GroEL locked in a closed conformation by an interdomain cross-link can bind ATP and polypeptide but cannot process further reaction steps.

It has been believed that when GroEL binds to GroES its apical domain moves upward and outward. To inhibit this “opening” movement, its equatorial and apical domains were cross-linked through a disulfide bond between mutationally introduced cysteine residues at the positions of Asp-83 and Lys-327. To avoid possible undesired cross-linking, we at first prepared a mutant GroEL (GroELNC; Cys-138 → Ser, Cys-458 → Ser, Cys-519 → Ser) in which all cysteine residues in wild-type GroEL were replaced by serine residues. GroELNC was fully functional as a chaperonin. We then introduced the above two point mutations into GroELNC to generate a mutant (GroELAEX; Cys-138 → Ser, Cys-458 → Ser, Cys-519 → Ser and Asp-83 → Cys, Lys-327 → Cys). Oxidized GroELAEX, which is locked in a “closed” conformation by an interdomain disulfide bond, can bind 6-7 mol of ATP, which remain bound without hydrolysis. This ATP-bound, oxidized GroELAEX can bind the stably nonnative substrate protein isopropylmalate dehydrogenase, whereas the nucleotide-free oxidized GroELAEX binds it with a weaker affinity. However, oxidized GroELAEX fails to process further reaction steps such as ATP hydrolysis, binding of GroES, dissociation of substrate protein from GroEL, and facilitating protein folding. When disulfide bonds in oxidized GroELAEX are reduced, GroELAEX exerts the ability to process all the reactions just as GroELNC and wild-type GroEL. Indications from these results are: hydrolysis of ATP may require opening movement of the apical domain; GroES binds to an open form of GroEL; and substrate polypeptide is released from GroEL coupled with either ATP hydrolysis or opening movement of the apical domain.

arranged in a dome shape (8). GroES binds to one end of the GroEL cylinder in the presence of ADP (or ATP), and it dissociates from GroEL during every ATPase cycle (9). Cryo-electron micrographs showed that the apical domain of GroEL in the GroEL-GroES complex undergoes an upward and outward movement as if on hinges, and the cleft between apical and equatorial domains opens (10). As a result of this "opening" movement of apical domains, the volume of the central cavity of the GroEL ring in contact with GroES becomes 2-fold greater than unliganded GroEL (10,11). Critical importance of an opening of apical domains has been further demonstrated by the recent finding that substrate polypeptides can be captured in the central cavity of GroEL underneath GroES (cis complex) and can fold to native structure in an ATP-dependent (or ADPdependent in some cases) manner (11)(12)(13).
Here, we have generated a mutant GroEL in which apical and equatorial domains can be cross-linked in a reversible manner and examined which step in the reaction cycle is dependent on the opening movement of the apical domain. Our results indicate that ATP hydrolysis, release of substrate polypeptide from GroEL, and GroES binding to GroEL cannot occur when opening of apical domains is inhibited.

Construction of Plasmids for Mutant
GroELs-A GroEL-expressing plasmid, pKY206, was a kind gift from Dr. K. Ito (Kyoto University) (14). For mutagenesis, a 1.7-kilobase MunI-SmaI fragment including the groEL gene was isolated from pKY206 and inserted into pTDT-7 (15) in EcoRI-SmaI sites. Single-stranded DNA of the plasmid pLWT1 was obtained by infecting E. coli CJ236 cells with helper phage M13K07 (Pharmacia Biotech). At first, cysteine residues of GroEL at positions 138, 458, and 519 were all replaced with serine residues by site-directed mutagenesis using oligonucleotides GAGTCAGAGCTCGGTACGGACA for Cys-138, ACGGTTCTTCTCCGGAGTTCAATAC for Cys-458, and GGTAACCATGGATTCGGTGGT for Cys-519 (16). The generated plasmid (pLNC1) containing a gene of the cysteine-less GroEL mutant (GroEL NC ) 1 was then used to introduce two cysteine residues at new positions by replacing Asp-83 and Lys-327. Oligonucleotides GTCGCT-GCAGCGCAATTGGCTTTAGAGGCA and AGTGGTGGTGTCGCAAT-TGATCACAACACGT were used to introduce mutations and a plasmid (pT7AEX) carrying a gene of the desired mutant GroEL (GroEL AEX ) was obtained. GroES-expressing plasmid pNM7626 was constructed as follows. A BstEII fragment (780 base pairs) containing the groES gene was isolated from pKY206 and treated by Klenow fragment. This BstEII * 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.
‡ To whom correspondence should be addressed. Fax: 81-45-924-5277; E-mail: myoshida@res.titech.ac.jp. fragment was inserted into pUC118. To attach a sulfhydryl-specific fluorescent probe, a cysteine residue was added at the C terminus of GroES (GroES 98C ). An oligonucleotide TCGTGCGCGGAGAATTCTTA-GCACGCTTCAACAA was used to generate this mutation and a plasmid pUCESC was obtained. The nucleotide sequences of the forgoing DNA probes were verified by dideoxy chain termination method using fluorescent DNA sequencer (ALF TM DNA sequencer II, Pharmacia).
Purification of Wild-type and Mutant GroELs-Wild-type GroEL (GroEL WT ) was purified from E. coli JM109 cells bearing the plasmids pKY206. GroEL NC and GroEL AEX were purified from E. coli BL21 (DE3) cells bearing the plasmids pLNC1 and pT7AEX, respectively. Wild-type GroES and GroES 98C were purified from E. coli JM109 cells bearing the plasmids pNM7626 and pUCESC, respectively. All cells were grown in 2 ϫ YT medium in the presence of 100 g/ml ampicillin at 37°C. When A 600 amounted to 0.5, 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside was added into the medium, and the culture was continued for another 12 h at 30°C. Collected cells were sonicated in 25 mM Tris-HCl, pH 7.5, containing 1 mM EDTA and 1 mM dithiothreitol (DTT), and the crude lysate was clarified by centrifugation at 40,000 rpm for 30 min at 4°C. In the case of GroEL purification, supernatant fractions were centrifuged in a sucrose gradient (10 -30%) in 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 1 mM DTT with a Hitachi P45AT-466 rotor at 21,000 rpm for 18 h at 4°C. The GroEL-enriched fractions of 25-30% sucrose were loaded to a Butyl-Toyopearl 650 column (Tosoh) equilibrated with 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM DTT, and a 10% saturation of ammonium sulfate, and the column was washed with the same buffer. Concentration of ammonium sulfate in the elution buffer, 25 mM Tris-HCl, pH 7.5, and 1 mM EDTA, was decreased linearly, and GroEL was eluted at 0% ammonium sulfate. For purification of GroES, supernatant fractions of sonicated lysates were chromatographed on a DEAE-Sephacel column (Pharmacia) equilibrated with 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 1 mM DTT. A linear gradient of NaCl (0 -0.5 M) was applied to the column, and GroES was eluted at 0.3 M NaCl. The fractions containing GroES were loaded to a Butyl-Toyopearl 650 column equilibrated with 25 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 1 mM DTT, and a 15% saturation of ammonium sulfate. The column was washed with 25 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , and a 15% saturation of ammonium sulfate, and the concentration of ammonium sulfate in the elution buffer was decreased linearly. GroES was eluted at an 8% saturation of ammonium sulfate. Butyl-Toyopearl 650 column chromatography was repeated once more. Behaviors of mutant GroEL and GroES during purification procedures were the same as those of corresponding wild-type ones. Purified GroEL and GroES were stored at 4°C as a suspension in 25 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , and 65% ammonium sulfate.
Folding Assay of Substrate Protein-Isopropylmalate dehydrogenase (IPMDH) from Thermus thermophilus (0.32 mg/ml) was denatured in 6 M guanidine HCl and diluted into a 120-fold dilution buffer (100 mM potassium phosphate, pH 7.8, and 1 mM MgCl 2 ) containing, when indicated, 10 mM DTT, 1 mM nucleotides, 0.3 M GroELs, 2 and 0.6 M GroES. The dilution buffer was preincubated for 10 min at 37°C prior to addition of denatured IPMDH. A final concentration of guanidine HCl was 50 mM, and this concentration is considered to have only little effect on the stability of GroEL and GroES in the protein-folding assay (17). The mixtures were incubated for 60 min at 37°C, and an aliquot was injected into the assay solution, which contained 100 mM potassium phosphate, pH 7.8, 1 mM MgCl 2 , 1.0 M KCl, 0.9 mM NAD ϩ , and 0.4 mM (2R*,3S*)-3-isopropylmalic acid. The increasing rate of absorbance at 340 nm was monitored at 60°C, and the activity of IPMDH was expressed as a percentage of that of the same amount of native IPMDH.
ATPase Assay-ATPase activities were assayed as described using malachite green to measure the amount of produced inorganic phosphate (18). The reaction was started by the addition of ATP (final concentration, 2 mM) to the assay solution containing 50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 50 mM KCl, 0.1 M GroEL, and, when indicated, 10 mM DTT and 0.2.M GroES. The assay solution was preincubated for 10 min at 25°C prior to addition of ATP. The reaction was terminated after a 20-min incubation at 25°C by addition of perchloric acid, and precipitated proteins were removed by centrifugation. The supernatant was reacted with a malachite green reagent, and absorbance at 630 nm was measured. One unit of activity is defined as the activity that hydrolyzes 1 mol of ATP/min.
Binding of Nucleotides to GroEL-Centrifuge gel filtration was used to quantify GroEL-bound ATP and ADP (19,20). GroEL (2.5 M) was incubated for 30 min at 25°C in 100 l of the solution containing 50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 50 mM KCl, 0.2 mM ATP, and, when indicated, 10 mM DTT and 5 M GroES. After cooled on ice for 1 min, the sample solution was loaded on a precentrifuged Sephadex G-50 (fine) column (Pharmacia), which was eqilibrated with 50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , and 50 mM KCl. After loading the sample solution, the column was centrifuged, and perchloric acid was added to the eluate. The precipitated proteins were removed by centrifugation, and the amount of nucleotides recovered in the supernatant were analyzed with HPLC using a COSMOSIL 5C-AR18 packed column (Nakalai Tesque Inc.) (21). The column was eluted with 20 mM phosphate, pH 6.9, 5 mM tetra-n-butyl ammonium hydrogen sulfate, and 20% methanol, and absorbance at 260 nm was monitored. The amount of nucleotide was calculated from the peak area of authentic nucleotide.
Binding of GroES to GroEL-GroES 98C was labeled by a fluorescent sulfhydryl-specific reagent, N-(7-dimethylamino-4-methyl coumarinyl) maleimide (22,23). N-(7-dimethylamino-4-methyl coumarinyl)maleimide (1 M) was mixed with 1 M GroES 98C in 50 mM sodium phosphate, pH 6.9, at 4°C for 12 h. Free N-(7-dimethylamino-4-methyl coumarinyl)maleimide was removed by gel filtration, and the fraction of labeled GroES 98C (GroES FL ) was concentrated with Centricon-30 (Amicon Inc.). The mixture of GroES FL and GroEL, each 0.5 M, was incubated at 25°C for 30 min with a 0.2 mM concentration of either ADP or ATP in 50 mM potassium phosphate, pH 6.9, and 5 mM MgCl 2 . DTT (10 mM) was included in the solution when indicated. After the incubation, these samples were loaded on a gel filtration HPLC column (G3000SWXL, Tosho) equilibrated with 50 mM potassium phosphate, pH 6.9, and 5 mM MgCl 2 . The column was eluted with the same buffer at a flow rate 0.5 ml/min. Elution was monitored with absorbance at 280 nm for GroEL and fluorescence at 470 nm for GroES FL (excitation light at 390 nm).
Other Analytical Methods-Proteins were analyzed by polyacrylamide gel electrophoresis (PAGE) on a 15 or 6% polyacrylamide gel in the presence or absence of 0.1% SDS, respectively (24). GroELs were preincubated for 10 min in 50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , and 50 mM KCl in the absence or presence of 10 mM DTT. An aliquot (2 g of protein) was mixed with the sample solutions with or without SDS and loaded on gels. The reducing reagent was omitted from the sample solutions, the running buffer for electrophoresis, and gels. Gels were stained by Coomassie Blue R-250. The concentrations of sulfhydryl groups were determined according to the method of Ellman (25). Wildtype and mutant GroELs (500 g each) were dissolved in 1 ml of 0.1 M Tris-HCl, pH 8.0, 1 mM EDTA, and, when indicated, 10 mM DTT and incubated for 30 min at 25°C. After the incubation, sample solutions were loaded on a Sephadex G-50 (fine) column, which was equilibrated with 0.1 M Tris-HCl, pH 8.0, and 1 mM EDTA, and subjected to centrifuge-gel filtration. Eluted solutions were immediately denatured in 6 M guanidine HCl, and 100 l of 10 mM 5,5Ј-dithiobis-(2-nitrobenzoic acid) (25) dissolved in 50 mM sodium phosphate, pH 7.0, was added. Absorbance at 412 nm was measured after 5 min. The value 13,380 M Ϫ1 cm Ϫ1 , used as molar absorption coefficient of 2-nitro-5-mercaptobenzoic acid in 6 M guanidine HCl, was used for calculation (26). Protein concentrations were assayed by the method of Bradford (27) with bovine serum albumin as a standard.

RESULTS
GroEL AEX Mutant-In the crystal structure of GroEL, Lys-327 in the apical domain is very close to Asp-83 in the equatorial domain (28), and we replaced these residues by cysteine residues, intending the formation of a disulfide bond between two domains (Fig. 1). Wild-type GroEL has three cysteine residues, Cys-138, Cys-458, and Cys-519, and none of them is essential for function of GroEL (29). We at first replaced them by serine residues to avoid cross-linking between undesired pairs of cysteines. GroEL NC showed apparently normal activity of GroEL, as described later. Then, two mutations, Asp-83 3 Cys and Lys-327 3 Cys, were introduced into GroEL NC , and GroEL AEX was generated. Mutant GroELs were well expressed in E. coli cells (about 50 mg/liter of culture medium) and purified by the same procedures as used for the purification of GroEL WT . Circular dichroism spectra in the absence and presence of 5 mM DTT were very similar to those of the wild-type GroEL (data not shown), indicating that the secondary structure of GroEL AEX was not changed significantly from that of the wild-type GroEL.
Numbers of total free sulfhydryl groups in GroELs were estimated from titration by 5,5Ј-dithiobis-(2-nitrobenzoic acid) (25). To expose all sulfhydryl groups, GroELs were denatured prior to the titration. The values 3.0 and 0.09 mol/mol of GroEL protomer were obtained for GroEL WT and GroEL NC , respectively. The sulfhydryl groups in the purified preparation of GroEL AEX , which had been stored in the buffer without a reducing reagent, was already fully oxidized by air oxygen, since the value 0.08 mol/mol of GroEL protomer was obtained. After the purified GroEL AEX was exposed to 10 mM DTT for 5 min at 25°C, the number of sulfhydryl groups in GroEL AEX increased up to 2.3 mol/mol of GroEL protomer. The reduced GroEL AEX was reoxidized automatically by air oxygen in the buffer without a reducing reagent within 5 min at 25°C, and the number of sulfhydryl groups in GroEL AEX dropped to 0.1 mol/mol of GroEL protomer. This rapid reoxidation may be a reflection of the close location of two cysteine residues in the native structure of the reduced GroEL AEX . Consistently, reoxidation of denatured, reduced GroEL AEX occurred only very slowly.
Cross-Linking between Apical and Equatorial Domains- Fig. 2A shows SDS-PAGE analysis of wild-type and mutant GroELs preincubated in the buffer with or without DTT. When the purified GroELs, which had been stored in the buffer without a reducing reagent, were preincubated in the buffer without DTT and electrophoresed, it was noticed that only Gro-EL AEX was electrophoresed at a slower mobility than GroEL WT and GroEL NC (Fig. 2A, left three lanes). The extent of the band shift of GroEL AEX corresponded to about a 6-kDa difference of apparent molecular size and was too small to assign this band as a GroEL AEX dimer (or oligomer) cross-linked through intersubunit disulfide bonds. Thus, the disulfide bond is an interdomain cross-link in the same GroEL subunit but not an intersubunit cross-link between two or more GroEL subunits. Fig.  2A also shows that GroEL AEX was not contaminated by Gro-EL WT , since it appeared as a single band in electrophoresis, and there was no detectable protein band at the position of GroEL WT . When GroEL AEX was preincubated with 10 mM DTT, the band shift of GroEL AEX was no longer observed, and all of the GroELs showed apparently the same electrophoretic mobility ( Fig. 2A, right three lanes). Standing DTT-treated Gro-EL AEX in the buffer without reducing reagent at 25°C for 5 min resulted in complete recurrence of the shifted electrophoretic mobility (data not shown).
The same samples were analyzed with native PAGE (PAGE in the absence of SDS; Fig. 2B). The purified GroEL AEX was electrophoresed at slightly faster mobility than GroEL WT and GroEL NC (Fig. 2B, left three lanes). When GroEL AEX was treated with 10 mM DTT, its electrophoretic mobility became the same as that of GroEL WT (Fig. 2B, right three lanes). The same electrophoretic mobility of mutant GroELs and GroEL WT is an indication of the normal tetradecameric structure of mutant GroELs. This was confirmed from the same retention time of mutant GroELs in gel filtration HPLC as that of GroEL WT , as described later. Combining results of sulfhydryl titration and PAGE, we concluded that Cys-83 and Cys-327 in the same GroEL AEX subunit were cross-linked through a disulfide bond in a reversible manner. The purified preparations of GroEL AEX stored in the solution without a reducing reagent were already oxidized by air oxygen (oxidized GroEL AEX ), but its disulfide bond was readily reduced to form free sulfhydryl groups (reduced GroEL AEX ) by incubation of oxidized GroEL AEX with 10 mM DTT for 5 min at 25°C. We then compared functional properties of oxidized and reduced GroEL AEX .
Oxidized GroEL AEX Can Bind a Substrate Polypeptide but Cannot Release It-Interaction of mutant GroELs with a substrate polypeptide was tested using IPMDH from T. thermophilus as a substrate protein (30). IPMDH is a homodimer of 37-kDa subunits that has no cysteine residue, and IPMDH activity is not influenced whether a reducing reagent is present (31). Under the condition of the experiment (at 37°C), IPMDH denatured in 6 M guanidine HCl can fold spontaneously on diluting into the dilution buffer, and we measured the arrest of folding by GroEL and the relief from the arrest by ATP. In the absence of DTT (Fig. 3A), GroEL NC and GroEL WT gave the same results. They captured unfolded IPMDH arrested its folding, and no IPMDH activity was recovered. When ATP was included in the dilution buffer, IPMDH was relieved from the arrest, and its activity was recovered to the magnitude (65%) similar to that attained by spontaneous folding (60%). ATP was not substitutable by ADP and AMP-PNP (1 mM each) to induce the relief from the arrest (not shown). Further inclusion of GroES in the dilution buffer improved the yield of recovered IPMDH activities by about 10%.
The results of oxidized GroEL AEX were very different from those of GroEL NC and GroEL WT (Fig. 3A). At first, folding arrest by oxidized GroEL AEX was incomplete, and about 30% of IPMDH activity, about one-half of the yield of spontaneous folding, was recovered. Interestingly, when ATP was present, the folding arrest became complete, and, in turn, relief from the arrest by ATP was not observed. ADP and AMP-PNP (1 mM

FIG. 2. Electrophoretic discrimination of oxidized GroEL AEX .
A, 13% SDS-PAGE of GroEL WT , GroEL NC , and GroEL AEX preincubated in the absence (left three lanes) or presence (right three lanes) of 10 mM DTT. GroELs (2 g each) were analyzed. The running buffer and gels for electrophoresis did not contain a reducing reagent. B, 6% native PAGE of GroEL WT , GroEL NC , and GroEL AEX preincubated in the absence (left three lanes) or presence (right three lanes) of 10 mM DTT. GroELs (2 g each) were analyzed. The running buffer and gels for electrophoresis did not contain a reducing reagent. Other detailed experimental conditions are described under "Materials and Methods." each) were also effective to make the arrest complete (not shown). Thus, oxidized GroEL AEX can bind substrate protein weakly (ϪATP) or tightly (ϩATP), and tightly bound substrate protein cannot dissociate from oxidized GroEL AEX even in the presence of ATP. Slight recovery of IPMDH activity (8%) was observed by further inclusion of GroES in the dilution buffer. In the presence of DTT (Fig. 3B), GroEL AEX , as well as GroEL NC , functioned normally, just as GroEL WT . The addition of DTT to the preformed arrested complex of oxidized GroEL (oxidized GroEL-IPMDH-ATP) triggered the start of recovery of IPMDH activity, which finally reached 60%. This ensures that the arrested IPMDH is an active intermediate in the folding. It is clear that the cross-link through a disulfide bond is responsible for the altered interactions of oxidized GroEL AEX with substrate polypeptide.
Oxidized GroEL AEX Cannot Hydrolyze ATP, Although It Can Bind ATP- Fig. 4 shows ATPase activities of wild-type and mutant GroELs. GroEL NC exhibited almost the same ATPase activity as that of GroEL WT under any conditions tested. As reported for GroEL WT (32,33), GroES inhibited ATPase activity of GroEL NC by about 50%. On the contrary, in the absence of DTT, oxidized GroEL AEX had no ATPase activity no matter whether GroES was present (Fig. 4A). The inhibited ATPase activity of oxidized GroEL AEX is merely due to the disulfide cross-link, since reduced GroEL AEX in the presence of DTT showed almost the same ATPase activity as that of GroEL WT , and suppression of ATPase activity by GroES was also observed for reduced GroEL AEX (Fig. 4B).
Loss of ATPase activity of oxidized GroEL AEX can be caused from impairment of either of two steps: binding of ATP or hydrolysis of bound ATP. Analysis of bound nucleotides after incubation of GroELs with ATP enabled us to distinguish these two possibilities. As reported already (20), GroEL WT alone did not bind nucleotide tightly, and only very small amount of bound nucleotide was detected (Fig. 5A). When GroES was present, GroEL WT retained 6.3 mol of ADP/mol of GroEL. DTT did not have an effect on the nucleotide binding properties of GroEL WT (Fig. 5, compare A and B). In any case, the amount of bound ATP on GroEL WT was very little. Nucleotide binding properties of GroEL NC were similar to those of GroEL WT , whereas the amount of bound ADP in the presence of GroES was somewhat larger than that in the presence of GroEL WT . On the contrary, oxidized GroEL AEX stably bound 6.0 (ϪGroES) and 6.3 (ϩGroES) mol of ATP/mol of GroEL (Fig. 5A). It also had 1.9 (ϪGroES) and 3.9 (ϩGroES) mol of bound ADP. As expected, when oxidized GroEL AEX was reduced with DTT, nucleotide binding properties became similar to those of Gro-EL WT ; bound ATP almost disappeared both in the absence and presence of GroES, and ADP was detected as a bound nucleotide species only in the presence of GroES (Fig. 5B). Thus, it is clear that oxidized GroEL AEX can bind ATP but cannot hydrolyze it. A comment should be added that a possibility of a single turnover of ATP hydrolysis by oxidized GroEL AEX has not been eliminated at the moment, since we did not know whether bound ADP in oxidized GroEL AEX was derived from catalyzed hydrolysis or a trace amount of contaminated ADP in the ATP solution.
Oxidized GroEL AEX Cannot Bind GroES-To know whether oxidized GroEL AEX can bind GroES, we have developed a convenient method to measure GroES binding. Wild-type GroES does not contain a cysteine residue, and we generated a mutant in which an additional cysteine residue was attached to the C terminus of GroES (GroES 98C ). GroES 98C was labeled with N-(7-dimethylamino-4-methyl coumarinyl)maleimide (22,23), a sulfhydryl group-specific fluorescent probe, and the labeled GroES (GroES FL ) was used for HPLC gel filtration assay for GroES binding. The C terminus of GroES is at the outside surface of the dome-shaped GroES (8,34) and labeling of the added cysteine residue may not interfere with the function of GroES. Indeed, GroES FL was fully functional in GroEL-and GroES-dependent rhodanese folding when it was used instead of wild-type GroES (data not shown). GroELs were incubated with a stoichiometric amount of GroES FL in the presence or absence of DTT and were analyzed with HPLC by monitoring absorbance at 280 nm (not shown) and fluorescence emission at 470 nm (Fig. 6). Mutant GroELs were eluted out at the same retention time as that of GroEL WT , 12.6 min, confirming their normal tetradecameric structure. Unliganded GroES FL was eluted at 17.6 min, but it was eluted at 12.6 min when associated with GroEL. The incubation mixtures contained either 0.2 mM ATP or ADP, but both nucleotides always gave the same effect on the binding of GroES FL to GroELs under the conditions tested. Just as observed for GroEL WT , GroEL NC was able to bind GroES FL in the absence and presence of DTT. Reduced GroEL AEX in 10 mM DTT was also able to bind GroES FL , but in the absence of DTT, oxidized GroEL AEX was unable to bind GroES FL in the presence of either ATP or ADP.

DISCUSSION
Locking the Apical Domain in a Closed Conformation-Cryoelectron micrographs showed that the apical domain of GroEL opens upward and outward when the GroEL-GroES complex is formed (10). Based on this contention, the x-ray crystal structure of GroEL (3) has been believed to represent a "closed" form of GroEL. Oxidized GroEL AEX has an interdomain cross-link between apical and equatorial domains within the same GroEL protomer (Fig. 1). According to the crystal structure of GroEL (28), the distance between C␣ carbons of Lys-327 in the apical domain and Asp-83 in the equatorial domain of wild-type GroEL is ϳ9 Å. The maximal length between the two C␣ carbons when they are connected through a disulfide bond is calculated to be ϳ7 Å. Therefore, once GroEL AEX is oxidized, GroEL should be locked in a closed form, and the apical domains cannot open upward and outward. Since GroEL AEX is fully functional when a disulfide bond is reduced, functional defects observed for oxidized GroEL AEX are not due to some irreversible change of conformation, and one can consider oxidized GroEL AEX as a model of the GroEL locked in a closed conformation.
GroEL in a Closed Conformation Can Bind ATP and a Substrate Polypeptide-The overall conformation of the ATP␥Sbound form of GroEL resembles the nucleotide-free structure (35), that is, a closed form. Therefore, it is not surprising that oxidized GroEL AEX can bind ATP (Fig. 5A). Opening movement of the apical domain is not necessary for any microscopic steps in acceptance and retention of ATP at nucleotide binding sites. Our results also indicate that access of ATP to the nucleotide binding sites from medium solution is not sterically hindered by the cross-link in oxidized GroEL AEX . The nucleotide binding site of GroEL is located at the side shoulder surface of the equatorial domain (35). From a careful observation of the GroEL crystal structure (28), it seems that ATP can approach the nucleotide binding site from other directions 3  tained 6 -7 mol of ATP/mol of GroEL. Because of positive cooperativity among GroEL protomers in the same ring (33,36), bound ATPs of oxidized GroEL AEX are probably on the GroEL protomers in the same ring rather than randomly distributed among 14 protomers.
Oxidized GroEL AEX can bind a substrate polypeptide tightly in the presence of ATP (Fig. 3). This implies that a substrate polypeptide can bind to GroEL in a closed form. Since ADP and AMP-PNP are also effective for the tight binding, occupation of nucleotide binding sites of one ring may induce stabilization of the binding of a polypeptide at the same (or opposite) ring. However, the real reason why ATP is required for the tight binding of a polypeptide to oxidized GroEL AEX is not known.
Opening of the Apical Domain May Be Necessary for ATP Hydrolysis-Oxidized GroEL AEX cannot hydrolyze ATP (Fig.  4A), indicating that opening movement is required for catalysis. This is an unexpected finding, because ATP␥S in the crystal structure interacts with residues only within the equatorial domain (35), and ATP hydrolysis appears to be more or less independent from movement of the apical domain. There is a possibility that residues, important for catalysis but not yet recognized in the known structure of a closed form of GroEL, come to the catalytic site when the apical domain moves upward and outward. Another possible explanation is that the distance between residues 83 and 327 becomes 2 Å shorter by cross-linking, and residues 87-91, which interact with the ␤and ␥-phosphates of ATP, are forced to dislocate from optimal positions for catalysis.
Polypeptide Release and GroES Binding-As oxidized Gro-EL AEX cannot hydrolyze ATP, all the reactions coupled with, or following, ATP hydrolysis should be damaged in oxidized Gro-EL AEX . Therefore, the observed functional defects of oxidized GroEL AEX can be direct or indirect consequences of the impaired ATPase activity. However, if a certain reaction is not dependent on ATP hydrolysis and still is not processed by oxidized GroEL AEX , the reaction may be dependent on opening movement of the apical domain. It is known that conditions required for the release of a substrate polypeptide from GroEL are varied from one substrate to another. Although release of some polypeptides from GroEL can be triggered only after the addition of ATP, some polypeptides are released by ATP hydrolysis or just binding of adenine nucleotides (37)(38)(39). In our experiments, release of nonnative IPMDH from GroEL WT appears to be coupled with ATP hydrolysis but not with mere ATP (or ADP) binding, since it is triggered by the addition of ATP but not by AMP-PNP or ADP. Given that it is really the case, persistent retention of nonnative IPMDH on oxidized Gro-EL AEX in the presence of ATP can be interpreted as a consequence of defective ATPase of oxidized GroEL AEX . The fact that, just as observed for oxidized GroEL AEX , mutant GroELs with defective ATPase, such as D87N, I150E, S151V, A383E, A405E, and A406E, cannot or can only poorly mediate ATPtriggered polypeptide release (5) agrees with the above contention. The manner by which ATP hydrolysis induces polypeptide release from GroEL is not known, but a possibility remains that the polypeptide captured by GroEL can dissociate from GroEL only when apical domains are in an open form, which is achieved by ATP hydrolysis (40).
Binding of GroES to GroEL occurs in the presence of adenine nucleotides, among which ADP is most effective (20). The dissociation constant of the GroEL-GroES complex in the presence of 0.2 mM ADP was reported to be 0.2 nM, whereas that in the presence of 2 mM ATP was 17 nM (41). AMP-PNP at 2.5 mM is nearly as effective as 0.2 mM ADP. Therefore, binding of GroES to GroEL is not necessarily coupled with ATP hydrolysis; rather, it has been thought that ATP stimulates the dissocia-tion of GroES from GroEL under physiological conditions. Then, the disability of oxidized GroEL AEX to bind GroES is not attributable to its defective ATPase activity but is likely due to its locked conformation in a closed form. In other words, opening movement of apical domains is necessary for GroEL to bind GroES. Probably, the conformation of GroEL is in a dynamic equilibrium between open and closed forms in the presence of ADP, and GroES interacts exclusively with the open form of GroEL.