Ursolic acid reverses liver fibrosis by inhibiting interactive NOX4/ROS and RhoA/ROCK1 signalling pathways

UA reverses liver damage and fibrosis in liver fibrotic mice

First, we used HE and Masson’s trichrome staining of mouse liver sections to determine the effects of UA on liver damage, fibrous septum formation and collagen deposition in mice with liver fibrosis (Figure 1A–1B). After CCl₄-induced mice were treated with UA, the effects of CCl₄ on liver hepatic lobule structure, collagen deposition and fibrous connective tissue hyperplasia were significantly reversed and were accompanied by decreased inflammatory cell infiltration (P<0.05) (Figure 1C–1D). Furthermore, as an important component of collagen deposition during liver fibrosis, the hydroxyproline content in the liver tissue was also detected (Figure 1E). Next, we assessed the levels of ALT, AST, and TBIL in the mouse serum to determine liver function (Figure 1F). Compared to those in the control group, the serum levels of ALT, AST and TBIL in the CCl₄ group mice were significantly increased. However, this increase was inhibited in UA-treated fibrotic mice. These results indicate that UA can reverse liver damage and fibrosis to a certain extent to protect the liver.

Figure 1. The effect of UA on CCl₄-induced liver injury and fibrosis is related to NOX4. (A) HE staining (100× magnification). (B) Masson’s trichrome staining (100× magnification). (C–D) Morphometrical analysis of the fibrotic score and fibrotic area. (E) Detection of the hydroxyproline content in the liver tissue by colorimetry. (F) Liver function indices in mouse sera. Data represent the mean ± SD for each group. *P < 0.05 and ***P < 0.001.

HSC activation is well established as the central driver of hepatic fibrosis in experimental and human liver injury, and HSC activation releases a large amount of ECM, which is deposited in the interstitial space of the liver, eventually leading to liver fibrosis [19]. As shown by both α-SMA staining and the TUNEL assay, the expression of α-SMA, a biomarker of activated HSCs [20], in the liver tissue of the CCl₄ group was significantly increased compared to that in the control group. Apoptosis in hepatocytes in the CCl₄ group also increased, but there was a significant decrease in apoptosis in the UA group (Figure 2A). This indicates that UA can inhibit the apoptosis of hepatocytes and the activation of HSCs, thereby inhibiting the progression of liver fibrosis.

Figure 2. The effect of UA on CCl₄-induced liver fibrosis-related indicators is related to NOX4. (A) Dual immunofluorescence staining of liver sections from mice in the control, CCl₄ and UA groups for nuclei (DAPI, blue), aHSCs (α-SMA, green), and apoptosis (TUNEL, red), and the merged images are shown. (B) Hepatic mRNA levels of collagen I, MMP-1, α-SMA, and TIMP-1 were measured by qRT-PCR. (C) Collagen I, MMP-1, α-SMA, and TIMP-1 protein expression was detected by a western blot. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

Liver fibrosis is often accompanied by changes in the expression of fibrosis-related factors. At the mRNA and protein levels, the expression levels of type I collagen and TIMP-1 in the CCl₄ group were significantly elevated compared with those in the control group. After UA treatment, this increase in the expression of type I collagen and TIMP-1 decreased, and the expression of MMP-1, an anti-fibrotic factor that promotes the degradation of extracellular matrix [21], was elevated (Figure 2B–2C), demonstrating that UV exerts a reversal effect on liver fibrosis.

UA relieves liver fibrosis by inhibiting the NOX4/ROS signalling pathway

NOX4 mediates the signal transduction of major pro-hepatic fibrosis factors, such as TGF-β, leading to HSC activation and hepatocyte apoptosis and thus plays an important role in liver fibrosis [22]. Our previous studies confirmed that NOX4 and ROS are upregulated in fibrotic liver tissue and that Rac1 expression is upregulated in activated HSCs. In CCl₄-induced liver fibrotic mice, the NOX4 expression level MDA content that assessed the accumulation of ROS and oxidative stress in the liver were significantly increased (Figure 3A–3D). To further validate the importance of NOX4 in liver fibrosis, we generated NOX4^-/- liver fibrotic mice and injected liver fibrotic mice with the NOX4 inhibitor AP (Supplementary Figure 1). Compared with the mice in the CCl₄ group, the mice in the NOX4^-/- and AP groups had a smaller fibrous septum and less collagen deposition, which were accompanied by decreased inflammatory cell infiltration (Figure 1A–1E). The levels of serum markers suggested that liver function damage was reduced in liver fibrotic mice after NOX4 inhibition (Figure 1F). The results of other experiments, such as α-SMA staining, TUNEL assay, MDA content, and liver fibrosis-related factor expression analyses, were similar (Figure 2A–2C), indicating that the inhibition of NOX4 can improve liver fibrosis, confirming that NOX4 is an important profibrotic factor in the progression of liver fibrosis.

Figure 3. Effect of UA on the expression of NOX4 in mice with liver fibrosis. (A) The effect of UA on NOX4 expression was determined by using IHC. (B) Hepatic mRNA levels of NOX4 were measured by qRT-PCR. (C) Hepatic protein levels of NOX4 were detected by a western blot. (D) Detection of MDA content in the liver by a commercial kit. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

To confirm that NOX4 is a target for UA intervention in fibrosis, we first verified the effect of UA on NOX4 expression in vivo by IHC (Figure 3A). The expression of NOX4 in the UA group was lower than that in the CCl₄ group. Changes in the mRNA expression of NOX4 also confirmed this hypothesis (Figure 3B). The mRNA expression level of NOX4 was significantly lower in the UA group than in the CCl₄ group (P<0.01). Changes in NOX4 expression at the protein level were also similar to those at the mRNA level (Figure 3C). Moreover, the MDA content in the UA group was lower than that in the CCl₄ group (Figure 3D), indicating the inhibitory effect of UA on NOX4/ROS. Next, we explored whether NOX4 is a target for the anti-fibrotic effects of UA. In the NOX4^-/-+UA group, changes to the fibrous septum, collagen deposition and inflammatory infiltration and serum indices were not more decreased than those in the NOX4^-/- group, although these changes were still less pronounced than those in the CCl₄ group (Figure 1A–1F). Furthermore, compared to those in the NOX4^-/- group, the number and area of α-SMA- and TUNEL-positive cells in the NOX4^-/-+UA group were not significantly changed. The changes in factors related to liver fibrosis were similar (Figure 2A–2C), suggesting that UA reverses liver fibrosis by inhibiting the NOX4/ROS pathway.

RhoA/ROCK1 is another target of the anti-fibrotic effect of UA

RhoA has been shown to regulate HSC activation, migration, adhesion, contraction, proliferation and apoptosis [9, 23, 24]. Compared with their expression in the control group, RhoA and ROCK1 expression in the CCl₄ group was not significantly altered (Figure 4A–4C). To examine the effects of RhoA on liver fibrosis in vivo, we constructed RhoA-inhibited liver fibrotic mice by injecting the mice with AAV and the RhoA inhibitor fasudil (Supplementary Figure 2). Compared with mice in the CCl₄ group, liver fibrotic mice in the RhoAi and FA groups exhibited a smaller fibrous septum, less collagen deposition and fewer infiltrated inflammatory cells (Figure 5A–5D). Correspondingly, the hydroxyproline content was lower in RhoA-inhibited mice with liver fibrosis than in control mice (Figure 5E). We next tested the levels of ALT, AST, and TBIL in mouse serum. The liver function in liver fibrotic mice in the RhoAi and FA groups exhibited significant recovery, unlike that in the CCl₄ group (Figure 5F). The results of dual immunofluorescence and the expression of liver fibrosis-related factors also show the importance of RhoA in liver fibrosis (Figure 6A–6C).

Figure 4. Effect of UA on the expression of RhoA/ROCK1 in liver fibrotic mice. (A) The effect of UA on RhoA/ROCK1 expression was determined by using IHC. (B) Hepatic mRNA levels of RhoA/ROCK1 were measured by qRT-PCR. (C) Hepatic protein levels of RhoA/ ROCK1 were detected by a western blot. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

We assessed whether UA reverses liver fibrosis by inhibiting RhoA-related signalling pathways. First, to verify the effect of UA on the expression of RhoA and ROCK1, the downstream target of RhoA, we conducted IHC analyses (Figure 4A). The expression of RhoA/ROCK1 in the UA group was lower than that in the CCl₄ group. Changes in the mRNA and protein levels of RhoA and ROCK1 were similar. In liver fibrotic mice, UA significantly reduced this increase in the expression of RhoA/ROCK1 in the liver (P<0.05) (Figure 4B–4C), indicating that UA can interfere with the expression of RhoA/ROCK. Next, we explored whether the anti-fibrotic effect of UA is related to its inhibition of RhoA/ROCK. Changes to the fibrous septum, collagen deposition and inflammatory infiltration in the RhoAi+UA group were not significant compared with those in the RhoAi group, although they were still less than those in the CCl₄ group (Figure 5A–5E). The levels of serum indicators also recovered to a certain degree compared with those in the RhoAi treatment group (Figure 5F). Compared to those in the RhoAi group, the number and area of α-SMA- and TUNEL-positive cells were decreased in the RhoAi+UA group (Figure 6A). The expression of liver fibrosis-associated factors also decreased in UA-treated liver fibrotic mice following RhoA inhibition (Figure 6B–6C). These results indicate that UA exerts anti-fibrotic effects by inhibiting the RhoA/ROCK1 signalling pathway.

Figure 5. The effect of UA on CCl₄-induced liver injury and fibrosis is related to RhoA. (A) HE staining (100× magnification). (B) Masson’s trichrome staining (100× magnification). (C–D) Morphometrical analysis of the fibrotic score and fibrotic area. (E) Detection of the hydroxyproline content in liver tissue by colorimetry. (F) Liver function indices in mouse sera. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

Figure 6. The effect of UA on CCl₄-induced liver fibrosis-related indicators is related to RhoA. (A) Dual immunofluorescence staining of liver sections from mice in the control, CCl₄, and UA groups stained for nuclei (DAPI, blue), aHSCs (α-SMA, green), and apoptosis (TUNEL, red), and the merged images are shown. (B) Hepatic mRNA levels of collagen I, MMP-1, α-SMA, and TIMP-1 were measured by qRT-PCR. (C) Collagen I, MMP-1, and TIMP-1 protein expression was detected by a western blot. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

NOX4/ROS is the upstream signalling pathway of RhoA/ROCK1 in liver fibrosis

The above results confirm that NOX4/ROS and RhoA/ROCK are two signalling pathways on which UA exerts its anti-fibrotic effects. We then sought to clarify the possible interaction mechanism between these two signalling pathways. First, we investigated the expression of RhoA/ROCK1 in NOX4-inhibited liver fibrotic mice. IHC analysis showed that the expression of RhoA/ROCK1 in the NOX4^-/- group was decreased compared with that in the CCl₄ group (Figure 7A). Similarly, dual immunofluorescence staining for α-SMA and RhoA showed decreased expression of RhoA in NOX4^-/- mice (Figure 7B). At the mRNA level, the expression of RhoA/ROCK1 in the NOX4^-/- group was significantly decreased compared to that in the CCl₄ group (P<0.01) (Figure 7C). Next, we determined the expression of NOX4 in RhoA-inhibited liver fibrotic mice. As shown by IHC, the expression of NOX4 in the RhoAi group was not significantly lower than that in the CCl₄ group (Figure 7D). A slight decrease in the MDA content in the livers of RhoA-inhibited liver fibrotic mice was observed (Figure 7E). Compared to that in the CCl₄ group, the mRNA level of NOX4 in the RhoAi group was decreased to a certain degree, but this difference was not statistically significant (Figure 7F), indicating that NOX4/ROS is the upstream signalling pathway of RhoA/ROCK1 in liver fibrosis.

Figure 7. Interaction between NOX4/ROS and RhoA/ROCK1 in liver fibrotic mice. (A) In liver fibrotic mice in which NOX4 expression was inhibited, RhoA/ROCK1 expression was detected by IHC. (B) Dual immunofluorescence staining of liver sections from mice in the control, CCl₄, NOX4^-/- and AP groups stained for nuclei (DAPI, blue), aHSCs (α-SMA, green), and RhoA (red), and the merged images are shown. (C) Hepatic mRNA levels of RhoA/ROCK1 were measured by qRT-PCR. (D) In liver fibrotic mice in which RhoA expression was inhibited, NOX4 expression was detected by IHC. (E) Detection of the MDA content in the liver by a commercial kit. (F) Hepatic mRNA levels of NOX4 were measured by qRT-PCR. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

Intestinal improvement in liver fibrotic mice following treatment with UA

Intestinal damage and destruction of intestinal barrier integrity are often accompanied by liver fibrosis [25–27]. Once the integrity of the intestine is destroyed, harmful factors in the intestine enter the liver through the gut-liver axis, which in turn further aggravates the progression of liver fibrosis [28, 29]. Therefore, we explored improvements to the intestinal barrier induced by UA. As they are key indicators of the integrity of the intestinal barrier, the expression of the tight junction (TJ) proteins ZO-1 and occludin was tested. The expression of ZO-1 and occludin in the ileum of liver fibrotic mice was lower than that in the control group, but their expression increased after UA treatment (P<0.05) (Figure 8A). These results show that UA has a protective effect on the intestinal barrier.

Figure 8. Effect of UA on intestinal microbiota dysbiosis in mice with CCl₄-induced liver fibrosis. (A) Ileal mRNA levels of the TJ proteins ZO-1 and occludin were measured by qRT-PCR. (B) The alpha diversity of each group was assessed by determining the Shannon index and the Chao1 index. (C) PCoA to determine the weighted UniFrac distance of the intestinal microbiota. (D) Composition of the intestinal microbiota of each group at the phylum level. (E) Composition of the intestinal microbiota of each group at the genus level. (F) Linear discriminant analysis effect size (LEfSe) prediction was used to identify the bacteria in each group with the most differential abundance. (G) Linear discriminant analysis (LDA) scores showed significant differences in the bacteria in each group. Only the bacteria whose abundance met an LDA threshold value of >2 are shown. Data represent the mean ± SD of each group. *P < 0.05 and ***P < 0.001.

The intestinal microbiota has also received attention as an important indicator of intestinal and systemic status. We explored disorders in the intestinal microbiota of animal models of liver fibrosis and their improvement following treatment with UA. Disorders in the intestinal microbiota in our animal models of liver fibrosis and improvements to the intestinal microbiota following UA treatment were detected by next-generation sequencing. First, we tested some indicators that represent alpha diversity (Figure 8B). The Shannon index and the Chao1 index, which were used to assess microbial abundance, in the CCl₄ group were significantly decreased compared to those in the control group (P<0.01). After UA treatment, these indices increased (P<0.01). In the NOX4^-/-+UA and RhoAi+UA groups, this recovery in the Shannon and Chao1 indices was pronounced, but it was not statistically significant. Next, we conducted beta diversity analysis, which was used to compare microbial community compositions and assess differences between microbial communities (Figure 8C). The principal coordinates analysis (PCoA) plot showed a difference in the microbial communities among the control, CCl₄ and UA groups. Interestingly, there was a slight difference in the microbial communities among the NOX4^-/-+UA group, the RhoAi+UA group and the UA group. The composition of the intestinal microbiota was also found to change (Figure 8D). At the phylum level, the operational taxonomic units (OTUs) of the beneficial bacteria Firmicutes were lower in the CCl₄-treated mice than in the control mice. However, after UA treatment, this decrease in the abundance of Firmicutes bacteria was reversed. Firmicutes are beneficial bacteria that protect the body through a variety of mechanisms [30–32]. The abundance of Verrucomicrobia showed the opposite trend. UA also improved microbial abundance at the genus level (P < 0.05). At the genus level, the abundances of Bacteroidales and Lachnospiraceae in the CCl₄ group were lower than those in the control group. In the UA group, the abundances of both Bacteroidales and Lachnospiraceae increased (Figure 8E). Furthermore, linear discriminant analysis effect size (LEfSe) was used to analyse the microbiota composition (Figure 8F–8G) and revealed the microbiota abundance within each group.

Experimental animal model and design

The wild-type (WT) C57BL/6 mice used in the experiments were from the Department of Laboratory Animal Science of Nanchang University, and NOX4 knockout C57BL/6 mice were purchased from the Jackson Laboratory (USA). All animals were maintained in a temperature-controlled environment (20–22 °C) with a 12 h light-dark cycle with free access to sterile food and water. Based on widely recognized research, carbon tetrachloride (CCl₄) was selected to induce liver fibrosis in mice [17, 18]. According to the principle of random allocation, C57BL/6 mice weighing 20 to 30 g were randomly divided into the control group [n = 8, gavage with olive oil (2 ml/kg) twice a week for 8 weeks], CCl₄ group [n = 8, gavage with CCl₄ at 2 ml/kg (20% olive oil dilution) twice a week for 8 weeks], and UA group [n = 8, mice that were administered CCl₄ for 4 weeks received continued CCl₄ and UA (40 mg/kg/day) gavage for 4 weeks]. NOX4 knockout mice were randomly divided into the NOX4^-/- group (n = 8, gavage with CCl₄ twice a week for 8 weeks) and the NOX4^-/- + UA group (n = 8, treatment consistent with that of the UA group). A group of C57BL/6 mice that received adeno-associated virus (AAV) (4 ml/kg) via tail vein injection for 1 week to inhibit RhoA were randomly divided into the RhoAi group (n = 8, gavage with olive oil twice a week for 8 weeks) and the RhoAi + UA group (n = 8, treatment consistent with that of the UA group). We also added 2 inhibitor groups: the apocynin (AP) group [n = 8, mice that were given CCl₄ by gavage for 4 weeks and CCl₄ plus AP (40 mg/kg/d) (NOX4 biological inhibitor) by gavage for 4 weeks] and the fasudil (FA) group [n = 8, mice received CCl₄ by gavage for 4 weeks and CCl₄ plus FA (10 mg/kg/d) (RhoA biological inhibitor) by gavage for 4 weeks]. The mice were subjected to experimental procedures approved by the Animal Care and Use Committee of Nanchang University. All procedures were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Measurement of blood indices

Serum samples were collected while sacrificing the mice, and the alanine aminotransferase (ALT), aspartate aminotransferase (AST), and total bilirubin (TBIL) levels were measured by using an automatic biochemical analyser (Department of Clinical Laboratory, First Affiliated Hospital of Nanchang University, China).

Histological analysis

The liver samples were fixed in 4% neutral-buffered formalin, embedded in paraffin and sectioned. Sections underwent haematoxylin and eosin (HE) staining, Masson’s trichrome staining, immunohistochemistry (IHC), TdT-mediated dUTP nick-end labelling (TUNEL) analysis, and immunofluorescent analysis and were evaluated by microscopy. We randomly selected five visual fields for observation, scored liver fibrosis using METAVIR scoring criteria, and evaluated intestinal mucosal damage using the Chiu scoring method. IHC was used to determine the localization and expression of related proteins. Liver fibrosis was estimated by Masson’s trichrome staining. Specimens were incubated with an appropriate antibody and were observed and photographed by confocal microscopy.

Hydroxyproline and malondialdehyde measurements

The hydroxyproline and malondialdehyde (MDA) contents of the liver tissues were determined using an assay kit from Nan-jing Jiancheng Bioengineering Institute (Nanjing, China) according to the manufacturer’s instructions.

Construction and injection of vectors for RhoA inhibition

We constructed an AAV for RhoA inhibition. AAVs carrying short hairpin RNA (shRNA) using CV266 as a vector were injected into the mouse via the tail vein. The final titre of AAV-shRhoA was 3.5 × 10¹² viral particles/ml.

Gut microbiota analysis

We used a stool DNA kit (Omega, China) to extract genomic DNA from the faeces collected while sacrificing mice, and then purity was verified by 1% agarose gel electrophoresis (BioFROXX, China). Primers (338F 5'-ACTCCTACGGGAGGCAGCAG-3' and 806R 5'-GGACTACHVGGGTWTCTAAT-3') were used to amplify the bacterial V3-V4 region of the 16S rRNA gene. Pyrosequencing of the PCR products was performed on an Illumina MiSeq Instrument. (Majorbio, China). Analysis of changes in intestinal microbiota by alpha diversity (Shannon and Chao1 index), beta diversity, and community composition were calculated.

Western blotting analysis

Liver tissue protein was obtained from tissue lysates for western blotting. The protein levels were determined using a BCA assay kit (Tiangen, Beijing, China). Denatured proteins were separated on 10% Tris-glycine polyacrylamide gels by SDS-PAGE and transferred to PVDF membranes. The PVDF membranes were incubated overnight at 4 °C with anti-collagen I (Abcam, Cat. ab138492), anti-MMP1 (Abcam, Cat. ab137332), anti-α -SMA (Abcam, Cat. ab32575), anti-TIMP1 (Abcam, Cat. ab109125), anti-NOX4 (Abcam, Cat. ab109225), anti-RhoA (Abcam, Cat. ab187027), anti-ROCK1 (Abcam, Cat. ab45171), and anti-GAPDH (Abcam, Cat. ab8245). The corresponding membrane-bound antibodies were detected by a hypersensitive chemiluminescence detection reagent.

Quantitative real-time RT-PCR (qRT-PCR)

For real-time PCR analysis, PCR was performed with a reaction mixture containing cDNA template, primers, and TB Green™ Fast qPCR Mix (TaKaRa) in a Step One Plus Real-Time PCR System (Thermo Fisher Scientific). The relative abundances of the target genes were obtained by comparison against a standard curve.

Statistical analysis

We used SPSS 23.0 software for data analysis. Image production and output were performed using GraphPad Prism 7.0 software. Each experiment was repeated 3 times to ensure confidence in the results. One-way analysis of variance (ANOVA), Student’s t test, Mann-Whitney rank sum test, or Kruskal-Wallis H test was used to analyse the significant differences between groups. P <0.05 was considered significant.