Methylation of ZNF331 is an independent prognostic marker of colorectal cancer and promotes colorectal cancer growth

Background ZNF331 was reported to be a transcriptional repressor. Methylation of the promoter region of ZNF331 has been found frequently in human esophageal and gastric cancers. The function and methylation status of ZNF331 remain to be elucidated in human colorectal cancer (CRC). Methods Six colorectal cancer cell lines, 146 cases of primary colorectal cancer samples, and 10 cases of noncancerous colorectal mucosa were analyzed in this study using the following techniques: methylation specific PCR (MSP), qRT-PCR, siRNA, flow cytometry, xenograft mice, MTT, colony formation, and transfection assays. Results Loss of ZNF331 expression was found in DLD1 and SW48 cells, reduced expression was found in SW480, SW620, and HCT116 cells, and high level expression was detected in DKO cells. Complete methylation of the ZNF331 in the promoter region was found in DLD1 and SW48 cells, partial methylation was found in SW480, SW620, and HCT116 cells, and unmethylation was detected in DKO cells. Loss of/reduced expression of ZNF331 is correlated with promoter region methylation. Restoration of ZNF331 expression was induced by 5-aza-2′-deoxycytidine (DAC) in DLD1 and SW48 cells. These results suggest that ZNF331 expression is regulated by promoter region methylation in CRC cells. ZNF331 was methylated in 67.1% (98/146) of human primary colorectal cancer samples. Methylation of ZNF331 was significantly associated with tumor size, overall survival (OS), and disease-free survival (DFS) (p < 0.01, p < 0.01, p < 0.05). Methylation of ZNF331 was an independent poor prognostic marker for 5-year OS and 5-year DFS (both p < 0.05). ZNF331 suppressed cell proliferation and colony formation in CRC cells and suppressed human CRC cell xenograft growth in mice. Conclusions ZNF331 is frequently methylated in human colorectal cancer, and the expression of ZNF331 is regulated by promoter region methylation. Methylation of ZNF331 is a poor prognostic marker of CRC.


Background
Colorectal cancer (CRC) is the third most commonly diagnosed cancer in males and the second in females worldwide [1]. The incidence has sharply increased in the past two decades in Eastern countries with changing environmental factors, such as lifestyle and diet [2,3]. Accumulation of aberrant genetic and epigenetic changes plays an important role in CRC initiation and progression [4].
Numerous prospective cohort epidemiology studies have identified specific dietary and lifestyle factors that either promote or protect against CRC [5][6][7]. Consumption of red meat and animal fats increases CRC risk, whereas dietary fiber decreases risk [8,9]. Other studies suggest that the human gut microbiome is relatively stable in individuals over time except in the case of certain events such as food poisoning/infection or international travel. This likely reflects the hegemony of long-term dietary patterns on our gut microbiome [10,11]. All these factors may cause colorectal epithelial cell epigenetic changes and further induces tumorigenesis. Thus, identification of new epigenetic biomarkers for diagnosis, prognosis, prediction, and targeting therapy for CRC is necessary.
Zinc finger protein 331 (ZNF331) was first identified from thyroid tumors [12]. It is also known as RITA (rearranged in thyroid adenoma), ZNF361, and ZNF463 [13]. The ZNF331 gene is located at chromosome 19q13.42, a region in which loss of heterozygosity (LOH) was detected in prostate cancer [14]. In our previous study, we found that the ZNF331 gene is frequently methylated in human esophageal squamous cell cancer (ESCC) and it serves as a tumor suppressor in ESCC [15]. The function of ZNF331 in human CRC remains unclear. In this study, we analyzed the epigenetic regulation and the function of ZNF331 in human CRC.

Human tissue samples and cell lines
A total of 146 cases of primary CRC and 10 cases of noncancerous colorectal mucosa were collected from the Chinese PLA General Hospital in Beijing between May 2009 and November 2013. All cancer samples were classified according to WHO Classification of Digestive Tumors: the 4th Edition. All samples were collected under the guidelines approved by the institutional review board at the Chinese PLA General Hospital. Among the patient cases, 98 cases were male and 48 cases were female. The median age was 60 years old (range 33-86 years old). All cancer samples were classified according to the TNM staging system (AJCC2010), which included tumor stage I (n = 17), stage II (n = 58), stage III (n = 52), and stage IV (n = 19). Six CRC cell lines, DLD1, SW48, HCT116, SW480, SW620, and DKO (DNMT1 and DNMT3b double knockout from HCT116 cells, a generous gift from Stephen Baylin), were examined in this study and maintained in 90% Roswell Park Memorial Institute (RPMI) 1640 media supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 mg/ml streptomycin. HEK-293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. All cell lines were cultured in an atmosphere of 5% carbon dioxide at 37°C.

Cell viability assay
ZNF331 stably expressed and unexpressed DLD1 and SW48 cells were plated into 96-well plates at a density of 2 × 10 3 cells/well. The cells (5 × 10 3 ) were plated into 96-well plates before and after knockdown of ZNF331 in DKO cells. The cell viability was measured by the MTT assay at 0, 24, 48, and 72 h (KeyGEN Biotech, Nanjing, China). Absorbance was measured on a microplate reader (Thermo Multiskan MK3, MA, USA) at a wave length of 492 nm. The results were plotted as means ± SD.
Colony formation assay ZNF331 stably expressed and unexpressed DLD1 and SW48 cell lines were seeded in six-well plates at a density of 200 cells per well. DKO cells before and after

Statistical analysis
Statistical analysis was performed using SPSS 18.0 software (SPSS, IL, USA). Either χ 2 or Fisher's exact tests were used to evaluate the relationship between methylation status and clinicopathological characteristics. The two-tailed independent samples' t test was applied to determine the statistical significance of the differences between the two experimental groups.
Survival rates were calculated by the Kaplan-Meier method, and differences in survival curves were evaluated using the log-rank test. Cox proportional hazards models were fit to determine independent associations of ZNF331 methylation with overall survival (OS) and disease-free survival (DFS) outcomes. Two-sided tests were used to determine significance, and p < 0.05 was considered statistically significant.

Results
The expression of ZNF331 is silenced by promoter region hypermethylation in human CRC cell lines The expression of ZNF331 was examined by semiquantitative reverse transcription PCR (RT-PCR) and real-time quantitative RT-PCR in CRC cells. ZNF331 was highly expressed in DKO cells. Loss of ZNF331 expression was found in DLD1 and SW48 cells, and reduced expression was found in HCT116, SW480, and SW620 cells (Fig. 1a, b, all p < 0.01). Methylation of the ZNF331 promoter region was detected by methylationspecific PCR (MSP). Complete methylation was observed in DLD1 and SW48 cells, partial methylation was detected in HCT116, SW480, and SW620 cells, and unmethylation was found in DKO cells (Fig. 1c). These data indicate that loss of/reduced expression of ZNF331 is correlated with promoter region methylation in CRC cell lines. To further determine whether expression of ZNF331 was regulated by promoter region methylation, CRC cells were treated with 5-aza-2′-deoxycytidine (DAC). As expected, restoration of ZNF331 expression was found in DLD1 and SW48 cells after DAC treatment, and increased expression was detected in HCT116, SW480, and SW620 cells (Fig. 1a, b, all p < 0.01). These results suggested that ZNF331 expression is regulated by promoter region methylation in human CRC cell lines. To validate the MSP results, bisulfite sequencing (BSSQ) was employed. Dense methylation was observed in the promoter region of ZNF331 in DLD1 and SW48 cells, while unmethylation was found in DKO cells (Fig. 1d).
ZNF331 is frequently methylated in human primary CRC, and methylation of ZNF331 is a poor prognostic marker Methylation of ZNF331 was detected by MSP in primary CRC samples and normal colorectal mucosa. ZNF331 was methylated in 67.1% (98/146) of primary CRC and 0% (0/10) of normal colorectal mucosa (Fig. 2a). Methylation of ZNF331 was significantly associated with tumor size (p < 0.05,    No association was found between ZNF331 methylation and CIMP-high or each marker (p > 0.05, Table 3). Kaplan-Meier plots indicated that methylation of ZNF331 was significantly associated with poor 5-year overall survival (OS) (p = 0.001, Fig. 2b) and 5-year disease-free survival (DFS) (p = 0.024, Fig. 2c). According to Cox proportional hazards model analysis, ZNF331 methylation was an independent prognostic factor for poor 5-year OS and 5-year DFS after adjusting for tumor differentiation and TNM stage (both p < 0.05, Table 4). The expression of ZNF331 was evaluated by immunohistochemistry in 28 cases of available CRC and matched adjacent paraffin tissue samples (Fig. 2d). Reduced ZNF331 expression was found in 19 cases of CRC samples compared to adjacent tissue samples (Fig. 2e, p < 0.01). Among the 19 cases in which reduced expression of ZNF331 was detected, 16 cases were methylated, while in the 9 cases in which ZNF331 was expressed, only 3 cases were methylated. Reduced expression of ZNF331 was associated with promoter region methylation (p < 0.01, Fig. 2f). These results further suggest that the expression of ZNF331 is regulated by promoter region methylation in human CRC.
To further validate the methylation status and regulation of ZNF331 in human colorectal cancer, ZNF331 mRNA expression and promoter region methylation data were extracted from Genotype-Tissue Expression (GTEx) databases and the Cancer Genome Atlas (TCGA) (http://xena.ucsc.edu/). status of 14 CpGs in the promoter region. Methylation levels of ZNF331 were increased in CRC samples compared to adjacent tissue samples according to available data from 356 cases of CRC and 21 cases of matched adjacent tissue samples (p < 0.001, Fig. 2h). In the 356 cases of CRC samples, reduced expression of ZNF331 was associated with promoter region hypermethylation (p < 0.0001, Fig. 2i, j). These data further support our results in this study.

ZNF331 suppresses cell proliferation in CRC cells
To evaluate the effects of ZNF331 on cell proliferation, cell viability was detected by MTT assays. The OD value were 0.507 ± 0.020 vs. 0.436 ± 0.023 in DLD1 and 0.693 ± 0.018 vs. 0.560 ± 0.021 in SW48 before and after restoration of ZNF331 expression (Fig. 3a). Cell viability was significantly reduced after restoration of ZNF331 expression in DLD1 and SW48 cells (both p < 0.001). The OD values were 0.520 ± 0.031 vs. 0.603 ± 0.022 (p < 0.001) before and after knockdown of ZNF331 in DKO cells (Fig. 3a). Cell viability was increased after knockdown of ZNF33 in DKO cells. Colony formation assays were performed to evaluate the effect of ZNF331 on clonogenicity. The clone numbers were 164 ± 12.1 vs. 106 ± 20.3 in DLD1 cells and 148.7 ± 6.5 vs. 107.7 ± 9.9 in SW48 cells before and after restoration of ZNF331 expression. The number of clones was reduced after re-expression of ZNF331 in DLD1 and SW48 cells (both p < 0.01). The effect of ZNF331 on clonogenicity was further validated by knockdown of ZNF331 in DKO cells. The clone numbers were 158.7 ± 18.0 vs. 198.7 ± 10.0 before and after knockdown of ZNF331 in DKO cells (p < 0.05, Fig. 3b, c). b a c Fig. 4 ZNF331 induces G0/G1 phase arrest in CRC cells. a Cell phase distribution in ZNF331 unexpressed and re-expressed DLD1 and SW48 cells and before and after siRNA knockdown of ZNF331 in DKO cells (*p < 0.05, ***p < 0.001.). b ZNF331-3FLAG, cyclin D1, and cyclin E1 were detected by Western blot in ZNF331 unexpressed and re-expressed DLD1 and SW48 cells. c Cyclin D1 and cyclin E1 were detected by Western blot in DKO cells before and after siRNA knockdown of ZNF331 The effects of ZNF331 on cell cycle were analyzed by flow cytometry. The cell phase distributions were 41.71 ± 0.32% vs. 44.22 ± 0.34% in G0/G1 phase, 41.35 ± 0.75% vs. 43.22 ± 0.83% in S phase, and 16.94 ± 0.53% vs. 12.55 ± 1.17% in G2/M phase before and after re-expression of ZNF331 in DLD1 cells. In the SW48 cells, the cell phase distributions were 45.53 ± 0.46% vs. 48.09 ± 0.93% in G0/ G1 phase, 37.83 ± 0.44% vs. 36.43 ± 0.71% in S phase, and 16.31 ± 0.53% vs. 15.81 ± 1.29% in G2/M phase before and after re-expression of ZNF331. G1/S phase arrest was induced by restoration of ZNF331 in DLD1 and SW48 cells (p < 0.001, p < 0.05, Fig. 4a). To further validate these results, siRNA knockdown technique was employed. The cell phase distributions before and after knockdown of ZNF331 were as follows: G0/G1 phase 54.04 ± 0.24% vs. 51.25 ± 0.93%, S phase 33.71 ± 0.83% vs. 35.56 ± 0.32%, and G2/M phase 12.25 ± 0.82% vs. 13.19 ± 0.73%. G1/S arrest was induced by ZNF331 in DKO cells (p < 0.05, Fig. 4a). To further validate the effects of ZNF331 on cell cycle, the levels of cyclin D1 and cyclin E1 were examined by Western blot. The expression levels of cyclin D1 and cyclin E1 were reduced after re-expression of ZNF331 in DLD1 and SW48 cells, while the levels of cyclin D1 and cyclin E1 expression were increased after knockdown of ZNF331 in DKO (Fig. 4b, c). Taken together, these results suggest that ZNF331 inhibits cell proliferation in CRC.

ZNF331 suppresses CRC cell xenograft growth in mice
To further explore the role of ZNF331 in CRC, a xenograft mouse model was employed (Fig. 5a). ZNF331 unexpressed and re-expressed DLD1 cells were inoculated into the nude mice subcutaneously (Fig. 5d). The tumor volumes were 289.03 ± 52.22 mm 3 in ZNF331 unexpressed DLD1 cell xenografts and 180.6 ± 50.28 mm 3 in ZNF331 re-expressed DLD1 cell xenografts. The tumor volumes were smaller in ZNF331 re-expressed DLD1 cell xenograft mice compared to ZNF331 unexpressed DLD1 cell xenograft mice (p < 0.05, Fig. 5b). The tumor weights were 0.14 ± 0.03 g vs 0.09 ± 0.02 g in ZNF331 unexpressed and re-expressed DLD1 cell xenografts. The tumor weights were significantly different between these groups (p < 0.05, Fig. 5c). These results further suggest that ZNF331 suppresses CRC cell growth.

Discussion
Classical zinc finger proteins (ZNFs) form the largest family of sequence-specific DNA-binding proteins and are encoded by 2% of human genes [20,21]. Different types of zinc finger motifs influence a great diversity of biological functions, including differentiation, development, metabolism, apoptosis, autophagy, and stemness maintenance [22]. In addition to DNA binding, zinc finger motifs may interact with RNA, protein, and lipids [23,24]. Thus, ZNFs may play more extensive roles in gene regulation.
As a member of this family, ZNF331 may serve as a transcriptional repressor [25]. ZNF331 was previously reported to suppress esophageal and gastric cancer growth [15,26]. In this study, we found that ZNF331 is frequently methylated in human colorectal cancer and the expression of ZNF331 is regulated by promoter region methylation. Our results were supported by TCGA data. Methylation of ZNF331 is a potential colorectal cancer detection marker. The CpG island methylator phenotype (CIMP) was first identified by Toyota et al. and has been extensively studied in colorectal cancer [27]. A cause or molecular mechanism for CIMP in colorectal cancer has not yet been identified. CIMP has been associated with environmental and lifestyle factors [28,29], while there is no universal standard or consensus with respect to defining CIMP [30]. Weisenberger and colleagues identified a robust five-gene panel that recognized a distinct, heavily methylated subset of colorectal tumors that were also characterized by the BRAF mutation and MSI [31]. By screening eight methylation markers, Ogino et al. identified a four-gene methylation marker panel, including RUNX3, CACNA1G, IGF2, and MLH1, as a CIMP-high panel [17]. To explore the relationship of ZNF331 methylation and CIMP, we detected the methylation status of RUNX3, CACNA1G, IGF2, and MLH1, as well as KRAS or BRAF mutations in our cohort. No association was found between ZNF331 methylation and KRAS or BRAF mutations. No association was found between ZNF331 methylation and RUNX3, CACNA1G, IGF2, and/or MLH1 methylation. Our further studies indicate that methylation of ZNF331 is significantly associated with poor 5-year OS and 5-year DFS in CRC patients. Cox proportional hazards model analysis demonstrates that methylation of ZNF331 is an independent prognostic factor for poor 5-year OS and 5-year DFS in CRC. ZNF331 suppressed colony formation, cell proliferation, and induced G1/S arrest in colorectal cancer cells. ZNF331 suppressed human colorectal cancer cell tumor growth in xenograft mice. These results suggest that ZNF331 is a potential tumor suppressor in human CRC.

Conclusions
ZNF331 is frequently methylated in human colorectal cancer, and the expression of ZNF331 is regulated by promoter region methylation. Methylation of ZNF331 is a poor prognostic marker in human colorectal cancer. ZNF331 may serve as a tumor suppressor in human colorectal cancer.