Impact of Helicobacter Pylori Infection on Duodenal Microbial Community Structure and Microbial Metabolic Pathways

Recent reports suggest that Helicobacter pylori infection may be related to the onset of certain diseases. However, the H. pylori-related factors that play a role in the etiology of these diseases have not been fully elucidated. This study aimed to elucidate the impact of H. pylori infection on the structure of commensal duodenal microbiota and their biofunctions.


Results
Thirteen subjects were positive for H. pylori, while thirty-four were negative. We observed 1404 bacterial operational taxonomic units from 23 phyla and 253 genera. In the H. pylori-positive group, we observed higher abundances of Proteobacteria and lower abundances of Actinobacteria and TM7 than that in the H. pylori-negative group. The abundances of 10 genera differed signi cantly between the H. pyloripositive and -negative groups. Microbiota features in the H. pylori-positive group was signi cantly in uenced by 12 taxa primarily belonging to Gammaproteobacteria. Microbial functional annotation collated using the Kyoto Encyclopedia of Genes and Genomes Orthology database showed that 12 microbial metabolic pathways (Synthesis and degradation of ketone bodies, Tryptophan metabolism, Nglycan biosynthesis, Ether lipid metabolism, Lipoarabinomannan (LAM) biosynthesis, Linoleic acid metabolism, alpha-Linolenic acid metabolism, Biotin metabolism, Carotenoid biosynthesis, Phenylpropanoid biosynthesis, Biosynthesis of siderophore group nonribosomal peptides, Biosynthesis of terpenoids and steroids) were signi cantly affected by H. pylori infection.

Conclusions
H. pylori infection disrupted the normal bacterial communities in the duodenum and changed the biofunctions of the commensal microbiota, primarily by upregulating speci c metabolic pathways. This alteration may be related to the onset mechanisms of the diseases suspected of being related to H. pylori infection.

Background
The presence of Helicobacter pylori infection results in the development of chronic gastritis or peptic ulcer and is also related to the development of certain gastric cancers. Recent epidemiological ndings suggest that the prevalence of cardiovascular disease, haematological disease, neurodegenerative disease, liver disease, and metabolic syndrome is high in patients with H. pylori infection [1][2][3][4][5]. However, the causal relationships between the pathogenesis factors in H. pylori infections and various other extra gastroduodenal diseases remain unknown [2,6,7].
The theory that the bioactivities of commensal gut microbiota markedly in uence host biofunction has attracted considerable attention, and studies aimed at the elucidation of the pathophysiology of various diseases have been conducted. Additionally, the duodenum plays a key role in the establishment of crosstalk between the gut and the central nervous system because the release of brain-gut hormones and neurotransmitters in the small intestine, including the duodenum, is regulated by food stimuli, dietary behaviours, and the information arising from the intraluminal environment. These hormones regulate widespread biofunctions, such as metabolism, biosynthesis, feeding behaviour, and gastrointestinal functions [8][9][10]. Schulz et al. (2016) have reported that H. pylori infection alters the duodenal microbiota based on evidence obtained from the analysis of reverse-transcribed 16S rRNA and that the same results were derived based on the investigation of duodenal biopsies and aspirates [11]. These ndings suggest that alteration in the duodenal microbiota induced by H. pylori infection is related to the onset of various other gastroduodenal diseases. This is because certain degradation products of digestion, attributed to duodenal microbial biofunction, serve as chemical effectors for host biofunctions [12,13].
Based on the above-mentioned information, we aimed to elucidate the impact of H. pylori infection on the structure of the commensal duodenal microbiota and their biofunctions using conventional microbial taxonomic diversity analyses and the novel linear discriminant analysis (LDA) effect size (LEfSe) algorithm method to discover metagenomic biomarkers that could explain differences among microbial communities [14]. We also performed metagenomic functional predictions using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database to infer the microbial genetic features associated with biological functions and metabolic pathways [15][16][17], and this approach enabled the extraction of speci c genetic information related to microbial biofunction from a chaotic metagenomic bin.

Results
All 47 samples provided su cient gene information for analysis. In uence of H. pylori on bacterial community structures Figure 4 illustrates the bacterial community structures at the phylum level. The relative mean abundances of Actinobacteria and TM7 were signi cantly higher in the H. pylori-negative compared to the H. pyloripositive group. In contrast, the abundance of Proteobacteria was signi cantly higher in the H. pyloripositive group, and Acidobacteria and Planctomycetes were evident only in the H. pylori-negative group.
The relative mean abundances of 14 (out of 18) detected phyla did not remarkably differ between the H. pylori-negative and H. pylori-positive groups. The data for the other four phyla were invalid for statistical analysis (Additional File 2).
The relative mean abundances of 10 genera signi cantly differed between the H. pylori-negative and H. pylori-positive groups (Table 1). Only the relative abundance of Neisseria was markedly higher in the H. pylori-positive group, and the relative abundances of the other nine genera (Rothia, {unknown order} TM7-3, Leptotrichia, {unknown genus} Lachnospiraceae, Megasphaera, {unknown genus} F16, Moryella, Filifactor, and Paludibacter) were markedly higher in the H. pylori-negative group. Furthermore, 188 and 143 genera were detected in the H. pylori-negative and H. pylori-positive groups, respectively. These differences were attributed to the different microbial community structures observed within each group. Speci cally, 60 and 15 genera were found only in the H. pylori-negative and H. pylori-positive groups, respectively. These results suggest that the duodenal microbial environment can maintain greater microbial diversity without H. pylori compared to that maintained with its presence. Helicobacter was detected only in the H. pylori-positive group, with an abundance of 2.79% ± 6.82% (Additional File 3).

In uence of H. pylori on biologically relevant features
The LDA score derived from LEfSe analyses indicated that 12 taxa markedly in uenced the biological features of the duodenal microbiota in the H. pylori-positive group (Fig. 5). These 12 taxa comprised three phyla (Streptophyta, Cyanobacteria, and TG5), one class (Gammaproteobacteria), one order In uence of H. pylori on duodenal microbial biofunctions Among 327 investigated KEGG pathways (Additional File 4), 163 were metabolic. The KEGG Orthology (ko)-abundance of 12 of these metabolic pathways markedly differed in the presence or absence of H. pylori infection (Table 2), and the ko-abundance of 9 of these 12 was remarkably greater in the H. pyloripositive group. This suggests that H. pylori generally promotes metabolic functional activities in duodenal microbes. The remaining pathways were associated with genetic information processing (n = 16), environmental information processing (n = 21), cellular processes (n = 19), organismal systems (n = 48), and human diseases (n = 60). Among these 164 pathways, 18 markedly differed between the individuals with and without H. pylori. However, it was not possible to infer whether these pathways were functional for microbial biofunctions and whether the pathways affected host biofunctions.

Discussion
Although H. pylori infection has long been suspected to result in the development of cardiovascular disease, haematologic disease, and metabolic syndrome, the roles of commensal microbiota in these diseases have remained unknown. The present study demonstrated that H. pylori infection markedly in uenced the relative abundances of three phyla and ten genera in the duodenal microbiota, and that the altered duodenal microbiota was characterised by increased Neisseria abundance and an enhanced impact of Gammaproteobacteria. The abundance of multiple commensal microbial metabolic pathways was also considerably altered. These results suggested that H. pylori altered the production of microbial metabolites, which might impact host biofunctions. Many studies have investigated the community structure and diversity of the gut microbiota before and after administration of therapeutic agents for various diseases. Although many comparative studies on the gut microbiota have associated differences in the gut microbiota with the development and pathophysiology of certain diseases, the results remain inadequate, especially, for determination of the factors originating from altered microbiota that are substantial etiological effectors.
The present α-and β-diversity analyses showed no signi cant differences between H. pylori-positive and H. pylori-negative groups (Figs. 2 and 3). However, these results do not necessarily indicate that an identical abundance or representation of bacterial species exists in each group. In fact, the duodenal bacterial community structures differed at the phylum level between the two groups, with greater abundance of Proteobacteria and lower abundances of Actinobacteria and TM7 (Saccharibacteria) in the H. pylori-positive group. The greater abundance of Proteobacteria is a predictable natural consequence because H. pylori belongs to Epsilonproteobacteria, which is a subclass of Proteobacteria. However, the bacterial community structure at the genus level and LEfSe results suggest that H. pylori infection altered the microbial features by increasing the abundance of Neisseria and by enhancing the impact of Gammaproteobacteria in the duodenum (Table 1 and Fig. 5). The increase in Neisseria in the duodenum is probably related to the gastric acid output attributed to atrophic gastritis induced by H. pylori. Intraoral indigenous bacteria in the genus Neisseria are not generally highly pathogenic, except for Neisseria gonorrhoeae and Neisseria meningitidis, which respectively cause gonorrhoea and meningitis [18]. However, excessive Neisseria proliferation in the duodenum may be somewhat pathogenic based on the Neisseria-associated changes induced in the microbial community structure [19,20]. The subclass Gammaproteobacteria consists of several medically important bacteria belonging to families, such as Enterobacteriaceae, Vibrionaceae, and Pseudomonadaceae.
Many studies examining the relationships between H. pylori infection and the development of extragastric diseases have identi ed increased short-chain fatty acid (SCFA) production induced by the proliferation of Bacteroidetes as a potentially important causative factor. These SCFAs induce the release of gut hormones, such as peptide YY and glucose-like peptide-1, the activation of host metabolic pathways, the elicitation of mucosal immune response, and trigger in ammation [21][22][23]. The present study found that H. pylori did not markedly change the abundance of Bacteroidetes in the duodenum (Fig. 4). The LDA scores also indicated that taxa belonging to Bacteroidetes did not signi cantly impact duodenal microbial features (Fig. 5). These ndings were consistent with those of previous analyses conducted using duodenal aspirates [8]. Although the notion that increased SCFA production leads to the development of various diseases is attractive, SCFAs are generated primarily as a result of the fermentation of nonhost-digestible dietary bres by the colonic microbiota. Therefore, other factors associated with upper gastrointestinal microbial functions warrant further exploration.
The KEGG pathway analysis provided substantial data for investigation of microbial function and showed that 12 bacterial metabolic pathways were affected by the presence or absence of H. pylori infection. Two pathways that were upregulated in the H. pylori-positive group, synthesis and degradation of ketone bodies (ko00072) and ether lipid metabolism (ko00565), are important for fatty acid degradation, butyrate and acetic acid synthesis, and the production of phosphocholine or seminolipid, which functions in the maintenance of mucosal integrity and immunohomeostasis [24,25].
Notably, the ko-abundance of the tryptophan metabolism pathway (ko00380) was markedly greater in the H. pylori-positive group, suggesting that an abnormal tryptophan provision from the intestine impaired serotonin production. Serotonin is a paracrine messenger expressed primarily in enterochroma n cells and enteric neurons. Various effects of serotonin on brain-gut interactions are exerted mainly through serotonin receptors on peripheral intrinsic (IPAN) and extrinsic (EPAN) primary afferent neurons [26,27]. Therefore, disruption of the continuous and subtle tryptophan provision attributed to the presence of altered microbiota may result in the development of nerve circuit disorders involving serotonin release. This information would help to clarify the causal relationships between H. pylori infection, the duodenal microbiota, and the pathophysiology of functional dyspepsia [28][29][30]. Additionally, issues related to serotonin production may alter local serotonin concentrations in portal blood, which can also affect the gut-liver axis [31].
The pathways of linoleic (ko00591) and α-linolenic (ko00592) acid metabolism were also upregulated in the H. pylori-positive group ( Table 2). Such an upregulation may cause an imbalance between the levels of the ω-3 and ω-6 fatty acids and affect the arachidonic acid cascade that is associated with the occurrence of in ammation [32]. The biotin metabolic pathway (ko00780) was also upregulated in the H. pylori-positive group. Bacteria synthesise biotin, which is an indispensable essential cofactor for fatty acid biosynthesis. The production of vitamin A may also be affected by H. pylori, because the biosynthetic pathway of the vitamin A precursor, carotenoid (ko00906), was upregulated in the H. pyloripositive group [33,34]. The phenylpropanoid biosynthesis pathway (ko00940) was also upregulated signi cantly in this group. However, the physiological signi cance of this upregulation in humans cannot be easily determined, because the roles of metabolites (chavicol, eugenol, lignin, and so on) originating from this pathway have not been fully elucidated. The terpenoid and steroid biosynthesis pathway (ko01062) was upregulated in the H. pylori-positive group. This may extensively affect host functions, because terpenoids are steroid precursors and closely related to cytochrome P450 that functions as an oxidase in terpenoid biosynthesis [35,36]. Yap et al. (2017) found 45 upregulated, and 551 downregulated serum metabolites 18 months after H. pylori eradication [37]. The affected metabolites were mapped to various biochemical pathways, including tryptophan metabolism, biosynthesis of unsaturated fatty acids, and linoleic acid metabolism. Although it has not been ascertained whether these alterations affect host biofunctions, our ndings con rmed that the metabolomic products reported by Yap et al. originated from microbial metabolic pathways affected by H. pylori infection.
The present study also has several study limitations and issues that need to be addressed in the future.
First, we could not exclude the possibility that few subjects in the H. pylori-negative group might have already experienced remarkable changes in the structure and biofunctions of their commensal duodenal microbiota due to a previous history of H. pylori infections. To eliminate this concern, it is necessary to analyse the changes in the microbial features before and after H. pylori eradication therapy in the same subject.
Furthermore, we could not completely evaluate the effect of gastric acid on the duodenal microbiota because we did not quantify gastric acid secretion. The extent of gastric mucosal atrophy caused by H. pylori infection is dependent on various factors, such as the age, duration of infection, differences between individual immune responses, and number of bacteria. The gastric acid output depends on the extent of gastric mucosal atrophy, and the extent of atrophic gastritis is closely related to a history of H. pylori infection [38]. In fact, the endoscopic ndings in this study indicated that an extended atrophic change was likely to be observed in the H. pylori-positive group. Additionally, the duodenal microbiota might be affected by a decrease in the gastric acid output.
Another limitation of this study is that contamination with gastric microbiota could not be completely excluded because of the sampling method. A concern has been raised that aspirate samples include only oating microbiota, a phenomenon which may exhibit a less remarkable association with host biofunctions, and that the microbiota present in biopsy samples (mucosa-associated microbial community structure) actually inhabit the mucosa, a phenomenon that may be considered to expound microbiota-host associations [39,40]. Our ndings suggest that microbial metabolite production may uctuate depending on changes in the commensal duodenal microbiota, and that this phenomenon may impact host biofunctions. These mechanisms may not be affected by aspects, such as the duodenal juice or mucosa, which are regions that are inhabited by microbes because microbial metabolites act as chemical effectors.

Conclusions
In conclusion, H. pylori infection changed aspects of the microbiota in the descending part of the duodenum. This dysbiosis altered the commensal microbial biofunctions, characterised mainly by the upregulation of microbial metabolic pathways, which might affect host biofunctions. The gut microbiota can be regarded as an independent organ within the gut lumen, and an investigation of biofunctions originating from this "commensal bacterial organ" would help to elucidate the aetiology of various diseases.

Participants
This study included 20 male and 27 female patients (mean age: 58.8 ± 11.3 years) who were screened for gastric cancer. We obtained information from all participants about treatment with gastric acid inhibitors, antibiotics, and a medical history of H. pylori eradication (Additional File 1). Patients treated with antibiotics within 4 weeks before sampling were excluded. The study protocol (Fig. 1) was implemented under the approval of the Ethics Committee of Toho University School of Medicine (authorization number: A16080), in accordance with current good clinical practice and the Declaration of Helsinki (2013). All participants provided written informed consent to participate before enrollment.

Patient and public involvement
This design of this study proceeded without public involvement. Patient involvement was restricted to sample collection at the time of enrollment. Patients were neither consulted to interpret the results, nor were invited to contribute to writing or editing this article.

Collection of duodenal uid samples and oesophagogastroduodenoscopy
Duodenal uid samples were collected from the descending part of the duodenum using a PW-2L-1 uororesin tube (Olympus, Tokyo, Japan) under standard video endoscopy with the Olympus GIF-XQ260 or GIF-XP260N video gastroscope (Olympus, Tokyo, Japan). The tube was sterilised and changed for each patient. Duodenal uid was aspirated immediately after injecting 5 mL of saline into the descending part of the duodenum, and the aspirate was immediately cryopreserved at − 80 ℃. One certi ed endoscopist (HZ) conducted all endoscopic procedures and sampling to avoid generation of bias. The endoscopic ndings were recorded simultaneously, and the extent of atrophic gastritis was evaluated in accordance with the Kimura-Takemoto classi cation for endoscopic atrophy [41].

Extraction of genomic DNA
Genomic DNA (gDNA) was extracted from duodenal uid using the PowerFecal DNA Isolation Kits (Mo Bio Laboratories, Inc., Carlsbad, CA, USA), as per the manufacturer's instructions.

Identi cation of H. pylori
We performed conventional nested polymerase chain reactions (PCR) using the extracted gDNA samples [42] and then sequenced amplicons to con rm the presence of H. pylori. The participants were then assigned to groups based on the presence or absence of H. pylori.

Bioinformatics and statistical analyses
The sequences were processed and clustered into OTU with 97% identity using the Greengenes database (version 13.5) as the reference [45]. We evaluated bacterial diversity by calculating α-and β-diversity from rare ed OTU tables. The α-diversity was determined using richness (based on the number of OTUs) and evenness (appraised by the Shannon diversity index) [46,47]. Differences in diversity indices between the H. pylori-positive and H. pylori-negative groups were compared using Welch's t-tests. The β-diversity was evaluated, based on the OTU table [48], as the unweighted UniFrac distance, which can be used to distinguish dissimilarities between the microbial pro les of two samples. The β-diversity results were analysed via permutational multivariate analyses of variance (PERMANOVA) using the CLC Genomics Workbench 10.0.1 and CLC Microbial Genomics Module 2.5 (Qiagen, Hilden, Germany). The relative abundances of phyla and genera in the H. pylori-positive and H. pylori-negative groups were compared based on the unrare ed OTU tables using Welch's t-tests.
The LEfSe algorithm can identify genomic taxa with a relative abundance that markedly differs between groups. We computed LEfSe using the Galaxy web application and work ow framework (https://huttenhower.sph.harvard.edu/galaxy/) to support high-dimensional class comparisons with a focus on metagenomic analysis.
Biofunctions of the duodenal microbiota were inferred via metagenomic functional annotation. The OTU abundance table was uploaded to the Piphillin server (https://piphillin.secondgenome.com/) with the KEGG Orthology (ko) database as the reference genomic database [49]. Thereafter, KEGG pathways were identi ed based on the gene information in the OTU, and this approach enabled interpretations of highlevel biofunctions of the microbiota.
The results of KEGG pathway analysis are identi ed by 'ko' followed by a ve-digit number and include quantitative information. Each genome copy number can represent an abundance of genome content in each OTU that is expressed as ko-abundance. Each ko-abundance corresponds to the abundance of a speci c KEGG pathway and quantitatively represents microbial biofunction characteristics. Here, we extensively investigated KEGG pathways classi ed under metabolism because the gut microbiota could be regarded as an independent organ in the gut lumen with metabolic functions that produces various metabolites [13]. Each ko-abundance was compared between the H. pylori-positive and H. pylori-negative groups using the Mann-Whitney U tests. All values with P < 0.05 were considered statistically signi cant.

Declarations
Ethics approval and consent to participate and Consent for publication: The study protocol (Fig. 1)       Differences in the bacterial community structures at the phylum level in Helicobacter pylori-positive and H. pylori-negative groups. *P < 0.05 and **P < 0.01 (Welch's t-tests).