Helicobacter-induced gastric inflammation alters the properties of gastric tissue stem/progenitor cells

This research was approved by Institutional Animal Experiment Committee at Yokohama City University (Approved#; F-A-14-043). C57BL6/J wild-type mice were obtained from CLEA Japan Inc. (Tokyo, Japan). Male mice aged ≥8 weeks were used in the chronic gastritis model induced by Helicobacter infection [28]. Mice were sacrificed at 3, 6, and 12-month post infection, and the stomachs were removed and subjected to histological analysis or mRNA expression analysis (n = 5 per each time point, Model 1 in Fig. 1a). For eradication study (Model 2 in Fig. 1a), we used 5 mice as continuous infected group and 5 mice as eradicated group. Gastric epithelial cells from corpus area were used for organoid culture unless otherwise indicated [22]. We further investigated the tumorigenic activity of the organoids from Helicobacter-infected or chemical carcinogen N-methyl-N-nitroso-urea (MNU)-treated mucosa by injecting them into immune-deficient NOD/SCID (NOD.CB17-Prkdc ^scid /J) mice purchased from Charles River Laboratories International, Inc. (Wilmington, MA, USA) (n = 4 per group). In MNU model, male 6-week-old wild-type mice were treated with water containing 240 ppm MNU (Sigma, St. Louis, MO, USA) in alternate weeks for a total of 5 weeks, as described previously [29]. Mice were sacrificed at 4 weeks after the final MNU dosing, and then isolated gastric epithelial cells from antrum area for subsequent organoid culture (Model 3 in Fig. 1a). Because MNU is a chemical carcinogen in the mouse gastric antrum, we used this model as a positive control to assess tumorigenicity. We first performed 3D organoid culture as described above, from mice either infected with H. felis, treated with MNU, or untreated control (Model 1 and 3 in Fig. 1a). After 4 days of culture, each group of organoids was injected subcutaneously into mice and the tumorigenicity and histological changes were analyzed after 2 months.

We used Helicobacter felis (H. felis, strain ATCC 49179) [30]. In brief, H. felis was cultured in trypticase soy broth at a titer of 1 × 10⁷ CFU/ml. The bacterial suspension was stored at −80 degree until use. 10 wild-type male mice aged 8 weeks were infected with 0.25 ml of H. felis (concentration; 1.0 × 10⁷ CFU/ml) suspension by oral gavage 3 times in a week. Four months after H. felis infection, mice were divided into two groups (5 mice each): one for continuous infection (Model 1 in Fig. 1) and the other for which H. felis was eradicated 2 months before sacrifice (Model 2 in Fig. 1) using a cocktail of antibiotics with tetracycline HCl (0.5 mg/30 g mouse/day), metronidazole (0.675 mg per 30 g mouse/day), and bismuth subsalicylate (0.185 mg per 30 g mouse/day) dissolved or suspended in a total volume of 500 μL sterile water, as described previously [31]. The cocktail was administered orally 5 days a week for 14 days. Eradication of H. felis was confirmed by histopathology and real-time PCR for H. felis flaB gene using gastric specimens as described previously [31]. We used 4 mice of 12 months’ post infection to analyze cRNA expression by microarray, and total 15 mice for immunohistochemical study (5 mice at 3 months, 5 mice at 6 months, and 5 mice at 12 months). Appropriate number of uninfected mice was used as experimental controls.

We cultured gastric epithelial cells in Matrigel in serum-free media for 3D organoid culture as described previously [32, 33]. Briefly, mouse gastric corpus was removed and cut into approximately 1-mm² pieces. The gastric epithelial tissues were washed in ice-cold phosphate-buffered saline (PBS) three times. Next, the tissues were incubated in ice-cold PBS containing 10 mM EDTA at 4 °C for 3 h. After incubation, the tissue samples in PBS were vigorously shaken and centrifuged, and then the sediments were enriched in gastric glands for subsequent experiments. Isolated gastric glands were seeded on pre-warmed Matrigel plates (24 wells) supplemented with B27, N2, N-acetylcysteine (final 0.5 mM, Invitrogen, Carlsbad, CA, USA), and gastrin (10 nM; Sigma-Aldrich) containing growth factors (50 ng/ml epidermal growth factor [Peprotech, Rocky Hill, NJ, USA], 1 mg/ml R-spondin1, 100 ng/ml noggin [Peprotech], 100 ng/ml fibroblast growth factor 10 [Peprotech] and 100 ng/ml Wnt3A [R&D Systems, Minneapolis, MN, USA]) without using any feeder cells as described [22]. To exclude the effect of bacterial/fungal contamination to the organoid phenotype, we used Penicillin-streptomycin (100×; Invitrogen, cat. no. 15140–122) in organoid growth media. In the cytokine experiments, after isolating organoids from uninfected wild-type mice stomachs, we added either tumor necrosis factor (TNF)-α (10 ng/ml) or interleukin (IL)-1β (10 ng/ml) to serum-free organoid culture media, and measured the size of the organoids and mRNA expression levels. When we measured the organoids’ size, we viewed 10 fields of each well by the original magnification of 100X. 3 replicates was performed per group, and the software we used was EVOS®XL Core Cell Imaging System (ThermoFisher Scientific) and Microsoft Excel to calculate the average number and size of organoids.

After culturing organoids from mouse gastric corpus infected with H. felis for 12 months or uninfected control, total cellular RNA was extracted from organoid using an acid guanidiumthiocyanate-phenol-chloroform method (ISOGEN Reagents; Nippon Gene, Tokyo, Japan) and column chromatography (RNeasy; Qiagen, Tokyo, Japan), according to the manufacturers’ instructions. Four RNA samples isolated from organoid, two experimental samples originated from mice infected with H. felis and two uninfected mice were used. For microarray analysis, cRNA mouse gene expression microarrays containing 39,430 Entrez Gene RNAs and 16,251 lincRNAs were used according to the manufacturer’s instructions. Genes were considered to be significantly upregulated in organoids infected with H. felis compared to control when they were upregulated in both arrays with an average log2 ratio exceeding 1 (two-fold). The data were confirmed by quantitative real-time polymerase chain reaction (qRT-PCR) analysis using primers (sequences were listed in Table 1).

The following antibodies, listed with the catalog number, company, and dilution, were used for immunohistochemical analysis: TFF2 (sc-15,334, Santa Cruz, 1:100), MUC2 (orb13709, Biorbyt, 1:100), DCLK1 (AP7219b, Abgent, 1:100), VILLIN (sc-7672, Santa Cruz, 1:100), CK19 (ab15463, abcam, 1:50), MUC4 (sc-33,654, Santa Cruz, 1:100), CD44v6 (MCA1967, Serotec, 1:100). Immunohistochemical staining was performed on 4 μm sections with avidin-biotin-peroxidase complex kits (Vector Laboratories, USA) and counterstained with hematoxylin. For each mouse, 5 well-oriented gastric glands were scored to quantify the number of positive cells per gland. Data were expressed as average number of positive cells +/− S.D. in each group. Glyceraldehyde 3-phosphate dehydrogenase expression was used as an internal control (GAPDH, Cell Signaling, 2118, 1:1000).

We first conducted MNU study in male C57BL6/J to generate organoids prior to subcutaneous injection into female NOD/SCID mice. Gastric epithelial cells were isolated and used for organoid culture. After 4 days of culture in Matrigel, we digested Matrigel with Cell Recovery Solution (Corning™ Cat No.354253), collect and centrifuge organoids, and re-suspended in PBS before injection. We injected the organoids subcutaneously into the flanks of 6-week-old female NOD/SCID mice. The sizes of the resulting tumors were measured using a caliper after sacrificing the mice. The tumors were isolated, paraffin-fixed, formalin-embedded, and subjected to histological analysis.

To assess the effect of chronic inflammation on the number of tissue stem/progenitor cells, we used a mouse gastritis model infected with H. felis for up to 12 months (Model 1 in Fig. 1). Histological analysis revealed severe gastric inflammation with lymphoid follicles, neck cell hyperplasia, oxyntic atrophy, and mucous metaplasia (SPEM), particularly in the gastric corpus, as described previously (Fig. 1b) [34]. We performed immunohistochemical analysis of TFF2, MUC2, CD44, DCLK1, and VILLIN, all of which have been proposed to be a tissue stem cell marker in gastrointestinal organs [20, 23, 35]. The expression levels of TFF2, MUC2, CD44, DCLK1, and VILLIN in H. felis-infected mice were significantly higher than those in uninfected control mice (Fig. 1b, c). Most of the cells positive for tissue stem cell markers were likely to be TFF2-positive, suggesting that certain cells in SPEM acquired stem cell properties. Those cells in SPEM also showed a proliferative marker, Ki67-positive (data not shown). These results indicated that chronic inflammation induced by H. felis infection affected the tissue stem cell property, which could be associated with gastric carcinogenesis.

To assess the change in gastric epithelial cells in the stomach after chronic inflammation, we utilized 3D organoid culture systems using cells isolated from mice infected with H. felis, or uninfected controls, as described in Methods. Gastric organoids grew slowly and started to show budding at approximately 10 days after seeding (Fig. 2a). The organoids showed a single-layered epithelial structure under confocal microscopy with E-cadherin immunocytochemistry (data not shown). In accordance with the increased tissue stem/progenitor cells, the number of organoids cultured from H. felis-infected gastric epithelial cells was significantly higher than those from uninfected mice (Fig. 2b). we used half of the stomach for each mouse to establish the organoids. In this experiments, we used five same aged- male mice to generate the organoids.

To characterize the organoids in H. felis-infected mouse stomachs, we analyzed the mRNA expression profiles of organoids of the two groups using microarray and qRT-PCR. Among approximately 60,000 spots, 197 genes were up-regulated by two-fold or greater, and 581 genes were down-regulated less than 0.5 times. The top 50 upregulated or downregulated genes are listed in Tables 2, Additional file 1: Table S1 and Table S2. Among these significantly altered genes, intestinal-related genes, such as Isx, Muc4, Muc16 and Villin, were upregulated, whereas gastric-related genes, such as Pepsinogen C and Gif were downregulated, demonstrating that H. felis infection caused intestinal alteration in gastric tissue stem/progenitor cells in vivo. Among the upregulated genes, we focused on genes reported to be associated with gastric carcinogenesis or genes known as stem cell marker, and confirmed that interleukin-1α, Villin and Muc4 expression levels were increased in organoids isolated from H. felis-infected mice compared with control mice by real-time qRT-PCR analysis. (Fig. 2c, Table 2).

Muc4, an intestinal mucin, has been shown to be associated with gastric carcinogenesis [36‐39]. Because Muc4 was upregulated in organoids obtained from H. felis-infected mice, we performed immunohistochemistry analysis to confirm whether MUC4 was expressed in mouse stomachs infected with H. felis. MUC4 protein was expressed in gastric mucosa after 3 months of H. felis infection; furthermore, the area of MUC4-positive mucosa expanded as infection continued (Fig. 2d).

Since early eradication of Helicobacter inhibits gastric cancer progression [31], we analyzed the effect of H. felis eradication on the gene expression profile of gastric organoids. Eradication of H. felis was confirmed by flaB gene expression in gastric tissue at sacrifice (Fig. 3a). The mice were sacrificed and the stomachs were used for immunohistochemical analysis or isolation for 3D–organoid culture as described above. The protein expression of candidate gastric cancer stem/progenitor cells was assessed by using immunohistochemistry. Among the markers analyzed, the number of cells positive for Cd44, Dclk1, Muc4, or Villin was significantly reduced after eradication of H. felis (Fig. 3b). The number of organoids was also significantly reduced after H. felis eradication whereas the size of organoid was not decreased after H. felis eradication (Fig. 3c). We also assessed the mRNA expression of the stem cell markers above in organoid, and found that after H. felis eradication, all markers except Villin, were significantly down-regulated (Fig. 3d), suggesting that H. felis-induced SPEM or intestinal phenotype were reversible by the eradication of H. felis.

Chronic inflammation induced by Helicobacter infection is considered as a key factor in gastric carcinogenesis [40]. When we measured inflammatory cytokines in the organoids generated from each mouse before and after H. felis eradication, all four pro-inflammatory cytokines were increased in the organoid isolated from H. felis infected mice, and downregulated after eradication of H. felis (Fig. 4a). To further analyze the effect of inflammatory cytokines on gastric tissue stem/progenitor cells, we used the organoid culture system to determine whether inflammatory cytokines directly affected the phenotype or mRNA expression profile of gastric organoids. The mRNA levels of Villin, IL-1α, Muc4, and Dclk1 were upregulated in organoids stimulated with inflammatory cytokines compared with untreated organoids. TNF-α administration demonstrated modest up-regulation of those four genes (Fig. 4b). On the other hands, stimulation with IL-1β resulted in marked upregulation of these genes (Fig. 4c). Immunostaining of organoids with an antibody against DCLK1 demonstrated that the number of DCLK1-positive cells was increased in organoids obtained from H. felis-infected stomachs (Fig. 4d). In addition, administration of each cytokine increased the size of organoid (Fig. 4e), suggesting that cytokine stimulation may play a role in inducing a genetic throwback from matured cell to the tissue stem/progenitor cell phenotype through the induction of stem/progenitor cell associated genes, which in turn accelerating their proliferation [41‐44].

Finally, we assessed the tumorigenic property of organoids cultured from stomach using NOD/SCID mice. At day 4 of organoid culture, the organoids from MNU-treated mice showed early proliferation of cells, and exhibited an irregular morphology compared with those isolated from H. felis-infected or uninfected mice (Additional file 1: Figure S1A). After 2 months of injection of organoids into NOD/SCID, only organoids from MNU-treated mice produced tumors, and no tumors developed with organoids obtained from mice infected with H. felis or control mice (Additional file 1: Figure S1B). We confirmed that these tumors contained epithelial cells by using CK19 immunohistochemical staining (Additional file 1: Figure S1C). These results suggested that chronic inflammation alone may not be sufficient to induce malignant transformation in gastric tissue stem/progenitor cells, and other factors such as additional genetic/epigenetic changes in tissue stem/progenitor cells would be required for carcinogenesis.