Hypoxia-induced methylation of a pontin chromatin remodeling factor

Jason S. Leea,1, Yunho Kima,1, Jinhyuk Bhinb, Hi-Jai R. Shina, Hye Jin Nama, Seung Hoon Leec, Jong-Bok Yoonc, Olivier Bindad, Or Gozanid, Daehee Hwangb,e, and Sung Hee Baeka,2

aDepartment of Biological Sciences, Creative Research Initiative Center for Chromatin Dynamics, Seoul National University, Seoul 151-742, South Korea; bDepartment of Chemical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 790-784, South Korea; cDepartment of Biochemistry and Translational Research Center for Function Control, Yonsei University, Seoul 120-749, South Korea; dDepartment of Biological Sciences, Stanford University, Stanford, CA 94305; and eSchool of Interdisciplinary Biosciences and Bioengineering, POSTECH, Pohang 790-784, South Korea

Edited by Gregg L. Semenza, Johns Hopkins University School of Medicine, Baltimore, MD, and approved July 12, 2011 (received for review April 19, 2011)

Pontin is a chromatin remodeling factor that possesses both ATPase such as ATP-dependent SWI/SNF as well as histone and DNA helicase activities. Although Pontin is frequently overex- modifying enzymes. Numerous enzymatic activities are associated pressed in human cancers of various types and implicated in onco- with coregulator functions and the activities are regulated by genic functions, the upstream signaling network leading to the posttranslational modifications, including methylation, acetyla- regulation of Pontin that in turn affects transcription of down- tion, phosphorylation, ubiquitylation, and small -like stream target has not been extensively studied. Here, we modifier (SUMO)-ylation (13–16). SUMOylation of Reptin has identify Pontin is methylated by G9a/GLP methyltransferases in been shown to be crucial for transcriptional repression of a KAI1 hypoxic condition and potentiates HIF-1α-mediated activation by metastasis suppressor (17, 18), whereas SUMOylation of increasing the recruitment of p300 coactivator to a subset of Pontin has been shown to function as a transcriptional coactivator HIF-1α target promoters. Intriguingly, Pontin methylation results of androgen receptor-mediated transcription in prostate cancer in the increased invasive and migratory properties by activating (19). Recently, we reported that Reptin chromatin-remodeling downstream target gene, Ets1. In contrast, inhibition of Pontin factor negatively regulates a subset of hypoxia-responsive genes methylation results in the suppression of tumorigenic and meta- (20). Biochemical purification of Reptin-binding proteins identi- BIOCHEMISTRY static properties. Together, our data provide new approaches by fied G9a, and hypoxia-induced Reptin methylation turned out to targeting Pontin methylation and its downstream targets for the participate in downregulating a subset of hypoxia target genes development of therapeutic agents for human cancers. involved in metabolism and tumor development using a gen- ome-wide analysis approach. In this manuscript, we provide evi- ∣ ∣ epigenetics transcriptional regulation covalent nonhistone modification dence that Pontin chromatin-remodeling factor is methylated by G9a and GLP in hypoxic condition. We address a detailed mo- efining the molecular mechanisms that coordinate specific lecular mechanism by which Pontin methylation mediates and Dupstream signal to diverse transcriptional responses remains elaborates the transcriptional regulation, thereby strongly activat- an important goal in biology. The main downstream effect of ing a subset of hypoxia target genes differentially regulated by signaling cascades is the modulation of transcription factors and Pontin compared to those of Reptin. coregulators functioning in the nucleus in response to specific upstream signals (1–4). Oxygen deficiency affects not only phy- Results siological processes such as those involved in embryonic develop- Pontin Is Methylated by G9a and GLP Methyltransferases. To screen ment, wound healing, and inflammation, but also in pathological for enzymes responsible for Pontin methylation, we performed conditions such as tumor progression, ischemic disease, and in vitro methyltransferase assays and found that out of sixteen atherosclerosis (5, 6). Many hypoxic responses are mediated by histone methyltransferases (HMTs), G9a was the only enzyme hypoxia inducible factor 1 (HIF-1), a heterodimeric transcription A α able to methylate Pontin (Fig. 1 ). Because G9a often functions factor that is comprised of an oxygen-regulated subunit (HIF- together with GLP (21, 22), we examined whether G9a and/or 1α or HIF-2α) and a constitutively expressed β subunit (HIF-1β) GLP are responsible for Pontin methylation. Further in vitro (7, 8). Under normoxic conditions, HIF-1α is unstable and subject HMT assays using G9a and GLP revealed that Pontin is methy- to degradation mediated by the von Hippel–Lindau E3 ligase. lated by both enzymes (Fig. 1 B and C). To determine whether However, under hypoxic conditions, HIF-1α is stabilized and it G9a/GLP can methylate Pontin in vivo, we examined Pontin translocates into the nucleus and binds to HIF-1β. The HIF-1α/β heterodimer is then able to bind to the hypoxia response element methylation by performing immunoprecipitation using anti- (HRE) that contains ACGTG as a core sequence. methyl lysine antibody by introducing either G9a or GLP. Pontin Although HIF-1α and HIF-2α share some common targets appears to have very low basal level of methylation, whereas its VEGF GLUT1 methylation status can be dramatically induced by G9a/GLP including and that are involved in regulating D angiogenesis and glycolytic pathway, HIF-2α appears to have its (Fig. 1 ). This was supported by coimmunoprecipitation assays unique targets during embryonic development by regulating factors such as Oct4 (9) and antioxidant enzymes, such as SOD2 Author contributions: J.S.L., Y.K., and S.H.B. designed research; J.S.L., Y.K., J.B., H.-J.R.S., (10). Both isoforms of HIFα interact with Sirt1 and the transcrip- H.J.N., S.H.L., and O.B. performed research; J.S.L., J.B., J.-B.Y., O.G., D.H., and S.H.B. tional activity of HIF-2α is enhanced upon deacetylation by Sirt1 analyzed data; and J.S.L., Y.K., and S.H.B. wrote the paper. whereas the transcriptional activity of HIF-1α is repressed by The authors declare no conflict of interest. deacetylation (11, 12). Therefore, although the two isoforms of This article is a PNAS Direct Submission. HIFα both execute hypoxic response, it appears that they have Data deposition: The data reported in this paper have been deposited in the Gene Expression distinct functions through differential regulatory mechanisms. Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE27813). Gene expression is not only influenced by presence of tran- 1J.S.L. and Y.K. contributed equally to this work. scription factors, but also by chromatin structure regulated by 2To whom correspondence should be addressed: E-mail: [email protected]. chromatin modifiers. The transcription of most genes is regulated This article contains supporting information online at www.pnas.org/lookup/suppl/ by the coordinate action of chromatin-remodeling complexes doi:10.1073/pnas.1106106108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1106106108 PNAS Early Edition ∣ 1of6 Downloaded by guest on September 28, 2021 A Autoradiogram of in vitro HMTase assay histone H3K9 suggesting that there exist a specificity amongst several HMTs in response to hypoxia in upregulating their activity D2 (Fig. 2A). We then examined whether Pontin methylation is HMTs: ET Mr(K) G9a SET8 EZH2 ASH1 NSD3 NSD1(SET) SETD5 S SETD4 PRDM2 PRDM4 ZNF298 SETDB2 PRMD14 SUV420H2 SUV39H2 100 induced in hypoxic condition, and found that Pontin methylation Pontin-me * 75 continued to increase with hypoxic exposure up to at least 12 h B C with corresponding accumulation of G9a and GLP in MCF7 breast cancer cells and HEK293 cells (Fig. 2B and Fig. S1).

M (K) M (K) To further establish that the increase in Pontin methylation is r r directly caused by the increase in G9a and GLP enzymatic activ- 91 91 Pontin-me Pontin-me # ity, we exposed cells to hypoxia in the presence and absence of * * 39 39 G9a and GLP inhibitor, BIX-01294 (23, 24). BIX-01294 signifi- Coomassie Autorad. Coomassie Autorad. cantly reduced hypoxia-mediated Pontin methylation further P G a demonstrating that the methyltransferase activity of G9a/GLP D E g I F GL Pontin G9 IgG Mock GLP α IP: α C α IP: α is required for Pontin methylation (Fig. 2 ). Dependency of Pon- α α IP: Kme Pontin-me αG9a Pontin tin methylation on the expression of G9a was also demonstrated Pontin α αG9a Pontin in G9a-deficient mouse embryonic fibroblasts (MEFs) (Fig. 2D). WCL WCL WCL GLP α αGLP G9a Pontin These data strongly suggest that G9a and GLP are required for hypoxia-mediated Pontin methylation. To identify specific methy- Fig. 1. G9a and GLP methylate Pontin. (A) Histone lysine methyltransferases lation sites on Pontin, we performed in vivo methyltransferase were prepared as GST-fusion proteins and incubated with GST-Pontin in the 3 assays on Pontin deletion mutants and found that a region con- presence of S-adenosyl-L-[methyl- H] methionine (SAM). Reaction products taining the amino acid residues between 251 and 350, was methy- were analyzed by autoradiogram. (B and C) In vitro methylation assay was E performed using GST-G9a SET domain (B) or GST-GLP SET domain (C) with lated (Fig. 2 ). Within this region, there were six lysine residues, Coomassie brilliant blue staining (Left). Asterisks (*) indicate nonspecific and we mutated each one to alanine, however, these single bands, and the hash mark (#) indicates automethylation. (D) HEK293 cells mutants (K265A, K267A, K268A, K274A, K281A, and K285A) were transfected with empty vector or with expression vector encoding G9a still retained some methylation (Fig. S2). Therefore we eventually or GLP. Cell lysates were immunoprecipitated with antibody against anti- had to mutate all 6 lysine residues to alanine to completely elim- methyl-lysine antibody followed by immunoblot analysis with anti-Pontin inate the residual methylation. We found that Pontin mutant antibody to detect methylated Pontin. (E and F) Coimmunoprecipitation that harbors lysine to alanine mutation (KA) did not show any assay of endogenous Pontin with G9a (E) or GLP (F). increase in methylation whereas Pontin WT showed significant increase in methylation by hypoxia (Fig. 2F). that showed strong Pontin interaction with both G9a and GLP at Next, to examine whether Pontin methylation is dynamic, E F the endogenous level (Fig. 1 and ). we performed in vivo methylation assay in normoxic or hypoxic condition as well as a dynamic normoxia-hypoxia-normoxia con- Pontin Methylation Is Induced by Hypoxia. We then asked whether dition. Immunoprecipitation assay with antimethyl lysine anti- Pontin methylation can be induced by hypoxia as we have re- body showed that Pontin methylation was induced by hypoxia, ported previously that Reptin methylation is induced by hypoxia and this methylation returned to its basal levels when the cells as a result of increased G9a protein level and corresponding in- were returned to normoxic condition (Fig. 2G). The extent of crease in its methyltransferase activity (20). Indeed, G9a protein Pontin methylation closely followed G9a protein levels as it was levels were increased in both HeLa and MCF7 cells exposed to elevated following hypoxia and returned to basal levels once the 6 h of hypoxia (1% O2) whereas there was no change in the levels cells were again exposed to normoxic condition. In an attempt to of Suv39h1; another histone methyltransferase that targets determine the functional link between Pontin methylation and

ABMCF7 C D HeLa MCF7 MCF7 1% O2 (h) 01 612 MEFs WT G9a-/- 1% O (6 h) - 1% O2 (h) 06 06 IP: αKme Pontin-me 2 ++ BIX-01294 --+ 1% O2 (6 h) - + - + αG9a α Pontin IP: αKme Pontin-me IP: αKme Pontin-me αGLP αHIF-1α αPontin αPontin WCL αSuv39h1 αGLP WCL αHIF-1α WCL αHIF-1α αTubulin αG9a αG9a αG9a E F G H Hypoxia 350 X X X -250 - Pontin WT KA M M PX N HP H N WT 15 1 251 351-456 GFP Pontin 1-150 1% O2 (6 h) - + -+ α IP:αKme Pontin-me Pontin IgG IP: Kme Pontin-me IP: α 1% O2 (6h)-+-+ -+-+-+ αPontin αPontin αHIF-1α IP:αKme Pontin-me α α αGFP Input αHIF-1α WCL αHIF-1α HIF-1 Input WCL α αHIF-1α αG9a αG9a Pontin

Fig. 2. Hypoxia-induced Pontin methylation is mediated by G9a and GLP. (A) Protein expression levels of G9a, GLP, and Suv39h1 were examined by immuno- blotting in HeLa and MCF7 cells exposed to normoxic and hypoxic conditions as indicated. (B and C) Pontin methylation was examined by immunoprecipitation of cell lysates from MCF7 cells exposed to hypoxia with anti-methyl-lysine antibody followed by immunoblotting with anti-Pontin, HIF-1α, G9a, and GLP anti- bodies for the indicated times (B) or in the presence and absence of an inhibitor of G9a and GLP, BIX-01294 (C). (D) Requirement of G9a for hypoxia-induced Pontin methylation was examined in WT and G9a-deficient MEFs exposed to normoxic and hypoxic condition. (E) HEK293 cells were cotransfected with plasmids encoding each GFP-tagged Pontin deletion constructs spanning the indicated amino acid residues exposed to normoxia or hypoxia for 6 h. Whole cell extracts were immunoprecipitated with antimethyl-lysine antibody followed by immunoblot assay using anti-GFP antibody to detect methylated Pontin. (F) Pontin methylation was examined as in (E) using constructs encoding Pontin WT or KA mutant exposed to normoxia or hypoxia for 6 h. (G) Dynamics of Pontin methylation was examined in 293HEK cells exposed to either normoxia, hypoxia, or hypoxia and normoxia and Pontin methylation was examined as in (E). (H) Endogenous interaction between Pontin and HIF-1α in MCF7 cells exposed to hypoxia using either anti-HIF-1α or normal IgG.

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1106106108 Lee et al. Downloaded by guest on September 28, 2021 hypoxic response, we examined the physical association between into several clusters (Fig. 3 E and F and Table S1). These clusters HIF-1α and Pontin at the endogenous expression level, and found represented mainly into Pontin-independent group (clusters 1 that Pontin is able to readily interact with HIF-1α during hypoxic and 2) and Pontin-dependent group (clusters 3 and 4) that condition (Fig. 2H). Together, these data indicate that both lost sensitivity to hypoxic responsiveness as a result of Pontin Pontin methylation and the interaction of Pontin with HIF-1α are knockdown. induced in hypoxic condition. Given that Pontin acts as an activator, we asked the question whether hypoxia and Pontin expression can synergistically act on Identification of Pontin-Dependent Target Genes by Microarray Ana- hypoxia-responsive genes. To this end, we examined cluster 3 to lysis. As Pontin is able to interact with HIF-1α, we hypothesized determine which targets are enriched. Upon that Pontin might influence the transcriptional activity of HIF-1α. searching for genes that are affected by transcription factors that To this end, we attempted to determine the transcriptional role of have been previously reported to modulate transcription upon Pontin methylation on the expression of hypoxia-responsive hypoxic stress such as HIF-1, NF-κB, and TCF/LEF (25), HIF- 3× genes. We first performed a reporter assay using HRE-lucifer- 1 scored the highest in terms of the number of its target genes ase and found that Pontin knockdown led to a reduction of hy- enriched in Pontin-dependent cluster 3 (from both 6 and 9 h of 3× A poxia-mediated increase in HRE-luciferase activity (Fig. 3 ). hypoxia) (Fig. 3G). In cluster 3, six out of 36 genes in 6 h dataset 3× Upon expressing Pontin WT, the activation of HRE-luciferase and four out of 34 genes in 9 h dataset were known HIF-1 target, activity was significantly increased whereas Pontin KA mutant and the promoters of 38 genes contained putative hypoxia re- B failed to have any effect (Fig. 3 ). We then performed a micro- sponse elements (Fig. 3H and Tables S1 and S2). These analyses array analysis from RNAs isolated from MCF7 cells expressing strongly support the idea that Pontin-dependent clusters con- either control shRNA (shNS) or Pontin shRNA (shPontin) in tained many HIF-1α targets (cluster 3). Intriguingly, genome- normoxic and hypoxic conditions to determine the effect of wide analysis of hypoxia-induced Pontin-dependent HIF-1α Pontin on the expression of hypoxic responsive genes across the target genes revealed that Pontin-dependent target genes do not whole genome (Fig. 3C). We used a Gaussian curve fitting to de- generally overlap with Reptin-dependent target genes upon termine the applicable cut-offs in determining genes that showed hypoxia (20) (Fig. S3 and Table S2). significant change in response to hypoxia. Approximately 23.5% of the differentially expressed genes were affected by Pontin Pontin Methylation Is Required for Transcriptional Activation of a Sub- knockdown, indicating that these were in fact regulated both D set of Hypoxia-Responsive Genes. To validate Pontin-dependent

by hypoxia and Pontin (Fig. 3 ). Hierarchical clustering was then BIOCHEMISTRY target genes identified from our microarray analysis (Fig. 3H performed and hypoxia-responsive genes could be categorized and Tables S1 and S2), we performed a quantitative RT-PCR ana- lysis on both Pontin-dependent and -independent genes. Pontin A B C Tot al array D 3X HRE luciferase 3X HRE luciferase knockdown led to an inhibition of hypoxia-mediated activation of 8 15 (48,324 probes/30,500 genes) Pontin-independent (76.5%) Ets1, KDM4B, and IGFBP3 transcripts (Pontin-dependent) 6 10 Hypoxia responsive BNIP3L HK2 WSB1 4 (721 probes/ 635 genes) whereas no effect was observed for , , and 5 2 transcripts (Pontin-independent) (Fig. 4 A and B). Further, intro- Fold change Fold change Pontin Pontin Pontin-dependent R 0 0 duction of shRNA-resistant Pontin WT (Pontin WT ) had an Mock WT KA -independent -dependent (23.5 %) shNS shPontin activating potential whereas shRNA-resistant Pontin KA mutant shNS shPontin shNS shPontin R E H ETS1* (Pontin KA ) appeared to have none. This activating function TMEM128 ZDBF2 QKI does not appear to be MCF7-specific event, as we have also ob-

1 ERGIC1 FLJ23356 served regulation of Ets1 transcript levels in HeLa cells (Fig. S4). Cluster NUCKS1* C17orf95 IFP38 Together, these data provide evidence that Pontin methylation CARS HIST2H2AC 2

) C9orf9 participates in hypoxia-driven activation of a subset of hypoxia

Pontin-independent HLXB9 Cluster

6h SESN2* shNS shPontin RPSAP47 target genes involved in oncogenesis and cell survival. F RPL32P11 CYP4F3 To gain further insight into how Pontin modulates expression er 3 (

t PLAC8 3 LOC100144603 RPL15P22 of a subset of HIF-1 target genes during hypoxic condition, we Cluster Clus LOC729742 TRIM66 FAM53C performed ChIP assays on Pontin-dependent and independent GADD45B*

4 ADAM15 promoters in MCF7 cells expressing either control shRNA KIFC2

Cluster HIST2H4B*

Pontin-dependent HIST1H2BK (shNS) or Pontin shRNA (shPontin). Hypoxia resulted in in- ZDHHC9* TMEM19 Ets1 G Transcription factor enrichment (Cluster 3) NAE1 creased recruitment of Pontin on promoter by binding to CCNT1 HIF-1 FAM158A HIF-1α that is recruited to its response element whereas knock- NFκB SRD5A1 NUDCD2 TCF/LEF SLC39A7 down of Pontin resulted in reduced level of Pontin recruitment 0123 -2 -1 0 1 2 without affecting HIF-1α recruitment (Fig. 4C). The recruitment Z score Fold log 2 of p300 to Ets1 promoter appeared to be HIF-1α and Pontin- Fig. 3. Pontin-dependent target identification by microarray analysis. dependent as knockdown of Pontin caused a marked decrease (A and B) Three × HRE-luciferase reporter assay with knockdown of Pontin in p300 recruitment. However, upon examining Pontin-indepen- (A) or overexpression with Pontin WT and KA mutant (B) under normoxia or dent target promoter; BNIP3L, Pontin recruitment was not hypoxia for 6 h. Values are expressed as mean SEM of three independent observed nor the facilitation of p300 recruitment to the promoter experiments. (C) Flow chart showing the strategy of cDNA microarray analysis (Fig. 4C). Similarly, exogenously introduced Pontin WT was and Pontin-dependent gene selection process. (D) Pontin-dependent genes Ets1 represent about 23.5% of all differentially expressed genes by hypoxia. able to enhance recruitment of p300 to promoter whereas (E and F) Identification of Pontin-independent genes (E) and Pontin-depen- Pontin KA mutant failed to recruit p300 to the promoter, suggest- dent up-regulated (cluster 3, red) and down-regulated (cluster 4, green) ing that Pontin methylation is crucial for the increased p300 lo- genes (F) by hierarchical clustering and comparing fold change of hypox- calization with HIF-1α (Fig. 4D). Further, to determine whether ia-induced genes from cells expressing shNS and shPontin. (G) Known target the recruitment of Pontin and p300 to Ets1 promoter occurs in κ genes for HIF-1, NF B and TCF/LEF were collected and the enrichment of HIF-1α-dependent manner, we performed ChIP assays in these genes was analyzed for each cluster is shown as a Z score within each HIF-1α-deficient MEFs. The recruitment of Pontin and p300 cluster. (H) Heatmap diagram of Pontin-dependent genes that are activated Ets1 α by hypoxia (6 h) showing gene names with known HIF-1α target genes to promoter was only observed in the presence of HIF-1 marked with an asterisk. Also see Table S1 for Pontin-dependent target genes (WT MEFs) whereas no significant recruitment was observed in −∕− at 9 h of hypoxia. HIF-1α-deficient (HIF-1α ) MEFs (Fig. 4E).

Lee et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on September 28, 2021 A Pontin-dependent genes 4 3 60 Ets1 KDM4B IGFBP3 3 2 40 NMX 2 HPX 1 1 20 Fold change Fold change Fold change 0 0 0 -- WTR KAR -- WTR KAR -- WTR KAR shNS shPon shNS shPon shNS shPon Pontin-independent genes B 8 4 3 BNIP3L HK2 WSB1 3 6 2 NMX 4 2 HPX 2 1 1 Fold change Fold change 0 0 Fold change 0 Fig. 4. Pontin methylation potentiates HIF-1α transcrip- -- R R --- R R - R R WT KA WT KA WT KA tional activity by enhancing interaction with p300. (A and shNS shPon shNS shPon shNS shPon B) Quantitative RT-PCR analysis of Pontin-dependent (A) and Pontin-independent (B) hypoxia target gene expres- C D E Ets1 promoter Promoter: Ets1 BNIP3L sions identified (both 6 and 9 h). Results are expressed as Promoter: Ets1 BNIP3L MEFs: WT HIF-1α -/- shNS shPon shNS shPon Flag-Pontin WTR KAR WTR KAR relative mRNA levels compared to shNS under normoxic 1% O2(hr) 0 606 1% O2(6 h) - + - + - + - + 1% O2 (6 h) - + - + - + - + condition. Values are expressed as mean SEM of three in- α αPontin αPontin Flag dependent experiments. (C and D) ChIP assays on the Ets1 α α α αp300 Pontin HIF-1 and BNIP3L promoters in MCF7 cells with Pontin knockdown αHIF-1α αp300 αp300 by shPontin (C) and reconstituted with exogenous shRNA- R R αpol II αHIF-1α αpol II resistant Pontin WT and KA mutant (D) upon hypoxia. Promoter occupancy of proteins indicated was analyzed. Input αpol II αIgG Input (E) ChIP analysis performed in WT and HIF-1α-deficient Pontin-dep. Pontin-indep. Input −∕− Pontin-dep. Pontin-indep. (HIF-1α ) MEFs examining the recruitment of various proteins to Ets1 promoter as indicated. (F) Endogenous in- FHG I IP: αp300 teraction between Pontin and p300 in MCF7 cells exposed IP: αp300 IP: αp300 IP: αp300 GFP Pontin to either normoxia or hypoxia using anti-p300 antibody. 1% O (6 h) - + GFP Pontin WT KA GFP Pontin WT KA -/- 2 MEFs WT HIF-1 (G and H) MCF7 cells were transfected with either Pontin 1% O 2 (6 h) -+ -+ αPontin 1% O2 (6 h) -+ -+ 1% O (6 h) -+ -+ αGFP 2 WT or KA mutant and immunoprecipitation assay was αHIF-1α α GFP αHIF-1α αHIF-1α performed using anti-p300 antibody. (I) Pontin WT or KA αHIF-1α αHIF-1 α WCL αPontin Input αGFP mutant was immunoprecipitated with anti-p300 antibody Input αGFP Input α GFP −∕− α in WT or HIF-1α MEFs followed by immunoblotting using αp300 p300 α α p300 p300 antibodies indicated.

To further investigate how methylated Pontin is involved in cells, but also the invasive potential during hypoxia as knockdown transcriptional activation, we tested the possibility that Pontin of Pontin greatly impaired invasion of these cells through Matri- shows altered interaction with other coactivators that may affect gel (Fig. 5C). Similarly, invasion through Matrigel by MCF7 cells HIF-1α transcriptional activity. As p300 conducts a critical role was markedly increased by introducing Pontin WT, whereas in HIF-1α-mediated transcriptional activation, we first examined Pontin KA did not appear to confer increase in invasive potential whether methylated Pontin shows increased interaction with (Fig. 5D). Together, these data confirmed that either Pontin p300. Coimmunoprecipitation assays revealed that Pontin's abil- knockdown or Pontin KA mutant overexpression significantly ity to interact with p300 was enhanced in hypoxic condition at the inhibited the invasive potential of MCF7 cells. endogenous level (Fig. 4F). This increase in p300 interaction was Pontin methylation-dependent as Pontin KA mutant did not show Ets1, which Is Regulated by Pontin Methylation, Is Responsible for significant difference in response to hypoxia (Fig. 4G). Moreover, Increased Cell Motility. As part of examining other processes affect- Pontin WT, but not Pontin KA mutant, was able to facilitate the ing metastatic potential of cancer cells, we examined the effect of interaction between HIF-1α and p300 (Fig. 4H). Pontin possesses hypoxia-mediated Pontin methylation on cell motility by perform- its ability to interact with p300 directly without HIF-1α,asit ing scratch-mobility assay. Over 36–48 h, control shRNA trans- shows increased binding with p300 in HIF-1α-deficient MEFs fected cells moved significantly into the scratch, whereas Pontin in a hypoxia-dependent manner (Fig. 4I). Collectively, these data knockdown inhibited this movement (Fig. 5E). Adding back indicate that Pontin methylation has a role in enhanced binding shRNA-resistant Pontin WT to these Pontin- knockdown cells re- of p300 and thereby able to increase transcriptional activity of sulted in rescuing Pontin knockdown phenotype, whereas intro- HIF-1α. ducing Pontin KA mutant failed to do so. These data strongly suggest that Pontin methylation is indeed important for confer- Pontin Methylation Enhances the Proliferative and Invasive Potential ring tumorigenic and metastatic potential of MCF7 cells. Because of Breast Cancer Cells. As many genes included in Pontin-depen- Ets1 was obtained from microarray analysis as a Pontin-depen- dent group contained genes that play a major role in prolifera- dent target gene that is responsible for cell motility and invasion tion, cell cycle, cell motility and invasion, we performed cell- (Fig. 3H), we decided to establish the contribution of Ets1 expres- based analyses to determine whether Pontin methylation plays sion to cell motility. We then performed a scratch-mobility assay an important role in regulating these processes. First, prolifera- using MCF7 cells stably expressing Pontin WTR or Pontin KAR tion of MCF7 cells was inhibited upon Pontin knockdown sug- with either control siRNA or Ets1 siRNA. Pontin-dependent in- gesting that Pontin expression participates in upregulating crease in cell motility was significantly decreased by Ets1 knock- proliferative potential during hypoxia (Fig. 5A). Consistently, down, demonstrating that Ets1 is required for Pontin-mediated ectopic expression of Pontin WTenhanced proliferation whereas increase in cell motility (Fig. 5F). Knockdown of Ets1 had little expression of Pontin KA mutant lost its ability to increase prolif- or no effect on cell motility of MCF7 cells stably expressing Pon- eration (Fig. 5B). We also performed cell invasion assays in hy- tin KAR. Collectively, our findings indicate that hypoxia-induced poxic condition to determine the invasive potential of MCF7 cells Pontin methylation is responsible for the regulation of a subset of expressing shPontin or shNS. Invasion assays revealed that Pontin HIF-1α target genes exemplified by Ets1 and thereby affects cell expression was not only important for proliferation of MCF7 proliferation, tumor growth and invasive properties.

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1106106108 Lee et al. Downloaded by guest on September 28, 2021 ) Proliferation assay ) Proliferation assay 3 A 8 B 3 10 shNS Mock 6 shPontin 8 Pontin WT 6 Pontin KA 4 4 2 2 Cell number ( x10 0 Cell number (0 x10 1246 1246 Time (days) Time (days) C Invasion assay D Invasion assay NMX HPX NMX HPX 40 25 NMX NMX Mock shNS 30 HPX 20 HPX Fig. 5. Pontin methylation increases tumorigenic proper- er 15 ties via Ets1. (A and B) Proliferation was monitored over 20 Pontin shPon WT 10 six days in hypoxic condition for MCF7 cells expressing shNS 10 lnumb

el or shPontin (A) and expression of either Pontin WT or KA Cell number Pontin C 5 mutant (B). (C and D) Photomicrographs (40×) from Matri- 0 KA shNS shPon 0 gel-coated transwell invasion assay of MCF7 cells expressing Motility assay Mock Pontin Pontin E WT KA shNS and shPontin (C) and Pontin WT or Pontin KA (D)in 0h 36h 48h G normoxic and hypoxic conditions for 24 h. Bar graph shows shNS 10 mean number of cells per filter and p value is shown from 8 student’s t-test analysis. Error bars represent SEM; n ¼ 5 shPon hyperactivation (Right). (E and F) Photomicrographs from a scratch-mobility 6 * ance (mm) *

st assay of MCF7 cells expressing either shNS or shPontin and shPon 4 shNS di R R R shPon reconstituted with shRNA-resistant Pontin WT or KA mu- +WT R ap 2 shPon+ WT R R G shPon+ KAR tant (E) or Pontin WT or KA mutant either transfected shPon 0 Pontin-dependent HIF-1 target +KAR 0 20 40 60 with control siRNA or Ets1 siRNA (F) in the hypoxia for Time (h) F 12 the indicated times (Left). Gap distance between the two shPon+WTR migrating fronts is measured and shown as a graph (Right). 10 sicont. Asterisk (*) is indicated for p < 0.001 by two-tailed students (mm) 8 shPon+WTR t-test. (G) Proposed model of transcriptional activation role

siEts1 ance 6 for methylated Pontin during hypoxia. Pontin is methylated * BIOCHEMISTRY shPon+KAR 4 WTR+sicont. by G9a/GLP by hypoxia and enhances the recruitment of sicont. WTR+siEts1 R Gap dist 2 KA +siEts1 p300 to a subset of hypoxia target promoters, thereby R KAR+sicont. α shPon+KA 0 potentiating the transcriptional activity of HIF-1 leading siEts1 0204060 to increase in hypoxia-mediated cellular migration, prolif- Time (h) eration, and invasion.

Discussion these enzymes. Indeed, Pontin methylation is dynamic; the extent In this study, we have identified Pontin as a nonhistone substrate of Pontin methylation was elevated following hypoxia and re- methylated by hypoxia-induced G9a and GLP, leading to hyper- turned to basal level once the cells are again exposed to normoxic activation of a subset of hypoxia target genes by enhanced binding condition. It can be speculated that certain demethylases may of p300 to HIF-1α and thereby increasing transcriptional activity function to oppose Pontin methylation for a dynamic on-off of HIF-1α (Fig. 5G). Although Pontin and Reptin share high switch for a subset of hypoxia-responsive genes. Further, hypoxia- structural homology, they have distinct functions in regulating induced Pontin methylation affects binding to HIF-1α and p300, their specific target genes as a coactivator and as a corepressor, and the association between methylated Pontin with HIF-1α or respectively (17–19). We found that this type of antagonistic reg- p300 appeared to be DNA-independent as treatment of ethidium ulation is also applied to the regulation of hypoxia target genes bromide did not alter their binding (Fig. S6). as well as the regulation of well-established Wnt target genes Together, our finding that G9a- and GLP-dependent Pontin (20, 26). Hypoxia-induced G9a is responsible for methylation methylation regulates a subset of hypoxia target genes represents α of both Pontin and Reptin, but the functional outcome appears a coordinate signaling pathway by which HIF-1 transcriptional to be in the opposite direction. G9a-mediated Reptin methylation activity can be modulated, thereby affecting cell migration, tumor α growth, and cell survival pathway that are important for the pro- exhibits high affinity for HIF-1 transcription factor as well as Ets1 for HDAC1 corepressor, functioning as a negative regulator gression of cancer. , which was selected as a Pontin-dependent on a subset of hypoxia target genes (20). However, G9a/GLP- target gene from microarray analysis, turned out to be important for exerting its effect on cell motility and growth regulation by mediated Pontin methylation exhibits enhanced binding to p300 Pontin methylation upon hypoxia. Our data shed light on the me- coactivator, and contributes to further activation of a subset of chanism of Pontin’s potential regulatory role in tumor progression hypoxia target genes. It appears that Pontin is recruited to the upon hypoxia and suggest a possibility of developing therapeutic target promoter at an early time point for transcriptional activa- α agents that target Pontin methylation, providing an avenue for tion of HIF-1 target genes, compared to Reptin recruited on powerful anticancer therapeutic approaches in the future. HIF-1α target promoters at a later time point for negative reg- ulation. Because the ATPase activity of Pontin is important in ex- Methods erting chromatin-remodeling function, we considered whether Antibodies. The following commercially available antibodies were used: anti- the ATPase activity is involved in regulating its methylation- Pontin, anti-GFP, anti-p300, anti-HDAC1, and anti-HIF-1α (Santa Cruz Biotech- dependent hypoxia targets. We found that there was no signifi- nology), anti-G9a, and anti-Suv39h1 (Upstate Biotechnology), antimethyl cant difference between Pontin WTand ATPase mutant in terms lysine and anti-GLP (Abcam), anti-FLAG (Sigma), and anti-RNA Polymerase of its methylation, transcriptional activity and recruitment to Ets1 II (Berkeley Antibody Company). promoter (Fig. S5) suggesting that Pontin’s ATPase activity is not In Vitro Methyltransferase Assay. All recombinant proteins (GST, GST-Pontin, involved in regulating the hypoxia targets identified in this study. GST-G9a SET, and GST-GLP SET) were purified from Escherichia coli. In vitro Because methylation and demethylation processes are dy- methyltransferase assays were performed with the reactions assembled in namic and thus the methylation status of nonhistone proteins as 5× lysine methyltransferase buffer including 3H-S-adenosylmethionine and well as histones is likely to be determined by the net effect of purified proteins and incubated overnight at 37 °C as described in (20).

Lee et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on September 28, 2021 Laemmli buffer was added to samples, and loaded on SDS-PAGE for autora- ACKNOWLEDGMENTS. We thank G. Semenza for providing HIF-1α null MEFs, diography. A. Tarakhovsky for G9a null MEFs, and Boehringer Ingleheim for G9a inhibi- tor. This work was supported by Creative Research Initiatives Program Quantitative Real-Time RT-PCR and ChIP Assays. Quantitative RT-PCR and (Research Center for Chromatin Dynamics, 2009-0081563) (S.H.B.), the Con- ChIP assays were conducted as previously described (20). Also see Table S3 verging Research Center Program (2010K001298) (D.H.), the Basic Science for primer sequences used. Research Program (3344-20100053) (J.S.L.), the National Junior Research Fellowship (NRF-2011-A01496-0001806) (H-J.R.S.), and Brain Korea 21 Statistical Analysis. Statistical differences in test and control samples were Fellowship (J.S.L., Y.K., and H.J.N.) from the National Research Foundation determined by Student’s t test or ANOVA using the Statview package (NRF) Grant funded by the Ministry of Education, Science, and Technology (Abacus Concepts, Inc.). (MEST) of Korea.

1. Semenza GL, Wang GL (1992) A nuclear factor induced by hypoxia via de novo protein 14. Yang XJ, Seto E (2008) Lysine acetylation: Codified crosstalk with other posttransla- synthesis binds to the human erythropoietin gene enhancer at a site required for tional modifications. Mol Cell 31:449–461. transcriptional activation. Mol Cell Biol 12:5447–5454. 15. Han SJ, Lonard DM, O’Malley BW (2009) Multi-modulation of nuclear receptor coac- 2. Karin M, Lawrence T, Nizet V (2006) Innate immunity gone awry: Linking microbial tivators through posttranslational modifications. Trends Endocrinol Metab 20:8–15. infections to chronic inflammation and cancer. Cell 124:823–835. 16. Baek SH (2011) When signaling kinases meet histones and histone modifiers in the 3. Kastan MB, et al. (1992) A mammalian cell cycle checkpoint pathway utilizing p53 and nucleus. Mol Cell 42:274–284. GADD45 is defective in ataxia-telangiectasia. Cell 71:587–597. 17. Kim JH, et al. (2005) Transcriptional regulation of a metastasis suppressor gene by 4. Behrens J, et al. (1996) Functional interaction of beta-catenin with the transcription Tip60 and beta-catenin complexes. Nature 434:921–926. factor LEF-1. Nature 382:638–642. 18. Kim JH, et al. (2006) Roles of sumoylation of a reptin chromatin-remodelling complex 5. Pugh CW, Ratcliffe PJ (2003) Regulation of angiogenesis by hypoxia: role of the HIF in cancer metastasis. Nat Cell Biol 8:631–639. system. Nat Med 9:677–684. 19. Kim JH, et al. (2007) SUMOylation of pontin chromatin-remodeling complex reveals 6. Semenza GL (2003) Targeting HIF-1 for cancer therapy. Nat Rev Cancer 3:721–732. a signal integration code in prostate cancer cells. Proc Natl Acad Sci USA 7. Wang GL, Semenza GL (1993) Characterization of hypoxia-inducible factor 1 and reg- 104:20793–20798. ulation of DNA binding activity by hypoxia. J Biol Chem 268:21513–21518. 20. Lee JS, et al. (2010) Negative regulation of hypoxic responses via induced Reptin 8. Ema M, et al. (1997) A novel bHLH-PAS factor with close sequence similarity to hypox- methylation. Mol Cell 39:71–85. ia-inducible factor 1alpha regulates the VEGF expression and is potentially involved in 21. Collins RE, et al. (2008) The ankyrin repeats of G9a and GLP histone methyltransferases lung and vascular development. Proc Natl Acad Sci USA 94:4273–4278. are mono- and dimethyllysine binding modules. Nat Struct Mol Biol 15:245–250. 9. Simon MC, Keith B (2008) The role of oxygen availability in embryonic development 22. Tachibana M, Matsumura Y, Fukuda M, Kimura H, Shinkai Y (2008) G9a/GLP complexes and stem cell function. Nat Rev Mol Cell Biol 9:285–296. independently mediate H3K9 and DNA methylation to silence transcription. EMBO J 10. Scortegagna M, et al. (2003) Multiple organ pathology, metabolic abnormalities, 27:2681–2690. and impaired homeostasis of reactive oxygen species in Epas1-/- mice. Nat Genet 23. Kubicek S, et al. (2007) Reversal of H3K9me2 by a small-molecule inhibitor for the G9a 35:331–340. histone methyltransferase. Mol Cell 25:473–481. 11. Dioum EM, et al. (2009) Regulation of hypoxia-inducible factor 2alpha signaling by the 24. Chang Y, et al. (2009) Structural basis for G9a-like protein lysine methyltransferase stress-responsive deacetylase sirtuin 1. Science 324:1289–1293. inhibition by BIX-01294. Nat Struct Mol Biol 16:312–317. 12. Lim JH, et al. (2010) Sirtuin 1 modulates cellular responses to hypoxia by deacetylating 25. Cummins EP, Taylor CT (2005) Hypoxia-responsive transcription factors. Pflugers Arch hypoxia-inducible factor 1alpha. Mol Cell 38:864–878. 450:363–371. 13. Wu SC, Zhang Y (2009) Minireview: Role of protein methylation and demethylation in 26. McDonald SL, Silver A (2009) The opposing roles of Wnt-5a in cancer. Br J Cancer nuclear hormone signaling. Mol Endocrinol 23:1323–1334. 101:209–214.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1106106108 Lee et al. Downloaded by guest on September 28, 2021