MicroRNAs (miRNAs) are a distinct class of small noncoding RNAs that posttranscriptionally repress expression of target genes through imperfect base pairing with the 3′ untranslated region. We previously reported amplification and overexpression of the miR-17-92 miRNA cluster at 13q31.3 in lung cancers, as well as growth inhibition by treatment with antisense oligonucleotides against miR-17-5p and miR-20a, constituents of miR-17-92, specifically in miR-17-92–overexpressing lung cancer cell lines. Although these findings clearly suggested important roles of miR-17-92 overexpression in lung cancers, only a few targets for the miR-17-92 cluster have been identified thus far. In this study, we identified hypoxia-inducible factor (HIF)-1α as a novel direct target for miR-17-92 through global expression profiling by mass spectrometric analysis using an isobaric tagging reagent, iTRAQ, combined with bioinformatic target prediction. This is the first report to describe negative regulation of HIF-1α by miRNA, which seemed to occur without disrupting the induction of HIF-1α for cellular adaptation to hypoxia. In addition, overexpression of c-myc led to down-regulation of HIF-1α and induction of miR-17-92, the latter of which was previously reported to be a transcriptional activation activity, suggesting that the induction of miR-17-92 may play a role at least in part in c-myc–mediated repression of HIF-1α. Together with previous reports on the functional negative regulation of c-myc by HIF-1α, our findings suggest the possible existence of an intricate and finely tuned circuit involving c-myc, miR-17-92, and HIF-1α that may play a role in cancer cell proliferation under normoxia in a cellular context–dependent manner. [Cancer Res 2008;68(14):5540–5]
MicroRNAs (miRNA) are a class of small noncoding RNAs of ∼22 nucleotides in length that are the products of sequential processing of primary RNA polymerase II transcripts mediated by two RNase III enzymes, Drosha and Dicer. miRNAs repress protein expression at the posttranscriptional level through imperfect base pairing with the 3′ untranslated region (UTR) of the target mRNA, leading to its reduced translation and/or degradation. Although miRNAs are thought to be involved in various biological processes, including development, differentiation, cell proliferation, and cell death, accumulating evidence suggests that alterations of their expression may play a role in the development of human cancers (1).
Lung cancer is the leading cause of cancer death in most economically developed countries, including Japan, and is characterized by the presence of various genetic and epigenetic alterations in cancer-related genes with coding capacity. In addition to these changes, emerging evidence suggests that miRNAs may also be involved in the development of lung cancers. We previously reported that members of the let-7 miRNA family are frequently down-regulated in lung cancers in association with a poor postoperative prognosis (2). This notion was further substantiated by the subsequent identification of RAS as a target gene for the let-7 family (3). In addition, we previously found amplification and overexpression of the miR-17-92 miRNA cluster at 13q31.3 in lung cancer (4), whereas antisense-mediated inhibition of miR-17-5p and miR-20a, constituents of miR-17-92, induced apoptosis in miR-17-92–overexpressing lung cancer cell lines (5), suggesting the existence of addiction to continued overexpression of miR-17-92 for cancer development. This miRNA cluster has also been suggested to play a role in B-cell lymphoma development (6). Together, these findings suggest possible important roles of miR-17-92 overexpression in human cancers, although only a few targets for the miR-17-92 cluster have been identified thus far (7).
In the present study, we identified hypoxia-inducible factor-1α (HIF-1α) as a novel direct target for miR-17-92 through global expression profiling by mass spectrometric analysis combined with bioinformatic target prediction. Together with previous reports of functional inhibition of c-myc by HIF-1α (8, 9), our results support the notion of the possible existence of a feedback loop between c-myc and HIF-1α via miR-17-92, although we also showed the involvement of miR-17-92–mediated HIF-1α inhibition of cell growth in a normoxic condition with considerable distinctions between lung cancer cell lines.
Materials and Methods
DNA constructs. A cDNA fragment containing the miR-17-92 cluster, also used in our previous report (4), was recloned into the EcoRV site of a modified pT2GN vector carrying a neomycin resistance gene. A 1,147-bp HIF-1α 3′ UTR containing target sites for miR-17-92 was amplified by PCR using primers with the SfiI site and cloned downstream of the luciferase coding sequence in the modified pGL3 vector. The following primers were used to amplify 3′UTR: 5′-AGTGGCCAGTAGGGCCTGGACACTGGTGGCTCATTA-3′ (forward) and 5′-AGTGGCCCTACTGGCCGCCTGGTCCACAGAAGATGT-3′ (reverse). To generate mutations in each predicted target site for miR-17-5p and miR-20a, four nucleotides, each corresponding to the seed sequences of miR-17-5p and miR-20a, were deleted using a QuikChange site-directed mutagenesis kit (Stratagene; Supplementary Fig. S1). The pMX vector carrying c-myc was a kind gift from Dr. Masao Seto (Aichi Cancer Center, Nagoya, Japan).
Cell culture, generation of stable cell lines, and transient transfection. Two immortalized lung epithelial cell lines, BEAS2B and HPL1D, were maintained as previously described (4). ACC-LC-172 and Calu6 lung cancer cell lines overexpressing miR-17-92 (5) were cultured in RPMI 1640 with 5% FCS. Cells were grown under hypoxic (1% O2) or normoxic (21% O2) conditions at 37°C/5% CO2. Stable clones of BEAS2B cells overexpressing the miR-17-92 cluster were generated by transfecting the pT2GN vector expressing the miR-17-92 cluster into BEAS2B cells using FuGENE6 (Roche), which was followed by selection in medium containing 200 μg/mL of geneticin (Invitrogen) for 7 d. The pMX vector expressing c-myc was then transfected using FuGENE6 into BEAS2B cells and selected with 100 μg/mL of hygromycin B (BD Biosciences) for 4 d.
Mass spectrometry analysis. An overview of the workflow is shown in Fig. 1A. Detailed methods can be found in the Supplementary Information section.
Dual-luciferase reporter assay. BEAS2B cells at 2 × 105 were plated in six-well culture plates 24 h before transfection. The cells were then transfected with a pGL3 vector (1.8 μg) together with a pRLTK vector (0.2 μg) for normalization of transfection efficiency. Cell lysates were collected and assayed after 48 h. Firefly and Renilla luciferase activities were determined using a Dual-Luciferase Reporter Assay System (Promega), with firefly luciferase activities calculated as the mean ± SD after being normalized by Renilla luciferase activities. Three independent experiments were done in triplicate.
RNA extraction, real-time reverse transcription-PCR, and Northern blot analysis. RNA samples were prepared using standard acid-phenol extraction procedures. First-strand cDNA samples were synthesized from total RNA using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems), according to the manufacturer's instructions. Real-time reverse transcription-PCR of miR-20a and 5S rRNA was done using the following primer sets: miR-20a, 5′-CGATGTAGAATCTGCCTGGTC-3′ (forward) and 5′-GGATGCAAACCTGCAAAACT-3′ (reverse); 5S, 5′-TCGTCTGATCTCGGAAGCTAA-3′ (forward) and 5′-TTAGCTTCCGAGATCAGACGA-3′ (reverse). Northern blot analysis of the miRNA samples was done using 5 μg of RNA, as previously described (2). The oligonucleotide probes used were the following: miR-20a, 5′-CTACCTGCACTATAAGCACTTTA-3′; 5S rRNA, 5′-TTAGCTTCCGAGATCAGACGA-3′.
Oligonucleotides and transfection experiments. BEAS2B cells were transfected with 10 nmol/L of the Pre-miR miRNA Precursor Molecule of miR-20a, Negative Control #1 (both from Applied Biosystems), 20 nmol/L of a small interfering RNA (siRNA) duplex (Sigma-Aldrich) targeting HIF-1α (siHIF-1α), 50 nmol/L of a siRNA duplex (Greiner) targeting HIF-2α (siHIF-2α), or the negative control (10) using Lipofectamine 2000 (Invitrogen). ACC-LC-172 and Calu6 cells were transfected with 20 nmol/L of siHIF-1α or the negative control in the presence of 20 nmol/L of locked nucleic acid (LNA) antisense oligonucleotides (Greiner) against miR-20a/scramble using Lipofectamine 2000. The transfection experiments and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays were performed as previously described (5).
Western blot analysis. Western blot analysis was performed using the following antibodies: anti-HIF-1α (BD Transduction Laboratories), anti-HIF-2α (Abcam), anti-HIF-1β (BD Transduction Laboratories), anti-phosphoglycerate kinase 1 (PGK1; Santa Cruz Biotechnology), anti-enolase 1 (ENO1; Santa Cruz Biotechnology), and anti-c-myc (9E10; Santa Cruz Biotechnology). α-Tubulin (Sigma-Aldrich) was used as a loading control.
To conduct a proteomic search for targets for the miR-17-92 cluster, we used differential tagging with iTRAQ followed by multidimensional liquid chromatography (LC) and tandem mass spectrometry (MS/MS) analysis (Fig. 1A). This proteomic comparison between a miR-17-92–transfected clone (miR-17-92 #57) and an empty vector–transfected clone (VC #1) from the BEAS2B normal human bronchial epithelial cell line allowed us to identify 479 proteins in the isobaric-tagged lysates, which resulted in the identification of eight proteins with more than a 2-fold down-regulation in the miR-17-92–transfected clone when compared with the empty vector–transfected clone (Fig. 1B). Western blot analysis of PGK1 and ENO1 confirmed their differential expressions (Fig. 2A). Interestingly, four of the eight proteins with significant down-regulation, PGK1, ENO2, ENO1, and triosephosphate isomerase 1 (TPI1), are in the glycolytic pathway. We also found that six of the eight proteins, which included eukaryotic translation initiation factor 3, subunit 10 𝛉, PGK1, β-galactoside–binding lectin precursor, ENO2, ENO1, and TPI1, lacked potential target sites for the miRNAs comprising the miR-17-92 cluster (Fig. 1B), according to the results of three different well-known target prediction programs, TargetScan, PicTar, and miRBase.
Down-regulation of multiple proteins lacking potential target sites for miR-17-92 suggested the possibility that miR-17-92 could target a key transcription factor that commonly regulates their expression. We consequently discovered that HIF-1α, a well-known molecule regulating the glycolytic pathway (11, 12), was computationally predicted to be a target for multiple miRNAs comprising the miR-17-92 cluster (Fig. 2B). As expected, Western blot analysis revealed a significant down-regulation of HIF-1α protein in BEAS2B cells stably overexpressing miR-17-92. HIF-2α carrying potential target sites was also found to be down-regulated, whereas the expression of HIF-1β did not show any obvious changes (Fig. 2C,, left). Transient transfection of the Pre-miR molecule of miR-20a into parental BEAS2B cells clearly reduced the expression of both HIF-1α and HIF-2α proteins but not of HIF-1β (Fig. 2C , right). These findings were consistent with the computational target predictions for HIF-1α and HIF-2α, but not for HIF-1β, as targets of miR-17-92. We also noted that inhibition of HIF-1α, but not HIF-2α, with the corresponding siRNA clearly repressed PGK1 expression (Supplementary Fig. S2), which is consistent with the notion that HIF-1α is a key regulator of glycolytic enzymes, such as PGK1, whereas HIF-2α does not regulate glycolytic enzymes (12).
To confirm a direct repression of HIF-1α by miR-17-92, a 1,147-bp fragment of the 3′ UTR of HIF-1α, which contained eight putative target sites for constituents of the miR-17-92 cluster, was cloned downstream of the luciferase coding sequence in a modified pGL3 vector. A dual-luciferase assay was then performed by transfecting pGL3 with or without the 3′ UTR of HIF-1α, together with a pRLTK control vector for normalization, into the miR-17-92–expressing clones miR-17-92 #31 and #57, or control clones, as well as VC #1 and VC #2. A significant decrease in luciferase activities in clones #31 and #57 was clearly observed, although it was less marked than the extent of reduction in HIF-1α expression shown in Western blot analysis, raising the possibility of both direct and indirect effects of miR-17-92 on HIF-1α expression. In contrast, pGL3 carrying the mutant 3′ UTR of HIF-1α with 4-bp deletions in the core of seed sequences of each target site for miR-17-5p and miR-20a did not show obvious decreases in luciferase activities when transfected into clones #31 and #57 (Fig. 2D).
Because HIF-1α is known to play a crucial role in various hypoxic responses, we also analyzed whether such negative regulation of HIF-1α could be indirectly provoked by hypoxia-induced up-regulation by miR-17-92. However, virtually no change in expression of miR-17-92 was observed, as shown in our results of Northern blot analysis of miR-20a, and robust HIF-1α induction was observed under hypoxia regardless of the constitutive expression levels of miR-17-92 (Fig. 3A). These findings indicate that miR-17-92 overexpression negatively regulates HIF-1α expression only under normoxia without disturbing the robust induction of HIF-1α under hypoxia, which is thought to confer a growth advantage to cancer cells in a hypoxic microenvironment. In this context, we observed a clear induction of HIF-1α under a hypoxic condition in the normal lung epithelial cell lines BEAS2B and HPL1D (Supplementary Fig. S3) as well as in miR-17-92–overexpressing ACC-LC-172 and Calu6 cells (Supplementary Fig. S4).
It has been shown that c-myc induces miR-17-92 by direct binding to the promoter region of C13orf25 (7), a vehicle of miR-17-92. We therefore examined whether c-myc could negatively regulate HIF-1α in association with induction of miR-17-92. Transient transfection of c-myc clearly led to up-regulation of miR-20a and down-regulation of HIF-1α in BEAS2B cells (Fig. 3B). Furthermore, inhibition of miR-20a with antisense LNA oligonucleotides was shown to alleviate the repression of HIF-1α protein expression caused by overexpression of exogenous c-myc in BEAS2B cells (Fig. 3C), suggesting that c-myc–mediated repression of HIF-1α may be mediated in part through the induction of miR-17-92. We previously reported that inhibition of miR-17-5p and/or miR-20a with antisense oligonucleotides induced growth inhibition and apoptosis in lung cancer cell lines overexpressing miR-17-92, such as ACC-LC-172 and Calu6 (5). Thus, we investigated whether the observed growth inhibition could be mediated in part by HIF-1α induction by antisense oligonucleotide-mediated inhibition of miR-17-92 components. Treatment with an antisense oligonucleotide under normoxia induced HIF-1α protein in both ACC-LC-172 and Calu6 cells (Fig. 4A). In contrast, we did not observe a clear induction of HIF-2α possibly because of its rather inefficient inhibition by endogenous levels of miR-17-92 expression compared with HIF-1α. Simultaneous treatment with siRNA against HIF-1α, which clearly overrode the miR-20a antisense oligonucleotide-mediated induction of HIF-1α, partially restored cell growth of ACC-LC-172 cells (Fig. 4A). In contrast, Calu6 cells showed a significant increase in cell growth when treated with siRNA against HIF-1α, although it was markedly abrogated by miR-20a antisense oligonucleotide treatment, suggesting that repression of HIF-1α by miR-17-92 might play a role that is dependent on the cellular context.
In the present study, we identified HIF-1α as a novel target for the miR-17-92 cluster through differential tagging with an isobaric tagging reagent, iTRAQ, followed by multidimensional LC and MS/MS analysis. Eight of the 479 proteins identified in the course of protein identification were shown to be down-regulated by overexpression of miR-17-92. Four of those eight that had down-regulation of greater than a 2-fold were those involved in the glycolytic pathway, whereas miR-17-92 seemed to regulate their expression via down-regulation of HIF-1α. The reason why we could not identify HIF-1α itself as a differentially expressed molecule may be related to the relatively low ability of the lysis buffer, which we optimized for iTRAQ labeling and MS/MS analysis; thus, nuclear proteins such as transcription factors might not have been sufficiently solubilized. It is also possible that the digested peptides from HIF-1α were eluted together with those from other highly abundant proteins during the on-line two-dimensional LC separation, making it difficult to detect HIF-1α–derived peptides.
HIF-1α is an oxygen-dependent transcription factor that transactivates genes involved in various biological processes, including angiogenesis, apoptosis, pH regulation, glucose metabolism, extracellular matrix metabolism, cell proliferation/survival, invasion, and metastasis (13). It is well established that von Hippel-Lindau (VHL) tumor suppressor–mediated ubiquitination plays a major role in the regulation of expression level of HIF-1α by proteasome-dependent degradation under normoxia. Conversely, oxygen depletion stabilizes HIF-1α followed by heterodimerization with constitutively expressed HIF-1β/aryl hydrocarbon receptor nuclear translocator. Recent reports also indicate that HIF-1α and c-myc act cooperatively in some pathways, whereas in others HIF-1α displaces c-myc binding, represses c-myc transcriptional activity, or degrades c-myc protein, resulting in a functional inhibition of c-myc (14). However, this is the first report of the involvement of miRNA in regulation of HIF-1α, which in turn suggests the possible existence of intricate and finely tuned mechanisms of HIF-1α expression that involve c-myc and miR-17-92.
Of interest, we observed repression of HIF-1α by miR-17-92 only under a normoxic condition, whereas HIF-1α was robustly induced under hypoxia regardless of the level of miR-17-92 expression. These findings suggest that miR-17-92 may play a role in fine tuning of the basal level expression of HIF-1α under normoxia, which is regulated by VHL-mediated, proteasome-dependent regulation. It was previously reported that HIF-1α was up-regulated in various human cancers possibly as a result of intratumoral hypoxia, which is thought to confer growth advantage to cancer cells (13). However, the contribution of HIF-1α toward tumor initiation and progression is not clear. For example, it was shown that HIF-1α is able to induce cell cycle arrest by activating p21 or p27 (8, 15) as well as apoptosis through stabilization of p53 or transactivation of BNIP3 in a certain cellular context. It has also been reported that HIF-2α, but not HIF-1α, is important in terms of cell growth promotion under normoxia (16, 17) and enhances c-myc transcriptional activity (18), whereas HIF-1α inhibits cell cycle progression by opposing c-myc (9). Moreover, it was shown that HIF-1α−/− tumors grew faster (19) and became more invasive than their HIF-1α+/+ counterparts when oxygen supply was adequate (20). The partial restoration of cell growth seen following simultaneous treatment with siRNA against HIF-1α and antisense oligonucleotide against miR-20a suggests that miR-17-92–mediated repression of HIF-1α may be advantageous for cell growth of ACC-LC-172 cells in a normoxic condition. However, that is probably dependent on cellular context because repression of HIF-1α did not seem to play a major role in the inhibition of cell growth of Calu6 cells following treatment with antisense oligonucleotide against miR-20a. In this regard, it is interesting to note that c-myc is amplified and overexpressed in ACC-LC-172 cells, whereas miR-17-92 itself is amplified and overexpressed without an obvious overexpression of c-myc in Calu6 cells (4).
In conclusion, we identified HIF-1α as a novel target for miR-17-92 using iTRAQ tagging and multidimensional LC and MS/MS analyses. Our results suggest the possible existence of an intricate and finely tuned circuit involving c-myc, HIF-1α, and miR-17-92 that plays a role in cancer cell proliferation under normoxia in a cellular context without interfering with the robust induction of HIF-1α for cellular adaptation to hypoxia.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Princess Takamatsu Cancer Research Fund grant 07-23903 and a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Masao Seto for providing the c-myc expression vector.