Expression and potential immune involvement of cuproptosis in kidney renal clear cell carcinoma

Background

Recently, a study of cuproptosis [1] has raised broad attention in the field of cancer cell death. Cuproptosis is a newly identified programmed cell death pathway mediated by intracellular free copper, that is distinguished from apoptosis, autophagy, necrosis, ferroptosis, pyroptosis, and oncosis. Although copper is an essential co-factor for core enzymes in cells [2], the concentration of intracellular copper is maintained at a very low level. Studies have revealed that the increase in the copper level in cells leads to the lethal fate of cells [3, 4], however, the exact mechanism of death inductions by copper has not been identified until the recent study of cuproptosis [1]. Cancer researchers are interested in cuproptosis because previous literature suggested that copper plays a role in different cancer types, including breast [5], head and neck [6], endometrial cancer[7], etc, yet, the blurt in the pathway of copper-cancer interactions prevent the filed from progressing. Now, a set of genes were identified in the cuproptosis pathway, enabling us to further explore the effect of copper in cancers.

Three anti-cuproptosis genes (MTF1, GLS, and CDKN2A) and seven pro-cuproptosis genes (FDX1, LIAS, LIPT1, DLD, DLAT, PDHA1, and PDHB) genes[1] with an anti-cuproptosis copper exporter ATP7B [8] and a pro-cuproptosis copper importer SLC31A1 [9] were studied in this study for a better insight into the role of cuproptosis in cancers. Our preliminary screening analysis based on these 12 cuproptosis genes across more than 9000 samples of 33 types of solid tumors identified kidney renal clear cell carcinoma (KIRC) as a cancer type most likely to be affected by cuproptosis. This study analyzed the multi-omic data to explore the cancer-noncancer expression pattern and potential immune involvement of the cuproptosis pathway in KIRC. The data sources and analysis methods of this study were provided in the supplementary materials. We hope this study can contribute to the application of cuproptosis in the clinical therapy of KIRC.

Results And Discussion

KIRC was a cancer type that might be affected by cuproptosis

We identified KIRC as the cancer type most likely to be affected by cuproptosis by comparing the cancer-noncancer expression gene set variation analysis (GSVA) scores of pro-cuproptosis genes and anti-cuproptosis genes respectively and analyzing the survival association of these scores across 33 types of solid tumors. Results showed that seven cancer types had higher pro-cuproptosis GSVA scores with lower anti-cuproptosis GSVA scores in cancer than in normal tissues, including BRCA, CHOL, COAD, HNSC, KIRC, KIRP, LIHC, READ, and SARC (S-Fig.1A), indicating that the cuproptosis pathway of these cancer types might be inactivated in tumorigenesis. The survival analysis of GSVA scores showed that KIRC was the only cancer type whose overall survival and disease-specific survival were significantly associated with both pro-cuproptosis and anti-cuproptosis GSVA scores (S-Fig.1B), suggesting that the alteration in cuproptosis pathway might affect the survival in KIRC. Therefore, this study focused on the KIRC.

Cancer-noncancer expression of cuproptosis molecules in KIRC

We specifically compared the cancer-noncancer expression of each cuproptosis molecule in KIRC. TCGA and GTEx data suggested that, among the 8 pro-cuproptosis genes, FDX1, DLD, DLAT, PDHA1, PDHB, and SLC31A1 were significantly down-regulated in KIRC compared to normal tissues, with an anti-cuproptosis gene, GLS, also down-regulated in KIRC, while among the other 3 anti-cuproptosis genes, CDKN2A and ATP7B were significantly up-regulated in KIRC (Fig.1A). The paired TCGA data expression analysis also showed the same cancer-noncancer expression pattern as the TCGA and GTEx data analysis (Fig.1B). To validate the cancer-noncancer expression pattern of these cuproptosis molecules at the protein level, we compared the protein expression levels of KIRC from TPTAC data, which included protein expression of KIRC and normal tissues of the cuproptosis set except for SLC31A1 and ATP7B. Results showed that the protein expression pattern was consistent with the mRNA expression except for LIAS and LIPT1, which were not significant at the mRNA level but down-regulated at the protein level in KIRC compared with normal tissues (Fig.1C). The HPA staining images data also showed that protein expression patterns were consistent with mRNA expression patterns except for LIAS and LIPT1. Protein expression patterns of LIAS and LIPT1 from TPTAC data were consistent with that from HPA data that they were down-regulated in KIRC compared with normal tissues (Fig.1D). The down-regulation of anti-cuproptosis molecules with the up-regulation of the anti-cuproptosis gene CDKN2A in KIRC inferred that KIRC might have a lower cuproptosis than normal tissues, at the same time, we suggested, that the down-regulation of anti-cuproptosis gene GLS in KIRC resulted from the negative feedback of the pathway.

Survival associations of cuproptosis genes in KIRC

In addition, we analyzed the overall survival association of the 12 cuproptosis genes specifically. Results showed that all the 8 pro-cuproptosis genes were significantly associated with better overall survival in KIRC, including FDX1, LIAS, LIPT1, DLD, DLAT, PDHA1, PDHB, and SLC31A1, while, among the 4 anti-cuproptosis genes, MTF1 and GLS were significantly associated with better overall survivals with CDKN2A and ATP7B associated with worse overall survivals (S-Fig.2). The survival beneficial effect of the 8 pro-cuproptosis genes indicated that the cuproptosis might cause more cancer cell death accounting for better survival in KIRP patients, while the anti-cuproptosis genes CDKN2A and ATP7B did the opposite. On the other hand, for the other 2 anti-cuproptosis genes, MTF1 and GLS, we suggested, their survival beneficial effect indicated that they might mediate the negative feedback of cuproptosis in KIRC, which was generally consistent with the results of the expression analysis.

KIRC subtypes clustering based on cuproptosis

To investigate the cuproptosis signature in KIRC, we clustered the TCGA KIRC samples based on the cuproptosis gene set. The consensus cumulative distribution function (CDF) calculation demonstrated that when the number of clusters (K) = 3, the delta area decreased remarkably, while K>3 did not further remarkably reduce the delta area, hence K = 3 was chosen as the optimum cluster number (Fig.2A). By a machine-learning algorithm called the non-negative matrix factorization (NMF) method, patients were clustered into three distinct subtypes, we named them C1, C2, and C3 (Fig.2B). The expression heatmap (Fig.2C) and expression PCA plotting (Fig.2D) also demonstrated the expression distinction among the three subtypes. The survival analysis of the three subtypes revealed that the C2 subtype was significantly associated with worse overall survival in KIRP (Fig.2E). These results further supported the effect of cuproptosis on KIRP survival and might be applied for clinical prognosis. Therefore, we utilized the machine-learning algorithm called least absolute shrinkage and selection operator (LASSO) regression to construct a prognostic model of KIRC based on the cuproptosis genes set and conducted an assessment of the risk model to demonstrate the prediction. However, the model did not show very good prediction accuracy with an AUC of time-dependent AUC of less than 0.7 (S-Fig.3). Furthermore, to further explore the potential involvement of cuproptosis in cancer biology, we conducted an intersection analysis among differentially expressed genes (DEGs) in C1, C2, and C3 and identified the common DEGs (Fig.2F). We then enriched these common DEGs in GO terms and found that cuproptosis associated with cell adhesion, cytokine, MAPK pathways, inflammatory, etc (Fig.2G).

The cuproptosis subtype difference in KIRC immunity

As the cytokine and inflammatory were part of the tumor immune microenvironment, we were especially interested in the potential role of cuproptosis in cancer immunity. Thus, we analyzed the immune cell infiltration levels in KIRC and compared the subtypes. Results revealed that the subtypes were significantly different in multiple immune cells (Fig.2H). The association between cuproptosis and immune cell infiltration levels inferred that the cuproptosis gene set might be applied in the prognosis of immune therapy. To study if the subtypes affect immune therapy, we compared the expression of immune checkpoints CD274, CTLA4, HAVCR2, LAG3, PDCD1, PDCD1LG2, SIGLEC15, and TIGIT [56] among subtypes. Results revealed that the expression among subtypes was different in all 8 immune checkpoints we analyzed (Fig.2I). These results suggested that cuproptosis might have a connection to immune therapy. To further explore the effect of the cuproptosis subtypes on immune therapy, we compared predicted immune checkpoint blockade (ICB) responses of the subtypes. Potential ICB response was predicted using the Tumor Immune Dysfunction and Exclusion (TIDE) algorithm using a set of gene expression markers to evaluate two different tumor immune escape mechanisms, including the dysfunction of tumor-infiltrating cytotoxic T lymphocytes (CTL) and the exclusion of CTL by immunosuppressive factors. A high TIDE score results in a poor response to the immune checkpoint blocking therapy (ICB) and worse survival in immune therapy. Our calculation showed that the C2 subtype had a significantly higher TIDE score than the other 2 subtypes (Fig.2J bottom panel). Only 22.4% (41 out of 183) of the C2 subtype KIRC patients were predicted to respond to the ICB treatment, while 41.0% (132 out of 322) of the C1 subtype KIRC patients were predicted to respond to ICB treatment and 56.0% (14 out of 25) of the C3 subtype KIRC patients were predicted to respond to ICB treatment (Fig.2J top panel). These results suggested that the patients with C2 subtype KIRC were more likely to respond to ICB therapy.

The cuproptosis signature was an immunotherapy biomarker for KIRC

Based on the above analysis of TCGA data, we proposed that the cuproptosis signature can be a predictive factor for immunotherapy. To validate this hypothesis, we analyzed a PD1 blockade immunotherapy cohort of KIRC and compared the survival prediction power of the cuproptosis signature and other standardized immunotherapy biomarkers, including TIDE score, MSI score, mutation (TMB), CD274, CD8, IFNG, T.Clonality, B. Clonality, and Merck 18. Results showed that the cuproptosis signature performed the best survival prediction of the PD1 blockade immunotherapy among all these indicators compared (Fig.2K). The prognosis of PD1 blockade immunotherapy of KIRC was found challenging because all the other standardized immunotherapy biomarkers failed to show significant survival association and, although the difference was subtle, the cuproptosis signature was the only biomarker significantly associated with survival (Fig.2L). These results suggested that the cuproptosis signature might have potential values as an immunotherapy biomarker for KIRC. Nevertheless, more studies are required to further support the application of cuproptosis signature in immunotherapy.

Expression and potential immune involvement of cuproptosis in kidney renal clear cell carcinoma

Abstract

Background

Results And Discussion

Conclusion

Declarations

References