Interactions of Chaperone a -Crystallin with the Molten Globule State of Xylose Reductase IMPLICATIONS FOR RECONSTITUTION OF THE ACTIVE ENZYME*

a -Crystallin is a multimeric protein that has been shown to function as a molecular chaperone. Present investigations were undertaken to understand its mechanism of chaperoning. For this functional in vitro analysis of a -crystallin we used xylose reductase (XR) from Neurospora crassa as the model system. Denaturation studies using the structure-perturbing agent guanidinium chloride indicated that XR folds through a partially folded state that resembles the molten globule. Fluorescence and delay experiments revealed that a -crystallin interacts with the molten globule state of XR (XR-m) and prevents its aggregation. Cold lability of a -crystallin z XR-m interaction was revealed by temperature shift experiments implicating the involvement of hydrophobic interactions in the formation of the complex. Reconstitution of active XR was observed on cool-ing the a -crystallin z XR-m complex to 4 °C or on addition of ATP at 37 °C. ATP hydrolysis is not a prerequisite for XR release since the nonhydrolyzable analogue 5 * -ad-enylyl imidodiphosphate (AMP-PNP) was capable of reconstitution of active XR. Experimental evidence has been provided for temperature- and ATP-mediated structural changes in the a -crystallin z XR-m complex that shed some light on the mechanism of reconstitution of active XR by this chaperone. The relevance of our finding to the role of a -crystallin in vivo is discussed. folding and protein conformation complex of reactions


Urmila Rawat ‡ and Mala Rao §
From the Division of Biochemical Sciences, National Chemical Laboratory, Pune 411008, India ␣-Crystallin is a multimeric protein that has been shown to function as a molecular chaperone. Present investigations were undertaken to understand its mechanism of chaperoning. For this functional in vitro analysis of ␣-crystallin we used xylose reductase (XR) from Neurospora crassa as the model system. Denaturation studies using the structure-perturbing agent guanidinium chloride indicated that XR folds through a partially folded state that resembles the molten globule. Fluorescence and delay experiments revealed that ␣-crystallin interacts with the molten globule state of XR (XR-m) and prevents its aggregation. Cold lability of ␣-crystallin⅐XR-m interaction was revealed by temperature shift experiments implicating the involvement of hydrophobic interactions in the formation of the complex. Reconstitution of active XR was observed on cooling the ␣-crystallin⅐XR-m complex to 4°C or on addition of ATP at 37°C. ATP hydrolysis is not a prerequisite for XR release since the nonhydrolyzable analogue 5-adenylyl imidodiphosphate (AMP-PNP) was capable of reconstitution of active XR. Experimental evidence has been provided for temperature-and ATP-mediated structural changes in the ␣-crystallin⅐XR-m complex that shed some light on the mechanism of reconstitution of active XR by this chaperone. The relevance of our finding to the role of ␣-crystallin in vivo is discussed.
The folding and assembly of a protein into its biologically active conformation is a complex succession of reactions involving the formation of secondary and tertiary structures and domains, the pairing of domains, and the oligomerization of folded monomers (1,2). Elucidation of the various processes that govern protein folding has been the focus of intense research for the past decades. Numerous in vitro protein folding experiments have demonstrated that many proteins successfully achieve their correct native structures in the complete absence of other cellular factors and without input of energy. This led to the expectation that protein folding is determined by the information encoded by the amino acid sequence and proceeds in vivo by the same spontaneous mechanism (1,2). However, in vivo folding and assembly of proteins occur in a highly complex heterogeneous environment, in which high concentrations of proteins in various stages of folding and with potentially interactive surfaces coexist that may change the folding potentials inherent in the sequence. Recently, a number of accessory proteins have been identified that affect the folding and subsequent assembly of proteins. These include the protein isomerases catalyzing cis-trans-isomerization of peptide bonds or disulfide exchange (3,4) and the polypeptide binding proteins termed as "molecular chaperones" (5). The molecular mechanisms by which the protein isomerases accelerate the rate-determining step in the folding of proteins are understood but perhaps not in detail (3,4). Elucidating the mechanistic details underlying the efficient refolding of proteins by chaperones now appears to be an important consideration for defining how proteins fold in vivo. The GroEL and GroES proteins from Escherichia coli are among the most detailed characterized chaperones (5). Horwitz (6) and other workers (7)(8)(9) have shown that ␣-crystallin acts as a molecular chaperone under various denaturing conditions including thermal inactivation, UV irradiation, or reduction of disulfide bonds. Until recently, ␣-crystallin was believed to be a lens-specific protein; however, now it has been reported to be present in many non-lenticular tissues (10,11). The expression of ␣-crystallin has been shown to be induced by thermal (11) or hypertonic stress (12). Numerous studies provide evidence that the ability of ␣-crystallin to suppress aggregation of damaged proteins plays a crucial role in maintaining the transparency of the ocular lens, and the failure of this function could contribute to the development of cataracts (13,14). Important recent development is the finding that ␣-crystallin shows extensive structural similarity with small heat shock proteins that are known to act in vitro as molecular chaperones (15,16). ␣-Crystallin has been shown to be functionally equivalent to the small heat shock proteins namely murine Hsp25 and human Hsp27 in refolding of ␣-glucosidase and citrate synthase in vitro (16). It was, however, unable to refold rhodanase denatured in 6 M GdmCl 1 (17). Recently ␣-crystallin has been reported to bind the temperature-induced molten globule state of proteins (18,19) and prevent photoaggregation of ␥-crystallin by providing hydrophobic surfaces (8). Despite the growing interest in the chaperone action of ␣-crystallin, little is known about its mechanism of chaperoning. For this functional in vitro analysis of ␣-crystallin we used xylose reductase (XR), from Neurospora crassa as the model system. This oxidoreductase plays a crucial role in the fermentation of xylose to ethanol (20) and has recently gained more importance due to its application in the synthesis of xylitol, an acariogenic non-caloric sweetener used in food products (21). Earlier we have undertaken structure-function studies to understand the contribution of essential amino acids in * This work was supported in part by the Department of Science and Technology. 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  the catalytic mechanism of XR, an enzyme that belongs to the important class of oxidoreductases; the conformation and microenvironment of the active site has also been assessed using fluorescent chemoaffinity labeling (22)(23)(24). Various studies on XR will assist in its biotechnological exploitation. Experimental evidence presented in this paper serves to implicate that the chaperone ␣-crystallin stabilizes the molten globule state of XR and thus restrains the non-native conformer from exploring unproductive pathways. Lowering the temperature to 4°C or presence of ATP at 37°C induces a conformational change in ␣-crystallin⅐XR-m complex that is accompanied by a concomitant internalization of hydrophobic surfaces previously exposed. This acts to reduce the hydrophobic interactions that facilitates the dissociation of the complex further allowing reconstitution of the active XR. This paper reports for the first time the mechanism of ␣-crystallin-mediated reconstitution of an active enzyme and the role of ATP in the function of ␣-crystallin as a molecular chaperone.
Xylose Reductase Purification and Assay-Purification of the Neurospora crassa xylose reductase was according to the procedure described earlier (22). The enzyme activity was measured spectrophotometrically at 28°C by monitoring the decrease in NADPH absorbance at 340 nm. The reaction mixture (1 ml) contained 50 mM sodium phosphate buffer, pH 7.2, 0.15 mM NADPH, and an appropriate amount of XR. The reaction was started by addition of D-xylose (final concentration 250 mM). Enzyme units are defined as mol of NADPH oxidized per min. Protein concentration has been determined by the method of Bradford (25) with bovine serum albumin as a standard. The molar concentration of XR was calculated assuming a M r of 60,000 (22).
Denaturation/Renaturation Studies of XR-All denaturation and renaturation experiments were carried out in 50 mM sodium phosphate buffer, pH 7.2. XR was incubated with varying concentrations of GdmCl for 1.5 h at 28°C, and the accompanying structural changes were investigated using fluorescence and circular dichroism measurements. Renaturation was initiated by rapidly diluting the sample (XR at varying states of denaturation) in the same buffer with or without ␣-crys-tallin, and the XR activity at various times of refolding was measured. When the effect of adenine nucleotides ATP, AMP-PNP, ADP, and AMP on the ␣-crystallin-assisted renaturation of XR was to be investigated these were added 1 h after initiation of renaturation process. Renaturation of XR was also carried out by substituting ␣-crystallin with bovine serum albumin (0.6 mg/ml). Other experimental details have been described in the legends to figures.
Circular Dichroism and Fluorescence Studies-Circular dichroism (CD) spectra were recorded on a JASCO J600 model spectropolarimeter. Changes in the secondary structure of XR induced by the denaturant GdmCl were monitored in the far UV region (200 -250 nm) using a 1-mm path length cell. The tertiary structure was monitored in the near UV region (250 -320 nm) using a 10-mm path length cell. The enzyme concentrations in these experiments was 0.5 mg/ml. Mean residue ellipticities [] (expressed as degree cm 2 dmol Ϫ1 ) were determined according to Ref. 26.
Fluorescence spectra were recorded with a Perkin-Elmer LS 50B spectrofluorimeter equipped with a Julabo F20 water bath or Aminco SPF-500 spectrofluorimeter. The excitation and emission wavelengths have been mentioned in the legends to figures. Corrections due to inner filter effect were made according to the formula F c ϭ F antilog ((A ex ϩ A em )/2), where F c and F are the corrected and uncorrected fluorescence intensities, and A ex and A em are the sample absorbance at the excitation and emission wavelengths, respectively (27).
Additional experimental details have been described in the text and figure legends.

RESULTS AND DISCUSSION
Folding Intermediates of XR-The CD spectrum of XR in the far UV region (200 -260 nm) exhibited a strong negative ellipticity in the region 215-222 nm and a weaker one at 208 nm, characteristic of a protein having an ␣-helix (data not shown). XR was incubated with increasing concentrations of the denaturant GdmCl, and the changes in the negative CD band in the far UV region were monitored. The mean residue ellipticities obtained at 220 nm [] 220 were normalized with respect to that in the absence of GdmCl and plotted against the respective GdmCl concentration (Fig. 1). A decrease in the negative ellipticity was observed with the addition of GdmCl, and at 1.4 M GdmCl the [] 220 decreased by almost 43% of that in the absence of GdmCl. Further increase in the denaturant resulted in a loss of negative ellipticity until there was a total loss of structure of the CD band in 6 M GdmCl indicating a considerable loss of secondary structure (Fig. 1). 6 M GdmCl converted XR into unfolded polypeptides, and this state has been referred to as XR-u. The CD spectrum of XR was also monitored in the near UV region (250 -320 nm) to study GdmCl-induced changes in the environment of tryptophan and tyrosine side chains. In the absence of the denaturant, the spectrum exhibited a broad negative band with a double minimum at 278 and 285 nm. However, the presence of 1.4 M GdmCl resulted in a decrease in the negative ellipticity in this region similar to that of the unfolded XR in 6 M GdmCl indicating that the aromatic residues in this state were no longer in an asymmetric environment (data not shown).
We used the fluorophore ANS to determine the relative amount of exposed hydrophobic surfaces in the folding intermediates of XR. ANS is not fluorescent in aqueous solutions ( em 525 nm); however, on addition of proteins containing hydrophobic pockets its emission maximum shifts to shorter wavelengths, and the emission intensity is enhanced. As shown in Fig. 2 the binding of ANS to XR was measured as a function of GdmCl. A maximum increase in the ANS fluorescence ( em 475 nm) was observed at 1.4 M GdmCl indicating maximum exposure of hydrophobic surfaces in this state of XR. At higher concentrations of the denaturant, a decrease in the intensity of the dye fluorescence was observed which was accompanied by a shift in the em toward red indicating unfolding of XR (Fig. 2). ANS has been widely used to detect the formation of molten globule-like intermediates in the folding pathways of several proteins (28). This state is characterized to be as compact as the native protein with solvent-accessible hydrophobic regions and appreciable amount of secondary structure but no rigid tertiary structure (29,30). It was thus evident from the CD studies that at 1.4 M GdmCl XR retains substantial amount of secondary structure ( Fig. 1) but very little tertiary structure (data not shown). Altogether our CD and ANS binding studies revealed that at 1.4 M concentration of GdmCl XR partially unfolded to its molten globule state which has been referred to as XR-m.
Chaperone-assisted Renaturation of XR-Attempts to refold XR from the XR-u state in the absence and presence of ␣-crystallin were unsuccessful (Fig. 3). Similar results were obtained for progressively less denatured states (XR denatured with 2-4 M GdmCl). Further investigations were carried out to study the influence of ␣-crystallin on the renaturation of XR-m. The refolding of XR-m was initiated at 28°C in the absence/presence of ␣-crystallin; after 30 min the samples were shifted to varying temperatures (Fig. 3), and the XR activity recovered at different time intervals was measured. As shown in Fig. 3, XR-m lacked the ability to spontaneously reconstitute active XR. However, in the presence of ␣-crystallin the renaturation process at 4°C followed a sigmoidal time course. As can be observed from inset of Fig. 3, there was no measurable XR activity for the first 15 min (lag phase). Thus similar to renaturation of oligomeric proteins (2), inactive XR monomers may be produced in an early folding step which then undergo additional folding and/or association prior to the assembly of XR into active oligomers. The rate of reactivation beyond the lag phase was slow, and a maximum 55% of the XR activity was recovered in 6 h. A value of 112 min was observed for t1 ⁄2 , where the activity recovered was half of the maximal extent. At 28°C, the renaturation process yielded a maximum 14% of the XR activity, whereas the values for the lag phase and t1 ⁄2 were similar to that observed at 4°C. ␣-Crystallin, however, failed to reconstitute active XR at 37°C (Fig. 3). These temperatureshift experiments thus revealed that the complex of ␣-crystal- respectively. The renaturation process was initiated at the same temperature by diluting 10 l of the sample into a final volume of 1 ml of 50 mM sodium phosphate buffer, pH 7.2, with or without ␣-crystallin (final concentration 0.6 mg/ml). After 30 min the samples were kept at varying temperatures, and 100-l aliquots of the refolding solution were withdrawn at various times of refolding and assayed for XR activity as described under "Experimental Procedures." E represents refolding of XR-m in the absence of ␣-crystallin and in its presence when the refolding solution was shifted from 28°C to the following temperatures: 4°C (q), 28°C (f), and 37°C (Ⅺ). ‚ and OE represents the refolding of XR-u and progressively less denatured states of XR (XR denatured with 2-4 M GdmCl) in the absence and presence of ␣-crystallin under the experimental conditions described above. The percentage activity recovered was determined with reference to native XR. The inset shows the early time course of ␣-crystallin-assisted renaturation demonstrating the lag phase (first 15 min).

FIG. 4. Reactivation of XR-m at varying concentrations of
␣-crystallin. XR-m was renatured (as described in the legend to Fig. 3) in the presence of varying concentrations of ␣-crystallin. After 30 min the samples were kept at 4°C, and aliquots withdrawn after 6 h were assayed for XR activity. The percentage activity recovered is with respect to native XR.
␣-Crystallin-Mediated Reconstitution of Active Enzyme lin and the bound XR is stable at 37°C but is cold-labile since lowering the temperature of the renaturation process from 28 to 4°C resulted in the reconstitution of active XR. Unlike the observations at 28°C, ␣-crystallin failed to reconstitute active XR when the refolding process was initiated at 4°C.
The ␣-crystallin-mediated renaturation of XR was examined as a function of the chaperone concentration. As shown in Fig.  4, 5% of the original XR activity was recovered at the lowest concentration of ␣-crystallin (0.05 mg/ml). The extent of renaturation increased in a concentration-dependent manner, and a maximum 55-57% of the original activity was recovered at ␣-crystallin concentration of 0.6 -0.8 mg/ml. The concomitant increase in the extent of renaturation with an increase in ␣-crystallin can be attributed to simple mass action effects, wherein an increase in the ␣-crystallin concentration would increase the collisional frequency so as to favor the formation of ␣-crystallin⅐XR-m complex as opposed to forming non-native XR. To test the specificity of ␣-crystallin, the renaturation of XR-m was also investigated in the presence of bovine serum albumin alone (0.6 mg/ml), under the conditions described for renaturation with ␣-crystallin. It was observed that unlike ␣-crystallin bovine serum albumin failed to mediate the reconstitution of active XR.
Dependence of Renaturation Process on the Time Interval between Initiation of Refolding and Addition of ␣-Crystallin-␣-Crystallin mediated reconstitution of active XR only from its XR-m state (Fig. 3), indicating that the chaperone probably traps the oxidoreductase in a conformation resembling the molten globule. Evidence for this observation was provided by delay experiments wherein refolding of XR-m was initiated in a solution lacking ␣-crystallin which was then added at the indicated times. As shown in Fig. 5, a concomitant decrease in the ability of ␣-crystallin to reconstitute active XR from the XR-m state was observed with an increase in the time between the dilution of XR-m and the addition of ␣-crystallin. The increase in XR-m concentration also resulted in a progressive decrease in the yield of reconstituted XR (Fig. 5) indicating that the loss of recoverable XR was due to aggregation and not due to some irreversible isomerization. These results reveal that when XR-m is diluted into a solution containing ␣-crystallin two competitive processes occur, namely aggregation or formation of ␣-crystallin⅐XR-m binary complex. Since aggregation is a second-order process it can be much faster than first-order folding (31) and hence predominant with increasing concentrations of XR-m. The delay experiments also revealed the inability of ␣-crystallin to redissolve XR aggregates formed in its absence.
␣-Crystallin Forms a Complex with Folding Intermediate-Fluorescence studies were performed to confirm that the XR bound to ␣-crystallin exists in the molten globule state. The tryptophanyl fluorescence of native, denatured, and ␣-crystallin-bound XR is shown in Fig. 6. Native XR exhibited an emission maximum at 326 nm, whereas in 6 M GdmCl the emission maximum was shifted to 350 nm which corresponds to the fluorescence maximum of tryptophan in aqueous solution. The XR bound to ␣-crystallin exhibited an emission maximum at 334 nm indicating that the tryptophans in the bound form of XR are more exposed to the solvent than the native enzyme. The increase in fluorescence intensity of the ␣-crystallin-bound XR may be attributed to the denaturant-induced changes in the microenvironment of the tryptophans of XR or may be due to interactions of the partially unfolded protein with ␣-crystallin. Altogether these results revealed that the conformation of XR bound to ␣-crystallin is neither native-like nor completely unfolded but a partially folded intermediate resembling the molten globule. Thus by sequestering the molten globule state of XR in the form of a stable binary complex, ␣-crystallin is able to ␣-Crystallin-Mediated Reconstitution of Active Enzyme suppress their interaction that would otherwise lead to aggregation.
Temperature Dependence of the Exposure of Hydrophobic Surfaces of ␣-Crystallin⅐XR-m Complex-The temperature dependence of the hydrophobic interactions in protein folding has been studied earlier by Baldwin (32). Maximum stabilization of these interactions is observed at high temperature where the enthalpy is the dominating factor in determining the stability, and as the temperature is decreased the interactions are weakened. Our temperature-shift experiments (Fig. 3) revealed an increase in the ␣-crystallin-mediated reconstitution of XR with the decrease in the temperature of the refolding solution implying that the hydrophobic interactions play a crucial role in the formation of ␣-cystallin⅐XR-m complex.
Attempts were made to correlate temperature-mediated alterations in the hydrophobic surfaces of the ␣-crystallin⅐XR-m complex to reconstitution of active XR, using ANS as a probe for apolar binding sites whose fluorescence is dependent on the hydrophobicity of the environment. As shown in Fig. 7, presence of ␣-crystallin⅐XR-m complex incubated at 37°C resulted in a blue shift in the ANS fluorescence from 525 to 475 nm accompanied by an increase in fluorescence intensity; however, in the presence of the complex incubated at 4°C a 35% decrease in the dye fluorescence was observed compared with that at 37°C (Fig. 7). These results indicate that at 37°C the complex exists in a state with hydrophobic binding sites that are accessible to ANS; however, a decrease in the incubation temperature to 4°C probably mediates a conformational change in the complex that is accompanied by internalization of the hydrophobic surfaces previously exposed. This further acts to weaken the hydrophobic interactions holding the ␣-crystallin⅐XR-m complex and thus reduces the affinity of ␣-crystallin for the substrate protein further, allowing reconstitution of active XR (Fig. 3). This observation also explains the cold lability of ␣-crystallin⅐XR-m complex and the inability of ␣-crystallin to reconstitute active XR when refolding was initiated at 4°C.
Fluorescent Chemoaffinity Labeling of XR Renatured in the Presence of ␣-Crystallin-Fluorescent chemoaffinity labeling studies were performed using o-phthalaldehyde as the chemi-cal initiator to shed some light on the conformation of XR renatured in the presence of ␣-crystallin. Chemoaffinity labeling is a powerful technique and combines some of the advantages associated with the photoactivated and electrophilic affinity labeling. o-Phthalaldehyde is a bifunctional agent that cross-links SH and NH 2 groups situated in close proximity to form an isoindole derivative that exhibits strong fluorescence (33). XR reacts with o-phthalaldehyde resulting in the formation of fluorescent XR-isoindole derivative at the active site ( ex 338 nm; em 410 nm) (23,24). As shown in Fig. 8, incubation of the ␣-crystallin renatured XR with o-phthalaldehyde resulted in the formation of XR-isoindole derivative as observed with the native enzyme. However, XR when renatured in the absence of ␣-crystallin failed to form the derivative. These results suggest that ␣-crystallin mediates refolding of XR to a conformation similar to that of the native enzyme.
Effect of Adenine Nucleotides and XR Substrates on ␣-Crystallin mediated Renaturation-The inability of ␣-crystallin to reconstitute active XR at 37°C (Fig. 3) cannot be attributed to lack of binding of the chaperone to XR as shown in Fig. 6 but may be due to either an inability to release the bound XR at high temperature or due to the polypeptide being released in a temperature-sensitive conformation. Earlier the role of ATP in the renaturation of the proteins rhodanese, ribulose-bisphosphate carboxylase/oxygenase, and ␤-protein (34 -36) by the chaperone GroEL has been reported. Recently evidence for the binding of ATP to ␣-crystallin was provided by 31 P NMR spectroscopy (37) and fluorescence studies (38). Hence, studies were undertaken to find out if ATP played any role in the chaperone function of ␣-crystallin. For this functional in vitro analysis refolding of XR-m was initiated in a buffer containing ␣-crystallin at 37°C and further incubated in the absence/presence of ATP. As shown in Fig. 9, ␣-crystallin mediated reconstitution of active XR was not observed in the absence of ATP. However, The sample was repetitively filtered through Centricon-100 microconcentrators (Amicon) with 100-kDa cut-off to separate the released XR (filtrate) from ␣-crystallin (retentate). Furthermore, 10 l of 3 mM o-phthalaldehyde was added to 2 ml of the filtrate, and the spectra were recorded after 30 min, with the excitation wavelength fixed at 338 nm. Similar experiments were repeated for renaturation of XR-m in the absence of ␣-crystallin. O and ---represent the isoindole spectra of XR renatured from XR-m state in the presence and absence of ␣-crystallin, respectively.

␣-Crystallin-Mediated Reconstitution of Active Enzyme
with ATP the renaturation process followed a sigmoidal time course, and values of 10 and 53 min were observed for the lag time ( Fig. 9 inset) and t1 ⁄2 (where the activity regain was half of the maximal extent), respectively. A maximum 22% of the XR activity was recovered in 1.5 h, and a further decrease in the yield may be attributed to temperature-mediated inactivation of the released XR.
The effect of XR substrates NADPH and xylose on the ␣-crystallin-mediated renaturation of XR-m was investigated. It was observed that the percentage of XR activity recovered in the presence of ATP and NADPH in 1.5 h was approximately 2.5-fold higher than that observed in the presence of ATP alone (Fig. 9). However, the lag time (10 min) (Fig. 9 inset) and t1 ⁄2 (53 min) values observed in both cases were identical indicating that the presence of NADPH did not alter the rate of the initial slow reaction (refolding of monomer and/or correct formation of dimer) nor the overall rate. The phosphorylated coenzyme, however, failed to mediate the reconstitution of active XR in the absence of ATP (Fig. 9). This ruled out the possibility of interaction of NADPH with XR bound to ␣-crystallin and altered the equilibrium between the bound and free enzyme. However, this may also be unlikely because the bound XR is in a non-native state (Fig. 6) and lacks enzymatic activity. In light of these data we propose that the release of ␣-crystallin-bound XR is mediated by ATP, and the phosphorylated coenzyme NADPH traps the free XR in a conformation stable at the physiological temperature. The XR substrate xylose failed to exert its effect on ␣-crystallin-mediated reconstitution of active XR in the presence of ATP (Fig. 9). This may be due to the fact that the enzyme follows an iso-ordered bi bi mechanism (22) wherein xylose binds to XR⅐NADPH binary complex.
The effects of adenine nucleotides upon the ␣-crystallin-mediated reconstitution of active XR were investigated. As shown in Fig. 10, addition of AMP-PNP, an ATP analog with a nonhydrolyzable ␤-␥ bond, resulted in a maximum 18% of XR activity in 1.5 h from its partially folded state compared with the maximum activity recovered in the presence of ATP. Further increases in the incubation period resulted in a decline in the yield of reconstituted XR which can be attributed to the thermal inactivation of the free XR. Hence, the experiments were repeated wherein the coenzyme NADPH was added 30 min after the addition of the adenine nucleotides ATP/AMP-PNP so as to stabilize the released XR. Under these conditions a maximum 53% of XR activity was recovered in the presence of AMP-PNP in 12 h compared to the maximum observed in the presence of ATP in 5 h and which did not increase with further incubation. These results indicated differential ability of the adenine nucleotides ATP and AMP-PNP to reconstitute active XR which may be attributed to their different binding constants. ␣-Crystallin-mediated reconstitution of active XR was not observed in the presence of the adenine nucleotides AMP or ADP (Fig. 10), implying the involvement of the P␥ of ATP upon binding to ␣-crystallin⅐XR-m complex. Altogether, these results supported the notion that ATP hydrolysis is not a prerequisite for release of XR bound to ␣-crystallin since the nonhydrolyzable analogue AMP-PNP was capable of reconstitution of the active XR. Instead, the release of XR may be mediated in part through the binding of ATP or AMP-PNP producing a similar conformational change in the chaperone which weakens its interactions with XR, further allowing reconstitution of the active enzyme. In contrast to 37°C (Fig. 9) the ␣-crystallinmediated renaturation of XR-m was observed in the absence of ATP at 4°C (Fig. 3). Also, when the renaturation was initiated at 37°C and later the temperature shifted to 4°C a maximum FIG. 9. Influence of ATP and XR substrates on the ␣-crystallinmediated reactivation of XR. Renaturation of XR-m (25 M) was initiated at 37°C by diluting 10 l of the sample into a final volume of 1 ml of phosphate buffer, pH 7.2, with ␣-crystallin 0.6 mg/ml preincubated for 2 h at 37°C. After 1 h the following additions were made, no nucleotide (E), ATP (1 mM) (q), NADPH (0.5 mM) (‚), ATP (1 mM) ϩ NADPH (0.5 mM) (OE), and xylose (250 mM) (f); furthermore, at the times indicated, 100-l aliquots of the refolding solution were withdrawn and added to the assay mixture to determine the XR activity as described under "Experimental Procedures." Ⅺ represents the renaturation of XR-u (XR at a concentration of 25 M denatured with 6 M GdmCl) in the presence of the nucleotides/XR substrates as described above. The percentage activity recovered is with respect to the control containing native XR. The inset shows the early time course of ␣-crystallin-assisted renaturation demonstrating the lag phase of 10 min for the renaturation process in the presence of ATP (q) or ATP ϩ NADPH (OE).

␣-Crystallin-Mediated Reconstitution of Active Enzyme
55% of the original XR activity was recovered in 24 h. These differences in the ATP requirement support the notion that the role of adenine nucleotide would seem to be linked with the need to provide a rapid dissociation pathway for the ␣-crystallin⅐XR-m complex at 37°C and is not essential for correct folding of XR. In light of these data it is tempting to speculate that a weak interaction exists between ␣-crystallin and the non-native XR. Tightly bound substrate proteins would probably reduce the flexibility of ␣-crystallin; hence, the binding of ATP alone or its hydrolysis would not be sufficient to induce conformational change necessary for the dissociation of the ␣-crystallin-protein complex.
Dependence of ␣-Crystallin-assisted Renaturation on the Order of Addition of Partially Folded XR and ATP to ␣-Crystallin-The highly selective nature of protein-ligand interaction provides a sensitive mechanism for the modulation of protein activity. Experiments were carried out to define the role of ␣-crystallin-ATP interaction on the structure and mechanism of action of the chaperone ␣-crystallin. As shown in Fig. 11 the order of addition of ATP and XR-m to ␣-crystallin resulted in a difference in the ␣-crystallin-mediated renaturation profiles. The renaturation process of XR-m carried out in the presence of preformed ␣-crystallin⅐ATP complex followed a sigmoidal time course, and values of 15 and 40 min were obtained for the lag phase (Fig. 11 inset) and t1 ⁄2 , respectively. However, when ␣-crystallin⅐XR-m complex was allowed to form prior to addition of ATP the renaturation process had a t1 ⁄2 of 53 min and an initial lag time of 10 min (Fig. 11, inset). Although the value of t1 ⁄2 was increased by 13 min in the latter case, the final extent of XR activity recovered was 5.5-fold higher than that observed with the preformed ␣-crystallin⅐ATP complex. This indicated that the ATP-free form of ␣-crystallin mediates reconstitution of active XR more effectively than the ATP-bound form.
The structural differences in ATP-free and -bound forms of ␣-crystallin⅐XR-m were probed by fluorescence spectroscopy using the hydrophobic probe ANS. As shown in Fig. 12, ANS exhibited an emission maximum at 525 nm; however, in the presence of ␣-crystallin⅐XR-m its fluorescence intensity increased and the emission maximum shifted to 475 nm characteristic of the transfer of ANS into a hydrophobic environment. Furthermore, a concomitant decrease in the intensity of the dye fluorescence ( em 475 nm) was observed in the presence of increasing concentrations of ATP (Fig. 12). These results imply that binding of the adenine nucleotide to ␣-crystallin⅐XR-m complex induces a conformational change that is accompanied by a concomitant internalization of hydrophobic surfaces previously exposed. This acts to reduce the hydrophobic interactions and thus the affinity of the chaperone for the substrate protein further allowing reconstitution of the active XR. Evidence for the ATP-␣-crystallin binding has earlier been provided by 31 P NMR spectroscopy (37) and fluorescence studies (38). ATP failed to influence the tryptophanyl fluorescence of XR in 1.4 M GdmCl (XR-m) indicating inability of the adenine nucleotide to bind XR-m (data not shown). Altogether these results imply that in the presence of ␣-crystallin⅐XR-m complex, ATP binds ␣-crystallin and not the bound XR which is quite likely because as shown in Fig. 6 the ␣-crystallin-bound XR exists in a non-native state. In light of these data we propose that ␣-crystallin operates by providing hydrophobic surfaces that interact with the molten globule state of XR, and the hydrophobic interactions play an important role in the formation of ␣-crystallin⅐XR-m complex.
Conformational changes have been proposed to play a major role in the binding of folding intermediates and in the discharge of polypeptides from molecular chaperones. One of the signals for inducing such structural changes is the hydrolysis of ATP as reported in case of the chaperone DnaK (39) and GroEL (34,35). However, reports are also available wherein the chaperones GroEL (36) and BiP (40) do not require ATP hydrolysis. Instead, the mere binding of the adenine nucleotide to the chaperone induces a typological change in the chaperone that weakens its interaction with the bound protein. This acts to release the protein, further allowing it to assume its native ␣-Crystallin-Mediated Reconstitution of Active Enzyme state. Our investigations reveal that the mechanism of chaperoning of ␣-crystallin also requires the binding of ATP to the chaperone and not its hydrolysis.
Earlier studies have indicated that chaperones functioned post-translationally before the formation of the folded functional enzyme. The present investigation was carried out to gain some insight into the conformation of XR interacting with the chaperone ␣-crystallin and the mechanistic details underlying the reconstitution of active enzyme. The conditions for the unfolding of native XR were sought in the belief that the unfolded enzyme or its folding intermediates would serve as a substrate for the ␣-crystallin-mediated reconstitution of active XR. Our denaturation studies using the structure-perturbing agent GdmCl revealed that the folding of XR involves an intermediate that resembles the molten globule. The existence of molten globule like intermediates has been demonstrated with several proteins, and these intermediates are known to be involved in various cellular functions such as membrane translocation of proteins (41,42), chaperone-assisted protein folding (5), and also in various genetic diseases (43,44). Interest in such intermediates is strong since they have been proposed to be an obligatory intermediate formed early in the folding pathway (45). A common feature of the molten globule state is the exposure of hydrophobic surfaces that lead to aggregation of proteins during folding. Our in vitro studies using XR revealed that the chaperone ␣-crystallin operates by interacting with the hydrophobic regions that appear on the surface of molten globule state of XR. This reduces the concentration of the free partially folded XR (XR-m) during renaturation and thus prevents loss of enzyme activity due to their hydrophobic aggregation. Lowering the temperature to 4°C or the presence of ATP at 37°C induces a conformational change in the ␣-crystallin⅐XR-m complex that is accompanied by a concomitant internalization of previously exposed hydrophobic surfaces. This acts to reduce the hydrophobic interactions involved in the formation of the complex and thus the affinity of the chaperone for the substrate protein further allowing reconstitution of the active XR. The results presented here are consistent with the notion that the complete folding of XR resulting in the formation of catalytically active dimer does not occur while it is bound to the surface of ␣-crystallin. Our investigation reveals for the first time the mechanism of ␣-crystallin-mediated reconstitution of an active enzyme, and the role of temperature and ATP in its mechanism of chaperoning. Earlier it has been reported that ␣-crystallin does not prevent the photoaggregation of ␥-crystallin at low temperatures. However, it can do so at temperatures above 30°C (8). Our present investigation also supports this view, since ␣-crystallin-mediated reconstitution of XR was observed when the refolding process was initiated at 28 and 37°C and not when initiated at 4°C which is attributed to the inability of the chaperone to prevent aggregation of XR-m at low temperature.
Delay experiments revealed the inability of ␣-crystallin to dissolve XR aggregates formed in its absence implying that the chaperone ␣-crystallin should be present during stress conditions. The dependence of protein aggregation reactions on temperature and concentration is known. Our results support the notion that one of the functions of ␣-crystallin in vivo may be to protect non-native protein from intracellular aggregation during high rate of protein synthesis and/or thermal stress. The inability of XR-u and XR-m to spontaneously reconstitute active XR under the conditions used in the present investigation is due to the fact that aggregation competes with the correct folding pathway. The kinetic competition between refolding and aggregation has been reported to be a major determinant for lower yields or irreversibility in refolding of proteins in vitro (46). However, refolding of XR may be possible under different experimental conditions, but regardless of this observation we are left with the fact that presence of ␣-crystallin resulted in a substantial amount of reconstitution of active XR from the XR-m state.
It has been reported that in E. coli a cascade of molecular chaperones mediate folding of proteins. The chaperone DnaK interacts with polypeptides in their extended conformation and prevents premature misfolding and aggregation after which GroEL stabilizes folding intermediates resembling the molten globule and mediates proper folding. The transfer of DnaK/ DnaJ-bound protein to GroEL requires GrpE as the coupling factor (47). Such a mechanism is likely to exist in eukaryotes also (47)(48)(49). Our investigation reveals that ␣-crystallin is able to reconstitute XR via interaction with its non-native conformer characterized by an increased surface hydrophobicity but a remarkably low degree of unfolding. The inability of ␣-crystallin to reconstitute XR from its extended conformation implies that in vivo other chaperones may be involved in binding to the unfolded polypeptides and prevent premature misfolding and aggregation, whereas the proper folding and assembly may depend on the subsequent transfer of the partially folded polypeptide to ␣-crystallin. Further evidence is provided by the observation that ␣-crystallin prevents aggregation of lens proteins induced by oxidative stress and UV radiation. These conditions are not likely to unfold protein molecules completely but induce formation of partially folded state with hydrophobic surfaces that result in its aggregation (19). For many years ␣-crystallin was thought to be a lens-specific structural protein where it played a role to facilitate proper transmission of light. However, recently ␣-crystallin has been demonstrated to be present in various non-lenticular tissues such as brain, spleen, and heart and also found in NIH 3T3 cells expressing Ha-ras and v-mos (10,11). ␣-Crystallin has been reported to be induced by thermal or hypertonic stress (11,12), and its expression is markedly increased in a number of neurological diseases such as Creutzfeld-Jacob disease, Alexander disease, and Lewy body disease (50 -52). Our present investigation on ␣-crystallin adds to the information available on its chaperone function which may assist to shed some light on its diverse roles in vivo.