SOCS3 Methylation Predicts a Poor Prognosis in HBV Infection-Related Hepatocellular Carcinoma

Suppressor of cytokine signaling 3 (SOCS3) plays crucial roles in JAK/STAT signaling pathway inhibition in hepatocellular carcinoma (HCC). However, the methylation status of SOCS3 in HBV infection-related HCC and the relationship between SOCS3 methylation and the clinical outcome remain unknown. Here, we reported that in HCC tumor tissues, two regions of the CpG island (CGI) in the SOCS3 promoter were subjected to methylation analysis and only the region close to the translational start site of SOCS3 was hypermethylated. In HCC tumor tissues, SOCS3 showed an increased methylation frequency and intensity compared with that in the adjacent non-tumor tissues. Moreover, SOCS3 expression was significantly down-regulated in HCC cell lines and tumor tissues, and this was inversely correlated with methylation. Kaplan–Meier curve analysis revealed that in patients with an hepatitis B virus (HBV) infection background, SOCS3 hypermethylation was significantly correlated with a poor clinical outcome of HCC patients. Our findings indicated that SOCS3 hypermethylation has already happened in non-tumor tissues and increased in both frequency and intensity in tumor tissues. This suggests that the methylation of SOCS3 could predict a poor prognosis in HBV infection-related HCC patients.


Introduction
Hepatocellular carcinoma (HCC) is a common malignancy worldwide with rapid progress and high short-term mortality. The initiation of HCC is intimately associated with chronically diseased liver tissue, induced by etiological factors such as hepatitis HBV or hepatitis C virus (HCV) infection, toxin exposure, excessive alcohol consumption, and/or other environmental or genetic factors [1]. Over 80% of HCC cases in China are attributed to chronic HBV infection [2,3]. Although surgical excision is still the most effective treatment so far, a large proportion of HCC patients show symptoms of intrahepatic metastases or post-surgical recurrence [1]. Therefore, the identification of more sensitive and specific biomarkers associated with HCC prognosis is essential to improve HCC management in China [4].
In the liver, the sequential progression to carcinoma is a multistep process with an accumulation of genetic and epigenetic alterations leading to the activation of oncogenes and the inactivation of tumor suppressor genes (TSG) [5,6]. DNA methylation, which is the first epigenetic regulation, has been suggested as one of the most important molecular mechanisms in hepatocarcinogenesis [5,7]. Mounting studies have revealed that HCC tumors exhibit specific DNA methylation signatures that are associated with major risk factors and tumor progression [8]. However, the methylation status of TSGs in HCC with different virus infection, especially HBV infection, is not fully understood.
The suppressor of cytokine signaling (SOCS) gene family has been found to play essential roles in suppressing tumor progression by inhibiting JAK/STAT, NFκB signaling, and promoting p53 signaling [9][10][11][12]. Among the SOCS family, SOCS3 has been reported to be hypermethylated in various kinds of cancers, including head and neck cancer, lung cancer, prostate cancer, Barrett esophagus carcinoma, and ulcerative colitis-related colorectal cancer [13][14][15][16][17][18][19]. In HCC, Niwa et al. revealed that SOCS3 was hypermethylated in 33.3% (6/18) of HCC tissues [20]. However, in our previous report, we did not observe SOCS3 hypermethylation in HBV-related HCC tissues [21]. These conflicting results may be due to the complicated etiological mechanism of HCC and the different detected loci in the SOCS3 promoter region. Therefore, it is imperative to explore the detailed methylation status of SOCS3 in HCC with different virus infection backgrounds and the relationship between its methylation and clinicopathology.
In this study, the methylation status of two loci of the CpG island (CGI) in the SOCS3 promoter was quantitatively investigated, with particular attention to the change in methylation intensities in primary HCC tissues with different virus infection backgrounds. Also, the clinicopathological significance of SOCS3 methylation status was statistically analyzed.

Suppressor of Cytokine Signaling 3 (SOCS3) CpG Island (CGI) Is Hypermethylated in the Region Close to the Translational Start Site
Using the online University of California Santa Cruz (UCSC) database, we searched for CGI at the SOCS3 promoter region and one CGI surrounding the SOCS3 gene promoter was predicted. This CGI region starts from −880 nt and ends at +1339 nt, covering the translation initiation site of the SOCS3 gene. In our previous study, we detected the methylation status in the CGI region from +173 to +296 nt (region 1) in HCC tissues and did not find any methylation in this region. However, Niwaet et al. observed obvious methylation in the CGI region in HCC tissues, which was close to the translational start site ( Figure 1A). To investigate whether the above conflicting results were due to the different detected regions in the SOCS3 promoter, we designed another pair of primers to amplify the CGI region from +797 to +885 nt (region 2), which was close to the translational start site and near the locus reported in the Niwaet et al. report [20].
Firstly, we detected the methylation statuses of both region 1 and region 2 in 20 pairs of the HCC tumor and non-tumor tissues. With the cut-off value setting at 10%, any tissue that had a hypermethylation intensity ≥10% would be judged as the hypermethylation status, and the higher value implicated a higher CGI hypermethylation status in intensity [15]. As shown in Figure 1B, at target region 2, 12 of the 20 tumor tissues and six of the non-tumor tissues showed hypermethylation. The methylation intensity in HCC tumor tissues was also found to be significantly higher than that in the adjacent non-tumor tissues (p = 0.0028). However, the methylation status at region 1 showed no hypermethylation in both tumor and non-tumor tissues. Next, we detected the methylation statuses of both regions in eight HCC cell lines. We found that region 2 of the SOCS3 CGI was hypermethylated in cell lines Huh-7, Hep3B, PLC/PRF/5, and SMMC7721, while it was unmethylated in SNU449, SNU182, SK-Hep1, and Huh-1 cells ( Figure 1C). Similar to that in HCC tissues, no methylation was detected in region 1 in all these cell lines (data not shown). These results indicated that it was the locus close to the translational start site that showed increased methylation, rather than that near the transcription start site.

SOCS3 Methylation Is Associated with Altered mRNA Expression in Hepatocellular Carcinoma (HCC) Cell Lines
To further investigate whether the hypermethylation of region 2 in the SOCS3 CGI could suppress SOCS3 gene expression, the expression levels of SOCS3 in these eight HCC cell lines were quantitatively measured by real-time qPCR. As shown in Figure 2A, the SOCS3 mRNA expression level was relatively lower in cell lines with high SOCS3 CGI methylation such as Huh-7, Hep3B, PLC/PRF/5, and SMMC7721. In contrast, in cell lines with SOCS3 CGI non-methylation, such as SNU449, SNU182, and SK-Hep1 cells, the mRNA expression of SOCS3 was relatively higher. Next, we processed Huh-7, Hep3B, and SK-Hep1 cells with the treatment of DNA demethylation reagent 5-aza-2ʹ-deoxycytidine. We found that after 5-aza treatment, the mRNA levels of SOCS3 significantly increased in Huh-7 and Hep3B cells with hypermethylation, but not in the SK-Hep1 cells with non-methylation ( Figure 2B). These data suggest that the methylation in CGI region 2 is associated with the down-regulation of SOCS3 expression in HCC cell lines. In addition, we noticed that the mRNA expression level of SOCS3 was also very low in Huh-1 cells with non-methylation, suggesting that besides methylation modulation, there might be other regulation mechanisms of SOCS3 expression, such as acetylation [22].

SOCS3 CGI Is Hypermethylated in Primary HCC Tumor Tissues
Next, we detected the methylation status of region 2 in SOCS3 CGI in an enlarged HCC sample. In the paired 127 HCC specimens, 61 (48.03%) of the tumor tissues and 32 (25.21%) of the adjacent non-tumor tissues showed methylation (p = 0.0002), which meant that in non-tumor tissues, the SOCS3 methylation had already happened. Meanwhile, the hypermethylation intensity in tumor tissues was significantly higher than that in the adjacent non-tumor tissues (26.63% vs. 8.31%, p < 0.0001, Figure 3A). In addition, in tumor tissues, there were 26.77% (34/127) specimens in which he SOCS3 methylation intensity was over 40%, while there were only 2.36% (3/127) in non-tumor tissues.
To test whether the SOCS3 CGI methylation in HCC tissues was associated with different virus infections, we analyzed the methylation frequency and intensity among 92 pairs of HCC tissues with HBV infection, 13 pairs with HCV infection, and 22 pairs with neither HBV nor HCV infection. In these three groups of HCC, the methylation frequencies in tumor tissues were 44.57% (41/92), 46.15% (6/13), and 63.63% (14/22), respectively; and in the adjacent non-tumor tissues, 25% (23/92), 0 (0/13), and 45.45% (10/22). Additionally, the methylation intensity of tumor tissues in each group was significantly higher than that in their corresponding non-tumor tissues (p < 0.0001, p = 0.048, and p = 0.0013). However, no marked difference was found among the tumor tissues with different virus infection backgrounds ( Figure 3B). Taken together, these results suggested that the frequency of SOCS3 CGI hypermethylation in HCC tissues was not related to the different virus infection backgrounds of HCC.

SOCS3 Is down-Regulated in Primary HCC Tumor Tissues
We also detected SOCS3 mRNA expression levels in 32 pairs of HBV-related HCC tumor tissues and non-tumor tissues ( Figure 4A). As expected, the level of SOCS3 mRNA in tumor tissues was statistically lower than that in non-tumor tissues (p = 0.0115). Among these 32 tumor tissues, the SOCS3 methylation intensity in five tumor tissues was extremely high (≥40%) and 10 specimens unmethylated (<3%). We found that the expression level of SOCS3 in tumor tissues with extremely high SOCS3 methylation was significantly lower than that in the tumor tissues without methylation ( Figure 4B). Comparing the SOCS3 expression levels of these two groups, they was inversely correlated with the CGI hypermethylation status. These data further demonstrated that the down-regulation of SOCS3 expression in HCC tissues was associated with SOCS3 methylation.

Methylation of SOCS3 Is Associated with the Poor Prognosis of HCC with Hepatitis B Virus (HBV) Infection
We then statistically analyzed the association between SOCS3 hypermethylation and patient clinicopathological features in HCC cases. As shown in Table 1, SOCS3 methylation was not correlated with the tumor size, tumor-node-metastasis (TNM) stage, cirrhosis background, portal vein tumor thrombosis, and tumor encapsulation of the HCC patients. It was worth noting that in the older group of HCC patients (≥55 years), the methylation frequency of SOCS3 in HCC tissues was higher than that in the younger group (<55 years). However, the difference of SOCS3 methylation between younger and older HCC groups was not statistically significant.
We then evaluated the prognostic value of the SOCS3 methylation status in HCC patients. The Kaplan-Meier curve analysis indicated that, for all the patients, there was no difference in clinical outcome between the methylated and unmethylated patients (p = 0.633, Figure 5A). However, in patients with an HBV infection background, SOCS3 hypermethylation was significantly correlated with the poor prognosis of HCC patients (p = 0.0457, Figure 5B), which meant that HCC patients with SOCS3 hypermethylation exhibited shortened overall survival compared with those with SOCS3 unmethylation, while no correlation was found in the rest of the patients with HCV or no virus infection background (data not shown).

Discussion
The aberration of SOCS3 in human malignancy has been noted in recent years, but the detailed methylation status of SOCS3 in HCC with different virus infection backgrounds has not been well understood. Here, we confirmed that SOCS3 was hypermethylated in HCC tissues and SOCS3 hypermethylation was significantly correlated with the poor prognosis of HCC patients with HBV infection.
Using Methylation-sensitive restriction enzyme-based quantitative PCR (MSRE-qPCR) assay, we found that it was the CGI region close to the translational start site of the SOCS3 gene that was frequently hypermethylated in HCC tumor tissues and cell lines. SOCS3 hypermethylation in HCC tissues was not related to the etiological mechanism of HCC. However, SOCS3 CGI methylation was associated with poor prognosis in HCC with an HBV infection background, while it was not associated with poor prognosis in HCC with HCV or no virus infection background, which indicated that SOCS3 might play a more powerful role in HBV infection-related HCC; the detection of the methylation of the SOCS3 gene was of value for the prognosis of HCC with HBV infection. Additionally, we also found that SOCS3 CGI hypermethylation occurred partially in non-tumor tissues, although its intensity and frequency were far lower than those in tumor tissues. Since most of these HCC patients had a cirrhosis background, it was reasonable to suppose a potential oncogenic role for SOCS3 hypermethylation during hepatocarcinogenesis from as early as the precancerous cirrhotic stage. Furthermore, no methylation was found in non-tumor tissues in HCC patients with HCV infection, suggesting that different hepatitis viruses might have different effects on the methylation status of SOCS3.
Aberrant methylation of gene promoter regions is one of the mechanisms for TSG inactivation in cancers. The decreased expression of SOCS3 has been reported to be correlated with poor patient survival in HCC tissues [23]. We also observed the frequent down-regulation of SOCS3 mRNA levels in HCC cell lines and tumor tissues, which was inversely associated with the SOCS3 methylation status. In addition, treatment by the demethylation agent could restore SOCS3 expression in the methylated cell lines. These findings confirmed that methylation modulation was an important inhibiting factor for SOCS3 expression. However, in Huh-1 cell lines and some HCC tissues with low expression of SOCS3, no methylation was detected, which indicated that other mechanisms could also lead to the down-regulation of SOCS3, including mutation, acetylation, and microRNA regulation, etc.
It has been reported that through inhibiting the JAK/STAT signaling pathway, the SOCS gene family plays an important role in suppressing tumor initiation and development [14]. As an essential member of this gene family, SOCS3 may function as a tumor suppressor gene. In previous reports, SOCS3 down-regulation was associated with vascular invasion and overall survival [24,25]. The silence of SOCS3 expression facilitated tumor formation and growth in lung and liver [26], while the loss of SOCS3 could promote aggressiveness in hepatocellular carcinoma [27]. However, the underlying mechanism of SOCS3 in the pathogenesis of HCC remains elusive. Further study should be conducted to determine the precise intracellular signaling and explore potential mechanisms of SOCS3 in the initiation and development of HCC.
In conclusion, our findings suggest that it is the CGI region close to the translational start site that is methylated in HCC, which might contribute to the aberrant expression of SOCS3. The hypermethylation of SOCS3 has already happened in non-tumor tissues and increased in tumors in both frequency and intensity. SOCS3 CGI methylation was associated with poor prognosis in HCC with a HBV infection background. However, it still needs a larger cohort to confirm this result.

Tissue Samples
Primary HCC and the adjacent non-tumor tissues were obtained from 127 patients who underwent routine curative surgical treatment at Henan Cancer Hospital in Zhengzhou, Henan Province, China, from 2008 to 2014. Among the 127 patients, 92 patients were infected with HBV, 13 patients with HCV, and 22 patients with no virus infection background. All patients were diagnosed by ultrasonography and computed tomography, and confirmed by liver biopsy. No patients had received radiotherapy or chemotherapy prior to surgery. The 127 patients were histologically diagnosed with HCC. The clinicopathological characteristics were shown in Table 2. The follow-up data were described in detail previously [28]. The study protocol was approved by the institute ethics committee, and informed consent was acquired from all patients and donors before the study commenced.

RNA Extraction and Real-Time RT-qPCR
Total RNA was prepared using total RNA isolation (TRIZOL) reagents (Invitrogen, Carlsbad, CA, USA) in accordance with the manufacturer's instructions. The expression level of SOCS3 was measured through the real-time RT-qPCR assay using the Roche lightcycler 480 (Roche, Mannheim, Germany) according to the manufacturer's instructions. The house-keeping gene HMBS was included in the assay as control reference to normalize the expression levels, and primers of HMBS have been described previously [30]. The SOCS3 qPCR primers were as follows: forward: GGAGACTTCGATTCGGGACC; reverse: GGTACTCGCTCTTGGAGCTG. Each experiment was conducted in duplicate or triplicate, and the comparative Ct method (2−∆Ct) was adopted to determine the relative expression levels of SOCS3.

Statistical Analyses
The differences between groups were analyzed by two-tailed paired or unpaired t-test. The χ2 test or Fisher's exact test were adopted to determine the association between the SOCS3 methylation status and the clinicopathological features. The Kaplan-Meier analysis with log-rank test was used to analyze the primary liver cancer (PLC) prognosis and survival time. All of the statistical analyses were performed by GRAPHPAD PRISM software (GraphPad Software, Inc., San Diego, CA, USA). In all cases, a p-value of less than 0.05 was considered significant.

Conclusions
Our findings suggested that SOCS3 hypermethylation has already happened in non-tumor tissues, while its frequency and intensity were increased in tumors. The methylation of SOCS3 predicted a poor prognosis in HBV infection-related HCC patients.