ERCC6L gene expression analysis
We began our study by examining the mRNA expression levels of ERCC6L in commonly diagnosed cancers. Compared with its expression levels in normal tissue compartments, ERCC6L was expressed at significantly higher levels in tumor samples from patients with cancers, including adrenocortical carcinoma (ACC), invasive breast carcinoma (BRCA), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), brain lower grade glioma (LGG), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), pancreatic adenocarcinoma (PAAD), bladder urothelial carcinoma (BLCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), colon adenocarcinoma (COAD), cholangiocarcinoma (CHOL), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), ovarian cancer (OV), prostate adenocarcinoma (PRAD), testicular germ cell tumors (TGCT), thyroid carcinoma (THCA), uterine carcinosarcoma (UCS), lung squamous cell carcinoma (LUSC), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), thymoma (THYM), skin cutaneous melanoma (SKCM) and uterine corpus endometrial carcinoma (UCEC) (Figs. 1A–I, S1A–H, J–N, and S2A–F). Notably, we also found that ERCC6L levels were specifically upregulated in tumor samples than paired normal tissue samples from the same individual in patients with LUSC, READ, STAD, UCEC, BRCA, KICH, KIRC, KIRP, LIHC, and LUAD (Figs. S2G–J and S3A–F). However, decreased ERCC6L levels were observed in patients with acute myeloid leukemia (LAML) (Fig. S1I). Additionally, we examined the ERCC6L protein levels in multiple common cancers. ERCC6L protein was elevated in eight of nine cancers that we investigated (Fig. S3). These results suggested that the expression of ERCC6L is elevated at both mRNA and protein levels in most of the cancers examined.
Prognostic value of ERCC6L for patient survival
We investigated whether ERCC6L could be used as a biomarker for the prognosis of cancer patients, given the pattern of ERCC6L upregulation in most cancers. As expected, high ERCC6L levels significantly correlated with unfavorable outcomes in patients with ACC, BRCA, KICH, KIRC, KIRP, LGG, LIHC, LUAD, PAAD, UCEC, SKCM, MESO, Sarcoma (SARC), and Uveal Melanoma (UVM) (Figs. 2A–I and S4A–F). However, the ERCC6L expression level was not correlated with the overall survival (OS) of patients with BLCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, LAML, OV, Pheochromocytoma/paraganglioma (PCPG), PRAD, TGCT, THCA, or UCS (Figs. S5A–O). Moreover, ERCC6L levels were negatively associated with higher survival rates in some cancer patients, such as LUSC, READ, STAD, and THYM (Figs. S6A–D). As clinical stages, like the T and N stages, are valuable parameters for predicting a patient’s survival, we sought to determine whether the ERCC6L expression level could also function as an independent predictor of survival. Surprisingly, the ERCC6L expression level was found to be a potential predictor of survival in patients with ACC, BRCA, KICH, KIRC, KIRP, LGG, LIHC, LUAD, and PAAD in multivariate and univariate Cox analyses of various clinical factors or its expression levels (Figs. 3A–I and S7A–I). Furthermore, the estimated ROC curve illustrated that the diagnostic sensitivity and specificity of ERCC6L expression levels were effective in patients with ACC, BRCA, KICH, KIRC, KIRP, LGG, LIHC, LUAD and PAAD (area under the curve > 0.6; Figs. 4A–I). Importantly, we observed that, in addition to OS, high levels of ERCC6L expression were associated with lower disease-specific survival and progression-free intervals in nine cancer types (i.e., ACC, BRCA, KICH, KIRC, KIRP, LGG, LIHC, LUAD, and PAAD; Figs. S8A–I and S9A–I).
Expression and prognostic value of ERCC6L for cancer patient survival in specific clinical stages
Then, we determined whether ERCC6L was also aberrantly expressed in various clinical stages of cancer and correlated with the survival of patients when classified into certain stages. Interestingly, we found that ERCC6L was generally more expressed in higher pathologic stages/grades (stage III or IV vs. stage I or II and grade 3, grade 4 vs. grade 1 or grade 2, T3 or T4 vs. T1 or T2, N2 or N3 vs. N1 or N0, and M1 vs. M0) in patients with the nine cancer types mentioned above (Figs. 5A–L, S10A–H, S11A–H, and S12A–H). In addition to the entire patient cohort, ERCC6L was associated with poor outcomes in the nine types of cancer patients at certain stages (Figs. 6A–M, S13A–E, S14A–L, S15A–J, and S16A–E). These findings imply that ERCC6L is upregulated in late pathological stages/status and is related to low OS rates when cancer patients are classified into subgroups based on their clinical parameters.
Analysis of ERCC6L-regulated processes
Next, we determined the biological processes or signaling pathways associated with the expression level of ERCC6L in cancers. To this end, overlapping co-expressed genes in the nine cancer types were used to identify a panel of ERCC6L-correlated genes (a total of 197) (Figs. S17A and B). First, the hub gene analysis among the 197 genes and protein–protein interaction network analysis was carried out to unravel the physiological function of ERCC6L more accurately (Figs. S18A and B). Subsequently, GO analysis was performed based on this set of ERCC6L-correlated genes. The three most highly correlated biological processes include organelle fission, nuclear division, and chromosome segregation (Fig. 7A). This gene set was also enriched in the cellular components, chromosome region, spindle, and condensed chromosomes (Fig. 7B). Moreover, the most correlated molecular functions of ERCC6L were found to be ATPase activity and catalytic activity, acting on DNA and tubulin binding (Fig. 7C). Furthermore, KEGG analysis demonstrated that the cell cycle was the key process regulated by ERCC6L in cancer patients (Fig. 7D). Importantly, we have analyzed the correlations between ERCC6L and tumor mutation burden (TMB), microsatellite instability (MSI) and neoantigens in these nine cancers (Fig. 8). We did find statistically significant associations in some cancer types. We systematically studied the processes and pathways correlated with ERCC6L expression in nine cancers.
Analysis of ERCC6L genetic variants
We then investigated ERCC6L genetic alterations in cancer patients. First, we analyzed CNV in the nine cancer types mentioned above. Most of the tested cancer types did not significantly differ in ERCC6L expression levels with different groups of CNV; similarly, ERCC6L mRNA expression levels did not significantly correlate with ERCC6L CNVs (Figs. S17A–I and S18A–I). Unexpectedly, we observed that in KIRP, CNV was negatively associated with ERCC6L mRNA expression levels and patients with a loss of CNV (indicating a decrease in ERCC6L mRNA levels) had poor clinical outcomes (OS, disease-free interval [DFS], progression-free survival [PFS], and disease-specific survival [DSS]; Figs. S18E and S19A–D). Subsequently, we determined if the changes in ERCC6L mRNA levels were related to the methylation of its promoter. Most cancer types, excluding PAAD, exhibited a negative correlation between ERCC6L mRNA expression levels and promoter methylation (Figs. S20A–I). In particular, a significant inverse correlation was observed for BRCA (Fig. S20B). However, we also observed a significant reduction in ERCC6L promoter methylation levels in BRCA tumors than in normal control tissue (Fig. 9A) and a negative correlation between ERCC6L mRNA levels and methylation status in BRCA patients (Fig. 9B). Furthermore, we found an inverse association between ERCC6L expression levels and methylation at the three most prevalent methylation sites in the promoter region, cg05279113, cg08304428, and cg25402895 (Figs. 9C–E). Notably, unlike ERCC6L mRNA levels, its hypermethylation was associated with a favorable prognosis (OS) for BRCA patients (Figs. 9F–H). Our results demonstrated that promoter hypomethylation may significantly cause elevated ERCC6L levels in BRCA patients and was associated with lower patient survival rates.
Immune cell infiltration analysis
To determine if ERCC6L is involved in immune cell infiltration, we quantified tumor purity in patients from each of the nine cancer categories. As displayed in Figs. S21A–I, a comprehensive analysis of the correlation between ERCC6L levels and infiltration of 24 immune cell types was conducted. We then identified the infiltrated immune cell types most positively or negatively associated with ERCC6L levels (Figs. 10A–I). Th2 cell infiltration positively correlated with ERCC6L expression levels across all studied cancer types. In general, immune cell infiltration was negatively associated with the ERCC6L levels, with various cell types being enriched in certain cancer types. For example, cytotoxic T cells were enriched in ACC and KICH, natural killer cells in BRCA and LGG, plasmacytoid dendritic cells in KIRC and PAAD, macrophages in KIRP, and mast cells in LUAD (Figs. 10A–I). In addition, the differential enrichment scores between the ERCC6L-low or -high groups were significantly altered, and the trends reflected the correlations depicted in Fig. 10 (Figs. 11A–I). Moreover, the StromalScore, ImmuneScore, and ESTIMATEScore analyses were performed, and significant correlations between ERCC6L and immune cell infiltration were mainly observed in BRCA, KIRC, and LUAD (Fig. S22). In conclusion, ERCC6L expression levels are strongly associated with the infiltration of various immune cells. Moreover, the types of infiltrating immune cells vary depending on the cancer type.
Somatic mutation and drug sensitivity analysis of ERCC6L
Somatic gene mutations contribute to cancer initiation in some cases [14]; therefore, we investigated the presence of somatic ERCC6L mutations in nine cancer types. The frequency of somatic mutations, including nonsense mutations, missense mutations, frameshift insertions, and frameshift deletions, resulting in single-nucleotide variants (SNVs), insertions, or deletions in the ERCC6L gene was 1.41% (Figs. 12A and B). Most somatic mutations were missense mutations, and C > A was the most prevalent class of SNV in the nine cancer types and BRCA when analyzed independently (Figs. 12B, S23, and S24). As indicated by the heatmap in Fig. S25, the SNV frequencies were markedly higher in LUAD and BRCA (8%) than in the other cancer types. Furthermore, significant decreases in patient OS and DSS rates were observed in LUAD patients with ERCC6L SNVs (Figs. 12C and D). In addition, by integrating data from the Cancer Therapeutics Response Portal and the Genomics of Drug Sensitivity in Cancer project, we identified correlations between ERCC6L mRNA expression levels and sensitivity to cancer therapeutic drugs (Figs. S26A and B).
