Knockdown of miR-21 in human breast cancer cell lines inhibits proliferation, in vitro migration and in vivotumor growth

In situ hybridization analysis was performed on fresh samples from BC or fibroadenoma (FA) tissues with paired normal adjacent tissues (NATs, > 2 cm from tumor tissues) obtained from Sun Yat-sen University Cancer Center (SYSUCC) (Guangzhou, China) between January and March 2009. For IHC staining of miR-21 predicted target genes, formalin-fixed paraffin-embedded tissues were obtained from 99 randomly selected BC patients without neoadjuvant therapy at SYSUCC from January 2000 to November 2004, from whom informed consent and agreement, and clinicopathological information was available. A pathologist reviewed slides from all blocks, selecting representative areas of invasive tumor tissue to be cored. Selected cores were analyzed in duplicate using a MiniCore Tissue Arrayer (Alphelys, Passage Paul Langevin, Plaisir, France) with a 1-mm needle. The diagnosis and histological grade of each case were independently confirmed by two pathologists based on World Health Organization classification [12]. The clinical stage was classified according to the American Joint Committee on Cancer (AJCC) tumor-lymph node-metastasis (TNM) classification system [13]. The study was approved by the Research Ethics Committee of SYSUCC (Reference number: YP-2009168). The clinicopathological characteristics and follow-up data of the patients are summarized in Table 1.

To study the spatial and temporal expression of miRNAs with high sensitivity and resolution, the miRNA chromogenic in situ hybridization (CISH) and fluorescein in situ hybridization (FISH) protocol [14] were optimized (Additional file 1).

MCF-7 and MDA-MB-231 cells were maintained in Dulbecco's modified Eagle's medium, supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin and 10% fetal bovine serum (GIBCO-Invitrogen, Carlsbad, CA, USA). For transfection, the LNA-antimiR-21 or LNA-control (Exiqon A/S, Skelstedet, Vedbaek, Denmark) were delivered at a final concentration of 50 nM using Lipofectamine 2000 reagent (Invitrogen).

Growing cells (2 × 10³ cells per well) were seeded into 96-well plates. At 24 h after LNA-transfection, cells were stained with 20 μl sterile MTT dye (5 mg/ml; Sigma-Aldrich Corp, St. Louis, MO, USA), followed by 4 h at 37°C. After supernatant removal, 150 μl of dimethyl sulphoxide (Sigma) was added and thoroughly mixed for 15 minutes. Absorbence was measured with a microplate reader (SpectraMax M5; Molecular Devices Corp., Silicon Valley, CA, USA) at 490 nm. For colony formation assays, cells were seeded in six-well plates (0.5 × 10³ cells per well) and cultured for two weeks. Colonies were fixed with methanol for 10 minutes and stained with 1% crystal violet (Sigma) for 1 minute. Each cell group was measured in triplicate.

Cells cultured in the presence of 50 nM LNA-antimiR-21 or LNA-control for 24 h were allowed to reach confluence before dragging a 1-mL sterile pipette tip (Axygen Scientific, Inc., Union City, CA, USA) through the monolayer. Cells were washed to remove cellular debris and allowed to migrate for 24 h or 48 h. Images were taken at time 0 h, 24 h and 48 h post-wounding using a digital camera system (Leica DFC 480; Leica Microsystems, Bannockburn, IL, USA). The motility of the cells was determined as repaired area percentage [15]. Each cell group was measured in triplicate.

Five- to six-week-old female BALB/c-nude mice (Slaccas Shanghai Laboratory Animal Co., Ltd., Shanghai, China) were used for experimental tumorigenicity assays. To facilitate estrogen-dependent xenograft establishment, each mouse received 17-estradiol (20 mg/kg; Sigma) intraperitoneally once a week. One week after treatment, equivalent amounts of MCF-7 cells, treated with PNA-antimiR-21 or PNA-control (100 nM for 48 h; Panagene, Inc., Yuseong-gu, Daejeon, Korea) without transfection reagents according to the manufacturer's protocol, were injected subcutaneously (10⁷ cells/tumor) into the left axilla of nude mice [16]. Mice were weighed, and tumor width (W) and length (L) were measured every day. Tumor volume was estimated according to the standard formula: V = ∏/6 × L × W², as described previously [17]. Animals were killed nine days after initial growth of the MCF-7 xenografts was detectable, and tumors were extracted. In all experiments, the ethics guidelines for investigations in conscious animals were followed, with approval from the local Ethics Committee for Animal Research.

MCF-7 and MDA-MB-231 cells were transfected either with LNA-antimiR-21 or with LNA-control at a final concentration of 50 nM. Total RNAs were isolated from MCF-7 cells 48 h post transfection and from MDA-MB-231 cells 36 h post transfection, respectively, using Trizol Reagent (Invitrogen). The mRNA expression profile was performed using human genome oligo array service V1.0 (Catalog Number 400010; CapitalBio, Beijing, China) as described [18]. Each sample was analyzed once, and the CapitalBio data preprocess, normalization and filtering were as previously described [18]. Ratios were defined as marginal signal intensity when there was a substantial amount of variation in the signal intensity within the pixels from 800 to 1,500. All the microarray data have been deposited to the Gene Expression Omnibus (GEO) [19] and are accessible through GEO Series accession number [GEO: GSE20627].

For validation of mRNA array and quantitative analysis of miR-21 as well as potential target genes, qRT-PCR was used as previously described [20]. The primers for qRT-PCR are in Additional file 2. The relative expression was calculated using the equation relative quantification (RQ) = 2^-ΔΔ^CT [21].

Predicted miR-21 targets were identified using the algorithms of TargetScan 5.1 [22], miRBase Targets V5 [23], miRNAMap 2.0 [24], PicTar [25] and miRanda 3.0 [26].

The 3' untranslated region (3' UTR) of mRNA sequence of ANKRD46 containing predicted miR-21 binding site was amplified by PCR. PCR primers were listed in Additional file 2. After amplification, PCR products were cloned into the pMIR-REPORT (Applied Biosystems, Foster City, CA, USA), resulting in the pMIR-REPORT-3'ANKRD46. Mutation of ANKRD46 was introduced in the predicted miR-21 binding site by a QuikChange site-directed mutagenesis kit (Stratagene, Foster City, CA, USA). Wild-type EIF4A2 and mutant EIF4A2 were cloned into pMD19-T Simple Vector by TaKaRa Biotechnology CO., LTD. (Dalian, Liaoning, China) and then were individually subcloned downstream of the luciferase coding sequence in the pMIR-REPORT (Applied Biosystems). All constructs were verified by DNA sequencing.

Cells were harvested and lysed in radioimmune precipitation buffer (Upstate, Lake Placid, NY, USA) at the indicted time post-transfection. Antibodies used for immunoblot analysis were against ANKRD46 (1:500 dilution; sc-87548, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), EIF4A2 (1:1000 dilution; ab31218, Abcam, Cambridge, UK) and GAPDH (1:3,000 dilution; sc-32233, Santa Cruz Biotechnology), as a loading control. All bands were detected using a SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA).

IHC and scoring of the estrogen receptor (ER), progesterone receptor (PR) and CerbB2 were performed as previously described [20]. Slides were incubated with primary antibodies against ANKRD46 (1:150 dilution; sc-87548, Santa Cruz Biotechnology); or EIF4A2 (1:700 dilution; ab31218, Abcam). All slides were processed simultaneously in identical conditions per the manufacturer's instructions. Three observers independently determined consensus scoring of EIF4A2 and ANKRD46 immunostaining using a semi-quantitative estimation according to the percentage of positive cells and the intensity of staining as described previously [27]. With these data, the composite score was obtained by adding the values of the staining intensity and relative abundance [28]. Samples with scores lower than the median score were grouped as low protein expression [29].

Spearman's rank correlation test was used for correlation analysis between predicted target gene protein levels and endogenous miR-21 levels measured previously by qRT-PCR [20]. Pearson's Chi-Square tests were used to compare target gene expression levels to clinicopathological characteristics. Survival curves were estimated by the Kaplan-Meier method and log-rank test. All analysis used SPSS 16.0 for Windows (SPSS Inc, Chicago, IL, USA). All tests were two-tailed, and the significance level was set at P < 0.05.

Expression of miR-21 was detected in the cytoplasm in cancerous and luminal epithelial cells, and occasionally in fibroblasts. In BC patients, an increase in miR-21 staining intensity was observed in BC and FA tissues compared with corresponding NATs (Figure 1a). Parallel detection by FISH is shown in Additional file 3. Consistent with the CISH results, quantitative analysis indicated that miR-21 expression was significantly increased by 4.44- to 2.02-fold in BC tissues compared with NATs (P = 0.019, n = 4), and increased in FA tissues by 3.03- to 1.89-fold (P = 0.008, n = 4, Figure 1b). FISH and qRT-PCR were used to measure miR-21 levels in five BC cell lines (MCF-7, MDA-MB-231, MDA-MB-453, MDA-MB-435, and SK-BR-3) and one non-tumorigenic epithelial cell line (MCF-10A). Consistent with previous findings [8, 10, 30], miR-21 overexpressed in MCF-7, MDA-MB-231 and MDA-MB-453 cell lines from 6.48- to 4.04-fold compared with MCF-10A cell line (P < 0.01, Figure 1d).

MCF-7 and MDA-MB-231 cell lines were selected to investigate miR-21 functions and targets by using sequence-specific functional inhibition of miR-21, because both cell lines express higher levels of miR-21 compared with MCF-10A cells. Optimal doses and time points for transfection of LNA reagents were determined by evaluating miR-21 levels using qRT-PCR (Additional file 4). Knockdown of miR-21 reduced miR-21 levels by 98% in MCF-7 cells, and 77% in MDA-MB-231 cells (P < 0.01) (Figure 2a). LNA-antimiR-21 led to a decrease in MCF-7 cell growth (Figure 2b) and proliferation (29%, P = 0.003, Figure 2c). Similar inhibition of cell growth and proliferation effects (51%, P = 0.011, Figure 2c) was also observed in MDA/LNA-antimiR-21 cells (data not shown). In vitro wound healing assays showed that wound repair in MCF/LNA-antimiR-21 and MDA/LNA-antimiR-21 was delayed compared with MCF/LNA-control and MDA/LNA-control cells (data not shown). Knockdown of miR-21 suppressed MCF-7 cell migration by up to 69% (P = 0.013), and MDA-MB-231 migration by 51% (P = 0.001), compared with the LNA-control at 24 h after wound scratch (Figure 2d). These data demonstrate the tumorigenic properties of miR-21 in regulating cell growth, proliferation and migration.

To address the potential effects of PNA-antimiR-21 in vivo on the growth of BC cells, equal numbers (3 × 10⁷) of MCF-7 cells treated with PNA-antimiR-21 or the PNA-control were subcutaneously injected into female nude mice (eight animals per treatment). As seen in Figure 3b and 3c, detectable tumor masses (0.011 ± 0.013 g, mean ± standard deviation, SD) were seen in only 5/8 (62.5%) of mice in the MCF/PNA-antimiR-21 group, while much larger tumors (0.036 ± 0.038 g, mean ± SD) were detected in all mice in the MCF/PNA-control group (P = 0.065, Mann-Whitney test). Both tumor weight and number showed that MCF/PNA-control cells formed larger tumors more rapidly (Figure 3a, c) than MCF/PNA-antimiR-21 cells in nude mice. PNA-antimiR-21 reduced miR-21 expression by 5.72 log₂-scale in MCF-7 cells 48 h post-treatment compared with that in the control (P < 0.01). miR-21 expression in xenograft tumors of PNA-antimiR-21 group was 0.96 log₂-scale higher than that of the control group (P < 0.05; Figure 3e). Notably, MCF/PNA-antimiR-21 tumor cells were decreased in mitotic and pathological mitotic stages compared with MCF/PNA-control cells, indicating a decreased in cell proliferative activity and apoptosis (Figure 3d).

It is known that animal miRNAs regulate gene expression by inhibiting translation and/or by inducing degradation of target. In our study, most modulated genes in the mRNA differential expression profiles changed by less than two-fold. Since mRNA levels regulated by less than two-fold may still be miRNA targets, we defined differentially expressed genes as no less than 1.3-fold change [31]. Comparative analysis of LNA-antimiR-21 and LNA-control mRNA profiles showed differential regulation of 394 genes in MCF/LNA-antimiR-21, of which 228 (58%) were up-regulated and 166 (42%) were down-regulated. 321 genes were differentially expressed in MDA/LNA-antimiR-21 cells, of which 190 (59%) were up-regulated, and 131 (41%) down-regulated (Figure 4a). The intersection of MCF/LNA-antimiR-21 and MDA/LNA-antimiR-21 consisted of 27 genes (18 up- and 9 down-regulated; Figure 4b and Table 2).

To further screen potential direct targets, we compared the 27 candidate mRNAs with miR-21 targets predicted by TargetScan 5.1, miRBase Targets V.5, miRNAMap 2.0, PicTar and miRanda 3.0. Of the 27 mRNAs, 3 were recognized by the algorithms (Table 2). Because miR-21 targets are expected to up-regulated for the LNA-antimiR-21 samples, ANKRD46 and EIF4A2, the two up-regulated genes upon miR-21 knockdown in the two cell lines and predicted by target prediction programs, were selected for further investigation.

The microarrays were validated by qRT-PCR assay. Consistent with the microarray results, qRT-PCR showed increased ANKRD46 and EIF4A2 mRNA levels in MCF-7 and MDA-MB-231 cell lines upon miR-21 inhibition, although the increase of EIF4A2 in MDA/LNA-antimiR-21 cells was relatively modest (Figure 4c). To determine whether miR-21 affects the expression of the potential endogenous target genes, we transfected MCF-7 and MDA-MB-231 cells with LNA-antimiR-21 or LNA-control. Western bolt showed that LNA-antimiR-21 led to up-regulation of endogenous ANKRD46 in both MCF-7 (P = 0.005) and MDA-MB-231 cells (P = 0.004), but no significant up-regulation of EIF4A2 protein (Figure 5c).

We further tested whether miR-21 could directly repress the identified mRNA targets through 3' UTR interactions (Figure 5a). Thus, the full-length 3' UTRs of the human genes ANKRD46 and EIF4A2 were cloned into the downstream of the luciferase gene (pMIR-REPORT), respectively. These vectors were then used to assess whether miR-21 could repress luciferase activity in 293T cells. ANKRD46 3' UTR showed a reduction to 54.8% of total luciferase reporter activity, in presence of miR-21, but EIF4A2 3' UTR did not display significant reduction of luciferase levels, compared with the miR-control (Figure 5b). These results suggest that miR-21 directly targets ANKRD46 in BC cells.

We examined ANKRD46 [NCBI: NP940683] and EIF4A2 [NCBI: NP001958] protein levels by IHC on TMAs constructed by the BC cases described in Materials and Methods. EIF4A2 was found in the cytoplasm, and ANKRD46 was seen in both the cytoplasm and nucleus (Figure 6a). ANKRD46 low expression was found in 47.5% (staining score < 4; median = 4); EIF4A2 low expression was found in 21.2% (staining score < 7; median = 7) of the 99 BC cases. Next, we investigated the negative regulation of endogenous EIF4A2 and ANKRD46 protein by endogenous miR-21 in vivo. In 99 successfully tested cases out of 113, endogenous miR-21 and EIF4A2 protein levels were inversely expressed in resected patient tumors (r_s = -0.283, n = 99, P = 0.005, Spearman's correlation analysis). However, no significant association between miR-21 and ANKRD46 (P = 0.181, Spearman's correlation analysis) was observed.

Using Pearson's Chi-Square test, we found that, in BC tissues, IF4A2 protein correlated with CerbB2 status (P = 0.019), and ANKRD46 protein correlated with ER (P = 0.021) and PR (P = 0.001, Table 3). No significant correlation was observed between EIF4A2, ANKRD46, or other parameters. The five-year overall survival rate of the 99 BC patients was 59.60% (Figure 6b). The five-year survival rate in patients with high EIF4A2 protein level was 64.10% (n = 78), significantly higher than those with a low EIF4A2 protein level (42.86%, n = 21; P = 0.044, log-rank test; Figure 6b). The five-year survival rate in patients with high ANKRD46 protein was 53.85% (n = 52), which was not statistically different from those with low ANKRD46 protein (65.96%, n = 47; P = 0.146, log-rank test; Figure 6b).