Hydrophilic Residues 526KNDAAD531 in the Flexible C-terminal Region of the Chaperonin GroEL Are Critical for Substrate Protein Folding within the Central Cavity*

The final 23 residues in the C-terminal region of Escherichia coli GroEL are invisible in crystallographic analyses due to high flexibility. To probe the functional role of these residues in the chaperonin mechanism, we generated and characterized C-terminal truncated, double ring, and single ring mutants of GroEL. The ability to assist the refolding of substrate proteins rhodanese and malate dehydrogenase decreased suddenly when 23 amino acids were truncated, indicating that a sudden change in the environment within the central cavity had occurred. From further experiments and analyses of the hydropathy of the C-terminal region, we focused on the hydrophilicity of the sequence region 526KNDAAD531 and generated two GroEL mutants where these residues were changed to a neutral hydropathy sequence (526GGGAAG531) and a hydrophobic sequence (526IGIAAI531), respectively. Very interestingly, the two mutants were found to be defective in function both in vitro and in vivo. Deterioration of function was not observed in mutants where this region was replaced by a scrambled (526NKADDA531) or homologous (526RQEGGE531) sequence, indicating that the hydrophilicity of this sequence was important. These results highlight the importance of the hydrophilic nature of 526KNDAAD531 residues in the flexible C-terminal region for proper protein folding within the central cavity of GroEL.

The final 23 residues in the C-terminal region of Escherichia coli GroEL are invisible in crystallographic analyses due to high flexibility. To probe the functional role of these residues in the chaperonin mechanism, we generated and characterized C-terminal truncated, double ring, and single ring mutants of GroEL. The ability to assist the refolding of substrate proteins rhodanese and malate dehydrogenase decreased suddenly when 23 amino acids were truncated, indicating that a sudden change in the environment within the central cavity had occurred. From further experiments and analyses of the hydropathy of the C-terminal region, we focused on the hydrophilicity of the sequence region 526 KNDAAD 531 and generated two GroEL mutants where these residues were changed to a neutral hydropathy sequence ( 526 GGGAAG 531 ) and a hydrophobic sequence ( 526 IGIAAI 531 ), respectively. Very interestingly, the two mutants were found to be defective in function both in vitro and in vivo. Deterioration of function was not observed in mutants where this region was replaced by a scrambled ( 526 NKADDA 531 ) or homologous ( 526 RQEGGE 531 ) sequence, indicating that the hydrophilicity of this sequence was important. These results highlight the importance of the hydrophilic nature of 526 KNDAAD 531 residues in the flexible C-terminal region for proper protein folding within the central cavity of GroEL.
The chaperonin GroEL (14-mer) from Escherichia coli binds denatured proteins and facilitates their folding in vivo and in vitro by encapsulating them within an isolated cavity formed in cooperation with the co-chaperonin GroES (7-mer) (1,2). Encapsulation by GroEL protects the denatured proteins from interactions with other misfolded proteins or aggregation prone species, providing the proper environment in which the denatured protein may fold spontaneously (3)(4)(5)(6). The unique quaternary structure (two heptameric rings stacked back to back) of GroEL enables a clever mechanism. The subunit structure (548 amino acid residues) is divided into three domains; the apical domain, the intermediate domain, and the equatorial domain. Each domain has a specific role in the chaperonin mechanism. The apical domain plays an important role in recognizing and binding denatured protein and the co-chaperonin GroES. The intermediate domain connects the apical and the equatorial domains, and the equatorial domain binds ATP and hydrolyzes it. This ATP hydrolysis controls the overall chaperonin mechanism, regulating binding and release of the substrate protein and GroES (7)(8)(9)(10)(11). The refolding substrate protein is encapsulated by GroEL-GroES and segregated from the surrounding environment and, under these conditions, folds correctly without forming aggregation.
Thus, the mechanism of GroEL-mediated protein folding is well characterized. However, some details regarding the specific roles of various structural elements in the GroEL subunit structure remain unclear. For example, as shown in Fig. 1, the final 23 amino acid residues of the C terminus are not clearly defined in x-ray crystallographic studies (9) due to high flexibility. However, these segments of GroEL oligomer appear to coalesce and block the central channel at the level of the equatorial domain in electron micrograph (12) and small-angle neutron scattering (13) experiments. Previously, it was reported that this C-terminal region is closely involved in the rate of ATP hydrolysis (14) and GroEL oligomerization (15,16). However, a mutant with 27-amino acid residues truncated from the C terminus was still functional and could support normal growth of the host cell (15,16). Moreover, the role of the C-terminal region in the mitochondrial homologue Hsp60 (572 amino acid residues) from Saccharomyces cerevisiae was studied by using C-terminal truncation and insertion mutants (17). From the results of refolding assays using these mutants, it was concluded that the predominantly hydrophobic residues in the C-terminal region are important for substrate folding inside the central cavity.
These results indicated that the C-terminal region of chaperonin is involved in multiple facets of the chaperonin mechanism, i.e. in ATP hydrolysis activity, stabilization of chaperonin quaternary structure, and substrate protein folding. However, exactly how these C-terminal amino acids participate in the chaperonin mechanism and, more importantly, which specific residues within this region are important are still unknown. Very recently, Tang et al. (18) showed that the chaperonin cage provided a physical environment optimized to catalyze the folding reaction of proteins within a specific size range by facilitating folding kinetics. This was determined by using GroEL mutants with elongated C-terminal amino acid residues. However, Farr et al. (19) performed similar experiments and reported no acceleration in folding rate in these tail-elongated single-ring GroEL mutants. This discrepancy may also be resolved by additional detailed information regarding the flexible C-terminal region.
In this study, we focused particularly on the final 23 amino acid residues ( 526 KNDAADLGAAGGMGGMGGMGGMM 548 ) at the C-terminal region of E. coli GroEL, which are invisible in x-ray crystal structures, and attempted to understand the role of this region in chaperonin function. To this end, we generated three types of C-terminal amino acid-truncated mutants (dC541, dC531, and dC525), in which the final 7, 17, and 23 residues were, respectively, truncated from the C terminus. From detailed studies on structural and functional characteristics of the mutants, the importance of a hydrophilic amino acid region, 526 KNDAAD 531 was highlighted. The critical importance of the hydrophilic nature of this region for chaperonin function was confirmed by modifying this sequence in a controlled manner. These results showed clearly that the hydrophilic nature of 526 KNDAAD 531 is critical in maintaining a proper environment for substrate protein folding within the central cavity. This is the first report that identifies specific amino acid residues within the crystallographically disordered C-terminal region that are critical to chaperonin function.

EXPERIMENTAL PROCEDURES
Preparation of Mutant Proteins-C-terminal truncated mutants of double ring and single ring GroEL, and NE, HB, SC, and SH mutants of double and single ring full-length GroEL were constructed by using the QuikChange site-directed mutagenesis kit (Stratagene), using either pUCESL (wild-type GroEL) (10) or pEL-SR1 (GroEL SR-1) (20) as the template. Amino acid sequences of the C-terminal region of all chaperonins used in this study are summarized in Table 1. The successful construction of each mutant was confirmed by DNA sequence analysis of the entire GroEL coding region, and protein expression was checked by SDS-PAGE (7.5% polyacrylamide gel).
Wild-type and all mutant proteins were expressed in E. coli JM109(DE3) and purified at room temperature as follows. After ultrasonic disruption and removal of nucleic acids by addition of streptomycin, proteins were recovered by ammonium sulfate precipitation (55%) and loaded onto a Superdex 200 HR 10/30 gel-filtration column (GE Healthcare Life Sciences), which had been equilibrated with purification buffer (50 mM Tris-HCl, pH 7.8, containing 2 mM EDTA, 2 mM dithiothreitol, and 20% methanol), at a flow rate of 0.5 ml/min. GroEL fractions were pooled and loaded separately onto a Resource-Q (6 ml) anionexchange column (GE Healthcare Life Sciences), which had been equilibrated with the purification buffer and eluted with a linear gradient (0 -0.5 M NaCl) at a flow rate of 2 ml/min. The purified GroEL proteins were desalted by dialysis against the purification buffer thoroughly, and then concentrated by a Centricon YM-100 centrifugal filter device (Millipore).
CD and 1-Anilino-8-naphthalenesulfonic Acid (ANS) 2 Fluorescence Measurements-CD spectra were measured on a Jasco J-720 spectropolarimeter equipped with a constant-temperature cell holder at 25°C. The buffer used was 50 mM Tris-HCl, pH 7.5, and protein concentrations of wild-type GroEL and C-terminal truncated mutants were 50 g/ml during measurement.
Measurements of ANS binding fluorescence were performed on a Jasco FP-6300 fluorescence spectrophotometer equipped with a constant-temperature cell holder at 25°C. GroEL and C-terminal truncated mutants (0.46 M) were mixed with ANS (5 M) in 50 mM Tris-HCl buffer, pH 7.5. The excitation wavelength was 371 nm, and the emission spectra were recorded at 400 -550 nm.
FRET from the FITC-labeled rhodanese bound on the apical domain to the TMR5MA-labeled mutants were measured by using the Jasco FP-6300 fluorescence spectrophotometer with an excitation wavelength of 494 nm. The emission spectra were monitored at 500 -650 nm. The FRET efficiency, E, is given by E ϭ 1 Ϫ F DA /F D , where F DA and F D are the fluorescence in the presence and absence of acceptor, respectively (23). Fluorescence peak areas of F DA (554 -600 nm) and F D (500 -546 nm) were used for determining the efficiency.
ATPase Activity Measurements-ATPase activities of GroEL proteins (0.1 M) were measured at 37°C, using the method described previously (24,25), in the presence or absence of GroES (0.1 M).
Refolding Assays-Refolding assays of rhodanese from bovine liver (Sigma) (25), enolase from yeast (Wako) (26), and malate dehydrogenase from pig heart (Roche Applied Science) (27) were carried out as previously reported. Refolding of ␣-glucosidase from S. cerevisiae (Wako) was performed as follows. ␣-Glucosidase (15 M) was denatured in 10 mM phosphate buffer, pH 6.8, containing 1 mM EDTA and 6 M guanidine hydrochloride at 25°C for 1 h. The refolding reaction was started by a 100-fold dilution into 100 mM Tris-HCl buffer, pH 7.6, containing 10 mM KCl and 10 mM Mg(CH 3 COO) 2 . The temperature and protein concentration of ␣-glucosidase during the refolding reaction were 25°C and 0.15 M, respectively. A 1.5-fold molar excess of GroEL (or C-terminal truncated mutants) and GroES oligomers relative to ␣-glucosidase were present in the refolding reaction mixtures. After 5-min incubation, ATP was added to a final concentration of 1 mM. At appropriate times during the refolding reaction, 200 l of the mixture was withdrawn and assayed for ␣-glucosidase activity by monitoring the conversion of p-nitrophenyl-␣-D-glucopyranoside into p-nitrophenol and D-glucose at 405 nm for 1 min. Refolding yield was determined as the percentage ratio of the specific activity of the refolding enzyme relative to that of native enzyme. Refolding assays for rhodanese in the presence of SR1 and truncated SR1 mutants was carried out as previously reported (25), except that in each case GroEL was exchanged for the SR1 variant.
Assessment of Trapping Ability of Rhodanese Refolding Intermediates-To examine whether the amounts of rhodanese encapsulated in the EL-WT cavity and EL-dC525 cavity were different, GroEL saturated with heat-denatured rhodanese was subjected to proteinase K digestion, as described previously (28,29). First, we incubated 2 M rhodanese at 60°C for 15 min in buffer (50 mM HEPES-KOH, pH 7.5, containing 100 mM KCl, 5 mM MgCl 2 , and 1 mM dithiothreitol) containing 0.5 M GroEL and 1 M GroES. The temperature was then shifted down to 25°C and NaF, BeCl 2 , and ATP (final concentrations, 10 mM, 1 mM, and 1 mM, respectively) were added. After 10-min incubation at 25°C, proteinase K was added to a final concentration of 1 g/ml. After 30-min incubation at 25°C, phenylmethanesulfonyl fluoride was added to a final concentration of 1 mM, and subsequently GroELrhodanese-GroES ternary complex was isolated by using a 100-kDa cut centrifugal filter device (Centricon YM-100, Millipore) and analyzed by SDS-PAGE (15% polyacrylamide gel). The quantitation of the bands was performed using the public domain software NIH Image (National Institutes of Health, available from rsb.info.nih.gov/nih-image/download.html).

Structural Characteristics and ATPase Activity of C-terminal Truncated GroEL Mutants
Using site-directed mutagenesis, we generated successive C-terminal truncated mutants of double ring GroEL (EL-dC541, EL-dC531, and EL-dC525), single ring variants of these mutants (SR1-dC541, SR1-dC531, and SR1-dC525), and analogs of each double ring mutant where the C-terminal amino acid was changed to cysteine (EL-WT/Met 548 3 Cys, EL-dC541/Met 541 3 Cys, EL-dC531/Asp 531 3 Cys, and EL-dC525/Pro 525 3 Cys). Amino acid sequences of the C-terminal region of each mutant are summarized in Table  1. Truncations were confirmed by analyzing subunit molecular weights with SDS-PAGE ( Fig. 2A). The extent of GroEL expression was similar for all the mutants constructed. We further examined the oligomeric state of each mutant by native-PAGE and size-exclusion chromatography (not shown), indicating that all of the C-terminal truncated mutants formed oligomeric states similar to wild-type GroEL. As shown in Fig. 2B, CD measurements also showed that the overall secondary structures of all of the mutants were quite similar to wild-type GroEL. This suggests that the flexible C-terminal region does not form a definite secondary structure. Thus, we concluded that deletions of up to 23 residues from the C-terminal amino acid residue of GroEL have no effect on overall chaperonin structure. This was consistent with results reported previously (15,16).
As a preliminary check of chaperonin function, we first examined the ATPase activities of all of the purified C-terminal truncated mutants in the presence or absence of an equimolar concentration of GroES (Fig. 2C). In the absence of GroES, the ATPase activities of all of the double ring GroEL mutants were decreased to about one-half of wild-type GroEL (EL-WT), suggesting that the ATP hydrolysis activity is disrupted by C-terminal truncation, as reported previously (14,30). Interestingly, Farr et al. (19) demonstrated that the ATPase activity was promoted by C-terminal extension (Gly-Gly-Met repeats). It is interesting to note that the C-terminal amino acid residues influence the ATP hydrolysis activity of GroEL in such a manner. In the presence of GroES, the ATPase activities of all of the mutants were additionally suppressed by ϳ50%, similar to EL-WT, suggesting that in each case GroEL and GroES formed GroEL-GroES complexes (31). These results suggested that the interactions of these mutants with GroES were all similar to EL-WT. Contrary to this, the ATPase activities of all of the single ring C-terminal truncated GroEL mutants (SR1-dCn) were almost the same as that of wild-type (SR1-WT), presumably due to missing inter-ring contacts. The ATPase activities of all of the SR1 mutants were strongly suppressed by the addition of GroES, which is also in agreement with the behavior of SR1-WT chaperonin (32).

FRET between Substrate Protein Bound to the Apical Domain and the Truncated C Terminus
To observe the orientation of the flexible C-terminal region of GroEL inside the central cavity, we performed FRET experiments. If the flexible C-terminal polypeptide normally extends and protrudes toward the interior of the central cavity, FRET efficiencies between the C-terminal residue and substrate protein bound at the apical domain would be expected to decrease in proportion to the length of amino acid residues truncated in our mutants. We used FITC to label a substrate protein (rhodanese, fluorescence donor) and TMR5MA (fluorescence acceptor) to label GroEL truncation mutants at the C-terminal cysteine residue introduced into EL-WT/Met 548 3 Cys, EL-dC531/Asp 531 3 Cys, and EL-dC525/Pro 525 3 Cys. FRET between fluorescein and tetramethylrhodamine moieties introduced to Cys 242 located at the top of the apical domain and Cys 527 at the vicinity of the C-terminal residue of GroEL was successfully observed in a previous study (33). Although wildtype GroEL has three cysteine residues at positions 138, 458, and 519, only the newly introduced cysteine residue at the C-terminal was labeled with fluorescence dyes under the relatively mild experimental conditions we used, because EL-WT was not labeled at all under the same conditions (Table 2). We measured FRET efficiencies between the FITC-labeled rhodanese bound to the apical domain and the TMR5MA-labeled C-terminal of wild-type chaperonin and mutants as shown in Fig. 3 (A-C) and determined the fluorescence energy transfer efficiencies (E). Interestingly, as shown in Fig. 3D, the FRET efficiencies were almost the same for all three truncated mutants. In our present experiments, we cannot estimate the distance between the fluorescein and the tetramethylrhodamine moieties quantitatively, because the specific positions of those introduced fluorescent dyes in the GroEL-refolding rhodanese complex cannot be determined topologically. However, FIGURE 2. Structural characteristics and ATPase activity of wild-type GroEL and C-terminal truncated mutants. A, SDS-PAGE (7.5% polyacrylamide gel). B, CD spectra of EL-WT and the truncated mutants in 50 mM Tris-HCl buffer, pH 7.5, at 25°C. The protein concentrations were 50 g/ml. EL-WT (circles), EL-dC541 (squares), EL-dC531 (diamonds), and EL-dC525 (triangles). C, ATPase activities of double and single ring GroEL, and their C-terminal truncated mutants in the absence and presence of GroES. Amounts of released inorganic phosphates during 1-h incubation at 37°C were determined.

TABLE 1 Amino acid sequence of the C-terminal region of various GroEL chaperonins used in this study
a The C-terminal amino acid sequence of wild-type GroEL (or single ring GroEL) and the truncated mutants. b The C-terminal amino acid sequences of the wild-type GroEL and truncated mutants whose C-terminal residue was changed to cysteine residue (marked by an underline). c The C-terminal amino acid sequence of NE, HB, SC, and SH mutants. The residues between 526 -531 are represented by bold and changed amino acids are marked by an underline.

TABLE 2 Number of fluorescent probes introduced per GroEL 14-mer
The number of fluorescent probes introduced per GroEL 14-mer was determined by using a molar extinction coefficient of TMR5MA at 541 nm (⑀ ϭ 95,000 cm Ϫ1 M Ϫ1 ).

Importance of Hydrophilic Site in GroEL C Terminus
we can evaluate qualitatively any large changes in the overall FRET efficiency. Thus, this result suggests that the distance between FITC-labeled rhodanese bound to the apical domain and the TMR5MA moiety at the C-terminal residue was independent of the length of C-terminal amino acid deletion. Our results also supported the notion that the C-terminal region of the wild-type GroEL does not protrude into the central cavity space, but rather forms a compact conformation that acts as a wall separating the double rings at the bottom of the cavity, as suggested previously (12,13).

Chaperonin Activities of Truncated Mutants
Double Ring GroEL Mutants-We evaluated the relative abilities of all the truncated double ring GroEL mutants to facilitate protein folding in the presence of ATP and GroES at 25°C compared with EL-WT. As substrate proteins, we used four different proteins, each of which displays distinct characteristics during refolding in the presence of GroEL/GroES. The first substrate protein used was rhodanese from bovine liver (monomer, M r 33,000), whose refolding yield absolutely depends on GroEL/GroES under "non-permissive" conditions (25). The second substrate protein, enolase from yeast (dimer, subunit M r 47,000), is a protein that shows a high spontaneous folding yield but whose refolding intermediate is trapped efficiently by GroEL (26). The third substrate protein, malate dehydrogenase from pig heart (dimer, subunit M r 35,000), is a protein that shows no spontaneous folding and whose refolding yield absolutely depends on GroEL/GroES (27,34). The fourth substrate protein, ␣-glucosidase from S. cerevisiae (monomer, M r 68,000), is a protein that is too large to be encapsulated within the central cavity but whose successful refolding depends on GroEL/GroES (35), presumably through a trans binding-mechanism (36). Typical refolding characteristics of rhodanese and ␣-glucosidase in the presence of wild-type and mutant chaperonins are shown in Fig. 4 (A and B), respectively. All data are summarized in Fig. 4C. Interestingly, in the case of the stringent proteins, rhodanese, and malate dehydrogenase, the refolding yields in the presence of chaperonin decreased markedly in the presence of the 23-amino acid truncated mutant (EL-dC525) relative to the other mutants and EL-WT. The specific amount of decrease was different depending on the substrate protein.
As can be seen in Table 1, the difference between EL-dC525 and EL-dC531 consists of the presence of residues 526 KNDAAD 531 at the C terminus of the latter mutant, which highlights the importance of these residues in the successful folding of stringent substrate proteins. In the case of ␣-glucosidase, refolding yields decreased gradually in proportion to the length of amino acid residues truncated. Finally, the refolding yield of enolase was almost independent of the truncations. This seemed to be reasonable considering that the refolding intermediates of enolase can interact with GroEL in the absence of ATP but refold with a high yield without the commitment of GroEL and GroES after release from the chaperonin (26).
Such results would also be observed if the truncated mutants have a decreased ability to initially trap (bind) refolding substrate intermediates. To see if the mutants were capable of trapping the refolding substrate as efficiently as the wild-type GroEL, we examined EL-dC525 for its ability to trap and encapsulate rhodanese in the presence of GroES, ATP, and BeF x . Under these conditions, a stable ternary complex of GroELrhodanese-GroES is formed (29). Isolation of the complex by using a 100-kDa cut centrifugal filter device was followed by SDS-PAGE analysis as shown in Fig. 4D. Even after the proteinase K treatment, a rhodanese band was observed in the GroELcontaining fractions. When GroES was absent during protease digestion, no rhodanese was observed. This demonstrated that rhodanese trapped by EL-dC525 was encapsulated within the GroEL-GroES cavity. The amount of rhodanese bound to EL-dC525 and the degree of protection from proteinase K digestion provided by EL-dC525 were almost the same as EL-WT (Fig. 4D), indicating that EL-dC525 was not impaired in the ability to trap rhodanese during refolding. These results suggested strongly that the decreased ability of these chaperonins to facilitate refolding was not due to a deficiency in the substrate trapping ability of the mutant or an impaired chaperonin cycle but rather arose from an environmental change inside the central cavity caused by deletion of the C-terminal amino acid residues. In other words, the residues of the C-terminal actively contribute in maintaining the environment of the central cavity for proper protein folding.
Single Ring GroEL Mutants-To assess the ability of proteins to fold within the chaperonin cavity during a single turn over cycle, we further examined refolding of rhodanese by using single ring GroEL variants of the C-terminal truncated mutants (SR1-dCn). As shown in Fig. 5A, due to the single turn over conditions formed by SR1-GroES cavity, the refolding yield of rhodanese even in the presence of SR1-WT was relatively low compared with the double ring wild-type GroEL (EL-WT). As shown in the figure, although SR1-dC541 and SR1-dC531 showed ϳ80% refolding yields relative to SR1-WT, the refolding yield of SR1-dC525 was not significantly different from the spontaneous yield, within the limits of error of our experiments. This result also clearly demonstrated that the deletion of the 23 amino acid residues greatly decreased the ability of proteins to refold inside the central cavity. To exclude the possibility of a reduced binding ability on the part of SR1-dC525, we analyzed the amount of rhodanese bound to SR1-dC525 using SDS-PAGE (Fig. 5B). The amount of rhodanese bound to SR1-dC525 was comparable to SR1-WT, as shown.

Hydrophobic Fluorescent Probe Binding Experiments
To understand how the C-terminal region contributes to the cavity environment, we examined the hydrophobicity of all truncated mutants, influenced by previous studies on the C-ter- triangles, EL-dC525; crosses, spontaneous refolding. C, summary of all refolding assays at 25°C. The refolding yield in the presence of EL-WT and GroES after 30 min (enolase) or 60 min (rhodanese, malate dehydrogenase, and ␣-glucosidase) incubation was set to 100%. Error bars are estimated using data from at least three independent experiments. D, the ternary complexes prepared as described under "Experimental Procedures" were treated with or without proteinase K. Amounts of 40 g of protein were loaded on the each lane. The amount of rhodanese encapsulated within GroEL-GroES was determined by densitometry and represented as a relative yield to EL-WT. minal region of Hsp60 homologue from S. cerevisiae, which showed that the predominantly hydrophobic C terminus of Hsp60 is a very important factor for proper protein folding inside the central cavity (17). The overall hydrophobicity of the EL-WT and all truncated double ring GroEL mutants were first assessed by ANS binding experiments. ANS is a probe that binds to hydrophobic patches and emits strong fluorescence (37). Very interestingly, as shown in Fig. 6A, ANS fluorescence intensities were increased slightly for EL-dC531 and greatly for EL-dC525 compared with EL-WT.
Next, to evaluate the hydrophobicity in the GroEL-GroES capsule state, we performed the same experiments using SR1 truncation mutants in the presence of GroES and ATP. As shown in Fig. 6B, whereas SR1-dC541 and SR1-dC531 showed almost the same ANS fluorescence as SR1-WT, SR1-dC525 displayed a large enhanced fluorescence, i.e. upon deletion of 23 amino acid residues, the hydrophobicity inside the cavity of SR1-GroES was increased. Again, because the difference between dC525 and dC531 amounted to the presence of 526 KNDAAD 531 in the latter mutant, these findings suggested that deletion of the charged and polar 526 KNDAAD 531 site in the C-terminal region within the cavity resulted in the conversion to a more hydrophobic environment.
Elucidation of Importance for GroEL Function of the Hydrophilic 526 KNDAAD 531 Residues in the Flexible C-terminal Region-To probe further the above results suggesting a sharply increased hydrophobic character inside the cavity of the truncation mutants EL-dC525 and SR1-dC525, the hydropathy index (hydrophobic and hydrophilic character) of the C-terminal amino acid region was examined by Kyte and Doolittle hydropathy analysis (38). As shown in Fig. 7A, it was found that the region highlighted in the experiments shown in Figs. 4 -6 ( 526 KNDAAD 531 ) represent a highly hydrophilic area of the GroEL C terminus. This hydrophilic region is excised  This finding prompted us to evaluate directly the importance of the hydrophilic 526 KNDAAD 531 region in chaperonin function, and we prepared two additional double ring or single ring GroEL mutants, whose initial 526 KNDAAD 531 sequence was altered to 526 GGGAAG 531 (neutral hydropathy sequence (NE)) and 526 IGIAAI 531 (hydrophobic sequence (HB)), respectively. As shown in Fig. 7 (B and C), the hydropathic characters around positions 526 -531 of NE and HB mutants were, respectively, shifted to a more hydrophobic nature compared with wild-type, i.e. to neutral and hydrophobic from hydrophilic.
The double ring and single ring NE and HB mutants (EL-NE, EL-HB, SR1-NE, and SR1-HB) were all purified, and the quaternary structures were confirmed to be normal. First, as shown in Fig. 8A, ANS binding experiments confirmed that SR1-NE was more hydrophobic than SR1-WT, and SR1-HB was the most hydrophobic of the three. To examine the direct effects of hydrophilicity change, refolding assays for rhodanese in the presence of SR1-NE and SR1-HB were performed. As shown in Fig. 8B, SR1-NE showed a refolding ability similar to SR1-dC525, and SR1-HB displayed a yield similar to spontaneous refolding. Refolding abilities of EL-NE and EL-HB for rhodanese were also similar to those of SR1-NE and SR1-HB, respectively (data not shown). In additional experiments, it was also confirmed that the low rhodanese refolding yields were not attributed to the differences in trapping/encapsulation yield of the refolding intermediates by the SR1-NE and SR1-HB (Fig.  8C). Finally, we tested whether EL-NE and EL-HB could rescue a conditionally GroE-deficient strain, E. coli KY1156, as described previously for GroEL C138W (39). When transformation of this strain with a plasmid that, respectively, encoded EL-WT, EL-NE, and EL-HB was performed and the transformants were cultivated at 37°C for 8 h, growth rates of EL-NE and EL-HB were notably slowed relative to EL-WT, as shown in Fig. 8D. This result suggested that EL-NE and EL-HB were also partially defective in function within E. coli cells.
Furthermore, to determine whether the hydrophilic nature of 526 KNDAAD 531 was important, or rather, the specific sequence of this region was necessary for chaperonin function, we prepared scrambled (SC) ( 526 NKADDA 531 ) and similar hydrophilic (SH) substitution ( 526 RQEGGE 531 ) mutants of this sequence and assayed the chaperonin function in vitro and in vivo. Hydropathic characteristics around positions 526 -531 as predicted by Kyte and Doolittle hydropathy analysis (38) and hydrophilicity examined by ANS binding experiments of SC and SH mutants were similar to those of wild-type GroEL (data not shown). As shown in Fig. 9, SR1-SC and SR1-SH mutants showed similar rhodanese refolding activities to wild-type SR1, and EL-SC and EL-SH mutants substituted for wild-type GroEL in E. coli KY1156 cells under restricted conditions, indicating that in fact the hydrophilic nature of this region was important for chaperonin function.

DISCUSSION
The C-terminal 23 amino acid residues of chaperonin GroEL are not defined structurally from x-ray crystallographic studies due to high flexibility (9). The specific role of this flexible C-terminal region in chaperonin function is still not clear; how are the C-terminal 23 amino acid residues involved in the protein folding mechanism specifically? In this study, to clarify this interesting question involving the role of these residues in chaperonin function, we generated various C-terminal amino acid truncated mutants and examined their characteristics in great detail. Because the importance of the hydrophilic 526 KNDAAD 531 region was apparent in our initial results, we then generated NE ( 526 GGGAAG 531 ), HB ( 526 IGIAAI 531 ), SC ( 526 NKADDA 531 ), and SH ( 526 RQEGGE 531 ) substitution mutants of this sequence region, to prove the importance in chaperonin function. The results of hydropathic analysis, the refolding assays for rhodanese of these single ring GroEL mutants, and in vivo activity of double ring GroEL mutants proved clearly that the hydrophilic nature of the 526 KNDAAD 531 segment contributes in maintaining a hydrophilic environment of the central cavity necessary for proper folding of protein. This is the first report that demonstrates an active involvement of specific hydrophilic residues within the flexible C-terminal region in chaperonin function.
Hydrophilicity of the C-terminal Region Is Closely Associated with Productive Protein Folding-The essence of chaperonin function is to sequester unfolded proteins away from misfolded and aggregation prone species (40) by providing the "Anfinsen's cage" (41). During chaperonin-mediated folding, many unfolded proteins fold spontaneously in the hydrophilic chaperonin cavity (3,4,32). The importance of the central cavity (chamber) was demonstrated in experiments comparing the chaperonin ability of intact chaperones versus mini-chaperones that cannot form the chamber; the intact chamber was absolutely required for the successful folding of several substrate proteins that could not fold spontaneously in the bulk solution (42).
In the present study, we clarified that the flexible C-terminal region also plays an important role to maintain a hydrophilic environment at the bottom of the chamber for productive protein folding. In previous studies, the C-terminal regions of yeast Hsp60 homologue with predominantly hydrophobic amino acid residues (17) and of E. coli GroEL with moderate hydrophobic amino acid residues (Gly-Gly-Met repeat) (16,18,43) have been shown to be important for protein folding. In this study, we reveal the importance of this region in more detail. As shown in Figs. 6 and 7A, a hydrophilic area ( 526 KNDAAD 531 ) that exists in the mutants dC541 and dC531 but is absent in the mutant dC525 was found to be of great importance. The specific nature regarding the role of this sequence region was confirmed using substitution mutants that altered the hydrophilicity of this sequence; the refolding of rhodanese in the presence of the NE and HB mutant chaperonins displayed a reduction in refolding yield that correlated with an increase in the relative hydrophobicity of the amino acid sequence from 526 to 531 (Fig. 8). This tendency was also observed in the refolding assays of malate dehydrogenase, albeit performed in the presence of the truncation mutant dC525 (Fig. 4). Moreover, both EL-NE and EL-HB contributed to a slow growth rate in E. coli cells (Fig.  8D). It seems that changes that cause a shift toward a hydrophobic environment in the vicinity of the C-terminal region of GroEL result in significant impairment of the chaperonin mechanism. Furthermore, the importance of the hydrophilic nature of this 526 KNDAAD 531 segment was confirmed by assessing the function of the SC and SH mutant chaperonins both in vitro and in vivo (Fig. 9). The 526 KNDAAD 531 segment therefore acts to promote a hydrophilic environment at the "floor" of the chaperonin chamber that is crucial to successful protein folding assistance.
Regarding the importance of hydrophilic residues in the interior of the GroEL chamber, previous studies have noted the existence of strategically placed hydrophilic residues in other positions. The importance of hydrophilic residues that line the interior of the GroEL roof portion was recently highlighted in studies performed by Tang et al. (18). Additional examples where substitution of certain hydrophobic amino acid residues to more hydrophilic residues result in an increase in the refolding yield of certain proteins have also been reported, as was the case for Tyr 71 of GroES (44). In the case of the present study and the above case of Tyr 71 in GroES, which is localized at the inner surface of the chamber lid, the facts that both regions are more or less static and (presumably) not significantly altered by various conformational changes in the GroEL subunit may result in a preference in these regions for hydrophilic amino acid residues that maintain a generally polar environment.
Possible Consequences of C-terminal Truncation to Inter-ring Communication in GroEL-It is interesting to note that a region that is important in modulating the inner environment of the chaperonin chamber may be found adjacent to another sequence region of the C terminus that was deemed important to the chaperonin mechanism. Previous studies have demonstrated that perturbing the length of the Gly-Gly-Met repeat (extending the length of the C terminus) results in alterations in the ATPase activity and overall protein folding assistance ability of GroEL (19). In our experiments, we found evidence to support the idea that the GroEL C-terminal amino acid region was localized more or less at the bottom of the chamber. This would imply that both the 526 KNDAAD 531 segment identified in our experiments and the following signature Gly-Gly-Met repeat segment (536 -547) of the GroEL C-terminal are placed in close proximity, on the floor of the chamber. The importance of these two segments in the chaperonin mechanism, although qualitatively different, seems well established; it remains to be FIGURE 9. Functional characterization of SR1-SC and SR1-SH, and in vivo activity of EL-SC and EL-SH. A, refolding of rhodanese in the presence of SR1-WT (closed circles), SR1-SC (closed squares), and SR1-SH (closed triangles). Error bars are estimated using data from at least three independent experiments. B, in vivo activity of EL-SC and EL-SH. Growth of E. coli KY1156 cells (39) containing EL-SC and EL-SH after 10-h incubation at 37°C. determined, however, whether the relative positioning of these two segments is also of importance. In our experiments, we showed that the refolding yields of ␣-glucosidase were decreased gradually in proportion to the length of truncated amino acid residues (Fig. 4). ␣-Glucosidase is a protein whose subunit molecular mass (68 kDa) exceeds the documented limits for encapsulation within the GroEL chamber (39,45), and the refolding of ␣-glucosidase is most likely mediated by the trans binding mechanism of GroEL, similar to that demonstrated for aconitase (36). Because ␣-glucosidase is unlikely to be encapsulated within the chamber and placed in proximity to the altered environment of the truncated chaperonins, the decrease in the refolding yield of ␣-glucosidase shown in Fig. 4B cannot be adequately explained by changes in hydropathy of the chamber of GroEL. A complex phenomenon was also observed in results where double ring GroEL NE and HB mutants showed ϳ120 and 60% refolding yields for ␣-glucosidase compared with the wild-type, respectively (data not shown), suggesting that the 526 KNDAAD 531 segment also affects refolding of ␣-glucosidase. These findings suggested that both the length (from the truncation of Gly-Gly-Met repeat) and hydropathy (from the 526 KNDAAD 531 mutation) of the C-terminal region are important for the folding of a large substrate protein that cannot be encapsulated by GroEL and GroES.
A possible role for the C-terminal region may involve ringring communication within the GroEL oligomer. It is well known that inter-ring communication is very important in chaperonin function (9,46). This behavior is explained mechanistically by conformational changes in the equatorial domains of one ring that affect the equatorial domains of the other ring of GroEL. Interestingly, it was reported recently in cryo-electron microscopic studies that the equatorial domain displayed subtle rotational changes upon substrate protein binding, in which concerted movements of both cis and trans equatorial domains upward and downward were observed (47). Furthermore, it was also reported that ATP hydrolysis in the cis-ring affected the trans-ring through inter-ring contacts and caused equatorial domain rotation within the trans-ring (48). These studies highlight the importance of the interactions between cis and trans equatorial domains. Because the equatorial domains play important roles in ring-ring communications, it is conceivable that the truncation of the C-terminal flexible region, including the Gly-Gly-Met repeat and 526 KNDAAD 531 , located in the vicinity of the inter-ring interface, presumably affects inter-ring interactions, thereby resulting in decrease of ATPase activity (Fig. 2C) and refolding yield of ␣-glucosidase (Fig. 4B). Thus, we may conclude that the flexible C-terminal region plays an important role in communicating between the rings in the equatorial domain, as well as in maintaining an optimum environment within the central cavity, and possibly, in the dynamic conformations that allow orderly protein encapsulation and release.