Ligand-induced conformational changes in the apical domain of the chaperonin GroEL.

Although the role of nucleotides in the catalytic cycle of the GroESL chaperonin system has been extensively studied, the molecular effects of nucleotides in modulating exposure of sites on GroEL has not been thoroughly investigated. We report here that nucleotides (ATP, ADP, or adenosine 5'-(beta, gamma-imino)triphosphate) in the presence of Mg2+ make the oligomer selectively sensitive to trypsin proteolysis in a fashion suggesting conformational changes in the monomers of one heptameric ring. The site of proteolysis in the monomer that is exposed upon nucleotide binding by the oligomer is in the apical domain (Arg-268). Further, complexes of GroEL with GroES or rhodanese display the same sensitivity to proteolysis, unlike the GroEL-GroES-rhodanese complex, which is protected from proteolysis. The influence of various cations on trypsin proteolysis is investigated to elucidate the differential effects that monovalent and divalent cations have on the oligomeric structure of the chaperonin. These results are discussed in relation to the molecular basis for the chaperonin activity.

The molecular chaperones are a class of proteins that have been shown to facilitate the in vivo folding and transport of nascent polypeptides (1,2). The chaperonins, one class of molecular chaperones, have been found in prokaryotes, mitochondria, and chloroplasts (3). One widely studied chaperonin from Escherichia coli, GroEL, has been demonstrated to promote the in vitro refolding and assembly of a variety of chemically denatured proteins, including rhodanese (4,5), ribulose-bisphosphate carboxylase/oxygenase (Rubisco) (6,7), and glutamine synthetase (8). The promiscuity of GroEL in polypeptide recognition and binding suggests that the hydrophobic interactions accounting for the complex formation vary from substrate to substrate (9 -11). The fact that some proteins require only K ϩ and ATP-Mg (or a non-hydrolyzable analog) for release of functional enzyme from the complex, whereas others also require the co-chaperonin GroES, supports this idea that complexes with some non-native proteins are stronger than others (12,13).
GroEL is a homotetradecamer (14-mer) of 57.2-kDa subunits, arranged as two stacked heptameric rings, with a central cavity at each end (14). The 2.8-Å x-ray crystal structure reveals that each monomer is organized into an equatorial, intermediate, and apical domain, with interactions between equatorial domains exclusively defining the heptamer-hep-tamer interface (15). Monomer-monomer interactions within each ring are mediated by the equatorial domains and by conserved interactions at the intermediate-apical domain interface (15,16). The apical domains line the opening to the central cavity at either end of the oligomer and contain a region (200 -263) that has been implicated by site-directed mutagenesis and photoincorporation of the hydrophobic probe bis-ANS 1 as the site of polypeptide binding (17,18).
Studies of the effects of nucleotides on GroEL have focused on regulation of the ATPase activity of the chaperonin (19,20), including the elements of positive and negative cooperativity in nucleotide binding/hydrolysis, and the debate over complex formation (21)(22)(23)(24)(25)(26). Several studies have shown that two GroES molecules can bind to a single GroEL oligomer in the presence of ATP to form symmetric complexes, while others have demonstrated asymmetric complexes (1:1 GroEL:GroES ratio) to be the functional, physiological unit. The few studies that have evaluated the conformational changes in GroEL due to ATP binding or hydrolysis have demonstrated quaternary structural changes consisting of monomer pivoting and apical domain reorientation upon ATP binding (27)(28)(29). Unfortunately, the results do not provide much information about the exposure of specific sites believed to mediate protein-protein interactions. Finally, nucleotides have also been shown to affect the affinity of GroEL for substrate, with the ADP-Mg complex displaying tight binding and the ATP-Mg complex displaying weak binding (30).
There have been several reports on ions (monovalent, divalent, and polyvalent) affecting structural changes in GroEL (31)(32)(33). Monovalent and polyvalent cations are reported to increase the exposure of hydrophobic surfaces on GroEL without disrupting the oligomeric structure (33). Divalent, but not monovalent, cations stimulate the ATPase activity of GroEL and induce structural changes that allow preferential crosslinking of the heptameric rings (31). These results suggest that cations of various charges may work at several different levels to produce structural changes in the GroEL oligomer.
In this report we show that proteolysis by trypsin can be used as a probe for conformational changes in the tetradecameric structure of GroEL. In a native state with Mg 2ϩ as the only ligand, GroEL is not susceptible to proteolysis by trypsin, whereas unliganded GroEL is rapidly digested, leaving only 10% of intact monomers within 30 min. Upon addition of nucleotides (ADP, ATP, or the ATP analog AMP-PNP) and Mg 2ϩ , exposure of one predominant cleavage site in the apical domain occurs. When ADP-Mg is liganded to GroEL, producing a state with high affinity for non-native protein, approximately half of the monomers become proteolyzed without loss of the oligomeric structure. Complexes of GroEL-ADP-Mg with GroES or denatured rhodanese are proteolytically sensitive in a manner similar to the nucleotide-liganded state. In contrast, a GroES-GroEL-rhodanese complex formed in the presence of ADP-Mg is significantly protected from proteolysis (ϳ90% protected). Monovalent and polyvalent cations also affect trypsin proteolysis of GroEL, protecting approximately half of the sites from cleavage compared with an unliganded state. These results suggest that nucleotide binding induces asymmetry in the GroEL tetradecamer, exposing specific sites in the apical domain of only one heptameric ring such that they are accessible to other proteins in solution, either protease or substrate proteins.

MATERIALS AND METHODS
Reagents and Proteins-All reagents used were of analytical grade. Trypsin was L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated type XIII purchased from Sigma. Low molecular weight markers were purchased from Bio-Rad. Hide powder azure was purchased from Calbiochem.
Rhodanese was prepared as described previously (34) and stored at Ϫ70°C as a crystalline suspension in 1.8 M ammonium sulfate. Rhodanese concentrations were determined using A 280 nm 0.1% ϭ 1.75 (35) and a molecular mass of 33 kDa (36). The chaperonin, GroEL, was purified from lysates of E. coli cells bearing the multicopy plasmid pGroESL (37). The purification was by a modified version of published protocols (20), excluding the Mono Q and hydroxyapatite columns. After purification, GroEL was dialyzed against 50 mM Tris-HCl, pH 7.5, and 0.1 mM dithiothreitol and then made 10% (v/v) in glycerol, rapidly frozen, and stored at Ϫ70°C. The co-chaperonin, GroES, was purified from lysates of E. coli cells bearing the multicopy plasmid pND5, using a previously published protocol (38). After purification, GroES was stored at 4°C in 70% ammonium sulfate. Prior to use, GroES was centrifuged in a microcentrifuge, and the pellet was resuspended in 50 mM triethanolamine hydrochloride, pH 8.0, and dialyzed overnight against the same buffer. The protomer concentration of GroEL was measured under denaturing conditions, and GroES was measured under non-denaturing conditions using the bicinchoninic acid assay (Pierce) according to the procedure recommended by the manufacturer.
Proteolysis-GroEL (10.4 M monomer) in 50 mM Tris-HCl, pH 7.8, 10 mM 2-mercaptoethanol, and ligands as specified in the text or figure legends was incubated with trypsin (5%, w/w) at room temperature for the times noted in the figure legends. The reactions were stopped one of two ways: 1) by adding phenylmethylsulfonyl fluoride to 3 mM from a stock solution in ethanol and allowing the sample to incubate on ice for 10 min prior to adding sample buffer; or 2) by adding heated sample buffer and immediately boiling for 5 min prior to analysis on denaturing polyacrylamide gels. The results obtained were the same regardless of which method was used for stopping the proteolysis reaction. The proteolytic activity of trypsin under various buffer and ionic conditions was tested using an ultrasensitive colorimetric method as described by Rinderknecht et al. (39).
For the experiments involving GroEL-rhodanese protein complexes, rhodanese was unfolded in 8 M urea as described previously (40) and diluted into a solution containing GroEL and all the other components of the buffered solution as indicated in the figure legends. In forming ternary complexes of GroEL-GroES-rhodanese, GroES was added to a solution already containing GroEL, nucleotide, and unfolded rhodanese. In each case, the final solutions were incubated at room temperature for at least 10 min to allow complex formation prior to proteolysis.
Amino Acid Sequence Analysis of Proteolytic Fragments-Trypsin digestion products of GroEL were electrophoresed as above and then electrophoretically transferred onto Immobilon-P membrane (Millipore Corp., Bedford, MA) using a semi-dry transfer system (American Bionetics, Inc., Hayward, CA) as described previously (44). The membrane was stained with 0.1% Ponceau S in water until the peptides were just visible. The desired bands were excised, washed, and allowed to air dry. The fragments were then analyzed on an Applied Biosystems 477A protein sequence analyzer (Applied Biosystems, Foster City, CA). Peptides were subjected to approximately 12 rounds of sequencing. The sequences of the fragments were compared with the published sequence of GroEL to determine their positions within the intact protein (3).
Scanning and Quantitation of Protein Gels-Coomassie-stained gels were imaged using a CCD 505 video camera (CCTV Corp., New York) under the control of NIH-Image software running on a Macintosh IIci microcomputer. Band densities were quantified using the densitometry options of the software.
Curve Fitting-Except where noted in the figure legends, data were fit using a non-linear least squares fitting procedure implemented in the software program, PSI Plot (Poly Software International, Salt Lake City, UT). Where data were fit to an exponential, the following equation was used: Y ϭ A 1 e Ϫkt ϩ A 2 , where Y represents the measured value, t represents time in minutes, A 1 represents the fraction of total protein susceptible to digestion (A 1 e Ϫkt therefore represents the fraction of protein digested at time t), and A 2 represents the fraction of protein resistant to proteolysis.
Terminology-The notations used to designate complexes imply neither stoichiometries nor ligand distribution but merely the substituents of the complex based on the reaction mix and conditions used in the complex formation. Fig. 1A represents a denaturing gel showing the time course for trypsin digestion of a GroEL-ADP-Mg complex. The parent band, representing a GroEL monomer, is digested, and a doublet appears consisting of bands with apparent molecular masses of ϳ31 and ϳ32 kDa. The intensities of the doublet bands increase over the time course and are approximately equal to each other at every time point, suggesting that they appeared together rather than one giving rise to the other upon further degradation. Other extremely faint bands appear but were not quantified since they do not represent a significant percentage of the digested GroEL monomer. The N-terminal sequences of the two bands of the doublet were determined, and they represent complementary sequences in a GroEL monomer; the upper band begins at glycine 269, and the lower band begins at the N terminus of the mature protein (alanine). Trypsin cleavage at Arg-268, in the apical domain, would produce two fragments of these approximate molecular masses and with the observed N-terminal sequences. Fig. 1B is a semi-log plot representing the percentage of parent band (monomeric GroEL) remaining at various times during the trypsin digestion of the GroEL-ADP-Mg complex. The curve was best fit by a single exponential with the following parameters: k ϭ 0.186 s Ϫ1 , A 1 ϭ 0.423, and A 2 ϭ 0.516, where A 2 represents the fraction of undigested protein remaining at long proteolysis times. The results indicate that within the first 15 min of proteolysis approximately half of the monomers in the tetradecamer were cleaved at Arg-268, while the other half of the monomer population remained intact. Longer periods of digestion (up to 60 min) did not produce significantly more than 50% digestion (data not shown). Fig. 2A shows a similar plot for the proteolysis of GroEL under different conditions. The upper curve (solid circles) represents trypsin proteolysis of the GroEL oligomer in the presence of 20 mM Mg 2ϩ . Very little digestion occurred (10% at most) without production of a doublet such as that seen in Fig.  1A. The lower curve (open squares) demonstrates the significant digestion of the unliganded GroEL oligomer. These data (open squares) were best fit by a single exponential with the following parameters: k ϭ 0.067 s Ϫ1 , A 1 ϭ 0.852, and A 2 ϭ 0.094. The digestion of unliganded GroEL also produced the doublet shown in Fig. 1A, but the percentage of the parent band converted to doublet was less, and many other weakly staining bands appeared (data not shown). These observations are consistent with the presence of a broader distribution of GroEL structures in the absence of ligand. Finally, in the absence of ligand, higher trypsin concentrations (10 -20%) completely digested the chaperonin, producing fragments that were too small (or too weakly staining) to be visualized on the 12% denaturing gels used (data not shown).

Oligomeric GroEL Can Be Proteolyzed by Trypsin in a Ligand-dependent Fashion-
To investigate the differences in proteolysis between unliganded and magnesium-liganded GroEL an experiment was conducted where the degree of proteolysis was measured at fixed times (15 or 30 min) at increasing concentrations of Mg 2ϩ . The results, graphed in Fig. 2B, show that an increasing percentage of the monomers were protected from trypsin proteolysis as the Mg 2ϩ concentration was increased from 0 to 20 mM, with maximum protection afforded by 10 -12 mM. This protection from trypsin proteolysis at increasing Mg 2ϩ concentrations is strikingly similar to the effect of increasing Mg 2ϩ concentrations on the cross-linking of heptameric rings in GroEL (31).
Trypsin proteolysis of GroEL in the presence of ATP-Mg or AMP-PNP-Mg produced results similar to those seen with ADP-Mg. Fig. 3 is a semi-log plot comparing the results with the three nucleotides, showing that all three produced an oligomeric structure that was sensitive to trypsin proteolysis. It is clear that nucleotide binding alone and not hydrolysis produced the conformational change necessary to expose the clip site in the apical domain. In each of the three cases the same digestion products appeared on denaturing gels, with the bands of the doublet described above representing the predominant species (data not shown).
Proteolyzed GroEL Retains Its Tetradecameric Structure- Fig. 4 shows a non-denaturing gel of samples digested with trypsin for various times in the presence of ADP-Mg (treated as in Fig. 1). In this gel system the band shown corresponds to tetradecameric GroEL, which runs as a sharp band near the top of the gel and can be easily distinguished from monomers (not present/seen) that run as a smeared band and migrate considerably further into the gel (45). The single band representing the GroEL tetradecamer is initially sharp but becomes increasingly fuzzy as the proteolysis progresses, and the band appears slightly but progressively higher in the gel. The staining intensity of the single band also increases by approximately 50% across the series. These results demonstrate that the proteolysis did not disrupt the oligomeric structure of the GroEL but certainly did produce a band with altered electrophoretic and staining properties, suggesting a heterogeneous population of structures.
There was a question of whether the non-denaturing gels were visualizing tetradecamers that had not fallen apart or tetradecamers that had reassembled during electrophoresis. This question was addressed by running samples treated in the same manner as above on a gel permeation column (TSK-4000), a system that resolves monomers from tetradecamers. The samples of ADP-Mg-liganded GroEL treated with trypsin for 30 min eluted in the same volumes as controls with native, unperturbed GroEL, or ADP-Mg-liganded oligomers that had not been proteolyzed, indicating that the tetradecamers remained intact after proteolysis (data not shown). Two-dimensional polyacrylamide gel electrophoresis was also performed. The bands from non-denaturing gels were excised, dissolved in SDS sample buffer, and electrophoresed on denaturing gels. The results were exactly the same as when samples were run directly on denaturing gels, as shown in Fig. 1A. These results correspond to GroEL standard (untreated) and 0, 1, 3, 6, 9, 12, 15, and 18 min of digestion, respectively. B, semi-log plot showing percent of undigested monomers remaining at increasing times of trypsin proteolysis. Data points (q) represent an average of four separate experiments, all performed as outlined in A. In each experiment the "0 min digestion" or "untreated GroEL standard" was taken as the 100% point for calculating the undigested monomer percentages. The line is the best fit exponential curve determined as outlined under "Materials and Methods," with parameters as stated in the text.  Fig. 1A. Data points (q) represent averages of two separate denaturing gels. Since this phenomenon is more complex than a simple tight binding ligand (31), the dashed line is not a fitted line but is simply drawn to guide the eye.
confirmed that the bands visualized on the non-denaturing gels consisted of cut and uncut monomers, which remained associated as oligomers after proteolysis.
Formation of a Ternary Complex Protects GroEL from Trypsin Proteolysis-Trypsin proteolysis of GroEL was used to probe various complexes of GroEL with GroES and/or unfolded rhodanese. Fig. 5 is a semi-log plot showing the trypsin digestion of GroEL in complex with GroES and/or non-native rhodanese. The two binary complexes, GroEL-GroES (1:1 molar ratio) (filled circles) and GroEL-rhodanese (1:2.5) (open diamonds), formed in the presence of ADP-Mg showed approximately the same digestion of GroEL as with nucleotide alone. Since both rhodanese and GroES were digested by trypsin in these experiments, it is difficult to draw simple conclusions from the results, although similar results for a GroEL-GroES complex have been previously reported (46). In contrast, the ternary complex of GroES-GroEL-rhodanese (2:1:2), formed in the presence of ADP-Mg, significantly protected the chaperonin from trypsin proteolysis (asterisks). These results are very similar to the protection from proteolysis displayed when the chaperonin was liganded only by Mg 2ϩ . This suggests either that an additional conformational change of GroEL occurs upon binding both proteins at one end of the oligomer, such that the unliganded heptamer is protected, or that the proteins bind at opposite ends of the chaperonin, physically blocking the sites of proteolysis on both heptameric rings.
Cations Protect GroEL from Trypsin Proteolysis- Fig. 6 shows the digestion time course of GroEL in the presence of either KCl (filled circles) or spermidine (open squares). Many faint proteolytic fragments were present on Coomassie-stained gels, but the doublet seen in Fig. 1 was either a very minor or non-existent digestion product (data not shown). Since the control for these experiments is proteolysis of the unliganded GroEL tetradecamer, the cations apparently protected approximately half of the sites from proteolysis. Although similar to the observations with Mg 2ϩ -liganded GroEL, these results are distinct in that the divalent cation protected almost 100% of the proteolytic sites whereas the monovalent or trivalent cations protected only 50%. In each of the experiments, the presence of proteolytic activity was verified as described under "Materials and Methods" to be certain that the cations did not adversely affect the protease (data not shown). DISCUSSION An understanding of how various ligands induce conformational changes in GroEL and the specific sites which become exposed is crucial to deciphering the molecular mechanism of chaperonin activity. We were able to follow such conformational changes by using the differential trypsin proteolysis of GroEL under various conditions. Initially, an important distinction arises, which should be carefully considered. Mg 2ϩ alone shifts the conformation of the chaperonin from a state that is 100% susceptible to proteolysis to one that is completely protected from trypsin. This is consistent with published studies showing that 50% of the protein is cross-linked as a heptamer at Mg 2ϩ concentrations of 10 mM or higher (31). Apparently, conditions that cause association of the monomers in one heptameric ring (such that cross-linking is facilitated) result in an oligomeric structure that is fully protected from trypsin proteolysis. Considering the concentration range of this effect and the fact that Mg 2ϩ concentrations in E. coli may vary between 20 and 40 mM (47), it is more reasonable that the magnesium-liganded state of GroEL is the form that should be thought of as the physiological state (control) rather than the unliganded form.
Previous reports have suggested that nucleotides bind asymmetrically to the tetradecamer occupying only seven sites, presumably of a single heptameric ring (29,48). The results presented here (Fig. 1, A and B) support those earlier studies by showing that 50% of the monomers become proteolytically sen- sitive upon nucleotide binding to the magnesium-liganded tetradecamer and extend those reports by identifying the specific site of exposure within the monomeric structure (Arg-268). Proteolysis at other sites in this region (203, 237, and 268) has also been demonstrated when GroEL is perturbed by 2.5 M urea (33). Significantly, this site (Arg-268) is in the region that has been identified by site-directed mutagenesis and bis-ANS incorporation as the putative protein binding region in the apical domain (17,18). This suggests that nucleotide binding to the magnesium-liganded tetradecamer exposes the regions of the apical domains of one heptamer to allow for interaction with substrate protein.
The results presented here of the effects of Mg 2ϩ and ADP-Mg on the structure of GroEL are perfectly consistent with a recent study evaluating the stability of GroEL to ureainduced dissociation (49). The report showed that 10 mM Mg 2ϩ alone stabilized GroEL against urea-induced dissociation, as measured by light scattering, bis-ANS binding, or intrinsic tyrosine fluorescence. The ADP-Mg complexed form of GroEL was more easily dissociated by urea, and exposure of hydrophobic surfaces occurred more readily. Finally, the GroEL-ADP-Mg complex was susceptible to limited chymotrypsin proteolysis at an apparent cleavage point in the apical domain. Taken all together, it seems clear that ligands such as Mg 2ϩ and nucleotides can coordinate and induce conformational changes in functionally important regions of the GroEL structure so that interactions with other proteins can be regulated.
Proteolytic analysis of complexes reveals no significant differences between nucleotide-bound GroEL and GroEL in a binary complex with GroES or rhodanese. These findings suggest that any additional conformational changes in GroEL upon binding protein (either co-chaperonin or non-native substrate) are local, confined to the specific site of interaction, rather than global. By contrast, formation of a ternary complex of GroEL-GroES-rhodanese effectively blocks proteolysis of GroEL, suggesting either that association of the two proteins with the chaperonin occurs at opposite ends of the tetradecamer or that binding of the two to one end of the oligomer results in global conformational changes that prevent proteolysis at the opposite end of the oligomer.
Local flexibility in the chaperonin has been shown to be an important characteristic that is frozen upon binding to nonnative proteins, producing a complex which stabilizes the tetradecamer (50,51). Although nucleotide binding induces interdomain and quaternary structural rearrangements in GroEL (27,28), the local flexibility of the protein is not affected by formation of a GroEL-ATP complex (50). This suggests that structural regulation in the chaperonin exists at two different levels: 1) exposure of the reactive sites in one heptameric ring by nucleotide binding and/or hydrolysis; and 2) structural rearrangements upon protein-protein interaction that stabilize regions of local flexibility.
The effects of ions on the GroEL structure also suggest this multilevel basis for structural regulation. Divalent cations (i.e. Mg 2ϩ and Mn 2ϩ ) stimulate the GroEL ATPase activity, stimulate rapid cross-linking of monomers within a heptameric ring by glutaraldehyde, and completely prevent trypsin proteolysis (31,32) (Fig. 2A). Monovalent and polyvalent cations increase the exposure of hydrophobic surfaces that bind bis-ANS and prevent trypsin proteolysis of the chaperonin by only 50% (31,33) (Fig. 6). With the exception of K ϩ , monovalent and polyvalent cations do not affect ATPase activity. Taken all together, divalent cations appear to mediate intermonomer and heptamer/heptamer interactions, whereas monovalent and polyvalent cations appear to affect the local structure of monomers in a fashion that may possibly be localized to the apical domain.