Information on genes and clinical characteristics for different tumors together with control tissues were acquired from the GTEx (https://gtexportal.org/) and TCGA (https://portal.gdc.cancer.gov/) databases. Information on cell lines was obtained from the CCLE database (https://portals.broadinstitute.org/). TIMER (https://cistrome.shinyapps.io/timer/) was used for information on immune infiltration scores. Data on the following tumors was acquired: ACC (Adrenocortical carcinoma); BLCA (Bladder Urothelial Carcinoma); BRCA (Breast invasive carcinoma); CESC (Cervical squamous cell carcinoma and endocervical adenocarcinoma); CHOL (Cholangiocarcinoma); COAD (Colon adenocarcinoma); COAD (Colon adenocarcinoma); READ (Rectum adenocarcinoma esophageal carcinoma); DLBC (Lymphoid Neoplasm Diffuse Large B-cell Lymphoma); ESCA (Esophageal carcinoma); GBM (Glioblastoma multiforme); HNSC (Head and Neck squamous cell carcinoma); KICH (Kidney Chromophobe); KIRC (Kidney renal clear cell carcinoma); KIRP (Kidney renal papillary cell carcinoma); LAML(Acute Myeloid Leukemia); LGG (Brain Lower grade Glioma); LIHC (Liver hepatocellular carcinoma); LUAD (Lung adenocarcinoma); LUSC (Lung squamous cell carcinoma); MESO (Mesothelioma); OV (Ovarian serous cystadenocarcinoma); PAAD (Pancreatic adenocarcinoma); PCPG (Pheochromocytoma and paraganglioma); PRAD (Prostate adenocarcinoma); READ (Rectum adenocarcinoma); SARC (Sarcoma); SKCM (Skin Cutaneous Melanoma); STAD (Stomach adenocarcinoma); STES (Stomach and Esophageal carcinoma); TGCT (Testicular germ cell tumors); THCA(Thyroid carcinoma); THYM (Thymoma); UCEC (Uterine Corpus Endometrial Carcinoma); UVM (Uveal Melanoma).

TRPM8 levels in various tumors and controls were examined using “edgeR” in Bioconductor and the Kruskal Wallis test. The “ggplot” package in R was used to draw violin plots. For integrated analysis of TRPM8 differential expression in 33 cancers, the gene expression data of GTEx and TCGA were downloaded from the University of California Santa Cruz (UCSC) Xena database (https://xenabrowser.net/datapages/). Both profiles from raw RNA-Seq data were recomputed using the UCSC Xena project and converted to a format as log2(TPM+1).

Cancer genomic information was obtained from cBioPortal (http://www.cbioportal.org/) (14), including copy-number variations, mRNA levels, deep deletions, and missense mutations with unknown implications. UALCAN (http://ualcan.path.uab.edu) was utilized for examining DNA methylation.

Univariate regression was performed to identify the relationships between TRPM8 level and patient outcomes and Kaplan-Meier curves were utilized to relate outcome to TRPM8 level. A bipartite technique was used to divide TRPM8 levels in tissue (tumor and control) into “high” and “low” groups for evaluation by univariate Cox regression. The “forestplot” package in R was used for visualization and PrognoScan (15) was used for meta-analysis of TRPM8 levels and the overall survival (OS) and disease-free survival (DFS) of patients.

The TISIDB database (http://cis.hku.hk/TISIDB/in-dex.php) (16) was used to investigate the link between TRPM8 and immune-associated genes and subtypes.

The presence and degree of immune infiltration, measured by the immune and stromal scores, are used to predict sentinel lymph node status and outcomes. The correlations between these scores and TRPM8 were determined, with significance levels of p < 0.05 and R > 0.20.

Scanneo was used to detect neoantigens and the products of mutated genes. This predicts the binding affinities of proteins using short (8-11 residues) epitopes, with low scoring proteins classified as neoantigens, which are then ranked. The association between TRPM8 levels and neoantigen numbers per tumor was determined. TRPM8 levels were also associated with over 40 immune checkpoint-associated genes, with significance set at p < 0.05 and R > 0.20.

The TMB is an indication of mutation numbers in a cancer cell. The MSI represents a predisposition to mutation caused by faulty DNA mismatch repair (MMR). Correlations between TRPM8 levels and TMB and MSI were assessed by Spearman’s rank correlation analysis.

Impaired MMR leads to increased numbers of somatic mutations. TCGA TRPM8 levels were correlated with those of five MMR genes, namely, MLH1, MSH2, MSH6, PMS2, and EPCAM. DNA methylation modulates gene expression and is conducted by methyltransferases. TRPM8 levels were correlated with those of four methyltransferases with significance assessed as above. The R package “ggplot” was utilized to visualize the data.

PPI networks of TRPM8 and its interactors were created with the STRING (http://string.embl.de/) (17, 18) and GeneMANIA (http://www.genemania.org) (19) databases.

Gene set enrichment analysis (GSEA) was used to investigate TRPM8 in relation to “biological process”, “molecular function”, or “cellular component” (20) using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and MsigDB (21) databases, using the Hallmark gene set. |NES| > 1, p < 0.05, and FDR < 0.25 were set as thresholds for GSEA.

The influence of TRPM8 on drug susceptibility was assessed using the GSCA (Gene Set Cancer Analysis) (http://bioinfo.life.hust.edu.cn/GSC-A/#/) (22) and Genomics of Drug Sensitivity in Cancer (GDSC) databases.

We performed IHC staining on paraffinized tumor tissue microarray sections according to a previously published protocol (23). Human KIRC tissue microarray (5-μm thick; cat. no. SP013-U1, Expectlab, Qingdao, Shandong, China) and human LIHC tissue microarray (5-μm thick; cat. no. SP014-T2, Expectlab) sections were deparaffinized, rehydrated, and subjected to antigen retrieval, blocking (3% H₂O₂, 5% goat serum), and incubation with primary rabbit anti-human TRPM8 antibody (1:200; cat. no. PB0882, Boster Biological Technology Co., Ltd., Wuhan, China) overnight at 4°C. Visualization of the antigen-antibody complexes was performed with an AEC kit (cat. no. ZLI-9036, Beijing Zhongshan Jinqiao Biotechnology Co., Ltd.), based on biotin-streptavidin-peroxidase complexation. Integrated IHC staining intensity was used to evaluate the TRPM8 level in each human tumor tissue point. The images were analyzed using Image Tool II software (University of Texas Health Science Center, San Antonio, TX, USA). The staining intensity of TRPM8 was evaluated using a grayscale ranging from 0 to 255.

The tissue microarrays used in this study are commercially available, which waived the need for ethical approval from the Ethics Committee of Jiangnan University.

Bioinformatic data were analyzed using R v 4.0.3. Statistical analysis of IHC results was performed using SPSS (statistical package for the social sciences) software (SPSS Inc., Chicago, IL, Version 17.0). Data were expressed as means ± SEM and analyzed using Student’s t-test for independent samples. P < 0.05 was considered significant.

The TRPM8 mRNA levels derived from the GTEx datasets for 31 cancer types are shown in Figure 1A , while Figure 1B illustrates TRPM8 levels in CCLE cell lines. The differences between tumor and adjoining control tissues (from TCGA) are seen in Figure 1C . The integration of the GTEx and TCGA data in 33 cancers ( Figure 1D ) indicates an elevation of TRPM8 levels in most solid tumors.

Significant differences in TRPM8 levels in relation to the clinical stage were seen in PRAD, KIRC, THCA, LIHC, LGG, ESCA, GBMLGG, and KICH ( Figure 2A ). TRPM8 levels also differed significantly according to molecular subtype in BRCA, KIRP, LGG, LIHC, OV, and PRAD ( Figure 2B ). ( Supplementary Figure S1A, S1B ).

Both genetic and epigenetic factors influence tumor development (24). The frequency of mutation in TRPM8 was investigated using cBioPortal on 10967 samples from 32 studies. It was found that the highest mutation numbers were seen in UCEC, melanoma, and STAD, with TRPM8 representing approximately 7% of mutations ( Figure 3A ). Two hundred and eighty-seven mutation sites were observed, including 239 missense, 23 truncating, 18 splice, and 7 fusion variants ( Figure 3B ) over the 1100-residue length. Deletions were seen in over a third of tumors ( Figure 3C ), resulting in reduced mRNA expression. There was an association between TRPM8 copy number and mRNA level ( Figure 3D ). Using UALCAN, lower DNA methylation of TRPM8 was observed in BRCA, ESCA, COAD, HNSC, KIRC, LIHC, LUAD, LUSC, READ, SARC, TGCT, THCA, and UCEC tumors relative to controls. This indicates that elevated TRPM8 mRNA is likely to be caused by both genetic and epigenetic changes ( Supplementary Figure S2 ).

Univariate analysis was performed to investigate the link between TRPM8 and OS in 33 tumors from TCGA. TRPM8 was significantly linked with OS in ACC (HR = 3.88, P = 0.042), HNSC (HR = 1.34, P = 0.002), KIRC (HR = 1.06, P < 0.001), LAML (HR = 11.02, P =0.001), LGG (HR = 1.05, P <0.001), and LIHC (HR = 1.23, P < 0.001) ( Figure 4A ). These findings indicate that TRPM8 is linked with poor outcomes, especially in LAML. The Kaplan-Meier curves for cancer types with significant TRPM8 association are shown in Figures 4B–G . Figures 4H, I shows the disease-free survival (DSS) and progression-free interval (PFI) which were determined to avoid bias from non-cancer mortality.

The prognostic worth of TRPM8 was then examined using PrognoScan. This showed significant value in five cancers, namely, colorectal, esophageal, breast, lung, and prostate cancers ( Figure 5 ). TRPM8 adversely affected three tumor types, namely, breast ( Figure 5A , RFS: Cox P = 0.022), esophageal ( Figure 5C , OS: Cox P = 0.001), and lung ( Figures 5D, E , OS: Cox P = 0.003; RFS: Cox P < 0.001) cancers but appeared to have a protective function in colorectal ( Figure 5B , OS: Cox P <0.001) and prostate cancers ( Figure 5F , OS: Cox P = 0.006). Supplementary Table 1 contains detailed information on these cohorts.

Thus, TRPM8 overexpression was significantly linked with reduced survival in several tumors, suggesting its possible usefulness as a prognostic indicator.

TRPM8 levels were associated with the degree of immune infiltration in various tumors, including BLCA, BRCA, and LGG. As seen in Figure 6A , there were significant correlations with CD4+ T cells (R= 0.129, P < 0.01), CD8+ T cells (R= 0.123, P < 0.05), neutrophils (R = 0.188, P <0.001), macrophages (R = 0.2, P < 0.001), and dendritic cells (R = 0.233, P < 0.001) in BLCA. As seen in Figures 6B, C , TRPM8 was significantly associated with all six tumor-infiltrating cells, including B cells (R = 0.191, P < 0.001), CD4+ T cells (R = 0.316, P < 0.001), CD8+ T cells (R = 0.224, P < 0.001), neutrophils (R = 0.349, P < 0.001), macrophages (R = 0.201, P < 0.001), and dendritic cells (R = 0.343, P < 0.001) in BRCA (B), and with B cells (R = 0.249, P < 0.001), CD4+ T cells (R = 0.194, P < 0.001), CD8+ T cells (R = 0.364, P < 0.001), neutrophils (R = 0.334, P < 0.001), macrophages (R = 0.298, P < 0.001), and dendritic cells (R = 0.349, P < 0.001) in LGG (C). Furthermore, analysis of the immune and stromal scores using the R package “estimate” showed that the highest correlations between TRPM8 and immune score were LGG (R = 0.405, P < 0.001), BRCA (R = 0.209, P < 0.001), and THCA (R = 0.230, P < 0.001) ( Figure 6D ). The highest correlations with stromal score were seen in BRCA (R = 0.209, P < 0.001), LGG (R = 0.405, P < 0.001), and THCA (R =0.230, P < 0.001), and estimated immune scores were BRCA (R = 0.209, P < 0.001), LGG (R = 0.405, P <0.001) and THCA (R = 0.23, P < 0.001). Evaluation of molecular subtypes showed significance in the C1 (wound healing), C2 (IFN-γ-dominant), C3 (inflammatory), C4 (lymphocyte depleted), C5 (immunologically quiet), and C6 (TGF-β dominant) subtypes in BRCA, CESC, KIRC, LGG, LIHC, LUAD, PRAD, and SARC ( Figure 6E ). Other tumor subtypes did not differ significantly ( Supplementary Figure S3 ). TRPM8 levels were highest in subtypes C6 in LUAD and BRCA, C4 in CESC and PRAD, and C3 in LGG. Different levels of expression may partly explain why TRPM8 appears to have different effects on prognosis in different cancers.

Figure 7 shows the six most significant associations between TRPM8 copy number variation and immune infiltration in tumors. TRPM8 deletion was related to higher infiltration and was linked to markedly raised numbers of infiltrates in SKCM ( Figure 7A ). In contrast, TRPM8 deletion was linked to lower infiltration in BRCA ( Figure 7B ), HNSC ( Figure 7C ), LUSC ( Figure 7D ), BLCA ( Figure 7E ), and STAD ( Figure 7F ). These results may suggest reasons for TRPM8 changes in prognostic prediction.

The levels of over 40 checkpoint genes were examined in different tumor types ( Figure 8A ). TRPM8 was found to be positively related to checkpoint gene levels in a number of tumors, including ACC, KIRC, KICH, and LGG. TRPM8 levels were also positively associated with neoantigen number in BRCA and PRAD (R = 0.10, P =0.0093; R = 0.14, P = 0.024, respectively) ( Figure 8B ).

The TMB represents the number of mutations in a cancer cell. TRPM8 levels were associated with TMB in THYM, BRCA, LGG, SARC, and SKCM ( Figure 9A ). TRMP8 was also associated with MSI in SARC, TGCT, BLCA, BRCA, and MESO, and negatively correlated with LUSC and PRAD ( Figure 9B ).

TRPM8 levels were positively correlated with most MMR genes in BLCA, BRCA, HNSC, KIRP, PRAD, and UVM ( Figure 10A ), as well as with the levels of methyltransferases in UVM, BLCA, BRCA, HNSC, KICH, and KIRP ( Figure 10B ), indicating that TRPM8 can modulate tumorigenesis through epigenetic means.

The PPI network created by STRING showed associations between TRPM8 and HSP90, Ubiquitinase USP10, and inositol triphosphate receptors (ITPR1, ITPR2, ITPR3), amongst others ( Figure 11A ). The network constructed by GeneMANIA showed interaction with TCAF2 (TRPM8 channel associated factor 2) ( Figure 11B ), which is involved in the regulation of TRPM8 activity in the membrane. Interactions with TRPM4 and CHAMP1 were also predicted.

The high and low TRPM8 expression groups were investigated for KEGG and Hallmark pathway enrichment (listed in Supplementary Tables 2 and 3 ). Enrichments were seen ( Figure 12 ) in the “primary immunodeficiency pathway” and “KRAS signaling pathway”. KEGG pathways included “complement and coagulation cascades”, “pentose and glucuronate interconversions”, “steroid hormone biosynthesis”, and “cytokine-cytokine receptor interaction”. “IL6/JAK/STAT3 signaling”, “Interferon (IFN) cell signaling pathway”, “inflammatory response”, “TNFalpha/NF-κB pathway”, “angiogenesis”, “epithelial-mesenchymal transition”, “hypoxia” and others were identified by Hallmark. Thus, TRPM8 appears to be involved in pathways associated with tumor physiology and immunity.

TRPM8 was observed to be linked to the response to Afatinib, Dabrafenib, and PLX4720, shown by GDSC ( Figure 13 ), suggesting the application of these findings in treatment.

IHC was used to verify TRPM8 levels in KIRC and LIHC tissues. 90 grade I−II and 42 grade III−IV KIRC tissues were examined, finding greater TRPM8 expression in grade III−IV material (p < 0.05) ( Figures 14A, B ). In LIHC samples (88 grade II, 53 grade III, and 3 grade IV), however, it was observed that TRPM8 levels were reduced in the higher grades, with lower TRPM8 expression seen in both grades III and IV compared with grade II (p < 0.001 vs. grade II for grade III, and p < 0.05 vs. grade II for grade IV) ( Figures 14C, D ).