3.1 The increased expression of SERINC2 in STAD
We initially analyzed the differential expression of SERINC2 in multiple malignancies and normal tissues via the RNA-seq data from TCGA. As shown in Fig. 1A, compared with adjacent the normal tissues, the expression level of SERINC2 was higher in BLCA (bladder urothelial carcinoma), BRCA (breast invasive carcinoma), CHOL (cholangiocarcinoma), ESCA (esophageal carcinoma), HNSC (head and neck squamous cell carcinoma), LIHC (liver hepatocellular carcinoma), LUAD (lung adenocarcinoma), LUSC (lung squamous cell carcinoma), STAD (stomach adenocarcinoma), THCA (thyroid carcinoma) and UCEC (uterine corpus endometrial carcinoma). Conversely, in COAD (colon adenocarcinoma), KICH (kidney chromophobe), KIRC (kidney renal clear cell carcinoma) and READ (rectum adenocarcinoma), the expression level of SERINC2 was significantly lower than that in normal tissues. We then used the GEPIA database to confirm the high expression levels of SERINC2 in STAD. Results revealed that the mRNA level of SERINC2 was remarkably higher in STAD than that in normal tissues (Fig. 1B and C).
3.2 Association of SERINC2 expression and mutation with the prognosis of STAD patients
We further explored whether the survival of STAD patients was affected by SERINC2. Kaplan-Meier Plotter tools were used to analyze the correlation between SERINC2 mRNA levels and the survival of patients with STAD. As shown in Fig. 2A-C, the patients with higher SERINC2 expression level had worse overall survival (OS), post-progression survival (PPS) and progression free survival (PFS) than those with lower SERINC2 level (P < 0.05). Generally, STAD was negative correlated with mRNA expression of SERINC2 in terms of PPS, PFS and OS. In order to better understand the relevance and potential mechanism of SERINC2 in STAD, we explored the relationship between the SERINC2 expression and clinical characteristics of STAD patients in the Kaplan-Meier plotter database. As shown in table1, the high SERINC2 expression level was significantly associated with worse OS, PPS and PFS in patient gender, stage, T stage, N stage, M stage and Lauren classification, with the exception of T stage 4, two types of Lauren classification (P < 0.05). There are significantly association between high SERINC2 expression and worse OS and PPS in stage N1 and N2. In addition, high SERINC2 expression was significantly related with worse PPS in stage N3. High SERINC2 expression was also significantly related with worse OS, PPS and PFS in stage N1 + N2 + N3. Generally, the above results indicated that SERINC2 expression level affected the prognosis of STAD patients with lymph node metastasis.
Occurrence and prognosis of most tumors are related to gene alterations. Here, we evaluated the association of SERINC2 mutation with the prognosis of STAD patients. Gene alterations in SERINC2 were found to occur in a total of 14 out of 478 sequenced cases in the data obtained from the OncoPrint schematic of cBioPortal (Fig. 2D). The information of the mutation sites, mutation types and case number is shown in the mutation diagram (Fig. 2E). In this diagram, different colors represent the corresponding mutation types. Overall, we found seven mutations of SERINC2, including four missense mutations, one splice mutation, one inframe mutations and one truncating mutation, in sequenced STAD patients. The mutation sites all occurred in Serinc. However, the KM curves analysis of the prognostic value between the SERINC2 altered group and the unaltered group showed no significance (Fig. 2F). These results suggested that the effect of SERINC2 on the prognosis of STAD was not due to its mutation.
3.3 SERINC2 expression is correlated with immune infiltration level in STAD patients
Given the key role of immune infiltration in the tumor development in diverse types of cancer, we further evaluated the relationship between SERINC2 expression and immune cell infiltration levels by using TIMER. Our results showed that the expression of SERINC2 was negatively correlated with the infiltration levels of CD8+ T cells (r = -0.25, P = 1.15E-6), CD4 + T cells (r = -0.212, P = 4.25E-5), macrophages (r = -0.288, P = 1.62E-8), neutrophils (r = -0.273, P = 8.91E-8) and dendritic cells (r = -0.314, P = 5.82E-10) in STAD (Fig. 3A). Meanwhile, we analyzed the correlation between SERINC2 copy number variation (CNV) and immune infiltration level using the somatic copy number alteration (SCNA) module in the TIMER database. The results revealed a closely association between SERINC2 CNV with the degree of infiltration of B cell, CD4+ T cell, macrophages, neutrophils and dendritic cell (Fig. 3B). Moreover, the correlation between SERINC2 expression and gene markers of different immune cell was analyzed through TIMER. As listed in table S1, multiple immune cell markers were significantly correlated with SERINC2 expression. The CD8+ T cell marker, the general T cell marker, the monocyte marker, the M2 Macrophage marker and the T cell exhaustion marker had strong correlations with SERINC2 expression. These findings strongly suggest that SERINC2 plays a specific role in immune infiltration in STAD.
3.4 Co-expression genes correlated with SERINC2 in STAD
To further explore the potential biological function of SERINC2 in STAD, we analyzed the co-expressed genes associated with SERINC2 in 415 STAD cases by mining the LinkedOmics database. Results showed that 4811 genes on the right of 0, represented by red dots in Fig. 4A, were positively associated with SERINC2. Moreover, there were 6382 genes, represented by dark green dots, having a negative correlation with SERINC2 (p < 0.05). The top 50 significant genes positively and negatively correlated with SERINC2 were shown by the heat map, respectively (Fig. 4B and C). The details of co-expressed genes is summarized in Table S2. Among the positive co-expression genes, DDR1, ELMO3, PTK6 were chose to confirm the correlation with SERINC2 through GEPIA database. SERINC2 expression was strongly correlated with DDR1 (r = 0.5362, p = 2.933e-32), ELMO3 (r = 0.529, p = 2.648e-31) and PTK6 (r = 0.5272, p = 4.575e-31) (Fig. 4D-F).
We further performed the GO and KEGG functional enrichment analysis of these co-expressed genes by GSEA in LinkedOmics. Biological process (BP) enrichment analysis showed that the co-expressed genes of SERINC2 mainly participated in epidermis development, cellular aldehyde metabolic process and isoprenoid metabolic process (Fig. 4G). Cellular component (CC) enrichment analysis showed that the co-expressed genes of SERINC2 mainly participated in cornified envelope, mitochondrial protein complex and respiratory chain (Fig. 4H). Molecular function (MF) enrichment analysis showed that the co-expressed genes mainly participated in oxidoreductase activity acting on CH-OH group of donors and a heme group of donors and electron transfer activity (Fig. 4I). KEGG pathway analysis indicated these genes were significantly correlated with immune signaling pathway and responses, including primary immunodeficiency, antigen processing and presentation, T cell receptor signaling pathway and natural killer cell mediated cytotoxicity (Fig. 4J). These results suggest that SERINC2 expression has a broad effect on the global transcriptome of STAD and immune signaling pathway.
3.5 The PPI network of co-expression gene was established
A protein-protein network was built by using the top 100 significantly positive co-expressed genes of SERINC2 in STRING database. As shown in Fig. 5A, PPI network was composed of 56 nodes and 76 edges. After using MCODE plug-in, S100A14, RAB25 and FXYD3, highlighted in yellow, was identified as the hub genes. We further performed survival analysis of the 3 hub genes using the Kaplan-Meier Plotter database. The expression level of RAB25 was significantly related to the OS of STAD. STAD patients with high expression of RAB25 had poor prognosis (Fig. 5B). Moreover, RAB25 was significantly up-regulated in STAD tissue (Fig. 5D). Based on the results, we speculated that the association between SERINC2 and RAB25 is a contributing factor for its prognostic value in STAD.