Increased Oxidative Stress and RUNX3 Hypermethylation in Patients with Hepatitis B Virus-Associated Hepatocellular Carcinoma (HCC) and Induction of RUNX3 Hypermethylation by Reactive Oxygen Species in HCC Cells

Promoter hypermethylation of the runt-related transcription factor 3 ( RUNX3 ) gene is associated with increased risk of hepatocellular carcinoma (HCC). Oxidative stress plays a vital role in both carcinogenesis and progression of HCC. However, whether oxidative stress and RUNX3 hypermethylation in HCC have a cause-and-effect relationship is not known. In this study, plasma protein carbonyl and total antioxidant capacity (TAC) in patients with hepatitis B virus (HBV)-associated HCC (n = 60) and age-matched healthy subjects (n = 80) was determined. RUNX3 methylation in peripheral blood mononuclear cells (PBMC) of subjects was measured by methylation-specific PCR. Effect of reactive oxygen species (ROS) on induction of RUNX3 hypermethylation in HCC cells was investigated. Plasma protein carbonyl content was significantly higher, whereas plasma TAC was significantly lower, in HCC patients than healthy controls. Based on logistic regression, increased plasma protein carbonyl and decreased plasma TAC were independently associated with increased risk for HCC. PBMC RUNX3 methylation in the patient group was significantly greater than in the healthy group. RUNX3 methylation in hydrogen peroxide (H 2 O 2 )-treated HepG2 cells was significantly higher than in untreated control cells. In conclusion, increase in oxidative stress in Thai patients with HBV-associated HCC was demonstrated. This oxidative increment was independently associated with an increased risk for HCC development. RUNX3 in PBMC was found to be hypermethylated in the HCC patients. In vitro , RUNX3 hypermethylation was experimentally induced by H 2 O 2 . Our findings suggest that oxidative stress is a cause of RUNX3 promoter hypermethylation in HCC cells.


Introduction
Liver cancer is a leading cancer in Thailand comprising of two main forms. One is cholangiocarcinoma, and the other is hepatocellular carcinoma (HCC). HCC in Thailand is principally associated with chronic hepatitis B virus (HBV) infection (Tangkijvanich et al., 1999), however molecular mechanism of HBV-associated HCC development in Thai patients is not fully understood. Beside chronic inflammatory activation, oxidative stress is considerably increased in HBV-associated HCC patients (Tsai et al., 2009, Nair et al., 2010, Zhao et al., 2011, and has been believed to play an important role in the development of viral-induced HCC (Marra et al., 2011, Higgs et al., 2014. In addition, oxidative stress is found to increase during the replication of HBV in cell culture model (Severi et al., 2006).
Oxidative stress, a condition with overwhelming generation of reactive oxygen species (ROS) and/or inadequacy of antioxidants, exerts a tumorigenic role to promote both genetic mutation and epigenetic alteration (Franco et al., 2008, Ziech et al., 2011). An epigenetic hallmark that is found in all carcinomas including HCC is alteration of DNA methylation (Herceg and Paliwal, 2011). There are two types of DNA methylation alterations in cancers viz. genome-wide or global hypomethylation and promoter hypermethylation of tumor suppressor genes (TSG). Silencing of TSG via DNA methylation is well recognized in the carcinogenesis of HCC (Sceusi et al., 2011). Runt-related transcription factor 3 (RUNX3) is one of TSG that vitally involves in the HCC carcinogenesis and progression. Promoter hypermetylation of RUNX3 is associated with increased risk for HCC (Yang et al., 2014, Zhang et al., 2015. RUNX3 expression is decreased in HCC tissues as well as HCC cell lines (Li and Jiang, 2011). Loss of RUNX3 expression is associated with the progression of tumor, as it is able to induce epithelialmesenchymal transition (EMT) in the low-EMT HCC cells (Tanaka et al., 2012). Up to now, cause and regulation of RUNX3 hypermethylation in HCC is not known.
Oxidative stress-induced aberrations of DNA methylation in HCC has been hypothesized and demonstrated (Nishida and Kudo, 2013). ROS-induced hypermethylation of E-cadherin promoter is demonstrated in HCC cell lines (Lim et al., 2008). We previously demonstrated an induction of RUNX3 hypermethylation by ROS in bladder cancer cells (Wongpaiboonwattana et al., 2013). Hitherto, causative relationship between ROS and RUNX3 hypermethylation in HCC, especially HBVassociated HCC, has not been investigated.
In the present study, oxidative stress and RUNX3 hypermethylation in HBV-associated HCC patients were investigated. Experimentally, whether ROS was able to induce RUNX3 hypermethylation in HCC cells was investigated.

Patients and specimen collection
A total of 140 subjects divided into HCC (n=60) and healthy (n=80) groups were recruited for the study. Means age between these two groups (52.33±7.91 vs. 50.59±5.54 years old) were not significantly different (Table 1). There were 52 (86.67%) men and 8 (13.33%) women in the HCC groups. The healthy control group was consisted of 50 (62.5%) men and 30 (37.5%) women (P=0.001 vs. HCC group). All HCC patients were serologically proof (including DNA test) to have chronic HBV infection, considered as HBV-associated HCC. HBsAg was positive, but anti-hepatitis C virus was negative in all cases. Of 60 patients, 52 had data of cirrhosis and staging. Most of patients (78.85%) had cirrhotic liver. According to Barcelona-Clinic Liver Cancer (BCLC) staging system, patients in stage 0, A, B and C were accounted for 2 (3.85%), 7 (13.46%), 19 (36.54%) and 24 (46.15%), respectively (Table 1).
Blood samples were preoperatively collected, representing as pre-treatment samples. Plasma was separated, and DNA was isolated from peripheral blood mononuclear cells (PBMC). Healthy subjects were blood donors at Thai Red Cross, Bangkok, Thailand. Blood samples leftover from the routine blood test were used for plasma and DNA isolations. Informed consents were received from all participants prior to collection of specimen. Research protocol was approved by the Ethics Committee, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand.

Protein carbonyl determination
Protein carbonyl content, an indicator of oxidative protein damage (protein oxidation), was measured in plasma samples and cell lysate. The procedure for protein carbonyl measurement in plasma was fully described in our previous report (Patchsung et al., 2012). For cell lysate, cells were lyzed with RIPA buffer, and the lysate was centrifuged at 4,500 xg for 10 min to collect supernatant. Protein concentration was determined by Bradford assay (Bio-Rad Laboratories, CA, USA). Protein carbonyl in cell lysate was measured in a similar way to the procedure performed in plasma samples. Experiments were performed at least in triplicate.

DNA extraction and bisulfite treatment
PBMC was isolated from blood samples. DNA was extracted using genomic DNA extraction kit (RBC Bioscience, New Taipei City, Taiwan) according to the manufacturer's procedure. DNA concentration was measured using spectrophotometer (NanoDrop 2000c, Wilmington, DE, USA). Bisulfite conversion of DNA (250 ng) was performed using EZ DNA Methylation-Gold™ kit (Zymo Research, Irvine, CA, USA). The bisulfite-treated DNA was collected and kept at -20 o C for further analysis.

Statistical analysis
Data are presented as mean±standard deviation (SD) or median (interquatile range, IQR) as appropriate. Twosample t-test or Mann-Whitney test was used to test the difference between two independent groups. One-way ANOVA followed by Tukey multiple comparison test or Kruskal-Wallis followed by Dunns test was used for testing the differences among three or more groups. Logistic regression was performed to obtain odds ratio (OR) adjusted for age and sex. Stata version 10 (College Station, TX) and GraphPad Prism 5 softwares (GraphPad, La Jolla, CA) was employed for graphs and statistical analyses. P value <0.05 was considered statistically significant.

Increased oxidative stress in HBV-associated HCC patients
Protein carbonyl and TAC were used as biomarkers for oxidative stress. Plasma protein carbonyl content in HBV-associated patients was significantly elevated, as compared to the age-matched healthy controls (Figure 1). In contrast, plasma TAC in HCC patients was significantly decreased, as compared to the healthy controls.
Because control group had more females than HCC group, we performed logistic regression to control confounding factors and quantify the strength of association of oxidative stress biomarkers with HCC. The β-coefficient of plasma protein carbonyl and TAC controlled for age and sex were 1.64 (95%CI: 0.55 -2.73, P=0.003) and -0.02 (95%CI:-0.02 --0.01, P<0.001), and their adjusted OR were 5.15 (95%CI: 1.73 -15.39) and 0.98 (95%CI: 0.98 -0.99), respectively (Table 2). These mean that every one-unit (nmol/mg protein) increase in plasma protein carbonyl the risk for HCC (odds of being HCC) is increased 5.15 times. For plasma TAC, every one-unit (VCEAC, mg/L) increase in its level the risk for HCC is reduced about 2% ((1 -0.98) x 100). These findings indicated that plasma protein carbonyl and TAC were independent predictors for the development of HBVassociated HCC.

Figure 1. Plasma Protein Carbonyl Content and TAC in Healthy (n=80) and HBV-Associated HCC Subjects
(n=60). Plasma levels of protein carbonyl and TAC (expressed as VCEAC) in HBV-associated patients were significantly higher than healthy controls. Data presented as median (IQR)

Figure 2. RUNX3 Methylation Levels in PBMC Compared between Healthy (n=20) and HCC (n=18).
Above panel: Representative gel of RUNX3 methylation detection by MSP in two healthy subjects (D033 and D092) and two HCC patients (HC045 and HC123). As indicated by M/U intensity ratio, RUNX3 methylation in PBMC of HCC patients was significantly higher than healthy individuals. Bars indicate medians and IQR. M: methylation (115 bp), U: unmethylation (124 bp) determination of RUNX3 methylation. Based on M/U intensity ratio presented herein, patients with HCC (n=18) had significantly increased RUNX3 methylation, as compared to the healthy individuals (n=20) ( Figure  2). Mean age (52.39 vs. 52.55 years) and sex distribution (85.00% vs. 94.44% females) between the HCC and control groups were not significantly different. Representative gel for RUNX3 methylation detection by MSP is shown in Figure 2.

Induction of RUNX3 hypermethylation by H 2 O 2 in HepG2 cell line
Causal relationship between oxidative stress and RUNX3 hypermethylation was investigated in cell culture model. H 2 O 2 was used as representative of ROS to stimulate oxidative stress in HepG2 cells (Figure 3). MTT assay showed that treatment with 10 µM H 2 O 2 for 72 h significantly increased cell survival, but treatments with 30 µM and more concentrations of H 2 O 2 gradually  Figure 3A). HepG2 cells treated with 20 µM H 2 O 2 for 72 h did not alter cell viability. Therefore, we opted to use the nonlethal concentration of H 2 O 2 at 20 µM for investigating the effect of ROS on RUNX3 methylation in HCC cells. Cells treated with 20 µM H 2 O 2 caused significantly increased in cellular protein carbonyl content, as compared to the untreated control cells ( Figure 3B). Co-treatment with NAC caused a significant decrease in protein carbonyl content, as compared to the H 2 O 2 -treated cells. Although significant differences were not revealed yet, cellular TAC trended to be decreased in the H 2 O 2 -treated cells and trended to be restored in the cells co-treated with H 2 O 2 and NAC ( Figure 3C). These indicated an increase in oxidative stress in HepG2 cells challenged with 20 µM H 2 O 2 .
RUNX3 M/U ratio in H 2 O 2 -treated HepG2 cells was significantly higher than the untreated controls ( Figure 4). The M/U ratio trended to be decreased in cells co-treated with H 2 O 2 and NAC, although significant difference was not observed yet. Representative gel image compared among these three cultured conditions is shown in Figure  4.

Discussion
Chronic hepatitis infection either with HBV or HCV is a primary etiologic cause of HCC. Mechanistic insight into hepatocarcinogenesis reveals that oxidative stress promotes both genetic mutation and epigenetic alteration (Nishida and Kudo, 2013). Several lines of evidences demonstrate an increase in oxidative stress in HBVassociated HCC (Tsai et al., 2009, Nair et al., 2010, Zhao et al., 2011. RUNX3 hypermethylation is also frequently detected in the HCC tissues (Yang et al., 2014, Zhang et al., 2015, suggesting a vital role in the HCC genesis. To date, the mechanism of how RUNX3 is hypermethylated in the HCC is not known. In this study, we demonstrated an increased oxidative stress and hypermethylation of RUNX3 promoter in patients with HBV-associated HCC. We additionally showed that increased extent of oxidative stress was independently associated with an increased risk for HCC development. Importantly, we experimentally demonstrated that H2O2 was capable of inducing RUNX3 hypermethylation in HepG2 cells, indicated that ROS was an inducer of RUNX3 hypermethylation. Oxidative stress-induced DNA methylation alteration in cancers gains more and more recognition (Nishida andKudo, 2013, Wu andNi, 2015). ROS-induced promoter hypermethylation of TSG in HCC is well demonstrated for E-cadherin (Lim et al., 2008). The authors conclude that Snail expression induced by H2O2 leads to recruiting histone deacetylase 1 (HDAC1) and DNA methyltransferase 1 (DNMT1), which subsequently causes hypermethylation of E-cadherin promoter. Similar mechanism that H 2 O 2 induces HDAC1 and DNMT1 expressions leading to hypermethylation of tumor suppressor caudal type homeobox-1 is also demonstrated in the colorectal cancer cells (Zhang et al., 2013). Additionally, increased expression and activity of HDAC1 and DNMT1 at the RUNX3 promoter resulting in RUNX3 hypermethylation is shown in the colon cancer cell lines challenged with H 2 O 2 (Kang et al., 2012). Whether the ROS-induced RUNX3 hypermethylation in HCC cells is mediated via this mechanism remains to be elucidated. The other mechanism for ROS-induced DNA hypermethylation is that H 2 O 2 induces the formation of a large silencing complex comprising of DNMT1, histone deacetylase SIRT1 and polycomb repressive complex 4 (O'Hagan et al., 2011). Such a large silencing complex relocalizes from non-GC-rich to GC-rich regions including promoter CpG islands, which in turn causes hypermethylation of the CpG-rich promoters. In this study, we clearly showed in HCC cells that H 2 O 2 is an inducer of RUNX3 hypermethylation, however the mechanism is unknown. Further study is awaiting to conduct to uncover the molecular mechanism of RUNX3 hypermethylation induced by ROS in HCC. Since RUNX3 hypermethylation is associated with HCC progression, and loss of RUNX3 is shown to induce EMT in the low-metastatic HCC cells (Tanaka et al., 2012), treatment with antioxidants might reestablish the unmethylated state of RUNX3, which in turn leads to re-expression of RUNX3 and deceleration of the tumor progression.
Limitations of the current study should be mentioned. We did not have data of oxidative stress and RUNX3 hypermethylation in HCC tissues to evaluate if they corresponded well with the measurements in blood samples. Samples size for detecting PBMC RUNX3 methylation was rather small. The transcript expression of RUNX3 did not measured. One cell line was investigated to demonstrate an induction of RUNX3 hypermethylation by ROS. The dose-dependent fashion of ROS-induced hypermethylation did not explored.
In conclusion, to the authors' knowledge this is the first study demonstrating an increase in oxidative stress coincided with hypermethylation of RUNX3 in patients with HBV-associated HCC. Increased degree of oxidative stress is an independent predictor for HCC development. It is also shown for the first time that ROS is able to induce the RUNX3 hypermethylation in HCC cells indicating a cause-and-effect relationship between oxidative stress and hypermethylation of RUNX3 promoter. Antioxidant regimen might be beneficial to restore the unmethylated state of RUNX3, in order to re-express this tumor suppressor protein in the HBV-related liver cancer.