Human bcl-2 Gene Attenuates the Ability of Rabbit Lens Epithelial Cells against H2O2-induced Apoptosis through Down-regulation of the αB-crystallin Gene*

It is well established that the proto-oncogene, bcl-2, can prevent apoptosis induced by a variety of factors. Regarding the mechanism by which BCL-2 prevents cell death, one theory suggests that it acts by protecting cells from oxidative stress. In the lens system, oxidative stress-induced apoptosis is implicated in cataractogenesis. To explore the possibility of anti-apoptotic gene therapy development for cataract prevention and also to further test the anti-oxidative stress theory of BCL-2 action, we have introduced the human bcl-2 gene into an immortalized rabbit lens epithelial cell line, N/N1003A. The stable expression clones of both vector- and bcl-2-transfected cells have been established. Treatment of the two cell lines with H2O2 revealed thatbcl-2-transfected cells were less capable of detoxifying H2O2 than the control cells. Moreover,bcl-2-transfected cells are more susceptible to H2O2-induced apoptosis. To explore whybcl-2-transfected cells have reduced resistance to H2O2-induced apoptosis, we examined the expression patterns of several relevant genes and found that expression of the αB-crystallin gene was distinctly down-regulated inbcl-2-transfected cells compared with that in vector-transfected cells. This down-regulation was specific because a substantial inhibition of BCL-2 expression through antisensebcl-2 RNA significantly restored the level of αB-crystallin and, moreover, enhanced the ability of thebcl-2-transfected cells against H2O2-induced apoptosis. Introduction of a mouse αB-crystallin gene into bcl-2-transfected cells also counteracted the BCL-2 effects. Down-regulation of αB-crystallin gene was largely derived from changed lens epithelial cell-derived growth factor activity. Besides, αB-crystallin prevents apoptosis through interaction with procaspase-3 and partially processed procaspase-3 to prevent caspase-3 activation. Together, our results reveal that BCL-2 can regulate gene expression in rabbit lens epithelial cells. Through down-regulation of the αB-crystallin gene, BCL-2 attenuates the ability of rabbit lens epithelial cells against H2O2-induced apoptosis.

BCL-2 was initially found to prevent interleukin-3-dependent cells from apoptotic death upon withdrawal of the cytokine (7). Since then, BCL-2 has been shown to prevent cell death induced by a large number of factors such as calcium ionophore, serum and growth factor depletion, and ␥-irradiation (reviewed in Refs. 8 -10).
Regarding the mechanism by which BCL-2 prevents cell death, one theory suggests that it acts by protecting cells from oxidative stress (11,12). BCL-2 could either reduce cellular generation of reactive oxygen compounds or block the activity of these compounds after they are formed (11,12). Supportive evidence for this theory comes from the finding that in certain cell lines, bcl-2-transfected cells show greater resistance to various pro-oxidant treatment than mock-transfected cells (13)(14)(15), and anti-oxidants protect some cells from apoptosis induced by non-oxidative agents (11, 16 -18).
However, in other cell lines, expression of BCL-2 does not protect the transfected cells against oxidative stress and oxidative stress-induced apoptosis (19,20). Why BCL-2 does not protect against oxidative stress-induced apoptosis in these cells remains largely unknown. Furthermore, it is also reported that oxygen depletion has no effect on the induction of apoptosis and that BCL-2 protects against apoptosis without inhibiting the production or activity of reactive oxygen compounds (21,22). Thus, depending on the types of cells and also the intracellular metabolic status, BCL-2 may provide protection through different mechanisms.
To study the mechanism by which BCL-2 prevents stressinduced apoptosis in the lens system, we have introduced the human bcl-2 gene into rabbit lens epithelial cells, N/N1003A (23). By using the established stable expression lines, pSFFV-N/N1003A (vector-transfected) and pSFFV-BCL-2-N/N1003A (bcl-2-transfected), we found that the bcl-2-transfected cells had an attenuated ability to metabolize H 2 O 2 and to resist H 2 O 2 -induced apoptosis compared with the vector-transfected cells. To understand this attenuation, we have examined expression of several relevant genes in these two types of cells. Whereas expression of the anti-oxidative stress genes was hardly changed, expression of the endogenous ␣B-crystallin gene was significantly down-regulated. This down-regulation is specific as demonstrated with antisense inhibition of BCL-2 expression. When BCL-2 expression is substantially inhibited through antisense bcl-2 RNA, the endogenous ␣B-crystallin in these double-transfected cells is significantly restored. Restoration of ␣B-crystallin expression enhanced the ability of the cells to resist H 2 O 2 -induced apoptosis. Moreover, an exogenous mouse ␣B-crystallin gene introduced into bcl-2-transfected cells also counteracted the BCL-2 effects. To understand how BCL-2 may down-regulate expression of ␣B-crystallin gene, we have examined the DNA binding activity of LEDGF, 1 a positive regulator of ␣B-crystallin (24,25), in vector-and bcl-2-transfected cells. Our results revealed that the DNA binding activity of LEDGF was also substantially down-regulated in BCL-2 expression cells. Moreover, overexpression of LEDGF in BCL-2 expression cells substantially up-regulated the level of ␣Bcrystallin. To understand the mechanism why ␣B-crystallin prevents apoptosis, we have conducted immunoprecipitationlinked Western blot analysis. Our results revealed that ␣Bcrystallin prevents induced apoptosis through interaction with both procaspase-3 and partially processed procaspase-3. Taken together, our results demonstrate that BCL-2 can down-regulate expression of the ␣B-crystallin gene in rabbit lens epithelial cells through modulating transactivity of LEDGF and possibly other transcription factors. Through down-regulated expression of ␣B-crystallin, which prevents apoptosis by preventing caspase-3 activation, BCL-2 attenuates the ability of N/N1003A cells to resist oxidative stress-induced apoptosis. Thus, our results reveal a unique mechanism explaining why BCL-2 is unable to prevent oxidative stress-induced apoptosis in these cells.

EXPERIMENTAL PROCEDURES
Chemicals-Various molecular biology reagents were purchased from Life Technologies, Inc.; Stratagene, La Jolla, CA; New England Biolabs, Beverly, MA; and Promega Biotech, Madison, WI. DNA and protein size markers were purchased from Life Technologies, Inc. Various antibodies were obtained from Roche Molecular Biochemicals and Transduction Laboratories, San Diego, CA. Radioactive compounds were obtained from Amersham Pharmacia Biotech. The culture medium and most other chemicals and antibiotics were purchased from Sigma and Life Technologies, Inc.
Cell Culture-The rabbit lens epithelial cells, N/N1003A, were grown in Eagle's minimum essential medium (Sigma catalog number M0643) containing 10% rabbit serum (Sigma catalog number R4505) as described previously (23). The medium was prepared in ion-exchanged double-distilled water to give an osmolarity of 300 Ϯ 5 mosmol supplemented with 26 mM NaHCO 3 and 50 g/ml gentamicin sulfate (Sigma catalog number G1397). Media and sera were sterilized by filtration through 0.22-m filters (Corning Glass catalog number 25942) with pH adjusted to 7.2. All cells were kept at 37°C and 5% CO 2 gas phase.
Analysis of H 2 O 2 Degradation-Various cell lines (see figure legends for details) were grown in MEM containing 10% rabbit serum with (transfected cell lines) or without (parental cell line) selection drugs until confluence. Then 3 ϫ 10 5 cells were plated into a 60-mm culture dish. After 12 h of growth, the media in different cultures were replaced with 10 ml of serum-free MEM containing 150 or 350 M H 2 O 2 (the concentration at the starting point). The treatment was continued for 90 min, and the H 2 O 2 concentration in the medium of different cultures was determined at 15, 30, 45, 60, and 90 min using a chemical method as described previously (31). The final results presented were averaged from four independent experiments. The standard deviation was shown in each figure.
Analysis of Induced Apoptosis-Various cell lines were grown in MEM containing 10% rabbit serum with (transfected cell lines) or without (parental cell line) selection drugs until confluence. Then 3 ϫ 10 5 cells were plated into a 60-mm culture dish. After 12 h of growth, the media in different cultures were replaced with 10 ml of MEM containing 1% serum, 150 M H 2 O 2 (the concentration at the starting point), 25 nM staurosporine, or 10 M camptothecin for 6 -18 h (see related figure legends for detail). The percentage of apoptotic cells and the apoptotic nature of cell death were determined using trypan blue exclusion, Hoechst staining, and DNA fragmentation as described previously (32). The quantitative results of apoptosis presented were averaged from three independent experiments. The standard deviation was shown in each figure.
Assay of Caspase-3 Activity-The caspase-3 activity in various cell lines after treatment by H 2 O 2 was assayed as described previously (30). The final results presented were averaged from three independent experiments. The standard deviation was shown in each figure.
RNA Preparation and Northern Blot-Total RNAs were extracted from N/N1003A and various stable expression cell lines as described previously (32) using an RNA extraction buffer, Trizol reagent (Life Technologies, Inc., catalog number 15596-026). For Northern blot, 25 g of total RNAs were used for each sample. Other procedures such as gel electrophoresis, pre-hybridization, hybridization, washing, and exposure were conducted as described previously (30 -32).
Protein Preparation and Western Blot-The total proteins were prepared from N/N1003A or various stable cell lines using 300 l of extraction buffer. The extraction buffer contained 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 9.1 mM Na 2 HPO 4 , 1.7 mM NaH 2 PO 4 , 150 mM NaCl, 10 l/ml PMSF stock solution (10 mg/ml in isopropyl alcohol), 30 l/ml aprotinin with pH of the preparation adjusted to 7.4. After homogenization by passing through a 21-gauge needle, additional 10 l of PMSF was added to each sample, which was incubated on ice for 30 min. After the cell lysate was centrifuged at 10,000 ϫ g for 20 min at 4°C, the supernatant of each sample was collected and stored in aliquots at Ϫ70°C. For each sample, the protein concentration was determined according to Peterson (38). Fifty micrograms of total proteins in each sample were resolved by 6 (for LEDGF), 10 (for GFP), or 12% (for ␣B-crystallin) SDS-polyacrylamide gel electrophoresis and transferred into supported nitrocellulose membranes. The protein blots were blocked with 5% milk in TBS (10 mM Tris⅐HCl, pH 8.0, 150 mM NaCl) overnight at 4°C and incubated with anti-human BCL-2 antibody or anti-GFP antibody (both from Roche Molecular Biochemicals), anti-caspase-3 antibody (Transduction Laboratory), anti-␣B-crystallin antibody (a kind gift from Dr. Joseph Horwitz), or anti-LEDGF antibody (a kind gift from Dr. Toshimichi Shinobara) at a dilution of 1 to 1000 to 2000 (g/ml) in 5% milk prepared in TBS. The secondary antibody was anti-mouse IgG (for anti-BCL-2, GFP, and caspase-3 antibodies) or anti-rabbit IgG (for anti-␣B-crystallin and anti-LEDGF antibodies) at a dilution of 1 to 1000 (Amersham Pharmacia Biotech). Immunoreactivity was detected with an enhanced chemiluminescence detection kit according to the company's instructions (ECL, Amersham Pharmacia Biotech).
Immunoprecipitation-linked Western Blot Analysis-Parental BCL-2 expression cells (pSFFV-Bcl-2-N/N1003), vector-transfected BCL-2 expression cells (pEGFP/pSFFV-Bcl-2-N/N1003A), and mouse ␣B-crystallin-transfected BCL-2 expression cells (pEGFP-m␣B/pSFFV-Bcl-2-N/ N1003A) were grown and treated with 150 M H 2 O 2 for 12 h as described above. At the end of treatment, the cells were harvested for extraction of total proteins. These protein samples were either used for Western blot analysis as described above or for immunoprecipitationlinked Western blot analysis. To conduct immunoprecipitation-linked Western blot analysis, 500 g of total proteins from vector-and mouse ␣B-crystallin-transfected BCL-2 expression cells were incubated with 10 g (in 50 l) of anti-GFP antibody or 50 l of normal mouse serum and 50 l of protease inhibitor mixture for 1 h on ice. After incubation, 50 l of protein A/G plus-agarose was added into each incubated sample. These samples were then incubated overnight in a 4°C refrigerator attached to a slow motion rotator. At the end of incubation, these samples were washed four times with RIPA buffer (1ϫ PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) by spinning down 5 min at 10,000 ϫ g. After the final wash, the pelleted samples were subjected to Western blot analysis as described above using specific anti-␣B-crystallin or anti-caspase-3 antibodies.
Analysis of Transient Gene Expression-For reporter gene activity, the construct of chloramphenicol acetyltransferase (CAT) reporter gene driven by the mouse ␣B-crystallin gene promoter (Ϫ427 to ϩ44 (39)), together with the control construct expressing ␤-galactosidase (40, both were kind gifts from Dr. Joram Piatigorsky), was introduced into both vector-and BCL-2-transfected cells using electroporation (30). The transfected cells were grown in 100-mm culture dishes and then harvested after 24 h of growth for assays of ␤-galactosidase and CAT activities as described previously (29,30).
For LEDGF-regulated ␣B-crystallin expression, the construct, pCI-GST-LEDGF (the LEDGF cDNA was a kind gift from Dr. Toshimichi Shinahara), and the vector, pCI, were introduced into BCL-2 expression cells using electroporation (30). The transfected cells were grown in 100-mm culture dishes and then harvested after 48 h growth for Western blot analysis of LEDGF and ␣B-crystallin expression.
Gel Mobility Shifting Assays-For gel mobility shifting assays, the following oligonucleotides were used: 5Ј-AAATATTTGGGGTTTTTTT-T-3Ј for LEDGF-binding site and 5Ј-AAATATTAAAAAATTTTTTT-3Ј for mutated LEDGF-binding site. Forty g of nuclear extracts prepared from pSFFV-N/N1003A and pSFFV-Bcl-2-N/N1003A cells were incubated with 1 ϫ 10 5 cpm of 32 P-labeled double-stranded synthetic oligonucleotides for 30 min at 30°C in a binding shifting buffer (1 g/ml poly(dI-dC), 25 mM HEPES, pH 7.9, 40 mM KCl, 3 mM MgCl 2 , 0.1 mM EDTA, 1 mM DTT, and 10% glycerol). For competition experiments, 50-fold of the unlabeled double-stranded synthetic oligonucleotides were preincubated with the nuclear extracts for 10 min before the labeled probe was added into the reaction. In the pre-cleared experiments, 2 g of antibody against LEDGF was pre-incubated with the nuclear extracts on ice for 10 min prior to addition of the 32 P-labeled oligonucleotides. After the binding reaction, the mixtures were loaded onto 5% native polyacrylamide gel electrophoresis and detected by autoradiography. BCL-2 expression construct (27) were introduced into the N/N1003A cells using electroporation (30). The stable expression clones of pSFFV-N/N1003A (vector-transfected) and pS-FFV-Bcl-2-N/N1003A (bcl-2-transfected) were obtained by G418 selection (400 g/ml). Following growth to confluence, both pSFFV-N/N1003A and pSFFV-Bcl-2-N/N1003A cells were harvested for extraction of total RNA and protein. Northern and Western blot analyses were conducted to confirm expression of the human bcl-2 in rabbit lens epithelial cells. As shown in Fig. 1, the mRNA and protein from the human bcl-2 gene were detectable only in the pSFFV-Bcl-2-N/N1003A cells but not in the pSFFV-N/N1003A cells. Thus, human BCL-2 was successfully expressed in rabbit lens epithelial cells.

Expression of Human bcl-2 in Rabbit Lens
Expression of Human BCL-2 in Rabbit Lens Epithelial Cells Attenuates the Ability of the Transfected Cells to Metabolize H 2 O 2 -After BCL-2 was successfully expressed in the rabbit lens epithelial cells, we next tested whether the expressed protein was functional. Previous studies (11)(12)(13)(14)(15) have suggested that BCL-2 can protect cells from induced apoptosis through reduction of cellular oxidative stress. Whether this is true in lens epithelial cells remains to be verified. To test this possibility, we exposed N/N1003A, pSFFV-N/N1003A, and pS-FFV-Bcl-2-N/N1003A cells to 150 and 350 M H 2 O 2 . Degradation of H 2 O 2 in the culture dishes containing either N/N1003A, pSFFV-N/N1003A, or pSFFV-Bcl-2-N/N1003A cells or containing medium alone was monitored within 90 min using a chemical method as previously described (31). Our results revealed that the pSFFV-Bcl-2-N/N1003A cells were less capable of me-tabolizing H 2 O 2 than the pSFFV-N/N1003A cells. As shown in Fig A, viability assays. N/N1003A cells, pSFFV-N/N1003A, and pSFFV-Bcl-2-N/ N1003A stable expression clones were cultured in MEM with 10% rabbit serum until confluence. Then 3 ϫ 10 5 cells were plated into 60-mm culture dish. After 12 h of growth, the media in three different cultures were replaced with 1% serum MEM containing 150 M H 2 O 2 (the concentration at the starting point). The treatment was continued for 18 h. Viability of the three different cultures was determined as described previously (32). B-D, Hoechst staining of parental (B), vector-transfected (C), and bcl-2-transfected (D) cells after 6 h of treatment with hydrogen peroxide. Hoechst staining was conducted as previously described (32). Apoptotic cells either had fragmented nuclei (arrows) or were detached from the culture dish so that empty space was left. Bar, 10 m.
pSFFV-Bcl-2-N/N1003A Cells Are Resistant to Apoptosis Induced by Staurosporine and Camptothecin-One possible explanation for the lack of protection of BCL-2 on pSFFV-Bcl-2-N/N1003A cells from H 2 O 2 -induced apoptosis could be that the BCL-2 expressed in rabbit lens epithelial cells is not functional. Because BCL-2 activity is defined by its ability to inhibit apoptosis, we tested this possibility by subjecting pSFFV-Bcl-2-N/ N1003A cells to staurosporine and camptothecin treatments that have been shown to induce apoptosis in different cells (41)(42)(43)(44). When pSFFV-Bcl-2-N/N1003A, pSFFV-N/N1003A, and the parental N/N1003A cells were treated with 25 nM staurosporine or 10 M camptothecin, the viability assay, Hoechst staining, and DNA fragmentation revealed that pSFFV-Bcl-2-N/N1003A cells are much more resistant to apoptosis induced by these two reagents. As shown in Fig. 4, A and  B, after a 6-h treatment by staurosporine, only 4% of pSFFV-Bcl-2-N/N1003A cells were apoptotic, whereas 19% of N/N1003A or pSFFV-N/N1003A cells were undergoing apopto-sis or were detached from the culture dish after death. Similar results were observed following treatment with camptothecin. After a 6-h treatment, there were about 6% apoptotic cells in pSFFV-Bcl-2-N/N1003A culture dish and 25% apoptosis in N/N1003A and pSFFV-N/N1003A cells (Fig. 4, C and D). Thus, BCL-2 expressed in rabbit lens epithelial cells can prevent apoptosis induced by non-oxidative stress factors.
Expression of the Genes Encoding Anti-oxidative Stress Enzymes Is Not Changed in pSFFV-Bcl-2-N/N1003A, pSFFV-N/ N1003A, and N/N1003A Cells-Another possible explanation for the lack of protection of pSFFV-Bcl-2-N/N1003A cells from H 2 O 2 -induced apoptosis could be that the expressed BCL-2 in rabbit lens epithelial cells affects the expression of the antioxidative stress genes. To test this possibility, total RNAs were prepared from pSFFV-Bcl-2-N/N1003A, pSFFV-N/N1003A, and N/N1003A cells. An equal amount of total RNAs (25 g per sample) from these cell lines was resolved with a 1.2% formaldehyde-agarose gel and transferred to supported nitrocellulose filters. The RNA blot was sequentially hybridized to the following [␣-32 P]dATP-labeled cDNA probes: catalase, glutathione peroxidase, ␤-actin, and GAPDH after the previous probe was stripped with Tris-EDTA buffer (1 mM EDTA, 10 mM Tris⅐Cl, pH 8.0) heated to 80°C. As shown in Fig. 5, the mRNA levels for catalase, glutathione peroxidase, ␤-actin, and GAPDH are very similar in the three RNA samples from pSFFV-Bcl-2-N/ N1003A, pSFFV-N/N1003A, and N/N1003A cells, respectively. Thus, BCL-2 has almost no effect on expression of the major anti-oxidative stress genes as well as housekeeping genes in rabbit lens epithelial cells.

FIG. 4. BCL-2 expression prevents N/N1003A cells from apoptosis induced by staurosporine (A and B) and camptothecin (C and D).
A and C, viability assays. N/N1003A cells, pSFFV-N/N1003A, and pSFFV-Bcl-2-N/N1003A stable expression clones were cultured in MEM with 10% rabbit serum until confluence. Then 3 ϫ 10 5 cells were plated into a 60-mm culture dish. After 12 h of culture, the media in three different cultures were replaced with 1% serum MEM containing 25 nM staurosporine (A) or 10 M camptothecin (C). The treatment was continued for 6 h. Viability of the three different cultures under treatment of these two reagents was determined as described previously (32). B and D, DNA fragmentation assays. The three different types of cells were cultured and treated for 12 h as described above. At the end of treatment, the cells were collected for extraction of the genomic DNA. The three DNA samples were separated in 2.0% agarose gel, stained with ethidium bromide, and photographed under UV illumination as previously described (30 -32).
FIG. 5. BCL-2 expression does not affect expression of the genes encoding catalase, glutathione peroxidase, ␤-actin, and glyceraldehyde-3-phosphatase dehydrogenase in N/N1003A as determined with Northern blot. Twenty five g of total RNAs extracted from N/N1003A, pSFFV-N/N1003A, and pSFFV-Bcl-2-N/ N1003A cells were denatured and separated on 1.2% formaldehydeagarose gel. Then the RNA samples were transferred to supported nitrocellulose membranes (Life Technologies, Inc.). The RNA blot was sequentially hybridized to [␣-32 P]dATP-labeled catalase, glutathione peroxidase, ␤-actin, and GAPDH cDNA probes, washed under high stringency conditions, and then exposed to x-ray film as described previously (32). The top panel shows ethidium bromide staining of the RNA gel to display the equal loading of the three samples.
Expression of the ␣B-crystallin Gene Is Distinctly Down-regulated in pSFFV-Bcl-2-N/N1003A Cells-To further explore why BCL-2 cannot prevent pSFFV-Bcl-2-N/N1003A cells from H 2 O 2 -induced apoptosis, we investigated whether expression of the lens crystallin genes was affected by BCL-2. Previous studies have revealed that ␣-crystallins are molecular chaperones (45,46) and can prevent apoptosis under a variety of conditions (47,48). The N/N1003A cells express only ␣B-crystallin as previous studies (49) have demonstrated. Thus, we analyzed expression of ␣B-crystallin gene in pSFFV-Bcl-2-N/N1003A, pSFFV-N/N1003A, and N/N1003A cells. First, Northern blot analysis with total RNA from the three cell lines demonstrated that the mRNA level for ␣B-crystallin was barely detectable in pSFFV-Bcl-2-N/N1003A cells. In contrast, both N/N1003A and pSFFV-N/N10003A cells had similar amounts of ␣B-crystallin mRNA that appeared as a prominent band (Fig. 6A). To further confirm this down-regulation, we used specific anti-␣B-crystallin antibody to conduct Western blot analysis. As shown in Fig.  6B, ␣B-crystallin protein was weakly detectable in pSFFV-Bcl-2-N/N1003A. In contrast, a strong ␣B-crystallin band was detected in the parental N/N1003A and pSFFV-N/N1003A cells. Thus, expression of BCL-2 in N/N1003A cells was associated with distinct down-regulation of expression of the ␣B-crystallin gene.

Inhibition of BCL-2 Expression Restores ␣B-crystallin Expression and the Ability against H 2 O 2 -induced Apoptosis in pSFFV-Bcl-2-N/N1003A
Cells-To confirm that BCL-2 actually represses expression of ␣B-crystallin gene in pSFFV-Bcl-2-N/N1003A cells, we prepared an antisense bcl-2 expression construct, pZeoSV-antisense-bcl-2 using the vector, pZeoSV. Both the vector and the antisense construct were introduced into pSFFV-Bcl-2-N/N1003A cells, and the transfected cells were selected with Zeocin (300 g/ml) and G418 (400 g/ml). After selection, stable expression clones of pZeoSV/pSFFV-Bcl-2-N/ N1003A and pZeoSV-antisense-bcl-2/pSFFV-Bcl-2-N/N1003A were obtained. Western blot analysis revealed that the antisense bcl-2 RNA was able to inhibit BCL-2 expression (top panel in Fig. 7A). When the same protein samples were analyzed for ␣B-crystallin expression, it was found that the level of ␣B-crystallin protein in pZeoSV-antisense-bcl-2/pSFFV-Bcl-2-N/N1003A cells was restored to the level close to that in pSFFV-N/N1003A cells (bottom panel in Fig. 7A). As in the pSFFV-Bcl-2-N/N1003A cells, the ␣B-crystallin protein in pZeoSV/pSFFV-Bcl-2-N/N1003A cells was down-regulated to a barely detectable level (bottom panel in Fig. 7A). Thus inhibition of BCL-2 expression in pSFFV-Bcl-2-N/ N1003A cells restores the level of ␣B-crystallin expression. Most importantly, when ␣B-crystallin expression was increased, the ability of the transfected cells to metabolize H 2 O 2 and to resist H 2 O 2 -induced apoptosis was also recovered (Fig. 7, C and F).
Expression of the Mouse ␣B-crystallin Counteracts the Effects of BCL-2 in Rabbit Lens Epithelial Cells-To further confirm that decreased expression of ␣B-crystallin accounts for the inability of the pSFFV-Bcl-2-N/N10003A cells to protect against oxidative stress-induced apoptosis, we prepared an expression construct in which the mouse ␣B-crystallin cDNA was inserted into a green fluorescence protein expression vector, pEGFP (48). Both the vector, pEGFP, and the expression construct, pEGFP-m␣B, were introduced into the BCL-2 expression cells, pSFFV-Bcl-2-N/N10003A. Expression of the GFP-mouse ␣B-crystallin fusion protein was confirmed by Western blot analysis using a specific anti-␣B-crystallin antibody (right lane in Fig. 8A) and also fluorescence microscopy (Fig. 8, B and C). Whereas the green fluorescence protein was distributed homogeneously in BCL-2 expression cells (Fig. 8B), the GFP-m␣B fusion protein was localized only in the cytoplasm (Fig. 8C). When the parental cells, vector-, and mouse ␣B-transfected BCL-2 expression cells were subjected to H 2 O 2 treatment, it was found that cells expressing both BCL-2 and ␣B-crystallin were more resistant to H 2 O 2 -induced apoptosis than the BCL-2 and GFP-expression cells or the BCL-2-transfected cells (Fig. 8D). Associated with the protection of cell death by exogenous mouse ␣B-crystallin expression, activation of caspase-3 found in pSFFV-Bcl-2-N/N1003A cells was largely repressed (Fig. 8E). These results suggest that the down-regulation of ␣B-crystallin in N/N1003A cells by BCL-2 is responsible for the attenuated ability of the pSFFV-Bcl-2-N/N1003A cells to protect against H 2 O 2 -induced apoptosis.
The Exogenous Mouse ␣B-crystallin Promoter Is Also Downregulated in BCL-2 Expression Cells-To determine whether BCL-2 can regulate ␣B-crystallin gene from another species, we have introduced a CAT reporter gene driven by the mouse ␣B-crystallin gene promoter (39) into both vector-and BCL-2transfected cells. As shown in Fig. 9, transient assays of relative CAT and ␤-galactosidase activities revealed that the mouse ␣B-crystallin gene promoter was also down-regulated by more than 2-fold. Thus, expression of human BCL-2 in rabbit A, Northern blot. Twenty five g of total RNAs extracted from N/N1003A, pSFFV-N/N1003A, or pSFFV-Bcl-2-N/N1003A cells were denatured, separated on 1.2% formaldehyde-agarose gel, and transferred to supported nitrocellulose membranes (Life Technologies, Inc.). The RNA blot was sequentially hybridized to ␣B-crystallin and GAPDH cDNA probes, washed under high stringency conditions, and then exposed to x-ray film as described previously (30 -32). B, Western blot. Fifty g of total proteins extracted from N/N1003A, pSFFV-N/N1003A, or pSFFV-Bcl-2-N/N1003A cells were separated by 12% polyacrylamide gel and transferred to nitrocellulose membranes and probed with anti-␣B-crystallin antibody as described under "Experimental Procedures." lens epithelial cells down-regulates both endogenous and exogenous ␣B-crystallin gene.

Down-regulation of ␣B-crystallin Is Largely Derived from Changed Activity of LEDGF in pSFFV-Bcl-2-N/N1003A
Cells-To explore how BCL-2 may down-regulate expression of ␣B-crystallin gene, we have examined the DNA binding activity of LEDGF, a nuclear transcription factor, which can positively regulate expression of ␣B-crystallin gene (24,25). As shown in Fig. 10A, when the nuclear extracts prepared from vector-and BCL-2-transfected cells were assayed, gel mobility shifting assays revealed that the vector-transfected cells contain a substantially higher level of binding activity to the LEDGF sites than the BCL-2-transfected cells (1st and 2nd lanes of Fig. 10A). This binding activity is contributed by LEDGF for three reasons. First, the unlabeled oligonucleotides containing LEDGF-binding site can compete off the labeled probe during the binding shifting assay (3rd and 4th lanes of Fig. 10A). Second, the pre-cleared nuclear extracts with anti-LEDGF antibody no longer gave DNA binding (5th and 6th lanes of Fig. 10A). Third, the oligonucleotides containing a mutated LEDGF site did not display any interaction with the nuclear extracts (7th and 8th lanes of Fig. 10A). Thus, downregulation of ␣B-crystallin gene is parallel to decreased activity of LEDGF. To demonstrate that the decreased LEDGF activity is responsible for down-regulation of ␣B-crystallin expression, we have introduced the LEDGF expression construct into BCL-2 expression cells. As shown in Fig. 10B, expression of an exogenous LEDGF cDNA in BCL-2 expression cells substantially increases the expression level of ␣B-crystallin. Thus, down-regulation of ␣B-crystallin gene is largely derived from BCL-2-modified LEDGF activity.

␣B-crystallin Prevents Apoptosis through Interaction with Procaspase-3 and Partially Processed Procaspase-3 in Rabbit
Lens Epithelial Cells-Our previous study (48) revealed that ␣B-crystallin can prevent apoptosis by repressing caspase-3 activation. In BCL-2 expression cells, ␣B-crystallin also prevents caspase-3 activation by H 2 O 2 (Fig. 8E). To explore how ␣B-crystallin represses caspase-3 activation, we have conducted immunoprecipitation-linked Western blot analysis. First, Western blot analysis revealed that the anti-GFP anti- body recognized both GFP and the fusion protein of GFP and mouse ␣B-crystallin (Fig. 11A). This anti-GFP antibody was used to conduct immunoprecipitation of the total proteins extracted from parental BCL-2 expression cells, pEGFP vector-, and mouse ␣B-transfected BCL-2 expression cells. The immunoprecipitated samples were then used for Western blot anal-ysis. As shown in Fig. 11B, the specific anti-␣B antibody identified the same 48-kDa band of GFP and mouse ␣B-crystallin fusion protein as the anti-GFP antibody did in pEGFP-m␣B/ pSFFV-Bcl-2-N/N1003A cells (top panel of Fig. 11B and also Fig. 11A). As expected, the specific anti-␣B antibody did not recognize any specific protein in the precipitated samples from pSFFV-Bcl-2-N/N1003A cells and pEGFP/pSFFV-Bcl-2-N/ N1003A cells. When the immunoprecipitated samples were analyzed with anti-caspase-3 antibody, it was found that in pEGFP-m␣B/pSFFV-Bcl-2-N/N1003A cells, two specific bands were recognized, one migrated slightly above 29 kDa and another below 29 kDa (bottom panel of Fig. 11B). The sizes of the two bands were equal to the 31-kDa procaspase-3 and the partially processed procaspase-3 (24 kDa), respectively. These two bands were disappeared when the anti-caspase-3 antibody was preincubated with purified recombinant caspase-3 protein (data not shown). Immunoprecipitation with normal mouse serum did not pull down either the 48 kDa GFP-m␣B-crystallin fusion protein or the procaspase-3 and partially processed procaspase-3 bands (data not shown). Together, our results suggest that mouse ␣B-crystallin can form a complex with both procaspase-3 and partially processed procaspase-3, which can be immunoprecipitated by anti-GFP antibody. DISCUSSION In the present communication, we have demonstrated the following. 1) BCL-2, when expressed in rabbit lens epithelial cells, can protect the transfected cells from apoptosis induced by a general protein kinase inhibitor, staurosporine, and also by a topoisomerase I inhibitor, camptothecin, suggesting that the expressed BCL-2 is functional in the lens epithelial cells.  (3) cells were grown to confluence in MEM with 10% rabbit serum. Then, 3 ϫ 10 5 cells were plated into 60-mm culture dishes. After 12 h of growth, the media in three different cultures were replaced with 1% serum MEM containing 150 M H 2 O 2 (the concentration at the starting point). The treatment was continued for 12 h. At the end of the treatment, each sample was collected for analysis of apoptosis (D) and caspase-3 activity assay (E) as described previously (30,48). 9. Regulation of the exogenous mouse ␣B-crystallin promoter linked to CAT reporter gene by BCL-2. The construct of CAT reporter gene driven by the mouse ␣B-crystallin gene promoter (39), together with the control construct expressing ␤-galactosidase (40), was introduced into both vector-and BCL-2-transfected cells using electroporation (30,48). The transfected cells were grown in 100-mm culture dishes and then harvested after 24 h of growth for assays of ␤-galactosidase and CAT activities as described previously (29 -30). The CAT activity is expressed as counts/min per microgram of protein after normalization against ␤-galactosidase activity.
BCL-2 prevents apoptosis in certain non-anti-oxidative stress pathway. 3) BCL-2 can specifically down-regulate expression of the ␣B-crystallin gene. 4) The down-regulation of ␣B-crystallin gene is largely derived from changed activity of LEDGF. 5) The down-regulation of ␣B-crystallin gene leads to attenuation of the BCL-2-transfected cells against H 2 O 2 -induced apoptosis. 6) ␣B-crystallin prevents apoptosis by interacting with procaspase-3 and partially processed procaspase-3 to repress caspase-3 activation.
The Protective Role of BCL-2-The protective role of BCL-2 has been documented extensively (8 -10) in many different cell and tissue types. Regarding the protection mechanism, one of the theories suggests that BCL-2 prevents cells from apoptosis by protecting them from oxidative stress. The supportive evidence is 2-fold. First, a number of laboratories have reported (11)(12)(13)(14)(15)) that bcl-2-transfected cells show a greater resistance to various pro-oxidant treatments than the mock-transfected cells. Second, antioxidants protect some cells from apoptosis induced by non-oxidative reagents (16 -18).
In the lens, oxidative stress and oxidative stress-induced apoptosis are implicated in cataractogenesis (31, 50 -52). Prevention of stress-induced apoptosis may lead to potential gene therapy strategy against cataractogenesis. Thus, we have studied the mechanism of BCL-2 action for two reasons. First, if BCL-2 could prevent oxidative stress-induced apoptosis in the lens, it would be a candidate for gene therapy in preventing cataractogenesis. Second, exploration of the BCL-2 action in lens cells may contribute to clarification of whether BCL-2 prevents apoptosis through anti-oxidative stress pathway. For these purposes, we have introduced the human bcl-2 gene into rabbit lens epithelial cells and established the stable line expressing hBcl-2. A mock control cell line transfected with the same vector was also established. When the two cell lines were subjected to treatment of H 2 O 2 at pathological concentrations in the eye as previously suggested (53), we found that the bcl-2-transfected cells are less capable of detoxifying H 2 O 2 than the control cells. Moreover, we found that bcl-2-transfected cells were more susceptible to H 2 O 2 -induced apoptosis when they were treated with a concentration previously shown to induce apoptosis of lens epithelial cells (31). Our results are consistent with cell lines such as EW-36 (19,20) but different from the results in other cell lines such as mouse neural cell line (11) or Pro-B-cell line (12). Such differences among different cell lines may reflect the intracellular property of these cell lines or the functional status of the expressed BCL-2 protein.
To confirm that human BCL-2 is functional in rabbit lens epithelial cells, we subjected the bcl-2-and vector-transfected cells to staurosporine and camptothecin treatment, both previously shown to induce typical apoptosis (41)(42)(43)(44). As expected, the bcl-2-transfected cells were more resistant to staurosporine-and camptothecin-induced apoptosis than the vectortransfected cells. Thus, our results suggest that BCL-2 prevents apoptosis in a non-anti-oxidative stress pathway in lens cells.
Next we explored why BCL-2 expression cells were less capable of degrading H 2 O 2 and more susceptible to H 2 O 2 -induced apoptosis. One of the possibilities tested is whether overexpression of BCL-2 in rabbit lens epithelial cells could down-regulate expression of the anti-oxidative genes as indirectly shown in the BCL-2 knockout mice (54). Hochman et al. (54) demonstrated that in the liver of the knockout mice, the activities for both catalase and glutathione are substantially increased over that in wild type mice as postnatal development proceeds, suggesting that BCL-2 may down-regulate expression of these enzymes in the wild type animal. Northern blot was conducted to analyze the expression of the genes encoding catalase and glutathione peroxidase, both of which are the major enzymes responsible for detoxifying H 2 O 2 (55). Our results revealed hardly any difference in the mRNA levels for the two enzymes between bcl-2-transfected and vector-transfected cells, thus excluding the possibility of differential expression of the antioxidative genes.
In the lens, another set of important proteins involved in cellular protection is the lens crystallin, especially the ␣-crystallin. In the rabbit lens epithelial cells, previous studies (49) have revealed that ␤and ␥-crystallins are not expressed in The reaction mixtures were then separated with 5% native polyacrylamide gel electrophoresis. The gel was dried and exposed to x-ray film for 3 h. Shown here is a typical result of three independent experiments. S, pSFFV-N/N1003A cells; B, pSFFV-Bcl-2-N/N1003A cells. B, Western blot analysis to demonstrate that LEDGF positively regulates expression of ␣B-crystallin in BCL-2 expression cells. The expression construct, pCI-GST-LEDGF, and its vector, pCI, were introduced into BCL-2 expression cells using electroporation (30). The transfected cells were grown in 100-mm culture dishes and then harvested after 48 h of growth for preparation of total proteins. Fifty g of total proteins from BCL-2 expression cells (1), vector-transfected BCL-2 expression cells (2), and LEDGF-transfected BCL-2 expression cells (3) were separated by 6 (for LEDGF) or 12% (for ␣B-crystallin) polyacrylamide gel and transferred to nitrocellulose membranes and probed with anti-LEDGF antibody (top panel) or anti-␣B-crystallin antibody (bottom panel) as described under "Experimental Procedures." this cell line. Regarding ␣-crystallins, only ␣B-crystallin is expressed. Therefore, we analyzed the expression of the ␣Bcrystallin gene. Both Northern blot and Western blot analyses revealed a distinct down-regulation of expression of ␣B-crystallin gene in BCL-2-expression cells.
Because ␣B-crystallin is a molecular chaperone (45,46) and also an anti-apoptotic protein (47,48), it is possible that the down-regulation of ␣B-crystallin accounts for the attenuated ability of the BCL-2 expression cells in preventing H 2 O 2 -induced apoptosis. To confirm this possibility, we used an antisense bcl-2 RNA to block BCL-2 expression. When BCL-2 expression was blocked, expression of ␣B-crystallin was increased (Fig. 7A). More importantly, the increase of ␣B-crystallin expression leads to the recovery of the ability of the double-transfected cells in their resistance against H 2 O 2 and H 2 O 2 -induced apoptosis (Fig.  7, B-D). The observation that ␣B-crystallin enhances the ability of the lens epithelial cells to protect against H 2 O 2 -induced apoptosis has been further confirmed by overexpression of an exogenous ␣B-crystallin gene in BCL-2 expression cells (Fig. 8).
Together, our results demonstrate that by down-regulating expression of ␣B-crystallin gene, human BCL-2 attenuates the ability of rabbit lens epithelial cells to resist H 2 O 2 -induced apoptosis.
Bcl-2 Regulates Gene Expression-In addition to its well established function in controlling cell survival, BCL-2 also has other important functions. One of these functions is to regulate expression of other genes. Initially, it is found that BCL-2 can increase the half-life of p21Bax (56), suggesting that BCL-2 modulates gene expression at the post-translational level. More recently, Feng et al. (57) have demonstrated that BCL-2 upregulates expression of a differentiation-specific gene encoding the proteoglycan aggrecan in rat chondrocyte cell line, IRC cells. This up-regulation occurs at the mRNA and protein levels. In contrast to its positive regulation on the aggrecan gene, we found that BCL-2 substantially down-regulates expression of the ␣B-crystallin gene in rabbit N/N1003A cells, although it has no effect on expression of either the anti-oxidative stress genes (coding for catalase and glutathione peroxidase) or the housekeeping genes (encoding ␤-actin and GAPDH). Such down-regulation occurs at both mRNA and protein levels and is BCL-2-dependent because the antisense bcl-2 RNA not only blocks BCL-2 expression but also abolishes ␣B-crystallin downregulation. BCL-2 also down-regulates expression from an exogenous mouse ␣B-crystallin gene promoter (Fig. 9). Thus, depending upon the cell type or the specific target gene, BCL-2 can exert either positive or negative regulation of expression of other genes. A recent study from Vairo et al. (58) further supports this point. In the mouse fibroblasts, Vairo et al. (58) demonstrate that BCL-2 up-regulates accumulation of p27 and p130 proteins but down-regulates the level of p107 protein from G 0 to S transition during the cell cycle of the fibroblasts. This differential regulation allows BCL-2 to retard fibroblast cells entering into the cell cycle (58).
How could BCL-2 regulate gene expression? It is well established that BCL-2 can modulate transactivities of different transcription factors. For example, NF-B, an important transcription factor mediating multiple signaling pathways (59), is positively regulated by BCL-2 (60 -62). By changing the affinity of IB␣ to NF-B, BCL-2 can up-regulate the transactivity of NF-B and expression of NF-B-responsive genes such as that encoding the matrix metalloproteinase-9 (61). The tumor suppressor, p53, is another target negatively regulated by BCL-2. Zhan et al. (62) found that in the human Burkitt's lymphoma WMN cell line, BCL-2 specifically suppresses the p53-mediated transactivation of p21 CIP1/WAF1 and GADD45 after treatment with methylmethane sulfonate or UV irradiation. In human kidney 293 cells and MCF7 cells, Froesch et al. (63) also observed that overexpression of BCL-2 down-regulates p53 transactivity without affecting nuclear accumulation of p53 protein. In BCL-2 expression rabbit lens epithelial cells, we have examined the DNA binding activity of LEDGF. This lens epithelial cell-derived growth factor is a transcription factor that positively regulates expression of ␣B-crystallin gene (24 -25). As expected, in BCL-2 expression cells where ␣Bcrystallin is distinctly down-regulated, the DNA binding activity of LEDGF is substantially decreased (Fig. 10A). This decreased LEDGF activity contributes substantially to the downregulation of ␣B-crystallin gene because expression of the exogenous LEDGF in BCL-2 expression cells significantly upregulates the expression level of ␣B-crystallin (Fig. 10B). Of course, BCL-2-modulated changes in other transcription factors may also contribute to down-regulation of ␣B-crystallin.
Mechanism by Which ␣B-crystallin Prevents Apoptosis-Since the first demonstration that ␣B-crystallin is able to prevent apoptosis induced by staurosporine (47), several laboratories have demonstrated that this molecular chaperone can provide cellular protection from a variety of stress conditions. FIG. 11. ␣B-crystallin prevents apoptosis through interaction with procaspase-3 and partially processed procaspase-3. A, Western blot analysis to demonstrate that the anti-GFP antibody recognizes both GFP and the fusion protein of GFP with mouse ␣B-crystallin. Fifty g of total proteins extracted from pSFFV-Bcl-2-N/N1003A, pEGFP/ pSFFV-Bcl-2-N/N1003A, and pEGFP-m␣B/pSFFV-Bcl-2-N/N1003A cells were separated by 10% polyacrylamide gel and transferred to nitrocellulose membranes and probed with anti-GFP antibody as described under "Experimental Procedures." B, immunoprecipitation-linked Western blot analysis. Five hundred g of total proteins from pSFFV-Bcl-2-N/N1003A, pEGFP/pSFFV-Bcl-2-N/N1003A, and pEGFP-m␣B/pSFFV-Bcl-2-N/ N1003A cells were incubated with 10 g (in 50 l) of anti-GFP antibody or 50 l of normal mouse serum and 50 l of protease inhibitor mixture for 1 h on ice. After incubation, 50 l of protein A/G plus-agarose was added into each incubated sample. These samples were then incubated overnight in a 4°C refrigerator attached to a slow motion rotator. At the end of incubation, these samples were washed 4 times with RIPA buffer (1ϫ PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) by spinning down for 5 min at 10,000 ϫ g. After the last wash, the pelleted samples were subjected to Western blot analysis as described above using specific anti-␣B-crystallin antibody (top panel) and anti-caspase-3 antibody (bottom panel).
For example, ␣B-crystallin has been shown to prevent apoptosis induced by UVA irradiation (64). In the transgenic mice overexpressing ␣B-crystallin, the expressed protein confers simultaneous protection against cardiomyocyte apoptosis during myocardial ischemia and reperfusion (65). In our recent study (48), we have demonstrated that ␣B-crystallin is able to prevent apoptosis induced by okadaic acid, and moreover, it does so by repressing caspase-3 activation. In the present communication, we have demonstrated that ␣B-crystallin also prevents cells from oxidative stress-induced apoptosis through repression of caspase-3 activation (Fig. 8E). How can ␣B-crystallin repress caspase-3 activation? The chaperone property of ␣B-crystallin suggests that it could bind to procaspase-3 to prevent caspase-3 activation by other proteases. A recent study (66) demonstrates that ␣B-crystallin indeed binds to the partially processed procaspase-3 intermediate. Here we present evidence to show that ␣B-crystallin can bind to both procaspase-3 and partially processed procaspase-3. Thus, one of the mechanisms for ␣B-crystallin to prevent apoptosis is to interact with procaspase-3 and partially processed procaspase-3 to repress caspase-3 activation.
It is also possible that ␣B-crystallin may prevent apoptosis through other mechanisms. For example, ␣B-crystallin is an autokinase (67,68), and it might modulate phosphorylation status of apoptosis regulators in BCL-2 family or other upstream death regulators. Recent studies (69 -72) have shown that the anti-apoptotic function of BCL-2 and BCL-X L and the proapoptotic ability of BAD and BIK can be dramatically changed by phosphorylation. It remains to be determined whether ␣B-crystallin can actually phosphorylate any of these cell death regulators.