Anti-apoptotic role of HIF-1 and AP-1 in paclitaxel exposed breast cancer cells under hypoxia

Recently, differential effects of hypoxia on drug-induced apoptosis according to the cancer cell lines have been reported and many studies showed that hypoxia can confer resistance against chemotherapy-induced apoptosis in numerous solid tumors including breast cancer [15, 18, 24]. In order to investigate the effect of hypoxia on paclitaxel- or epirubicin-induced apoptosis, MDA-MB-231 breast cancer cells were incubated under normoxia or hypoxia with or without these two drugs and caspase 3 activity and DNA fragmentation were assessed (Figure 1). Hypoxia per se did not induce apoptosis since no increase in caspase 3 activity or DNA fragmentation were observed after 16 hours incubation. Paclitaxel at 50 μM and epirubicin at 10 μM did trigger apoptosis as shown by an increase in caspase 3 activity in normoxic conditions. Hypoxia did not modify the epirubicin-induced apoptosis. However, it markedly inhibited the paclitaxel-increase in caspase 3 activity. An increase in DNA fragmentation (Figure 1) as well as of nucleus fragmentation as evidenced by DAPI staining (data not shown), other important features of apoptotic cells, were also observed in the presence of paclitaxel under normoxia, which were significantly decreased by hypoxia. The fact that epirubicin showed no DNA fragmentation while caspase 3 was activated may be due to a different kinetics: caspase 3 is indeed activated before DNA fragmentation occurs so that it may occur later than 16 hours. In addition, in order to investigate whether apoptosis is really involved in taxol-induced cell death, the effect of a pan-caspase inhibitor has been studied on overall cell death. Overall cell death has been assessed by quantifying LDH release. Results showed that zVAD-fmk did inhibit taxol- and epirubicin-induced cell death (data not shown). This inhibition was however partial probably because in cell culture, when apoptosis is inhibited, cells die through necrosis, thus also increasing LDH release. Altogether, these data demonstrate that hypoxia is able to protect MDA-MB-231 cells against the chemotherapeutic agent paclitaxel-induced apoptosis.

In order to ensure that the hypoxia protection against paclitaxel was not due to an inactivation of the drug in hypoxic conditions, the paclitaxel effect on microtubules was investigated by immunofluorescence. The localization of the pro-apoptotic protein Bim was also studied (Figure 2). The antimicrotubule drug paclitaxel is known to interfere with microtubule function by binding to the tubulin polymer and stabilizing it against depolymerization. Figure 2 shows that on one hand neither epirubicin, nor hypoxia alone influenced the network of microtubules. On the other hand, in the presence of paclitaxel, we observed a strong modification of the microtubules which became shorter and thicker to form bundles. Hypoxia did not modify this paclitaxel-induced modification of the microtubule network. In the presence of paclitaxel, both under normoxia and hypoxia, we also observed a redistribution of the protein Bim from a perinuclear localization to clusters localized at the microtubule extremities. We thus concluded that hypoxia did not influence the effect of paclitaxel which remains active in hypoxic conditions since microtubule rearrangement and Bim translocation were still observed.

Recently, the influence of hypoxia in chemoresistance has been recognized. Hypoxic conditions elicit cellular responses designed to improve cell survival through an adaptive process. The main transcriptional factor activated under hypoxic conditions is HIF-1 (hypoxia-inducible factor-1). Regulation of gene expression through HIF-1 (hypoxia-inducible factor-1) but also via other transcription factors plays an important role in this process. We investigated here whether HIF-1 could be implicated in the hypoxia-induced protection against the apoptosis induced by paclitaxel.

We first evaluated the stabilization of the sub-unit HIF-1α. HIF-1 DNA binding activity as well as transcriptional activity using respectively an ELISA like TransAM assay and a reporter assay and by following the mRNA level of the HIF-1 inducible gene LDHA. Figure 3 shows that hypoxia induced an increase in HIF-1α stabilization, that was lower in the presence of paclitaxel and much higher in the presence of epirubicin compared to the control cells (Figure 3A). DNA binding activity of HIF-1α was increased in hypoxic conditions and neither paclitaxel nor epirubicin influenced this activity (Figure 3B). The transcriptional activity of HIF-1 was not modified by taxol or epirubicin under normoxic conditions while it was increased by hypoxia alone or combined with paclitaxel but not epirubicin. This was directly determined by reporter assay but also indirectly by studying gene expression of a HIF-1 inducible gene, LDHA (Figure 3C, D). HIF-1 thus remains active in presence of paclitaxel under hypoxic conditions. It has to be noted that under hypoxia, epirubicin inhibited HIF-1 activity while HIF-1alpha protein level was upregulated. We hypothesized that the cell "tried" to compensate for not having a HIF-1 active by upregulating further HIF-1alpha level. The mechanism for this has not been studied. The inhibitory effect of epirubicin on HIF-1 was also observed by Semenza's team in human hepatocellular carcinoma cell line Hep3B and in human embryonic kidney cell line HEK293 [25].

In order to study the implication of HIF-1 in the hypoxia anti-apoptotic effect, we then used siRNA directed against the HIF-1α sub-unit. Additional file 1 shows that a concentration of 50 nM of siRNA was high enough to specifically inhibit the synthesis and activity of HIF-1α but not HIF-2α while the RISC-Free control siRNA had no effect.

Cells were transfected with anti-HIF-1α or control siRNA before being incubated under normoxia or hypoxia in the presence of one or the other agent. The activity of the caspase 3 was then assessed. If HIF-1 plays an anti-apoptotic role, an increase in the paclitaxel-induced caspase 3 activity is expected when inhibited by specific siRNA. Figure 4 shows that, on one hand, the negative control siRNA did not influence caspase 3 activity. On the other hand, this activity was markedly increased for HIF-1α siRNA transfected cells incubated under normoxia as well as under hypoxia, with paclitaxel or epirubicin. No modification was observed in control cells. The caspase 3 activity measured for HIF-1α siRNA transfected cells incubated under hypoxia with paclitaxel reached similar level than the activity measured for non transfected cells incubated in normoxia with paclitaxel, indicating that the protection was reversed in this case.

These results clearly demonstrate an implication of HIF-1 in the hypoxia-induced protection against the apoptosis induced by paclitaxel.

Apoptosis is a complex event controlled by numerous pro- and anti-apoptotic genes. In order to understand by which mean HIF-1 protects MDA-MB-231 cells against paclitaxel-induced apoptosis, we studied the expression of 93 genes involved in apoptosis in cells transfected or not with HIF-1α siRNA or with control siRNA and incubated with or without paclitaxel under normoxic or hypoxic conditions, using Taqman Human Apoptosis Arrays.

Gene expression data for the 77 genes detected after real time PCR reactions are presented in additional file 2. Figure 5 shows gene expression modifications for 8 of these genes which expression seemed to be regulated by HIF-1 and that could be involved in the hypoxia-induced protection against apoptosis induced by paclitaxel.

Some of the genes are well known HIF-1 target genes such as GAPDH and BNIP3 whose expression was enhanced in non transfected cells or cells transfected with control siRNA and incubated under hypoxia with or without paclitaxel and which was inhibited by HIF-1α siRNA. Although BNIP3 is known to play pro-apoptotic function, a putative role in autophagy is more recently conferred to this protein which could thus participate to the hypoxia protection [26, 27]. BAK, CASP3, CASP8, CASP10 and TNFRSF10A are five pro-apoptotic genes for which we observed a lower expression under hypoxic conditions when compared to normoxic conditions, for non-transfected cells and cells transfected with the control siRNA. The downregulation of these genes by hypoxia was inhibited when cells were transfected with HIF-1α siRNA, and particularly in the presence of paclitaxel. In this case, hypoxia alone did not influence expression of these genes. We can therefore conclude that HIF-1 is responsible for the decrease of expression of these five pro-apoptotic genes, which could also participate to the hypoxia protection.

Hypoxia also increased the expression of the anti-apoptotic gene MCL-1. siRNA directed against HIF-1α inhibited this effect indicating that HIF-1 could participate to the anti-apoptotic effect of hypoxia by increasing MCL-1 expression under hypoxia in cells incubated in the presence of paclitaxel.

In order to validate our data, we also performed single SYBR Green quantitative real time PCR assays for two of the selected genes (Figure 5). Good correlation between relative transcript abundance data obtained by micro fluidic card and by real time RT-PCR was observed.

The transcription factor AP-1 was also studied for its potential role in the protection effect of hypoxia against the paclitaxel-induced apoptosis. It was already shown that this factor could play a role in the hypoxia-induced anti-apoptotic effect [18]. AP-1 is composed of two proteins belonging to the jun or fos family and c-jun is the most frequent subunit. Epirubicin totally inhibited the expression of c-jun both under normoxia and hypoxia while paclitaxel decreased its expression, interestingly more under normoxia than under hypoxia (Figure 6A). The DNA binding activity of AP-1 was significantly increased under hypoxia comparatively to normoxia (Figure 6B). Epirubicin decreased AP-1 binding activity both in normoxia and hypoxia while this activity was not modified by paclitaxel. These results indicate that hypoxia was able to increase the expression of c-jun and DNA binding activity of AP-1 in the presence of paclitaxel.

In order to verify a possible implication of AP-1 in the protective role of hypoxia against paclitaxel-induced apoptosis, c-jun siRNA was used to inhibit the expression of this transcriptional factor. Additional file 3 shows that a concentration of 50 nM of siRNA was high enough to inhibit the synthesis and activity of c-jun while the RISC-Free control siRNA had no effect. Cells were transfected with c-jun or negative control siRNA before being incubated under normoxia or hypoxia with one or the other chemotherapeutic agent. The activity of the caspase 3 was then assessed. Specific inhibition of c-jun decreased the hypoxia-induced protection against paclitaxel-induced apoptosis (Figure 7). The negative control siRNA did not influence the caspase 3 activity measured. On the other hand, caspase 3 activity was markedly increased in c-jun siRNA transfected cells incubated under normoxia or hypoxia in the presence of paclitaxel and even more in the presence epirubicin while it was not modified in control cells. The caspase 3 activity measured for c-jun siRNA transfected cells incubated under hypoxia with paclitaxel reached a level higher than the activity measured for non transfected cells incubated under normoxia with paclitaxel.

We already point out a possible role of the anti-apoptotic protein Mcl-1 belonging to the Bcl-2 family members in the hypoxia protection against paclitaxel-induced apoptosis via the activity of HIF-1. The effect of the c-jun inhibition on Mcl-1 expression was also investigated. Hypoxia as well as paclitaxel increased Mcl-1 expression but epirubicin markedly decreased it both at the mRNA and the protein levels (Figure 8). c-jun siRNA decreased Mcl-1 mRNA expression while negative control siRNA had no effect (Figure 9A). Similar results have been observed for Mcl-1 protein abundance (Figure 9B).

All together, these results show that hypoxia was able to increase the expression of c-jun and DNA binding activity of AP-1 and that c-jun was then able to increase Mcl-1 expression which may participate to the hypoxia-induced protection against apoptosis induced by paclitaxel.

Human Breast Cancer cells MDA-MB-231 were maintained in culture in 75-cm2 polystyrene flasks (Costar) with 15 ml of Roswell Park Memorial Institute medium (RPMI 1640, Invitrogen) containing 10% of fetal calf serum (Invitrogen) and incubated under an atmosphere containing 5% CO₂.

For hypoxia experiments (1% O₂), cells were incubated in serum-free CO₂-independent medium (Invitrogen) supplemented with 1 mM L-glutamine (Sigma) with or without paclitaxel (Invitrogen) at 50 μM or epirubicin (Calbiochem) at 10 μM. Normoxic control cells were incubated in the same conditions but in normal atmosphere (20% O₂).

Silencing of HIF-1α or c-jun expression was achieved using siGENOME SMARTpool human HIF1A/c-jun from Dharmacon. RISC-free control siRNA purchased from Dharmacon was used to control for non-specific effects. For siRNA experiments, 2 × 10⁶ cells were seeded in 75-cm² polystyrene flasks (Costar) with 10 ml of RPMI 1640 medium (RPMI) containing 10% of fetal calf serum and incubated 24 hours under an atmosphere containing 5% CO₂. Cells were then transfected 24 hours under standard culture conditions with 50 nM siRNA using the DharmaFECT 1 (Dharmacon) transfection reagent according to the manufacturer's instructions.

The transfection media were removed and replaced by culture media for 8 hours. Cells were then trypsinized and seeded at the appropriate density in flasks or microplates, depending on the experiment to perform.

The fluorogenic substrate Ac-DEVD-AFC was used to measure caspase 3 activity according to Lozano et al [61]. Cell extracts were prepared as described by Wellington et al [62]. Cells were seeded in 6-well plates (300,000 cells/well) one day before the incubation. After the incubation, the medium was recovered and centrifuged at 1,000 g for 5 min. Cells still attached to the well were scrapped in 200 μl cold PBS and recovered into a microtube. Pelleted cells were resuspended in 100 μl PBS at 4°C and also added to the microtube. The samples were centrifuged at 1,000 g for 5 min. at 4°C and the pellet resuspended in 50 μl of lysis buffer (10 mM Hepes/KOH, pH 7.0, 10% sucrose, 2 mM EDTA, 0.1% CHAPS, 5 mM dithiothreitol and 10 μg/ml aprotinin). After incubation at 4°C on a rotating wheel for 15 min, the lysates were centrifuged at 13,000 g for 5 min at 4°C and the supernatants were recovered for the assay.

The protein concentration was measured and 10 μg proteins completed to 50 μl with lysis buffer were mixed with 13 μM Ac-DEVD-AFC (BD Pharmingen) and 50 μl reaction buffer (40 mM PIPES, pH 7.2, 200 mM NaCl, 2 mM EDTA, 0.2% CHAPS, 0.10% sucrose and 10 mM dithiothreitol). The reaction was allowed to take place for 1 hour at 37°C and the fluorescence generated by the release of the fluorogenic group AFC on cleavage by caspase 3 was measured by excitation at 400 nm and emission at 505 nm.

The measurement of cytoplasmic histone-associated DNA fragments (mono- and oligonucleosomes) after induction of cell death was performed with the «cell death detection ELISA» (Roche Molecular Biochemicals). After treatment, cells were lysed with the incubation buffer and the cytoplasmic fraction recovered. ELISA was performed according to the manufacturer's protocol.

MDA-MB-231 cells were seeded at 30,000 cells/well in 24-well plates and 24 hours later were incubated for 16 hours with or without paclitaxel or epirubicin under normoxic or hypoxic conditions. Immunofluorescence staining was performed as described in [18].

Primary antibody for Bim staining was rabbit anti-Bim (# 2819 Cell Signaling) (1/100 dilution). Primary antibody for Mcl1 staining was rabbit anti-Mcl1 (# 819 Santa Cruz) (1/100 dilution). Alexa Fluor-488-conjugated anti rabbit IgG antibody (Molecular Probes) was used at 1/1,000 dilution. Primary antibody for α-tubulin staining was mouse anti-α-tubulin (# T5186 Sigma) (1/100 dilution). Alexa Fluor-546-conjugated anti-mouse IgG antibody (Molecular Probes) was used at 1/1,000 dilution.

Cells, seeded and incubated in 25 cm2 flasks, were scrapped in 200 μl of lysis buffer (Tris 40 mM pH 7.5, KCl 150 mM, EDTA 1 mM, triton X-100 1%) containing a protease inhibitor mixture («Complete» from Roche Molecular Biochemicals, 1 tablet in 2 ml H₂O, added at a 1/25 dilution) and phosphatase inhibitors (NaVO₃ 25 mM, p-Nitrophenyl Phosphate (PNPP) 250 mM, β-glycerophosphate 250 mM and NaF 125 mM, at a 1/25 dilution). Western blot analysis was performed as described in [18] using mouse anti-HIF-1α antibody (# 610958 Becton Dickinson) used at 1/1,500 dilution or rabbit anti-MCL1 antibody (# 4572 Cell Signaling) used at 1/5000 dilution. Mouse anti-β-actin antibody (A5441 Sigma) (final dilution 1/100,000) and mouse anti-α-tubulin (Sigma) (final dilution 1/3000) were used for normalization. Mouse and rabbit IgG horseradish peroxidase-linked antibody (Amersham Pharmacia Biotech) was used at 1/300,000 dilution as secondary antibodies.

Nuclear protein extractions in high salt buffer were prepared as previously described [63]. Briefly, cells seeded in 75 cm2 flasks (Corning) were incubated with or without paclitaxel at 50 μM or epirubicin at 10 μM under normoxic or hypoxic conditions for 16 hours. At the end of the incubation, cells were rinsed with PBS containing 1 mM Na₂MoO₄ and 5 mM NaF. They were then incubated on ice for 3 minutes with 10 ml cold Hypotonic Buffer (HB 1X, 20 mM HEPES, 5 mM NaF, 1 mM Na₂MoO₄, 0.1 mM EDTA) and harvested in 500 μl HB containing 0.5% NP-40 (Sigma). Cell lysates were centrifuged 1 minute at 13,000 rpm and sedimented nuclei were resuspended in 50 μl HB containing 20% glycerol and a protease inhibitor cocktail (Roche) and phosphatase inhibitors (1 mM Na₃VO₄, 5 mM NaF, 10 mM p-nitrophenylphosphate, 10 mM β-glycerophosphate). Extraction was performed for 30 minutes at 4°C by the addition of 50 μl HB containing 20% glycerol, 0.8 M NaCl and protease/phosphatase inhibitors. Nuclear extracts from the supernatant were recovered after centrifugation for 10 minutes at 13,000 rpm.

DNA-binding assays using TransAM ELISA kit (Active Motif) for detecting transcription factor DNA binding activity was performed according to the manufacturer's recommendations. Briefly, 5 μg of nuclear proteins were incubated for 1 hour in a 96-well plate coated with a double-stranded oligonucleotide containing the consensus sequence recognized by the transcription factor to be assayed. The transcription factor bound to DNA was detected using a specific primary antibody (rabbit anti-c-Jun (SC-1964 Santa Cruz), mouse anti-HIF-1α (H72320 Transduction Laboratories)). Colorimetric reaction was then performed with a HRP-conjugated anti-rabbit or anti-mouse IgG antibody and absorbance was measured at 450 nm in a spectrophotometer.

MDA-MB-231 transfections were performed in 24-well plates (35,000 cells per well) with SuperFect reagent (Qiagen). 1384 ng of the reporter plasmid pGL3(PGK-HRE6)-tk-Luc containing binding sites for HIF-1 upstream of the firefly luciferase gene [64] were co-transfected with 115.4 ng of normalization vector (pCMVβ vector coding for the β-galactosidase, Promega) in RMPI without serum for 5 hours.

The transfection media were removed and replaced by culture media for 24 hours. Cells were then incubated under hypoxia for 16 hours. After hypoxia incubation, β-galactosidase was assayed in parallel to the firefly luciferase activity, assayed in a luminometer using the Luciferase Reporter Assay System (Promega). Results are expressed as means of the ratio between the firefly luciferase activity and the β-galactosidase activity.

After the incubation, total RNA was extracted using the Total RNAgent extraction kit (Promega). mRNA contained in 2 μg total RNA was reverse transcribed using SuperScript II Reverse Transcriptase (InVitrogen) and oligodT primers according to the manufacturer's instructions. Forward and reverse primers for RPL13A, C-JUN, MCL-1, LDHA, BNIP3 and TNFRSF1B were designed using the Primer Express 1.5 software (Applied Biosystems). Amplification reaction assays contained 1× SYBR Green PCR Mastermix (Applied Biosystems) and primers (Applied Biosystems) at the optimal concentrations. A hot start at 95°C for 5 minutes was followed by 40 cycles at 95°C for 15 seconds and 65°C for 1 minute using an ABI PRISM 7000 SDS thermal cycler (Applied Biosystems). RPL13A was used as the reference gene for normalization and mRNA expression level was quantified using the threshold cycle method.

After the incubation, total RNA was extracted using the Total RNAgent extraction kit (Promega). mRNA contained in 2 μg total RNA was reverse transcribed using the "High Capacity cDNA Archive" kit from Applied Biosystems according to the manufacturer's instructions. 100 ng of retrotranscribed total RNA in 50 μl were then mixed with 50 μl of the "Taqman Universal PCR master Mix" (Applied Biosystems) and loaded into one of the 8 fill ports the microfluidic array. "TaqMan Human Apoptosis Arrays" are 384-well micro fluidic cards that contain assays for 93 human genes in addition to 3 endogenous controls. They enable to perform 96 real-time PCR reactions simultaneously for 4 samples, allowing the detection of 96 genes. mRNA expression level was quantified using the threshold cycle method with 18S as the reference gene.

SigmaStat software (Jandle Scientific, Germany) was used for the statistical analysis. Data are presented as means ± SD and were evaluated by one-way ANOVA, using the Holm-Sidak method.