Packing-induced Conformational and Functional Changes in the Subunits of α-Crystallin*

The heteroaggregate α-crystallin and homoaggregates of its subunits, αA- and αB-crystallins, function like molecular chaperones and prevent the aggregation of several proteins. Although modulation of the chaperone-like activity of α-crystallin by both temperature and chaotropic agents has been demonstrated in vitro, the mechanism(s) of its regulationin vivo have not been elucidated. The subunits of α-crystallin exchange freely, resulting in its dynamic and variable quaternary structure. Mixed aggregates of the α-crystallins and other mammalian small heat shock proteins (sHSPs) have also been observedin vivo. We have investigated the time-dependent structural and functional changes during the course of heteroaggregate formation by the exchange of subunits between homoaggregates of αA- and αB-crystallins. Native isoelectric focusing was used to follow the time course of subunit exchange. Circular dichroism revealed large tertiary structural alterations in the subunits upon subunit exchange and packing into heteroaggregates, indicating specific homologous and heterologous interactions between the subunits. Subunit exchange also resulted in quaternary structural changes as demonstrated by gel filtration chromatography. Interestingly, we found time-dependent changes in chaperone-like activity against the dithiothreitol-induced aggregation of insulin, which correlated with subunit exchange and the resulting tertiary and quaternary structural changes. Heteroaggregates of varying subunit composition, as observed during eye lens epithelial cell differentiation, generated by subunit exchange displayed differential chaperone-like activity. It was possible to alter chaperone-like activity of preexisting oligomeric sHSPs by alteration of subunit composition by subunit exchange. Our results demonstrate that subunit exchange and the resulting structural and functional changes observed could constitute a mechanism of regulation of chaperone-like activity of α-crystallin (and possibly other mammalian sHSPs) in vivo.

The heteroaggregate ␣-crystallin and homoaggregates of its subunits, ␣Aand ␣B-crystallins, function like molecular chaperones and prevent the aggregation of several proteins. Although modulation of the chaperonelike activity of ␣-crystallin by both temperature and chaotropic agents has been demonstrated in vitro, the mechanism(s) of its regulation in vivo have not been elucidated. The subunits of ␣-crystallin exchange freely, resulting in its dynamic and variable quaternary structure. Mixed aggregates of the ␣-crystallins and other mammalian small heat shock proteins (sHSPs) have also been observed in vivo. We have investigated the time-dependent structural and functional changes during the course of heteroaggregate formation by the exchange of subunits between homoaggregates of ␣Aand ␣B-crystallins. Native isoelectric focusing was used to follow the time course of subunit exchange. Circular dichroism revealed large tertiary structural alterations in the subunits upon subunit exchange and packing into heteroaggregates, indicating specific homologous and heterologous interactions between the subunits. Subunit exchange also resulted in quaternary structural changes as demonstrated by gel filtration chromatography. Interestingly, we found time-dependent changes in chaperone-like activity against the dithiothreitol-induced aggregation of insulin, which correlated with subunit exchange and the resulting tertiary and quaternary structural changes. Heteroaggregates of varying subunit composition, as observed during eye lens epithelial cell differentiation, generated by subunit exchange displayed differential chaperone-like activity. It was possible to alter chaperone-like activity of preexisting oligomeric sHSPs by alteration of subunit composition by subunit exchange. Our results demonstrate that subunit exchange and the resulting structural and functional changes observed could constitute a mechanism of regulation of chaperone-like activity of ␣-crystallin (and possibly other mammalian sHSPs) in vivo.
␣-Crystallin, the major lens protein of the mammalian lens, is a member of the family of ubiquitous small heat shock proteins (1)(2)(3)(4). Like most other small heat shock proteins, ␣-crystallin exists as a high molecular mass complex of a large number of subunits and prevents the aggregation of a number of proteins like a molecular chaperone (5). In most mammals, the two types of subunits of ␣-crystallin, ␣Aand ␣B-crystallins, are present at a molar ratio of 3:1 (6). They are encoded by evolutionarily related genes and have Ͼ50% sequence identity (7). ␣Aand ␣B-crystallins form homoaggregates of high molecular mass and display chaperone-like activity, albeit to different extents (8,9). The demonstration of their independent, non-lenticular expression in tissues such as the heart, thymus, lung, kidney, retina, and brain (10 -13) and overexpression of ␣B-crystallin under stress (14 -19) and in disease conditions (20 -25) suggests individual roles for ␣Aand ␣B-crystallins. In the lens, ␣Aand ␣B-crystallins are expressed differentially, with ␣B-crystallin being expressed earlier in the lens epithelial cells; the ratio of ␣Ato ␣B-crystallin increases during differentiation to fiber cells (26 -30). The significance of this differential expression is not clearly understood. It is not clear as to how the two subunits interchange after expression, leading to the formation of the heteroaggregate. The arrangement of ␣Aand ␣B-crystallin subunits in the ␣-crystallin heteroaggregate is also speculative.
␣-Crystallin is normally isolated as an oligomeric complex with an average of 40 subunits and a molecular mass of 8 ϫ 10 5 Da (6), although its size distribution can vary from 3 ϫ 10 5 to 1.5 ϫ 10 6 Da, depending on the age of the tissue from which it was isolated (31)(32)(33)(34)(35)(36) and isolation conditions such as temperature, pH, and ionic strength (37)(38)(39)(40). Cryoelectron microscopy studies of human recombinant ␣B-crystallin showed that it has a variable quaternary structure (41). The quaternary structure of ␣-crystallin is also dynamic. Calf lens ␣-crystallin that is isolated into five subpopulations with distinct molecular masses returns to its original distribution upon mixing (42), indicating a dynamic equilibrium between ␣-crystallin oligomers. Homoaggregates of ␣Aand ␣B-crystallins are known to exchange subunits among themselves and with the ␣-crystallin heteroaggregate (43)(44)(45)(46). Exchange of subunits between the native and phosphorylated forms of ␣-crystallin has also been detected (43). Subunits of HSP27 1 exchange freely with homoaggregates of ␣A-crystallin (47). These findings suggest that the subunits of ␣-crystallin are freely exchangeable, indicating a dynamic structure for ␣-crystallin and therefore its polydisperse character. Moreover, the demonstration of mixed aggregates of ␣B-crystallin and HSP27 in human heart and in vitro (48 -51) suggests a dynamic quaternary structure for most mammalian small HSPs. The relevance of subunit exchange with regard to the functionality of small HSPs has not been investigated.
Earlier studies from our laboratory have shown that the chaperone-like activity of ␣-crystallin is temperature-dependent (52)(53)(54). ␣-Crystallin shows negligible protection against DTT-induced insulin aggregation or UV light-induced ␥-crys-* This work was supported in part by the Department of Biotechnology, Government of India. 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 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a student fellowship from the University Grants Commission.
§ To whom correspondence should be addressed. tallin aggregation at temperatures below 30°C, but offers significant protection above 30°C. The enhanced chaperone-like activity at elevated temperatures is due to a temperatureinduced structural perturbation that leads to the concomitant exposure of appropriately placed hydrophobic surface(s), which sequester aggregation-prone proteins (52). Interestingly, the exchange of subunits between ␣-crystallin molecules is also temperature-dependent (43,44). The subunits of ␣-crystallin do not exchange at 4°C, but do so at 37°C. This result indicates that the structurally perturbed form of ␣-crystallin, which displays higher chaperone-like activity, exchanges subunits. Earlier studies on the subunit exchange of ␣-crystallin have investigated the dependence of subunit exchange on experimental parameters such as temperature, pH, salt concentration, and the presence of calcium ions (43-46). The functional significance of subunit exchange is, however, not known.
In this study, we have investigated the subunit exchange between ␣Aand ␣B-crystallin homoaggregates and studied the consequent time-dependent structural and functional changes. Our results indicate that the packing of subunits in ␣-crystallin involves specific interactions between ␣Aand ␣Bcrystallins that alter the structure of the individual subunits. We found that packing-induced structural changes in the subunits affect the chaperone-like activity of the proteins. We also demonstrate in vitro that altering the composition of heteroaggregates by the incorporation of ␣B-crystallin subunits by subunit exchange can modulate chaperone-like activity. We feel that this is of significance, as it is known that ␣B-crystallin expression increases under a variety of stress conditions. ␣B-Crystallin expressed under stress conditions can incorporate into preexisting oligomeric small HSPs to increase/modulate its chaperone-like activity. Although the modulation of the chaperone-like activity of ␣-crystallin in vitro by temperature (52)(53)(54) and chaotropic agents (55,56) is known, its regulation in vivo remains largely unexplored. Our results demonstrate that the rearrangement of subunits in ␣-crystallin, due to its dynamic quaternary structure, may provide a possible mechanism to regulate its chaperone-like activity in vivo.

EXPERIMENTAL PROCEDURES
Isolation of ␣-Crystallin-Calf lens ␣-crystallin was isolated and purified as described earlier (57). The fractions corresponding to ␣-crystallin were pooled and concentrated at 4°C using an Amicon ultrafiltration unit with M r 30,000 cutoff. The concentrated solution of ␣-crystallin was stored in Tris-HCl buffer at 4°C. ␣-Crystallin was used as the heteroaggregate of ␣A-and ␣B-crystallins. The ratio of ␣Ato ␣Bcrystallin was ϳ3:1 (w/w) in this heteroaggregate as determined by urea-polyacrylamide gel electrophoresis.
Subunit Exchange between Homoaggregates-Homoaggregates of ␣Aand ␣B-crystallins, filtered separately through 0.45-m Millipore filters, were mixed at various ratios (5:1, 3:1, 1:1, 1:3, and 1:5 (w/w) ␣A/␣B-crystallin) to a final concentration of 1 mg/ml in 10 mM sodium phosphate buffer (pH 7.4) containing 100 mM NaCl. The exchange of subunits was initiated by incubation at 37°C and stopped by cooling to 4°C. To monitor the rate of subunit exchange, the homoaggregates were mixed at a ratio of 3:1 (w/w) ␣A/␣B-crystallin and incubated at 37°C. Samples were taken at various intervals of time (20,40, and 60 min and 1.5, 2, 2.5, 3, 4, 6, 8, 12, and 24 h) and transferred to 4°C to stop further exchange of subunits. As a control, a sample was kept at 4°C for 24 h without exposure to a temperature of 37°C. This was done to check if any residual exchange occurred at 4°C. These samples were used for CD, gel filtration, and protection assays to measure chaperonelike activity. The extent of exchange was monitored by native isoelectric focusing in the pH range of 3-10. Incubation at 37°C for 24 h was sufficient to ensure complete exchange.
Isoelectric Focusing of Homo-and Heteroaggregates-Homoaggregates and heteroaggregates resulting from subunit exchange experiments were analyzed on acrylamide gels (3% T, 3% C) containing 13% (v/v) glycerol and an ampholyte mixture of the following composition: 1.81% (v/v) Pharmalytes 6 -8, 2.26% (v/v) Pharmalytes 3.5-10, and 1.81% (v/v) Pharmalytes 2.5-5. Gels (10 ϫ 20 cm with a thickness of 1 mm) were cast between GelBond film (supported on a glass plate) and a glass plate treated with dimethyldichlorosilane. Flat bed electrophoresis was carried out in an Amersham Pharmacia Biotech electrophoresis unit using strips soaked in 1 M phosphoric acid and 1 M sodium hydroxide and an Amersham Pharmacia Biotech constant power supply unit. Limit settings were 2500 V and 70 mA. Prefocusing was carried out for 45 min at a 30-watt constant power supply. Subsequently, the samples were focused for 30 min at 15 watts and then for 120 min at 30 watts. A Julabo thermostatted water bath was used to maintain a constant temperature of the gel bed at 4°C. The pH gradients were determined by simultaneously running pI markers (Amersham Pharmacia Biotech). These were further confirmed by cutting the gel to 0.5-cm pieces and extracting the ampholytes with 2 ml of water for 48 h. The pH of the resulting solution was measured. After the gels were fixed for 1 h in an aqueous solution of 30% (v/v) ethanol and 11.5% (w/v) trichloroacetic acid and rinsed with 2% (v/v) acetic acid, they were stained for 45 min in a solution of 1.15% (w/v) Coomassie Blue dye (BDH, page blue 83) in 25% (v/v) ethanol and 8% (v/v) acetic acid until a clear background was obtained.
Assay for Chaperone-like Activity-Insulin at a concentration of 0.2 mg/ml (in 10 mM sodium phosphate buffer (pH 7.4) containing 100 mM NaCl) in the presence or absence of a 0.1 mg/ml concentration of the various reconstituted proteins was equilibrated at the required temperature for 10 min with constant stirring in the cuvette using a Julabo thermostatted water bath. The actual temperature in the cuvette was monitored with a Physitemp Microthermocouple thermometer system. The reduction of insulin was initiated by the addition of 30 l of 1 M DTT to 1.5 ml of sample, and the extent of aggregation was monitored by measuring the scattering at 90°with the excitation and emission monochromators in a Hitachi-4000 fluorescence spectrometer set at 465 nm and excitation and emission band passes at 1.5 nm. Results are expressed as percentage protection. Percentage protection is calculated as follows: (I t Ϫ I tϩ␣ )/I t , where I t is the intensity of scattered light for target protein insulin and I tϩ␣ is the intensity of scattered light in the presence of homoaggregates or heteroaggregates of ␣Aand ␣B-crystallins.
Circular Dichroism Measurements-The CD measurements were carried out in 10 mM phosphate buffer using a Jasco J-715 spectropolarimeter. ␣Aand ␣B-crystallin concentrations were 1.0 mg/ml in 10 mM sodium phosphate and 100 mM NaCl (pH 7.4). Water-jacketed 1and 0.01-cm path length cells were used for recording near-and far-UV CD spectra, respectively. ␣Aand ␣B-crystallins at a ratio of 3:1 (w/w) at a final concentration of 1 mg/ml were used for the temperature-dependent subunit exchange and the associated CD measurements. Spectra of the mixture of homoaggregates (3:1 ␣A/␣B-crystallin) were recorded prior to and after subunit exchange (by incubating at 37°C). Sample temperature was maintained at 15°C in water-jacketed sample cells of 1 and 0.01 cm using a Julabo thermostatted water bath. In another experiment, subunit exchange was carried out in the waterjacketed sample cell by maintaining the sample temperature at 37°C. Spectra were acquired at 15-min intervals initially and later at 0.5-and 1-h intervals. The shutter was kept closed during incubation intervals to prevent possible photodamage to the protein. The sample temperature was monitored using a Physitemp Microthermocouple thermometer.

RESULTS AND DISCUSSION
␣B-Crystallin and, to a much lesser extent, ␣A-crystallin are found as homoaggregates in a variety of tissues, including the brain, spleen, and heart (10 -13). In the lens, however, the two proteins are found to exist only as a heteroaggregate, ␣-crystallin, although there has been a report showing that ␣Bcrystallin may exist in minute amounts as an independent homoaggregate in the lens (59). ␣-Crystallin from the mammalian lens is a heteroaggregate of ␣Aand ␣B-crystallins present at a molar ratio of 3:1. The significance of this ratio, if any, is not clear. Sun and Liang (45) found that ␣A-crystallin at a ratio of 3:1 ␣A/␣B-crystallin is most efficient in prevention of thermal aggregation of ␣B-crystallin. The composition of the ␣-crystallin heteroaggregate varies with age and during the differen-tiation of the lens epithelial cells to fiber cells (26 -30) and between various species (60 -63). It is known that homoaggregates of ␣Aand ␣B-crystallins can exchange subunits among themselves as well as with the ␣-crystallin heteroaggregate (43-46). This observation led to the proposal of a "dynamic quaternary structure" model of ␣-crystallin that can freely exchange subunits (43). Moreover, it is possible to produce heteroaggregates of varying subunit composition by co-refolding ␣Aand ␣B-crystallins in various ratios (64,65). The presence of two types of subunits in the ␣-crystallin heteroaggregate and the dynamic and variable nature of this aggregate raise questions about the packing of subunits in it. Are the interactions between the subunits nonspecific, dictated mainly by hydrophobic interactions, or are there any specific and preferential interactions? Are the subunits packaged in the heteroaggregate more like rigid spheres, or do the subunits show a degree of structural flexibility during packaging in the heteroaggregate? These questions gain relevance in the light of the observations that ␣-crystallin as well as both homoaggregates of its individual subunits differ in their stability and the efficiency of their chaperone-like function (9). The exchange of subunits occurs efficiently only at temperatures close to 37°C, but not below 15°C (43, 44). We have used this temperature dependence of subunit exchange to control the extent of subunit exchange between homoaggregates of ␣Aand ␣B-crystallins by allowing exchange at 37°C and stopping it at 4°C. We hypothesized earlier that ␣B-crystallin, due to the higher chaperone-like activity and propensity for structural change, may modulate the chaperone-like activity of ␣-crystallin either directly or by inducing packing changes in the heteroaggregate (9). Structural perturbation of the individual subunits in the heteroaggregate may translate to quaternary structural alterations, which could modulate the chaperone-like activity of ␣-crystallin. Conversely, packing alterations of the subunits may reflect in tertiary structural changes within the subunits. We have tried to address these questions by studying the time-dependent structural and functional changes upon heteroaggregate formation by subunit exchange between homoaggregates.
If there are no specific interactions between the subunits in the ␣-crystallin heteroaggregate, then the near-UV CD spectrum of the heteroaggregate should be an algebraic sum of those of the ␣Aand ␣B-crystallin homoaggregates. In other words, the spectrum synthesized by the algebraic summation of the spectra of ␣Aand ␣B-crystallin homoaggregates would indicate if homologous interactions (␣A-␣A, ␣B-␣B), as seen in homoaggregates, are equivalent to heterologous interactions (␣A-␣B), as seen in the ␣-crystallin heteroaggregate. Fig. 1 shows the near-UV CD spectra of ␣Aand ␣B-crystallin homoaggregates and the ␣-crystallin heteroaggregate. Also shown are the near-UV CD spectra of a mixture of ␣Aand ␣Bcrystallin homoaggregates at a ratio of 3:1 (w/w) immediately upon mixing and after allowing for subunit exchange by incubation at 37°C for 24 h. The near-UV CD spectra of ␣Aand ␣B-crystallin homoaggregates show differences mainly in the 270 -290 nm region, where tryptophan and tyrosine contribute to the observed signal (Fig. 1A). The near-UV CD spectrum of a mixture of ␣Aand ␣B-crystallin homoaggregates (3:1, w/w) does not overlap with that of the ␣-crystallin heteroaggregate (Fig. 1B). This non-additive nature suggests alterations in the subunit structure upon incorporation into the heteroaggregate. The spectrum of the mixture (3:1 (w/w) ␣A/␣B-crystallin) bears a closer resemblance to that of the ␣A-crystallin homoaggregate (Fig. 1A). This was expected due to the predominance of ␣A-crystallin in the mixture. At this point in time, the proteins exist as homoaggregates, with little or no subunit exchange having taken place. A spectrum synthesized by the algebraic addition of the individual homoaggregate spectra ((3 ϫ ␣Acrystallin spectrum ϩ ␣B-crystallin spectrum)/4) overlaps with that of the mixture of homoaggregates (Fig. 1B), but not with that of ␣-crystallin (Fig. 1A). Incubating the mixture at 37°C allows for subunit exchange and generates heteroaggregates. The near-UV CD spectrum of the mixture following subunit exchange for 24 h is almost identical to that of the native ␣-crystallin heteroaggregate (Fig. 1B). The main difference in the near-UV CD spectra of the homoaggregates before and after subunit exchange is in the 270 -290 nm region. The near-UV CD spectra of the 3:1 (w/w) mixture of ␣Aand ␣Bcrystallins did not show any change when incubated at 4°C for 24 h (data not shown). In an earlier study, we investigated the temperature-dependent changes in the structure and function of ␣Aand ␣B-crystallin homoaggregates (9). We found that unlike ␣A-crystallin and contrary to the well known temperature stability of the ␣-crystallin heteroaggregate (57,58), ␣Bcrystallin loses considerable tertiary structure at 45°C. Interestingly, the maximum CD signal change observed was also in the 270 -290 nm region. To monitor the changes in tertiary structure as a function of subunit exchange, we recorded the near-UV CD spectra of the homoaggregates as a function of time of exchange at 37°C. Fig. 2 plots the change in ellipticity at 284 and 265 nm as a function of time as the subunit exchange proceeded. The change was much higher at 284 nm than at 265 nm. This is because, at 265 nm, the ellipticity value of the ␣-crystallin heteroaggregate is not different from the algebraic sum (3:1) of those of the homoaggregates. The observed change in ellipticity at 284 nm upon subunit exchange saturated by 2 h.
The effect of subunit exchange on the secondary structure of the subunits was monitored by far-UV CD (Fig. 3). The far-UV CD spectra of ␣Aand ␣B-crystallin homoaggregates show only minor differences from the spectrum of the ␣-crystallin heteroaggregate (9). The spectrum of the 3:1 (w/w) mixture of ␣Aand ␣B-crystallin homoaggregates is almost similar to that of ␣-crystallin (data not shown). Upon subunit exchange by incubating the mixture at 37°C, no significant changes were observed. Taken together, these results indicate that a change from homologous subunit packing (as in homoaggregates) to heterologous subunit packing (as in ␣-crystallin) alters the tertiary, but not the secondary, structures of ␣Aand ␣Bcrystallins. The secondary structure of ␣-crystallin is exceedingly stable against perturbation by temperature (57, 58) and, to a lesser extent, by chaotropic agents (56). The primary ␤-pleated secondary structure of HSP25 is stable up to a pH of 4.5 and at a neutral pH up to a temperature of 60°C (66). The tertiary structure of the ␣-crystallins is more labile and is lost at 60°C (9) and in 0.8 M guanidinium chloride (56). Using cryoelectron microscopy, Haley et al. (41) showed that the outer surfaces of ␣B-crystallin homoaggregates have an extremely variable and dynamic structure. The interactions involved in quaternary structure formation must be predominantly on the surface of the proteins. Increasing the temperature would increase the dynamics of the flexible elements on the surface. This could be a reason why the subunits of ␣-crystallin freely exchange at or above 37°C, but not at lower temperatures. Here, it is interesting to note that electron microscopy studies of two-dimensional crystals of GroEL showed that it has unusual plasticity (67). Perhaps such a high degree of flexibility is required for molecular chaperones to have a low specificity, enabling them to bind a variety of protein folding intermediates. This could constitute a general mechanism by which a few chaperones could interact with the folding intermediates of a large number of proteins.
To ascertain if the time course of subunit exchange correlates with the time course of the tertiary structural changes observed upon mixing homoaggregates, we used the difference in the pI values of the ␣Aand ␣B-crystallin homoaggregates as a tool to follow the course of subunit exchange. Fig. 4 shows the time course of heteroaggregate formation upon mixing ␣Aand ␣B-crystallin subunits at a ratio of 3:1 (w/w) as monitored by native isoelectric focusing. Homoaggregates of ␣A-crystallin focused to a pI range of 4.5-5.2, whereas those of ␣B-crystallin focused to a pI range of 6.4 -6.9. The broad pI ranges of the homoaggregates of ␣Aand ␣B-crystallins focused in the absence of urea indicate that the homoaggregates have charge microheterogeneity. Earlier studies on the native isoelectric focusing of ␣-crystallin and homoaggregates of its subunits also revealed a high extent of charge microheterogeneity (43). This may be caused by heterogeneity in the structure of the aggregates with respect to number, folding, or packing of subunits. Previous Monte Carlo simulation of the charge frequency distribution demonstrated that a random combination of subunits could explain the observed charge microheterogeneity of ␣-crystallin (68). Isoelectric focusing of ␣Aand ␣B-crystallin subunits in the presence of 7 M urea does not show such charge heterogeneity (69). It is interesting to note that the pI values observed for ␣Aand ␣B-crystallin homoaggregates are lower by ϳ0.8 -1.1 and 0.4 -0.7 units, respectively, as compared with their corresponding subunits in 7 M urea. This is indicative of the masking of positive charges, probably due to the folding or aggregation of the subunits. Upon incubation of a mixture of ␣Aand ␣B-crystallin homoaggregates at 37°C, the bands in the acidic region (␣A-crystallin homoaggregates) became more basic, whereas those in the basic region (␣B-crystallin homoaggregates) moved to a more acidic pI (Fig. 4). At a ratio of 3:1 (w/w) ␣A/␣B-crystallin, the pI of the resulting heteroaggregate is not very different from that ␣A-crystallin. The formation of heteroaggregates could, however, be followed by monitoring the shift of the ␣B-crystallin homoaggregates. As the incubation proceeded, more protein focused to regions of an intermediate pI.
Most of the change was completed in 2 h, and only minor alterations in the focusing pattern were observed when the incubation was extended to 24 h. A similar experiment with exchange of subunits between homoaggregates at a ratio of 1:1 (w/w) showed a similar trend (data not shown). From these experiments, we were able to conclude that subunit exchange is concomitant with the observed changes in tertiary structure upon mixing of homoaggregates. Polydispersity is one of the striking characteristics of the ␣-crystallin heteroaggregate. The average molecular mass of the ␣-crystallin heteroaggregate is 800 kDa, whereas those of the homoaggregates of its subunits, ␣Aand ␣B-crystallins, are 600 and 450 kDa, respectively. The molecular mass of the ␣-crystallin heteroaggregate is affected by isolation conditions such as pH, ionic strength, and temperature (37)(38)(39)(40). We used gel filtration chromatography to study the effect of subunit exchange on oligomer size and polydispersity. Fig. 5 shows the gel filtration profile of a 3:1 (w/w) mixture of homoaggregates (␣A/␣B-crystallin) on a water-jacketed Superose 6B column run at 4°C. A mixture of the homoaggregates, injected immediately after mixing, eluted as a broad peak because the ratio of ␣Ato ␣B-crystallin was 3:1, and the two peaks did not resolve fully. Incubation of this mixture at 37°C to facilitate subunit exchange caused a change in elution volumes. With increasing incubation time and the generation of heteroaggregates, the proteins eluted earlier, indicating an increase in oligomer size. Subunit exchange at 37°C for 4 h increased the size of the heteroaggregate considerably. Upon further incubation, only a minor change in subunit size was observed. The heteroaggregate obtained upon complete exchange after 24 h eluted as a homogeneous peak with a narrower size distribution as compared with a mixture of homoaggregates. Subunit exchange between homoaggregates of ␣Aand ␣B-crystallins generates a discreet population of oligomers. The resultant heteroaggregates generated upon subunit exchange are larger than the homoaggregates from which they are generated. A balance of charge and hydrophobic interactions would dictate the size of a soluble aggregate. As the sizes of the aggregates become smaller, the surface/volume ratio of aggregates increases, but the total internal volume (and therefore, burial of hydrophobic surface) decreases. The charges on the surface of the proteins are essential for its solubility, whereas the burial of hydrophobic surfaces is energetically favorable. In the cell cytoplasm, where the pH is close to 7 and ionic strength is ϳ150 mM, the ␣-crystallins bear a high effective charge of 50. Their high molecular mass is associated with a high excluded volume fraction equivalent to twice the dry cell volume fraction (70), suggesting an open, non-compact quaternary structure. The large excluded volume fraction and high charge combine to produce the strong repulsive behavior, responsible for lens transparency. Image-restrained analysis of cryoelectron microscopy data of ␣B-crystallin yielded a similar "open" structure with large voids between subunits (71). The highly charged C-terminal tail of ␣-crystallin is responsible for the solubility of the heteroaggregate. Siezen et al. (34) have shown that successive truncation of residues from the C-terminal tail of ␣A-crystallin produces successively larger aggregates. The charge on the ␣-crystallin heteroaggregate is important for the solubility of the ␣-crystallin-target protein complex. Trunca-tion of the C terminus of ␣-crystallin decreases its chaperonelike activity (73,74) and is believed to be one of the posttranslational modifications that compromise its molecular chaperone-like function in vivo. The homoaggregates of ␣Aand ␣B-crystallins differ with respect to their charge and hydrophobicity. Subunit exchange between homoaggregates to form heteroaggregates alters the balance between the charges and hydrophobicity. The alteration of surface charge in the heteroaggregate due to altering subunit composition, as during lens differentiation, could have implications for the functions of ␣-crystallin in both the maintenance of lens transparency and its chaperone-like activity. The formation of a larger heteroaggregate is probably a compensatory mechanism by which the burial of hydrophobic surfaces in a larger aggregate could lead to its stabilization. This might be a reason why in the lens, where both ␣Aand ␣B-crystallins are expressed, they are found to exist together as heteroaggregates of ␣-crystallin.
Since subunit exchange between homoaggregates leads to tertiary and quaternary structural changes, it was of relevance to check if these affect the functionality of the heteroaggregate. We studied the chaperone-like activity of the proteins as a function of subunit exchange. Fig. 6 shows the percentage protection offered against DTT-induced insulin aggregation as a function of time allowed for subunit exchange. Interestingly, the protective ability decreased with subunit exchange. It fell to ϳ50% in ϳ5-6 h of exchange. Further incubation at 37°C resulted in only a minor change in chaperone-like activity. The resulting chaperone-like activity observed upon complete subunit exchange is comparable to that of native ␣-crystallin at 37°C (9). In contrast, the homoaggregate mixture held at 4°C showed no change in the protective ability during the same period. The fall in the chaperone-like activity closely followed the time course of subunit exchange and the resulting changes in the tertiary structure of the subunits. It is important to note that the rate of subunit exchange was strongly dependent on the concentration of the interacting homoaggregates. Subunit exchange at a protein concentration of 0.1 mg/ml during the time period of the protection assay was not significant. Moreover, binding of denatured proteins to ␣-crystallin, as in this protection assay, decreases the rate of subunit exchange considerably (44). Previous studies have shown that changes in ␣-crystallin oligomer size at elevated temperatures are slow and require many hours to take effect (37,76), and tertiary structural alterations have been proposed to be responsible for the temperature-dependent enhancement of its chaperone-like activity (58). This result demonstrates that alteration in subunit packing leads to alteration in chaperone-like activity. The mechanism of the temperature-induced enhancement of the chaperone-like activity of ␣-crystallin can be attributable to alteration of subunit packing within the oligomer. Such changes in packaging could alter the relative accessibility of the individual subunits or lead to the generation/exposure of target protein-binding sites.
The subunits of ␣-crystallin are very prone to form aggregates. In fact, single subunits exist only in concentrated solutions of denaturing agents, and they re-aggregate upon dilution of these agents or after their removal by dialysis. The ratio of subunits is not crucial to form aggregates. Not only dissociated ␣-crystallin subunits (77)(78)(79)(80), but also isolated ␣Aand ␣Bcrystallin subunits re-associate (64,81,82) to homoaggregates. Moreover, it is possible to generate heteroaggregates of varying subunit composition by co-refolding ␣Aand ␣B-crystallins mixed at various ratios (64,65). These results suggest the equivalence of ␣Aand ␣B-crystallins in the ␣-crystallin heteroaggregate.
The ␣-crystallin in the lens epithelial cells has a subunit composition of 1:3 ␣A/␣B-crystallin; but upon differentiation to fiber cells, the ratio becomes 3:1 ␣A/␣B-crystallin (26 -30). The significance of this alteration of the composition of ␣-crystallin is not clearly understood. Subunit exchange between ␣Aand ␣B-crystallin homoaggregates mixed at various ratios also generates heteroaggregates of varying subunit composition (43-45). The composition of the ␣-crystallin heteroaggregate varies between species (60 -63) and during ontogenesis in a single species (72). Kangaroo ␣-crystallin has a subunit composition of 9:1 (w/w) ␣A/␣B-crystallin, whereas dogfish ␣-crystallin has a ratio of 1:4. Moreover, ␣B-crystallin expression is controlled by a heat shock promoter and is known to be overexpressed in a variety of stress (14 -19) and diseased (20 -25) conditions. In light of the differential structural and functional properties of ␣Aand ␣B-crystallins, we investigated the effect of subunit composition on the chaperone-like activity of heteroaggregates. We generated heteroaggregates of varying subunit composition (5:1, 3:1, 1:1, 1:3, and 1:5 (w/w) ␣A/␣B-crystallin) by allowing subunit exchange between homoaggregates, mixed at the corresponding ratios, to proceed for 24 h at 37°C. Fig. 7 shows the native isoelectric focusing pattern of the resulting heteroaggregates. The native pI of the heteroaggregates is dictated by their composition of acidic and basic subunits. The heteroaggregates also show charge microheterogeneity, such as that observed for the individual homoaggregates. The heteroaggregates with subunit compositions of 5:1 and 1:5 (w/w) ␣A/␣B-crystallin focused at positions comparable to those of the ␣Aand ␣Bcrystallin homoaggregates, respectively. The change in pI with respect to subunit composition was more significant between the ratios of 3:1 and 1:3 (w/w) ␣A/␣B-crystallin.
We performed gel filtration chromatography to study the effect of subunit composition on the molecular mass of the heteroaggregates (Fig. 8). The molecular mass of the heteroaggregate was dependent on the ratio of ␣A to ␣B-crystallin subunits. There was a sudden decrease in the estimated molecular mass of the heteroaggregate when the amount of ␣Bcrystallin subunits predominated in the heteroaggregate. Thomson and Augusteyn (64) reported that, upon co-refolding mixtures of ␣Aand ␣B-crystallin subunits at various ratios, a sharp increase in the sedimentation coefficient of the resulting heteroaggregate was observed once ␣B-crystallin predominated in the heteroaggregate. Interestingly, both studies show an alteration in the molecular mass at the same subunit stoichiometry. The differences may be due to different mechanisms involved in heteroaggregate formation by co-refolding of subunits as compared with by the exchange of prefolded subunits in homoaggregates.
One important difference between the chaperone-like activities of ␣Aand ␣B-crystallin homoaggregates is the temperature dependence of their protective abilities (9). At elevated temperatures (ϳ63°C), the protective abilities of the two proteins are comparable, although at physiological temperatures, ␣B-crystallin displays higher activity. Even below physiological temperatures, ␣B-crystallin shows significant protective ability, whereas ␣Aand ␣-crystallins do not. The sensitivity of ␣B-crystallin structure to temperature-induced perturbation may be responsible for these properties. We investigated the effect of altering subunit composition on the temperature-de- pendent chaperone-like activity of the heteroaggregate. We used the DTT-induced aggregation of insulin as a model of non-thermal aggregation to study the chaperone-like activity of the heteroaggregates at various temperatures. Fig. 9 shows the abilities of the various heteroaggregates to prevent the DTTinduced aggregation of insulin as a function of temperature. At a protein concentration of 0.1 mg/ml heteroaggregate and 0.2 mg/ml insulin, all the ratios protected at 40°C. The proteins offered lower protection as the temperature was reduced. The extent of the decrease in protective ability as the temperature was reduced, however, was dependent on the subunit composition of the heteroaggregates. Increasing the content of ␣Bcrystallin in the heteroaggregate increased the effectiveness of the heteroaggregate at lower temperatures. This result demonstrates that incorporation of ␣B-crystallin into a heteroaggregate could provide a means of increasing the chaperone-like activity of the heteroaggregate. This is of relevance since, as mentioned earlier, ␣B-crystallin synthesis is induced under stress and disease conditions. Interestingly, HSP27 is also expressed in the eye lens (75).
Our studies demonstrate that packing of subunits in the variable oligomer assembly provides a mechanism for the modulation of chaperone-like activity. The accompanying tertiary structural changes, upon subunit exchange, due to specific interactions between the subunits may be responsible for this modulation. The dynamic nature of the ␣-crystallin oligomer may provide a mechanism to increase its efficiency by incorporation of ␣B-crystallin during stress conditions. Due to their homology to ␣-crystallin and their ability to exchange subunits and to form mixed oligomers, mammalian small HSPs might use a similar form of activity modulation.