Comprehensive Analysis of DNA Methylation Regulator DNA Methyltransferase (DNMT) Family and Ten-Eleven-Translocation (TET) Enzymes Family in Pan-Cancer

Background: DNA methyltransferase (DNMT) family and ten-eleven-translocation (TET) family enzymes play pivotal roles in regulating DNA methylation, and are closely related to diverse cancers. This study was designed to clarify the specic roles of DNMT and TET genes in pan-cancers. Methods: The expression, mutation, copy number variations (CNVs), cancer-related pathways, immune cell inltration correlation, and prognostic potential of DNMT/TET genes were systematically investigated in 33 cancer types using next-generation sequence data from the Cancer Genome Atlas database. Results: DNMT3B was more highly expressed in the majority of tumors analyzed than in normal tissues. Most DNMT/TET genes were frequently mutated in uterine carcinosarcoma, and TET1 and TET2 showed higher mutation frequencies in various cancer types. DNMT3B exhibited inclusive copy number amplication in almost all cancers, such as stomach adenocarcinoma(STAD) and colon adenocarcinoma(COAD)l, while most DNMT/TET genes displayed highly copy number deletion in kidney chromophobe (cid:0) KICH (cid:0) . DNMT/TET genes were mainly involved in the following cancer-related pathways: UV response DN, mitotic spindle, cholesterol homeostasis, TGF beat signaling, xenobiotic metabolism, G2/M checkpoint, and E2F targets. DNMT/TET genes were signicantly correlated with NK cells, CD4 positive T cells, and Tfh cells. Additionally, Most DNMT/TET genes were signicantly associated with lower survival rates of adrenocortical carcinoma (ACC), mesothelioma, and liver hepatocellular carcinoma (LIHC), but played a protective role in thymoma (THYM). Furthermore, overexpression of most DNMT genes, except for DMAP1, was associated worse prognoses in pan-cancer. Conclusion: These results suggest that DNMT/TET genes can serve as potential predictors for prognosis and treatment in pan-cancer, providing new insight for future study.

Accumulating evidence suggests a close association between epigenetic changes and the onset and progression of multiple tumor types. DNA methylation plays a critically important in activating or silencing gene transcription [16][17][18]. Methylation of promoter region CpG islands is important in transcriptional regulation, and leads to tumor suppressor gene silencing in carcinogenesis [17,19].
DNMT/TET gene activity plays a pivotal role in the dynamic regulation of DNA methylation patterns [20,21], which is essential in tumorigenesis [2]. Abnormal methylation silences the secreted frizzled-related protein (SFRP) gene family, activating the classic Wnt pathway, and in uencing the occurrence and development of colorectal cancer [22]. In breast cancer, mir-22 inhibits TET genes expression, resulting in loss of function. This leads to increased promoter region methylation, tumor development, and metastasis [23].
The past decade had seen increasingly rapid advances in the eld of DNA methylation in tumorigenesis.
However, the speci c roles of DNMT/TET genes in tumorigenesis have not been fully elucidated. Given the role of epigenetic changes in cancer, it is important to understand the expression, mutations, and CNVs in DNA methylation regulators like the DNMT/TET genes and their prognostic potential. This study was designed to analyze DNMT/TET genes data from 33 types of cancers in The Cancer Genome Atlas (TCGA). We postulate that this work will generate fresh insight into the way DNMT/TET genes in uence cancer.

Collection of DNMT/TET genes
We collected six key DNMT family members and three key TET family members [13][14][15]

Analysis of differentially expressed genes in pan-cancer
Using the Deseq2 package in R, we identi ed differential DNMT/TET genes expression in pan-cancer, using |Log2 Fold Change|>1 and adjusted P-values < 0.05. TBtools software was used to draw a heatmap. [24] Protein-wide omics data across 20 cancer types from protein expression data Protein expression data were derived from Human Protein Atlas Datasets (https://www.proteinatlas.org/), DNMT/TET protein expression were systematically analyzed in twenty-one cancer types, including SKCA, Genome-wide mutation and CNVs in DNMT/TET genes across pan-cancer Mutation frequencies and CNVs were calculated. Using GraphPad Prism software, mutation and TPM expression were analyzed by Mann-Whitney U test, and the correlation between CNVs and gene expression was analyzed using the Kruskal-Wallis test. P<0.05 was considered to be statistically signi cant.
Correlation between DNMT/TET genes, oncogenic pathway activity, and immune cell in ltration The correlation between gene expression and cancer-related pathways were determined using Gene Set Variation Analysis (GSVA). This is a gene set enrichment (GSE) method to estimate pathway activity variation based on a sample expression matrix in an unsupervised, non-parametric manner. The Spearman Correlation Coe cient (SCC) was calculated to assess the correlation between gene expression and pathway activity, and immune cell in ltration. |SCC| >0.25 and adjusted P-value< 0.05 were considered to be a signi cantly correlation. The results were visualized by Cytoscape v3.7.1.

Clinical signi cance of DNMT/TET genes
The median expression of each gene was selected as a cut-off value to divide patients into two groups. Kaplan-Meier analysis, with the log-rank test, was used to compare survival rates between the two groups. Additionally, gene expression and prognostic value associations were analyzed using the GEPIA2(Gene Expression Pro ling Interactive Analysis 2) database. P-value< 0.05 was considered as statistically signi cant.

DNMT/TET genes expression pro les in pan-cancer
We summarized the chromosomal positions of all DNMT/TET family members across the published literature (Table 1). We then analyzed the gene expression differences in TCGA. The results showed heterogeneous expression of DNTM and TET genes in different cancer types. DNMT3B was highly expressed in the majority of tumors; DNMT3L was expressed in low levels in CHOL, KICH, KIRP, and KIRC, and had increased expression in UCEC READ, ESCA, and LUSC. (Figure 1a). The Log2FC and adjusted Pvalue of the results are listed in Supplementary Table 1. We visualized the differential expression of

DNMT/TET genetic alterations in pan-cancer
We examined the DNMT/TET gene mutation frequency in 33 cancer types. Most DNMT/TET genes were frequently mutated in UCEC, and rare mutations were observed in CHOL, PCPG, and THCA. TET1 and TET2 had higher mutation frequencies than did other genes in pan-cancer, and the overall average mutation frequency of DNMT/TET genes in different cancer types ranged from 0% to 34.9% (Figure 4a). We also investigated DNMT/TET gene CNVs in pan-cancer ( Figure 4b). DNMT3B exhibited inclusive copy number ampli cation in STAD, COAD, READ, BLCA, ESCA, and LUSC, but almost no copy number deletions in multiple cancer types. Most DNMT/TET genes displayed copy number deletions in KICH, but copy number variations were rarely observed in LIHC and LAML.
Next, we analyzed the relationship between DNMT/TET gene mutations and expression levels in human cancers and visualized the statistically signi cant results ( Figure 5). DNMT1 expression level changes were related to its mutation in COAD, ESCA, STAD, UCEC, and PRAD. Similarly, mutations in DNMT3L and TET2 also affected their expression levels in COAD. The details were shown in Supplementary Table 3.
The association between CNVs and DNMT/TET gene expression levels in multiple cancer types is shown in Figure 6. We found that almost all expression changes in DNMT/TET genes were associated with CNVs.
The correlation between DNMT/TET genes and cancer-related pathways We investigated the correlation between DNMT/TET gene expression and cancer related pathways to clarify the molecular signi cance of DNMT/TET genes in carcinogenesis. DNMT/TET genes were signi cantly correlated with multiple carcinogenic pathways, playing both activation and suppression roles (Figure 7a). We concluded that DNMT/TET genes are mainly involved in the following cancerrelated pathways: UV response DN, mitotic spindle, cholesterol homeostasis, TGF beta signaling, xenobiotic metabolism, G2/M checkpoint, and E2F targets (Supplementary Table4). We also found that DNMT3A, TET1, TET2, and TET3 were more likely to be related to the tumor occurrence and progression pathways (Figure 7b).

The correlation between DNMT/TET genes and immune in ltration
The correlation between DNMT/TET gene expression and immune cell in ltration in pan-cancer were assessed. DNMT/TET genes were signi cantly correlated with NK cells, CD4 positive T cells, and Tfh cells. Notably, the correlation between DNMT/TET genes and immune cell in ltration tended to be positive. In particular, TET2, TRDMT1, and TET3 showed more stronger correlation. (Figure 8)

Discussion
Previous studies veri ed an intimate relationship between DNMT/TET genes and cancers. However, the potential implication of these genes in pan-cancer is not well studied. To illustrate the pivotal role of DNMT and TET genes in cancer, we performed a thorough pan-cancer analysis of the expression and transcriptome patterns of core DNMT and TET genes using data obtained from TCGA. Our results show that expression of DNMT/TET genes is heterogeneous in certain cancers and paracancer tissues, and affects patient prognosis. The identi ed correlation between DNMT/TET gene expression, immune cell in ltration, and cancer-related pathways highlights the signi cance of DNMT/TET genes in tumorigenesis.
Our systematic analysis of TPM data of nine key DNMT/TET genes from TCGA database provides an overview of DNMT/TET gene expression pro les in 33 human cancer types. DNMT/TET gene expression in different cancer types was heterogeneous and DNMT3B was expressed in most cancer types. A review of the available literature supports our results, showing that expression of DNMT genes is increased in a host of malignant tumors, including PRCA, leukemia, BRCA, PAAD, and COCA [25]. Additionally, TET genes expression is substantially decreased in BRCA, LIHC, LUCA, PAAD, and PRCA [26]. DNMT3B expression was increased in BRCA, [27] PRCA, [28] and COAD [29]. We found that DNMT3L expression levels were increased in COAD, READ, and LUCA, and decreased in CHOL, PRAD, and BLCA. These results are consistent with those showing that DNMT3L is highly expressed in LUCA [30] and gastrointestinal stromal tumor [31]. It is worth noting that our ndings show that overexpression of DNMT3B and DNMT3L may have important impacts on numerous cancers. The underlying mechanisms of DNMT3B and DNMT3L in cancer deserve further investigation, and may provide new ideas for cancer treatment.
DNMT or TET genes mutations may be responsible for the occurrence and progression of tumors [32,33].
Through analyzing DNMT/TET genes mutations in multiple cancers we found that the total average mutation frequency ranged from 0% to 34.9%, and that TET1 and TET2 had relatively high mutation frequencies in pan-cancer. TET genes were frequently mutated in various cancers. TET genes mutations were previously detected in hematopoietic malignancies, melanoma, and some solid tumors [15,34]. TET2 mutations were frequently detected in myeloid malignancies (~15%) [35]. Moreover, genetic studies have shown that TET2 is essential for the self-renewal and differentiation of hematopoietic stem cells in mouse models, and have con rmed that TET2 inactivation leads to the development of myeloid malignancies [36][37][38]. TET2 mutation has been detected in AML, CMML, MPD, and MDS [39], and TET2 mutation frequency is signi cantly higher than that of TET1 and TET3. In solid tumors, TET genes mutations were observed in COCA [40] and CCRCC [41]. Interestingly, most DNMT/TET genes showed high mutation frequencies in UCEC, which is a cancer type considered to have high global mutation burden [42]. DNMT/TET genes mutations were rare in CHOL, PCPG, and THCA. Furthermore, TET genes usually functioned as cancer suppressors. Indeed, TET genes were reported to be targets of oncogenic miRNAs [23], and reduced TET genes expression is often detected in cancer cells [26]. Furthermore, TET genes mutations may interfere with TET genes expression, which plays a pivotal role in human tumors [43,44]. Indeed, our results show that TET2 and TET3 mutations were signi cantly related to their expression in some cancers.
Sporadic CNVs signi cantly affect genomic stability in cancer, and CNVs are observed in nearly 80% of cancers [45,46]. We noticed that DNMT3B exhibited extensive CNV in different cancers and almost all DNMT/TET genes showed copy number deletion in KICH. Additionally, signi cant copy number changes were usually related to alterations in the expression of the corresponding genes [47]. We found that CNVs and DNMT/TET genes expression levels were associated in multiple cancer types. Together, these ndings strongly suggest that genetic changes might contribute to DNMT/TET genes expression changes in carcinogenesis. The potential regulation mechanism related to this deserves further exploration.
We found that most DNMT/TET genes were positively correlated with each other, especially DNMT3A-TET1, TET2-TET3, and TET1-TET3. DNMT/TET genes might have a synergistic effect in tumorigenesis, although additional investigation is necessary to determine the mechanism of interaction between these genes in oncogenesis.
Our results indicate that DNMT/TET genes expression is signi cantly associated with multiple cancerrelated pathways, especially in UV response DN, mitotic spindle, cholesterol homeostasis, TGF beat signaling, xenobiotic metabolism, G2/M checkpoint, and E2F targets. Individual DNMT/TET genes presented different associations with distinct cancer-related pathways, suggesting that each DNMT/TET gene may have different functions. These results are consistent with those of a previous study showing that DNMT/TET genes have distinct expression patterns and functions [9,48]. The methylation of mouse oocyte genome was less than 50%, which is signi cantly lower than that of the sperm genome, and is related to DNMT1 cytoplasmic retention [49,50]. TET1 was highly expressed in mouse ESCs [101]. Unlike TET1, TET3 expression was mainly con ned to oocytes and zygotes, where it appeared to contribute to the active demethylation or conversion of 5mC to 5hmc in male prokaryotes after fertilization [51]. Further investigations are required to analyze the common and diverse functions of these genes in tumorigenesis.
Analysis of the relationship between pan-oncogene expression and immune cell in ltration showed that NK, CD4 + T, and Tfh cells were most signi cantly associated with DNTM/TET genes, and with TET2, TRDMT1, and TET3 in particular. The TET protein family has an important in uence on maintaining immune system homeostasis by driving Treg cells, and that changes in this may result in the occurrence of cancers [52,53]. Another study suggested that TET2 regulates Th1 cells, playing a pivotal role in prevention of excessive in ammation in experimental autoimmune encephalomyelitis (EAE) [54].
Together, these ndings suggest that further study of DNMT/TET genes may provide novel anti-cancer therapeutics.
Lastly, we evaluated the value of DNMT/TET genes in pan-cancer prognosis. Most DNMT/TET genes were correlated with poor prognosis in ACC, MESO, and LIHC, and were correlated with good prognosis in THYM. Survival analysis also showed that high DNMT genes expression, except for DMAP1, was signi cantly associated with poor prognosis in pan-cancer and that TET1 and TET2 showed no statistical signi cance with overall survival in pan-cancer. Early studies showed that DNMT genes were poor prognostic factors in diverse cancers [14], including AML [55], STAD [56], and LUCA [57]. More interestingly, our results revealed that TRDMT1 and DNMT3A showed distinctly disparate prognoses in pan-cancer. DNMT3A expression is related to poor survival rate in hematologic cancers [58], and CMML [59], and in AML DNMT3A acts as a tumor suppressor [60]. This could be explained by different DNMT3A expression patterns in different cancers. Taken together, these ndings suggest that DNMT genes can serve as prognostic predictors in pan-cancer. Future studies are needed to reveal the critical role of DNMT/TET genes in the prognosis of various tumors.

Conclusion
The heterogeneous expression of DNMT/TET genes in pan-cancer is closely related to their mutation, CNVs, and cancer-related pathways, including UV response DN, mitotic spindle, cholesterol homeostasis, TGF beta signaling, xenobiotic metabolism, G2/M checkpoint, and E2F targets. These expression levels of DNMT/TET genes are also relevant to immune in ltrating cells, including NK cells, CD4 positive T cells, and Tfh cells. Furthermore, DNMT/TET genes could serve as prognostic indicators, and might provide new insights for future studies.    The association between DNMT/TET genes expression levels and mutation frequency in pan-cancer.

Figure 6
The association between DNMT/TET genes expression levels and copy number variations in different cancers.   Survival analysis for DNMT/TET genes in pan-cancer. The median expression level was selected as the cut-off value.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.