Sensitivity to HDAC inhibitors in parental and eribulin-resistant TNBC cells
To evaluate potential growth-inhibitory effects by HDAC inhibitors (VOR and RICO) on TNBC cells in vitro, MDA-MB-231, Hs578T, and MDA-MB-157 cells were treated with VOR or RICO for 72 h, and cell viability was measured by performing WST assays (Additional file 1; Table S1). The IC50 of VOR for MDA-MB-231, Hs578T, and MDA-MB-157 cells was 1.8 ± 0.4 µM, 1.3 ± 0.5 µM, and 1.6 µM ± 0.3 µM, respectively. On the other hand, the IC50 of RICO for these three cell lines was 2.0 ± 0.5 µM, 1.8 ± 0.3 µM, and 2.4 ± 0.4 µM, respectively. There were no significant differences in the IC50 of VOR or RICO among these three cell lines (Additional file 2; Fig. S1).
Next, we investigated the growth-inhibitory effects of HDAC inhibitors on eribulin-resistant TNBC cells (MDA-MB-231/E, Hs578T/E, MDA-MB-157/E), which were established previously in our laboratory [30]. The IC50 of VOR for MDA-MB-231/E, Hs578T/E, and MDA-MB-157/E cells was 2.0 ± 0.5 µM, 1.8 ± 0.3 µM, 1.8 ± 0.4 µM, respectively. Meanwhile, the IC50 of RICO for these three eribulin-resistant TNBC cell lines was 1.8 ± 0.5 µM, 2.2 ± 0.3 µM, and 2.2 ± 0.6 µM, respectively. Thus, no significant differences in the IC50 of VOR or RICO were observed among the three eribulin-resistant TNBC cell lines (Additional file 2; Fig. S1). Moreover, no cross-resistance to eribulin and VOR or RICO was observed in both between these parental cells and eribulin-resistant cells.
Low concentrations of VOR or RICO enhance the anti-tumor effect of eribulin in MDA-MB-231 and Hs578T cells
Next, we analyzed whether the co-administration of low concentrations of VOR or RICO could enhance the anti-tumor effect of eribulin on TNBC cells. The concentrations of co-administrated VOR or RICO were determined to be 0.2 µM and 0.5 µM because we confirmed that these concentrations do not affect cell growth as a single agent before this experiment (Additional file 2; Fig. S1). The growth-inhibitory effect of eribulin was enhanced when low concentrations (0.2 or 0.5 µM) of VOR or RICO were simultaneously added to MDA-MB-231 and Hs578T cells. However, in MDA-MB-157 cells, low-dose VOR or RICO did not enhance sensitivity to eribulin (Fig. 1a). Isobologram analysis of VOR and eribulin results demonstrated that each experimental data point was located below the diagonal line for MDA-MB-231 cells and Hs578T cells, indicating that VOR and eribulin acted synergistically (Additional file 3; Fig. S2). In contrast, when we examined the growth-inhibitory effect of combined eribulin and HDAC inhibitors on MCF7 cells, RICO was not found to enhance eribulin sensitivity (Additional file 4; Fig. S3).
Next, we examined the induction of apoptosis after single treatment with eribulin, VOR, or RICO, as well as a combination of eribulin with VOR or RICO, in MDA-MB-231 and Hs578T cells. Whereas the administration of 0.5 µM VOR or RICO for 48 h did not induce apoptosis in MDA-MB-231 and Hs578T cells, 1 nM of eribulin induced apoptosis significantly compared to that in cells treated with DMSO alone. Notably, the addition of 0.5 µM VOR or RICO to 1 nM eribulin significantly enhanced the induction of apoptosis compared to that induced by monotherapy comprising 1 nM of eribulin (Fig. 1b).
Next, to gain further insight into apoptosis induction by eribulin and HDAC inhibitors, we analyzed alterations in the levels of Bcl-2, which is known as an anti-apoptotic protein, in TNBC cell lines (Fig. 1c). Whereas the administration of VOR or RICO did not change the expression of Bcl-2, treatment with eribulin (1 nM) decreased the expression of Bcl-2. Furthermore, the addition of VOR or RICO to eribulin enhanced this decrease in Bcl-2 expression compared to that induced by eribulin monotherapy, which was concordant with the results of apoptosis assays mentioned previously herein. These results indicate that low concentrations of VOR or RICO enhance the anti-tumor effect of eribulin by augmenting eribulin-mediated induction of apoptosis in MDA-MB-231 and Hs578T cells.
Low concentrations of VOR or RICO restore eribulin-resistance in eribulin-resistant MDA-MB-231 and Hs578T cells
As we found that low concentrations of HDAC inhibitors could enhance sensitivity to eribulin in the parental MDA-MB-231 and Hs578T cells, we next examined whether HDAC inhibitors could restore eribulin sensitivity in three eribulin-resistant TNBC cell lines (MDA-MB-231/E, Hs578T/E, MDA-MB-157/E). The co-administration of a low concentration (0.2 µM or 0.5 µM) of VOR or RICO partially restored eribulin sensitivity in MDA-MB-231/E and Hs578T/E cells, though this did not reach the level of eribulin sensitivity in parental cells. However, as in parental cells, a low concentration of VOR or RICO did not alter the sensitivity of eribulin in MDA-MB-157/E cells (Fig. 2a).
Next, we examined whether VOR or RICO could enhance the induction of apoptosis in eribulin-resistant MDA-MB231 and Hs578T cells, as was observed for their parental cells. Eribulin monotherapy for 48 h (3 nM for MDA-MB-231/E cells, 70 nM for Hs578T/E cells) induced apoptosis significantly in both eribulin-resistant cell lines. Whereas the addition of 0.5 µM of VOR or RICO for 48 h did not induce apoptosis in the eribulin-resistant MDA-MB-231 and Hs578T cells, the combination of eribulin with the same dose of VOR or RICO significantly enhanced the induction of apoptosis in these cell lines (Fig. 2b). These results indicate that a low concentration of VOR or RICO can enhance apoptosis induced by eribulin and restore sensitivity to eribulin in the eribulin-resistant MDA-MB-231 and Hs578T cells, as well as in their parental cells.
EMT is not involved in enhanced eribulin sensitivity induced by HDAC inhibitors
As we previously reported that EMT induction enhances eribulin sensitivity in a subset of TNBC cell lines (MDA-MB-231 and Hs578T) [14], we also examined whether VOR or RICO treatment would alter EMT markers in these cell lines. ZEB1, vimentin, and Slug were studied as mesenchymal markers, whereas E-cadherin was studied as an epithelial marker. Western blotting demonstrated that the administration of 0.5 µM VOR or RICO did not alter the expression of these mesenchymal markers in MDA-MB-231 and Hs578T cells. E-cadherin expression was not detected in both cell lines treated with DMSO alone and 0.5 µM of VOR or RICO (Fig. 3a).
We previously reported that the siRNA-mediated knockdown of ZEB1, which serves as a transcriptional activator of mesenchymal differentiation, confers eribulin resistance to MDA-MB-231 and Hs578T cells, indicating that the EMT–MET axis is involved in eribulin sensitivity [14]. However, the involvement of the EMT–MET axis in sensitivity to HDAC inhibitors has not been elucidated. Hence, we tested whether ZEB1 inhibition by siRNA could alter VOR or RICO sensitivity in three TNBC cell lines. The inhibition of ZEB1 expression by siRNA was confirmed by western blotting (Fig. 3b). In all three cell lines, no difference in the growth-inhibitory effect of VOR or RICO was observed between cells treated with control siRNA and those treated with siRNA targeting ZEB1 (Fig. 3c). These results indicate that the administration of HDAC inhibitors does not alter the EMT–MET phenotype in TNBC cells; moreover, the EMT–MET axis is less likely to be associated with TNBC cell sensitivity to HDAC inhibitors.
Low concentrations of HDAC inhibitors and eribulin increase the expression of acetylated α-tubulin in MDA-MB-231 and Hs578T cells, but not in MDA-MB-157 cells
Next, to further investigate the mechanism underlying the increase in eribulin sensitivity induced by VOR and RICO, we examined differences in acetylated α-tubulin expression caused by VOR or RICO treatment for 48 h among these cell lines. Western blotting demonstrated that VOR or RICO addition increased the expression of acetylated α-tubulin in a dose-dependent manner in MDA-MB-231 and Hs578T cells. In contrast, although 0.2 and 0.5 µM VOR or RICO did not alter the expression of acetylated α-tubulin, 5 µM VOR or RICO drastically increased the expression of acetylated α-tubulin in MDA-MB-157 cells. A similar change in the expression of acetylated α-tubulin observed in parental TNBC cell lines was induced in their eribulin-resistant sublines (Fig. 4a).
Then, we examined alterations in the expression of acetylated α-tubulin induced by eribulin in TNBC cell lines because paclitaxel, which is another anti-tubulin agent, has been reported to increase the expression of this marker [33]. The expression of acetylated α-tubulin was increased in MDA-MB-231 and Hs578T cells in a dose-dependent manner after the addition of 0.5, 1, and 2 nM eribulin. However, eribulin at these concentrations did not alter the expression of acetylated α-tubulin in MDA-MB-157 cells (Fig. 4b).
Next, we analyzed the alteration of acetylated α-tubulin expression when eribulin was administered in combination with VOR or RICO to the parental TNBC cell lines. Combination therapy comprising 0.5 nM eribulin with 0.5 µM VOR or RICO additively upregulated the expression of acetylated α-tubulin in MDA-MB-231 and Hs578T cells, whereas no effect on the expression of acetylated α-tubulin was observed in MDA-MB-157 cells (Fig. 4c). These results indicate that both eribulin and HDAC inhibitors might increase the acetylation of α-tubulin in MDA-MB-231 and Hs578T cells but not in MDA-MB-157 cells.
Increasing the expression of acetylated α-tubulin through VOR or RICO pre-treatment enhances eribulin sensitivity in TNBC cells
Next, we examined whether the upregulation of acetylated α-tubulin induced by VOR or RICO pre-treatment could alter the sensitivity to eribulin. Results showed that 5 µM VOR or RICO could increase the expression of acetylated α-tubulin in MDA-MB-231, Hs578T, and MDA-MB-157 cells. However, all three cell lines treated with 5 µM VOR or RICO for 72 h did not survive, as demonstrated in Fig. S1, whereas the same dose of VOR or RICO for 48 h did not inhibit cell proliferation, and the cells grew similarly to those without treatment after passage (Additional file 5; Fig. S4). Therefore, 5 µM VOR or RICO was used for this experiment. After pre-treating MDA-MB-231, Hs578T, and MDA-MB-157 cells with 5 µM VOR or RICO for 48 h, the pre-treated cells were seeded in a 96-well plate and tested for sensitivity to eribulin (Fig. 5a). After incubation for another 72 h, cell viability was measured. As a result, pre-treatment with VOR or RICO for 48 h enhanced sensitivity to eribulin in all three cell lines (Fig. 5b).
Furthermore, the effect HDAC inhibitor pre-treatment on sensitivity to eribulin was tested in eribulin-resistant TNBC cells. As shown in Fig. 5c, pre-treatment with VOR or RICO enhanced eribulin sensitivity in the three eribulin-resistant TNBC cell lines (MDA-MB-231/E, Hs578T/E, and MDA-MB-157/E cells), as observed in their parental cell lines (Fig. 5c). These results indicate that the increase in the expression of acetylated α-tubulin mediated by VOR or RICO pre-treatment enhances TNBC cell sensitivity to eribulin.
Increased expression of acetylated α-tubulin induced by eribulin treatment in clinical TNBC specimens
As our in vitro results suggested the possibility that α-tubulin acetylation might be increased by eribulin treatment in a subset of TNBC cells (Fig. 4b), we next analyzed whether eribulin would increase the acetylation of α-tubulin in clinical TNBC specimens. The expression of acetylated α-tubulin was evaluated by immunohistochemical staining in clinical specimens obtained from breast cancer patients who underwent neoadjuvant treatment with eribulin. The tissue sections were obtained by core needle biopsy before the initiation of treatment and after four courses of treatment with eribulin from 26 breast cancer patients who enrolled in the JONIE-3 study. Regarding ER expression, 16 cases were ER-positive and eight were ER-negative. Of the eight cases of ER-negative breast cancer, six were TNBC. ER-negative cases tended to exhibit higher baseline expression of acetylated α-tubulin compared to that in ER-positive cases though there was no statistical significance (Fig. 6a). We then analyzed the change in acetylated α-tubulin expression with eribulin treatment in five TNBC cases because one case presented with a clinically complete response to neoadjuvant treatment with eribulin and thus a post-treatment specimen could not be obtained. In two cases that showed a high H-score before eribulin treatment (275 and 300), high expression of acetylated α-tubulin was maintained throughout treatment with eribulin. In the other three cases, H-scores were increased by eribulin treatment (Additional file 6; Table S2). The H-scores of six cases at each point, the rate of H-score change, and responses to eribulin treatment are shown in Table S2, and the representative findings of immunohistochemical analyses are shown in Fig. 6b–e.
Next, to investigate whether altered α-tubulin acetylation, induced by eribulin, was associated with ER status, we analyzed eribulin-mediated changes in acetylated α-tubulin expression in ER-positive and ER-negative patients. Although there was no significant change in acetylated α-tubulin expression in ER-positive breast cancer specimens (p = 0.994), the expression of acetylated α-tubulin significantly increased in ER-negative breast cancer specimens after treatment with eribulin (p = 0.012; Fig. 6f). Notably, no cases of ER-negative breast cancer were associated with decreases in the expression of acetylated α-tubulin expression, whereas this decreased in 37.5% of ER-positive breast cancers (Fig. 6g). These results indicate that the increased acetylation of α-tubulin induced by eribulin is correlated with ER signaling.
Furthermore, we examined whether altered acetylated α-tubulin expression induced by eribulin treatment is associated with the response to eribulin in TNBC because increased acetylated α-tubulin expression resulted in higher sensitivity to eribulin in vitro. Two of three TNBC patients who showed a partial response (PR) exhibited > 2-fold positive conversion of acetylated α-tubulin expression. Another patient with a PR had a high level of acetylated α-tubulin expression (H-score: 300) before treatment, and thus acetylated α-tubulin could not be upregulated. In contrast, two patients who did not show a response to eribulin (stable disease) maintained acetylated α-tubulin expression during eribulin treatment (Additional file 6; Table S2). These results suggest that an increase in acetylated α-tubulin expression was associated with a favorable response to eribulin treatment in TNBC patients.