Binding of Destabilized βB2-Crystallin Mutants to α-Crystallin THE ROLE OF A FOLDING INTERMEDIATE

Age-related changes in protein-protein interactions in the lens play a critical role in the temporal evolution of its optical properties. In the relatively non-regenerating environment of the fiber cells, a critical determinant of these interactions is partial or global unfolding as a consequence of post-translational modifications or chemical damage to individual crystallins. One type of attractive force involves the recognition by α-crystallins of modified proteins prone to unfolding and aggregation. In this paper, we explore the energetic threshold and the structural determinants for the formation of a stable complex between α-crystallin and βB2-crystallin as a consequence of destabilizing mutations in the latter. The mutations were designed in the framework of a folding model that proposes the equilibrium population of a monomeric intermediate. Binding to α-crystallin is detected through changes in the emission properties of a bimane fluorescent probe site-specifically introduced at a solvent exposed site in βB2-crystallin. α-Crystallin binds the various βB2-crystallin mutants, although with a significantly lower affinity relative to destabilized T4 lysozyme mutants. The extent of binding, while reflective of the overall destabilization, is determined by the dynamic population of a folding intermediate. The existence of the intermediate is inferred from the biphasic bimane emission unfolding curve of a mutant designed to disrupt interactions at the dimer interface. The results of this paper are consistent with a model in which the interaction of α-crystallins with substrates is not solely triggered by an energetic threshold but also by the population of excited states even under favorable folding conditions. The ability of α-crystallin to detect subtle changes in the population of βB2-crystallin excited states supports a central role for this chaperone in delaying aggregation and scattering in the lens.

Age-related changes in protein-protein interactions in the lens play a critical role in the temporal evolution of its optical properties. In the relatively non-regenerating environment of the fiber cells, a critical determinant of these interactions is partial or global unfolding as a consequence of post-translational modifications or chemical damage to individual crystallins. One type of attractive force involves the recognition by ␣-crystallins of modified proteins prone to unfolding and aggregation. In this paper, we explore the energetic threshold and the structural determinants for the formation of a stable complex between ␣-crystallin and ␤B2-crystallin as a consequence of destabilizing mutations in the latter. The mutations were designed in the framework of a folding model that proposes the equilibrium population of a monomeric intermediate. Binding to ␣-crystallin is detected through changes in the emission properties of a bimane fluorescent probe site-specifically introduced at a solvent exposed site in ␤B2-crystallin. ␣-Crystallin binds the various ␤B2-crystallin mutants, although with a significantly lower affinity relative to destabilized T4 lysozyme mutants. The extent of binding, while reflective of the overall destabilization, is determined by the dynamic population of a folding intermediate. The existence of the intermediate is inferred from the biphasic bimane emission unfolding curve of a mutant designed to disrupt interactions at the dimer interface. The results of this paper are consistent with a model in which the interaction of ␣-crystallins with substrates is not solely triggered by an energetic threshold but also by the population of excited states even under favorable folding conditions. The ability of ␣-crystallin to detect subtle changes in the population of ␤B2-crystallin excited states supports a central role for this chaperone in delaying aggregation and scattering in the lens.
In the inner regions of the lens, transparency and refractivity are dependent on the stability, high solubility, and packing of three families of proteins, collectively referred to as crystallins (1)(2)(3). One of these families, the ␣-crystallins, consists of two small heat-shock proteins (sHSP) that can recognize and bind non-native states of proteins. The other two families, the ␤and ␥-crystallins (4 -6), are evolutionary related structural proteins that have close to 30% sequence identity and similar polypeptide chain folds. The double Greek key motif characteristic of their structures has been found to occur in microbial stress proteins (7,8). Subunits of ␤-crystallins assemble into dimers and higher oligomers, whereas ␥-crystallins are monomeric.
The ␤-crystallin family of the vertebrate lens consists of six distinct gene products that segregate into two classes, basic and acidic (9). Despite extensive sequence similarity, one of the hallmarks that distinguish members of the two classes is Nand C-terminal extensions of variable lengths, the age-related truncation of which adds another dimension of molecular diversity (10 -13). The structures of ␤B2-crystallin and extension truncation mutants have been determined to high resolution (5, 14 -18). Wild type ␤B2-crystallin structure reveals an all-␤sheet fold consisting of two domains with identical double Greek key topology. ␤B2-crystallin homodimers assemble as a result of intermolecular domain pairing. Compelling evidence suggests that, at low concentrations, some ␤-crystallins may populate a monomeric state in solution (19 -21). Two models for the conformation of monomeric ␤-crystallin have been proposed. The first is a ␥-crystallin-like closed conformation with intra-subunit pairing between the N-and C-terminal domains (20), whereas the second has an open conformation with an unfolded N terminus (21,22).
In the low protein turnover environment of lens fiber cells, the ␤-crystallins are subject to extensive modifications, either programmed or as a result of oxidative and other types of damage (12,13,(23)(24)(25)(26). Regardless of their origin, these modifications are expected to shift the equilibrium between dimeric, monomeric, and unfolded ␤-crystallin. The consequences of the altered equilibrium may include changes in solubility of the crystallins and the balance of forces that defines their mutual interactions, both of which may lead to aggregation and subsequent loss of transparency.
One of the hypothesized mechanisms by which the lens delays the onset of scattering is through the chaperone activity of the resident small heat-shock protein, ␣-crystallin (3,27,28). Whether due to age-related modifications or as a consequence of changes in the physico-chemical environment, ␤and ␥-crystallins that significantly populate non-native states associate with the ␣-crystallins and are sequestered; hence, their aggregation is suppressed. Although previous studies have demonstrated a reduction in light scattering by heat-denatured ␤and ␥-crystallins in the presence of ␣-crystallin (29 -33), the energetic and structural aspects of this interaction have not been investigated under conditions that mimic the lens environment. Rarely do site-specific modifications cause complete or global unfolding; therefore, ␣-crystallins must detect the increased excursions of substrate proteins to non-native states under conditions where the folded state is still the predominantly populated state.
In the context of a mechanistic study of ␣-crystallin chaperone activity, we have developed an equilibrium binding assay using T4L 1 as a model substrate (34). A set of destabilized T4L mutants with similar structures of the folded state but progressively decreasing free energy of unfolding was constructed. Differential binding of ␣-crystallin to these mutants suggests that ␣-crystallin recognizes and binds excited states of T4L (34 -37). Complex formation occurs under conditions that favor the folded state with equilibrium constants of the folding reaction of at least 10 4 . ␣Aand ␣B-crystallin can "sense" the change in stability of T4L through thermodynamic coupling of the binding reaction to the equilibria that describe the excursion from the folded state to the excited states (34,38). Binding occurs through two different modes, each characterized by substantially different affinity and number of binding sites.
To test the chaperone mechanism of lens transparency, we extended this approach to the ␤-crystallins. ␤B2-crystallin mutants with different degrees of destabilization of the native state were constructed. The rationale for using ␤B2-crystallin is the availability of a crystal structure to guide in the design of the mutations (14). The sites selected for this study were either buried at the contact interface of the dimer or in the fold of a subunit. In the context of the folding model of ␤-crystallin, the former are likely to increase the equilibrium population of the monomeric intermediate. This paper reports the thermodynamic and solution characteristics of these mutants and the analysis of their interactions with the ␣-crystallins.

EXPERIMENTAL PROCEDURES
Cloning and Site-directed Mutagenesis-The cDNA of mouse ␤B2crystallin (a generous gift from Drs. J. F. Hejtmancik and Y. Sergeev) was cloned between the NdeI and XhoI sites of the plasmid pET-20(bϩ). The cloned DNA was verified by DNA sequencing and determined to have an identical DNA sequence to that deposited in the GenBank TM under accession number P26775. A cysteine-less ␤B2-crystallin was generated in which Cys-38 and Cys-67 were mutated to leucine and serine, respectively. Overlap, extension site-directed mutagenesis was performed to generate the PCR fragments G61D/F74A, L165A, and E167L using either a cysteine-less template or a variant containing a unique cysteine at position 66. The fragments were then digested and subcloned into the pET 20(bϩ) vector between the NdeI and XhoI sites. All mutant constructs were sequenced to confirm the substitution and the absence of nonspecific mutations. Single-site mutants are named by specifying the original residue and the number of that residue followed by the new residue.
Protein Expression and Purification-␣A-crystallin, ␣B-crystallin, and ␣B-crystallin phosphorylation mimics were expressed and purified as described previously (34,38). ␤B2-crystallin plasmids were used to transform competent Escherichia coli BL21(DE3) cells. Cultures, inoculated from overnight seeds, were grown to mid-log phase at 37°C, and the expression of ␤B2-crystallin was induced by the addition of 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside. After 3 h of induction, the cells were harvested by centrifugation and resuspended in lysis buffer (20 mM Tris, 0.1 mM EDTA, 0.02% NaN3, and 10 mM dithiothreitol, pH 8). The resuspended cells were disrupted by sonication, and the DNA was precipitated by the addition of 0.017% polyethyleneimine. The lysates were then centrifuged at 15,000 ϫ g. ␤B2-crystallin in the supernatant was purified using anion exchange chromatography followed by size exclusion chromatography.
All ␤B2-crystallin mutants expressed in E. coli formed inclusion bodies even at expression temperatures as low as 28°C. Therefore, the harvested cells were resuspended in lysis buffer containing 1% Triton X-100. The resuspended cells were disrupted by sonication and centrifuged at 15,000 ϫ g. The pellet containing the inclusion bodies was washed repeatedly to remove trace amounts of Triton X-100. The inclusion bodies were then solubilized in 8 M urea and incubated at room temperature for 1 h. The solubilized protein was refolded by dilution into cold refolding buffer (15 mM MES, 35 mM Tris, 50 mM NaCl, 0.1 mM EDTA, 0.02% NaN 3 , 10 mM dithiothreitol, and 10% glycerol, pH 8). Refolded ␤B2-crystallin mutants were further purified by size exclusion chromatography. Purified ␤B2-crystallin mutants were reacted with the fluorescent probe bimane as described earlier and illustrated in Scheme 1 (39,40). The samples were then desalted into a buffer consisting of 9 mM Tris, 6 mM MES, 50 mM NaCl, and 10% glycerol, pH 8. Where needed, bimane-labeled mutants are designated as such by the addition of the suffix BI.
Conformational Stability of ␤B2 Mutants-The conformational stability of ␤B2-crystallin mutants was determined by denaturant unfolding tryptophan fluorescence at room temperature. ␤B2-crystallin solutions at a constant concentration of 2 M were titrated with increasing urea concentration. The unfolding curves were obtained by following the changes in tryptophan emission at 320 nm as a function of urea concentration on a PTI L-format spectrofluorometer. All unfolding curves were obtained at pH 8.
Circular Dichroism-Far-and near-UV CD measurements were performed on a Jasco 810 spectropolarimeter equipped with a thermostated cell holder. Far-UV CD spectra were recorded between 190 and 260 nm using a 1-mm pathlength cell. For near-UV CD studies, a 1-cm pathlength cell was used, and the spectra were recorded between 250 and 360 nm. Protein samples were prepared in buffer containing 9 mM Tris, 6 mM MOPS, and 50 mM NaCl, pH 8. All measurements were corrected by background subtraction, and each spectrum presented is the average of 10 scans.
Binding of ␤B2 Mutants to ␣-Crystallin-Binding studies of bimanelabeled ␤B2-crystallin mutants were carried out on a PTI L-format spectrofluorometer equipped with an RTC2000 temperature controller and a sample holder containing a Peltier heater/cooler. Samples containing 5 or 10 M ␤B2-crystallin and varying concentrations of ␣-crystallin were incubated at the desired temperature for 2 h. The fluorescence emission spectra of the samples were recorded in the 400 -500 nm range after excitation of the bimane molecule at 380 nm. The data was plotted as the emission intensity at 460 nm versus the molar ratio of ␣-crystallin to ␤B2-crystallin.

RESULTS
General Methodology-The folding model of ␤B2-crystallin, proposed by Jaenicke and co-workers (21,22), involves the population of a partially unfolded monomeric intermediate. Equilibrium sedimentation studies of ␤A3-crystallin molecular weight were interpreted in terms of a reversible equilibrium involving a monomeric intermediate with a ␥-crystallin-like conformation (20). Therefore, the folding equilibrium of ␤-crystallin can be written as shown in Equation 1, where D is the native dimer, I is the putative monomeric intermediate, and U is the unfolded state. Regardless of its conformation, it is expected that the equilibrium population of I can be enhanced by selective destabilization of the dimer. A direct approach to destabilize D is by disruption of the interactions at the monomer-monomer interface. Two mutants of ␤B2-crystallin were designed for this purpose. The L165A substitution reduces the buried surface area between the two monomers. The original leucine is involved in Van der Waals (VDW) contacts with Ile-43 and Val-81 from the other subunit. The second mutant, E167L, disrupts an ionic interaction with Arg-79 from the paired domain. Unlike L165A, the mutation does not significantly reduce the molar volume of the side chain.
The third ␤B2-crystallin mutant was designed to disrupt the packed core of the N-terminal domain. The mutation G61D/ F74A simultaneously reduces the buried hydrophobic surface area and introduces a negative charge. The two sites are proximal in the folded structure, such that the removal of the bulky phenylalanine side chain allows the accommodation of the new aspartate side chain. A molecular model of the three mutants is shown in Fig. 1.
To detect complex formation, an environmentally sensitive reporter group, such as a spin label or a fluorescence probe, is site-specifically incorporated at a solvent-exposed site in the substrate. This is achieved via site-directed mutagenesis to introduce a unique cysteine followed by reaction with the spectroscopic probe. Because mouse ␤B2-crystallin has two endogenous cysteines, a cysteine-less variant, hereafter referred to as WT*, was constructed. Cysteine 38, located in a buried environment, was replaced with a leucine, while cysteine 67, located in a surface loop and having a partially water-exposed side chain, was substituted with a serine. WT*-␤B2-crystallin is expressed in the soluble fraction of E. coli and has an identical elution volume on a Superdex 75 column as the WT. Farand near-UV CD spectra of the WT* are consistent with a lack of significant effects on the structure due to the cysteine substitutions (data not shown). All three destabilizing mutations were constructed in the WT* background as well as in a variant of WT*, where a unique cysteine is introduced at the solvent exposed site, Asn-66, in a surface loop.
Structure, Stability, and Solubility of the ␤B2-Crystallin Mutants-In contrast to WT and WT*, all three ␤B2-crystallin mutants were found in inclusion bodies when expressed in E. coli. Following isolation of the inclusion bodies, the mutants were refolded with Ͼ80% efficiency, purified, and then reacted with monobromobimane as described under "Materials and Methods." Analytical SEC was used to verify that the mutants assemble into a dimer. Table I shows that at a 2 M concentration, all three mutants have significant changes in their elution volume, although not to an extent consistent with a shift to a monomer. Similar changes were obtained when the injected sample concentration was 10 M. This may reflect changes in the hydrodynamic radius, the molecular mass, an enhanced interaction with the column matrix, or a shift in the putative dimer/monomer equilibrium. The last interpretation is supported by the fact that the largest shift occurs for the two interface mutants ␤B2-E167L and ␤B2-L165A.
To determine the consequences of the mutations on the thermodynamic stability of ␤B2-crystallin, unfolding curves were constructed from the changes in tryptophan emission intensity as a function of increased urea concentration. As reported previously, fluorescence unfolding curves of WT-␤B2-crystallin do not show a discrete intermediate (21,22). Nevertheless, the non-coincidence of the curves obtained by two spectral probes, fluorescence and CD, was interpreted as an indication of a multi-state equilibrium (22). Equation 1 predicts that the dimer to intermediate transition is bimolecular and, thus, is concentration dependent. Quantitative analysis of the unfolding curves (41) in Fig. 2 is hindered by the large number of parameters to be determined and the monophasic nature of the corresponding tryptophan unfolding curves that provide little restriction on the values of ⌬G and m (the denaturant dependence of ⌬G) for the intermediate. Therefore, the curves in Fig.  2 were compared using C1 ⁄2 , the urea concentration at the tran- sition midpoint. Although C1 ⁄2 is a qualitative estimator of protein stability, it can be unequivocally interpreted in cases where it reflects large right or left shifts in unfolding curves. The left shift in the denaturation curve of ␤B2-L165A indicates a significant destabilization of this mutant due to the loss of the two leucine side chains in the dimer. In contrast, the loss of the ionic interaction between Glu-167 and Arg-79 does not affect the overall stability of the dimer, as the denaturation curve of this mutant is right-shifted relative to that of the WT. The loss of the phenylalanine side chain and the introduction of a charge in the core of the N-terminal domain result in a relatively small change in stability. WT-␤B2, WT*-␤B2, and the Asn-66BI variant have virtually superimposable unfolding curves (data not shown). The unfolding curves of the ␤B2-L165A, ␤B2-E167A, and ␤B2-G61D/F74A constructed in the WT* background reproduced the trends of Fig. 2 and have C1 ⁄2 values similar to those of their bimane-labeled variants.
CD spectroscopy was used to determine whether the mutations resulted in gross changes in the structure of ␤B2-crystallin. The superimposable far-UV CD spectra of the WT and the mutants in Fig. 3, panels a and c, indicate little if any change in the overall secondary structure of the protein. The near-UV CD spectra, compared in Fig. 3, panels b and d, also have a similar overall shape consistent with the lack of significant alteration in the environments of aromatic residues. The relatively weak signal-to-noise is the consequence of an interesting change in the apparent surface properties of the two interface mutants. Although both ␤B2-L165A and ␤B2-E167L were soluble at low concentrations for days, they tended to precipitate when concentrated to more than ϳ1 mg/ml. This is not the case for ␤B2-G61D/F74A, where the mutations are not located at the monomer-monomer interface.
Bimane Emission Intensity in Unfolded ␤B2-Crystallin and in the Complex with ␣-Crystallin-To directly observe complex formation between the ␣-crystallins and their substrates, we have previously utilized electron paramagnetic resonance spectroscopy to monitor the changes in the spectrum of a nitroxide reporter group upon binding to ␣-crystallins (34). However, the limited solubility of the ␤B2-crystallin mutants precludes the use of this approach at the required concentrations. An alternative detection method, introduced by Sathish et al. (42), is based on the change in the spectral properties of the fluorescent probe bimane. In general, the emission max of this probe in proteins reflects the solvent accessibility at the site of attachment (39,40). However, upon binding of bimane-labeled T4L to ␣-crystallin, the blue shift in max is accompanied by a decrease in the emission intensity. The latter has been shown to arise from a binding-induced transition in the conformation of T4L (42).
The use of bimane-labeled ␤B2-crystallin allows for a convenient assay to screen for binding. Bimane-labeled ␤B2-crystallin mutants are incubated with ␣B-crystallin and the phosphorylation mimics of ␣B, namely S45D, S45D/S59D (D2), and S19D/S45D/S59D (D3). Previous studies have revealed that the Ser to Asp substitutions result in the activation of two-mode binding manifested by increased extent of binding and higher affinities (38). After incubation, the ␤B2-mutant/chaperone mixture is loaded on a size-exclusion column. The complex is detected by the presence in the chaperone peak of bimane emission at 470 nm. The stability of the complex on the time scale of SEC suggests a relatively slow off rate. Fig. 4 compares the extent of binding between ␤B2-L165A and the various ␣B-crystallins at the same molar ratio. The D3 form has the highest extent of binding, as inferred from the drop in the intensity of the ␤B2-crystallin peak and the concomitant increase in the intensity of the chaperone peak.
Complex formation results in a change in the fluorescence characteristics of the bimane. This is illustrated in Fig. 5 by the emission spectra of ␤B2-L165A in the free and bound states. The increase in quantum yield upon binding is accompanied by a blue shift in the max , suggesting transfer to a more buried environment. The sign of the change in the intensity is opposite to that observed upon binding of bimane-labeled T4L to either the high or the low affinity mode of ␣-crystallin (42). To inves-  tigate the origin of the increased fluorescence, the emission intensity at 470 nm was measured as a function of urea concentration. For all mutants, unfolding results in an increase in the emission at 460 nm without significant changes in the max , as shown by the unfolding curves of Fig. 6. Furthermore, for ␤B2-E167L the curve is biphasic, suggesting the presence of an equilibrium intermediate where the bimane emission characteristics are distinct from those of the folded and unfolded states, consistent with the folding model of Equation 1.
Binding Isotherms of ␤B2-Crystallin Mutants to ␣-Crystallin-The rather small increase in intensity observed in the presence of ␣Aand ␣B-crystallin in Fig. 5 indicates that the concentrations used in the binding assay are significantly smaller than the dissociation constant. The limited solubility of ␤B2-L165A and ␤B2-E167L prevents the use of the high concentration necessary for quantitative analysis. Among the ␣-crystallins and their variants, only ␣B-D2 and ␣B-D3 bind ␤B2-L165A to an extent that an isotherm can be constructed. Because this mutant precipitates at 37°C (in the absence of ␣-crystallin), binding isotherms were obtained at 23 and 30°C and the latter is shown in Fig. 7a. Superimposed on the binding isotherm to ␣B-D3 is a non-linear, least squares fit based on a single mode binding model. A fit using a two-mode model slightly improves 2 , but the resulting value of n 2 is 0.07, which is significantly different from the value obtained from analysis of T4L binding. Similarly, a single mode binding model provides a satisfactory, although not unique, fit to the binding isotherm of ␣B-D2 (data not shown). Although the binding of ␤B2-L165A is expected based on its marginal stability, the range of the K D (Table II), when compared with those obtained with T4L mutants (42), suggests a markedly reduced affinity.
In contrast to ␤B2-L165A, the stability of ␤B2-E167L relative to the WT, deduced from the midpoints of the unfolding transitions, suggests that this protein will not significantly bind ␣B-D2 or ␣B-D3. Fig. 7b shows that, at 37°C, this mutant binds with an affinity similar to that of ␤B2-L165A (Table II), and SEC experiments similar to those reported in Fig. 4 indicate that this mutant binds ␣B-D1 and both WT ␣-crystallins at 37°C (data not shown). In contrast, to obtain measurable binding of ␤B2-G61D/F74A, a concentration of 10 M and a large excess of ␣B-D3 is required (Fig. 7b).
The differential recognition of ␤B2-E167L relative to ␤B2-WT and ␤B2-G61D/F74A is not correlated with changes in overall stability, as defined by the equilibrium population of U, because the Trp unfolding curve of ␤B2-G61D/F74A is leftshifted relative to that of ␤B2-E167L. Rather, the bimane unfolding curves (Fig. 6) indicate that binding is triggered by the dynamic population of a folding intermediate. Consistent with this interpretation, the unfolding curves of ␤B2-E167L and ␤B2-G61D/F74A (Fig. 6) beyond 2 M urea, which according to Equation 1 describe the I to U transition, are very similar. If ␣-crystallin binds I, then ␤B2-L165A should have the highest equilibrium population of I. The monophasic shape of its unfolding curve is thus interpreted as predominantly representing the I to U transition.
Because damaged ␤B2-crystallins in the lens exist in a background of native proteins, we tested the effect of the presence of the WT on complex formation with ␣B-D3. Neither the preincubation of ␤B2-E167L with WT nor the addition of the latter after complex formation significantly affected the binding (data not shown).

DISCUSSION
That the ␣-crystallins bind destabilized ␤B2-crystallin mutants lends support to the hypothesis that their chaperone activity plays a critical role in the long-term maintenance of lens transparency. Although previous studies have shown that ␣-crystallin can suppress the heat-induced aggregation of ␤-crystallins (29 -32), the novel aspect of this work is that binding occurs to ␤B2-crystallin mutants of similar stability to the WT and under equilibrium conditions where the mutants are predominantly in the folded state. Furthermore, the use of site-directed mutants more closely mimics the localized agerelated damage that may occur in the lens than the extreme temperatures used to unfold and aggregate the ␤-crystallins. The finding that the complex with ␣-crystallin is stable even in the presence of WT ␤B2-crystallin is critical because, in the lens, damaged proteins exist in a background of stable proteins.
Determinants of ␤B2-Crystallin Stability-Despite the different nature of the two interface mutations manifested by their differential effects on the stability curves, both result in changes in the apparent surface properties of the ␤B2-crystal-lin molecule as inferred from the tendency to aggregate at high concentrations. The change in the surface properties for ␤B2-E167L occurs despite the lack of a significant change in its stability. Presumably, residual hydrophobic interactions occur between the introduced Leu and Arg-79 side chains. Because ␤B2-L165 and ␤B2-E167 are buried at the interface between the subunits, it is unlikely that the mutants cause loss of solubility of the dimer. An alternative interpretation, based on Equation 1, is that the mutations increase the equilibrium population of the monomeric intermediate I by reducing the stability of the dimer. The limited solubility and the tendency to bind ␣-crystallin reflect the partially unfolded structure of I. Such a mutationinduced shift in the dimer-monomer equilibrium can account for the shift in the retention times in SEC. Furthermore, it is directly supported by the biphasic unfolding curve of ␤B2-E167L reported by the bimane probe (Fig. 6). The lack of the explicit detection of this intermediate in the tryptophan unfolding curve may be due to the existence of five such residues in ␤B2-crystallin. Consequently, the population of this intermediate may not affect their environments in a similar fashion.
Despite its location at a solvent-exposed site in a surface loop, the emission intensity of the bimane is sensitive to the conformational state of the ␤B2-crystallin molecule. A similar effect is observed in T4L, where global unfolding significantly decreases the bimane emission intensity at a surface-exposed site (42). The mechanistic origin of this effect has not been investigated, although it is likely to involve a change in the average distance between the bimane probe and various Trp and Tyr residues in the folded versus unfolded states. Because Asn-66BI is in close proximity to Tyr-62, the increase in the bimane emission intensity in the intermediate may reflect the partial unfolding of the N terminus. For both T4L and ␤B2crystallin, the sign of the intensity change due to binding and unfolding is the same.
Another remarkable result is the rather limited destabilization observed in ␤B2-G61D/F74A. Phe to Ala substitutions in the buried core of T4L as well as the introduction of negative charges result in at least a 3-5 kcal/mol change in the free energy of unfolding (43). The tolerance of ␤B2-crystallin to the substitution can also be interpreted within the context of Equation 1. If I has an unfolded N terminus and the transition between D and I occurs under ambient temperatures, then the contribution of this region to the overall stability of the protein FIG. 7. a, binding isotherms of ␤B2-L165A to ␣B-D3 and ␣B-D2 at 30°C, pH 8. The superimposed solid line is a non-linear least squares fit using a one-mode binding model. B, binding isotherms of ␣B-D3 to ␤B2-E167L and ␤B2-G61D/F74A at 37°C, pH 8. The superimposed solid line is a non-linear least squares fit using a one-mode binding model. may not be significant. The core mutations do not affect the solution behavior of ␤B2-crystallin. ␣-Crystallin/␤B2-Crystallin Interactions-Both ␣-crystallin subunits bind the ␤B2-crystallin mutants. However, despite the large decrease in the stability of ␤B2-L165A, the extent of binding to WT ␣Aand ␣B-crystallin is not sufficient for quantitative analysis. This suggests that the K D values associated with binding are in the tens of micromolar range. Furthermore, it suggests a significantly lower affinity than that reported for T4L. If binding was determined by an absolute energetic threshold, the opposite result is predicted. ␤B2-crystallin unfolding curves indicate significantly lower overall stability than T4L (34).
The limited solubility of the mutants prevented the use of the higher concentrations needed for the construction of binding isotherms. Therefore, insight into the thermodynamics of binding was gleaned from binding curves to ␣B-D3 and ␣B-D2, both of which bind ␤B2-crystallin mutants to a larger extent. The use of these variants is solely for the purpose of enhancing binding and is thermodynamically equivalent to increasing the temperature without the complications of aggregation. The conclusions obtained from these curves are applicable to both ␣-crystallin subunits (38).
Analysis of binding isotherms reveals that the highest affinity of ␣B-D3 and ␣B-D2 is for the most destabilized ␤B2crystallin mutant consistent with previous studies on the binding of T4L-destabilized mutants (38). However, a novel finding of the present work is that ␣-crystallin can interact differentially with proteins that have similar overall stability. Whereas ␣B-D3 binds ␤B2-E167L, as shown in Fig. 7, it does not significantly interact with the WT, although the overall stabilities, reflected in Trp unfolding curves, are similar (Fig. 6). An increase in the equilibrium population of an intermediate with some degree of unfolding, such as I of Equation 1, can be at the origin of the differential recognition. This is consistent with previous results with model substrates demonstrating that ␣-crystallins can recognize states distinct from the globally unfolded state, U (34, 44 -47). According to Fig. 2, there is little difference in the equilibrium population of U between ␤B2-E167L and ␤B2-WT. Presumably, the limited binding of ␤B2crystallin relative to T4L reflects intrinsic differences in the energetic distribution of the excited states relative to the native state.
It is significant in this context that binding to the ␤B2crystallin mutants appears to primarily occur through the high affinity mode. Previous work suggested that high affinity binding by ␣-crystallin is selective for compact substrate states, whereas the low affinity binding involves more extensively unfolded states (38). Thus, the observed single mode binding (Fig. 7) provides another line of evidence consistent with the recognition of the compact monomeric intermediate observed in the unfolding curves. The lack of a contribution by low-affinity binding may reflect a relatively low equilibrium population of the extensively unfolded states, even in marginally stable ␤B2crystallin mutants such as ␤B2-L165A.
Concluding Remarks-The results of this paper demonstrate that the mechanistic outline of ␣-crystallin chaperone activity, deduced from studies with T4L (34,38,42), is general and applies to the interaction with native lens proteins. The large K D suggests a lower intrinsic affinity for ␤B2-crystallin than for T4L. This is not surprising, considering the need to avoid attractive interactions with native lens proteins. Because the intermediate that triggers binding has also to be populated by the WT, the low affinity may reflect a negative design element to avoid significant binding to ␣-crystallin. It is noted that macromolecular associations are likely to be enhanced in the crowded molecular environment of the lens, where protein concentrations can reach 500 mg/ml.
The results confirm the previously proposed mechanism of the "sensor" in the ␣-crystallins (34). Recognition and binding are not triggered by an absolute energetic threshold, but rather by one that is defined relative to the ladder of excited states that a protein can populate. Therefore, the nature of the mutations may be critical for the recognition by the chaperone as exemplified by the mutant ␤B2-E167L. The results also provide insight into the determinants of stability in the ␤-crystallin family as well as the equilibrium that describes their folding reaction.