Prognostic Potential of DNA Methylation and Transcript Levels of HIF1A and EPAS1 in Colorectal Cancer

Published OnlineFirst May 13, 2014; DOI: 10.1158/1541-7786.MCR-14-0054

Molecular Cancer Chromatin, Gene, and RNA Regulation Research

Agnieszka Anna Rawłuszko-Wieczorek1, Karolina Horbacka2, Piotr Krokowicz2, Matthew Misztal1, and Paweł Piotr Jagodzinski 1

Abstract Hypoxic conditions during the formation of colorectal cancer may support the development of more aggressive tumors. Hypoxia-inducible factor (HIF) is a heterodimeric complex, composed of oxygen-induced HIFa and constitutively expressed HIFb subunits, which mediates the primary transcriptional response to hypoxic stress. Among HIFa isoforms, HIF1a (HIF1A) and endothelial PAS domain–containing protein 1 (EPAS1) are able to robustly activate hypoxia-responsive gene signatures. Although posttranslational regulation of HIFa subunits is well described, less is known about their transcriptional regulation. Here, molecular analysis determined that EPAS1 mRNA was significantly reduced in primary colonic adenocarcinoma specimens compared with histopathologically nonneoplastic tissue from 120 patients. In contrast, no difference in HIF1A mRNA levels was observed between cancerous and noncancerous tissue. Bisulfite DNA sequencing and high-resolution melting analysis identified significant DNA hypermethylation in the EPAS1 regulatory region from cancerous tissue compared with nonneoplastic tissue. Importantly, multivariate Cox regression analysis revealed a high HR for patients with cancer with low EPAS1 transcript levels (HR, 4.91; 95% confidence interval, CI, 0.42–56.15; P ¼ 0.047) and hypermethylated EPAS1 DNA (HR, 33.94; 95% CI, 2.84–405.95; P ¼ 0.0054). Treatment with a DNA methyltransferase inhibitor, 5-Aza-20-deoxycytidine (5-aza-dC/Decitabine), upregulated EPAS1 expression in hypoxic colorectal cancer cells that were associated with DNA demethylation of the EPAS1 regulatory region. In summary, EPAS1 is transcriptionally regulated by DNA methylation in colorectal cancer. Implications: DNA methylation and mRNA status of EPAS1 have novel prognostic potential for colorectal cancer. Mol Cancer Res; 12(8); 1112–27. 2014 AACR.

Introduction mechanism (1, 2). In normoxic conditions, HIFa is hydrox- ylated at specific residues, which results in proteasomal Immense proliferation of tumor cells and their inadequate a perfusion results in hypoxia, which is a hallmark of many degradation (2). There are three isoforms referred to as: HIF1a, HIF2a (officially designated as endothelial PAS solid tumors, including colorectal cancer (1). Mechanisms – — a by which tumor cells alter their expression profile to adjust to domain containing protein 1 EPAS1) and HIF3 , which low oxygen tension involve hypoxia-inducible factor (HIF; are encoded by the HIF1A, EPAS1, and HIF3A genes, respectively (2). Among them, HIF1a or EPAS1 may bind ref. 1). HIF is a heterodimeric transcription factor assembled b from a and b subunits. It recognizes the hypoxia response together with and other coactivators to HRE and activate element (HRE) and promotes expression of many genes HIF-dependent gene transcription (2). Many articles describe aberrant HIF1a or EPAS1 protein levels and their involved in glucose metabolism, angiogenesis, or metastasis – (1). The b subunit of HIF is constitutively expressed, association with colorectal cancer prognosis (3 10). How- whereas HIFa is mainly controlled by a posttranslational ever, there are only few articles about the same issues on the HIF1A and EPAS1 mRNA level (8, 11, 12). Moreover, relatively little is known about transcriptional regulation of Authors' Affiliations: Departments of 1Biochemistry and Molecular Biol- HIF1A and EPAS1. It should be noted that both of them 2 ogy and General and Colorectal Surgery, Poznan University of Medical possess a CpG island in their promoter region. DNA Sciences, Poland methylation within the CpG island associates the gene Note: Supplementary data for this article are available at Molecular Cancer transcriptional repression, and aberrant DNA methylation Research Online (http://mcr.aacrjournals.org/). patterns are observed during colorectal tumorigenesis (13). Corresponding Author: Agnieszka Anna Rawłuszko-Wieczorek, Poznan University of Medical Sciences, S´wiecickiego˛ 6 Street, Poznan 60-781, To date, only one article indicates DNA methylation of Poland. Phone: 486-1854-6516; Fax: 48618546510; E-mail: HIF1A in colorectal cancer (14), and there is no scientific [email protected] reports about this type of epigenetic regulation of EPAS1 doi: 10.1158/1541-7786.MCR-14-0054 expression. Therefore, we aimed to examine DNA methyl- 2014 American Association for Cancer Research. ation and mRNA levels of the HIF1A and EPAS1 genes in

1112 Mol Cancer Res; 12(8) August 2014

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

primary cancerous and histopathologically unchanged colo- Table 1. Demographic and histopathologic rectal tissues from the same 120 patients. Moreover, we classification of patients with colorectal cancer evaluated the impact of EPAS1 DNA methylation and transcript level in colorectal cancer tissue with respect to 0 patient survival. We also assessed the effect of 5-Aza-2 - Number of deoxycytidine (5-dAzaC), an inhibitor of DNA methyltrans- Features patients ferases, on DNA methylation level of the EPAS1 gene and on Total number of patients 120 the EPAS1 transcript, as well as protein level in HCT116 Gender (female/male) 54/66 and DLD-1 colorectal cancer cells under hypoxic and Mean (SD) age at radical surgical 67.94 12.45 normoxic conditions. resection of colon (y) Colorectal cancer localization Materials and Methods Proximal colon (cecum to 45 transverse) Antibodies and reagents Distal colon (splenic flexure 22 a Rabbit polyclonal (Rp) anti-HIF1 (NB100-449) and to sigmoid) anti-EPAS1 (NB100-122) Abs were provided by Novus Rectum 53 Biologicals. Rp anti-GAPDH Ab (FL-335) and goat anti- Histologic grade – rabbit horseradish peroxidase (HRP) conjugated Ab were G1 7 provided by Santa Cruz Biotechnology. 5-dAzaC was pur- G2 77 chased from Sigma-Aldrich Co. G3 36 TNM classification Patient material I17 Primary colonic adenocarcinoma tissues were collected IIA 47 between June 2009 and March 2013 from 120 patients who IIC 6 underwent radical surgical resection of the colon at the IIIA 3 Department of General and Colorectal Surgery, Poznan IIIB 34 University of Medical Sciences, Poland (Table 1). The IIIC 13 histopathologically unchanged colonic mucosa located at least 10 to 20 cm away from the cancerous lesions was obtained from the same patients. One set of samples was Branch, Gliwice, Poland. These cells were cultured in immediately snap-frozen in liquid nitrogen and stored at DMEM GibcoBRL containing 10% heat-inactivated FBS 80C until DNA/RNA isolation. The other set of samples and 2 mmol/L glutamine. To determine the effect of 5- was directed for histopathologic examination. Histopatho- dAzaC on DNA methylation, transcript and protein level of logic classification was performed by an experienced pathol- the HIF1A and EPAS1 genes, the HCT116 and DLD-1 cells ogist. No patients received preoperative chemo- or radiother- were cultured for 24 hours in DMEM GibcoBRL supple- apy. An informed consent was obtained from all participating mented with 10% FBS from Sigma-Aldrich Co. Cells were individuals. The procedures of the study were approved by then cultured under normoxic or hypoxic (1% O2) condi- the Local Ethical Committee of Poznan University of Med- tions, either in the absence or in the presence of 5-dAzaC. 5- ical Sciences. dAzaC was at concentrations of 1.00 and 5.00 mmol/L for 6, 24, and 48-hour time frames. Hypoxic conditions were Measurement of overall and disease-free survival achieved using a MCO-18M multi-gas cell culture incuba- Follow-up data were available for 80 patients, who were tor, Sanyo, modified to permit flushing the chamber with a fi observed from August 11, 2009, until death or October 15, humidi ed mixture of 5% CO2, 94% N2. These cells were 2013, whichever came first. Nine patients were excluded used for DNA and RNA isolation, quantitative real-time from further analysis because they did not fulfill criteria given PCR (qRT-PCR), Western blotting, and high-resolution below. Disease-free survival (DFS) is defined as the time melting (HRM) analysis. elapsed from surgery to the first occurrence of any of the following events: recurrence or distant metastasis of colo- DNA isolation and bisulfite modification rectal cancer, development of a second noncolorectal malig- Genomic DNA from tissues of patients with colorectal nancy. In overall survival (OS) analysis, deaths from any cancer and cell lines were isolated using the DNA Mam- cause without clinical documentation of cancer related event malian Genomic Purification Kit purchased from Sigma- were excluded from the study. Aldrich Co. A total of 500 ng of genomic DNA was subjected to bisulfite conversion of cytosine to uracil, Cell culture according to the EZ DNA Methylation Kit procedure from DLD-1 colon cancer cells were obtained from the ATCC, Zymo Research Corporation. The position of the CpG and HCT116 cells were kindly provided by the Department islands and binding sites of transcription factors located in of Experimental and Clinical Radiobiology, Maria Skło- the HIF1A and EPAS1 promoter were determined by online dowska-Curie Cancer Center, Institute of Oncology programs (15–17).

www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1113

Rawłuszko-Wieczorek et al.

DNA methylation evaluation by bisulfite sequencing Chomczynski and Sacchi (26). The RNA samples were The DNA fragments containing CpG dinucleotides locat- quantified and reverse-transcribed into cDNA. qRT-PCR ed in the promoter region of the HIF1A and EPAS1 genes was carried out in the Light Cycler480 Real-Time PCR were amplified from the bisulfite-modified DNA by the System, Roche Diagnostics GmbH using SYBR Green I as primer pairs (Supplementary Table S1A) complementary to the detection dye. The target cDNA was quantified by the the bisulfite DNA-modified sequences in 5 patients. Clin- relative quantification method using a calibrator for the icopathologic parameters for these patients are given in primary tissues. The calibrator was prepared as a cDNA Supplementary Table S1B. PCR amplification was per- mix from all of the patients' samples, and successive dilutions formed by FastStart Taq DNA Polymerase from Roche were used to create a standard curve as described in Relative Diagnostic GmbH. The PCR products were purified using Quantification Manual Roche Diagnostics GmbH. For the Agarose Gel DNA Extraction Kit, Roche Diagnostic amplification, 1 mL of (total 20 mL) cDNA solution was GmbH with subsequent cloning into pGEM-T Easy Vector added to 9 mL of IQ SYBR Green Supermix, Bio-Rad System I, Promega and transformation into TOPO10 E. coli Laboratories Inc. with primers (Supplementary Table strain cells. The plasmid DNA isolated from five positive S1A). To prevent amplification of sequences from genomic bacterial clones was used for commercial sequencing of the DNA contamination, primers and/or amplicons were cloned fragment of DNA. The results of bisulfite sequencing designed at exon/exon boundaries and covered all gene splice were assessed and presented using BiQ analyzer software and variants. The quantity of HIF1A and EPAS1 transcripts in Bisulfite sequencing Data Presentation and Compilation each sample was standardized by the geometric mean of two (BDPC) web server, respectively (18, 19). internal controls: porphobilinogen deaminase (PBGD) and human mitochondrial ribosomal protein L19 (hMRPL19; DNA methylation assessment by HRM analysis Supplementary Table S1A). The selection of internal control Methylation level of DNA fragments located within the genes was made as previously (27). The HIF1A and EPAS1 CpG island of the HIF1A and EPAS1 genes was determined transcript levels in the patients' tissues were expressed as by RT-PCR amplification of bisulfite-treated DNA, fol- multiplicity of the cDNA concentrations in the calibrator. In lowed by HRM profile analysis by Light Cycler480 Real- HCT116 and DLD-1 cells, transcript levels were presented Time PCR System, Roche Diagnostics GmbH. For PCR as multiplicity of the respective controls. amplification, 1 mL of the bisulfite-treated DNA from patients, HCT116, DLD-1 cells, or standards, and primers Western blotting analysis (Supplementary Table S1A) was added to 19 mL of 5 X Hot HCT116 and DLD-1 cells were treated with lysis RIPA FIREPol EvaGreen HRM Mix, Solis BioDyne Co. Stan- buffer, and proteins were resuspended in the sample buffer dardized solutions of DNA methylation percentage were and separated on 10% Tris-glycine gel using SDS-PAGE. prepared by mixing methylated and nonmethylated bisul- Gel proteins were transferred to a nitrocellulose membrane, fite-treated DNA from Human Methylated/Non-methylat- which was blocked with 5% milk in Tris/HCl saline/Tween ed DNA Set, Zymo Research Corp. in different ratios. To buffer. Immunodetection of bands was performed with Rp determine the percentage of methylation, the HRM profiles anti-HIF1a and -EPAS1 Ab, followed by incubation with of patients DNA PCR products were compared with HRM goat anti-rabbit HRP-conjugated Ab. To ensure equal profiles of standard DNA PCR products (20, 21). HRM protein loading of the lanes, the membrane was stripped methylation analysis was performed using Light Cycler480 and incubated with Rp anti-GAPDH Ab (FL-335), followed Gene Scanning software, Roche Diagnostics GmbH. Each by incubation with goat anti-rabbit HRP-conjugated Ab. PCR amplification and HRM profile analysis was performed The bands were revealed using SuperSignal West Femto in triplicate. The HRM results were compared with those Chemiluminescent Substrate, Thermo Fisher Scientific, and obtained from bisulfite sequencing for analyzed genes in Biospectrum Imaging System 500, UVP Ltd. The amount of reconstituted samples. A similar pattern of DNA methyla- analyzed proteins was presented as the protein-to-GAPDH tion was observed between these two methods. The meth- band optical density ratio. For HCT116 and DLD-1 cells ylation for each patient was presented as a percentage of cultured in the absence of 5-dAzaC, the ratio of EPAS1 to methylation in amplified fragments located in the CpG GAPDH was assumed to be 1. island of HIF1A and EPAS1. Because low level of methylation may not demonstrate significant biologic effects and we Statistical analysis were not able to quantify all the CpG dinucleotides within The normality of the observed patient data distribution the analyzed CpG island, the percentage results were divided was assessed by the Shapiro–Wilk test, and the Mann– into three groups: 0% to 1% methylation, 1% to 10% Whitney U test was used to compare the median values. methylation, and 10% to 100% methylation for statistical The c2 test was used to examine the significance in DNA analysis (22–25). methylation. Fisher exact probability test was used for data that do not fulfilled criteria for Cochran theorem. To Reverse transcription and quantitative real-time PCR evaluate the association between different ranges of DNA analysis methylation (0%–1% methylation, 1%–10% methylation, Total RNA from tissues of patients with colorectal cancer and 10%–100% methylation) and the ratios of cancerous and cell lines were isolated according to the method of tissue EPAS1 mRNA level to histopathologically unchanged

1114 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

EPAS1 mRNA level, the nonparametric Kruskal–Wallis test colorectal cancer (P < 0.00001; Fig. 1F and Table 2). was used. Survival curves were plotted using the Kaplan– Patients were also stratified by gender, age, histologic grades, Meier method, and survival differences were achieved using and TNM stages for DNA methylation analysis. We the log-rank test. Multivariate Cox proportional hazard observed higher DNA methylation within the analyzed model was used to estimate the adjusted HR. Data groups region of the EPAS1 CpG island in primary cancerous tissue for cell lines were assessed by ANOVA to evaluate if there for a majority of subgroups and no distinctive subgroup were significant differences (P < 0.05) between the groups. biased DNA methylation (Supplementary Table S3). More- For all experimental groups that fulfilled the initial criteria, over, we observed that an increase in DNA methylation level individual comparisons were performed, by post hoc Tukey of EPAS1 in region chr2: 46 526 762-46 526 905 correlated test with the assumption of two-tailed distribution. Statis- to a decrease in the ratio of cancerous to histopathologically tically significant results were indicated by P < 0.05. Statis- unchanged tissue EPAS1 mRNA level (P ¼ 0.0036; Fig. 2). tical analysis was performed with STATISTICA 10.0 software. DNA hypermethylation and low mRNA level of the EPAS1 gene are prognostic factors for patients' OS with colorectal cancer Results To investigate the effect of transcript and DNA methyl- DNA hypermethylation of EPAS1 regulatory region is ation level of EPAS1 on patients' survival, we carried out associated with a decrease in EPAS1 mRNA level in retrospective clinical analysis of 71 patients. The median primary cancerous tissue compared with survival was 36 months (range, 9–51 months). On the basis histopathologically unchanged tissue from patients with of RQ-PCR data, the EPAS1 mRNA level in histopatho- colorectal cancer, whereas there is neither DNA logically unchanged and cancerous tissue was subdivided methylation nor transcript changes of HIF1A into three groups: low, intermediate, and high EPAS1 To compare the HIF1A and EPAS1 transcript and DNA transcript levels. Univariate analysis of OS revealed that methylation levels in the HIF1A and EPAS1 promoter patients with low mRNA expression level of EPAS1 in regions in cancerous and histopathologically unchanged histopathologically unchanged tissue had a significant tissues from 120 patients with colorectal cancer, we used increase in risk of death compared with patients with an RQ-PCR and bisulfite DNA sequencing followed by HRM intermediate and/or high expression level (Fig. 3A). This analysis, respectively. We found significantly lower levels of related to survival: 33 months in low EPAS1 mRNA the EPAS1 transcript (P ¼ 0.000011) in primary cancerous subgroup versus 36 in intermediate and high EPAS1 mRNA than in the histopathologically unchanged tissues in patients subgroups (Fig. 3A). Moreover, the Kaplan–Meier analysis with colorectal cancer (Fig. 1A). Moreover, we observed revealed benefit of a high EPAS1 transcript level in significantly lower levels of the EPAS1 transcript in cancer- histopathologically unchanged tissue of a 7-month median ous tissues in different age groups, genders, colorectal cancer increase in survival compared with the intermediate and 14- localizations, histologic grades, and tumor–node–metastasis month increase compared with the low EPAS1 mRNA (TNM) stages (Supplementary Table S2). There was no subgroup in patients not treated with postoperative chemo- significant difference in the level of the HIF1A transcript therapy (Fig. 3B). Analysis of cancerous tissue disclosed lack between primary cancerous and histopathologically of impact of EPAS1 mRNA level on OS (Fig. 3A and B). unchanged tissues in 120 patients with colorectal cancer Furthermore, there was no evidence of impact of EPAS1 (P ¼ 0.87; Fig. 1A). We also undetected DNA methylation mRNA level on DFS in both the histopathologically within the HIF1A promoter in the analyzed regions (chr14: unchanged and cancerous tissues (Supplementary Fig. 62 161 804-62 162 333 and chr14: 62 162 250-62 163 074 S1). Impact of DNA methylation of EPAS1 regulatory using bisufite sequencing; chr14: 62 161 655-62 161 825 region was done by comparison of two groups: absent DNA and chr14: 62 162 301- 62 162 427 using HRM methylation and present DNA methylation in the EPAS1 analysis; Fig. 1B and C). Moreover, we did not observe gene regulatory region. Of note, 1% to 10% and 10% to DNA methylation in the regulatory region of the EPAS1 100% subgroups were merged into one because of limited gene in cancerous and histopathologically unchanged tissues number of patients in 10% to 100% subgroup (n ¼ 4). We in regions chr2: 46 524 336-46 524 767 and chr2: 46 524 found that patients with DNA hypermethylation of EPAS1 751-46 525 189 using bisulfite sequencing; chr2: 46 524 in cancerous tissue compared with histopathologically 636-46 524 769 and chr2: 46 524 969-46 525 075 using unchanged had shorter OS rate compared with patients HRM analysis (Fig. 1D and E). However, study of EPAS1 with no changes in DNA methylation status (Fig. 3C). gene regulatory region chr2: 46 526 521-46 527 161 Although, result was statistically insignificant, it suggests revealed significant DNA hypermethylation in cancerous that there may have been a reduction in the risk of death tissue compared with histopathologically unchanged tissue, for patients with the hypomethylated EPAS1 gene regu- using bisulfite sequencing in 5 patients (Fig. 1F). In keeping latory region. The analysis in the group of patients without with the bisulfite sequencing data, we observed significantly postoperative chemotherapy and analysis of impact of higher DNA methylation within EPAS1 regulatory region EPAS1 DNA methylation on DFS did not reveal statis- chr2: 46 526 762-46 526 905 in cancerous compared with tically significant data (Fig. 3C and D; Supplementary Fig. histopathologically unchanged tissue from 120 patients with S1). However, multivariate Cox regression analysis with

www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1115

Rawłuszko-Wieczorek et al.

P = 0.87 P = 0.000011 A 5.0 7

4.5 6 4.0 5 3.5

3.0 4

2.5 3

Log [HIF1A mRNA] 2.0 mRNA] Log [EPAS1 2 1.5

1 1.0 Median 0.5 0 25%–75% Histopathologically unchanged tissue Primary cancerous tissue Histopathologically unchanged tissue Primary cancerous tissue Min-Max

B HIF1A.1 chr14: 62 161 804 Histopathologically unchanged tissue chr14: 62 162 333 CpG dinucleotides 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 P1 P2 P3 P4 P5 Cancerous tissue CpG dinucleotides 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 P1 P2 P3 P4 P5 HIF1A.1 chr14: 62 161 655 chr14: 62 161 825 Normalized melting curves

100,000 90,000

80,000 70,000 100% 60,000

50,000 50% 40,000

Relative signal (%) Relative 30,000 20,000 10% Cancerous tissue 10,000 1% 0% Histopathologically unchanged tissue 0,000 74 74.5 75 75.5 76 76.5 77 77.578 78.579 79.5 80 80.5 81 81.5 Temperature (ºC)

C chr14: 62 162 250 HIF1A.2 Histopathologically unchanged tissue chr14: 62 163 074 CpG dinucleotides 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 P1 P2 P3 P4 P5 Cancerous tissue CpG dinucleotides 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 P1 P2 P3 P4 P5 HIF1A.2 chr14: 62 162 301 chr14: 62 162 427 Normalized melting curves

100,000 90,000 100% 80,000 70,000 60,000 50,000 50% 40,000

Relative signal (%) Relative 30,000 20,000 10% Cancerous tissue 1% 0% 10,000 Histopathologically unchanged tissue 0,000 75 75.5 76 76.5 77 77.578 78.5 79.57980 80.5 81 81.5 82 82.5 83 83.5 Temperature (ºC)

Figure 1. DNA methylation of promoter region and transcript levels of HIF1A and EPAS1 in primary cancerous and histopathologically unchanged tissues from patients with colorectal cancer. The cancerous and histopathologically unchanged tissues from 120 patients with colorectal cancer were used for RNA and DNA isolation. A, total RNA was reverse-transcribed, and cDNAs were investigated by RQ-PCR relative quantiﬁcation analysis. The HIF1A and EPAS1 mRNA levels were corrected by the geometric mean of PBGD and hMRPL19 cDNA levels. The amounts of mRNA were expressed as the decimal logarithm of multiples of these cDNA copies in the calibrator. The P value was evaluated by the Mann–Whitney U test. (Continued on the following page.)

1116 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

D EPAS1.1 chr2: 46 524 336 Histopathologically unchanged tissue chr2: 46 524 767 CpG dinucleotides 1234567 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 P1 P2 P3 P4 P5

Cancerous tissue CpG dinucleotides 1234567 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 P1 P2 P3 P4 P5

EPAS1.1 chr2: 46 524 636 chr2: 46 524 769 Normalized melting curves

100,000

90,000

80,000 100%

70,000

60,000

50,000 50% 40,000

Relative singnal (%) Relative 30,000

20,000 10% 1% 10,000 Cancerous tissue 0% Histopathologically unchanged tissue 0,000

72.5 73 73.5 74 74.5 75 75.5 76 76.5 77 77.5 Temperature (ºC) E EPAS1.2 chr2: 46 524 751 Histopathologically unchanged tissue chr2: 46 524 189 CpG dinucleotides 12345678 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 P1 P2 P3 P4 P5

Cancerous tissue CpG dinucleotides 12345678 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 P1 P2 P3 P4 P5

EPAS1.2 chr2: 46 524 969 chr2: 46 525 075 Normalized melting curves

100,000

90,000

80,000

70,000

60,000 100% 50,000 50% 40,000 10% Relative singnal (%) Relative 30,000 1% 20,000 Cancerous tissue

10,000 Histopathologically unchanged tissue 0%

0,000

74.5 75 75.5 76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 Temperature (ºC)

Figure 1. (Continued.) B to F, primary cancerous and histopathologically unchanged tissues from the same patients with colorectal cancer (P1–P5; Supplementary Table S1B) were used for genomic DNA isolation followed by bisulfite conversion of cytosine to uracil. The HIF1A regions containing 49 CpG dinucleotides (chr14: 62 161 804-62 162 333; B) and 70 CpG dinucleotides (ch14: 61 162 250-62 163 074; C) as well EPAS1 regions containing 49 CpG dinucleotides (chr2: 46 524 336-46 524 767; D), 44 CpG dinucleotides (chr2: 46 524 751-46 525 189; E), and 37 CpG dinucleotides (chr2: 46 526 521-46 527 161; F) were then amplified by a pair of primers complementary to the bisulfite DNA-modified sequence (Supplementary Table S1A). (Continued on the following page.) www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1117

Rawłuszko-Wieczorek et al.

F EPAS1.3 chr2: 46 526 521 Histopathologically unchanged tissue chr2: 46 527 161 CpG dinucleotides 12345678 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 P1 P2 P3 P4 P5 Cancerous tissue CpG dinucleotides 12345678 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 P1 P2 P3 P4 P5

EPAS1.3 chr2: 46 526 762 chr2: 46 526 905 Normalized melting curves

100,000

90,000 100% 80,000

70,000 Cancerous tissue 60,000

50,000

40,000 50%

Relative singnal (%) Relative 30,000 10% Histopathologically unchanged tissue 20,000 1% 0% 10,000

0,000

76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 Temperature (ºC)

Figure 1. (Continued.) The PCR products were purified with subsequent cloning into a plasmid vector. Plasmid DNA isolated from five positive bacterial clones was used for commercial sequencing. The results of bisulfite sequencing were assessed and presented using BiQ analyzer software and BDPC web server (18, 19). Black and gray boxes, methylated and unmethylated CpG dinucleotide, respectively. Black rectangles, regions amplified in HRM analysis by specific primers (Supplementary Table S1A). Bottom panels in B to F, HRM profiles of standard and example of patient DNA (patient P1 from bisulfite sequencing). Methylation percentage of three DNA fragments within the HIF1A and EPAS1 CpG island was determined by RT-PCR amplification of bisulfite-treated standard and patient DNA, followed by comparison of their HRM profiles. DNA standards were prepared by mixing different ratios of methylated and nonmethylated bisulfite-treated DNA. HRM methylation analysis was performed using Light Cycler480 Gene Scanning software, Roche Diagnostics GmbH. Each PCR amplification and HRM profile analysis was performed in triplicate.

respect to age, gender, and postoperative chemotherapy 46 526 905 and no DNA methylation in chr2: 46 524 status revealed that mRNA level and DNA methylation 636-46 524 769 and chr2: 46 524 969-46 525 075 (Fig. are both independent prognostic factors for patient's 4A). We revealed a lower level of the HIF1A and EPAS1 survival (Table 3). Low EPAS1 mRNA level in histopath- transcript in HCT116 cells compared with DLD-1 cells in ologically unchanged tissue and DNA hypermethylation both hypoxic and normoxic conditions (Fig. 4B). The in cancerous tissue compared with histopathologically HIF1A transcript level was not induced upon hypoxia in unchanged have signiﬁcant HR equal to 4.91 and both cell lines (Fig. 4B). However, we observed a signif- 33.94, respectively (Table 3). Neither EPAS1 mRNA icant induction of the EPAS1 transcript level upon hyp- status nor DNA methylation were associated with DFS oxia in DLD-1 cells, with no changes in HCT116 cells in multivariate analysis (Table 3). under the same conditions (Fig. 4B). In both analyzed cell lines, hypoxic conditions induced HIF1a and EPAS1 EPAS1 gene regulatory region is hypermethylated in protein level (Fig. 4B). HCT116 colorectal cancer cells in normoxic and hypoxic conditions 5-dAzaC induced DNA demethylation of EPAS1 gene To evaluate DNA methylation and expression level of the regulatory region, EPAS1 transcript, and protein HIF1A and EPAS1 genes in HCT116 and DLD-1 colorectal contents in HCT116 cells; did not affect EPAS1 DNA cancer cells, we performed HRM analysis, RQ-PCR, and methylation or expression level in DLD-1 cells under Western blotting. We observed no DNA methylation of the hypoxic conditions HIF1A promoter region in the analyzed regions using HRM To assess the effect of 5-dAzaC on DNA methylation and analysis under hypoxic and normoxic conditions in the EPAS1 gene expression level, we used HRM analysis, HCT116 and DLD-1 cells (Fig. 4A). Moreover, we detected RQ-PCR, and Western blotting. We observed no effect of DNA hypomethylation in the EPAS1 CpG island in DLD-1 5-dAzaC treatment on DNA methylation status in the cells(Fig.4A).Nonetheless,wedetectedahighlevelof analyzed region of the EPAS1 promoter in DLD-1 cells DNA methylation in HCT116 in the chr2: 46 526 762- under hypoxic and normoxic conditions (Fig. 4C). On the

1118 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

Table 2. DNA methylation level of the EPAS1 gene regulatory region in primary cancerous tissue and histopathologically unchanged tissue sample from patients with colorectal cancer

0% to 1% 1% to 10% 10% to 100% EPAS1 methylation methylation methylation Pa Histopathologically unchanged tissue 113 5 2 <0.00001 Primary cancerous tissue 77 29 14

NOTE: The primary cancerous and histopathologically unchanged tissue samples from the same patients were used for genomic DNA isolation, followed by bisulfite conversion of cytosine to uracil. The DNA fragments of the CpG island were then amplified pairs of primers complementary to the bisulfite DNA-modified sequence (Supplementary Table S1A). To determine the percentage of methylation, the HRM profiles of the patients' DNA PCR products were compared with HRM profiles of the prepared standard PCR products (Fig. 1F). DNA methylation of the EPAS1 regulatory region for each patient was calculated as a percentage of methylation in amplified fragment. ac2 test. contrary, using HRM analysis, we noticed significant DNA patient prognosis by three independent studies, but three demethylation in chr2: 46 526 762-46 526 905 region of the consecutive articles showed lack of such association in EPAS1 CpG island in HCT116 cells cultured for 48 hours in colorectal cancer (3–9). Moreover, Yoshimura and collea- the presence of 5.00 mmol/L 5-dAzaC under both hypoxic gues demonstrated strong positive immunohistochemical and normoxic conditions (Fig. 4C). The changes in DNA staining of EPAS1 in advanced colorectal cancer compared methylation level were accompanied by 5-dAzaC–induced with low-grade tumors (6). Nonetheless, a study conducted expression of EPAS1 in HCT116 cells. We observed that 5- by two other research teams described the opposite results dAzaC resulted in a progressive increase in the EPAS1 (9, 10). The HIF1A and EPAS1 transcript level was not transcript level in HCT116 cells. For HCT116, we found analyzed extensively. In the esophageal squamous cell car- approximately a 2.11- and 2.88-fold increase in the EPAS1 cinoma, pancreatic, gastric, cervical and colon cancers, transcript level at 48 hours of incubation under normoxic and hypoxic conditions, respectively (Fig. 4D). Despite the absence of DNA methylation in DLD-1, we noticed pro- P = 0.036 gressive increase in mRNA level in DLD-1 cells under 1.6 normoxic conditions (Fig. 4D). Alterations in the EPAS1 1.5 transcript level in HCT116 cells were associated with an 1.4 increased EPAS1 protein level in hypoxic conditions (Fig. 1.3 4D). Densitometric analysis of Western blotting bands 1.2 indicated an approximately 2.31-fold increase in EPAS1 1.1 protein level in HCT116 cells, incubated with 5.00 mmol/L 1.0 5-dAzaC for 48 hours as compared with the respective 0.9 controls under hypoxic conditions (Fig. 4D). These changes 0.8 were not observed for the HCT116 cells under normoxic 0.7 conditions (Fig. 4D). Incubation of DLD-1 cells with 5- to mRNA level Ratio of cancerous EPAS1 0.6 dAzaC at various concentrations for different time periods 0.5 Median histopathologically unchanged tissue EPAS1 mRNA level histopathologically unchanged tissue EPAS1 0%–1% 1%–10% 10%–100% 25%–75% did not significantly increase EPAS1 protein content under Methylation Min–Max either hypoxic or normoxic conditions (Fig. 4D).

Figure 2. Ratio of cancerous to histopathologically unchanged tissue Discussion EPAS1 mRNA level in three ranges of EPAS1 DNA methylation status: 0% to 1%, 1% to 10%, and 10% to 100%. Methylation percentage of three HIFa initiates adaptive responses that maintain proper DNA fragments within the EPAS1 CpG island (Supplementary Table S1A) metabolism and pH homeostasis, thereby reinforcing tumor was determined by RT-PCR amplification of bisulfite-treated standard growth and metastasis (1). Many immunohistochemical and patient DNA, followed by comparison of their HRM profiles. DNA a methylation for each patient was calculated as a percentage of studies reported a correlation between the HIF1 or EPAS1 methylation in amplified fragments. The samples were divided into three protein and patient survival, response to chemotherapy, groups for statistical analysis: 0% to 1% methylation, 1% to 10% expression of oncogenes and genes involved in angiogenesis methylation, and 10% to 100% methylation (27). To evaluate the in breast, head and neck, cervix, gastric, hepatocellular, and statistically significant difference in the ratio of cancerous EPAS1 mRNA glioma cancers (28–32). The level of HIF1a and EPAS1 level to histopathologically unchanged tissue EPAS1 mRNA level between the three DNA methylation ranges (0%–1% methylation, protein was also determined in colorectal cancer, but the 1%–10% methylation, and 10%–100% methylation), the nonparametric results are inconclusive. HIF1a was correlated with poor Kruskal–Wallis test was used.

www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1119

Rawłuszko-Wieczorek et al.

A Histopathologically unchanged tissue Cancerous tissue Event Censored Event Censored

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4 Survival probability Survival probability

0.2 0.2

P = 0.022 P = 0.32 Low EPAS1 mRNA 0.0 0.0 lnt EPAS1 mRNA 0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55 60 High EPAS1 mRNA Time (months) Time (months) N Events Median OS N Events Median OS Low EPAS1 mRNA 15 4 36 Low EPAS1 mRNA 18 239 lnt EPAS1 mRNA 39 2 36 lnt EPAS1 mRNA 36 5 34.5 High EPAS1 mRNA 17 1 33 High EPAS1 mRNA 17 0 37 B Histopathologically unchanged tissue Cancerous tissue Event Censored Event Censored

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4 Survival probability Survival probability

0.2 0.2

P = 0.42 P = 0.31 0.0 0.0 Low EPAS1 mRNA 0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55 lnt EPAS1 mRNA Time (months) Time (months) High EPAS1 mRNA

N Events Median OS N Events Median OS Low EPAS1 mRNA 9 3 24 Low EPAS1 mRNA 11 1 35 lnt EPAS1 mRNA 22 1 31 lnt EPAS1 mRNA 21 4 21 High EPAS1 mRNA 11 1 38 High EPAS1 mRNA 10 0 36.5

Event Censored Event Censored

CD1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4 Survival probability Survival probability

0.2 0.2

P = 0.098 DNA methylation P = 0.87 DNA methylation 0.0 of EPAS1 present 0.0 of EPAS1 present 0 5 10 15 20 25 30 35 40 45 50 55 DNA methylation 5 10152025 30 35 40 45 50 55 DNA methylation Time (months) of EPAS1 absent Time (months) of EPAS1 absent

N Events Median OS N Events Median OS DNA methylation of DNA methylation of EPAS1 present 47 336 EPAS1 present 25 3 33 DNA methylation of DNA methylation of 24 4 33 17 2 31 EPAS1 absent EPAS1 absent

1120 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

Table 3. Multivariate analysis of DNA methylation and transcript level of EPAS1 in patients with colorectal cancer

OS DFS

Variable HR (95% CI) P HR (95% CI) P EPAS1 mRNA High 1 1 Intermediate 0.65 (0.049–8.64) 0.21 0.39 (0.083–1.86) 0.09 Low 4.91 (0.42–56.16) 0.047 1.24 (0.28–5.59) 0.24 EPAS1 DNA methylation Absent 1 1 Present 33.94 (2.84–405.95) 0.0054 1.62 (0.32–8.36) 0.56 Gender Female 1 1 Male 27.89 (1.56–497.24) 0.022 1.98 (0.38–10.40) 0.42 Age (y) Below 60 1 1 Above 60 0.86 (0.079–9.23) 0.89 1.42 (0.29–6.82) 0.66 Therapy No 1 1 Yes 0.19 (0.027–1.33) 0.095 1.39 (0.44–4.37) 0.57

HIF1A mRNA levels were increased in cancerous compared EPAS1-specific target genes seems to be the interaction of with noncancerous tissue (8, 11, 12, 33, 34). However, other N-transactivation domain of HIF protein with different studies reported a constant HIF1A mRNA level in tumor coactivators (41, 42). In renal cell carcinoma, tumor pro- cells and suggested mainly posttranslational regulation of gression and metastasis were predominantly dependent on HIF1A expression (35–37), which is consistent with our EPAS1 (43). However, an in vitro study in colorectal cancer observations. We have not detected significant HIF1A cells revealed the induction number of genes associated with transcript changes between cancerous and histopathologi- glycolysis and angiogenesis by HIF1a and tumor-suppressor cally unchanged tissues isolated from 120 patients. Only one genes such as cyclin G2 or angiopoietin-like 4 by EPAS1 (10). publication described a correlation of elevated EPAS1 Xenograft studies support the hypothesis of a protective mRNA level in blood plasma with poor outcome for the function of EPAS1 in colorectal cancer. Silencing of EPAS1 patients with colorectal cancer (38). We found reduced expression in a mouse model is associated with a more EPAS1 mRNA level in primary cancerous compared intensive development of colorectal cancer (10). In a with histopathologically unchanged tissues. Moreover, we KRAS-driven non–small cell lung carcinoma mouse model, observed a higher risk of death of patients with colorectal the loss of EPAS1 expression resulted in increased tumor cancer with low EPAS1 mRNA level in histopathologically growth and progression (39). Moreover, siRNA knockdown unchanged tissues. In keeping with our results, reduced of EPAS1 reduced apoptosis in glioblastoma cells (44). Our EPAS1 transcript level was observed in a non–small cell observation illustrated an increased risk of death in patients lung carcinoma relative to normal tissue (39). with low EPAS1 mRNA level in histopathologically Data suggest that the importance of HIF1a and EPAS1 in unchanged tissues, which supports the idea of EPAS1 as response to hypoxia may differ among tumor types and having a tumor-protective role. Obviously, large multicenter different stages of tumor progression. Moreover, recent studies on various patient populations with extended follow- articles indicate that these two HIFa subunits exhibit up need to confirm these results. distinct roles in hypoxic conditions (31). HIF1a and EPAS1 For the first time, we examined a relationship between may regulate the expression of many of the same target epigenetic silencing of EPAS1 and clinical prognosis of genes, but each has unique responsive genes as well (40). The patients with colorectal cancer. We found that a reduced mechanism responsible for the activation of HIF1a-or EPAS1 mRNA level was associated with DNA

Figure 3. The Kaplan–Meier survival analysis among patients with colorectal cancer according to the EPAS1 mRNA level and DNA methylation of the EPAS1 promoter. A and B, patients were subdivided into three groups: low, intermediate, and high EPAS1 transcript levels in histopathologically unchanged and cancerous tissue. C and D, patients were subdivided into two groups: absent DNA methylation and present DNA methylation in the EPAS1 gene regulatory region in cancerous tissue compared with histopathologically unchanged. The analysis was performed with the division into groups: A and C: all patients included in the analysis (n ¼ 71); B and D: patients without postoperative chemotherapy (n ¼ 42). N, number of patients.

www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1121

Rawłuszko-Wieczorek et al.

Normalized melting curves Normalized melting curves

100,000 100,000

90,000 90,000

80,000 80,000

70,000 100% 70,000 100%

60,000 50% 60,000

50,000 50,000 50%

40,000 40,000 DLD-1 Hypoxia DLD-1 Hypoxia Relative signal (%) Relative signal (%) 30,000 DLD-1 Normoxia 30,000 DLD-1 Normoxia 10% 20,000 10% 20,000 10,000 HCT116 Hypoxia 1% 0% 10,000 HCT116 Hypoxia 1% 0% HCT116 Normoxia HCT116 Normoxia 0,000 0,000

74 74.5 75 75.5 76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 72.5 73 73.5 74 74.5 75 75.5 76 76.5 77 77.5 Temperature (°C) Temperature (°C)

Normalized melting curves Normalized melting curves

100,000 100,000 100% 90,000 90,000 100% 80,000 80,000 50%

70,000 70,000 10% 60,000 60,000

50,000 50,000 50% DLD-1 Hypoxia 1% 40,000 40,000 DLD-1 Normoxia DLD-1 Hypoxia Relative signal (%) 30,000 DLD-1 Normoxia Relative signal (%) 30,000

20,000 10% 20,000 HCT116 Hypoxia HCT116 Normoxia 10,000 HCT116 Hypoxia 1% 0% 10,000 0% HCT116 Normoxia 0,000 0,000

75 75.5 76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 82 82.5 83 83.5 74.5 75 75.5 76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 Temperature (°C) Temperature (°C)

Normalized melting curves

100,000 100% 90,000 HCT116 Hypoxia 80,000

70,000 HCT116 Normoxia 60,000 50% 50,000

40,000 DLD-1 Hypoxia

Relative signal (%) 0% 30,000 DLD-1 Normoxia 10%

20,000 1%

10,000

0,000

76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 Temperature (°C)

Figure 4. DNA methylation, expression level of the HIF1A and EPAS1 genes as well effect of 5-dAzaC on DNA methylation and expression of EPAS1 in HCT116 and DLD-1 colorectal cancer cells. HCT116 and DLD-1 cells were cultured under normoxic or hypoxic (1% O2) conditions for 48 hours. Cells were then used for DNA isolation followed by bisulfite modification, RNA and protein isolation. A, methylation percentage of DNA fragments within the HIF1A and EPAS1 CpG island (Supplementary Table S1A) in HCT116 and DLD-1 cells under hypoxic and normoxic conditions was determined by RT-PCR amplification of bisulfite- treated standard and cell line DNA, followed by comparison of their HRM profiles. (Continued on the following page.)

hypermethylation in cancerous tissues, and EPAS1 DNA ment of the EPAS1 CpG island is the region of a transcrip- methylation was a prognostic factor for patient survival in tion factor binding and different types of epigenetic mod- multivariate Cox regression analysis. An investigated frag- iﬁcations such as histone acetylation (17). Hence, it may be

1122 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

1.00 DLD-1 HCT116 0.75

HIF1α (92 kDa) 0.50 1 45.9 1.13 78.2

GAPDH (36 kDa) HIF1A mRNA level 0.25

of the respective controls of the respective NH NH 0.00 DLD-1 HCT116

1.75

1.50 DLD-1 HCT116 1.25

1.00 EPAS1 (97 kDa) 0.75 1 4.3 1.13 2.53 0.50

EPAS1 mRNA level EPAS1 GAPDH (36 kDa) 0.25 of the respective controls of the respective NH NH 0.00 DLD-1 HCT116 C EPAS1.3 chr2: 46 526 762 chr2: 46 526 905

Normalized melting curves

100% 100,000

90,000 HCT 116 Hypoxia control

80,000 HCT 116 Normoxia control

70,000 HCT 116 Normoxia 5-dAzaC 60,000 HCT 116 Hypoxia 5-dAzaC 50,000 50% 40,000

Relative singnal (%) Relative 30,000 DLD-1 Hypoxia 5-dAzaC DLD-1 Normoxia 5-dAzaC 20,000 1% 10%

10,000 DLD-1 Hypoxia control DLD-1 Normoxia control 0% 0,000

76 76.5 77 77.5 78 78.5 79 79.5 80 80.5 81 81.5 Temperature (ºC)

Figure 4. (Continued.) B, cells were cultured in DMEM either in hypoxic (1%O2; H) or normoxic (N) conditions for 48 hours. After incubation, the cells were used for total RNA isolation followed by reverse transcription and protein isolation. The HIF1A and EPAS1 cDNA levels were determined by RQ-PCR relative quantification analysis. RQ-PCR results were standardized by the geometric mean of PBGD and hMRPL19 cDNA levels. HIF1A and EPAS1 cDNA levels are expressed as a multiplicity of these cDNA copies in the cell line's calibrator. Proteins were separated by 10% SDS-PAGE, and transferred to a membrane that was then immunoblotted with specific primary and secondary Ab. The band densitometry readings were normalized to GAPDH loading control. The ratio of HIF1a or EPAS1 to GAPDH for DLD-1 in normoxic conditions was assumed to be 1. C, HCT116 and DLD-1 cells were cultured under normoxic or hypoxic (1% O2) conditions either in the absence or in the presence of 5-dAzaC at a concentration of 5.00 mmol/L for 48 hours. Cells were then used for DNA isolation followed by bisulfite modification. Methylation percentage of DNA fragment within the EPAS1 CpG island (chr2: 46 526 762-46 526 905) in HCT116 and DLD-1 cells under hypoxic and normoxic conditions was determined by RT-PCR amplification of bisulfite-treated standard and cell line DNA, followed by comparison of their HRM profiles. (Continued on the following page.) www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1123

Rawłuszko-Wieczorek et al.

D HCT116 DLD-1 3.0

3 2.5

2.0 2 1.5

1.0 1 EPAS1 mRNA levels EPAS1 mRNA levels EPAS1 0.5 of the respective controls of the respective controls of the respective

0 0.0 6 24 48 6 24 48 Time (h) Time (h) Normoxia Hypoxia

HCT116

5-dAzaC concentration (μmol/L) 0.0 1.0 5.0 0.0 1.0 5.0 0.0 1.0 5.0 EPAS1 (97 kDa) Hypoxia 1.00 0.85 1.32 1.00 1.21 1.98 1.00 1.11 2.31 GAPDH (36 kDa)

EPAS1 (97 kDa) 1.00 1.09 1.07 1.00 1.21 1.39 1.00 1.23 1.33 Normoxia GAPDH (36 kDa)

6 h 24 h 48 h

DLD-1

5-dAzaC concentration (μmol/L) 0.0 1.0 5.0 0.0 1.0 5.0 0.0 1.0 5.0 EPAS1 (97 kDa)

1.00 1.21 1.01 1.00 0.77 0.85 1.00 0.99 0.91 Hypoxia GAPDH (36 kDa)

EPAS1 (97 kDa)

1.00 0.96 0.87 1.00 0.92 0.87 1.00 0.94 1.04 Normoxia GAPDH (36 kDa)

6 h 24 h 48 h

Figure 4. (Continued.) D, HCT116 and DLD-1 cells were cultured in DMEM for 6, 24, and 48 hours either in the absence or in the presence of 5-dAzaC at a concentration of 1.00 or 5.00 mmol/L under hypoxic or normoxic conditions. After incubation, the cells were used for total RNA isolation and protein isolation. Total RNA was reverse-transcribed, and EPAS1 cDNA levels were determined by RQ-PCR relative quantiﬁcation analysis. EPAS1 cDNA levels are expressed as a multiplicity of the respective controls. Each sample was determined in triplicate, and results are presented as the mean SE from three experiments , P < 0.01; , P < 0.001. The cell protein was separated by 10% SDS-PAGE, and transferred to a membrane that was then immunoblotted with speciﬁc primary and secondary Ab. The band densitometry readings were normalized to GAPDH loading control. The ratio of EPAS1 to GAPDH for control was assumed to be 1.

1124 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

recognized as a putative enhancer or a promoter region. Data DNA demethylation of the EPAS1 CpG island in HCT116 about DNA methylation of the HIF1A promoter region are cells, regardless of oxygen concentration. An increase in the ambiguous. The absence of DNA methylation of the HIF1A EPAS1 transcript and protein levels under hypoxic conditions promoter was observed in advanced uterine cervical carci- suggests the role of EPAS1 DNA methylation in HCT116 noma and datasets available at ENCODE project, whereas cells. In normoxic conditions, the increase of EPAS1 protein DNA hypermethylation was observed in an immature was not observed in both cell lines, probably due to oxygen- hematopoietic cell line HMC-1 and normal colon tissues dependent degradation. However, an increase of EPAS1 isolated from 20 patients with colorectal cancer (14, 34, 45). mRNA in DLD-1 cells was observed under normoxic con- In our studies, DNA methylation of the HIF1A promoter ditions after 5-dAzaC treatment, despite the lack of DNA was undetected in a group of 120 patients. The difference to methylation. The absence of mRNA and protein upregula- previous studies in colorectal cancer may result from the used tion in DLD-1 cells upon hypoxia suggests that other factors methods for determining DNA methylation, and may sug- must be involved in the induction of EPAS-1 gene expression gest complexity in epigenetic regulation of HIF1A. More- besides DNA methylation. Moreover, discrepancies of the over, many environmental factors like tobacco smoking, results obtained from the two cell lines may be a result of their diet, and physical activity may affect DNA methylation different genetic background. status in colorectal cancer (46). Further experiments need In conclusion, our findings present epigenetic transcrip- to verify the potential impact of these factors on EPAS1 tional downregulation of EPAS1 in patients with colorectal DNA methylation. In addition, even though we have not cancer and the HCT116 cell line. In addition, low EPAS1 detected subgroup-biased DNA methylation in patients, we mRNA level in histopathologically unchanged tissues and cannot exclude potential impact of gender, age, and other DNA hypermethylation in cancerous tissues compared with clinicopathologic features in different patient populations. the histopathologically unchanged might be an independent The main limitation of our studies is the lack of association prognostic factor and potentially useful for selecting patients of DNA methylation status and mRNA level with protein with a higher risk of death after resection. The clinical value expression of analyzed genes. However, because the amount of changes in EPAS1 mRNA and DNA methylation levels of samples was limited, we preferred to evaluate HIF1A, needs to be confirmed by large longitudinal studies as well as EPAS1 DNA methylation, and transcript level as the prog- verified in other cancer types. nostic significance of transcript, and DNA methylation has not been investigated previously. We assessed DNA methyl- Disclosure of Potential Conflicts of Interest ation and expression level of the HIF1A and EPAS1 genes No potential conflicts of interest were disclosed. under hypoxic and normoxic conditions in DLD-1 and HCT116 colorectal cancer cell lines. We undetected changes Authors' Contributions in DNA methylation of the HIF1A promoter region and the Conception and design: A.A. Rawłuszko-Wieczorek HIF1A mRNA level under normoxic and hypoxic conditions Development of methodology: A.A. Rawłuszko-Wieczorek a Acquisition of data (provided animals, acquired and managed patients, provided in both analyzed cell lines. However, an increase of HIF1 facilities, etc.): A.A. Rawłuszko-Wieczorek, K. Horbacka, P. Krokowicz protein in hypoxia was observed when compared with nor- Analysis and interpretation of data (e.g., statistical analysis, biostatistics, compu- tational analysis): A.A. Rawłuszko-Wieczorek moxic conditions, which indicates that the main regulator of Writing, review, and/or revision of the manuscript: A.A. Rawłuszko-Wieczorek, HIF1A expression in colorectal cancer is oxygen-dependent M. Misztal, P.P. Jagodzinski posttranslational modification. The EPAS1 transcript level Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Horbacka remained stable in normoxic and hypoxic conditions in Study supervision: P.P. Jagodzinski HCT116 cells, whereas EPAS1 protein level was higher in fi hypoxic conditions. On the other hand, a signi cant increase Acknowledgments in both the amount of EPAS1 transcript and protein in The authors thank the Institute of Molecular Biology and Biotechnology, Adam hypoxic conditions was observed in DLD-1 cells. EPAS1 Mickiewicz University, for access to the MCO-18M multi-gas cell culture incubator, mRNA changes in DLD-1 were associated with DNA hypo- Sanyo. methylation of EPAS1 CpG island. The same region (chr2: 46 526 762-46 526 905) was hypermethylated in HCT116 cells, Grant Support This work is supported by grant 2012/05/N/NZ5/00844 from the National which may explain the lack of mRNA induction under Science Center, Poland. hypoxic conditions. HCT116 and DLD-1 cells were also The costs of publication of this article were defrayed in part by the payment of page incubated for different time periods with 5-dAzaC under charges. This article must therefore be hereby marked advertisement in accordance with hypoxic and normoxic conditions. 5-dAzaC may induce 18 U.S.C. Section 1734 solely to indicate this fact. expression of many genes and inhibit growth of colorectal Received January 29, 2014; revised April 7, 2014; accepted April 25, 2014; cancer cell lines (47, 48). We observed 5-dAzaC-induced published OnlineFirst May 13, 2014.

References 1. Webb JD, Coleman ML, Pugh CW. Hypoxia, hypoxia-inducible factors 2. Jokilehto T, Jaakkola PM. The role of HIF prolyl hydroxy- (HIF), HIF hydroxylases and oxygen sensing. Cell Mol Life Sci lases in tumour growth. J Cell Mol Med 2010;14:758– 2009;66:3539–54. 70.

www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1125

Rawłuszko-Wieczorek et al.

3. Rasheed S, Harris AL, Tekkis PP, Turley H, Silver A, McDonald PJ, et al. for the semi-quantitative determination of CST6 promoter methylation Hypoxia-inducible factor-1alpha and -2alpha are expressed in most in clinical samples. BMC Cancer 2012;12:486. rectal cancers but only hypoxia-inducible factor-1alpha is associated 25. Oster B, Thorsen K, Lamy P, Wojdacz TK, Hansen LL, Birkenkamp- with prognosis. Br J Cancer 2009;100:1666–73. Demtroder K, et al. Identification and validation of highly frequent CpG 4. Schmitz KJ, Muller CI, Reis H, Alakus H, Winde G, Baba HA, et al. island hypermethylation in colorectal adenomas and carcinomas. Int J Combined analysis of hypoxia-inducible factor 1 alpha and metal- Cancer 2011;129:2855–66. lothionein indicates an aggressive subtype of colorectal carcinoma. Int 26. Chomczynski P, Sacchi N. Single-step method of RNA isolation by J Colorectal Dis 2009;24:1287–96. acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Bio- 5. Rajaganeshan R, Prasad R, Guillou PJ, Scott N, Poston G, Jayne DG. chem 1987;162:156–9. Expression patterns of hypoxic markers at the invasive margin of 27. Rawluszko AA, Bujnicka KE, Horbacka K, Krokowicz P, Jagodzi Ski colorectal cancers and liver metastases. Eur J Surg Oncol PP. Expression and DNA methylation levels of prolyl hydroxylases 2009;35:1286–94. PHD1, PHD2, PHD3 and asparaginyl hydroxylase FIH in colorectal 6. Yoshimura H, Dhar DK, Kohno H, Kubota H, Fujii T, Ueda S, et al. cancer. BMC Cancer 2013;13:526. Prognostic impact of hypoxia-inducible factors 1alpha and 2alpha in 28. Birner P, Schindl M, Obermair A, Plank C, Breitenecker G, Oberhuber colorectal cancer patients: correlation with tumor angiogenesis and G. Overexpression of hypoxia-inducible factor 1alpha is a marker for an cyclooxygenase-2 expression. Clin Cancer Res 2004;10:8554–60. unfavorable prognosis in early-stage invasive cervical cancer. Cancer 7. Kuwai T, Kitadai Y, Tanaka S, Onogawa S, Matsutani N, Kaio E, et al. Res 2000;60:4693–6. Expression of hypoxia-inducible factor-1alpha is associated with 29. Birner P, Gatterbauer B, Oberhuber G, Schindl M, Rossler K, Prodinger tumor vascularization in human colorectal carcinoma. Int J Cancer A, et al. Expression of hypoxia-inducible factor-1 alpha in oligoden- 2003;105:176–81. drogliomas: its impact on prognosis and on neoangiogenesis. Cancer 8. Furlan D, Sahnane N, Carnevali I, Cerutti R, Bertoni F, Kwee I, et al. Up- 2001;92:165–71. regulation of the hypoxia-inducible factor-1 transcriptional pathway in 30. Birner P, Schindl M, Obermair A, Breitenecker G, Oberhuber G. colorectal carcinomas. Hum Pathol 2008;39:1483–94. Expression of hypoxia-inducible factor 1alpha in epithelial ovarian 9. Baba Y, Nosho K, Shima K, Irahara N, Chan AT, Meyerhardt JA, et al. tumors: its impact on prognosis and on response to chemotherapy. HIF1A overexpression is associated with poor prognosis in a cohort of Clin Cancer Res 2001;7:1661–8. 731 colorectal cancers. Am J Pathol 2010;176:2292–301. 31. Keith B, Johnson RS, Simon MC. HIF1alpha and HIF2alpha: sibling 10. Imamura T, Kikuchi H, Herraiz MT, Park DY, Mizukami Y, Mino- rivalry in hypoxic tumour growth and progression. Nat Rev 2012;12: Kenduson M, et al. HIF-1alpha and HIF-2alpha have divergent roles 9–22. in colon cancer. Int J Cancer 2009;124:763–71. 32. Semenza GL. Defining the role of hypoxia-inducible factor 1 in cancer 11. Hoffmann AC, Mori R, Vallbohmer D, Brabender J, Klein E, Drebber U, biology and therapeutics. Oncogene 2010;29:625–34. et al. High expression of HIF1a is a predictor of clinical outcome in 33. Matsuyama T, Nakanishi K, Hayashi T, Yoshizumi Y, Aiko S, Sugiura Y, patients with pancreatic ductal adenocarcinomas and correlated to et al. Expression of hypoxia-inducible factor-1alpha in esophageal PDGFA, VEGF, and bFGF. Neoplasia 2008;10:674–9. squamous cell carcinoma. Cancer Sci 2005;96:176–82. 12. Ma J, Zhang L, Ru GQ, Zhao ZS, Xu WJ. Upregulation of hypoxia 34. Luczak MW, Roszak A, Pawlik P, Kedzia H, Lianeri M, Jagodzinski PP. inducible factor 1alpha mRNA is associated with elevated vascular Increased expression of HIF-1A and its implication in the hypoxia endothelial growth factor expression and excessive angiogenesis and pathway in primary advanced uterine cervical carcinoma. Oncol Rep predicts a poor prognosis in gastric carcinoma. World J Gastroenterol 2011;26:1259–64. 2007;13:1680–6. 35. Minet E, Ernest I, Michel G, Roland I, Remacle J, Raes M, et al. 13. Kim MS, Lee J, Sidransky D. DNA methylation markers in colorectal HIF1A gene transcription is dependent on a core promoter cancer. Cancer Metastasis Rev 2010;29:181–206. sequence encompassing activating and inhibiting sequences locat- 14. Koslowski M, Luxemburger U, Tureci O, Sahin U. Tumor-associated ed upstream from the transcription initiation site and cis elements CpG demethylation augments hypoxia-induced effects by positive located within the 50UTR. Biochem Biophys Res Commun autoregulation of HIF-1alpha. Oncogene 2011;30:876–82. 1999;261:534–40. 15. CpGPlot/CpGReport/Isochore. Available from: http://www.ebi.ac.uk/ 36. Dery MA, Michaud MD, Richard DE. Hypoxia-inducible factor 1: Tools/seqstats/emboss_cpgplot/ regulation by hypoxic and non-hypoxic activators. Int J Biochem Cell 16. CpG Island Searcher. Available from: http://cpgislands.usc.edu/ Biol 2005;37:535–40. 17. UCSC Genome Bioinformatics Site. Available from: http://genome. 37. Galban S, Kuwano Y, Pullmann R Jr, Martindale JL, Kim HH, Lal A, et al. ucsc.edu/ RNA-binding proteins HuR and PTB promote the translation of hyp- 18. Bock C, Reither S, Mikeska T, Paulsen M, Walter J, Lengauer T. BiQ oxia-inducible factor 1alpha. Mol Cell Biol 2008;28:93–107. Analyzer: visualization and quality control for DNA methylation data 38. Mohammed N, Rodriguez M, Garcia V, Garcia JM, Dominguez G, Pena from bisulfite sequencing. Bioinformatics (Oxford, England) C, et al. EPAS1 mRNA in plasma from colorectal cancer patients is 2005;21:4067–8. associated with poor outcome in advanced stages. Oncol Lett 19. Rohde C, Zhang Y, Jurkowski TP, Stamerjohanns H, Reinhardt R, 2011;2:719–24. Jeltsch A. Bisulfite sequencing Data Presentation and Compilation 39. Mazumdar J, Hickey MM, Pant DK, Durham AC, Sweet-Cordero A, (BDPC) web server–a useful tool for DNA methylation analysis. Nucleic Vachani A, et al. HIF-2alpha deletion promotes Kras-driven lung tumor Acids Res 2008;36:e34. development. Proc Natl Acad Sci U S A 2010;107:14182–7. 20. Wojdacz TK, Dobrovic A. Methylation-sensitive high resolution melting 40. Hu CJ, Wang LY, Chodosh LA, Keith B, Simon MC. Differential roles of (MS-HRM): a new approach for sensitive and high-throughput assess- hypoxia-inducible factor 1alpha (HIF-1alpha) and HIF-2alpha in hyp- ment of methylation. Nucleic Acids Res 2007;35:e41. oxic gene regulation. Mol Cell Biol 2003;23:9361–74. 21. Rawluszko AA, Horbacka K, Krokowicz P, Jagodzinski PP. Decreased 41. Hu CJ, Sataur A, Wang L, Chen H, Simon MC. The N-terminal expression of 17-beta-hydroxysteroid dehydrogenase type 1 is asso- transactivation domain confers target gene specificity of hypoxia- ciated with DNA hypermethylation in colorectal cancer located in the inducible factors HIF-1alpha and HIF-2alpha. Mol Biol Cell 2007;18: proximal colon. BMC Cancer 2011;11:522. 4528–42. 22. Hsieh CL. Dependence of transcriptional repression on CpG methyl- 42. Lau KW, Tian YM, Raval RR, Ratcliffe PJ, Pugh CW. Target gene ation density. Mol Cell Biol 1994;14:5487–94. selectivity of hypoxia-inducible factor-alpha in renal cancer cells is 23. Shao Y, Zhang W, Zhang C, Wu Q, Yang H, Zhang J, et al. High- conveyed by post-DNA-binding mechanisms. Br J Cancer 2007;96: resolution melting analysis of BLU methylation levels in gastric, colo- 1284–92. rectal, and pancreatic cancers. Cancer Invest 2010;28:642–8. 43. Carroll VA, Ashcroft M. Role of hypoxia-inducible factor (HIF)-1alpha 24. Dimitrakopoulos L, Vorkas PA, Georgoulias V, Lianidou ES. A closed- versus HIF-2alpha in the regulation of HIF target genes in response to tube methylation-sensitive high resolution melting assay (MS-HRMA) hypoxia, insulin-like growth factor-I, or loss of von Hippel-Lindau

1126 Mol Cancer Res; 12(8) August 2014 Molecular Cancer Research

DNA Methylation and mRNA Level of HIF1A and EPAS1 in Colorectal Cancer

function: implications for targeting the HIF pathway. Cancer Res 46. Migliore L, Migheli F, Spisni R, Coppede F. Genetics, cytogenetics, and 2006;66:6264–70. epigenetics of colorectal cancer. J Biomed Biotechnol 2011;2011: 44. Acker T, Diez-Juan A, Aragones J, Tjwa M, Brusselmans K, Moons L, 792362. et al. Genetic evidence for a tumor suppressor role of HIF-2alpha. 47. Momparler RL. Pharmacology of 5-Aza-20-deoxycytidine (decitabine). Cancer Cell 2005;8:131–41. Semin Hematol 2005;42(3 Suppl 2):S9–16. 45. Walczak-Drzewiecka A, Ratajewski M, Pulaski L, Dastych J. DNA 48. Ghoshal K, Motiwala T, Claus R, Yan P, Kutay H, Datta J, et al. methylation-dependent suppression of HIF1A in an immature hemato- HOXB13, a target of DNMT3B, is methylated at an upstream CpG poietic cell line HMC-1. Biochem Biophys Res Commun 2010;391: island, and functions as a tumor suppressor in primary colorectal 1028–32. tumors. PloS ONE 2010;5:e10338.

www.aacrjournals.org Mol Cancer Res; 12(8) August 2014 1127

Prognostic Potential of DNA Methylation and Transcript Levels of HIF1A and EPAS1 in Colorectal Cancer

Agnieszka Anna Rawluszko-Wieczorek, Karolina Horbacka, Piotr Krokowicz, et al.

Mol Cancer Res 2014;12:1112-1127. Published OnlineFirst May 13, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-14-0054

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2014/05/13/1541-7786.MCR-14-0054.DC1

Cited articles This article cites 45 articles, 9 of which you can access for free at: http://mcr.aacrjournals.org/content/12/8/1112.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/12/8/1112.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/12/8/1112. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.