Expression of COL6A in pan-cancer
Based on published data, a total of seven COL6A members, including COL6A1, COL6A2, COL6A3, COL6A4P1, COL6A4P2, COL6A5, and COL6A6, were analyzed in the present study. To explore the intrinsic expression profiles of COL6A, we first examined gene expression levels in all 33 TCGA tumor tissues. Our results indicated striking inter-cancer heterogeneity in the expression levels of COL6A1/2/3 in pan-cancer. Further analysis found that COL6A1 and COL6A2 were relatively highly expressed in all cancer types compared with other COL6A members. By contrast, COL6A3 was moderately expressed, and COL6A4P1/4P2/5/6 had lower expression levels (Fig. 1A).Using the HPA database, we subsequently tested mRNA expression in normal tissues. The results showed that the mRNA expression levels of COL6A1/2/3 were generally higher in normal tissues compared to COL6A5/6, which were mainly enriched in the lung tissues (Additional file 2). In addition, coexpression analysis revealed significantly positive correlations between COL6A1 and COL6A2 (R = 0.86), followed by COL6A2-COL6A3 (R = 0.82), COL6A1-COL6A3 (R = 0.72), and COL6A5-COL6A6 (R = 0.51) (Fig. 1B).
We also analyzed RNA-seq data from the TCGA database to detect the differential expression of COL6A between tumor tissues and normal tissues in 18 tumors. The fold change (log FC) of COL6A members in different cancer types is summarized in detail in Additional file 1: Table S3. Our results revealed that COL6A was abnormally expressed in a variety of tumor types, either down-expressed or up-expressed in different tumors. For instance, COL6A1 was overexpressed in some specific tumors, including BRCA, CHOL, ESCA, GBM, HNSC, KICH, and KIRC, while the lower expression of COL6A1 was discovered in BLCA, PRAD, THCA, and UCEC. The other COL6A members showed different expression trends in 18 tested cancer types, depending on the specific tumor type. Interestingly, we also noticed that there were significantly opposite expression trends of different COL6A members in the same tumor. For pan-lung: LUAD and LUSC, low expression levels of COL6A5 and COL6A6 were observed, which was opposite to the expression levels of COL6A3, COL6A4P1 and COL6A4P2, and the expression levels of COL6A1 and COL6A2 had no significant changes. The significant overexpression of COL6A5 and downregulation of COL6A6 were observed in BRCA (Fig. 1C).
Prognostic role of COL6A in pan-cancer
The dysregulated COL6A members were found in pan-cancer, but their prognostic value remained unclear. Therefore, we further explored the association of expression levels of COL6A members with overall survival by Cox regression models, and HR > 1 was considered as a poor prognostic factor. The results revealed that COL6A1/2/3 were prognostic risk factors with HR > 1 in multiple cancer types (Fig. 2 and Additional file 1: Table S4). Specifically, these three genes were predicted poor prognosis in patients with BLCA, GBM, KIRC, KIRP, LGG, and MESO. Notably, COL6A4P1 was associated with lower survival in BRCA, LGG, and LIHC, while it was predicted to have a better prognosis for KICH and PCPG. COL6A5 correlated with poor prognosis of PCPG and UCEC, but favored survival of patients with HNSC and LUAD. In addition, increased COL6A6 expression was predicted poor prognosis for KIRP, READ, KICH, and UCEC, and a better survival rate in LUAD. We further used Kaplan–Meier survival curve to evaluate the prognosis risk of COL6A in 33 cancer types. The results showed that COL6A members were still significantly associated with patients’ overall survival in most tumor types. It is also worth noting that all COL6A members played adverse prognostic roles in patients with KIRC (Additional file 3).
Association of COL6A with immune subtypes, TME, and stemness scores in pan-cancer
After that, we focused on three major collagen VI isoforms, including COL6A1, COL6A2, and COL6A3, as a further research object. To understand their association with immune components, we analyzed the expression levels of COL6A1/2/3 in six immune subtypes from 33 tumors. The results indicated that COL6A1/2/3 were associated with immune subtypes (all p < 0.001), and had similar expression profiles in six immune subtypes. More specifically, they had the highest expression in C6, followed by C1, C2, and C3, while the lowest expression in C4 and C5 (Fig. 3A). These results suggested that the role of COL6A1/2/3 in inhibiting or promoting cancers may be associated with their immune effects.
We further investigated the relationship between the expression levels of COL6A1/2/3 and infiltrating stromal cells and immune cells in 33 cancers. ESTIMATE program was used to calculate stromal scores and immune scores. The results indicated that COL6A1/2/3 expression had a strong positive association with stromal scores and immune scores in multiple tumor types (Fig. 3B and C), but had a negative correlation with immune scores in TGCT and THCY, indicating that elevated expression levels of three genes were correlated with lower tumor purity.
Furthermore, we explored the relationship between COL6A1/2/3 and stem cell-like characteristics of 33 cancer types based on mRNA expression and stemness scores. The results indicated that COL6A1/2/3 had strong negative correlations with RNAss in most tumors. However, we also noticed that there was no significant association between COL6A1 expression and RNAss in KICH (Fig. 3D). Moreover, we found that there were differences between expression levels of COL6A1/2/3 and DNAss in various cancer types. Specifically, they were negatively associated with DNAss in BLCA, LIHC, and TGCT, while positively correlated with DNAss in CHOL, THCA, and THYM (Fig. 3E).
Drug sensitivity analysis of COL6A in pan-cancer
In order to explore the effects of COL6A1/2/3 on drug treatment, we analyzed their expression in NCI-60 cell lines, and conducted the Pearson’s correlation test to investigate the association between gene expression and drug sensitivity. The results showed that increased expression of COL6A1/2/3 was associated with drug sensitivity of distinct cell lines to multiple chemotherapeutic drugs (Fig. 4 and Additional file 1: Table S5). For instance, COL6A2 was related to cell sensitivity to bleomycin, zoledronate, taurosporine and simvastatin. COL6A3 was positively associated with zoledronate. However, COL6A1/2/3 were also associated with the resistance to several drugs. Moreover, we also noticed that different genes had similar associations with the same drug. For example, COL6A1/2/3 were all positively associated with staurosporine, while negatively correlated with by-products of CUDC-305. These findings indicated that COL6A1/2/3 could serve as potential treatment targets.
Underlying molecular mechanism analysis of COL6A
To further understand the potential molecular mechanism of COL6A1/2/3 in cancers, we first identified the effects of methylation patterns of COL6A1/2/3 in pan-cancer by GSCALite. Our results showed that the methylation levels of COL6A1/2/3 in tumor tissues were significantly lower than in normal tissues, as follows: COL6A1 in ESCA, BRCA, UCEC, and PRAD; COL6A2 in LIHC, HNSC, KIRP, and BLCA; COL6A3 in HNSC, BRCA, UCEC, COAD, and PRAD. However, these genes had higher methylation levels in tumor tissues, including COL6A1 in KIRC, HNSC, THCA, KIRP, and COAD; COL6A2 in KIRC, LUSC, and BRCA; COL6A3 in LIHC and KIRC (Fig. 5A). In addition, the expression of COL6A1/2/3 was mainly negatively correlated with methylation, except for COL6A3 in LIHC (Fig. 5B). Survival analyses revealed that the hypermethylation levels of COL6A1/2/3 were risk factors to predict prognosis in most cancer types, but the hypermethylation level was identified as a better prognostic factor for COL6A3 in DLBC and COL6A1 in SARC (Fig. 5C).We further constructed the miRNA-to-gene network of COL6A1/2/3. Our results showed that COL6A members were regulated by more than one miRNA. To be specific, COL6A1 was regulated by 67 miRNAs, COL6A2 was regulated by 21 miRNAs, while COL6A3 was regulated by 67 miRNAs. In addition, we also observed that the same miRNA could regulate multiple genes, such as hsa-miR-29-3p, which regulated COL6A1 and COL6A2 (Additional file 4).
Role of COL6A in colorectal cancer
Considering the heterogeneity of tumors from the same origin, we further investigated the role of COL6A1/2/3 in COAD and READ. Based on the immunohistochemical results from the HPA database, we studied the protein expression of COL6A1/2/3 in colorectal cancer tissues (Additional file 5). Besides, the CCLE database was used to show the expression of COL6A1/2/3 in colorectal cancer cell lines (Additional file 5). The expression patterns of COL6A1/2/3 were similar in six immune subtypes of COAD. More specifically, COL6A1/2/3 expression was relatively high in C6 while low in C4 (Fig. 6A). COL6A2 expression varied in six immune subtypes of READ and was higher in C4 (p < 0.05) (Fig. 6B). Very interestingly, there was no C5 subtype of COAD and READ. Figure 6C shows that the expression of COL6A1/2/3 in patients with COAD was significantly negatively associated with the RNAss and DNAss (p < 0.05), while positively correlated with stromal scores, immune scores, and estimate scores, and similar results were observed in READ (Fig. 6D).