miR-489 expression is lost in tamoxifen resistance, predicts breast cancer aggressiveness and is regulated by estrogen/ERα axis
To specifically identify miRNAs that are clinically relevant in endocrine resistance, we analyzed miRNA screening datasets of endocrine resistant models previously published by three independent laboratories including our own (Fig. 1A)(11,12,16). Since all three models are derived through completely different processes, they represent independent tamoxifen resistant models with different resistance mechanisms. The MCF7-HER2 cell line acquired resistance through the activation of the HER2 oncogenic pathway while MCF7-TAM and MCF7:2A represent acquired resistance through long-term culture in tamoxifen-containing and estrogen-deprived media respectively. miRNAs that are dysregulated in all three cell lines suggest their potential roles in the regulation of multiple mechanisms involved in tamoxifen resistance, which may therefore be more clinically relevant with promising therapeutic applicationto address tamoxifen resistance. Although many miRNAs were dysregulated in these cell lines, we only found few miRNAs that were aberrantly expressed in all three cell lines (Fig. 1B). Out of these miRNAs, miR-135b, miR-33b, and miR-505 showed an opposite expression pattern among these cell lines. miR-378a-3p and miR-218 were significantly upregulated in all three cell lines while miR-342-5p and miR-489 were significantly downregulated. Intriguingly, miR-489 was one of the top downregulated miRNAs in all three datasets, suggesting its role in tamoxifen resistance. We validated these results using qRT-PCR and indeed found significant downregulation of miR-489 in both resistant cell lines (Fig. 1C). To determine whether the expression of miR-489 was associated with endocrine resistance in patient cohorts, we examined miR-489 expression in hormone-therapy treated ER+ breast cancer patients. We observed a statistically significant association between lower miR-489 expression and poorer overall survival in these patients. In fact, miR-489 expression remained an independent prognostic factor in hormone-therapy treated breast cancer patients obtained from two independent datasets, GSE19783 and METBRIC (Fig. 1D). These results suggest that the loss of miR-489 may promote tamoxifen resistance.
Previously we observed that the average expression of miR-489 was notably higher in luminal cells compared to basal cells (13,16). Furthermore, analysis of miR-489 expression on 13 different breast cancer cell lines also demonstrated that it was expressed at a higher level in hormone positive luminal breast cancer cell lines (Fig. 1E) compared to the cell lines from other subtypes. miR-489 is an intragenic microRNA located in the intron region of CALCR. Analysis of primary breast tumors revealed that miR-489 expression positively correlated with the expression of CALCR, ESR1 and ER responsive genes such as PGR (Fig. 1F), suggesting that miR-489 expression may be regulated by estrogen signaling. To examine this hypothesis, we stimulated three ER+ breast cancer cell lines (T47D, MCF7, and BT474) with estrogen or ethanol for indicated time periods and measured the expression of miR-489 and its host gene CALCR. We found significant upregulation of miR-489 and CALCR in all three cell lines treated with estrogen (Fig. 1G). Estrogen regulation of miR-489 was further investigated in complete media or estrogen deprived media. As expected, depletion of estrogen drastically reduced the expression of miR-489 and CALCR like other classical ER target genes such as trefoil factor 1 (TFF1), progesterone receptor (PGR), and C-X-C motif chemokine ligand 12 (CXCL12) (Fig. 1H). In summary, this data strongly suggests that miR-489 is an estrogen regulated miRNA in breast cancer and may play a regulatory role in tamoxifen resistance.
miR-489 restoration overcomes tamoxifen resistance.
Since miR-489 was lost in tamoxifen resistant tumors and cell lines, we asked whether restoration of miR-489 would sensitize the resistant cell lines. Our previous studies have shown that overexpression of miR-489 inhibited breast cancer cells proliferation(13,16). We first tested whether Tamoxifen-resistant cells were still sensitive to miR-489 mimics. Notably, we observed that tamoxifen resistant cell lines, MCF7-TAM and MCF-HER2, were just as sensitive to miR-489 mimics as their sensitive counterparts, as opposed to their different sensitivities to tamoxifen (Fig. 2A, 2B). These results further bolstered the possibility that miR-489 might target pathways involved in resistance and could potentially sensitize these resistant cell lines to tamoxifen. Indeed, forced expression of miR-489 significantly sensitized both resistant cell lines to tamoxifen. Tamoxifen alone had no significant effect on the growth of both resistant cell lines at 5 µM, while combination with miR-489 led to around 40% growth inhibition of MCF7-TAM (Fig. 2C) and 30% growth inhibition in MCF7-HER2 cells (Fig. 2D). To determine whether miR-489 suppresses cell growth independently or has a synergy effect with tamoxifen, we performed a synergy analysis. In MCF7-TAM cells, the combination of miR-489 and tamoxifen achieved synergistic effects at high levels (fraction reduction > 0.2). However, in MCF7-HER2 cells, this combination showed a slight synergy (nearly additive effect) at most levels (Fig. S1A and 1B, additional file 2). In both cell lines, this combination could dramatically reduce the dose of either miR-489 or tamoxifen required to achieve the same level of growth inhibition. These results indicated that miR-489 and tamoxifen can synergistically inhibit cell growth in a cell-line dependent manner.
To further assess the role of miR-489 in tamoxifen resistance, we inhibited endogenous miR-489 in tamoxifen sensitive cells MCF7-Vec and MCF7-WT cells. As expected, inhibition of miR-489 significantly increased tamoxifen resistance in both sensitive cell lines. At the highest concentration tested, inhibition of miR-489 increased survival by 25% and 40% in sensitive counterpart of MCF7-Vec and MCF7-WT, respectively (Fig. 2E, 2F). Similarly, colony formation assays also revealed that forced expression of miR-489 significantly reduced the survival and colony forming ability in resistant cell lines, while inhibition of endogenous miR-489 enhanced the survival of sensitive counterpart and promoted tamoxifen resistance. (Fig. 2G). To rule out the possibility of the off-target effect of the miR-489 inhibitor, we utilized CRISPR/ Cas9 gene editing to create a miR-489 knockout cell line (Additional file, Fig. S2A). We validated these knock out cells using genotyping and sequencing to ensure the deletion of miR-489. As expected, miR-89 knockout cells exhibited increased growth rate (Additional file, Fig. S2B-2C). Consistently, miR-489 knockout cells also showed significant resistance to tamoxifen, evidenced by MTT based cell viability and colony formation assays (Fig. 2H and 2I). These results provide direct evidence to support that loss of miR-489 contributes to development of tamoxifen resistance.
miR-489 acts as an endogenous negative feedback regulator to balance estrogen signaling
To elucidate the underlying pathways targeted by miR-489 to induced tamoxifen sensitization, we re-examined the gene expression profiles of T47D cells transfected with miR-489 mimics and scrambled RNA(13). Interestingly, gene expression analysis revealed enrichment of multiple pathways involved in estrogen signaling and tamoxifen resistance (Fig 3A, 3B). We observed enrichment of ErbB signaling pathway and several stress associated pathways including endoplasmic reticulum (ER) stress and lysosomal pathways. All of these pathways have been previously reported to be involved in tamoxifen resistance (4,9,17-21). Meanwhile, these results are also in accordance with our previous studies showing the role of miR-489 in HER2 signaling, metabolic stress, and autophagy regulation (13,14,16). Interestingly, we noticed significant enrichment of estrogen-dependent gene expression and ESR-mediated signaling. Upon further analysis, we found substantial downregulation of estrogen responsive genes (Fig. 3C). This data suggested that estrogen regulated miR-489 might function as a negative regulator of estrogen signaling. To determine the clinical significance of this data, we applied a miR-489 gene expression signature to a gene expression profile obtained from patient datasets. Consistent with our microarray results, we found a strong inverse correlation between the estrogen responsive gene signature and the miR-489 signature and between the PI3K-ERBB2 signature and the miR-489 signature (Additional file 2, Fig. S3A-B). In addition, we noticed that low miR-489 expression in ER+ breast cancer was indicative of worse overall survival (Additional file 2, Fig. 3C), further supporting an essential tumor suppressive role of miR-489 in ER+ breast cancer.
To examine how miR-489 negatively regulated estrogen signaling, we measured transcriptional activity of estrogen receptors by performing a Luciferase reporter assay with a T47D-ERE-Luc reporter cell line. We observed the inhibition of estrogen receptor transcriptional activity upon miR-489 overexpression and increased activity upon inhibition of endogenous miR-489 (Fig. 3D). Consistently, gene expression levels of estrogen responsive genes were further increased upon estrogen stimulation in knock out cells compared to wild type cells (Fig. 3E). We then validated the microarray results by performing qRT-PCR analysis on ER+ and ER- cell lines. As expected, miR-489 caused down regulation of estrogen responsive genes only in ER+ cell lines including T47D, MCF7, and BT474 cells, but did not affect or in some instances increased the expression of these genes in ER- cell lines such as AU565 and HCT116 cells (Fig. 3F-G). These results suggest that miR-489 regulates the expression of these ERα-downstream genes by inhibiting estrogen signaling.
miR-489 inhibits ER-induced cell proliferation and cancer stem cells expansion
Our previous studies have shown that miR-489 inhibits proliferation of all breast cancer cell lines including ERα+ cell lines (13,16). However, it remains unknown whether miR489-mediated growth inhibition in ER+ breast cancer cells is due to its effects on ER signaling. To investigate this, we examined the effects of miR-489 on estrogen-induced cell proliferation in MCF7 and T47D cells. Both cell lines showed poor proliferation when treated with the vehicle (ethanol), while estrogen treatment enhanced proliferation by more than 2-fold and 4-fold in MCF7 and T47D cell lines respectively. Restoration of miR-489 completely abolished estrogen-induced proliferation in both cell lines while inhibition of endogenous miR-489 further increased estrogen-induced proliferation by more than 2 fold in MCF7 and more than 3 fold in T47D cell lines (Fig. 4A-B). Simar patterns were observed in colony formation assay by modulating miR-489 expression in MCF7 and T47D cells (Fig. 4C-D). Forced expression of miR-489 almost completely inhibited estrogen induced colony formation of both cells. Interestingly, inhibition of endogenous miR-489 drastically enhanced estrogen-mediated colony formation (Fig. 4C).
Estrogen treatment has been previously shown to enhance the population of stem-like cells in ERα cell lines(22). These so-called cancer stem cells are thought to be responsible for tumor relapse(23). We hypothesized that miR-489 may inhibit estrogen-induced population of cancerous stem-like cells by its effect on estrogen signaling. Therefore, we studied the effect of miR-489 on the estrogen-induced cancerous stem cell population using colony formation assay, mammosphere assay and flow cytometry. Consistent with previous results, estrogen increased the cancerous stem-like cell population (CD44+CD24-) by 3-fold in MCF7 cells and 10-fold in T47D cells. Inhibition of endogenous miR-489 further increased the cancerous stem-like cell population by more than 3-fold in MCF7 cells and by more than 11-fold in T47D cells (Fig. 4E). Similarly, we observed increased MFE upon estrogen treatment and miR-489 inhibition not only increased MFE but also increased the mammosphere size. Forced expression of miR-489 almost completely prevented mammosphere formation (Fig. 4F). Together, these results suggest that estrogen regulated miR-489 is a feedback regulator that is able to confine estrogen-induced tumor cell growth and inhibit the population of cancerous stem-like cells.
miR-489 inhibits estrogen induced signaling by targeting p38 and PTPN11
We further seek to elucidate the molecular mechanism responsible for miR-489 mediated inhibition of estrogen-ERα axis. Multiple mechanisms have been identified for regulation of estrogen-ERα mediated gene expression(24,25). Direct inhibition of ERα or its co-factors, inhibition of kinases that activate ERα, and inhibition of estrogen-induced nuclear localization of ERα have been previously reported to regulate estrogen induced gene transcription (26). We first examined if miR-489 exerted its effects by affecting localization of estrogen receptors. Interestingly, forced expression of miR-489 strongly promoted translocalization of estrogen receptors from the nucleus to the cytoplasm in MCF7 and T47D cell lines (Fig. 5A-B). In contrast, inhibition of endogenous miR-489 or knockout resulted in increased nuclear localization (additional file 2, Fig. S4A). Searching through factors which have been reported to regulate localization of estrogen receptors, we found that one of these factors, p38 MAPK, is a potential miR-489 target (26) (Fig. 5C). Interestingly, forced expression of miR-489 significantly downregulated total p38 MAPK (Fig. 5D). We then performed 3’UTR assay to examine if p38 MAPK is a direct target of miR-489. Forced expression of miR-489 significantly reduced the luciferase activity of wild type constructs but did not affect luciferase activity of constructs with a mutant miR-489 binding site (Fig. 5E). This result confirms that p38 MAPK is a direct target of miR-489. Next, we tested if the p38 MAPK inhibitor, SB203580, can phenocopy the effect of miR-489 on estrogen receptor localization. We transfected a control siRNA or miR-489 mimic for 72 hours or treated with DMSO or 10 μM SB203580 for 24 hours in hormone starved cells followed by treatment with estrogen to examine estrogen-induced nuclear localization of estrogen receptors. Indeed, T47D cells treated with p38 MAPK inhibitor phenocopied the effect of miR-489 on ER localization (Additional file 2, Fig. S4B). However, co-treatment of MCF7 cells with a p38 inhibitor following miR-489 transfection could not further decrease ERα phosphorylation or enhance cytoplasmic translocation of ERα (Fig. 5F), indicating that miR-489 affects nuclear translocation of ERα at least partially by downregulating p38 MAPK.
Considering that phosphorylation of the ERα protein is also a critical step for its transcription activity, we investigated changes in the phosphorylation status upon modulation of miR-489 expression. Western blot analysis of total ERα protein and its phosphorylated forms suggest that forced expression of miR-489 reduced the phosphorylation of ERα at both S118 and S167 sites (Fig. 5G). This data suggests that miR-489 may further regulate ligand-dependent activation of ERα through inhibition of kinases that phosphorylate ERα at residues S118 and S167. MAPK and AKT have been known to regulate ERα phosphorylation at S118 and S167 respectively. Indeed, miR-489 restoration significantly reduced the activated form of these kinases while its inhibition or knockout enhanced their activation (Fig. 5G).
All these three kinase pathways have been shown to be regulated by SHP2 and promote increase in tamoxifen resistance (4,27). It is also known that the PTPN11 gene encoding SHP2 is the direct target of miR-489 (16,28). To test whether miR-489 inhibits phosphorylation of all these kinases through regulating SHP2, we examined the effect of miR-489 modulation on SHP2 expression and the downstream signaling pathways in MCF7-HER2 and its isogenic cell line, MCF7-Vec, in the presence or absence of the SHP2 inhibitor, RMC4550 (Fig. 5H). Western blot analysis confirmed the upregulation of HER2, SHP2, pAKT and pERK in MCF7-HER2 cell line compared to MCF7-Vec. Transfection of miR-489 resulted in downregulation of SHP2 and p38 expression and decreased phosphorylation of AKT and MAPK. SHP2 inhibitor treatment alone inhibited phosphorylation of MAPK and AKT but had no effect on p38 expression or phosphorylation. SHP2 inhibitor treatment of miR-489-transfected cells showed some degree of synergy to inhibit phosphorylation of MAPK and AKT. Overall, our data suggested that miR-489 inhibits phosphorylation of MAPK and AKT at least partially though SHP2 but directly targets p38.
Pharmacological inhibition of p38 MAPK, PI3K-Akt and MAPK phenocopies the effect of miR-489 in ER+ breast cancer cell lines
Next, we examined whether pharmacological inhibition of all three responsible kinase signaling pathways phenocopies the effect of miR-489. We inspected estrogen-induced transcription and proliferation after inhibition of all three kinases. To a variable extent, we observed significant inhibition of transcriptional activity of ERα using an ERE-reporter cell line (Fig. 6A). qRT-PCR analysis also showed downregulation of estrogen responsive genes (Fig. 6B). Consistently, these data were also supported by diminished estrogen induced proliferation upon inhibition of all three kinases (Fig. 6C-6D). However, downregulation of endogenous miR-489 or miR-489 knockout was partially able to rescue cells from the growth inhibitory effect of kinase inhibitors (Fig. 6E-F). This data suggests that miR-489 may exert its inhibitory effect on estrogen signaling partially, if not completely, by simultaneously inhibiting p38 MAPK, AKT, and ERK signaling pathways.
Intriguingly, we observed that p38 MAPK inhibition had a significant inhibitory effect only in the presence of estrogen and did not affect estrogen-independent growth (Fig. 6G), indicating p38 MAPK activation may be estrogen dependent. Therefore, we tested if estrogen activates p38 MAPK and then mediates nuclear translocation. Since estrogen induced ERα nuclear translocation occurs within 5-30 minutes(29), we performed a time course of estrogen treatment on MCF7 and T47D cell lines. We observed a sharp increase in phospho-p38 MAPK and its downstream target, phosphor-ATF2, upon estrogen treatment in both cell lines (Fig. 6H). These results are consistent with previous studies that showed estrogen mediated activation of p38 MAPK in various tissues(26) (30).This provided evidence of a positive feedback loop between the E2-ERα axis and p38 MAPK in breast cancer cells such that binding of E2 leads to activation of p38 MAPK and activation of p38 MAPK leads to nuclear translocation of ERα which is necessary for its function as a transcription factor (Fig. 6I). As p38 MAPK activation was estrogen dependent, we therefore suspected p38 MAPK inhibition might have a pronounced effect in pre-menopausal women as compared to post-menopausal women (additional file 2, Fig. S5A). Indeed, we observed a higher p38 MAPK gene signature score in pre-menopausal luminal patients compared to post-menopausal luminal patients (Fig. 6J). Furthermore, luminal pre-menopausal patients with higher tumor grade also showed higher p38 MAPK gene signature (additional file 2, Fig. S5B). Additionally, our correlation analysis of miR-489 and p38 MAPK signature showed statistically significant inverse correlation in pre-menopausal patients. However, this correlation was reduced in post-menopausal patients (Fig. 6J). More importantly, high p38 MAPK expression in pre-menopausal patients predicted poor survival more significantly compared to post-menopausal patients (Fig. 6K). In summary, these results suggest that miR-489 regulates tamoxifen resistance by targeting multiple kinase signaling pathways and therefore could potentially be used as a therapeutic sensitizer to treat resistance patients.