Integrated Analysis of Gut Microbiome and Liver Metabolome to Evaluate the Effects of Fecal Microbiota Transplantation on Lipopolysaccharide/D-galactosamine-Induced Acute Liver Injury in Mice

Acute liver failure (ALF) refers to the occurrence of massive hepatocyte necrosis in a short time, with multiple complications, including inflammatory response, hepatic encephalopathy, and multiple organ failure. Additionally, effective therapies for ALF are lacking. There exists a relationship between the human intestinal microbiota and liver, so intestinal microbiota modulation may be a strategy for therapy of hepatic diseases. In previous studies, fecal microbiota transplantation (FMT) from fit donors has been used to modulate intestinal microbiota widely. Here, we established a mouse model of lipopolysaccharide (LPS)/D-galactosamine (D-gal) induced ALF to explore the preventive and therapeutic effects of FMT, and its mechanism of action. We found that FMT decreased hepatic aminotransferase activity and serum total bilirubin levels, and decreased hepatic pro-inflammatory cytokines in LPS/D-gal challenged mice (p < 0.05). Moreover, FMT gavage ameliorated LPS/D-gal induced liver apoptosis and markedly reduced cleaved caspase-3 levels, and improved histopathological features of the liver. FMT gavage also restored LPS/D-gal-evoked gut microbiota dysbiosis by modifying the colonic microbial composition, improving the abundance of unclassified_o_Bacteroidales (p < 0.001), norank_f_Muribaculaceae (p < 0.001), and Prevotellaceae_UCG-001 (p < 0.001), while reducing that of Lactobacillus (p < 0.05) and unclassified_f_Lachnospiraceae (p < 0.05). Metabolomics analysis revealed that FMT significantly altered LPS/D-gal induced disordered liver metabolites. Pearson’s correlation revealed strong correlations between microbiota composition and liver metabolites. Our findings suggest that FMT ameliorate ALF by modulating gut microbiota and liver metabolism, and can used as a potential preventive and therapeutic strategy for ALF.


Introduction
Acute liver failure (ALF) leads to hepatocytes apoptotic, large amounts inflammatory response, and liver damage, thereby leading to multiple organ failure [1,2]. Numerous factors can stimulate acute liver injury, including hepatitis viruses, drugs, and toxins [3]. At present, the effective therapeutic strategies for ALF are still lacking [4], of which liver transplantation is still the most effective therapy for ALF. There is an urgent need for new therapies and drugs due to the lack of liver donors and extremely high costs [5]. Lipopolysaccharide (LPS), a characteristic component of Gram-negative bacteria, stimulates Kupffer cells, leading to the release of numerous pro-inflammatory cytokines including interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) [6,7]. Dgalactosamine (D-gal) is a hepatocyte disrupting agent, which used to enhance the liver

Experimental Protocol
Mice were intraperitoneally injected with LPS and D-gal, purchased from Sigma-Aldrich, to induce ALF. After 1 week of acclimation, the animals were separated into six groups randomly with 10 mice in each group: (1) control, (2) LPS/D-gal, (3) LPS/D-gal combined with FMT preventive treatment for 7 days (FMT-pre7d group), (4) LPS/D-gal + FMT preventive treatment for 14 days (FMT-pre14d group), (5) LPS/D-gal + FMT treatment for 7 days (FMT-treat7d group), and (6) LPS/D-gal + FMT treatment for 14 days (FMT-treat14d group). Prevention groups were gavaged with FMT (0.2 mL, 1 × 10 9 CFU/mL) once a day for 7 or 14 days before LPS/D-gal administration. Additionally, the treatment groups were gavaged with FMT (0.2 mL, 1 × 10 9 CFU/mL) once a day for 7 or 14 days after LPS/D-gal administration. The preventive and therapeutic effects of FMT were assessed over two time periods to determine its effect on acute liver injury over time.
Correspondingly, the control and LPS/D-gal model groups were gavaged 0.2 mL PBS once a day. After 14 days, the preventive and therapeutic groups were injected with D-gal and LPS intraperitoneally at the doses of 400 mg/kg and 50 µg/kg body weight, respectively. Meanwhile, the control group was injected the same dose of PBS.
On day 14, the mice were anesthetized and sacrificed 6 h after injection. Professors with Medical Laboratory Animal and animal Experiment Certificate of Zhejiang Province handled the experimental animals. The serum, colonic content, and liver tissues were collected immediately from all mice. A part of the liver was fixed using 4% paraformaldehyde solution, and the rest of liver tissue was saved at −80 • C.
Feces from control mice were collected daily as donors. Then, we placed the collected fresh fecal material in sterile tubes, and homogenized in sterile normal saline. Centrifuged at 6000× g for 10 min, pass the homogenate through 70 µm filter, and quantified (OD 600 value of 1.0, predicted to be 10 9 CFU/mL). Eventually, the fecal bacterial slurry was diluted to 5.0 × 10 9 CFU/mL using 30% sterile medical glycerin PBS, and stored at −80 • C [23].

Histology and Immunohistochemistry
The livers were fixed using 4% paraformaldehyde, and embedded using paraffin, and stained using hematoxylin and eosin (H and E) or myeloperoxidase (MPO) antibodies. Histopathological changes in each mouse were examined using an optical microscope (Nikon, Tyoko, Japan).

Microbial 16S rRNA Gene Sequencing Analysis
Colon contents were taken on day 14 for microbiota analysis. Extracted fecal DNA from the microbial community using DNA Kit (Omega Biotek, Norcross, GA, USA). The, we utilized 338F and 806R universal PCR amplification to change the regions of 16S rRNA genes V3-V4. The activation was 2 min at 95 • C, denaturation of 30 s at 95 • C, annealing of 30 s at 50 • C, and extension of 30 s at 72 • C, then extension of 10 min at 72 • C followed by 27 cycles. PCR amplicon products were isolated using a 2% agarose gel and a DNA gel extraction kit to purify. We used the QuantiFluor™-ST system (Promega, Madison, WI, USA) to quantify the PCR products. Sequencing was performed on an Illumina MiSeq platform (Illumina, San Diego, CA, USA) at Majorbio, Shanghai, China.
The following criteria were used for demultiplexing and filtering raw reads by QIIME (version 1.9.1): (1) At each site with a mean quality score < 20, 300 bp reads were trimmed, and reads less than 50 bp were deleted after a sliding window of 10 bp; (2) Reject sequences with indistinct reads or two nucleotide mismatches in primer mates. Sequences with at least 97% of matching nucleotides are clustered in the operational taxonomic units (OaTUs). Every 16S rRNA gene sequence was explained using the RDP classifier (version 2.11), compared to the SILVA 16S rRNA database (version 132), and set a comparison threshold of 70% [24,25].
LC-MS uses a Waters ACQUITY ultraperformance liquid chromatography system with a triple time-of-flight mass spectrometer (Waters, Milford, CT, USA; AB SCIEX TripleTOF 5600 System, Framingham, MA, USA). Separate chromatographic on Waters ACQUITY UPLC R BEH C18 (100 mm × 2.1 mm, 1.7 µm) at a flow rate of 0.4 mL/min preheated to 40 • C. The mobile phase consists of ultrapure water including 0.1% formic acid (phase A), and a mixture of acetonitrile: isopropanol with 0.1% formic acid according 1:1 volume (phase B), and injection volume is 20 µL. The gradient started with 5% phase B and 95% phase A, increases to 20% phase B in 3.0 min, increases to 95% in 6.0 min, and holds for 4.0 min, respectively. Thereafter, the gradient turned to the original chromatographic status within 0.1 min and hold 2.9 min. The scanning scope was m/z 50-1000 and the resolution was 30,000 resolutions for the full-scan mode. Using nitrogen as the carrier and a rate of 900 L/h. Quality parameters are set to ion original temperature is 120 • C, desolvation temperature is 500 • C, electrospray capillary voltage is 1.0 kV, collision voltage is 6 eV, and injection voltage is 40 V. We prepared quality control (QC) samples by mixing all sample extraction aliquots to evaluate system stability and were analyzed after each of the five samples throughout the analytical run. Principal component analysis suggested that the QC samples were clustered closely, verifying the good repeatability of UPLC-triple TOF-MS/MS method.
Metabolomics raw data from LC-MS were analyzed, and calibration was done using the Progenesis QI software (Waters, Milford, USA). Peak intensities were normalized to exclude peaks with the relative standard deviation > 30% in QC samples. Comparing its mass spectrometry information to public and commercial databases (http://www.hmdb.ca/; https://metlin.scripps.edu/; https://i-sanger.com) (accessed on 23 August 2021), annotating metabolites. The data from the positive and negative ion modes were then integrated into the SIMCA-P 14.0 (Umetrics, Umeå, Sweden) for further exploration. Principle component analysis (PCA) using an unsupervised method was applied to obtain an overview of the metabolic data, general clustering, trends, or outliers were visualized. Overall differences between treatments were distinguished using orthogonal partial least squares discriminant analysis (OPLS-DA) and Student's t-test. The changed metabolites among treatments were analyzed using the importance in the projection (VIP) values > 1.0 and p-value < 0.05. To evaluate the potentially affected metabolic pathways, metabolic pathways enrichment, and topological analysis were conducted on the Kyoto Encyclopedia of Genes and Genomes (KEGG) library [26,27].

Western Blot Analysis
Separate proteins electrophoretically on 10% SDS-polyacrylamide gels and the bands were transferred on polyvinylidene difluoride membranes. We blocked the membrane for 1 h using 5% defatted skim milk and incubated with Tris-buffered saline containing anticaspase-3 at 1:1000 dilution overnight, then incubated with secondary antibody. Finally, we used the ChemiDoc MP (Bio-Rad, Hercules, CA, USA) imaging system to obtain images [28].

Statistical Analysis
Statistical calculations using the GraphPad Prism (version 8.0, San Diego, CA, USA), and expressed as mean ± SEM. We used one-way ANOVA and Tukey's multiple comparison test to analyze. We used Spearman's rank correlation test to analyze the correlation of variants. The statistical significance was set at p < 0.05.

FMT Relieved LPS/D-Gal-Induced Liver Injury in Mice
To evaluate liver injury, several hepatic enzymes in liver homogenates were examined. Hepatic plasma ALT and AST, indicators of ALF [29], were remarkably improved in the LPS/D-gal group. Interestingly, FMT prevention and treatment decreased the elevated ALT and AST levels significantly ( Figure 1A,B). Furthermore, total bilirubin (TBIL) level, an indicator of liver function, were decreased significantly in the FMT treatment groups than that in the LPS/D-gal group ( Figure 1C). As the H and E staining showed, the LPS/D-gal challenge acute hepatic injury accompanied by a range of pathological manifestations, such as hemorrhagic necrosis, hepatic structural destruction, and numerous inflammatory cell infiltration. However, these morphological changes improved in the livers of the FMT prevention and treatment group mice ( Figure 1D). In conclusion, these results indicate that pretreatment and treatment with FMT ameliorated ALF in LPS/D-gal-caused mice. Moreover, we also found that the effect of the 14 d groups was better than that of the 7 d groups, and the 14 day treatment group had the best results.

FMT Ameliorated Inflammatory Response in LPS/D-Gal-Challenged Mice
The hepatocellular injury caused by LPS/D-gal is linked with liver and systemic inflammation [30]. Serum IL-6, IL-1β, and TNF-α levels were improved significantly in the LPS/D-gal group, and pretreatment and treatment with FMT effectively suppressed these pro-inflammatory cytokines levels, which further decreased with prolongation of FMT

FMT Ameliorated Inflammatory Response in LPS/D-Gal-Challenged Mice
The hepatocellular injury caused by LPS/D-gal is linked with liver and systemic inflammation [30]. Serum IL-6, IL-1β, and TNF-α levels were improved significantly in the LPS/D-gal group, and pretreatment and treatment with FMT effectively suppressed these pro-inflammatory cytokines levels, which further decreased with prolongation of FMT gavage (Figure 2A-C). Moreover, the anti-inflammatory cytokines IL-10 and IL-22 decreased significantly in the LPS/D-gal group and recovered in FMT prevention and treatment groups ( Figure 2D,E). Thus, FMT by gavage suppressed LPS/D-gal caused systemic and liver pro-inflammatory responses in mice. A notable increase in MPO, a marker of neutrophil infiltration, was found in the LPS/D-gal group, and treatment with FMT repressed neutrophil infiltration caused by LPS/D-gal ( Figure 2F).

FMT Inhibited Apoptosis of Hepatocytes in LPS/D-Gal-Induced Mice
LPS/D-gal caused by ALF is characterized by excessive apoptosis of hepatocytes [31]. Hence, hepatocyte apoptosis was detected using TUNEL staining and caspase activation in the liver. As shown in Figure 3A, numerous TUNEL-positive hepatocytes were ob-

FMT Inhibited Apoptosis of Hepatocytes in LPS/D-Gal-Induced Mice
LPS/D-gal caused by ALF is characterized by excessive apoptosis of hepatocytes [31]. Hence, hepatocyte apoptosis was detected using TUNEL staining and caspase activation in the liver. As shown in Figure 3A, numerous TUNEL-positive hepatocytes were observed in LPS/D-gal-treated liver tissues. However, the improved number of positive hepatocytes was lower in the FMT prevention and treatment groups. Similarly, the level of active caspase-3 was upregulated in LPS/D-gal-induced mice, whereas this induction was inhibited in the FMT prevention and treatment groups ( Figure 3B). Therefore, FMT by gavage suppressed LPS/D-gal-treated hepatocyte apoptosis and protected against ALF. Moreover, the FMT treatment 14 d group showed a lower positive TUNEL than the FMT treatment 7 d group, suggesting that prolonged FMT administration have better therapeutic effects against liver injury. FMT treatment 7d group, suggesting that prolonged FMT administration have better therapeutic effects against liver injury.

FMT Modulated Gut Microbiota Composition in LPS/D-Gal-Induced Mice
We further analyzed the influence of FMT therapy on the colonic microbial composition of LPS/D-gal challenged mice by 16S rDNA gene sequencing. We found decreased α-diversity indices, including Shannon, Sobs, Chao, and ACE indices of the gut microbiota of LPS/D-gal group than those of control group (p < 0.05). Aforementioned indices were increased in FMT prevention and treatment groups than in the LPS/D-gal group, although not obviously ( Figure 4A-D), suggesting that FMT increased bacterial richness (as determined by rising Sobs, Chao, ACE, and Shannon indices) and evenness (as determined by rising Shannon index) in LPS/D-gal-induced mice. Then, we analyzed the β-diversity with the Bray-Curtis principal coordinate analysis (PCoA). PCoA exposed remarkable differences in gut microbiota between the LPS/D-gal and control groups (R 2 = 0.4343, p = 0.002, Figure 4E). The LPS/D-gal and FMT treatment groups also showed different gut microbiota (R 2 = 0.3216, p = 0.001, Figure 4G), but there was no significant difference between

FMT Modulated Gut Microbiota Composition in LPS/D-Gal-Induced Mice
We further analyzed the influence of FMT therapy on the colonic microbial composition of LPS/D-gal challenged mice by 16S rDNA gene sequencing. We found decreased α-diversity indices, including Shannon, Sobs, Chao, and ACE indices of the gut microbiota of LPS/D-gal group than those of control group (p < 0.05). Aforementioned indices were increased in FMT prevention and treatment groups than in the LPS/D-gal group, although not obviously ( Figure 4A-D), suggesting that FMT increased bacterial richness (as determined by rising Sobs, Chao, ACE, and Shannon indices) and evenness (as determined by rising Shannon index) in LPS/D-gal-induced mice. Then, we analyzed the β-diversity with the Bray-Curtis principal coordinate analysis (PCoA). PCoA exposed remarkable differences in gut microbiota between the LPS/D-gal and control groups (R 2 = 0.4343, p = 0.002, Figure 4E). The LPS/D-gal and FMT treatment groups also showed different gut microbiota (R 2 = 0.3216, p = 0.001, Figure 4G), but there was no significant difference between LPS/D-gal and FMT prevention groups (R 2 = 0.2974, p = 0.001, Figure 4F). These results indicated that FMT modulates gut microbiota composition of the ALF mice are induced by LPS/D-gal.

FMT Altered Liver Metabolome in LPS/D-Gal-Induced Mice
Analysis of the liver metabolome using LC-MS-based non-targeted metabolomics. PCA showed a distinct variation in liver metabolome of the control and LPS/D-gal groups ( Figures 5A,B), suggesting that hepatic injury induced metabolic disturbances. The liver metabolome of FMT prevention and treatment group mice were slightly overlapped with Subsequently, we assessed the relative abundance of bacteria on phylum and genus levels ( Figure 4H,I). Bacteroides and Firmicutes were the most abundant, accounting for over 80% of all microorganisms in all groups. In Figure 4J, the relative abundance of Bacteroidota was reduced (p < 0.01), and those of Firmicutes (p < 0.01) and Desulfobacterota (p < 0.01) were increased in LPS/D-gal-challenged mice. FMT treatment inverted this change by improving the relative abundance of Bacteroidetes (p < 0.001) and reducing the abundance of Firmicutes (p < 0.01) and Desulfobacterota (p < 0.05). Compared with the FMT treatment groups, FMT prevention groups mainly showed an increased abundance of Bacteroidetes (p < 0.001) and a reduced abundance of Firmicutes (p < 0.01). Furthermore, within the phylum Bacteroidota ( Figure 4K), the abundance of norank_f_Muribaculaceae (p < 0.01), Alistipes (p < 0.05), and Prevotellaceae_UCG-001 (p < 0.01) was significantly decreased in LPS/D-gal than control group and accounted for the majority of reduced abundance of Bacteroidetes in LPS/D-gal group. Improved abundance of Firmicutes in LPS/D-ga-induced mice was mostly explained by the improved abundance of Lactobacillus (p < 0.05) and unclassified_f_Lachnospiraceae.

FMT Altered Liver Metabolome in LPS/D-Gal-Induced Mice
Analysis of the liver metabolome using LC-MS-based non-targeted metabolomics. PCA showed a distinct variation in liver metabolome of the control and LPS/D-gal groups ( Figure 5A,B), suggesting that hepatic injury induced metabolic disturbances. The liver metabolome of FMT prevention and treatment group mice were slightly overlapped with those of control group and were distinct from that of LPS/D-gal group, clearly indicating the role of FMT in preventing and treating hepatic injury. Furthermore, PLS-DA exhibited a better discriminative ability than PCA and was used to analyze the metabolic profiles. The OPLS-DA score plot showed significant distinctions in the liver metabolite content of the six groups ( Figure 5C,D). The hierarchical clustering heatmap in Figure 5E shows the changing trends of liver metabolites, and indicating significant differences in the control, LPS/D-gal, and FMT prevention and treatment groups.
Based on variable VIP > 1, fold change (FC) ≥ 1.2 or ≤0.8333, p < 0.05, and with KEGG annotations. As shown in Table 1, compared with the control group, there were 25 significantly differentially expressed metabolites in the LPS/D-gal group; of which, 11 were upregulated and 14 were downregulated-vs. ( Table 1). The levels of these metabolites tended to restore in the FMT prevention and treatment groups. We found that caffeine, L-2-Aminoadipic acid, cortolone, thymine, and 12(R)-HETE were significantly downregulated in the FMT-pre7d group, and cortolone, N-acetyl-alpha-D-glucosamine 1-phosphate, and taurocholate were significantly down-regulated in the FMT-pre14d group. Ne-acetyllysine, trichloroethanol glucuronide, thiamine, 2-(S-glutathionyl)acetyl glutathione, stachyose, and 5-L-Glutamyl-L-alanine were significantly up-regulated in both the FMT-pre7d and 14d groups. These metabolites were further studied. KEGG enrichment analysis revealed that FMT prevention mainly affected lysine degradation, metabolism of xenobiotics by cytochrome P450, the biosynthesis of plant secondary metabolites and other pathways ( Figure 5G,H). Compared with the LPS/D-gal group, both the FMT-treat7d and FMT-treat14d groups exhibited a decrease in metabolites of nucleotide sugar and amino sugar metabolites, such as UDP-N-acetyl-alpha-D-glucosamine, N-acetyl-alpha-d-glucosamine 1-phosphate, and CDP-glucose, and an increase in metabolites issued from xenobiotic metabolism by cytochrome P450 (trichloroethanol glucuronide, 2-(S-Glutathionyl)acetyl glutathione, and 1-methylguanosine), and starch and sucrose metabolites (UDP-glucose, CDP-glucose, and maltose). In addition, we found that cholesterol metabolites (number of metabolites = 2, p = 0.0005) were down-regulated in the FMT-treat 7 d group significantly and arachidonic acid metabolites (number of metabolites = 4, p = 0.0002) were notably down-regulated in the FMT-treat 14 d group (Figure 5I,J).

Gut Microbiota Is Associated with Liver Metabolites and Inflammatory Markers
The correlations between gut microbes, liver metabolites, and markers of inflammation were analyzed to further explore the potential mechanism of action of FMT. Figure 6A,B displayed the correlation between differential gut microbes on phylum, and genus and inflammatory markers. Firmicutes and Desulfobacterota showed a positive relationship with AST, ALT, TBIL, IL-6, IL-1β, and TNF-α, and a negative correlation with IL-10 and IL-22 levels. Similar trends were observed at the genus level, including Lactobacillus, Lachnospiraceae_UCG-006 unclassified_f_Lachnospiraceae, Desulfovibrio, etc. Opposite correlations were observed in Bacteroides, unclassified_o__Bacteroidales, no-rank_f__Muribaculaceae, and Prevotellaceae_UCG-001, etc. Figure 6C showed the correlation between metabolic biomarkers and inflammatory markers. Thymine, CDP-glucose, and N-acetyl-alpha-D-glucosamine 1-phosphate were negatively correlated with antiinflammatory cytokines and direct correlation with pro-inflammatory cytokines. In particular, unclassified_o__Bacteroidales, unclassified_f__Prevotellaceae, and norank_f__Muribaculaceae, among others positively correlated with metabolites, such as trichloroethanol glucuronide, stachyose, and maltose, that were notably downregulated in LPS/D-gal group, and negatively correlated with some metabolites that were significantly upregulated in LPS/D-gal group. Conversely, Lachnoclostridium, Faecalibaculum, Dubosiella, and Lachnospiraceae_UCG-006 showed opposite trends ( Figure 6C).

Discussion
ALF, fulminant hepatic failure, has a high mortality rate and resource cost, which is caused by the stimulation of drugs and toxins. Clinical manifestations of ALF include

Discussion
ALF, fulminant hepatic failure, has a high mortality rate and resource cost, which is caused by the stimulation of drugs and toxins. Clinical manifestations of ALF include rapid hepatic injury, derangements in coagulopathy, hepatic encephalopathy, and multi-organ failure, which pose severe threats to a patient [32,33]. D-Gal, a hexosamine derived from galactose, is a hepatotoxic poison, and intraperitoneal administration of which will induce multiple hepatocyte damage and inflammation, is analogous to the pathological status of clinical viral hepatitis [34,35]. Combined use of LPS and D-gal has been diffusely applied to set up animal models of ALF [36]. Considering the limitations of the existing ALF treatments, there is urgent to find effective treatments to ameliorate ALF-related treatment difficulties. There is increasing evidences that the dynamic changes of intestinal microbiota perform an important role in the occurrence and development of liver injury. In view of this, the regulation of intestinal microbiota are promising diagnostic, prognostic, and therapeutic tools. In this regard, FMT is the most powerful tool for resetting intestinal microbiome disturbances caused by liver diseases. [37,38]. Our study mainly explored the guarding role of FMT and its potential mechanism of action in an LPS/D-gal-induced ALF, and found that FMT can be used as a safe, effective, and low-cost treatment for ALF. However, there is no clear protocol for the standardized donor screening, preparation, pretreatment, administration, or long-term stability of FMT, and the exact mechanism involved in its action has not been elucidated. Furthermore, histopathological assessment and aminotransferase activities revealed that FMT intervention also alleviated LPS/D-gal-challenged hepatic failure. FMT decreased hepatic ALT and AST levels significantly to attenuate liver failure, which was confirmed by the decrease in TUNEL-positive hepatocytes, indicating reduced hepatocyte apoptosis. LPS, on administration, binds TLR4 to activate it, which in turn activates the NF-κB via intermediate proteins, which then induces the expression of inflammatory cytokines [39,40]. We found that FMT restrained LPS/D-gal-caused hepatic neutrophil infiltration, and the making of pro-inflammatory cytokine and chemokine. In ALF, hepatocytes and other cells produce excessive inflammatory mediators, resulting partial and systemic inflammation [41]. Sustained making of cytokines including TNFα, IL-1β, IL-6, IL-10, and IL-22, causes hepatocellular inflammation and apoptosis, and eventually leads to ALF [42,43]. According to the research, inhibition of inflammatory cytokines making may be a potential tactic for the therapy of ALF [43,44]. We found that FMT intervention downregulated the pro-inflammatory cytokines levels significantly and upregulated the anti-inflammatory cytokines, thereby reducing LPS/D-gal-induced inflammation, consistent with a previous study [45,46].
Studies have shown that the TNF-α is the main pro-apoptotic factor inducing extensive hepatocyte apoptosis in LPS/D-gal caused ALF [47,48]. TNF-α induces the neutrophil migration, that casts an important role in hepatocyte necrosis in the late stage of ALF [49]. TNF-α administration has been improved to accelerate hepatic failure [50], while TNF-α production inhibition [51] or TNF-α knockout [48] efficiently prevents acute liver injury. We found that FMT inhibited TNF-α production, a key regulator of apoptosis, to ameliorate the development of LPS/D-gal-induced ALF. Daubioul et al. [52] demonstrated in an initial research of seven patients with nonalcoholic steatohepatitis (NASH), prebiotic feeding for 8 weeks significantly reduced levels of liver inflammatory markers. Another study found the modulation of gut microbiota significantly decreased inflammatory markers (TNF-α and C reactive protein) [53]. To sum up, these data suggest that gut microbiota composition and metabolic activity may help suppress hepatic and systemic inflammation. The infiltration of neutrophils into liver tissue increases after the development of ALF, which can promote liver injury by producing inflammatory mediators [54]. Therefore, we measured the level of the neutrophil infiltration index MPO. We observed that FMT reduced MPO activity, indicating decreased neutrophil filtration. TNF-α induces apoptosis and necrosis of hepatocytes by activating caspase-3, which is a key executive molecule in apoptotic pathways [55,56]. FMT inhibited the expression of caspase-3, which was raised in LPS/D-gal challenged liver, indicating that FMT may adjust cellular apoptosis of ALF by suppressing caspase-3 expression. Finally, we found that prolonged intragastric administration of FMT moderated ALF caused by LPS/D-gal, and the effect was slightly better in the treatment groups than in the prevention groups.
To probe the mechanism of action of FMT, we evaluated the FMT-mediated changes in intestinal microbiota and liver metabolic pathways in mice. The intestinal microbiota is closely associated with health and numerous host functions, including growth, development, immunity, metabolism, and disease [57][58][59]. The liver secretes bile acids that end up in the small intestine. The portal vein that enters the liver is rich in nutrients, and is the main blood provide to liver. The gut-liver axis makes reference to the interplay between the gut (including the microbes in it) and the liver [11]. Accumulating studies revealed that the composition and metabolites of gut microbes can regulate and promote the development of the human immune system, such as through probiotics, prebiotics, antibiotics, FMT, diet regulation, and management of bacterial complexes [60,61].
FMT can replenish a healthy colonic microbial condition and restore colonization [62]. It is currently an established clinical therapy for Clostridium difficile infection [63] and was explored as a therapy for ameliorating inflammatory bowel disease [64], metabolic syndrome [65], and several liver diseases [66]. To elucidate the mechanisms by which FMT manifests its effects, we studied the alterations of colonic microbiota composition and diversity. We found that FMT significantly altered the relative abundance of some phylum and genera in intestinal microbiota, some of which are supported by previous studies. Using 16S rDNA sequencing, it found that LPS/D-gal decreased the abundance of Bacteroidota and improved that of Firmicutes and Desulfobacterota significantly. In contrast, gavage of FMT decreased the Firmicutes/Bacteroidetes ratio, which can assess the gut microbial community stability [67]. Therefore, FMT may alleviate liver injury by modulation of gut microbiota composition. Yu et al. [68] also identified an improved abundance in Firmicutes, and a reduction in Bacteroidetes after S. boulardii intervention. Firmicutes and Bacteroidetes phyla comprise the vast majority of the dominant human gut microbiota [69]. Members of Bacteroidota are widely considered as candidate probiotics for the treatment of immune dysfunction, intestinal colitis, metabolic disorders, and cancer [67,70]. The abundance of norank_f_Muribaculaceae increased in the FMT prevention groups, while the abundance of norank_f_Muribaculaceae, unclassified_o_Bacteroidales, and Prevotellaceae_UCG-001 increased in the FMT treatment groups. Muribaculaceae have favorable influences on gut dysregulation through immune modulation and regulation of intestinal homeostasis [71]. The LPS/D-gal-induced reduction in Muribaculaceae, which was negatively correlated with AST, ALT, and pro-inflammatory cytokines, was significantly improved after treatment with FMT. Although some strains of the Prevotellaceae appear to be inflammatory pathogens, they act a key role in immunomodulatory, especially those of Prevotella [12]. Thus, the reduced abundance of Prevotellaceae_UCG-001 in the FMT treatment groups suggested that FMT promoted the abundance of microbes regulating immunity. Alistipes, a comparative new genus of Bacteroidetes, is highly relevant to immune modulation, dysbiosis, and inflammation [72,73]. Studies have revealed that Alistipes abundance is decreased in liver fibrosis diseases, including NASH and nonalcoholic fatty liver disease (NAFLD) [73], as we found. In this research, Alistipes likely contributed to worsening liver injury and inversely correlated with IL-1β level.
The reduction in member of Bacteroidota induced by LPS/D-gal was negatively related to pro-inflammatory cytokine levels, which may affect the host immune system. After treatment with FMT, LPS/D-gal caused the depletion of Muribaculaceae, Bacteroidales, and Prevotella in mice to improve. Therefore, FMT may promote immune homeostasis by regulating the intestinal microbiota, which, in turn, modifies the making of pro-and anti-inflammatory cytokines. Firmicutes abundance shows positively correlated with proinflammatory cytokines and marker enzymes of liver injury. The increase in Firmicutes was mainly manifested through the increased abundance of Lactobacillus and unclassi-fied_f_Lachnospiraceae in the LPS/D-gal group. Lachnospiraceae and Lactobacillaceae can sugars and hydrolyze starch to generate short-chain fatty acids (SCFAs) [74,75]. Large amounts of SCFAs do not ever confirm favorable effects [76]. The gut microbiota of patients with NAFLD is enriched with Lachnospiraceae, as do patients with NASH or obvious hepatic fibrosis [77]. In our research, the gavage treatment of FMT obviously attenuated the abundance of Lachnospiraceae, which was positively correlated with pro-inflammatory cytokines. Lactobacillus provides the host with multiple benefits, such as suppression of pathogens and improvement of immune response [78]. Jiang et al. [12] found that pretreatment with probiotic Lactobacillus reuteri DSM 17,938 in rats could alleviate intestinal dysbiosis, reduced the inflammatory factors, and alleviated D-gal caused liver injury. However, our results showed a negative impact of Lactobacillus for it was positively correlated with pro-inflammatory cytokines, and a significantly negative correlation with IL-10 and IL-22. FMT pretreatment significantly attenuated the abundance of Lactobacillus, which is the same as that observed by Yan R et al. [79], who found Lactobacillus was heavily enriched in the D-gal-induced group and depleted in Lactobacillus casei strain Shirota (LcS) group.
The metabolomic PLS-DA analysis showed the mice treated with LPS/D-gal had unique metabolic characteristics compared to control group. A total of 25 differentially accumulated metabolites were identified. Through KEGG enrichment analysis, these metabolites were mainly enriched in primary bile acid biosynthesis, bile secretion, cholesterol metabolism, amino sugar and nucleotide sugar metabolism, secondary bile acid biosynthesis, and starch and sucrose metabolism. KEGG analysis revealed that prevention with FMT mainly improved lysine degradation, the metabolism of xenobiotics by cytochrome P450, and the biosynthesis of plant secondary metabolites. The levels of L-2aminoadipic acid and N-acetyl-alpha-D-glucosamine 1-phosphate, intermediate products of lysine metabolism, were elevated in the LPS/D-gal induced live. Several amino acids are known to have favorable influences on the treatment of liver steatosis, and lysine is known to regulate lipid metabolism in liver [80]. FMT significantly restored the levels of these two metabolites by altering lysine metabolism to alleviate LPS/D-gal-induced liver failure. Recent studies reported the active role of the hexosamine biosynthesis pathway (HBP) in host antiviral immunity of hepatitis B and C virus [81,82]. About 2-5% of the total glucose is transformed to uridine diphosphate N-acetylglucosamine (UDP-GlcNAc)-the last product of HBP [83]. D-gal can exhaust the uridine phosphate in hepatocytes, reducing the production of nucleic acid and protein synthesis [84]. We also found that N-acetylalpha-D-glucosamine 1-phosphate, UDP-N-acetyl-alpha-D-glucosamine, and CDP-glucose were markedly downregulated in the FMT treatment groups than LPS/D-gal group; therefore, FMT modulates HBP markedly. Pearson's correlation analysis revealed that the three metabolites mentioned above positively correlated with pro-inflammatory cytokines. The cytochrome P450 metabolism pathway has important function in detoxification, cell metabolism, and homeostasis of exogenous drugs [85]. Metabolomics results showed that the metabolites of xenobiotics metabolism by cytochrome P450, including 2-(S-Glutathionyl) acetyl glutathione and trichloroethanol glucuronide, which were negatively correlated with pro-inflammatory cytokines, were downregulated remarkably in LPS/D-gal group, and were restored in FMT prevention and treatment groups. Several studies have revealed that CYP450 enzyme are essential in the detoxification of Aristolochic acid I (AAI)-induced liver injury in mice (87). We also found that arachidonic acid metabolism was obviously down-regulated in FMT-treat14d group. As a plentiful lipid mediator in the human, arachidonic acid is essential in the inflammatory metabolic network, and its metabolites have pro-inflammatory features [86,87]. CYP450 hydrolyzes arachidonic acid, and the main metabolite of which is important in pro-inflammatory reactions [88,89]. Correlation analysis revealed arachidonic acid metabolites were positively related to IL-1β, IL-6, and TNF-α levels.

Conclusions
In summary, this study demonstrated the preventive and therapeutic effects of FMT against ALF caused by LPS/D-gal; indeed, we found that with prolongation of intragastric administration of FMT, liver failure caused by LPS/D-gal had markedly improved, and the results of treatment were significantly better than those of prevention. Furthermore, changes in the gut microbiome and hepatic metabolism have also been shown to contribute to understanding of the potential mechanism of FMT therapy for ALF. In conclusion, this study suggests that FMT could be a potential therapeutic approach to treat ALF and that the gut microbiome may be a potential therapeutic target for ALF.

Conflicts of Interest:
The authors declare no conflict of interest of investigators, ethics committee members, or disclosure of research results in this study.