NMR supplement 492 nature structural biology ? NMR supplement ? july 1998 used for quantitative characterization of structural and dynamic aspects of protein and nucleic acid hydration in solution. NMR data on dynamics and solvation have so far in most instances been collect- ed with small proteins and nucleic acid fragments, but this fundamental informa- tion applies to any molecular size and can be used for the interpretation of crystal structures with regard to structure-func- tion correlations in physiological fluids 9 . However, although NMR data on molecu- lar dynamics have long attracted keen interest by theoreticians 10,11 , the inherent complexities of dynamic macromolecular struc- tures have so far limited the use of this information in the prac- tice of analyzing biochemical measurements and designing novel projects. In a certain sense these NMR data — on dynamics and solvation — appear to be ahead of their time: here, then, is an open avenue for the future. Very recent observations also indi- cate that our view of internal mobility of proteins may still be largely incomplete and that additional insight on low frequency motions can be anticipated from novel NMR approaches 12–14 , in particular also from the residual anisotropic interac- tions reviewed by Prestegard 15 in his contribution to this volume. NMR spectroscopy is special among the techniques of structural biology in its abil- ity to observe and characterize unfolded polypeptide chains in solution 16,17 . Applications of interest to the ‘protein folding problem’ are discussed in the pre- sent NMR special issue in articles by Dyson and Wright 18 , and by Dobson and Hore 19 . In addition, recent structure determinations revealed the existence of In the first Nature Structural Biology spe- cial issue on nuclear magnetic resonance spectroscopy (NMR), Gerhard Wagner recounted the development of NMR from 1946 to the present 1 . The history of NMR includes, besides spectacular achievements in physics, chemistry, materials science and medicine 2 , the determination of the complete three- dimensional structure of a globular protein in solution in 1984 3 and subse- quently the evolution of NMR into a key technique for structural biolo- gy. The achievements of NMR with biological macromole- cules reviewed by Wagner 1 and the conditions under which they have been attained repre- sent the platform for future advances, which are the prime focus of this article. A remarkable field The increasingly important role that NMR plays in structural biology is illustrated by Table 1, which lists the number of novel NMR structures of proteins and nucleic acids published annually during the period 1990–1996. In the table these numbers are fur- ther placed in perspective by comparison with the corre- sponding data for the other techniques of structure determi- nation. It is quite obvious that, so far, nearly all atomic resolution struc- tures of biological macromolecules have been solved either by X-ray diffraction in single crystals or by NMR in solution 4 , where the number of new crystal struc- tures published exceeds the number of new NMR structures about three- to four- fold. The harvest of new structures in 1996 exceeds the corresponding numbers for 1990 by more than four-fold. From 1994 onward the total number of new struc- tures published per year exceeds the total number of three-dimensional structures of proteins and nucleic acids that were known at the end of 1989, so that we now have an extensive data base of three- dimensional structures in the Brookhaven Protein Data Bank. Impressive testimony to the amazing recent evolution of structural biology, the numbers in Table 1 are only a partial reflection of the increasing prominence of NMR, since in addition to structure deter- mination this technique can provide a wealth of supplementary, unique infor- mation that is complementary to crystal structure data 5 . Wagner 1 presented a sur- vey on NMR characterization of molecu- lar dynamics of proteins and nucleic acids in solution, and the present issue includes a further thorough review of this area by Kay 6 . In addition, direct observation of nuclear Overhauser effects (NOE) between water protons and macromolecu- lar hydrogen atoms 7 and measurement of nuclear magnetic relaxation dipersion 8 are The second decade — into the third millenium Kurt Wüthrich NMR spectroscopy is one of the principal experimental techniques of structural biology, with abilities to determine atomic resolution structures as well as investigate dynamics and intermolecular interactions of biological macromolecules. There is plenty of room for continued progress of this young branch of science, based on further technical advances as well as innovative funding strategies and project organization. Fig. 1 NMR structure of the recombinant murine prion protein. In a first step the structure of the fragment 121–231 (which includes a globular domain from residues 126–226 38 ) was solved. In the intact protein with residues 23–231, the segment 23–126 forms an extended coil, where high mobility of the individual residues is manifested by rotational correlation times shorter than 1 ns for the 15 N- 1 H groups (adapted from ref. 20). NMR supplement nature structural biology ? NMR supplement ? july 1998 493 on dissection of their modular architec- ture, which is an approach used in a wide variety of structural studies by NMR as well as X-ray diffraction. Finally, the solid state NMR techniques presented by Griffin 31 in this issue carry the promise of information on very large systems. A new NMR experiment, TROSY (transverse relaxation-optimized spec- troscopy) 32 , promises a further several- fold increase of the molecular size accessible with solution NMR. TROSY makes use of the fact that at 1 H frequen- cies in the range 900–1,000 MHz nearly complete cancellation of transverse relax- ation effects can be achieved for one of the four multiplet components observed for 15 N- 1 H moieties (Fig. 2), and TROSY exclusively observes this narrow compo- nent. Fig. 3 shows that already at 750 MHz, TROSY yields significantly narrow- er spectral linewidths and improved sensi- tivity for observation of 15 N- 1 H groups in an oligomeric protein of 110,000 M r than native, folded proteins that contain long flexible coils attached to well structured globular domains. A striking example is the prion protein (Fig. 1), with a globular domain containing a -helical and b -sheet secondary structure, and an N-terminal domain of nearly equal size that forms a highly mobile extended coil 20 . The length of this extended coil exceeds the diameter of the globular domain by almost ten-fold. A similar structure consisting of a globular domain and a flexibly extended coil has been observed for a yeast heat-shock tran- scription factor 21 . Considering the practi- cal difficulties in preparing proteins with long extended coils for structural studies, one is tempted to speculate that this struc- ture type has so far largely escaped detailed characterization and may be quite com- mon in nature. In this context it is interest- ing that the crystal structure of the nucleosome core particle 22 contains numerous extended polypeptide segments in the multimolecular aggregate, indicat- ing that formation of the oligomeric struc- ture may start with subunits that contain sizeable extended coils. The existence of extended polypeptide coils under physio- logical conditions is expected to depend on protective chaperoning by other macromolecules. The assembly of oligomeric structures — for example, the formation of the nucleosome core particle, or processes such as those leading to dis- ease-related amyloid formation — could thus be governed in subtle ways by chaper- one systems that ensure maintenance and controlled release of the flexible coil struc- tures in the healthy cell. Flexible polypep- tide coils attached to globular domains could well become yet another focus for future NMR studies, considering that their characterization by alternative methods is very limited and NMR has the potential for further refinement of this class of structures. Large molecules The title “NMR structures ... beyond 20,000 M r ” (M r = relative molecular mass) used by Clore and Gronenborn 23 in the first Nature Structural Biology special issue on NMR reflects on a stigma that has traditionally been attached to solution NMR, that is “its use is limited to small molecular sizes”. Actually, during the last few years the size limit for de novo protein structure determin- ation has been moved to about 30,000 M r by the use of stable isotope labeling with 13 C, 15 N and 2 H 24,25 combined with the use of triple-resonance experiments 26,27 and het- eronuclear-resolved 28 and -edited 29 NMR. More importantly, there is no a priori rigid size barrier: As men- tioned above, much of the NMR data on internal mobility and hydration apply to any molecular size. In the article by Campbell and Downing 30 NMR studies with very big proteins are based Fig. 2 Frequency dependence from 100–1800 MHz of the full resonance line width at half height for amide groups in TROSY experiments calcu- lated for three correlation times of t c = 20, 60 and 320 ns, which represent spherical pro- teins with molecular weights of 50,000, 150,000 and 800,000 M r . a, 1 H N linewidth. b, 15 N linewidth. (The calcula- tion used Ds ( 15 N) = 155 p.p.m. and Ds ( 1 H N ) = 15 p.p.m.; axial symmetry was assumed for both tensors; the angle between the principal tensor axis and the N–H bond was assumed to be 15° for 15 N and 10° for 1 H N ; d N–H = 0.104 nm; effects of long-range dipole- dipole couplings with spins outside of the 15 N- 1 H moiety were not considered. Table 1 Annual publications of novel atomic resolution structures of proteins and nucleic acids (ref. 4) Method X-ray crystallography NMR Other methods Year (single crystals) (solution) (crystals, fibers) 1990 109 23 2 1991 123 36 - 1992 168 61 - 1993 207 59 - 1994 352 100 2 1995 394 102 - 1996 461 112 - a b NMR supplement 494 nature structural biology ? NMR supplement ? july 1998 number of NMR groups, and that X-ray crystallography has already made the step to ‘big science’: in 1996 the majority of X- ray structure determinations, including essentially all structures with M r above 100,000, made use of high intensity syn- chrotron X-ray sources in addition to local X-ray equipment 4 . In contrast, the NMR community is oriented to doing ‘small science’, with the groups being equipped with commercial spectrom- eters and working individually on the preparation of isotope-labeled proteins and nucleic acids. the corresponding, conventional NMR experiments. Theory (Fig. 2) predicts that TROSY experiments with suitably isotope-labeled systems can yield infor- mative data on proteins in particles with molecular weights of several hundred thousand M r , such as membrane proteins solubilized in micelles or lipid vesicles, proteins attached to nucleic acid frag- ments, or homo-oligomeric proteins. For example, direct correlation experiments of the kind shown in Fig. 3 enable chemi- cal shift mapping of intermolecular con- tacts in very large aggregates, and will thus extend the use of SAR by NMR 33 (where SAR stands for ‘structure-activity relationships’) to larger systems. The TROSY principle is readily applicable for improvement of a wide variety of more complex NMR experiments, also with aromatic 13 C- 1 H groups 34 . Big science NMR When comparing the contributions from X-ray crystallography and NMR in Table 1, one has to consider that the number of research groups in macromolecular crys- tallography exceeds by a large margin the Fig. 3 Comparison of conventional and TROSY-type 15 N- 1 H correlation spectra of the uniformly 15 N- and 2 H-labeled 110,000 M r protein 7,8-dihydro- neopterin aldolase from Staphylococcus aureus in H 2 O solution. This protein is a homo-octamer with subunits of 121 amino acid residues. a, TROSY- HSQC. b, Conventional HSQC. c, a’( 1 H) and b’( 1 H) show cross sections along the 1 H frequency axis w 2 through three peaks identified by upper case letters in (a) and (b) respectively. d, a”( 15 N) and b”( 15 N) show cross sections along the 15 N frequency axis w 1 through the same three peaks. (Protein monomer concentration 0.4 mM, pH 5.5, T 20 °C, 1 H frequency 750 MHz, recording time per spectrum 1 h). In both dimensions the positions of cor- responding peaks in (a) and (b) differ by 1 J( 15 N, 1 H)/2. a b c d NMR supplement nature structural biology ? NMR supplement ? july 1998 495 1. Wagner, G. Nature Struct. Biol. 4, 841–844 (1997). 2. Grant, D.M. & Harris. R.K. (eds) Encyclopedia of nuclear magnetic resonance (Wiley, New York; 1996). 3. Williamson, M.P., Havel, T.F. & Wüthrich, K. J. Mol. Biol. 182, 295–315 (1985). 4. Hendrickson, W.A. & Wüthrich, K. (eds) Macromolecular structures. (Current Biology, London; 1991–1997). 5. Wüthrich, K. NMR in structural biology (World Scientific, Singapore; 1995). 6. Kay, L.E. Nature Struct. Biol. 5, 514–517 (1998). 7. Otting, G., Liepinsh, E. & Wüthrich, K. Science 254, 974–980 (1991). 8. Venu, K., Denisov, V.P. & Halle, B. J. Am. Chem. Soc. 119, 3122–3134 (1997). 9. Wüthrich, K. Acta Crystallogr. D 51, 249–270 (1995). 10. Gelin, B.R. & Karplus, M. Proc. Natl. Acad. Sci. USA 72, 2002–2006 (1975). 11. Smith, P.E., van Schaik, R.C., Szyperski, T., Wüthrich, K. & van Gunsteren, W.F. J. Mol. Biol. 246, 356–365 (1995). 12. Tolman, J.R., Flanagan, J.M., Kennedy, M.A. & Prestegard, J.H. Nature Struct. Biol. 4, 292–297 (1997). 13. Tjandra, N. & Bax, A. Science 278, 1111–1114 (1997). 14. Akke, M., Liu, J., Cavanagh, J., Erickson, H.P. & Palmer, A.G. Nature Struct. Biol. 5, 55–59 (1998). 15. Prestegard, J.H. Nature Struct. Biol. 5, 518–523 (1998) 16. Wüthrich, K. Curr. Opin. Struct. Biol. 4, 93–99 (1994). 17. Shortle, D.R. Curr. Opin. Struct. Biol. 6, 24–30 (1996). 18. Dyson, H.J. & Wright, P.E. Nature Struct. Biol. 5, 499–503 (1998). 19. Dobson, C. & Hore, P.J. Nature Struct. Biol. 5,. 504–507 (1998). 20. Riek, R., Hornemann, S., Wider, G., Glockshuber, R. & Wüthrich, K. FEBS Lett. 413, 282–288 (1997). 21. Cho, H.S. et al. Prot. Sci. 5, 262–269 (1996). 22. Luger, K., M?der, A.W., Richmond, R.K., Sargent, D.F. & Richmond, T.J. Nature 389, 251–260 (1997). 23. Clore, G.M. & Gronenborn, A.M. Nature Struct. Biol. 4, 849–853 (1997). 24. LeMaster, D.M. Progr. NMR Spectr. 26, 371–419 (1994). 25. Kainosho, M. Nature Struct. Biol. 4, 858–861 (1997). 26. Bax, A. & Grzesiek, S. Acc. Chem. Res. 26, 131–138 (1993). 27. Kay, L.E. & Gardner, K.H. Curr. Opin. Struct. Biol. 7, 722–731 (1997). 28. Fesik, S. & Zuiderweg, E.R.P. J. Magn. Reson. 78, 588–593 (1988). 29. Otting, G. & Wüthrich, K. Q. Rev. Biophys. 23, 39–96 (1990). 30. Campbell, I.D. & Downing, A.K. Nature Struct. Biol. 5, 496–499 (1998). 31. Griffin, R.G. Nature Struct. Biol. 5, 508–513 (1998). 32. Pervushin, K., Riek, R., Wider, G. & Wüthrich, K. Proc. Natl. Acad. Sci. USA 94, 12366–12371 (1997). 33. Shuker, S.B., Hajduk, P.J., Meadows, R.P. & Fesik, S.W. Science 274, 1531–1534 (1996). 34. Pervushin, K., Riek, R., Wider, G. & Wüthrich, K. J. Am. Chem. Soc., in the press. 35. Normile, D. Science 278, 1700–1702 (1997). 36. Service, R.F. Science 279, 1127–1128 (1998). 37. Pennisi, E. Science 279, 978–979 (1998). 38. Riek R. et al. Nature 382, 180–182 (1996). Introduction of the traits of big science to biomacromolecular NMR would mean the loss of the opportunity to perform all the steps of a structure determination in the individual laboratories, a comfort that most of us presently enjoy. In return we would get access to NMR centers with more powerful equipment than the individual researcher could afford, as well as support from chemistry and microbiology laboratories specialized in efficient preparation of isotope-labeled precursor molecules and production of NMR quantities of labeled macromolecules. In a successful big science scenario the newly established NMR centers would need to be comparable in function (if not in size) with the high intensity light sources of the X-ray crystallographers, with instrumenta- tion that significantly outperforms the equipment likely to be available to individ- ual research groups. (This demand con- trasts with the present situation, where the equipment of regional and national NMR centers consists of spectrometers identical to those available in the laboratories of most of the leading individual research groups.) The centralized equipment should provide high quality data for automated spectral analysis, to enable improvement of quality and efficiency of NMR structure determi- nation, including structure refinement against raw NMR data. Much of the aforementioned demands for improved equipment in big science NMR centres could probably be met by improved performance at the presently available field strengths corresponding to 1 H resonance frequencies up to 800 MHz, which can be anticipated from on-site con- struction of specialized NMR equipment and from operation in a suitably shielded and stable environment. Nonetheless, past experience in general and more specifically the TROSY experiment 32 (Fig. 2) indicate additional advantages of higher fields, in particular in extending the upper size range amenable to analysis by NMR. Politically acceptable big science NMR centers would be laid out primarily for high-throughput, high-quality structure determination. Facilities operating at highest fields would, however, also initiate further development of spectroscopic techniques for the benefit of continued progress of biomolecular NMR. The future Considering that the current use of NMR in structural biology still shows typical signs of a young, emerging field of research, the results obtained (Table 1) are truly remarkable and are an encourage- ment for the future. NMR with biological macromolecules could experience an important boost with the creation of cen- tralized high-performance facilities for data collection, which have recently been extensively discussed in Japan 35 and in the USA 36 . These projects are intimately relat- ed to genome sequencing projects, and envisage determination of a comprehen- sive set of all naturally occurring protein folds 35,37 . NMR will have an indispensable role in any such ambitious venture, since experience indicates that a sizeable frac- tion of all proteins will not be amenable to structure determination in crystalline form, and increasingly there will also be demands for additional, NMR-specific information. Considering the current rate of progress in structural biology (Table 1) an immediate start of construction of high performance NMR centers should be encouraged by the necessary innovations in funding. Acknowledgments Support by the Schweizerischer Nationalfonds is gratefully acknowledged. The author thanks K. Pervushin, R. Riek and G. Wider for the preparation of the figures and for discussions. Kurt Wüthrich is at the Institut für Molekularbiologie und Biophysik, Eidgen?ssische Technische Hochschule H?nggerberg,CH-8093 Zürich, Switzerland. Correspondence should be addressed to K.W. Fax: +41 1 633 11 51.