The animal experiments were reviewed and approved by the Animal Ethics Committee of Central South University (Permit number: 2019sydw0056) and conducted according to the Guide for Animal Care and Use. Male C57BL/6J mice (8 weeks old, 18–20 g) were purchased from SLAC Laboratory (Hunan, China). All mice were housed in one cage for 1 week of acclimation, thus ensuring a similar microbial composition at baseline. Then mice were randomly divided into two groups (n = 20 per group): 1) the CCl₄ group, mice were intraperitoneally (i. p.) injected with CCl₄ solution (Macklin Biochemical, China) (20% in olive oil, 3 ml/kg body weight, twice a week); 2) the normal control (NC) group, mice were i. p. injected with an equal volume of olive oil (Macklin Biochemical, China). All mice were maintained in a temperature and light controlled specific pathogen-free facility (23°C, 12 h dark-light cycle, and 50% humidity) with free access to feed and water. To observe the evolution of gut microbiota in parallel with the development of cirrhosis, mice in the NC and CCl₄ groups were sacrificed at different time points (5 and 15 weeks after intervention, n = 10 per group at each time point).

Blood samples were obtained and centrifuged (3,000 rpm, 15 min) to collect serum. The serum levels of aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma-glutamyl transferase (γ-GT), total bilirubin (TBIL), and total bile acid (TBA) in mice were detected by an automated chemistry analyzer, Chemray 240 (Rayto Life Science, China).

After fixing in formalin, liver tissues were embedded in paraffin. In order to assess liver injury, the 3-μm-thick liver sections were stained with hematoxylin and eosin (H&E). Additionally, the tissues were stained with Masson’s trichrome or Sirius red, and morphometric analysis was performed. Briefly, five randomly selected 40 × fields were analyzed by the ImageJ software (NIH, United States) (Syn et al., 2012).

The feces of mice were freshly collected at the 5th and 15th week of the intervention and immediately stored at −80°C until analysis. The total DNA was extracted from feces samples of mice, and the DNA quality and concentration was evaluated by 1% agarose gel and a Qubit® dsDNA Assay Kit (Life Technologies, CA, United States), respectively. Then, the DNA samples were fragmented and prepared for PCR amplification. Next, the PCR products were purified and used for library construction. And the library preparations were sequenced using an Illumina HiSeq platform.

The acquired raw data were preprocessed by Readfq to obtain clean data which were further assembled and analyzed using SOAPdenovo software (Luo et al., 2012), predicted by MetaGeneMark (Qin et al., 2014), and annotated by DIAMOND software (Gao et al., 2021). The α diversity indexes reflecting microbial richness (Chao1) and diversity (Shannon index) were calculated by the R vegan package. The PERMANOVA was used to evaluate β-diversity [principal coordinates analysis (PCoA)] based on Canberra distance. The linear discriminant analysis (LDA) effect size (LEfSe) was performed to identify the differential taxa between groups, and the microbiota with p values <0.05 and LDA scores >2 were considered statistically significant.

Data were given as means with standard deviation (mean ± SD), or medians with ranges. Statistical analysis was performed in SPSS software package version 22.0 (SPSS Inc., United States) and GraphPad Prism 6 (GraphPad Software, United States). The Student’s t-test or Mann-Whitney U test was used for group comparison, and p value <0.05 was considered as a significant difference.

As shown in Figure 1A, the H&E staining results showed that 5-weeks CCl₄ intervention caused liver injury indicated by the structural disorder of the hepatic cord, swollen hepatocytes, and mild inflammatory cell infiltration without obvious hepatocyte necrosis. Although the serum parameters reflecting hepatocytes injury including ALT, AST, and TBIL were unchanged (Supplementary Figures S1A–C), the serum ALP indicating cholangitis was significantly increased in mice with 5-weeks CCl₄ compared with those with vehicles (Supplementary Figure S1E). And the serum TBA and γ-GT were also unchanged in CCl₄-treated mice (Supplementary Figures S1D,F). Following liver injury, the Masson’s trichrome and Sirius red staining results presented collagen deposition in the portal area which was extended into the hepatic lobule (Figures 1B,C), and the morphometric analysis results showed that the percentage of the positive area of Masson’s trichrome and Sirius red staining was significantly higher in CCl₄-treated mice than NC after 5 weeks of intervention (Figures 1D,E). These results showed that 5-weeks CCl₄ intervention induced liver fibrosis in mice.

The 15-weeks CCl₄ administration resulted in more severe liver injury indicated by a stretched and deformed bile duct, moderate inflammatory cell infiltration, and hepatocytes enlargement shown by H&E staining (Figure 1A). Meanwhile, the serum cholangitis markers ALP and γ-GT were increased and the serum levels of TBIL and TBA were higher in CCl₄-treated mice compared with NC (Supplementary Figures S1C–F). Similar to findings after 5 weeks, the ALT and AST levels were unchanged (Supplementary Figures S1A,B) and there was no obvious hepatocyte necrosis (Figure 1A). Parallel with the progression of liver injury, Masson’s trichrome and Sirius red staining exhibited marked fibrogenesis and pseudo-lobule formation in mice with 15-weeks CCl₄ treatment (Figures 1B,C). And the positive staining area of Masson’s trichrome and Sirius red staining was much higher in CCl₄-treated mice than NC (Figures 1D,E). These results evidenced that mice with 15-weeks CCl₄ administration developed liver cirrhosis without necrosis, which was consistent with previous findings that showed that CCl₄ may induce liver cirrhosis by causing bile duct injury and abnormal bile excretion rather than hepatocyte necrosis (Horikoshi et al., 2015; Li et al., 2018).

In order to truly reveal the microbial alterations in the development of liver cirrhosis, metagenomic sequencing was performed. There was no significant difference between CCl₄-treated mice and NC in Chao l and Shannon indexes after 5 and 15-weeks intervention (Supplementary Figures S2A,B), indicating that microbial richness and diversity were unchanged during the development of liver cirrhosis. However, as shown in Figures 2A,B, the PCoA plots displayed the distinct microbial composition between mice with 5 or 15-weeks CCl₄ treatment (PERMANOVA, p = 0.021 and 0.011, respectively). Furthermore, LEfSe analysis based on microbes from phylum, class, order, family, and genus was conducted to identify the microbiota associated with the development of liver cirrhosis. As shown in Figure 2C, mice with 5-weeks CCl₄ showed an increase of microbiota from Verrucomicrobia including Verrucomicrobiae, Verrucomicrobiales, Verrucomicrobiaceae, and Akkermansia compared with those with vehicles. Meanwhile, the gut microbiota from Actinobacteria including Bifidobacteriales, Bifidobacteriaceae, and Bifidobacterium, and Lactobacillaceae as well as its genera Lactobacillus were also significantly increased in CCl₄-treated mice. The class Gamaproteobacteria as well as Enterobacteriales and its common pathogens Enterobacteriaceae were enriched in mice with 5-weeks CCl₄ intervention. Reversely, the microbes of Deltaproteobacteria including Desulfovibrionales, Desulfovibrionaceae, and Desulfovibrio were decreased in the CCl₄-treated group compared to NC. The genera Barnesiella and Capnocytophaga were also reduced in mice with 5-weeks CCl₄ treatment. These results showed altered microbial composition in the liver fibrosis mice induced by 5-weeks CCl₄.

For the liver cirrhosis mice induced by 15-weeks CCl₄, gut microbiota from Lactobacillales including Streptococcaceae and Streptococcus were decreased compared with NC. Formicrobes from Bacteroidetes and Dysgonomonas were increased while taxa of Sphingobacteriia (Sphingobacteriales, Sphingobacteriaceae), Flavobacteriia (Flavobacteriales, Flavobacteriaceae), and Rikenellaceae and its genera Alistipes were decreased in mice of the CCl₄ group compared to NC. Unlike the findings of 5-weeks intervention, Barnesiella was increased in CCl₄-treated mice. Notably, consistent with the trend in liver fibrosis mice, liver cirrhosis mice also presented with enriched microbiota from Verrucomicrobia including Verrucomicrobiae, Verrucomicrobiales, Verrucomicrobiaceae, and Akkermansia, and the depletion of Deltaproteobacteriaas well as its order Desulfovibrionales, family Desulfovibrionaceae, and genus Desulfovibrio (Figure 2D). These findings support the vital role of gut microbiota, especially Verrucomicrobia and Deltaproteobacteria, in the development of liver cirrhosis.

Due to the distinct microbiota from phylum to genus levels during the development of liver cirrhosis, we further analyzed the microbes at species level. Among all of the predominant species with the average prevalence of more than 0.1% in any groups, 13 species were significantly altered in mice with 5-weeks or 15-weeks CCl₄ treatment when compared with NC. As shown in Figures 2E,F, four species including Akkermansia muciniphila, Bacteroides dorei, Prevotella dentalis, and Bacteroides vulgatus were progressively increased while two species including Bacteroides fragilis and Bacteroides uniformis were progressively decreased during the development of CCl₄-induced liver cirrhosis, and showed a significant difference between CCl₄-treated mice and NC after 15-weeks intervention, indicating modulation of these six species might prevent liver cirrhosis. The significant increase of Parabacteroides goldsteinii, and Bacillus cereus, and decrease of Bacteroides coprocola only presented in mice with liver fibrosis. While the liver cirrhosis mice were featured with increased Parabacteroides merdae and decreased Bacteroides ovatus. And Barnesiella intestinihominis and Alistipes shahii showed an opposite alteration trend in liver fibrosis and liver cirrhosis mice when compared with NC. These findings further suggest that gut dysbiosis is involved in the development of liver cirrhosis, suggesting microbial intervention might be a potential target for preventing and treating liver cirrhosis.

The compositional alterations of gut microbiota prompted us to further analyze the metabolic disorders of microbial community by fecal metabolomics, which provided a functional readout of microbiota. As shown in Figures 3A,B, the PLS-DA plots based on untargeted fecal metabolomics data of both positive and negative modes showed distinct metabolomics profiles of mice with liver fibrosis and liver cirrhosis from NC. The fecal metabolites with the top 15 VIP values of both positive and negative modes were chose as the differential metabolites involved in the development of liver cirrhosis (Figures 3C,D).

As shown in Figure 3C; Supplementary Table S1, among the 30 differential metabolites in liver fibrosis mice with 5-weeks CCl₄ intervention, 8 metabolites belong to purine metabolism and 7 were metabolites of pyrimidine metabolism. Among these 15 purine and pyrimidine metabolites, 3 metabolites including adenine, cytosine, and deoxycytidine were enriched in mice with liver fibrosis, while 7 purine metabolites (deoxyinosine, 2-hydroxyadenine, inosine, hypoxanthine, N6-methyladenosine, adenosine, and deoxyguanosine), and 5 pyrimidine metabolites (2′-deoxyuridine, thiamine, thymine, thymidine, and ribothymidine) were decreased in mice with CCl₄-treatment compared with those with vehicles. Four organic compounds [3-(3-Hydroxyphenyl)propanoic acid, daidzein, cellobiose, and zearalenone] were decreased, and D-mannitol was increased in mice with CCl₄ compared with NC. The primary bile acid cholic acid and the secondary bile acid lithocholic acid were increased while the conjugated taurocholate was decreased in CCl₄-treated mice. The lipid metabolites of palmitic acid and pentadecanoic acid were also significantly altered in CCl₄-treated mice. The CCl₄ treatment also disturbed the metabolism of pyruvate, vitamin B6, betaine, histidine, and inositol, indicated by altered DL-lactate, 4-pyridoxic acid, betaine aldehyde, histamine, and myo-Inositol compared with NC. The pathway enrichment analysis using these 30 metabolites showed that liver fibrosis mice were characterized by remarkable disturbance of purine (p < 0.001) and pyrimidine metabolism (p = 0.005) (Figure 3E).

As shown in Figure 3D and Supplementary Table S2, consistent with the liver fibrosis mice, the liver cirrhosis mice induced by 15-weeks CCl₄ treatment also showed obvious alterations in purine and pyrimidine metabolites, and pathway enrichment also pointed out the dysregulation of purine (p = 0.034), and pyrimidine (p = 0.006) metabolic pathways (Figure 3F). Similar to the trend in the liver fibrosis mice, cytosine and adenine were increased, and 2′-deoxyuridine, thymidine, thymine, and deoxyinosine were decreased in mice with liver cirrhosis compared with NC. Meanwhile, the enrichment of cholic acid, lithocholic acid, and the depletion of zearalenone, myo-Inositol, and 4-pyridoxic acid were also shown in mice with 15 weeks of CCl₄ treatment. Unlike the trend at 5 weeks, adenosine and taurocholate were increased in mice with 15-weeks CCl₄ treatment. Besides, 5-methylcytosine of pyrimidine metabolites and deoxyadenosine of purine metabolites were increased, while uracil of pyrimidine metabolites, and the primary bile acid chenodeoxycholate were decreased in mice with 15-weeks CCl₄ compared with those with vehicles. Four lipid metabolites including arachidonic acid, phytosphingosine, cis-9-palmitoleic acid, and sphingosine were increased in CCl₄-treated mice. The mice with 15-weeks CCl₄ treatment also showed the disorders of amino acid metabolism indicated by the increased N-acetyl-D-glucosamine, L-citrulline, L-tryptophan, p-hydroxyphenylacetic acid, L-isoleucine, Glycyl-L-leucine, and decreased tryptamine. The ribitol of organic compounds and L-malic acid of pyruvate metabolism were also obviously altered in feces of mice with 15-weeks CCl₄ treatment. Taken together, these metabolomics findings revealed the distinct metabolic function of gut microbiota during the development of liver cirrhosis. Notably, the differential metabolites mainly pointed to the disorders of purine and pyrimidine metabolism.

Due to the obvious disturbance of purine and pyrimidine metabolism, we compared all detected metabolites involved in purine and pyrimidine metabolism pathways. As shown in Figures 4A,B,E, in total seven pyrimidine metabolites were detected in feces of all mice. For two pyrimidine deoxyribonucleosides, deoxycytidine was significantly increased while the 2′-deoxyuridine was decreased in mice with 5-weeks CCl₄-induced liver fibrosis. The thymidine of pyrimidine nucleosides, a known microbial metabolite was decreased, while its downstream dTMP was increased in liver fibrosis mice, indicating the enhanced microbial conversion from thymidine to dTMP. Consistently, the depletion of 2′-deoxyuridine and thymidine was also presented in the liver cirrhosis mice with 15-weeks CCl₄ treatment. And the nucleobase uracil, thymine, and uracil derivative dihydrouracil were comparable between CCl₄-treated mice and NC after 5 or 15 weeks of intervention.

In terms of compounds related to purine metabolism, 10 metabolites were identified in our study (Figures 4C,D,F). The nucleotide AMP was significantly increased and its nucleoside adenosine was decreased in mice with liver fibrosis compared with NC. The inosine which could be deaminated from adenosine was also decreased in liver fibrosis mice. While the adenosine was increased in mice with liver cirrhosis. The adenine is the purine base of nucleoside deoxyadenosine, which could be deaminated into deoxyinosine. The mice with liver fibrosis showed the increase of adenine and decrease of deoxyinosine in feces. Interestingly, the liver cirrhosis mice exhibited increased adenine and deoxyadenosine, and decreased deoxyinosine, indicating the enhanced synthesis, and blocked degradation of deoxyadenosine in microbial community during the development of liver cirrhosis. The purine nucleoside xanthosine, also a microbial metabolite, as well as its base xanthine were decreased in mice with liver fibrosis or liver cirrhosis compared with NC. The hypoxanthine and the end product of purine metabolism uric acid were unchanged in CCl₄-treated mice. These findings supported the fact that the disturbance of pyrimidine and purine metabolism was involved in the development of liver fibrosis.