miR-497 and miR-34a retard lung cancer growth by co-inhibiting cyclin E1 (CCNE1)

Cyclin E1, encoded by the CCNE1 gene, promotes G1/S transition, chromosome instability, and oncogenesis. Here, we show that miR-497 and miR-34a target the 3′-UTR of CCNE1. miR-497 and miR-34a are downregulated in cancer cells and their ectopic expression inhibited cell proliferation and colony formation in vitro, and inhibited tumor growth in a xenograft model. The effect of simultaneous overexpression of miR-497 and miR-34a on the inhibition of cell proliferation, colony formation, and tumor growth, and the downregulation of cyclin E1 was stronger than the effect of each miRNA alone. The synergistic actions of miR-497 and miR-34a partly correlated with cyclin E1 levels. When cells stably expressing CCNE1 were transfected with the Hi-miR-497/34a plasmid, there was no effect on colony formation, compared with that of cells transfected with either Hi-miR497 or Hi-miR34a. These results indicate cyclin E1 is downregulated by both miR-497 and miR-34a, which synergistically retard the growth of human lung cancer cells.


IntroductIon
Cancer is a complex disease caused by the progressive accumulation of genetic and epigenetic alterations in cells, which allow the cells to evade normal and environmental controls. Much progress has been made in the treatment of lung cancer in the last 10 years. Nonetheless, lung cancer is currently the most common cause of cancer-related death throughout the world, and therefore remains an unresolved medical issue. An understanding of the processes and pathogenesis of cancer at the systemic, cellular, and molecular levels is one of the most ambitious goals of cancer research. MicroRNAs (miRNAs) are epigenetic regulators that play a pivotal role in the acquisition of tumorigenic properties by cells.
miRNAs are a class of endogenous noncoding RNAs of approximately 22 nucleotides (nt) that regulate mRNA stability and translation [1][2][3]. A wide range of biological functions are controlled by miRNAs, including cell proliferation, differentiation, and apoptosis [4][5][6]. There is strong evidence that miRNAs can act as oncogenes or tumor suppressors, with key roles in cancer initiation, progression, and therapy. In an attempt to understand the mechanisms underlying cancer, an increasing number of studies have reported that individual miRNAs exert their functions in specific cancers. Many recent studies reported that some miRNAs cooperatively control a variety of biological processes, including cell development and differentiation, apoptosis, and the cell cycle [7][8][9][10]. When an mRNA or several different mRNAs involved in a specific biological process are targeted by several miRNAs, the miRNAs act cooperatively. Recent studies have demonstrated that miRNAs encoded in miRNA clusters function synergistically in cancer, e.g., the miR-17-92 cluster [11][12][13] and the miR-15a-16-1 cluster [14,15]. Ventura et al. reported that miR-17-92 and miR-106b-25 double-knockout mice have a more severe phenotype than miR-17-92 single-knockout mice [16]. www.impactjournals.com/oncotarget In our previous study [17], we profiled miRNA expression in lung cancer cells treated with resveratrol, a potential cancer chemopreventive agent. We found nine miRNAs (miR-622, -512-5p, -504, -497, -34a, -302d, -302b*, -29c, -20b) upregulated in the cells. Only miR-622, -497, and -34a inhibited the proliferation of lung cancer cells, whereas the remaining six miRNAs had no effect on cell proliferation in our preliminary experiments. miR-497 and -34a have a similar seed sequence, indicating that they might share similar functions; therefore, in the present study, these miRNAs were selected for the identification of potential targets and to explore their roles in lung cancer. Bioinformatics analysis suggested that both miRNAs target the CCNE1 gene, which encodes Cyclin E1. Cyclin E1, a member of the conserved cyclin E family, activates CDK2 [18] and regulates the transition of mammalian cells from quiescence to S phase. Transgenic cyclin E1 triggers dysplasia and multiple pulmonary adenocarcinomas [19,20], and the overexpression of cyclin E1 was suggested to contribute to cancer development or tumorigenesis in various types of cancer, including breast, colon, and lung cancers [21]. Increased expression of cyclin E1 is a useful marker of poor prognosis in lung cancer [22]. These data suggest that cyclin E1 is a potential target for the treatment of lung cancer. Based on previous data, we designed a study to test the hypothesis that cyclin E1 expression is coregulated by miR-497 and miR-34a in lung cancer. results mir-497 and mir-34a inhibit the proliferation of human lung cancer cells miR-34a is downregulated in lung cancer tissues and cells [23,24]; however, few reports have examined the expression of miR-497 in lung cancer. Although miR-497 is downregulated in lung cancer [25,26], its specific role remains to be determined. Analysis of the expression of miR-497 and miR-34a in lung cancer cells showed that the levels of miR-497 and miR-34a ( Figure 1a)  The cell viability of A549, H460, and H1299 lung cancer cells was decreased by 66.71 ± 1.65% , 46.36 ± 1.96% and 72.10 ± 4.02 %, respectively, in response to miR-34a overexpression, and by 60.71 ± 4.63% , 74.94 ± 3.58%, and 73.71 ± 6.50%, respectively, in response to miR-497 overexpression (Figure 1b). Downregulation of the expression of miR-34a ( Figure S1a) or miR-497 ( Figure S1b) with inhibitors had no effect on the growth of A549, H460, and H1299 cells ( Figure S1c) because the endogenous levels of these miRNAs in these cells are low. To identify the phase of the cell cycle at which the miRNAs exert their proliferation-inhibitory effect, cell-cycle distribution was analyzed by flow cytometry. Transfection with miR-497 or miR-34a mimics caused cell-cycle arrest at G 0 /G 1 phase in A549, H1299, and H460 lung cancer cells (Figure 1c). Typical histograms of the cell-cycle arrest induced by miR-497 or miR-34a in A549 cells are shown in Figure 1d.

mir-497 and mir-34a suppress colony formation and tumorigenesis
The effect of miR-497 and miR-34a on the colony forming ability of A549 cells was assessed. Cells transfected with miR-497 or miR-34a mimics showed fewer (31.33 ± 2.44 and 21.00 ± 4.00 colonies per well, respectively) and smaller colonies than those observed in the control groups (71.00 ± 9.33 colonies per well) ( Figure  2a).
The effect of miR-497 or miR-34a on tumorigenicity was examined in vivo. The bilateral inguino-abdominal flanks of nude mice were inoculated subcutaneously (s.c.) with A549 cells transfected with normal control (NC) (left flank) or miR-497 (right flank) mimics or with NC (left flank) or miR-34a (right flank) mimics. At 5 weeks, the average volumes (588.39 ± 117.34 mm 3 ) and weights (308.57 ± 26.53 mg) of tumors in groups of seven mice injected with cells transfected with miR-497 mimics were lower than those (1293.16 ± 198.57 mm 3 and 427.14 ± 365.31 mg) in the control group injected with the NC mimic (Figure 2b), while the average volumes (190.25 ± 67.79 mm 3 ) and weights (72.86 ± 31.84 mg) of tumors in mice injected with cells transfected with miR-34a mimics were lower than those (913.14 ± 455.23 mm 3 and 287.20 ± 131.09 mg) in the control group injected with the NC mimic ( Figure 2c). This evidence collectively suggests that miR-497 and miR-34a inhibit cell growth and might function as tumor suppressors.

CCNE1 is a direct target of mir-497 and mir-34a
To confirm that CCNE1 is targeted by miR-497 and miR-34a, we investigated the effects of miR-497 and miR-34a on cyclin E1 levels by immunoblotting. Overexpression of miR-497 or miR-34a in A549, H460, and H1299 lung cancer cells by transfection with miR-497 or miR-34a mimics ( Figure S3a and S3b) markedly reduced the levels of cyclin E1 protein ( Figure 3a). However, real-time quantitative polymerase chain reaction (real-time qPCR) showed no changes in CCNE1 mRNA levels in response to miR-497 or miR-34a upregulation ( Figure S3c). This indicates that the negative correlation between cyclin E1 expression and miR-497 or miR-34a levels is due to post-transcriptional modulation of cyclin E1 expression.
The wild-type CCNE1 3′-UTR and a mutant CCNE1 3′-UTR were subcloned downstream from the Renilla luciferase gene in the psiCHECK2 vector to determine whether they bound directly to seed regions of the miRNAs. In Figure 3b, the wild-type sequence (α, wt-CCNE), mutant sequences at nt 247-253 (β, mt-CCNEβ) or nt 485-491 (γ, mt-CCNEγ), and mutant sequences at both binding sites (δ, mt-CCNEδ) in the CCNE1 3′-UTR were inserted individually into the vector, generating four plasmids as follows: wt-CCNE, mt-CCNEβ, mt-CCNEγ, and mt-CCNEδ. The binding sites between miR-34a and the wild-type or mutant CCNE1 were examined ( Figure  3c). When the wt-CCNE plasmid was cotransfected with the miR-34a mimic, the luciferase activity of A549 cells was markedly reduced by 34.56 ± 1.13% (Figure 3d). When the mt-CCNEβ plasmid was cotransfected with the miR-34a mimic or an inhibitor, the luciferase activity did not differ from that of the control. These data confirm that miR-34a directly targets CCNE1. To examine the interaction between miR-497 and CCNE1, the binding sites in CCNE1 were completely mutated, or mutated individually at each nucleotide ( Figure 3e). When the wt-CCNE plasmid was cotransfected with the miR-497 mimic, the luciferase activity was markedly reduced by 40.09 ± 1.99% (Figure 3f). Moreover, when either plasmid (mt-CCNEβ or mt-CCNEγ) with only one mutant site was cotransfected with the miR-497 mimic, the luciferase activity was slightly reduced, whereas cotransfection of the mt-CCNEδ plasmid, which contained two completely mutated binding sites, with the miR-497 mimic reversed the miR-497 induced reduction of luciferase activity (Figure 3f). However, because the endogenous levels of miR-34a and miR-497 are low in A549 cells, transfection with miR-497 and miR-34a inhibitors did not affect the levels of luciferase activity (Figures 3d and 1f). cyclin e1 mediates the effects of mir-497 or mir-34a on cell growth Because CCNE1 was identified as a direct target of miR-497 and miR-34a, we investigated whether the effects of miR-497 and miR-34a on cell proliferation were mediated by the modulation of CCNE1 expression. Cell proliferation and cell-cycle distribution were examined in A549 cells after CCNE1 knockdown and overexpression of the miRNAs. The efficiency of RNA interference was confirmed by real-time qPCR (Figure 4a, upper panel) and immunoblotting (Figure 4a, lower panel). Knockdown of CCNE1 significantly inhibited cell viability and G 0 / G 1 arrest in A549 cells, whereas miR-497 or miR-34a overexpression did not enhance the effect of CCNE1 knockdown (Figure 4b and 4c). A typical histogram of the cell cycle is shown in Figure 4c. These results support the assumption that the effects of miR-497 and miR-34a on cell growth are mediated by its modulation of CCNE1 expression.
Oncotarget 13157 www.impactjournals.com/oncotarget coexpression of miR-497 and miR-34a enhanced the effect of each individual miRNA on the modulation of cyclin E1 expression. A reporter gene activity assay was used to test this hypothesis in A549 cells. Cotransfection of the wt-CCNE plasmid with Hi-miR497/34a reduced luciferase activity to a greater extent than cotransfection of wt-CCNE with Hi-miR497 or Hi-miR34a alone (Figure 6b). However, when the mt-CCNEδ plasmid was cotransfected with Hi-miR497/34a, the luciferase activity did not differ from that of cells cotransfected with either Hi-miR497 or Hi-miR34a alone (Figure 6b). This suggests that the synergistic effects of miR-497 and miR-34a are correlated with the levels of cyclin E1. Transfection of Hi-CCNE1a cells with Hi-miR-497/34a did not affect colony formation ( Figure 6c) compared with that of cells transfected with either Hi-miR497 or Hi-miR34a, as shown in Figure 6d. Therefore, overexpression of CCNE1 abolished the growth retardation induced by miR497 and miR34a in A549 cells, and CCNE1 mediates the synergistic effects of miR-497 and miR-34a.

dIscussIon
The major finding of our study is that miR-497 and miR-34a synergistically inhibit the expression of the same gene, CCNE1, and the function of its encoded protein, cyclin E1, thereby impeding the growth of lung cancer cells. Several studies have demonstrated that miR-34a retards lung cancer cell growth or induces apoptosis by targeting TGFβR2 [23], Axl [27], Notch-1 [28], or HDM4 [29], whereas miR-497 does so by targeting HDGF [26] in lung cancer. Our study extends the results of others by showing that miR-34a and miR-497 cotarget CCNE1 in lung cancer cells. To the best of our knowledge, this is the first study to demonstrate the cooperative effect of miR-497 and miR-34a, which belong to different miRNA families, on inhibiting cancer cell growth.
In our previous study [17], we reported that an antitumor agent, resveratrol, upregulates the expression of several miRNAs (including miR-497 and miR-34a), and thus might inhibit the proliferation of lung cancer Oncotarget 13158 www.impactjournals.com/oncotarget cells. This prompted us to investigate whether these two miRNAs function as antitumor agents. Some miRNAs that are downregulated in cancer are found in regions with frequent loss of heterozygosity [30]. miRNA-497 and miRNA-34a are located on chromosomes 17p13.1 and 1p36, respectively. The loss or deletion of chromosome 17p13.1 or 1p36 has been reported in various types of cancer, including lung cancer [31,32], suggesting that the downregulation of miRNA-497 or miRNA-34a in these cancers arises from genomic DNA loss or deletion. miR-34a, located at 1p36, has been extensively studied as a microtumor suppressor [33,34], and miR-497, located at 17p13.1, is deleted in 93% of small-cell lung cancers [35]. Son et al. reported that miR-497 is downregulated in lung cancer [25]; however, its role in lung cancer remains unclear. Here, we studied the function of miR-497 in lung cancer cells, and showed that miR-497 and miR-34a inhibit cell growth in vitro and in vivo, supporting our hypothesis that the two miRNAs function as tumor suppressors. Several studies have reported that miR-497 and miR-34a are potential anticancer agents based on their ability to target oncogenes [24,25,36,37]. In the present study, we found that they cotarget CCNE1, repressing the expression of cyclin E1.
Cyclin E1 is an essential regulator of cell cycle G 1 progression and entry into S phase. It is a nuclear protein that was first identified by its ability to complement the proliferative defects of cyclin-deficient yeast cells. It is overexpressed in various cancers, including lung, breast, and gastrointestinal tract tumors [38][39][40]. Several mechanisms regulate its expression in tumors. Although gene amplification might be involved, the expression of cyclin E1 is also modulated post-transcriptionally by miRNAs. Cyclin E1 (or cyclin E) is repressed by miR-15b, miR-16, miR-34c, and miR-145 [41,42] in other cancer cells. In the present study, we showed for the first time that both miR-34a and miR-497 bind to the 3′-UTR region of CCNE1, as confirmed by immunoblotting and luciferase activity assays. Previous studies showed that miR-34a targets the oncogenes CDK4, CDK6, CCND1, MET, and BCL2 [36,43,44], whereas miR-497 targets the oncogenes CCND2 and BCL2 [37,45] in various types of cancer. These and our results, together with the strong possibility that more miR-497 or miR-34a targets will be discovered soon, suggest that these two miRNAs regulate proliferation-related mRNAs (including CCNE1), and thus function as tumor suppressors.
The protein encoded by CCNE1, a newly identified cotarget for miR-34a and miR-497, plays a key role in regulating the growth of lung cancer cells, and its effect is mediated by the cooperative action of the two miRNAs. Several studies have focused on the effects of the miRNA family axis on cancer because many miRNAs contain the same 7-nt or 8-nt seed sequence. Most single miRNA mutations do not markedly affect the corresponding phenotype, whereas knocking out multiple or all miRNA family members can affect several phenotypes in insects [46,47]. This phenomenon, which reflects the complexity of miRNAs and especially the miRNA-miRNA network, allows them to perform many biological functions. Here, we found that two miRNAs from different families synergistically inhibit the growth of lung cancer cells, which indicates that the two miRNA families share similar biological functions. Biochemical and bioinformatics studies showed that the 5′ ends of miRNAs, designated as "seed sequences", play critical roles in target recognition and posttranscriptional repression [3,48,49]. The recognition is based on the complementary base paring between the seed sequence and the target mRNA. Studies have shown that G:U base pairing is tolerated in seed complementarity [3,50]. Therefore, miR-497 and miR-34a, with seed sequences 5′-AGCAGCA-3′ and 5′-GGCAGUG-3′, respectively, can bind to the same sequence (3′-UCGUCGU-5′) in the 3′ UTR of CCNE1; however, there are three-base variations between their seed regions. This was verified in our luciferase activity assay. Our results indicated that miR-497 and miR-34a synergistically inhibit cell proliferation, predominantly by repressing the expression of their cotarget, CCNE1. However, the relationship between miR-497 and miR-34a is complex and requires further research.
Activation of cyclin E:Cdk2 results in retinoblastoma tumor-suppressor gene (RB) inactivation by hyper-phosphorylation, induction of E2F target gene transcription, and cell cycle G1/S transition [18]. Cell cycle dysregulation (including the overexpression of cyclins) occurs frequently in neoplasia or cancer. Multiple reports suggested that cyclin E is a promising therapeutic target in lung [20,51], ovarian [52], and breast cancers [53,54]. Li et al [53] showed that cyclin E siRNA delivered by intratumoral injection effectively inhibits cyclin E expression in vivo and results in tumor suppression. However, there are limitations to the therapeutic application of siRNAs, such as their stability, off-target side effects, and interferon responses to foreign nucleic acids. Unlike siRNAs targeting a single molecule, the miRNA approach might have a greater potential as cancer therapeutics because of the endogenous nature of miRNAs and their capacity to simultaneously regulate many different oncogenes across multiple pathways. Furthermore, evidence indicates that a combination of two tumor-suppressive miRNAs is superior to single miRNAs for the repression of oncogene expression, the inhibition of proliferation and invasion of cancer cells in culture, and the inhibition of tumor proliferation in vivo. Our data indicated that the tumor-suppressive miR-34a and miR-497 could co-inhibit cyclin E1, suggesting that cyclin E1 is a key target mediating the anti-tumor effect of miR-34a and miR-497 in lung cancer. Both miRNAs modulate other oncogenic genes that may mediate their tumor suppressor function. Chek1 (putative target of miR-497 identified by the algorithm DIANA miRPath v.2.0), cdc25a (miRwww.impactjournals.com/oncotarget 497) [55], and cdk6 (miR-497and miR-34a) [55,56] are involved in the indirect or direct regulation of the cyclin E1 downstream genes cdk2, RB and E2f3 (target for miR-34a, [33,57]). Therefore, the combination of miR-34a and miR-497 could be superior to each individual miRNA in its ability to retard lung cancer cell growth to some extent.
In conclusion, we showed that miR-497 and miR-34a act cooperatively to regulate certain aspects of tumorigenesis, including the growth of lung cancer cell lines, especially through their cooperative effect on the downregulation of cyclin E1 expression.

MAterIAls And Methods cell culture and reagents
The lung cancer cell lines A549, H1299, H460, H446, and QG56, purchased from the Chinese Academy of Sciences Cell Bank of Type Culture Collection (Shanghai, China), were cultured in RPMI-1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; Sijiqing Co., Ltd, Hangzhou, China). Normal human bronchial epithelial cells (16HBE), kindly provided by Dr. Xujun from the Guangzhou Institute of Respiratory Diseases, were maintained in minimum essential medium supplemented with 10% FBS (Sijiqing Co., Ltd). All cells were incubated at 37°C in a humidified atmosphere of 5% CO 2 . Antibody directed against cyclin E1 (Santa Cruz Biotechnology, Santa Cruz, CA), Alexa-Fluor-680conjugated goat anti-mouse IgG antibody (Molecular Probes, Invitrogen), and IRDye 800-conjugated anti-rabbit IgG antibody (Li-Cor, Lincoln, NE) were purchased for immunoblotting.

Immunoblotting analysis
Total cell lysates were prepared with cell lysis buffer (CST, Beverly, MA) containing proteinase inhibitors (1% cocktail and 1 mmol/L PMSF, both from Kangcheng, China). Cell proteins (50 μg) were separated by 12% SDS-PAGE and transferred to PVDF membranes (Millipore, Bedford, MA). The membranes were blocked, incubated with anti-cyclin E1 or anti-glyceraldehyde phosphate dehydrogenase (GAPDH) antibody, and then with secondary antibodies. Immunoblotting analyses were performed with the Odyssey Infrared Imaging System (Li-Cor).

luciferase assays
The sequences of the CCNE1 3′-UTR and the CCNE1 3-UTRs in which the putative binding sites had been mutated were amplified with specific primers (listed in Table S1), and verified by DNA sequencing. These gene fragments were then subcloned downstream from the Renilla luciferase gene in the psiCHECK2 vector (Promega, Madison, WI) to generate the wildtype CCNE1 plasmid (wt-CCNE), the partial mutant CCNE1 plasmids (mt-CCNEβ and mt-CCNEγ), and the complete mutant CCNE1 plasmid (mt-CCNEδ). For luciferase assays, A549 cells were transfected with wt-CCNE, mt-CCNEβ, mt-CCNEγ, or mt-CCNEδ in 24well plates using Lipofectamine 2000 (Invitrogen). The transfection mixtures contained 100 ng of plasmid and 50 nM synthetic mimic, inhibitor, Hi-miR497, Hi-miR34a, or Hi-miR497/34a. A549 cells were also transfected with the psiCHECK-2 vector as the normalization control.

soft-agar colony formation
Transfected cells were propagated on soft agar. A base layer (1.5 mL) of agar (Amresco, Solon, OH) (0.6% agar in RPMI-1640 with 10% FBS) was allowed to solidify in a six-well flat-bottomed plate before the addition of 2 mL of a cell suspension containing 2000 cells in 0.3% agar in RPMI-1640 with 10% FBS. The colonies were allowed to grow for 28 days at 37°C under 5% CO 2 and visualized with an inverted microscope (Olympus IX71, Olympus, Tokyo, Japan).

tumor xenograft model
Five-week-old male nude mice (BALB/c nu/ nu), purchased from the Medical Animal Experimental Center of Guangdong Province, were used to examine the tumorigenicity of the transfected cells. The animal protocol was approved by the Animal Care and Use Committee. The transfected cells were propagated and 5 × 10 6 cells were inoculated s.c. into the dorsal flanks of the mice. Tumor size was measured weekly and tumor volume was estimated as described in our previous report [17]. Tumors were removed and weighed 5 weeks after the injection of tumor cells.

expression vector construction
The sequences of the miR-497 precursor (MIR497) and the miR-34a precursor (MIR34A) were amplified with oligonucleotide primers synthesized by Invitrogen ( Figure S4). MIR497 and MIR34A were then inserted individually into the pcDNA™6.2-GW/EmGFPmiR vector (Invitrogen) using the BLOCK-iT™ Pol II miR RNAi Expression Vector Kit (Invitrogen), to generate the expression constructs Hi-miR497 and Hi-miR34a, respectively. These were verified by DNA sequencing with the primer 5′-CTCTAGATCAACCACTTTGT-3′ in an ABI Prism 373 Genetic Analyzer (Applied Biosystems). The empty pcDNA™6.2-GW/EmGFPmiR vector was used as the negative control (mock). To investigate the cooperative activities of miR-497 and miR-34a, we used the isocaudomers BamHI (NEB, Beverly, MA) and BglII (NEB) to ligate both MIR497 and MIR34A into the vector to generate the expression construct Hi-miR497/34a, which was confirmed by DNA sequencing.
A clone of the green fluorescent protein (GFP)tagged cDNA [including complete 3′-UTR encoding human CCNE1 (NM_001238.2)], was bought from Origene (Rockville, MD) as transfection-ready plasmid DNA. The neomycin-resistance gene is expressed downstream from the SV40 promoter in the same vector, which permits the positive selection of transfected cells. To construct cells stably expressing CCNE1, A549 cells were transfected with the plasmid DNA and selected with neomycin (0.5 mg/mL).

statistical analysis
All assays were performed in triplicate in three independent experiments, and all data are expressed as the mean ± SD. Statistical differences between two groups were evaluated with the two-tailed Student's t test and analysis of variance (ANOVA) was used for multiple comparisons. In all cases, P < 0.05 was considered significant. All statistical tests were performed with the statistical analysis software IBM SPSS Statistics 20 for Mac (International Business Machines Corporation, New York, NY).