A total of 48 gastric mucosa biopsy tissues were collected from the First Affiliated Hospital of Sun Yat-sen University from June 1 to July 1, 2021. Of these, 34 tissue samples were collected from patients with superficial gastritis (SG), 9 from those with erosive gastritis (EG), 3 from those with atrophic gastritis (AG), and 2 from those with intestinal metaplasia (IM). All samples were obtained from the antrum through endoscopic examination. The inclusion criteria for participants in this study were as follows: over 18 years old and under 65 years old; no history of malignant diseases, cardiopulmonary diseases, metabolic diseases, or autoimmune diseases; no history of gastrectomy or reflux esophagitis; no history of H. pylori eradiation, antibiotics, proton pump inhibitor, or H₂ receptor antagonist treatment within the past 3 months. Written informed consent was obtained from all patients, and this study was conducted according to the guidelines of the Declaration of Helsinki and approved by the ethics committee of the First Affiliated Hospital of Sun Yat-sen University (project number 2020-164).

The mucosal samples from each patient were divided into two pieces aseptically; one was stored in the Sample Protector (TAKARA, Beijing, China) and transferred to −80 °C until DNA extraction, while the other one was stored in the H. pylori protection fluid (Huankai Biology, Guangzhou, Guangdong, China) and transferred to the laboratory for H. pylori isolation.

H. pylori strains were isolated from the mucosal tissues according to the method as previously described (Li et al., 2022). In brief, the mucosal samples were inoculated on a H. pylori selective agar (Huankai Biology) at 37 °C in a microaerophilic incubator (Binder, Tuttlingen, Germany) containing 10% O₂, 5% CO₂, and 85% N₂ for up to 10 days. The small gray and translucent colonies were picked and purified on a new H. pylori selective agar for genomic DNA extraction. The taxonomic analysis of the H. pylori-like colonies was performed using 16S rRNA sequencing with the universal primers (27F: 5′-AGAGTTTGATCCTGGCTCAG-3′; 1492R: 5′-ACGGCTACCTTGTTACGACTT-3′) (Garrity, 2016). In all, 18 strains of H. pylori were isolated from the 48 patients in this study.

The PCR amplification of dupA of the H. pylori strains was carried out according to the method described by Xue using the dupA primer (dupA−F: 5′-GACGATTGAGCGATGGGAATAT-3′; dupA−R: 5′- CTGAGAAGCCTTATTATCTTGTTGG-3′) (Xue et al., 2021). The genomic DNA of H. pylori was extracted using the microbial DNA extraction kit (Huankai Biology). PCR amplification was carried out in a 25-μL reaction mix containing 0.2 μM forward and reverse primers, 50 ng genomic DNA template, 12.5 μL PrimeSTAR Max DNA Polymerase (TAKARA), and nuclease-free water. Samples were denatured at 98 °C for 1 min, followed by 35 cycles at 98 °C for 10 s, 55 °C for 15 s, and 72 °C for 5 s, with a final elongation at 72 °C for 30 s. The amplified products were analyzed by electrophoresis at 120 V for 25 min in a 1.5% agarose gel containing 1 × Tris-acetate-EDTA (TAE) buffer, stained with GoldView Staining Dyes (Solarbio, Beijing, China), and visualized using a gel documentation system (Bio-Rad, Hercules, CA, USA). Samples showing a monoclonal brand of 971 bp were considered as dupA⁺ H. pylori strains.

The expression of dupA was examined according to the method of Alam (Alam et al., 2012). Total RNA of H. pylori was extracted by TRIzol reagent according to the manufacturer’s protocol (TAKARA). cDNA was generated using the Master Mix cDNA Synthesis Kit (Accurate Biotechnology, Changsha, Hunan, China). Quantitative reverse transcription PCR (qRT-PCR) of dupA was performed using SYBR Green I (Accurate Biotechnology) using the primers of DupAsetIF/DupAsetIR, with rpsT being the reference (Table 1). The mean Ct of triplicate reactions was determined and the expression levels of dupA were measured using the -ΔCt method. The relative expression of dupA in H. pylori isolates was calculated as 2^-ΔΔCt, where -ΔΔCt =-ΔCt_sample - ΔCt_reference. H. pylori strain P6, which was isolated from the SG patient, was selected as the reference strain in the qRT-PCR assay.

Microbial DNA of the gastric mucosa was extracted using the QIAamp PowerFecal Pro DNA kit (Qiagen, Hilden, Germany). 16S rRNA gene was amplified using the V3–V4 hypervariable region primers 338F 5′-ACTCCTACGGGAGGCAGCAG-3′ and 806R 5′-GGACTACHVGGGTWTCTAAT-3′, with the amplicon length of 468 bp (Fadrosh et al., 2014). PCR was performed using the UCP Multiplex PCR Kit (Qiagen), which involved denaturation at 95 °C for 2 min, followed by 30 cycles at 95 °C for 5 s, 55 °C for 15 s, and 72 °C for 30 s, with a final elongation at 72 °C for 30 s. The amplicon libraries for next-generation sequencing were generated using the QIAseq Ultralow Input Library Kit (Qiagen).

The amplicon libraries were validated and pooled by the Agilent high sensitivity DNA kit (Agilent, Santa Clara, CA, USA). Paired-end sequencing was conducted on the MiSeq platform with the MiSeq Reagent Kit version V3 (Illumina, San Diego, CA, USA).

Quality filtering of the data was performed using the CLC Genomic Workbench version 20.0 (Qiagen). The trimmed sequences were matched to those in the Greengenes database (v13.5) with similarity cutoff of 97% and clustered into operational taxonomic units (OTUs).

Taxonomic analysis was performed using the Microbiome Analyst software (Chong et al., 2020). Microbial diversity was explored based on the Chao1 and Shannon indices at the genus level using analysis of variance (ANOVA). In addition, microbial composition was investigated using non-metric multidimensional scaling (NMDS) at the genus level among different groups. Linear discriminant analysis effect size (LEfSe) was used to determine the key taxonomic differences among groups, with the threshold on the logarithmic linear discriminant analysis score of 2.0 for discriminative features and alpha value of 0.05 for the pairwise Wilcoxon test (Segata et al., 2011).

Metagenome prediction of the mucosal microbiome was performed using the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) (Langille et al., 2013). The statistical differences between the groups were examined in the second and third levels of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis using ANOVA. Correlation analysis was conducted between the key genera and metabolic pathways in the gastric mucosa using Pearson’s correlation test.

Network analysis was performed to identify the co-occurring relationship using Pearson’s correlation test, and results were visualized using Cytoscape v3.9.0 (Shannon et al., 2003). The core genera in gastric microbiome were represented as nodes (red for key genera in AG/IM, blue for key genera in EG, grey for genera in all groups), relative abundance (RA) as node size, and edges depicted correlations (green for positive correlation, grey for correlation without significance).

The clinical characteristics of the study cohort are summarized in Table 2. No significant differences were found in age and gender among the SG, EG, and AG/IM patients.

In this study, 972 OTUs were observed in the gastric mucosa of 48 patients with gastritis. These gastric microbes were distributed into 11 phyla, 19 classes, 35 orders, 65 families, and 87 genera. The most commonly detected phyla in the gastric microbiome were Proteobacteria (all samples), Firmicutes (89.58% of samples), Bacteroidetes (81.25% of samples), Actinobacteria (9.17% of samples), and Fusobacteria (77.08% of samples). The most commonly detected genera in the gastric microbiome were Streptococcus (83.33% of samples), Prevotella (81.25% of samples), Neisseria (75.00% of samples), Fusobacterium (75.00% of samples), Veillonella (70.83% of samples), and Rothia (70.83% of samples).

The microbial diversity of each sample was evaluated at the genus level based on Chao1 and Shannon indices. A comparison of alpha diversity revealed that the highest richness of microbes was observed in the mucosa of SG patients, while the lowest was observed in AG/IM patients (P < 0.001 for both Chao1 and Shannon indices) (Figure 1A).

Moreover, we found that the microbial composition of the mucosa differed among SG, EG, and AG/IM patients (P = 0.016). The gastric microbial composition of AG/IM patients was different from that of SG patients (PERMANOVA, R² = 0.371, P = 0.011) and EG patients (PERMANOVA, R² = 0.140, P = 0.002) based on the Bray–Curtis index. However, no difference in the microbial composition of the mucosa was observed between the SG and EG patients (PERMANOVA, R² = 0.010, P = 0.736) (Figure 1B).

To identify the dominant microbes in different patients with gastritis, LEfSe analysis was performed. The results showed that Pasteurellaceae and Gemellaceae were the most prevalent families in SG patients, whereas Prevotellaceae, Veillonellaceae, Clostridiaceae, Actinomycetaceae, and Rubrobacteraceae were common in EG patients, and Helicobacteraceae was the most prevalent family in AG/IM patients (Figure 1C).

Overall, 5739 microbial metabolites in 276 metabolic pathways were identified by PICRUSt in the gastric microbiome in this study. Among them, a higher number of genes involved in cellular processes and fewer genes involved in microbial metabolism were detected in the gastric microbes of AG/IM patients compared to the SG and EG patients (P = 0.042 and P = 0.026, respectively). As microbial metabolites might exert carcinogenic effects on epithelial cells (Imai et al., 2021; Singhal et al., 2021), we further analyzed the changes in the 12 metabolic pathways in patients with AG/IM. Our results showed that the microbial activities of carbohydrate metabolism, enzyme families, nucleotide metabolism, biosynthesis of other secondary metabolites, and xenobiotic biodegradation and metabolism were significantly lower in the gastric microbes of AG/IM patients, whereas the activities of energy metabolism, glycan biosynthesis, and metabolism of cofactors and vitamins were higher (P < 0.05) (Figure 2A).

We further analyzed the functional changes of 20 xenobiotics in the gastric microbiome and found that the microbial activities of drug metabolism associated with cytochrome P450 and other enzymes, ethylbenzene degradation, aminobenzoate degradation, naphthalene degradation, and caprolactam degradation were significantly decreased in AG/IM patients (P < 0.05), while the microbial activities associated with polycyclic aromatic hydrocarbon degradation increased (P = 0.013). Correlation analysis revealed that microbial dysfunction was related to changes in key microbes in the gastric mucosa (Figure 2B).

Although OTUs of H. pylori were found in 23 patients based on amplicon sequencing, only 21 H. pylori strains were isolated from mucosal samples. Of these, 12 strains were isolated from patients with SG, six from those with EG, and three from those with AG/IM (Figure 3A).

PCR analysis revealed that among these H. pylori isolates, six strains were dupA⁺ and 15 were dupA⁻ . Further investigation revealed that 83.33% of the dupA⁺ H. pylori strains were isolated from the mucosal samples of the EG patients and 16.67% were isolated from the SG patients, indicating that the infection rate of the dupA⁺ H. pylori strain was higher in EG patients than in both SG and AG/IM patients (P = 0.002) (Figure 3B). qRT-PCR of the six dupA⁺ strains revealed that the relative level of dupA transcript was higher in the EG isolated strains than the SG isolated strains (P = 0.002) (Figure 3C).

DupA is an H. pylori virulence factor that increases the risk of mucosal ulcers (Queiroz et al., 2011). As all dupA⁺ H. pylori strains were found in SG and EG patients, we further investigated the influence of these strains on the gastric microbiome in these patients.

Microbial diversity analysis revealed that genus richness decreased after dupA⁺ H. pylori infection (P < 0.01 for both Chao1 and Shannon indices), and similar changes occurred in dupA⁻ H. pylori infection (P < 0.01 for both Chao1 and Shannon indices) (Figure 4A). Further investigation showed that dupA⁻ H. pylori infection induced a more severe loss of biodiversity than dupA⁺ H. pylori infection (P = 0.041 for the Chao1 index and P = 0.044 for Shannon index) (Figure 4A). Moreover, the microbial composition changed as a result of both dupA⁺ H. pylori and dupA⁻ H. pylori infection (P < 0.01 for both groups compared with the uninfected group in NMDS analysis). Differences in microbial composition were also observed in dupA⁺ H. pylori- and dupA⁻ H. pylori-infected patients (P = 0.015) (Figure 4B).

ANOVA revealed that the RA of Leptotrichia, Bulleidia, Capnocytophaga, Gemella, Lactobacillus, Aggregatibacter, Atopobium, and Ochrobactrum was higher in dupA⁺ H. pylori-infected patients than in dupA⁻ H. pylori-infected patients (P < 0.05) (Figure 4C).

As the commensal microbes in the human body compete and cooperate with each other and act as important players in maintaining our health (Faust et al., 2012), further exploration of the gastric microbial network might help to better understand the influence of H. pylori infection on the gastric mucosa. In this study, we observed a complex correlation network among the key genera in mucosal microbes without H. pylori infection, most of which were in a mutually beneficial relationship (Figure 5A). Infection with dupA⁺ H. pylori resulted in reduced microbial diversity, weakened interactions between microbes, and the disappearance of mutual promotion between probiotic bacteria such as Streptococcus and Prevotella (Figures 5A, B). Most significantly altered gastric microbiome was observed in dupA⁻ H. pylori-infected patients, with a loss of most probiotics in the gastric microbiome (Figure 5C).