Chaperonin structure and conformational changes

Helen R. Saibil

Department of Crystallography,
Birkbeck College London, Malet
St, London WC1E 7HX, UK

Published in 'The Chaperonins', RJ Ellis, Ed., Academic Press and reproduced by permission.

I. Introduction: Methods for structural studies of chaperonins
A. Electron microscopy: 1. Negative staining; 2. Cryo EM and 3D reconstruction
B. X-ray crystallography

II. Arrangement of subunit domains in the cpn60 oligomer
A. Low resolution studies of chaperonins by EM: 1. Comparison of GroE and TCP-1 chaperonins; 2. 3D reconstructions from cryo EM images

III. The crystal structures of GroEL and GroES
A. Subunit structure and oligomeric contacts in GroEL
B. Structure of GroES
C. Functional sites in GroEL defined by mutagenesis: 1. Nucleotide binding pocket; 2. Polypeptide substrate binding surface; 3. GroES binding surface

IV. Conformational changes in cpn60 and its complexes studied by cryo EM
A. Effects of nucleotides
B. GroEL-GroES complexes: hinge rotations in GroEL
C. Ligand complexes: binary and ternary

V. Coordination of chaperonin function: structural aspects of the molecular mechanism
A. Hypotheses about the molecular basis of chaperonin action
B. Mutational probes of allosteric interactions

References

I. Introduction: Methods for structural studies of chaperonins

The large size and 7-fold symmetry of the GroE chaperonins presents special challenges for structural work. For X-ray crystallography, it is a major achievement to crystallize and solve a structure of this size; this has recently been achieved by Braig et al. (1994). On the other hand, low resolution studies by electron microscopy are facilitated by the high and odd-numbered symmetry, as well as the large size and distinctive features of the cpn60 structure. The two approaches can provide complementary information, since there are large conformational changes and transient complexes which can be captured by cryo electron microscopy (EM) (Saibil, 1994), which provides low resolution density envelopes into which atomic structure data can be fitted and manipulated. This combination of X-ray and cryo EM data has been very effective in other areas, notably for actomyosin and virus-receptor complexes, and it has great potential for cpn60. Fig. 1 shows a diagram of the cpn60 14-mer, with the subunits shown in outline, based on the crystal structure.

Fig. 1. The cpn60 14-mer. Subunit outlines based on the X-ray crystal structure of GroEL. Diagram adapted from Braig et al. (1994).

A. Electron microscopy

The EM methods for sample preparation and imaging will be briefly explained in the following sections. Very small amounts of protein are needed for EM study. Unlike the situation in optical microscopy, the depth of focus in transmission EM is much greater than the specimen thickness, so that the observed image is a projection through the whole structure. In order to get 3D information, images must be obtained of the specimen in different orientations so that the different views can be combined into a 3D density map. This is done by recording images with different tilt angles, and/or by taking advantage of molecular symmetry to generate other views. Transmission EM images are 2D projections of 3D structures, and 3D reconstruction can be done by computed tomography, if enough views at different angles are available.

1. Negative staining

In routine transmission EM, samples are imaged under high vacuum and therefore must be dehydrated. Protein structures can be imaged by allowing them to dry in a thin film of heavy metal stain, effectively encasing them in an electron dense medium. The image is formed by the stain, which reveals the surface shape of the protein by its exclusion. Although crude and simple to apply, negative staining often gives quite a good idea of the shape and symmetry of a protein structure to about 20 Å resolution. Its main drawbacks are variable distortion due to dehydration and lack of internal structure. Drying usually causes flattening of the 3D structure which poses particular problems for 3D reconstruction from negative stain images.

2. Cryo EM

A technique increasingly used by structural biologists is cryo EM. The aqueous protein sample is applied to an EM grid, but instead of staining and dehydration, a thin layer of suspension is very rapidly cooled by plunging into a cold liquid (typically ethane, cooled to liquid nitrogen temperature). This traps the protein in the native, hydrated state, embedded in vitrified water. Rapid cooling prevents the formation of ice crystals, which would otherwise damage the protein by removing water from its surface, and the protein is preserved in the native state as long as the grid is kept below about -150/C (Dubochet et al., 1988). There are very significant advantages to this approach, mainly that the native protein electron density is imaged, and is directly comparable to that determined by X-ray crystallography, even though the resolution is normally much lower, of the order of 20-30 Å in single molecule work (as opposed to 2D crystal data which can go to atomic resolution). A disadvantage of cryo EM of unstained, frozen-hydrated proteins is the extremely low image contrast. This, combined with the need to limit the electron dose because of radiation damage, means that the signal to noise ratio is very low. In order to extract reliable information, it is necessary to average many copies of the protein image, by computer processing. This is essential in any case for 3D reconstruction, which requires the combination of many different views.

B. X-ray crystallography

Crystallization and structure determination for a large protein with 7-fold symmetry is an enormous task. Highly purified protein must be available in large quantities to find good crystallization conditions and grow suitable crystals for data collection. Wild-type GroEL is difficult to crystallize, but Braig et al. (1994) found that a double mutant, Arg13®Gly, Ala126®Val formed good crystals. The C2221 crystal form contained one half molecule per asymmetric unit; a molecular 2-fold corresponded to a lattice 2-fold, reducing the problem of finding the molecule to that of positioning the 7-fold axis, known to be perpendicular to the 2-fold. Three heavy atoms per subunit were located in the heptameric ring to yield a 6 Å map by single isomorphous replacement; then 7-fold averaging was exploited to extend the phases from 6 to 2.8 Å. Because 7-fold symmetry does not fit into a crystal lattice, the 7 subunits make different packing interactions with neighbouring rings, and the more deformable parts of the structure deviate from exact 7-fold symmetry. Such regions are not resolved by methods that rely on 7-fold averaging. Most of the structure, with the notable exception of the termini and a region near the middle of the sequence, was clearly resolved by this procedure. Considerable work lies ahead, in the refinement of the structure and in the analysis of GroEL in complex with nucleotides, GroES and the various possible substrate combinations. All of these structures should illuminate the molecular mechanism of chaperonin-assisted protein folding.

II. Arrangement of subunit domains in the cpn60 oligomer

A. Low resolution studies of chaperonins by EM

The distinctive structure of cpn60 was observed in the 1970s by negative stain EM. Two views, a round one with 7-fold symmetry and a rectangular or square one with 4 stripes of density, were reported by Hendrix and by Hohn et al. in 1979, when GroEL was described as a factor required for bacteriophage S assembly. Choroplast cpn60 was shown to have similar structure (Pushkin et al., 1982), and a 3D model was proposed, but the relationship between side and end views was not understood until the work of Hutchinson et al. (1989). By recording highly tilted views, these authors were able to show that the 7-fold axis seen in the (round) end view was perpendicular to the stripes in the side view. This established the basic layout of cpn60: in order to explain the 4 stripes, the 14-mer had to be a cylinder formed of 2 rings of 7 subunits, in which each subunit was divided into 2 major domains, so that each ring contributed 2 stripes, or layers of density. The chloroplast protein, unlike the other GroE cpn60s, is formed of two distinct subunit types with an I7J7 composition (see Gatenby chapter). Figs. 2a & b show averaged cryo EM views of GroEL. The negative stain appearance of these views is very similar. Despite the variable flattening of the structure in negative stain, a 3D reconstruction correctly showed the 2 major subunit domains to be connected by a thin bridge of density on the outside of the cylindrical oligomer, forming a cage-like structure with side openings (Saibil et al., 1993).

1. Comparison of GroE and TCP-1 chaperonins

Fig 2 shows averaged, cryo EM end and side views of GroEL (Figs 2a and 2b, respectively). Although they are both double ring complexes with 4 layers of density perpendicular to the symmetry axis, there are 7 subunits forming the rings in the eubacterial chaperonin and 8 in the eukaryotic cytosolic protein. In archaebacterial members of the TCP1 subfamily of chaperonins, there are 8 or 9 subunits in the ring (Phipps et al., 1991; Marco et al., 1994b). GroEL has 14 identical subunits, whereas there are up to 9 different gene products in CCT preparations (Kubota et al., 1994). These different gene products are all related in sequence, and they are likely to have very similar structures. An apparent structural difference between GroEL and CCT is the difference in relative masses of the 2 subunit domains, with the apical domains appearing relatively smaller than the equatorial domains in the latter case. Also, the CCT oligomer appears to have a much larger central channel. These structural features were shown by negative stain EM of the archaebacterial protein (Phipps et al., 1991).

Fig. 2. Cryo-EM averaged end and side views of GroEL (a,b), from Chen et al. (1994).

2. 3D reconstructions from cryo EM images

Cryo EM has proved more reliable than negative stain in defining the dimensions and particularly the domain reorientations in the various cpn60 complexes. Fig. 3 shows 3D reconstructions of GroEL, GroEL-ATP and GroEL-GroES-ATP from cryo EM. The GroEL cylinder (a) is progressively elongated, with a small but distinct increase in length with ATP binding (b), and a very substantial elongation, particularly visible in the upper ring of GroEL in the complex with GroES (c). At the same time, the end cavities are widened. GroES is seen as a flat, disk-like object above the GroEL. With this surface contour display, the contacts between GroEL and GroES are not seen.

Fig. 3. Surface-rendered views of 3D reconstructions of GroEL, GroEL-ATP and GroEL-GroES-ATP from cryo-EM. The GroES ring is seen as a disk above the GroEL. Reprinted with permission from Nature (Chen et al., 1994). Copyright 1994 Macmillan Magazines Limited.

III. The crystal structures of GroEL and GroES

A. Subunit structure and oligomeric contacts in GroEL (Braig et al., 1994)

Each subunit is divided into 3 distinct domains (Fig. 4): the N terminus begins in the equatorial domain, and continues up through the intermediate domain. The central part of the sequence forms the apical domain, after which the chain returns down through the intermediate domain and the crystallographically visible density terminates near the N terminus in the equatorial domain. Both N and C termini (a total of 30 residues in total are not seen at the termini, presumably due to disorder) face the central channel, which is continuous through the 14-mer in the crystal structure. However, the missing 30 residues per subunit appear to form a central constriction to the channel as seen in cryo EM images (Chen et al., 1994). Mutation of Lys4®Glu (sequence begining at Met1) completely blocks oligomer assembly (Horovitz et al., 1993), suggesting an important structural role for this apparently disordered region. The equatorial domain, containing N and C terminal parts of the chain, forms the backbone of the oligomeric structure. It provides most of the contacts holding the rings together (including a parallel J-strand interaction between adjacent subunits), and also the only contacts between the 2 rings in the 14-mer, an important route of allosteric communication. This domain is a well-defined bundle of 7 I-helices.

At the top of the equatorial domain, there is a well-defined junction with the intermediate domain, a small, antiparallel domain connecting the top and bottom large domains. It comprises a pair of crossed helices and a small J-strand region that contacts the neighbouring apical domain. This is followed by another junction leading to the large apical domain, which contains I and J structures and has poorly defined density in regions lining the channel, the site of residues involved in substrate and GroES binding, as identified by mutagenesis (see below). A conspicuous feature is that much of the apical domain surface is not in close contact with neighbouring domains, and that there is little barrier to its movement, both locally in some parts and in overall orientation of the domain. It is this part of the structure that deviates most from 7-fold symmetry in the crystal.

Very interesting potential hinge sites are found at the upper and lower domain junctions. These provide the possibility of large hinge movements allowing rigid body rotations (Gerstein et al., 1994) of the apical and equatorial domains. In these regions of exposed antiparallel chain, 3 of the 4 sequences contain conserved glycine residues. Fig. 6a shows a schematic diagram representing the subunit domain arrangement in the oligomer, in which the potential hinge sites are marked with black dots.

Fig. 4. Atomic resolution structure of the GroEL subunit (Braig et al., 1994). (a) alpha-carbon and (b) full structure, colour coded in rainbow colours from N (red) to C (blue). In this view, the equatorial domain is the large mass containing both termini at the bottom of the subunit, connected by the antiparallel intermediate domain (cyan and yellow) to the large apical domain (green and cyan) at the top. Images kindly provided by Dr. Kerstin Braig.

B. Structure of GroES

GroES was isolated and characterized as a ring structure with approximately seven 10-kDa subunits that formed a complex with GroEL, by Chandrasekhar et al. (1986) in their study of bacteriophage S assembly. Crystallographic analysis of GroES has recently been completed (J.F. Hunt, A.J. Weaver, S. Landry, L. Gierasch & J. Deisenhofer, unpublished; Fig 5). Each subunit consists of a J barrel region that forms most of the contacts around the ring, and a J-hairpin pointing slightly upwards and towards the center of the ring. The J-hairpin region is loosely packed, with little intersubunit contact, and forms the roof of the dome-like structure of the oligomer. Earlier NMR work showed that a mobile domain in GroES became ordered upon binding to GroEL (Landry et al., 1993). This region (residues 17-32) is not seen on the X-ray structure of GroES, but is expected to form at least part of the binding contact. The visible ends of the mobile loop point downwards and radially outwards from the bottom of the J-barrel domain (dashed arrows, Fig 5b). EM reconstruction of the GroEL-GroES complex shows that the dome caps the opened apical domains of GroEL (Figs. 3 & 6). In the complex, the mobile loop is expected to be in contact with GroEL, because its accessibility to trypsin is reduced in the complex and a synthetic peptide with the loop sequence binds to GroEL (Landry et al., 1993). Furthermore, all the GroES mutations originally isolated by their inability to support bacteriophage S growth map to the mobile loop region of the sequence (Georgopoulos et al., 1973).

Fig. 5 Ribbon diagrams of the GroES structure viewed (a) from above and (b) from the side, showing the ß-barrel structure of the subunit with a ß-hairpin forming the roof of the dome-like heptamer. The N and C termini are labeled and point radially outward. In (a), the positions of two glutamic acid residues in the ß-hairpin are indicated as negative charges. The broken ends of the disordered mobile loop, which is not seen in this map, are indicated by dashed arrows in (b). Images kindly provided by Dr John Hunt.

C. Functional sites in GroEL defined by mutagenesis

An extensive mutational study (Fenton et al., 1994) has enabled the mapping of several important functional sites onto the atomic structure, including nucleotide, substrate and GroES binding sites. Results from other mutant studies (e.g. Baneyx & Gatenby, 1992; Yifrach & Horovitz, 1994) can now also be interpreted in terms of the 3D structure.

Fig. 6. Schematic diagram of the subunit arrangement in a hypothetical slice through the oligomer, showing the major functional sites. (a) GroEL, based on the crystal structure with bound substrate (shaded light to dark) as imaged by cryo-EM. A, apical domain; I, intermediate domain; E, equatorial domain. The apical domains form a ring of hand-like structures with the substrate binding sites on the fingers protruding into the central channel. Sites of potential hinge rotations are indicated by black dots. The notch in the equatorial domain represents the ATP-binding cleft. The interaction across the equatorial plane is shown as pairs of wavy surfaces. (b) The folding complex GroEL-MDH-GroES-ATP, rotating the subunit domains and adding the GroES (dark gray) and substrate (shaded) densities according to cryo EM observations.

1. Nucleotide binding pocket

The helix bundle in the equatorial domain contains the ATP binding pocket which is bordered by a highly conserved sequence motif containing Asp87, GDGTT, and lined by other stretches of highly conserved residues (Fenton et al., 1994). The structure with nucleotide bound has been solved by Boisvert, D.C., Wang, J., Otwinowski, Z., Horwich, A.L. & Sigler, P.B. (unpublished), and confirms that nucleotide occupies this pocket. Mutations in Asp87 completely abolish ATPase activity, although ATP analogue binding is relatively unaffected. The ATP-binding pocket is adjacent to the lower hinge region, and mutations in the intermediate domain just beyond this hinge region, also abolish ATPase activity (residues 150, 151, 152, 405 and 406), as does mutation of residue 383, near the upper hinge region. These strategic locations of key residues around the hinge regions strongly suggest that hinge movements are involved in the hydrolysis mechanism.

2. Binding surface for polypeptide substrates

The mutagenesis study defines a set of hydrophobic residues on the apical domain, lining the channel, that provide the binding surface for substrates, a region that is flexible in the GroEL structure. The binding surface is shown as a wavy profile on the apical domains in the diagram (Fig. 6a). Single amino acid substitutions, in some cases severe changes, of these residues (including residues 199, 203, 204, 234, 237, 259, 263, and 264) abolish binding of the substrates ornithine transcarbamylase and dihydrofolate reductase. By cryo EM, density of the substrate malate dehydrogenase (MDH) is found in exactly the same region, held between the ends of the claw-like domains and protruding outwards (Chen et al., 1994). The ring of 7 binding sites is easily accessible from outside, being very near the surface of the cylinder, but the inward-facing orientation may protect GroEL from self-aggregation. Mutation of residue 152, near the lower hinge region, also has a strong effect on substrate binding. The ability of monomeric cpn60, missing 78 N-terminal residues, to bind substrates and partially promote their folding is consistent with the apical location of substrate binding sites (Taguchi et al., 1994).

3. GroES Binding Surface

Many of the same mutations to the apical channel-lining surface also interfere with GroES binding, suggesting that the binding sites for GroES and substrate overlap. In addition, the GroES binding surface appears to extend further over the top surface of the apical domain. Looking at the GroEL crystal structure, it would appear impossible to fit the GroES heptamer (diameter ~70 Å, Hunt et al., unpublished) into the cavity to contact most of these binding sites. However, an explanation for this is found in the low resolution structure of the GroEL-GroES complex from cryo EM (Fig 6b and 7a&c). A large rotation of at least the upper hinge region in the GroEL subunits is clearly evident, to give a 50-60/ reorientation of the apical domains, placing the substrate binding surfaces adjacent to the bases of the J-barrel domains of the GroES subunits (Fig 5b).

Fig. 7. (a) Side view of the GroEL-GroES complex averaged from 200 cryo EM images (Chen et al., 1994). (b) Side view average of "football" complexes from cryo-EM (S. Chen, A. Roseman, and H. Saibil, unpublished observations). (c) Section through a 3D reconstruction of the complex in (a) showing the large reorientation of the apical domains in contact with GroES, forming the contact with the same region as the substrate binding site. The GroEL cylinder is greatly elongated. (d) Section of a 3D reconstruction of the folding complex GroEL-MDH-GroES-ATP trapped by vitrification after 12 s of folding from cryo-EM (Chen et al., 1994). The MDH substrate density is found in the opposite ring to GroES.

IV Conformational changes in cpn60 and its complexes studied by cryo EM

A. Effects of nucleotides

Binding and hydrolysis of ATP drive the key functional cycle of cpn60: in the ADP-bound form, or without nucleotide, cpn60 has a high affinity for unfolded substrate, and in the ATP-bound form, it has lower affinity for substrate (Staniforth et al., 1994). Alternation between these states has been proposed as the basis of the assisted folding mechanism (Jackson et al., 1993). These two states have different conformations in cryo EM images, with a 5-10/ opening of the apical domains in the presence of ATP. The superposed outlines of sections through the 3D reconstructions of GroEL and GroEL-ATP are shown in Fig 8. The apical domains are seen to open out slightly, elongating the cylinder and widening the binding cavity. A structural change induced by ATP binding was originally observed by negative stain EM, and it appeared to cause an inward rotation of the subunits (Saibil et al., 1993). The discrepancy can be partly explained by the collapse of the oligomer in negative stain, which is most evident in molecules that are noticeably flattened (larger in diameter relative to equivalent cryo EM views). This behaviour suggests that ATP binding makes the hinge regions more flexible. Non-hydrolysable analogues of ATP appear to cause the same type of cavity opening as ATP (Hunter, A.S., Roseman, A.M., Wood, S.P. & Saibil, H.R., unpublished observations), but ADP causes a more subtle change (Langer et al., 1992). Recent work shows that ATP induces asymmetry of the GroEL 14-mer (Bochkareva and Girshovich, 1994).

Fig. 8. Superposed outlines of sections through GroEL and GroEL-ATP (shaded) from cryo-EM (Chen et al., 1994), showing the slight but distinct outward rotation of the apical domains in the ATP form. These displacements have the effect of lengthening the cylinder and widening the binding cavity.

B. Cpn60-cpn10 complexes: hinge rotations in cpn60

In the presence of nucleotide, a complex is formed between cpn60 and cpn10 (Chandrasekhar et al., 1986). Cpn10 binds tightly to one ring of cpn60 after ATP hydrolysis, trapping ADP in the 7 sites on that ring (Todd et al, 1994). ATP binding by cpn60 is cooperative, and this positive cooperativity is increased by cpn10, suggesting that all 7 sites in the opposite ring of cpn60 act together. With each round of ATP hydrolyis, the cpn10 is released and rebinds. The biggest structural change in cpn60 is seen in its complex with cpn10. From negative stain EM studies of the complex it was immediately apparent that cpn10 binding was asymmetric and that it appeared to cause a significant expansion in cpn60 (Saibil et al., 1991; Ishii et al., 1992; Langer et al., 1992). The asymmetry was surprising, since the two rings of cpn60 are identical. It implies strong negative cooperativity of cpn10 binding, and accords well with the observation that the rate of chaperonin-assisted refolding of rubisco in vitro as a function of cpn10/cpn60 molar ratio saturates as the cpn10 7-mer concentration approaches that of the cpn60 14-mer (Azem et al., 1994), although not with the interpretation of those authors. However, it was also shown that double-ended cpn10 binding could also take place (Harris et al., 1994; Azem et al., 1994; Schmidt et al., 1994; Todd et al., 1994), and that the "football" form was favoured under certain ionic conditions (high pH and Mg2+ seem to be important). The second cpn10 binds more weakly (Todd et al., 1994).

Averaged side views and sections through the GroEL-GroES complex shows large displacements of the apical domains, consistent with a large rotation of the apical domains in contact with GroES about the upper hinge region (Fig. 7a,c). This dramatic rotation occurs at both ends of GroEL in the case of the football structures (Fig 7b).

C. Ligand complexes: binary and ternary

Substrates bound to GroEL are thought to be in a partially folded form, with native secondary structure, but with expanded and incomplete tertiary structure (the 'molten globule') (Goloubinoff et al., 1989; Martin et al., 1991; van der Vies et al., 1992). The position of substrate binding in the central channel in end projection was first suggested by negative stain EM of GroEL complexes with rhodanese and alcohol oxidase (Langer et al., 1992), and by scanning transmission EM of complexes with dihydrofolate reductase labelled with colloidal gold particles (Braig et al., 1993). More recently, a central location for actin binding in TCP1 has been demonstrated by negative stain EM (Marco et al., 1994a). Ishii et al. (1994) used antibody labelling of the substrate 3-isopropylmalate dehydrogenase to show that it could be found at the opposite end of cpn60 from the cpn10 in the asymmetric complex. However, the substrate density is not seen in negative stain side views of GroEL complexes.

By cryo EM, the substrate malate dehydrogenase was directly observed as additional density at one end of GroEL, held between the apical domains, in the mouth of the central channel (Chen et al., 1994; diagrammed in Fig 6a). One-sided binding was observed even in the presence of greater than a 2-fold molar excess of MDH, implying a negative cooperativity of substrate binding. Because the vitrification procedure traps the sample rapidly from the native state, it is possible to capture transient complexes whose lifetime is longer than the mixing, blotting and freezing time (about 10 sec with normal procedures). The transient complex GroEL-GroES-ATP-MDH was captured and imaged within 12 sec of starting the folding reaction. It showed extra density, as compared to the 'empty' GroEL-GroES-ATP complex, in the apical cavity remote from the bound GroES (Chen et al., 1994; Fig 6b, 7c & d). Samples vitrified after longer periods of folding show a progressive emptying of the binding site (Roseman, A.M., Chen, S., Burston, S.G., Clarke, A.R. & Saibil, H.R., unpublished). The size and shape of the extra density are more suggestive of the native MDH subunit structure than of a highly expanded state.

V. Molecular basis of chaperonin function

A. Hypotheses about the mechanism of chaperonin action

There is general agreement that cpn60 binds folding intermediates with exposed hydrophobic surfaces and suppresses aggregation, and that different substrates have varying requirements for release from cpn60-cpn10 complexes into a form committed to folding to the native state. (Gatenby & Viitanen, 1994; Hartl & Martin, 1995). It is also generally held that cycles of ATP binding and hydrolysis are central to the binding and release mechanisms (Jackson et al., 1993; Todd et al., 1994). The main arguments about the precise function of cpn60 and the role of cpn10 center around the conformational state of the bound polypeptide substrate and the sequence of events during folding. In particular, does the substrate fold while bound to cpn60, is it unfolded as a result of the binding interaction, and does folding take place in free solution? The requirement for cpn10 depends on the substrate and the folding conditions; ATP alone is often sufficient to induce substrate release but results in aggregation under non-permissive folding conditions if cpn10 is absent (Schmidt et al., 1994). Studies with the E. coli proteins show that cpn10 is bound and released with each round of ATP hydrolysis (Martin et al, 1993; Todd et al, 1994), and there is strong evidence that the substrate also goes through cycles of binding and release, since it can be trapped in a non-native state by mutant cpn60, added after the start of the reaction, that binds it irreversibly (Weissman et al, 1994). But the temporal and spatial relationships between cpn10 and substrate binding are still unclear. For example does the substrate get enclosed in the cavity formed by cpn10 binding, or is does it only bind on the opposite ring of cpn60? The volume of the cavity is only large enough to accommodate a non-compact protein subunit of up to ~40 kD. The requirements for folding vary greatly between different substrates, which makes it difficult to arrive at a single interpretation of the results. Broadly speaking, one school of thought holds that chaperonins are essential for preventing aggregation and that substrates fold in the protected environment (Martin et al., 1993), while another proposes that binding is only transient and that the interaction with chaperonins serves to unfold misfolded structures (Jackson et al., 1993; Zahn et al., 1994) which are then released to a fresh start to fold, in solution, to a productive form (Todd et al., 1994; Weissman et al., 1994). In the latter model, they would continue to be released and rebound until a form committed to folding is reached. The time course of commitment to folding has been measured for rhodanese and glutamine synthetase by removing the GroEL by immunoprecipitation at different times during refolding (Fisher and Yuan, 1994). These experiments show that the GroE system is needed for subunit folding, but is not required during oligomeric assembly of folded monomers to produce the active, multimeric protein (Fisher, 1994). For rhodanese, the GroE system is needed throughout the folding period, until full regain of enzyme activity.

B. Mutational probes of allosteric interactions

A very striking property of cpn60 is the negative cooperativity between the rings. ATP, ADP, cpn10 and protein substrates all bind asymmetrically: once a ligand is bound, the two rings are in some way different.

In the crystal structure, Arg 197 in the apical domain is shown to be involved in a contact with Glu386 in the neighbouring intermediate domain. Mutation of R197®Ala reduces both the positive cooperativity of the ATPase within rings and its negative cooperativity between rings, leading to the proposal, now backed up by the crystal structure, that it is involved in both intra- and inter-subunit allosteric interactions (Yifrach and Horovitz, 1994). These residues are near the upper (apical) hinge, whereas the ATP-binding site is next to the lower hinge.

Other mutations, around the hinge regions, and at a site at the equatorial plane that forms a contact between the two rings, influence GroES binding. Glu461, which forms a charge pair contact between the two rings with Arg452, is particularly remote from the apical domain, and likely to be important in negative cooperativity.

A set of particularly interesting mutations point to allosteric sites affecting GroES binding and ATP hydrolysis. In addition to the sites that are likely to be in direct contact with these ligands, mutations of residues throughout the intermediate domain interfere with GroES binding and ATPase activity. The intermediate domain with its two potential hinges is thus heavily implicated in the allosteric interactions at all stages of the reaction cycle. Not surprisingly, mutations that affect GroES binding and ATP hydrolyis also affect polypeptide release and folding. Another line of communication, via a direct contact between apical and equatorial domains, is suggested by the effect of mutating Gly45, a residue in the equatorial domain that comes quite close to the apical domain. Mutating Gly45®Glu causes defective release without reducing ATPase or GroES binding. Intriguingly, other mutations with these effects are also found near the top of the apical domain (Glu238), and towards the bottom of the equatorial domain (Asp25). It appears that binding and release are blocked by very different mutations. Binding may be relatively straightforward to understand but release is complex, involving ATPase and GroES mechanisms. The wide distribution of sites affecting substrate release suggests that global rearrangements accompany the release step.

References

Azem, A., Kessel, M. and Goloubinoff, P. (1994). Characterization of a functional GroEL14(GroES7)2 chaperonin hetero-oligomer. Science 265, 653-656.

Baneyx, F. and Gatenby, A.A. (1992). A mutation in GroEL interferes with protein folding by reducing the rate of discharge of sequestered polypeptides. J. Biol. Chem. 267, 11637-11644.

Bochkareva, ES and Girshovich, AS (1994) ATP induces non identity of two rings in chaperonin GroEL. J. Biol. Chem. 269, 23869-23871.

Braig, K., Otwinowski, Z., Hegde, R., Boisvert, D.C., Joachimiak, A., Horwich, A.L. and Sigler, P.B. (1994). The crystal structure of the bacterial chaperonin GroEL at 2.8 Å. Nature 371, 578-586.

Braig, K., Simon, M., Furuya, F., Hainfeld, J.F. and Horwich, A.L. (1993). A polypeptide bound by the chaperonin groEL is localized within a central cavity. Proc. Natl. Acad. Sci. USA 90, 3978-3982.

Chandrasekhar, G.N., Tilly, K., Woolford, C., Hendrix, R. and Georgopoulos, C. (1986). Purification and properties of the groES morphogenetic protein of Escherichia coli. J. Biol. Chem. 261, 12414-12419.

Chen, S., Roseman, A.M., Hunter, A., Wood, S.P., Burston, S.G., Ranson, N., Clarke, A.R. and Saibil, H.R. (1994). Location of a folding protein and shape changes in GroEL-GroES complexes imaged by cryo-electron microscopy. Nature 371, 261-264.

Dubochet, J., Adrian, M., Chang, J.-J., Homo, J.-C., Lepault, J., McDowall, A.W. and Schultz, P. (1988). Cryo-electron microscopy of vitrified specimens. Quart. Rev. Biophys. 21, 129-228.

Fenton, W.A., Kashi, Y., Furtak, K. and Horwich, A.L. (1994). Residues in chaperonin GroEL required for polypeptide binding and release. Nature 371, 614-619.

Fisher, M.T. and Yuan, X. (1994). The rates of commitment to renaturation of rhodanese and glutamine synthetase in the presence of the GroE chaperonins. J. Biol. Chem.

Fisher, M. (1994). The effect of groES on the groEL-dependent assembly of dodecameric glutamine synthetase in the presence of ATP and ADP. J. Biol. Chem. 269, 13629-13636.

Gatenby, A.A. and Viitanen, P.V. (1994). Structural and functional aspects of chaperonin-mediated protein folding. Annu. Rev. Plant Physiol. Plant Mol. Biol. 45, 469-491.

Georgopoulos, C.P., Hendrix, R.W., Casjens, S.R. and Kaiser, A.D. (1973). Host participation in bacteriophage lambda head assembly. J. Mol. Biol. 76, 45-60.

Gerstein, M., Lesk, A.M. and Chothia, C. (1994). Structural mechanisms for domain movements in proteins. Biochemistry 33, 6739-6749.

Goloubinoff, P., Christeller, J.T., Gatenby, A.A. and Lorimer, G.H. (1989). Reconstitution of active dimeric ribulose bisphosphate carboxylase from an unfolded state depends on two chaperonin proteins and Mg-ATP. Nature 342, 884-889.

Harris, J.R., Zahn, R. and Plückthun, A. (1994). Transmission electron microscopy of GroEL, GroES and the symmetrical GroEL/ES complex. J. Struct. Biol. 112, 216-230.

Hartl, F.-U. and Martin, J. (1995). Molecular chaperones in cellular protein folding. Curr. Opinion in Struct. Biol. 5, 92-102.

Hendrix, R. (1979). Purification and properties of GroE, a host protein involved in bacteriophage assembly. J. Mol.Biol. 129, 375-392.

Hohn, T., Hohn, B., Engel, A., Wurtz, M. and Smith, P.R. (1979). Isolation and characterization of the host protein GroE involved in bacteriophage lambda assembly. J. Mol.Biol. 129, 359-373.

Horovitz, A., Bochkareva, E.S. and Girshovich, A.S. (1993). The N terminus of the molecular chaperone GroEL is a crucial structural element for its assembly. J. Biol. Chem. 268, 9957-9959.

Hutchinson, E.G., Tichelaar, W., Hofhaus, G., Weiss, H. and Leonard, K. (1989). Identification and electron microscopic analysis of a chaperonin oligomer from Neurospora crassa mitochondria. EMBO J. 8, 1485-1490.

Ishii, N., Taguchi, H., Sumi, M. and Yoshida, M. (1992). Structure of holochaperonin studied with electron microscopy. FEBS Letters 299, 169-174.

Ishii, N., Taguchi, H., Sasabe, H. and Yoshida, M. (1994). Folding intermediate binds to the bottom of bullet-shaped holo-chaperonin and is readily accessible to antibody. J. Molec. Biol. 236, 691-696.

Jackson, G.S., Staniforth, R.A., Halsall, D.J., Atkinson, T., Holbrook, J.J., Clarke, A.R. and Burston, S.G. (1993). Binding and hydrolysis of nucleotides in the chaperonin catalytic cycle: implications for the mechanism of assisted protein folding. Biochemistry 32, 2554-2563.

Kubota, H., Hynes, G., Carne, A., Ashworth, A. and Willison, K. (1994). Identification of six Tcp-1-related genes encoding divergent subunits of the TCP-1-containing chaperonin. Current Biol. 4, 89-99.

Landry, S.J., Zeilstra-Ryalls, J., Fayet, O., Georgopoulos, C. and Gierasch, L.M. (1993). Characterization of a functionally important mobile domain of GroES. Nature 364, 255-258.

Langer, T., Pfeifer, G., Martin, J., Baumeister, W. and Hartl, F.U. (1992). Chaperonin-mediated protein folding: GroES binds to one end of the GroEL cylinder, which accommodates the protein substrate within its central cavity. EMBO J. 11, 4757-4765.

Marco, S., Carascossa, J.L. and Valpuesta, J.M. (1994a). Reversible interaction of J-actin along the channel of TCP1 cytoplasmic chaperonin. Biophys. J. 67, 364-368.

Marco, S., Ureña, D., Carrascosa, J.L., Waldmann, T., Peters, J., Hegerl, R., Pfeifer, G., Sack-Kongehl, H. & Baumeister, W. (1994b). The molecular chaperone TF55. Assesment of symmetry. FEBS Letters 341, 152-155.

Martin, J., Langer, T., Boteva, R., Schramel, A., Horwich, A. and Hartl, F.-U. (1991). Chaperonin-mediated protein folding at the surface of groEL through a 'molten globule'-like intermediate. Nature 352, 36-42.

Martin, J., Mayhew, M., Langer, T. and Hartl, F.U. (1993). The reaction cycle of GroEL and GroES in chaperonin-assisted protein folding. Nature 366, 228-233.

Phipps, B.M., Hoffman, A., Stetter, K.O. and Baumeister, W. (1991). A novel ATPase complex selectively accumulated upon heat shock is a major cellular component of thermophilic archaebacteria. EMBO J. 10, 1711-1722.

Pushkin, A.V., Tsuprun, V.L., Solojeva, N.A., Shubin, V.V., Evstigneeva, Z.G. and Kretovich, W.L. (1982). High molecular weight pea leaf protein similar to the groE protein of Escherichia coli. Biochim. Biophys. Acta 704, 379-384.

Saibil, H.R. (1994). How chaperonins tell wrong from right. Nature Structural Biology 1, 838-842.

Saibil, H.R., Zheng, D., Roseman, A.M., Hunter, A.S., Watson, G.M.F., Chen, S., auf der Mauer, A., O'Hara, B.P., Wood, S.P., Mann, N.H., Barnett, L.K. and Ellis, R.J. (1993). ATP induces large quaternary rearrangements in a cage-like chaperonin structure. Current Biol. 3, 265-273.

Schmidt, M., Buchner, J., Todd, M..J, Lorimer, G.H. and Viitanen, P.V. (1994). On the role of GroES in the chaperonin-assisted folding reaction. J. Biol. Chem. 267, 10304-10311.

Schmidt, M., Rutkat, K., Rachel, R., Pfeiffer, G., Jaenicke, R., Viitanen, P., Lorimer, G. and Buchner, J. (1994). Symmetric complexes of GroE chaperonins as part of the functional cycle. Science 265, 656-659.

Staniforth, R.A., Burston, S.G., Atkinson, T., and Clarke, A.R. (1994). Affinity of chaperonin-60 for a protein substrate and its modulation by nucleotides and chaperonin-10. Biochem. J. 300, 651-658.

Taguchi, H., Makino, Y. and Yoshida, M. (1994) Monomeric chaperonin-60 and its 50-kDa fragment possess the ability to interact with non-native proteins, to suppress aggregation and to promote protein folding. J. Biol. Chem. 269, 8529-8534.

Todd, M.J., Viitanen, P.V. and Lorimer, G.H. (1994). Dynamics of the chaperonin ATPase cycle: Implications for facilitated protein folding. Science 265, 659-666.

van der Vies, S.M., Gatenby, A.A. and Georgopoulos, C. (1994). Bacteriophage T4 encodes a co-chaperonin that can substitute for Escherichia coli GroES in protein folding. Nature 368, 654-656. [not currently cited]

Weissman, J.S., Kashi, Y., Fenton, W.A. and Horwich, A.L. (1994). GroEL-mediated protein folding proceeds by multiple rounds of binding and release of nonnative forms. Cell 78, 693-702.

Yifrach, O. and Horovitz, A. (1994). Two lines of allosteric communication in the oligomeric chaperonin GroEL are revealed by the single mutation Arg196®Ala. J. Mol. Biol. 243, 397-401.

Zahn, R., Spitzfaden, C., Ottiger, M., Wüthrich, K. and Plückthun, A. (1994). Destabilization of the complete protein secondary structure on binding to the chaperone GroEL. Nature 368, 261-265.

List of Figures

1. line drawing of subunits in oligomer

2. Cryo-EM averaged end and side views of GroEL (a,b) from Chen et. al ., (1994)

3. 3D recontructions of EL, EL ATP, EL ES ATP

4. Colour: backbone & full chain rainbow colour-coded structures of GroEL subunit

5. Ribbon diagrams of GroES structure, top & side views

6. EL-MDH vs EL-ES-MDH subunit domain & hinge diagram

7. Cryo EM of ELES, football side views; EL ES vs EL ES MDH sections

8. Cryo EM section of GroEL vs GroEL-ATP

Figure Legends

1. The cpn60 14-mer. Subunit outlines based on the X-ray crystal structure of GroEL. Diagram adapted from Braig et al. (1994).

2. Cryo EM averaged end and side views of GroEL (a,b), from Chen et al., (1994); and the same views from preliminary data on CCT (c,d) (Chen, Liou, Roseman, Willison & Saibil, unpublished).

3. 3D reconstructions of GroEL, GroEL-ATP and GroEL-GroES-ATP from cryo EM. From work of Chen et al. (1994). (a) Surface-rendered view of the GroEL-GroES-ATP complex at 30 Å resolution determined from cryo EM. The GroES ring is seen as a disk above the GroEL.

4. Atomic resolution structure of the GroEL subunit (Braig et al., 1994). (a) I-carbon and (b) full structure, colour coded in rainbow colours from N (red) to C (blue). In this view, the equatorial domain is the large mass containing both termini at the bottom of the subunit, connected by the antiparallel intermediate domain (cyan and yellow) to the large apical domain (green and cyan) at the top. Images kindly provided by Dr Kerstin Braig.

5. Ribbon diagrams of the GroES structure viewed (a) from above and (b) from the side, showing the J-barrel structure of the subunit with a J-hairpin forming the roof of the dome-like heptamer. The N and C termini are labelled and and point radially outwards. In (a), the positions of two glutamic acid residues in the J hairpin are indicated as negative charges. The broken ends of the disordered mobile loop, which is not seen in the map, are indicated by dashed arrows in (b). Images kindly provided by Dr John Hunt.

6. Schematic diagram of the subunit arrangement in a hypothetical slice through the oligomer, showing the major functional sites. (a) GroEL, based on the crystal structure with bound substrate (shaded light to dark) as imaged by cryo EM. A, apical domain; I, intermediate domain; E, equatorial domain. The apical domains form a ring of hand-like structures with the substrate binding sites on the fingers protruding into the central channel. Sites of potential hinge rotations are indicated by black dots. The notch in the equatorial domain represents the ATP-binding cleft. The interaction across the equatorial plane is shown as pairs of wavy surfaces. (b) the folding complex GroEL-MDH-GroES-ATP, rotating the subunit domains and adding the GroES (dark gray) and substrate (shaded) densities according to cryo EM observations.

7. (a) Side view of the GroEL-GroES complex averaged from 200 cryo EM images (Chen et al, 1994). (b) Side view average of "football" complexes from cryo EM (Chen, S., Roseman, A., Saibil, H., unpublished observations). (c) Section through the complex in (a) showing the large reorientation of the apical domains in contact with GroES, forming the contact with the same region as the substrate binding site. The GroEL cylinder is greatly elongated. (d) section of a 3D reconstruction of the folding complex GroEL-MDH-GroES-ATP trapped by vitrification after 12 sec of folding from cryo EM (Chen et al., 1994). The MDH substrate density is found in the opposite ring to GroES.

8. Superposed outlines of sections through GroEL and GroEL-ATP (shaded) from cryo EM (Chen et al, 1994), showing the slight but distinct outward rotation of the apical domains in the ATP form. These displacements have the effect of lengthening the cylinder and widening the binding cavity.