Introduction

A central paradigm in the chronic inflammation-driven pathogenesis of Helicobacter pylori-associated gastric cancer is the organism’s inherent ability to establish successful colonization niches in the gastric mucus and also it’s ability to adhere to the gastric epithelium [1]. A complex interplay exists between Hp and the host’s gastric mucosa leading to persistent Hp colonization within the stomach often for a lifetime [1, 2]. The gastric lumen-surface epithelial interface is characterized by its’ acidic pH, lack of nutrients, high viscosity mucus, continuous clearance of luminal contents by peristaltic action, epithelial cell turnover, infiltrating inflammatory cells and the host’s innate and adaptive immune responses [3,4,5]. Upon gastric infection, Hp encounters the protective mucous layer that is composed primarily of mucins and many other glycoproteins [3,4,5]. Mucins are high molecular mass, heavily glycosylated, oligomeric substances secreted on the cell surface and associated with glycoproteins present within the mucous and the surface epithelium; nearly 21 different mucin forms have been identified in humans and to a large extent in mice [4,5,6]. The secreted mucins, MUC5AC and MUC6, and the cell surface mucin, MUC1, form the major gastric mucins in the human stomachs [4, 5], and have similar expression profiles in mice [7]. However, in the diseased stomach secondary to Hp-induced mucosal alterations, de novo expression of intestinal mucins namely MUC2, MUC4, MUC5B and MUC13 has been documented in humans and animal models [8,9,10,11,12]. MUC5AC/Muc5ac is the major mucin in the stomach secreted by mucous cells on the surface foveolar epithelium of humans and mice and is localized along the adherent mucous layer [7]. Muc6 is primarily expressed within deeper glands and mucous neck cells, whereas Muc1 is primarily distributed along the foveolar cell surface epithelium mimicking the Muc5ac expression pattern and is also expressed minimally within mucous neck cells [10, 12, 13].

In mice and Mongolian gerbils, H. pylori mostly resides within 25 µM of the mucosal surface with high densities of motile organisms in the juxta-mucosal region (10 µM distance) where the pH gradient is less acidic than the luminal space [14]. This site predilection is reflective of the close interaction between Hp adhesins with host mucins and other cell surface receptors [15]. Blood group fucosylated antigens such as Lewis a and b (Lea and Leb) and H-type antigens are known to be highly expressed in epithelial cell surfaces and associated with gastric mucins, chiefly MUC5AC [16]. Interestingly, Leb is the major receptor for one of the best characterized adhesins of Hp, namely the blood group binding antigen, BabA (1, 2 3, 15−17). As MUC5AC/Muc5ac is a major binding factor for H. pylori in the stomach [3, 17, 18], we sought to further investigate the role of Muc5ac in H. pylori pathogenesis in a mouse model deficient in Muc5ac. Briefly, the study was conducted using a well-characterized chronic Hp SS1 infection model in Muc5ac−/− and WT mice to compare the differences in gastric pathology scores, colonization densities and gastric cytokine responses at two post infection time points, specifically at 16 and 32 weeks post infection (wpi).

Methods

Experimental mice and bacterial infection protocol

Muc5ac−/− mice were obtained from the National Institute of Environmental Sciences (NIEHS) where they had been developed by deleting exons 21–31, the exons coding for the major glycosylation sites in Muc5ac and maintained on a C57/BL6-129 Ola background for 15 generations. The genotype of these animals was confirmed using PCR, Northern blot and stomach Muc5ac immunohistochemistry analyses prior to shipping these mice to MIT. AT MIT, the mice were re-derived by embryo transfer prior to experimental usage. Five-week-old male and female C57BL/6-129 0la mice (referred to as wild type or miceWT) were obtained from Taconic Farms (Germantown, NY). The WT mice as well as Muc5ac−/− mice were housed under specific pathogen-free (SPF) conditions (free of Helicobacter spp., Citrobacter rodentium, Salmonella spp., endoparasites, ectoparasites, and exogenous murine viral pathogens) in an AAALAC accredited facility. All animal housing conditions and experimental protocols were approved by the MIT Committee on Animal Care. The various groups of mice (Table 1) at 7 weeks of age were orally gavaged with either Hp strain SS1 (Hp SS1) as described earlier [19, 20] every other day for 6 days with three doses of 0.2 ml (∼2 × 108 organisms) or three sham doses of 0.2 ml sterile Brucella broth.

Table 1 Animals and experimental groups
Full size table

Gross necropsy and tissue collection protocols

After dosing, the mice were euthanized at 16 wpi and 32 wpi by CO2 euthanasia followed by gross necropsy and tissue collection procedures as previously described with slight modifications [21, 22]. Briefly, the stomach was opened along the lesser curvature and evaluated for any gross abnormalities. Representative longitudinal sections of the stomach extending from the squamous forestomach to the pyloric duodenum were collected and fixed in 10% neutral buffered formalin. These fixed sections were then processed by standard paraffin embedding, 5 µM sectioning and hematoxylin and eosin (H&E) staining protocols. Representative sections of the stomach corpus and antrum were also collected for RNA/DNA extraction as described below. Additionally, archival stomach sections from the original WT and Muc5ac−/− male mice at 10–12 months of age were obtained from NIEHS. The corpus and antrum from these mice were also evaluated to compare the gastric phenotypes of WT and Muc5ac−/− mice housed under different environmental conditions.

Histopathological evaluation

H&E slides of the stomach were graded by a comparative pathologist (S.M.) blinded to sample identity separately for both the corpus and antrum/pylorus. In the corpus, the categories scored were inflammation, epithelial defects, atrophy, hyperplasia, pseudopyloric metaplasia, dysplasia, and mucous metaplasia on an ascending scale from 0 to 4 based on severity using criteria previously described with slight modifications [21, 22]. Gastric histological activity indices (GHAI) were generated by computing the sum of all individual sub-categorical scores except hyalinosis which can be a spontaneous nonspecific alteration in murine stomach. Adapting the same grading scheme the antrum/pylorus was also scored for inflammation, epithelial defects, hyperplasia and dysplasia with the sum of all these categories represented as total antral pathology scores [21, 22]. Briefly, inflammation was evaluated on the basis of amount and extent of leukocyte infiltration in the mucosa and submucosa of the stomach. Epithelial defects were scored on the basis of the extent of surface epithelial degeneration, tethering, erosions to full thickness ulcerations. Hyalinosis is a nonspecific change characterized by the accumulation of brightly eosinophilic droplets and/or extracellular crystals was scored the extent of gastric mucosal involvement [21, 22]. Oxyntic atrophy was defined as a reduction in the mass of chief and parietal cells in the gastric corpus and graded on the extent of loss via cell loss and/or metaplastic transformation. Epithelial hyperplasia refers to expansion of the gastric mucosal thickness by proliferation of surface foveolar-type epithelium and/or antral-type glandular units and appropriately graded in relation to the extent of oxyntic loss [21, 22]. Mucous metaplasia is usually associated with parietal and chief cell loss leading to expansion of gastric mucous neck cells (PAS+/Alcian blue/TFF2 + foamy cells with marginated nucleus resembling Brunner’s glands) secreting a mixture of neutral and acidic mucins in the gastric corpus mucosa [21, 22]. Pseudopyloric metaplasia is a preneoplastic change defined as replacement of the oxyntic mucosa by glands resembling antral phenotype with cells being more columnar and lacking typical cytoplasmic granules of oxyntic cells as well as the absence of cytoplasmic mucous appearance of mucous metaplasia. SPEM is considered to be a molecular equivalent term for pseudopyloric metaplasia due to expression of Trefoil factor 2 (Tff2) but is also seen in murine mucous metaplasia [21, 22].

Gastric epithelial dysplasia is histologically an unequivocal neoplastic change without evidence of stromal invasion occurring in sessile, flat, depressed or elevated/polypoid mucosal lesions and is defined by both architectural abnormalities (haphazard glandular arrangement, loss of vertical orientation, back-to-back gland associations, branching, infolding, and piling up of cells) and cytological atypia (cellular pleomorphism, anisocytosis, anisokaryosis, ill-defined cellular junctions, loss of nuclear polarity, pencil or cigar shaped nuclei, hyperchromatic nuclei, increased nuclear–cytoplasmic (N–C) ratio, visible mitosis, bizarre mitotic figures) [21, 22]. The term gastric adenoma was restricted to well-demarcated or circumscribed, polypoid or raised nodular lesions comprising of tubular and/or villous structures, lined by dysplastic epithelium. Dysplasia/neoplasia was graded on its extent and severity of changes as follows: score of 0 for normal or mild simple epithelial hyperplasia with no atypia; score of 1 for one or few epithelial hyperplastic foci with mild architectural atypia (equivalent to indefinite for dysplasia or reactive epithelium); a score of 2 for atypical hyperplasia comprising of coalescing proliferative epithelial lesions with glandular architectural abnormalities (equivalent to indefinite for dysplasia); a score of 2.5 for low-grade dysplasia/intraepithelial neoplasia or low-grade adenoma and all of these proliferative lesions were defined by severe glandular architectural abnormalities and borderline cytological atypia; a score of 3 is equivalent to high-grade dysplasia/ intraepithelial neoplasia or a high-grade adenoma characterized by severe architectural and cytological atypia; a score of 3.5 for intramucosal invasive neoplasia (intramucosal carcinoma) characterized by high-grade dysplastic lesions with unequivocal invasion into the lamina propria or muscularis mucosa (dysplasia score of 3.5; and a score of 4 for submucosal invasive neoplasia (submucosal carcinoma) for true unequivocal invasion into the gastric submucosa or beyond. This 4-point scoring system was adapted to reflect the limitations of the mouse models [21, 22] and correlated to human gastric dysplasia diagnosis and grading schemes [23].

Immunohistochemistry

Six representative formalin-fixed and paraffin-embedded stomach histological sections (3 M, 3 F) from all the groups at 32 wpi were used to perform a panel of specific immunohistochemical stains using the following rabbit polyclonal or mouse monoclonal antibodies to either human- (reactive with mouse) or mouse-specific antigens: (a) rabbit origin- MUC5AC and MUC2 (Santa Cruz Biotechnology, Dallas, TX, USA), TFF2 (in-house rabbit polyclonal TFF2 antibody referenced in 21), MUC1, Clusterin and TFF1 (Life Span BioSciences, Seattle, WA, USA), NSE (Neuron-specific enolase) (Thermo-Fischer Scientific Inc., Waltham, MA, USA), Chromogranin A (Novus Biologicals, Littleton, CO, USA and AbCam, Cambridge, UK), Cytokeratin 20 (CK20) (AbCam, Cambridge, UK); (b) mouse origin: S100 (BioCare Medical, Pacheco, CA, USA). Additionally, gastric mucous neck cells were identified using fluorescein isothiocyanate (FITC)-conjugated Griffonia (Bandeiraea) simplicifolia lectin II (GSL II) and slides were mounted with Vectashield plus DAPI (4,6-diamidino-2-phenylindole) (Vector Laboratories, Burlingame, CA, USA). For chromogranin A peptide blocking immunohistochemistry, to identify nonspecific staining if any, the chromogranin primary antibodies were mixed with varying concentrations of the full-length human (full length) chromogranin A peptide (ab138920) AbCam, Cambridge, UK) for 1 h at room temperature prior to following routine immunostaining protocols. For routine immunohistochemistry, unstained paraffin embedded and sectioned stomach slides were incubated at 60 °C for 30 min, followed by de-paraffinization and rehydration prior to staining with specific antibodies as described earlier with slight modification [21]. The antigens were retrieved at a low or high pH, depending on the antibody, by incubation for 45 min at 95 °C and the slides were stained sequentially using an automated stainer (Lab Vision Autostainer 360; Thermo Fisher Scientific, Waltham, MA, USA) incorporating a 10-min Ultravision peroxidase block, for a 30-min Rodent block R, and primary antibodies for 30 min. This was followed by appropriate species-specific secondary polymer-based antibodies, and subsequent development with diaminobenzidine (DAB), counterstained with hematoxylin. Slides were dehydrated and cleared, and then cover slipped for qualitative assessment of the expression pattern of the various markers.

Quantitation of Hp SS1

At necropsy, gastric contents were removed by rinsing with sterile PBS and 1-cm segments of piece of gastric tissue containing the corpus and antrum for isolation of RNA or DNA were collected and frozen in liquid nitrogen immediately and stored at –70 °C prior to use. DNA from the samples was extracted using the High Pure PCR Template Preparation Kit (Roche, Applied Science, Indianapolis, IN). Colonization levels of Hp SS1 from the glandular portion of the stomach (antrum and corpus) were quantitatively measured using the primers and probe complementary to the Hp ureB gene qPCR in the 7500 FAST Real-time PCR System (Life Technologies) as described previously [20]. The copy numbers of the Hp genome as measured by qPCR were then normalized to μg of mouse chromosomal DNA using the 18S rRNA gene-based primers and probe mixture (Life Technologies) [19, 20].

Expression of select gastric genes

For relative mRNA quantitation of selected genes, total RNA the gastric tissue was prepared using Trizol reagent according to the manufacturer’s recommendations (Invitrogen, Carlsbad, CA). Five micrograms of total RNA from each sample was converted into cDNA using the High Capacity cDNA Archive kit (Life Technologies). Levels of Ifnγ, Tnfα, Il-1β, Il-17A, Il-10, Il-22 and Tff2 mRNA in the gastric tissues were examined by qPCR using commercial primers and probes (TaqMan Gene Expression Assays) in the 7500 FAST Real-time PCR System. Transcript levels were normalized to the endogenous control glyceraldehyde-3-phosphate dehydrogenase mRNA (Gapdh) [19, 20].

Statistical analysis

Semi-quantitative histopathological scores from the corpus and antrum for different sub-categories and cumulative GHAI scores were compared across groups by the Kruskal−Wallis one-way analysis of variance with Dunn’s posttest and between groups by the Mann−Whitney U test using Prism software (GraphPad, San Diego, CA) and graphs were represented as the median with the range. Quantitative data values generated for colonization levels of Helicobacter species and cytokine mRNA levels in the tissues were analyzed using the two-tailed Student’s t test. P values of <0.05 were considered significant.

Results

Muc5ac −/− mice are predisposed to spontaneous antro-pyloric epithelial proliferation that is independent of Hp infection

The gastric antrum/pylorus of the experimental mice were scored for inflammation, edema, epithelial hyperplasia, dysplasia, and a cumulative total antral pathology index score was generated for group and time-wise comparative analysis (Fig. 1a–d). Interestingly, uninfected Muc5ac−/− mice at 32 wpi (39 weeks of age) had significantly higher total antral pathology index (P = 0.0008) and dysplasia scores (P = 0.0019) compared to uninfected WT mice (Fig. 1b, d). The histopathological changes in the antrum and pylorus of these Muc5ac−/− mice included mild to moderate inflammation, epithelial defects including erosions and dilated glands, epithelial hyperplasia and progressive dysplasia resulting in various categorical lesions as enumerated in Table 2. These lesions included normal with no hyperplasia/dysplasia (1/12), simple epithelial hyperplasia with no dysplasia (2/12), indefinite for dysplasia (reactive epithelium and/or atypical hyperplasia) (7/12), low-grade adenoma (1/12) or high-grade adenoma (1/12) (Fig. 2g, h) compared to lack of similar range of proliferative and dysplastic lesions in the uninfected WT mice (Fig. 1e, f and Table 2). Within the large adenoma in the pyloric region of the Muc5ac−/− mouse, there were multiple characteristic micronodular areas of divergent morphological appearance resembling rosettes that were presumed to be foci of possible neuroendocrine differentiation (Fig. 1g, h, arrows). The antro-pyloric inflammation and proliferative changes were also observed in uninfected Muc5ac−/− mice at 16 wpi, but at a lesser magnitude with no significant dysplastic changes indicative of an age-related progression in hyperplasia and dysplasia (Fig. 1a, c).

Fig. 1
figure 1

Gastric antro-pyloric spontaneous pathology in Muc5ac−/− mice: Cumulative antrum/pylorus pathology index scores in the different groups of mice at 16 wpi (a) and 32 wpi (b). Antro-pyloric dysplasia score chart for different groups of mice at 16 wpi (c) and 32 wpi (d). (*) = p < 0.05, (**) = p < 0.01, (***) = p < 0.001, Uninf Uninfected, Hp Hp-infected. e, f Representative H&E images of the uninfected gastric antrum/pylorus at 32 wpi time point. e WT uninfected, low mag, normal. f WT uninfected, high mag, normal. g Muc5ac−/− uninfected, (low mag) showing a large pyloric adenoma with multinodular (arrows) pleomorphic appearance. h Muc5ac−/− uninfected: Higher magnification image of the micro nodules within the adenoma showing more basophilic neoplastic cells forming rosettes (neuroendocrine-like pattern) adjacent to dysplastic glandular epithelium (stars). Bars: e, g 400 µm, g 160 µm, h 80 µm

Full size image
Table 2 Distribution of various dysplastic lesions in the stomach of Muc5ac−/− and WT mice
Full size table
Fig. 2
figure 2

Immunohistochemical analysis of the antrum/pylorus for various mucins and trefoil factors: Panels show staining pattern of various markers in normal uninfected WT mice and polypoid adenoma in the antro-pyloric region of uninfected Muc5ac−/− mice at 39 weeks of age. Note lack of epithelial Muc5ac and Muc2 staining in Muc5ac−/− antrum/pylorus and increased staining intensity for Muc1, TFF2 and TFF2 within the adenoma/hyperplastic antrum/pylorus. Bars: 160 µm: All WT images. 400 µM: All Muc5ac−/− images. Arrows indicate micro adenomas with divergent differentiation within the large adenoma in the pylorus

Full size image

The various antral pathological alterations were also observed in the Hp-infected Muc5ac−/− mice at both time points but were of lesser degree, resulting in significantly lowered total pathology index scores at 32 wpi (P = 0.0467) as compared to their uninfected counterparts but with comparable median dysplasia scores though high grade adenoma was noted only in the uninfected cohort (Fig. 1b, d, Table 2). The WT mice (both infected and uninfected) did not show similar antro-pyloric proliferative and dysplastic changes as observed in the uninfected Muc5ac−/− mice at 32 wpi time point (Fig. 1, Table 2). At 16 wpi, Hp-infected WT mice also showed significantly (P = 0.009) higher total antral pathology index compared to uninfected WT but it was primarily due to mild extension of H. pylori-induced corpus inflammation and epithelial defects into the antrum (Fig. 1a). This was not significant at 32 wpi (Fig. 1b).

Given the fact that spontaneous gastric antro-pyloric lesions were noted in 10-month-old Muc5ac−/− mice (males only), we sought to rule out any possible institution-related environmental or microfloral effects in these mice by analyzing uninfected (Hp free) archival gastric tissues from mice maintained at NIEHS. We observed similar pathological alterations in the gastric antrum and pylorus of 11-month-old Muc5ac−/− mice (Fig. S1). Compared to WT (n = 9), Muc5ac−/− male mice (n = 11) developed significantly higher cumulative antrum/pylorus total pathology scores (Fig. S1A), with higher scores for inflammation, epithelial defects, hyperplasia and dysplasia (Fig. S1B, Table 2) including low-grade adenoma (1/11) and high-grade adenoma (1/11) (shown in Fig. S1, C&D). We did not attempt to further classify the nature of antro-pyloric hyperplasia and adenoma into sub-categories such as foveolar, pyloric gland and intestinal type using additional markers such as Muc5ac (for foveolar adenoma but not useful in Muc5ac−/− mice) or Muc6 (no reliable commercially available antibody that works on murine tissues) but nevertheless, on the basis of morphology and lack of Muc2 immunostaining, most of these proliferative epithelial lesions were considered to be of either foveolar or mixed foveolar and pyloric glandular phenotype. Overall, we identified an age-associated phenotype of gastric antro-pyloric hyperplasia and dysplasia/adenoma progression independent of environmental effects in association with loss of Muc5ac function in murine stomach.

Immunohistochemical expression profiles of mucins and trefoil factors in the gastric antrum/pylorus of WT and Muc5ac −/− mice

Given the importance of mucins and trefoil factors in gastric homeostasis and H. pylori pathogenesis [24,25,26,27,28], we assessed the immunohistochemical expression patterns of these proteins in relation to the spontaneous antro-pyloric lesions in Muc5ac−/− mice. Of note, similar proliferative lesions involving the antrum have been documented in Hp-infected Tff2−/− mice on a CF57BL background [21]. In addition, mice deficient in Tff1 have also been shown to develop spontaneous antro-pyloric adenomas and carcinomas [29]. The gastric antrum/pylorus of uninfected Muc5ac−/− mice did not express Muc5ac as expected, but showed moderate to strong expression of Muc1, Tff2 and Tff1 within the hyperplastic and neoplastic glandular and surface epithelium (Fig. 2). The increase in expression levels of these proteins was correlated with increased epithelial mass as a result of epithelial hyper-proliferation. It should be noted that the occasionally observed neuroendocrine-like distinct micronodular foci within one large adenoma in the pylorus did not stain with the markers, Muc1, Tff2, and Tff1, suggestive of altered differentiation within the tumor. Additional characterization of these presumptive neuroendocrine-like foci by immunohistochemistry using markers specific for neuroendocrine tumors as in humans [30] showed that these foci exhibited mild to moderate positivity with Neuron-Specific Enolase (NSE) (Fig. S2). With Chromogranin A immunostaining, these foci showed only nonspecific background extracellular staining as demonstrated by a Chromogranin A blocking peptide incubation protocol in Fig. S2. In addition, these neuroendocrine foci or rosettes and surrounding proliferative villo-tubular glands failed to react with the neuronal marker, S100 (not shown). Chromogranin A is typically present in most gastric neuroendocrine tumors [30] and its absence in our case was intriguing and may require additional testing but was not pursued due to tissue limitations and rarity of these neuroendocrine foci within gastric antro-pyloric proliferative lesions of Muc5ac−/− mice. However, cytokeratin 20 (CK20), an epithelial marker for intestinal epithelium/goblet cells as well for gastric foveolar epithelium, was not expressed within the antro-pyloric rosettes (NSE positive areas) of Muc5ac−/− stomach as shown in Fig. S2. Other than Muc2, which is not typically expressed in the normal stomach, all other mucins and trefoil factors displayed their typical immunostaining pattern in the antrum/pylorus of the WT murine stomach (Fig. 2), with the expression patterns of Muc5ac and Muc1 more or less similar in distribution. In addition, Hp infection did not alter the overall antro-pyloric expression patterns of mucins (Muc1, Muc5ac, Muc2) and trefoil factors (Tff1 and Tff2) in both WT or Muc5ac−/− mice (data not shown).

Hp infection led to attenuation of gastric corpus mucous metaplasia in Muc5ac −/− mice

Hp-infected WT and Muc5ac−/− mice at both 16 wpi and 32 wpi had significantly higher corpus GHAI scores than their uninfected counterparts (Fig. 3a, c) (WT Hp vs. WT uninfected, 16 wpi, 32 wpi: P = 0.0005, P = 0.0002/Muc5ac−/− Hp vs. Muc5ac−/− uninfected, 16 wpi, 32 wpi: P = 0.0013, P = 0.029). Representative images of the gastric corpus from various groups are depicted in Fig. 3e−h. However, there was no significant difference in corpus GHAI between the two infected groups at both time points, though Hp-infected Muc5ac−/− mice at both time points showed a trend towards lesser overall corpus pathology. Within each infected group there was also no appreciable gender differences on the severity of gastric corpus pathology. Gastric corpus mucous metaplasia was also significantly (P = 0.0012–0.0021) increased by Hp infection in WT mice as compared to their uninfected counterparts at both 16 wpi and 32 wpi (Fig. 3b, d). However, in the Hp-infected Muc5ac−/− mice, mucous metaplasia was significantly increased during Hp infection only at 16 wpi (Fig. 3b, P = 0.0033) when compared to uninfected Muc5ac−/− mice. Interestingly, at both time points, Hp-infected Muc5ac−/− mice had significantly lower median corpus mucous metaplasia scores as compared to Hp-infected WT mice (Fig. 3b, d) (16 wpi and 32 wpi; P = 0.0057 and P = 0.0016, respectively). There was an overall age-related increase in background inflammation and associated changes like atrophy and metaplasia in uninfected WT and Muc5ac−/− mice at 32 wpi as compared to 16 wpi, as reflected by the increased median GHAI scores at 32 wpi (Fig. 3a, c).

Fig. 3
figure 3

Gastric corpus pathology in Hp-infected WT and Muc5ac−/− mice: Charts showing corpus gastric histological activity indices (GHAI) at 16 wpi (a) and 32 wpi (c) and sub-categorical mucous metaplasia scores at 16 wpi (b) and 32 wpi (d). (*) = p < 0.05, (**) = p < 0.01, (***) = p < 0.001. Representative H&E images of the gastric corpus at 32 wpi: e WT, uninfected; f Muc5ac−/−, uninfected; g WT, Hp-infected, arrows indicate mucous metaplasia; h Muc5ac−/−, Hp-infected, arrows indicate metaplasia with dysplasia. Bar: 160 µm, all images. Hp infection in both WT and Muc5ac−/− gastric corpus showed prominent inflammation, oxyntic loss, metaplasia, hyperplasia and mild dysplasia as compared to their uninfected counterparts with a significant reduction in corpus mucous metaplasia in Hp-infected Muc5ac−/− vs. Hp-infected WT

Full size image

Immunohistochemical analysis of mucins and trefoil factors in the gastric corpus of WT and Muc5ac −/− mice and its correlation with Hp infection

As shown in Fig. 4a, strong Muc5ac immunopositivity was observed primarily in the surface and foveolar epithelial pits of the gastric corpus of uninfected WT stomach, with infrequent staining of the glandular neck region. The corpus of Muc5ac−/− mice, as expected, did not show any specific Muc5ac immunostaining, although mild nonspecific background was observed within the epithelial stroma (Fig. 4b). Muc5ac immunopositivity was seen in the apical foveolar pits of uninfected WT stomach and there was also a reduction in Muc5ac staining within foveolar pits of the gastric corpus overlying hyperplastic and mucous metaplastic regions of Hp-infected WT mice (Fig. 4c, d). Muc1 was highly expressed in foveolar pits and also moderately within deeper glands of the gastric corpus in both uninfected WT and Muc5ac−/− mice (Fig. 4e). In Hp-infected WT and Muc5ac−/− mice, Muc1 immunostaining was seen prominently within most hyperplastic foveolar and glandular regions (Fig. 4e, g) but was frequently reduced in areas of mucous metaplasia especially in the corpus of WT Hp-infected stomachs (Fig. 4h).

Fig. 4
figure 4

Immunohistochemistry of Muc5ac and Muc1 in the gastric corpus. Muc5ac IHC: a–d, Prominent Muc5ac expression in the gastric foveolar region of the WT uninfected control (a) and lack of expression in uninfected Muc5ac−/− mice at 32 wpi (b). Muc5ac expression in WT Hp-infected stomach in areas of pseudopyloric metaplasia or mild mucous metaplasia (c, MM mild) compared to a reduction in surface foveolar expression in areas of severe glandular mucous metaplasia (d, MM severe 32 wpi. Bars: 160 µm, all images. Muc1 IHC: e–h Muc1 expression is robust in the upper 1/3rd and lower 1/3rd of normal corpus (e, WT uninfected, Bar: 160 µm) and is enhanced in hyperplastic glands of Hp-infected WT and Muc5ac−/− gastric corpus (f, g, Bar: 400 µm). A slight reduction in intensity of Muc1 staining is seen in areas of glandular mucous metaplasia (arrows) as shown in the higher magnification of inset (h, Bar: 80 µm)

Full size image

Trefoil factors are closely associated with gastric mucosal biology, as determined by the ability of TFF1 to bind to MUC5AC [24] and by the increased expression of TFF2 in SPEM [21]. In both uninfected WT and Muc5ac−/− mice (Fig. 5a−d), Tff1 staining was primarily restricted to corpus surface foveolar epithelium. In Hp-infected WT and Muc5ac−/− mice there was a slight reduction in Tff1 immunoreactivity in areas where the metaplastic changes extended up to the surface epithelium. Tff2 expression in the WT uninfected stomach was observed within mucous neck cells of the corpus (Fig. 5e−h), basal aspects of antral/pyloric glands and within duodenal Brunner’s glands (shown in Fig. 3). The corpus of uninfected Muc5ac−/− mice (Fig. 5f, h) showed similar Tff2 expression patterns as seen in the WT uninfected stomach. In both the infected WT and Muc5ac−/− stomachs, there was an aberrant increase of Tff2 immunoreactivity within metaplastic (mucous and pseudopyloric types of metaplasia) and hyperplastic glands of the corpus (Fig. 5g, h).

Fig. 5
figure 5

TFF1 and TFF2 Immunohistochemistry in the stomach: a–d Tff1 expression is similar in the gastric corpus of both infected and control mice with localization to foveolar pits at 32 wpi. e–h TFF2 expression is increased in infected animals with strong positivity in metaplastic glands (mucous metaplasia and pseudopyloric type) in both WT and Muc5ac−/− mice compared to their uninfected controls at 32 wpi. Bar: 160 µm, all images

Full size image

GSII Lectin and Clusterin immunostaining confirms that gastric mucous metaplasia is inhibited in Hp-infected Muc5ac −/− mice compared to WT Hp-infected mice

We used Clusterin and GSL II immunostaining to characterize the gastric metaplasia observed in both WT and Muc5ac−/− Hp-infected mice. The lectin, Griffonia simplicifolia (GSL II), is a well-characterized marker of mucin and it labels the corpus neck cells and antral basal glands [21]. GSL II staining is increased following Hp infection in response to loss of oxyntic cells and expansion and/or transformation to mucous metaplasia [21]. In our study, GSL II lectin staining was restricted to a thin layer of corpus mucous neck cells in uninfected mice (Fig. 6, rectangular boxes), but was prominently increased in areas of corpus metaplasia following Hp infection in both WT and Muc5ac−/− mice (Fig. 6, middle and bottom panels). Interestingly, in metaplastic regions, the expression of GSL II expanded from the neck region into the hyperplastic foveolar epithelium as well as deep into the basal aspect of mucosa where chief cells are normally located. Within Hp-infected groups, gastric corpus GSL II lectin staining was more extensive and intense in the WT mice as compared to Muc5ac−/− and this correlated with higher levels of mucous metaplasia observed in the Hp-infected WT group (Fig. 6, bottom panels). GSL II labeling was also observed in a proportion of glands classified as pseudopyloric metaplasia on the basis of the typical morphology of these metaplastic tubules lacking a mucous filled cytoplasmic appearance (Figs. 6 and 3g, h).

Fig. 6
figure 6

GSL II-Lectin Fluorescent staining in the stomach. In uninfected gastric corpus of WT and Muc5ac−/− mice, GSL II fluorescence (FITC, green) was observed within the mucous neck cell region only (white rectangle box) at 32 wpi (top two images). In both the infected groups, there is increased GSL II positive staining throughout the mucosa though in WT corpus (left middle and bottom images), the proportion of glands involved and their intensity of staining is greater than that of Muc5ac−/− corpus (right middle and bottom images). Bottom two images are higher magnification images from the boxed square regions. Blue—DAPI nuclear counter stain

Full size image

SPEM is a preneoplastic metaplastic phenotype that is relevant in mouse gastric cancer models. Mice do not typically manifest intestinal metaplasia as noted in humans during the Correa stepwise cascade of gastric carcinogenesis [31]. Clusterin is a recently identified marker of SPEM in murine models [31] and we sought to explore its expression levels within the different experimental groups. Clusterin labeling was mostly absent or sparse within the gastric corpus epithelium of both the uninfected WT and Muc5ac−/− mice, but was prominently observed multifocally within the metaplastic glands of the Hp-infected WT and Muc5ac−/− mice (Fig. 7). However, the degree of Clusterin positivity within glands was variable depending upon the type of metaplasia (i.e. pseudopyloric metaplasia vs. mucous metaplasia or mixed forms). A strong positivity for Clusterin was noted in a significant proportion of metaplastic glands of pseudopyloric type, and only infrequently observed within glands of the mucous metaplasia type (Fig. 7). This correlated with an appreciable decrease in overall clusterin staining (SPEM transformation) in WT-infected stomachs in which mucous metaplasia was more prominent as compared to infected Muc5ac−/− mice.

Fig. 7
figure 7

Immunohistochemistry of Clusterin in stomach. In both the uninfected WT and Muc5ac−/− gastric corpus epithelium, clusterin is rarely detected except patchy staining of a few foveolar cells at 32 wpi. In Hp-infected animals, Clusterin staining is enhanced in a significant proportion of the glands showing pseudopyloric metaplasia (Muc5ac−/−, low and high magnification, middle and bottom right images respectively) and mostly absent in areas of severe mucous metaplasia (WT stomach, middle and bottom left images). A subset of glands undergoing mucous metaplasia also express the SPEM marker, Clusterin (arrows, left bottom image), Bars: top four images—160 µm, bottom two images—80 µm

Full size image

Gastric Hp colonization dynamics in WT and Muc5ac −/− mice

At 16 wpi, the levels of gastric H. pylori, as measured by qPCR in both male and female Muc5ac−/− mice, were significantly (P < 0.045) higher than their male and female WT counterparts (Fig. 8). Additionally, the levels of gastric Hp in male Muc5ac−/− male mice were significantly higher than female Muc5ac−/− mice at 16 wpi but the differences were not statistically significant between infected WT males and WT females. At 32 wpi, there was no significant difference in gastric Hp levels between genotypes or genders.

Fig. 8
figure 8

Gastric Hp colonization levels. Bacteria abundance was determined in gastric tissues by qPCR using a specific primer set for Hp DNA at 16 and 32 weeks post-infection (wpi). Selected p values are displayed in the panels. WT wild type, M male, F Female, Muc5 Muc5ac−/− mice, Hp Hp SS1 infection, C uninfected controls

Full size image

Hp-associated gastric gene expression

Gastric cytokines, including Tnfα, Ifnγ, Il-1β, Il-17A, Il-22, and Il-10, have been shown to play an important role in influencing histological manifestations of Hp infection. In general, infected mice had significantly higher mRNA levels of these cytokines compared to genotype-matched uninfected mice (P < 0.05). The H. pylori-associated increase in Tnfα mRNA levels in Muc5ac−/− mice were significantly lower than that of gender-matched WT mice at 16 wpi (P = 0.017 and P = 0.036, male and female respectively) and 32 wpi—males only, P = 0.025). Additionally, there was significantly lower gastric Il-17A mRNA levels at 16 wpi in infected Muc5ac−/− males (P = 0.025) compared to their Hp-infected WT counterparts. The mRNA levels of gastric Ifnγ, Il-1β, Il-10 and Il-22 did not show any significant differences between the infected genotypes. Trefoil factor 2 (Tff2), another marker for SPEM, was comparable among the groups (Fig. 9, panels 1 and 2) though there was an enhancement of gastric Tff2-positive areas in infected mice as compared to the uninfected mice in both genotypes (Fig. 5e−h).

Fig. 9
figure 9figure 9

(Panels 1 and 2): mRNA expression levels of inflammatory murine cytokines in the gastric corpus. Differential expression of inflammatory cytokines was determined in gastric tissues by qPCR using gene-specific primer sets at 16 and 32 wpi. Selected p values are displayed in the panels. WT wild type, M male, F Female, Muc5 Muc5ac−/− mice, Hp Hp SS1 infection, C uninfected controls

Full size image

Discussion

This study was designed to elucidate the role of Muc5AC in Hp pathogenesis utilizing Muc5ac−/− mice in a well-established chronic Hp infection murine model. Our results have illustrated several interesting observations. First, Muc5ac−/− mice are spontaneously predisposed to develop an Hp independent gastric phenotype of antro-pyloric inflammation, hyperplasia and dysplasia, including development of adenomas. Further, Muc5ac−/− mice, in general, have slightly attenuated Hp-induced corpus pathology compared to WT mice, in particular a significant reduction in oxyntic mucous metaplasia. At the cytokine mRNA level, most notable in males, gastric TNFa and IL-17A levels were significantly higher in Hp-infected WT B6-129 mice as compared to infected Muc5ac−/− mice, which correlated with previously observed trends for these cytokines in mouse models of Hp- associated gastritis (19, 20, and 21). Interestingly, Muc5ac−/− mice (predominantly males) during the initial course of infection (16 wpi) displayed significantly higher gastric Hp colonization levels and an associated significant reduction in gastric corpus mucous metaplasia compared to Hp-infected WT mice, but these differences in colonization levels were not noted at the end of the study (32 wpi).

Trefoil factors are small secreted proteins in mammals that share a conserved trefoil domain of 42–43 amino acids in humans (TFF1, TFF2 and TFF3) and their homologs (Tff1, Tff2 and Tff3) in mice [5, 9, 24,25,26,27]. In the stomach, TFF1, also known as ps2/TFF1, is expressed in the superficial foveolar epithelium of the human corpus and antrum and binds to MUC5AC, the primary mucin in this location [24]. TFF1 interacts with both the gastric MUC5AC and intestinal MUC2 through their von Willebrand factor C (VWFC) cysteine rich domains, and this interaction is largely responsible for the gastrointestinal protective effect of TTF1 [27]. In biopsies from children, it was also shown that TFF1 colocalized with Hp in the superficial mucous layer of the surface epithelium and within pits of the gastric mucosa, thus mimicking the localization of MUC5AC [24, 28]. The deficiency of Tff1 in mice has been reported to cause the development of spontaneous antro-pyloric adenomas and carcinoma [29]. TFF2 is expressed in the mucous neck cells in the fundus or corpus, in the basal glands of the antrum and pylorus and in Brunner’s glands of the duodenum [21, 26, 32]. Mice deficient in Tff2 have accelerated progression to premalignant dysplasia in the antrum/pylorus of Hp-infected mice [21]. In contrast, TFF3 is not expressed in the stomach but elaborated primarily by goblet cells in the intestinal tract [25, 26]. Thus, the expression pattern of TFF1 and TFF2 mimics those of MUC5ac/MUC1 and MUC6, respectively, in normal stomach of humans and mice, as well as in human gastric carcinoma patients and in mouse models [7,8,9, 21, 25, 28, 32].

Our findings of spontaneous antro-pyloric proliferative lesions in uninfected Muc5ac−/− mice is similar in some respects to the spontaneous antro-pyloric lesions observed in an earlier study with Tff1−/− (pS2) mice on a similar C57BL/6 background [29]. We observed spontaneous antro-pyloric hyperplasia and dysplasia in nearly 70% of Muc5ac−/− mice housed in two different institutions (NIEHS and MIT) at about 9–10 months of age with at least 17% of mice developing adenomas but no carcinomas were noted. However, in the earlier study by Lefebvre, the antro-pyloric dysplastic lesions were more rapid and severe, with 100% penetrance affecting both Tff1−/− sexes equally and progressed to antral carcinoma (30%) at about 5 months of age [29]. Of interest, we observed strong expression of Tff1 within the antral/pyloric hyperplastic/adenomatous glands of Muc5ac−/− mice. This suggests that Tff1 was unaffected from the deficiency of Muc5ac−/− and it likely did not play a key role in the development of spontaneous antro-pyloric proliferative lesions in Muc5ac−/− mice. The Tff1−/− mice used in the earlier study [29] were not assessed for Muc5ac expression levels, and hence our data raise the possibility of Muc5ac downregulation in Tff1−/− mice as a co-contributor in antro-pyloric carcinogenesis. In a manner similar to Muc5ac−/− mice, A4gnt–/– mice that are deficient in the gastric αGlcNAc-terminated glandular mucin are also prone to pyloric carcinogenesis via a hyperplasia−dysplasia−neoplasia pathway [33], thus implying that alterations in gastric mucins either in the surface or within glands region predisposes to pyloric tumors. Interestingly, one of the adenomas in the gastric antro-pyloric region of Muc5ac−/− mice also showed well-demarcated foci with histomorphological features of presumptive neuroendocrine differentiation; however, it was not positive for ChrgA, unlike the typical expected findings of tissue ChrgA positivity with frequent elevated plasma levels of ChrgA of human patients with gastric neuroendocrine tumors or carcinoids [30]. In humans, gastric neuroendocrine tumors (GNETs), also called gastric carcinoids, constitute less than 2% of all gastric neoplasms and are mainly derived from enterochromaffin-like (ECL) cells in the glands of the gastric mucosa. They can be solitary or multiple tumors, rarely metastatic and are often diagnosed as an incidental finding with or without overt clinical symptoms. The incidence of GNETs has increased in recent decades and is attributed to increased awareness, screening, and diagnosis via upper gastrointestinal endoscopy [30].

The relevance of these Hp-independent antro-pyloric spontaneous tumors in the absence of Muc5ac needs further exploration for its translational significance in humans. Nevertheless, these lesions in terms of location mimic rare instances of human pyloric gland adenomas located in the gastric outlet [34]. Although cancers have been reported in association with gastric hyperplastic polyps in humans, these are extremely rare [35]. Unlike foveolar gland adenomas, where cells have an apical mucous cap and express MUC5AC in humans, pyloric gland adenomas typically do not have cells with a distinct mucin cap and tend to label with both MUC5AC and MUC6 and pose a risk for progression to carcinoma [34]. Gastric hyperplastic polyps are infrequently detected in human endoscopies and can be single (predominant) or multiple, and are more common in females than males. The polyps may be flat, sessile or pedunculated with a histological spectrum of fundic gland polyps, hyperplastic polyps and adenomatous polyps [35]. Gastric polyps are increasingly recognized in middle-aged humans and their distribution is now altered from its predominant antral location in past descriptions to favor a more corpus location [35]. Large antral polyps or adenomas in humans can lead to gastric outlet obstruction and associated clinical symptoms [35].

The corpus gastric histopathological index scores in Hp-infected WT and Muc5ac−/− mice at 16 wpi and 32 wpi trended in a manner similar to that of historical data in C57B/6-129 mice [19, 21] and with significantly higher scores than their respective uninfected controls. A comparison of the two infected groups indicates a trend towards slightly lower overall gastric pathology index and importantly a significant reduction in mucous metaplasia in Hp-infected Muc5ac−/− mice. Oxyntic metaplastic changes in mice following Hp infection are not typical of intestinal metaplasia seen in Hp-infected humans and more often are a combination of mucous metaplasia and pseudopyloric metaplasia, both of which express markers of SPEM [21, 22, 25]. Mucous metaplasia can occur spontaneously and independent of Hp infection [21, 22] but we have also frequently observed its enhancement following Hp infection [19,20,21]. Given the importance of Muc5ac in mucin/mucous formation, we consider mucous metaplasia, though murine-specific, to be an important parameter to include in our analysis as it also shares many features of the molecular aspects of SPEM noted in pseudopyloric metaplasia. In WT Hp- infected mice, in areas with corpus oxyntic changes, particularly with mucous metaplasia, there was a decrease in Muc5ac immunostaining of the overlying foveolar epithelial surface pits. In the same WT Hp-infected mice, Muc5ac staining was enhanced within hyperplastic regions of the antrum and this corresponds to altered Muc5ac expression profile observed in Hp-infected patients [8, 9, 26].

Our results from GSL II and Clusterin immunohistochemistry has shown that SPEM occurs in a significant proportion of glands characterized morphologically as pseudopyloric metaplasia. It also occurs to a certain extent in glands undergoing mucous metaplasia and it is likely that the presence of Muc5ac aids in an unidentified manner to transformation of oxyntic cells to mucous metaplasia phenotype. Recently, it was shown using Muc1-deficient mice that MUC1 protects against Hp-induced gastric pathology mice by regulation of the NLRP3 inflammasome [36]; however, we noted strong expression of Muc1 in Hp-infected corpus and within spontaneous antral adenomas of Muc5ac−/− mice. Tff2, which is a marker for SPEM and binds to MUC6 [37], was shown to be highly expressed in metaplastic (both mucous and pseudopyloric type) gastric corpus of Hp-infected WT and Muc5ac−/− groups which is in agreement with our earlier findings in the corpus of Hp-infected wild-type mouse stomach [21]. In our previous study in Tff2−/− mice, there was strong evidence for a protective role by Tff2 in Hp-induced IFN-ɣ-associated gastric corpus pathology (6 months of age) and late onset (19 months of age) antral carcinogenesis [21]. In this study, the antral proliferative lesions strongly expressed Tff2 and there was no difference in Tff2 gene expression levels on the basis of qPCR data across all experimental groups.

The inflammatory gastric pathology in Hp-infected WT and Muc5ac−/− mice correlated with an increase in cytokines, including Tnfα, Ifnγ, Il-1β, Il-17A, and Il-22, as previously documented [19,20,21]. However, Tnfα at both post-infection time points and Il-17A at 16 wpi (males only) was significantly reduced in the stomach of Hp Muc5ac−/− mice as compared to their Hp-infected WT, whereas there was no significant difference in Ifnγ levels between the two infected groups. This contrasts the documented significant increase in gastric Ifnγ levels in Hp-infected Tff2−/− mice on a C57BL/6 background [21]. We also observed significantly enhanced Hp gastric colonization levels in Muc5ac−/− mice at 16 wpi, although the colonization levels were similar in Muc5ac−/− and WT mice at 32 wpi. Previous data with Hp infection studies in WT C57BL/6 have shown either an inverse correlation between grade of inflammation and colonization density [20, 21], or a lack of any significant correlation between inflammation and Hp colonization in C57BL/6 mice [38] or Tff2−/− C57BL/6 mice [21], respectively. In C57BL mouse studies with the gastric H. pylori and entero-hepatic H. bilis [19] or H. muridarum [20], it was shown that H. bilis or H. muridarum intestinal colonization was associated with a significant attenuation of Hp- induced gastric pathology in a setting of higher Hp gastric colonization levels. It was hypothesized that the inhibitory effect was likely mediated due to anti-inflammatory effects of regulatory T (Foxp3+ TREG) cells [20, 21]. It is also possible that the underlying reason for this enhancement of Hp colonization in Muc5ac−/− mice at 16 wpi is linked to the inherent ability of Hp to impair mucin production and turnover [39] which might be more pronounced in the absence of Muc5ac, thus leading to reduced bacterial clearance and enhanced Hp adherence. Muc1, another gastric epithelial surface mucin was shown to play a beneficial role in a mouse model of Hp infection by blocking downstream activation of nuclear factor-κB inflammatory signaling pathway [36]. Further, Hp-infected Muc1−/− mice showed significantly enhanced gastric Hp colonization, coupled with increased tissue transcript levels of TNF-α and keratinocyte chemoattractant (KC) [40]. Interestingly, in our study, in Muc5ac−/− mice, enhanced Hp gastric colonization was observed in the setting of attenuated inflammation and reduced Tnfα tissue levels, raising the possibility that Muc5ac may enhance bacterial clearance and limit persistent colonization within deeper glands, while at the same time serving as a critical receptor for Hp, activating downstream signaling cascades and host immune responses.

In conclusion, we have shown that Muc5ac−/− mice are predisposed to the development of spontaneous antro-pyloric proliferative lesions, and could serve as a useful model to study antro-pyloric proliferative dysplastic lesions. In addition, Hp infection is mildly attenuated in the absence of Muc5ac in tandem with a reduction in gastric Il17a and/ or Tnf-α expression. Further studies are needed to clarify the roles of Muc5ac in Hp colonization and adherence mechanisms, as well to explore the interaction between Muc5ac and other innate local factors in maintaining gastric homeostasis.