New techniques in structural NMR anisotropic interactions

Transcription

1 . 1. Palmer, A. Curr. Opin. Biotech. 4, (1993). 2. Kay, L.E., Torchia, D.A. & Bax, A. Biochemistry 28, (1989). 3. Pascal, S.M., Yamazaki, T., Singer, A.U., Kay, L.E. & Forman-Kay, J.D. Biochemistry 34, Tjandra, N., Feller, S.E., Pastor, R.W. & Bax, A. J. Am. Chem. Soc. 117, Bruschweiler, R., Liao, X. & Wright, P.E. Science 268, Zheng, Z., Czaplicki, J. & Jardetzky, O. Biochemistry 15, Lee, L.K., Rance, M., Chazin, W.J. & Palmer, A.G. J. Biomol. NMR 9, Lipari, G. & Szabo, A. J. Am. Chem. Soc. 104, (1982). 9. Yamazaki, T., Muhandiram, R. & Kay, L.E. J. Am. Chem. Soc. 116, (1994). 10. Engelke, J. & Ruterjans, H. J. Biomol. NMR 5, Cordier, F., Brutscher, B. & Marion, D. J. Biomol. NMR 7, (1996). 12. Dayie, K.T. & Wagner, G. J. Am. Chem. Soc. 119, Fischer, M.W.F. et al. J. Am. Chem. Soc. 119, Zeng, L., Fischer, M.W.F. & Zuiderweg, E.R.P. J. Biomol. NMR 7, (1996). 15. LeMaster, D.M. & Kushlan, D.M. J. Am. Chem. Soc. 118, (1996). 16. Muhandiram, D.R., Yamazaki, T., Sykes, B.D. & Kay, L.E. J. Am. Chem. Soc. 117, Yang, D., Mittermaier, A., Mok, Y.K. & Kay, L.E. J. Mol. Biol. 276, (1998). 18. Werbelow, L.G. & Grant, D.M. Adv. Magn. Reson. 9, (1977). 19. Vold, R.L. & Vold, R.R. Progress in NMR Spectroscopy 12, (1978). 20. Daragan, V.A. & Mayo, K.H. Biochemistry 32, (1993). 21. Daragan, V.A. & Mayo, K.H. J. Magn. Reson. Ser. B 107, Akke, M. & Palmer, A.G. J. Am. Chem. Soc. 118, (1996). 23. Akke, M., Liu, J., Cavanagh, J., Erickson, H.P. & Palmer, A.G. Nature Struct. Biol. 5, (1998). 24. Denisov, V.P. & Hale, B. J. Mol. Biol. 245, Denisov, V.P., Peters, J., Horlein, H.D. & Halle, B. Nature Struct. Biol. 3, (1996). 26. Tolman, J.R., Flanagan, J.M., Kennedy, M.A. & Prestegard, J.H. Nature Struct. Biol. 4, Steinbach, P.J. et al. Biochemistry 30, (1991). 28. Feher, V.A., Baldwin, E.P. & Dahlquist, F.W. Nature Struct. Biol. 3, (1996). 29. Beeser, S.A., Goldenberg, D.P. & Oas, T.G. J. Mol. Biol. 269, Housset, D., Kim, K.S., Fuchs, J., Woodward, C. & Wlodawer, A. J. Mol. Biol. 220, (1991). 31. Nicholson, L.K. et al. Nature Struct. Biol. 2, Wlodawer, A. & Erickson, J.W. A. Rev. Biochem. 62, (1993). 33. Sommerville, R. Prog Nucleic Acid Res 42, 1 38 (1992). 34. Gryk, M.R., Jardetzky, O., Klig, L.S. & Yanofsky, C. Protein Science 5, (1996). 35. Ogata, K. et al. Nature Struct. Biol. 3, (1996). 36. Ogata, K. et al. Nature. Struct. Biol. 2, Pawson, T. Nature, Kuriyan, J. & Cowburn, D. Curr. Opin. Str. Biol. 3, (1993). 39. Pascal, S.M., et al. Cell 77, (1994). 40. Lee, C.H., et al. Structure 2, (1994). 41. Kay, L.E., Muhandiram, D.R., Wolf, G., Shoelson, S.E. & Forman-Kay, J.D. Nature Struct. Biol. 5, (1998). 42. Akke, M., Bruschweiler, R. & Palmer, A. J. Am. Chem. Soc. 115, (1993). 43. Yang, D. & Kay, L.E. J. Mol. Biol. 263, (1996). 44. Philippopoulos, M. & Lim, C. J. Mol. Biol. 254, Farrow, N.A., Zhang, O., Forman-Kay, J.D. & Kay, L.E. Biochemistry 34, Li, Z., Raychaudhuri, S. & Wand, A.J. Prot. Sci. 5, (1996). 47. Wand, A.J., Urbauer, J.L., McEvoy, R.P. & Bieber, R.J. Biochemistry 35, (1996). 48. Mandel, A.M., Akke, M. & Palmer, A.G. Biochemistry 35, (1996). 49. Yang, D., Mok, Y.-K., Forman-Kay, J.D., Farrow, N.A. & Kay, L.E. J. Mol. Biol. 272, Alexandrescu, A.T. et al. Prot. Sci. in the press (1998). 51. Stivers, J.T., Abeygunawardana, C., Mildvan, A.S. & Whitman, C.P. Biochemistry 35, (1996). 52. Gagne, S.M., Tsuda, S., Spyracopoulos, L., Kay, L.E. & Sykes, B.D. J. Mol. Biol. in the press (1998). New techniques in structural NMR anisotropic interactions J.H. Prestegard Structure determination of biomolecules by NMR has traditionally been based on nuclear Overhauser effects (NOEs). Now there are additional sources of information that can complement NOEs in cases where positioning of remote parts of molecules is important, and where extension to larger and more complex systems is desired. The traditional NMR approach to structure determination of biomolecules is based on interpretation of rates of magnetization transfer between pairs of protons in terms of distance constraints. The approach requires not only the measurement of magnetization transfer (NOEs or nuclear Overhauser effects), but the resolution and assignment of NMR signals to specific protons, of specific residues, in a known protein sequence. Assignment of resonances and measurement of adequate numbers of NOEs have always been obstacles that made structure determination time consuming and limited to relatively small proteins (<10,000 M r for early homonuclear studies). The limitations have been pushed back over the years with additional structural information from scalar coupling constants and chemical shifts. Assignment strategies based on the use of through bond connectivites between 13 C and 15 N sites in isotopically enriched proteins have also made it possible to assign resonances in increasingly larger proteins. However, full structure determinatons have still remained confined to reasonably compact systems of molecular weights less than 30,000 40,000 M r 1,2. There have been, within the last two years, experiments reported that could dramatically change the range of applicability of NMR structural methods. Interestingly, they share an origin in anisotropic magnetic interactions that are not normally observable in high resolution NMR spectra. One important class of experiments yields structural constraints that are orientational, rather than distance based. The experiments rely on the measurement of residual dipolar couplings, and, in some cases, chemical shift anisotropy (CSA) 3 5. The measurements can be made with great efficiency, and when combined with other recent discoveries that take advantage of interference between the same dipoledipole and CSA interactions 6,7,8, it appears that NMR may be poised to take another large step forward in applicability to larger, more complex systems. Residual dipolar interactions The dipole-dipole interaction, the leading term of which is described in equation (1), is actually the basis of the NOE effect: D ij = -ξ (3 cos 2 θ - 1) ij Izi I zj (1) 2 The interaction constant, ξ ij, contains factors that describe the magnitudes of magnetic moments for a pair of nuclei i and j, and the internuclear distance dependence that shows up in NOE measurements. The spin operator, I zi I zj, has the same form as a first order through-bond spin-spin coupling interaction suggest- nature structural biology NMR supplement july

2 Fig. 1 Dipolar coupled 15 N- 1 H spin pair in an amide bond. The bond length, r, is assumed fixed and the primary variable is the angle, θ, between the magnetic field, B 0, and the internuclear vector. ing that the dipole interaction can in principle add to the multiplet splittings normally seen in high resolution NMR spectra. If it were this simple, measurement would be extraordinarily easy. The feature of most interest for our discussion is the dependence on the angle between the vector connecting the interacting nuclei and the applied magnetic field of the spectrometer, θ. As can be imagined from a look at Fig. 1, knowing θ for a bonded pair of nuclei, such as a backbone amide 15 N and 1 H in a protein, could be very useful in defining a molecular structure. The bar over the θ dependent term, however, denotes a time average. Normally in solution we assume that the time average results from a molecular tumbling that uniformly samples directions in space (isotropic sampling). The (3cos 2 θ 1) term in equation (1) then averages to zero, and no static contributions to spin interaction energies remain. This is why we are often relegated to measuring the dipole-dipole interaction indirectly through spin relaxation based phenomena such as the NOE. Sampling of orientational space, however, need not be isotropic. If molecules have preferred orientations in the presence of a magnetic field, a non-uniform Boltzmann distribution will result, and there will be secular contributions. These will appear as new multiplet splittings of resonances in the case of non-bonded nuclei, or as changes in splittings of resonances in the case of bonded nuclei 9. The utility of residual dipolar couplings in molecular structure determination of soluble molecules was anticipated years ago in works by Bothner-By and McLean 10. The reason for preferred molecular orientations in these early works was the interaction between the applied magnetic field and an orientation dependent Fig. 2 Induced protein orientation by dilute phospholipid bicelles. The protein tumbles rapidly, but anisotropically, in large aqueous interbicelle spaces. induced magnetic moment in the molecule under study. However, departures from isotropic averaging in solution, due to these interactions are normally small (less than a part in a thousand), and measurement with precision was difficult. About two years ago an example of measurement at a useful level of precision for a protein in solution appeared 3. There were several reasons for reemergence of this measurement. One was simply that a paramagnetic protein with a particularly high magnetic susceptibility anisotropy was used (cyanometmyoglobin). A second was that the 15 N dimension in the heteronuclear single quantum coherence (HSQC) experiment used on the 15 N enriched protein offered particularly high resolution. And a third was that new, very high field magnets, became available (orientation due to field induced interactions rises approximately as the square of the magnetic field). The utility of residual dipolar couplings in structural analysis had also been recognized earlier in the liquid crystal community 11. Liquid crystals orient in the presence of a magnetic field for the same reasons isolated magnetically anisotropic molecules do, but here molecules form large cooperative domains. The net orientational effect is large, and residual dipolar couplings are large. In fact, couplings are usually so large that spectra suffer from second order effects that make analysis of spectra from even small molecules dissolved in liquid crystals difficult. A few years ago applications of liquid crystal NMR technology reemerged in the structural biology community, particularly the segment involved in membrane protein structure determination. Fragments of lipid bilayers, a well accepted building block of biological membranes, were found to form liquid crystal arrays when prepared at weight % lipid to aqueous buffer using mixtures of long chain phospholipids such as dimyristoylphosphatidylcholine (DMPC) and lipids with detergent-like properties such as dihexanoylphosphatidylcholine (DHPC) 12. The bilayer fragments appear to be small discoidal particles a few hundred Å in diameter, and have become known as bicelles 13. At ºC they orient with the normals of the bilayer surfaces perpendicular to the magnetic field. Proteins and peptides that associate with lipid bilayers attach to the oriented discoidal surfaces and themselves become oriented. Applications share some of the difficulties associated with strong dipolar couplings in other liquid crystal media. However, use of isotopically 518 nature structural biology NMR supplement july 1998

3 Fig. 3 Segments from a proton coupled, nitrogen decoupled, 15 N- 1 H HSQC spectrum of a 0.4 mm solution of a barley lectin fragment in a 5% DMPC/DHPC 3:1 bicelle (doped with a positively charged amphiphile). Left is an isotropic spectrum at 25 o C, right is an oriented spectrum at 35 o C. Both increases and decreases in couplings are observed. labeled peptides, ( 15 N, 13 C), again allows avoidance of some of these problems 14, and the physical properties of bicelles continue to be improved 15. The general area of membrane protein structure determination by solid state NMR on samples including bicelles was highlighted in the last supplement series 16. An important event for the present discussion is that the areas of field induced orientation of single molecules in solution and liquid crystal induced orientation of bicelle associated molecules have merged to provide a new and very promising route to utilization of residual dipolar couplings. As noted above, one area suffers from splittings that are too small and the other from splittings that are too large. It may seem obvious that a suitable compromise should exist, but one cannot easily increase further the alignment of isolated oriented molecules, and one cannot in general decrease domain sizes or degrees of order in liquid crystals without producing heterogeneous mixtures of different phases. It turns out that at least at certain temperatures, and with certain lipid mixtures, one can dilute the bicelle medium used for membrane protein studies by a factor of five or six to produce a cooperatively oriented homogeneous medium with aqueous spaces between bicelles large enough to accommodate soluble proteins (Fig. 2) 4. Proteins in aqueous spaces in the dilute bicelle preparations adopt some of the order of the surrounding bicelles, and the level of order can be tuned to yield a very favorable ±10 Hz residual dipolar splitting by adjusting the concentration of the bicelles in the mixture. At this point an understanding of the mechanism of ordering may not be complete, but in the initial application the level of order is consistent with a simple weak collisional model that depends only on the physical barriers provided by the bicelles and on the non-spherical shape of the dissolved proteins. If this mechanism holds, there will be minimal effects on molecular tumbling rates, and hence, line widths. More importantly, structures determined will be truly representative of molecules in bulk aqueous solution. Reports currently emerging include applications to not only proteins, but nucleic acids and carbohydrates. It is likely that any variation in mechanistic understanding will have a minor effect on applications to structure determination. An example of the type of data attainable in a bicelle medium is shown in Fig. 3. This is an HSQC spectrum of a two domain fragment of a carbohydrate binding protein from barley 17, that has been ordered in a 5% 3:1 DMPC/DHPC bicelle dispersion. Decoupling of protons during 15 N evolution has been omitted from the normal HSQC sequence so that a doublet appears for each correlated amide proton amide nitrogen pair. A region of the spectrum that shows two doublets that increase and decrease splittings respectively, upon orientation is shown. The bicelle preparations have a convenient property in that an isotropic phase containing the same lipids, presumably in more symmetric micelles, is obtained by lowering the temperature slightly. Spectra under these conditions (left of Fig. 3) show only scalar through-bond couplings of 93 ±2 Hz. Comparing isotropic and oriented spectra, Lys 53 shows a negative 5 Hz residual dipolar contribution and Lys 95 shows a positive 2 Hz contribution. These are indicative of an average N-H vector orientation perpendicular and parallel to the magnetic field respectively a useful structural constraint. In the above illustration it is clear that measurements can be made directly from the frequency domain spectrum. These measurements are straightforward and very efficient compared to NOE measurements. Fig. 4 Chemical shift anisotropy of an amide carbonyl carbon. Chemical shift tensor elements, δ 11, δ 22, and δ 33, are taken to be 223, 79 and 55 p.p.m. respectively. Shifts observed in oriented samples are functions of the angle between the magnetic field, B 0, and the principle shift tensor axes. nature structural biology NMR supplement july

4 Spectra such as the one shown can be acquired in as little as one hour on 1 mm protein samples. There are also a number of options for making these measurements more precise In addition, measurements are not confined to 15 N- 1 H pairs, but extend to 13 C- 1 H pairs and even nonbonded 1 H- 1 H pairs 4. Chemical shift anisotropy Residual dipolar coupling is not the only anisotropic spin interaction that can provide useful structural information. Chemical shifts are also anisotropic. Nuclei which are part of various molecular functional groups resonate at different frequencies depending on shielding by the local electronic environment. Electronic environments are seldom isotropic, and hence, shielding is different for different orientations of functional groups. For the case of a molecule with an axially symmetric chemical shift tensor, the contribution to spin energy levels, which leads to an offset in resonance position from that seen with isotropic averaging, is given in equation (2): C i = δ (3 cos2 θ - 1) γi B 0 I zi (2) 2 The coefficient δ is the difference in chemical shift in directions parallel and perpendicular to the symmetry axis, and γ i B 0 I zi is the Zeeman interaction operator. It is significant that an angular dependence analogous to that seen for the residual dipole interaction occurs. An illustration of anisotropic chemical shift effects is given for the 13 C of the carbonyl group of the peptide in Fig. 4; if the peptide preferred an orientation with the field perpendicular to the plane, the carbonyl carbon would resonate at a higher field (or lower frequency) than in a case where isotropic sampling of orientations occurred. Measured chemical shift offsets are clearly of structural value. Chemical shift offsets are easily measured in the well oriented cases represented by membrane associated peptides, and both 15 N and 13 C chemical shift offsets have been used to help deduce structural properties 12,15,23. In weakly oriented solution systems, and in dilute bicelle systems, measurements are more difficult. Small changes in resonance position due to temperature and other factors must be considered. Nevertheless, useful measurements have been made even in very weakly oriented solution samples 5. Structure determination How can structure be determined from all this new orientational information? Published examples utilizing orientational information from simple solution and dilute bicelle studies have been confined to refinement of existing protein structures. It is possible to add orientational penalty functions to force fields used in simulated annealing protocols, and in this Fig. 5 Properly oriented domains in a two zinc-finger DNA complex have common orientations of independently determined principle orientation tensors. The structure shown is modeled on a crystal structure of the TramTrack DNA complex by Fairall et al. 25 way combine orientational constraints with the distance constraints used in conventional NMR structure determinations. A good example is the work on a solution of a DNA binding domain from the transcription factor GATA-1 bound to a 16 base pair double helical DNA 24. A conventional NMR structure determination had previously been completed. Adding orientation constraints improved consistency with accepted values of backbone torsion angles in several cases and did significantly change the structure of a loop that had been poorly defined by the distance constraint data. It was also pointed out that it was possible to deduce the orientation of the DNA grove binding α-helix from the small negative values of the couplings for N H bonds in this protein segment. Currently there are few examples of determining structures using orientational data from solution or bicelle studies as a primary structure determinant. There are some fundamental problems in that error functions used with orientational data are more complex than those used with distance data because the inverse of the basic angular function of equations (1) and (2) is multivalued. However, there are viable alternative approaches that can be gleaned from work occurring on more ordered membrane samples 12,23. One can in principle determine, within a degeneracy of two, the orientation of any rigid substructure of a molecule from five independent orientational measurements. Such a substructure can be a peptide bond if 15 N- 1 H, 13 C- 1 H, and 13 C- 13 C dipolar vectors are combined with chemical shift anisotropy effects. Then, peptide bonds can be oriented one at a time. Such a procedure has recently been used to suggest a structure for a membrane bound myristoylated peptide from the ARF-1 protein 12. Substructures can also be single domains within multidomain proteins. If the structures of individual domains are known from conventional NMR or X-ray studies, the entire domain can be treated as a rigid substructure. Even if only N-H dipolar interaction vectors are used, there is plenty of information to determine orientation relative to a principle orientational frame. For a well structured molecule, domains should see a common principle orientational frame. This requirement is illustrated in Fig. 5 for a two domain transcription factor bound to a DNA double helix (taken from the X- ray structure of Fairall et al. 25 ). If the domains were positioned properly, the principle orientational frames, determined independently from orientational NMR data in solution, would differ only by translation. If they were not, domains could be reoriented to achieve a common frame. The positioning of the domains is entirely independent of how far apart they may be, thus illustrating the fact that residual dipolar data and CSA data can be ideal complements to short range dis- 520 nature structural biology NMR supplement july 1998

5 tance constraints from NOEs. A combination of NOE and residual dipolar data has recently been used to determine the relative orientation of two zinc finger domains from the transcription factor Sp1 as they bind to a 14 base pair DNA double helix (V.A. Narayan & J.P. Caradonna, pers. comm.). Larger molecules and the future Why is there so much excitement about the prospects for using residual dipolar coupling and other data from anisotropic interactions to determine structures of biomolecules? Some certainly arises from the improvements in precision that might arise in refinement of existing NMR structures 24. But, we believe that more will arise from the unique contributions that can be made in cases where NOEs are hard to obtain. The example involving positioning of remote domains in DNA binding transcription factors is a good illustration of what we mean. But, another arises when biomolecules are simply large. In conventional NMR structure determinations of proteins, side chain side chain NOEs play a particularly important role in determining a three dimensional fold. Frequently only side chain protons of sequentially remote residues come close enough to one another to yield an NOE. Yet, assignments of side chain proton resonances, or those of their attached carbons, are hard to establish because they require multiple transfers through short lived 13 C magnetization to get from the more easily assigned peptide backbone resonances to those of the side chain extremities. Assignment of backbone resonances can often be accomplished with much larger proteins than assignment of side chain resonances. In fact, resolution and assignment of backbone amide proton and nitrogen resonances extends to particularly large systems when perdeuteration of non-exchangeable protons is employed 26. While relatively few NOEs are observed that connect backbone amide protons to sequentially remote parts, observed short and intermediate range NOEs do define secondary structure elements. Under these circumstances the orientational constraints for N-H vectors in secondary structure elements may be used to supplant some side chain side chain NOEs. Interference of anisotropic interactions There are some reasons to believe that further improvement in an ability to Fig. 6 Line widths for the two lines of a proton coupled 15 N doublet as a function of spectrometer operating frequency. A correlation time of 20 ns and a nitrogen CSA of -160 p.p.m. have been assumed. Note the minimum line width predicted at a frequency slightly above 1 GHz. resolve and assign amide nitrogen and proton resonances are at hand 6. In fact, with suitable advances in magnetic field strengths it may be possible to consider NMR based structure determinations of proteins several times the size of those studied today. One reason is an expected dramatic narrowing of resonances in newly devised experiments. The expected advance is actually based on interference of the same two interactions that we have been discussing above, the dipole-dipole interaction and the dipole-csa interaction. In addition to shifts in resonance position and changes in doublet splitting, dipole-dipole and CSA-dipole interactions contribute to line broadening. The fact that the two contributions interfere is well known and can be seen in the simple proton coupled 15 N- 1 H cross peaks of Fig. 4. The cross peaks belonging to a 1 H coupled doublet are not of the same width. The downfield doublet component is actually narrower; in fact, it is narrower than the single cross peak that appears when protons are decoupled and the two components collapse to a single peak. Differences are small here because they are obscured by a relatively short acquisition time and unresolved long range couplings to protons that could be replaced with deuterons. Nevertheless, the differences are real. The reason that differential line broadening occurs is that the nitrogen spin relaxes in response to the square of the total fluctuation in the local magnetic field as opposed to the sum of the squares of the individual fluctuations 27. When local field fluctuations come from two sources such as the magnetic dipole of the attached proton and the chemical shift anisotropy of the nitrogen, it is the vector sum that counts. For the two lines of a nitrogen doublet, the magnetic dipoles of the protons project in opposite directions. So, the dipolar and CSA fields add for one line and subtract for the other. Hence, one line relaxes efficiently and the other less efficiently. At fields corresponding to 500 MHz for protons the CSA contribution in an amide 15 N- 1 H pair is smaller than the dipolar contribution, but large enough to produce lines of observably different width. However, the CSA field is itself field dependent. At higher fields the sizes of the CSA and dipolar terms become more comparable and eventually become equal. If the CSA tensor were axially symmetric and colinear with the dipolar vector, exact cancellation would occur at a little over 1 GHz, and the line width of the narrower line would be determined predominantly by remote spin interactions. The line widths of the two lines as a function of spectrometer operating fre- nature structural biology NMR supplement july

6 quency, or equivalently magnetic field, are plotted in Fig. 6. Clearly there would be tremendous advantage in observing only the narrow peaks in coupled HSCQ spectra. Wüthrich and coworkers have devised pulse sequences for the selective acquisition of the narrower peaks in two dimensional HSQC spectra and illustrated the first applications of this new methodology 6. Improvement in resolution (and improvement in sensitivity that can come with it) is key to doing NMR based structure determination of larger molecules. Increasing the molecular weight of a protein by a factor of two approximately doubles the number of peaks, and doubles the line width of each peak. Decreasing the line width by a factor of two to four would largely compensate for the degradation in resolution and sensitivity on doubling the size of a protein. Coupling these improved HSQC experiments with some of the assignment and NOE experiments traditionally used in protein structure determination will certainly be possible. It may also be possible to couple them with experiments used to determine residual dipolar couplings and CSA offsets. We should mention that the above is not the only case where interference between dipole-dipole and CSA interactions promises to improve structure determination of biomolecules in solution. There is an inherent geometry dependence of this interference. In the above example, it is important that the approximate symmetry axis of the nitrogen CSA tensor is nearly parallel to the 15 N- 1 H dipolar vector. When it is not, effects vary. Some authors have used these effects to advantage in determining torsion angles along protein backbones 7,8. The information returned is similar to that obtained through three bond scalar coupling, but applications to larger molecules will be possible. Other authors have used interference in spin relaxation to get extra information needed to deal with anisotropic motion in proteins 28. A caveat about motion All of the above discussion about structure presumes an ability to separate motional effects from structural effects. In many cases, analysis of observables is predicated on the assumption of a rigid structural model. Proteins and other biomolecules are not rigid. We can, to a certain extent, ignore this fact in NOE based determinations because models that include rapid uncorrelated centrosymmetric motions are degenerate with rigid models. This is not the case with residual dipolar and other anisotropic data. Motions almost always reduce the magnitude of their effects. Moreover, all motions from those on picosecond time scales to those on millisecond time scales can contribute. In early work on residual dipolar couplings in a paramagnetic form of myoglobin, deviations from rigid model predictions were observed and found consistent with averaging by internal motions 29. At the current time it appears that the effects of these motions on structures determined can be minimized by using a combination of spin relaxation and amide exchange rate experiments to identify residues which may be most susceptible to motional averaging and eliminate data from these residues when attempting to produce structural models. Nevertheless, there may be reason for caution in interpretation, particularly for applications that refine to high resolution. There are of course good as well as bad aspects to the existence of motional effects. A new probe of motional properties that is sensitive to a broad range of motions is interesting in itself. These more interesting aspects of motional effects are discussed in another article in this supplement 30. Acknowledgments Support from the NIH and NSF is gratefully acknowledged. We also thank J. Losonczi, A. Fowler, M. Fischer and H. Al-Hashimi for their help in preparing the figures. J.H. Prestegard is at the Complex Carbohydrate Research Center, University of Georgia, Athens, Georgia , USA. Correspondence should be addressed to J.H.P. jpresteg@ccrc.uga.edu 1. Clore, G.M. & Gronenborn, A.M., Nature Struct. Biol. 4, Wagner, G., Nature Struct. Biol. 4, Tolman, J.R., Flanagan, J.M., Kennedy, M.A., & Prestegard, J.H. Proc. Natl. Acad. Sci. USA 92, Tjandra, N. & Bax, A. Science 278, Ottiger, M., Tjandra, N. & Bax, A. J. Am. Chem. Soc. 119, Pervushin, K., Riek, R., Wider, G. & Wüthrich, K. Proc. Natl. Acad. Sci. USA 94, Reif, B., Hennig, M. & Griesinger, C. Science, 276, Brutscher, B., Skrynnikov, N.R., Bremi, T., Brueschweiler, R. & Ernst, R.R. J. Magn. Reson. 130, (1998). 9. Prestegard, J.H., Tolman, J.R., Al-Hashimi, H.M. & Andrec, M. In Modern techniques in protein NMR (N.R. Krishna & L.J. Berliner, eds) (Plenum, New York, 1998) in the press. 10. Bastiaan, E.W., Maclean, C., Van Zijl, P.C.M. & Bothner-By, A.A. Ann. Rep. NMR Spect. 19, (1987). 11. Emsley, J.N. & Lindon, J.C. (Pergamon Press, Oxford, 1975). 12. Sanders II, C.R., Hare, B.J. Howard, K.P. & Prestegard, J.H. Prog. Nucl. Magn. Resonan. 26, (1994). 13. Sanders II, C.R. & Landis, G.C. Biochemistry, 34, Losonczi, J.A. & Prestegard, J.H. Biochemistry 37, (1998). 15. Prosser, R.S.,Hunt, S.A, DiNatale, J.A. & Vold, R.R. J. Am. Chem. Soc. 118, (1996). 16. Opella, S. J. Nature Struct. Biol. 4, Weaver, J.L. & Prestegard, J.H. Biochemistry. 37, (1998). 18. Ottiger, M., Delaglio, F. & Bax, A. J. Magn. Reson. 131, (1998). 19. Tjandra, N., Grzesiek, S. & Bax, A. J. Am. Chem. Soc. 118, (1996). 20. Tolman, J.R. & Prestegard, J.H. J. Magn. Reson (1996). 21. Bax, A. & Tjandra, N. J. Biomol. NMR. 10, Yang, D., Tolman, J.R., Goto, N.K. & Kay, L.E. J. Biolmol. NMR 11, (1998) in the press. 23. Kovacs, F.A. & Cross, T.A. Biophys. J. 73, Tjandra, N., Omichinski, J.G., Gronenborn, A.M., Clore, G.M. & Bax, A. Nature Struct. Biol. 4, Fairall, L., Schwabe, J.W., Chapman, L., Finch, J.T. & Rhodes, D. Nature. 366, (1993). 26. Venters, R.A., Metzler, W.J., Spicer, L.D., Mueller, L., & Farmer III, B.T. J. Am. Chem. Soc. 117, Werbelow, L.G. & Grant, D.M. Adv. Magn. Reson. 9, (1977). 28. Fischer, M.W., et. al. J. Am. Chem. Soc. 119, Tolman, J.R., Flanagan, J.M., Kennedy M. A. & Prestegard, J.H. Nature Struct. Biol. 4, Kay, L.E. Nature Struct. Biol. 5, (1998). 522 nature structural biology NMR supplement july 1998