Low expression of SLC17A2 in HCC. Firstly, we aimed to determine the distinctions in SLC17A2 expression between normal and tumor tissues of pan-cancer. The TIMER “DiffExp” module revealed that SLC17A2 expression was relatively higher in renal clear cell carcinoma and renal papillary cell carcinoma. In addition, SLC17A2 expression was notably lower in cholangiocarcinoma, renal chromophobe cell cancer and liver HCC tissues than in normal tissues (p < 0.001, Fig. 2A).
In addition, we also used the Oncomine database to explore the differential expression of SLC17A2 in 7 studies, and found that SLC17A2 expression was significantly lower in HCC tissues than in normal tissues (Fig. 2B).
Furthermore, 377 patients in the TCGA database were classified according to the following clinical pathology factors: age, sex, pathological stage and grade, survival time and disease state. The TCGA data showed that the expression of SLC17A2 was remarkably lower in HCC tissues than in paracancerous tissues (Fig. 2C, P < 0.001). Then GEO datasets from the National Center for Biotechnology Information (NCBI) were retrieved, and the RNA-seq data of SLC17A2 were obtained from the GSE45267 dataset, which collates 87 gene expression profiles of tissue samples from 61 patients: 46 primary HCC tissues and 41 paracancerous tissues. A t-test was performed on the GEO data and a boxplot was drawn (Fig. 2D, P < 0.0001). In addition, two other databases obtained from the GEO, GSE14520 (Fig. 2E, P < 0.0001) and GSE54236 (Fig. 2F, P < 0.01), were used for further verification. In conclusion, all the results confirmed that SLC17A2 was expressed at low levels in HCC.
Validation of SLC17A2 protein expression by immunohistochemical staining. The expression level of SLC17A2 protein in the cancer tissues and adjacent normal tissues of 32 patients with liver cancer was verified by immunohistochemistry (Fig. 3). The results showed that the expression level of SLC17A2 in cancer tissue was significantly lower than that in adjacent tissues (n = 60, P < 0.0001, Table I). In conclusion, all the results confirmed that SLC17A2 was expressed at low levels in HCC.
In addition, we mined cBioPortal data, which revealed that SLC17A2 expression was altered in 4% (16/400) of samples, including 14 cases of amplification, and 2 cases of a missense mutation (Fig. 4A). In summary, amplification is the most common type of SLC17A2 mutation in HCC.
Prognostic significance of SLC17A2 in HCC patients. By mining data in the TCGA and generating a Kaplan-Meier survival curve, which collated 344 patients to investigate the relationship between SLC17A2 expression and the prognosis of HCC patients, we concluded that patients with higher SLC17A2 mRNA expression had longer overall survival times than those with lower SLC17A2 mRNA expression (P = 0.0078, Fig. 4B). The same conclusion was draw by analyzing data from OncoLnc (P < 0.05, Fig. 4C) and GEPIA (P = 0.0059, Fig. 4D). In addition, Cox regression analysis was performed on the subgroups according to clinical-pathology factors, and the univariate Cox analysis indicated that SLC17A2, T-stage 3–4, M stage and TNM stage were significant factors influencing the overall survival of patients with liver cancer. In the multivariate analysis, the expression level of SLC17A2 was a significant independent prognostic biomarker for HCC (HR = 0.594, P = 0.01, Table II).
Potential diagnostic value of SLC17A2 in HCC patients. We explored the diagnostic value of SLC17A2 in HCC, and SLC17A2 expression was interrelated to HCC. In addition, the ROC curve was plotted and the area under the curve (AUC) was 0.708. The 95% CI was between 0.656 and 0.760 (P < 0.0001) (Fig. 4E), which revealed the diagnostic value of SLC17A2 in the HCC population.
Biological signaling pathways related to SLC17A2 according to GSEA. GSEA was used to predict the biological pathways related to SLC17A2. The first six enrichment pathways associated with SLC17A2 were fatty acid metabolism, valine leucine and isoleucine degradation, glycine serine and threonine metabolism, histidine metabolism, butanoate metabolism and peroxisome (Fig. 5A), suggesting that SLC17A2 is involved in many biological processes of HCC.
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analyses. First, the cBioPortal and UALCAN databases were screened, and 1142 genes (those with a Spearman's correlation coefficient < 0.3 or > 0.3 were removed) and 288 genes (those with a Pearson’s correlation coefficient < 0.3 or > 0.3 were removed) that were coexpressed with SLC17A2 were obtained, respectively, and intersected with a Venn diagram (http://bioinformatics.psb.ugent.be/webtools/Venn/); ultimately, 158 coexpressed genes were obtained (Fig. 4F).
GO and KEGG analyses of the 158 coexpressed genes were performed with the “ggplot2” and “clusterProfiler” packages of R programming. GO analysis indicated that the 158 genes are mainly involved in fatty acid and cellular amino acid metabolism and carboxylic acid and organic acid catabolism (biological processes) microbodies and peroxisomes (cellular components) and coenzyme binding and oxidoreductase activity (molecular functions) (Fig. 5B). The top 30 pathways analyzed are shown in Fig. 6. The intersecting genes mainly participate in the peroxisome, complement and coagulation cascade, carbohydrate, fatty acid and amino acid metabolism, cytochrome P450 and chemical carcinogenesis pathways.
Construction of the PPI network and selection of hub genes. After identifying the disease-related genes, the STRING database was used to acquire the PPI network. Then, the network we obtained was further processed through Cytoscape, and the unconnected nodes were removed (Fig. 7A-B). In addition, we used the cytoHubba plug-in of Cytoscape to screen the top 10 genes through the Closeness algorithm, and the genes are shown in Fig. 7C. The top 10 hub genes are as follows: CAT, APOB, CYP3A4, AOX1, EHHADH, MAPK14, HSD17B4, PLG, NR1H4, CYP2E1.
Overall survival analysis according to the top three hub genes. We used the GEPIA database to perform overall survival analysis. The analysis of the CAT, APOB and CYP3A4 suggested a significant correlation between the survival time and CAT (P < 0.05) and CYP3A4 (P < 0.05) expression (Fig. 7D-F).
SLC17A2 expression is related to immune cells in the hepatocellular cancer microenvironment. Tumor-infiltrating immune cells (TIICs) are a part of the complex microenvironment and are related to the development of tumors. We used the TIMER database to estimate SLC17A2 expression and the enrichment of TIICs, including B lymphocytes, CD4 + T cells, CD8 + T cells, regulatory T cells (Treg), dendritic cells, neutrophils and macrophages. We found that SLC17A2 expression was positively associated with the infiltration levels of CD4 + T cells (R = 0.153, p = 4.38e − 3), naive CD8 + T cells (R = 0.271, p = 3.1e − 7), and naive B cells (R = 0.128, p = 0.017) and negatively correlated with the levels of Treg cells (R = -0.171, p = 0.001) in HCC (Fig. 8). These results indicate that SLC17A2 plays a significant role in the immune infiltration of HCC.
Correlations between SLC17A2 and immune cell markers. Finally, to deepen our understanding of immunity, in addition to analyzing common types of immune cells, we used the GEPIA database to further explore the correlations between SLC17A2 and markers of immunocyte to identify subtypes of immune cells. We analyzed both innate immune cells such as macrophages, neutrophils, NK cells, and dendritic cells, and various functional adaptive immune cells, such as B cells, CD8 + T cells, Th1 cells, Th2 cells, Th17 cells and Tregs, in HCC (Table III). The results suggested that SLC17A2 expression is closely associated with most makers of various immune cell types in HCC.
Excitingly, we found that SLC17A2 expression was implicated in the levels of some gene marker sets of CD8 + T cells, Th1 cells, Th2 cells, Th17 cells, Tregs, B cells, neutrophils, macrophages and NK cells. More specifically, we found that ABT1, ZNF22, and ZEB1 of CD8 + T cells, DPP4 of Th1 cells, MICAL3 and BACH2 of B cells, SGMS1 of macrophages, CYP4F3, HIST1H2BC, and KIAA0329 of neutrophils, IL17A and STAT3 of Th17 cells, and CDC5L, ALDH1B1, and ARL6IP2 of NK cells are significant positively correlated with SLC17A2 expression in HCC (P < 0.05). We also discovered the significant negative correlations between marker genes of Th2 and Treg cells and SLC17A2 expression (P < 0.05), in accordance with the results from the TIMER database (Fig. 9). In conclusion, SLC17A2 may affect the development of HCC by regulating the infiltration of immune cells with different phenotypes and may explain the effect of SLC17A2 on the prognosis of HCC.