The Importance of a Mobile Loop in Regulating Chaperonin/ Co-chaperonin Interaction

Chaperonins are universally conserved proteins that nonspecifically facilitate the folding of a wide spectrum of proteins. While bacterial GroEL is functionally promiscuous with various co-chaperonin partners, its human homologue, Hsp60 functions specifically with its co-chaperonin partner, Hsp10, and not with other co-chaperonins, such as the bacterial GroES or bacteriophage T4-encoded Gp31. Co-chaperonin interaction with chaperonin is mediated by the co-chaperonin mobile loop that folds into a β-hairpin conformation upon binding to the chaperonin. A delicate balance of flexibility and conformational preferences of the mobile loop determines co-chaperonin affinity for chaperonin. Here, we show that the ability of Hsp10, but not GroES, to interact specifically with Hsp60 lies within the mobile loop sequence. Using mutational analysis, we show that three substitutions in the GroES mobile loop are necessary and sufficient to acquire Hsp10-like specificity. Two of these substitutions are predicted to preorganize the β-hairpin turn and one to increase the hydrophobicity of the GroEL-binding site. Together, they result in a GroES that binds chaperonins with higher affinity. It seems likely that the single ring mitochondrial Hsp60 exhibits intrinsically lower affinity for the co-chaperonin that can be compensated for by a higher affinity mobile loop.

The GroEL (Hsp60; Cpn60) and GroES (Hsp10; Cpn10) families of molecular chaperones play an essential role in mediating folding of various substrates (1)(2)(3). A great amount of data exists on the functional and structural properties of Escherichia coli's GroEL and GroES and their homologues. Electron microscopy and x-ray crystallography revealed that GroEL is a tetradecamer composed of two rings, stacked back-to-back (4 -7). The smaller member, GroES, functions as a heptamer and binds to either or both ends of GroEL depending on the nucleotide distribution (8,9). GroEL displays a weak ATPase activity that is further reduced in the presence of co-chaperonin (10,11). GroEL captures its substrates via its apical domain, which presents a hydrophobic surface (12). GroES binding to GroEL through its seven mobile loops locks GroEL into a new conformation, in which most of the previously exposed hydrophobic residues of the apical domain now are involved in intersubunit contacts, leaving the interior wall of GroEL lined with hydrophilic residues (13). The result is the release of the substrate into the GroEL-GroES cavity, also referred to as the Anfinsen cage (14). Thus, an ideal folding environment has been created for the substrate, which is given ϳ10 s to attain a foldingcompetent conformation (3). The timing is dependent on the rate of ATP hydrolysis (15). GroES discharge is effected by binding of substrate and ATP to the opposite (trans) ring of GroEL (16 -18). Substrate is released in a folded form or in a form that is competent to fold, or, if it is not folded, it rebinds to GroEL, at which point the cycle is repeated (19 -21).
Although the above mechanism has been established in large part by using the E. coli chaperonin system, it probably applies to most members of the class I chaperonin family (22). However, one member, the human mitochondrial Cpn60 (Hsp60), differs in its quaternary structure from the bacterial Cpn60. GroEL oligomerizes as a tetradecamer, whereas human Hsp60 has been characterized as a heptamer (23). Such differences in structure have mechanistic implications. GroEL mutants that form only a single ring bind substrate and GroES irreversibly (24), presumably because the driving force for substrate discharge originates from ATP and/or substrate binding to the (missing) trans ring. If mitochondrial Hsp60 only forms one ring, then what drives substrate and Cpn10 discharge? This problem was addressed by Nielsen and Cowan (28), who showed that the mammalian mitochondrial pair functions by a mechanism that differs at the level of co-chaperonin and substrate release. While GroEL/ADP/GroES has a dissociation constant (K d ) in the nanomolar range, Hsp60/ADP/Hsp10 is such a weak complex that its K d could not even be measured.
Still a mystery at the co-chaperonin level is why GroES is excluded from a productive interaction with Hsp60 while both Hsp10 and GroES function with GroEL. It has previously been proposed that a mitochondrial N-terminal acetylation may be necessary to allow interaction with Hsp60 (25). This hypothesis was disproved when recombinant mouse Hsp10 was expressed in bacteria, where N-terminal acetylation does not occur, and the resulting purified protein was shown to function in conjunction with Hsp60 as well as GroEL (26). Another conjecture was that the differences of pI could be important in determining functional specificity, since bacterial GroES is typically acidic (pI 5.0), while Hsp10 is basic, with a pI close to 9.0 (27). Since the mobile loop of the co-chaperonin mediates interaction with the chaperonin, we hypothesized that the specificity may lie in a few amino acid differences between the GroES and Hsp10 mobile loops. Conversely, Nielsen and Cowan (28) demonstrated that, by swapping the co-chaperonin-binding apical domains between GroEL and Hsp60, the chimeric equatorial-Hsp60-apical-GroEL protein acquired the ability to interact with GroES.
We have extensively studied the structure and function of the mobile loops from GroES, Gp31, and Hsp10 co-chaperonins (29 -32). The mobile loop is an unstructured, ϳ20-residue-long segment that mediates Cpn10/Cpn60 interaction. Binding captures the mobile loop in a ␤-hairpin conformation, which presents three highly conserved hydrophobic residues at the binding interface. Both flexibility and the GroEL-bound ␤-hairpin structure are conserved among co-chaperonins. Mutations encoding substitutions throughout the GroES and Hsp10 mobile loops have been identified that alter affinity by affecting the flexibility of the mobile loop (33)(34)(35)(36)(37). Mutants that allow better binding are probably sequence alterations that disfavor mobility, while mutants that weaken binding either increase the entropy, and therefore flexibility of the mobile loop, or promote an incorrect mobile loop conformation (32).
Here, we show that the essential Cpn60 binding element lies within the mobile loop sequence of the Hsp10 co-chaperonin. Our studies lead us to conclude that these sequence variations dictate the affinity differences between GroES and Hsp10, and thus it is affinity rather than a distinct conformation that confers Hsp10's preference for Hsp60.

EXPERIMENTAL PROCEDURES
Bacteria and Bacteriophage-E. coli groEL(E191G) (originally named groEL44), is isogenic to E. coli B178 (a galE derivative of W3110), which is sup ϩ , i.e. nonpermissive for bacteriophage T4 amber (am) mutants (38,39). For bacteriophage T4-GT7 transduction experiments, B178(), which carries a prophage, thus making it less susceptible to T4-GT7 infection, was used as the recipient strain. DH5␣ supE and CJ236 were used for cloning and site-directed mutagenesis purposes (40,41 Subcloning of the Human Chaperonin Genes into the pBAD System-pBADhsp10 was created by subcloning the hsp10 cDNA from pJG-10 (31) into the pBAD vector either alone or with the human chaperonin gene, hsp60, directly downstream. To subclone in the vector alone, hsp10 was PCR-amplified using a primer for the 5Ј-end that contains an NdeI site and a primer for the 3Ј-end that contains the recognition site for EcoRI. For cloning with hsp60 downstream, hsp10 was amplified by PCR using the same 5Ј primer but a different 3Ј primer containing the MscI recognition sequence. Plasmid pBADgroEShsp60 was constructed by subcloning the hsp60 gene from pETcpn60 by cutting with NcoI, blunt-ending, and then cutting with HindIII. pETcpn60 was a kind gift from Dr. P. Viitanen (Dupont). The vector pBADgroES has EcoRI and HindIII sites at the gene's 3Ј-end. pBADgroES was first cut with EcoRI, blunt-ended, and then digested with HindIII. All constructs were tested for protein overexpression, following transformation into MC1009 bacteria.
Transduction and Complementation Experiments-T4-GT7 transduction experiments were performed as described in Ref. 43 except that in the donor strain, OF3465, a kind gift from Dr. O. Fayet (CNRS, Toulouse), the groESgroEL operon is deleted and replaced by an ⍀-chloramphenicol resistance (Cam R ) cassette. Transductants were selected on medium containing 50 g/ml ampicillin or 20 g/ml kanamy-cin, 12 g/ml tetracycline, and 0.02% L-arabinose and then scored for Cam R .
To test the growth of B178()⌬groESgroEL:Cam R ϫ pBAD groES(hsp10ml)groEL versus ϫ pBAD groES(hsp10ml)hsp60 (Fig. 4), single colony isolates were resuspended in LB medium, and 10-fold serial dilutions were spot-tested on LB plates supplemented with 50 g/ml ampicillin, 15 g/ml chloramphenicol, and 0.1% arabinose and incubated for 24 -36 h at the temperatures indicated in Fig. 4. Bacteriophage growth was tested by spot testing aliquots of serial dilutions of various bacteriophage on LB plates containing 50 g/ml ampicillin, 15 g/ml chloramphenicol, and 0.1% arabinose and seeded with 0.2 ml of culture grown for 24 h in LB medium containing 50 g/ml ampicillin, 15 g/ml chloramphenicol, and 0.02% arabinose. Plates were incubated for 20 h at 37°C. Complementation experiments were performed as described previously (44).
Protein Expression and Purification-Human Hsp60 was overexpressed by induction with 1 mM isopropyl-1-thio-␤-D-galactopyranoside from the pET vectors transformed in BL21. Transformants were grown to midlog phase and induced for 3 h before harvesting. Hsp60 was purified exactly as described in Ref. 45.
GroES and GroES(Hsp10ml) were overexpressed in LMG-190 cells, following transformation by plasmid pgroES and pgroES(hsp10ml), respectively. Cells were grown to midlog phase at 37°C and induced with the addition of 0.05% arabinose. After 4 -5 h of growth, cells were harvested. The purification protocol used is essentially a procedure described by Richardson et al. (32) with the following exceptions. For acid precipitation, following the DEAE-Sepharose column, fractions containing GroES were precipitated by ammonium sulfate and resuspended in buffer containing 50 mM sodium succinate (pH 4.6). Following overnight dialysis against the low pH buffer at 4°C, the precipitated proteins were removed by centrifugation, and the GroES-containing supernatant was ammonium sulfate-precipitated. Essentially pure GroES and GroES(Hsp10ml) protein was obtained following fractionation by HPLC Superdex G200 gel filtration.
Hsp10 was purified as described by Landry et al. (31). All proteins were stored at Ϫ80°C in a buffer containing 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 15% (v/v) glycerol. Protein concentrations were measured by either absorption at 280 nm using molar extinction coefficients determined by quantitative amino acid analysis or by the Bradford protein assay method, standardized with known concentrations of either GroEL, GroES, or Gp31.
Citrate Synthase Refolding-The chaperonin-dependent renaturation of pig heart citrate synthase (referred to here as citrate synthase) was performed as described previously (32,48). The following protein concentrations (given for monomers) were used: 4.2 M GroEL or Hsp60 and a 4.2 M concentration (or a fraction of the value as indicated in Fig.  3A) of co-chaperonins, and 0.2 M citrate synthase. Citrate synthase at 33 M was denatured for 30 min at 25°C in a solution containing 6 M guanidine hydrochloride, 3 mM dithiothreitol, and 2 mM EDTA. The refolding buffer contained 10 mM MgCl 2 , 2 mM ATP, 1 mM oxaloacetic acid and 20 mM potassium phosphate, pH 7.4. The co-chaperonin was added last to the refolding reaction, and its addition indicated the starting time of the reaction. The refolding reaction was performed at 30°C and in a total volume of 400 l, and citrate synthase activity was measured after 60 min by measuring the disappearance of acetyl-CoA at 323 nm. For the time-based measurements, aliquots were taken exactly at times following the addition of the co-chaperonin to the refolding mix as indicated in Fig. 3B.

GroES Does Not Assist Hsp60 in Refolding Citrate
Synthase-We have reconfirmed the previously published finding that human mitochondrial Hsp60 in combination with E. coli GroES is unable to assist substrate refolding (23). Chemically denatured pig heart citrate synthase was diluted into renaturation buffer supplemented with ATP and GroELor Hsp60, and either no co-chaperonin, GroES, Hsp10, or Gp31 was added (Fig. 1). Citrate synthase is a substrate that requires the assistance of chaperonin, co-chaperonin, and Mg-ATP (48,49). We found that GroEL with any of the three co-chaperonins, GroES, human Hsp10, or bacteriophage T4 Gp31, assists citrate synthase refolding. In contrast, Hsp60 helps refold citrate synthase only when partnered with Hsp10. Neither GroES nor Gp31 increased the levels of refolded citrate synthase significantly above those of Hsp60 alone, which, in fact, inhibits citrate synthase refolding. Not surprisingly, Gp31 has closer functional similarity to GroES than Hsp10 because it interacts with the same GroEL partner in vivo as GroES. However, GroES and Gp31 share much less sequence similarity (ϳ14%) than GroES and Hsp10 (ϳ44%; Ref. 50; Fig. 2A).
GroES Carrying the Mobile Loop of Hsp10 Assists Hsp60 in Refolding Citrate Synthase-Since it has been shown that GroES does not assist Hsp60 in refolding substrates because GroES does not bind to Hsp60, then it is likely that elements responsible for the defective partnership lie within the binding interface of the chaperone proteins (23). If the mobile loop contains the essential information for binding to Cpn60, then providing the correct mobile loop sequence capable of recognizing Hsp60 should restore interaction. We constructed the following two chimeric proteins: GroES containing the Hsp10 mobile loop, referred to as GroES(Hsp10ml), and Hsp10 with the GroES mobile loop, referred to as Hsp10(ESml) (Fig. 2C). These proteins were generated from plasmid gene constructs made by site-directed mutagenesis (for details, see "Experimental Procedures").
We first characterized GroES(Hsp10ml) by purifying the pro-tein and testing for its ability to assist GroEL or Hsp60 in the refolding of citrate synthase. The purification properties of GroES(Hsp10ml) were very similar to those of wild type GroES. The hybrid protein was functional with GroEL to the same extent as GroES, Gp31, and Hsp10, yielding approximately an 80% recovery of native citrate synthase activity. GroES(Hsp10ml) was completely indistinguishable from Hsp10 in its ability to help Hsp60 in the citrate synthase refolding assays (Fig. 3). Using different ratios of co-chaperonin to chaperonin as well as measuring the rates of citrate synthase refolding, no differences between Hsp10 and GroES (Hsp10ml) could be detected. These results clearly indicate that all of the necessary information for co-chaperonin interaction with mitochondrial Hsp60 lies within the mobile loop sequence. GroES(Hsp10ml) and Human Hsp60 Can Substitute for GroEL/GroES in E. coli Growth-Our subsequent approach was to test whether combinations between human and bacterial chaperonins can substitute for the bacterial GroEL machinery for E. coli and/or bacteriophage growth. We used a genetic system to knock out the chromosomal groESgroEL operon while maintaining viability by providing the cells with a plasmid carrying the desired chaperonin genes. OFB3465 is a strain that contains a chloramphenicol resistance-encoding cassette (Cam R ) in the place of the groESgroEL operon (51). Viability of the OFB3465 is maintained by a plasmid-encoded groESgroEL operon. In addition, a nearby Tn10 transposon insertion (encoding tetracycline resistance (Tet R )) is 90% cotransducible with the Cam R -encoding cassette. We grew a bacteriophage T4-GT7 lysate on this strain and used it to transduce the B178() recipient strain, transformed with various chaperonin plasmid constructs. We first selected for the inheritance of the Tet R marker. The number of Tet R transductants is indicative of the transduction efficiency, since inheritance of the Tn10 transposon alone does not produce a defective phenotype. We then scored for coinheritance of the Cam R cassette. Inheritance of the Cam R marker is only possible if the genes provided in trans can completely substitute for groESgroEL in bacterial viability. A 90% coinheritance is expected when the plasmid-encoded genes are able to functionally substitute for the chromosomal groESgroEL genes, as was observed when the recipient strain carried the pBADgroESgroEL plasmid (Table  I).
In the absence of any chaperonin genes, such as the pBAD vector alone, no Cam R -resistant colonies were recovered. We tested the ability of GroES(Hsp10ml) to substitute for GroES for E. coli growth. The hybrid co-chaperonin, when coexpressed along with E. coli GroEL from plasmid pBADgroES (hsp10ml)groEL, supports cell viability in the presence of chloramphenicol; i.e. recipient cells bearing either plasmid exhibit a T4-GT7 cotransduction frequency of 90% (Table I).
While these experiments were in progress, it was shown that human Hsp60 can substitute for GroEL in E. coli (52). However, Hsp60 requires the presence of Hsp10 and is unable to function in vivo with GroES. Using the T4-GT7 transduction system, we were able to delete the groESgroEL operon in the presence of plasmids pBADhsp10hsp60 and pBADgroES (hsp10ml)hsp60, but not in the presence of pBADgroEShsp60 (Table I). Contrary to this, we did not recover any Cam R transductants from strains transformed with plasmid pBAD hsp10(ESml)hsp60, providing further proof that the Hsp10 mobile loop sequence carries the essential elements that dictate Hsp60 chaperonin binding specificity.
How efficiently do GroES(Hsp10ml) and Hsp60 or GroES (Hsp10ml) and GroEL function in E. coli? We compared the colony-forming units of cells bearing the deletion of the endogenous groESgroEL operon and carrying either pBAD groES(hsp10ml)groEL or pBADgroES(hsp10ml)hsp60. Cells carrying plasmid-encoded GroEL grow better at higher temperatures, compared with cells carrying plasmid-encoded Hsp60 (Fig. 4). We also examined the ability of both strains to support bacteriophage growth. We found that only bacteriophage T5 efficiently formed plaques when GroEL was replaced by Hsp60. Bacteriophage exhibited a reduced efficiency of plaque formation on a strain expressing Hsp60, and bacteriophages T4 and its distant relative RB49 were unable to form plaques on this strain because Gp31 does not interact with Hsp60, in agreement with previously published results (52).
GroES(Hsp10ml) Possesses a Higher Affinity for GroEL (E191G)-We hypothesized that sequence alterations in the mobile loop affect co-chaperonin interaction with Cpn60 function by changing the structure and dynamics of the mobile loop. Using NMR techniques, Landry et al. (54) showed that the mobile loop of Hsp10 is less dynamic than that of GroES. Thus, it may exhibit a greater affinity than GroES for GroEL. Likewise, GroES(Hsp10ml) should exhibit a higher affinity than GroES for GroEL.
To test this hypothesis, we transformed a low affinity GroEL mutant strain, groEL(E191G), with plasmids expressing either GroES or GroES(Hsp10ml) (32). Bacteriophage does not form plaques on groEL(E191G) (38). Overexpression of GroES did not overcome the block on bacteriophage ; however, overexpression of Hsp10 or GroES(Hsp10ml) in groEL(E191G) allowed growth of bacteriophage (Table II). This result suggests that Hsp10 and GroES(Hsp10ml) restore interaction with the low affinity GroEL(E191G) enough to assist bacteriophage maturation.
A GroES Mutant Protein Carrying Three Amino Acid Substitutions in Its Mobile Loop Can Functionally Interact with Hsp60 -If a high affinity co-chaperonin is sufficient to create functional interaction with Hsp60, then key residues in the Hsp10 mobile loop might be identified by potentially large individual contributions to high affinity. The amino acid sequences of the GroES and Hsp10 mobile loops are similar, albeit with a few significant differences (Fig. 2B). These sequence differences affect two aspects of the mobile loop that regulate its affinity for Cpn60. First, the sequence of the hydrophobic tripeptide that makes direct contact with Cpn60 should be more hydrophobic in Hsp10 than GroES. Hsp10's middle tripeptide residue, methionine, is more hydrophobic than valine (53) found at the equivalent position in GroES. Second, the amino acids at two positions in Hsp10, which regulate the balance between mobile loop flexibility and a preference toward the chaperonin-bound conformation, favor the chaperonin-bound conformation more than the residues in the GroES sequence that occupy the equivalent positions. A distinguishing feature is the presence of a highly conserved proline residue following the hydrophobic tripeptide in mitochondrial mobile loops. A Pro residue at this position should reduce the conformational dynamics of the mobile loop, thus making the Hsp10 mobile loop less flexible than its bacterial counterpart. Indeed, it is known that mutations at this position, P33S or P33H, reduce the affinity of yeast Hsp10 for GroEL (33,34). Most likely, substitution at this site results in increased conformational dynamics at position 33, which in turn increases the entropic cost of binding to GroEL (31,54). Finally, the nature of the residue at position 21 in GroES and the corresponding position 26 in Hsp10 is important in determining ␤-sheet propensity. Threonine, as found in the Hsp10 sequence, exhibits a higher ␤-sheet propensity than serine (55) at the equivalent site in GroES (32). The three features described above, differentiating the Hsp10 and GroES mobile loops, support the idea that Hsp10 has evolved to possess a higher affin- ity mobile loop.
To test whether these three mobile loop features are necessary and sufficient to specify Hsp60 interaction, we created the following GroES mutants: GroES(S21T,V26M,T28P)HisC, GroES(S21T,T28P)HisC, GroES(V26M)HisC, and GroES(T28P) HisC. The proteins were expressed with C-terminal histidine (His tags) and purified from strains expressing Gp31 instead of wild type GroES. Because Gp31 and GroES do not form mixed oligomers, we were able to obtain pure, homogenous His-tagged mutant proteins. As controls, we also purified wild type GroES and GroES(Hsp10ml), each with C-terminal His-tags. All six proteins were tested for their ability to assist either GroEL or Hsp60 in refolding citrate synthase. All co-chaperonins when paired with GroEL helped recover ϳ80% of the denatured citrate synthase (Fig. 5). As already shown with the non-Histagged proteins, GroES (with a His tag) and Hsp60 provided no assistance for renaturation of citrate synthase, while the use of the His-tagged version of GroES(Hsp10ml) and Hsp60 resulted in ϳ80% renaturation of citrate synthase. Neither of the single point mutants, GroES(V26M) or GroES(T28P), were functional with Hsp60 in the refolding assay. Interestingly, the double mutant, GroES(S21T,T28P), helped recover partial citrate synthase activity, while the triple mutant, GroES (S21T,V26M,T28P), was almost as active as Hsp10 and  Hsp10ml)) is completely indistinguishable from Hsp10 in its ability to help Hsp60 in the refolding assays. A, comparison of Hsp10 and GroES(Hsp10ml) in combination with Hsp60 by testing for citrate synthase refolding activity by using different ratios of co-chaperonin to chaperonin. B, comparison of Hsp10 and GroES(Hsp10ml) in combination with Hsp60 by testing for citrate synthase refolding activity by measuring the rates of refolding.

TABLE I Deletion of the groESgroEL operon by T4-GT7-mediated transduction
in the presence of different combinations of human and bacterial chaperonins Transductants were first selected on LB-agar plates containing 12 g/ml tetracycline, following 2-3 days of incubation at 30°C. Subsequently, individual transductants were tested for Cam R by streaking onto LB-agar plates supplemented with 15 g/ml chloramphenicol and incubated for 2 days.  GroES(Hsp10ml) in its partnership with Hsp60 in assisting citrate synthase refolding. Therefore, it is likely that all three mobile loop substitutions are necessary to bestow an Hsp10specific activity.
The Mutant GroES Protein Can Only Partially Support Bacteriophage Growth-Do the mobile loops of GroES (S21T,V26M,T28P), GroES(Hsp10ml), and Hsp10 differ from that of wild type GroES simply because they have higher affinity for chaperonin; or is the mechanism more complicated, involving both affinity differences and a more specific conformational adaptation necessary for interaction with Hsp60? As described above, the co-chaperonin affinity can be qualitatively measured using an in vivo complementation assay. The low affinity groEL(E191G) mutant strain was transformed with the following plasmids: pESHisC, pES(hsp10ml)HisC, pES(T28P) HisC, pES(V26M)HisC, pES(S21T,T28P)HisC, and pgroES (S21T,V26M,T28P)HisC, and protein expression was induced with isopropyl-1-thio-␤-D-galactopyranoside. The ability of bacteriophage to form plaques was measured at 37°C (Table II). Transformants expressing the single mutant GroES (T28P)HisC allowed formation of medium size plaques, although transformants expressing GroES(V26M)HisC or GroES(S21T, T28P)HisC behaved like wild type GroESHisC. GroES(S21T,V26M,T28P)HisC and GroES(T28P)HisC behaved equally in partially suppressing growth at higher temperatures, in contrast to their differing abilities to interact with Hsp60 in vitro. Furthermore, only GroES(Hsp10ml)HisC was able to efficiently support growth at all temperatures. Although this assay does not measure affinity directly, it demonstrates that the GroES triple mutant lacks a component necessary for completely restoring interaction with GroEL(E191G) at least at higher temperatures.
It is possible that this in vivo assay is far more sensitive than the in vitro citrate synthase refolding assay. Therefore, to discern subtle functional differences between the entire Hsp10 mobile loop and that of GroES with the S21T, V26M, and T28P changes, we created the construct pBADgroES(S21T,V26M, T28P)hsp60 and tested its ability to support E. coli growth (Table I). Compared with pBADhsp10hsp60 and pBADgroES (hsp10ml)hsp60, pBADgroES(S21T,V26M,T28P)hsp60 functions equally well for supporting E. coli growth in the absence of the endogenous groES and groEL genes. DISCUSSION Our understanding of the mechanism of chaperonin-assisted folding has greatly increased over recent years, yet there are many details that remain unresolved or under debate. For example, it is well accepted that the regulation of the timing of co-chaperonin binding to chaperonin depends on nucleotide hydrolysis (15). On the other hand, the co-chaperonin plays a critical role in controlling both the efficiency and specificity of chaperonin-assisted folding. Beyond playing a structural role in the chaperone machine, the co-chaperonin also acts as an allosteric modulator (56). We have previously proposed that the GroEL-binding co-chaperonin mobile loop has evolved a finely tuned balance between flexibility and conformational preference for a 3:5 ␤-hairpin loop that is optimized for continued cycles of binding and release (30,32). Here, we have shown that co-chaperonin affinity for GroEL is dictated by a balance between disorder and structure of the mobile loop, as well as the hydrophobicity of the conserved internal tripeptide, which constitutes the actual binding site.
Before dissecting the various features of the mobile loop, we first set out to identify the essential and specific co-chaperonin feature in Hsp10 that distinguishes it from GroES in its ability to interact with Hsp60. Because GroES does not bind to Hsp60, we hypothesized that it is the mobile loop that contains the essential information for binding to chaperonin. Therefore, providing GroES with the correct mobile loop sequence capable of recognizing Hsp60 should restore interaction. Indeed, the mobile loop of Hsp10 grafted into GroES, resulting in GroES(Hsp10ml), is sufficient to provide functional specificity. Furthermore, GroES(Hsp10ml) is able to function in vivo for growth of E. coli and bacteriophage T5 when partnered with Hsp60.
From our indirect affinity measurements of the mobile loops, we concluded that the Hsp10 mobile loop sequence restored interaction between GroES and Hsp60 because the Hsp10 mobile loop increases GroES affinity. This observation is in agreement with previously published NMR data showing that the mobile loop of Hsp10 is less flexible than that of GroES (54).
What are the specific features of the mobile loop sequence FIG. 6. GroES assumes Hsp10 specificity with three amino acid substitutions in the region of the mobile loop that forms a ␤ hairpin upon binding to GroEL. Each substitution in GroES(S21T,V26M,T28P) results in higher chaperonin affinity. Threonine has the highest propensity for ␤-sheet structure of any amino acid at a solvent-exposed site (green). NMR experiments suggest that a proline at this site reduces the conformational flexibility of the mobile loop (red). The hydrophobic tripeptide makes direct contact with GroEL (purple). Met is more hydrophobic than Val. The illustrated structure corresponds to the GroEL-bound mobile loop peptide conformation (PDB:1EGS) determined by trNOE NMR spectroscopy (30), except the N-terminal NH 2 is replaced with threonine. that determine co-chaperonin affinity? We approached this question by mutational analysis and found that three substitutions in the GroES mobile loop are necessary and sufficient to acquire Hsp10-like specificity, as summarized in Fig. 6. These three residues most likely provide a combination of increased hydrophobicity and reduced flexibility in the Hsp10 mobile loop, explaining why Hsp10 exhibits higher chaperonin affinity than GroES. Since the single ring Hsp60 obviously cannot utilize binding of GroES and ATP to the trans ring as a driving force for the discharge of cis GroES, the single ring chaperonin may have evolved an intrinsically lower affinity for co-chaperonin. The high affinity mobile loop of Hsp10 may compensate for an intrinsically low co-chaperonin affinity in single ring Hsp60 (28).
We propose that the difference in affinity between Hsp60 and GroEL is achieved by altered domain-domain interactions within Hsp60, very likely analogous to the substitution that reduces co-chaperonin-affinity in GroEL(E191G) (32). Thus, the actual co-chaperonin-binding sites of GroEL and Hsp60 could be nearly identical, which would explain the simultaneous increase in affinity for both Hsp60 and GroEL(E191G) obtained by the S21T, V26M, and T28P substitutions in GroES. Nevertheless, Hsp60 function may require tighter binding than would be obtained with a GroES-like mobile loop; thus, the Hsp10 mobile loop has acquired a higher affinity for chaperonins through the incorporation of these residues in the native sequence.
Increased demands on the chaperonin machine imposed by bacteriophage have helped to reveal the importance of a self-consistent mobile loop structure. The GroES(Hsp10ml)-GroEL(E191G) combination can support growth, but the GroES(S21T,V26M,T28P)-GroEL(E191G) combination is less efficient. Perhaps the chaperonin affinity of the GroES triple mutant is still not quite correct. Previously, we have shown that chaperonin affinity must be delicately balanced to support growth of bacteriophage T4 (32); thus, this combination of substitutions may produce an affinity below or above the level required for optimal chaperonin function. In either case, other residues in the GroES or Hsp10 mobile loops could further modulate the affinity toward the appropriate level. Alternatively, appropriate chaperonin affinity is not sufficient for optimal function, and the internal workings of the mobile loop are under additional constraints, such as might be required for an important intermediate in the chaperonin reaction cycle. Such sophisticated constraints could explain the unique requirement for the GroES homolog, Gp31, in the assembly of bacteriophage T4 Gp23 into capsids (57). In contrast, chaperonin function for cellular growth either is less constrained or more easily compensated by the other chaperone systems.