miR-34a and miR-15a/16 are co-regulated in non-small cell lung cancer and control cell cycle progression in a synergistic and Rb-dependent manner

The NSCLC cell lines A549, H2009, H1299 and H358 were obtained from the American Type Culture Collection, Rockville, MD. All cell lines were cultured in Iscove's modified Dulbecco's medium supplemented with 2 mM L-alanyl-L-glutamine, 1% penicillin/streptomycin and 5% foetal bovine serum (Sigma) at 37°C and 5% CO₂.

Cells were seeded in culture flasks 24 h prior to transfection. Co-transfections with plasmid DNA were performed using Effectene reagent (Qiagen), all other transfections were performed using HiPerFect (Qiagen). If not otherwise specified, transfection was performed using 20 nM of hsa-pre-miR-34a, a mixture of 10 nM hsa-pre-miR-15a and 10 nM hsa-pre-miR-16 or 20 nM pre-miR miRNA precursor control 1 (Ambion). Si RNAs against Rb, or CCNE1 (siGENOME SMARTpool, Dharmacon) were used at 60 nM or 7.8 nM, respectively. Control transfections were performed using non-targeting Pool 2 (siGENOME). pCMV-Rb [19] or empty control plasmid were used at 125 ng/ml. Transfection efficiency of short RNAs and plasmid DNA was monitored using siGloGreen transfection indicator (Dharmacon) or an RFP-expression plasmid, respectively.

Cell cycle analysis was performed by flow cytometry essentially as described [7]. For cell death analysis, floating and adherent cells were harvested, combined, washed with PBS and stained with 10 μg/ml propidium idode (Sigma). Apoptotic cells were detected using an antibody directed against cleaved caspase 3 (clone 5A1E, 1:100, Cell Signaling) as described [20]. Cells were analysed using an LSR II flow cytometer (BD Biosciences) and FlowJo 8.8.4 software (Tree Star). As a positive control, cells were treated with UV (400 mJ) using a UV Stratalinker 1800 (Stratagene).

Total RNA was extracted from cultured cells using the miRVana RNA isolation kit according to the manufacturer's instructions (Ambion). TaqMan miRNA assays (Applied Biosystems) were performed as described [7] using a Real-Time PCR system 7500 (Applied Biosystems). miRNA levels were normalized to the level obtained for RNU48. Quantification of Bcl2 was done using a TaqMan assay (Applied Biosystems); all other mRNAs were quantified using Quantitec primer assays (Qiagen). mRNA levels were normalized to the level obtained for GAPDH. Changes in expression were calculated using the ΔΔCt method.

Western blot analysis was performed as described [7]. Monoclonal antibodies against Rb (clone 3C8, QED Bioscience) and phospho-Rb (Ser807/811, Cell Signaling) were diluted 1:1000, monoclonal antibody against Bcl2 (clone 124, Dako) was diluted 1:300, and monoclonal antibody against α-tubulin (clone B512, Sigma) was diluted 1:5000. Secondary goat anti-mouse-HRP and goat anti-rabbit-HRP antibodies (Biorad) were used at 1:5000 or 1:7000, respectively.

Formalin-fixed paraffin-embedded tissues from 11 adenocarcinomas and 12 squamous cell carcinomas were used for miRNA expression analysis. Tumour tissues and corresponding normal tissues from bronchiolar or alveolar epithelium, respectively, was collected by laser capture microdissection as described previously [7]. Stroma components including connective tissues, inflammatory cells and blood vessels were excluded. Nucleic acids were subjected to a heat-treatment in order to remove methylol groups introduced during formalin fixation and subjected to real-time PCR as described [7]. All experiments using human specimens were done according to the ethical guidelines of the Institute of Pathology, University of Bern, and were reviewed by the institutional review board.

Statistical analyses were performed using the GraphPAD prism software. Statistical differences were calculated using unpaired two-tailed student's t-test. A probability of p ≤ 0.05 was considered statistically significant. Statistical significance of correlation was assessed by the Pearson test.

The finding that miR-15a/16 and miR-34a share many common targets involved in G₁ progression prompted us to analyse if these miRNAs are co-regulated in NSCLC. Tumour tissues of adenocarcinomas and squamous cell carcinomas and matched normal tissues from alveolar or bronchial epithelium, respectively, were collected by laser capture microdissection and compared for the expression of both miRNAs by real-time PCR. We have shown previously that miR-15a and miR-16 are frequently down-regulated in NSCLC [7]. Here we demonstrate that the expression of miR-16 is significantly correlated to the expression of miR-34a in adenocarcinomas (p = 0.018), but not in squamous cell carcinomas of the lung. Both miRNAs were down-regulated in 82% (9/11), and up-regulated in 18% (2/11) of adenocarcinomas (Figure 1A). In contrast, miR-34a was significantly down-regulated in 10/12 squamous cell carcinoma samples while miR-16 was down-regulated in 5/12 tumour samples. Both miRNAs were also significantly down-regulated in the NSCLC cell lines A549, H2009, H1299 and H358 (data not shown).

Co-repression of miRNAs may be due to defects in miRNA processing. Notably, reduced expression of Dicer has been detected in NSCLC [21]. To address this possibility, the same tumour tissues were analysed for the expression of miR-21, which is frequently up-regulated in lung cancer [22, 23]. As shown in Figure 1B, miR-21 was up-regulated or expressed at normal levels in 9 of 11 adenocarcinomas and 7 of 12 squamous cell carcinomas. Thus, abrogation of miRNA processing may only account for a small subgroup of NSCLC samples.

We next investigated the possibility that both miRNAs are linked because they are able to mutually regulate their expression. To this end, miR-15a/16 or miR-34a were overexpressed by transfection with miRNA precursors (pre-miRNA) in a Rb-deficient NSCLC cell line, H2009, and analysed for the expression of the miRNA counterpart. Since H2009 is refractory to cell cycle arrest induced by miR-15a/16[7] and miR-34a (see below), secondary effects on miRNA expression as a consequence of the G₁-G₀ arrest can be excluded using this cell line. However, neither miR-15a/16 nor miR-34a was able to affect the expression of its counterpart, while CDK6 mRNA, a target common to both miRNAs, was significantly down-regulated (Figure 1C).

One miRNA can affect the expression of hundreds of proteins [24], which often renders it difficult to identify the relevant targets. We have previously shown that miR-15a/16-induced cell cycle arrest depends on the expression of Rb. This indicates that components of the cell cycle machinery upstream of Rb, including cyclin D1 (CCND1), cyclin D3 (CCND3), CDK4 and CDK6[7, 25, 26], are the most relevant targets of miR-15a/16. To investigate if miR15a/16 and miR-34a share redundant functions, miR-34a was analysed using the same set of experiments as we have described previously for miR-15a/16[7].

To investigate cell cycle arrest, the NSCLC cell lines A549, H358, H1299 and H2009 were transfected with pre-miRNA-34a and treated with nocodazole 24 h post-transfection. Nocodazole traps cells at the G₂-M phase, but ~30% of the transfected A549, H358 or H1299 cells accumulated in G₁-G₀ (Figure 2A and 2B) indicating that miR-34a induces an arrest in this phase of the cell cycle. In contrast, Rb-deficient H2009 cells were completely refractory to miR-34a-induced arrest (Figure 2A and 2B). However, known targets of miR-34a including CDK4, CDK6, CCND1 and c-Met were significantly down-regulated in these cells (Figure 2C). Rb reconstitution into Rb-deficient NSCLC lines restores G₁ arrest mechanisms [27]. To investigate if miR-34a-induced cell cycle arrest depends on the expression of Rb, the latter gene was reintroduced into H2009 cells. Co-transfection of H2009 cells with pre-miR-34a and an empty control plasmid induced cell cycle arrest in 3.1 ± 1% of the population (Figure 3A). Consistent with previous findings [27], the percentage of cells in the G₁-G₀ phase of the cell cycle increased to 16 ± 1% upon transfection with a Rb expression plasmid (p = 0.001). This indicates that Rb per se is able to induce cell cycle arrest in a significant proportion of the population. However, Rb plasmid in combination with pre-miR-34a induced cell cycle arrest in significantly more cells (26 ± 1%, p = 0.01) than Rb plasmid in combination with the precursor control. The expression of Rb protein was not affected by pre-miR-34a. In contrast, phospho-Rb was reduced by 50% under these conditions (Figure 3B). In conclusion, the ability of miR-34a to induce cell cycle arrest depends on the expression of Rb. Complementary experiments were performed in A549 cells, in which the Rb gene was knocked down by RNA interference. The knock-down expressed three times less Rb protein than the control (Figure 3C). As expected, the knock-down was significantly more resistant to miR-34a-induced cell cycle arrest (22 ± 1% in G₁-G₀) than the control (43 ± 1% in G₁-G₀, p < 0.001) (Figure 3D). In conclusion, there is a significant degree of redundancy between miR-34a and miR-15a/16 in their ability to induce cell cycle arrest in NSCLC cells.

We next addressed the question if miR-15a/16 and miR-34a act together to induce cell cycle arrest. A549 cells were transfected with increasing concentrations of pre-miR-15a/16 or pre-miR-34a and analysed for cell cycle arrest (Figure 4A). From the slope of the dose-response curves it can be deduced that pre-miR-34a was more efficient than pre-miR-15a/16 in inducing cell cycle arrest. We next assessed the concerted action of miR-15a/16 and miR-34a precursors on cell cycle arrest. Transfection with 2.5 nM pre-miR-15/16 or transfection with 0.63 nM pre-miR-34a resulted in a G₁-G₀ arrest of A549 cells in 10.9 ± 0.6% and 20.1 ± 1.6% of the population, respectively. Interestingly, a mixture with half the concentrations of pre-miR-15/16 and pre-miR-34a was more efficient (25.9 ± 2.2%) than each pre-miRNA alone in inducing a G₁-G₀ arrest, p ≤ 0.02 (Figure 4A). Consistent results were obtained over a four-fold concentration range. Thus, these results clearly indicate that miR-15a/16 and miR-34a act synergistically to induce cell cycle arrest in G₁-G₀.

Both pre-miRNAs displayed saturation for cell cycle arrest at a concentration of 20 nM (data not shown). Interestingly, a synergistic effect was also obtained at this concentration: cells transfected with 20 nM pre-miR-15a/16 or 20 nM pre-miR-34a in each case gave rise to a G₁-G₀ arrest in about 37% of the population. In contrast, co-transfection of cells with both pre-miRNAs at half the concentration (10 nM each) resulted in a G₁-G₀ arrest in 54.6 ± 2.2% of the population (Figure 4B, 24 h), p < 0.0005. The synergistic effect was observed 24 h and 48 h post-transfection (Figure 4B).

miR-15a/16 and miR-34a are both able to down-regulate the anti-apoptotic protein Bcl2 (Figure 5A). In addition, both miRNAs have been shown to induce apoptosis in different cell systems [13, 15]. We therefore wondered if these miRNAs also act synergistically to induce apoptosis. In line with previous findings we were unable to detect any significant increase in propidium iodide (PI)-positive A549 or H2009 cells on transfection with miR-15a/16 (Figure 5B and Additional file 1). Likewise, no significant increase in cleaved caspase-3-positive cells was observed (Figure 5C). In contrast, miR-34a elicited a 2-3-fold increase in PI-positive cells and an eight-fold increase in cleaved caspase-3-positive cells 72-96 h post-transfection (Figure 5B and 5C). Surprisingly, a mixture of both pre-miRNAs at half the concentration was less efficient than miR-34a alone in inducing cell death. In conclusion, no synergism elicited by miR-15a/16 and miR-34a exists for cell death in NSCLC cells.

We next investigated the mechanism underlying the synergistic action of miR-34a and miR-15a/16. It is possible that the observed effect is due to a more efficient down-regulation of targets common to both miRNAs by a combined action of miR-34a and miR-15a/16. To circumvent adverse secondary effects on target gene expression of individual G₁ proteins as a consequence of cell cycle arrest, the cell line H2009 was again used for the experiment. Transfection with serial dilutions of pre-miRNAs allowed the establishment of a direct relationship between the amount of input pre-miRNA and the steady-state level of mRNA of the target genes CDK4, CDK6 and CCND1, respectively (Figure 6A and Additional file 2).

Interestingly, transfection of H2009 cells with miR-34a or miR-15a/16, or co-transfection of H2009 cells with a mixture of both pre-miRNAs at half the concentration gave rise to comparable dose-response curves for all the three target genes (Figure 6A). From these results we may conclude that miR-15a/16 and miR-34a act in an additive rather than in a synergistic manner on individual mRNAs. Comparable results were obtained in H2009 and A549 cells excluding the possibility that the concerted action on individual mRNA targets depends on the expression of Rb (data not shown).

Although miR-15a/16 and miR-34a share many common targets, other targets exist which are unique to miR-15a/16 or miR-34a (Additional file 2). Since no synergistic effect was observed for individual mRNA targets, we hypothesized that the observed effect may be due to the fact that more targets involved in G₁-S progression are repressed by the combined action of both miRNAs. Target specificity was re-evaluated using the cell line H2009, which is refractory to miRNA-induced arrest. In agreement with reports from the literature [7, 16], down-regulation of c-Met mRNA was specific for miR-34a, whereas down-regulation of CCND3 and CCNE1 mRNAs were specific for miR-15a/16 (Figure 6B and Additional file 2). In contrast, CCNE2 mRNA, which contains a target site for miR-34a[16], and CCNA1 mRNA, which contains no target site, were virtually unaffected.

To address the possibility that the synergistic effect was due to an increased number of targets, CCNE1, a target unique to miR-15a/16 (Figure 6B and Additional file 2), was knocked down in A549 cells by RNA interference resulting in 80.3 ± 6.6% less CCNE1 mRNA. We would expect that the synergistic effect would be reduced under these conditions, given that one of the unique targets, CCNE1, has been removed. This was indeed the case. The percentage of cells in the G₁-G₀ phase increased to 36% on co-transfection with CCNE1 siRNA and pre-miR-control (Figure 7A, striped columns). However, CCNE1 siRNA had only a low impact on cell cycle arrest of cells overexpressing miR-15a/16 (Figure 7A, white columns), which can be explained by the fact that CCNE1 was already down-regulated by miR-15a/16. In contrast, the percentage of A549 cells in G₁-G₀ increased almost two-fold on co-transfection with pre-miR-34a and CCNE1 siRNA relative to cells co-transfected with pre-miR-34a and si control (Figure 7A, grey columns), suggesting that CCNE1 siRNA and pre-miR-34a act together to induce cell cycle arrest in a more efficient manner. Notably, CCNE1 siRNA in combination with pre-miR-34a (59.0 ± 3.1%; grey column) and CCNE1 siRNA in combination with both pre-miRNAs (61.6 ± 5.3%; black column; p = 0.5) induced a G₁-G₀ arrest with the same efficiency. Thus, the synergistic effect exerted by the combined action of miR-15a/16 and miR-34a was clearly abrogated. In contrast, si control in combination with both pre-miRNAs together gave rise to almost two times more cells in G₁-G₀ than si control in combination with pre-miR-34a or pre-miR-15a/16, respectively (p < 0.004). The observed effects were not cell-line-specific, since comparable results were obtained for A549 (Figure 7A) and H1299 cells (Figure 7B). In conclusion, the synergistic effect of miR-15a/16 and miR-34a is due to the fact that more miRNA targets are down-regulated by the combined action of both miRNAs.