MicroRNA miR-210 modulates cellular response to hypoxia through ...

[Cell Cycle 8:17, 2756-2768; 1 September 2009]; ©2009 Landes Bioscience

Report

MicroRNA miR-210 modulates cellular response to hypoxia through the MYC antagonist MNT Zhan Zhang,1,* Hong Sun,1 Hongyue Dai,1 Ryan M. Walsh,1 Maki Imakura,1 Janell Schelter,1 Julja Burchard,1 Xudong Dai,1 Aaron N. Chang,1 Robert L. Diaz,1 Joseph R. Marszalek,2 Steven R. Bartz,3 Michael Carleton,1 Michele A. Cleary,1 Peter S. Linsley1 and Carla Grandori1,* 1Rosetta

Inpharmatics LLC, a wholly-owned subsidiary of Merck & Co., Inc.; Seattle, WA USA; 2Department of Cancer Pathways; Merck Research Laboratories; Merck & Co., Inc.; Boston, MA USA; 3Sirna Lead Discovery; Merck Research Laboratories; Merck & Co., Inc.; San Francisco, CA USA

Key words: miR-210, hypoxia, HIF, MYC, MNT

The hypoxia-inducible factor (HIF) pathway is essential for cell survival under low oxygen and plays an important role in tumor cell homeostasis. We investigated the function of miR-210, the most prominent microRNA upregulated by hypoxia and a direct transcriptional target of HIFs. miR-210 expression was elevated in multiple cancer types and correlated with metastasis of breast and melanoma tumors. miR-210 overexpression in cancer cell lines bypassed hypoxia-induced cell cycle arrest and partially reversed the hypoxic gene expression signature. We identified MNT, a known MYC antagonist, as a miR-210 target. MNT mRNA contains multiple miR-210 binding sites in the 3' UTR and its knockdown phenocopied miR-210 overexpression. Furthermore, loss of MYC abolished miR-210-mediated override of hypoxia-induced cell cycle arrest. Comparison of miR-210 and MYC overexpression with MNT knockdown signatures also indicated that miR-210 triggered a “MYC-like” transcriptional response. Thus, miR-210 influences the hypoxia response in tumor cells through targeting a key transcriptional repressor of the MYC-MAX network.

Introduction Intratumoral hypoxia is a hallmark of most solid tumors and results from increased oxygen consumption and/or insufficient blood supply. Many of the hypoxia-induced cellular responses are mediated through the hypoxia-inducible factors (HIFs),1,2 which regulate genes involved in angiogenesis, survival, cell metabolism, invasion and other functions.3 HIFs are basic-helix-loop-helix-PAS protein family transcription factors that bind hypoxia-regulated elements (HREs) of specific target genes. HIFs function as obligate *Correspondence to: Zhan Zhang; 401 Terry Ave. N; Seattle, WA USA; Tel.: 425.270.5892; Email: [email protected]/Carla Grandori; Department of Pharmacology; University of Washington School of Medicine; 815 Mercer; Seattle, WA; Tel.: 206.543.8119; Email: [email protected] Submitted: 06/23/09; Accepted: 06/29/09 Previously published online as a Cell Cycle E-publication: http://www.landesbioscience.com/journals/cc/article/9387

2756

heterodimers containing an α-subunit (HIF-1α or HIF-2α) and a β-subunit (HIF-1β). In the presence of oxygen, the α-subunits are degraded by the Von Hippel-Lindau (VHL) tumor suppressor, part of an E3 ubiquitin ligase complex. During hypoxia, HIF-α protein is protected from degradation and translocates to the nucleus, where it binds to constitutively expressed HIF-1β and activates HIF target genes.4 Whereas HIF-1α is ubiquitously expressed, HIF-2α expression is limited to specific tissues, including kidney, heart, lungs and endothelium.5 Although HIF-1α and HIF-2α regulate distinct genes, they also share targets including VEGF and ADRP (Adipose Differentiation-related protein).6,7 HIF-1α and HIF-2α also exert opposite effects on the activity of the c-MYC oncoprotein. Stabilization of HIF-1α by hypoxia causes cell cycle arrest at G1/S by inhibition of c-MYC function, and several mechanisms underlying this effect have been demonstrated: HIF-1α induces MXI1, the c-MYC antagonist;8 HIF-1α displaces Myc from target gene promoters9,10 and promotes proteasome-dependent degradation of c-MYC.8,11 Moreover, HIF-1α inhibits c-MYC function by binding MAX and thereby competing with MYC.6,7 In contrast, HIF-2α promotes cell cycle progression by enhancing c-MYC activity by binding and stabilizing MYC/ MAX complexes.6,7,12 However, recent data also indicate that MYC-induced lymphomagenesis is inhibited by knock-down of HIF-1α, implying that MYC and HIF-1α can collaborate, an outcome that could not be predicted by a simple model of HIF-1α being an antagonist of MYC.11,13 Hypoxia alters the expression of hundreds of mRNAs required for many aspects of tumorigenesis, and the HIF transcription factors play a central role in this response.6,14 Recently, the effect of hypoxia on microRNA expression was reported.15-19 microRNAs are a novel class of gene modulators that can each regulate as many as several hundred genes with spatial and temporal specificity.20,21 These non-coding RNAs have been proposed to contribute to oncogenesis by functioning either as tumor suppressors (e.g., miR-15a/miR16-1) or oncogenes (e.g., miR-155 and the miR-17-92 cluster).16,22,23

Cell Cycle

2009; Vol. 8 Issue 17

miR210 overrides hypoxia-induced cell cycle arrest by activating c-MYC

Because of the importance of the hypoxic response in tumorigenesis and the potential critical roles of microRNAs in gene regulation, we investigated the function of miR-210, the most prominent hypoxia-induced microRNA (reviewed in ref. 19). Our studies show that miR-210 is directly induced by HIFs and it modulates the cell cycle and hypoxia gene expression signature by targeting MNT, a transcriptional repressor known to antagonize MYC function.24,25 In linking hypoxia to tumor severity, we show that miR-210 levels correlate with poor prognosis in breast and melanoma cancer samples.

Results miR-210 expression is upregulated by hypoxia in vitro and correlates with hypoxic signatures in human tumors. To discover hypoxia-modulated microRNAs in tumor cells we measured the expression of ~200 microRNAs in a panel of cancer cell lines. Figure 1A shows a graphical comparison of the microRNA levels in the colon cancer cell line, HT29, grown under normal and hypoxic conditions. In these cells, miR-210 (highlighted red) was upregulated ~19 fold by hypoxia treatment (the number of miR-210 copies per 10 pg of RNA increased from 165 to 3,075). Most other microRNAs fall on the diagonal of the graph, indicating similar expression levels in cells grown under normal (X axis) and hypoxic (Y axis) conditions (Table S1). Similar results were obtained in other colon lines (HCT116, DLD1 and RKO), cervical lines (HeLa, and ME-180), a glioma line (U251), and kidney lines (786-O and RCC4), as well as in primary cells such as HMECs (human mammary epithelial cells) and HFFs (human foreskin fibroblasts). miR-210 copies increased in most cell lines tested and with an average upregulation of ~8 fold by 24 hrs (Figs. 1E and S1A). These findings agree with a previous report of microRNAs induced by hypoxia.19 However, the copy number of miR-210/cell varied among different cancer lines, being highest in ME-180, HeLa and HT29, while lowest in normal HFFs (Fig. 1E). Furthermore, miR-210 levels were constitutively elevated in cells with defective VHL function, RCC4-pBABE and 786-0-pBABE, which contain elevated HIFs levels independently of hypoxic stimulation (Figs. 1E, and 2E and F). Reconstitution of VHL function, restored miR-210 inducibility in both cell lines (Fig. 2E and F) indicating that miR-210 expression depends on HIFs. miR-210 is located within the intron of a non-coding gene on human chromosome 11 (Fig. S2). The expression of its nascent primary transcript (pri-miR-210) was also upregulated by hypoxia as early as 4 hours post-hypoxia treatment in all cell lines tested: HCT116, HeLa, HT29, ME-180, U251, 786-O, H1299, MCF7, A549, PC-3, Hep3B, HuH7, DLD1, RKO and HFF (Fig. S1C). To determine whether miR-210 levels correspond to the hypoxic response in human tumors, we compared miR-210 levels with levels of an hypoxia-induced gene set (Zhang Z, unpublished data) in human tumors and matched adjacent normal tissue samples. As shown in Figure 1B, miR-210 levels in tumors showed significant positive correlation with transcripts upregulated by hypoxia. In addition, pri-miR-210 was expressed at high levels in multiple tumor types, including kidney, breast and lung compared with www.landesbioscience.com

adjacent normal tissues (Figs. 1C and S3). As hypoxia-responsive genes serve as robust biomarkers for breast cancer patient outcome,14 we asked whether pri-miR-210 levels had similar predictive power. Figure 1D shows a Kaplan-Meier curve depicting the relationship between metastasis-free survival time and the level of pri-miR-210 in a set of breast cancer samples previously analyzed by microarray-based gene expression profiling.26 The results of this analysis indicated that upregulation of pri-miR-210 is correlated with the metastatic potential of these breast cancer tumor samples. Similar results were obtained by analyzing a melanoma dataset (Fig. S3C). miR-210 is a direct target of both HIF-1α and HIF-2α. The early kinetics of pri-miR-210 upregulation was consistent with miR-210 serving as a direct target of HIF (Fig. S1C). To test the effects of loss of either the α subunits (HIF-1α or HIF-2α) or the common β subunit (HIF-1β) on the level of miR-210, we suppressed these genes by RNA interference. As shown in Figure 2A and B, under hypoxic conditions, silencing of HIF-1β or HIF-1α in HCT116 cells reduced expression of pri-miR-210 (represented by Contig63649_RC on the signature plots) by 76% and 64%, respectively. In 786-O-pBABE cells, which are predominantly HIF-2α dependent for HIF activity, silencing of HIF-2α reproducibly reduced the expression of pri-miR-210 by 45% (Fig. 2C). Consistent with miR-210 being a direct transcriptional target of HIF, overexpression of a stabilized variant of HIF-2α (not recognized by the VHL complex for degradation,27) in a stable subclone of 786-O cells expressing wild type VHL (786-O WT7) induced expression of pri-miR-210 3.37-fold (Fig. 2D). In each experiment, the specificity of the knockdown was verified by measuring the non targeted HIF (see Fig. 2A–D). We also measured the effect of HIF silencing on the expression of mature miR-210 in HCT116 cells by PE-qPCR and found that the copy number was reduced 30% and 45% by silencing of HIF-1α or HIF-1β, respectively, but not HIF-2α (EPAS1) because these cells express primarily HIF-1α (Fig. S1B). Chromatin immunoprecipitation (ChIP) assays (Figs. S4 and S5) indicated that under hypoxic conditions, HIF-1α or HIF-2α binding was enriched at HIF consensus binding sites in the miR-210 promoter19 consistent with both HIFs directly inducing miR-210 transcription. miR-210 overrides hypoxia-induced cell cycle arrest and modulates hypoxia-induced gene expression changes. Hypoxia treatment induces cell cycle arrest at the G1/S transition.10,12,28 We tested whether miR-210 impinged on cell cycle progression during hypoxia treatment of HCT116 Dicerex5 cells HCT116 cells with homozygous disruption of the Dicer helicase domain.29 We chose this cell line because of its dampened background of endogenous microRNAs.29,30 Like the parental HCT116, these cells arrest in response to hypoxia. As shown in Figure 3A, transfection of a miR-210 mimic reduced the fraction of cells in G1 from ~60% to ~21% and increased the number of cells in S phase from 12% to 22% and G2/M from 22% to 46% under hypoxic conditions in comparison to a mock control. This effect was dose dependent and could be observed at a concentration as low as 0.5 nM (Fig. S6). In contrast, synthetic miR-210 duplexes with seed region mutations (miR-210 mt) did not affect the cell cycle profiles (Fig. 3A). Thus,

Cell Cycle

2757


Figure 1. miR-210 is upregulated by hypoxia and its expression correlates with the hypoxic gene expression signature in human tumors. (A) miR-210 expression under normoxia (21% O2) or hypoxic condition (1% O2) (24 hr) in HT29 cells. Expression of ~200 miRNAs was quantified by quantitative RT-PCR. Results are shown in a logarithmic-scale dot plot of copy number per 10 pg of RNA. The full data set is presented in Table S1. (B) miR-210 levels positively correlate with hypoxia upregulated genes in human tumors (n = 29). miR-210 copy number and the levels of genes in the hypoxiainduced gene set in 29 tumors and 28 adjacent normal tissues were expressed as ratios to the expression levels in a pool of normal samples of each tissue type.30 Correlations were calculated between expression ratios of miR-210 and transcripts upregulated by 24 hrs of hypoxia treatment in tissue culture cells. As a control, correlations were also calculated for ~200 random permutations of expression ratios (random transcripts). (C) pri-miR-210 is overexpressed in breast tumor tissues. The level of pri-miR-210 was determined by microarray gene expression profiling in 75 pairs of matched tumor and normal (i.e., adjacent non-involving) samples. The combined p-value indicates the probability that the expression of pri-miR-210 in normal samples is the same as that in tumor samples. The data plotted on the Y-axis are reported as log10 values of expression intensity/common reference. (D) Levels of pri-miR-210 predict breast cancer outcome.55 The Kaplan-Meier curve shows time on the X-axis of the and the Metastasis free probability (probability of the patients not developing metastasis) on the Y-axis. (E) miR-210 copy number in response to hypoxia or normoxia (24 hr unless indicated otherwise) in multiple cell lines.

miR-210 specifically overrides the cell cycle arrest induced by hypoxia. Under normoxic conditions, in which HIFs are de-stabilized and not functional, miR-210 also decreased the fraction of cells in G1 from 45% to 19% and increased G2/M from 32% to 54% 2758

(Fig. 3A). These results suggest that miR-210 may act independently of HIFs to regulate cell cycle progression. To understand how miR-210 functions, we examined gene expression changes upon transfection of miR-210 duplexes by

Cell Cycle

2009; Vol. 8 Issue 17


Figure 2. pri-miR-210 is upregulated by HIF. (A–D) Cells were transfected with HIF siRNAs or infected with a retroviral vector pBABE expressing stabilized HIF-2α. 24 hrs after transfection cells were exposed to hypoxia for another 24 hrs before their gene expression profiles were determined. Significantly upregulated (red) or downregulated (green) genes were determined as described previously.56 The pri-miR-210 (Contig63649_RC) is highlighted with red arrow. HCT116 cells transfected with a HIF-1β siRNA pool (A) or a HIF-1α siRNA pool (B) were compared with Mock transfected cells. (C) RNA isolated from 786-O-pBABE-VHL cells transfected with a HIF-2α siRNA pool were compared with Mock transfected cells. (D) RNA isolated from 786-O WT7 cells infected with a stabilized HIF-2α variant was compared with cells infected with pBABE empty vector. (E and F) miR-210 copy number in response to hypoxia or normoxia (24 hr) in RCC4 and 786-O cells.

www.landesbioscience.com

Cell Cycle

2759


Figure 3. miR-210 regulates cell cycle progression and partially reverses the hypoxia signature. (A) miR-210 overrides hypoxia-induced cell cycle arrest at G1/S. HCT116 Dicerex5 cells were transfected with either miR-210 duplexes, miR-210 containing mismatches at position 5 and 6 (miR-210 mt) or mock treated as a control. 24 hrs post-transfection, cells were exposed to hypoxia or normoxia for another 24 hrs before analysis of the cell cycle distribution. The percentages of cells in G1, S or G2/M phase are shown on each panel. 2N, cells having diploid DNA content; 4N, cells having tetraploid DNA content. (B) miR-210 loss-of-function leads to accumulation of cells in G1. 786-O-pBABE cells were mock-transfected or transfected with anti-miR-210 or anti-miR-185. Cells were treated with nocodazole 30 hrs after transfection and cell cycle distribution was analyzed 16 hrs after Nocodazole treatment. Proliferating cells accumulated in the G2 phase and only cells arrested in G1 or S prior to nocodazole treatment retained those states. This experiment is representative of three performed with consistent results. (C) miR-210-regulated genes under hypoxia are negatively correlated with the hypoxia signature. HCT116 Dicerex5 cells were exposed to hypoxia for 24 hrs or they were transfected with miR-210 24 hrs prior to hypoxia treatment for another 24 hrs. Microarray analysis was performed on two independent experiments to identify common signature genes (p < 0.01). Venn diagrams summarize the significant overlap among the following gene sets: miR-210 upregulated genes and genes downregulated by hypoxia; miR-210 downregulated genes and genes upregulated by hypoxia. The p value for this overlap is shown on the upper right corner of each panel. (D) miR-210 targets are upregulated by anti-miR-210. ME-180 cells were transfected with a luciferase siRNA, miR-210, or anti-miR-210. 24 hours after transfection, the cells were exposed to hypoxia for another 24 hrs before microarray analysis. Shown is a heat map representation of gene expression changes of corresponding to consensus miR-210-downregulated transcripts. The color bar represents log10 expression ratio (samples from treated cells/samples from mock-treated cells) -0.7 (teal) to +0.7 (magenta).

microarray analysis. Transfection of miR-210 partially reversed the hypoxia-induced signature in several cell lines. As shown in Figure 3C, in HCT116 Dicerex5 cells miR-210 upregulated transcripts overlapped significantly with transcripts downregulated by hypoxia (p value