• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Jun 7, 2005; 102(23): 8144–8149.
Published online May 26, 2005. doi:  10.1073/pnas.0500048102
PMCID: PMC1149413

The T4-encoded cochaperonin, gp31, has unique properties that explain its requirement for the folding of the T4 major capsid protein


The morphogenesis of bacteriophage T4 requires a specialized bacteriophage-encoded molecular chaperone (gp31) that is essential for the folding of the T4 major capsid protein (gp23). gp31 is related to GroES, the chaperonin of the Escherichia coli host because it displays a similar overall structure and properties. Why GroES is unable to fold the T4 capsid protein in conjunction with GroEL is unknown. Here we show that gp23 binds to the GroEL heptameric ring opposite to the ring that is bound by gp31 (the so-called trans-ring), while no binding to the trans-ring of the GroEL-GroES complex is observed. Although gp23 can be enclosed within the folding cage of the GroEL-gp31 complex, encapsulation within the GroEL-GroES complex is not possible. So it appears that folding of the T4 major capsid protein requires a gp31-dependent cis-folding mechanism likely inside an enlarged “Anfinsen cage” provided by GroEL and gp31.

Keywords: chaperones, protein folding, viral capsid assembly

The Escherichia coli chaperonin GroEL and its cochaperonin GroES constitute a molecular machine that assists the folding of nonnative and misfolded polypeptides (1-3). The GroEL chaperonin is composed of 14 57-kDa subunits that are organized into two heptameric rings that are stacked back to back, thus forming a double-ring structure. Each GroEL subunit is composed of three domains: an equatorial domain that contains the nucleotide-binding site and forms the interface between the two rings, an apical domain that binds substrate polypeptide and GroES, and an intermediate domain that connects the equatorial and apical domain (4). The cochaperonin GroES forms a single-ring structure of seven identical 10-kDa subunits (5). Each GroES subunit possesses a mobile loop with a hydrophobic tripeptide that is required for the interaction with GroEL. The mechanism by which the E. coli chaperonin machine assists polypeptide folding can be compared with that of a two-cylinder reciprocal engine (6). The chaperonin-folding cycle starts with the binding of a substrate protein to the hydrophobic inner surface of one of the GroEL rings. After cooperative binding of ATP and GroES to the substrate-bound ring (cis-ring), the apical domains reorient and sequester the hydrophobic surfaces, resulting in the displacement of the substrate into an enclosed and enlarged cavity in which it is free to fold (7, 8). Subsequent ATP hydrolysis in the cis-ring primes the complex for release of the bound GroES. Binding of substrate and ATP to the unliganded GroEL trans ring induces the release of GroES, substrate, and ADP from the cis ring. If the substrate has not folded to its native conformation, it will rebind to GroEL to undergo a further round of folding. The cycling time of ≈8 s is largely determined by the rate of ATP hydrolysis in the cis ring (7). The E. coli chaperonin machine assists folding by preventing aggregation (9), limiting conformational “choices” of the substrate (10, 11), and/or actively unfolding misfolded substrate polypeptides (12-14). It is estimated that ≈10-15% of the newly synthesized polypeptides in E. coli with a mass between 20 and 60 kDa transit through the GroEL-GroES chaperonin machine (15).

E. coli bacteriophages such as λ, HK97, T5, Mu, and PRD1 strictly depend on GroEL-GroES for growth (16). In many cases it is the structural proteins of these viruses that require the folding assistance of the GroEL-GroES chaperonin machine. Bacteriophage T4 forms an intriguing exception to this imperative. Folding of the T4 major capsid protein (gp23) requires an altered chaperonin complex, i.e., the E. coli cochaperonin GroES needs to be replaced by the bacteriophage-encoded cochaperonin, gp31 (17-19). Why GroES, in conjunction with GroEL, is incapable of helping the T4 capsid protein to fold is unknown. When GroES is replaced by gp31 in E. coli, thus creating a situation in which gp31 is the sole cochaperonin, the bacterium remains perfectly viable (20). This finding signifies that the hybrid chaperonin complex, GroEL-gp31 has no problem folding all of the E. coli proteins that normally depend on GroEL-GroES. In vitro, refolding of substrates such as ribulose bisphosphate carboxylase/oxygenase (Rubisco) and citrate synthase is also facilitated by the hybrid chaperonin complex (19-21). It thus appears that gp31 is capable of performing the same function as GroES but in addition can assist the folding of the T4 major capsid protein in conjunction with GroEL.

Based on structural differences between GroES and gp31, it has been postulated that the folding cavity of the GroEL-gp31 complex may be larger than that of the E. coli chaperonin machine (22) to accommodate the larger or different-shaped capsid protein. The vast majority of the E. coli proteins that require GroEL-GroES for folding were found to be <60 kDa, suggesting that larger proteins cannot be accommodated within the folding cavity of the GroEL-GroES complex (15). With a molecular mass of ≈56 kDa, the bacteriophage T4 major capsid protein approaches this size limit and may therefore be better aided by a chaperonin machine with a somewhat larger folding cavity. Although moderately-sized proteins like rhodanese (33 kDa), ornithine transcarbamoylase (36 kDa), or even Rubisco (50.5 kDa) can be enclosed within the cis-cavity of GroEL-GroES, a large protein like aconitase (82 kDa) cannot and appears to use a trans-folding mechanism (23, 24). It is possible that gp31 exerts its function in the folding of the T4 major capsid protein by enforcing folding from the trans position and/or by changing the kinetics of the folding cycle. To gain insight into the mechanism by which the T4 capsid protein is folded by using the GroEL-gp31 folding machine, we determined conditions to monitor refolding of the capsid protein in vitro and analyzed the nature of the binding of gp23 to the chaperonin complexes GroEL-gp31 and GroEL-GroES.

Materials and Methods

Expression and Purification of Proteins. GroEL and GroES were overexpressed in E. coli MC1009 by using a groE operon-based plasmid. Expression was induced with l-arabinose (0.001% wt/vol) for ≈16 h. Purification was as described (25-27). For GroES the MonoQ-Sepharose chromatography was replaced by a heat treatment, i.e., 10 min at 72°C. GroES was recovered in the supernatant after centrifugation (Sorvall, SS34 rotor) at 11,000 rpm for 2 h at 4°C. The single-ring GroEL mutant, SR1, was expressed and purified from E. coli as described (28). Plasmid pSR1 was a gift from F.-U. Hartl, Max Planck Institute, Martinsried, Germany. gp31 was expressed in E. coli and purified as described (29). E. coli BL21(DE3) containing the isopropyl β-d-thiogalactoside-inducible pET2331 (a gift from L. Black, University of Maryland, Baltimore) was used to produce polyheads, i.e., large tube-like structures composed exclusively of gp23. Polyheads were isolated as described (16) and dissociated by dialysis against 10 mM Tris·HCl (pH 9.0) for 4 h at 4°C. The gp23 solution was cleared by centrifugation (Beckman, Ti50.2 rotor) at 25,000 rpm for 45 min at 4°C. Rubisco from Rhodospirillum rubrum was expressed in E. coli and purified as described (30).

Miscellaneous. The protein concentrations are final concentrations of GroEL 14-mer, SR1 7-mer, GroES 7-mer, gp31 7-mer, gp23 monomer, and Rubisco monomer. Substrate proteins were unfolded in 6 M urea in an Eppendorf thermomixer at 25°C. Gp23 was unfolded for at least 1 h. Rubisco was unfolded in the presence of 10 mM DTT for at least 2 h. Refolding experiments were carried out by using buffer A (50 mM Tris·HCl, pH7.5/10 mM KCl/10 mM MgCl2). Binding experiments (i.e., in the presence of ADP) involving GroEL were carried out in buffer B (50 mM Tris·HCl, pH 7.5/50 mM KCl/5 mM MgCl2) and experiments by using SR1 in buffer C (50 mM Tris·HCl, pH 7.5/5 mM KCl/12 mM MgCl2).

Gel Filtration Chromatography. Gel filtration analysis was performed by using an AKTA-explorer chromatography system (Amersham Pharmacia) equipped with a TSKG4000SWxl (TosoHaas, Montgomeryville, PA) column and fitted with an RF-10Axl fluorescence detector (Shimadzu). Tryptophan fluorescence emission was monitored in line with gel filtration at 340 nm. Excitation was at 295 nm. Before injection, protein samples were passed over a 0.45-μm membrane filter [poly(vinylidene difluoride) durapore, Millipore], and aliquots of 100 μl were applied to the column at room temperature by using a flow of 0.75 ml/min. After gel filtration, protein in the peak fractions corresponding to the elution positions of GroEL, gp23, gp31, and GroES was precipitated with acetone [80% (vol/vol) final concentration] and analyzed by SDS/PAGE and Western blotting.

SDS/PAGE and Western Blot Analysis. To separate the chaperonin from the cochaperonin(s) and the chaperonin from the substrate, a 15% or a 7% polyacrylamide gel was used, respectively. Proteins were visualized with Coomassie brilliant blue (CBB). Because gp23 (56 kDa) and Rubisco (50.5 kDa) often migrate too close to the GroEL (57 kDa) band in the acrylamide gel to be detected with CBB, we used Abs raised in guinea pigs and rabbits to detect gp23 and Rubisco, respectively. Primary Abs were detected by using secondary Abs conjugated to horseradish peroxidase and visualized by enhanced chemiluminescence.

Chaperonin-Assisted Refolding. Gp23 was unfolded as described above. Binary complexes of GroEL and gp23 were prepared by subsequently adding four equal portions of unfolded gp23 (final concentration of 1 μM) to buffer A containing 1.3 μM GroEL. After incubation for 5 min at 25°C, either GroES (3.9 μM) or gp31 (3.9 μM) was added followed by addition of nucleotide (ATP or ADP) to a final concentration of 5 mM. The mixtures were incubated at 25°C for the time indicated and analyzed by gel filtration chromatography at room temperature in buffer A containing 0.01% (vol/vol) Tween-20 and 0.25 mM nucleotide (ATP or ADP). Peak fractions were analyzed by SDS/PAGE and Western blotting as described above.

Binding of Substrate to Asymmetric GroEL-GroES and GroEL-gp31 Complexes. First, binary complexes of GroEL with gp31 and GroEL with GroES were prepared. GroEL (0.8 μM) was incubated with either GroES (11.4 μM) or gp31 (11.4 μM) in buffer B containing 1 mM ADP for 10 min at 25°C. Next, unfolded substrate (gp23 or Rubisco) was added to a final concentration of 0.65 μM, followed by incubation for 10 min at 25°C. The protein mixtures were subjected to gel filtration chromatography at room temperature in buffer B containing 0.25 mM ADP. Peak fractions were analyzed by SDS/PAGE and Western blotting as described above.

Limited Proteolysis of Various Chaperonin(-Substrate) Complexes. Binary complexes of GroEL and cochaperonin were produced as follows: GroEL (0.8 μM) or SR1 (0.8 μM) was incubated in buffer B or C, respectively, for 10 min at 25°C in the presence of either GroES (1.9 μM), gp31 (1.9 μM), or no cochaperonin, with or without ADP (1 mM). Ternary complexes were obtained in two ways that differ in the order of addition of cochaperonin and unfolded substrate (either gp23 or Rubisco) to the incubation mixture containing GroEL: (i) Binary complexes of GroEL and cochaperonin were preformed as described above followed by addition of the unfolded substrate to a final concentration of 0.65 μM. Incubation was for 5 min at 25°C. (ii) GroEL-substrate complexes were preformed followed by addition of cochaperonin. Unfolded substrate (0.65 μM) was rapidly mixed with buffer B containing GroEL (0.8 μM) and incubated for 5 min at 25°C. Subsequently, cochaperonin (GroES or gp31) was added to a final concentration of 1.9 μM, followed by addition of ADP (1 mM), and complex formation was allowed to proceed for 10 min at 25°C. Protein mixtures without cochaperonin or in the absence of ADP served as controls. Ternary SR1 complexes were produced in buffer C followed by analysis as described for GroEL (in ii). Removal of gp31 from ternary GroEL complexes was achieved by addition of EDTA (50 mM) and from ternary SR1 complexes by addition of EDTA (100 mM) followed by incubation for 3 min at 0°C. Protease treatment of the various protein mixtures was performed by addition of proteinase K to a final concentration of 1.2 μg/ml. After incubation for 10 min at 25°C with rotary shaking, proteolysis was stopped by adding PMSF (1 mM). Protein mixtures were analyzed by SDS/PAGE and Western blotting.


Refolding of the T4 Major Capsid Protein. Folding of the bacteriophage T4 major capsid protein (gp23) in vivo requires the hybrid GroEL-gp31 chaperonin system, which possesses a unique folding activity that is lacking from the GroEL-GroES system. To investigate the functional difference(s) between these two systems, their various constituents were produced in E. coli and purified to homogeneity. Because gp23 is a structural protein, refolding could not readily be determined by using an activity assay. We therefore turned to gel filtration chromatography and immunodetection. Native gp23 was present in a hexameric form when kept at room temperature because it behaves on a gel filtration column as a protein with an apparent Mr of ≈348 kDa (see Fig. 1A; calculated size of a gp23 hexamer is 336 kDa). Because it is unlikely that incorrectly refolded gp23 monomers will be able to assemble into oligomers, the formation of gp23 hexamers was taken as proof of proper refolding (see also ref. 31).

Fig. 1.
Chaperonin-assisted refolding of the capsid protein, gp23. Binary complexes of GroEL and gp23 were incubated with (i) ATP, (ii) GroES and ATP, (iii) gp31, (iv) gp31 and ATP, and (v) gp31 and ADP. Mixtures i-iv were incubated for 1 min and mixture v was ...

First, refolding of gp23 in the absence of chaperonins was analyzed. To this end, urea-unfolded gp23 was diluted into buffer without urea and allowed to refold at 25°C. The protein sample was then subjected to gel filtration chromatography. Remarkably, gp23 was not detected in any of the column fractions (Fig. 1A, spontaneously “refolded” gp23). In contrast, native gp23 eluted from the column at a position indicative of a hexamer (Fig. 1A, native gp23). Possibly, on removal of denaturant by dilution, unfolded gp23 forms unstable folding intermediates that form aggregates, which are removed by filtration before chromatography and/or stick to the column during gel filtration. This behavior indicates that gp23 does not refold spontaneously under the conditions tested. We then established that unfolded gp23 binds to GroEL in vitro in the absence of nucleotide and cochaperonin, whereas native gp23 does not bind to GroEL (ref. 31 and data not shown). Next, the fate of GroEL-bound gp23 was determined upon addition of ATP or upon addition of cochaperonin and ATP. Fig. 1A shows the gel filtration profiles of several refolding mixtures after 1 min of refolding. It is apparent that correctly refolded gp23 (indicated by an inverted triangle) is only formed when GroEL, gp31, and ATP were present [Fig. 1A, (iv) GroEL-gp23 + gp31 + ATP]. When ATP was omitted from the folding reaction, refolding of gp23 was not observed [Fig. 1A, (iii) GroEL-gp23 + gp31, no ATP], and gp23 was found in complex with GroEL-gp31 (Fig. 1B, lanes 9-11). Note that the small amount of ATP that is present in the elution buffer did allow ternary complex formation but did not allow productive folding. Similarly, when ATP was substituted by ADP, folded hexameric gp23 was not detected, and gp23 was associated with GroEL-gp31 [Fig. 1A, (v) GroEL-gp23 + gp31 + ADP and Fig. 1B, lanes 17-19]. Note that incubation with ADP was for 1 h, ruling out the possibility that gp23 refolds slowly in the presence of ADP. In contrast to GroEL-gp31, neither GroEL-GroES nor GroEL alone did facilitate productive refolding of gp23. Rather, gp23 was found associated with GroEL-GroES or GroEL (Fig. 1B, lanes 5 and 2, respectively). Even when folding was allowed to proceed for 60 min, gp23 was still bound (data not shown), suggesting that it is unable to fold or be released. In conclusion, refolding of gp23 in vitro was achieved by GroEL-gp31 only in the presence of ATP. Under these conditions refolding is fast and efficient.

Binding of the Capsid Protein to Asymmetric Chaperonin Complexes. So what might be the difference between the GroEL-GroES and GroEL-gp31 chaperonin machine? Based on studies with GroEL-GroES and substrate proteins such as Rubisco, it has become commonly accepted that the asymmetric bullet-shaped Gro-EL(ADP7)-GroES complex is an important intermediate in the folding cycle as it binds substrate with high affinity (7). Such a complex consists of one GroEL ring that has ADP and GroES bound to it, whereas the opposite (trans) GroEL ring, which is free of ligands, can bind a nonnative polypeptide. It was shown previously that limited proteolysis could be used to assess whether asymmetric GroEL-GroES complexes are formed. Proteinase K treatment of such complexes results in removal of 16-aa residues from the C termini of the GroEL subunits that are not in contact with GroES, whereas the GroEL subunits in contact with GroES remain intact (32, 33). We investigated whether gp31 was likewise able to protect the GroEL subunits from truncation. When GroEL-gp31 complexes formed in the presence of ADP were treated with proteinase K, half of the GroEL subunits remain undigested, indicating that asymmetric complexes were indeed formed (Fig. 2B, lanes 11 and 12). In the absence of ADP, however, proteolysis protection of GroEL subunits by gp31 was not observed (Fig. 2B, lanes 9 and 10). Identical results were obtained when gp31 was substituted by GroES (Fig. 2B, lanes 5-8). Note that when gp31 and GroES are in complex with GroEL both are less susceptible to proteinase K than unbound cochaperonin (Fig. 2C, compare lanes 12 and 10 and lanes 8 and 6). Electrospray ionization-MS analysis revealed that, as reported for GroEL-GroES, 16-aa residues from the C terminus of the GroEL subunits were removed in the proteinase K-treated GroEL-gp31 complexes (E. van Duijn and A. J. R. Heck, personal communication).

Fig. 2.
Formation of asymmetric GroEL-gp31 complexes. (A) Schematic representation of the experimental procedure. GroEL was incubated with GroES or gp31 in the presence or absence of ADP and subsequently treated with proteinase K as indicated. The protein mixtures ...

A number of “typical” GroEL substrates (e.g., ornithine transcarbamoylase, rhodanese, or methylmalonyl CoA mutase) have been shown to be strongly susceptible to proteolysis when bound to the GroEL trans-ring of GroEL(ADP7)-GroES complexes (32). Does the T4 capsid protein gp23 behave as a typical GroEL substrate in this respect? To answer this question, we first tested whether unfolded, purified gp23 is able to bind to the GroEL trans-ring of ADP bullets of GroEL-GroES and GroEL-gp31. First, binary complexes of GroEL and the cochaperonin were made by using a large excess of cochaperonin to ensure that all GroEL 14 mers were occupied. Proteinase K treatment revealed that under these conditions only one cochaperonin molecule was bound to each GroEL 14-mer (data not shown). Subsequently, unfolded gp23 or unfolded Rubisco (control) was added, and after incubation at 25°C, the mixtures were subjected to gel filtration chromatography and the proteins in the peak fractions identified by SDS/PAGE and Western blotting. In Fig. 3C, lane 4, it can be seen that gp23 coelutes with GroEL-gp31, suggesting that trans-ternary complexes were formed. Interestingly, no gp23 was found in complex with GroEL-GroES (Fig. 3C, lane 2), indicating that gp23 is unable to bind to the trans-ring of the GroEL(ADP7)-GroES bullet even though it binds efficiently to free GroEL (Fig. 1 and ref. 31). In contrast, the typical GroEL substrate Rubisco did form trans-ternary complexes with both GroEL-gp31 and GroEL-GroES in the presence of ADP (Fig. 3B, lanes 4 and 2, respectively). These results indicate that binding of gp31 to GroEL changes the GroEL trans-ring to a conformation that is capable of binding the T4 capsid protein. Under conditions where trans-ternary complexes are formed (see above), proteinase K treatment will lead to the complete digestion of gp23. As is shown in Fig. 4B Bottom, lane 4, this is indeed the case.

Fig. 3.
Formation of trans-ternary gp23-GroEL-gp31 complexes. (A) Schematic representation of the experimental procedure. (B and C) Unfolded Rubisco (B) or unfolded gp23 (C) was added to preformed GroEL-GroES (lanes 1 and 2) or GroEL-gp31 (lanes 3 and 4) complexes. ...
Fig. 4.
Gp23 bound to the trans-ring of the GroEL-gp31 complex is susceptible to proteolysis. (A) Schematic representation of the experimental procedure. gp31 is represented by the black cap. (B) SDS/PAGE analysis and CBB staining (Top and Middle) or immunodetection ...

gp31 Protects GroEL-Bound gp23 from Proteolysis, but GroES Does Not. To determine whether gp23 can be encapsulated in the GroEL-gp31 folding chamber, we performed a set of experiments that was inspired by Weissman and coworkers (31). Those authors demonstrated that binding of GroES to a GroEL-substrate complex results in enclosure of the substrate within the cis-cavity of GroEL-GroES (formation of cis-ternary complexes), resulting in protection of the substrate against proteolysis. Resistance of a substrate to digestion by proteinase K therefore can be used as an experimental paradigm for enclosure of the substrate within the cis-cavity of the folding machine. First, binary complexes of GroEL and gp23 or GroEL and Rubisco (control) were formed, followed by addition of ADP and either GroES or gp31. After incubation, the mixtures were treated with proteinase K. The remarkable result was that gp23 is protected from degradation in the presence of gp31, but remains sensitive to proteolysis in the presence of GroES (Fig. 5C Lower, compare lanes 4 and 8). When gp31 was removed from the GroEL-gp23-gp31 complexes by incubation with EDTA, gp23 became again susceptible to proteolysis (Fig. 5C Lower, compare lanes 8 and 9). Together with the observation that gp23 remains bound to GroEL-gp31 in the presence of ADP (Fig. 1B, lanes 17-19), this finding indicates that the proteolysis protection of GroEL-bound gp23 in the presence of gp31 and ADP is the result of encapsulation rather than folding. When complex formation of GroEL and gp31 was precluded (i.e., in the absence of ADP), gp31 did not protect gp23 against proteolytic attack (Fig. 5C Lower, lane 6). These results demonstrate that gp23 can be encapsulated in the cis-cavity of GroEL-gp31, but that cis-ternary GroEL-gp23-GroES complexes cannot be formed, even though unfolded gp23 and GroES bind efficiently to GroEL separately (data not shown). The inability to form cis-ternary GroEL-gp23-GroES complexes is most likely resulting from the fact that GroES cannot properly bind over GroEL-bound gp23. Because this conclusion is based on negative data on proteinase K protection of gp23, we cannot rigorously rule out that GroES can encapsulate gp23 but that the stability of the cis-ternary complex is so low that gp23 remains susceptible to digestion. As a control the proteinase K susceptibility of Rubisco, a substrate that can be refolded by either the GroEL-GroES or the GroEL-gp31 complex (19), was determined. Fig. 5B Lower, lanes 4 and 8 shows that in the presence of ADP, both GroES and gp31 protected GroEL-bound Rubisco from proteolysis. In the absence of ADP (when ternary complexes were not formed), Rubisco was completely digested (Fig. 5B Lower, lanes 2 and 6). Thus, Rubisco can be encapsulated in either of the two chaperonin machines.

Fig. 5.
GroEL-bound gp23 is protected from proteolysis by gp31 but not by GroES. (A) Schematic representation of the experimental procedure. The cochaperonin (GroES or gp31) with or without ADP was added to preformed binary complexes of GroEL and unfolded substrate ...

The Capsid Protein Is Encapsulated in the SR1-gp31 Chaperonin Complex. To further substantiate that protection against proteolysis of gp23 in the ternary complex is the result of encapsulation in the cis-cavity, proteolysis protection assays were performed by using the SR1 mutant of GroEL. SR1 binds substrate and GroES like WT GroEL, but consists of just a single heptameric ring (28). Hence, substrate and GroES obligatory bind to the same ring, resulting in the formation of cis-only complexes. When gp23 was bound to SR1 and the complex treated with proteinase K, the capsid protein was readily degraded (Fig. 6C, lane 2). However, on addition of ADP and gp31, SR1-bound gp23 became resistant to proteolysis (Fig. 6C, lane 13). As a result of the obligatory formation of cis-only SR1-gp23-gp31 complexes, the proteolysis protection was 100% (Fig. 6C, compare lanes 12 and 13). Contrary to gp31, GroES did not protect SR1-bound gp23 from proteolysis in the presence of ADP (Fig. 6C, lane 8). Similarly, when ATP was used instead of ADP, gp31 protected SR1-bound gp23 from proteolysis, whereas GroES did not (Fig. 6C, compare lanes 14 and 9). When gp31 was removed from the cis-only SR1-gp23-gp31 (ADP) complexes by incubation with EDTA at 0°C, proteinase K treatment resulted in complete digestion of gp23 (Fig. 6C, lane 15). Proteinase K treatment of SR1 resulted in the removal of 16-aa residues from the C terminus of the SR1 subunits (Fig. 6B Upper; E. van Duijn and A. J. R. Heck, personal communication), which in case of the SR1-gp23-gp31 complex must have occurred at the plane of the equatorial domains of the SR1 subunits where the flexible C termini are readily accessible (Fig. 6B Upper, compare lane 13 and 14). This result is in contrast to the GroEL 14-mer where the C termini of the cis-ring are protected by the subunits of the GroEL trans-ring. In conclusion, gp23 can be enclosed within the SR1-gp31 cavity, whereas (stable) ternary SR1-gp23-GroES complexes cannot be formed.

Fig. 6.
Gp23 is encapsulated in the SR1-gp31 complex. (A) Schematic representation of the experimental procedure. The cochaperonin (GroES or gp31) with or without nucleotide (ADP or ATP) was added to preformed binary complexes of SR1 and unfolded gp23. The obtained ...


The experiments described in this article suggest a possible explanation for the puzzling observation that the bacteriophage T4-encoded cochaperonin, gp31, together with the E. coli chaperonin GroEL is able to fold the T4 major capsid protein, gp23, whereas the host cochaperonin GroES is unable to perform this function. We show that in vitro, nonnative gp23 binds to the trans-ring of GroEL(ADP)-gp31 complexes and not to similar complexes formed by GroEL and GroES (see Fig. 3). This finding can be explained by the fact that binding of gp31 to GroEL induces specific conformational changes in the GroEL trans-ring such that gp23 binds with high affinity to the GroEL-gp31 “bullet.” Binding of GroES to GroEL may lead to a GroEL trans-ring with a different conformation, such that gp23 can no longer bind. This idea is supported by preliminary cryo-electron microscopy data of GroEL(ADP)-gp31 binary complexes, which shows small but distinct differences between the GroEL trans-rings of the two chaperonin complexes can be seen (D. K. Clare, P.J.B., H.H., S.M.V., and H. R. Saibil, unpublished results). In addition, we demonstrate another remarkable aspect of the GroEL-gp31 folding machine and its dedicated substrate, gp23, which relates to the differences in the actual size and properties of the folding chamber of the GroEL-gp31 complex as compared with that of GroEL-GroES. In the current view on GroEL-GroES-mediated folding, productive folding takes place while the substrate resides in the cis-cavity of the chaperonin complex. Here, we report that gp23 can be captured in the GroEL-gp31 folding chamber, but that encapsulation in the GroEL-GroES folding cage is not possible. Which could be the essential distinction(s) between the two cochaperonins to explain this distinct behavior? A comparison of the crystal structures of gp31 and GroES calls the attention to a number of differences. Compared with GroES, the gp31 monomer has a 12-residue insertion starting at position 82, a longer mobile loop (22 vs. 16 aa), no roof β-hairpin, and no tyrosine at position 71, a residue that is present in all GroES-like cochaperonins and protrudes into the GroEL-GroES folding cavity. These differences, either by themselves or in combination, have been postulated to lead to a slight expansion of the GroEL-gp31 folding cage relative to the GroEL-GroES cavity (22). Preliminary cryo-electron microscopic images of the GroEL(ADP)-gp31 complex show that gp31 protrudes less far into the cavity than GroES when compared with the GroEL(ADP)-GroES crystal structure (D. K. Clare, P.J.B., H.H., S.M.V., and H. R. Saibil, unpublished results), suggestive of a larger folding cavity. Other explanations for the specific cochaperonin requirement of gp23 have been put forward, e.g., the kinetics of the GroEL-gp31 folding cycle might differ from that of the GroEL-GroES cycle. Clearly, our results do not exclude this possibility. In fact, it is likely that coevolution of gp23 and gp31 has led to both changes in the nature of the (gp31 part of the) GroEL-gp31 folding cage, as well as to an optimal tuning of the timer of the folding cycle. In this respect it is interesting to recall some of the findings obtained by Wang et al. (34), who performed directed evolution on the GroEL-GroES system to optimize it for the folding of GFP. They observed that improved GroEL/S variants displayed changes in the polarity of the folding cage as well as in the kinetics of the folding cycle.

Could the chaperonin function of gp31 be exerted by a trans-folding mechanism? For several reasons this does not seem likely. Our data indicate that gp23 can be encapsulated in the cis-cavity of the GroEL-gp31 complex, which strongly suggests that gp23 folds in cis. Moreover, we show that gp23 is folded fast and efficiently by GroEL-gp31. For GroEL-GroES, trans-folding of substrates like malate dehydrogenase and Rubisco is considerably less efficient than cis-folding (24). If the same would hold true for the GroEL-gp31 system, a trans-folding mechanism would most likely not be able to supply the “folding power” required during T4 infection. Finally, Sun et al. (35) have produced a set of single-ring GroEL mutants that support the growth of bacteriophage T4. These mutants form a chaperonin containing only a single cavity, hence, trans-folding is not possible, and gp23 must be folded in cis.

Interestingly, mutant bacteriophages T4 (called T4bypass31) that propagate in the presence of gp31 amber mutations contain mutations in the gene encoding gp23 (18, 36), suggesting that the mutated capsid protein may fold independently of the gp31 cochaperonin. Alternatively, such mutant capsid proteins could fold without the help of the chaperonin system altogether or use GroES in combination with GroEL, either by a classical cis-folding or a trans-folding mechanism (23, 24). It will be interesting to determine the folding pathway of these mutant capsid proteins. We now have all of the tools available.


We thank E. van Duijn for electrospray ionization-MS analysis, E. H. Kroezinga for excellent experimental help, Dr. H. A. Raué for critical reading of the manuscript, Dr. L. W. Black for pET2331, Dr. A. Richardson (University of Geneva, Geneva) for pAR1 (gp31 expression) and Dr. F.-U. Hartl for pSR1.


Author contributions: H.v.H. and S.M.v.d.V. designed research, P.J.B. and B.W.F. performed research; P.J.B., H.v.H., and S.M.v.d.V. analyzed data; and P.J.B., H.v.H., and S.M.v.d.V. wrote the paper.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: CBB, Coomassie brilliant blue; Rubisco, ribulose bisphosphate carboxylase oxygenase; SR1, single ring 1.


1. Hartl, F. U. & Hayer-Hartl, M. (2002) Science 295, 1852-1858. [PubMed]
2. Sigler, P. B., Xu, Z., Rye, H. S., Burston, S. G., Fenton, W. A. & Horwich, A. L. (1998) Annu. Rev. Biochem. 67, 581-608. [PubMed]
3. Bukau, B. & Horwich, A. L. (1998) Cell 92, 351-366. [PubMed]
4. Braig, K., Otwinowski, Z., Hegde, R., Boisvert, D. C., Joachimiak, A., Horwich, A. L. & Sigler, P. B. (1994) Nature 371, 614-619.
5. Hunt, J. F., Weaver, A. J., Landry, S. J., Gierasch, L. & Deisenhofer, J. (1996) J. Biol. Chem. 271, 238-243. [PubMed]
6. Lorimer, G. (1997) Nature 388, 720-721. [PubMed]
7. Rye, H. S., Roseman, A. M., Chen, S., Furtak, K., Fenton, W. A., Saibil, H. R. & Horwich, A. L. (1999) Cell 97, 325-338. [PubMed]
8. Xu, Z., Horwich, A. L. & Sigler, P. B. (1997) Nature 388, 741-750. [PubMed]
9. Hemmingsen, S. M., Woolford, C., van der Vies, S. M., Tilly, K., Dennis, D. T., Georgopoulos, C. P., Hendrix, R. W. & Ellis, R. J. (1988) Nature 333, 330-334. [PubMed]
10. Coyle, J. E., Texter, F. L., Ashcroft, A. E., Masselos, D., Robinson, C. V. & Radford, S. E. (1999) Nat. Struct. Biol. 6, 683-690. [PubMed]
11. Brinker, A., Pfeifer, G., Kerner, M. J., Naylor, D. J., Hartl, F. U. & Hayer-Hartl, M. (2001) Cell 107, 223-233. [PubMed]
12. Shtilerman, M., Lorimer, G. H. & Englander, S. W. (1999) Science 284, 822-825. [PMC free article] [PubMed]
13. Ellis, R. J. & Hartl, F. U. (1996) FASEB J. 10, 5-9. [PubMed]
14. Park, E. S., Fenton, W. A. & Horwich, A. L. (2005) FEBS Lett. 579, 1183-1186. [PubMed]
15. Houry, W. A., Frishman, D., Eckerskorn, C., Lottspeich, F. & Hartl, F. U. (1999) Nature 402, 147-154. [PubMed]
16. Andreadis, J. D. & Black, L. W. (1998) J. Biol. Chem. 273, 34075-34086. [PubMed]
17. Laemmli, U. K., Beguin, F. & Gujer-Kellenberger, G. (1970) J. Mol. Biol. 47, 69-85. [PubMed]
18. Doermann, A. H. & Simon, L. D. (1984) J. Virol. 51, 315-320. [PMC free article] [PubMed]
19. van der Vies, S. M., Gatenby, A. A. & Georgopoulos, C. (1994) Nature 368, 654-656. [PubMed]
20. Keppel, F., Rychner, M. & Georgopoulos, C. (2002) EMBO Rep. 3, 893-898. [PMC free article] [PubMed]
21. Richardson, A., Schwager, F., Landry, S. J. & Georgopoulos, C. (2001) J. Biol. Chem. 276, 4981-4987. [PubMed]
22. Hunt, J. F., van der Vies, S. M., Henry, L. & Deisenhofer, J. (1997) Cell 90, 361-371. [PubMed]
23. Chaudhuri, T. K., Farr, G. W., Fenton, W. A., Rospert, S. & Horwich, A. L. (2001) Cell 107, 235-246. [PubMed]
24. Farr, G. W., Fenton, W. A., Chaudhuri, T. K., Clare, D. K., Saibil, H. R. & Horwich, A. L. (2003) EMBO J. 22, 3220-3230. [PMC free article] [PubMed]
25. Quaite-Randall, E. & Joachimiak, A. (2000) Methods Mol. Biol. 140, 29-39. [PubMed]
26. Quaite-Randall, E. & Joachimiak, A. (2000) Methods Mol. Biol. 140, 41-49. [PubMed]
27. Voziyan, P. A., Jadhav, L. & Fisher, M. T. (2000) J. Pharmacol. Sci. 89, 1036-1045. [PubMed]
28. Horwich, A. L., Burston, S. G., Rye, H. S., Weissman, J. S. & Fenton, W. A. (1998) Methods Enzymol. 290, 141-146. [PubMed]
29. van der Vies, S. M. (2000) in Methods in Molecular Biology: Chaperonin Protocols, ed. Schneider, S. (Humana, Totowa, NJ), Vol. 140, pp. 51-61. [PubMed]
30. Pierce, J. & Gutteridge, S. (1985) Appl. Environ. Microbiol. 49, 1094-1100. [PMC free article] [PubMed]
31. van Duijn, E., Bakkes, P. J., Heeren, R. M. A., van den Heuvel, R. H. H., van Heerikhuizen, H., van der Vies, S. M. & Heck, A. J. R. (2005) Nat. Methods, 2, 371-376. [PubMed]
32. Weissman, J. S., Hohl, C. M., Kovalenko, O., Kashi, Y., Chen, S., Braig, K., Saibil, H. R., Fenton, W. A. & Horwich, A. L. (1995) Cell 83, 577-587. [PubMed]
33. Langer, T., Pfeifer, G., Martin, J., Baumeister, W. & Hartl, F. U. (1992) EMBO J. 11, 4757-4765. [PMC free article] [PubMed]
34. Wang, J. D., Herman, C., Tipton, K. A., Gross, C. A. & Weissman, J. S. (2002) Cell 111, 1027-1039. [PubMed]
35. Sun, Z., Scott, D. J. & Lund, P. A. (2003) J. Mol. Biol. 332, 715-728. [PubMed]
36. Simon, L. D. & Randolph, B. (1984) J. Virol. 51, 321-328. [PMC free article] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...