MRVI1 and NTRK3 Are Potential Tumor Suppressor Genes Commonly Inactivated by DNA Methylation in Cervical Cancer

The abnormally methylated tumor suppressor genes (TSGs) associated with cervical cancer are unclear. DNA methylation data, RNA-seq expression profiles, and overall survival data were downloaded from TCGA CESC database. DMGs and DEGs were obtained through CHAMP and DESeq packages, respectively. TSGs were downloaded from TSGene 2.0. Candidate hypermethylated/down-regulated TSGs were further evaluated and pyrosequencing was used to confirm their difference in methylation levels of selected TSGs in cervical cancer patients. A total of 25946 differentially methylated CpGs corresponding to 2686 hypermethylated genes and 4898 hypomethylated genes between cervical cancer and adjacent normal cervical tissues were found in this study. Besides, 693 DEGs (109 up-regulated and 584 down-regulated) were discovered in cervical cancer tissues. Then, 192 hypermethylated/down-regulated genes were obtained in cervical cancer compared to adjacent tissues. Interestingly, 26 TSGs were found in hypermethylated/down-regulated genes. Among these genes, low expression of MRVI1 and NTRK3 was associated with poor overall survival in cervical cancer. Moreover, GEO data showed that MRVI1 and NTRK3 were significantly decreased in cervical cancer tissues. The expression levels of MRVI1 and NTRK3 were negatively correlated with the methylation levels of their promoter CpG sites. Additionally, elevated methylation levels of MRVI1 and NTRK3 promoter were further verified in cervical cancer tissues by pyrosequencing experiments. Finally, the ROC results showed that the promoter methylation levels of MRVI1 and NTRK3 had the ability to discriminate cervical cancer from healthy samples. The study contributes to our understanding of the roles of MRVI1 and NTRK3 in cervical cancer.

The abnormally methylated tumor suppressor genes (TSGs) associated with cervical cancer are unclear. DNA methylation data, RNA-seq expression profiles, and overall survival data were downloaded from TCGA CESC database. DMGs and DEGs were obtained through CHAMP and DESeq packages, respectively. TSGs were downloaded from TSGene 2.0. Candidate hypermethylated/down-regulated TSGs were further evaluated and pyrosequencing was used to confirm their difference in methylation levels of selected TSGs in cervical cancer patients. A total of 25946 differentially methylated CpGs corresponding to 2686 hypermethylated genes and 4898 hypomethylated genes between cervical cancer and adjacent normal cervical tissues were found in this study. Besides, 693 DEGs (109 up-regulated and 584 down-regulated) were discovered in cervical cancer tissues. Then, 192 hypermethylated/down-regulated genes were obtained in cervical cancer compared to adjacent tissues. Interestingly, 26 TSGs were found in hypermethylated/down-regulated genes. Among these genes, low expression of MRVI1 and NTRK3 was associated with poor overall survival in cervical cancer. Moreover, GEO data showed that MRVI1 and NTRK3 were significantly decreased in cervical cancer tissues. The expression levels of MRVI1 and NTRK3 were negatively correlated with the methylation levels of their promoter CpG sites. Additionally, elevated methylation levels of MRVI1 and NTRK3 promoter were further verified in cervical cancer tissues by pyrosequencing experiments. Finally, the ROC results showed that the promoter methylation levels of MRVI1 and NTRK3 had the ability to discriminate cervical cancer from healthy samples. The study contributes to our understanding of the roles of MRVI1 and NTRK3 in cervical cancer.

INTRODUCTION
Cervical cancer is the fourth most common cancer and the fourth leading cause of cancer death in women (1). The prognosis varies depending on the stage of cervical cancer. Compared with patients with early stage of cervical cancer, the five-year survival period of patients with advanced cervical cancer is much shorter (2). Therefore, the identification predictive biomarkers can help effective targeted therapy and treatment decisions.
Epigenetic processes can be reversed and this principle makes it a potential target for therapeutic intervention (3). Epigenetic variations could change the expression of tumor suppressor genes (TSGs) in cervical cancer (4). DNA methylation is an important part of epigenetics (5,6) and the regulatory effect of DNA methylation on gene expression has been studied extensively (7,8). DNA methylation levels could be detected by techniques, including pyrosequencing, methylation-specific polymerase chain reaction, methylation-sensitive highresolution melting, multiplex ligation-dependent probe amplification (MLPA), and Combined bisulfite restriction analysis (COBRA) and MethyLight (9). Aberrant methylation of TSGs could silence the expression of TSGs to consequently promote tumor formation (10). During recent decades, there have been a massive number of studies about TSGs in cervical cancer (11)(12)(13). For example, compared with the control samples, the promoter methylation frequency of TSG (including RARB, CADM1, PAX1, and DAPK1) in patients with invasive cervical cancer is higher (14). The silencing of TSGs is thought to be an early, driving event in the oncogenic process. Even after human papilloma virus (HPV) clearance, the silencing of TSGs by DNA hypermethylation could trigger carcinogenesis of the cervix (15). However, changes in DNA methylation and related abnormal TSGs expression have not been systematically elucidated in cervical cancer.
Gene methylation profiling and gene expression profiling have been utilized to investigate DNA methylation and gene expression in the molecular mechanism, biological process, and biomarker (16)(17)(18). Combined analysis of gene expression and DNA methylation data may contribute to identifying potential biomarkers of cervical cancer for treatment. Therefore, in this study, Illumina HumanMethylation450K methylation data and RNA-seq expression profiles from the Cancer Genome Atlas-Cervical Cancer (TCGA-CESC) were integrated for identifying the DMGs and DEGs in cervical cancer. First, TSGs among hypermethylated/down-regulated genes were found. Second, cervical cancer prognosis-related genes were selected and used as candidate cervical cancer-related TSGs. Then, expression levels of these TSGs were subsequently verified in three independent data sets from the Gene Expression Omnibus (GEO) database. Moreover, cervical cancer tissues and paired adjacent normal cervical tissues were collected to verify the methylation levels of these TSGs. Finally, receiver operating characteristic (ROC) curve analysis was used to assess the development of candidate cervical cancer related TSGs. This study aims to find prognostic and diagnostic TSGs related to cervical cancer through data analysis and experimental verification.

Differential Methylation and Expression Analysis
The workflow of our study is displayed in Figure 1. TCGA-CESC was used to identify aberrantly methylation-regulated genes. 12611 hypermethylated CpG sites, which were correspond to 2686 genes, were found in cervical cancer than that in adjacent normal cervical tissues. On the contrary, 13335 hypomethylated CpG sites, which were correspond to 4898 genes, were discovered in cervical cancer compared to adjacent normal cervical tissues (Figures 2A, B). Additionally, a total of 693 DEGs (109 upregulated and 584 down-regulated) were obtained from TCGA-CESC ( Figures 2C, D). Then, 192 hypermethylated/downregulated genes ( Figure 2E) and 60 hypomethylated/upregulated genes ( Figure 2F) were identified. Hypermethylated/ down-regulated genes were particularly focused in the current study.

GO and KEGG Pathway Enrichment of the Down-Regulated DEGs With Hypermethylation
The top 15 significant GO enrichments of biological processes were illustrated in Figure 3A, including extracellular structure organization, multicellular organismal signaling, extracellular matrix organization, and actin filament-based process. There were 13 enrichment pathways, such as vascular smooth muscle contraction, cGMP-PKG signaling pathway, calcium signaling pathway, focal adhesion, ECM-receptor interaction, proteoglycans in cancer, apelin signaling pathway ( Figure 3B).

Identification of Candidate TSGs
26 TSGs were discovered in hypermethylated/down-regulated genes ( Figure 4A). A total of 2361 cervical cancer survival-related genes were found by Kaplan-Meier analysis using RNA expression data. After integrated TSGs and survival-related genes, 2 overlapping genes (MRVI1 and NTRK3) were discovered and considered as the cervical cancer candidate TSGs ( Figure 4B).

Survival Analysis and Validation of MRVI1 and NTRK3 Expression
The expression levels of MRVI1 and NTRK3 genes were obtained in this study ( Figure 5A). As shown in Figures 5B, C, the expression levels of MRVI1 (P = 0.002) and NTRK3 (P = 0.029) were significantly lower in cervical cancer than those in adjacent normal cervical tissues. In addition, patients with low expression of MRVI1 (P = 0.026) and NTRK3 (P = 0.025) had significantly worse survival rates ( Figures 5D, E).

Validation of MRVI1 and NTRK3
Expression Levels by the GEO Database  Table 1). These results suggested that the expression levels of MRVI1 and NTRK3 could distinguish between cervical cancer patients and healthy controls.

Correlation Analysis of Promoter Region Methylation Level and Gene Expression Level
A total of 5 CpG sites were located in the promoter regions of MRVI1 (cg24365867, cg24541550, cg16014606, and cg15283950) and NTRK3 (cg14384532) ( Figure 7A). The methylation levels of these CpG sites were up-regulated in cervical tumors compared to controls (p < 0.05, Figures 7B-F). Further correlation analysis showed that the methylation levels of these CpG sites were negatively associated with gene expression for these two genes (p < 0.05, Figure 8).

Verification of Differences in Promoter Methylation Levels of MRVI1 and NTRK3 by Pyrosequencing Experiments
In order to verify the differential methylation levels of MRVI1 and NTRK3 between cervical cancer and adjacent normal cervical tissues, pyrosequencing experiments were conducted. As shown in Table 2 and Figure 9A, the methylation levels of cg24365867, cg24541550, cg16014606, and cg15283950 of MRVI1 gene in cervical cancer were significantly higher than that in adjacent normal cervical tissues (p < 0.05). Compared with adjacent normal tissues, a significantly elevated methylation level of cg14384532 on NTRK3 was also found in cervical cancer tissues (p < 0.05, Table 2 and Figure 9A). In addition, ROC

DISCUSSION
Cervical cancer is one of the most common types of cancer and represents a major global health challenge (1). Since aberrant DNA methylation occurs very early during tumorigenesis (19), it could therefore be used as an early diagnostic biomarker (20). In this study, hypermethylated and significantly lower expressions of TSGs MRVI1 and NTRK3 were discovered in cervical cancers than that in normal cervical tissues using the bioinformatics. The differences of MRVI1 and NTRK3 expressions between cervical cancer specimens and normal cervical tissues were further verified via three GEO datasets. Besides, the low expression of MRVI1 and NTRK3 was negatively associated with high methylation levels of promoter CpG sites. Moreover, promoter hypermethylation levels of MRVI1 and NTRK3 were also found in our clinical cervical cancer samples. ROC curve analyses proved the diagnostic value of MRVI1 and NTRK3 in cervical cancer. Furthermore, low expression of MRVI1 and NTRK3 was associated with poor prognosis of cervical cancer. These results enhanced our understanding of the DNA methylation pattern of TSGs in cervical cancer. MRVI1 is a protein-coding gene, which has been widely studied in cancer (21,22). MRVI1 was reported to regulate the cellular release of calcium signal (23), which plays an important role in cancer cell proliferation invasiveness (24). One study discovered that MRVI1 was transcriptionally activated by p53, and p53induced inhibition of colorectal cancer prognosis was depended on MRVI1 (25). Zhu et al. found that the MRVI1-AS1/ATF3 signalling pathway could increase paclitaxel chemosensitivity by modulating the Hippo-TAZ signalling pathway in nasopharyngeal cancer (21). Another research found that miR-940 could promote proliferation and metastasis of endometrial carcinoma through the regulation of MRVI1 (22). High expression of MRVI1 had a better prognosis than that of the low expression of MRVI1 in endometrial carcinoma (22). Unfortunately, the role of MRVI1 in cervical cancer has not yet been reported. In the current study, the overall survival of cervical cancer patients with low MRVI1 expression was also significantly shorter than those with high MRVI1 expression, which is consistent with previous endometrial carcinoma study.
NTRK3 encodes the TrkC protein, a member of neurotrophic tropomyosin receptor kinase (Trk) family, which autophosphorylates and motivates various signalling pathways such as MAPK and PI3K/AKT pathways (26). Trk aberrations, including gene fusion, gene overexpression, and single nucleotide variation, are involved in the pathogenesis of many cancers, among which NTRK3 gene fusion is extremely confirmed for oncogenic event (27). Unusual activation of NTRK3 and its fusion proteins may balance the epithelialmesenchymal transition (EMT), oncogenicity, and tumor growth rate via triggering various signalling pathways (28). ETV6-NTRK3 gene fusion acted as a potent oncogene driver and had been presented in the majority of cases of infantile fibrosarcoma (29). Oncogenic fusions in NTRK family receptor tyrosine kinases had been identified in several cancers and could serve as therapeutic targets, for instance in spitz tumors (30), fibrosarcoma (31), gastrointestinal stromal tumors (32), and inflammatory myofibroblastic tumors (33). Conversely, NTRK3 expression was a good prognosis factor in a variety of cancers and more specifi cally in melanomas (34), neuroblastomas (35), and colorectal cancer (36). NTRK3 expression and activation had been shown to trigger apoptosis in medulloblastoma cells (37). In recent years, SPECC1L-NTRK3 gene fusion was found in cervical sarcoma patients (38). However, the research on NTRK3 gene in cervical cancer is rare. Depending on the present study, NTRK3  expression was significantly lower in cervical cancer specimens than that in normal cervical tissues, and low NTRK3 expression was associated with a poor prognosis. These findings suggested that NTRK3 might likewise serve as a tumor suppressor gene in cervical cancer. In present study, two TSGs (MRVI1 and NTRK3) were identified via bioinformatics. Nevertheless, a total of 26 hypermethylated/ down-regulated TSGs have been discovered, the rest TSGs should be further studied. Although, the hypermethylation levels of MRVI1 and NTRK3 were verified in 9 cervical cancer tissues by pyrosequencing, the large number of clinical samples should be collected in further study. Hypermethylated and down-regulated expression levels of TSGs MRVI1 and NTRK3 have been identified in the current study; however, the detail epigenetic regulatory mechanism under cervical cancer still needs further investigation. In summary, our results revealed that hypermethylation in the promoter regions of MRVI1 and NTRK3 genes might lead to low expression in cervical cancer. Low expression levels of MRVI1 and NTRK3 were associated with poor prognosis of cervical cancer. The methylation levels and expression levels of MRVI1 and NTRK3 had the ability to effectively discriminate cervical cancer from healthy samples. Therefore, MRVI1 and NTRK3 genes may play important roles in the occurrence and prognosis of cervical cancer. It could be further explored and validated as a therapeutic target for cervical cancer. In conclusion, the down-regulation of MRVI1 and NTRK3 may drive cervical cancer through hypermethylation of their promoters. Further studies are needed to draw more attention to the roles of these TSGs in cervical cancer.

Differential Methylation and Gene Expression Analysis
Between cervical cancer tissues and normal cervical tissues, significant DMGs and DEGs were identified using DESeq package (42) and CHAMP package of R (43), respectively. The false discovery rate (FDR) was adopted to avoid the occurrence of false-positive results. FDR < 0.05 and |Log2 Fold change (Log2FC)| > 1 were used to select significant DMG or DEG.

Gene Ontology (GO) and KEGG Pathway Analysis
GO and KEGG pathway enrichment analysis of hypermethylated/ down-regulated genes was performed using the g:Profiler program (44).

Searching for TSGs Associated With Cervical Cancer
Among hypermethylated/down-regulated genes, TSGs were identified based on TSGene 2.0 (45). With the median expression level as the demarcation point, 291 patients with clinical data in TCGA were divided into low-risk group and high-risk group. Kaplan-Meier analysis in the survival package of R (46) was used to compare the difference in overall survival between the two groups. Prognostic-related TSGs were considered as candidate genes for cervical cancer. To solve the problem of a small number of normal tissues in TCGA-CESC, the expression levels of cervical cancer candidate TSGs in three GEO datasets (GSE29570, GSE39001, and GSE52903) were further compared by T test using R.

Pyrosequencing Experiment
Nine pairs of cervical cancer specimens and adjacent normal cervical tissues were obtained from the Second Affiliated Hospital of Wenzhou Medical University. Cervical cancer patients were diagnosed by experienced pathologists based on the results of surgically removed specimens (Supplementary Table 1). Human genomic DNA was extracted from tissue samples using the Genomic DNA Extraction Kit (Qiagen, Dusseldorf, Germany). DNA concentrations were determined by the Infinite F200 Tecan microplate reader (Tecan, männedorf, Switzerland). Primers were designed using the PyroMark Assay Design Software 2.0 and bisulfite-treated DNA PCR-amplified using the PyroMark PCR kit prior to analysis on a PyroMark Q96 according to manufacturer's instruction (Qiagen, Dusseldorf, Germany). Sequences of the PCR primers were shown in Table 3. Amplification was carried out as follows: 95°C for 3 min, followed by 40 cycles of 94°C for 30 s, 56°C for 30 s, and 72°C for 1 min, with a final elongation step at 72°C for 7 min. Raw data were analyzed using PyroMark Q96 software (Qiagen, Dusseldorf, Germany). The research protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Wenzhou Medical University. Written informed consents were obtained from all subjects.

Statistical Analysis
Pearson correlation coefficient was used to correlate promoter methylation levels with candidate TSGs expression levels. ROC curves were used to compare the sensitivity and specificity of the candidate TSGs expression levels and promoter methylation levels in the prediction of cervical cancer. All the data were analyzed using R scripts. A two-tailed p value < 0.05 was considered statistically significant.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.