NMR supplement nature structural biology ? NMR supplement ? july 1998 499 Equilibrium NMR studies of unfolded and partially folded proteins H. Jane Dyson and Peter E. Wright Multidimensional NMR studies of proteins in unfolded and partially folded states give unique insights into their structures and dynamics and provide new understanding of protein folding and function. Acknowledgments A.K.D. thanks the Wellcome Trust and The Queen’s College, Oxford for support. I.D.C. is also supported by the Wellcome Trust. The authors acknowledge P. Handford for critical reading of the manuscript. The Oxford Centre for Molecular Sciences is funded by the Biology and Biotechnology Sciences Research Council, Engineering and Physical Sciences Research Council and Medical Research Council. Iain D. Campbell and A. Kristina Downing are at the Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, England and the Oxford Centre for Molecular Sciences, New Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QT, England. Correspondence should be addressed to I.D.C. email: idc@bioch.ox.ac.uk 1. Hendrickson, W.A. & Wüthrich, K. Macromolecular structures (Current Biology, London, 1996). 2. Cowburn, D. & Riddihough, G. Nature Struct. Biol. 4, 761–762 (1997). 3. Doolittle, R.F. Annu. Rev. Biochem. 64, 287–314 (1995). 4. Kuriyan, J. & Cowburn, D. Annu. Rev. Biophys. Biomol. Struct. 26, 259–288 (1997). 5. Bork, P., Schultz, J. & Ponting, C.P. Trends Biochem. Sci. 22, 296–298 (1997). 6. Rhodes, D. & Burley, S.K. Curr. Opin. Struct. Biol. 7, 73–75 (1997). 7. Chothia, C. & Jones, E.Y. Annu. Rev. Biochem. 66, 823–862 (1997). 8. Potts, J.R. & Campbell, I.D. Matrix Biology 15, 313–320 (1996). 9. Bork, P., Downing, A.K., Kieffer, B. & Campbell, I.D. Quarterly Rev. Biophys. 29, 119–167 (1996). 10. Pereira, L. et al. Human Mol. Gen. 2, 961–968 (1993). 11. Sakai, L.Y., Keene, D.R. & Engvall, E. J. Cell Biol. 103, 2499–2509 (1986). 12. Collod-Béroud, G. et al. Nucleic Acids Res. 26, 229–233 (1998). 13. Hynes, R.O. Fibronectins. (Springer-Verlag, New York, 1990). 14. Yuan, X., Downing, A.K., Knott, V. & Handford, P.A. EMBO J. 16, 6659–6666 (1997). 15. Yuan, X. et al. Prot. Sci., in the press (1998). 16. Leahy, D.J., Aukhil, I. & Erickson, H.P. 2.0 Cell 84, 155–164 (1996). 17. Downing, A.K. et al. Cell 85, 597–605 (1996). 18. Sakai, L.Y., Keene, D.R., Glanville, R.W. & B?chinger, H.P. J. Biol. Chem. 266, 14763–14770 (1991). 19. Cardy, C.M. & Handford, P.A. J. Mol. Biol. 276, 855–860 (1998). 20. Handford, P.A. et al. J. Biol. Chem. 270, 6751–6756 (1995). 21. Main, A.L., Harvey, T.S., Baron, M., Boyd, J. & Cell 71, 671–678 (1992). 22. Grant, R.P., Spitzfaden, C., Altroff, H., Campbell, I.D. & Mardon, H.J. J. Biol. Chem. 272, 6159–6166 (1997). 23. Copié, V. et al. J. Mol. Biol. 277, 663–682 (1998). 24. Wiles, A.P. et al. J. Mol. Biol. 272, 253–265 (1997). 25. Bax, A. & Tjandra, N. J. Biomol. NMR 10, 289–292 (1997). 26. Tjandra, N. & Bax, A. Science 278, 1111–1114 (1997). 27. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. & Higgins, D.G. Nucleic Acids Res. 25, 4876–4882 (1997). 28. Merritt, E.A. & Murphey, M.E.P. Acta Crystallogr. 50, 869–873 (1994). 29. Kraulis, P.J. J. Appl. Crystallogr. 24, 946–950 (1991). 30. Spitzfaden, C., Grant, R.P., Mardon, H.J. & Campbell, I.D. J. Mol. Biol. 265, 565–579 (1997). In recent years NMR has developed into one of the two leading technologies, together with X-ray crystallography, for determining the three-dimensional struc- tures of folded proteins at atomic resolu- tion. However, NMR is unequaled in its ability to characterize the structure and dynamics of unfolded and partially folded states of proteins. Such non-native pro- tein states do not adopt unique three- dimensional structures in solution but fluctuate rapidly over an ensemble of con- formations. Structural characterization of non-native states is of great interest because of their importance in protein folding, in the transport of proteins across membranes, in cellular processes such as signal transduction, and in the develop- ment of amyloid diseases (Fig. 1). Knowledge of the structure of protein folding intermediates is of central impor- tance for a detailed understanding of pro- tein folding mechanisms. Likewise, characterization of the ensemble of con- formations sampled by denatured pro- teins can provide insights into the nature of the free energy landscape at the very beginning of the folding process. Finally, it is now recognized that many proteins and protein domains are only partially structured or are unstructured under physiological conditions and only become structured upon binding to their biologi- cal targets. Knowledge of the structural propensities of these domains is essential to a proper understanding at the molecu- lar level of their biological functions and interactions. Understanding the fundamental molec- ular mechanisms by which proteins fold into the complex structures required for biological activity remains one of the cen- tral challenges in structural biology. NMR has emerged as an especially important tool for studies of protein folding because of the unique structural insights it can provide into many aspects of the folding process 1 . Applications range from direct or indirect characterization of kinetic folding events (reviewed in the accompa- nying article by Dobson and colleagues 2 ) to structural and dynamic characteriza- tion of equilibrium folding intermediates, partly folded states, peptide fragments, and fully denatured states of proteins. For most proteins, refolding is very rapid and any intermediates formed are populated only transiently and are therefore difficult to study by direct real-time NMR experi- ments. An especially powerful method for obtaining site-specific information on the structure of folding intermediates is hydrogen exchange pulse labeling com- bined with 2D NMR detection 3,4 . A typi- cal experiment involves rapid dilution of denatured protein in H 2 O buffer to initi- ate refolding; the protein is allowed to refold for a short period (typically mil- liseconds–seconds) before mixing with a high pH labeling buffer in D 2 O solution. Amide protons that become involved in hydrogen bonded secondary structures during the refolding period are protected from exchange, whereas amide protons in regions of the polypeptide that remain unfolded are exchanged with deuterium by the labeling pulse. After quenching of exchange and completion of folding, 2D NMR supplement NMR spectra are acquired to identify the protected amides and monitor the pro- gressive stabilization of hydrogen bonded secondary structure during kinetic refold- ing. Although its importance should not be underestimated, the primary limitation of the pulse labeling method is that it pro- vides information only on the location of amide protons that become protected from exchange during folding; the nature of the structures that give rise to protec- tion must be deduced indirectly and ele- ments of structure that are insufficiently stable to protect amides from exchange will go undetected. Fortunately, for some proteins, partially folded states that corre- spond closely to kinetic folding interme- diates can be stabilized at equilibrium, thereby opening the way to direct NMR analysis. In addition, direct NMR studies of fully denatured states provide valuable insights into the nature of the conforma- tional ensemble at the starting point of protein folding, while studies of peptide fragments reveal the intrinsic conforma- tional propensities of the polypeptide chain and identify potential folding initia- tion sites. The challenge of assignments Characterization of unfolded and partial- ly folded states of proteins by NMR pre- sents special challenges because the polypeptide chain in such states is inher- ently flexible and rapidly interconverts between multiple conformations. Consequently, the chemical shift disper- sion of most resonances is poor and sequence-specific assignment of reso- nances is difficult (Fig. 2). Exceptions are the backbone 15 N and 13 C' (that is, car- bonyl carbon) resonances, which are influenced both by residue type and by the local amino acid sequence and there- fore remain well-dispersed, even in fully unfolded states 5,6 . Multi-dimensional triple resonance NMR experiments which establish sequential connectivities through the well-resolved 15 N and 13 C' resonances provide a robust method for obtaining unambiguous resonance assignments 7–9 . The lack of 1 H and aliphatic 13 C chemical shift dispersion for unfolded or partially folded proteins means that it is extremely difficult to assign unambiguously the NOEs that could provide key information on sec- ondary structure and tertiary contacts. Fortunately, recently developed NMR experiments help to overcome this prob- lem by transferring the NOE information to the relatively well-resolved 15 NH or 13 C' resonances 6 . One significant advantage in NMR studies of proteins in highly unfolded states is that resonances are generally nar- row due to the rapid fluctuations of the polypeptide chain. As a result, high quali- ty 2D and 3D spectra can be obtained at surprisingly low protein concentrations; indeed, our own experience is that excel- lent data can be obtained at concentra- tions of 0.1 mM or lower. In addition, sequential assignments can be made using triple resonance experiments that other- wise may be unsuitable for a folded protein of comparable molecular weight. NMR spectroscopy of partially folded proteins can be even more challenging in that resonances are at least as broad as those of native folded proteins but with the limited dispersion found in com- pletely unfolded states; in many cases, NMR studies are impeded by severe resonance broadening that results from conformational fluctuations on a milli- second–microsecond time scale 8,10 . Structural characterization Once resonance assignments have been completed, detailed information on the conformational propensities of the polypeptide chain can be readily derived from chemical shifts, NOEs or coupling constants. The patterns and relative inten- sities of the sequential and medium range NOEs provide information on the propensity of the polypeptide to populate the a and b regions of f,y space or to form ordered helical structures 11,12 . The deviations of chemical shifts from random coil values, especially for 13 Ca and 1 Ha , provide a convenient and sensitive probe of the secondary structural propensities 13 . Main chain coupling constants also give insights into the conformational ensemble populated by an unfolded or partly folded protein 14 . Careful analysis of NMR data for unfolded proteins and peptide frag- ments of proteins has led to a description of the random coil state as a statistical dis- tribution of backbone dihedral angles in f,y space 15 . It is becoming increasingly clear that many unfolded proteins do not simply form statistical random coils but exhibit measurable propensities to popu- late native-like backbone conformations. Secondary structure NMR is particularly useful for determining secondary structural propensities on a residue-by-residue basis in unfolded and partly folded proteins; this is necessary for an understanding of the local interactions that are likely to participate in the initiation of protein folding 1,12 . Information obtained under non-denaturing or very weakly denaturing conditions is most relevant since it more closely relates to the condi- tions prevailing at the start of a protein folding reaction. For many proteins, the unfolded state can only be obtained in solutions of strong denaturants which will have a pronounced effect on the popula- tion of residual structured conformers. The ensemble of conformations sampled by a polypeptide can differ significantly between denaturing and non-denaturing conditions 16,17 and subtle differences in the location of residual structure have been observed for different denaturants 7 . Because their tendency towards structure formation is governed by local rather than long-range interactions, short linear pep- tide fragments of proteins are an ideal vehi- Fig. 1 Schematic diagram summarizing the roles of unfolded, partially folded proteins, and mis- folded proteins in biology. 500 nature structural biology ? NMR supplement ? july 1998 NMR supplement nature structural biology ? NMR supplement ? july 1998 501 cle for elucidation of the intrinsic propen- sities of sequences to fold under non-dena- turing conditions 12 . Studies of peptide fragments of proteins and of proteins that are unfolded under non-denaturing or weakly denaturing conditions show that the intrinsic conformational propensities of the polypeptide backbone frequently reflect the secondary structure found in the native folded protein 8,9,18–20 . In other words, the conformations populated by the unfolded polypeptide are not distrib- uted randomly over the low energy regions of f,y space but are biased in a way that reflects the secondary structural propensi- ties of the local amino acid sequence. In addition, turn-like structures are frequent- ly populated in unfolded states of proteins and in peptide fragments. The observation of conformational preferences for forma- tion of secondary structure or hydropho- bic clusters in short peptides shows that local interactions determined by the amino acid sequence bias the conforma- tional search toward specific structured forms, even in the absence of stabilization by long-range interactions with the remainder of the protein. The molten globule Molten globules are compact states that contain native-like secondary structure but which lack the unique side chain interactions that characterize the tertiary structure of the native protein. Equilibrium molten globules are formed by many proteins under partially denatur- ing conditions. Unfortunately, the confor- mational heterogeneity and complex dynamics of these species frequently result in extremely broad and featureless NMR spectra which make direct NMR structur- al analysis difficult. Nevertheless, numer- ous NMR experiments have been devised to provide structural information on molten globule states, including hydrogen exchange measurements 21 , magnetization transfer experiments 10 , and denaturant titrations to allow residue-specific charac- terization of the hydrophobic core 22 . The molten globule state formed by apomyo- globin at pH 4 is exceptional in the quality of the NMR spectra that it yields; this species is stable at relatively high tempera- ture where there is sufficient internal motion to give rise to narrow resonances and permit use of multidimensional NMR experiments 9 . As a consequence, it has been possible to make complete backbone NMR assignments and obtain highly detailed insights into secondary structure and backbone dynamics. The apomyoglo- bin molten globule is of particular interest and importance because it corresponds closely to an intermediate formed during kinetic refolding of the protein 23 . High quality NMR spectra can often be obtained from partially folded compact species formed by denaturation of pro- teins with alcohols 24 ; however, the rele- vance of such states to protein folding remains to be established. Tertiary structure Characterization of residual tertiary struc- ture in unfolded and partially folded pro- teins is extremely challenging given their intrinsic flexibility. While observation of a long-range NOE between two protons definitively indicates that they must be in close proximity in at least some structures in the conformational ensemble, determi- nation of the nature of the folded struc- ture is difficult unless an extensive network of NOEs can be observed. Newly developed methods for assigning NOE peaks in partly folded states may eventual- ly provide sufficient data in favorable cases to allow a detailed description of highly populated structures 20 . For the 434 repres- sor, for example, enough NOEs were observed to permit distance geometry cal- culations of the three-dimensional struc- a b c Fig. 2 1 H- 15 N HSQC spectra of apomyoglobin at three pHs, illustrating the decrease in resonance dispersion in the 1 H dimension as the protein unfolds. Note that the 15 N dimension remains relatively well-dispersed, an important factor in successful assignment of resonances of unfolded pro- teins. a, pH 2.0 (acid-unfolded state); b, pH 4.0 (equilibrium molten globule intermediate state); c, pH 6.0 ( folded native apoprotein). (Reproduced from ref. 9 with permission). NMR supplement 502 nature structural biology ? NMR supplement ? july 1998 ture of a local hydrophobic cluster 25 . However, it may often prove to be the case that backbone or side chain conforma- tional averaging is so extensive in partially folded states that observation of long- range NOEs is difficult, precluding deter- mination of the folding topology by conventional NOE-based methods. The paucity of long-range NOEs in a denatured fragment of staphylococcal nuclease (termed D 131D ) led Gillespie and Shortle to develop an innovative method to obtain long-range distance constraints by measuring the enhancement of amide pro- ton relaxation induced by paramagnetic nitroxide spin labels 26,27 . Spin labels were coupled to unique cysteine residues intro- duced at 14 different sites on the polypep- tide chain and ~700 long-range distance constraints were derived from measure- ments of T 2 relaxation enhancement (that is, broadening of the resonances of protons close in space to the spin label). The calcu- lated ensemble of structures of this dena- tured state has a global topology that is very similar to that of the native folded pro- tein 26,27 (Fig. 3). These results suggest that the correct folding topology can be estab- lished in denatured states even in the absence of cooperative interactions and a tightly packed hydrophobic core. This spin labeling approach is highly promising and should be generally applicable to the eluci- dation of the folding topologies of other partially folded proteins. Dynamics Unfolded and partially folded proteins are highly flexible. 15 N spin relaxation mea- surements can be used to probe the dynamics of the polypeptide backbone in these species. Interpretation of 15 N relax- ation rates and { 1 H}- 15 N heteronuclear NOEs is not straightforward because the motions are complex and the common assumption of isotropic tumbling with a single correlation time is unlikely to be valid. Nevertheless, valuable insights into the backbone motions can be obtained, using either an extended model-free analysis or reduced spectral density map- ping 28,29 . On the basis of the relaxation measurements reported to date, it is clear that unfolded states of proteins vary con- siderably in their dynamical properties. At one extreme, the backbone fluctuations show little variation as a function of sequence 30 while for other proteins there are clear indications of local interactions that lead to motional restriction 29,31 . For molten globule states and other partially folded species, the molecular motions are highly heterogeneous and relaxation mea- Fig. 3 Ca backbone superposition of residues 56–140 of folded staphylococcal nuclease (thick tube) and five structures calculated for the fragment D 131D (thin line). The three helices are col- ored red and the three b -strands are shown in yellow, yellow-green, and orange. (Reproduced from ref. 27 with permission). Fig. 4 Schematic diagram illustrating the increas- ing restriction of backbone flexibility as myoglo- bin folds to increasingly structured and increasingly compact states, from the acid-unfold- ed state (U acid ), to the pH 4.1 molten globule state (I MG ), to native apomyoglobin (N apo ), and finally to fully folded holomyoglobin (N holo ). Except for holomyoglobin, the structures are purely schematic, shown only to indicate the location of secondary structure in the various partly folded states of apomyoglobin, as indicated by NMR data 9 . The polypeptide fluctuates over an ensem- ble of conformations in all of these states, and no single structure suffices to describe its behavior. The smoothed { 1 H- 15 N} heteronuclear NOE at each residue is shown, on a color scale from dark blue (least flexible) to red (most flexible). The F helix region of apomyoglobin is colored gray, since no NMR information is available for it 9 . (Adapted from ref. 9 with permission). NMR supplement nature structural biology ? NMR supplement ? july 1998 503 surements can provide key insights into the structural organization of such states. For example, 15 N relaxation studies of the pH 4 molten globule state of apomyoglo- bin show that backbone motions are highly restricted within a compact hydrophobic core formed by packing of three helices whilst other parts of the chain remain highly fluctional 9 . Similarly, nuclear spin relaxation studies of a partial- ly folded state of ubiquitin formed in 60% methanol reveal the presence of three loosely coupled secondary structural ele- ments with enhanced mobility relative to the native protein 28 . Intrinsically unstructured proteins It is now recognized that many proteins are intrinsically unstructured under phys- iological conditions 32 . While this has long been known for certain polypeptide hor- mones such as glucagon, there is an increasing awareness that many eukaryot- ic proteins or protein domains involved in signal transduction, transcriptional acti- vation, nucleic acid recognition or cell cycle regulation adopt stable folded struc- tures only upon binding to their molecu- lar targets. Indeed, many genes in eukaryotic genomes contain regions of low sequence complexity that encode bio- logically functional domains which would not be expected to fold spontaneously into ordered structures in the absence of addi- tional stabilizing interactions. NMR is the method of choice for characterization of such domains, many of which probably do not exist as statistical random coils but will exhibit intrinsic conformational propensities that may presage the confor- mation stabilized upon binding. Recent examples of functional yet unfolded pro- tein domains include the anti-sigma factor FlgM 33 , the SH3 domain of the Drosophila signal transduction protein Drk 16 , a fibronectin-binding protein from Staphylococcus aureus 34 , and the kinase inducible transactivation domain (KID) from the transcription factor CREB 35 . In the latter example, NMR analysis shows that the phosphorylated KID domain is intrinsically unstructured but undergoes a folding transition to form a pair of a - helices upon binding its target domain from the CREB binding protein (CBP) 35 . Future perspectives There can be little doubt that NMR will continue to make major contributions to the understanding of the molecular mechanisms of protein folding and mis- folding, provide insights into the subtle relationship between amino acid sequence and protein structure, and lead to new understanding of the behavior and biological function of intrinsically unstructured protein domains. Clearly it is of vital importance to understand the kinetics of collapse and structure forma- tion that accompany the folding process, and NMR has important contributions to make in this field 2 . However, the intrinsically long time scale of the NMR experiment makes real-time kinetic observations problematic. For many pro- teins, equilibrium NMR studies can pro- vide valuable and extensive information on the conformational propensities of unfolded or partly folded states, informa- tion that is directly applicable to an understanding of the folding process. Recent studies of apomyoglobin pro- vide an illustrative example of the funda- mental insights into protein folding mechanisms that can potentially be obtained from equilibrium NMR experi- ments. By careful manipulation of the solution conditions, several states of apomyoglobin that differ in structural content and degree of compaction can be stabilized for NMR analysis 9 . In this way, detailed insights can be obtained, at the level of individual residues, into the pro- gressive accumulation of secondary struc- ture and increasing restriction of backbone dynamics as the chain collapses during folding to form more compact states (Fig. 4). Studies such as this are only a beginning, and the prospects for obtain- ing even deeper insights into the folding topology, hydration, and dynamics of the compact, partially folded states formed by apomyoglobin and other proteins are excellent. Future work in the area will be aided in no small part by the continued development of novel NMR methodolo- gies and improvements in NMR instru- mentation, especially the anticipated development of spectrometers operating at or above 900 MHz which will provide greater sensitivity and dispersion and, as a consequence, more detailed insights into the nature of unfolded and partially fold- ed states. Acknowledgments We thank D. Eliezer, P. Jennings and S. Cavagnero for assistance with preparation of the figures. This work was supported by grants from the National Institutes of Health. H. Jane Dyson and Peter E. Wright are at the Department of Molecular Biology and Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, USA. Correspondence should be addressed to P.E.W. email: wright@scripps.edu or H.J.D. email: dyson@scripps.edu 1. Dyson, H.J. & Wright, P.E. Annu. Rev. Phys. Chem. 47, 369–395 (1996). 2. Dobson, C.M. Nature Struct. Biol. 5, 504–507 (1998). 3. Roder, H., El?ve, G.A. & Englander, S.W. Nature 335, 700–704 (1988). 4. Udgaonkar, J.B. & Baldwin, R.L. Nature 335, 694–699 (1988). 5. Braun, D., Wider, G. & Wüthrich, K. J. Am. Chem. Soc. 116, 8466–8469 (1994). 6. Zhang, O., Forman-Kay, J.D., Shortle, D. & Kay, L.E. J. Biomol. NMR 9, 181–200 (1997). 7. Logan, T.M., Thériault, Y. & Fesik, S.W. J. Mol. Biol. 236, 637–648 (1994). 8. Alexandrescu, A.T., Abeygunawardana, C. & Shortle, D. Biochemistry 33, 1063–1072 (1994). 9. Eliezer, D., Yao, J., Dyson, H.J. & Wright, P.E. Nature Struct. Biol. 5, 148–155 (1998). 10. Baum, J., Dobson, C.M., Evans, P.A. & Hanley, C. Biochemistry 28, 7–13 (1989). 11. Dyson, H.J. & Wright, P.E. Ann. Rev. Biophys. Biophys. Chem. 20, 519–538 (1991). 12. Wright, P.E., Dyson, H.J. & Lerner, R.A. Biochemistry 27, 7167–7175 (1988). 13. Wishart, D.S. & Sykes, B.D. Meth. Enz. 239, 363–392 (1994). 14. Smith, L.J. et al. J. Mol. Biol. 255, 494–506 (1996). 15. Smith, L.J., Fiebig, K.M., Schwalbe, H. & Dobson, C.M. Folding & Design 1, R95–R106(1996). 16. Zhang, O. & Forman-Kay, J.D. Biochemistry 36, 3959–3970 (1997). 17. Pan, H., Barbar, E., Barany, G. & Woodward, C. Biochemistry 34, 13974–13981 (1995). 18. Dyson, H.J., Merutka, G., Waltho, J.P., Lerner, R.A. & Wright, P.E. J. Mol. Biol. 226, 795–817 (1992). 19. Dyson, H.J. et al. J. Mol. Biol. 226, 819–835 (1992). 20. Zhang, O., Kay, L.E., Shortle, D. & Forman-Kay, J.D. J. Mol. Biol. 272, 9–20 (1997). 21. Hughson, F.M., Wright, P.E. & Baldwin, R.L. Science 249, 1544–1548 (1990). 22. Schulman, B.A., Kim, P.S., Dobson, C.M. & Redfield, C. Nature Struct. Biol. 4, 630–634 (1997). 23. Jennings, P.A. & Wright, P.E. Science 262, 892–896 (1993). 24. Harding, M.M., Williams, D.H. & Woolfson, D.N. Biochemistry 30, 3120–3128 (1991). 25. Neri, D., Billeter, M., Wider, G. & Wüthrich, K. Science 257, 1559–1563 (1992). 26. Gillespie, J.R. & Shortle, D. J. Mol. Biol. 268, 158–169 (1997). 27. Gillespie, J.R. & Shortle, D. J. Mol. Biol. 268, 170–184 (1997). 28. Brutscher, B., Brüschweiler, R. & Ernst, R.R. Biochemistry 36, 13043–13053 (1997). 29. Farrow, N.A., Zhang, O., Forman-Kay, J.D. & Kay, L.E. Biochemistry 36, 2390–2402 (1997). 30. Frank, M.K., Clore, G.M. & Gronenborn, A.M. Prot. Sci. 4, 2605–2615 (1995). 31. Alexandrescu, A.T. & Shortle, D. J. Mol. Biol. 242, 527–546 (1994). 32. Plaxco, K.W. & Gross, M. Nature 386, 657–659 (1997). 33. Daughdrill, G.W., Hanely, L.J. & Dahlquist, F.W. Biochemistry 37, 1076–1082 (1998). 34. Penkett, C.J., Redfield, C., Dodd, I., et al. J. Mol. Biol. 274, 152–159 (1997). 35. Radhakrishnan, I. et al. Cell 91, 741–752 (1997).