Age-related cataract is a result of crystallins, the predominant lens proteins, forming light-scattering aggregates. In the low protein turnover environment of the eye lens, the crystallins are susceptible to modifications that can reduce stability, increasing the probability of unfolding and aggregation events occurring. It is hypothesized that the α-crystallin molecular chaperone system recognizes and binds these proteins before they can form the light-scattering centres that result in cataract, thus maintaining the long-term transparency of the lens. In the present study, we investigated the unfolding and aggregation of (wild-type) human and calf βB2-crystallins and the formation of a complex between α-crystallin and βB2-crystallins under destabilizing conditions. Human and calf βB2-crystallin unfold through a structurally similar pathway, but the increased stability of the C-terminal domain of human βB2-crystallin relative to calf βB2-crystallin results in the increased population of a partially folded intermediate during unfolding. This intermediate is aggregation-prone and prevents constructive refolding of human βB2-crystallin, while calf βB2-crystallin can refold with high efficiency. α-Crystallin can effectively chaperone both human and calf βB2-crystallins from thermal aggregation, although chaperone-bound βB2-crystallins are unable to refold once returned to native conditions. Ordered secondary structure is seen to increase in α-crystallin with elevated temperatures up to 60 °C; structure is rapidly lost at temperatures of 70 °C and above. Our experimental results combined with previously reported observations of α-crystallin quaternary structure have led us to propose a structural model of how activated α-crystallin chaperones unfolded βB2-crystallin.
- circular-dichroism (CD)
The transparency and refractive power of the vertebrate eye lens is achieved by the high concentration and regular spatial distribution of crystallin proteins [1,2]. In the low protein turnover environment of the lens, transparency is maintained by the intrinsic stability, solubility and longevity of these crystallins; disturbances in lens transparency inevitably result in cataract .
The three classes of crystallins, α, β and γ, are members of two large gene families: the α-crystallins, which are members of the widely distributed small heat-shock protein family and function as both molecular chaperones and structural proteins , and the more lens-specific structural βγ-crystallins . Lens optical quality and refractive power stems from a gradient of protein concentration along the optical axis, formed from differing proportions of these crystallin classes. Relative amounts of crystallin present in young human lenses are approx. 28% (α), 43% (β) and 28% (γ) [6–8]. The core of the lens is enriched in γ-crystallins, with the predominantly expressed being γC, γD and γS in humans [6,8]. Mammalian β-crystallins can be divided into acidic (βA1, βA2, βA3 and βA4) and basic (βB1, βB2 and βB3) forms; βB2 is the predominant β-crystallin in the human lens, contributing 14–20% of total lens protein [7,8]. Lens α-crystallin is formed from two polypeptide chains, αA- and αB-crystallins, which have sequence identity of approx. 56%; ratios of αA- to αB-crystallins in the human lens vary with age. Both α- and βγ-crystallins undergo extensive post-translational modifications, with deamidation appearing to correlate with loss of solubility .
α-Crystallins are present in the lens as polydisperse hetero-oligomers of αA- and αB-crystallins. Tardieu and colleagues [10,11] reported that native calf lens α-crystallin assemblies consist of 40–45 subunits, with an external dimension of 170 Å (1 Å=0.1 nm), while others have reported a range of assembly sizes . This diversity, as well as diversity arising from post-translational modifications, has so far prevented crystallization. Homologous small heat-shock proteins from archaea, wheat and tapeworm have revealed the structure of the conserved α-crystallin domain, but in each instance variation in modes of association results in substantial differences in oligomeric assembly [13–15]. The chaperone function for plant small heat-shock proteins is thought to be driven by the temperature-regulated increase in oligomer dissociation, resulting in the population of a sub-assembly species (possibly a dimer) that provides the binding sites for non-native substrates [14,16,17]. Similarly, dissociation of mammalian HSP27 (heat-shock protein 27) oligomers to smaller multimeric species is correlated with binding competency: the interaction between HSP27 and destabilized T4 lysozyme has been used to propose a model in which the folding equilibrium of T4 is coupled with the oligomeric equilibrium of HSP27 . This coupling offers a novel explanation for certain congenital cataracts, allowing gain-of-function mutations in αA-crystallin to promote unfolding of substrate proteins . The active form of α-crystallin remains undetermined. Lens α-crystallin subunit exchange is extremely rapid and has been shown to increase with elevated temperature [11,20]. It has been proposed that transient dissociation of the oligomeric form to yield free subunits, as a result of phosphorylation or an increase in temperature, is associated with the activation of binding or chaperone activity [21,22]. Contrastingly, in yeast small heat-shock protein Hsp26 dissociation of the oligomer is not required for chaperone activity . It has been suggested that the structural changes of α-crystallin in response to increased temperature support a possible ‘activation’ step in the chaperone mechanism .
Monomeric γ-crystallins and oligomeric β-crystallins both consist of two domains, with each domain composed of two intercalated Greek key motifs . These domains pair intra-molecularly in γ-crystallins, via a V-shaped linker peptide, while in the case of βB2-crystallins the linker is extended and domains pair inter-molecularly, promoting oligomeric assembly . However, in truncated βB1-crystallin, the domains pair intra-molecularly, like γ-crystallins, with this oligomer using a new interface for assembly [26,27]. Like the α-crystallins, the oligomeric β-crystallins engage in subunit exchange, creating diversity .
Long-term stability is essential for crystallins to remain folded, thus minimizing aggregation . βB1-crystallin and especially γ-crystallins typically exhibit high thermodynamic stability and, sometimes, kinetic stabilization to prevent protein unfolding [29–31]. However, aging crystallins are susceptible to modification that can lower stability and/or solubility and increase the likelihood of aggregation, leading to cataract as the binding capacity of α-crystallin is exceeded [32,33]. Recent studies have sought to address the structural mechanisms of βγ-crystallin unfolding and aggregation, and α-crystallin chaperone activity. In the case of human γD-crystallin, it has been shown to unfold by a three-state mechanism at pH 7 in which the N-terminal domain unfolds first, followed by the C-terminal domain. During refolding, when the C-terminal domain folds first and acts as a nucleating centre for folding of the N-terminal domain, an aggregation-prone intermediate occurs [30,34,35]. Similarly, dimeric rat βB2-crystallin at low concentrations unfolds by a three-state model, in which the N-terminal domain unfolds and the dimer dissociates, resulting in a monomeric partially folded intermediate that unfolds in a second transition; at higher protein concentrations, the unfolding appears to be two-state . In the presence of chemical denaturants, α-crystallins have lower conformational stability than γ- and βB1-crystallins , but broadly similar to βB2-crystallin and acidic β-crystallins . McHaourab and co-workers  have studied the association between destabilized mouse βB2 mutants and αA/αB-crystallin, and have suggested that population of a folding intermediate, rather than a specific energy threshold alone, can trigger chaperone activity.
In the present study, we have examined the stability of partially folded intermediate states of βB2-crystallins. Two orthologues of βB2-crystallin, human and calf, have been studied because of differences in their thermal aggregation properties; calf βB2 unfolds reversibly, with the unfolded form remaining soluble at high temperatures , while human βB2 becomes less soluble on unfolding . These proteins have 97% sequence identity, differing by one residue in the N- and C-terminal extensions, and four residues in the C-terminal domain (Figures 1a and 1b). The present study examines how these crystallins differentially unfold and aggregate and the structural changes that occur when they interact with calf lens α-crystallin following thermal challenge.
Recombinant human βB2-crystallin  and calf lens βB2- and calf lens α-crystallins  were supplied by Dr O. Bateman (Birkbeck College). Calf lens α-crystallin was purified, as described in , by gel filtration on Sephacryl S-300HR, taking care to exclude the high-molecular-mass shoulder of α-crystallin. Bovine α-crystallin has been shown to consist of αA- and αB-crystallin at an approximate ratio of 3:1 . Samples were buffer-exchanged into 10 mM sodium phosphate buffer (pH 7.0), by using Amicon Microcon 10 microconcentrator vessels, and frozen into individual aliquots.
Fluorimetry and light scattering
Fluorescence emission spectra were collected using a thermostatically controlled Hitachi F-2500 spectrometer, calibrated using a rhodamine B standard. Tryptophan residues were excited at 280 nm. Emission spectra were collected from 300 to 450 nm at a scan speed of 600 nm/min. Excitation and emission slit widths were set to 10 nm, except where stated differently in the text.
Light scattering was measured with λexcitation and λemission set to 360 nm, and slit widths set to 10 nm, unless otherwise noted. Samples were at 0.2 mg/m concentration, in a 1-cm-path-length quartz cuvette (Hellma, Müllheim, Germany). Baseline data of buffer alone were collected under identical parameters and subtracted. Samples for fluorescence and light-scattering experiments were in 10 mM sodium phosphate buffer (pH 7.0).
CD measurements and analyses
CD spectra were collected at 0.2 nm steps from 280 to 178 nm, using an Aviv 62DS (Aviv Biomedical) instrument calibrated using 99% pure CSA (camphor-10-sulfonic acid) (Sigma–Aldrich) . Three scans of each sample and the equivalent baseline were collected. Sample spectra were averaged across the three repeats, and averaged baselines of buffer (10 mM sodium phosphate buffer, pH 7.0) or urea were subtracted. Sample concentrations were 0.2 mg/ml, determined by absorbance at 280 nm, and the cell path length was 0.5 mm.
The amount of protein lost from a solution of βB2-crystallin by aggregation was measured by heating 0.2 mg/ml aliquots in 10 mM sodium phosphate buffer (pH 7.0) to increasing temperatures for 10 min, removing the heated sample, cooling on ice for 5 min and centrifuging at 14000 g for 5 min in a bench-top microfuge. The UV absorbance at 280 nm of the supernatant was measured before heating and after incubation at each temperature.
Thermal denaturation experiments were performed as follows: samples at a protein concentration of 0.2 mg/ml, in 10 mM sodium phosphate (pH 7.0) were incubated for 10 min at a given temperature before CD measurements were obtained at a single wavelength (195 nm). Refolded samples were first unfolded at the elevated temperature for 10 min and then refolded for 15 min at 20 °C. In each experiment, fresh samples were used for each temperature step.
Chemical denaturation was performed as follows: stock solutions of 7 M urea (>99.5% purity; Sigma–Aldrich) were prepared in 10 mM sodium phosphate buffer (pH 7.0) and the exact urea concentration was determined by measuring the refractive index using an Abbe Mark II refractometer (Reichert). βB2-crystallin samples were incubated at 20 °C for 16 h in urea solutions of increasing concentration, at 0.2 mg/ml. Samples were centrifuged for 5 min in a bench-top microfuge prior to data collection. CD measurements were obtained at a single wavelength (222 nm, because the absorbance of the urea prevented measurements being obtained at 195 nm).
Preparation of the crystallin and chaperone complexes
Samples of α-crystallin and βB2-crystallin were incubated together at 4 °C, with each component at a concentration of 0.2 mg/ml (equivalent to molar concentrations of 10.1 and 8.6 μM for α-crystallin and βB2-crystallin respectively, based on the molecular mass of individual polypeptide chains; a 1:1.17 excess of α-crystallin), for 48 h prior to experiments. The buffer for these experiments was 10 mM sodium phosphate (pH 7.0). Solutions of α-crystallin and βB2-crystallin were incubated at 60 °C in a circulating water bath, with separate control samples of α-crystallin, and human and bovine βB2-crystallins. Samples were removed from incubation after 16 h, cooled on ice for 5 min, centrifuged at 14000 g for 5 min in a bench-top microfuge, loaded into demountable quartz cuvettes of appropriate path length (Hellma), and equilibrated for 10 min at 20 °C in the CD instrument prior to data collection.
Unfolding and refolding of βB2-crystallin
The studies of Bateman et al.  indicated that calf and human βB2-crystallins exhibit different solubilities at elevated temperature, with human βB2 becoming insoluble at higher temperatures. The aggregation of human and calf βB2 was investigated to identify the structure/stability basis for these observations.
Aliquots of βB2-crystallin were incubated at increasing temperature, cooled and centrifuged to remove aggregated material and then UV absorbance was measured to monitor changes in concentration due to heating. Calf βB2 remains soluble following incubation at temperatures up to 90 °C, while human βB2-crystallin aggregates extensively above 50 °C (Figure 2a). Non-centrifuged samples of human βB2 exhibit intense light scattering above 50 °C, while scattering from calf βB2 is minimal (Figure 2b). These observations confirm those of Bateman et al.  and reflect substantially different behaviour of the two βB2 orthologues.
βB2 unfolding during thermal and chemical denaturation was measured using CD. Calf βB2-crystallin unfolds in an apparent two-state process with a Tm of approx. 62 °C, as indicated by the sigmoid unfolding curve of Figure 3(a) and isodichroic point in Figure 4(a). Urea-induced unfolding occurs with a C1/2, urea (concentration of urea when the protein is half unfolded) of approx. 1.6 M (Figure 3b). Conversely, human βB2 unfolds through two distinct transitions as monitored by the onset of light scattering and aggregation (Figure 3a). Tm values for the two transitions are 64 and 78 °C respectively, and the mid-point of the inter-phase plateau is 74 °C (Figure 3a). Urea denaturation confirms the biphasic unfolding of human βB2, with approximated C1/2, urea values for the transitions of 2.2 and 3.4 M respectively (Figure 3b). Both thermal and chemical denaturation curves show the intermediate as ∼33% folded. As the relative stability of human βB2 exceeds that of calf, it appears that the less stable protein is in fact more soluble at higher temperature, from which it can be inferred that the unfolded state exhibits greater solubility than the partially folded state.
Refolding of βB2-crystallins was assessed by recording CD spectra of proteins at elevated temperatures and then refolding from each of these temperature steps at 20 °C. Spectra of refolded calf βB2 samples are similar to those of unheated βB2 at 20 °C, confirming a return to the folded state (Figure 4b). Differences in CD magnitude are due to a small loss of material during the heating, cooling and data collection process, indicating that an aggregation-prone earlier pathway exists alongside the productive refolding pathway. During unfolding, CD spectra of human βB2 are very similar to their fully folded and fully unfolded calf counterparts; however, an isodichroic point is not observed during human βB2 unfolding (Figure 4c). Above 60 °C, human βB2 does not fully refold to the native structure, appearing partially folded (Figure 4d). Refolded spectra show increased absorbance due to light scattering, indicating that the spectra may include contributions from insoluble aggregate material as well as soluble protein. Previous biophysical characterization of human βB2-crystallin has shown no change in molecular mass upon refolding from denaturing concentrations of urea .
At concentrations used in the present study, calf βB2 unfolds in a two-state manner (N2↔2U) similar to that proposed for higher concentrations of rat βB2-crystallin ; this cannot rule out a sparsely populated monomeric intermediate state. Human βB2 unfolds according to the three-state model proposed by Fu and Liang , but as unfolding is not fully reversible the unfolding process is better described as (N2→2I→2U). In the present study, the human βB2 intermediate appears ∼33% folded, as defined by the magnitudes of the CD signals at 195 and 222 nm, in agreement with published fluorimetry data [44,45], suggesting that the intermediate (monomeric with unfolded N-terminal domain) may also exhibit loss of structure within the predominantly folded C-terminal domain. Population of this intermediate marks the divergence of an aggregation-prone pathway, similar to the proposed mechanism for the aggregation of human γD-crystallin during refolding .
As studies of rat βB2 mutants showed that removal of the extensions does not alter the stability or unfolding profiles of the protein , it is unlikely that sequence differences in these could contribute to the increased stability of human βB2. Comparison of human  and bovine (PDB accession number 2BB2 ) βB2-crystallin structures reveals no differences in hydrogen-bonding at sites of sequence variation other than one additional hydrogen bond between Asp172 and Ser174 of human βB2. The increased stability of human βB2 may therefore be due to stabilization of the loop between strands C4 and D4.
CD and tryptophan fluorimetry were used to study aggregated material, before and after centrifugation, obtained from unfolded samples of human βB2. The spectrum of the aggregate, determined as the difference between samples before and after centrifugation, is consistent with the aggregate having a high content of β-sheet, and appears more structured than the soluble, partially refolded, βB2 (Figure 5). Fluorimetry λmax emission measurements were collected from native (328 nm), unfolded (347 nm) and refolded (338 nm pre/post-centrifugation) human βB2 samples, suggesting partial burying of tryptophan residues in both the refolded and aggregated states (results not shown). The failure of the soluble material to fully refold may indicate that it is trapped in a low-energy non-native state.
Stability and secondary structure of α-crystallin
Chaperone assays frequently rely on temperature to promote substrate protein aggregation [4,47] and to activate α-crystallin . Accordingly, identifying the maximum temperature at which α-crystallin remains folded is essential for such assays. In this instance, the unfolding temperature was determined using CD (Figure 6a). At 20 °C, our preparation of α-crystallin is estimated to be only 4% α-helical (Table 1) and shows only a minor increase during higher temperature incubations. A small increase in the CD signal at 200 nm was observed at temperatures up to 60 °C; above this temperature, the CD signal at 200 nm shifts to lower wavelengths and decreases in magnitude, indicative of a loss in ordered structure. Deconvolution estimates suggest an increase in unordered conformation above 60 °C, at the expense of β-sheet and turn conformations. The tertiary/quaternary structure of α-crystallin has been extensively characterized over a similar temperature range and has been shown to double rapidly and irreversibly in assembly size between 60 and 69 °C , corresponding to the ‘activated’ form, while precipitating once above 70 °C . Considered in conjunction with the CD data, this suggests that increases in ordered α-crystallin secondary structure may correlate with assembly into the higher-molecular-mass activated complex. The rapid increase in unordered secondary structure above 60 °C is indicative of the aggregating state and is unlikely to relate to its chaperone mechanism. α-Crystallin appears stable below this temperature; after a 16 h incubation at 60 °C, the CD spectrum is unaltered except for an increased 193 nm signal, due to the presence of a small amount of the lower wavelength peak, and scattering from the solution did not increase during incubation (Figure 6b). The increase in CD signal at 193 nm is irreversible upon cooling to 20 °C and is indicative of structural changes associated with the previously proposed concept of a chaperone activation mechanism [11,49].
Activity of α-crystallin
α-Crystallin was shown to be functional by its ability to suppress light scattering from a 1:1.17 molar ratio (based on monomer size) solution of α-crystallin and βB2 at 60 °C and above (Figure 7a). At 60 °C, unchaperoned human βB2 begins to scatter light after ∼450 s and reaches a plateau at ∼3300 s. When incubated with α-crystallin, light scattering remains constant and low over a 1 h period (Figure 7b).
Conformation of βB2-crystallins bound to α-crystallin
The conformation of βB2-crystallin bound to α-crystallin was investigated using CD spectroscopy. Incubation was extended to 16 h to promote increased βB2-crystallin association with α-crystallin without exceeding 60 °C, and was found to promote appreciable binding of both human and calf βB2 to α-crystallin. Solutions were cooled and centrifuged, and CD spectra were recorded to examine the complex under conditions that favour refolding. Spectra of solutions of βB2 and α-crystallin (βB2-α) prior to heating are identical with composite spectra of α-crystallin and βB2-crystallin alone; there is no spectral indication of secondary structure change due to interaction. Light scattering did not increase during heat incubation, confirming that the complex of chaperone bound to substrate protein remained soluble. Spectra of chaperoned human and calf βB2-crystallins were obtained by subtracting the post-incubation spectrum of α-crystallin from the post-incubation spectra of human and calf βB2-α solutions (Figure 8), and compared with the spectra of unchaperoned βB2-crystallins after equivalent incubations. Unchaperoned human βB2 and calf βB2 are not stable during this incubation; spectra of both appear only partially folded and of lower concentration following incubation. Chaperoned human and calf βB2 are retained in solution to a higher concentration, but their CD spectra resemble those of partially refolded human βB2-crystallin, rather than the more structured CD spectra of aggregate material shown in Figure 5. The failure of chaperoned calf βB2 to refold under conditions that have been shown to favour full refolding following shorter incubation lends support to the idea that the aggregation-prone pathway involves a non-native low-energy intermediate state, which is recognized and bound by α-crystallin prior to aggregation. Binding of α-crystallin to states that are aggregation-prone, but not yet precipitating, has also been described for α-lactalbumin, in which the chaperone engages in kinetic competition with the aggregation pathway .
A model for βB2-crystallin unfolding and chaperone association
By performing structural studies of both substrate protein (βB2) and α-crystallin, it is possible to unite models for βB2-crystallin unfolding [36,44] and α-crystallin chaperone activity [11,37,49]. Our observations lead us to conclude that calf and human βB2 share broadly similar unfolding pathways, but the overall reversibility of calf βB2 unfolding is due to sparse population of the partially unfolded intermediate state. It is interesting that these experiments do not conform to the assumption that destabilization enhances aggregation. If it is assumed that the additional stability of human βB2 derives from its C-terminal domain, then the stability difference between the N and C domains becomes more marked. Thus, under conditions when the N-terminal domain unfolds, human βB2 is more likely to have a folded C-terminal domain with an interaction interface exposed that can act as an aggregation-prone intermediate.
Our studies of the secondary structure of α-crystallin show changes occurring at temperatures similar to those associated with tertiary/quaternary structure changes, in which increases in molecular mass are reported at 60 and 66 °C [11,48]. This leads us to conclude that upon thermal activation, α-crystallin undergoes an irreversible increase in ordered secondary structure, associated with assembly into the reported high-molecular-mass (80-mer) form. It is this form that associates with partially folded βB2 intermediates prior to aggregation in our experiments (Figure 9). This finding is in agreement with a model in which population of a folding intermediate, rather than absolute energetic thresholds, can promote chaperone association . In the case of human and calf βB2-crystallins, the intermediate states recognized by α-crystallin are unable to correctly refold and, if not bound by α-crystallin, will form insoluble aggregates. This mechanism has consequences for in vivo cataract formation: substrate protein sequestered by α-crystallin must remain associated to avoid aggregation. This reduces the binding capacity of the α-crystallin present in the lens, thus eventually overwhelming the chaperone system.
We thank Dr Orval Bateman and Dr Myron Smith (Birkbeck College) for generously supplying protein samples and making available unpublished crystal structures. This work was supported by BBSRC (Biotechnology and Biological Sciences Research Council) grants to B. A. W. C. S. acknowledges the financial support of the MRC. P. E. was supported by a BBSRC studentship.
Abbreviations: HSP, heat-shock protein
- © The Authors Journal compilation © 2008 Biochemical Society