Screening of DEGSA total of 24 samples from the GEO dataset (GSE17251) were used for identifying the DEGs using the R version (4.3.2). Overall, 96 DEGs were upregulated and 10 were downregulated with the fold change threshold of 0.5 in the treated group compared to the untreated group (Fig. 1A). In addition, the top 20 upregulated genes related to lipid metabolism in HCC are represented in a heatmap plot (Fig. 1B).Figure 1Screening of the differentially expressed genes among treated and untreated groups with PPARα agonist (WY-14643). (A) Volcanic plot for differently expressed genes (DEGS) in GSE17251. (B) Heat map of top 20 differently expressed genes (DEGS) in GSE17251.Weighted correlation network analysis (WGCNA) of DEGSWGCNA analysis was carried out to identify the gene clusters related to lipid metabolism. Lipid metabolism-Related Genes were extracted from the Molecular Signatures Databasev7.5.1 (MSigDB). The soft threshold value for the dataset was chosen with a cutoff R2 value of 0.85. A total of eighteen co-expressed gene modules were identified. Next different color was assigned to each module. The grey60 module has the fewest genes with 68 genes, while the turquoise module has been detected to be the largest module as well Table 3. (Fig. 2A–D).Table 3 Different Modules and their genes count.Figure 2Identification of PPARα agonist (WY-14643) gene modules in GSE17251 using WGCNA analysis. (A) Hierarchical clustering of samples to detect the outlier samples. (B) Hierarchical clustering of modules and heatmap of traits. (C) Scale independence and mean connectivity. (D) Dendrogram of GSE17251 dataset according to (1-TOM) matrix.The WGCNA network construction to identify PPARα treatment-related moduleAll candidate genes from datasets related to PPARα treatment were subjected to WGCNA analysis to determine whether or not the potential gene modules were connected to PPARα treatment. Among all the distinct modules that have been identified (Table 3), two modules (brown and pink modules) had a significant negative correlation with PPARα treatment based on the heatmap plot of the adjacencies (Fig. 3A–C).Figure 3Detecting the PPARα agonist associated modules with WGCNA. (A) Module-trait heatmap to identify the correlation between the gene module and the PPARα agonist treatment in the GSE17251 dataset, while each module contains the corresponding correlation coefficient and P-value. (B) The adjacency heatmap of the trait for each module. (C) Scatter plot of module pink and brown with negative correlation with PPARα agonist treatment.KEGG pathway and GO enrichment analysis of candidate hub genesOverlapping genes among the pink and brown modules and the DEGS (P-value < 0.05) were detected by the Vann diagram. Based on our GO enrichment and KEGG pathway results overlapping hub genes were primarily associated with PPAR signaling, Fatty acid metabolism, uptake, transport, and degradation (Fig. 4A–D).Figure 4Screening of the candidate hub genes. (A) 140 overlapping hub genes were detected by the Venn diagram. (B–D) GO enrichment analysis and KEGG pathway analysis of candidate hub genes. PM = pink module, BM = brown module, DEG = differently expressed genes.Top PPI network clusters and validation of final hub genes expression levelsThe PPI network of the overlapping genes was detected by the STRING online tool and the result was inserted into Cytoscape software version (3.7.1)’s MCODE plugin to find the hub clusters. Next, the Cytohubba plugin was used to detect the top 10 hub genes with the highest maximal clique centrality (MCC) (Fig. 5A). The expression levels of each hub gene were validated using expression data using GEPIA2 and UALCAN database. According to the expression plot, the expression of five genes out of 10 genes was significantly meaningful in HCC samples (P Value < 00.01) (Fig. 5B–G). The biological roles of selected hub genes were examined using a functional enrichment analysis, GSEA. Based on our results, selected DEGs are significantly related to alcohol liver disease, and PI3k-AKT signaling (Fig. 5H). The Go enrichment analysis of lipid metabolism-related hub genes indicates that the selected hub genes are mainly enriched in fatty acid metabolism, degradation, and transport (Fig. 5I).Figure 5Validation of candidate hub genes. (A) Top 10 hub genes by Cytohubba plugin. (B) Expression plot of top 10 hub genes through GEPIA2 database (C–G). Box plot expression of 5 hub genes in HCC with meaningful P Value < 00.01. (H) GESA analysis of the hub genes using cluster Profiler package (I) GO enrichment analysis of lipid metabolism-related Hub genes.Hub genes validation and TF-miRNA regulatory network evaluationThe q-PCR result of each hub gene was by the TCGA data indicating the over-expression of ACSL3 and downregulation of (ACSL1, ACOX1, CPT2, and SLC27A2) in HepG2 cells compared to normal control samples (Fig. 6A). Network Analyst database provided evidence of the relationship between transcription factors (TFs), miRNAs, and final hub genes (Fig. 6B). Information on transcription factors and gene targets obtained from ENCODE ChIP-seq data. This algorithm uses only the peak intensity signal < 500 and the predicted regulatory potential score < 1 (BETA Minus). Next, cytoscape 3.8 was used to plot the network after the nodes were filtered according to their degree.Figure 6Validation of hub gene expression and visualization of TF-miRNA network for each hub gene. (A) The qRT-PCR evaluation of hub genes mRNA levels in HEPG2 cells compared to the normal control cell line. (B) Diagram of core miRNA and TF regulatory network of hub genes filtered by degree, where red and orange represent genes; yellow represents predicted miRNAs; light-orange represents TF. The data represent the means ± SD. **P < 0.01, *** < 0.001 vs. control.The significant association of ACSL3 with HCC progressionThe function of each hub gene on clinical outcomes was detected, however, only ACSL3 was significantly associated with different pathways activity, worst clinical outcomes including stages (I-IV), and lower Overall survival rate (OS) in HCC patients (HR = 1.61, P = 0.011) (Fig. 7A–C). Moreover, the protein levels of immunohistochemistry (IHC) staining obtained from the Human Protein Atlas (HPA) database showed that the expression of the ACSL3 was significantly higher in tumor tissues than in normal tissues (Fig. 7D), which was consistent with that at the transcriptional level. The higher expression of ACSL3 was also positively associated with immune markers such as T cells, NKT cells, CD4, and monocytes and negatively associated with NK, TH2, and CD4 naive in liver cancer patients (Fig. 7E). ACSL3 expression is also associated with different biological pathways (Fig. 7F). The correlation of ACSL3 over-expression and drug sensitivity also indicates that ACSL3 is positively associated with different anti-cancer drugs such as navitoclax (Fig. 7G).Figure 7Higher ACSL3 expression is associated with poor clinical outcomes. (A) The over-expression of ACSL3 was related to poor OS. (B) KM plot of ACSL3 in liver cancer among 364 patients from km plot database. (C) Association of ASCL3 with poor stage in LIHC tumor. (D) The IHC staining of ACSL3 levels in normal human tissues compared to HCC tissues using the human protein atlas database. (E–G) Association of ACSL3 expression with immune markers, biological pathways, and drug sensitivity.ACSL3 is a potential target for chemotherapyWe analyzed the ACSL3 for immune checkpoint inhibitors (ICIs) chemotherapy in cancers. Since, ROC analysis is used in diagnostic test evaluation, the ROC plotters of the ACSL3 with strong correlation to resistance especially chemotherapy in all samples (n = 167) were evaluated as shown in (Fig. 8A,B). Based on the ROC analysis, we find that ACSL3 has relatively high AUC and low P-value with anti-PD1 (P = 3.8 e − 3, AUC = 0.744) and anti-PD-L1 therapy (P = 3.5e − 2, AUC = 0.589) which can suggest ACSL3 as a potential target of any chemotherapy. The qRT-PCR results and western blot results of ACSL3 expression in HUH7 and HepG2 cell lines also indicated a higher ACSL3 level in compared to the LO2 cell line (Fig. 8C,D).Figure 8ACSL3 is a potential biomarker. (A,B) Box-plot and ROC curve of ACSL3 in liver cancer. (C,D) The q-RCR and western blot results of ACSL3 expression in HepG2 and HUH7 cells compared to LO2 cells.
