EMP1 as a Potential Biomarker in Liver Fibrosis: A Bioinformatics Analysis

Liver fibrosis is a wound-healing response to chronic injury, which may result in cirrhosis and liver failure. Studies have been carried on the mechanisms and pathogenesis of liver fibrosis. However, the potential cell-specific expressed marker genes involved in fibrotic processes remain unknown. In this study, we combined a publicly accessible single-cell transcriptome of human liver with microarray datasets to evaluate the cell-specific expression patterns of differentially expressed genes in the liver. We noticed that EMP1 (epithelial membrane protein 1) is significantly active not only in CCl4 (carbon tetrachloride)-treated mouse liver fibrosis but also in BDL (bile duct ligation)-induced liver fibrosis and even in human fibrotic liver tissues such as alcoholic hepatitis, NASH (nonalcoholic steatohepatitis), and advanced stage liver fibrosis. Furthermore, we demonstrated that EMP1 is a specific fibrotic gene expressed in HSCs (hepatic stellate cells) and endothelial cells using the Protein Atlas single-cell transcriptome RNA-sequencing clustering. Its expression was significantly elevated in fibrotic HSCs or CCl4 and NASH-induced fibroblasts. Previous research revealed that EMP1 plays a role in proliferation, migration, metastasis, and tumorigeneses in different cancers via a variety of mechanisms. Because HSC activation and proliferation are two important steps following liver injury, it would be interesting to investigate the role of EMP1 in these processes. All of this information suggested that EMP1 could be used as a novel fibrotic liver marker and a possible target in the future.


Introduction
Liver fibrosis is an abnormal accumulation of extracellular matrix (ECM) proteins, such as collagen, that occurs in the majority of chronic liver disease, and cirrhosis, liver failure, and portal hypertension are all symptoms of advanced liver fibrosis [1]. Liver inflammation is caused by chronic liver damage that disrupts the physiological architecture of the liver tissue [2]. During the injury, hepatocytes apoptosis and emit damage-associated patterns, which recruit and activate lymphocytes and macrophages, as well as promote pro-fibrotic myofibroblast activation [3]. These myofibroblasts are primarily derived from transdifferentiation resident hepatic stellate cells (HSCs) [4]. However, the fate of liver might either be an anti-fibrotic, scar-dissolving stage, or an unconstrained fibrosis-promoting stage by liver non-parenchymal cells [5]. A number of pathways and mediators, such as autophagy, endoplasmic reticulum stress, oxidative stress, retinol and cholesterol metabolism, epigenetics, and receptor-mediated signals, demonstrate the complexities of HSC activation [6,7]. The novel possible marker genes that may be important in these processes, on the other hand, remain unexplored.
In this study, we combined single-cell RNA sequence data from liver samples with a conventional microarray dataset from public database to identify marker genes that are expressed in HSCs and respond during the fibrotic process. We discovered that EMP1 (epithelial membrane protein 1) is significantly activated not only in CCl 4 (carbon tetrachloride)-treated mice liver, but also in BDL (bile duct ligation)-induced mice liver fibrosis and even in human fibrotic liver tissues such as alcoholic hepatitis, NASH (nonalcoholic steatohepatitis), and also liver with advanced stage disease. Furthermore, we proved EMP1 is a particular fibrotic gene expressed in HSCs and endothelial cells (EC) utilizing the single-cell transcriptome RNA-sequencing clustering result from ProtinAtlas database, and its expression was dramatically elevated in both CCl 4 and NASH-induced fibroblasts. Previous studies found that EMP1 regulates cancer cell migration and proliferation, and we suspect it has a similar role in activated HSCs [8,9]. The epithelial membrane proteins (EMPs) are encoded by the peripheral myelin protein 22 kDa (PMP22) gene family, and involved in tumor cell migration, growth, and differentiation. EMP1 is also known as CL-40, tumor-associated membrane protein. The EMP1 protein consists of 157 amino acids and is a glycoprotein containing four highly conserved hydrophobic transmembrane domains localized on the membrane. All of this information showed that EMP1 might be employed as a novel fibrotic liver marker and could be a potential target in the future.

Materials and Methods
2.1. Gene Expression Profile Data. All gene expression microarray matrix were collected from the NCBI (National Center for Biotechnology Information) GEO (Gene Expression Omnibus), which is a database repository of high throughput gene expression data and hybridization arrays, chips, and microarrays [10].

Function Enrichment Analysis.
We uploaded the DEGs into Metascape (https://metascape.org/) and conducted Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses, and Minimal Common Oncology Data Elements (MCODE) enrichment analysis [22]. The results of GO and KEGG analysis were visualized in the R ggplot2 package (v 3.3.3). Meanwhile, DEGs were submitted into the STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) database (v11.0) using the medium confidence (interaction score >0.400) parameter [23]. Functional annotation enrichment was also performed in STRING and top enriched pathways were visualized in R using ggplot2 package (v 3.3.3).

Liver Tissue
Single-Cell Analysis. Because numerous cells are involved in the course of liver fibrosis, the expression of common genes in these cells was examined to see if they induce fibrosis by controlling the function (migration) of these cells. In this study, we used the Protein Atlas [24] and PanglaoDB databases [25] to investigate for specific expression patterns of EMP1. Protein Atlas is an interactive open-access database (http://www.proteinatlas.org/) to allow genome-wide exploration of the impact of individual proteins on clinical outcomes. PanglaoDB (https://panglaodb.se/) is a database for the scientific community interested in investigating single-cell RNAsequencing research from mice and humans.
2.5. GEPIA Database Analysis. The GEPIA (Gene Expression Profiling Interactive Analysis) [26] is a website that contains the sequenced RNA expression data of 9736 cancers and 8587 normal samples from the TCGA and GTEx (the genotype-tissue expression) projects (http://gepia.cancer-pku .cn). Pearson's correlation was utilized to validate the strong positive correlations of EMP1 with other fibrotic-related genes,

Fibrotic-Related Gene Expression Patterns in CCl 4 -
Induced Mouse Liver Samples. To investigate the mechanisms of liver fibrosis, we analyzed the DEGs in three GEO datasets between CCl 4 -induced fibrotic and control samples. In GSE141821, 45 upregulated and 7 downregulated DEGs were found (|log2 (foldchange)| >1, adjusted P = 0:05; Figure 1(a) and 1(b)). On the other hand, Figure 1(c) and 1(d) demonstrate that in GSE55747 fibrotic mouse liver samples, there were 1817 upregulated and 1778 downregulated DEGs with a |log2 (foldchange)| > 1 and adjusted P < 0:05. In GSE80601, following analysis, 1778 downregulated and 1817 upregulated DEGs were found in CCl4-induced fibrotic liver tissues (|log2 (foldchange)| > 1, adjusted P < 0:05; Figures 1(e) and 1(f)). We identified 62 upregulated genes and 5 downregulated genes by intersecting the DEGs from three datasets (Figures 1(g) and 1(h)). These genes encode a number of ECM proteins, including Col3a1, Col4a1, and Col1a1. In the meanwhile, genes encoding ECM degrading enzymes including Timp2, Mmp2, and Adamts5 were considerably upregulated. Acute-phase protein expression of Saa3 (serum amyloid A3) significantly elevates in response to acute and chronic inflammatory stimuli. This finding highlights the fact that the ECM is altered after CCl 4 treatment. Also, Cd14, Cd53, and Cd68, which are markers for monocytes and macrophages, were considerably increased. It implies an increase of macrophage contents and increased immune filtration level following the injury. Cytochrome P450 gene family members, such as Cyp7b1 and Cyp2d40 are among downregulated genes. In addition, after the injury, Hsd3b5 (hydroxydelta-5-steroid dehydrogenase, 3 beta-and steroid deltaisomerase 5) was down regulated. The function of normal liver cells is impacted by liver injury. The upregulated and downregulated DEGs were shown using Venn diagrams (Figures 1(g) and 1(h), respectively).
The enrichment analysis was performed by Metascape ( Figure 2(a)). Pathways involved includes ECM organization (GO: 0030198), positive regulation of cell migration (GO: 0030335), collagen degradation (R-MMU-1442490), and positive regulation of response to external stimuli (GO: 0032103). Figure 2(b) depicts the MCODE analysis for PPI (protein-protein interaction) models by Metascape. The ECM protein cluster is part of the core PPI interaction network (Col1a1, Col3a1, Col4a1, Col6a1, Lum, and Sparc; Figure 2(b)). Furthermore, we used STRING for functional annotation enrichment, and the top enriched pathways were displayed in R using the ggplot2 tool (v 3.3.3). GO:BP enrichment indicated that pathways, such as stress response and cytokine are altered (Figure 2(c)), whilst KEGG enrichment indicated an alteration in actin cytoskeleton regulation and leukocyte transendothelial activation (Figure 2(d)). These enrichment pathway findings are mostly consistent with the fibrosis process in the liver.

EMP1 Is a Marker Gene for CCl 4 -Treated Liver Samples.
We noticed that CCl 4 treatment caused a rapid increase in  7 Gastroenterology Research and Practice liver tissues treated with CCl 4 (n =6) and control liver tissues (n = 5) treated with oil for both 6 weeks from Balb/c mice revealed that Emp1 expression significantly increased following CCl 4 treatment (log 2 FC = 2.461, P = 2:39 × 10 −4 ; Figure 3(a)). In GSE27640 after 18 weeks of CCl 4 treatment in the mouse model, the Emp1 gene was significantly upregulated (log 2 FC = 1.673, P = 2:61 × 10 −3 ) compared to normal control liver tissues. Erlotinib, an EGF receptor inhibitor that can reduce liver fibrosis and the development of hepatocellular carcinoma, suppressed the increased expression of Emp1 in GSE27640 (Figure 3(b)). Meanwhile, in GSE73499 following 3, 6, and 9 weeks of CCl 4 administration, the expression of the Emp1 gene        (Figure 3(c)). When compared to samples of rat livers treated with a control vehicle, the expression of Emp1 in the liver tissue of rats treated with CCl 4 was significantly upregulated (log 2 FC = 1.416, P = 0:035) in GSE139994. Taken together, these findings suggested that the liver tissues response to CCl 4 -induced acute and chronic injury triggers a rapid increase in Emp1 in mouse model.

EMP1 Is a Marker Gene for Fibrotic Liver Samples.
Aside from the CCl 4 -treatment model, similar findings have been made in other liver fibrosis models, including human clinical fibrotic liver samples caused by conditions like NAFLD (nonalcoholic fatty liver [NAFL] disease). For instance, when compared to the control group, Emp1 is rapidly activated in GSE40041 in acute or chronic damage mouse models by sham or BDL for either 48 hours or 28 days, respectively (log 2 FC = 1.152, P = 3:26 × 10 −4 ; Figure 3(e)). According to the GSE28619 dataset, hepatic gene expression profiling was examined by microarray in patients with alcoholism (n = 15) and normal livers (n = 7), and EMP1 is considerably higher in alcoholic hepatitis tissues (GSE28619; log 2 FC = 2.73, P = 1:08 × 10 −5 ; Figure 3(f)). EMP1 is also substantially more highly expressed in alcoholic steatosis and cirrhosis than in moderate acute alcoholic hepatitis, according to the set of data (GSE103580; (log 2 FC = 0.325, P = 0:032; Figure 3(g)). Moreover, EMP1 is greater higher expressed in samples of NAFLD liver biopsy tissues recovered from advanced (fibrotic stage 3-4) compared to moder-ate samples (fibrotic stage 0-1) in GSE49541 (log 2 FC = 1.152, P = 3:26 × 10 −4 ; Figure 3(h)). Furthermore, in the different stages of rat fibrotic liver from GSE65220, Emp1 has the highest expression in NASH compared to other stages, such as healthy, NAFL, and NAFLD with T2DM (type 2 diabetes; log 2 FC = 2.157, P = 9:92 × 10 −3 ; Figure 3(i)). All of this evidence suggests that EMP1 expression increased with the severity of the fibrotic liver malignancy and that it may play a role in the regulation of the liver injury response.

Dominant Expression of EMP1 in HSCs from Liver
Single-Cell Transcriptome. Using The Human Protein Atlas (https://www.proteinatlas.org/), we investigated the expression profiles of different cell types in liver tissues to determine which cell type expresses EMP1 the most. We discovered that EMP1 is exclusively expressed in EC (purple c-12 cluster) as well as Ito (perisinusoidal fat-storing cells, also known as stellate cells) of the liver as mesenchymal cells (grass green c-4 cluster) in the single-cell transcriptome clustering results for normal liver tissues. The data was displayed in a UMAP (Uniform Manifold Approximation and Projection plot; Figure 4(a)). According to the clustering results, EMP1 expression is lower in Kupffer cells (brick red c-3 and c-6 clusters) and other cell types including immune cells and hepatocytes (c-2; Figure 4(a)). Ito cells had an average expression value TPM (transcripts per kilobase) of 161.3, while EC had an average TPM of 181.9, and EMP1 expression is low in all other cell types (Figure 4(b)). The heatmap

13
Gastroenterology Research and Practice depicts the highest levels of EMP1 gene expression as well as other marker genes from various cell types, such as CD34 for EC, CD3E for T cells, and CD163 for macrophages. EMP1 and FSCN1 (Fascin actin-bundling protein 1) have the most similar expression patterns according to the dataset (Figure 4(c)).

EMP1 Expression Elevated in Activated HSCs and Liver
Fibroblast. Since we know that the activation of HSCs into proliferative, fibrogenic myofibroblasts has long been recognized as the primary cause of hepatic fibrosis in both experimental and human liver injury. GSE120281, an mRNA profile of quiescent (Q) or fibrotic (F) HSC and liver sinusoidal EC (SEC) generated by RNA sequencing in control or CCl 4 -treated mice, was investigated. In fibrotic HSCs, compared to quiescent HSCs, Emp1 expression is markedly increased in response to CCl 4 treatment (log 2 FC = 4.895, P = 0:0008; Figure 4(d)). GSE134512 dataset that analyzed the transcriptome of activated fibroblasts from NASH livers and discovered that Emp1 was significantly elevated in both CCl 4 -induced and NASH-induced fibroblasts when compared to activated HSCs (log 2 FC = 2.116, P = 0:00228; Figure 4(e)).
3.6. EMP1 Expression Correlated with ECM Proteins. GEPIA was used to confirm the significantly positive association of EMP1 expression with COL1A2 (collagen type I alpha 2 chain), TGFB1 (transforming growth factor beta 1), MMP19 (matrix metallopeptidase 19), and VIM (vimentin) in normal liver samples (r > 0:60, P < 0:05; Figure 5(a)-5(d)). Then, we double-checked our findings using the PanglaoDB database. EMP1 was shown to be highly expressed in fibroblasts, EC, followed by keratinocytes, but not in HSCs ( Figure 5(e)). These results indicate that EMP1 may assist in the fibrotic progression after activation and transdifferentiation of HSCs in the development of liver fibrosis, therefore increasing liver fibrosis.

Discussion
The fundamental cause of liver fibrosis is the activation of HSCs, which then transform from quiescent, vitamin-Astoring cells into proliferative, fibrogenic myofibroblasts [28]. Deactivation of HSCs by Tcf21 (transcription factor 21), has been proved to suppress hepatic fibrosis progression in mice [29]. However, HSCs activation is a rather complex process that has yet to be fully understood.
In this work, we first demonstrated that EMP1 is one of the most elevated genes after CCl 4 -induced liver damage in mouse model. Then, we discovered that EMP1 levels are also elevated in various liver samples after injury, such as the  BDL model and alcoholic hepatitis in humans. EMP1 is expressed in fibroblasts and EC, according to single-cell transcriptome study of liver tissues. Furthermore, we observed that EMP1 expression was much greater in activated fibrogenic fibroblasts and was associated with typical fibrotic genes. The EMPs, which includes EMP1, EMP2, and EMP3 are encoded by the growth arrest-specific 3 (GAS3)/peripheral myelin protein 22 kDa (PMP22) gene family [9]. The EMPs family has four putative transmembrane domains with about 160 amino acid residues [30].
Although the genes in this family are commonly implicated in cancer cell migration, proliferation, and differentiation, few studies have demonstrated their role in fibrosis. He et al. previously discovered that abnormal upregulation of PMP22 in TGF-β-activated HSCs and CCl 4 -induced hepatic fibrosis model in mice, as well as the pro-fibrotic role of PMP22 through aggravating TGF-β-induced HSC activation [31]. It would be interesting to look into and compare the role of the entire gene family in fibrotic processes in the liver. Extensive research has been conducted into the role of EMP1 in pathogenesis and tumorigenesis of various cancer. In breast cancer, for example, EMP1 functions as a new marker for lobular and ductal invasive breast carcinoma differentiation, as well as a putative link with breast cancer invasion promotion [32]. With increasing histologic grade, the expression of EMP1, EMP2, and EMP3 decreases in the epithelial component and increases in the stromal component of phyllodes tumors [33]. In prostate cancer, EMP1 is highly expressed in patients with a higher Gleason score, and increasing EMP1 levels significantly increases cancer cell migration, resulting in tumor metastasis, implying that EMP1 may play an important role as a positive regulator of tumor metastasis [8]. In ovarian cancer, EMP1 was discovered to play a critical role as a negative regulator in ovarian serous tumors, and decreased EMP1 expression in serous tumors associated with increased disease severity [34]. EMP1 is also a biomarker of gefitinib resistance, and it has been linked to a lack of complete or partial response to gefitinib in lung cancer patient samples, as well as clinical progression to secondary gefitinib resistance [35]. In acute lymphoblastic leukemia, EMP1 is a novel poor prognostic factor in pediatric leukemia that regulates prednisolone resistance, cell proliferation, migration, and adhesion [36]. In glioma, EMP1 regulates the cell proliferation, migration, and stemness through PI3K-AKT signaling and CD44 [37]. Given that the majority of research agree that EMP1 can enhance tumor growth, invasion, and migration, we hypothesize that EMP1 could also promote activated HSC proliferation and migration following injury. Stellate cell activation is characterized by proliferation and migration [38], and reducing proliferation may modify hepatic fibrosis, according to many published results [39][40][41][42]. However, further validating tests are required in the future to validate the hypothesis.
In our enriched analysis, the pathways involved includes ECM organization, positive regulation of cell migration, collagen degradation, and positive regulation of response to external stimuli. The enrichment analysis indicated that pathways such as stress response and cytokine are altered, whilst KEGG enrichment indicated an alteration in actin cytoskeleton regulation and leukocyte transendothelial activation. These enrichment pathway findings are mostly consistent with the fibrosis process in the liver.
There were several limitations in the present study. First, the included datasets were based on mice; we lack validation in human samples. Second, we did not perform the analysis to study the expression of EMP1 in different subgroups by different clinical factors such as sex and age. Third, lack of experiments to verify, and our analysis was performed using data sets from public databases but was not validated in larger data sets.
Taken together, these bioinformatic data imply that EMP1 might be used as a marker gene during liver fibrosis after injury and may have a potential role in HSCs, however additional validating experiments is needed in the future.

Data Availability
The simulation experiment data supporting this research article are available from the corresponding author or first author on reasonable request.