Betaine ameliorates intestinal injury by targeting the LPS/TLR4/MyD88 pathway and microbial communities of intestinal tract in acute liver failure

Background: A growing body of evidence revealed that the gut microbiome has a marked impact in acute liver failure (ALF). Microbes and their products will translocate into enterocoelia and enter into circulation system from the damaged intestinal lumen. It will further aggravate liver injury by enhancing the spread of inflammation, tissue damage and sepsis. Betaine is a hepatoprotective drug which has anti-inflammatory and anti-oxidant effects. Here, we evaluated the impact of betaine on gut microbiota composition in ALF animal experiment. The potential protective effect of betaine by inhibiting Toll-like receptor 4 (TLR4) responses was explored as well. Results: Eighteen mice were randomly divided into normal, model, and betaine groups. The ALF-induced intestinal epithelial barrier disruption internal models were induced by D-galactosamineD-Galand lpopolysaccharide (LPS). Betaine was administered intragastrically 1 week before exposure of D-Gal/LPS. LPS were solely applied with a rat small intestinal cell line IEC-18 to establish ALF-induced intestinal epithelial barrier disruption external model. Mice in the model group developed severe intestinal epithelial tissue injury than normal group, increased significantly in the protein levels of TLR4, MyD88, TRAF6 and TNF-a and the mRNA levels of TLR4 and MyD88, and decreased significantly in the protein and mRNA levels of (ZO)-1 and occludin. Whereas, all above indicators were improved significantly by administration of betaine than that in model. The degree of liver tissue pathological damage, liver function and serum inflammatory cytokines in betaine group were significantly reduced than that in model group. The permeability of mice intestinal epithelial and IEC-18 cell in models was improved in g-Enterorhabdus compared group. down-regulated the relative abundance of g-Enterorhabdus in model. The relative of g-Bacteroides was the highest in normal group and the least in model group. The relative of g-Prevotella was almost identical in normal and betaine group, and it was decreased in model group.

betaine group than models. A total of 509 Operational Taxonomic Units (OTUs) were produced from mice fecal samples according to 16s rRNA gene sequence analysis.
There were 156 core microbiomes in fecal samples. There existed a total of 24 species contained 11 species at the genus level which had a significant difference between groups. The increased relative abundance of g-Enterorhabdus in the model group was detected compared with normal group. Betaine down-regulated the relative abundance of g-Enterorhabdus in model. The relative abundance of g-Bacteroides was the highest in normal group and the least in model group. The relative abundance of g-Prevotella was almost identical in normal and betaine group, and it was decreased in model group.
Conclusion: Betaine effectively improved the intestinal mucosal barrier in acute liver failure. The mechanism was probably related to inhibit the LPS/TLR4/MyD88 signaling pathways, improved the intestinal mucosal barrier and then maintained the gut microbiota composition.

Background
Acute liver failure (ALF), characterized by massive liver necrosis associated with severe impairment of hepatic function, is an intractable and high mortality disease in clinical practice [1]. As for the pathogenesis of liver failure, the majority of scholars support the theory of 'two-hit hypothesis'. The primary liver injure is caused directly or indirectly by various pathogen (such as hepatitis virus, ethanol, drugs and hepatotoxicants, etc). Intestinal endotoxemia, as the secondary liver injury, also plays an important role in the occurrence and development of ALF [2]. In addition, the liver failure patients are susceptible to merge endotoxemia. It's mainly due to the damaged intestinal mucosal barrier which led to a significant quantities 4 of endotoxin (lipopolysaccharide, LPS) produced by overgrowth of gram-negative bacteria [3].
The integrated intestinal wall barrier, mainly consist of intestinal mucosal epithelium, tight connection and intrinsic membrane under the epithelium [4]. The tight connection between epithelial cells, contains transmembrane proteins occludin, claudins, junctional adhesion molecules, as well as the cytoplasm protein (ZO)-1. They are the most imperative part of the intestinal wall barrier [5]. The intestinal mucosal permeability will be increased, when these junctions are altered [6]. There is a close relationship between the weakened intestinal mucosal immune function and damaged mucosal mechanical barrier bacterial. Both of them contribute to the endotoxin and excessive intestinal pathogens translocation [7].
Endotoxin is a LPS component which recognized as part of the outer membrane of gram-negative bacteria. Tool like receptors (TLRs), are pattern recognition receptors (pattern recognition receptor, PRR). They play an important role in intestinal mucosal immune system via mediating signal transduction. Particularly, TLR4-mediated recognition of LPS occurs in the formation of a receptor multimer composed of two copies of the TLR4-MD2-LPS complex. And then the complex recruits the TIR domain-containing adaptors TIRAP (Mal) and MyD88 (MyD88dependent pathway). Once MyD88-dependent pathway is activated, the downstream signaling molecules is further activated, which lead to release a lot of inflammatory mediators such as TNF-α, IL-1β and IL-18 [8][9][10]. The previous study has indicated that the transcription of TLR4 is enhanced in ALF mice [11].
Increased evidences have been shown a close correlation between the gut microbiome and liver disease [12]. The liver possesses immune surveillance against multifarious pathogens which from the gut, as well as intestinal mucosal immunity 5 and the intestinal microbiome also affect liver function [13]. The mutual effect between the gut and the liver is regared as "gut-liver axis". The integrity of the intestinal epithelial barrier has been demonstrated to be regulated by gut microbiota [14]. A compromised intestinal mucosal barrier contributes the bacteria translocation into portal vein and then reach to the liver. It leads more inflammation, liver cell apoptosis, and rapid progression to multiple organ failure [15]. The pathogenic Proteus is increasing in ALF rats and markedly depleting the Coriobacteriaceae, Bacteroidales and Allobaculum in ALF rats [16]. Accumulating evidences point to gut microbiota being involved in this process. Moreover, the Bacteroidetes, Ruminococcaceae, Porphyromonadaceae and Lachnospiraceae are decreased in fecal microbial communities in acute-on-chronic liver failure (ACLF) patients compared to the levels before disease onset [17]. Besides, the levels of Firmicutes are ascended in ACLF patients fecal. The regulation of the intestinal microecology by microbial ecological agents has been proposed as an emerging therapeutic strategy for liver failure as well [18].
Betaine, as an essential human nutrient, exists in many tissue and organs, which are especially abundant in liver and kidney. It participates in methyl cycle. It is often used as feed additives to replace the biological function of methionine, to reduce the cost of feedstuffs [19]. Many studies have shown that betaine has many pharmacological functions such as antioxidant and anti-inflammatory effects [20,21]. The retrospective study suggests that higher betaine intakes may be associated with a lower risk of primary liver cancer [22]. There is a favorable relationship between the blood betaine concentration and the severity of nonalcoholic fatty liver disease (NAFLD) in the community-based participants [23]. And our previous study have also indicated that betaine effectively prevents alcohol-6 induced and high-fat-diet-induced liver injury from inhibition of the expression of TLR4 [24,25]. However, the effects of betaine on acute liver failure and related mechanisms are still unknown. Especially, the influence of betaine on the gut microbial ecosystem need further to study.
At present, the application of high-throughput next generation sequencing techniques allowed us to characterize the overall structure of the complex gut microbial ecosystem at the OUT level. In this study, we employed the D-Gal/ LPS in vivo and LPS in vitro to establish ALF-induced intestinal epithelial barrier disruption internal and external models respectively. Betaine was employed as an intervention agent. The relationship between the expression of key molecules in intestinal TLR4mediated signaling pathway and intestinal tight connection protein were examined.
The gut microbial ecosystem was also studied to explore potential therapeutic effect of betaine in acute liver failure.

Betaine ameliorated the liver damage and improved liver function and serum inflammatory mediators in ALF mice
ALF mice model was successfully established. The liver pathological changes and serum biochemical indicators were evaluated. As HE staining shown in Fig. 1A, the structure of liver lobules and the arrangement of liver cells in normal group was clear and neat, the necrosis of liver cells and the infiltration of inflammatory cells were not observed. On the contrary, the liver lobular structure in ALF model group was deranged, and massive hepatocyte necrosis surrounded by inflammatory cells infiltration. Compared with ALF model group, the degree of hepatocyte necrosis and inflammation was significant reduced in betaine intervention groups. As shown in Fig. 1C-E. The serum ALT, AST and TBIL in models were significantly increased than that in normal group (P < 0.05). Compared with the model group, the ALT, AST and TBIL levels were greatly lowered in betaine groups (P < 0.05). The levels of inflammatory cytokines, including TNF-α, IL-1β and IL-18 in the ALF group were significantly increased compared with normal group. They were significantly decreased in betaine group than that in ALF model group (P <0.05) (Fig. 1F-H).

Betaine ameliorated the small intestine damages in ALF mice
As shown in

Betaine inhibited TLR4/MyD88 pathway in ALF mice
The effect of betaine on the TLR4/MyD88 pathway was assessed. The protein levels of TLR4, MyD88, TRAF6 and TNF-α and the mRNA levels of TLR4 and MyD88 were significantly higher in model group than that in normal group (P < 0.05). Compared with model group, the protein levels of TLR4, MyD88, TRAF6 and TNF-α and the mRNA levels of TLR4 and MyD88 were significantly decreased in betaine group ( Fig.   2A-C and Fig. 3C).

Betaine improved the expression of (ZO)-1 and occludin in ALF mice
As shown in Fig. 2C-D and Fig. 3D, the mRNA and protein expression of (ZO)-1 and occludin were both significantly decreased in that in model group than normal (P < 8 0.05). Compared with the model group, the expressions of (ZO)-1 and occludin were significantly elevated in betaine groups (P < 0.05).

Betaine suppressed the TLR4/MyD88 pathway in LPS stimulated IEC-18 cell
As shown in Fig. 2E-H, the protein levels of TLR4, MyD88, TRAF6 and TNF-a were significantly increased in model group, compared with normal group (P < 0.05).
Compared with models, the protein levels of TLR4, MyD88, TRAF6 and TNF-a were dramatically suppressed by betaine administration (P < 0.05). In addition, the protein expressions of TLR4, MyD88, TRAF6 and TNF-a in medium dose in betaine group were lower than low dose betaine group. They were the most low in high dose betaine group (P < 0.05). The mRNA levels of TLR4 and MyD88 had a statistical difference among high dose, medium dose and low dose betaine group as well ( Fig.3E, P < 0.05). It showed a dose-depend manner. Compared with the model group, the expression of (ZO)-1 and occludin protein and mRNA were significantly improved in betaine intervention groups (P < 0.05 ).

IEC-18 cell
Moreover, the proteins and mRNA expression of (ZO)-1 and occludin in medium dose betaine group were elevated than low dose betaine group, and were the highest in high dose betaine group (P < 0.05). The expression improvement showed a dosedepend manner.

Betaine elevated the TEER value in LPS stimulated IEC-18 cells
As shown in Fig. 3B, the TEER value in model group was greatly decreased, 9 compared with the normal group, (P < 0.05). After being intervened with betaine, the TEER value were significant elevated (P < 0.05). There was a statistical difference among high dose, medium dose and low dose betaine group as well (P < 0.05). It showed a dose-depend manner.

OTUs analysis
The OTU network analysis for fecal samples of mice provided the existence of some core OTUs and a common microbial composition among each group. A total of 509 OTUs were produced from 15 fecal samples (Additional file 1). As shown in Fig

Alpha diversity analysis
Alpha diversity contains the observed species index, the chao1 index, the shannon index and simpson index. The observed species index indicated the actual number of OTU observed and chao1 index is used to estimate the total number of OTUs contained in a sample. Both of them indicate species richness of the sample. The simpson index and shannon index are applied to estimate the species diversity. As shown in Fig. 4B-F, the fecal bacteria of normal group had the highest level of species richness and diversity. However, the species diversities of microbiota in betaine and model group did not have notable differences.

Beta diversity analysis
Different from Alpha diversity analysis, Beta diversity analysis is appropriate to compare the differences species diversity among a pair of samples. UniFrac compares species community differences using phylogenetic evolution information.
The results can be used as an index to measure Beta diversity, which has taken into account the evolutionary distance between species. UniFrac results are divided into weighted UniFrac and unweighted UniFrac. Weighted UniFrac considers the abundance of sequences, and unweighted UniFrac does not. The Beta diversity analyses include the Anosim analyses and Principal Coordinates Analysis (PCoA).
Anosim analysis is a nonparametric test used to detect whether the difference between two or more groups is significantly greater than the difference within the intragroup to judge whether the grouping is reasonable. As shown in Fig.5A  species at the genus level which had a significant difference between groups (P < 0.05). At the genus level, they were g-Anaeroplasma, g-Anaerovorax, g-Bacteroides, g-Desulfovibrio, g-Dorea, g-Enterorhabdus, g-Helicobacter, g-Odoribacter, g-Parabacteroides, g-Paraprevotella and g-Prevotella. Among them, the increased relative abundance of g-Enterorhabdus in the model group was detected compared with normal group (P < 0.05). Betaine down-regulated the relative abundance of g-Enterorhabdus ( P < 0.05). The relative abundance of g-Bacteroides was highest in normal group and least in model group (P < 0.05). The relative abundance of g-Prevotella is almost same in normal and betaine group, and was decreased in betaine group (P < 0.05).

Species classification and abundance analysis
A sequence with the highest abundance is selected from each OTU as a representative sequence of the OTU. Using the RDP method, the representative sequence is aligned with the 16S database to classify each OTU to corresponding Bacteroidetes) was significant higher in ALF model group than normal group, and was reduced by betaine pretreatment in ALF model group (P < 0.05). There existed many differences of gut microbiome in mice fecal. These results indicated altering the gut microbiome might be crucial therapeutical target in ALF mice.

Discussion
Since 1998, when Marshall proposed the gut-liver axis, many attentions have been paid to the role of gut in liver disease [26]. Systemic endotoxemia, characterized by 13 increased plasma LPS concentrations, increases the intestinal permeability such as altering tight junctions and thus lead to more endotoxins entering into the portal vein and activating the Kupffer cells in the liver. It results in producing proinflammatory cytokine and acute inflammatory response proteins which persistently aggravates hepatic insufficiency and/or failure [2]. Our previous study have indicated intestine injure plays a crucial role in the progress of ALF as well [27]. All these results support that if intestinal injury was reduced, and the liver lesion would be alleviated in ALF. Gut-liver axis pays an essential role in the pathogenesis in ALF.
Epithelial cells are held together by the apical junctional complex which including transmembrane proteins (such as claudins and occludin) and cytosolic scaffold proteins such as (ZO)-(1-3) and cingulin. The study has proved that inflammatory conditions contribute to significant disturbance of mucosal barrier function with a decreased expression and redistribution of claudin, (ZO)-1, and occludin which lead to increasing permeability [28]. LPS, which can trigger a powerful inflammatory response, is recognized as a causal or complicating factor of multiple serious diseases. One of the mechanisms which preventing LPS from intestine enter into the systemic circulation is through TLRs, which can activate the immune system [29].
However, the continuous stimulation of TLR signaling does not always provide positive effects for the host and it can enhance hepatic injury in NASH, alcoholic liver disease, and chronic viral hepatitis [30 -32]. There are 11 TLRs which have been identified in mammals, whereas LPS binds mainly TLR4. The MyD88 dependent pathway, as one of the downstream signaling events mediated by TLR4, lead to the recruitment of numerous molecules, which results in the activation of transcription factors nuclear factor-B (NF-B), tumor necrosis factor (TNF-) and other proinflammatory factors [8]. It has proved that TNF-can down-regulate the expression of the transmembrane tight junction strand protein occludin, paralleling the barrier disturbance detected electrophysiologically which lead to an increase in intestinal permeability [33].
The intestinal epithelial barrier is not a static physical barrier but rather strongly interacts with the gut microbiome and cells of the immune system. Under physiological conditions, there is a dynamic regulation of tight junction components. However, sustained inflammation or infections can result in dysregulation in the expression of adhesion molecules, leading to barrier breach and disturbance of gut microbes [34]. The studies demonstrated that TLRs are innate pattern recognition receptors involved in host defense, normal over commensal bacteria and the maintenance of tissue integrity [35,36]. The experiment has demonstrated that MyD88-negative mice with a defined microbial consortium which representing bacterial phyla normally present in human gut [37]. It has been found that B.
adolescentis exhibited anti-inflammatory properties in D-Gal treated rats.
Accumulating evidence points to gut microbiota being involved in this process. The study has shown that S. boulardii significantly decreased in the relative abundance of phylum Bacteroidetes, and increased the relative abundance of Firmicutes and Proteobacteria in ALF mice [38].
Betaine, an oxidative metabolite of choline, can protect rats from induction of the LPS hepatotoxicity [39]. The present study showed that betaine exerts positive effects on gut-liver axis, including inhibition of potent inflammatory response and maintain the gut integrity. Our data showed that the TLR4 level was low in the  [40]. The PCoA analysis showed that normal group and betaine group had more similar microbial composition. The relative abundance of Firmicutes in ALF model mice fecal was significantly higher than normal group. And the relative abundance of Bacteroidetes in ALF model mice fecal was apparently lower than normal group. But the bataine decreased the relative abundance of Firmicutes and increased the relative abundance of Bacteroidetes in ALF model. The previous study has been conformed the levels of Firmicutes were positive correlation with ACLF severity and Bacteroidetes were inversely correlated [8]. It exactly suggested that betaine played a role in potential modifying gut bacterial community. The taxon in gut microbitoa Coriobacteriaceae, Lachnospiraceae, Enterorhabdus and

Cell culture
The rat small intestinal cell line IEC-18 was grown in DMEM medium mixed with 10 % FBS in an incubator at 37 °C, 5 % CO 2 , and saturated humidity. LPS (1 μg/ml) were applied with cells in model group, low dose, medium dose and high dose betaine group. The cellular model was divided into normal group, model group, and betaine group. IEC-18 cells were seeded in 6-well plates. After 12 hours betaine (400 μg/ml, 600μg/ml and 800 μg/ml) was added respectively into low dose, medium dose and high dose betaine group, when model was conducted. After 24 h, the cells were detected by Western Blot and Quantitative real-time PCR.

Transepithelial electrical resistance (TEER) measurement
IEC-18 cell was made into 2.5×10 5 cells/ml single cell suspension. The lower cell chambers and the upper cell chambers were added with 1.5 ml DMEM complete medium and 1 ml cell suspension respectively. The upper and lower chambers in model and betaine group were replaced by the complete medium and complete medium contained betaine. After 2 h, betaine group and LPS group were added LPS at the same time. Normal group was added complete medium. Then all cells were incubated for 24 h. The calibrated Millipore Millicell ERS-2 cell resistance meter was used to detect resistance value. The measured resistance value was multiplied by the area of the filter to obtain an absolute value of TEER, expressed as Ω•cm 2 . And the TEER values were measured as follows: TEER = (measured resistance valueblank value) ´ single cell layer surface area (cm 2 ).

Animal groups
Eighteen male specific pathogen free (SPF) mice, weighing 20 ± 2g, were purchased from the Experimental Animal Center of Wuhan University. This study was approved by the Ethics Committee of Renmin Hospital of Wuhan University. All animals were allowed access to food and water freely throughout the acclimatization and experimental periods. And they were kept in temperature (22 ± 2 °C) with a 12 h light/dark cycle. After acclimation for 1 week, animals were randomly divided into three groups: normal, model, betaine group. The ALF model was induced by D-Gal (400 mg/kg) and LPS (100 μg/ml) by intraperitoneal injection. Betaine groups were administrated intragastrically with betaine 800 mg/kg per day one week before ALF model was conducted. The other mice were administrated intragastrically with same amount of saline. All mice were sacrificed in 24 h after the mice of model and betaine groups were given D-Gal/LPS.

Assessment of liver function and serum inflammatory mediators
Blood samples were collected. The serum alanine aminotransferase (ALT), aspartate aminotransferase (AST) and total bilirubin (TBil) levels were assayed using routine biochemical methods by a Hitachi Automatic Analyzer (Hitachi, Inc, Japan). The serum inflammatory cytokine levels of tumour necrosis factor α (TNF-α), interleukin-1β (IL-1β) and IL-18 were detected by enzyme-linked immunosorbent assay (ELISA) kits (eBioscience, CA, USA)

Histological examinations
Liver and small intestine specimens were fixed in 10 % formaldehyde for 24 h, embedded in paraffin, sliced into sections of 5 μm thickness and stained with hematoxylin-eosin (HE). Histological assessment was evaluated under BX 51 light microscope (Olympus, Japan).

Western blotting
The IEC-18 cells and small intestine specimens extracts were subjected to 10 % sodium dodecyl sulfate-polyacrylamidem gelelectrophoresis (SDS-PAGE) and  (Table 1) were designed and 20 synthesized by Tsingke (Wuhan, China). β-actin was set as the housekeeping gene.

Intestinal permeability
The everted sac method was used as previously described to evaluate the small intestinal mucosal barrier function [27]. First, we everted the intestine segment in ice-cold Krebs buffer (pH 7.4), gently distended by injecting 1.

DNA extraction and 16s rRNA gene sequencing
Fecal samples of mouse were collected 1 hour before all animals sacrificing. And they were frozen at -80 °C. DNA was extracted using the QIAGEN extraction kit (Universal Biotech Company, Shanghai, China) and total DNA quality was assessed using the Thermo Qubit.

Sequencing library construction
The V3-4 region of the 16s rRNA gene was amplified using custom barcoded primers and sequenced as described previously using an Illumina MiSeq sequencer [41,42].
Briefly, the V3-4 domain of the 16S rRNA gene was amplified using primers F (5′-NNNNNNNNGTGTGCCAGCMGCCGCGGTAA-3′) and R (5′-GGACTACHVGGGTWTCTAAT-3′). Using diluted genomic DNA as a template, PCR was performed using Taq DNA Polymerase (Vazyme Biotech Company, Nanjing, China) to ensure the accuracy and efficiency of amplification. The PCR product library was examined by Fragment Analyzer. After the library quality inspection was eligible, the corresponding ratios were mixed according to the volume required in each sample. The mixed library was subjected to gelatinization purification (cutting range: 500-750 bp) using a QIAquick gel recovery kit (Universal Biotech Company, Shanghai, China). After recycling, the library was examined and quantified using a Fragment Analyzer and an Applied Biosystems QuantStudio 6 real-time PCR instrument. Sequencing was performed using Illumina MiSeq PE300.
After performing quality control of the original data, the Usearch software was used to de-chimerize and cluster the data. When Usearch clusters, Reads were first sorted according to the abundance from large to small, and the standard clustering of 97 % similarity is obtained. Each Operational Taxonomic Units (OTU) was considered to represent a species. Next, the Reads of each sample were randomly leveled and the corresponding OTU sequence was extracted. Then we used Qiime software to do the dilution curve of Alpha diversity index, selected reasonable sampling parameters according to the dilution curve, and analyzed the obtained OTU. A read was extracted as a representative sequence from OTU. Next, the representative sequence was aligned with the 16S database to classify each OTU by using the RDP method. After categorization, the OTU abundance table was obtained according to the number of sequences in each OTU. Finally, the subsequent analysis was performed according to the OTU abundance table [43].

22
The all results were presented as mean ± SE. Comparisons of the measurement data among groups were assessed with Student's t test and one-way analysis of variance (ANOVA). P value < 0.05 was considered statistically significant. Calculations were performed with the SPSS16.0 statistical software package. Alpha diversity was applied in analyzing complexity of species diversity for single sample through 4 indices including observed species, chao1, shannon and simpon. Beta diversity analysis was used to evaluate differences of samples in species complexity. LEfSe was used by linear discriminant analysis (LDA) to estimate the influence of each species abundance to identify communities or species that have significant