Helicobacter pylori infection alters gastric and tongue coating microbial communities

Abstract Objective Infection with Helicobacter pylori (H pylori), especially cytotoxin‐associated gene A‐positive (CagA+) strains, has been associated with various gastrointestinal and extragastric diseases. The aim of this study was to characterize H pylori‐induced alterations in the gastric and tongue coating microbiota and evaluate their potential impacts on human health. Design The gastric mucosa and tongue coating specimens were collected from 80 patients with chronic gastritis, and microbiota profiles were generated by 16S rRNA gene sequencing. Samples were grouped as H pylori negative (n = 32), CagA‐negative H pylori infection (n = 13), and CagA‐positive H pylori infection (n=35). The comparison of bacterial relative abundance was made using a generalized linear model. Functional profiling of microbial communities was predicted with PICRUSt and BugBase. Microbial correlation networks were produced by utilizing SparCC method. Results Significant alterations of the gastric microbiota were found in the H pylori+/CagA+ samples, represented by a decrease in bacterial diversity, a reduced abundance of Roseburia, and increased abundances of Helicobacter and Haemophilus genera. At the community level, functions involved in biofilm forming, mobile element content, and facultative anaerobiosis were significantly decreased in gastric microbiome of the H pylori+ subjects. The presence of CagA gene was linked to an increased proportion of Gram‐negative bacteria in the stomach, thereby contributing to an upregulation of lipopolysaccharide (LPS) biosynthesis. The number of bacterial interactions was greatly reduced in networks of both tongue coating and gastric microbiota of the H pylori+/CagA+ subject, and the cooperative bacterial interactions dominated the tongue coating microbiome. Conclusions Infection with H pylori strains possessing CagA may increase the risk of various diseases, by upregulating LPS biosynthesis in the stomach and weakening the defense of oral microbiota against microorganisms with pathogenic potential.


| INTRODUC TI ON
Highly acidic condition together with various antimicrobial chemicals and enzymes make stomach a hostile environment for most microorganisms. However, investigations of gastric fluid and biopsy samples by next-generation sequencing (NGS) demonstrate that the human gastric microbiota is a diverse ecosystem that dominated by five phyla, including Actinobacteria, Bacteroidetes, Firmicutes, Fusobacteria, and Proteobacteria. 1,2 There is a trend of gastric microbiota alterations in people with Helicobacter pylori (H pylori) infection; however, inconsistent results have been obtained. [3][4][5][6] With a small Colombian patient cohort (n = 40), Yang et al showed that the overall gastric microbiota composition was largely independent of the H pylori infection and carriage of the cag pathogenicity island. 3 In contrast, another study with a small cohort (n = 30) demonstrated that the abundances of several genera in the gastric microbial community are significantly different between H pylori negative and H pylori positive individuals. 4 Infection with H pylori is a leading risk factor for the development of gastric cancer, making it a type I carcinogen. 7 However, only approximately 3% of H pylori-infected patients developed gastric cancer. 8 The differential host responses to H pylori colonization indicate that the pathogenicity of H pylori strains and other factors implicated in disease development.
It has been suggested that the cytotoxin-associated gene A (CagA) gene with its products is linked to increased pathogenicity of H pylori strains. As one of the most virulent pathogenicity islands genes, CagA is able to perturb multiple host signaling pathways by acting as a hub or extrinsic scaffold protein, in turn potentiating malignant transformation. 9 CagA can interact with multiple cell components and activate multiple signaling molecules downstream of receptor tyrosine kinase growth factors, therefore causing a set of complex cellular alterations, for example, enhanced proliferation and attenuated apoptosis, changes in epithelial cell morphology and polarity, and prevention of the assembly of apical junctions. 9 A serologic response to CagA in H pylori-infected patients was found to be strongly associated with peptic ulceration. 10 Oncogenic potential of CagA has been proved in animal experiments. 11,12 People infected by cagA -positive H pylori strains have a higher risk of developing gastric carcinoma compared to those infected with cagA-negative strins. 13 CagA-positive H pylori strains also participate in the inflammatory response, through the production of certain cytokines such as IL-1β and IL-8, and activation of NF-κB. 14 Thus, infection of CagA+ H pylori strains can alter both the local and systematic environments, which might consequently influence the local and distant microbiota in the host. Indeed, with a mouse model, the intestinal microbiota has proven to be altered by the presence of H pylori in the stomach. 15 To date, the impact of H pylori infection on local and distant microbial populations in humans remains to be explored. In particular, it is essential to illuminate the role of non-H pylori organisms in the development of gastrointestinal diseases, so as to improve diagnosis and treatment strategies. The oral microbiome helps host against invasion of opportunistic microorganisms. The oral microbiome also impacts the microbial communities that colonize the gastrointestinal tract, 16 and its imbalances contribute to not only oral diseases but also risk of gastrointestinal disorders, adverse pregnancy outcomes, cardiovascular disease, diabetes, rheumatoid arthritis, and nervous systemic diseases. 17 In this study, we sought to elucidate the relationships between CagA status of H pylori infection and the compositional, functional, and ecological changes in both gastric and oral microbiome communities.

| Cohort
Eighty Chinese participants diagnosed with chronic nonatrophic gastritis were enrolled in this study, who were diagnosed by gastroscopy  Table S1.

| Sample collection, DNA isolation, and 16S rRNA gene sequencing
Sample collections were performed in the morning, and all the volunteers had fast (no food, fluids, or water) before the procedure. In order to overcome variations potentially caused by different gastric regions, all mucosal biopsy samples were collected from the antrum region of the stomach. Tongue coating samples were collected by scraping the tongue dorsum three times with sterilized cotton swabs.
All samples were enclosed in sterile plastic tubes with RNAlater (Qiagen, German) immediately after collection and transported to HRK-biotech laboratory with dry ice. Both gastric biopsy and tongue coating samples were stored at −80°C until tested.
Samples were centrifuged at 3420 X g for 10 minutes, and supernatant was removed. Homogenization was employed using 0.1 mm zirconium beads and bead beating for 5 minutes at 1500 g (Scientz-48, High-throughput Tissue Grinder, Scientz, China). Total DNA was isolated using QIAamp DNA Mini Kit (Qiagen, Germany), following the manufacturer's instructions. DNA was quantified using a Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA).

| Data analysis
The forward and reverse reads were merged using VSEARCH 19 with a minimum overlap set to 10 bp. Sequences were quality fil- orthologs. 21 The functional differences between groups were determined with Linear Discriminant Analysis Effect Size (LEfSe). Organism level microbiome phenotypes were predicted and compared with BugBase. 22 The proportions of six phenotypic categories, including Gram staining, oxygen tolerance, ability to form biofilms, mobile element content, pathogenicity, and oxidative stress tolerance, were compared among different groups of patients with gastritis.

| Microbial association network analysis
SparCC algorithm 23 was utilized to elucidate microbial interactions for each sample group. SparCC obtains the P-value through permutation-based approaches that iteratively correct spurious correlation coefficients, and limits false discovery rates. OTUs that occur in at least 50% of the samples within one group and with an average relative abundance of at least 1% were selected to infer the correlation network. Taxon-taxon correlation coefficients were estimated as the average of 20 inference iterations and 200 shuffled matrices for the P-value calculation. This set of iterative procedures was applied separately to each sample group of gastric mucosa and tongue coating ecologies. Correlation coefficients with magnitude ≥0.3 and P-value <0.05 were selected for visualization in Cytoscape V.3.6.1.

| Infection with CagA+ H pylori strains influences the gastric microbial community structure
It was shown that pyrosequencing method detectedH pylori sequences in about 60% of samples that were H pylori negative by a F I G U R E 1 Microbial diversity and richness reduced in gastric biopsies of H pylori+/CagA+ samples. Bacterial community richness was defined by the observed number of OTUs (A), and alpha diversity was calculated using the Shannon index (B), Simpson index (C), and inverse-Simpson index (D). Statistical significance was determined by paired student t-test, *P < 0.05, **P < 0.01, and ***P < 0.001 combination of conventional testing (histology, rapid urease test, and culture), 24 which emphasizes the use of NGS as a more sensitive method for determining H pylori infection. Following the proposed cutoff value of determining H pylori infection, 24 we labeled samples with >1% H pylori relative abundance as H pylori positive (H pylori+).
However, there is still a certain possibility to have false-negative case due to uneven distribution of the organism through the gastric mucosa. To reduce the false-negative rate, we also labeled samples with ≤1% H pylori relative abundance as H pylori+ if the patients were positive for both 13 C-urea breath test and serological CagA. The rest of samples were grouped as H pylori negative (H pylori−). We noticed that some low relative abundances of H pylori sequences were detected when breath test and/or serology were positive. This is likely due to the uneven distribution of H pylori in the stomach.  Figure S1).

| Relative microbial abundances in the human stomach are influenced by the presence of CagA+ H pylori strains
Bacterial profiles in the mucosa revealed that the gastric microbiota As for the tongue coating microbiota, the most prevalent phyla were Bacteroidetes (30.74%), Proteobacteria (28.53%), Firmicutes (21.74%), Actinobacteria (11.12%), and Fusobacteria (6.29%) ( Table   S2). The proportion of Actinobacteria in the H pylori+/CagA− group was significantly lower than other groups ( Figure 3B). The genera Akkermansia, Meganomas, and Ruminococcus were found to be more

| Changes in the functional capacity of gastric microbiota is associated with H pylori infection and CagA gene expression
The functional capacities of the gastric mucosa and tongue coating microbiome were predicted based on 16S data using PICRUSt and BugBase. Statistically significant KEGG pathways for each group

| Alterations of gastric and tongue coating microbial interactions are associated with H pylori infection and CagA gene expression
The human body contains many diverse microorganisms competing and cooperating with one another, thus acting as an ecosystem supporting health or promoting disease. Most of microbial interactions are niche-specific 25 ; therefore, the disease-specific microenvironment may also shape microbial interaction networks. We next inferred taxonomic correlations among the three sample groups, using the SparCC algorithm. OTUs that occur in ≥50% of the samples within one group and with a relative abundance ≥1% were selected to infer the correlation network.
Among taxa colonizing the gastric mucosa, the number of interactions was highest in the H pylori− group (n = 24), which was sig- Taken together, infection with H pylori, especially the CagA+ strains, was associated with a reduced complexity of bacterial interactions in both gastric and tongue coating microbial communities. As a result, the oral microbiome structure is more likely to be disturbed such as by invasion of alien microorganisms, since the community response hinged on the complex interspecific interactions.

| DISCUSS IONS
Former studies on mice gastric microbiome indicated that H pylori infection significantly affects the population structure of the gastric and intestinal microbiota, alters gastric immune and inflammatory responses, and causes distant effects via altered hormones and immunity. 15 Because physiology and immunity were substantially in- The metabolites of gut microbiota play crucial roles in determining the biochemical profile of the diet, and thus modulating host metabolism. However, the metabolism of gastric microbiota (especially non-H pylori species) and their interplay with host metabolism in health and disease remain poorly understood.
We found that the functions involved in LPS biosynthesis were upregulated within patients infected by CagA+ strains. LPS (also termed endotoxin) is the major surface molecule of most Gramnegative bacteria that play a key role in host-pathogen interactions with the innate immune system. Indeed, the proportion of Gram-negative taxa was significantly higher in the stomach of H pylori+/CagA+ patients. In the gastric mucosa, LPS is capable of suppressing the elimination of H pylori by interfering the activity of innate and adaptive immune cells, diminishing the inflammatory response, and affecting the adaptive T lymphocyte response, thus facilitating the development of chronic gastritis. 26 Additionally, the exposure of LPS has been found in close associations with several diseases such as endotoxemia, autoimmune and allergic disease, 27 obesity, 28 and nervous systemic diseases (including autism 29

and
Alzheimer's disease 30 ). It has been hypothesized that most of LPS is derived from the gut microbiota, and enters blood circulation due to an increased intestinal barrier permeability. However, it is As an open system, the oral cavity contains a variety of microbial habitats including teeth, tongue, gingival sulcus, cheek, and palates which contribute to a vast ecological complexity. The diversity and composition of tongue coating have been found more stable comparing to teeth and supragingival plaque which were more susceptible to oral hygiene habits. 31 In addition, most of the highly abundant taxa found in saliva were derived from the tongue. Therefore, we collected tongue coating samples to access the oral microbiota in this study. Although H pylori infection had a subtle influence on the microbial diversity and structure of tongue coating microbiota, the complexity of bacterial interaction network was greatly reduced.
Our data indicated that the total number of interactions were significantly declined in the oral microbiota of the H pylori+ group, especially for the CagA+ patients. Additionally, oral microbiota of CagA+ patients was dominated by co-occurrence relationships, further indicating a low network complexity since cooperation is destabilizing for the community. 32 Thus, the oral microbiota of CagA+ patients may be more tolerant to the invasion of alien species.
In conclusion, our study identified the influences of H pylori infection on gastric and tongue coating microbial populations. Infection with CagA+ H pylori strains reduced the diversity of gastric microbiota and altered its composition and functions. The upregulation of LPS biosynthesis likely resulted from Gram-negative bacteria enrichment in the stomach of CagA+ patients may increase the chance of developing various diseases that associated with LPS exposure.
CagA+ H pylori infection was also associated with an unstable tongue coating bacterial community, which suggests a weak resistance to invasion by microorganisms with pathogenic potential.
F I G U R E 6 H pylori infection altered the gastric and tongue coating microbial interaction networks. Correlation networks of gastric (A-C) and tongue coating microbiota (D-E) in H pylori−, H pylori+/CagA−, and H pylori+/CagA+ were inferred by using SparCC. Visualization was applied to a subset of correlations with strengths greater than 0.3. Node size represents mean relative abundance of taxon in each sample group; metacommunity markers are denoted by node numbers accordingly. Node colors indicate the members of the same bacterial phylum

D I S CLOS U R E S O F I NTE R E S T S
The authors declare no competing financial interests.

AUTH O R S' CO NTR I B UTI O N S
GX analyzed data and drafted the manuscript. GJ, XY, WH, and LY collected human samples. GJ and YD performed DNA isolation and sequencing. ZY designed the study and revised the paper.