Bioinformatics analysis of hepatic fibrosis based on hepatic stellate cells

Yang Zheng, Jia-Hui Wang, Jia-Ru Wang, Tie-Jian ZhaoDepartment of Medicine, Faculty of Chinese Medicine Science Guangxi University of Chinese Medicine, Nanning 53001, China.

2College of Nursing, Guangdong Medical University, Dongguan 523000, China. 3Department of Physiology, College of Basic Medicine, Guangxi University of Chinese Medicine, Nanning 530021, China.

# Yang Zheng and Jia-Hui Wang are the co-first authors of this paper.

Abstract Objective: To analyze differentially expressed genes in human hepatic stellate cells (HSCs) based on data from the GEO database and to identify important target genes for hepatic fibrosis (HF). Methods: In GEO database,microarray GSE11954 of the GEO database was used to obtain data on differential gene expression in human HSCs and was analyzed using GEO2R, using a P value of ＜ 0. 01 and log2FC value of ≥ 2 for the screening. The genes were input into the DAVID database for enrichment analysis of genes and pathways, followed by protein interaction analysis and module analysis. The results were compared with the results found through text mining.Results: Two hundred sixty two differentially expressed genes (DEGs) were identified. The results of gene bulk enrichment showed that the functional molecules encoded by the DEGS were mainly located in the cytoplasm,extracellular matrix and nucleosome, while the molecular functions were mainly related to "regulating actin binding", "protein kinase binding" and "kinase activity". The biological processes they were found to be involved in "regulating cell division", "immune response" and "collagen decomposition reaction". KEGG signaling pathway analysis found that they were mainly involved in "cell cycle signaling pathway", "ECM receptor interaction signaling pathway", "p53 signaling pathway" and "FOXO signaling pathway". Text mining results suggested that MMP1 and ETV6 are potential molecular targets for HF therapy. Conclusion: The results of bioinformatics analysis identified targets and signaling pathways involved in the pathogenesis of HF, but these require further experimental verification.

Keywords: Hepatic stellate cells, Liver fibrosis, Bioinformatics, Differentially expressed genes, Gene enrichment analysis

Background

Hepatic fibrosis (HF) is a disease in which the liver is damaged due to various physical and chemical factors,resulting in inflammation, necrosis of hepatocytes and excessive deposition of extracellular matrix (ECM) in the necrotic zone. Some studies have shown that the activation and transformation of hepatic stellate cell(HSC) into myofibroblast is an important link between the genesis and development of HF, and that myofibroblast is the main source of ECM. HF slowly progresses from hepatitis B to liver cirrhosis. The intermediate pathological processes of liver cancer transformation are of great significance for the treatment and reversal of HF [1-3]. HSCs, as a cell model of liver fibrosis, have become the focus of attention [4-6]. China is a country that is greatly affected by liver disease, and HF results in a large disease burden for patients with liver disease [7].Although some progress has been made in the study of the etiology, diagnosis and treatment of HF, the current understanding of the molecular mechanism of HF is not deep enough, leading to an inability to formulate better treatment strategies for patients. Therefore, it is still of great clinical significance to further study and determine the therapeutic targets of HF. The GEO database [8] is an online database for species specific gene expression. In this study, the data set GSE11954[9], which included activated HSC samples and senescent HSC samples, was obtained from the GEO database. Through the analysis and comparison between the two sets of sample data, important targets of HF were identified. Gene function enrichment analysis, protein-protein interaction (PPI) network and module analysis were performed. Then, text mining of therapeutic drugs and their target genes was carried out and the results were preliminarily verified. The results of this study can provide us with a preliminary understanding of the molecular mechanism of HF and provide reliable HF target genes for use in further experiments.

Materials and methods

Gene chip data

GSE11954 gene expression data derived from GEO database (http://www.ncbi.nlm.nih.gov/geo) was of an expression profiling by array study type performed on a GPL570 platform (Affymetrix Human Genome U133 Plus 2.0 Array). The GSE11954 data set included data on human HSCs, in which 2 cases were activated HSCs (n = 2) and 2 cases are senescent HSCs (n = 2).

Identifying DEGs

Morpheus software (http://software.broadinstitute.org/morpheus/) was used to analyze the gene chip data.The differentially expressed genes were analyzed and screened used the online analysis tool, GEO2R [10].The screening conditions were set as follows: a P value of ＜ 0.01 and log2FC value of ≥ 2. Omicshare(http://www.omicshare.com/tools/index.phpl) software was used to make volcanic map and to observe the differentially screened genes.

Enrichment Analysis of Differential Gene function and Pathway

The DAVID [11] database (http://david.ncifcrf.gov/)was used to annotate the differentially expressed genes.The enrichment of gene function included molecular functions, biological processes and cellular components [12]. The enrichment analysis of the signaling pathway mainly used the KEGG [13]database (http:// www.genome.jp/), a database containing KEGG signaling pathways that is used to check the pathways involved with a gene cluster and its related functions.

The Establishment and Module Analysis of PPI Network

Using the String [14] database(http://www.string-db.org/), the association between known and predicted proteins can be retrieved and used to predict PPI information. The differentially expressed genes were imported into the String database to obtain data of differential gene interactions.Cytoscape 3.5.0 [15] (http://www.cytoscape.org/) was used to show the relationship between differential genes. The resulting PPI network was analyzed in modules, and the Mcode [16] clustering algorithm in Cytoscape was used to analyze each module. Modular analysis can be used to identify other linked gene groups. In addition, the results of the modular analysis were analyzed to identify gene functions and enrichment pathways.

Text mining

"Liver fibrosis" was used as a keyword on the treatment target database [17] (http://db.idrblab.org/ttd/)to search for drugs commonly used for the treatment of liver fibrosis. The Drug Gene Interaction Database [18](DGIdb; http://dgidb.genome.wustl.edu/) can be used to identify molecular targets involved in drug action,and these known molecular targets may be used to compared with that of the screened genes.

Figure 2 Volcano diagram of all genes. Red represents up-regulated genes, green represents down-regulated genes, and black represents unregulated genes

Results

Screening results

The heat map created using Morpheus software was used to observe the expression of all genes, as shown in Figure 1. The difference in gene expression between activated hepatic stellate cells and senescent hepatic stellate cells was identified. Two hundred and sixty two genes were screened using the online analysis tool,GEO2R, of which 119 genes were upregulated and 143 genes downregulated, as shown in Figure 2.

Gene function enrichment analysis

Most of the selected genes were found to be located in the cytoplasm, extracellular matrix and nucleosome.These genes play a molecular role through the regulation of actin binding, protein kinase binding,kinase activity, protein binding and receptor tyrosine kinase binding. In addition, these genes may be involved in the pathogenesis of HF through the regulation of cell division, immune response, collagen decomposition and extracellular matrix tissue. The results of gene function enrichment analysis are shown in Figure 3.

Figure 3 The first ten gene function enrichment in biological processes. the differentially expressed genes between activated hepatic stellate cells and aging hepatic stellate cells were analyzed in molecular function,cellular composition.

KEGG signaling path analysis

The screened genes were found to be involved in the cell cycle signaling pathway, viral carcinogenic signaling pathway, ECM receptor interaction signaling pathway, p53 signaling pathway, FOXO signaling pathway, complement and coagulation cascade signaling pathway, the pathway of protein digestion and absorption signaling. The results of the KEGG analysis are shown in Figure 4.

Figure 4 KEGG signaling pathway analysis. Differentially expressed genes between activated hepatic stellate cells and aging hepatic stellate cells,.

Analysis of PPI Network and Modules

The PPI network constructed in Cytoscape was found to include 181 nodes and 2,294 interactions, as shown in Figure 5. The genes with more interactions are known as central genes and may be involved in HF.Among them, DNA topoisomerase Ⅱ α (TOP2A),cyclin dependent kinase 1 (CDK1), cyclin B1(CCNB1), cyclin A (CCNA2), cyclin B2 (CCNB2),BUB1 mitotic checkpoint serine/threonine kinase(BUB1), cell division cycle 20 (CDC20), actin family member 20A (KIF20A), PDZ binding kinase (PBK),centrosomal protein 55 (CEP55) were identified as the top ten. The Mcode analysis identified seven modules,as shown in Figure 5. The first three modules were analyzed for path enrichment. The signaling pathways enriched by these three modules may be involved in regulating the process of HF genesis, as shown in Table 1.

Figure 5 The protein-protein interaction network constructed by Cytoscape. The size of the point is proportional to the interaction. Red dots indicate up-regulation, blue dots indicate down-regulation, and gray lines indicate protein interactions.

Figure 6 Common drugs and target genes for treatment of liver fibrosis. Red represents disease, yellow represents medicine, blue target gene.

Common therapeutic drugs and their targets

Commonly used drugs for HF are Amikacin,Cri20tinib and Ribavirin, and their target genes were identified to include TP53, EGFR, BCL2, MMP1,ADK, SMAD4 and EML4, using the DGIDB database,as shown in Figure 6.

Table 1 Enriched pathways of modules 1-3.

Discussion

The liver usually undergoes normal regeneration after liver damage. However, when liver injury and inflammation continue and progress, the liver is not able to regenerate normally, and this leads to fibrosis.Further development of liver fibrosis leads to cirrhosis,where liver cells do not function normally due to the formation of fibrous scars and regenerative nodules,resulting in decreased hepatic blood supply [19-20].The etiology of HF in any chronic liver injury involves hepatitis B and hepatitis C, alcohol consumption, fatty liver, cholestasis and autoimmune hepatitis, while the etiology of HF includes numerous factors and its pathogenesis is extremely complex [21]. Various inflammatory and fibroblast pathways are involved in the activation of HSC. The activation of HSC is the central link and the main mechanism of HF [22-23].Therefore, this study is based on the activation and aging of differences in HSC expression between the entry points. This study attempted to identify important target genes of HF, and 262 differentially expressed genes were identified to further clarify the regulatory role of these genes in HF.

Using gene enrichment analysis, cellular components showed that these differentially expressed genes were mainly located in the cytoplasm,extracellular matrix and nucleosome. In terms of molecular functions, these differentially expressed genes were found to play a molecular role in regulating actin binding, protein kinase binding, kinase activity,protein binding and receptor tyrosine kinase binding.In terms of biological processes, these differentially expressed genes were found to be involved in regulating cell division, immune response, collagen decomposition and extracellular matrix tissue. The KEGG analysis showed that these differentially expressed genes are involved in cell cycle signaling pathways, viral carcinogenic signaling pathways, ECM receptor interaction signaling pathway, p53 signaling pathway, FOXO signaling pathway, complement and coagulation cascade signaling pathway, protein digestion and absorption signaling pathway. It has been reported that the p53 signaling pathway can promote hepatic fibrosis by inducing apoptosis of hepatocytes,and the level of p53 in patients with hepatic fibrosis has been found to be significantly elevated [24-25].The FOXO family represents a set of transcription factors. These transcription factors are necessary for many stress-related transcriptional processes, including antioxidant responses, glycometabolism, cell cycle control, apoptosis and autophagy. FOXO molecules,especially FOXO1 and FOXO3, play an important role in nonalcoholic fatty liver disease. The deletion of the FOXO4 gene may aggravate hepatic steatosis and liver injury induced by diet, suggesting that FOXO exerts a protective effect on the liver [26-27]. Studies have shown that ITGA6 and CD44 molecules in the ECM receptor interaction signaling pathway play an important role in regulating the proliferation and invasion of cancer cells [28].

A PPI network of differences was constructed, in which TOP2A, CDK1, CCNB1, CCNA2, CCNB2,BUB1, CDC20, KIF20A, PBK and CEP55 were identified as the top 10 pivotal genes. TOP2A has been used as a target for cancer treatment and a biomarker for predicting responses. In addition, the expression of TOP2A is a valuable prognostic marker for the progression and recurrence of small cell lung, ovarian,colon, breast, prostate and nasopharyngeal carcinomas[29-32]. CDK1 is a typical kinase and a central regulatory factor in the G2 phase and mitosis process.The liver specific deletion of CDK1 is well tolerated,and liver regeneration after partial hepatectomy proceeds unimpaired. This suggests that regeneration can be driven by undivided cell growth. The absence of CDK1 does not affect the progression of the S phase[33]. It has been found that under conditions of natural killer T cell deficiency, the expression of CCNB1 is decreased and liver regeneration is inhibited [34]. The newly discovered FXR/mir-22/CCNA2 pathway may be a new mechanism that could explain the anti-proliferation effect of FXR. CCNA2 is the target gene of mir-22. FXR regulates the expression of mir-22 in order to regulate the induction of CCNA2 in tumors [35]. Several studies have shown that the expression of CCNB2 in tumor tissue directly affects the invasion and metastasis of the tumor tissue, as well as its prognosis [36-37]. Mir-490-5p can regulate the transforming growth factor β/Smad signaling pathway by inhibiting BUB1. This was found have led to the inhibition of hepatoma cell proliferation, invasion and migration, which in turn decreased cell viability and increased apoptosis [38]. It has been reported that silencing the expression of CDC20 will activate apoptosis and autophagy. Targeted inhibition of CDC20 expression is an ideal strategy for the treatment of HCC. KIF20A is an important target gene in the pathogenesis of HCC. The proliferation and growth of hepatoma cells are essential, and studies have shown that KIF20A is a potential target for future therapeutic interventions. PBK is also an independent prognostic biomarker of HCC [39, 40]. PBK plays an important role in the proliferation of cancer cells and the maintenance of mitotic fidelity. In many cancers,PBK is often upregulated, causing it to drive tumorigenesis and metastasis. CDK1 is phosphorylated by PBK during mitosis [41]. CEP55 has been studied in many types of cancers, including breast, lung, colon and liver cancer. The overexpression of CEP55 is closely related to the staging, invasiveness, metastasis and poor prognosis of many types of tumors. Therefore,it has been listed as a prognostic marker of various tumors [42].

Modular analysis showed that the occurrence of HF is associated with "Cell cycle","Progesterone-mediated oocyte maturation", "oocyte meiosis", "p53 signaling pathway", "Proteoglycans in cancer", "ECM-receptor interaction", "Focal adhesion","Neuroactive ligand-receptor interaction","MicroRNAs in cancer", "Bladder cancer". "Systemic lupus erythematosus", "Alcoholism" and "Viral carcinogenesis". In this study, HF was analyzed using bioinformatics methods. Further analysis of differential genes helped to identify useful targets and pathways in the pathogenesis of HF. These results may provide a theoretical basis for the clinical study of therapeutic targets. One of the limitations of this study is that most of the selected genes and pathways have not been experimentally verified. However, MMP1 and ETV6 were verified through text mining and may be potential molecular targets for HF therapy, while the other central genes need further verification. This could be the focus of future research on HF.