BCL-xL Is a Target Gene Regulated by Hypoxia-inducible Factor-1α*

The transcription factor hypoxia-inducible factor-1α (HIF-1α) plays pivotal roles in physiology and pathophysiology. Constitutive or hypoxia-induced HIF-1α overexpression is observed in many types of cancers including prostate adenocarcinoma, in which it is associated with resistance to apoptosis and therapeutic agents. BCL-xL, a hypoxia-responsive, anti-apoptotic protein of the Bcl-2 family, is also overexpressed in prostate carcinoma and many other cancers. Despite this connection, whether BCL-xL expression is directly regulated by HIF-1α is not known. We used prostate cancer PC-3 cell with constitutive high HIF-1α level as a model to address this important question. We first generated prostate cancer PC-3 cells in which HIF-1α was stably knocked-down (HIF-KD) by using small interference RNA. BCL-xL was dramatically decreased in HIF-KD PC-3 cells, in parallel with sensitization to apoptosis with caspase-3 activation as well as decreased cell proliferation. We then demonstrated that HIF-1α directly regulated BCL-xL transcription by binding to a hypoxia-responsive element in the BCL-xL promoter (-865 to -847) by reporter gene assay, chromatin immunoprecipitation, and electrophoretic mobility shift and supershift assays. HIF-1α-dependent BCL-xL overexpression may be an important mechanism by which HIF-1α protects prostate cancer cells from apoptosis and leads to treatment resistance.

Hypoxia and HIF-1␣ overexpression are implicated in the pathogenesis of many cancers, including prostate carcinoma (3)(4)(5), in which it is associated with advanced clinical stage and treatment failure (6). HIF-1␣ overexpression has been identified in both prostate adenocarcinoma tissue (5,7) and cell lines (8).
Although acute hypoxia may lead to cell death, prolonged hypoxia results in resistance to apoptosis as well as to radiotherapy and chemotherapy (4,9,10), the mechanism of which is not well understood. Only recently have a few apoptosis regulators been identified as HIF-1␣ target genes, most notably the anti-apoptotic Mcl-1 (11) and BIRC5/survivin (12). Although pro-apoptotic molecules BNIP3, NIX (13,14), and Noxa (15) are also responsive to HIF-1␣, hypoxia-induced apoptosis-resistant phenotype eventually predominates.
BCL-xL (BCL2-like 1 or BCL2L1), a major anti-apoptotic protein of the Bcl-2 family, is also overexpressed in prostate carcinoma and many other cancers. BCL-xL overexpression is associated with the hormone-refractory phenotype and renders prostate cancer cells apoptosis-resistant, whereas BCL-xL knock-down increases sensitivity to chemotherapeutic agents (16,17). Despite the correlation of BCL-xL overexpression with HIF-1␣ in some tumors (18) and the observation that BCL-xL is a key molecule underlying hypoxia-driven cell death resistance (10), the mechanism by which hypoxia induces BCL-xL expression is unclear, as it has not been elucidated if HIF-1␣ directly regulates BCL-xL.
We tested the hypothesis that BCL-xL is under HIF-1␣ regulation using prostate cancer PC-3 cell as a model, in which HIF-1␣ level is constitutively high. We show that stable knockdown of HIF-1␣ by small interference RNA (siRNA) results in a dramatic decrease of BCL-xL with consequent increase in The Real-time PCR Master Mix containing SYBR Green (ToYoBo) was used for real-time PCR on Light Cycler 2.0 (Roche Diagnostics), and data were recorded and analyzed by the Light Cycler software 4.05. Copy number of target genes (relative to ␤-actin) was defined by 2 Ϫ⌬⌬Ct , where ⌬⌬Ct ϭ ⌬Ct HIF-KD Ϫ ⌬Ct HIF-CON ϭ (Ct HIF-KD-target Ϫ Ct HIF-KD-actin ) Ϫ (Ct HIF-CON-target Ϫ Ct HIF-CON-actin ).
Hypoxia Mimetic Treatment of PC-3 Cells-PC-3 cells were cultured in 6-well plates in fetal calf serum-free media and treated with 0, 200, 400 M CoCl 2 for 4 h. Cells were collected for Western analysis of HIF-1␣ and BCL-xL.
Total proteins resolved by SDS-polyacrylamide (Sigma) gel electrophoresis were electroblotted to polyvinylidene difluoride membrane (Amersham Biosciences), blocked with 5% nonfat milk and 0.1% Tween 20, and incubated with primary and secondary antibodies at room temperature for 2 and 1.5 h, respectively. Signals were detected by exposure to x-ray films after treatment with the SuperSignal enhanced chemiluminescence kit (Pierce).
Ultraviolet (UV) Irradiation Induced Cell Death-Cells in culture plates were briefly exposed to UV irradiation in a UV cross-linker (UVC-500, Hoefer, San Francisco, CA) at 120 mJ/cm 2 for 30 s. Cells were then cultured as appropriate for subsequent assays.
Cell Viability Assay-Cells were cultured in 96-well plates and measured by tetrazolium-based MTT (Sigma) cell proliferation assay. The working concentration of MTT was 1 mg/ml.
Terminal Deoxynucleotidyltransferase-mediated Biotinylated dUTP Nick end-labeling (TUNEL)-TUNEL was performed by using in situ cell death detection kit (Roche Diagnostics). Cells were fixed with 4% paraformaldehyde and permeabilized in 0.1% Triton-100, 0.1% sodium citrate, incubated with TUNEL reaction mixture then with alkaline phosphatase-conjugated anti-fluorescein antibody, stained with nitro blue tetrazolium/ 5-bromo-4-chloro-3-Indolyl phosphate and counterstained with methyl green. Reaction without terminal transferase was used as negative control. The Apoptotic index was represented as number of TUNEL(ϩ) cells/total number of cells (%).
Reporter Gene Assay for HIF-1␣-dependent BCL-xL Promoter Activity-The basic pGL3 luciferase reporter vector (Promega, Madison, WI) was used to construct reporter plasmids with various lengths of the BCL-xL promoter. Four plasmids were constructed in which the BCL-xL promoter spanned Ϫ1075 to ϩ617 (relative to the transcription start site) or truncated fragments of which were inserted upstream of the luciferase gene. The reporter constructs were designated as pGL1642 (Ϫ1075 to ϩ617), pGL1281 (Ϫ664 to ϩ617), pGL828 (Ϫ211 to ϩ617), and pGL621 (Ϫ4 to ϩ617), respectively. Two additional plasmids with HRE1 and HRE2 site-specific mutation were constructed: pGL828-MUT (Ϫ211 to ϩ617, with CGTG at Ϫ78 to Ϫ75 of HRE1 mutated to TCGG) and pGL1642-MUT (Ϫ1075 to ϩ617, with CGTG at Ϫ858 to Ϫ855 of HRE2 mutated to TCGG). Each reporter construct and the pRL-CMV plasmid (Promega) containing the Renilla luciferase gene as internal control were used in dual reporter gene assay for studying HIF-1␣-dependent gene expression. Cells were transfected with plasmids by using Lipofectamine 2000 (Invitrogen). Four hours after transfection, the medium was replaced by fresh medium. Thirty-six hours after transfection, cells were treated with 400 M CoCl 2 for 12 h, and luciferase activity was determined by using Luminometer TD-20/20 (Turner Designs, Sunnyvale, CA).
Chromatin Immunoprecipitation-Cells were lysed, and nuclei were pelleted. The extract was sonicated, and supernatants were collected and treated with sheared salmon sperm DNA (Invitrogen) and protein A/G-Sepharose (Santa Cruz). Immunoprecipitation was performed overnight at 4°C with 3 g of HIF-1␣ monoclonal antibody or the control isotype IgG2b (Lab Vision Corp., Fremont, CA) or no antibody and then with protein A/G-Sepharose and salmon sperm DNA. Precipitates were washed, and extracted with 1% SDS and 0.1 M NaHCO 3 . Eluates were pooled and heated. DNA fragments were purified and used as template for PCR. The promoterspecific primers used were: BCL-xL, 5Ј-CGA GCA GTC AGC CAG GTA G-3Ј and 5Ј-GAC GGC GAA GGC TCC TAT TG-3Ј; VEGF (as positive control), 5Ј-GTT CCC TGG CAA CAT CTG G-3Ј and 5Ј-GAC ATC AAA GTG AGC GGC AG-3Ј.
PC-3 cells were harvested after 6 h of incubation with 400 M CoCl 2 . Nuclear extract was prepared, and 10 g was incubated with 100 pmol of biotinylated probe in a 10-l reaction mixture (with 0.5 g of poly[dI-dC]) for 30 min at room temperature. For competition assays, a 50-fold excess of unlabeled wild-type probe or mutant probe was used. For supershift assays, 1.0 g of monoclonal anti-HIF-1␣ antibody was added to the reaction mixture and incubated at 4°C overnight. The mixture was electrophoresed at 4°C on 6% PAGE for 3 h and transferred to nylon membranes (Roche Diagnostics) by electroblotting. After baking and blocking, horseradish peroxidase-labeled streptavidin (1:1000, Zymed Laboratories Inc.) was added and incubated at room temperature for 2 h. Signals were detected by exposure to x-ray films after treatment with the SuperSignal enhanced chemiluminescence kit (Pierce).
Statistical Analysis-Statistical analysis was performed by using the SPSS 10 software package (Chicago, IL).

RESULTS
HIF-1␣ siRNA Significantly Decreased Expression of HIF-1␣ and Its Target Genes-HIF-1␣ mRNA and protein overexpression in prostate cancer cells and primary prostate adenocarcinoma tissues was validated by conventional RT-PCR (Fig. 1A) and Western blot (Fig. 1B) analysis, respectively. In normal prostate epithelium, HIF-1␣ mRNA and protein were undetectable (Fig. 1, A and B).
The two HIF-siRNA constructs showed comparable interference efficiency, with HIF-siRNA1 being more potent (Fig. 1, C, D, and E). In contrast, the control constructs with scrambled sequences had no effect on HIF-1␣ expression. The G418-selected PC-3 cells with stably transfected HIF-siRNA1 and HIF-siRNA2 (HIF-KD1 and HIF-KD2 cells, respectively) and the control plasmids (RNAi-CON1 and RNAi-CON2 cells, respectively) were largely homogeneous, as shown by the co-expression of green fluorescence protein (Fig. 1F).
HIF-1␣ mRNA (Fig. 1G) and protein (Fig. 1H) were significantly reduced in the HIF-KD cells, with consequent downregulation of known HIF-1␣ target genes BNIP3, GLUT-1, and GAPDH (Fig. 1, G and H). The control constructs had no effect on mRNA and protein expression of HIF-1␣ or its target genes (Fig. 1, G and H). Quantitative PCR analysis of mRNA of HIF-1␣ and its target gene GLUT-1 further validated the interference effect of HIF-1␣ siRNA, as both of which were significantly reduced (not shown).
HIF-siRNA also sensitized PC-3 cells to the anti-androgen drug flutamide, treatment by which further inhibited growth rate of HIF-KD cells (Fig. 2B). In contrast, RNAi-CON cell growth was exuberant and essentially remained the same with or without flutamide treatment.
HIF-1␣ siRNA Dramatically Decreased BCL-xL Expression-To elucidate the mechanisms by which HIF-1␣ siRNA inhibited cell proliferation and promoted apoptosis in PC-3 cells, we examined the effect of HIF-1␣ siRNA on major apoptosis regulators: the Bcl-2 family, the IAP family, and the caspase family. Most prominently, BCL-xL expression was significantly down-regulated by HIF-1␣ siRNA (Fig. 3, A-D), whereas the other examined members of the Bcl-2 family showed little change.
BCL-xS, which is encoded by the same gene locus, also showed a decrease in expression upon HIF-1␣ siRNA. As the base level of BCL-xS was extremely low, the effect appeared less dramatic than BCL-xL (Fig. 3A).
As expected, HIF-1␣ siRNA also induced a significant decrease of survivin, a documented HIF-1␣ target (Fig. 3, A and B). Other members of the IAP family, including CIAP1, CIAP2, and XIAP, were not down-regulated. Moreover, HIF-1␣ siRNA had no effect on expression level of the examined caspases (Fig. 3, A and B).

Overexpression of BCL-xL in HIF-KD Cells Promoted Cell
Growth and Inhibited Cell Death-To further demonstrate the importance of HIF-induced BCL-xL expression in apoptosis resistance, BCL-xL was artificially overexpressed in HIF-KD1 cells (HIF-KD1-xL, Fig. 4, A-E), which counteracted the effects of HIF-1␣ siRNA, resulting in enhanced cell growth (Fig. 4C) and inhibition of UV irradiation-induced cell death and caspase 3 activity (Fig. 4, D and E).

Inhibition of the PI3K/Akt Pathway Resulted in Down-regulation of Both HIF-1␣ and BCL-xL-The
down-regulation of BCL-xL by HIF-1␣ siRNA was dramatic given the constitutive high BCL-xL mRNA (Fig. 5A) and protein (Fig.  5B) level in prostate cancer cell lines. Because PI3K/Akt is a major signaling pathway that controls HIF-1␣ level, we tested if PI3K inhibitor LY294002 could lead to decrease of BCL-xL. As shown in Fig. 5C, both HIF-1␣ and BCL-xL were simultaneously reduced upon PI3K/Akt inhibition in a dosedependent manner.

Hypoxia Mimetic CoCl 2 Boosted Concomitant BCL-xL and HIF-1␣ Expression in PC-3 Cells-Treatment of PC-3 cells with the hypoxia mimetic CoCl 2 resulted in additional increase of BCL-xL and
HIF-1␣ simultaneously, further supporting the potential dependence of BCL-xL on HIF-1␣ in response to hypoxia (Fig. 5D).
Potential HREs on BCL-xL Promoter-The above experiments provided important clues to relationship between HIF-1␣ and BCL-xL. Because BCL-xL is a gene regulated by NF-B, which could potentially be activated by PI3K/Akt signaling, reduction of BCL-xL expression by PI3K/Akt inhibition might be the effect of either HIF-1␣ or NF-B inhibition or both. This prompted us to investigate whether BCL-xL was directly regulated by HIF-1␣.
Potential HRE was searched in the human BCL-xL promoter region. Four short HRE consensus motifs were identified within the ϳ1000-bp region preceding the transcriptional start site. Two of them, starting at positions Ϫ78 and Ϫ858 (Fig. 6A), fit the extended consensus and were highly conserved across species (Fig. 6B). Reporter Gene Assay of Putative HRE Activity-The reporter constructs were shown in Fig. 6C. The pGL1642 (Ϫ1075 to ϩ617) contained both HREs (HRE2 and HRE1) flanking two NF-B binding sites, whereas pGL1281 (Ϫ664 to ϩ617) differed by lacking HRE2. The pGL828 (Ϫ211 to ϩ617) contained HRE1 only, and pGL621 (Ϫ4 to ϩ617) contained no HRE. Two constructs were prepared with site-specific mutation of the respective HREs: pGL828-MUT and pGL1642-MUT.  PC-3 cells were transiently transfected with one of the six constructs, with PRL-CMV co-transfection as internal control. The hypoxia mimetic CoCl 2 was used to simulate hypoxia and to further boost HIF-1␣ as a means of showing hypoxia-induced gene transcription in the reporter gene assay.
The experiment showed significantly higher luciferase activity of the HRE-bearing constructs (Fig. 6C) than base line (pGL3). The effects were more dramatic with the HRE2-bearing construct upon CoCl 2 treatment, whereas mutation of the HRE (notably HRE2) core sequence resulted in a significant reduction of transcriptional activity. These experiments indicated that the reporter gene transcription was under control of BCL-xL promoter containing HRE, particularly HRE2, which responded to the hypoxia mimetic.
Chromatin Immunoprecipitation Assay Displayed HIF-1␣ Interaction with BCL-xL Promoter-To show HIF-1␣ physically bind to BCL-xL promoter, we first used chromatin immunoprecipitation assay of PC-3 cells treated by CoCl 2 . Using the chromatin fraction pulled down by anti-HIF-1␣ antibody as template, a PCR fragment corresponding to Ϫ1025 to Ϫ821 (containing HRE2) of BCL-xL promoter was detected (Fig. 6D) and verified by sequencing. This fragment was not detected when isotype control IgG2b or no antibody was used for the pulldown assay (Fig. 6D).
EMSA and Super Shift Assay Demonstrated HIF-1␣ Binding to the Promoter Region of BCL-xL-To further confirm binding of HIF-1␣ to the putative HRE in human BCL-xL promoter, we performed EMSA with two oligonucleotide probes, each containing one of the extended HRE consensus sequence, designated BCL-xL-Pro1 (Ϫ89 to Ϫ64) and BCL-xL-Pro2 (Ϫ865 to Ϫ847), respectively. Two probes with HRE core sequence mutations were also prepared for competition experiments and were designated as BCL-xL-Pro1-MUT and BCL-xL-Pro2-MUT, respectively. A known HIF-1␣ binding oligonucleotide derived from the VEGF receptor gene promoter (VEGF-Pro) was used as positive control (Fig. 6E) together with a corresponding HRE mutation probe, VEGF-Pro-MUT, for competition assays.
The experiments showed that the labeled BCL-xL-Pro2 (Fig. 6E, lanes  1 and 2), but not BCL-xL-Pro1 (not shown), caused gel mobility shift when incubated with nuclear proteins from PC-3 cells. The shift could be suppressed in the competition experiments with excess unlabeled wild-type probe BCL-xL-Pro2 (Fig. 6E, lane 3) but not with the HRE-mutated probe BCL-xL-Pro2-MUT (Fig. 6E, lane 4). When HIF-1␣ monoclonal antibody was included in the binding reaction, a supershift band was observed as well with BCL-xL-Pro2 (Fig. 6E,  lane 5) but not BCL-xL-Pro1 (not shown). Gel mobility shift and supershift were also shown with the VEGF-Pro positive control (Fig. 6E,  lanes 6 and 7) together with competition and supershift assays (Fig. 6E, lanes 8 -10). These results further demonstrated HIF-1␣ binding to the HRE2 in the Ϫ865 to Ϫ847 region of BCL-xL promoter.

DISCUSSION
We generated prostate cancer PC-3 cells in which HIF-1␣ was stably knocked-down by using siRNAs, which resulted in a significant decrease of the anti-apoptotic molecule BCL-xL. We then showed that HIF-1␣ directly regulated BCL-xL gene transcription. These novel findings point to HIF-1␣-dependent BCL-xL overexpression as an important mechanism by which HIF-1␣ protects prostate cancer cells from apoptosis and leads to treatment failure.
Hypoxia is common in solid tumors (4), including prostate carcinoma (3), in which the extent of hypoxia is correlated with clinical stage and treatment failure (6). A hypoxia-mediated increase in HIF-1␣ plays critical roles in tumorigenesis and progression of many cancers through HIF-1␣-dependent activation of genes that promote cancer cell survival, proliferation, spreading, and angiogenesis. Overexpression of HIF-1␣ and its target genes has been observed in a variety of solid tumors, for example, tumors of the brain (25), kidney and the urinary tract (26), and lung (27) as well as prostate (5,7).
High levels of HIF-1␣ has been observed in prostate cancer tissue and cell lines (5,7,8). Up-regulation of HIF-1␣ might be an early event in prostate carcinogenesis, as high grade prostate intraepithelial neoplasia showed higher a HIF-1␣ level than benign epithelium (5). It is noteworthy that, although HIF-1␣ overexpression is often hypoxia-dependent, prostate cancer cells have constitutively high HIF-1␣ level, which could be further increased by hypoxia (8). HIF-1␣ gene amplification (28) and P582S polymorphism or mutation in the oxygen-dependent domain (29) might contribute to overexpression of HIF-1␣ at normoxic conditions. Hypoxia and HIF-1␣ overexpression contribute to resistance to radiotherapy and chemotherapy (4). For example, the multidrug resistance 1 gene has been observed to be hypoxia-respon- sive and is regulated by HIF-1␣ (30). Treatment of LNCaP cells with the androgen receptor antagonist Casodex results in up-regulation of a subset of hypoxia-related genes, including membrane metallo-endopeptidase and cyclin G2, which might be involved in development of the androgen-independent phenotype (31).
Of more clinical interest is that silencing of HIF-1␣ gene results in sensitization of cancer cells to therapeutic agents. For example, HIF-1␣ knock-down increases sensitivity to 5-fluorouracil, doxorubicin, and gemcitabine in pancreatic cancer cell (36). Our finding that HIF-1␣ knock-down renders the androgen-independent PC-3 cells more sensitive to UV or flutamide treatment also supports that HIF-1␣ is a potential therapeutic target in androgen-independent prostate cancer.
Promotion of apoptosis is a recurrent theme of HIF-1␣ knock-down (11,(32)(33)(34)(35)(36)(37)(38)(39)(40). However, the underlying mechanisms have been less clear. Caspases could be increased or activated upon HIF-1␣ siRNA, but it most probably reflects the activation of caspase-dependent pathways rather than transactiva-FIGURE 6. HRE in BCL-xL promoter; binding to and transcriptional regulation by HIF-1␣. Two potential HRE sites (Site 1 and Site 2, starting at position Ϫ78 and Ϫ858 of human BCL-xL promoter, respectively) identified by sequence analysis (A) were conserved across species (B). Dual reporter gene assays were performed with expression constructs carrying various lengths of BCL-xL promoter (range relative to transcription start site) (C). The promoter regions were inserted into respective reporter gene (firefly luciferase) constructs, and the upper panel shows restriction analysis of the inserts (which have also been verified by sequencing). pGL1642 (Ϫ1075 to ϩ617) contained HRE2 and HRE1 flanking two NF-B binding sites. pGL1281 (Ϫ664 to ϩ617) lacked HRE2. pGL828 (Ϫ211 to ϩ617) contained HRE1 only, and pGL621 (Ϫ4 to ϩ617) contained no HRE. In pGL1642-MUT and pGL828-MUT, the respective HRE was mutated (cross). PC-3 cells were co-transfected with one of the six constructs together with pRL-CMV (which carried the Renilla luciferase gene controlled by CMV promoter, as internal control). CoCl 2 was used to simulate hypoxia and to boost HIF-1␣ level. The reporter gene activity, represented by relative luciferase activity (firefly/Renilla), was significantly increased over the base line when HRE2 was present, and the cells were stimulated with CoCl 2 . (C). In contrast, pGL1642-MUT significantly reduced the reporter gene activity. Constructs with HRE1 only (pGL828) resulted in slightly higher reporter gene activity over pGL621, but the difference was not significant, and mutation of HRE1 (pGL828-MUT) had little effect, indicating HRE1 was not effectively regulated by HIF-1␣. Also notice that the presence of the NF-B binding site resulted in significantly higher reporter gene activity over constructs lacking NF-B binding site, indicating significant contribution of this binding site to BCL-xL promoter activity. The chromatin immunoprecipitation assay was used to show HIF-1␣ binding to the BCL-xL promoter (D). PCR using chromatin (Input) pulled down by anti-HIF-1␣ antibody (HIF-Ab) as template yielded the BCL-xL-Pro fragment (Ϫ1025 to Ϫ821 containing HRE2) of the BCL-xL promoter, which was verified by sequencing. When isotype control IgG2b or no antibody (No Ab) was used for the pulldown assay, no PCR product was observed. PCR of VEGF promoter (VEGF-Pro) was used as positive control. EMSA was performed to further confirm HIF-1␣ binding to HRE (E). Oligonucleotide probes BCL-xL-Pro1 (Ϫ89 to Ϫ64) and BCL-xL-Pro2 (Ϫ865 to Ϫ847) contained HRE1 and HRE2, respectively. The HRE core sequence was mutated in the corresponding BCL-xL-Pro1-MUT and BCL-xL-Pro2-MUT probes. A known HIF-1␣ binding oligonucleotide derived from the VEGF receptor gene promoter (VEGF-Pro) was used as positive control together with a corresponding HRE mutation probe, VEGF-Pro-MUT. The biotin-labeled BCL-xL-Pro2 (E, lanes 1 and 2), but not BCL-xL-Pro1 (not shown), caused gel mobility shift when incubated with nuclear proteins from PC-3 cells. The shift was suppressed by competition with excess unlabeled wild-type probe BCL-xL-Pro2 (E, lane 3) but not with the HRE-mutated probe BCL-xL-Pro2-MUT (E, lane 4). When HIF-1␣ monoclonal antibody (HIF-Ab) was included in the binding reaction, a supershift band was observed with BCL-xL-Pro2 (E, lane 5) but not BCL-xL-Pro1 (not shown). Gel mobility shift and supershift were also shown with the VEGF-Pro positive control (E, lanes 6 and 7) together with competition and supershift assays (E, lanes 8 -10).
Although BCL-xL has been found to be a key molecule involved in hypoxia-induced resistance to cell death (10) and BCL-xL overexpression has been associated with increased HIF-1␣ in tumors such as non-small cell lung cancer (18), the mechanism by which hypoxia induces BCL-xL up-regulation and the relationship between HIF-1␣ and BCL-xL has not been known.
Our study, thus, provides the first evidence that HIF-1␣ directly regulates BCL-xL transcription by interacting with HRE in the BCL-xL promoter. It has been shown that in the androgen-independent PC-3 cell, BCL-xL is more responsible for apoptosis-resistance than the prototypic Bcl-2 (16). Our data, therefore, indicate that HIF-1␣-dependent overexpression of BCL-xL in PC-3 cells is one of the major mechanisms by which prostate cancer cells, particularly androgen-independent cells, resist apoptosis and chemotherapy. Recently, the IAP family member survivin has been identified as a transcriptional target of HIF-1␣ (12). Being up-regulated in many cancers, survivin is reported to be involved in the regulation of both apoptosis and cell division. Thus, HIF-1␣ overexpression (either constitutive or hypoxia-induced) may promote tumorigenesis by exerting double effects on key members of major gene families controlling cell death and proliferation; that is, the inhibition of cell death by up-regulating BCL-xL and survivin and promotion of cell proliferation by up-regulation of survivin. Elucidation of HIF-1␣-dependent BCL-xL expression may provide a new dimension for understanding BCL-xL regulation.