Suppression of cell migration is promoted by miR-944 through targeting of SIAH1 and PTP4A1 in breast cancer cells

In order to confirm the clinical relevance of miR-944 in breast cancer, we quantified its expression by qRT-PCR in a set of clinical specimens obtained from a cohort of 40 patients (discovery cohort) from the FUCAM institution. Clinical features of breast tumors including hormonal receptor status, tumor size, histology, clinical stage, and tumor grade are summarized in Table 1. Results indicated that miR-944 expression was significantly (p < 0.05) diminished in tumors in comparison with adjacent normal tissues (Fig. 1a). Our results were validated by the analysis of 776 matched normal/tumor samples at The Cancer Genome Atlas (TCGA) (validation cohort), since the average expression of miR-944 was 8.16 in normal tissues versus 3.04 in tumors (Fig. 1b). To strengthen these data, we further analyzed the TGCA data for miR-204 and miR-10b, two miRNAs that have been previously reported as down-regulated and up-regulated, respectively, in breast cancer. As expected, miR-204 was suppressed, whereas miR-10b was overexpressed in the validation cohort (Additional file 2). On the other hand, miR-944 expression was significantly lower (8 to 9-fold) in MCF-7, MDA-MBD-231, MDA-MB-45, ZR-45, and T47-D breast cancer cell lines in comparison with non-tumorigenic MCF-10A breast cell line (Fig. 1c). Taken all together, our results confirmed that miR-944 was significantly suppressed in breast tumors.

To define the functions of miR-944 we restored its expression using RNA mimics in triple negative MDA-MB-231 (highly metastatic) and oestrogen responsive MCF-7 (poorly invasive) breast cancer cells (Additional file 3). First, the effect of diverse concentrations of miR-944 precursor on cell viability was evaluated by MTT assays. Results showed minimal changes (less than 5 %) in cell viability of MDA-MB-231 transfected with 50 nM miR-944 precursor in comparison with scramble transfected and non-transfected controls. Using 100 nM and 200 nM miR-944 precursor, we observed a 10 % reduction on cell viability relative to controls (Fig. 2a). Similar results were obtained in MCF-7 cells (Fig. 2e). Then, we performed scratch/wound-healing assays in both breast cancer cell lines to evaluate the contribution of miR-944 in tumor cell migration. Data indicated that cell monolayers restoration was delayed in both MDA-MB-231 and MCF-7 cells transfected with miR-944 precursor (50 nM) when compared with non-treated and scramble-transfected cells at 24 h (Fig. 2b and f). In addition, transwell chamber assays showed that the number of migratory cells was significantly (p < 0.05) reduced in MDA-MB-231 (4-fold) and MCF-7 (8-fold) cells that ectopically express miR-944 (Fig. 2c and g) in comparison with control cells. Moreover, miR-944 significantly (p < 0.05) inhibited the ability of metastatic MDA-MB-231 cells to invade matrigel in vitro (Fig. 2d).

As cell migration may involve the coordinated expression and association of proteins driving the epithelial-mesenchymal transition (EMT), cytoskeleton organization and reinforcement of focal adhesions, we decided to determine if miR-944 contributes to these cellular processes. We first analyzed the expression of proteins modulating the EMT, including SIP1, ZEB1 and BMP2, by Western blot assays. Results showed no significant changes in the expression of these proteins in miR-944 expressing cells (data not shown). Then, we examined the organization of cytoskeleton in MDA-MB-231 and MCF-7 cells by analyzing the distribution of F-actin labeled with rhodamine-phalloidin using confocal microscopy. As α-actinin-1 is an actin-crosslinking protein that reinforces focal adhesions its subcellular distribution was also examined. As depicted in Fig. 3a (upper panels), MDA-MB-231 control cells were featured by an axial F-actin cytoskeleton organization, and the presence of structures, such as membrane ruffles (MR) and filopodia (F) associated to a migrating phenotype were evident. Interestingly, the ectopic expression of miR-944 induced a dramatic effect on overall cell morphology since spread area was increased (Fig. 3a bottom panel). Moreover, F-actin was redistributed in a radial mode towards the periphery of the cell, as well as in the central zone; and the membrane ruffles and filopodia structures were lost. Based on these morphological differences, we next analyzed the strengthening of adhesion-related structures. MDA-MB-231 cells transfected with miR-944 precursor exhibited a robust signal of α-actinin-1 and an increase in the number of contact points with F-actin in multiple points of cell body, indicative of the reinforcement of focal adhesions. Remarkably, these cells displayed enrichment in α-actinin-1-rich blebs at the rear end of the cell suggesting a strong adhesive process (Fig. 3b bottom panel). Likewise, restoration of miR-944 expression in MCF-7 cells induced changes in actin cytoskeleton organization and loss of the axial pattern in a similar manner as in MDA-MB-231 cells (Fig. 3c bottom panel). In addition, α-actinin-1 was redistributed and accumulated in focal points at the end or front of cells indicative of focal adhesions formation (Fig. 3d, bottom panel), although in a less extend in comparison with MDA-MB-231 cells transfected with miR-944 precursor.

In order to identify potential target genes of miR-944 that may explain the phenotypic changes described above, we carried out a transcriptional profiling of MDA-MB-231 cells that ectopically express miR-944 using DNA microarrays. Results evidenced that 1197 genes were significantly downregulated and 144 were upregulated (fold change >1.5; Additional file 4). Some of these modulated genes are well known cancer-related genes including MAPK1, IGF1R, SIAH1, PRKCA, RAC1, NOTCH2, MMP14, PAK1 and PTP4A1, among others. Classification of the set of repressed genes based on GO categories showed that 15 genes are involved in cell migration and invasion processes (Table 2).

Data from DNA microarrays led us to the identification of potential new target genes for miR-944. Surprisingly, no genes involved in EMT and focal adhesions were found as directly modulated, thus we focused on genes involved in cell migration and cytoskeleton dynamics. Interestingly, the cell migration-related SIAH1, PTP4A1 (also known as PRL-1), and PRKCA genes were repressed after transfection of miR-944 precursor. These genes are key regulators of cell migration and cancer progression in diverse types of cancer [9‐11]. Therefore, we investigated if SIAH1, PTP4A1 and PRKCA genes are direct targets of miR-944 using luciferase reporter assays. We identified the complementary site for miR-944 in the 3´UTR sequence of each gene and cloned it downstream of the luciferase coding region in the pmiR-report vector (Fig. 4a). Results showed that forced expression of miR-944 and co-transfection of pmiR-LUC-PRKCA-3´UTR did not result in significant differences in luciferase activity (Fig. 4b). In contrast, the co-transfection of miR-944 and pmiR-LUC-SIAH1-3´UTR or pmiR-LUC-PTP4A1-3´UTR plasmids significantly reduced the luciferase activity (p < 0.001 and p < 0.05, respectively) in comparison with controls (Fig. 4c and d). Because of its relevant role in migration of cancer cells we next focused in the analysis of the SIAH1 protein. Western blot assays revealed that SIAH1 protein levels were reduced in MDA-MB-231 cells transfected with miR-944 in comparison to non-transfected control cells (Fig. 4e). Congruently, the expression of SIAH1 was significantly increased in 53 % of miR-944 deficient breast tumors in comparison with normal adjacent tissues (Fig. 4f and 4g).

To determine if targeted inhibition of SIAH1 affects cell migration we proceeded to knock-down its expression using RNA interference. Two specific short hairpin RNAs (dubbed as shSIAH1.1 and shSIAH1.2) targeting the human SIAH1 gene were designed and cloned into the pSilencer vector (Additional file 1). Both constructs were individually introduced into MDA-MB-231 cells and SIAH1 expression was analyzed by Western blot 48 h after transfection. Results showed that shSIAH1.2 sequence down-regulated the SIAH1 expression (Fig. 5a), whereas no significant effect was observed with shSIAH1.1 interfering sequence (data not shown). The expression of GADPH used as a control, did not show significant changes between treatments. Densitometric analysis of immunodetected bands showed that silencing induced by shSIAH1.2 construct was effective since this sequence suppressed SIAH1 expression by 42 % (Fig. 5b). The effect of SIAH1 silencing in cell migration was evaluated in MDA-MB-231 cells by scratch/wound-healing assays. Results showed that restoration of cell monolayers was significantly (p > 0.05) delayed in SIAH1-deficient cells when compared with scramble-transfected cells and non-treated control cells at 24 h (Fig. 5c).