Administration of Lactobacillus salivarius LI01 or Pediococcus pentosaceus LI05 prevents CCl4-induced liver cirrhosis by protecting the intestinal barrier in rats

Alterations in the gut microbiome have been reported in liver cirrhosis, and probiotic interventions are considered a potential treatment strategy. This study aimed to evaluate the effects and mechanisms of Lactobacillus salivarius LI01, Pediococcus pentosaceus LI05, Lactobacillus rhamnosus GG, Clostridium butyricum MIYAIRI and Bacillus licheniformis Zhengchangsheng on CCl4-induced cirrhotic rats. Only administration of LI01 or LI05 prevented liver fibrosis and down-regulated the hepatic expression of profibrogenic genes. Serum endotoxins, bacterial translocations (BTs), and destruction of intestinal mucosal ultrastructure were reduced in rats treated with LI01 or LI05, indicating maintenance of the gut barrier as a mechanism; this was further confirmed by the reduction of not only hepatic inflammatory cytokines, such as TNF-α, IL-6, and IL-17A, but also hepatic TLR2, TLR4, TLR5 and TLR9. Metagenomic sequencing of 16S rRNA gene showed an increase in potential beneficial bacteria, such as Elusimicrobium and Prevotella, and a decrease in pathogenic bacteria, such as Escherichia. These alterations in gut microbiome were correlated with profibrogenic genes, gut barrier markers and inflammatory cytokines. In conclusion, L. salivarius LI01 and P. pentosaceus LI05 attenuated liver fibrosis by protecting the intestinal barrier and promoting microbiome health. These results suggest novel strategies for the prevention of liver cirrhosis.

Bacterial translocation and identification. Samples from the left lobe of the liver and mesenteric lymph nodes (MLN) were collected, weighed, and milled under aseptic conditions in glass homogenizers that had been autoclaved at 121 °C for 15 min. After appropriate dilution with physiological saline, the milled MLN samples and samples from arterial blood or portal blood were separately incubated under aerobic and anaerobic conditions on brain heart infusion agar (BHI, Oxoid, Thermo Fisher Biochemicals Ltd., Beijing, China) at 37 °C for 48 to 72 h.
Histological examination. For histological evaluation, all liver samples and the terminal ileums were immediately fixed in 10% neutral buffered formalin at the time of death. Paraffin-embedded samples were cut into 4-μm-thick sections. Subsequently, the liver sections were stained with Masson's trichrome (MT). Staining intensity was quantified by histomorphometry with Image-Pro Plus 6.0. Sections from the ileum were processed for H&E and Alcian blue staining and analysed by a pathologist who was blind to the groups. For scoring the degree of injury to the terminal ileum, the inflammatory cell infiltration (score, 0-4), goblet cell depletion (score, 0-4), mucosa thickening (score, 0-4), destruction of architecture (score, 0 or 3-4) and loss of crypts (score, 0 or 3-4) were also semi-quantitatively assessed 12 . Immunohistochemical staining and immunofluorescence. Liver sections (4 µm) were dewaxed, rehydrated regularly and treated with 3% H 2 O 2 . The specimens were incubated overnight at 4 °C with anti--SMA (1:500, Invitrogen). The cryostat sections of the terminal ileums were incubated with anti-ZO-1 (1:100, Invitrogen) antibodies overnight at 4 °C, followed by incubation with FITC-conjugated goat anti-rabbit secondary antibody (1:100, Beyotime) for 1 h at room temperature. The sections were then washed in PBS, mounted with mounting medium (Beyotime) and photographed under a fluorescence microscope (Eclipse 80i; Nikon, Tokyo, Japan). LI05 group, cirrhotic rats treated with P. pentosaceus LI05; MY group, cirrhotic rats treated with C. butyricum; ZCS group, cirrhotic rats treated with B. licheniformis; LGG group, cirrhotic rats treated with L. rhamnosus GG; NS group, cirrhotic rats treated with normal saline. (B) Survival times of the Control, NS, LI01, LI05, MY, ZCS, and LGG groups. LI01 group vs. NS group (P < 0.1), LI05 vs. NS group (P < 0.05). (C) Positive α-SMA or MT staining was analysed with Image-Pro Plus 6.0 software. (D) Representative images of liver histology and fibrillar collagen deposition assessed by MT staining. Activated HSCs evaluated by immunohistochemical staining of α-SMA from the Control, NS, LI01, LI05, MY, ZCS, and LGG groups. Scale bar: 250 μm. All data are given as the mean ± SEM. * P < 0.05 and ** P < 0.01 vs. normal saline-treated cirrhotic rats in the NS group.
DNA and RNA extraction. DNA was extracted from the caecal contents using a Qiagen stool kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Total RNA was isolated from liver samples and intestinal segments using an RNeasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The integrity of the DNA and RNA was verified by ethidium bromide staining. DNA and RNA extracts were stored until use at −20 °C and −80 °C, respectively.
Real-time RT-PCR. The relative mRNA expression was measured in triplicate via the comparative cycle threshold method using a 7500 real-time PCR system (Applied Biosystems). The primer sequences (Integrated DNA Technologies) are shown in Supplementary Table 1. We used the housekeeping gene β-actin as the internal standard.
Sequencing and bioinformatics. DNA from the isolated caecal contents was used as a template for the amplification of the 16 S rRNA V3 and V4 hypervariable region using the fusion dual barcoded PCR primers F319 (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806 R (5′-GGACTACHVGGGTWTCTA AT-3′). Duplicate PCR products were pooled in equimolar concentrations and then purified using the Qiagen PCR purification kit (Qiagen, Hilden, Germany). Sequencing was performed on an Illumina MiSeq Instrument (Illumina Inc., San Diego, CA) using the 300 bp paired-end protocol.
Overlapping paired-end reads from the original DNA fragments were merged using FLASH (version 1.2.10). Quality control was conducted using FastQC (version 0.10.1). The assembled sequences were clustered using the CD-hit-est-based clustering method 14 . PyNAST software (http:/qiime.org/pynast/) was used to analyse and calculate the numbers of sequences and operational taxonomic units (OTUs) for each sample. Next, the sequences were grouped into various OTUs using Felsenstein-corrected similarity matrices; the sequences within an OTU shared at least 97% similarity. The Ribosomal Database Project (RDP) classifier, which is available from the RDP website (http://rdp.cme.msu. edu/classifier/), was used to classify the 16 S rDNA into distinct taxonomic categories by aligning the sequences to a curated database of taxonomically annotated sequences 15 . All 16 S rDNA sequences were mapped to the RDP database using BLASTN to determine the taxonomic assignments. Sequences sharing greater than 97% identity were used to associate a group of OTUs to specific species, whereas those with less than 97% identity were considered novel reads. The microbial diversity in individual conjunctival samples was estimated using rarefaction analysis, the Shannon diversity index (SDI), and the Chao 1 index. Principal coordinate analysis (PCoA) was performed for each sample using OTUs and the ade4 package within the R program (R version 2.15, http://www.R-project.org).

Statistical analysis.
For results of biochemical assays and histology score, the Kolmogrov-Smironv test was used to check for normality. Statistical differences between groups were estimated by Kruskal-waills test followed by the Mann-Whitney U test. The frequency of bacterial translocation was compared using the chi-squared test. Analysis of Similarities (ANOSIM) was used to test for clustering of microbial communities using weighted and unweighted UniFrac distance matrices. Wilcoxon rank sum test combined with the Benjamini-Hochberg method was applied to compare bacteria taxa. Correlations between variables were computed using the Spearman rank correlation. The values were presented as mean ± SEM if normally distributed; otherwise the values were presented as median (25th and 75th). P < 0.05 was considered significant. The data were analysed using SPSS version 20.0 for Windows (SPSS Inc., Chicago, IL, USA).

Results
L. salivarius LI01 and P. pentosaceus LI05 prolonged the survival time and protected against weight loss of CCl 4 -induced liver cirrhosis. Morality during this study was 53% in the MY group, 47% in the ZCS group, 27% in the LI01 group, 31% in the LI05 group (two rats were suffocated to be removed the group), 47% in the LGG group, and 47% in the NS group (Fig. 1B). Compared with the rats in NS group, the LI05 groups showed significantly increased survival time (P < 0.05, 92.3 ± 3.0 vs. 78.5 ± 7.6 days). We also observed a trend of increased survival time with LI01 supplementation (P < 0.1, 89.1 ± 4.2 vs. 78.5 ± 7.6 days; Table 1). No significant differences were found between the other probiotic groups. Body weight was an important indicator Values are expressed as the mean ± SEM. * P < 0.05, ** P < 0.01 compared with the NS group.
of rat health status during CCl 4 administration. Notably, the body weights in the LI01 group (414.3 ± 11.2 g) and the LI05 group (412.3 ± 7.4 g) were significantly higher than those in the NS group (362.2 ± 9.2 g; P = 0.003 and P = 0.001, respectively) at the 6th week before ascites formation (Table 1), whereas the other probiotic groups showed no differences.
L. salivarius LI01 and P. pentosaceus LI05 improved liver functions and ameliorated liver fibrosis. Liver function analysis (Table 1) revealed that administration of P. pentosaceus LI05 significantly relieved the increase in serum ALT, AST and GGT (P < 0.05 and P < 0.01, respectively). Treatment with L. salivarius LI01 slightly decreased ALT (P = 0.07) and significantly decreased GGT level compared with the rats in NS group (P < 0.01). Conversely, treatment with C. butyricum, B. licheniformis or L. rhamnosus GG did not have a significant effect. After continuous CCl 4 administration, typical pathological features of liver fibrosis were confirmed. Evaluation of the collagen content by MT staining of cirrhotic rats revealed increased collagen around the portal triad, lobules surrounded by bundles of collagen fibres, and the presence of large fibrous septa (Fig. 1C,D). Conversely, liver fibrosis and collagen deposition were only ameliorated in the rats treated with L. salivarius LI01 or P. pentosaceus LI05.
L. salivarius LI01 and P. pentosaceus LI05 inhibited hepatic stellate cell (HSC) activation and attenuated the hepatic inflammatory response. The excessive production of extracellular matrix (ECM) during the development and progression of liver fibrosis is predominantly due to HSCs. Using immunohistochemical staining of α-SMA, a specific marker of activated HSCs, we found that the α-SMA-positive areas were substantially reduced by administration of probiotic LI01 or LI05 (Fig. 1C,D), whereas the livers in other strain-treated groups exhibited no differences in α-SMA compared to those of the NS group.
The following profibrogenic genes were significantly down-regulated in various groups compared with their expression in the NS group: Collagen α1 showed decrease in group LI01 and 1.5-fold decrease in group LI05; TIMP-1 showed 1.7-fold decrease in group ZCS, 1.6-fold decrease in group LI01 and 1.9-fold decrease in group LI05; TGF-β showed 1.5-fold decrease in group LI01 and 1.4-fold decrease in group LI05; and INOS-2 showed 1.9-fold decrease in group LI01, 1.8-fold decrease in group LI05 and 1.9-fold decrease in group LGG ( Fig. 2A).  TLR family members TLR2, TLR4,  TLR5 and TLR9. Gene expression was determined by quantitative PCR analysis of the total mRNA extracted from liver fragments. Results are expressed as fold change relative to NS group. All data are given as the mean ± SEM. * P < 0.05 and ** P < 0.01 vs. normal saline-treated cirrhotic rats in the NS group.
Inflammatory cytokines play a key role in signal activation of liver fibrosis. As shown in Fig. 2B, IL-6 mRNA level showed 2.2-fold down-regulation in group LI01 and 2.6-fold down-regulation in group LI05 compared with its expression in cirrhotic rats from the NS group; TNF-α mRNA level showed 1.5-fold down-regulation in group ZCS, 1.9-fold down-regulation in group LI01 and 1.6-fold down-regulation in group LI05; IL-17A mRNA level showed 3.2-fold down-regulation in group LI01 and 5.2-fold down-regulation in group LI05; IL-10 expression was unchanged except in the Control group (7.1-fold up-regulated, P < 0.01). Notably, the administration of C. butyricum or L. rhamnosus GG did not significantly affect inflammatory cytokine regulation.
L. salivarius LI01 and P. pentosaceus LI05 reduced plasma endotoxin levels and modulated TLR gene expression. Endotoxin is believed to be a key factor that contributes to the progression of liver damage.
We tested the plasma endotoxin levels to evaluate the effect of the different probiotic strains on endotoxaemia (Table 1). Rats from the NS group (1.2 ± 0.1 EU/ml) showed significantly higher plasma endotoxin levels than those of rats from the Control group (0.4 ± 0.1 EU/ml, P < 0.01). A significant decrease in endotoxaemia was observed in the LI01 group (0.5 ± 0.1 EU/ml) and the LI05 group (0.5 ± 0.1 EU/ml) compared to that in the NS group (P < 0.01). There were no significant differences between the other probiotic groups and the NS group.
Endotoxin is an important ligand that activates specific Toll-like receptors (TLRs). Mechanistic investigations revealed that probiotic treatment strongly induced the expression of TLRs (Fig. 2C), which are involved in regulation of inflammation and bacterial infection. We observed significantly down-regulated expression of TLR2 in the LI01 group; TLR4 in all groups except MY and ZCS, and TLR5 and TLR9 in the LI01 and LI05 groups compared with the expression of these receptors in the cirrhotic rats in the NS group.
L. salivarius LI01 and P. pentosaceus LI05 strongly reduced BT. As shown in Table 2, only treatment with L. salivarius LI01 decreased BT in all evaluated tissues compared with that of the cirrhotic rats treated with normal saline as follows: mesenteric lymph node (MLN) (4/11, P = 0.026), arterial blood (4/11, P = 0.026), portal blood (4/11, P = 0.026) and liver (3/11, P = 0.04). Moreover, treatment with P. pentosaceus LI05 decreased the BT incidence in the MLN (3/9, P = 0.024), arterial blood (3/9, P = 0.024) and portal blood (2/9, P = 0.007), treatment with C. butyricum decreased the BT incidence in the portal blood of the MY group (2/7, P = 0.02), and treatment with L. rhamnosus decreased the BT incidence in the arterial blood of rats in the LGG group (3/8, P = 0.039). However, B. licheniformis showed no significant effects on BT incidence. L. salivarius LI01 and P. pentosaceus LI05 improved the intestinal barrier integrity and protected the intestinal microstructure. We then focused on the intestinal barrier to explore the mechanism underlying the observed effect on BT. All cirrhotic groups demonstrated histological abnormalities of the terminal ileum; the level of inflammation was estimated in H&E-and Alcian blue-stained tissue sections of the terminal ileum (Fig. 3A,B), Although there were no differences in goblet cell depletion, mucosal thickening and crypts between each probiotic group, we observed L. salivarius LI01 or P. pentosaceus LI05 improved ileal inflammation marked by lower infiltration of leukocytes and more complete villi architecture compared to the rats in the NS group (Fig. 3E, Supplementary Table 2). Bacteria can penetrate the inner mucosal layer and contact the epithelium to elicit a beneficial or harmful effect. Therefore, we detected the ultrastructure of the ileal mucosa in the experimental groups to evaluate the microscopic changes in the intestinal mucosal integrity (Fig. 3C). The microvilli of the intestinal epithelial cells in the NS, MY, ZCS, and LGG groups were ruptured, sparse and stunted. However, the injured ultrastructure was substantially ameliorated in the LI01 and LI05 groups. Since the loosening of tight junctions is a major factor that contributes to pathological BT, we speculated that the probiotic strains LI01 and LI05 protected the intestinal barrier by enhancing these connections. To address this hypothesis, we measured ileal ZO-1 expression by quantitative PCR and immunofluorescence staining. In contrast with the other experimental groups, groups LI01 and LI05 showed higher levels of ZO-1 mRNA expression, which were verified by clear, uniform positive distributions of ZO-1 protein at the apical region of the intestinal epithelium (Fig. 3D,F). Notably, all probiotic strains could promote antimicrobial peptide β-defensin-1 secretion compared to the lowest mRNA level in the NS group (Fig. 3G).
L. salivarius LI01 and P. pentosaceus LI05 increased the microbial richness and ameliorated microbiome dysbiosis. To gain further insights into the processes involved in BT reduction by probiotics, we investigated its impact on microbiota composition of caecal content by next-generation sequencing. The number of observed OTUs and the Chao1 index, the estimated microbial richness value, were significantly higher in  Table 2. Incidence of bacterial translocation in the experimental groups. Results are shown as the number of rats with positive culture in different tissues; percentages are parenthesized. * P < 0.05, ** P < 0.01 compared with the NS group. MLN, mesenteric lymph nodes.
the ZCS, LI01 and LI05 groups than those in the NS group (P < 0.01, Table S3). However, the overall microbial diversity was not significantly different between experimental groups based on the Shannon and Simpson indices (Table S3). Cirrhosis and probiotic treatment had a significant impact on the bacterial communities based on the PCoA plots (Fig. 4A, Figure S1). The unweighted and weighted UniFrac PCoA, which measured the phylogenetic similarities between microbial communities, showed a marked difference between the Control group and the NS group (Unweighted, P = 0.001, r = 0.82; Weighted, P = 0.002, r = 0.39; Table S4). Furthermore, weighted and Unweighted UniFrac analysis showed that the microbiota of the LI01 and LI05 group clustered separately from the NS group (ANOSIM, P < 0.01). In contrast, weighted metrics showed the microbiota of MY and LGG group had several intersection points with the NS group, indicating the closer microbial communities in these groups (ANOSIM, P > 0.05). On the other hand, the unweighted UniFrac analysis showed that the distinct microbial structure of the LI01 and LI05 group significantly separate from other probiotic groups (ANOSIM, P < 0.01, Table S4). Despite the highly diverse bacterial communities and interindividual differences, cirrhosis clearly affected the intestinal microbial community. Compared with the controls, the rats with cirrhosis in the NS group demonstrated a marked increase in the Firmicutes/Bacteroidetes (F/B) ratio (1.7130 ± 0.3771 vs. 0.8183 ± 0.1113, P < 0.01). Additionally, only the probiotic strains LI01 (0.8287 ± 0.0630, P < 0.01) and LI05 (0.8329 ± 0.1512, P < 0.05) significantly decreased the ratio (Fig. 4B). Although Lactobacillus comprised a small proportion of microbial sequences, we found that its relative abundances were significantly higher in the LI01 and LI05 microbiome compared with those of the NS microbiome (Fig. 4C). The ten most abundant taxa that differed between the Control and NS microbiome at the family and genus levels are presented in Fig. 4D. Among these discriminatory taxa, 5 bacterial families and 4 genera were more abundant in the Control microbiome, and the other 8 microbial taxa were strongly enriched in the NS microbiome, including the families Sutterellaceae, Enterobacteriaceae, Erysipelotrichaceae, and Pasteurellaceae and the genera Parasutterella, Escherichia/Shigella, Odoribacter, Pseudoflavonifractor. Next, we used linear discriminant analysis effect size (LEfSe) to compare the estimated phylotypes of the probiotics and NS microbiome. LEfSe demonstrated that compared to the NS group, the phylum Bacteroidetes and the genera Elusimicrobium, Paraprevotella, and Prevotella were both enriched in LI01 and LI05 group, whereas the NS group was characterized by a preponderance of Firmicutes, Enterobacteriaceae, Oscillospira, Oscillibacter, Flavonifractor and Escherichia/Shigella (Fig. 5A-D).

Discussion
The major finding in the present study was that long-term administration of L. salivarius LI01 or P. pentosaceus LI05 prolonged the survival time and substantially ameliorated CCl 4 -induced liver cirrhosis in rats. Compared with widely used probiotic strains (L. rhamnosus GG, C. butyricum MIYAIRI or B. licheniformis Zhengchangsheng), the probiotic strains LI01 and LI05 achieved this effect through several complementary mechanisms, including the following: i) alleviation dysbiosis of gut microbiota; ii) improvements in the intestinal barrier function; iii) decreased BT; and iv) reduced liver inflammatory response.
In response to liver injury, quiescent HSCs are activated and develop a myofibroblast-like phenotype that proliferates and expresses intermediate filament α-SMA and profibrotic genes 16 . We found that in the LI01 and LI05 Figure 5. Treatment with L. salivarius LI01 or P. pentosaceus LI05 ameliorated microbiome dysbiosis in cirrhotic rats. (A-D) LEfSe prediction was used to identify the most differentially abundant taxa in the LI01 or LI05 group compared to the NS group. (A,C) The cladogram shows the phylogenetic relationship between the microbial taxa that were higher in probiotic samples compared to those in the cirrhotic rats in the NS group. (B,D) LDA scores showed significant bacterial differences between the NS and probiotic groups. Only the taxa meeting a significant LDA threshold value of >3.6 are shown. Red indicates the LI01 or LI05 group, and green indicates the NS group. groups, HSC activation was inhibited, the hepatic inflammatory response was decreased, and the expression of inflammatory mediators and profibrogenic cytokine genes, including TNF-α, TGF-β, INOS-2, IL-6 and IL-17A, were down regulated. Our results are consistent with other studies in which probiotics showed anti-inflammatory effects in experimentally induced cirrhosis 17,18 . TGF-β is a major factor in the promotion of fibrogenesis in the liver and other tissues. Proinflammatory cytokines, such as IL-6, TNF-α and IL-17A, are key stimulators of HSC activation 19,20 . The down-regulation of inflammatory cytokine expression may be related to lower endotoxin or bacterial DNA levels from the gut 21 . Furthermore, the down-regulation of INOS-2 gene expression indicated another potential mechanism mediating the anti-inflammatory effects of LI01 and LI05, which are likely related to their antioxidative inhibition of proinflammatory signalling in the chronically injured liver. Liver function is indicated by ALT, AST and GGT levels, which were reduced in the groups treated with the LI01 and LI05 strains, resulting in the alleviation of both hepatocyte damage and liver inflammation.
HSCs provide a link between the gut and the liver through their high expression of TLRs, which are pattern recognition receptors (PRRs) involved in sensing and eliminating antigens and thereby promote HSC activation and fibrosis 22 . The tissue sections with mild liver fibrosis in the LI01 and LI05 groups showed depressed TLR family gene expression. TLR2 promotes hepatic fibrosis-mediated injured intestinal permeability, leading to increased liver exposure to bacterial products from the gut microflora, which promote liver injury, inflammation, and fibrosis 23 . TLR4 plays a key role in the development of liver fibrosis; lipopolysaccharide (LPS) ligand from the gut microbiota activates TLR4 on HSCs, influencing downstream signalling and resulting in myofibroblast proliferation and ECM production 22 . TLR5 and its ligand bacterial flagellin are directly involved in the progression of fibrosis via activation of the NF-κ B and MAPK signalling pathways 24 . TLR9, which is activated by bacterial DNA, is also required for liver fibrosis through the induction of IL-1β, which in turn activates the IL-1 receptor on HSCs, resulting in their activation 25 . Additionally, ECM is involved in hepatic fibrosis and is regulated by the specific inhibitor tissue inhibitor of metalloproteinases (TIMPs) 26 . The decreased expression of TIMP-1 in several probiotic strains suggested their beneficial effects on ECM regulation during fibrogenesis.
In the clinic, intestinal barrier function is impaired in patients with cirrhosis and parallels cirrhosis progression 27 . The improved hepatic fibrosis in the LI01 and LI05 groups was associated with their protective effect on the integrity and microstructure of the intestinal barrier and the resultant reduction in BT. This finding is similar to the results of previous studies showing that modulating the gut microbiota to correct intestinal barrier function and reduce BT and endotoxaemia could contribute to improving liver fibrogenesis 17,28 . BT is considered to play a pivotal pathophysiological role in the development of severe complications, such as ascites, renal failure and hepatic encephalopathy 29 . The mechanism by which probiotics reduced BT can be simply summarized as gut flora homeostasis, which improves intestinal barrier functions and reduces inflammatory immune responses 18,30 . Interestingly, in our present study, five probiotic strains increased the ileal expression of the antimicrobial peptide β-defensin-1. Antimicrobial peptides act as intestinal immune defences by maintaining intestinal barrier homeostasis 31 . The increased expression of β-defensin-1 may partially explain the decreased BT in the blood following treatment with L. rhamnosus GG or C. butyricum.
Multiple pathogenic factors accelerating liver injuries were associated with gastrointestinal microbiota dysbiosis. In this study, since we only sequenced partial 16 S rDNA sequences, and the OTUs could not be assigned to the species level with high confidence, none of the five probiotics strains and their genera were observed to be enriched in the caecal microbiome. This is likely because of no significant increases in the amount of these strains in the rat caecum 1 week after probiotic withdrawal. Importantly, the abundances of Sutterellaceae and Escherichia/Shigella (LPS-producing bacteria), Erysipelotrichaceae (inflammation-related bacteria), Pasteurellaceae and Pseudoflavonifractor (opportunistic pathogens) were highly enriched in the cirrhotic rats, whereas beneficial bacteria, such as Akkermansia, were decreased [32][33][34][35] . Probiotic interventions have been shown to effectively modulate gut barrier integrity and gut microbiota in animals by tempering chronic inflammation and metabolic disorders 35 . An increased F/B ratio, caused by an expansion of Firmicutes and/or a contraction of Bacteroidetes has been widely considered a signature of gut dysbiosis 36,37 . The reduction of F/B ratio in LI01 or LI05 group demonstrated L. salivarius LI01 and P. pentosaceus LI05 produces beneficial effects on dysbiosis. Specifically, L. salivarius LI01 and P. pentosaceus LI05 alleviated microbial dysbiosis to promote intestine health, such as the depletion of Escherichia/Shigella, Oscillospira, Oscillibacter, and Flavonifractor and the enrichment of Prevotella, Paraprevotella, Lactobacillus, and Elusimicrobium. These gut bacteria that are specifically altered in probiotic LI01-and LI05-treated rats, especially enriched beneficial microbial taxa and inhibited opportunistic pathogens, may be involved in the amelioration of liver cirrhosis. Consistent with this conclusion, increases in liver injury indicators, profibrogenic genes, and inflammatory cytokines are negatively associated with enrichments of the beneficial taxa in probiotic LI01-and LI05-treated rats, such as the intensive association of Prevotella with IL-6 and TNF-α; Elusimicrobium with IL-6 and IL-17; Lactobacillus with GGT, Collagen α1 and TIMP-1. As reported, Prevotella, which was reported to be more abundant in healthy subjects than that in NAFLD patients, has been suggested to exert protective effects against the development of NAFLD 38 . Elusimicrobium minutum produces acetate and alanine, which may be beneficial for liver damage and the glucose-alanine cycle 39 . Symbiotic-containing Lactobacillus paracasei significantly reduced the expression of α-SMA, collagen, and inflammatory cytokines and improved liver inflammation and fibrosis in the same experimental rat model of liver cirrhosis 17 . However, the Oscillibacter-like organisms, including Oscillibacter and Oscillospira, were identified as potentially harmful gut microbes that mediated high saturated fat diet-induced gut dysfunction; additionally, consistent with our results, a negative correlation was found between the abundance of Oscillibacter and barrier function 40 . LPS is a major product from Escherichia/Shigella, and the overgrowth of these bacteria can impair intestinal permeability and result in endotoxaemia, which is associated with worsening disease severity and complications in cirrhosis 41 . Thus, the reduction in endotoxin level in probiotic LI01-or LI05-treated rats and the strong positive correlation between Escherichia/Shigella and TLRs suggested that the beneficial microbiome regulation of L. salivarius LI01 and P. pentosaceus LI05 ameliorated liver fibrosis. Additionally, gut microbial richness has been correlated with health 42 , and our previous study found a lower abundance of microbial diversity and richness in acute-on-chronic liver failure patients 43 . More speciose bacterial communities reduce pathogens more effectively. Thus, based on the Chao1 index and OTUs, the increase in microbial richness following long-term treatment with the specific probiotics LI01 and LI05 may play an important role in maintaining intestinal homeostasis.
In conclusion, the interaction of gut bacteria with different probiotic strains influences the progression of liver damage, which provides strong evidence that gut microbiome homeostasis contributes to the improvement of liver cirrhosis. The current study potentially provides new insights into the prevention and treatment of liver cirrhosis via selection of an appropriate probiotic strain to manipulate the gut microbiota.