RAB7B protein as a possible treatment target in liver scarring based on protein and data analyses

From the GEO database, the microarray data sets (GSE77627 and GSE139602) were retrieved. GSE77627 contained 22 liver cirrhosis samples and 14 healthy samples, while GSE139602 included 20 cirrhotic samples and 6 healthy samples. The two data sets were merged to generate a completely new data set of liver cirrhosis (LC data set) by the R packages "Limma" and "sva". In addition, 75 mitophagy-related genes in Table S1↗ were curated from the KEGG and Reactome databases.

The R package "Limma" was used to find DEGs between cirrhotic liver samples and healthy controls in the LC data set. Differential expression requires an adjusted p-value (adj. p) < 0.05 and an absolute value of log2 (fold change) (log2FC) > 1. The R package "ggplot2" for volcano plots and "pheatmap" for heatmap representation were used to visualize DEGs.

The R package "clusterProfiler" was used to conduct GO and KEGG analyses to investigate the biological processes and functions associated with DEGs. The threshold for enrichment significance was adj. p < 0.05.

The STRING database and GeneMANIA software were used to build PPI networks. The GeneMANIA database offers functional bioinformatics tools, including analysis of physical interactions, gene enrichment, localization and coexpression analysis, and subcellular prediction. To measure functional gene relationships, the STRING database combines information from several sources, including protein–protein interactions, coexpression, gene neighborhood, and fusion. ¹⁹

Immune cell infiltration quantification was performed via CIBERSORT, assessing 28 immune subsets in the LC data set. A deconvolution algorithm called CIBERSORT was utilized to describe the makeup of immune cells in a tissue. The Wilcoxon test was used to assess statistical significance (p < 0.05). The R package "heatmap" and "vioplot" tools were used to depict the differences in immune cell infiltration. To confirm the link between genes and immunological infiltration, Spearman's correlation analysis was also carried out. The R tool "corrplot" was used to show the results.

WGCNA investigates the connection between gene networks and illnesses and finds gene modules with significant biological relevance.The LC data set's disease-related modules were found by the R package "WGCNA". The R package "VennDiagram" program was used to find shared genes between module genes and mito-DEGs. ²²

The single-cell RNA sequencing data set GSE137720 was analyzed via the scLiverDB.The expression of RAB7B at the single-cell level was analyzed. ²³

Eight-week-old male C57BL/6 mice weighing between 17 and 23 g were obtained from Hangzhou GemPharmatech Co., Ltd. and were randomized to either the liver cirrhosis group or the normal control group. Every animal experiment was carried out in compliance with institutional policies and authorized by Zhejiang University's Animal Care and Use Committee. All animal studies were conducted in full compliance with the ARRIVE guidelines 2.0. CCl₄-induced liver cirrhosis model using the previously reported technique. In short, mice received twice weekly intraperitoneal injections of sterile CCl₄ (1:3, dissolved in corn oil) for 6 weeks. After 48 h postfinal injection, the animals were anesthetized under isoflurane, and euthanized by cervical dislocation. Liver tissues were extracted for histological analysis.

Tissue samples were fixed with a 4% paraformaldehyde, embedded in paraffin. Histopathological changes were assessed using Masson staining, Sirius red, and immunohistochemistry.

LX-2 cells were purchased from Wuhan Pricella Biotechnology Co., Ltd. Cells were cultured in DMEM (Gibco, USA) with an addition of 2% fetal bovine serum (FBS). Cells were cultured at 37 °C with 5% CO₂.

Following the manufacturer's instructions, Lipofectamine3000 (Invitrogen, USA) was used to transfect cells with either RAB7B-targeting siRNA (si-RAB7B) or negative control siRNA (si-NC). Table S2↗ displays the sequences of every RNA interference product.

Standard transwell assays were used to conduct the cell migration assay.For quantitative analysis, three randomly selected fields of vision were photographed. ImageJ was utilized to quantify the quantity of migrating cells. ²⁵

Under the manufacturer's instructions, the EdU Cell Proliferation Kit (Beyotime, China) was utilized to assess cell proliferation. Using fluorescence microscopy, pictures were taken.

To put it simply, cells were blocked and fixed before being treated with α-SMA antibody and a secondary antibody conjugated to AF594. Nuclear counterstaining was performed with DAPI. Images were acquired using confocal microscopy (Zeiss LSM900, Germany).

The Quick RNA Extraction Kit (Accurate Biology, China) was used to extract total RNA following the manufacturer's instructions. The 2^‑ΔΔCt technique was used to quantify the data with β-actin as the reference gene. Table S3↗ lists the primer sequences that were employed.

Western blot assays were carried out as described in prior research. A 20 μg protein sample was separated by SDS-PAGE followed by Western blot. Table S4↗ details the primary antibodies utilized in the experiment.

The Connectivity Map (CMap) database is a resource that leverages differential gene expression profiles to predict potential therapeutic compounds.The top compounds were chosen based on their highest absolute connectivity scores (|score| ≥ 95). The 3D structure of the target protein was retrieved from the PDB. The protein structure was further processed using PyMOL software to remove water molecules and extract the target protein. Protein–protein docking was then performed with the GRAMM.Subsequently, LigPlot+ software was employed to analyze the chemical bonds involved in the docking results, and PyMOL was used for visualization. ^{26 27}

The data were statistically analyzed using SPSS 22.0 (SPSS Inc.) and GraphPad Prism 10 (GraphPad, USA) software. Pearson and Spearman correlation analyses were used to determine the relationship between metrics expression in cirrhosis data sets and liver fibrosis. Continuous data was analyzed between groups using two-tailed Student's t-tests or one-way ANOVA, with mean ± standard deviation. P-values <0.05 were considered statistically significant.

The human LC data set (GSE77627 and GSE139602) were merged and batch effects were removed. Figure illustrates the data sets and flowcharts used in this study. Distribution boxplots and principal component analysis revealed the successful elimination of batch effects (Figure S1A–D↗). Comparative analysis identified 2149 DEGs (|log2 FC| > 1, p < 0.05) between cirrhotic and healthy samples. This included 1044 upregulated and 1105 downregulated genes (FigureA). Heatmaps showing the expression patterns of these DEGs in liver cirrhosis and control samples are displayed (FigureB). The KEGG enrichment analysis found considerable enrichment in pathways associated with MAFLD and bile secretion (FigureC). GO enrichment analysis highlighted significant enrichment in process related to fatty acid metabolic processes, mitochondrial matrix, electron transport chain activity and oxidoreductase (FigureD–F).

These DEGs were intersected with mitophagy-related genes, and finally, seven mito-DEGs (NRAS, TOMM20, RAB7B, HIF1A, CITED2, FIS1, and CSNK2B) were identified (FigureA). Among them, NRAS, TOMM20, HIF1A, CITED2, FIS1, and CSNK2B were down-regulated and RAB7B was up-regulated in cirrhotic tissues relative to normal tissues (FigureB). Further, correlation analysis of these seven mito-DEGs revealed a strong positive correlation between NRAS, TOMM20, HIF1A, CITED2, FIS1, and CSNK2B, while a strong negative correlation was observed with RAB7B (FigureC). In addition, ROC analysis was performed for all seven mito-DEGs. The results indicated that all seven mito-DEGs yielded ROC curves with an AUC of greater than 0.7, suggesting a high accuracy in predicting liver cirrhosis formation (FigureD–J). GeneMANIA was used to predict 20 probable transcriptional coregulators of these seven key genes with PPI network construction of 27 total nodes (FigureK). Enrichment analysis about important genes and putative coregulators was conducted using the STRING database. GO enrichment analyses were primarily associated with mitochondrial protein localization, mitochondrial function, ATP metabolism, and mitochondrial outer membrane (FigureL). Additionally, KEGG enrichment analysis displayed that each of these genes was involved in mitophagy in animals (FigureM).

To analyze differences in the immune microenvironment, we further examined the infiltration of 28 immune cell types to completely comprehend liver cirrhosis and immune cells by the CIBERSORT algorithm. (FigureA). Distinct immune cell distribution patterns emerged between cirrhotic and control groups, with significant intercellular correlations (FigureB,C). Elevated infiltration of activated CD4+ T, eosinophil, immature dendritic cell, MDSCs, and natural killer T cell was observed in the liver cirrhosis. Conversely, patients with cirrhosis had reduced numbers of effector memory CD8+ T, central memory CD8+ T, macrophage, memory B cell, monocyte, natural killer cell, and regulatory T cell cells. Significant gene–immune correlations were identified for all seven mitophagy-DEGs. The results indicated that CITED2 and RAB7B were most strongly correlated with natural killer T cells (r = −0.741 and 0.750, p < 0.001 and p < 0.001, respectively) and memory B cells (r = 0.689 and −0.724, p < 0.001 and p < 0.001, respectively); CSNK2B exhibited the strongest association with monocyte (r = 0.753, p < 0.001); FIS1 and NRAS were the most strongly correlated with memory B cell (r = 0.647 and 0.786, p < 0.001 and p < 0.001, respectively); and HIF1A and TOMM20 displayed the highest correlation with regulatory T cells (r = 0.773 and 0.779, p < 0.001 and p < 0.001, respectively) (FigureD). These results establish functional crosstalk between mitophagy regulators and immune microenvironment in cirrhosis pathogenesis.

In the LC data set, WGCNA identified an optimal soft-thresholding power of 13 (FigureA). A total of six modules were identified, with the red module comprising 743 genes strongly correlated with liver cirrhosis (FigureB). Furthermore, a scatter plot revealed a positive association (r = 0.59, p < 0.001) across the red module membership and gene importance for liver cirrhosis (FigureC). As consequently, the red module was regarded as the crucial module for next analysis. Venn analysis revealed RAB7B and NRAS at the mitophagy-module intersection (FigureD).

To identify and screen candidate genes associated with liver fibrosis development, we employed microarray data sets from public databases. The analysis of the GSE25097 data set indicated higher expression levels of RAB7B and NRAS in cirrhotic livers compared to normal livers (FigureA). In addition, the GSE84044 data set indicated increased expression of RAB7B and NRAS in patients with significant fibrosis (Scheuer score S2–4) than those without significant fibrosis (Scheuer score S0–1) (FigureB). Notably, RAB7B exhibited consistent upregulation across the LC data set, GSE25097 and GSE84044, whereas NRAS showed a divergent regulation. Additionally, we performed a correlation analysis between RAB7B, NARS, and key liver fibrosis markers. The findings revealed a significant positive correlation (r = 0.739 and 0.468, p < 0.001 and p < 0.001, respectively) between RAB7B and collagen alpha 1 (COL1α1) (FigureC, D), while NRAS exhibited a weaker association with COL1α1 (r = 0.604 and 0.418, p < 0.001 and p < 0.001, respectively) (FigureC, D). Based on these findings, we selected RAB7B for further investigation.

According to the bioinformatic analyses above, RAB7B could have a role in the process of liver cirrhosis. The cell types that express RAB7B in the liver were identified using the Human Protein Atlas. The findings showed that HSCs, cholangiocytes, and plasma cells were the primary locations for RAB7B expression in the liver (FigureA). In addition, we found that RAB7B is predominantly localized in the cell population of myofibroblasts with Stmn1 expression, as revealed by single-cell RNA sequencing analysis of liver tissues from carbon tetrachloride-induced mouse models (FigureB–D). Given the pivotal role of HSCs activation in liver cirrhosis, we investigated RAB7B expression dynamics during HSCs activation. After extensive research, it was shown that the human HSC-derived LX-2 cell line had profibrotic characteristics. We employed to assess RAB7B regulation under activation conditions. The findings demonstrated that when HSCs are cultivated in DMEM supplemented with 10% fetal bovine serum (FBS), they display a pro-fibrotic phenotype. We then evaluated RAB7B expression using a model that was 10% FBS-stimulated. After 24 h of incubation with 10% FBS, we found that LX-2 cells expressed more RAB7B and COL1α1 than cells grown in a control condition with 2% FBS (FigureE,F). In an LX-2 cell activation model induced by transforming growth factor-β (TGF-β), RAB7B mRNA levels were markedly increased at 24- and 48 h postexposure. (FigureG). In addition, COL1α1 was significantly upregulated following TGF-β stimulation (FigureG). Western blot also confirmed RAB7B protein elevation (FigureH). Collectively, these data establish RAB7B upregulation as a hallmark of HSC activation.

To further validate the results obtained from the previous data set, we induced liver cirrhosis using CCl₄ in mice. Sirius red, Masson, and α-SMA staining demonstrated significant collagen fiber deposition around the hepatic lobules and within the confluent region of liver cirrhosis mice (FigureA). Immunohistochemistry analysis further confirmed that RAB7B expression was markedly upregulated in CCl₄-induced cirrhotic livers (FigureA). To validate the expression of RAB7B in liver cirrhosis, we quantified mRNA levels in liver specimens from CCl₄-induced cirrhotic mice. The findings showed that liver cirrhosis was associated with elevated expression of COL1α1, an activation marker for HSCs, and persistently higher levels of RAB7B expression than normal control mice. (FigureB). Moreover, correlation analysis revealed a positive relationship between RAB7B and COL1α1 expression (r = 0.502, p = 0.030) in CCl₄-treated mice (FigureC). According to Western blot, cirrhotic livers have higher levels of RAB7B protein expression than normal liver tissues. (FigureD). These findings collectively demonstrate that RAB7B expression was upregulated in cirrhotic livers, providing strong evidence of a substantial association with liver cirrhosis.

Building on established evidence of RAB7B upregulation in activated HSCs, our hypothesis was that the activation phenotype of HSCs would be suppressed by RAB7B expression. To test this hypothesis, Western blot and qRT-PCR were used to confirm siRNA-mediated RAB7B knockdown. The results exhibited that si-RAB7B had a significant interference efficiency as evidenced by a reduction in RAB7B mRNA and protein levels (FigureA,B). It has been demonstrated that activated HSCs exhibit altered phenotypes, including migration toward the site of injury, increased cell proliferation, and expression of α-SMA., RAB7B knockdown impaired migration in activated LX-2 cells according to the transwell experiment. TGF-β stimulation increased LX-2 cell migration by 3.0-fold, which was attenuated by 63% with RAB7B silencing (FigureC,D). EdU proliferation assays demonstrated that RAB7B knockdown reduced TGF-β induced LX-2 cell proliferation by 45% (FigureC,E). Furthermore, immunofluorescence analysis revealed that inhibition of RAB7B effectively suppressed the expression of α-SMA (FigureC,F). Notably, RAB7B silencing abrogated TGF-β mediated COL1α1 upregulation, showing 61% suppression compared to stimulated controls (FigureG).

A total of 150 upregulated and 150 downregulated DEGs were submitted to the CMap database to screen for potential small-molecule therapeutics for liver cirrhosis. Through this analysis, six candidate drugs with high connectivity scores were discovered, including mofezolac, parbendazole, N-formylmethionylalanine, SKF-89976A, LDN-193189, and celastrol (FigureA). The structures of the identified compounds were retrieved from the PubChem database and shown in FigureB–G.

Molecular docking analysis was performed to identify potential interactions between RAB7B and selected therapeutic compounds. The calculated binding energies for the RAB7B-drug complexes were as follows: mofezolac (−6.4 kcal/mol, FigureA), parbendazole (−7.0 kcal/mol, FigureB), N-formylmethionylalanine (−5.3 kcal/mol, FigureC), SKF-89976A (−6.8 kcal/mol, FigureD), LDN-193189 (−9.1 kcal/mol, FigureE), and celastrol (−8.6 kcal/mol, FigureF). These results indicate strong molecular interactions between RAB7B and the six candidate compounds.

To further clarify the role of RAB7B in mitophagy, we performed functional and regulatory analyses in LX2 cells. TGF-β stimulation induced p62 accumulation, LC3-II reduction, TOMM20 elevation, and decreased PINK1 and Parkin, indicating impaired mitophagy. Notably, RAB7B knockdown reversed these alterations, with reduced p62 and TOMM20 and increased LC3-II, PINK1, and Parkin, suggesting that RAB7B deficiency promotes mitophagy and prevents mitochondrial accumulation (FigureA,B). Consistently, MitoTracker and LysoTracker staining showed enhanced colocalization of mitochondria and lysosomes after RAB7B knockdown, further supporting its inhibitory role in mitophagy (FigureC). In addition, we examined RAB7B expression under altered mitophagy states. Chloroquine (CQ)-mediated mitophagy inhibition led to elevated p62, LC3-II, TOMM20, and RAB7B, whereas CCCP-induced mitophagy activation decreased p62, TOMM20, and RAB7B while increasing LC3-II (FigureD–G). Together, these findings indicate that RAB7B negatively regulates mitophagy and that its expression is dynamically modulated by mitophagy activity.