Prognostic potential and mechanism of SORT1 and its co‐expressed genes in hepatocellular carcinoma based on integrative analysis of multiple database

Abnormal SORT1 expression has been reported in various cancers. However, the expression and function of SORT1 in hepatocellular carcinoma (HCC) remain to be explored. This study aims to explore the expression and function of SORT1 and to identify its co‐expressed genes in HCC. Various gene expression databases were applied in our analysis. We found SORT1 was up‐regulated in HCC tumor tissues and high SORT1 expression level was associated with worse overall survival (OS). Co‐expressed genes with SORT1 and its potential regulators were explored using LinkedOmics. Functional network analysis of co‐expressed genes by Metascape revealed that they participated in aberrant lipid metabolism, AMPK signaling pathway, and PPAR signaling pathway which were all strongly linked to the pathogenesis of HCC. In addition, co‐expression genes were analyzed by Cytoscape to identify their hub genes, which included CYB5A, CYP2C9, CYP3A5, CYP4A11, and POR. The mRNA expression level of CYP2C9, CYP3A5, and CYP4A11 were down‐regulated in HCC tumor tissues via GEPIA. High hub genes expression level was associated with better OS and progression free survival (PFS) in HCC. The correlations between SORT1 and hub genes with cancer immune infiltrates were investigated by TIMER. Notably, SORT1 and hub genes expression was positively correlated with infiltrating levels of different immune cells. Our findings suggested that high SORT1 expression level predicted dismal prognosis in HCC and its possible mechanism was immune‐related.

smoking, and type 2 diabetes. 2 Obesity and type 2 diabetes are usually associated with abnormal lipid metabolism which is emerging as a new pathway to explain the mechanism of HCC. In highly proliferating tumor cells, enhanced adipogenesis provides building blocks, signaling molecules, and bioactive mediators to ensure growth, proliferation, and persistence. The progression of HCC is associated with complex metabolic reprogramming and immunosuppression driven by tumor associated macrophages (TAMs) and mediated by biologically active lipid mediators. 3 Existing targeted drugs show unsatisfactory efficacy because of complex clinical and biological behaviors of HCC and increasing drug resistance. 4 In addition, lack of effective prognostic biomarkers that are specific for tumor subtype and stage remains an obstacle in the current understanding of HCC. Thus, it is necessary to discover novel biomarkers and drug targets to predict the prognosis of HCC and to develop novel-targeted therapy.
SORT1 encodes a member of the vacuolar protein-sorting 10 (VPS10)-related sortilin family of proteins which were initially identified in searching for new lipoprotein receptors that resemble the low-density lipoprotein receptor (LDLR). 5 Sortilin is a 95 kDa protein receptor acting as part of the ectodomain of VPS10P. 6 It plays an important role in trafficking different proteins to cell surface or subcellular compartments such as lysosomes and endosomes. Several genome-wide association studies (GWAS) found the encoding locus of SORT1 at 1p13.3 was related to plasma levels of cholesterol and risk of myocardial infarction in humans indicating the potential role of SORT1 in systemic cholesterol homeostasis. [7][8][9] In recent years, the increasing number of researches indicated abnormal lipid metabolism was strongly linked to the pathogenesis of HCC. 10,11 However, the relationship between SORT1 and HCC remained unexplored and further analysis via bioinformatics is required.
This research aims to investigate the prognostic potential and mechanism of SORT1 and its co-expressed genes in HCC. Results showed that high SORT1 expression level was associated with inferior outcome and its possible mechanism included metabolic pathways and immune response in HCC patients. These findings shed light on risk stratification of HCC and targeted therapy development.

| HCCDB database
HCCDB is a public HCC gene expression analysis platform which contains up to about 4000 clinical samples from TCGA and GEO database. 12 It provides a comprehensive resolution of gene expression patterns in HCC patients. We analyzed the mRNA expression level of SORT1 via this platform.

| Oncomine database
Oncomine (https://www.oncomine.org/resource/login.html) is a useful database and data-mining platform. Currently, Oncomine contains 715 gene expression datasets comprising 86 733 samples that can be integrated to discover novel biomarkers for diagnosis and treatment of different diseases. 13 In this study, SORT1 gene was selected as our research object and its mRNA expression level was analyzed in HCC and normal liver tissue.

| UALCAN database
UALCAN (http://ualcan.path.uab.edu) provides comprehensive analysis of cancer OMICS data and additional information about selected genes by linking to various databases such as TCGA. 14 Using UALCAN we analyzed the expression level of SORT1 between HCC and normal liver tissues, and between sub-groups classified by stage, grade, and other clinical characteristics.

| GEPIA database
The Gene Expression Profiling Interactive Analysis (GEPIA) database (http://gepia.cancer-pku.cn/) was used for co-expression genes expression and survival analysis based on TCGA and the GTEx projects. 15

| LinkedOmics database
The LinkedOmics database (http://www.linkedomics.org/login.php) is a free-access database which can be used to analyze the relationship between target genes and their co-expression genes. 16 With this tool, SORT1 co-expression genes were exported and visualized.

| c-BioPortal database
The cBio Cancer Genomics Portal (http://cbioportal.org) is mainly used to explore cancer genomics. 17 Various genomic alterations of SORT1 and hub genes in HCC were analyzed using this tool.

| Kaplan-Meier plotter database
The Kaplan-Meier plotter (http://kmplot.com/analysis/) assesses the effects of more than 50 000 genes on survival from GEO and TCGA databases. 18 Here, this tool was used to explore the effects of SORT1 and hub genes on survival in HCC.

| TIMER database
TIMER is a free platform for systematic exploration of immune infiltrates from TCGA database (https://cistrome.shinyapps.io/timer/). 19 Here, we analyzed the relationship of SORT1 and hub genes with tumor-infiltrating immune cells.

| Metascape database
Metascape (http://metascape.org/) is an online gene annotation tool that can be applied for enrichment analysis. 20 In this study, to explore the function of co-expressed genes, biological process (BP), and KEGG pathway enrichment analysis were performed using Metascape. In addition, enrichment-bubble diagram was plotted on http://www. bioinformatics.com.cn.

| PPI network construction and module analysis
Search Tool for the Retrieval of Interacting Genes (STRING; http:// string-db.org; version 11.5) was used to predict protein interaction (PPI) network. 21 SORT1 associated PPI network was constructed and displayed via Cytoscape (version 3.8.2) plug-in APP Molecular Complex Detection (MCODE). The critical genes in top one module were shown in this study.

| Statistical analysis
The p < .05 was defined as statistically significant between groups with different expression level of SORT1 with t-test. Overall survival (OS) and progression-free survival (PFS) of HCC patients were explored using Kaplan-Meier curves. Significance in survival time difference by log-rank test was defined as p < .05.

| SORT1 co-expression network construction, enrichment pathway analysis, PPI network construction, and module analysis in HCC
To explore the molecular mechanism of SORT1 in HCC, we used Linke-dOmics tool to analyze SORT1 co-expression networks. Exactly 2022 genes were shown to be positively related with SORT1, while 842 genes were shown to be significantly negatively related (FDR < 0.05) ( Figure 4A).

| Expression and prognostic value of five hub genes in HCC and their correlation with SORT1
We evaluated the transcription levels of CYB5A, CYP2C9, CYP3A5, CYP4A11, and POR via GEPIA database ( Figure 6A-E). Results

| Association of SORT1 and hub genes with immune infiltration level in HCC
We used TIMER to explore the relationship between SORT1 and hub genes with immune infiltration level in HCC. Results showed that   22 In addition, the elevated expression of SORT1 was found to predict worse prognosis in B acute lymphoblastic leukemia. 23 Researches which directly linked SORT1 and HCC were scarce and to clarify the signaling events accompanying abnormal SORT1 expression, we explored its coexpression network. Our results showed that SORT1 co-expression genes participate in monocarboxylic acid metabolic process, cellular carbohydrate metabolic process, and regulation of transforming growth factor beta receptor signaling pathway. Among them, the results of biological process (BP) showed that they were significantly enriched in oxidation-reduction process, long-chain fatty acid metabolic process, WNT signaling pathway, and planar cell polarity pathway. KEGG analysis showed that metabolic pathways, PPAR signaling pathway, and AMPK signaling pathway were significantly enriched. These pathways are strongly linked to the pathogenesis of HCC in previous studies. [24][25][26][27] Abnormal SORT1 expression probably influences various signaling pathways and biochemistry metabolism processes in HCC via its co-expression network. To conclude, our findings indicate a complex regulatory network of SORT1 in HCC, more in vitro and in vivo experiments are needed to validate the above pathways and the regulatory mechanism of SORT1 and its co-expressed genes.
We also found a strong negative correlation between SORT1 and its hub genes including CYB5A, CYP2C9, CYP3A5, CYP4A11, and POR. Analysis of transcriptome from GEPIA database confirmed that the mRNA expression levels of CYP2C9, CYP3A5, and CYP4A11 were significantly higher in adjacent normal tissues than HCC tissues, which were consistent with previous studies. [28][29][30] Moreover, high

| CONCLUSION
With the assistance of databases analysis, we found increased SORT1 expression level in HCC regardless of disease stage, tumor pathological grade, gender, race, or age and predicted worse survival. Enrichment pathway analysis of SORT1 co-expression genes suggested that aberrant lipid metabolism, AMPK signaling pathway, PPAR signaling pathway, regulation of transforming growth factor beta receptor signaling pathway, WNT signaling pathway, and planar cell polarity pathway were strongly linked to the pathogenesis of HCC. For mechanism study, SORT1

ACKNOWLEDGMENTS
We thank all developers of public databases and softwares.