Integrated Bioinformatic Analysis of DNA Methylation and Immune Infiltration in Endometrial Cancer

Background Endometrial cancer greatly threatens the health of female. Emerging evidences have demonstrated that DNA methylation and immune infiltration are involved in the occurrence and development of endometrial cancer. However, the mechanism and prognostic biomarkers of endometrial cancer are still unclear. In this study, we assess DNA methylation and immune infiltration via bioinformatic analysis. Methods The latest RNA-Seq, DNA methylation data, and clinical data related to endometrial cancer were downloaded from the UCSC Xena dataset. The methylation-driven genes were selected, and then the risk score was obtained using “MethylMix” and “corrplot” R packages. The connection between methylated genes and the expression of screened driven genes were explored using “survminer” and “beeswarm” packages, respectively. Finally, the role of VTCN1in immune infiltration was analyzed using “CIBERSORT” package. Results In this study, 179 upregulated genes, and 311 downregulated genes were identified and found to be related to extracellular matrix organization, cell–cell junctions, and cell adhesion molecular binding. The methylation-driven gene VTCN1 was selected, and patients classified to the hypomethylation and high expression group displayed poor prognosis. The VTCN1 gene exhibited highest correlation coefficient between methylation and expression. More importantly, the hypomethylation of promoter of VTCN1 led to its high expression, thereby induce tumor development by inhibiting CD8+ T cell infiltration. Conclusions Overall, our study was the first to reveal the mechanism of endometrial cancer by assessing DNA methylation and immune infiltration via integrated bioinformatic analysis. In addition, we found a pivotal prognostic biomarker for the disease. Our study provides potential targets for the diagnosis and prognosis of endometrial cancer in the future.


Introduction
Endometrial cancer, the second most common female malignancy, greatly threatens the health of women. According to the statistics in 2018, the numbers of new cases and deaths in the world were over 380,000 and 89,000, respectively [1]. With the increase in the prevalence of obesity and ageing population, the incidence of endometrial cancer is increasing [2]. Approximately 70% of endometrial cancer are confined to uterus when diagnosed; these cases are mainly treated by removing the uterus, which is associated with better prognosis and approximately 75% survival rate [3,4]. However, 10~15% of endometrial cancer extend beyond the uterine tissues when diagnosed. The survival rate in these cases is less than 5~15% [5]. It is highly desirable to develop novel targets for the intervention and treatment of endometrial cancer patients.
Cancer is generally considered to be an epigenetic disease caused by abnormal gene expression. The epigenetic alternation plays a critical role in the progression of endometrial cancer [6,7]. DNA methylation, a common kind of epigenetic change, can activate or silence some genes to promote or inhibit related signaling pathways [8]. Abnormal DNA methylation, including hyper-and hypomethylation, can both result in the occurrence of disease [9]. DNA methylation in cancers, such as lung cancer [10], prostate cancer [11], and breast cancer [12], have been widely studied.
With the development of single-cell technology, the role of tumor microenvironment (TME), which is consisted of immune cells, fibroblasts, endothelial cells, inflammatory mediators, and extracellular matrix, has attracted increasing attention in the study of endometrial cancer [13,14]. Immunoresponse is an important factor for the prognostic value in endometrial cancer [15]. Usually, the increased number of cytotoxic T lymphocytes (CTLs, CD8 + T cells) is an independent prognostic indicator for better prognosis in endometrial cancer [15]. CIBERSORT is broadly performed to explore the abundance of immune cells in normal and tumor tissues [16,17]. Many recent studies have examined the role of immune cells in TME in the mechanism of cancers, such as prostate cancer [18], clear cell renal cell carcinoma [19], and endometrial cancer [20]. However, few have studied the disease mechanism and prognosis biomarkers related to DNA methylation and immune cell infiltration.
In this study, we explored the prognostic biomarkers of DNA methylation and immune cell infiltration. VTCN1, exhibits est correlation coefficient between methylation and expression, was identified. VTCN1 (a T cell activation suppressor 1), also known as B7-H4, can regulate T cell activation in non-small-cell lung cancer [21], hepatocellular carcinoma [22], and prostate cancer [23]. To further explore whether VTCN1 is involved in regulating the occurrence and development of endometrial cancer through immune factors, we analyzed the level of immune cell infiltration in normal and tumor tissues. The results showed that VTCN1 is indeed negatively correlated with CD8 + T cells in endometrial cancer, and there was lower T cell infiltration in tumor tissues in the high VTCN1 expression group. Finally, we demonstrated that VTCN1 was upregulated in tumors, and hypomethylation and high expression of VTCN1 were associated with poor prognosis. Our study demonstrated that VTCN1 may be involved in the occurrence and development of endometrial cancer by inhibiting CD8 + T cell infiltration. This finding may contribute to understand the mechanism of disease and provide a prognostic biomarker for endometrial cancer.

Data Download and Preprocessing.
The latest RNA-Seq (35 normal samples and 422 tumor samples) and DNA methylation data (46 normal samples and 422 tumor samples) were downloaded from the UCSC Xena dataset (https://xenabrowser.net/datapages/). The data were processed and normalized via R software. The differentially expressed genes (DEGs) were analyzed using the "limma" package. The DEGs were screened out with the criteria of jlogFCj > 2 and adj-P value < 0.05. Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were analyzed using the "org.Hs.e.g.db" package. A value of P < 0:05 was defined as statistically significant.
2.3. The Risk Score of the Above Genes. The clinical data were downloaded from the UCSC Xena dataset. Risk scores were calculated via the "survival" package. The survival curve and heat map were generated based on the high-and lowrisk scores. Subsequently, 5 genes were selected, including TSPYL5, KLF9, GYPC, VTCN1, and PGR. The risk score for each patient was performed as our previous article [24]: E denotes the normalized expression level of the gene, and b denotes the corresponding regression coefficient.

Survival Analysis.
A survival analysis including the highly methylated genes and genes with low expression was implemented in R software.
2.5. The Evaluation of the Protein Expression. The expression of GYPC, VTCN1, and PGR genes in the paired tumor and normal groups was analyzed by the "limma" and "ggpubr" packages. The protein expression was analyzed by The Human Protein Atlas (https://www.proteinatlas.org/).

Immune Cell Infiltration.
CIBERSORT is a deconvolution algorithm that uses a set of reference gene-expression values (a signature with 547 genes) considered a minimal representation for each cell type. Based on those values, cell type proportions in data from bulk tumor samples with mixed cell types are inferred using support vector regression. CIBERSORT can be applied to distinguish 22 human immune cells, including B cells, T cells, NK cells, macrophages, DCs, and myeloid subsets, based on the high specificity and sensitivity of the gene expression profile. To determine whether there is a correlation between tumor immune cells infiltrationand immunerelated gene expression, tumor infiltration with six types of immune cells (CD4 + T cells, CD8 + T cells, B cells, neutrophils, macrophages, mast cells, and dendritic cells) was analyzed by CIBERSORT.
2.7. The VTCN1 Gene Expression. The genes related to VTCN1 were selected via the STRING database (https:// string-db.org/). The GO pathways of these genes were analyzed using the Metascape dataset (http://metascape.org/ gp/index.html#/main/step1). The expression of VTCN1 was divided into two groups according to the wilcox test function. A visualization of the correction ofVTCN1expression with immune cells was generated using the "vioplot" package. Six patients with advanced endometrial cancer and normal tissues were collected. Fresh tissues are stored at -80°C for quantitative analysis. As we recently published, [

Functional Analysis of DEGs.
According to the criteria of jlogFCj > 2 and adj-P value < 0.05, there were 490 DEGs, among which there were 179 upregulated genes and 311 downregulated genes. The volcano plot is shown in Figure S1. Second, Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed via R software. The results indicated that endometrial cancer was involved in the cell cycle, P53 signaling pathway, and focal adhesion in KEGG pathway ( Figure 1(a)). GO function included biological process (BP), cellular component (CC), and molecular function (MF). The GO analysis  showed that endometrial cancer was related to extracellular matrix organization, cell-cell junctions, and cell adhesion molecular binding (Figure 1(b)).

Survival Analysis.
Survival analysis was performed with the product-limit method (Kaplan-Meier analysis). Additionally, the log-rank test (Mantel-Cox test) was used to compare the difference in the survival status between the high-(268 patients) and low-risk groups (269 patients) using the "Survminer" package. The Cox model was applied to build a risk model to obtain the risk value of patients, and   Figure 3(a)). Obviously, a lower risk indicates a better prognosis, and a higher risk indicates a poorer prognosis (Figure 3(b)). The heat map of the TSPYL5, KLF9, GYPC, VTCN1, and PGR expression is shown in Figure 3(c). Patients were classified to a high expression group and a low expression group based on the expression levels of PGR, GYPC, and VTCN1. The survival curves of the above 13 methylation-driven genes were generated using survival software. Three genes (PGR, VTCN1, and GYPC) were considered statistically significant (P < 0:05) (Figures 4(a)-4(c)). The hypomethylation and high expression group of VTCN1 displays poor prognosis, while the lower expression group showed better prognosis (P < 0:05). The negative correlations of PGR, VTCN1, and GYPC expression and methylation are displayed in Figures 4(d)-4(f) . The expression of PGR and GYPC is shown in Figure S2. Since the VTCN1 gene possesses the highest correlation coefficient between methylation and expression, we selected the VTCN1 gene for the follow-up study.

The Mechanism of the VTCN1 Gene in Immune
Regulation. VTCN1 is closely related to immunity. We fur-ther explored the mechanism by which VTCN1 is involved in endometrial cancer. The relative proteins of VTCN1 were identified using the STRING database: B7RP1, BTLA, CD28, CD80, CD86, CTLA4, ICOSL, IL4, IL6, and PDCD1LG2 ( Figure 5(a)). These genes are mainly involved in lymphocyte costimulation, regulation of T cell activation, proliferation, B cell activation, and immune response-regulating cell surface receptor signaling pathways ( Table 1). The percentage of immune cells in normal and tumor tissues was analyzed. Box plots according to the stromal score, immune score, estimate score, and tumor purity are displayed in Figure 5(b). The box plot of the immune cell percentage in the two groups was analyzed using the "ggpubr" package. The percentages of B native cell and CD4 memory resting T cells and M2 macrophages were clearly lower in the tumor group than in the normal group. Tregs and M1 macrophages were more abundant in the tumor group than in the normal group (P < 0:05). Likewise, the higher expression of VTCN1 exhibited a positive correlation with the abundance of resting memory CD4 T cells, while the higher expression of VTCN1 was negatively correlated with the abundance of T cells and activated memory CD4 T cells ( Figure 5(c)). The connection of CD8 + T cells and VTCN1 is shown in Figure 5(d). The expression of VTCN1 was visualized using the "Beeswarm" package. The VTCN1 expression was higher in the tumor group than in the normal group (Figures 5(e) and 5(f)). The protein expression in the Human Protein Atlas (https://www.proteinatlas.org/) further confirmed the results ( Figure 5(g)).

Discussion
Endometrial cancer is a lethal female reproductive malignant tumor. The incidence of endometrial cancer is usually second only to cervical cancer among gynecological diseases in China [26]. The average age of onset of endometrial   cancer is 63 years old, and it usually occurs in postmenopausal women, and in women with obesity and diabetes. Traditional treatments usually include surgical resection, radiotherapy, and chemotherapy. However, it is challenging to treat patients with advanced endometrial cancer, which often has a poor prognosis [27]. Currently, the mechanism of disease and prognostic biomarkers of endometrial cancer is unclear. DNA methylation and immune cell infiltration often participate in the development of various cancers, including gastric cancer [28], clear cell renal cell carcinoma [29], and colorectal cancer [30]. In our study, we found that high methylation and low expression of PGR and GYPC were associated with poor prognosis, while low methylation and high expression of VTCN1 were associated with poor prognosis. In addition, immune regulation driven by the high expression of VTCN1 in tumors may promote the development of endometrial cancer by inhibiting CD8 + T cell infiltration.
VTCN1 (B7-H4) belongs to the B7 family and functions as a cell surface transmembrane protein, negatively regulating the T cell-mediated immune response via interaction with a receptor protein on the surface of T cell to inhibit T cell activation and proliferation and cytotoxic factor production [31,32]. Several recent studies have shown that VTCN1 is often overexpressed in tumor tissues of ovarian [33], lung 11 BioMed Research International [34], and breast cancers [35]. Miyatake et al. demonstrated that VTCN1 is overexpressed in high-risk uterine endometrial cancer and negatively correlated with tumor T cell infiltration [36]. In our study, we found that VTCN1 is downregulated in tumor tissue via DNA methylation analysis. In addition, we performed the CIBERSORT algorithm to analyze the immune cell distribution in normal and tumor tissues. Subsequently, the relationship between VTCN1 expression and immune cell infiltration was analyzed using bioinformatic methods. The results indicated that the expression of VTCN1 is inversely correlated with CD8 + T cell infiltration.

Conclusion
In summary, our study first revealed the mechanism of endometrial cancer combining DNA methylation and immune cell infiltration. Hypomethylation of the VTCN1 promoter leads to its high expression, which can cause tumor development by inhibiting CD8 + T cell infiltration. Furthermore, the VTCN1 expression was higher in the tumor group than in the normal group, and hypomethylation and high expression of VTCN1 indicated poor prognosis. Our study explains the mechanism of immune infiltration and provides potential targets for the diagnosis and prognosis of endometrial cancer.

GO:
Gene Ontology KEGG: Kyoto Encyclopedia of Genes and Genomes DEGs: Differentially expressed genes.

Data Availability
The datasets analyzed during the current study are available in the TCGA and UCSC repository.

Conflicts of Interest
The authors declare that they have no competing interests.

Authors' Contributions
DFF, DZM, and WJL made the design of this study. DFF, DZM, and YCY did the acquisition and analysis of data. YMQ, YDY, LSY, and ZYJ did the interpretation of data. DFF and WJL drafted the article. LHX collected the samples. YCY and CYX revised this study critically. All authors approved of the version to be published. All authors read and approved the final manuscript. Fangfang Dai, Jinglin Wu, and Zhimin Deng have contributed equally to this work and share first authorship. Immune response-regulating cell surface receptor signaling pathway 12 BioMed Research International