Decoding the contributions of gut microbiota and cerebral metabolism in acute liver injury mice with and without cognitive dysfunction

Abstract Aims Patients with acute liver injury (ALI) can develop cognitive dysfunction (CD). The study investigated the role of gut microbiota and cerebral metabolism in ALI mice with and without CD. Methods Male C57BL/6 mice that received thioacetamide were classified into ALI mice with (susceptible) or without (unsusceptible) CD‐like phenotypes by hierarchical cluster analysis of behavior. The role of gut microbiota was investigated by 16S ribosomal RNA gene sequencing and feces microbiota transplantation (FMT). 1H‐[13C] NMR and electrophysiology were used to detect the changes in cerebral neurotransmitter metabolic and synaptic transition in neurons or astrocytes. Results Apromixlay 55% (11/20) of mice developed CD and FMT from the susceptible group transmitted CD to gut microbiota‐depleted mice. Alloprevotella was enriched in the susceptible group. GABA production was decreased in the frontal cortex, while hippocampal glutamine was increased in the susceptible group. Altered Escherichia. Shigella and Alloprevotella were correlated with behaviors and cerebral metabolic kinetics and identified as good predictors of ALI‐induced CD. The frequencies of both miniature inhibitory and excitatory postsynaptic currents in hippocampal CA1 and prefrontal cortex were decreased in the susceptible group. Conclusion Altered transmitter metabolism and synaptic transmission in the hippocampus and prefrontal cortex and gut microbiota disturbance may lead to ALI‐induced CD.


| INTRODUC TI ON
Acute liver injury (ALI) refers to the impaired liver function that occurs in a short period of time and is mainly manifested by liver cell necrosis and elevation of hepatic enzymes in the blood. 1 The ALI can be caused by a variety of insults, such as acetaminophen-induced hepatotoxicity, drug-induced liver injury, and hepatitis viral infection. 2,3 In the early stage of ALI, some patients may recover on their own, while others will even develop hepatic encephalopathy (HE) to varying degrees. 4,5 HE is associated with higher mortality and serves as a prognostic indicator for ALI. 6,7 Minimizing the progression of HE in ALI is critical to improving patient outcomes. 8 The pathogenesis of HE is closely related to abnormal metabolism in the brain. Hyperammonemia and brain energy metabolism disorder may be responsible for the HE. 6,9 Studies are showing that increased levels of lactate and glutamine in the brain could lead to swelling of astrocytes in cirrhotic rats with HE. 10,11 Glutamine accumulation, dysregulated cerebral glucose metabolism, and changes in cerebral metabolism were also observed in the frontal lobe of comatose rats with acute liver failure. 12,13 Besides, altered glutamate and GABAergic neurotransmission is also reported to cause cognitive impairment in liver disease. 14 Recently, microbiota-host interactions have been demonstrated in several liver diseases, including ALI and HE. [15][16][17] Treatment and prevention strategies of ALI based on microbial modulation have been proposed. 18 Microbiome-targeted therapy, such as rifaximin 19 and fecal microbiota transplantation (FMT), show encouraging therapeutic outcomes in patients with HE, 20 although the mechanisms of action remain elusive. Thus, in this study, we investigated whether the shaping of the gut microbiota could affect cerebral metabolism and synaptic transmission and thereby regulate ALI-associated cognitive dysfunction (CD).

| Behavioral test
The open field test (OFT), Y-maze test (Y-maze), and novel object recognition test (NORT) were orderly performed to evaluate the cognition of mice. Animals' behavior was recorded by video cameras fixed on the ceiling and analyzed by ANY-MAZE software. OFT: OFT device is made of a gray polyethylene box (L × W × H: 40 × 40 × 40 cm).
Mice were placed in the center and allowed to move freely for 5 min.
The total distance traveled was analyzed. NORT: Two identical objects were placed in an open field, 6 cm away from the walls, 21 each mouse was allowed to explore freely for 5 min and then returned to the cage for 2 h. After one object was replaced with a novel object, mice were put back into the open field and allowed to explore freely for 5 min. The time mice spent exploring the novel object (TN) and the familiar object (TF) were recorded. The recognition index = TN/ (TN + TF). Y-maze: the Y-maze device consisted of three arms at an angle of 120° each (L × W × H: 30 cm × 8 cm × 15 cm). In the first learning stage, mice were placed in the starting arm (randomly assigned), while the new arm (randomly assigned) was blocked and allowed to explore freely for 10 min. The mouse was then returned to the cage for 2 h. In the testing stage, mice were placed in the same starting arm (the new arm was open) and allowed to explore for 5 min. The time spent on each arm was analyzed

| Fecal samples collection and Fecal microbiota transplantation (FMT)
Fecal samples were collected after all behavioral tests. Immediately after defecation, fecal samples were collected and stored at −80°C. FMT was applied as described before. 21 Briefly, mice were treated with water containing broad-spectrum antibiotics (1 g/L: Metronidazole; Ampicillin; Neomycin Sulfate; and Vancomycin 0.5 mg/kg) for 2 weeks to eliminate gut microbes. Then, the antibiotics were removed 24 h before FMT. Feces collected from the three Conclusion: Altered transmitter metabolism and synaptic transmission in the hippocampus and prefrontal cortex and gut microbiota disturbance may lead to ALIinduced CD.

K E Y W O R D S
acute liver injury, cerebral neurotransmitter metabolic, cognitive dysfunction, microbiota, synaptic transmission groups were resuspended in PBS (phosphate-buffered solution) at 0.2 g/ml. Gut microbiota-depleted mice were infused intragastrically with 0.2 ml of suspension or PBS once daily for 14 days.

| Tissue harvest and hematoxylin and eosin (H&E) staining
Mice were deeply anesthetized with sodium pentobarbital (i.p. 100 mg/kg) and rapidly decapitated to harvest blood. The upper serum was sent to the clinical laboratory of Tongji Hospital for the determination of total bilirubin (TBIL), alanine aminotransferase (ALT), and aspartate aminotransferase (AST) levels. The left lobe of the liver was fixed in a 4% neutral formaldehyde solution. The specimen was embedded in paraffin and sliced into 5 μm thick sections, then stained with hematoxylin and eosin. The liver damage was estimated by histological activity index analysis. 22

| 16S rRNA gene sequencing
16S rRNA gene sequencing of fecal samples was performed at OEBiotech Co. Ltd. DNA was extracted using DNA Extraction Kit (Tiangen Biotechnology Co., Ltd.). V3-V4 variable regions of 16S rRNA genes were amplified with universal primers 343 F and 798 R. Purified PCR products were used for sequencing. All representative reads were annotated and blasted against Unite database (ITSs rDNA).

| Sample preparation for 1 H-[ 13 C] NMR study
According to our previous research. 23  26 NaHCO 3 , and 10 glucose), a single brain slice was transferred onto the recording chamber and perfused with running 37°C ACSF equilibrated with 95% O 2 and 5% CO 2 at a rate of 2 ml/min. Only 1-2 brain slices per animal were used, and only 1 neuron was recorded in each slice.
Miniature inhibitory postsynaptic currents (mIPSCs) and miniature excitatory postsynaptic currents (mEPSCs) was recorded according to our previous research. 27

| Statistical analysis
Statistical analyses were performed using SPSS 17.0 (IBM) and GraphPad Prism 8.0. Y-maze results were hierarchically clustered by using the Ward method and Euclidean distance square as distance measurement and standardizing with the Z-score. 21 Data were analyzed by one-way analysis of variance, followed by post hoc Bonferroni's test. p < 0.05 was considered to indicate a statistically significant difference. The correlation analysis was performed using the psych package in the R environment and displayed by Cytoscape

| Effect of acute liver injury on cognitive function
Schematic diagram of the experimental design is illustrated in Figure 1A. Seven days after the first TAA injection, the animals (n = 20) manifested differential spatial learning and memory as evidenced by the different time spent in the novel arm in the Y-maze test, indicating that some mice had developed CD after ALI. We then divided animals into the mice with CD (susceptible, n = 11) and mice without CD (unsusceptible, n = 9) by the hierarchical cluster analysis of the Y-maze test results ( Figure 1B). The susceptible group, but not the unsusceptible group spent significantly less time in the novel arm than the control group ( Figure 1C). But ALI did not affect short-term memory ( Figure 1D). The total travel distance of the susceptible and unsusceptible groups showed a downward trend, indicating that the mice's spontaneous movement decreased in these two groups ( Figure 1E). Moreover, there were no differences in the total arm entries among the control, susceptible, and unsusceptible groups ( Figure S1A). The susceptible only manifested decreased novel arm entries while identical start arm, and familiar arm entries compared with the control and unsusceptible groups ( Figure S1B). These results suggested that the overall exploration of the Y-maze of the three groups is identical and is not affected by reduced locomotor activity.
Spatial memory impairment is raised by the reduced exploration of the novel arm of the susceptible group. Both susceptible and unsusceptible groups exhibited inflammatory cell immersion, bridge necrosis, and punctate necrosis in the livers ( Figure 1F,G). In addition, serum ALT and AST levels of susceptible and unsusceptible groups are significantly higher than in the control group ( Figure 1H,I).

| ALI-induced cognitive dysfunction is transmissible to intact mice by FMT
In order to assess the role of fecal microbiota for cognition, gut microbiota-depleted mice were transplanted with stool suspensions from different groups of mice. PBS gavage was applied in the FMT-free group. Compared to FMT-control mice (transplanted with control stool) and FMT-unsusceptible mice (transplanted with unsusceptible group stool), FMT-susceptible mice (transplanted with susceptible group stool) spent significantly less time in the novel arm ( Figure 2A). There was no significant difference in the NORT and OFT tests between these four groups ( Figure 2B,C). Besides, there was no significant difference in the serum ALT, AST, and TBIL levels, which indicated that changes in cognitive behaviors were not caused by liver injury after FMT ( Figure 2D-F). These results showed that the gut microbiota encompasses risk factors that could lead to ALI-associated CD phenotype. Thus, we performed a 16S rRNA sequencing to examine the difference in fecal microbiota among control, susceptible, and unsusceptible groups.

| Alternation of fecal microbiota in ALI mice with or without cognitive dysfunction
The Venn diagram shows that 259, 188, and 125 OTUs are unique to control, susceptible, and unsusceptible groups, respectively, revealing a great difference among the three groups ( Figure 3A The DAM in the unsusceptible group is enriched for Carotenoid biosynthesis ( Figure 3F).

| Reprogramming of cerebral metabolism in ALI mice with cognitive dysfunction
With the infusion of [1-13 C]-glucose, we traced the metabolic process of glucose in neurons and astrocytes by assessing metabolites labeled with 13 C at different positions. 28 In the FC, compared with the control group, the enrichment of GABA 3 (p = 0.0238), Glu 3 (p = 0.035), GABA 2 (p = 0.007), and Gln 4 (p = 0.001) were decreased in the susceptible group, while only the Glx 3 (overlapping signals of Gln 2 and Gln 3 , p = 0.041) enrichment was decreased in the unsusceptible group ( Figure 4A). In the PC, the GABA 2 (p = 0.043) enrichment in the susceptible group was decreased compared with the unsusceptible group, and the Gln 4 enrichment (p = 0.019) in the unsusceptible group was decreased compared with the control group ( Figure 4B). In the HIP, the Glx 3 (p = 0.049) enrichment differed between the two ALI groups. The GABA 2 enrichment (p = 0.045) was increased in two ALI groups compared with the control group.
The susceptible group had higher Gln 4 enrichment (p = 0.005) than the unsusceptible and the control groups ( Figure 4C). In the STR, the control group had significantly higher GABA 3 (p = 0.001) and lower NAA (p = 0.012) than the two ALI groups. The enrichment of Gln 4 (p = 0.028) and Glx 3 (p = 0.034) in the susceptible group were increased compared with the unsusceptible group ( Figure 4D).
In addition, the Glx 3 (p = 0.020) enrichment was decreased in the unsusceptible group compared with the control group in PON ( Figure S2D). Taken together, these findings provide evidence that neurotransmitter metabolism, particularly GABA metabolism, is changed in various brain regions associated with cognitive function after ALI.

| Correlation among metabolites, microbiota, and behaviors in ALI mice
We conducted a correlation analysis between microbiota and major neurochemicals, followed by network construction using the correlation coefficient ( Figure 5A, Appendix S1). This created a highly complex network, implying that gut microbes and neurochemical metabolism may have a complicated interaction. MCODE

| The mIPSCs and mEPSCs of CA1 and Prl were suppressed in ALI mice
Network analysis highlighted neurochemical metabolism in the HIP and FC as essential nodes of the network (higher MCC scores). Meanwhile, both HIP and FC have been widely implicated in cognition. We speculated that dysregulated neurotransmitter metabolism could alter neurotransmission in these brain regions, thereby impairing the cognitive function of affected animals. In  Figure 6B,E). There is no difference in the amplitudes of miniature postsynaptic currents (mPSCs) between groups, except that the susceptible group has smaller amplitudes of mIPSCs compared with the control group ( Figure 6C). The results showed that excitatory and inhibitory synaptic transmission can be suppressed by ALI and that cognitive impairment develops with aggravated inhibition of neurotransmission in CA1. Given that the reduction in frequencies generally indicates a presynaptic mechanism, the results are consistent with the finding that ALI causes dysregulated neurotransmitter metabolism. In addition, the decreased amplitudes of mIPSCs indicated that a postsynaptic mechanism might also be involved in ALI-associated cognitive impairment. A similar pattern of changes in the frequencies of mIPSCs and mEPSCs was also observed in PrL ( Figure 6G-K). However, the amplitudes of the mPSCs are comparable between all groups ( Figure 6G,I,J,L).
Taken together, these results indicate that suppressed synaptic transmission in CA1 and Prl caused by reprogrammed neurotransmitter metabolism may be the mechanism of ALI-induced CD in mice.

| DISCUSS ION
Clinical studies have shown that some ALI patients develop different degrees of HE and progress to ALF, eventually leading to the need for liver transplantation or death. 31,32 TAA is widely used to develop animal models of ALI and HE, 33,34 which are believed to share similar manifestations in humans. 35 In our study, we induced ALI in mice by multiple injections of low-dose TAA, resulting in hepatocyte necrosis and elevated plasma ALT and AST. Clustering behavioral performance in the Y-maze test identified spatial memory impairment in 55% (11/20) of ALI mice, similar to that observed in patients with chronic liver disease. 36 Microbiota dysbiosis is involved in liver disease 16,18 and its complications. 37 For example, Ruminococcaceae was found to be reduced in patients with HE. 38 Transplantation with stool rich in Ruminococcaceae improved cognition and reduced hospitalizations in HE patients. 39 In this study, we found that control mice have  (1 -4) in the metabolites. Note: The letters before the metabolite name represent the different brain regions, CE, cerebellum; FC, frontal cortex; HIP, hippocampus; PC, parietal cortex; PON, medulla-pons; STR, striatum; TC, temporal cortex; TH, thalamus. The data were analyzed by oneway ANOVA followed by Tukey's multiple comparisons test. Data are presented as the mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001. disease patients. 41 However, Erysipelotrichaceae correlates significantly with brain inflammation. 42 Consistent with these reports, Erysipelatoclostridium and Erysipelotrichaceae were increased in the unsusceptible and susceptible groups, respectively. Alloprevotella is increased in hepatitis B and alcoholic fatty liver, 43,44

and increased
Alloprevotella in the duodenum could predict the risk of cirrhosis and hepatocellular carcinoma. 45 Similarly, in our result, Alloprevotella is increased in the susceptible group. Impaired brain energy metabolism and neuronal dysfunction due to glutamine accumulation-induced swelling of astrocytes is a major mechanism of HE. 46,47 Besides, abnormal glutamine accumulation may also affect the synthesis of neurotransmitters like glutamate and GABA. 48 Neurotransmitter alterations due to energy metabolism disorders are reported to be directly involved in the pathogenesis of HE. 49 In vitro 1 H-[ 13 C] NMR studies in rats have shown that the enrichments of Gln 4 and Clu 4 are increased in the FC of HE rats, while GABA 2 enrichment is unchanged. 11,12 In our study, the Gln 4 enrichment in the susceptible group is increased in subcortical regions such as HIP and STR, but not in cortical regions. This may be explained by the findings that higher ammonia uptake is observed in subcortical brain regions in MHE patients, making subcortical metabolism more sensitive to ammonia. 46 In addition, GABA 2 enrichment is decreased in cortical areas such as FC and PC in the susceptible group and is increased in HIP. The results are consistent with previous reports that in the rats with hyperammonemia, extracellular GABA content is increased in HIP but decreased in cortical areas. 50 The gut microbiota has been reported to modulate neurotransmitters. 51 However, there is no report on the relationship between gut microbiota and the metabolic dynamics of specific neurotransmitters in the brain. Escherichia produces GABA in the human gut. 52 The increase in Escherichia.Shigella has been described as a pro-inflammatory flora in F I G U R E 5 Correlation network analysis among microbiota, neurochemical enrichments, and behavioral test. (A) The Spearman test was used to assess the correlation between major neurochemical enrichments and major microbiota in the genus. Module identification of correlation network by Molecular Complex Detection (MCODE) plugins (degree ≥ 10, Node Score ≥ 0.4, K-Core ≥ 2, max depth: 100). (B) Top 10 hub candidates identified by Maximal Clique Centrality (MCC). (C) Correlation network analysis of microbiota, neurochemical enrichments, and behavioral test. Note: Metabolites having a mean enrichment > 0.05 (Lac, Alanine, GABA 3 , Glu 3 , Glx 3 , GABA 2 , Glu 4 , Gln 4 , Asp 3 , GABA 4 , Glx 2 , and Asp 2 ) and top 100 microbiotas were selected as major neurochemicals and major microbiotas to report. The spearman coefficient (r) was adjusted by FDR and ranged from positive (light pink edge) to negative (light blue edge). Node size is proportional to the node's MCC score, indicating its interaction value. Nodes are colored according to behavior, microbiota in the genus, and brain regions of metabolites. (D) ROC curve of major microbes. cognitively impaired elderly 53 and post-operative mice with cognitive dysfunction 54 and is negatively correlated to cognition in Alzheimer's disease patients. 55 Consistently, we found that increased abundance of Escherichia.Shigella is negatively correlated with FC-GABA 2 and all three behaviors. Alloprevotella is reported to be negatively correlated to cognition in Phenylketonuria mice 56 and related to transmitter metabolism in hippocampus. 57 In this study, the increased Alloprevotella is significantly negatively correlated with Y-maze and positively correlated to HIP-Gln 4 . We proposed that Alloprevotella and Escherichia.
Shigella could be a potential biomarker to predict cognitive impairment in ALI (AUC = 0.872 and 0.733). These findings suggest that specific microbiota correlates to metabolic dynamics of brain transmitters and behavior.
Maintenance of cognitive function depends on normal synaptic transmission. Previously reported that neurotransmission disorders are responsible for the cognitive function alteration in HE. 14 We found that the frequencies of both mIPSCs and mEPSCs in CA1 and PrL were decreased in the susceptible group compared to the unsusceptible group. These changes were consistent with the reprogrammed metabolic kinetics, highlighting a presynaptic mechanism underlying ALI-induced cognitive dysfunction. In addition, the correlation of behaviors, microbes, and metabolites provides a novel insight into the perspective of gut microbiota and brain metabolism for the cognitive dysfunction in ALI-induced CD mice. However, the effects of specific microbes and metabolites have not been validated.
The development of therapeutic strategies targeting Escherichia.
Shigella and Alloprevotella have great application prospects for preventing and treating ALI-induced CD.

| CON CLUS ION
This work revealed altered gut microbiota and cerebral metabolism in acute liver injury mice with and without cognitive dysfunction.

ACK N OWLED G M ENT
This study was supported by the National Natural Science Foundation of China (Nos. 81873467 and 21921004).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in the supplementary material of this article.