Leu309 Plays a Critical Role in the Encapsulation of Substrate Protein into the Internal Cavity of GroEL*

In the crystal structure of the native GroEL·GroES·substrate protein complex from Thermus thermophilus, one GroEL subunit makes contact with two GroES subunits. One contact is through the H-I helices, and the other is through a novel GXXLE region. The side chain of Leu, in the GXXLE region, forms a hydrophobic cluster with residues of the H helix (Shimamura, T., Koike-Takeshita, A., Yokoyama, K., Masui, R., Murai, N., Yoshida, M., Taguchi, H., and Iwata, S. (2004) Structure (Camb.) 12, 1471-1480). Here, we investigated the functional role of Leu in the GXXLE region, using Escherichia coli GroEL. The results are as follows: (i) cross-linking between introduced cysteines confirmed that the GXXLE region in the E. coli GroEL·GroES complex is also in contact with GroES; (ii) when Leu was replaced by Lys (GroEL(L309K)) or other charged residues, chaperone activity was largely lost; (iii) the GroEL(L309K)·substrate complex failed to bind GroES to produce a stable GroEL(L309K)·GroES·substrate complex, whereas free GroEL(L309K) bound GroES normally; (iv) the GroEL(L309K)·GroES·substrate complex was stabilized with BeFx, but the substrate protein in the complex was readily digested by protease, indicating that it was not properly encapsulated into the internal cavity of the complex. Thus, conformational communication between the two GroES contact sites, the H helix and the GXXLE region (through Leu309), appears to play a critical role in encapsulation of the substrate.

In the crystal structure of the native GroEL⅐GroES⅐substrate protein complex from Thermus thermophilus, one GroEL subunit makes contact with two GroES subunits. One contact is through the H-I helices, and the other is through a novel GXXLE region. The side chain of Leu, in the GXXLE region, forms a hydrophobic cluster with residues of the H helix (Shimamura, T., Koike-Takeshita, A., Yokoyama, K., Masui, R., Murai, N., Yoshida, M., Taguchi, H., and Iwata, S. (2004) Structure (Camb.) 12, 1471-1480). Here, we investigated the functional role of Leu in the GXXLE region, using Escherichia coli GroEL. The results are as follows: (i) crosslinking between introduced cysteines confirmed that the GXXLE region in the E. coli GroEL⅐GroES complex is also in contact with GroES; (ii) when Leu was replaced by Lys (GroEL(L309K)) or other charged residues, chaperone activity was largely lost; (iii) the GroEL(L309K)⅐substrate complex failed to bind GroES to produce a stable GroEL(L309K)⅐GroES⅐substrate complex, whereas free GroEL(L309K) bound GroES normally; (iv) the GroEL(L309K)⅐GroES⅐substrate complex was stabilized with BeF x , but the substrate protein in the complex was readily digested by protease, indicating that it was not properly encapsulated into the internal cavity of the complex. Thus, conformational communication between the two GroES contact sites, the H helix and the GXXLE region (through Leu 309 ), appears to play a critical role in encapsulation of the substrate.
Chaperonins are a subclass of molecular chaperones capable of mediating ATP-dependent folding of polypeptides to their native states (1)(2)(3)(4). GroEL is the best characterized chaperonin; it is found in the cytoplasm of Escherichia coli and is essential for cell viability and growth at all temperatures (5). The complete functional cycle of GroEL is dependent on the presence of ATP and the co-chaperonin GroES (6 -11). GroEL is a large cylindrical protein complex comprising two heptamer rings of identical 57-kDa subunits stacked back to back (12). GroES is a dome-shaped, single heptamer ring of 10-kDa subunits (13). GroEL binds a wide variety of substrate proteins in non-native states and forms a binary complex (14 -18), which then binds ATP and GroES to the same (cis) GroEL ring to form the cis-ternary complex (8,9). The binding of GroES induces the encap-sulation of the substrate protein into an enlarged cavity (the ciscavity) inside the cis-ring, which is capped by GroES. In the ciscavity, non-native protein initiates folding without the risk of aggregation (8,9,19). Based on studies of crystal structures and mutagenesis, it is thought that the residues of GroEL involved in binding of GroES are overlapped, to a large extent, with those for binding of the substrate protein (16). Therefore, it might appear that binding of GroES results in freeing of the unfolded protein into the cis-cavity through deprivation of its binding sites. However, simple competition between substrate protein and GroES for the same binding sites does not explain how the release of substrate protein always results in encapsulation into the cis-cavity rather than diffusion into the bulk solution. Analysis of an intermediate in the process of encapsulation may help clarify the mechanism by which GroEL operates at this critical stage.
We recently determined a crystal structure of the native GroEL⅐GroES complex purified from Thermus thermophilus, the ciscavity of which is filled with cellular proteins (20). The structure shows several significant differences to the GroEL⅐GroES complex of E. coli, which was obtained by reconstitution of purified GroEL and GroES, in the presence of ADP (19). A new contact region between GroEL and GroES was identified in the T. thermophilus GroEL⅐GroES structure (Fig. 1A). In E. coli GroES, residues 24 -27 are part of a mobile loop structure (comprising residues 24 -30) that interacts with helices H and I at the apical domain of GroEL, located in the inner rim of the central cavity. In the GroEL⅐GroES of T. thermophilus, the same interactions are observed. The region 305 GFKLE 309 of GroEL (corresponding to the E. coli GroEL sequence 306 GMELE 310 ) makes contact with the sequence 33 PDT 35 , in the mobile loop of the adjacent GroES ( 28 TGS 30 in E. coli GroES). The residues Gly 305 , Leu 308 , and Glu 309 of T. thermophilus GroEL are well conserved across species, and hereafter, we refer this region as the GXXLE region. This region also has an intrasubunit interaction with the H helix; the side chain of Leu 308 (Leu 309 in E. coli GroEL) points at the N terminus of the H helix to form a hydrophobic cluster with other residues (Fig. 1, A, circle, and B). Fenton et al. (16) reported that a Leu 309 mutant of E. coli GroEL (GroEL(L309K)) was unable to assist folding. The aim of this study was to examine the contribution of the GXXLE region to chaperonin function, using E. coli GroEL and GroES. We investigated the GXXLE region in the E. coli GroEL⅐GroES complex to determine whether it is also in contact with GroES and whether or not Leu 309 in E. coli GroEL plays a role for the efficient encapsulation of substrate protein into the cis-cavity.
Strains and Plasmids-E. coli XL2-Blue (Stratagene) was used for site-directed mutagenesis and cloning. E. coli GroEL mutants were generated using QuikChange site-directed mutagenesis (Stratagene). The mutated groEL gene fragment was amplified using PCR, and the mutation containing pET-EL plasmid was used as a template. PCR products were digested with NcoI and HindIII and ligated into the NcoI/HindIII site of pTV118N (Takara), forming pTV-EL. The wild type groES gene fragment was amplified using PCR, digested with SalI and EcoRI, and ligated into the SalI/EcoRI site of pSTV29, forming pSTV-ES. E. coli MM100 (supplied by Dr. M. Masters) was used for complementation experiments (25). Mutated GroEL and wild type GroES were co-expressed (from expression plasmids pTV-EL and pSTV-ES, respectively) in E. coli MM100.   (20). In the complex, one GroEL subunit (GroEL 1 ) makes contact with two GroES subunits, GroES 1 (magenta) in the known contact region (helices H and I) and GroES 2 (pink) in the undescribed contact region (arrows). The novel contact region in GroEL (the sequence 305 GFKLE 309 in T. thermophilus; shown in red) is located immediately behind helices H and I. The GroES sequences that are interacting with the conventional and novel contact regions are in the same stretch of the loop region of GroES. Side chains of the T. thermophilus GroEL residues (Lys 307 and Leu 308 ) and GroES (T35) are shown as stick models. A magnified view around the highly conserved Leu 308 is shown by a circle. Hydrophobic residues Leu 220 , Val 222 , Val 226 , Leu 232 , Ile 235 , and Ile 300 are drawn as stick models in yellow. B, schematic drawing of the adjacent GroEL-GroES contacts. One GroEL subunit in the heptameric cis-ring interacts with two adjacent GroES subunits, at the H-I helices (black arrows) and at the GXXLE region (red arrows). The two GroES contact sites can communicate with each other through a hydrophobic cluster formed by Leu in the GXXLE region and residues at the entrance formed by the H helix (red arrows).
ATP. The flow rate was 0.5 ml/min, and elution was monitored by an in-line fluorometer (excitation at 550 nm, emission at 570 nm).
Protease Sensitivity of Substrate Protein in the GroEL⅐GroES Complex-GroEL that had been saturated with denatured rhodanese was prepared as described previously (21,27). The reaction mixtures contained 1 mM nucleotide, 10 mM NaF, 2 mM BeCl 2 , 20 mM Na 2 S 2 O 3, 1 M rhodanese-saturated GroEL, 2.0 M GroES, and 1 mM DTT in HKM buffer. Unbound GroES and substrate proteins were removed by ultrafiltration (Microcon YM-100) at 90 min after initiation of the reaction. Chymotrypsin (final concentration, 1 g/ml) and glycerol (final concentration, 10% v/v) were added to 25 l of the mixture containing 1 mM DTT and 15 g of protein in HKM buffer. Following incubation for 20 min at 25°C, components with a molecular mass of Ͻ100 kDa were removed by ultrafiltration (Microcon YM-100). An aliquot of the resulting solution was analyzed by 13% SDS-PAGE. The intensity of band staining was quantified using the NIH Image program and calibrated using known protein concentrations.

Conservation of Novel Contacts in the E. coli GroEL⅐GroES Complex-To investigate whether or not the GXXLE region is in contact
with GroES in the E. coli GroEL⅐GroES complex, we conducted a series of cross-linking experiments using mutants of E. coli GroEL and GroES. We replaced Met 307 , Glu 308 , Leu 309 , and Glu 310 in GroEL, and Thr 28 , Gly 29 , and Ser 30 in GroES with Cys. The removal of DTT from the reaction mixture containing GroEL and GroES led to the generation of a single high molecular mass band, corresponding to cross-linked GroEL-GroES (Fig. 2, A and B). Binding of GroES to GroEL is known to be ATP (or ADP)-dependent, and the cross-linking was only successful when ATP was present. Among the mutants, the combination GroEL(E308C)/GroES(S30C) was most efficiently cross-linked, followed by the combination of GroEL(E308C)/GroES(G29C). We found efficient intrasubunit cross-linking between L309C and E232C at the N terminus of the H helix (Fig. 2C), and cross-link formation was not affected by nucleotides or GroES. These results confirm that the topological arrangement of the GXXLE region in the GroEL⅐GroES complex of E. coli is similar to that of the T. thermophilus complex, where the mobile loop of GroES and the H helix interact with the GXXLE region.

Effect of Mutation of Leu 309 on Growth of E. coli-
The results above indicate that Glu 308 of GroEL and Gly 29 /Ser 30 of GroES are in close proximity. Next we investigated a role for the highly conserved Leu 309 in chaperonin function. We replaced Leu 309 of GroEL with Val, Ala, Asn, Asp, and Lys. E. coli MM100 (25), a strain in which expression of chromosomal GroEL-GroES is under control of the P BAD promoter (arabinose induction), was co-transformed with the expression plasmids encoding the GroEL Leu 309 mutants and wild type GroES. Transformants were cultured on LB plates, in the absence of arabinose (Fig. 3). Cells with expression plasmids encoding the mutants GroEL(L309A) and GroEL(L309V) grew normally, as did cells expressing wild type GroEL. In contrast, as previously described by Fenton et al. (16), cells expressing the mutant GroEL(L309K) could not grow in the absence of arabinose, Similarly, cells expressing the mutants GroEL(L309N) or GroEL(L309D) were unable to rescue GroEL-deficient E. coli MM100. Thus, the GroEL mutants, in which Leu 309 was replaced by polar residues, could not support growth of GroEL-deficient E. coli MM100, indicating a critical role for Leu 309 in chaperonin function in vivo.
Chaperone Activity of the Leu 309 Mutants-The GroEL Leu 309 mutants were purified, and their properties were examined. ATPase

FIGURE 2. Cross-linking between cysteine-incorporated E. coli mutants of GroEL and GroES.
A and B, intermolecular cross-linking between mutants of GroEL and GroES. The mixture containing purified E. coli GroEL and GroES mutants with 1 mM DTT was subjected to ultrafiltration to remove DTT gradually and in the absence (A) or presence (B) of 1 mM ATP to induce cross-linking. C, intrasubunit cross-linking between the GXXLE region and the H helix in GroEL. The GroEL(E232C/L309C) double cysteine mutant was incubated with 20 M CuCl 2 in the absence or presence of GroES and 1 mM nucleotides. DTT was included in the leftmost sample (lane 1). Products were analyzed by SDS-PAGE in the absence of a reducing agent, and gels were stained with Coomassie Blue. activities of the mutants were similar to those of the wild type (data not shown). Protein folding activity was tested using GFP. Denatured GFP was diluted into a solution of GroEL and GroES. Upon dilution, denatured GFP was bound efficiently to the GroEL mutant, because no spontaneous GFP folding occurred (Fig. 4A, inset). GFP started folding upon the addition of ATP, and regardless of mutations, a similar yield of folded protein (ϳ70%) was achieved after 200 s (Fig. 4A). In the parallel experiments, hexokinase was included in the mixtures to eliminate ATP and to prevent the secondary turnover of the GroEL reaction cycle. Excess trap-GroEL (GroEL(N265A)) (8) was added prior to ATP addition, to capture unfolded proteins in the bulk solution. Under these conditions, only folding of proteins encapsulated in the cis-cavity during the first round of the GroEL reaction cycle would be observed. Under the single cycle reaction conditions, the yield of folded GFP differed among the mutants (Fig. 4B). Two mutants, GroEL(L309A) and GroEL(L309V), retained wild type-like folding activity, whereas GroEL(L309N), GroEL(L309D), and GroEL(L309K) gave significantly reduced yields of folded protein. These results indicate that the GroEL mutants in which Leu 309 was replaced by polar residues tend to fail in encapsulating unfolded proteins into the cis-cavity. In addition, we tested the effect of these mutations on the folding of stringent substrate proteins, such as MDH and rhodanese, the folding of which depends on the GroEL/GroES system. GroEL(L309V) and GroEL(L309A) mediated efficient folding of both proteins, with the former giving a better yield of folded protein than wild type GroEL (Fig. 4C). As expected, substitution of Leu 309 by polar residues, particularly Lys and Asp substitutions, resulted in a significant decrease in the yields of reactivated proteins (MDH and rhodanese).
Binding of Substrate Protein and GroES to GroEL(L309K)-In the light of previous investigations of GroEL(L309K) by Fenton et al. (16), we sought to analyze further the properties of GroEL(L309K). The mutant protein was mixed with denatured MDH Cy3 and analyzed using gel filtration HPLC with elution in a buffer containing ATP (Fig. 5A). Like the wild type protein, GroEL(L309K) bound and retained MDH Cy3 . The binding of MDH Cy3 to GroEL(L309K) was not affected by the presence of GroES in the mixture (Fig. 5B). Next we investigated the binding of GroES Cy3 . In the absence of denatured MDH, GroEL(L309K) formed a complex with GroES Cy3 (Fig. 5C). However, in the presence of denatured MDH, GroES Cy3 was primarily eluted as the free GroES heptamer (Fig. 5D). Under the same conditions, the wild type GroEL formed a complex with both MDH Cy3 (Fig. 5B) and GroES Cy3 (Fig. 5D). These results demonstrate that GroEL(L309K) can bind denatured substrate protein or GroES individually but cannot form a GroEL⅐GroES⅐MDH cis-ternary complex that has sufficient stability to survive gel filtration.
cis-Ternary Complexes of GroEL(L309K) Formed in BeFx and Their Sensitivity to Protease Digestion-In general, BeF x can mimic the phosphate of enzyme-bound nucleotides and stabilize transient complexes in ATP-and GTP-metabolizing proteins (28). In the case of GroEL, BeF x stabilizes the cis-ternary complex of GroEL; a 1:2:2 GroEL⅐GroES⅐substrate protein complex with double cis-cavities and a 1:1:2 GroEL⅐GroES⅐substrate protein complex with a single cis-cavity are formed with ATP and ADP, respectively (27). Each cavity contains a substrate protein that is able to fold and that is protected from the attack by protease (27). In anticipation that BeF x would also stabilize the cisternary complex of GroEL(L309K), we included BeF x in the reaction mixtures that contained rhodanese as a substrate protein. The complex was isolated using ultrafiltration and analyzed with SDS-PAGE (Fig. 6,  lanes 1-6). In a control experiment, 1 mol of free wild type GroEL or GroEL(L309K) (without GroES, nucleotide and BeF x ) bound 2 mol of rhodanese (lanes 1 and 2). The GroEL(L309K) complex that formed in FIGURE 4. Chaperone activities of GroEL Leu 309 mutants. Chaperone activity of GroEL (WT) and the mutants, in which Leu 309 of GroEL was replaced by the alternative residues (A, V, N, D, and K). A, GroEL-assisted GFP folding for multiple reaction cycles. Denatured GFP was diluted into buffer containing GroEL, GroES, and 200 mM glucose. ATP (0.8 mM) was added to initiate folding. Recovery of GFP fluorescence was monitored with a fluorometer; GFP fluorescence measurements at 200 s, following the addition of ATP, are shown. Folding in the presence of bovine serum albumin instead of GroEL and GroES is shown as a spontaneous folding (Spont.). Inset, a representative example of the folding assay. B, GroEL-assisted GFP folding for a single reaction cycle. Denatured GFP was diluted as above (A), with 0.04 unit/l hexokinase, and after incubating for 5 min, a GroEL trap mutant (GroEL(N265A)) was added. A single round of folding was initiated by the addition of ATP and monitored as above (A). Other experimental procedures are the same as those in A, except that spontaneous folding was done in the buffer containing the trap-GroEL. Inset, a representative example of the folding assay under the single cycle reaction conditions. C, GroEL-assisted folding of MDH and rhodanese. Denatured MDH or rhodanese were diluted in buffer containing GroEL and GroES. The recovery of activity was measured at 42 min (MDH) and at 40 min (rhodanese), following the initiation of folding, by the addition of ATP. Spontaneous folding (Spont.) levels were determined by measurement of folding in the presence of bovine serum albumin, instead of GroEL and GroES. For the purposes of comparison, the intensity of fluorescence (GFP) or activity (rhodanese and MDH) of the same amounts of native protein was considered to be 100% folding.
the presence of ATP and BeF x , GroEL(L309K)⅐GroES⅐rhodanese, had an apparent composition of 1:2:1 (lane 4). This indicated that, for the GroEL(L309K) mutant, one of the two rhodanese molecules in the 1:2:2 GroEL⅐GroES⅐rhodanese complex (lane 3) had dissociated. The GroEL(L309K) complex, formed in the presence of ADP and BeF x , had an apparent composition of 1:1:2 (GroEL(L309K)⅐GroES⅐rhodanese) (lane 6). This composition is the same as found for wild type (lane 5). In this complex, one substrate occupies the cis-cavity, whereas the other is positioned on the opposite (trans) GroEL ring, without GroES. In parallel experiments, the isolated complexes were treated with chymotrypsin, reisolated using ultrafiltration, and analyzed with SDS-PAGE (Fig. 6,  lanes 7-12). The rhodanese molecules that had bound to free GroEL (wild type or GroEL(L309K)) were digested completely (lanes 7 and 8). Rhodanese molecules in the 1:2:2 GroEL⅐GroES⅐rhodanese complex were fully protected from digestion (lane 9). Rhodanese molecules in the 1:2:1 GroEL(L309K)⅐GroES⅐rhodanese complex were digested completely (lane 10). For the complexes formed in the presence of ADP and BeF x , only one rhodanese molecule in the wild type 1:1:2 GroEL⅐GroES⅐rhodanese complex was digested (lane 11). Both rhodanese molecules in the 1:1:2 GroEL(L309K)⅐GroES⅐rhodanese complex were digested (lane 12). Thus, the substrate proteins in the GroEL(L309K)⅐GroES complexes are all accessible to attack by chymotrypsin. Similar results were obtained when MDH was used as a substrate protein (data not shown). It appears, therefore, that GroEL(L309K) can form a relatively stable ternary complex with GroES and substrate proteins in the presence of BeF x ; however, it cannot prop-  2) or GroES(S30C) and GroEL(E308C/E309K), in which a second mutation, GroEL(E308C), was introduced into GroEL(L309K) (lanes 3 and 4). The mixture containing 1 mM DTT and the two mutants was subjected to ultrafiltration as described in the legend to Fig. 2 in the presence of either ATP (lanes 1 and 3) or ATP and BeF x (lanes 2 and 4). FIGURE 5. Binding of substrate protein and GroES to GroEL(L309K) and wild type GroEL. A and B, Binding of MDH Cy3 to wild type (WT) GroEL or GroEL(L309K) in the absence (A) or presence (B) of GroES. Denatured MDH Cy3 was diluted in the buffer containing GroEL and incubated for 2 min. GroES was added to B, and ATP was added to both reactions, which were incubated at 25°C for 2 min. The mixtures were applied to a gel filtration HPLC column and eluted with a buffer containing 0.2 mM ATP. Fluorescence (excitation at 550 nm, emission at 570 nm) was monitored in-line. GroEL and complexes with GroES with the substrate or with both GroES and substrate were eluted at approximately 12 min (indicated by a dotted line). C and D, binding of GroES Cy3 to GroEL in the absence (C) or presence (D) of denatured MDH. The reactions were prepared as above, and then GroEL and free GroES Cy3 were eluted at 12 and 16 min, respectively (indicated by dotted lines). FIGURE 6. Sensitivity of substrate protein trapped in the GroEL complexes to protease digestion. GroES, ATP, and BeF x or GroES, ADP, and BeF x were added to GroEL (for both wild type (WT) and the L309K mutant) that had been saturated with denatured rhodanese. After 90 min, the aliquots underwent one of the two following treatments: ultrafiltration (100-kDa cut) and SDS-PAGE (lanes 1-6); or ultrafiltration (100-kDa cut), chymotrypsin treatment, a second ultrafiltration (100-kDa cut), and SDS-PAGE (lanes 7-12). The gels were stained with Coomassie Blue.