Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cell. Author manuscript; available in PMC Oct 14, 2012.
Published in final edited form as:
Cell. Oct 14, 2011; 147(2): 382–395.
doi:  10.1016/j.cell.2011.09.032
PMCID: PMC3236086

In vivo identification of tumor suppressive PTEN ceRNAs in an oncogenic BRAF-induced mouse model of melanoma



We recently proposed that competitive endogenous RNAs (ceRNAs) sequester microRNAs to regulate mRNA transcripts containing common microRNA recognition elements (MREs). However, the functional role of ceRNAs in cancer remains unknown. Loss of PTEN, a tumor suppressor regulated by ceRNA activity, frequently occurs in melanoma. Here, we report the discovery of significant enrichment of putative PTEN ceRNAs among genes whose loss accelerates tumorigenesis following Sleeping Beauty insertional mutagenesis in a mouse model of melanoma. We validated several putative PTEN ceRNAs and further characterized one, the ZEB2 transcript. We show that ZEB2 modulates PTEN protein levels in a microRNA-dependent, protein coding-independent manner. Attenuation of ZEB2 expression activates the PI3K/AKT pathway, enhances cell transformation, and commonly occurs in human melanomas and other cancers expressing low PTEN levels. Our study genetically identifies multiple putative microRNA decoys for PTEN, validates ZEB2 mRNA as a bona fide PTEN ceRNA, and demonstrates that abrogated ZEB2 expression cooperates with BRAFV600E to promote melanomagenesis.


Melanoma is estimated to affect more than 70,000 people in the US in the year 2011 and, despite extensive research and clinical efforts, remains fatal in the majority of patients with metastatic disease (http://www.cancer.gov/). Aberrant activation of the MAPK signaling pathway plays a central role in melanoma development, as exemplified by the frequent occurrence of activating mutations in BRAF (Brose et al., 2002; Davies et al., 2002). Genetic and molecular analyses have demonstrated that oncogenic BRAFV600E represents an initiating event in the evolution of melanoma (Davies et al., 2002). Indeed, 80% of human nevi harbor a BRAFV600E mutation (Pollock et al., 2003). Moreover, mouse models of BRAFV600E develop melanoma only after a long latency and with incomplete penetrance (Dankort et al., 2009; Dhomen et al., 2009; FAK, DP, DAT in preparation), suggesting that additional mutations are required for the formation of frank malignancy. Silencing of the tumor suppressor PTEN represents one such mutation and is observed in approximately 30% of human melanoma cases (Tsao et al., 2004). In mice, complete or partial PTEN loss dramatically accelerates BRAFV600E-induced melanoma (Dankort et al., 2009), thus highlighting the oncogenic potential of combined hyperactivation of PI3K/AKT and MAPK signaling.

MicroRNAs (miRNAs) have been shown to regulate PTEN and thus contribute to cell transformation mediated by aberrant activation of the PI3K/AKT pathway (Poliseno et al., 2010a). miRNAs are endogenous, non-coding ~22 nucleotide RNA molecules that bind to microRNA response elements (MREs) contained in their target mRNAs (Bartel, 2009; Thomas et al., 2010). This association recruits the RNA-induced silencing complex (RISC) to target mRNA transcripts, thereby antagonizing their stability and/or translation (Bartel, 2009; Thomas et al., 2010). miRNA-mediated modulation of mRNA levels is conserved in most eukaryotic organisms and is considered a mechanism to fine-tune gene expression. In recent years, numerous examples of abnormal gene regulation by miRNA mis-expression have been demonstrated to contribute to pathological conditions ( mRNAs harbor multiple MREs and thus can be regulated by several miRNAs, while miRNAs are known to target dozens of mRNA transcripts. The fact that distinct RNA molecules can be targeted by common miRNAs led us to propose that related, highly homologous mRNAs, such as gene-pseudogene pairs, may act as miRNA decoys for each other. Pseudogenes are considered “junk DNA” as they lack a protein-coding function (D'Errico et al., 2004). However, by binding to common miRNAs, pseudogene mRNAs may maintain the balance between their ancestral genes and such miRNAs. Indeed, we have recently demonstrated that the PTEN pseudogene transcript PTENP1 regulates the levels of PTEN through sequestration of shared miRNAs (Poliseno et al., 2010b).

On this basis, we further hypothesized that the concept of gene regulation by competition for common miRNAs is not limited to pseudogenes and can be extended to mRNAs and long non-coding RNAs, and have termed RNA molecules that act as miRNA decoys as “competitive endogenous RNAs” (ceRNAs) (Salmena et al., 2011). Importantly, we proposed that the mRNA and the protein encoded by ceRNA genes may be involved in distinct biological processes (Salmena et al., 2011). Employing bioinformatics-guided prediction methods of MRE overlap, we have discovered that multiple mRNAs serve as ceRNAs for PTEN (Tay et al., this issue). Importantly, the proteins encoded by PTEN ceRNAs have thus far not been associated with the regulation of PTEN, suggesting that in some instances mRNAs and the proteins they encode may be involved in distinct biological processes.

Our recent work suggests that mRNAs may act as tumor suppressors or oncogenes through their ceRNA activity. However, whether aberrant ceRNA expression is associated with cancer development in general, and whether loss of PTEN ceRNAs promotes BRAFV600E-induced melanoma in vivo in particular is unknown. Here, we report a striking enrichment for PTEN ceRNAs among genes that were identified in a transposon mutagenesis screen in an oncogenic BRAF-driven mouse model of melanoma. Detailed functional analysis of one such putative PTEN ceRNA, ZEB2, validated its protein-independent and miRNA-dependent ability to regulate PTEN expression. Moreover, we show that ZEB2 and PTEN are co-regulated and that ZEB2 levels are commonly attenuated in human cancers.


Identification of putative PTEN ceRNAs using in vivo Sleeping Beauty insertional mutagenesis

Oncogenic BRAFV600E is an initiating mutation in melanoma, and while some other mutations such as loss of PTEN commonly occur in melanoma, the full spectrum of tumor-promoting genetic events remains to be determined. To this end, we performed a forward genetic screen utilizing the Sleeping Beauty transposon system in a B-RafV619E-driven mouse model of melanoma (Suppl. Figure 1 and Extended Experimental Procedures), in which B-RafV619E corresponds to human BRAFV600E. A detailed description and in-depth analysis of this screen will be reported elsewhere (FAK, DP, DAT; unpublished). Briefly, we created mutant mice that carried the following alleles: LoxP-STOP-LoxP-B-RafV619E (LSL-B-RafV619E, FAK, DP, DAT; unpublished), Tyrosinase-CreERt2 (TyrCreERt2, Bosenberg et al., 2006), rosa26-LoxP-STOP-LoxPSleepingBeauty13 (LSL-SB, PAP-M and DAT; submitted), and T2Onc (Collier et al., 2005). Treatment of compound mutant LSL-B-RafV619E; TyrCreERt2; LSL-SB; T2Onc mice with 4-OH Tamoxifen activated the melanocyte-specific Cre, leading to excision of the STOP cassettes and expression of endogenous oncogenic B-Raf and SB13. Sleeping Beauty transposase-mediated ‘hopping’ of the T2Onc transposon resulted in insertional mutagenesis, thereby accelerating melanoma development (Figure 1A, Suppl. Figure 1 and data not shown). 454 sequencing of genomic melanoma DNA identified 320 genes with a significant enrichment of transposon insertions, termed common insertion sites (CIS). Importantly, PTEN was among the most significant CIS (data not shown), thus demonstrating the ability of our Sleeping Beauty mutagenesis approach to identify key genes altered in melanoma.

Figure 1
(related Table S1 and Figure S1): Identification of putative PTEN ceRNAs. (A) Schematic outline of our hypothesis: oncogenic BRAF mediates the transformation from melanocytes to nevi, and additional loss of PTEN promotes progression to melanoma. PTEN ...

As loss of PTEN expression cooperates with BRAFV600E (Dankort et al., 2009; Tsao et al., 2004) and PTEN expression is regulated by ceRNAs (Poliseno et al., 2010b; Tay et al., this issue), we sought to determine whether Sleeping Beauty identified putative PTEN ceRNAs that cooperate with B-RafV619E to accelerate melanoma development (Figure 1A). We performed a mutually targeted MRE enrichment (MuTaME) analysis (Figure 1B, see also Tay et al., this issue) using the TargetScan prediction algorithm (Friedman et al., 2009; Grimson et al., 2007; Lewis et al., 2005; www.targetscan.org) for MREs located in 3′UTRs. We postulated that the ability of any given mRNA to act as a miRNA decoy for PTEN increases with the number of MREs they share. The 3′UTR of murine PTEN was predicted to contain MREs for 39 different miRNAs (Suppl. Table 1). We set a stringent cut-off of at least 7 shared MREs between PTEN and putative PTEN ceRNAs for the MuTaME analysis (Figure 1B). Using these conditions we identified 33 candidate PTEN ceRNAs among the 320 CIS discovered in the Sleeping Beauty screen (Figure 1C). Notably, this represented a significant enrichment of putative PTEN ceRNAs (p=7.76×10-11) as the MRE-based overlap with PTEN expected by chance was only ~9 genes. Thus, our Sleeping Beauty approach uncovered putative PTEN ceRNAs that promote cancer in an in vivo model of melanoma.

We ranked the 33 putative PTEN ceRNAs according to their similarity with PTEN (Figure 2A). The similarity score is based on a Poisson distribution: we used a linear combination, with equal and opposite weights, of the dissimilarity based on distinct occurrences and the additive similarity as defined previously (van Helden, 2004). This approach utilized all miRNAs predicted by TargetScan and the total number of their MREs located in the 3′UTRs to obtain the similarity score for each CIS. Furthermore, most candidate PTEN ceRNAs contain several MREs for the same miRNAs, and thus the total numbers of MREs shared with PTEN are even greater (Suppl. Table 1).

Figure 2
(related to Figure S2): Putative PTEN ceRNAs modulate PTEN expression levels. (A) List of candidate PTEN ceRNAs identified by Sleeping Beauty. The 33 putative PTEN ceRNAs are ranked by their similarity score, which is based on the identity and number ...

PTEN ceRNAs modulate expression of PTEN

To examine whether the identified putative PTEN ceRNAs regulate expression of PTEN, we performed RNAi-mediated gene silencing in human melanoma cells using pools of 4 siRNAs to reduce off-target effects. We selected 8 putative PTEN ceRNAs (AFF1, DCBLD2, JARID2, MBNL1, RBM9, TNRC6a, TNRC6b, and ZEB2) and depleted them in WM35 cells (Figure 2B). Knock-down of 7 genes led to reduction of PTEN, 6 of which significantly attenuated PTEN expression (AFF1, JARID2, MBNL1, RBM9, TNRC6a, TNRC6b, ZEB2). A reduction of PTEN mRNA levels following PTEN ceRNA silencing was also evident in WM35 cells (Suppl. Figure 2A). In a second human melanoma cell line, A375, decreased PTEN expression was also observed following PTEN ceRNA knockdown, albeit at lower levels (Suppl. Figure 2C, D). The knock-down efficiencies are shown in Suppl. Figures 2B and 2E. Interestingly, depletion of PTEN resulted in downregulation of the putative PTEN ceRNA in some instances (Suppl. Figure 2B,E), indicating that the regulatory relationship between PTEN and PTEN ceRNAs might be reciprocal. Importantly, we found that CNOT6L, a PTEN ceRNA that we predicted using the rna22 algorithm and validated in prostate cancer (Tay et al., this issue), was a CIS identified by Sleeping Beauty. Similar to its ceRNA function in prostate cancer cells (Tay et al., this issue), siRNA-mediated knock-down of CNOT6L significantly reduced PTEN expression in both human melanoma cell lines (Figure 2B, Suppl. Figure 2C).

Critically, the T2Onc transposons inserted in PTEN and the putative PTEN ceRNAs in both orientations throughout the genes (Suppl. Figure 2F). This insertion pattern is indicative of gene repression via the polyadenylation signals rather than gene activation by the MSCV promoter (Collier et al., 2005). Thus, these CIS are likely tumor suppressors. Taken together, these results indicate that the candidate ceRNAs identified by Sleeping Beauty may indeed act as tumor suppressive ceRNAs for PTEN.

ZEB2 silencing results in attenuated PTEN expression

As silencing of several of the putative PTEN ceRNAs led to reduced PTEN protein levels, we sought to further validate and characterize the ability of a CIS mRNA to function as a PTEN ceRNA. None of the 7 CIS that reduced PTEN levels in WM35 melanoma cells codes for a known tumor suppressor protein. The ZEB2 protein, however, has been well established as an activator of the epithelial-to-mesenchymal transition (EMT) (Gregory et al., 2008; Vandewalle et al., 2005), and therefore plays a critical role in the progression of epithelial cancers. Whether the ZEB2 protein is pro-tumorigenic in melanoma is unknown; however, as a PTEN ceRNA, the ZEB2 mRNA may be tumor suppressive. Given the potentially distinct function of the ZEB2 protein and transcript, we decided to examine whether ZEB2 mRNA exerts a tumor suppressive function in melanoma through its competition for PTEN-targeting miRNAs.

Transposon insertions in the 5′-3′ orientation in the ZEB2 locus could theoretically allow for overexpression of the ZEB2 3′UTR, which would potentially sequester miRNAs from PTEN and result in elevated PTEN levels. To ascertain that the ZEB2 3′UTR is not overexpressed in melanomas with such insertions, we analyzed the expression ratio of the ZEB2 5′UTR and 3′UTR. Importantly, we observed a 1:1 ratio of expression in all cases, thus excluding overexpression of the 3′UTR (Figure 2C). As an additional control, we tested the effects of transposon insertion in the PTEN locus. Similarly, tumors with transposon insertions in the 5′-3′ orientation did not display an increase in the PTEN 3′UTR : 5′UTR ratio (Figure 2C), indicating that the transposon acts as a gene trap in these cases. We further tested whether ZEB2 protein levels are affected by transposon insertions. PTEN and ZEB2 are readily detectable by immunoblotting in melanomas without transposon insertions (lanes 1-4, Figure 2D). In contrast, transposon insertion in either PTEN (lanes 5-7) or ZEB2 (lanes 8-10) reduced expression of both proteins (Figure 2D). These data confirm that transposon insertion in PTEN and ZEB2 represses gene expression and indicate that ZEB2 reduction modulates protein levels of PTEN. Moreover, the decrease in ZEB2 expression in tumors with a PTEN transposon insertion suggests that the PTEN-ZEB2 miRNA decoy mechanism may be reciprocal.

To further examine the regulation of PTEN by ZEB2 via miRNA sequestration, we used pools of 4 siRNAs to deplete PTEN or ZEB2 in a primary murine melanoma cell line, TB13602, isolated from a LSL-B-RafV619E; TyrCreERt2 mouse. Knock-down of ZEB2 in TB13602 cells reduced PTEN protein levels by approximately 60% (Figure 3A). RNAi-mediated silencing of PTEN led to a slight attenuation of ZEB2 expression (Figure 3A), similar to the results obtained in melanomas with PTEN or ZEB2 CIS (Figure 2D).

Figure 3
(related to Figure S3): ZEB2 depletion downregulates PTEN. (A) ZEB2 silencing lowers PTEN protein levels in murine melanoma cells TB13602. Western analysis for PTEN and ZEB2 expression and HSP90 as loading control is shown in the left panel. Quantification ...

As PTEN is a major antagonist of PI3K/AKT signaling, we examined if ZEB2-mediated reduction of PTEN activates this pathway. ZEB2 depletion had no effect on AKT phosphorylation at steady state levels (data not shown); however, upon starvation and re-stimulation, ZEB2-depleted cells displayed elevated AKT activation compared to control cells (Figure 3B).

We next analyzed if ZEB2 acts as a PTEN ceRNA in human melanoma cells. In 4 human melanoma cell lines PTEN protein levels were reduced following knock-down of ZEB2 (Figure 3C,D, Suppl. Figure 3A). Moreover, depletion of ZEB2 led to AKT activation in these cell lines (Figure 3C,E, Suppl. Figure 3A). Importantly, ZEB2 siRNAs without potential seed matches to PTEN efficiently lowered PTEN levels in mouse and human melanoma cells (Suppl. Figure 3B), indicating that PTEN downregulation is not due to off target effects of the ZEB2 siRNA pools. Interestingly, ZEB2 depletion had only minor and statistically insignificant effects on PTEN mRNA levels in TB13602 and A375 cells (Figure 3F,G), while it significantly reduced PTEN transcript in WM35 cells (Figure 3H). These data suggest that the miRNAs mediating the functional interaction of PTEN and ZEB2 regulate mRNA stability in a cell line dependent fashion, while their control of translation is a more universal phenomenon.

Repression of PTEN by ZEB2 loss is 3′UTR and miRNA dependent

These in vitro and in vivo data support notion that ZEB2 acts as a PTEN ceRNA. They do not exclude, however, the transcriptional regulation or protein stability as potential mechanisms of ZEB2 loss-mediated PTEN reduction. We addressed these possibilities by ectopic expression of a Luciferase-PTEN3′UTR reporter construct in TB13602 cells, followed by knock-down of PTEN or ZEB2. Silencing of either PTEN or ZEB2 increased the availability of shared miRNAs, thereby suppressin the Luciferase-PTEN3′UTR reporter as measured by diminished Luciferase activity (Figure 4A). Importantly, as the Luciferase-PTEN3′UTR reporter construct is expressed from a CMV promoter and does not code for a PTEN peptide, reduced Luciferase activity was dependent on the PTEN 3′UTR, thus eliminating transcriptional regulation and protein stability as explanations for the ZEB2-PTEN functional interaction.

Figure 4
(related to Figure S4): Regulation of PTEN by ZEB2 is 3′UTR- and miRNA-dependent. (A) Effect of ZEB2 depletion on Luciferase activity of a Luciferase-PTEN3′UTR reporter. Knock-down of PTEN (positive control) and ZEB2 reduces Luciferase ...

To exclude the involvement of RNA-binding proteins that could regulate PTEN mRNA translation through its 3′UTR, we used cells deficient for the miRNA biogenesis protein Dicer to examine whether ZEB2 controls PTEN levels in a miRNA-dependent manner. In Dicer wild-type HCT116 colon carcinoma cells, siRNA against ZEB2 lowered PTEN levels, similar to our observation in melanoma cells (Figure 4B). In contrast, ZEB2 depletion failed to reduce PTEN expression in Dicer null HCT116 cells (Figure 4B). Moreover, ZEB2 knock-down decreased PTEN mRNA only in Dicer wild-type cells but not in Dicer deficient cells (Figure 4C). Conversely, PTEN silencing resulted in diminished ZEB2 mRNA levels in Dicer wild-type HCT116 cells, which was rescued in Dicer deficient HCT116 cells (Figure 4C). Similar to PTEN attenuation, AKT activation by ZEB2 silencing is only evident in Dicer wild-type HCT116 cells (Figure 4D). These data indicate that regulation of PTEN expression by ZEB2 is indeed miRNA dependent.

As ablation of ZEB2 mRNA reduces PTEN expression due to increased availability of common miRNAs, overexpression of ZEB2 mRNA should have the opposite effect.

However, ectopic expression of full-length ZEB2 mRNA would also increase ZEB2 protein. We therefore overexpressed only the ZEB2 3′UTR to uncouple potential effects of elevated ZEB2 protein on PTEN expression, such as transcriptional regulation, from the miRNA-based regulation via MREs. This approach was also used to show that the 3′UTRs of other putative PTEN ceRNAs increase PTEN levels in prostate cancer (Poliseno et al., 2010b, Tay et al., this issue). We expressed the 3′UTRs of ZEB2 or PTEN in A375 and TB13602 melanoma cells. Cells were co-transfected with the Luciferase-PTEN3′UTR reporter construct to measure the ceRNA activity of ectopically expressed 3′UTRs on transcripts containing the PTEN 3′UTR. Critically, the ZEB2 3′UTR significantly increased the activity of the Luciferase-PTEN3′UTR reporter (Figure 4E, F). Moreover, PTEN protein levels were increased when the PTEN and ZEB2 3′UTRs were overexpressed in A375 cells (Figure 4G) and Dicer wild-type HCT116 cells (Figure 4H). Elevated expression of PTEN was rescued by Dicer deficiency in HCT116 cells (Figure 4H). Similarly, overexpression of the ZEB2 3′UTR significantly increased the activity of the Luciferase-PTEN 3′UTR reporter in HCT116 cells and Dicer deficiency partially negated this effect. Critically, overexpression of the ZEB2 coding sequence had no effect on PTEN protein levels in HCT116 cells (Suppl. Figure 4). Thus, depletion as well as overexpression of ZEB2 mRNA alters the balance between PTEN mRNA and common miRNAs, thereby impacting PTEN expression.

Analysis of miRNAs common to PTEN and ZEB2

To elucidate which of the miRNAs predicted to target PTEN and ZEB2 mediate the observed cross-talk, we further characterized the effect of these miRNAs on PTEN and ZEB2 expression. First, we interrogated which MREs were predicted to be common between the 3′UTRs of PTEN and ZEB2. TargetScan predicted that PTEN and ZEB2 are targets for 9 shared miRNA families, and that they contain a total of 14 and 16 MREs for these miRNAs, respectively (Figure 5A). Of these miRNAs, several have been validated as PTEN-targeting miRNAs (miR-25, miR-32, miR-92ab, miR-141, miR-144, miR-363, miR-367) (Lee et al., 2010; Poliseno et al., 2010a; Zhang et al., 2010), while only the miR-200 family has been validated as repressors of ZEB2 (Gregory et al., 2008; Korpal et al., 2008; Park et al., 2008). We next delivered miRNA mimics for one member of each miRNA family predicted to target PTEN and ZEB2 to TB13602 cells and analyzed PTEN and ZEB2 protein levels. Of the 10 miRNAs tested, four targeted PTEN and ZEB2 to some extent (miR-181, miR-200b, miR-25, miR-92a) (Figure 5B). We next determined the expression levels of these 10 miRNAs in melanoma cells. Interestingly, miRNA expression levels were similar between the 3 different melanoma cell lines analyzed (Figure 5C) and all four miRNAs that target PTEN and ZEB2 were expressed in the melanoma cell lines (Figure 5C).

Figure 5
Characterization of miRNAs that are common to PTEN and ZEB2. (A) miRNA Response Elements (MREs) shared by PTEN and ZEB2. This table depicts the 9 miRNAs that the 3′UTRs of PTEN and ZEB2 have in common, as well as the number of sites for each miRNA. ...

Next, we examined whether the 4 suppressive miRNAs (miR-181, miR-200b, miR-25, miR-92a) that are expressed in melanoma associate with the 3′UTRs of PTEN and ZEB2. To this end, we performed RNA immunoprecipitations (RIPs) in melanoma cells using the PTEN and ZEB2 3′UTRs as baits. Notably, all four miRNAs associated with the 3′UTRs of PTEN and ZEB2, while a control miRNA with no predicted MRE in PTEN and ZEB2 bound neither 3′UTR (Figure 5D). Furthermore, knock-down of either PTEN or ZEB2 increased miRNA availability, as determined by increased association of miR-92a with the MS2-PTEN 3′UTR bait mRNA (Figure 5E). Taken together, ZEB2 sequesters at least four miRNAs (miR-181, miR-200b, miR-25, miR-92a) to regulate PTEN levels and ZEB2 downregulation increases the availability of at least one of these miRNAs.

ZEB2 displays tumor suppressive properties in melanoma cells

As decreased PTEN expression drives the formation of numerous types of cancer, we determined whether aberrant control of PTEN by loss of ZEB2 enhanced oncogenic transformation of melanoma cells. Indeed, knock-down of either PTEN or ZEB2 increased proliferation of TB13602 cells (Figure 6A) and WM35 cells (Figure 6B), while neither PTEN nor ZEB2 silencing significantly altered the proliferation rate of A375 cells (Figure 6C). Importantly, anchorage-independent growth in soft agar was enhanced by depletion of PTEN or ZEB2 in all three melanoma cell lines (Figure 6D-F). We next determined whether this ceRNA cross-talk would be operational in vivo. To this end, we lentivirally delivered a short hairpin against ZEB2 to TB13602 cells and examined their oncogenic properties. shRNA-mediated knock-down of ZEB2 lowered expression of PTEN (Figure 6G), accelerated proliferation (Figure 6H), and increased the in vivo growth of xenografted tumors in nude mice (Figure 6I). These data support the notion that ZEB2 has tumor suppressive activity in melanoma cells, which is, at least in part, due to ZEB2 mRNA-mediated regulation of PTEN expression.

Figure 6
ZEB2 displays tumor suppressive properties in melanoma cells. (A-C) Proliferation curves of TB13602 (A), WM35 (B), and A375 (C) melanoma cells treated with siPTEN, siZEB2 and NC is shown. (D-F) Anchorage-independent growth in soft agar. Representative ...

Functional cross-talk of ZEB2 and PTEN in human cancer

We reasoned that if PTEN and ZEB2 expression is linked through several shared miRNAs, then their mRNA expression levels might be co-regulated. We interrogated mRNA levels in a set of human primary melanoma samples (Halaban et al., 2009) and found that PTEN and ZEB2 expression indeed significantly correlated (Figure 7A). Using the same set of expression data, we asked whether PTEN and ZEB2 expression is diminished in melanomas compared to normal melanocytes. Neither PTEN nor ZEB2 mRNA was significantly decreased in melanomas in this dataset (Suppl. Figure 5A), which is in line with the finding that PTEN expression is lost in only 30% of melanomas (Tsao et al., 2004). We therefore examined whether decreased ZEB2 expression would be evident in tumors with reduced PTEN levels. We subdivided the tumor specimens into two groups: melanomas with PTEN expression above average (“PTEN high”), and tumors with PTEN expression below average (“PTEN low”). Intriguingly, when compared to normal melanocytes, ZEB2 expression was significantly decreased only in the “PTEN low” subset (Figure 7B, Suppl. Figure 5A), suggesting that ZEB2 and PTEN mRNAs may co-regulate each other in melanoma.

Figure 7
(related to Figure S5): Functional cross-talk of PTEN and ZEB2 in human cancer. (A) mRNA expression of PTEN and ZEB2 correlates in human melanoma. Graph showing mRNA expression levels for PTEN and ZEB2 in human melanoma samples, where each dot corresponds ...

We further determined whether the PTEN-ZEB2 relationship is specific to melanoma or whether other tumors types display a similar mode of PTEN regulation. Indeed, we identified significant co-expression of PTEN and ZEB2 in primary prostate cancer (Figure 7C). Moreover, ZEB2 expression is significantly reduced in primary prostate cancer samples with a “PTEN low” expression profile (i.e. PTEN expression level below average) when compared to normal prostatic epithelium or all prostate tumors (Figure 7D, Suppl. Figure 5B). We analyzed 3 additional mRNA expression datasets for melanoma, colon carcinoma and glioblastoma and found that both PTEN and ZEB2 were significantly downregulated in these tumors when compared to normal tissue (Figure 7E-G). Finally, by interrogating human-mouse conserved and human-specific mRNA co-expression networks we found that PTEN and ZEB2 mRNA show a significant correlation of co-expression in several human tissues (Suppl. Figure 5C). Taken together, our data demonstrate that ZEB2 sequesters miRNAs from PTEN and thereby regulates PTEN expression. This mode of PTEN regulation was identified in murine and human melanomas, as well as human prostate, colon and brain cancer.


We report here the identification, functional characterization and relevance of ceRNAs in cancer biology in vivo. By means of a forward genetics approach using Sleeping Beauty insertional mutagenesis in a mouse model of melanoma, we discovered multiple putative ceRNAs for the tumor suppressor PTEN. Further in vitro characterization validated the EMT regulator ZEB2 as a PTEN ceRNA, and human cancer data corroborated its functional relationship with PTEN. We therefore provide evidence that aberrant regulation of PTEN via miRNA competition by ceRNAs contributes to melanoma development.

The results presented here further support our hypothesis that protein-coding mRNAs communicate and co-regulate each other through competition for miRNAs that target both transcripts (Salmena et al., 2011; Tay et al., this issue). Importantly, as regulation through miRNA sequestration is solely based on MREs, our findings ascribe a predictable and protein coding-independent function to mRNA molecules. Furthermore, we establish a means of regulatory interaction between mRNAs that is based on MREs as functional units. MREs can be found on both protein-coding and non-coding mRNAs, thus challenging the notion that mRNAs are mere blueprints for peptide synthesis. While other examples of coding-independent functions of mRNAs have been previously described (Candeias et al., 2008; Jenny et al., 2006), our findings bestow on any RNA molecule functions that may be predicted based on their MRE sequence (Salmena et al., 2011). Our findings of a microRNA-mediated and MRE-dependent gene regulatory ceRNA dimension are corroborated by two other studies demonstrating ceRNA activity for protein-coding (Sumazin et al., this issue) and non-coding RNA (Cesana et al., this issue) molecules.

Using bioinformatics approaches and in vitro experimental validation, we have successfully predicted and characterized several ceRNAs that regulate the tumor suppressor PTEN (Tay et al., this issue). In the work presented here, we undertook an unbiased genetic in vivo approach to uncover mutational events that cooperate with oncogenic BRAF to promote melanoma development. Not only did this approach identify PTEN, a known cooperator of oncogenic BRAF in melanoma, but it also generated a list of CIS that was significantly enriched for putative PTEN ceRNAs. Thus, our genetic approach represents an alternative and more functional way of identifying PTEN ceRNAs, which complements our bioinformatics-based ceRNA prediction method.

We have previously used the rna22 prediction algorithm in conjunction with 10 validated, 3′UTR-binding PTEN miRNAs (miR-17-5p, miR-19a, miR-19b, miR-20a, miR-20b, miR-26a, miR-26b, miR-93, miR-106a, miR-106b) to predict PTEN ceRNAs (Tay et al., this issue). To screen the murine melanoma CIS for candidate PTEN ceRNAs, we instead employed TargetScan, as this algorithm considers MRE conservation between mammals. TargetScan predicted PTEN as a target for 39 different miRNAs (Suppl. Table 1), which were used to screen the Sleeping Beauty CIS to identify putative PTEN ceRNAs. However, we also compared our list of 320 CIS with the rna22-generated list of putative PTEN ceRNAs and found an overlap of 4 genes: CNOT6L, MEF2A, PDS5B, and ROCK2. These 4 putative PTEN ceRNAs were not predicted by TargetScan, which is due to the fact that TargetScan considers the miR-17-5p/20ab/93/106ab/519, miR-19ab, and miR-26ab families each as single MREs, while we considered them as individual MREs for our rna22-based prediction of PTEN ceRNAs. Thus, while CNOT6L, MEF2A, PDS5B, and ROCK2 share more than 7 MREs with PTEN if they are considered individually (rna22 approach), they fall under the threshold of 7 MREs if each miRNA family is considered as one MRE (TargetScan approach). Thus, rna22 and TargetScan prediction algorithms, as well as the use of only validated or all predicted MREs, can successfully be used to predict PTEN ceRNAs. These findings are of great relevance as they validate and generalize our MuTaME predictions irrespective of the algorithms employed.

We have proposed a set of rules and a methodology to validate and characterize ceRNAs (Tay et al., this issue). By applying this methodology, we have confirmed that ZEB2 indeed acts as a ceRNA for the tumor suppressor PTEN. Indeed, we found a significant correlation between the expression of ZEB2 and PTEN in human melanomas and prostate cancer (Figure 7). Moreover, compared to benign tissues, ZEB2 expression was significantly reduced in tumors with low PTEN expression levels (Figure 7). Thus, ZEB2 has tumor suppressive properties. The ZEB2 protein promotes EMT by repressing expression of E-Cadherin (Vandewalle et al., 2005), and thus may be involved in promoting cancer progression and metastasis in some instances of epithelial cancers. Indeed, the miR-200 family regulates expression of ZEB2 (Gregory et al., 2008; Korpal et al., 2008; Park et al., 2008), and is commonly associated with EMT and cancer progression (Mezzanzanica et al., 2010; O'Day and Lal, 2010; Pang et al., 2010). In melanoma cells, miR-200 does not appear to repress EMT and invasion but rather different miR-200 family members mediate alternative modes of melanoma cell migration (Elson-Schwab et al., 2010). It has been reported that ZEB2 expression is suppressed by p53-mediated up-regulation of miR-200 (Kim et al., 2011), suggesting that p53 deficiency promotes EMT through induction of ZEB2. However, p53 mutations are rare in melanoma (www.sanger.ac.uk/genetics/CGP/cosmic/), making up-regulation of ZEB2 via this route unlikely. Moreover, whether EMT is involved in progression of cancers of non-epithelial origin, such as melanoma is unclear. We hypothesized that in certain instances mRNA and protein encoded by the same gene may exert different biological effects (Salmena et al., 2011). ZEB2 powerfully exemplifies such a case, acting as a tumor suppressor by regulating PTEN expression through its mRNA in melanoma, while the protein promotes tumor progression and metastasis by controlling EMT in epithelial cancers.

Insertional mutagenesis screens using retroviruses or transposons have been widely used to discover cancer genes. Our work encourages the integration of the classic “protein dimension” with an additional “ceRNA dimension” when analyzing the “hits” of such screens. By doing so, genes that are initially classified as false positives based on their protein function, may in fact be cancer-promoting genes by means of their ceRNA activity. In addition, previously performed insertional mutagenesis screens could be re-interrogated for the presence of ceRNAs of prominent cancer genes using bioinformatics MRE prediction methods described here and in Tay et al. This may reveal a so far unappreciated genetic dimension with a role in cancer development and the pathogenesis of other human conditions.

Experimental Procedures


The 3′UTRs of mouse and human ZEB2 and human PTEN were amplified by PCR from genomic DNA and cloned into pCDNA3 or pCMV according to standard procedures. Generation of PTEN3′UTR-psiCHECK-2 is described by Tay et al. Primer sequences are available upon request.

Cell culture and transfection

A375, WM35, 451Lu and WM278 human melanoma cells were obtained from M. Herlyn (Wistar Institute) and cultured as previously reported (Tsao et al., 2004). TB13602, HCT116 wild-type and Dicer-/- cells were grown in DMEM supplemented with 10% FCS, penicillin/streptomycin and glutamine at 37°C in a humidified atmosphere with 5% CO2. Cells were transfected with 100nM siRNAs and Dharmafect 1 or 1.5ug of plasmid DNA and Lipofectamine 2000 in 12 well plates according to the manufacturer's recommendations for transfection.

Western blot analysis

Cells were lysed in RIPA buffer containing Complete Mini protease inhibitors (Roche) and a Phosphatase Inhibitor cocktail (Sigma). 5-20 μg of total protein were subjected to SDS–PAGE on 4-12% Bis-Tris acrylamide NuPAGE gels in MOPS SDS buffer (Invitrogen). The following primary antibodies were used: HSP90 (BD), PTEN, phospho-AKT (p473), total-AKT (all Cell Signaling), ZEB2, actin (both Santa Cruz Biotechnologies). Subsequently, membranes were incubated with secondary, HRP-tagged antibodies (Amersham) and signals were visualized with ECL or ECL plus (Amersham).

RNA extraction and real-time PCR

For real-time PCR analyses, total RNA was extracted using the Qiashredders and RNeasy Mini Kit (both Qiagen) according to the manufacturer's recommendations. cDNA was synthesized using the High Capacity cDNA Reverse Transcriptase Kit according to the manufacturer's instructions (Applied Biosystems) and analyzed by real-time PCR using taqman gene expression assays (Applied Biosystems) on a LightCycler 480 System (Roche Applied Science).

Luciferase assays

TB13602 or A375 cells were transfected with 150ng of empty psiCHECK2 vector or psiCHECK2-PTEN3′UTR and either 100nM siRNA or 1μg 3′UTR vector constructs using Lipofectamine 2000 according to manufacturer's recommendations. In all transfections, firefly luciferase activity was used as a normalization control for transfection efficiency. 72 hours after transfection, luciferase activities were measured consecutively with the dual luciferase reporter system (Promega) using a luminometer (Promega).

Cell proliferation

Cells were plated in triplicates in 12-well plates at a final density of 2×104/well. On days 1-4 after plating, cells were washed with PBS, fixed in 4% PFA and stained with crystal violet. The dye was extracted with 10% acetic acid and absorbance at 595nm was determined.


1×106 TB13602 cells were mixed with Matrigel and injected into the flanks of NCR nude mice. Tumor growth was measured after 14 days and the volume calculated using the formula 0.5×L×W×H.

Statistical analysis

In vitro data were analyzed using unpaired Student's t-test. Values of P < 0.05 were considered statistically significant. *P < 0.05; **P < 0.01; ***P < 0.001. The mean ± standard error of three or more independent experiments performed in triplicates is reported.

Supplementary Material




We thank Pandolfi and Tuveson laboratory members for critical discussions. qRT-PCR analysis was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (NIH Award #UL1 RR 025758 and financial contributions from Harvard University and its affiliated academic health care centers). The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic health care centers, the National Center for Research Resources, or the National Institutes of Health. We thank B. Vogelstein for DICER-/- cells and M. Bosenberg for TyrCreERt2 mice. We thank F. Connor and other members of the Tuveson lab for assistance, and the animal care staff and histology core at CRI. Mice were maintained in compliance with the UK home office regulations. This research was supported by the University of Cambridge, Cancer Research UK, The Li Ka Shing Foundation, Hutchison Whampoa Limited, the NIHR Cambridge Biomedical Research Centre and The Wellcome Trust. YT was supported by a Special Fellow Award from The Leukemia & Lymphoma Society, DP received an International Fellowship in Cancer Research from the Italian Association for Cancer Research (AIRC), PAPM was supported by the Fundación Ibercaja, UA received a fellowship from the Fondazione per la Ricerca Biomedica ONLUS of Torino, and SMT was supported by a Department of Defense Breast Cancer Research Program fellowship. PP and UA received support from AIRC under grant IG-9408. This work was supported by the Yale SPORE in Skin Cancer funded by the National Cancer Institute grant number 1 P50 CA121974 to RH and by NIH grant R01 CA-82328-09 to PPP.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Bartel DP. MicroRNAs: target recognition and regulatory functions. Cell. 2009;136:215–233. [PMC free article] [PubMed]
  • Bosenberg M, Muthusamy V, Curley DP, Wang Z, Hobbs C, Nelson B, Nogueira C, Horner JW, 2nd, Depinho R, Chin L. Characterization of melanocyte-specific inducible Cre recombinase transgenic mice. Genesis. 2006;44:262–267. [PubMed]
  • Brose MS, Volpe P, Feldman M, Kumar M, Rishi I, Gerrero R, Einhorn E, Herlyn M, Minna J, Nicholson A, et al. BRAF and RAS mutations in human lung cancer and melanoma. Cancer Res. 2002;62:6997–7000. [PubMed]
  • Candeias MM, Malbert-Colas L, Powell DJ, Daskalogianni C, Maslon MM, Naski N, Bourougaa K, Calvo F, Fahraeus R. P53 mRNA controls p53 activity by managing Mdm2 functions. Nat Cell Biol. 2008;10:1098–1105. [PubMed]
  • Collier LS, Carlson CM, Ravimohan S, Dupuy AJ, Largaespada DA. Cancer gene discovery in solid tumours using transposon-based somatic mutagenesis in the mouse. Nature. 2005;436:272–276. [PubMed]
  • D'Errico I, Gadaleta G, Saccone C. Pseudogenes in metazoa: origin and features. Brief Funct Genomic Proteomic. 2004;3:157–167. [PubMed]
  • Dankort D, Curley DP, Cartlidge RA, Nelson B, Karnezis AN, Damsky WE, Jr., You MJ, DePinho RA, McMahon M, Bosenberg M. Braf(V600E) cooperates with Pten loss to induce metastatic melanoma. Nat Genet. 2009;41:544–552. [PMC free article] [PubMed]
  • Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, et al. Mutations of the BRAF gene in human cancer. Nature. 2002;417:949–954. [PubMed]
  • Dhomen N, Reis-Filho JS, da Rocha Dias S, Hayward R, Savage K, Delmas V, Larue L, Pritchard C, Marais R. Oncogenic Braf induces melanocyte senescence and melanoma in mice. Cancer Cell. 2009;15:294–303. [PubMed]
  • Elson-Schwab I, Lorentzen A, Marshall CJ. MicroRNA-200 family members differentially regulate morphological plasticity and mode of melanoma cell invasion. PLoS One. 2010;5 [PMC free article] [PubMed]
  • Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome Res. 2009;19:92–105. [PMC free article] [PubMed]
  • Gregory PA, Bert AG, Paterson EL, Barry SC, Tsykin A, Farshid G, Vadas MA, Khew-Goodall Y, Goodall GJ. The miR-200 family and miR-205 regulate epithelial to mesenchymal transition by targeting ZEB1 and SIP1. Nat Cell Biol. 2008;10:593–601. [PubMed]
  • Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel DP. MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell. 2007;27:91–105. [PMC free article] [PubMed]
  • Halaban R, Krauthammer M, Pelizzola M, Cheng E, Kovacs D, Sznol M, Ariyan S, Narayan D, Bacchiocchi A, Molinaro A, et al. Integrative analysis of epigenetic modulation in melanoma cell response to decitabine: clinical implications. PLoS One. 2009;4:e4563. [PMC free article] [PubMed]
  • Jenny A, Hachet O, Zavorszky P, Cyrklaff A, Weston MD, Johnston DS, Erdelyi M, Ephrussi A. A translation-independent role of oskar RNA in early Drosophila oogenesis. Development. 2006;133:2827–2833. [PubMed]
  • Kim T, Veronese A, Pichiorri F, Lee TJ, Jeon YJ, Volinia S, Pineau P, Marchio A, Palatini J, Suh SS, et al. p53 regulates epithelial-mesenchymal transition through microRNAs targeting ZEB1 and ZEB2. J Exp Med. 2011;208:875–883. [PMC free article] [PubMed]
  • Korpal M, Lee ES, Hu G, Kang Y. The miR-200 family inhibits epithelial-mesenchymal transition and cancer cell migration by direct targeting of E-cadherin transcriptional repressors ZEB1 and ZEB2. J Biol Chem. 2008;283:14910–14914. [PMC free article] [PubMed]
  • Lee DY, Jeyapalan Z, Fang L, Yang J, Zhang Y, Yee AY, Li M, Du WW, Shatseva T, Yang BB. Expression of versican 3'-untranslated region modulates endogenous microRNA functions. PLoS One. 2010;5:e13599. [PMC free article] [PubMed]
  • Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005;120:15–20. [PubMed]
  • Mezzanzanica D, Bagnoli M, De Cecco L, Valeri B, Canevari S. Role of microRNAs in ovarian cancer pathogenesis and potential clinical implications. Int J Biochem Cell Biol. 2010;42:1262–1272. [PubMed]
  • O'Day E, Lal A. MicroRNAs and their target gene networks in breast cancer. Breast Cancer Res. 2010;12:201. [PMC free article] [PubMed]
  • Pang Y, Young CY, Yuan H. MicroRNAs and prostate cancer. Acta Biochim Biophys Sin (Shanghai) 2010;42:363–369. [PubMed]
  • Park SM, Gaur AB, Lengyel E, Peter ME. The miR-200 family determines the epithelial phenotype of cancer cells by targeting the E-cadherin repressors ZEB1 and ZEB2. Genes Dev. 2008;22:894–907. [PMC free article] [PubMed]
  • Piro RM, Ala U, Molineris I, Grassi E, Bracco C, Perego GP, Provero P, Di Cunto F. An atlas of tissue-specific conserved coexpression for functional annotation and disease gene prediction. Eur J Hum Genet. 2011 [PMC free article] [PubMed]
  • Poliseno L, Salmena L, Riccardi L, Fornari A, Song MS, Hobbs RM, Sportoletti P, Varmeh S, Egia A, Fedele G, et al. Identification of the miR-106b~25 microRNA cluster as a proto-oncogenic PTEN-targeting intron that cooperates with its host gene MCM7 in transformation. Sci Signal. 2010a;3:ra29. [PMC free article] [PubMed]
  • Poliseno L, Salmena L, Zhang J, Carver B, Haveman WJ, Pandolfi PP. A coding-independent function of gene and pseudogene mRNAs regulates tumour biology. Nature. 2010b;465:1033–1038. [PMC free article] [PubMed]
  • Pollock PM, Harper UL, Hansen KS, Yudt LM, Stark M, Robbins CM, Moses TY, Hostetter G, Wagner U, Kakareka J, et al. High frequency of BRAF mutations in nevi. Nat Genet. 2003;33:19–20. [PubMed]
  • Salmena L, Poliseno L, Tay Y, Kats L, Pandolfi PP. A ceRNA Hypothesis: The Rosetta Stone of a Hidden RNA Language? Cell. 2011;146:353–358. [PMC free article] [PubMed]
  • Thomas M, Lieberman J, Lal A. Desperately seeking microRNA targets. Nat Struct Mol Biol. 2010;17:1169–1174. [PubMed]
  • Tsao H, Goel V, Wu H, Yang G, Haluska FG. Genetic interaction between NRAS and BRAF mutations and PTEN/MMAC1 inactivation in melanoma. J Invest Dermatol. 2004;122:337–341. [PMC free article] [PubMed]
  • van Helden J. Metrics for comparing regulatory sequences on the basis of pattern counts. Bioinformatics. 2004;20:399–406. [PubMed]
  • Vandewalle C, Comijn J, De Craene B, Vermassen P, Bruyneel E, Andersen H, Tulchinsky E, Van Roy F, Berx G. SIP1/ZEB2 induces EMT by repressing genes of different epithelial cell-cell junctions. Nucleic Acids Res. 2005;33:6566–6578. [PMC free article] [PubMed]
  • Zhang L, Deng T, Li X, Liu H, Zhou H, Ma J, Wu M, Zhou M, Shen S, Niu Z, et al. microRNA-141 is involved in a nasopharyngeal carcinoma-related genes network. Carcinogenesis. 2010;31:559–566. [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...