Addition of a histone deacetylase inhibitor redirects tamoxifen-treated breast cancer cells into apoptosis, which is opposed by the induction of autophagy

Chemicals and antibodies

Valproic acid, chloroquine, fulvestrant and 17β-estradiol (E2) were purchased from Sigma-Aldrich (St. Louis, MO). Vorinostat (suberoylanilide hydroxamic acid) was provided by Aton Pharma (Whitehouse Station, NJ). 4-OH-tamoxifen was purchased from Calbiochem (San Diego, CA). ERα and PR antibodies were purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Bcl-2, Bax, Bak and PARP antibodies were purchased from Cell Signaling technology (Danvers, MA). Beclin-1 antibody was purchased from ProSci Inc. (Poway, CA). HDAC2 antibody was purchased from Upstate Biotechnology (Chicago, IL). LC3 antibody was purchased from Novus Biologicals (Littleton, CO). Cytochrome C antibody was provided with the ApoAlert Cell Fractionation kit (Clontech, Mountain View, CA). GAPDH antibody was purchased from Chemicon (Temecula, CA).

Cell Culture

All cell lines were purchased from the American Type Culture Collection (Manassas, VA) and maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Fisher Scientific, Atlanta, GA) with 10% heat inactivated fetal bovine serum (FBS, Sigma-Aldrich), 2 mM glutamine, and 50 unit/mL penicillin and 50 μg/mL streptomycin (Fisher Scientific). Cells were incubated in a humidified atmosphere with 5% CO₂ at 37°C. For experiments evaluating the effect of estrogen concentration, cells were maintained in phenol red free DMEM supplemented with 10% dextran-charcoal stripped FBS (Invitrogen, Carlsbad, CA). Prior to the initiation of these experiments, cells were conditioned by growing them in the absence of estrogen 48 hours.

siRNA Depletion

siRNA duplexes for ERα (ID#42835) and Beclin-1 (sc-29797) were purchased from Ambion (Austin, TX) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Cells were transfected by nucleofection using the Nucleofector transfection kit according to the manufacturer’s recommendations (Amaxa, Gaithersburg, MD). Briefly, cells (4×10⁶) were suspended in 0.1 mL electroporation buffer containing 1 μM siRNA and pulsed with a cell line specific protocol as described by the manufacturer. Pulsed cells were suspended in 0.5 ml complete media without antibiotics and incubated at 37°C prior to seeding. Experiments were conducted as indicated the following day. The Silencer negative control #2 siRNA (Ambion, ID#4613), a nonsense siRNA duplex, was used as a control.

Cell Proliferation, Viability, and Apoptosis Assays

5×10³ cells/well were seeded in 96-well plates and drugged the following day as indicated. Each treatment was conducted at least in triplicate. At the indicated time, cell proliferation was assayed using Celltiter 96 AQueous one solution as per manufacturer’s instructions (Promega, Madison, WI). Absorbance was measured at 492 nm. Cell viability was determined by dye exclusion assay. Briefly, adherent and floating cells were combined and trypan blue solution (Hyclone, Logan, UT) was added 1:1 (v/v) and incubated for 1 minute at RT. Cells excluding dye were scored as viable. A minimum of 100 cells were scored for each treatment, which was conducted at least in triplicate. To assay for apoptosis, cells were harvested and washed in PBS. Cell nuclei were stained with 0.5 μg/ml bis-benzimide. Using fluorescent microscopy, cells were scored as apoptotic when nuclear chromatin appeared condensed or fragmented. Two hundred cells were counted per experiment in at least six different fields.

Western Blot Analysis

Cells were harvested, washed with ice-cold PBS, and solubilized using SDS lysis buffer (0.1% SDS, 1% triton X-100, 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 10% glycerol, Halt Protease and Phosphatase inhibitor cocktail (Thermo Scientific, Rockford, IL), and 1 mM PMSF) by passing the cell suspension through a 20-gauge syringe needle. The lysate protein concentration was determined by Bradford assay. Protein was separated by SDS-PAGE using 4–12% Bis-tris NuPage gels (Invitrogen) and transferred to nitrocellulose membranes for 1 hour at 100 V at 4°C. Membranes were blocked in tris-buffered saline containing 0.05% Tween 20 (TBST) and 5% nonfat milk (w/v) for 1 hour at room temperature (RT), followed by incubation with the indicated primary antibody (1:1000 v/v) in TBST with 5% nonfat milk, overnight at 4°C. Membranes were washed thrice for 10 minutes with TBST, followed by incubation with the appropriate secondary antibody (1:10000) in TBST with 5% nonfat milk for 1 hour at RT. Protein was visualized by chemiluminescent autoradiography using the Amersham ECL plus detection system (GE Healthcare, Buckinghamshire, UK).

Cell fractionation

The cytoplasmic fraction was separated from mitochondria using the ApoAlert Cell Fractionation Kit according to the manufacturer’s protocol (Clontech, Mountain View, CA). Briefly, ~3×10⁷ cells were harvested by trypsinization, washed with ice-cold PBS, followed by washing with Cell Wash Buffer. Washed cells were incubated in Cell Fractionation buffer on ice for 10 minutes. Cells were then dounce homogenized, 50 strokes. Cell homogenate was centrifuged at 700 × g for 10 minutes at 4°C to pellet debris. The supernatant was centrifuged at 10000 × g for 25 minutes at 4°C to pellet mitochondria. The supernatant was collected and saved as the cytoplasmic fraction. Cytoplasmic cytochrome C was quantified by western blot.

HDAC inhibitors increase both apoptosis and autophagy in tamoxifen treated breast cancer cells

Our previous work showed that the addition of an HDAC inhibitor potentiated the effects of tamoxifen in ER-positive breast cancer cells [18]. Forty-eight hours following co-administration of an HDAC inhibitor (e.g. valproic acid or vorinostat) and tamoxifen (10 μM), T47D and MCF-7 cells exhibit a substantial increase in apoptotic nuclei and cell death compared to untreated cells and cells treated with either agent alone. (Supplemental Figure 1 and data not shown). Tamoxifen is a known inducer of autophagy in breast cancer cells and recently has been implicated in the development of tamoxifen resistance. This raised the possibility that the synergy observed with the addition of an HDAC inhibitor may in part be due to inhibition of autophagy by HDAC inhibitors. To evaluate this possibility, T47D and MCF-7 breast cancer cells were treated with VPA and tamoxifen and examined for the induction of apoptosis and autophagy over time measuring PARP cleavage and LC3 (specific to autophagosomes) (Figure 1A and data not shown). As seen previously, tamoxifen treatment was associated with an increase of LC3, yet therapeutic doses of VPA or other HDAC inhibitors (vorinostat and panobinostat) did not induce the expression of LC3 by themselves. However, the addition of VPA to tamoxifen was associated with LC3 induction beyond the effects seen with tamoxifen alone. A notable increase in LC3 was observed following 8 hours treatment and further increased over time. The anti-tumor activity of tamoxifen has been shown previously to be primarily anti-proliferative rather than apoptotic. Consistent with these findings, neither tamoxifen nor VPA treatment alone was associated with PARP cleavage (Figure 1) or apoptotic nuclei (supplemental figure). However, when treated with the combination, cells exhibited significant PARP cleavage by 48 hours, which increased further with time.

To determine whether these findings were specific to valproic acid, we tested whether other HDAC inhibitors increase the autophagy observed with tamoxifen treatment. T47D cells were treated with the pan-HDAC inhibitor, vorinostat (1 μM) or panobinostat (30 nM), tamoxifen (10 μM), or the combination of each HDAC inhibitor with tamoxifen. Following 48 hours treatment, LC3 levels were evaluated by western blot (Figure 1B). Alone, LC3 levels in cells treated with either HDAC inhibitor were comparable to untreated cells, while tamoxifen treatment induced a modest increase in LC3. However, when combined with tamoxifen, both vorinostat and panobinostat elicited increased LC3 levels compared to cells treated with tamoxifen alone, with the panobinostat combination yielding a more robust increase.

Together, these results demonstrate that when combined with tamoxifen, HDAC inhibitors potentiate tamoxifen-induced LC3 expression, a component specific to autophagosomes and a hallmark of autophagy. Furthermore, LC3 induction precedes apoptosis induced by the addition of an HDAC inhibitor to tamoxifen.

In the presence of an HDAC inhibitor, cells undergo apoptosis despite tamoxifen-induced autophagy

The evaluation of LC3 induction and PARP cleavage in cells treated with an HDAC inhibitor and tamoxifen suggested a temporal order of induction, with autophagy preceding apoptosis. To determine whether autophagy induction is a ubiquitous event that precedes apoptosis in all cells or whether autophagy and apoptosis are independent processes occurring in different cell populations, we treated T47D cells with VPA (0.5 mM) and tamoxifen (10 μM) and evaluated LC3 and PARP cleavage over time by flow cytometry (Figure 2). Thus, the induction of both autophagy and apoptosis could be examined in individual cells. Following 24 hours treatment with tamoxifen or tamoxifen + VPA, 4–6% of cells exhibited elevated LC3 relative to untreated cells, while PARP cleavage was negligible. This trend was similar at 48 hours. By 96 hours however, 11.5% and 12.2% of cells expressed increased LC3 in the tamoxifen and tamoxifen + VPA treatment respectively. However, at this time point, apoptosis measured by cleaved PARP was evident in 21% of the cells treated with the combination, while with tamoxifen treatment alone the percentage of cells with increased PARP cleavage was comparable to untreated cells. Consistent with western blot analysis, treatment with VPA alone did not significantly affect the percentage of cells expressing elevated levels of LC3 or cleaved PARP. Surprisingly, only 5–7% of cells co-expressed of LC3 and cleaved PARP when treated with the combination of VPA and tamoxifen. Taken together, these results do not support a temporal progression from autophagy to apoptosis when cells are treated with an HDAC inhibitor and tamoxifen. Rather, the presence of an HDAC inhibitor evokes the induction of apoptosis in cells exposed to tamoxifen despite the induction of autophagy as a survival strategy.

HDAC inhibitors and tamoxifen induce autophagy in part through ER signaling

Tamoxifen is a selective ER modulator that competes with estrogen for ER binding. The therapeutic efficacy of tamoxifen is thought to be the result of altered ER mediated gene transactivation [1, 24]. Thus, we sought to examine the effect of modulating ER signaling on the observed induction of autophagy with combined HDAC inhibitor and tamoxifen treatment. We first evaluated the effect of high and low dose estrogen on the induction of autophagy. For these experiments, T47D cells were maintained in estrogen stripped media, to which no or 1 nM estradiol was added together with VPA (0.5 mM), tamoxifen (10 μM), or the combination for 48 hours (Figure 3A). In the absence of additional estradiol, both tamoxifen only treatment and combined treatment with tamoxifen and VPA induced autophagy. However, the marked difference in LC3 expression between these two conditions, seen when estrogen is not stripped from the growth medium, does not appear to be as pronounced when estrogen is depleted. When grown in the presence of high estradiol (1 nM), LC3 induction by tamoxifen alone is substantially reduced, comparable to untreated and VPA treated cells. However, the presence of high estrogen does not substantially abrogate LC3 induction when cells are treated with the combination of VPA and tamoxifen. As previously shown [25], treatment with tamoxifen, regardless of estrogen level, stabilizes the ER. These results suggest that tamoxifen stimulated autophagy is mediated through ER signaling. With the addition of an HDAC inhibitor, the competitive inhibitory effect of estrogen on autophagy induction is overcome.

We next sought to determine whether combining ER depletion with an HDAC inhibitor could mimic the observed induction of autophagy. To this end, we treated T47D cells with VPA (0.5 mM), the selective ER down regulator fulvestrant (0.1 μM), or the combination for 48 hours in the presence low and high estrogen levels (Figure 3B). As expected, cells treated with fulvestrant or fulvestrant and VPA exhibited a substantial decrease in ER at both low and high estrogen levels compared to VPA alone or untreated cells. However, an increase in LC3 was not observed. To further explore the importance of the ER for the induction of autophagy, we depleted ER by siRNA in T47D cells and treated them with VPA (0.5 mM), tamoxifen (10 μM), or the combination for 48 hours (Figure 3C). In support of the experiment with fulvestrant, we found that cells treated with tamoxifen or the combination of tamoxifen and VPA had reduced induction of LC3 when ER was depleted compared to cells transfected with a scramble siRNA control. The requirement of ER for tamoxifen and VPA to fully induce autophagy suggests that this response would not be seen in ER-negative breast cancer cells. To test this hypothesis, we treated the ER-negative MDA-MB-231 cell line with VPA (0.5 mM), tamoxifen (10 μM) or the combination for 48 hours (Figure 3D). Cells treated with tamoxifen or tamoxifen and VPA exhibited a slight increase in LC3 compared to untreated cells, but not comparable to that seen with ER-positive T47D and MCF-7 cell lines (Figure 1 and data not shown). Together, these results argue that the full induction of autophagy seen with tamoxifen and an HDAC inhibitor requires tamoxifen mediated ER signaling, and not simply the inhibition or degradation of the ER.

Treatment with tamoxifen and an HDAC inhibitor alters the balance of autophagy and apoptosis inhibitors and their drivers

Several studies have shown that tamoxifen treatment may induce autophagy in ER-positive breast cancer cells, possibly contributing to the development of anti-estrogen resistance [19, 22, 23]. This is believed to be in part the result of up-regulating Beclin-1 mRNA and its protein product, an essential component of the autophagic program that when over-expressed can drive production of autophagosomes [26]. Furthermore, it has been shown that the induction of autophagy can effectively inhibit apoptosis. Several groups have postulated that this is due to induction of apoptosis inhibitors, such as BCL2, and depression of apoptosis inducers, such as Bax. To define the link between autophagy and apoptosis, we examined the effects of HDAC inhibition on the balance of autophagy and apoptosis inducers and inhibitors.

As HDACs are key regulators of gene expression, we hypothesized that their inhibition may further result in increased Beclin-1 expression and thus contribute to the observed induction of autophagy. To test this hypothesis, we treated T47D cells with VPA (0.5 mM), tamoxifen (10 μM), or the combination for 48 hours and harvested cells to evaluate Beclin-1 protein and RNA levels (Figure 4A and B). As previously shown, tamoxifen treated cells had both increased Beclin-1 protein and RNA compared to untreated cells. Cells treated with VPA also had increased Beclin-1 protein, less robust than with tamoxifen, however, RNA levels were unchanged compared to untreated cells. In cells treated with the combination of VPA and tamoxifen, the induction of Beclin-1 exceeded that of either treatment alone. Thus, the effect of VPA on Beclin-1 is predominantly post-translational, as treatment with the combination does not significantly effect Beclin-1 mRNA compared to treatment with tamoxifen alone.

When ER was depleted by siRNA, induction of Beclin-1 protein was substantially reduced in cells treated with tamoxifen and the combination of VPA and tamoxifen compared to siRNA scramble controls (Figure 3C). Together these results suggest that tamoxifen induces Beclin-1 expression, at least in part, through increased transcription mediated by ER signaling, while VPA acts on Beclin-1 protein stability resulting in a more than additive interaction.

Pro-autophagic signaling of Beclin-1 is regulated by Bcl-2, which binds to the BH3 domain of Beclin-1 through its BH3 binding groove, inhibiting autophagy signaling. Perturbing this interaction, through stress response pathways or by other means, results in autophagy induction [27]. Thus, we measured Bcl-2 protein levels following treatment with VPA (0.5 mM), tamoxifen (10 μM), or the combination for 48 hours (Figure 4 C). Indeed, when treated with the combination of VPA and tamoxifen, a substantial decrease in Bcl-2 protein was observed. Treatment with either VPA or tamoxifen alone resulted in a negligible change in Bcl-2 protein level.

In addition to regulating autophagy through the inhibition of Beclin-1, Bcl-2 also inhibits the pro-apoptotic function of Bax and Bak. When inhibition is relieved, Bax and Bak act to compromise mitochondrial membrane potential, leading to cytochrome C release into the cytosol where it subsequently promotes activation of execution caspases and cellular digestion [28, 29]. Elevated levels of Bax and Bak, relative to Bcl-2 and Bcl-2 like apoptotic inhibitors, can also lead to cytochrome C release and apoptosis. To determine whether combining an HDAC inhibitor and tamoxifen effect levels of these downstream pro-apoptotic proteins, we evaluated their expression in T47D cells by western blot following 48 hours treatment with VPA (0.5 mM), tamoxifen (10 μM), or the combination (Figure 4D and E). The combination resulted in both increased Bax and Bak compared to untreated cells. Elevated levels of Bax induced by the combination of VPA and tamoxifen, while either drug alone had negligible effects on Bak levels compared to untreated cells. To determine whether the drug combination further results in compromised mitochondia and cytochrome C release, we fractionated the cytosol from the mitochondria in these cells and measured cytochrome C in the cytosol by western blot (Figure 4F). Compared to untreated cells, only cells treated with the combination of VPA and tamoxifen exhibited increased cytochrome C levels, consistent with our observation that only the combination results in PARP cleavage (Figure 1A).

Together, these results suggest that with the addition of an HDAC inhibitor, tamoxifen treated cells exhibit enhanced autophagic and apoptotic response. In part, this is due to perturbing the balance between the master inhibitor, Bcl-2, and downstream drivers of autophagy, Beclin-1, and apoptosis, Bax and Bak.

Autophagy inhibits the cytotoxicity of VPA and tamoxifen in T47D cells

It has recently been demonstrated that when ER-positive breast cancer cells are treated with tamoxifen, autophagy is induced. In these cells, inhibition of the autophagic program increases cytotoxicity and thus suggests autophagy is promoting survival in response to the insult. However, cells experiencing prolonged elevated autophagy may undergo autophagic cell death. Therefore, we sought to determine whether VPA and tamoxifen induced autophagy was contributing to or inhibiting the cytotoxicity of this combination. First, we evaluated the effect of combining chloroquine with VPA and tamoxifen. Chloroquine preferentially accumulates in the lysosome, raising the pH, and inhibiting subsequent fusion with autophagosomes and degradation of their contents [30]. T47D cells were treated with VPA (0.5 mM), tamoxifen (10 μM), or the combination and either 0 or 10 μM chloroquine. In cells treated with chloroquine, autophagosome turnover was inhibited, exhibited by substantially elevated LC3 levels for all treatments compared to cells not treated with chloroquine (Figure 5A). An assay of cell viability revealed that treatment with chloroquine increased cytotoxicity of tamoxifen when administered alone (12% vs. 43% cell death) or the combination of VPA and tamoxifen (39% vs. 78% cell death) (figure 5B). To more directly assess the role of autophagy, Beclin-1 was depleted by siRNA in T47D cells prior to treatment with VPA and tamoxifen (Figure 5C). Consistent with chloroquine inhibition, when Beclin-1 was depleted, cell death increased nearly 20% with VPA and tamoxifen treatment compared to cells transfected with scramble siRNA and treated with VPA and tamoxifen (Figure 5D).

The findings from these experiments suggest that cells in response to tamoxifen exhibit growth arrest, while a subset of cells induce autophagy in an attempt to escape the effects of tamoxifen. Unlike tamoxifen alone, the addition of an HDAC inhibitor can effectively induce apoptotic cell death by synergistic depletion of Bcl-2 and up-regulation and activation of pro-apoptotic genes such Bax and Bax. However, our data further suggests that a subpopulation of cells may protect themselves from apoptosis by autophagic induction. To avoid the emergence of a resistant cell subpopulation, autophagy must be inhibited. Thus, adding an autophagy inhibitor to an HDAC inhibitor and tamoxifen may represent a more effective treatment regimen for ER-positive tumors.