Comprehensive Analysis of m5C RNA Methylation Regulator Genes in Clear Cell Renal Cell Carcinoma

Background Recent research found that N5-methylcytosine (m5C) was involved in the development and occurrence of numerous cancers. However, the function and mechanism of m5C RNA methylation regulators in clear cell renal cell carcinoma (ccRCC) remains undiscovered. This study is aimed at investigating the predictive and clinical value of these m5C-related genes in ccRCC. Methods Based on The Cancer Genome Atlas (TCGA) database, the expression patterns of twelve m5C regulators and matched clinicopathological characteristics were downloaded and analyzed. To reveal the relationships between the expression levels of m5C-related genes and the prognosis value in ccRCC, consensus clustering analysis was carried out. By univariate Cox analysis and last absolute shrinkage and selection operator (LASSO) Cox regression algorithm, a m5C-related risk signature was constructed in the training group and further validated in the testing group and the entire cohort. Then, the predictive ability of survival of this m5C-related risk signature was analyzed by Cox regression analysis and nomogram. Functional annotation and single-sample Gene Set Enrichment Analysis (ssGSEA) were applied to further explore the biological function and potential signaling pathways. Furthermore, we performed qRT-PCR experiments and measured global m5C RNA methylation level to validate this signature in vitro and tissue samples. Results In the TCGA-KIRC cohort, we found significant differences in the expression of m5C RNA methylation-related genes between ccRCC tissues and normal kidney tissues. Consensus cluster analysis was conducted to separate patients into two m5C RNA methylation subtypes. Significantly better outcomes were observed in ccRCC patients in cluster 1 than in cluster 2. m5C RNA methylation-related risk score was calculated to evaluate the prognosis of ccRCC patients by seven screened m5C RNA methylation regulators (NOP2, NSUN2, NSUN3, NSUN4, NSUN5, TET2, and DNMT3B) in the training cohort. The AUC for the 1-, 2-, and 3-year survival in the training cohort were 0.792, 0.675, and 0.709, respectively, indicating that the risk signature had an excellent prognosis prediction in ccRCC. Additionally, univariate and multivariate Cox regression analyses revealed that the risk signature could be an independent prognostic factor in ccRCC. The results of ssGSEA suggested that the immune cells with different infiltration degrees between the high-risk and low-risk groups were T cells including follicular helper T cells, Th1_cells, Th2_cells, and CD8+_T_cells, and the main differences in immune-related functions between the two groups were the interferon response and T cell costimulation. In addition, qRT-PCR experiments confirmed our results in renal cell lines and tissue samples. Conclusions According to the seven selected regulatory factors of m5C RNA methylation, a risk signature associated with m5C methylation that can independently predict prognosis in patients with ccRCC was developed and further verified the predictive efficiency.

Clear cell renal cell carcinoma (ccRCC) is the most common histological subtype of RCC, accounting for 70% [2]. Surgical treatment is still an effective way to treat ccRCC, because it is not sensitive to radiotherapy and chemotherapy [3,4]. However, nearly a third of patients were reported experiencing a recurrence after surgery with a median survival time of 1.9 years, resulting in a worse overall survival (OS) [5]. Due to the unclear biological progression and molecular mechanism of ccRCC, it is difficult to predict the prognosis of patients with ccRCC accurately. Therefore, we urgently need to find the potential developmental mechanisms of ccRCC and search for new therapeutic targets with better diagnostic and prognostic value.
Recently, increasing studies have suggested that m 5 C methylation also plays a vital role in pathological conditions such as cancer [15]. Chen et al. revealed that m 5 C methylation could promote the pathogenesis of human bladder urothelial carcinoma by stabilizing oncogene mRNAs [16]. Another study indicated that gene signatures of m 5 C methylation genes might predict prognoses of patients with head and neck squamous cell carcinoma [9]. However, the potential mechanisms of m 5 C RNA modification involved in the development of ccRCC were still unknown.
In this present research, we aim to explore the association between ccRCC patients with the expression of RNA m 5 C modification genes in the TCGA-KIRC cohort. Furthermore, the relationships between the genes and the clinicopathological parameters were also analyzed. Then, we screened out seven m 5 C RNA methylation-related genes to construct a risk signature model to predict the overall survival (OS) of the patients with ccRCC. The prediction accuracy of this risk signature was further validated in the testing group and the entire cohort, respectively.

Materials and Methods
2.1. Data Acquisition and Preprocessing. All public gene expression data, along with their clinical annotations such as age, gender, grade, stage, TNM classification, and survival status, were acquired from TCGA (http://cancergenome.nih .gov/). The gene expression profiles of 539 ccRCC cases and 72 normal controls were analyzed for further research, and 513 ccRCC patients with complete clinical information, including OS, were further selected from the 539 ccRCC cases (Table 1).

Identification of Differentially Expressed m 5 C-Related
Genes in TCGA Database. We screened differential expression genes (DEGs) according to the adjusted p value < 0.05, log 2 fold change > 1 or < −1 between the ccRCC tissue samples and normal renal tissue samples. The m 5 C RNA methylation regulator genes were further screened from DEGs. Univariate Cox regression analysis was used to evaluate the m 5 C RNA methylation regulator genes significantly related to OS according to p value < 0.05.

Construction of PPI Network and Correlation Analysis.
Protein-protein interaction (PPI) network was constructed for these twelve m 5 C methylation regulators using the STRING online database (http://string-db.org/). Moreover, by using the Pearson correlation analysis, we performed the correlation analysis among these regulators and the different clinicopathological parameters were also analyzed 2.5. m 5 C-Related Prognostic Signature Generation and Prediction. Univariate cox analysis was carried out to identify possible prognostic m 5 C RNA methylation regulators in ccRCC. Then, m 5 C RNA methylation-related risk score was constructed by using the LASSO Cox Regression algorithm in the training cohort. The m 5 C-related risk signature was calculated with the following formula: where n, coef ðiÞ, and xðiÞ represent the number of genes, coefficient, and the relative expression value of each gene, respectively. On the basis of the median score in the training cohort, we divided all ccRCC patients into the low-risk and high-risk groups. Log-rank test and Kaplan-Meier curve were applied, respectively, for revealing whether the risk score can differentiate the OS in ccRCC patients. To evaluate the predictive ability of these risk model, we further analyzed the receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) by the package of "survi-valROC" in R language.
2.6. Validations of the Seven-Gene Risk Score in the Testing and the Entire Cohort. The seven-gene risk score was further verified in the testing and entire cohort. Log-rank test and Kaplan-Meier curve were applied, respectively, for revealing whether the risk score can differentiate the OS in ccRCC patients. ROC and AUC were further analyzed in the testing and entire cohort.

Construction and Validation of Nomogram.
The nomogram was applied to predict survival of ccRCC patients. Selected seven m 5 C RNA methylation regulators genes and survival states were used to build the nomogram using R "rms" packages. The calibration curve was used to evaluate the accuracy of the nomogram in differentiating between patient groups.
2.8. GO and KEGG Pathway Enrichment Analyses. GO and KEGG enrichment analyses of seven selected genes were performed with R packages "clusterProfiler," "enrichplot," and "ggplot2." Only terms with both p and q values of < 0.05 were considered significantly enriched.

Single-Sample Gene Set Enrichment Analysis (ssGSEA).
To further evaluate the prognostic value of this risk model in TCGA, single-sample Gene set enrichment analysis (ssGSEA) was performed in two different risk score groups. When the false detection rate (FDR) was less than 0.25 and the normalized p value was less than 0.05, it was considered to be significantly enriched.    International Journal of Genomics 2.14. Statistical Analysis. All statistical data and figures were analyzed by R 4.0.3. Using Wilcoxon's test, we assessed the different expressions of m 5 C RNA methylation regulators between ccRCC patients and normal tissues. The correlation analysis among m 5 C RNA methylation regulators was carried out by the Pearson correlation analysis. The chisquare test was performed to evaluate the association between the risk score and clinicopathological parameters. All statistical results with p < 0:05 were considered statistically significant.

Functional Annotation and Pathway Enrichment
Analyses of Twelve m 5 C-Related Genes. To identify the function and potential pathway involved in ccRCC of m 5 C RNA methylation genes, we perform GO and KEGG enrichment analyses. Our results revealed that these m 5 C RNA methylation genes were remarkably enriched in the biological processes (BP) associated with methylation, macromolecule methylation, RNA methylation, and RNA modification. Among the molecular function (MF) analyses, they were significantly enriched in chromosomal region, heterochromatin, and organellar large ribosomal subunit. Through the cellular component (CC), m 5 C RNA methylation genes were significantly enriched in methyltransferase activity, methyltransferase activity, RNA methyltransferase activity, and rRNA methyltransferase activity (Figures 2(a) and 2(b)). KEGG pathway analysis found that m 5 C RNA methylation genes were mainly enriched in cysteine and methionine metabolism, microRNAs in cancer, mRNA surveillance, spliceosome, and RNA transport (Figures 2(c) and 2(d)).

Construction of a PPI Network and Correlation Analysis.
To further explore the underlying mechanism, the PPI network was constructed on the basis of the STRING database.      International Journal of Genomics PPI network analysis showed that NSUN5, NSUN3, NSUN4, NSUN7, NOP2, and TET2 were the essential genes ( Figure 3(a)). In correlation analysis, we found that these genes were strongly correlated at the transcriptional level ( Figure 3(b)). Among the regulatory factors of m 5 C RNA methylation, the positive interaction between NSUN3 and TET2 (r = 0:7) and the negative interaction between NSUN5 and TET2 (r = −0:31) were the most significant ( Figure 3(c)). In addition, a network diagram showed the interaction, function, and prognostic value between twelve m 5 C RNA methylation regulators ( Figure 3(d)).

Genetic Alterations of m 5 C RNA Methylation Regulators
and Association with Immune Cell Infiltration. Then, we investigate the relationship between twelve m 5 C RNA methylation regulators and CNV mutations. We observed widespread CNV on twelve m 5 C RNA methylation regulators through copy number variation (CNV) analysis. Among these genes, NOP2, ALYREF, NSUN2/3, and TRDMT1 showed high CNV amplification frequency. In contrast, NSUN4/5/7, TET2, and DNMT1/3A/3B had significantly high CNV deletion frequency (Figure 4(a)). The locations of CNV of twelve m 5 C RNA methylation regulators on chromosomes are shown in Figure 4(b). Nevertheless, the effects of CNV of the m 5 C RNA methylation regulators-based signatures were further analyzed to clarify the association with different immune cell infiltrations. Adopting the TIMER database, the CNV of the identified m 5 C regulators signatures, including deep deletion, arm-level deletion, diploid/normal, arm-level gain, and high amplification, significantly affected the infiltration levels of B cells, CD4+ T cells, CD8+ T cells, neutrophils, macrophages, and dendritic cells in ccRCC (Figures 4(c)-4(n)). These results illustrated the underlying mechanisms that m 5 C RNA methylation regulators had pivotal regulatory effects on the tumor immune microenvironment for ccRCC patients.

Consensus
Cluster Analysis to Define m 5 C Subtypes. We used unsupervised consensus clustering analysis to identify two subtypes based on the m 5 C RNA methylation regulator expression profiles. Clearly, k = 2 was assumed to be the most appropriate choice for dividing the tumor samples into two different clusters (Figures 5(a)-5(c)). Additionally, the principal component analysis (PCA) was performed to compare the transcriptional profile between the two clusters, and the result was shown in Figure 5(d). For comparing the overall survival (OS) of ccRCC patients between the two clusters, the Kaplan-Meier method was applied. The result indicated that the ccRCC patients in cluster 1 had a significantly longer OS than cluster 2 (p = 0:006) ( Figure 5(e)). The comparison of clinicopathological features between the two clusters indicated a significant difference in terms of the stage (p < 0:01), grade (p < 0:001), and TNM stage (p < 0:001), while there was no significant difference in age and gender between the two clusters ( Figure 5(f)). The clinicopathological features between two clusters were shown in Table 2.
In conclusion, the results above suggested that the clustering was intimately connected with the clinicopathological characteristics of ccRCC.

Prognostic Risk Signature of m 5 C RNA Methylation
Regulators. Then, we performed univariate Cox regression analysis based on the expression levels of these regulators from TCGA to investigate the prognostic value of these twelve m 5 C RNA methylation regulators in ccRCC. The results suggested that eleven regulators were significantly associated with overall survival (OS) (p < 0:05), among which, TET2, NSUN3, NSUN4, NSUN7, and TRDMT1 were considered as protective genes with HR < 1, while NSUN5, NSUN2, DNMT3B, DNMT3A, ALYREF, and NOP2 were considered as risky genes with HR > 1( Figure 6(a)). Furthermore, ten genes were intersected from differentially expressed genes and genes related to OS by the Venn diagram ( Figure 6  Cox regression analysis, including these ten genes, was performed to predict the prognosis of ccRCC through m 5 C RNA methylation regulators better. Consequently, seven genes, composed of NOP2, DNMT3B, NSUN3, NSUN5, TET2, NSUN2, and NSUN4, were screened for the construction of a prognostic risk signature via LASSO Cox regression analysis (Figures 6(c) and 6(d)). The expression levels of the seven genes selected out in various cancers on the basis of the TCGA database were consistent with the result we obtained in ccRCC ( Figure S1B-H). The risk score for each patient was calculated with the following formula (Table 3): m 5 C Risk Signature = 0:358712202257608 * NOP2 + 0:229671617272167 * NSUN2 + 0:150827601647872 * NSUN5 + 0:453815651228113 * DNMT3B + ð− 0:00949630611801698Þ * NSUN3 + ð−0:604181882439325Þ * TET2 + ð−0:09915438936325Þ * NSUN4: For the sake of evaluating the prognostic value of the risk signature, 267 random selected ccRCC patients with com-plete follow-up information in the training cohort were divided into the high-risk group (135 patients) and the low-risk group (132 patients) according to the calculated median risk score above, and the overall survival of the two groups was compared. The result indicated that the low-risk ccRCC patients had a better prognosis with a longer OS (Figure 6(e)). The time-dependent ROC curve analysis suggested that the prognostic signature with the AUC values for the 1-, 2-, and 3-year survival in the training cohort were 0.792, 0.675, and 0.709, respectively ( Figure 6(f)). The result of PCA in the training cohort was shown in Figure S5A, and the result of stochastic neighbor embedding (t-SNE) was displayed in Figure S5B. The heatmap showed higher expression levels of the seven riskrelated m 5 C RNA methylation regulators (NOP2, DNMT3B, NSUN3, NSUN5, TET2, NSUN2, and NSUN4) in the highrisk group compared to the low-risk group (Figure 6(g)).
The distributions of the risk scores and their survival status    10 International Journal of Genomics were displayed (Figure 6(h)). These results, taken together, suggested that this risk score was a good predictor for the prognosis of ccRCC.

m 5 C Risk Signature Can Be an Independent Prognostic
Factor. So as to explore whether the risk signature can act as an independent prognostic factor, the univariate and multivariate Cox regression analyses were performed in the training cohort. The results indicated that the risk score was significantly related to the worse OS with HR = 2:796 (p < 0:001, 95% CI 1.977−3.954) (  (Table 4). In addition, the risk score was also related to the worse OS with HR = 1:779 (p = 0:002, 95% CI 1.594−2.147) according to the multivariate Cox analysis (Table 4). In conclusion, these results above suggested that signature-based risk score can be an independent prognostic factor for ccRCC.
3.8. Validation of the Risk Signature in the Testing Cohort and Entire Cohort. To further verify the prognostic accuracy of the risk score, we tested it in the testing cohort and the whole cohort. The results of the univariate Cox regression analysis indicated that the risk score in the testing cohort, and the entire cohort was significantly related with the worse          (Table 4). Similarly, patients were divided into two groups according to the median risk score, respectively. Kaplan-Meier curves analysis in the testing cohort demonstrated that the patients in low risk had longer OS than those in high risk (Figure 7(a)). To detect the effectiveness of this model, the ROC analysis of 1/2/3-year OS was carried out in the testing cohort. The AUC values for the 1-, 2-, and 3-year survival in the testing cohort were 0.660, 0.642, and 0.683, respectively (Figure 7(b)). The results of PCA and t-SNE in the testing cohort were shown in Figure S5C and Figure S5D, respectively. The heatmap in the testing cohort showed that higher expression levels of the seven riskrelated m 5 C RNA methylation regulators in the high-risk group compared to the low-risk group (Figure 7(c)). The distributions of the risk scores and the patient survival status between the low-and high-risk groups in the testing cohort were displayed in Figure 7(d). Similar results were further verified in the entire cohort. The result of Kaplan-Meier curves analysis in the entire cohort showed that the patients in low risk had longer OS than those in high risk (Figure 8(a)). The ROC values of 1-/2-/3-year OS in the entire cohort were 0.740, 0.664, and 0.699, respectively (Figure 8(b)). The result of PCA in the entire cohort was shown in Figure S5E. The result of t-SNE in the entire cohort was shown in Figure S5F. The heatmap in the entire cohort showed that these seven risk-related m 5 C RNA methylation regulators were highly expressed in the highrisk group compared to the low-risk group (Figure 8(c)). The distributions of the risk scores and the patient survival status between the low-and high-risk groups in the entire cohort were displayed (Figure 8(d)).

Construction of Nomogram in the Training Cohort, the
Testing Cohort, and the Entire Cohort. In order to predict the 1-year, 2-year, and 3-year overall survival of each patient, the nomograms were designed in the training cohort, testing cohort and whole cohort, respectively. The expression signature for the seven risk-related genes was used as variables. Figure 9(a) presented the seven variables of the training cohort, and the calibration curve compared well with the ideal model was shown in Figure 9(b). The nomograms for 1-, 2-, and 3-year OS were also constructed in the testing and entire cohort. The results in the testing cohort and the entire cohort were similar with that in the training cohort and were displayed in Figures 9(c) and 9(d) and Figures 9(e) and 9(f), respectively. These results above, taken together, suggested that m 5 C risk signature is a good predictor for the prognosis in ccRCC.

Kaplan-Meier Survival Curves of m 5 C RNA Methylation Regulators in ccRCC Patients.
We further analyzed the association between the seven selected m 5 C RNA methylation regulator genes and the OS and disease-free survival (DFS) of ccRCC patients in the TCGA database. Kaplan-Meier survival curves and log-rank test showed that the OS of ccRCC patients with higher expression of DNMT3B, NOP2, NSUN2, and NSUN5 was significantly shorter compared to those with lower expression. In contrast, ccRCC patients with higher expression of NSUN3, NSUN4, and TET2 will have longer OS than those with lower expression ( Figure S2A-G). Moreover, higher expression of NSUN3, NSUN4, and TET2 was correlated with significantly longer disease-free survival (DFS). However, there were any significant differences in the DFS of ccRCC patients with differential expression of NOP2, NSUN2, NSUN5, and DNMT3B ( Figure S3A-G). Overall, our results demonstrated that these seven m 5 C RNA methylation regulators are potential prognostic biomarkers that can accurately predict survival outcomes of ccRCC patients.

GSEA Analysis Reveals Potential Signaling Pathways
Related to Risk Score. To further understand the biological    14 International Journal of Genomics function and potential signaling pathways between the highrisk and low-risk groups, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEEG) pathway analyses were performed. The results of GO analysis indicated that the high-risk group are enriched in the regulation of various enzyme activities, such as negative regulation of hydrolase activity, negative regulation of proteolysis and so on (Figure 10(a)). Moreover, KEGG pathway analysis suggested that these genes were associated with various cancer-related pathways, including ERBB pathway, MAPK pathway, mTOR pathway, renal cell carcinoma, pathway in cancer, TGF-β pathway, and Wnt pathway, which gives a clue of the underlying mechanism in the pathogenesis of ccRCC (Figure 10(b) and S4, Table 5). Further, ssGSEA analysis was applied to explore the different infiltration degrees of immune cell types, immune-related functions, and immune-related pathways between the low-risk group and the high-risk group. The results indicated that the immune cells with different infiltration degrees between the two groups were T cells including follicular helper T cells, Th1 cells, Th2 cells, and CD8+ T cells (Figure 10(c)). The main differences in immune-related functions between the two groups were the interferon response and T cell costimulation ( Figure 10(d)). The above results revealed that these seven selected genes were tumor-related, and the risk score can independently predict prognosis in patients with ccRCC. was downregulated in renal cell lines. However, NSUN3 mRNA expression level showed no significant difference (Figure 11(a)). Furthermore, we explored these m 5 C RNA methylation-related genes' expression level in clinical tissue samples, which was consistent with our results in the TCGA database and cell lines (Figure 11(b)). In addition, m 5 C RNA modification levels in the cell lines were measured by Global RNA Methylation Assay Kit. m 5 C RNA modification levels in renal cancer cell lines (786-O, Caki-1) were dramatically higher than that in HK-2 cell lines (Figure 11(a)).

Discussion
Renal clear cell carcinoma (ccRCC), the most common type of adult kidney carcinoma, is characterized by poor prognosis and high risk of metastasis and recurrence [17]. No effective therapeutic strategies for ccRCC patients with advanced stage or metastasis were founded, and the rate of 5-years disease-free survival in patients with metastasis is only 12% [18]. Therefore, identifying the effective diagnostic and prognostic biomarkers for early diagnosis and accurate    prognosis is urgent to improve and prognose survival outcomes of ccRCC patients. The malignant progression and prognosis of ccRCC were associated with epigenetic modifications, including DNA methylation [19], histone modification [20], microRNA changes [21,22], and RNA modification [23,24]. The m 5 C methylation, the second common methylation modification, plays essential roles in various cellular, biological, and pathological processes. The methylation of cytosine is regulated by the genes that we call "writers," "erasers," and "readers." The "writers" can upregulate m 5 C methylation, and the "erasers" can reverse the level of m 5 C methylation. The "readers" can bind to an m 5 C site and modulate differential biological signals.
Recent research revealed that m 5 C methylation is associated with prognosis in many cancers [9,16]. The molecular mechanisms associated with the prognosis of ccRCC are still unknown. Therefore, we explored the relationship between m 5 C methylation regulators and the prognosis of ccRCC in this study.
In the present study, we found that eleven of twelve m 5 C RNA methylation regulators were abnormally expressed in ccRCC, among which, ten genes were associated with the prognosis. According to these ten selected m 5 C RNA meth-ylation regulators, two clusters with significant differences of OS were distinguished in the ccRCC patients of TCGA. Moreover, on the basis of the training cohort, seven genes (NOP2, NSUN2, NSUN3, NSUN4, NSUN5, TET2, and DNMT3B) were screened for the construction of a prognostic risk signature. The results showed that the risk score was effective in predicting the clinical outcomes of ccRCC. Similarly, the independent prognostic value of this seven-gene risk score was represented again in the testing and the entire cohort, indicating that the risk signature had good performance in prognosis prediction. Univariate and multivariate Cox regression analyses in the whole cohort showed that age, stage, grade, TNM stage, and risk score were significantly associated with OS. the risk score could also predict the prognosis of the ccRCC patients with different clinicopathological parameters. The above results, taken together, suggested that the clinical outcomes were worse in the high-risk ccRCC patients than in the low-risk.
In our study, the higher expression level of NSUN4, NSUN3, and TET2 presented a better prognosis in ccRCC patients, indicating that these three genes might inhibit the progression of ccRCC. NSUN4 is a dual functional mitochondrial protein, enabling 12S rRNA methylation and coordinating mitochondrial assembly [25]. NSUN4 participates in cell proliferation and differentiation, protein biosynthesis, and cancer [26]. He et al. reported a high expression of NSUN4 in advanced liver cancer [27], contrary to our results. The reason may be that advanced tumors require a lot of energy and tumor cells improve their mitochondrial activity and obtain more energy by self-upregulating NSUN4 and other genes. NSUN3 was a putative methyltransferase in mitochondria [28], whose aberration may lead to a variety of diseases [29,30]. Many studies have reported that TET2 can inhibit tumors, and its mutation can induce the occurrence and development of tumors [31][32][33][34]. Our results are consistent with the above results, indicating that TET2 is a tumor suppressor gene.

18
International Journal of Genomics In contrast, the higher expression level of the other genes NOP2, DNMT3B, NSUN2, and NSUN5, the worse prognosis of ccRCC patients may have, which indicated that these four genes might promote the development of ccRCC. The high expression of NSUN5 can promote the proliferation of colon cancer cells [35], and the high expression of NOP2 can promote the metastasis of prostate cancer and the proliferation of liver cancer cells [36,37]. We came to the same conclusion about NSUN5 and NOP2 genes in our study. DNMT3B is expressed as an oncogene in various of tumors, including leukemia, liver cancer, and bladder can-cers [38][39][40]. Several studies have reported that NSUN2 can promote the development and metastasis of tumors [41][42][43].
To further understand the biological function and potential signaling pathways of these genes, we performed GO, KEGG, and ssGSEA analyses between tissues with different risk scores. The results above indicated that patients in the high-risk group might have the following changes in their bodies that result in a poor prognosis. Firstly, many articles have reported recently that the occurrence and development of tumors are related to inflammatory stimulation. Patients in the high-risk group may experience large inflammatory

19
International Journal of Genomics responses, resulting in more inflammatory cell infiltration and ultimately the inhibition of several key enzyme activities, which can induce cell apoptosis and mutation [44][45][46].
Secondly, the functions of various T cells including follicular helper T cells, Th1 cells, Th2 cells, and CD8+ T cells were suppressed in the high-risk group, causing the tumor cells

20
International Journal of Genomics to escape immunity and resulting in a poorer prognosis [47][48][49]. Thirdly, various cancer-related pathways, including ERBB pathway, MAPK pathway, mTOR pathway, pathways in cancer, TGF-β pathway, and Wnt pathway, were more likely to be activated, leading to poor outcomes. The mRNA expression level of m 5 C RNA methylation regulators and the global m 5 C RNA methylation level were measured in vitro and tissue samples, which is consistent with our above in silico analysis. Some limitations of this study are noteworthy. The results of the current study, obtained by bioinformatics analysis, were not entirely accurate so that more experimental and clinical studies were still needed to further verify our results and find out the potential mechanism of m 5 C RNA methylation in ccRCC.

Conclusions
Our results demonstrated that eleven out of twelve m 5 C RNA methylation regulators are dysregulated between ccRCC tissues and normal tissues, among which ten genes were associated with prognosis. We defined m 5 C molecular subtypes and constructed a m 5 C methylation-related risk signature in the training cohort, by which the OS rate of ccRCC patients can be forecasted independently. The efficiency of the risk signature was further proved in the testing and whole cohort, respectively. The overall survival rate of patients with high risk may be lower. In addition, we found out the potential pathways of m 5 C RNA methylation-related genes in ccRCC and verified our results in vitro and clinical tissue samples.   Figure S5: the analyses of PCA and t-SNE between the high-risk and low-risk patients in the training cohort, testing cohort, and entire cohort, respectively. (A, B) The results of PCA (A) and t-SNE (B) between the high-risk (red) patients and the low-risk (blue) patients in the training cohort. (C, D) The results of PCA (C) and t-SNE (D) between the high-risk (red) patients and the low-risk (blue) patients in the testing cohort. (E, F) The results of PCA (E) and t-SNE (F) between the highrisk (red) patients and the low-risk (blue) patients in the entire cohort. Table S1: oligonucleotide sequences used in this study. (Supplementary Materials)