Functional Characterization of an Archaeal GroEL/GroES Chaperonin System SIGNIFICANCE OF SUBSTRATE ENCAPSULATION*

In all three kingdoms of life chaperonins assist the folding of a range of newly synthesized proteins. As shown recently, Archaea of the genus Methanosarcina contain both group I (GroEL/GroES) and group II (ther-mosome) chaperonins in the cytosol. Here we report on a detailed functional analysis of the archaeal GroEL/ GroES system of Methanosarcina mazei ( Mm ) in comparison to its bacterial counterpart from Escherichia coli ( Ec ). We find that the groESgroEL operon of M. mazei is unable to functionally replace groESgroEL in E. coli . However, the Mm GroES protein can largely complement a mutant Ec GroES protein in vivo . The ATPase rate of Mm GroEL is very low and the dissociation of Mm GroES from Mm GroEL is 15 times slower than for the Ec GroEL/ GroES system. This slow ATPase cycle results in a prolonged enclosure time for model substrate proteins, such as rhodanese, in the Mm GroEL:GroES folding cage before their release into the medium. Interestingly, optimal functionality of Mm GroEL/GroES our group mechanism has from studies of the bacterial Ec GroEL/ A functional analysis of the archaeal Mm GroEL/GroES system comparison to its bacterial counterpart now offers the opportunity to define general features of group I function that have been conserved in evolution. Moreover, it may provide insight into how the archaeal Mm GroEL/GroES system functionally adapted because of co-evolution with the group II chaperonin. Here we show that Mm GroEL/GroES is unable to functionally replace Ec GroEL/ GroES in E. coli . However, the Mm GroES protein can largely complement a mutant Ec GroES protein in vivo . The ATPase rate of Mm GroEL is very low and the dissociation of Mm GroES from Mm GroEL is 15 times slower than for the Ec GroEL/ GroES system. Indeed, model substrates such as rhodanese accumulate in a folded state inside the Mm GroEL/GroES folding cage before they are released into the medium. Optimal functionality of Mm and substrate the of sulfate. Interestingly, in the absence proteins as malate dehydrogenase fail to be encapsulated by GroES and rather cycle on and off the GroEL trans ring in a non-productive reaction. These results indicate that the basic and suggest length reaction cycle growth Archaea. Additionally, the only folded with thermosome, which is not normally located within the same

A subset of newly synthesized proteins in the cytosol, as well as in mitochondria and chloroplasts, depend on chaperonins, a family of structurally related molecular chaperones, for folding assistance (1)(2)(3)(4)(5)(6)(7)(8)(9). In contrast to other types of molecular chaperones that act more generally in de novo folding, such as the Hsp70s, the chaperonins form large cylindrical double-ring structures in which a single molecule of unfolded protein is transiently enclosed and allowed to fold unimpaired by aggregation. The chaperonins have been divided into two distinct classes, group I and group II, with members of both groups exhibiting only limited sequence homology but an overall similar architecture of the oligomeric ring complexes (8 -13).
Group I chaperonins, also known as Hsp60s or Cpn60s, are generally found in the bacterial cytosol (e.g. GroEL in Escherichia coli) as well as in mitochondria (mtHsp60) and chloroplasts (Rubisco 1 subunit binding protein), and typically form homo-oligomeric complexes of two stacked heptameric rings. They cooperate functionally with cofactors of the Hsp10 (GroES) family, single heptameric rings of ϳ10-kDa subunits that bind to the ends of the Hsp60 cylinder. GroEL of E. coli and its cofactor GroES have been extensively analyzed in terms of their structure and function. GroEL is composed of 14 identical 57-kDa subunits. Each subunit consists of three domains: the equatorial domain contains the ATP binding site and mediates most intersubunit contacts within and between rings. It is connected via an intermediate hinge-like domain to the apical domain. The apical domains expose a number of hydrophobic residues toward the ring cavity for the binding of unfolded protein substrate and provide binding regions for flexible sequence loops on the subunits of GroES.
The ATP-dependent interactions of GroEL with protein substrate and GroES have been studied extensively (reviewed in Refs. 2-4, 6, and 7). Briefly, binding of 7 ATP and GroES to GroEL results in the displacement of GroEL-bound substrate protein into an enclosed folding cage, the so-called cis complex. Proteins up to ϳ60 kDa can become enclosed and fold in the confined environment of the cage in the time it takes the 7 ATP to be hydrolyzed to ADP (ϳ10 -15 s at 25°C). Upon completion of this first round of ATP hydrolysis, binding of 7 ATP to the opposite ring of GroEL (the trans ring) transmits an allosteric signal to the cis ring, causing the dissociation of the 7 ADP and GroES. At this point, folded protein leaves the cage, whereas incompletely folded intermediates are rapidly recaptured by the same or another GroEL complex for a subsequent folding cycle. Recently, it has been shown that GroEL can assist in the folding of certain proteins too large to be enclosed by GroES using a mechanism that involves cycling on and off the trans ring (14,15).
In contrast to the group I chaperonins, the group II chaperonins of the archaeal cytosol (known as thermosome) are composed of 1 to 3 types of different subunits and form double-ring structures with 8-or 9-fold symmetry (16). Significantly greater structural complexity is exhibited by the group II chaperonin of the eukaryotic cytosol (known as CCT for chaperonin containing TCP-1 or as TRiC for TCP-1 ring complex), which consists of eight orthologous subunits per ring. The group II chaperonins are generally independent of a Hsp10 (GroES)-like cofactor but cooperate with the molecular chaperone prefoldin (also called GimC) in Archaea and eukaryotes (17)(18)(19)(20)(21). Their mechanism of action involves protein encapsulation by helical extensions of the apical chaperonin domains that provide a "built in" GroES function (22).
It has generally been assumed that groups I and II chaperonins do not co-exist in the same cellular compartment. However, the recent genome sequencing of three mesophilic archaea, Methanosarcina barkeri (M. barkeri; ϳ2.8 Mbp, US DOE Joint Genome Institute), Methanosarcina acetivorans (M. acetivorans, ϳ5.8 Mbp) (23), and Methanosarcina mazei Gö1 (M. mazei, ϳ4.1 Mbp) (24) revealed the presence of both group I and group II chaperonin genes that are simultaneously expressed to similar levels in the cytosol (25). The M. mazei (Mm) and E. coli (Ec) GroEL/GroES systems exhibit 54% sequence identity and 72% similarity.
So far, almost all our knowledge of the group I chaperonin mechanism has come from studies of the bacterial EcGroEL/ GroES system. A functional analysis of the archaeal MmGroEL/GroES system in comparison to its bacterial counterpart now offers the opportunity to define general features of group I chaperonin function that have been conserved in evolution. Moreover, it may provide insight into how the archaeal MmGroEL/GroES system functionally adapted because of coevolution with the group II chaperonin. Here we show that MmGroEL/GroES is unable to functionally replace EcGroEL/ GroES in E. coli. However, the MmGroES protein can largely complement a mutant EcGroES protein in vivo. The ATPase rate of MmGroEL is very low and the dissociation of MmGroES from MmGroEL is 15 times slower than for the EcGroEL/ GroES system. Indeed, model substrates such as rhodanese accumulate in a folded state inside the MmGroEL/GroES folding cage before they are released into the medium. Optimal functionality of MmGroEL/GroES and substrate encapsulation requires the presence of ammonium sulfate. Interestingly, in the absence of ammonium sulfate, stringently GroEL-dependent proteins such as malate dehydrogenase fail to be encapsulated by GroES and rather cycle on and off the GroEL trans ring in a non-productive reaction. These results indicate that the archaeal GroEL/GroES system has preserved the basic encapsulation mechanism of bacterial GroEL and suggest that it has adjusted the length of its reaction cycle to the slower growth rates of Archaea. Additionally, the release of only folded protein from the GroEL/GroES cage would avoid non-productive interactions of the GroEL substrates with the thermosome, which is not normally located within the same compartment.
EXPERIMENTAL PROCEDURES Proteins E. coli GroEL and GroES and M. mazei GroEL and GroES were purified as described (25). EcGroES and MmGroES were cloned into pET-22b (Novagen) resulting in C-terminal His-tagged constructs, allowing immobilization on the NTA (nitrilotriacetate) biosensor chip. The His-tagged GroES proteins were overexpressed by induction in BL21(DE3) with 1 mM isopropyl-␤-D-galactopyranoside. Supernatants were applied on Ni 2ϩ -NTA (Qiagen) and the GroES fractions were further purified by chromatography on

Bacterial and Bacteriophage Genetic Manipulations
The parental E. coli strain used in this study is B178, a W3110 galE sup ϩ derivative. B178 and MC4100⌬ara714 have been previously described (26 -28). The mutants groEL44 and groES619 and all the bacteriophages used in this study have also been described (26,27,29,30). The EcgroES and EcgroESgroEL genes were cloned under the arabinose-inducible promoter of pBAD22 (31), using standard molecular cloning techniques (a kind gift of Dr. France Keppel). The MmgroES-groEL operon was amplified from a genomic library of M. mazei Gö1 (24) by PCR using primers described in Ref. 25 and inserted into pET22b using NdeI and BamHI restriction sites. The genes MmgroES, MmgroEL, and MmgroESgroEL together with the ribosomal binding site of the corresponding pET22b (Novagen) plasmids (25) were excised with XbaI and HindIII and inserted into the arabinose-inducible promoter of pBAD18 (31).
Classical bacteriophage P1-mediated transduction was used to determine whether the MmgroESgroEL operon can replace that of E. coli. A P1 lysate was grown on E. coli strain AR189 (27) whose chromosomally encoded groESgroEL operon has been deleted and substituted with a chloramphenicol resistance (Cam R ) encoding cassette. This Cam Rencoding cassette is ϳ50 -60% linked to a nearby Tn10 transposon, encoding resistance to tetracycline (Tet R ), i.e. when one selects first for the Tet R marker in the recipient strain, then ϳ50% of such transductants should simultaneously become Cam R , provided the essential EcgroESgroEL operon (32) is carried on a plasmid. Accordingly, the bacteriophage P1 lysate grown on strain AR189 was used to infect wild-type E. coli B178 strains carrying various plasmid constructs (Tables I and II). Tet R transductants were selected on LB-agar plates supplemented with 12.5 g/ml tetracycline and 5 mM sodium citrate (to prevent re-infection and killing by the P1 bacteriophage). Following incubation for 2 days at 30°C, Tet R colonies were tested for inheritance of the Cam R marker by cross-streaking the Tet R transductants on LB-agar plates supplemented with 5 mM sodium citrate and either 12.5 g/ml tetracycline or 10 g/ml chloramphenicol, followed by incubation overnight at 30°C.
Genetic complementation experiments were carried out as follows: isogenic E. coli wild-type MC4100 or groES619 or groEL44 bacterial strains were first transformed with a series of pBAD constructs created by standard PCR methods carrying various combinations of the E. coli (Ec) and M. mazei (Mm) genes by selecting for plasmid-encoded ampicillin resistance (Amp R ) at 30°C. Bacterial transformants were grown at 30°C overnight in LB broth supplemented with 100 g/ml ampicillin without arabinose. For complementation of the bacterial temperaturesensitive phenotype at 43°C, bacterial cultures were serially diluted 10-fold and 3-l aliquots spot-tested on LB-agar plates supplemented with 0.1% arabinose and 100 g/ml ampicillin and incubated overnight at 30 and 43°C for 16 h. For complementation of bacteriophage , T4 or T5 growth, 0.3-ml aliquots of overnight cultures in LB broth supplemented with 100 g/ml ampicillin were added to 3 ml of LB soft agar (0.66% agar) and evenly distributed over LB-agar plates containing 0.1% arabinose. When the soft agar solidified, 3-l aliquots of 10-fold dilutions of various bacteriophage lysates were spotted on the bacterial lawn and, following drying of the plates, incubated at 37°C for 16 h.

ATPase Assay
GroEL (1 M oligomer) was added to Buffer A (20 mM MOPS, pH 7.5, 100 mM KCl, 5 mM MgCl 2 ) or Buffer A, 0.5 M ammonium sulfate (AS) and incubated for 5 min at 37°C. Where indicated, GroES was present at a 2-fold molar excess over GroEL. The reaction was initiated at 37°C by the addition of 2 mM ATP. The kinetics of the ATPase activities of EcGroEL and MmGroEL were followed for 0-5 and 0 -25 min, taking time points every 1 and 5 min, respectively. ATPase activity at the various time points was stopped by the addition of ϳ15 mM CDTA. Quantification of liberated inorganic phosphate was performed by the malachite green assay (33) after incubation for 30 min at 25°C, and absorption was measured at 640 nm. A calibration curve of inorganic phosphate (0 -20 nmol) was measured in parallel. No spontaneous ATP hydrolysis was observed under the conditions used.

Refolding Assays
Rhodanese Refolding-Rhodanese was denatured by incubation for 30 min at 37°C in denaturing buffer (6 M guanidinium HCl, 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM MgCl 2 , and 5 mM dithiothreitol) and diluted 100-fold into Buffer A or Buffer A, 0.5 M AS at 37°C in the absence or presence of various concentrations of the chaperonins (as specified in the figure legends). At the times indicated, chaperonin action was stopped with CDTA (50 mM) and rhodanese activity measured at 460 nm at 25°C, as previously described (34,35).
Malate Dehydrogenase Refolding-MDH was denatured for 30 min in Buffer A containing 3 M guanidinium HCl, 5 mM dithiothreitol and diluted 100-fold into Buffer A or Buffer A, 0.5 M AS at 37°C in the absence or presence of various concentrations of the chaperonins (as specified in the figure legends). At the times indicated, aliquots were withdrawn and chaperonin-assisted refolding stopped by the addition of 50 mM CDTA. Enzyme activities were measured essentially as described (36,37) at 25°C in assay mixture (Buffer A, 1 mg/ml bovine serum albumin (Sigma), 0.22 mM ␤-NADH (Sigma), 0.55 mM oxaloacetate (Sigma), 1 mM CDTA), after 60 min incubation at 20°C. The time dependent oxidation of ␤-NADH was monitored for ϳ1 min at 340 nm.

Size Exclusion Chromatography
At 5 and 45 min into the chaperonin-assisted refolding assays of rhodanese or MDH, an aliquot of 50 l was withdrawn from the reaction and further ATP hydrolysis was inhibited by the addition of 25 mM glucose and 0.3 units/l hexokinase (Roche Diagnostics). The sample was then applied onto a Superose 6 3.2/30 (Amersham Biosciences) size exclusion column equilibrated in Buffer A. Fractions were analyzed for enzyme activity and/or resolved by 16% SDS-PAGE followed by Coomassie Blue staining and immunoblotting with anti-rhodanese or anti-MDH antibodies.

Protease Protection
MDH was denatured as described above and diluted 100-fold into Buffer A or Buffer A, 0.5 M AS in the presence of an equimolar concentration of GroEL chaperonin at 37°C. Protein aggregates were removed by centrifugation for 10 min at 10,000 ϫ g. The supernatant was divided into two equal portions, one portion receiving 4 mM AMP-PNP (or 2 mM ADP) and the other 4 mM AMP-PNP (or 2 mM ADP), 1 M GroES, followed by incubation for 5 min. Treatment with 1.5 g/ml proteinase K (from Tritirachium album, Roche) was followed for 0 -15 min at 25°C and the proteinase K action stopped with 1 mM phenylmethylsulfonyl fluoride. Half of each reaction was resolved by 8% SDS-PAGE followed by Coomassie Blue staining for GroEL, whereas the other half was resolved by 16% SDS-PAGE and transferred to nitrocellulose and immunoblotted with MDH antibodies.

Surface Plasmon Resonance
EcGroES and MmGroES, modified with a C-terminal His 6 tag, were immobilized (ϳ30 response units) on a chelating NTA biosensor chip (BIAcore Inc. (38)) using a BIAcore 2000 instrument at 37°C. Approximately 20 -40% of the immobilized GroES was competent in GroEL binding. Association and dissociation of the GroEL chaperonins in Buffer A, 2 mM ATP was performed for 480 s at a flow rate of 20 l/min, with an association phase for 8 min followed by dissociation for 15-20 min. Kinetic analyses with 25 to 500 nM of the GroEL chaperonins were performed essentially as described in Ref. 39 using BIAevaluation software 3.2.

Partial Exchangeability of Components between the Archaeal and Bacterial GroEL/GroES Systems in Vitro-
The monomeric model substrate rhodanese (ϳ35 kDa) has been shown to require the EcGroEL/GroES system and ATP for successful refolding from denaturant at concentrations where the unfolded protein tends to aggregate in the absence of chaperones. In a recent study we showed that the archaeal group I chaperonin of M. mazei, MmGroEL, is as efficient as EcGroEL in preventing the aggregation of denatured rhodanese (25). This effect was optimal at a nearly equimolar ratio of MmGroEL to rhodanese. Refolding of rhodanese by MmGroEL/MmGroES, as followed by the regain of enzymatic activity, occurred with somewhat slower kinetics but the same final yield as with EcGroEL/EcGroES and was dependent on the complete chaperonin system, that is, MmGroEL, MmGroES, and hydrolysable ATP (Ref. 25 and Fig. 1). MmGroES was found to functionally cooperate with EcGroEL, resulting in rhodanese refolding with similar overall kinetics as those observed with MmGroEL/MmGroES but with a ϳ25% lower yield (Fig. 1). Surprisingly, the combination of MmGroEL with EcGroES proved to be completely inactive in rhodanese refolding (Fig. 1). As will be shown below, this is because of the inability of EcGroES to efficiently bind MmGroEL.
The MmgroE Operon Is Unable to Functionally Replace the Endogenous groE Operon in E. coli-Given the high homology between the corresponding groES and groEL genes of E. coli and M. mazei (25), we asked whether these chaperonin systems can substitute for each other in vivo. We have previously shown that the EcgroESgroEL operon cannot be deleted under all conditions tested (see "Experimental Procedures" for detailed description of the methodology used). However, the chromosomally encoded EcgroESgroEL operon can be deleted provided a wild-type operon is present in trans (27,32,43). Table I shows that, as expected, the chromosomally encoded EcgroESgroEL operon can be deleted as long as the recipient strain carries the same operon on a plasmid construct. In this case, 56% of the  Tet R transductants simultaneously inherited the nearby Cam R -encoding cassette (Table I). In contrast, when the E. coli recipient strain carried a plasmid encoding the MmgroES-groEL operon, none (0/72) of the Tet R -encoding transductants simultaneously inherited the nearby Cam R -encoding allele. Because M. mazei is a strictly anaerobically growing organism, this same type of P1 transduction experiment was repeated under anaerobic growth conditions. Table I shows that, even under these conditions, the MmgroESgroEL operon failed to replace its EcgroESgroEL counterpart.
The MmgroES Gene Can Functionally Replace Its E. coliencoded Homologue-Following our failure to substitute the entire groESgroEL operon of E. coli by that of M. mazei, we asked whether the individual MmgroES or MmgroEL genes can functionally complement their counterparts in E. coli. This was done by transforming isogenic E. coli strains, either wildtype or carrying the groES619 or groEL44 mutations, with plasmids encoding various constructs of wild-type genes from either E. coli or M. mazei (Table II). E. coli carrying either the groES619 or groEL44 mutant alleles cannot propagate the and T5 bacteriophages or form colonies at 43°C (26,44). In addition, groEL44 mutant bacteria cannot propagate bacteriophage T4 at 37°C (Table II). The following observations were made: (a) the MmgroES wild-type gene complemented the groES619 defect in bacterial growth at the nonpermissive temperature of 43°C and the growth of bacteriophage T5 at 37°C. In contrast, MmgroES did not efficiently substitute for EcgroES in bacteriophage growth at 37°C. These results indicate that the MmGroES protein can indeed carry out most of the essential in vivo functions of EcGroES. Surprisingly, expression of the MmgroES gene in trans could also largely correct many of the defects exhibited by the groEL44 mutation, namely bacterial temperature sensitivity at 43°C and resistance to bacteriophages and T5 at 37°C. As expected, MmgroES did not correct bacteriophage T4 growth at 37°C because T4 needs a bacteriophage-encoded GroES-like cochaperone (Gp31) for the proper processing of its Gp23 capsid protein (43)(44)(45). (b) In contrast to the MmGroES results, the expression of the MmGroEL protein in trans did not significantly correct any of the defects exhibited by groEL44 mutant bacteria in terms of bacteriophage or bacterial growth (Table  II). In sharp contrast to MmGroES, expression of MmGroEL from a multicopy plasmid, either alone or in combination with MmGroES, interfered with E. coli growth at 43°C (Table II) and partially with bacteriophage T4 growth at 37°C. Bacteriophage growth was tested on LB-agar plates in the presence of 0.1% arabinose, followed by 16 h of incubation at 37°C. Bacterial growth was tested under the same conditions but at 43°C (see "Experimental Procedures" for details). ϩϩϩ, wild type size colonies or plaques, with an efficiency of plating of ϳ1.0; ϩϩ, good growth, but colonies or plaques smaller than wild type, but still with an efficiency of plating of ϳ1.0; ϩ, some growth, but colonies or plaques are very small, with an efficiency of plating of ϳ0.1⅐1.0; Ϫ, no detectable colony or plaque formation, with an efficiency of plating less than 0.0001.
Bacterial growth at 43°C with 0.1% arabinose T5 T4 In summary, these in vivo studies demonstrate that the MmGroES protein can complement in vivo for most of EcGroES functions. Furthermore, MmGroES must exhibit a higher overall affinity for EcGroEL because it efficiently suppressed the defects of the EcGroEL44 mutant protein, previously shown to exhibit diminished ability to bind to its cochaperone (44,45).
The Archaeal and Bacterial GroEL/GroES Systems Cycle with Different Kinetics-In vitro experiments were performed to explore the possible basis for the functional differences between the archaeal and bacterial chaperonin systems. One parameter that influences the production of refolded protein by chaperonin is the rate of GroES cycling on and off GroEL. Surface plasmon resonance (SPR) experiments were performed by immobilizing C-terminal His-tagged EcGroES or MmGroES on a NTA biosensor chip. Dissociation rates were measured with the two forms of GroEL in the presence of ATP at 37°C. Under the experimental conditions used, EcGroEL dissociated from EcGroES with a rate constant of ϳ4.1 ϫ 10 Ϫ2 s Ϫ1 (Fig.  2A), corresponding to a half-time for dissociation of ϳ35 s. Interestingly, MmGroEL was found to dissociate from immobilized MmGroES ϳ15 times more slowly, with a rate constant of ϳ2.6 ϫ 10 Ϫ3 s Ϫ1 (Fig. 2B). Because the folded subunits of oligomeric substrate proteins can associate and become biologically active only after leaving the GroEL/GroES cage, the M. mazei chaperonin system would be about 15 times less efficient in supplying growing cells with active enzyme. This difference may provide an explanation for the failure of the M. mazei system to replace the corresponding E. coli system in vivo (Table I).
The results of the SPR analysis also readily explain the failure of MmGroEL to cooperate with EcGroES in rhodanese refolding. As shown in Fig. 2A, MmGroEL does not bind to EcGroES, whereas EcGroEL binds to MmGroES and is released in the presence of ATP with the same slow kinetics as observed for MmGroES (Fig. 2B). This finding indicates that the intrinsic binding properties of MmGroES alone must contribute significantly to the slow cycling rate on and off GroEL. Moreover, the SPR measurement demonstrates that MmGroES has a higher affinity for EcGroEL than EcGroES, consistent with the observation in vivo that MmGroES complements E. coli mutants groES619 and groEL44 (Table II).

MmGroEL Has a Much Lower ATPase Activity Than
EcGroEL-The rate of GroES cycling on and off GroEL is linked with the ATPase activity of GroEL. We therefore compared the ATPase activity of MmGroEL to that of EcGroEL. The MmGroEL oligomer has an ATPase activity (ϳ4.5 ATP min Ϫ1 ) ϳ20 times lower than its bacterial homologue (Fig. 2C), consistent with the ϳ15 times slower rate of GroEL/GroES cycling measured for the M. mazei chaperonin system by SPR (Fig. 2B). As reported previously for EcGroEL/GroES, binding of GroES inhibits the ATPase activity of GroEL by ϳ50% (34,46). Interestingly, MmGroES has a much stronger effect than EcGroES on reducing the ATPase activity of EcGroEL (Fig.  2C), consistent with the observation that the off-rate of EcGroEL from MmGroES is much slower than that for EcGroES.
The Archaeal Chaperonin Refolds Rhodanese But Releases It Very Slowly-Both the SPR kinetic analyses and the ATPase assays indicated a slower cycling of the archaeal chaperonin system compared with its bacterial counterpart. Nevertheless, MmGroEL/GroES appears to support the folding of rhodanese to its enzymatically active state as efficiently as EcGroEL/ GroES, as shown in Fig. 1.
It seemed possible that rhodanese accumulates inside the MmGroEL-GroES complex in its active form but is not readily released into the medium. We therefore investigated the release of rhodanese from the GroEL cavity during ongoing refolding with MmGroEL/GroES or EcGroEL/GroES. Refolding reactions were stopped after 5 min by the addition of glucose/ hexokinase, to rapidly convert the ATP in the reaction to ADP, thereby arresting the chaperonin as a stable GroEL-GroES-ADP complex. The reactions were then analyzed by size exclusion chromatography and rhodanese enzyme activities were determined in the fractions. Considerably more rhodanese activity co-eluted with the MmGroEL-GroES complex than with EcGroEL:GroES, indicating that the MmGroEL/GroES system releases folded rhodanese more slowly than EcGroEL/GroES (Fig. 3). It has been shown previously that monomeric rhodanese acquires full enzymatic activity while being enclosed in the GroEL/GroES cage (35,47).
MmGroEL/GroES Fails to Produce Active, Dimeric Malate Dehydrogenase-To further investigate the mechanism of the archaeal chaperonin system, we used the dimeric protein MDH (ϳ45 kDa subunits) as a stringently GroEL-dependent protein. This enzyme is known to require the chaperonin for successful refolding from denaturant, as shown with EcGroEL/GroES. In contrast to rhodanese, MDH must dimerize to become functionally active and thus is an appropriate substrate to investigate the efficiency of protein release from the archaeal chaperonin. Strikingly, under the conditions optimal for refolding with EcGroEL/GroES, the MmGroEL/GroES system failed to produce any active MDH (Fig. 4A). In addition, active MDH was also not produced by the hybrid EcGroEL/MmGroES system (data not shown). These findings were surprising as we failed to detect any aggregated MDH protein in the chaperonin reactions (data not shown). To investigate whether MDH was retained by MmGroEL during refolding, as seen with rhodanese, refolding reactions were stopped after 5 and 45 min and subjected to size exclusion chromatography, followed by SDS-PAGE analysis of the fractions and immunoblotting for MDH. Even after 45 min virtually all MDH was still bound to MmGroEL/GroES, whereas free native MDH was observed with EcGroEL/GroES already after 5 min of refolding (Fig. 4B).
Ammonium Sulfate Enhances the ATPase Activity of MmGroEL and the Release of Folded Substrate-We next looked for conditions that could enhance the ATPase activity and therefore the refolding activity of the MmGroEL/GroES system. It had been reported that AS has a stimulatory effect on the ATPase of both EcGroEL (48) and the archaeal thermosome from Methanopyrus kandleri (49). Indeed, titration experiments showed that a concentration of 0.5 M AS was optimal in stimulating the ATPase activity of MmGroEL (data not shown), more than doubling the rate compared with conditions in the absence of AS (Fig. 5A). This stimulated ATPase activity was inhibited by GroES, suggesting that AS did not cause a functional uncoupling of the GroEL ATPase. Notably, the ASstimulated ATPase activity was still 5-fold slower than the unstimulated rate of EcGroEL.
SPR experiments revealed that in the presence of 0.5 M AS, MmGroEL dissociated from MmGroES with a rate of ϳ1.1 ϫ 10 Ϫ2 s Ϫ1 , i.e. about 4 times faster than in its absence (Fig. 5B). This dissociation rate is ϳ4-fold slower than that observed for the dissociation rate of EcGroEL from EcGroES (ϳ4.1 ϫ 10 Ϫ2 s Ϫ1 ) measured in the absence of AS (Fig. 5C). Approximately 2-fold higher levels of EcGroEL binding to EcGroES were observed in the presence of AS, presumably because AS increased the association rate for EcGroEL binding more strongly than the dissociation rate, which was accelerated by a factor of ϳ2.5 (Fig. 5C).
The presence of AS caused a slight increase in the rate of rhodanese folding for both GroEL/GroES systems, whereas the yield of refolding remained unchanged (Fig. 6A). Analysis by size exclusion chromatography showed that AS also accelerated the release of folded rhodanese from MmGroEL/GroES (Fig. 6B). Strikingly, in the presence of AS the archaeal chaperonin system also supported the production of active dimeric MDH at a rate similar to that observed with EcGroEL/GroES in the absence of AS (Fig. 7A). Interestingly, the rate of MDH renaturation by EcGroEL/GroES was also significantly enhanced by AS and an increase in refolding yield was consistently observed under these conditions. The basis for this effect remains to be investigated but it is likely because of an increased efficiency of GroEL/GroES function. An effect on MDH itself cannot be ruled out, however, although AS did not measurably improve the ability of MDH to refold spontaneously (Fig. 7A).
Analysis of MDH refolding reactions with MmGroEL/GroES in the presence and absence of AS demonstrated that the production of free MDH was strictly dependent on AS (Fig. 7B). Release from GroEL allows dimerization of folded monomers and subsequent acquisition of MDH activity.
Based on the result obtained with rhodanese (Fig. 3), it was possible that the inability of MmGroEL/GroES to assist MDH refolding in the absence of AS resulted from inefficient release of folded MDH from the cis cavity of the MmGroEL-GroES complex. To address this possibility, we performed proteinase K protection assays to investigate whether MDH was enclosed in the GroEL-GroES complex under these conditions (35,50,51). As expected, addition of proteinase K to GroEL-bound MDH in the absence of GroES resulted in the rapid digestion of the unfolded protein, independent of the presence of AS (Fig. 8,  A-D, left panels). When EcGroES was added to the EcGroEL-MDH complex in the presence of the non-hydrolysable ATP analog AMP-PNP, with or without AS, a substantial amount of the bound MDH was protected against proteolysis, indicating that this protein was encapsulated in the EcGroEL/GroES cage (Fig. 8, A and B, right panels). The same result was obtained when the experiment was performed in the presence of ADP (data not shown). In contrast, addition of MmGroES to MmGro-EL:MDH under the same conditions resulted in MDH protection only when AS was also added (Fig. 8, C and D, right  panels). In the absence of AS, virtually all the MDH protein was digested by proteinase K and was therefore bound to the trans ring of the MmGroEL-GroES complex (see also Fig. 4B). MmGroES was indeed bound to MmGroEL in trans to MDH, because half of the GroEL subunits were protected against cleavage of their flexible C-terminal sequences by proteinase K (50), as revealed by SDS-PAGE and Coomassie Blue staining (Fig. 8E, right panel). In contrast, in the absence of GroES, the majority of GroEL was C-terminal truncated under the conditions of mild proteolysis used (Fig. 8E, left panel). These effects were independent of the presence of AS.

MDH Bound in Trans to MmGroEL/GroES Can Refold upon Addition of Ammonium Sulfate or Transfer to the Bacterial
Chaperonin System-Having established that MDH is bound to the trans ring of MmGroEL:GroES in the absence of AS, we next asked whether this MDH is cycling on and off the trans ring in the presence of hydrolysable ATP. To address this possibility, we added an excess of EcGroEL and EcGroES 5 min into a refolding reaction of MDH with MmGroEL/GroES in the absence of AS. The MDH bound to the trans ring of MmGroEL: GroES was readily transferred to the bacterial chaperonin system and refolded with the same efficiency as with EcGroEL/ GroES alone ( Fig. 9; see also Thus, in the absence of AS, MDH cycles from the trans ring of MmGroEL but does not refold. Interestingly, when instead of the bacterial chaperonin, AS was added to a reaction containing MmGroEL, MmGroES, MDH, and ATP after 5 or 15 min incubation, MDH refolding was observed at a rate and yield similar to that measured when AS was present at the onset of the reaction (Fig. 9). Thus, addition of AS initiates encapsulation of MDH in the MmGroEL/GroES cage and this results in efficient refolding. DISCUSSION Thus far, three species of the archaeal genus Methanosarcina have been found to contain both group I (GroEL/GroES) and group II (thermosome) chaperonins in the cytosol (25). The groESgroEL operon is believed to have been acquired by these species by lateral gene transfer from an unknown bacterial source, along with ϳ1000 protein-encoding genes of bacterial origin (ϳ30% of the M. mazei genome) (24). To gain insight into how GroEL/GroES may have adapted to this unusual biological context, we have performed a detailed functional analysis of the M. mazei GroEL/GroES system in comparison with that of E. coli. Consistent with a significant diversion from its bacterial ancestor, the archaeal groESgroEL operon failed to replace the E. coli groESgroEL operon. As a likely explanation for this finding, our measurements showed that MmGroEL has a much slower ATPase activity and a correspondingly slower cycling rate for GroES than the bacterial system. This results in a prolonged enclosure time for the unfolded protein substrate in the GroEL:GroES folding cage, suggesting that the M. mazei chaperonin has adapted to the slower growth rate of this organism and/or to a subset of slow folding substrate proteins. For the model substrates tested, productive folding was observed only upon encapsulation of the unfolded protein in the M. mazei GroEL cavity by GroES. The emergence of only folded proteins from the GroEL/GroES cage may prevent adverse interactions of the GroEL substrates with the thermosome, which is not normally located within the same compartment.
The main functional parameter that differs between the archaeal and the bacterial GroEL/GroES systems is the rate of GroES cycling on and off GroEL. The slower cycling rate of the M. mazei system limits the capacity of the chaperonin to release folded protein into the cytosol and this probably explains why MmGroEL/GroES fails to functionally replace the E. coli chaperonin, which is adapted to the faster growth rates of this organism. Properties of both MmGroEL and MmGroES contribute to slow cycling. The ATPase activity of MmGroEL is severalfold reduced compared with that of EcGroEL, independently of the stimulating effect of AS. It has frequently been observed that archaeal ATPases are less active than their bacterial counterparts (49,(52)(53)(54)(55)(56). A relevant example in this context is the thermosome, whose ATPase is also activated by AS in vitro (49). A comparison of the ATPase centers of M. mazei and E. coli GroEL based on the structures of EcGroEL and EcGroEL/GroES/ADP (57-60) reveals no obvious differences that may account for the differential ATPase rates, because most critical residues participating in nucleotide binding and hydrolysis are conserved. The only obvious difference is the exchange of alanine 481 in EcGroEL, which contacts the purine ring through van der Waals interactions, by a lysine in MmGroEL. Whether this structural difference is relevant, directly or indirectly, with regard to the stimulatory effect of AS remains to be explored.
MmGroES exhibits an ϳ15-fold slower dissociation rate from MmGroEL compared with the E. coli GroEL/GroES pair, resulting in an increased GroES affinity of MmGroES for GroEL. Interestingly, this effect is also observed upon interaction of MmGroES with EcGroEL and thus reflects, at least in part, a property of GroES independent of the intrinsic ATPase activity of the respective GroEL partner. The association of GroES with GroEL is mediated by flexible loop sequences extending from the base of the GroES oligomer (60 -62). Upon binding to GroEL, these sequences assume a ␤-hairpin structure (60,63). GroES proteins with loop sequences that are more disordered in solution tend to exhibit a lower affinity for GroEL as a result of the greater entropic penalty incurred upon binding (27,63). 2 On the other hand, loop sequences that are more ordered in solution and exhibit higher hydrophobicity in the residues that contact GroEL bind with higher affinity. Indeed, the sequence of the tripeptide in the loop that contacts GroEL directly is more hydrophobic in MmGroES (residues IYI at positions [25][26][27] than in EcGroES (residues IVL). The tripeptide in MmGroES is similar in hydrophobicity to that in human Hsp10 (IML), which binds more strongly to EcGroEL than EcGroES (27). Moreover, the highly conserved proline residue in mitochondrial Hsp10 following the hydrophobic tripeptide was shown to reduce the dynamics of the mobile loop, making it less flexible than its bacterial counterpart, which has a threonine at this position (27). The GroES proteins of the Methanosarcina species all have a proline after the hydrophobic tripeptide. Additionally, the residue at position 21 in the Methanosarcina GroES is a threonine as in mitochondrial Hsp10. This residue was found to increase the affinity of Hsp10 for EcGroEL by increasing the propensity of the sequence to assume a ␤-hairpin structure in solution (27,65,66). EcGroES has a serine at the corresponding position and a serine to threonine mutation in EcGroES results in a higher affinity for GroEL (27).
These features of the MmGroES mobile loop very likely determine the increased affinity for GroEL and may explain why MmGroES restores a functional interaction with the mutant GroEL44 protein, allowing bacterial growth at 43°C and the propagation of bacteriophages and T5 at 37°C. GroEL44 does not bind appreciably to Gp31, the T4 bacteriophage analog of GroES (27), and also has a drastically reduced affinity for EcGroES. 2 The observation that overexpression of MmGroEL interferes with E. coli growth, especially at 43°C, and with bacteriophage T4 growth at 37°C is consistent with the possibility that some bacterial substrates and the T4-encoded Gp23 capsid protein may bind the M. mazei chaperonin but may not fold in a timely fashion.
Our functional analysis of the M. mazei chaperonin system also lends strong support for the significance of protein encapsulation as a major mechanistic feature of chaperonin-assisted folding. The unusual functional dependence of M. mazei GroEL/GroES on AS allowed us to compare two modes of action of the chaperonin in vitro, substrate encapsulation in the cis cavity and cycling from the GroEL trans ring. In the absence of AS, the GroEL model substrate MDH went through ATP-dependent cycles of binding and release from the MmGroEL trans ring in a manner non-productive for folding. Productive folding was only achieved upon encapsulation of MDH in the GroEL cis cavity by GroES, and in vitro this step was critically dependent on addition of AS. In contrast, encapsulation of the smaller protein rhodanese occurred independently of AS. We therefore suggest that the presence of AS facilitates the GroES-mediated displacement of MDH from its multiple attachment sites on GroEL into the chaperonin cavity. The fact that MmGroES supports bacteriophage T5 growth on groES619 mutant bacteria and bacteriophage and T5 growth on groEL44 mutant bacteria suggests that the in vivo environment of the bacterial cytosol can replace the effect exerted by AS in vitro.
Whereas stringent GroEL substrates such as Rubisco and MDH reach the native state rapidly and in a highly efficient manner only through the encapsulation mechanism (64), it was recently shown for E. coli GroEL that some of these proteins can fold, albeit slowly and with lower yield, by cycling from the trans GroEL ring when cis encapsulation is disabled (15). The trans mechanism is thought to allow GroEL assisted folding of certain proteins too large to be encapsulated (14). Based on our findings, the M. mazei GroEL/GroES system does not support such a reaction, at least under the experimental conditions in which trans cycling was observed. It seems possible that in M. mazei the GroEL/GroES system has lost the ability to support trans folding because of the simultaneous presence of the thermosome, which functions independently of a GroES-like cofactor and may be able to assist the folding of larger proteins in a domain-wise manner.