WALTZ-16 sequence for spin-locking. The rf<br />
phases are cycled as 4>i=x, -x, y, -y; 2=-x, -x, -y -y;<br />
fa=-x, -x, -y, -y; 4=x, x, y, y; sl=y, y, -x, -x; 5=x, x,<br />
y, y; acq=x, -x, y, -y. For the same values of l\ and t2,<br />
5 is incremented by 180 degrees and another FID is<br />
acquired and stored in a different memory location. These<br />
seperated FIDs can be suitably combined to obtain a<br />
sensitivity-enhanced 3D NOESY-TOCSY spectrum<br />
(Figure 4).<br />
are stored in separate memory blocks for subsequent<br />
processing. Adequate combinations of these FIDs would<br />
restore the two orthogonal components, Iiy and l[x. If<br />
both the Iiy and 1^ FIDs are processed to yield absorptive<br />
spectra, they can be combined to produce a spectra with<br />
peaks doubled in size compared to each individual<br />
spectrum. The key to sensitivity enhancement lies in the<br />
fact that noise components in the two spectra are<br />
statistically independent and the post-processing combinations<br />
are then equivalent to signal averaging, reducing<br />
noise in the process (19).<br />
Bulletin of Magnetic Resonance<br />
With the sensitivity-enhanced NOESY-TOCSY, we<br />
utilized delay-compensated sine modulation along both ti<br />
and t2 to eliminate baseline distortions. Baseline<br />
adjustments along F3 were achieved in the frequency<br />
domain by use of Fourier baseline reconstruction (8). An<br />
absorptive spectrum can be obtained from Ijy if a sine<br />
transform is applied along both tj and t2- With the Iix<br />
components, an absorptive spectrum can be obtained only<br />
if a sine transform is used along t\ and a cosine transform<br />
used along t2. However, the points of the 1^ matrix for t2<br />
= 0 can not be sampled due to the finite widths of the 90<br />
degree pulses. Therefore, the missing first point was<br />
estimated via a linear prediction algorithm (17).<br />
Alternatively, the first data point can be left as zero during<br />
cosine transformation. Baseline offsets as a result can be<br />
corrected afterwards in the frequency domain.<br />
Sensitivity-enhanced NOESY-TOCSY was applied to<br />
an anticoagulant pep tide (2). The concentration of the<br />
peptide was 6 mM and the thrombin concentration was 0.5<br />
mM in an aqueous solvent of 90% H2O and 10% D2O at<br />
pH 5.5. The experiment composed of four scans for each<br />
of the two FIDs with each pair of fixed values for ti and<br />
t2- The data were acquired on a Briiker AMX-500 MHz<br />
NMR spectrometer and the 3D FID matrix (I(y or 1^ components)<br />
was of the sizes 512(t3) x 160(ti) x 62(t2).<br />
Residual solvent signals were suppressed by linear<br />
prediction time-domain convolution (8). The data along ti<br />
was extended to 200 and along t2 to 80 by linear prediction<br />
(17). Kaiser windows were used in all dimensions to<br />
reduce the effects of error propagation in linear prediction.<br />
Frequency-domain baseline adjustments were applied only<br />
along the F3 spectral dimension. The final sizes of the<br />
spectra were 512(F3) x 256(Fj) x 128(F2). Figure 4A is<br />
the F1-F2 plane sliced through the frequency of one of the<br />
well-resolved 6CH2 protons of Pro along the F3<br />
dimension of the Iiy 3D spectral matrix. The corresponding<br />
IiX spectrum (Figure 4B) also contains similar<br />
information but with somewhat reduced intensities for<br />
some of the crosspeaks. This is probably due to the fact<br />
that the 1^ components travel through different transfer<br />
pathways and are much more sensitive to prolonged delays<br />
and/or phase offsets both before and after the WALTZ-16<br />
spin-lock pulse. Nonetheless, combination of the two<br />
spectra still produced a sensitivity-enhanced spectrum<br />
(Figure 4C). This is evident if one compares the NOE<br />
crosspeaks between the aCH and PCH protons of ProfiO<br />
and the NH proton of Glufi! (60A/61N and 60B/61N of<br />
Figure 4) and if one inspects the selected slices (Figure 5)<br />
through the sensitivity-enhanced spectrum compared to the.,<br />
original spectrum. |
Vol. 14, No. 1-4<br />
A<br />
t<br />
608/6 IN<br />
.' i |,<br />
608/63E<br />
600/630 1<br />
.6 ft* fom<br />
\ ft<br />
60A/6IN AX<br />
0 /<br />
6.0 4.0<br />
F1( ppm )<br />
a /^ 59C/600<br />
608/6 IN<br />
2.0 8.0 6.0 4.0<br />
F"K ppm)<br />
2.0<br />
Figure 4. 3D transferred NOES Y-TOCSY spectra of an anticoagulant<br />
peptide, DSS-F-E-E-I-P-E-E-Y-L-QGS. (A) the<br />
plane was extracted from the conventional 3D NOESY--<br />
TOCSY Iiy spectrum. Only positive levels above 0.008<br />
are plotted. (B) the same plane as in (4A), but extracted<br />
from the 1^ spectrum. Only negative levels below -0.008<br />
are plotted. (C) the same plane as in (4A) and (4B), but<br />
from the combined spectra of both the and the Ijy components.<br />
The contour levels are above 0.012.<br />
79
80<br />
10.0<br />
o.o<br />
Figure 5. Selected slices from the sensitivity-enhanced 3D<br />
spectrum (top trace in each box) compared to those from<br />
the conventional spectrum (bottom trace in each box).<br />
Top trace of Box A, ID spectrum sliced through the PCH2<br />
chemical shift of Pro GO (labelled as 60B in Figure 4C);<br />
bottom trace of Box A, the corresponding slice from<br />
Figure 4A. Box B, slices corresponding to the aCH<br />
chemical shift of Pro^o (Labelled as 60A in Figure 4).<br />
6 Summary<br />
We have optimized procedures for both acquisition and<br />
processing of homonuclear 2D and 3D spectra in aqueous<br />
solutions. These include a new 3D NOESY-TOCSY<br />
pulse sequence that can be used with sine modulation to<br />
simplify spectral processing and to improve spectral<br />
baselines. It is also demostrated that orthogonal<br />
components of the spin-locked magnetizations can be<br />
suitably combined during processing to achieve sensitivity<br />
enhancements for 3D NOESY-TOCSY. These improved<br />
schemes are not limited to transferred NOE experiments.<br />
They should be of general applicability for resonance<br />
assignments and structure determination of dilute proteins.<br />
7 References<br />
Bulletin of Magnetic Resonance<br />
!R Ni, Y. Konishi, R. B. Frazier, H. A. Scheraga, and S.<br />
T. Lord, Biochemistry 28,3082 (1989).<br />
2 F. Ni, Y. Konishi, and H. A. Scheraga, Biochemistry<br />
29, 4479 (1990).<br />
3A. P. Campbell, and B. D. Sykes, J. Mol. Biol. 222,<br />
405 (1991).<br />
4 S. J. Landry, R. Jordan, R. McMacken, and L. M.<br />
Gierasch, Nature 355,455 (1992).<br />
5 V. Sklenar, and J. Feigon, J. Am. Chem. Soc. 112,<br />
5644 (1990).<br />
6G. W. Vuister, R. Boelens, and R. Kaptein, /. Magn.<br />
Reson. 80,176 (1988).<br />
TF. Ni, J. Magn. Reson. 96,651 (1992).<br />
8F. Ni, /. Magn. Reson. (1992a), in press.<br />
9F. Ni,7. Magn. Reson. (1992b), in press.<br />
10<br />
G. Otting, H. Widmer, G. Wagner, and K.<br />
Wuthrich, J. Magn. Reson. 66,187 (1986).<br />
11D. Marion, and A. Bax, 7. Magn. Reson. 79, 352<br />
(1988).<br />
12K. A. Carpenter, and F. Ni, J. Magn. Reson.<br />
(1992), in press.<br />
13 W. Dietrich, C. H. Riidel, and M. Neumann, J.<br />
Magn. Reson. 97,1(1991).<br />
14<br />
R. R. Ernst, G. Bodenhausen, and A. Wokaun, "<br />
Principles of Nuclear Magnetic Resonance in One and<br />
Two Dimensions", Clarendon Press, Oxford, 1987.<br />
15<br />
D. Marion, and A. Bax, J. Magn. Reson. 83, 205<br />
(1989).<br />
i&W. H. Press, B. P. Flannery, S. A. Teukolsky, and<br />
W. T. Vetterling, "Numerical Recipes'. The Art of<br />
Scientific Computing," Chap.13 and 14, Cambridge Univ.<br />
Press, Cambridge, 1986.<br />
«F. Ni, and H. A. Scheraga, /. Magn. Reson. 70,<br />
506 (1986).<br />
l«E, T. Olejniczak, and H. L. Eaton, /. Magn. Reson.<br />
87,628 (1990).<br />
19 J. Canavagh, and M. Ranee, J. Magn. Reson. 88,<br />
72 (1990).
Vol. 14, No. 1-4 81<br />
TIME-RESOLVED SOLID-STATE NMR: SMALL MOLECULES AND ENZYMES<br />
IN RAPIDLY FROZEN SOLUTION<br />
INTRODUCTION<br />
Jeremy N. S. Evans§**, Richard J. Appleyard§ and Wendy Shuttleworth§<br />
Departments of Biochemistry/Biophysics§ and Chemistry^,<br />
Washington State University, Pullman, WA 99164-4660. U. S. A.<br />
An important aspect of understanding enzymatic reaction<br />
mechanisms is the determination of the molecular structures<br />
of enzyme-bound substrates and products. NMR spectroscopy<br />
is in a unique position in that it is the only relatively<br />
high resolution structural technique which can focus<br />
on specific parts of the molecule, such as enzyme-bound<br />
intermediates. It has been widely applied to the study of enzyme<br />
mechanisms in both the solution and solid states [1-3]<br />
at ambient temperatures and at sub-zero temperatures [4].<br />
However, the current molecular weight limit in solution is<br />
around 50 kDa [2]. In contrast, with solid-state NMR spectroscopy<br />
[5] anisotropic and dipolar broadening can be<br />
reduced by cross-polarization magic angle sample spinning<br />
(CP-MASS), and there is no known molecular weight limit.<br />
We have therefore sought to develop a new technique called<br />
time-resolved solid-state NMR spectroscopy [6] which involves<br />
applying CP-MASS NMR to studying rapidly<br />
freeze-quenched enzyme-substrate mixtures.<br />
The rapid freeze-quench method involves rapidly mixing<br />
enzyme and substrate together and freezing by spraying the<br />
mixture directly into a secondary cryogen such as liquid<br />
propane cooled to ~ 85 K. CP-MASS NMR of the enzymesubstrate<br />
mixture is carried out as a function of mixing<br />
time, and the transient enzyme-bound species may be<br />
detected. Since it is largely unknown how the NMR spectrum<br />
of solutes are affected by the structure of water in<br />
frozen solution, we have studied this with a small molecule,<br />
glycine [7], and report the results in preliminary form here.<br />
We have subsequently applied these methods to the direct<br />
observation of an enzyme-intermediate complex by timeresolved<br />
solid-state NMR spectroscopy [6] and report our<br />
preliminary results here.<br />
MATERIALS AND METHODS<br />
NMR spectroscopy was carried out using a wide-bore,<br />
T Chemagnetics CMX-400 spectrometer, operating at<br />
iPO.l MHz for *H and 100.6 MHz for 13 C. The small<br />
nolecule studies were carried out using zirconia rotors (7<br />
m) in a double resonance, variable temperature (VT),<br />
ic angle sample spinning (MASS) Chemagnetics proto-<br />
Pencil-rotor probe. The enzyme studies were carried out<br />
*? Dotv °SI-368 double resonance probe using a 5mm<br />
jpnire rotor. Stable spinning was achieved with a<br />
"lagnetics spinning speed controller, which uses a<br />
'"-cessor controlled valve on the drive gas line to<br />
the speed ±5 Hz. The MAS probe was of a triple<br />
mel (dnve, bearing and VT) design. Boil-off N2 gas<br />
was cooled by an exchange dewar filled with liquid nitrogen<br />
(280 kPa, -110K). The sample cooling was performed by<br />
cooling the VT line, with the drive and bearing lines at approximately<br />
room temperature (-293K). Stable VT operation<br />
could be achieved for over 24h with this apparatus.<br />
The nuclear relaxation constants were determined by<br />
established techniques [8-10]. The proton Zeeman spinlattice<br />
relaxation, Ti H , was determined using the inversion<br />
recovery technique [11], with 13 C detection via cross<br />
polarization. The carbon Zeeman spin-lattice relaxation,<br />
Ti c , was determined using a modified cross polarization and<br />
inversion recovery technique [12]. The rotating-frame spinlattice<br />
relaxations, Tip H and Tjp C , were determined using<br />
an appropriate spin lock (50kHz) of varied length, with 13 C<br />
detection via cross polarization [9,13]. Unlabelled glycine<br />
was purchased from Sigma (St Louis, MO). 99% [1-<br />
13 C] Glycine was purchased from Cambridge Isotope<br />
Laboratories (Cambridge, MA), and 99% [2- 13 C]Glycine<br />
was purchased from MSD Isotopes (Canada). Propane gas<br />
was purchased from Bemzomatic (Medina, NY). The crystalline<br />
glycine sample was re-crystallized from a saturated<br />
aqueous solution, using ethanol as the triturant. The dimensions<br />
of the crystals were consistent with the a-form. A 1M<br />
1:1 solution (pH 7.5) of [l- 13 Ci]glycine and [2-<br />
13 Ci]glycine was used for the frozen samples. The slow<br />
frozen samples were prepared in situ. A 200 pL sample was<br />
loaded into the rotor and spun slowly (
82 Bulletin of Magnetic Resonance<br />
TABLE 1 Activation Energies for motions in different states of glycine from Ti and T\p data.<br />
Ea (kj mol* 1 )<br />
Crystalline<br />
Slow Frozen<br />
Fast Frozen<br />
(10"<br />
Crystalline<br />
Slow Frozen<br />
Fast Frozen<br />
1 Ks" 1 )<br />
(10 5 Ks 4 )<br />
(10" 1 Ks" 1 )<br />
(10 5 Ks" 1 from Ti<br />
)<br />
H data<br />
/romTj<br />
23.5<br />
21.1<br />
19.4<br />
from Tip** data<br />
28.2<br />
24.6<br />
23.1<br />
c Cl<br />
data<br />
23.5<br />
16.5<br />
19.5<br />
from Tip c data<br />
Cl<br />
13.6<br />
4.4<br />
5.2<br />
BL21(XDE3)(pLysS)(pWS230) and purified by literature<br />
methods [14].<br />
RESULTS<br />
The structures of frozen water have been extensively studied<br />
in electron microscopy [15] and materials science [16],<br />
and has been shown to be greatly affected by the freezing<br />
rate used. The main problem with water is that due to its<br />
strong propensity for hydrogen bonding, it is incredibly hard<br />
not to form crystalline hexagonal ice (I^). Two other forms<br />
of frozen water known are cubic ice (Ic) and vitreous water<br />
(Iv), also referred to as amorphous solid water (ASW). ^ ice<br />
is a metastable form of ice (with respect to 1^ ice) produced<br />
when Iv is heated. In Iv, the water molecules are randomly<br />
distributed throughout the solid phase having been unable to<br />
form an Ih lattice during the freezing process.<br />
The freeze-quench apparatus used in these studies is a<br />
modified version of that used in rapid freeze quench ESR<br />
spectroscopy studies [17,18]. It involves firing a solution<br />
at a pre-determined flow-rate through a fine nozzle into a<br />
receptacle containing a secondary cryogen (e.g. liquid<br />
propane) cooled by immersion in a primary cryogen (usually<br />
liquid nitrogen). The cooling rate is dependent on the size of<br />
the droplets created by the exit nozzle and can be determined<br />
by the study of a reaction with a known rate. ESR studies<br />
[17,18] have estimated the freezing time by this technique to<br />
be 2-6 ms and the cooling rate to be -10 5 Ks' 1 .<br />
We have investigated how nuclear relaxation rates of a<br />
solute in frozen solution are affected by the freezing rate,<br />
since they have been shown to be a sensitive probe of<br />
molecular motion in the solid state [19,20]. We have chosen<br />
to study glycine because of its simple mode of motion that<br />
has been characterized previously using NMR relaxation<br />
techniques [21]. Since the relaxation properties of pure ice<br />
have also been studied [22,23], this provides a good model<br />
for characterization of the molecular motions of solute<br />
molecules in frozen solution.<br />
Previous studies [21,24] of polycrystalline glycine have<br />
shown that the main source of Tj H relaxation is the random<br />
modulation of the proton magnetic dipolar interaction by the<br />
C2<br />
30.2<br />
17.6<br />
19.4<br />
C2<br />
0.26<br />
3.3<br />
1.6<br />
reorienting ammonium group (-NH3+). The activation<br />
energy of the -NH3 + rotation in glycine was determined to be<br />
28.6 kJ mol" 1 which agrees well with our T^ and Ti c Ea<br />
values for crystalline glycine shown in Table 1. The Ea values<br />
from the Tip C data (Table 1) display a deviation from<br />
this value, which is even more evident for the slow and fast<br />
frozen states. However, since the latter fits are less welldefined,<br />
these Ea values should be regarded as less reliable.<br />
The crystal structures of a-glycine has been solved to a<br />
high degree of accuracy by X-ray [25] and neutron diffraction<br />
[26] techniques. The glycine molecule is in the zwitterion<br />
form. The dipolar glycine molecules are linked by two short<br />
-N-H--O- hydrogen bonds to form layers connected in an<br />
antiparallel manner by weaker -N-H •••0- hydrogen bonds.<br />
The close packing nature of the crystalline lattice, coupled<br />
with the hydrogen bond interactions, will produce large barriers<br />
to rotation for both the -NH3"*" and -CO2' groups as<br />
observed.<br />
The relaxation process in the frozen solution is more<br />
complicated due to the presence of an additional relaxation<br />
pathway provided by the ice lattice. Previous studies [22,23]<br />
have shown the measured activation energy for the Tj and<br />
Tip processes in pure crystalline ice to be 59.8 kJ mor 1 .<br />
This gives a xc (-10°C) = 7.5 (is, and it has been concluded<br />
[23] that the dominant magnetic relaxation mechanism must<br />
be the diffusion of Schottky defects through the ice lattice.<br />
Although there are insufficient data points for the number of<br />
variables to fit two correlation times, it is clear that the diffusion<br />
due to Schottky defects is not detected since no activation<br />
energy of ~60 kJ mol" 1 is evident from the Ti dau<br />
for the frozen solutions given in Table 1. The field strengu<br />
used is much higher than those in previous studies<br />
Therefore the (Ti)min occurs at a temperature above tin<br />
melting point of the ice and cannot be detected.<br />
Table 1 clearly shows a drop in Ea for -NH3 group rot*<br />
tion of glycine calculated from the Ti H and Tip H data in tr<br />
order, crystalline; slow-frozen solution; fast-frozen soluua<br />
The structure of the frozen solution samples is less cle|<br />
Many studies [16,27] have been made of water frozen unfl
Vol. 14, No. 1-4 83<br />
a variety of different freezing rates. Much effort has been<br />
made to develop techniques of freezing with rates that<br />
approach 10 5 - 10 10 Ks- 1 , the estimated [28-31] rates<br />
necessary to form vitreous ice, Iv (amorphous solid water).<br />
It is now generally accepted that freezing rates >10 5 Ks" 1<br />
are required before Iv will form, however there are other<br />
factors to consider, such as sample size, freezing medium<br />
and so on. For solutions frozen at rates well below the 10 5 -<br />
10 10 Ks -1 range, the frozen sample is thought to be made<br />
up of bulk hexagonal ice and aggregated solute. For a slowfrozen<br />
1M glycine solution, aggregates of crystalline<br />
glycine will form along with the Ih, i.e. the freezing rate is<br />
slow enough that the crystalline states have time to form.<br />
The crystalline glycine is distributed throughout the Ih<br />
lattice, resulting in a large surface to volume ratio. The<br />
glycine molecules exposed to the ice front have fewer lattice<br />
interactions, giving the -NH3 + group a lower energy barrier<br />
to rotation. This explains the drop in Ea observed in Table I<br />
between crystalline and slow frozen glycine.<br />
Even freezing rates of -10 5 Ks' 1 have been shown [32]<br />
not to prevent completely the segregation of solute from<br />
solvent, nor to form Iv, in both pure water and dilute aqueous<br />
solutions. However, the inclusion of a solute increases<br />
the propensity for vitrification [16], and is the basis for the<br />
use of cryoprotectants. In sufficiently large quantities, a<br />
solute will depress the free energy of the liquid water relative<br />
to the Ih lattice and reduce the driving force for crystallization.<br />
It is therefore unclear what the Iv / Ih ratio is under the<br />
conditions used here. However, once the secondary cryogen<br />
(liquid propane) is removed under vacuum at 223 K, any Iv<br />
that might have been formed at 85 K would readily undergo<br />
a phase transition to Ih at -160 K [33]. Therefore, at the<br />
temperatures used in this NMR study, it is unlikely that Iv<br />
is present. Rapid cooling has been shown [34] to cause a<br />
highly dispersed Ih phase regardless of the degree of vitrification.<br />
A rapidly frozen 1M glycine solution will consist of<br />
.' highly dispersed ice and glycine phases and the very high<br />
^ surface: volume ratio achieved renders the system metastable.<br />
'Under these conditions, maturation can take place with the<br />
^crystal size distribution broadening and shifting to larger<br />
fSfy?^ 1 dimensions. However, the rate is dependent on the<br />
'" ' temperature and is not significant within the timeof<br />
the relaxation studies. It can be rationalized that<br />
very high surface: volume ratio contributes to the<br />
r drop in Ea observed between the slow frozen and fast<br />
solutions. It is also possible that the glycine is<br />
*nt predominantly in the amorphous form in the rapidly<br />
*en sample, since the freezing rate is fast enough to<br />
Jtnt the glycine from forming a microcrystalline<br />
" ""The barriers to the -NH3+ group rotation in the<br />
"i state will be lower since there will not be such a<br />
» of ordered hydrogen bonding, and this will<br />
'- further to the drop in Ea.<br />
applied these methods to a well-characterized<br />
g-€nolpyravylshikimate-3-phosphate (EPSP) syn-<br />
thase (EC 2.5.1.19), which catalyzes the penultimate step in<br />
the aromatic amino acid biosynthetic pathway in higher<br />
plants and bacteria. EPSP (4) is formed from shikimate-3phosphate<br />
(S3P, 1) and phosphoenolpyruvate (PEP, 2) (see<br />
Scheme 1). The enzyme is a monomer with molecular<br />
weight Mx = 46,000 and the cloned E.coli gene has been<br />
used to generate a hyperexpressing strain [14], so that the<br />
bacterial enzyme is available in gram quantities.<br />
Furthermore, EPSP synthase is the primary site of action of<br />
the herbicide glyphosate [35], or N-phosphonomethylglycine.<br />
This is a broad spectrum post-emergence herbicide<br />
with worldwide applications in agriculture and horticulture.<br />
This enzyme has been extensively studied by kinetic and<br />
biophysical methods in the last 5 years. The direct observation<br />
of the enzyme-intermediate (E«I) complex was first reported<br />
by our laboratory [1,2], later confirmed by another<br />
laboratory [36]. There are only a handful of enzymes for<br />
which the full kinetic and thermodynamic profile has been<br />
determined, and EPSP synthase is one of this select group<br />
[37].<br />
Concerns about the fate of the protein under these conditions<br />
of freezing have been addressed [38], and at the protein<br />
concentrations and freezing rates (10 5 K s 1 , or millisecond<br />
time regime) employed here, there is significant dispersal of<br />
the solute in the frozen water [7]. Furthermore the frozen<br />
water is probably largely amorphous [39], with the protein<br />
itself acting in a similar manner to a cryoprotectant [39]<br />
thereby reducing the formation of hexagonal ice that is<br />
detrimental to the protein. We have found that the specific<br />
activity of EPSP synthase employed in the experiments<br />
reported here was found to be unchanged before and after<br />
rapid freezing at these high protein concentrations.<br />
Figure 1 shows 13 C CP-MASS solid-state NMR spectra<br />
of EPSP synthase«S3P mixed with [2- 13 C]PEP under<br />
steady-state conditions (in the presence of the product, inorganic<br />
phosphate) and under pre-steady state conditions where<br />
/ indicates the time elapsed from the start of the reaction.<br />
The intermediate (E«I) is clearly visible at 104 ppm [1,2]<br />
under steady-state conditions and its build-up demonstrated as<br />
the reaction proceeds under pre-steady state conditions. It is<br />
worth noting that the intensities of the E«I resonance correlate<br />
well with the concentrations observed by chemical<br />
quench methods [37]. On allowing the pre-steady state reaction<br />
to proceed for a few minutes, the turnover of intermediate<br />
(E«I) to product (E«EPSP) is evident. In addition to the<br />
resonance due to the E«I complex, the resonance due to the<br />
E'EPSP product complex builds up at 155 ppm, and one<br />
tentatively assigned to the E-PEP substrate complex appears<br />
transiently at 151 ppm. Note that under the conditions<br />
which these spectra were obtained, the free small molecules<br />
(substrate and product) are not detected due to their relative<br />
isotropic motion in frozen solution. Although the spin<br />
locking fields used provide excellent cross-polarization for<br />
rapidly-frozen solutions of EPSP synthase and enzymebound<br />
species also provide very poor cross-polarization for<br />
PEP and EPSP.
84 Bulletin of Magnetic Resonance<br />
240.0 160.0 80.0 0.0<br />
ppm<br />
Figure 1 The 9.4 T 13 C CP-MASS solid-state NMR<br />
spectra of EPSP synthase at 233 K under conditions indicated:<br />
steady-state (EPSP synthase (4 mM) in 20 mM phosphate<br />
buffer, pH 7.8,15 % D2O in the presence of S3P (9.1<br />
mM) and [2- 13 C]PEP (7.6 mM) slow frozen in the NMR<br />
rotor over 90 s; rapidly mixed and freeze-quenched after time<br />
1 (EPSP synthase (4 mM) in 50 mM tris buffer containing<br />
5 mM P-mercaptoethanol, pH 7.8, in the presence of S3P<br />
(40 mM) and rapidly mixed with [2- 13 C]PEP (40 mM) and<br />
sprayed into liquid propane at -85 K.<br />
DISCUSSION<br />
These relaxation studies have provided evidence that the<br />
distribution of the Ih and the solute becomes more<br />
dispersive as the freezing rate is increased. This is<br />
manifested as a lowering of the activation energy for the<br />
-NH3 + group rotation as the amount of glycine free from<br />
crystalline packing forces increases. This is rationalized by<br />
two factors: the increase in surface: volume ratio in the<br />
dispersed phase; and the increase in the percentage of<br />
amorphous glycine present This demonstrates the difference<br />
between fast and slow freezing and suggests that at 10 5 Ks"<br />
1 , either some Iv or highly dispersed Ih is forming. This is<br />
encouraging from the point of view of studying rapidlyfrozen<br />
proteins, since this suggests minimal possible<br />
damage through Ih formation. The characterization of the<br />
freezing rate also indicates that it will be sufficient for<br />
trapping transient intermediates in an enzymatic reaction.<br />
Time-resolved solid-state NMR spectroscopy provides a<br />
major technological advance in the study of enzymatic reaction<br />
mechanisms. One important consideration in any<br />
attempt to detect transient intermediates in addition to their<br />
lifetimes, is their pre-steady state concentrations. This is<br />
dependent upon the kinetics and thermodynamic stability of<br />
intermediates of each individual enzyme. Some enzymes,<br />
like EPSP synthase, have unusually stable intermediates.<br />
However, we would expect that the majority of enzymes are<br />
evolving towards "perfection" [40], and stabilize highly unstable<br />
intermediates. Furthermore, when coupled with the<br />
elegant solid-state NMR distance measurements that have<br />
been introduced recently [41] this technique will be uniquely<br />
able to "map out" molecular conformations of intermediates<br />
and enzyme active site-intermediate distances as an enzymatic<br />
reaction proceeds. This will provide the crucial missing<br />
structural details which Laue X-ray diffraction and allied<br />
techniques cannot provide, and enable the complete definition<br />
of the molecular events of enzyme catalysis.<br />
ACKNOWLEDGEMENTS<br />
We should like to thank Prof. R.C.Bray (University of<br />
Sussex) for helpful discussions in the design of the freezequench<br />
apparatus, and Dr. Yves Dupont of Biologic Co. for<br />
implementing some of these designs. We are also grateful to<br />
Drs. Allan Palmer and Jim Frye of Chemagnetics/Otsuka<br />
Electronics Ltd. for help with the custom design of the lowtemperature<br />
equipment for the NMR spectrometer, member!<br />
of WSU technical services instrument shop and Fret<br />
Schuetze of WSU electronics shop for numerous customiza<br />
tions, and Dr. Dave Cleary for help with the ESR spec<br />
troscopy. This work was supported in the early stages by th;<br />
donors of the Petroleum Research Fund (American ChemicJ<br />
Society) and National Institutes of Health grant RR O60Q|<br />
and mainly by National Institutes of Health graf<br />
GM43215. The WSU NMR Center is supported by NJ<br />
grant RR 0631401, NSF grant CHE-9115282 and Battd<br />
Pacific Northwest Laboratories Contract No.12- 097718-/<br />
L2.
Vol. 14, No. 1-4 85<br />
REFERENCES<br />
11] P.N.Barlow,R.J.Appleyard, BJ.O.Wilson and<br />
J.N.S .Evans, Biochemistry, 28, 7985 and p. 10093<br />
[2]<br />
[ 3]<br />
[4]<br />
[5]<br />
[6]<br />
[7]<br />
[8]<br />
[9]<br />
[10]<br />
[11]<br />
[12]<br />
[13]<br />
[14]<br />
[15]<br />
U6]<br />
[17]<br />
B US]<br />
[19]<br />
:[20]<br />
(1989).<br />
J. N. S. Evans, "NMR and Enzymes", in "Pulsed<br />
Magnetic Resonance, Optics and Imaging (Honoring<br />
E. L. Hahn)", (Ed. D. Bagguley), Oxford University<br />
Press, in press (1992).<br />
J.N.S.Evans, CBurton, P.E.Fagerness,<br />
N.E.Mackenzie & A.I.Scott, Biochemistry 25, 905<br />
(1986); JP.G.Malthouse, Prog, in NMR Spectroscopy<br />
18, pp 1-59 (1985); PHosch, Ibid. 18, 123 (1986);<br />
K.Kanamori & J.D.Roberts, Acc.ChemJRes. 152, 35<br />
(1983); W.W.Bachovchin, Biochemistry 25, 7751<br />
(1986).<br />
J.P.G.Malthouse, M.P.Gamcsik, A.S.F.Boyd,<br />
KE.Mackenzie, and Ai.Scott, JAmer.Chem.Soc.<br />
104, 6811 (1982); N.E.Mackenzie, J.P.G.Malthouse<br />
and A.I.Scott, BiochemJ. 219, 437 (1984).<br />
S.O.Smith, LPalings, V.Copie, D.P.Raleigh,<br />
J.Courtin, J.A.Pardoen, Jlugtenburg, R.A.Mathies<br />
and R.G.Griffin, Biochemistry 26, 1606 (1987).<br />
J.N.S.Evans, RJ.Appleyard & W.A,Shuttleworth,<br />
submitted (1992).<br />
RXAppleyard and JJsf.S.Evans, submitted (1992).<br />
J. S. Frye, Concepts Magn. Res. 1, 27 (1989).<br />
C. A. Fyfe, "Solid State NMR for Chemists," p. 48,<br />
CFC Press, Canada, 1983.<br />
D. W. McCall, Ace. Chem. Res. 4, 223 (1971).<br />
R. L. Void, J. S. Waugh, M. P. Klein and D. E.<br />
Phelps, J. Chem. Phys. 48, 3831 (1968).<br />
D. A. Torchia, /. Magn. Res. 30, 613 (1978).<br />
E. 0. Stejskal, J. Schaefer, M. D. Sefcik and R. A.<br />
McKay, Macromolecules 14,275 (1981).<br />
W.A.Shuttleworth, CD.Hough, KP.Bertrand and<br />
J.N.S .Evans, Protein Engineering 5, in press (1992).<br />
H. Plattner and L. Bachmann, Int. Rev. Cytol. 79,<br />
237 (1982).<br />
C. A. Angell and Y. Choi, /. Microsc. 141, 251<br />
(1986).<br />
R. C. Bray, D. J. Lowe, C. Capeillere-Blandin and E.<br />
M. Fielden, Biochem. Soc. Trans. 1, 1067 (1973).<br />
D. P. Ballou and G. A. Palmer, Anal. Chem. 46,<br />
1248 (1974).<br />
A. M. P. Goncalves, Prog. Solid St. Chem. 13, 1<br />
(1980).<br />
A. Weiss, Angew. Chem. Internal. Edit. 11, 607<br />
(1972).<br />
E. R. Andrew, W. S. Hinshaw, M. G. Hutehins and<br />
R. O. I. Sjoblom, Mol. Phys. 31, 1479 (1976a).<br />
[22] D. E. Barnaal and I. J. Lowe, /. Chem. Phys. 48,<br />
4614 (1968).<br />
[23] M. Weithase, F. Noack and J. von Schiitz, Z. Phys.<br />
246, 91 (1971).<br />
[24] E. R. Andrew, W. S. Hinshaw, M. G. Hutehins and<br />
R. O. I. Sjoblom, Mol. Phys. 34, 1695 (1977).<br />
[25] R. E. Marsh and J. Donohue, Advanc. Protein Chem.<br />
22, 235 (1967).<br />
[26] P.-G. Jonsson and A. Kvick, Ada Cryst. B28, 1827<br />
(1972).<br />
[27] E. Mayer, /. Appl. Phys. 58, 663 (1985).<br />
[28] P. T. Sargeant and R. Roy, Mater. Res. Bull. 333,<br />
265 (1968).<br />
[29] D. Tumbull, Contemp. Phys. 10, 473 (1969).<br />
[30] N. H. Fletcher, Rep. Prog. Phys. 34, 913 (1971).<br />
[31] D. Uhlmann, /. Non-Cryst. Solids 7, 337 (1972).<br />
[32] P. Briiggeller and E. Mayer, Nature 288, 569 (1980).<br />
[33] D. R. MacFarlane and C. A. Angell, J. Phys. Chem.<br />
88, 759 (1984).<br />
[34] A. Calvelo, "Development in Meat Science," p. 125,<br />
Applied Science Publishers, London, 1981.<br />
[35] H.C.Steinrucken and KAmrhein, EurJ.Biochem.<br />
143, 351 (1984).<br />
[36] K.S.Anderson et a!., Biochemistry 29, 1460 (1990).<br />
[37] K.S.Anderson, J.A.Sikorski & K.A.Johnson,<br />
Biochemistry 27, 7395 (1988).<br />
[38] See for example, Proteins at Low Temperatures (Ed.<br />
O. Fennema), Am. Chem. Soc. Adv. in Chem. Series<br />
No. 180 (1979); RFranks, Biophysics and<br />
Biochemistry at Low Temperatures, Cambridge<br />
University Press (1985).<br />
[39] L. Bachmann & E. Mayer, in Cryotechniques in<br />
Biological Electron Microscopy (Ed. R.A. Steinbrecht<br />
and K. Zierold), Springer-Verlag, Berlin p.3. (1987);<br />
P.Douzou, Cryobiochemistry: An Introduction,<br />
Academic Press (1977).<br />
[40] WJ.Albery and JJl.Knowles, Biochemistry 15, 5631<br />
(1976).<br />
[41] D.P.Raleigh, M.H.Levitt & R.G.Griffin,<br />
Chem.PhysJLett. 146, 71 (1988); D.P.Raleigh,<br />
F.Creuzet, S.K.Das Gupta, M.H.Levitt &<br />
R.G.Griffin, JAm.Chem.Soc. Ill, 4502 (1989);<br />
T.Gullion & J.Schaefer, JMagnReson. 81, 1%<br />
(1989); G.R.Marshall et aL.JAm.Chem.Soc. 112,<br />
963 (1990); S. M. Holl, R. A. McKay, T. Gullion, J.<br />
Schaefer, J. Magn. Reson. 89, 620 (1990); V.Copi6 et<br />
al.. Biochemistry 29, 9176 (1990).
86<br />
Abstract<br />
Coupled methyl groups in dimethyl sulphide<br />
M.R. Johnson, S. Clough, A.J. Horsewill and I.B.I. Tomsah<br />
Bulletin of Magnetic Resonance<br />
Department of Physics, University of Nottingham, Nottingham. NG7 2RD.<br />
Low field methyl tunnelling spectra of dimethyl sulphide<br />
have been measured using field cycling NMR and indicate<br />
the existence of two distinct methyl groups with tunnel.<br />
frequencies of 100kHz and 750kHz. Spectra show a significant<br />
broadening of the Larmor peak at those fields at<br />
which the Larmor and tunnel frequencies are equal. These<br />
are resonances between the rotational and spin dynamics<br />
of the methyl groups. An associated" change in the intensity<br />
and position of the 100kHz sidebands suggests that<br />
methyl group coupling may be responsible for these effects.<br />
Narrow 100kHz sidebands are restored by irradiating<br />
with an external RF field of this frequency which we<br />
suggest is due to decoupling of the rotor pairs. Measurements<br />
on partially deuterated dimethyl suphide show that<br />
the coupling cannot be intra-molecular.<br />
1 Introduction<br />
The methyl group tunnel frequency wt depends on the<br />
height of the potential barrier hindering rotation due to<br />
the molecular environment. Three stationary states are<br />
distributed equally in three potential wells and form an<br />
energy singlet and doublet split by ftwt. Excitations consist<br />
of a superposition of a pair of eigenfunctions to form<br />
a partially localised wavepacket which rotates in a definite<br />
sense at a frequency given by the energy splitting,<br />
ie ±wt or 0. With the aid of external fields the rotation<br />
frequency of a wavepacket can be changed by exerting a<br />
rotational impulse her. The rotation frequency is then determined<br />
by the difference between two of the three values<br />
(2w where i, y and z are the spin states, a or /?, at ,,<br />
proton sites which are defined by the hindering P° ten .*|<br />
[1]. The protons themselves are each equally ditbufr<br />
of
Vol. 14, No. 1-4<br />
between the three sites satisfying proton indistinguishabilitj.<br />
In this represenoation the matrix elements cf the<br />
tunnelling interaction reflect the balance of clockwise and<br />
anti-clockwise rotation of an undriven methyl group.<br />
< xyz\HR\zxy >=< zxy\HR\xyz >= A (2)<br />
A is the overlap integral between states localised in neighbouring<br />
wells of the three-fold hindering potential.<br />
Away from the anti-crossings the methyl group eigenstates<br />
are linear combinations of the localised states and<br />
are the delocalised representations of the Cz symmetry<br />
group,<br />
1 [\xyz> + exp(t2irn/3)\zxy (3)<br />
where n = 0 for an A-state and n = ±1 for Ea and £&<br />
states. Temporarily ignoring the dipole-dipole interaction,<br />
these symmetrised states have energies<br />
(4)<br />
E{\, m) = u>Lm - 2A cos(2;rn/3) (5)<br />
where OIL is the Larmor frequency and m is the spin component<br />
of \xyz > in the direction of the magnetic field.<br />
Mixtures of these eigenstates are partially localised, rotating<br />
wavepackets which have energies resulting from the<br />
differences between the rotational and magnetic energies<br />
of the component states E(Xi,rm) — E(Xj ,mj). Magnetic<br />
and rotational wavepacket energies add when the Larmor<br />
precession and the rotation have the same sense and they<br />
subtract when these senses are opposite.<br />
The exact eigenstates and eigenvalues of the methyl<br />
' group in the magnetic field are calculated numerically by<br />
;:"'diagonalising H. The field dependence of the energy levels<br />
Mis shown in figure 1. The level anti-crossings can be seen<br />
^ . 1.2 and 2.4mT for the 100kHz methl group and 8.8 and<br />
|17.6mT for the .750kHz rotor, where the A and E symmetry<br />
states reflect off each other and exchange symmedue<br />
to the finite non-secular parts of the intra-methyl<br />
1 dipole-dipole interaction.<br />
a energy levels diagram of figure 1 can be transformed<br />
1 diagram of energy differences, or wavepacket ener-<br />
|..as shown in figure 2 from which frequency spectra at<br />
magnetic fields can be predicted by taking horsections.<br />
In this way the level anti-crossings are<br />
manifest themselves as resonant motional broad-<br />
M the Larmor peak. Symmetry mixing at the<br />
sings results in stationary wavepackets with en-<br />
Pjroportional to u>£ evolving into rotating wavepackets<br />
~" T gy Proportional to wfc (x to y in figure 2). This<br />
he ability of the non-secular parts of the dipoleraction<br />
for converting spin angular momentum<br />
•tional angular momentum of the methyl group.<br />
18<br />
A(3/2)<br />
^(-3/2)<br />
A(-3/2)<br />
B(mT)<br />
Figure 1: Energy levels of dimethyl sulphide in a magnetic<br />
field showing resonances between tunnelling motion and<br />
spin dynamics as level anti-crossings at 1.2, 2.4 ,8.8 and<br />
17.6mT<br />
15-; '1:-.<br />
10-<br />
Reld(mT)<br />
5 - • .?• .:•?" ,v*<br />
0 S~=r<br />
200 400 600 800<br />
Frequency(kHz)<br />
1000<br />
Figure 2: Field/frequency plane of dimethyl sulphide<br />
showing resonant broadening of the Larmor peak at the<br />
level anti-crossings, notably at 9mT and 18mT<br />
87
88<br />
3 Low field, field cycling experiments<br />
Measuring magnetisation at low field using a Faraday law<br />
detector results in poor signal to noise. Thus in order to<br />
measure a low field frequency spectrum either a different<br />
type of detector [5] or a different technique such as field<br />
cycling must be employed. Field cycling has been used<br />
in this investigation. Each cycle of the experiment begins<br />
with the preparation of a standard magnetisation by<br />
destroying all magnetisation with a train of 90° pulses followed<br />
by the preparation period of 20 seconds in a field<br />
of 0.6T, The field is then switched rapidly in about 2 seconds<br />
to the chosen magnetic field where the initial magnetisation<br />
is destroyed by spin-lattice relaxation by an<br />
amount proportional to the relaxation, rate T{~ and the<br />
time at low field, 15 seconds. Anomalously rapid relaxation<br />
can be stimulated by an external RF field of scanning<br />
frequency / if this frequency matches the characteristic<br />
wavepacket frequencies in figure 2 and can therefore<br />
excite these wavepackets. The remnant magnetisation is<br />
measured by a single 90° pulse after a rapid field switch<br />
to 0.6T. The cycle is repeated increasing / each time, producing<br />
a frequency spectrum which is a flat plateau in<br />
which holes are drilled where f is equal to the frequencies<br />
predicted in figure 2. These spectra are inverted to give<br />
peaks for presentation.<br />
A slightly more complicated experiment entails the application<br />
of a second RF field of fixed frequency fsU at<br />
low field , which is applied on the same coils as the scanning<br />
field in an alternating sequence of short bursts of 0.5<br />
seconds of each field. This is a stirring experiment which<br />
may affect the intensity of spectral peaks by depopulating<br />
energy levels and enhancing transitions which would otherwise<br />
become suppressed as the populations of the initial<br />
and final states become equal [6].<br />
4 Sample preparation<br />
Dimethyl sulphide was obtained from the Aldrich Chemical<br />
Company and was used in an unsealed sample tube.<br />
The deuterated sample was prepared in the Chemistry Department<br />
by mixing aqueous solutions of CHzS~ No. and<br />
CD2I as outlined in [7]. It was also used in an unsealed<br />
sample tube.<br />
5 Low field frequency spectra of<br />
dimethyl sulphide<br />
A typical low field spectrum of dimethyl sulphide, measured<br />
at 3.5rnT, is shown in figure 3 and it is in excellent<br />
agreement with the corresponding slice through figure 2,<br />
indicated by the broken line. The Larmor peak occurs at<br />
175kHz indicating a true magnetic field of 4.1mT and the<br />
Am = 2 version of this peak occurs at 350kHz. These are<br />
Bulletin of Magnetic Resonance<br />
250 BOO 750 1000<br />
Frequency [kHz]<br />
1250<br />
Figure 3: Frequency spectrum of dimethyl sulphide<br />
mesaured at 3.5mT<br />
labelled 1 and 2 respectively. They are each flanked by a<br />
pair of 100kHz sidebands (labelled 1± and 2±) and also<br />
by a pair of 750kHz sidebands (labelled 1± and 2±). Since<br />
the Larmor frequency is much smaller than the larger tunnel<br />
frequency, the reflected low frequency sidebands give<br />
rise to a 750kHz spectrum which appears as a Am = 0<br />
peak split into five peaks, each separated by the Larmor<br />
frequency.<br />
Frequency spectra of dimethyl sulphide have been measured<br />
at many magnetic fields up to 35mT and a selection<br />
of these are shown in figure 4. The spectra are aligned by<br />
their Larmor peaks and only the frequency range incorporating<br />
the 100kHz sidebands is covered. Away from the<br />
level anti-crossings of the 750kHz rotor, that is below 7mT,<br />
between 10 and 14mT, and above 20mT, the Larmor peak<br />
and sidebands are well defined and reasonably symmetrical.<br />
However in the vicinity of the level anti-crossings,<br />
between 7 and lOmT and between 14 and 20mT, this discrete<br />
structure is lost as the Larmor peak broadens, the<br />
low frequency sideband virtually disappears and the high<br />
frequency sideband is poorly defined. Where a frequency<br />
separation of the Larmor peak and a sideband can be determined<br />
it is apparent that rotation frequency of these<br />
wavepackets is less 100kHz and is sometimes as small a?<br />
70kHz. . 1<br />
Although these effects occur at the level anti-crossing<br />
of the 750kHz rotor, the field range over which they occ^<br />
and the extent of the frequency broadening are too gre^<br />
for them to be explained by dipolar broadening alone. Ful<br />
thermore the the dramatic change in the sideband mtei<br />
sity and frequency is not predicted by the dipole-dipole"<br />
teraction. That it is the 100kHz spectrum which is so pr'<br />
foundly affected by the level anti-crossings of the 750K<br />
rotor suggests that these spectra may be evidence of sp
Vol. 14, No. 1-4 89<br />
Z<br />
o<br />
m zo<<br />
:<br />
111<br />
DC<br />
Q<br />
LU<br />
h-<br />
DC<br />
LU<br />
• \<br />
:<br />
v"V<br />
• ".}'<br />
• ' : I'.<br />
I<br />
I<br />
• I<br />
i '."<br />
I<br />
I<br />
I<br />
I V<br />
coz-1OO co'z ©2+100<br />
22.5<br />
20<br />
• 4'~.~*~ 15<br />
• • !•<br />
'.. ' i.<br />
• r-^ '12.5<br />
Si W9<br />
A 5<br />
• •"?• y '<br />
E<br />
Q<br />
UJ<br />
U_<br />
o<br />
LU<br />
z<br />
CD<br />
<<br />
FREQUENCY/kHz<br />
4: Field dependence of the Larmor peak and the<br />
coupling between the 100kHz and 750kHz rotors.<br />
The 100kHz sidebands can be restored in a stirring experiment<br />
with an external RF field with a stirring frequency<br />
of 100kHz, as shown in the spectra in figure 5. It<br />
appears that stirring at 100kHz drives rotation of methyl<br />
wavepackets at this frequency and therefore prevents coupling<br />
which has been seen to modify this frequency, as<br />
shown in figure 4.<br />
In order to probe the methyl group coupling a partially<br />
deuterated sample of dimethyl sulphide, in which<br />
one methyl group per molecule is fully deuterated, was<br />
investigated. Figure 6 shows a low field spectrum of the<br />
deuterated sample measured at 4mT which is very similar<br />
to the spectrum of the fully protonated sample shown in<br />
figure 3, one difference being the increase in the large tunnel<br />
frequency from 750kHz to 780kHz. Field dependent<br />
spectra of the deuterated sample covering the frequency<br />
range which incorporates the Larmor peak and the 100kHz<br />
sidebands are shown in figure 7 and they too are almost<br />
identical to the spectra from the fully protonated sample<br />
of figure 4. It therefore appears that the coupling of<br />
protonated rotors persists in the deuterated sample, suggesting<br />
that the packing of the molecules in the solid regarding<br />
the relative positions of protonated and deuterated<br />
methyl groups is random. Deuteration reduces the<br />
number of coupled protonated rotors and slightly distorts<br />
the packing of the molecules, decreasing the magnitude<br />
of the hindering potential of the 750kHz rotor which becomes<br />
780kHz rotor. These spectra indicate clearly that<br />
the anti-crossing effects cannot arise from intra-molecular<br />
methyl group coupling since each molecule has a protonated<br />
and a deuterated methyl group. One new feature of<br />
these spectra which may arise from coupling of protonated<br />
and deuterated rotors is a small, sharp, field independent<br />
peak at 200kHz which is indicated by a V in figure 7.<br />
6 Discussion - Coupled rotating<br />
wavepackets<br />
Tunnelling of small molecular rotors like CH3, CH4 and<br />
NHf is generally single particle in character. Few examples<br />
exist,of coupled tunnelling of rotors (see [8] and references<br />
therein) perhaps the most notable example being of<br />
coupled methyl groups in lithium acetate [9]. In both of<br />
these papers the coupling of methyl groups is mechanical,<br />
being propagated by the modulation of the hindering potential<br />
which has the three-fold symmetry of the individual<br />
methyl groups. The coupled states of the system are a<br />
direct product of the symmetrised eigenstates of each rotor.<br />
Lithium acetate has weakly hindered methyl groups,<br />
the tunnelling spectrum of the coupled system has been<br />
observed using neutron scattering, and it is thought that<br />
such a coupling is unlikely to be observable in strongly hindered<br />
methyl groups, although the computational method<br />
in [8] has been extended to consider such systems.<br />
This kind of coupling is in stark contrast to that seen
90<br />
G)z-100<br />
"FREQUENCY/kHz<br />
Figure 5: Stirred spectra of dimethyl sulphide showing '<br />
restoration of the 100kHz sidebands<br />
Bulletin of Magnetic Resonance<br />
O 250 BOO 7S0 1000 1250<br />
Frequency (kHz]<br />
Figure 6: Frequency spectrum of partially deuterated<br />
dimethyl sulphide mesaured at 4mT<br />
in dimethyl sulphide. An analysis in terms of uncoupled<br />
rotors was pursued in previous sections because first, the<br />
coupling only occurs at those magnetic fields corresponding<br />
to level anti-crossings and secondly, the form of the<br />
spectra indicate that the coupled states are not simple<br />
products, with three-fold symmetry, of the eigenstates of<br />
the individual rotors. Furthermore, these methyl groups<br />
are very strongly hindered in comparison with the methyl<br />
groups in lithium acetate.<br />
It is noted from figure 2 that at the level anti-crossings<br />
there is a matching of the frequencies associated with<br />
the rotational and magnetic evolution of the states of<br />
the 100kHz and 750kHz rotors. If the stable states of<br />
the methyl groups at these magnetic fields are rotating<br />
wavepackets then the frequency of evolution of the spins<br />
on each proton site is modulated by the rotation in a way<br />
that depends upon the composition bf the wavepacket.<br />
For example a coherent mixture of A and Ea states with<br />
the same spin quantum number is a partially localised<br />
wavepacket, which precesses clockwise at the tunnel frequency<br />
and consequently modulates the longitudinal component<br />
of the spins at this frequency. A similar mixed<br />
symmetry state, but composed of states with spin quanne<br />
turn numbers differing by unity modulates the transverse<br />
component of the spin states at each proton site at a frequency<br />
equal to the Larmor frequency plus or minus the<br />
tunnel frequency, depending on whether the rotation and<br />
Larmor precession have the same or opposite senses. Thus,<br />
providing the proton sites of the two methyl groups arft<br />
close enough together, the rotational motions may be<br />
pled by the spin dynamics and angular momentum may<br />
transferred between the groups. This results in an imb<br />
ance of clockwise and anti-clockwise rotation, the unic l u
Vol. 14, No. 1-4 91<br />
ide<br />
O<br />
<<br />
CO<br />
LU<br />
<<br />
Hi<br />
DC<br />
Q<br />
LU<br />
DC<br />
LU<br />
,v<br />
J •<br />
J<br />
v/AV \<br />
V<br />
00<br />
C0z-100<br />
/ ; •<br />
i<br />
ll<br />
V /<br />
\ '<br />
1/<br />
v \<br />
v 21<br />
\ 20<br />
v<br />
\<br />
19<br />
18<br />
16<br />
u<br />
10<br />
8<br />
h-<br />
E<br />
Q<br />
_1<br />
UJ<br />
LL<br />
g<br />
FREQUENCY/kHz<br />
7: Field dependence of the Larmor peak and the<br />
z sidebands in partially deuterated dimethyl sul-<br />
tunnel frequency being replaced by a pair of discrete rotation<br />
frequencies. A reduced rotation frequency is clearly<br />
observed in figure 4.<br />
Taken as evidence not only of coupled methyl groups<br />
but also for rotating wavepackets at 4K, these results are<br />
very significant. The wavepackets do not have three-fold<br />
symmetry which throughout the history of methyl dynamics<br />
has wrongly been regarded as an essential prerequisite<br />
for satisfying the indistinguishabilty of the methyl protons<br />
(e.g. [10]). Each basis function \xyz > accomodates proton<br />
indistinguishability since each proton has an equal amplitude<br />
at each proton site and therefore any combinations<br />
of these functions, ranging from delocalised states which<br />
ressemble spin symmetry species to completely localised<br />
functions, satisfy proton indistinguishability requirements.<br />
7 Acknowledgements<br />
The authors are grateful to the B.P. Venture Research<br />
Unit for supporting this work. I.B.I.T. would like to express<br />
his gratitude to the Sudanese government for his<br />
fellowship. We would also like to thank Dr D.K. Knight<br />
of the Chemistry Department for preparing the deuterated<br />
sample.<br />
References<br />
[1] S. Clough, A.J. Horsewill, M.R.Johnson and<br />
I.B.I.Tomsah (1992) submitted to Molec. Phys.<br />
[2] P.J. McDonald, G.J. Barker, S. Clough, R.M. Green<br />
and A.J. Horsewill (1986) Molec. Phys. 57,901<br />
[3] S. Clough, A. Heidemann, A.J. Horsewill, A.J. Lewis<br />
and M.N.J. Paley (1982)J. Phys. C 15,2495<br />
[4] S. Clough, A.J. Horsewill, P.J. McDonald and F.O.<br />
Zelaya (1985)Phys.Rev. Lett. 55,1794<br />
[5] C. Connor, A. Chang and A. Pines (1986)Rev. Sci.<br />
Instrum. 61,1059<br />
[6] M.J. Barlow, S. Clough, P.A. Debenham and A.J.<br />
Horsewill (1992)J. Phys. C 4,4165<br />
[7] L.Pierce and M. Hayashi (1960)J. Chem. Phys. 35,479<br />
[8] W. Hausler and A. Huller (1985) Z. Phys. B 59,177<br />
[9] S. Clough, A. Heidemann, A.J. Horsewill and M.N.J.<br />
Paley (1984) Z. Phys. B 55,1<br />
[10] J.H. Freed (1965)J. Chem. Phys. 43,1710
92 Bulletin of Magnetic Resonance<br />
INTRODUCTION<br />
NMR RELAXATION STUDIES OF MICRODYNAMICS IN<br />
CHLOROALUMINATE MELTS<br />
Pamela A. Shaw, W. Robert Carper, Charles E. Keller<br />
Room temperature molten salts consisting of mixtures of<br />
A1C13 and l-ethyl-3-methylimidazolium chloride (MEIC1),<br />
are of interest as aprotic solvents for studying a wide range<br />
of both organic and inorganic compounds [1-7]. These<br />
chloroaluminate molten salts possess considerable potential<br />
as battery electrolytes and various types of electrochemical<br />
agents [8-10].<br />
The composition of a chloroaluminate melt has a<br />
considerable effect on its physical properties. The<br />
variations in physical properties of the melt are due to a<br />
combination of factors including ion-ion interactions [4],<br />
and Lewis acid-base properties. Chloroaluminate melts<br />
with A1C13 present in excess (mole fraction, N, of A1C13 ><br />
0.5) are termed acidic with A1C14' and A12C17" the<br />
predominant anions.<br />
The use of NMR relaxation methods provides useful<br />
information about the dynamics and structure of various<br />
chemical systems and chloroaluminate systems in<br />
particular. In a previous work[ll], 13 C NMR relaxation<br />
measurements were used to investigate the motion and<br />
interactions of the MEI cation. The results indicate that<br />
A1C14' in a Na + 0.22MEI + 0,7gAlCl4" melt forms a complex by<br />
interacting with the C-2, C-4 and C-5 hydrogens on the<br />
MEI + ring. This investigation was followed by studies<br />
[12,13] in which the Dual Spin Probe method [14]<br />
supported the existence of MEI(AlCl4)n
Vol. 14, No. 1-4 93<br />
The quadrupole coupling constant, QCC, is given by:<br />
QCC = [e 2 Qq/h] (4)<br />
The DSP method has been applied to chloroaluminate<br />
melts[12,13] and has provided evidence that the ring<br />
hydrogens of MEI + interact with the tetrachloroaluminate<br />
anion. The existence of these complexes has been<br />
supported by linear plots of 13 C dipolar relaxation<br />
rates(R, dd ) vs. quadrupolar 27 A1 relaxation rates(R,) that<br />
pass through the origin as predicted by equation (5):<br />
where a = [3TT 2 /10][(2I + 3)/I 2 (2I -<br />
QCC = x-<br />
= R,( 27 Al)/aX 2<br />
(5)<br />
+ (z 2 /3)], and<br />
In this study, the DSP method is applied to melts<br />
containing MEIC1, A1C13 and EtAlCl2. The inclusion of<br />
EtAlCl2 provides a "baseline" as there is a covalent bond<br />
between the ethyl group and aluminum in EtAlCl2. The<br />
existence of covalent bonding(or complexation) between<br />
quadrupolar and dipolar nuclei in a molecule results in a<br />
linear plot of eqn. (5) that passes through the origin. In<br />
the MEICl-EtAlCl2 melts reported herein, we observe a<br />
linear plot of eqn (5) that passes through the origin when<br />
applied to the terminal CH3 carbon in EtAlCl2 and one of<br />
the peaks in the 27 A1 NMR of the melts.<br />
EXPERIMENTAL<br />
Materials<br />
The l-ethyl-3-methylimidazolium chloride (MEIC1) and<br />
. chloroaluminate molten salts were prepared as described<br />
. previously [1]. Ethylaluminum dichloride (EtAlCy was<br />
•^.obtained from Aldrich. All materials were stored under<br />
ft^hydrous helium gas atmosphere in a dry box. All<br />
molten salt preparations and manipulations were performed<br />
• the dry box. Samples were loaded into 5 mm sample<br />
s > capped in the dry box, removed, and sealed<br />
Jiately with a torch.<br />
Measurements<br />
and 27 A1 NMR spectra were recorded on a Varian<br />
300- spectrometer at 75.43 or 78.15 MHz.<br />
measurements were calibrated against<br />
or ethylene glycol and are accurate to within<br />
§£Pulse widths(90°) were typically 8.6 (75.43 MHz)<br />
1^(78.15 MHz) jts. Longitudinal relaxation times<br />
ured by the the inversion-recovery method<br />
(180°-r-90°-T) with T> 10T,. At least 12 delay times(r)<br />
were used and the results fitted to a three parameter<br />
exponential. NOE measurements were made using the<br />
gated decoupler method[18]. It is likely that the error in<br />
the NOE measurements is in the 5-10% range[18].<br />
RESULTS AND DISCUSSION<br />
The ability of both A1C13 and EtAlCl2 to form C2H<br />
dimers[19,20] led us to examine the 27 A1 spectra of: (1)<br />
neat EtAlCl2, (2) mixtures of MEICl-EtAlCl2 and (3)<br />
ternary melts (N = AlCl3/MEICl/EtAlCl2)[21]. The neat<br />
EtAlCl2 27 A1 NMR spectrum contains two peaks [21].<br />
Peak 1 is a broad downfield peak that domi-nates the<br />
spectrum. The second peak (upfield) overlaps peak 1 and<br />
is only a fraction of peak 1 in total peak area. Peak 2<br />
collapses into peak 1 as the temperature is lowered from<br />
60 to 25°C. These two aluminum sites are consistent with<br />
the extent of monomer-dimer formation in liquid<br />
EtAlCl2[21].<br />
The MEICI-EtAlCl2 (N = 0.5/0.5) melt 27 A1 NMR<br />
spectrum also has two peaks. In this case, peak<br />
1 (downfield) is very broad while peak 2 is very sharp, and<br />
has a low peak area. Peak 2 increases slightly in area and<br />
peak 1 broadens as the temperature is lowered from 70 to<br />
0°C. We have previously[21] made the tentative assignments<br />
of EtAlCl3- for peak l(downfield) and EtjAljClj" for<br />
peak 2.<br />
0.25<br />
0.20<br />
0.15<br />
O<br />
a.<br />
O o.io<br />
O CO<br />
0.05<br />
0.00<br />
EtAia<br />
10 20 30 40<br />
27AI R1<br />
Fig. 1. °C Dipolar Rl's vs "AI Rl's(25 to 70°C) for AI<br />
peak 1 (127-131 ppm from A1(H2O)6 3+ ).
94<br />
In this study, we first apply the DSP method to the CH3<br />
carbon in EtAlCl2 and 27 A1 NMR peaks 1 and 2 from<br />
several melt combinations and neat EtAlCl2. Fig. 1<br />
contains the results for 27 A1 peak 1 (downfield) and Fig. 2<br />
contains the results for 27 A1 peak 2. The fact that both<br />
plots are linear and pass through the origin, indicate that:<br />
(1) the DSP method is appropriate for these systems and<br />
(2) the species associated with each peak contains EtAlCl2.<br />
Furthermore, the slopes of these lines can be used to<br />
cn<br />
DIPOL<br />
13C<br />
. u<br />
0 .20<br />
0 .15<br />
0 .10<br />
0 .05<br />
n nn<br />
| EtAlClj /<br />
r<br />
I J .25/.40/.35 /<br />
/<br />
y<br />
• I * /<br />
t 1 A35/.40/.25<br />
i ; /<br />
i/<br />
0 240 480 720 960 1200<br />
27 Al R1<br />
Fig. 2. "C Dipolar Rl's vs 27 A1 Rl's(25 to 70°C) for Al<br />
peak 2 (102.5-103.0 ppm from Al(H2O)6 3+ ).<br />
calculate the relative quadrupole coupling constants for the<br />
EtAlCl2 -containing species in solution. The QCC values<br />
obtained from Fig. l(Al peak 1) are 171, 119, 106 and 93<br />
MHz for the (.5/.5), (.35/.40/.25), (.25/.40/.35) melts and<br />
neat EtAlCl2, respectively. The QCC values obtained from<br />
Fig. 2(A1 peak 2) are 6.9, 20, 11 and 93 MHz for the<br />
(.5/.5), (.35/.40/.25), (.25/.40/.35) melts and neat<br />
EtAlCl2(repeated).<br />
Results of the Dual Spin Probe method (eqn. [5]) applied<br />
to the (.5/.5), (.35/.40/.25) and (.25/.40/.35) melts<br />
indicate interactions between the Al-containing species in<br />
peak 2(102.5-103.0 ppm relative to A1(H2O)6 3+ ) and both<br />
the NCH3 and ethyl terminal CH3 groups of MEI + . Fig.<br />
3 contains the plots for the NCH3 group in each melt and<br />
Fig. 4 contains data for the terminal CH3 on the MEI ethyl<br />
group.<br />
cc<br />
0.50<br />
0.40<br />
o.oo<br />
.5/.5<br />
Bulletin of Magnetic Resonance<br />
• / .35/.40/.<br />
40/.25<br />
0 64 128 192 255 320<br />
27AI R1<br />
Fig. 3. 13 C Dipolar Rl's vs. 27 A1 Rl's(25 - 70 C) for<br />
NCH3 carbon vs Al peak 2(25 - 70°C).<br />
0.55<br />
0.44<br />
T .5/.5<br />
» •<br />
I / .35/.40/.: 40/.25<br />
64 128 192 256 320<br />
27AI R1<br />
Fig. 4. l3 C Dipolar Rl's for ethyl CH3 carbon vs 27 A1<br />
Rl's(25 - 70°C) for Al peak 2.<br />
The QCC's obtained from the slopes in Fig. 3(MEI<br />
NCH3) are 1.7, 2.3 and 4.4 MHz for the (.5/.5),<br />
(.35/.40/.25) and (.25/.40/.35) melts. The QCC's froffl|
Vol. 14, No. 1-4 95<br />
Fig. 4(terminal CH3 on the MEI ethyl group) are 1.6, 6.9<br />
and 1.3 MHz for the (.5/.5), (.35/.40/.25) and<br />
(.25/.40/.35) melts.<br />
Finally, there is no correlation between the ring hydrogen<br />
dipolar Rl's and any of the 27 A1 peak Rl's. This result is<br />
directly opposite to that found in MEIC1-A1C13 systems<br />
[11,12].<br />
CONCLUSIONS<br />
Application of the DSP probe method to these mixed<br />
melt systems indicates a lack of complexation between the<br />
ring hydrogens of MEI + and any of these aluminum<br />
containing species. These and previous results[21] suggest<br />
that the formation of various charged dimers containing<br />
EtAlCl2 takes precedence over the formation of complexes<br />
between EtAlCl3" and the MEI + ring hydrogens. However,<br />
there is evidence of interaction between the various Alcontaining<br />
species and the CH3 groups(NCH3 and terminal<br />
CH3 in the ethyl group) of MEI + in these melts.<br />
ACKNOWLEDGMENTS<br />
This work was partially supported by a National<br />
Research Council and Summer Faculty Research<br />
Associateship to W. R. C. and a summer Graduate<br />
Fellowship to P. A. S.<br />
REFERENCES<br />
[1] J. S. Wilkes, J. A. Levisky, R. A. Wilson and C. L.<br />
, Hussey, Inorg. Chem., 21 1263 (1982).<br />
(2| J. S. Wilkes, J. S. Frye and G. F. Reynolds, Inorg.<br />
.•Chem., 22(1983)3870.<br />
jfI3| A. A. Fannin, L. A. King, J. A. Levisky and J. S.<br />
'"Vilkes.J. Phys. Chem., 55(1984)2609.<br />
| 4 1 A. A. Fannin, D. A. Floreani, L. A. King, J. S.<br />
toders, B. J. Piersma, D. J. Stech, R. L. Vaughn, J. S.<br />
jfllkes and J. L. Williams, J. Phys. Chem., 88 (1984)<br />
514.<br />
M. Dieter, C. J. Dymek, N. E. Heimer, J. W.<br />
and J. s. Wilkes, J. Amer. Chem. Soc, 110<br />
2722.<br />
J - D ymek and J. J. P. Stewart, Inorg. Chem., 28<br />
,1472.<br />
[7] J. A. Boon, J. A. Levisky, J. L. Pflug and J. S.<br />
Wilkes, J. Org. Chem., 51 (1986) 480.<br />
[8] C. J. Dymek, J. L. Williams, D. J. Groeger and J. J.<br />
Auborn, J. Electrochem. Soc, 131 (1989) 2887.<br />
[9] C. J. Dymek and L. A. King, J. Electrochem. Soc.,<br />
132 (1985) 1375.<br />
[10] C. L. Hussey, T. B. Scheffler, J. S. Wilkes and A.<br />
A. Fannin, J. Electrochem. Soc, 133 (1986) 1389.<br />
[11] W. R. Carper, J. L. Pflug, A. M. Elias and J. S.<br />
Wilkes, J. Phys. Chem. 96 (1992) 3828.<br />
[12] W. R. Carper, J. L. Pflug and J. S. Wilkes,<br />
Inorganica Chimica Acta 193 (1992) 201.<br />
[13] W. R. Carper, J. L. Pflug and J. S. Wilkes,<br />
Inorganica Chimica Acta (in press).<br />
[14] J. J. Dechter and U. Henriksson, J. Magn. Res., 48<br />
(1982) 503.<br />
[15] A. Abragam, "Principles of Nuclear Magnetism",<br />
Oxford University Press, Oxford (1961).<br />
[16] K. F. Kuhlmann and D. M. Grant, J. Amer. Chem.<br />
Soc, 90 (1968) 7355.<br />
[17] B. Lindman and S. Forsen, in "NMR Basic Principles<br />
and Progress," P. Diehl, E. Fluck and R. Kosfeld,<br />
Editors, Vol. 12, p. 22, Springer-Verlag, New York<br />
(1976).<br />
[18] D. Neuhaus and M. Williamson, "The Nuclear<br />
Overhauser Effect in Structural and Conformational<br />
Analysis", VCH Publishers, New York (1989).<br />
[19] J. Weidlein, J. Organomet. Chem., 27(1969)213.<br />
[20] B. Gilbert, Y. Chauvin and I. Guibard, Vib.<br />
Spectros., 1 (1991)299.<br />
[21] W. R. Carper, C. E. Keller, P. A. Shaw, M. P. and<br />
J. S. Wilkes, in "Eighth International Symposium on<br />
Molten Salts", Electrochem. Soc, New York (in press).
96<br />
Structure and Dynamics of a<br />
Membrane Bound<br />
Polypeptide<br />
T. A. Cross, R.R.Ketchem, W. Hu, K.-C. Lee,<br />
N.D.Lazo & C.L. North<br />
Department of Chemistry &<br />
Institute of Molecular Biophysics<br />
Florida State University, Tallahassee, FL 32306-3006<br />
INTRODUCTION:<br />
Orientational constraints can be used to build-up<br />
three dimensional structures of biological macromolecules<br />
in much the same way that distance constraints are used<br />
today. In an anisotropic environment where molecular<br />
motions do not average NMR signals to their isotropic<br />
average the resonant frequencies become orientation<br />
dependent. Consequently, chemical shift frequencies, dipolar<br />
couplings, and quadrupolar interactions are all dependent on<br />
the specific orientation of the interaction tensor with respect<br />
to the magnetic field. In unoriented samples this results in<br />
broad spectral lineshapes, which arc useful in their own<br />
right, but if the samples are aligned so that all molecules<br />
have the same orientation with respect to the magnetic field,<br />
then sharp line spectra can be obtained that reflect the<br />
orientation dependence of the spin interactions.<br />
to interpret these frequencies for orientational<br />
constraints two pieces of information are critical. The first<br />
is a refined knowledge of the static tensor element<br />
magnitudes and orientation with respect to the molecular<br />
frame. The tensor element magnitudes represent the unique<br />
frequencies of the orthogonal axes of the interaction<br />
ellipsoid. In other words, the magnitudes represent the<br />
resonant frequency when the individual axes are aligned with<br />
the magnetic field. These values can, in many cases, be<br />
readily obtained from spectra of unoriented samples. The<br />
second critical characterization is a knowledge of the<br />
molecular motions that result in averaging both the tensor<br />
element magnitudes and orientation. Consequently, a<br />
detailed picture of dynamics is needed to characterize the axis<br />
about which motions are occurring, the amplitude of the<br />
motions and whether the motions are continuous or<br />
discontinuous, such as a flip of an aromatic ring. In fact, it<br />
Bulletin of Magnetic Resonance<br />
is this separation of dynamics and structure that represents<br />
the most difficult challenge for the biological solid state<br />
NMR spectroscopist. By using low temperature (120 K)<br />
experiments, torsional motions are essentially eliminated<br />
except in specific cases such as methyl group three site<br />
jumps. Once both the static tensors and motional averaging<br />
of the tensors are characterized it is possible to achieve very<br />
high resolution quantitative constraints. Here we present<br />
both an overview of the structural constraints and the<br />
dynamic characterizations achieved for the channel forming<br />
polypeptide, gramicidin A in a fully hydrated lipid bilayer.<br />
Gramicidin A is a hydrophobic polypeptide of 15<br />
amino acid residues with both end groups blocked so that<br />
there are no formal charges. As a dimer it forms a helical<br />
channel with a 4A pore that accommodates a single file of<br />
water molecules and cations that are stripped of all but two<br />
waters in the primary hydration sphere. The peptide linkages<br />
that line the channel are thought to help solvate the cations<br />
during transport by rotating the carbonyl oxygens toward<br />
the channel axis. While crystal structures of gramicidin A<br />
have been achieved in organic solvents (Langs, 1988;<br />
Wallace & Ravikumar, 1988; Langs et al., 1991) no crystal<br />
structure of the channel conformation has been achieved.<br />
However, much is known about the channel conformation.<br />
The backbone folding motif is a (5-sheet type of structure<br />
that has been wound into a helix (Urry, 1971). This is<br />
possible because the amino acids of gramicidin alternate in<br />
stereochemistry between D and L configurations. The helix<br />
sense has been determined from the orientational constraints<br />
of the 15 N amide sites (Nicholson & Cross, 1989).<br />
Recently, the first backbone torsion angles for the channel<br />
conformation have been determined from orientational<br />
constraints (Teng et al., 1991). Complementing this work
Vol. 14, No. 1-4 97<br />
f<br />
13 C Chemical Shift<br />
15 N Chemical Shift<br />
15 N- 13 C Dipolar<br />
^NMH Dipolar<br />
14 N- 13 C Dipolar<br />
15 N- 15 N Dipolar<br />
13 C- 13 C Dipolar<br />
2 H Quadrupolar<br />
Figure 1: A model of the gramicidin channel dimer showing the interaction tensors that have been studied. The model<br />
structure is that of Lomize et al., 1992. Each tensor is represented by a orthogonal set of three unit vectors that have<br />
been oriented correctly for the specific site of interest. All backbone and tryptophan indole I5 N chemical shift and I5 N-<br />
! H dipolar tensors have been studied with the exception of the ethanolamine blocking group. About half of the 15 N-<br />
13 Cj dipolar interactions in the backbone have been studied, four of the backbone carbonyl l ^C chemical shift tensors<br />
and in so doing the 14 N electric field gradient tensor has been characterized. Four sidechains have been deuterated and the<br />
quadrupole splittings obtained. 15 N spin diffusion has been observed between selectively labeled sites. To achieve this<br />
data approximately 50 separate syntheses of gramicidin A have been performed (Fields et al., 1989).
98<br />
are solution NMR studies of gramicidin in SDS micelles<br />
that have also shown the same backbone folding motif<br />
(Arseniev & Barsukov, 1986; Lomize et al., 1992).<br />
RESULTS & DISCUSSION:<br />
Fig. 1 illustrates the large number of nuclear spin<br />
interactions that have been studied in gramicidin. For each<br />
interaction the tensor represented by an orthogonal set of<br />
unit vectors is placed on each atomic site that has been<br />
studied. More than 40 different isotopically labeled (Fields<br />
et al., 1989) gramicidins have been synthesized and more<br />
than 100 different structural constraints have been<br />
developed. The structural task is one of determining the<br />
torsion angles. If it is assumed that the peptide linkages are<br />
planar then there are two backbone torsion angles for each<br />
amino acid residue, one for valine sidechains and two each<br />
for leucine and tryptophan sidechains. There are two<br />
additional torsion angles in the ethanolamine blocking<br />
Val<br />
Ala<br />
Leu,<br />
End View<br />
Side View<br />
Figure 2: A partial structure for the gramicidin channel<br />
dimer, experimentally derived. The N-terminal four peptide<br />
planes have been oriented with respect to the bilayer normal<br />
through a combination of ^N-'H, 15 N- 13 Ci dipolar and<br />
15 N chemical shift interactions in uniformly aligned bilayer<br />
samples. The experimentally verified symmetry of the two<br />
monomers has permitted the docking of these two partial<br />
structures to form a turn of the helix that supports previous<br />
models with 6.3 residues per turn of the channel helix. The<br />
amino acid residues have been identified for one of the<br />
monomers and the amide protons and oxygens labeled.<br />
Bulletin of Magnetic Resonance<br />
group of the carboxy terminus. Therefore, the structural<br />
problem comes down to finding the solution for 52 torsion<br />
angles.<br />
Fig. 2 shows the N-terminal / N-terminal<br />
backbone junction for the channel as determined by<br />
orientational constraints. For each peptide linkage plane the<br />
* S N - !Hand 15 N - 13 Ci dipolar interactions, which have<br />
their unique static tensor elements directed along the<br />
internuclear vectors, were obtained. These two vectors define<br />
the orientation of the plane with some ambiguity that is<br />
minimized by an interpretation of the 15 N chemical shift for<br />
each plane. The orientation of the chemical shift tensor<br />
elements with respect to the molecular frame has been<br />
determined for each of these i 5 N sites (Teng & Cross, 1989;<br />
W. Mai, W. Hu, C. Wang and T.A. Cross- unpublished<br />
results).<br />
This structure clearly shows the right-handed<br />
helical sense and the ^-helical class of torsion angles, in<br />
that the orientation of adjacent peptide planes alternates<br />
between parallel and antiparallel with respect to the channel<br />
axis. Furthermore, like many of the computationally refined<br />
structures (e.g. Roux and Karplus, 1988; Chiu et al., 1991)<br />
and unlike the original structural model (Urry, 1971) many<br />
of the carbonyl oxygens are rotated in toward the channel<br />
axis. The prime exception to this observation is the formyl<br />
oxygen of the amino terminal blocking group. Here we<br />
suspect that this is an artifact of the assumption that this<br />
peptide linkage is planar. Preliminary evidence suggests that<br />
the © torsion angle for this plane is far from 180°, the<br />
value for normal trans-planar linkages.<br />
Fig. 3 shows that we have also studied the<br />
sidechains of gramicidin A. In (A) the chemical shift powder<br />
pattern of a dry powder sample of 15 N indole labeled<br />
tryptophan is shown. The chemical shift tensor elements<br />
agree fairly well with those from the dry powder sample of<br />
gramicidin shown in (B). However the tensor elements for<br />
the fast frozen hydrated lipid bilayer preparation of<br />
gramicidin is very different. Both o"22 and a33 tensor<br />
elements differ by 10 ppm from the dry samples. This<br />
sample was frozen in liquid propane to avoid distortions in<br />
the lipid bilayer which occur when a bilayer preparation is<br />
slowly cooled through its phase transition temperature. The<br />
sample was then transferred to liquid nitrogen and from there<br />
into our low temperature NMR probe. It is unlikely that<br />
the former two samples have the nitrogen bound hydrogen<br />
involved in a hydrogen bond. However, there is some<br />
electrophysiological evidence that these hydrogens for each<br />
of the tryptophan rings in the gramicidin channel state are<br />
hydrogen bonded to the carbonyl oxygens of the ester<br />
linked lipids. Consequently, it may be that the substantial<br />
changes seen in the chemical shift tensor element<br />
magnitudes reflect hydrogen bonding of the indole group.<br />
Further indirect evidence, given below, for hydrogen<br />
bonding comes from the orientation of the indole N-H with
Vol. 14, No. 1-4<br />
respect to the bilayer surface.<br />
200 100<br />
ppm<br />
Figure 3: Powder pattern spectra of 1 5 N labeled indole. The<br />
spectra were obtained at 40 MHz for 15 Nona spectrometer<br />
that has been home built around a Ghemagnetics data<br />
acquisition system and an Oxford 400/89 superconducting<br />
magnet. A] Dry powder sample of the amino acid,<br />
tryptophan obtained at room temperature. Spectral<br />
simulation yields tensor elements: (?i i = 35, o~22 = 104, and<br />
C33 = 158 ppm. B] Dry powder of 15 N-Trp9 gramicidin A:<br />
On = 36, 022 = 106, and a33 = 161 ppm. C] Fast frozen<br />
sample of 1S N-Trp9 gramicidin A in fully hydrated lipid<br />
bilayers: a^ = 36, G22 = 116, and a33 = 171 ppm.<br />
The 2 H quadrupolar spectrum (Fig. 4) of d5-Trpj r<br />
gA provides an example of the spectroscopic data that has<br />
been used for structural constraints. The arrows point to a<br />
linewidth of 1.4 kHz that represents an uncertainty for the<br />
0<br />
orientation of ±0.2°. Not only does this spectrum describe a<br />
single conformation for this tryptophan ring, but it also<br />
dictates that the orientation for all of the gramicidin<br />
molecules in the sample is remarkably uniform (Moll &<br />
Cross, 1990). For this sample there are five deuterons on<br />
the tryptophan ring and five quadrupole splittings are<br />
observed. The analysis of this data combined with the 15 N<br />
chemical shift and 15 N-iH dipolar interaction for the indole<br />
nitrogen (W. Hu, K.-C. Lee and T.A. Cross - unpublished<br />
results) yields two possible orientations for this ring. Each<br />
of these structures has the same orientation with respect to<br />
the channel axis and the orientation is similar to that shown<br />
in Fig. 1 for Trpn, where the N-H points towards the<br />
bilayer surface. In recent computational (Meulendijks et al.,<br />
1989) and electrophysiological (O'Connell et al., 1990)<br />
studies it has been suggested that the indoles of gramicidin<br />
A are hydrogen bonded to the carbonyl oxygens of the esterlinked<br />
lipids. These results indicate that such hydrogen<br />
bonding may be present and this may be one of the prime<br />
reasons why this conformation is present in lipid bilayers<br />
rather than the double-helical structures that dominate<br />
organic solutions (Veatch and Blout, 1974; Zhang et al.,<br />
1992). Furthermore, the tryptophan dipole moment is<br />
oriented primarily along the channel axis, rather than radial<br />
to this axis. This has significant implications for cation<br />
transport by gramicidin A.<br />
Figure 4: 2 H NMR spectrum of ds-Trpj ] gramicidin A in an<br />
oriented lipid bilayer preparation. The arrows indicate a<br />
resonance linewidth of 1.4 kHz. The assignment of two of<br />
these quadrupole splittings is clear from a knowledge of<br />
other spin interactions in this ring system: £3 = 19 2; and<br />
T|2 = 9 9 kHz.<br />
Detailed dynamics have come from a combination<br />
of lineshape simulation and relaxation studies. Below the<br />
gel to liquid crystalline phase transition temperature of the<br />
bilayers all large amplitude motions cease. Both the global<br />
motion about the bilayer normal and the local backbone<br />
-80<br />
99
100<br />
motions become slower than the chemical shift frequency<br />
scale (Nicholson et al., 1989; 1991). This is because the<br />
backbone motions are dependent upon motion of the C a-Cp<br />
axis and hence, if the sidechain conformations are frozen in<br />
the gel phase the backbone motions will be greatly<br />
impeded. When oriented samples are lowered through the<br />
phase transition temperature the broadened resonance<br />
represents a considerable orientational dispersion. Instead of<br />
the backbone having a single conformation in this static<br />
environment a range of conformational substates has been<br />
trapped (Frauenfelder et al., 1988; Nicholson et al., 1989).<br />
This set of substates represents the range of orientations<br />
over which local motions occur above the phase transition<br />
temperature. From the observed lineshape it has been<br />
possible to determine the orientation of the axis about<br />
which the local backbone motions occur with respect to the<br />
magnetic field (Nicholson et al., 1991). This axis has been<br />
shown to be coincident with the Ca-Ca axis for each<br />
peptide linkage. Since the number of conformational<br />
substates is greater than three, the local motion can be<br />
modeled as a diffusion within a gaussian well. Furthermore,<br />
the amplitude of the motion could be estimated.<br />
Once such a motional model has been<br />
experimentally developed for a specific backbone site in the<br />
gramicidin channel conformation it is possible to interpret<br />
relaxation data for these same i*N sites. Tj relaxation times<br />
were obtained from oriented samples of fully hydrated lipid<br />
bilayers. To fix the frequency of the local motions it has<br />
been necessary to measure the relaxation times at two<br />
different field strengths. In Fig. 5 an analysis of such data is<br />
shown. The global correlation time (tp) and the local<br />
motion correlation time (tj) are variables as well as the<br />
amplitude of the local motion. However, we have a<br />
previous estimate of the motional amplitude (±15°,<br />
Nicholson et al., 1991) as well as the global correlation<br />
time (200 ns, Seelig and Macdonald, 1987). Consequently,<br />
it is possible to achieve a unique solution for the local<br />
motion correlation time (10 ns). A similar correlation time<br />
has been reported for an even numbered site (Leu4) in the<br />
gramicidin channel (North and Cross, 1992)<br />
This is a remarkably slow correlation time for a<br />
motion of a molecular group with nominally such a small<br />
molecular weight. Roux and Karplus (1988) have analyzed<br />
the normal modes in the gramicidin channel and concluded<br />
that fluctuations occur with frequencies of 4.6 to 20 cm'<br />
corresponding to harmonic oscillator periods of 2 to 8 ps,<br />
approximately, a factor of lO^-lO 4 slower than the<br />
determination reported here. Similarly, estimates from<br />
molecular dynamics indicate frequencies for these local<br />
motions that are in a similar range to those of the normal<br />
mode analysis.<br />
The experimental evidence presented here is not<br />
A.<br />
-7.0<br />
Log %. -9.0<br />
B.<br />
Log<br />
c.<br />
-11.0<br />
-7.U<br />
T i " 9 -°<br />
-11.0<br />
-7.0<br />
LogTj -9.0<br />
-11.0<br />
Bulletin of Magnetic Resonance<br />
i!! 1<br />
j||<br />
iii<br />
•>t<br />
ii!<br />
•IJJX,<br />
-7.0 -6.0<br />
.f |||1|lW<br />
iii<br />
HI<br />
Hi<br />
Hi i<br />
IP' i<br />
lit} I<br />
a<br />
° 4.7 T<br />
• 9.4 T<br />
' lll|ltI| ! ll "3J l lJJ;;<br />
° 4.7 T<br />
• 9.4 T<br />
• 1 ' • • r— 1 1 1 , . 1<br />
-7.0 -6.0<br />
iii*<br />
ill<br />
f iii<br />
111<br />
Hi<br />
ill<br />
-7.0 -6.0<br />
i<br />
° 4.7 T<br />
• 9.4 T<br />
Figure 5: T, relaxation times that were obtained from 15 N-<br />
Gly2 gramicidin A in oriented lipid bilayers at 20 and 40<br />
MHz have been analyzed to achieve a solution for local and<br />
global correlation times as well as the local motional<br />
amplitude. A] calculated for 12° rms deviation; B] 15°; and<br />
C] 18°. Because of previous estimates for the amplitude and<br />
global correlation time, it is possible to determine the local<br />
correlation time as 10 ns with an rms amplitude of 15°.<br />
unique for slow motions in polypeptide backbones, in fact<br />
most of the experimental evidence suggests much slower<br />
motions than the computational techniques. Analysis of Tj<br />
and NOEs from the backbone of the filamentous virus fd<br />
suggested correlation times of 1 ns (Cross and Opella,<br />
1982). A similar timescale has been reported for collagen
Vol. 14, No. 1-4 101<br />
(Sarkar et al., 1985). More recently, Cole and Torchia<br />
(1991) have reported local motional frequencies in the<br />
backbone of crystalline staphyloccal nuclease in the ns or<br />
near-ns timescale.<br />
For the gramicidin channel there appear to be two<br />
primary reasons for the discrepancy between the<br />
computational and experimental frequencies. First, it maybe<br />
as suggested by Venkatachalam and Urry (1984) that the<br />
local motions are correlated along the polypeptide backbone.<br />
Computational studies have argued that the extent of such<br />
correlations are limited to nearest neighbors. The second<br />
reason is that the lipid environment may damp the backbone<br />
motions severely. Evidence has already been presented that<br />
the lipid environment can reduce the motional frequencies<br />
below the kHz range when the lipids are in the gel phase.<br />
Above the phase transition temperature it is well known<br />
that the lipid environment damps global motional<br />
frequencies. For instance, the 200 ns global correlation time<br />
for gramicidin (1880 daltons) would be consistent with a<br />
protein of 50,000 daltons or greater in aqueous solution.<br />
Furthermore, evidence of specific gramicidin - lipid<br />
interactions described here suggests an additional mechanism<br />
for damping by the lipid environment.<br />
The transit time for cations to move between<br />
dipeptide carbonyl sites in the polypeptide backbone can be<br />
estimated from kinetic measurements (Anderson, 1983) and<br />
analyses of the energetics (Roux and Karplus, 1991) to be<br />
in the range of 10 ns. Consequently, it is now likely that<br />
there is a correlation between kinetics and the local<br />
dynamics of the polypeptide backbone. This unique<br />
correlation between structure, dynamics and function<br />
emphasizes the importance of pursuing such detailed studies<br />
of polypeptides and proteins in functional native-like<br />
environments.<br />
ACKNOWLEDGEMENTS:<br />
We are indebted to the staff of the FSU NMR<br />
facility, Joseph Vaughn, Richard Rosanske, and Thomas<br />
Gedris for their skillful maintenance, modification and<br />
service of the NMR spectrometers. This effort has been<br />
supported in large part through grant #AI-23007 from the<br />
National Institutes of Health. TAC also gratefully<br />
acknowledges support of the Alfred P. Sloan Foundation for<br />
a Research Fellowship.<br />
REFERENCES:<br />
Anderson, O. S. (1983) Biophys. J. 41:119.<br />
Arseniev, A.S. and Barsukov, V.F. (1986) in Chemistry of<br />
Peptides and Proteins, Vol. 3; W. Voelter.E. Bayer,<br />
Y.A. Ovchinnikov, V.T. Ivanov, Eds. Walter de Gryter &<br />
Co. Berlin, pgs. 127-158.<br />
Chiu, S.W., Nicholson, L.K., Brenneman, M.T., Teng, Q.,<br />
Subramaniam, S., McCammon, J.A., Cross, T.A.,<br />
Jakobsson, E. (1991) Biophys. J. 60:974-978.<br />
Cole, H.B.R. and Torchia, D.A. (1991) Chem. Phys.<br />
158:271.<br />
Cross, T.A. and Opella, S.J. (1982) J. Mol. Biol. 159:543-<br />
549.<br />
Fields, C.G., Fields, G.B., Noble, R.L. and Cross, T.A.<br />
(1989) Int. J. Peptide Protein Res. 33:298-303.<br />
Frauenfelder, H., Parak, F. and Young, R.D. (1988) Annu.<br />
Rev. Biophys. Biophys. Chem. 17:451-479.<br />
Langs, D.A. (1988) Science 241:188-191.<br />
Langs, D.A., Smith, G.D., Courseille, C, Precigoux, G.<br />
and Hospital, M. (1991) Proc. Natl. Acad. Sci. USA<br />
88:5345-5349.<br />
Lomize, A.L., Orechov, V. Yu. and Arseniev, A.S. (1992)<br />
Bioorgan. Khimia 18:182-200.<br />
Meulendijks, G.H.W.M., Sonderkamp, T, Dubois, J.E.,<br />
Nielen, R.J., Kremers, J.A. and Buck, H.M. (1989)<br />
Biochim. Biophys. Acta 979:321-330.<br />
Moll, F. Ill, and Cross, T.A. (1990) Biophys. J. 57:351-<br />
362.<br />
Nicholson, L.K. and Cross, T.A. (1989) Biochemistry<br />
28:9379-9385.<br />
Nicholson, L.K. LoGrasso, P.V. and Cross, T.A. (1989) J.<br />
Am. Chem. Soc. 111:400-401.<br />
Nicholson, L.K., Teng, Q. and Cross, T.A. (1991) J. Mol.<br />
Biol. 218:621-637.<br />
North, C.,L. and Cross, T.A. (1992) J. Magn. Res. - in<br />
press.<br />
O'Connell, A. M., Koeppe, R. E. II and Anderson, O. S.<br />
(1990) Science 250:1256-1258.<br />
Roux, B. and Karplus, M. (1988) Biophys. J. 53:297-309.<br />
Roux, B. and Karplus, M. (1991) J. Phys. Chem. 95:4856.<br />
Sarkar, S.K., Sullivan, C.E. and Torchia, D.A. (1985)<br />
Biochemistry 24:2348-2354.<br />
Seelig, J. and Macdonald, P.M. (1987) Ace. Chem. Res.<br />
20:221-228.<br />
Teng, Q. and Cross, T.A. (1989) J. Magn. Res. 85:439-<br />
447.<br />
Teng, Q., Nicholson, L.K. and Cross, T.A. (1991) J. Mol.<br />
Biol. 218:607-619.<br />
Urry, D.W. (1971) Proc. Natl. Acad. Sci. USA 68:672-676.<br />
Veatch, W.R. and Blout, E.R. (1974) Biochemistry<br />
13:5257-5264.<br />
Venkatachalam, CM. and Urry, D.W. (1984) J. Comp.<br />
Chem. 5:64.<br />
Wallace, B.A. and Ravikumar, K. (1988) Science 241:182-<br />
187.<br />
Zhang, Z., Pascal, S. and Cross, T.A. (1992) Biochemistry<br />
- in press.
102 Bulletin of Magnetic Resonance<br />
The Role of Metal Ions in<br />
Processes of Conformational<br />
Selection during Ligand-<br />
Macromolecule Interactions<br />
E.Gaggelli, N.Gaggelli, G.Valensin<br />
Department of Chemistry, University of Siena<br />
Pian dei Mantellini 44<br />
Siena 53100, Italy<br />
and<br />
A.Maccotta<br />
Department of Chemistry, University of Basilicata<br />
Via N.Sauro 85<br />
Potenza 85100, Italy<br />
1 Introduction<br />
The interaction with macromolecules<br />
plays a major role in<br />
eliciting the biochemical activity<br />
of relatively small flexible<br />
ligands. Several processes are<br />
involved in such interaction<br />
such that the key and hole<br />
assumption may often fail. One<br />
of the processes not carefully<br />
considered so far is that of<br />
conformational selection, in<br />
which the macromolecule<br />
stabilizes the conformation of<br />
the ligand at the bound site<br />
after a selection among several<br />
conformational arrangements.<br />
The conformation at the bound<br />
state may or may be not connected<br />
to some low-energy<br />
conformation or to the conformation<br />
stabilized at the solid<br />
state. The comprehension of<br />
this process is expected to<br />
provide a valuable aid in<br />
rationale drug design where<br />
synthesized molecules are<br />
sought yielding the same or<br />
even more specific responses<br />
than the natural ligands.<br />
Investigation of this pro-
Vol. 14, No. 1-4 103<br />
cess requires to delineate the<br />
change in conformation when<br />
going from the free to the<br />
bound state and, from this<br />
point of view, NMR is the<br />
technique of choice.<br />
Here we present evidence<br />
that NMR allows to<br />
detect and delineate the<br />
change in conformation experienced<br />
by a flexible ligand,<br />
the dipeptide carnosine (p -<br />
alanyl-L-histidine), when it<br />
binds to the serum protein<br />
albumin. We show also that<br />
the process of conformational<br />
selection is favoured by the<br />
presence of divalent metal<br />
ions, such as Ca(II) and Cu(II),<br />
that stabilize, in the metal<br />
complex, a conformation of the<br />
ligand very close to that assumed<br />
in the bound state.<br />
2 NMR Parameters<br />
The preferred conformation<br />
assumed by the ligand in its<br />
free state in solution can be<br />
easily delineated by measuring<br />
dipolar interaction energies<br />
between pairs of homo- or<br />
hetero-nuclear spin. Since such<br />
interaction terms are functions<br />
of tc/r 6 , distances can be calculated<br />
if the reorientational<br />
dynamics can be characterized,<br />
even in some approximate<br />
way. This last purpose can be<br />
accomplished by measuring<br />
and interpreting the 13 C-NMR<br />
spin-lattice relaxation rates,<br />
that are, in general, determined<br />
by the one bond (r = 1.09<br />
A [1]) 13C-1H dipole-dipole interaction.<br />
Once the motional correlation<br />
time(s) is (are) determined,<br />
relevant geometric<br />
features can be obtained by<br />
one or more of the following<br />
experiments:<br />
a) evaluation of the iH-pH}<br />
n.O.e. if spectral resolution<br />
is not limited and crosscorrelation<br />
effects can be<br />
neglected;<br />
b) measurement of singleand<br />
double-selective l H-<br />
NMR spin-lattice relaxation<br />
rates of selected proton<br />
pairs; these provide a<br />
means of calculating absolute<br />
values of pairwise<br />
dipolar cross-relaxation<br />
terms, Ojj [2-4];<br />
c) measurement of the * 3 C -<br />
pH} n.O.e. upon selective<br />
presaturation of resolved<br />
proton resonances [5];<br />
d) evaluation of relative<br />
values of cross-relaxation<br />
terms from intensities of<br />
cross peaks in 2D NOESY<br />
maps [6].<br />
All these methods are<br />
very efficient in providing the<br />
desired information on the<br />
preferred conformation in<br />
solution of any 'NMR visible'<br />
ligand and, eventually, the<br />
change in conformation caused<br />
by the presence of metal ions.<br />
In this last case, if the metal is<br />
paramagnetic, a great piece of<br />
structural information is<br />
gained by investigating the<br />
paramagnetic effects on nu-
11 ;<br />
104 Bulletin of Magnetic Resonance<br />
clear relaxation rates and chemical<br />
shifts [7,8] or on 2D<br />
spectra [9,10].<br />
In order to investigate<br />
the eventual change in the<br />
conformation of the ligand<br />
when it binds to a macromolecule<br />
either in the presence or<br />
in the absence of metal ions,<br />
the previously outlined NMR<br />
methods, at least not all of<br />
them, are not as efficient as in<br />
the free state. The concentration<br />
of the macromolecule is a<br />
limiting factor, since it must be<br />
kept quite small if spectral<br />
distortion is to be avoided. As<br />
a consequence exchange of the<br />
ligand between the free (bulk)<br />
and the bound state must be<br />
taken into consideration, yielding:<br />
Pobs = xfPf + xbPb<br />
where P is any observed NMR<br />
parameter, f and b refer to the<br />
free and bound states and the<br />
x's are molar fractions. It<br />
follows that the change in P<br />
from the free to the bound<br />
state is expressed by:<br />
AP = xbPb<br />
where Xb is usually of the<br />
order 0.01-0.1. The consequence<br />
is that spin-lattice<br />
relaxation rates and chemical<br />
shifts are no longer suitable<br />
parameters for delineation of<br />
geometric features of the<br />
bound state, unless a paramagnetic<br />
centre, either intrinsic<br />
or extrinsic, is present. One is<br />
then left with measurements<br />
of ID or 2D transferred n.O.e.<br />
or, which we prefer, of singleand<br />
double-selective proton<br />
spin-lattice relaxation rates. In<br />
fact, in absence of spectral<br />
distortions, the same measurements<br />
can be easily accomplished<br />
for the free ligand as<br />
well as in the system where<br />
the ligand is exchanging<br />
between the two states, and<br />
absolute values of the crossrelaxation<br />
rate can be separately<br />
obtained for the free and<br />
the bound ligand.<br />
3 Free Carnosine<br />
The relevant features of<br />
carnosine in water solution can<br />
be summarized as follows:<br />
a) reorientational dynamics<br />
can be interpreted in<br />
terms of a principal correlation<br />
time (Tc = 58 ps)<br />
describing reorientation<br />
around a molecular axis<br />
passing through the imidazole<br />
ring, coupled with<br />
segmental motion of the<br />
amino-terminal moiety<br />
and librational motion of<br />
the ring (Tg = 10 ps);<br />
b) predominance of the g~<br />
rotamer around the C6-C7<br />
bond (CH2-CH segment of<br />
the histidyl residue);<br />
c) folding of the fi-alanyl<br />
moiety towards the imidazole<br />
ring.
Vol. 14, No. 1-4 105<br />
4 Calcium Complex<br />
Calcium forms two complex<br />
speies with carnosine in solution:<br />
a 1:1 complex where the<br />
carbonyl and carboxyl oxygens<br />
are the metal binding atoms<br />
and a dimeric complex where<br />
the two carbonyl and one carboxyl<br />
oxygens and the imidazole<br />
nitrogen are the four<br />
coordinated atoms. The overall<br />
dissociaton constant of the<br />
complexes is Kd = 0.04 mol<br />
dm"3. In the monomer complex<br />
the dipeptide retains the<br />
conformation detected by NMR<br />
as the 'preferred' one in the<br />
free state in solution. In the<br />
dimer species extensive intermolecular<br />
interactions are<br />
favoured and the conformation<br />
of the dipeptide is less folded<br />
than in the free state or in the<br />
1:1 complex.<br />
It is concluded that calcium<br />
ions stabilize a particular<br />
geometric arrangement that is<br />
itself representing the 'preferred'<br />
conformation in solution,<br />
as it raises from motional averaging<br />
of all the particular<br />
conformations assumed by the<br />
flexible peptide.<br />
5 Interaction with HSA<br />
The interaction of carnosine<br />
with human serum albumin<br />
(HSA) can be detected and<br />
delineated by measuring<br />
selective and double-selective<br />
proton spin-lattice relaxation<br />
rates of carnosine protons in<br />
the presence of low molar<br />
fractions of the protein [11,12].<br />
Detection of binding is allowed<br />
by appreciable relaxation rate<br />
enhancements of selective<br />
relaxation rates of the imidazole<br />
protons, as well as of the<br />
His Ha:<br />
AR sel = pbRb sel<br />
Even at very low fractions of<br />
bound carnosine the selective<br />
relaxation rate at the bound<br />
site is so fast that the observed<br />
rate undergoes enhancements<br />
as high as 50-100 %. The effect<br />
allows also a titration of the<br />
binding process, yielding an<br />
apparent dissociation constant<br />
of 2.5xlO- 4 mol dm- 3 .<br />
The observed enhancements<br />
are consistent not only<br />
with a very tight binding of<br />
the whole peptide molecule to<br />
the protein but also with<br />
occurrence of dipolar interaction<br />
between ligand* and protein<br />
protons, although there is<br />
no possibility of obtaining<br />
quantitative estimations of<br />
such interactions.<br />
More information can be<br />
obtained by measuring the<br />
double-selective proton spinlattice<br />
relaxation rates within<br />
the His Ha-Hpi-H(32 moiety. As<br />
in the free state in solution,<br />
such measurements yield the<br />
dipolar cross-relaxation rate<br />
between the excited protons,<br />
e.g.:
106<br />
a,Bl sel<br />
aa.pl = R« - Ra<br />
where the first term on the<br />
right hand defines the doubleselective<br />
spin-lattice relaxation<br />
rate measured on Ha when<br />
both Ha and Hpi are excited.<br />
The cross-relaxation rate<br />
in the bound ligand is obtained<br />
by:<br />
Gobs - Gf<br />
Ob =<br />
As a consequence of binding,<br />
the cross-relaxation rate changes<br />
from positive to negative<br />
values and allows to gain geometric<br />
information on the<br />
bound molecule.<br />
In fact, substitution of<br />
the motional correlation time<br />
of the protein in the equation:<br />
ob = -0.1<br />
provides a means of evaluating<br />
proton-proton distances in the<br />
bound peptide. It comes out<br />
that at least the investigated<br />
moiety retains the conformation<br />
that was shown to be<br />
stabilized by calcium ions with<br />
exclusive occurrence of the g~<br />
rotamer.<br />
6 Effect of Calcium<br />
The same experiments used to<br />
detect and delineate binding of<br />
carnosine to HSA can be repea-<br />
h<br />
Bulletin of Magnetic Resonance<br />
ted in the presence of Ca(II) at<br />
equimolar ratios with the<br />
ligand. No substantial change is<br />
observed, as far as the selective<br />
relaxation rate enhancement<br />
and the change in the<br />
double-selective relaxation<br />
rates are concerned. An<br />
appreciable change can be<br />
however observed in the<br />
ligand-protein dissociation<br />
constant that is now measured<br />
at 2.0xl0- 5 mol dm" 3 .<br />
It is therefore possible to<br />
conclude that the effect of the<br />
metal ion is to stabilize the<br />
conformation that occurs at the<br />
bound site. In absence of the<br />
ion, such conformation has to<br />
be selected among all the<br />
several conformations that are<br />
possibly assumed by the<br />
flexible ligand in solution. This<br />
process leads to an appreciable<br />
reduction of the binding constant.<br />
7 Effect of Copper<br />
It is important to underline<br />
that the same experiments<br />
cannot be carried on when<br />
using a paramagnetic ion such<br />
as copper. It is still possible to<br />
delineate the geometry of the<br />
metal complex, but, only of<br />
that having the maximum<br />
number of ligands in the<br />
coordination sphere. One is in<br />
fact forced to work at very low<br />
[metal]/[ligand] ratios.<br />
By the same way, it is<br />
possible to detect and delineate<br />
the ternary complexes
Vol. 14, No. 1-4 107<br />
formed in the presence of the<br />
protein but there is no way of<br />
shedding light on the geometrical<br />
and conformational features<br />
of the ligand bound to<br />
the macromolecule.<br />
References<br />
[1] Dill,K. and Allerhand,A.<br />
J.Am.Chem.Soc.lOl, 4376<br />
(1979).<br />
[2] Hall,L.D. and Hill, H.D.W.<br />
J.Am.Chem.Soc. 98, 1269<br />
(1976).<br />
[3] Gaggelli, E., Kushnir, T.,<br />
Navon, G. and Valensin, G.<br />
Magn.Reson.Chem., in the<br />
press.<br />
[4] Marchettini, N. and<br />
Valensin, G. J.Phys.Chem.<br />
94, 4508 (1990)<br />
[5] Niccolai, N., Rossi, C,<br />
Mascagni, P., Neri, P. and<br />
Gibbons, W.A. Biochem.<br />
Biophys. Res. Commun.<br />
124, 739 (1984).<br />
[6] Jeener, J., Meier, B.H.,<br />
Bachmann, P. and Ernst,<br />
R.R. J.Chem.Phys. 71,<br />
4546 (1979).<br />
[7] Niccolai, N., Tiezzi, E. and<br />
Valensin,G. Chem.Rev. 82,<br />
359 (1982)<br />
[8] Bertini, I. and Luchinat, C.<br />
"NMR of paramagnetic<br />
species in biological<br />
systems", Benjamin Cummings,<br />
Menlo Park, 1986.<br />
[9] Gaggelli, E., Tiezzi, E. and<br />
Valensin, G. J.Chem.Soc,<br />
Faraday Trans. II 84,<br />
141 (1988).<br />
[10] Gaggelli, E., Gaggelli, N.,<br />
Maccotta,A. and Valensin,<br />
G. Inorg.Chem., submitted.<br />
[11] Valensin, G., Valensin, P.E.<br />
and Gaggelli, E. in "NMR<br />
spectroscopy in drug research"<br />
(Jaroszewski, J.W.,<br />
Schaumburg,K. and Kofod,<br />
H. eds.), Munksgaard,<br />
Copenhagen, 1988, p.409.<br />
[12] Gaggelli, E., Di Perri, T.,<br />
Orrico, A., Capecchi, P.L.,<br />
Laghi Pasini, F. and<br />
Valensin, G. Biophys.<br />
Chem. 36, 209 (1990).
108 Bulletin of Magnetic Resonance<br />
1. Introduction<br />
Detection and Characterization Of CFC, HCFC AND HFC<br />
Gases in Foamed Insulation by High Field NMR Imaging<br />
Leslie H. Randall<br />
Alberta Research Council, PO Box 8330, Station F,<br />
Edmonton, Alberta, T6H 5X2.<br />
In recent years, there has been considerable concern<br />
regarding the environmental impact of<br />
chlorofluorocarbons (CFC's). CFC's have been<br />
widely used as blowing agents for both cellular<br />
polyurethane and polystyrene insulation applications.<br />
[1] In many non-critical applications, GFC's are<br />
being replaced by non-fluorochemical blowing agents,<br />
but in insulation based applications where the final<br />
performance of the product is dependent on the<br />
superior insulating characteristics of a closed cell<br />
network which retains the fluorochemical blowing<br />
agent, the use of similar blowing agents will probably<br />
need to continue. Hydrochlorofluorocarbons<br />
(HCFC's) have been identified as intermediate<br />
replacements for the CFC's, but hydrofluorocarbons<br />
(HFC's) are likely to become the blowing agents of<br />
choice in insulation based products. The eventual<br />
environmental fate of the blowing agent combined<br />
with the dependency of product performance on the<br />
fluorocarbon distribution within the cellular structure<br />
requires an accurate knowledge of its spatial<br />
distribution over a period of time in order to optimize<br />
performance. Currendy, there is no reliable and<br />
readily accessible technique with which the<br />
Colin A. Fyfe, Zhiming Mei<br />
University of British Columbia,<br />
Dept. of Chemistry and Pathology,<br />
Vancouver, B.C., V6T 1Y6.<br />
and<br />
Steve Whitworth<br />
Du Pont Canada Research Centre,<br />
Kingston, Ontario, K7L 5A5.<br />
distribution of these gases can be detected or monitored.<br />
Microscopic imaging has recently emerged as an excellent<br />
technique by which the distribution of mobile fluids in<br />
polymeric materials can be monitored. [2] - [8] In principal, it<br />
should be possible to perform similar experiments on samples<br />
which contain gaseous materials, the limiting factor being the<br />
signal to noise. In the present study, we demonstrate that 19 F<br />
NMR microscopic imaging is ideally suited for measuring the<br />
changes that occur in the spatial distribution as a function of<br />
time and yields quantitatively reliable information which will<br />
be critical to the fabrication of optimized insulating materials.<br />
2. Experimental<br />
Aged foam samples were provided by Dupont Canada and<br />
contained either a single fluorinated gas or a mixture of gases.<br />
NMR measurements were made on a Bruker MSL 400<br />
spectrometer equipped with a microimaging system. All<br />
experiments were performed using the microimaging probe<br />
supplied except that the probehead was modified by replacing<br />
the vertical saddle proton rf coil by a 16 mm horizontal<br />
solenoid coil that was tuned to fluorine (376.13 MHz). The<br />
nonselective 90° and 180° rf pulses were 12.5 us and 25 ps<br />
respectively. Quadrature phase cycling was used in all<br />
spectroscopic measurements.
Vol. 14, No. 1-4 109<br />
One dimensional 'H NMR spectra were obtained<br />
by the standard one pulse method and by the Carr-<br />
Purcell spin-echo experiment The Carr-Purcell NMR<br />
sequence was used to determine the T2 spin-spin<br />
relaxation times. [9] The inversion-recovery pulse<br />
sequence was used to determine the Tj spin-lattice<br />
times. [10] ID quantitative spectroscopic data was<br />
obtained by using a short 1 us ring-down delay.<br />
The spin-echo imaging sequence [11] was used<br />
for all samples. For samples which contained a single<br />
blowing agent, non-selective 90° and 180° rf pulses<br />
were used. This reduced the echo time to 2 ms which<br />
was advantageous, since spin-spin relaxation time<br />
constant, T2 for the fluorinated gases absorbed into<br />
the foam matrix was less than 5 ms. Images were<br />
composed of 128 phase encoding steps. The number<br />
of transients per experiment was typically 160. The<br />
in-plane resolution was typically 270 pm using a<br />
frequency encode gradient on the order of 7 G/cm.<br />
Due to the short Tj relaxation time of the fluorinated<br />
gases, the recycle delay was 100 ms which allowed<br />
the acquisition of an image in under 30 minutes.<br />
Concentration profiles were obtained using a<br />
frequency selective gradient in a time period on the<br />
order of 20 seconds.<br />
Samples which contained a mixture of gases were<br />
examined by a spin-echo imaging sequence in which<br />
the initial excitation pulse was frequency selective<br />
(Gaussian shaped, 300 ps duration). The echo time<br />
for this experiment was typically 3 ms.<br />
3. Results and Discussion<br />
To compare the aging characteristics, rectangular<br />
samples 1 cm x 1 cm x 2 cm in size were cut from<br />
foam boards which had been aged for several months.<br />
A variety of fluorinated gases were examined for both<br />
polyurethane and polystyrene (high and low density)<br />
foams. To devise the most appropriate imaging<br />
protocol, it is important to know the relaxation<br />
parameters of the 19 F nuclei. Tj was typically 10 on<br />
the order of 10 ms while T2 was on the order of 4<br />
ms. (Table 1). Thus, 19 F imaging will be very<br />
efficient in that the experiment can be repeated very<br />
quickly due to the short Tj values. The data must also<br />
be acquired with a short time between excitation and<br />
data acquisition due to the short T2 value. Using a<br />
spin-echo sequence comprised of hard 90° and 180°<br />
pulses, the echo time was reduced to 2 ms.<br />
Figure 1 shows the 19 F image of a sample of<br />
polystyrene foam of rectangular cross-section which<br />
has been cut from the outside edge of a sheet of<br />
foamed insulation. The sheet has been aged at room<br />
temperature for 14 months. The concentration of gas<br />
(CH3CF2C1) decreases from the outside edge inwards. The<br />
quantitative distribution of cell gas is also obtained from an<br />
examination of the image projection (Figure IB). This shows<br />
the actual distribution as a function of distance. Although<br />
there is only minute quantity of fluorinated gas present in the<br />
sample, an image is easily obtained at high magnetic fields due<br />
to the high sensitivity of the 19 F nucleus and its short spinlattice<br />
relaxation time (10 ms).<br />
Figure 1. Concentration of CH3CF2C1 in a sample of insulating<br />
foam which has been aged 14 months,<br />
(a) Projection, (b) Spin echo image.<br />
Edge of foam<br />
Table 1: I9 F NMR Relaxation Behaviour of<br />
CFC, HFC and HCFC gases in Insulating Foams<br />
Foam Cell Gas<br />
Polyurethane CFC13<br />
T2<br />
5<br />
3.6 5.4 -5.0<br />
Polyurethane CF3CHC12 13.3 3.7 -85.9<br />
Polystyrene CF2C12 4.4 4.5 -13.5<br />
Polystyrene CH3CF2C1 4.9 4.2 -53.1<br />
Polystyrene CF3CH2F 7.4 3.0 -85.0<br />
-246.4<br />
Tj and T2 are in ms.
110 Bulletin of Magnetic Resonance<br />
Polystyrene samples which had been formed<br />
using a mixture of blowing agents (CF2C12 and<br />
CH3CF2C1) was also examined. The distribution of<br />
the two gases was compared in samples which had<br />
been aged two months and 14 months (Figure 2).<br />
The concentration and distribution of CF2C12 gas in<br />
the two samples is similar. However, the<br />
concentration of CH3CF2C1 as a function of distance<br />
from the foam edge has substantially decreased. This<br />
implies that the loss of CF2C12 is somehow slowed by<br />
the presence of the CH3CF2C1. These observations<br />
have been verified by one- dimensional spectroscopic<br />
measurements in which a calibrated standard has been<br />
used.<br />
Figure 2. (a) Concentration of CF2C12 gas<br />
(i) after 2 months and (ii) after 14 months<br />
Figure 2. (b) Concentration of CH3CF2C1 gas<br />
(i) after 2 months and (ii) after 14 months<br />
In a separate study, a cylindrical piece of insulating foam<br />
which had been formed using CH3CF2C1 was then treated to<br />
CF3CH2F in an oven at 80° C. Chemical shift selective<br />
imaging was performed on the sample and revealed that the<br />
post-treatment gas has penetrated the outside edge of the foam<br />
and has formed a ring, approximately 2-3 mm in thickness.<br />
(Figure 3a) The initial blowing gas, CH3CF2C1 has an almost<br />
uniform distribution in the centre of the foam, which decreases<br />
smoothly to the outside edge and includes the volume occupied<br />
by the CF3CH2F gas (Figure 3b). The loss of CH3CF2C1 is<br />
substantially less in these samples as compared with samples<br />
that had been placed in the oven (in air). Thus, in the<br />
presence of a fluorinated gas, the blowing agent is lost at a<br />
much slower rate.<br />
Figure 3. (a) Concentration of CH3CF2C1 gas.<br />
Figure 3. (b) Concentration of CF3CH2F gas.
Vol. 14, No. 1-4 11!<br />
4. Conclusions<br />
These data are typical of those we have obtained on<br />
a variety of fluorocarbon gas/foam matrices and<br />
clearly indicate that 19 F NMR microscopic imaging is<br />
ideally suited for measuring the distribution of<br />
fluorinated hydrocarbons in polystyrene and<br />
polyurethane foams. Presently, a protocol is under<br />
development which will ensure that quantitatively<br />
reliable information is being obtained. This will<br />
allow us to monitor the changes that occur in the<br />
spatial distribution as a function of time (accelerated<br />
aging tests) or as a function of blowing agent. This<br />
information can then be used to optimize the<br />
fabrication of these insulating materials.<br />
5. References<br />
1. Proceedings of the Polyurethane World Congress,<br />
1987.<br />
2. Rothwell, W. P., Holeck P. R. & Kershaw, J. A.<br />
/. Polym. Sri. Polym. Lett. Ed., 22, 241 (1984).<br />
3. Blackband, S. & Mansfield, P. J. Phys. C: Solid<br />
State Phys., 19, L49 (1986).<br />
4. Marcei, T. H., Donstrup, S. & Rigamonti, A. J.<br />
Mol. Liquids, 38, 185, (1988).<br />
5. Weisenberger, L. A. & Koenig, J. L. Appl.<br />
Spectrosc, 42, 1117(1989).<br />
6. Weisenberger, L. A. & Koenig, J. L.<br />
Macromolecules, 23, 2445 (1990).<br />
7. Webb A. G. & Hall, L. D. Polym. Commun. 11,<br />
422 (1990).<br />
8. Webb A. G. & Hall, L. D. Polym. Commun. 11,<br />
425 (1990).<br />
9. H. Y. Carr and E. M. Purcell, Phys. Rev., 94, 630<br />
(1954).<br />
10. R. L. Void, J. S. Waugh, M. P. Klein and D. E.<br />
Phelps, J. Chem. Phys., 48, 383 (1968).<br />
11. Edelstein, W.A.; Hutchinson, J.M.S.; Johnson G.;<br />
Redpath, T. Phys. Med. BioL, 25, 751 (1980).
112<br />
MYSTERIOUS NEGATIVE PEAKS<br />
IN THE 1 H{ 1 H} NOE DIFFERENCE SPECTRA<br />
OF SOME THIOPYRAN COMPOUNDS<br />
Thioyrans 1-4 were formed in the course of<br />
regiospecific and stereoselective Diels-Alder<br />
cycloadditions involving 2-(N-Acylamino)-lthia-l,3-dienes.<br />
The constitution and the main<br />
conformational features of thiopyrans 1-4 were<br />
established using classical high field NMR<br />
methods: 2D ^ H and 13 C- X H shift correlation,<br />
selective 13 C{ l U} and ^H} NOE difference<br />
experiments [1], [2].<br />
Conformational aspects:<br />
a) ring pseudorotation:<br />
In the case of compounds 3 and 4 the<br />
thiopyran ring exhibits a slow ring flip between<br />
the relevant half-chair forms. This gives rise to<br />
saturation transfer effects in the 1 H{ 1 H} NOE<br />
difference experiments, as illustrated for 4 in<br />
Fig. 1.<br />
Csaba Szantay, Jr.<br />
Chemical Works of Gedeon Richter Ltd.<br />
H-1475, Budapest, POB 27, Hungary.<br />
Bulletin of Magnetic Resonance<br />
In 1 and 2 the ring conformational equilibrium<br />
is completely shifted towards the form<br />
in which H-2 and H-3 are antiperiplanarly arranged<br />
(Fig. 2).<br />
b) The C(3) sidechain:<br />
The sidechain shows free rotameric mobility<br />
at room temperature. The conformational features<br />
were characterized in terms of the<br />
measured NOEs and MM calculations [1]. The<br />
relative contributions of the major sidechain<br />
conformations are depicted in Fig. 2.<br />
Ac2N-<br />
AcN-<br />
anti syn<br />
Ph<br />
H<br />
s cis—trans<br />
H P \ h f<br />
. trans—cis<br />
\h<br />
Ph<br />
H<br />
0-<br />
i %<br />
1<br />
A-<br />
~100%<br />
2<br />
Ac2N—(/<br />
AcN—V<br />
Et S<br />
Figure 2. The main rotameric forms of the<br />
C(3) sidechain in compounds 1-4.<br />
Q
Vol. 14, No. 1-4 113<br />
minor<br />
NHMe<br />
major<br />
H-5<br />
U.I<br />
NH<br />
HMe<br />
H-3<br />
NMe<br />
NHMe<br />
•» ifr<br />
COMe<br />
Figure 1. One of the several NOE difference spectra of compound 4 showing the presence of<br />
saturation transfer as a result of a slow ring interconversion.<br />
H-5<br />
JL<br />
G.00 5.S0<br />
Ph<br />
-14X<br />
H-2<br />
PPH S.00<br />
4.SO<br />
+9X<br />
H-4<br />
JL<br />
Figure 3. A segment of the NOE difference<br />
spectra showing negative peaks in<br />
the case of 2. (CDC13,25 °C, 400 MHz).<br />
Ph<br />
2 PPH<br />
-30°C<br />
+7X<br />
' I ' • ' ' I ' ' • • I ' • « " T " t ' • • ! • ! •<br />
6.0 5.S 5.0 4.5 4.0<br />
PPH<br />
Figure 4. A segment of the NOE difference<br />
spectra showing negative peaks in<br />
the case of 1. (CDC13, -30 °C, 400 MHz).
t<br />
,!<br />
11<br />
114<br />
The mystery:<br />
During these invetigations a highly unusual<br />
phenomenon was observed in the ^^H} NOE<br />
difference spectra of compounds 1 and 2, in that<br />
the H-3 and H-2 protons are connected by massive<br />
negative peaks while all other NOEs are<br />
positive as illustrated in Figs. 3 and 4.<br />
The facts:<br />
a) The observed effect involves H-2 and H-3<br />
only.<br />
b) The phenomenon appears to be specifically<br />
linked to the presence of the C = O unit in<br />
the sidechain, and is compeletely absent in 3<br />
and 4.<br />
c) As measured at 400 and 300 MHz, respectively,<br />
the negative peaks showed no field dependence<br />
beyond that attributable to<br />
experimental errors.<br />
d) The intensity of the negative enhancements<br />
increases markedly with decreasing<br />
temperature and viscosity, and tends towards<br />
zero at elevated temperatures in accordance<br />
with the antiperiplanar arrangement of H-2 and<br />
H-3:<br />
Negative enhancements at 400 MHz,<br />
measured at various temperatures using<br />
otherwise identical experimental<br />
parameters:<br />
1<br />
-30°C<br />
-20%<br />
CDCb<br />
+ 25°C<br />
-3%<br />
+25°C<br />
-14%<br />
+50°C<br />
0%<br />
+ 50°C<br />
-4%<br />
DMSO<br />
+25°C<br />
-14%<br />
e) The nonselective Tx relaxation times of all<br />
ring protons in 1 (CDC13, +25°C, 400 MHz)<br />
were measured to be ca. 1 s. This accords with<br />
the size and expectedly fast tumbling of the<br />
molecule.<br />
Bulletin of Magnetic Resonance<br />
Conceivable (but rejectable) explanations:<br />
The effect seems to be inexplicable in terms<br />
of the well understood cross-relaxation or conceivable<br />
saturation transfer processes:<br />
a) The trivial case of irradiation spillover can<br />
be discounted because: 1) This would be incompatible<br />
with the observed temperature dependence<br />
of the negative peaks; 2) Fig. 4. illustrates<br />
a situation in which the signal due to H-3 is<br />
"halfway" in between those of H-5 and H-2, of<br />
which only the latter shows the negative effect.<br />
(The irradiation power levels used in all experiments<br />
were: 54L (Bruker AM 400) and<br />
DLP = 30 (Varian VXR-300).<br />
b) A three-spin effect is geometrically unjustified,<br />
and is incommensurate with the magnitude<br />
of these negative enhancements.<br />
c) The possibility of saturation transfer may<br />
be considered as being the result of some form<br />
of chemical exchange in 1 and 2. Possible options<br />
are: 1) direct proton transfer between H-2<br />
and H-3. However, *H and 2 H NMR studies on<br />
the C(3)- 2 H labeled isotopomer of 1 have conclusively<br />
shown that any possibility of a<br />
stereoselective H-2—H-3 proton transfer can be<br />
discounted. 2) A possible chemical exchange<br />
may also involve restricted ring pseudorotation<br />
or restricted C(3) sidechain mobility, both<br />
being slow on the chemical shift time-scale. Assuming<br />
that in such a hypothetical situation the<br />
H-3 signal of the minor conformer lies directly<br />
underneath the H-2 signal of the major conformer<br />
(and vice versa !), this could lead to<br />
negative peaks that are (vaguely) similar to<br />
those observed in the NOE difference spectra.<br />
However, such a situation can be ruled out on<br />
account of several quite obvious considerations,<br />
the most trivial being that 1) the multiplet patterns<br />
of the observed negative peaks do not<br />
conform to those expected for either of the<br />
above possibilities for slowly interconverting<br />
conformational species; 2) the phenomenon is<br />
observed only in relation to H-2 and H-3.
Vol. 14, No. 1-4<br />
Moreover, the observed temperature and<br />
viscosity dependence of the effect is contrary to<br />
that normally expected for saturation transfer:<br />
1) The longer correlation times associated with<br />
higher solution viscosity (in our case in DMSO)<br />
or lower temperatures are, in the extreme narrowing<br />
limit, related to more efficient Ti relaxation<br />
which works against saturation transfer. 2)<br />
Decreased temperatures usually diminish the<br />
efficiency of the exchange process.<br />
d) A major contribution from scalar relaxation<br />
[3] which could be perceived as being<br />
brought about by the somewhat labile character<br />
of H-3 can be ruled out since the J(H-2,H-3)<br />
couplings show no sign of the rapid modulation<br />
that would result in the collapse of the relevant<br />
rmiltiplet patterns.<br />
e) Strong coupling effects [4] are not present,<br />
and would not give negative peaks of this size.<br />
f) Anisotropic reorientation in the intermediate<br />
region between fast and slow tumbling<br />
can be considered. This carries the possibility of<br />
exhibiting small positive and small negative enhancements<br />
simultaneously. However, all the<br />
115<br />
positive enhancements measured between spatially<br />
analogously related protons in 1-4 were of<br />
similar magnitude, and in line with that expected<br />
for a fastly tumbling molecule.<br />
The phenomenon therefore still awaits adequate<br />
rationalization. The question of how the<br />
effect may be related to the geometrical and<br />
constitutional specifics of these molecules is<br />
still being pursued.<br />
1. I.T. Barnish, C.W.G. Fishwick, D.R. Hill,<br />
Cs. Szantay Jr., Tetrahedron, 45 (1989) 6771.<br />
2. Cs. Szantay Jr., I. Moldvai, C.W.G. Fishwick,<br />
D.R. Hill, Tetrahedron Lett, 32 (1991)<br />
2529.<br />
3. D. Neuhaus, M.P. Williamson, The<br />
Nuclear Overhauser Effect in Structural and<br />
Conformational Analysis, VCH Publishers, New<br />
York, 1989, pp. 203-207.<br />
4. J. Keeler, D. Neuhaus, M.P. Williamson, /.<br />
Magn. Reson., 73 (1987) 45.
116 Bulletin of Magnetic Resonance<br />
H-1 and C-13 NMR Spectra of the Carbanions<br />
Produced from Phenylpropene Derivatives<br />
Akihiro Yoshino, Kensuke Aoki, Masahiro Ushio,<br />
and Kensuke Takahashi<br />
Department of Applied Chemistry, Nagoya Institute of Technology,<br />
Gokiso-cho, Showa-ku, Nagoya 466, Japan<br />
Abstract l,l-Diphenyl-2-methyl-l-propene produces an equimolar mixture<br />
of two carbanions in contact with excess potassium-sodium alloy in<br />
tetrahydrofuran. The carbanions can be interpreted as the products of disproportionation<br />
reaction between two radical anions produced by one-electron<br />
reduction of the phenylpropene with alkali metal.<br />
1 Introduction<br />
Various carbanions can be prepared<br />
from substituted ethylenic<br />
compounds (1) in contact with alkali<br />
metal in tetrahydrofuran(THF).<br />
In these reactions a radical anion<br />
(2) is formed as shown in Scheme 1<br />
Scheme 1<br />
and its reactivity or stability will<br />
depend on the nature of the substituents.<br />
The three routes of reactions<br />
can be considered as follows.<br />
(1) If the ethylene has bulky<br />
substituents on both Ci and C2, the<br />
corresponding ethylene dianion is<br />
produced (Scheme l).[l-4] (2) If<br />
the ethylene has no or sterically<br />
small substituent such as only one<br />
methyl group on either Ci or C2, the<br />
corresponding dimer dianion is produced<br />
(Scheme 2).[5-7] (3) In the<br />
present paper, three title phenyl<br />
R, © • _R<br />
2 X >C,-C2 '>C,-CH< 3 + >C,-C2< '<br />
^ R4THF R/ V R4 R/ 2K R4<br />
2 5 6<br />
Scheme 3<br />
carbanions, 5 and 6 respectively,<br />
whose structures are confirmed by<br />
!H and 13 C NMR spectra. One of<br />
these anions(5) is a phenylalkyl<br />
carbanion and the other is a phenylallyl<br />
carbanion (6). Thus the result<br />
may be interpreted as a dispro-
Vol. 14, No. 1-4 117<br />
portionation reaction between two<br />
anion radicals generated first by<br />
one-electron reduction of the<br />
starting phenylpropene with alkali<br />
metal. As far as we know, this type<br />
of simultaneous formation of two<br />
different carbanions has not yet<br />
been reported. The conditions of<br />
these reactions will be discussed in<br />
terms of steric bulkiness of the<br />
substituents.<br />
2 Experimental<br />
The starting materials were prepared<br />
from dehydration of the corresponding<br />
phenylpropanols which<br />
were prepared by Grignard reaction<br />
and followed by dehydration with<br />
anhydrous acetic acid. The starting<br />
materials dissolved in THF or THFd8<br />
were kept in contact with<br />
potassium-sodium alloy in vacuum<br />
at room temperature for about 24 h.<br />
The resulting dark red solutions<br />
were filtered, concentrated if necessary,<br />
and then sealed into a 5-mm<br />
NMR sample tube. Their concentrations<br />
were about 1 M. The *H and<br />
13 C NMR measurements were carried<br />
out at 22°C using Varian XL-<br />
200 or Unity-400 spectrometer.<br />
The chemical shifts were evaluated<br />
from the upfield peak of THF or<br />
THF-dg, used as an internal reference.<br />
This peak was taken as 1.79<br />
or 1.75 for l H and 26.4 or 26.0 ppm<br />
for 13 C resonances, respectively,<br />
from TMS.<br />
3 Results and Discussion<br />
1 H NMR Spectra of the Carbanions.<br />
A typical l H NMR spectrum and<br />
chemical shifts are shown in Fig. 1<br />
and Table 1, respectively. In Fig. 1,<br />
there are three signals in the region<br />
from 0 to 3 ppm, which are assigned<br />
to one methyl and one isopropyl<br />
groups. There are two<br />
ethylenic signals in the region from<br />
3 to 4 ppm. One problem in the<br />
spectrum is the origin of hydrogen<br />
in the isopropyl group. To clarify<br />
the origin, the experiments using<br />
two different solvents, THF and<br />
THF-d8 were carried out. Since<br />
these two spectra are similar each<br />
Table 1. *H Chemical Shifts of Related Carbanions and Their Precursors.<br />
No.<br />
1 a<br />
1 b<br />
1 c<br />
5 a<br />
5 •b<br />
5<br />
6<br />
6<br />
6<br />
c<br />
a<br />
b<br />
c<br />
R?<br />
Ph<br />
Ph<br />
CH3<br />
Ph<br />
Ph<br />
CH3<br />
Ph<br />
Ph<br />
CH3<br />
R 4<br />
CH3<br />
C2H5<br />
CH3<br />
CH3<br />
C2H5<br />
CH3<br />
CH3<br />
C2H5<br />
CH3<br />
H2<br />
2 .904<br />
2 .591<br />
2 .540<br />
t rans ci s<br />
1<br />
H3<br />
1.862<br />
1.107<br />
.580 a) 1.<br />
1.115<br />
1.114<br />
0.875<br />
780<br />
4 .035<br />
4<br />
3<br />
b> 4.<br />
.375 b) 4.<br />
.141 b> 695<br />
811<br />
3. 343<br />
a )<br />
b)<br />
b)<br />
b)<br />
Ho<br />
( H o )<br />
1<br />
1<br />
1<br />
6.741<br />
6.756<br />
5.084<br />
5.368<br />
7.119<br />
7.091<br />
6.584<br />
H m<br />
(H»- )<br />
7.24<br />
7.25<br />
7.14<br />
6.593<br />
6.563<br />
6.076<br />
6.044<br />
6.711<br />
6.661<br />
6.656<br />
Hp<br />
1<br />
1<br />
5.760<br />
5.726<br />
4.364<br />
6.121<br />
5.966<br />
5.735<br />
a) Trans(R4) and cis(Ra) configurations are defined for<br />
b) They are defined for Ci.<br />
Hothe<br />
CH3<br />
1.842<br />
1.950<br />
0.872<br />
1.287<br />
1.870<br />
1.068<br />
1.858<br />
1.971<br />
r s<br />
CH2<br />
2.802<br />
1.467<br />
1.628<br />
2.252
118<br />
6Ho<br />
5Ho<br />
l6Hm 6HMe 5H3<br />
Ph CH3 Ph CHs<br />
\ - / \ - /<br />
C-C H + C-C<br />
/ 1 2 \ / 1 2 ^<br />
P h 3 C H i P h 3 C-H<br />
- /<br />
H<br />
5a 6a<br />
8 4<br />
6 I ppm<br />
Fig.l !H NMR spectrum of an equimolar mixture<br />
of 5a and 6a in THF-d8 at 200MHz.<br />
5Cm<br />
6C2<br />
6Ci 5Ci<br />
1 f<br />
160<br />
6Cm<br />
6Co P h e n , P h C H ,<br />
\- / \- /<br />
C-CH + C-C<br />
5CO / 1 2 \ /12V<br />
Ph 3CHj Ph H 3 C-H<br />
5a 6a<br />
6Cp<br />
Bulletin of Magnetic Resonance<br />
5Cp 6CMe<br />
5C3<br />
5C2<br />
6C3<br />
5C1<br />
Jt<br />
l l l 1 1 i i<br />
80<br />
Fig .2. 13 C NMR spectrum of a mixture of 5a<br />
and 6a in THF-d8 at 50.3 MHz.<br />
l i i i i i i<br />
61 ppm<br />
I<br />
0<br />
0
Vol. 14, No. 1-4<br />
other, it must be concluded that the<br />
CH hydrogen in the isopropyl group<br />
of 5a is coming from another radical<br />
anion (2), but not from the solvent.<br />
This is also supported by an<br />
experimental fact that the product<br />
ratio of 5 and 6 is 1 to 1. Another<br />
point of interest is that two<br />
methylene hydrogens of 5b give<br />
different chemical shifts, 1.467<br />
and 1.628 ppm, respectively. This<br />
is ascribed to the presence of an<br />
asymmetric center on C2.<br />
13 C NMR Spectra of the Carbanions.<br />
A typical 13 C NMR spectrum and<br />
chemical shifts are shown in Fig. 2<br />
and Table 2, respectively. In Fig. 2,<br />
there are 15 signals for a mixture<br />
of 5a and 6a except for two solvent<br />
peaks. One methyl signal is overlapped<br />
with a more shielded solvent<br />
peak. The carbon species can be<br />
differentiated by the DEPT technique.<br />
The assignment of each peak<br />
is shown in Fig. 2.<br />
IH and 13 C NMR Chemical Shifts.<br />
All the *H and 13 C except for C2 and<br />
Cj of the carbanions (5 and 6) show<br />
upfield shifts as compared with<br />
those of the starting materials.(1)<br />
These upfield shifts are mainly<br />
Table 2. 13 C Chemical Shifts of Re<br />
No. R2 R.4 C<br />
(C<br />
la Ph CH3 138.48 131.24<br />
1 b Ph C2H5 138 .47 136.74<br />
1 c CH3 CH3 146.11 127.47<br />
5 a<br />
5 b<br />
5 c<br />
6 a Ph<br />
6 b Ph<br />
6 c CH3<br />
Ph CH3<br />
Ph C2H5<br />
CH3 CH3<br />
90.26<br />
89.56<br />
78.19<br />
30.30<br />
38.24<br />
29.10<br />
CH3 89.61 147.54<br />
C2H5 88.65 156.93<br />
CH3 80.24 144.30<br />
ascribed to the extra negative<br />
charges on carbons of the carbanions.<br />
Since the carbanions have their<br />
stability due to partial delocalization<br />
of the extra charge into the<br />
phenyl rings, phenyl rings are<br />
necessary for formation of a stable<br />
carbanion. In fact, tetramethylethylene<br />
does not react with a<br />
potassium-sodium alloy in a<br />
similar condition used for the<br />
present study.<br />
The extent of charge delocalization<br />
can be thus estimated by comparison<br />
of the *H or 13 C NMR chemical<br />
shifts. The *H and 13 C chemical<br />
shifts of Hp and Cp of 6a are less<br />
shielded than those of 5a, by about<br />
0.361 and 5.49 ppm, respectively.<br />
These shift differences divided<br />
by 10.7 and 160 ppm, respectively,<br />
gave about 0.034 unit of extra<br />
charge difference on Cp. [8]<br />
Therefore, the chemical shift<br />
difference between Hp or Cp of 6a<br />
and 5a is explained in terms of<br />
their localized excess charges. The<br />
same is true for 6b and 6c in<br />
comparison with 5b and 5c, respectively.<br />
The especially large<br />
lated Carbanions and Their Precursors.<br />
C<br />
) (c<br />
V-» o t h e r s<br />
CH3<br />
22.71 144.22 130.58 128.61 126.85 ---<br />
19.26 144.25 130.17 128.64 126.83 13.64<br />
20<br />
21<br />
22<br />
19<br />
21<br />
64<br />
07<br />
11<br />
46<br />
76<br />
130.40 128.71 126.89<br />
131.14 129.08 128.71 126.55 22.33<br />
CH2<br />
119<br />
29.20<br />
144.28 118.75 128.90 107.91 —<br />
145.02 118.77 128.81 107.80 14.19 29.45<br />
135.54 105.30 130.75 88.50 12.09<br />
105.47 130.25<br />
93.16 148.38 124.86 128.61 113.40 24.76<br />
97.66 148.25 123.21 128.61 112.11 15.87 29.94<br />
76.36 146.24 118.64 128.59 107.46 19.39<br />
27.07
120<br />
difference between 6c and 5c, is<br />
ascribed to the delocalized capacity<br />
for excess charge consisting of only<br />
one phenyl ring. Therefore, Ci of 5c<br />
is more shielded by about 10 ppm<br />
than that of 5a or 5b. The same is<br />
true for Ci and C3 in 6c; they are<br />
more shielded than those in 6a or<br />
6b. Thus the p-orbital of Ci can<br />
conjugate with those in both phenyl<br />
and allyl groups in 6., while that of<br />
5 can only conjugate with phenyl<br />
rings.<br />
For the allyl anions the average<br />
shift of the outer carbons (Ci and<br />
C3) is shielded by about 60 ppm<br />
than that of the center carbon (C2),<br />
as shown in Table 2. This is a characteristic<br />
nature of the allyl anions.[12]<br />
Dunkelbium and Brenner reported<br />
on the same carbanion (6a) with<br />
lithium as a counterion.[13]<br />
Although their chemical shifts of<br />
6a were deshielded by about 0.1<br />
ppm, their *H NMR data are consistent<br />
with ours in consideration of<br />
different counterion.<br />
Internal Rotations. Three internal<br />
rotations can be considered around<br />
the bonds of Ci to Ci, Ci to C2, and<br />
C2 to C3 for 5 and 6 (these will<br />
hereafter be referred to as rotations<br />
A, B, and C). In the *H and 13 C<br />
NMR spectra of 5c, two signals for<br />
Ho, Co, Cm, Hm were observed and<br />
their chemical shifts are given in<br />
Tables 1 and 2, respectively. This<br />
is interpreted as a restricted rotation<br />
A for the phenyl ring of 5c<br />
even at room temperature. A similar<br />
observation was presented earlier<br />
in a methylbenzyl anion.[14] A<br />
chemical shift difference of 0.284<br />
ppm for Ho of 5c is smaller than<br />
that of 0.43 ppm for Ho of methyl-<br />
Bulletin of Magnetic Resonance<br />
benzyl anion. The cause of the<br />
shifts cannot be clarified yet; the<br />
shift differences for Ho, Co, Cm,<br />
and Hm changed to 0.284, 0.17,<br />
0.50, and 0.032 ppm. These four<br />
sites are adjacent to each other.<br />
But the values change alternatively.<br />
Therefore the origin of the shift<br />
may not be explained simply. On the<br />
other hand, the internal rotations A<br />
of the phenyl groups of 5a and 5b<br />
were not restricted at room temperature.<br />
At lower temperature,<br />
however, the rotation B was inhibited<br />
at -60°C for 5b. For 5a the<br />
aromatic carbon signals were significantly<br />
broadened at -80°C but<br />
were not split. In comparison with<br />
this behavior, however, those aromatic<br />
carbon signals of 6a and 6b<br />
were observed sharply even at<br />
-80°C. Further study is necessary.<br />
Two Ho signals of 5c at 5.084 and<br />
5.368 ppm were correlated with<br />
two Co signals at 105.30 and<br />
105.47 ppm, respectively, in a 2D<br />
CH COSY spectrum for their assignment<br />
purpose. [15] Similarly, two<br />
Hm signals (6.044 and 6.076 ppm)<br />
are correlated with two Cm signals<br />
(130.25 and 130.75 ppm, respectively).<br />
Which one of the two Ho's<br />
locates near to methyl or isopropyl<br />
group? NOE of the methyl signal<br />
(1.287 ppm) was observed on the<br />
more shielded Ho signal (ca. 20 %).<br />
Therefore the methyl group on Ci is<br />
nearer to Ho than Ho'. NOE of another<br />
methyl signal (0.875 ppm)<br />
was observed on the less shielded<br />
Ho" signal (ca. 10 %).<br />
Condition of Disproportionation<br />
Reaction. The condition can be<br />
discussed in terms of bulkiness of<br />
the substituents. Four groups are
Vol. 14, No. 1-4 121<br />
used as stubstituents on Ci and C2;<br />
namely H, CH3, C2H5, and CeHs. They<br />
are differentiated to have approximately<br />
1, 2, 3, and 5 A in their diameters<br />
as spherical models. For<br />
simplified discussion, we consider<br />
that 2 has bulkier substituents on<br />
Ci than on C2. Thus the C2 site will<br />
be more reactive than the Ci site<br />
for dimerization. In this occasion<br />
discussion will be rather limited to<br />
the bulkiness of substituents on C2.<br />
In cases where substituents are<br />
two C6H5 or one CgHs and one H,<br />
monomer dianions are formed<br />
(Scheme 1). Dimerization of the<br />
radical anion occurs in cases where<br />
the substituents are two hydrogens<br />
or one H and one CH3 (Scheme 2). In<br />
cases where the substituents are<br />
two CH3, disproportionate occurs<br />
(Scheme 3). Therefore bulkiness of<br />
the substituents may control the<br />
progress of the reaction. Two alkyl<br />
groups are large in size for dimerization,<br />
and small for dianion formation,<br />
and may be suitable for<br />
disproportionation. For disproportionation<br />
reaction, the substituent<br />
must have a hydrogen to be abstracted.<br />
One CH3 on C2 of la is<br />
substituted by C2H5 in order to investigate<br />
which hydrogen in two<br />
substituents, either CH3 or C2H5, is<br />
easily abstracted. From analyses of<br />
a mixture of the products, it is<br />
concluded that the hydrogen-leaving<br />
power is about ten times stronger<br />
in CH3 than in C2H5.<br />
Acknowledgement<br />
The authors wish to thank Mr. K.<br />
Kushida and A. Ono (Varian<br />
Instruments, Ltd.) for their kindness<br />
in measuring several 2D CH<br />
COSY spectra at 100.6 MHz.<br />
4 References<br />
1) K.Takahashi,Y.Inoue, and R.Asami,<br />
Org. Magn. Reson., 3, 349 (1971).<br />
2) M.Morikawa, H.Matsui,A.Yoshino,<br />
and K.Takahashi,£«//. Chem.Soc.Jpn.,<br />
57,3327(1984).<br />
3) H.Fujiwara,A.Yoshino,<br />
Y.Yokoyama, and K.Takahashi,<br />
Bull.Chem.Soc.Jpn., 65, No.8, in<br />
press (1992).<br />
4) Y. Yokoyama, T.Koizumi, and<br />
O.Kikuchi, Chem. Lett., 1991, 2205.<br />
5) K.Takahashi and R.Asami,<br />
Bull.Chem.Soc.Jpn., 41, 231 (1968).<br />
6) K.Takahashi,M.Takaki, and<br />
R.Asami, J.Phys.Chem.,<br />
75,1062(1971).<br />
7) Y.Yokoyama, et.,<br />
Bull.Chem.Soc.Jpn., 61, 1557(1988).<br />
8) The charge calculation was followed<br />
by using an empirical equation<br />
proposed by Fraenkel et al.[9]<br />
and proportional factors presented<br />
by Schaefer et al.[10,ll]<br />
9) G.Fraenkel, et al.,<br />
J.Am.Chem.Soc.,82, 5846(1960).<br />
10) T.Schaefer and W.G.Schneider,<br />
Can. J. Chem., 41, 966 (1963).<br />
11) H.Spiesecke and W.G.Schneider,<br />
Tetrahedron Letters, 1961, 468.<br />
12) D.H.O'Brien/'Comprehensive<br />
Carbanion Chemistry," ed by<br />
E.Buncel and T.Durst, Elsvier, New<br />
York (1980), Vol 5A, p.273.<br />
13) E.Dunkelblum and S.Brenner,<br />
Tetrahedron Letters, 1973, 669.<br />
14) K.Takahashi, et al.,<br />
Org.Magn.Reson., 3, 539 (1971).<br />
15) The 2D CH COSY spectra were<br />
measured at 75.4 MHz with a<br />
Hitachi R-3000 spectrometer. The<br />
authors wish to thank Mr. M. Tamura<br />
(Instrument Division Hitachi, Ltd.)<br />
for his kindness in measuring the<br />
spectra.
122 Bulletin of Magnetic Resonance<br />
Intracellular pH and Inorganic Phosphate<br />
Effects on<br />
Skeletal Muscle Force<br />
E. R. Barton-Davis 1 , R. W. Wiseman 2 ,<br />
and<br />
M. J. Kushmerick 1 ' 2<br />
Depts. of Physiology and Biophysics* and of Radiology 2 ,<br />
University of Washington,<br />
Seattle, WA 98195, USA<br />
INTRODUCTION<br />
The molecular mechanisms of skeletal muscle<br />
fatigue are not fully understood. Intracellular pH<br />
(pHi) and inorganic phosphate (Pi) have been<br />
implicated as causes of peripheral fatigue. Both a<br />
rise in [H + ] (decreased pHi) and elevated Pi<br />
result in the inhibition of force generation in<br />
skinned fibers (2) and sometimes in whole<br />
muscle preparations (1, 3). During exercise, an<br />
accumulation of H + and Pi occur simultaneously<br />
over time, which confounds the interpretation of<br />
the relative role of each with regard to skeletal<br />
muscle fatigue. The specific protonation state of<br />
Pi might also inhibit force generation, because<br />
the relative amount of Pi species (H2PO4" and<br />
HPO4 2 ") is dependent on pHi, and these amounts<br />
change over the physiologic range of pHi (pKa =<br />
6.79) (3). In order to evaluate the role these<br />
metabolites have in muscle force, the effects of<br />
pHi and Pi on force have been dissociated by<br />
independently manipulating each.Changes in the<br />
amount of CO2 in the perfusate (PCO2)<br />
manipulates pHi (1), and exogenous pyruvate<br />
lowers Pi (6). Performing simultaneous NMR<br />
spectroscopic and mechanical measurements<br />
can assess the effects of these treatments. We<br />
have compared the relationship between force<br />
and pHi in the presence and absence of inorganic<br />
phosphate and interpreted the relative<br />
contribution of Pi and pHi on muscle force<br />
generation.<br />
METHODS<br />
Isolated soleus from male Swiss Webster mice<br />
were attached to a Harvard Apparatus isometric<br />
force transducer and placed in a bath containing<br />
MOPS Ringers (llmM glucose, pH=7.4, 95%<br />
O2, 5% CO2) at 23°C. Tetanic force, as well as<br />
rise and relaxation time, were measured at<br />
optimum length every 10 minutes. Stimulation<br />
was at supramaximal voltage and fusion<br />
frequency for 1 second. The collateral muscle<br />
was placed in a custom designed NMR probe (5)<br />
and superfused from the same source of Ringers<br />
as the mechanics bath for 31 P-NMR spectra were<br />
acquired on a General Electric 7 Tesla GN300 at<br />
121 MHz. Intracellular pH measurements were<br />
made using the chemical shift of Pi relative to<br />
PCr and the equation<br />
pHi = 6.79 + log{(.89 - 8)/(8 - 3.19)}.<br />
Mechanical and metabolic measurements were<br />
repeated with control (11 mM glucose) and<br />
substrate (20mM pyruvate) solution (PYR),<br />
varying pHi by equilibration with 5%, 25%, and<br />
50% CO2, which resulted in PCO2 values of 22,<br />
174, and 266 torr respectively. All isometric<br />
contractions were bracketted by llmM Glucose<br />
5% CO2 treatment to normalize data. Statistical<br />
analysis was performed using 2-way ANOVAS<br />
with relative force as the dependent variable, and<br />
pHi and Pi as independent variables. Paired<br />
comparisons of means were performed between<br />
each PCO2 l eve * aa ^ substrate treatment. Values<br />
of p
Vol. 14, No. 1-4<br />
RESULTS<br />
PYR lowered Pi levels from ~5 mM to
124 Bulletin of Magnetic Resonance<br />
TABLE II: Total, Diprotonated, and<br />
Monoprotonated Phosphate Levels in<br />
Soleus<br />
Glue<br />
5%<br />
CO2<br />
Glue<br />
25%<br />
CO2<br />
Glue<br />
50%<br />
CO2<br />
Pyr<br />
5%<br />
CO2<br />
Pyr<br />
25%<br />
CO2<br />
Pyr<br />
50%<br />
CO2<br />
pHi<br />
7.19<br />
+.015<br />
6.81<br />
+.018<br />
6.54<br />
+.013<br />
7.22<br />
+.04<br />
6.74<br />
+.04<br />
6.48<br />
+0.0<br />
DISCUSSION<br />
Total<br />
Pi<br />
(fi«/g)<br />
6.76<br />
+.73<br />
(n=22)<br />
7.74<br />
+2.10<br />
(n=4)<br />
6.74<br />
+.05<br />
(n=2)<br />
2.64<br />
+1.01<br />
(n=2)<br />
1.69<br />
(n=l)<br />
N.D.<br />
H2PO4<br />
(M«/g)<br />
1.91<br />
+.22<br />
3.68<br />
+.95<br />
4.25<br />
+.02<br />
0.69<br />
+.22<br />
0.88<br />
N.D.<br />
HPO4-<br />
(uste)<br />
4.85<br />
+.52<br />
4.06<br />
+1.16<br />
2.48<br />
+.07<br />
1.96<br />
+.78<br />
0.81<br />
N.D.<br />
Data presented in mean ± SE<br />
Decreased [Pi] increased force at all PCO2<br />
levels, and did enhance the dependence of force<br />
generation on pHi. The effect of pHi on force<br />
concurs with skinned fiber data (2), but does not<br />
agree with results from hypercapnic<br />
manipulation in perfused muscle preparations<br />
(1). There are several possible explanations for<br />
this discrepency. One possibility is that there is<br />
inadequate O2 delivery to the isolated muscle<br />
preparation. However, since the mouse soleus is<br />
small (~1 mm in diameter), there is little chance<br />
of O2 limitation during hypercapnia because the<br />
diffusional distances are less than 500p.m. It is<br />
also possible that at low pHi, a 1 second tetanic<br />
stimulation was not sufficiently long enough for<br />
the muscle to attain maximum force . However,<br />
longer stimulations did not appreciably increase<br />
force (data not shown). Force measured at 1<br />
second tetani was not significantly different from<br />
that of longer tetani. Hence, the results observed<br />
give an accurate presentation of the effects pHi<br />
has on tetanic force in isolated whole muscle.<br />
It has been suggested that a specific protonation<br />
state of phosphate is correlated to muscle force<br />
(3). The diprotonated species (H2PO4") is<br />
thought to lower force by binding more tightly to<br />
actomyosin crossbridges. Because force<br />
production is correlated with the release of Pi<br />
from the crossbridge, the higher binding affinity<br />
of H2PO4" would result in a decrease in force. As<br />
pHi decreases, there is a greater proportion of<br />
H2PO4" present, indicating that this<br />
phenomenon is an indirect effect of pHi on<br />
skeletal muscle force. We examined the<br />
dependence of force on pHi, HPO4 2 ", and<br />
H2PO/. The slopes of each regression line show<br />
a higher correlation between force and pHi<br />
(R^.966 in normal [Pi]) (Figure 1) than between<br />
force and either phosphate species (H2PO4",<br />
R^.138; HPO4 2 ", R 2 =.O35) alone (Table II). In<br />
these experiments, the primary effect of pHi on<br />
force does not appear to be mediated by a<br />
specific species of phosphate.<br />
Table I shows that pHi and Pi influence other<br />
properties during tetanic stimulation. Proton<br />
accumulation could alter kinetics of force<br />
development (rise and relaxation times) in<br />
several ways. Enzymes, including Ca 2+ -ATPase,<br />
actomyosin ATPase, and those in glycolysis,<br />
such as phosphofructokinase, are inhibited when<br />
pHi falls. The decline in enzymatic activity<br />
inhibits the speed at which a muscle can develop<br />
force or relax after stimulation. Low Pi causes<br />
faster rise yet slower relaxation times by<br />
different mechanisms which are not fully<br />
understood. Faster kinetics are expected since the<br />
phosphate release step in the crossbridge cycle<br />
would be more favorable due to lower Pi.<br />
However, this hypothesis can not explain why<br />
relaxation rates are significantly slower than in<br />
control conditions. It would seem plausible that<br />
Ca 2+ uptake at the sarcoplasmic Ca 2+ -ATPase<br />
would be enhanced by the increase in AG for<br />
ATP breakdown caused by low free Pi. Yet this<br />
is in direct opposition to our observations. If<br />
phosphorylation of the Ca 2+ -ATPase is<br />
necessary to stimulate Ca 2+ transport, this type<br />
of upregulation is not as effective as in normal<br />
conditions because free Pi has been depleted.
Vol. 14, No. 1-4 125<br />
Ca 2+ would remain in the sarcoplasm for a<br />
longer period of time, and relaxation would<br />
slow. It seems that there are two independent<br />
mechanisms responsible for the effect of low [Pi]<br />
on rise and relaxation.<br />
By maintaining muscles in a resting state, our<br />
experimental design avoids introducing factors<br />
arising from constant stimulation, specifically, a<br />
concomitant rise of proton and phosphate levels.<br />
The independent diminution of Pi in this<br />
preparation allows the examination of the relative<br />
contribution of pHi on force generation. In so<br />
doing, we conclude that the relation between<br />
skeletal muscle force and pHi is affected by the<br />
depletion of Pi.<br />
REFERENCES<br />
(1) Adams et al. 1991. AJP, 260 (29), C805 -<br />
C812.<br />
(2) Chase & Kushmerick. 1988. Biophysical<br />
J., 53,935-946.<br />
(3) Dawson et al. 1986. Biophysical J., 49,<br />
286a.<br />
(4) Kushmerick et al. (in press) PNAS.<br />
(5) Wiseman et al. 1990. Biophysical J., 59(2),<br />
517a.<br />
(6) Wiseman et al. 1991. SMRM Abstracts,<br />
10th Annual Mtg., 284.
126 Bulletin of Magnetic Resonance<br />
Introduction<br />
A simple model for the influence of motion<br />
on the NMR line shape.<br />
M. Goldman, T. Tabti, C. Fermon, J.F. Jacquinot and G. Saux.<br />
A problem as old as NMR is that of the influence<br />
of motion on the shape of the NMR absorption<br />
signal, or equivalently the shape of the FID.<br />
Although it is known that motion means<br />
narrowing, only two limiting cases are well<br />
defined.<br />
In the rigid solid, one has a clean theoretical<br />
expression for the FID function, in terms of the<br />
secular part 3TSS of the spin-spin interactions, most<br />
of the time essentially dipolar (1). This is<br />
insufficient to know the FID shape, nor that of its<br />
Fourier transform, the absorption f signal, but one<br />
can compute the first few moments of the latter<br />
and obtain reasonably good value for its frequency<br />
width Ad)0 (1-3).<br />
The other extreme is that of a fast motion whereby<br />
the spin-spin interactions have a vanishing average<br />
value and are modulated at a rate x' 1 which is fast<br />
on the time scale of the solid-state FID decay rate<br />
AGV.<br />
ACOOTC [1]<br />
Service de Physique de FEtat Condense,<br />
C.E.N. SACLAY, 91191 GIF SUR YVETTE<br />
FRANCE.<br />
In that case, relaxation theory predicts that the<br />
transverse magnetization decay is exponential,<br />
with a relaxation time T2 related in a precise way<br />
to the magnitude of the spin-spin interactions and<br />
the nature of the motion (1,4). The absorption line<br />
is then Lorentzian, and its half-width at half<br />
intensity T^ x is of the order of:<br />
2<br />
IC<br />
[2]<br />
Things are much more confuse in the intermediate<br />
case, that is for motions rates for which<br />
Ao)oxc«l. In that domain tjiere exists only<br />
qualitative models of the Anderson-Weiss (5) or of<br />
the Kubo-Tomita type (6), which are admittedly<br />
very crude and whose sole ambition is to provide a<br />
general physical feeling of how the signals are<br />
evolving from slow to fast motion.<br />
In this article, we attempt to correlate the FID<br />
shape under intermediate rate motion to that of the<br />
rigid solid. We use for that purpose a simpleminded<br />
model that is not new: it is the so-called<br />
Strong Collision Model (7). It has been used in<br />
particular for (i.SR studies (8) but in conjunction<br />
with other approximation. This is at variance with<br />
our approach, which treats the rigid lattice FID as<br />
an empirical information. This model is not
Vol. 14, No. 1-4<br />
rigorous either, but it fits remarkably well the<br />
experimental results, as will be shown later.<br />
The theoretical model<br />
The scenario of the Strong Collision Model is the<br />
following. Let us consider a spin system whose<br />
FID signal is Gr(t) when it is rigid, with the usual<br />
normalization:<br />
G,(0) = l [3]<br />
This system undergoes sudden motions at random<br />
time intervals with probability X per unit time.<br />
After each motion, the FID starts anew with the<br />
same shape as initially, but with an initial value<br />
equal to Gr immediately before the jump. In other<br />
words all correlations developed in the system by<br />
its previous evolution are lost, and the only<br />
"memory" left is that of the transverse polarization<br />
at the time of the sudden motion. This results in a<br />
FID signal Gm(t) (where m stands for motion)<br />
which is the weighted average of all possible<br />
random successions of initial rigid-state FBD's.<br />
A pulse being applied at time 0, the possibilities at<br />
time t are:<br />
i) No jump between 0 and t. This corresponds to a<br />
FED signal shape Gr(t) and its probability to occur<br />
is equal to exp(-Ai). Its contribution to Gm(t) is<br />
then:<br />
Gr(f)«p(-X0<br />
ii) One jump only at a time between f, and /, +dtv<br />
The corresponding probability is:<br />
x exp[-X(t - tx)] = X exp(-Xt)dt1<br />
and the signal at time t is then:<br />
mt egrating over the time f, we obtain the<br />
contribution to Gm(t):<br />
127<br />
iii) Two jumps between 0 and /. By an extension of<br />
the preceding argument, the contribution to Gm(t)<br />
is:<br />
JT Gr{t2)Gr(t-tx-t2)dt2<br />
etc...<br />
By taking the Laplace transforms:<br />
we obtain:<br />
that is:<br />
e<br />
= [G(t)exp(-zt)dt<br />
n=0<br />
[4]<br />
[5]<br />
[6]<br />
which is identical with Eq.(18) of ref.(8).<br />
The same result can be obtained much more simply<br />
by noting that after a jump the subsequent FID is<br />
that of the mobile system normalized to the<br />
transverse polarization at the time of the jump.<br />
The two possibilities being no jump between 0 and<br />
/, or at least one jump, we obtain :<br />
[7]<br />
whence the following relations between Laplace<br />
transforms:<br />
[8]<br />
from which Eq.[6] follows.<br />
The advantage of the first treatment is to make<br />
explicit use of the assumption that each partial FID<br />
between jumps has the same shape. The simplest<br />
way of interpreting Eq.[6] is through the<br />
consideration of memory functions (2,3,9). The<br />
memory function K of a function G(t) is defined<br />
through:
128 Bulletin of Magnetic Resonance<br />
[9]<br />
This form yields remarkably simple expressions for<br />
the Laplace transforms. That of the left-hand side<br />
is equal to: z0(z)-G(O) = z8(z)-l, where we<br />
have used Eq.[3], and that of the right-hand side is<br />
equal to -cp(z)9(z), where cp(z) is the Laplace<br />
transform of K(t):<br />
Then, Eq.[9] yields:<br />
z8(z)-l = -(p(z)8(z)<br />
Now, Eq.[6] yields:<br />
1 1<br />
or else:<br />
1<br />
1<br />
[10]<br />
[11]<br />
"•— A* [12]<br />
-z-X<br />
whence, according to Eq.[ll]:<br />
[13]<br />
[14]<br />
This corresponds to the following relations<br />
between memory functions:<br />
[15]<br />
We obtain the remarkably simple result that in the<br />
Strong Collision Model, the rate of decay of the<br />
"memory" shows up simply by an extra<br />
exponential decay of the memory function.<br />
Constraints and limitations to the model<br />
The most questionable assumption underlying the<br />
model is that following a motion, everything is lost<br />
but the transverse magnetization. This will be<br />
discussed in a forthcoming article.<br />
A key assumption in the formulation of the model<br />
is that the shape of the rigid-state FID is not<br />
modified by a motion, which implies that the<br />
motions do not change the form of the spin-spin<br />
interactions. This is only possible if the nuclear<br />
environment of each individual spin is the same<br />
after as before a motion, the only change being<br />
that it is not the same nuclei that occupy the same<br />
relative sites. This case corresponds to atomic or<br />
molecular motion induced by the diffusion of<br />
vacancies at low concentrations in a single crystal.<br />
One must make the distinction, first introduced by<br />
Eisenstadt and Redfield (10), between a jump and<br />
an encounter. A given portion of the crystal<br />
experiences the sudden arrival of a vacancy, which<br />
performs many jumps before disappearing for<br />
away (This is the standard expression. It is evident<br />
that it is the atoms, or molecules, that jump into<br />
vacant sites.). The whole process is sudden, in the<br />
sense that it takes place in a time too short for the<br />
spin system to undergo any significant evolution.<br />
That portion of the crystal was vacancy-free at the<br />
end of the encounter, so that the form of the spinspin<br />
interactions is indeed the same. The motions<br />
referred to in describing the model are in fact<br />
encounters, and not individual jumps.<br />
In a powder, each crystallite obeys the relation<br />
[15], but with different rigid-lattice memory<br />
functions Kr(t). The motion rate X being<br />
independent of crystal orientation, we have on the<br />
average:<br />
and:<br />
[151<br />
[14 1 ]<br />
These relations being linear, it seems that we might<br />
use powder samples, and not solely single crystals<br />
to test the model. Unfortunately, this is not the ,<br />
case because all one can observe in a powder is the<br />
average FID (or average absorption signal). We<br />
have, in place of Eq.[8]:
Vol. 14, No. 1-4<br />
This is a non-linear relation, and we have in<br />
general:<br />
[16]<br />
It is therefore impossible to test the model with<br />
powders.<br />
Experimental study<br />
We have tested the model with a single crystal of<br />
Hexamethylethane (HME) which is the most<br />
symmetrical octane molecule.<br />
Single crystals of HME are obtained from the<br />
liquid state by the Stockbarger method. Due to the<br />
high vapour pressure of the solid, the crystal is<br />
enclosed in a sealed tube.<br />
This molecular crystal experiences a first-order<br />
phase transition at 152.5 K whereby the molecules<br />
undergo rapid reorientation up to the melting point<br />
at 374 K. As a consequence, the inter-molecular<br />
dipolar interactions average to zero, whereas the<br />
average dipolar interactions between protons of<br />
.1<br />
129<br />
different molecules are the same as if each proton<br />
was located at the centre of gravity of its<br />
molecule. These centres form a body centred cubic<br />
structure with a unit cell of size 7.69 A.<br />
NMR relaxation measurements reveal the<br />
existence in this phase of a thermally activated<br />
translational diffusion of the molecules through the<br />
motion of vacancies (11). Analysis of these<br />
measurements yields the value of the average time<br />
x between successive motions, as a function of<br />
temperature (12).<br />
The proton FID's were observed at 91 MHz with a<br />
home-made pulse spectro-meter. We have used<br />
0.5 (is pulses with a repetition time of 0.3 sec. The<br />
FID signal were sampled in 2048 channels with a<br />
time of O.3(xs per channel. The FID signals were<br />
extrapolated back to the origin through a fourthorder<br />
Taylor expansion approxi-mation:<br />
4!<br />
The main test consists in checking whether the<br />
model is actually able to account for the<br />
variation of the FID shape with temperature. This<br />
is done with the help of Eq.[6] for z = i(0. The<br />
results of the fits are given Figure 1.<br />
100 200 300 400 500 600*<br />
time (jxsec)<br />
figure 1: Free Induction Decay Signals at different temperatures: open circles: 283 K, open squares<br />
292.5 K, full circles: 303 K, full squares : 311.3 K. Solid curves are calculated with our model.
130 Bulletin of Magnetic Resonance<br />
The only fit parameter is X,. The values of X are<br />
plotted as a function of 1/T in figure 2, together<br />
with values of 1/T taken from ref.(12) . It is seen<br />
that both X and 1/x obey an Arrhenius law with the<br />
same activation energy.<br />
The ratio (Xi)' 1 ~ 1.6 is compatible with X being<br />
equal to the decay rate of the secular dipolar autocorrelation<br />
function.<br />
2<br />
§<br />
I<br />
1E+06 T<br />
1E+05 •:<br />
1E+04 -:<br />
1E+03 -:<br />
1E+02<br />
3.1 32 3.3 3.4 35 3.6 3.7 3.8<br />
1000 ,„_..<br />
figure 2 : Memory decay rates A. (white circles, present<br />
work) and average jump rates f* (black circles, deduced<br />
fromref. (12)).<br />
Conclusion<br />
The naive strong collision model is remarkably<br />
successful in accounting for the variation of the<br />
FID with motion of intermediate rate. There is<br />
however a surprising and unexplained fact : the<br />
rate X entering eq.[7] is smaller than the average<br />
rate between encounters. This rate X is found<br />
approximately equal to the rate of decay of the<br />
dipolar auto-correlation function, which is well<br />
known from theory to involve several encounters<br />
(13).<br />
References<br />
(1) A. ABRAGAM, The principles of nuclear magnetism,<br />
Oxford University Press, Oxford (1961).<br />
(2) M. MEHRING, High resolution NMR in solids, 2 nd Ed.<br />
Springer-Verlag, Berlin (1983). Appendix G.<br />
(3) A. ABRAGAM and M. GOLDMAN, Nuclear<br />
magnetism: order and disorder, Oxford University Press,<br />
Oxford (1982) chap 1.<br />
(4) N. BLOEMBERGEN, E.M. PURCELL and R.V.<br />
POUND, Phys. Rev. 73 679 (1948).<br />
(5) P. W. ANDERSON and P.R. WEISS, Rev. Mod. Phys.<br />
25, 269 (1953).<br />
(6) R. KUBO and K. TOMTTA, J. Phys. Soc. Japan 9, 888<br />
(1954).<br />
(7) R. KUBO, J. Phys. Soc. Japan 9, 935 (1954).<br />
(8) R. S. HAYANO, Y. J. UEMURA, J. IMAZATO, N.<br />
NISHJDA, T. YAMAZAKI and R. KUBO, Phys. Rev. B20,<br />
850 (1979).<br />
(9) H. MORI, Prog. Theor. Phys. 33,423 (1965).<br />
(10) M. EISENSTADT and A.G. REDFTELD Phys. Rev.<br />
132, 635 (1963).<br />
(11) J.M. CHEZEAU, J. DUFOURQ and J.H. STRANGE,<br />
Molec. Phys. 20, 305 (1971).<br />
(12) A.R. BRICHTER and J.H. STRANGE, Molec. Phys.<br />
37, 181 (1979).<br />
(13) D. WOLF, Phys.Rev. B10,2710 (1974).
Vol. 14, No. 1-4 131<br />
1. Introduction<br />
THE EFFECT ON Ti OF CORRELATED WATER<br />
MOTIONS IN THE POLAR PHASE OF COLEMANITE<br />
Proton magnetic resonance in the ferroelectric<br />
colemanite, CaB3O4(OH)3.H2O, is dominated by<br />
the dynamical motion of the water molecules.<br />
Absorption lineshape [1] and spin lattice<br />
relaxation studies [2,3] have shown that in the non<br />
polar phase (above 270 K) the water molecules<br />
undergo two kinds of reorientational motion; a<br />
180° flipping motion about the H-O-H bisectrix<br />
which merely exchanges the hydrogen positions,<br />
together with a jumping motion in which one of<br />
the water hydrogens takes up one of two possible<br />
sites. This latter motion is also accompanied by a<br />
jump of some of the hydroxyl hydrogens between<br />
two possible sites.<br />
H',<br />
H'84(O —<br />
^N 2.16 A<br />
J. Sun and A. Watton<br />
Dept. of Physics and Astronomy,<br />
University of Victoria<br />
Victoria, BC<br />
Canada V8W 3P6<br />
He*<br />
l g- 1: Projection along the b axis of colemanite<br />
mowing the relative configuration of a pair of<br />
[ater molecules and neighbouring hydroxyl<br />
jrorogens. The smaller water hydrogen -<br />
f?S? yl hydrogen distances, which make a<br />
111 d Dt contribution t0 the relaxation, are<br />
In the polar phase, as illustrated by the<br />
hydrogen positions in Fig. 1, the jumping motion<br />
freezes out accompanied by a slight<br />
rearrangement of the heavy atom network [4]-<br />
The predominant lineshape and relaxation<br />
mechanism in this phase is the remaining 180°<br />
flipping motion of the water groups. This<br />
accounts very well for most of the observed<br />
features of the absorption lineshape and BPP type<br />
temperature dependence of the spin lattice<br />
relaxation time (Fig. 2). However there still<br />
remains a small discrepancy between the<br />
0.13±0.01 s value observed for the Ti minimum<br />
at a Larmor frequency of 30 MHz and the 0.09 s<br />
value predicted by this model.<br />
T (K)<br />
500 300 200 150 125 100<br />
100<br />
i r<br />
0.1<br />
Fig. 2: Temperature dependence of the proton<br />
spin lattice relaxation time in powdered<br />
colemanite at a Larmor frequency of 30 MHz.
132<br />
Although experimental values for the Ti minimum<br />
smaller than model predictions can usually<br />
be accounted for by additional mechanisms not<br />
included in the model, this is clearly not the<br />
situation here and the interpretation is not so<br />
straightforward. It appears that the flipping<br />
mechanism in the theoretical model, assuming this<br />
motion is an appropriate one, is not as effective as<br />
it should be. One possible reason for this reduced<br />
efficiency is that since the water molecules occur<br />
in adjacent pairs their flips are correlated with<br />
each other. That is to say that a flip of one water<br />
molecule is accompanied by a simultaneous flip<br />
of the other one in the pair. Such a correlation<br />
would result in the number of intermolecular<br />
configurations available to the proton pairs being<br />
reduced over those available in uncorrelated<br />
motion. It would seem reasonable then that the<br />
corresponding reduction in the fluctuations of the<br />
intermolecular dipole interaction could lead to a<br />
reduction in relaxation time. We have therefore<br />
extended the theoretical model for water flipping<br />
to include correlations within adjacent water pairs.<br />
The goal was to see whether such correlations<br />
would lead to an increase of the theoretical Ti and<br />
if so whether this could, by itself, account for the<br />
larger value observed.<br />
2. Model Calculation<br />
The spin lattice relaxation time resulting from<br />
the reorientations of the water molecules is given<br />
by [5,6]<br />
£ = f y 4 * 2 X<br />
where the J(co)'s represent the spectral density of<br />
the fluctuating dipolar interaction between protons<br />
i and j, and coo is th e Larmor frequency.<br />
In the formalism which we will adopt the water<br />
dynamics are described by a dimensionless<br />
transition matrix [7]<br />
Y--A-<br />
" 0<br />
where the matrix A has elements Xjj, i^ j, which<br />
are the probabilities/unit time of the interproton<br />
vector ?j- making a transition to r • among the n<br />
possible position vectors (k[[ is chosen so that<br />
Bulletin of Magnetic Resonance<br />
In a similar way the proton dipolar interaction is<br />
described by a geometrical matrix A with<br />
elements given by<br />
l-3cos 2 a;,-<br />
3 3<br />
where r^, r • =1 r^ I, I r • I and ay is the angle between<br />
r- andr: .<br />
X has eigenvalues XJ and is diagonalized with a<br />
matrix T.i.e.<br />
jj<br />
If A is then transformed by the same matrix T,<br />
AT = TAT~ l<br />
the spin lattice relaxation time can be written as<br />
where<br />
M(©0)= X -^<br />
In a 180° flip of a water molecule the<br />
intramolecular proton-proton vector merely<br />
changes sign leaving its contribution to the dipolar<br />
interaction invariant and hence contributing<br />
nothing to the relaxation. As a result only the<br />
intermolecular couplings need be considered, of<br />
which there are two dominant kinds; the<br />
interaction between water pairs and the waterhydroxyl<br />
interaction.<br />
In uncorrelated water motions each water-water<br />
interproton vector can occupy the four positions<br />
illustrated schematically in Fig. 3 and labelled,<br />
arbitrarily 1,2,3 and 4. The transition matrix is<br />
then<br />
-2<br />
0<br />
1<br />
0<br />
-2<br />
1<br />
1<br />
1<br />
-2<br />
1 1 0 - 2<br />
where X is the probability/unit time of any water a<br />
molecule, all assumed dynamically equivalent, jf<br />
making a 180° flip. However, if the water -<br />
motions are correlated in the sense that the water<br />
molecules in each pair reorient in unison then the<br />
transition matrix for the same inter-water proton<br />
vectors becomes<br />
1<br />
1<br />
0
Vol. 14, No. 1-4<br />
"-1 1 0 0"<br />
1 - 1 0 0<br />
0 0 - 1 1<br />
0 0 1-1<br />
Fig. 3: Schematic representation of a pair of<br />
water molecules showing the four positions<br />
occupied by each inter proton vector between<br />
water molecules during uncorrelated motions.<br />
The choice of label numbers is arbitrary.<br />
In either case the water-hydroxyl interproton<br />
vectors have only two values for each waterhydroxyl<br />
group and their corresponding transition<br />
matrix is given by<br />
In fact, for correlated water motions the waterwater<br />
interproton vectors can be partitioned into<br />
two sets {1,2} and {3,4} each of which has a<br />
transition matrix of the simple 2x2 form above.<br />
Applying the diagonalization and transformation<br />
^procedure described previously to these matrices<br />
^produces the following expression for Ti resulting<br />
ifrpm correlated 180° flips of the water molecules,<br />
= 7.0<br />
4coQr<br />
where x=l/\ is the correlation time for the 180°<br />
up motion.<br />
his gives a Ti minimum of 0.10s for a Larmor<br />
fluency of 30 MHz which is greater than that<br />
^"jcorrelated motion by about 10% but still<br />
nilicantly short of the 0.13s observed. It<br />
133<br />
appears that at least the simple correlated model<br />
considered here is inadequate to fully account for<br />
the minimum in Ti and that some other<br />
mechanism is presumably in effect.<br />
In colemanite, as shown by Fig. 1, some<br />
hydroxyl hydrogens are quite close to the water<br />
molecules and there are only two water hydrogens<br />
for every three hydroxyl hydrogens in each<br />
molecular unit. As a result the hydroxyl-water<br />
coupling and the water-water coupling make<br />
comparable contributions to the proton relaxation.<br />
Since the hydroxyl-water contribution is<br />
independent of any correlations in the water<br />
dynamics it seems reasonable that there is not a<br />
large difference in the overall Ti between the<br />
correlated and uncorrelated motions, but it is<br />
gratifying that the change is in the right direction<br />
and is consistent with the reduction in the dipolar<br />
fluctuation discussed earlier.<br />
References<br />
1. A. Watton, H.E. Petch and M.M. Pintar,<br />
Can. J. Phys. 48, 1081 (1970)<br />
2. R. Blinc, M. Brenman, S.R. Miller and J.S.<br />
Waugh, J. Phys. Chem. Solids 23_, 156<br />
(1962)<br />
3. A. Watton, H.E. Petch and M.M. Pintar,<br />
Can. J. Phys. 51, 1005 (1973)<br />
4. F.N. Hainsworth and H.E. Petch, Can. J.<br />
Phys. 44, 2083, (1966)<br />
5. N. Bloembergen, E.M. Purcell and R.V.<br />
Pound; Phys Rev. 73, 679 (1948)<br />
6. A. Abragam, The Principles of Nuclear<br />
Magnetisim (Oxford University Press, New<br />
York, 1961)<br />
7. A. Watton, Phys. Rev. BI7, 945 (1978)
134 Bulletin of Magnetic Resonance<br />
MEASUREMENT OF DEUTERON SPIN RELAXATION<br />
TIMES IN LIQUID CRYSTALS by a BROADBAND<br />
EXCITATION SEQUENCE<br />
I. Introduction<br />
Ronald Y. Dong<br />
Department of Physics and Astronomy, Brandon University<br />
Brandon, Manitoba R7A 6A9<br />
Liquid crystals are composed of flexible organic<br />
molecules and capable of forming different<br />
ordered structures in their mesophases. Nuclear<br />
spin relaxation [1], [2] is a powerful technique<br />
that provides useful information on the<br />
molecular dynamics of liquid crystals. There<br />
are collective director fluctuations, molecular<br />
reorientation and internal rotations in flexible<br />
end chains. Recently internal dynamics<br />
of mesogenic molecules has attracted much attention.<br />
Both theoretical [3], [4] and experimental<br />
[5], [6] studies have been carried out.<br />
Experimentally carbon-13 and deuteron may<br />
be used to probe internal dynamics of flexible<br />
mesogens. In aligned samples of deuterated<br />
liquid crystals, deuterium NMR spectroscopy<br />
yields well-resolved spectral lines having different<br />
quadrupolar splittings for various atomic<br />
sites. These quadrupolar splittings result from<br />
incomplete averaging by anisotropic reorientation<br />
of molecules in the mesophases. For a single<br />
deuteron spin (1=1), there are five independent<br />
spin relaxation times [7]. These are two<br />
spin-lattice relaxation times and three independent<br />
spin-spin relaxation times. Since the deuterium<br />
Zeeman (T\z) and quadrupolar (T\Q)<br />
spin-lattice relaxation times are given by [7],<br />
[8]<br />
rp -l<br />
~3Ji(u;0),<br />
4J2(2w0)<br />
they can be used to separate the two spectral<br />
densities of motion Ji(wo) and J2(2u>o)- Accurate<br />
determination of these spectral parameters<br />
as a function of temperature and the Larmor<br />
frequency (u>o) is necessary for testing various<br />
motional models.<br />
Both T\z and T\Q can be simultaneously measured<br />
by the Jeener-Broekaert (J-B) method [9]<br />
or the selective-inversion method [5]. However a<br />
separate experiment has to be performed for the<br />
deuterons on each labelled site in order to maximize<br />
their quadrupolar order for better signalto-noise<br />
considerations. The J-B sequence has<br />
been used to determine T\z and T\Q in several<br />
nematogens [10]. The pulse sequence was modified<br />
using an additional 45° pulse to produce<br />
the net effect of subtracting the equilibrium Moc<br />
signal from the J-B signal. Here we examine the<br />
modification of the J-B sequence (Table 1) to<br />
produce [11] a broadband excitation sequence<br />
(Table 2) in order to minimize the number of<br />
separate experiments required to give T\z an< l<br />
T\Q for various deuterons on an alkyl chain. Recently<br />
this broadband J-B excitation sequence<br />
was vised to create quadrupolar order with same<br />
efficiency on all the labelled sites in a liquid -j<br />
crystal [12].
Vol. 14,i No. 1-4<br />
\<br />
X<br />
-y<br />
X<br />
-y<br />
X<br />
-y<br />
X<br />
-y<br />
TABLE 1<br />
J-B Sequence with Phase-cycling<br />
Receiver<br />
7 3 Aqu T 2 4>s Aqu T Aqu T Phase<br />
-y y y X +•<br />
0<br />
-X X X y<br />
90<br />
-y y y -X<br />
0<br />
-X X X -y +<br />
y +<br />
90<br />
90<br />
X<br />
0<br />
-y<br />
90<br />
-x +<br />
0<br />
-x +<br />
0<br />
-y<br />
90<br />
X<br />
0<br />
y + 90<br />
y -y -y<br />
90<br />
X -X -X<br />
0<br />
y -y -y<br />
90<br />
)X -X -X<br />
0<br />
-y +<br />
-X<br />
y<br />
x +<br />
2. Experimental<br />
; The deuterium Txz and T\Q were measured on<br />
a home-built superheterodyne coherent pulsed<br />
|INMR spectrometer operating at 15.3 and 46.05<br />
^MHz for deuteron with a Varian 15 in electromagnet<br />
and a 7.1 Tesla Oxford superconductmagnet.<br />
The TT/2 pulse width of ca. 4.5 //s<br />
^produced by an Amplifier Research Model<br />
power amplifier. Pulse control, signal coltiou,<br />
Fourier transformation and data pro-<br />
«mg were performed using a General Electric<br />
1280 computer [10]. Both the J-B sequence and<br />
the broadband J-B excitation sequence (Figure<br />
1) were used with the appropriate phase-cycling<br />
[9] of RF and receiver phases to rid of the unwanted<br />
double quantum coherence (see Tables 1<br />
and 2). Several nematogens (5CB, MBBA and<br />
60 CB) were employed to check the spin relaxation<br />
times obtained by the two different multipulse<br />
sequences.<br />
135<br />
3. Results and Discussion<br />
In figure 2 we show a comparison of the<br />
J-B sequence (2(a)) and the broadband J-B<br />
excitation sequence (2(b)) for a set of partially<br />
relaxed spectra at 15.3 MHz in the nematic<br />
phase of 4-n-pentyl-dn-4'-cyano-2,3,5,6d4-biphenyl<br />
(5CB-di5). Minimal phase correction<br />
was required and the baseline of each spectrum<br />
has been corrected. In figure 2( a) the J-B<br />
sequence was set to maximize the quadrupolar<br />
order of the C4 methylene deuterons. In comparison<br />
with the J-B sequence, we found that<br />
maximum quadrupolar order was created for all<br />
the chain deuterons with r = 5/j.s (or an excitation<br />
bandwith of ca. 75 kHz). In table 3 we<br />
summarize the T\z and T\Q measured at 33.2°C<br />
in 5CB-di5 by the two different J-B sequences.<br />
As seen in this table, their values agree with<br />
each other for all the labelled sites within experimental<br />
errors.<br />
a)<br />
b)<br />
90, 67.5, 45, 45,<br />
9 ^ 4>
136 Bulletin of Magnetic Resonance<br />
a)<br />
1 32 4 5R<br />
R>-c5oll<br />
A 70m ^ h Hijft A, A M A 70m<br />
.50 ft W iiii A A I il 60<br />
A 45<br />
Figure 2 Plots of partially relaxed spectra at 33.2°C and 15.3 MHz. (a) Using the J-B sequence, (b) Using the<br />
broadband J-B excitation sequence.<br />
b)<br />
a)<br />
12 34 5<br />
0 0 DO<br />
6R<br />
R<br />
_ ! , , 1 1 , 1 1 1 r<br />
40000 20000<br />
= 36.2°c<br />
, . 1 1 . r—j-<br />
0 -20000 -40000 Hz<br />
Figure 3 (a) A typical deuterium NMR spectrum of 60CB-d2i showing the peak assignments; (b) A schematic<br />
diagram of a 60CB-d2i molecule.
Vol. 14, No. 1-4 137<br />
>N<br />
CO<br />
c<br />
0)<br />
D<br />
o<br />
20<br />
15 -<br />
10 -<br />
0<br />
o J1<br />
o<br />
oJ2<br />
o<br />
•o<br />
AA J<br />
AA<br />
J 2<br />
AA AA<br />
0<br />
O #)<br />
A Ai<br />
A j^\<br />
t<br />
o<br />
A A A<br />
A A A<br />
40 50 60 70<br />
T(°C)<br />
.We 4 Plots of spectral densities versus temperature in the nematic phase of 60CB-d2i. Open symbols are obtained<br />
^ e broadband J-B excitation sequence, while closed symbols by the J-B sequence. O and A denote data for C4<br />
[a '^6, respectively.<br />
1
138<br />
TABLE 3<br />
Comparison of T\ z and Tiq in ms measured<br />
at 15.3 MHz and 33.2°C<br />
for 5CB-d^<br />
12.6<br />
(13.5)<br />
10.4<br />
(10.1)<br />
c2<br />
26.9<br />
(25.8)<br />
21.5<br />
(21.8)<br />
c3<br />
30.0<br />
(30.7)<br />
22.8<br />
(26.4)<br />
ct<br />
55.4<br />
(48.9)<br />
51.9<br />
(46.1)<br />
159<br />
(137)<br />
107<br />
(104)<br />
Ring<br />
8.5<br />
(9.1)<br />
10.9<br />
(11.4)<br />
* Ti values in parentheses were obtained by the J-B sequence,<br />
while those without parentheses were obtained by the broadband<br />
J-B excitation sequence.<br />
Figure 3 shows the molecular structure of<br />
60CB-d2i and a typical deuterium NMR spectrum<br />
for this mesogen at 15.3 MHz. We have<br />
used both pulse sequences to measure spectral<br />
densities JI(JJJQ) and J2(2u;o) for all the labelled<br />
sites except the ring R!, because of excessive<br />
overlapping of its signal with that from the<br />
methyl (C6) at high temperatures. The agreement<br />
between the two methods are extremely<br />
good. As an example, we show in figure 4<br />
plots of spectral densities versus temperature<br />
for C\ and C&. Since the relaxation times of the<br />
methyl deuterons are much longer than those of<br />
the ring R! deuterons, the overlapped doublet<br />
signals can still be used to determine T\z and<br />
T\Q for the methyl deuterons as long as t is chosen<br />
larger than 40 ms in the pulse sequence.<br />
Finally MBBA-di3 has been studied at 15.3<br />
MHz using the J-B sequence [10]. For comparison<br />
we summarize in Table 4 the results obtained<br />
using the broadband J-B excitation sequence<br />
at 26°C and at 15.3 and 46 MHz. Thus<br />
both Ji(wo) and J2(2u>o) show frequency dependence.<br />
The frequency dependence of J2(2w0) is<br />
weaker; it is negligible for the methine deuteron<br />
(Co). Currently we are analyzing the temperature<br />
and frequency dependences of the measured<br />
spectral parameters in MBBA using models<br />
[3], [4] proposed for flexible mesogens.<br />
In conclusion, the relaxation data can be effectively<br />
obtained in liquid crystals by using the<br />
broadband J-B excitation sequence.<br />
Bulletin of Magnetic Resonance<br />
TABLE 4<br />
Measurements of Ji(u>o) an d ^2(2wo) ins ' for<br />
MBBA-di3 at 26°C (u>o = Larmor frequency<br />
in MHz) using broadband excitation<br />
15.3 46<br />
Co 49.8<br />
40.7<br />
14.6<br />
13.7<br />
1.73<br />
36.2<br />
21.0<br />
9.0<br />
7.5<br />
1.58<br />
15.3 46<br />
Acknowledgments<br />
24.9 27.7<br />
15.1 10.2<br />
6.7 5.4<br />
5.1 4.0<br />
1.05 0.98<br />
The financial support of the Natural Sciences<br />
and Engineering Council of Canada is gratefully<br />
acknowledged. We thank Ms. L. Friesen for her<br />
assistance in carrying out some experiments.<br />
References<br />
1. C.G. Wade, Annu. Rev. Phys. Chem. 28,<br />
47 (1977) and references therein.<br />
2. R.R. Void, in "Nuclear Magnetic Resonance<br />
of Liqud Crystals" J.W. Emsley,<br />
Ed., Reidel, Dordrecht (1985).<br />
3. A. Ferrarini, G.J. Moro and P.L. Nordio,<br />
Liq. Cryst. 8, 593(1990).<br />
4. R.Y. Dong, Phys. Rev. A 43, 4310 (1991).<br />
5. P.A. Beckmann, J.W. Emsley, G.R. Luckhurst<br />
and D.L. Turner, Mol. Phys. 50, 699<br />
(1983); 59, 97 (1986).<br />
6. R.Y. Dong and G.M. Richards, J. Chem.<br />
Soc. Faraday Trans. 88, ni the press.<br />
7. J.P. Jacobsen, H.K. Bildsor and K. Schumburg,<br />
J. Magn. Reson. 23, 153 (1976).<br />
8. S.B. Ahmad, K.J. Packer and J.M. Ramsden,<br />
Mol. Phys. 33, 857 (1977); R.R. Void<br />
and R.L. Void, J. Chem. Phys. 66, 4018<br />
(1977).<br />
9. R.L. Void, W.H. Dickerson and R.R. Void,<br />
J. Magn. Reson. 43, 213 (1981).<br />
10. R.Y. Dong and G.M. Richards, J. Chem.<br />
Soc. Faraday Trans. 84, 1053 (1988) and<br />
references therein.<br />
11. S. Wimperis, J. Magn. Reson. 86, 46<br />
(1990).<br />
12. G.L. Hoatson, J. Magn. Reson. 94, 152<br />
(1991).
Vol. 14, No. 1-4 139<br />
Introduction<br />
CARBON-13 RELAXATION MECHANISMS AND MOTIONAL STUDIES<br />
The carbon-13 relaxation rates of several symmetric<br />
top halomethane species, CH3I [1], CClBr3<br />
[2], CHBr3 [3] and three others that have been<br />
shown to be quasi-symmetric top, CH2Br2 [4,5],<br />
CH2C12 [6], CH2I2[7] were determined. These<br />
results were evaluated at two field strengths,<br />
2.1 and4.7Tesla, at UNC-Wilmington and East<br />
Carolina University, respectively, and at an identical<br />
set of temperatures at each site. With<br />
these data and several theoretical models we<br />
were able to determine, or calculate, the contributions<br />
for all plausible relaxation modes.<br />
IN SELECTED HALOMETHANE MOLECULES<br />
Art A. Rodriguez , Tim Davis<br />
Stokes-Einstein-Debye (SED) [8,9] theory of<br />
rotational diffusion and several variants to characterize<br />
the anisotropic reorientation of spheroids<br />
were also used to investigate for goodness of fit<br />
for hydrodynamically controlled rotational motion.<br />
In the hydrodynamic model, terms called<br />
"stick " and "slip" that attempt to describe<br />
the involvement of probe molecules and the<br />
solvent are exploited. The stick limit is normally<br />
encountered where the solute molecular radius is<br />
'•much larger than that of the solvent, while the<br />
^slip condition is approached as solvent radius<br />
lears or exceeds that of the solute molecule. The<br />
'-extended diffusion model is used to deterinertial<br />
properties of rotational diffusion<br />
Department of Chemistry, East Carolina University<br />
Greenville, NC 27858, USA<br />
and<br />
Lewis E. Nance.<br />
Department of Chemistry, UNC-Wilmington<br />
Wilmington, NC 28403, USA<br />
[4] from J-diffusion to free rotation of the molecule.<br />
Separation of R, 510 from R/ 01 allowed cal-<br />
culation of Tin, Q for Brand 'Jr.Rr for CClBr,,<br />
IBr<br />
C-Br<br />
CH2Br2, and CHBr3 by use of plots of derived<br />
sc 2<br />
T^vs. -s.A(O .<br />
Experimental Section<br />
Separation of Relaxation Mechanisms<br />
Tj values were obtained by the inversion recovery<br />
method using (Mo,cos6, T\), a three parameter<br />
fit for magnetization, M(x), as shown<br />
below.<br />
M{x) = Mo [ 1 - (1 - cos G) exp(-T/ri)] (1)<br />
Partitioning of relaxation rates into contributions<br />
by specific mechanisms was generally<br />
made with the following set of relationships in<br />
mind: When present, contribution to the relaxation<br />
rate by the scalar mechanism of the "second<br />
kind" (SC) is greater at 2.1 T than at 4.7<br />
T while the reverse will be true for chemical<br />
shift anisotropy (CSA). Relaxation is more efficient<br />
at higher temperatures for SC while it is<br />
less so for CSA. Dipole-dipole (DD) relaxation<br />
is more rapid at lower temperatures while that of<br />
spin-rotation (SR) is less. These contributions
140<br />
can be summed as follows where R, tot is the experimental<br />
rate of relaxation:<br />
JL<br />
tot<br />
CSA SC<br />
jCSA 4.<br />
(2)<br />
Integration of the C 13 signals with and without<br />
decoupling (pulse delay set to a full 10 Tt<br />
[10]) for the protonated species, CH3I, CH2I2,<br />
CHBr3, CH2Br2, and CH2C12 permitted determination<br />
of NOE values, r\, and allowed direct<br />
calculation of the H contribution to R, DD [10] by<br />
the following relationship.<br />
The evaluation of %C{C-H) [11] follows immediately<br />
from equation (4).<br />
(3)<br />
XC(C-H) = (y 2 cfHh 2 /2ii 2 r 6 CH)R° D {C-H) (4)<br />
The dipole-dipole contribution to C fromBr<br />
can then be calculated from the relationship below<br />
[12].<br />
•)DD, •>DD<<br />
R" U (C-H)IR\ )U {C-Bf) =<br />
2 (rCH) 6<br />
(5)<br />
A value for the quadrupole coupling constant,<br />
(2ne 2 Qq/h), orT, Q for Br allows a calculation<br />
of xc(C-Br) in equation (6) [13] or (7) [14] respectively.<br />
Rr~-<br />
3 2/+3<br />
40<br />
If<br />
(6)<br />
(7)<br />
R, DD subtracted from R, tot leaves R, 0 *" to be<br />
partitioned among the other contributions. The<br />
correlation time, xc, that was acquired from<br />
R, DD (C-H)canbe used in calculating Ri CSA by<br />
Bulletin of Magnetic Resonance<br />
equation (8) [12] if aper and cpar, the perpendicular<br />
and parallel components of the sheilding<br />
tensor, are known.<br />
(8)<br />
A classic case for relaxation by the CSA<br />
mode was that of CH3I [1]. We found that the<br />
CSA component to relaxation was significant<br />
since Rj tot was much greater at 4.7 T than at 2.1<br />
T and that this value decreased with an increase<br />
in temperature. Using the relationship<br />
CSA _ CSA<br />
(9)<br />
we utilized the fact that the variance in Rj tot<br />
going from 2.1 to 4.7 T must be due to the CSA<br />
mechanism since the SR mechanism is independent<br />
of field strength. A plot of R, tot vs. v 2 at<br />
303K has at the intercept R, tot = R, 1 and at this<br />
point the contribution of R, CSA to the value of<br />
R, tot vanishes. These results show the R, CSA<br />
contributions at 303K of 7.52 x 10" 3 (T, CSA =<br />
133s) at 4.7 T and 1.52 xlO" 3 (T, CSA =659s) at<br />
2.1 T.<br />
The Tc(C- H) values were also employed in<br />
conjunction with the J-extended diffusion theory<br />
computer program by McClung to calculate Xj,<br />
and its reduced form, x}, for symmetric tops<br />
[15,16,17 ]. The equation for R, is as<br />
follows [12]:<br />
Rf t = (2KlkT/h 2 )C 2 effxJ<br />
(10)<br />
A second program written by McClung relates<br />
R, SR to xj. This program requires values<br />
for Ix,Iz,Xj and the spin-rotation tensor components<br />
, Cx and Cz, which are not generally available.<br />
These spin-rotation components can<br />
be approximated by the method of Flygare [18],<br />
which uses the facts that op( n CO) = -259.5 |<br />
ppm and 8( 13 CO) - -182.2 ppm, and that the |
Vol. 14, No. 1-4 141<br />
difference between Adp and A5 for n C0 and<br />
the particular halomethane of interest, for<br />
instance, (Aap = ap(halomethane) - ap(CO),<br />
etc.) are the same . In general one must assume<br />
that Ac is zero. In this manner the relationship<br />
IZCZ ~ IXCX holds if Aa is small compared to<br />
aP.<br />
Scalar relaxation of the "second kind" is expected<br />
to be a prevalent relaxation mechanism in<br />
the Br bearing halomethanes. It is predominantly<br />
the 79 Br isotope instead of 81 Br that makes<br />
the greatest contribution to relaxation of C-13<br />
[19,20]. This is due to the fact that the<br />
(G)/-G)S) term in equation (7) is smaller for<br />
79 Br. Scalar coupling is greater at 2.1 T than at<br />
4.7 T since the Aw term is smaller in the lower<br />
magnetic field. R, sc also increases with temperature.<br />
A plot of Aco 2 vs. T, sc gives<br />
rrQ _<br />
IBr = Jm/b<br />
and<br />
J= \)m)<br />
(11)<br />
(12)<br />
where m and b are the slope and intercept, and<br />
N is the number of Br atoms attached to the<br />
halomethane carbon atom.<br />
Rotational Motion<br />
In the limit of extreme narrowing , the small-step<br />
diffusion theory [13,21,22] predicts the relationship<br />
between xc and the diffusion constants<br />
Dx and Dz in a symmetric top environment by<br />
the following equation [4]<br />
X - °- 3sin 2 6cos 2 e 0.7Ssin 4 6<br />
5DX+D2 2<br />
where 8 is the angle between the reorienting<br />
vector and the unique axis of rotation. This is<br />
the axis with the lowest moment of inertia. For<br />
example, this is the C3 axis in CH3I, but not in<br />
CHBr3 where the C-H vector is at 90° to this<br />
unique axis.<br />
( We have applied the J-extended diffusion<br />
.theory of Gordon [23], as expanded to symmet-<br />
ric tops byMcClung [15, 16], an inertial model,<br />
to generate Xj values using our previously<br />
determined xc(C-H), or xc, parameters. In this<br />
model molecules undergo a period of free rotation<br />
generating angular momentum, then upon<br />
hard molecular collision randomize both magnitude<br />
and direction of this vector. In the smallstep<br />
diffusion limit, Xj « xc • Here the reduced<br />
correlation time, x}, is found to obey the<br />
following equation:<br />
t} = Ty(^-)T«l (14)<br />
In addition, under these conditions, the rotational<br />
diffusion constants, DzandDx of equation (13)<br />
are related to Xy by<br />
Dz = Z-Xj and Dx = 4H (15)<br />
In the inertial model Xy = xc. The period of<br />
time for rotation through 1 radian, X/, as determined<br />
by the equipartition principle is given as<br />
[24]:<br />
y=(~)^ (16)<br />
Thus one is able to follow with this program<br />
by McClung the limits from small-step diffusion<br />
to free rotation where the rotations are inertially<br />
controlled.<br />
At the other extreme, molecular shape instead<br />
of inertial effects may dictate the mode of rotational<br />
diffusion of a molecular species. Extension<br />
of the SED theory to prolate and oblate<br />
spheroids has produced many equations as<br />
boundary conditions for stick and slip rotational<br />
motion are considered for Dx and Dz. For<br />
instance, Perrin [25] was able to show a relationship<br />
be tween the Stokes diffusion constant,<br />
Ds, by solving the Navier-Stokes equation<br />
assuming a stick boundary condition.<br />
Dt = A-Ds = A. kT (17)<br />
The factors fPiX and fPtZ are functions of p = bla<br />
(1 for oblate spheroid). The
142 Bulletin of Magnetic Resonance<br />
average molecular radius is used for r and the<br />
bulk viscosity for T\.<br />
Hu and Zwanzig [26] tackled the rotational diffusion<br />
problem by assessing the fact that the<br />
Perrin values , using the stick boundary condition,<br />
did not fit the experimental work well.<br />
They instead assumed a slip boundary condition<br />
in solving the Navier-Stokes equation. A<br />
separate value for a friction coefficient, £*,<br />
was obtained and tabulated for each axis ratio<br />
for prolate and oblate spheroids. Under this<br />
treatment, motion parallel to the top axis would<br />
experience no tangential stress and would have<br />
the motion of a free rotor.<br />
180<br />
(18)<br />
There would be some solvent displacement possible<br />
about the x-axis for perpendicular rotational<br />
motion.. The expression used here is<br />
Dx = ±LDs<br />
(19)<br />
where in our work /#z = ^*/8.<br />
Gillen [27] and Griffiths [28] note that the<br />
Dx motion in some molecules is diffusionally<br />
controlled and as such may be treated by<br />
the Gierer- Wirtz micro viscosity model [29]. The<br />
parallel motion, Dz, is treated in the slip boundary<br />
conditions, as essentially frictionless rotation.<br />
(20)<br />
The Gierer-Wirtz factor, fGw, is 0.1633 for neat<br />
liquids.<br />
Tanabe [30, 31] has extended the Hynes, Kapral,<br />
Weinberg (HKW) model [32] to include<br />
nonspherical molecules in solution and in neat<br />
form.<br />
(21)<br />
The a factor is zero for Dz but is fitted to a Hu-<br />
Zwanzig coefficient for Dx since even in the<br />
case of slip ((3=0) there is a finite friction coefficient.<br />
Results<br />
In CH2Br2 [4, 5] and CHBr3 [3 ], DD and SC of<br />
the second kind were shown to be the important<br />
relaxation modes. The dipolar contribution to R,<br />
fell off as the temperature was raised, while the<br />
scalar coupling rate increased. There was a<br />
larger value forRj at any particular temperature<br />
at 2.1 T than found at 4.7 T. R, SR and R, CSA were<br />
calculated and found to be negligible . The dipole-<br />
dipole relaxation contribution from Br to<br />
C-13 was calculated by equation (5) and was<br />
much less than the experimental error. In CQBr3<br />
[3 ] the relaxation results were clearly scalar.<br />
Average values of l JcBr and T\Br for CQBr3,<br />
CH2Br2, and CHBr3 are respectively: 75 Hz and<br />
4.3 x 10 6 ; 53 Hz and 4.3 x 10' 6 ; 49.7 Hz and<br />
8.03 x 10" 7 .<br />
The rotational diffusional motion of CH2Br2<br />
andCHBr3 as treated by the several models<br />
above, matches very closely that predicted by<br />
the J-extended diffusion model.<br />
The trend towards faster relaxation rates in<br />
CH3I [1] at 4.7 T compared to 2.1 furnishes positive<br />
evidence for the importance of CSA rather<br />
than SC in this molecular species. SRandDD<br />
are also found to significant.<br />
The approach taken here was that of Gillen [27]<br />
and Griffiths [28]. Motion about the top axis in<br />
CH3I fit closely the Gierer-Wirfz microviscosity<br />
model with an average difference of less than<br />
1% compared to experimental results. The excellent<br />
correlation indicates that the tumbling motion<br />
is hindered by viscous drag while the motion<br />
about the top axis is dominated by inertial<br />
effects.<br />
The modes of relaxation in CH2I2 [7] were as<br />
follows: DD; the dominant mechanism, SR; a<br />
very small contribution, and SC; about 18% at all<br />
temperatures.<br />
The J-extended diffusion model was a good<br />
predictor of rotational motion for both CH2I2<br />
andCH2Cl2[4,6].
Vol. 14, No. 1-4 143<br />
References<br />
1. Manuscript in preparation.<br />
2. Submitted to /. ofMolec. Spectroscopy.<br />
3. Manuscript in preparation.<br />
4. D. N. Dixon and A. A. Rodriquez, J. of<br />
Molec. Liq. 44, 79 (1990).<br />
5. P. B. Simcox, A. A. Rodriguez, L. E, Nance,<br />
The J. of Physical Chem. 96, (1992).<br />
6. A. A. Rodriguez, S. J. H. Chen and M.<br />
Schwartz,/. Magn. Reson. 74, 114 (1987).<br />
7. L.E. Nance, M. R. Nealey, and A. A.<br />
Rodriguez, Magn. Reson. Chem. 28, 11<br />
(1990).<br />
8. R. T. Boere 1 and R. G. Kidd, Annu. Rep.<br />
NMR<br />
Spectrosc, edited by G. A. Webb, Academic<br />
Press, New York, 13, 319 (1982).<br />
9. P. Debye, Polar Molecules, Dover, New<br />
York 1929.<br />
10. M. L. Martin, J.-J. Delpuech and G. L.<br />
Martin, Practical NMR Spectroscopy,<br />
Heyden, London (1980).<br />
11. T. C. Farrar,and E. D. Becker, Pulse and<br />
Fourier Transform NMR, Chap. 4,<br />
Academic Press, New York, (1971).<br />
12. E. D. Becker, High Resolution NMR:<br />
Theory and Applications, 2nd ed., Chap. 8.<br />
Academic Press, New York (1980).<br />
13. W. T. Huntress, J. Chem. Phys. 48, 3524<br />
(1968).<br />
14. A. Abragam, Principles of Nuclear<br />
Magnetism, Chap. 8. Oxford University<br />
Press, Oxford (1961).<br />
15. R. E. D. McClung, /. Chem. Phys. 57, 5478<br />
(1972).<br />
16. R. E. D. McClung, Adv. Mol. Relaxation<br />
Interact. Processes 10, 83 (1977).<br />
17. R. E. D. McClung, J. Chem. Phys. 51, 3842<br />
(1969).<br />
18. W. H. Flygare, J. Chem. Phys. 41,7903<br />
(1964).<br />
19. C. R. Lassigne and E. J. Wells, J. Magn.<br />
Reson. 27, 215 (1977).<br />
.20. T. C. Farrar, S. J. Druck, R. R. Shoup and E.<br />
D. Becker, J. Am. Chem. Soc. 94, 669<br />
(1972).<br />
1- H. Shmizu, J. Chem. Phys. 40, 754 (1964).<br />
22. D. E. Woessner, J. Chem. Phys. 37, 647<br />
(1962).<br />
23. R. G. Gordon,/. Chem. Phys. 44, 1830<br />
(1966).<br />
24. George C. Levy, Topics in Carbon-13 NMR<br />
Spectroscopy, Vol 1, Chapter 3,<br />
Wiley-Interscience, New York, (1974).<br />
25. E. Perrin, /. Phys. Radium, 5,497 (1934).<br />
26. C. M. Hu and R. Zwanzig, J. Chem. Phys.,<br />
60 4354 (1974).<br />
27. K. T. Gillen and J. E. Griffiths, Chem. Phys.<br />
Lett., 17 359 (1972).<br />
28. J. E. Griffiths, Chem. Phys. Lett., 21 354<br />
(1973).<br />
29. A. Gierer and K. Wirtz, Naturforsch, A8<br />
532 (1953).<br />
30. K. Tanabe, Chem. Phys., 31 319 (1978).<br />
31. K. Tanabe and J. Hiraishi, Molec. Phys., 39<br />
493 (1980).<br />
32. J. T. Hynes, R. Kapral and M. Weinberg, /.<br />
Chem. Phys., 69 2725 (1978).
144<br />
An Efficient Large Sample Volume System<br />
for Solid State NMR<br />
Ronald J. Pugmire, Yi Jin Jiang, Mark S. Solum and David M. Grant<br />
Department of Chemistry and Fuels Engineering<br />
University of Utah<br />
Salt Lake City, Utah 84112<br />
ABSTRACT<br />
U.S.A.<br />
An efficient large sample volume<br />
system has been developed to carry out<br />
MAS solid state NMR experiments. The<br />
system components are primarily<br />
zirconia and macor and no background<br />
1 3 C is observed. The stator design<br />
employs separate air bearing and drive<br />
systems and is run using dry air. At a<br />
bearing pressure of about 32 psi, the<br />
rotor can be spun in a stable manner<br />
from less than one hundred Hz (with a<br />
driving pressure of 5 psi) to 4.3 KHz (24<br />
psi driving pressure). This low gas<br />
pressure feature makes the system easy<br />
to operate. The volume of the rotor is<br />
1.8 cm 3 and it can hold 1.1 g of HMB.<br />
The S/N ratio obtained is a factor of 4.6<br />
better than the rotor previously<br />
designed and used in our laboratory<br />
(volume 0.6 cm 3 : 0.28g HMB). This<br />
increased sample size allows us to obtain<br />
the same S/N ratio in a MAS spectrum<br />
with a factor of 21 saving in<br />
spectrometer time. The time saving<br />
achieved with this rotor system is<br />
extremely useful in obtaining data on<br />
biological samples and polymers, and is<br />
especially useful when experiments on<br />
fossil fuels require the use of the Bloch<br />
delay technique. Examples of relevant<br />
applications will be discussed.<br />
Bulletin of Magnetic Resonance<br />
INTRODUCTION<br />
A large MAS spinner system has been<br />
designed to achieve spinning speeds<br />
from less than a hundred Hz to<br />
approximately 4.3 KHz with a relatively<br />
large sample volume of about 1.8 cm^.<br />
The reason for employing a large<br />
sample volume was to improve the S/N<br />
ratio in l3 c NMR experiments. This *is<br />
extremely important for obtaining<br />
quantitative 13 C Bloch decay (BD)<br />
spectra on coal samples with an<br />
acceptable amount of spectrometer<br />
time, due to their reasonably long T\<br />
relaxation times.<br />
In order to avoid a * 3 C background<br />
signal the spinner and stator system<br />
are constructed from zirconia and<br />
macor respectively both of which<br />
contain no carbon.[l] Attention has<br />
also been focused on increasing the S/N<br />
ratio by using a probe electronic circuit<br />
which optimizes the sensitivity of the<br />
l3 C observe channel.[2,3] The result is<br />
a system which gives a S/N ratio for a<br />
* 3 C CP/MAS spectrum of l.lOg sample of<br />
hexamethylbenzene (HMB) that has a<br />
factor of 4.6 improvement over the S/N<br />
ratio of a spectrum taken on our<br />
traditional spinner system, using 0.28g<br />
of HMB and the same spectral<br />
parameters.
t.!-.<br />
* * • •<br />
K<br />
Vol. 14, No. 1-4 145<br />
DESIGN AND CONSTRUCTION OF THE<br />
SPINNER<br />
A cross-sectional drawing of the<br />
spinner system is shown in Figure 1.<br />
There are eighteen flutes in the Up of<br />
the rotor. They are driven by a set of<br />
eight air jets (with a diameter of 0.025<br />
in.) that are mounted on the stator,<br />
along with a set of eight bearing holes<br />
of the same diameter located in the<br />
middle of the stator. These air holes are<br />
spaced at 45' intervals around the<br />
circumference of the stator. The<br />
clearance between the rotor and stator<br />
is approximately 0.002 in. The stator<br />
housing is machined from Kel-F which<br />
has a stepped surface, as shown in<br />
Figure 1, in order to separate the<br />
driving and bearing gases. This design<br />
leads to ease in assembling the system,<br />
and also increases operational safety.<br />
The spinning speed of the rotor as a<br />
function of the driving gas pressure,<br />
with the rotor containing 1.1 Og of HMB<br />
and a bearing gas pressure of 32 psi is<br />
shown in Figure 2. Figures 3 through 5<br />
are a series of photographs of this<br />
system. Due to the high rotational<br />
energy of the spinning system, the<br />
probe must either be in the bore of the<br />
magnet or behind an explosion screen<br />
when in operation.<br />
EXPERIMENTAL RESULTS<br />
Some typical results of data obtained<br />
with the large sample spinning system<br />
are shown below. Figure 7 shows the<br />
CP/MAS 13 C spectrum of 1.1 Og of HMB<br />
taken with only 12 scans and with 21 Hz<br />
line broadening applied. The S/N ratio<br />
of this spectrum is approximately 517.<br />
Figure 8 is a comparison of the spectral<br />
results using this new large rotor and<br />
the standard rotor system previously<br />
used in our laboratory. The size of the<br />
standard rotor (it holds about 0.28g of<br />
HMB) is fairly typical of those<br />
commercially available. In Figure 8a,<br />
CP/MAS spectrum is shown of l.lOg of<br />
HMB consisting of 20 scans with no line<br />
broadening. A S/N ratio of 130 is<br />
obtained. Figure 8b shows a 20 scan<br />
spectrum taken with the smaller<br />
standard rotor system which exhibits a<br />
S/N ratio of 28, again, with no line<br />
broadening. The comparison of Figures<br />
8a and 8b demonstrate that an<br />
improvement in the S/N ratio of about<br />
4.6 can be achieved which provides a<br />
factor of 21 in terms of spectrometer<br />
time. Figure 9 demonstrates that there<br />
are no l^C background signals in BD<br />
and cross polarization (CP) experiments.<br />
This is an important feature in the<br />
quantitative application of CP NMR<br />
experiments. This rotor/stator system<br />
has been used to carefully compare the<br />
BD and CP spectra of the eight coals in<br />
the Argonne Premium Coal Safhple<br />
Bank and these data are presented<br />
elsewhere.[4]<br />
ACKNOWLEDGMENT<br />
This work was supported through the<br />
Advanced Combustion Engineering<br />
Research Center at the University of<br />
Utah and Brigham Young University<br />
which is supported by the NSF, 23<br />
industrial firms and DOE/PETC.<br />
Additional support was provided<br />
through the Consortium for Fossil Fuel<br />
Liquefaction Science at the Univeristy<br />
of Kentucky, Auburn University,<br />
University of Pittsburgh, University of<br />
West Virginia, and the University of<br />
Utah.
1<br />
i<br />
146<br />
REFERENCES<br />
1. Ming Zhang and Gary E. Maciel; J.<br />
Magn. Reson., 85, 156 (1989).<br />
2. Yi Jin Jiang, Ronald J. Pugmire,<br />
and David M. Grant; /. Magn.<br />
Reson.,11, 485 (1987).<br />
3 Yi Jin Jiang, Warner R.<br />
Woolfenden, Anita M. Orendt,<br />
Karen L. Anderson, Ronald J.<br />
Pugmire, and David M. Grant;<br />
Poster MB, 30th ENC, Asilomar,<br />
California, April 1989.<br />
4. Mark S. S,olum, Yi Jin Jiang,<br />
Ronald J. Pugmire, and David M.<br />
Grant, submitted for publication.<br />
Figure 1: Cross-sectional view of the spinner system:<br />
I. Stator, 2. Rotor. 3. Turbine driving jet, 4. Housing<br />
- upper pan, 5. NMR coil terminal plug, 6. Bearing gas<br />
inlet plug, 7. Bearing gas orifice, 8. Light fiber mount.<br />
9. Screw for fixing light fiber. 10. Light fiber.<br />
II. Housing - lower pan, 12. Driving gas inlet plug.<br />
13. Bearing gas exit holes, 14. Mounting nng.<br />
5000<br />
Bulletin of Magnetic Resonance<br />
Driving Gas Pressure (PSD<br />
Figure 2" Plot of the spinning speed of :he<br />
arge volume rotor system as a function ot<br />
he driving pressure used. Bearing pressure<br />
vas 32 psi.<br />
Figure 3: Photograph of the large sample rotor (a) as<br />
compared to the typical rotor (b) and the /mm<br />
rotor supplied by Doty Scientific (c).
if<br />
&<br />
Figure 4: Photograph of the large volume rotor and the<br />
stator.<br />
Figure 5: Photograph of the spinner system.<br />
£' Figure 6: Photograph of the probe containing this<br />
spinning system.<br />
A<br />
B<br />
Figure 7: CP/MAS spectrum of l.lOg of HMB. taken with 12 sc<br />
5ms contact time. Is delay time. 21 Hz line broadening<br />
Spectrum at 25.12 MHz on a Bruker CXP-100.<br />
(b)<br />
CP<br />
2280 scans<br />
lms contact time<br />
1 sec. recycle time<br />
BD<br />
944 scans<br />
180 sec. recycle<br />
delay<br />
CP/MAS spectra of HMB taken at 25.12 MHz:<br />
(t)l«r|e volume (1.8 cm 3 ) system, l.lOg HMB.<br />
20 scans, 3m* contact time. Is delay, no line<br />
broadening, S/N ratio 130; (b) small (0.63 cm 3 )<br />
rotor system. 0.28| HMB, 20 scan*. 3ms contact<br />
time. Is delay, no line broadening. S/N ratio 28.<br />
330 230 130 30<br />
PPM from TMS<br />
-70<br />
Figure »• Comparison of the spectral<br />
Iineshapes obtained for BD and CP<br />
experiments on Pittsburgh #8 Argonne<br />
Premium Coal. A is the spectrum and B is the<br />
background obtained by taking the spectrum<br />
with an empty rotor under identical<br />
conditions to A.<br />
147
148<br />
Bulletin of Magnetic Resonance<br />
Magnetic Resonance Spectroscopic Investigations<br />
of Poly(p-Phenylene Sulfide/Disulfide), PPS/DS<br />
Douglas W. Lowman and David R. Fagerburg<br />
Research Laboratories, Eastman Chemical Company<br />
P. O. Box 1972, Kingsport, TN 37662-5150 USA<br />
Introduction<br />
Recently a new process for the<br />
synthesis of the commercially important<br />
semi-crystalline, engineering<br />
thermoplastic poly(p-phenylene<br />
sulfide/disulfide), PPS/DS, was presented<br />
[1]. This process, based on the reaction of<br />
p-diiodobenzene (DIB) and sulfur at<br />
elevated temperatures (Figure 1),<br />
generates a polymeric material containing<br />
para-substituted aromatic groups<br />
connected by sulfide and disulfide<br />
linkages.<br />
+ S<br />
230 to 300°C<br />
I + S »•<br />
Melt, air bleed<br />
230 to 250°C<br />
• —<br />
Melt, air bleed<br />
Pro-Polymer + '2<br />
PPS/DS<br />
Solid-Stats<br />
Build-Up<br />
(240°C, Nj)<br />
Figure 1. Synthesis of PPS/DS<br />
PPS/DS is not easily characterized by<br />
conventional magnetic resonance<br />
techniques operating under normal<br />
conditions due to its high thermal stability<br />
and solvent resistance. Wade and<br />
coworkers [2] recently reported on their<br />
application of high-temperature solution-<br />
state carbon-13 NMR to commercially<br />
available poly(phenylene sulfide), PPS We<br />
have extended this study by examining<br />
PPS/DS with epr spectroscopy as well as<br />
high-temperature solution-state carbon-13<br />
NMR.<br />
The PPS/DS chemistry is accomplished<br />
under conditions that are in certain<br />
respects considerably less stringent than<br />
those previously employed for other PPS<br />
synthetic routes [3,4]. Free radicals have<br />
been observed in PPS previously by<br />
several workers [5-8] but not under the<br />
conditions of synthesis. Since it was<br />
anticipated that radicals would play an<br />
important role in our chemistry, we<br />
examined this reaction process directly<br />
employing high-temperature epr<br />
spectroscopy under conditions of the<br />
reaction. In this report we present<br />
evidence for sulfur and carbon radical<br />
formation during PPS/DS synthesis.<br />
Experimental<br />
EPR spectra were collected on a Bruker<br />
ER 200D SRC EPR spectrometer operating<br />
on X-band (9.65 GHz) in the general<br />
temperature range of 230 to 300°C and<br />
using 3-mm O.D. glass epr tubes with a 200<br />
gauss sweep width, 100 sec sweep time,<br />
modulation frequency of 100 KHz and<br />
modulation of 40 milligauss. Measurement<br />
of g values was accomplished in a dual<br />
cavity with the resonance of<br />
diphenylpicrylhydrazyl (g value = 2.0037)<br />
as the reference.<br />
PPS/DS synthesis in the epr tube was<br />
accomplished in a manner similar to that
Vol. 14, No. 1-4 149<br />
20 G<br />
Figure 2. EPR spectra of perdeuterated PPS/DS<br />
at (A) 235°C, g(iso) = 2.0073, and (B) 290°C, g =<br />
2.0042.<br />
Figure 3. EPR spectra of melting<br />
perdeuterated PPS/DS<br />
Table 1. Linkage and end group<br />
species found in PPS/DS. Letters<br />
and numbers are used in Figure 4<br />
for resonance assignments.<br />
B<br />
D<br />
G PPS<br />
Table 2. Chemical shift assignments<br />
in ppm for iodo-terminated PPS<br />
(structure shown in Figure 5)<br />
Cl<br />
C2<br />
C3<br />
C4<br />
C5<br />
C6<br />
92.3<br />
138.4<br />
132.7<br />
135.7<br />
135.1<br />
131.8
150<br />
A1<br />
140 138 136 134 132<br />
Chemical Shift, ppm<br />
130 128<br />
Figure 4. Solution-state carbon-13 NMR spectrum of PPS/DS<br />
previously reported [1]. Perdeuterated<br />
PPS/DS (PPS-d4) (degree of polymerization,<br />
DP, of 25) was synthesized in a manner<br />
similar to that previously reported [1].<br />
Iodo-terminated PPS (DP of 9) was prepared<br />
according to the previously reported<br />
method [1] but employing a 53 mole %<br />
excess of DIB.<br />
High-temperature solution-state<br />
carbon-13 NMR employed a hightemperature<br />
10-mm probe system from<br />
Doty Scientific, Inc. (Columbia, SC) on a<br />
JEOL Model GX-400 NMR spectrometer.<br />
Samples were dissolved in N-cyclohexylpyrrolidinone<br />
(CHP) under a blanket of<br />
argon and examined at either 230 or 260°C.<br />
Field/frequency stabilization was<br />
accomplished with glyme-d6 ne ld<br />
concentrically in a 5-mm capillary.<br />
Solid-state NMR spectra were collected<br />
on a Varian XL-300 NMR spectrometer.<br />
Discussion<br />
EPR Spectroscopy. During the<br />
synthesis of PPS/DS in the epr<br />
spectrometer, two distinctly different<br />
resonances are observed (Figure 2). At<br />
235°C a triplet resonance is observed, g<br />
Values for the components of the triplet<br />
resonance are 2.0033 ± 0.0002 (gi), 2.0072 ±<br />
0.0001 (g2) and 2.0113 ± 0.0001 (53). The<br />
Bulletin of Magnetic Resonance<br />
126 124<br />
isotropic g value is 2.0073 This resonance<br />
is assigned to a sulfur-centered radical<br />
cation. The isotropic g value is in excellent<br />
agreement with g values of 2.0079 and<br />
2.0076 reported by Murray and coworkers<br />
[9] for a sulfur-centered radical cation in<br />
two ASF5-doped PPS samples.<br />
Above 280°C, the triplet resonance is<br />
replaced by a singlet resonance with a g<br />
value of 2.0042. This resonance is assigned<br />
to an aryl radical.<br />
It is interesting that the melting<br />
process for PPS/DS can be easily monitored<br />
by epr (Figure 3). Using PPS-d4, two<br />
overlapping triplet resonances are<br />
observed at 230°C. The triplet resonance<br />
arises from hyperfine coupling between<br />
the carbon-centered radical and the<br />
deuterium attached to the carbon. The two<br />
triplet resonances arise from free radicals<br />
being in two different environments. At<br />
230°C, the larger triplet resonance arises<br />
from radicals in the solid phase while the<br />
smaller triplet resonance arises from<br />
radicals in a melted phase. As the<br />
temperature is increased, the solid-phase<br />
component decreases in intensity until it<br />
disappears just above 280°C. The DSC<br />
melting point for this polymer is 279°C.<br />
It is unlikely that the carbon radical<br />
observed here is mechanistically involved<br />
in the synthesis of linear para-substituted<br />
PPS/DS since these radicals would produce
Vol. 14, No. 1-4<br />
J8 3 2<br />
180 160 140 120 100<br />
CHP<br />
Carbonyl<br />
160 140<br />
Chemical Shift, ppm<br />
151<br />
Solid-State NMR Spectrum<br />
Solution-State NMR Spectrum<br />
120 100<br />
Figure 5. Solid-State and Solution-State Carbon-13 NMR<br />
spectra of an Iodo-Terminated PPS<br />
1
152<br />
aromatic substitution other than parasubstitution.<br />
Our PPS/DS chemistry has<br />
been shown to produce exclusively parasubstituted<br />
polymer [10].<br />
NMR Spectroscopv. Solution-state<br />
carbon-13 NMR at 260°C in CHP provides<br />
high-resolution NMR spectra of PPS/DS<br />
that enable an analysis for various linkage<br />
groups, such as sulfide and disulfide, as<br />
well as numerous end groups of the type p-<br />
X-phenyl, where X = H, SH or I. Resonances<br />
for the species in Table 1 are assigned in<br />
the spectrum shown in Figure 4.<br />
Assignments are based on comparison with<br />
spectra for model compounds and analysis<br />
of several PPS/DS spectra. Species A, B and<br />
C were assigned by comparison of spectra<br />
from PPS or PPS/DS with and without .<br />
excess sulfur as well as model compounds<br />
such as diphenyl sulfide and diphenyl<br />
disulfide. Based on the thermal history of<br />
the synthesis, excess sulfur is expected to<br />
be present only as disulfide linkages.<br />
Species D was assigned by comparison with<br />
the spectrum of thianthrene. Species E<br />
assignments are discussed below. Species F<br />
and G were assigned based on previous<br />
assignments [2]. Several resonance<br />
assignments in Figure 4, labelled with "?",<br />
have not been made conclusively.<br />
Complete assignments await synthesis of<br />
appropriate model compounds.<br />
Solution-state carbon-13 NMR of an<br />
iodo-terminated PPS polymer (Species E,<br />
Table 1) at 230°C in CHP clearly shows<br />
sufficient detail to confirm the degree of<br />
polymerization of 9 and to assign all<br />
resonances in the terminal iodophenyl<br />
group (Figure 5, Bottom). The assignments<br />
for these carbons are shown in Table 2. By<br />
comparison, the room temperature solidstate<br />
NMR spectrum of this polymer is not<br />
very informative (Figure 5, top).<br />
Conclusions<br />
PPS/DS does not lend itself to study by<br />
conventional magnetic resonance<br />
techniques operating under normal<br />
conditions. Through the use of several<br />
magnetic resonance techniques, we have<br />
begun to understand some of the<br />
limitations to studying PPS/DS by these<br />
techniques and to enhance our<br />
understanding of the chemistry of this<br />
Bulletin of Magnetic Resonance<br />
unusual polymer. We have presented here<br />
data on the first direct observation of free<br />
radicals generated under conditions of<br />
synthesis of PPS/DS and have expanded the<br />
applicability of the high-temperature<br />
solution-state carbon-13 NMR technique.<br />
Solution-state NMR has provided evidence<br />
for several linkage groups and end groups<br />
in PPS/DS enabling us to better appreciate<br />
the chemistry of this polymer. Room<br />
temperature solid-state NMR has not<br />
proven to be useful in these studies.<br />
References<br />
1. Rule, M.; Fagerburg, D. R.; Watkins, J. J.;<br />
Lawrence, P. B.; Makromol. Chem.,<br />
Rapid Commun. 1991. 12, 221<br />
2. Wade, B.; Abhiraman, A. S.; Wharry, S.;<br />
and Sutherlin, D.; J. Poly. Sci: Pt B:<br />
Polym. Phys., 28 1233 (1990)<br />
3. Fahey, D. R.; Ash, C. E. Macromolecules<br />
1991,24, 4242<br />
4. Lopez, L. C; Wilkes, G. L. J. Macromol.<br />
Sci., Rev. Macromol. Chem. Phys. 1989,<br />
C29, 83<br />
5. Ma, C.-C. M.; Hsiue, L.-T.; Liu, W.-L. J.<br />
Appl. Polm. Sci. 1990,39, 1399<br />
6. Hill, D. J. T.; Hunter, D. S.; Lewis, D. A.;<br />
O'Donnell, J. H.; Pomery, P. J. Radiat.<br />
Phys. Chem. 1990, 36, 559<br />
7. Kreja, L.; Rozploch, F.; Warszawski, A.<br />
Angew. Makromol. Chem. 1988,160, 163<br />
8. Kispert, L. D.; Files, L. A.; Frommer, J. E.;<br />
Shacklette, L. W.; Chance, R. R. J. Chem.<br />
Phys. 1983,57, 4858<br />
9. Murray, D. P.; Kispert, L. D.; Frommer, J.<br />
E. J. Chem. Phys. 1985, 83, 3681<br />
10. Rule, M.; Fagerburg, D. R.; Watkins, J.<br />
J.; Lawrence, P. B.; Zimmerman, R. L.;<br />
Cloyd, J. D.; Makromol. Chem.,<br />
Macromol. Symp. 1992,54/55, 233
Vol. 14, No. 1-4 153<br />
Application of 2-D HETCOR NMR to Investigate Polymer Blend Heterogeneity<br />
Introduction:<br />
Most NMR techniques for determining<br />
domain structures in polymeric systems are<br />
based upon the classic Goldman-Shen<br />
experiment [1]. Domain sizes are calculated<br />
from the time for spin diffusion to transfer<br />
magnetization from one region of the sample<br />
to another. These experiments generally<br />
require resolution of the individual<br />
components in the proton spectrum,<br />
although recent experiments [2] demonstrate<br />
that, in some cases where there is<br />
inadequate proton chemical shift resolution,<br />
it is possible to follow the evolution of spin<br />
diffusion when the constituents have significant<br />
differences in proton lineshapes.<br />
An alternative approach to analyzing<br />
domain structures entails application of<br />
heteronuclear ^C- 1 !! NMR. Of the several<br />
techniques introduced to measure heteronuclear<br />
correlated spectra in the solid state,<br />
a particularly effective method has been proposed<br />
by Burum and Bielecki [3]. In the<br />
basic 2-D experiment, crosspeaks occur primarily<br />
for carbon-proton distances of less<br />
than ~3 A. By incorporation of a spin<br />
diffusion evolution time, crosspeak intensities<br />
reflect longer range couplings, thus<br />
I &.. 1- Goldman, M.; Shen, L. Phys. Rev. 1966,<br />
^ 144,^21. J<br />
U. 2 - C^P^ll, G.C.; VanderHart, D.L. J.<br />
Magn. Reson. 1992,96,69-93.<br />
\ 3 - Burum, D.P.; Bielecki, A. J. Magn.<br />
Reson. 1991,94,645-652.<br />
S. Kaplan<br />
Xerox "Webster Research Center<br />
800 Phillips Road 1004-39D<br />
Webster, NY 14580, USA<br />
enabling conformational and domain size<br />
analyses [4]. This latter experiment consists<br />
(Fig. 1) of an evolution period during which<br />
homo- and heteronuclear interactions are<br />
suppressed by simultaneous application of<br />
BLEW-24 ( J H) and BB-24 ( 13 C) pulse<br />
sequences, a waiting period without rf<br />
during which exchange of magnetization<br />
(spin diffusion) takes place, a mixing period<br />
for isotropic cross polarization transfer of<br />
proton magnetization to carbons utilizing<br />
WIM-24 sequences on both nuclei, and<br />
finally, carbon acquisition with proton<br />
decoupling. In this paper we examine the<br />
90° 63° 90° 90°<br />
BLEW-24 WIM-24<br />
13C BB-24 WIM-24<br />
Figure 1. The HETCOR Experiment<br />
cw<br />
Decoupling<br />
applicability of the HETCOR technique to<br />
investigate domain sizes in a two component<br />
blend.<br />
Results and Discussion:<br />
We have chosen for the present study a blend<br />
of an aromatic diamine, N,N'-diphenyl-N-<br />
4. Simpson, J.H.; Ruggeri, G.; Rice, D.M.;<br />
Karasz, F.E., submitted.
154<br />
N'-bis(3-methylphenyl)-[l,l'-biphenyl]-4,4'diamine<br />
(TPD) in a bisphenol A polycarbonate<br />
matrix. Pure TPD is highly crystal-<br />
TPD<br />
line, with a melting point of Tm = 167 °C and<br />
a glass transition temperature of Tg=63 °C<br />
[5]. A well mixed 50/50 blend with polycarbonate<br />
(Tg=137 °C) cast from methylene<br />
chloride is amorphous and shows a single Tg<br />
of ~89 °C. Since the aromatic region of the<br />
13 C spectrum of the blend has severe overlap,<br />
we have focused our analysis on the aliphatic<br />
region, where the individual component<br />
methyl resonances are resolved.<br />
Figure 2 shows contour plots of the<br />
carbon aliphatic region in the two dimensional<br />
HETCOR spectra as a function of the<br />
spin diffusion mixing time. The 13 C peaks at<br />
22 ppm and 32 ppm are from TPD and polycarbonate<br />
methyl carbons, respectively.<br />
Crosspeaks in the 20 us mixing time plot are<br />
due primarily to short range directly bonded<br />
carbon-proton couplings. However, with<br />
increasing mixing times, longer range correlations,<br />
e.g., between the methyl carbons<br />
and aromatic protons, intensify. Ultimately,<br />
for very long spin diffusion times, the<br />
relative signal intensities for the contours<br />
associated with a particular carbon will<br />
reflect quantitatively the chemical shift distribution<br />
of protons that are within an<br />
effective range of spin diffusion from the protons<br />
directly bonded to that carbon. Figure 3<br />
shows the volume integral fraction of<br />
aliphatic protons for the polycarbonate and<br />
TPD methyl carbons as a function of spin<br />
diffusion time. Both curves asymptote to an<br />
5. Prest, W.M., unpublished data.<br />
Bulletin of Magnetic Resonance<br />
40 30 20 40 30 20<br />
ppm (13Q<br />
ppm OH)<br />
Figure 2. HETCOR contour plots of the aliphatic<br />
carbon region of a 50/50 (wt/wt) blend<br />
as a function of spin diffusion time.<br />
aliphatic volume fraction of 0.3 (point c),<br />
which corresponds exactly to the fraction of<br />
protons in the entire sample that are from<br />
methyl groups. The results are very<br />
different from measurements on a physical<br />
mixture of the same components, where<br />
separate asymptotes, corresponding to the<br />
individual aliphatic proton fractions, are<br />
observed for each component (polycarbonate<br />
at point a and TPD at point b). Also shown<br />
in Figure 2 is the difference response<br />
between these two curves. The short term<br />
initial rise can be attributed to<br />
intramolecular and the long term decay to<br />
intermolecular proton spin diffusion. From<br />
the ratio of these rates (-10) and the known<br />
intramolecular proton-proton distances (0.3<br />
nm) the intermolecular distances can be
Vol. 14, No. 1-4 155<br />
Aliphatic Fraction<br />
1.0<br />
0.8'<br />
0.6<br />
0.4 -<br />
: •fW*t<br />
0.2 r.<br />
•* * •<br />
—O— PC Icompatible 1<br />
-••-TPD J Blend i<br />
D PC 1 Physical z .<br />
• TPD J Blend j<br />
• 4i<br />
m :<br />
(a)<br />
(c)<br />
.. Difference (X5) =<br />
(b)<br />
0<br />
0 0.5 1.0 1.5 2.0<br />
Spin Diffusion Time (ms)<br />
"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<br />
Figure 3. TPD and polycarbonate methyl<br />
carbon correlations with protons.<br />
estimated to be about 1 nm (0.3XV10),<br />
indicative of intimate molecular level<br />
mixing. This treatment of the data permits<br />
estimation of interdomain separation<br />
without knowledge of the spin diffusion<br />
constant, which is only assumed to be equal<br />
in both domains.<br />
An alternative approach to analyzing the<br />
results of this experiment is shown in Figure<br />
4. Plotted here is a measure of the similarity<br />
(mean square difference) of the proton slices<br />
for each of the methyl carbon resonances as a<br />
function of spin diffusion time. This plot is a<br />
direct measure of the diffusion of spin order<br />
between the two blend constituents. In spite<br />
of scatter, the best fit for the data is a single<br />
exponential with a decay rate of 350 ms.<br />
Employing the standard equation for<br />
determining domain sizes from spin diffusion<br />
i- times, r = (nDt)- 1/2 , where n represents the<br />
domain dimensionality, t is the measured<br />
if', spin diffusion time, and D is the spin diffusion<br />
constant, typically 5X10" 12 cm2 sec" 1 ,<br />
• domain sizes in the range of 0.8-1.4 nm are<br />
• estimated.<br />
0.2 r<br />
0 -<br />
0 0.5 1.0 1.5 2.0<br />
Spin Diffusion Time (ms)<br />
Figure 4. Intermolecular correlations between<br />
TPD & polycarbonate methyl protons.<br />
Heating the blend overnight at 110 °C induced<br />
some phase separation, as evidenced<br />
by a change in appearance from clear to<br />
opaque. HETCOR diffusion spectra of this<br />
sample showed negligible differences from<br />
the data of the clear films. We conclude that<br />
a very small amount of phase separation<br />
occurred, e.g., near impurities or on the<br />
surface only.<br />
In summary, it is shown that the two<br />
dimensional heteronuclear correlated spin<br />
diffusion experiment is capable of extending<br />
the range of applicability of domain size<br />
measurements to blends comprised of<br />
components that are structurally very<br />
similar.<br />
Acknowledgements:<br />
The author wishes to thank J. Yanus for<br />
supplying the blend sample used for the<br />
current study and D. Rice (Univ. of<br />
Massachusetts) and D. Burum (Bruker<br />
Instruments, Inc.) for helpful discussions on<br />
the HETCOR experiment.
156 Bulletin of Magnetic Resonance<br />
NEW HIGH RESOLUTION NMR STUDIES<br />
IN<br />
POLYCRYSTALLINE TETRACYANOQUINODIMETHANE<br />
M.T. Nunes<br />
ICTPOL/CFMC-INIC, Av.Prof. Gama Pinto 2, 1699 Lisboa Codex, Portugal;<br />
Dep. Quimica, ICEN, LNETI 2686 Sacavem Codex Portugal<br />
A. Vainrub<br />
Institute of Chemical and Biophysics, Estonian Academy of Sciences,<br />
200001 Tallinn, Estonia<br />
M. Ribet, F. Rachdi, P. Bernier<br />
Groupe de Dynamique des Phases Condensees, U.S.T.L., 34095 Montpellier<br />
Cedex 05, France<br />
M. Almeida<br />
Dep. Quimica, ICEN, LNETI 2686 Sacavem Codex, Portugal<br />
and<br />
G.Feio<br />
ICTPOL/CFMC-INIC,Av.Prof. Gama Pinto 2,1699 Lisboa Codex,Portugal<br />
Introduction<br />
Tetracyanoquinodimethane (TCNQ) is<br />
a well-known electron acceptor<br />
that has been widely used to<br />
prepare charge transfer complexes,<br />
some of them exhibiting very high<br />
electrical conductivity; locally<br />
resolved "C Knight shifts [1] were<br />
measured, accordingly [2].<br />
NC<br />
NC<br />
CH==CH<br />
CH=CH<br />
TCNQ<br />
\c==c.<br />
CN<br />
CN<br />
In a static magnetic field of<br />
4.7T, high resolution NMR studies<br />
were performed on neutral TCNQ<br />
using different polycrystalline<br />
samples: natural isotopic<br />
abundant, selectively enriched on<br />
13 15<br />
C isotope content and N enriched<br />
[3]. Consequently, a complete<br />
assignment of the 13 C resonances<br />
was reported; the principal<br />
components of the chemical<br />
shielding tensor, the shielding<br />
anisotropy and the shielding<br />
asymmetry factor were also<br />
obtained for the 13 C nuclei in the<br />
fragments C-(CN)2, by graphical<br />
analysis and by a computer fit of<br />
the line intensities of CP/MAS<br />
spectra [4]. However, large<br />
uncertainties were reported on CN<br />
groups data. 14 N quadrupole effects<br />
were observed on 13 C spectra of CN<br />
groups and the "N quadrupole<br />
coupling constant was obtained (*"<br />
3.84±0.12MHz). Using different<br />
experimental conditions (like a<br />
static magnetic field of 7.0T) we<br />
highlight here<br />
cyano groups.<br />
new<br />
3<br />
C data on
Vol. 14, No. 1-4 157<br />
RESULTS and DISCUSSION<br />
When applying the numerical method<br />
[4] to the spectra obtained at<br />
4.7T, two different data sets were<br />
found for the principal components<br />
of the 13 C shielding tensor in TCNQ<br />
cyano groups [ 3 ]; only one of the<br />
corresponding asymmetry factors<br />
(t]= 0 and 0.4) was characteristic<br />
of an axially symmetric system.<br />
Aiming to a clarification of this<br />
problem, the improvement of the<br />
experimental data for the<br />
application of the numerical<br />
method seems to be the next step;<br />
in particular, the acquisition of<br />
13 C spectra displaying an higher<br />
number of spinning sidebands and a<br />
better signal/noise is required.<br />
These conditions are fulfilled by<br />
running the spectra at higher<br />
static magnetic field but<br />
selecting the same MAS rate and a<br />
longer recovering delay (600s);<br />
Figure 1 shows typical 13 C CP/MAS<br />
high resolution NMR spectra of<br />
selectively enriched TCNQ to 99%<br />
13 C isotope content on CN groups.<br />
Again, two data sets are obtained<br />
by a computer fit of the<br />
intensities of the lines, one of<br />
them corresponding to an axial<br />
symmetry for the cyano groups, in<br />
close agreement with that obtained<br />
at 4.7T [3 ] :<br />
ffll<br />
ppm<br />
228<br />
210<br />
a22<br />
ppm<br />
190<br />
208<br />
ppm<br />
-80<br />
-80<br />
Aa<br />
ppm<br />
-289<br />
-289<br />
0.2<br />
0.0<br />
However, regarding the other set,<br />
a much lower deviation from axial<br />
symmetry is now obtained (axx-o22<br />
equal to 38ppm instead of 69ppm).<br />
Indeed, Clayden and co-workers<br />
already pointed out that similar<br />
simulated MAS spectra<br />
Too 1«5 us ito iio M o -So<br />
FIG.l. 13 C CP/MAS high resolution<br />
NMR spectra of TCNQ cyano groups<br />
obtained at 75.47 (on the top) and<br />
50.13 MHz (on the bottom) with the<br />
same spinning rate: 3.01 kHz.<br />
are obtained when the chemical<br />
shift tensor is axial or nearaxial<br />
(ii0.2. On the<br />
light of these conclusions the<br />
present results are definitely<br />
acceptable.<br />
For the acquisition of the<br />
spectra previously reported [ 3 ],<br />
too short relaxation delays were<br />
used possibly inducing a distorted<br />
lineshape for the sidebands<br />
envelope; anisotropic molecular<br />
motion would influence distinctly<br />
the relaxation times of magnetic<br />
equivalent 13 C nuclei on TCNQ<br />
molecules perpendicular and
158<br />
parallel to the static magnetic<br />
field. In fact, if 1.5s is now<br />
used instead of 600s for that<br />
duration period, at least three<br />
fits are obtained (depending on<br />
the sidebands used) with much<br />
higher residual values; comparing<br />
the data sets in this case, large<br />
uncertainties are observed for all<br />
the principal components of the<br />
shielding tensor.<br />
The study of 13 C TCNQ relaxation<br />
in a wide temperature range are<br />
now in progress and will certainly<br />
contribute to the understanding of<br />
these results; also, additional<br />
insight on the role of TCNQ on<br />
charge transfer complexes (like<br />
DMTM(TCNQ)2 [2]) will be obtained.<br />
To obtain the 14 N quadrupole<br />
coupling constant from X3 C CP/MAS<br />
spectra, run at 7.0T, the<br />
procedure previously described was<br />
used [3]; a C-N dipolar splitting<br />
close to 230 Hz should be measured<br />
now, which is in fact the case.<br />
References<br />
1 M.Mehring and J.Spengler,<br />
Phys.Rev.Lett. ,53., 2441 (1984)<br />
2 for a recent nmr study on these<br />
materials see: F.Rachdi, T.Nunes,<br />
M.Ribet, P.Bernier, M.Helmle,<br />
M.Mehring and M.Almeida,<br />
Phys.Rev.B, 45(14), 8134 (1992)<br />
3 T.Nunes, A.Vainrub, M.Ribet,<br />
F.Rachdi, P.Bernier and M.Almeida,<br />
J.Chem.Phys.,96(11) , 8021 (1992)<br />
4 J.Herzfeld and A.E.Berger,<br />
J.Chem.Phys., 73_, 6021 (1980)<br />
5 N.J.Clayden, C.M.Dobson, L.Lian<br />
and D.J.Smith,<br />
J.Magn.Reson.,69/476 (1986)<br />
Bulletin of Magnetic Resonance
Vol. 14, No. 1-4 159<br />
USE OF NMR RELAXATION<br />
MEASUREMENTS TO DERIVE THE<br />
BINDING SITE OF PLASTOCYANIN<br />
IN COMPLEXES WITH<br />
CYTOCHROME-F AND C<br />
Sandeep Modi 1 , Ewen McLaughlin 1 , Derek S. Bendall 1 ,<br />
S. He 2 and J.C. Gray 2<br />
Departments of Biochemistry 1 and Plant Sciences 2<br />
University of Cambridge, Cambridge, CB2 1QW<br />
England (U.K.)<br />
1 Introduction<br />
Plastocyanin (PC) is a small (Mr 10<br />
500), 'blue' copper protein which<br />
transfers electrons from cytochrome<br />
/ to the primary electron<br />
donor of photosystem I (P700) in<br />
the photosynthetic electron transport<br />
chain. It reacts rapidly with<br />
cytochrome / in vitro and also<br />
with several other cytochromes,<br />
although somewhat more slowly<br />
[1].. The crystal structures of both<br />
oxidized and reduced poplar<br />
plastocyanin have been determined<br />
and show that redox changes<br />
cause only small changes at the Cu<br />
site, leaving the structure of the<br />
rest of the molecule essentially<br />
unchanged [2,3]. The Cu atom and<br />
its ligands (His-37, Cys-84, His-87<br />
and Met-92) are located in a<br />
hydrophobic pocket near one end<br />
of the molecule (the "northern"<br />
end) such that only the imidazole<br />
ring of His-87 (the northern<br />
histidine) is accessible to solvent<br />
(Fig 1).<br />
Studies with small molecules,<br />
such as [Fe(CN)6] 3 " and<br />
[Co(phen)3]3+, have identified two<br />
reaction sites on plastocyanin; one<br />
close to the copper ligand His-87<br />
at the northern hydrophobic patch<br />
and the other close to the more<br />
remote Tyr-83 at the eastern<br />
acidic patch [4-8]. Chemical modification<br />
of the acidic amino acid<br />
residues, inhibitory effects of<br />
small molecules and ionic strength
160 Bulletin of Magnetic Resonance<br />
effects on electron transfer<br />
strongly suggest that cytochrome /<br />
binds at the remote eastern site<br />
[9,10]. However the pathway of<br />
electron transfer from cytochrome<br />
/ to the copper site in plastocyanin<br />
C<br />
L12 H87<br />
E60<br />
Fig. 1 : Structure of Plastocyanin. The Cu<br />
ligands and the side-chains of Tyr83 and the<br />
residues of the eastern acidic patch are shown<br />
here with the coordinates of poplar<br />
plastocyanin from the Brookhaven Database,<br />
except for substitution of Glu45 (as in pea<br />
plastocyanin) for Ser45. ^-Strands are<br />
shaded.<br />
is not clear. One possibility is that<br />
cytochrome / binds to the negative<br />
charges of the eastern acidic patch<br />
and donates electrons via a<br />
tunneling pathway starting at<br />
Tyr-83 [5,11]. However, the<br />
distance from Tyr-83 to the<br />
copper ion is approximately 12A,<br />
whereas at the northern site the<br />
copper ion is only 6A from the<br />
surface of the molecule [11].<br />
The recent development of<br />
expression systems for the small<br />
blue copper protein, plastocyanin<br />
[12-15], has provided a valuable<br />
tool for study of the molecular<br />
details of its interaction with its<br />
native reaction partners (cytochrome<br />
/ and photosystem I in the<br />
photosynthetic electron transport<br />
chain). To examine the pathway of<br />
electron transfer from cytochrome<br />
/ to plastocyanin we have altered<br />
Tyr-83 to Phe-83 and Leu-83 by<br />
site-directed mutagenesis of the<br />
pea plastocyanin gene [13-14].<br />
Measurements of binding constants<br />
and electron transfer rates<br />
indicate, that Tyr-83 not only<br />
forms part of the main route of<br />
electron transfer from cytochrome<br />
/ to plastocyanin but is also involved<br />
in binding to cytochrome /.<br />
Nuclear magnetic resonance<br />
spectroscopy has been established<br />
as a very convenient and effective<br />
technique for structural studies of<br />
proteins. For haemproteins, the<br />
characteristic hyperfine-shifted<br />
NMR spectrum of a paramagnetic<br />
haemprotein carries the signature<br />
of the electronic and structural<br />
properties of the haem group.<br />
Measurements of spin - lattice
Vol. 14, No. 1-4 161<br />
relaxation (Tj) and spin-spin<br />
relaxation (T2) times provides<br />
useful methods for the<br />
determination of dissociation<br />
constants and distances of various<br />
nuclei from the paramagnetic<br />
centre in a protein-protein<br />
complex. Relaxation measurements<br />
were carried out to get<br />
information about the relative<br />
dispositions of the two proteins<br />
(PC and cytochromes) in their<br />
complexes.<br />
2 Assignment of<br />
Proton NMR resonances<br />
for pea<br />
Plastocyanin<br />
Proton NMR measurements were<br />
carried out on a Bruker AM 500-<br />
MHz FT-NMR spectrometer at<br />
300K in 50 raM phosphate buffer<br />
(pH 6.0). Proton chemical shifts<br />
were referred to a proton signal of<br />
dioxan as a reference at 3.74 ppm.<br />
Proton NMR resonances for pea PC<br />
were assigned using 2D-NMR<br />
spectroscopy at 300K (pH = 6.0).<br />
The conformation of the Phe-83<br />
plastocyanin was examined by<br />
^-NMR. A ID spectrum in H2O,<br />
compared with that of the wildtype<br />
protein, clearly demonstrated<br />
the disappearance of the amide<br />
resonance of Tyr-83 at 9.40 ppm,<br />
and the appearance of a new<br />
resonance at 9.37 ppm which is<br />
likely to be that of Phe-83 [13],<br />
although positive identification<br />
must await further 2D<br />
experiments. Overall the close<br />
similarity between the spectra of<br />
the mutant and wild-type proteins<br />
indicated that the replacement of<br />
Tyr-83 with Phe-83 had not led to<br />
major conformational changes<br />
throughout the protein. Insufficient<br />
protein of the Leu-83<br />
mutant was available for NMR<br />
analysis, so its conformation was<br />
examined by CD between 190 and<br />
260 nm. The spectrum obtained<br />
for the oxidized protein was<br />
essentially identical to that of the<br />
wild-type. The Phe-83 mutant<br />
protein also gave a closely similar<br />
spectrum. These results confirm<br />
that the three proteins had<br />
identical gross conformations.<br />
3 Kinetics<br />
surements<br />
Mea-<br />
Electron transfer from reduced<br />
cytochrome to oxidized plastocyanin<br />
was monitored at 422 nm<br />
with an Applied Photophysics<br />
stopped-flow spectrophotometer<br />
(SF.17MV). The rate of binding of<br />
plastocyanin and cytochrome was<br />
measured by following the
162 Bulletin of Magnetic Resonance<br />
increase in absorbance of oxidized<br />
cytochrome at 410 nm in the<br />
stopped-flow spectrophotometer.<br />
K& was determined by taking<br />
advantage of the increased absorbance<br />
of the Soret band of<br />
cytochrome on binding to<br />
plastocyanin<br />
The results reported in Table I<br />
[13,14] demonstrate convincingly<br />
that Tyr-83 of plastocyanin is part<br />
of the main tunneling pathway for<br />
the electron between the haem<br />
rings of both cytochrome c and<br />
cytochrome / and the Cu atom of<br />
plastocyanin. A leucine residue in<br />
this position is much less effective<br />
and it seems likely that the<br />
facilitation of electron transfer by<br />
tyrosine or phenylalanine is due<br />
to the aromatic nature of the ring,<br />
as has been proposed in other<br />
proteins. A striking difference<br />
between the kinetics of reduction<br />
of plastocyanin by the two<br />
cytochromes is that the wild-type<br />
protein and the Phe-83 mutant<br />
behave identically towards<br />
cytochrome c, but not towards<br />
cytochrome /. In the latter case<br />
the rate of reduction of the<br />
mutant protein is about seven<br />
times slower, a difference that can<br />
be ascribed entirely to weaker<br />
Table I<br />
Kinetic parameters for reduction of pea plastocyanin by cytochromes c and/<br />
Plastocyanin<br />
Cytochrome<br />
Wild type<br />
Phe-83<br />
Leu-83<br />
k2 (x 10-6)<br />
(M-V 1 )<br />
c as donor<br />
3.26<br />
3.28<br />
0.421<br />
Cytochrome/as donor<br />
Wild type<br />
Phe-83<br />
Leu-83<br />
40.6<br />
5.43<br />
0.955<br />
*A<br />
(M-l)<br />
1253<br />
1295<br />
1260<br />
9890<br />
1270<br />
968<br />
•*a(xlO-6)<br />
(MrV 1 )<br />
20.0<br />
22.7<br />
21.7<br />
43.5<br />
5.86<br />
1.27<br />
*.a(xl0< *) k( (x 10-3)<br />
(s- 1 )<br />
16.0<br />
17.5<br />
17.2<br />
4.40<br />
4.61<br />
1.31<br />
(s- 1 )<br />
3.11<br />
2.96<br />
0.340<br />
Values are given as mean + standard deviation. Data for cytochrome/and c are from<br />
[13].and [14] respectively.<br />
62<br />
58<br />
4.0
Vol. 14, No. 1-4 163<br />
binding. We therefore predict that<br />
when the structure of cytochrome<br />
/ becomes known it will reveal a<br />
surface residue in the region of<br />
the exposed haem edge which is<br />
capable of hydrogen bonding to<br />
the -OH of Tyr-83.<br />
4 NMR relaxation<br />
(Ti) measurements<br />
Protein concentrations were<br />
determined from the following<br />
absorption coefficients: reduced<br />
horse heart cytochrome c, £550nm<br />
= 2.76 x 10 4 M-icm" 1 ; reduced<br />
oil-seed rape cytochrome /,<br />
£554nm = 2.6 x 10 4 M-icm-l;<br />
oxidised plastocyanin, £597nm =<br />
4.7 x 103 M-lcm-1. Proton NMR<br />
relaxation measurements were<br />
carried out on a Bruker AM 500-<br />
MHz FT NMR spectrometer at<br />
300K. The samples were in 0.01 M<br />
phosphate buffer (containing 90<br />
raM NaCl) at pH 6.0 (volume, 0.4<br />
ml). Proton NMR spectra of Cd-PC<br />
were obtained by accumulation of<br />
about 160 transients at 16K data<br />
points in quadrature mode. To<br />
facilitate relaxation measurements,<br />
a redox-inactive form of<br />
plastocyanin was prepared, in<br />
which Cu was replaced by Cd. 2D<br />
NMR spectra of Cd-PC showed that<br />
the conformations of the two<br />
proteins are essentially the same.<br />
For the relaxation time<br />
measurements, samples were<br />
treated with Chelex 100 (Bio-Rad)<br />
to remove any traces of free metal<br />
ions. To obtain the longitudinal<br />
relaxation time (Tiobs), the<br />
inversion recovery method with<br />
180^-T -90° pulse sequence was<br />
used.<br />
5 Determination of<br />
the Apparent Dissociation<br />
Constant of<br />
Cd-PC Binding to<br />
Cytochromes using<br />
1H-NMR Tl Measurements<br />
Longitudinal proton relaxation<br />
times of Cd-PC were measured in<br />
the presence of various<br />
concentrations of cytochrome (cor<br />
f) to find the binding constants<br />
and the distances from various<br />
protons of Cd-PC to the<br />
cytochrome iron atom. Observed<br />
longitudinal relaxation time<br />
( T lobs) of Cd-PC proton<br />
resonances can be considered as<br />
the sum of the relaxation rates of<br />
the bound and free Cd-PC<br />
fractions and is related to Kj), T\ 5<br />
and Tif, where Kj) is the apparent<br />
dissociation constant of the<br />
Cyt/Cd-PC complex, Tjb is the Ti
164<br />
of the Cyt/Cd-PC complex, and Tif<br />
is the Ti of the Cd-PC in the<br />
absence of the cytochrome. KTJ and<br />
Tib for Cd-PC binding to<br />
cytochrome was obtained from the<br />
above data. Kj) obtained from NMR<br />
relaxation measurements agreed<br />
very well with the value obtained<br />
from optical spectroscopy (Table<br />
I).<br />
6 Determination of<br />
Distance using<br />
Measurements<br />
The Solomon and Bloembergen<br />
equations were used to determine the<br />
distance (r) of individual protons of<br />
bound Cd-PC from the ferric centre<br />
of cytochromes. The distances for<br />
these protons were used to get the<br />
relative position and conformation of<br />
PC with respect to the ferric ion of<br />
cytochromes. Our initial results show<br />
that the ferric ion of cytochrome is<br />
very near to the Tyr-83 residue of<br />
PC, which is consistent with our<br />
kinetics studies. This work is still in<br />
progress.<br />
7 References<br />
1. P.M. Wood, Biochim. Biophys.<br />
Acta 357, 370 (1974).<br />
2. J.M. Guss, and H.C. Freeman, J.<br />
Mol. Biol. 169, 521 (1983).<br />
Bulletin of Magnetic Resonance<br />
3. J.M. Guss, P.R. Harrowell, M.<br />
Murata, V.A. Norris and H.C. Freeman,<br />
J. Mol. Biol. 192, 361 (1986).<br />
4. DJ. Cookson, M.T. Hayes and P.E.<br />
Wright, Biochim. Biophys. Acta<br />
591, 162 (1980).<br />
5. A.G. Sykes, Chem. Soc. Rev. 14,<br />
283 (1985).<br />
6. A.G. Sykes, Struct. Bonding, 75,<br />
175 (1990).<br />
7. F.A. Armstrong, H.A.O. Hill and<br />
C. Redfield, J. Inorg. Biochem. 28,<br />
171 (1986).<br />
8. J.D. Sinclair-Day and A.G. Sykes<br />
J. Chem. Soc. Dalton Trans. 2069<br />
(1986).<br />
9. G.P. Anderson, D.G. Sanderson<br />
and E.L. Gross, Biochim. Biophys.<br />
Acta, 894, 386 (1987).<br />
10. C. Beoku-Betts, S.K. Chapman<br />
and A.G. Sykes Inorg. Chem. 24,<br />
1677 (1985).<br />
11. P.M. Colman, H.C. Freeman,<br />
J.M. Guss, M. Murata, V.A. Norris,<br />
J.A.M. Ramshaw and M.P. Venkatappa,<br />
Nature, 272, 319 (1978).<br />
12. M. Nordling, T. Olausson and<br />
L.G. Lundberg, FEBS Lett., 276, 98<br />
(1990).<br />
13. S. He, S. Modi, D.S. Bendall and<br />
J.C. Gray, EMBO J, 10, 4011 (1991).<br />
14. S. Modi, S. He, D.S. Bendall and<br />
J.C. Gray, Biochim. Biophys. Acta<br />
(in press).<br />
15. S. Modi, M. Nordling, L.G. Lundberg,<br />
O. Hansson and D.S. Bendall,<br />
Biochim. Biophys. Acta (in press).
Vol. 14, No. 1-4<br />
1 Introduction<br />
Metal-peptide Interaction:<br />
Influence of the Aminoacid Sequence on the Binding of Co(II)<br />
to Glycyltryptophan and Tryptophylglycine<br />
Studied by ^NMR and Fluorescence<br />
A. Spisni, G. Sartor, L. Franzoni<br />
Institute of Biological Chemistry, University of Parma<br />
Via Gramsci, 14,<br />
43100 Parma, Italy<br />
A. Orsolini, P. Cavatorta<br />
Department of Physics, Section of Biophysics, University of Parma<br />
Viale delle Scienze,<br />
43100 Parma, Italy<br />
and<br />
M. Tabak<br />
Instituto de Fisica e Quimica de Sao Carlos, University of Sao Paulo<br />
Av. Dr. Carlos Botelho, 1465,<br />
13560 Sao Carlos (SP), Brazil<br />
Aim of this work is to investigate the nature of the<br />
interactions between transition metals and peptides.<br />
Peptides bind metal ions in various manner<br />
depending on their aminoacid composition, on the pH of<br />
the solution as well as on the metal/peptide ratio. Among<br />
the various aminoacids those with a charged side-chain<br />
are the most efficient for metal binding, though, ions can<br />
also be coordinated by the peptidic nitrogen as well as by<br />
the terminal arnino group. It is well known that the<br />
binding of transition metals to peptides decreases the<br />
apparent pKa of both the terminal amino group and of the<br />
peptidic NH. The fluorescence of the aromatic<br />
aminoacids (Trp, Tyr and Phe) can be used to monitor<br />
those changes. It is well recognized that fluorescence<br />
spectroscopy is suitable for the study of peptides and<br />
proteins, and that Trp is the best internal probe due to its<br />
' Hgh quantum efficiency and molar absorption as<br />
compared to Tyr and Phe. Stemming from these<br />
considerations, conformational changes of proteins or<br />
Peptides can be monitored following the modifications of<br />
.the Trp fluorescence. Similarly, the interaction of Trp<br />
•*ith transition metal ions can be easily detected by<br />
measuring the fluorescence quenching induced by the<br />
metal ions themselves.<br />
Recently [1], [2] we have been studying the<br />
interaction of Cu(II) and Ni(II) with Trp and Gly-Trp<br />
and we found not only that the quenching of Trp<br />
fluorescence is mainly due to a ground state interaction<br />
but also that, for the two metals, the formation of the<br />
metal complex with the dipeptide and the AA is<br />
different, both in terms of binding constants and<br />
stoichiometry.<br />
Another technique that can be used for the study of<br />
metal-peptide complexes is high resolution 'HNMR.<br />
The possibility of a transition metal to act as a line<br />
broadening or as a shift reagent is strictly associated to<br />
its electronic relaxation time. Paramagnetic ions with<br />
a short relaxation time are responsible for changes in<br />
chemical shift without line broadening while, if the<br />
relaxation time is long enough, the effect is a strong<br />
line broadening with no changes of the chemical shift.<br />
Therefore the study of the chemical shift variation as<br />
a function of the metal concentration can lead to<br />
interesting results relevant for a better understanding of<br />
the complexe's stoichiometry. At the same time<br />
relaxation studies will provide valuable information on<br />
their geometry.<br />
165
166<br />
2 Materials and Methods<br />
Tryptophylglycine (Trp-Gly) and glycyltryptophan (Gly-<br />
Trp) were obtained from Sigma Co. S. Louis, MO, their<br />
purity was checked by gas-chromatography. CoCl2-6H2O<br />
was obtained from Merk, Darmstad, Germany and used<br />
without further purification.<br />
*HNMR experiments were carried out using a Bruker<br />
AMX 400 spectrometer, operating at 9.41 T, static<br />
fluorescence experiments were carried out using a Perkin<br />
Elmer MPF 44A spectrofluorimeter, time resolved<br />
fluorescence experiments were carried out with a time<br />
correlated single photon counter equipped with an<br />
Edinburgh F199 nanosecond flash lamp operating in a N2<br />
flux of 1 2/min, a Philips XP2020Q fast photomultiplier<br />
and an EG&G Ortec fast NIM electronics. Time resolved<br />
fluorescence data analysis was carried out using the<br />
Global Analysis [3].<br />
The temperature for all the experiments was 25°C.<br />
Peptide's solutions for *HNMR were freshly prepared<br />
in double distilled water, with 10% D2O, at a final<br />
concentration of 10 mM. Solutions for fluorescence<br />
experiments were obtained by appropriate dilution of 1<br />
mM stock solution in double distilled water. No buffers<br />
were used. CoCl2 solutions were 3 M for 'HNMR and<br />
1 M for fluorescence experiments.<br />
Distinct protocols were used to obtain the pH<br />
titration of the two peptides in : HNMR and fluorescence<br />
experiments. In 'HNMR experiments small aliquots (y.£)<br />
of 0.1 M HC1 or 0.1 M NaOH were added to 0.5 m£ of<br />
the sample in order to obtain a given pH. As for the<br />
fluorescence measurements, additions of 1 M HCl or 1 M<br />
NaOH were made on 50 m£ in order to avoid dilution<br />
artifacts, the same pHmeter was used. The data from pH<br />
titrations, for both *HNMR and fluorescence, were fitted<br />
using the Henderson-Hasselbach equation in order to<br />
obtain the pK, values .<br />
In the case of fluorescence experiments correction for<br />
the inner filter effect was made using Parker's equation<br />
[4]-<br />
3 Results and Discussion<br />
Fluorescence quenching experiments demonstrate that the<br />
complexes fonned by Gly-Trp and Trp-Gly with Co(II)<br />
have distinct properties. The binding of the metal ion is<br />
strictly influenced by the aminoacid sequence and by pH.<br />
Moreover Co(II)like other paramagnetic transition metals<br />
such as Cu(II) and Ni(II), is known to influence the pK,<br />
of the ionizable groups in aminoacids and peptides [1],<br />
[2].<br />
Bulletin of Magnetic Resonance<br />
In the case of the binding of metal ions to<br />
fluorescent aminoacids a non fluorescent ground state<br />
complex is formed, thus, it is possible to calculate the<br />
binding constant from the variation of the fluorescence<br />
intensity. Co(II) binding to Trp-Gly and Gly-Trp<br />
produces biphasic quenching curves [5], indicating that<br />
at least two different complexes are formed in each<br />
case. In table I the binding constants of Co(II) for the<br />
two peptides are reported together with the fraction of<br />
the two complexes present at the given pHs.<br />
Unfortunately, fluorescence spectroscopy is not<br />
able to give the required informations for the direct<br />
determination of the stoichiometry of the complexes<br />
and of their geometry. To overcome these limitations<br />
J HNMR has been used.<br />
The pK, values of the ionizable groups<br />
(carboxylate and amino group) in the absence and in<br />
the presence of Co(II) have been obtained from<br />
fluorescence data. These values turn out to be quite<br />
different with respect to those obtained using the same<br />
technique and reported in a recent publication [6].<br />
To verify these values, the determination of the<br />
two pK,s was carried out by means of 'HNMR. We<br />
studied the pH dependence of the proton chemical<br />
shifts for the two dipeptides in the absence of the metal<br />
ion. Figure 1A reports the plot of the data for the two<br />
a protons of glycine and figure IB for the NH proton<br />
and for the indolic one. From these data the pK,s of<br />
the carboxylate and of the amino group (Table II) were<br />
calculated. The pK,s values of the amino group<br />
obtained from pH dependence of the chemical shift oi<br />
the various protons were in good agreement with those<br />
obtained from fluorescence experiments. In the case oi<br />
the indolic and the NH proton resonances, as they<br />
disappear above pH 7.5, probably because of their fasi<br />
exchange with H2O, the fitting of the data was obtained<br />
imposing, as pK^ values, the average values obtained<br />
from all the other protons. As can be seen the fitting is<br />
quite satisfactory.<br />
When Co(II) is present, while with fluorescence<br />
spectroscopy it is possible to carry out a pH titration oi<br />
the dipeptide-metal complex up to pH 12, NMR i*<br />
limited to pH 7. In fact, due to its high sensitivity<br />
fluorescence spectroscopy allows to operate at (iw<br />
concentration for the dipeptides and for Co(II), thus<br />
avoiding the precipitation of the complex at high pn<br />
NMR, being less sensitive, requires concentrations ii<br />
the range of 10 mM, therefore, above pH 7.5 w<<br />
observe the formation of a mixed complex with OH<br />
that precipitates, with a consequent disappearing of th<<br />
signal.
Vol. 14, No. 1-4 167<br />
•>"• .<br />
TABLE I: Binding constants (K} and K^) obtained from fluorescence quenching experiments, /, and f2 represent the<br />
fraction of complex formed<br />
Dipeptides<br />
Trp-Gly<br />
pH 3.2<br />
Gly-Trp<br />
pH 3.2<br />
Trp-Gly<br />
pH 8.2<br />
Gly-Trp<br />
pH 8.2<br />
3.9 -<br />
fi<br />
1<br />
0.36<br />
0.22<br />
0.55<br />
0<br />
2810.6<br />
896.6<br />
900.6<br />
0 2 4 8 8 10 12 14 0 2 4<br />
f2<br />
—<br />
0.64<br />
0.78<br />
0.45<br />
K, (M- 1 )<br />
Figure 1 A Glycine a protons chemical shifts of glycyltryptophan (•,•) and tryptophylglycine (O,E).<br />
B Indolic proton (», v) and NH(*, A) chemical shifts of glycyltryptophan ( T , A ) and tryptophylglycine (V,A).<br />
—<br />
33.5<br />
131.3<br />
12.1
168<br />
TABLE II: pK,,s obtained from fluorescence and from 1 H NMR experiments<br />
pKal<br />
P*^<br />
P*^<br />
pK,,<br />
P*^<br />
4.0<br />
3.8<br />
3.7<br />
s 3 - 8<br />
a, ' 3.5<br />
3.3<br />
3.Z<br />
Gly-Trp Gly-Trp:Co ++<br />
2.74<br />
8.27<br />
—<br />
3.13<br />
8.30<br />
Fluorescence intensities<br />
1.85<br />
7.46<br />
9.66<br />
a H NMR Chemical Shifts<br />
2.89<br />
3.1<br />
0 1 2 3 4 5 6 7<br />
pH<br />
—<br />
Bulletin of Magnetic Resonance<br />
Trp-Gly Trp-Gly:Co ++<br />
2.59<br />
7.73<br />
—<br />
2.86<br />
7.74<br />
1 ; I I I<br />
2 3 4<br />
PH<br />
B 10.2<br />
Figure 2 A Glycine a protons chemical shifts of glycyltryptophan (•,•) and tryptophylglycine (O,Q) in presence q<br />
Co(II) (1:5 peptide metal ratio). ><br />
B Indolic proton (^.v) and NH (^,A)chemical shifts of glycyltryptophan (•,*) and tryptophylglyctoe (v,i<br />
in presence of Co(II) (1:5 peptide metal ratio).<br />
10.0<br />
8.8<br />
K><br />
8.0<br />
7.8<br />
7.6<br />
7.4<br />
1Z<br />
2.51<br />
7.05<br />
9.15<br />
2.88<br />
—
Vol. 14, No. 1-4 169<br />
Figure 3 Top<br />
Bottom<br />
""I"'<br />
-SO<br />
WG ilOmM) CO • 1; 5 pH-7.15<br />
SW-30
170<br />
In figure 2A and 2B the 'HNMR titration of the glycine<br />
a protons and of the indolic and NH protons in the<br />
presence of Co(II) (1:5 ratio) are reported. It can be seen<br />
that above pH 7.5 no signal was detectable due to the<br />
formation of a bluish precipitate and that, as a<br />
consequence only the carboxylate pK, has been calculated<br />
and reported in Table II.<br />
Despite these limitations, we have been able to detect<br />
the proton NMR spectrum for the two dipeptides<br />
complexes (Figure 3). A group of resonances is shifted<br />
between -50 ppm to -100 ppm and two or three peaks<br />
appear between 80 ppm and 130 ppm. Interestingly,<br />
though the assignment of the peaks is still to be<br />
completed, it can be seen that the NMR profile for the<br />
two complexes is quite different suggesting that Trp-Gly<br />
and Gly-Trp, indeed, are forming two distinct complexes.<br />
The half line width is approximatively 200 Hz for Gly-<br />
Trp and 1000 Hz for Trp-Gly indicating that Co(II) is<br />
either closest or more tightly bound to Trp-Gly as<br />
compared to the Gly-Trp. Indeed the Trp-Gly binding<br />
constants suggests an average high affinit of this peptide<br />
respect to Gly-Trp. The NMR spectra of the two<br />
peptide-metal complexes present one single peak with<br />
the same chemica shift.A that peak tend to disappear<br />
upon D2O addition we believe it is a proton bound to a<br />
nitrogen atom. The NMR spectra of the two complexes<br />
posses other peculiar characteristics. The chemical shifts<br />
of their resonance lines are insensitive to the pH variation<br />
suggesting a high stability of the complexes. Moreover,<br />
the NMR spectra disappear below pH 5.2 for Trp-Gly<br />
and below pH 4.15 for Gly-Trp. We do believe these<br />
evidences are an indication both of the need for the<br />
carboxylic group to be ionized as well as of the<br />
requirement for a small fraction (1/1000) of NH2 (the pK,<br />
for the amino group are 8.27 and 7.73 respectively) in<br />
order to have metal binding.<br />
In conclusion, these preliminary results indicate that,<br />
because of the complementarity of NMR and<br />
fluorescence spectroscopy, it is possible to better evaluate<br />
the pK,s of the dissociable groups and the metal binding<br />
properties of small peptides. We believe that such an<br />
integrated approach can be relevant for the study of more<br />
complex macromolecules such as polypeptides and<br />
proteins.<br />
4. References<br />
Bulletin of Magnetic Resonance<br />
1. Tabak M., Sartor G. and Cavatorta P., /. of<br />
Luminescence, 43., 355 (1989).<br />
2. Tabak M., Sartor G., Neyroz P. and Cavatorta P.,<br />
J. of Luminescence, 46, 291 (1990).<br />
3. Knutson J.R., Beechem J.M. and Brand L., Chem.<br />
Phys. Lett. 102, 501 (1983).<br />
4. Parker C.A. Photoluminescence of solutions,<br />
Elsevier, (1968).<br />
5. Sartor G. Franzoni L., Cavatorta P., Tabak M. and<br />
Spisni A., manuscrip in preparation.<br />
6. Chen F., Knutson J.R., Ziffer H. and Porter D.,<br />
Biochemistry, 2Q, 5184 (1991).<br />
Acknowledgments This work was supported<br />
by a CNR Grants #92.00752.CT04 and #92.02243.ctl4,<br />
by MURST 60% (SG) and MURST 40% (SA)
Vol. 14, No. 1-4 171<br />
INTRODUCTION<br />
ASSIGNMENTS OF THE *H NMR SPECTRUM OF A CONSENSUS<br />
DNA-BINDING PEPTIDE FROM THE HMG-I PROTEIN<br />
Jeremy N. S. Evans§$*, Mark S. Nissen§ and Raymond Reeves§<br />
Departments of Biochemistry/Biophysics§ and Chemistry*,<br />
Washington State University, Pullman, WA 99164-4660. U. S. A.<br />
The HMG-I subfamily [1-3] of high mobility group<br />
(HMG 1 ) chromatin proteins [4] consists of DNA-binding<br />
proteins that preferentially bind to stretches of A'T-rich sequence<br />
both in vitro [5-8] and in vivo [9]. Recently, members<br />
of the HMG-I family have been suggested to bind in<br />
vitro to the narrow minor groove of A*T-DNA by means of<br />
an 11 amino acid peptide binding domain (BD) which, because<br />
of its predicted structure is called the "A*T-hook motif<br />
[10]. The HMG-I proteins are specific substrates for the<br />
ii i i • / \ -, Acdc2/cdc28 , . , ,<br />
cell cycle regulating enzyme(s) p34 kmase (also<br />
known as histone HI kinase) both in vivo and in vitro [11-<br />
13]. The sites of phosphorylation by cdc2 kinase are the<br />
threonine residues at the amino terminal ends of the A*Thook<br />
motifs and such modifications have been demonstrated<br />
to reduce markedly the affinity of binding of the HMG-I proteins<br />
to DNA [12,13].<br />
HMG-I proteins are also of considerable biological interest<br />
because they are expressed at elevated levels in actively<br />
proliferating cells and have been observed to be a<br />
characteristic feature of undifferentiated [1,7] or neoplastically<br />
transformed cellular phenotypes [14,15]. High HMG-I<br />
levels have been found to be a consistent feature of rat and<br />
mouse malignant cells [14-17] and have been suggested to<br />
be protein markers for both neoplastic transformation [15]<br />
and metastatic potential [18]. The HMG-I proteins have also<br />
been implicated in control of DNA replication [19,20] and<br />
the regulation of gene transcription [8,21,22]. It is known<br />
from their primary sequences that the HMG-I proteins have<br />
the overall structure typical of Ptashne-type transcriptional<br />
activator proteins possessing both a DNA binding domain(s)<br />
and a highly acidic COOH terminus [23]. In vitro HMG-I-<br />
Ijke proteins have been demonstrated to increase transcription<br />
of isolated ribosomal genes [21] and to alter the conformation<br />
and stability of A-T-rich regions of DNA [24]<br />
properties often associated with DNA-binding gene regulatory<br />
proteins.<br />
As an initial attempt to determine in molecular detail the<br />
interaction of the BD peptide with A»T rich DNA, we have<br />
examined a synthetic 13 residue BD peptide by NMR<br />
spectroscopy. In this paper we report the assignments of the<br />
j<br />
HMG - Wfi 11 mobility group; BD,<br />
domain; DQF, double quantum filtered;<br />
C0Trelated<br />
spectroscopy; NOESY, nuclear<br />
! effCt spectroscopy; ROESY, rotatingei<br />
VT 3USer effect spectroscopy; TOCSY, total<br />
elated spectroscopy.<br />
resonances for the peptide at 295 K and pH 3.4, and provide<br />
preliminary evidence on the peptide structure.<br />
MATERIALS AND METHODS<br />
Peptide Synthesis. The 13mer BD peptide<br />
(VPTPKRPRGRPKK) was synthesized by solid-phase synthesis<br />
(on a Departmental Applied Biosystems model 431A<br />
peptide synthesizer), and purified by reverse-phase HPLC on<br />
a Vydak C4 column (1 x 25 cm) using, a water (containing<br />
0.1% trifluoroacetic acid)-acetonitrile gradient under standard<br />
conditions. The 13mer BD peptide eluted at 15% acetonitrile<br />
(Rt = 13 mins @ 1.5 mL min" 1 ).<br />
NMR Spectroscopy. High field Fourier transform (FT)<br />
NMR studies were performed on a Varian VXR-500S (11.75<br />
T, 500MHz J H) NMR spectrometer. Deuterium was used<br />
for locking the field. *H NMR chemical shifts were referenced<br />
externally to samples of similar dielectric constant<br />
containing sodium 3-(trimethylsilyl) propanoate-2,2,3,3-d4<br />
(TSP) in D2O buffer (5H = 0.00 ppm). Sample temperature<br />
was maintained with a Varian variable temperature control<br />
unit, using gaseous nitrogen (from boil-off liquid nitrogen)<br />
cooled using an FTS XR-85 cryo-cooler. The majority of<br />
samples of peptide were maintained at 295 K, except where<br />
the amide exchange rates were being measured, when the<br />
sample was maintained at 277 K. Data was downloaded to<br />
either a Silicon Graphics 4D25TG or a-4D70GT workstation,<br />
and converted from Varian format to FELIX format using<br />
the VNMR2FELIX conversion program (a gift from<br />
Darrell Davies, University of Utah). The output from this<br />
was processed using FELIX (Hare Research Ltd.). All 2D<br />
data was obtained using the hypercomplex phase sensitive<br />
method [25] and processed as 2K x 2K complex data sets<br />
with baseline correction and sine-bell squared weighting<br />
functions in both dimensions.<br />
DQF-COSY were recorded with the pulse sequence tQ-<br />
90°-f7-90 o -S-90 o -f2, where tj is the evolution time, t2 is<br />
the acquisition time, and 8 is a fixed delay of 3 \is [26].<br />
TOCSY spectra were recorded with the pulse sequence /#-<br />
90°-fi-SLx-(MLEV-17)-SLx-/2, where SLX was a 4 ms<br />
trim pulse along the x axis [27]. The MLEV-17 spin-locking<br />
pulse sequence was repeated to give a mixing time of<br />
40 ms. ROESY spectra were recorded with the pulse sequence<br />
/o-90 0 -r;-90 0 -SLx(30°)-90°-t2, where SLx(30°) is a<br />
small spin-lock pulse repeated to give a mixing time of 200<br />
ms [28]. NOESY spectra were recorded with the pulse
172 Bulletin of Magnetic Resonance<br />
TABLE 1 Sequential Assignments of HMG-I 13mer Binding Domain Peptide, pH 3.4,295K<br />
Residue<br />
VI<br />
P2<br />
T3<br />
P4<br />
K5<br />
R6<br />
P7<br />
R8<br />
G9<br />
RIO<br />
Pll<br />
K12<br />
K13<br />
NH<br />
_<br />
_ 8.44<br />
• _<br />
8.10<br />
8.35<br />
8.54<br />
8.41<br />
8.28<br />
— 8.44<br />
7.24<br />
otCH<br />
4.23<br />
4.58<br />
4.60<br />
4.33<br />
4.23<br />
4.68<br />
4.46<br />
4.34<br />
3.92, 4.04<br />
4.68<br />
4.45<br />
4.32<br />
3.27<br />
sequence ta-90°-tj-90°-Tm-90 o -t2 [29,30] with mixing<br />
times, TW, of 100, 300, 400, and 600 ms. Solvent suppression<br />
was achieved by presaturation of the H2O resonance for<br />
all the 2D experiments.<br />
Sample Preparation. The BD peptide was dried with successive<br />
cycles of lyophilization and re-hydration with either<br />
H2O or D2O and then dissolved in one of the following:<br />
Buffer (i) 25 mM potassium phosphate, 0.01% (w/v) NaN3,<br />
in 10% v/v D2O/H2O, pH 3.4; or Buffer (ii) 25 mM potassium<br />
phosphate, 0.01% (w/v) NaN3, in 99% v/v<br />
D2O/H2O, pH 3.4. pH titrations were carried out by careful<br />
4.0 3.0 2.0<br />
(ppm)<br />
tPx<br />
m 6*<br />
f\j<br />
I<br />
&CH<br />
2.36<br />
1.91, 2.37<br />
4.16<br />
1.91, 2.35<br />
1.76, 1.87<br />
1.77, 1.87<br />
1.93, 2.34<br />
1.73, 1.87<br />
1.76, 1.88<br />
1.94, 2.34<br />
1.79, 1.84<br />
1.74, 1.89<br />
Others<br />
1.03, 1.14<br />
2.08<br />
1.31<br />
2.04<br />
1.43<br />
1.71<br />
2.05<br />
1.44, 1.50<br />
1.73<br />
2.05<br />
1.51<br />
1.49<br />
3.67, 3.80<br />
3.73, 3.90<br />
3.03, 3.25<br />
3.05, 3.25, 7.55<br />
3.68, 3.87<br />
3.05, 3.25<br />
3.25<br />
3.66, 3.84<br />
3.05<br />
3.05<br />
addition of small quantities of HC1 or NaOH and measurement<br />
with a 4 mm pH electrode (Ingold Co.).<br />
RESULTS<br />
The 13mer peptide was studied by NMR at 277 and 295 K<br />
and at a variety of pH values. The linewidths of the ID 500<br />
MHz NMR spectra were not sensitive to concentration in<br />
the range 0.1-20 mM, indicating negligible aggregation of<br />
the peptide [31]. The use of 2-dimensional double-quantum<br />
filtered correlated spectroscopy (DQF-COSY) and total<br />
correlated spectroscopy (TOCSY) has enabled us to assign<br />
resonances to amino acid spin types, although distinctions<br />
1.0 4.0 3.0 2.0<br />
(ppm)<br />
Figure 1 500 MHz *H DQF-COSY (A) and TOCSY (B) NMR spectra of the aliphatic resonances of the 13mer BD<br />
(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<br />
G9NH<br />
•°KI2NHa<br />
R8nHa<br />
• K-JNHa<br />
»T3NHO<br />
* »Rl 6 hours.<br />
Interestingly, the slowly exchanging amide resonances also<br />
do not titrate over the pH range 2.2 to 7.1, with the exception<br />
of the K5 NH, and all appear to be located along one<br />
side of the peptide molecule.<br />
DISCUSSION<br />
A model for how the consensus binding domain peptide<br />
from the HMG-I protein binds to the minor groove of A«T<br />
rich DNA has been proposed by this laboratory [10] on the<br />
basis of molecular modelling. In order to test this model, we<br />
have initiated NMR studies of a 13mer BD peptide in<br />
solution. While no
HI<br />
i ill,<br />
I 1<br />
174<br />
the 13mer peptide). These amides might be expected to be<br />
less accessible to solvent, and potentially may participate in<br />
hydrogen-bonds with the carbonyl groups from (i, i + 2)<br />
residues. At this stage, there is no additional corroborating<br />
evidence from medium-range or long-range NOEs either to<br />
support or refute these speculations. However, in the<br />
presence of A«T rich DNA, the conformational lability of<br />
the peptide backbone and side-chains would be expected to be<br />
reduced significantly, and longer range NOEs detectable.<br />
ACKNOWLEDGEMENTS<br />
We should like to acknowledge Gerhard Munske for the<br />
synthesis of the 13mer peptide, Wendy Shuttleworth for<br />
help with some sample preparations, Darrell Davies<br />
(University of Utah) for the VNMR2FELIX conversion<br />
software. Supported in part by an American Cancer Society<br />
Institutional Grant IGR-119L (JNSE), NIH Grant R01<br />
GM46352 (RR), and the WSU NMR Center is supported by<br />
NIH grant RR 0631401, NSF grant CHE 9115282 and<br />
Battelle Pacific Northwest Laboratories Contract No. 12-<br />
097718-A-L2.<br />
REFERENCES<br />
[ 1] Lund, T., Holtlund, J., Fredriksen, M. & Laland, S.<br />
(1983) FEBS Lett. 152, 163-167.<br />
t 2] Johnson, K., Lehn, D., Elton, T., Barr, P. & Reeves,<br />
R. (1988) J. Biol. Chem. 263, 18338-42.<br />
[ 3] Johnson, K., Lehn, D. & Reeves, R. (1989) Mol.<br />
Cell. Biol. 9,2114-23.<br />
[ 4] Goodwin, G. & Bustin, M. (1988) in Kahl, G.<br />
Architecture of Eukaryotic Genes, 187-205, VCH<br />
Weinheim, Germany.<br />
[ 5] Strauss, F. & Varshavsky, A. (1984) Cell 37, 889-<br />
901.<br />
[ 6] Solomon, M., Strauss, F. & Varshavsky, A. (1986)<br />
Proc. Nail. Acad. Sci. U.SA. 83, 1276-1280.<br />
[ 7] Elton, T., Nissen, M. & Reeves, R. (1987)<br />
Biochem. Biophys. Res. Commun. 143, 260-265.<br />
[ 8] Reeves, R., Elton, T., Nissen, M., Lehn, D. &<br />
Johnson, K. (1987) Proc. Natl. Acad. Sci. U.SA.<br />
84, 6531-6535.<br />
[ 9] Disney, J., Johnson, K., Magnuson, N., Sylvester,<br />
S. & Reeves, R. (1989) / Cell Biol. 109,1975-82.<br />
[10] Reeves, R. & Nissen, M. (1990) J. Biol. Chem.<br />
265, 8573-8582.<br />
[11] Lund, T. & Laland, S. (1990) Biochem. Biophys.<br />
Res. Commun. 171, 342-7.<br />
[12] Reeves, R., Langan, T. & Nissen, M. (1991) Proc.<br />
Natl. Acad. Sci. USA88, 1671-5.<br />
[13] Nissen, M., Langan, T. & Reeves, R. (1991) /.<br />
Biol. Chem. 266, 19945-19952.<br />
Bulletin of Magnetic Resonance<br />
[14] Giancotti, V., Pani-D'Andrea, P., Berlingieri, M. T.,<br />
Di Fiore, P. P., Fusco, A., Veccio, G., Philip, R.,<br />
Crane-Robinson, C, Nicolas, R. H., Wright, C. A., &<br />
Goodwin, G. H. (1987) EMBO J. 6,1981-1987.<br />
[15] Giancotti, V., Buratti, E., Perissin, L., Zorzet, S.,<br />
Balmain, A., Portella, G., Fusco, A., & Goodwin,<br />
G. H. (1989) Exp. Cell Res. 184, 538-45.<br />
[16] Elton, T. & Reeves, R. (1986) Anal. Biochem.157, 53-<br />
62.<br />
[17] Vartiainen, E., Palvimo, J., Mahonen, A., Linnala-<br />
Kankkunen, A. & Maenpaa, P. (1988) FEBS Lett<br />
228, 45-48.<br />
[18] Bussemakers, M., van de Ven, W., Debruyne, F. &<br />
Schalken, J. (1991) Cancer Res. 51,606-611.<br />
[19] Grummt, F., Hoist, A., Muller, F., Wegner, F.,<br />
Schwender, S., Luksza, H., Zastow, G., & Klavinius,<br />
A. (1988) Cancer Cells 6,463-466.<br />
[20] Wegner, M., Zastrow, G., Klavinius, A., Schwender,<br />
S., Muller, F., Luksza, J., Hoppe, J., Wienberg, J. &<br />
Grummt, F. (1989) Nucleic Acids Res. 17, 9909-9932<br />
[21] Yang-Yen, H. & Rothblum, L. (1988) Mol. Cell. Bio<br />
8, 3406-3414.<br />
[22] Eckner, R. & Bimstiel, M. (1989) Nucleic Acids Res.<br />
17, 5947-59.<br />
[23] Ptashne, M. (1988) Nature 335,683-689.<br />
[24] Lehn, D., Elton, T., Johnson, K. & Reeves, R. (198*<br />
Biochem. Int. 16, 963-971.<br />
[25] States, D. J., Haberkorn, R. A., & Ruben, D. J. (198:<br />
/. Magn. Reson. 48, 286-293.<br />
[26] Ranee, M., S0rensen, O.W., Bodenhausen, G., Wagne<br />
G., Ernst, R. R., & Wuthrich, K. (1983) Biochem.<br />
Biophys. Res. Commun. 117, 479-485.<br />
[27] Bax, A., & Davis, D. G. (1985) J. Magn. Reson. 65,<br />
355-360.<br />
[28] Kessler, H., Griesinger, C, Kerssebaum, R., Wagner,<br />
R., & Ernst, R. R. (1987) J. Am. Chem. Soc. 109,<br />
607-609.<br />
[29] Jeener, J., Meier, B. H., Bachmann, P., & Ernst, R. 1<br />
(1979) /. Chem. Phys. 71, 4546-4553.<br />
[30] Bodenhausen, G., Kogler, H., & Ernst, R. R. (1984;<br />
/. Magn. Reson. 58, 370.<br />
[31] Dyson, H. J., & Wright, P. E. (1991) Annu. Rev.<br />
Biophys. Biophys. Chem. 20, 519-538.<br />
[32] Wuthrich, K., Billeter, M., & Braun, W. (1984) /.<br />
Mol. Biol. 180, 715-740.<br />
[33] Mayo, K. M., Parra-Diaz, D., McCarthy, J. B., &<br />
Chelberg, M. (1991) Biochemistry 30, 8251-8267.<br />
[34] Grathwohl, C. & Wuthrich, K. (1976) Biopolymers<br />
15, 2025-2041.<br />
[35] Grathwohl, C. & Wuthrich, K. (1981) Biopolymers<br />
20,2623-2633.
Vol. 14, No. 1-4 175<br />
Solution Structure of the<br />
DNA-binding Domain of GAL4<br />
from Saccharomyces cerevisiae<br />
James D. Baleja, V. Thanabal, Ted Mau, and Gerhard Wagner<br />
Department of Biological Chemistry and Molecular Pharmacology<br />
1 Introduction<br />
The GAL4 transcriptional<br />
activator protein has long been a<br />
favorite for the study of<br />
transcription in eukaryotic biology.<br />
Genetic studies reveal a modular<br />
architecture for the protein with<br />
different functions associated with<br />
each module [1]. A DNA-binding<br />
domain of the protein recognizes and<br />
binds to a sequence of DNA termed<br />
the Upstream Activating Sequence<br />
(UASQ). Other parts of the protein are<br />
relevant for activation. They interact<br />
with the transcriptional machinery<br />
including RNA polymerase to activate<br />
transcription. The UASQ is near the<br />
genes that encode the proteins<br />
required for galactose utilization.<br />
Upon presentation of galactose to the<br />
yeast cell, this DNA site is specifically<br />
bound by GAL4, the transcription<br />
function of RNA polymerase is<br />
activated, and enzymes required for<br />
galactose utilization are produced [2].<br />
As a dimer of 881 amino acids,<br />
GAL4 is too large for the<br />
determination of a high-resolution<br />
NMR structure and we have instead<br />
studied a fragment containing the Nterminal<br />
65 amino acid residues<br />
Harvard Medical School<br />
Boston, Massachusetts 02115<br />
including its DNA-binding domain<br />
[3]. In the absence of DNA, GAL4(65)<br />
is monomeric in solution,<br />
presumably because it does not have<br />
the amino acid residues necessary for<br />
dimerization [4]. GAL4(65) is dimeric<br />
when bound to any of four DNA sites<br />
present in the UAS [5]. Each of these<br />
binding sites is approximately twofold<br />
palindromic, and like many<br />
other dimeric DNA-binding proteins,<br />
each monomer of the GAL4 dimer<br />
interacts with a half-site of DNA.<br />
The structure and dynamics of the<br />
monomeric DNA-binding domain of<br />
GAL4 (residues 1-65; Figure 1), an<br />
investigation of the binding to a DNA<br />
half-site (Figure 2), and the structure<br />
of the resultant protein-DNA complex<br />
are presented in this paper.<br />
1 11 21<br />
MKLLSSIEQA CDICRLKKLK CSKEKPKCAK<br />
31 41 51<br />
CLKNNWECRY SPKTKRSPLT RAHLTEVESR<br />
61<br />
LERLE<br />
Figure 1. Primary amino acid<br />
sequence for the DNA-binding<br />
domain of GAL4.
176<br />
1 2 3 4 5 6 7 3 9 10<br />
C C G G A G G A C T<br />
G G C C T C C T G A<br />
20 19 18 17 16 15 14 13 12 11<br />
Figure 2. Nucleotide sequence for the<br />
one-half DNA-binding site of GAL4.<br />
2 1H, 15N, and n^Cd NMR<br />
resonance assignments<br />
Assignment of specific nuclei to<br />
the observed NMR resonance<br />
frequencies is the first step in<br />
determining the structure of a<br />
protein using NMR techniques [6].<br />
Assignments for the 1 H, ^$N, and<br />
113 Cd NMR resonances of GAL4(65)<br />
were made using homonuclear and<br />
hetero-nuclear NMR experiments<br />
and by following standard protocol<br />
([6], Baleja and Wagner, unpublished<br />
results).<br />
Zinc is required for the DNAbinding<br />
activity of GAL4 [4]. It can be<br />
replaced by NMR-active 113 Cadmium<br />
without loss of DNA-binding. The<br />
113 Cd NMR spectrum (Figure 3)<br />
shows that two metal ions are<br />
coordinated by a Cys - (X)2 - Cys - (X>6<br />
- Cys - (X)6 - Cys - (X)2 - Cys - (X)6 -<br />
Cys motif using six cysteines and<br />
forming a bimetal-thiolate cluster<br />
[7]. Hetero-nuclear correlation<br />
experiments between the l* 3 Cd and<br />
*H define the liganding of the two<br />
central metal ions (Figure 3).<br />
Having assigned resonance frequencies<br />
to specific nuclei, the<br />
correspondence of cross peaks to the<br />
protons can be made and local<br />
structural information was can then<br />
be determined. A section of the twodimensional<br />
NOESY spectrum is<br />
shown in Figure 4. The region shows<br />
the cross peaks among the amide<br />
protons of the protein backbone and<br />
protons of the aromatic sidechains.<br />
BuJietin of Magnetic Resonance<br />
Figure 3. Heteronuclear ^ C d H<br />
correlation experiments. The standard<br />
reverse INEPT pulse sequence<br />
[8] was followed by a short MLEV-17<br />
TOCSY transfer [9] and proton<br />
detection./<br />
Indicating the presence of two ahelices,<br />
there are a series of close<br />
approaches between amide proton<br />
resonance frequencies for several<br />
sequential residues. Other NOEs arise<br />
between amino acid residues more<br />
distant in sequence and define the<br />
unique topological features for the<br />
protein. NOE intensities at 56, 150, and<br />
250, and 500 millisecond mixing times<br />
were converted to the corresponding<br />
interproton distances using the usual<br />
inverse r 6 relationship [3]. In case of<br />
overlap in the homonuclear NOESY,<br />
spectrum analysis was also made<br />
using a three-dimensional ^JSJ.IJJ.IH<br />
NOESY-HMQC spectrum recorded on a<br />
uniformly ^N-labeled protein<br />
(Figure 5, [10]). From the NOESY data,
Vol. 14, No. 1-4<br />
residues 9-40 are observed to form a<br />
well-defined, compact globular cluster.<br />
The terminal residues 1-8 and 41-<br />
66 show little persistent structure in<br />
solution. There are many interresidue<br />
NOE crosspeaks for the<br />
central core (30 per residue, on<br />
average), but only a few weak NOE<br />
cross peaks for protons in the<br />
disordered region. In addition, the<br />
amide protons of the flexible trails<br />
are in rapid exchange with solvent<br />
[10] and have narrow resonance<br />
lines, indicating that these residues<br />
have considerable conformational<br />
mobility in the absence of DNA. This<br />
unstructured character for parts of<br />
the GAL4 protein may be vital to its<br />
biological function. On the other<br />
hand, at least some aspects of this<br />
mobility may merely be a<br />
consequence of the truncation used<br />
for GAL4 in this study. Nonetheless,<br />
our picture of GAL4 is one in which<br />
two N-terminal recognition modules<br />
are connected by flexible linkers to a<br />
dimeric core [4],<br />
-©V<br />
o<br />
K23<br />
o<br />
J28<br />
R15<br />
D1i! L1<br />
O<br />
E37<br />
OC11<br />
S41<br />
DS22<br />
>C21<br />
C38,<br />
9.0 8.5 8.0 7.5 7.0<br />
6)2 (ppm)<br />
6.5 6.0<br />
Figure 4. NOESY spectrum of amide<br />
and aromatic protons of GAL4(65).<br />
Sample concentration was 1.5 mM in<br />
0.2 M NaCl, 20 mM sodium phosphate,<br />
pH 7.0 at 25°C. Numbers indicate the<br />
residues for which sequential strong<br />
amide to amide cross peaks are<br />
observed which are indicative of ahelices.<br />
K&O T»K27<br />
,©113<br />
C31
178<br />
3 Structure of the GAL4<br />
DNA-binding domain<br />
Sets of interproton distances and<br />
torsion angles formed the basis for<br />
structure determination. Distances<br />
were determined from the NOESY data<br />
[3]. § torsion angles were interpreted<br />
from measured NH-oc coupling constants<br />
and x 1 angles were derived<br />
from ^N-j} and a-(5 coupling constants<br />
[3]. Sulfur-Cd liganding distances<br />
were imposed to be 2.35 to 2.45 A, Cys<br />
Cp-Cd distances to be less than 3.4 A,<br />
and sulphurs liganding the same Cd<br />
to be between 3.3 and 4.2 A. 614<br />
distance and 41 torsion angle<br />
measurements were used with the<br />
distance geometry package DG-II to<br />
generate a set of structures (0.6 A<br />
backbone atom rmsd) in agreement<br />
with the NMR data [3]. A schematic of<br />
the structure (Figure 6) shows the<br />
two central metal ions coordinated by<br />
the six cysteines. The DNA-binding<br />
domain consists of an a-helix and an<br />
extended structure, then a sharp turn<br />
that contains a cis-proline bond, and<br />
then a 2nd a helix followed by an<br />
extended structure. If the central<br />
metal binding subdomain of GAL4 is<br />
split, and the two halves<br />
superimposed, a striking correspondence<br />
between each part is revealed.<br />
The conformation of the 13 residue<br />
segment from residues 10 to 22 is<br />
almost identical to that of residues 27<br />
to 39, with an rmsd of 0.8 A for the<br />
backbone atoms. Although the<br />
structural integrity of the protein<br />
would be provided mainly by the way<br />
in which the two metal ions are<br />
liganded, some hydrophobic packing<br />
is observed with the side chains of<br />
W36 and Y40 [3]. Slowly exchanging<br />
amides of GAL4, in both the free and<br />
DNA-bound forms [10], can be<br />
attributed, in part, to hydrogen<br />
bonding to carbonyl oxygens within<br />
the a-helices, and to sulfurs of the<br />
cysteinyl side-chains [12].<br />
Bulletin of Magnetic Resonance<br />
Figure 6. Structure of the central<br />
core for the DNA-binding domain of<br />
GAL4. Cysteines that ligand the two<br />
central metal ions are numbered.<br />
4 Formation of a GAL4-DNA<br />
complex<br />
Imino protons of one half of the<br />
consensus dimeric GAL4 DNAbinding<br />
were monitored in forming<br />
the complex between the protein and<br />
DNA (Figure 7). One imino proton is<br />
present for each base-pair of a 10<br />
base-pair DNA duplex.<br />
_JV<br />
13.8 13.6 13.4 13.2 13.0 12.8 12.6 ppm<br />
Figure 7. Titration of DNA with GAL4.<br />
Experimentally, eight imino proton<br />
resonances are observed, since the<br />
imino protons on the base-pairs at<br />
each end of the duplex exchange<br />
rapidly with the bulk solvent at room<br />
temperature. Each imino resonance<br />
has been assigned to a specific base-
Vol. 14, No. 1-4<br />
pair. Resonances shift as the<br />
environment around each proton<br />
changes upon addition of GAL4. All<br />
resonances broaden as the more<br />
slowly protein-DNA complex is<br />
formed. The equilibrium dissociation<br />
constant is 169± 13 |J.M. Resonances<br />
are in fast exchange between the<br />
protein-DNA complex and the free<br />
components indicating rapid equilibrium<br />
between free and bound<br />
forms.<br />
NOESY and TOCSY two-dimensional<br />
data for the protein-DNA complex<br />
14.0 12.0 10.0 8.0<br />
F2<br />
6.0<br />
(ppm )<br />
have been collected (Figure 8). These<br />
spectra are promising for full<br />
analysis since the resonance lines<br />
are not too broad for resonance<br />
assignment, chemical shift dispersion<br />
is good, and solubility of the<br />
protein-DNA complex is adequate. The<br />
DNA resonances have been completely<br />
re-assigned and the protein<br />
resonances have been partially reassigned<br />
within the protein-DNA<br />
complex.<br />
4.0 2.0<br />
Figure 8. NOESY spectrum of the GAL4:DNA complex. The concentration of the<br />
complex was approximately 0.5 mM. Buffer conditions were 0.15 M NaCl, 20 mM<br />
PO4, pH 7, 25°C.<br />
5 NMR structure of<br />
GAL4-DNA complex<br />
the<br />
Several NOE contacts have been<br />
observed between protein and DNA<br />
(dashed lines, Figure 9) yielding a<br />
preliminary stucture for this GAL4-<br />
DNA complex. The recognition helix<br />
for the free form of the protein was<br />
docked onto a B-DNA. The amino acid<br />
residues responsible for recognizing<br />
DNA are part of the metal-binding<br />
cluster and interact with edges of<br />
'base-pairs exposed in the<br />
major<br />
groove of the DNA.<br />
The structure of the DNA-binding<br />
domain of GAL4 (residues 1-65) bound<br />
to full site DNA containing two GAL4<br />
binding sites has been determined<br />
crystallographically. In agreement<br />
with our observed intermolecular<br />
NOE cross peaks (Figure 8), the Cterminal<br />
end of the first a-helix of<br />
the metal binding cluster (residues 9-<br />
40) provides for sequence-specific<br />
re-cognition of DNA by GAL4. When<br />
bound to full site DNA, parts of<br />
GAL4(65), before unstructured in<br />
solution, adopt a regular conformation.<br />
Residues 41 to 49 form a<br />
linker region which interacts with<br />
179
180<br />
the backbone of the DNA. In addition,<br />
residues 50-64 form a small dimerization<br />
domain using a coiled-coil type<br />
packing arrangement [4]. Our preliminary<br />
results on an intact<br />
dimerization element for GAL4<br />
(residues 50-106) indicate that the ahelical<br />
character for the coiled-coil<br />
of the protein extends beyond residue<br />
64 (Baleja, Marmorstein, and Wagner,<br />
unpublished results). The same<br />
conformation (1.1 A rmsd) is observed<br />
for the recognition module of<br />
GAL4(65) in solution using NMR<br />
techniques as for the central metalbinding<br />
cluster of GAL4(65) bound to<br />
DNA using crystallographic techniques.<br />
Thus the core of the DNAbinding<br />
domain changes little upon<br />
binding DNA.<br />
Figure 9. Preliminary structure for a<br />
GAL4-DNA complex.<br />
6 Summary<br />
The DNA-binding domain of GAL4<br />
has many interesting features. It is a<br />
novel DNA-binding motif with a<br />
globular two-metal cluster that has<br />
hydrogen-bonds to sulfur and twofold<br />
internal symmetry. GAL4 reads<br />
the sequence of DNA through the<br />
amino acids present at the C-terminal<br />
Bulletin of Magnetic Resonance<br />
end of the first a-helix. The two DNAreading<br />
modules of intact GAL4 are<br />
tethered by flexible linkers to the<br />
central body of the protein. Once<br />
bound to DNA through the Nterminal<br />
recognition domains, the Cterminal<br />
portion of GAL4 is in the<br />
correct position to interact with the<br />
components of the transcriptional<br />
machinery to bring about transcriptional<br />
activation.<br />
7 References<br />
1. Keegan, L., Gill, G., and Ptashne,<br />
M., Science 231, 699 (1986).<br />
2. Johnston, M., Micro. Rev. 51, 458<br />
(1987).<br />
3. Baleja, J. D., Marmorstein, R.,<br />
Harrison, S. C, and Wagner G.,<br />
Nature 356, 448 (1992).<br />
4. Marmorstein, R., Carey, M.,<br />
Ptashne, M., and Harrison, S. C,<br />
Nature 356, 408 (1992).<br />
5. Carey, M., Kakidani, H.,<br />
Leatherwood, J., Mostashari, F.,<br />
and Ptashne, M., J. Mol. Biol. 209,<br />
423 (1989).<br />
6. Wuthrich, K. NMR of Proteins<br />
and Nucleic Acids (Wiley, New<br />
York, 1986).<br />
7. Pan, T., and Coleman, J. E.,<br />
Biochemistry 209, 3023 (1990).<br />
8. Briihwiler, D., and Wagner, G., J.<br />
Magn. Reson. 69, 546 (1986).<br />
9. Bax, A., and Davis, D. G., J. Magn.<br />
Reson. 65, 355 (1985).<br />
10. Mau, T., Baleja, J. D., and Wagner,<br />
G., Protein Science (submitted).<br />
11. Bodenhausen, G., and Ruben, D. J.,<br />
Chem. Phys. Lett. 69, 185 (1980).<br />
12. Kraulis, P., Raine, A. R. C<br />
Gadhavi, P. L., and Laue, E. D.,<br />
Nature 356, 448 (1992).
Vol. 14, No. 1-4<br />
STRUCTURAL INVESTIGATION OF FOLIC ACID BY<br />
NMR PROTON RELAXATION AND MOLECULAR<br />
MECHANICS ANALYSIS<br />
1 Introduction<br />
Claudio Rossi, Alessandro Donati, Sergio Ulgiati*<br />
and Maria Rosaria Sansoni<br />
Department of Chemistry, University of Siena, Pian dei Mantellini, 44<br />
53100 Siena ITALY<br />
•Department of Chemistry, University of Sassari, Via Vienna, 2<br />
07100 Sassari ITALY<br />
Folic acid N-[4-(2-Amino-4-hydroxypteridinyl-(6)-methylamino)-benzoyl]-Laminoglutaric<br />
acid (Figure 1) is a<br />
fundamental coenzyme involved in onecarbon<br />
unit transfer processes 1 . The solid<br />
state conformation of folic acid has been<br />
defined but there have been few<br />
investigations on the structure of this<br />
coenzyme in solution^.<br />
12<br />
by Nuclear Magnetic Resonance (NMR).<br />
Information on dynamical motion,<br />
magnetic dipolar connectivities and<br />
energy minimization calculations are<br />
combined in order to define the solution<br />
structure of folic acid.<br />
2 Experimental<br />
Two-dimensional COSY, NOESY and Hetcor<br />
Figure 1: Structure and numbering of folic acid.<br />
f ole is also fundamental in reductive<br />
zymatic processes in which<br />
fchydrofolate (the reduced form of folic<br />
) is oxidized to dihydrofolate and folate.<br />
•aihydrofolate reductase NADPH-<br />
•ndent enzyme controls the biological<br />
of tetrahydrofolate 3 .<br />
present paper the conformational<br />
of this molecule were analyzed<br />
experiments were obtained by (3C/2-ti -7C/2-<br />
AT)n, 4 (rc/2-tl-TC/2-tm-JU/2AT)n 5 a n d<br />
pulse sequences respectively. Spin-lattice<br />
relaxation rates were measured using the<br />
(180°-T-90°-t)n pulse sequence. The NMR<br />
measurements were performed using a 0.12<br />
mol.dm' 3 DMSO-d6 solution at 27°C. NMR<br />
spectra were recorded on a Varian XL-200<br />
181
182<br />
and a Bruker AMX-600 spectrometers<br />
operating at 200 and 600 MHz respectively.<br />
Molecular mechanics calculations were<br />
computed by the MacroModel program,<br />
version 2.5^ implemented on a Vax 11/750<br />
computer. The force field used was that<br />
reported by Weiner et al^.<br />
3 Results and Discussion<br />
In the present paper the NMR properties of<br />
folic acid were investigated in depth in<br />
order to define the molecular structure in<br />
DMSO-d6 solution.<br />
As the proton and carbon assignments of<br />
folic acid refer only to water solution at<br />
basic pH^^O, both proton and carbon<br />
chemical shifts in DMSO-d6 were<br />
determined by a method based on<br />
conventional two-dimensional COSY and<br />
Hetcor experiments. Figures 2 and 3 show<br />
the COSY and Hetcor spectra of folic acid.<br />
The complete proton assignments and<br />
chemical shifts of protonated carbons can<br />
be determined from these sets of data. The<br />
quaternary carbons were assigned by a<br />
frequency-dependent selective protoncarbon<br />
NOE experiment 11 > 1 ^.<br />
The results obtained are reported in Table 1.<br />
The strategy for structural analysis was<br />
based on a combined approach. First we<br />
studied the dynamical properties of folic<br />
acid in solution which led to two possible<br />
scenarios:<br />
i) the molecular motion is subject to<br />
different degrees of freedom, in which case<br />
each molecular moiety behaves<br />
independently as a consequence of the lack<br />
of "ordered" elements;<br />
ii) the molecule is subject to overall<br />
dynamical reorientation, characterized by<br />
a single rotational correlation time, %c.<br />
These conditions, for molecules of the size<br />
of folic acid, are verified whenever noncovalent<br />
interactions stabilize specific<br />
conformations.<br />
The appropriate method for dynamical<br />
investigation is based on analysis of the<br />
carbon spin-lattice relaxation rate, Ric.<br />
The experimental Ric> calculated for<br />
protonated carbons, are reported in Table 1.<br />
From these data and using the Allerhand's<br />
approach 1 3 a unique correlation time<br />
value of 3x10'^ s was calculated. In Table 1<br />
the selective and non-selective proton<br />
Bulletin of Magnetic Resonance<br />
relaxation rates are also reported. These<br />
experimental values confirm the dynamical<br />
region of the isotropic molecular motion of<br />
folic acid in solution. A second set of<br />
structural information can be derived from<br />
the study of the extent of the dipolar<br />
magnetization transfer in the protonic<br />
environment.<br />
In this case the NOESY spectrum can<br />
provide the complete network of the<br />
proton-proton dipolar interactions, which<br />
is related to internuclear distances. The<br />
analysis of the NOESY spectrum enables us<br />
to identify proton pairs in which the<br />
cross-relaxation contribution is<br />
significant. These include the NH(ifj)-<br />
H(12/16). NH(io)-H(9), H(i2/16)-H(9),<br />
H(12/16)-H(13/15) and H( i 8)-H(2 lb).<br />
Different cross-peaks due to exchange<br />
contributions between two different sites<br />
can also be detected in the NOESY spectrum.<br />
These cross-peaks are related to the<br />
NH(1O)/NH2(2) — HOD exchange.<br />
Information on the extent of dipolar<br />
interactions can be used as experimental<br />
"constraints" in theoretical energy<br />
minimization calculations. The presence of<br />
an exchange process selectively restricted<br />
to the NH(iO)/NH2(2) — HOD protons<br />
suggests the involvement of other<br />
exchangeable nuclei such as NH(i8) in non<br />
covalent interactions (e.g. hydrogenbonds),<br />
important for the stabilization of<br />
the conformation of folic acid in solution.<br />
Further evidence of the slow chemical<br />
exchange process of NH(J8) with respect to<br />
NH (10)/NH2(2) can be obtained from<br />
saturation transfer experiments.<br />
By irradiating the HOD resonance for a<br />
sufficient period of time, with a selective<br />
frequency, a strong reduction in signal<br />
intensity is observed on protons involved<br />
in exchange phenomena. The experimental<br />
findings show that the NH(1O)/NH2(2) .<br />
signal is drastically affected by HOD ;<br />
saturation whereas NH(18) does not show ]<br />
significant intensity variation. This is<br />
further evidence of the importance of |<br />
NH(i8) in the stabilization of selective ;<br />
conformation by hydrogen bond with the J<br />
C(23) carboxyl group of glutamic acid. Thi<br />
hypothesis is confirmed by the selectivity j<br />
in NH(i8)-H(2l) dipolar connectivities,<br />
observed in the NOESY spectrum. In fact t<br />
specific NH(i8)-H(21b) cross-peak<br />
observed, suggesting the stabilization of ^<br />
unique conformation in solution.
Vol. 14, No. 1-4<br />
F2 (PPM)<br />
9 8 7 6 5 4 3 2 }<br />
Figure 2: COSY spectrum of 0.12 mol.dm-3 folic acid DMSO-d6 solution at 27°C.<br />
12S 1M<br />
Figure 3: 1H-13C Hetero correlation<br />
' solution at 27°C.<br />
I I '—I—I—|—i—t-<br />
spectrum of folic acid in DMSO-d6<br />
183
184 Bulletin of Magnetic Resonance<br />
Table 1<br />
Proton and carbon NMR parameters of 0.12 mol.dm'3 folic acid solution at 27°C.<br />
Nuclei<br />
2<br />
4<br />
4a<br />
6<br />
7<br />
8a<br />
9<br />
10<br />
11<br />
12<br />
13<br />
14<br />
15<br />
16<br />
17<br />
18<br />
19<br />
20<br />
21A<br />
21B<br />
22<br />
23<br />
8 ppm<br />
8.75<br />
. .<br />
4.59<br />
7.02<br />
6.74<br />
7.75<br />
7.75<br />
6.74<br />
8.22<br />
4.44<br />
2.15<br />
2.01<br />
2.42<br />
—<br />
8 ppm<br />
156.160<br />
161.274<br />
127.945<br />
148.610<br />
148.610<br />
153.823<br />
45.922<br />
150.793<br />
111.216<br />
128.998<br />
121.321<br />
128.998<br />
111.216<br />
166.438<br />
51.762<br />
173.744<br />
26.045<br />
26.045<br />
30.439<br />
173.932<br />
Further evidence of the structure assumed<br />
by folic acid can be obtained from<br />
molecular mechanics calculations using<br />
the Macro Model program and experimental<br />
NMR constraints, a low energy<br />
conformation of -135.5 KJ/mol was<br />
computed after several iterative and<br />
minimizzation cycles.<br />
Figure 4: Solution structure of folic acid<br />
as determined by NMR experimental<br />
constraints and subsequent energy<br />
minimization calculations.<br />
Ric<br />
s-1<br />
0.11<br />
0.21<br />
0.12<br />
0.55<br />
11.9<br />
0.73<br />
6.12<br />
5.70<br />
0.43<br />
5.70<br />
6.12<br />
0.38<br />
6.07<br />
0.37<br />
11.75<br />
11.75<br />
11.60<br />
0.30<br />
RlNS<br />
s-1<br />
0.77<br />
5.40<br />
2.80<br />
2.00<br />
1.87<br />
1.87<br />
2.00<br />
4.71<br />
1.62<br />
5.50<br />
5.55<br />
5.00<br />
—<br />
RlSE<br />
s-1<br />
0.75<br />
--<br />
5.30<br />
1.46<br />
1.34<br />
1.34<br />
1.46<br />
1.17<br />
5.40<br />
5.40<br />
4.90<br />
The minimized structure calculated (Figure<br />
4), shows a NH(18)-C(i9)-C(21)-H(21b)<br />
torsion angle of 315°. This is in agreement<br />
with the experimental NOESY data and<br />
confirms the conformation of folic acid<br />
stabilized in solution by the NH(i8)-C(23)<br />
hydrogen bond.<br />
4 References<br />
1) R.L. Blakley; "The Biochemistry of Folic<br />
Acid and Related Pteridines"; North-<br />
Holland publishers, Amsterdam, (1969).<br />
2) D. Mastropaolo, A. Camerman and H.<br />
Camerman; Science 210, 334 (1980).<br />
3) J.M. Blaney, C. Hansch, C. Silipo and A.<br />
Vittoria; J. Am. Chem. Soc. £4, 303 (1984).<br />
4) A.D. Bax, R. Freeman and G. Morris; J-<br />
Magn. Reson. 42, 169 (1981).
Vol. 14, No. 1-4<br />
5) D.J. States, R.A. Haberkorn and D.J.<br />
Ruben; J. Magn. Reson. 4JL 286 (1982).<br />
6) A.D. Bax and G. Morris; J. Magn. Reson.<br />
42, 501 (1981).<br />
7) C. Still; Macromodel, Columbia Universty<br />
Molecular Modelling System (1987).<br />
8) S. Weiner, P.A. Kollman, D.A. Case, U.C.<br />
Singh, C.Ghio, G. Alagona and P.K. Weiner;<br />
J. Am. Chem. Soc. 106, 765 (1984).<br />
185<br />
9) W. Frick, R. Weber and M. Viscontini;<br />
Helv. Chim. Acta 52, 2658 (1974).<br />
10) M. Poe; Method in Enzymology 6JL 483<br />
(1980).<br />
11) N. Niccolai, C. Rossi, V. Brizzi and W.A.<br />
Gibbons; J. Am. Chem. Soc. 106, 5732 (1984).<br />
12) C. Rossi, N. Niccolai and F. Laschi; J.<br />
Phys. Chem. 9_i, 3903 (1987).<br />
13) A. Allerhand, R.A. Komoroski; J. Am.<br />
Chem. Soc. 9_5_, 8828 (1973).
186<br />
Bulletin of Magnetic Resonance<br />
Characterization of Water-in-Bitumen Emulsions<br />
in Model Porous Media by NMR Microscopic Imaging Techniques<br />
1. Introduction<br />
Leslie H. Randall and George E. Sedgwick<br />
Alberta Research Council,<br />
Oil Sands and Hydrocarbon Recovery Division<br />
PO Box 8330, Station F, Edmonton, Alberta, T6H 5X2.<br />
Production from thermal (steam enhanced) oil<br />
recovery processes is-complicated by the presence of<br />
water-in-oil emulsions. [1], [2] Critical to monitoring<br />
the in-situ formation and flow of these water-in-oil<br />
emulsions is the ability to distinguish between the<br />
various components (bitumen, water/steam and<br />
emulsified water) present in a core flood experiment.<br />
In principle, NMR imaging is ideally suited to<br />
monitor the spatial distribution of absorbed fluids, and<br />
in this regard, the viability of the NMR imaging<br />
technique to examine the distribution of fluids in<br />
reservoir rock samples has recently been<br />
demonstrated. [3] - [15]. In general, several fluids or<br />
phases may be present and the ability to distinguish<br />
these components is of prime interest.<br />
In the majority of these studies, the NMR<br />
imaging technique has been applied to samples which<br />
contain a low viscosity crude oil and water/brine. In<br />
the case of heavy oil production, the fluids have a<br />
high and varying viscosity, (ie: bitumen, water and<br />
emulsions of bitumen and water). Under these<br />
circumstances, it is possible that these differences in<br />
viscosity will lead to the ability to discriminate<br />
between phases via differences in NMR relaxation<br />
behaviour. In addition, the influence that the solid<br />
matrix has on the relaxation behaviour of water<br />
dispersed as an emulsion may be quite different from<br />
water absorbed into the porous medium. In an effort<br />
to evaluate the feasibility of characterizing water-inbitumen<br />
emulsions by NMR microimaging techniques,<br />
and<br />
Colin A. Fyfe<br />
University of British Columbia,<br />
Dept. of Chemistry and Pathology,<br />
Vancouver, B.C., V6T 1Y6.<br />
the one-dimensional NMR spectra, the relaxation time<br />
constants and the spin-echo images for a series of samples<br />
consisting of water, bitumen and water-in-bitumen emulsions<br />
absorbed into glass beads were examined and are presented<br />
herein.<br />
2. Experimental<br />
All water-in-oil (bitumen) emulsions were prepared from Cold<br />
Lake bitumen which has been ultracentrifuged to remove solid<br />
particles. The emulsion samples were created by passing a<br />
heated mixture of bitumen and water through an auxiliary sand<br />
column at a suitable flow rate. The emulsions were checked<br />
under an optical microscope to ensure that the water phase was<br />
well dispersed prior to packing. Eight samples were prepared<br />
each consisting of a different fluid/porous medium mixture.<br />
Sample 1 contained 10 mL of Cold Lake Bitumen. Sample 2<br />
contained 10 mL of a 20 % (w/w) water-in-bitumen. The<br />
remaining samples contained fluid absorbed into a glass bead<br />
matrix. Two different sizes of glass beads were used. The<br />
small glass beads were determined to be 88 -104 pm in<br />
diameter which represents fine sand. The large glass beads<br />
were 0.8 - 1.0 mm in diameter. Sample 3 contained<br />
approximately 10 g of the large glass beads with 5 mL ol<br />
distilled water. Sample 4 had a similar composition to sampl<<br />
3 except that the smaller glass beads were used. Sample -<br />
contained 10 g of the large glass beads with 5 mL of a water<br />
in-oil emulsion (20% w/w water). Sample 6 had a simila<br />
composition except that the smaller glass beads were used. T<<br />
ensure that the composition of the water-in-bitumen emulsio)<br />
was maintained in the glass bead matrix, samples 5 and 6 wep<br />
prepared by mixing the emulsion with the glass beads by haflj
Vol. 14, No. 1-4 187<br />
and then placing the appropriate amount of the<br />
mixture at the bottom of a 10 mm NMR tube. The<br />
samples were then spun at low speeds on a bench-top<br />
centrifuge for 10 minutes to obtain a uniform packing.<br />
In an effort to compare the NMR behaviour of the<br />
two types of fluids directly, Sample 7 was composed<br />
of two regions, the bottom layer contained emulsion<br />
in large glass beads and the top layer contained<br />
distilled water in large glass beads. Sample 8 was<br />
identical in composition to sample 7 except that the<br />
smaller glass beads were used. The contents of the<br />
samples are summarized in Table 1.<br />
Table 1. Composition of Samples<br />
Sample Fluid Matrix<br />
1 Bitumen<br />
2 Emulsion<br />
3 Distilled Water<br />
4 Distilled Water<br />
5 Emulsion<br />
6 Emulsion<br />
7 Emulsion/Water<br />
8 Emulsion/Water<br />
None<br />
None<br />
88 - 104 urn<br />
0.8 - 1.0 mm<br />
0.8 - 1.0 mm<br />
88 - 104 urn<br />
0.8 - 1.0 mm<br />
88 - 104 um<br />
All NMR measurements were made on a Bruker<br />
MSL 400 spectrometer equipped with a microimaging<br />
system using the proton microimaging probe<br />
equipped with a vertical 12 mm saddle coil. The<br />
nonselective 90° rf pulse length was 14.5 ps.<br />
Quadrature phase cycling was used in all the<br />
spectroscopic measurements. ID J H NMR spectra and<br />
Carr-Purcell spin-echo (90-tau-180) NMR spectra [16]<br />
were obtained to characterize the samples. The spinecho<br />
sequence was also used to determine the average<br />
spin-spin relaxation times (T2). The inversionrecovery<br />
sequence [17] was used to determine the<br />
average Tj spin-lattice relaxation times.<br />
After evaluating the relaxation time constants<br />
and the NMR lineshapes for several of the samples,<br />
was determined that the spin-echo imaging<br />
luence [18] was an appropriate choice. The spinio<br />
imaging pulse sequence employs a 90-tau-180 rf<br />
sequence in which a hard 180° pulse sequence<br />
* 110 refocus ^ eff ects of field inhomogeneity.<br />
selection was performed by using a selective<br />
Pulse with an appropriate Gz gradient. Echo<br />
fes for the spin-echo imaging experiments varied<br />
from 4.5 ms to over 100 ms and the actual echo times are<br />
indicated in the text. The slice thickness was typically 2.1<br />
mm. The phase encoding gradient was incremented through<br />
256 experiments. The frequency encode gradient was 5.8<br />
G/cm resulting in an in-plane resolution of 95 um. The<br />
samples used in this study had a porosity of approximately 30<br />
-35 % and the imaging experiments required 1-4 hours to<br />
acquire. Images presented in this paper follow the convention<br />
in which an inverse gray scale is used to indicate relative<br />
intensity. The darker the region on the image, the higher the<br />
concentration of water.<br />
3. Results and Discussion<br />
The relaxation behaviour and linewidths were investigated<br />
(Table 2) to determine the appropriate imaging sequence for<br />
the bitumen and water-in-bitumen emulsions. A ID *H NMR<br />
spectrum of Cold Lake Bitumen consisted of a large peak (Vj^<br />
= 690 Hz) which is assigned to the heavy oil component<br />
(bitumen) and a small shoulder which is attributed to trace<br />
amounts of connate water. The spin-spin relaxation time (T2)<br />
of the oil component was determined to be 1.3 ms which<br />
indicates that quantitative spin-echo imaging of the bitumen<br />
component in the samples will not be possible.<br />
The linewidth determined for distilled water placed in the<br />
small glass beads (Sample 3) is approximately 1000 Hz, an<br />
increase of more than 2 orders of magnitude over that observed<br />
for a distilled water phantom. Similar line broadening was<br />
observed for the other samples (Table 2). The line broadening<br />
is a result of the magnetic susceptibility differences between<br />
the solid matrix and the absorbed fluids. These line-widths<br />
indicate that the application of the gradient echo sequence [18]<br />
would be difficult due to the short T2* values (T2* = 300-800<br />
ps). The spin-spin relaxation parameter for the distilled water<br />
absorbed into either glass bead matrix (Table 2) is markedly<br />
reduced from that observed for bulk water. An NMR image<br />
of Sample 4 with an echo time of 4.5 ms demonstrates that the<br />
distribution of water in these types of samples are possible.<br />
(Figure la) The image with an echo time of 104 ms (Figure<br />
lb) is more interesting. It shows a large decrease in signal<br />
intensity, particularly where the distilled water is in contact<br />
with the glass beads. Due to the non-uniform packing of the<br />
large glass beads, small pockets of water on the order of 100 -<br />
400 um in diameter can be observed.<br />
An examination of the NMR relaxation parameters of the<br />
20% (w/w) water-in-bitumen emulsion (Sample 5) absorbed<br />
into the large glass bead matrix reveals that the T2 relaxation<br />
time constant of the water phase is unaffected by placing the<br />
emulsion into the glass bead matrix. The image obtained with<br />
a 104 ms echo time (Figure 2) displays a strong uniform signal<br />
intensity. The bitumen component of the sample has a T2<br />
relaxation time constant on the order of a millisecond, and thus<br />
does not contribute to the intensity of the image.
i ii<br />
'I,' , ',<br />
il i I<br />
i i<br />
188 Bulletin of Magnetic Resonance<br />
Figure 1 (a) Spin echo images of sample 4.<br />
a) Echo time = 4.5 ms. (b) Echo time = 104 ms<br />
During a core-flood experiment it is likely that both<br />
water and emulsified water phases will be present.<br />
The ability to distinguish between bulk absorbed<br />
water and emulsified water is considered essential to<br />
analyzing the formation and flow of the emulsions in<br />
sand packs. To this end, a sample which contains<br />
both distilled water and water-in-bitumen emulsion in<br />
a glass bead pack was examined (Samples 7 and 8).<br />
The conditions of the imaging experiments were<br />
chosen such that only the water component of the<br />
samples were observed. Using an echo time of 4.5<br />
ms the distribution of water in sample 7 (Figure 3a)<br />
can be obtained. The distilled water plus large glass<br />
beads are in the top half of the NMR tube and this<br />
results in a much higher signal intensity due to the<br />
higher water concentration. To discriminate between<br />
the emulsified water phase and the absorbed bulk<br />
water, an echo time of 104 ms was used. (Figure 3b)<br />
Under these conditions, only water which has a high<br />
mobility or low surface contact will be observed (ie:<br />
water which is emulsified will be favoured). This<br />
results in an image in which small pockets of water<br />
are observed in the distilled water region, while the<br />
emulsion containing region appears nearly uniform.<br />
The effect of the smaller grain size of the<br />
relaxation parameters was examined using sample 8.<br />
The smaller glass beads have a more uniform pore<br />
size distribution and are more representative of the<br />
matrix used in core flood experiments. In such<br />
samples, the pore size is typically on the order of 30<br />
pm which means that all of the water will be in<br />
intimate contact with the solid matrix.<br />
Figure 2. Spin-echo image of sample 5.<br />
Echo time = 104 ms.<br />
Figure 3. Spin echo images of sample 7.<br />
a) Echo time = 4.5 ms. (b) Echo time = 104 ms<br />
The T2 of the distilled water component is reduced to 4.9 ms,<br />
on the same order of magnitude as the echo time (4.5 ms).<br />
This means that the intensity of the water will be strongly T2<br />
weighted and the region which contains the distilled water will<br />
no longer have an intensity which is much higher than the<br />
emulsion. (Figure 4a) When the echo time is increased to 34<br />
ms, (Figure 4b) the region which contains the distilled water<br />
disappears in a uniform manner. The smaller grain size of the<br />
solid matrix has enlarged the relaxation time differences<br />
between the bulk absorbed water and the emulsified water and<br />
thus the contrast between the two physical states are enlarged.
Vol. 14, No. 1-4 189<br />
Figure 4. Spin echo images of sample 8.<br />
a) Echo time = 4.5 ms. (b) Echo time = 34.5 ms<br />
3. Summary<br />
In the present study, the feasibility of examining<br />
heavy oil emulsion samples has been demonstrated.<br />
Bitumen, water and emulsified water placed into glass<br />
beads are easily distinguished in an NMR imaging<br />
experiment on the basis of their relaxation times. The<br />
spin-spin relaxation time constant, T2 observed for the<br />
water component of the emulsion samples placed in<br />
contact with glass beads was dramatically different<br />
than that found for distilled water. The NMR<br />
experiment is therefore sensitive to the physical state<br />
of the water, ie: whether the water is emulsified and<br />
therefore 'protected' from the solid matrix. This<br />
difference in relaxation times was exploited to provide<br />
water-selective images of the emulsified water phase<br />
in the presence of oil and non-emulsified water<br />
phases. Good S/N images can be obtained with high<br />
in-plane resolution (50-100 pm) can be obtained on a<br />
microimaging system in reasonable time periods.<br />
Saturation profiles can be obtained in a manner of<br />
seconds which would allow for the continuous<br />
monitoring of the formation of an emulsion under<br />
flow conditions. The ability to monitor the formation<br />
and flow of water emulsions will be important to<br />
understanding enhanced oil recovery processes.<br />
Table 2: Proton Relaxation Times and Linewidths of Water<br />
and Bitumen<br />
Sample<br />
1<br />
2 water<br />
bitumen<br />
3<br />
4<br />
5 water<br />
bitumen<br />
6 water<br />
bitumen<br />
Ti(s)<br />
0.71<br />
2.2<br />
0.62<br />
2.4<br />
2.3<br />
1.6<br />
0.58<br />
1.5<br />
0.45<br />
T2 (ms)i<br />
1.3<br />
200*<br />
1.8<br />
29<br />
4.9<br />
495<br />
1.3<br />
484<br />
1.4<br />
v l/2<br />
690<br />
150<br />
650<br />
350<br />
1000<br />
The water component of emulsion samples show a variation<br />
in T2. The reasons for this variation are under investigation<br />
but are believed to be due to the size distribution of the water<br />
droplets.<br />
Acknowledgements<br />
The authors wish to acknowledge G. A. Kissel and D. Vu of<br />
the Alberta Research Council for the invaluable technical<br />
assistance in preparing the samples. The financial assistance<br />
of NSERC is acknowledged (CAF).<br />
4. References<br />
1. Thermal Recovery of Oil and Bitumen, Butler, R. B.<br />
Prentice Hall, New Jersey, 1991.<br />
2. Emulsions; Fundamentals and Applications in the Petroleum<br />
Industry, Schram, L. L., Ed. American Chemical Society, 1992.<br />
3. Rothwell, R. P.; Vinegar, H. J. Applied Optics, 24, 1985,<br />
3969-3972.<br />
4. Blackband, S.; Mansfield, P.; Barnes, J. R.; Clague, A. D.<br />
H.; Rice, S. A. Soc. Pet. Eng. Form. Evaln.l, 1986, 31-34.<br />
5. Hall, L. D.; Rajanayzgmi, V.; Hall, C. /. Magn. Res. 68,<br />
1986, 185-188.<br />
6. Baldwin, B. A.; Yamanashi, W. S. Mag. Res. Imaging, 6,<br />
1988, 493-500.
190 Bulletin of Magnetic Resonance<br />
7. Hall, L. D.; Rajanayagmi, V. /. Magn. Res. 74,<br />
1987, 139-147.<br />
8. Chen, J. D.; Dias, M. M.; Patz, S.; Schwartz, L. M.<br />
Phys. Rev. Lett. 61, 1988, 1489-1492.<br />
9. Edelstein, W. A.; Vinegar, H. J.; Tutunjian, P. N.;<br />
Roemer, P. B.; Mueller, O. M. SPE preprint 18272,<br />
63rd Annual Technical Conference, Houston, 1988,<br />
101-112.<br />
10. Mandava, S. S.; Watson, A. T.; Edwards, C. M.<br />
Amer. Inst. Chem. Eng. 36, 1990, 1680-1686<br />
11. Majors, P. D.; Smith, J. L.; Kovarik, F. S.;<br />
Fukushima, E. /. Magn. Res. 89, 1990, 470-478.<br />
12. Woessner, D. E.; Gleeson, J. W.; Jordan, C. F.<br />
SPE preprint 20493, 65th Annual Technical<br />
Conference, New Orleans, 1990, 247-253.<br />
13. Osment, P. A.; Packer, K. J.; Taylor, M. J.;<br />
Attard, J. J.; Carpenter, T. A.; Hall, L. D.; Herrod, N.<br />
J.; Doran, S. J. Phil. Trans. R. Soc. Lond. A. 333,<br />
1990, 441-452<br />
14. Dereppe, J. M.; Moreaux, C; Schenker, K. J.<br />
Magn. Res. 91, 1991, 596-603.<br />
15. Dechter, J. J.; Komoroski, R. A.; Ramaprasad, S.<br />
/. Magn Res. 93, 1991, 142-150.<br />
16. H. Y. Carr and E. M. Purcell, Phys. Rev., 94,630<br />
(1954).<br />
17. R. L. Void, J. S. Waugh, M. P. Klein and D. E.<br />
Phelps, J. Chem. Phys., 48, 383 (1968).<br />
18. Edelstein, W.A.; Hutchinson, J.M.S.; Johnson G.;<br />
Redpath, T. Phys. Med. Bioi, 25, 751 (1980).<br />
19. A. Haase, J. Frahm, D. Matthaei, W. Hanicke and<br />
K. D. Merboldt, /. Magn. Reson., 67, 258 (1986).
Vol. 14, No. 1-4 191<br />
COMPUTER GRAPHICS FOR PULSE SEQUENCE<br />
ANALYSIS<br />
Introduction<br />
Jonathan Callahan, Debbie Mattiello and Gary P. Drobny<br />
The development of software for<br />
the simulation of NMR experiments<br />
has increased at an enormous pace in<br />
the last decade. Its usefulness has<br />
spread far beyond the analysis of<br />
lineshapes and spectra. Today,<br />
search-and-optimize strategies are<br />
used to develop new pulse sequences<br />
while other programs measure the<br />
performance of pulse sequences on<br />
spin systems of interest [1, 2]. Up to<br />
now, most of the output from these<br />
programs has been displayed as twodimensional<br />
hard-copy output. With<br />
the arrival of relatively inexpensive<br />
graphics workstations, the possibility<br />
of visualizing the time development of<br />
the density operator has spurred us to<br />
develop graphics software in<br />
conjunction with our ongoing<br />
development of simulation software.<br />
Current projects which benefit<br />
from graphical analysis include the<br />
development of "time-suspension"<br />
sequences for use with solids imaging,<br />
development of improved sequences<br />
for the creation of Zeeman or<br />
quadrupolar order in deuterium NMR,<br />
and analysis of artefacts seen in<br />
imaging experiments in the presence<br />
of flow.<br />
Chemistry Dept. University of Washington<br />
Seattle, Washington 98195 USA<br />
Time Suspension<br />
"Time suspension" sequences are<br />
multi-pulse sequences which remove<br />
the effects of both the dipolar coupling<br />
and the chemical shift Hamiltonians<br />
[3-5]. They are useful in pulsedgradient<br />
imaging experiments where<br />
imaging gradients are applied only<br />
during the multi-pulse windows [6-8].<br />
Currently implemented pulse<br />
sequences apply average Hamiltonian<br />
theory and use carefully cycled<br />
"wahuha" subcycles to achieve the<br />
suppression of internal Hamiltonians<br />
[3]. Like other multi-pulse<br />
experiments these sequences perform<br />
best at or near resonance and show<br />
decreased line-nawowing as one<br />
moves off resonance. A new "timesuspension"<br />
sequence developed with<br />
computer search-and-optimization<br />
techniques (CDIS-4) [9] achieves<br />
comparable reduction of the internal<br />
Hamiltonians but shows<br />
complementary behavior: poor line<br />
narrowing on resonance but improved<br />
performance off resonance. In an<br />
effort to understand this behavior we<br />
calculate the evolution of the density<br />
operator for a two-spin system under<br />
the influence of effective Hamiltonians<br />
defined by the multi-pulse
192 Bulletin of Magnetic Resonance<br />
propagators associated with each<br />
sequence.<br />
With perfect wahuha type<br />
sequences the trajectory of the net<br />
magnetization should be around the<br />
base of a cone whose axis is along the<br />
cube diagonal in spin space (ie. in the<br />
nodal plane of the coupling ternsor).<br />
When chemical shift offsets are small<br />
(
Vol. 14, No. 1-4 193<br />
greater than 10KHz the conventional<br />
sequence shows modulated rotation<br />
(Fig. 1C) whereas the new sequence<br />
traces out a somewhat bent circle<br />
lying in the nodal plane of the<br />
chemical shift tensor (Fig.ID). From<br />
such pictures we hope to gain a better<br />
understanding of the effect of error<br />
terms which are not easily amenable<br />
to analytical treatment.<br />
Deuterium<br />
The deuterium quadrupole is an<br />
excellent probe of dynamics and as<br />
such is synthetically incorporated into<br />
DNA and other biologically important<br />
molecules [10]. The same quadrupole<br />
which allows one to use deuterium as<br />
a probe also presents formidable<br />
experimental difficulties when the<br />
strength of the quadrupole approaches<br />
the strength of the rf field. This<br />
situation is realized in some of our<br />
labeled oligonucleotides. In this<br />
regime, the effective axis about which<br />
the magnetization is rotated (the sum<br />
of rf and quadrupolar terms) is<br />
substantially different from the rf<br />
axis.<br />
Current pulse sequences which<br />
convert Iz magnetization into -Iz or<br />
into quadrupolar order are found to<br />
be inefficient at higher values of the<br />
quadrupole coupling. For an inversion<br />
pulse this means incomplete inversion<br />
at the shoulders of the powder<br />
pattern. This is unacceptable in<br />
experiments where one attempts to<br />
measure the orientation dependence<br />
of spin-lattice relaxation time. Such<br />
measurements are necessary to prove<br />
or disprove particular dynamical<br />
models.<br />
An example of non-uniform<br />
excitation is given in Fig. 2) which<br />
shows the development of spin<br />
coherences during a 180 degree pulse<br />
at five different values of the<br />
quadrupole coupling (OKHz, +/- 62.5<br />
KHz and +/- 125KHz). The operator<br />
basis of Vega and Luz [11] is used<br />
because of the simple form of the<br />
quadrupolar operator in that basis:<br />
Ix<br />
Jx=IyIZ+IzIy<br />
Kx=Iy 2 -Iz 2 Ky=Iz IVy=lZ —1X<br />
x-lx -1 ^y-Iy -1<br />
2 -Ix 2 K2=Ix2-Iy 2<br />
QZ=Iz 2 -I 2<br />
Rotation about Ix during the<br />
pulse is modified by (0q dependent<br />
rotation about the Qz axis. This causes<br />
a buildup of "antiphase" Jy and zeroquantum<br />
Jz. By the end of the pulse it<br />
is clear that a simple n pulse is<br />
ineffective at creating -Iz over the<br />
entire range of couplings (+/-125 KHz).<br />
We are again using search and<br />
optimize strategies to find pulse<br />
sequences which create -Iz or Qz<br />
evenly over a broad range of<br />
quadrupole couplings [12]. With its<br />
small operator basis, deuterium NMR<br />
provides us with an excellent system<br />
on which to develop interactive<br />
computer aided pulse sequence<br />
design. With user control of pulse<br />
amplitude and phase, rapid calculation<br />
of the time evolution of the spin 1<br />
density operator and a graphical view<br />
of Hilbert space we will soon have the<br />
opportunity to design pulse<br />
sequences and shaped pulses<br />
intelligently rather than relinquishing<br />
our insight to the cpu.
194<br />
Qz<br />
\lx<br />
Jx<br />
Iz<br />
iy<br />
Bulletin of Magnetic Resonance<br />
Figure 2) Evolution of spin coherences for deuterium with ()(•), +/- 62.5(AY) and +/-<br />
KHz quadrupole coupling. The evolution is depicted in the operator basis of Vega and Lus<br />
Rotation about Ix due to the rf pulse is seen but rotation about quadrupole operator Qz is als<br />
evident. For +/- 125 KHz coupling the simple K pulse is very inefficient at creating -Iz.<br />
Flow Imaging<br />
Another simulation program<br />
which benefits from graphical display<br />
calculates the response of flowing<br />
spins in an NMR imaging experiment.<br />
Our interest in this area focuses on the<br />
artefacts that arise when spins move<br />
from a region with one gradient<br />
strength to a region with anothe<br />
during an imaging experiment. Ou<br />
simulation incorporates the effects c<br />
flow and of arbitrarily comple<br />
imaging sequences on an array c<br />
spins 1/2. When following an<br />
particular spin through time we kee<br />
track of its three-dimensional positio<br />
and its magnetization vector. Thus w
Vol. 14, No. 1-4 195<br />
have for each spin seven parameters<br />
(3 space, 3 spin and time) which<br />
describe its state.<br />
In order to understand how<br />
artefacts develop during the<br />
experiment we display the system as<br />
an array of vectors whose position<br />
corresponds to spatial position and<br />
whose orientation corresponds to<br />
magnetization state. From the z-axis<br />
we observe spatial flow of spins in the<br />
spatial x-y plane and also the<br />
magnitude and phase of transverse<br />
magnetization in the superposed spin<br />
x-y plane.<br />
A<br />
B<br />
c<br />
/*-<br />
In Fig. 3) we see how the sliceselect<br />
portion of an imaging<br />
experiment can be perturbed by flow<br />
in the direction of the slice-select<br />
gradient. As the rate of flow<br />
increases, the width of the slice<br />
remains fairly constant but the phase<br />
order of the spins in that slice<br />
deteriorates. Such phase disorder will<br />
lead to artefacts along the phaseencode<br />
dimension at the edges of the<br />
slice. As it now exists our simulation<br />
will allow us to evaluate imaging<br />
sequences on flow geometries<br />
Flow / Imaging Gradient<br />
Figure 3) Excitation profiles for the slice-select portion of an imaging experiment. In this<br />
simulation the direction of flow is along the slice-select gradient causing spins to change<br />
their frequency during the sine excitation pulse. The flow velocities are in arbitrary units<br />
but three regimes are displayed: A) static spins; B) slow flow; C) moderate flow.
iii.:<br />
196<br />
of interest. A better understanding of<br />
the formation of such errors will aid<br />
us in the development of improved<br />
imaging sequences.<br />
Conclusion<br />
We have extended the computer<br />
techniques available to the NMR<br />
spectroscopist by presenting<br />
simulated data in multi-dimensional<br />
animations. With these animations it<br />
is much easier to see the development<br />
of spin coherences which are the<br />
result of experimental imperfection or<br />
which are intended by design. Our<br />
original goal with computer graphics<br />
was to enhance our own<br />
understanding of the experiments we<br />
perform and to aid us in experimental<br />
development In the process we have<br />
found animated simulations to be a<br />
generally useful pedagogical tool for<br />
explaining all types of NMR<br />
experiments. With computer designed<br />
pulse sequences containing unusual<br />
phases and non-analytical shapes<br />
becoming more and more common we<br />
hope to bring some of the intuition<br />
back to experimental design.<br />
References<br />
[1] S. J. Glaser and G. P. Drobny,<br />
.Adv. Mag. Res., 14, 35 (1990)<br />
[2] H. Liu, S. J. Glaser, and G. P.<br />
Drobny, J. Chem. Phys., 93(111),<br />
7543 (1990)<br />
[3] D. G. Cory, J. B. Miller, and A. N.<br />
Garroway, /. Mag. Res.,9Q, 205<br />
(1990)<br />
Bulletin of Magnetic Resonance<br />
[4] P. Caravatti, L. Braunschweiler,<br />
and R. R. Ernst, Chem. Phys. Lett.,<br />
100(4), 305 (1983)<br />
[5] P. Mansfield and P. K. Grannell,<br />
Phys. Rev. B., 12(9), 3618<br />
(1975)<br />
[6] D. G. Cory and W. S. Veeman, J.<br />
Mag. Res., 84, 392 (1989)<br />
[7] J. B. Miller, D. G. Cory, and A. N.<br />
Garroway, Chem. Phys. Lett.,<br />
164(1), 1 (1989)<br />
[8] J. B. Miller, D. G. Cory, and A. N.<br />
Garroway, Philos. Trans. R. Soc.<br />
London, Ser. A., 333(1632), 413<br />
(1990)<br />
[9] J. Iwamiya, S. Sinton, J. Callahan,<br />
and G. P. Drobny, in abstracts of<br />
the 33 rd ENC, Asilomar, CA USA,<br />
221 (1991)<br />
[10] T. M. Alam and G. P. Drobny,<br />
Chem. Rev., 91, 1545 (1991)<br />
[11] A. J. Vega and Z. Luz, /. Chem.<br />
Phys., $6, 1803 (1987)<br />
[12] D. Mattiello, J. Callahan, T. M<br />
Alam, and G. P. Drobny, in<br />
Proceedings of the 13 th <strong>ISMAR</strong><br />
Meeting, Vancouver, B.C Canada<br />
(1992)
Vol. 14, No. 1-4 197<br />
NMR INVESTIGATION OF THE SIMULTANEOUS<br />
FERMENTATION OF XYLOSE AND GLUCOSE BY A<br />
SELECTED STRAIN OF KLEBSIELLA PLANTICOLA (Gil).<br />
C.Rossi*, A.Lepri*, M.P.Picchi*, S.Bastianoni*, D.Medaglini 0 ,<br />
M.Vanassina 0 and E.Cresta*.<br />
•Department of Chemistry, University of Siena, Pian dei Mantellini<br />
44, 53100 Siena, ITALY.<br />
°Department of Molecular Biology, University of Siena, 53100<br />
Siena, ITALY<br />
1. INTRODUCTION<br />
The hydrolysis of hemicellulose<br />
yields a mixture of sugars of which<br />
D-xylose and D-glucose are the<br />
major constituents.O) This sugar<br />
mixture is an excellent substrate for<br />
growing microorganisms and yields<br />
high energy products such as<br />
ethanoH 2 ' 3 ). In developing this<br />
project two main problems have to<br />
be analyzed in detail: i) the isolation<br />
and identification of microorganisms<br />
whose metabolism can be sustained<br />
by the hemicellulose-derived sugar<br />
mixture, ii) the characterization of<br />
the metabolism, and the selection of<br />
specific metabolic pathways of<br />
microorganisms growing on sugar<br />
mixtures. It has already been shown<br />
that the use of selectively carbon-13<br />
enriched substrates enables the "in<br />
vivo" metabolization process of<br />
microorganisms and tissues^ 4 ' 5 ) to be<br />
studied by NMR. This technique was<br />
applied to the study of the<br />
simultaneous fermentation of xylose<br />
and glucose by a newly isolated<br />
Klebsiella planticola (G 11) strain.<br />
In the present investigation [2- 13 C]glucose<br />
and [l- 13 C]-xylose were<br />
used as carbon-13 enriched<br />
substrates. Isotopic enrichment in<br />
different positions of the sugar chain<br />
enabled us to: i) separate the xylose<br />
from the glucose signals in the<br />
carbon spectrum and ii) calculate<br />
the contribution of each sugar to<br />
end-product yield.<br />
2. EXPERIMENTAL<br />
Klebsiella planticola Gil was<br />
isolated from the soil of a corn field<br />
and selected for its capacity of<br />
growing on a mixed sugar<br />
substrate^ 6 ). Identification of the<br />
bacterium was performed on the<br />
basis of taxonomic characters and<br />
biochemical behaviour. The<br />
microorganism was cultivated in a<br />
nitrogen atmosphere at pH 7.5 and<br />
35°C on a mineral medium<br />
containing 0.2 g/1 of yeast extract.<br />
The metabolism of the bacterium<br />
was investigated by "in vivo" NMR<br />
spectroscopy, using selectively
198<br />
carbon-13 enriched sugar<br />
substrates. The cell culture used for<br />
microbatch NMR experiments had an<br />
initial Optical Density (O D) of 0.5.<br />
The microorganism was grown on a<br />
NMR coaxial tube containing D2O as<br />
lock signal on the external section<br />
and positioned permanently in the<br />
magnetic cavity.during fermentation<br />
[l-^CJ-xylose and R-^CJ-glucose<br />
obtained from Cambridge Isotope<br />
Laboratories were used as enriched<br />
substrates.<br />
c)<br />
Bulletin of Magnetic Resonance<br />
The sugar metabolic process was<br />
followed by recording carbon<br />
spectra at 30 minute intervals until<br />
the end fermentation.<br />
3. RESULTS AND DISCUSSION<br />
The NMR spectra of the enriched<br />
sugar and the end-products of sugar<br />
fermentation are shown in Figure 1.<br />
The metabolization of xylose and<br />
glucose followed two different and<br />
iBU|iiii|Dii|iuuiiiijiiii|uiijuii|au|iiiiiniijiiiiiau|iiiiiiiii|iiiijuii|in inii|iui IIIII|INI IIIII|IIII jaiijiiimiii|iiiiiuii|iiii iu<br />
180 160 140 120 100 80 60 40 20 PPM 180 160 140 120 100 80 60 40 20 PPM<br />
Figure 1 - 1 3 C-NMR spectra obtained at the begining (left) and at the end<br />
(right) of fermentation by K. planticola Gil. Total sugar lOg/1; pH<br />
7.5; Temperature 35 °C. a) [2- 13 C]-glucose fermentation; b) [1- 13 C]xylose<br />
fermentation and c) simultaneous [l-l 3 C]-xylose and [2-<br />
* 3 C]-glucose fermentation. End-product signals detected by * 3 C-<br />
NMR were:l) [2- 13 C]-lactic acid, 2) [2- 13 C]-ethanol, 3) formic acid,<br />
4) [2- 13 C]-succinic acid 5) [l- 13 C]-lactic acid 6) [I- 13 C]-acetic acid<br />
and 7) [l- 13 C]-ethanol.<br />
The end-products of fermentation<br />
were identified on the basis of the<br />
NMR chemical shifts. Analysis of the<br />
dependence of chemical shift on pH<br />
enabled the preliminary<br />
identification of carboxylic and non<br />
carboxylic end-products.<br />
independent pathways: "ethanol and<br />
mixed acids" for xylose( ? ) and<br />
"Emboden-Meyerhof" for glucose.<br />
Diagrams of the sugar pathways are<br />
shown in Figure 2. Glucose<br />
metabolism was also analyzed using<br />
[1- 13 C] enrichment. The same end-
Vol. 14, No. 1-4<br />
products as<br />
enrichment<br />
detected.<br />
a)<br />
for [2- 13 C]-glucose<br />
(Figure 1A) are<br />
i<br />
Xylulose<br />
5, phosphate<br />
Pentose phosphate pathway<br />
biomass have been detected. The<br />
increase in uptake rate with<br />
increasing xylose concentration in<br />
b)<br />
Fructose<br />
1,6 diphosphote<br />
Dihydroxyaceton<br />
phosphate<br />
Figure 2 - Metabolic pathways of xylose (a) and glucose (b) fermentation<br />
by Klebsiella planticola Gil.<br />
The metabolization of xylose by<br />
Klebsiella planticola Gil was<br />
interesting from the view points of<br />
the metabolization rate and the endproducts<br />
of the process. Following<br />
the xylose metabolization process by<br />
NMR in a range of sugar<br />
concentration from 5 to 100 g/1, the<br />
uptake rate can be calculated. This<br />
parameter it is very important<br />
because is correlated with the<br />
efficiency of the process of sugar<br />
transport through the cell<br />
membrane. Table 1 shows the xylose<br />
uptake rate during the first three<br />
hours of fermentation. Sugar<br />
consuption rates for Klebsiella<br />
planticola ATCC 33531, in the range<br />
of 0.8-1.6 g.l^.h" 1 per gram of<br />
AoeBc<br />
Acid<br />
our case is indicative of a "low<br />
affinity" mechanism (not previously<br />
detected in Klebsiella species)<br />
similar to the "facilitated diffusion"<br />
transport process described in<br />
Candida sheabetaeS 9 ^<br />
Figure 3 A shows the effect of<br />
glucose on the metabolization of<br />
xylose. The NMR spectra show that<br />
glucose does not interfere with<br />
xylose metabolism and its uptake<br />
rate. Figure 3B reports the results of<br />
a similar experiment, in which<br />
glucose was added after two hours<br />
of xylose fermentation. The addition<br />
of glucose did not affect xylose<br />
fermentation. It is very unusual for<br />
two substrates to be metabolized at<br />
199
200 Bulletin of Magnetic Resonance<br />
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<br />
170 165 160 155 150PPM45 95 90 85 BO 75 PPH7C<br />
Figure 3)- ^C NMR spectra recorded at different stages of sugar<br />
fermentation, a) Simultaneous fermentation of xylose<br />
and glucose b) The effect of addition of glucose after 2<br />
hours of xylose metabolization. Spectra were recorded<br />
every 30 minutes.<br />
TABLE 1<br />
Xylose uptake rate calculated from NMR spectra during the<br />
first three hours of fermentation.<br />
Xylose<br />
concentration<br />
g/1<br />
5<br />
10<br />
20<br />
40<br />
50<br />
80<br />
100<br />
a) per gram of dry weight biomass<br />
Uptake rate a<br />
g.l-i.h-l<br />
1<br />
2.1<br />
3.5<br />
4.6<br />
5.1<br />
6.2<br />
7.0
Vol. 14, No. 1-4<br />
the same time. The phenomenon has<br />
been hypothesized before but has<br />
only ever been demonstrated in<br />
Candida sheabetae. The non diauxic<br />
growth of Klebsiella planticola Gil is<br />
very important because it could be<br />
used to ferment sugar mixtures like<br />
that obtained from the hydrolysis of<br />
hemicellulose.<br />
If the property of good xylose<br />
uptake can be combined with good<br />
end-product yields by selection or<br />
genetic engineering^ 10 ' 11 ), it will<br />
constitute a further advance in the<br />
development of bioethanol<br />
production.<br />
4. REFERENCES<br />
1) T.E.Timell; Adv. Carbohydr. Chem.<br />
19, 247 (1964).<br />
2) K.Skoog and B.Hahn-Hgerdal; Enz.<br />
Microb. Technol., 10, 66 (1988).<br />
3) H.Shneider; Critical Review in<br />
Biotecnology, 9, 18 (1989).<br />
4) K.Ugurbil, T.R.Brown, J.A.Den<br />
Hollander, P.Glynn and R.Shulman;<br />
Proc. Acad.<br />
(1978).<br />
Sci. USA, 75., 3742<br />
5) J.A.Den Hollander, T.R.Brown,<br />
K.Ugurbil and R.Shulman; Proc. Acad.<br />
Sci. USA, 76, 6096 (1979).<br />
6) E.Cresta, SJez, A.Lepri A.Pisani,<br />
C.Rossi and G.Sabatini; "NMR<br />
investigation of xylose bioconversion<br />
by Klebsiella sp. strain isolated from<br />
soil". In "Biomass for Energy and<br />
Industry" G.Grassi, G.Gosse and G.dos<br />
Santos eds., Elsevier Applied<br />
Sciences, 2, 253 (1990).<br />
7) C.S.Gong, L.F.Chen, M.C.Flikinger<br />
and G.T.Tsao; Adv. Biochem. Eng.,<br />
20,93 (1981).<br />
8) J.Tolan and R.K. Finn; Appl.<br />
Environ. Microbiol., 53_, 2039 (1987).<br />
9) C.Lucas and N.van Uden; Appl.<br />
Microbiol. Botechnol., 2J3_, 491<br />
(1986).<br />
10) F.Alterthun and L.O.Ingram;<br />
Appl. Environ. Microbiol., 55_ 1943<br />
(1989).<br />
11) J.S.Tolan and R.K.Finn; Appl.<br />
Environ. Microbiol., 51 2039 (1987).<br />
201
202 Bulletin of Magnetic Resonance<br />
Interleukin-1 Receptor Antagonist Protein:<br />
Solution Secondary Structure from NOE's and<br />
1H« and 13C« Chemical Shifts<br />
Brian J. Stockman, Terrence A. Scahill, Annica Euvrard, Nancy A. Strakalaitis,<br />
David P. Brunner, Anthony W. Yem, and Martin R. Deibel, Jr.<br />
1 Introduction<br />
The Upjohn Company, 301 Henrietta St., Kalamazoo, MI 49007<br />
Interleukin-la and interleukin-ip are two<br />
polypeptides which share a significant<br />
number of inflammatory, immunological<br />
and pathological properties (for a review<br />
see [1]). Importantly, these dissimilar 17<br />
kDa proteins bind to two classes of interleukin-1<br />
receptors, resulting in the<br />
mediation of several immune and inflammatory<br />
responses and in the induction of a<br />
variety of biological changes in neurologic,<br />
metabolic, hematologic, and endocrinologic<br />
systems [1]. In addition to IL-loc and IL-1(3,<br />
an interleukin-1 receptor antagonist<br />
protein (termed either IRAP or IL-lra) has<br />
been isolated, characterized, cloned and<br />
expressed in E. coli [2-4]. This newer<br />
member of the IL-1 gene family is a<br />
naturally occurring inhibitor of the<br />
interleukin-1 receptor [2,4], and represents<br />
the first described naturally occurring<br />
cytokine that functions entirely as a<br />
specific receptor antagonist.<br />
Site-directed mutagenesis [5-7] and<br />
protein modification [6] studies have<br />
identified three regions of IL-1 that are<br />
involved in either receptor binding or<br />
transmission of the biological response<br />
upon binding. For IRAP, it can be hypothesized<br />
that the regions of structure important<br />
for receptor binding are maintained, but<br />
that the region or regions responsible for<br />
eliciting the response are somehow<br />
different. To this end, we have begun an<br />
intensive program to determine the solution<br />
structure of IRAP using NMR spectroscopy.<br />
Since the solution [8-12] and crystalline<br />
[13,14] structures of IL-lp have been<br />
determined, direct comparisons can be made<br />
between IRAP and IL-lp. This may lead to a<br />
correlation between structural and biological<br />
differences.<br />
2 Methods<br />
Expression of IRAP was carried out using E.<br />
coli K-12 strain DU379. Fermentation media<br />
were supplemented with (15NH4)2SO4, [ 13 Ci]and/or<br />
[i5N]-L-methionine, 1 5 NH4C1, and<br />
[ 13 C]-D-glucose (stable isotopes were<br />
obtained from Cambridge Isotope Laboratories,<br />
Isotec, and/or MSD Isotopes) as<br />
required to produce either 15 N- or doubly<br />
13 C/ 15 N-enriched IRAP. Analysis of<br />
resolved *H resonances indicated that both<br />
13 C and 15 N were incorporated at an<br />
enrichment level greater than 95%.<br />
Samples for NMR spectroscopy contained 2<br />
mM IRAP, 50 mM 2H4-ethanolamine and 300<br />
mM NaCl at pH 6.4. Trace amounts of PMSF<br />
and NaN3 were added to prevent any<br />
protease digestion or bacterial growth in<br />
the sample.<br />
All NMR spectra were recorded at<br />
27 °C on a Bruker AMX-600 spectrometer<br />
equipped with a triple-resonance probe and<br />
a multi-channel interface. Threedimensional<br />
1H-15N NOESY-HMQC and TOCSY-<br />
HMQC experiments were recorded with<br />
slight modification of the methods of<br />
Zuiderweg and Fesik [15] and Marion et al.<br />
[16]. Three-dimensional 1H-15N-13C HNCA<br />
and HN(CO)CA triple resonance experiments<br />
were recorded with constant-time 15 N<br />
evolution as described by Grzesiek and Bax<br />
[17]. Detailed acquisition parameters have<br />
been described elsewhere [18]. Three-
Vol. 14, No. 1-4 203<br />
6.0<br />
Figure 1. Region of the slice corresponding the 1H-13H-13C-1H TOCSY spectrum of IRAP.<br />
to 13 C frequencies of 18.7 and 51.8 ppm in Several assigned correlations are labeled.<br />
dimensional 1H-13C-13C-1H COSY [19], m<br />
13C-1H TOCSY [20], and 1H-13C NOESY-HMQC<br />
[21] experiments were also recorded.<br />
3 Results<br />
Assignment of the majority of the backbone<br />
!H, 13 c, and 15 N resonances of IRAP was<br />
accomplished by analysis of four threedimensional<br />
data sets. First, 1H-15N NOESY-<br />
HMQC and TOCSY-HMQC experiments were<br />
recorded on uniformly 15 N-enriched IRAP.<br />
Then, two *H- 15 N- 1 3 C triple resonance<br />
experiments were recorded, the so-called<br />
HNCA and HN(CO)CA experiments [22,23].<br />
Redundant sequential connectivities<br />
obtained from the heteronuclear data sets<br />
simplified and increased the reliability of<br />
the assignments. During the assignment<br />
process, NOE's indicative of secondary<br />
structure were identified.<br />
For many residues, magnetization<br />
transfer in the 1H-15N TOCSY-HMQC spectrum<br />
extended resonance assignments to at<br />
least one X HP resonance and sometimes even<br />
further down the side chain. In cases of<br />
favorable resolution, such as for high-field<br />
shifted resonances, two-dimensional DQF-<br />
COSY and TOCSY spectra confirmed and/or<br />
extended these side-chain assignments.<br />
Extensive side-chain assignments, however,<br />
will require ^C-directed strategies [19,20]<br />
and are currently in progress. A represen-<br />
tative slice from the 1H-13C-«C-»H TOCSY<br />
spectrum of IRAP is shown in Figure 1.<br />
Once the majority of correlations<br />
were assigned in the !H- 15 N HSQC spectrum<br />
recorded in J H2O, an identical spectrum was<br />
recorded after exchanging the protein into<br />
2 H2O. Only 50 1H- 15 N correlations remained<br />
after six hours in 2 H2O solvent. As discussed<br />
below, each of these residues was found to<br />
participate in the p-sheet framework of<br />
IRAP.<br />
4 Discussion<br />
During analysis of the 1H-15N NOESY-HMQC<br />
data set, NOE's indicative of the solution<br />
secondary structure [24] of IRAP were<br />
identified. The majority of these- were<br />
classified as cross-strand NOE's between<br />
residues involved in ($ -sheet structure. They<br />
are manifested in the NOESY-HMQC spectrum<br />
as a third 1 H a NOE to an amide proton (the<br />
others being the interresidue and intraresidue<br />
1 H« *s), or as weak, non-sequential !H N -<br />
!H N NOE's. Stretches of residues giving rise<br />
to these types of NOE's also had other<br />
characteristics associated with p-sheet<br />
residues: low field 1 H a and !H N chemical<br />
shifts, strong ^N- 1 !!" coupling (manifested<br />
by intense DQF-COSY and TOCSY correlations),<br />
and reduced } H N exchange rates. In<br />
addition, 22 intense iHa-iH« NOE's, characteristic<br />
of antiparallel p-sheet [24], were
204 Bulletin of Magnetic Resonance<br />
-iY^k XII 146-152<br />
119-122<br />
X 120-125<br />
III30_34<br />
II23-27<br />
VI.5-73<br />
VH76.84<br />
Figure 2. Schematic diagram of the topology strand NOE's. Dashed lines indicate interof<br />
the p-sheet framework of IRAP. Double- strand hydrogen bonds inferred from<br />
arrowhead lines identify assigned inter- analysis of *H N exchange rates.<br />
11-19
Vol. 14, No. 1-4 205<br />
1.20<br />
1.00<br />
0.80<br />
0.60<br />
0.40<br />
A5 0.20<br />
0.00<br />
-0.20<br />
-0.40<br />
-0.60<br />
-0.80 -L i i....?...J I !.........! n i [in! i.J<br />
-0.50 --<br />
-1.00 -•;•<br />
-1.50 --<br />
-2.00 --<br />
10 20 30<br />
Alpha proton<br />
iv I | v | [ vi i | vii<br />
40 50 60 70 80 90<br />
Residue<br />
Alpha carbon<br />
: IXj<br />
100 110 120 130 140 150<br />
-2.50 --<br />
rrni_n mm —yyj |_yj L^L<br />
-3. 1<br />
DQ DL J US — IEJ CO OD<br />
10 20 30 40 50 60 70 80 90 140 150<br />
Residue<br />
Figure 3. Comparison of 'H" (top) and 13 C a shown at the bottom of each plot, p-strands<br />
(bottom) A5 chemical shifts for IRAP and IL- are boxed, while helical regions are denoted<br />
ip. Locations of NOE-defined secondary by lines,<br />
structure elements in both proteins are
206 Bulletin of Magnetic Resonance<br />
identified in the two-dimensional NOESY<br />
spectrum recorded in 2 H2O and confirmed in<br />
the 1H-13C NOESY-HMQC spectrum. Analysis<br />
of this pattern of NOE's results in alignment<br />
of the 12 p-sheet strands as shown in Figure<br />
2. Arrows indicate observed cross-strand<br />
NOE's, while dashed lines indicate hydrogen<br />
bonds inferred from !H N exchange rates.<br />
The p-sheet strands have been<br />
presented in Figure 2 in a manner that<br />
allows easy comparison to the P-sheet<br />
framework elucidated for IL-lp in solution<br />
by Driscoll et al. ([9], Figure 5). Comparison<br />
of the two figures illustrates how the<br />
overall topology of the two proteins is<br />
identical, but in several regions is composed<br />
of different stretches of the primary<br />
sequence. Strands II and III, which are<br />
adjacent strands connected by a fiveresidue<br />
turn in IL-lp, are shifted by six<br />
residues in the primary sequence and are<br />
connected by a four-residue turn in IRAP.<br />
Similarly, strands I and IV in IRAP are<br />
shifted by six arid five residues, respectively.<br />
The consequence of shifting the<br />
residues that comprise these portions of the<br />
P-sheet is that the N-terminal six residues of<br />
IRAP have no structural counterpart in the<br />
IL-ip structure. Structurally significant<br />
shifts of one residue are seen for strands VI,<br />
VII, and XII.<br />
As expected for predominantly Psheet<br />
proteins, large positive (downfield)<br />
deviations from random coil chemical shifts<br />
are observed for the l H a resonances [25],<br />
and large negative (upfield) deviations<br />
from random coil chemical shifts are<br />
observed for the !3C a resonances [25,26].<br />
Comparison of the secondary 'H a and 13 C a<br />
chemical shifts of IRAP and IL-ip also<br />
illustrates the differences and similarities<br />
in location of secondary structure elements,<br />
as shown in Figure 3. Note the excellent<br />
agreement of the out-of-phase appearance<br />
of the plots over the first 50 residues with<br />
the five- or six-residue offset in location of<br />
the P-strands. Also note how the in-phase<br />
sections of the plots correspond to P-strands<br />
at identical positions in both proteins.<br />
While the solution secondary<br />
structure of IRAP is dominated by antiparallel<br />
p-sheet, short stretches of strong !H N -<br />
!H N NOE's between adjacent residues,<br />
indicative of a helical conformation, were<br />
also observed. As shown in Figure 2, these<br />
regions involve residues 40-45 and 86-89.<br />
In addition, NOE's from 1HN of L89 to 1H« of<br />
186 and from IH* of E44 to 1H« of V41, both<br />
medium-range NOE's characteristic of a<br />
helical conformation [24], were observed.<br />
These were the only NOE's of this type<br />
unambiguously assigned. In addition, the<br />
i 3 C a chemical shifts in these two stretches<br />
are shifted downfield slightly compared to<br />
their random coil values, as would be<br />
expected for a helical conformation [26].<br />
These residues correspond to residues 35-40<br />
and 87-90 in IL-ip, the former of which is a<br />
3 io helix in solution [9]. Isolated strong !H N -<br />
!H N NOE's between two or three sequential<br />
residues were also observed, and locate turn<br />
conformations at positions: 18-22, 27-30, 62-<br />
65, and 116-118.<br />
5 Acknowledgements<br />
We thank Dr. Paul E. Fagerness, Dr. Eldon L.<br />
Ulrich, Dr. Melinda Roy, Kathleen A. Farley,<br />
Ron L. VanZanten, and Cindy A. Granatir for<br />
their continuing contributions. We thank<br />
Dr. Ad Bax (NIH) for advice and for supplying<br />
preprints of the constant-time HNCA<br />
and HN(CO)CA pulse sequences.<br />
6 References<br />
1. Dinarello, C. A. (1989) Adv. Immunology<br />
44, 153-205.<br />
2. Hannum, C. H., et al. (1990) Nature 343,<br />
336-340.<br />
3. Eisenberg, S. P., et al. (1990) Nature 343,<br />
341-346.<br />
4. Carter, D. B., et al. (1990) Nature 344, 633-<br />
638.<br />
5. MacDonald, H. R., et al. (1986) FEBS Lett.<br />
209, 295-298.<br />
6. Wingfield, P., et al. (1989) Eur. J. Biochem.<br />
179, 565-571.<br />
7. Gehrke, L., et al. (1990) J. Biol. Chem. 265,<br />
5922-5925.<br />
8. Driscoll, P.C., et al. (1990a) Biochemistry<br />
29, 3542-3556.<br />
9. Driscoll, P. C, et al. (1990b) Biochemistry<br />
29, 4668-4682.<br />
10. Clore, G. M., Wingfield, P. T., &<br />
Gronenborn, A. M. (1991) Biochemistry 30,
Vol. 14, No. 1-4 207<br />
2315-2323.<br />
11 Clore, G. M, &. Gronenborn, A. M. (1991)<br />
/. Mol. Biol. 221, 47-53.<br />
12. Tate, S., et al. (1992) Biochemistry 31,<br />
2435-2442.<br />
13. Finzel, B. C, et al. (1989) J. Mol. Biol. 209,<br />
779-791.<br />
14. Priestle, J. P., Schar, H. P., & Griitter, M.<br />
G. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 9667-<br />
9671.<br />
15. Zuiderweg, E. R. P., & Fesik, S. W. (1989)<br />
Biochemistry 28, 2387-2391.<br />
16. Marion, D., et al. (1989) Biochemistry 28,<br />
6150-6156.<br />
17. Grzesiek, S., & Bax, A. (1992)7. Magn.<br />
Reson. 96, 432-440.<br />
18. Stockman, B. J., et al. (1992)<br />
Biochemistry 31, 5237-5245.<br />
19. Ikura, M., Kay, L. E., & Bax, A. (1991) J.<br />
Biomol. NMR 1, 299-304.<br />
20. Clore, G. M, et al. (1990) Biochemistry<br />
29, 8172-8184.<br />
21. Ikura, M., et al. (1990) J. Magn. Reson.<br />
86, 204-209.<br />
22. Ikura, M., Kay, L. E., & Bax, A. (1990)<br />
Biochemistry 29, 4659-4667.<br />
23. Bax, A., & Ikura, M. (1991) J. Biomol.<br />
NMR 1, 99-104.<br />
24. Wiithrich, K. (1986) NMR of Proteins and<br />
Nucleic Acids, Wiley, New York.<br />
25. Wishart, D. S., Sykes, B. D., & Richards, F.<br />
M. (1991) /. Mol. Biol. 222, 311-333.<br />
26. Spera, S., & Bax, A. (1991) J. Am. Chem.<br />
Soc. 113, 5490-5492.
208<br />
1 Introduction<br />
Green's Function Calculation of<br />
Effective Nuclear Relaxation Times in<br />
Metals and Disordered Metals<br />
M. Martin-Landrovc and J A. Moreno<br />
Departamento de Fisica and Centro de Resonancia Magnetica,<br />
Facultad de Cicncias, Universidad Central de Venezuela,<br />
Apartado 47586, Caracas 1041-A, Venezuela.<br />
The theoretical derivation of expressions for die<br />
magnetic nuclear relaxation times at very low<br />
temperatures and the description of the behaviour<br />
with temperature for such relaxation times,<br />
has been of major interest among the researchers<br />
in the field, specially because of the recent experimental<br />
possibility to obtain measurements of<br />
nuclear magnetic properties at such low temperatures.<br />
There has been a considerable amount of<br />
work in the area of nuclear magnetism [1], but a<br />
comprehensive theoretical interpretation of<br />
NMR relaxation times, at arbitrary temperatures,<br />
is still lacking. Recently, Sbibata et al. has published<br />
a series of papers [2], [3] concerning the<br />
theoretical determination of the nuclear spin lattice<br />
relaxation time for a system of nuclear spins<br />
interacting with conduction electrons ina a metal.<br />
Using a theory of nonlinear spin relaxation [4], [5]<br />
they predicted a multicxponential spin-lattice relaxation<br />
behaviour.<br />
In the case of disordered metals and high temperatures,<br />
where a Korringa law is aplicable, Warren<br />
[6] predicted an enhancement of the relaxation<br />
rate, which in some cases [7] could be as large<br />
as 6,500. More recently [8], Gdtze and Ketterle<br />
derived expressions for die Warren enhancement<br />
factor by means of normalized Kubo response<br />
functions [9].<br />
ED the present work, we make use of the two-times<br />
Green's function formalism in the regime of<br />
the Linear Response Theory to derive the temperature<br />
behaviour of nuclear relaxation times [10]<br />
for nuclei in metals and disordered metals. The<br />
Bulletin of Magnetic Resonance<br />
results obtained are in complete agreement with<br />
those derived by Sbibata in the assumption of an<br />
effective unique relaxation time [2], [3] and with<br />
experimental evidence [11], [12]. Also, there is<br />
agreement between our results and those derived<br />
by Gdtze and Ketterle [8] in the high temperature<br />
regime, where Korringa law is valid but additionally<br />
we obtained expressions for the enhanced<br />
relaxation rate which are valid in the whole temperature<br />
range. The organization of this paper is<br />
as follows, in section 2, the general formalism is<br />
derived, in section 3, we work out the Hamiltonian<br />
of die system from which the equation for the<br />
Green's function < < I°/I°> > to, which contains<br />
all the information relevant to the spin-lattice<br />
relaxation, is derived. This equation is then<br />
solved including terms up to second order in the<br />
electron nucleus interaction and the disorder parameters.<br />
Finally in section 4 we discuss the relaxation<br />
times formulas.<br />
2 Green's Functions and the Relaxation<br />
Rate<br />
We will consider a system which can be modelled<br />
by the total Hamiltonian:<br />
H = Hs + HSL + HL (1)<br />
where Hs represents the nuclear spin Hamiltonian,<br />
HL is the Hamiltonian for the heat bath and<br />
HSL represents the couplig between both<br />
systems, usually under the condition<br />
HL > Hs > HSL which occurs commonly in NMR<br />
experiments. In order to consider the evolution of
Vol. 14, No. 1-4 209<br />
the system toward thermal equilibrium, it is necessary<br />
to prepare the system in an initial nonequilibrium<br />
state. This can be achieved by the<br />
adiabatic switching on of a pertrubation, which in<br />
our case is a magnetic field along a particular<br />
direction, and suddenly at time t—0, this perturbation<br />
is switched off letting the system to evolve<br />
freely according to die Hamiltonian H. The perturbation<br />
can be written as:<br />
where:<br />
Hi' = - M .<br />
H\* = Q(-t)e Et Hi<br />
(2)<br />
(3)<br />
Within the linear response theory, the magnetization<br />
for t>0 is given by the following expression<br />
[13]:<br />
H\ r dt<br />
which is written in terms of the two times retarded<br />
Green's function. By using the causality property<br />
[13] of the retarded Green's function, it is<br />
posible to write the equation for the displacement<br />
of magnetization from the equilibrium situation<br />
as:<br />
1 [+OO > lT ) H\ _ lm ,<br />
where < 8Af(/> = - o.Itcan<br />
be shown that:<br />
(4)<br />
. H\ (6)<br />
so that taking the Laplace transform of eq. (5), it<br />
can be written as:<br />
/(z) . (7)<br />
(5)<br />
where:<br />
with GM ( ><br />
- iz)<br />
We are interested in the asymptotic behaviour<br />
of < 8 M(f > since in that regime is that the relaxation<br />
rates are experimentally measured. According<br />
to Tauber's theorem [14], [15] an asymptotic<br />
espanskm of f(t) can be written as:<br />
*« 0<br />
where Z\ represents the poles of the function<br />
f(z), n* ( v) the order of the pole and Ck (y) the<br />
coefficient in a Laurent expansion of f(z) around<br />
the pole. In the particular case of f(z) having only<br />
first order poles, eq. (9) can be written:<br />
(8)<br />
(10)<br />
Let us suppose that the Green's function can be<br />
written in the general form:<br />
)<br />
o> - acoo -<br />
(11)<br />
where Qxand Waare complexfunctions. According<br />
to eq. (8) the poles of the function f(z) are<br />
related to the poles of the Oieen's function so that<br />
only those poles that are located at the upper<br />
complex plane must be considered [16]. In general<br />
it is necessary to make a complete analysis of<br />
the pole structure of the Green's function at the<br />
upper complex plane in order to describe the total<br />
aymptotic behaviour of the function f(t). As a simplifying<br />
assumption, we will consider the following<br />
zeroth-order approximation to the pole<br />
structure. We will first assume that the function<br />
Ox(ci>) does not contribute with any pole. This<br />
(V)
210 Bulletin of Magnetic Resonance<br />
analyzed with some care. Secondly that the function<br />
Wo. can be approximated by its value at die<br />
frequency aoo. This second assumption ussually<br />
is a very strong one since the whole pole structure<br />
is sometimes collapsed to a single first order pole,<br />
but nevertheless in many cases, this approximation<br />
gives the right relaxation behaviour correlating<br />
quite well with the experimental results. The<br />
relaxation rates obtained under this assumptions<br />
can be written in terms of the imaginary part of<br />
the function Wa evaluated at the frequency aoao,<br />
that is:<br />
and:<br />
T2<br />
j - ImWo(O) (12)<br />
too) - (- too)<br />
(13)<br />
3 Hamiltonian and Equation for the<br />
Green's Function Go (coX<br />
For tile system considered in this work, the Hamiltonian<br />
can be written as in equation (1) where<br />
the different terms are:<br />
vv'tt<br />
X Vz «*v*+ ~ «v'f-<br />
with:<br />
r<br />
(14)<br />
(15)<br />
The lattice Hamiltonian can be written as the<br />
sum of two terms [8]:<br />
HL = 2- £v (ks) at k s av k s<br />
\ks<br />
where:<br />
p + (q)V(q) (17)<br />
X at- §.*«* + (18)<br />
and represents the electron density fluctuations<br />
for wavevector q. The disorder in the lattice is<br />
represented by the Fourier transform U(q) of the<br />
random potential. This coefficient satisfies the<br />
symmetry property:<br />
U*(q)= V(-q) (19)<br />
The equation for the Green function Go(
Vol. 14, No. 1-4 211<br />
and:<br />
with:<br />
W(to) = § **'<br />
Att - A** A** -<br />
At* -<br />
Gf-+ (0 JU<br />
Ajt* -<br />
oi<br />
(22)<br />
(23)<br />
X {»(* *-)(!- *(*+)) + «(lt+)(i- «(*:*-))}<br />
At* -<br />
2<br />
- OJO)+<br />
at-- at+ ,<br />
i<br />
(24)<br />
o<br />
(25)<br />
(26)<br />
(27)<br />
(28)<br />
21 1 + I<br />
\k-g At •+,.*]<br />
(29)<br />
As it was discussed previously [17], the longitudinal<br />
relaxation rate is proportional to the imaginary<br />
part of the function W ( 0<br />
ImAt*(w)= * q> 0<br />
V(q)\ Z ,<br />
&t',k-<br />
(30)<br />
(31)<br />
(32)<br />
Equation (31) represents the dynamical shift in<br />
electronic energy due to the disorder and equation<br />
(32) corresponds to the electronic relaxation<br />
rate function, which in the limit of a>-»O,gives the<br />
electronic relaxation rate 1/J> . The imaginary<br />
part of W (to ) now become s:<br />
1ml<br />
x r<br />
kk<br />
XI-<br />
(At*-ReAt*) 2 +(/mAt*) 2<br />
(A**- RcAt*) 2 + (<br />
I<br />
(33)<br />
In the case of a perfect metal, that is for U(q) -<br />
0, we get from equation (33), the following expression<br />
for the relaxation rate [17]:
212<br />
k* '<br />
[n(*+)(!-<br />
+ n(*'-)(!-«(*+))]5( At*•) (34)<br />
which exhibits a Korringa behaviour at high<br />
temperatures and attains a maximum for the relaxation<br />
time at very low temperatures, comparable<br />
with nuclear spin energies [17]. in the case of<br />
disorder we get for the relaxation rate:<br />
njk - XI- n(k+<br />
(At*- ReAt*) 2 + (ImA**) 2<br />
I (A*i- ReAi"*) 2 -<br />
in the limit goes to 0.<br />
4 Conclusions<br />
(35)<br />
In die whole temperature range, equation (34)<br />
can be calculated numerically, and the result for<br />
the case of a perfect metal was obtained in reference<br />
17, where it can be appreciated that T i<br />
shows a maximum at a temperature that is approximately<br />
half of the nuclear spin temperature,<br />
which is consistent with the result obtained by<br />
Shibata and co-workers in the supposition that<br />
the relaxation process can be described by a unique<br />
effective relaxation time T l as the observed<br />
experimental behaviour is. Also there is a correspondence<br />
between both results in the limit of low<br />
temperature, leading us to think that our calculation,<br />
even in its simpler approximation, is quantitatively<br />
correct From the experimental point of<br />
view there are not enough data to decide whether<br />
a single or a multiexponential relaxation takes<br />
place, but the general tendency is to believe that<br />
even though the process seems actually to be multiexponential,<br />
it could be described by an effective<br />
relaxation time T 'i, which is die time that characterizes<br />
the evolution of observables, in particu-<br />
Bulletin of Magnetic Resonance<br />
lar, the longitudinal magnetization [11], [12]. The<br />
temperature dependence shown experimentally<br />
by this time T 'i, agrees completely with the behaviour<br />
calculated in reference 17. Also, equation<br />
(35) takes into account the enhancement of the<br />
relaxation rate for the whole temperature range,<br />
showing at high temperatures a departure from<br />
Korringa's law proportional to the electronic relaxation<br />
time as Warren proposed. The result obtained<br />
for the enhanced relaxation rate shows<br />
that this enhancement will be present even at low<br />
temperatures, as equation (35) is valid in the whole<br />
temperature range. The approximation assumed<br />
in this work, besides its simplicity, takes into<br />
account the main features present in die temperature<br />
behaviour of relaxation times, witiiin the<br />
limits of Linear Response Theory, and it can be<br />
extended to consider more realistic models or<br />
systems.<br />
Acknowledgments<br />
This work was partially supported by CONICTT<br />
and die Consejo de Desarrollo Cientffico y Humanistico<br />
of die Universidad Central de Venezuela.<br />
We also dutnk die collaboration of Fundacion<br />
Polar for die necessary funding to present<br />
diis work.<br />
References<br />
[1] Abragam A. and Goldman M., in Nuclear<br />
Magnetism: Order and Disorder (Oxford, Clarendon,<br />
1982).<br />
[2] Shibata, F. and Hamano, Y., Solid St. Cotnmun.,<br />
44,921 (1982).<br />
[3] Shibata, F. and Hamano, Y., J. Phys. Soc.<br />
Japan, 52,1410(1983).<br />
[4] Shibata, F., J. Phys. Soc. Japan, 49,15 (1980).<br />
[5] Shibata, F. and Asou, M., J. Phys. Soc. Japan,<br />
49,1234(1980).<br />
[6] Warren, W.W., Phys. Rev. B, 3,3708 (1971).<br />
[7] Warren, W.W. and Brennerrt, G.F., Amorphous<br />
and Liquid Semiconductors, (London, Taylor<br />
and Francis, 1974), 2, p. 1074.<br />
[8] Gdtze, W. and Ketterle, W., Z. Phys. B, 54,49<br />
(1983).<br />
[9] Kubo, R., J. Phys. Soc. Japan, 12,570 (1957).
Vol. 14, No. 1-4 213<br />
[10] Mardn Landrove, M. and Moreno, J.A., Acta.<br />
Oent Venez., 37,387 (1986).<br />
[11] Bacon, F., Barclay, A., Brewer, W.D., Shirley,<br />
D.A. and Templeton, JJE., Phys. Rev. B, 5,<br />
2397 (1972).<br />
[12] Brewer, WJD., Shirley, D.A. and Tcmplcton,<br />
J.E., Phys. Lett. A, 27,81 (1968).<br />
[13] Zubarev, D.N., Nonequilibrium Statistical<br />
Thermodynamics (Consultants Bureau, New<br />
York, 1974).<br />
[14] Berg, h., Introduction to the Operational<br />
Calculus (John WHey, New York, 1967).<br />
[15] Krasnov, M.L., Kiselev, A.I. and Makarenko,<br />
G.I., Functions of Complex Variable, Operational<br />
Calculus and Stability Theory (Mir Publishers,<br />
Moskow, 1983).<br />
[16] Paley, R.E.A.C. and Wiener, W., Fourier<br />
Transforms in the Complex Domain (A.M.S., New<br />
York, 1934).<br />
[17] Martin Landrove, M. and Moreno, J.A.,<br />
Phil. Mag., 58,103 (1988).
214<br />
1 Introduction<br />
The extraction of molecular<br />
potentials and rates by modeling the<br />
dependence of spectra on thermodynamic<br />
state is one of the major contributions of<br />
magnetic resonance to molecular physics.<br />
We argue here that this entire endeavor is<br />
conceptually suspect due to the implicit<br />
factorization of spin and spatial degrees of<br />
freedom in calculating stochastic averages.<br />
All such averages have heretofore neglected<br />
spin—dependent energies in the potential for<br />
nuclear motion and are therefore not results<br />
of equilibrium statistical mechanics. We<br />
develop an averaging procedure from<br />
equilibrium statistical mechanics and find<br />
that it predicts large, previously<br />
unrecognized contributions to motionally<br />
averaged spin Hamiltonians which are<br />
directly proportional to spatial terms in the<br />
energy. We discuss the present state of the<br />
experimental evidence for the accepted<br />
average, with particular emphasis on<br />
indirect scalar couplings averaged by<br />
conformer equilibria, and find that the<br />
quality of presently available data cannot<br />
resolve the issue conclusively. The weakness<br />
of the theoretical basis of the accepted<br />
theory is also outlined.<br />
2 The Traditional Stochastic Average<br />
Stochastic Averaging Revisited<br />
D.H. Jones, N.D. Kurur, and D.P. Weitekamp<br />
Arthur Amos Noyes Laboratory of Chemical Physics<br />
California Institute of Technology<br />
Pasadena, CA 91125, USA<br />
For over forty years it has been<br />
accepted 1 " 36 that spin states are transported<br />
between spatial states with spinindependent<br />
rates. This unexamined<br />
assumption was clearly stated in the seminal<br />
work of Bloembergen, Purcell and Pound: 1<br />
"The atom or molecule is simply a vehicle<br />
Contribution No. 8724<br />
Bulletin of Magnetic Resonance<br />
by which the nucleus is conveyed from point<br />
to point. We thus neglect the reaction of<br />
magnetic moments of the nuclei upon the<br />
motion." This notion is the" basis for all<br />
existing formalisms 1 " 36 for calculating the<br />
magnetic resonance lineshapes of spin<br />
systems undergoing spatial rate processes —<br />
most importantly, chemical exchange. A<br />
well—known consequence is that the average<br />
value of a spin—Hamiltonian parameter<br />
(chemical shift, scalar coupling, dipolar<br />
coupling, etc.) in the fast—exchange limit is<br />
given by<br />
= SpnX:<br />
n n, (1)<br />
where Xn is the value of this parameter in<br />
the spin Hamiltonian of the nth spatial<br />
manifold and pn is viewed as the probability<br />
of the system being in that manifold,<br />
irrespective of spin state. The sum may be<br />
over molecular eigenstates or over' large<br />
groups of them, as when n indexes molecular<br />
conformers. The traditional prescription,<br />
which neglects spin energies, is to express<br />
the molecular partition function q as a sum<br />
of parts qn associated with each indexed<br />
manifold. The probability for each manifold<br />
is pn = qn/ and the ratio of two such<br />
probabilities is (in the absence of work<br />
terms)<br />
Pn/pn' = exp[-AAnn' /RT] (2a)<br />
= exp[(-AUnn'+TASnn')/RT], (2b)<br />
where AAnn'= -RTln(qn/qn') is the<br />
difference in the molar Helmholtz free<br />
energies of the manifolds. In Eq. 2b, the free<br />
energy differences have been divided into
Vol. 14, No. 1-4 215<br />
differences in energy AUnn' and entropy<br />
ASnn'- The connection to molecular<br />
energies is AUnn' = N^(En-En'), where En<br />
is the common spatial contribution to free<br />
energy of the n th manifold of spin states and<br />
N 4 is Avogadro's number.<br />
3 An Alternative Stochastic Average<br />
The actual molecular energies may be<br />
written as E? = En+E (n), where 7 indexes<br />
a spin eigenstate within the n th manifold.<br />
Since spin Hamiltonians are constructed as<br />
traceless, the spin—dependent contributions<br />
E (n) sum to zero in each manifold.<br />
Nevertheless, the absence of the spin energy<br />
terms in the pn indicates unambiguously<br />
that Eq. 1 is not derivable without<br />
approximation from equilibrium statistical<br />
mechanics.<br />
We address the following questions.<br />
What is the exact equilibrium expression for<br />
the stochastically averaged parameters? For<br />
which systems will it differ measurably from<br />
Eq. 1? Do existing experiments decide the<br />
issue conclusively?<br />
To proceed, we specialize to the case<br />
that the spin Hamiltonians in all<br />
significantly occupied spatial manifolds are<br />
mutually commuting. Then the spin<br />
eigenbasis {I7)} is independent of spatial<br />
state and is the basis needed to describe the<br />
stochastically averaged spectrum. The<br />
spatially-averaged energy of a particular<br />
such spin eigenstate 17) is<br />
E =SE2exp(-A2/RT)/Sexp(-A2/RT) (3a)<br />
I n n<br />
= SpM (3b)<br />
n<br />
Note that Eq. 3 uses the complete<br />
manifold free energy, A 2 = NAEZ-TSn,<br />
including spin terms, according to the<br />
prescription of equilibrium statistical<br />
mechanics. Each such summation is<br />
restricted to a particular spin state and the<br />
distribution among spatial states for each<br />
spin state is assumed to be the equilibrium<br />
distribution at the lattice temperature.<br />
Thus p2 is the conditional probability of<br />
being in the n th spatial manifold, given that<br />
the spin state is [7). As in the traditional<br />
formulation, no specification of the<br />
distribution of population among spin states<br />
is needed. Our hypothesis is that the<br />
spectral line positions for sufficiently fast<br />
exchange between spatial manifolds are the<br />
Bohr frequencies, ^^(E^-E^/h<br />
corresponding to differences between the<br />
average energies of Eq. 3. The<br />
corresponding motionally averaged<br />
spin—Hamiltonian parameters are those<br />
which generate this spectrum.<br />
As a simple illustration, consider a<br />
spin Hamiltonian with only two distinct<br />
eigenvalues, ±Xn/2 (in Hz), for each n.<br />
Then the proposed alternative to Eq. (1) is<br />
(with 7 = ±)<br />
(X) =<br />
(<br />
= S [(p£ -<br />
n<br />
(4a)<br />
(4b)<br />
(4c)<br />
For the usual high—temperature case where<br />
hXn
i "i<br />
t! I<br />
m<br />
216<br />
depends on spin, spatial energy differences<br />
between manifolds contribute to the<br />
statistically averaged frequencies for spin<br />
transitions. The product of a small<br />
spin—dependent probability difference<br />
multiplied by a large spatial energy<br />
difference gives a readily measured<br />
contribution to the magnetic resonance line<br />
position.<br />
4 Test Systems<br />
In both formulations it is possible<br />
ideally to predict the fast exchange<br />
observations without adjustable parameters.<br />
This would constitute a purely experimental<br />
test, the quality of which depends only on<br />
the experimental uncertainties. In practice<br />
there seems to be no case where NMR<br />
spectra of individual molecular eigenstates<br />
have been obtained separately and also as a<br />
thermal average.<br />
Thus, it seems necessary to look to<br />
those situations where n indexes conformers.<br />
The experimental concept is simple and<br />
well-known. At low temperature, the spin<br />
Hamiltonian for each conformer can be<br />
determined, since, in the limit of negligible<br />
chemical exchange between them, separate<br />
spectra are seen for each. The relative areas<br />
of these spectra at each slow—exchange<br />
temperature provide the relative populations<br />
and thus the free—energy difference between<br />
conformers. A linear fit to the temperature<br />
dependence of this free—energy difference<br />
allows it to be decomposed into two terms,<br />
which can be viewed as an energy difference<br />
and an entropy difference if one additionally<br />
assumes that the temperature dependence of<br />
these is negligible over the experimental<br />
range.<br />
Since a conformer is a set of<br />
molecular eigenstates, additional dependence<br />
on thermodynamic state (eg. temperature<br />
dependence) is possible due to averaging<br />
within this set. Theoretically, one has<br />
precisely the same problem in deciding how<br />
to take this average as for the averaging over<br />
conformers. However, if one can measure any<br />
such temperature dependence within the<br />
slow—exchange regime and extrapolate to<br />
fast—exchange, then the stochastic theory<br />
need not enter at this level.<br />
Thus, we arrive at a set of criteria<br />
which need to be met for a compelling, fully<br />
experimental test of any theory relating<br />
Bulletin of Magnetic Resonance<br />
slow-exchange and fast—exchange spectra:<br />
i) The system must have<br />
state—dependent rates such that<br />
measurements in both the slow— and<br />
fast-exchange regimes are possible. For<br />
fluids this typically requires barriers between<br />
conformers on the order of 10 kcal/mole.<br />
ii) The state dependence of the<br />
conformer spin Hamiltonians and the<br />
free-energy differences must be measured in<br />
the slow—exchange region to allow<br />
extrapolation through the fast—exchange<br />
region. This is often the major source of<br />
uncertainty because of the small<br />
temperature range corresponding to<br />
slow-exchange. The difference in<br />
thermodynamic state between these regimes<br />
ideally is small or even zero, so as to<br />
minimize the propagation of errors due to<br />
phenomenological extrapolation. Using<br />
different NMR transitions or field conditions<br />
to measure the same spin Hamiltonian<br />
parameter can help in this regard; since the<br />
criterion for motional collapse varies with<br />
the transition observed, there is no minimum<br />
8.<br />
8 1 4.<br />
o<br />
•a o<br />
GO<br />
6. -<br />
1 1<br />
- 'V^<br />
1 __l<br />
—<br />
0.<br />
f1 1 1 1<br />
0. 200. 400. 600. 800. 1000.<br />
Temperature (K)<br />
Figure 1. Traditional (dashed) and<br />
alternative (solid) stochastic averages for a<br />
simple two—conformer, two—spin system.<br />
Simulations have the general features of<br />
some substituted ethanes: a gauche coupling<br />
of 2.0 Hz, a trans coupling of 20.0 Hz, the<br />
trans conformer with a free energy 200<br />
cal/mol greater than the doubly—degenerate<br />
gauche conformer.
Vol. 14, No. 1-4 217<br />
difference in thermodynamic state between<br />
fast and slow exchange.<br />
iii) Some or all of the fast-exchange<br />
data should fall outside the error bars on the<br />
predictions of one of the theories, thereby<br />
disproving it. The accepted and alternative<br />
theories embodied in Eqs. 1 and 3,<br />
respectively, have identical predictions for<br />
mutual exchange and whenever the occupied<br />
conformers are degenerate in spatial energy<br />
or in spin Hamiltonians. For two—site<br />
problems, the theories will typically differ<br />
measurably when the conformer free energies<br />
differ by > 102 cal/mole. When this<br />
difference exceeds 10 3 cal/mole, sensitivity<br />
will usually preclude observing the<br />
slow-exchange spectrum of the minor<br />
conformer. Figure 1 is a numerical<br />
comparison of the two theories for the simple<br />
case of two conformers and a two—spin<br />
system. It demonstrates that the predictions<br />
of the theories are different by a magnitude<br />
that should be measurable. There is also a<br />
qualitative difference: the sign of the<br />
temperature dependence of the<br />
fast—exchange spin Hamiltonian parameter<br />
is opposite to that of the accepted theory<br />
over part of the temperature range. This<br />
behavior is inconsistent with Eq. 1 regardless<br />
of how the pn are calculated.<br />
The above conditions are not<br />
extremely restrictive; a substantial fraction<br />
of the molecules whose conformer equilibria<br />
have been studied by solution-state NMR<br />
fall into this range of free—energy differences.<br />
Since the accepted theory has been in<br />
increasing use for four decades, it might be<br />
expected that it would have substantial and<br />
diverse experimental support. While it is<br />
difficult to have confidence in the<br />
completeness of a search through such a<br />
large literature, we are as yet unaware of<br />
any data set that meets the criteria above.<br />
Thus, no theory has presently been<br />
evaluated by this seemingly reasonable<br />
standard.<br />
The only theories ever considered<br />
previously are of the form of Eq. 1.<br />
Numerous authors have noted failures in its<br />
application, 19 " 21 but these have usually been<br />
plausibly attributed to inadequacies in the<br />
data, most commonly uncertainties of<br />
conformer assignment or unmeasured<br />
temperature dependence of a conformer spin<br />
Hamiltonian.<br />
5 Substituted Ethanes<br />
Substituted ethanes in solution 14 " 24<br />
are the most studied systems and include<br />
cases which nearly meet the criteria of the<br />
previous section. Figure 2 illustrates, with<br />
the example of 1-fluoro-l,1,2,2—<br />
tetrachloroethane, the trans and gauche<br />
rotational isomers which interconvert at<br />
convenient rates. Since the two gauche<br />
conformers are mirror—images, there are<br />
only two magnetically distinct conformers,<br />
one with the H and F atoms trans to one<br />
another and the other a degenerate pair of<br />
gauche rotamers at the other two staggered<br />
positions of the dihedral angle of rotation<br />
about the carbon—carbon single bond.<br />
Increasing temperature carries the system<br />
from the slow—exchange limit to the<br />
fast—exchange limit without a change in<br />
composition or phase. The fast—exchange,<br />
three—bond vicinal coupling (Jrrp) is known<br />
to be temperature dependent and this has<br />
been attributed to the averaging, according<br />
to Eqs. 1 and 2, between distinct values Jt<br />
and Jg in the trans and gauche conformers,<br />
respectively. 15 " 16 Unlike chemical shifts,<br />
scalar couplings do not require a nominally<br />
temperature—independent reference<br />
resonance to compensate for the usual<br />
uncontrolled shifts of internal field with<br />
temperature. Also, scalar couplings are<br />
generally believed to be less sensitive to<br />
intermolecular interactions which could<br />
provide a confounding mechanism of<br />
temperature dependence.<br />
For the particular case of<br />
1—fluoro—1,1,2,2—tetrachloroethane, a test as<br />
E(
218 Bulletin of Magnetic Resonance<br />
described of the stochastic averaging theories<br />
is prevented by the failure to resolve the<br />
coupling Jg. 37 This quantity has an<br />
experimental upper bound of 2 Hz. If it is<br />
assumed that Jg is independent of<br />
temperature, then the data are consistent<br />
with the accepted theory. 16 However a<br />
temperature dependence of 0.005-O.01 Hz/K<br />
between 150 K, where slow-exchange<br />
observations have been made on the 19 F<br />
resonances, and 300 K, where the *H<br />
spectrum is motionally averaged, would<br />
allow our alternative stochastic average in<br />
the form<br />
(5)<br />
to fit the data as well or better.<br />
Reason to suspect such a temperature<br />
dependence can be found from a close<br />
reading of the literature on the related<br />
system 1—fluoro—1,1,2,2—tetrabromo—<br />
ethane. 22 A value of Jg = 2.4 ± 0.3 Hz in<br />
dimethylether at 188 K can be measured<br />
from published data, 39 but has been reported<br />
as 1.7 Hz at 180 K in the same solvent. 23 In<br />
CFCI3 this coupling has been tabulated as<br />
1.15 Hz from 171 to 178 K, 22 but our recent<br />
fits of the spectrum (at 171 K only) in<br />
reference 23 indicate Jg = 1.5 ± 0.3 Hz.<br />
Thus the reported absence of temperature<br />
dependence to three significant figures is<br />
dubious and further experimental work is<br />
needed.<br />
6 Discussion<br />
The present conformer model is by no<br />
means the most refined version of Eq. 3<br />
possible, but has the advantage of employing<br />
only quantities that are experimentally<br />
measured on the same sample. Continuous<br />
classical—mechanical forms of the present<br />
theory are readily written down and the<br />
differences from the accepted theory persist.<br />
Alternatively, the quantum partition<br />
functions could be modeled. Such<br />
modifications would introduce unmeasured<br />
parameters. Such extensions might be<br />
warranted after improvements in the<br />
experimental data base.<br />
It is of interest to note that the same<br />
theoretical prediction for (Jjrp) obtained by<br />
use of Eqs. 3 and 5 also results if the same<br />
conformer model is evaluated using Eq. 4<br />
with Xn = Jn and + and — indicating,<br />
respectively, triplet and singlet zero-field<br />
eigenstates. Thus, no measurable field<br />
dependence of (Jjjp) is predicted by the<br />
alternative theory, which is not obvious<br />
because of the entanglement of different spin<br />
Hamiltonian parameters in the spin energies.<br />
The theoretical justification for Eq. 1<br />
and related propositions is also weaker than<br />
has been appreciated. Such an average<br />
follows from dynamic models 1 " 36 based on<br />
the assumption that spin states in<br />
superposition are transported between<br />
different spatial states in perfect concert.<br />
Any such model exists in a truncated<br />
Liouville space that excludes superpositions<br />
of states that differ in both their spin and<br />
spatial factors. Whether such a truncated<br />
space suffices to describe magnetic resonance<br />
lineshapes is an open question. What is<br />
clear is that such a space cannot describe the<br />
approach to equilibrium of the total system,<br />
since this requires spin—dependent rates<br />
between spatial manifolds. Thus a full<br />
dynamic solution is needed in this complete<br />
Liouville space. One result of such a full<br />
solution will be the equilibrium average spin<br />
energies of Eq. 3. Less clear is under what<br />
dynamic assumptions either these energies or<br />
those that follow from Eq. 1 will describe the<br />
fast—exchange spectrum. In any case the<br />
problem is richer than has been appreciated.<br />
The accepted idea of how to calculate<br />
a stochastically averaged spectrum is<br />
universal in the literature of magnetic<br />
resonance, underlying the interpretation of<br />
average chemical shifts, dipolar couplings,<br />
tunnel splittings, quadrupole couplings, and<br />
hyper-fine interactions. In most situations<br />
the number of unknowns is such that it is<br />
not possible to verify the form of the<br />
stochastic average, but valuable information<br />
could be obtained if the correct form were<br />
known. If the traditional ideas are generally<br />
incorrect, many thousands of experiments<br />
need to be reinterpreted with forms of the<br />
present theory to in fact obtain the<br />
quantitative information on molecular<br />
structure that they were designed to yield.<br />
Ultimately, the choice of theory will be<br />
decided by a preponderance of data. The<br />
present work indicates clearly that the issue<br />
must be reopened and is a first step in the<br />
reexamination of the experimental basis of<br />
Eq. 1. Accurate experimental measurements
Vol. 14, No. 1-4<br />
of systems such as those discussed here will<br />
be needed in order to discriminate between<br />
the stochastic theories, as opposed to fitting<br />
free parameters according to one or the other<br />
theory. Precisely the same issue of how to<br />
calculate a stochastic average also arises in<br />
the (ab initio) theoretical calculation of a<br />
measurable spin Hamiltonian from<br />
expectation values of the underlying<br />
molecular eigenstates, which are almost<br />
never sufficiently long—lived to measure<br />
individually by magnetic resonance.<br />
Application of the correct statistical<br />
prescription will often be needed to in fact<br />
test experimentally whether the quantummechanical<br />
part of the calculation is<br />
adequate.<br />
This work was supported by the<br />
National Science Foundation<br />
(CHE-9005964). DHJ holds a Department of<br />
Education Graduate Fellowship. DPW is a<br />
Camille and Henry Dreyfus<br />
Teacher—Scholar.<br />
*N. Bloembergen, E.M. Purcell, and R.V.<br />
Pound, Phys. Rev. 73, 679 (1948).<br />
2<br />
E.R. Andrew and R. Bersohn, J. Chem.<br />
Phys. 18, 159 (1950).<br />
3<br />
E.L. Hahn and D.E. Maxwell, Phys. Rev.<br />
88, 1070 (1952).<br />
220<br />
1 Introduction<br />
Magnetic Resonance of Trapped Ions<br />
by Spin-Dependent<br />
Cyclotron Acceleration<br />
Pedro J. Pizarro and Daniel P. Weitekamp<br />
Arthur Amos Noyes Laboratory of Chemical Physics<br />
California Institute of Technology, 127-72<br />
Pasadena, CA 91125 USA<br />
The frequencies of motion of ions trapped by static<br />
magnetic and electric fields may be detected through the<br />
charge induced on the trap plates. In a homogeneous<br />
magnetic field, these frequencies have long been used for<br />
high resolution measurements of the charge-to-mass ratio.<br />
The most important chemical application of this is mass<br />
spectroscopy via Fourier transform ion cyclotron resonance<br />
(ICR). 1 ' 2 ' 3 Single-ion sensitivity in the detection of the<br />
axial trapping frequency has also been demonstrated<br />
recently for masses of chemical interest by D. Pritchard. 4 - 5<br />
We have recently proposed methods for transferring this<br />
exquisite sensitivity to the detection of the internal<br />
spectroscopy of ions, in particular the magnetic resonance<br />
spectra. 6 ' 7 This presentation focuses on design issues for<br />
the most promising such method, in which spin-dependent<br />
cyclotron acceleration is imposed and the resulting change<br />
in ion orbit is detected as a change in the axial trapping<br />
frequency.<br />
As in the electron g-factor measurement of H.<br />
Dehmelt, 8 the shift in axial trapping frequency is<br />
proportional to the strength of a static magnetic bottle field<br />
gradient. However, that direct effect of a spin flip is<br />
impractically small for nuclear spin flips of ions. In the<br />
present case, the transverse magnetic moment is coupled to<br />
a radiofrequency gradient to provide an accelerating force.<br />
A precedent is M. Bloom's deflection of neutral molecular<br />
beams by radiofrequency field gradients (the "transverse<br />
Stern-Gerlach effect"). 9 ' 10<br />
We have derived both semiclassically and quantummechanically<br />
the conditions under which a magnetic field<br />
gradient modulated at both the Larmor and cyclotron<br />
Contribution No. 8714<br />
Bulletin of Magnetic Resonance<br />
frequencies will lead to cyclotron acceleration proportional<br />
to the transverse magnetic moment of a coherent state of<br />
the particle and radiation field. In the presence of a<br />
magnetic bottle, the corresponding shift in the axial<br />
trapping frequency due to this spin-dependent work can be<br />
made much larger than the shift due directly to a spin flip.<br />
This effect has been incorporated into a proposed<br />
experimental procedure in which the spin-flip probability,<br />
resulting from a period of high-resolution magnetic<br />
resonance, controls the presence or absence of a net axial<br />
frequency shift between two detection periods. A data<br />
reduction algorithm based on the fast Fourier transform<br />
allows rapid conversion of the "before" and "after" signals<br />
from one or many trapped ions into a point of the magnetic<br />
resonance spectrum or interferogram. Simulated signals,<br />
including the anticipated noise from both the detection<br />
circuit and intrinsic quantum fluctuations in the number of<br />
spins flipped, indicate the method is practical.<br />
2 Ion Confinement in a Penning Trap and a<br />
Magnetic Bottle<br />
An ion in a homogeneous static magnetic field Bo,<br />
whose direction defines the z-axis, will circle in the<br />
transverse (x-y) plane at frequency ©c=qB0/m. Axial<br />
(z-axis) trapping can be obtained by adding a static electric<br />
field E = (vo/d 2 )(xx/2+yy/2-zz). Here Vo is a DC<br />
trapping potential and d is a characteristic linear<br />
dimension dependent on the details of the electrode<br />
geometry (e.g., hyperbolic, cubic, cylindrical). An ion so<br />
trapped undergoes three different types of harmonic<br />
translational motion: 11 axial oscillation at frequency
Vol. 14, No. 1-4 221<br />
a, = ^qV0/md 2 , rapid cyclotron motion in the<br />
transverse plane at frequency ©+, and slower magnetron<br />
motion of frequency co_, with ©± = y{ ©c + J© 2 ~2© 2 I.<br />
In ICR, one monitors the radiofrequency voltage at © +<br />
induced on a capacitor by the cyclotron orbit of a group of<br />
ions coherently excited by an oscillating electric field<br />
resonant with the cyclotron motion. Single ion sensitivity<br />
has been achieved for detection of both ©z 4 and, indirectly,<br />
©+, 5 using axial detection. The charge-to-mass ratio is the<br />
only structural quantity measured on trapped ions by ICR<br />
to date.<br />
In order to describe how other quantities, in particular<br />
the NMR spectrum, may be encoded into trapped ion<br />
signals, it is necessary to analyze the case of ion motion in<br />
the presence of a magnetic field gradient. In particular we<br />
consider the introduction of a magnetic bottle field of the<br />
form (in cylindrical coordinates, f = zz+pp+)<br />
AB = B2f(z 2 -p 2 /2)z-zpp]. n > 12 The presence of AB<br />
couples the axial, cyclotron and magnetron motions to the<br />
spin. A classical analysis of the ion motion in a known<br />
spin state is adequate though the principal results can be<br />
confirmed with the trapping frequencies from first-order<br />
quantum perturbation theory. 11 To a very good<br />
approximation the axial amplitude is a simple onedimensional<br />
sinusoid even in the presence of the magnetic<br />
bottle. The potential energy for the axial motion is<br />
The spin magnetic moment operator has been replaced by<br />
its two high-field eigenvalues (upper and lower signs)<br />
assuming a single spin 1/2 nucleus with gyromagnetic ratio<br />
y. The only coupling to the transverse motion is through<br />
the mechanical magnetic moment \xm = (q/2)[x(dy/dt)y(dx/dt)].<br />
This quantity is however a constant of the<br />
motion, 13 so that the axial motion is separable with a<br />
parametric dependence on the ion's transverse orbit. The<br />
ellipsis denotes terms quadratic or higher in B2/Bo. We<br />
have performed exact three-dimensional trajectory<br />
calculations to confirm that Eq. 1 suffices for the times and<br />
orbits of the numerical examples discussed later. The axial<br />
frequency for each ion, including corrections to ©z due to<br />
the magnetic bottle, can be written from Eq. 1 as<br />
mco,<br />
A key illustration of this coupling was the use of the spindependent<br />
shift of the axial frequency to measure the g-<br />
(i)<br />
factor of the electron 8 ' 11 cooled at 4 K to the ground state<br />
of its cyclotron motion. The straightforward generalization<br />
of this to ions is not practical, since the shift is inversely<br />
proportional to mass at fixed observation frequency (oz.<br />
Extending this shift to a 100 amu ion with a proton<br />
magnetic moment under conditions similar to those used in<br />
the single electron experiments yields an unpractically<br />
small 4 nHz shift. A more difficult problem is that the<br />
three trapping frequencies and the Larmor frequency<br />
become inhomogeneously broadened due to the wide range<br />
of nm values present in a thermal ensemble. Minimizing<br />
this by ion cooling methods would be time-consuming and<br />
reducing this range to be less than the spin magnetic<br />
moment requires lower temperatures as mass increases,<br />
since this quantum decreases inversely with mass.<br />
3 Spin-Dependent Cyclotron Acceleration:<br />
IRICE<br />
Rather than attempt to measure the small spindependent<br />
term in the axial motion, we derive a form of<br />
spin-dependent cyclotron acceleration analogous to<br />
cyclotron excitation via ICR: this is mternally resonant /'on<br />
cyclotron excitation (IRICE). The resonant electric field of<br />
ICR is replaced by an oscillating magnetic gradient with<br />
components at the cyclotron and Larmor (©0) frequencies.<br />
This field is constructed by arranging two orthogonal<br />
quadrupole coils: one is parallel to the x-axis with current<br />
proportional to cos(©0t+n/2)cos(©+t), and the other is<br />
directed along the y-axis with current proportional to<br />
cos(©0t-Hi)cos(©+t-Hc/2). 14 With gradient field strength G<br />
for each, the total magnetic field is<br />
t = {B0 +Gzsin[(©0+© +<br />
-Gcos(oooO sm ( to +<br />
A quantum-mechanical description of a spin-1/2 magnetic<br />
moment in this field shows that the eigenstates of spin lie<br />
in the transverse plane, aligned such that the spin-<br />
dependent force Fs=(fi»v)B resonant with the ion<br />
cyclotron motion is<br />
Like the force in ICR excitation, this is resonant with the<br />
cyclotron motion, but with explicit spin dependence due to<br />
a gradient dipole force. 9 ' 10 * 15<br />
Neglecting the magnetron mode, the transverse ion<br />
motion is described as a cyclotron oscillation of radius p+<br />
and phase + (the sense of rotation used here is appropriate<br />
for a positively charged ion):<br />
(3)<br />
(4)<br />
(5)
222<br />
—- = -ra+p+fsin((a+t+
pa*:'<br />
Vol. 14, No. 1-4 223<br />
' 4 R.M. Weisskoff, G.P. Lafyatis, K.R. Boyce, E.A. Cornell,<br />
R.W. Flanagan, Jr., and D.E. Pritchard, J. Appl. Phys. 63,<br />
4599 (1988).<br />
5 E.A. Cornell, R.M. Weisskoff, K.R. Boyce, R.W.<br />
Flanagan, Jr., G.P. Lafyatis, and D.E. Pritchard, Phys. Rev.<br />
Lett. 63, 1674 (1989).<br />
6 D.P. Weitekamp and P.J. Pizarro, U.S. Patent No.<br />
4,982,088 (1991).<br />
7 C.R. Bowers, S.K. Buratto, P.J. Carson, H.M. Cho, J.Y.<br />
Hwang, L. J. Mueller, P.J. Pizarro, D.N. Shykind, and D.P.<br />
Weitekamp, SPIE Proc. 1435,36 (1991).<br />
8 R.S. Van Dyck, Jr., P.B. Schwinberg, and H.G. Dehmelt,<br />
in New Frontiers in High Energy Physics, edited by B.<br />
Kursunoglu, A. Perlmutter, and L. Scott (Plenum, New<br />
York, 1978); in Atomic Physics 9, edited by R.S. Van<br />
Dyck, Jr. and E.N. Fortson (World Scientific, Singapore,<br />
1984).<br />
9 M. Bloom and K. Erdman, Can. J. Phys. 40, 179 (1962).<br />
10 M. Bloom, E. Enga, and H. Lew, Can. J. Phys. 45, 1481<br />
(1967).<br />
n L.S. Brown and G. Gabrielse, Rev. Mod. Phys. 58, 233<br />
(1986).<br />
12 R.S. Van Dyck, Jr., F.L. Moore, D.L. Farnham, and P.B.<br />
Schwinberg, Rev. Sci. Instrum. 57, 593 (1986).<br />
13 G. Schmidt, Physics of High Temperature Plasmas. 2nd.<br />
ed. (Academic Press, New York, 1979).<br />
14 Note that this field configuration has different symmetry<br />
from that suggested in Ref. 9; the "Transverse Stern-<br />
Gerlach" experiment on ions suggested there uses a field<br />
whose symmetry does not in fact lead to a linear change in<br />
the cyclotron radius over an extended period of time.<br />
15 J.P. Gordon and A. Ashkin, Phys. Rev. A 21, 1606<br />
(1980).
224<br />
Bulletin of Magnetic Resonance<br />
Coordination Modes of Histidine Moiety in Copper (II) Dipeptide<br />
Complexes Detected by Multifrequency ESR<br />
1 Introduction<br />
R.Basosi, R.Pogni, G.Delia Lunga<br />
Department of Chemistry - University of Siena<br />
Pian dei Mantellini 44,53100 Siena, Italy<br />
The copper complexes of histidinecontaining<br />
peptides naturally occurring in<br />
blood plasma [1,2] can be used as a model<br />
for metal/protein interactions.In fact a low<br />
molecular weight peptide containing<br />
histidine or even histidine itself may<br />
compete with ceruloplasmin for copper in<br />
the blood, significantly altering free ligand<br />
levels. For this reason, the coordination<br />
behaviour of the histidine moiety in copper<br />
(II) dipeptides has been the subject of<br />
much investigation [3-7]. From this point<br />
of view the study under physiological<br />
conditions, namely room temperature and<br />
neutral pH, is crucial for maintaining the<br />
biological significance of the model.<br />
Unfortunately a definitive speciation has so<br />
far been hindered by the strong<br />
dependence on the ligand-to-metal ion ratio<br />
and the multiplicity of possible equilibria,<br />
function of pH, in solution.<br />
ESR may be used to study<br />
paramagnetic metal complexes, and<br />
multifrequency ESR in combination with<br />
computer simulation is a decisive tool for<br />
determining the chemical structure of<br />
copper biosystems under physiological<br />
conditions. For ESR spectra the<br />
requirement of a good fit at different<br />
frequencies put constraints on the precision<br />
of magnetic parameters and allows<br />
definitive assignments, despite the lack of<br />
resolution typical of room temperature<br />
spectra. In this paper the above method<br />
was used in the study of complexes of<br />
copper(II) and Glycylhistidine (GlyHis),<br />
Histidylglycine (HisGly) and p-Alanyl-L-<br />
Histidine or Carnosine (Cam) formed in<br />
solution in a large range of pH,<br />
concentration and metal/ligand ratios. The<br />
aim was to show how a computer-aided<br />
multifrequency ESR approach can provide<br />
valuable information on the systems under<br />
investigation, avoiding undesirable effects<br />
due to changes in the physical state of<br />
samples.<br />
2 Material and methods<br />
Glycyl-L-Histidine, L-Histidylglycine and<br />
p-Alanyl-Histidine (Carnosine) from<br />
Sigma Chemical Co. were used without<br />
further purification.<br />
Solutions were made with distilled<br />
water and the uncorrected pH was adjusted<br />
with HC1 or NaOH and determined with<br />
an "LCD" Model pH meter. Isotopically<br />
pure 63 CuO (from Oak Ridge National<br />
Laboratory, Oak Ridge , TN) was used for<br />
the EPR experiments. Stock solutions<br />
with 1:2 molar ratio were prepared, [Cu]<br />
= 10- 2 M, [peptide] = 2 xlO" 2 M.<br />
X-band EPR spectra were obtained<br />
with a Bruker 200D SRC X-band<br />
spectrometer, and S-band spectra were<br />
obtained with a microwave bridge by<br />
Medical Advances Inc., Milwaukee, USA.<br />
All the bridges were equipped with loopgap<br />
resonators (Jagmar, Krakow, Poland)<br />
operating at v = 9.5 GHz for the X-band<br />
bridge and v = 4.04 GHz for the S-band<br />
bridge. Microwave frequencies were<br />
measured with an XL Microwave Model<br />
3120 counter. The spectrometer was<br />
interfaced with a Compaq Deskpro<br />
486/50L computer with an 8-megabyte<br />
memory and a 50 MHz clock. The data<br />
were acquired using the EPR data system<br />
CS-EPR produced by Stelar Inc., Mede,<br />
Italy.<br />
The CUKOS program for the<br />
simulation of EPR spectra of fast tumbling<br />
copper complexes in an isotropic<br />
environment, written in QuickBasic, is<br />
based on Kivelson's theory of linewidth<br />
[8] and includes the second order shift<br />
equation of Bruno et al. [9] and the further<br />
assumption of Lorentzian lineshapes. A<br />
Monte Carlo calculation method was added
Vol. 14, No. 1-4 225<br />
to the program. The Monte Carlo method<br />
randomly varied selected spectral<br />
parameters within defined limits in order to<br />
fit experimental data.<br />
3 Results and Discussion<br />
In recent years, data for copper(II)<br />
complexes of histidine and histidine<br />
containing peptides has been generated by<br />
a variety of techniques such as infraredspectroscopy<br />
[10], x-ray diffraction [11],<br />
nuclear magnetic resonance [12], circular<br />
dichroism [13,14], thermodynamic [3] and<br />
potentiometric studies [15,16] and ESR at<br />
liquid nitrogen temperature [17,18].<br />
Table I<br />
between two species. Aggregation can<br />
occur [21] and often tedious empirical<br />
approaches for the formulation of good<br />
glasses must be employed. In water,<br />
freezing can change the local pH [20,22],<br />
and Wilson and Kivelson [8] found that<br />
the isotropic g- and A-values of copper<br />
acetylacetonate are actually temperature<br />
dependent. All of these complications can<br />
be avoided by the use of multifrequency<br />
ESR of copper complexes above 0 °C.<br />
ESR spectra obtained for Cu-Histidine<br />
complexes with a 1:1,1:2 and 1:100 molar<br />
ratios, in the pH range 5-9 are reported in<br />
Refs. [17,18]. At pH= 7.4 and high molar<br />
ratios the ESR pattern shows a<br />
Parameters Used to Simulate ESR Spectra for Cupric Ion complex with Dipeptides a<br />
system giso Ag AisO( 63 Cu) AA A N isO xc(ps)b Ref.<br />
63 Cu-His<br />
3N<br />
4N<br />
63 Cu-GlyHis<br />
63 Cu-(GlyHis)2<br />
63 Cu-(HisGly)2<br />
63 Cu-B-Ala-His<br />
2.098<br />
2.104<br />
2.1134<br />
2.107<br />
2.114<br />
2.1468<br />
0.162<br />
0.159<br />
0.171<br />
0.166<br />
0.235<br />
0.221<br />
61.50<br />
61.60<br />
69.32<br />
80.36<br />
54.4<br />
62.91<br />
137.4<br />
143.0<br />
174.95<br />
195.87<br />
193.79<br />
190.00<br />
a Values for the hyperfine and superhyperfine splittings arc given in G.<br />
11.1<br />
11.0<br />
12.4<br />
13.5<br />
11.03<br />
12.5<br />
b Correlation times in ps (10~ 12 sec.). 63 Cu magnetogyric ratio 0.70904 s^<br />
A conventional ESR spectroscopic<br />
approach is to freeze the sample and extract<br />
the magnetic parameters by analysis of the<br />
frozen solution powder pattern. There are<br />
many reasons why this is undesirable.<br />
Vanng&rd [19] called attention to changes<br />
in coordination. Falk et al. [20] observed<br />
the temperature dependence of equilibrium<br />
85.00<br />
85.00<br />
50.64<br />
88.23<br />
75.00<br />
45.00<br />
18<br />
18<br />
this work<br />
this work<br />
this work<br />
this work<br />
septet 1:3:6:7:6:3:1 consistent with mixed<br />
glycine-like and histamine-like<br />
coordination. At higher pH, this species is<br />
in equilibrium with a complex in which<br />
both the histidines are coordinated in the<br />
histamine way (4N). The magnetic<br />
parameters for the two species are<br />
reported in Table 1.
226 Bulletin of Magnetic Resonance<br />
Fig. 1 shows ESR spectra at X-band for<br />
the complex Cu-GlyHis in a fast tumbling<br />
regime. All the experimental spectra are<br />
paired with the simulations for different<br />
species.<br />
100 G<br />
Fig. 1. Exp. ( ) and simulated (- • - ) ESR<br />
spectra for Cu-GlyHis at pH = 7.3. a) monomeric<br />
species, b) 62 % monomeric species + 38 % biscomplex,<br />
c) bis-complex.<br />
Fig. la represents the monomeric species<br />
obtained for a 1:1 molar ratio and pH=7.3.<br />
This species is readly obtained for a molar<br />
ratio of 1:2 at lower pH. Fig. lc shows the<br />
ESR spectrum of the bis-complex. This<br />
species is obtained when there is a large<br />
excess of the ligand with respect to the<br />
metal. Magnetic parameters for the above<br />
two species are reported in Table 1. With a<br />
molar ratio of 1:2 and physiological<br />
conditions, the two species occur<br />
simultaneously present in solution and the<br />
overall ESR spectrum is reported in Fig.<br />
lb paired with the simulated spectra. The<br />
best fit in Fig. lb is obtained for a pattern<br />
in which 62% of the monomeric species is<br />
considered together with 38% of the biscomplex.<br />
As the fit is reasonable, the<br />
concomitant presence of the two species<br />
can be supposed. A similar best fit<br />
procedure was performed for S-band. Fig.<br />
2 shows the expansion of the second<br />
derivative of the mi = +3/2 component for<br />
the monomeric species of Cu-GlyHis in<br />
the X and S-bands.<br />
A<br />
Fig. 2. Expanded 2nd derivative experimental mj =<br />
+3/2 component ( ~) paired with<br />
simulation (-••-) for the monomeric species Cu-<br />
GlyHis at: a) X-band, b) S-band.<br />
The procedure proposed [23] is based on<br />
the relative intensities of the patterns of the<br />
three center lines in order to discriminate<br />
between three and four nitrogen<br />
coordination. The simulations are adjusted<br />
until the best fit is obtained. In this case a
Vol. 14, No. 1-4 227<br />
three nitrogens coordination (amino,<br />
peptide and imidazole) from the ligand<br />
molecule accounts for a very stable<br />
monomeric species. The process of<br />
simulation was initiated by reading the<br />
approximate values of all parameters from<br />
the spectra. The g-tensor values were<br />
adjusted by fitting the spectra at X-band<br />
with the highest sensitivity to these<br />
changes. The simulations for liquid phase<br />
EPR spectra started with a Monte Carlo<br />
calculation method. This method randomly<br />
varied selected spectral parameters within<br />
defined limits in order to fit experimental<br />
data. When a good fit was obtained at one<br />
frequency the magnetic parameters were<br />
tested at the other frequency and the<br />
parameters were varied by an iterative<br />
process until the best fit was obtained in<br />
the X- and S-bands for the two physical<br />
states of the samples. The use of second<br />
derivative displays was a crucial step in<br />
obtaining a good set of parameters. In<br />
Fig. 2, second derivative, (or second<br />
harmonic), display emphasizes sharp<br />
features and discriminates against broad<br />
features. This display is particularly useful<br />
for analyzing superhyperfine patterns.<br />
Shoulders in the spectrum become peaks<br />
with precisely defined turning points that<br />
are useful for accurate measurements of<br />
coupling constants. The correlation times<br />
(xc) reported in Table 1 are consistent with<br />
the proposed speciation at pH=7.3.<br />
Despite the fact that HisGly is chemically<br />
similar to GlyHis, ESR spectra obtained<br />
under similar conditions are strikingly<br />
different. The relative parameters are<br />
reported in Table 1. A big difference in<br />
Aiso is evident: 54.4 G for Cu-(HisGly)2<br />
and 80.4 G for Cu-(GlyHis)2. Data<br />
obtained at low temperature confirmed this<br />
assignment. In the case of the Cu-HisGly<br />
complex, if excess HisGly is present, there<br />
may be bis-complex formation in which<br />
deprotonation of the peptide-NH linkage<br />
is suppressed [24]. Voelter et al.<br />
confirmed this hypothesis with 13 C NMR<br />
experiments at pH 7, at which the glycine<br />
moiety is not involved in copper ligation<br />
[25]. The ligation is pure histamine-like<br />
coordination with two imidazole nitrogens<br />
and two amine nitrogens in the first<br />
coordination sphere of copper. This<br />
arrangement is in agreement with the EPR<br />
results previously obtained for histidine<br />
[17,18]. In fact in the case of Cu-HisGly,<br />
the carboxyl group is blocked by peptide<br />
bonding in favour of a total histamine-like<br />
coordination whereas in the histidine<br />
complex a mixture of histamine-like<br />
glycine-like structures has been proposed<br />
on the basis of experimental findings<br />
[17,18,25].<br />
Fig. 3 shows the ESR spectra obtained for<br />
Cu-Carn with a 1:2 molar ratio at pH=5.6<br />
paired with its simulation.<br />
Fig. 3. Experimental ESR spectrum for Cu-Carn at pH<br />
= 5.6 (——) paired with simulation (-•-).<br />
Low pH was explored because at pH=7 a<br />
very distorted ESR pattern, previously<br />
attributed to a predominant dimeric<br />
species, is obtained. If we increase the<br />
ligand concentration up to 1:100 molar<br />
ratio, a very different spectrum arises<br />
which is shown in Fig. 4 with its<br />
simulation.<br />
Fig. 4. Experimental Cu-Carn ESR spectrum ( )<br />
for 1:100 molar ratio paired with<br />
simulation (-•-).<br />
This spectrum is attributed to the Cu-<br />
(Carn)4 complex and has very different
228 Bulletin of Magnetic Resonance<br />
magnetic parameters. In this case, our data<br />
confirms results reported in the literature<br />
[26] for experiments at low temperature.<br />
4 Conclusions<br />
The ESR features of the above copperdipeptide<br />
complexes show very different<br />
coordination behaviour for homologue<br />
Species under similar solution conditions.<br />
At physiological pH, the predominant<br />
histidine species present in solution are a<br />
histamine-like and a mixed histamine-like<br />
glycine-like complex. Under the same<br />
conditions HisGly only shows the<br />
histamine-like coordination whereas for<br />
GtyHis, two species (monomeric and the<br />
bis-complex) characterized by different<br />
magnetic parameters occur simultaneously.<br />
For Carnosine, a dimeric species is<br />
dominant and is possibly in equilibrium<br />
with a monomeric form.<br />
A good characterization of the real species<br />
present in solution can be achieved only by<br />
exploiting the sensitivity of a computer<br />
aided multifrequency ESR approach. A<br />
crucial role is played by a procedure based<br />
on the high selectivity of the second<br />
derivative display.<br />
5 References<br />
1) CJ. Gubler, M.E. Lahey, G.E.<br />
Cartwright, M.M. Wintrobe,<br />
J.ClinJnvest., 32, 405, 1987<br />
2) D.R. Williams, C. Furnival, P.M. May<br />
in "Inflammatory Diseases and Copper",<br />
J.RJ. Sorenson Ed., Humana Press,<br />
Clifton, New Jersey, 45,1982<br />
3) G. Brookes, L.D.Pettit, J.C.S.Dalton<br />
1975,2112<br />
4) R.P.Agarwal, D.D.Perrin, J.C.S.<br />
Dalton 1975,268<br />
5) RJ. Sundberg, R.B. Martin,<br />
ChemRev. 1974, 74(4), 472<br />
6) H. Sigel, "Metal ions in biological<br />
systems", 1973, 63<br />
7) D.B. McPhail, B.A. Goodman,<br />
J.Chem.Soc.Faraday Trans. 1, 1987,<br />
83(12), 3683<br />
8) R. Wilson, D. Kivelson, /. Chem.<br />
Phys., 1966, 44, 4445; 1966, 44, 154<br />
9) G.V. Bruno, J.K. Harrington, M.P.<br />
Eastman, J.Phys.Chem., 1977, 81, 1111<br />
10) R.H.Carlson, T.L.Brown, Inorg.<br />
Chem., 1966, 5, 268<br />
11) N.Camerman, J.K.Fawcett, T.P.A.<br />
Kruck, et al., J.A.C.S., 1978, 100:9,<br />
2690<br />
12) H.Sigel, B.McCormick, J.A.C.S.,<br />
1971,93:8,2041<br />
13) L.Casella, M.Gullotti, J.Inorg.<br />
Biochem., 1983, 18,19<br />
14) E.W.Wilson, M.H.Kasperian,<br />
R.B.Martin, J.A.C.S., 1970, 92:18, 5365<br />
15) D.D.Perrin, V.S.Sharma, J.Chem.<br />
Soc. (A), 1967, 724<br />
16) T.P.A. Kruck, B.Sarkar, Can.J.<br />
Chem., 1973, 51, 3563<br />
17) R.Basosi, G.Valensin, E.Gaggelli et<br />
al., Inorg.Chem., 1986, 25, 3006<br />
18) M.Pasenkiewicz-Gierula, R.Basosi et<br />
al., Inorg. Chem., 1987, 26(6), 801<br />
19) T.Vanngard, in: Biological<br />
Applications of Electron Spin Resonance<br />
(H.M. Swartz, J.R.Bolton and D.C.Borg,<br />
eds) p. 411, Wiley-Interscience, New<br />
York (1972<br />
20) K.E.Falk, E.Ivanova, B.Roos,<br />
T.Vanngard, Inorg.Chem., 1970, 9, 556<br />
21) A.Saryan, K.Mailer, C.Krishnaruti, et<br />
al., Biochem. Pharmacol., 1981, 30, 1595<br />
22) Y.Orii, M.Morita, /. Biochem.<br />
(Tokyo), 1977, 81, 163<br />
23) J.S. Hyde, W.E. Antholine, W.<br />
Froncisz, R. Basosi, In Advanced<br />
Magnetic Resonance Techniques in<br />
Systems of High Molecular Complexity;<br />
N. Niccolai, G. Valensin, Eds.;<br />
Birkhausen: Boston, 1986; p 363<br />
24) I. Sdvagd, E. Farkas, A. Gergely<br />
J.Chem.Soc.Dalton Trans., 1982, 2159<br />
25) W. Voelter, G. Sokolowski, U.<br />
Weber, U. Weser, Eur.J.Biochem.,<br />
1975, 58, 159<br />
26) C.E. Brown, W.E. Antholine, W.<br />
Froncisz, J.C.S.Dalton, 1980, 590
Vol. 14, No. 1-4<br />
An EPR and ab initio Study of a Phosphaalkene Radical Anion, and Comparison<br />
with other Phosphorus-Containing Radical Ions<br />
M. Geoffroy*, G. Terron, A. Jouaiti<br />
Departement de Chimie Physique, Universite de Geneve, 30 Quai E. Ansermet, 1211, Geneve<br />
(Switzerland)<br />
P. Tordo<br />
Universite de Provence, CNRS-URA 1412,13397 Marseille Cedex 13 (France)<br />
and<br />
Y. Ellinger<br />
Ecole Normale Superieure et Observatoire de Paris, 24 rue Lhomond, 75005 (France).<br />
1 Introduction<br />
Electron Paramagnetic Resonance (EPR) is an<br />
efficient method to obtain information about the<br />
spin densities in organic radicals containing a<br />
heteroatom and, when used in conjonction with<br />
ab initio calculations, this spectroscopy can<br />
yield a precise description of the structure of<br />
these species. In the present study, our purpose<br />
is to compare the spin delocalization on the fol-<br />
lowing three paramagnetic moieties (-P= 13 C
230<br />
a P=P bond and to record the EPR spectrum of<br />
the corresponding frozen solution. Diphosphines<br />
are well known stable molecules and we have<br />
measured the 31 P anisotropic hyperfine<br />
interaction after freezing a solution of<br />
tetra,2,4,6,trimethylphenyldiphosphine [3] HI<br />
previously oxidized in an electrolytic cell.<br />
2 Experimental<br />
The various compounds have been obtained by<br />
adapting already published syntheses:<br />
ArP=C(H)Ar' [1], ArP=PAr [2] and Ar"2P-PAr"2<br />
[3] (where Ar: trit-t-butyl phenyl., An phenyl,<br />
Ar": 2,4,6, trimethylphenyl ). The electrolyses<br />
were performed in the EPR cavity of an X-band<br />
Bruker spectrometer by using solutions of the<br />
organophosphorus compound in presence of<br />
tetra-n-butyl amonium hexafluor phosphate. The<br />
31 P, 13 C and *H isotropic and anisotropic<br />
coupling constants were obtained after<br />
simulation of the experimental spectrum with a<br />
program using second order perturbation.<br />
The ab initio calculations were carried out on<br />
a Silicon Graphics (Iris 4D) and a Vax 6700<br />
Bulletin of Magnetic Resonance<br />
computer by using G-82 and G-90 Gaussian<br />
programs [4]. The 6-31G* basis set was<br />
generally used and annihilation of the spin<br />
contamination was performed for the UHF<br />
calculations. The coupling constants were<br />
obtained by calculating the expectation values<br />
of the Fermi contact interaction and of the<br />
hyperfine dipolar interaction. ROHF calculations<br />
were also carried out, in particular when the<br />
UHF method led to final values which<br />
were not equal to 0.75.<br />
3 Results and Discussion<br />
EPR spectra..<br />
(ArP=C(H)Ar')~. The liquid solution spectrum<br />
is characterized by a large splitting of 152 MHz<br />
and a rather broad linewidth exhibiting a poorly<br />
resolved additional structure [5]. Full deuteration<br />
of the C(H)C6H5 fragment only affects the<br />
linewidth which is then equal to 8 MHz . The<br />
spectrum obtained with ArP=C(H)C6D5 clearly<br />
shows a coupling of 11 MHz with the ethylenic<br />
proton while the spectrum obtained with<br />
ArP= 13 C(H)C6H5 indicates a 13 C coupling equal<br />
to 16 MHz. From these couplings and from the<br />
observed linewidth one can conclude that an<br />
appreciable spin delocalization occurs onto the<br />
phenyl ring and the simulated spectra are<br />
consistent with the following three additional<br />
proton couplings: 11MHz, 7MHz and 7MHz.<br />
The frozen solution spectrum obtained with<br />
(ArP=C(H)Ar')" is characterized by an axial<br />
hyperfine coupling with 31 P: T//=455MHz and<br />
Tx =1 MHz; the full deuteration of the<br />
fragment only decreases the
Vol. 14, No. 1-4 231<br />
linewidth whereas the use of the 13 C(H)C6H5<br />
fragment leads to an additional hyperfine<br />
interaction: 13 C-T// =47 MHz, 13 C-T±=1 MHz.<br />
(ArP=PAr)~ : As already mentioned [6,7,8], the<br />
liquid solution spectrum exhibits a 31 P isotropic<br />
coupling constant equal to 158 MHz. The<br />
frozen solution spectrum has still not been<br />
reported; it is characterized (Fig. 1) by an axial<br />
coupling tensor : 31 P: T//=458 MHz and T± =10<br />
MHz. No additional J H coupling is observed<br />
with this compound.<br />
100G<br />
Fig 1. EPR spectrum obtained with a frozen solution of<br />
(ArP=PAr)"<br />
(Ar"2P-PAr"2) + .Only the liquid solution spectrum<br />
has been previously reported [9], the<br />
splittings observed with the frozen solution<br />
spectrum (Fig. 2) show that the two phosphorus<br />
nuclei are equivalent and that their hyperfine<br />
tensors exhibit an axial symmetry: T/y= 796<br />
MHz and T± =356 MHz.<br />
100 G<br />
Fig. 2. EPR spectrum obtained with a frozen solution of<br />
(Ar"2P-PAr"2) + .<br />
A'<br />
Assuming a positive sign for all the hyperfine<br />
eigenvalues leads to the isotropic and<br />
anisotropic coupling constants shown in Table 1<br />
together with the spin densities calculated by<br />
using the atomic parameters obtained from [10].<br />
Table 1. Experimental coupling constants (MHz) and<br />
spin densities<br />
radical<br />
T<br />
il 7± Ps V<br />
ArP=CHR- 31p 152 303 -152 0.01 0.41<br />
ArP=PAT<br />
Ar'2PPAr' 2<br />
13 C<br />
31p<br />
31p<br />
31p<br />
31p<br />
16<br />
158<br />
158<br />
502<br />
502<br />
ab initio calculations.<br />
31<br />
300<br />
300<br />
293<br />
293<br />
-15<br />
-148<br />
-148<br />
-146<br />
146<br />
0.00<br />
0.01<br />
0.01<br />
0.04<br />
0.04<br />
0.18<br />
0.41<br />
0.41<br />
0.39<br />
0.39<br />
Phosphaalkene radical onion: In the Cs<br />
symmetry, the UHF optimized structure for<br />
(HP=CH2)- is characterized by HPC=97.5°, P-C<br />
=1.791 A, P-H= 1.428 A; using the ROHF<br />
method does not significantly affect these<br />
parameters. The structure of (HP=C(H)Ar')-<br />
was optimized (ROHF calculations) by fixing<br />
the geometry of the phenyl group and by<br />
assuming a planar structure: HPC=96.6°, H-P=<br />
1.420 A and P-C =1.761 A.<br />
Diphosphene onion and diphosphine cation.<br />
The optimized structures of these two species<br />
are known [11,12,13] and agree with our UHF<br />
results : for (HP=PH)": HPH=95.8°, P-P=<br />
2.133A (Cs symmetry) and for (H2P-PH2)+:<br />
HPH= 102.2°, PPH=104.09° and P-P= 2.164 A
232<br />
=0.750<br />
symmetry). For these two ions<br />
The various hyperfine coupling constants<br />
resulting from UHF calculations are given in<br />
Table 2 . For both the phosphaalkene and the<br />
diphosphene anions, the "parallel" 31 P coupling<br />
eigenvectors are oriented perpendicular to the<br />
molecular plane (x structure), whereas for the<br />
diphosphine cation these eigenvectors make an<br />
angle of 45° with the P-P bond direction in<br />
accordance with the orientation of the n" orbital.<br />
Table 2. Calculated 31 P hyperfine coupling constants<br />
(MHz).<br />
radical<br />
(HP=CH2)-<br />
(HP=PH)-<br />
(H2PPH2)+<br />
^iso<br />
25<br />
70<br />
445<br />
T ll<br />
130<br />
279<br />
328<br />
T il.<br />
-60<br />
-136<br />
-153<br />
T ±2<br />
-70<br />
-142<br />
-174<br />
For the radical ions containing no phenyl ring,<br />
the various spin densities calculated by using<br />
the UHF method are very similar to those<br />
obtained from ROHF calculations. These spin<br />
densities are shown in Table 3.<br />
Table 3. Calculated spin densities.<br />
radical<br />
(HP=CH2)" a<br />
(HP=CHAiO" a<br />
(HP=PH)' a<br />
+ b<br />
(H2P-PH2)<br />
phosphorus<br />
Ps Pp<br />
0.00 0.22<br />
0.00 0.44<br />
0.00 0.49<br />
0.06 0.42<br />
adjacent carbon or<br />
phosphorus<br />
Ps Pp<br />
0.00 0.77<br />
0.00 0.33<br />
0.00 0.49<br />
0.06 0.42<br />
a ) ROHF calculations, b ) UHF calculations.<br />
phenyl<br />
0.22<br />
Bulletin of Magnetic Resonance<br />
These data show that the substitution, in the<br />
phosphaalkene anion, of an ethylenic hydrogen<br />
atom by a phenyl ring drastically decreases the<br />
spin density on the carbon atom and increases<br />
the spin density on the phosphorus atom. The<br />
spin delocalisation onto the phenyl ring is in<br />
good accordance with the variation of the<br />
linewidth observed after deuteration of the<br />
benzene ring.<br />
The calculated hyperfine tensors for (HP=PH)"<br />
and (H2P-PH2) + reasonably agree with the<br />
values measured for (ArP=PAr)~ and<br />
(Ar"2PPAr"2) + . For (HP=CH2 )", the isotropic<br />
and anisotropic coupling constants are quite dif-<br />
ferent from those measured for (ArP=CF£Ar')-,<br />
but, as indicated by the spin densities calculated<br />
for (HP=CHAr')-> this difference is due to the<br />
spin delocalisation induced by the phenyl ring.<br />
0<br />
0 / 0<br />
In summary, the frozen solution EPR spectra<br />
confirm the x* structure for phosphaalkene and<br />
diphosphene radical anions ( negligible contri-<br />
bution of the s orbitals of the phosphorus or<br />
carbon atom to the SOMO) and the non-<br />
bonding character of the SOMO for the<br />
diphosphine cation.
Vol. 14, No. 1-4 233<br />
References<br />
* R. Appel, J. Menzel, F. Knoch and P.Volz, Z. anorg.<br />
allg. chem. 534,100 (1986).<br />
2 M. Yoshifuji, I. Shima, N. Inamoto, K.Hirotsu and T.<br />
Higuchi, J. Am. Chem. Soc. 103, 4587 (1981).<br />
3 B. I. Stepanov, E. N. Karpova and A. I. Bokanov, Zh.<br />
Obshch. Khim., 39, 1544 (1969).<br />
4 F.M.J. Frisch, M. Head-Gordon, G.W. Trucks, J.B.<br />
Foresman, H.B. Schlegel, K. Raghavachari, M. Robb,<br />
J.S. Binkley, C. Gonzalez, D.J. Defrees, D.J. Fox, R.A.<br />
Whiteside, R. Seeger, C.F. Melius, J. Baker, R.L.Martin,<br />
L.R. Kahn,J.J.P. Stewart, S. Topiol and J.A. Pople.<br />
GAUSSIAN 90, Gaussian Inc., Pittsburg, PA (1990).<br />
5 M. Geoffroy, A. Jouaiti, G. Terron, M. Cattani-<br />
Lorente, and Y. Ellinger. J. Phys. Chem. (in press).<br />
6 B. Cetinkaya,A. Hudson, M. F. Lappert and H.<br />
Goldwhite, J. Chem. Soc. Chem. Commun. 609 (1982).<br />
7 A.J. Bard, A.H Cowley, J.E. Kilduff, J. K. Leland,<br />
N.C. Norman and M. Palkulski, J. Chem. Soc.., Dalton<br />
Trans., 249 (1987).<br />
8 M. Calcasi, G. Grouchi, J. Escudie, C. Couret, L. Pujol<br />
and P. Tordo, J. Am. Chem. Soc., 108, 3130 (19886).<br />
9 M. Culcasi, G. Grouchi,and P. Tordo, J. Am. Chem.<br />
Soc.,107, 7191 (1985).<br />
10 J.R. Morton and K.F. Preston. JMagn. Reson., 30,<br />
577 (1978).<br />
11 M.T. Nguyen, J. Chem. Phys., 91, 2679 (1987).<br />
12 T. Clark, J.Am.Chem.Soc, 107,2598,(1985).<br />
13 D. Feller,E.Davidson and W.T.Borden, J. Am.Chem.<br />
Soc, 107, 2596 (1985)
234 Bulletin of Magnetic Resonance<br />
Conformational Substate Distribution in<br />
Myoglobin as studied by EPR Spectroscopy<br />
Anna Rita Bizzarri 1 ^ and Salvatore Cannistraro 1 ' 2 '<br />
' INFM-CNR, Dipartimento di Fisica dell'Universita', Perugia, Italy<br />
2 ) Dipartimento di Scienze Ambientali, Sezione Chimica e Fisica, Universita'<br />
della Tuscia, Viterbo, Italy<br />
Introduction<br />
It has recently been shown that EPR, in<br />
connection with the aid of a computer<br />
simulation approach, can be successfully<br />
applied to investigate the structural<br />
heterogeneity displayed by metallo-proteins [1-<br />
6]. The g-strain effect characterizing the low<br />
temperature EPR spectra of metallo-proteins<br />
can be interpreted by taking into account for the<br />
presence of an ensemble of molecules frozen in<br />
many slightly different structures [6-8].<br />
Different experimental and theoretical<br />
approaches point out that a protein molecule<br />
can assume a very large number of different<br />
substates, called conformational substates (CS)<br />
[9,10] whose sampling is important for the<br />
biological functionality of the protein [11]. At<br />
physiological temperature, proteins fluctuate<br />
among CS; such a behaviour affecting the<br />
kinetic response of the molecules. By<br />
decreasing the temperature, the protein<br />
solution undergoes a glass-like transition and<br />
the fluctuations among CS are progressively<br />
suppressed [12]. Below the glass-temperature,<br />
T „, the molecules are frozen in many different<br />
CS whose distribution may be modulated by<br />
external agents such as pressure, pH, solvent<br />
composition [13-15]. However, the role of the<br />
solvent on the dynamical coupling between the<br />
protein and the CS distribution is still open.<br />
To get further information on this aspect, we<br />
have analyzed the high and low spin ferric<br />
myoglobin (Mb) samples in different<br />
conditions. The EPR spectra of high spin ferric<br />
Mb samples have been interpreted in terms of a<br />
distribution of the crystal field parameters A ,,<br />
A 2 connected with the energy differences of<br />
the low-lying electronic states of the ferric ion;<br />
whereas, the EPR spectra of the low spin Mb<br />
samples have been analyzed in terms of a<br />
distribution of the tetragonal and rhombic<br />
splitting parameters, A and V. An accurate<br />
computer simulation of the spectra has allowed<br />
us to extract the parameters characterizing<br />
these distributions which, in turn, have been put<br />
into a relationship to the CS distribution. The<br />
effects on these distributions as induced by<br />
different solvent compositions and by different<br />
cooling rates are analyzed.<br />
Materials and experimental<br />
methods<br />
Mb EPR samples were prepared by dissolving<br />
commercial (Sigma Chem. Co.) lyophilized<br />
horse skeletal muscle Mb in 0.2 M phosphate<br />
buffer. The highest concentration of Mb in the<br />
solutions was about 5 mM. Final pH for Mb<br />
solutions was about 6.8. Ferricyanide was used<br />
to oxidize the heme iron to the ferric valence<br />
state and solutions were dialysed several times<br />
against buffers to remove the oxidant.<br />
Approximately a twofold molar excess of<br />
sodium azide was used to convert metMb to the<br />
low spin form. Samples in mixed waterglycerol<br />
solvent were prepared by adding<br />
aliquots of glycerol to Mb solutio until the<br />
required concentration was reached. A fast<br />
cooling rate (Fast) has been obtained by<br />
dipping the the samples into liquid nitrogen at<br />
77K; while in the slow cooling rate ( Slow), the<br />
system was cooled with a rate of 0.5 deg/min<br />
from300Kto WOK<br />
All the EPR spectra were recorded at 77 K by an<br />
X-band Varian E109 spectrometer equipped with<br />
a variable temperature control which was also<br />
used to cool the samples in a controlled way. To<br />
calculate the experimental g-values, a<br />
magnetic field calibration was performed with
Vol. 14, No. 1-4<br />
a Magnion Precision . NMR gaussmeter<br />
Mod.G-542; the microwave frequency being<br />
measured with a Marconi 2440 counter.<br />
The acquisition of EPR data was carried out on<br />
a HP 86A personal computer through a home<br />
made interface connected to a IEEE 488 bus [16].<br />
To run both simulations and bestfit programs,<br />
the same microcomputer was switched to an<br />
intelligent terminal of the main frame<br />
computer (VAX 8350), through a serial interface<br />
and an HP terminal emulator.<br />
Analysis of the EPR spectra<br />
It is well-known that the ferric ion, in Mb heme<br />
complexes, is placed in a crystalline electric<br />
field of cubic symmetry in which four ligands<br />
are provided by the four nitrogen atoms of the<br />
porphyrin ring, the fifth ligand is the nitrogen<br />
of the proximal histidine, finally, in the sixth<br />
coordination site different ligands can be<br />
bound. In general, the presence of a weak<br />
ligand causes the ferric ion to assume a high<br />
spin state, S=5/2; while a strong ligand<br />
determines a low spin state, S=l/2. In this paper<br />
we consider metMb in which the weak ligand<br />
HgO + is present, and azide Mb samples with the<br />
strong sixth ligand N% [17].<br />
The EPR spectrum, at 77 K, of metMb is<br />
characterized by two resonances, one at g ~ 6<br />
and a weaker one at g ~2 (spectrum not shown).<br />
This system can be described by the spin<br />
hamiltonian<br />
Hs=ge(l H-S+D[SZ 2 -S (S + 1)/3]<br />
where ge is the value for the free electron; D<br />
and E are the tetragonal and the rhombic zerofield<br />
splittings, respectively. For heme<br />
proteins, the condition of large zero field<br />
splitting is satisfied ( D -10 cm "*) [18] and<br />
only transitions within the lowest Kramers<br />
doublet occur; a fictitious spin S=l/2 can then be<br />
used to fully represent the spin Hamiltonian of<br />
the system, which for axial symmetry (g = g<br />
= gx and gz = g | |) can be expressed by<br />
H S = 9| | P H z SZ+QJLPI H X S x +H y s y) ( 2 )<br />
where g| |~ 2 and g ~ 6 are the g-values<br />
which are observed in the experimental<br />
spectra. Splitting of the in-plane value into two<br />
values, gx and gv, may result in a broadening<br />
(as in our case) or even in a splitting of the g-<br />
235<br />
6 line [19,20]. High order corrections, arising<br />
from spin-orbit mixing of the excited quartet<br />
states into the lowest Kramers doublet lead,<br />
under the assumption of a four-state model<br />
[21,22] (Fig. 1) to the following expression for<br />
gx and gy<br />
gxy=3ge±24E/D-18.7 (E/D) 2 - 12 n 2 (3)<br />
where the tetragonal zero-field splitting D is<br />
given by<br />
D = J L<br />
and the rhombic zero-field splitting E<br />
-2<br />
(4)<br />
2 (5)<br />
the spin-orbit mixing of excited quartet states<br />
into the lowest Kramers doublet is<br />
Jl<br />
~ 2 _ ^ / 1 . 1 ,1\<br />
\ \<br />
(6)<br />
i is the effective spin-orbit coupling constant<br />
(i = 300 cm ** ) which is reduced from the freeion<br />
valued = 420 cm ~* ; A |, A 2 and y are the<br />
energy differences between the low-lying<br />
electronic states of high spin ferric heme (see<br />
Fig. 1). •<br />
V///////////////A<br />
z<br />
V///////777/7//I/<br />
Z '/////// '///////A<br />
777777777777/7////<br />
Figure 1 Energy level diagram of the low-lying electronic<br />
states of high-spin heme. It has been assumed A-p 2000 cm"<br />
1 ,A2 ~ 6000 cm' 1 ' Y -60 cm" 1 . The shaded regions<br />
indicate the variability of the energy levels (not in scale).<br />
The low temperature (77 K) EPR spectra of<br />
azide Mb samples are characterized by three<br />
absorption lines to which three principal<br />
different g- values (about gx = 2.8, gy = 2.2 and
236<br />
gz = 1.7 ) correspond.<br />
Within the Griffith's model [23], the ground<br />
state electronic configuration is a ^T2 state<br />
that can be described by one hole in the shell<br />
made by the iron dxz, dyZ) dxy orbitals.<br />
Owing to the presence of a rhombic distortion,<br />
the orbitals are split into three Kramers<br />
doublets with energies respectively of -V/2,<br />
V/2 and A ( see Fig.2).<br />
7/77/7/7/////////A<br />
7/ 7777777777//////<br />
7/7777/7/777777/<br />
Figure 2 Energy hole levels of the low-lying d-orbitals for<br />
the low spin ferric ion. The shaded regions indicate the<br />
variability of the energy levels (not in scale).<br />
Accordingly, the EPR spectra of azide Mb<br />
samples can be described by the spin<br />
Hamiltonian associated to S=l/2<br />
H S=P ( 9x H x V9y Hy Sy+QzHzSz) ( 7 )<br />
The three principal g-values are given by the<br />
expressions<br />
gx=2[2 AC - B 2 + k B(C- A) (2 1/2 )]<br />
gy= 2[2 AC + B 2 + k B(C+ A) (2 1/2 )]<br />
gz=2[A 2 -B 2 +C 2 +k ( A 2 - C 2 )] (8)<br />
where k is the orbital reduction factor and A,<br />
B and C are the coefficient characterizing the<br />
Kramers doublet of the lowest energy<br />
where 11* >, !, <<br />
wavefunctions within t^ 2 |-1±> are the<br />
T values<br />
for V and A have been assigned, A, B and C<br />
can be calculated by solving for ^/ + and \j/~ ,the<br />
secular equations associated with the matrix<br />
which takes into account for both the spin-orbit<br />
Bulletin of Magnetic Resonance<br />
coupling and the distortion field; then,<br />
assigned a value for k, the g-values can be<br />
determined from eq.(8).<br />
In a general way, once the g values are<br />
known, the EPR spectra can be generated by<br />
computer simulation with the aid of a suitable<br />
model. The derivative field-swept EPR<br />
absorption spectrum, related to randomly<br />
oriented paramagnetic centers with S=l/2, can<br />
be reproduced by the expression<br />
dS(vc,H) vc h p(e, <br />
(10)<br />
where the 1 / g(e.) is the Aasa-Vanngard [24]<br />
correction, C is a constant that encompasses all<br />
the instrumental parameters, P(8, ) is the<br />
orientation dependent transition probability<br />
which, for an S =1/2 system, can be exactly<br />
expressed by [25]<br />
cos 2 + g 4 y sin 2 8 cos 2 + g 4 z cos :<br />
[g 4 x sin 2<br />
(11)<br />
finally f(H ) is the lineshape function<br />
(residual linewidth [26] centered at the<br />
resonance field HQ and with a linewidth<br />
parameter a^ measured in magnetic field<br />
units.<br />
The integration over 8,4> in eq.(10) takes into<br />
account for the random orientation of the<br />
molecular axes with respect to the magnetic<br />
field.<br />
Use of eq.(10) is not, however, sufficient to<br />
reliably reproduce the EPR spectra of metalloproteins.<br />
It is known in fact that EPR spectra<br />
of metallo-proteins are characterized by a<br />
large inhomogeneous broadening resulting<br />
into a spread of the g-tensor values (g-strain)<br />
[1,4,6]. Such an effect can be interpreted by<br />
taking into account the presence of the CS<br />
distribution [8]; the heterogeneity<br />
corresponding to the presence of a frozen<br />
ensemble of molecules in different CS could<br />
entail a spread of the low-lying electronic state<br />
energies of the metal ion and, in turn, a<br />
modulation of the g-values [7,22]. On such a<br />
ground, and accordingly to previous works<br />
[7,8], it has been assumed that the low-lying<br />
electronic state energies of the ferric iron are<br />
distributed.<br />
In definitive, our spectra of metMb samples<br />
have been simulated by introducing two<br />
independent gaussian distributions for the
Vol. 14, No. 1-4<br />
crystal field parameters A j, A2; on the other<br />
hand, the azide Mb samples have been<br />
simulated by considering two independent<br />
gaussian distributions for the energy<br />
differences A and V.<br />
In both cases, the resulting simulated spectra<br />
can be visualized as a superposition, weighed<br />
in a proper way, of different spectra each one of<br />
them corresponds to a different g-tensor: the<br />
final expression of eq.(lO) convoluted with two<br />
gaussians is<br />
dS(vc,<br />
dH r 2%ar dH<br />
r 2- r ;<br />
(12)<br />
where T, and ?2 refer to the gaussian<br />
distributions.<br />
Computer-synthesized spectra have been used to<br />
fit the experimental EPR spectra; a<br />
minimization procedure of the x^-function,<br />
based on a simulated annealing approach [27],<br />
has been followed to extract the parameters A 0 ,,<br />
and and<br />
A2<br />
characterizing the two gaussian distributions<br />
for high and low spin Mb samples, respectively;<br />
the chi-square function being<br />
i = 1 a. (13)<br />
where I^PCHj) is the derivative of the<br />
experimental EPR absorption spectrum<br />
sampled at 500 discrete points of the magnetic<br />
field, I 8im (Hj,p) is the simulated spectrum,<br />
finally cTj is the standard deviation calculated<br />
for the i-th experimental point of the EPR<br />
spectrum by repeated runs.<br />
Results and Discussion<br />
Fig.3 shows two examples of the experimental<br />
and the corresponding simulated spectra for the<br />
analyzed metMb and azide Mb samples. The<br />
parameters A 0 ,, A°2,
238 Bulletin of Magnetic Resonance<br />
TABLE 1 Central values and variances of the gaussian distributions of the crystal field parameters Ax A2<br />
for high spin ferric Mb samples, and A V for low spin ferric Mb samples obtained by simulations (through<br />
eq.(12) ) of the experimental 77 K EPR spectra of Mb frozen solutions. Gly means that the sample has<br />
been prepared in 1:1 (by volume) water-glycerol mixture. Fast means that the sample has been submitted<br />
to a fast cooling rate ( about 50 deg/min). Slow means that the sample has been submitted to a slow<br />
cooling rate ( about 0.4 deg/min ).<br />
SAMPLE<br />
High spin Mb (Fast)<br />
High spin Mb (Slow)<br />
High spin Mb+Gly (Fast)<br />
High spin Mb+Gly (Slow)<br />
Low spin Mb (Fast)<br />
Low spin Mb (Slow)<br />
Low spin Mb+Gly (Fast)<br />
Low spin Mb+Gly (Slow)<br />
A? cm- 1<br />
2266<br />
2250<br />
2194<br />
2248<br />
Ao<br />
3.03<br />
3.01<br />
3.08<br />
3.05<br />
parameters distributions in both the high and<br />
low spin case and in fast and in slow cooled<br />
samples; such an effect can be interpreted in<br />
terms of a decrease in the structural<br />
heterogeneity of the protein as induced by<br />
glycerol [7,8]. Different molecular<br />
mechanisms could be invoked to interpret such<br />
an effect; it is possible that addition of glycerol<br />
could result into a viscosity-induced damping<br />
of the protein motion; on the other hand,<br />
changes in the dielectric properties of the<br />
solvent could result into a different shielding of<br />
the electrical charges of the amino acid<br />
residues [30,31] and then into a modification of<br />
the protein dynamics; moreover, glycerol, by<br />
decreasing the ice-crystal dimensions, could<br />
minimize the freezing strains [32].<br />
In the high spin Mb samples, the slow cooling<br />
rate induces, in presence and in absence of<br />
glycerol, a narrowing of the crystal field<br />
parameters distributions A , and A2- Such an<br />
effect, which has been observed also in high<br />
spin ferric Hb samples [7] can be interpreted in<br />
different ways. First of all, a sort of<br />
"condensation" could take place in the<br />
molecules populating the frozen CS distribution<br />
[7]; moreover, the cooling rate could induce<br />
some modifications in the state of the hydration<br />
water, as it has been observed by calorimetric<br />
259<br />
239<br />
280<br />
232<br />
CTA<br />
0.13<br />
0.14<br />
0.09<br />
0.10<br />
A£ cm- 1<br />
5759<br />
5750<br />
5500<br />
5417<br />
Vo<br />
2.00<br />
2.01<br />
2.03<br />
2.02<br />
aA, cm- 1<br />
936<br />
919<br />
549<br />
516<br />
*v<br />
0.06<br />
0.08<br />
0.05<br />
0.07<br />
measurements [33], and consequently affect the<br />
CS distribution; finally, it cannot be ruled out<br />
the possibility that cooling rate, acting on the<br />
crystal growth, modifies the freezing straininduced<br />
effects that might be present in low<br />
temperature heme-proteins [34,35]. It is aspected<br />
that all these mechanisms should be operative<br />
also in low spin Mb samples, in which,<br />
however, it has been observed that the slow<br />
cooling rate induces an increase of the<br />
variances a ^ and 0y.<br />
The different behaviour registered in this case<br />
requires a deeper investigation of the role<br />
played by the strong sixth ligand in connection<br />
with the freezing procedure.<br />
In particular, we can speculate about the<br />
possibility that the freezing procedure might<br />
affect, in some way, the average position and<br />
also the spread of the N'g ligand. In this<br />
context, it should be noted that the different<br />
number of ligand orientation, as induced by<br />
different cooling rates, have been observed in<br />
oxycobalt Mb [34].Therefore, different cooling<br />
rates could result into different arrangements<br />
of the ligand with respect to the metal ion.
Vol. 14, No. 1-4<br />
Conclusions<br />
EPR results to be a suitable spectroscopy to<br />
study the structural heterogeneity displayed<br />
by the ferric Mb samples in both the high<br />
and the low spin configuration as derived<br />
from an ensemble of different structures. In<br />
particular, it can fruitfully be applied to<br />
investigate the dynamical coupling between<br />
the solvent and the protein CS distribution.<br />
References<br />
[I] D. O. Hearshen, W. R. Hagen, R. H.<br />
Sands, H. J. Grande, H. L. Crespi, I. C.<br />
Gunsalus, W. R. Dunham, J. Magn.<br />
Resonance 69 (1986)440.<br />
[2] C. More, P. Bertrand and J.P. Gayda, J.<br />
Magn. Resonance 73 (1987) 13.<br />
[3] A. S. Yang and B. J. Gaffney, Biophys. J. 51<br />
(1987)55.<br />
[4] J.C. Salerno: Biochem. Soc. Trans. 13<br />
(1985) 611.<br />
[5] A. S. Brill, F. G. Fiamingo, D. A. Hampton,<br />
J. Inorg. Biochem. 28 (1986) 137.<br />
[6] S. Cannistraro, J. Phys. France 51 (1990)<br />
131.<br />
[7] A.R. Bizzarri and S. Cannistraro, Appl.<br />
Magn. Res. 2(1991)627.<br />
[8] A. R. Bizzarri and S. Cannistraro, Biophys.<br />
Chem. 42(1991)79.<br />
[9] H. Frauenfelder, F. Parak, R. D. Young<br />
Ann. Rev. Biophys. Biophys. Chem. 17 (1988)<br />
451.<br />
[10] V.I.Goldanskii and Y. F. Krupyanskii,<br />
Quart. Rev. Biophys. 22 (1989) 39.<br />
[II] A. Ansari, J. Berendzen, S. F. Bowne, H.<br />
Frauenfelder, I. E. T.Iben, T. B. Sauke, E.<br />
Shyamsunder, R. D. Young, PNAS USA 82<br />
(1985) 5000.<br />
[12] H. Frauenfelder, in Proteins and glasses<br />
(E. Luscher, G. Fritsch and G. Jacucci eds.)<br />
Amorphous and Liquid Materials, Nato series,<br />
Martinus Nijhoff Publishers, Dordrecht (1987).<br />
[13] M. K. Hong, D. Braunstein, B. R. Cowen,<br />
H . Frauenfelder, I. E. T. Iben, J. R. Mourant,<br />
P. Ormos, R. Scholl, A. Schulte, P. J. Steinbach,<br />
A. H. Xie, R. D. Young, Biophys. J. 58 (1990)<br />
429.<br />
[14] H. Frauenfelder, N. A. Alberding, A.<br />
Ansari, D. Braunstein, B. R. Cowen, M. K.<br />
Hong, I. E. T. Iben, J. B. Johnson, S. Luck, M.<br />
C. Marden, J. R. Mourant, P. Ormos, L.<br />
Reinisch, R. Scholl, A. Schulte, E.<br />
Shyamsunder, L. B. Sorensen, P. J. Steinbach,<br />
A. H. Xie, R. D. Young, K. T. Yue, J. Phys.<br />
Chem. 94(1990) 1024.<br />
[15]E.E. Di Iorio, U.R. Hiltpold, D. Filipovic,<br />
K.H. Winterhalter, E. Gratton, E. Vitrano, A.<br />
Cupane, M. Leone and L. Cordone, Biophys. J.<br />
59(1991)742.<br />
[16] G. Giugliarelli, P. Tancini and S.<br />
Cannistraro, J. Phys. E (Sci. Instrum) 22<br />
(1989)702.<br />
[17] M. Kotani Adv. Quantum Chem. 4 (1968)<br />
227.<br />
[18] J.F. Gibson, in ESR and NMR of<br />
Paramagnetic Species in Biological and<br />
related Systems, , eds. I. Bertini and R.S.<br />
Drago, Reider Publishing Company (1979).<br />
[19] J. Peisach, W.E. Blumberg, S. Ogawa,<br />
E.A. Rachmilewitz and R. Oltzik, J. Biol.<br />
Chem. 25(1971)3342.<br />
[20] S. Cannistraro, Chem. Phys. Lett. 122<br />
(1985)165.<br />
[21]C.P. Scholes, J. Chem.Phys.52(1970)4890.<br />
[22] F.G. Fiamingo, A. S. Brill, D. A.<br />
Hampton, R. Thorkildsen, Biophys. J. 55<br />
(1989)67.<br />
123] J.H.E. Griffith, Theory of Transition of<br />
Metal Ions, Cambridge Univ. Press, London<br />
and New York (1961).<br />
[24] R. Aasa and T. Vanngard, J. Magn.<br />
Resonance 19(1975) 308.<br />
[25] A. Isomoto, H. Watari, M. Kotani, J.<br />
Phys. Soc. Japan 29 (1970) 1571.<br />
[26] S. Cannistraro and G. Giugliarelli, Mol.<br />
Phys. 58(1986) 173.<br />
[27] S. Kirkpatrick, C. D. Gelatt, M. P. Vecchi,<br />
Science 220(1983)671. .<br />
[28] C. T. Migita, K. Migita and M. Iwaizumi,<br />
Biochim. Biophys. Acta 743(1983)290.<br />
[29] C. More, J.P. Gayda and P. Bertrand J.<br />
Magn. Resonance 90(1990)486.<br />
[30] G.P. Singh, H. J. Schink, H. V.<br />
Lohneysen, F. Parak, S. Hunklinger, Z. Phys.<br />
B55(1984)23.<br />
[31] G. P. Singh, F. Parak, S. Hunklinger, K.<br />
Dransfert, Phys. Rev. Lett. 47(1981) 685.<br />
[32]W.R. Hagen, J. Magn. Resonance44(1981)<br />
447.<br />
[33] W. Doster, A. Bachleitner, R. Dunau, M.<br />
Hiebl, E. Luscher, Biophys. J. 50 (1986) 213.<br />
[34] H. Hori, M. Iketa-DSaito and T.<br />
Yoketani, Nature 288 (1980) 501.<br />
[35]M.R. Ondrias and D.L. Rousseau, Science<br />
213(1981)2066.<br />
239
240 Bulletin of Magnetic Resonance<br />
1 Introduction<br />
Effect of Paramagnetic Ions in<br />
Aqueous Solution<br />
for Precision Measurement<br />
of Proton Gyromagnetic Ratio<br />
Ae Ran Lim, Chang Suk Kim<br />
Korea Research Institute of Standards and Science,<br />
Taejon 305-606, Korea<br />
and<br />
Sung Ho Choh<br />
The measurement of proton gyromagnetic ratio }'p'<br />
has been the object of an intensive experimental<br />
program for several decades [l]-[3]. The<br />
gyromagnetic ratio of the proton is defined as a<br />
resonance frequency «p divided by a magnetic field<br />
Bo [4], when a spherical water sample at 25°C is<br />
applied by a magnetic field. The ?P' for a defined<br />
pure water sample is somewhat difficult to measure<br />
because of the weak absorption due to the long<br />
relaxation time of the proton [5]. In order to<br />
reduce the relaxation time, paramagnetic ions<br />
are added to the water sample [6]. However,<br />
as the concentration of paramagnetic ions<br />
increases, the resonance point of *H shifts and<br />
differs according to the sample shapes.<br />
The purpose of present work is to investigate the<br />
effect of paramagnetic ions(Fe 3 *, Mn 2 *, Co 2 *, and<br />
Cu 2 *) on the l H NMR in paramagnetic aqueous<br />
solutions [Fe(NO3)3- 9H2O, FeCl3, MnCl2-4H2O,<br />
CoCh-2H2O, and CuCl2-2H2O]. !H NMR in<br />
aqueous solution containing paramagnetic ions was<br />
measured as a function of concentration for the<br />
fixed spherical and cylindrical sample shapes, and<br />
the spinning cylindrical sample shape. The<br />
magnetic susceptibility per unit volume of the<br />
paramagnetic solution has also been measured as a<br />
function of concentration at room temperature.<br />
The interaction between the ' H nucleus and<br />
Department of Physics, Korea University,<br />
Seoul 136-701, Korea<br />
paramagnetic ion is discussed in terms of the<br />
shift of >H resonance point measured with two<br />
sample shapes and the magnetic susceptibility of<br />
the solution. From these experimental results,<br />
we discuss the paramagnetic solution having a<br />
short relaxation time and nearly zero shift of<br />
resonance point to implement the precision<br />
measurement of proton gyromagnetic ratio in a<br />
low magnetic field.<br />
2 Experimental Method<br />
*H NMR experiment was performed by a<br />
Brucker model MSL 200 FT pulse spectrometer.<br />
l H NMR of paramagnetic aqueous solutions in<br />
high magnetic field of 4.7 and 11.74 T, and<br />
frequency of 200.13 and 500 MHz was<br />
measured with two sample shapes at room<br />
temperature. The magnetic susceptibility has<br />
been measured using the Gouy magnetic balance.<br />
2.1 *H NMR and Relaxation Time in<br />
Paramagnetic Aqueous Solutions<br />
The paramagnetic aqueous solutions [Fe(NO3)3-<br />
9H2O, FeCb, MnCl2-4H2O, CoCl2-2H2O,<br />
CuCl2-2H2O] were prepared by dissolving<br />
paramagnetic ions of various concentration in
Vol. 14, No. 1-4 241<br />
distilled water. The linewidth and the shift of 'H<br />
resonance point were measured at room<br />
temperature according to the shape of the<br />
sample tube and the paramagnetic ion<br />
concentration of aqueous solution. The<br />
linewidth and the resonance point of *H NMR<br />
in pure water were also measured.<br />
The spin-lattice relaxation time (7"i) and<br />
spin-spin relaxation time (T2*) were determined<br />
from the signal of l H NMR at room<br />
temperature by the inversion recovery method<br />
and the inverse of linewidth, respectively.<br />
2.2 Magnetic Susceptibility of Paramagnetic<br />
Aqueous Solution<br />
The density of paramagnetic aqueous solution<br />
was measured with the mass and volume, and<br />
the magnetic susceptibility per unit volume was<br />
obtained with the susceptibility per unit mass.<br />
3 Experimental Results<br />
3.1 Shift of !H Resonance Point and Linewidth<br />
In case of cylindrical sample tube, the<br />
resonance point of r H NMR was shifted to the<br />
negative direction with respect to that of pure<br />
water according to the concentration of<br />
paramagnetic ion as shown in Figure 1. It was<br />
nearly unchanged with the variation of<br />
concentration of paramagnetic ion in<br />
Fe(NO3>3-9H2O solution. However, the identical<br />
Fe 3+ ion in FeCb shows the largest shift of the<br />
resonance point. Since the distribution of<br />
valence electrons is influenced by the chemical<br />
bqnding of an atom, it could be explained that<br />
the displacement of nuclear magnetic resonance<br />
frequency depends upon the chemical<br />
environment [7].<br />
Figure 2 shows the linewidth of } H NMR<br />
according to the concentration of paramagnetic<br />
ions. Here we have used the Lorentzian<br />
absorption lineshape, and the linewidth<br />
corresponds to the full width at the half<br />
maximum. The Co 2+ and Cu 2+ ions are<br />
almost ineffective to the linewidth. Whereas the<br />
paramagnetic aqueous solutions containing Fe 3+<br />
ion influence the linewidth as a function of<br />
concentration of paramagnetic ions. The trend<br />
in aqueous solution containing the Mn 2+ ion<br />
differs from those in other paramagnetic ions.<br />
For the case of spherical sample shape, the<br />
shift of resonance point as a function of<br />
concentration of paramagnetic ions is shown in<br />
OO<br />
o<br />
5<br />
-10-<br />
> -15-1<br />
B = 4.7 T<br />
gy-;wHyt-w3<br />
Fe(NO3)3-9H2O<br />
FeCb<br />
-20-<br />
0 To" 20 30<br />
Ion Concen.(10 20 ions/cc)<br />
Figure 1. The frequency shift of 'H NMR signal as<br />
a function of concentration of paramagnetic ions in<br />
aqueous solution contained in a fixed cylinder (* is<br />
the resonance point of 'H in pure H2O).<br />
e(NO3) 9H2O<br />
0<br />
Ion Concen.(10 20 ions/cc)<br />
Figure 2. The linewidth of 'H NMR signal as a<br />
function of concentration of paramagnetic ions in<br />
aqueous solution contained in a fixed cylinder (* is<br />
the linewidth of 'H in pure H2O, 0.15 kHz).<br />
Figure 3. The resonance point of 'H NMR was<br />
shifted to the positive direction compared with<br />
that of pure water. The frequency shift was<br />
nearly unchanged with the variation of the<br />
concentration of paramagnetic Cu 2+ ion. The<br />
linewidth of *H NMR in the spherical shape is<br />
the same as that in the cylindrical one as shown<br />
in Figure 2, i.e. the linewidth has no difference<br />
between the cylindrical and spherical samples<br />
within the experimental error.
242 Bulletin of Magnetic Resonance<br />
00<br />
O<br />
o<br />
20-<br />
15-<br />
B = 4.7 T<br />
B"<br />
MnCl2-4H2O<br />
•CuCU-2H;O<br />
0 10 20 30<br />
Ion Concen.(10 20 ions/cc)<br />
Figure 3. The frequency shift of 'H NMR signal as<br />
a function of concentration of paramagnetic ions in<br />
aqueous solution contained in a sphere (* is the<br />
resonance point of 'H in pure H2O).<br />
However, the shift of the resonance point of<br />
l H in case of spinning cylindrical sample<br />
shows the similar trend as that of the fixed<br />
spherical sample, but with the larger shift than<br />
the fixed sphere as shown in Figure 4. Each<br />
100-<br />
B = 11.74 T /A^MnClr4H,O<br />
FeClj<br />
/<br />
Fe(NOj)j-9HzO<br />
CoCl;-2H2O<br />
CuCl2-2H2O<br />
0 10 20 30<br />
Ion Conceti.(10 20 ions/cc)<br />
Figure 4. The frequency shift of *H NMR signal as<br />
a function of concentration of paramagnetic ions in<br />
aqueous solution contained in a spinning cylinder (*<br />
is the resonance point of 'H in pure H2O).<br />
line in these figures was determined by the least<br />
square fit with the experimental data.<br />
3.2 l H Relaxation Time<br />
Figure 5 shows the *H spin-lattice relaxation<br />
time (Ti) for the cylindrical and spherical<br />
samples measured by the inversion recovery<br />
method. As the concentration of paramagnetic<br />
ions increased, the relaxation time was<br />
shortened. The 'H relaxation time of<br />
paramagnetic ions containing Co 2 * or Cu 2+ was<br />
longer than that of Mn 2 + or Fe 3 + . The 'H<br />
spin-lattice relaxation time of 2.51 s measured in<br />
pure water was consistent with the previously<br />
reported value of 2.3 s at 20°C and 29 MHz<br />
[8]. Figure 6 shows the spin-spin relaxation<br />
time (7*2 *) for the cylindrical and spherical<br />
samples obtained with the inverse linewidth of<br />
the resonance line. This result shows the<br />
similar trend as that of Ti. However, the J H<br />
relaxation time 7z* is shorter than Ti in<br />
aqueous solutions.<br />
10-5. 10 20<br />
10 21<br />
! T c(N03)-9HzO<br />
1022<br />
Ion Concentration(ions/cc)<br />
Figure 5. Spin-lattice relaxation time Ti of 'H due<br />
to the paramagnetic ions in aqueous solution<br />
contained in the fixed cylindrical and spherical<br />
shapes. Both shapes have the same values within<br />
experimental error.<br />
3.3 Magnetic Susceptibility<br />
The magnetic susceptibility per unit volume<br />
of the paramagnetic aqueous solution was<br />
obtained by the Gouy magnetic balance as a<br />
function of concentration of paramagnetic ions
Vol. 14, No. 1-4 243<br />
o<br />
GO<br />
"B<br />
H<br />
o<br />
•X<br />
i<br />
'E,<br />
10" 4 -<br />
10 -5-<br />
B = 4.7 T<br />
ESS 3SS1<br />
10 20 10 21<br />
Ion Concentration(ions/cc)<br />
Figure 6. Spin-spin relaxation time r2* of >H due<br />
to the paramagnetic ions in aqueous solution<br />
contained in the cylindrical and spherical shapes.<br />
Both shapes have the same values within<br />
experimental error.<br />
at room temperature as shown in Figure 7.<br />
The susceptibility was proportional to the<br />
concentration of paramagnetic ions.<br />
4 Analysis and Discussion<br />
10-4-<br />
MnCh-4H2O<br />
FeClj<br />
«CoCl2-2H2O<br />
Fe(NO3)3-9H2O<br />
CuCU- 2HjO<br />
10-6-<br />
1020 1021 1022<br />
Ion Concentration(ions/cc)<br />
Figure 7. The magnetic susceptibility of aqueous<br />
solution as a function of concentration of<br />
paramagnetic ions.<br />
The l H resonance point in pure water differs<br />
from that in the paramagnetic aqueous<br />
solution. The paramagnetic aqueous solution<br />
induced the shift of resonance point due to the<br />
presence of paramagnetic ions. In this study,<br />
we have tried to search a suitable paramagnetic<br />
solution, having the short relaxation time and<br />
nearly zero shift of resonance point in order to<br />
obtain the correct proton resonance frequency in<br />
a low magnetic field.<br />
4.1 The Shift of 'H Resonance Point and<br />
Interaction Factor<br />
For a liquid, the time averaged field at a<br />
nucleus may be divided into three significant<br />
components<br />
Bav = Bo B' + B" (1)<br />
where Bo is the external magnetic field, which is<br />
the main component in Bav. B' is the magnetic<br />
shielding field at the nucleus due to the induced<br />
motion of the electrons in the atom or molecule.<br />
B" is the magnetization field due to the<br />
paramagnetic ions to shorten the spin-lattice<br />
relaxation time T\ of the nuclear spin system.<br />
The dipole interaction between the *H nucleus<br />
and paramagnetic ion is given by [9]<br />
(2)<br />
The field B\ is ascribed to the induced magnetic<br />
dipoles on the surface of a small hypothetical<br />
sphere with its center at the nucleus. This is<br />
the so-called Lorentz or cavity field and has the<br />
value (4H/3)M, where W is the magnetization.<br />
The field Bz is the familiar demagnetizing field,<br />
defined by #2 = -aW, where a is the<br />
demagnetizing factor. The value of a is 47?/3 and<br />
2n for the spherical and infinite cylindrical<br />
sample perpendicular to the field, respectively. It<br />
might be expected that the remaining field #3<br />
due to those paramagnetic ions inside the<br />
hypothetical sphere would be exactly zero.<br />
However, it is found experimentally that #3 may<br />
differ significantly from zero. Therefore, we<br />
define an "interaction factor" q=B3/M. The<br />
expression for B" hence becomes [10]<br />
B" = [(4B/3) - a + q]M (3)<br />
The magnetization H was obtained from the<br />
susceptibility per unit volume according to the<br />
concentration of paramagnetic ions. Also, the
V '<br />
244 Bulletin of Magnetic Resonance<br />
shift of resonance point(B") for paramagnetic<br />
ions with respect to resonance point of proton in<br />
pure water was measured from *H NMR<br />
experiment in various paramagnetic solutions.<br />
Using the magnetic field induced to the 'H<br />
nucleus and the value of magnetization, we<br />
calculated the interaction factor q from eq.(3)<br />
for the fixed spherical and cylindrical samples.<br />
A summary for the various paramagnetic ions is<br />
given in Table 1. The consistency of the<br />
Table 1. Experimental values of the interaction<br />
factor q for the fixed spherical and cylindrical<br />
samples, obtained with eq.(3).<br />
paramagnetic<br />
ions<br />
Fe 3 *<br />
Fe 3 '<br />
Mn 2i<br />
Co 2t<br />
Cu 2 '<br />
chemical<br />
compound<br />
Fe(NO3)3-9HjO<br />
FeCl3<br />
MnCI2-4H2O<br />
CoCl2-2H2O<br />
CuCl2-2H2O<br />
Q<br />
cylinder<br />
2.19<br />
0.85<br />
1.70<br />
1.28<br />
0.39<br />
value<br />
sphere<br />
1.77<br />
0.86<br />
1.21<br />
1.06<br />
0.85<br />
experimental data is indicated by the agreement<br />
between the interaction factors for the<br />
corresponding cylindrical and spherical cases.<br />
The amount of disagreement can be attributed<br />
partly to the experimental error and partly to<br />
the meniscus effect and the lack of perfect<br />
sphericity of the spherical sample.<br />
4.2 Relaxation Time<br />
The spin-lattice relaxation time measured by the<br />
inversion recovery method with a pulse sequence<br />
of 180°(2 /is) - t - 90°( 1 us) - 5 fis(Td) - free<br />
induction decay. The ringing down delay-time Ta<br />
was used to remove the effect of the pulse and<br />
the free induction decay was measured with<br />
time t.<br />
The spin-lattice relaxation time obtained with<br />
the inversion recovery method decreases as the<br />
concentration of paramagnetic ions increases.<br />
The relaxation time measured with the<br />
spherical sample is similar to that with the<br />
cylindrical sample. The 'H relaxation time of<br />
paramagnetic solution containing Co 2 + or Cu 2 *<br />
shows longer than that containing Mn 2 * or<br />
Fe 3+ . The spin-lattice relaxation time of J H in<br />
various paramagnetic solutions is shorter than<br />
that in pure water because of the interaction<br />
between the nuclear spin and paramagnetic<br />
ions. When the number of paramagnetic ions<br />
was increased, the shortening mechanism of the<br />
relaxation time could be understood as follows.<br />
If the number of paramagnetic ions is increased,<br />
the nuclear spin is coupled more with the<br />
magnetic field produced by the paramagnetic<br />
ions. This magnetic interaction between the<br />
nuclear spin and magnetic field of the<br />
paramagnetic ions can contribute to the decrease<br />
in the spin-lattice relaxation time [11].<br />
In case of the short spin-lattice relaxation<br />
time Ti, the following relation generally holds<br />
[12] :<br />
(V)" 1 =<br />
(4)<br />
where Tz is the "natural" spin-spin relaxation<br />
time, and Tz' is the time due to the field<br />
inhomogeneity. The value of Tz" is measured<br />
from the full width at half maximum of the<br />
NMR lineshape.<br />
The linewidth of 'H NMR was brodened when<br />
the concentration of paramagnetic ions was<br />
increased. In case of the aqueous solution<br />
containing Fe 3+ ion, the linewidth was<br />
remarkably increased according to the<br />
concentration of paramagnetic ions. The<br />
linewidth could be broadened by the magnetic<br />
dipole field produced by the paramagnetic ions<br />
at the site of *H nucleus. Normally the dipole<br />
field of the paramagnetic ions has the field<br />
strengths of several thousands times greater than<br />
that due to the magnetic monents of the nucleus,<br />
but it is averaged out at the site of ] H nucleus.<br />
Consequently only a small effect, the linewidth<br />
broadening is occured in the magnetic resonance<br />
[13].<br />
5 Conclusion<br />
For the spherical and cylindrical samples, B"<br />
would be always zero and positive,<br />
respectively, if q were zero. The deviation of<br />
the shift of resonance point between the<br />
experimental results and the theoretical<br />
prediction (q = 0) could be understood as an<br />
effect due to an additional interaction between<br />
the paramagnetic ions and the 'H nucleus.<br />
The spin-lattice and spin-spin relaxation times of<br />
l H NMR in paramagnetic aqueous solution were<br />
shortened as the concentration of paramagnetic ions<br />
was increased.<br />
From these experimental results, we found<br />
that the paramagnetic solution having the<br />
short relaxation time and nearly zero shift of<br />
resonance point is CUO22H2O aqueous<br />
solution. Therefore, the aqueous solution<br />
containing Cu 2+ would be the best candidate to<br />
implement the precise determination of the
Vol. 14, No. 1-4 245<br />
proton gyromagnetic ratio.<br />
Acknowledgement<br />
This work was supported by the Ministry of<br />
Science of Technology and in part the KOSEF<br />
through the SRC of Excellence Program<br />
(1991-94).<br />
References<br />
[1] E.R.Williams and P.T.Olsen, Phys. Rev.<br />
Lett. 42, 1575 (1979).<br />
[2] E.R.Williams, G.R.Jones, J.S. Song,<br />
W. D. Phillips, and P. T. Olsen, IEEE<br />
Trans. Instrum. Meas. IM-38(2), 233<br />
(1989).<br />
[3] H. Nakamura, N. Kasai and H. Sasaki,<br />
IEEE Trans. Instrum. Meas. IM-36, 196<br />
(1987).<br />
[4] N. Bloembergen, Nuclear Magnetic<br />
Relaxation (W. A. Benjamin, New York,<br />
1961), Chap. 4.<br />
[5] J. H. Simpson and H. Y. Carr, Phys. Rev.<br />
Ill, 1201 (1958).<br />
[6] N. Bloembergen, E. M. Purcell, and R. V.<br />
[7]<br />
Pound, Phys. Rev. 73, 679 (1948).<br />
J. T. Arnold, S. S. Dharmatti and M. E.<br />
Packard, J. Chem. Phys. 19, 509 (1951).<br />
[8] N. Bloembergen and W. C. Dickinson,<br />
Phys. Rev. 79, 179 (1950).<br />
[9] W. C. Dickinson, Phys. Rev. 77, 736<br />
[10]<br />
(1950).<br />
A. R. Lim, S. H. Choh, Saemulli 26, 381<br />
(1986).<br />
[11] A. Abragam, The Principles of Nuclear<br />
Magnetism(Oxford Univ. Press, Oxford,<br />
1961), Chap. 3.<br />
[12] D. Pines and C. P. Slichter, Phys. Rev.<br />
100, 1014 (1955).<br />
[13] W. C. Dickinson, Phys. Rev. 81, 717<br />
(1951).
246 Bulletin of Magnetic Resonance<br />
1. Introduction<br />
Magnetic Resonances of 23 Na and 14 N Nuclei<br />
in Single and Multi-Domain<br />
.Crystals of Ferroelectric NaN02<br />
Sung Ho Choh and Kee Tae Han<br />
Department of Physics, Korea University, Seoul<br />
136-701, Korea.<br />
Betsuyaku [1] and Kanashiro et al [2] reported<br />
that the two satellite lines of 23 Na NMR<br />
in a ferroelectric NaNCh crystal were asymmetric,<br />
and that even the central line for Bo//a<br />
deviates appreciably from the first derivative of<br />
the Gaussian line shape. Kanashiro et al [2] tried<br />
to explain this effect in terms of the nuclear<br />
dipole coupling between the Na atoms. Hughes<br />
and Pandey [3] also studied the same effect, and<br />
atempted to explain it by means of the magnetic<br />
dipolar third moments. They concluded that the<br />
asymmetry arises from the perturbation of the<br />
magnetic dipolar coupling, and this is the<br />
intrinsic property of NaNO2 crystal from the<br />
similarity of experimental results obtained with<br />
two different crystals.<br />
In connection with these reports [1-3], the<br />
present work is focused on the 23 Na NMR line<br />
shape as well as the electric field gradient<br />
(e.f.g.) at 23 Na and 14 N sites in NaNO2 crystal<br />
by employing the NMR and NQR technique. Two<br />
NaNO2 crystals prepared here have different<br />
domain states each other: one has a multi-domain<br />
and the other has a single domain state,<br />
respectively.<br />
2. Experimental<br />
A. Procedure<br />
The NaNO2 is body centered orthorombic and<br />
its space group is C2v 20 -Im2m in the ferro-<br />
electric phase at room temperature [4],. The used<br />
sample crystals were grown from the melt. Virgin<br />
crystals of NaNC>2 turned out to be in the<br />
multi-domain state (herein named by Sm). The<br />
crystal of single domin (Ss sample) was prepared<br />
by applying an external electric field of 3 kV/cm<br />
along the ferroelectric axis of a virgin crystal for<br />
8 hrs near its Tc of 163.5°C. Their domain states<br />
were confirmed by empolying optical polarizing<br />
microscopy and etching technique [5].<br />
The NMR signals for 23 Na nucleus were observed<br />
with an rf-frequency of 6 MHz by using<br />
the cw-NMR (Varian WL-112). They were recorded<br />
by setting the magnetic field modulation at<br />
35 Hz with an approximate amplitude of less than<br />
one-third of the resonance line width. The<br />
pulse-NMR (Bruker MSL 200) system was also<br />
employed to investigate the exact line shape of<br />
23 Na with an inversion recovery sequence, and<br />
NQR measurements were made by using the<br />
Robinson type spectrometer [6]. All the<br />
measurements were made at room temperature.<br />
B. Results<br />
Since the 23 Na nucleus has a spin of 1 = 3/2,<br />
three resonance lines are typically observed: one<br />
central and two satellite lines. The line shape of<br />
23 Na NMR in two single crystals of NaNO2 had<br />
been measured [7] by using the cw-NMR<br />
spectrometer. In the Ss sample, as shown in Fig.<br />
1, the asymmetric satellite lines were observed<br />
for Bo//a and Bo//c, consistent with the previous<br />
reports [1-3]. However, in the Sm, the observed<br />
satellite lines was nearly symmetric (see Fig. 2).
Vol. 14, No. 1-4<br />
(a) Bo//c-axis<br />
(b) Bo//a-axis<br />
-4— -+-<br />
510.0 511.0 531.0 531.5 552.5 553.5 (mT)<br />
505.0 510.0 530.5 531.5 558.0 559.0 (mT)<br />
Fig. 1. In the Ss sample, asymmetric satellite lines were observed for (a) Bo//c and<br />
(b) Bo //a with the cw-NMR spectrometer.<br />
(a) Bo//c-axis<br />
(b) Bo//a-axis<br />
510.0 511.0 531.0 552.5 553.5 (mT)<br />
505.0 510.0 530.5 531.5 558.0 559.0 (mT)<br />
Fig. 2. In the Sm sample, symmetric satellite lines were observed for (a) Bo//c and<br />
(b) Bo//a with the cw-NMR spectrometer.<br />
247
248 Bulletin of Magnetic Resonance<br />
The line width of the satellite lines (ABS) in Ss<br />
was found to he broader than that in Sm. whereas<br />
the central line in these two samples is symmetric<br />
and its line shape is nearly the same; it was 0.29<br />
mT (0.21 mT) for Bo//a (B0//c). Meanwhile, the<br />
quadrupole coupling constant (Qcc) and asymmetry<br />
parameter (7?) for 23 Na of these two samples are<br />
nearly the same within the experimental accuracy<br />
(AQcc : + 0.003 MHz, AT? : ± 0.003) [7].<br />
For the 14 N NQR measurements in the two<br />
samples, there were no detectable differences in<br />
their line widths (At 1 * and AP~ ) and NQR parameters.<br />
The quadrupole parameters at the 23 Na<br />
and 14 N nuclei in the Sm and Ss samples are<br />
summarized in Table 1 together with the line<br />
width. The only difference between the Sm and Ss<br />
samples is in the line shape of 23 Na NMR and<br />
its line width.<br />
The line shape of 23 Na NMR in the two<br />
samples obtained with the pulse-NMR is shown in<br />
Fig. 3 and 4, respectively. In both of two<br />
samples, the central line has one peak and its<br />
line shape is a Gaussian as well known [1-3]. So<br />
the central line is not shown in Fig. 3 and 4,<br />
where PL S(III) (i4i s(m) ) is the satellite line of low<br />
(high) frequency, and superscript s(m) stands for<br />
the single domain (multi-domain) state. In Fig. 3,<br />
(a) shows that each satellite line obtained with Ss<br />
has two peaks (PL S ' and VL S , m s and v\\ s ' ),<br />
and (b) displays their first derivative curves.<br />
However, in the Sm sample, the satellite line (vi m<br />
or PH" 1 ) has only one peak and it is well fitted<br />
with a Gaussian line shape as shown in Fig. 4,<br />
which displays VL m representatively since the two<br />
satellite lines all are symmetric. The inner set<br />
(vi s and vn s ) of the two couples obtained with<br />
Ss [Fig. 3 (a)] corresponds to those (^L m and fH m )<br />
with the Sm sample (Fig. 4). The outer set (vi s '<br />
and m s ' ) in Ss is newly observed additionally.<br />
3. Discussion<br />
Betsuyaku and Kanashiro et al [1-2] had<br />
measured the NMR line shapes for Bo //a and<br />
Bo//c. Their data for Bo//a was distinctively<br />
asymmetric but those for Bo//c was hardly asymmetric.<br />
They, by considering the dipolar interaction,<br />
argued that this effect arised from the<br />
geometrical arrangement of Na atoms in NaNO2<br />
crystal having two fold symmetry along the<br />
b-axis. Namely, the dipole coupling between the<br />
two Na nuclei along the a-axis contribute considerably<br />
to the asymmetry of the satellite lines<br />
because the interatomic distance along the a-axis<br />
is about two-thirds of that along the c-axis.<br />
Kanashiro et al [2] determined the sign of the<br />
Qcc at Na in NaNCh by comparing the line<br />
shape of the satellites for Bo //a with the<br />
calculated line spectrum of two interacting I =<br />
3/2 spins. However, Hughes et al indicated that<br />
there was some uncertainty in the determination<br />
by Kanashiro et al since the resonances showed a<br />
substantial inhomogeneous quadrupole broadening,<br />
and no steps were taken to confirm that the<br />
asymmetry was indeed associated with the dipolar<br />
interaction and not with the quadrupole<br />
broadening. Hughes et al confirmed the sign of<br />
Qcc determined by Kanashiro et al by comparing<br />
the magnetic dipolar third moment with the<br />
observed asymmetry, and reported that the<br />
asymmetry due to the quadrupole broadening was<br />
dominant at some crystal orientations by measuring<br />
quantitatively the asymmetry of inhomo-<br />
Table 1. The quadrupole parameters and the line widths at room temperature.<br />
samples<br />
Sm<br />
Ss<br />
23 Na NMR<br />
Qcc(MHz) .»?(%) ABs(Bo//a) ABs(Bo//c)<br />
1.094 11.0 0.30 mT 0.23 mT<br />
1.094 11.0 0.38 mT 0.25 mT<br />
14 N NQR<br />
Qcc (MHz) 7?(%) At> + (kHz) Ai""(kHz<br />
5.490 35.7 0.66 0.53<br />
5.490 35.7 0.67 0.53
Vol. 14, No. 1-4<br />
(b)<br />
240 230<br />
(KHz)<br />
-250 (KHz)<br />
240 230 (KHz) -240 -250 (KHz)<br />
Fig. 3. The satellite lines of 23 Na NMR in the Ss sample observed for B0//c with the<br />
pulse-NMR spectrometer, (a) Each satellite line has two peaks, (b) the first derivative<br />
curve of (a)<br />
m<br />
I<br />
240 230 KHz<br />
Fig. 4. The satellite line of 23 Na NMR in the Sm sample observed for Bo//c with the<br />
pulse-NMR spectrometer. The satellite line has only one peak and is well fitted<br />
with a Gaussian line shape, where + denotes signal trace and the dashed line is<br />
a best fitted Gaussian line shape.<br />
\<br />
249
250 Bulletin of Magnetic Resonance<br />
geneous quadrupole broadening. They also suggested<br />
that the asymmetry was the intrinsic property<br />
of this crystal by observing the similar quadrupolar<br />
broadening for two different crystals.<br />
As can be seen in Fig. 1, our NMR results<br />
show that the satellite line shape in Ss is<br />
asymmetric, which is in accordance with the<br />
previous reports [1-3]. However, those in Sm is<br />
nearly symmetric (Fig. 2). The exact line shapes<br />
of Fig. 3 and 4 obtained with the pulse-NMR<br />
system may provide a decisive clue for the origin<br />
of the observed asymmetry. Fig. 3(a) shows that<br />
each satellite line in Ss has two peaks, and the<br />
line intensity of the inner set is stronger than<br />
that of the outer one. Fig. 3(b) displays their<br />
first derivative curves in Fig. 3(a). It can be<br />
expected to obtain the asymmetric satellite line of<br />
Fig. 1 by drawing a curve as an envelope of the<br />
satellite line made of two peaks of Fig. 3(b).<br />
Whereas, in the Sm sample, the satellite line has<br />
only one peak as shown in Fig. 4 and it is found<br />
to be well fitted with a Gaussian line shape.<br />
Our experimental results shown in Fig. 4<br />
suggest that the asymmetry due to the dipole<br />
coupling is quite small or very weak based on the<br />
fact that the satellite line in the Sm is well fitted<br />
with a Gaussian. Two peaks in the satellite line<br />
of Fig. 3 may imply that there are possibly two<br />
different absorption centers in the Ss sample.:<br />
one is a normal set of the satellite lines<br />
corresponding to those in Sm and the other is a<br />
new one (vi s ' and PH S ' ). Such a new set might<br />
have orignated from the strained part of the<br />
crystal during the process of the polarization<br />
reversal to make the virgin crystal into a single<br />
domain state. Meanwhile, as listed in Table 1,<br />
the Qcc and r? for 23 Na in these two samples<br />
with the cw-NMR were the same within the experimental<br />
uncerntainty [7]. NQR results such as<br />
line widths and quadrupole parameters of 14 N in<br />
Sm and Ss are also similar to each other. The<br />
appreciable difference between the Sm and Ss<br />
samples is only in the line shape of 23 Na NMR.<br />
These facts on the e.f.g. at 23 Na and 14 N sites<br />
supports the proposal of the existence of two<br />
absorption regions in the Ss sample. The asymmetry<br />
and broad line width of the satellites in<br />
the Ss sample can be understood by means of the<br />
formation of a new region in the crystal due to<br />
the strain induced by the polarization reversal.<br />
From our experimental results, one can<br />
propose that the asymmetric line shape of the<br />
satellite line of 23 Na is due to an imperfection<br />
caused by the applied electric field rather than<br />
the magnetic dipole coupling in the crystal. In<br />
general, crystal imperfection may be produced<br />
during the crystal growth [8] or by the external<br />
stress [9], and due to the external electric field<br />
as the present case [10].<br />
4. Conclusion<br />
We observed the symmetric line shape as well<br />
as the asymmetric one in the 23 Na NMR in the<br />
ferroelectric NaNO2 crystal by employing the<br />
cw-NMR and pulse NMR technique. The pulse<br />
NMR spectrometer provided a better resolved resonance<br />
line shape than the cw-NMR. Each satellite<br />
line obtained with Ss has two peaks, while<br />
only one peak in the Sm sample. Thus the<br />
observable asymmetry might have originated from<br />
the existence of two absorption regions in the<br />
crystal such as a normal region and a new one<br />
abnormally induced in NaNO2 crystal. Such a<br />
new region seems to be due to the internal strain<br />
induced during the process of polarization reversal<br />
by the external electric field. These results imply<br />
that the asymmetry is not the intrinsic property of<br />
an NaNO2 crystal but due to a kind of crystal<br />
imperfection caused by strain.<br />
Acknowledgements<br />
This work is supported by the Korea Science<br />
and Engineering Foundation through the SRC of<br />
Excellence Program (1991-94). Authors are<br />
grateful to Professor R. Blinc of University of<br />
Ljublijana for his helpful discussion during his<br />
visit to Korea in February 1992.
Vol. 14, No. 1-4<br />
References<br />
[1] H. Betsuyaku, J. Phys. Soc. Japan. 27,<br />
1485 (1969)<br />
[2] T. Kanashiro, T. Ohno, and M. Satoh, J.<br />
Phys. Soc. Japan. 54, 2720 (1985).<br />
[3] D. G. Hughes and L. Pandey, J. Mag.<br />
Reson. 75, 272 (1987)<br />
[4] M. I. Kay and B. C. Frazer, Acta cryst.<br />
J4, 56 (1961)<br />
[5] K. T. Han, Ph. D. thesis, Korea Univ.<br />
(1992)<br />
[6] J. Lee and S. H. Choh, Rev. Sci. Instr.<br />
53, 232 (1982)<br />
[7] K. T. Han, H. W. Shin, I. W. Park and<br />
S. H. Choh, J. Korean. Phys. Soc. 25,<br />
67 (1992)<br />
[8] S. H. Choh, J. Lee and K. H. Kang,<br />
Ferroelectrics 36, 297 (1981)<br />
[9] K. T. Han, T. H. Yeom and S. H. Choh,<br />
Ferroelectrics 107, 349 (1990)<br />
[10} Private communication with Prof. R. Blinc<br />
of University of Ljublijana<br />
251
252<br />
1. Introduction<br />
Knight Shifts and<br />
Spin Dynamics in Disordered<br />
Systems<br />
M.J.R. Hoch and S.T. Stoddart<br />
Department of Physics and<br />
Condensed Matter Physics Research Unit,<br />
University of the Witwatersrand, Johannesburg<br />
The metal insulator (MI) transition is a<br />
problem that has received much attention in<br />
solid state physics. Heavily doped<br />
semiconductors have featured prominently in<br />
this work with Si:P the archetypal system.<br />
In order to explain the observed low<br />
temperature properties, such as the magnetic<br />
susceptibility, of MI systems in the vicinity<br />
of the critical concentration, n(;, of dopant<br />
atoms, a phenomenological two—fluid model<br />
has been proposed [I]. The present work is<br />
concerned with interpreting 29 Si NMR<br />
relaxation time measurements and Knight<br />
shifts for Si:P and Si:(P,B) in the vicinity of<br />
nc. The results are analyzed in terms of<br />
available theory in the context of the twofluid<br />
model.<br />
2. The Two-Fluid Model and the Bhatt-<br />
Lee Theory<br />
For the just metallic or just insulating<br />
phases of MI systems, the two—fluid model<br />
distinguishes between two types of electron<br />
spins. As the transition is traversed, the<br />
proportions of the two fluids change. The<br />
fluids are comprised of localized moments<br />
associated with isolated dopant atoms, or<br />
small clusters of dopant atoms, on the one<br />
hand, and delocalized moments on the other.<br />
In broad terms, the localized moments<br />
dominate in determining the magnetic<br />
properties in the vicinity of nc, while the<br />
delocalized moments determine the electrical<br />
properties, such as the conductivity.<br />
Bulletin of Magnetic Resonance<br />
For n < nc, the localized moments<br />
constitute a disordered antiferromagnetic<br />
system. The exchange Hamiltonian may be<br />
written in the usual way as<br />
H = Hi - 1 ' -J '<br />
where Jij is the exchange coupling between<br />
spins i and j and Si and Sj are the spin<br />
operators.<br />
In order to explain the behaviour of<br />
the magnetic susceptibility x with<br />
temperature for n < nc,Bhatt and Lee [2]<br />
have developed a theory in which the<br />
exchange coupling Jy between nearest<br />
neighbour pairs of localized moments is used<br />
to separate the moments into two groups.<br />
In simple terms, spin pairs with Jjj >> kT<br />
are tightly coupled or frozen in the singlet<br />
state and effectively do not contribute to X-<br />
The remaining spins do contribute and the<br />
susceptibility may be written in terms of the<br />
Curie law susceptibility as<br />
n Curie (1)<br />
Numerical procedures were used to<br />
determine ne(T), the effective number of<br />
spins at temperature T. Good quantitative<br />
agreement was obtained with available<br />
experimental susceptibility data. Bhatt and<br />
Lee used the following form for the J<br />
distribution in their calculations :<br />
P(J) « J-tt, with 0.6 < a < 0.8. This led to<br />
X « T-«, as observed.
Vol. 14, No. 1-4<br />
3.<br />
29 Si Spin Relaxation and Localized<br />
Electron Spin Dynamics (n < nc)<br />
Previous work [3] has provided strong<br />
evidence that localized moments dominate in<br />
determining the 29 Si spin lattice relaxation<br />
times at low temperatures. These moments<br />
constitute an exchange coupled reservoir to<br />
which the nuclear spin system is coupled via<br />
the dipolar interaction. Hoch and Holcomb<br />
[3] have analyzed available Ti results using<br />
a model which allows for spin diffusion<br />
to the localized moments, which exhibit<br />
fluctuations in orientation with a frequency<br />
related to the strength of the exchange<br />
coupling to neighbouring spins. Frozen spin<br />
pairs have been excluded by introducing an<br />
effective number of spins in the spirit of the<br />
Bhatt—Lee approach to the susceptibility.<br />
Available T( data, measured at various<br />
fields B, at 1.5 K and 13.5 mK were fitted<br />
reasonably well by choosing the spectral<br />
function for the spin fluctuations to have the<br />
form f(w) « »/w and using calculated values<br />
for other quantities, such as the spin<br />
diffusion coefficient D.<br />
The best fits were obtained by keeping<br />
the diffusion barrier radius, b, constant<br />
independent of the field used. Calculations,<br />
however, suggest that b should vary as in B.<br />
We now propose that the form of the<br />
spectral function can be obtained from the<br />
form of the J distribution used by Bhatt and<br />
Lee. An outline of the treatment is given<br />
below. Converting the J distribution into a<br />
r distribution and integrating over all<br />
correlation times for the unfrozen spins gives<br />
J(u/) = P(T) dr<br />
with P(T) « V 7 " 2 "" • The form of the J(w,r)<br />
may be chosen in various ways corresponding<br />
to different possible forms for the<br />
correlation function. Our calculations<br />
suggest that the form of J(w) is not very<br />
sensitive to this, and for simplicity, we use<br />
an exponential correlation function, leading<br />
to a Debye form for J(w,r). If we put a - 1,<br />
this leads to J(w) « 70;, as used in the work<br />
referred to above. Numerical integration is,<br />
in general, necessary for other values of a.<br />
For values of a < 1, we obtain a<br />
spectral function which varies as l /uP- and<br />
therefore more slowly with frequency than<br />
the l /w form. This permits b to vary when<br />
fitting the experimental data. Further<br />
details will be published elsewhere.<br />
The approach to the spectral function<br />
for an amorphous antiferromagnet, outlined<br />
above, is consistent with the ideas of the<br />
253<br />
Bhatt-Lee theory. Susceptibility measurements<br />
as a function of n and T and nuclear<br />
relaxation measurements as a function of n,<br />
B and T can be explained in terms of the<br />
properties of the localized fluid component.<br />
4. 29Si Knight Shifts (n > nc)<br />
Knight shifts have previously been measured<br />
as a function of dopant concentration n at<br />
4.3 K and 1.5 K [4,5,6] in various magnetic<br />
fields. At high concentrations the Knight<br />
shifts tend towards the Pauli susceptibility<br />
behaviour (xP « n^*), as expected for a<br />
metal. At lower concentrations the values<br />
fall below the Pauli susceptibility<br />
predictions.<br />
Using a tight binding approximation<br />
based on the approach given by Kaveh<br />
and Liebert [7], we have given a semiquantitative<br />
explanation [8] for the observed<br />
Knight shift behaviour for both Si:P and<br />
Si:(P,B). The delocalized moment fluid<br />
determines the Knight shift.<br />
In order to see whether there is any<br />
temperature dependence of the Knight shift,<br />
we made measurements on two just metallic<br />
samples at temperatures down to 50 mK in<br />
an Oxford dilution refrigerator. The samples<br />
were in the form of a stack of wafers, which<br />
were well anchored thermally to the high<br />
purity copper tail used in the refrigerator.<br />
The field of 1 T was supplied by a high<br />
homogeneity superconducting solenoid.<br />
The results are shown in Figure 1.<br />
10<br />
10<br />
10 -6<br />
10 -2<br />
,Si:P<br />
oSi:P<br />
(n/nc = 1.6)<br />
(n/nc = 1.6) Ref. 5<br />
• Si:(P,8)(n/nc = 1.1)<br />
°Si:(P.BHn/ne a 1.1) Ref. 6<br />
10 1<br />
Temperature (K)<br />
Figure 1<br />
The mean Knight shift for Si:P (n/nc = 1.6)<br />
andSi:(P,B) (n/nc = 1.1) as a function of<br />
temperature down to 50 mK.<br />
Within experimental uncertainty it can be<br />
seen that there is no temperature<br />
dependence of the mean Knight shift <br />
over the temperature range 4 K — 50 mK for<br />
either Si:P or Si:(P,B).<br />
In terms of the two-fluid model this<br />
aro irarxr \irpaV<br />
10
254 Bulletin of Magnetic Resonance<br />
interactions between the localized and<br />
delocalized spins. The local susceptibility of<br />
the delocalized electrons does not change<br />
with temperature.<br />
5. Conclusion<br />
Measurements of the 29 Si relaxation rates<br />
and Knight shifts as a function of donor<br />
concentration, magnetic field and<br />
temperature have provided evidence which<br />
supports the two—fluid model for the MI<br />
transition in Si:P. The Ti measurements<br />
provide information on the spin dynamics of<br />
the localized moments, while the Knight<br />
shifts probe the properties of the delocalized<br />
moments.<br />
The relaxation rate<br />
interpreted using ideas<br />
Bhatt—Lee theory for<br />
susceptibility. The form<br />
results may be<br />
based on the<br />
the magnetic<br />
of the spectral<br />
function for spin fluctuations in the<br />
amorphous antiferromagnet system has been<br />
deduced using an accepted form for the<br />
distribution of exchange couplings.<br />
Knight shifts may be explained using a<br />
tight binding model for the delocalized fluid.<br />
No temperature dependence of has<br />
been found over the range 4 K — 50 mK.<br />
This may be interpreted to mean that<br />
interactions between the two fluids, which<br />
occupy spatially distinct regions in the<br />
sample, are very weak.<br />
6. References<br />
1. H. Alloul and P. Dellouve, Phys. Rev.<br />
Lett. 59, 578 (1987).<br />
S. Sachdev, R.N. Bhatt and<br />
M.A. Paalanen, J. Appl. Phys. 63,<br />
4285 (1988).<br />
2. R.N. Bhatt and P.A. Lee, Phys. Rev.<br />
Lett. 48, 344 (1982).<br />
3. M.J.R. Hoch and D.F. Holcomb, Phys.<br />
Rev. B 38, 10550 (1988).<br />
4. S. Kobayashi, Y. Fukagawa, S. Ikehata<br />
and W. Sasaki, J. Phys. Soc. Jpn. 45,<br />
1276 (1978).<br />
5. M.J. Hirsch and D.F. Holcomb, Phys.<br />
Rev. B 33, 25201 (1986).<br />
6. M.J.R. Hoch, U. Thomanschefsky and<br />
D.F. Holcomb, Physica B 165/166, 305<br />
(1990).<br />
7. M. Kaveh and A. Liebert, Phil. Mag.<br />
Lett. 58, 247 (1988).<br />
8. S.T. Stoddart and M.J.R. Hoch - to be<br />
published in Phys. Rev. B.
Vol. 14, No. 1-4 255<br />
Numerical Design and Evaluation of Broadband<br />
Pulse Sequences for 1=1 spin systems<br />
Debra Lynne Mattiello, Jonathan Callahan, Todd Alam^ and Gary Drobny<br />
Chemistry Department, University of Washington,<br />
Seattle WA USA 98195<br />
INTRODUCTION<br />
Quadrupolar coupling constants of<br />
170 kHz result from reduced motional<br />
averaging in biological polymers. The<br />
decay rates of Zeeman and quadrupolar<br />
order are direct measures of the spectral<br />
densities of motion. Information on the<br />
orientation-dependence of relaxation rates<br />
of Zeeman and quadrupolar order, Tlz and<br />
Tlq respectively, assists in understanding<br />
the dynamics of molecules. 1 Sensitivity and<br />
dynamic range considerations mandate<br />
optimum efficiency and uniformity over<br />
large spectral widths. The design of<br />
composite pulses to create broadband<br />
excitation and inversion is well established<br />
for deuterium NMR. 2 " 11 The numerical<br />
optimization of existing composite inversion<br />
pulse sequences was performed in order to<br />
increase the uniformity of excitation over<br />
spectral widths of 250 kHz. 2 " 7<br />
Pulse sequences to create<br />
quadrupolar order over moderately broad<br />
spectral widths have recently been<br />
proposed. 12 " 14 The pulse sequence design<br />
f Present Address: University of New Mexico,<br />
Albuquerque, NM<br />
of Wimperis is a good starting place from<br />
which to numerically optimize for the<br />
creation of quadrupolar order. 13 Broadband<br />
quadrupolar order is transferred to<br />
detectable magnetization with a 45 degree<br />
pulse and an additional refocusing pulse<br />
eliminates large phase corrections. 15<br />
Numerical optimization of the conversion<br />
from quadrupolar order to well-behaved<br />
transverse magnetization is worthy of<br />
investigation.<br />
A program developed in our<br />
laboratory for the numerical optimization of<br />
pulse sequences has been modified for 1=1<br />
spin systems. 15 " 18 This research is part of an<br />
ongoing investigation into the local and<br />
global dynamics of oligonucleotides utilizing<br />
solid state deuterium NMR. At the present<br />
time, the dynamics of the sugar rings of<br />
DNA are under study. The lower levels of<br />
hydration of DNA exhibit rigid-lattice<br />
lineshapes. Information on orientation<br />
dependence and the substantiation of<br />
motional models requires increased<br />
sensitivity and large spectral windows.
256<br />
METHODS<br />
Solid state deuterium NMR spectra<br />
were obtained at 76.72 MHz on a homebuilt<br />
spectrometer controlled by a DEC<br />
micro VAX II. Both the inversion<br />
sequences and the broadband Jeener-<br />
Broekaert experiments were phase cycled to<br />
eliminate double quantum coherence. 19 ' 20<br />
Phase shifting was accomplished with a<br />
homebuilt digital phase shifter. The sample,<br />
a labeled nucleoside, 2"-deutero-2'deoxyguanosine,<br />
was prepared by Jerome<br />
Shiels at the University of Washington.<br />
Either 2K or 4K scans were taken with a<br />
recycle delay of 5.0 seconds. The field<br />
strength was 100 kHz and the dwell time<br />
was 200 nanoseconds. Each spectrum was<br />
acquired with 4096 points.<br />
Calculations were accomplished on a<br />
DEC UNIX 3100 workstation. The strategy<br />
thus far has been to parameterize the pulse<br />
sequence and generate random trial pulse<br />
sequences. The basis set proposed by Vega<br />
and Luz was used. The coherences of<br />
interest were single basis elements rather<br />
than linear combinations of basis<br />
elements. 19 The expectation value of -Iz<br />
Qz quadrupolar order, or Iy was compared<br />
with the target function over a specific<br />
spectral width. These quality factors<br />
quantify the performance of the pulse<br />
sequence as a function of the quadrupolar<br />
frequency. Excitation profiles of the<br />
expectation value as a function of reduced<br />
frequency illustrate overall smoothness and<br />
breadth. Evolution profiles of the elements<br />
of the density operator as a function of time<br />
demonstrate the effects of rf pulses and<br />
evolution under a strong quadrupolar<br />
Hamiltonian. Additional three-dimensional<br />
graphics developed in our laboratory assist<br />
in visualizing the transfer of coherences. 22 "<br />
RESULTS<br />
Bulletin of Magnetic Resonance<br />
Measurement of spin-lattice<br />
relaxation times and investigation of the<br />
orientation-dependence of Tt in solids are<br />
important tools for understanding<br />
dynamics. 1 Inversion pulse lengths of more<br />
than 4 microseconds compromise the<br />
inversion breadth and uniformity across<br />
wide line deuterium powder patterns- 7 A<br />
variety of composite pulse schemes have<br />
been proposed for spectra with widths<br />
approaching the Rabi frequency of the<br />
radiofrequency pulse. The composite<br />
excitation triplet designed by Levitt, Suter<br />
and Ernst and supercycled in the method<br />
proposed by Levitt in order to create<br />
broadband inversion of deuterium<br />
lineshapes was numerically optimized. •<br />
The sequence consists of the composite<br />
excitation pulse, 45o9O18O135o, supercycled<br />
in the triplet form, O0 O9Q 3>O Only the<br />
flip angles of the excitation triplet were<br />
parametrized. The sequence has already<br />
been shown to invert rigid-lattice deuterium<br />
spectra with an rf field of 139 kHz while a<br />
weaker field led to a small loss in<br />
performance. 9<br />
Higher order pulses would lead to<br />
longer total pulse lengths. The total<br />
duration of the composite pulse should be<br />
short to avoid irreversible loss of<br />
magnetization during the excitation. 9<br />
Experimental spectra acquired with the<br />
optimized composite excitation pulse of<br />
43010018Q142Q, supercycled as above (B)<br />
and the Levitt triplet (A) are displayed in<br />
figure 1.<br />
Recent application of Tycko's use of<br />
the Magnus expansion for the design of<br />
composite pulses has led to the<br />
development of new excitation schemes. 8 ' 11
Vol. 14, No. 1-4 257<br />
B<br />
P<br />
460 n6<br />
kHz<br />
-46a<br />
fig-1<br />
The modified Jeener-Broekaert pulse<br />
sequence was founded on the model of a<br />
composite excitation pulse broadband with<br />
respect to rf field strength 13 ' 14<br />
The multipulse sequence eliminated<br />
the frequency selection of the original<br />
Jeener-Broekaert experiment. The<br />
broadband sequence has been utilized to<br />
measure the spectral densities of liquid<br />
crystals. 21 Increased signal sensitivity<br />
becomes highly desired as one goes to<br />
broader linewidths, lower Larmor<br />
frequencies and smaller biological samples<br />
with dilute nuclei.<br />
Computer search routines to create<br />
broadband quadrupolar order were<br />
performed with 10, 8 and 7 parameters.<br />
The 10 parameter search consisted of 4<br />
pulses, 3 phases, and 3 seperate delays. The<br />
8 parameter search had a single delay<br />
parameter. An assortment of the resulting<br />
sequences were tested experimentally. The<br />
"91" sequence was a 10 parameter search<br />
optimized over a spectral width of 300 kHz.<br />
It proved to perform well in breadth and<br />
sensitivity. Experimental spectra obtained<br />
with the Wimperis sequences A and B and<br />
the "91" pulse sequence, C, found in this<br />
study, are shown in figure 2.<br />
400 -400<br />
A: 900 -4.0p. -67.5270 -4.0n -4590 -2.0p. -4590<br />
B: 900 -4.0(i -75270- 4.0M. -52.590 -2.0M -4590<br />
C: 1130 -5.5n -72 284 -3.2n -52142 -3.9M -59112<br />
CONCLUSION<br />
fig-2<br />
The use of solid state deuterium<br />
NMR as a probe of dynamics in DNA offers<br />
the luxury of selective labeling. The same<br />
luxury introduces some of the limitations of<br />
site-selective wide line spectroscopy. The<br />
technique enables one to focus on the<br />
dynamics of single positions within large<br />
molecules. The small, precious samples are<br />
dilute in the observe nuclei yet one can<br />
monitor the onset of motion in both a local<br />
and global fashion as water is added to the<br />
spaces in DNA. The technique requires high<br />
power, short pulses for broad lineshapes,<br />
resistant samples, and extensive phase<br />
cycling and signal averaging. The dry DNA<br />
has longitudinal relaxation times of several<br />
seconds making signal intensity even more<br />
elusive. For these reasons, this investigation<br />
aims to numerically optimize existing pulse<br />
sequences for the creation of selected<br />
coherences over static deuterium powder<br />
linewidths of 250 kHz. Analysis of the
258<br />
evolution of the system provides clues for<br />
producing the very best tailored excitation.<br />
REFERENCES<br />
1 R.R. Void and R.L. Void, "Advances in Magnetic<br />
and Optical Resonance". Vol. 16, Academic Press,<br />
San Diego (1991)<br />
2 R. Tycko. Phys. Rev. Lett. 51. 775, (1983)<br />
-* M.H. Levitt. D. Suter. and R.R. Ernst, J. Chem.<br />
Phys.,S0, 3064 (1984)<br />
4 R. Tycko, E. Schneider. A. Pines. J. Chem. Phys.,<br />
81, 680 (1984)<br />
5 R. Tycko, H.M. Cho, E. Schneider, A. Pines, J.<br />
Mag. Res., 61, 90 (1985)<br />
6 M.H. Levitt. Prog, in NMR Spec. 18, 61 (1986)<br />
7 D J Simonivitch. DP Raleigh. E.T. Olejniczak.<br />
and R.G. Griffin../. Chem. Phys.,84, 2556 (1986)<br />
1 S. Wimperis and G. Bodenhausen. J. Mag. Res.,<br />
69.264(1986)<br />
9 N.J. Healon. R.R. Void, and R.L.Vold, J. Mag.<br />
Res..17. 572(1988)<br />
10<br />
DP. Raleigh. E.T. Olejniczak and R.G. Griffin,<br />
J. Mag. Res.,$l, 455 (1989)<br />
11 S. Wimperis../. Mag. Res..83.509 (1989)<br />
12 J. Jeener and P. Broekaert. Phys. Rev. , 157, 232<br />
(1967)<br />
13 S. WimperisJ. Mag. Res.. 86, 46 (1990)<br />
S. Wimperis and G. Bodenhausen. Chem. Phys.<br />
Lett. .132. 194(1986)<br />
15<br />
G. Hoatson.J. Mag. Res.. 94. 152-159 (1991)<br />
16<br />
S.J. Glaser and G.P. Drobny. "Advances in<br />
Magnetic Resonance" (W.S. Warren. ED) Vol 14.<br />
Academic Press. San Diego. 1990<br />
17 H. Liu. S.J. Glaser. G.P. Drobny. J. Chem. Phys..<br />
93.7543.(1990)<br />
18 B. Ewing. S.J. Glaser and G.P. Drobny, Chem.<br />
Phys. Lett. J47. 121 (1990)<br />
19 A.J Vega and Z. Luz. ./. Chem. Phys.. 86. 1803-<br />
1813(1987)<br />
Bulletin of Magnetic Resonance<br />
20 R.R. Void and G. Bodenhausen. J. Mag. Res..<br />
39, 363 (1980)<br />
21 G. Hoatson. T. Tse, R.L. Void. J. Mag. Res., 98.<br />
342 (1992)<br />
zl J. Callahan. D. Mattiello and Gary Drobny,<br />
<strong>ISMAR</strong> conference, Vancouver , B.C., July 1992,<br />
Poster 54
Vol. 14, No. 1-4 259<br />
1 Introduction<br />
A BASIC Program to Calculate<br />
the Evolution of Cartesian Product Operators<br />
Stefano Mammi<br />
Biopolymer Research Center, National Research Council<br />
Via Marzolo 1,35131 Padova, Italy<br />
The introduction of the product operator<br />
formalism [1] has greatly improved the<br />
description of multiple-pulse NMR<br />
experiments allowing the understanding of the<br />
fate of the magnetization in a direct manner.<br />
The use of this formalism has become<br />
widespread [2] because of its advantages ower<br />
both the complete density matrix approach<br />
and the method of vector diagrams. It is<br />
much simpler than the former and it permits a<br />
clear description of the results while<br />
remaining rigorous. With respect to the<br />
latter, it allows one to visualize all the states<br />
of the magnetization and to follow their<br />
evolution over very complicated pulse<br />
sequences.<br />
The rules that govern the evolution of<br />
product operators are very simple and lend<br />
themselves to automation by means of<br />
computer programs. Automation is especially<br />
desirable considering that the description of<br />
any two-dimensional experiment leads quickly<br />
to very long expressions which can be affected<br />
by trivial mistakes.<br />
Recently, computer programs that perform<br />
such calculations have been reported in the<br />
literature. Among these are the program by<br />
Nakashima and McClung [3] and the more<br />
recent one by Shriver [4]. The first was<br />
written in FORTRAN 77 and describes the<br />
evolution of product operators in the<br />
spherical basis [3]. While this approach is<br />
extremely useful in following coherence<br />
pathways and thus deriving phase cycling<br />
schemes, many times the use of the cartesian<br />
basis is preferable as in the development of<br />
new pulse sequences.<br />
The second program [4] was written in the<br />
new computer language Mathematica whose<br />
major advantage is the easy simplification of<br />
algebraic expressions. This program was<br />
written for Macintosh systems and is not yet<br />
available for IBM personal computers. The<br />
input seems to be stepwise and rather<br />
cumbersome.<br />
The program presented here runs within<br />
MS-DOS and describes the evolution of<br />
cartesian product operators. The program,<br />
named "EVOLVE", was written using the<br />
QuickBasic (C) 4.50 language. The input is<br />
from a file which contains all the information<br />
pertaining to the spin system and the pulse<br />
sequence, and the output is filed separately.<br />
2 Features of the Program<br />
A sample input file is presented in Fig. 1. In<br />
this example, an HMQC experiment {5] is<br />
applied to a system composed of an X<br />
nucleus and two protons, only one of which is<br />
coupled to the heteronucleus.<br />
The first information in the input file is the<br />
spin system which can be made of up to four<br />
spins, denoted by capital letters. The initial<br />
magnetization is entered next.
260 Bulletin of Magnetic Resonance<br />
mi Spin System<br />
HHX<br />
### Initial Operators CAxis(Sp#)Axis(Sp#) : coeffj<br />
### Terms in sine or cosine must nave exactly 6 characters in parenthesis<br />
z(2): *1<br />
### Coupling Constants (Spin 1,Spin 2,J)<br />
2,3.90<br />
### Sequence (1 Line for Name, N Lines for Pulse Sequence)<br />
HMQC - 2 spins + 1 spin<br />
P(90)H PH1<br />
D1<br />
P(90)X PH2<br />
DO<br />
P(180)H PH3<br />
DO<br />
P(90)X PH4<br />
D1<br />
A0 PH5 DEC(X)<br />
### Phase Cycles (as Brutcer, separated by commas)<br />
PH1,0<br />
PH2,0,1,2,3<br />
PH3.0<br />
PH4.0<br />
PH5,0,3,2,1<br />
### Delays: D#=Num/Oen*J(Sp#,Sp#) (#, Numerator, Denominator, Spin 1, Spin 2)<br />
1,1,2,2,3<br />
### Do you want to skip the Evolution under Chemical Shift?<br />
N<br />
### Print out only the Observable Operators?<br />
Y<br />
Figure 1. Sample input file for an HMQC sequence applied to a (H + HX) spin system.<br />
In listing the operators, the spins are<br />
numbered from one to four to prevent<br />
ambiguities among like spins. Any number of<br />
operators can be listed with appropriate<br />
coefficients as required. It is not necessary to<br />
start with equilibrium magnetization: for<br />
example, the magnetization of any spin can be<br />
neglected. Moreover, it is possible to follow<br />
the evolution of a specific operator, e.g.,<br />
2x(2)z(3), generated after the first Dl in the<br />
sequence of Fig 1, by restricting the initial<br />
input to just that operator with its own<br />
coefficients, e.g., +cos(Q 2*D1), utilizing an<br />
appropriately shortened pulse sequence.<br />
Next, the scalar coupling network is described<br />
in terms of the spins which are coupled and<br />
the relevant coupling constant.<br />
The format for the pulse sequence is very<br />
similar to the Bruker one. This allows for<br />
simple transcription of sequences already in<br />
use and for easy modification of any part of<br />
the phase cycling scheme. Each pulse is<br />
written as a "P" followed by the flip angle in<br />
parenthesis, by the spin(s) to which it is<br />
applied and by the phase cycle to be used.<br />
Only 90° and 180° pulses are currently<br />
accepted. Each delay is written as a "D"<br />
followed by a single digit. The acquisition is<br />
referred to as "AQ" followed by its own phase<br />
cycle. Up to ten different phase cycles can be<br />
utilized each containing up to 128 steps. Each<br />
step is recorded as a number according to<br />
the usual notation: 0= +x; 1 = +y; 2 =<br />
-x; 3 s -y.<br />
Many sequences require decoupling during<br />
specific delays, including the acquisition.
Vol. 14, No. 1-4 261<br />
z(2):<br />
-- 90 (H)+x --><br />
y(2): -1<br />
- D1 --><br />
2x(2)z
262 Bulletin of Magnetic Resonance<br />
EVOLVE will "decouple" any spin if an<br />
appropriate statement is added on the same<br />
line of any delay.<br />
EVOLVE was written specifically for<br />
sequences in which some delays are inversely<br />
proportional to certain scalar coupling<br />
constants as in heteronuclear correlation<br />
experiments. Proper space is provided in the<br />
input file for defining such cases.<br />
The user is then asked to indicate if<br />
evolution under both coupling and chemical<br />
shift should be taken into account or if the<br />
latter should be neglected. Finally, the user<br />
chooses whether all the resulting operators or<br />
just the observable ones should be printed out<br />
after the acquisition step.<br />
The program runs through the pulse<br />
sequence as many times as required by the<br />
phase cycle. If only the observables are<br />
chosen as output, the final intensities of the<br />
signal are written in two separate files, one for<br />
the real part and one for the imaginary part.<br />
At the end of the phase cycle, these files are<br />
read and final simplifications are carried out.<br />
In Fig. 2, a portion of the output obtained<br />
with the input file of Fig. 1 is reported, Le., the<br />
result from the first of the four steps of the<br />
phase cycle and the signal obtained at the end<br />
of the four step cycle. It can be seen that the<br />
term y(2) generated by the first 90° pulse gives<br />
rise only to antiphase terms at the end of the<br />
first Dl, because Dl = 1/2J23.<br />
The signal from the proton not coupled to<br />
the X-nucleus is present in the first<br />
acquisition, but is canceled out at the end of<br />
the four step cycle while the single quantum<br />
terms from the proton coupled to the Xnucleus<br />
add up to generate the final signal.<br />
The BASIC language does not have built-in<br />
routines for the simplification of algebraic<br />
expressions. A sizable portion of the program<br />
is devoted to such routines. Beside the more<br />
trivial expressions containing n/2, EVOLVE<br />
is able to handle terms containing sin(^/4) or<br />
cos(n/4), most commonly encountered in<br />
heteronuclear sequences. The program does<br />
not evaluate these expressions; rather, it<br />
simplifies them according to the rules it<br />
knows. This entails a longer computation<br />
time but with the advantage of eliminating all<br />
numerical coefficients. This compromise was<br />
found satisfactory.<br />
The only terms that would be useful to have<br />
simplified and that the program is unable to<br />
handle at this stage are those including sin(20)<br />
or cos(20) terms, encountered for example<br />
when there is a 180° pulse in the middle of a<br />
delay. This is not a serious limitation of the<br />
BASIC language, especially considering that<br />
even in Mathematica this simplification must<br />
be explicitly requested by the user.<br />
3 Conclusions<br />
EVOLVE has been applied to numerous<br />
complicated pulse sequences avoiding lengthy<br />
and monotonous calculations and providing<br />
the results in a way suitable for<br />
straightforward analysis.<br />
A copy of the program, including the source<br />
code, can be obtained by sending a 5.25" or<br />
3.5" diskette and return postage to the author.<br />
4 References<br />
[1] O. W. Stfrensen, G. W. Eich, M. H.<br />
Levitt, G. Bodenhausen, and R. R. Ernst,<br />
Prog. NMR Spectrosc 16,163 (1983).<br />
[2] See for example H. Kessler, M. Gehrke,<br />
and C. Griesinger, Angew. Chem. Int. Ed.<br />
Engl 27,490 (1988).<br />
[3] T. T. Nakashima and R. E. D. McClung, /.<br />
Magn. Reson. 70,187 (1986).<br />
[4] J. W. Shriver, /. Magn. Reson. 94, 612<br />
(1991).<br />
[5] A. Bax, R. H. Griffey, and B. L. Hawkins<br />
/. Magn. Reson. 55,301 (1983).
Vol. 14, No. 1-4 263<br />
SELECTIVE LONG-RANGE POLARI<br />
ZATION TRANSFER via DEPT.<br />
Introduction<br />
T. Parella, F. Sanchez-Ferrando* and A. Virgili.<br />
Departament de Quimica. Universitat Autonoma de Barcelona,<br />
08193 Bellaterra, Barcelona, Spain.<br />
The structural assignment of organic compounds<br />
containing quaternary carbons and/or heteroatoms is<br />
often greatly helped by measurements of long-range<br />
proton-carbon coupling constants, "J^, with particular<br />
emphasis on two-bond and three-bond couplings which<br />
usually show values in the range 3-10 Hz [1,2], depending<br />
on carbon hybridization, torsion angles, substttuent<br />
electronegativity and orientation, etc.<br />
Selective ID NMR methods, such as the selective<br />
INEPT method proposed by Bax [3], have long been<br />
used to reveal connectivities with a given proton.<br />
Thus, application of a selective INEPT (or, more<br />
frequently, refocussed INEPT) pulse sequence on a<br />
well resolved proton, with delays optimized for a longrange<br />
heteronuclear coupling (usually around 5-7<br />
Hz), results in a ID carbon spectrum displaying large<br />
intensity enhancements at the carbons coupled (at<br />
long range) with the perturbed proton. Furthermore,<br />
the intensity of these carbons shows a maximum when<br />
the true "iCH value is used for the delay optimization<br />
of the selective INEPT sequence. We have recently<br />
shown the use of this dependence to obtain a quick<br />
estimate of "J^, in several polycyclic derivatives [4].<br />
Selective 2D methods, however, are to be<br />
preferred because they can easily yield accurate values<br />
for the desired long-range coupling constants, provided<br />
the perturbed proton appears as a well resolved multiplet<br />
in the proton spectrum. Thus, the selective spin flip<br />
method first proposed by Bax and Freeman [5] has<br />
been widely used for the determination of "J^ values<br />
from a given proton.<br />
In recent years a number of methods have been<br />
suggested which combine the intensity enhancements<br />
obtained from selective polarization transfer with the<br />
easy measurement of "J^ characteristic of the selective<br />
spin flip method. Thus, Jippo et al. proposed a 2D<br />
selective INEPT method [6] which consists in a<br />
conventional 2D INEPT pulse train containing a<br />
selective spin echo sequence. We have modified this<br />
method by delivering all decoupler pulses in the<br />
selective mode [4], and in this way we have suppressed<br />
the artefact peaks otherwise observed with this method.<br />
Another combination of polarization transfer<br />
and selective spin flip was proposed by Uhrin et at. [7].<br />
In this method, a standard non-selective DEPT<br />
preparation period (optimized for I JCH) is followed by<br />
a conventional selective (or semiselective) spin flip<br />
sequence, yielding a J-resolved 2D spectrum displaying<br />
the desired long range couplings. More recently,<br />
Poppe and van Halbeek [8] have introduced inverse<br />
detection, either in ID or 2D methods, by combining<br />
selective polarization transfer with long range<br />
heteronuclear coupling measurements in lH-detected<br />
sequences.<br />
We now report two new u C-detected, 2D Jresolved<br />
sequences, based on a DEPT pulse train,<br />
which allow fast and accurate measurements of longrange<br />
heteronuclear coupling constants from well<br />
resolved protons. Both sequences are particularly<br />
useful for the determination of couplings to quaternary<br />
carbons, and their main feature is that all proton<br />
pulses are delivered as soft, selective pulses (20-30<br />
ms).
264 .<br />
Parametrization<br />
Starting from the 1D-SDEPT sequence (Scheme<br />
1), we have derived the new sequences 2D-SDEPT1<br />
(Scheme 2) and 2D-SDEPT2 (Scheme 3), in which<br />
SDEPT stands for Selective DEPT. Since ail proton<br />
pulsesareselective (yBjln=20-30 Hz), both sequences<br />
achieve a selective polarization transfer from the<br />
pulsed proton to long-range coupled carbons, provided<br />
that the fixed interpulse delay A is optimized for longrange<br />
couplings. Both sequences yield comparable<br />
results in similar experiment times.<br />
scheme 1<br />
scheme 2<br />
90° 180<br />
scheme 3<br />
180° 4>° 180°<br />
i i i i<br />
i A i A < Evolution •<br />
90" 180°<br />
180° 90° 180°<br />
In both sequences the incrementable evolution<br />
period t,, which brackets a selective spin echo moiety,<br />
allows only the evolution of heteronuclear long range<br />
couplings which will therefore be detected in Fl<br />
dimension in the final 2D J-resolved spectrum.<br />
Bulletin of Magnetic Resonance<br />
We have studied the dependence of carbon<br />
signal intensities on several experimental parameters,<br />
such as the delay A or the pulse angle of the last<br />
selective proton pulse in the DEPT part of the sequence,<br />
using camphor (Fig. 1) as a readily available, rigid<br />
model compound. These determinations have been<br />
most conveniently carried out using the ID selective<br />
DEPT sequence 1D-SDEPT.<br />
Fig.l<br />
The extent of polarization transfer is a function<br />
of both, the interpulse delay A and the pulse angle $.<br />
Fig. 2 shows the intensity of the quaternary C-l carbon<br />
signal of camphor when pulsing the H-4 methine<br />
proton, for 4>=45° and for =90°, as a function of<br />
interpulse delay A, using sequence 1D-SDEPT. In this<br />
case, the heteronuclear coupling constant involved is<br />
3 JH4^,= 4.4 Hz. As expected, the quaternary carbon C-<br />
1 intensity follows a typical sinus function, and maximum<br />
signal is obtained for (j>=90° and A=50-90 ms, a range<br />
which does not contain the theoretical value, (2* 3 JH4^<br />
C1)'—115 ms. Instead, this range of maximum signal is<br />
centered around the practical value (3* 3 JH4^1) l =75<br />
ms.<br />
1000-<br />
800-<br />
~ 600<br />
f<br />
£ 400<br />
200<br />
0<br />
0.02<br />
-i 1 « 1 1 1—r<br />
0.04 0.06 0.08 0.1<br />
Fig. 2<br />
Delay (s)
Vol. 14, No. 1-4<br />
However, when pulsing on a methyl proton the<br />
optimum delay has to be modified. Fig. 3 shows the<br />
variation of the quaternary O7 carbon signal of camphor<br />
when pulsing the methyl H-10 protons as a function of<br />
the pulse angle, for three A values, using also sequence<br />
1D-SDEPT. In this case, the heteronuclear coupling<br />
constant involved is 2 JH1(>C7=4.1 Hz (sign not determined).<br />
If the optimization is carried out as in a conventional<br />
DEPT, using A=(2J)S the intensities follow the typical<br />
sin*cos 2
266<br />
The results (Fig. 7) easily allow the assignment<br />
of the four quaternary carbons coupled to the H-5<br />
proton. The particular example shown was obtained<br />
using sequence 2D-SDEPT1, but the alternative sequence<br />
2D-SDEPT2 gave the same results. We also performed<br />
a comparison between 2D-SDEPT1 and our previous<br />
[4] sequence 2D-SINEPT. Both methods can yield the<br />
desired long-range heteronuclear coupling constants<br />
in less than one hour of accumulation (using a 400<br />
M Hz spectrometer), with excellent sensitivity even for<br />
quaternary carbons. However, the SDEPT method is<br />
to be preferred, because it is less sensitive to mismatch<br />
between the delay A and the long range coupling "J^.<br />
b)<br />
d)<br />
6'7'<br />
5'6 4' 87a' 7 4a 3a' 8a 5<br />
6 8 7<br />
'.58 128 1!!<br />
Fig. 7<br />
a) Decoupled "C spectrum; b) 1D-SDEPT spectrum after pulsing on<br />
H-5. Only long-range coupled carbons with it are presents. Then, The<br />
2D- J spectra (d) and its internal projection (c) obtained with the pulse<br />
sequence of the scheme 2.<br />
Acknowledgements.<br />
Financial support from DGICYT through project<br />
number PB89-0304 is gratefully acknowledged. We<br />
also thank the Servei de Ressonancia Magnetica Nuclear,<br />
UAB, for allocating instrument time to this project. A<br />
grant (to T.P.) from Universitat Autonoma de Barcelona<br />
is gratefully acknowledged.<br />
References<br />
Bulletin of Magnetic Resonance<br />
1.- P.E. Hansen, Progress in NMR Spectroscopy, 1980,<br />
14, 175.<br />
2.- J.L. Marshall, "Carbon-Carbon and Carbon-Proton<br />
NMR Couplings. Application to Organic<br />
Stereochemistry and Conformational Analysis", VCH<br />
Publishers, Deerfield Beach, Florida, 1982.<br />
3.- A. Bax, J. Magn. Reson., 1984, 57,314.<br />
4.- T. Parella, F. Sanchez-Ferrando and A. Virgili,<br />
Magn. Reson. Chem., 1992, in press.<br />
5.- A. Bax and R. Freeman, J. Am. Chem. Soc, 1982,<br />
104, 1099.<br />
6.- T. Jippo, O. Kamo and K. Nagayama, J. Magn.<br />
Reson., 1986,66, 344.<br />
7.- D. Uhrin, T. Liptaj, M. Hricovini and P. Capek, J.<br />
Magn. Reson., 1989,85,137.<br />
8.- L. Poppe and H. van Halbeek, J. Magn. Reson.,<br />
1991,92,636.
Vol. 14, No. 1-4 267<br />
Computer Simulations of High Resolution NMR Spectra<br />
1 Introduction<br />
Scott A. Smith, William E. Palke, and J. T. Gerig<br />
Department of Chemistry, University of California<br />
The evolution of a spin system during the application<br />
of an RF pulse is a central aspect of high<br />
resolution NMR spectroscopy. An appropriately<br />
applied field can lead to saturation or decoupling<br />
effects. When a field is used to maintain a spinlocked<br />
condition, both scalar and dipolar interactions<br />
come into play, leading to coherence transfers<br />
and nuclear Overhauser effects that form the<br />
basis of TOCSY, ROESY and related experiments.<br />
These experiments have important roles<br />
in structural studies of biological macromolecules<br />
in solution . The accompanying bad<br />
news is that the interpretation of these experiments<br />
is tricky; computer simulations can be<br />
very helpful in assisting the analysis of these<br />
spectra.<br />
The theoretical formalism of relaxation in<br />
the presence of an RF field can be written elegantly<br />
in superoperator notation , or in a more<br />
conventional notation laden with superscripts<br />
and subscripts. The latter approach displays<br />
more details and shows how the presence of the<br />
RF field introduces new combinations of frequencies<br />
into the relaxation expressions. Of<br />
course, it can be shown that both methods generate<br />
identical results. Details of the theoretical<br />
developments formulated in this lab are given in<br />
reference 7 and in a manuscript in preparation.<br />
We have used our theoretical results to<br />
extend the program GAMMA 8 to include the<br />
effects of relaxation in the presence of an RF<br />
field and present here the results of several calculations<br />
of double resonance or rotating frame<br />
experiments that illustrate the capabilities<br />
Santa Barbara, CA 93106, USA<br />
thereby developed. For the examples presented,<br />
only dipole-dipole relaxation was included, and<br />
isotropic diffusional tumbling was assumed with<br />
a correlation time of 1.0 nsec. In all cases, the<br />
spectrometer proton frequency was 500 MHz.<br />
2 Applications<br />
Presaturation - The first example is a presaturation<br />
experiment on a three spin system whose<br />
parameters were chosen to mimic three protons<br />
in the trans conformation of a glycine residue.<br />
Details are given in Table 1.<br />
Table 1: Spin System 1 Parameters<br />
R12 = 1.75A<br />
R13 = 3.00A<br />
R23 = 2.53A<br />
J12 = -16.3Hz<br />
J13=4.0Hz<br />
J23 = 12.0 Hz<br />
v = -700 Hz<br />
t> = -950Hz<br />
v = 950 Hz<br />
The pulse sequence consisted of a pre-irradiation<br />
period followed by an analyzing 90° pulse, as<br />
shown in Figure 1.<br />
Pre-saturation<br />
Figure 1. Pre-saturation Pulse Sequence<br />
The initial pulse length was chosen to be long<br />
enough to establish a steady state, and the RF frequency<br />
chosen to match the chemical shift of<br />
proton 2. The resulting spectra for several RF<br />
field strengths are shown in Figure 2. Notice the
268 Bulletin of Magnetic Resonance<br />
750<br />
960 950 940<br />
V3<br />
-690-700-710 -940-950-960<br />
V2<br />
Figure 2. Simulation of Presaturation. Shading is<br />
used to highlight NOE enhanced intensities.<br />
Units on both axes are Hz.<br />
non-symmetrical effects on spins 1 and 3.<br />
Decoupling - The second example illustrates the<br />
decoupling of a two spin proton system with a<br />
chemical shift difference of 1000 Hz and a spin<br />
coupling constant of 10 Hz. The protons were<br />
2A apart. After an initial 90° pulse, an RF field<br />
of a specified magnitude was applied during the<br />
collection of the FID. The pulse sequence is<br />
shown in Figure 3. Some spectra are shown in<br />
Figure 4. Notice the slight oscillation in the peak<br />
height as a function of the strength of the decoupling<br />
field.<br />
Figure 3. Decoupling Pulse Sequence<br />
ROESY/TOCSY Simulations - Two-dimensional<br />
rotating frame experiments may produce coherence<br />
transfer (TOCSY) and spin-spin cross<br />
relaxation (ROESY) effects simultaneously, and<br />
1125 1100<br />
v (Hertz)<br />
1075<br />
(Hertz)<br />
Figure 4. Simulation of Decoupling. Spectra at the<br />
first and last field strengths are shown to<br />
the upper left.<br />
ambiguities of interpretation arise when bothkinds<br />
of effects are present. Griesinger et al.<br />
have described a strategy for suppressing<br />
ROESY effects when doing TOCSY experiments<br />
9 . The interpretation of ROESY experiments<br />
is complicated because coherence transfer<br />
(COSY) and indirect interactions (HOHAHA)<br />
can lead to spectral effects that interfere with or<br />
imitate rotating frame Overhauser features 10 .<br />
Methods for suppressing these unwanted effects<br />
have been proposed , and it has been suggested<br />
that use of a train of short RF pulses to generate<br />
an effective spin-locking field leads to diminution<br />
of COSY-type cross peaks .<br />
Here we simulate several variations of<br />
ROESY experiments. First, consider the pulse<br />
sequence shown in Figure 5 in which a continuous<br />
spin-locking pulse is applied.<br />
JC/2 Spin Lock<br />
Figure 5. CW ROESY Pulse Sequence<br />
We apply this to an equilateral triangle of protons<br />
as described in Table 2. The 2D spectra that
Vol. 14, No. 1-4 269<br />
1<br />
1<br />
C-l. 0 sec<br />
M<br />
• 1 11<br />
150 100 50 0 -50 -100 -150 150 100 50 0 -50 -100 -150<br />
150 100 0 -50 -100 -150<br />
Figure 6. 2D ROESY spectra for differing spin-lock<br />
times.<br />
result from several different spin-lock times are<br />
shown in Figure 6. A point to notice here is that<br />
the ROESY peaks grow more slowly than the<br />
Table 2: Spin System 2 Parameters<br />
Rl2=V3A<br />
Ri3=V3A<br />
R23=V3A<br />
J12 = 0<br />
Jl3 = 0<br />
J23 = 4.0 Hz<br />
v = 140 Hz<br />
v = -90 Hz<br />
?? = -140Hz<br />
D -1.1 sec<br />
150 100 50 0 -50 -100 -150<br />
©<br />
.©<br />
-o<br />
o<br />
• o<br />
©<br />
1 V)<br />
-o<br />
COSY peaks, reaching their maximum intensity<br />
well after the latter have begun to subside.The-<br />
ROESY peaks also persist much longer and<br />
decay more smoothly than both the COSY and<br />
diagonal peaks for protons 2 and 3. The plots for<br />
1.0 and 1.1 sec. spin-lock times show oscillation<br />
in both the latter features while the ROESY<br />
peaks decay smoothly.<br />
Interesting contrasts to these results arise in<br />
changing to an isosceles triangle configuration.<br />
The spin system is given in Table 3. In this<br />
geometry, the ratio of the 1-3 distance to the 1-2<br />
©
270 Bulletin of Magnetic Resonance<br />
distance is \2. Thus, the direct 1-2 dipole-dipole<br />
interaction is 8 times stronger than the corresponding<br />
1-3 interaction. Keeping a fixed<br />
Table 3: Spin System 3 Parameters<br />
Ri2=V3A<br />
Ri3=V6A<br />
R23=V3A<br />
J12 = 0<br />
J13 = 0<br />
J23 = 0 -10 Hz<br />
v = 140 Hz<br />
v = -90 Hz<br />
v = -U0Uz<br />
geometry and a spin-lock time of 0.1 sec, the 2-<br />
3 spin-spin coupling is varied from 0 to 10 Hz as<br />
we repeat the pulse sequence of Figure 5. Slices<br />
through the resulting 2D spectra at 140 Hz are<br />
shown in Figure 7.<br />
J = 10 Hz<br />
J = 8Hz<br />
= 6Hz<br />
J = 4Hz<br />
J = 2Hz<br />
= 0Hz<br />
J)LJL_<br />
150 100 50 0 -50 -100 -150<br />
Figure 7. The effect of J23 variation on ROESY/<br />
TOCSY experiment.<br />
Indeed, the simulation with J23 = 0 shows an 8:1<br />
ratio of NOE peak intensities. However, variation<br />
of J23 generates a ROESY-like peak in the<br />
location of the 1-3 interaction that becomes even<br />
larger than the true 1-2 ROESY peak for some<br />
values of J23- This situation provides an excellent<br />
example of a spectrum that cannot directly<br />
yield reliable information about internuclear distances.<br />
In the next example, the frequency of the<br />
spin-locking RF field is varied. In this simulation<br />
spin system of Table 3 is used with J23 = 4 Hz. It<br />
is subjected to a 1.0 sec. spin-locking pulse as<br />
depicted in the ROESY pulse sequence in Figure<br />
5. The frequency of the RF field is indicated by<br />
the arrow on each of the stacked plots in Figure<br />
8. While the 1-2 ROESY peak is relatively unaffected<br />
by the frequency of the RF field, the 1-3<br />
peaks which are generated by indirect<br />
HOHAHA effects show extreme sensitivity both<br />
in magnitude and sign. As has been suggested in<br />
the literature, several experiments should be run<br />
with different RF frequencies to discriminate<br />
between these interactions.<br />
\<br />
1<br />
150 100 50 0 -50 -100 -150<br />
Figure 8. The effect of varying the applied RF frequency<br />
on ROESY/TOCSY experiment.<br />
i<br />
1 1<br />
As a final demonstration, we investigate the<br />
suggestion that dividing the spin-locking period<br />
into a sequence of numerous pulse-delay steps<br />
can emphasize the ROESY interactions. Again,<br />
the three spin proton system in Table 3 is used<br />
with J23 = 4 Hz. It was subjected to the ROESY<br />
pulse sequence utilizing a pulse train as shown<br />
in Figure 9. The length of the pulse and delay<br />
steps were varied keeping the average yBj con-<br />
I
Vol. 14, No. 1-4 271<br />
Figure 9. ROESY Pulse Sequence with pulse train<br />
150 100 50 0 -50 -100 -150<br />
Figure 10. Simulation of ROESY/TOCSY experiment<br />
using a spin-locking pulse train.<br />
o<br />
.©<br />
o<br />
-o<br />
150 100 50 0 -50 -100 -150<br />
Figure 11. Simulation of ROESY/TOCSY using continuous<br />
irradiation for spin-locking.<br />
o<br />
•o<br />
stant at 2000 Hz by adjusting the number of<br />
pulses per second. This simulation has been performed<br />
with pulse lengths that correspond to<br />
individual rotations of 180°, 30°, and 10°, and<br />
also with CW irradiation. The contour plot using<br />
30° pulses is presented in Figure 10. For comparison,<br />
the plot which was simulated using continuous<br />
irradiation during the spin lock is shown<br />
in Figure 11. Cross sections of these contour<br />
plots taken at -90 Hz are shown in Figure 12.<br />
= 10deg<br />
= 180deg<br />
150 100 50 0 -50 -100 -150<br />
Figure 12. Cross sections of ROESY/TOCSY simulations<br />
using a spin-locking pulse train.<br />
Shortening the pulse length indeed suppress the<br />
COSY peaks, but they remain even for the shortest<br />
pulses. Also, the limit of the sequence of<br />
shorter and more frequent pulses gives the same<br />
result as continuous irradiation. These two processes<br />
appear to become the same once the delay<br />
between pulses is sufficiently short that a negli-
272 Bulletin of Magnetic Resonance<br />
gible precession occurs before the next pulse.<br />
3 Conclusions<br />
It is clear that full computer simulation may be<br />
essential for the correct interpretation of ROESY<br />
experiments. Our initial results indicate that<br />
there may be additional information about structure<br />
and dynamics that can be extracted from<br />
classical one-dimensional multiple irradiation<br />
experiments with the aid of simulations. An<br />
appreciable expansion of this program in terms<br />
of the size of spin systems that can be handled is<br />
probably needed to make such applications practical;<br />
this along with other expansions of the program's<br />
capabilities is planned.<br />
4 References<br />
1 - L. Braunschweiler and R. R. Ernst, J. Magn.<br />
Reson. 53, 521 (1983).<br />
2 - A. Bax and D. G. Davis, /. Magn. Reson.<br />
63,207 (1985).<br />
3 - A. A. Bothner-By and R. Shukla, /. Magn.<br />
Reson. 77, 524 (1988).<br />
4 - M. Ranee and J. Cavanaugh, J. Magn.<br />
Reson. 87, 363 (1990).<br />
5 - S. J. Glaser and G. P. Drobny, Advan. Magn.<br />
Reson. 14, 35 (1990).<br />
6 - R. R. Ernst, G. Bodenhausen, and A. Wokaun,<br />
"Principles of Nuclear Magnetic Resonance<br />
in One and Two Dimensions",<br />
Clarendon Press, Oxford (1987); J. Jeneer,<br />
Advan. Magn. Reson. 10, 2 (1982); M.<br />
Ravikumar, R. Shukla, and A. A. Bothner-<br />
By, J. Chem, Phys. 95, 3092 (1991).<br />
7 - S. A. Smith, W. E. Palke, and J. T. Gerig, /.<br />
Magn. Reson. in press (1992).<br />
8- S.A. Smith, T. Levante, B.H. Meier, and<br />
R.R. Ernst, manuscript in preparation.<br />
9 - C. Griesinger, G. Otting, K. Wiithrich, and<br />
R.R. Ernst, J. Am. Chem Soc. 110, 7870<br />
(1988).<br />
10-D. Neuhaus and M. P. Williamson, "The<br />
Nuclear Overhauser Effect in Structural and<br />
Conformational Analysis". VCH, New York<br />
(1989) p. 312-327.<br />
11 - H. Kessler, C. Griesinger, R. Kessebaum, K.<br />
Wagner, and R.R. Ernst, J. Am. Chem Soc.<br />
109,607-609(1987).
Vol. 14, No. 1-4 273<br />
1. Introduction<br />
Variation of 13 C NMR Linewidths of<br />
Metallocenes as a Function of Magic<br />
Angle Sample Spinning Frequency<br />
In this paper we report high-resolution solid state 13 C<br />
NMR investigations of various metallocenes<br />
[(Ti5-C5H5)M(Ti5-C5H5); M = Fe, Ru, Ni], carried out as<br />
a function of magic angle sample spinning (MAS)<br />
frequency. Specifically, we focus on how the linewidth of<br />
the isotropic peak varies with MAS frequency at fixed<br />
temperature. Unexpectedly, it has been found that the<br />
linewidth increases as the MAS frequency is increased, and<br />
it is demonstrated that the underlying reason is that the *H<br />
decoupling becomes less efficient as the MAS frequency is<br />
increased. It is suggested that molecular motion within<br />
these solids is an important factor underlying this<br />
phenomenon.<br />
It is well known that, at room temperature, there is<br />
substantial molecular motion in crystalline metallocenes<br />
[1-3]. In ferrocene, for example, it has been shown [4-6]<br />
that there is rapid reorientation of the cyclopentadienyl<br />
(C5H5) rings via a five-fold jump mechanism, with<br />
correlation time xc = 5 X 10~ 12 s at 293 K. The<br />
correlation time for ring reorientation in nickelocene at<br />
room temperature is essentially the same as that for<br />
ferrocene, whereas the ring reorientation in ruthenocene is<br />
associated with a significantly longer correlation time (tc ~<br />
5 X 10- 10 s at 293 K) [4,5]. It has been suggested [2]<br />
that there may be some additional slower molecular<br />
motions in crystalline ferrocene at ambient temperature.<br />
All experiments reported in this paper were carried out at<br />
constant temperature, and hence the correlation time for<br />
molecular motion for a given metallocene can be assumed<br />
to be constant for the series of experiments discussed here.<br />
Before presenting our results, we discuss briefly relevant<br />
aspects of magic angle sample spinning (MAS) and high<br />
power *H decoupling in relation to the measurement of<br />
high-resolution 13 C NMR spectra for organic solids, with<br />
Ian J. Shannon, Kenneth D.M. Harris*, s* S. Arumugam<br />
Department of Chemistry<br />
University of St. Andrews<br />
St. Andrews<br />
Fife KY16 9ST<br />
Scotland<br />
particular emphasis on the application of the technique to<br />
systems in which there is substantial molecular motion.<br />
2. Theory<br />
* Author to whom all correspondence should be addressed.<br />
In solid state 13 C NMR of organic materials, the two<br />
major sources of line-broadening are chemical shift<br />
anisotropy (CSA) and direct ^C- 1 !! dipole-dipole<br />
interaction. Magic angle sample spinning (MAS) [7,8]<br />
will transform an NMR line broadened by these effects into<br />
a set of comparatively narrow, equally-spaced lines<br />
comprising an isotropic peak and spinning sidebands. The<br />
spacing between adjacent lines in this set is equal to the<br />
MAS frequency vr. The chemical shift of the isotropic<br />
peak is independent of vr, and only this line remains when<br />
the anisotropic interactions have been averaged completely<br />
(i.e. at sufficiently large vr).<br />
As discussed fully elsewhere [7-9], there is an important<br />
difference concerning the way in which CSA and direct<br />
dipole-dipole interaction are affected by MAS. For an<br />
NMR line broadened by CSA, relatively slow MAS is<br />
generally sufficient to transform this line into the set of<br />
narrow, equally-spaced lines discussed above, whereas a<br />
line broadened by direct dipole-dipole interaction will be<br />
narrowed significantly only if vr is in the region of (or<br />
greater than) the magnitude of the dipole-dipole interaction.<br />
The underlying reason for this difference [9] is that CSA<br />
gives rise to inhomogeneous broadening of the spectral<br />
line, whereas dipole-dipole interaction is a source of<br />
homogeneous broadening; the value of vr required to<br />
achieve effective line-narrowing is larger (relative to the<br />
linewidth of the static sample) in the case of homogeneous<br />
broadening. For most organic solids with natural isotopic<br />
abundance, the only appreciable dipole-dipole interaction<br />
directly affecting the 13 C NMR spectrum is that between
274<br />
13 C and 1 H. Since the magnitude of this interaction<br />
(typically ca. 30 kHz for rigid organic solids) is generally<br />
much larger than the MAS frequencies that can be obtained<br />
on conventional instruments, substantial averaging of this<br />
interaction cannot be achieved using MAS. For this<br />
reason, high power *H decoupling is generally applied (in<br />
addition to MAS) during acquisition of the 13 C spectrum<br />
in order to eliminate line-broadening due to direct ^C-^H<br />
dipole-dipole interaction.<br />
In 13 C NMR spectroscopy of many crystalline<br />
metallocenes (such as ferrocene and ruthenocene), GSA and<br />
direct ^C- l U dipole-dipole interaction are the important<br />
sources of line-broadening.<br />
A detailed publication [10] has considered, from both<br />
theoretical and experimental standpoints, the linewidth of<br />
the isotropic peak, recorded under MAS conditions, for a<br />
system that is subject to line-broadening by CSA. It was<br />
also shown that, at fixed temperature, the linewidth of the<br />
isotropic peak decreases as the' MAS frequency (vr) is<br />
increased, approaching a limiting value at sufficiently large<br />
vr. Another paper [11] has considered the linewidth of a<br />
spin system S, dipolar coupled to an unlike spin system /,<br />
under conditions of isotropic molecular motion and<br />
decoupling of the / spins. Using coi to denote the<br />
decoupler field strength, it was shown that, in the limit of<br />
long correlation time (i.e. (0itc » 1) the linewidth is<br />
proportional to (o>i)~ 2 , whereas in the limit of short<br />
correlation time (i.e. coixc « 1) the linewidth is<br />
independent of ca i. Thus, for a sample at fixed<br />
temperature (and hence fixed xc), the linewidth should<br />
either decrease as the decoupler field strength is increased<br />
(long correlation limit) or remain independent of the<br />
decoupler field strength (short correlation limit). The<br />
effects of anisotropic molecular motion on the measured<br />
spectrum were also discussed briefly in ref. 11.<br />
In the studies discussed in this paper, both MAS<br />
frequency and *H decoupler field strength are important in<br />
controlling the linewidth of the isotropic peak in the 13 C<br />
NMR spectrum. If the separate effects discussed above can<br />
be combined in a simple way, then it should be expected<br />
that: (a) at fixed temperature and fixed decoupler field<br />
strength, the linewidth should decrease with increasing<br />
MAS frequency (up to a limiting value, beyond which the<br />
linewidth should be effectively independent of the MAS<br />
frequency); and (b) at fixed temperature and fixed MAS<br />
frequency, the linewidth should either decrease or remain<br />
constant as the decoupler field strength is increased,<br />
depending on the motional regime (i.e. long or short<br />
correlation limit) of the sample at the temperature of<br />
interest.<br />
It is shown here that, for ferrocene and ruthenocene at<br />
room temperature, the effects of MAS frequency and *H<br />
decoupler field strength on the 13 C NMR linewidth cannot<br />
be combined in this simple way, since the effective<br />
decoupler field strength is modulated by altering the MAS<br />
frequency. Nickelocene is paramagnetic, and for this<br />
Bulletin of Magnetic Resonance<br />
reason it is not valid to consider nickelocene in the same<br />
way as ferrocene and ruthenocene in relation to the NMR<br />
properties discussed here.<br />
3. Experimental<br />
13 C NMR spectra were recorded at 125.758 MHz on a<br />
Bruker MSL500 spectrometer using a Bruker doublebearing<br />
magic angle spinning probe capable of MAS<br />
frequencies between ca. 1 kHz and 12 kHz with stability<br />
better than ca. ± 10 Hz. All spectra were recorded at room<br />
temperature (293 ± 2 K) with the samples contained in<br />
zirconia rotors (4 mm external diameter).<br />
The 13 C "single pulse" sequence was used to record the<br />
spectra, with high power *H decoupling applied during<br />
acquisition. Typical parameters were: 13 C 90° pulse<br />
length = 3.5 \xs; recycle delay = 10 s for ferrocene and 20 s<br />
for ruthenocene.<br />
The *H decoupler field was set on resonance for the *H<br />
of each metallocene. An accurate assessment of the<br />
decoupler field strength was made by measuring (for<br />
adamantane) the length of the *H 90° pulse (t9o( 1 H)) for<br />
the *H r.f. power level used in the experiments involving<br />
*H decoupling. For the discussion of results, tgo^H) has<br />
been converted to a decoupling frequency vi via:<br />
1<br />
vi is related to the decoupler field strength Hi and to the<br />
parameter coi used in ref. 11 by the equations:<br />
_ Y( 1 H) Hi _ ©!_<br />
L 71 Z %<br />
The linewidth of the isotropic peak in the 13 C NMR<br />
spectrum was measured as the full width at half maximum<br />
height, and the experimental error in the measured<br />
linewidth is estimated to be less than ca. ± 5 Hz.<br />
4. Results and Discussion<br />
Initially we present the main results for ferrocene [12], and<br />
then consider the other metallocenes studied. Highresolution<br />
13 C NMR spectra of ferrocene were recorded<br />
initially at two different decoupler field strengths (vi =<br />
64.9 kHz and 26.6 kHz), and at several MAS frequencies<br />
(vr) ranging from ca. 1 kHz to 11 kHz. The spectrum at<br />
vr
Vol. 14, No. 1-4<br />
increases as vr is increased. The relationship between A<br />
and vr is approximately linear, particularly at the higher<br />
decoupler field strength, and the gradient is greater at the<br />
lower decoupler field strength. At fixed vr, the linewidth<br />
A is smaller at higher decoupler field strength, as shown in<br />
Fig. 1.<br />
A/Hz<br />
400<br />
300<br />
a n<br />
vr / kHz<br />
Fig. 1 Linewidth (A) versus MAS frequency (vr) for the<br />
isotropic peak in the 13 C NMR spectrum of ferrocene.<br />
The spectra were recorded at 'H decoupler field<br />
strengths corresponding to Vi = 64.9 kHz (•) and Vj =<br />
26.6 kHz (A).<br />
Our observation that the linewidth of the isotropic peak<br />
increases as the MAS frequency is increased conflicts with<br />
the discussion in Section 2 that, under conventional<br />
conditions, A should decrease, or remain constant, as vr is<br />
increased. We propose that the observed increase in A<br />
with increasing vr for ferrocene arises as a result of MAS<br />
indirectly modulating the efficiency of the *H decoupling<br />
in such a way that the effective decoupler field strength is<br />
decreased, leading to line-broadening, as vr is increased.<br />
This conclusion is supported by the results of three further<br />
experiments.<br />
(1) l^C NMR spectra of ferrocene were recorded for a<br />
series of different decoupler field strengths (with vi in the<br />
range 20 kHz to 80 kHz) at fixed vr. As expected, A<br />
decreases as vi is increased at fixed vr (Fig. 2).<br />
Furthermore, in the limit of sufficiently high decoupler<br />
field strength, A becomes essentially independent of both<br />
vr and vi, and converges to a limiting value of ca. 100 Hz.<br />
From this it can be concluded that both values of vr (5.06<br />
kHz and 9.05 kHz) used to record the data shown in Fig. 2<br />
are sufficiently rapid to remove any Vr-dependent sources<br />
of line-broadening due to CSA; i.e. at sufficiently high<br />
decoupler field strength, A is essentially independent of vr<br />
at these values of vr. This is consistent with the view<br />
that, under the conditions of the experiments shown in Fig.<br />
1, the property that does depend on vr is the efficiency of<br />
the *H decoupling (and not the ability of MAS to remove<br />
275<br />
the line-broadening effects due to CSA). As discussed<br />
fully elsewhere [12], there is a linear relationship between<br />
A and (vi)~ 2 .<br />
5001<br />
400<br />
A/Hz 300-<br />
200<br />
o o<br />
20 SO 70 80<br />
/ kHz<br />
Fig. 2 Linewidth (A) versus Vj for the isotropic peak in the<br />
13 C NMR spectrum of ferrocene. The spectra were<br />
recorded at MAS frequencies vr = 5.06 kHz (O) and vr =<br />
9.05 kHz (•).<br />
(2)<br />
13 C NMR spectra of ferrocene were recorded as a<br />
function of vr, but with no *H decoupling field applied;<br />
the variation of A with vr in these experiments is shown<br />
in Fig. 3. Under these conditions, A decreases as vr is<br />
increased, as predicted from the discussion in Section 1 and<br />
from ref. 10. This relationship between A and vr reflects<br />
the direct effect of MAS on the linewidth of the isotropic<br />
peak for a system that is subject to line-broadening by<br />
CSA and by direct 13 C-*H dipole-dipole interaction. At<br />
the lowest value of vr studied (ca. 3 kHz), A is ca. 592 Hz,<br />
and it is clear that slow MAS alone can substantially<br />
average the dipole-dipole interaction (as well as CSA) in<br />
A/Hz<br />
6001<br />
400"<br />
300<br />
vr / kHz<br />
Fig. 3 Linewidth (A) versus MAS frequency (vr) for the<br />
isotropic peak in the 13 C NMR spectrum of ferrocene,<br />
with no 'H decoupling applied.<br />
12
276<br />
this system. Undoubtedly, this is a consequence of the<br />
fact that the direct ^C- 1 !! dipole-dipole interaction is<br />
already extensively averaged by molecular motion.<br />
(3)<br />
13 C NMR spectra were recorded (using the 13 C<br />
"single pulse" method with no *H decoupling field applied)<br />
for perdeuterated ferrocene (denoted ferrocene-dio) as a<br />
function of MAS frequency. Over the range vr« 1 kHz to<br />
12 kHz, the linewidth of the isotropic peak is independent<br />
of vr (Fig. 4), and the value (A = 117 Hz) is close to the<br />
limiting linewidth obtained in our experiments for<br />
undeuterated ferrocene. In 13 C NMR spectroscopy of<br />
ferrocene-dio, the principal source of line-broadening is<br />
CSA and, furthermore, the CSA should be substantially<br />
the same in ferrocene-dio and in undeuterated ferrocene.<br />
The fact that A is essentially independent of vr for<br />
ferrocene-dio thus strongly supports the view that the<br />
increase of A with vr for undeuterated ferrocene is not<br />
arising from CSA being modulated by MAS.<br />
A / Hz<br />
200l<br />
ISO-<br />
100<br />
50<br />
vr / kHz<br />
10 12<br />
Fig. 4 Linewidth (A) versus MAS frequency (vr) for the<br />
isotropic peak in the 13 C NMR spectrum of<br />
ferrocene-djo, with no J H decoupling applied.<br />
High-resolution 13 C NMR spectra of ruthenocene and<br />
nickelocene were also recorded as a function of MAS<br />
frequency in order to establish whether the behaviour<br />
observed for ferrocene is also exhibited by other<br />
structurally-related systems. In Fig. 5, the relationships<br />
between A and vr for ruthenocene and ferrocene are<br />
compared, with all spectra recorded at the same *H<br />
decoupler field strength (vi = 64.9 kHz).<br />
It is clear that ruthenocene exhibits the same general<br />
trend as ferrocene, with A increasing as vr is increased,<br />
although the relationship for ruthenocene is apparently less<br />
linear than that for ferrocene. At the higher values of vr<br />
studied, the gradient 3A/3vr is larger for ruthenocene.<br />
These small differences in behaviour between ferrocene and<br />
ruthenocene presumably reflect small differences in the<br />
Bulletin of Magnetic Resonance<br />
dynamic properties of these solids at 293 K (as suggested<br />
in ref. 5).<br />
2501<br />
200"<br />
A /Hz lso-<br />
50-<br />
c ! 1<br />
vr / kHz<br />
10 12 14<br />
Fig. 5 Linewidth (A) versus MAS frequency (yr) for the<br />
isotropic peak in the 13 C NMR spectra of ferrocene (•)<br />
and ruthenocene (A), at fixed *H decoupler field<br />
strength corresponding to Vj = 64.9 kHz.<br />
The linewidth in the 13 C NMR spectrum of nickelocene<br />
is substantially greater (A = 15.7 kHz at vr = 5 kHz) than<br />
that in spectra recorded for ferrocene and ruthenocene under<br />
the same conditions. The linewidth for nickelocene is not<br />
significantly affected by increasing the MAS frequency,<br />
although there is a measurable decrease to A = 15.4 kHz at<br />
vr = 11 kHz. The very large 13 C NMR linewidth<br />
observed for nickelocene, even under conditions of MAS<br />
and high power *H decoupling, is due to the paramagnetic<br />
properties of this molecule (which contains two unpaired<br />
electrons). For this reason, it is not expected that the<br />
NMR properties of nickelocene will be comparable, in any<br />
way, to those of ferrocene and ruthenocene.<br />
5. Conclusions<br />
The increase in linewidth of the isotropic peak in the 13 C<br />
NMR spectra of ferrocene and ruthenocene as the MAS<br />
frequency is increased (at fixed temperature and fixed *H<br />
decoupler field strength) is due to an indirect effect in which<br />
the effective *H decoupler field strength is decreased as vr<br />
is increased. Considering the results of high-resolution<br />
13 C NMR experiments for several crystalline organic<br />
solids carried out in our laboratory, it is clear that, for<br />
systems in which there is no appreciable molecular<br />
motion, A is essentially independent of vr. This fact<br />
tends to suggest that the molecular motions present within<br />
crystalline ferrocene and ruthenocene are important in<br />
relation to the observed increase in the linewidth of the<br />
isotropic peak with increasing MAS frequency.
Vol. 14, No. 1-4<br />
A similar anomalous relationship between isotropic 13 C<br />
NMR linewidth and temperature has been observed recently<br />
by Muller [13] in studies of thiourea inclusion compounds;<br />
in this case, linewidths for 13 C environments in the guest<br />
molecules (which undergo substantial molecular motion)<br />
have been observed to increase with increasing temperature,<br />
and again this effect has been attributed to an "interference"<br />
between the molecular motion and the efficiency of the *H<br />
decoupling.<br />
In view of the fact (see Section 1) that cyclopentadienyl<br />
ring reorientation occurs on a timescale of the order of ca.<br />
10~ 12 s for ferrocene and ca. 10~ 10 s for ruthenocene at<br />
room temperature, it is not clear whether ring reorientation<br />
is indeed the dynamic process responsible for influencing<br />
the value of A in our experiments; it might be expected<br />
that a motion occurring at a frequency comparable to vi<br />
and/or vr (and thus in the approximate frequency range 10 3<br />
Hz to 10 6 Hz) would be required. At present, we make no<br />
attempt to assign the dynamic process that is important in<br />
giving rise to the observed NMR phenomena for ferrocene<br />
and ruthenocene at room temperature. However, it may be<br />
that the slower motion in ferrocene alluded to in ref. 2 is<br />
influential in this regard. Future experiments, at different<br />
temperatures, will investigate the variation of A with vr at<br />
different values of the correlation time for this motion, and<br />
other experimental investigations of the dynamic properties<br />
of crystalline metallocenes are in progress. We are also<br />
currently investigating various theoretical implications of<br />
the results reported here.<br />
Finally, it is relevant to strike a cautionary note in regard<br />
to recording high-resolution solid state 13 C NMR spectra<br />
for systems that are subject to line-broadening by CSA and<br />
direct ^C-*H dipole-dipole interactions. In view of the<br />
results reported here, it is clear that there are circumstances<br />
under which optimum resolution can be obtained by<br />
recording the spectrum at high *H decoupler field strength<br />
and low MAS frequency, rather than at high *H decoupler<br />
field strength and high MAS frequency.<br />
Acknowledgements<br />
277<br />
We are grateful to the S.E.R.C. (studentship to US) and<br />
the Nuffield Foundation (research grant to KDMH) for<br />
financial support.<br />
References<br />
[I] D. Braga, Chem. Rev., 92 (1992) 633.<br />
[2] C.H. Holm and J.A. Ibers, /. Chem. Phys., 30<br />
(1959) 885.<br />
[3] B.T.M. Willis, Ada Cryst., 13 (1960) 1088.<br />
[4] A.B. Gardner, J. Howard, T.C.Waddington, R.M.<br />
Richardson and J. Tomkinson, Chem. Phys., 57<br />
(1981) 453.<br />
[5] A.J. Campbell, CA. Fyfe, D. Harold-Smith and<br />
K.R. Jeffrey, Mol. Cryst. Liq. Cryst., 36 (1976) 1.<br />
[6] A. Kubo, R. Ikeda and D. Nakamura, /. Chem. Soc,<br />
Faraday Trans. 2, 82 (1986) 1543.<br />
[7] E.R. Andrew, Int. Rev. Phys. Chem., 1 (1981) 195.<br />
[8] E.R. Andrew, Phil. Trans. Roy. Soc. (Lond.) A, 299<br />
(1981)505.<br />
[9] M.M. Maricq and J.S. Waugh, /. Chem. Phys., 70<br />
(1979) 3300.<br />
[10] D. Suwelack, W.P. Rothwell and J.S. Waugh,<br />
/. Chem. Phys., 73 (1980) 2559.<br />
[II] W.P. Rothwell and J.S. Waugh, /. Chem. Phys., 74<br />
(1981)2721.<br />
[12] I.J. Shannon, K.D.M. Harris and S. Arumugam,<br />
Chem. Phys. Lett., in press.<br />
[13] K. Muller,/. Phys. Chem., 96(1992) 5733.
278<br />
1 Introduction<br />
The structural role of water in silicate glasses:<br />
1H and 29si NMR evidence<br />
J. Kiimmerlen, T. Schaller, A. Sebald and H. Keppler<br />
Bulletin of Magnetic Resonance<br />
Bayerisches Geoinstitut, Universitat Bayreuth, Postfach 101251<br />
W-8580 Bayreuth, Germany<br />
Over the last years various spectroscopic<br />
methods [1-6] have been used to<br />
investigate hydrous silicate and aluminosilicate<br />
glasses and to clarify the structural<br />
role of water in such systems. Especially,<br />
the existence of molecular water and/or<br />
Si-OH/Al-OH species, i.e. the H2O-induced<br />
depolymerisation of the Si/Al or Si<br />
network in such glasses has been discussed<br />
controversially in the literature.<br />
We have chosen two hydrous Na2Si4O9<br />
glasses with different H2O-contents as an<br />
Al-free model system.<br />
The Al-free Na2Si4O9 system is particularly<br />
well suited for high-resolution<br />
solid-state NMR investigations as the<br />
various Q-species can easily be resolved in<br />
the respective 29 Si MAS and CP/MAS<br />
spectra.<br />
Various high resolution solid state NMR<br />
techniques were used: 2!? Si singie pulse<br />
MAS, *H -> 29 Si-CP/MAS and *H-CRAMPS.<br />
Only the combined use of all these methods<br />
provides an insight into the interactions<br />
between the Si-network and the proton<br />
system. Additional modifications of these<br />
standard techniques were then used to<br />
confirm and to refine the picture of the<br />
Na2Si4O
Vol. 14, No. 1-4 279<br />
selectively and to obtain separate<br />
information about the dipolar protoninteractions<br />
("dipolar dephasing").<br />
29 Si dipolar dephasing<br />
10% H2O<br />
I • .I . . . I . . . I . . .I...I<br />
-70 -90 -110 ppm<br />
dephasing time<br />
1 ms<br />
5 ms<br />
10 ms<br />
Fig. 1: 29 Si-CP/MAS dipolar dephasing<br />
spectra of Na2Si4O9 glass with<br />
9.1% H2O. The glass with the<br />
lower water content shows<br />
similar decays of the signal<br />
components.<br />
3 Results and Discussion<br />
The 29 Si-MAS and 29 Si-CP/MAS spectra<br />
of the hydrous glasses show the presence<br />
°f Q4» Q3 and Q2 silicon species. The higher<br />
water content corresponds to a higher<br />
relative intensity of the Q2 signal and to a<br />
lower relative intensity of the Q4 signal,<br />
respectively. Furthermore, the 29 Si-<br />
CP/MAS experiments show significant<br />
differences in the time dependence of the<br />
29 Si magnetization of the two glasses. For<br />
the glass with the higher water content<br />
the signal intensity as a function of<br />
contact time is describable by a<br />
biexponential curve according to the usual<br />
thermodynamic model used to describe I-S<br />
CP-dynamics. The cross polarization times<br />
Tis and the relaxation times Tip show a<br />
trend for (Q4) > (Q3) >(Q2). Both Tis and Tip<br />
are significantly longer for the glass with<br />
the lower water content. The 29 Si CP-data<br />
of the Q3 and Q4 species for the lower H2Ocontent<br />
material can only be described by<br />
double-biexponential magnetization<br />
curves. This leads to the following<br />
conclusions: i) in the glass with the<br />
higher water content a uniform *H spin<br />
lattice is established, ii) due to the dilution<br />
of the protons in the other glass at least<br />
two ^H spin baths are present and iii) in<br />
the lower H2O-content glass the average<br />
interatomic distances between 29 Si and *H<br />
for the two different reservoirs must be<br />
significantly different. In consequence,<br />
one can assume the presence of both Si-OH<br />
groups and molecular water in these<br />
hydrous glasses. In fact, quantitative<br />
determination of the Q2, Q3 and Q4 species<br />
in both hydrous glasses from 29 Si MAS<br />
spectra and comparison of these results<br />
with the respective stoichiometric<br />
requirements if fully in accord with the<br />
depolymerisation of the silicate network.<br />
Results of the 29 Si-CP/MAS experiment<br />
with interrupted decoupling ("dipolar<br />
dephasing") are illustrated in Fig. 1. The<br />
decay of the 29 Si magnetization is mainly<br />
determined by the strength of the dipolar<br />
interaction between the 29 Si nuclei and<br />
the nearest surrounding proton system.<br />
Obviously, for the loss of magnetization<br />
characterized by a decay time T2 the<br />
relation T2(Q2)
280<br />
'H dipolar dtphasing<br />
dtphasing tinw<br />
»%H20<br />
20 10 0 ppm 20 10 0 ppm<br />
Fig. 2: ^-dipolar dephasing CRAMPS<br />
spectra of both glasses (450<br />
scans). (The spectra on the left<br />
hand side show an experimental<br />
artefact at ca. 18ppm)<br />
To confirm these CP results various J H<br />
multiple pulse experiments were used. The<br />
^H-CRAMPS spectra clearly show two well<br />
resolved signals at ca. 4.7 ppm and 12.2<br />
ppm (with respect to TMS = 0.0 ppm). The<br />
chemical shifts and the results of the CP<br />
experiments allow tentative assignment of<br />
these signals as molecular water (4.7 ppm)<br />
and protons of Si-OH groups (12.2 ppm).<br />
This assignment was confirmed by the *H<br />
"dipolar dephasing" CRAMPS experiment<br />
(jc/2-T-rc-T-MREV8). As illustrated in Fig. 2,<br />
in both glasses the signal at 4.7 ppm decays<br />
much faster than the less shielded signal,<br />
i.e., the dipolar * H - 1 H interaction is<br />
significantly stronger. This experimental<br />
fact corroborates the model proposed<br />
above. Comparing the decay rates found<br />
for the two different glasses, differences<br />
in these rates were found (Fig. 3). Again,<br />
these differences are in good agreement<br />
with the results of the standard 29 s i-<br />
CP/MAS experiments.<br />
. »0-<br />
3<br />
C<br />
v<br />
100<br />
>*, 90<br />
V)<br />
C<br />
Fig. 3:<br />
Bulletin of Magnetic Resonance<br />
a water protons<br />
A Si-OH protons<br />
a D<br />
5 10 15 20 25<br />
dephasing time (/is)<br />
Signal intensities in the * H<br />
dipolar dephasing CRAMPS<br />
experiment for the<br />
glass with<br />
a) 4.9 % H2O (top)<br />
b) 9.1 % H2O (below)<br />
Additionally, the longitudinal relaxation<br />
times Ti of both *H signal components in<br />
both glasses were measured. The use of a<br />
JC-T-JC/2-T-MREV8 pulse sequence provides<br />
a uniform Ti of 0.70 ± 0.05 s for the glass<br />
with the higher water content. In contrast,<br />
the signal components of the other<br />
glass decay with Ti(SiOH) = 0.85 ± 0.05 s and<br />
Ti(H2O) = 0.97 ± 0.05 s, respectively. These<br />
data also confirm the interpretation of the<br />
CP experiments and provide an additional<br />
support for the following conclusions:
Vol. 14, No. 1-4 " 281<br />
(1)H2O does depolymerize the silicate<br />
network,<br />
(2) both Si-OH and molecular water are<br />
present, and<br />
(3) in the lower H2O-content glass the<br />
more dilute proton system, forming<br />
these components is best described as<br />
"separate" sub-systems.<br />
To answer further questions concerning<br />
the interactions between Si-OH and<br />
H2O protons and to address questions<br />
concerning the structural role of cations<br />
like e.g. Na + in such silicate networks, ID<br />
and 2D-spin diffusion experiments are in<br />
preparation.<br />
Acknowledgment: Support of this work<br />
by the Deutsche Forschungsgemeinschaft<br />
and the Alexander von Humboldt-<br />
Foundation is gratefully acknowledged.<br />
Literatur:<br />
[I] Bartholomew, R.F., Butler, B.L., Hoover,<br />
H.L., Wu, C.K.J., J. Am. Ceram. Soc. 1980,<br />
63, 481<br />
[2] Kohn, S.C., Dupree, R., Geochim.<br />
Cosmochim. Acta 1987, 53, 2925<br />
[3] McMillan, P., Holloway, J.R., Contrib.<br />
Mineral. Petrol. 1987, 97, 320<br />
[4] Mysen, B.O. Virgo, D., Chem. Geol. 1986,<br />
64, 2623<br />
[5] Stolper, E.M., Am. Mineral. 1989, 74,<br />
1247<br />
[6] Eckert, H., Yesinowski, J.P., Silver,<br />
L.A., Stolper, E.M., J. Phys. Chem. 1988,<br />
92, 2055<br />
[7] Kiimmerlen, J., Merwin, L.H., Sebald,<br />
A., Keppler, H., J. Phys. Chem. (in<br />
press)<br />
[8] Oppella, S.J., Frey, M.H., J. Am. Chem.<br />
Soc. 1979, 101, 5854<br />
[9] Bodenhausen, G., Stark, R.E., Ruben,<br />
D.J., Griffin, R.G., Phys. Lett. 1979, 67,<br />
424<br />
[10] Rhim, W.K., Elleman, D.D., Vaughan,<br />
R.W., j. Chem. Phys. 1973, 59, 3740<br />
[II] Bronnimann, C.E., Zeigler, R.C.,<br />
Maciel, G.E., J. Am. Chem. Soc. 1988,<br />
110, 2023
282<br />
High-Resolution Solid-State NMR Study<br />
of Microstructures in Layered Aluminosilicate<br />
Bulletin of Magnetic Resonance<br />
Shigenobu Hayashi, Takahiro Ueda, Kikuko Hayamizu, and Etsuo Akiba<br />
1. INTRODUCTION<br />
National Chemical Laboratory for Industry,<br />
Tsukuba, Ibaraki 305, Japan<br />
Kaolin ite, Al4Si4Qo(OH)8> is a layered<br />
aluminosilicate with a dioctah edr al 1:1 layer<br />
structure consisting of an octahedral aluminum<br />
hydroxide sheet and a tetrahedral silica sheet.<br />
Figure 1 shows the structure of kaolin ite. Ihe crystal<br />
structure is not fully understood because of the<br />
absence of a large single crystal.<br />
In the present paper, we have traced 2 ^Si, 27 A1,<br />
and !H NMR spectra of various kaolinites, using<br />
high-resolution solid-state techniques. Analyzing<br />
the spectra theoretically, correlations between the<br />
NMR data and the local structures are discussed<br />
quantitatively.<br />
2. EXPERIMENTAL<br />
Totally eight samples were used. Six samples<br />
were natural, which were Kanpaku kaolin (called<br />
Nl; Hinckley index 1.4), API No.9 standard kaolin ite<br />
specimen (N2; 1.4), Georgia kaolin (N3; 0.7), Hakone<br />
Cbwakudani kaolin (N4; 0.4), Kibushi clay (N5; 0.2),<br />
and Gairome clay (N6; 0.2). Two synthetic samples<br />
were used,wh ich were synth esized at 290°C(51; 0.9)<br />
and220°C(52;0.8).<br />
NMR spectra were traced at room temperature by<br />
aBruker MSL400 (a static m agn etic field of 9.4 T) and<br />
a JBXGSH200 (4.71). Th e lin e sh apes of th e spectra<br />
were analyzed using computer programs written by<br />
ourselves.<br />
3. RESULTS AND DISCUSSION<br />
3.1. 2 ^Si spectra<br />
Figur e 2 sh ows 29 Si O7 MAS NMR spectr a of th e<br />
sample Nl, which has the highest crystallinity and<br />
the lowest content of paramagnetic impurities<br />
Fig. 1. Projection of the structure of kaolin ite from<br />
the(100) direction.<br />
among the eight samples studied. The spectra have<br />
two signals at -90.8 and -91.4 ppm from<br />
tetramethylsilane, being ascribed to Q'(GAl). The<br />
line shapes do not depend on the contact time, and<br />
the two peaks have the same cross relaxation time<br />
between ^H and 29 Si, which is 2.0 HIS. Maximum<br />
intensities are obtained at the contact time of 8 ms.<br />
The field dependence experiments demonstrate<br />
clearly th at two in equivalent Si sites are present.<br />
The linewidth in the 29 Si CP/MAS spectra is<br />
originated from the dipole-dipole interaction with<br />
27 Al,the chemical shift dispersion due to structural<br />
disorders, and the anisotropic magnetic<br />
susceptibility due to paramagnetic impurities.<br />
For Kanpaku kaolin,the chemical shift dispersion<br />
is 0.39 ppm, while the contributions of the dipolar<br />
interaction are0.08and033ppm for the fields of 9.4<br />
and4.7T,respectively,being estimatedfrom th e field
Vol. 14, No. 1-4 283<br />
B<br />
-85 -90 -95<br />
ppm<br />
79.496MHz 104.263MHz<br />
39.683MHz<br />
Fig. 2. 29 Si O7 MAS NMR spectra of th e sample JV1,<br />
measured at (A) 79.496 MHz and (B) 39.683<br />
MHz. Chemical shifts are expressed with<br />
r esp ect to tetr am eth y lsilan e.<br />
dependence experiments. The effect of<br />
paramagnetic impurities is negligible. The<br />
contribution of the dipole-dipole interaction with<br />
27 Al spin sis calculated theoretically from the crystal<br />
structure. The estimated linewidths at 9.4Tare 0.14<br />
and 0.09 ppm for Si(l) and Si(2), respectively,<br />
whereas those at 4.7 Tare 055 and 0.35 ppm. These<br />
values are in excellent agreement with the values<br />
estimated experimentally.<br />
Qher kaolinites, with lower crystallinities and/or<br />
higher contents of paramagnetic impurities, have<br />
broader linewidths due to the structural disorders<br />
an d th e par am agn etic impurities.<br />
3.2. 27 Al spectra<br />
Figur e 3 sh ows 27 A1 DE^ MAS NMR spectr a of th e<br />
sampleiv"l,measuredatdifferent fields. The Al atom<br />
in the kaolinite structure is coordinated by six<br />
oxygen atoms,and they give a signal around 0 ppm<br />
with respect to 1M A1(NQ)3 aqueous solution. The<br />
spectrum at the lower field gives the broader signal<br />
atthelower frequency position,which suggests that<br />
the signal is the central transition, being broadened<br />
by the second-order quadrupole interaction. The<br />
B<br />
52.051MHz<br />
100 50 0 -50 -100<br />
ppm<br />
Fig. 3. 27 A1 Hy MAS NMR spectra of the sample M,<br />
measured at (A) 104.263 MHz and (B) 52.051<br />
MHz. Chemical shifts are expressed with<br />
r espect to 1 M A1(NC$)3 aqueous solution.<br />
line shapes are simulated by our computer<br />
programs. The observed spectra cannot be<br />
simulated by one component, but can be simulated<br />
much better by two components with equal<br />
intensities. The obtained parameters for Al(l) are a<br />
chemical shift of 7.8 ppm, a quadrupole coupling<br />
constant of 3.36 MHz, an dan asymmetry factor of the<br />
quadrupole interaction of 055, while they are 7.8<br />
ppm, 2.88 MHz, and 1.00 for Al(2). The crystal<br />
structure indicates the presence of two in equivalent<br />
Al sites with a population ratio of 1:1.<br />
The 27 A1 spectra are also recorded for .the other<br />
kaolin ites. The same quadrupole coupling<br />
parameters as those in the sample JV1 can well<br />
explain th e lin e sh apes of th e oth er kaolin ites.<br />
A small fraction of tetrahedral Al is observed at<br />
about 70 ppm for several samples, which might be<br />
ascribed to impurities.<br />
3.3. 1 H static spectra<br />
Figure 4Ashows an 1 H static NMR spectrum of<br />
the sample JV1. The spectrum consists of two<br />
components with different line shapes; a narrow<br />
Lorentzian line with a width of 1.4 kHz and a broad<br />
Gaussian with a28.9kHzwidth.
284 Bulletin of Magnetic Resonance<br />
40 -40<br />
20 10 -10<br />
ppa<br />
Fig.4. *H NMR spectra of the sample Nl, measured at<br />
400.136 MHz. (A) The ordinary single-pulse<br />
sequen ce is used for th e static sample. (B) Th e<br />
CRAMPS spectrum measured with the BR24<br />
pulse sequence in the quadrature phase<br />
detection mode. Chemical shifts are expressed<br />
with respect to tetramethylsilane.<br />
The narrow component can be ascribed to water<br />
molecules adsorbed on the outer surface. This<br />
component is easily diminished by evacuation, and<br />
they grow up gradually in the air atmosphere.<br />
Qi the other hand,the broad component can be<br />
ascribed to the hydroxyl groups in the kaolinite<br />
structure, whose second moment is 105 kHz 2 . The<br />
second moment estimated from the crystal structure<br />
is 92 kHz 2 , in which iH-iH and 1 H- 27 A1<br />
contributions are 67 and 25 kHz 2 , respectively. The<br />
calculated second moment agrees with the<br />
experimental value. These results demonstrate that<br />
the hydrogen atoms in the CH group is in a rigid<br />
lattice state at room temperature.<br />
The *H static NMR spectra of other kaolin ites also<br />
con sist of th e two compon en ts.<br />
3.4. 1 E CRAMPS<br />
The CRAMPS technique is successfully applied to<br />
the kaolinite samples. Figure 4Bshows l H CRAMPS<br />
spectra of the sample JV1. The BR24 pulse sequence is<br />
used in the quadrature phase detection mode.<br />
Considerably large spinning sidebands are observed<br />
on both sides of the central peak, which are caused<br />
by the strong dipole-dipole interaction between *H<br />
and 27 A1 spins. The linewidth in the static state, 29<br />
kHz, is reduced to about 600 Hz by the use of the<br />
CRAMPS technique, where the reduction factor is<br />
about 50. The chemical shift is 2.8 ppm from<br />
tetramethylsilane.<br />
The chemical sh ift of the hydroxyl groups in the<br />
kaolin ites changes slightly depending on the<br />
sample, which might reflect the strength of the<br />
hydrogen bonding or the acidity of the hydrogen .<br />
Acknowledgement<br />
The authors wish to express their thanks to Dr. R.<br />
Miyawaki at Government Industrial Research<br />
Institute, Nagoya and Dr. K. Kuroda at Waseda<br />
University for kindly supplying the kaolin ite<br />
samples.
Vol. 14, No. 1-4 285<br />
1 Introduction<br />
Broadline NMR of Structural Ceramics<br />
Magic angle spinning (MAS) has become the most widely<br />
used NMR technique for the study of inorganic solids.<br />
This popularity has come about because MAS removes<br />
the effects of chemical shift anisotropy, permitting acquisition<br />
of chemical shift spectra in these materials.<br />
However, there are two weaknesses to the MAS<br />
technique, which are accentuated in the study of structural<br />
ceramics.<br />
The first difficulty with MAS is that rotation at the<br />
magic angle does not remove the effects of second order<br />
quadrupolar broadening of the central (+1/2 «—» -1/2)<br />
transition [1]. This has limited the applicability of NMR<br />
for half-integral quadrupolar nuclei like 27 A1, U B, 17 O,<br />
and 91 Zr, which are important constituents of ceramics.<br />
While recent work has shown that the effects of second<br />
order quadrupolar broadening are reduced by working in<br />
larger magnetic fields [2], or by employing more sophisticated<br />
spinning techniques [3], these solutions can be<br />
difficult and expensive to implement.<br />
The other important limitation of MAS, also true of<br />
other sample spinning techniques, is that the physical<br />
form of the sample is restricted to fine powders, homogeneous<br />
cylinders, or chunks of material packed in an NMR<br />
inert powder of similar density. All of these forms present<br />
problems when working with ceramics. Many ceramics<br />
are challenging to machine or grind due to their extreme<br />
hardness. Even when grinding is possible, other<br />
requirements may limit the use of powders. For example,<br />
in determining the effect of long term or repeated heating<br />
of a ceramic, heating a powdered sample may provide<br />
unreliable data because of surface oxidation. Packing<br />
large chunks of sample in a powder with similar density<br />
becomes time consuming when a number of samples have<br />
to be studied, or the sample is being subjected to heating<br />
which changes its chemical or phase composition.<br />
We have used broadline NMR of static samples as an<br />
alternative to MAS for samples with large quadrupolar<br />
splittings. In favorable cases, the resulting powder<br />
pattern, produced by the first order quadrupolar coupling<br />
Chuck Connor<br />
Defence Research Establishment Pacific<br />
FMO Victoria, B.C., VOS 1B0, Canada<br />
[1], directly yields information about the electronic environment<br />
of the nuclei under investigation. Although the<br />
broadline technique for quadrupolar nuclei is not new,<br />
having been first applied over forty years ago [4], it is a<br />
simple, useful technique which is often overlooked. It is<br />
generally applicable only to nuclei with large quadrupolar<br />
splittings, which provide the most difficulty for MAS;<br />
accordingly the broadline technique can serve as a complement<br />
to MAS. The work reported here demonstrates<br />
that, for many ceramics, broadline spectra can provide<br />
useful structural information.<br />
2 Experimental<br />
All spectra were recorded on a Bruker MSL-300<br />
spectrometer, equipped with the BC-131 5 MHz 9 bit<br />
digitizer. The spectrometer operates at 96 MHz for U B,<br />
and 78 MHz for 27 A1. A standard Bruker multinuclear<br />
solenoid probe was used for the broadline spectra. Free<br />
induction decays were collected after a single 1 \is pulse.<br />
Although the excitation profile of this pulse drops to zero<br />
at 1 MHz, reasonable sensitivity is still maintained for<br />
satellites 600 to 700 kHz off resonance. The delay time<br />
before the start of data collection was typically 5 u.s.<br />
Usually the first 2 or 4 points of the FID were discarded<br />
because of pulse breakthrough. The relatively long<br />
deadtime usually prohibits observation of the broad<br />
pedestal portion of the powder pattern, but the cusps,<br />
which by themselves are sufficient to characterize the<br />
quadrupolar splitting, do not appear greatly affected. The<br />
width of the spectral window was 2.5 MHz (+1.25 MHz)<br />
for U B and 1.67 MHz (±0.833 MHz) for 27 A1. A filter<br />
bandwidth of 1 MHz, the largest available on the MSL-<br />
300 in quadrature mode, was used. Additional data collection<br />
parameters used for individual spectra are noted in<br />
the figure captions.<br />
Quadrupolar splittings were measured from a point on<br />
the outside edge of one satellite, at about 80% of the<br />
satellite peak height, to the corresponding point on the<br />
other satellite of the pair. The value 80% was obtained by
286 Bulletin of Magnetic Resonance<br />
comparing simulated powder patterns with and without<br />
Gaussian broadening. Measurements between the maxima<br />
of the satellites give values that are smaller than the true<br />
value, because dipolar broadening shifts the position of<br />
the maxima toward the central transition.<br />
3 Results and Discussion<br />
The 27 A1 (/ = 5/2) resonance in corundum (a-Al2O3) has<br />
been observed by a variety of NMR techniques [4], [5].<br />
For comparison, we show an 27 A1 powder pattern in<br />
Figure 1. For this sample, excellent sensitivity is obtained<br />
500000 0<br />
HERTZ<br />
-500000<br />
Fig. 1. 27 A1 powder pattern of a-Al2O3. 32 scans were coadded,<br />
using a recycle delay of eight seconds. The upper trace is<br />
a magnification of the lower trace by ten.<br />
with about four minutes of signal averaging. From the<br />
splitting of each pair of satellites, we find<br />
e 2 qQ/h = 2.42+0.02 MHz and r| = 0.0, in fair agreement<br />
with the values determined by Pound [4]. The slight<br />
reduction in precision of these numbers, as compared with<br />
Pound's work, is offset by the speed and ease with which<br />
the results can be obtained on a standard solids NMR<br />
spectrometer.<br />
One of the many applications for NMR of ceramics is<br />
to follow high temperature phase changes and reactions.<br />
We have studied the calcination of gibbsite (A1(OH)3) to<br />
form a-Al2O3, which is a complex and poorly understood<br />
process [6]. Results of preliminary work to determine the<br />
applicability of broadline NMR to this problem are shown<br />
in Figure 2. The ^Al spectrum of the unheated material<br />
(determined by X-ray diffraction to be at least 90%<br />
gibbsite) shows five pairs of satellite transitions, with<br />
splittings of 96 kHz, 397 kHz, 517 kHz, 608 kHz, and<br />
960 kHz. The peaks near ±700 kHz, which appear in<br />
many 27 A1 spectra, are probably artifacts. Assuming three<br />
sites are present, the splittings can be paired as follows:<br />
397 and 608 kHz, from a site with e 2 qQ/h = 2.1 MHz and<br />
500000<br />
0<br />
HERTZ<br />
-500000<br />
Fig. 2. 27 A1 powder patterns of gibbsite (A1(OH)3) before<br />
heating (lower trace), after heating at 700 C for 35 minutes<br />
(middle trace), and after further heating at 1070 C for 16 hours<br />
(upper trace). Each trace represents 1 to 2 hours of signal<br />
averaging, with a one second recycle delay.<br />
T] = 0.5; 517 and 960 kHz, from a site with<br />
e^qQ/h = 3.2 MHz and Y| = 0.2. The other possible pairing<br />
gives one site with e*qQlh = 1.8 MHz (r\ = 0.7) and one<br />
with e 2 qQlh = 3.3 MHz (r\ = 0.5). Only one pair of<br />
satellites from the remaining site is visible, with a splitting<br />
of 96 kHz. Two possibilities may explain this absence of<br />
a second pair. If e 2 qQ/h = 0.36 MHz and r\ ~ 1, the<br />
second pair of satellites will overlap the observed pair. If<br />
e*qQ/h = 0.32 MHz and r\ - 0, the second pair will have a<br />
splitting of about 48 kHz, and will not be resolved from<br />
the central transition. An accurate quantification of the<br />
relative population of the three sites from this data is<br />
difficult, but one can say the three sites are roughly equally<br />
populated. The nuclei with the larger couplings are<br />
probably in octahedral environments, as expected from the<br />
reported structure, in which all the aluminum nuclei are in<br />
octahedral environments [7]. The relatively large values<br />
of r| can be attributed to hydrogen bonding, which distorts<br />
the octahedral symmetry of the aluminum sites in gibbsite.<br />
The smaller quadrupolar coupling is about an order of<br />
magnitude less than that typically observed for octahedral<br />
sites (cf. a-Al2O3), so is probably due to a slightly<br />
distorted tetrahedral site. From a simulation of MAS<br />
results at 6.35 T [2], it appears that only two sites are<br />
observed by MAS, both of which are octahedral.<br />
Conversion of A1(OH)3 to CC-A12O3 involves several<br />
intermediate phases such as boehmite, X-A12O3, y-Al2O3,<br />
K-A12O3, 6-Al2O3, and 8-Al2O3, with formation of oc-<br />
A12O3 reportedly occurring at 1140 C [6]. The middle
Vol. 14, No. 1-4<br />
trace in Figure 2 shows the effect of heating gibbsite at<br />
700 C for 35 minutes. The satellites attributed to gibbsite<br />
have disappeared, and the central transition is flanked by<br />
broad features indicative of amorphous material. The<br />
width of the central transition, which may be attributed to<br />
the second order quadrupolar interaction, corresponds to<br />
values of e 2 qQ/h ranging up to 4.4 MHz. After further<br />
heating for 16 hours at 1070 C, the upper spectrum in<br />
Figure 2 was obtained. The sharp features indicate a<br />
substantial amount of amorphous material has converted<br />
to a-Al2O3. This contradicts the notion that conversion to<br />
a-Al2O3 occurs only at temperatures above 1140 C.<br />
However, the conversion is quite sluggish at 1070 C.<br />
These broadline spectra demonstrate conversion of<br />
gibbsite to a-Al2O3, but unfortunately information on the<br />
intermediate phases cannot be obtained from these spectra.<br />
Chemical shift spectra obtained with sample spinning will<br />
probably not be useful in determining which phases are<br />
present in the amorphous intermediate stage, considering<br />
the large quadrupolar couplings which appear to be<br />
present.<br />
Another problem of interest is the oxidation of<br />
ceramics at high temperatures. When boron nitride (BN)<br />
is heated in an oxidizing environment, conversion to<br />
boron oxide gradually takes place. Since both BN and<br />
B2O3 have large quadrupole couplings [8], [9], one can<br />
expect difficulty in resolving the signals from the two<br />
materials in MAS spectra. In fact, the second order<br />
quadrupolar broadening is on the order of 50 ppm, based<br />
on a Larmor frequency of 96 MHz, which is a significant<br />
fraction of the expected range of chemical shift for n B.<br />
To demonstrate that broadline NMR may prove more<br />
useful in monitoring the oxidation of BN, we prepared a<br />
mixture of roughly equal amounts of BN and B(OH)3.<br />
Boric acid was used instead of boron oxide because the<br />
fine powder required for MAS work is not hygroscopic, as<br />
is the B2O3 powder, and the boric acid was readily available.<br />
Boric acid has a quadrupolar splitting of 1282 kHz,<br />
similar to that of B2O3 (1308 kHz).<br />
U B (/ = 3/2) MAS and broadline spectra of the boron<br />
nitride/boric acid mixture were recorded, and are shown in<br />
Figures 3 and 4 respectively. From the MAS spectrum,<br />
one can see that the single peak observed is a composite of<br />
several resonance lines. However, one would have<br />
difficulty determining which boron species are present in<br />
the sample, or even how many species there are. This<br />
spectrum should be contrasted with the broadline<br />
spectrum shown in Figure 4. One immediately sees that<br />
two boron sites are present, with quadrupolar splittings of<br />
1462±10 kHz and 1256±10 kHz. By comparison with<br />
previous work [8], [9] the species are identified as boron<br />
nitride and boric acid respectively. Based on this model<br />
mixture, we expect to be able to monitor the oxidation of<br />
boron nitride.<br />
A variety of approaches, including double-rotation<br />
(DOR), dynamic-angle-spinning (DAS) [3], and MAS in<br />
larger static fields [2], have been used to improve the<br />
50000 0<br />
HERTZ<br />
-50000<br />
Fig. 3. MAS spectrum of n B in a mixture of boron nitride and<br />
boric acid, with a 5 kHz spinning rate. 2296 scans were<br />
collected, with a recycle delay of one second.<br />
resolution of chemical shift spectra beyond that attainable<br />
by conventional MAS. Broadline NMR may compare less<br />
favorably with these new techniques than it does with<br />
MAS. However, these techniques have not been widely<br />
exploited in materials research, and, like MAS, they suffer<br />
from the physical limitations imposed on the sample as<br />
discussed in the Introduction.<br />
Titanium boride (TiB^ has been presented as a candidate<br />
for lightweight armor [10]. The structure consists of<br />
alternating planar layers of boron and titanium, with each<br />
boron nucleus trigonally bound to three other boron nuclei<br />
[11]. To yield the 2:1 stoichiometry, each boron layer<br />
contains twice as many nuclei as a titanium layer. The<br />
crystal structure suggests only one type of boron site is<br />
present, and previous workers have reported only one site,<br />
with a quadrupolar splitting of 180*10 kHz [9]. However,<br />
500000 0<br />
HERTZ<br />
-500000<br />
AJL.<br />
Fig. 4. U B powder pattern of the mixture examined in Figure<br />
3. This spectrum was obtained after about 16 hours of signal<br />
averaging (56989 scans, one second recycle delay).<br />
287
288<br />
we see evidence for at least two sites in breadline U B<br />
NMR spectra (Figure 5), with quadrupolar splittings of<br />
500000<br />
0<br />
HERTZ<br />
-500000<br />
Fig. 5. n B powder pattern of titanium boride (TiBj), from 324<br />
co-added scans with a recycle delay of one second.<br />
177 kHz and 355 kHz. In addition, a weak, poorly<br />
resolved pair of satellites may be present with a splitting<br />
of 195 kHz. It appears that roughly four fifths of the<br />
boron nuclei are located in the 177 kHz site, one fifth in<br />
the 355 kHz sites, with a much smaller fraction in the<br />
195 kHz site.<br />
X-ray diffraction confirmed the sample was greater<br />
than 90% TiB2, with most of the remainder consisting of<br />
titanium oxides. The general features of the pattern<br />
matched that of A1B2, which also has planar layers. The<br />
diffraction pattern showed no similarity to patterns from<br />
compounds containing puckered boron layers, such as<br />
RhB2, TcB2, and Ru2B3. Thus it appears the two sites are<br />
in equivalent geometrical positions within the boron<br />
plane. The larger quadrupolar splitting could be due to<br />
decreased donation of electrons from the titanium atoms<br />
to the boron it orbitals in about one fifth of the boron<br />
nuclei [9]. This heterogeneity within the boron layers<br />
may explain why titanium boride does not fracture along<br />
the interface between layers, as does graphite and many<br />
other layered materials.<br />
4 Summary<br />
For many ceramics, the second order quadrupolar broadening<br />
in MAS spectra exceeds the dispersion provided by<br />
the chemical shift. The far greater dispersion from the<br />
first order quadrupolar interaction can be used to obtain<br />
resolved spectra from these samples. We have provided<br />
examples of several structural ceramics for which<br />
broadline spectra, obtained with a standard solids<br />
spectrometer, yield useful structural information. Thus<br />
the technique shows the potential to complement MAS in<br />
the study of inorganic materials.<br />
5 References<br />
Bulletin of Magnetic Resonance<br />
1. A. Abragam, Principles of Nuclear Magnetism<br />
(Clarendon Press, Oxford, 1961), Chapter VII.<br />
2. e.g. D.E. Woessner, Am. Mineral. 74, 203 (1989).<br />
3. B.F. Chmelka, K.T. Mueller, A. Pines, J. Stebbins,<br />
Y. Wu, and J.W. Zwanziger, Nature, 339, 42<br />
(1989).<br />
4. R.V. Pound, Phys. Rev. 79,685 (1950).<br />
5.a) D. Lee and P.J. Bray, /. Magn. Reson. 94, 51<br />
(1991).<br />
b) J. Haase and H. Pfeifer, /. Magn. Reson. 86, 217<br />
(1990).<br />
c) HJ. Jakobsen, J. Skibsted, H. Bilds0e, and N.C.<br />
Nielsen,/. Magn. Reson. 85,173 (1989).<br />
6. Engineered Materials Handbook, Vol. 4, Ceramics<br />
and Glasses (ASM International, 1991), p.112.<br />
7. A.F. Wells, Structural Inorganic Chemistry<br />
(Clarendon Press, Oxford, 1962), p. 552.<br />
8.a) A.H. Silver, J.Chem.Phys. 32, 959 (1960).<br />
b) C. Connor, J. Chang and A. Pines, J. Chem. Phys.<br />
93, 7639 (1990).<br />
9. PJ. Bray, AIP Conf. Proc. 140,142 (1986).<br />
10. D.J. Viechnicki, M.J. Slavin, and M.I. Kliman,<br />
Ceramic Bulletin, 70,1035 (1991).<br />
11. T. Lundstrom in Boron and Refractory Borides,<br />
edited by V.I. Matkovich (Springer-Verlag, Berlin,<br />
1977), p. 351.
Vol. 14, No. 1-4 289<br />
PERMEABILITY OF LIPOSOMAL MEMBRANES TO MOLECULES OF<br />
ENVIRONMENTAL INTEREST: RESULTS FROM NMR EXPERIMENTS<br />
EMPLOYING SHIFT AGENTS<br />
INTRODUCTION:<br />
by<br />
F.G. Herring*, W.R. Cullen, J.C. Nelson and P.S. Phillips,<br />
Department of Chemistry, University of British Columbia,<br />
Vancouver, B.C., Canada V6T 1Z1<br />
Dimethylarsinic acid (DMA) is a<br />
widely used pesticide and is an important<br />
intermediate in the marine bio-cycling of<br />
arsenic. Studies into the uptake mechanism<br />
of this organo-arsenical have shown that it<br />
enters cells by slow passive diffusion (1).<br />
The work presented in this article describes a<br />
NMR technique that has been developed to<br />
measure the rate of diffusion of compounds<br />
through a phospholipid bilayer. A similar<br />
method has been described by Prestegard<br />
et.al. (2) for the measurement of maleic acid<br />
diffusion constants.<br />
The diffusion rates of molecules<br />
across the membrane of liposomes have been<br />
measured using a variety of techniques, the<br />
most common of which is radio-labelling.<br />
The technique demonstrated in this study has<br />
many advantages over radio-labelling some<br />
of which are as follows; the cost and<br />
difficulties associated with working with<br />
radio-labelled compounds are eliminated, the<br />
method is readily automated, and sampling is<br />
eliminated. The NMR method is applicable<br />
to any water soluble compound which has a<br />
*H resonance signal which does not overlap<br />
*The abbreviation (DMA) does not<br />
distinguish between the protonated (DMAH)<br />
and the unprotonated (DMA") forms of the<br />
acid.<br />
with any other peaks or which can be shifted<br />
either upfield or downfield agent. It can be<br />
used on molelecules which have a<br />
permeability through the phospholipid bilayer<br />
of 10'^cm/s or less.<br />
We present here an investigation of<br />
DMA transport in a model membrane system<br />
by using this NMR method as applied to<br />
diffusion across the membranes of extruded<br />
large unilamella vesicles (LUVs) (3). The<br />
study contributes to our understanding of<br />
diffusive transport across bilayer membranes.<br />
It also illustrates the potential of NMR<br />
spectroscopy for membrane permeability<br />
studies in large unilamella vesicles.<br />
MATERIALS AND METHODS:<br />
Dry egg phosphatidyl-choline was<br />
hydrated with a buffered solution (in D2O)<br />
of DMA at the appropriate pH. The solution<br />
was then subjected to several freeze-thaw<br />
cycles using liquid nitrogen to enhance<br />
entrapment of DMA and to increase the<br />
degree of unilamellarity of the phospholipid<br />
bilayer. The multi-lamella suspension was<br />
extruded through polycarbonate filters with a<br />
200 nm pore diameter under high pressures<br />
to ensure that the vesicles used were all<br />
approximately the same size (3,4). The<br />
resultant solution of LUVs was then passed<br />
down a Sephadex column and eluted with<br />
buffer (in H2O) to remove most of the DMA
290<br />
that was not encapsulated; this procedure<br />
establishes the desired concentration gradient<br />
for diffusion studies. The eluted LUVs were<br />
added to a NMR tube which already<br />
contained Mn 2+ , HEPES, TSP, and glucose.<br />
NMR spectra were obtained at appropriate<br />
time intervals by using a Bruker AM<br />
4OO.The water signal was suppressed by presaturation.<br />
The FIDs were processed with a<br />
line-broadening of 10 Hz. The DMA and<br />
HEPES peaks were integrated and the ratio<br />
of the integrals taken as a measure of the<br />
efflux of DMA, this procedure was adopted<br />
to account for instrument variability.<br />
THEORY:<br />
-dnin/dt = dnout/dt = k (nin/Vin-n0Ut/V0Ut)<br />
Inside peak:<br />
nt. 11 =neq. +(no. 1 .neq. II \e-(l+f)kt/V.<br />
e<br />
in " in v in m> in<br />
Outside peak:<br />
P = k/A<br />
°out<br />
(1)<br />
(4)<br />
Equation (1) is the basic rate<br />
equation for the flux of particles across the<br />
bilayer assuming that the rate of appearance<br />
of the particles on the outside is equal to the<br />
rate of disappearance of the particle on the<br />
inside. Equations (2) and (3) are the<br />
equations which relate the number of<br />
particles on either side of the membrane as a<br />
function of time. Equation (4) relates the<br />
internal volume to the external volume.<br />
Equation (5) relates the rate constant to the<br />
permeability coefficient which is the standard<br />
measure of the rate of diffusion of a<br />
compound through a membrane.<br />
For the integral of the composite<br />
methyl resonance (DMAH and DMA") we<br />
have:<br />
(5)<br />
Bulletin of Magnetic Resonance<br />
where kobs = a ICDMAH Vjn, is the<br />
observed rate constant of diffusion for<br />
DMAH and a = Ka / (Ka + [H+]).<br />
Equation (6) is valid if it is assumed that<br />
^DMA"
Vol. 14, No. 1-4 291<br />
0.25<br />
both sides of the membrane. The membrane<br />
is impermeable to these three additives over<br />
the time scale of an experiment. It should be<br />
noted that DMA is a weak acid which is<br />
about 80% dissociated at pH = 7 (pKa =<br />
6.28).<br />
The decrease of the NMR peak for<br />
those molecules effusing and a corresponding<br />
increase in the peak for DMA outside is<br />
displayed in Fig 1. At the equilibrium position<br />
(the last spectrum shown) the ratio of<br />
the integrals of these peaks correspond to<br />
the ratio of the volume inside to outside (eq<br />
(4))-<br />
The amplitude ratios (total integral of<br />
DMAH and DMA" to the integral of the<br />
HEPES buffer) of both the inside and outside<br />
peaks for the 25 spectra which make up a<br />
single experimental run are shown as a function<br />
of time in Fig. 2. An iterative fit of<br />
these curves using a spreadsheet program<br />
(QPRO) permits the estimation of the rate<br />
constant.<br />
10 15<br />
TIME(SECS.)<br />
(Thousands)<br />
Fig. 2<br />
20 25<br />
In order to demonstrate that the<br />
transport is dominated by the neutral species<br />
(DMAH) as suggested above, we performed<br />
experiments at different pH. The rate of<br />
change of the integral ratio will depend upon<br />
the concentration of DMAH. The integral<br />
ratio will decrease faster due to the increased<br />
fraction (a) of DMAH contributing to the<br />
single methyl resonance observed for both<br />
DMA" and DMAH. The table illustrates the<br />
results we obtained.<br />
Table<br />
The pH dependence of the<br />
observed rate constant<br />
pH<br />
7.00<br />
7.15<br />
7.40<br />
7.73<br />
7.97<br />
(/cm 3 s-l*10" 5 )<br />
14.50<br />
8.89<br />
5.65<br />
2.39<br />
1.56
292<br />
Equation (6) gives the true rate<br />
constant for DMAH from the measured<br />
observed rate constant. The true rate<br />
constant and permeability coefficient are<br />
1.08.x 10" 3 cm 3 /s and 3 x 10' 8 cm/s<br />
respectively. Similar studies for monomethyl<br />
arsonic acid (MMAH) show that the<br />
permeability is 2 x 10"*" cm/s. This<br />
difference is consistent with the general rule<br />
that replacing a hydroxyl group with a<br />
methyl group will increase the permeability<br />
of the molecule by approximately two to<br />
three orders of magnitude(6).<br />
ACKNOWLEDGMENTS:<br />
This work is supported by operating<br />
grants from the NSERC of Canada. The<br />
exceptional technical assistance of Alice Ho<br />
and Anna Mason is appreciated. Alice and<br />
Bulletin of Magnetic Resonance<br />
Anna also received Summer '92 Studentships<br />
from NSERC.<br />
REFERENCES:<br />
(1) W.R. Cullen, B.C. McBride, A.W.<br />
Pickett, Appl. Organomet. Chem. 4,<br />
119,(1990).<br />
(2) J.H. Prestegard, J.A. Cramer and<br />
D.B. Viscio, Biophys. J. (Biophysical<br />
Society) 26, 575, (1979).<br />
(3) M.J. Hope, M.B. Bally, G. Webb and<br />
P.R. Cullis, Biochim. Biophys. Acta<br />
872. 55-65, (1985).<br />
(4) L.D. Mayer, M.J. Hope and<br />
P.R.Cullis, Biochim. Biophys. Acta<br />
858, 181,(19861<br />
(5) W.C. Stein, Channels Carriers and<br />
Pumps, Academic Press, New York,<br />
(1990).
Vol. 14, No. 1-4<br />
NUCLEAR MAGNETIC RESONANCE PARTITIONING STUDIES OF<br />
SOLUTE ACTION IN LIPID MEMBRANES<br />
Lan Ma, Theodore F. Taraschi, and Nathan Janes<br />
Department of Pathology and Cell Biology,<br />
Thomas Jefferson University, 1020 Locust St., Philadelphia, PA 19107<br />
INTRODUCTION: Lipid theories of<br />
anesthesia implicate perturbation of membrane<br />
lipids as the locus for acute anesthetic<br />
action. [1] Chronic exposure to alcohols and<br />
anesthetics induces an adaptive response in<br />
membrane phospholipids that confers resistance<br />
to many of the acute actions of alcohols<br />
and anesthetics. [2]<br />
We have proposed a colligative thermodynamic<br />
reformulation of the Meyer-Overton<br />
hypothesis for anesthetic action. [3,4] This<br />
reformulation implicates configurational<br />
entropy (Scf), the entropy imparted by a solute<br />
upon a membrane structure in the partitioning<br />
process, as the driving force of solute action<br />
on cooperative membrane equilibria. Solute<br />
potency is determined by the competing contributions<br />
of configurational and thermal<br />
entropy (ASt). Equilibria most susceptible to<br />
solute action (where dilute concentrations of<br />
solutes induce a perturbation equivalent to a<br />
large change in temperature) involve large<br />
changes in configurational entropy and small<br />
changes in thermal entropy according to the<br />
following relation. [3]<br />
AT/Tm = AScf/ASt (1)<br />
AT is the perturbation of the midpoint temperature,<br />
Tm, from its value in the absence of<br />
solute. The thermal entropy of an equilibrium<br />
is deduced from calorimetry and is approximately<br />
constant for solute levels of biological<br />
relevance. The remaining unknowns are the<br />
configurational entropy, which is determined<br />
from the partitioning of the solute, and the<br />
perturbation of the equilibrium midpoint.<br />
The colligative thermodynamic framework<br />
implicates solute partitioning as the energetic<br />
force that drives perturbations of cooperative<br />
membrane equilibria by altering the relative<br />
free energies of membrane states. Tests of<br />
the framework require simultaneous measures<br />
of solute partitiomng and membrane structure<br />
over a range of solute concentrations and<br />
temperatures.<br />
293<br />
Spin label* partitioning protocols have often<br />
been used in ESR studies of membrane structure.<br />
[5] Such studies are designed so that the<br />
spin label partitioning probe is a<br />
nonperturbing reporter of membrane structure.<br />
To study solute action on membranes,<br />
however, a partitioning probe should serve the<br />
multifarious role of membrane perturbant,<br />
reporter of perturbations, and reporter of<br />
solute partitioning. Since NMR methods are<br />
not limited to dilute solute levels, such flexibility<br />
is offered. Furthermore, complementary<br />
structural information is available from<br />
simultaneous wideline X H [6], 2 H [7], or 31 P<br />
[8] studies.<br />
PARTITIONING APPROACH TO SOLUTE<br />
ACTION: In this abstract, we describe a H<br />
NMR partitioning approach based on the<br />
uncharged local anesthetic alcohol, benzyl<br />
alcohol. Benzyl alcohol is a clinically used<br />
topical bacteriostatic agent. A variety of<br />
commercial pharmaceutical agents prepared<br />
for injection contain benzyl alcohol for its<br />
preservative properties and for pain relief.<br />
The partitioning approach is based on (/) the<br />
sensitivity of the ring proton chemical shift to<br />
the polarity of its environment and («) the<br />
sensitivity of the ring proton linewidth to<br />
membrane binding. The chemical shift of the<br />
ring resonances in a hydrophobic environment<br />
are shielded and resolved from the ring resonances<br />
of the aqueous alcohol. The sensitivity<br />
of the ring proton resonance to its environment<br />
provides a means of discriminating the<br />
aqueous alcohol resonance from the partitioned<br />
alcohol resonance. The dependence of<br />
the three chemically distinct ring proton<br />
chemical shifts on their environment is shown<br />
in Table 1 for benzyl alcohol (5 mole fraction<br />
%) in a variety of bulk solvents at 22°C. The<br />
resonance exhibits a diamagnetic shift in<br />
hydrophobic solvents. A modest correlation<br />
between the chemical shift and Hildebrandt's<br />
solubility parameter ($*) for the solvent is<br />
evident.
294<br />
SOLVENT<br />
Water<br />
Methanol<br />
1-Propanol<br />
1-Butanol<br />
1-Octanol<br />
1-Decanol<br />
Acetone<br />
Methylene Chloride<br />
Chloroform-dj<br />
Carbon Tetrachloride<br />
Hexane<br />
I<br />
7.8<br />
I<br />
7.6<br />
I<br />
7.4<br />
7.2<br />
PPM<br />
5*<br />
23.4<br />
14.5<br />
11.9<br />
11.4<br />
10.3<br />
9.9<br />
9.8<br />
9.2<br />
8.6<br />
7.3<br />
7.0 6.8<br />
7.41<br />
7.32<br />
7.29<br />
7.29<br />
7.27<br />
7.27<br />
7.34<br />
7.29<br />
7.32<br />
7.19<br />
7.20<br />
Figure 1: The ring proton resonances of benzyl alcohol<br />
are shifted upfield and broadened upon binding to<br />
lecithin membranes.<br />
Further discrimination stems from the<br />
motional restrictions imparted by the membrane<br />
environment that is reflected in the<br />
spin-spin relaxation. The ring resonances<br />
corresponding to the free and bound drug are<br />
shown in Figure 1 for a lecithin model membrane<br />
in the L_ state. The resonance of the<br />
bound agent is oroadened due to immobilization<br />
in the membrane. The T2 of the bound<br />
agent is approximately 6 msec, while the T2 of<br />
the free agent is more than three orders of<br />
magnitude greater (11 sec). The different<br />
TABLE 1<br />
(ppm from TMS)<br />
7.41<br />
7.32<br />
7.25<br />
7.24<br />
7.21<br />
7.21<br />
7.30<br />
7.29<br />
7.32<br />
7.16<br />
7.20<br />
7.41<br />
7.23<br />
7.17<br />
7.16<br />
7.12<br />
7.12<br />
7.21<br />
7.29<br />
7.32<br />
7.16<br />
7.13<br />
Bulletin of Magnetic Resonance<br />
S (ppm) avg<br />
7.41<br />
7.29<br />
7.24<br />
7.23<br />
7.20<br />
7.20<br />
7.28<br />
7.29<br />
7.32<br />
7.17<br />
7.18<br />
relaxation properties allow for spectral editing<br />
based on T2 using spin echoes.<br />
The partitioning of benzyl alcohol into<br />
membranes is modest. Consequently, the<br />
lipid to water ratios of the sample must be<br />
large to obtain accurate simulations of the<br />
broad bound resonance, while a sample size<br />
and geometry consistent with high Zeeman<br />
field homogeneity must be maintained. In<br />
practice, to reduce sample demands, an internal<br />
acetate standard was used to determine<br />
the aqueous alcohol concentration in a dilute<br />
membrane suspension. Since the lipid concentration<br />
is known, the intramembrane concentration<br />
is obtained by difference to yield<br />
the partition coefficient. To ensure that the<br />
integrated aqueous resonance is not contaminated<br />
by the broad bound resonance, a<br />
CPMG sequence is used to delay acquisition<br />
by 25 msec in order to filter the broad bounddrug<br />
resonance. This filtering method also<br />
removes the dipolar broadened lipid resonance<br />
to improve baseline definition.<br />
Since the colligative thermodynamic framework<br />
equates the action of solute and temperature<br />
through entropy, precise temperature<br />
regulation is required. The aqueous benzyl<br />
alcohol resonance exhibits a temperature<br />
dependent chemical shift. In order to maintain<br />
a consistently reproducible temperature,<br />
we have taken advantage of this temperature<br />
dependence. The chemical shift differences<br />
between the HOD and free benzyl alcohol<br />
resonances as a function of temperature is<br />
shown in Figure 2.
Vol. 14, No. 1-4<br />
Q.<br />
0.<br />
UI<br />
o<br />
cc<br />
UI<br />
u.<br />
u.<br />
a<br />
u.<br />
X<br />
in<br />
2.9<br />
2.8<br />
2.7<br />
2.6<br />
2.5<br />
2.4<br />
2 ^<br />
-<br />
-<br />
-<br />
c<br />
i 1<br />
1 , 1<br />
) 10 20<br />
Figure 2: The chemical shift difference between t<br />
benzyl alcohol ring proton resonance and the HOD resonance<br />
shows a temperature dependence.<br />
1<br />
1<br />
30<br />
TEMPERATURE (C)<br />
-<br />
_<br />
-<br />
_<br />
1<br />
40 5<br />
ANALYSIS OF PARTITIONING: The<br />
degree of anesthetic partitioning into a membrane<br />
system is sensitive to and characteristic<br />
of the state of the lipid assembly. The equilibrium<br />
constant, K , is deduced from the<br />
partitioning changes characteristic of the<br />
interchange between membrane states. The<br />
temperature dependence of the partitioning<br />
exhibits the following functional form for a<br />
state change between two membrane structures.<br />
[3]<br />
KP =<br />
C =<br />
expC(T-Tm)<br />
AHvH<br />
„<br />
RTT m<br />
(2)<br />
(3)<br />
The partition coefficient for the membrane<br />
states a and ft are Kp« and Kp", respectively.<br />
These partition coefficients are not necessarily<br />
constant and may be altered to include a<br />
temperature dependence. The total partition<br />
coefficient is Kp. The midpoint temperature<br />
is Tm. A fit of the experimental data to this<br />
function yields partition coefficients for each<br />
phase, the midpoint temperature, and the<br />
van't Hoff enthalpy (AHvH), a measure of the<br />
cooperativity of the equilibrium.<br />
295<br />
TEMPERATURE (K)<br />
Figure 3: The molal partition coefficient of benzyl alcohol<br />
into multilamellar DPPC membranes is shown as a<br />
function of temperature for two concentrations of benzyl<br />
alcohol. The fit corresponds to the theoretical multiparameter<br />
least-squares analysis described in the text. The<br />
derivative of the fit to the data is shown offset below.<br />
The percent mole fraction intramembrane benzyl alcohol<br />
concentrations at the L<br />
p,<br />
p p equilibrium<br />
midpoint are as follows: Panel A: Lp , = 0.23%, Pp ' =<br />
1.1%; Panel B: 2.2%, 16.8%; The mole fraction benzyl<br />
alcohol concentrations at the P^/ -* La equilibrium<br />
midpoint are as follows: Panel A: Pp, = 0.84%, La=<br />
2.1%; Panel B: 11.7%, 24.2%; For comparative purposes,<br />
general anesthetic intramembrane concentrations<br />
are considered less than 5 mole fraction percent.<br />
The analytical framework presented is not<br />
specific to the partitioning analysis. It is<br />
broadly applicable to any technique in which<br />
the observable is characteristic of each state.<br />
ALCOHOL ACTION IN MODEL<br />
MEMBRANES: The lecithin membrane,<br />
DPPC (l,2-dipalmitoyl-.stt-glycero-3-phosphocholine),<br />
adopts three well-studied structures<br />
or phases, a gel-structure (L«,), a ripplestructure<br />
(Pp,), and a fluid bilayer-structure<br />
(La). [9] Since the interchange between these<br />
three membrane structures is driven by
296<br />
entropy, changes in solute, temperature and<br />
pressure alter the energetic balance to favor a<br />
given structure. The Lp, •* P^/ equilibrium<br />
(pretransition) exhibits an equilibrium midpoint<br />
temperature determined by calorimetry<br />
as 34.8°C. [9] This change in state is accompanied<br />
by a small change in thermal entropy<br />
(12.5 J mol" 1 K- 1 ). The P^, -* La (main transition)<br />
exhibits an equilibrium midpoint temperature<br />
determined by calorimetry as 41.0°C.<br />
This change in state is accompanied by a relatively<br />
large change in thermal entropy (85.6 J<br />
mol' 1 K*i). [9] The large difference between<br />
the thermal entropy changes associated with<br />
these two equilibria provides a simple system<br />
in which to test the predictions of the colligative<br />
thermodynamic framework, that solute<br />
action occurs through entropy and that equilibria<br />
characterized by a small thermal<br />
entropy change should be most susceptible to<br />
perturbation.<br />
The temperature dependence of benzyl<br />
alcohol partitioning at two substantially different<br />
alcohol concentrations is shown in<br />
Figure 3. Figure 3A corresponds to benzyl<br />
alcohol concentrations below that required for<br />
general anesthesia; whereas, the concentration<br />
in Figure 3B is near that required for<br />
local anesthesia. The partition coefficients<br />
obtained by the NMR method are in excellent<br />
agreement with direct radiolabel measures.<br />
[10] Two discontinuities correspond to the<br />
two membrane equilibria. It is these changes<br />
in partitioning that provide the configurational<br />
entropy by which solutes perturb equilibria.<br />
The low entropy Le, -+ Ppr equilibrium<br />
exhibits greater sensitivity to the alcohol<br />
than the high entropy P^, -*• La, as qualitatively<br />
predicted by the thermodynamic model.<br />
The quantitative test for the model is<br />
shown in Figure 4. The partitioning method<br />
provides intramembrane solute concentrations,<br />
which, in turn, provide the magnitude of<br />
the configurational entropy imparted to each<br />
membrane structure. This contribution<br />
lowers the free energy of each state according<br />
to the magnitude of the partitioning, and<br />
thereby alters the difference in free energy<br />
and shifts the equilibrium. The experimental<br />
points are in good agreement with the theoretical<br />
predictions (represented by the lines)<br />
at dilute alcohol concentrations for which the<br />
thermodynamic treatment is derived and<br />
which corresponds to pharmacological levels<br />
Bulletin of Magnetic Resonance<br />
1 10<br />
INTRAMEMBRANE HOLE FRACTION U) DIFFERENCE<br />
Figure 4: The dependence of the equilibrium midpoint<br />
temperature of DPPC on the presence of benzyl alcohol.<br />
The benzyl alcohol intramembrane concentration difference<br />
between the initial and final states at the equilibrium<br />
midpoint is shown. Data are presented for the low<br />
entropy Lp, -+ Fg, (pretransition; filled circles) and<br />
the high entropy Fp, -*• La (main transition; open circles)<br />
equilibria. The coUigative thermodynamic predictions<br />
(eq. 1) are represented by the lines. The solid<br />
portions of the lines designate the average<br />
intramembrane concentrations at the midpoint which<br />
correspond to the range of pharmacological relevance<br />
for general anesthesia.<br />
Particularly striking is the dramatic contrast<br />
in benzyl alcohol sensitivity exhibited by these<br />
two equilibria. At average intramembrane<br />
concentrations of 5 m.f.%, the low entropy<br />
equilibrmm is perturbed by approximately<br />
12°C, whereas the high entropy equilibrium is<br />
perturbed by approximately 1°C. Not only<br />
does this observation support the predictions<br />
of the thermodynamic model, but it demonstrates<br />
that remarkably low intramembrane<br />
concentrations of nonspecific solutes can precipitate<br />
quite substantial effects upon membrane<br />
structure.<br />
ALCOHOL ACTION IN LIPOSOMES<br />
MADE FROM RATS CHRONICALLY<br />
EXPOSED TO ANESTHETICS: Rat liver<br />
microsomes obtained from rats exposed to<br />
nitrous oxide or fed ethanol were isolated and<br />
liposcmes formed from the extracted phospholipids.<br />
Shown in Figure 5 are representative<br />
benzyl alcohol partitioning traces for the<br />
liposomes prepared from the ethanol-fed and<br />
control animals. The partition coefficient into
Vol. 14, No. 1-4<br />
UJ<br />
FFIC]<br />
UJ<br />
o<br />
t i<br />
TION (<br />
IRTI<br />
«*.<br />
Q.<br />
40LAL<br />
40<br />
35<br />
30<br />
25<br />
• •! 1<br />
• •<br />
-i—' 1 ' r~ —i 1 1 r—<br />
\mr •<br />
^ T<br />
1 , 1 , 1<br />
i . \<br />
20 30 40 50 60 70<br />
TEMPERATURE<br />
Figure 5: Benzyl alcohol partitioning traces are shown<br />
for liposomes made from rat-liver microsomal-phospholipids.<br />
The samples represented in the lower trace<br />
(filled triangles) were prepared from chronically<br />
ethanol-fed rats. The samples represented in the upper<br />
trace (filled circles) were prepared from their pair-fed<br />
littermates.<br />
the control samples is larger than the treated<br />
samples. This difference in partitioning is<br />
characteristic of 'membrane tolerance'. [2] A<br />
structural equilibrium is apparent in the control<br />
samples near 37°C that is lacking in the<br />
samples obtained from treated animals. Similar<br />
results are obtained from the chronic<br />
nitrous oxide paradigm. These results evidence<br />
an adaptive response to the chronic<br />
presence of anesthetic agents that results in<br />
altered domain structure in the reconstituted<br />
system. Similarly, structural lipid domains are<br />
predicted at the anesthetic locus in our thermodynamic<br />
reformulation of the Meyer-<br />
Overton hypothesis.<br />
CONCLUSIONS: Alcohols and anesthetics<br />
act through the entropy imparted by partitioning<br />
to modify membrane architecture.<br />
Analysis of anesthetic action requires simultaneous<br />
measures of solute partitioning and<br />
membrane structure over a wide range of<br />
-<br />
-<br />
297<br />
solute concentrations. NMR partitioning<br />
methods offer unique advantages in such<br />
inquiry since the solute can serve the multifaceted<br />
role of perturbant, reporter of membrane<br />
perturbations, and reporter of solute<br />
partitioning.<br />
METHODS: Spectra were obtained on a Bruker 8.5T<br />
AM spectrometer operating at 360 MHz with deuterium<br />
lock. Lipids were dried under N2, evacuated (
7. Smith, R.L. and Oldfield, E. Science 225,<br />
280-288 (1984).<br />
8. Taraschi, T.F., Lee, Y.-C, Janes, N., and<br />
Rubin, E. Annals New York Acad. Sci.<br />
625,698-706(1991).<br />
9. Chen, S.C. and Sturtevant, J.M.<br />
Biochemistry 20. 713-718 (1981V<br />
10. Colley, CM. and Metcalfe, J.C. FEBS<br />
Letters 24,241-246 (1972).<br />
11. Ellingson, J.S., Janes, R, Taraschi, T.F.,<br />
and Rubin, E. Biochim. Biophys. Acta<br />
1062,199-205 (1991).<br />
Bulletin of Magnetic Resonance
Vol. 14, No. 1-4<br />
ABSTRACT<br />
WEAK MOLECULAR INTERACTIONS: NMR SPECTROSCOPY<br />
OF ORIENTED MOLECULES<br />
C.L. Khetrapal<br />
National Institutes of Health, Bethesda, MI)-20892, USA<br />
"Indian Institute of Science, Bangalore 560 012, India<br />
NMR spectroscopy of oriented<br />
molecules is employed to study<br />
weak molecular interactions. The<br />
information is obtained from the<br />
changes in the molecular order<br />
and the structure as a result of<br />
the complexation.<br />
Specific results on the pi<br />
and charge transfer complexes<br />
formed by iodine, chloroform, and<br />
silver nitrate with nitrogen<br />
heterocycles, aromatic systems<br />
and acetonitrile are reported. A<br />
method for the determination of<br />
the order parameter and the<br />
structure of the 'complexed'<br />
species is presented and its<br />
utility demonstrated. The use of<br />
mixture of liquid crystals of<br />
opposite diamagnetic anisotropies<br />
to investigate the extraordinary<br />
symmetry of Buckminster<br />
Fullerene , Cgo> 1S illustrated.<br />
INTRODUCTION<br />
Application of NMR to<br />
weak molecular interactions<br />
study<br />
from<br />
changes in chemical shifts,<br />
indirect spin-spin coupling<br />
constants and line width is quite<br />
well known and has been employed<br />
since the early days of NMR.<br />
However, the use of direct<br />
dipolar couplings/order parameters<br />
derived from the spectra<br />
of molecules oriented in thermotropic<br />
liquid crystals is relatively<br />
new (1) and provides more<br />
quantitative results since<br />
changes in molecular structure if<br />
any as a result of complexation<br />
* Address for Correspondence<br />
can also be determined directly<br />
and used for such a purpose.<br />
Results on the pi and the<br />
charge transfer complexes formed<br />
by chloroform, iodine, silver<br />
nitrate with aromatic systems,<br />
nitrogen heterocycles and acetonitrile<br />
are described in this<br />
communication. A study of<br />
Buckminster Fullerene (C, )<br />
oriented in mixture of liquid<br />
crystals of opposite diamagnetic<br />
anisotropies is also reported in<br />
order to establish the extraordinary<br />
symmetry of the molecule<br />
indicating, thereby, negligible<br />
solvent-solute interactions.<br />
2. METHODOLODY<br />
Information on weak molecular<br />
complexes has been obtained<br />
from the changes in the degree of<br />
order as well as the molecular<br />
structure produced as a result of<br />
the complex formation.<br />
The degree of order of a<br />
molecule dissolved in a "hematic<br />
liquid crystal usually decreases<br />
with the increase of temperature<br />
or concentration. Any abnormal<br />
change in the degree of order is<br />
attributed to the formation of<br />
the complex(es). Recently, we<br />
have investigated metal ion<br />
ligand interactions between monovalent<br />
metal ions such as Li+ and<br />
Ag+ in LiCIO , LiBF and AgNC<br />
with ligands like aoetonitri1e?<br />
dimethyl sulphoxide, pyridine and<br />
acetone in thermotropic liquid<br />
crystals (2-4). Interactions of<br />
iodine (and bromine) with<br />
299
300 Bulletin of Magnetic Resonance<br />
pyridine (4-6), pyrimidine (7)<br />
and quinazoline (8) have also<br />
been investigated. Work on chloroform-benzene<br />
pi complexes has<br />
also been undertaken. We have<br />
also employed the use of mixture<br />
of liquid crystals of opposite<br />
diamagnetic anisotropies in order<br />
to find out if any detectable<br />
distortions in the spherical symmetry<br />
are present in Buckminster<br />
Fullerene , C 59, as result of<br />
solvent-solute interactions (9).<br />
Some such results are reported<br />
below briefly.<br />
LiCIO •-! igand<br />
1igands used in interactions:<br />
(a)<br />
The 1 igands "'used in such studies<br />
are dimethyIsulphoxide, acetonitrile<br />
and pyridine. The liquid<br />
crystals employed are Schiff<br />
bases such as N-(p-methoxy (or<br />
ethoxy) benzylidene)-p-n-butyl<br />
aniline (MBBA or EBBA) and their<br />
deuterated (-d2) analogue with<br />
deuterium substituted at position<br />
ortho to the -N= group in the<br />
butyl aniline ring. The results<br />
indicate ' the formation of two<br />
types of complexes in these cases<br />
with one having "isotropic-1ike"<br />
structure. The "isotropic-1ike"<br />
complex may be of the type ML4<br />
where M is the metal ion and L is<br />
the 1igand and the other could be<br />
the one containing different<br />
ratios<br />
gands.<br />
of the metal ions to li-<br />
(b) LiBF -Ligand solutions: The<br />
1igand used in such studies was<br />
acetonitrile. The liquid crystals<br />
employed were trans, trans-4-npropyl(1,1'-bicyclohexyl)-4'carbonitrile<br />
(ZLI-1184) and<br />
trans-4-pentyl-1-(4-cyanophenyl)<br />
cyclohexane (S-1114). The results<br />
indicate the formation of tetrahedral<br />
1ithium-acetonitrile complexes<br />
and the coordination of<br />
lithium from LiBF^ and LiClC^<br />
with the cyano group of ZLI-1184.<br />
No evidence was found to support<br />
either lithium ion complexation<br />
with S-1114 or the tetrafluorobo-<br />
rate moiety in these systems.<br />
(c) AgJNO3-ligand complexes: In<br />
this case AgNC^-CI^CN (CD3CN)<br />
complexes have been investigated<br />
in two different liquid crystals,<br />
namely, ZLI-1167 (a ternary eutectic<br />
mixture of propyl, pentyl<br />
and heptyl bicyclohexyl carbonitrile)<br />
and S-1114 from both the<br />
proton and the deuteron NMR<br />
studies. The results indicate the<br />
formation of different types of<br />
AgNO -CHoCN complexes, i.e., ML4,<br />
ML 3.?..etc.<br />
(d) Benzene-chloroform complex:<br />
It has been studied from the<br />
reduction of molecular order of<br />
benzene upon addition of iodine.<br />
From the shape considerations,<br />
the order parameter of such a<br />
complex should be either opposite<br />
in sign to or smaller in magnitude<br />
than that of benzene under<br />
comparable conditions. The observed<br />
spectrum of benzene in<br />
ZLI-1167 containing iodine which<br />
arises from the average orientation<br />
of the "complexed" and the<br />
"free" benzene, therefore, has an<br />
order parameter which is smaller<br />
in magnitude than that of the'<br />
"uncomplexed" benzene. The change<br />
of the chemical shift of the<br />
benzene carbon is also consistent<br />
with this observation.<br />
(e) Iodine(bromine)-pvridine<br />
charge transfer complex: Such<br />
complexes using NMR spectroscopy<br />
of oriented systems were first<br />
studied in 1973 and 1983 (5, 6)<br />
from the drastic change of the<br />
order parameter of the C2~axis of<br />
symmetry of pyridine. We have<br />
however, detected the formation<br />
of the "inner" complexes<br />
[(PYlfl"] as well as those of<br />
the type PY.I2 unambiguously. The<br />
order parameters of the<br />
"complexed" species have also<br />
been determined (4).
Vol. 14, No. 1-4<br />
(f) Extraordi nary symmetry<br />
Buckmi nster Fullerene. C,_<br />
of<br />
in<br />
REFERENCES<br />
nematic solutions: The "^ C-NMR<br />
spectrum of C6o was studied in a<br />
mixture of nematic liquid<br />
crystals of opposite diamagnetic 1.<br />
anisotropies. The method makes<br />
use of the change in the anisotropic<br />
parameters resulting from<br />
the switching of the order parameters<br />
at the critical point in<br />
the mixture; the change is by a 2.<br />
factor of 2 or -1/2 depending<br />
upon the direction of approach of<br />
the critical point (10). By an<br />
appropriate adjustment of the<br />
concentration and temperature, it<br />
is possible to observe both the 3.<br />
types of orientations to coexist.<br />
Even in tetrahedral molecules<br />
such as methane and tetramethysilane,<br />
the coexistence of 4.<br />
the two spectra has been observed<br />
(11). On the other hand, the<br />
observation of a single line at<br />
the same position as in the isotropic<br />
phase at the critical 5.<br />
point where the coexistence of<br />
the two spectra in methane or<br />
tetramethylsi 1ane is observed,<br />
indicates the absence of any 6.<br />
detectable distortions in the<br />
molecule. This has actually been<br />
observed in a solution of C6Q in<br />
a 1:1 mixture of liquid crystals<br />
ZLI-1167 and S-1114 containing 7.<br />
tetramethylsilane, at 332.4 K.<br />
The spectrum clearly shows the<br />
coexistence of two spectra for<br />
tetramethylsi 1ane whereas for the 8.<br />
C60 only a single line at its<br />
isotropic position (140.37 ppm<br />
with respect to TMS) is observed.<br />
The- results, therefore, establish 9.<br />
that there are no detectable<br />
distortions from spherical symmetry<br />
in C in the nematic solvents.<br />
TcP® our knowledge this is<br />
the first molecule where no detectable<br />
distortation from spherical<br />
symmetry have been observed.<br />
It should, therefore, serve<br />
as an ideal reference for the 10<br />
study of the chemical shift<br />
sotropy.ani-<br />
301<br />
P. Diehl, H.R. Wasser, G.A.<br />
Nagana Gowda, N. Suryaprakash<br />
and C.L. Khetrapal,<br />
Chem. Phys. Lett. 159. 199<br />
(1989)<br />
G.A. Nagana Gowda, N. Suryaprakash,<br />
R.G. Weiss, C.L.<br />
Khetrapal and P. Diehl,<br />
Magn. Reson. Chem. .28, 642<br />
(1990)<br />
G.A. Nagana Gowda, R.G.<br />
Weiss and C.L. Khetrapal,<br />
Liq. Cryst. 10, 659 (1991)<br />
C.L. Khetrapal, G.A. Nagana<br />
Gowda and N. Suryaprakash,<br />
Spect. Chim. Acta. (in the<br />
press)<br />
C.A. Veracini, M. Longeri<br />
and P.L. Barili, Chem. Phys.<br />
Lett. 19., 592 (1973)<br />
D. Catalano, C.A. Veracini,<br />
P.L. Barili and M. Longeri,<br />
J. Chem. Soc. Perk Trans II,<br />
171, 1983<br />
N. Suryaprakash, R. Ugolini<br />
and P. Diehl, Magn. Reson.<br />
Chem. 29., 1024 (1991 )<br />
N. Suryaprakash, C.L.<br />
Khetrapal and P. Diehl (To<br />
be published)<br />
C.N.R. Rao, T. Pradeep, R.<br />
Seshadri, R. Nagarajan, V.N.<br />
Murthy, G.N. Subbanna, F.<br />
D'Souza, V. Krishnan, G.A.<br />
Nagana Gowda, N. Suryaparakash,<br />
C.L. Khetrapal and<br />
S.V. Bhat, Ind. J. Chem. 31.<br />
F5 (1992).<br />
C.L. Khetrapal and A.C.<br />
Kunwar, Chem. Phys. Lett.<br />
.82, 170 ( 1981 ) .
302 Bulletin of Magnetic Resonance<br />
11. C.L. Khetrapal, A.C. Kunwar<br />
and M.R. Lakshminarayana,<br />
Mol. Cryst. Liq. Cryst. 111.<br />
189 (1984).
Vol. 14, No. 1-4 303<br />
Introduction:<br />
Structural Studies of Collagen by Solid State NMR.<br />
Richard J. Wittebort and Anne Marie Clark<br />
University of Louisville, Department of Chemistry,<br />
Louisville, KY 40292 USA<br />
Collagen is one of the most abundant<br />
proteins nature and has the same structure and<br />
virtually the same composition for diverse<br />
species. It provides the organic matrix for teeth<br />
and bones, and gives tendons and blood<br />
vessels their strength. Collagen has three<br />
polypeptide chains, two al chains and one a2<br />
chain, coiled together in a 3-1 helix. Each<br />
chain is formed from a repeating (Gly-X-Y)<br />
triad with Gly always in the first position. The<br />
other two positions can have any amino acid<br />
but frequently (about 30%) have proline (Pro)<br />
and hydroxyproline (Hyp). Hyp always<br />
occurs in position 3 while Pro usually occurs in<br />
position 2 and only rarely in position 3 in adult<br />
animals. The principle difference between the<br />
types of chains is their direction of orientation.<br />
Several models have been developed to<br />
explain the three dimensional structure of<br />
collagen. From steric considerations, it is<br />
thought that the collagen triple helix has Gly<br />
alpha hydrogens pointing inside. Despite<br />
numerous attempts to solve the X-ray structure,<br />
collagen's lack of long range order has made it<br />
impossible to determine a unique structure from<br />
diffraction methods. 1 " 5<br />
Collagen is a uniaxially ordered fiber<br />
with helical symmetry about the fiber axis. In<br />
order to do structural studies by solid state<br />
NMR, the fiber axis must be aligned with the<br />
magnetic field. Opella used solid state<br />
techniques to determine the three dimensional<br />
structure of bactreiophage fd coat protein. 6<br />
This experiment requires at least one direction<br />
of orientation and the molecular sites of interest<br />
must be immobilized and uniformly oriented<br />
with respect to the magnetic field. In single<br />
crystals, any arbitrary sample orientation can be<br />
studied; however, for uniaxially oriented -<br />
samples the direction of orientation must be<br />
along the applied field to obtain single crystal-<br />
like spectra. The bacteriophages are well suited<br />
for this type of experiment. They<br />
spontaneously align with the magnetic field and<br />
their large size and rod-like shape immobilize
304<br />
the protein subunits. The alpha helical<br />
structure determined agrees well with X-ray<br />
structures of alpha helical proteins.<br />
The three dimensional structure of<br />
polypeptides can be described as a series of<br />
connected peptide planes. By selectively<br />
labelling specific sites and determining their<br />
orientation, structural constraints will be added<br />
to the three dimensional structure of collagen.<br />
The frequency of resonance lines in<br />
solid state NMR spectra depend on the<br />
orientation of the local molecular environment<br />
relative to the magnetic field Ho. The observed<br />
splitting can be used to determine the angle a<br />
bond makes with the magnetic field by the<br />
following equation, assuming an axially<br />
symmetric interaction:<br />
Av = vn(3cos 2 6-1) Eqn. 1<br />
Av is the observed splitting, 0 is the angle<br />
between the bond and the applied field and vn<br />
is the quadrupolar coupling constant. It has<br />
been found that in order to determine the<br />
orientation of a peptide plane at least two labels<br />
per plane must be examined.<br />
Experimental:<br />
The solenoidal coil geometry typically<br />
used in solid state NMR probes is<br />
Bulletin of Magnetic Resonance<br />
unsatisfactory since a bundle of fibers is only<br />
conveniently placed in the coil in the<br />
perpendicular orientation. Attempts were made<br />
to wrap tendons around several small slides<br />
and stack them in the sample holder but<br />
satisfactory alignment was not obtained. A<br />
different probe design is called for by this<br />
experiment. One probe design is based on a<br />
small flat coil geometry suggested by Opella. 7<br />
Small, flat frames were made for us upon<br />
which to wind the coil. A series of rat-tail<br />
tendons were tied together and wrapped around<br />
a 1 x 1 cm polycarbonate card. The card was<br />
then placed in the coil frame and the coil wound<br />
around it. This arrangement consistently gives<br />
a 90 degree pulse width of 2.7 JIS at 1 kW<br />
power. This design has the advantage of<br />
allowing the fiber axis to be placed at any angle<br />
relative to Ho. Proper turn spacing for good<br />
RF field homogeneity and consistent<br />
inductance is insured by using a frame with set<br />
holes. This also allows the sample to be kept<br />
in the center of the coil, away from the edges<br />
where the field homogeneity is poor. We<br />
obtained satisfactory results with this probe.<br />
We were, however, limited to a small sample
Vol. 14, No. 1-4 305<br />
size due to the polycarbonate cards and<br />
resultant poor filling factor. We had to retain<br />
the cards to keep the fibers aligned; otherwise,<br />
they shrink during changes in humidity and<br />
temperature.<br />
We are principally interested in<br />
structural information obtainable from samples<br />
with the fiber axis along Ho, that is, the parallel<br />
orientation. From that premise, we designed a<br />
probe using a Helmhotz coil similar to liquid<br />
NMR probes. The probe was constructed<br />
based on our circuit for a double resonance<br />
design with a 0.5 x 1 cm saddle. The probe<br />
has a 90 pulse width of 3.2 microseconds with<br />
400 watts power at the 2 H frequency<br />
(38.8 MHz). This probe is capable of holding<br />
ten times the sample as the flat probe and has<br />
provisions for stretching the sample.<br />
We have labelled selected sites in the<br />
collagen molecule by incorporating deuterated<br />
amino acid into the rat tail tendon. Torchia<br />
incorporated labelled glycine into 1/3 of the Gly<br />
positions in rat tail tendon by injecting it into<br />
rats. 8 We slightly modified his injection<br />
scheme in order to label selected amino acids in<br />
rat tail tendon. Given the low natural<br />
abundance of deuterium, we were confident<br />
that the only signal we would see would be<br />
from our covalently bound label.<br />
The first covalently labelled position we<br />
tried was alpha-d-Pro. The solutions for<br />
injection were 1.3 M in the amino acid and<br />
were neutralized with NaOH. Ten rat pups<br />
were injected once a day for 21 days and then<br />
sacrificed. The rat tail tendon was extracted for<br />
use in our experiments. The degree of<br />
incorporation of d-Pro in the tendon was<br />
determined by GC/MS and found to be 5.4%<br />
which agrees well with our NMR results. No<br />
incorporation of deuterium in any amino acids<br />
other than Pro and Hyp was seen. The<br />
oriented fibers were run on the Helmholtz coil<br />
probe. A splitting of Av=l 17 kHz was<br />
observed and the angle of the C-D bond relative<br />
to the magnetic field determined.<br />
The glycine amides were labelled by<br />
exchange. Samples were heated at 40°C and at<br />
constant humidity to completely exchange the<br />
labile H's. At 78% relative humidity, exchange<br />
reaches a constant level after 24 hours, at 38%<br />
relative humidity three days are required to<br />
reach a constant level of exchange. The<br />
samples were cooled to room temperature at<br />
constant humidity for an hour and then back
306<br />
exchanged in liquid H2O. Samples that were<br />
not heated to label the amide positions showed<br />
an extremely rapid loss of the solid echo peak.<br />
The heated sample, after back-exchange, had<br />
10% of the solid echo remaining, indicating the<br />
exchanged sample has a significant amount of<br />
bound water. The doublet splitting of 155 kHz<br />
shows that the hydrogen bonded amides are in<br />
fact nearly perpendicular to the fiber axis in<br />
agreement with the predictions from the X-ray<br />
structure. Recent packing studies predict the<br />
collagen molecule is tilted 4 to 5 degrees off the<br />
fiber axis.<br />
Results and Conclusions:<br />
Bulletin of Magnetic Resonance<br />
The asymmetric lineshape of the back<br />
exchanged sample indicates a distribution of<br />
orientation about 6=89 degrees. The spectral<br />
simulation for a Gaussian distribution of<br />
orientations P(6) a exp(-6 2 /2a) with 6=90 and<br />
o=17 matches our experimental data quite well.<br />
We have successfully labelled selected<br />
positions and determined their orientation<br />
relative to the fiber axis. By carefully choosing<br />
our labels, we will be able to determine the<br />
orientation of various peptide planes in collagen<br />
and gradually build the three-dimensional<br />
structure.<br />
Table I. Angle between X-D bond and the fiber axis comparison of solid state NMR results and<br />
various models derived from X-ray crystallography data.<br />
Deuteron Experiment Fraser's 4 Ramachandran's 1 Ramachandran's 2 Yonath's 5<br />
Rich-Crick bridging water two bonded<br />
ProCa-D 90<br />
GlyN-D 89<br />
References:<br />
81.3<br />
89.4<br />
1 Ramachandran, G. N. and Kartha, G.,<br />
Nature, 1955, 776, 593.<br />
2 Ramachandran, G. N. and<br />
Chandrasekaran, R., Biopolymers, 1968,<br />
6,1649.<br />
3 Rich, A. and Crick, F. H. C, Nature,<br />
1955,276,915.<br />
4 Fraser, R. D. B., /. Mol Biol., 1979,<br />
129, 463.<br />
73.8<br />
78.7<br />
73.1<br />
78.4 83.1<br />
5 Yonath, A. and Traub, W., J.Mol. Biol,<br />
1969,43,461.<br />
6 Opella, S. J., Quart. Rev. Biophys.,<br />
1987,79,7.<br />
7 Bechinger, B. and Opella, S. J., /. Magn.<br />
Reson., 1991, 95, 585.<br />
8 Jelinski, L. W. and Torchia, D. A., /.<br />
Mol. Biol, 1979,133, 45.
Vol. 14, No. 1-4 307<br />
Calender of Forthcoming<br />
Conferences in Magnetic<br />
Resonance<br />
March 8-14, 1993<br />
1993 Keystone Symposia on Molecular & Cellular<br />
Biology, Taos, New Mexico, USA<br />
Frontiers of NMR in Molecular Biology - III;<br />
Organizers: Thomas L. James, Stephen W. Fesik<br />
and Peter E. Wright.<br />
Information:<br />
Keystone Symposia<br />
Drawer 1630<br />
Silverthorne, CO 80498<br />
Phone: 303-262-1230<br />
March 14-18, 1993<br />
Experimental Nuclear Magnetic Resonance Conference,<br />
The Adam's Mark Hotel, St. Louis, Missouri,<br />
USA<br />
There will be sessions covering high resolution<br />
and multi-dimensional NMR in liquids, materials<br />
imaging, biological imaging, hardware, NMR<br />
in solids, calculations and data processing, and that<br />
old standby, miscellaneous. The deadline for submitting<br />
poster abstracts will be December 30,<br />
1992.<br />
Contact:<br />
ENC<br />
815 Don Gaspar<br />
Santa Fe, New Mexico 87501<br />
Phone: 505-989-4573<br />
FAX: 505-989-5073<br />
April 1993<br />
High Resolution NMR Spectroscopy (a residential<br />
school), University of Sheffield, England<br />
For information contact:<br />
Ms. L. Hart<br />
The Royal Society of Chemistry<br />
Burlington House<br />
Piccadilly, London W1V 0BN<br />
England<br />
Tel: 071-437-8656<br />
September 6-10, 1993<br />
Second International Conference on Magnetic<br />
Resonance Microscopy, Heidelberg, Germany<br />
The program includes plenary lectures, as well<br />
as oral and poster contributions selected from the<br />
submitted papers. The papers will be judged solely<br />
on the basis of an abstract. Preregistration deadline:<br />
April 30, 1993.<br />
Organization and Program: Winfried Kuhn<br />
(St. Ingbert), Bernhard Blumich (Mainz).<br />
International Advisory Board: J. L. Ackerman<br />
(Charlestown), L. J. Berliner (Columbus),<br />
P. T. Callaghan (Palmerston), W. Edelstein<br />
(Schenectady), A. N. Garroway (Washington),<br />
A. Haase (Wiirzburg), W. E. Hull (Heidelberg),<br />
L. W. Jelinski (Ithaca), J. L. Koenig (Cleveland),<br />
G. A. Johnson (Durham), P. Jonson (London),<br />
P. C. Lauterbur (Urbana), P. Mansfield (Nottingham),<br />
B. Maraviglia (Rome), G. D. Mateescu<br />
(Cleveland), J. M. Pope (Kensington), V. Sarafis<br />
(Richmond), S. Sarkar (King of Prussia)<br />
For further information please write to:<br />
Dr. Winfried Kuhn<br />
Fraunhofer Institute<br />
Ensheimer Str. 48<br />
DW-6670 St. Ingbert, Germany<br />
phone: +49-6894-89738<br />
FAX: +49-6894-89750<br />
or<br />
Dr. Bernhard Blumich<br />
Max-Planck Institute for Polymer Research<br />
Postfach 3148<br />
D-6500 Mainz, Germany<br />
phone: +49-6131-379125<br />
FAX: +49-6131-379100<br />
The editor would be pleased to receive<br />
notices of future meetings in the field of<br />
magnetic resonance so that they could be<br />
recorded in this column.
308<br />
Recent Magnetic Resonance Books<br />
1 Magnetic Resonance Spectroscopy in Biology<br />
and Medicine (1992). Edited by J. D. De Certaines,<br />
W. M. M. J. Bovee and F. Podo. Contents: Presents<br />
the experimental and basic aspects of functional and<br />
pathological tissue characterization of MRS. A balance<br />
is drawn between the basic science, practical<br />
technologies and biomedical applications. Covers<br />
recent developments in the field: localization, 2D<br />
NMR, spectroscopic imaging, data quantification<br />
and quality assessment, as well as the basic principles<br />
of magnetic resonance spectroscopy. Pergamon<br />
Press, ISBN 0-08-0410170 (flexicover) $70.00; ISBN<br />
0-08-0410189 (hardcover) $170.00.<br />
1 In Vivo Magnetic Resonance Spectroscopy I.<br />
Probeheads and Radiofrequency Pulses, Spectrum<br />
Analysis (1992). Edited by M. Rudin, Springer, 345<br />
pp. ISBN 0-387-54547-6 (hardcover) $119.00.<br />
l In Vivo Magnetic Resonance Spectroscopy II.<br />
Localization and Spectral Editing (1992). Edited by<br />
M. Rudin and J. Seelig, Springer, 368 pp. ISBN<br />
0-387-55022-4 (hardcover) $119.00.<br />
^In Vivo Magnetic Resonance Spectroscopy III.<br />
In Vivo MR Spectroscopy: Potential and Limitations<br />
(1992). Edited by M. Rudin and J. Seelig,<br />
Springer, 293 pp. ISBN 0-387-55029-1 (hardcover)<br />
$98.00.<br />
1 Annual Reports on NMR Spectroscopy. Volume<br />
24 (1992). Contents: Developments in solid state<br />
NMR. Solid state NMR imaging. NMR studies of<br />
interfacial phenomena. NMR measurements of intracellular<br />
ions in living systems. 199Hg NMR parameters.<br />
Applications of NMR methods in coal<br />
research.<br />
1 Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 24 No. 3(1992). Contents: Structural<br />
characterization of noncrystalline solids and<br />
glasses using solid state NMR.<br />
1 Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 24 No. 2 (1992). Contents: 129 Xe<br />
J New additions to the list.<br />
Bulletin of Magnetic Resonance<br />
NMR as a probe for the study of microporous solids:<br />
A critical review. Simulation of 2D NMR spectra for<br />
determination of solution conformations of nucleic<br />
acids.<br />
Progress in Biophysics & Molecular Biology.<br />
Volume 57 No. 1 (1992). Contents: ENDOR and<br />
EPR of metalloproteins. Free energy transduction<br />
in polypeptides and proteins based on inverse temperature<br />
transitions.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 24 No. 1 (1992). Contents: 13 C<br />
NMR spectroscopy of oleanane triterpenoids.<br />
NMR at Very High Field (1991). Guest editor:<br />
J. B. Robert, Springer, 168 pp. ISBN 0-387-52946-2<br />
(hardcover) $79.00.<br />
1 Transition Metal Nuclear Magnetic Resonance<br />
(1991). Edited by P. S. Pregosin. Contents: The<br />
book contains a collection of review articles concerned<br />
with measuring, understanding and using the<br />
nuclear magnetic resonance spectra of the metals of<br />
Groups 3-12. The reader is provided with a view<br />
on how these nuclei are currently being approached,<br />
and what information can be obtained. The authors<br />
have liberally reproduced spectra as well as correlations<br />
relating metal-NMR data to different physical<br />
characteristics of their molecules. 364 pp. ISBN<br />
0-444-88176-X $169.00.<br />
Chemical Reviews. Volume 91 No. 7 (1991).<br />
Contents: Low-temperature solid-state NMR of proteins.<br />
Structure and dynamics of solid polymers<br />
from 2D- and 3D-NMR. NMR under high gas pressure.<br />
Nuclear magnetic resonance at high temperature.<br />
Gas-phase NMR spectroscopy. Selective<br />
excitation in high-resolution NMR. Application of<br />
the linear prediction method to NMR spectroscopy.<br />
High-resolution fluorine-19 magnetic resonance of<br />
solids. NMR determination of enantiomeric purity.<br />
Solid-state NMR studies of molecular sieve<br />
catalysis. Pulsed electron-nuclear double resonance<br />
methodology. Multidimensional NMR and data processing.<br />
One- and two-dimensional high-resolution<br />
solid-state NMR studies of zeolite lattice structures.<br />
Solid-state NMR studies of DNA structure and dynamics.<br />
Spin-lattice relaxation of coupled nuclear
Vol. 14, No. 1-4 309<br />
spins with applications to molecular motion in liquids.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 23 No. 2 (1991). Contents: Solvent<br />
signal suppression in NMR.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 23 No. 3 (1991). Contents: Modern<br />
methods of NMR data processing and data evaluation.<br />
X H NMR magic angle spinning (MAS) studies<br />
of heterogenous catalysis.<br />
NMR - Basic Principles and Progress. Volume<br />
23: Deuterium and Shift Calculation (1991).<br />
Eds.: P. Diehl, E. Fluck, H. Giinther, R. Kosfeld,<br />
J. Seelig. Contents: M.L. Martin, G.J. Martin,<br />
Nantes, France: Deuterium NMR in the Study of<br />
Site-Specific Natural Isotope Fractionation (SNIF-<br />
NMR); H.-H. Limbach, Freiburg, FRG: Dynamic<br />
NMR Spectroscopy in the Presence of Kinetic Hydrogen/Deuterium<br />
Isotope Effects; W. Kutzelnigg,<br />
U. Fleischer, M. Schindler, Bochum, FRG: The<br />
IGLO-Method: Ab-initio Calculation and Interpretation<br />
of NMR Chemical Shifts and Magnetic Susceptibilities.<br />
Approx. 270 pp. 92 figs. 45 tabs.<br />
ISBN 3-540-52949-7.<br />
NMR - Basic Principles and Progress. Volume<br />
24: High Pressure NMR (1991). Eds.: P. Diehl,<br />
E. Fluck, H. Giinther, R. Kosfeld, J. Seelig, J.<br />
Jonas, University of Illinois, Urbana, IL (Guest-<br />
Ed.). Contents: D. Brinkmann, Zurich, Switzerland:<br />
Solid-State NMR Studies at High Pressure;<br />
K.O. Prins, Amsterdam, The Netherlands: High<br />
Pressure NMR Investigations of Motion and Phase<br />
Transitions in Molecular Systems; J. Jonas, Urbana,<br />
IL: High Pressure NMR Studies of the Dynamics<br />
in Liquids and Complex Systems; E.W. Lang, H.-<br />
D. Liidemann, Regensburg, FRG: High Pressure<br />
NMR Studies on Water and Aqueous Solutions;<br />
J.W. Akitt, A.E. Merbach, Lausanne, Switzerland:<br />
High Resolution Variable Pressure NMR for Chemical<br />
Kinetics; H. Yamada, Kobe, Japan: Glass Cell<br />
Method for High-Pressure, High-Resolution NMR<br />
Measurements. Applications to the Studies of Pressure<br />
Effects on Molecular Conformation and Structure.<br />
Approx. 270 pp. 148 figs. 28 tabs. ISBN<br />
3-540-52938-1.<br />
NMR - Basic Principles and Progress. Volume<br />
25: NMR at Very High Field (1991). Eds.: P. Diehl,<br />
E. Fluck, H. Giinther, R. Kosfeld, J. Seelig, J.B.<br />
Robert, CNRS, Grenoble,-France (Guest-Ed.). Contents:<br />
R. Freeman, Cambridge, UK, J.B. Robert,<br />
Grenoble, France: A Brief History of High Resolution<br />
NMR; E.W. Bastiaan, C. MacLean, Amsterdam,<br />
The Netherlands: Molecular Orientation<br />
in High-Field High-Resolution NMR; D. Canet,<br />
Vandoeuvre-les-Nancy, France, J.B. Robert, Grenoble,<br />
France: Behaviour of the NMR Relaxation Parameters<br />
at High Fields; D. Marion, Orl ans, France:<br />
Structural Studies of Biomolecules at High Field; U.<br />
Haeberlen, Heidelberg, FRG: Solid State NMR in<br />
High and Very High Magnetic Fields. Approx. 175<br />
pp. 44 figs. 10 tabs. ISBN 3-540-52946-2.<br />
Modern NMR Techniques and Their Application<br />
in Chemistry (Practical Spectroscopy Series Volume<br />
11). Edited by Alexander I. Popov and Klaas Hallenga,<br />
Marcel Dekker, Inc. (1991). ISBN 0-8247-<br />
8332-8<br />
Annual Reports of NMR Spectroscopy. Volume<br />
23 (1991). Contents: NMR studies of isolated spin<br />
pairs in the solid state. The oxidation-state dependence<br />
of transition-metal shieldings. The Cinderella<br />
nuclei. Permutation symmetry in NMR relaxation<br />
and exchange. Nuclear spin relaxation in organic<br />
systems and solutions of macromolecules and aggregations.<br />
NMR of coals and coal products.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 23 No. 1 (1991). Contents:<br />
Nuclear magnetic resonance imaging in the solid<br />
state. Applications of three-and four-dimensional<br />
heteronuclear NMR spectroscopy to protein structure<br />
determination. Angiography and perfusion<br />
measurements by NMR.<br />
EPR Imaging and in vivo EPR (1991). Edited<br />
by Gareth R. Eaton, Sandra S. Eaton, and Keiichi<br />
Ohno, CRC Press, Boca Raton, FL. 320 pages,<br />
$89.95, ISBN: 0-8493-4923-0.<br />
Basic One-and Two-dimensional NMR Spectroscopy<br />
by Horst Friebolin (1991). VCH, New York.<br />
344 pages.
310<br />
Advances in Magnetic and Optical Resonance<br />
Volume 16 (1991). Contents: Laser excitation and<br />
detection of magnetic resonance. Deuterium relaxation<br />
in molecular solids. On the growth of multiple<br />
spin coherences in NMR of solids.<br />
1 Progress in Biophysics & Molecular Biology<br />
Volume 56 No. 1 (1991). Contents: An evaluation<br />
of computational strategies for use in the determination<br />
of protein structure from distance constraints<br />
obtained by nuclear magnetic resonance.<br />
1 Radiospectroscopy of Natural Substances by B.<br />
F. Alekseev, Y. V. Bogachev, V. Z. Drapkin, A. S.<br />
Serdjuk, N. B. Strakhov and S. G. Fedin, Engl. Tr.<br />
Norell Pr., New Jersey, 1991.<br />
1 Electron Paramagnetic Resonance of Exchange<br />
Coupled Systems by A. Bencini and D. Gatteschi,<br />
Springer Verlag, Berlin, 1990.<br />
Modern Pulsed and Continuous Wave Electron<br />
Spin Resonance by L. Kevan and M. K. Bowman<br />
(1990). Wiley, New York.<br />
1 Transition Ion Electron Paramagnetic Resonance<br />
by J. R. Pilbrow, Clarendon Press, Oxford,<br />
1990.<br />
1 Electron Paramagnetic Resonance of Exchange<br />
Coupled Systems by A. Bencini and D. Gattechi<br />
(1990). Springer, 287 pp. ISBN 0-387-50944-5<br />
(hardcover) $83.00.<br />
1 Isotope Effects in NMR Spectroscopy by S.<br />
Berger, J. M. Risley, N. M. Sergeyev and<br />
R. L. Van Etten (1990). Springer, 173 pp. ISBN<br />
0-387-51286-1 (hardcover) $83.00.<br />
17 O NMR Spectroscopy in Organic Chemistry<br />
(1990). Edited by David W. Boykin. This book provides<br />
a comprehensive review of the application of<br />
17 O NMR spectroscopy to organic chemistry. Topics<br />
include the theoretical aspects of chemical shift,<br />
quadrupolar and J coupling; 17 O enrichment; the<br />
effect of steric interactions on 17 O chemical shifts of<br />
functional groups in flexible and rigid systems; the<br />
additions to the list.<br />
Bulletin of Magnetic Resonance<br />
application of 17 O NMR spectroscopy to hydrogen<br />
bonding investigations; mechanistic problems in organic<br />
and bioorganic chemistry; and 17 O NMR spectroscopy<br />
of oxygen monocoordinated to carbon in<br />
alcohols, ethers, and derivatives. CRC Press, Inc.,<br />
Florida. ISBN: 0-8493-4867-6.<br />
Advances in Magnetic and Optical Resonance.<br />
Volume 15 (1990). Contents: Iterative methods in<br />
the design of pulse sequences for NMR excitation.<br />
Electron-nuclear polarization transfer in the nuclear<br />
rotating frame. Multipole NMR. Solid state and solution<br />
NMR of nonclassical transition metal polyhydrides.<br />
Low-frequency magnetic resonance with<br />
a dc SQUID.<br />
Advances in Biophysical Chemistry. Volume<br />
1 (1990). Contents: Stable-isotope-assisted protein<br />
NMR spectroscopy in solution.<br />
31 P and<br />
1<br />
H two-dimensional NMR and NOESY-distance<br />
restrained molecular dynamics methodologies for<br />
defining sequence-specific variations in duplex<br />
oligonucleotides: A comparison of NOESY two-spin<br />
approximation and the relaxation matrix analyses.<br />
NMR study of B- and Z-DNA hairpins of d[(CG)3]<br />
in solution. Molecular dynamics simulations of carbohydrate<br />
molecules. Diversity in the structure of<br />
hemes.<br />
Biological Magnetic Resonance. Volume 9<br />
(1990). Contents: Phosphorus NMR of membranes.<br />
Investigation of ribosomal 5S ribonucleic acid solution<br />
structure and dynamics by means of highresolution<br />
nuclear magnetic resonance spectroscopy.<br />
Structure determination via complete relaxation<br />
matrix analysis (CORMA) of two-dimensional nuclear<br />
overhauser effect spectra: DNA fragments.<br />
Methods of proton resonance assignment for proteins.<br />
Solid-state NMR spectroscopy of proteins.<br />
Methods for suppression of the H2O signal in proton<br />
FT/NMR spectroscopy: A review.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Vol. 25 pt. 5 (1990). Contents: Solid<br />
state NMR techniques for the study of surface phenomena.<br />
A primer on isotopic labeling in NMR investigations<br />
of biopolymers. Vanadium-51 NMR.
Vol. 14, No. 1-4 311<br />
One-dimensional and Two-dimensional NMR<br />
Spectra by Modern Pulse Techniques. Koji Nakanishi.<br />
(1990). University Science Books, Mill Valley,<br />
CA. 234 p.<br />
Annual Reports on NMR Spectroscopy. Volume<br />
22 (1990). Contents: Metal-ion NMR studies of ion<br />
binding. NMR studies of ligand-macromolecule interactions.<br />
Applications of NMR in the analysis of<br />
agrochemicals and pesticides. NMR nuclear shielding<br />
and the electronic structures of macromolecules.<br />
207 Pb-NMR parameters. Nuclear spin relaxation in<br />
diamagnetic fluids part 1. General aspects and inorganic<br />
applications.<br />
Fourier Transforms in NMR, Optical, and Mass<br />
Spectrometry: A User's Handbook. By A. G. Marshall<br />
and F. R. Verdun (Ohio State University). Elsevier:<br />
Amsterdam and New York. 1990. xvi + 450<br />
pp. $107.25. ISBN 0-444-87360-0.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy.<br />
Volume 22, pt. 1 (1990). Contents: Scaling<br />
in one and two dimensional NMR spectroscopy<br />
in liquids. Oligosaccharide conformations: Application<br />
of NMR and energy calculations. Relaxation<br />
matrix analysis of 2D NMR data.<br />
Progress in Magnetic Resonance Spectroscopy.<br />
Volume 22, Part 3 (1990). Contents: NMR parameters<br />
of alkynes. Improved methods for quantitative<br />
spectral analysis of NMR data.<br />
Advances in Magnetic Resonance. Volume 14<br />
(1990). Contents: Measurement of dipole-dipole<br />
cross correlation by triple-quantum filtered twodimensional<br />
exchange spectroscopy. Assessment<br />
and optimization of pulse sequences for homonuclear<br />
isotropic mixing. Spin-1/2 description of spins 3/2.<br />
Optical pumping measurements of nuclear cross relaxation<br />
and electrix doublets.<br />
Quarterly Review of Biophysics. Volume 23<br />
(Number 1) February 1990. Contents: Biosynthetic<br />
incorporation of 15 N and 13 C for assignment and interpretation<br />
of nuclear magnetic resonance spectra<br />
of proteins. Heteronuclear niters in two-dimensional<br />
[1H, 1H]- NMR spectroscopy: combined use with<br />
isotope labelling for studies of macromolecular con-<br />
formation and intermolecular interactions.<br />
Quarterly Reviews of Biophysics. Volume 23<br />
(Number 2) May 1990. Contents: Heteronuclear<br />
three-dimensional NMR spectroscopy of isotopically<br />
labelled bioilogical macromolecules. Deuterium labelling<br />
in NMR structural analysis of larger proteins.<br />
Use of deuterium labelling in NMR studies of<br />
antibody combining site structure.<br />
Principles of Nuclear Magnetic Resonance in<br />
One and Two Dimensions. Richard R. Ernst and<br />
Geoffrey Bodenhausen. Oxford University Press.<br />
1990. 640 pp. paper $39.95<br />
A Dictionary of Concepts in NMR. S.W.<br />
Homans. Oxford University Press. 1990. 352 pp."<br />
$80.00<br />
Nuclear Magnetic Resonance: Principles and<br />
Theory. Ryozo Kitamaru. Elsevier, New York,<br />
1990.<br />
Quantum Description of High-Resolution NMR<br />
in Liquids. Maurice Goldman. Oxford University<br />
Press. 1990. 288 pp. $65.00<br />
Modern Pulsed and Continuous-wave Electron<br />
Spin Resonance. Edited by Larry Kevan and<br />
Michael K. Bowman. Wiley, New York, 1990. 440<br />
p.<br />
Principles of Magnetic Resonance, Second Ed.<br />
by C. P. Slichter, Springer, New York, 1990. 655 p.<br />
Soviet Scientific Reviews Section B: Chemistry<br />
Reviews. Volume 14, Part 2 (1990). Contents:<br />
Pulsed NMR study of molecular motion in solids.<br />
Progress in Nuclear Magnetic Resonance Spectroscopy,<br />
Volume 22 No. 6 1990. Contents: Fieldcycling<br />
relaxometry of protein solutions and tissue.<br />
Implications for MRI. Solid state NMR studies of<br />
local motions in polymers.<br />
Nuclear Magnetic Resonance, Volume 20 1989/<br />
1990. Contents: NMR books and reviews. Theoretical<br />
and physical aspects of nuclear shielding. Applications<br />
of nuclear shielding. Theoretical aspects
312<br />
of spin-spin couplings. Applications of spin-spin<br />
couplings. Nuclear spin relaxation in liquids and<br />
gases. Solid state NMR Multiple pulse NMR Natural<br />
macromolecules. Synthetic macromolecules.<br />
Conformational analysis. Nuclear magnetic resonance<br />
spectroscopy of living systems. Nuclear magnetic<br />
resonance imaging of living systems. NMR of<br />
paramagnetic species. NMR of liquid crystals and<br />
micellar solutions.<br />
1 Spin Labeling: Theory and Applications.<br />
Edited by L. J. Berliner and J. Reuben, Academic<br />
Press, New York, Vol. 3, 1989.<br />
1 Advanced EPR: Applications in Biology and<br />
Biochemistry. Edited by A. J. Hoff, Elsevier, Amsterdam,<br />
1989.<br />
1 Pulsed EPR: A New Field of Applications.<br />
Edited by C. P. Keijzers, E. J. Reijerse and J.<br />
Schmidt, North Holland, Amsterdam, 1989.<br />
1 Electron Spin Resonance, Specialist Periodical<br />
Report, Vol. 1 IB, Royal Chemical Society, London,<br />
1989.<br />
Nuclear Magnetic Resonance: Structure and<br />
Mechanism. Edited by Norman J. Oppenheimer and<br />
Thomas L. James, Academic Press, New York, 1989.<br />
507 p. (Methods in Enzymology).<br />
NMR Spectroscopy and Polymer Micro structure.<br />
The Conformation Connection. Alan E. Tonelli.<br />
VCH, New York, 1989. x 252 pp., illus. $69.50.<br />
Methods in Stereochemical Analysis.<br />
Annual Reports on NMR Spectroscopy, Vol. 21.<br />
Edited by G. A. Webb, Academic Press, London,<br />
1989. ISBN: 0-12-505321-5.<br />
EPR of Exchange-Coupled Systems. Alessandro<br />
Bencini and Dante Gatteschi. Springer-Verlag,<br />
Berlin, 1989. 287 pages. ISBN: 0-387-50944-5.<br />
Nuclear Magnetic Resonance, Vol. 18, Specialist<br />
Periodical Reports, G. A. Webb, Senior Reporter,<br />
Royal Society of Chemistry, London, 1989. 511<br />
x New additions to the list.<br />
pages. ISBN: 0-85186-412-0.<br />
Bulletin of Magnetic Resonance<br />
Advances in Magnetic Resonance Imaging.<br />
Edited by Ephraim Feig, IBM Research Division,<br />
Thomas J. Watson Research Center. Ablex Publishing<br />
Corporation. 1989. 272 pp. $55.00<br />
Advances in Magnetic Resonance. Volume 13<br />
(1989). Contents: Single crystal nuclear magnetic<br />
resonance studies of high temperature superconductors.<br />
Deuterium nuclear magnetic resonance and<br />
molecular dynamics in alkane/urea inclusion compounds.<br />
1 H nuclear magnetic resonance imaging of<br />
solids with magic-angle spinning. Two-dimensional<br />
nuclear magnetic resonance experiments for studying<br />
molecular order and dynamic in static and rotating<br />
solids. Electrophoretic nuclear magnetic resonance<br />
experiments for studying molecular order<br />
and dynamics in static and rotating solids. Electrophoretic<br />
nuclear magnetic resonance. Ultraslow<br />
atomic motion by site-selective excitation of highly<br />
resolved nuclear magnetic resonance lines in dilute<br />
spin systems. Two-dimensional hybrid experiments<br />
for the measurement of small anisotropies in magicangle<br />
spinning nuclear magnetic resonance.<br />
Analytical NMR. Edited by L. D. Field and S.<br />
Sternhell. Wiley, New York, 1989, 250 p.<br />
Modern NMR spectroscopy. A workbook of<br />
chemical problems by Jeremy K.M. Sanders, Edwin<br />
C. Constable and Brian K. Hunter. Oxford University<br />
Press, New York, 1989, 1.18 p.<br />
Introduction to Pulse NMR Spectroscopy, Second<br />
Edition., Thomas C. Farrar. The Farragut<br />
Press, Madison, Wisconsin 53705, 1989. 211 pages.<br />
$39.95 (hard cover); $24.95 (paperback).<br />
Nuclear Magnetic Resonance. Volume 19<br />
(1988/1989) Contents: Theoretical and physical aspects<br />
of nuclear shielding. Applications of nuclear<br />
shielding. Theoretical aspects of spin-spin couplings.<br />
Applications of spin-spin coupling. Nuclear<br />
spin relaxation in liquids. Solid state NMR Multiple<br />
pulse NMR Conformational analysis. Nuclear<br />
magnetic resonance of living systems. Oriented molecules.<br />
Heterogeneous systems.
Vol. 14, No. 1-4 313<br />
1 Electron Nuclear Double Resonance Spectroscopy<br />
of Radicals in Solution, by H. Kurreck, B.<br />
Kirste and W. Lubitz. VCH Publishers, New York,<br />
1988.<br />
Quantum Description of High-Resolution NMR<br />
in Liquids. Maurice Goldman. Clarendon Press,<br />
Oxford University Press, New York, 1988.<br />
Electron Spin Resonance. Volume 11B (1988).<br />
Contents: In vivo detection of free radical metabolites<br />
by spin trapping. Theoretical aspects of ESR<br />
Transition metal ions. Recent developments of EN-<br />
DOR spectroscopy in the study of defects in solids.<br />
Inorganic and organometallic radicals and clusters<br />
prepared in a rotating cryostat by metal vapour<br />
techniques. Inorganic and organometallic radicals.<br />
Metalloproteins. Complexes of paramagnetic metals<br />
with paramagnetic ligands.<br />
Nuclear Magnetic Resonance<br />
Spectroscopy. Frank A. Bovey, Lynn Jelinski and<br />
Peter A. Mirau. Academic, New York, 1988. 653 p.<br />
Coherence and NMR, Michael Munowitz. Wiley,<br />
New York, 1988. 289 pages. $39.95.<br />
Biomedical Magnetic Resonance Imaging, Principles,<br />
Methodology, and Applications. Edited by<br />
Felix W. Wehrli, Derek Shaw, and J. Bruce Kneeland.<br />
VCH Publishers, New York, 1988. 601 pages.<br />
$95.00.<br />
Interpretation of Carbon-13 NMR Spectra, F. W.<br />
Wehrli, A. P. Marchand, and S. Wehrli. Wiley, New<br />
York, 1988. 484 pages. $89.95. ISBN 0-471-91742-<br />
7.<br />
Introduction to NMR Spectroscopy. By R. J.<br />
Abraham (University of Liverpool) et al. John Wiley<br />
and Sons: Chicester and New York. 1988. xiii<br />
+ 271 pages. $44.95. ISBN 0-471-91893-8.<br />
Nuclear Magnetic Resonance. Volume 18<br />
(1987/1988) Contents: Theoretical and physical<br />
aspects of nuclear shielding. Theoretical aspects<br />
of spin-spin couplings. Applications of spin-spin<br />
1 New additions to the list.<br />
coupling. Nuclear spin relaxation in liquids and<br />
gases. Solid state NMR Multiple pulse NMR Natural<br />
macromolecules. Synthesis macromolecules.<br />
Conformational analysis. Nuclear magnetic resonance<br />
of living systems. NMR of paramagnetic<br />
species. NMR of liquid crystals and micellar solutions.<br />
The Nuclear Overhauser Effect in Stereochemical<br />
and Conformational Analysis. David Neuhaus<br />
and Michael Williamson. Contents: Integrated account<br />
of the theory, experimental practice and applications<br />
of the nuclear Overhauser effect (NOE). Describes<br />
such recent developments as NOE difference<br />
spectroscopy, NOESY, heteronuclear NOE, and rotating<br />
frame NOE. Covers experimental design in<br />
depth and the underlying theory of NOE. Assumes<br />
a graduate level knowledge of NMR spectroscopy.<br />
496 pp. 1987. $95.00. .<br />
Modern NMR Spectroscopy: A Guide for<br />
Chemists. Jeremy K.M. Sanders and Brian K.<br />
Hunter. Oxford University Press. 1987.<br />
Stereochemical Analysis of Alicyclic Compounds<br />
by C-13 NMR Spectroscopy, James K. Whitesell and<br />
M. A. Minton. Chapman & Hall, London, 1987. 231<br />
pages. $60.00.<br />
Frequency Synthesizers, Theory and Design, 3rd<br />
Edition, Vadim Manassewitsch. Wiley, New York,<br />
1987. 608 pages. $52.95.'<br />
Carbon-13 NMR Spectroscopy. High-Resolution<br />
Methods and Applications in Organic Chemistry<br />
and Biochemistry, 3rd Edition, Eberhard Breitmaier<br />
and Wolfang Voelter. VCH Publishers,<br />
New York/Weinheim, Germany, 1987. 515 pages.<br />
$135.00. ISBN 0-89573-493-1 (3-527-26466-3 at<br />
Weinheim).<br />
Two-dimensional NMR Spectroscopy: Applications<br />
for Chemists and Biochemists. Edited by<br />
William R. Croasmun and Robert M. K. Carlson.<br />
VCH Publishers: New York and Weinheim. 1987<br />
xx + 511 pages. $95.00 ISBN 0-89573-308-0.
314<br />
Instructions for Authors<br />
Because of the ever increasing difficulty of keeping<br />
up with the literature there is a growing need for<br />
critical, balanced reviews covering well-defined areas<br />
of magnetic resonance. To be useful these must<br />
be written at a level that can be comprehended by<br />
workers in related fields, although it is not the intention<br />
thereby to restrict the depth of the review.<br />
In order to reduce the amount of time authors must<br />
spend in writing we will encourage short, concise<br />
reviews, the main object of which is to inform nonexperts<br />
about recent developments in interesting aspects<br />
of magnetic resonance.<br />
The editor and members of the editorial board<br />
invite reviews from authorities on subjects of current<br />
interest. Unsolicited reviews may also be accepted,<br />
but prospective authors are requested to<br />
contact the editor prior to writing in order to avoid<br />
duplication of effort. Reviews will be subject to critical<br />
scrutiny by experts in the field and must be<br />
submitted in English. Manuscripts should be sent<br />
to the editor, Dr. David G. Gorenstein, Chemistry<br />
Department, Purdue University, West Lafayette, Indiana<br />
47906, USA/(317) 494 7851. Fax No. 317 494<br />
0239.<br />
MANUSCRIPTS must be submitted in triplicate<br />
(one copy should be the original), on approximately<br />
22x28 cm paper, type-written on one side<br />
of the paper, and double spaced throughout. If the<br />
manuscript cannot be submitted on computer tapes,<br />
floppy disks, or electronically (see below), please<br />
type with a carbon ribbon using either courier 10<br />
or 12, gothic 12, or prestige elite type face with 10<br />
or 12 pitch. All pages are to be numbered consecutively,<br />
including references, tables, and captions to<br />
figures, which are to be placed at the end of the<br />
review.<br />
ARRANGEMENT: Considerable thought<br />
should be given to a logical ordering of the subject<br />
matter and the review should be divided into<br />
appropriate major sections, and subsections, using<br />
Roman numerals, capital letters, and Arabic numerals<br />
respectively. A table of contents should be included.<br />
TABLES: These are to be numbered consecutively<br />
in the text with Arabic numerals. Their place<br />
of insertion should be mentioned in the text, but<br />
they are to be placed in order at the end of the<br />
Bulletin of Magnetic Resonance<br />
paper, each typed on a separate sheet. Each table<br />
should be supplied with a title. Footnotes to tables<br />
should be placed consecutively, using lower case letters<br />
as superscripts.<br />
FIGURES are also to be numbered consecutively<br />
using Arabic numerals and the place of insertion<br />
mentioned in the manuscript. The figures are<br />
to be grouped in order at the end of the text and<br />
should be clearly marked along the edge or on the<br />
back with figure number and authors' names. Each<br />
figure should bear a caption, and these should be<br />
arranged in order and placed at the end of the text.<br />
Figures should be carefully prepared in black ink<br />
to draftsman's standards with proper care to lettering<br />
(typewritten or freehand lettering is not acceptable).<br />
Graphs should include numerical scales<br />
and units on both axes, and all figures and lettering<br />
should be large enough to be legible after reduction<br />
by 50-60%. Figures should be generally placed on<br />
sheets of the same size as the typescript and larger<br />
originals may be handled by supplying high-contrast<br />
photographic reductions. One set of original figures<br />
must be supplied; reproduction cannot be made<br />
from photocopies. Two additional copies of each figure<br />
are required. Complex molecular formula should<br />
be supplied as ink drawings.<br />
. REFERENCES to the literature should be<br />
cited in order of appearance in the text by numbers<br />
on the line, in parentheses. The reference list is to be<br />
placed on separate sheets in numerical order at the<br />
end of the paper. References to journals should follow<br />
the order: author's (or authors') initials, name,<br />
name of journal, volume number, page, and year of<br />
publication. The abbreviation of the journal follows<br />
that used in the Chemical Abstracts Service Source<br />
Index. References to books should include in order:<br />
author's (or authors') initials, name, title of book,<br />
volume, edition if other than the first, publisher,<br />
address, date of publication, and pages.<br />
FOOTNOTES should be used sparingly and<br />
only in those cases in which the insertion of the<br />
information in the text would break the train of<br />
thought. Their position in the text should be<br />
marked with a superscript Arabic numeral and the<br />
footnotes should be typed at the bottom of the relevant<br />
page in the text, separated from the latter by<br />
a line.<br />
SYMBOLS AND ABBREVIATIONS:<br />
Mathematical symbols should be typewritten wher-
Vol. 14, No. 1-4 315<br />
ever possible. Greek letters should be identified in<br />
pencil in the margin. In reviews containing a number<br />
of mathematical equations and symbols, the author<br />
is urged to supply a list of these on a separate<br />
sheet for the assistance of the printer; this will not<br />
appear in print. Standard abbreviations will follow<br />
the American Chemical Society's HANDBOOK<br />
FOR AUTHORS names and symbols for units.<br />
PERMISSIONS: It is the responsibility of the<br />
author to obtain all permissions concerned with the<br />
reproduction of figures, tables, etc, from copyrighted<br />
publications. Written permission must be obtained<br />
from the publisher (not the author or editor) of the<br />
journal or book. The publication from which the<br />
figure or table is taken must be referred to in the<br />
reference list and due acknowledgement made, e.g.<br />
reprinted by permission from ref. (00).<br />
REPRINTS: Thirty reprints of a review will<br />
be supplied free to its senior author and additional<br />
reprints may be purchased in lots of 100.<br />
INSTRUCTIONS FOR SUBMITTING<br />
MANUSCRIPTS ON FLOPPY DISKS<br />
OR ELECTRONICALLY: If you have used a<br />
word processor to type your manuscript, please forward<br />
your manuscript after review and revision, in a<br />
computer readable form. Tape should be unlabeled,<br />
in a standard IBM format, 1600 BPI, 80x80 blocks<br />
and ASCII image. Floppy disks or tapes readable on<br />
IBM, DEC, or Macintosh computers are also acceptable.<br />
For direct submission over the phone lines use<br />
300 - 2400 baud ASCII transmission. Submission<br />
via INTERNET (address David@chem.purdue.edu)<br />
is also recommended. Additional details will be provided<br />
for the appropriate hand-shake requirements.<br />
Please supply us with the code for interpreting superscripts,<br />
greeks, etc. on your word processor.
INTERNATIONAL SOCIETY OF MAGNETIC RESONANCE<br />
(<strong>ISMAR</strong>)<br />
<strong>ISMAR</strong> MEMBERSHIP APPLICATION<br />
Name and Address (Telephone & Fax numbers, e-mail) of <strong>ISMAR</strong> Member<br />
Membership Fee for 1991, $20.00 $.<br />
Membership Fee for 1992, $20.00 $<br />
Membership Fee for 1993, $20.00 $<br />
Membership Fee for 1994, $20.00 $<br />
If this is a new membership application, please check Q<br />
If you wish to be registered as a member of the Division of Biology and Medicine,<br />
in addition to <strong>ISMAR</strong> itself, at no extra charge, please check box. D<br />
BULLETIN OF MAGNETIC RESONANCE<br />
The Quarterly Review Journal of the<br />
International Society of Magnetic Resonance<br />
Telephone: (317)494-7851 DAVID G. GORENSTEIN, Editor<br />
FAX: (317)494-0239 Department of Chemistry<br />
Internet: david@adam.chem. Purdue University<br />
purdue.edu W. Lafayette, IN, U.SA.47907<br />
Subscription to Bulletin of Magnetic Resonance: <strong>ISMAR</strong> Member Rate<br />
Volume 13 (1991), $25.00 $<br />
Volume 14 (1992), $25.00 $<br />
Volume 15 (1993), $25.00 $<br />
Volume 16 (1994), $25.00 $<br />
Subscription to Bulletin of Magnetic Resonance: Library & NON-<strong>ISMAR</strong> Member Rate<br />
Volume 13 (1991), $80.00 $<br />
Volume 14 (1992), $80.00 $<br />
Volume 15 (1993), $80.00 $<br />
Volume 16 (1994), $80.00 $<br />
TOTAL $<br />
Send this completed form with your check made payable to:<br />
International Society of Magnetic Resonance<br />
Professor Regitze R. Void, Treasurer<br />
Department of Chemistry, 0342<br />
University of California, San Diego<br />
9500 Gilman Drive<br />
La Jolla,CA 92093-0342<br />
(619) 534-0200; Fax (619) 534-7042<br />
Bitnet: rrvold@ucsd.bitnet