Synovial cell death is regulated by TNF-α-induced expression of B-cell activating factor through an ERK-dependent increase in hypoxia-inducible factor-1α

B-cell activating factor (BAFF) has a role in the maturation and maintenance of B cells and is associated with rheumatoid arthritis (RA). Here, we investigated whether tumor necrosis factor (TNF)-α-induced BAFF expression controls the survival of fibroblast-like synoviocytes (FLS) and whether their survival can be regulated by TNF-α-mediated upregulation of hypoxia-inducible factor (HIF)-1α using MH7A synovial cells transfected with the SV40 T antigen. More TNF-α-treated cells died compared with the control. Survival was increased by incubation with Z-VAD but inhibited after transfection with BAFF-siRNA. Both BAFF and HIF-1α expression were enhanced when MH7A cells were treated with TNF-α. TNF-α-induced BAFF expression decreased in response to HIF-1α-siRNA, whereas it increased under hypoxia or by overexpressing HIF-1α. The HIF-1α binding site on the BAFF promoter (−693 to −688 bp) was confirmed by chromatin immunoprecipitation assay to detect the −750 to −501 bp and −800 to −601 bp regions. The BAFF promoter increased in response to TNF-α treatment or overexpression of HIF-1α. However, TNF-α-induced BAFF expression and promoter activity decreased after treatment with the ERK inhibitor PD98059. Cell death was enhanced by PD98059 but was inhibited by overexpression of HIF-1α. Taken together, our results demonstrate that BAFF expression to control synovial cell survival was regulated by HIF-1α binding to the BAFF promoter, and suggest for the first time that HIF-1α might be involved in the production of inflammatory cytokines to regulate the physiological function of rheumatic FLS.

Synovial hyperplasia and destruction of cartilage and bone are characteristics of rheumatoid arthritis (RA). 1 The synovial membrane is thin in a normal joint and consists of only a few cells. However, many cell types, including immune cells and synoviocytes, occur in a rheumatoid synovial membrane. 2 Recruitment and accumulation of immune cells in joint tissue induces inflammation 3 and the abnormal increase in the number of synoviocytes causes low oxygen tension. 4,5 Both inflammation and hypoxia are major microenvironmental features of RA.
Hypoxia-inducible factor-1α (HIF-1α), which is a regulator of angiogenesis, tumor growth, and glucose metabolism, is a wellknown transcription factor in cancers. 6,7 HIF-1α also has an important role in the pathogenesis of RA. 8 High expression levels of HIF-1α are detected in the intimal synovium of patients with RA and are localized in the nucleus and cytoplasm of synoviocytes. 9 HIF-1α is normally degraded under normoxic conditions by the ubiquitin-proteasome pathway; 10 however, it accumulates under normoxic conditions in an inflammatory environment. 11 Various immune cells, including macrophages, T cells, B cells, and plasma cells are recruited to the layer that lines the synovium during the progression of RA. 12 Although angiogenesis occurs, a malfunctioning vascular system maintains the hypoxic conditions. 13,14 Hypoxia-exposed macrophages produce additional quantities of proinflammatory cytokines, such as tumor necrosis factor (TNF)-α. 15 RA is an immune disorder associated with many cytokines. Synovial macrophages and synoviocytes produce abundant pro-and anti-inflammatory cytokines, such as TNF-α, interleukin (IL)-1, IL-6, and transforming growth factor-β. 3,16,17 TNF-α regulates other cytokines, destroys joint tissue, 18,19 and stabilizes HIF-1α under normoxic conditions. 20 Fibroblast-like synoviocytes (FLS), which are components of the synovial membrane, have a crucial role in initiating RA. RA-FLS develop cancer cell-like characteristics, such as anchorage-independent growth, loss of contact inhibition, and an invasive phenotype. 21 They also produce and release proinflammatory cytokines, matrix metalloproteinases, and growth factors that affect other cells. 22 TNF-α-stimulated FLS express B-cell activating factor (BAFF), originally known as a B-cell proliferation and survival factor. 23 BAFF is produced by immune cells (monocytes, dendritic cells, and macrophages), 24 and also by non-immune cells, such as salivary gland epithelial cells, 25 prostate epithelium, 26 and FLS. 27 Excess levels of BAFF are detected in patients with an autoimmune disease, and particularly in the synovial tissue of patients with RA. 28 Increased B-cell survival in response to BAFF can be detrimental to patients with an autoimmune disorder. An excess B-cell response increases circulating autoantibody levels in patients with certain autoimmune disorders, such as systemic lupus erythematosus, Sjögren's syndrome, and RA. 27,29,30 We reported previously that TNF-α-induced BAFF controls RA angiogenesis by regulating vascular endothelial growth factor (VEGF) expression in synoviocytes. 31 Although HIF-1α and BAFF are highly expressed in the joints of patients with RA, the relationship between these two  factors is not understood. In this study, we investigated whether TNF-α-induced BAFF expression controls synovial cell survival, and whether TNF-α regulates HIF-1α and BAFF expression through the extracellular-regulated kinase (ERK) pathway in TNF-α-stimulated MH7A synovial cells. We also describe the mechanism of action underlying BAFF expression, which is regulated in a HIF-1α-dependent manner in TNF-α-stimulated FLS.

Results
hBAFF expression is associated with the survival of synovial cells. FLS have a crucial role in initiating RA, and RA-FLS develop cancer cell-like characteristics; 21 thus, we examined hBAFF expression and the effect of hBAFF on survival of synovial cells using MH7A synovial cells. Synovial cells isolated from patients with RA were treated with TNF-α for 1, 3, 6, 9, 12 h, and hBAFF expression was highest after the 6 h treatment (data not shown). We also confirmed that hBAFF expression was increased by stimulating FLS from patients with RA or MH7A synovial cells with TNF-α for 6 h ( Figure 1a). TNF-α-induced BAFF was also observed by immunostaining and fluorescence-activated cell sorting analysis. hBAFF proteins were detected in the intracellular compartment and on the synovial cell surface (Figures 1b  and c). Then, we studied the role of hBAFF expression in synovial cells. As synovial hyperplasia is a characteristic of RA, we examined cell death in TNF-α-treated MH7A cells using the trypan blue exclusion assay. The percentage of dead cells increased significantly after treatment with TNF-α ( Figure 1d). In contrast, the percentage of dead cells decreased significantly after incubating the cells with TNF-α in the presence of Z-VAD (Figure 1e). hBAFF expression was enhanced by incubating the cells with TNF-α in the presence of Z-VAD (Figure 1f). We confirmed a role for hBAFF in the survival of synovial cells by inhibiting BAFF expression using BAFF-siRNA ( Figure 1g). The percentage of dead cells increased significantly after transfection with hBAFF-siRNA ( Figure 1h). These data demonstrate that hBAFF expression could be associated with the survival of synovial cells.
hBAFF expression increases following TNF-α treatment of RA-FLS, MH7A cells. As HIF-1α is associated with the pathogenesis of RA 8,9 and BAFF controls RA angiogenesis, 31 we investigated whether BAFF expression is regulated by HIF-1α in FLS. We examined hBAFF and HIF-1α expression levels under normoxic conditions, and MH7A cells were treated with various concentrations of TNF-α for different times ( Figure 2). When MH7A cells were treated with various concentrations of TNF-α for 6 h, hBAFF, VEGF, and HIF-1α transcript levels increased (Figure 2a). A significant increase in hBAFF expression was confirmed by real-time quantitative polymerase chain reaction (qPCR; Figure 2b). The hBAFF promoter, as judged by a luciferase activity assay, was also significantly and dose-dependently enhanced after a 6 h stimulation with TNF-α (Figure 2c), which was confirmed by measuring the hBAFF protein level in MH7A cells (Figure 2d). HIF-1α protein levels under normoxic conditions also increased in response to TNF-α treatment (Figure 2e). In addition, when MH7A cells were treated with TNF-α for different durations, hBAFF, VEGF, and HIF-1α transcript levels increased (Figure 2f). A significant increase in hBAFF expression was confirmed by real-time qPCR (Figure 2g). Transcriptional activity of the hBAFF promoter was enhanced significantly and time-dependently after stimulation with TNF-α (Figure 2h), which was confirmed by measuring the hBAFF protein level in MH7A cells ( Figure 2i). These results suggest that HIF-1α may directly or indirectly regulate hBAFF expression.
TNF-α-treated hBAFF expression is dependent on HIF-1α expression. We examined the association between HIF-1α and hBAFF expression. When MH7A cells were transfected with HIF-1α-siRNA, hBAFF expression was inhibited according to reverse transcriptase polymerase chain reaction (RT-PCR; Figure 3a). When pGL3-hBAFF-Luc plasmids were co-transfected with HIF-1α-siRNA into MH7A cells, hBAFF promoter activity was significantly inhibited after stimulation with TNF-α for 6 h (Figure 3b). We observed no changes in the toxicity of HIF-1α-siRNA-transfected control cells, as judged by a trypan blue exclusion assay (data not shown). No differences in total cell number were detected without TNF-α treatment before or after transfection with HIF-1α-siRNA (data not shown). These data demonstrate that BAFF expression could be associated with HIF-1α in TNF-αstimulated synovial cells.
To confirm whether hBAFF expression is regulated by HIF-1α, changes in hBAFF promoter activity were measured in HIF-1α-transfected MH7A cells. When pSG5-HIF-1α was overexpressed in MH7A cells transfected with the pGL3-hBAFF-Luc plasmid, glutathion-S-transferase (GST) protein from pSG plasmids was detected and BAFF expression was enhanced in pSG5-HIF-1α-overexpressing cells (Figure 3c). We observed no changes in toxicity in (e and f) MH7A cells were treated with TNF-α for 3 days (e) or 6 h (f) in the presence or absence of Z-VAD. Dead cells were estimated with trypan blue exclusion assay (e). hBAFF transcripts were measured by RT-PCR (f, left). Each band was quantified by using ImageJ 1.34 (g, right). (g and h) MH7A cells were transfected with hBAFF-siRNA and treated with TNF-α. Then, hBAFF transcripts were measured by RT-PCR (g, left). Each band was quantified by using ImageJ 1.34 (g, right). Dead cells were estimated with trypan blue exclusion assay (h). Data were the representative of four experiments. Data in the bar or line graph represent the means ± S.E.M. *Po0.05; **Po0.01; significant difference as compared with TNF-α-untreated control. # Po0.05; ## Po0.01; significant difference as compared with TNF-α-treated and Z-VAD-untreated control (e and f right) or TNF-α-treated and scrambled siRNA-treated control (g and h right) pSG5-HIF-1α-transfected control cells, as judged by the trypan blue exclusion assay (data not shown). No differences were detected in total cell number without TNF-α treatment before or after transfection with pSG5-HIF-1α (data not shown). hBAFF promoter activity in MH7A cells increased significantly by the overexpression of HIF-1α in MH7A cells (Figure 3d), and a significant effect of HIF-1α on hBAFF promoter was confirmed in HEK293T cells (Figure 3e).
To re-confirm the effect of HIF-1α on hBAFF expression, the MH7A cells were incubated under hypoxic conditions for various times. As shown in Figure 4a, hBAFF and HIF-1α protein levels increased, as well as hBAFF, hVEGF, and HIF-1α transcript levels ( Figure 4b). A significant increase in hBAFF expression was confirmed by an increase in hBAFF promoter activity as judged by luciferase activity assay ( Figure 4c). These results suggest that hBAFF production could be controlled by HIF-1α expression in synovial cells.
hBAFF expression is enhanced by HIF-1α binding to the MH7A cell hBAFF promoter. Given that hBAFF promoter construct ranges from 0.25 to 1.0 kb, 32 we examined where HIF-1α affects the hBAFF promoter. As shown in Figure 5a, when MH7A cells were transfected with different sizes of hBAFF promoter plasmids, hBAFF promoter activity increased significantly after treatment with TNF-α. hBAFF promoter activity increased in response to overexpression of HIF-1α when each size of hBAFF promoter plasmid was co-transfected with pSG5-HIF-1α ( Figure 5b). Transcriptional activity of the 1.0 kb hBAFF promoter was higher than that of the 0.25 or 0.5 kb promoters, suggesting that HIF-1α binds the hBAFF promoter from − 501 to − 1000 bp.
To examine HIF-1α binding to 0.25 kb fragments of the 1.0 kb hBAFF promoter, we performed a chromatin immunoprecipitation (ChIP) assay. As shown in Figure 6a, HIF-1α was associated from − 750 to − 501 bp. However, no association of HIF-1α was found from − 1000 to − 751 bp (data not shown). Then, we deleted the − 1000 to − 800 bp fragment, beginning at 1.0 kb, to confirm hBAFF promoter activity via HIF-1α binding, which resulted in a 0.8 kb hBAFF promoter. When MH7A cells were transfected with the 0.8 kb hBAFF promoter plasmids, hBAFF promoter activity increased significantly after treatment with TNF-α (Figure 6b). Activity of the 0.8 kb hBAFF promoter increased significantly after incubation under hypoxic conditions (Figure 6c) or by co-transfection with pSG5-HIF-1α (Figure 6d), suggesting that HIF-1α binds with the hBAFF promoter from − 501 to − 750 bp.
We analyzed a TRANSFAC database to examine HIF-1α binding to the hBAFF promoter. The results showed that HIF-1α bound from − 693 to − 688 bp on the hBAFF promoter ( Figure 7a). Then, we deleted − 801 to − 1000 bp and − 600 to − 1 bp, beginning at 1.0 kb, which resulted in a 0.2 kb hBAFF promoter. When MH7A cells were transfected with 0.2 kb hBAFF promoter plasmids, hBAFF promoter activity increased significantly after treatment with TNF-α ( Figure 7b). Activity of the 0.2 kb hBAFF promoter increased significantly after co-transfection with pSG5-HIF-1α (Figure 7c). The ChIP assay showed that HIF-1α bound from − 800 to − 601 bp of the hBAFF promoter in MH7A cells treated with TNF-α (Figure 7d). These results suggest that TNF-α-mediated BAFF expression was regulated by HIF-1α binding to the hBAFF promoter in synovial cells.
hBAFF expression decreases by inhibiting TNF-αinduced ERK with PD98059. We assessed ERK phosphorylation to determine whether signaling molecules control hBAFF expression. When the MH7A cells were treated with TNF-α, ERK phosphorylation increased significantly in TNF-αtreated cells (Figure 8a). When the MH7A cells were treated with TNF-α in the presence or absence of the ERK inhibitor PD98059 (Figure 8b, left and right bottom), hBAFF and HIF-1α protein levels decreased in response to PD98059 (Figure 8b, left and right top). In addition, hBAFF and HIF-1α transcript levels were inhibited by PD98059 treatment (Figure 8c). PD98059 significantly inhibited hBAFF expression according to real-time qPCR (Figure 8d). Transcriptional activity of the 1.0 kb hBAFF promoter was enhanced by TNF-α stimulation, but was also inhibited significantly by treatment with PD98059 ( Figure 8e); this suggests that hBAFF expression may be regulated by HIF-1α via activation of ERK. The results were confirmed by co-transfection of the 1.0 or 0.8 kb hBAFF promoter plasmid with the pSG5-HIF-1α plasmid. hBAFF promoter activity, which was increased by overexpression of HIF-1α, was significantly inhibited by treatment with PD98059 ( Figures 8f and g). These data suggest that hBAFF expression may be associated with HIF-1α binding to the hBAFF promoter through an ERK-mediated increase in HIF-1α level.
Survival of synovial cells was recovered by PD98059 treatment or HIF-1α overexpression. We examined the association between HIF-1α and death of synovial cells.  nce of PD98059, cell death was increased by pre-treatment with PD98059 compared with that in the TNF-α-induced cell death control (Figure 8i). These results suggest that the survival of synovial cells may be regulated by TNF-α-induced hBAFF expression via activation of ERK and the expression and binding of HIF-1α to the hBAFF promoter.

Discussion
RA is characterized by synovial hyperplasia and destruction of cartilage and bone. 1 Various immune cells are recruited and accumulate in joint tissues. 3 Synovial macrophages and FLS in patients with RA secrete abundant pro-and antiinflammatory cytokines, including TNF-α. 3,16,17,22 Then, RA deteriorates in response to increased production of TNF-α and other cytokines. 18,19 The abnormal increase in the number of synoviocytes decreases oxygen tension, 4,5 stabilizes HIF-1α, and induces VEGF expression. 9,13,14 HIF-1α is also stabilized by TNF-α under normoxic conditions. 20 In addition, TNF-αstimulated FLS express BAFF, a well-known B-cell proliferation and survival factor, 23 which may contribute to RA angiogenesis by regulating VEGF expression. 31 Autoimmunity can deteriorate due to an increase in B-cell survival and excess B-cell responses, caused by the expression of BAFF, which increases in response to TNF-α stimulation. 27,29,30 Although HIF-1α and BAFF are highly expressed in the joints of patients with RA, little has been reported about the relationship between these two factors. Here, we report the effect of TNF-α-induced BAFF expression on the survival of synovial cells, as well as the mechanism of action of BAFF expression in a HIF-1α-dependent manner in FLS. BAFF was originally known as a B-cell proliferation and survival factor. 23 BAFF is not only produced by myeloid cells, 24,33 but also by non-lymphoid cell types. 27,31,[34][35][36][37] However, with the exception of B cells, little is known about the role of BAFF on other types of cell, including myeloid cells and non-lymphoid cells. We observed a significant increase in hBAFF expression, in the intracellular compartment and on the cell surface, when synovial cells were treated with TNF-α (Figures 1a-c and Figure 2). As synovial hyperplasia is a characteristic of RA, our data show that TNF-α-induced death of synovial cells was reduced by inhibiting apoptosis with Z-VAD treatment, which enhanced hBAFF expression (Figures 1d-f). The role of hBAFF in the survival of synovial cells was confirmed by BAFF-siRNA (Figures 1g and h), suggesting that hBAFF expression might be associated with regulating the survival of synovial cells in patients with RA.
Additional proinflammatory cytokines, such as TNF-α, are produced by hypoxia-exposed macrophages. 15 We observed the changes of BAFF transcripts at 1, 3, 6, 9 h-incubated synocytes with TNF-α ( Figure 2f) and we choose only the time point to show the highest hBAFF expression level at 6 hincubation with TNF-α. Even though protein generally changes following transcript changes, our data showed that the transcriptional increase looks like at the same time to protein increase. Then, we just would like to figure out the regulatory mechanism of hBAFF on synovial cell death via HIF-1α.
HIF-1α accumulates even under normoxic conditions in an inflammatory condition. 11 The high expression level of HIF-1α detected in synoviocytes may have a role in the pathogenesis of RA. 8,9 Therefore, TNF-α-induced BAFF expression could be regulated by the association between HIF-1α and the BAFF promoter. Our data show that the expression of HIF-1α and VEGF increased under various experimental conditions used to treat synoviocytes with TNF-α ( Figure 2). Although transfection with HIF-1α-siRNA inhibited hBAFF expression, overexpression of pSG5-HIF-1α increased hBAFF expression ( Figure 3). Our results also show that the hBAFF promoter was activated by the binding of HIF-1α. Increased hBAFF expression was confirmed under hypoxic conditions (Figure 4), suggesting that hBAFF could be a novel target gene for HIF-1α-mediated transcription.
A previous report showed that BAFF expression is associated with inhibited lung function and hypoxia in patients with chronic obstructive pulmonary disease, 38 suggesting a link between hypoxia and the induction of BAFF in FLS. As HIF-1α is a well-known transcription factor, 6,7 we investigated the role of HIF-1α in the regulation of BAFF promoter activity and BAFF expression by overexpressing HIF-1α. Our data show that HIF-1α may be associated with the − 750 to − 501 bp region of the hBAFF promoter ( Figures 5 and 6). HIF-1α binding to the hBAFF promoter was demonstrated by detecting the − 800 to − 601 bp region in MH7A cells treated with TNF-α ( Figure 7). These results suggest that HIF-1α may regulate hBAFF expression directly or indirectly, and may also be associated with the pathogenesis of RA. Thus, hBAFF expression could be regulated by the binding of HIF-1α to a specific site on the BAFF promoter.
BAFF expression is also regulated by various signaling pathways, including the activation of nuclear factor-kappa B by lipopolysaccharide (LPS)-induced production of reactive oxygen species (ROS), 39 the CREB-binding protein/p300, [40][41][42] JAK/STAT, 35 LPS-induced protein kinase A-mediated CREB, Figure 8 hBAFF expression was dependent on TNF-α-induced ERK activation. (a-d) MH7A cells were stimulated with TNF-α in the absence (a) or presence (b-d) of PD98059, ERK inhibitor. Cell lysates were prepared and western blotting was performed to detect p-ERK, ERK (a and b), hBAFF, or HIF-1α (b, left). Each protein band was quantified by using ImageJ 1.34 (b, right). RNA was isolated with TRIzol TM . hBAFF and HIF-1α transcripts were measured by RT-PCR (c) and real-time qPCR normalized to GAPDH expression (d). (e-g) MH7A cells were transfected with pGL3-hBAFF-Luc plasmid and stimulated with TNF-α in the presence or absence of PD98059 (e). MH7A cells were co-transfected with 1.0 kb (f) or 0.8 kb (g) pGL3-hBAFF-Luc and pSG5-HIF-1α plasmids. Then, cells were incubated with TNF-α in the presence or absence of PD98059 (f and g). Luciferase activity of hBAFF promoter was measured by using luminometer. (h and i) MH7A cells were transfected with pSG5 or pSG5-HIF-1α (h) or pretreated with PD98059,ERK inhibitor (i). Then, MH7A cells were stimulated with TNF-α for 72 h. Dead cells were estimated with trypan blue exclusion assay. Data were the representative of four experiments. Data in the bar graph represent the means ± S.E.M. *Po0.05; **Po0.01, significant difference as compared with TNF-α-untreated control (b, right, d, h and i) or pSG5 plasmid-transfected control (h) at each time point (f and g). # Po0.05; ## Po0.01, significant difference as compared with TNF-α-treated and PD98059-untreated control (b, right, d, and i), PD98059-untreated control (d) at each time point (e) with pSG5 or pSG5-HIF-1α plasmid-transfected control (f and g), or TNF-α-treated and pSG5-plasmid-transfected control (h) the Epac1-mediated Rap1, 43,44 and ROS-dependent protein kinase C (PKC)/c-Fos pathways. Both the ERK and phosphoinositide 3-kinase pathways are involved in IL-1β and TNF-α-induced HIF-1α expression in FLS. 45 Our data show that hBAFF expression could be associated with HIF-1α binding to the hBAFF promoter through the ERK-mediated increase in HIF-1α level (Figure 8). The ERK/HIF-1α pathway in our study could be an additional signaling pathway to control hBAFF expression.
Taken together, although it has not been cleared of all possible signaling molecules for TNF-α-induced BAFF expression except ERK, our data suggest that the survival of synovial cells could be regulated by BAFF expression through the association between HIF-1α and the BAFF promoter. In addition, it appears that TNF-α-induced BAFF expression in FLS may protect synovial cells by themselves and mediate a crosstalk to protect B cells from apoptosis in the inflammatory microenvironment of RA. These results may for the first time provide a novel role for BAFF in cells other than B cells. Data also suggest that HIF-1α could be a novel target for treating RA.
Cell cultures. MH7A synovial cells isolated from intra-articular soft tissues of the knee joints of RA patients were obtained from the Riken cell bank (Ibaraki, Japan) through Dr. Ho-Geun Yoon of Yonsei University (Seoul, Korea). Briefly, MH7A is a cell line established by transfection with the SV40 T antigen. 47 MH7A cells were cultured in RPMI 1640 (Gibco BRL, USA) supplemented with 10% heat-inactivated FBS, penicillin (final concentration, 100 U/ml), streptomycin (P/S, final concentration, 0.1 mg/ml) and L-glutamine at 37°C in an atmosphere of 5% CO 2 in air.
Measurement of hBAFF promoter activity. The MH7A cells were transfected with pGL3-hBAFF-Luc and pcDNA-lacZ for monitoring transfection efficiency by β-galactosidase assay. Luciferase activity was determined by incubating cell extracts with luciferase substrate (Promega). Luminescence was measured using luminometer (Berthold Technologies, Oak Ridge, TN, USA). Luciferase units of experimental vector were normalized to the control vector in each sample. 31,32,48 Hypoxia treatment. For incubation under hypoxic conditions, cells were placed in an atmosphere of 1% O 2 , 5% CO 2 , 10% H 2 , and 84% N 2 with intermittent flushing with nitrogen, sealed, and then maintained in a humidified incubator at 37°C in a hypoxia chamber (Forma Anaerobic System, Thermo Electron Corporation). Hypoxia-treated cells were collected in the hypoxia chamber to prevent the rapid degradation of hypoxia-responsive molecules.
siRNA transfection. siRNA for HIF-1α (GenBank accession number NM_001530.2 and NM_181054.1) and BAFF was purchased from Bioneer (Daejeon, Korea). When cells were 70-80% confluent in a 12-well plate, the medium was changed with serum-free RPMI 1640. Transfection was performed with lipofectamine 2000 transfection reagent according to the manufacturer's protocol (Invitrogen, Carlsbad, CA, USA). Briefly, 20 pmol of each siRNA or 1 μl of lipofectamine 2000 transfection reagent diluted in 50 μl serum-free RPMI 1640 was mixed after the incubation for 5 min. The mixture was incubated for the additional 20 min and carefully added to the cells. Then, the cells were incubated for 4 h and the medium was changed again with a complete RPMI 1640 containing 10% FBS, P/S, and Lglutamine. The cells were collected at 48 h for RNA isolation and protein extraction.
Chromatin immunoprecipitation assay. Chromatin immunoprecipitation (ChIP) assays were performed as described previously. 49,50 The cells were crosslinked with final concentration 1% formaldehyde for 10 min at room temperature. Then 125 mM glycine was added to quench unreacted formaldehyde. The cells were gathered and sonicated to make DNA fragments with a size range of 200 to 1000 bp. The cell extracts were immunoprecipitated using 2 μg anti-HIF-1α or rabbit IgG control (Abcam, Cambridge, UK) for each sample suspended in 450 μl