Effects of Gabexate Mesylate on the Gut Microbiota and Metabolomics in Rats with Sepsis

Introduction

Sepsis affects the lives and health of millions of people worldwide annually. Sepsis results in life-threatening organ dysfunction caused by loss of control over infection.¹ One of the main causes of organ failure in sepsis is the inflammatory response due to infection.² According to statistical data, the sepsis mortality rate in hospitalized patients is 26.7%, and the sepsis mortality rate in the intensive care unit is 41.9%.³ Despite recent decades, no treatment has been approved for sepsis other than antibiotics and supportive care.⁴ Finding a promising drug to reduce sepsis mortality remains a serious challenge.

Gabexate mesylate (GM) is a non-peptidic serine protease inhibitor,⁵ which can inhibit the coagulation cascade and be used clinically as an anticoagulant in patients with acute pancreatitis⁶ and disseminated intravascular coagulation (DIC).⁷ Abnormal coagulation can lead to a large increase in plasma plasminogen activator inhibitor 1 (PAI-1)⁸ level, the formation of microthrombi and pulmonary vascular injury.⁹ GM can also inhibit various serine proteases produced during inflammation,¹⁰ which has a certain anti-inflammatory effect. It directly inhibited the activation of NF-κB in lipopolysaccharide (LPS)-stimulated macrophages and human monocytes, resulting in decreased production of inflammatory cytokines such as TNF-α, IL-6 and high mobility group protein 1 (HMGB1).¹¹ Besides, GM has a protective effect in acute lung injury caused by endotoxin,¹² ischemia/reperfusion,¹³ and smoke inhalation.¹⁴ It alleviated LPS-induced coagulation abnormalities and pulmonary vascular injury by inhibiting TNF-α production.¹² Vitro experiments demonstrated that GM inhibited the function of activated neutrophils. However, inflammatory mediators produced by neutrophils, such as granulocyte elastase and superoxide free radicals, as well as neutrophil aggregation, are associated with pulmonary vascular injury.¹⁵ It also attenuates LPS-induced lung injury by inhibiting the expression of leukocyte adhesion molecules in endothelial cells.¹⁶ Studies have found that GM indirectly inhibits the expression of HMGB1 and reduces lung injury by inhibiting PAI-1 and protease activated receptor 2 (PAR-2).¹¹ Since PAI-1 levels are associated with the severity and prognosis of multiple organ failure in patients with sepsis, GM can also partially prevent the development of organ dysfunction through anticoagulation.¹⁷ Some studies have shown that GM significantly reduced the neutrophil infiltration scores in lung, liver and kidney, and alleviated organ dysfunction in septic rats.^18,19 But the potential mechanisms by which GM can be used for sepsis treatment have not been elucidated.

The gut is the main origin of microorganisms that lead to sepsis. Gut microbiota regulates the homeostasis of healthy hosts through immune function and intestinal barrier.²⁰ Decreased diversity of the microbiome,²¹ thinning of the gut mucus layer and increased gut permeability, allow pathogens to thrive in the gut lumen.²² Several studies reported that the gut microbiota dysbiosis increased susceptibility to sepsis²³ and organ dysfunction, which associated with sepsis progression.²⁴ Gut bacterial content determines the severity of systemic damage in sepsis, suggesting that dysbiosis also affects sepsis prognosis.²⁵ Increasing evidence indicates that gut microbiota plays a key role in pulmonary immunity and is associated with airway homeostasis.²⁶ For example, intestinal microbes can enhance host defense against pneumonia through increased IL-17 and upregulation of granulocyte-macrophage colony-stimulating factor (GM-CSF) signaling.²⁷ The gut microbiota derived metabolites, such as short-chain fatty acids (SCFAs), can induce the production of immune cells and cytokines, which subsequently enter the systemic circulation to regulate immune and inflammatory responses in the lung.²⁸ Lung SCFA levels are dependent on functional gut microbiota, and SCFA (propionic acid) reprograms the metabolism of alveolar macrophages. In addition, oral SCFA can reduce airway hyperresponsiveness and increase regulatory T cells in bronchoalveolar lavage fluid.²⁹ It can target alveolar macrophages to remotely influence lung immune responses, modulating asthma³⁰ and respiratory infections.³¹ However, SCFA (acetate) supplementation reduced the bacterial burden after influenza attack, alleviated pulmonary pathology and improved survival.³²

The gut microbiota plays a key role in sepsis and septic shock, and gut dysbiosis could contribute to the onset of sepsis by producing harmful metabolites and altering host physiological processes. Microbe-derived SCFAs also contribute to intestinal mucosal homeostasis and enhance clearance of pathogenic microorganisms by immune cells such as macrophages.³³ These bacterial metabolites are essential elements of the microbiota–host interaction. Dysbiosis leads to reduced levels of gut microbiota-derived SCFAs, which may be due to reduction of Parabacteroides, Prevotella, and Bifidobacterium abundance and further triggers immune cascades in the respiratory tract.³⁴ Limited evidence suggests that microbiome-based therapies, including probiotics and fecal microbiota transplantation, may reduce sepsis risk and improve septic outcomes.^35,36

In this study, we aimed to investigate the effect of GM on regulation of the gut microbiota, identify the metabolites involved and explore the protective effect on sepsis. Since the current treatment of sepsis is facing huge challenges, our findings may provide evidence for new therapeutic strategies for sepsis.

Materials and Methods

Instruments and Reagents

The following instruments and reagents were used: bench-type high-speed centrifuge (Eppendorf), PCR instrument (Bio-Rad), electrophoresis instrument (Tanon), Illumina NovaSeq sequencer (Illumina), Vanquish ultra-high-performance liquid chromatograph (Thermo Fisher Scientific), high-resolution mass spectrometry Q Exactive HFX (Thermo Fisher Scientific), Column ACQUITY UPLC HSS T3 (2.1 mm × 100 mm, 1.7 μm) (Waters); Magnetic Bead Method Soil and Fecal DNA Extraction Kit (Tiangen Biochemical Technology [Beijing] Co., Ltd., DP712), Phusion® High-Fidelity PCR Master Mix with GC Buffer (Thermo Fisher Scientific, F532S), Phusion® High-Fidelity DNA polymerase (New England Biolabs, M0530S) NEB Next® Ultra DNA Library Prep Kit (Illumina, E7370L), methanol (CNW Technologies, CAS 67–56-1), acetonitrile (CNW Technologies, CAS 75–05-8), ammonium acetate (Sigma-Aldrich, CAS 631–61-8), ammonia (Fisher Chemical, CAS 1336–21-6), Ultra Pure Water (Watsons) and GM (0.1 g, Changzhou Siyao Pharmaceuticals Co., Ltd.).

Experimental Animals and Grouping

Sixty 6–8-week-old adult male Sprague–Dawley (SD) rats weighing 200–250 g (https://www.vitalriver.com/). All rats were housed in a 12/12-h day and night animal room for a week before the experiment and allowed to eat and drink ad libitum. The temperature in the animal room was controlled from 20 ℃ to 25 ℃, and the relative humidity was controlled from 40% to 60%. SD rats were randomly divided into sham operation control (SC) group, sepsis after cecal ligation and puncture (CLP) group, and intraperitoneal injection of GM 1 h after CLP group. The establishment of the CLP model was based on our previous studies.^37,38 The rats were weighed and anesthetized by intraperitoneal injection of 30 mg/kg pentobarbital. A 2-cm incision was made on the midline of the abdomen, the cecum was exposed, and the feces were squeezed in the cecum to the free end. It was ligated at the middle and lower one-third of the cecum, and an 18-G needle was used to puncture the middle of the cecal ligation. After the puncture, a small amount of feces is squeezed out, and the cecum is returned to the original position of the abdomen. The muscle and skin layer were sutured with No. 5 and No. 3 surgical sutures, respectively. Immediately after the procedure, subcutaneous injection of 50 mL/kg of warm saline was used for resuscitation, and the rats were irradiated with warm light for 30 min. The SC group underwent the same procedure without ligation and puncture. The GM group received intraperitoneal injection of GM (40 mg/kg) 1 h after CLP. One hour after CLP, rats in the SC and CLP groups were intraperitoneally injected with normal saline. The experiments were conducted according to the guidelines established by the National Institutes of Health (http://grants1.nih.gov/grants/olaw/) and approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University.

Sample Collection

The survival rate of the rats was determined 24 h after the operation. Colonic feces of the surviving rats in each group (two tubes of samples were obtained from each rat) were squeezed and placed into 2-mL cryopreservation tubes separately and stored in liquid nitrogen. The perfused lung tissues were collected, and the perfusion procedure was performed similar to that in the previous study.³⁷ Lung tissues were fixed with 4% paraformaldehyde, embedded in paraffin overnight, stained with hematoxylin and eosin (HE), and observed under a microscope (Olympus, Japan).

Fecal DNA Extraction and High-Throughput 16S rRNA Sequencing

DNA was extracted from the fecal samples using sodium dodecyl sulfate; then, the DNA was diluted to 1 ng/μL sterile water, and the primers were 341F:CCTAYGGGRBGCASCAG and 806R:GGACTACNNGGGTATCTAAT, targeting the 16S V3–V4 region. The DNA purity and concentration were verified by electrophoresis in 2% agarose gel and purified by magnetic spheres. Library construction was qualified and sequenced on NovaSeq 6000.

16S rRNA Sequencing Data Analysis

The original data were spliced and filtered to obtain valid data. Based on the valid data, the clusters of operating taxonomic units (OTUs) and species classification were analyzed with 97% consistency. Subsequently, the species was recorded on the representative sequence of each OTU to obtain species-related information and species-based abundance distribution. Chao1, ACE, Shannon, Simpson, Good’s coverage, and phylogenetic diversity (PD) whole tree indices were calculated using QIIME software. Principal component analysis (PCA) was performed using the ade4 package and ggplot2 package of R software (version 2.15.3); principal coordinate analysis (PCoA) was performed using WGCNA and stats and ggplot2 packages; and R software was used to draw the rarefaction curve and analyze the differences between the groups of α- and β-diversity. Parametric and nonparametric tests were used in the data analysis, t-test and Wilcoxon test were used for two-group comparison, and Tukey’s and Wilcoxon tests were used for three-group comparison. Linear discriminant analysis (LDA) effect size (LEfSe)³⁹ was performed by comparing two or more groups and examining markers that are statistically different between groups. LEfSe used the Kruskal–Wallis rank-sum test to detect characteristics with significantly different abundance levels among the specified taxa, and LDA was performed to estimate the effect size, emphasizing statistical significance and biological relevance. Tax4Fun was used to predict the microbiota function. Correlations between different species and functional pathways were calculated using Spearman rank correlations, and correlations were presented using a heatmap.

Fecal Sample Preparation for Metabolomic Analysis

Moreover, 25 mg of thawed stool sample was obtained and mixed with 500 μL of extract (methanol: acetonitrile: water = 2:2:1). Then, 20 μL of 0.3 mg/mL 2-chloro-D-phenylalanine was added. All samples were ground and ultrasonically extracted in ice water for 5 min, and the steps were repeated two to three times and allowed to stand at −40 ℃ for 1 h. Then, the supernatant was used in the analysis by centrifugation at 12,000 rpm for 15 min at 4 ℃. Each 20 μL of sample was mixed to obtain a quality control (QC) sample, and 200 μL of QC sample was collected for QC analysis.

LC-MS Metabolomics-Based Data Acquisition

Chromatographic separation of metabolites was performed using ultra-high-performance liquid chromatography (UPLC) and UPLC BEH Amide column (2.1 mm × 100 mm, 1.7 μm) (autosampler temperature, 4℃; injection volume, 2 μL). Liquid chromatography phase A consisted of an aqueous phase containing 25 mmol/L ammonium acetate and 25 mmol/L ammonia water, and phase B contained acetonitrile. The conditions of the ionization source of the electrospray were set as follows: the flow rate of the gas in the sheath is 30 Arb, the flow rate of the auxiliary gas is 25 Arb, and the capillary temperature is 350℃. The resolution of mass spectrometry for primary information acquisition was set to 60,000, and secondary information acquisition was set to 7500. The collision energy was 10/30/60 in normalized collision energy mode, and the spray voltage was 3.6 kV (positive) or −3.2 kV (negative). The Thermo Q Exactive HFX mass spectrometer collects mass spectral data as raw data.

Metabolomics Data Analysis

After processing the raw data for peak identification, peak extraction, peak alignment, and integration, a dataset containing mass-to-charge ratio, retention time, and peak intensity was obtained. Then, the data were preprocessed, normalized and matched with an internal mass spectrometry database (BiotreeDB V2.1), and metabolites were annotated with a cutoff point of 0.3. Unsupervised PCA was performed to visualize metabolic changes between groups. Discriminant analysis of orthogonal partial least squares (OPLS-DA) was applied to visualize the maximum difference between the two groups. The variable importance in projection (VIP) was calculated, and a high VIP indicated that the metabolite has a large contribution to the grouping. Data were analyzed using one-way analysis of variance and Tukey’s post hoc test (SPSS software version 21.0, Chicago, IL). Metabolites with VIP > 1 and P < 0.05 were selected as differential metabolites. Heatmaps were used to show trends in differential metabolites. Pathway analysis was performed using MetaboAnalyst 3.0, which is based on database sources, including the KEGG database (http://www.genome.jp/kegg/) and Human Metabolome Database (https://hmdb.ca/).

Results

Composition of the Gut Microbiota in the Three Groups

The rarefaction curve of each sample was close to saturation, the amount of sequencing data was reasonable, and the sequencing depth was sufficient (Figure 1A). Sample valid sequences with 97% identity were clustered into OTUs. The gut microbiota was annotated with representative OTUs, with a total of 2126 OTUs. Moreover, 660, 616 and 850 OTUs were identified in the SC, CLP, and GM groups, respectively, of which 495 OTUs were shared by three groups (Figure 1B). In the phylum-level gut microbiota, the predominant phyla in the SC and GM groups were Firmicutes, Proteobacteria, and Verrucomicrobiota in the CLP group (Figures 1C). The Firmicutes–Proteobacteria ratio was significantly decreased in the CLP group (Figures 1D). The predominant genera in the SC and GM groups were Lactobacillus, Ruminococcus, and Blautia. The predominant genera in the CLP group were Escherichia–Shigella, Akkermansia, and Enterococcus (Figure 1E).

Figure 1 Effect of GM on the gut microbiota of rats with sepsis. (A) Rarefaction curves plotted for each sample. (B) The Venn diagram of the OTUs of each group. (C) Histogram of relative abundance of top ten species at the phylum level. (D) The ratio of Firmicutes to Proteobacteria in the SC, CLP, and GM groups; *compared with the CLP group, P < 0.05. (E) Relative abundance of the top ten species at the genus level. (F) PcoA analysis based on weighted Unifrac distance.

α-Diversity Analysis

The Simpson index represented the diversity and evenness of species distribution. The Wilcoxon rank-sum test showed that the diversity and evenness (Simpson index) of species distribution in the SC and GM groups were higher than those in the CLP group. The Shannon index, which measures the total number of microbial communities and their proportions, was also higher in the SC and GM groups than in the CLP group. The Chao1 index and ACE were used to evaluate the total number of species and OTUs, respectively. Interestingly, the Chao1 index and ACE were the lowest in the GM group. Good’s coverage index represents the sequencing depth. Good’s coverage rates of the three groups were all greater than 99.8%, indicating that the sequencing depth of the gut microbiota was ideal. The PD whole tree index represents the genetic relationship of the species in the community. The PD whole tree index in the SC and GM (P < 0.01) groups was higher than in CLP group (Table 1).

Table 1 Diversity Analysis of a Single Sample with Different Groups Under the 97% Consistency Threshold

β-Diversity Analysis

To assess the degree of similarity between microbial communities, conducted sampling comparative analysis (β-diversity analysis) for different groups. The PCoA results showed that samples from the SC, CLP, and GM groups formed three distinct clusters, among which samples from the SC and GM groups tended to cluster together, indicating that the two groups had similar species composition. The CLP group was far away, indicating that it had a larger community difference compared with the SC and GM groups (Figure 1F). Through ANOSIM and Adonis nonparametric tests, the obtained value of Pr (>F) is P < 0.05. It indicated a high degree of sample difference and statistical significance of PCoA analysis results between SC-CLP and CLP-GM groups. The results are shown in Supplementary Table 1 and Supplementary Table 2.

Differential Microbiota Analysis in the SC, CLP, and GM Groups

To identify the specific microbiota associated with the SC, CLP, and GM groups, we used LEfSe analysis. According to the criteria of LDA > 4 and P < 0.05, the differential microbiota was screened. There are three differential genera, namely, Lactobacillus, Romboutsia, and Escherichia–Shigella, between the CLP and SC groups (Figure 2A). Similarly, there were five differential genera between the CLP and GM groups, namely, Lactobacillus, Pygmaiobacter, Erysipelotrichaceae_UCG_003, Escherichia–Shigella, and Akkermansia (Figure 2B). The only differential genera between the SC group and GM group was Parasutterella (Figure 2C).

Figure 2 Comparison of abundances of biomarkers with statistical differences in different groups. (A) Linear discriminant analysis (LDA) relative abundance map of differential bacterial genera between the CLP and SC groups. The columns represent the relative abundance of each sample, while the solid and dotted lines represent the mean and median, respectively. (B) Linear discriminant analysis (LDA) relative abundance map of differential bacterial genera between the CLP and GM groups. (C) Linear discriminant analysis (LDA) relative abundance map of differential bacterial genera between the SC and GM groups.

Correlation Analysis of Genera Between the CLP and GM Groups

The gut microbiota is interconnected to maintain homeostasis, and studying the interactions between different genera contributes to finding the important role of microbiota of the CLP and GM groups (P < 0.05). Notably, the GM-enriched genera had higher associations than the CLP group-enriched genera. Akkermansia was strongly positively correlated with Desulfovibrio. Lachnospiraceae UCG-006 closely related to lipid metabolism was strongly negatively correlated with Lactobacillus in the CLP group. Moreover, Lactobacillus was strongly negatively correlated with Desulfovibrio, which was considered to be toxic to the intestinal epithelium (Supplementary Figure 1). In the GM group, Ruminococcus was strongly negatively correlated with common opportunistic pathogens Escherichia–Shigella and Enterobacteria, and Candidatus Saccharimonas with anti-inflammatory property was strongly negatively correlated with Escherichia–Shigella (Supplementary Figure 2). The enriched Muribaculum in the CLP group was strongly negatively correlated with the enriched Lactobacillus in the GM group and strongly positively associated with the enriched Lachnospiraceae UCG-006 in the CLP group. The enriched Akkermansia in the CLP group was strongly positively correlated with Terrisporobacter, Proteus, and Odoribacter. Candidatus Saccharimonas enriched in the GM group was strongly negatively correlated with Escherichia–Shigella enriched in the CLP group (Supplementary Figure 3).

Functional Alterations of GM on Gut Microbiota in the CLP Group

To characterize the functional alterations of gut microbiota following GM application in rats with sepsis, we used Tax4Fun to analyze 16S rRNA and predicted functional composition profiles. Eighteen pathways were differentially enriched (P < 0.05) between the CLP and GM groups (Supplementary Figure 4). Then, we investigated the correlation between different species (n = 27) and KEGG pathways (n = 44) using Spearman correlation analysis (Student’s t-test) (Supplementary Figure 5). CLP-enriched Akkermansia was negatively correlated with phospholipase D signaling pathway and carbohydrate metabolism signaling pathway. GM-enriched Lactobacillus is associated with several pathways, such as cell cycle, tight junction, phosphotransferase system, isoflavone biosynthesis, caffeine metabolism, thyroid hormone synthesis, mineral absorption, and cellular motility and secretion, and positively correlated with transport and cancer pathways. GM-enriched Ruminococcus was positively correlated with amino sugar and nucleotide sugar metabolism, while Lactobacillus was negatively associated with apoptosis, lipopolysaccharide biosynthesis, steroid hormone biosynthesis, transport, sphingolipid metabolism, and pathways in cancer, fluid shear stress, and atherosclerosis. The dominant Phascolarctobacterium in the CLP group was positively correlated with G protein coupled receptors. Moreover, the dominant Colidextribacter in the CLP group was positively correlated with polyketide sugar unit biosynthesis, terpomycin biosynthesis, and N glycan biosynthesis (Supplementary Figure 6).

Metabolite Analysis of the SC, CLP, and GM Groups

Six samples and a QC sample were run during the running process. The aggregation of the QC samples was good, indicating that the method was stable and experimental data had high quality (Supplementary Figure 7). The total ion chromatograms of the SC, CLP, and GM groups were observed (Supplementary Figure 8). In PCA, the SC group was closer to the GM group and was significantly separated from the CLP group (Figures 3A and D). Using OPLS-DA, the SC, CLP and CLP, GM groups can be divided according to their metabolic differences, and the distinction is significant (Figures 3B and C, 3E and F).

Figure 3 (A–C) represent positive ion mode; (D–F) represent negative ion mode. (A and D) represent the PCA diagrams of the SC, CLP and GM groups. (B and E) represent the OPLS-DA diagrams of the SC and CLP groups. (C and F) represent the OPLS-DA diagrams of the GM and CLP groups.

Differential Metabolite Identification

Multivariate statistical analysis was used to screen differential metabolites (VIP > 1.0 and P < 0.05). In the positive ion mode, the SC group had 28 differential metabolites compared with the CLP group, eight were significantly upregulated in the SC group, and 20 were significantly downregulated in the SC group (Figure 4A). The GM group had 26 differential metabolites compared with the CLP group: 13 were significantly upregulated and 13 were significantly downregulated (Figure 4B). In the negative ion mode, the SC group had 17 differential metabolites compared with the CLP group: 12 were significantly upregulated and 5 were significantly downregulated (Figure 4C). The GM group had 20 differential metabolites compared with the CLP group: 16 were significantly upregulated and 4 were significantly downregulated (Figure 4D). There were 15 significantly changed metabolites in the SC, CLP, and GM groups. The detailed information of each metabolite is shown in Table 2. As shown in Figure 4E, 15 metabolites are mainly divided into 11 fatty acids and 4 other classified metabolites, namely, 10E, 12Z-octadecadienoic acid, 9,10-epoxyoctadecenoic acid, 15-KETE, and alpha-linolenic acid. These four fatty acids are grouped together, and the relative abundance is lower in the CLP group but higher in SC and GM groups.

Table 2 Detailed Information of Differential Metabolites of the SC, CLP, and GM Groups (VIP > 1, P < 0.05)

Figure 4 Heatmap of fecal differential metabolites in positive and negative ion mode. Red indicates highly expressed, and blue indicates expressed at a low level. (A) SC and CLP group (positive ion mode). (B) GM group and CLP group (positive ion mode). (C) SC and CLP group (negative ion mode). (D) GM group and CLP group (negative ion mode). (E) Heatmap of common differential metabolites of the three groups.

Metabolic Pathway Analysis of Key Differential Metabolites

Using MetaboAnalyst 3.0 and the KEGG database, the different metabolite pathways between the CLP and GM groups were analyzed. It was found that GM mainly regulates sphingolipid metabolism, histidine metabolism, steroid biosynthesis, glycerophospholipid metabolism, primary bile acid biosynthesis, and alpha-linolenic acid metabolism (Supplementary Figure 9). They are consistent with the predicted pathway of Tax4Fun function, and the GM group was negatively related to apoptosis, lipopolysaccharide biosynthesis, steroid hormone biosynthesis, and sphingolipid metabolism. Therefore, changes in gut metabolites after GM application may be related to gut microbiota dysbiosis.

Links Between Different Genera and Metabolites

To further investigate the microbiota–intestinal metabolite relationship after GM application, Spearman correlation analysis was used to determine the connections between 27 different intestinal genera and differential metabolites between the CLP and GM groups. Some CLP-rich genera (Akkermansia and Escherichia–Shigella) were positively associated with Cortisol and Palmitoylethanolamide, and negatively correlated with Choline, Deoxycholic acid, 10E,12Z-Octadecadienoic acid, Chenodeoxycholic acid, [4]-Gingerdiol 3,5-diacetate, Oleoyl glycine and 9,10-Epoxyoctadecenoic acid. Moreover, the GM-rich genus (Lactobacillus, Pygmaiobacter and Erysipelotrichaceae_UCG-003) were positively associated with Deoxycholic acid and Chenodeoxycholic acid, and negatively correlated with Calystegin A3, Palmitoylethanolamide, [10]-Gingerdione, 12-Hydroxy-11-methoxy-8,11,13-abietatrien-20-oic acid. Notably, these 3 metabolites (Palmitoylethanolamide, Deoxycholic acid and Chenodeoxycholic acid) have significant correlation with CLP- genus (Akkermansia and Escherichia–Shigella) and GM-rich genus (Lactobacillus, Pygmaiobacter and Erysipelotrichaceae_UCG-003) (Figure 5).

Figure 5 Heatmap between different genera of intestinal microbiota and metabolites in CLP and GM groups. Red, positive correlation; blue, negative correlation. × mean P < 0.05. (A) Positive ion mode. (B) Negative ion mode.

Effects of GM on Lung Damage and Survival in Rats with Sepsis

The mortality rate in the CLP group (10/20) at 24 h after operation was higher than in the SC group (0/20) (P < 0.001) and GM group (4/20) (P < 0.05) (Figure 6A). Lung histopathological scores were conducted according to Osman criteria,⁴⁰ and inflammatory infiltration, edema, and structural damage of the lung tissue in the CLP group were significantly worse than those in SC (P < 0.001) and GM (P < 0.001) groups (Figure 6B and C).

Figure 6 (A) The mortality rate in the SC, CLP, and GM groups at 24 h after operation; Comparison with CLP, *P < 0.05, ***P < 0.001. (B) The histopathological section of the lung tissue in the SC, CLP, and GM groups, respectively (bar = 100 μm). (C) The lung histopathological scores in the SC, CLP, and GM groups; Comparison with CLP, ***P < 0.001.

Discussion

The study demonstrates that GM could attenuate sepsis-related lung injury and reduce sepsis mortality by modulating the gut microbiota and its metabolites in septic rats. The abundance of gut microbiota associated with SCFA production and metabolites related to bile acid metabolism were all increased in GM-treated septic rats. Besides, the alterations in the gut microbiota of GM were significantly correlated with these metabolites. Our study supports the possibility of GM in sepsis treatment because GM protects against acute lung injury in sepsis.

The gut microbiota could prevent the colonization of intestinal pathogens, and studies have shown that impaired gastrointestinal function in sepsis leads to significant changes in intestinal flora, manifested as dysbacteriosis.^41,42 Studies have found that infiltrating neutrophils and other immune cells can release proteolytic enzymes and activate PAR, resulting in distribution changes of tight junction-related proteins, such as ZO-1 and occludins.⁴³ The intestinal epithelial barrier integrity is disrupted⁴⁴, promoting harmful bacteria into the systemic circulation and resulting in the imbalance of gut microbiota.⁴⁵ Mice with high proteolytic activity had higher intestinal permeability and lower relative abundance of gut microbiota.⁴⁵ Theoretically, inhibition of serine proteases could alleviate disruption of intestinal homeostasis and have beneficial effects. Previous studies have shown that another serine protease inhibitor, ulinastatin, protects the integrity and function of the intestinal barrier in sepsis, reducing leakage of gut bacteria.⁴⁶ GM also can inhibit the function of various serine proteins and activated neutrophils, the production of PAR-2 and leukocyte adhesion molecules to reduce sepsis-induced inflammation. In this study, we observed an increase in the abundance of Lactobacillus, Pygmaiobacter and Erysipelotrichaceae_UCG-003 after GM treatment. Lactobacillus as a natural gastrointestinal flora have various beneficial effects, including interference with intestinal pathogens, maintaining gut barrier and regulating mucosal and systemic immune responses.⁴⁷ It was widely used in the prevention of Clostridium difficile infection.⁴⁸ And recent studies found Lactobacillus significantly improved intestinal dysbiosis,⁴⁹ reduced the risk of late-onset sepsis.⁵⁰ In addition, both Pygmaiobacter and Erysipelotrichaceae_UCG_003 are the main bacteria producing butyric acid, which is a type of SCFA.^51,52 And SCFAs are potent anti-inflammatory substances involved in mediating immune responses and inducing the formation of regulatory T cells in sepsis.^53,54 Therefore, GM-enriched genera contribute to gut microbiota homeostasis and regulate the inflammatory response. Besides, compared to CLP group, the Escherichia–Shigella and Akkermansia were significantly reduced in GM group. Escherichia–Shigella as a common pathogenic bacteria can promote the production of LPS⁵⁵ and induce macrophage death.⁵⁶ Interestingly, Akkermansia was considered to be a new generation probiotic, but it was significantly reduced after the use of GM. However, it has been reported that excessive Akkermansia can disrupt host mucins, increase intestinal permeability,⁵⁷ induce inflammation and promote colon tumorigenesis.^58,59 Increased Akkermansia abundance and decreased SCFA-producing bacteria also aggravate colorectal cancer.⁶⁰ Additionally, Seibert et al found that Akkermansia abundance was significantly increased at peak viral infection and allergic diseases,^61,62 which was also consistent with our findings.

The gut microbiota synthesizes, regulates and degrades a variety of metabolites as functional supplements for host metabolism, especially for dietary components that cannot be metabolized by the host.⁶³ Nearly 10% of metabolites in blood and more than 50% of metabolites in feces and urine originate from or are modified by gut microbiota.^64,65 Microbial products, such as SCFAs, may influence host metabolism by regulating appetite, lipogenesis, glucongenesis, inflammation, and other functions.⁶⁶ A byproduct of intestinal metabolism, bile acids are known to regulate hepatic metabolism,⁶⁷ which are also important for clearing bacteria and metabolic adaptation to inflammation in sepsis.⁶⁸ Deoxycholic acid activates the G protein-coupled receptor (TGR5), which affects neuroinflammation, cognitive or behavioral and became a therapeutic target for neurological outcomes in sepsis survivors.⁶⁹ Cholestasis is a common complication of sepsis, and elevated plasma Chenodeoxycholic acid levels predict sepsis-related mortality.⁷⁰ The Chenodeoxycholic acid derivative obeticholic acid could improve bile acid homeostasis and inhibit levels of inflammatory cytokines, which indicated bile acids were protective against sepsis.⁷¹ Our study found that both deoxycholic acid and chenodeoxycholic acid were markly increased after treatment with GM, but Palmitoylethanolamide was decreased. Palmitoylethanolamide is an endogenous fatty acid amide that selectively activates peroxisome proliferator-activated receptor α (PPARα). Oxidation of fatty acids under starvation or fasting conditions stimulates transcription of PPARα-regulated genes and activates the oxidative system.⁷² PPARγ is mainly involved in the regulating energy balance and lipid biosynthesis. Among them, the PPARγ2 isoform, as a receptor of 9, 10-DHOME, can prevent lipotoxicity through different mechanisms and enhance the lipid buffering capacity of liver and muscle.⁷³ This may be the reason why 9, 10-DHOME was significantly increased and Palmitoylethanolamide was decreased after GM treatment. Taken together, the improved survival rate of septic rats treated with GM may be attributed to changes in metabolites. However, the mechanisms underlying GM-induced changes in metabolite remain to be investigated.

Our study is the first to elucidate the relationship between intestinal microbiota and metabolomics after GM administration in sepsis. However, the study also has some limitations. First, the experiments had a small sample size, and the research results need to be verified with more animal experiments and clinical studies. In addition, we did not use pseudo-germ-free mice and fecal microbiota transplantation to verify a cause-consequence between microbiota and sepsis. Furthermore, although the effect of GM on lung damage using HE-stained sections was observed, a more complete experimental design is needed to explore the molecular mechanism by which GM improves sepsis prognosis.

In conclusion, this is the first study to elucidate that GM may improve sepsis-related lung injury and regulate gut microbiota and related metabolites, which provided a new idea for the treatment of sepsis.