sign in sign up
<<
Home , Biomolecular structure, Folding funnel, Hydrophobic collapse, Protein, Protein folding

Brazilian Journal of Medical and Biological Research (2001) 34: 419-435 folding and its implications in and medicine 419 ISSN 0100-879X

Protein folding: a perspective for biology, medicine and

J.M. Yon Institut de Biochimie, Biophysique Moléculaire et Cellulaire, UMR CNRS, Université de Paris-Sud, Orsay, France

Abstract

Correspondence At the present time, is an extremely active field of Key words J.M. Yon research including aspects of biology, chemistry, , com- · Protein folding Institut de Biochimie, Biophysique puter science and physics. The fundamental principles have practical · landscape Moléculaire et Cellulaire · Misfolding and UMR CNRS, Université de Paris-Sud applications in the exploitation of the advances in research, in Orsay the understanding of different and in the design of novel France with special functions. Although the detailed mechanisms of Fax: +1-6985-3715 folding are not completely known, significant advances have been E-mail: jeannine.yon@.u-psud.fr made in the understanding of this complex process through both experimental and theoretical approaches. In this review, the Presented at the XIV Annual Meeting of concepts from Anfinsen’s postulate to the “new view” emphasizing of the Federação de Sociedades de Biologia Experimental, Caxambu, MG, the concept of the of folding is presented. The main Brazil, August 25-28, 1999. rules of protein folding have been established from experi- ments. It has been long accepted that the in vitro refolding process is a good model for understanding the mechanisms by which a nascent polypeptide chain reaches its native conformation in the cellular Received June 1, 2000 environment. Indeed, many denatured proteins, even those whose Accepted February 5, 2001 bridges have been disrupted, are able to refold spontane- ously. Although this assumption was challenged by the discovery of molecular chaperones, from the amount of both structural and func- tional information now available, it has been clearly established that the main rules of protein folding deduced from in vitro experiments are also valid in the cellular environment. This modern view of protein folding permits a better understanding of the aggregation processes that play a role in several pathologies, including those induced by and Alzheimer’s disease. and de novo with the aim of creating proteins with novel functions by application of protein folding rules are making significant progress and offer perspectives for practical applications in the development of pharmaceuticals and medical diagnostics.

Introduction makes the of the problem all the more urgent. the is only the An important challenge in molecular bi- first step; it is essential to know the , ology is finding the rules that determine how the and the of the nascent polypeptide chain acquires its three- that is coded by the gene. For diagnosis of the dimensional and functional structure. Indeed, true cause of a disease, and for an approach the rapid progress in genome sequencing, to a rational treatment, after having located including that of the genome, which the gene responsible for a disease, it is essen- has aroused great interest in medical circles, tial to know the structure of the protein for

Braz J Med Biol Res 34(4) 2001 420 J.M. Yon

which it . folded denatured and reduced Protein folding, the second of into the fully active . He stated the the genetic message, completes the informa- fundamental principle of protein folding in tion transfer from DNA to the active protein 1973: the folding of a protein is determined product. In other words, for a complete un- by its amino sequence. These historical derstanding of this process, it is necessary to aspects have been reviewed by Ghélis and decipher the folding , the second part of Yon (2). The evidence gathered over many the genetic message: DNA®RNA®poly- years supports this principle even for folding chain ¾ ?®active protein. assisted by molecular chaperones. There are about 100,000 proteins in the The fundamental questions are the fol- and about 1011 in all organ- lowing. How does the sequence code for the isms. To become biologically active these pro- fold, given that the backbone of all proteins teins must fold into a stable three-dimensional has the same composition; in other words, structure. In spite of the great diversity of how do the side chains dictate the overall proteins, the number of folds is relatively small, fold? How does a given sequence find its less than 700 observed to date, while protein specific native structure in a finite time among domains exhibit only 32 different architec- the enormous number of possible conforma- tures according to a recent classification (1). tions that a polypeptide chain could adopt? Nature has created complexity through the How is the folding process initiated and combination of a small number of simple ele- what is (are) the pathway(s) of folding? And ments, such as the two most common elements last, are the main rules of protein folding of secondary structure observed in proteins: deduced from in vitro studies valid for fold- a-helices and ß-strands. ing in vivo? The question of the mechanism of pro- At the present time, protein folding is an tein folding has intrigued scientists for many extremely active field of research involving decades. As far back as 1929, a time when aspects of biology, chemistry, biochemistry, nothing was known about , computer science and physics. The funda- Wu had analyzed the reverse of the folding mental principles have practical applications process, denaturation. Shortly thereafter, in the exploitation of the recent advances in Northrop in 1932, and Anson and Mirsky in genome research, in the understanding of 1934-35 succeeded in reversing the denatur- several pathologies and in the design of novel ation of several proteins, such as hemoglo- proteins with special functions. Although bin, chymotrypsinogen, and trypsinogen. the detailed mechanisms of folding are not During the decade between 1950 and 1960, completely known, significant advances have several thermodynamic studies of the un- been made in the understanding of this com- folding-refolding process and the role of plex process through both experimental and noncovalent in protein stability theoretical approaches (3). were published by several authors, among them, Linderstrøm-Lang, Lumry, Klotz and The fundamentals of protein folding Kauzmann. In 1959, Kauzmann proposed from the Anfinsen postulate to the that the is the driving new view force in directing the folding process. Later, the determination of the three-dimensional The Anfinsen postulate and the Levinthal of proteins by X-ray diffraction paradox provided a new basis on which to study the folding process. Significant progress began The remarkable achievement of C. to be made when Anfinsen successfully re- Anfinsen and his group in refolding dena-

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 421 tured and reduced ribonuclease into a fully tal observations (6, and references therein) active enzyme marked the beginning of the have been proposed. modern era of the protein folding problem. The classical nucleation-propagation From his results, the author concluded that model, which applies to helix-coil transi- “all the information necessary to achieve the tions, involves a nucleation step followed by native conformation of a protein in a given a rapid propagation, the limiting step being environment is contained in its the nucleation process. This model has been sequence” (4). The thermodynamic control proposed to explain the folding of ribonu- of protein folding is a corollary to the clease A but was forsaken after new kinetic Anfinsen postulate; it means that the native studies of the refolding of ribonuclease were structure is at a minimum of the Gibbs free performed (6). More recently a nucleation- energy. This statement was discussed by condensation model, different from the clas- Levinthal in a consideration of the short time sical one, has been proposed by Fersht (7). required for the folding process in vitro as This model proposes a mechanism involving well as in vivo. Indeed, for a 100-amino acid a weak local nucleus which is stabilized by polypeptide chain, if we assume only two long range interactions. possible conformations for each , A stepwise sequential and hierarchical there are 1030 possible conformations for the folding process, in which several stretches chain as a whole. If only 10-11 second is of structure are formed and assemble at dif- required to convert one conformation into ferent levels following a unique route, has another, a random search of all conforma- been supported by several authors for many tions would require 1011 years, an irrealistic years (6,8). According to this model, the first time in a biological context where the fold- event, nucleation, is followed by the forma- ing time is of the order of seconds or min- tion of secondary structures which associate utes. Thus, it is clear that evolution has to generate supersecondary structures, then found an effective solution to this combina- domains and eventually the active mono- torial problem. This is referred to as the mer; the association of domains induces the Levinthal paradox and has dominated dis- last conformational refinements which gen- cussions for the last 30 years. Different erate the functional properties. Such a hier- mechanisms have been suggested in order to archy of protein folding corresponds to the solve the Levinthal paradox. Among them is hierarchy of protein structure. Similarly, the Wetlaufer’s proposed model in which pro- framework model assumes that the second- tein folding is under kinetic control rather ary structure is formed in an early step of than thermodynamic control. This states that folding, before the tertiary structure, empha- the protein is trapped in an energetic mini- sizing the role of short range interactions in mum which is not the global minimum, a directing the folding process (9). high energetic barrier preventing the protein A modular model of folding was sug- from reaching the latter (see Ref. 2). gested on the basis of the three-dimensional structures of proteins. This model assumes Folding models and pathways of protein that not only domains, but also subdomains folding can be considered as folding units which fold independently into a native structure, In order to understand how the polypep- forming structural modules that assemble to tide chain could overcome the Levinthal para- yield the native protein (10,11). dox, different folding models arising from The -collision model of folding theoretical considerations (5, and references was developed in 1976 by Karplus and therein), folding simulations, or experimen- Weaver and reconsidered in 1994 in the

Braz J Med Biol Res 34(4) 2001 422 J.M. Yon

of more recent experimental data (12). In der equilibrium conditions. The two-state such a model, nucleation occurs simulta- approximation, which applies to very coop- neously in different parts of the polypeptide erative transitions, is frequently valid for chain generating microstructures which dif- small proteins. Such an approximation al- D fuse, associate and coalesce to form sub- lows the determination of Go, the free en- structures with a native conformation. These ergy of denaturation. This value varies be- microstructures have a lifetime controlled tween 5 and 15 kcal/mol for most proteins by segment diffusion, so the folding of a studied thus far, indicating a relatively low polypeptide chain containing 100 to 200 stability (for reviews, see 2,15). amino can occur within a very short The existence of intermediates has been time, less than a second. According to this shown from kinetic studies for most proteins model, folding occurs through several diffu- even when the two-state approximation de- sion-collision steps. scribes the overall denaturation process. For The model implies many proteins, monophasic unfolding kinet- that the first event of protein folding consists ics and multiphasic refolding kinetics are of a collapse via long range hydrophobic observed. This is one of many ways in which interactions and occurs before the formation the experimental evidence proves the occur- of a secondary structure (13). The idea origi- rence of intermediates in the folding path- nates from the work of Kauzmann consider- way. The structural characterization of such ing the hydrophobic effect as the driving intermediates is a prerequisite to solving the force in protein folding and stabilization. folding problem. However, while kinetic Later, Dill and co-workers proposed that the studies demonstrate the existence of inter- formation of stretches of secondary struc- mediates and their location on the folding tures occurs simultaneously with the hydro- pathway, these intermediates are often too phobic collapse. poorly populated to obtain detailed struc- The jigsaw puzzle model was introduced tural information. Two major impediments in 1985 by Harrison and Durbin (14). This to characterizing these species are thus the model admits the existence of multiple fold- high and the rapidity of the ing routes to reach a single solution. Accord- process, especially in the early events of ing to this hypothesis, the identification of protein folding. folding intermediates represents a kinetic In spite of these inherent difficulties, much description rather than a structural one, each effort has been devoted to characterizing intermediate consisting of heterogeneous spe- these transient species. Kinetic trapping of cies in rapid equilibrium. It was the subject intermediates during the refolding of disul- of controversy until recently but presents fide-bridged proteins was developed for ly- some similarities with the diffusion-colli- sozyme (16) and used for bovine pancreatic sion model proposed by Karplus and Weaver. inhibitor (BPTI) (17,18). An elegant method using differential chemical labeling Detection and characterization of has been elaborated by Ghélis (19) and ap- intermediates in protein folding plied to the refolding of . In the past decade, technological advances The unfolding-refolding transition under have improved our approaches to the study equilibrium transition has often been de- of the protein folding process. Several meth- scribed as a two-state process in which only ods have been developed allowing some of the unfolded and the native species are sig- the intermediates to be characterized, par- nificantly populated. The intermediates are ticularly stopped-flow mixing devices cou- generally unstable and poorly populated un- pled to , and NMR using

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 423 rapid hydrogen-deuterium exchange associ- served in the early steps of the folding pro- ated with a mixing system allowing for the cess are not always identical to those ob- pulse labeling of transient species. This served in the native structure. In the refold- method is very informative, yielding resi- ing of the ß2-subunit of synthase due-specific information (20-22). Protein (29), a rapid formation of non-native sec- engineering has also been successfully em- ondary structures precedes a reorganization ployed to stabilize intermediates or to probe yielding the native structure. Similarly, non- particular regions of a protein during the native secondary structures are formed dur- folding process (23-25). Another approach ing the refolding of ß-lactoglobulin (30). frequently used is the study of protein frag- Even in the case of the paradigmatic molten ments (10,26). Transient folding species have globule of a-lactalbumin, some differences been found to accumulate at low pH for have been observed by NMR . several proteins, and especially for a-lactal- This appears as a heteroge- bumin, permitting the study of their struc- neous species. The helical domain is struc- tural properties. tured, but has highly fluctuating hydropho- The formation of secondary structures in bic interactions, whereas the ß-sheet is rather the early steps of protein folding has been disordered. Furthermore, the side chains of observed for many proteins (24,27). Such Tyr103, Trp104 and His107 are packed early species with a high content of second- within a hydrophobic cluster in a region that ary structures were named “the molten glob- differs in structure from the native one (31). ule” by Ohgushi and Wada (28). Ptitsyn (27) The folding pathway of hen lyso- has suggested that it is a general intermediate zyme includes transient steps corresponding in the folding pathway of proteins. The lit- to a reorganization of secondary structures erature being rather confusing concerning (32). the structural characteristics of the molten An intermediate preceding the molten globule state, Goldberg and colleagues (29) globule state has been identified by Ptitsyn have introduced the term “specific molten (27), and Uversky and Ptitsyn (33). This globule”, and defined its characteristics. The species, less compact than a molten globule, specific molten globule is a rather compact has a significant secondary structure content intermediate with a high content of native but smaller than that of a molten globule, and secondary structure, but a fluctuating ter- displays hydrophobic regions accessible to a tiary structure. It contains an accessible hy- . It has been called a “pre-molten drophobic surface susceptible to binding a globule” by Jeng and Englander (34). Its hydrophobic dye, aniline naphthalene sul- occurrence has been observed during the fonate. Since the tertiary structure is not cold denaturation of ß-lactamase and car- stabilized, the aromatic residues can rotate bonic anhydrase, and also during the refold- in a symmetrical environment and are acces- ing of several proteins (35). The rapid for- sible to the solvent, as assessed by the ab- mation of transient intermediates, either mol- sence of near UV circular dichroism. The ten or pre-molten, or both, with a high con- formation of a molten globule as an early tent of secondary structure and a small folding intermediate has been reported for amount of fluctuating tertiary structure, is several proteins, among them a-lactalbu- supported by a great number of observations min, carbonic anhydrase, ß-lactamase, the (22,35). However, since these transient states a - and ß2-subunits of tryptophan synthase, are formed within the dead-time of a stopped- bovine growth , and phosphoglyc- flow device, it is possible that their forma- erate (24,27,29). tion might be preceded by an earlier event. However, the secondary structures ob- According to another point of view, the

Braz J Med Biol Res 34(4) 2001 424 J.M. Yon

first event in the folding of a polypeptide used to study the refolding of cold-denatured chain consists of a hydrophobic collapse proteins. Other rapid techniques such as nano- preceding the formation of secondary struc- second laser photolysis, optical electron trans- ture or occurring simultaneously, and being fer and dynamic NMR methods have been followed by a rearrangement of a small num- reported. They allow detection of very fast ber of condensed states. This view empha- events which occur on a time scale of less sizes both the hydrophobic effect and the than a microsecond. For comparison, it has role of long range interactions in the initia- been shown that the time scale for helix tion of the folding process. Several experi- formation is 10-7 s, and of the order of 10-6 mental data are consistent with a hydropho- for the formation of a ß-hairpin or a 10- bic collapse during the early stages of fold- residue loop (37). The upper limit for the ing. For example, residual microstructures rate of protein folding has been evaluated to persisting during the denaturation process be around 1 µs (38), in agreement with theo- have been characterized for several proteins. retical estimates (39). The initial collapse of These microstructures may be involved in apomyoglobin into a compact state is com- the folding process as hydrophobic nucle- plete in less than 20 µs. Moreover, the use of ation centers. They have been observed in a laser T-jump method has allowed the de- the folding of 434 repressor, in FK506 bind- tection of two very fast phases during the ing protein, in staphylococcal nuclease, in folding of this protein. The first one occur- dihydrofolate reductase, and in a peptide ring in 250 ns is a local collapse around from BPTI (for a review, see 15). In phos- Trp14 and consists of non-native hydropho- phoglycerate kinase and two of its mutants bic contacts and helix formation. During the containing only one of the two tryptophan second phase within 5 µs, helix A makes residues, residual microstructures around contact with two other helices, H and G. W308 and W333 have been detected by These two rapid phases are followed by a emission spectroscopy during slower final phase of 0.9 s generating the the guanidinium/HCl-induced unfolding of native protein as shown by Ballew et al. (40). the protein; these microstructures consist of The refolding of c also starts by hydrophobic clusters (25). a rapid collapse occurring within 50 µs fol- Classical rapid mixing techniques such lowed by a massive chain condensation (41). as stopped-flow, continuous flow and It appears therefore that the very fast quenched-flow are limited to the millisec- events of protein folding consist of a hydro- ond time scale, preventing analysis of the phobic collapse accompanied or not by the events occurring within the initial burst phase formation of secondary structures, depend- of folding. Most of the secondary structures ing on the protein. Intermediate events on are formed within the dead-time of a stopped- the folding time scale occur after the forma- flow device. This is particularly inconve- tion of the molten globule and before the nient as a crucial part of the folding problem rate-limiting step of the folding process which is the characterization of the very early inter- generates the native and functional struc- mediates. Recent technical advances to im- ture. In these intermediate phases, the ap- prove the resolution time of kinetic studies pearance of - or -binding sites have been made (36). Submillisecond mix- can be observed. For example, in lactalbu- ing techniques have been developed and min, Ca2+-binding sites appear before com- applied to the refolding of cytochrome c. pletion of the native structure (42). Non-mixing techniques such as the classical In the final rate-limiting step, the protein T-jump, nanosecond infrared laser-induced achieves its native conformation with the T-jump and picosecond T-jump have been of functional properties. These

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 425 final events correspond to the precise order- semble. Rather, the interactions between ing of the elements of secondary structure, domains are required during the folding pro- the correct packing of the hydrophobic core, cess to allow the structural adjustments that the correct domain pairing in multidomain yield a functional protein. proteins, the reshuffling of disulfide bonds, In many proteins, the N and C termini are cis-trans isomerization, and subunit spatially adjacent in the folded state. In phos- assembly in oligomeric proteins. For several phoglycerate kinase, for example, circular proteins, the rate-limiting step consists of the permutations have been obtained by protein reorganization of misfolded species (for re- engineering that introduce a discontinuity views, see 2,6,15,43). Figure 1 illustrates a into either one or the other domain. Such a folding pathway with kinetic competition change can modify the sequence of events in between correct folding and a side reaction the folding process, but does not prevent the leading to the formation of aggregates. protein from achieving a native and func- tional conformation. However, it significantly Domains and subdomains in protein folding decreases the stability of the enzyme (by 4 kcal/mol), suggesting that the continuity of Domains are compact substructures Figure 1 - Illustration of a sche- within a protein . They have been Unfolded Molten Folded matic pathway for protein fold- Intermediate protein globule protein ing with a side reaction and the considered as folding units by Wetlaufer, formation of aggregates. forming structural modules that fold inde- pendently and assemble to generate the na- Misfolded tive structure. Many experimental data have species shown that domains behave as autonomous folding units, as reviewed by Ghélis and Yon (2), Yon (15) and Jaenicke (8,43). In our Aggregates laboratory, we have studied the role of struc- tural domains in whose structure is organized into two do- mains of approximately the same size (Fig- ure 2). The independently expressed engi- neered N- and C-domains have quasi-native structures and are capable of refolding coop- eratively. The C-domain binds the ATP sub- strate with the same affinity as the native protein. However, even though these iso- lated domains can refold independently, it has not proved possible via a variety of experimental procedures to reproduce func- tionally active protein from the two separate domains. The ability of independently folded domains to associate to yield a functional protein has been observed only with a few proteins, such as , elastase, and methionyl-t-RNA synthetase (reviewed in Ref. 15). Thus, it seems that, for many pro- teins, the correct folding of isolated domains Figure 2 - Structure of phosphoglycerate kinase. The two are colored in is not sufficient to yield a functional en- and Cys97 in magenta.

Braz J Med Biol Res 34(4) 2001 426 J.M. Yon

the domains is required for the stability but review, see Ref. 15). not for the correct folding of phosphoglycer- ate kinase (15). It is likely that during evolu- The new view: the energy landscape and the tion structural patterns of proteins consisting of continuous domains have been selected on the basis of thermodynamic stability. Many phenomenological models have However, proteins have enough plasticity to been thus proposed over the years; more find their functional fold, even when the recently a unified model of protein folding continuity of the domains has been artifi- based on the effective energy surface of the cially disrupted. polypeptide chain has emerged. This so- It has been proposed that subdomains, called new view of protein folding arises which represent compact regions smaller than from theoretical studies. Although simpli- domains within a protein, might also fold fied to take into account the computational autonomously, forming folded modules that limitations, several models have been pro- assemble to generate the native protein (11). posed to overcome the Levinthal paradox by These subdomains are proposed to be con- simulation of folding from to densed states in which the close packing of the native structure. There are essentially the in the hydrophobic core of the two approaches, lattice models and molecu- molecule has not yet taken place. Oas and lar dynamics simulations. Kim have reported that fragments from BPTI In the lattice models, the protein chain is corresponding to subdomains fold autono- represented as a string of beads on a two- mously and then associate. The results ob- dimensional square lattice, or on a three- tained on the folding and complementation dimensional cubic lattice (Figure 3). The of fragments from are also consist- interactions between residues (the beads) ent with a modular model. In contrast, this provide the energy function for Monte Carlo model cannot account for the folding of simulations. In such models, the aim is not to chymotrypsin inhibitor 2. Likewise, examine the folding of a particular amino fragments smaller than a domain cannot be acid sequence. Instead, the idea is to include considered as folding units in the SH2 do- the most essential features of proteins, i.e., main of proteins p60 and p85, in staphylo- the heterogeneous character of the interac- coccal nuclease, in tryptophan synthase, or tions (hydrophobic and polar) and the exist- in cytochrome c. Several pairs of structural ence of long range interactions to explore the fragments from phosphoglycerate kinase general characteristics of the possible folds. have been obtained. Some of them, smaller Such an approach has played an important than a domain, have a quasi-native structure; role in science. Lattice models were but their unfolding transition exhibits a very first applied to protein folding by Go and co- low cooperativity, as discontinuities in the workers and simple exact models for pro- ß-sheet regions perturb the folding process. teins were initiated by Dill and co-workers Furthermore, several pairs of adjacent frag- (for a review, see 44) and have been used by ments have been found to give a functional several researchers. In these models, the na- complementation. Among them, at least one tive state corresponding to the energy mini- of the two fragments was not significantly mum is a compact globule with a native fold folded when isolated. This demonstrates that which has the characteristics of the experi- the association of individual fragments with mentally observed molten globule with sec- a non-native structure can favor subsequent ondary structure, but not tertiary interac- rearrangements to functional native struc- tions. From the lattice simulations, insights tures through long range interactions (for a into possible folding scenarios have been

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 427 obtained, providing a basis for exploring the during the past few years from both experi- general characteristics of folding for real ment and theory through the use of simpli- proteins. The exploration of such models fied mechanical models and is illustrated by supplies useful information that can be sub- the concept of the folding funnel introduced mitted to experimental tests. by Wolynes and co-workers (46). The model Using such a 3 x 3 x 3 cubic lattice model, is represented in terms of an energy land- Karplus and co-workers (45) have studied scape and describes the thermodynamic and the folding of a 27mer (a polypeptide of 27 kinetic behavior of the transformation of an amino acids). Two hundred random amino ensemble of unfolded to a pre- acid sequences were generated, each charac- dominantly (Figure 4). As un- terized by an matrix for all pairs derlined by Wolynes et al. (46): “to fold, a of beads. Each was submitted to Monte protein navigates with remarkable ease Carlo simulations. The results indicated that through a complicated energy landscape”. A 30 of the 200 sequences found their native wide variety of folding behavior emerges state very easily, whereas 146 never found from the energy landscape depending on the the native state. The difference between these energetic parameters and conditions. The two sets of sequences, strongly folding and authors have suggested that the folding rate non-folding, was not to be found in their is slowed by ripples in the energy landscape structural features. Instead, it was found corresponding to local minima populated by in the large energy gap between the native transiently stable intermediates. According and the excited states in the strongly folding to this model, there are parallel micropath- sequences. Under these conditions, the fold- ways, each individual polypeptide chain fol- ing can be very fast. The energy gap concept lowing its own route. Towards the bottom of focuses on small fragments and their role the folding funnel, the number of protein as early folding units. Thus, the Levinthal conformations decreases as does the chain paradox can be overcome by the simulta- . The steeper the slope, the faster the neous and fast formation of microstructures folding. In Figure 5 (left), an idealized smooth in several regions of the protein, restricting funnel is represented in three dimensions; in the number of possible conformations dur- this case, the protein can fold very quickly ing their subsequent assembly to the final exhibiting a two-state behavior. In contrast, structure. in Figure 5 (right), a rugged energy land- The so-called “new view” has evolved scape with kinetic traps formed by energy

Figure 3 - Lattice models. A, ABSquare; B, cubic.

Braz J Med Biol Res 34(4) 2001 428 J.M. Yon

barriers is represented; in this case folding Unfolded states can be multistate and slower (47). When local energy barriers are high enough, pro- tein molecules can be trapped and possibly Entropy aggregate. The new view has progressively replaced the idea of folding pathways. The energy landscape metaphor provides a conceptual framework for understanding both two-state and multistate kinetics, and also protein mis- folding and subsequent aggregation. This view is similar to the jigsaw puzzle model Q (14).

Energy Molten globule Many experimental results are consistent states with this view. There is an increasing amount of evidence showing that the initially ex- tended polypeptide chain folds through a heterogeneous population of partially folded intermediate species in fluctuating equilib- rium. It has been reported that hen egg lyso- zyme and cytochrome c refold according to parallel alternative pathways (45). Hetero- geneous species have also been detected during the refolding of phosphoglycerate ki- nase. Furthermore, for the first time, it was Native state shown that transient oligomeric species are formed very rapidly, but that they neverthe-

Figure 4 - A schematic two-dimensional representation of the folding funnel. The width of less yield monomeric native protein during the funnel represents entropy, and depth the energy. Q is the fraction of native contacts the slow step of folding. The associations indicated for each collection of states. occur via the N-terminal domain (48). More

Figure 5 - Three-dimensional rep- resentation of folding funnels (see text) according to Dill and Chan (47; reproduced with per- mission).

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 429 recently, transient oligomeric species have mits the release of the folded protein. The also been reported during the folding of the molecular chaperones by their transient as- 102-residue monomeric protein U1A, and sociation with nascent, stress-destabilized or chymotrypsin inhibitor 2 (49). A transient translocated proteins, have a role in prevent- dimer has been characterized by small angle ing improper folding and subsequent aggre- X-ray scattering in the case of the refolding gation. They do not interact with folded of cytochrome c (50). proteins. They transiently associate with an From the convergence of theoretical and early intermediate on the folding pathway, experimental studies, a unified view of the probably a molten globule or a pre-molten folding process has progressively emerged, globule, by hydrophobic interactions. They providing also an explanation of the aggre- do not carry information capable of directing gation processes observed in several patholo- a protein to assume a structure different from gies. This view has replaced the classical that dictated by the amino acid sequence. one of a hierarchical and sequential pathway Furthermore, they increase the yield but not and is now quite generally accepted. Even the rate of folding reactions; in this respect Baldwin recognized in 1994 (51): “In retro- they do not act as catalysts. spect, it seems that Harrison and Durbin may Most small chaperones bind transiently have had the right idea”. to small hydrophobic regions of nascent poly- peptide chains, preventing aggregation and Folding in the biological environment premature folding. In contrast, large chaper- ones such as the GroEL-GroES system in The main rules of protein folding have or TriC in completely been established from in vitro experiments. sequester partially folded molecules in a cen- It has been accepted that the in vitro refold- tral cage. This cage is provided by the ing process is a good model for understand- heptameric double ring of GroEL and is ing the mechanisms by which a nascent poly- capped by GroES to prevent the premature peptide chain reaches its native conforma- release of the folding protein. This cage is tion in the cellular environment. Indeed, many large enough to accommodate proteins up to denatured proteins, even those whose disul- about 70 kDa. Figure 6, according to Wang fide bridges have been disrupted, are able to and Weissman (52), illustrates the GroEL refold spontaneously. This view was never- reaction . The non-native protein binds theless challenged by the discovery of mo- to the apical domains of the unoccupied lecular chaperones in 1987. From that mo- upper ring of GroEL-GroES. ATP and GroES ment on, a great amount of information has bind to the same ring sequestering the pro- been gathered. The structural data now avail- tein. The binding of GroES induces a large able together with functional studies led to in GroEL and ATP significant progress in the understanding of induces a conformational change the mechanisms used by the ma- in the bottom ring allowing it to bind a non- chinery to assist protein folding. There are native protein. This promotes subsequent now more than 20 protein families that have binding of ATP and GroES in the lower ring, been described as molecular chaperones. It disrupting the upper complex and ejecting is now well established that neither further GroES and releasing the protein. If the pro- information nor external energy is required tein has not reached the native state, it is to achieve the correct folding in chaperone- subjected to a new cycle. The hydrolysis is assisted systems. The ATP hydrolysis by required in some cases but not all for the GroEL is only used to induce the conforma- release of the protein. The energy of ATP is tional change of the chaperone which per- used to provoke a conformational change in

Braz J Med Biol Res 34(4) 2001 430 J.M. Yon

the chaperone allowing the release of the play a helper role in the folding of proteins in protein. As concluded by Hartl: “The role of vivo. Protein disulfide isomerase, an abun- ATP hydrolysis is mainly to induce confor- dant component of the lumen of the endo- mational change in GroEL that results in plasmic reticulum, catalyzes the formation GroES cycling at a physiologically relevant of disulfide bonds in secretory proteins. An- rate”. Therefore molecular chaperones as- other enzyme, peptidyl-prolyl-cis-trans isom- sist the in vivo folding without violation of erase, catalyzes the cis-trans isomerization Anfinsen’s postulate. The same mechanisms of X-Pro peptide bonds. These only direct the in vivo and in vitro folding pro- accelerate the process, which can also be cesses. achieved in their absence in vitro under ad- In vitro experiments on protein folding equate conditions and are thus adequately have shown that misfolding and subsequent classified as enzymes. aggregation result from a kinetic competi- tion between correct folding and an off- and its pathway process. It results from the occur- implications in human rence of local minima in the folding funnel described in the previous section. When the Protein aggregation is a widespread phe- formation of the correct structure is kineti- nomenon that can occur under particular cally favored in vivo, molecular chaperones conditions or following certain in are not required. This was shown not only the polypeptide chain (54). It occurs com- for monomeric, but also for dimeric proteins monly when proteins are overexpressed in such as the p22 Arc repressor studied by foreign , with the high concentra- Sauer and co-workers. The probability of tion of material leading to the formation of incorrect interactions will be greater if the . The role of molecular chap- rate of is much slower than the erones in cells is to prevent aggregation and rate of folding. In such a case, the partially the ability of normal proteins to fold rapidly synthesized polypeptide chain could undergo is an important evolutionary development to incorrect folding missing long range interac- minimize aggregation. tions. In vivo, fewer than 15% of biosynthe- Another consequence of abnormal pro- sized proteins need the assistance of molec- tein folding yields aggregates with the ap- ular chaperones to achieve their native struc- pearance of fibrils, leading to the ture (reviewed in Ref. 53). spongiform encephalopathies. These severe Other accessory molecules are able to pathologies include -associated diseases

Figure 6 - The GroES-GroEL (I) (II) (III) (IV) cycle according to Wang and ADP Weissman (52). Inf is the un- Inf N folded protein, N the folded one. GroES ATP A is the apical domain (in blue) which binds the unfolded pro- A t~6 s tein and GroES; I is the interme- I diate domain (in red) and E is the ATP ATP ADP ADP equatorial domain (in magenta) E GroEL ADP ADP ATP ATP which binds and hydrolyzes ATP (reproduced with permission).

ADP ATP

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 431 such as scrapie in sheep, mad cow disease in infectious or sporadic disorders, all involv- cattle and Creutzfeld-Jacob in . ing the conformational change of the prion Alzheimer’s disease is also characterized by protein. The normal cellular protein PrPc, the presence of amyloid fibrillar deposits in whose function remains unknown, is con- the brain . However, Alzheimer’s dis- verted into the pathological one PrPsc, through ease is a more complex process during which a structural transformation in which parts of the amyloid protein precursor is initially the a-helices of the native structure are con- cleaved by a or ß secretase, and then by g verted into ß-strands. The three-dimensional secretase generating of 40 and 42 structures of normal prion protein from ham- amino acids, Aß(40) and Aß(42), which ag- sters, mice, and recently humans have been gregate into amyloid structures. There are 16 solved by NMR studies (56,57). Prions are human amyloidogenic proteins known to transmissible particles devoid of abnormally self-assemble into fibrils 60-100 and composed of PrPsc. The data are con- Å in width and of variable length. They are sistent with a mechanism in which PrPsc characterized by a cross-ß repeat structure acts as a template for the conversion of (55). Table 1 summarizes these different nascent PrPc into further molecules of PrPsc. amyloidogenic proteins and the correspond- Many investigations have led to the deter- ing pathologies. mination of the amino acids involved in The tremendous number of results ob- the species barrier, i.e., in the impossibil- tained on the prion protein has served to ity of transmission from a species to an- firmly establish the concept of “protein only” other (Figure 7). This unusual disease introduced in 1982 by Prusiner who discov- mechanism has shown how an aberrant ered the protein, and emphasized the role of conformational change in a protein can be protein conformational change and misfold- propagated, and underlines the importance ing in human pathologies (for a review, see of protein folding for understanding the 56). Prion diseases may result from genetic, cause and propagation of a disease.

Table 1 - Amyloidogenic proteins and the corresponding diseases (according to Kelly (55)).

Clinical syndrome Precursor protein Fibril component

Alzheimer’s disease Amyloid protein precursor ß-Peptide 1-40 to 1-43 Primary systemic Immunoglobulin light chain Intact light chain or fragments Secondary systemic amyloidosis amyloid A Amyloid A (76-residue fragment) Senile systemic amyloidosis Transthyretin or fragments Familial amyloid polyneuropathy I Transthyretin Over 45 transthyretin variants Hereditary cerebral amyloid angiopathy C minus 10 residues

Hemodialysis-related amyloidosis ß2-Microglobulin ß2-Microglobulin Familial amyloid polyneuropathy III A1 Fragments of Finnish hereditary systemic amyloidosis Gelsosin 71-Amino acid fragments of gelsosin Type II diabetes Islet amyloid polypeptide (IAPP) Fragment of IAPP Medullary carcinoma of the Calcitonin Fragments of calcitonin Spongiform encephalopathies Prion Prion or fragments thereof Atrial amyloidosis Atrial natriuretic factor (ANF) ANF Hereditary non-neuropathic systemic amyloidosis Lysozyme Lysozyme or fragments thereof Injection-localized amyloidosis Insulin Hereditary renal amyloidosis Fibrinogen fragments

Braz J Med Biol Res 34(4) 2001 432 J.M. Yon

Protein folding in biotechnology: teins with functions unprecedented in na- and design ture. This research is based on our under- standing of the principles of protein folding. Two main applications concerning pro- The conception of new proteins represents a tein folding are involved in : burgeoning field of research. Genetic engi- protein engineering and the de novo design neering provides a powerful methodology to of proteins with novel functions. Recombi- redesign existing proteins. The aim of de nant proteins of pharmaceutical interest, such novo protein design is to create completely as growth hormone, insulin, and antihemo- new proteins with determined activities. The philic factor VIII, are commonly expressed first step consists of choosing a function, and and overproduced in E. coli or other cells. finding the amino acids with a favorable Since the overexpression of in foreign spatial arrangement capable of generating hosts often results in the formation of inclu- the desired function. Then it is necessary to sion bodies, further treatments including un- find a polypeptide scaffold capable of sup- folding and refolding are required. These porting the reactive groups in the appropri- proteins can also be modified by genetic ate orientation. In the following step, it is engineering to increase their stability for necessary to determine an amino acid se- storage which is an important industrial prob- quence able to fold into an adequate and lem. stable three-dimensional structure, which De novo protein design has recently presents the desired geometry of the binding emerged with the hope of constructing pro- site. Some folds such as triose phosphate

Figure 7 - Structure of prion pro- A B teins. A, NMR structure of ham- ster recombinant PrP (90-231); Helix C D227 helices are colored in pink, the disulfide bond in yellow and con- Helix B D172 served hydrophobic regions in red. B, NMR structure of recom- Helix C binant PrP (90-231) indicating a S2 van der Waals rendering of the T215 c Helix A residues thought to bind PrP Q168 S1 with a protein X. C, PrP residues Q219 governing the transmission of Helix A prions across species are indi- Helix B cated. D, Schematic diagram of PrP (29-231) showing the flex- A113 S231 ibility of the N-terminal region. E, Plausible model for the ter- C D E sc tiary structure of PrP (repro- 1.0 duced from Ref. 56 with permis- V121 0.8 sion of the author and the owner Helix B Q186 0.6 0.4 Q172 Q168 National Academy of Sciences, 0.2 USA). 0.0 I184 I203 -0.2 -0.4 T215 Helix C -0.6 -0.8 I205 N155 -1.0 M139 Q129

N143

M138 S231 Helix A W145 R148

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 433 isomerase barrel, three- and four-helix bun- been designed, resulting in the desired fold dles, and immunoglobulin fold appear fre- but showing a marginal stability, some of quently in proteins with highly divergent them displaying the characteristics of mol- sequences. They are highly designable and ten globules. Several designed helix-bundle may be easily modified without perturbing peptides that adopt multiple conformations their three-dimensional structure. in solution have been crystallized in only At this stage, several different strategies one of these conformations. These motifs may be used. One consists of using as a can serve as a starting point in protein de- scaffold a protein of known structure with sign. properties close to those desired. This is Functionalization of designed polypep- called the local conception. The other, the tides has been successfully obtained in the global conception, consists of the design of a field of , metal ion and bind- structure by analogy with one of the classical ing, and introduction of cofactors (for a re- folds in the . However, view, see 59). A 14-residue polypeptide that since in general a large number of sequences forms a bundle-like structure catalyzes the can fold into the same three-dimensional decarboxylation of oxaloacetate with a low structure, it is only necessary to arrive at one catalytic activity, a residue acting as of them. Genetic methods can be used to nucleophile. Hydrolysis and transesterifica- screen a great number of randomized se- tion reaction of paranitrophenyl esters have quences to find those that fold into a given been accomplished by designed four-helix three-dimensional structure. Combinatorial bundles formed from 42-residue polypep- computational algorithms provide a power- tides in which residues have been ful complementary approach to genetic meth- introduced. The successful design of a four- ods for exploring the sequence space. They protein that binds four heme consist of the exploration of a large number groups with high affinity has been reported. of side-chain combinations that can fit to- The structure is well defined as shown by gether to stabilize a given backbone fold and NMR spectroscopy. necessarily include a potential energy func- A number of natural proteins have been tion. Automated design of functional pro- redesigned with important changes in their teins capable of generating a sequence com- sequence. The strategies generally used are patible with the template fold and specific based on genetic selection with the help of for some purpose is being developed. Once computational methods and the construction the sequence is conceived, the recombinant of consensus sequences. of- protein can be produced from the corre- fers a powerful method to select the highest sponding gene. Another strategy uses pep- affinity binders. tide synthesis, for example in the design of It is clear that de novo protein design monomeric ß-sheets, or helical bundles. The represents a growing field of research that different methods for de novo protein design will be useful both in testing the principles of have been reviewed by de Grado et al. (58). protein folding and in offering the perspec- Most accomplishments have concerned tive to design new proteins with practical the synthetic design and structural character- applications for pharmaceuticals and diag- ization of secondary and supersecondary nostics. structures such as helices, helix bundles, ß- I will conclude this review with a citation hairpins, and ß-sheets. They have widely from Dobson, Šali and Karplus (45): “An contributed to the understanding of second- understanding of folding is important for the ary structures in proteins. A large number of analysis of many events involved in cellular parallel or antiparallel helix bundles have regulation, the design of proteins with novel

Braz J Med Biol Res 34(4) 2001 434 J.M. Yon

functions, the utilization of sequence infor- Acknowledgments mation from the various genome projects, and the development of novel therapeutic The author thanks Charles Robert for strategies for treating or preventing human carefully reading the manuscript and revis- diseases that are associated with the failure ing the English, and Charis Ghélis for help- of proteins to fold correctly.” ful discussions.

References

1. Thornton JM, Orengo CA, Todd AE & 4028-4030. plementation of engineered fragments Pearl FM (1999). Protein folds, functions 15. Yon JM (1997). Protein folding: concepts from yeast phosphoglycerate kinase. Pro- and evolution. Journal of Molecular Biol- and perspectives. Cellular and Molecular tein Engineering, 6: 313-325. ogy, 293: 333-342. Sciences, 53: 557-567. 27. Ptitsyn OB (1995). Molten globule and 2. Ghélis C & Yon JM (1982). Protein Fold- 16. Wetlaufer DB & Ristow S (1973). Acquisi- protein folding. Advances in Protein ing. Academic Press, New York. tion of the three-dimensional structure of Chemistry, 47: 83-229. 3. Dobson CM & Karplus M (1999). The fun- proteins. Annual Review of Biochemistry, 28. Ohgushi M & Wada A (1983). Molten damentals of protein folding: bringing to- 42: 135-158. globule state: a compact form of protein gether theory and experiment. Current 17. Creighton TE (1974). Experimental stud- with mobile side-chains. FEBS Letters, Opinion in , 9: 92-101. ies of protein folding and unfolding. Prog- 164: 21-24. 4. Anfinsen CB (1973). Principles that gov- ress in and , 29. Chaffotte AF, Cadieux C, Guillou Y & Gold- ern the folding of protein chains. Science, 33: 231-297. berg ME (1992). A possible folding initial 181: 223-230. 18. Weissman JA & Kim PS (1992). Reexami- intermediate: the C-terminal proteolytic 5. Karplus M & Sali A (1995). Theoretical nation of the folding of BPTI: predomi- domain of tryptophan synthase ß-chain studies of protein folding and unfolding. nance of native intermediates. Nature, folds in less than 4 milliseconds into a Current Opinion in Structural Biology, 5: 336: 42-48. condensed state with non-native-like sec- 58-73. 19. Ghélis C (1980). Transient conformational ondary structure. Biochemistry, 31: 4303- 6. Kim PS & Baldwin RL (1990). Intermedi- states in proteins followed by differential 4308. ates in protein folding reactions of small labeling. Biophysical Journal, 32: 503-514. 30. Shiraki K, Nishikawa K & Goto Y (1995). proteins. Annual Review of Biochemistry, 20. Roder H, Elöve GA & Englander SW Trifluoroethanol induced stabilization of 59: 631-660. (1988). Structural characterization of fold- the a-helical structure of ß-lactoglobulin: 7. Fersht AR (1997). Nucleation mechanisms ing intermediates in cytochrome c by H- implication for non-hierarchical protein in protein folding. Current Opinion in exchange labeling and proton NMR. Na- folding. Journal of Molecular Biology, 245: Structural Biology, 7: 3-9. ture, 335: 700-704. 180-194. 8. Jaenicke R (1987). Folding and associa- 21. Baldwin RL (1993). Pulse H/D exchange 31. Alexandrescu AT, Evans PA, Pitkeathly M, tion of proteins. Progress in Biophysics studies of folding intermediates. Current Baum J & Dobson CM (1993). Structure and Molecular Biology, 49: 117-237. Opinion in Structural Biology, 3: 84-91. and dynamics of the acid-denatured mol- 9. Ptitsyn OB & Rashin AA (1973). Stage- 22. Dobson CM (1991). Characterization of ten globule state of a-lactalbumin: a two- wise mechanism of protein folding. protein folding intermediates. Current dimensional NMR study. Biochemistry, Doklady Akademii Nauk SSSR, 213: 473- Opinion in Structural Biology, 1: 22-27. 32: 1707-1718. 475. 23. Matouschek A & Fersht AR (1991). Pro- 32. Radford SE, Dobson CM & Evans PA 10. Wetlaufer DB (1981). Folding of protein tein engineering in analysis of protein fold- (1992). The folding of hen lysozyme in- fragments. Advances in Protein Chemis- ing and stability. Methods in Enzymology, volves partially structured intermediates try, 34: 61-92. 202: 82-112. and multiple pathways. Nature, 358: 302- 11. Chothia C (1984). Principles that deter- 24. Ballery N, Desmadril M, Minard P & Yon 307. mine the structure of proteins. Annual JM (1993). Characterization of an inter- 33. Uversky VN & Ptitsyn OB (1996). Further Review of Biochemistry, 53: 537-572. mediate in the folding pathway of phos- evidence on the equilibrium “pre-molten 12. Karplus M & Weaver DL (1994). Protein phoglycerate kinase; chemical reactivity globule state”: four-state guanidinium folding dynamics: the diffusion-collision of genetically introduced cysteinyl resi- chloride unfolding of carbonic anhydrase models and experimental data. Protein dues during the folding process. Biochem- B at low . Journal of Molecu- Science, 3: 650-668. istry, 32: 708-714. lar Biology, 255: 215-228. 13. Dill KA (1985). Theory for the folding and 25. Garcia P, Desmadril M, Minard P & Yon 34. Jeng MF & Englander SW (1991). Stable stability of globular proteins. Biochemis- JM (1995). Evidence for residual struc- submolecular folding units in a non-com- try, 24: 1501-1509. tures in the unfolded form of yeast phos- pact form of cytochrome c. Journal of 14. Harrison SC & Durbin R (1985). Is there a phoglycerate kinase. Biochemistry, 34: Molecular Biology, 221: 1045-1061. single pathway for the folding of a poly- 397-404. 35. Fink AL (1995). Compact intermediate peptide chain? Proceedings of the Na- 26. Pecorari F, Minard P, Desmadril M & Yon states in protein folding. Annual Review tional Academy of Sciences, USA, 82: JM (1993). Structure and functional com- of Biophysics and ,

Braz J Med Biol Res 34(4) 2001 Protein folding and its implications in biology and medicine 435

24: 495-522. and Molecular Biology, 71: 155-241. 52. Wang JD & Weissman JS (1999). Think- 36. Plaxco KW & Dobson CM (1996). Time- 44. Dill KA, Bromberg S, Yue K, Fiebig KM, ing outside the box: new insights into the relaxed biophysical methods in the study Yee DP, Thomas PD & Chan HS (1995). mechanisms of GroEL-mediated protein of protein folding. Current Opinion in Principles of protein folding: a perspec- folding. Nature Structural Biology, 6: 597- Structural Biology, 6: 630-636. tive from simple exact models. Protein 600. 37. Eaton WA, Muñoz V, Thompson PA, Science, 4: 561-602. 53. Ellis RJ & Hartl FU (1999). Principles of Henry ER & Hofrichter J (1998). Kinetics 45. Dobson CM, Šali A & Karplus M (1998). protein folding in the cellular environment. and dynamics of loops, a-helices, ß-- Protein folding: A perspective from theory Current Opinion in Structural Biology, 9: pins, and fast-folding proteins. Accounts and experiment. Angewandte Chemie, 102-110. of Chemical Research, 31: 745-753. International Edition in English, 37: 868- 54. Yon JM (1996). Protein aggregation. In: 38. Hagen SJ, Hofrichter J, Szabo A & Eaton 893. Meyers RA (Editor), Encyclopedia of Mo- WA (1996). Diffusion-limited contact for- 46. Wolynes PG, Onuchic JN & Thirumalai D lecular Biology and Molecular Medicine. mation of an unfolded cytochrome c: esti- (1995). Navigating the folding routes. Sci- Vol. V. VCH, Weinheim, 73-93. mating the maximum rate of protein fold- ence, 267: 1619-1620. 55. Kelly JW (1996). Alternative conforma- ing. Proceedings of the National Acade- 47. Dill KA & Chan HS (1997). From Levinthal tions of amyloidogenic proteins govern my of Sciences, USA, 93: 11615-11617. paradox to pathways to funnel. Nature their behavior. Current Opinion in Struc- 39. McCammon JA (1996). A speed limit of Structural Biology, 4: 10-19. tural Biology, 6: 11-17. protein folding. Proceedings of the Na- 48. Pecorari F, Minard P, Desmadril M & Yon 56. Prusiner SB (1998). Prions (Nobel Lec- tional Academy of Sciences, USA, 93: JM (1996). Occurrence of transient multi- ture). Proceedings of the National Acade- 11426-11427. meric species during the refolding of a my of Sciences, USA, 95: 13363-13383. 40. Ballew RM, Sabelko J & Gruebele M monomeric protein. Journal of Biological 57. Zahn R, Liu A, Lühre T, Riek R, von (1996). Observation of distinct nanosec- Chemistry, 271: 5270-5276. Shroetter C, Wider G & Wüthrich K (2000). ond and microsecond protein folding 49. Silow M, Tan YJ, Fersht AR & Oliveberg NMR solution structure of the human events. Nature Structural Biology, 3: 923- M (1999). Formation of short-lived protein prion protein. Proceedings of the National 926. aggregates directly from the coil in two- Academy of Sciences, USA, 97: 145-150. 41. Shastry MCR, Sander JM & Roder H state folding. Biochemistry, 38: 13006- 58. De Grado WF, Summa CM, Pavone V, (1998). Kinetic and structural analysis of 13012. Nastri F & Lombardi A (1999). De novo submillisecond folding events in cyto- 50. Segel DJ, Eliezer D, Uversky V, Fink AL, design and structural characterization of chrome c. Accounts of Chemical Re- Hodgson KO & Doniac S (1999). Transient proteins and . Annual Re- search, 31: 717-725. dimer in the refolding kinetics of cyto- view of Biochemistry, 68: 779-819. 42. Kuwajima K (1996). The molten globule chrome c characterized by small-angle X- 59. Balzer L (1998). Functionalization of de- state of a-lactalbumin: a review. FASEB ray scattering. Biochemistry, 38: 15353- signed folded polypeptides. Current Opin- Journal, 10: 102-109. 15359. ion in Structural Biology, 8: 466-470. 43. Jaenicke R (1999). Stability and folding of 51. Baldwin RL (1994). Matching speed and domain proteins. Progress in Biophysics stability. Nature, 369: 183-184.

Braz J Med Biol Res 34(4) 2001