Combined histone deacetylase inhibition and tamoxifen induces apoptosis in tamoxifen-resistant breast cancer models, by reversing Bcl-2 overexpression

Previous studies have shown that continuous exposure of ER-positive breast cancer cells to Tam results in suppression of classical ER genomic signaling activity without loss of ER expression [14-17]. This suggests a direct role for the ER in driving resistance [18]. To mimic the emergence of hormone therapy resistance, we generated in vitro Tam-resistant cell lines TAMR^M and TAMR^T (see Materials and methods). Resistance to Tam was reflected in TAMR^M and MCF7 total cell counts, recorded over time with and without 10 μM Tam treatment and expressed as fold increase over time compared to baseline (0 h) (Figure 1A). In the absence of Tam, TAMR^M cells grew considerably faster than parental MCF7 cells, as illustrated by the doubling times (Dts, 21.5 versus 33.1 hours). Treatment with 10 μM Tam arrested MCF7 cell growth. However, with the same concentration of Tam, TAMR^M cell growth was only modestly reduced (23.9% reduction in doubling time). Consistent with elevated proliferation rate, TAMR^M cells exhibited greater BrdU incorporation compared to MCF7 cells, both in the absence and presence of Tam (Figure 1B). To test whether TAMR^M cells had altered sensitivity to estrogen and the anti-estrogen fulvestrant, cells were incubated for 48 hours in estrogen-free media and then treated with vehicle (C), E2 1 nM, Tam 1 μM or Ful 0.1 μM. As expected, E2 stimulated the growth of MCF7 cells, whereas the absence of E2 or the presence of anti-estrogens Tam or Ful inhibited growth (Figure 1C). In contrast, the presence or absence of E2 or Tam did not significantly affect TAMR^M cell growth. Furthermore, treatment with Ful only partially inhibited TAMR^M cell growth. Thus, consistent with previous reports, development of Tam resistance in TAMR^M cells resulted only in partial cross-resistance to Ful [19,20]. Treatment with increasing Ful concentrations for 24 hours depleted ER protein in both cell lines (Figure 1D). However, comparable doses of Ful up to 500 nM resulted in substantially more ER loss in MCF7 compared to TAMR^M cells, consistent with the growth assay. This suggested the TAMR^M cells were less sensitive to Ful inhibition compared to MCF7 cells. The increased proliferation and decreased sensitivity of TAMR^M cells to Tam was also observed in in vivo tumor xenografts (Figure 1E). Similar to cells grown in vitro, TAMR^M-derived tumors did not require E2 for growth, unlike MCF7 cells (Figure S1 in Additional file 1). The relative proliferation of TAMR^T cells as compared to T47D cells, treated with increasing Tam doses in complete media for 72 hours demonstrates the Tam resistance of this TAMR^T cell line (Figure 1F).

Several mechanisms of resistance have been attributed to the development of acquired hormone resistance, including altered expression of key proteins regulating proliferation and death. Consistent with prior preclinical models and in the clinic, ER expression was maintained in both our tamoxifen-resistant cell lines [21,22]. Evaluation of ER expression in TAMR^M cells revealed that both mRNA (ESR1) and protein were higher compared to MCF7 cells, 4- and 2-fold, respectively (Figure 2A and B).

To determine whether increased ER expression affected its activity, ER target gene expression was examined. Despite the elevated levels of ER in both resistant models, expression of classic ER response genes (for example PGR and TFF1) were substantially reduced (Figure 2A and E). It has been previously shown that the pro-growth and survival signals mediated by estrogen signaling are in part through the regulation of c-Myc, p21 and Bcl-2 expression [23-25]. Consistent with the higher proliferation capacity of TAMR^M cells, the major cell cycle checkpoint protein p21 was suppressed (approximately 90%) (Figure 2B) and c-Myc substantially elevated (approximately 300%) compared to MCF7 cells. Furthermore, Bcl-2 expression increased nearly 600%. Consistent with these in vitro findings, the TAMR^M xenografts expressed higher Bcl-2 and c-Myc proteins and lower p21 protein compared to the MCF7 xenografts. However there was no significant increase in ER protein in the TAMR^M xenografts (Figure 2C). Comparing the basal level of proteins in the TAMR^T cells, we found that these resistant cells also displayed higher ER and Bcl-2 proteins and substantially reduced PGR protein as compared to T47D parental cells (Figure 2D). Basal mRNA levels of BCL-2 and PGR corresponded with the proteins, but ESR1 mRNA was not elevated in the TAMR^T-resistant cells although ER protein was found to be elevated (Figure 2E). This suggests that the relative mechanisms leading to ER protein upregulation likely differ in the two resistant models.

External stimulation by E2 results in ER dimerization and activation of estrogen response elements (EREs) or co-factor recruitment to non-ERE sites, which drives ER target gene expression and proliferation of ER-positive MCF7 cells. To determine the role of the ER in the accelerated growth rate of TAMR^M cells, we conducted siRNA-mediated knockdown of ER (ESR1) mRNA in both MCF7 and TAMR^M cells, and found comparable ER depletion resulted in an approximately 60% growth reduction in both cell lines (Figure 3A). This suggests the ER remains an important mediator of cell growth in TAMR^M cells despite substantial reduction in sensitivity to estrogen and anti-estrogens (Figure 1). We next tested whether this reduced sensitivity was the result of altered ER-mediated signaling. To study relative ligand sensitivity of the ER in TAMR^M and MCF7 cells, we transfected each with a luciferase reporter plasmid (pERE-luc-tk) driven by a promoter containing an ERE. As expected, treatment of MCF7 cells with E2 increased luciferase activity with increasing dose, while Tam and Ful decreased luciferase activity below basal levels. On the other hand, luciferase activity in TAMR^M cells remained unaffected by E2 or anti-estrogen treatment (Figure 3B). To evaluate whether the ER-mediated target gene transcription was altered in TAMR^M cells, mRNA from a set of known ER-responsive genes (for example PGR, TFF1, CDH1, c-MYC, H.CTSD, GREB1, TRIM25, and BCL2), were quantified following siRNA-mediated depletion of the ER and compared to similar depletion in MCF7 cells (Figure 3C). In MCF7 cells, ER depletion resulted in downregulation of all target mRNAs to varying degrees. Interestingly in TAMR^M cells, all but three transcripts (for example TFF1, CDH1 and c-MYC) were downregulated following ER depletion, suggesting that, independent of E2, ER continued to regulate gene expression in TAMR^M cells. To test whether ER-mediated expression was truly E2 independent, TAMR^M cells were stimulated with E2 and mRNA levels of these select genes were quantified and compared to MCF7 cells (Figure 3D). Consistent with siRNA-mediated depletion of ER, E2 stimulation did not significantly increase TFF1 and CDH1 mRNA, while c-MYC, H.CTSD and BCL-2 mRNA were only slightly increased. PGR and GREB1 mRNA were significantly increased with E2 stimulation; however, their increase was substantially less than the increase observed in MCF7 cells. Taken together, these results suggest that ER remains important for TAMR^M cell growth, but has differential transactivation properties at target genes compared to MCF7 cells. Furthermore, the ER in TAMR^M cells has significantly reduced sensitivity to E2.

We have previously shown that the addition of an HDAC inhibitor to hormonal therapy can induce apoptosis in breast cancer cells and reverse hormone therapy resistance in clinical studies [11,26,27]. For this study, the potent hydroxamic-type pan-HDAC inhibitor PCI-24781 (abexinostat) was primarily used. To evaluate the efficacy of this combination, TAMR^M cells were treated with increasing concentrations of PCI, with and without 10 μM Tam, and assayed for viability after 72 hours (Figure 4A). At clinically feasible concentration of PCI (200 nM), combination with Tam resulted in cell death in about 54% of the TAMR^M cells, greater than the additive effect of either agent alone. The PT drug combination was most effective in inhibiting colony formation. (Figure 4B). Consistent with an elevated proliferation rate (Figure 1A), untreated TAMR^M cells formed colonies within 14 days, while MCF7 cells failed to form detectable colonies within this time (DNS).

In MCF7 cells, combined HDAC and ER inhibition results in reduced proliferation, with cells accumulating in G1 of the cell cycle. As such, the effect of PCI and Tam alone and in combination on cell cycle progression after 24-hour treatment was examined. Untreated TAMR^M cells had a 2-fold greater G2-M population (Figure 4C) compared to parental MCF7 cells, consistent with accelerated growth. Unlike in MCF7 cells, Tam treatment did not result in a G1 shift in the TAMR^M cells. Treatment with PCI, however, resulted in a similar G1 accumulation for both TAMR^M and MCF7 cells. When treated in combination, both MCF7 and TAMR^M cells exhibited cell cycle distributions comparable to PCI treatment alone. We next evaluated the effect of the PT combination on cell proliferation after 72 hours of treatment. PCI alone comparably restricted growth of both cell lines. While Tam alone significantly reduced the growth of parental cells (80%), it only modestly reduced TAMR^M cell growth (30%) (Figure 4D). The combination treatment resulted in an enhanced anti-proliferative effect, comparable in both cell lines. The combined effect of HDAC and ER inhibition on TAMR^M cell viability was not limited to PCI. Combinations with Tam and several other HDAC inhibitors, including valproic acid, panobinostat, entinostat and vorinostat, resulted in a greater than additive reduction in cell viability (Figure 4E). To exclude the possibility of any non-ER mediated effect of Tam we have demonstrated enhanced apoptotic cell death by HDAC inhibition in ER-depleted TAMR^M cells (Figure S2 in Additional file 2). We further evaluated whether treatment with PCI and Tam could induce cell death in TAMR^T cells. After 72 hours of treatment, PCI alone induced 30% cell death, Tam alone did not lead to significant death, and the combination resulted in 60% cell death, suggesting an enhanced effect in presence of both drugs (Figure 4F). In summary, combined HDAC and ER inhibition induced growth arrest and cell death in our acquired Tam-resistant breast cancer models. While growth arrest appeared to be primarily the effect of HDAC inhibition, significant cell death required both agents.

As shown in Figure 2, the major anti-apoptotic protein, Bcl-2, and proliferation driver, c-Myc, are significantly overexpressed in TAMR^M cells. Additionally, a key cell cycle break, p21, is downregulated. We thus hypothesized that the efficacy of combined HDAC and ER inhibition may be the result of reversing the altered expression of these key proteins. To test this hypothesis, we first treated TAMR^M cells with increasing concentrations of PCI for 24 hours and evaluated protein expression of ER, p21, c-Myc, and Bcl-2 (Figure 5A). With increasing dose, ER and c-Myc protein were significantly downregulated, while p21 was upregulated. Although modest, Bcl-2 protein was reproducibly reduced when treated with 100 nM or more PCI. As HDACs regulate both ER protein stability and transcription, the effect of PCI treatment on ER mRNA was evaluated. As seen previously with MCF7 cells, HDAC inhibition using PCI reduced TAMR^M cell transcription of ER mRNA in a dose- and time-dependent manner (Figure S3 in Additional file 3). To determine the effect of combined HDAC and ER inhibition on the expression of these key proteins, MCF7 and TAMR^M cells were treated with vehicle, 200 nM PCI, 10 μM Tam, or the combination for 48 hours and evaluated for c-Myc, p21, and Bcl-2 expression (Figure 5B). In MCF7 cells, treatment with either single agent reduced c-Myc and Bcl-2 and increased p21 expression. In TAMR^M cells, PCI elicited similar changes to the expression of these proteins compared to MCF7 cells, while Tam had no effect on p21 or c-Myc expression, but slightly reduced Bcl-2. PCI treatment reduced ER in both cell lines. Though Tam treatment stabilized ER in MCF7 cells, ER was unaltered in TAMR^M cells. With the PT combination, p21 expression was further increased in MCF7 cells compared to single agent treatment. Due to the significant single agent effect on c-Myc and Bcl-2 expression, a combination effect was difficult to determine in MCF7 cells. In TAMR^M cells, combination treatment resulted in similar changes to ER, p21 and c-Myc expression compared to PCI treatment alone. In contrast, PT combination resulted in a greater than additive decrease in Bcl-2 expression.

ER regulates BCL-2 transcription in TAMR^M cells (Figure 3C) similar to that observed in parental MCF7 cells [23]. Depletion of ER using increasing concentrations of ESR1-directed siRNA (Figure 5C) resulted in a commensurate decrease of BCL-2 mRNA in the resistant cells, demonstrating that the ER continues to regulate BCL-2 transcription. However, physiological levels of E2 (0.1 nM) or E2 + Tam (0.1 nM E2 and 10 μM Tam) fail to modulate BCL-2 transcription in TAMR^M cells, in contrast to MCF7 cells (Figure 5D). To find out whether this was due to reduced sensitivity of the ER to ligands, TAMR^M cells were treated with increasing concentrations (0.1, 1 and 10 nM) of E2. A statistically significant increase in BCL-2 mRNA was observed with only 10 nM E2, but not with lower E2 doses (Figure 5E). To examine the influence of PCI on the ER response to E2, BCL-2 mRNA was evaluated in the presence of 200 nM PCI and increasing E2 doses. PCI (200 nM) slightly reduced BCL-2 mRNA, potentially by downregulating ER expression, which was significantly reversed with the addition of a very high E2 (10 nM) dose (Figure 5E). This suggested that the ER in TAMR^M cells are less sensitive to E2 and require more E2 to modulate expression of BCL-2. To evaluate whether HDAC inhibition enabled Tam to reduce ER-mediated BCL-2 transcription, TAMR^M cells were treated with DMSO (C), 10 μM Tam, 200 nM PCI or PT for 24 hours and BCL-2 mRNA expression was determined (Figure 5F). Tam alone had no significant effect on BCL-2 mRNA. Treatment with PCI reduced BCL-2 mRNA modestly (approximately 30%), even though the ER was reduced by approximately 75% at this dose. Higher concentrations of PCI, beyond the clinically feasible dose of 200 nM, did significantly deplete BCL-2 mRNA (Figure S4 in Additional file 4), consistent with the increased rates of cell death at those doses. However, combined PT treatment reduced BCL-2 mRNA levels by approximately 70% (Figure 5F). To determine whether these effects were specific to PCI, other HDAC inhibitors, valproic acid (2 mM) and panobinostat (20 nM) were evaluated. Both inhibitors alone did not significantly decrease BCL-2 mRNA, however, in combination with Tam, BCL-2 mRNA was significantly reduced.

The combination of HDAC and ER inhibition causes cell death in MCF7 and TAMR^M cells (Figure 4). HDAC inhibitors are known to induce pro-apoptotic proteins, which we have previously shown with PCI-treated MCF7 cells [13]. Thus, we hypothesized that downregulation of Bcl-2 (Figure 5) combined with upregulation of pro-apoptotic factors together drive TAMR^M cells into apoptosis. To test this hypothesis, TAMR^M cells were treated with increasing concentrations of PCI for 72 hours and evaluated for expression changes in key apoptotic proteins (Figure 6A). In response to increasing PCI doses, several pro-apoptotic proteins, including Bax, Bak, Bok and cleavage of Bid, exhibited pro-apoptotic expression changes accompanied by a modest increase in PARP cleavage. To evaluate the effect of combined HDAC and ER inhibition on the expression of these proteins and the induction of apoptosis, MCF7 and TAMR^M cells were treated with 200 nM PCI (P), 10 μM Tam (T), or the combination for 72 hours. Changes to apoptotic proteins in both MCF7 and TAMR^M cells were primarily seen with PCI treatment and not Tam, with the combination treatment eliciting similar changes to PCI treatment alone (Figure 6B). An exception seen in both MCF7 and TAMR^M cells was Bax expression, which exhibited a greater than additive increase. Furthermore PARP cleavage was significantly enhanced in both cell lines with PT combination, confirming apoptotic cell death. Taken together these results suggest that HDAC inhibition can reverse Tam resistance by depletion of the major pro-survival protein Bcl-2 and upregulation of pro-apoptotic factors, driving cells into apoptosis.

Several studies have demonstrated that overexpression of the pro-survival protein Bcl-2 confers resistance to apoptosis by apoptotic inducers in several cancers, including breast [28-30]. Similarly, apoptosis induction in TAMR^M was much lower compared to MCF7 cells when challenged with increasing doses of the apoptotic stress inducers, Tam or the anthracycline doxorubicin (Figure S5 in Additional file 5). To establish Bcl-2 as a major contributor to apoptotic resistance in TAMR^M cells, the Bcl-2 family member inhibitor ABT.263 was employed. Treatment of TAMR^M cells with doses of ABT.263 up to 10 μM resulted in modest PARP cleavage (Figure 7A), consistent with previous reports that demonstrate similar inhibitors alone are insufficient to induce apoptosis in breast cancer cells [31]. However, when combined with PCI (200 nM), 1 μM ABT.263 was sufficient to substantially induce PARP cleavage and cell death in TAMR^M cells (Figure 7A and B). Furthermore, combined ABT.263 and PCI treatment resulted in cell death comparable to combined Tam and PCI treatment in both MCF7 and TAMR^M cells (Figure 7B). To specifically implicate Bcl-2 as the key target, Bcl-2 was first depleted in TAMR^M cells. Analysis of mRNA expression revealed about 50% transient knockdown of BCL-2 mRNA (Figure 7C). These cells were then treated with PCI for 72 hours. Combined Bcl-2 depletion and HDAC inhibition was sufficient to induce significant cell death in TAMR^M cells (Figure 7D). However, genetically depleting BCL-2 in MCF7 cells followed by PCI treatment did not lead to additional cell death compared to PCI treatment alone. Together, these results demonstrate that Bcl-2 and HDAC inhibition is sufficient to induce cell death in TAMR^M cells.