The Effect of groES on the groEL-dependent Assembly of Dodecameric Glutamine Synthetase in the Presence of ATP and ADP*

The yields of active dodecameric glutamine synthe- tase (GS) are significantly increased when in vitro folding is initiated in the presence of the Escherichia coli groE chaperonins and ATP (37 “C). To observe the effects of chaperonins and ATP on GS assembly, the GS assem- bly intermediates were separated by nondenaturing gel electrophoresis, visualized by Western analysis, and studied as a function of time. The form of GS that was initially released from groEL is monomeric. After the monomers formed dimers, active GS oligomers were as- sembled by the association of assembly competent dimers with higher order even-numbered oligomers un- til the dodecamer was formed. When ATP was added to the groEL.GS complex (no groES), a groEL-GS complex remained visible for up to 30 min after the renaturation was initiated. This slow disappearance of the groEL-GS complex is consistent with observed lags in both the GS activity regain profile and the assembly-dependent in- crease in GS tryptophan fluorescence. When groES was present, the addition of ATP resulted in the disappearance of observable complex at early sample times (e2 min). Concomitantly, the rates of the regain of GS activity and the GS-dependent increase in tryptophan fluo- rescence intensity showed substantial accelerations. These results indicate that groES facilitates GS assem- bly from groEL by inducing the rapid release of GS from groEL, which in turn

The yields of active dodecameric glutamine synthetase (GS) are significantly increased when in vitro folding is initiated in the presence of the Escherichia coli groE chaperonins and ATP (37 "C). To observe the effects of chaperonins and ATP on GS assembly, the GS assembly intermediates were separated by nondenaturing gel electrophoresis, visualized by Western analysis, and studied as a function of time. The form of GS that was initially released from groEL is monomeric. After the monomers formed dimers, active GS oligomers were assembled by the association of assembly competent dimers with higher order even-numbered oligomers until the dodecamer was formed. When ATP was added to the groEL.GS complex (no groES), a groEL-GS complex remained visible for up to 30 min after the renaturation was initiated. This slow disappearance of the groEL-GS complex is consistent with observed lags in both the GS activity regain profile and the assembly-dependent increase in GS tryptophan fluorescence. When groES was present, the addition of ATP resulted in the disappearance of observable complex at early sample times (e2 min). Concomitantly, the rates of the regain of GS activity and the GS-dependent increase in tryptophan fluorescence intensity showed substantial accelerations. These results indicate that groES facilitates GS assembly from groEL by inducing the rapid release of GS from groEL, which in turn increases the concentration of assembly competent GS monomers. In addition, groES can initiate renaturation of GS from the groEL-GS arrested complex in the presence of ADP. When chaperonin-dependent GS renaturation was initiated with ATP or ADP ( 2 2 mm), the rates were identical. Since ATP hydrolysis is not absolutely required, the combined binding energies of groES and ATP (or ADP) appear to be sufficient to weaken the binding affinity of groEL for GS subunits and facilitate the release and refolding of assembly competent GS monomers from groEL.
The Escherichia coli groE chaperonin system (groEL and groES) facilitates the folding and assembly of various oligomeric proteins both in vivo and in vitro (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). Experimental evidence suggests that the groE system directs protein folding *This project is supported in part by Grant BRSG SO7RRO5373 awarded by the Biomedical Research Support Grant Program Division of Research Resources, National Institutes of Health. This work was presented in part at the ASBMBBiophysical Society Meeting, Houston, T X , February [9][10][11][12][13]1992 and the 1992 FASEB Fall Symposia on "Molecular Chaperones," Keystone, CO. 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. and assembly reactions by largely preventing the formation of inappropriate aggregates (3,12). However, the mechanism of chaperonin-assisted protein folding and assembly remains unknown.
When the chaperonin system is present, the renaturation of active E. coli dodecameric glutamine synthetase (GS)' is substantially enhanced at physiological temperatures and high GS concentrations (6,7). Since GS and groEL are both from E. coli, the interactions observed in vitro may correlate with in vivo assembly reactions. Indeed, even dimeric ribulose bisphosphate and P22 tailspike protein, which are foreign to E. coli cells, have been shown to interact with the groE system both in vitro and in vivo (1,2,13,14). The groE chaperonins facilitated the folding of rhodanese from both unfolded protein in denaturants and from de novo synthesis of the enzyme using a n in vitro E. coli translation system (15).
The folding and assembly of tetrameric succinyl-CoA synthetase and P-galactosidase do not require the groE system either in vivo or in vitro (16,17). Thus, not all oligomeric proteins require the groE chaperonins to fold and assemble. To date, all proteins that interact with the groE system in vitro also appear to interact with these same chaperonins in vivo and this may be a general property.
There is evidence from in vivo experiments that GS can interact with chaperonins. GS subunits are transiently associated with resident chloroplast groEL-like chaperonins after import (18). Endogenous GS subunits are associated with groEL prior to purification, and the resulting complexes have the same physical properties as the exogenously prepared complexes presented herein. ' Although initial experiments with the isolated system demonstrated that groEL and ATP were sufficient to enhance GS refolding yields, the renaturation of GS showed a marked acceleration in formation when groES was present. Following the formation of a stable groEL.GS complex (renaturation arrest), GS renaturation exhibited a dramatic lag in the renaturation rate profile with the incomplete chaperonin system (groEL, MgATP, K+), whereas the lag was significantly shortened in the presence of the complete system (groEL, groES, MgATP, K+) (6). These differences in the renaturation rate profiles contain information about groES function. The fate of the groEL.GS complex can be resolved by nondenaturing gradient-pore electrophoresis (6) and visualized by Western analysis. This resolved complex and transient GS assembly intermediates were observed in a time-dependent manner in the absence or presence of groES.
The results presented here suggest that GS is released from the groEL complex as a monomer in the presence of ATP and groES. groES facilitates the release of the bound "arrested" GS conformer as evidenced by 1) the complete disappearance of an electrophoretically resolved complex, 2) the increase in the GS activity regain rate, and 3) the assembly-dependent increase in GS tryptophan fluorescence. Furthermore, in the presence of DOES, ADP can replace ATP to facilitate GS renaturation from stable groEL.GS complexes with no loss in active GS yields.

EXPERIMENTAL PROCEDURES
Materials-Glutamate, ATP, EDTA, and dithiothreitol were purchased from Sigma. Monopotassium ADP was purchased from Boehringer Mannheim. Ultra-pure urea, guanidine HC1, and Tris-HC1 were purchased from ICN Biochemical. Ultra-pure ammonium sulfate was purchased from Bethesda Research Laboratories. GS was purified from E. coli containing the overexpression plasmid (YMClO/pgln6) (19). Protein concentrations were determined by absorbance measurements at 290 nm using A,, of a 0.1% solution (1 cm) = 0.387 (20) after the base line was corrected for light scattering. GS activity was monitored using the y-glutamyl transferase assay method (21). The GS preparations contained approximately 0.7-1.5 adenylylated subunitddodecamer and possessed transferase activities of 120-130 unitdmg.
groE Purification-The E. coli chaperonins, groEL and groES, were isolated from lysates of cells containing an overexpression plasmid (gift of Dr. George Lorimer). These proteins were purified by using a combination of purification procedures described previously (3, 6). After isolating groEL from the Q-Sephrarose column, the protein was pooled by collecting those fractions which possessed the highest second derivative r value indicative of a lower tryptophan content. The ratio r = (A"(288 nm) -A(284 nm))/(A(294 f 1 nm) -A"(291 f 1 nm)), where A ( h ) is the second derivative absorbance at wavelength A (22). The pooled fractions were extensively dialyzed against 50 nm Tris-HC1, 0.5 nm EDTA, pH 7.5, and were incubated with Cibracon blue-agarose resin ("Gel blue, Bio-Rad) that had been equilibrated in the same buffer overnight at 4 "C on a gentle rocker. A large amount of groEL does not bind to the Cibracon blue column and is easily separated from the resin by washing with equilibration buffer (6). At this stage of the purification, the second derivative spectmm no longer contained the characteristic tryptophan indole contribution of a trough and peak centered around 291-292 and 293-295 nm, respectively. To obtain samples used for the tryptophan fluorescence experiments, 3 m MgATP and 50 nm KC1 was added to the groEL sample for 5 min at room temperature. The ATP-treated protein was applied onto a S-400 gel filtration column at 4 "C and eluted with buffer containing 50 nm Tris-HC1, 50 nm KC1,0.5 nm EDTA, pH 8.5. The collected protein was dialyzed against the appropriate buffer prior to experiments. A 0.1% solution of groEL in 6 M guanidine HCl had an absorption at 280 nm of 0.195. Normalized absorptions prior to ATP addition and after gel filtration were essentially the same, indicating that no bound nucleotide remained with groEL.
The purified groES had a small amount of tryptophan contamination as assessed by the fluorescence emission spectrum (excitation at 297 nm). However, when groES was added to groEL at equimolar concentrations, there'was little change in the residual Trp steady state emission spectrum. In addition, the final fluorescence product from refolded GS was virtually the same f groES (see Fig. 5B). Like purified groEL, a second derivative analysis of the uv absorption spectrum indicated that the aromatic fingerprint region did not show the characteristic trough (-291 n m ) and peak (-295 nm) contribution from the indole tryptophan (23). Instead, the second derivative spectrum has the characteristic peak trough pattern of tyrosine in the region from 270 to 300 nm. The groES samples show a single band on Coomassie-stained SDSpolyacrylamide gel electrophoresis.
Nucleotide Solutions-For the groEL-dependent GS refolding assays, stock ADP (Boehinger Mannheim) solutions (50-100 m, pH 7.1) were incubated at 37 "C in the presence of hexokinase (25 unitdml) and 5 nm glucose for 2 h to remove any residual ATP. The presence of ATP was tested by monitoring the ATP dependent GS activity inhibition that results from the GS catalyzed ATP-dependent phosphorylation of Lmethionine sulfoximine (MetSox). L-Methionine sulfoximine phosphate (MetSox-P) is a transition state analog to y-glutamylphosphate and completely inactivates GS with a binding constant of Ka > 10" (24). In addition, this sensitive enzymic assay was used to determine whether groEL or groES contained adenylate kinase as a contaminant. When hexokinasdglucose-treated ADP and MetSox were added to 0.5 w native GS oligomer in the presence of 1 p groEL and groES oligomers, no detectable decreases in GS activity were observed within the time of assay used for GS renaturation (2 h). ATP and AMP-PNP (Sigma) were used without further treatment or purification.
Antibody Production-Antibodies raised to native GS were prepared in sheep as described by Hohman and Stadtman (25). Antibodies raised against groEL were prepared in rabbits (26). Antibody exchange immunochemistry with groEL was performed as described by Hammarback and Vallee (27).
Denaturation and Renaturation of GS-The procedures used to denature and renature GS were previously described (6) with the following modifications. GS was exclusively denatured in 6 M guanidine HCl, 50 m Tris-HC1,l nm EDTA, and 0.5 nm dithiothreitol at 0 "C for 4-5 h at pH 7.5. All GS renaturation reactions were performed at 25 "C in dithiothreitol at pH 7.5. The protein concentrations used in these experiments was 0.5 w GS subunits and 1 w groEL and groES oligomers unless specified.
Native Electrophoretic Separation of GS Renaturation Reaction Products-Nondenaturing gradient-pore gel electrophoresis was used to examine the time-dependent changes of the protein species present during groEL-dependent GS renaturation. At specified times during renaturation, the protein species present were separated using a precast continuous 4-15 or %25% polyacrylamide gradient (Phastgel). Following formation, equal aliquots of the groEL.GS complex were put into separate Eppendorf tubes, and the renaturation reactions were initiated with 3-5 ~M A T P at staggered times at 37 "C. These timed samples were applied simultaneously onto the gradient gel carrying current using a Pharmacia Phast electrophoretic separation system. This method was used to rapidly separate (15-20 min) the various transient assembly intermediates of GS formed during the spontaneous and groEL-assisted renaturation of initially unfolded GS. Starting with previously unfolded GS monomers, no significant assembly could be observed when freshly mixed GS monomers (100 x diluted from denaturant) were immediately loaded onto the Phastgel system. In comparison with the large amount of monomer present, faint diffuse bands migrating in the same region of tetramers and hexamers could be detected using Western analysis ( Fig. U?, lane 1 ). Comparing these bands with those observed during the time-dependent chaperonin-assisted assembly of GS indicated that little if any assembly occurred during electrophoresis. This conclusion was also corroborated by assembly arrest experiments using the methodology outlined by Maurizi and Ginsburg (24) (see "Results").
Western Analysis-The gel was removed from the plastic backing, and the proteins were electroblotted onto nitrocellulose soaked in 12 nm Tris, 96 IMI glycine, 20% methanol, pH 8.5, using a Phast Transfer system. The proteins were transferred for 45 min at 20 V, 25 mA with constant power of 1 watt. Native dodecameric GS, groEL.GS, and the transient GS assembly intermediates were visualized by Western analysis using primary antibodies raised against GS and the appropriate secondary antibody linked with alkaline phosphatase (Pierce Chemical Co.). After transfer, the polyacrylamide gel was stained with Coomassie Blue to assess the amount of protein that remained on the gel.
Fluorescence Measurements-Since groEL and groES do not contain tryptophan (28), the time-dependent changes in GS tryptophan fluorescence could be monitored as the chaperonin-dependent assembly reaction proceeded. In order to enhance the GS tryptophan fluorescence changes, L-methionine sulfoximine and ATP were present during the assembly reaction (24, 29). The formation of L-methionine sulfoximine phosphate at the GS active sites increases the tryptophan fluorescence intensity of GS and serves as a spectral monitor of the formation of the GS active site (30).
Fluorescence Instrumental Parameters-The tryptophan fluorescence emission was monitored at 336 nm with the excitation monochromators set at 297 nm using a Hitachi 1050 spectrofluorometer. In the presence of 3 IMI ATP, the absorbance of the renaturation mixture at 297 nm was lower than 0.1, thus insuring no significant inner filter effects (31). Contributions from light scattering were kept at a minimum by setting polarizers at the magic angle configuration (vertical for excitation, 54.7 O from the vertical for emission). The excitation and emission slit widths were set at 5 nm. At these settings, the fluorescence appears to be sensitive to saturation effects so the fluorescence was monitored using 10-s exposure times at intervals between 30 and 60 s. The solution was stirred magnetically throughout the GS assembly reaction. groEL when the renaturation mixture was separated by nondenaturing gel electrophoresis (see Fig. 7 in Ref. 6). In order to identify the protein components present in the slower migrating protein band, the proteins were electroblotted onto a nitrocellulose membrane, probed with polyclonal antibodies raised against either E. coli GS or groEL, and visualized by Western blot analysis (Fig. 1). The Western analysis shows that the slower migrating band just above native groEL consists of a complex between groEL and GS subunits (Fig. 1). For groEL alone, the extraneous higher and lower bands observed in both the Coomassie Blue-stained and the Western analysis appears to result from aggregated and dissociated groEL. Affinity-purified antibodies to groEL, isolated using the antibody exchange method of Hammarback and Vallee (27), cross-reacted with all the species observed when groEL alone was run on a 4-15% nondenaturing gel (Fig. 1C). Anti-GS antibodies did not cross-react with groEL or groES. The gel used in the electroblotting procedure was stained with Coomassie Blue, and a substantial amount of higher molecular mass oligomeric proteins remained on the gel (Fig. lA ). This observation indicated that a significant amount of the protein was not electroblotted onto the nitrocellulose transfer membrane. Therefore, this observation, coupled with the possibility that the transfer eficiencies vary for the resolved transient oligomeric species, suggest that the information concerning the resolved bands observed on the Western blots is primarily qualitative in nature. The shift in molecular mass resulting from the complex formation and hence the stoichiometry between groEL and GS cannot be accurately assessed by nondenaturing gradient-pore gel electrophoresis.
Generating Standards for Resolving the Assembly Intermediates of GS-The intermediate oligomeric species observed during the chaperonin-dependent and independent renaturation comigrated with dissociated GS molecular mass standards. The dissociated GS molecular mass standards were generated by diluting native dodecameric GS into a low ionic strength, low M e or Mn2+ solution, at 0 "C. This treatment caused the native dodecameric structure to dissociate into seven distinct species. The identity of these species is consistent with the formation of the monomer and even-numbered oligomers such as dimers, tetramers, hexamers, octamers, decamers, and undissociated dodecamer (Fig. 2, A and C and Fig. 4, A and B, lane   1 ). The dissociation of dodecameric GS occurs through the disruption of the heterologous side to side intra-ring subunit contacts as opposed to disrupting the up-down inter-ring contacts (24,321. Calculated GS subunit-subunit free energies, derived from the x-ray crystallographic model, suggest that the larger binding free energies associated with the up-down subunit contacts result from the hydrophobic interactions at this subunit interface (33).

Time-dependent Resolution of the GS Assembly Reaction-
Using a Pharmacia Phast system, one can obtain snapshots of the GS assembly reaction by initiating the GS renaturation reaction at staggered times prior to loading the reaction mixtures onto the phastgel (see "Experimental Procedures"). The Pharmacia Phastsystem is designed to load all the samples simultaneously onto the top of precast pre-electrophoresed gels. Western analysis was used to visualize and resolve the transient GS assembly intermediates without interference from groEL or groES (Figs. 2 and 4) and was consequently preferable to silver staining.
The formation of transient oligomers was also examined by initiating the renaturation in the presence of methionine sulfoximine and ATP. In the presence of these substrates, GS forms a n extremely tight binding transition state analog, Met-Sox-P, in situ which prevents dissociation of the subunits that make up the active site. Once formed, this analog remains bound to the active site of GS, and the binding is essentially irreversible under these solution conditions (24). To ensure that further assembly does not occur during electrophoresis, a 1000fold molar excess of 5,5-dithiobis(2-nitrobenzoic acid) (DTNB) to GS was added before the samples were separated on native  From left to right, lanes 3-8 are the separated GS oligomer species after ATP was added to the preformed complex and allowed to assembly after 2,5,10,15,30, and 120 min, respectively. The groEL.GS complex slowly disappears when ATP is added. GS monomers migrated toward the bottom of the 4-15% gel and were not easily resolved. B, unfolded GS added to the preformed groEL.ATP complex (5 m ATP). From left to right, lanes 2-7 contained the time-dependent transient GS assembly intermediates that were formed when the GS renaturation reactions were initiated by adding unfolded GS monomers to a preformed groEL.ATP complex (5 m ATP). The groEL.ATP complex was incubated for 1 min prior to the addition of unfolded GS monomers. Lanes 2-7 contain GS assembly intermediates formed and applied onto the gels after folding had proceeded for 2, 5, 10, 15, 30, and 120 min, gels. This procedure modified surface sulfhydryl groups at the subunit interfaces and prevents further assembly. This same strategy was used to isolate dissociated GS intermediates generated from the native dodecamer (24). The same time-dependent pattern of transient intermediate formation was observed when MetSox-P and DTNB were present with the exception of the resolution of monomers and dimers (gel not shown). Maurizi and Ginsburg reported that monomeric and dimeric units of GS precipitated in the presence of DTNB (24). Since there was an increase in the amount of material that did not migrate into the separating gel following the MetSox-P and DTNB treatment, the precipitated material was presumed to be primarily comprised of modified GS monomers and dimers. By comparison, resolving the assembly species by directly applying the renaturing solution onto native gels showed no difference in the time-dependent appearance of the GS intermediates compared with the MetSox-P, DTNB treatment. This suggests that assembly during electrophoresis is minimal (see also lane 1, Fig. 1B). The untreated separation method has the advantage of resolving the monomer and dimer GS intermediates. Therefore, the time-dependent formation of assembled GS intermediates was examined without adding MetSox-P and DTNB.
Upon separating the time-staggered renaturation mixtures on a 4-15% gradient gel, the electrophoretically resolved groEL.GS immunoreactive complex slowly disappears with time after ATP addition ( Fig. 2A, lanes 3-8). The disappearance of the complex is accompanied by a time-dependent appearance of dodecameric GS. The preformed complex decreased yet remained visible for at least 0.5 h following ATP addition. The higher order oligomers of GS became prevalent at later time intervals of >5 min ( Fig. 2A and Fig. 4A, lane 5). Since the transient oligomers migrated precisely with the even-numbered molecular mass standards (dimer to dodecamer) generated from native GS (Fig. 2, A and C, and Fig. 4, A and B, lane 1 ), the increase in the higher order oligomers occurred by the combination with dimeric units.
The Amount of Resolved groEL.GS Complex Depends on the Order of Addition of ATP and GS to groEGEarlier experimental results suggested that the binding affinity of partially or unfolded GS for groEL may be dependent on whether ATP is initially present (6). By Western analysis, the amount of groEL.GS complex was found to be substantially higher when unfolded GS was added to the nucleotide free form of groEL (e.g. at 2 min) than when ATP (5 mM) was added to groEL prior to the addition of unfolded GS (compare complexes at various times shown in Fig. 2, A and B). In order to visualize the groEL.GS complex when the ATP-bound groEL was the initial species, the development reaction of the Western blot (Fig. 2B was allowed to continue for a longer period of time. This difference in resolved material may be linked to lower binding affinities of partially folded GS for ATP-bound groEL than for the ATP free form. The differences in the amounts of resolvable complex for these two conditions are consistent with the differences in the initial lag in GS activity regain profiles presented previously (see Fig. 5 in Ref. 6). When the ATP concentration respectively. Lane 1 contains the dissociated GS molecular mass standards. The increase in intensity is due to longer incubation times (%fold) during development of the Western analysis to visualize the groEL.GS complex. C, ATP added to the preformed groEL.GS complex in the presence of groES. Lane 1, dissociated GS molecular mass markers.

Glutamine Synthetase Assembly 13633
was increased from 5 to 10 mM and added to the preformed groEL.GS complex (no groES), no differences were observed in the GS renaturation rate profiles. (Fig. 2C, lane 2 , time = 0, no ATP). When ATP is added (3-5 mM) to the groEL.GS complex in the presence of groES, the complex is no longer observed at the initial times (compare lane 3, T = 2 min 2 groES in Fig. 2, A and C ) . This observation may suggest that the affinity of groEL for unfolded or partially folded GS is significantly decreased when the complete chaperonin system (groEL, groES, MgATP) is present. In addition, the GS assembly intermediates appear much earlier in the presence of the complete chaperonin system (Fig. 2C). The rapid appearance of GS oligomeric intermediates is most likely due to a concentration-dependent increase in the association of released GS subunits.

The Addition of groES to the Preformed groEL.GS Complex Accelerates the Formation of Active GS Oligomers-The groEL.GS complex is readily formed in the presence of groES
GroES Accelerates GS Assembly from Stable groEL.GS Complexes in the Presence of ATP-The groES-induced acceleration of GS assembly from preformed groEL.GS complexes was observed through 1) enhanced GS activity regain profiles (Fig.  3 A ) and 2 ) more rapid changes in the GS assembly-dependent tryptophan fluorescence (Fig. 3B). The enhanced activity regain profiles are consistent with the increase in electrophoretically resolved transient GS oligomer intermediates. Since each subunit of GS has 2 tryptophan residues near the active site or subunit-subunit interface, the time-dependent tryptophan fluorescence changes, coupled with the activity regain profiles, directly reflect the GS assembly process (30, 33). The fluorescence experiments were performed in the presence of 2 mM L-methionine sulfoximine phosphate (MetSox). In the presence of ATP, GS phosphorylates MetSox which binds at the newly formed active site and enhances the tryptophan fluorescence signal. With the complete chaperonin system, there was no significant lag in the fluorescence signal. Since the complex is no longer prominently observed on the electrophoresis gels at early time points, groES appears to accelerate the release of GS from groEL. Even though groEL and groES are highly purified, some contaminating tryptophan fluorescence, albeit it small, is still present. Control experiments showed that when ATP is added to the same concentration for groEL and groES in the absence of GS, no time-dependent fluorescence changes occurred.
GS Monomers Are Released from groEL-The assembly of GS appears to progress from monomer to dimer and is followed by a time-dependent ordered increase of even-numbered higher molecular mass species up to the dodecamer (see Western analysis in Fig. 4). In order to resolve the lower molecular mass GS assembly intermediates and monomers, the time-dependent renaturation mixtures were separated using an 8-25% polyacrylamide continuous gradient. A quantitative assessment of the different species formed during the time course of the assembly was not possible under these experimental conditions due to the incomplete immunotransfer of the protein species and the variable immunotransfer efficiencies of higher molecular species onto the nitrocellulose blotting paper. However, relative amounts of the same molecular mass species reveal important differences between the various renaturation conditions. All seven transient GS assembly intermediates are observed during the renaturation (Fig. 4 ) . When the chaperonin proteins and ATP were present, GS monomer was detectable by Western analysis throughout most of the renaturation reaction (Fig. 4, A  and B). In contrast, smaller concentrations of GS monomers were observed at the initial time points during the spontaneous renaturation and assembly of GS (Fig. 4C, lanes 3-6) and were no longer visible at 15 min (Fig. 4C, lane 7). A significant amount of aggregated material remained in the stacking gel and did not migrate from the loading position. This material probably resulted from misfolded and highly aggregated GS. These data are in agreement with the hypothesis that chaperonins prevent the rapid formation of misfolded or inappropriately aggregated GS oligomers (3). The chaperonins appear to be able to increase the concentration of assembly competent monomers.
ADP Initiates GS Refolding from a groEL.GS Complex in the Presence of groES-Previous studies showed that ADP alone does not support the renaturation of GS from groEL.GS complexes (Table I, see Ref 6). In experiments presented here, ATP-free ADP (see "Experimental Procedures") was added to the groEL.GS complex (37 "C). No GS activity was observed until the groES was added at the 30-min time point. The extent of the GS activity regain was the same as was observed with ATP ( Fig. 5 A ) . GroES addition initiates the renaturation of GS from an arrested groEL-GS complex in the presence of 3 mM ADP (Fig. 5 A ) . In addition, the renaturation rates of GS from a groEL.GS complex + groES were the same when either 3 mM are the separated GS oligomer species when ATP was added to the preformed groEL.GS complex and were allowed to renature for 1,2,5, 10, 30, and 120 min incubation times prior to electrophoretic separation. Upon ATP addition, the amount of released monomers increases initially (lune 3 ) . B , 5 mM ATP added to the preformed groELGS complex in the presence of groES (same times as in A). The higher order oligomers appeared much earlier in the renaturation (e.g. lune 3, time: 1 min) compared with the same time in A. C, for the spontaneous GS renaturation reaction, from left to right: lune 1, native GS; lune 2, dissociated native GS molecular mass standards; lunes 3-8 are the electrophoretically separated oligomeric species after the renaturation proceeded for 2, 5, 10, 15, 30, and 120 min after being initiated by 100-fold rapid dilution. In the presence of ADP (3 m~) , no renaturation was observed until groES (1 p~) was added (A) as indicated by the arrow. The yield of active GS formed is the same as was observed when ATP was added. B, comparison of ADP and ATP induced GS renaturation from a preformed groELGS complex in the presence of groES. Prior to nucleotide addition, the preformed complex was divided into equal aliqouts. When either 3 m~ ATP (A) or ADP (+) are added to the groELG3 complex, the renaturation rates are the same. Under similar conditions, the AMP-PNP-induced GS renaturation was slower (W. ATP or ADP were present (Fig. 5B). When AMP-PNP is added to the preformed groEL.GS complex + groES, the final activity regain is lower, and the rates of GS renaturation are slower than with ADP or ATP (Fig. 5B). DISCUSSION Numerous studies have suggested that chaperonins release monomers which then assemble into higher order oligomers (10, 11,341. The results presented here indicate that the chaperonins do not directly aid in the assembly of dodecameric glutamine synthetase. Chaperonins appear to release higher levels of monomers than the levels observed during the spontaneous renaturation (Fig. 4). Once these monomers form assembly competent dimers, the dimers provide the secondary building blocks for higher order even-numbered GS oligomer assembly. Although GS can fold and assemble in the absence of the chaperonins in uitro, the active GS yields are small at physiological temperatures and high GS concentrations. Chaperonins appear to markedly increase the amount of active GS formed under both of the above mentioned conditions by maintaining high levels of assembly competent monomers (6, 7). A direct comparison between species generated during the spontaneous and chaperonin-dependent GS renaturations indicated that larger amounts of resolved monomers were initially observed when chaperonins were present (Fig. 4).
The chaperonin-assisted reconstitution of GS does not absolutely require both ATP and groES. When ATP alone is added to the preformed groEL-GS complex, the electrophoretically resolved complex slowly disappears. Active GS oligomers, where the tetramer is the minimal active unit (241, appear more prominently at later time points ( Fig. 2A and Fig. 4B, 5 min). This observation is consistent with the long lag in the activity regain profile when groES is absent from the chaperonin-dependent GS renaturation (Fig. 3). There is also a distinct lag in the GS-dependent tryptophan intensity increase following assembly. Each E. coli GS subunit contains 2 tryptophans CTrp", T r p l S 8 ) located at the N-terminal portion of the molecule. Since the GS active site is formed from the heterologous side-side contacts between the N-and C-terminal domains of two GS subunits, the observed changes in tryptophan fluorescence reflect changes in the tryptophan microenvironments during assembly and active site formation (30,35,36).
If ATP is added to groEL prior to initiating the refolding of GS, the amount of the electrophoretically resolved complex is substantially decreased and the oligomeric intermediates are observed at earlier time points (Fig. 2 B ) . These data are consistent with the diminished inactive lag period in the observed activity regain profile (see Fig. 5 in Ref. 6). The source for this observed difference may come from the binding affinity differences between the partially or unfolded GS with nucleotide free and bound groEL. This view is compatible with the hypothesis of Clarke and co-workers (4) who propose that the ATP-bound form of groEL has a lower binding affinity for partially or completely unfolded proteins than does the nucleotide-free form.
In the presence of groES and ATP, the groEL.GS complex is no longer readily visible prior to GS assembly. GroES appears to increase the release of bound GS from arrested groEL.GS complexes and thereby increasing concentration of assembly competent monomers. This leads to an acceleration in the rate of assembly of GS from arrested groEL-GS complexes (Fig. 3).
This conclusion logically follows from the observed increases in active higher order GS assembly intermediates at early times and the early increases in both GS activity regain and tryptophan fluorescence intensities during GS assembly. These results lend direct support for the preliminary studies of Girshovich and co-workers (37) who suggested that the groES lowers the affinity of ATP-bound groEL for rhodanese. The question remains as to how groES can facilitate this folding event in the presence of ATP Recently, Kawata and co-workers (38, 39) have demonstrated that the refolding yields of tryptophanase and enolase were enhanced with ADP and both chaperonins. The chaperonin-assisted folding of GS is also observed when ADP is added to a fully arrested groEL.GS complex in the presence of groES. ATP hydrolysis is not required for efficient release of GS from groEL. Presumably, the formation of the ternary complex groEL,,-groES,.ADP, can alter the conformation of groEL to initiate the GS release (40-42). One explanation would be that the ligand binding free energies of groES and ATP may have a negative cooperative effect on the binding equilibria between the bound GS and groEL, resulting in an enhanced release rate. This remains to be tested.
There is a substantial variability in the solution requirements for groE chaperonin-assisted folding of various substrates (2-4, 6, 12, 43). For example, the correct folding of ribulose-bisphosphate decarboxylase, citrate synthetase, rhodanese, and ornithine decarboxylase absolutely require the complete groE system and ATP (2, 3, 11, 12, 44). The importance of ATP hydrolysis during the chaperonin-assisted folding of these proteins is unclear. It has been proposed that chaperonin-dependent ATP hydrolysis permits the folding protein to rebind to the ADP-bound or nucleotide free form of the groEL chaperonin if it has not acquired the correct folded structure after release (12,45). After rebinding, another round of ATP binding and hydrolysis takes place, and the folding protein again dissociates from the groEL surface and continues to fold.
Recent electron microscopy data indicate that incompletely folded proteins may bind in the central channel of groEL (46). In addition, the diameter of the central channel of groEL and the conformation of the groEL changes when ADP or ATP are present (40,41). From these data, it has been proposed that folding proteins remain sequestered inside the central cavity of groEL where they undergo a series of release and rebinding reactions driven during multiple rounds of ATP hydrolysis by groEL (47). However, this hypothesis still requires more direct experimental support. Theoretical models that simulate the renaturation rate profiles from experimental data for chaperonin-assisted ribulose-bisphosphate carboxylase assembly (2) cannot distinguish between mechanisms that either allow protein folding to occur on the surface of groEL or invoke a release and rebinding reaction scheme (48). It is conceivable that both processes can occur simultaneously.
It is interesting to note that a majority of the in vitro protein substrates that absolutely require the complete chaperonin system reside in the mitochondria (3,5, 11,12,44). Since these proteins are synthesized in the cytosol, they must be maintained in a transport competent state (i.e. partially unfolded) prior to transport into the mitochondria. Recently, Martinez-Camon and co-workers have demonstrated that the mature form of mitochondrial aspartate transaminase folds slower than its cytoplasmic counterpart cytoplasmic aspartate transaminase (49). These investigators also demonstrated that at high temperatures, the mitochondrial aspartate transaminase absolutely requires the complete system, whereas the cytoplasmic aspartate transaminase can fold with groEL and ATP alone (50). Perhaps some slower folding proteins may require the entire chaperonin system, because they exist as a partially folded intermediates for longer periods of time. If partially folded proteins bind to groEL at multiple sites, then the degree of unfolding may determine the binding affinity and hence the requirement of the complete chaperonin system.
It is highly likely that groEL has a wide array of binding affinities for various proteins. For example, the binding free energies for a protein-groEL complex may be variable because of the extent of unfolding (51). Also, groEL.P22 tailspike complexes dissociate as the temperature is decreased which may reflect a temperature dependence for binding phenomenon (14). Thus, the binding equilibria between proteins and chaperonins may be a principle factor which relates to the different solution requirements of various proteins. In order to determine the frequency of these different chaperonin requirements in protein folding within a given organism, it will be necessary to identify those endogenous proteins that require some or all of the endogenous chaperonins to augment their folding.
In vivo, natural selection may favor organisms that synthesize cytoplasmic proteins which rapidly fold and assemble at normal growth temperatures to either minimize or, in some cases, not require (16) any interactions with chaperonins. For those proteins that do interact, groEL probably has variable binding affinities for various nascent polypeptides, it is possible that groEL utilizes the binding free energies of ATP, ADP, and groES to modulate its binding affinity and allow folding to proceed. However, there are numerous examples of proteins that function as molecular chaperones which do not require Glutamine Synthetase Assembly ATP for their action (52-55). Unlike groEL, these molecular chaperones interact transiently with their particular substrates and do not entirely arrest the folding reaction. Here the release appears to be triggered by the folding or formation of higher order oligomers.
At this time, proposed mechanisms of chaperonin function are highly speculative. If the rebinding and release hypothesis for chaperonin facilitated folding is correct, then proteins which interact with groEL and fold in the presence of either the incomplete (groEL and ATP alone) or the complete (groEL, groES, and ATP (or ADP)) system may only require a single binding and release event to attain elevated levels of correctly folded structure. Since it is also possible that some folding (or unfolding) may occur while the protein is complexed with groEL, the structural population of GS subunits that initially binds to groEL may not be the same conformer population that is eventually released from groEL following activation by groES and ATP (or ADP) (7). Gray and Fersht (56) have suggested that some refolding of barnase can occur while it is in contact with groEL. In this case, an increase in the groEL concentration alone slows the spontaneous folding rate to a constant lower level. In general, the ease (i.e. rate) at which initial conformer populations are shifted to folding or assembly-competent forms on groEL may depend on the binding strength and the number of different binding interactions that define the groEL-protein folding intermediate complex. As pointed out by Gray and Fersht (561, this rate of folding is probably determined by the nature and lifetime of the folding intermediate and not by groEL. This feature may in turn dictate whether a protein undergoes a single release or must rebind (or remain bound) to groEL. Studies aimed at examining the release reaction and probing the energetics aspects and structural nature of the groEL-bound folding intermediates will be useful in elucidating the mechanism of chaperonin-assisted folding.