Hypoxia Regulates Alternative Splicing of HIF and Non-HIF Target Genes

Published OnlineFirst May 21, 2014; DOI: 10.1158/1541-7786.MCR-14-0149

Molecular Cancer Chromatin, Gene, and RNA Regulation Research

Hypoxia Regulates Alternative Splicing of HIF and non-HIF Target Genes

Johnny A. Sena1, Liyi Wang2, Lynn E. Heasley2, and Cheng-Jun Hu1,2

Abstract Hypoxia is a common characteristic of many solid tumors. The hypoxic microenvironment stabilizes hypoxia- inducible transcription factor 1a (HIF1a) and 2a (HIF2a/EPAS1) to activate gene transcription, which promotes tumor cell survival. The majority of human genes are alternatively spliced, producing RNA isoforms that code for functionally distinct proteins. Thus, an effective hypoxia response requires increased HIF target gene expression as well as proper RNA splicing of these HIF-dependent transcripts. However, it is unclear if and how hypoxia regulates RNA splicing of HIF targets. This study determined the effects of hypoxia on alternative splicing (AS) of HIF and non-HIF target genes in hepatocellular carcinoma cells and characterized the role of HIF in regulating AS of HIF- induced genes. The results indicate that hypoxia generally promotes exon inclusion for hypoxia-induced, but reduces exon inclusion for hypoxia-reduced genes. Mechanistically, HIF activity, but not hypoxia per se is found to be necessary and sufficient to increase exon inclusion of several HIF targets, including pyruvate dehydrogenase kinase 1 (PDK1). PDK1 splicing reporters confirm that transcriptional activation by HIF is sufficient to increase exon inclusion of PDK1 splicing reporter. In contrast, transcriptional activation of a PDK1 minigene by other transcription factors in the absence of endogenous HIF target gene activation fails to alter PDK1 RNA splicing.

Implications: This study demonstrates a novel function of HIF in regulating RNA splicing of HIF target genes. Mol Cancer Res; 12(9); 1233–43. 2014 AACR.

Introduction forms expression will be more informative than quantifying Hypoxia is a common characteristic of many solid tumors. total transcripts to appreciate the functional consequence of The hypoxic microenvironment stabilizes hypoxia-inducible gene regulation. transcription factor 1a (HIF1a) and 2a (HIF2a) that are Exon arrays contain probe sets interrogating every exon normally degraded under normoxia. The stabilized HIF1a of the transcriptome, thereby permitting the distinguish- and HIF2a proteins translocate to the nucleus, where they ing of different isoforms of a gene as well as measuring dimerize with aryl hydrocarbon receptor nuclear translocator gene expression levels. So far, two studies have examined (ARNT) to form HIF1a–ARNT (HIF1) and HIF2a– the effect of hypoxia in regulating RNA splicing using ARNT (HIF2) complexes. HIF1 and HIF2 activate gene exon arrays (6, 7). However, both studies used endothelia transcription to promote tumor progression and metastasis cells, a normal cell type. So far, hypoxia-mediated RNA (1–4). Thus, HIF-mediated transcriptional response is very splicing changes have not been measured in cancer cells. In important in cancer biology. addition, these previous studies in endothelial cells Recently, it was found that 95% of human genes are focused on non-HIF target genes and reported that hyp- alternatively spliced and RNA isoforms produced from a oxia, like other stresses, inhibits RNA splicing of non-HIF single gene code proteins with distinct or opposing functions target genes by promoting exon skipping and/or intron (5). Thus, it becomes increasingly clear that assessing iso- inclusion (6, 7). Thus, role of hypoxia in regulating RNA splicing of HIF-induced genes, the most important genes in hypoxia response is still unknown. Furthermore, none Authors' Afﬁliations: 1Molecular Biology Graduate Program and 2Depart- of these studies have addressed the molecular mechanisms ment of Craniofacial Biology, School of Dental Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado about how hypoxia regulates alternative splicing (AS) of HIF target genes. In this study, we used exon arrays to Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/). assess global effects of hypoxia on RNA splicing of HIF and non-HIF target genes in Hep3B cells, a cancer cell Corresponding Author: Cheng-Jun Hu, Department of Craniofacial Biol- ogy, School of Dental Medicine, University of Colorado Anschutz Medical line.Inaddition,westudiedifHIForhypoxiaper se is Campus, Aurora, CO 80045. Phone: 303-724-4574; Fax: 303-724-4580; important for RNA splicing of several HIF target genes. E-mail: [email protected] These ﬁndings have important implications for our under- doi: 10.1158/1541-7786.MCR-14-0149 standing of how HIFs promote tumorigenesis in response 2014 American Association for Cancer Research. to hypoxia.

www.aacrjournals.org 1233

Sena et al.

Materials and Methods induced, (ii) genes that were predicted to undergo hypox- Cell culture ia-induced AS, and (iii) genes that were hypoxia induced and Hep3B cells were cultured in Minimum Essential Medi- exhibiting AS. um with Earle's Balanced Salt Solution (Hyclone) contain- Knockdown of endogenous mRNA using small ing 10% FBS, 2 mmol/L L-glutamine, 1 mmol/L sodium interfering RNAs pyruvate, 100,000 U/L penicillin/streptomycin, 1.5 g/L fi sodium bicarbonate, and 1 nonessential amino acids Control (Qiagen; 1027281) or siRNAs speci c for human (NEAA). HEK293T, RCC4, RCC4T, and UM-SCC- ARNT (Qiagen; equal mix of SI00304220, SI00304234, fi and SI03020913), HIF1a (Qiagen; SI02664053), and 22B cells were grown in high-glucose Dulbecco's Modi ed a Eagle Medium (DMEM; Hyclone) with 10% FBS, 2 HIF2 (Qiagen; SI00380212) mRNAs were transfected into Hep3B cells at 40% confluency using HiPerFect mmol/L L-glutamine, 100,000 U/L penicillin/streptomycin, and 1 NEAA. SK-N-MC cells were grown in RPMI-1640 Transfection Reagent (Qiagen) according to the manufac- turer's protocol. Thirty-two hours after transfection, cells (Hyclone) with 10% FBS, 2 mmol/L L-glutamine, 100,000 were cultured at 21% or 1.5% O2 for 12 to 16 hours and U/L penicillin/streptomycin, and 1 NEAA. Before hyp- fi oxia treatment, 25 mmol/L HEPES was added to growth then collected for mRNA or protein puri cation. media and cells were incubated under normoxia (21% O2)or hypoxia (1.5% O ) for 12 to 16 hours. All parental cell lines Viral transduction of Hep3B cells 2 The GFP-Flag, HIF1aTM-Flag (triple mutation to make were purchased from ATCC. After completing the experi- a ments, the parental (Hep3B, HEK293T, RCC4, UM-SCC- HIF protein active under normoxia, with Flag tag), and fi HIF2aTM-Flag lentiviral constructs were described (11). 22B, and SKNMC) and modi ed cell lines (RCC4T) were a a authenticated by DNA profiling or "fingerprinting" by the The GFP-Flag, HIF1 TM-Flag, or HIF2 TM-Flag plas- University of Colorado DNA Sequencing & Analysis Core. mid was cotransfected with psPAX2 packaging vector (Addgene), and pMD2.G envelope vector (Addgene) into HEK293T cells. Twenty-four hours after transfection, Exon array analysis of AS transfection media were replaced with complete Hep3B The genome-wide effect of hypoxia on AS was determined media. The following day, viral media were filtered to in Hep3B cells using the Affymetrix GeneChip Human remove possible 293T cells and viral particles were concen- Exon 1.0 ST array. Hep3B cells were cultured, as described trated using high-speed ultracentrifugation at 26,000 g for above, under normoxic (21%) or hypoxic (1.2%) conditions 3 hours. The viral pellet was resuspended in 1 mL of PBS and for 12 hours. RNA was prepared using a Qiagen RNeasy Plus added to 6 cm Hep3B cells (30% confluency) for 6 hours. RNA extraction kit. RNA was reverse transcribed and The following day, a second round of concentrated virus was assessed for hypoxic induction of HIF target genes using added to the same Hep3B cells. Twenty-four hours after the qPCR. RNAs that passed quality control were submitted to second viral transduction, Hep3B cells were harvested for the Gene Expression Core at the University of Colorado RNA and protein analysis. Comprehensive Cancer Center for exon array analysis. Three independent replicate samples were analyzed for each Plasmid constructs treatment. The Core performed target labeling, hybridiza- The CA9P/ADM, PAI1P/ADM, and PAI1PmHRE/ fi tion, and chip scanning. To assess AS events, raw CEL les ADM constructs were described previously (11). The were analyzed using the freely available AltAnalyze software CA9P/PDK1, PAI1P/PDK1, and PAI1PmHRE/PDK1 (http://www.altanalyze.org; ref. 8) with the following constructs were generated by replacing the ADM gene with settings: species, human; gene and exon set, both core; and PDK1 minigene that was PCR amplified from human gene and exon level normalization FIRMA. The array data genomic DNA and contained exon 3 to exon 5, including fi can be accessed at GEO (GSE57613). Raw CEL les were introns 3 and 4. The HIF1aTM-Flag, HIF2aTM-Flag, alternatively analyzed using easyExon (http://microarray. HIF1aDBD-Flag, HIF1aDBD/VP16-Flag, HIF1aDBD/ ym.edu.tw:8080/tools/module/easyexon/index.jsp?mod- E2F-Flag, and upstream stimulatory factor 2 (USF2) expres- ¼ fi e home), with default settings and using RMA ltering (9). sion plasmids were described previously (11–13). The G5/ In addition, AltAnalyze software was used for gene expres- PDK1 minigene was generated by replacing the ADM gene sion analysis. Genes predicted to undergo AS by AltAnalyze with PDK1 minigene in the G5/ADM construct that was and easyExon were then compared with genes that were described previously (11). The Gal4 DNA binding domain induced by hypoxia (AltAnalyze) using Venny (http:// (Gal4), Gal4/HIF1aTAD, Gal4/VP16, and Gal4/E2F1 bioinfogp.cnb.csic.es/tools/venny/) to identify genes that have been described in ref. (11). were predicted to undergo AS and hypoxia induction. PDK1 splicing reporter assay Functional clustering of hypoxia-inducible genes and Typically, HEK293T cells were grown to approximately alternatively spliced genes using DAVID bioinformatics 30% confluency in 6-well plates and cotransfected with resources 200 ng of splicing reporters (CA9P/PDK1, PAI1P/PDK1, DAVID Bioinformatics (10) was used to create function- PAI1PmHRE/PDK1, or G5/PDK1) and 1.8 mg of His ally annotated clusters of (i) genes that were hypoxia (empty vector control), HIF1aTM, HIF2aTM, USF2,

1234 Mol Cancer Res; 12(9) September 2014 Molecular Cancer Research

Hypoxia Regulates Alternative Splicing

HIF1aDBD, HIF1aDBD/E2F1, or HIF1DBD/VP16 ation. Asterisks indicate statistical significance as follows: expression constructs or Gal4 expression construct for , P < 0.05; , P < 0.01. Controls for statistical analysis G5/PDK1. Forty-eight hours after transfection, cells were are specified in each figure. All experiments were per- collected for RNA or protein preparation. Results were the formed at least three separate times. average of at least three experiments. Results Protein analysis Genome-wide exon array analysis determines that Whole-cell lysates were prepared and quantified for pro- hypoxia alters RNA splicing of HIF and non-HIF target tein concentration. Western blot analysis was performed genes in Hep3B cells using standard protocols with the following primary anti- To assess the effects of hypoxia on RNA splicing, Affy- bodies: anti-CA9 (H-120) pAb (SC-25599; Santa Cruz metrix GeneChip Human Exon 1.0 S T arrays were used to Biotechnology), anti-ANGPTL4 (H-200) pAb (SC- probe RNAs isolated from normoxic or hypoxic Hep3B 66806; Santa Cruz Biotechnology), anti-Flag mAb cells. Hep3B cells were chosen because of their high expres- (F3165; Sigma), anti-Gal4DBD pAb (SC-577; Santa Cruz sion levels of HIFa proteins and high hypoxic induction of Biotechnology), or anti-beta Actin pAb (SC-1616; Santa HIF target genes (12–14). AltAnalyze (FIRMA) and easy- Cruz Biotechnology). Exon software predicted that hypoxia regulated AS of about 2,005 (Fig. 1A, yellow and gray) and 3,919 genes, respec- RNA preparation and reverse transcription PCR or tively (Supplementary File S1; AS by Firma or AS by easy- quantitative PCR Exon). Venn diagrams using Venny indicated that 1,012 RNA was isolated from cells using the RNeasy Plus mini genes were predicated to undergo AS by both programs, kit (Qiagen), which removes DNA, then was reverse tran- whereas an additional 957 and 2,634 genes were predicated scribed using the iSCRIPT Advanced cDNA synthesis kit to undergo AS by FIRMA and easyExon, respectively. (Bio-Rad) containing both oligo-dT and random hexamer as FIRMA was used to further characterize the hypoxia-medi- primers because using both random hexamers and oligo-dT ated splicing events, as FIRMA predictions of AS are more allows efficient reverse transcription of the whole mRNA stringent than easyExon. FIRMA analysis indicated that transcript as well as 18 rRNA. Alternative splicing analysis hypoxia produced 3,059 AS events in 2,005 genes, averaging was then validated by RT-PCR using forward and reverse 1.53 splicing events per gene (Fig. 1A, all genes). As primers flanking AS events predicted by both AltAnalyze and expected, alternative cassette-exons were the most abundant easyExon software. Genes that expressed multiple transcripts type of splicing events, making up 51% of all splicing events. were Topo TA cloned and submitted to the Gene Expression Other AS events included alternative 50 splice-sites (16%), Core at the University of Colorado Cancer Center for alternative 30 splice-sites (14%), intron retention (11%), sequencing. qRT-PCR using iQ SYBR Green Supermix mutually exclusive exon (4%), and others (4%). (Bio-Rad) in triplicate on the CFX384 Real-Time System In addition, AltAnalyze determined that hypoxia induced (Bio-Rad) was used to quantify the levels of RNA isoforms. 2,439 genes, for at least 1.4 fold, when compared with All primer sets for qRT-PCR designed to measure mRNA normoxic Hep3B controls (Fig. 1A, blue and gray; also see levels were validated for their specificity and amplification Supplementary File S1, gene expression). Venny analysis of efficiency (85%–110%) using melt curve analysis, qRT- hypoxia-inducible genes and AS genes indicated that 134 PCR product sequencing, and standard dilution analysis. genes were hypoxia induced and alternatively spliced and flDD qRT-PCR results were normalized using the Pfaf CT these 134 genes were called Hx AS genes (Fig. 1A, in gray). method using18s rRNA and b-actin as reference genes Interestingly, these 134 Hx AS genes had 238 splicing because their expression are not affected by hypoxia and events, averaging 1.78 AS events per gene (Fig. 1A, Hx- untreated normoxia samples, GFP lentivirus, or empty induced genes), suggesting that AS is enriched in hypoxia- vectors (His) as controls. The RNA ratio of splice variants induced genes, compared with the total population of genes was calculated by dividing the expression level of the full- that underwent AS (1.53 AS events per gene). Similar to the length (FL) isoform by the expression level of the DE total population of AS genes, alternative cassette-exons were isoform. The expression level for each isoform was deter- the most abundant type of splicing events (62%) in Hx AS mined by the following equation: Expression level ¼ genes. Other AS events for these Hx AS genes included D D efficiency Ct target (control sample)/efficiency Ct alternative 50 splice sites (16%), alternative 30 splice sites target reference reference (control sample), where the efficiency ¼ 10 1/slope (13%), intron retention (9%), and mutually exclusive exons (slope generated by standard curve analysis). At least three (0%). However, Hx AS genes exhibited 1.2-fold enrichment independent experiments were performed to generate in alternative cassette-exon inclusion and a 1.22-fold reduc- the results presented in the Figures. All the primers used tion in intron retention in comparison with the total pop- for RT-PCR and qRT-PCR were included in Supplemen- ulation of AS genes. tary File S3. Because cassette-exon AS is the most abundant splicing event, hypoxia-mediated cassette-exon regulation was fur- Statistical analysis ther analyzed. In the total population of cassette-exon genes One-way analysis of variance was performed unless oth- (all cassette, Fig. 1B), 44% of cassette-exon genes exhibited erwise stated. Error bars in figures indicate standard devi- exon inclusion (454 out of 1,022 genes), whereas 56% of the

www.aacrjournals.org Mol Cancer Res; 12(9) September 2014 1235

Sena et al.

ES: 1.5 ES: 1.3 ES: 7.1 ES: 2.2 ES: 1.5 P = 0.023 P = 0.036 −8 ES: 2.5 −4 ES: 1.4 ES: 1.2 P = 2.35x10 P = 4.99x10 P = 0.021 Bloodvesse Leucine- ES: 5.6 P = 0.001 P = 0.015 ES: 1.2 P = 0.01 Hexose L-ascorbic Membrane l morpho- Zipper P = 1.45x10−8 P = 0.047 metabolic Oxidoreduct acid - bounded genesis Gluconeo Transcription Basement process Glycolysis ase activity binding vesicle -genesis factor Cell death membrane

A ES: 6.1 Vascular ATP-binding/protein ES: 6.1 −8 P = 5.08x10 −7 development kinase activity P = 1.53x10 2,305 134 1,871 ES: 4.7 ES: 5.7 Pleckstrin homology P = 5.8x10−6 − P = 4.08x10 8 Glycolysis Oxidoreductase ES: 3.8 − activity P = 3.41x10 4 ES: 5.4 Gluclose − P = 7.82x10 7 metabolism All Genes Hx Induced Genes Total Events 3,059 238 Rho protein signal ES: 3.4 ES: 3.7 L-ascorbic Total Genes 2,005 134 transduction P = 9.78x10−5 −5 P = 4.21x10 acid binding Events/Gene 1.53 1.78 ES: 3.4 Cytoskeleton Leucine-zipper Exon skipping P = 3.81x10−4 ES: 3.2 organization P = 2.22x10 −4 transcription factor Exon inclusion B ES: 3.3 Zinc finger −4 ES: 3.2 G-protein coupled 100% P = 2.36x10 − P = 1.22x10 4 receptor 90% 24% 80% 48% 20 Glycolysis ES: 3.2 56% P = 1.31x10−4 ES: 3.1 Gluconeogenesis 70% 273 P = 1.83x10 −4 60% 568 75% Positive regulation of glucose ES: 3.1 ES: 3.0 50% 275 P = 5.36x10−4 −4 Transmembrane transport P = 1.88x10 40% 76% 30% 63 ES: 2.8 ES: 2.9 Oxidoreductase 52% Cell death 20% 44% P = 4.55x10 −4 P = 0.0011 activity 299 25% 10% 454 92 ES: 2.7 ES: 2.8 Nuclear lumen 0% P = 5.92x10 −4 P=0.003 Cell adhesion/cadherins All cassette No change Hx-induced Hx-reduced AS AS AS AS

Figure 1. Hypoxia regulates gene expression and RNA splicing in Hep3B cells. A, Venn diagrams of genes that are hypoxia induced (blue and gray), genes that are alternatively spliced under hypoxia (yellow and gray), and genes whose expression are induced by hypoxia and alternatively spliced under hypoxia (gray). Left column, top 10 enrichment pathways whose genes are induced at least 1.4 fold by hypoxia. Top row, top 10 enrichment pathways whose genes that are hypoxia induced and undergo hypoxia-induced AS. Right column, top 10 enrichment pathways whose genes undergo hypoxia-induced AS. Numbers next to each box represent the DAVID enrichment score (ES), higher score representing more enriched, followed by the Fisher exact P value, smaller score representing more enriched. P values equal to or less than 0.05 are considered strongly enriched. Boxes of the same color, excluding white, represent overlapping or functionally related gene pathways. White boxes are unique pathways for each group. Additional statistics and gene list associated with each group can be found in Supplementary File S2. Below the Venn diagram in (Fig. 1A are the splicing events per genes for all alternatively spliced genes (all genes, left) and hypoxia-induced genes (Hx-induced genes, right). B, summary of exon skipping or inclusion events in all cassette AS genes, genes whose expression are not changed during hypoxia (no change AS), genes induced by hypoxia (Hx-induced AS), and genes reduced by hypoxia (Hx-reduced AS).

genes exhibited exon skipping, suggesting that hypoxia induced (Fig. 1A, left column), the entire population of AS generally promotes exon skipping in Hep3B cells. However, genes (Fig. 1A, right column), and the Hx AS genes (Fig. 1A, 76% of cassette-exon Hx AS genes underwent exon inclu- top; see Supplementary File S2 for the gene lists). As sion (63 out of 83 genes), whereas 24% of genes exhibited expected, hypoxia induced the expression of genes in path- exon skipping, suggesting that hypoxia promotes exon ways allowing cancer cells to adapt to a hypoxic microen- inclusion of hypoxia-inducible genes (Fig. 1B, Hx-induced vironment such as vascular development, glycolysis, oxido- AS). In contrast, hypoxia promoted exon skipping of 75% of reductase activity, and cell adhesion/cadherins (Fig. 1A, left hypoxia-reduced cassette-exon genes (Hx reduced, Fig. 1B), column). On the other hand, when analyzing all AS genes, whereas the number of exon inclusion to exon skipping pathways involved in oxidoreductase activity, glycolysis, and events was equal (50%) for genes whose expression was not positive regulation of glucose transport that were enriched in changed by hypoxia (no change, Fig. 1B). gene expression were also enriched in AS (Fig. 1A, right column and Supplementary File S2). However, for the same Functional clustering of hypoxia-mediated AS genes pathway, the genes involved in AS could be different from using DAVID reveals new pathways regulated by hypoxia the genes enriched in expression analysis, demonstrating the To determine whether hypoxia-induced AS was enriched importance of these pathways in the hypoxia response. in speciﬁc cellular pathways, DAVID (10) was used to create Importantly, several novel cellular pathways, including functional annotation clusters of genes that were hypoxia ATP-binding/protein kinase activity, pleckstrin homology

1236 Mol Cancer Res; 12(9) September 2014 Molecular Cancer Research

Hypoxia Regulates Alternative Splicing

(related to rho protein signal transduction and cytoskeleton Expression of multiple isoforms and hypoxia-regulating organization), rho protein signal transduction, cytoskeleton RNA splicing of additional HIF target genes, pyruvate organization, zinc fingers, cell death, and nuclear lumen dehydrogenase kinase 1 (PDK1, FL, and DE4), WNK lysine genes, were also enriched (Fig. 1A, right column and deficient protein kinase 1 (WNK1, FL, and DE11–12), and Supplementary File S2). Correspondingly, Hx AS genes prolyl 4-hydroxylase alpha polypeptide II (P4HA2, FL, and were largely overlapped with pathways identified in gene DE2), were also confirmed by RT-PCR and qRT-PCR expression and/or AS (Fig. 1A, top). (Supplementary Fig. S2A–C). CA9, ANGPTL4, PDK1, WNK1, and P4HA2 exhibited RT-PCR and qRT-PCR validation of AS for select HIF preferential induction of the FL isoform. However, AltA- target genes identified in exon array analysis nalyze (FIRMA) predicts that hypoxia promotes exon skip- Next, we sought to validate the exon array data using RT- ping of some HIF target genes, including procollagen-lysine, PCR and qRT-PCR. We focused on Hx AS genes as the 2-oxoglutarate 5-dioxygenase 2 (PLOD2, FL, and DE14) hypoxia-inducible genes are the most important genes in the and enolase 2 (ENO2, FL, and DE8), which are also hypoxia response. First, we selected Hx AS genes as deter- confirmed (Supplementary Fig. S2D and S2E). These data mined by both AltAnalyze FIRMA and easyExon. Next, we demonstrated that hypoxia promotes exon inclusion for used easyExon to identify regions of potential AS. Then we most HIF target genes; however, hypoxia promotes exon used RT-PCR to identify the RNA isoforms and qRT-PCR skipping for some HIF target genes such as PLOD2 and to quantify the levels of these isoforms in normoxic and ENO2. hypoxic Hep3B cells or other cancer cells. For example, carbonic anhydrase 9 (CA9), a well-known HIF target gene Hypoxia changes isoform ratios of HIF target genes at the (15, 16), was predicted to undergo AS in exon 2 and in a transcriptional level region spanning exons 6 to 10 as determined by the splicing After confirming AS of several HIF target genes, we index (Fig. 2A). AS was considered significant if the splicing addressed the molecular mechanism on how hypoxia reg- index was greater than þ1.5 or less than 1.5 (negative SI ulates AS of HIF target genes. Although AS was proposed to indicates exon inclusion by hypoxia) using easyExon. RT- explain the observed FL to skipping isoform ratio changes in PCR was not able to detect an AS event in exon 2. RT-PCR, the above sections, the hypoxia-induced isoform ratio varia- using primers that amplify exons 4 to 10 (Fig. 2B), followed tions could be caused by RNA splicing or other mechanisms. by Topo TA cloning and sequencing confirmed that Hep3B For instance, preferential degradation of exon skipping iso- cells expressed two CA9 isoforms, a full-length isoform (FL), forms under hypoxia might explain the hypoxia-induced and an isoform in which exons 8 and 9 were skipped isoform ratio changes as exon skipping RNA isoforms may (DE89; Fig. 2C, left). qRT-PCR determined that hypoxia incur premature termination codons and such transcripts are induced the expression of both isoforms but significantly often targeted for nonsense-mediated decay (NMD; ref. 18). favored FL in Hep3B cells; thus, hypoxia increased the FL/ In the seven genes analyzed, only RNA of CA9 DE89 (E8–9, DE89 ratio by 9.3 fold (Fig. 2D). Similar CA9 isoform 172 bp), but not ANGPTL4 (E4, 114 bp), PDK1 (E4, expression patterns and increased CA9 FL/DE89 ratio by 138 bp), WNK1 (E11–12, 609 bp), P4HA2 (E2, 136 bp in hypoxia were observed in a neuroblastoma cell line, 50-UTR), PLOD2 (E13, 114 bp), and ENO2 (E8, 198 bp) SKNMC, suggesting that this splicing event is not cell-type had premature stop codon. To test the possibility that specific (Fig. 2E and F). In addition, Western blot analysis of preferential instability of CA9 DE89 contributes to the CA9 in Hep3B cells confirmed that both FL and DE89 increased CA9 FL/DE89 ratio under hypoxia, Hep3B cells protein isoforms were expressed and FL CA9 protein was were placed under hypoxia for 16 hours to increase the levels preferentially induced by hypoxia (Fig. 2C, right). The larger of both CA9 FL and DE89 transcripts, followed by treatment CA9 FL band in hypoxic Hep3B cells was likely the result of with actinomycin D to inhibit de novo transcription. Cells posttranslational modifications and was also observed pre- were then placed back under normoxia or hypoxia for 0, 2, 4, viously (17). or 8 hours to allow RNA decay. Using qRT-PCR, both CA9 Angiopoietin-like 4 (ANGPTL4) was predicted to under- FL and DE89 transcripts were found to be very stable, as go AS in exon 4 (Supplementary Fig. S1A). RT-PCR, Topo 90% of the FL and DE89 transcripts were detected even after TA cloning, and sequencing confirmed that ANGPTL4 8 hours. Moreover, CA9 FL and DE89 transcripts exhibited expressed a FL isoform, and an exon 4 skipping isoform similar stability under normoxia and hypoxia at every time (DE4) in Hep3B cells (Supplementary Fig. S1C, left). qRT- point (data not shown). ANGPTL4 FL and DE4 transcripts PCR determined that hypoxia induced the expression of both were far less stable than CA9 transcripts as only 40%, 28%, isoforms but FL was favored; thus, hypoxia increased the FL/ and 20% of the transcripts remained after 2, 4, and 8 hours. DE4 ratio by 4.9 fold (Supplementary Fig. S1D). In addition, However, ANGPTL4 FL and DE4 transcripts also exhibited hypoxia increased the ANGPTL4 FL/DE4 ratio by 3.5 fold in similar stability under normoxia and hypoxia (data not UM-SCC-22B, a head and neck squamous cell carcinoma shown). Furthermore, actinomycin D treatment in Hep3B cell line (Supplementary Fig. S1E and S1F). The expression cells blocked hypoxic induction of HIF target genes and of ANGPTL4 FL and DE4 protein isoforms and preferential blocked splicing change of HIF target genes, indicating that induction of ANGPTL4 FL protein by hypoxia were also the hypoxia-mediated isoform shift required active transcrip- observed in Hep3B cells (Supplementary Fig. S1C, right). tion. These data supported the idea that transcription

www.aacrjournals.org Mol Cancer Res; 12(9) September 2014 1237

Sena et al.

200 A 12 D Hep3B Fold ratio Hx Δ 11 FL: E89 change Nx 16.68 11.28 1 10 150 Hx 154.36 41.42 9.3 9 8 100 Vs Hx 7 Intensity (log) Nx *P < 0.05 * P < 0.01 6 * 50 5 * 0.5 * * Normalized fold expression Normalized fold 0.0 0 –0.5 –1.0 CA9FL CA9ΔE89

Splicing index –1.5 5 ′ 3 ′ HxNx E FL

*3168067 *3168068 *3168071 *3168072 *3168073 *3168076 *3168078 *3168079 *3168083 *3168085 *3168086 SKNMC cell Probeset ΔE89 B CA9 E10/7 CA9 F Fold ratio E4 E7 SKNMC cell 120 FL:ΔE89 change 192 876543 10 11Nx 3.70 ± 0.57 1 100 Hx 50.8 ± 15.28 13.7 80 E8/7 E10 C Hep3B 60 HxNx NX HX 40 FL FL 50 kDa * * Δ 20 * * ΔE89 E89

Nonspecific expression Normalized fold * 37 kDa 0 CA9 anti-CA9 CA9FL CA9ΔE89

Figure 2. Hypoxia regulates AS of the HIF target gene, CA9. A, easyExon analysis of differentially expressed probes for the CA9 gene in Hep3B cells. Lines, probe intensities for normoxia or hypoxic samples. Differentially expressed probes were considered positive for AS if the splicing index was greater than 1.5 or less than 1.5 (negative SI indicates exon inclusion is increased by hypoxia). Vertical numbers below the splicing index indicate Affymetrix probe IDs. B, diagram of the conﬁrmed CA9 gene isoforms. Gray boxes, constitutive exons; black boxes, alternatively spliced exons. Solid arrows, primer used for RT-PCR (E4//E10) and qRT-PCR (E7/E10/7 for CA9DE89, and E7/E8/7 for CA9 FL). C, RT-PCR and Western blot analysis of CA9 mRNA and protein in normoxic and hypoxic Hep3B cells. CA9 FL proteins exhibit two bands with different sizes. D, qRT-PCR analysis of CA9FL and DE89 transcripts in normoxic and hypoxic Hep3B cells. E, RT-PCR analysis of CA9 mRNAs in normoxic and hypoxic SKNMC cells. F, qRT-PCR analysis of CA9FL and DE89 transcripts in normoxic and hypoxic SKNMC cells. Numbers in D and F here and other qRT-PCR panels represent the ratio of the FL isoform versus the exon-skipping isoform the standard deviation. Fold of ratio change is also indicated. One-way analysis of variance was performed for this and other studies in this article unless otherwise stated. , P ¼ 0.05; , P ¼ 0.01. Controls for statistical analysis are speciﬁed in each Figure.

regulation, not posttranscriptional regulation is responsible FL and exon skipping isoforms of CA9, ANGPTL4, and for the hypoxia-induced increased CA9 and ANGPTL4 FL/ PDK1 and prevented the splicing changes of these genes exon-skipping ratio. (Fig. 3B–D). In contrast, HIF2a knockdown only mildly reduced hypoxic induction of CA9FL (1.44 fold) and HIF activity, not hypoxia per se, is necessary to change AS PDK1DE4 (1.6 fold), similarly reduced hypoxic induction of HIF target genes of ANGPTL4FL and DE4. Thus, HIF2a knockdown only To test whether hypoxic stress or HIF activity is respon- reduced the CA9FL/DE89 ratio by 1.33 fold (Fig. 3B), sible for the splicing changes of HIF target genes, ARNT, maintaining the ANGPTL4FL/DE4 ratio (Fig. 3C), but HIF1a, or HIF2a mRNA levels were reduced by 80% using enhanced the PDK1FL/DE4 ratio 1.5 fold (Fig. 3D). Knock- siRNAs in normoxic or hypoxic Hep3B cells (data not down of ARNT and HIF1a also inhibited hypoxic induc- shown). ARNT and HIF1a, but not HIF2a knockdown tion of the FL and exon skipping isoforms of WNK1, dramatically reduced the hypoxic induction of CA9, PLOD2, ENO2, and P4HA2 in Hep3B cells and prevented ANGPTL4, and PDK1, consistent with the idea that CA9, splicing ratio changes for these genes (data not shown). These ANGPTL4, and PDK1 are primarily regulated by HIF1 in data suggested that HIF activity, but not hypoxia per se is Hep3B cells (Fig. 3A). qRT-PCR conﬁrmed that ARNT and necessary for increased gene expression as well as hypoxia- HIF1a knockdown signiﬁcantly reduced the levels of both mediated splicing changes of these HIF target genes.

1238 Mol Cancer Res; 12(9) September 2014 Molecular Cancer Research

Hypoxia Regulates Alternative Splicing

A C 200

150 siRNA Control ARNT HIF1α HIF2α * Nx Hx Nx Hx Nx Hx Nx Hx Figure 3. HIF activity, but not FL 100 CA9 hypoxia per se is necessary to ΔE89 promote AS of HIF target genes. FL 50 A, RT-PCR analysis of CA9, ΔE4 ANGPTL4 * * ANGPTL4, and PDK1 RNA * * * * FL * * * transcripts in normoxic or hypoxic Δ PDK1 0 E4 expression fold Normalized Hep3B cells targeted with control ANGPTL4FL ANGPTL4ΔE4 a a (Ctrl), ARNT, HIF1 , or HIF2 B siRNA siRNAs. B to D, qRT-PCR analysis D Ctrl Nx of FL and exon-skipping isoforms 90 Vs Ctrl Hx *P < 0.05 Ctrl Hx 25 of CA9 (B), ANGPTL4 (C), and 80 * P < 0.01 * ARNT Nx 70 PDK1 (D) in the normoxic or ARNT Hx 20 * hypoxic Hep3B cells targeted with 60 HIF1α Nx control or siRNA to ARNT, HIF1a, 50 HIF1α Hx 15 * or HIF2a. 40 HIF2α Nx 30 10 20 * 5 * 10 * * * ** * * * * * * *

Normalized fold expression 0 0 Normalized fold expression Normalized fold CA9FL CA9ΔE89 PDK1FL PDK1ΔE4

HIF activity is sufficient to regulate AS of HIF target FL/DE89 ratio in RCC4 cells was already high even under genes normoxia (Supplementary Fig. S3B). Similar findings were Next, we wanted to determine whether HIF activity is observed for AS of ANGPTL4; however, hypoxia was still sufficient for hypoxia-regulated AS of HIF target genes. To able to enhance the ANGPTL4 FL/DE4 ratio in RCC4 test this, normoxic Hep3B cells were transduced with (Supplementary Fig. S3C). Although hypoxia was able to lentiviruses expressing normoxia active, flag-tagged, HIF1a, increase the expression of the PDK1FL and DE4 isoforms or HIF2a, or GFP as a negative control (Fig. 4A). HIF1a and to enhance the PDK1FL/DE4 ratio in RCC4T cells, and HIF2a transduction induced the expression of CA9, PDK1 FL/DE4 ratio was not elevated in normoxic or ANGPTL4, and PDK1 as determined by RT-PCR (Fig. 4B). hypoxic RCC4 cells (Supplementary Fig. S3D). More importantly, qRT-PCR determined that HIF1a and Taken together, these data suggested that HIF activity is HIF2a increased both FL and exon skipping isoforms of necessary and sufficient to regulate AS of a subset of HIF CA9 (Fig. 4C), ANGPTL4 (Fig. 4D), and PDK1 (Fig. 4E). target genes independent of hypoxia. However, FL transcripts of CA9, ANGPTL4, and PDK1 were preferentially induced (Fig. 4C–E). Thus, HIF1 or PDK1 splicing reporters recapitulate splicing changes HIF2 increased the FL/exon skipping ratio for CA9 by 14.5 observed for the endogenous PDK1 gene when activated or 6.0 fold (Fig. 4C), ANGPTL4 by 3.41 or 5.7 fold by HIF under normoxia (Fig. 4D), and PDK1 by 1.43 or 1.2 fold (Fig. 4E). Next, we wanted to see if similar splicing changes could be To further validate these results, the RNA splicing of CA9, observed in splicing reporter gene. We selected the PDK1 ANGPTL4, and PDK1 were examined in RCC4 cells, a gene for splicing reporter model as both the FL and DE4 renal cell carcinoma cell line that expresses constitutively isoforms were highly expressed in 293T cells (data not active HIF1a and HIF2a proteins due to loss of functional shown). Exons 3 to 5, including introns 3 and 4 of the pVHL (15, 19). In addition, the RNA splicing of the above PDK1 gene, were PCR amplified from human genomic three genes were also assessed in RCC4T cells in which DNA and placed downstream of the CA9 promoter, a HIF1 functional pVHL is reintroduced into RCC4 cells and, target gene promoter, or the PAI1 promoter, a HIF2 target therefore, HIF proteins are only active under hypoxia gene promoter (Fig. 5A). RT-PCR using minigene-specific (15, 19). RT-PCR and qRT-PCR analysis determined that primers determined that the PDK1 minigene expressed both hypoxia increased the levels of CA9FL and DE89 by 41.8 FL and DE4 isoforms in 293T and Hep3B cells (Fig. 5B). In and 7 fold, respectively, in RCC4T cells, increasing the FL/ addition, the PDK1DE4 isoform was the dominantly DE89 ratio by 6.2 fold (Supplementary Fig. S3B). In expressed isoform in both 293T and Hep3B cells (Fig. contrast, hypoxia did not induce the levels of CA9FL and 5B). Furthermore, PDK FL isoforms were also highly DE89 isoforms nor did hypoxia enhance the FL/DE89 ratio expressed in 293T cells (Fig. 5B). These data demonstrated in RCC4 cells (Supplementary Fig. S3B). However, the CA9 that the PDK1 splicing reporter recapitulated the expression

www.aacrjournals.org Mol Cancer Res; 12(9) September 2014 1239

Sena et al.

A D 140 Fold ratio FL:ΔE4 change * α α 120 GFP * 3.61 GFP HIF1 Flag HIF2 Flag α 1 HIF1 12.31 100 HIF2α 3.4 130 kDa 20.55 5.7 80 * 95 kDa * Figure 4. HIF activity is sufﬁcient to Anti-Flag 60 promote AS of HIF target genes. A, Western blot analysis of Flag- GFP HIF1α HIF2α 40 * B * * tagged proteins in normoxic FL 20 CA9 Hep3B cells transduced with ΔE89 FL Normalized fold expression 0 lentiviruses expressing GFP, or ΔE4 ANGPTL4 ANGPTL4FL ANGPTLΔE4 Flag-tagged, normoxia-active HIF1a or HIF2a proteins. B, RT- FL ΔE4 PDK1 PCR analysis of CA9, ANGPTL4, and PDK1 transcripts in normoxic Fold ratio Hep3B cells transduced with GFP, C E a Fold ratio FL:ΔE4 change or normoxia-active HIF1 ,or FL:ΔE89 HIF2a proteins. C to E, qRT-PCR change 8 GFP 0.14 1 120 α analysis of FL and exon-skipping GFP 3.22 1 HIF1 * 0.20 1.43 * α 7 HIF2α 100 * HIF1 46.75 14.5 0.12 –1.2 RNA isoforms of CA9 (C), HIF2α 19.26 6 6 ANGPTL4 (D), and PDK1 (E) in the 80 Vs GFP 5 * above described cells. 60 *P < 0.05 4 * P < 0.01 * 3 40 * * 2 * * 20 * * 1 0 0 Normalized fold expression Normalized fold expression CA9FL CA9ΔE89 PDK1FL PDK1ΔE4

patterns of the endogenous PDK1 gene in 293T and Hep3B These data indicated that HIF transactivation domain is not cells (Supplementary Fig. S2A and data not shown). required for AS of the PDK1 minigene. To assess AS of the PDK splicing reporter in response to In summary, these data indicated that HIF1 or HIF2- HIF activation, 293T cells were transfected with the PDK1 mediated activation of the PDK1 minigene is sufficient to minigene and normoxia active HIF1a expression plasmids. increase PDK1 FL/DE4 ratio. HIF1a induced the levels of total PDK1 minigene transcripts, FL, and DE4 isoform by 3.65, 24.9, and 10.6 fold, Activation of endogenous HIF target genes contributes to respectively, and enhanced the FL/DE4 ratio by 2.34 fold increased FL/DE4 ratio of PDK1 splicing reporter (Fig. 5D). Next, similar experiments were performed using As stated above, transcription activation of PDK1 by HIF the PAI1P/PDK1 splicing reporter (Fig. 5A). Normoxia or HIFDBD hybrid constructs is sufficient to increase the active HIF2a induced the expression of PDK1FL and DE4 PDK1 minigene FL/DE4 ratio. However, HIF and the isoforms by 4.22 and 2.36 fold, respectively, thus enhancing fusion constructs can activate endogenous HIF target genes; the FL/DE4 ratio 2.08 fold (Fig. 5E). These results indicated therefore, it is possible that activation of endogenous HIF that the PDK1 splicing reporters recapitulate splicing target genes may increase FL/DE4 ratio. To rule out or changes observed for the endogenous PDK1 gene when confirm this possibility, we used USF2 and the PAI1P/ activated by HIF under normoxia. PDK1 minigene as USF2 can activate the PAI1 promoter Some transcription factors have dual roles in RNA splicing (13), but not endogenous HIF target genes. USF2 induced and gene transcription by recruiting splicing factors and the expression of the PDK1FL and DE4 isoforms by 2.95 transcription cofactors using their transactivation domains and 2.96 fold, respectively, but was not able to enhance the (20–22). To determine whether the activation domain of FL/DE4 ratio even though USF2 was able to activate the HIF1a protein is important for PDK1 pre-mRNA splicing, minigene to a similar extent as HIF2a (USF2 ¼ 1.97 and fusion constructs containing the HIF1a DNA binding HIF2a ¼ 1.52-fold induction of total transcripts; Fig. 5E). domain fused to the transactivation domains from the VP16 To test if activation of endogenous HIF target gene could or E2F1 transcription factors were used to activate CA9P/ regulate AS of the PDK1 minigene, HIF-binding site on the PDK1 splicing reporter (Fig. 5C). HIF1aDBD/VP16 PAI1 promoter was mutated to produce the PAI1PmHRE/ induced the expression of PDK1FL (12.2 fold) and PDK1 minigene (Fig. 5F). Interestingly, although HIF2a PDK1DE4 (6.4 fold), and enhanced the FL/DE4 ratio by was not able to activate the expression of PAI1PmHRE/ 1.9 fold (Fig. 5D). Similarly, HIF1aDBD/E2F1 induced PDK1 as assessed by total as well as FL transcripts, HIF2a PDK1FL (21.7 fold) and PDK1DE4 (9.4 fold), and increased the FL/DE4 ratio by reducing the levels of the increased FL/DE4 expression ratio by 2.3 fold (Fig. 5D). PDK1 DE4 transcript (Fig. 5F).

1240 Mol Cancer Res; 12(9) September 2014 Molecular Cancer Research

Hypoxia Regulates Alternative Splicing

A 72 bp 60 bp 185 bp D CA9P/PDK1 E3 E4 E5 * 30 * Fold ratio change 449 bp 1,702 bp * * Ctrl 1.00 25 HIF1α 2.34 CA9P Vs His E3 E4/5 20 HIF1αDBD/VP16 1.90 * α *P < 0.05 * HIF1 DBD/E2F1 2.32 * P < 0.01 15 * E3 E4 E5 * * * * * HBS Vector R * total 10 * * * E3/5 * * PAI1P 5 * E4/5 E3 0 Normalized fold expression Normalized fold PDK1FL Δ PDK1 Total E3 E4 E5 PDK1 E4 UBS HBS Vector R total E PAIP/PDK1 E3/5 Fold ratio change 5 * 293T Hep3B * Ctrl 1.00 HIF2α 2.08 B 4 FL USF2 –1.12 Δ E4 CA9P/PDK1 * 3 * * * * 2 * C 1 0 PDK1FL PDK1 ΔE4 PDK1 Total F Fold ratio PAIPmHRE/PDK1 change 3 Ctrl 1.00 DBD/E2F1 Flag Flag DBD/VP16 Flag * 2.5 * HIF2α 1.66 * USF2 2 –1.09 Ctrl HIF1 α HIF1 α 130 kDa HIF1 α 1.5 * 95 kDa 1 72 kDa anti- 0.5 56 kDa Flag 0

Normalized fold expression Normalized fold expression PDK1FL PDK1ΔE4 PDK1 Total 43 kDa

Figure 5. Transcription activation is not sufficient to regulate splicing of a PDK1 minigene. A, schematic of a PDK1 splicing reporter driven by CA9 or PAI1 promoter. Arrows, primers used for RT-PCR (E3/vector R) and qRT-PCR (E3/5, E4/5, or total with Vector R for DE4, FL, or total transcript respectively). B, RT- PCR analysis of PDK1 transcripts expressed from CA9P/PDK1 in 293T and Hep3B cells. C, Western blot analysis of Flag-tagged HIF1a, HIF1aDBD/ VP16TAD, and HIF1aDBD/E2F1TAD proteins in 293 T cells for experiments whose results are presented in Fig. 5D. HIF1aDBD/E2F1TAD construct produces two proteins. D, qRT-PCR analysis of minigene-specific PDK1 isoforms in 293 T cells cotransfected with CA9P/PDK1 minigene and HIF1a or HIF1aDBD expression constructs. E, qRT-PCR analysis of minigene-specific PDK1 isoforms in 293 T cells cotransfected with PAI1P/PDK1 minigene and HIF2a or USF2 expression constructs. F, qRT-PCR analysis of minigene-specific PDK1 isoforms in 293 T cells cotransfected with PAI1PmHRE/PDK1 minigene and HIF2a or USF2 expression constructs.

To further validate that transcription activation of endog- 3.29 fold. Importantly, cotransfected HIF1a and HIF2a enous HIF target genes increases the PDK1 FL/DE4 ratio, increased the expression of endogenous HIF target genes, the PDK1 splicing reporter was placed under the control of a including CA9 and LOX (data not shown), but not the promoter containing ﬁve copies of the Gal4 DNA binding expression of G5/PDK1 (Supplementary Fig. S4C, same element (Supplementary Fig. S4A, 5xUAS). Fusion con- amount of PDK1 total transcripts with or without HIF). structs containing the Gal4 DNA binding domain fused to These data demonstrated that activation of endogenous HIF the transactivation domains of normoxia active HIF1a, target genes contributes to AS of the PDK1 minigene. VP16, or E2F1 were used to activate the G5P/PDK1 minigene in the presence or absence of normoxia-active Discussion HIF1a and HIF2a that activate endogenous HIF target Most analyses of hypoxia-mediated gene expression genes (Supplementary Fig. S4B). Interestingly, although changes are conducted at the gene level using traditional Gal4/HIF1aTAD, Gal4/VP16TAD, and Gal4/E2F1TAD microarrays (15, 23–26). However, due to signiﬁcant func- increased the PDK1 FL/DE4 ratio by 1.27, 2.44, and 1.47 tional difference of the proteins encoded by RNA isoforms fold, activation of endogenous HIF target genes further and the fact that a majority of genes express multiple RNA increased the PDK1 FL/DE4 ratio to 2.04, 3.28, and isoforms (5), gene expression analysis by exon array or RNA

www.aacrjournals.org Mol Cancer Res; 12(9) September 2014 1241

Sena et al.

deep sequencing is necessary to fully understand gene increased FL/exon-skipping ratio of several HIF target genes, expression programs. Our exon array analysis in normoxic including CA9 and ANGPTL4, is the result of AS, not and hypoxic Hep3B cells not only confirms hypoxic induc- preferential degradation of exon-skipping isoforms. More- tion of previously identified hypoxia-inducible genes, but over, we determined that HIF activity, but not hypoxic stress also reveals significant roles of hypoxia in regulating RNA per se is necessary and sufficient to regulate AS of CA9, splicing of genes whose transcription are induced by hypoxia PDK1, and ANGPTL4 pre-mRNAs. Furthermore, we or genes whose transcription are not changed or reduced by found that activation of endogenous HIF target genes hypoxia. contributes to AS of PDK1 minigenes. Here, we first reported novel hypoxia-regulated genes/ Our studies also indicated that there is an opposing effect pathways that cannot be identified using traditional micro- on HIF target genes versus non-HIF target genes for RNA arrays as expression levels of these genes are typically not splicing in which hypoxia primarily promotes exon inclusion changed by hypoxia or reduced by hypoxia. In addition, we of HIF target genes, but inhibits exon inclusion of genes found that hypoxia primarily promotes exon skipping of these repressed under hypoxia. This opposing effect is similarly non-HIF target genes in Hep3B cells, a result consistent with observed for gene transcription and protein translation in what is reported in endothelial cells (6, 7). We proposed that which, transcription and protein translation of non-HIF increased exon skipping of these non-HIF target genes also target genes are generally reduced, whereas HIF target genes have functional consequences. For example, genes involved in are transcriptionally induced and protein translation of genes ATP binding and protein kinase activity are not hypoxia involved in hypoxia response such as HIF1A, HIF2A, and induced, but exhibit AS. Increased exon skipping of these VEGF are maintained due to internal ribosome entry sites genes in hypoxic Hep3B cells, presumably acts to reduce ATP (IRES). Our data indicated that exon inclusion of HIF target usage to maintain cellular ATP levels. AS of some of these non genes is due to HIF activity, but not hypoxic stress. However, hypoxia-induced genes are not associated with enriched path- we do not know if exon skipping of hypoxia-repressed genes ways identified by DAVID Bioinformatics; however, their is dependent on HIF activity or hypoxic stress. This is a very splicing are also altered and expected to be functionally interesting question that should be assessed in future studies. important. For example, the KRAS 4B isoform (exon 4a is In summary, this study suggests that hypoxia regulates AS skipped in KRAS 4B, but included in KRAS 4A) is induced in of hypoxia-induced genes, hypoxia-reduced genes, and genes hypoxic Hep3B, thus reducing the KRAS 4A/4B ratio. whose transcription is not changed by hypoxia. In addition, Interestingly, a decreased KRAS 4A/4B ratio is often observed HIF activity, not hypoxic stress is found to be necessary and in human colorectal cancer, but not in normal colon (27, 28). sufficient to regulate AS of a subset of hypoxia-inducible Furthermore, decreased KRAS 4A/4B ratios through reduc- genes. Moreover, activation of endogenous HIF target genes tion of KARS 4A expression or increased KRAS 4B expression contributes to AS of some HIF target genes. These findings promotes DMH-induced colonic tumorigenesis in in vivo significantly increase our understanding of how cells regulate mouse models independent of KRAS mutations (29), dem- gene transcription as well as RNA splicing to adapt to a onstrating the functional consequence of reduced KRAS 4A/ hypoxic microenvironment. In the future, it will be inter- 4B ratio in tumor biology. esting to examine the role of hypoxia in regulating RNA More importantly, we reported here, for the first time the splicing in primary tumors. role of hypoxia in regulating RNA splicing of hypoxia- inducible genes. We found that hypoxia promotes exon Disclosure of Potential Conflicts of Interest fl inclusion for 75% of Hx AS genes. Although no functional No potential con icts of interest were disclosed. differences have been examined for the isoforms of most HIF Authors' Contributions target genes, FL CA9 protein was reported to be a mem- Conception and design: J.A. Sena, L. Wang, C.-J. Hu brane-associated, tumor-promoting protein by regulating Development of methodology: J.A. Sena, L. Wang, C.-J. Hu intercellular pH via transporting protons out of the cells Acquisition of data (provided animals, acquired and managed patients, provided D facilities, etc.): J.A. Sena, L. Wang during hypoxia. In contrast, the CA9 E89 protein is a Analysis and interpretation of data (e.g., statistical analysis, biostatistics, compu- soluble protein and is not able to transport protons out of tational analysis): J.A. Sena, L. Wang, C.-J. Hu Writing, review, and or revision of the manuscript: J.A. Sena, L.E. Heasley, the cells (17). Thus, it makes sense that CA9 FL is prefer- C.-J. Hu entially induced by hypoxia. In addition, preferential induc- Administrative, technical, or material support (i.e., reporting or organizing data, tion of nerve growth factor receptor TrkA DE679 isoform, constructing databases): J.A. Sena, C.-J. Hu but not FL leads to nerve growth factor–independent recep- Study supervision: C.-J. Hu tor activity and neuroblastoma tumor-promoting activity fi Grant Support (30). Again, speci c promotion of Cysteine-Rich protein 61 This work was supported by the National Cancer Institute (RO1CA134687, FL, but not intron-3 retaining isoform (producing no to C.-J. Hu) and Cancer League of Colorado (to C.-J. Hu). J. A. Sena was supported by "Research Supplemental to Promote Diversity" (RO1CA134687-S3 and functional protein) promotes tumor angiogenesis (31). RO1CA134687-S4). While isoform ratio changes of several HIF target genes The costs of publication of this article were defrayed in part by the payment of page have been reported under hypoxia (17, 30–34), none of charges. This article must therefore be hereby marked advertisement in accordance with these studies have addressed the molecular mechanisms 18 U.S.C. Section 1734 solely to indicate this fact. about how hypoxia regulates isoform ratio changes of HIF Received March 19, 2014; revised April 21, 2014; accepted May 6, 2014; target genes. We demonstrated in this study that the published OnlineFirst May 21, 2014.

1242 Mol Cancer Res; 12(9) September 2014 Molecular Cancer Research

Hypoxia Regulates Alternative Splicing

References 1. Finger EC, Giaccia AJ. Hypoxia, inflammation, and the tumor micro- 19. Maxwell PH, Wiesener MS, Chang GW, Clifford SC, Vaux EC, Cock- environment in metastatic disease. Cancer Metastasis Rev 2010;29: man ME, et al. The tumour suppressor protein VHL targets hypoxia- 285–93. inducible factors for oxygen-dependent proteolysis. Nature 1999;399: 2. Kaelin WG Jr. Cancer and altered metabolism: potential importance of 271–5. hypoxia-inducible factor and 2-oxoglutarate-dependent dioxy- 20. Auboeuf D, Dowhan DH, Kang YK, Larkin K, Lee JW, Berget SM, et al. genases. Cold Spring Harb Symp Quant Biol 2011;76:335–45. Differential recruitment of nuclear receptor coactivators may deter- 3. Keith B, Johnson RS, Simon MC. HIF1alpha and HIF2alpha: sibling mine alternative RNA splice site choice in target genes. Proc Natl Acad rivalry in hypoxic tumour growth and progression. Nat Rev Cancer Sci U S A 2004;101:2270–4. 2012;12:9–22. 21. Auboeuf D, Dowhan DH, Li X, Larkin K, Ko L, Berget SM, et al. CoAA, a 4. Semenza GL. Hypoxia-inducible factors in physiology and medicine. nuclear receptor coactivator protein at the interface of transcriptional Cell 2012;148:399–408. coactivation and RNA splicing. Mol Cell Biol 2004;24:442–53. 5. Wang ET, Sandberg R, Luo S, Khrebtukova I, Zhang L, Mayr C, et al. 22. Cheng D, Cote J, Shaaban S, Bedford MT. The arginine methyltrans- Alternative isoform regulation in human tissue transcriptomes. Nature ferase CARM1 regulates the coupling of transcription and mRNA 2008;456:470–6. processing. Mol Cell 2007;25:71–83. 6. Weigand JE, Boeckel JN, Gellert P, Dimmeler S. Hypoxia-induced 23. Hu CJ, Iyer S, Sataur A, Covello KL, Chodosh LA, Simon MC. Differ- alternative splicing in endothelial cells. PloS ONE 2012;7:e42697. ential regulation of the transcriptional activities of hypoxia-inducible 7. Hang X, Li P, Li Z, Qu W, Yu Y, Li H, et al. Transcription and splicing factor 1 alpha (HIF-1alpha) and HIF-2alpha in stem cells. Mol Cell Biol regulation in human umbilical vein endothelial cells under hypoxic 2006;26:3514–26. stress conditions by exon array. BMC Genomics 2009;10:126. 24. Wang V, Davis DA, Haque M, Huang LE, Yarchoan R. Differential gene 8. Emig D, Salomonis N, Baumbach J, Lengauer T, Conklin BR, Albrecht up-regulation by hypoxia-inducible factor-1alpha and hypoxia-induc- M. AltAnalyze and DomainGraph: analyzing and visualizing exon ible factor-2alpha in HEK293T cells. Cancer Res 2005;65:3299–306. expression data. Nucleic Acids Res 2010;38:W755–62. 25. Sowter HM, Raval RR, Moore JW, Ratcliffe PJ, Harris AL. Predominant 9. Chang TY, Li YY, Jen CH, Yang TP, Lin CH, Hsu MT, et al. easyExon-a role of hypoxia-inducible transcription factor (Hif)-1alpha versus Hif- Java-based GUI tool for processing and visualization of Affymetrix 2alpha in regulation of the transcriptional response to hypoxia. Cancer exon array data. BMC Bioinformatics 2008;9:432. Res 2003;63:6130–4. 10. Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment 26. Elvidge GP, Glenny L, Appelhoff RJ, Ratcliffe PJ, Ragoussis J, Gleadle tools: paths toward the comprehensive functional analysis of large JM. Concordant regulation of gene expression by hypoxia and 2- gene lists. Nucleic Acids Res 2009;37:1–13. oxoglutarate-dependent dioxygenase inhibition: the role of HIF- 11. Sena JA, Wang L, Pawlus MR, Hu CJ. HIFs enhance the transcriptional 1alpha, HIF-2alpha, and other pathways. J Biol Chem 2006;281: activation and splicing of adrenomedullin. Mol Cancer Res 2014;12: 15215–26. 728–41. 27. Plowman SJ, Berry RL, Bader SA, Luo F, Arends MJ, Harrison DJ, et al. 12. Hu CJ, Sataur A, Wang L, Chen H, Simon MC. The N-terminal K-ras 4A and 4B are co-expressed widely in human tissues, and their transactivation domain confers target gene specificity of hypoxia- ratio is altered in sporadic colorectal cancer. J Exp Clin Cancer Res inducible factors HIF-1alpha and HIF-2alpha. Mol Biol Cell 2007;18: 2006;25:259–67. 4528–42. 28. Abubaker J, Bavi P, Al-Haqawi W, Sultana M, Al-Harbi S, Al-Sanea N, 13. Pawlus MR, Wang L, Ware K, Hu CJ. Upstream stimulatory factor 2 and et al. Prognostic significance of alterations in KRAS isoforms KRAS- hypoxia-inducible factor 2alpha (HIF2alpha) cooperatively activate 4A/4B and KRAS mutations in colorectal carcinoma. J Pathol HIF2 target genes during hypoxia. Mol Cell Biol 2012;32:4595–610. 2009;219:435–45. 14. Warnecke C, Zaborowska Z, Kurreck J, Erdmann VA, Frei U, Wiesener 29. Luo F, Ye H, Hamoudi R, Dong G, Zhang W, Patek CE, et al. K-ras exon M, et al. Differentiating the functional role of hypoxia-inducible factor 4A has a tumour suppressor effect on carcinogen-induced murine (HIF)-1alpha and HIF-2alpha (EPAS-1) by the use of RNA interference: colonic adenoma formation. J Pathol 2010;220:542–50. erythropoietin is a HIF-2alpha target gene in Hep3B and Kelly cells. 30. Tacconelli A, Farina AR, Cappabianca L, Desantis G, Tessitore A, Faseb J 2004;18:1462–4. Vetuschi A, et al. TrkA alternative splicing: a regulated tumor-promot- 15. Hu CJ, Wang LY, Chodosh LA, Keith B, Simon MC. Differential roles of ing switch in human neuroblastoma. Cancer Cell 2004;6:347–60. hypoxia-inducible factor 1alpha (HIF-1alpha) and HIF-2alpha in hyp- 31. Hirschfeld M, zur Hausen A, Bettendorf H, Jager M, Stickeler E. oxic gene regulation. Mol Cell Biol 2003;23:9361–74. Alternative splicing of Cyr61 is regulated by hypoxia and significantly 16. Pawlus MR, Wang L, Hu CJ. STAT3 and HIF1alpha cooperatively changed in breast cancer. Cancer Res 2009;69:2082–90. activate HIF1 target genes in MDA-MB-231 and RCC4 cells. Onco- 32. Gang H, Hai Y, Dhingra R, Gordon JW, Yurkova N, Aviv Y, et al.A gene 2014;33:1670–9. novel hypoxia-inducible spliced variant of mitochondrial death gene 17. Barathova M, Takacova M, Holotnakova T, Gibadulinova A, Ohrada- Bnip3 promotes survival of ventricular myocytes. Circ Res 2011;108: nova A, Zatovicova M, et al. Alternative splicing variant of the hypoxia 1084–92. marker carbonic anhydrase IX expressed independently of hypoxia 33. Kikuchi Y, Kakeya T, Nakajima O, Sakai A, Ikeda K, Yamaguchi N, et al. and tumour phenotype. Br J Cancer 2008;98:129–36. Hypoxia induces expression of a GPI-anchorless splice variant of the 18. Lejeune F, Maquat LE. Mechanistic links between nonsense-mediated prion protein. FEBS J 2008;275:2965–76. mRNA decay and pre-mRNA splicing in mammalian cells. Curr Opin 34. Elias AP, Dias S. Microenvironment Changes (in pH) Affect VEGF Cell Biol 2005;17:309–15. Alternative Splicing. Cancer Microenviron 2008;1:131–9.

www.aacrjournals.org Mol Cancer Res; 12(9) September 2014 1243

Hypoxia Regulates Alternative Splicing of HIF and non-HIF Target Genes

Johnny A. Sena, Liyi Wang, Lynn E. Heasley, et al.

Mol Cancer Res 2014;12:1233-1243. Published OnlineFirst May 21, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-14-0149

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2015/06/05/1541-7786.MCR-14-0149.DC1

Cited articles This article cites 34 articles, 13 of which you can access for free at: http://mcr.aacrjournals.org/content/12/9/1233.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/12/9/1233.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/12/9/1233. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.