Systematic pan-cancer analysis identifies cuproptosis-related gene DLAT as an immunological and prognostic biomarker

Pan-cancer analysis of DLAT expression

First, we utilized the Genotype Tissue-Expression (GTEx) dataset to evaluate the mRNA expression of DLAT in normal samples. Notably, the expression of DLAT was observed to be lowest in blood, as demonstrated in Figure 1A, while it was found to be highest in bone marrow. Furthermore, Figure 1B showed that the DLAT expression level was the lowest in uveal melanoma (UVM), while it was highest in kidney chromophobe (KICH) among the tumor tissues within the TCGA dataset. Subsequently, using the Cancer Cell Line Encyclopedia (CCLE) database of tumor cell lines, we determined that adrenal cancer had the highest DLAT gene expression (Figure 1C).

We also compared the tumor’s DLAT expression with that of normal tissues premised on TCGA data. Figure 1D demonstrates that DLAT expression was upmodulated in 7 distinct cancer types: cholangiocarcinoma (CHOL), esophageal carcinoma (ESCA), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), stomach adenocarcinoma (STAD), and stomach and esophageal carcinoma (STES). Contrastingly, the expression of DLAT was shown to be low in nine different malignancies: breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), colon adenocarcinoma/rectum adenocarcinoma (COADREAD), head and neck squamous cell carcinoma (HNSC), pan-kidney cohort (KIPAN), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), prostate adenocarcinoma (PRAD), and thyroid carcinoma (THCA).

Besides, we investigated whether a correlation existed between DLAT expression and tumor stages or grades. As per the findings, the expression of DLAT was significantly related to tumor stage in seven cancer types, namely, COADREAD, KIPAN, KIRC, LIHC, LUAD, rectum adenocarcinoma (READ), and THCA (Supplementary Figure 1). As for the DLAT expression in different tumor grades, significant differences were observed in five cancer types, including glioma (GBMLGG), KIPAN, KIRC, brain lower grade glioma (LGG), and pancreatic adenocarcinoma (PAAD) (Supplementary Figure 2).

In addition, we utilized the Human Protein Atlas (HPA) database to compare the expression of DLAT in normal and tumor samples at the protein level. Strong staining of DLAT was observed in STAD tissues, in contrast to the normal stomach, which had only weak immunohistochemistry (IHC) staining (Figure 2). DLAT staining was found to be moderate in samples of normal liver tissue, but the staining was strong in samples of tumor tissues. The DLAT staining in normal lung tissues was weak, whereas that of LUSC tissues was moderate and strong in LUAD tissues. Conversely, DLAT staining was strong in samples of normal colon and prostate tissues but only moderate in tumor tissue samples. DLAT staining was moderate in normal breast tissue samples but weak in cancer tissues.

Prognostic relevance of DLAT

After that, we examined the role of DLAT expression in tumor survival outcomes. For overall survival (OS), DLAT functioned as a risk factor in bladder urothelial carcinoma (BLCA), BRCA, GBMLGG, LGG, LIHC, PAAD, and UVM based on the analyses of Cox regression. In contrast, it functioned as a protective factor in COAD, COADREAD, KIPAN, KIRC, and READ. The Kaplan-Meier (KM) analysis on the OS illustrated that high levels of DLAT expression were related to dismal prognoses in patients suffering from BRCA, GBMLGG, LGG, LIHC, and PAAD, but a favorable prognosis was found in COADREAD, KIPAN, KIRC, KIRP, and READ (Figure 3B–3K). Subsequently, Cox regression analysis of the disease-specific survival (DSS) showed that DLAT was a risk factor for six distinct cancers, including BLCA, GBMLGG, LGG, LIHC, PAAD, and UVM. On the other hand, as shown in Figure 4A, it was a protective factor in COADREAD, KIPAN, KIRC, and KIRP. Figure 4B–4E show that a high level of DLAT expression was associated with a favorable DSS for the COADREAD, KIPAN, KIRC, and KIRP based on the results of KM analysis. However, poor DSS was found in patients with high DLAT expression levels in GBMLGG, LGG, PRAD, and primary skin cutaneous melanoma (SKCM-P) (Figure 4F–4I). Additionally, the findings result from Cox regression analysis showed that DLAT was a risk factor for disease-free interval (DFI) in PAAD (Figure 5A). Further KM analysis demonstrated that patients with PAAD whose DLAT expression was higher exhibited a poorer DFI in contrast with those whose DLAT expression was lower, and reversely in GBMLGG and LGG (Figure 5B–5D). Lastly, for progression-free interval (PFI), DLAT served as a risk factor in patients with adrenocortical carcinoma (ACC), BLCA, cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), GBMLGG, LIHC, PAAD, SKCM-P, and UVM but was found to be a protective factor in KIPAN and KIRC according to the Cox regression analysis findings (Figure 6A). Furthermore, an unfavorable PFI was found in patients exhibiting high levels of DLAT expression in ACC, GBMLGG, PAAD, pheochromocytoma and paraganglioma (PCPG), and SKCM-P based on the results of KM analysis, while opposite results were found in COADREAD, KIPAN, and KIRC, as depicted in Figure 6B–6I.

Genetic alteration of DLAT

Both DNA methylation and genetic alterations have been shown to be associated with tumorigenesis and tumor progression. First, we evaluated the frequency of DLAT alterations in cancer patients utilizing the cBioPortal database. Notably, patients with SKCM had the highest frequency of DLAT alterations, accounting for 4.73 percent, in comparison to other types of cancer. The “deep deletion” type was prevalent in most cancers, including BRCA, ESCA, HNSC, KIRC, KIRP, PRAD, sarcoma (SARC), skin cutaneous melanoma (SKCM), testicular germ cell tumors (TGCT), uterine carcinosarcoma (UCS), and UVM (Figure 7A). Subsequently, we assessed the degree of association between DLAT mRNA expression and promoter methylation and found significant correlations in seven malignancies (Table 1). Interestingly, an inverse relationship was observed between promoter methylation level and DLAT expression level in BRCA, CESC, LIHC, and PCPG (Figure 7B). Furthermore, the relationship between DLAT expression level and CNV was also analyzed. The results indicated that DLAT expression levels were positively linked to CNV in nearly all cancer types, except for GBMLGG, acute myeloid leukemia (LAML), LGG, and UVM (Table 1). Figure 7C displays the six tumor types with the highest correlation coefficients.

Gene set enrichment analyses (GSEA) of DLAT

We conducted a GSEA analysis to identify the pathways in which DLAT may be involved, in order to better understand the potential biological processes underlying DLAT expression in different cancer tissues. Interestingly, our results showed that several types of tumors exhibited similar GSEA results. The findings demonstrated that DLAT is implicated in immune modulation-associated pathways in pan-cancer, such as “Antigen activates B Cell Receptor (BCR) leading to generation of second messengers,” “CD22 mediated BCR regulation,” “FCGR activation,” “FCGR3A-mediated IL10 synthesis,” “FCERI mediated Ca+2 mobilization,” “FCERI mediated MAPK activation,” “complement cascade,” “Role of LAT2/NTAL/LAB on calcium mobilization,” and “Immunoregulatory interactions between a Lymphoid and a non-Lymphoid cell” (Figure 8A–8F). As per these findings, DLAT possibly performs an instrumental function in modulating cancer immunity.

Tumor microenvironment analysis

The TME, which enables the survival of tumor cells, plays a crucial role in the development and progression of cancers. Therefore, we utilized the immune and stromal scores derived from the ESTIMATE algorithm to assess any potential associations between DLAT expression and TME components. The result revealed a negative correlation between DLAT expression and the immune score across 27 different cancer types, including ACC, BRCA, CESC, ESCA, glioblastoma multiforme (GBM), GBMLGG, HNSC, LAML, LGG, LUAD, LUSC, mesothelioma (MESO), KIPAN, KIRC, KIRP, PCPG, PRAD, SARC, SKCM, metastatic skin cutaneous melanoma (SKCM-M), SKCM-P, STAD, STES, TGCT, THCA, thymoma (THYM), and uterine corpus endometrial carcinoma (UCEC). Moreover, in 14 different types of cancer, including ACC, CESC, ESCA, GBM, KIPAN, LUAD, LUSC, PCPG, SARC, SKCM-P, STAD, STES, THCA, and UCEC, DLAT expression was found to be inversely related to stromal scores (Table 2). The five tumor types with the highest correlation coefficients are shown in Figure 9.

Analysis of immune cell infiltration

Following this, we conducted an inquiry to examine the potential relationship between DLAT expression and the degree of infiltration by various immune cells. According to the results from the TIMER algorithm, DLAT expression had a significant correlation with the levels of infiltrating neutrophils in 25 different cancer types, macrophages in 24, B cells in 22, CD8+T cells in 21, dendritic cells in 19, and CD4+T cells in 16 (Figure 10A). Furthermore, we found that the six cancer types with the strongest association between DLAT expression and immune infiltration levels were BRCA, COAD, COADREAD, GBMLGG, LGG, and THCA. In the case of BRCA, COAD, and COADREAD, all six immune cell types were positively linked with DLAT expression. As for GBMLGG, LGG, and THCA, B cells, CD8+T cells, macrophages, neutrophils, and dendritic cells were positively correlated with DLAT expression, whereas CD4+T cells showed a negative correlation.

Consistent with the aforementioned findings, further analyses employing the CIBERSORT algorithm corroborated that DLAT expression was significantly associated with immunological infiltration levels in most cancer types (Figure 10B). Furthermore, there were statistically significant links between the levels of DLAT expression and the various subtypes of infiltrating macrophages. DLAT expression exhibited a positive correlation with the levels of M0 macrophage infiltration in GBM, GBMLGG, LGG, LIHC, STAD, and STES, while showing a reverse correlation in KIPAN, KIRC, and UVM. DLAT expression was positively linked to the macrophage M1 subtype in BLCA, BRCA, COADREAD, KIPAN, KIRC, LGG, PAAD, PRAD, READ, STAD, and THYM, while showing a negative correlation in GBM and THCA. The degree of infiltrating macrophage M2 was negatively associated with DLAT expression only in LGG, whereas it was positively associated with 19 cancer types, including COAD, COADREAD, BRCA, HNSC, KIPAN, KIRC, KIRP, LIHC, LUAD, LUSC, ovarian serous cystadenocarcinoma (OV), PAAD, SKCM, SKCM-M, SKCM-P, STAD, TGCT, THCA, and THYM.

Analyses of tumor mutational burden and microsatellite instability

TMB and MSI have proven to be effective predictors of immune treatment responses across a multitude of tumors. We subsequently investigated whether DLAT expression correlated with TMB and MSI in various cancers (Figure 11). The results indicated a significant correlation between DLAT expression and TMB in ten different types of cancer. Specifically, DLAT expression was positively correlated with TBM in GBMLGG, LAML, LUAD, STES, STAD, THYM, and UCEC, but inversely correlated with TMB in KIPAN, KIRC, and THCA (Figure 11A). Additionally, we found that DLAT expression was positively correlated with MSI in KIPAN, STAD, STES, and UCEC, but negatively correlated with MSI in BRCA, lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), GBMLGG, HNSC, LUSC, PRAD, and THCA (Figure 11B).

Immune-related genes analyses

We further investigated whether there existed a co-expression between DLAT and immune-related genes in a diverse array of cancers. The results indicate that DLAT was closely associated with the majority of immune-related genes, as evidenced by Figure 12. Chemokines, such as CCL28, CXCL8, and CXCL16, and chemokine receptors, including CCR1, CCR8, and CXCR2, were positively linked to DLAT expression in various cancers. Notably, there was a robust relationship between the expression of DLAT and major histocompatibility complex (MHC) genes in all cancer types, particularly in KIRC, LIHC, PRAD, TGCT, and UVM. In addition, immunosuppressive genes and immune activation genes were both strongly linked to DLAT expression.

Data processing and differential expression analysis

RNA sequencing and associated phenotype data were extracted from TCGA (a total of 10535 samples) utilizing UCSC Xena (https://xena.ucsc.edu/). In addition, DLAT expression data from tumor cell lines and normal tissues were acquired from the CCLE database (https://portals.broadinstitute.org/ccle/) and the GTEx database (https://commonfund.nih.gov/GTEx) respectively for a multidimensional investigation.

Cancer types with fewer than three normal samples in the TCGA database were excluded for a robust differential expression analysis. Finally, the DLAT expression levels in 26 distinct human cancer types were analyzed. The R package “ggpubr” was employed to visually display the data.

Immunohistochemistry staining

The IHC data of DLAT expression were extracted and analyzed based on the HPA database (http://www.proteinatlas.org/).

DLAT gene expression and survival

To explore the potential prognostic value of DLAT in cancer, the survival data of OS, DSS, DFI, and PFI downloaded from TCGA were analyzed. For each different form of cancer, we used both the Cox regression analysis and the KM survival method to analyze the survival data. The threshold value for categorizing patients into two groups was determined according to each tumor’s median value of DLAT expression. The R packages “forestplot,” “survminer,” and “survival” were employed to visually represent the data.

DLAT genetic alteration analysis

To explore DLAT genomic alterations across the TCGA cancer types, we applied the “Cancer Types Summary” section of the cBioPortal website (www.cbioportal.org) [31]. In addition, the methylation data (merged HM27 and HM450), as well as the CNV data, were also downloaded from the cBioPortal website. The R package “ggpubr” was used to analyze and visually display the relationship between the levels of DLAT expression and its promoter methylation levels or CNV.

Tumor mutation burden and microsatellite instability

TMB refers to the overall amount of somatic coding mutations present in a particular kind of cancer and has been shown to correlate highly with immunotherapy’s efficacy [32–34]. We estimated each patient’s TMB score after downloading the data on somatic mutations from the UCSC Xena database, which included information on all TCGA patients. MSI, which suggests a deficiency in DNA mismatch repair, is a biomarker that indicates favorable responsiveness to immunotherapy [35]. The MSI data in this research was extracted from a published study [36]. As part of this research, we determined if there was a connection between DLAT expression and TMB or MSI in various cancers. The “ggpubr” R package was utilized for visualizations of the results.

Association of DLAT expression with immunity

ESTIMATE is an algorithm to infer the abundance of immune and stromal components within tumors. With this study, the R package “estimate” was used to calculate the ImmuneScore and StromalScore of each tumor sample. Furthermore, we used two algorithms, including TIMER and CIBERSORT, to analyze the link between DLAT expression and the infiltration of different immune cell types in various tumors. Additionally, we utilized the “limma” R package to carry out co-expression analyses of DLAT and immune-related genes, including those encoding chemokines, chemokine receptor proteins, MHC, immunostimulators, and immune inhibitors. To display the findings, the “ggpubr,” “reshape2,” and “RColorBreyer” packages were utilized.

Gene set enrichment analyses

DLAT’s biological functions in tumors were investigated via GSEA. The R package “clusterProfiler” and Reactome gene sets obtained through the GSEA website (https://www.gsea-msigdb.org/gsea/downloads.jsp) were applied for GSEA.

Statistical analysis

Expression data in this study were Log2(x+1) transformed. T-tests were utilized to estimate the differences in levels of DLAT expression between tumor and normal samples. The differential analyses of survival outcomes between high- and low-expression of DLAT were conducted using the log-rank test. The relationship between two variables was examined using Spearman’s test. All statistical analyses were done in R (Version 4.0.3). The statistical significance level was set at p < 0.05.