Helicobacter pylori Induced Gastric Immunopathology Is Associated with Distinct Microbiota Changes in the Large Intestines of Long-Term Infected Mongolian Gerbils

Background Gastrointestinal (GI) inflammation in mice and men are frequently accompanied by distinct changes of the GI microbiota composition at sites of inflammation. Helicobacter (H.) pylori infection results in gastric immunopathology accompanied by colonization of stomachs with bacterial species, which are usually restricted to the lower intestine. Potential microbiota shifts distal to the inflammatory process following long-term H. pylori infection, however, have not been studied so far. Methodology/Principal Findings For the first time, we investigated microbiota changes along the entire GI tract of Mongolian gerbils after 14 months of infection with H. pylori B8 wildtype (WT) or its isogenic ΔcagY mutant (MUT) strain which is defective in the type IV secretion system and thus unable to modulate specific host pathways. Comprehensive cultural analyses revealed that severe gastric diseases such as atrophic pangastritis and precancerous transformations were accompanied by elevated luminal loads of E. coli and enterococci in the caecum and together with Bacteroides/Prevotella spp. in the colon of H. pylori WT, but not MUT infected gerbils as compared to naïve animals. Strikingly, molecular analyses revealed that Akkermansia, an uncultivable species involved in mucus degradation, was exclusively abundant in large intestines of H. pylori WT, but not MUT infected nor naïve gerbils. Conclusion/Significance Taken together, long-term infection of Mongolian gerbils with a H. pylori WT strain displaying an intact type IV secretion system leads to distinct shifts of the microbiota composition in the distal uninflamed, but not proximal inflamed GI tract. Hence, H. pylori induced immunopathogenesis of the stomach, including hypochlorhydria and hypergastrinemia, might trigger large intestinal microbiota changes whereas the exact underlying mechanisms need to be further unraveled.


Introduction
Chronic Helicobacter (H.) pylori infection represents a significant health burden affecting approximately half of the world's population [1]. Even though the vast majority of infected individuals remains asymptomatic or displays rather minor unspecific symptoms, the pathogen can cause chronic gastritis, peptic ulcer, gastric adenocarcinoma or mucosa-associated lymphoid tissue lymphoma [2,3]. As a consequence of epidemiological studies, in 1994 the WHO has declared H. pylori a class I carcinogen [4]. H. pylori produce a plethora of virulence factors inducing gastric immunopathology. Among these, the vacuolating cytotoxin A (VacA) and cytotoxin-associated antigen A (CagA) have been studied intensively [5,6,7,8,9].
Strains carrying a cag pathogenicity island (cagPAI) are able to translocate the effector protein CagA via the type IV secretion system (T4SS) into the host cell. CagA gets tyrosine phosphorylated by host kinases and interferes with the signal transduction pathways regulating cell polarity, inflammation, proliferation, and apoptosis [5,10]. The H. pylori mutant DcagY is defective in the T4SS, thus representing a less virulent strain, which lacks the ability to induce a corpus dominant atrophic gastritis as confirmed in our previous Mongolian gerbil study [9]. As compared to mice, the Mongolian gerbil infection model has been proven to better mimic key features of pathophysiology and long-term sequelae in H. pylori infected humans (reviewed in [11,12,13,14]). Following H. pylori infection of gerbils, like in humans the gastric inflammatory process starts in the antrum mucosa and further expands to the corpus whereas in infected mice the antrum is rarely involved. Importantly, H. pylori type I strains expressing CagA, VacA and a functional type IV secretion system only persist in a Mongolian gerbil, but not murine stomach in a genetically stable fashion. In addition, infected gerbils develop severe gastric lesions such as peptic ulcer and gastric adenocarcinoma [11,12,13,14].
Recent reports have highlighted that H. pylori infection is associated with changes in the gastric microenvironment, which in turn affects the gastric microbiota composition. In our own study, for instance, we have shown that following H. pylori infection bacterial species usually restricted to the lower intestinal tract (such as Bacteroides/Prevotella spp. and clostridia, among others) were present in stomach samples of H. pylori infected mice, whereas Lactobacillus spp. were predominant in control animals [15]. Furthermore, in a previous report Mongolian gerbils with gastritis and duodenitis harbored significantly higher Bacteroides spp. numbers at sites of inflammation 12 weeks following H. pylori infection [16].
So far, microbiota changes in H. pylori infected Mongolian gerbils have only been investigated at sites of inflammation itself [16,17,18]. Therefore, for the first time we investigated potential changes of the microbiota composition distal to the inflammatory process, particularly in long-term H. pylori infection. Results represent a comprehensive survey of the microbiota composition within the entire gastrointestinal (GI) tract of Mongolian gerbils 14 months following H. pylori wildtype infection as compared to its isogenic DcagY mutant strain applying cultural and molecular methods.

Results
Inflammatory Responses in the Gastrointestinal Tract of Mongolian Gerbils Following Long-term H. pylori Infection Acute and chronic GI inflammation is accompanied by shifts within the intestinal microbiota composition at sites of inflammation. In the presented study, we were interested whether long-term H. pylori infection results in luminal microbiota changes at sites of gastric inflammation or beyond in the distal GI tract. To address this question Mongolian gerbils were orally infected with H. pylori B8 wildtype (WT) or its isogenic B8DcagY mutant (MUT) strain. Fourteen months thereafter, either H. pylori WT and MUT strain could be re-isolated at comparable loads from the stomach in 66.7% and 100% of infected Mongolian gerbils, respectively. In H. pylori WT and MUT infected animals, gastric pathology was characterized by lymphoid aggregates in the antrum (100% with either strain) and corpus (100% and 33.3%, respectively), erosions (100% and 83.3%, respectively), ulcers (75.0% and 0%, respectively), parietal cell atrophy (100% and 33.3%, respectively), and metaplastic changes (90.9% and 33.3%, respectively), gastritic cystica profunda (75.0% and 33.3%, respectively), and focal dysplasia (25.0% and 0%, respectively) ( Fig. 1, Table 1). After the establishment of a severe pangastritis in H. pylori WT strain infected animals, gerbils developed significant increases in pH values from 1.5 to 4 (hypochlorhydria) and plasma gastrin concentrations (hypergastrinemia) (refer to [9]).
Of note, significant histopathological signs of inflammation could not be observed in hematoxylin and eosin (H&E) stained paraffin sections derived from the small and large intestinal tract of infected animals (not shown). To further assess quantitative pro-inflammatory immune cell responses in the colonic mucosa of infected Mongolian gerbils, we performed in situ immunohistochemical stainings of colonic paraffin sections with an antibody against CD3 to visualize T lymphocytes. H. pylori WT infected gerbils displayed significantly higher numbers of T lymphocytes in their colonic mucosa as compared to MUT infected or naïve animals 14 months post infection (p.i.) (Fig. 2).
Distinct Microbiota Changes in the Large, but not small Intestinal Tract of Mongolian Gerbils Following Longterm H. pylori Infection We next performed a comprehensive cultural survey of the microbiota composition in the entire GI tract of long-term H. pylori infected Mongolian gerbils. Fourteen months following H. pylori WT or MUT infection, the total bacterial load increased by almost two orders of magnitude in the stomach lumen due to increased numbers of Lactobacillus spp. as compared to uninfected controls, but only reached statistical significance in the H. pylori WT group (p,0.01) (Fig. 3). Interestingly, only in 30-40% of naïve gerbils Gram-negative species such as E. coli, Proteus sp. and Bacteroides/ Prevotella spp. could be cultured from the gastric lumen (with a median of ,10uCFU per g for either bacterial species). During 14 months of long-term infection with the H. pylori WT or MUT strain, however, a trend towards higher gastric Gram-negative commensal loads was observed (approximately four and two orders of magnitude, respectively; not significant, n.s.) when compared to uninfected gerbils (Fig. 3).
In proximal parts of the small intestine such as duodenum (Fig. 4) and jejunum (Fig. 5) no differences in the luminal microbiota composition of infected versus non-infected gerbils could be observed. In the ileal lumen of H. pylori WT infected animals, however, lower Lactobacillus spp. numbers (approximately 1.5 orders of magnitude; p,0.05) were detected as compared to naïve or with the H. pylori MUT strain infected gerbils at 14 months p.i. (Fig. 6).
Strikingly, in the distal compartment of the GI tract, such as the caecum (Fig. 7) and colon (Fig. 8), H. pylori WT, but not MUT strain infected gerbils harbored significantly higher E. coli and Enterococcus spp. loads (1.0-1.5 orders of magnitude; p,0.01) 14 months following infection when compared to naïve animals. In addition, numbers of Bacteroides/Prevotella spp. were up to 100 times higher in colonic lumen of H. pylori WT infected animals as compared to naïve controls (Fig. 8E). Notably, no significant differences in microbiota composition could be observed in Mongolian gerbils suffering from distinct gastric immunopathology 14 months following infection, irrespective whether H. pylori WT strain could be re-isolated or not at necropsy (not shown).
We finally applied molecular methods for detection of fastidious or uncultivable bacterial species in the entire GI tract. Genetic fingerprints generated by PCR-based denaturing gradient gel electrophoresis (DGGE) and subsequent sequence analyses revealed that most prominent bacterial DNA bands occurring in caecal and colonic specimens 14 months following H. pylori WT, but not MUT strain infection referred to enterobacteria (Fig. 9). Thereby the increased abundance of commensal E. coli upon longterm H. pylori WT infection as detected by culture could be independently confirmed. Strikingly, Akkermansia, an uncultivable species involved in mucus degradation, could exclusively be detected in the caecal and colonic lumen of long-term H. pylori WT, but neither MUT infected (not shown) nor naïve gerbils ( Fig. 9). Interestingly, genetic fingerprints from luminal samples taken from either stomach, duodenum, jejunum or ileum were comparable in H. pylori WT, MUT strain infected and naïve gerbils, respectively (not shown), further supporting results derived from culture. Taken together, long-term H. pylori WT infection of Mongolian gerbils leads to distinct quantitative as well as qualitative changes of the microbiota composition within the uninflamed distal, but not proximal inflamed GI tract.

Discussion
In the study presented here we performed a comprehensive survey of microbiota changes along the entire GI tract in Mongolian gerbils following long-term infection with H. pylori for 14 months comparing B8 wildtype with its isogenic cagY mutant strain defective in the type IV secretion system. Most notably, we for the first time extended the microbiota analyses to locations distal ''classical'' sites of inflammation (such as stomach and duodenum) to the lower small and large intestines. Our presented data are well in line with our studies in other rodent inflammation models revealing that acute and chronic inflammation of the small as well as large intestinal tract in mice is associated with distinct shifts of the microbiota composition towards overgrowth with commensal E. coli, Bacteroides/Prevotella spp. and enterococci at sites of inflammation [19,20,21,22,23]. Given that in the presented study the significant microbiota shifts were exclusively detected following infection with the H. pylori WT strain able to translocate the effector protein CagA into the host cells, it is somewhat surprising, though, that at site of gastric inflammation no statistical significant differences in microbial colonization densities could be observed. As we could show in previous studies [9,10], H. pylori induced gastritis in antrum mucosa of Mongolian gerbils is cagPAI-    independent, whereas a severe inflammation of the corpus mucosa is induced via a cagPAI-dependent mechanism. Only H. pylori WT infected gerbils developed a pangastritis with a complete atrophy of the parietal cells resulting in a significant up-rise of the intragastral pH value (hypochlorhydria) followed by hypergastrinemia. Moreover, CagA translocation and hypochlorhydria have been shown to be pivotal prerequisites for H. pylori induced longterm sequelae in humans such as atrophic gastritis, dysplasia or gastric adenocarcinoma [24,25,26,27]. Furthermore, most of these animals also developed a mucous gland metaplasia producing atypical mucus. This is in sharp contrast to H. pylori MUT strain infected gerbils that do not develop such sequelae. One must take into account, however, that the observed lack of statistical significant differences in gastric bacterial colonization densities might be due to the relatively high standard deviations in either group since there was a trend towards four orders of magnitude higher (median) E. coli and Bacteroides/Prevotella spp. loads in H.
pylori WT, but not MUT strain infected gerbils at 14 months p.i. when compared to naïve animals. In a previous molecular study of the gastric microbiota composition we could demonstrate that E. coli and Bacteroides/ Prevotella spp. were exclusively abundant in stomachs of mice 8 weeks following H. pylori infection, but absent in uninfected controls [15]. A possible explanation for the gastric abundance of commensals that are usually restricted to the intestines could be that following chronic H. pylori infection the ''acidic barrier'' breaks (and thus the pH within the stomach lumen increases). Subsequently Gram-negative species such as E. coli and Bacteroides/ Prevotella spp. perorally ingested by the coprophagous rodents via their own feces have an advantage to grow in the formerly hostile environment and win the intraluminal competition with other commensal bacteria for niches and nutrients in the stomach. In contrast to mice, however, Mongolian gerbils tend not to coprophage their own feces if supplied with excess food. Therefore, gerbils kept under SPF conditions will not ingest intestinal commensals. This would explain the lack of significant abundance of Gram-negative intestinal commensals in the stomach of gerbils here.
In the past some studies attempted to profile the microbiota of H. pylori-infected Mongolian gerbils at sites of inflammation [16,17,18,28]. Yin et al. previously showed that gerbils in which the pathogen could still be re-isolated 12 weeks following H. pylori infection harbored higher Bacteroides, Enterococcus and Staphylococcus spp., but tended to harbor lower lactobacilli loads in their inflamed stomach and duodenum as compared to animals having expelled the pathogen in the meantime [16]. Furthermore, the gastric and duodenal microbiota composition in H. pylori-negative gerbils following infection appeared to be similar to naïve, uninfected animals. The authors concluded that the abundant species had adapted to the H. pylori induced changes of the intraluminal milieu at site of infection and inflammation, whereas the lactobacilli population had not [16]. Whether the lower lactobacilli densities was a consequence of H. pylori infection (due to milieu changes and lack of adaptation) or had initially rather facilitated the pathogen's establishment in the stomach cannot be answered. In contrast, in our study 14 months following H. pylori infection, gerbils exhibited even significantly (two orders of magnitude) higher lactobacilli and total bacterial loads in the stomach as compared to uninfected animals most likely as a consequence of the hypochlorhydria and, hence, increased pH within the gastric lumen. In addition, a different repertoire of expressed b-defensins [29] and/or abundance of B lymphocyte subsets [30] in lymphoid aggregates of H. pylori-infected as compared to naïve gerbils might have contributed to the observed differences in gastric lactoballi loads. Whereas in our study gerbils had been infected for over one year, Yin et al. had performed their analyses 12 weeks p.i. Of note, the microbiota changes in the interplay with pathogen and host responses are a dynamic system and substantially change over time. Given that here we have not assessed the commensal microbiota composition to an earlier time point, results derived following short-term H. pylori infection might differ for several (patho-) physiological reasons. In our study, applied molecular methods (such as PCR-based DGGE analysis with subsequent sequencing of DNA bands of interest) not only confirmed culture-based results (e.g. enterobacterial such as E. coli overgrowth). Strikingly, Akkermansia, an uncultivated species involved in mucus degradation [31], could be detected in the caecum and colon of H. pylori WT, but not MUT strain infected or naïve gerbils. This suggests that the accumulation of Akkermansia is driven by H. pylori induced inflammation and, hence, depends on a functional Cag pathogenicity island. Thus, intestinal abundance of bacterial species further compromising host barrier function by mucus degradation, might be one important puzzle piece in the complex pathogenesis of H. pyloriinduced sequelae, also in humans.
Overall, it is most surprising, however, that in our study even though the H. pylori induced immunopathology was prominent within the stomach, the microbiota shifts became exclusively overt in the large and not small intestinal tract where no mucosal histopathological changes of the intestinal mucosa could be observed. Interestingly, we were able to detect increased numbers of CD3+ cells in the large (but not small) intestinal mucosa and lamina propria in long-term H. plyori-WT, but not MUT strain infected Mongolian gerbils. It is tempting to speculate that recruited T lymphocytes might subsequently induce the release of cytokines and chemokines, which in turn promote further inflammatory responses leading to a compromised epithelial barrier function. As a consequence, a broken epithelial barrier might lead to changes within the intraluminal milieu subsequently promoting the luminal growth of intestinal commensals such as Gram-negative species as we could show in murine acute and chronic inflammation of the small as well as large intestinal tract [19,20,21,22,23]. In addition, one might speculate that via the neuroendocrine route, for instance, mediators (e.g. gastrin, somatostatin, cytokines, low proton concentration, among other factors) released in the stomach during H. pylori infection and induced immunopathology [32,33] exert their effect in the lower intestinal tract. Of note, patients suffering from H. pylori infection and hypochlorhydria occasionally complain about rather unspecific intestinal symptoms such as irregular bowel movements, flatulence or abdominal pain of otherwise unknown origin [34,35], which might be attributed to bacterial microbiota changes (e.g. overgrowth of some commensals and disappearance of others from the complex bacterial community) in the lower intestine during the disease process.
Taken together, long-term infection of Mongolian gerbils with an H. pylori WT strain displaying an intact type IV secretion system leads to distinct shifts of the microbiota composition in the distal uninflamed, but not proximal inflamed GI tract. Hence, H. pylori induced immunopathogenesis of the stomach, including hypochlorhydria and hypergastrinemia, might trigger large intestinal microbiota changes whereas the exact underlying mechanisms need to be further unraveled.

Animals and Infection Experiments
Outbred Mongolian gerbils from our own breeding colony (Max von Pettenkofer-Institute, LMU Munich, Germany) were bred and maintained under specific pathogen free (SPF) conditions and housed in SEALSAFE IVC cages (H-Temp, Tecniplast, Hohenpeissenberg, Germany) in an air-conditioned biohazard room (room temperature, 2365%; 12/12 hours light/dark cycle) with free access to a commercial gerbil diet (ssniff Gerbil, SSNIFF, Soest, Germany) and sterile tap water. Female animals at the age of 8-12 weeks were challenged orogastrically three times over five consecutive days with approximately 10 9 viable H. pylori B8 wildtype or H. pylori B8DcagY mutant strain. Age-matched control animals were incubated with identical volumes of sterile Brucella broth alone. Animals were sacrificed after defined time of infection (14 months) by isofluran treatment (Abbott, Germany). Gastrointestinal samples from each Mongolian gerbil were collected in parallel for immunohistological, microbiological, and molecular analyses. The stomach was opened along the great curvature, and the gastric tissue conserved separately in antrum and corpus as previously described [10]. In addition luminal contents of stomach, duodenum, jejunum, ileum, caecum, and colon were collected under sterile conditions in sterile PBS and immediately stored on ice.

Histopathology and Immunohistochemistry of Gastrointestinal Tissues
Histopathological changes were determined in GI tissue samples immediately fixed in 5% formalin and embedded in paraffin. Sections (5 mm) were stained with H&E and examined by light microscopy (100 x and 400 x magnification). In situ immunohistochemical analysis of paraffin sections taken from the colon was performed as described previously [38]. A primary antibody against CD3 (#N1580, Dako, Denmark, dilution 1:10) was used to visualize T lymphocytes in the colonic mucosa in situ (200 x magnification).

Cultural Analysis of the Luminal Gastrointestinal Mícrobiota
Cultural analysis of the intestinal microbiota along the entire GI tract was performed as described previously [20]. Luminal contents taken from stomach, duodenum, jejunum, ileum, caecum, and distal colon were resuspended in PBS, weighted, and 100 ml aliquots of serial dilutions plated onto solid media (Oxoid). Bacteria were grown at 37uC for 2 days under aerobic or for 4 days under anaerobic conditions, and total numbers determined by colony counting on Columbia blood agar. Bile esculin, McConkey, anaerobe 5% sheep blood agar supplemented with kanamycin and vancomycin, and Rogosa (Merck) media were used for quantitative identification of enterococci, enterobacteria (such as E. coli and Proteus spp.), Bacteroides/Prevotella spp., and lactic acid bacteria (mainly lactobacilli), respectively. Bacteria were subcultivated and further investigated by Gram-staining and by biochemical analysis with the API20E, API50 CH, and API Rapid ID 32A systems (Biomérieux) and confirmed by sequence analyses in cases of inconclusive identification. Results were expressed as colony forming units (CFU) per g of luminal content. The detection limit of viable pathogens was <25 CFU per g.

Molecular Microbiota Analysis
Molecular detection of fastidious or uncultivable bacterial communities in stomach, duodenum, jejunum, ileum, caecum, and distal colon was performed as previously described [20]. Respective luminal contents were removed, resuspended in PBS, and centrifuged (16,000 g/10 min/4uC). The sediment was resuspended in 0.5 ml of lysis buffer (500 mM Tris (pH 9.0), 20 mM EDTA, 10 mM NaCl, 1% SDS) and incubated with proteinase K (2 mg/ml; Sigma-Aldrich) for 1 h at 56uC. After bead beating with zirconium-silica beads (0.2 mm), the total DNA isolated by phenol-chloroform extraction served as template for PCR amplification.

Statistical Analysis
Medians and levels of significance were determined using Mann-Whitney-U Test. Two-sided probability (P) values #0.05 were considered significant. Experiments were repeated twice.