PD-MSCs were obtained from CHA University (Prof. Yong Soo Choi, CHA University, Seongnam, Korea). The PD-MSC line was cultured in α-MEM containing 1 μg/ml heparin, 25 mg/ml fibroblast growth factor, 10% (v/v) fetal bovine serum, and 100 U/ml penicillin. The cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂. The rat gastric mucosal cells, RGM-1, were kindly given by Prof. Hirofumi Matsui (University of Tsukuba, Japan) and were maintained at 37°C in a humidified atmosphere containing 5% CO₂. RGM-1 cells were cultured in Dulbecco’s modified Eagle’s medium containing 10% (v/v) fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin. For co-culture experiments, PD-MSCs were seeded in a six-well transwell system and cultured for 24 h before co-culture with RGM-1 cells. Just prior to co-culture, PD-MSCs were washed with PBS three times and co-cultured with RGM-1 in the RGM-1 cell culture medium.

H. pylori strain ATCC43504 (American Type Culture Collection, cagA+ and vacA s1-m1 type strain) was used for the in vitro cell model and Sydney strain (SS1, a cagA+ and vacA s2-m2 strain adapted for mice infection) for the in vivo model. H. pylori bacteria (Figure 1A and Figure 4A) were cultured at 37°C in a BBL Trypticase soy (TS) agar plate with 5% sheep blood (TSAII; BD Biosciences, Franklin Lakes, NJ) under microaerophilic conditions (BD GasPaK EZ Gas Generating Systems, BD Biosciences) for 3 days. The bacteria were harvested in clean TS broth, centrifuged at 3,000×g for 5 min, and resuspended in the broth at a final concentration of 10⁹ colony-forming units (CFUs)/ml. In all experiments, cultures grown for 72 h on TS agar plates were used.

Five-week-old male C57BL/6 mice (WT mice) were purchased from Orient (Seoul, Korea), and they were housed in a cage maintained in a 12 h/12 h light/dark cycle under specific-pathogen-free conditions (n = 50). C57BL/6 mice were purchase from Central Lab Animal Inc. (Seoul, Korea). Six-week-old female C57BL/6 mice were fed sterilized commercial pellet diets (Biogenomics, Seoul, South Korea) and sterile water ad libitum and housed in an air-conditioned biohazard room at a temperature of 24°C. We divided 50 mice into four groups: Group 1 (n = 10), WT mice in the vehicle control group; Group 2 (n = 20), WT mice in the H. pylori–infected disease control group; Group 3 (n = 10), WT mice in the H. pylori–infected disease group administered 1x10⁷/100 ml PD-MSCs; and Group 4 (n = 10), WT mice in the H. pylori–infected disease group administered CM obtained from PD-MSCs, 100 μl concentrated from PD-MSC culture. We maintained these four groups up to 36 weeks, respectively. All groups were given intraperitoneal injections of pantoprazole, 20 mg/kg (Amore-Pacific Pharma, Seoul, Korea), as the proton pump inhibitor (PPI), three times per week, to increase successful H. pylori colonization through lowered gastric acidity, and then, each mouse was intragastrically inoculated with a suspension of H. pylori containing 10⁸ CFUs/ml or with an equal volume (100 μl) of clean TS broth using gastric intubation needles. The H. pylori–infected mice were fed a special pellet diet based on AIN-46A containing 7.5% NaCl (high-salt diet, Biogenomics, Seongnam, Korea) for a total of 36 weeks (Figure 1A and Figure 4A) to promote the H. pylori–induced carcinogenic process in all infected animals. Randomized groups of mice (n = 10) were sacrificed after 36 weeks of H. pylori infection, respectively, based on our previous experience that CAG was generated at 24 weeks and gastric tumorigenesis was generated after 36 weeks (Park et al., 2014). The body weight was checked in all mice every 3 days up to observational periods. The stomachs of mice were opened along the greater curvature and washed with ice-cold PBS. The numbers of either erosions/ulcers or protruded nodule/mass were determined under the magnified photographs (Figure 1C). Stomachs were isolated and subjected to histologic examination, ELISA, western blotting, and RT-PCR. All animal studies were carried out in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of CHA University, CHA Cancer Institute, after IRB approval (IRB 17-0901).

The results are expressed as mean (standard deviation (SD)). Statistical analyses were conducted with GraphPad Prism (GraphPad Software, La Jolla, CA) and SPSS software (version 12.0; SPSS Inc., Chicago, IL). Statistical significance between groups was determined by a multi-variate Kruskal–Wallis test. Statistical significance was accepted at p < 0.05.

Detailed experimental procedures for gross lesion index, index of histopathologic injury, immunohistochemical staining, terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling (TUNEL) staining, RT-PCR, western blotting, cytokine protein array, preparation of cytosolic and nuclear extracts, RNA interference, zymography, bacterial DNA extraction from mouse stool samples, bacterial metagenomic analysis using DNA from stool samples, and analysis of bacterial composition in the microbiota can be found in the Supplementary Materials.

Since we already have an established H. pylori–induced CAG mouse, which relies on an H. pylori infection–initiated, high-salt diet–promoted mouse model (Nam et al., 2004a; Nam et al., 2004b; Park et al., 2014; Jeong et al., 2015; Han et al., 2016; An et al., 2019), we went on to design this experiment to document the ameliorating action of PD-MSCs or their conditioned medium (CM) against H. pylori–induced CAG. We administered ten doses of PD-MSCs or their CM following 22 weeks of H. pylori infection and then evaluated these animals over a nine-week period until they were terminated for end point analysis at 36 weeks (Figure 1A). Our CAG model was initiated following four PPI injections, which facilitates the successful colonization of H. pylori in the lowered gastric pH and of cultured H. pylori into mice who were then fed a 7.5% salt AIN-76A pellet diet until 20–24 weeks, when the control mice (Group 2) showed significant changes in CAG, manifested with erosions, ulcers, and a very thin atrophied gastric wall (Figure 1C). To encourage the rejuvenating effects of the PD-MSCs (1 × 10⁷/100 μl PD-MSCs, Group 3) or their concentrated CM (100 μl, concentrated from PD-MSC culture), they were were administered to groups of mice via the oral route approximately 10 times before the mice were sacrificed at 36 weeks and subjected to histological examination. In addition, each mouse was evaluated for body weight over the total course of treatment as this is an effective marker for atrophic gastritis. Figure 1B clearly shows that the mean body weights of the H. pylori control mice significantly decreased between week 24 and the end point of the experiment (p < 0.01), while there were no significant changes in body weight in the groups treated with PD-MSCs or their CM. There were also significant changes in the gross lesions of the stomach in the control animals at 36 weeks, with these animals presenting with multiple scattered nodular and elevated mass-like lesions, thinned corpus and pylorus walls, and some scattered nodular changes, signifying the development of typical CAG. The gross lesion index revealed that there were significant changes in the H. pylori–infected stomach and that intervention with PD-MSCs or their concentrated CM (p < 0.01) significantly ameliorated the severity of these lesions (Supplementary Table 1; Group 4, p < 0.01, Figure 1C). The resected stomachs from each group were also subjected to pathological evaluation, and the total pathological scores describing gastric inflammation, gastric atrophy, and tumorigenesis were significantly decreased in the PD-MSC–treated group compared to the control group (Supplementary Table 2; p < 0.001, Figure 1D). Gastric atrophy is generally associated with a loss of parietal cells and gastric glands and increased inflammatory cell infiltration. Our evaluations revealed that Group 2 demonstrated a typical increase in these parameters (p < 0.001), but that intervention with PD-MSCs or their CM significantly ameliorated these effects (p < 0.05, Figure 2A; Supplementary Figure 1A). Further investigation of the proton pump in the parietal cells, analyzed via immunostaining of H⁺/K⁺-ATPase, revealed a significant decrease in its expression in the control group (p < 0.01) but an increase in its expression in Groups 3 and 4 (p < 0.05, Figure 2B; Supplementary Figure 1B). These results suggest that PD-MSCs and their CM exert a significant mitigating effect on H. pylori–induced CAG.

Lgr5+ cells have been identified as a possible source of stemness in the stomach (Hata et al., 2018; Sigal et al., 2019; Tang et al., 2019). Figure 2C shows that the expression of Lgr5+ was significantly decreased in Group 2 (p < 0.05, Figure 2C) and that there was a loss of leucine-rich repeat-containing G-protein–coupled receptor 5+ (Lgr5+) cells following chronic H. pylori infection. On the contrary, this trend was reversed following treatment with either PD-MSCs or their CM (p < 0.01, Figure 2C). Ki-67–mediated evaluation of cellular proliferation revealed that there were significantly fewer Ki-67 cells in the control group when compared to the healthy control. However, Groups 3 and 4 showed significant increases in Ki-67 expression (p < 0.01, Figure 2D; Supplementary Figure 1C). Musashi-1 expression in the stomach reflects stemness. Musashi-1 expression was increased in Groups 3 and 4 (p < 0.05, Figure 2E). In addition, the expressions of cyclin A, cyclin E, p-ERK, and smad2/3, which are all implicated in the pathogenesis of H. pylori–associated CAG, were evaluated in response to treatment with PD-MSCs or their CM. PD-MSCs significantly increased cyclin E expression, and PD-MSCs and their concentrated CM significantly decreased cell growth suppressive smad2/3 expression (Figure 2F; Supplementary Figure 2). The expression of PDGF mRNA in the resected stomach tissues was significantly increased in response to treatment with either PD-MSCs or their concentrated CM. The expression of HGF mRNA was increased in response to treatment with PD-MSCs (Figure 2G).

Chronic H. pylori infection induces considerable levels of apoptosis, a loss of parietal cells in the corpus, decreased somatostatin-secreting D cells in the antrum, and robust apoptosis in the epithelial cells, which are responsible for atrophic gastritis, peptic ulcer disease, and mucosal erosions (Alzahrani et al., 2014; Zhao et al., 2020). Figure 3A shows that 36 weeks of chronic H. pylori infection led to considerable levels of apoptosis (p < 0.001). However, animals treated with PD-MSCs showed significantly decreased levels of apoptosis, even during chronic H. pylori infection (p < 0.01, Figure 3A). Western blot against the central apoptosis-related molecules showed a significant increase in Bax in response to CAG conditions (Group 2). However, the levels of Bcl-2 were significantly increased and Bax was significantly decreased in Group 3 (p < 0.05, Figure 3B). Given this, we extended our investigation to include an exploration of the anti-apoptotic effects of PD-MSCs using a transwell co-culture system. Figures 3C,D demonstrate that H. pylori infection significantly decreased RGM-1 cell viability, while the addition of PD-MSCs significantly decreased the expression of H. pylori–induced apoptotic executors. Further evaluations of the autophagy response in cells treated with PD-MSCs or their CM demonstrated that there was a significant increase in LC3B-II and ATG5 autophagosomes (Figures 3E,F; Supplementary Figure 3) and that this increase was lost in response to LC3B siRNAs (Figure 3G).

Increased expression of COX-2 is known to be responsible for perpetuated gastric inflammation and gastric carcinogenesis in chronic H. pylori infections (Resende et al., 2011; Thiel et al., 2011; Cheng and Fan, 2013; Echizen et al., 2016). When we measured COX-2 mRNA and COX-2 protein expressions (Figure 4A), we noted a significant increase in both COX-2 mRNA and COX-2 protein in Group 2 (p < 0.001). However, COX-2 expression was significantly decreased in both Groups 3 and 4 (p < 0.01, Figure 4A). COX induction can lead to an increase in 15-PGDH as part of the hormetic response designed to retain homeostasis. Expression of 15-PGDH is known to exert some tumor suppressive effects and has been linked to reducing tumorigenesis in response to H. pylori–mediated CAG. However, the mean expression of 15-PGDH in Group 2 was significantly lower than that in Group 1 (p < 0.01; Figure 4B), while 15-PGDH levels were significantly increased in Groups 3 and 4 compared to Group 2 (p < 0.05, Figure 4B). The results of the immunohistochemical staining of 15-PGDH were further confirmed by western blot against this protein in each group (p < 0.01, Figure 4C).

Oncogenic perpetuated gastric inflammation following H. pylori infection is the root cause of both CAG and gastric carcinogenesis (Piazuelo et al., 2010; Kim E.-H. et al., 2011). The expressions of inflammatory mediators such as IL-1β, IL-6, IL-8, TNF-α, and NOX-1 were all significantly increased in Group 2. However, administration of either the PD-MSCs or their concentrated CM led to a significant decrease in the expression of these inflammatory mediators (Figure 5). In this experiment, a protein array comprising several cytokines and chemokines was compared between the groups. The expressions of IL-1β, RANTES, IFN-γ, IL-17, IL-6, and TNF-α were all significantly increased in Group 2, but their expressions were all consistently attenuated in Group 3, signifying the contribution of the PD-MSCs to both the anti-inflammatory and anti-mutagenesis responses (Figure 5A). The RT-PCR evaluating the expression of the inflammatory mediators was then validated using a protein array experiment (Figure 5B). NF-κB activation and STAT3 phosphorylation are known to be involved in the progression of gastric inflammation following H. pylori infection. When we compare the expression of NF-κB and the phosphorylation of STAT3 between the groups (Figure 5C), we observed a significant increase in the activation of NF-κB κ and STAT3 in Group 2 and that the addition of PD-MSCs or their CM reversed these effects almost entirely. Infiltrating macrophages are the primary source of these inflammatory mediators following transcriptional activation, and we examined the expression of NF-κB, p65, and F4/80 through immunohistochemical staining. Figures 5C,D show that the highest expression levels of NF-κB and F4/80 (Figure 5E) were seen in Group 2 and that their expression was significantly decreased in Groups 3 and 4 (p < 0.01).

Acute and chronic H. pylori infection leads to significant changes in atrophic gastritis (Correa and Piazuelo, 2008; Piazuelo et al., 2010; Correa and Piazuelo, 2011; Wroblewski et al., 2015), and H. pylori infection is defined as a class I carcinogen. Although eradication of H. pylori and non-anti-microbial interventions have been evaluated in the prevention of gastric cancer, the data suggest that the rejuvenation of precancerous atrophic gastritis seems to be the best way to prevent malignancy. Homeostasis seems to be very important in achieving rejuvenation in these tissues, and we hypothesize that IL-10 and the regulation of the inflammasome are critical to the success of therapeutic interventions using PD-MSCs. H. pylori infection is associated with increased inflammasome activity (Figure 6A; Supplementary Figure 4A) as H. pylori infection leads to increased NOD, LRR, and pyrin domain–containing protein 3 (NLRP3) and IL-1β expressions. PD-MSCs significantly increased inflammasome activation in the presence of H. pylori infection in the transwell co-culture system. However, H. pylori infection led to a significant decrease in IL-1β/IL-18 activity, while the administration of PD-MSCs in the presence of H. pylori infection led to significant inhibition of IL-18 and IL-1β secretion (Figure 6B; Supplementary Figure 4B). Under these conditions, IL-10 mRNA expression was significantly induced in response to PD-MSCs (Figure 6C; Supplementary Figure 4C), and where IL-10 induction was not feasible, the inhibitory action of PD-MSCs on IL-1β was significantly reduced (Figures 6D,E). This suggests that the significant anti-inflammatory actions of PD-MSCs were largely reliant on the concerted activity of various mechanisms for maintaining homeostasis. These actions were further supported by the significant inhibitory action of PD-MSCs on MMP, as seen in Figures 6F,G, which revealed a significant attenuation in H. pylori–induced MMP activity in response to these cells. Activated proteases, especially matrix metalloprotease (MMP), have been implicated in the propagation and aggravation of gastritis and the development of CAG or ulcers. This is supported by the fact that the expressions of MMP-2 and MMP-2 activity are significantly increased in Group 2 (p < 0.001). The protein array revealed that the expression of TIMP-1 was significantly increased in Group 3 (p < 0.001, Figure 6G), and these results were validated by a significant decrease in the expression of MMP-2 mRNA and MMP-2 activity in the RT-PCR and zymography assays (Figure 6F).

Changes in the microbiome in response to H. pylori infection are responsible for various gastric pathologies as bacterial overgrowth is closely associated with the changes in gastric atrophy and decreased gastric acidity (Huang et al., 2020; Lahner et al., 2020; Stewart et al., 2020; Ye et al., 2020). Although not clearly defined, overt changes in the intestinal microbiota do occur as the gastric pathology progresses toward atrophic gastritis. Figure 7A shows that the changes in the microbiota can be defined according to the group. Clear delineation was observed in PDA. Phylum analysis clearly showed that the phyla in Group 2 were quite different from those in Group 1 and Groups 3 and 4. Since the average gastric pathology was CAG, we speculated that the genera in Group 2 were also different from those in Group 1. However, Groups 3 and 4 showed a similar pattern to Group 1, suggesting that changes in CAG in Groups 3 and 4 might improve the community composition of the microbiome in these mice and facilitate their return to a more normal profile (Figure 7B,C). A detailed analysis of the genera in these samples (Figure 7D) revealed significant changes in the gastric microbiota in Group 2, but not in Groups 3 and 4 when compared to Group 1. The heatmap in Figure 7E and Supplementary Table 3 summarize the detailed changes in the gastric microbiota of these animals. We concluded that the administration of PD-MSCs and their CM significantly rejuvenated H. pylori–associated CAG, leading to the expectation that MSCs can be used as potential cell therapeutics to reverse precancerous atrophic changes after chronic H. pylori infection.