A mutation in GroEL interferes with protein folding by reducing the rate of discharge of sequestered polypeptides.

GroEL140, a mutant Escherichia coli chaperonin unable to support bacteriophage lambda head assembly, was purified to near homogeneity and compared to wild type GroEL (cpn60). GroEL140 exhibited a 1.5-fold lower ATPase activity relative to the wild type protein. The hydrolysis of ATP by both polypeptides was fully inhibited by an excess of ATP gamma S and partially inhibited by ADP and 5'-adenylylimidodiphosphate, suggesting that adenine nucleotides display different affinities for the ATP binding site of chaperonins. GroEL140 was more sensitive to trypsin digestion compared to wild type GroEL indicating that the mutation destabilized the conformation of the mutant. The proteolytic susceptibility of both chaperonins was similarly enhanced upon addition of ATP, ADP or non-hydrolyzable ATP analogs, providing evidence (i) of a conformational change in the chaperonin structure which is likely to drive the protein discharge process, and (ii) that hydrolysis of ATP is not required to achieve topological modifications. GroEL140 retained its ability to bind non-native ribulose bisphosphate carboxylase/oxygenase (Rbu-P2-carboxylase), but released bound proteins upon addition of ATP and GroES (cpn 10) 6-7-fold less efficiently compared to GroEL. This functional defect was shown to be related to a suboptimal, but not an absence of, interaction with GroES since (i) GroEL140 and GroES were unable to form a complex isolatable by size exclusion chromatography, and (ii) increasing the incubation time or the concentration of GroES enhanced the amount of refolded Rbu-P2-carboxylase discharged from GroEL140-Rbu-P2-carboxylase binary complexes. Pulse-chase experiments involving a double immunoabsorption technique confirmed that Rbu-P2-carboxylase remained associated two times longer with GroEL140 than with GroEL in vivo.

Protein folding has recently emerged as a central issue in both fundamental and applied research. The isolation and characterization of molecular chaperones which can modulate the folding and subsequent oligomerization, transport, and stability of polypeptides (reviewed in Refs. [1][2][3][4][5] has certainly contributed to a renewed interest in the pathways proteins follow to reach their final conformation. Molecular chaperones have been conserved throughout evolution and have been identified in most of the organisms and cellular compartment studied.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ T o whom correspondence should be addressed. Tel.:  One of the most abundant molecular chaperones in bacteria is chaperonin GroEL (also known as cpn60 or hsp6O in different organisms), a molecule composed of 14 identical subunits (Mr -57,000) which associate to form a double stack of heptameric rings in prokaryotic and plant cells (2,6). Recent investigations have, however, shown that the mammalian mitochondrial homolog of GroEL can exist as a functional single heptameric ring (7). GroEL, which is essential for cell survival ( 8 ) , was initially identified for its requirement in the assembly of X and T4 phage heads, and that of T5 tails (9). Genetic evidence also suggests that GroEL is involved in DNA replication (10,11), cell division (12), and protein transport (13-15). GroEL displays a weak ATPase activity which is dependent upon the presence of magnesium and potassium ions (16), and promotes its own assembly (17). Compelling genetic (14,15,(18)(19)(20) and biochemical (16,21,22) evidence indicates that GroEL functionally interacts with the heptameric cochaperonin GroES (also known as cpnl0).
GroEL and related chaperonins associate with non-native polypeptides, thereby preventing the proteins from entering unproductive aggregation pathways, and are able to discharge bound proteins in a biologically active conformation upon addition of ATP and GroES (16,21,(23)(24)(25)(26)(27). However, many of the mechanistic details of this process remain obscure. In this paper, we have purified a mutant GroEL protein and compared it to the wild type in an effort to gain some insight on the mechanism of action of chaperonins. The mutant GroELllO was found to qualitatively behave as the wild type protein in its ability to bind non-native proteins, to promote ATP hydrolysis, and in its susceptibility to proteolysis. Nevertheless, it was less efficient and more sensitive to proteolysis than wild type GroEL. Furthermore, the mutant chaperonin was inefficient in discharging bound proteins in a biologically active form. We determined that the decreased yields were related to an impaired ability of GroEL140 to bind GroES. The implications of these experiments for the mechanism of action of chaperonins are discussed.
GroELl4O Purification-Two shake flasks containing 1 liter of LB medium (Difco) supplemented with 0.2% glucose were inoculated at a 1:20 dilution with an overnight inoculum of CG714 cells grown in LB medium at 30 "C. Cultures were incubated with agitation at 30 "C to late exponential phase (OD, = 0.8) and rapidly shifted to 42 "C by transfer to a 60 "C water bath. Incubation was continued at 43 'C for an additional 50 min, at which point the cells were pelleted by centrifugation at 10,000 rpm X 10 min. The pellet was resuspended in 20 ml of buffer I (50 mM Tris-HC1, pH 7.2, 0.1 mM EDTA, 1  and equilibrated with buffer I. The column was developed at 2 ml/ min with a 525-m1 gradient from 0 to 1 M NaCl in buffer I at 4 "C. Fractions eluting between 55 and 62% NaCl were pooled, concentrated by ultrafiltration using a YM-30 membrane (Amicon), and loaded on a Superose 6-HR 10/30 column (Pharmacia) developed in buffer 11 (100 mM Tris-HC1, pH 7.6, 10 mM MgCIa, 10 mM KCl, 0.01% Tween 20) at 0.5 ml/min using the Pharmacia fast protein liquid chromatography system. The fractions eluting between 12 and 13 ml (the column void volume is 10 ml) were pooled, concentrated using Amicon PM-10 microconcentrators, and rechromatographed on Superose 6. The GroEL140 peak fractions were concentrated, supplemented with 10% glycerol, frozen in liquid nitrogen, and stored at -80 "C. The protein concentration was 45 p M (protomer) as estimated by amino acid analysis and video scanning. GroELl4O appeared about 95% pure on overloaded SDS gels and migrated at the same position as wild type GroEL on 6% non-denaturing gels both in the presence and absence of Mg-ATP. Proteolytic Digestion-All proteolytic digestion experiments were performed in buffer I1 at 37 "C using a 0.17 mg/ml solution of trypsin (Sigma) in a total reaction volume of 100 pl. Briefly, 8 pl of GroEL (64 p M protomer) or GroEL140 (45 p M protomer) were mixed in buffer I1 in Eppendorf tubes held at 0 "C. For some experiments 3.2 pl of purified GroES (308 p~ protomer) was added to the reaction mixture. This GroES concentration corresponded to a 3.8 or 5.5 excess of (GroES)7 over ( G~O E L )~~ or (Gr0EL140)~~, respectively. At zero time, 2.5 mM Mg-ATP (Sigma) or 5 mM of either ADP (Sigma), AMP-PNP (Sigma), or ATPyS (Boehringer Mannheim) was added to the tube. This step was followed by addition of 8 p1 of 0.17 mg/ml trypsin. A 12.5-p1 sample was immediately withdrawn, quenched with 10 mM phenylmethylsulfonyl fluoride, and placed on ice. Time course samples were obtained and treated similarly. An equal volume of 2 X loading buffer was added to the samples which were boiled for 5 min and resolved on 12.5% polyacrylamide gels. The wet gels were digitized using a Cohu camera and the Quickcapture software. The amount of protein present in the GroEL band was estimated by digital integration using the Image software. The zero time point was assigned a 100% value.
ATPase Assays-The ATPase activity assays were performed essentially as described by Viitanen et al. (16). Reactions were carried out in a 160-pl final volume. The assay buffer (buffer 111) consisted of 100 mM Tris-HC1, pH 7.7,l mM DTT, 10 mM MgC12, and 100 mM KC1. Samples contained 0.1 p~ (protomer) of the appropriate dilutions of GroEL and GroELl4O and 9.375 p~ of ATP supplemented with 1.25 pCi of [y-32P]ATP at 3,000 Ci/mmol (Du Pont-New England Nuclear). A 25-pl sample was taken immediately and further samples withdrawn every 5 min. Phosphate extraction was carried out as described (16). Potassium chloride was omitted from buffer I11 for potassium-free experiments, and the samples were supplemented with 100 mM KC1 (final concentration) 30 s following the 10 min time point as indicated by the arrow on Fig. 1. For the nucleotide inhibition experiments, 1 mM of ADP, AMP-PNP, or ATPyS was added to the samples immediately after addition of ATP. All experiments included control reactions carried out in the absence of GroEL or GroEL140 to correct for the acid hydrolysis of [y-32P]ATP. The turnover numbers were estimated by least squares fit of the initial velocity profiles with correlation coefficients greater than 0.95.
Size Exclusion Chromatography-[35S]Rbu-P~-carboxylase was isolated from DS410(pRR2119) minicells as described (30). Purified unlabeled Rbu-Pa-carboxylase was added to the purified concentrated radioactive fractions to a final concentration of 10.35 p M for a specific activity of 780 Ci/mol. Rbu-P2-carboxylase was acid denatured by dilution in 10 mM HC1 to a final concentration of 1.66 p M (protomer). The unfolded protein (20 pl) was loaded at 0 "C by rapid mixing onto GroEL or GroEL140 (2.4 p~ final concentration of protomer) diluted in buffer I1 + 5 mM DTT. Samples were incubated on ice for 5 min The abbreviations used are: DTT, dithiothreitol; AMP-PNP, 5' adenylylimidodiphosphate in lithium salts; ATPyS, adenosine 5'43- (3-thiotriphosphate) in tetralithium salts; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; GroELI4, the 14-mer form of the GroEL molecule; GroEL14014, the 14-mer form of the GroEL140 mutant molecule; GroES,, the 7-mer form of the GroES molecule. and at room temperature for 10 min. For some experiments 8.9 p~ of GroES (final concentration of protomer) and/or 2.5 mM ATP were added after 5 min of incubation on ice, and the samples were transwas 300 pl. Exactly 200 p1 were injected on a TSK G3000SW (LKB) ferred at room temperature for 20 min. The final reaction volume column equilibrated with buffer I1 or buffer I1 + 0.25 mM ATP for the experiments including ATP treatment. The column was developed at 1 ml/min using the Pharmacia fast protein liquid chromatography system. Fractions (0.5 ml) were collected directly in scintillation mixture and counted. The experiment of Fig. 4 was performed by mixing GroEL (1.2 p M protomer) and GroES (2.8 p~ protomer) in the presence of 2.5 mM ATP in buffer I + 5 mM DTT. The samples (200 pl) were resolved by size exclusion chromatography on a Superose 6-HR 10/30 column equilibrated with buffer I1 and developed at 0.5 ml/min. The fractions corresponding to the GroEL or GroEL140 peaks were pooled, and the proteins were recovered by acetone precipitation. Samples were run on a 15% SDS gel, transferred for 1 h to nitrocellulose, and probed with anti-GroES (1:2000 dilution) and goat-anti-rabbit IgG conjugated with alkaline phosphatase. The bands were identified by color reaction.
Rbu-P2-carboxylase Refolding Assays-Chaperonin-mediated Rbu-P2-carboxylase refolding was determined essentially as described by Goloubinoff et al. (21). Purified Rbu-P2-carboxylase was acid denatured by 312.5-fold dilution in 10 mM HCl (final concentration 10 pM) followed by incubation at room temperature for 30 min. The refolding buffer (buffer IV) consisted of 42 mM Tris-HC1, pH 7.7, 8.4 mM MgC12, 11.7 mM KC1, 0.26 mg/ml bovine serum albumin and 2.1 mM DTT. The final reaction volume was 427.5 pl with a final concentration of 1.9 p~ (protomer) of GroEL or GroEL140. Acidunfolded Rbu-P2-carboxylase (0.09 p~ of protomer) was loaded by rapid mixing onto the chaperonins diluted in buffer IV. This operation was performed at 0 "C and was followed by an additional 5-min incubation step on ice. At zero time 2.8 p~ GroES (protomer) and 9 mM ATP were added to the samples. The high ATP concentration was necessary to guarantee that some unhydrolyzed ATP would still be present after 6 h of incubation. Aliquots (85.5 pl) were withdrawn at the times indicated in Fig. 5 and assayed for Rbu-P2-carboxylase activity. For the incomplete reaction experiments, neither GroES, nor ATP were initially provided. ATP was added to the samples at a 9 mM final concentration 5 min after the 90-min point was taken, and the mixture was supplemented with 2.8 p~ GroES (final concentration of protomer) 5 min after the 270-min point. The experiment of Fig. 5B was performed in a 85.5-p1 final reaction volume as described above but with the concentrations of GroES indicated in the text.
The amount of refolded Rbu-Pa-carboxylase was estimated as in Goloubinoff et al. (21) with some modifications. Samples were mixed with 11.8 p1 of buffer V (20 mM Tris-HC1, pH 7.7, 300 mM NaHC03, 34 mM MgCla, 67 nCi/pl of ['4C]Na2C03 (55 mCi/mmol, Du Pont-New England Nuclear)) and incubated at room temperature for 10 min. Next, 3 pl of 25 mM purified ribulose bisphosphate were added, and the mixture was incubated for 30 min at 23 "C. The reaction was stopped by addition of 210 p1 of acetic acid, and the liquid-phase evaporated on a hot plate. The samples were resuspended in 500 pl of water and the 14C incorporated into stable acid products was determined by liquid scintillation counting.
Zmmunoisolutwn of GroEL-Rbu-P2-carboxylase Binary Complexes formed in Viuo-Strains B178 and CG714, transformed with the Rbu-Pz-carboxylase expression plasmidpRR2119 (29), were grown at 30 'C in M9 minimal medium containing 0.2% glucose, 1% methionine assay medium (Difco), 50 pg/ml ampicillin to an OD, = 0.2. Rbu-P2-carboxylase expression was induced by the addition of 1 mM isopropyl-P-D-thiogalactoside for 1 h. Cells were then labeled by incubating 3 ml of culture with 100 pCi of [35S]methionine for 2 min, followed by the addition of 1 ml of medium containing 1% methionine. At 0, 1, and 4 min after addition of the chase, 1-ml samples were removed and rapidly chilled in the presence of 100 pg/ml chloramphenicol, and the cells were pelleted in a refrigerated microcentrifuge for 30 s. Pellets were resuspended in 200 pl of buffer VI (30 mM Tris-HCl, pH 8.0, 20% sucrose, 5 mM EDTA, 100 pg/ml lysozyme) for 10 min on ice, followed by the addition of 1 ml of buffer VI1 (50 mM Tris-HC1, pH 7.5, 20 mM MgCla, 5 mM KCl, 0.01% Tween 20) for an additional 30 min. NaCl was added to a final concentration of 0.2 M, and the samples were centrifuged for 10 min in a microcentrifuge at 2 "C. The supernatants were incubated with shaking for 2 h on ice with 5 mg of Protein A-Sepharose previously incubated with anti-GroEL serum. The Sepharose beads with bound GroEL and any trapped Rbu-Pa-carboxylase were washed three times with cold buffer VI1 containing 0.2 M NaC1, and the immunoabsorbed binary complexes between GroEL or GroELllO and substrate proteins were disrupted and released by two 100-p1 washes of 1% SDS at 90 "C.
The SDS samples were pooled, diluted with 1 ml of buffer VI11 (50 mM Tris-HC1, pH 7.8, 150 mM NaCl, 2 mM EDTA, 1% Triton Xloo), and incubated with anti-GroEL and anti-Rbu-P2-carboxylase to identify the two radiolabeled proteins of interest. After subsequent absorption to Protein A-Sepharose and washing (31), the proteins were identified by SDS-PAGE and fluorography using radiolabeled GroEL and Rbu-P2-carboxylase as markers. GroEL and Rbu-Pzcarboxylase bands were excised from dried gels, incubated with NCS tissue solubilizer (Amersham Corp.) at 50 "C overnight, neutralized with acetic acid and counted in Econofluor 2 (Du Pont-New England Nuclear).

RESULTS
Purification of GroEL140-Escherichia coli strain CG714 encodes a mutant GroEL protein and was initially isolated for its inability to support bacteriophage X head morphogenesis (28). The mutation in GroEL140 was recently mapped.' It corresponds to the single amino acid substitution SerZa1 + PheZa1. Transcription of the groE operon, including the mutant groELl40 gene, is under the control of the Ea3' and Eo7' promoters (9). Neidhardt et al. (32) have shown that a temperature shift from 37 to 46 "C increases the synthesis of native GroEL from 2 to 10% of the total cellular protein, primarily because of enhanced expression by the Ea3' promoter (9). Therefore, to maximize groEL140 expression, CG714 was grown to late exponential phase in LB medium at 30 "C, rapidly shifted to 42 "C, and incubated for 50 min at this temperature (see "Materials and Methods"). GroEL140 was isolated from disrupted cells by DEAE-Sephacel ionexchange chromatography followed by two gel filtration steps. The concentrated GroEL140 protein migrated as wild type GroEL on non-denaturing SDS gels and was estimated to be more than 95% pure by video scanning of overloaded SDS gels (data not shown).
ATPase Actiuity and Inhibition-Bacterial GroEL displays a weak ATPase activity which fully depends on the presence of magnesium and potassium ions (16). Since the binding and hydrolysis of ATP appear to play a major role in the release of folding intermediates associated with several molecular chaperones, we first determined the influence of the single amino acid substitution in GroEL140 on the ATPase activity of this protein. Under the experimental conditions chosen (see "Materials and Methods"), the ATP turnover was 1.45 X lo-' pmol of ATPlpmol of GroEL140 (protomer)/s with a standard deviation of 0.011 x lo-*. This result was about 1.5fold lower compared to the turnover number obtained with wild type GroEL (2.11 x lo-' f 0.073 X lo-', Fig. 1). When potassium ions were omitted, no ATPase activity could be detected with either GroEL or GroEL140. The ATPase activity, however, was fully restored upon addition of 100 mM KC1 to the reaction mixture (arrow in Fig. 1). This result indicated that neither purified preparations were contaminated with an extraneous, potassium independent, ATPase activity. ATP analogs have been reported to influence the release of folding intermediates complexed with GroEL (24,26,27)? To verify that these nucleotides could indeed interact with the ATP-binding site of cpn60, the above experiment was repeated in the presence of 1 mM AMP-PNP, ATP-yS, or ADP. Table I shows that all adenine nucleotides examined were able to inhibit the ATPase activity of both GroEL and GroEL140, presumably by competing for the ATP-binding site. ATPyS fully inhibited hydrolysis in both the wild type and the mutant protein, suggesting that this ATP analog C. Georgopoulos, personal communication.
' F. Baneyx, unpublished results. mutant (closed symbols) chaperonins (0.1 p~ final protomer concentration) were determined in buffer containing 10 mM MgCl, and 9.375 p~ ATP as described under "Materials and Methods." The assay was performed in the presence of 100 mM KC1 with GroEL (0) and GroELllO (0). To demonstrate the K' dependence, the experiment was repeated in potassium-free buffer and 100 mM KC1 was added to the reaction mixture 10.5 min following addition of ATP (arrow).
could efficiently recognize the ATP-binding site on both the wild type and mutant chaperonins. ADP was slightly less effective and conferred 80 and 70% inhibition in GroEL and GroEL140, respectively. AMP-PNP inhibition was inefficient with GroELl4O but still appreciable with wild type GroEL ( Table I). The above result suggests a poor binding of AMP-PNP to GroEL140 possibly due to modifications in the conformation of the mutant protein.
Proteolytic Digestion-Conformational changes have been proposed to play a major role in the binding of folding intermediates and in the discharge of polypeptides from molecular chaperones. However, very little information is currently available on the various conformational states of GroEL. We have used protease digestion to determine the extent of topological modifications resulting from the binding and hydrolysis of ATP by GroEL. Fig. 2 shows that although native GroEL is resistant to trypsin digestion under the experimental conditions chosen (panel A , see "Materials and Methods"), about 50% of GroEL140 was degraded following 30 min of incubation with trypsin at 37 "C (panel E ) . Addition of 2.5 mM ATP to the reaction mixture significantly enhanced the proteolytic susceptibility of both GroEL and GroEL140, reducing their half-lives to 22 and 3.6 min, respectively. Since both the binding and hydrolysis of ATP are likely candidates for inducing a structural change, the experiment was repeated in the presence of 5 mM ADP or 5 mM of the non-hydrolyzable ATP analogs AMP-PNP and ATP-yS. A comparable, or in some instances an enhanced, susceptibility to proteolysis was observed when the adenine nucleotides were incubated with GroEL or GroEL140. These results suggest that the binding, and not the hydrolysis of ATP, was responsible for exposing more arginine and lysine residues. The fact that GroEL140 is consistently more susceptible to trypsin digestion relative to the wild type can be interpreted as a more loosely assembled conformation directly resulting from the single amino acid substitution in this protein.
In contrast, the addition of a 3.8-fold molar excess of ( G~O E S )~ over (~p n 6 0 )~~ had essentially no effect on the ATPdependent degradation of GroEL or GroEL140 by trypsin (Fig. 2, panel C ) . It is interesting to note that in the presence of adenine nucleotides, the degradation of GroEL and Gro-EL140 by trypsin apparently occurs in a two-step process consisting of a rapid hydrolysis phase followed by a slower degradation stage. The first phase is completed within 5 min with GroEL140 and about 10 min with native GroEL. However, the rate of degradation of both proteins in the second, ~~p u i r e~ A b i~i~ of ~~t u n t GroEL peron on in to Bind GroES  slower phase appears to be almost identical.
Sinding and Release of Rbu-A-carboxylase to GmEL and G m E L l~~-R~s p i r~~u m rubrum R b u -P~-c~b o~l a s e consists of two identical monomers (L1) which associate to yield a biologically active enzyme (Lz) (33). Non-native Rbu-Pzcarboxylase tightly interacts with GroEL but can be mostly released in an active form when GroES, Mg-ATP, and K' ions are present (16, 21). The extent of Rbu-Pz-carboxylase refolding can be readily determined in an enzyme activity assay which measures the incorporation of 14C into acid stable products (34). In addition, the physical existence of the discharged LZ enzyme was recently demonstra~d by size exclusion chromatography (30). Therefore, we decided to use R. rubrum Rbu-Pz-carboxylase as a substrate to study the interaction of non-native polypeptides with GroEL140. Fig. 3 shows that acid-denatured [35S]Rbu-Pz-carboxylase forms a stable binary complex with GroEL140 eluting in the void volume of the TSK3000 size exclusion column. About 70% of the total radioactivity was present in the GroEL140 peak (compared to about 80% for wild type GroEL). This implied that the mutation in GroEL140 did not si~ificantly reduce the affinity of non-native polypeptides for GroEL14O. However, the mutant appeared to be compromised in its ability to discharge Rbu-Pz-carboxylase. Addition of 2.5 mM ATP to the binary complex released over 75% of the radioactivity associated with GroEL but only 18% of that complexed with GroELl4O (Table 11). In both cases, however, most of the released material did not fold in a monomeric or dimeric form and remained trapped in the column (Table 11). This result was previously observed (30) and indicates that the mere presence of ATP weakens the affinity of GroEL for Rbu-Pzcarboxylase. When both ATP and GroES were added, a significant increase in Rbu-Pz-carboxylase release was observed, and two peaks corresponding to the monomeric and dimeric forms of Rbu-Pz-carboxylase were detected (Fig. 3 ,Ref. 30). Under these conditions, more than 50% of the protein loaded onto GroEL140 was discharged compared to over 90% for GroEL (Table If). Furthermore, while 82% of the material dissociated from GroEL was recovered either in dimeric (62%) or monomeric form (20%), only 48% of the Rbu-Pz-carboxylase released from GroEL140 was obtained in a dimeric (26%) or monomeric form (22%).

I1 Release of ~S J R b u -P z -c a r b o x y l e loaded onto CroEL or GroELIlO by ATP or ATP and GroES Acid-denatured [RLS]
Rbu-P2-carboxylase was loaded onto GroEL or GroEL140 and incubated with no additives, 2.5 mM ATP, or 2.5 mM ATP and a 7-fold molar excess of (GroES): over (GroEL),, as descrihed under "Materials and Methods." Resuks are expressed as percent of the radioactivity associated with the GroEL/CroEL140 peak. I n Vitro Interaction between GroELl4O and GroES-Since the low efficiency of Rbu-Pz-carboxylase refolding by Gro-EL140 could only be marginally attributed to its reduced ATPase activity, and was not related to a poor binding of folding intermediates to the mutant chaperonin, we investigated the interaction of the cochaperonin GroES with the mutant GroEL140. In the presence of ATP, GroES forms a stable complex with GroEL a t a 1 to 1 or 2 to 1 ratio respectively,Ref. 16). The dissociation constant for the complex is relatively low since it can be isolated by size exclusion chromatography (Ref. 16;Fig. 4,lane 2). (Gr0EL140)~~ was incubated with a 5.6-fold excess of (GroES ) 7 in the presence of 2.5 mM ATP and resolved on a Superose 6 column equilibrated with 0.25 mM ATP as described under "Materials and Methods." The fractions corresponding to the GroEL140 peak were acetone precipitated and visualized by Western blotting using anti-GroES as a primary antibody (see "Materials and Methods"). The experiment was repeated with wild type GroEL as a control. Fig. 4 shows that GroES could not be detected in the GroEL140 peak ( l a n e 3 ) , while a band migrating at the same position as purified CroES was evident in the fractions isolated from the GroEL peak (10 kDa arrow, lanes 1 and 2). Since the anti-GroES antiserum is slightly cross-reactive with GroEL, faint bands corresponding to the GroEL and GroEL140 proteins were also visible on the blot (Fig. 4, 60 kDa arrow). Our inability to isolate a complex between GroES and GroEL140 could be explained by (i) a total absence of binding between GroES and the mutant chaperonin, or (ii) a weaker interaction between these proteins resulting in a larger dissociation constant. Nevertheless, since GroEL140 was able to yield a t least some dimeric (LJ Rbu-P2-carboxylase (Table 11) and because both GroES and ATP are required to achieve optimal results (Table 11, Ref. 21), we suspected that the two proteins could still interact, albeit suboptimally. This hypothesis was tested as follows.  Rbu-Pz-carboxylase unfolded by incubation in 10 mM HCl was loaded onto an excess of GroEL140 or GroEL as a control, and supplemented with a 3-fold molar excess of CroES (7mer over 14-mer) and 9 mM ATP. The reaction mixture was incubated a t room temperature for increasing periods of time and the amount of refolded Rbu-Pz-carboxylase estimated periodically. Fig. 5A shows that Rbu-Pz-carboxylase refolding by wild type GroEL was essentially complete after 3 h of incubation a t room temperature. GroELl4O was much less efficient in achieving refolding. After 90 min of incubation, the Rbu-P2-carboxylase activity for the GroEL14O sample was only 15% of that obtained with the wild type GroEL sample. Although the amount of active Rbu-Pz-carboxylase refolded by GroEL140 steadily increased with time, it was

Impaired Ability of Mutant GroEL Chaperonin to Rind GroES
still only 33% of the activity obtained with GroEL after 6 h of incubation. When an incomplete reaction mixture consisting solely of Rbu-P2-carboxylase loaded onto GroEL and GroEL140 was incubated under similar conditions, essentially no enzymatic activity could be detected. As expected, addition of 9 mM ATP to the samples (Fig. 5A, black arrow) and further incubation for 3 h only generated traces of Rbu-P2-carboxylase activity. However, when a 3-fold molar excess of (GroES), was added to the reaction mixtures (Fig. 5A, white arrow), a significant amount of Rbu-P2-carboxylase activity was obtained following an additional 90 min of incubation step. This result was consistent with the observation that, in addition t o ATP, GroES is necessary to achieve Rbu-P2-carboxylase refolding. The amount of Rbu-P2-carboxylase activity from the GroEL sample represented about 28% of that obtained after 90 min of incubation when a complete reaction mixture was used. The lower yield probably reflects the unproductive discharge of GroEL-bound Rbu-P2-carboxylase by ATP in the 3-h incubation period preceding GroES addition. As expected from the results in Table 11, discharge of GroEL140-bound Rbu-P2-carboxylase by ATP alone was much less efficient compared to wild type GroEL. Therefore, the amount of active Rbu-P2-carboxylase obtained from the GroEL140 sample following addition of GroES were comparable to that detected when a complete reaction mixture was used. In a separate experiment, we tested the influence of increasing concentrations of GroES on the GroEL140-mediated folding of Rbu-P2-carboxylase. Fig. 5R shows that as the molar excess of (GroES)? over (Gr0EL140)~~ increased, more Rbu-P2-carboxylase could be discharged from the mutant chaperonin. Interestingly, the same amount of active Rbu-P2-carboxylase (15% of the activity of the native Rbu-P2-carboxylase control) was obtained by either quadrupling the incubation time or quadrupling the excess of GroES. Thus, the above results demonstrate that GroEL140 is still able to interact with GroES and discharge biologically active Rbu-P2-carboxylase although the yield of the process is between 6-7-fold lower compared to native GroEL.
In Vivo Interaction between CroELI40 and Rbu-P2-carboxylase-In uitro experiments with purified proteins may not reflect the true behavior of complex systems in uiuo. In an attempt to determine the in uiuo role of the RrOEL140 mutation on cellular function, we transformed CG714 and the parental E. coli strain B178 with pRR2119, a plasmid encoding the R. rubrum Rbu-P2-carboxylase gene under the control of the lac promoter (29). Cells growing exponentially in methionine-free medium were induced with isopropyI-fi-D-thiogalactoside, pulsed with ["S]methionine, and chased for increasing amounts of time (see "Materials and Methods"). The binary complexes between folding intermediates and wild type or mutant GroEL were isolated by affinity adsorption onto Protein A-Sepharose beads which had been previously incubated with antiserum raised against GroEL. It has previously been reported that GroEL is a promiscuous molecular chaperone which interacts with about 50% of the E. coli proteins unfolded by GdnHCl treatment (30). When the proteins adsorbed onto anti-GroEL-protein A-Sepharose beads were resolved by SDS-PAGE and fluorography, a t least 20 discrete bands were observed confirming that GroEL also acts in a promiscuous fashion in vivo (data not shown). Since we were primarily concerned with the interaction of the chaperonins with Rbu-P2-carboxylase overproduced from pRR2119, the different proteins adsorbed onto the immunobeads were dissociated by treatment with heat and SDS. This was followed by a second incubation with anti-Rbu-P2-carboxylase (or anti-GroEL as a control) serum, Finally, Protein A-Sepharose beads were added to the mixture. The polypeptides released by heat-SDS treatment were resolved by SDS-PAGE and visualized by fluorography. By using this approach, we were able to determine the amount of Rbu-P,-carboxylase associated with GroEL a t different times following initiation of the chase (Fig. 6). Quantitative determination of the binding was performed by counting the radioactivity in the Rbu-P2-carboxylase bands a t different time points and normalizing the result to the radioactivity in the GroEL bands at the same times. Fig. 6 shows that Rbu-P,-carboxylase remained associated with GroEL140 about twice as long as with wild t-ype GroEL, suggesting that the process of polypeptide release was also less efficient in uiuo. Also note that immediately following the 2- min pulse (panel A, 0 rnin chose), there was significantly more Rbu-P2-carboxylase associated with GroEL140 compared with GroEL, as would be anticipated if Rbu-P2-carboxylase has a longer occupancy time on the mutant GroEL protein. The increased time of interaction between the mutant chaperonin and non-native proteins may explain the slower growth characteristics of CG714 compared to the parental strain B178.

DISCUSSION
In spite of the growing body of information dealing with molecular chaperones, very little is known about the in uiuo function and mechanism of action of these proteins. The E. coli chaperone DnaK, its mammalian homolog hsp72/73, and the related endoplasmic reticulum protein Hip, have heen extensively studied, and some clues regarding their mode of action have been recently obtained. Specific conformational changes in the molecule resulting from the binding and hy- drolysis of ATP (35), together with an interaction (in E. coli) with proteins DnaJ and GrpE (36) seem to drive the process of polypeptide binding and release. On the basis of oligopeptide interaction, it was further concluded that BiP preferentially binds protein domains exposing hydrophobic but flexible side chains (37). Although similar experiments have not been reported for the GroEL-related chaperonin class, there is some preliminary evidence that hydrophobic sites may be involved in the formation of the binary complexes between non-native proteins and chaperonins (24). Nevertheless, polypeptides associated with GroEL appear to maintain a greater amount of secondary structure relative to those bound to DnaK (25,38,39). Hence, there is no guarantee that these molecular chaperones function in a similar fashion.
In this paper, we have attempted to shed some light on the mechanism of action of bacterial chaperonins by comparing and contrasting the behavior of wild type GroEL and a single amino acid substitution mutant, GroEL140. Because of the peculiar "double doughnut" organization of the GroEL protein (6), it is likely that even minor changes in the protomers induce more pronounced changes on the structure and function of the assembled molecule.
We first tested GroEL140 for its ability to hydrolyze ATP in a magnesium-potassium-dependent manner since a weak ATPase activity requiring these ions appears to be a typical characteristic of chaperonins. Fig. 1 shows that, like GroEL, GroELl4O was fully dependent upon added potassium to achieve ATP hydrolysis. However, the ATPase activity of the mutant was reproducibly 1.5-fold lower compared to that of the wild type. Although we cannot rule out the possibility that this small decrease in activity was related to the presence of inactive enzyme in the purified GroEL14O stock solution, we believe that this change was a direct result of the mutation since significant structural modifications were observed in GroEL140 by protease accessibility experiments (Fig. 2, see  below). However, the GroEL140 mutation is probably not located in a region essential for ATP hydrolysis since the mutant chaperonin retained ATPase activity.
Non-native proteins forming binary complexes with GroEL can usually be discharged in a biologically active conformation by addition of ATP and GroES (16,20,21,[23][24][25][26][27]30). However, in some instances, non-hydrolyzable ATP analogs have been shown to achieve the same result albeit with a lower efficiency (24,26,27). ADP or the adenine nucleotides AMP-PNP and ATPrS were able to inhibit the ATPase activity of both GroEL and GroEL140, suggesting that these nucleotides compete with ATP for the ATP-binding site. Surprisingly, even though a 100-fold molar excess of nucleotides over ATP was used in these experiments, full inhibition was only observed with ATPrS. Hence, it is likely that the latter nucleotide possesses a higher affinity for GroEL. This result is consistent with the fact that ATP+ is more efficient than AMP-PNP at discharging GroEL-bound dihydrofolate reductase (26) as well as other model proteins? It should also be noted that ADP inhibited the ATPase activity of both GroEL and GroEL14O better than AMP-PNP. This implies that ADP retains a reasonably high affinity for the ATP-binding site and that an additional event (possibly mediated by GroES) may be necessary to release ADP and allow ATP binding to take place.
Several models have been proposed to describe the mode of action of chaperonins (9,40). They usually include a step where the conformation of GroEL changes in order to drive the release of the bound protein. An attractive signal for the induction of such structural modifications is the binding and/ or hydrolysis of ATP. In fact, evidence for ATP-mediated conformational changes in DnaK (35) and BiP (41) has been presented. Sensitivity to proteolytic degradation can provide information of the conformational changes undergone by a protein if protease cleavage sites become more accessible or are buried during the process. While purified GroEL was essentially resistant to proteolytic degradation by trypsin under our experimental conditions, the half-life of GroEL140 was significantly shorter, suggesting that the two proteins adopted different conformations. Since neither the migration pattern on non-denaturing gels or size exclusion columns, nor the ATPase activity, nor the ability to bind unfolded polypeptides were significantly affected, we concluded that GroEL140 could assemble in a conformation that was similar to that of the wild type but was probably less tightly packed as a result of the mutation. The presence of ATP, ATP analogs, or ADP made both proteins much more susceptible to trypsin hydrolysis. Therefore, the mere binding of nucleotides to GroEL induces a topological change in the molecule, although we cannot rule out that a subsequent conformational change is caused by ATP hydrolysis. This observation is consistent with the results of Kassenbrock and Kelly (41) who determined that nucleotide binding but not hydrolysis alter the conformation of BiP. In contrast, DnaK conformational changes seem to be dependent upon ATP hydrolysis (35). Interestingly, the binding of GroES did not affect the trypsin digestion pattern of GroEL suggesting that (i) either the binding of the cochaperonin did not impart a significant conformational change in the GroEL-ATP(ADP) complex or (ii) that the residues affected by the binding were not recognized or accessible to trypsin.
The poor efficiency of Rbu-Pz-carboxylase refolding by GroEL140 was shown to be related to a suboptimal interaction between the mutant chaperonin and GroES. Obviously, the mutation affected the affinity of GroES for GroEL140 since a stable complex between these proteins could not be isolated by size exclusion chromatography. However, the interaction between the two chaperonins was not completely abolished since (i) the presence of GroES allowed the formation of a Lz Rbu-Pz-carboxylase dimer as judged by gel filtration chromatography, and (ii) increasing the contact time between GroES and GroEL140-Rbu-Pz-carboxylase or the molar excess of GroES similarly enhanced the recovery of biologically active Rbu-Pz-carboxylase. In agreement with the in uitro results, in vivo pulse-chase experiments showed that overexpressed Rbu-Pz-carboxylase remained associated with Gro-EL140 about twice as long compared with GroEL. The magnitude of the effect is comparable to the 6-7-fold lower yield of refolded Rbu-Pz-carboxylase obtained by in vitro experiments involving wild type and mutant chaperonins. Nevertheless, release of labeled Rbu-Pz-carboxylase bound to either GroEL or GroEL140 is completed within a few minutes in vivo, while the process is slower in uitro. This raises the possibility that cellular factors other than GroES could be necessary to achieve optimal turnover of polypeptides associated with chaperonins.