The R116C Mutation in αA-crystallin Diminishes Its Protective Ability against Stress-induced Lens Epithelial Cell Apoptosis*

αA-crystallin is a small heat-shock protein expressed preferentially in the lens and is detected during the early stages of lens development. Recent work indicates that the expression of αA-crystallin enhances lens epithelial cell growth and resistance to stress conditions. Mutation of the arginine 116 residue to cysteine (R116C) in αA-crystallin has been associated with congenital cataracts in humans. However, the physiological consequences of this mutation have not been analyzed in lens epithelial cells. In the present study, we expressed wild type or R116C αA-crystallin in the human lens epithelial cell line HLE B-3. Immunofluorescence and confocal microscopy indicated that both wild type and R116C αA-crystallin were distributed mainly in the cytoplasm of lens epithelial cells. Size-exclusion chromatography indicated that the size of the αA-crystallin aggregate in lens epithelial cells increased from 500 to 600 kDa for the wild type protein to >2 MDa in the R116C mutant. When cells were exposed to physiological levels of UVA radiation, wild type αA-crystallin protected cells from apoptotic death as shown by annexin labeling and flow cytometric analysis, whereas the R116C mutant had a 4- to 10-fold lower protective ability. UVA-irradiated cells expressing the wild type protein had very low TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling) staining, whereas cells expressing R116C mutant had a high level of TUNEL staining. F-actin was protected in UVA-treated cells expressing the wild type αA-crystallin but was either clumped around the apoptotic cells or was absent in apoptotic cells in cultures expressing the R116C mutant. Structural changes caused by the R116C mutation could be responsible for the reduced ability of the mutant to protect cells from stress. Our study shows that comparing the stress-induced apoptotic cell death is an effective way to compare the protective abilities of wild type and mutant αA-crystallin. We propose that the diminished protective ability of the R116C mutant in lens epithelial cells may contribute to the pathogenesis of cataract.

␣A-crystallin is a small heat-shock protein expressed preferentially in the lens and is detected during the early stages of lens development. Recent work indicates that the expression of ␣A-crystallin enhances lens epithelial cell growth and resistance to stress conditions. Mutation of the arginine 116 residue to cysteine (R116C) in ␣A-crystallin has been associated with congenital cataracts in humans. However, the physiological consequences of this mutation have not been analyzed in lens epithelial cells. In the present study, we expressed wild type or R116C ␣A-crystallin in the human lens epithelial cell line HLE B-3. Immunofluorescence and confocal microscopy indicated that both wild type and R116C ␣Acrystallin were distributed mainly in the cytoplasm of lens epithelial cells. Size-exclusion chromatography indicated that the size of the ␣A-crystallin aggregate in lens epithelial cells increased from 500 to 600 kDa for the wild type protein to >2 MDa in the R116C mutant. When cells were exposed to physiological levels of UVA radiation, wild type ␣A-crystallin protected cells from apoptotic death as shown by annexin labeling and flow cytometric analysis, whereas the R116C mutant had a 4to 10-fold lower protective ability. UVA-irradiated cells expressing the wild type protein had very low TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling) staining, whereas cells expressing R116C mutant had a high level of TUNEL staining. Factin was protected in UVA-treated cells expressing the wild type ␣A-crystallin but was either clumped around the apoptotic cells or was absent in apoptotic cells in cultures expressing the R116C mutant. Structural changes caused by the R116C mutation could be responsible for the reduced ability of the mutant to protect cells from stress. Our study shows that comparing the stress-induced apoptotic cell death is an effective way to compare the protective abilities of wild type and mutant ␣A-crystallin. We propose that the diminished protective ability of the R116C mutant in lens epithelial cells may contribute to the pathogenesis of cataract.
␣-Crystallin is a major lens protein and belongs to the small heat-shock protein family of molecular chaperones. It is a polymeric complex of two ϳ20-kDa subunits, ␣A and ␣B, that share a significant sequence identity, and can be isolated from lens fiber cells as a heteroaggregate of ␣A and ␣B-polypeptides in a 3:1 ratio with an average oligomeric mass of 600 -800 kDa. The protein has a dynamic polymeric structure with a hollow interior and undergoes inter-aggregate subunit exchange (1). The expression of ␣-crystallin in a variety of tissues outside the lens and its ability to prevent thermal aggregation of proteins in a manner similar to molecular chaperones suggest that it has general cellular functions over and above its role in light refraction (1,2).
To understand the physiological functions of ␣Aand ␣Bcrystallin, knockout mice have been generated (3,4). Examination of lenses from these mice suggests that ␣Aand ␣Bcrystallin may have dissimilar and highly specific cellular functions. Disruption of the ␣A-crystallin gene causes earlyonset cataract in mice, whereas disruption of the ␣B-crystallin gene does not result in altered lens morphology or transparency. Lens epithelial cells derived from mice lacking ␣A-crystallin have a 50% slower growth rate in vitro, whereas lens epithelial cells derived from ␣B-crystallin knockout mice demonstrate hyperproliferation and genomic instability (5,6). These findings have led to the suggestion that ␣Aand ␣Bcrystallin may have distinct functions in regulating cellular growth.
Other studies indicate that the expression of ␣Aor ␣Bcrystallin enhances the resistance of cells to a variety of stress conditions. ␣Aand ␣B-crystallin have the ability to protect cells from apoptosis induced by cytokines, protein kinase C inhibitor staurosporine, and UVA 1 radiation (7-10). The ability of ␣-crystallin to prevent apoptosis may be linked to an inhibition of caspase activity (11). Additional studies have shown that intermediate filaments are the physiological targets of small heat-shock proteins (12). ␣B-crystallin and HSP27 associate with intermediate filaments thereby preventing their undesired aggregation (13). In lens fiber cells exposed to thermal stress, ␣B-crystallin was shown to associate with lens-specific intermediate filament proteins phakinin and filensin (14). ␣Bcrystallin has been shown to associate with actin in ischemic cardiac cells where it may prevent aggregation of filaments (15).
Molecular chaperones prevent undesired protein aggregation by binding to non-native intermediates that may arise in response to cellular stress or during protein translation in vivo. It has been shown that proteins subjected to heat treatment in vitro in the presence of ␣-crystallin retain a high level of native secondary structure (16). Although the mechanism of chaperone action is not completely understood, it involves the formation of a complex between ␣-crystallin and substrate proteins (17)(18)(19). A binding site for substrate proteins has been identified in the sequences of both ␣Aand ␣B-crystallins (20,21). Recent work suggests that the protein structure of ␣B-crystallin is stabilized by complex formation with ␣A-crystallin (22).
Small heat-shock protein sequences contain a highly conserved amino acid sequence called the ␣-crystallin domain, and the importance of particular residues within that domain has been demonstrated recently by evidence that mutations in ␣Acrystallin and ␣B-crystallin are responsible for two genetic disorders. In ␣A-crystallin, substitution of arginine 116 by cysteine was found to be the cause of one form of autosomal dominant cataract (23). In ␣B-crystallin, substitution of arginine 120 by glycine was found to be the cause of another autosomal dominant disease, desmin-related myopathy, as well as cataracts (24). The recombinant mutant proteins have been purified from bacterial expression systems and are characterized by higher oligomeric mass than the wild type proteins, as well as altered secondary and tertiary structures and significantly reduced chaperone-like activity (25)(26)(27).
Lens epithelial cells in the germinative zone near the equator divide and elongate throughout life and are crucial for maintaining proper growth, transparency, and refractive properties of the lens. As new cells elongate to form lens fibers, they wrap around the lens periphery and meet at the sutures (28). ␣Aand ␣B-crystallin transcripts have been detected at early stages of lens development, and a marked increase in expression of the proteins accompanies differentiation (29,30), but the specific roles of ␣Aand ␣B-crystallin in lens epithelial cells have not been identified.
Recent work in our laboratory demonstrates that UVA radiation, which is known to cause cell death by production of reactive oxygen species and induces cataract in animals, induces lens epithelial cell death and that the expression of ␣A-crystallin prevents cell death (5,10). Because the R116C mutation in ␣A-crystallin results in a form of congenital human cataract, it is important to investigate the function of the mutant protein in lens epithelial cells. Furthermore, to understand the physiological function of ␣A-crystallin, it is essential to identify proteins that ␣A-crystallin associates with in lens epithelial cells.
In the present study, we expressed wild type and R116C ␣A-crystallin in human lens epithelial cell line HLE B-3 to examine the effect of expression of the mutant protein on UVA radiation-induced apoptotic cell death. We show that expression of R116C␣A had a 4-to 10-fold lower ability to protect cells than the wild type protein. Our work also shows that comparing the in vitro stress-induced apoptotic cell death is an effective way to measure the protective abilities of the wild type and mutant ␣A-crystallin.

MATERIALS AND METHODS
Vectors and Mutagenesis-Sequences encoding human wild type ␣Acrystallin were cloned into pCIneo mammalian expression vector (Promega) and transfected into human lens epithelial cells (HLE B-3) as described previously (5,10). Mutagenesis of wild type ␣A to R116C was performed using the QuikChange mutagenesis kit according to the manufacturer's recommendations (Stratagene). Each construct was confirmed by sequencing on an ABI 310 genetic analyzer using dRhodamine dideoxy termination (PerkinElmer Life Sciences).
Cell Culture and Transfection-Human lens epithelial cells with extended life span (HLE B-3) were cultured in 20% fetal bovine serum containing Eagle's minimum essential medium and gentamicin (50 g/ml) as described previously (31). These cells were passaged through at least 11 passages, after which they cease to produce ␣A-crystallin, as determined by immunoblot analysis (32). Wild type or R116C ␣Acrystallin cDNA can thus be reintroduced into these cells by cDNA transfection, and stable cell lines expressing different levels of ␣Acrystallin can be generated. Transfection was carried out using the Promega ProFection kit according to the manufacturer's recommendations, and cells were treated with a Me 2 SO shock treatment after 16 h of transfection to increase transfection efficiency (10). After 48 h, cells were trypsinized and subcultured 1:3, and Geneticin was added at a concentration of 0.5 mg/ml for colony selection. Geneticin-resistant colonies were isolated, and individual colonies were expanded into mass cultures. Each colony, presumably derived from a single cell, was examined by quantitative immunoblot analysis and compared with cells transfected with vector only. Clonal cell lines were used for up to four passages for various assays. The expression of ␣A-crystallin was unchanged over four passages as determined by immunoblot analysis. Cell lines expressing wild type ␣A-crystallin or the R116C mutant from 0.1 to 2.5 ng/g cellular protein were used.
Immunofluorescence-To demonstrate the expression of wild type or R116C ␣A-crystallin in individual cells of a culture, cells were labeled and immunofluorescence and confocal microscopy were used as described previously (5,6,10). Briefly, cells were plated on coverslips, fixed with 4% p-formaldehyde for 1 h, and permeabilized in 0.1% Triton X-100, and nonspecific binding was blocked by incubation in 10% normal goat serum for 30 min. To visualize the distribution of ␣A-crystallin, cells were incubated overnight with a 1:50 dilution of a monoclonal antibody to bovine ␣A-crystallin (a gift from Dr. Paul Fitzgerald). A lissamine-rhodamine-conjugated goat anti-mouse IgG was used as a secondary antibody. To visualize the distribution of actin, cells were incubated in a 1:50 dilution in PBS of a fluorescein phalloidin (Molecular Probes) methanolic stock solution (100 units/ml ethanol). Cells were stained with fluorescein phalloidin for 20 min, washed three times for 5 min with PBS, and viewed. Lens epithelial cells were viewed using a Zeiss LSM 410 confocal microscope equipped with an argon-krypton laser.
The distribution of ␣B-crystallin was also examined in cultures expressing wild type ␣A-crystallin or the R116C mutant. To visualize the distribution of ␣B-crystallin, cells were incubated overnight in a 1:100 dilution of an antibody to bovine ␣B-crystallin (Nova Castra). An Alexa 488-conjugated goat anti-rabbit IgG was used as the secondary antibody.
Western Blotting-Western blotting was used to examine the expression of ␣Aand ␣B-crystallin in transfected HLE B-3 lens epithelial cultures (5,10). The antibody used for Western blot analysis of ␣Acrystallin was a polyclonal antiserum to bovine ␣A-crystallin (at a dilution of 1:1000). Immune complexes were detected using 125 I-protein A (5,31,32). The detected proteins were quantified with the Storm 860 phosphorimaging system (Molecular Dynamics) using the ImageQuant program (10).
FPLC Gel Filtration-FPLC was carried out on a Superose 6 column (FPLC system, Amersham Biosciences, Inc.). HLE B-3 cell extracts were prepared in gel filtration buffer (50 mM Tris-HCl, 1 mM EDTA, and 100 mM NaCl) by homogenization of 10 6 cells/ml. The extract was centrifuged at 10,000 ϫ g at 4°C. The sample solution was filtered through a 0.45-m filter before application to the column. The column was eluted at 0.3 ml/min. Aliquots of the collected fractions were placed in a 96-well plate for detection of ␣A-crystallin by enzyme-linked immunosorbent assay (ELISA). A polyclonal antibody specific to ␣A-crystallin (1:1000) and a fluorescence-labeled secondary antibody (Alexa 488 goat anti-rabbit IgG, Molecular Probes, Inc.) were used, and the immune complexes were analyzed on an HTS 7000 PerkinElmer Life Sciences fluorescence plate reader.
Stress Conditions-Cells were exposed to UVA radiation comparable to in vivo levels as described previously (5,10). Cells were washed twice with phosphate-buffered saline (PBS), and a layer of PBS was added to controls and cells to be irradiated. UVA radiation at 365 nm was obtained from a mercury-xenon arc lamp using a monochromator, focusing lens, and beam turner to give a fluence rate of 0.007 watts/cm 2 and a total fluence of 37.8 J/cm 2 . Cells were processed 4 h after UVA irradiation in normal tissue culture medium.
Analysis of Cell Death-Cells were labeled with Annexin V-FITC (PharMingen) and propidium iodide according to the manufacturer's recommendations, and flow cytometry was performed to assess apoptotic cell death as described previously (5,10). Briefly, cells were washed with PBS and exposed to UVA irradiation as described above and incubated in normal medium for 4 h. To distinguish apoptotic cells from necrotic cells, attached cells were trypsinized, combined with cells floating in the medium, washed with PBS, and resuspended in 0.5 ml of annexin-binding buffer. 5 l of Annexin V-FITC and 10 l of a 50 g/ml propidium iodide solution were added, and samples were analyzed in a Becton Dickinson FACScan as described previously (10). The flow cy-tometer had a laser excitation beam at 488 nm. Band pass filters at 530 nm and 685 nm were used to collect the fluorescence emission, and data collected in each window were designated FL1-H and FL2-H, respectively. Unlabeled cells, cells labeled with annexin only, and cells labeled with PI only were used as controls to adjust the compensation between the flow cytometer and the detectors and to set the quadrants. Data analyses were done with Cell Quest software. Percent protection by different levels of ␣A-crystallin expression was determined as the ratio of live cells in treated culture of cells with no ␣A-crystallin and with different levels of ␣A-crystallin expression.
TUNEL labeling was also used to examine cell death in cultures exposed to UVA stress conditions. Cells were fixed in 4% paraformaldehyde, pH 7.4, for 30 min and permeabilized for 30 min in 0.1% Triton X-100/PBS at room temperature, and apoptotic nuclei were detected using a TUNEL-labeling reaction according to the manufacturer's instructions (Roche Biochemicals), as described previously (5). To examine the actin cytoskeleton, cells were co-labeled with Texas-red phalloidin. Images were recorded on a confocal microscope. In other experiments, TUNEL-labeled cells were co-labeled with propidium iodide to count the total number of nuclei. Percent apoptosis was determined as the ratio of TUNEL-labeled cells to total number of cells multiplied by 100.
Immunoprecipitation-A co-immunoprecipitation assay was used to investigate proteins associating with ␣A-crystallin in lens epithelial cells and to investigate the effect of stress on the interaction of ␣Acrystallin with cellular proteins. Cells were lysed for 30 min on ice with immune precipitation buffer containing 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 0.1% sodium deoxycholate, and protease inhibitor mixture (Sigma Chemical Co.) and centrifuged for 10 min at 10,000 ϫ g (32). Supernatants were treated with a primary antibody to ␣A-crystallin and immunoprecipitated with Protein A/Gagarose beads (Santa Cruz Biotechnologies). A monoclonal antibody to ␣A-crystallin (1:20) or a rabbit polyclonal antibody to ␣A-crystallin (1:200) was used. Immunoprecipitates were washed three times with a lysis buffer containing 20 mM Tris-HCl, 1% Triton X-100, 5 mM EDTA, 50 mM NaCl, and protease inhibitor mixture (Sigma), resuspended in SDS-PAGE sample buffer, and analyzed on 15% acrylamide gels as described previously (5,6,10,32). Proteins were stained with the SilverQuest silver staining kit (Invitrogen). Gels were also stained with Bio-Safe Coomassie Blue G-250 (Bio-Rad).
MALDI Mass Spectrometric Analysis-Stained bands were excised from the polyacrylamide gel and were digested with Promega (Madison, WI) modified porcine trypsin overnight at 37°C in ammonium bicarbonate buffer, using a protocol developed in the Washington University Protein Chemistry Laboratory. The recovered tryptic peptides were dried and dissolved in 5 l of 50% acetonitrile made 0.1% in trifluoroacetic acid. 0.5 l of this solution was co-crystallized on the MALDI target plate with 0.5 l of a saturated solution of ␣-cyano-4-hydroxycinnamic acid (33,34). The MALDI spectrum was acquired on a Per-Septive Biosystems (Framingham, MA) Voyager DE-PRO MALDI-TOF mass spectrometer running Voyager version 5.1 software. The peptide masses from the resultant mass spectrum were submitted to Protein Prospector MS-Fit peptide mass fingerprinting analysis and searched against the NCBI non-redundant protein data base.

Expression of Wild Type and R116C ␣A-crystallin in Lens
Epithelial Cells-To compare their relative ability to protect lens epithelial cells from UVA radiation-induced stress, wild type or R116C␣A-crystallin were expressed in the human lens epithelial cell line HLE B-3. Sequences encoding human wild type ␣A-crystallin were cloned into the mammalian expression vector pCIneo and transfected into the HLE B-3 cell line at passages higher than 11. (Cells at passage 11 or higher have no detectable level of ␣A-crystallin by immunoblot analysis (32), allowing the effect of ␣A-crystallin expression to be studied against a null background.) Mutagenesis of wild type ␣A-crystallin to R116C was carried out using the QuikChange mutagenesis kit from Stratagene. Each construct was confirmed by sequencing. Cell lines were expanded from Geneticin-resistant clones and cell lysates were analyzed by immunoblot analysis using an antibody to ␣A-crystallin. Fig. 1 shows representative immunoblot analysis of individual cell lines expressing different levels of wild type ␣A-crystallin or the R116C mutant. Cell lines expressing wild type ␣A-crystallin or the R116C mutant ranging between 0.1 and 2.5 ng/g cellular protein were used to study the effect of UVA radiation.
The intracellular distribution of ␣A-crystallin in transfected HLE B-3 cells was examined by immunofluorescence. Wild type ␣A-crystallin was distributed in the cytoplasm of these cells, with lower levels around the periphery (Fig. 2A). The actin cytoskeleton was also labeled uniformly in these cells. In a majority of the cells, there was little or no ␣A-crystallin detected in the nucleus. The overall morphology of the R116C ␣A-expressing cells was similar, but cells contained more vacuoles than the wild type ␣A-crystallin-expressing cells. Like the wild type ␣A-crystallin, R116C ␣A-crystallin was also distributed mainly in the cellular cytoplasm (Fig. 2B).
The distribution of ␣B-crystallin was examined in cell lines expressing wild type ␣A-crystallin and the R116C mutant to determine whether the mutation in ␣A-crystallin affected the localization of ␣B-crystallin. Consistent with previous data, HLE B-3 cells expressed ␣B-crystallin in their cytoplasm. As shown in Fig. 3, ␣B-crystallin (green) was co-localized with ␣Acrystallin (red) in the cytoplasm of the transfected cells, as indicated by the yellow color. The distribution of ␣B-crystallin in wild type and R116C ␣A-crystallin-expressing cells was similar.
Size Distribution of Wild Type and R116C ␣A-crystallin-Several studies in the literature indicate that the R116C ␣Acrystallin has a higher aggregate molecular mass than the wild type protein (25)(26)(27). The R116C mutation also alters the tertiary structure and chaperone-like activity in preventing in vitro protein aggregation (25)(26)(27). Because changes in tertiary and quaternary structure may have an effect on the protective ability of the protein, we determined the size distribution of wild type ␣A-crystallin and the R116C mutant expressed in HLE B-3 cells. Cells expressing wild type ␣A-crystallin or the R116C mutant were treated with hypotonic shock, and the water-soluble proteins were separated by Superose 6 FPLC chromatography and analyzed by ELISA using an antibody specific to ␣A-crystallin (Fig. 4). This analysis revealed that the wild type ␣A-crystallin eluted at a size of ϳ600 kDa, whereas the R116C␣A eluted from the column at the void volume, indicating a size of Ͼ2 MDa. In addition to the major peak at the void volume, a minor peak with a molecular mass of 500 -600 kDa was also observed for the R116C mutant, and may represent heterogeneity in aggregate size of the mutant. Effect of UVA Radiation on Wild Type and R116C ␣A-crystallin-expressing Cells-We next compared the protective effect of expression of wild type ␣A-crystallin and the R116C mutant on transfected HLE B-3 cells exposed to UVA radiation. Equal numbers of cells expressing wild type or R116C ␣A-crystallin were exposed to physiological levels of UVA radiation, incubated for 4 h, and labeled with annexin V-FITC and propidium iodide (PI). The number of cells in the annexin-positive/PInegative, annexin-positive/PI-positive, annexin-negative/PIpositive, and unlabeled cells was determined by flow cytometry. Less than 1% of the wild type ␣A-expressing cells were labeled with annexin, as compared with 5% of the R116C ␣A-expressing cells (lower right quadrant of Fig. 5, A and C). This indicates that wild type ␣A-crystallin-expressing cells had a lower basal level of apoptosis than R116C ␣A-crystallinexpressing cells. Cells were exposed to 37.8 J/cm 2 of UVA (365 nm) radiation as described previously (5,10). UVA-induced annexin labeling was much higher in R116C ␣A-crystallinexpressing cells, but expression of wild type ␣A-crystallin diminished UVA radiation-induced annexin labeling (Fig. 5B). After UVA exposure, there were 19% cells in the early stages of apoptosis in the R116C ␣A-crystallin expressing cells as compared with only 1% in the wild type ␣A-crystallin-expressing cells (lower right quadrants of Fig. 5, B and D). The number of cells that had already died by apoptosis was also lower in UVA-treated wild type ␣A-crystallin-expressing cultures as compared with those expressing R116C ␣A-crystallin (Fig. 5, B  and D, upper right quadrant). Concurrently, unlabeled cells representing the live population (lower left quadrants of Fig. 5) decreased only slightly in wild type ␣A-crystallin expressing cells (from 90% to 89%) but decreased from 83% to 63% in R116C ␣A-crystallin expressing cells. These observations indicate that the expression of wild type ␣A-crystallin protects HLE B-3 cells from UVA radiation-induced apoptotic cell death, whereas R116C ␣A-crystallin was less protective.
We also analyzed the protective effects of expression of different levels of ␣A-crystallin or R116C mutant in stably transfected HLE B-3 cells. The relative level of expression of ␣Acrystallin in the transfected cell lines was determined by immunoblot analysis (Fig. 1). Quantitative analysis using several clonal cell lines revealed that the R116C ␣A-crystallin had a 4-to 10-fold lower protective ability than wild type ␣Acrystallin against UVA radiation-induced apoptotic cells death (Fig. 6). These results coincide with previous reports indicating that the chaperone-like activity of the R116C mutant is significantly reduced (25)(26)(27).
To investigate the effect of UVA stress on the actin cytoskeleton, apoptotic cells were identified by TUNEL staining, and F-actin was visualized with Texas Red phalloidin staining. TUNEL labeling was not detected in untreated wild type ␣Acrystallin expressing cells (Fig. 7A). A low level of TUNEL labeling could be seen in untreated R116C mutant-expressing cells, and the cytoskeleton had a number of actin rings in these cells (Fig. 7C). After exposure to UVA stress, there was sparse TUNEL labeling in cells expressing wild type ␣A-crystallin, and the actin cytoskeleton was not affected significantly by UVA treatment (Fig. 7B). TUNEL-positive cells were abundant in UVA-treated R116C mutant-expressing cells, and actin was Note that wild type ␣A-crystallin eluted at a molecular mass ϳ500 -600 kDa. Note also that the R116C mutant eluted at the void volume corresponding to a molecular mass Ͼ 2 MDa. either absent or was clumped around some of the TUNELlabeled cells (Fig. 7D). Quantitative analysis showed that the number of TUNEL-positive cells increased 3-4-fold in the R116C mutant-expressing cells exposed to UVA radiation as compared with cells expressing wild type ␣A-crystallin exposed to UVA radiation (Fig. 7E). Cells transfected with vector only were also exposed to UVA radiation as described previously (5,10). Comparison of the mock transfected and R116C mutantexpressing cells indicated that the percent apoptosis in the R116C mutant-expressing cells was similar to that of mock transfected (vector only) cells. This indicates that the R116C mutant has a lower ability to prevent apoptosis than wild type ␣A-crystallin.
The cellular localization of ␣A-crystallin in UVA-treated clonal cell lines expressing wild type or R116C mutant was also analyzed. ␣A-crystallin was labeled with a monoclonal antibody to ␣A-crystallin and an Alexa 568 goat anti-mouse IgG as secondary antibody. F-actin was visualized by fluorescein phalloidin staining. Confocal microscopy showed that UVA treatment did not dramatically change the cellular localization of ␣A-crystallin in wild type or R116C-expressing cells, and the protein was detected mainly in the cytoplasm (data not shown).
Co-immunoprecipitation-Cells expressing wild type ␣Acrystallin were exposed to UVA radiation. Proteins that associate with ␣A-crystallin in untreated cells and cells exposed to UVA radiation were immunoprecipitated with an ␣A-crystallin-specific antibody. Extensive washing of the immunoprecipitates was done to ensure that the nonspecific interactions between ␣A-crystallin and cellular proteins did not occur. Immunoprecipitated proteins were separated on SDS-PAGE and silver-stained (Fig. 8A). Several protein bands corresponding in molecular mass to cytoskeletal proteins appeared to associate with ␣A-crystallin in lens epithelial cells both in control and UVA-treated cells. In addition, proteins with low molecular mass (11)(12)(13)(14)(15)(16)(17)(18) were observed by silver staining.
The immunoprecipitated proteins were analyzed by MALDI mass spectrometric analysis. The MALDI mass spectra obtained for the ϳ54-kDa protein shown on the gel and peptide mass data base searches of the labeled signals by MS-Fit analysis gave strong data for vimentin, which came up across many Wild type or R116C mutant ␣A-crystallin-expressing cells were irradiated with UVA radiation (365 nm, 37.8 J/cm 2 ), incubated for 4 h in normal growth medium, and labeled with annexin and propidium iodide (PI). The distribution of live and dead cells in the cultures was determined by FACS analysis of annexin and PI-labeled cells. Fluorescence of annexin is shown on the x axis, and PI on the y axis. Annexin labeling (lower right quadrants) represents the population undergoing apoptosis. Annexin and PI double labeling (upper right quadrants) represent cells that have already died by apoptosis. Live cells are unlabeled with annexin or PI, and are shown in the lower left quadrants. Each dot represents an individual cell. 10,000 cells were analyzed in each sample. A, control (unirradiated) wild type ␣A-crystallin-expressing cells. B, wild type ␣A-crystallinexpressing cells after UVA irradiation and 4-h incubation in normal growth medium. C, control (unirradiated) R116C mutant ␣A-crystallinexpressing cells. D, R116C mutant ␣A-crystallin-expressing cells after UVA irradiation and 4-h incubation in normal growth medium. Note that the basal level of apoptosis was higher in R116C mutant-expressing cells than in wild type ␣A-crystallin-expressing cells as shown by the higher number of dots in the lower right quadrants of untreated (control) cells. An increase in annexin-labeled cells (i.e. cells undergoing apoptosis) in the lower right quadrants of UVA irradiation of the R116C mutant-expressing cells was observed. Note also that there was no significant increase in the annexin labeling of the wild type ␣A-crystallin-expressing cells after UVA stress. Cells double-labeled with annexin and PI (i.e. cells that had died by apoptosis) shown in the upper right quadrants also increased noticeably for the UVA-treated R116C mutant-expressing cells.
organisms and was in good agreement with the molecular weight on the silver-stained gel (Fig. 8, B and C). The MALDI mass spectrum obtained for the ϳ42-kDa protein from the silver-stained gel and peptide mass data base searches of the labeled signals by MS-Fit analysis showed that the observed spectrum corresponded to actin, a very well conserved protein, and the mass was in good agreement with the molecular weight observed on the gel (data not shown). The mass spectrometric data for the low molecular mass proteins gave a strong match for two histones. The protein with a molecular mass of ϳ14 kDa on the gel was identified to be histone H2B (molecular mass 13.94 kDa) by MALDI mass spectra and MS-Fit analysis. The protein with a molecular mass of ϳ11 kDa on the gel was identified to be histone H4 (molecular mass 11.36 kDa) by MALDI spectra and MS-Fit analysis. The other bands in the low molecular weight region were also identified to be histones, although with less confidence than histones H2B and H4. Two different antibodies (a monoclonal and polyclonal) to ␣A-crystallin gave the same result in the immunoprecipitation assay (data not shown). Non-immune rabbit serum did not co-immunoprecipitate these proteins (data not shown). The association of histones with ␣A-crystallin suggests that a small amount of ␣A-crystallin may be present in the nucleus.

DISCUSSION
Mutation of a conserved arginine residue (Arg-116) in ␣Acrystallin sequence has been associated with autosomal dominant congenital cataract (23), but the effect of expression of the mutant protein in mammalian cells is unknown. We have shown here that both wild type ␣A-crystallin and the R116C mutant were distributed mainly in the cytoplasm of transfected human lens epithelial cells (HLE B-3). The aggregate size of the R116C mutant was 4-fold higher than that of the wild type protein, consistent with previous studies on the protein expressed in bacterial systems (25,27). The cells expressing R116C mutant were more vacuolar than cells expressing wild type ␣A-crystallin and had a higher basal level of apoptosis as indicated by annexin labeling and flow cytometry, suggesting that expression of the mutant protein had a mild cytotoxic effect on cells. The actin cytoskeleton in cells expressing the R116C mutant also showed subtle changes, with increased formation of actin rings, which have been detected previously in stressed cells (5). Because ␣A-crystallin is normally associated with ␣B-crystallin in the lens, we also examined the distribution of ␣B-crystallin. The results showed that the R116C mutation in ␣A-crystallin did not markedly affect the distribution of ␣B-crystallin in lens epithelial cells.
The ability of ␣A-crystallin to protect cells from physiological levels of UVA radiation was significantly diminished by the R116C mutation. The expression of wild type ␣A-crystallin almost completely protected lens epithelial cells from UVA radiation-induced apoptotic cell death. In striking contrast, R116C ␣A-crystallin expression had a 4-to 10-fold reduced protective ability. The results obtained with annexin labeling and flow cytometric analysis were consistent with TUNEL labeling of cells exposed to UVA stress. The distribution of ␣A-crystallin in cells exposed to UVA stress was not altered dramatically. Previous studies on the R116C mutant protein expressed in bacterial systems indicate that the mutant protein has an altered tertiary structure, a higher aggregate size, and a reduced ability to protect nonspecific aggregation of denatur-  7. Merged confocal images of TUNEL and Texas Red phalloidin staining after UVA-induced apoptosis in lens epithelial cells. Cells were exposed to 37.8 J/cm 2 UVA radiation, incubated in 20% fetal bovine serum-Eagle's minimum essential medium for 4 h, and processed for TUNEL labeling. A, wild type ␣A-expressing cells not exposed to UVA radiation had no detectable TUNEL labeling. B, wild type ␣A-expressing cells exposed to UVA radiation had very sparse TUNEL labeling shown in green (arrows). C, R116C mutant-expressing lens epithelial cells untreated with UVA radiation had a small but significant number of TUNEL-positive cells. D, R116C mutant-expressing lens epithelial cells treated with UVA radiation had a large number of TUNEL-positive cells. Note that the F-actin fluorescence shown in red was absent or clumped around some of the TUNEL-positive cells. E, mock transfected, wild type ␣A-crystallin-expressing, or R116C mutant-expressing cells were exposed to UVA radiation at 365 nm (37.8 J/cm 2 ) and incubated in normal medium for 4 h. Cells were TUNELlabeled and stained with propidium iodide to enable a count of the total number of cells in each microscopic field. The number of TUNELlabeled cells and PI-labeled cells were counted in 10 randomly selected fields from three or more cultures. The asterisks indicate that the difference between the TUNEL labeling was statistically significant for UVA-treated cells expressing wild type ␣A-crystallin and those expressing R116C mutant or mock transfected cells.
FIG. 8. Analyses of proteins associating with ␣A-crystallin. A, cells expressing wild type ␣A-crystallin (2.0 ng/g of protein) were untreated or exposed to UVA radiation, lysed, and immunoprecipitated with a polyclonal antibody to ␣A-crystallin. Immunoprecipitates were analyzed by SDS-PAGE. Proteins were detected by silver staining using a kit from Invitrogen. Lane 1, low molecular weight markers (Amersham Biosciences, Inc.); 1 g per protein.
Only an outline of the protein bands was detected by silver staining. Lane 2, control (untreated) cells expressing wild type ␣A-crystallin. Lane 3, UVA radiation-exposed cells expressing wild type ␣A-crystallin. Lanes 4, native human ␣-crystallin isolated from 1-year-old human lens, 20 ng of protein. Major proteins associating with ␣A-crystallin included ϳ54-kDa and ϳ42-kDa proteins and bands with molecular mass Ͻ20 kDa. B, MALDI mass spectra of the ϳ54-kDa protein associating with ␣A-crystallin. C, MS-Fit peptide mass fingerprinting analysis against NCBI peptide data base search identified the protein to be vimentin. ing proteins (25,27). Our results suggest that structural alteration of the R116C mutant may result in a decreased chaperone activity toward proteins in lens epithelial cells.
Because the lens is exposed to light-induced stress due to its location along the optical axis of the eye, it is likely that mechanisms may be present to protect the lens from a lifetime of exposure to environmental and metabolic stress. UVA radiation, which increases oxidative stress by production of reactive oxygen species and causes cataract in animals, induces apoptosis of lens epithelial cells, with a prominent effect occurring 4 h after exposure (5). UVA radiation affects cytoskeletal structures in lens epithelial cells (35). We have shown recently that ␣A-crystallin expression protects lens epithelial cells from UVA radiation stress and protects the actin cytoskeleton (5). This observation is consistent with previous work indicating that ␣A-crystallin interacts with actin in lens cells (36 -38). ␣Aand ␣B-crystallins stabilize actin filaments and prevent their depolymerization and have been proposed to act as chaperones for actin (39). The present study showed that the R116C mutant had a significantly lower ability to protect lens epithelial cells from apoptotic cell death. F-actin was disrupted after UVA radiation of lens epithelial cells expressing the R116C mutant but was protected in cells expressing the wild type protein. It has been reported that ␣B-crystallin may also contribute to the stability of the cytoskeletal organization and interact with cytoplasmic intermediate filament bundles during mitosis or heat-shock (14, 40 -42). Other studies have identified a specific association between ␣B-crystallin and intermediate filaments in human pathologies involving intermediate filament aggregates, and the R120G missense mutation in ␣B-crystallin has been shown to co-segregate with desmin-related myopathy, characterized by adult onset accumulation of desmin aggregates (24,(43)(44)(45)(46).
The mechanism by which ␣A-crystallin protects lens epithelial cell apoptosis may be a complex mechanism involving protection of cellular proteins, including the cytoskeletal proteins.
Our studies indicate that ␣A-crystallin protects cells from apoptotic death, and structural changes caused by the R116C mutation may be responsible for the decreased protective ability of the mutant. Under the conditions of UVA stress, some of the R116C mutant may become insoluble, thereby decreasing its protection. ␣A-crystallin may also exert a protective effect by either indirectly affecting the activation of anti-apoptotic factors, such as Bcl-2, or prevent the activation of pro-apoptotic factors, such as caspases. Previous studies on the mechanism by which small heat-shock proteins prevent apoptosis indicate that different mechanisms are involved in the protection by ␣B-crystallin and HSP27. ␣B-crystallin was shown to inhibit the autoproteolytic maturation of caspase-3 (11). In contrast, HSP27 inhibits apoptosis clearly by disrupting the cytochrome c-dependent activation of pro-caspase-9, an event that was only weakly inhibited by ␣B-crystallin (11). It has also been reported that intermediate filament proteins such as keratins, lamins, and vimentin undergo caspase-mediated proteolysis in an apoptosis-related manner (47). Further studies are in progress to investigate the role of ␣A-crystallin in inhibiting the action of caspases.
Our immunoprecipitation data suggest that ␣A-crystallin may associate with cytoskeletal proteins and histones in lens epithelial cells. The possible association of ␣A-crystallin with histones in lens epithelial cells needs to be investigated further. Our studies suggest that a small amount of ␣A-crystallin may be present in the nucleus, but it was not detected by the antibody. The phosphorylation of certain histones is dependent on the activation of caspases, and therefore may be linked with caspase-induced signaling pathways (48 -50). Histone H2B has been reported to undergo phosphorylation in response to apoptotic signals in mammalian cells, and this modification initiates around the time of nucleosomal DNA fragmentation (48 -51). It has been suggested that histone phosphorylation may facilitate access of the damaged DNA to repair mechanisms (51). ␣A-crystallin has been shown to undergo phosphorylation in vivo and act as an autokinase (52). It is also noteworthy that histone transcripts accumulate in lens epithelial cells after UVC irradiation (53). The phosphorylation of heat-shock protein 60 regulates its attachment to histone H2B in the plasma membrane (54), and a nucleosome assembly protein has recently been shown to function as a histone chaperone (55). The possible association of ␣A-crystallin with histones in lens epithelial cells is also consistent with its potential role in the regulation of cell growth (5). Current studies in our laboratory are investigating whether ␣A-crystallin directly affects lens epithelial cell growth by preventing apoptosis in vivo.
It is interesting to note that the physiological substrates of the chaperone GroEL include essential components of transcription and translational machinery and metabolic enzymes (56). GroEL interacts with 10 -15% of all newly synthesized proteins under normal growth conditions of Escherichia coli and with 30% of all cytoplasmic proteins under heat stress (57). GroEL substrates contain a preferred structural motif with several protein domains having ␣␤-folds containing ␣-helices and buried ␤-sheets with large hydrophobic surfaces (57). These proteins are expected to fold slowly and may be prone to aggregation. Most GroEL substrates were larger that 20 kDa, but a majority were smaller than 60 kDa. Small heat-shock proteins on the other hand, appear to preferentially interact with intermediate filament proteins and actin (43).
In summary, we have demonstrated that wild type ␣A-crystallin protects lens epithelial cells from apoptotic cell death and that mutation of arginine 116 to cysteine, which has been associated with autosomal dominant cataract in humans, results in a protein with a dramatically lower protective ability. Our work also shows that comparing the in vitro stress-induced apoptotic cell death is an effective way to measure the protective abilities of the wild type and mutant ␣A-crystallin. Although it remains to be determined whether, like ␣B-crystallin and HSP27, ␣A-crystallin affects caspase activities, the present studies show that the cell transfection system used here can serve as a useful model for understanding the effects of mutations in ␣A-crystallin under physiological conditions.