UALCAN (http://ualcan.path.uab.edu/analysis.html) is an online professional database for analyzing tumor gene expression and methylation levels. We used the UALCAN database to analyze the expression and methylation levels of SHMT1 and SHMT2 in healthy individuals and patients with KICH, KIRC, and KIRP. Furthermore, we used the Student’s t-test for comparative analysis, and the difference was considered significant at a p-value 0.05 [19, 20].

Gene Expression Profiling (GEPIA) (http://gepia.cancer-pku.cn/index.html) is a free online platform for analyzing the correlation between gene expression levels and the pathological tumor stage and prognostic value. We used the GEPIA database to analyze the correlation between pathological tumor stage and the prognostic value of SHMT1 and SHMT2 expression levels in patients with KICH, KIRC, and KIRP. Comparative analysis was performed using Student’s t-tests, with a p-value less than 0.05 considered significant [19, 20].

cBioPortal (http://cbioportal.org) is an online professional database used to analyze genetic alterations in tumors. We used the cBioPortal database to analyze genetic alterations in SHMT1, SHMT2, and their neighboring genes (NG). A total of 66, 446, and 280 KICH, KIRC, and KIRP samples were analyzed, respectively, and mRNA expression z-scores were obtained relative to all samples (log RNA Seq V2 RSEM) using a z-score threshold of $\pm$2.0 [19, 20].

STRING (https://string-db.org/cgi/input.pl) is an online professional database for analyzing protein-protein interactions (PPI). We used the STRING database to build a low-confidence level (0.150) PPI network and screening criteria for species defined as humans. Finally, K-means cluster analysis was performed on SHMT1, SHMT2, and their NG [19, 20].

GeneMANIA (http://www.genemania.org) is a free professional tool for analyzing gene interactions. We used the GeneMANIA database to explore the interactions between SHMT1, SHMT2, and their altered NG [19, 20].

Metascape (https://metascape.org) is a professional-free tool for analyzing gene ontology (GO) functions and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment. We used the Metascape database to analyze the GO function and KEGG pathway enrichment of SHMT1, SHMT2, and their altered NG in patients with KICH, KIRC, and KIRP [19, 20].

TRRUST (https://www.grnpedia.org/trrust/) is an online professional database that analyzes transcriptional gene regulators. We used the TRRUST database to analyze the transcriptional regulators of SHMT1, SHMT2, and their altered NG in patients with RCC.

LinkedOmics (http://www.linkedomics.org/) is a free online platform for analyzing microRNA (miRNA) target enrichment and differentially expressed genes associated with tumor genes. We used the LinkedOmics database to analyze the miRNA target enrichment and differentially expressed genes associated with SHMT1 and SHMT2 [19, 20].

TIMER (https://cistrome.shinyapps.io/timer/) is a specialized database for systematically analyzing tumor genes associated with infiltrating immune cells. We used the TIMER database to analyze the correlation between SHMT1 and SHMT2 expression and immune cell infiltration levels [19, 20].

We compared the SHMT1 and SHMT2 expression levels based on sample type, individual cancer stage, sex, and patient age among patients with RCC. Results showed that SHMT1 transcript levels were significantly downregulated in patients with KICH, KIRC, and KIRP, based on sample type, individual cancer stage, sex, and patient age (p 0.05; Fig. 1). The transcript level of SHMT1 in stage 4 cancer was significantly lower than in stage 1 among patients with KIRC and KIRP (p 0.05; Fig. 1F,J). In patients with KIRC, SHMT1 transcript levels were significantly upregulated in females compared to males (p 0.001; Fig. 1G). However, SHMT1 transcript levels in patients with KIRP were significantly downregulated in females compared to that in males (p 0.001; Fig. 1K). In addition, results indicated that SHMT2 transcript levels were significantly upregulated in patients with KICH, KIRC, and KIRP, based on sample type, individual cancer stage, sex, and patient age (p 0.05; Fig. 2). In patients with KIRP, SHMT2 transcript levels were significantly upregulated in females compared to that in males (p 0.05; Fig. 2K).

Moreover, we evaluated the correlation between the differential expression levels of SHMT1 and SHMT2 and the pathological stage of RCC. A significant correlation was found between the expression of SHMT1 (p = 0.0142; Fig. 3B) and pathological stage of KIRC. The expression of SHMT1 (p = 0.00326; Fig. 3C), SHMT2 (p = 7.32 $\times$ 10${}^{-8}$; Fig. 3F), and the pathological stage of KIRP also had a significant correlation. Moreover, results showed a significant correlation between the expression of SHMT2 (p = 0.000135; Fig. 3D) and pathological stage of KICH.

Finally, the GEPIA database was used to assess the prognostic value of SHMT1 and SHMT2 expression in patients with RCC. Results indicated that patients with KIRC, and those with KIRP, with high SHMT1 expression had a longer overall and disease-free survival rate than those with low SHMT1 expression (p 0.05; Fig. 4C–F). In contrast, patients with KICH with low SHMT2 expression had longer overall and disease-free survival times than those with high SHMT2 expression (p 0.05; Fig. 5A,B). However, patients with KIRC with high SHMT2 expression had longer overall survival times than those with low SHMT2 expression (p 0.05; Fig. 5C).

The genetic alterations in SHMT1 and SHMT2 were assessed in 66, 446, and 280 patients with KICH, KIRC, and KIRP, respectively, using the TCGA database. Results indicated that SHMT1 expression was altered by 9% in patients with KICH, and the types of genetic alterations mainly included high- and low RNA levels (Fig. 6A). In addition, SHMT1 and SHMT2 expression levels were altered by 4% in patients with KIRC. The types of genetic alterations mainly included missense and truncating mutations as well as high- and low RNA levels (Fig. 6E and Fig. 7E). Meanwhile, among patients with KIRP the SHMT1 expression was altered by 6%, and the types of genetic alterations mainly included missense mutations, deep deletions, and low RNA levels (Fig. 6J). SHMT2 expression was altered by 3% in patients with KICH, and the type of genetic alteration mainly included high RNA levels (Fig. 7A). Meanwhile, the expression of SHMT2 was altered by 7% among patients with KIRP, and the types of genetic alterations mainly included missense mutations, amplification, and high and low RNA levels (Fig. 7J).

Subsequently, we assessed the promoter methylation levels of SHMT1 and SHMT2 in patients with RCC using the UALCAN database. Results indicate that SHMT1 promoter methylation levels were significantly upregulated in patients with KIRP, based on sample type, individual cancer stage (stages 1 and 3), and patient age (41–60 years and 61–80 years) (p 0.05; Fig. 6K,L,N). SHMT1 promoter methylation levels were significantly upregulated in male patients with KIRP (p 0.001; Fig. 6M). However, SHMT2 promoter methylation levels were significantly downregulated in patients with KIRC and KIRP based on sample type, individual cancer stage, sex, and patient age (p 0.05; Fig. 7F–N).

We evaluated alterations in the NG of SHMT1 and SHMT2 in RCC using the cBioPortal database. Results showed gene alteration frequencies $\ge$16.67%, $\ge$16.67%, and $\ge$11.76% for the 50 most frequently altered NG of SHMT1 in patients with KICH, KIRC, and KIRP, respectively (Tables 1,2,3). Furthermore, in patients with KICH, KIRC, and KIRP, the 50 most frequently altered NG of SHMT2 showed 100.00%, $\ge$15.79%, and $\ge$10.00%, respectively (Tables 4,5,6). In patients with KICH, KIRC, and KIRP, the most frequently altered NG of SHMT1 were AP5Z1 (33.33%), MUC16 (20.00%), and STAG2 (23.53%), respectively (Tables 1,2,3). However, AARD (100.00%), PAPPA2 (31.58%), and CDKN2A (55.00%) were the most frequently altered NG of SHMT2 among patients with KICH, KIRC, and KIRP, respectively (Tables 4,5,6).

We further evaluated the potential interactions between SHMT1, SHMT2, and their NG. Results indicated that 42 nodes, 146 edges, and 3 clusters were obtained in the constructed PPI network of SHMT1 and its NG among patients with KICH (Fig. 8A). SHMT1 and its NG were linked to a complex interaction network (71 genes and 267 edges) through physical interactions, co-expression, pathways, shared protein domains, and predictions (Fig. 8B). Furthermore, we found that 40 nodes, 106 edges, and 3 clusters were obtained in the constructed PPI network of SHMT1 and its NG among patients with KIRC (Fig. 8C). SHMT1 and its NG were linked to a complex interaction network (70 genes and 389 edges) through co-expression, pathway, genetic interactions, co-localization, and shared protein domains (Fig. 8D). Furthermore, 44 nodes, 198 edges, and 3 clusters were obtained in the constructed PPI network of SHMT1 and its NG among patients with KIRP (Fig. 8E). SHMT1 and its NG were linked to a complex interaction network (71 genes and 361 edges) through co-expression, physical interactions, shared protein domains, co-localization, and prediction (Fig. 8F). Conversely, 34 nodes, 206 edges, and 3 clusters were obtained in the constructed PPI network of SHMT2 and its NG in KICH (Fig. 9A). SHMT2 and its NG were linked to a complex interaction network (56 genes and 154 edges) through co-expression, physical interactions, and shared protein domains (Fig. 9B). In addition, 38 nodes, 146 edges, and 3 clusters were obtained in the constructed PPI network of SHMT2 and its NG in KIRC (Fig. 9C). SHMT2 and its NG were linked to a complex interaction network (71 genes and 151 edges) through co-expression, co-localization, physical interactions, shared protein domains, and prediction (Fig. 9D). A total of 43 nodes, 148 edges, and 3 clusters were obtained in the constructed PPI network of SHMT2 and its NG among patients with KIRP (Fig. 9E). SHMT2 and its NG were linked to a complex interaction network (71 genes and 219 edges) through co-expression, physical interactions, co-localization, shared protein domains, and prediction (Fig. 9F).

We further performed GO and KEGG pathway enrichment analyses of SHMT1, SHMT2, and their top 50 altered NG in patients with RCC using the Metascape database. Results indicated that biological processes related to SHMT1 and its NG among patients with KICH mainly included cardiac muscle cell development, pyrimidine-containing compound metabolic processes, and multicellular organismal signaling (Fig. 10A). Their cellular components included the AP-type membrane coat adaptor complex, anchored membrane component, and synaptic membrane (Fig. 10B). Clathrin adaptor-, metalloendopeptidase-, and GTPase regulator activities were involved to their molecular functions (Fig. 10C). KEGG pathway analysis revealed that lysosome activity, human immunodeficiency virus 1 infection, and endocytosis were enriched (Fig. 10D). However, our results showed that synapse assembly, canonical Wnt signaling pathway regulation, and myeloid cell differentiation regulation were the biological processes related to SHMT1 and its NG among patients with KIRC (Fig. 10E). Neuronal cell bodies, sites of polarized growth, and extrinsic membrane components were related to their cellular components (Fig. 10F). Lipid binding and active transmembrane transporter- and lipid transporter activity were related to their main molecular functions (Fig. 10G). KEGG pathway analysis revealed that SHMT1 and its NG were associated with the coronavirus disease 2019 and pathways of neurodegeneration (Fig. 10H). Furthermore, biological processes related to SHMT1 and its NG among patients with KIRP mainly included a positive regulation of peptidyl-serine phosphorylation of the STAT protein, nucleocytoplasmic transport, and cellular response to hypoxia (Fig. 10I). The cellular components were the external side of the plasma membrane, axon, and chromosomal region (Fig. 10J). Type I interferon receptor binding, integrin binding, and protein heterodimerization were associated with their main molecular functions (Fig. 10K). RIG-I-like receptor signaling pathway, cell cycle, and actin cytoskeleton regulation were enriched according to the KEGG pathway analysis (Fig. 10L).

The biological processes related to SHMT2 and its NG among patients with KICH mainly included plasma membrane organization, cell-cell junction organization, and positive regulation of organelle organization (Fig. 11A). Their cellular components included the ubiquitin ligase complex, chromosomal region, and mitochondrial membrane (Fig. 11B). Cell adhesion molecule and phospholipid bindings were involved to their molecular functions (Fig. 11C). However, we found that calcium ion transmembrane import into the cytosol, positive regulation of lipid localization, and regulation of acute inflammatory response were biological processes associated to SHMT2 and its NG among patients with KIRC (Fig. 11D). The apical plasma membrane, perinuclear region of the cytoplasm, and sarcolemma were related to their cellular components (Fig. 11E). Protein domain-specific binding, enzyme inhibitor activity, and endopeptidase activity were associated to their main molecular functions (Fig. 11F). KEGG pathway analysis revealed that SHMT2 and its NG were associated with cAMP and calcium signaling pathways (Fig. 11G). Furthermore, biological processes related to SHMT2 and its NG among patients with KIRP mainly included the regulation of the transforming growth factor-beta receptor signaling pathway, defense response to Gram-negative bacteria, and positive binding regulation (Fig. 11H). Their cellular components included the leading-edge membrane, extrinsic membrane component, and spindle (Fig. 11I). Hormone activity, integrin, and protein kinase regulatory activity was related to their main molecular functions (Fig. 11J). MiRNAs involved in cancer and neuroactive ligand-receptor interactions were enriched according to KEGG pathway analysis (Fig. 11K).

We used the TRRUST database to analyze the key regulatory factors of SHMT1 and SHMT2 in patients with RCC. We found that MYC was the critical transcription factor of SHMT1 and its NG among patients with KICH (p 0.05) (Table 7). PRODH and SHMT1 were the primary regulatory genes of MYC. In addition, STAT1 was the critical transcription factor of SHMT1 and its NG among patients with KIRC (p 0.05) (Table 7). STAT1 regulated the functions of STAT2 and TLR3. Furthermore, PPARG, NR3C1, HDAC1, STAT1, and MYC were the critical transcription factors of SHMT1 and its NG among patients with KIRP (p 0.05) (Table 7). SLC2A4, TP53, and TXNIP were the primary regulatory genes of PPARG. NR3C1 regulated the functions of ATP1B1 and TP53. Moreover, TP53 and TXNIP were the primary regulatory genes of HDAC1. STAT1 regulated the functions of FGFR3 and TP53. SHMT1 and TP53 were the central regulatory genes of MYC. However, we found that AR was the key transcription factor of SHMT2 and its NG among patients with KICH (p 0.01) (Table 8). ATAD2 and NDRG1 were the primary regulatory genes of AR. In addition, SREBF2, MYCN, and SP3 were the key transcription factors of SHMT2 and its NG among patients with KIRC (p 0.05) (Table 8). SREBF2 regulated the functions of ABCA1 and LRP1. DDB1 and EFNB3were the central regulatory genes of MYCN. SP3 regulated the functions of ABCA1 and ALOX12. Furthermore, FOXM1, DNMT1, REST, and E2F1 were the key transcription factors of SHMT2 and its NG in KIRP (p 0.05) (Table 8). CDC25B and CDKN2A were the central regulatory genes of FOXM1. DNMT1 regulated the functions of BMP2 and CDKN2A. Moreover, AVP and CHGB were the primary regulatory genes of REST. E2F1 regulated the functions of BMI1 and CDKN2A.

miRNA targets of SHMT1 and SHMT2 were analyzed using the LinkedOmics database. We found that (GCACTTT) miR-17-5P, miR-20A, miR-106A, miR-106B, miR-20B, and miR-519D; (TTTGCAC) miR-19A and miR-19B; and (AAGCCAT) miR-135A and miR-135B were the top three miRNA targets of SHMT1 among patients with KICH (p 0.001; Table 9). In addition, the top three miRNA targets of SHMT1 among patients with KIRC were (AAGTCCA) miR-422B and miR-422A, (GCTTGAA) miR-498, and (GGGGCCC) miR-296 (p 0.001; Table 9). Furthermore, (CAGGTCC) miR-492, (ACTGAAA) miR-30A-3P and miR-30E-3P, and (CTTTGTA) miR-524 were the top three miRNA targets of SHMT1 in KIRP (p 0.001; Table 9). However, (AAGCAAT) miR-137, (GGGCATT) miR-365, and (AACATTC) miR-409-3P were the top three miRNA targets of SHMT2 in KICH (p 0.001; Table 10). The top three miRNA targets of SHMT2 in KIRC were (ACTGAAA) miR-30A-3P and miR-30E-3P, (CATGTAA) miR-496, and (AGCATTA) miR-155 (p 0.001; Table 10). Moreover, (ATGTACA) miR-493, (GGGCATT) miR-365, and (CTTTGCA) miR-527 were the top three miRNA targets of SHMT2 in KIRP (p 0.001; Table 10).

The mRNA sequencing data from patients with KICH (n = 66), KIRC (n = 533), and KIRP (n = 290), using the LinkedOmics database, were analyzed. Results indicated that the expression of 19,216 genes was correlated with SHMT1 expression in KICH. The 10,757 and 8459 genes were positively and negatively correlated with SHMT1 expression, respectively (Fig. 12A). We screened 50 genes whose expression levels were significantly correlated with SHMT1 expression in KICH (p 0.05; Fig. 12B,C). PRODH (Pearson correlation coefficient (PCO) = 0.7353, p = 2.05 $\times$ 10${}^{-12}$; Fig. 13A), SH3KBP1 (PCO = 0.7242, p = 6.321 $\times$ 10${}^{-12}$; Fig. 13B), and EML6 (PCO = 0.7169, p = 1.288 $\times$ 10${}^{-11}$; Fig. 13C) were the top three genes whose expression levels were positively correlated with SHMT1 expression. In addition, the expression levels of 20,158 genes were correlated with SHMT1 expression in KIRC. A total of 9036 and 11,122 genes were positively and negatively correlated with SHMT1 expression, respectively (Fig. 12D). We screened 50 genes whose expression levels were significantly correlated with SHMT1 expression in KIRC (p 0.05; Fig. 12E,F). Among the screened genes, ACSM2B (PCO = 0.7968, p = 2.754 $\times$ 10${}^{-118}$; Fig. 13D), ACSM2A (PCO = 0.792, p = 7.393 $\times$ 10${}^{-116}$; Fig. 13E), and PDZK1 (PCO = 0.7817, p = 5.842 $\times$ 10${}^{-111}$; Fig. 13F) were the top three genes whose expression levels were positively correlated with SHMT1 expression. Furthermore, the expression levels of 20,023 genes were correlated with SHMT1 expression in KIRP, of which 8091 and 11,932 genes were positively and negatively correlated with SHMT1 expression, respectively (Fig. 12G). In addition, we screened 50 genes whose expression levels were significantly positively and negatively correlated with SHMT1 expression in KIRP (p 0.05; Fig. 12H,I). Among the screened genes, SMTNL2 (PCO = 0.8186, p = 2.592 $\times$ 10${}^{-71}$; Fig. 13G), ACSM2A (PCO = 0.7928, p = 6.781 $\times$ 10${}^{-64}$; Fig. 13H), and ACSM5 (PCO = 0.7911, p = 1.956 $\times$ 10${}^{-63}$; Fig. 13I) were the top three genes whose expression levels were positively correlated with SHMT1 expression.

Our results showed that the expression levels of 19,216 genes were correlated with SHMT2 expression in KICH. Moreover, 10,436 and 8780 genes were positively and negatively correlated with SHMT2 expression, respectively (Fig. 14A). We screened 50 genes whose expression levels were significantly positively and negatively correlated with SHMT2 expression in KICH (p 0.05; Fig. 14B,C). PSAT1 (PCO = 0.7665, p = 6.356 $\times$ 10${}^{-14}$; Fig. 15A), ASNS (PCO = 0.7594, p = 1.458 $\times$ 10${}^{-13}$; Fig. 15B), and YARS (PCO = 0.7573, p = 1.871 $\times$ 10${}^{-13}$; Fig. 15C) were the top three genes whose expression levels were positively correlated with SHMT2 expression. However, the expression levels of 20,158 genes were correlated with SHMT2 expression among patients with KIRC; of which 10,443 and 9715 genes were positively and negatively correlated with SHMT2 expression, respectively (Fig. 14D). Additionally, we screened 50 genes whose expression levels were significantly positively and negatively correlated with SHMT2 expression among patients with KIRC (p 0.05; Fig. 14E,F). NDUFA4L2 (PCO = 0.6471, p = 1.481 $\times$ 10${}^{-64}$; Fig. 15D), RALGPS1 (PCO = –0.5957, p = 1.684 $\times$ 10${}^{-52}$; Fig. 15E), and NOL3 (PCO = 0.5789, p = 5.235 $\times$ 10${}^{-49}$; Fig. 15F) were the top three genes whose expression levels were positively or negatively correlated with SHMT2 expression. Furthermore, the expression levels of 20,023 genes were correlated with SHMT2 expression among patients with KIRP; of which 9759 and 10,264 genes were positively and negatively correlated with SHMT2 expression, respectively (Fig. 14G). We also screened 50 genes whose expression levels were significantly positively and negatively correlated with SHMT2 expression among patients with KIRP (p 0.05; Fig. 14H,I). CDK4 (PCO = 0.6734, p = 1.154 $\times$ 10${}^{-39}$; Fig. 15G), GK5 (PCO = –0.6554, p = 5.696 $\times$ 10${}^{-37}$; Fig. 15H), and C12orf52 (PCO = 0.6466, p = 9.859 $\times$ 10${}^{-36}$; Fig. 15I) were the top three genes whose expression levels were positively or negatively correlated with SHMT2 expression.

The TIMER database was used to evaluate the relationship of immune cell infiltration levels with SHMT1 and SHMT2 expression among patients with RCC. Results showed that SHMT1 expression was positively correlated with B-cell (Cor = 0.308, p = 1.25 $\times$ 10${}^{-2}$) and dendritic cell infiltration levels (Cor = 0.421, p = 4.82 $\times$ 10${}^{-4}$; Fig. 16A) in KICH. The SHMT1 expression was positively correlated with B-cell infiltration level in KIRC (Cor = 0.161, p = 5.42 $\times$ 10${}^{-4}$; Fig. 16B). In addition, SHMT1 expression was positively correlated with macrophage infiltration levels in KIRP (Cor = 0.257, p = 3.97 $\times$ 10${}^{-5}$; Fig. 16C). However, B-cell (Cor = –0.227, p = 2.56 $\times$ 10${}^{-4}$; Fig. 16C) and CD8+ T-cell (Cor = –0.226, p = 2.58 $\times$ 10${}^{-4}$; Fig. 16C) infiltration levels were negatively correlated with SHMT1 expression in KIRP. The SHMT2 expression was positively correlated with CD4+ T-cell (Cor = 0.363, p = 2.98 $\times$ 10${}^{-3}$), neutrophil (Cor = 0.397, p = 1.07 $\times$ 10${}^{-3}$), and dendritic cell (Cor = 0.312, p = 1.14 $\times$ 10${}^{-2}$) infiltration levels in KICH (Fig. 16D). SHMT2 expression was positively correlated with B-cell (Cor = 0.229, p = 2.19 $\times$ 10${}^{-4}$; Fig. 16F) and CD8+ T-cell (Cor = 0.319, p = 1.69 $\times$ 10${}^{-7}$) infiltration levels in KIRP (Fig. 16F).