CD248 is highly expressed in pancreatic cancer and results in a poor prognosis
In this study, we first analyzed the transcript levels of CD248 gene in multiple tumor tissues through TCGA and GTEx datasets and observed the gene expression in 27 tissues (Fig. 2a). According to the data, the expression of CD248 was significantly different in several tumor tissues compared with normal tissues (p˂0.05), among which, the expression of CD248 was significantly higher in pancreatic cancer than in normal tissues. Next, for the pancreatic cancer aspect, we combined the data with survival analysis of CD248 to obtain Kaplan-Meier curves (Fig. 2c). The results showed that the upregulation of CD248 leads to poor prognosis in PDCA. Therefore, CD248 can be considered as a proto-oncogene and a potential therapeutic target for PDCA, and CD248 expression can be used as a potential diagnostic indicator for PDCA.
E334K and R41C mutations of CD248 in PDCA
We then used cBioPortal to determine the type and frequency of CD248 mutations in PDCA based on sequencing data from patients in the TCGA database. We analyzed the mRNA transcripts and copy-number alterations of CD248 in 176 patients. The result shows that copy-number alterations include amplification, gain, diploid and shallow deletion, and diploid expression was the most abundant (Fig. 2d). In PDCA, the frequency of somatic mutations in CD248 was low at 1.1%. The two known mutation types, both of which were missense mutations and both replicated as diploid, were the E334K and R41C mutations (Fig. 2b).
PPI of CD248 in PDCA
To identify the biological interaction network of CD248 in PDCA, we searched for co-expression target genes related to CD248 in pancreatic cancer through cBioPortal in the TCGA database. A total of 104 related genes (Spearman's Correlation>0.7, p<0.05) were selected to obtain the gene co-expression network for protein interaction analysis (Fig. 3).
Enrichment analysis: CD248 was related to tumor angiogenesis
Then, in order to analyze the biological classification of co-expressed genes, we used String, Cytoscape and other tools to perform GO function and KEGG pathway enrichment analysis. The concentration of GO biological process (BP) is mainly related to regulation of biological process, multicellular organismal process, developmental process, anatomical structure development, multicellular organism development, system development, positive regulation of biological process, etc. (Fig. 4a). The enriched GO molecular function (MF) items are mainly associated with ion binding, protein binding, metal ion binding, protein-containing complex binding, signaling receptor binding, structural molecule activity, calcium ion binding, extracellular matrix structural constituent and integrin binding and so on are related to each other (Fig. 4b). The enriched GO celluar component (CC) items are mainly related to extracellular region, intracellular organelle lumen, endomembrane system, extracellular matrix, endoplasmic reticulum, extracellular space, endoplasmic reticulum lumen and supramolecular polymer, etc. (Fig. 4c). The enrichment of KEGG pathways includes protein digestion and absorption, focal adhesion, human papillomavirus infection, PI3K-Akt signaling pathway, ECM-receptor interaction, relaxin signaling pathway, AGE-RAGE signaling pathway in Diabetic complications, microRNAs in cancer, proteoglycans in cancer, etc. (Fig. 4d). All in all, these data suggest that CD248 is involved in many fundamental life processes and may play roles in regulating signal transduction, protein binding and vascular development in tumor.
LncRNA-miRNA-gene network axes of CD248 in PDCA
Finally, to further explore the target of CD248 in PDCA, we analyzed the miRNAs negatively associated with CD248 in the TCGA database and the lncRNAs negatively associated with the corresponding miRNAs. First and foremost, we screened the following 70 miRNAs, which were negatively correlated with the expression of CD248 in pancreatic cancer, by GraphPad Prism analysis (Table 1). Then, we used ENCORI and SangerBox to analyze the differences in expression of these 70 miRNAs in normal tissues and pancreatic cancer and their relationship with pancreatic cancer prognosis. We obtained one miRNA (hsa-miR-200c-3p) with differential expression in normal tissues and in tumor tissues and its high expression predicted a better prognosis (Fig. 5a-c). We then used ENCORI to analyze the lncRNAs associated with miRNAs and obtained 15 lncRNAs (AC008040.1, AC015813.6, AC025569.1, AC027279.4, AC055822.1, AL049796.1, AL353796.1, AP001486.2, HELLPAR, LINC01140, LINC01303, MSC-AS1, OIP5-AS1, RRN3P2, ZNF433-AS1) were negatively correlated with hsa-miR-200c-3p, but only three of the above 15 lncRNAs (AC008040.1, AC055822.1, RRN3P2) were differentially expressed in normal tissues versus in pancreatic cancer tissues (Fig. 5d-i). Therefore, we obtained three new lncRNA-miRNA-gene network axes, namely AC008040.1-hsa-miR-200c-3p-CD248 axis, AC055822.1-hsa-miR-200c-3p-CD248 axis and RRN3P2-hsa-miR-200c-3p- CD248 axis, are potential pathways and potential therapeutic targets for regulating pancreatic cancer progression.