Investigation of “Microbial-mitochondrial” cross-talk to explore the compatibility mechanism of Xiaojianzhong Tang against chronic atrophic gastritis rats

2.1. Reagents and materials

Sodium deoxycholate was purchased from Beijing Aoboxing Biotechnology Co., Ltd. (Beijing, China). Ammonia was purchased from Tianjin Damao Chemical Reagent Factory (Tianjin, China). Teprenone was purchased from Self-Defense Pharmaceutical Co., Ltd. (Nanjing, Jiangsu, China). Saline was purchased from North China Pharmaceutical Co., Ltd. (Shijiazhuang, Hebei). Neutral formalin fixative (10%) was purchased from Beijing Solebold Technology Co., Ltd. The pepsin assay kit was purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, Jiangsu, China). Mass spectrometry grade -methanol, formic acid, and acetonitrile were purchased from Thermo Fisher Co., Ltd. (Waltham, MA, USA). All the chemical reagents applied in this experiment were of analytical rank.

GZ, BS, DZ, and GC were purchased from Tongrentang Pharmacy (Taiyuan, Shanxi, China). YT was purchased from Lingui Tianxiang Food Raw Material Co., Ltd. (Guilin, Guangxi, China). SJ was purchased from Xuxi farmers ‘ market (Taiyuan, Shanxi, China). All the above medicinal materials were identified by Professor Xuemei Qin of the Modern Research Center of TCM of Shanxi University, and all the herbs were in accordance with the “Pharmacopoeia of the People ‘s Republic of China 2020” [5]. The samples were preserved in the cold storage of the Center for Modern Research of TCM, Shanxi University.

2.2. Preparation of XJZ and its subtracted formulas

Previous studies were screened for the most effective dose of XJZ [17]: GZ (18 g), GC (12 g), DZ (30 g), BS (36 g), SJ (18 g), and YT (60 g). First, 1740 mL (10 times the quality of medicinal material) of water was added to the medicinal material, soaked for 1 hr, and then heated and refluxed for 1 hr. The filtrate was collected and the above extration was repeated again. The two filtrates were combined, and the combination was initially concentrated under reduced pressure, and then YT was added to dissolve. XJZ-P, XJZ-A, and XJZ-C were prepared by subtracting the “principal drug (YT)”, “assistant drug (GZ and BS),” and “complement drug (DZ, SJ and GC)” from XJZ, respectively. The composition of XJZ and its subtracted formulas have been shown in Table S1. Additionally, the chemical analysis was performed by Ultra-High Performance Liquid Chromatography Quadrupole/Time-of-Flight Mass Spectrometry (UHPLC-Q/TOF-MS), where a total of 785 components were characterized in XJZ including 385 compounds from GC, 99 compounds from GZ, 89 compounds from DZ, 151 compounds from BS, and 64 compounds from SJ [26]. The typical chromatograms of XJZ have been showed in Figure S1.

2.3. Animal experiment

2.5. Metabolomics study of gastric tissue

2.6. Metabolomics study of mitochondrial from gastric tissue

2.7. Metabolite traceability analysis

Deep MetOrigin Analysis was performed at Metorigin (https://metorigin.met-bioinformatics.cn) with species set as Rat.

2.8. Construction of protein-protein interaction (PPI) network

Enzymes in co-metabolic pathways were collected in Metorigin. The enzyme information was imported into String (https://cn.string-db.org/), the species was set to Rat, the minimum required interaction score was set to medium confidence (0.400), and the hide disconnected nodes in the network. Import PPI network information into Cytoscape 3.10.1 for plotting.

2.9. Molecular docking

3D chemical structures of microbial metabolites were available for download in HMDB. The structure of the enzyme was downloaded from the PDB (http://www.rcsb.org/pdb/home/home.do) with the following information: purine-nucleoside phosphorylase (Pnp, PDB ID: 7zsl), glutathione reductase (Gsr, PDB ID: 3dk9), isocitrate dehydrogenase (Idh1, PDB ID: 4ja8), xanthine dehydrogenase/oxidase (Xdh, PDB ID: 2e1q), nicotinamide adenine dinucleotide kinase (Nadk, PDB ID: 7r4m), glutathione peroxidase (Gpx7, PDB ID: 3kij). Auto Blind Docking was performed at CB-Dock2 (https://cadd.labshare.cn/cb-dock2/php/index.php) and the docking score was recorded [28]. Docking scores less than -5 were considered to be combined pairs [29].

3.1. Pharmacodynamic evaluation

There was no obvious abnormality in the appearance of rats in the control group. The rats in the CAG group were apathetic, with yellow and dull hair, and thin body. The appearance of CAG could be improved by each experimental drug. The weight of rats in the model group was decreased significantly, and the other groups showed different degrees of improvement. The degree of improvement in each group was as follows in descending order: XJZ>XJZ-C>XJZ-P=Pos>XJZ-A (Figure 1a).

Close

Figure 1.

Pharmacodynamic results. (a) Line graph of body weight change of rats. (b) Histogram of food intake of rats. (c) Histogram of pH of gastric juice of rats. (d) Histogram of PA. (e) Histopathological sections of stomach in rats stained with hematoxylin-eosin at 200X magnification. (Compared with model group, * p < 0.05, ** p < 0.01, *** p < 0.001, n=6).

Export to PPT

Food intake of the model group was significantly lower than that of the control group. Food intake could be significantly increased by each experimental drug (Figure 1b).

The pH of the gastric juice was significantly higher in the model group compared to the control group. The pH of gastric juice could be significantly reduced by each experimental drug, with XJZ being the most effective (Figure 1c).

The PA was significantly reduced in the model group as compared with the control group. The PA improved with each experimental drug, with XJZ group being the best. The rest of the groups in descending order of action were as follows: XJZ-A, XJZ-P, XJZ-C and Pos (Figure 1d).

There was no abnormality in the gastric mucosa of rats in the control group. In the model group, the chromatin staining was light, the cell morphology and arrangement were irregular, the mucosal muscularis layer was atrophied, the submucosal layer was loose, and the mucosa was thinned. The pathological changes of the CAG could be improved with each experimental drug, among which the effect of XJZ was the most effective. The rest of the groups in descending order of action were as follows: XJZ-A, XJZ-C, Pos, and XJZ-P (Figure 1e).

3.2. XJZ and its subtracted formulas altered the microbiota of gastric mucosal

16S rRNA sequencing was performed to evaluate the effects of XJZ and its subtracted formulas on the gastric mucosal microbiota. In the Figure S2, the number of taxa has been shown in Figure S2(a). Venn diagram of ASV/OUT has been shown in Figure S2(b). The richness and diversity of the model rats were significantly decreased compared with control group (Figure 2a). While the alpha diversity could be modulated by XJZ, XJZ-A, and XJZ-C, but there was no significant modulation by XJZ-P. In the Beta diversity results, the control group was clearly separated from the model group; the colony structure was significantly regressed by XJZ, XJZ-A, and XJZ-C, and there was no significant regression by XJZ-P (Figure 2b). Heat maps of the phylum and genus level have been shown in Figure S2(c-d). The differential microbes in the model group compared to the control group included Actinobacteria, Fusobacteria, Tenericutes, Firmicutes, and Verrucomicrobia (Figure S2e). Actinobacteria, Verrucomicrobia, Fusobacteria, and Firmicutes could be significantly regulated by XJZ. The differential microbes in the model group compared to the control group included Cetobacterium, Corynebacterium, Jeotgalicoccus, Lactobacillus, Methylobacterium, Olsenella, Petrimonas, Rothia, and Streptococcus. Olsenella, Petrimonas, Rothia, and Streptococcus (Figure S2f). Corynebacterium, Cetobacterium, Jeotgalicoccus, Lactobacillus, Methylobacterium, Petrimonas, and Streptococcus could be significantly regulated by XJZ. The number of phylums regulated by XJZ-P, XJZ-A, and XJZ-C were 0, 1, and 3, respectively. The number of genera regulated by XJZ-P, XJZ-A, and XJZ-C were 3, 4, and 6, respectively.

Close

Figure 2.

Effect of XJZ and its subtracted formulas on gastric mucosal microbiota. (a) Alpha Diversity. (b) Beta diversity. (c) Analysis of microbial metabolic pathways in the gastric mucosa, metabolic pathways with log_0.05 P values greater than 1 were considered statistically significant (A: Sulfur metabolism; B: Glycine, serine and threonine metabolism; C: Valine, leucine, and isoleucine biosynthesis; D: Secondary bile acid biosynthesis; E: Primary bile acid biosynthesis; F: Vitamin B6 metabolism; G: Nucleotide excision repair; H: Glycerophospholipid metabolism; J: Insulin signaling pathway; K: Pantothenate and CoA biosynthesis; L: Galactose metabolism; M: Pyrimidine metabolism; N: Glyoxylate and dicarboxylate metabolism; O: Peptidoglycan biosynthesis; P: D-Alanine metabolism) (Compared with model group, * p < 0.05, ** p < 0.01, *** p < 0.001, n=6).

Export to PPT

The 16S rRNA gene sequences were used to predict metabolic pathways in the KEGG database, and a total of 15 differential metabolic pathways were predicted (Figure 2c). Each subtracted formula regulated the differential pathways weaker than XJZ. XJZ regulated all 15 differential metabolic pathways. No differential metabolic pathways were regulated by XJZ-P. After subtracting the “principal drug” from XJZ, microbial metabolic pathways were no longer regulated by XJZ-P. Nine of the 15 differential metabolic pathways were regulated by XJZ-A. After subtracting the “assistant drug” from XJZ, K, L, M, N, O, and P were no longer regulated by XJZ-A. Twelve of the 15 differential metabolic pathways could be regulated by XJZ-C. After subtracting the “complement drug” from XJZ, M, N, and O were no longer regulated by XJZ-C.

For CAG-induced gastric mucosal microbial disorders, the regulatory effects from high to low were XJZ, XJZ-C, XJZ-A, and XJZ-P.

3.3. Metabolomics analysis of rat gastric tissue

Metabolomic analysis of gastric tissues based on Waters Primer Synapt XS IMS-TOF technology was performed to investigate the effects of XJZ and its subtracted formulas on the metabolism of gastric tissues in CAG rats. LC-MS chromatograms of each group of gastric tissues in positive mode have been shown in Figure S3(a). The QC samples were all within 2-fold standard deviation (± 2SD) (Figure S3b). In the PLS-DA score plot, the control group was completely separated from the model group suggesting that CAG caused metabolic disturbances in gastric tissues (Figure 3a). In the permutation test results of PLS-DA, Q2 was smaller than R2 and the value of the intersection of Q2 with the Y-axis was negative, proving that the model was applicable (Figure 3b).

Close

Figure 3.

Metabolomics analysis of gastric tissue. (a) PLS-DA of gastric tissue metabolomics. (b) Permutation test result of PLS-DA. (c) OPLS-DA of gastric tissue metabolomics (d) Permutation test result of OPLS-DA.

Export to PPT

In the OPLS-DA score plot, the control group was completely separated from the model group (Figure 3c). The permutation test of OPLS-DA proved that the model was applicable (Figure 3d). Seventeen differential metabolites were identified, including cellobiose, chenodeoxycholic acid 3-sulfate, D-alanine, D-alanyl-d-alanine, glycoursodeoxycholic acid 3-sulfate, diaminopimelic acid, D-glutamic acid, deoxyglycocholate, phenylpyruvic acid, urobilinogen, L-homoserine, N2-succinyl-ornithine, hydroxypropionic acid, indoleacetic acid, taurodeoxycholic acid, N-methylnicotinamide, and glycocholic acid (Table S2). Nine compounds were significantly decreased and 8 compounds were significantly increased in the model group compared to the control group (Figure S3c). Seventeen differential metabolites were modulated by XJZ; glycoursodeoxycholic acid 3-sulfate and N-methylnicotinamide were modulated by all treatment groups with the same trend of modulation. The number of differential metabolites regulated by XJZ-P, XJZ-A, and XJZ-C were 3, 7, and 6, respectively. The results showed that the effects on the metabolic disorders of gastric tissue were XJZ, XJZ-A, XJZ-C, and XJZ-P in descending order of regulation.

3.4. Differential metabolite traceability and the STA-Sankey network

Six metabolites were host-associated and all six were “host-microbe” co-metabolites; 17 metabolites were microbe-associated, 11 of which were microbe-specific (Figure 4a and 4b, Table S3). 8 and 10 metabolite pathways belonged to microbiota and host-microbiota, respectively, of which 4 and 2 metabolic pathways were significantly (p < 0.05) associated with CAG, respectively (Figure 4c).

Close

Figure 4.

MetOrigin analysis of differential metabolites in gastric tissues. (a) Venn diagram for traceability analysis. (b) Histogram of traceability analysis. (c) Functional analysis of differential metabolite traceability, metabolic pathways with log_0.05 P values greater than 1 were considered statistically significant (1: D-Amino acid metabolism. 2: Peptidoglycan biosynthesis. 3: Lysine biosynthesis. 4: Beta−Alanine metabolism. 5: Sulfur metabolism. 6: Phenylalanine, tyrosine and tryptophan biosynthesis. 7: Propanoate metabolism. 8: Phenylalanine metabolism. 9: Carbon fixation pathways in prokaryotes. 10: Glycine, serine and threonine metabolism. 11: Primary bile acid biosynthesis. 12: Tryptophan metabolism. 13: Cysteine and methionine metabolism. 14: Teichoic acid biosynthesis. 15: Arginine and proline metabolism. 16: Pyrimidine metabolism. 17: Porphyrin metabolism). (d) Heatmap of microbial-metabolite correlation.

Export to PPT

Significant metabolic pathways, including D-amino acid metabolism (involved metabolites: phenylpyruvicacid and D-glutamicacid) and Beta-alanine metabolism (involved metabolites: hydroxypropionicacid) were co-metabolised; D-amino acid metabolism (involved metabolites: D-alanyl-D-alanine, diaminopimelic acid, D-alanine), peptidoglycan biosynthesis (involved metabolites: D-alanyl-D-alanine; D-alanine), lysine biosynthesis (involved metabolites: diaminopimelic acid, L-homoserine), and sulfur metabolism (involved metabolite: L-homoserine) were microbial metabolism. Of interest, D-amino acid metabolism was the most prominent metabolic pathway in both co-metabolism and microbial metabolism, with a total of five involved metabolites, and was the most relevant pathway for CAG pathogenesis.

Cetobacterium was negatively correlated with indoleacetic acid and positively correlated with D-glutamic acid; Streptococcus was negatively correlated with D-alanine, N-methylnicotinamide, glycoursodeoxycholicacid 3-sulfate, and positively correlated with chenodeoxycholic acid 3-sulfate, taurodeoxycholic acid; Olsenella was negatively correlated with glycoursodeoxycholicacid 3-sulfate, N2-succinyl-ornithine and positively correlated with chenodeoxycholic acid 3-sulfate, taurodeoxycholic acid; Lactobacillus was positively correlated with hydroxypropionic acid and N-methylnicotinamide and negatively correlated with cellobiose and D-glutamic acid (Figure 4d). STA-Sankey network diagrams of D-Amino acid metabolism, beta-Alanine metabolism, Peptidoglycan biosynthesis, Lysine biosynthesis and Sulfur metabolism have been shown in Figure S4.

Based on the regulation of metabolites by each experimental drug, it was clear that all five metabolic pathways could be regulated by XJZ. 2 metabolites could be regulated by XJZ-P. D-amino acid metabolism, Beta−alanine metabolism, and peptidoglycan biosynthesis were not regulated by XJZ-P after subtracting “principal drug” from XJZ, suggesting that the “principal drug” plays a major role in the regulation of the above 3 metabolic pathways by XJZ. Three metabolic pathways could be regulated by XJZ-A. Lysine biosynthesis and sulfur metabolism were not regulated by XJZ-A after XJZ subtracted the “assistant drug”, suggesting that the “assistant drug” plays a major role in the regulation of the above 2 metabolic pathways by XJZ. Two metabolic pathways could be regulated by XJZ-C. Peptidoglycan biosynthesis, lysine biosynthesis, and sulfur metabolism were not regulated by XJZ-C after subtracting the “complement drug” from XJZ, suggesting that the “complement drug” played a major role in the regulation of the above 3 metabolic pathways by XJZ.

3.5. Metabolomics analysis of mitochondria

LC-MS chromatograms of each group of mitochondrial in positive mode have been shown in Figure S5(a). The QC samples were all within 2-fold standard deviation (± 2SD) (Figure S5b). The PLS-DA results for all the experimental groups have been shown in Figure 5(a), where the control and model groups were significantly separated. Permutation test has been shown in Figure 5(b), Q2 was less than R2 and Q2 intersects the Y-axis negatively, proving that the model was applicable.

Close

Figure 5.

Analysis of mitochondrial metabolomics. (a) PLS-DA of mitochondrial metabolomics. (b) Permutation test result of PLS-DA. (c) OPLS-DA of mitochondrial metabolomics. (d) Permutation test result of OPLS-DA. (e) Results of metabolic pathway enrichment analysis, metabolic pathways with log_0.05 P values greater than 1 were considered statistically significant, (A: Purine metabolism; B: Glutathione metabolism; C: Biosynthesis of various other secondary metabolites; D: Nicotinate and nicotinamide metabolism; E: Aminoacyl-tRNA biosynthesis; F: Ether lipid metabolism; G: Biotin metabolism; H: Beta-Alanine metabolism; I: Pantothenate and CoA biosynthesis; J: Phenylalanine, tyrosine and tryptophan biosynthesis; K: Lysine biosynthesis; L: Phenylalanine metabolism; M: Glycerophospholipid metabolism; N: Lysine degradation; O: D-Amino acid metabolism; P: Pyrimidine metabolism).

Export to PPT

In the OPLS-DA score plot, the control group was completely separated from the model group (Figure 5c). The permutation test of OPLS-DA proved that the model was applicable (Figure 5d). A total of 18 differential metabolites were identified, including glycerophosphocholine, deoxyguanosine, thymidine, L-carnitine, carnosine, guanine, pantothenic acid, glycylleucine, oxidized glutathione (GSSG), 1-methylnicotinamide, hypoxanthine, nicotinamide adenine dinucleotide phosphate (NADP), coenzyme Q10, lysine, phenylalanine, guanosine, adenosine, and Gamma-linolenyl carnitine (Table S4). Six compounds were decreased and 12 compounds were increased in the model group. The number of metabolites regulated by XJZ, XJZ-P, XJZ-A, and XJZ-C were 15, 7, 10, and 13, respectively (Figure S5c). For CAG-induced metabolic disorders, the modulatory effects were XJZ, XJZ-C, XJZ-A, and XJZ-P in descending order.

A total of 16 metabolic pathways were enriched, of which 5 metabolic pathways were significantly correlated with CAG (Figure 5e). Metabolic pathways included purine metabolism (involved metabolites: guanine, hypoxanthine, guanosine, adenosine and deoxyguanosine), glutathione metabolism (involved metabolites: GSSG and NADP), biosynthesis of various other secondary metabolites (involved metabolites: lysine and phenylalanine), nicotinate and nicotinamide metabolism (involved metabolites: 1-methylnicotinamide and NADP), and aminoacyl-tRNA biosynthesis (involved metabolites: lysine and phenylalanine). The most significant pathway was purine metabolism, suggesting that CAG was most affected by this pathway. All 5 pathways were regulated by XJZ, XJZ-A, and XJZ-C. Three metabolic pathways were regulated by XJZ-P, including nicotinate and nicotinamide metabolism, purine metabolism, and glutathione metabolism. Therefore, the mitochondrial metabolic pathway was most strongly regulated by “principal drug”.

3.6. “Microbiota-mitochondria” cross-talk

By combining 16S rRNA and metabolomics, changes in gastric mucosal microbial and gastric tissue metabolites were found, with all gastric tissue metabolites derived from microbes, as well as changes in mitochondrial metabolites. Many studies had found that microbes might influence mitochondrial function through metabolites [30,31]. Gastric tissue metabolites-mitochondrial metabolites correlation analysis revealed significant correlations between multiple microbial metabolites and multiple mitochondrial metabolites, suggesting that microbial metabolites could potentially influence mitochondrial metabolites (Figure 6a). We assumed that microbial metabolites could enter mitochondria to affect mitochondrial enzymes and mitochondrial function, and therefore collected enzyme proteins in the mitochondrial pathway for PPI and molecular docking analysis.

Close

Figure 6.

”Microbiota-mitochondria” cross-talk. (a) Correlation heat map of gastric tissue metabolites-mitochondrial metabolites. (b) PPI network diagram of mitochondria, the larger the degree, the darker the color. (c) Heat map of molecular docking scores.

Export to PPT

The gene information of enzymes involved into mitochondrial metabolic pathway has been shown in Table S5. There were 26 nodes and 76 edges in the PPI network (Figure 6b). The top 6 of degree in the PPI network were selected as the key genes (degree ≥ 8) of “microbial-mitochondrial” cross-talk, including Gsr, Pnp, Idh1, Gpx7, Nadk, Xdh.

The molecular docking of microbial metabolites with key targets in mitochondria was shown in Figure 6(c), there were a total of 102 dockings, with 83 docking scores <-5 and 30 docking scores < -7.5. Mitochondrial targets were potential targets for microbial metabolites such as chenodeoxycholicacid3-sulfate, urobilinogen, and glycoursodeoxycholic acid 3-sulfate.

3.7. Regulation of CAG by components of the XJZ

The regulation of CAG by the subtracted formula was reduced after subtracting a certain part of the drug in XJZ. When a part of the drug was subtracted, the loss of a regulating effect could be regarded as this part of the drug having this regulating effect. The “principal drug,” “assistant drug,” and “complement drug” had different roles in the treatment of CAG by XJZ, but also co-regulate in many ways. For gastric microbiota, the number of genera individually regulated by “principal drug,” “assistant drug,” and “complement drug” were 2, 1, and 0, respectively. For gastric tissue metabolites, the number of gastric tissue metabolites individually regulated by “principal drug,” “assistant drug,” and “complement drug” were 4, 0, and 1, respectively. Eight gastric tissue metabolites were regulated by the “principal drug,” the “complement drug,” and the “assistant drug” together. For mitochondrial metabolites, the number of mitochondrial metabolites individually regulated by “principal drug,” “assistant drug,” and “complement drug” were 4, 1, and 0, respectively. Three mitochondrial metabolites were regulated by the “principal drug,” the “complement drug,” and the “assistant drug” together.

“Principal drug” could regulate Corynebacterium, Jeotgalicoccus, Lactobacillus, Streptococcus, and affect 14 microbial metabolites, such as deoxyglycocholate, phenylpyruvic acid, cellobiose, and taurodeoxycholic acid. These microbial metabolites could potentially affect Pnp, Xdh, Nadk, and influence purine metabolism and nicotinate and nicotinamide metabolism. It was worth noting that nicotinate and nicotinamide metabolism was only regulated by the “principal drug”.

“Assistant drug” could regulate Petrimonas, Streptococcus, Methylobacterium, Lactobacillus, and affect 10 microbial metabolites, such as deoxyglycocholate, cellobiose, and indoleacetic acid. These microbial metabolites could potentially affect Pnp, Xdh, Gsr, Idh1, Gpx7, and influence purine metabolism and glutathione metabolism.

“Complement drug” could regulate Streptococcus, Methylobacterium, and affect 11 microbial metabolites, such as cellobiose, chenodeoxycholic acid 3-sulfate, and N-methylnicotinamide. These microbial metabolites could potentially affect Gsr, Idh1, Gpx7, and influence glutathione metabolism.

Therefore the “principal drug” had the greatest role in the treatment of CAG with XJZ, the “assistant drug” and the “complement drug” alone had a lesser role, and both mainly played a role together with the “principal drug” (Figure 7).

Close

Figure 7.

(a) Network of regulation by XJZ, “principal drug”, “complement drug” and “assistant drug” (Red V: Drugs; Triangles: gastric mucosa microbes; Green hexagon: gastric tissue metabolites; Rhombus: mitochondrial metabolites; Purple rectangle: mitochondrial metabolite pathway; Green circle: potential target.). (b) Upset plots of genera-level microbes. (c) Upset plots of gastric tissue metabolites. (d) Upset plots of mitochondrial metabolites.

Export to PPT

A multi-factorial approach was used in the study to establish a rat model of CAG, which were induced by combination administrations of sodium deoxycholate and ammonia coupled with starvation and fatigue. Deoxycholic acid is one of the most toxic components of reflux bile. 0.1% ammonia could mimic the damage to the gastric mucosa caused by high ammonia in the presence of Helicobacter pylori infection [32]. Several pharmacodynamic indicators showed that the rats in the model group had the disease characteristics of CAG, indicating that the model construction was successful.

Destruction of the gastric mucosa led to the death of acid-secreting wall cells and a decrease in gastric acid secretion, which could lead to overgrowth of certain bacteria [33], which was in accord with our work. The different microbes induced by CAG mainly involved Actinobacteria, Fusobacteria, Tenericutes, Firmicutes, and Verrucomicrobia at the phylum level, which were usually harmless to humans. Actinobacteria were widely distributed, produced abundant secondary metabolites, and played an important role in the energy cycle of ecosystems [34,35]. As an important microbial resource, Actinobacteria had important applications in the fields of medicine and biotechnology. Actinobacteria produced metabolites, such as alkaloids, terpenoids, esters, peptides, etc. [36,37], which were known to be anti-carcinogenic, anti-tumor, and anti-bacterial [38,39]. In the present study, reduced abundance of Actinobacteria in CAG rats reduced their effective metabolites, leading to exacerbation of CAG. Verrucomicrobia could utilize mucosal mucus as a source of nutrients [40], and its abundance was reduced in CAG rats due to the destruction of gastric mucosa by ammonia and deoxycholic acid. Systematic studies on Tenericutes were lacking, but their effects on CAG could not be ruled out.

The different microbes at the genera level in the control group compared to the model group included Cetobacterium, Corynebacterium, Jeotgalicoccus, Lactobacillus, Methylobacterium, Olsenella, Petrimonas, Rothia, and Streptococcus. Cetobacterium was a probiotic that produces vitamins, acetic acid, and butyric acid. CAG led to a decrease in Cetobacterium, which produced less acetic acid, leading to a block in participation in the tricarboxylic acid cycle [41]. The majority of Corynebacterium were pathogenic [42] and susceptible to infection in immunocompromised organisms, and their presence was increased in CAG. Lactobacillus was increased in CAG rats and it was a bacterium that produces bile acid hydrolase (BSH). Bound bile acids were increased in CAG rats [43] and BSH-producing bacteria were tolerant to bound bile acids [44]. Methylobacterium was pathogenic bacteria which was positively associated with liver disease [45] and could also cause kidney infections [46]. Olsenella was a probiotic reduced in CAG rats, and its elevation alleviated high-fat diet-induced obesity[47], hypercholesterolemia [48]. Streptococcus was a pathogenic bacterium whose cell wall complexes, secondary metabolites (hemolysin, autolysins, etc.) are harmful [49]. Rothia, Jeotgalicoccus and Petrimonas remain to be studied further.

To further explore the changes in gastric tissue metabolites, gastric tissue metabolomics analysis was performed. The results showed that CAG induced the metabolite disorders in gastric tissues, as well as the gastric mucosal flora. A total of 17 differential metabolites were identified, which were associated with the host and gastric mucosal flora.

Differential metabolite pathway analysis identified 5 metabolic pathways that were significantly associated with CAG and all of these pathways might be associated with microbes. D-amino acids regulated glucose metabolism by regulating insulin secretion [50,51]. In the present study, it was D-glutamate that was involved in the metabolism of D-amino acids, which for some time was thought to be metabolized by D-aspartate oxidase. However, a recent study showed that D-glutamate was also converted to 5-OXO-D-proline in mitochondria via a new D-glutamate cyclase [52]. 5-OXO-D-proline was associated with glutathione (GSH) in mitochondria [20], which was involved in mitochondrial redox reactions as a reducing substance in mitochondria. Peptidoglycan was an essential component of the cell wall of bacteria and was necessary for the maintenance of cell morphology, size, osmotic pressure stability, and cell survival [53,54]. In the host, peptidoglycan inhibited hexokinase (HK), which dissociated HK from mitochondria and therefore activated NLR family pyrin domain containing 3 (NLRP3) inflammasomes. NLRP3 inflammasomes regulated a number of mitochondrial functions including mitochondrial motility [55], mitochondrial division and growth [56], and mitochondrial autophagy [57]. Lysine biosynthesis could regulate the synthesis of lysine, which was important for regulating metabolic homeostasis in the body, enhancing the absorption of other proteins, promoting growth and development, and improving human dietary nutrition and animal nutrition [58]. Beta-alanine metabolism was a way of regulating Beta-alanine content. Beta-alanine was not involved in protein synthesis and was the only Beta-amino acid present in nature, involved in the composition of pantothenic acid, coenzyme A. In mice, Beta-alanine increased superoxide dismutase (SOD) and glutathione peroxidase (GSH-Px) activity in muscle, decreases oxidation products such as malondialdehyde (MDA), and improved muscular endurance [59]. Sulfur metabolism in microbes was associated with microbial resistance [60]. Sulfur existed in bacteria mainly in inorganic or organic form, inorganic substances included sulfate, thiosulfate and hydrogen sulfide (H₂S), etc.; organic sulfur substances included various amino acids. H₂S produced by bacteria was the substance that most directly affects the host, and H₂S could stimulate the expression of complex IV (cytochrome c oxidase, CcO) in the mitochondria of trophoblast progenitor cells [61], and increased ATP production [62]. The above results suggested that gastric tissue metabolites were altered. Metabolites and metabolic pathways were associated with the gastric mucosal microbiota and potentially linked to mitochondrial metabolic alterations.

To investigate the changes in mitochondrial metabolites, mitochondrial metabolomics analysis was further performed, where total of 18 mitochondrial differential metabolites were identified. The differential metabolites were subjected to pathway analysis and five metabolic pathways were significantly correlated with CAG. In purine metabolism, purines were involved in the synthesis of a variety of biomolecules in the organism, such as nucleic acid synthesis, cell signals, cofactors, and co-transporters [63]. Purine metabolism was involved in the NLRP3 inflammatory pathway and the Cyclic adenosine monophosphate-protein kinase A (cAMP-PKA) pathway [64], which might account for its influence on CAG. Glutathione metabolism mainly involved the production of dihydronicotinamide-adenine dinucleotide phosphate (NADPH), glutathione, and GSSG, and was closely related to mitochondrial inflammation, oxidative stress, and apoptosis [65]. In biosynthesis of various other secondary metabolites, phenylalanine generates Cyclo (L-Trp-L-Phe). Cyclo (L-Trp-L-Phe) showed antimicrobial, antitumor, antiviral, and antioxidant effects, but its role in CAG is unknown. In this experiment, nicotinate and nicotinamide metabolism involved the synthesis of NADP and nicotinamide. NADP produced nicotinamide under the catalysis of ADP-ribosyl cyclase or cyclic ADP-ribose hydrol, and nicotinamide produced 1-methylnicotinamide under the catalysis of nicotinamide N-methyltra ferase catalyzed to produce 1-methylnicotinamide. The various types of niacinamide were collectively known as niacin, which is a vitamin. In CAG rats, the levels of 1-methylnicotinamide and NADP were significantly reduced, reflecting the fact that the overall level of niacin in their organisms was significantly decreased compared to normal rats. Lack of nicotinate could lead to a variety of diseases, such as enteritis, dermatitis, neuroinflammation [66]. Lack of nicotinate could be an important cause of CAG. In this study aminoacyl-tRNA biosynthesis involved the synthesis of L-phenylalanyl-tRNA and L-LySyI-tRNA, and the key to the pathway was aminoacyl-tRNA synthetase, which catalyzed the synthesis of aminoacyl-tRNA from amino acids and their corresponding tRNAs [67]. In CAG, mitochondrial phenylalanine and lysine were elevated, which suggested that their corresponding aminoacyl-tRNAs were blocked, which might lead to cessation of biological responses and mitochondrial dysfunction.

It had been shown that there existed a cross-talk between microbiota and mitochondria [68]. Mitochondria are essential organelles for ATP production and the main producers of reactive oxygen species. Bajpai et al. found that there was an overlap between abnormalities in intestinal flora and mitochondrial dysfunction, and that the intestinal flora might influence health and disease by regulating mitochondrial function [69]. Thiele et al. screened 2626 metabolites from 7439 reactions and found that 437 were from mitochondria, of which 325 overlapped with gut flora metabolites [31]. Microbiota can directly or indirectly affect mitochondria. Direct effects were manifested by microbial metabolites, such as SCFAs, H₂S, NO, and secondary bile acids that could modulate mitochondrial function and affect host metabolism and health. Escherichia coli, Streptomyces and Pseudomonas aeruginosa in the intestine secreted xanthine oxidase, which converts purines in the intestine to uric acid [70]. Uric acid was reducibility and had the function of scavenging many reactive oxygen species such as oxygen radicals and peroxynitrite produced by mitochondria [71]. CAG caused thinning of gastric mucosa, reduction of barrier function, and entry of a large amount of lipopolysaccharide (LPS) produced by bacteria into gastric tissue cells. LPS could induce NF-κB-mediated inflammation [72] and increase intracellular ROS and activate the ROS/p38/MAPK signaling pathway, which affected apoptosis [73]. SCFAs produced by microbiota affect mitochondrial function, for example, the lack of butyrate led to a significant reduction in nicotinamide adenine dinucleotide (NAD), resulting in autophagy in germ-free mouse colon cells [30]. It had also been shown that microbes regulated mitochondrial function by indirectly inducing genes of SIRT1, FIAF, and FXR [74].

To further explore the effects of microbial metabolites on mitochondrial enzymes, molecular docking was used. Gsr was an oxidoreductase enzyme that was an important component of the human redox system. With NADPH as a cofactor, Gsr catalyzed the formation of reduced GSH from GSSG, which is beneficial to the body’s antioxidant activity [75]. Pnp was a key enzyme in the purine remediation pathway, catalyzing the phosphorylation of nucleosides, such as inosine and guanosine and participating in the degradation of ATP [76]. Idh1 was located in the mitochondrial matrix and converted isocitrate to α-ketoglutarate (α-KG) and NADP to NADPH. α-KG was a cofactor for many dehydrogenases and was involved in the regulation of various key biological processes, including nucleic acid repair, hypoxia response and chromatin modification and fatty acid metabolism. α-KG could regulate various key biological processes, including nucleic acid repair, hypoxia response, chromatin modification and fatty acid metabolism. During the conversion of isocitrate to α-KG, Idh1 catalyzed the generation of NADPH, and NADPH and reduced nicotinamide adenine dinucleotide (NADH) were reductants for lipid biosynthesis and antioxidants that protect cells from oxidative stress and radiation damage [77]. Gpx was an important oxidoreductase containing eight family members that played important roles in scavenging peroxides and maintaining the functional integration of cell membranes [78]. High expression of Gpx7 could alleviate cellular damage caused by ROS, and Gpx7 also reduced IL-1β-induced cellular inflammation, apoptosis, and extracellular matrix degradation through the regulation of iron death [79]. Nadk was the rate-limiting enzyme that phosphorylated NAD to NADP and is highly conserved throughout biological evolution. NADP was important in cellular redox homeostasis [80]. The main role of Xdh was to act as a catalyst for the conversion of hypoxanthine to xanthine, and ultimately to uric acid [81]. Multiple microbial metabolites could dock with the six enzyme proteins mentioned above, suggesting that there was a potential cross-talk between microbiota and mitochondria.

XJZ enhanced the antioxidant and anti-inflammatory capacity of piglet liver through Nrf2/NF-κB pathway [82]. XJZ reduced mucosal damage, increased ZO-1 and occludin expression, decreased epithelial cell apoptosis, decreased the expression of autophagy-related proteins LC3 and beclin1, increased the expression of p62, activated the PI3K/AKT/mTOR/ULK1 (ser757) signaling pathway, inhibited AMPK/ULK1 (ser317) signaling pathway, antagonized the imbalance of redox homeostasis, and protected the gastric mucosa [83]. In this study, it was found that the effect of XJZ against CAG was the result of the participation of all parts. The “principal drug” had the strongest effect, and the “assistant drug” and the “complement drug” were weaker and more auxiliary to the “principal drug” compared to compared to the “principal drug”. Although the effect of the “principal drug” was the strongest, the effect of XJZ was reduced without the assistance of the “assistant drug” and the “complement drug”. The treatment of TCM compound was a holistic action of all the parts, with possible synergies between the parts [84].

Additionally, in other studies of CAG, metabolomics analysis had been conducted using samples, such as serum and urine, and many CAG biomarkers had been identified. Interestingly and similarly, differences in energy metabolism, such as glycerophospholipid metabolism, citrate cycle, and alanine, aspartate, and glutamate metabolism, were also observed in serum and urine between normal and CAG rats in these studies [85,86]. Serum samples provided a more practical and simpler choice for disease monitoring, which was very important in clinical practice. Compared with serum or urine samples, the gastric tissue can offer a unique perspective on localized metabolic information. The metabolic profiles of gastric tissue could offer a direct description of its pathological process during the onset and progression of CAG, and facilitate the understanding of the underlying mechanism of CAG. Our current research focused on gastric tissue to gain a deeper understanding of the metabolic links between the gastric tissues and microbes, as well as the cross-talk between gastric mucosal microbes and mitochondria. These findings might provide new ideas for the mechanism of XJZ and its subtracted formulas in the treatment of CAG. The connection between gastric tissue and serum would be performed to provide a more comprehensive and deeper lights into the efficacy of XJZ against CAG, as well as its compatibility.

In future experiments, molecular biology experiments will be performed to validate the results of molecular docking and Metorigin. The mechanism of cross-talk between mitochondria and microbes needs to be further investigated. In addition, we need to go deeper into the function and efficacy of the compounds in XJZ.