Differential Protective Activity of αA- and αB-crystallin in Lens Epithelial Cells*

αA- and αB-crystallins are molecular chaperones expressed at low levels in lens epithelial cells, and their expression increases dramatically during differentiation to lens fibers. However, the functions of αA- and αB-crystallins in lens epithelial cells have not been studied in detail. In this study, the relative ability of αA- and αB-crystallin, in protecting lens epithelial cells from apoptotic cell death was determined. The introduction of αA-crystallin in the transformed human lens epithelial (HLE) B-3 lens epithelial cell line (which expresses low endogenous levels of αB-crystallin) led to a nearly complete protection of cell death induced by staurosporine, Fas monoclonal antibody, or the cytokine tumor necrosis factor α. To further study the relative protective activities of αA- and αB-crystallins, we created a cell line derived from αA−/−αB−/− double knockout mouse lens epithelia by infecting primary cells with Ad12-SV40 hybrid virus. The transformed cell line αAαBKO1 derived from αA/αB double knockout cells was transfected with αA- or αB-crystallin cDNA contained in pCIneo mammalian expression vector. Cells expressing different amounts of either αA-crystallin or αB-crystallin were isolated. The ability of αA- or αB-crystallin to confer protection from apoptotic cell death was determined by annexin labeling and flow cytometry of staurosporine- or UVA- treated cells. The results indicate that the anti-apoptotic activity of αA-crystallin was two to three-fold higher than that of αB-crystallin. Our work suggests that comparing the in vitro annexin labeling of lens epithelial cells is an effective way to measure the protective activity of αA- and αB-crystallin. Since the expression of αA-crystallin is largely restricted to the lens, its greater protective effect against apoptosis suggests that it may play a significant role in protecting lens epithelial cells from stress.

␣Aand ␣B-crystallins are molecular chaperones expressed at low levels in lens epithelial cells, and their expression increases dramatically during differentiation to lens fibers. However, the functions of ␣Aand ␣B-crystallins in lens epithelial cells have not been studied in detail. In this study, the relative ability of ␣Aand ␣B-crystallin, in protecting lens epithelial cells from apoptotic cell death was determined. The introduction of ␣A-crystallin in the transformed human lens epithelial (HLE) B-3 lens epithelial cell line (which expresses low endogenous levels of ␣B-crystallin) led to a nearly complete protection of cell death induced by staurosporine, Fas monoclonal antibody, or the cytokine tumor necrosis factor ␣. To further study the relative protective activities of ␣Aand ␣B-crystallins, we created a cell line derived from ␣A؊/؊␣B؊/؊ double knockout mouse lens epithelia by infecting primary cells with Ad12-SV40 hybrid virus. The transformed cell line ␣A␣BKO1 derived from ␣A/␣B double knockout cells was transfected with ␣Aor ␣B-crystallin cDNA contained in pCIneo mammalian expression vector. Cells expressing different amounts of either ␣A-crystallin or ␣B-crystallin were isolated. The ability of ␣Aor ␣B-crystallin to confer protection from apoptotic cell death was determined by annexin labeling and flow cytometry of staurosporine-or UVA-treated cells. The results indicate that the anti-apoptotic activity of ␣A-crystallin was two to threefold higher than that of ␣B-crystallin. Our work suggests that comparing the in vitro annexin labeling of lens epithelial cells is an effective way to measure the protective activity of ␣Aand ␣B-crystallin. Since the expression of ␣A-crystallin is largely restricted to the lens, its greater protective effect against apoptosis suggests that it may play a significant role in protecting lens epithelial cells from stress.
␣-Crystallin, a member of the small heat shock protein (sHSP) 1 family of molecular chaperones, is an aggregate of two polypeptides, ␣Aand ␣B-crystallins, that share 55% amino acid sequence identity. The two ϳ20-kDa subunits form soluble aggregates with an average molecular mass of 600 -800 kDa and can be isolated from lens fiber cells as a heteroaggregate containing ␣Aand ␣B-polypeptides in a 3:1 ratio. The ␣-crystallins have dynamic polymeric structures with a hollow interior and undergo inter-aggregate subunit exchange (1,2).
Initially regarded as lens-specific, the expression of these proteins (especially ␣B) has since been observed in other tissues, and the two subunits share a ϳ40% sequence identity with HSP27 (3)(4)(5). ␣-Crystallin is the most abundant soluble protein in the lens and plays an important role in establishing the refractive index of the cytoplasm. However, the extralenticular expression of ␣-crystallins, their autokinase activity, phosphorylation patterns, association with neurodegenerative diseases, and ability to protect against heat shock, argue for a more generalized cellular function of ␣-crystallins, over and above their role in light refraction (5)(6)(7)(8)(9). Like other sHSPs, ␣Aand ␣B-polypeptides can act as molecular chaperones in vitro, preventing protein aggregation induced by heat and other stresses (1). Of the two subunits, ␣A-crystallin expression is considered to be more lens-specific, but only ␣B-crystallin expression is induced by stress (3)(4)(5)(6)(7)(8). However, the in vivo functions and mechanism of chaperone action of the ␣-crystallin aggregates are largely enigmatic.
Lens growth begins with the division of lens epithelial cells in the germinative zone near the equator. As new cells elongate to form lens fibers, they wrap around the lens periphery and meet at the sutures. Transcripts for ␣Aand ␣B-crystallins can be detected in lens epithelial cells at early stages of mouse lens development (10), and a marked increase in ␣A-crystallin accompanies differentiation (11), but the specific cellular functions of the ␣-crystallins in the lens epithelium are largely unknown.
Recent work on lens epithelial cells derived from ␣Aor ␣B-crystallin knockout mice suggest that they play an important role in regulating cellular growth (9,12), consistent with the roles of other small heat shock proteins in normal growth and development (13). Because lens epithelial cells grow throughout life, and 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 these cells from a lifetime of exposure to metabolic and environmental stress. ␣Aand ␣B-crystallin expression in lens epithelial cells could presumably be helpful in providing protection for these cells.
Recent studies in our laboratory demonstrated that UVA radiation, which increases oxidative stress by production of reactive oxygen species and produces cataract in animals, induces lens epithelial cell death, and that the expression of ␣A-crystallin prevents cell death (9). Because ␣A-crystallin is largely restricted to the lens, it is important to investigate its role in protecting lens epithelial cells from diverse stress conditions. TNF␣ is a cytokine that kills cells via apoptosis or necrosis depending on cell type (14). Its mechanism of action involves an increase in reactive oxygen species in cells upon stimulation of TNF␣ receptors (14). The expression of ␣Bcrystallin and HSP27 prevents TNF␣-induced cell death in murine L929 cells, as well as cell death induced by crosslinking of Fas (a member of the TNF and nerve growth family of receptors) and staurosporine, a protein kinase C inhibitor (14 -16). Proteins of the HSP70 and HSP90 chaperone families also regulate apoptosis (17,18). The mechanism by which molecular chaperones prevent cell death presumably involves an inhibition of factors that can induce cytochrome c release from mitochondria, and by inhibition of the activation of caspases (19). The study of physiological roles of the molecular chaperones ␣Aand ␣B-crystallin can be assisted with the development of mammalian tissue culture systems in which the cloned crystallin genes can be expressed and the protective activities of the proteins under various stress conditions can be probed. In the present work, we describe cell lines derived from ␣Aand ␣B-crystallin knockout mouse lens epithelial cultures in which the two proteins can be re-introduced. These cell lines, as well as transformed human lens epithelial cell lines, allowed us to investigate the relative protective activity of ␣Aand ␣B-crystallin in preventing apoptosis of lens epithelial cells induced by staurosporine, TNF␣, anti-Fas, and UVA radiation.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-Mouse and human lens epithelial cell cultures were used in this study. The wild-type mouse strain was the 129SvJ mouse. ␣AKO-127 knockout mice were made by targeted disruption of the ␣A-crystallin gene (20). ␣BKO-168 knockout mice were made by the targeted disruption of the ␣B-crystallin gene. The knockout vector eliminated the common promoter region of the ␣B-crystallin and HSPB2 genes, as well as almost all of the coding regions of these two genes (21). Northern blot analysis demonstrated the absence of HSPB2 mRNA in tissues normally expressing this gene (data not shown) confirming disruption of this gene.
␣A/␣BKO-127 mice lacking both ␣Aand ␣B-crystallins were generated by cross breeding ␣AKO-127 and ␣BKO-168 mice (21). Mouse lenses were obtained from 6-to 12-week-old mice. Capsule epithelia of wild-type and knockout lenses were dissected and cultured in 20% fetal bovine serum-minimal essential medium in 24-well tissue culture plates as described earlier (9). Cells were passaged using Trypsin-EDTA and plated in 35-mm plates. In some experiments, cells were grown on sterile glass coverslips to facilitate microscopic examination. Cultures were fed twice weekly. We can differentiate two different kinds of colonies from ␣B-crystallin knockout mouse lens epithelial cells. When we generated primary cultures of ␣B knockout lens epithelial cells, we noticed that, in a small proportion of cases, highly proliferative cells are generated. These cells are the subject of another study (12). However, in the current work, we have used only the cells with normal growth.
Extended life span mouse lens epithelial cell lines were also used in this study. These transformed cells were created by infection of primary mouse lens epithelial cultures with Ad12-SV40 hybrid virus using a published procedure (22). To investigate the expression of the HSPB2 gene in transformed mouse lens epithelial cells lines, we used reverse transcription-polymerase chain reaction with primers specific for the HSPB2 gene (23) and RNA from transformed lens epithelial cells. Wild-type mouse hind limb skeletal muscle RNA was used as a positive control. The results of our study showed that, as in the case of lens tissue and the primary lens epithelial cells, the transformed mouse lens epithelial cell lines did not express the HSPB2 gene (data not shown).
The pCIneo␣A expression vector used for human ␣A-crystallin expression has been described previously (9). The expression vector for human ␣B-crystallin was made by taking a full-length cDNA clone containing the complete coding sequence of human ␣B-crystallin from an IMAGE clone (24) (from Genome Systems, Inc., St. Louis, Gen-Bank accession number N35834). The ␣B cDNA was cloned into the EcoRI and NotI site of the mammalian expression vector, pCIneo (Promega). The pCIneo␣A or pCIneo␣B was stably transfected into the transformed double knockout mouse lens epithelial cell line, ␣A␣BKO1, by standard calcium phosphate protocols as described previously (9) using the Promega ProFection kit, except that 16 h after transfection cells were given a Me 2 SO shock according to the manufacturer's instructions. This procedure increased the transfection efficiency to approximately 40%. 2 days post-transfection, the cells were trypsinized, subcultured 1:3, and Geneticin (0.5 mg/ml) was then added for colony selection. Geneticin-resistant colonies were isolated, and individual colonies were expanded into mass cultures. Each colony is presumably derived from a single cell. Once a mass population of transfected cells was obtained by expansion of the individual colonies, they were used in passage 1 or 2 for assays. Cultures were examined by quantitative immunoblot analysis for the expression of ␣Aor ␣B-crystallin (9) and compared with mock transfected cells (cells transfected with vector lacking ␣Aor ␣B-crystallin cDNA insert). The expression of ␣Aor ␣B-crystallin did not change with passage under these conditions, as determined by immunoblot analysis.
In this study, we investigated stably transfected mouse lens epithelial clonal cell lines expressing between 0.42 and 1. 2). Mock transfected clones were also used from cells that were transfected with the pCIneo vector alone. Mouse clonal cell lines were grown to 80 -90% confluence before treatment with staurosporine or UVA radiation as described below.
Human lens epithelial cells with extended life span (HLE B-3 cells) have been described previously (22). They were derived from an infant human lens epithelial culture by Ad12-SV40 hybrid virus infection and propagated through at least 11 passages. After passage 11, HLE B-3 cells cease to produce ␣A-crystallin (as determined by immunoblot analysis) (25). Therefore, ␣A-crystallin cDNA can be reintroduced into these cells by cDNA transfection, and stably transfected cell lines expressing different amounts of ␣A-crystallin can be generated. The HLE B-3 cells were cultured in 20% fetal bovine serum-minimal essential medium according to published procedures (22,25). Transfection of HLE B-3 cells with pCI-neo␣A was carried out as described above. The concentration of ␣A-crystallin in transfected cell lines derived from human lens epithelial clonal cell lines ranged between 0.12 and 2.10 ng of human ␣A per g of cellular protein. Human lens epithelial cell lines 113␣A (0.12), 405␣A (0.45), 504␣A (0.70), 809␣A (1.40), and 815␣A (2.10) were used. Mock transfected cell lines 406, 505, and 811 were also used.
Induction of Apoptosis-Cell death was induced by one of the following agents: staurosporine (STP), Fas monoclonal antibody, the cytokine TNF␣, or UVA radiation. Optimal conditions were determined for treatment with different apoptosis-inducing agents by varying the concentration and time of incubation. Staurosporine (Sigma) was prepared as a 0.5 mM stock solution and used at 0.5 M concentration. Cells were treated with staurosporine for 2 h, because readily detectable annexin labeling was observed under these conditions. Human recombinant TNF␣ (1.08 ϫ 10 8 units/mg, Sigma) was also used to induce apoptosis. We titrated the cells with TNF␣ and found that the most effective concentration for triggering apoptosis was 2000 units/ml for 3 h. The apoptosis-inducing antibody to Fas (clone DX2 from PharMingen) was used at a concentration of 0.25 g/ml for 3 h. HLE B-3 cells have been shown to express TNF␣ receptors and Fas by immunofluorescence and immunoblotting. 2 Cells were treated with UVA radiation at 365 nm at a fluence of 37.8 J/cm 2 as described previously (9). UVA-irradiated cells were incubated for 4 h after irradiation in normal tissue culture medium at 37°C and then processed for annexin labeling as described below. In some experiments, UVA-treated cells were incubated for varying times in normal tissue culture medium at 37°C and processed for TUNEL labeling, as described below.
Thermal Stress-HLE B-3 cells were grown to confluence in 35-mm tissue culture plates and exposed to mild thermal stress by incubation at 43°C for 1 h in a water bath. After heat treatment, cells were incubated at 37°C in a water-saturated air incubator containing 5% CO 2 for times ranging from 0.5 to 24 h, and proteins were analyzed by immunoblotting.
Analysis of Cell Death-Annexin V-FITC (PharMingen), a Ca 2ϩ -dependent phospholipid-binding protein that binds to the plasma membrane of cells in early stages of apoptosis, was used. Propidium iodide (PI) binds to cells that have ruptured their membranes and died. Cells were incubated in medium alone or medium containing the staurosporine, TNF␣, or monoclonal antibody to Fas for 2 or 3 h. To distinguish apoptotic cells (annexin-positive/PI-negative) from necrotic cells (annexin-negative/PI-positive) or those cells that had already died via apoptosis (annexin-positive/PI-positive), attached cells were trypsinized, combined with cells floating in the medium, and labeled with annexin V-FITC and PI according to the manufacturer's guide (PharMingen). Briefly, cells were washed with PBS and resuspended in 500 l of annexin-binding buffer. Annexin V-FITC (5 l) and propidium iodide (10 l of a 50 g/ml stock solution) were added, and flow cytometry was performed using a Becton Dickinson FACScan as described previously (9). The flow cytometer 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 fluorescence, 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 fluorescence detectors and to set the quadrants. Data analyses were performed with Cell Quest software.
Live cells did not get labeled with either annexin or PI and were identified in the lower left quadrant of the FACS data, as shown in Fig.  1. Annexin-positive but PI-negative cells were interpreted as cells undergoing apoptosis. These cells were identified in the lower right quadrant. Cells that had already died by apoptosis were both annexinpositive and PI-positive, and these cells were identified in the upper right quadrant. PI-positive but annexin-negative cells, identified in the upper left quadrant, were defined as cells that had already died by necrosis. Percentage of protection by ␣Aor ␣B-crystallin expression under each stress condition was determined by the ratio of unlabeled, live cells in the mock transfected and ␣Aor ␣B-crystallin-expressing clones multiplied by 100.
In some experiments, the TUNEL assay was used to detect apoptotic cells. Cells were fixed in 4% paraformaldehyde, pH 7.4, for 30 min, permeabilized for 30 min in 0.1% Triton X-100/PBS at room temperature, and washed for 20 min in PBS as described previously (9). Apoptotic nuclei were detected using a TUNEL-labeling reaction according to the manufacturer's instructions (Roche Molecular Biochemicals). To determine the proportion of apoptotic cells in a culture, cells were co-labeled with 4Ј,6-diamidino-2-phenylindole (DAPI), a DNA-binding fluorophore, which fluoresces in the UV region. TUNEL-positive nuclei and DAPI-labeled nuclei were visualized using a fluorescence microscope (Axioskop; Zeiss). Images were recorded on a cooled chargecoupled device camera (Photometrix), and the number of TUNEL-positive nuclei and total nuclei were counted using image analysis software (Metamorph, Universal Imaging Corp.).
Western Blotting-Cells (ϳ10 6 ) were washed in PBS, detached with trypsin, and treated with endonuclease (Sigma) for 30 min to degrade DNA. Cells were incubated in lysis buffer containing protease inhibitors leupeptin, pepstatin, and aprotinin (10 g/ml each), 50 mM Tris-HCl, pH 7.45, 150 mM NaCl, 1% Triton X-100, and 0.5% sodium deoxycholate for 30 min on the ice as described previously (9). Insoluble material was removed by centrifugation, and lysates were run on SDS-polyacrylamide gel. After transfer to Immobilon membranes (Millipore) immunoblot analysis was carried out as described previously (9,25). The antibody used for immunoblot analysis of ␣B-crystallin was a rabbit polyclonal antiserum to the 21-amino acid C-terminal peptide of human ␣B raised in rabbits. The antibody was used at a dilution of 1:1000. To detect HSP27, a polyclonal antibody to human HSP27 (SPA-800) was obtained from StressGen Biotechnologies and used at 1:1000 dilution. The antibody to ␣A-crystallin was a monoclonal against bovine ␣Acrystallin (a gift from Dr. Paul Fitzgerald). Immune complexes to ␣Acrystallin, ␣B-crystallin, and HSP27 were detected using 125 I-protein A. The detected proteins were quantified with the Storm 860 phosphorimaging system (Molecular Dynamics) using the ImageQuaNT program.
To determine the -fold change in protein levels induced by thermal stress, relative protein levels were plotted as -fold change from control, untreated cells.
Immunofluorescence-Western immunoblots gave the level of ex-pression of ␣Aor ␣B-crystallin in the cell population as a whole. However, to demonstrate how many cells in the culture expressed the proteins, we also examined the expression of ␣Aand ␣B-crystallin in different cells of a culture by immunofluorescence, as described previously (9). Briefly, cells were fixed for 30 min in 4% paraformaldehyde/ PBS and permeabilized in 0.1% Triton X-100 for 30 min, 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 at 4°C in a 1:100 dilution of a monoclonal antibody against bovine ␣A-crystallin (a gift from Dr. Paul Fitzgerald). A lissaminerhodamine-conjugated goat anti-mouse IgG was used as the secondary antibody. To visualize the distribution of ␣B-crystallin, cells were incubated overnight at 4°C in a 1:100 dilution of a polyclonal antibody raised against bovine ␣B-crystallin (a gift from Dr. Joseph Horwitz). This primary antibody was used, because it gave a very low background in immunocytochemistry with the ␣B-crystallinϪ/Ϫ mouse lens slices. An Alexa 488 -conjugated goat anti-rabbit IgG was used as the secondary antibody. To visualize the organization of F-actin, fixed and permeabilized cells were incubated in a 1:50 dilution in PBS of a fluorescein phalloidin (or Texas Red phalloidin, Molecular Probes Inc.) methanolic stock solution (100 units/ml of methanol). Cells were stained with fluorescein phalloidin for 20 min, washed 3 ϫ 5 min in PBS, and viewed. Lens epithelial cells were viewed using a Zeiss LSM 410 confocal microscope equipped with an argon-krypton laser.

RESULTS
Human Lens Epithelial Cells-The expression of ␣A-crystallin in lens epithelial cells has been shown to prevent UVA radiation-induced apoptotic cell death. We examined whether ␣A-crystallin expression also protects cells from various other inducers of apoptotic death. HLE B-3 cells were transfected with pCIneo␣A vector carrying human ␣A-crystallin cDNA or mock transfected with the vector alone. Mock and ␣A-expressing cells were treated with staurosporine (0.5 M), incubated for 2 h, and labeled with annexin and propidium iodide (PI). The number of cells in the annexin-positive/PI-negative, annexin-positive/PI-positive, annexin-negative/PI-positive, and annexin-negative/PI-negative channels was determined. Cells in the early stages of apoptosis, as identified by their annexin labeling, are shown in the lower right quadrants of Fig. 1. Cells incubated in culture medium alone had very few (Ͻ5%) cells in the early stages of apoptosis (Fig. 1, A and B), and the ␣Aexpressing cells had a lower basal level of apoptosis than the mock cells. Expression of ␣A-crystallin significantly diminished staurosporin-induced annexin labeling. There were 43% annexin-positive/PI-negative cells in the mock clone as compared with 5% in the ␣A-expressing clone after staurosporine treatment. Cells that had already died by apoptosis (annexinpositive/PI-positive, upper right quadrant) also decreased in staurosporine-treated ␣A-expressing cells as compared with mock cells (Fig. 1, C and D). Concurrently, unlabeled cells representing the live population (lower left quadrant) decreased only slightly in the ␣A-expressing cells (from 82% to 75%) but decreased dramatically (from 75% to 18%) in the mock cells. These observations indicate that the expression of ␣Acrystallin protected HLE B-3 cells from apoptotic cell death induced by staurosporine.
Mock transfected and ␣A-expressing HLE B-3 cells were also treated with TNF␣ or Fas monoclonal antibody and labeled with annexin and propidium iodide. Annexin labeling was 18% in the TNF␣-treated mock cell line as compared with 3% in the ␣A-crystallin-expressing clone (data not shown). ␣A-crystallin expression also decreased annexin labeling in Fas monoclonal antibody-treated cells (36% annexin-labeling for mock cells and 5% for ␣A-expressing cells, data not shown). These results indicate that, by introducing ␣A-crystallin in cells, apoptosis in response to a variety of agents (as measured by annexin labeling) decreased significantly.
We next analyzed the protective effects of expression of different levels of ␣A-crystallin in stably transfected HLE B-3 cell lines exposed to stress-inducing agents. The relative expression of ␣A-crystallin in transfected cell lines was determined by immunoblot analysis (9) and immunofluorescence. As shown in Fig. 2 (D-F), the individual cells of a given culture expressed similar amounts of the protein, giving a measure of homogene-ity in a culture. Fig. 2 (A-C) shows the percentage of protection from cell death induced by staurosporine, TNF␣, or anti-Fas in transfected HLE B-3 cell lines and indicates that the relative protection against cell death increased with an increase in expression level of ␣A-crystallin under all three conditions. This suggests that ␣A-crystallin protected against apoptosis induced by a variety of pathways. ␣A-crystallin may be exerting its protective effect either by activating anti-apoptotic factors (such as Bcl-2), or by preventing the activation of proapoptotic factors (such as caspases).
The HLE B-3 lens epithelial cell line used in this study expressed endogenous levels of the small heat shock proteins, ␣B-crystallin and HSP27, as detected by immunoblotting (Fig.  3). Thermal stress-induced enhancement of expression of these proteins has been demonstrated in a number of non-lens cell lines (5, 6), but it is not known whether HLE B-3 cells are capable of increasing the expression of small heat shock proteins. We investigated the effect of mild thermal stress on the expression of ␣B-crystallin and HSP27. Exposure of HLE B-3 cells to 43°C for 1 h, followed by periods of recovery at 37°C ranging from 1 to 24 h, led to a small increase in HSP27 expression 1 and 4 h after heat shock. HSP27 expression was enhanced considerably between 8 and 18 h after thermal stress (Fig. 3, C and D). In contrast, ␣B concentration decreased immediately after heat shock and remained lower than control for up to 1 h before increasing to levels above control between 4 and 18 h after thermal stress (Fig. 3, A and B). These results indicate that, as has been reported in other cells, the transformed HLE B-3 lens epithelial cells have the capacity to enhance their expression of heat shock proteins upon exposure to thermal stress. ␣B-crystallin and HSP27 accumulated at prolonged times after thermal stress. Only ␣B-crystallin decreased immediately after stress.
Mouse Lens Epithelial Cells-Primary cultures of ␣B-crys-tallinϪ/Ϫ mouse lens epithelial cells were compared with wildtype primary cells in their response to UVA radiation-induced cell death. There was no difference between the susceptibility of ␣B-crystallinϪ/Ϫ and wild-type cells to UVA-induced TUNEL labeling (Fig. 4A). This observation is in striking contrast to the results we reported previously with ␣A-crystal-linϪ/Ϫ lens epithelial cells, which showed a significantly higher level of apoptosis by UVA radiation (9), suggesting that ␣A-crystallin is more effective in preventing cell death by UVA radiation than is ␣B-crystallin. Furthermore, when the double knockout primary cells were exposed to UVA radiation,

FIG. 3. Western blot analysis of ␣Bcrystallin and HSP27 expression in HLE B-3 cells.
A, ␣B-crystallin immunoblot. Cells were exposed to heat shock at 43°C for 1 h and then incubated for recovery at 37°C. Lane 1, control cells; lanes 2-6, cells exposed to 43°C for 1 h and then incubated at 37°C for recovery times of 0, 1, 4, 8, and 18 h, respectively. B, -fold increase in ␣B-crystallin expression at various times of recovery after heat shock (quantified by the Storm phosphorimaging system). The data are representative of three experiments. C, HSP27 immunoblot. Lane 1, control cells; lane 2-6, cells exposed to 43°C for 1 h and then incubated at 37°C for recovery times of 0, 1, 4, 8, and 18 h, respectively. D, -fold increase in HSP27 expression at various times of recovery after heat shock. The data are representative of three experiments. TUNEL-labeled, and analyzed for extent of labeling, the proportion of TUNEL-labeled cells was similar to that in ␣Acrystallin knockout cells reported previously. Together, these observations indicate that knocking out ␣A-crystallin increased UVA-induced apoptotic cell death, whereas knocking out ␣B-crystallin did not have a significant effect.
To determine whether primary cultures of wild-type mouse lens epithelial cells expressed a uniform level of ␣Aor ␣Bcrystallin, we examined the expression of the proteins in wildtype lens epithelial cultures by immunofluorescence. As shown in Fig. 4B, the level of expression of ␣A-crystallin was not uniform in individual cells of a primary culture, with some cells expressing a low level and others expressing higher levels of ␣A-crystallin. Similar results were obtained when ␣B-crystallin expression was examined (data not shown).
To investigate the relative protective abilities of ␣Aand ␣B-crystallin in more detail, we reasoned that a cell line lacking both ␣Aand ␣B-crystallin expression would be useful for transfection with either pCIneo␣A or pCIneo␣B, to probe the effect of expression of each protein against a null background. To increase the growth potential of the mouse lens epithelial cultures, and enable us to transfect them readily, primary mouse lens epithelial cells of wild-type, ␣A-crystallin knockout, ␣B-crystallin knockout, and ␣A/␣B-crystallin double knockout mice were infected with Ad12-SV40 hybrid virus and propagated to make cell lines (Fig. 5A). The expression of ␣Aand ␣B-crystallins in transformed cell lines created from wild-type, ␣A-crystallin knockout, ␣B-crystallin knockout, and ␣A/␣B double knockout cells was analyzed by immunoblotting (Fig.  5B). As expected, the transformed cell line from the double

FIG. 4. Effect of UVA exposure and subsequent incubation on the proportion of TUNEL-positive cells in primary cultures of wild-type, ␣B؊/؊ and ␣A؊/؊␣B؊/؊ mouse lens epithelial cells.
A, primary lens epithelial cell cultures were exposed to 37.8 J/cm 2 UVA radiation (365 nm) and incubated in normal medium for the indicated times. Cells were TUNEL-labeled and stained with DAPI to enable a count of the total number of cells in each microscopic field. The number of TUNEL-positive cells and DAPI-stained cells were counted in 10 randomly selected fields from three or more cultures at each time point.
The asterisks indicate that the difference between TUNEL labeling was statistically significant for UVA-treated ␣A/␣B-crystallin double knockout cells and wild-type cells. B, immunofluorescence of ␣A-crystallin and F-actin in primary cultures of wild-type mouse lens epithelial cells as detected by confocal microscopy. Red, ␣A-crystallin; green, F-actin. Note that the expression of ␣A-crystallin was dissimilar in different cells of a given culture. knockout mouse lens (␣A␣BKO1) was shown to lack the expression of both ␣Aand ␣B-crystallin proteins. This cell line was then used for transfection with the pCIneo␣A or pCIneo␣B containing human ␣Aor ␣B-crystallin cDNA, and clones expressing different levels of ␣Aor ␣B-crystallin were isolated (Fig. 6). Cells were exposed to staurosporine (0.5 M) for 2 h, labeled with annexin, and analyzed by FACS analysis. Cells expressing 0.42 ng of ␣A-crystallin/g of cellular protein decreased annexin labeling by 25% as compared with mock transfected cells (Fig. 7A). In contrast, cells expressing 0.35 ng of ␣B-crystallin had the same level of annexin labeling as mock transfected cells. Cells expressing 0.90 ng of ␣A-crystallin/g of cellular protein reduced annexin labeling to background levels. In contrast, cells expressing the same level of ␣B-crystallin did not prevent staurosporine-induced annexin labeling. However, cells expressing 2.2 ng of ␣B-crystallin were effective in decreasing annexin labeling to background levels. This finding indicates that even a small amount of ␣A-crystallin expression was protective against cell death. Under the conditions tested, the ability of low concentrations of ␣A-crystallin to prevent apoptotic cell death was significantly higher than that of ␣Bcrystallin. At higher concentration, ␣B-crystallin provided the same protective ability as did ␣A-crystallin.
Additional experiments were carried out to compare the relative ability of ␣Aand ␣B-crystallin to protect lens epithelial cells from stress. Mouse lens epithelial cell clones expressing ␣Aor ␣B-crystallin were also exposed to UVA radiation. As shown in Fig. 7B, ␣A-crystallin decreased annexin labeling of UVA-treated cells at lower concentrations than ␣B-crystallin. However, some differences were noted between the level of protection by ␣B-crystallin between the two methods used to induce apoptosis. ␣B-crystallin provided a greater protection against UVA stress than it did against staurosporine-induced apoptosis. This may be attributed to different mechanisms of apoptosis in UVA radiation and staurosporine-induced cell death. DISCUSSION ␣Aand ␣B-crystallins are synthesized as major components of all vertebrate lenses (5). Although both subunits are highly expressed in lens fiber cells, they are also readily detected in lens epithelial cells (10). Because the epithelial cells represent the most metabolically active region of the lens, we focused on cultured lens epithelial cells derived from ␣A-, ␣B-, or ␣A-/␣B-crystallin knockout mice. A recent study of lens epithelial cells derived from ␣B-crystallin knockout mice indicated that the loss of ␣B-crystallin predisposes lens epithelial cells to genomic instability (12), which may explain its broad tissue distribution as well as its overexpression in growing tissues, in diseased states, and under stress conditions (3-5, 26 -28). In contrast, lens epithelial cells from ␣A-crystallin knockout mice were reportedly delayed in growth and were more sensitive to UVA radiation (9). The present results provide new insight into the distinctive cellular functions of ␣Aand ␣B-crystallins under physiologically relevant conditions.
One of the main findings of this study is that ␣A-crystallin expression in lens epithelial cells imparts a greater resistance to various apoptotic inducers than the expression of ␣B-crystallin. At a given expression level, the anti-apoptotic activity of ␣Awas significantly higher than that of ␣B-crystallin. Moreover, ␣A-crystallin, which has been shown previously to confer resistance to thermal stress and UVA radiation (8,9), was also found to prevent apoptosis induced by TNF␣, staurosporine, and anti-Fas. Lens epithelial cells are proximal to sources of stress and our results suggest that an important in vivo function of ␣A-crystallin may be to protect these cells from stressinduced apoptosis. Although the mechanism by which ␣A-crystallin exerts this protective effect is unknown, it may indirectly affect either the activation of anti-apoptotic factors, such as Bcl-2, or prevent the activation of pro-apoptotic factors, such as caspases (19).
Previous work on prevention of stress-induced apoptosis by sHSPs has been restricted to ␣B-crystallin and HSP25/27 (14 -16). Because ␣A-crystallin expression is largely restricted to the lens tissue, and because lens epithelial cells are continually exposed to sources of environmental stress, we sought to determine the relative protective activity of ␣A-crystallin and other sHSPs. When primary cultures of wild-type or ␣B-crystallin knockout mouse lens epithelial cells were irradiated with physiological levels of UVA radiation, no significant increase in apoptosis occurred during the 24-h post-irradiation period. After UVA exposure, the number of apoptotic cells was indistinguishable between wild-type and ␣B-crystallin knockout cells, suggesting that loss of ␣B-crystallin did not increase the incidence of apoptosis. In striking contrast, our previous study with UVA-treated ␣A-crystallin knockout lens epithelial cells had demonstrated that the loss of ␣A-crystallin resulted in a 40-fold increase in apoptosis (9). We also exposed ␣A/␣B double knockout lens epithelial cultures to UVA radiation. In these double knockout cells, a 40-fold increase in apoptotic cells was observed following UVA radiation. Taken together, the results obtained with primary cultures of ␣A knockout, ␣B knockout, and ␣A/␣B double knockout lens epithelial cells indicate that, in the presence or absence of ␣A-crystallin, the loss of ␣Bcrystallin does not increase apoptosis. This is consistent with the notion that the anti-apoptotic activity of ␣A-crystallin is higher than that of ␣B-crystallin. This observation also suggests that HSP25/27 present in lens epithelial cells contributes to maintaining their protection at a certain level.
␣Aand ␣B-crystallin have pleomorphic cellular functions, including the ability to bind cytoskeletal elements (29,30), bind membranes (24), and translocate to the nucleus (31), although the proteins that interact with ␣Aand ␣B-crystallin in the lens epithelium in vivo are unknown. One of the unresolved question in lens research is why are both ␣Aand ␣B-crystallin needed in the lens (2)? Recently, it has been shown that a targeted disruption of a mouse ␣A-crystallin gene induces cataract (20). Interestingly, dense inclusion bodies consisting mainly of ␣B-crystallin were found in the lens fiber cells of these mice suggesting that ␣B-crystallin is unstable at high concentrations in the lens in the absence of its aggregation partner, ␣A-crystallin. Other experiments suggest that mixing of ␣Aand ␣B-crystallin increases the stability of the system (2). Although ␣B-crystallin is widely distributed in tissues such as muscle, heart, brain, lung, and kidney, ␣A-crystallin is largely restricted to the lens, with low levels detected in the thymus and spleen (3)(4)(5). This distinct tissue distribution and the stress inducibility of only ␣Band not ␣A-crystallin supports the notion that the two proteins may have unique cellular functions in the lens.
Recent in vitro studies demonstrate that the chaperone-like activity of ␣-crystallins is sensitive to substrate, buffer conditions, and temperature (32)(33)(34)(35)(36). Although the activity of ␣Bcrystallin is relatively insensitive to temperature, a marked temperature dependence of chaperone activity of ␣A-crystallin has been reported (32)(33)(34)(35)(36). Using several protein substrates, ␣Aand ␣B-crystallin were found to have similar chaperone activities at physiological temperatures (ϳ37°C) when inhibition of lactalbumin aggregation was measured, whereas the chaperone-like activity of ␣B-crystallin was greater than that of ␣A-crystallin when the substrate was alcohol dehydrogenase or insulin (33). Another factor that may contribute to their different activities is the surface hydrophobicity of the ␣Aand ␣B-crystallins (32)(33)(34)(35)(36). However, neither the differences in vitro chaperone activity toward selected substrate proteins, nor the different surface hydrophobicity of ␣Aand ␣B-crystallin are sufficient to explain the greater anti-apoptotic activity of ␣Aover that of ␣B-crystallin found in the present study (32)(33)(34)(35)(36). These findings provide a basis for future studies on identification of in vivo substrates of ␣Aand ␣B-crystallin and other factors that determine their in vivo protective activity.
The assessment of relative anti-apoptotic activities of ␣Aand ␣B-crystallins in mouse lens epithelial cells using annexin labeling was further supported by our findings in human lens epithelial cells. The extended-life span HLE B-3 cells used for transfection studies did not express endogenous ␣A-crystallin after 11 passages but continued to express low endogenous levels of ␣B-crystallin (25) and were able to accumulate ␣Bcrystallin and HSP27 in response to thermal stress (Fig. 3). The ␣A-crystallin-expressing stably transfected human lens epithelial cell lines used in this study were significantly more resistant to staurosporine, TNF␣, and an antibody to Fas, than were the mock transfected cells, consistent with a previous report of the enhanced resistance of ␣A-crystallin expressing lens epithelial cells to UVA radiation (9). Furthermore, the protective effect of ␣A-crystallin increased with its level of expression. A factor that may influence the protective activity of ␣A-crystal-lin in the presence of ␣B-crystallin in HLE B-3 cells is the stoichiometry of the ␣A/␣B aggregate (2). It has been shown that the in vitro stability and the chaperone-like activity of the 3:1 ␣A/␣B heteroaggregate is substantially greater than that of aggregates containing higher or lower ratios of the two subunits or of homopolymers containing either ␣Aor ␣B-crystallin (2,37). Quantitative immunoblot analysis of HLE B-3 cells (data not shown) indicated that the stoichiometric composition of the ␣A/␣B-crystallin expression in the stably transfected HLE B-3 cells containing Ͼ1 ng of ␣A-crystallin/g of cellular protein is approximately 3:1. Such heteroaggregates were presumably present in ␣A-crystallin-expressing stably transfected HLE B-3 cells, which express endogenous ␣B-crystallin. The present results indicate that the expression of ␣A-crystallin, on a background of low ␣B-crystallin, is helpful in enhancing resistance of lens epithelial cells to apoptosis induced by a variety of agents and supports the idea that ␣A-crystallin is an important anti-apoptotic protein in lens epithelial cells.
In summary, the present studies indicate that ␣A-crystallin has a higher capacity than ␣B-crystallin to prevent lens epithelial cell death. This may indicate that the threshold concentration of ␣A-crystallin necessary for protection against stressinduced apoptosis in the lens epithelium is less than that of ␣B-crystallin. Because their expression in the lens epithelium is at much lower levels than in the fiber cells, it remains to be determined whether the 2-to 3-fold difference in protective activity is of significance for their function in the in vivo lens epithelium. The current work suggests that comparing the annexin labeling of lens epithelial cells is an effective way to measure the protective activity of ␣Aand ␣B-crystallin under physiologically relevant conditions. Taken together with our previous study (12), which demonstrated that the lack of ␣Bcrystallin predisposes cultured lens epithelial cells to genomic instability, these studies provide a basis for further investigating the distinct physiological functions of ␣Aand ␣B-crystallins in lens epithelial function.