TCGA database and genotypic tissue expression (GTEx) database were used to obtain gene expression profiles. An analysis of 31 normal tissues was performed using mRNA data obtained from the GTEx project. Cancer cell lines were analyzed in 31 tissues according to their expression levels, and then the Kruskal Wallis test was performed on the mRNA data between adjacent tissues and tumor tissues, as well as healthy tissues and tumor tissues, to determine the difference of DDX58 expression. DDX58 expression levels were compared between healthy tissues and tumor tissues, as well as between adjacent tumor tissues and tumor tissues.

HPA¹ contains normal tissue and tumor tissue protein levels of human gene expression profile information. In this study, we compared the expression of DDX58 protein in normal tissues and cancer tissues of four different organs by HPA. The significance of the difference was calculated using the Wilcoxon test. p < 0.05 suggests that the expression of tumor tissue is different from that of normal tissue.

The expression level of DDX58 was examined using univariate COX regression analysis to determine whether it was associated with tumor prognosis in various cancers. According to the median of DDX58, samples were divided into two groups based on their expression levels: high-and low-expression groups of DDX58. In order to determine the importance of survival differences, a log rank test was used, with a threshold of p = 0.05. What’s more, we used the limma package to learn the relationship between DDX58 and T stage in pan-cancer.

We constructed gene–gene interaction network through GeneMANIA² and built PPI network through STRING database.³ We had further constructed a gene disease network on the OPENTARGET platform. Gene ontology (GO) terminology, Kyoto Encyclopedia of Genes and Genomes (KEGG) and GSEA were used to gene enrichment analysis. The term “GO” refers to molecular function (MF), cellular components (CC), and biological processes (BP). Use the “ClusterProfiler” package to perform GO, KEGG analysis, and GSEA. The TIMER⁴ was a comprehensive online database, analysis of a wide variety of cancer types related to immune infifiltrating. In this study, we used TIMER to determine the relationship between DDX58 expression and ACE2.

As a form of DNA chemical modification, DNA methylation controls gene expression by changing epigenetics without changing DNA sequence. To analyze the methylation level of tumor and normal tissues, we obtained them from the methylation module of UALCAN database. Later, from UCSC⁵ database, we have downloaded a standardized pan-cancer dataset: TCGA Pan Cancer (PANCAN, N = 10,535, G = 60,499), from which we further extracted the expression data of DDX58 gene and 44 marker genes of three kinds of RNA modified m6A genes in each sample. We filtered the samples from: Primary Blood Derived Cancer-Peripheral Blood, Primary Tumor. Further, log2 (x + 1) transformation has been performed for each expression value. Next, we had calculated the spearman correlation between DDX58 and marker genes of five different immune pathways.

The total number of substitutions, insertions and deletions per megabase in the coding region of tumor gene exons was used to calculate the tumor mutation load (TMB). We got the expression data of DDX58 gene in every sample from the previously downloaded datasets, combined with the previously screened samples. In addition, we also had download the Simple Nucleotide Variation dataset of level4 of all TCGA samples processed through MuTect2 software from GDC (https://portal.gdc.cancer.gov/; Beroukhim et al., 2010). To calculate the tumor mutation burden (TMB), we used the TMB function of the R software package maftools (version 2.8.01). Then we integrated the TMB and gene expression data of the samples. Finally, we obtained the expression data of 37 cancer species after removing those with fewer than three samples in a single cancer species. Through cBioPortal resources,⁶ we had analyzed the genetic changes of DDX58 in the TCGA dataset (Reimer et al., 2021). The gene changes and mutation sites of DDX58 were obtained in the “Oncoprint,” “Cancer Type Summary,” and “Mutations” sub modules.

Biological phenomena such as gene fusion, deletion mutation and point mutation are called new antigens encoded by mutated genes in tumor cells. We had calculated the binding affinity score of epitopes with 8–11 amino acids of a certain length and the epitopes with a score less than 500 nm were defined as new antigens. Then, we ranked the predicted new antigens according to antigenicity index value, affinity and mutation allele frequency. In each tumor sample, scannedo was used to count the new antigens and analyze the relationship between DDX58 expression and new antigens. The immune checkpoint genes had been extracted and analyzed along with the DDX58 expression to further investigate their relationship.

From UCSC (see Footnote 5) database we had downloaded a standardized pan-cancer dataset: TCGA Pan Cancer (PANCAN, N = 10,535, G = 60,499). Based on the previously extracted expression data and screened samples, we obtained MSI (Microsatellite instance) scores of each tumor from the previous study (Gounder et al., 2022). Next, the MSI and gene expression data of the samples were integrated.

We screened the metastatic samples from the following sources: Primary Blood Derived Cancer - Peripheral Blood (TCGA-LAML), Primary Tumor, and TCGA-SKCM. The gene expression profiles of each tumor were extracted, mapped to GeneSymbol, and further analyzed using the Timer method of the R software package IOBR (version 0.99.9, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8283787/; Li et al., 2017). The B cell, T cell CD4, T cell CD8, Neutrophil, Macrophage and DC infiltration scores of each patient in each tumor were reevaluated according to gene expression.

To study the drug sensitivity of pan-cancer patients to DDX58, the CallMinerTM database was used⁷ to get activity data and RNA seq expression profile of NCI-60 compounds. In order to analyze and select drugs approved by FDA or clinical trials, R packages “impute,” “limma,” “ggplot2,” and “ggpubr” were used for analysis.

The analysis of gene disease network interaction showed that DDX58 was mainly related to genetic, familial or genetic disease, immune system disease, infectious disease, benign tumor, etc. In particular, DDX58 had a certain relationship with benign tumor (Figure 1A). Subsequently, we investigated the role of human DDX58 expression in pan-cancer. A comparison of the expression levels of DDX58 in tumors and normal tissues was performed using the TCGA database. As compared to normal tissues, DDX58 was found to be highly expressed in BRCA, ESCA, STES, KIPAN, STAD, HNSC, KIRC, LIHC, CHOL, while it was low expressed in LUAD, COAD, READ, KIRP, LUSC, KICH (Figures 1B,C). At the same time, DDX58 protein levels in four different organs and tissues also showed significant differences (Figure 1D).

Next, an analysis of DDX58 expression and cancer prognosis was conducted using univariate Cox regression. According to the forest map of pan-cancer, the expression of DDX58 had a significant impact on the OS of LGG, KIRC, SKCM, MESO, TGCT, PAAD, LUAD patients (Figure 2). In addition, we also analyzed the expression of DDX58 in different cancer T stages, and the results showed that there were significant differences in the expression of DDX58 in different stages of 10 cancers (Supplementary Figure S1).

In order to understand the DDX58 related network, the STRING and GeneMANIA protein-protein and gene-gene interaction networks that interact with DDX58 were used (20 potential related genes were selected respectively; Figures 3B,C). We obtained 8 genes from the intersection of two data sets (Figure 3A) and carried out GO and KEGG analysis on 9 genes including DDX58 (Figure 3D). We found that BP was enriched in negative regulation of type I interchange production, regulation of type I interchange production, type I interchange production. CC was mainly enriched in phagophore assembly site membrane, phagophore assembly site, phagocytic vascular membrane. MF was significantly enriched in protein tag, Lys63 specific dehydrogenase activity, and Lys48 specific dehydrogenase activity. KEGG analysis showed that many related pathways were significantly enriched, including RIG-I-like receiver signaling pathway, NF kappa B signaling pathway, Influenza A. In addition, it could be seen from the correlation analysis with AEC2 (SARS CoV-2 receptor) that there was a positive correlation between the expression of DDX58 and ACE2 in many cancers (Supplementary Figure S2).

According to promoter methylation analysis, DDX58 is hypermethylated in a variety of cancer types (Figure 4A). DDX58 methylation seems to be correlated with the level of DNA methyltransferase mRNA expression in various cancers (all p < 0.05; Figure 4B). As we all know, DNA methylation is the result of DNA methyltransferase, which plays a role by covalently binding to the methyl at the 5′ carbon position of cytosine, a CpG dinucleotide in the genome. A correlation was found between methyl related genes and various cancers. There was a positive correlation between the expression of DDX58 in pan-cancer and methyl-related genes, which meant that DDX58 may mediate tumor genesis and progression by regulating epigenetic status. Moreover, it was worth noting that the correlation coefficient was higher in DLBC and UVM.

Based on the cBioPortal database, we found that there were higher DDX58 gene changes in LUSC, UCEC, STAD, and SKCM, and mutation was the main type (Figure 5A). It further proved the type, location and quantity of DDX58 gene modification. R244K/I changes were detected in 4 patients with DDX58 (Figure 5B). Then the 3D structure of DDX58 protein at this mutation site was mapped (Figure 5C). The most common type of mutation found in pan-cancer analysis were gain and diploid (Figure 5D). In addition, TRAJ6, TMEM158, YY1P2, TTN, TAF1L, TP53, TOPORS, MUC16, ACO1, RYR2 gene changes were more common in the altered group than in the unchanged group (Figure 5E).

TMB and MSI are effective prognostic biomarkers and indicators of immunotherapeutic response in many tumors. From these two analyses, we can conclude the relationship between DDX58 and immunotherapy prognosis of specific cancer types.

The tumor cell genome’s TMB is usually expressed as the total number of non-synonymous mutations within an average 1 M base region. In some cases, it is also expressed directly as the number of somatic mutations. Base substitution, frameshift mutation, deletion mutation, insertion mutation and other mutation types are the most common mutation type. In tumor cells, TMB is a quantifiable indicator of mutation frequency. The correlation between DDX58 and TMB was calculated for each tumor. Ten tumors showed a significant correlation, including a significant positive correlation in 6 tumors, such as GBMLGG (N = 650; R = 0.1430, p = 0.0002), COAD (N = 282; R = 0.1328, p = 0.0257), COADREAD (N = 372; R = 0.1103, p = 0.0334), KIPAN (N = 679; R = 0.194, p = 3.4926e-7), UCS (N = 57; R = 0.3099, p = 0.01895), BLCA (N = 407; R = 0.0996, p = 0.0444), significantly negative correlation in 4 tumors, for example: BRCA (N = 981; R = −0.0669, p = 0.0361), HNSC (N = 498; R = −0.1226, p = 0.0061), THCA (N = 489; R = −0.2149, p = 0.00001), UVM (N = 79; R = −0.3177, p = 0.0043; Figure 6A).

DDX58 expression correlated with MSI in different types of cancer and we had calculated their spearman correlation in each tumor. A significant correlation was observed in 10 tumors, including significant positive correlation in 3 tumors, such as COAD (N = 285; R = 0.1526, p = 0.0098), COADREAD (N = 374; R = 0.1068, p = 0.0389), THYM (N = 118; R = 0.1868, p = 0.0428), and the significant negative correlation in 7 tumors, such as GBMLGG (N = 657; R = −0.1286, p = 0.0009) LGG (N = 506; R = −0.0878, p = 0.0483), KIPAN (N = 688; R = −0.3528,p = 1.351e-21), PRAD (N = 495; R = −0.1286, p = 0.0041), THCA (N = 493; R = −0.0967, p = 0.0317), PAAD (N = 176; R = −0.1677, p = 0.0260), DLBC (N = 47; R = −0.4937, p = 0.0004; Figure 6B).

It was worth noting that the absolute coefficients associated with TMB or MSI in the COAD cohort were relatively high compared with other cancer types, suggesting that the it may be sensitive to immunotherapy.

To determine whether this pathway affects the tumor immune microenvironment, we studied the expression of DDX58 with the degree of immune cell infiltration in each cancer type. Using the data collected from TCGA and the six types of immune cells available in TIMER database (B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages and dendritic cells) for analysis, the results indicated that there was significant correlation in multiple tumors (Figure 7). It was worth noting that CD8 + T cells had the highest DLBC correlation coefficient. Their corresponding linear regression diagram showed that the high expression of DDX58 may be related to the increased level of immune cell infiltration. Similarly, DDX58 also affected the expression of immune checkpoints in different cancers (Supplementary Figure S3).

The database was further analyzed to determine whether DDX58 expression was correlated with drugs using CellMiner™ (Figure 8). Our results indicated that the expression of DDX58 was positively correlated with the sensitivity to Cediranib, VE−821, Itraconazole, JNJ−42756493, IWR−1, Linsitinib. And the expression of DDX58 was negatively correlated with the drug sensitivity of geldanamycin analysis, Tanespimecin, TYROTHRICIN, Panobinostat, Alvespimycin, Quisinostat, XR−5944, Lapiphone, Paclitaxel,Tamoxifen.