Conformational Properties of Substrate Proteins Bound to a Molecular Chaperone -Crystallin*

-Crystallin, the major protein of the ocular lens, acts as a molecular chaperone by suppressing the nonspe- cific aggregation of damaged proteins. To investigate the mechanism of the interaction between (cid:97) -crystallin and substrate proteins, we prepared a tryptophan-free mutant of human (cid:97) A-crystallin and assessed the conformation of thermally destabilized proteins captured by this chaperone using fluorescence spectroscopy. The fluorescence emission characteristics of bound sub- strates (rhodanese and (cid:103) -crystallin) and the results of fluorescence quenching experiments indicate that the proteins captured by (cid:97) -crystallin are characterized by a very low degree of unfolding. In particular, the struc- ture of rhodanese bound to (cid:97) A-crystallin appears to be considerably more native-like compared to that of the enzyme bound to the chaperonin GroEL. We postulate that (cid:97) -crystallin (and likely other small heat shock proteins) recognize preferentially the aggregation-prone conformers that occur very early on the denaturation pathway. With its ability to capture and stabilize these early non-native structures, (cid:97) -crystallin appears to be uniquely well suited to chaperone the transparency properties of the ocular lens. highly homologous 20-kDa subunits, A B. The A and B chains noncovalently self-associate to form a large macromolecular complex of approximately 40 sub-units purified gel permeation chromatogra- phy on a Sephacryl S-300 column Biotech Inc.). in fractions eluting a position bovine (cid:97) -crystallin. Purified recombinant (cid:97) A-crystallins on SDS-gel electrophoresis single band corre- sponding by exclusion chromatography on a Sephacryl S-300 The molar ratio of (cid:97) -crystallin to the substrate protein in a given complex was estimated by subtracting the amount of the unbound substrate (as determined spectrophotometrically) from the initial amount of this protein in the incubation mixture. Fluorescence Measurements and Quenching Experi- ments— Tryptophan spectra were measured on an SLM using the excitation wavelength of 295 nm. Quenching experiments were performed by titrating the solution of free- and W9F (cid:97) A-crystallin-bound proteins with freshly prepared 5 M solutions of acrylamide or potassium iodide. Fluorescence intensities were measured at the wavelength corresponding to the emission max- imum of each protein and were corrected for dilution, blanks, and the inner filter effect. The effective quenching constants ( K SV ) eff (equal to f a K SV ) were calculated from the inverse slopes of the F 0 / (cid:68) F versus 1/[Q] plots according to the modified Stern-Volmer equation: F 0 / (cid:68) F (cid:53) 1/ a SV [Q], where F 0 and F are the fluorescence intensities in the absence and presence of the quencher, [Q] is the molar concentration of the quencher, and (cid:68) F (cid:53) F 0 (cid:50) F (24, 25). The fraction of quenchable fluorescence, f a , was obtained from the ordinate intercept of the linear portion of the F 0 / (cid:68) F versus 1/[Q] plot (24).

␣-Crystallin, the major protein of the ocular lens, acts as a molecular chaperone by suppressing the nonspecific aggregation of damaged proteins. To investigate the mechanism of the interaction between ␣-crystallin and substrate proteins, we prepared a tryptophan-free mutant of human ␣A-crystallin and assessed the conformation of thermally destabilized proteins captured by this chaperone using fluorescence spectroscopy. The fluorescence emission characteristics of bound substrates (rhodanese and ␥-crystallin) and the results of fluorescence quenching experiments indicate that the proteins captured by ␣-crystallin are characterized by a very low degree of unfolding. In particular, the structure of rhodanese bound to ␣A-crystallin appears to be considerably more native-like compared to that of the enzyme bound to the chaperonin GroEL. We postulate that ␣-crystallin (and likely other small heat shock proteins) recognize preferentially the aggregation-prone conformers that occur very early on the denaturation pathway. With its ability to capture and stabilize these early non-native structures, ␣-crystallin appears to be uniquely well suited to chaperone the transparency properties of the ocular lens.
␣-Crystallin, the major protein of the vertebrate eye lens, consists of two types of highly homologous 20-kDa subunits, ␣A and ␣B. The A and B chains noncovalently self-associate to form a large macromolecular complex of approximately 40 subunits (1,2). Spectroscopic data provided strong evidence that the secondary structure of ␣-crystallin is dominated by ␤-sheets (3,4). Much less specific information is available regarding the tertiary and quaternary structure of the protein. While a number of structural models have been proposed for ␣-crystallin oligomers (2), none of them is generally accepted.
Believed for many years to be strictly lens-specific proteins, ␣Band ␣A-crystallin have recently been found in many nonlenticular tissues (5)(6)(7)(8). Furthermore, ␣B-crystallin has been associated with a number of neurodegenerative disorders (5,6,8). Another important recent development is the rapidly growing evidence that ␣-crystallin belongs to a family of small heat shock proteins (sHSPs). 1 This is indicated by extensive structural similarities between ␣-crystallin and other sHSPs (8,9) as well as by the recent findings that ␣B-crystallin is inducible by various stress conditions (10,11) and that ␣Band ␣Acrystallin are able to confer cellular thermoresistance (12,13). Despite the abundance of sHSPs in both eukaryotic and prokaryotic organisms (14), no obvious function has been associated with these ubiquitous proteins. A new light on the potential physiological role of ␣-crystallin and related sHSPs has been shed by recent findings that these proteins act in vitro as molecular chaperones by preventing the aggregation of other proteins under conditions of thermal stress or other insults (1,(15)(16)(17)(18)(19)(20). The chaperone function of ␣-crystallin and other sHSPs is likely to be of considerable importance in vivo. In the context of eye research, it has been postulated that the ability of ␣-crystallin to suppress the aggregation of damaged proteins may be critical for maintaining the transparency of the ocular lens and that aging-related deterioration of the chaperone function could contribute to the development of cataracts (1,2,15).
Despite the rapidly growing interest in the chaperone function of ␣-crystallin and related sHSPs, very little is known about the mechanism by which these proteins interact with their substrates. Recent results indicate that, in contrast to other known chaperones, ␣-crystallin has very low affinity for folding intermediates formed during protein refolding reactions in vitro and that its substrate specificity is limited to non-native structures that occur on the denaturation pathway only (21). However, the critically important information about the specific conformational state(s) of the non-native structures that are recognized by ␣-crystallin is still missing. In this work, we have cloned and overexpressed ␣A-crystallin mutant in which the sole Trp residue has been replaced with Phe. Fluorescence spectroscopy experiments with the complexes formed between the Trp-free ␣A-crystallin and various substrate proteins allowed us, for the first time, to gain a direct insight into the conformation of proteins bound to this chaperone.

Reagents and Proteins-bis-ANS was purchased from Molecular
Probes. Acrylamide, bovine pancreas insulin, and bovine liver rhodanese were obtained from Sigma. The low molecular mass ␤-(␤ L ) and ␥-crystallin fractions were isolated from young bovine lenses and purified as described previously (21).
Cloning and Overexpression of Human ␣A-Crystallins-Complementary DNA clones encoding human ␣A-crystallin were constructed using RNA PCR (Perkin Elmer). First strand cDNA synthesis was carried out on human lens total RNA by priming with a downstream primer/ adaptor (GGCTGCTATCTAA) designed to anneal in the 3Ј-untranslated region of the ␣A-crystallin mRNA. For amplification of target sequences by PCR, an upstream primer (ATGGACGTGACCATCCAG) was designed to anneal at the translational initiation codon. Sequences of primer binding sites were deduced from the human crystallin gene sequence deposited by Jaworski and Piatigorsky, GenBank TM accession number X14789. Both PCR primer/adaptors contained sequences that permitted ligation-free cloning of the PCR product into the pDIRECT cloning vector (PCR-Direct, Clontech). Site-directed mutagenesis was carried out using a PCR-based system (Life Technologies, Inc.). Nucle-otide sequence analysis of both strands of the wild type and mutant clones confirmed their structures. For overexpression studies, wild type and W9F mutant coding sequences were transferred into the expression plasmid pMON20,400 and cultured in shake flasks as described previously (22). Crystallins were extracted from host cells essentially as described by Merck et al. (23) and were purified by chromatography on a TMAE-fractogel ion exchange column (EM Separations) using a linear gradient (0 -0.5 M) of NaCl. Fractions corresponding to a single predominant peak were further purified by gel permeation chromatography on a Sephacryl S-300 column (Pharmacia Biotech Inc.). Virtually all protein from this column was found in fractions eluting at a position similar to that observed for bovine ␣-crystallin. Purified recombinant ␣A-crystallins showed on SDS-gel electrophoresis a single band corresponding to a molecular mass of approximately 20 kDa.
Preparation of the Complexes between ␣A-Crystallin and Substrate Proteins-One ml of a solution containing 0.25 mg of ␣A-crystallin and a known amount of the substrate protein in 50 mM phosphate buffer, pH 7.2, was incubated for 1 h at a temperature which induces thermal denaturation of a given substrate (55, 60, and 65°C for rhodanese, ␤ L -crystallin, and ␥-crystallin, respectively). While in the absence of the chaperone, the thermally denaturating proteins aggregate, ␣-crystallin prevents the aggregation process by forming stable, water-soluble complexes with the aggregation-prone proteins (1,15,16). After incubation, the samples were cooled down to room temperature and the complexes were separated from the remaining free substrate proteins by repetitive filtration using a 100-kDa cut-off Microcon microconcentrators (Amicon) until no tryptophan fluorescence was detectable in the filtrate. The absence of unbound substrates in the retentate was further verified by size exclusion chromatography on a Sephacryl S-300 column. The molar ratio of ␣-crystallin to the substrate protein in a given complex was estimated by subtracting the amount of the unbound substrate (as determined spectrophotometrically) from the initial amount of this protein in the incubation mixture.
Tryptophan Fluorescence Measurements and Quenching Experiments-Tryptophan fluorescence spectra were measured on an SLM 8100 spectrofluorometer using the excitation wavelength of 295 nm. Quenching experiments were performed by titrating the solution of free-and W9F ␣A-crystallin-bound proteins with freshly prepared 5 M solutions of acrylamide or potassium iodide. Fluorescence intensities were measured at the wavelength corresponding to the emission maximum of each protein and were corrected for dilution, blanks, and the inner filter effect. The effective quenching constants (K SV ) eff (equal to f a K SV ) were calculated from the inverse slopes of the F 0 /⌬F versus 1/[Q] plots according to the modified Stern-Volmer equation: , where F 0 and F are the fluorescence intensities in the absence and presence of the quencher, [Q] is the molar concentration of the quencher, and ⌬F ϭ F 0 Ϫ F (24,25). The fraction of quenchable fluorescence, f a , was obtained from the ordinate intercept of the linear portion of the F 0 /⌬F versus 1/[Q] plot (24).
Fluorescence Measurements with bis-ANS-␣A-Crystallin (0.1 mg/ ml) with or without bound substrate was incubated with 20 M bis-ANS for 2 h at room temperature, and fluorescence emission spectra of protein-bound dye were recorded using the excitation wavelength of 390 nm. Since binding of bis-ANS to ␣-crystallin at room temperature is dependent on the thermal history (preincubation temperature) of the sample (19), prior to incubation with the probe, control ␣-crystallin was subjected to the same treatment (i.e. 1-h incubation at elevated temperature) as used to prepare ␣-crystallin complex with a given substrate protein.

RESULTS
The fluorescence characteristics of tryptophan residues depend strongly on the microenvironment and thus provide a sensitive probe of the conformational state of proteins. Measurements of intrinsic fluorescence properties of substrate proteins have been used previously to study protein interaction with the chaperonin GroEL (26 -28) and the heat shock protein DnaJ (29), providing critical information about the conformation of folding intermediates that are captured by these chaperones. Fluorescence studies with proteins bound to ␣-crystallin are hampered by the fact that, in contrast to tryptophanfree GroEL and DnaJ, ␣-crystallin itself contains Trp residues whose emission spectrum overlaps with those of substrate proteins. Therefore, in order to facilitate characterization of the conformational states of proteins bound to ␣-crystallin, we have cloned and overexpressed in Escherichia coli the mutant of ␣A in which the sole Trp at position 9 was replaced with Phe. As anticipated, the conservative Trp 3 Phe substitution had a negligible effect on the structural and functional properties of the protein. Thus, the Sephacryl S-300 size exclusion chromatography profiles for the wild type ␣A-crystallin and the W9F mutant were essentially identical. Furthermore, the mutation fully preserved the chaperone function of the protein as indicated by a very similar ability of both the wild type and Trpfree ␣A-crystallin to suppress the thermal aggregation of rhodanese and ␥-crystallin (data not shown).
In general, when the protein unfolds, it exposes buried tryptophans to the aqueous solvent; this results in the shift of the fluorescence emission maximum toward longer wavelength. The fluorescence spectrum of native rhodanese has a maximum at 332 nm (Fig. 1A, trace 1) and is consistent with a largely apolar environment of eight Trp residues. Upon unfolding in 6 M guanidine HCl, the fluorescence maximum of rhodanese shifts to 352 nm, a wavelength characteristic of tryptophans fully exposed to water. The spectrum of rhodanese bound to W9F ␣A-crystallin has a maximum at 337 nm (Fig. 1A, trace 2). A small (5 nm) red-shift in the fluorescence spectrum indicates that the tertiary structure of the chaperone-bound enzyme is somewhat looser that that of the native form; however, the average environment of Trp residues in the bound protein appears to be still fairly hydrophobic. Notably, the position of the emission maximum for rhodanese associated with ␣A-crystallin is much closer to that observed for the native enzyme than for its chemically denatured form. The fluorescence properties of ␥-crystallin bound to ␣A-crystallin were very similar to those of rhodanese. ␥-Crystallin contains four Trp residues which are uniformly distributed throughout the molecule. As shown in Fig. 1B, unfolding of bovine ␥-crystallin in guanidine HCl results in a 23 nm red-shift of the fluorescence maximum (from 330 nm in the native form to 353 nm). ␥-Crystallin bound to ␣A-crystallin shows an emission maximum at 337 nm, i.e. red-shifted only by 7 nm from that of its native state. These observations suggest that the conformation in which both proteins are captured by ␣-crystallin is characterized by a rela- ␣-Crystallin as a Molecular Chaperone 10450 tively low degree of unfolding. Furthermore, the conformation of ␣-crystallin-associated proteins appears to be remarkably stable; there is very little change (2-3 nm) in fluorescence maxima of bound rhodanese or ␥-crystallin upon heating of the complexes to temperatures well above those at which free proteins undergo thermal denaturation (Table I).
Further insight into the tertiary structure of ␣-crystallinbound substrate proteins may be obtained from fluorescence quenching experiments. Acrylamide and iodide were used as a complementary set of water-soluble quenchers of tryptophan fluorescence. Neutral acrylamide, although relatively polar, has the ability to penetrate into protein interior (25). Iodide is a highly hydrated, negatively charged molecule; its quenching ability is limited almost exclusively to surface-exposed tryptophans and also depends on the location of the neighboring charged groups. In general, the interpretation of fluorescence quenching data for multitryptophan proteins is complicated by factors such as heterogeneous emission and varied accessibility of individual tryptophan residues to the quenchers. Therefore, we have used in our analysis the effective Stern-Volmer constant (K SV ) eff ; this effective parameter represents weighted average of the quenching constants of individual Trp residues and may contain contributions from both dynamic and static quenching processes (25). Fig. 2 shows representative Stern-Volmer plots for the quenching of Trp fluorescence of rhodanese and ␥-crystallin under various experimental conditions. The quenching parameters (K SV ) eff and f a are summarized in Table  I. Unfolding of both rhodanese and ␥-crystallin in 6 M guanidine HCl at 25°C results in a greatly increased solvent exposure of Trp residues, as indicated by about a 6-fold increase in (K SV ) eff for acrylamide. This may be contrasted with only a 2-fold increase of the above parameter observed for the proteins bound to ␣-crystallin. A very similar trend may be found when the characteristics of iodide quenching are compared, although in this case the analysis is somewhat more complicated since only a fraction of the emission arises from iodide-quenchable fluorophores (as indicated by a downward curvature of the respective Stern-Volmer plots). For both proteins, the fraction of iodide-accessible tryptophans in ␣-crystallin-bound conformers is larger compared to the native states. However, the increase in (K SV ) eff values for bound proteins is very modest, especially when compared with that observed upon unfolding of the proteins in guanidine HCl. Altogether, the results of the quenching experiments support the notion that the structure of ␣-crystallin-bound proteins is characterized by a relatively low degree of unfolding. Although somewhat loosened, this structure is clearly closer to the native conformation than that of chemically denatured proteins. It is also worthwhile noting that heating of the complexes to temperatures as high as 75°C has only a small effect on the quenching parameters of bound proteins. This is consistent with the lack of major changes in max at high temperatures and further indicates that the conformation of ␣-crystallin-associated proteins is highly stable.
The fluorescence of bis-ANS is strongly dependent on the polarity of the environment: it is very weak in water and increases dramatically upon binding to hydrophobic sites of proteins. This compound has been widely used as a probe to assess the exposure of hydrophobic surfaces in proteins (19,30). As shown previously, bis-ANS has considerable affinity for ␣-crystallin; this indicates the presence of surface-exposed hydrophobic patches (19). In contrast, essentially no binding of the probe could be detected to native rhodanese, ␥and ␤ Lcrystallin. However, when the same proteins were bound to the chaperone, the fluorescence of bis-ANS in the presence of the complexes was substantially higher than that for ␣-crystallin alone. Furthermore, the increase in fluorescence intensity appeared to correlate with the amount of bound substrates (Fig.  3). This strongly indicates that, in contrast to the native states, the conformation of chaperone-associated proteins is characterized by high affinity for the hydrophobic probe.

DISCUSSION
Employing site-directed mutagenesis and fluorescence spectroscopy, we have explored the structural properties of nonnative proteins that are bound to ␣A-crystallin. The present data clearly show that the fluorescence characteristics of Trp residues in ␣-crystallin-bound conformers are much closer to those of native proteins than the fully unfolded ones. This strongly suggests that ␣-crystallin stabilizes aggregation-prone proteins in a conformation which, although compromised, remains relatively compact and is characterized by a low degree of unfolding.
The specific conformational features of substrate proteins bound to various functionally different chaperones have been the subject of many recent studies (for review see Refs. 31 and 32). For example, DnaK and PapD are believed to bind polypeptides in an extended conformation. In contrast, substrate proteins bound to the chaperones GroEL and DnaJ appear to be in a partially folded conformation, often characterized as a molten globule state with a largely preserved secondary structure and a collapsed, highly flexible tertiary structure (26 -29, 32). It is informative to compare the fluorescence characteristics of the same substrate, rhodanese, bound to ␣-crystallin and the two latter chaperones.The fluorescence emission maximum for rhodanese bound to GroEL or DnaJ occurs at 342-343 nm (26,29), i.e. it is red-shifted by 10 -11 nm compared to that of the native enzyme. In contrast, max maximum for rhodanese associated with ␣-crystallin occurs at a wavelength as low as 337 nm, indicating a substantially more hydrophobic (native-like) environment of Trp residues as compared with that of rhodanese bound to GroEL or DnaJ. A more compact (less accessible) tertiary structure of the substrate protein associated with ␣-crystallin than that bound to GroEL is further supported by the quenching data for rhodanese stabilized by these different chaperones (cf. data for GroEL in Refs. 26 and 27 and present results for ␣-crystallin). A potential complication in the interpretation of fluorescence data for substrate proteins is caused by the possibility that the chaperone itself could affect the environment of tryptophan residues by shielding them from the aqueous environment. This indeed may be a factor for GroEL- However, since ␣-crystallin can accommodate as many as 40 substrate molecules per 800-kDa oligomer (one substrate protein per subunit), geometric considerations point to the location of substrate proteins on the surface of the chaperone. In this case, the role of shielding effects mentioned above would be much less significant. We and others have recently shown that ␣-crystallin has substrate specificity strikingly different from other chaperones: it binds non-native protein structures formed on the denaturation pathway, but apparently has no affinity for intermediates formed on the refolding pathway (21,34). Our present results complement the previous findings by providing crucial information about the conformational state of proteins captured by this chaperone. In view of these combined data, one can postulate that ␣-crystallin preferentially recognizes non-native structures characterized by an increased surface hydrophobicity but a remarkably low degree of unfolding, i.e. the aggregation-prone conformers that occur very early on the denaturation pathway. Such a unique substrate specificity would be fully consistent with the putative physiological role of ␣-crystallin as a "junior chaperone" specifically designed to suppress irreversible aggregation of proteins under stress conditions. The above function of ␣-crystallin may be of particular importance in the ocular lens (15). Major physiologically relevant factors known to induce aggregation of lens proteins include ultraviolet radiation and oxidative stress. While the above insults are highly unlikely to cause extensive unfolding of protein molecules, they may induce perturbations in the tertiary structure, leading to an increased surface hydrophobicity and, eventually, protein aggregation (17). With its ability to capture and stabilize these early non-native structures, ␣-crystallin appears to be uniquely well suited to chaperone the transparency properties of the ocular lens.