Bifidobacterium pseudocatenulatum LI09 and Bifidobacterium catenulatum LI10 attenuate D-galactosamine-induced liver injury by modifying the gut microbiota

The gut microbiota is altered in liver diseases, and several probiotics have been shown to reduce the degree of liver damage. We hypothesized that oral administration of specific Bifidobacterium strains isolated from healthy guts could attenuate liver injury. Five strains were tested in this study. Acute liver injury was induced by D-galactosamine after pretreating Sprague-Dawley rats with the Bifidobacterium strains, and liver function, liver and ileum histology, plasma cytokines, bacterial translocation and the gut microbiome were assessed. Two strains, Bifidobacterium pseudocatenulatum LI09 and Bifidobacterium catenulatum LI10, conferred liver protection, as well as alleviated the increase in plasma M-CSF, MIP-1α and MCP-1 and bacterial translocation. They also ameliorated ileal mucosal injury and gut flora dysbiosis, especially the enrichment of the opportunistic pathogen Parasutterella and the depletion of the SCFA-producing bacteria Anaerostipes, Coprococcus and Clostridium XI. Negative correlations were found between MIP-1α / MCP-1 and Odoribacter (LI09 group) and MIP-1α / M-CSF and Flavonifractor (LI10 group). Our results indicate that the liver protection effects might be mediated through gut microbiota modification, which thus affect the host immune profile. The desirable characteristics of these two strains may enable them to serve as potential probiotics for the prevention or adjuvant treatment of liver injury.

rats had died. To examine the effects of the Bifidobacterium strains on D-GalN-induced acute liver injury, we first compared the liver function of Bifidobacterium-treated rats with that of positive acute liver injury control (PC) rats. Among the five Bifidobacterium strains used in this study, oral administration with B. longum LI06, B. pseudocatenulatum LI08, B. pseudocatenulatum LI09 and B. catenulatum LI10 significantly reduced the D-GalNinduced increase in alanine aminotransferase (ALT) and total bile acid, but B. longum LI07 did not ( Table 1). The increase in glutamyltransferase induced by D-GalN was significantly reduced by B. longum LI06, B. pseudocatenulatum LI08 and B. catenulatum LI10. In addition, only administration with B. pseudocatenulatum LI09 and B. catenulatum LI10 relieved the increase in aspartate aminotransferase (AST) and glycylproline dipeptidyl aminopeptidase. Therefore, B. pseudocatenulatum LI09 and B. catenulatum LI10 seemed to more efficiently ameliorate the destruction of liver function induced by D-GalN.
To further confirm the effects of these Bifidobacterium strains on liver injury, we examined the liver morphology of rats in different groups. Twenty-four hours after the induction of liver injury, all groups, except the negative acute liver injury control (NC) group, demonstrated hepatic histological abnormalities (Fig. 1). Administration of B. longum LI07, B. pseudocatenulatum LI09 and B. catenulatum LI10 significantly ameliorated D-GalN-induced hepatic degeneration or necrosis and inflammatory cell infiltration. The results from two aspects, liver function and liver histology, both indicate the protective effects of B. pseudocatenulatum LI09 and B. catenulatum LI10 against hepatic injury.

B. pseudocatenulatum LI09 and B. catenulatum LI10 reduced the D-GalN-induced increase in plasma inflammatory cytokines.
Since liver injury is aggravated by inflammatory molecules, we next examined whether these two strains could relieve inflammation by measuring a total of 24 different plasma cytokines. Twenty-four hours after D-GalN treatment, the plasma levels of all detected cytokines increased in the PC group compared with those in the NC group. Administration of B. pseudocatenulatum LI09 significantly alleviated the D-GalN-induced increase in the proinflammatory cytokine interleukin (IL)-1β and the immunomodulatory cytokine IL-10 ( Fig. 2). Among the chemokines, the increase in macrophage inflammatory protein 1 alpha (MIP-1α) and monocyte chemoattractant protein 1 (MCP-1) was attenuated when the rats were treated   Table 1. Effects of pretreatment with five Bifidobacterium strains on liver function during D-GalN-induced acute liver injury. Values are expressed as the mean ± SD. The serum levels of aspartate aminotransferase (AST), alanine aminotransferase (ALT), glutamyltransferase (GGT), glycylproline dipeptidyl aminopeptidase (GPDA), total bile acid (TBA), total bilirubin (TB) and albumin (ALB) were determined after 24 h of D-galactosamine (D-GalN) administration. Compared with the PC group, *p < 0.05, **p < 0.01, ***p < 0.001.
with either B. pseudocatenulatum LI09 or B. catenulatum LI10. However, the increase in another chemokine, MIP-3α, was only ameliorated in B. pseudocatenulatum LI09-treated rats. In addition, the increase in macrophage colony-stimulating factor (M-CSF) was also alleviated in the groups treated with these two strains. These results suggest that the reduction of chemokine levels may play a pivotal role in the process by which Bifidobacterium improves liver injury. By comparison, Administration of none of the B. longum LI06, B. longum LI07 or B. pseudocatenulatum LI08 could alleviate the increase of all these chemokines simultaneously (see Supplementary Fig. S1).

B. pseudocatenulatum LI09 and B. catenulatum LI10 ameliorated the D-GalN-induced bacterial translocation.
To determine how the inflammatory reaction could be attenuated, we investigated the gut barrier function, since oral administration of Bifidobacterium strains first and directly impacts gut. We examined the bacterial translocation to evaluate the protective effects of these two strains on D-GalN-induced destruction of the gut barrier. As shown in Table 2, administration of either B. pseudocatenulatum LI09 or B. catenulatum LI10 reduced the population of bacteria that translocated to mesenteric lymph nodes (MLNs), indicating the restorative effects of these two strains on gut barrier function. By contrast, administration of none of the B. longum LI06, B. longum LI07 or B. pseudocatenulatum LI08 could cut down the population of bacteria which translocated to MLNs (see Supplementary Table S1).

B. pseudocatenulatum LI09 and B. catenulatum LI10 attenuated the D-GalN-induced terminal ileum injury.
Since the integrity of the intestinal mucosa is an important structure that supports the gut barrier function, we explored whether the intestinal mucosa integrity was damaged during D-GalN-induced liver injury and whether administering these strains improved the condition. As evaluated by haematoxylin and eosin (H&E) staining 17 , pretreatment with B. pseudocatenulatum LI09 or B. catenulatum LI10 alleviated the D-GalN-induced histological abnormalities of the terminal ileum by decreasing the incidence of subepithelial Gruenhagen's space and epithelial lifting (Fig. 3a). In comparison, pretreatment with none of the B. longum LI06, B. longum LI07 or B. pseudocatenulatum LI08 could alleviate the histological abnormalities significantly (see Supplementary Fig. S2). We further observed the intestinal mucosal ultrastructure by transmission electron microscopy. Pretreatment with these two strains improved the ruptured, sparse, and stunted phenotypes of the intestinal epithelial cell microvilli (Fig. 3b).  were filtered for downstream analysis. A total of 5000 reads were randomly chosen from each sample to equilibrate the sequencing depth. Based on ≥97% sequence identity, 497 qualified operational taxonomic units (OTUs) were clustered. Four OTUs were discarded because of their low coverage (fewer than 5 samples). The species diversity calculated by the Simpson diversity indices and the community richness determined by the Chao1 indices were similar between the PC group and the LI09 (p = 0.161, p = 0.332, respectively) or LI10 group (p = 0.161, p = 0.180, respectively) (Fig. 4a). Actually, the Simpson indices and Chao1 indices were also similar between the PC group and the LI06 or LI07 group. However, the Simpson index of LI08 group was higher than that of PC group, although the Chao1 indices were similar between these two groups (see Supplementary  Fig. S3). An unweighted UniFrac principal coordinate analysis (PCoA) was used to analyse the overall structural  changes of microbial communities (Fig. 4b). PC samples were roughly separated from NC samples (p = 0.013) along PC2, which explained 8.7% of the total variations. When the rats were administered with either B. pseudocatenulatum LI09 or B. catenulatum LI10 before D-GalN treatment, the gut microbiome was distinctly separated from that of the PC samples (p < 0.001, p = 0.015, respectively) and did not differ from that of the NC samples (p = 0.050, p = 0.607, respectively) along PC2, demonstrating that oral administration of B. pseudocatenulatum LI09 or B. catenulatum LI10 ameliorates the D-GalN-induced alterations of the gut microbiome. D-GalN-induced acute hepatic injury clearly affected the intestinal microbial structure at the phylum level (Fig. 5a). Compared with the NC group, the PC group demonstrated a marked depletion in Firmicutes, Verrucomicrobia and TM7 and a significant enrichment in Bacteroidetes, Proteobacteria and Fusobacteria. Oral B. pseudocatenulatum LI09 administration ameliorated the D-GalN-induced Proteobacteria enrichment, and administration of B. catenulatum LI10 alleviated the D-GalN-induced TM7 depletion.

B. pseudocatenulatum
Furthermore, acute liver injury had a much wider impact on the gut microbiome at the family and genus levels (Fig. 5). Odoribacter, which belongs to family Porphyromonadaceae of phylum Bacteroidetes, was depleted in the PC group, and administration of either B. pseudocatenulatum LI09 or B. catenulatum LI10 attenuated the D-GalN-induced Odoribacter depletion.
Administration of B. pseudocatenulatum LI09 or B. catenulatum LI10 contributed considerably to preventing the D-GalN-induced alterations in the abundance of Clostridiales of phylum Firmicutes. Specifically, genus Clostridium_sensu_stricto, the main component of family Clostridiaceae_1, was enriched in the PC group compared with that in the NC group, and administration of either B. pseudocatenulatum LI09 or B. catenulatum LI10 ameliorated its enrichment. Conversely, genus Clostridium XI, the main member of Peptostreptococcaceae, was depleted in the PC group, and this depletion was alleviated by pretreatment with either of these two strains. Only B. pseudocatenulatum LI09 attenuated the D-GalN-induced depletion of genus Gracilibacter. In addition, pretreatment with B. pseudocatenulatum LI09 or B. catenulatum LI10 also ameliorated the depletion of genera Dorea, Anaerostipes and Coprococcus, which all belong to family Lachnospiraceae. Although there was no difference in the abundance of family Ruminococcaceae between the PC and NC groups, genus Saccharofermentans, within this family, was enriched in the PC group; this enrichment was alleviated when the rats were pretreated with B. catenulatum LI10. Genus Flavonifractor, another Ruminococcaceae member, was depleted in the PC group, and treatment with B. catenulatum LI10 attenuated its depletion.
B. pseudocatenulatum LI09 or B. catenulatum LI10 administration significantly prevented the D-GalN-induced enrichment of opportunistic pathogens belonging to phylum Proteobacteria. Parasutterella, which was the primary contributor to the enrichment of class Betaproteobacteria, was enriched in the PC group, and administration of either B. pseudocatenulatum LI09 or B. catenulatum LI10 ameliorated its enrichment. In addition, the D-GalN-induced enrichment of another Proteobacteria member, genus Helicobacter, was attenuated in the group treated with B. catenulatum LI10.
By comparison, administration of none of the B. longum LI06, B. longum LI07 or B. pseudocatenulatum LI08 could alleviate the depletion of Odoribacter, Clostridium XI, Dorea, Anaerostipes and Coprococcus and the enrichment of Clostridium_sensu_stricto and Parasutterella simultaneously (see Supplementary Table S2).
Correlations between involved inflammatory cytokines and gut bacterial genera were found in the groups treated with these two strains. To verify that alleviating the alterations in the gut microbial composition plays an important role in reducing the increase in inflammatory cytokines, a systems biology approach (correlation-network analysis) was used to identify key linkages between the involved inflammatory cytokines and gut bacterial genera in acute liver injury (Fig. 6). In the B. pseudocatenulatum LI09-treated group, the chemokines MIP-1α and MCP-1 were negatively correlated with Odoribacter, while in the B. catenulatum LI10-treated group, the chemokine MIP-1α, along with M-CSF, was negatively correlated with Flavonifractor. Moreover, in both groups, a positive association was found between MCP-1 and M-CSF.

Discussion
Acute liver failure, a severe form of liver damage caused by various factors, is an infrequent but life-threatening disease, with death occurring in up to 50% of the cases 18,19 . The standard therapy for liver failure, liver transplantation, has improved the survival rate by approximately 40%; however, this option is strongly limited because of the shortage of donor livers 20,21 . Thus, an artificial liver system (ALS), with the primary aim of detoxifying blood, The principal coordinate analysis (PCoA) plot shows the beta diversity of the gut microbiome with unweighted UniFrac distance derived from 16S-based sequencing data. Composition profiles of the caecal flora in the LI09 or LI10 group was compared with those in the PC and negative acute liver injury control (NC) groups.
is widely used in clinics until the patient's own liver regenerates or a donor liver becomes available; nevertheless, the cost and safety of ALS remain a concern 22 . In this study, oral administration of B. pseudocatenulatum LI09 or B. catenulatum LI10 had extensive beneficial effects on induced liver damage in experimental rats, including reducing ALT, AST, glycylproline dipeptidyl aminopeptidase and total bile acid levels, as well as hepatic inflammation and necrosis, and attenuating systemic inflammatory responses, which could be of great importance in the prevention or adjuvant treatment of clinical acute liver failure in the future.
By contrast, administration of B. longum LI06 or B. pseudocatenulatum LI08 did not demonstrate significant protective effects on induced liver damage. And administration of B. longum LI07 showed beneficial effects only in the aspect of liver histology, which is a subjective indicator. Therefore, we didn't consider B. longum LI07 as a candidate probiotics either.
Two strains from the same species, B. pseudocatenulatum LI08 and B. pseudocatenulatum LI09, demonstrated different effects. A similar situation was encountered by LX Lv et al. 23 , who reported that Lactobacillus salivarius LI01 were beneficial in the prevention of acute liver injury and Lactobacillus salivarius LI02 did not. Lactobacillus rhamnosus is another example. There are more than twenty kinds of strains of this species stored at American type culture collection (ATCC), but only the strain Lactobacillus rhamnosus GG (ATCC 53103) has been studied extensively on its various health benefits and become the world's most studied probiotic bacterium. These conflicting findings may be explained by different sources of the strains which carry different genes and therefore have differences in their phenotypes.
There is increasing evidence suggesting that systemic inflammatory response syndrome (SIRS) occurs in acute liver failure through the release of damage-associated molecular patterns (DAMPs) and proinflammatory cytokines as a result of massive hepatic cell necrosis, which in turn plays a key role in the clinical course and outcome in acute liver failure patients 24 . In our study, strong inflammatory responses, as evidenced by significant increase in the levels of both proinflammatory cytokines (TNF-α, IL-1α, IL-1β, IL-2, and IL-6) and anti-inflammatory cytokine (IL-10), as well as chemokines (MCP-1, MIP-1α and MIP-3α) and colony-stimulating factor (M-CSF), were observed during acute liver injury progression. Lower levels of MCP-1, MIP-1α and M-CSF were seen at 24 h in both the LI09 and LI10 groups than in the PC group. MCP-1 and MIP-1α, both important members of the CC-chemokine family, recruit and activate monocytes/macrophages to the injured tissue area and regulate proinflammatory cytokines and adhesion molecules. They have been reported to play an important role Green indicates those bacterial taxa whose abundance were higher in the PC group, and red indicates those bacterial taxa whose abundance were higher in the other group. In addition, yellow indicates those bacterial taxa whose abundance show non-significant differences between the two groups. (b) The relative abundance at the bacterial family level in the LI09 or LI10 group was compared with that in the PC group. (c) The relative abundance at the bacterial genus level in the LI09 or LI10 group was compared with that in the PC group. Values are expressed as the median with interquartile range. *p < 0.05 compared with the PC group.
in the early inflammatory response. M-CSF affects cells of the mononuclear phagocytic lineage in several ways, including regulating their survival, proliferation, and differentiation and as a major chemoattractant for these cells 25 . In the absence of M-CSF, monocytes could not be recruited to the sites of injured tissue due to the lack of MCP-1 synthesis 26 . Here, our data showed that administration of either B. pseudocatenulatum LI09 or B. catenulatum LI10 reduced the increased levels of MCP-1, MIP-1α and M-CSF (positively correlated with MCP-1), which may have led to the protection against liver injury, as suggested by the alleviation of the increase in serum ALT & AST and the amelioration of hepatic inflammation and necrosis.
The innate immune response is triggered by the recognition of translocated bacteria/bacterial products, followed by the release of chemokines and cytokines, leading eventually to bacterial killing 7 . In our study, bacterial translocation in the groups treated with these two strains was evaluated. MLNs are the first line of defence once the gut barrier has been breached. We found that the population of translocated bacteria to MLNs did decrease when rats were treated with either B. pseudocatenulatum LI09 or B. catenulatum LI10, which is consistent with the lower inflammatory response in these two groups.
Generally, only traces of gut-derived bacteria and their components can traffic in the portal vein to the liver due to the gut barrier, which is weakened upon the occurrence of liver disease. A disrupted intestinal barrier facilitates this translocation.
The integrity of the intestinal mucosa is one of the two important pillars that support the gut barrier function. In a liver disease situation, intestinal venous congestion will happen, followed by enterocyte necrosis or apoptosis, which manifest as damaged intestinal mucosal integrity. In this study, the incidence of subepithelial Gruenhagen's space and epithelial lifting determined by H&E staining and the ruptured, sparse, and stunted microvilli of the intestinal epithelial cells determined by transmission electron microscopy indeed dropped in the LI09 and LI10 groups.
The intestinal colonization resistance, or the control of the growth of opportunistic microorganisms, is the other pillar. The normal flora acts as a barrier against colonization of potentially pathogenic microorganisms and against overgrowth of already present opportunistic microorganisms (mainly aerobic gram-negative bacteria) 27,28 . The ecological balance of the gut flora is disrupted in hepatic injury, since bile is secreted abnormally and intestinal peristalsis slows down. Probiotic strains or their metabolites can modify the gut microbiological ecology. In our study, oral administration of B. pseudocatenulatum LI09 or B. catenulatum LI10 alleviated the D-GalN-induced alterations in the gut microbiome, as evidenced by the observation that the gut microbiome in samples pretreated with either B. pseudocatenulatum LI09 or B. catenulatum LI10 was significantly separated from that in PC samples and did not differ from that in NC samples in the PCoA analysis.
Bacterial phylotypes whose abundance was enriched in rats with acute liver damage were mostly associated with phylum Proteobacteria, including Betaproteobacteria and Epsilonproteobacteria. Genus Parasutterella, belonging to class Betaproteobacteria, was significantly enriched in PC samples. A high prevalence of these potentially pathogenic phylotypes has also been demonstrated in patients with Crohn's disease 29 . In this study, administration of either B. pseudocatenulatum LI09 or B. catenulatum LI10 ameliorated the enrichment of genus Parasutterella, indicating the beneficial effects of these two strains. In addition, genus Helicobacter, belonging to Figure 6. Correlation-network analysis of plasma cytokines and gut bacterial genera of D-GalN-sensitized rats who received B. pseudocatenulatum LI09 or B. catenulatum LI10. A Spearman correlation analysis was performed, and only correlations with p < 0.05 are displayed. Blue nodes represent bacterial genera, and yellow nodes represent cytokines. The ellipses indicate that the depletion of the bacterial genera was alleviated compared to that in the PC group, and the round rectangles indicate that the enrichment of the bacterial genera was alleviated or the increase of the cytokines was alleviated compared to that in the PC group. A red line connecting nodes represents positive correlation, and a green line represents negative correlation. Values of the corresponding correlation coefficients are indicated by the thickness of the line; the thicker the line, the greater the coefficient.
class Epsilonproteobacteria, was also found to be significantly enriched in PC samples. The most widely known species of this genus is Helicobacter pylori, and some strains of this species, as well as several strains of non-H. pylori Helicobacter, are pathogenic to humans, as they are strongly associated with peptic ulcers, chronic gastritis and gastric cancer. The enrichment of genus Helicobacter was alleviated in the LI10 group.
Bacterial phylotypes with reduced abundance in rats with acute liver damage were strongly associated with class Clostridia (phylum Firmicutes), including Lachnospiraceae, Peptostreptococcaceae and Gracilibacteraceae, but not Clostridiaceae_1. Lachnospiraceae is known to participate in fibre fermentation in the human intestine 30 . The by-products of carbohydrate fermentation, such as butyrate, nourish cells lining the colon. In addition, these short-chain fatty acids (SCFA) seem to have anti-inflammatory effects by inducing regulatory T cells (Tregs) and, as a result, calibrate our immune system. The absence of these fermentation-related bacteria causes a decline in SCFA production and often correlates with diseases, such as asthma and inflammatory bowel disease 31 . Genera Dorea, Anaerostipes and Coprococcus, within family Lachnospiraceae, were all found to be depleted in the PC group. The latter two are both butyrate-producing bacteria. A decline of Coprococcus has been demonstrated in patients with intestinal 32, 33 , neuropsychological 34,35 , infectious 36 , and atopic diseases 37 , as well as liver diseases [38][39][40] . A depletion of Dorea has been found in patients with intestinal 41 and infectious diseases 36 , along with liver disease 42 , and a suppression of Anaerostipes has been shown in patients with paratuberculosis infection 43 . In addition, genus Clostridium XI, a SCFA-producing microorganism, within the family Peptostreptococcaceae, was also found to be significantly depleted in the PC group. The depletion of Dorea, Anaerostipes, Coprococcus and Clostridium XI in D-GalN-sensitized rats was attenuated by both the B. pseudocatenulatum LI09 and B. catenulatum LI10 strains, which further demonstrated the regulatory capacity of these two strains. Although there was no difference in the abundance of family Ruminococcaceae between the PC and NC groups, genus Flavonifractor, capable of cleaving the flavonoid C-ring 44 , within this family, was depleted in the PC group. A decline of a species belonging to Flavonifractor has also been demonstrated in obese individuals 45 . Treatment with B. catenulatum LI10 attenuated this depletion. Conversely, genus Clostridium_sensu_stricto (family Clostridiaceae_1), another member of class Clostridia, was found to be enriched in the PC group, which is consistent with some infant illnesses such as food allergies [46][47][48] and necrotizing enterocolitis 49 . Its enrichment was ameliorated by either B. pseudocatenulatum LI09 or B. catenulatum LI10.
In addition, genus Odoribacter, another butyrate producer 50 , belonging to family Porphyromonadaceae (phylum Bacteroidetes), was also depleted in PC samples. The results of this study suggested that the abundance change of this genus in rats with acute liver injury is mostly accordance with metabolic syndrome 50,51 . Its depletion was alleviated in both the LI09 and LI10 groups.
In short, these two strains both played a role in attenuating the depletion of health-promoting bacteria and ameliorating the enrichment of pathogenic bacteria, therefore partly maintaining the intestinal colonization resistance, which contributes to protecting the gut barrier function.
The regulation of the levels of MIP-1α and MCP-1 in the LI09 group and the levels of MIP-1α and M-CSF in the LI10 group might be associated with increased Odoribacter and Flavonifractor respectively, as evidenced by the significantly robust correlations, which are likely due to these two strains. As discussed above, the butyrate-producing genus Odoribacter and flavonoid-cleaving genus Flavonifractor, were both depleted in metabolic syndrome, which is closely associated with chronic low-grade inflammation. Seeing that MCP-1, MIP-1α and M-CSF are important chemoattractants for mononuclear phagocytic cells in the inflammatory response, it is indeed possible that B. pseudocatenulatum LI09 and B. catenulatum LI10 ultimately down-regulate the cytokines MCP-1, MIP-1α and M-CSF by modulating Odoribacter and Flavonifractor respectively. However, the underlying mechanisms will be possibly investigated in the future in view of the limited isolates and undeveloped culture techniques of Odoribacter and Flavonifractor at this stage.
In conclusion, our study shows the beneficial effects of B. pseudocatenulatum LI09 and B. catenulatum LI10 against liver injury induced by D-GalN, which might be mediated through the protection of the gut barrier function, specifically by alleviating intestinal flora dysbiosis and preventing intestinal epithelial cell damage. The restored gut barrier function may contribute to reducing bacterial translocation and finally down-regulating the overactive immune response. We will next evaluate their roles in attenuating liver damage in other experimental models of liver injury to facilitate the development of probiotic products in the future.

Methods
Strains and culture conditions. Five strains were used in this study: B. longum LI06 (CGMCC 10385), B. longum LI07 (CGMCC 10386), B. pseudocatenulatum LI08 (CGMCC 10387), B. pseudocatenulatum LI09 (CGMCC 10388), and B. catenulatum LI10 (CGMCC 10389). All strains were isolated from healthy volunteers, whose fecal samples obtained were immediately placed in an anaerobic environment before homogenized and plated on trypticase-phytone-yeast (QingDao RiShui, Ltd., Qingdao, China) agar. Five isolates were screened out through identifying with 16 s rDNA sequencing, stored at −80 °C and deposited in the China General Microbiological Culture Collection Center (CGMCC). The bacterial strains were revived following a standard approach (https://www.atcc.org/How_to_Revive_Cultures.aspx#bacteria2) and anaerobically cultured for 36 h at 37 °C. The cells were harvested by centrifugation at 6,000 g for 10 min. Subsequently, these cells were washed twice with normal saline and resuspended to a concentration of 3 × 10 9 colony-forming units/ml in normal saline for further use.

Animals and experimental design. Fifty-nine male Sprague-Dawley rats (from the Experimental Animal
Center of Zhejiang Province, China), with an initial weight of 250-350 g, were randomized into seven groups with different treatments as follows: PC group (8 rats) and NC group (6 rats) both received 1 ml of physiologic saline by daily gavage; while LI06 group (9 rats), LI07 group (9 rats), LI08 group (9 rats), LI09 group (9 rats) and LI10 group (9 rats) received 1 ml (3 × 10 9 colony-forming units/ml) of B. longum LI06, B. longum LI07, B. pseudocatenulatum LI08, B. pseudocatenulatum LI09 and B. catenulatum LI10 strains by daily gavage respectively. All animals were fed normal food (standard rat chow), kept at room temperature (22 °C) with a controlled 12-h light/dark cycle, and orally administered saline or bacteria through an oro-gastric tube once daily for 7 days. Acute liver injury was induced on the 8th day by an intraperitoneal injection of D-galactosamine (G0500, Sigma, Saint Louis, MO, USA) at a dose of 1.1 g/kg body weight in all groups except the NC group. All experimental procedures were performed in accordance with the 2011 National Institutes of Health Guide for the care and use of laboratory animals. The study protocol was approved by the Animal Care committee of Zhejiang University, China.
Sample collections. Twenty-four hours after the induction of liver damage, the animals were anaesthetized and subjected to laparotomy through a large midline incision under aseptic conditions. Blood samples were collected from the inferior vena cava for liver function tests and to measure plasma cytokine levels. MLNs of the ileocaecal area were picked for bacterial translocation assay and caecal contents were obtained for gut bacterial microflora analysis. Tissue samples from the left lobe of the liver and ileum biopsied from a site approximately 2 cm away from the ileocaecal valve were harvested for histologic evaluation. All of the fifty-nine rats were analyzed for each following assay.
Liver function tests. Blood samples were centrifuged at 3,000 g for 10 min to separate the serum for analysis.
Concentrations of ALT, AST, glutamyltransferase, glycylproline dipeptidyl aminopeptidase, total bile acid, total bilirubin and albumin in the serum were quantified using a Hitachi 7600 automatic analyser according to the manufacturer's instructions (Hitachi, Tokyo, Japan) 52 .
Plasma cytokine analysis. Blood samples were centrifuged at 3,000 g for 10 min to separate the plasma and stored at −80 °C until analysis. The plasma (20 μl) was analysed using magnetic bead suspension arrays with the Bio-Plex Pro Rat Cytokine 24-Plex Panel (Bio-Rad) according to the manufacturer's instructions. The 24-plex contains antibodies specific for erythropoietin, granulocyte colony-stimulating factor, granulocyte-macrophage colony-stimulating factor, growth-regulated oncogene-keratinocyte chemoattractant, IFN-gamma (γ), interleukin (IL)-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12(p70), IL-13, IL-17A, IL-18, monocyte chemoattractant protein 1 (MCP-1), macrophage colony-stimulating factor (M-CSF), macrophage inflammatory protein 1 alpha (MIP-1α), MIP-3α, Regulated on Activation, Normal T Cell Expressed and Secreted, tumour necrosis factor alpha (TNF-α), and vascular endothelial growth factor. The samples were examined using a Bio-Plex 200 analyser, and the results were calculated using the Bio-Plex manager 6.0 software 53 . Values for granulocyte colony-stimulating factor, granulocyte-macrophage colony-stimulating factor and IFN-γ from most samples were below the detection limit, and these cytokines were excluded from the data analysis. Additionally, a few values for IL-1α, IL-2, IL-4, IL-6, IL-13, IL-17A, IL-18, erythropoietin, Regulated on Activation, Normal T Cell Expressed and Secreted and TNF-α were below the detectable limit. Since the standard curve lower limits of all these cytokines were above 0.1 pg/ml, and the statistical analysis was performed by rank sum test, those undetectable values were assumed to be 0.1 pg/ml for statistical purposes 54 .
Bacterial translocation. A one-fold amount of sample from the MLNs was weighed and milled in a sterile glass homogenizer containing a nine-fold amount of physiological saline. Then, 50 μl of this homogenate (1/10 dilution) was plated on brain-heart infusion agar (BHI, Oxoid, Thermo Fisher Biochemicals Ltd., Beijing, China) in duplicate within 30 min of the sample collection and incubated for 72 h at 37 °C under either aerobic or anaerobic condition. Bacterial translocation was thought to occur if bacteria grew in the culture medium with the MLN homogenate. The number of colony-forming units from each plate was counted, and the number of translocated bacteria was expressed as CFU per gram of tissue.
Histological evaluation. Tissue samples from the left lobe of the liver and the terminal ileum were immediately fixed in 10% neutral formalin when taken from the anesthetized rats and then embedded in paraffin, cut into 2-μm sections, stained with haematoxylin and eosin (H&E), and analysed by a pathologist who was blind to the groups. At least three slides were studied from each specimen.
The tissue damage of the liver was semiquantitatively assess by a histological score consisting of two categories from the Histological Activity Index 55 , intralobular degeneration and focal necrosis (score, 0-1 or 3-4) and portal inflammation (score, 0-1 or 3-4), which reflect the acute injury of livers. Intestinal mucosal lesions were classified as described by Chiu et al. 17 . Score 0 is defined as normal mucosa and score 1 is the development of subepithelial Gruenhagen's space at the tip of the villus. This space is more extended in score 2. In score 3, there is a massive epithelial lifting down the side of the villus. And in score 4, the villus is denuded of epithelium. It is characterized by a loss of the villus itself in score 5.
Electron microscopy. Ileal mucosal specimens fixed in 2.5% glutaraldehyde were post-fixed, dehydrated and embedded in epoxy resin. Ultrathin sections were made and post-stained in uranyl acetate and lead citrate 56 . The ileal mucosal ultrastructure was analysed on a Philips Tecnai 10 electron microscope (Philips, Eindhoven, the Netherlands). The length, diameter, linear density of microvilli was observed to learn the situation of microvilli damage 57 . 16 S rRNA sequencing. Caecal content samples (200 mg/aliquot) were immediately frozen upon collection from euthanized rats and stored at −80 °C before analysis. DNA extraction was performed using a QIAamp ® Fast DNA Stool Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions.
The bacterial 16 S rRNA V3-V4 regions were amplified by PCR using Phusion ® High-Fidelity DNA Polymerase (New England Biolabs, Ipswich, MA, USA). The universal primer pairs were 319 F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806 R (5′-GGACTACHVGGGTWTCTAAT-3′). The reactions were hot started at 98 °C for 30 s, followed by 40 cycles of 98 °C for 10 s, 52 °C for 30 s, and 72 °C for 45 s, with a final extension step at 72 °C for 10 min. The amplified products from different samples were purified using the QIAquick Gel Extraction kit (QIAGEN, Hilden, Germany) and then mixed accordingly to achieve equal amounts in the final mixture. An amplicon library was constructed with a DNA sample preparation kit (Illumina, San Diego, CA).
Sequencing was conducted on an Illumina MiSeq platform (Illumina, San Diego, CA) according to the manu-facturer′s instructions. Raw sequence reads were assigned to each sample according to their unique barcode pairs. Overlapping paired-end reads were merged to form tags using FLASH (version 1.2.8) 58 . The following quality control criteria were used: (1) an exact match to at least one end of barcodes and primers; (2) no undetermined bases in the tags; and (3) the number of mismatches in overlap region was no more than 3.
Operational taxonomic units (OTUs) were clustered with a 97% similarity cut-off using Usearch 59 . Alpha diversity was determined using the R program 60 "rich" and "diversity" packages. Principal coordinate analysis (PCoA) as a standard multivariate statistical technique was performed with the R program "ade4" package to explain differences among microbial communities. The longest sequence in each OTU was chosen as the representative sequence for identification using the RDP (Ribosomal Database Project) database v.11.3 61 . Taxonomic ranks were assigned to each tag using RDP classifier v.2.10.1 62 , using 0.8 as the confidence coefficient.
Statistical Analysis. A t-test (for those with normal distribution) or Mann-Whitney U test (for those with skewed distribution) was used to determine differences between two groups. The Spearman rank correlation coefficient was used for linear correlation analysis. The data were analysed using SPSS 20.0 and GraphPad Prism 7 and presented as the mean ± SD or median with interquartile range. A p-value < 0.05 derived from a two-tailed test was considered statistically significant. The network was presented after the correlation analysis using Cytoscape 63 .