ABSTRACT
Diverse RNA transcripts acting as competing endogenous RNAs (ceRNAs) can co-regulate each other's expression by competing for shared microRNAs. CCR2 protein, the receptor for CCL2, is implicated in cancer progression. However, we found that a higher CCR2 mRNA level is remarkably associated with prolonged survival of breast cancer patients. These conflicting results prompted us to study the non-coding function of CCR2 mRNA. We found that the CCR2 3′ untranslated region (UTR) inhibited MDA-MB-231 and MCF-7 cell metastasis by repressing epithelial–mesenchymal transition (EMT) in vitro, and suppressed breast cancer metastasis in vivo. Mechanistically, the CCR2 3′UTR modulated the expression of the RhoGAP protein STARD13 via acting as a STARD13 ceRNA in a microRNA-dependent and protein coding-independent manner. The CCR2 3′UTR blocked the activation of RhoA–ROCK1 pathway, which is the downstream effector of STARD13, and thus decreased the phosphorylation level of myosin light chain 2 (MLC2) and formation of F-actin. Additionally, the function of the CCR2 3′UTR was dependent on STARD13 expression. In conclusion, our results confirmed that the CCR2 3′UTR acts as a metastasis suppressor by acting as a ceRNA for STARD13 and thus inhibiting RhoA–ROCK1–MLC–F-actin pathway in breast cancer cells.
This article has an associated First Person interview with the first author of the paper.
INTRODUCTION
Worldwide, breast cancer is the most common malignancy among women (Sui et al., 2015). Metastasis is a complex process leading to dissemination of cancer cells from the primary tumor to distant organs (Gilkes et al., 2014), and it is the cause of the vast majority of deaths of patients with breast cancer (Papageorgis et al., 2015). Thus, there is an urgent need to elucidate the molecular mechanisms contributing to breast cancer metastasis.
CCL2 is a member of the CC chemokine family and acts on its target cells by binding to the cognate cysteine-cysteine chemokine receptor 2 (CCR2), a member of the seven-transmembrane G protein-coupled receptor family, which is the main signaling partner of CCL2 (Lee et al., 2009). Recently, the CCL2–CCR2 axis has attracted increasing interest due to its association with tumor progression (Lim et al., 2016). On one hand, CCL2 can be synthesized by metastatic tumor cells and stromal cells in the tumor microenvironment, which is critical for recruiting a subpopulation of CCR2-expressing monocytes or macrophages that enhance the subsequent extravasation of tumor cells (Qian et al., 2011). On the other hand, CCR2 is expressed in various cancer types and regulates CCL2-induced breast cancer cell survival and motility through MAPK- and Smad3-dependent mechanisms (Fang et al., 2012). In addition, CCL2–CCR2 can induce STAT3 activation and epithelial–mesenchymal transition (EMT) of PCa cells (Izumi et al., 2013), and the CCL2–CCR2 axis might promote metastasis of nasopharyngeal carcinoma by activating the ERK1/2–MMP2/9 pathway (Yang et al., 2016). These results support the hypothesis that the CCR2 protein, as a receptor together with its cognate ligand CCL2, promotes cancer metastasis. However, a Kaplan–Meier analysis of disease-specific survival for breast cancer patients stratified by CCR2 mRNA expression showed that higher expression of CCR2 mRNA is significantly correlated with longer survival (see Results). Therefore, we postulated that the CCR2 transcript may perform a different function to its protein in breast cancer progression.
The competing endogenous RNAs (ceRNAs) hypothesis posits that all RNA transcripts can modulate each other's expression by competing for shared microRNAs (miRNA), thus acting as ceRNAs (Cesana et al., 2011; Karreth et al., 2011; Salmena et al., 2011; Sumazin et al., 2011; Tay et al., 2011, 2014). Multiple non-coding RNA species, including small non-coding RNAs, pseudogenes (Zheng et al., 2015), long non-coding RNAs (lncRNAs) (Cesana et al., 2011; Chou et al., 2016; Liu et al., 2013) and circular RNAs (circ RNAs) can possess ceRNA activity. mRNA transcripts may act as tumor suppressors or oncogenes through their ceRNA activity, and this activity may be complementary or even distinct from their protein-coding function (Tay et al., 2014). For example, the ZEB2 protein, promotes the progression of glioma (Li et al., 2016a) and lung cancer (Jiao et al., 2016); however, as a PTEN ceRNA, the ZEB2 mRNA displays tumor-suppressive properties in melanoma cells (Karreth et al., 2011). To determine the potentially distinct function of the CCR2 transcript, we examined whether CCR2 mRNA exerts a tumor-suppressive function in breast cancer through its ceRNA activity.
STARD13 (also called DLC2), a unique Rho GTPase-activating protein (RhoGAP), is underexpressed in some types of cancer and suppresses cytoskeleton reorganization, cell migration and transformation by inhibiting RhoA activity through its RhoGAP domain (Ching et al., 2003; Leung et al., 2005; Lin et al., 2010). Active (GTP-loaded) RhoA binds to and activates Rho-associated coiled-coil-forming kinase 1 (ROCK1). RhoA-bound ROCK1 activates myosin light chain 2 (MLC2) by phosphorylating it at S19 (denoted pMLCS19), increasing the level of pMLCS19, which is required to coordinate the formation of stress fibers (Gilkes et al., 2014). Increased formation of the filamentous actin (F-actin) cytoskeleton and enhanced contractility are critical to cell migration (Stricker et al., 2010). miRNAs that target CCR2 mRNA have not been reported before. Here, TargetScan (www.targetscan.org/) was used to predict the miRNA-binding sites in the CCR2 3′-untranslated region (3′UTR), and three microRNA-125b (miR-125b)-binding sites were found to exist in the CCR2 3′UTR. Our laboratory has previously shown that miR-125b can promote metastasis of MCF-7 and MDA-MB-231 cells by targeting STARD13 (Tang et al., 2012). Furthermore, our recent study described a STARD13-correlated ceRNA network that possesses a metastasis-suppressive role in breast cancer (Li et al., 2016b). Thus, we deduced that CCR2 mRNA acts as a tumor suppressor in breast cancer through its ceRNA activity to promote STARD13 expression in a protein-independent but miRNA-dependent way.
This study reveals that the CCR2 3′UTR inhibits EMT and modulates STARD13 expression in MDA-MB-231 and MCF-7 cells. The CCR2 3′UTR blocked RhoA–ROCK1 signaling, which acts as the downstream effector of STARD13, and thus decreased the phosphorylation of MLC and the formation of F-actin. Taken together, these results delineate a molecular mechanism by which the CCR2 3′UTR regulates STARD13 expression and further inactivates the RhoA–ROCK1 pathway, which is required for breast cancer metastasis.
RESULTS
The CCR2 3′UTR displays tumor-suppressive properties in breast cancer cells
We first investigated the roles of the CCR2 3′UTR in breast cancer development. First, we performed a Kaplan–Meier survival analysis, by using the KM-plotter (Györffy et al., 2010), which revealed that CCR2 mRNA level was positively correlated with the longer overall survival (OS) (P=0.0011) (Fig. 1A), distant metastasis-free survival (DMFS) (P=0.018) (Fig. 1B) and relapse-free survival (RFS) (P=2.9×10−15) (Fig. 1C) of breast cancer patients. The results seem to contradict the idea that CCR2 enhances CCL2-induced breast cancer cell survival and motility, which promoted us to explore whether CCR2 mRNA holds distinct functions other than its protein-coding function. Cell migration was measured in MCF-7 and MDA-MB-231 cells transfected with the CCR2 3′UTR or empty vector through wound healing and transwell migration assays. Overexpression of the CCR2 3′UTR remarkably decreased the migration of MCF-7 and MDA-MB-231 cells compared with control vector (Fig. 1D,E). Adhesion of tumor cells to extracellular and basement membranes plays a critical role in the initial step in the invasive process (Tang et al., 2012). The adhesion ability of the CCR2 3′UTR-transfected cells was decreased in MCF-7 and MDA-MB-231 cells (Fig. 1F). A similar trend was observed in Transwell migration assays (Fig. 1G,H). Additionally, Matrigel invasion chambers were used to examine the role of the CCR2 3′UTR towards tumor invasiveness. The invasive ability of the CCR2 3′UTR-transfected cells was markedly decreased compared to cells treated with control vector (Fig. 1G,I). In addition, cell cycle analysis showed that no relevant changes were detected upon CCR2 3′UTR ectopic expression in the percentage of cells in G0/G1, G2/M and S phases (Fig. S1A–D). Further cell apoptosis analysis indicated that the CCR2 3′UTR overexpression had no significant effect on cell apoptosis (Fig. S1E–H). These results suggest that ectopic expression of the CCR2 3′UTR suppresses breast cancer cell metastasis and has no effect on cell cycle progression and apoptosis.
Ectopic expression of the CCR2 3′UTR inhibits the EMT process
EMT is the most critical step in tumor metastasis (Biddle and Mackenzie, 2012; Kong et al., 2008). We further sought to determine whether the CCR2 3′UTR could reverse the EMT phenotype. As shown in Fig. 2A,B, the ectopic expression of the CCR2 3′UTR or STARD13 3′UTR resulted in upregulation of the epithelial marker E-cadherin and downregulation of mesenchymal markers, including N-cadherin, and α-smooth muscle actin (α-SMA, also known as ACTA2) in MCF-7 cells. Furthermore, TGF-β1-induced EMT was used as test model for the effect of CCR2 3′UTR overexpression in epithelial MCF-7 cells. At 48 h after TGF-β1 treatment, MCF-7 cells exhibited lower levels of E-cadherin and higher N-cadherin, which was reversed upon CCR2 3′UTR overexpression (Fig. 2C). After TGF-β1 treatment, MCF-7 cells underwent a dramatic morphological change, from a cobblestone-like epithelial structure to fibroblastoid spindle-shaped cells. This morphological transition is accompanied by E-cadherin downregulation. However, transfection with the CCR2 3′UTR significantly reduced the acquisition of mesenchymal characteristics during the TGF-β1-induced EMT of MCF-7 cells (Fig. 2D). The ectopic expression of the CCR2 3′UTR or STARD13 3′UTR downregulated mesenchymal markers expression including N-cadherin and vimentin in MDA-MB-231 cells, whereas α-SMA expression was not affected in MDA-MB-231 cells (Fig. 3A,B). Identical results were obtained by immunofluorescence staining analysis (Fig. 3C,D). Taken together, these results demonstrate that the CCR2 3′UTR could block the TGF-β1-induced EMT of MCF-7 cells and induce mesenchymal–epithelial transition (MET) in MCF-7 and MDA-MB-231 cells.
The CCR2 3′UTR modulates expression of STARD13
Based on the fact that a series of common miRNA response elements (MREs) exist in the 3′UTR of CCR2 and STARD13, we assumed that CCR2 is a ceRNA candidate for STARD13, which has proven to be a tumor suppressor (Tang et al., 2012). We first examined whether the CCR2 3′UTR could modulate STARD13 expression. Ectopic expression of the CCR2 3′UTR significantly upregulated the mRNA levels of CCR2 and STARD13 in MDA-MB-231 cells (Fig. 4A) and MCF-7 cells (Fig. S2A). Similarly, the STARD13 3′UTR also upregulated the mRNA levels of CCR2 and STARD13 (Fig. 4B; Fig. S2B). Moreover, STARD13 and CCR2 protein levels were increased in the CCR2 3′UTR-transfected (Fig. 4C; Fig. S2C) and STARD13 3′UTR-transfected (Fig. 3D; Fig. S2D) breast cancer cells. In contrast, knockdown of CCR2 reduced STARD13 mRNA levels and knockdown of STARD13 led to downregulation of CCR2 mRNA levels in MDA-MB-231 cells (Fig. 4E) and MCF-7 cells (Fig. S2E). Importantly, STARD13 and CCR2 protein levels were reduced following knockdown of CCR2 or knockdown of STARD13 in MDA-MB-231 cells (Fig. 4F) and MCF-7 cells (Fig. S2F). These results indicate that CCR2 and STARD13 can regulate each other's expression.
CCR2 is a potential target of miR-125b
To elucidate which miRNA mediates the CCR2–STARD13 crosstalk, we analyzed the common MREs in the 3′UTRs of STARD13 and CCR2. TargetScan 6.2 predicted that CCR2 and STARD13 shared sites for four miRNA families and contained a total of six and nine MREs for these miRNAs, respectively, with miR-125b being one of the shared miRNAs (Fig. 5A). Our previous study has indicated that miR-125b induces breast cancer metastasis by targeting STARD13 directly (Li et al., 2016b), whereas the miR-125b–CCR2 axis needed to be validated. Quantitative real-time RT-PCR (qRT-PCR) analysis confirmed the transfection efficiency of miR-125b mimics and miR-125b inhibitor in MDA-MB-231 (Fig. 5B,C) and MCF-7 cells (Fig. S3A,B). TargetScan 6.2 analysis indicated that STARD13 contains four miR-125b-binding sites on its 3′UTR, and two of these four sites were predicted to be eight nucleotides long (Fig. 5D). To validate whether CCR2 is regulated by miR-125b, TargetScan 6.2 was adopted to predict targets of miR-125b, and the 3′UTR of CCR2 was predicted to contain three complementary sites for the seed region of miR-125b. One of these three sites, starting at nt 169 of the the CCR2 3′UTR, was predicted to be eight nucleotides long. CCR2 3′UTR fragments containing the wild-type version of this miR-125b target site (CCR2-WT) or with a version with a mutant miR-125b-binding site (CCR2-MUT) were introduced downstream of the luciferase reporter gene in a pMIR-reporter vector (Fig. 5D). CCR2-WT or CCR2-MUT was then co-transfected with miR-125b mimics in MDA-MB-231 and MCF-7 cells (Fig. 5E; Fig. S3C), and we found that the relative luciferase activity of CCR2-WT was attenuated, while the activity of CCR2-MUT was unaffected. To validate the direct binding between miR-125b and the CCR2 3′UTR and STARD13 3′UTR, we performed an RNA immunoprecipitation (RIP) assay using antibody against Ago2, which is the core component of the RNA-induced silencing complex, to pulldown miRNAs associated with the CCR2 3′UTR or STARD13 3′UTR. qRT-PCR analysis of results from the RIP assay demonstrated that the amount of miR-125b pulled down (Ago2 enrichment) from MDA-MB-231 cells stably transfected with the CCR2 3′UTR [MDA-MB-231(C)] or with the STARD13 3′UTR [MDA-MB-231(S)] was significantly higher than that from MDA-MB-231 cells stably transfected with an empty vector [MDA-MB-231(V)] (Fig. 5F). CCR2 and STARD13 mRNA levels were significantly decreased or increased with upon transfection with miR-125b mimics or inhibitor, respectively, in MDA-MB-231 (Fig. 5G,H) and MCF-7 cells (Fig. S3D,E). Furthermore, overexpression of miR-125b resulted in a significant reduction in the protein levels of CCR2 and STARD13 (Fig. 5I; Fig. S3F) and knockdown of miR-125b led to an increase in protein levels (Fig. 5J; Fig. S3G). Taken together, these data indicate that CCR2 is a potential target of miR-125b.
miRNA dependency of the CCR2–STARD13 interaction
To further confirm that CCR2 acts as a STARD13 ceRNA, we must confirm that the CCR2–STARD13 interaction is indeed mediated by miRNAs. Thus, siRNA against DICER (also known as DICER1), a miRNA biogenesis protein, was utilized. The efficient knockdown of DICER (siDICER) and the concomitant downregulation of miR-125b have been revealed by our earlier study (Li et al., 2016b). siRNA against CCR2 (siCCR2) failed to inhibit STARD13 mRNA and protein expression when co-transfected with siDICER, whereas the CCR2 mRNA level was decreased. Similarly, STARD13 knockdown diminished the STARD13 mRNA level but did not affect the CCR2 mRNA and protein level in siDICER-treated cells as compared with negative control (NC) cells (Fig. 6A,B; Fig. S4A,B). Moreover, in the presence of siDICER, the CCR2 3′UTR-mediated or STARD13 3′UTR-mediated increase of STARD13 and CCR2 protein expression was lost (Fig. 6C,D; Fig. S4C,D). To confirm the pivotal roles of miRNAs in the regulation between CCR2 and STARD13, we performed an RIP assay on Ago2. As shown in Fig. 6E, overexpression of the CCR2 3′UTR in MDA-MB-231 cells led to the increased enrichment of Ago2 on the CCR2 3′UTR but substantially decreased enrichment on the STARD13 3′UTR, and this result suggests that the CCR2 3′UTR can compete with STARD13 transcripts for the Ago2-based miRNA-induced repression complex. Notably, ectopic expression of the CCR2 3′UTR with mutation of all three predicted miR-125b-binding sites did significantly upregulate the protein level of CCR2 but caused no significantly change in STARD13 protein level (Fig. 6F). Importantly, a wound healing assay showed that the CCR2 3′UTR with mutation of all three predicted miR-125b-binding sites caused no significant decrease in the migration of MDA-MB-231 cells (Fig. 6G). These data indicate that the CCR2 3′UTR-mediated regulation of STARD13 is indeed dependent on miRNA and the miR-125b target sites within the CCR2 3′UTR, especially, are necessary for the CCR2 3′UTR to inhibit the metastasis of breast cancer cells.
The CCR2 3′UTR and STARD13 3′UTR suppress RhoA activity, MLC phosphorylation and formation of stress fibers
To study whether the CCR2 3′UTR could regulate the downstream signals of STARD13, immunofluorescence staining of filamentous actin (F-actin) with Rhodamine-labeled phalloidin was performed, and this revealed that stress fiber formation was markedly inhibited in the CCR2 3′UTR- and STARD13 3′UTR-transfected cells (Fig. 7A; Fig. S5A). Additionally, the levels of pMLCS19 were decreased significantly in the CCR2 3′UTR-transfected (Fig. 7B; Fig. S5B) and STARD13 3′UTR-transfected (Fig. 7C; Fig. S5C) cells. To further determine whether RhoA or ROCK1 was involved in the CCR2 3′UTR- or STARD13 3′UTR-mediated inhibition of MLC phosphorylation, qRT-PCR and western blot analyses were performed, and revealed that the CCR2 3′UTR and STARD13 3′UTR had no effect on expression of RhoA and ROCK1 in MDA-MB-231 cells (Fig. 7D–G) and in MCF-7 cells (Fig. S5D–G). Furthermore, a G-LISA RhoA activation assay revealed that CCR2 3′UTR-transfected cells had a lower basal level of RhoA activity (Fig. 7H). In summary, these data indicate that the CCR2 3′UTR and STARD13 3′UTR inhibit the formation of F-actin and MLC phosphorylation through regulating RhoA activity rather than by altering RhoA and ROCK1 expression in human breast cancer cells.
The decreased formation of stress fibers and MLC phosphorylation mediated by CCR2 3′UTR is dependent on STARD13
We observed that the CCR2 3′UTR could inhibit formation of F-actin and MLC phosphorylation. To further ascertain whether these observed effects are dependent upon STARD13 expression, we undertook immunofluorescence staining of F-actin, and showed that siRNA against STARD13 (siSTARD13) could reverse the CCR2 3′UTR-mediated inhibition on stress fiber formation (Fig. 7I; Fig. S5H), suggesting that the CCR2 3′UTR induces cell migration via STARD13. Furthermore, blocking STARD13 expression with siSTARD13 abolished the CCR2 3′UTR-induced decrease of pMLCS19 in MDA-MB-231 (Fig. 7J) and MCF-7 cells (Fig. S5I). Taken together, these results demonstrate that the decreased MLC phosphorylation and formation of stress fibers induced by the CCR2 3′UTR are dependent on STARD13.
The CCR2 3′UTR inhibits breast cancer metastasis in a xenograft model
The antitumor function of the CCR2 3′UTR in breast cancer was further confirmed in vivo. qRT-PCR analysis demonstrated the efficient overexpression of the CCR2 3′UTR in stably transfected cells (Fig. 8A). MDA-MB-231(C) cells expressed more CCR2 and STARD13 protein (Fig. S6A) and less pMLC protein (Fig. S6B) than MDA-MB-231(V) cells. The morphology in the MDA-MB-231(C) cells, which are stably transfected with the CCR2 3′UTR, was round in shape, rather than the angular and spindular shape seen in MDA-MB-231(V) cells, and the stress fiber formation was inhibited in the MDA-MB-231(C) cells (Fig. S6C). Additionally, MDA-MB-231 cells stably transfected with the CCR2 3′UTR had a significantly decreased level of cell migration (Fig. S6D). We intravenously transplanted MDA-MB-231(V) or MDA-MB-231(C) into nude mice. As seen in Fig. 8B, micro-computed tomography (micro-CT) analysis revealed that the CCR2 3′UTR overexpression strikingly decreased the lung metastasis of MDA-MB-231 cells in vivo. The Carestream noninvasive optical imaging system was used for whole-animal imaging, and animals injected with cells stably transfected with the CCR2 3′UTR showed a reduced metastasis signals in ventral views (Fig. 8C). Notably, the CCR2 3′UTR suppressed pulmonary metastasis of MDA-MB-231 cells, as shown by the bioluminescence of the excised tissues (Fig. 8D). Again, histopathological analysis confirmed the inhibitory effect of the CCR2 3′UTR on pulmonary metastasis (Fig. 8E). Our in vivo data further strengthen the finding that the CCR2 3′UTR has an anti-metastatic effect in breast cancer.
DISCUSSION
The present study shows that the CCR2 3′UTR induces a miRNA-dependent increase of STARD13 expression, leading to downstream pathway inactivation that is manifested by the decrease of RhoA activation, actin polymerization and MLC phosphorylation, and thus suppresses breast cancer metastasis (Fig. S7).
CCR2 is viewed as receptor of CCL2, and the CCL2–CCR2 axis has been shown to play crucial roles in cancer metastasis. Targeting this signaling pair is considered as a therapeutic intervention. However, therapeutics obstructing this chemokine receptor pair have displayed disappointing results in the clinic (Lim et al., 2016). Our analyses found that breast cancer patients with higher CCR2 mRNA expression have a longer overall survival, distant metastasis-free survival and a longer relapse-free survival. This suggests that high CCR2 mRNA is a marker for good prognosis in breast cancer. We analyzed the inconsistency between results found for CCR2 transcript and CCR2 protein for the first time, and demonstrated that CCR2 mRNA and protein exert different biological effects. The ceRNA hypothesis was used to explain the molecular mechanism of CCR2 mRNA action. The ceRNA hypothesis provides a new angle for studying the multilayered complexity of genes. Although it has been proposed that ceRNA regulation is rare and can only exist in certain regimes of target and miRNA abundance (Bosson et al., 2014; Denzler et al., 2014, 2016), the discovery of functional ceRNA regulation in diverse cancer progression by multiple independent groups suggests that ceRNA may represent a widespread layer of gene regulation (Fang et al., 2013; Kumar et al., 2014; Tay et al., 2014). In order to reveal the function of a gene, we should integrate the classic ‘protein dimension’ with an additional ‘ceRNA dimension’ (Karreth et al., 2011).
It has recently been reported that CCR2 knockdown in 4T1 and MCF-7 cells can block CCL2-induced wound closure (Fang et al., 2012). In contrast, we proved that overexpression of the CCR2 3′UTR suppressed breast cancer cells migration and invasion. One possible explanation for this discrepancy is that cells were not stimulated with CCL2 in our experiments. In addition, the CCR2 3′UTR had an influence on biomarkers of EMT, which indicated that the CCR2 3′UTR could affect breast cancer cell metastasis through suppressing the EMT process. Notably, a nude mice xenograft model verified the inhibitory role of the CCR2 3′UTR on breast cancer metastasis in vivo.
Our previous studies have implicated STARD13 as a metastatic suppressor in MCF-7 and MDA-MB-231 cells in vitro and in vivo, and miR-125b can negatively regulate the expression of STARD13 by targeting STARD13 directly (Li et al., 2016b; Tang et al., 2012). Although a series of experiments were performed to validate that CCR2 and STARD13 could regulate each other's expression, the ceRNA crosstalk is complex and multi-level. We have to point out that this study did not concentrate some additional considerations of ceRNA activity, such as the abundance of key players, potential interplay with RNA-binding proteins (RBPs) and RNA editing, which may have effects on ceRNA crosstalk (Tay et al., 2014). Analyzing the miRNA-binding sites located in the CCR2 3′UTR has suggested that numerous miRNAs could bind with the CCR2 3′UTR. The change in the expression level of CCR2 could affect the level of these potential CCR2-binding miRNAs, which in turn could affect the level of other targets. In addition to the direct interactions through shared miRNAs, secondary indirect interactions may also have a profound effect on ceRNA regulation. Originally, ceRNAs were defined as target RNAs of miRNAs that cross talk via competition over binding to their common miRNA regulators. However, by the same logic, miRNAs that share targets compete over binding to their common targets and therefore also exhibit ceRNA-like behavior. Thus, the term ceRNA should be expanded to encompass both types of competing elements. This suggests that perturbation effects could propagate in the network through a cascade of coregulated target RNAs and miRNAs that share targets, leading to mutual effects between distant components in the network, i.e. distant ceRNAs (Nitzan et al., 2014). There is increasing evidence confirming that ceRNAs crosstalk in large interconnected networks (Tay et al., 2014). Further studies are needed to accurately identify metastasis-related miRNA and mRNAs in the CCR2-correlated ceRNA network.
TargetScan was employed because this algorithm considers MRE conservation between mammals, TargetScan predicted that CCR2 and STARD13 are targets for four shared miRNA families; among them, the number of miR-125b MREs in the CCR2 3′UTR and STARD13 3′UTR was more than that found for the other miRNAs, therefore miR-125b was used as a model to explore the CCR2–STARD13 interaction. As expected, our results showed that miR-125b can inhibit CCR2 expression. siDICER was used to present an ideal system to evaluate miRNA-dependent effects, and the results showed that the CCR2–STARD13 is interaction is an miRNA-mediated and MRE-dependent regulation. It was specifically found that miR-125b target sites within the CCR2 3′UTR were necessary for the CCR2–STARD13 interaction. Overexpression of the CCR2 3′UTR caused a different fold-increase in the CCR2 or STARD13 mRNA compared to that of the protein; a possible explanation for this is that, in contrast to the transcription process, the translation process is regulated by other mechanisms such as phosphorylation, acetylation, ubiquitylation and proteolysis of core components of the translation machinery (de Sousa Abreu et al., 2009). In addition, and more remarkably, TargetScan list two transcript variants of CCR2, only one of which contains miR-125b target sites; the 3′UTR sequence we study belongs to CCR2 (ENSG00000121807.5), which TargetScan considers as the less-prevalent transcript; therefore, considerations such as the percentage of the transcript variants to the function of CCR2 and the mutual regulation of the two transcript variants still need further study. Importantly, further experiments should focus on whether there a particular 3′UTR variant that can repress CCR2 translation, while stabilizing the 3′UTR allowing it to act as a ceRNA, and whether there are cancer-specific regulatory pathways that promote CCR2 protein degradation, while stabilizing the mRNA.
As STARD13 is a major antagonist of RhoA–ROCK1 signaling, we examined whether CCR2 can inactivate this pathway via acting as a STARD13 ceRNA. Intriguingly, the CCR2 3′UTR inhibits RhoA activity rather than the expression of total RhoA and ROCK1. These findings rule out the possibility of a direct action between the CCR2 3′UTR and RhoA, and proved that the CCR2 3′UTR-mediated inhibition of the RhoA activity is dependent on a RhoGAP-containing protein instead acting through another mechanism. RhoA and ROCK1 expression can phosphorylate and inhibit myosin phosphatase (MYPT) and activate focal adhesion kinase (FAK, also known as PTK2) in hypoxic breast cancer cells (Gilkes et al., 2014). Further research is needed to confirm whether the phosphorylation of MYPT and FAK will be affected by the CCR2 3′UTR. The CCR2 3′UTR-induced decrease of pMLCS19 and inhibition of stress fiber formation were abolished by STARD13 siRNA, indicating that STARD13 acted as a downstream effector of the CCR2 3′UTR. Since our previous study has confirmed that a STARD13-correlated ceRNA network can inhibit breast cancer metastasis (Li et al., 2016b), further investigation should be carried out to confirm the regulation between the CCR2 and STARD13-correlated ceRNA network.
Although several clinical trials targeting the CCL2–CCR2 signaling axis have been established, the CCL2–CCR2 signaling network is undeniably far more complex than that suggested by the current researches (Lim et al., 2016). There is still has a long way to go before CCL2- or CCR2-targeted therapies can be set up successfully. The non-coding function of mRNA might create functional complexity and diversification in both physiological and pathological conditions, and the unappreciated genetic dimension should be considered when designing treatment strategy.
MATERIALS AND METHODS
Cell lines and culture
Human breast cancer cell lines MCF-7 and MDA-MB-231 were obtained from the ATCC (Manassas, VA) and maintained in high-glucose Dulbecco's modified Eagle's medium (DMEM) and L15 medium (Gibco), respectively. Both media were supplemented with 10% fetal bovine serum (FBS) (Gibco), 80 U/ml penicillin and 0.08 mg/ml streptomycin, and cells were incubated at 5% CO2 at 37°C. L15 medium with 0.6 µg/ml puromycin was used to incubate MDA-MB-231(C) cells (stably transfected with the CCR2 3′UTR), MDA-MB-231(V) cells (stably transfected with an empty vector) and MDA-MB-231(S) (stably transfected with the STARD13 3′UTR).
MicroRNA, siRNA, plasmids and transfection
miRNA mimics and inhibitor, negative control (NC) mimics and inhibitor, siRNA and NC siRNA were synthesized by Biomics Biotechnology, China. Sequences of siRNA were: CCR2 siRNA (sense strand, 5′-CAUCAAUCCCAUCAUCUAU-3′), STARD13 siRNA (sense strand, 5′-CACCUUUCCAUCUCCUAAU-3′), DICER siRNA (sense strand, 5′-AAGGCUUACCUUCUCCAGGCU-3′). The CCR2 3′UTR (NM_001123396.1) and STARD13 3′UTR were cloned into the pSilencer 4.1 vector (Ambion) and the pCDNA 3.1 vector (Invitrogen), respectively, and the constructs were verified by DNA sequencing. The recombinant plasmids were denoted the CCR2 3′UTR and STARD13 3′UTR, respectively. The CCR2 3′UTR with mutation of all three predicted miR-125b-binding sites were cloned into the pSilencer 4.1 vector and was denoted the CCR2 3′UTR MUT. Lipofectamine 2000 Reagent (Invitrogen) was used for the miRNA mimics and inhibitor, mimics NC, inhibitor NC, siRNA, NC and plasmid transfection. At 48 h after transfection, the mRNA expression was assessed by qRT-PCR, and the protein expression was tested by western blotting. At 24 h after transfection, cells were treated with TGF-β1 (PeproTech) at a concentration of 10 ng/ml for 48 h.
Adhesion assay
Wells of microwell plate were coated with matrix gel (BD Biosciences) at 37°C for 4 h, and were then blocked for 1 h with 1.2% BSA in PBS. Cells transfected with the CCR2 3′UTR or pSilencer 4.1 were suspended at concentration of 0.6×105 cells/well in serum-free medium. The colorimetric MTT-assay was used to determine the number of adherent cells after 1 h adhesion.
Wound healing assay
The wound healing assay was conducted as described previously (Chou et al., 2016). MCF-7 and MDA-MB-231 cells were seeded in six-well plates and transiently transfected with the CCR2 3′UTR or pSilencer 4.1. Cells were allowed to grow up to 90% confluency in complete medium. A vertical wound was made to cells with a sterile pipette tip, and the wounded monolayer was washed with PBS to remove cell debris, and serum-free medium was used to maintain cells. Phase-contrast images were taken of each sample at 0 h and 12 h (MDA-MB-231) or 24 h (MCF-7) at the same position with respect to the wound. The distances that cells migrated into wound surface were calculated for three different times. The migration rate was taken as (D0−Dt)/D0, where D0 stands for the distance measured at 0 h and Dt refers to the distance measured at 12 h (MDA-MB-231) and 24 h (MCF-7).
Transwell migration and invasion assay
The migration assay was performed using a Transwell insert chamber (a pore size of 8 µm; Millipore) and the invasion assay was performed using BD Biosciences Matrigel invasion chambers (a pore size of 8 µm; BD Biosciences). After transfection, 105 cells in serum-free medium were placed into the upper chamber, and the lower chamber was filled with complete medium containing 20% FBS. For the migration assay, cells were allowed to migrate at 37°C for 24 h (MDA-MB-231) or 36 h (MCF-7). For the invasion assay, cells were allowed to invade at 37°C for 36 h (MDA-MB-231) or 48 h (MCF-7). The migrated or invaded cells were fixed in methanol for 30 min and stained with 0.1% Crystal Violet for 30 min. The stained cells were photographed, and the dye was eluted with glacial acetic acid and quantified by measuring with Microplate Reader [optical density (OD) at 570 nm].
Flow cytometric analysis of the cell cycle and apoptosis
Cells were harvested at 48 h after transfection. The cell cycle and apoptosis were analyzed by using a FACS caliber machine (Becton Dickinson) with a cell cycle assay kit (Vazyme) and apoptosis detection kit (Vazyme), respectively. The cell cycle distribution was determined via Propidium Iodide (PI) staining. A total of 20,000 gated events were acquired to assess the proportions of cells in different stages of the cell cycle and cell cycle profiles were calculated by using the ModFit LT 4.0 software. For the apoptosis assay, the cells were stained with Annexin V–FITC and PI. 10,000 cells were collected and FlowJo software was used to analyze the data, and the data were expressed as a percentage.
Luciferase reporter assay
To test whether miR-125b bound directly to the 3′UTR of CCR2 mRNA, the pMIR-REPORT™ miRNA Expression Reporter Vector System (Applied Biosystems), consisting of an experimental firefly luciferase reporter vector and an associated β-gal reporter control plasmid were used. A portion of the 3′UTR of CCR2 mRNA containing a wild-type (WT) or mutant (MUT) miR-125b-binding site was cloned respectively into the firefly luciferase reporter vector. For the luciferase assay, MCF-7 or MDA-MB-231 cells in 24-well plates were co-transfected with recombinant firefly luciferase reporter plasmids (CCR2-WT or CCR2-MUT), miR-125b mimics or mimics NC and β-gal reporter control plasmid using Lipofectamine 2000 reagent. Luciferase activity was measured according to the manufacturer's protocol and normalized to β-gal activity.
mRNA and miRNA quantification
Total RNA of cells after transfection was isolated by using TRIzol reagent (Invitrogen). cDNA was synthesized by using the M-MLV (Promega) following standard protocols. qRT-PCR was performed by using the Applied Biosystems StepOnePlus™ system. EzOmics SYBR qPCR mix and miRNA-125b qPCR kit were purchased from Biomics. MiR-125b levels were normalized to levels of U6 snRNA. The primers for various genes were designed and synthesized as follows:
GAPDH, 5′-AAGGTCGGAGTCAACGGATT-3′ and 5′-CTGGAAGATGGTGATGGGATT-3′; CCR2, 5′-CTGTCCACATCTCGTTCTCGGTTTA-3′ and 5′-CCCAAAGACCCACTCATTTGCAGC-3′; STARD13, 5′-AGCCCCTGCCTCAAAGTATT-3′ and 5′-ATGGGCGTCATCTGATTCTC-3′; RhoA, 5′-GGACCCCAGAAGTCAAGCAT-3′ and 5′-GAGCAGCTCTCGTAGCCATT-3′; ROCK1, 5′-AACATGCTGCTGGATAAATCTGG-3′, and 5′-TGTATCACATCGTACCATGCCT-3′; the CCR2 3′UTR: 5′-ATGCCTCATTACCTTGTG-3′ and 5′-CCATTCATCTGTGCCTGT-3′; the STARD13 3′UTR, 5′-TTAAAGCTAGATGAGGAGTTGCC -3′, and 5′-TTGACTTGGGGTCTGTAGTGATG-3′.
The relative gene expression level of each group was calculated by using the 2−ΔΔct method.
Western blotting
Cells were harvested and lysed in buffer RIPA (Beyotime) on ice. Protein concentration was measured by using the BCA method. 30 µg protein of each sample was fractionated by 10% or 12% SDS-PAGE and transferred electrophoretically onto polyvinylidene difluoride membranes (Millipore). The membranes were blocked with 8% BSA in Tris-buffered saline with 0.1% Tween 20 at 37°C for 1 h and then blotted with primary antibodies overnight at 4°C. Antibodies against the following proteins were used: E-cadherin (ab40772), N-cadherin (ab76011) and vimentin (ab92547) (all 1:5000, Abcam); α-SMA (WL0002a; 1:1000, Wanleibio); CCR2 (12199), MLC (8505), pMLCS19 (3671) and RhoA (2117) (all 1:1000, Cell Signaling Technology); β-actin (TA-09 1:500, ZSGB-BIO); and STARD13 (AP19692c; 1:1000, ABGENT). After incubation with horseradish peroxidase (HRP)-conjugated secondary antibody (ZSGB-BIO), the protein bands were detected by using ECL chemiluminescence reagent (Tanon). Protein expression levels were quantified by density analysis using Quantity One Software and normalized to levels of β-actin.
RNA-binding protein immunoprecipitation assay
The RIP assay was performed using the Protein A/G Agarose Resin 4FF (YEASEN) following the manufacturer's protocol. Briefly, MDA-MB-231(C) cells (stably transfected with the CCR2 3′UTR), MDA-MB-231(V) cells (stably transfected with an empty vector) and MDA-MB-231(S) (stably transfected with STARD13 3′UTR) cells were lysed with NP-40 lysis buffer (Beyotime). Then, 100 µl cells extract was incubated with NP-40 buffer containing Protein A/G Agarose Resin 4FF conjugated to human anti-Ago antibody (Cell Signaling Technology) at 4°C overnight. Agarose resin was isolated by centrifugation and incubated with proteinase K (Beyotime) to dissociate the Ago2–RNA complex from the agarose resin. qRT-PCR was performed to detect CCR2, STARD13 and miR-125b levels.
RhoA GTPase assay
Cells were assayed for RhoA GTPase activity by performing a RhoA G-LISA (Cytoskeleton) according to the commercial protocol. Briefly, lysates were prepared and lysate protein concentration was measured. Lysis buffer was then added to equalize the cell extracts to give identical protein concentrations in each sample. Samples were incubated in the wells provided and then managed following the technical guide. Absorbance at 490 nm was measured using a microplate spectrophotometer.
Immunofluorescence and F-actin visualization
Cells were fixed in 4% paraformaldehyde for 20 min, permeabilized with 0.2% Triton X-100 for 20 min at 4°C, and blocked with 3% BSA in PBS for 1 h at room temperature. For immunofluorescence, appropriate primary antibodies [E-cadherin (ab40772), N-cadherin (ab76011) (both 1:100, Abcam) and vimentin (ab92547, 1:200, Abcam)] were used to blot overnight at 4°C and FITC-conjugated secondary antibody was used. For F-actin visualization, F-actin was stained with Rhodamine–phalloidin (Cytoskeleton) for 30 min at room temperature. The nuclei were stained with DAPI for 30 min and observed through confocal microscopy. Cells were captured by adopting confocal laser scanning. Images were analyzed using OLYMPUS FLUOVIEW ver.3.0 Viewer software and the fluorescence intensity was quantified by using Image J software.
In vivo metastasis study
All experiments were carried out under the ethical approval of Ethics Committee for Animal Experimentation of China Pharmaceutical University. Six-week-old female nude mice were purchased from the Model Animal Research Center of Nanjing University. Intravenous injection with MDA-MB-231 cells that were stably overexpressed the CCR2 3′UTR or empty vector was performed, and 5×106 cells in 150 µl PBS were injected into six animals per group. At 8 weeks after injection, animals were scanned using the SkyScan 1176 micro-CT scanner. For bioluminescent analysis, Nano-GloTM substrate containing furimazine (20 µg in 100 µl PBS, Promega) was injected intraperitoneally into mice before imaging. Images of in vivo tumor growth and ex vivo lung metastasis were taken using the Carestream noninvasive optical imaging system. Mice were killed, and then lungs were sectioned and stained with H&E.
Statistical analysis
The data are presented as the mean±s.d. and statistical evaluation for data analysis was determined by using an unpaired Student's test. P<0.05 was considered to indicate a statistically significant result. *P<0.05, **P<0.01, ***P<0.001; ns indicates no significant differences from control. Survival analysis was performed using the Kaplan–Meier plotter tool (http://kmplot.com/analysis/).
Footnotes
Author contributions
Conceptualization: L.Z., T.X.; Methodology: J.H., X.L., X.G., Q.G., Y.X., L.Z.; Software: J.H., X.G., Q.G., C.X., Z.Z., L.Z.; Validation: J.H., Y.X., T.X.; Formal analysis: J.H., Q.G., C.X.; Investigation: J.H., X.L.; Resources: L.Z., T.X.; Data curation: X.L., X.G.; Writing - original draft: J.H.; Writing - review & editing: X.L., Z.Z.; Supervision: Y.X.; Project administration: L.Z., T.X.; Funding acquisition: T.X.
Funding
This work was supported by the National Major Special Program of New Drug Research and Development (no. 2013ZX09301303-005), the Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions and Qing Lan Project, National Natural Science Foundation of China (no. 81372331), and the Fundamental Research Funds for the Central Universities (no. 2015ZD004).
References
Competing interests
The authors declare no competing or financial interests.