The folding of most newly synthesized proteins in the cell requires the interaction of a variety of protein cofactors known as molecular chaperones. These molecules recognize and bind to nascent polypeptide chains and partially folded intermediates of proteins, preventing their aggregation and misfolding. There are several families of chaperones; those most involved in protein folding are the 40-kDa heat shock protein (HSP40; DnaJ), 60-kDa heat shock protein (HSP60; GroEL), and 70-kDa heat shock protein (HSP70; DnaK) families. The availability of high-resolution structures has facilitated a more detailed understanding of the complex chaperone machinery and mechanisms, including the ATP-dependent reaction cycles of the GroEL and HSP70 chaperones. For both of these chaperones, the binding of ATP triggers a critical conformational change leading to release of the bound substrate protein. Whereas the main role of the HSP70/HSP40 chaperone system is to minimize aggregation of newly synthesized proteins, the HSP60 chaperones also facilitate the actual folding process by providing a secluded environment for individual folding molecules and may also promote the unfolding and refolding of misfolded intermediates.
The basic paradigm of molecular chaperones is that they recognize and selectively bind nonnative, but not native, proteins to form relatively stable complexes (48). In most cases, the complexes are dissociated by the binding and hydrolysis of ATP. In addition, there are “specific” molecular chaperones that typically are involved in the assembly of particular multiprotein complexes. Molecular chaperones comprise several highly conserved families of unrelated proteins; many chaperones are also heat shock (stress) proteins. The ubiquitous role of molecular chaperones continues to unfold with more discoveries each year. In the context of in vivo protein folding, chaperones prevent irreversible aggregation of nonnative conformations and keep proteins on the productive folding pathway. In addition, they may maintain newly synthesized proteins in an unfolded conformation suitable for translocation across membranes and bind to nonnative proteins during cellular stress, among other functions. It is likely that most, if not all, cellular proteins will interact with a chaperone at some stage of their lifetime.
The focus of this review is on the functional contribution of chaperones to in vivo protein folding and assembly, especially those chaperones that are promiscuous, in that they show broad specificity for binding nonnative proteins. In addition to several specialized review articles on chaperones, e.g., References 11, 49, 50, 53, 76, 77,90, 144, 164, 190, 191, 226, 228, 252, there have been two recent monographs published on the subject (60,150). This review is of necessity selective; the main goal is to furnish an up-to-date overview of the role of the major molecular chaperones involved in protein folding. In view of the vast array of literature on molecular chaperones, and the ease of access to literature citations using the Internet, this article should not be viewed as exhaustive.
Why do we need chaperones? After all, a basic tenet of in vitro protein folding has been the seminal work of Anfinsen (2), which demonstrated that formation of the native protein from the unfolded state is a spontaneous process determined by the global free energy minimum. The results indicated that the native state of small globular proteins is determined by their amino acid sequence. However, the experimental conditions necessary to successfully fold many proteins, especially larger ones, in vitro, are very constrictive, usually requiring very low protein concentration and long incubation times and are usually unphysiological (e.g., relatively low temperatures). In contrast, most cells operate at ambient or homeothermically set temperatures (e.g., 37°C) where the hydrophobic effect will be stronger and thus protein denaturation and aggregation will be bigger problems, and the time-frame available for successful folding is short. Thus there is the need for additional factors for the successful folding of many proteins in vivo. When one considers the crowded cellular environment within a cell, it becomes clear that in vitro folding experiments at low protein concentrations are poor models for what happens in the cell, where a newly synthesized protein is in an environment with little or no “free” water, very high concentrations of other proteins and metabolites, and typically membranes, cytoskeletal elements, and other cellular components. Thus the need for chaperones 1) to prevent aggregation and misfolding during the folding of newly synthesized chains,2) to prevent nonproductive interactions with other cell components, 3) to direct the assembly of larger proteins and multiprotein complexes, and 4) during exposure to stresses that cause previously folded proteins to unfold, becomes evident. In the few cases where folding has been studied both in vivo and in vitro, it appears that the folding pathways are similar (148,190).
Cells have solved the problem of misfolding and aggregation, to a considerable extent at least, through the participation of molecular chaperones in the in vivo folding process. Many investigations in the past few years have confirmed the critical role of molecular chaperones in protein folding in the cell. Although much has been learned about the function of chaperones in protein folding, and the general outline of the process is thought to be understood, there are still many important unresolved issues, and new chaperones and cochaperones are still being discovered.
The molecular chaperones involved in the folding of newly synthesized proteins recognize nonnative substrate proteins predominantly via their exposed hydrophobic residues. The major chaperone classes are 40-kDa heat shock protein (HSP40; the DnaJ family), 60-kDa heat shock protein [HSP60; including GroEL and the T-complex polypeptide 1 (TCP-1) ring complexes], 70-kDa heat shock protein (HSP70), and 90-kDa heat shock protein (HSP90). All these chaperones can prevent the aggregation of at least some unfolded proteins. For HSP60 and HSP70, their activity is modulated by the binding and hydrolysis of ATP. The HSP70 (DnaK inEscherichia coli) bind to nascent polypeptide chains on ribosomes, preventing their premature folding, misfolding, or aggregation, as well as to newly synthesized proteins in the process of translocation from the cytosol into the mitochondria and the endoplasmic reticulum (ER). The HSP70 are regulated by HSP40 (DnaJ or its homologs). The HSP60 are large oligomeric ring-shaped proteins known as chaperonins that bind partially folded intermediates, preventing their aggregation, and facilitating their folding and assembly. This family is composed of GroEL-like proteins in eubacteria, mitochondria, and chloroplasts and the TCP-1 (CCT or TRiC) family in the eukaryotic cytosol and the archaea. The HSP60 (GroEL inE. coli) are large, usually tetradecameric proteins with a central cavity in which nonnative protein structures bind. The HSP60 are found in all biological compartments except the ER. The HSP60 are regulated by a cochaperone, chaperonin 10 (cpn10) (GroES in E. coli). In addition to preventing aggregation, it has been suggested that HSP60 may permit misfolded structures to unfold and refold. The HSP90 are associated with a number of proteins and play important roles in modulating their activity, most notably the steroid receptors. A number of other proteins involved in the folding of many newly synthesized proteins are often considered to be molecular chaperones; these include protein disulfide isomerase and peptidyl prolyl isomerase, which catalyze the rearrangement of disulfide bonds and isomerization of peptide bonds around Pro residues, respectively, and are perhaps better considered to be folding catalysts rather than chaperones. As mentioned previously, there are also a number of more specific chaperones that are involved in the folding/assembly of only one, or a very limited number, of particular substrate proteins.
Chaperones are catalysts in the sense that they transiently interact with their substrate proteins but are not present in the final folded product, and also in that they increase the yield of folded protein. However, there is no good evidence that they actually enhance the spontaneous rate of folding itself, although they may appear to do this by minimizing off-pathway reactions.
A brief perusal of the literature demonstrates that our knowledge of molecular chaperones is growing at an enormous pace. To put these new discoveries in context, a few more general points are worthy of note. Although in many respects the field of molecular chaperones can now be considered a mature one, in that it has passed its first decade of life, and the broad outlines, at least, are reasonably well established, there are still many outstanding questions. Furthermore, there are many areas of considerable controversy, and many of these relate to fundamental questions. For example, we do not yet know with certainty whether all newly synthesized proteins interact with chaperones, although it is likely that they do. We certainly do not know much about all the interactions between the various chaperones themselves, as well as with newly synthesized proteins or other chaperone target proteins. As discussed in this review, there are significant controversies concerning which chaperones interact first with nascent polypeptides, and even whether all nascent polypeptides interact with chaperones. The GroEL family of chaperones has been intensively studied, especially in the context of in vitro protein folding, yet it is not clear just how important a role this family (the cpn60 chaperonins and their TCP-1 eukaryotic homologs) play in the folding of most proteins in the cell. We are only now beginning to get a picture of the apparently ubiquitous role of the HSP90 family in many critical processes in the cell, especially those involving protein-protein interactions. Recently, several new “accessory” proteins have been discovered, which apparently act as “cochaperones.” Again, their significance to protein folding and denaturation in the cell in general is unclear at the present time; they may be highly specialized or may turn out to be critical in a broad range of cellular processes involving chaperones. Although some of the chaperones clearly are important in preventing protein aggregation, there is as yet no good evidence that chaperones play a role significant in the opposite side of this equation, namely, in solubilizing protein aggregates, although it would seem likely that this may in fact be a function of some chaperones. Even at the level of the specific mechanisms of chaperone function, there are many controversial aspects, and those in the field know there have been some quite rancorous discussions over competing mechanisms. Thus the molecular chaperone field is one in which there are still many outstanding questions, including some quite fundamental ones. Consequently, chaperone scientists are likely to remain busy for a long time to come.
We begin with a brief review of the current understanding of in vitro protein folding and the potential for aggregation and misfolding.
II. IN VITRO PROTEIN FOLDING
Despite the fact that in vitro folding may not exactly mimic folding in the cell, it is minimally a good model for in vivo protein folding and has the critical advantage that a very wide variety of biophysical methods may be applied to provide a detailed knowledge of the folding pathway, kinetics, and energetics. Significant increases in our understanding of the folding process have occurred in the past few years, especially through the application of sophisticated new techniques, and these have been summarized in recent reviews (28, 29, 43, 44,51, 56, 57, 177,186, 255, 267). Both in vivo and in vitro, proteins fold remarkably rapidly, indicating that the folding pathway is directed in some way. Many studies have revealed intermediates during in vitro protein folding experiments; it is not clear, and is very difficult to establish experimentally, whether these are on- or off-pathway species. Although it is becoming apparent that in some cases these may be off-pathway species (208), some appear to be true intermediates on the productive folding pathway, consistent with rugged energy landscapes (248).
Small proteins may, under appropriate conditions, fold to the native state within a few tens of milliseconds with no detectable intermediates (177, 210). Such folding is consistent with smooth funnel energy landscape models (248), i.e., no intermediates, but could also reflect very fast folding with intermediates of sufficiently short lifetimes that they are not detected by current methods (44). However, for many systems there is substantial experimental data to support the presence of partially folded intermediates during folding. Although stopped-flow circular dichroism kinetics investigations reveal substantial secondary structure formation within a few milliseconds of the initiation of folding, most proteins take much longer to achieve the native state (seconds or longer).
The earliest stages of folding involve hydrophobic collapse to a relatively compact state and formation of metastable secondary structure. It is not clear if collapse or secondary structure occur simultaneously or if one precedes the other. It is most likely that both proceed concurrently. Certainly secondary structural units may be formed on a microsecond time scale (24). There is no conclusive data yet available on how fast the collapse occurs.
The nature of this initial collapsed state will vary depending on the conditions and the particular protein, but in general, it will consist of a very large number of substates. Further condensation will lead to one or more particularly stable intermediates; again depending on the particular protein, the intermediate(s) will have regions of unique structure, especially in terms of the compactness, amount of secondary structure, and topology. It is very likely that in most cases these intermediates will consist of a core of nativelike structure with the remainder of the protein in varying degrees of disorder. Regions of the nonordered chain are probably flickering in and out of their nativelike secondary structure conformation. At least some proteins fold via a hierarchical path in which additional structural units coalesce to an initially formed core with nativelike structure (61).
It is now clear, based on investigations of transient and equilibrium intermediates in vitro, that partially folded intermediates, as found with newly synthesized proteins in the cell, are particularly prone to aggregate, probably via specific intermolecular interactions between hydrophobic surfaces of structural subunits (59,255). The intermediates are more prone to aggregate than the unfolded state because in the latter the hydrophobic side chains are scattered relatively randomly in many small hydrophobic regions, whereas in the partially folded intermediates, there will be large patches of contiguous surface hydrophobicity that will have a much stronger propensity for aggregation. The tendency of partially folded intermediates to associate or aggregate is exacerbated as the protein concentration increases. The growing recognition of the critical importance of protein aggregation has resulted in a number of reviews (42, 59, 112,253-255).
A. Molecular Chaperones and Protein Aggregation
Both in vivo and in vitro the transition of a protein from the unfolded to folded state frequently results in the formation of partially folded intermediate states that have a very strong propensity to aggregate. In vivo this may lead to formation of inclusion bodies, especially when overexpression occurs. Members of the HSP60 and HSP70 molecular chaperone families seem to be most directly, and most generally, involved in preventing this. Current understanding of the role of HSP70 in protein folding suggests that the chaperone sequesters the unfolded or partially folded protein, thereby preventing its aggregation, but does not actively participate in the folding process; subsequent binding of ATP leads to release of the substrate protein in a nonnative conformation (144, 146,166, 167). The E. coli HSP60 chaperone GroEL and its eukaryotic homologs facilitate protein folding by binding partially folded intermediates (or partially folded domains of large multidomain proteins) in their large central cavity (see sect. vB). Folding can thus occur in a situation where aggregation is precluded (144, 229). The general outline is summarized in Figure1.
It is likely that a significant factor in the formation of in vivo aggregates, such as inclusion bodies, is a lack of available molecular chaperones, usually due to either the rapid rate of protein synthesis, the formation of long-lived folding intermediates, or a combination of both. Either situation could lead to saturation of the available chaperones. The longer a protein takes to fold spontaneously, the longer it is likely to remain associated with the HSP60 and HSP70 chaperones. Some proteins fold fast and may have partially folded intermediates that have little propensity to aggregate, thus requiring little or no chaperone assistance and little tendency to form inclusion bodies.
Several experiments have been conducted in which overexpression of various combinations of the DnaK and GroEL chaperone systems decreases the amount of aggregation (8, 72,81, 134, 229). For example, newly synthesized proteins in E. coli were shown to aggregate extensively when the rpoH mutation was present (81). This mutation in the RNA polymerase ς32-subunit, which is responsible for heat shock promotor recognition, leads to a lack of heat shock proteins. Although growth is normal at 30°C, on elevating the temperature to 42°C, the cell is unable to produce sufficient chaperones and massive aggregation is observed. Overproduction of either GroEL and GroES, or DnaK and DnaJ, significantly decreases the aggregation at 42°C. If overexpressed together, the four chaperones are able to suppress most of the aggregation. The data suggest that the GroEL/GroES and the DnaK/DnaJ chaperone systems have complementary functions in the folding and assembly of most proteins. In addition, for in vitro aggregating systems, the presence of various chaperones increases the yield of soluble or native protein (16, 18,97, 221). There have been conflicting reports as to whether the DnaK or GroEL systems, individually or together, yield the optimal amount of renaturation. It appears that in some cases all the chaperones are required for maximal suppression of aggregation, whereas in others either the DnaK system alone, or the GroEL system alone, was effective (229). It is possible that the two systems interact at different stages of folding, and thus different results may be observed depending on the particular system (16).
III. MOLECULAR CHAPERONES INVOLVED IN IN VIVO PROTEIN FOLDING
The major classes of general chaperones are the HSP40, HSP60, HSP70, HSP90, 100-kDa heat shock protein (HSP100), and the small heat shock proteins. Recent investigations have shown that not only do the major classes of chaperones often function with protein cofactors, but direct interactions between members of the HSP40, HSP70, and HSP90 families may be frequent. This section provides a brief description of the main families of molecular chaperones involved in protein folding in the cell.
A. Small Heat Shock Proteins and α-Crystallins
The small heat shock protein (HSP) and α-crystallin family consists of 12- to 43-kDa proteins that assemble into large multimeric structures and contain a conserved COOH-terminal region termed the α-crystallin domain. Many of the small HSP are produced only under stress conditions. They have been shown to function in vitro as chaperones by preventing protein aggregation in an ATP-independent manner. Several recent reviews have been published (19,47, 113, 197). The role of small HSP in protein folding in vivo is unclear, but it seems unlikely that they are major players; this probably reflects the fact that release of bound, denatured proteins from the small heat shock proteins is very slow or nonexistent. For the α-crystallins in the eye lens, a major role is to bind denatured proteins and prevent their aggregation (which would result in cataracts). The small HSP bind denatured proteins tightly, but there is little evidence at present that they normally release the bound material subsequently. It has been proposed that their major function may be in times of stress when they bind denatured proteins and prevent their aggregation. Subsequently, when the stress is removed, these complexes may provide a reservoir for the HSP70 chaperone machinery to renature the bound proteins (47). The small HSP exhibit high affinity for partially folded intermediates but show no apparent substrate specificity and are only functional in the oligomeric form (131). Little is known about the mechanism of action of the small HSP; it has been suggested that the substrate protein coats the outside of the large chaperone multimer (133) and that hydrophobic interactions are critical in substrate binding. Several models have been proposed for the quaternary structure of the small HSP, but no consensus exists. A model in which small HSP prevent protein aggregation and may facilitate substrate refolding in conjunction with other molecular chaperones has recently been proposed (133).
B. HSP40 Family
The HSP40 or DnaJ family consists of over 100 members, defined by the presence of a highly conserved J domain of ∼78 residues (DnaJ from E. coli has 376 amino acids) (131). Proteins in this family typically consist of several domains, e.g., DnaJ contains at least four conserved regions representing potential functional domains (the J domain, which is linked by a Gly/Phe-rich region to a domain of unknown function, followed by a zinc-finger region, and ending with the COOH-terminal domain, also of unknown function). Much variability is seen in the non-J domains of members of this family. The best studied examples are DnaJ from E. coli and several homologs from yeast, such as Mdj1 and Ydj1 (33, 34, 189). The best defined role thus far for the HSP40 is as a cochaperone for HSP70; however, even this function is not well understood, and there is evidence to indicate that DnaJ and other members of the HSP40 family are chaperones in their own right, binding to at least some unfolded proteins and nascent chains (94). The details of the putative role of DnaJ in protein folding are described in sections iv andv. In E. coli, DnaK, DnaJ, and GrpE cooperate synergistically in a variety of biological functions, including protein folding. The properties of DnaJ and its homologs have been reviewed previously (23, 34, 131,260).
Little is known about the structural features of DnaJ that are involved in its interaction with DnaK and unfolded proteins. Analysis of DnaJ fragments showed that both the NH2-terminal J domain and the adjacent glycine/phenylalanine-rich region are required for interactions with DnaK (117) and to stimulate the ATPase activity of DnaK (220). The G/F motif of DnaJ is also involved in modulating the substrate binding activity of DnaK (246). However, only complete DnaJ is functional with DnaK and GrpE in refolding denatured firefly luciferase. Binding experiments and cross-linking studies indicate that the zinc fingerlike domain is required for DnaJ to bind to nonnative proteins (220).
Nuclear magnetic resonance spectroscopy has been used to determine the three-dimensional structure of the J domain in DnaJ from E. coli and humans (101, 181,222). The structure is dominated by two long helices, with a hydrophobic core of highly conserved side chains. The residues believed responsible for the specificity of the interaction between DnaJ and its homologs with their corresponding HSP70 partners comprise a conserved His-Pro-Asp sequence that extends out from the core of the structure (171, 181). A peptide containing this sequence inhibited the Ydj1 stimulation of HSP70 ATPase activity but did not prevent binding of nonnative substrate proteins, indicating that DnaJ interacts with HSP70 at a site distinct from the peptide binding site (238). The adjacent Gly/Phe-rich domain in DnaJ is disordered and flexible in solution (222).
The major effect of DnaJ on the functional cycle of DnaK is the significant stimulation of the ATPase rate-limiting step, γ-phosphate cleavage, leading to stabilization of DnaK-ADP-substrate protein complexes (146). Both prokaryotic and eukaryotic forms of HSP40 interact with HSP70 in the presence of ATP to suppress protein aggregation (32). It has been proposed that HSP40 is required for the efficient binding of substrate protein to HSP70 through the stimulation of its ATPase activity (see sect.v A) (147).
It has been suggested DnaJ acts directly as a molecular chaperone in that it binds to certain denatured substrate proteins such as firefly luciferase (130, 220, 221), and even some specific folded proteins such as the ς32-heat shock transcription factor or the λP DNA replication protein, but not “normal” native proteins (41, 262). However, DnaJ binds to ς32 at a different site than that to which DnaK binds. The yeast DnaJ homolog Ydj1 was found to bind to denatured rhodanese but not unfolded reduced carboxymethylated α-lactalbumin (32, 35). As discussed in section iv, DnaJ or HSP40 has been proposed to bind to nascent polypeptides to prevent their premature folding and to target HSP70 to them (70). However, unambiguous data to support this role are scant. In yeast, DnaJ and its homologs are required not only for protein folding but also for selective ubiquitin-dependent degradation of abnormally folded proteins (132).
Significant specificity in the interactions between members of the HSP70, DnaJ, and GrpE families has been observed (40). As noted, the interaction between a given HSP70 and its interacting DnaJ is determined by the J domain (198). Recently, evidence for interactions between DnaJ homologs and HSP90 have been reported (118). It has also been suggested that DnaJ possesses an active dithiol/disulfide group and may catalyze protein disulfide formation, reduction, and isomerization (38).
C. HSP60 Family
Under the rubric of the HSP60 or chaperonin family, we consider both the GroEL and TCP-1 ring complex families. Unfortunately, different research groups have used different names for the TCP-1 ring complex, e.g., TRiC (for TCP-1 ring complex) and CCT (for chaperonin containing TCP-1). Other members include the Rubisco subunit binding protein and thermophilic factor 55 from archaea. GroEL and its homologs are found in prokaryotes, chloroplasts, and mitochondria, whereas TCP-1 and its homologs are found in the eukaryotic cytosol. Many of the HSP60 chaperones are also known as chaperonins (cpn60) and are ring-shaped oligomeric protein complexes with a large central cavity in which nonnative proteins can bind. In bacteria, at least, HSP60 require a cochaperonin, GroES (cpn10), for full function. The term chaperonin was originally coined by Ellis (48) to refer to non-heat-induced HSP60.
GroEL is probably the most studied of all molecular chaperones; in combination with its cochaperonin GroES and ATP, it facilitates protein folding, not only by preventing aggregation but also by simultaneously allowing partially folded intermediates to fold in an environment conducive to stabilizing the native state. It has been suggested that GroEL may also function by unfolding misfolded states so as to allow their productive refolding (268, 269). Members of the HSP60 family are also involved in the assembly of large multiprotein complexes such as Rubisco (27,243). The availability of a high-resolution crystallographic structure, in conjunction with mutagenesis studies, has helped in the elucidation of the details of the reaction cycle (see sect. v B). However, there are still many points of controversy, reflecting the complexity of the mechanism of this large chaperone. Recent reviews include References 53, 105, 108, 144.
The structure of the E. coli chaperonin GroEL has been solved by X-ray crystallography (9, 12,263) and electron microscopy (196) and consists of 14 identical subunits in two stacked heptameric rings, each containing a central cavity. Substantial structural information about GroEL, GroES, and related chaperonins is available from the chaperonin web home page:http://bioc09.uthscsa.edu/∼seale/Chap/struc.html. Each subunit consists of three domains: the equatorial, the intermediate, and the apical. The latter, forming the mouth of the central cavity, undergoes major conformational changes on binding of ATP and the cochaperonin GroES, which lead to substantial changes in the hydrophobic nature of the cavity (263) (Fig. 2). In particular, the relatively hydrophobic cavity lining to which the unfolded substrate protein binds before GroES binding becomes much more polar, coincident with a substantial increase in the size of the cavity. The hydrophobic polypeptide-binding site on the cavity-lining surface of the apical domain was identified with the help of various mutants (54). These same residues are also essential for binding of the cochaperonin GroES, which is required for productive polypeptide release.
The identity of amino acid residues at the nucleotide-binding sites of GroEL/GroES was determined by photoaffinity labeling with 2-azido-ATP (13). The labeled site is located at the GroEL/GroEL subunit interface, and labeling of the cochaperonin GroES occurred through a conserved proline. The 2.4-Å crystal structure of the bacterial chaperonin GroEL complexed with adenosine 5′-O-(3-thiotriphosphate) bound to each subunit shows that ATP binds in a pocket with a unique nucleotide-binding motif, whose primary sequence is highly conserved among chaperonins (9).
The 45-Å-diameter cavity in GroEL is large enough to accommodate proteins of 40–50 kDa and, as noted, is larger when capped by the GroES heptamer. Conformational changes are observed on binding nucleotides or GroES. The hydrophobic nature of the central cavity in GroEL (in the absence of the cochaperonin) presumably accounts for the lack of affinity for native proteins. GroES enhances the cooperativity of ATP binding and hydrolysis by GroEL and is necessary for the release and folding of many GroEL substrates. The crystallographic structure of GroES is known to high resolution (110). GroES has a highly mobile and accessible polypeptide loop whose mobility and accessibility are lost upon formation of the GroES/GroEL complex (128, 129).
The TCP-1 is a heteroligomeric 970-kDa complex containing several structurally related subunits of 52–65 kDa found in the eukaryotic cytosol. These are assembled into a ring complex that resembles the GroEL double ring (69, 121,187). In vitro, the TCP-1 ring complex appears to function independently of a small cochaperonin protein such as GroES. Thus far, TCP-1 complexes have been shown to be involved in the folding of very few proteins in the eukaryotic cytosol.
The major difference between TCP-1 complexes and GroEL is the heteroligomeric nature of the TCP-1 ring complex; at least eight subunit species that are encoded by unique genes are known (216). The genes are calculated to have diverged around the starting point of the eukaryotic lineage and share ∼30% amino acid identity. It has been proposed that this complexity may have evolved to cope with the folding and assembly of complex proteins in eukaryotic cells (122). Although each TCP-1 subunit is highly diverged from each other, individually they are quite homologous, suggesting that each subunit has a specific, independent function (30a,121).
D. HSP70 Family
The HSP70 are a family of molecular chaperones that are involved in protein folding and several other cellular functions and that exhibit weak ATPase activity. The HSP70 chaperones are composed of two major functional domains. The NH2-terminal, highly conserved ATPase domain binds ADP and ATP very tightly (in the presence of Mg2+ and K+) and hydrolyzes ATP, whereas the COOH-terminal domain is required for polypeptide binding. Cooperation of both domains is needed for protein folding. Several recent reviews summarize the role of HSP70 molecular chaperones in protein folding (58, 75, 83,90, 100, 144). Many of the functions of the E. coli HSP70, DnaK, require two cofactors, DnaJ (see sect. iii B) and GrpE (see sect.iii J). The majority of in vitro studies on HSP70 have been with DnaK.
The HSP70 family is very large, with most organisms having multiple members; most eukaryotes have at least a dozen or more different HSP70, found in a variety of cellular compartments. Some of the better known mammalian members are HSC70 (or HSP73), the constitutive cytosolic member; HSP70 (or HSP72), the stress-induced cytosolic form; BiP (or Grp78), the ER form; and mHSP70 (or mito-HSP70, or Grp75), the mitochondrial form. In yeast the homologs of HSC70 and BiP are known as Ssa1–4 and Kar2. In E. coli, the major form of HSP70 is DnaK. Here we will use the term HSP70 to refer to any member of the family.
The crystallographic structures of the bovine HSC70 ATPase domain, the DnaK peptide-binding domain complexed with a peptide substrate, and most recently the human HSP70 ATPase domain have been determined (63, 214, 272). The ATPase domain, which is structurally similar to actin and hexokinase, consists of four smaller domains forming two lobes with a deep cleft within which the MgATP and MgADP bind. The structure of the peptide-binding domain consists of a β-sandwich subdomain followed by α-helical segments. The peptide is bound to DnaK in an extended conformation through a channel defined by loops from the β-sandwich. An α-helical domain (the flap or latch) is believed to stabilize the complex but does not contact the peptide directly. Only five residues of the substrate protein make significant contacts with HSP70, explaining the previously observed specificity for short, hydrophobic peptides, with a strong preference for hydrophobic residues such as Leu in the central region and a strong unfavorable interaction with negatively charged residues (7, 64,80, 225). A model in which the flap over the substrate binding pocket could be in either an open conformation (to allow entry and egress of substrates) or closed conformation (to form a stable complex) has been suggested to account for the high- and low-affinity states of HSP70 (272).
Recently, the substrate specificity of DnaK has been mapped out in detail by screening an immobilized peptide library (192), following up on earlier peptide-scanning experiments (7). DnaK binding sites in protein sequences occurred statistically every 36 residues. In the folded proteins, these sites are mostly buried, and the majority are found in β-sheets. The binding motif consists of a hydrophobic core of four to five residues enriched particularly in Leu, but also in Ile, Val, Phe, and Tyr, and two flanking regions enriched in basic residues. Acidic residues are excluded from the core and disfavored in flanking regions. On the basis of these data, an algorithm was established that predicts DnaK binding sites in protein sequences with high accuracy (192).
The HSP70 preferentially bind unfolded or partially folded proteins and do not bind normal native proteins (although there are a few specific interactions with proteins in their native states, such as clathrin and ς32). It is likely that only some newly synthesized proteins require the assistance of chaperones. In coimmunoprecipitation studies with anti-HSP70 antibodies and pulse-chase labeling, it was observed that smaller proteins were disproportionately absent, suggesting that they may fold more rapidly, either with or without the assistance of HSP70 (5).
In fact, there is some evidence to support the notion that HSP70 may not interact with short-lived partially folded intermediates (218). The HSP70 inhibits the refolding of the mitochondrial isozyme of aspartate aminotransferase (AAT), but not the cytosolic homolog. This has been attributed to HSP70 binding to a long-lived early folding intermediate in the folding of mitochondrial AAT, for which the analogous cytosolic isozyme intermediate is shorter lived and rapidly transforms to a more nativelike species that does not bind to HSP70 (3). Because there will always be a kinetic competition between spontaneous folding and chaperone binding, intermediates with shorter lifetimes than that required for binding to HSP70 would not form a complex with the chaperone (Fig. 3).
The rapid binding kinetics for substrate proteins to DnaK-ATP (199) suggest that ATP-bound DnaK is the primary form initiating interaction with substrates for chaperone activity. The resulting DnaK-ATP-substrate complexes, however, are also characterized by rapid dissociation of bound substrate but can be stabilized by hydrolysis of the ATP (stimulated to a small extent by the substrate itself, or to a large extent by DnaJ; Ref. 146). The ATP-induced protein-HSP70 complex dissociation results from a conformational change induced in HSP70 by ATP binding. This conformational change decreases the affinity of HSP70 for nonnative substrate proteins and leads to their dissociation (166). Because the binding of ATP occurs in the NH2-terminal domain and peptide binding is in the COOH-terminal domain, it is clear that strong coupling between the two functional domains must exist.
Under appropriate conditions, DnaK undergoes autophosphorylation (137). It is not yet clear if this is of physiological significance. At the moment, there is no good evidence that it is. However, GrpE and synthetic peptides have been observed to inhibit the phosphorylation (the effectiveness of a given peptide correlated with its affinity for DnaK), whereas DnaJ had no effect on the reaction (169). Human HSP70 is phosphorylated in vitro in the presence of divalent ions, with calcium being the most effective. Two calcium ions were found in the human ATPase domain structure, and calcium binding may facilitate phosphorylation (214).
Various techniques have shown that HSP70 adopts at least three significantly different conformations, one in the absence of nucleotide, one with ADP bound, and one with ATP bound. Binding of nucleotides or polypeptides alters the conformations of both the nucleotide- and polypeptide-binding domains, further indication that the conformations of these two domains are highly coupled (71).
Recently, a new pair of DnaK/DnaJ-like chaperones has been discovered in E. coli (242). Sequence differences between HSC66 and HSC20 compared with other HSP70/HSP40 members suggest that these chaperones may have different peptide binding specificity and be subject to different regulatory mechanisms. In particular, the high level of constitutive expression and lack of significant response to temperature changes suggest that HSC66 and HSC20 may play an important role in the folding of certain newly synthesized proteins under normal cellular conditions.
Details of the mechanism by which HSP70 interact with newly synthesized and nonnative proteins are given in sections iv andv A.
E. HSP90 Family
Members of the HSP90 family are highly conserved, essential proteins found in all organisms from bacteria to humans. Examples include the cytosolic form in eukaryotes, HSP90, the ER form, Grp94, and the E. coli homolog HtpG. Mammalian HSP90 exist as dimers. Although there are a number of similarities between the activities of HSP90 and HSP70, the former has several identified specific interactions, for example, with cytoskeleton elements, signal transduction proteins (including steroid hormone receptors), and protein kinases (such as the mitogen-activated protein kinase system). HSP90 is frequently found in complexes with other chaperones. In vitro, HSP90 exhibits chaperone activity with diverse proteins, suggesting a general function. The properties of HSP90 have been reviewed (10, 11, 20,113, 175, 265).
Recently, the crystal structure of the NH2-terminal domain of the yeast HSP90 was solved to reveal a dimeric structure based on a highly twisted 16-stranded β-sheet. The opposing faces of the β-sheet in the dimer define a potential peptide-binding cleft, suggesting that the N domain may serve as a molecular “clamp” in the binding of ligand proteins to HSP90 (179).
There has been a long-standing controversy as to whether HSP90 binds or hydrolyzes ATP. The crystal structures of complexes between the NH2-terminal domain of the yeast HSP90 with ADP/ATP unambiguously show a specific adenine nucleotide binding site, homologous to the ATP-binding site of DNA gyrase B. This site is the same as that identified for binding the antitumor agent geldanamycin, suggesting that geldanamycin acts by blocking the binding of nucleotides to HSP90 and not the binding of incompletely folded substrate proteins as previously suggested. These results strongly suggest the direct involvement of ATP in the function of HSP90 (82, 178).
Even though HSP90 is one of the most abundant chaperones in the cell, its in vivo functions are poorly understood, and little is currently known about its role in chaperoning the folding of newly synthesized proteins, although there are hints that it does not function alone but is associated with several other cofactors. For example, HSP90 performs at least part of its function in a complex with members of the prolyl isomerase family, FKBP52 and p23 (10), and the steroid receptor complex consists of HSP90, HSP70, p48, the cyclophilin Cyp-40, and the associated proteins p23 and p60 (45). Although neither Cyp-40 nor p23 can refold unfolded substrates, in in vitro folding experiments they interact with nonnative proteins and maintain a folding-competent intermediate (67).
A temperature-sensitive mutant of HSP90 in yeast, which rapidly and completely loses activity on shift to high temperatures, has been used to examine the functions of HSP90 in vivo. The results suggested that HSP90 is not required for the de novo folding of most proteins but is required for a specific subset of proteins that have greater difficulty reaching their native conformations (153). In vitro, in the absence of nucleotide, HSP90 can maintain nonnative substrate in a “folding-competent” state that refolds upon addition of HSP70, DnaJ homolog, and nucleotide (66).
F. HSP100 Family
The heat-inducible members of the HSP100 (or Clp) family of proteins have a number of very intriguing properties and share a common function in helping organisms to survive extreme stress (78). They perform a diverse set of functions, including proteolysis. They are highly conserved, present in all organisms, and contain ATP and polypeptide binding sites. Both HSP104 and ClpA form six-membered ring complexes; the diameter of the interior of the rings is much smaller than in GroEL, making it unlikely that the HSP100 function analogously to HSP60. The basic mechanisms by which these chaperones function are not understood. There is some suggestion that HSP104 may act in concert with HSP70 and DnaJ homologs to increase the yields of renatured protein (78). It should be noted that no human analogs of HSP104 have been found.
Unlike HSP60 and HSP70, which are unable to resolubilize aggregated proteins in vitro (with the exception of RNA polymerase), HSP104 has been observed to solubilize thermally aggregated proteins both in vivo and in vitro (170). Interestingly, ClpA can substitute for the ATP-dependent chaperone function of DnaK and DnaJ in the in vitro activation of the plasmid P1 RepA replication initiator protein (257). Another unusual feature of HSP104 is its role in triggering a prionlike disorder in yeast, involving the extrachromosomal elements PSI+ and URE3 (37).
G. Calnexin and Calreticulin
Calnexin is a transmembrane molecular chaperone that resides in the ER. Calreticulin, which has sequence homology with calnexin, is a soluble ER chaperone. Both proteins are involved in the folding and assembly of nascent proteins in the ER in a calcium-dependent manner and play an important role in glycoprotein maturation and quality control in the ER (6, 93,119, 259).
Most proteins that enter the ER are cotranslationally modified by the addition of a complex carbohydrate structure that undergoes subsequent modification by selective removal of individual hexose residues (219). Both calreticulin and calnexin transiently interact with many newly synthesized proteins in the ER, with some overlap between those proteins that bind to calreticulin and those that bind to calnexin. The specificity of the interaction is determined by the nature of the oligosaccharide and requires the trimming of glucose residues from the asparagine-linked core glycans by glucosidases. Calnexin transiently interacts with newly synthesized glycoproteins, specifically recognizing a monoglucosylated intermediate (162). Its major role appears to be to monitor glycoprotein folding and prevent incompletely folded proteins from leaving the ER. As proposed by Helenius and co-workers (92), carbohydrate processing and folding occur simultaneously; calnexin recognizes and binds the monoglucosylated glycoprotein intermediate of the nascent chain. After the remaining glucose is removed, the glycoprotein is released from calnexin; if it is incompletely folded, it is reglycosylated and rebinds to calnexin. If it is folded it no longer binds to calnexin. Calreticulin is also specific for monoglucosylated glycans (172). The interactions of calreticulin and calnexin with denatured proteins are highly dependent on divalent metal ions or polyamines (261). Calnexin facilitates the folding and assembly of class I histocompatibility molecules and prevents formation of aggregates, showing that it functions as a molecular chaperone (241). Protein folding in the ER is also discussed in section iv A.
H. Protein Disulfide Isomerase
Protein disulfide isomerase (PDI) is a critical cofactor in the folding of many proteins that are found in the ER (65,77, 143, 176). Many secreted proteins have multiple disulfide bonds, presenting potential problems for correct disulfide pairing during folding. In vitro studies of the refolding of reduced proteins show that disulfide bond formation occurs rapidly and is followed much more slowly by thiol-disulfide rearrangement leading to the correct disulfide pairings. Thus catalysis of oxidative folding is necessary in vivo to rapidly generate the correct disulfide bonds in newly synthesized proteins. In the eukaryotic ER, PDI fulfills this function. Its concentration can reach close to millimolar levels. The properties of PDI have recently been reviewed (245). In addition to strong affinity for unfolded proteins and peptides, it binds many relatively hydrophobic molecules such as steroid and thyroid hormones. Hence, it is not surprising that PDI has been reported to have chaperone-like activity at high concentrations (such as inhibition of aggregation) distinct from its disulfide bond interactions (22,180, 209).
Protein disulfide isomerase has two catalytic sites situated in two domains homologous to thioredoxin, one near the NH2terminus and the other near the COOH terminus. The thioredoxin domains, by themselves, can catalyze disulfide formation, but they are unable to catalyze disulfide isomerizations (36).
I. Peptidyl Prolyl Isomerase/Trigger Factor
Under in vitro (and presumably in vivo) conditions, prolinecis-trans isomerization may become rate limiting in the folding of proteins; in many cases, the presence of peptidyl prolyl isomerase (PPI) will enhance the rate of folding. Peptidyl prolyl isomerases are ubiquitous enzymes found in virtually all organisms and subcellular compartments. Three unrelated families are known: the cyclophilins, the FK506-binding proteins (FKBP), and the parvulins (200). The former two families are also known as immunophilins. The trigger factor is a PPI with somewhat similar activity, and weak homology, to FKBP. Trigger factor is an abundant cytosolic protein originally identified by its ability to maintain the precursor of a secretory protein in a translocation-competent form (31).
Structural studies of the E. coli trigger factor reveal a modular structure, composed of three stably folded domains, of which the catalytic one is homologous to FKBP (270). Trigger factor binds partially folded intermediates tightly. Although the isolated catalytic domain of the trigger factor retains full prolyl isomerase activity toward short peptides, its activity toward protein substrates is dramatically reduced, indicating that the polypeptide binding site extends beyond the FKBP domain (201).
Trigger factor has several chaperone-like functions: it binds to nascent cytosolic and secretory polypeptide chains, and it catalyzes protein folding in vitro (98). Trigger factor interacts with GroEL in vivo and promotes its binding to at least some polypeptides; GroEL-trigger factor complexes show much greater affinity for partially folded intermediates than GroEL alone (116). On the basis of studies showing that trigger factor was cross-linked to all tested nascent chains derived from both secreted and cytosolic proteins, it appears that trigger factor may act as a general molecular chaperone in protein synthesis (99,240).
J. HSP70 Cochaperones
In addition to DnaJ and GrpE, which function as cochaperones with DnaK, and have been known for several years, other protein cofactors that interact with HSP70 have been discovered recently. These include Hip (HSC70-interacting protein), BAG-1, and auxilin. The existence of these cofactors illustrates the complexity of the HSP70 chaperone machinery in cells.
GrpE is a key component of the HSP70 chaperone system for protein folding in bacteria and mitochondria. GrpE acts as a nucleotide exchange factor to control the ATPase activity of DnaK in its reaction cycle, although the details of its mechanism remain unclear. GrpE has high affinity for monomeric native DnaK, as well as the isolated ATPase domain (185, 202). GrpE has no affinity for ATP or ADP, nor the oligomeric states of DnaK. The nucleotide exchange properties of GrpE are a consequence of the binding of GrpE to DnaK, leading to a conformational change involving the opening of the nucleotide cleft on DnaK, resulting in a low-affinity state for nucleotides. Recently, the crystal structure of GrpE bound to the ATPase domain of the molecular chaperone DnaK has been determined (89). A dimer of GrpE binds asymmetrically to a single molecule of DnaK. The structure of the nucleotide-free ATPase domain complexed with GrpE closely resembles that of the nucleotide-bound mammalian HSP70 homolog, except for an outward rotation of one of the subdomains of the protein. Two long α-helices extend away from the GrpE dimer and suggest an additional role for GrpE in peptide release from DnaK. The functional aspects of GrpE are given in section v A.
Hip is a novel tetrameric cochaperone involved in the regulation of eukaryotic HSC70, distinct from that of bacterial HSP70. It appears to play a role in forming stable HSP70 complexes with substrate proteins. One Hip oligomer binds the ATPase domains of at least two HSC70 molecules, dependent on activation of the HSC70 ATPase by HSP40. Although hydrolysis remains the rate-limiting step in the ATPase cycle, Hip stabilizes the ADP state of HSC70 that has a high affinity for substrate protein. Hip also appears to be a chaperone in its own right, in that it binds to some unfolded proteins (15,104).
BAG-1 is a recently discovered regulator of HSP70 (216,271). BAG-1 is an antiapoptotic protein and also interacts with several steroid hormone receptors that require the molecular chaperones HSP70 and HSP90 for activation. The action of BAG-1 is similar to that of GrpE in bacterial cells, in that it binds to the ATPase domain of HSP70 and, in cooperation with HSP40, stimulates the rate of ATP hydrolysis by increasing the rate of release of ADP from HSP70 (103). BAG-1 can be coimmunoprecipitated with HSP70 from cell lysates (223). BAG-1 inhibited the HSP70-mediated in vitro refolding of an unfolded protein substrate. The binding of BAG-1 to one of its known cellular targets, Bcl-2, in cell lysates was found to be dependent on ATP, consistent with the possible involvement of HSP70 in complex formation. The identification of HSP70 as a partner protein for BAG-1 may explain the diverse interactions observed between BAG-1 and several other proteins, including steroid hormone receptors and certain tyrosine kinase growth factor receptors.
Auxilin is a 100-kDa cofactor involved in the HSP70-mediated uncoating of clathrin-coated vesicles (239). Clathrin-coated vesicles transport selected integral membrane proteins from the cell surface and the trans-Golgi network to the endosomal system. Before fusing with their target, the vesicles must be stripped of their coats. Auxilin binds with high affinity to assembled clathrin lattices and, in the presence of ATP, recruits HSP70. The presence of a J domain at its COOH terminus indicates that auxilin is a member of the DnaJ family: deletion of the J domain results in the loss of cofactor activity.
A 16-kDa cytosolic protein, called p16, which copurifies with HSC70 from fish liver, has been identified as a member of the Nm23/nucleoside diphosphate kinase family (135). p16 may modulate HSC70 function by maintaining HSC70 in a monomeric state and by dissociating unfolded proteins from HSC70 either through protein-protein interactions or by supplying ATP indirectly through phosphate transfer.
Hop is a recently discovered 60-kDa protein that can form a physical link between HSP70 and HSP90, thus modulating their activities (114). Hop is involved in the refolding of denatured protein in rabbit reticulocyte lysate and stimulates the refolding by HSP70 and Ydj-1 in a purified refolding system. Optimal refolding was observed in the presence of both Hop and HSP90. Hop preferentially formed a complex with ADP-bound HSP70 and also appears to bind to the ADP-bound form of HSP90.
K. Specialized Chaperones
Some molecular chaperones may be highly specific in that they interact with only one, or a very limited number, of target proteins; examples are PapD (127), which is involved in the assembly of bacterial pili, and HSP47, which is involved in the folding and processing of procollagen in the ER. There are many large and complex protein machines in cells: in some of these cases, specific molecular chaperones are involved in their assembly (206). Some of the best-studied systems are bacteriophage capsids and bacterial pili and flagella.
The 47-kDa HSP (HSP47) is an ER-resident chaperone found in collagen-producing cells, where it interacts with procollagen. It has been proposed that it functions as a chaperone regulating procollagen chain folding and/or assembly, but the mechanism is not well understood (152). It is likely that its main function is to prevent aggregation and misfolding of newly synthesized procollagen chains until the correct COOH-terminal associations have been made to yield the collagen triple helix. When HSP47-procollagen complexes reach the cis-Golgi network, the chaperone rapidly dissociates. The major interaction site on procollagen has been shown to be the pro-alpha 1 N-propeptide (109).
Receptor-associated protein (RAP) is another example of a specialized molecular chaperone, in this case for the low-density lipoprotein receptor-related protein (LRP), a large receptor that binds multiple ligands. The major role of RAP is to facilitate correct folding of LRP and to prevent the premature interaction of ligands with LRP (161).
The production of native α/β-tubulin heterodimer depends on the action of cytosolic chaperonin and at least five protein cofactors. These reactions do not depend on ATP hydrolysis (230,231). The β-tubulin monomer release factor, p14, which catalyzes the release of β-tubulin monomers from intermediate complexes, has recently been shown to be a member of the DnaJ family (141).
There are also a number of chaperones involved in protein export, such as SecB from E. coli (88). SecB has two functions: it maintains precursors of some exported proteins in a conformation compatible with export, by preventing them from aggregating or from folding to their native state in the cytoplasm, and it delivers both nascent and completed precursors to SecA, one of the components of the export apparatus associated with the plasma membrane. Only those polypeptides that fold slowly interact significantly with SecB, even though it is able to bind a wide variety of nonnative proteins. Complexes between SecB and substrate proteins are in rapid equilibrium with the free states (236). Thus, unlike the HSP70 and HSP60, in which hydrolysis of ATP is coupled to the binding and release of substrate proteins, SecB does not form stable complexes with substrate proteins. This may reflect the fact that SecB does not mediate protein folding but is specialized for the protein export pathway.
IV. INTERACTIONS OF NASCENT CHAINS WITH CHAPERONES
Fundamental questions in protein biogenesis include at which stage the nascent protein first interacts with molecular chaperones, the identity of the chaperones, and the role of the chaperones in facilitating protein folding. Do they just prevent aggregation and misfolding, or do they play a more active role in the actual folding process? The involvement of chaperones in both co- and posttranslational folding is now clear.
Considerable controversy continues regarding which chaperones are involved in interactions with nascent polypeptide chains. The initial evidence for chaperoning came from studies on the assembly of immunoglobulin light and heavy chains and the involvement of the protein now known as BiP (an HSP70) (85,151). In a study with major implications for in vivo protein folding and assembly, Welch and co-workers (5) demonstrated that cytosolic forms of HSP70 bind cotranslationally to nascent polypeptide chains and to newly synthesized proteins in the normal (unstressed) cell in an ATP-dependent manner. The association of cytosolic HSP70 with nascent polypeptide in translating ribosomes has subsequently been confirmed in a number of organisms (154).
There have been several reports that a high-molecular-weight complex of proteins including various chaperones is associated with nascent (or unfolded) polypeptide chains during chain elongation in vitro and in vivo. Early evidence was observed in the renaturation of firefly luciferase in cell-free translation systems (70,97, 205). Chaperone-stabilized luciferase was associated with high-molecular-weight complexes overlapping the distributions of HSP70, HSP90, and the chaperonin TRiC on gel filtration columns (160). Molecular chaperones that have been implicated include HSP70 (5, 86); HSP70 and HSP40 (123, 154); HSP70, HSP40, and the TCP-1 ring complex chaperonin (70); and HSP70 and HSP90 (46, 205).
In a clever new approach, an antibody to puromycin was used to identify a population of truncated nascent polypeptides that were then probed by immunoprecipitation and chemical cross-linking with several antibodies that recognize the cytosolic chaperones HSP70, CCT (TRiC), HSP40, p48 (Hip), and HSP90, as a means of identifying chaperones bound to the nascent chains (46). The results showed that HSP70 is the predominant chaperone bound to nascent polypeptides. The interaction between HSP70 and nascent polypeptides is apparently dynamic under physiological conditions but can be stabilized by depletion of ATP or by chemical cross-linking. Interestingly, the cytosolic chaperonin CCT (TRiC) was found to bind primarily to full-length, newly synthesized actin and tubulin. Other studies have also implicated the TCP-1 ring complex in the synthesis and assembly of tubulin and actin (69, 215,264).
This investigation also demonstrated that nascent polypeptides have a strong propensity to bind to many proteins nonspecifically in cell lysates. It is likely that this nonspecific binding is responsible for the reports of additional components in contact with nascent polypeptides (46).
Several studies provide support for cotranslational interactions of molecular chaperones with nascent polypeptide chains, especially HSP70 and perhaps HSP40. The interaction of DnaJ with nascent ribosome-bound polypeptide chains as short as 55 residues was reported using firefly luciferase and chloramphenicol acetyltransferase in cross-linking experiments (97). These investigations showed that both folding and subsequent mitochondrial translocation required DnaK, DnaJ, and GrpE and led to the proposal that DnaJ protects nascent polypeptide chains from aggregation and, in cooperation with HSP70, controls their productive folding once a complete polypeptide or a polypeptide domain has been synthesized. Both HSP70 and HSP40 were shown to be associated with nascent polypeptide chains in translating ribosomes, whereas GroEL, although transiently associated with newly synthesized proteins, was absent from the ribosomes, suggesting that HSP70 and HSP40 play an early role in protein folding, whereas GroEL acts at a later stage (70,72, 173).
Investigations using fluorescent-labeled rhodanese (by the cotranslational incorporation of a coumarin derivative at the NH2 terminus of the nascent protein) demonstrated the accumulation of full-length but enzymatically inactive polypeptides on the ribosomes. These polypeptides could be activated and released by subsequent incubation with the chaperones DnaJ, DnaK, GrpE, GroEL, GroES, and ATP and release factor. Changes in fluorescence indicated that DnaJ bound to the nascent protein and appeared to be essential for folding of ribosome-bound rhodanese into the native conformation (87, 124, 126).
Further support that folding of nascent proteins can take place on the ribosome comes from studies on partially folded intermediate states of bacteriophage P22 tail-spike protein and the β-subunit of tryptophan synthase, which can be detected while still bound to ribosomes using monoclonal antibodies to the intermediates. The rapid appearance of the intermediates suggests that the nascent chains start folding during their elongation on the ribosomes. The newly synthesized incomplete chains were shown to interact with DnaK but not GroEL while still bound to the ribosome (235).
There has been considerable discussion as to whether GroEL interacts with newly synthesized proteins in a cotranslational or posttranslational manner. As noted above, several studies indicated that DnaK and DnaJ are involved at an early stage in the folding of newly synthesized protein and that GroEL acts at a later stage (72). In a recent investigation in which rhodanese was synthesized in both in bacterial and wheat germ translation extracts, only posttranslational stable complexes with GroEL were found (184). Further evidence consistent with the HSP70 chaperone machinery interacting with newly synthesized proteins before GroEL (or concurrently) comes from investigations on the synthesis of chloramphenicol acetyltransferase in a system genetically depleted of DnaK and DnaJ. Most of the chloramphenicol acetyltransferase failed to assemble into active trimers and accumulated either in a complex with GroEL or as inactive monomer. The addition of DnaK and DnaJ to the system before the start of protein synthesis led to increased formation of native chloramphenicol acetyltransferase (244). Another investigation supporting a place for GroEL in the later stages of the folding of newly synthesized proteins made use of temperature-sensitive lethal mutations in the GroEL gene. After a shift to a nonpermissive temperature, the rate of general translation in the mutant cells was reduced, but a specific group of cytoplasmic proteins failed to fold to their native states (107). The much more limited specificity demonstrated for TCP-1 chaperonins, compared with GroEL, suggests significantly different roles for these two classes of chaperonins in the biosynthesis of proteins. It is likely that the functional differences reflect underlying structural differences.
Bukau and co-workers (16) have recently suggested that during the folding of newly synthesized proteins, DnaK and GroEL do not act in sequence, but rather the two chaperone systems form a “lateral network of cooperating proteins.” There are data to support both this and the sequential models, so the question remains unresolved at present. Another source of controversy relates to the question of how many newly synthesized proteins require the assistance of HSP60 (chaperonins) in folding. Lorimer has calculated that for E. coli there is only sufficient GroEL to assist ∼5% of newly translated proteins under normal conditions (142). It is therefore likely that most newly synthesized proteins in E. coli fold without the assistance of GroEL, and this implies that most proteins fold fast enough that sequestration on DnaK to minimize the concentration of nonchaperone-bound protein suffices to prevent aggregation.
Thus, whereas the overall outline of the process of chaperone-mediated folding of newly synthesized proteins is clear, the details are as yet incompletely resolved. A nascent polypeptide will interact with HSP70 and possibly other chaperones (probably HSP40) as it emerges from the ribosome. The lifetime of HSP70 complexes with substrate proteins under in vivo conditions is not well established but is likely to be comparable to the time for folding of many newly synthesized proteins. Dissociation of the newly synthesized chain from HSP70 after release of the nascent chain from the ribosome sets up a kinetic competition between rebinding to HSP70, binding to HSP60, spontaneous folding, aggregation, or possible even proteolysis (Fig.3).
It has also been reported that there may be significant differences between folding in prokaryotes and eukaryotes (155). In a eukaryotic translation system, two-domain engineered polypeptides were observed to fold by sequential and cotranslational folding of their domains. However, in E. coli, folding of the same proteins was found to be posttranslational and to lead to intramolecular misfolding of the concurrently folding domains (155). In addition, differences between the in vitro and in vivo nature of the interactions of chaperones with actin during refolding from denaturant have been reported (68).
A. Folding in the Endoplasmic Reticulum
The ER is a key compartment in cells that are specialized for protein export and contains many chaperones that are essential for the production of functional proteins for export (250). Folding begins with the insertion of a preprotein into the lumen of the ER and can occur either posttranslationally, in which case the preprotein is completely synthesized on cytosolic ribosomes before being translocated, or cotranslationally, in which case membrane-associated ribosomes direct the nascent polypeptide chain into the ER concomitant with polypeptide elongation (14). The ER has excellent quality control mechanisms (involving chaperones) that recognize and selectively retain misfolded proteins, which are then either degraded or refolded (30, 149). The concentration of the ER HSP70, BiP, is increased by elevated levels of misfolded proteins in the ER. How the levels of misfolded molecules are monitored and how this information is used to regulate the synthesis of BiP are still poorly understood. Likewise, the mechanisms by which oxidizing potential of the ER environment is regulated, and the misfolded proteins are degraded, are also unknown.
Although some of the major chaperones involved in protein folding in the ER are well studied, e.g., BiP and PDI, it is apparent that more have yet to be characterized. For example, several calcium-dependent putative chaperones have recently been identified using affinity chromatography with denatured-protein columns and elution with ATP (159). These proteins were identified as BiP (grp78), HSP90 (grp94), calreticulin, a novel 46-kDa protein that binds azido-ATP, as well as three members of the thioredoxin superfamily: PDI, ERp72, and a previously reported 50-kDa protein (p50). Because the release of HSP90, PDI, ERp72, calreticulin, and p50 was stimulated by Ca2+, these proteins appear to function as Ca2+-dependent chaperones (159).
Evidence is accumulating that the ER HSP70 chaperone machinery is similar to that in the cytosol and bacteria, in that at least two DnaJ homologs have been found in the ER. For example, a yeast DnaJ homolog, Scj1p, is located in the lumen of the ER where it can interact with Kar2p (the HSP70 of the yeast ER) via the conserved J domain (198). Undoubtedly, chaperone-mediated folding in the lumen of the ER is complex, as revealed by the observation that the interaction of BiP with immunoglobulin light chains during folding suggests that light chains undergo both BiP-dependent and BiP-independent folding steps and that BiP must release the light chains before disulfide bond formation can occur in them (94).
B. Mitochondrial Import/Folding
Molecular chaperones play a critical role in targeting proteins to the mitochondria and the subsequent folding of the imported protein. In support of the endosymbiont theory on the origin of mitochondria, the chaperones of the mitochondria show a high degree of similarity to bacterial molecular chaperones, including a GrpE homolog (mGrpE) (193). The mitochondrial HSP70 (mHSP70) mediates protein transport across the inner membrane and protein folding in the matrix. These two reactions are carried out by two different mHSP70 complexes. The ADP-bound form of mHSP70 favors formation of a complex on the inner membrane; this “import complex” contains mHSP70, its membrane anchor Tim44, and mGrpE (106). The ATP-bound form of mHSP70 favors formation of a complex in the matrix; this “folding complex” contains mHSP70, the mitochondrial DnaJ homolog Mdj1, and mGrpE. A more detailed discussion of the role of chaperones in mitochondrial import and folding can be found in recent reviews (106, 156, 193).
V. MECHANISMS OF CHAPERONE FUNCTION
Considerable effort has been expended over the past few years to understand the mechanistic details of chaperone function. Great progress has been made, although considerable further study is necessary. The two best understood systems are those of HSP70 and GroEL. Even with these, the complexity of the systems, especially due to the interactions with cochaperones and other cofactors, has often led to apparently conflicting hypotheses. An additional source of potential discrepancies in behavior of the chaperones results from the effects of low concentrations of critical contaminants; for example, it has recently been shown that samples of HSP70 and HSP90 are often contaminated with low levels of DnaJ or HSP40, which may profoundly affect the experimental observations (204).
It is convenient to consider the mechanism of action of both HSP70 and GroEL in terms of their reaction cycles. Both of these chaperones require cochaperones for their full function, GroES in the case of GroEL, and HSP40 (or DnaJ) in the case of HSP70. Several theoretical models have been proposed to account for the effects of chaperonins on protein folding (25, 207, 232).
A. HSP70 Reaction Cycle
Several models have been proposed for the reaction cycle of HSP70 (4, 16, 74, 90,146, 147, 166, 174,195, 221). The DnaK cycle has been the most studied and is considered here. The reaction cycles for other HSP70 appear to be similar, with the exception that the cofactor GrpE will only be present in bacteria and mitochondria (273).
Although the general features of the HSP70 reaction cycle are established, there is considerable discussion about the details. Many observations indicate that the maximal functional effect of HSP70 requires the presence of DnaJ (or its homologs) (and GrpE in the case of prokaryotes and mitochondria) (73, 102,136, 203, 217, 221,249, 258, 274). The reports that DnaJ or HSP40 may bind at least some unfolded substrates are another source of confusion. It is now well established that GrpE and its homologs are nucleotide exchange factors and stimulate the ATPase cycle of DnaK or mHSP70 by increasing the rate of ADP release (39, 221) and that DnaJ and its homologs function to increase the rate of hydrolysis of HSP70-bound ATP (146, 221). Several studies have suggested that the action of DnaJ and GrpE with DnaK requires substoichiometric levels of the two cofactors (174). This is also consistent with the physiological molar ratios, in which DnaK is in large excess.
It has been shown that HSP70 discriminates between folded and unfolded proteins, normally binding only the latter (168). In fact, it is likely that HSP70 can distinguish between relatively unfolded intermediates and strongly nativelike intermediates and binds only the former. The fact that HSP70 binds to certain proteins in their native state, e.g., clathrin, is assumed to arise from the presence of accessible, unfolded loops. For a given unfolded substrate protein, there will be several potential HSP70 binding sites along the polypeptide chain, of different affinity for the chaperones (192). Both the conformational state of the substrate protein bound to HSP70 and the conformation of the substrate protein on ATP-induced release have been shown to be substantially unfolded (167).
The nature of the bound nucleotide affects the conformation of the chaperone and particularly its affinity for substrate protein. Thus complexes with ATP have low affinity for substrate and those with bound ADP have high affinity (166, 167,174). The high affinity of HSP70 for nucleotides means that these chaperones will be found as binary complexes with ATP and ADP in the cell. Although the ATP complex binds substrate proteins/peptides much more rapidly than the ADP complex (146, 199, 224), the resulting ternary complex, HSP70-ATP-substrate, also releases the substrate protein very rapidly, and thus no productive complexes with unfolded substrate result (166). In contrast, the HSP70-ADP complex, although binding substrate protein at a slower rate, forms a relatively stable ternary complex, HSP70-ADP-substrate (167). The formation of small amounts of substrate complex when HSP70-ATP and substrate protein are mixed arises from ATP hydrolysis occurring during the reaction (stimulated by the presence of the substrate protein). Thus the formation of relatively long-lived complexes between unfolded proteins and HSP70 requires the presence of the HSP70-ADP-substrate complex. This explains, at least in part, the need for DnaJ and its homologs, since DnaJ significantly stimulates the rate of hydrolysis of ATP bound by HSP70, thus leading to formation of HSP70-ADP (18, 146).
The release of the substrate protein from the HSP70-ADP-substrate complex is triggered by the binding of ATP, which induces a conformational change in the peptide-binding domain (17, 84, 137, 165,166, 227). On the basis of the crystallographic structure of the peptide-binding domain, the conformational change presumably involves the raising of the flap or latch, which is hypothesized to help maintain the substrate peptide bound (272). The cycle is completed by rebinding of another substrate protein molecule to the HSP70-ATP complex, or the hydrolysis of the ATP, leading to formation of DnaK-ADP and another conformational change. That substrate protein dissociation precedes ATP hydrolysis was demonstrated by comparison of the corresponding rates, the rate for ATP hydrolysis being significantly slower than that for substrate dissociation (166).
There appear to be several potential pathways for substrate proteins to enter the DnaK reaction cycle: via binding to DnaK-ATP, to DnaK-ATP-DnaJ, to DnaJ (which then binds to DnaK-ATP), to DnaK-ADP, and possibly to DnaK-ADP-DnaJ. The concentrations of the two ternary complexes are expected to be quite low so these are probably not major entry points. The same goes for DnaK-ADP under normal (nonstress) conditions when the levels of DnaK-ATP greatly exceed those of DnaK-ADP. The majority of the data suggest that the DnaK-ATP complex will normally be the main portal for entry to the cycle.
Most of the proposed reaction cycles of HSP70 fall into two broad classes: 1) those which propose that it is only DnaK (HSP70) with bound ATP which interacts with the unfolded substrate, and that the interaction of this ternary complex with DnaJ (HSP40) leads to rapid ATP hydrolysis (146), and 2) those which postulate that DnaJ first interacts with an unfolded (nascent) polypeptide, targeting it for binding to DnaK (74,90, 221). Although there have been several reports that DnaJ (or HSP40) binds to some unfolded proteins (70, 95, 125, 126,130, 203, 221), unambiguous evidence that DnaJ or its homologs will bind to unfolded proteins in general is currently lacking (240).
In the absence of the cochaperones, substrate protein will cycle on and off the ATP complex and accumulate only in the ADP complex. Although there are conflicting reports regarding the rate-limiting step in the intrinsic HSP70 ATPase activity, the evidence is strongly in favor of rate-limiting cleavage of the γ-phosphate of ATP, both in the absence and presence of DnaJ and substrate protein (115,146, 147, 227). Both polypeptide substrates and DnaJ homologs stimulate the ATPase activity of HSP70 inE. coli, yeast, and human cytosol (147,273). In the case of the yeast HSP70, Ssa1, the DnaJ homolog Ydj1 also accelerated release of ATP from Ssa1 (273), suggesting a possible explanation for the lack of a GrpE homolog in eukaryotic cytosol.
Thus the major pathway in the DnaK reaction cycle is likely to be the following (Fig. 4 A).1) DnaK-ATP binds the unfolded substrate protein; the resulting complex may dissociate or bind DnaJ. 2) The latter complex will undergo rapid DnaJ-stimulated hydrolysis of the ATP to yield a “stable” DnaK-ADP-substrate protein complex (due to the conformational change induced by the ATP→ADP transition), which may or may not also contain the DnaJ. 3) The ADP dissociates, catalyzed by GrpE, and is replaced by ATP. 4) This induces a conformational change to the low-affinity form which results in dissociation of the substrate protein, leaving a DnaK-ATP complex.5) The latter can then either restart the cycle by binding a substrate protein, or it can undergo ATP hydrolysis to yield a DnaK-ADP complex. This would have to dissociate the ADP and rebind an ATP before entering the productive cycle again. The rates for several of the key steps in the DnaK cycle have been reported (4, 84, 117, 163,174, 224). Because of the complexity of the system, the measured rates will be very sensitive to the concentrations of all the species involved, as well as the temperature and pH.
The alternative class of models in which DnaJ (or HSP40) acts as the initial chaperone will involve 1) the unfolded substrate protein binding to DnaJ, which will then 2) interact with DnaK-ATP, to form a transient HSP70-HSP40-U-ATP complex. This rapidly 3) undergoes hydrolysis of its ATP, resulting in the formation of a stable HSP70-HSP40-U-ADP complex. It is likely that HSP40 dissociates rapidly from such a complex. Displacement of ADP by ATP (catalyzed by GrpE in bacteria and mitochondria) 4) triggers the release of substrate protein, thus completing the reaction cycle (Fig. 4 B).
Some of the newly synthesized proteins released by the HSP70 will fold spontaneously to the native state at a sufficiently fast rate that they neither aggregate nor bind to another chaperone molecule (either HSP70 or chaperonin) before they are fully folded. However, for some proteins, further interaction with a chaperonin, such as GroEL, is apparently required for complete folding (90,96).
B. GroEL Reaction Cycle
The GroEL cycle is by far the most studied and best understood chaperonin reaction cycle, yet there are still outstanding questions. A comprehensive review has been published recently (53). For GroEL, the folding reaction is driven by cycles of binding and release of the cochaperone GroES, which alternate with binding and release of the nonnative protein substrate (62, 234). These cycles are driven by ATP binding and hydrolysis that control the conformation of the chaperonin and its affinity for nucleotides and the cochaperonin GroES. There are three major functional states: one in which the unfolded substrate is bound tightly, another in which the substrate protein is trapped in the cavity capped by GroES but in which folding can proceed because the substrate protein is not bound to the walls of the cavity, and a final state in which the substrate protein is “ejected” regardless of whether it is folded or not. Partially folded protein will rebind to the chaperonin, continuing the cycle until folding is complete (145). A distinction has been made between released nonnative conformations that are committed to folding and those that are not. It is assumed that the isolation of a partially folded intermediate in the GroES-capped, relatively polar cavity will lead to significant folding occurring, without competition from aggregation. Mutant chaperonins that are able to trap (bind but not release) substrate protein have proven very useful in such investigations (21, 52, 256).
Although both symmetric and asymmetric complexes of GroEL with GroES have been observed (211, 234,237), only the latter are believed to be physiologically functional (194, 233) (although the existence of transient symmetric complexes cannot be ruled out). In the asymmetric complexes, the GroEL ring with GroES attached is known as the cis-ring, the opposing (distal) ring is thetrans-ring. The recently determined structure of the GroEL-GroES-(ADP)7 complex revealed that the large rigid-block movements of the intermediate and apical domains in thecis-ring allowed bound GroES to stabilize a folding chamber with ADP confined to the cis-ring (26,188, 263). The conformational changes in the apical domains doubled the volume of the central cavity and resulted in burial of the hydrophobic peptide-binding residues at the interface with GroES and between the GroEL subunits. These structural changes result in the enlarged central cavity having a polar surface that favors protein folding (26, 263). The conformational changes induced in GroEL upon binding of ATP have been observed by several techniques including electron microscopy imaging, X-ray crystallography, and fluorescence labeling (26,140, 237, 263). In addition, it has been shown that binding of nucleotide to one GroEL ring is strongly favored by GroES binding to the other ring (211). The nucleotides bind to a site near the top of the equatorial domain facing the cavity (9). Under physiological concentrations of chaperonins (equimolar GroEL and GroES) and nucleotides, the predominant species is the asymmetric GroEL-GroES complex (20).
A recent model for the GroEL reaction cycle is shown in Figure5. It is generally assumed that the asymmetric GroEL-GroES complex is the state that binds substrate protein (211). This species has ADP bound in the ring that is capped by GroES. The substrate protein thus binds to the hydrophobic cavity of the distal ring (20, 53,145, 251). This trans-complex then converts to the cis-asymmetric complex in which GroES caps the cavity containing the substrate protein. This intermediate may arise either by a transient symmetric complex with two GroES lids, or by a complex in which the trans-GroES dissociates first. Binding of ATP in the cis-ring leads to the major conformational changes, especially in the apical domain leading to an increase in the size of the cavity and conversion of its surface to a more polar environment. This leads to dissociation of the substrate protein from its hydrophobic interactions with the lining of the cavity, favoring the folding reaction. Simultaneously, hydrolysis of the ATP in the trans-ring leads to the release of GroES and the opportunity for the substrate protein to exit the cavity. If the released substrate protein has not reached the native state, it may rebind for another cycle (53, 55).
Interactions between the two back-to-back rings in GroEL result in the allosteric regulation of ATP hydrolysis, binding, and release of folding substrates and the cochaperonin GroES. Allosterism in ATP hydrolysis can be described by a model in which each ring of GroEL is in equilibrium between a low-affinity (T) and high-affinity (R) state for ATP, and in which the GroEL double ring is in equilibrium between three states: TT, TR, and RR. Electron microscopy (26, 188) images of all three allosteric states, TT, TR, and RR, have been obtained for various complexes (26). Unfolded substrate proteins bind preferentially to the T state and stimulate the ATPase activity of GroEL by both a direct effect on GroEL and a shift in the equilibrium from the RR state toward the more active TR state (266). GroES promotes the T to R transition of the ring distal to GroES in the GroEL-GroES complex. Owing to the relatively low affinity of the R conformation for nonfolded proteins, this transition leads to release of protein substrates from trans-ternary complexes of GroEL, GroES, and protein substrate. The role of this release mechanism may be to assist the folding of relatively large proteins that cannot formcis-ternary complexes and/or to facilitate degradation of damaged proteins that cannot fold (111, 266). GroEL undergoes a conformational change that is partly maintained after ATP hydrolysis, as long as ADP and Pi are bound to the GroEL ring (140).
There have been several investigations of the rates for individual steps in the GroEL reaction cycle that demonstrate the importance of nucleotide binding and hydrolysis, and GroES binding, on the rate of substrate protein release (79, 91,138, 139, 157, 182,212, 234). Horwich and co-workers (21) have shown that under normal conditions the rate of hydrolysis of ATP in the ring trans to the bound GroES determines the rate of release of the GroES and hence the rate of dissociation of the substrate protein. This has been estimated to have a half-life of ∼15 s (21). Confirmation that this “timer” sets the length of time for which the folding substrate protein remains in the GroEL cavity comes from observations on the folding of mitochondrial malate dehydrogenase; trapping experiments show that its dwell time on the complex is only 20 s (182). This is in good agreement with both the rate of ATP turnover and the dwell time of GroES on the complex but is much shorter than the time taken for the substrate to commit to the folded state.
Evidence is accumulating to indicate that GroEL is able to unfold misfolded conformations (1, 158,183, 213, 232, 234,268). The protection factors for the backbone amide protons of cyclophilin A bound to GroEL have been calculated from measurements of the rates of hydrogen/deuterium exchange using NMR (158); in contrast to the native structure, similar protection factors were found throughout the sequence consistent with complete unfolding of the substrate protein. Clarke and co-workers (183) studied the GroEL-facilitated folding of mitochondrial malate dehydrogenase and showed that the chaperonin accelerated the dissociation of a misfolded intermediate formed by reversible aggregation of an early partially folded intermediate, through a repeated binding and release cycle coupled to ATP hydrolysis. It is likely that the apparent “unfoldase” activity of GroEL actually arises from its preferential affinity for the unfolded conformation (247). Thus, through mass action, misfolded intermediates will be unfolded and given a new chance to fold productively in the GroEL cavity.
The key factors in the chaperonin cycle therefore are as follows:1) nonnative substrate protein binds to thetrans-ring of GroEL, in which ADP and GroES are bound in the “opposite” GroEL ring. Binding is facilitated by the hydrophobic surfaces of the apical domain lining the cavity in the GroEL ring.2) Subsequent ATP binding to the cis-ring leads to release of the ADP and GroES, followed by 3) binding of ATP and GroES to the cis-ring results in the massive conformational change leading to the enlarged cavity. This conformational change triggers the release of the substrate protein from the surface of the apical domain and also “starts the clock.”4) Hydrolysis of the ATP in the cis-ring weakens the interaction between GroES and the cis-ring, and binding of ATP in the trans-ring leads to the complete release of the GroES and substrate protein with a half-life of ∼15 s (194).
VI. CONCLUDING REMARKS
Molecular chaperones recognize and bind to nascent polypeptide chains and partially folded intermediates of proteins, preventing their aggregation and misfolding. The folding of most newly synthesized proteins in the cell will involve interaction with one or more chaperones. The chaperones most generally implicated in protein folding are the HSP40 (DnaJ), HSP60 (GroEL), and HSP70 (DnaK) families. Recent investigations using a wide variety of techniques ranging from genetics to biophysics have begun to unravel the complexities of these chaperone machines. At the heart of the general protein folding machinery of the cell are the reaction cycles of HSP60, HSP70, and their cochaperones. For both these chaperone systems, the binding of ATP triggers a critical conformational change ultimately leading to release of the bound substrate protein. Although both chaperone systems minimize aggregation of newly synthesized proteins, the HSP60 chaperones also facilitate the actual folding process by providing a secluded environment for individual folding molecules and may also promote the unfolding and refolding of misfolded intermediates. Different cellular locations, with their different roles in the production of new proteins, have specific chaperone systems tailored to the demands of the specific location (e.g., ER, mitochondria). Because of the critical nature of chaperones in maintaining orderly functioning of the cell, substantial redundancy is found in that multiple versions of chaperones are usually present. For selected proteins, additional specific chaperones are required for their folding and assembly. Although we now have what appears to be a good picture of the general outline of in vivo chaperone-mediated protein folding, it is clear that there are still a very large number of unanswered questions, especially regarding the molecular details.
- Copyright © 1999 The American Physiological Society