MicroRNAs (miRNAs) are small noncoding RNA molecules that negatively control expression of target genes in animals and plants. The microRNA-21 gene (mir-21) has been identified as the only miRNA commonly overexpressed in solid tumors of the lung, breast, stomach, prostate, colon, brain, head and neck, esophagus and pancreas. We initiated a screen to identify miR-21 target genes using a reporter assay and identified a potential miR-21 target in the 3′-UTR of the programmed cell death 4 (PDCD4) gene. We cloned the full-length 3′-UTR of human PDCD4 downstream of a reporter and found that mir-21 downregulated, whereas a modified antisense RNA to miR-21 upregulated reporter activity. Moreover, deletion of the putative miR-21-binding site (miRNA regulatory element, MRE) from the 3′-UTR of PDCD4, or mutations in the MRE abolished the ability of miR-21 to inhibit reporter activity, indicating that this MRE is a critical regulatory region. Western blotting showed that Pdcd4 protein levels were reduced by miR-21 in human and mouse cells, whereas quantitative real-time PCR revealed little difference at the mRNA level, suggesting translational regulation. Finally, overexpression of mir-21 in MCF-7 human breast cancer cells and mouse epidermal JB6 cells promoted soft agar colony formation by downregulating Pdcd4 protein levels. The demonstration that miR-21 promotes cell transformation supports the concept that mir-21 functions as an oncogene by a mechanism that involves translational repression of the tumor suppressor Pdcd4.
MicroRNAs (miRNA) are short 20–25 nucleotide RNA molecules that negatively regulate gene expression in animals and plants. Over 700 human miRNA genes have been identified (Griffiths-Jones et al., 2006), yet the physiological function of only a handful of them has been experimentally determined. The pioneering work on let-7 and RAS in lung cancer (Johnson et al., 2005) and other studies on cancer-related miRNAs have led to a suggested framework to comprehend the role of miRNAs in cancer. miRNA-mediated tumorigenesis results from either downregulation of a tumor suppressing miRNA or upregulation of oncogenic miRNA. Recent reports have identified miR-21 as the only miRNA that is overexpressed in nine types of solid tumors (lung, breast, stomach, prostate, colon, brain, head and neck, esophagus and pancreas) (Chan et al., 2005; Iorio et al., 2005; Diederichs and Haber, 2006; Roldo et al., 2006; Volinia et al., 2006), as well as in diffuse large B-cell lymphoma (Lawrie et al., 2007), chronic lymphocytic leukemia (Calin et al., 2005), uterine leiomyomas (Wang et al., 2007) and malignant hepatocytes (Meng et al., 2007), supporting its involvement in cancer pathogenesis. Increased expression of miR-21 has been implicated in various processes involved in carcinogenesis, including inhibition of apoptosis (Chan et al., 2005), promotion of cell proliferation (Roldo et al., 2006) and stimulation of tumor growth (Si et al., 2007). In addition, increased miR-21 has been associated with chemoresistance in human cholangiocarcinoma cell lines (Meng et al., 2006).
The genes targeted by miR-21 have been under intense study since miRNAs generally function through downregulating target gene mRNAs, thus reducing protein expression of the gene target. The phosphatase and tensin homolog deleted on chromosome 10 (PTEN) gene was first selected as a potential miR-21 target based on its well-characterized role in tumor biology (Meng et al., 2006, 2007). PTEN is a tumor suppressor gene encoding a phosphatase that regulates cell cycle, Akt and p53 activity (Li and Ross, 2007). Another tumor suppressor gene that is downregulated by miR-21 is tropomyosin 1 (TPM1) (Zhu et al., 2007). Recently, two independent reports revealed that yet another tumor suppressor gene, programmed cell death 4 (PDCD4) is a target of miR-21 in colon cancer (Asangani et al., 2007) and MCF-7 human breast cancer cells (Frankel et al., 2008). Pdcd4 was first discovered to inhibit tumor promoter-induced neoplastic transformation in the JB6 murine epidermal model of neoplastic transformation (Cmarik et al., 1999). Pdcd4 inhibits AP-1 transactivation (Yang et al., 2001), stalls the translational machinery (Yang et al., 2003), decreases benign and malignant tumor progression (Jansen et al., 2005) and controls lymphoma initiation and autoimmune inflammation (Hilliard et al., 2006). In this study, we used a reporter screening assay to identify miR-21 target genes. We identified PDCD4 as an evolutionarily conserved and clinically significant target for this miRNA. Further, our results revealed that miR-21 promotes cell transformation by decreasing Pdcd4 protein levels in MCF-7 and JB6 cells.
Results and discussion
PDCD4 3′-UTR contains an MRE for miR-21
Four different computational methods were employed to identify over 100 genes that are predicted to be targets of miR-21 by at least two out of four methods: Miranda, TargetScan, rna22 and PicTar (John et al., 2004; Krek et al., 2005; Lewis et al., 2005; Miranda et al., 2006). Eight of the genes annotated as cancer relevant (Table 1) (Ashburner et al., 2000) were arbitrarily selected and subjected to further screening to assess whether their putative miRNA regulatory elements (MREs) for miR-21 reduced the expression of an upstream reporter when miR-21 levels were downregulated by transfecting HEK-293T cells with a 2′-O-methylated antisense miR-21 (anti-miR-21) (Meister et al., 2004). A random sequence RNA was used as a negative control for anti-miR-21. Figure 1a shows that only two of the eight putative MREs for miR-21 were responsive to inhibition of mir-21 expression. These results are likely to reflect the inherent limitation of computational target prediction. Notably, Rluc reporter expression increased (>25%) in the presence of the MREs from the RASA1 and PDCD4 gene. Thus, these genes were selected for further characterization by cloning the full-length 3′-UTRs of the RASA1 and PDCD4 genes (Table 1) downstream of the reporter and performing a similar assay (Figure 1b). Only the PDCD4 3′-UTR reporter showed upregulation when mir-21 gene expression was inhibited. The lack of induction of reporter activity from the native, full-length 3′-UTR of RASA1 in the presence of anti-miR-21 (Figure 1b) may indicate the putative MRE for miR-21 is inaccessible, perhaps due to folding, in the context of the full 3′-UTR compared to the short MRE (Figure 1a). To confirm the importance of the MRE in the context of the 2 kb full-length 3′-UTR of the PDCD4 gene, a substitution mutation (PDCD4 3′-UTRM) and a deletion mutant (PDCD4 3′-UTRS) were generated and cloned downstream of the reporter gene. Reporter expression did not change significantly with either mutation in the cells transfected with anti-miR-21 (Figure 1b). These results support the conclusion that the 3′-UTR of PDCD4 mRNA contains a functional binding site for miR-21.
Pdcd4 protein is downregulated by miR-21 through translational repression
When mir-21 expression was inhibited by anti-miR-21 in HEK-293T cells (∼50% Figure 2a), the protein level of Pdcd4 was increased ∼3-fold while that of RASA1 was unchanged (Figure 2a). The PDCD4 mRNA level did not change significantly, that is, <10%, in cells transfected with the anti-miR-21 compared to the negative control (Figure 2a), suggesting that the reduction in Pdcd4 protein is likely a result of translational inhibition.
miR-21 promotes neoplastic transformation in MCF-7
To determine whether miR-21 promotes transformation, we overexpressed miR-21 in MCF-7 human breast cancer cells. The MCF-7 cell line was selected because it is one of the very few cancer cell lines from the NCI-60 panel that expresses high levels of Pdcd4 (Jansen et al., 2004) (Figure 2b), while other breast cancer cell lines such as MDA-MB-231 express considerable lower levels of this protein (Jansen et al., 2004). Second, the MCF-7 cell line represents an earlier stage of breast cancer compared to other highly malignant cell lines like MDA-MB-231 (Yu et al., 2001; Chen et al., 2004). Third, miR-21 is less abundant in MCF-7 than in MDA MB-231 breast cancer cells (Frankel et al., 2008). The miR-21 level in MCF-7 cells transfected with pSIF-miR-21 was ∼80% higher than that with the vector control (data not shown), while the Pdcd4 protein was decreased about ∼50% (Figure 2b). The PDCD4 mRNA level in MCF-7 cells did not change in the transfected cells (data not shown), indicating translation repression. Soft agar colony formation assays in MCF-7 cells transfected with pSIF-miR-21 revealed ∼100% more colonies than cells transfected with the control vector (Figure 3a). This is the first report to demonstrate that increased expression of miR-21 promotes anchorage-independent transformation in a human breast cancer cell line. A recent report showed that antisense to miR-21 reduced MCF-7 tumor growth as xenografts in nude mice (Si et al., 2007), consistent with the results reported here.
miR-21 promotes neoplastic transformation in murine JB6 cells
The interaction of miR-21 and PDCD4 MRE is highly conserved among animals. The miR-21s from human, chimpanzee, monkey, mouse, rat, chicken and pig are 100% identical and 95% identical to that of zebrafish, puffer fish and cow(Griffiths-Jones et al., 2006). Both human and mouse PDCD4 genes are predicted to be a target of miR-21 by four widely used computational methods, Miranda (John et al., 2004), TargetScan (Lewis et al., 2005), PicTar (Krek et al., 2005) and RNA22 (Miranda et al., 2006). The 3′-UTR sequences of mammalian PDCD4 vary substantially with that of human only 50% identical to mouse and no significant similarity could be found between human and zebrafish. However, the sequences complementary to the miR-21 seed sequence (UAGCUUAU) are fully conserved in the 3′-UTR of PDCD4 genes in human, chimpanzee, rhesus monkey, mouse, rat, dog, cow, zebrafish and puffer fish (only these nine 3′-UTRs are reported). The conservation of both miR-21 and MRE (PDCD4) indicates that gene regulation of PDCD4 by miR-21 is likely conserved in vertebrates, particularly in mammals (Figure 3b), allowing further assessment of the physiological role of miR-21 in a murine neoplastic transformation model.
JB6 cell lines are a well-characterized murine epidermal model of neoplastic transformation (Cmarik et al., 1999). In response to tumor promoters such as 12-O-tetradecanoylphorbol 13-acetate (TPA) JB6 transformation-sensitive (P+) cell lines, such as Cl22, irreversibly undergo neoplastic transformation and acquire a transformed (Tx) phenotype. Cell lines established from anchorage-independent colonies of JB6 such as RT101 are tumorigenic. Promotion-resistant (P−) variants, such as SC21, do not undergo cell transformation and are isolated without selection from the same original population of JB6 epidermal cells. Both the mRNA and protein levels of Pdcd4 are greater in P− than P+ cells (Cmarik et al., 1999). Overexpression of antisense PDCD4 in P− cells reduced the endogenous PDCD4 expression and renders P− cells sensitive to tumor promoter, that is, a P+ phenotype (Cmarik et al., 1999), while overexpression of sense Pdcd4 in P+ cells renders them transformation resistant or P− (Yang et al., 2001). Supplementary Figure S1 shows that the endogenous level of miR-21 in Cl22 (P+) cells is higher than that in SC21 (P−) cells, suggesting that miR-21 may promote neoplastic transformation in JB6 cells.
To address the role of miR-21 regulation of Pdcd4 protein in the JB6 model of transformation, JB6 Cl22 (P+) and SC21 (P−) cells were transfected with the miR-21 expression plasmid or the vector (pSIF) control, resulting ∼2.9- and 1.5-fold higher miR-21 expression compared to control (Supplementary Figure S1) and a >2-fold decrease in Pdcd4 protein in Cl22 (P+) and SC21 (P−) cells (Figure 2b), respectively. Overexpression of miR-21 in Cl22 cells resulted in a slight increase (∼20%) in colony formation in vehicle-treated cells (Figure 2c). Overexpression of miR-21 in SC21 (P−) cells significantly increased colony formation (200%) in vehicle-treated cells. Importantly, in response to TPA promotion, the number of colonies formed by Cl22 cells transfected with miR-21 was 65% greater than vector control (Figure 3c). Furthermore, overexpression of miR-21 increased the size of the colonies formed by Cl22 cells (Figure 3d). Overexpression of miR-21 in SC21 (P−) cells treated with TPA resulted in an even greater (250%) increase in colony number (Figure 3c). The gain of transformed phenotype in response to increased miR-21 expression indicates that miR-21 promotes neoplastic transformation (Figure 3) through downregulating the PDCD4 gene (Figure 2b). This is the first demonstration that overexpression of the miR-21 gene in P− cells shifted them toward the P+ phenotype. We noted that SC21 and Cl22 cells treated with TPA expressed similar amount of miR-21 compared to vehicle control (Supplementary Figure S1), indicating that this chemical tumor promoter does not affect the endogenous miR-21 expression. This result was further supported by a reporter experiment in which the expression of a luciferase gene driven by the promoter for the miR-21 gene (Cai et al., 2004) did not respond to TPA treatment (data not shown). Taken together, our results suggest that miR-21 promotes neoplastic transformation of JB6 epidermal cells either on its own or in combination with chemical tumor promoters.
Overexpression of miR-21 may target many genes in addition to PDCD4 in the CL22 and SC21 cells. To address whether the reduction in Pdcd4 protein that we observed with miR-21 overexpression (Figure 2b) is sufficient to promote cell transformation in this model system, we transfected Cl22 and SC21 cells with five different short hairpin siRNAs (shRNAs) designed by the RNAi consortium (Root et al., 2006) targeting murine PDCD4 (Supplementary Figure S2). shRNA no. 5 consistently reduced PDCD4 expression in both cell lines to a level comparable to that with miR-21 overexpression (Supplementary Figure S2 Figure 2b). Transfection of Cl22 and SC21 cells with siRNA no. 5 against PDCD4 increased colony induction in both cell lines in response to TPA (Supplementary Figure S2). As seen with overexpression of miR-21 (Figure 3c), the increase in anchorage-independent colonies was more significant in the SC21 (P−) cells than in the Cl22 (P+) (Supplementary Figure S2). These data substantiate our results demonstrating that the reduction in PDCD4 by increased miR-21 expression in these cells (Figure 2b) can account for the increased transformation measured (Figure 3).
To further demonstrate that miR-21 promotes colony formation by specifically targeting PDCD4 via the miR-21-binding site in its 3′-UTR, we cloned the wild-type and miR-21 MRE mutant 3′-UTR of PDCD4 downstream of its coding sequence. These constructs were transfected into mouse epidermal JB6 RT101-transformed cells, which lack endogenous Pdcd4 protein expression (Yang et al., 2004), but that have relatively high endogenous miR-21 (Supplementary Figure S1). As seen in Supplementary Figure S3, Pdcd4 protein expression in RT101 cells transfected with the PDCD4 expression vector with the native 3′-UTR was lower than in the cells transfected with the mutant 3′-UTR, while their mRNA levels were about the same (Supplementary Figure S3), suggesting that blocking the miR-21/PDCD4 interaction increases Pdcd4 protein levels. The increase in Pdcd4 expression with the mutant 3′-UTR resulted in decreased RT101 colony formation (Supplementary Figure S3), that is, the mutation rescued the transformation suppressor activity of Pdcd4. These results further support the conclusion that the miR-21-binding site in the 3′-UTR of PDCD4 plays an important role in Pdcd4 translational repression by miR-21 and cell transformation.
Although we initiated this study with eight putative miR-21 target genes that are cancer relevant (Table 1), our results confirmed only PDCD4 as a bona fide miR-21 gene target. Recently, Zhu et al. (2007) reported miR-21 targets TPM1. However, the interaction of miR-21/TPM1 is not conserved. miR-21 is not predicted to target mouse TPM1 (John et al., 2004; Krek et al., 2005; Lewis et al., 2005; Miranda et al., 2006); and the TCGAATG sequence, which is proposed to be crucial to the miR-21/TPM1 (human) interaction (Zhu et al., 2007), is not present in the mouse TPM1 3′-UTR. Another tumor suppressor gene PTEN is reported to be regulated by miR-21 (Meng et al., 2006, 2007); however, PTEN is not a predicted to be an miR-21 target by any of the highly regarded computational methods (John et al., 2004; Krek et al., 2005; Lewis et al., 2005; Miranda et al., 2006). Although inhibition of miR-21 expression increased PTEN expression in malignant cholangiocytes and human hepatocellular carcinoma cells (Meng et al., 2006, 2007), we note that a critical mutational analysis of the putative MRE in the PTEN gene was not included in these studies. Such an analysis is indispensable to establish whether PTEN is a direct or indirect target of miR-21. Overexpression of miR-21 in MCF-7 cells was recently reported to decrease Pdcd4 protein but resulted in only a subtle decrease in PTEN protein levels (Frankel et al., 2008). This report along with the results presented here indicates that miR-21 regulates Pdcd4 protein levels in MCF-7 breast cancer cells. Another new report demonstrated an inverse correlation of miR-21 and Pdcd4 protein in colon cancer and showed that overexpression of miR-21 in Colo206f cells reduced Pdcd4 protein and increased invasion in a chicken embryo membrane assay (Asangani et al., 2007). Taken together, these reports along with the data reported here indicated that PDCD4 is a functional target of miR-21 in various aspects of tumor progression: cell proliferation in breast cancer (Frankel et al., 2008), invasion/intravasation/metastasis in colon cancer(Asangani et al., 2007), and, as demonstrated for the first time here, neoplastic transformation, a model of tumor induction.
In summary, this study demonstrates that the translation of the tumor suppressor gene PDCD4 is negatively regulated by the miRNA miR-21 in HEK-293T, MCF-7 and JB6 cell lines, and provides evidence that the mir-21 gene functions as an oncogene to promote cell transformation. Reduced PDCD4 expression has been reported in at least six human tumor types or cancer cell lines (lung, brain, renal, breast, colon and pancreas) (Chen et al., 2003; Jansen et al., 2004; Ma et al., 2005; Lee et al., 2006), out of the nine solid tumors in which miR-21 is overexpressed (Chan et al., 2005; Iorio et al., 2005; Diederichs and Haber, 2006; Roldo et al., 2006; Volinia et al., 2006), indicating that mir-21/PDCD4 is likely to be a clinically significant oncogene:tumor suppressor pair in the induction and progression of human carcinomas.
YL is supported by the career development program and a pilot grant from the Center for Genomics and Integrated Biology at University of Louisville funded by NIEHS P30ES014443. This research was supported in part by NCI R21-CA124811 to CMK. ML is supported by PUJIANG program from the Committee of Shanghai Science and Technology (06PJ14105).