Helicobacter pylori-induced adrenomedullin modulates IFN-γ-producing T-cell responses and contribute to gastritis

Adrenomedullin (ADM) is a multifunctional peptide that is expressed by many surface epithelial cells, but its relevance to H. pylori-induced gastritis is unknown. Here, we found that gastric ADM expression was elevated in gastric mucosa of H. pylori-infected patients and mice. In H. pylori-infected human gastric mucosa, ADM expression was positively correlated with the degree of gastritis, accordingly, blockade of ADM resulted in decreased inflammation within the gastric mucosa of H. pylori-infected mice. During H. pylori infection, ADM production was promoted via PI3K-AKT signaling pathway activation by gastric epithelial cells in a cagA-dependent manner, and resulted in increased inflammation within the gastric mucosa. This inflammation was characterized by the increased IFN-γ-producing T cells, whose differentiation was induced via the phosphorylation of AKT and STAT3 by ADM derived from gastric epithelial cells. ADM also induced macrophages to produce IL-12, which promoted the IFN-γ-producing T-cell responses, thereby contributing to the development of H. pylori-associated gastritis. Accordingly, blockade of IFN-γ or knockout of IFN-γ decreased inflammation within the gastric mucosa of H. pylori-infected mice. This study identifies a novel regulatory network involving H. pylori, gastric epithelial cells, ADM, macrophages, T cells, and IFN-γ, which collectively exert a pro-inflammatory effect within the gastric microenvironment. Author summary H. pylori infect almost half the world’s population. Once infected, most of people carry the bacteria lifelong if left untreated, so that persistent H. pylori infection can lead to chronic gastritis, peptic ulceration and ultimately gastric cancer. H. pylori infection is accompanied with increased inflammation in gastric mucosa, but the mechanisms of chronic gastritis induced by H. pylori infection remains poorly understood. We studied a multifunctional peptide known as adrenomedullin (ADM) in gastric epithelial cells, which was known as a key factor of regulating gastrointestinal physiology and pathology. Here, we found that gastric ADM expression was elevated in gastric mucosa of H. pylori-infected patients and mice, and was positively correlated with the degree of gastritis. ADM production was promoted via PI3K-AKT signaling pathway activation by gastric epithelial cells in a cagA-dependent manner. Blockade of ADM during H. pylori infection resulted in decreased gastric inflammation that was characterized by the increased IFN-γ-producing T cells which was induced via the phosphorylation of AKT and STAT3 by ADM derived from gastric epithelial cells. ADM also induced macrophages to produce IL-12, which promoted the IFN-γ-producing T-cell responses. These data demonstrate that H. pylori-induced ADM modulates FN-γ-producing T-cell responses and contribute to gastritis.

1 Introduction 2 Helicobacter pylori (H. pylori) is a gram-negative bacterium that infects more than half of the world's 3 population [1]. H. pylori is an important factor of chronic gastritis, peptic ulcer and other digestive 4 system diseases, and has been classified as a class I carcinogen by WHO [2]. During the infection, 5 gastric epithelial cells produce a variety of cytokines that are involved in the inflammatory gastric 6 environment after contacting with H. pylori [3]. Besides, many immune cells such as neutrophils, 7 lymphocytes and plasma cells are releasing inflammatory factors in the stomach of H. 8 pylori-infection [4][5][6]. Inflammatory reaction to H. pylori infection shows special characteristics rarely 9 seen in other organs or biological systems, and the mixed acute and chronic inflammatory reactions 10 contribute to H. pylori-associated gastritis and take place simultaneously during H. pylori infection. 11 Adrenomedullin (ADM) is a small active hormone which is expressed throughout the 12 gastrointestinal tract [7]. ADM that consists of 52 amino acids is structurally similar to calcitonin regulating gastrointestinal physiology and pathology. For example, it has been reported that 18 over-expression of ADM in the stomach can inhibit gastric acid secretion [11]. In other studies, ADM 19 can protect mucosal as an endothelial cell growth factor by promoting mucosal healing [12], and 20 has anti-inflammatory effects in a mouse DSS-induced colitis model [13]. However, the relationship 21 between ADM and gastric inflammation especially in H. pylori-associated gastritis is presently 22 unknown.

23
In our study, we have, for the first time, demonstrated a pro-inflammation role of ADM in H. pylori 24 infection and H. pylori-induced gastritis. ADM is increased in gastric mucosa of both patients and 25 mice infected with H. pylori. H. pylori induces ADM production from gastric epithelial cells in a 26 cagA-dependent manner via activating PI3K-AKT signaling pathway. ADM derived from gastric 27 epithelial cells promotes IFN-γ-producing T cells via the phosphorylation of AKT and STAT3. 28 Besides, ADM also induces macrophages to produce IL-12, which promotes IFN-γ-producing T-cell 29 responses. In vivo, the increased ADM promotes inflammation and IFN-γ-producing T-cell 30 responses in the stomach during H. pylori infection, which contributes to H. pylori-associated 1 gastritis. These findings provide a novel regulatory network involving ADM within the gastric 2 microenvironment and a potentially new target of ADM in the treatment of H. pylori-associated 3 gastritis. 3 To evaluate the potential role of adrenomedullin (ADM) in H. pylori-associated pathology, first we 4 compared the ADM levels in gastric tissues. Notably, compared to uninfected donors, the overall 5 level of ADM mRNA was higher respectively in gastric mucosa of H. pylori-infected patients (Fig 1A). 6 Next, ADM expression was positively correlated with H. pylori colonization (Fig 1B), suggesting 7 induction of ADM by H. pylori. 8 The presence of cagA is strongly associated with the development of H. pylori-associated 9 gastritis [14]. Notably, we found that ADM mRNA expression in cagA-positive patients was 10 significantly higher than that in cagA-negative individuals (Fig 1C). Immunohistochemical staining 11 ( Fig 1E) and western blot analysis ( Fig 1F) showed that ADM protein in cagA-positive patients was 12 also significantly higher than that in cagA-negative individuals. Consistent with our findings in 13 humans, the levels of ADM mRNA ( Fig 1D) and protein ( Fig 1E) were almost only detected in WT H.  (Fig 1G). Taken together, these findings suggest that ADM is increased in H. pylori-infected 20 gastric mucosa of patients and mice.   Next, we used two strains of H. pylori to infect AGS cells, an immortalized human gastric 1 epithelial cell line, and found that H. pylori significantly increased ADM expression (Fig 2D). Moreover, H. pylori-infected AGS cells could increase ADM mRNA expression and ADM protein 3 production in a time-dependent ( Fig 2E) and infection dose-dependent manner ( Fig 2F). Notably, 4 compared to uninfected or the ones infected with ΔcagA, WT H. pylori-infected AGS cells (Fig 2G) 5 and human primary gastric epithelial cells (Fig 2H) also potently increased ADM mRNA expression 6 and ADM protein production. Similar observations were made when another human gastric 7 epithelial cell line HGC-27 cells (S1B and C Fig). Collectively, these results demonstrate that H. 8 pylori infection induces ADM expression in gastric epithelial cells, implying that induction of ADM in 9 these cells are a major cause of increased ADM within the H. pylori-infected gastric mucosa.  12 To see which signaling pathways might operate in the induction of ADM in gastric epithelial cells, we 13 first per-treated AGS cells with corresponding inhibitors, and then stimulated AGS cells with H. 14 pylori. We found that only blocking PI3K-AKT pathway with Wortmannin effectively suppressed 15 ADM mRNA expression and ADM protein production in/from H. pylori-infected gastric epithelial cells 16 (Fig 3A and B and S2 Fig). Furthermore, AKT, a direct PI3K-AKT pathway downstream substrate, 17 was predominantly phosphorylated in AGS cells after stimulated with H. pylori, and this was more 18 noticeable when infected with a WT H. pylori compared to ΔcagA (Fig 3C), and this phosphorylation 19 was abolished when PI3K-AKT signal transduction pathway was blocked with inhibitor Wortmannin 20 ( Fig 3C). Furthermore, H. pylori induced AKT phosphorylation in AGS cells in a dose-dependent 21 ( Fig 3D) as well as in a time-dependent manners ( Fig 3E). These data imply that activation of 22 PI3K-AKT signaling pathway is crucial for ADM induction by H. pylori in gastric epithelial cells.

In vivo blockade of adrenomedullin significantly reduced inflammation and IFN-γ-producing
25 T-cell responses in the stomach during H. pylori infection 26 To understand the possible biological effects of ADM induction during H. pylori infection, we 27 compared ADM expression within the gastric mucosa with the severity of gastritis observed in 28 patients infected with H. pylori. Notably, higher ADM expression was strongly associated with more 29 severe gastritis (Fig 4A). This led us to hypothesize that ADM might exert pro-inflammatory effects 30 during H. pylori infection and thus contribute to gastritis. To test this hypothesis in vivo, we conducted a series of loss-of-function experiments involving 2 ADM, and evaluated the inflammatory response in gastric mucosa on day 21 p.i.. Compared with 3 mice treated with control IgG, mice treated with neutralizing antibodies against ADM showed 4 significantly less inflammation in gastric mucosa ( Fig 4B). Furthermore, neutralization of ADM 5 significantly reduced the level of IFN-γ-producing T-cell responses ( Fig 4C) and IFN-γ production 6 ( Fig 4D) in gastric mucosa but had no impact on Ly6G -CD11b + monocytes, Ly6G + CD11b + 7 neutrophils, CD19 + B cells, and NK1.1 + natural killer cells (NK cells), IL-4-producing T cells and 8 IL-17-producing T cells in gastric mucosa (S3 Fig). As for the control of bacteria growth by 9 inflammatory immune response [16], finally, we compared the levels of bacterial colonization in 10 gastric mucosa on day 21 p.i. and found that neutralization of ADM effectively increased H. pylori 11 colonization ( Fig 4E). Collectively, these results suggest that ADM has pro-inflammatory effects 12 during H. pylori infection in vivo probably by regulating IFN-γ-producing T-cell responses.

Adrenomedullin promotes IFN-γ-producing T-cell responses via PI3K-AKT and STAT3
15 activation 16 Next, we were therefore interested to know if and how ADM induces IFN-γ-producing T-cell 17 responses. To begin, we found that an IFN-γ-producing T-cell infiltration as well as the expression  To further evaluate the contribution of ADM to the induction of IFN-γ-producing T-cell responses, 24 we stimulated T cells with ADM and found that ADM was able to potently induce T cell proliferation 25 and IFN-γ production in a dose-dependent manner ( Fig 5C and S4D Fig). To see which signaling 26 pathways might operate in the induction of IFN-γ-producing T cells by ADM, we first per-treated T 27 cells with corresponding inhibitors, and then stimulated T cells with ADM. We found that blocking 28 PI3K-AKT pathway with Wortmannin or STAT3 phosphorylation with FLLL32 effectively suppressed 29 T cell proliferation and IFN-γ production ( Fig 5D and S4C and E Fig). Furthermore, AKT and STAT3 30 were predominantly phosphorylated in T cells after stimulated with ADM in a time-dependent 1 manner ( Fig 5F and S4F Fig), and this phosphorylation was abolished when PI3K-AKT signal 2 transduction pathway was blocked with Wortmannin or STAT3 phosphorylation was abolished with 3 FLLL32 (Fig 5F and S4F Fig). Finally, to ascertain whether ADM from H. pylori-stimulated gastric 4 epithelial cells contribute to the induction of IFN-γ-producing T-cell responses, we abolished 5 ADM-ADM receptor interaction with ADM fragment 22-52 (an ADM receptor antagonist) (AMA) on T 6 cells, and then stimulated T cells with the culture supernatants from WT H. pylori-stimulated primary 7 gastric epithelial cells. As expected, blocking ADM-ADM receptor interaction effectively inhibited T 8 cell proliferation and IFN-γ production ( Fig 5E). Taken together, these findings suggest that ADM 9 promotes IFN-γ-producing T-cell responses via activation of PI3K-AKT and STAT3 signaling 10 pathways. 13 It has previously been shown that IFN-γ-producing T cells are induced by , and that  To further evaluate the contribution of ADM-regulated macrophages to the induction of 26 IFN-γ-producing T-cell responses, we stimulated macrophages with ADM and then co-cultured with 27 T cells, and found that ADM-stimulated macrophages was able to potently induce T cell proliferation 28 and IFN-γ production ( Fig 6D and S5C Fig). To see whether IL-12 from ADM-stimulated 29 macrophages might operate in the induction of IFN-γ-producing T cells, we added with neutralizing 30 antibodies against IL-12 in co-culture system above, and found that blocking IL-12 effectively 1 suppressed T cell proliferation and IFN-γ production induced by ADM-stimulated macrophages ( Finally, we conducted a series of loss-and gain-of-function in vivo experiments involving 5 and evaluated the inflammatory response and bacterial colonization in gastric mucosa on day 21 6 p.i.. Compared with mice treated with control isotype IgG, mice treated with neutralizing antibodies 7 against IFN-γ showed significantly less inflammation and higher H. pylori colonization in gastric 8 mucosa ( Fig 6F). Conversely, injection of IFN-γ significantly increased inflammation and reduced H. 9 pylori colonization in gastric mucosa ( Fig 6F). Similarly, compared to WT mice, IFN-γ -/mice showed 10 significantly less inflammation and higher H. pylori colonization in gastric mucosa ( Fig 6G). 11 Collectively, these results suggest that ADM-stimulated macrophages promote IFN-γ-producing 12 T-cell responses and IFN-γ contributes to inflammation during H. pylori infection. More than 50% of the world's population infects H. pylori in their upper gastrointestinal tracts, which 3 is more common in developing countries than Western countries [1]. H. pylori infection is a threat to 4 human health, for example, the long-term colonization of H. pylori in the stomach can change PH of 5 the stomach, promote chronic gastritis, gastric ulcers, even gastric cancer [2]. However, until now, 6 the mechanism of H. pylori-associated chronic gastritis remains unclear, and it is believed that the 7 interplays between host and bacterial virulence factors [14,20] System (T4SS) [23], and the following persistent inflammation are likely the underlying causes.

10
In this study we demonstrated a multistep model of inflammation during H. pylori infection within 11 the gastric mucosa involving interactions among H. pylori, gastric epithelial cells, T cells and 12 macrophages via ADM (Fig 7). In vivo and in vitro, we established that H. pylori-associated 13 virulence factor cagA was essential to inducing ADM expression in gastric epithelial cells, which in 14 turn promoted gastric inflammation, CD3 + T cell proliferation and IFN-γ production. Moreover, the 15 increased ADM induced macrophages to secrete IL-12, which also promoted IFN-γ-producing T-cell 16 responses. To our knowledge, this study is the first time to demonstrate the pro-inflammatory role of 17 ADM and its association with macrophages and T cells in H. pylori-induced gastritis.

18
ADM is a biologically active peptide which was isolated in 1993 by Kitamura et al [24] from human 19 pheochromocytoma. It was discovered that ADM was able to activate platelet adenylate cyclase 20 and exerted a long-lasting hypotensive effect in rats [24]. The expression of ADM is related to many 21 factors, for example, inflammatory factors IL-1β and TNF-α can induce ADM expression in some 22 cell types [25]. Some studies showed that plasmatic ADM levels were raised in many infectious  pylori-infected patients, it is possible that ADM might serve as a novel diagnostic and prognostic 3 biomarker for H. pylori-associated gastritis. Future clinical studies are necessary to investigate and 4 verify the ADM-associated mechanisms in humans, which may lead to the application of novel 5 pharmacologic approaches to resist this gastric inflammation.  Table 1). H. pylori infection was determined by [ 14 C] urea 6 breath test and rapid urease test of biopsy specimens taken from the antrum, and subsequently 7 conformed by real-time PCR for 16s rDNA and serology test for specific anti-H. pylori antibodies 8 (Abs). For isolation of human primary gastric epithelial cells, fresh non-tumor gastric tissues (at 9 least 5-cm distant from the tumor site) were obtained from gastric cancer patients who underwent 10 surgical resection and were determined as H. pylori-negative individuals as above at the Southwest 11 Hospital. None of these patients had received chemotherapy or radiotherapy before sampling.

12
Individuals with atrophic gastritis, hypochlorhydria, antibiotics treatment, autoimmune disease, 13 infectious diseases and multi-primary cancer were excluded. The study was approved by the Ethics

22
In some cases, ADM fragment 22-52 (an ADM receptor antagonist) (1 μg/ml) was added into the 23 culture supernatants and incubated for 2 h before stimulation. After 5-day co-culture, cells were 24 collected for flow cytometry and western blot, and the supernatants were harvested for ELISA. Human CD14 + cells were isolated from PBMCs of uninfected donors, and cultured as described 28 previously. 43 To get macrophages, bead-purified CD14 + cells (2×10 4 cells/well in 96-well plates) 29 were stimulated with GM-CSF (100 ng/ml) for 5 days. Macrophages were stimulated with ADM (100 30 nM) for 24 h, and the cells were collected for real-time PCR, and the supernatants were harvested 18 1 for ELISA. In other cases, after stimulated with ADM (100 nM) for 24 h, macrophages were then 2 co-cultured with CFSE-labeled T cells (2×10 5 cells/well in 96-well plates) in new complete 3 RPMI-1640 medium containing rhIL-2 (20 IU/ml), anti-CD3 (2 μg/ml), and anti-CD28 (1 μg/ml) 4 antibodies. In some cases, anti-human IL-12 Ab (20 μg/ml) or control IgG was added into the 5 co-culture supernatants. After 5-day co-culture, T cells were collected for flow cytometry, and the 6 supernatants were harvested for ELISA. Gastric tissue samples were fixed with paraformaldehyde and embed with paraffin, and then were 10 cut into 5 µm sections. For immunohistochemical staining goat anti-mouse ADM Abs was used as 11 primary Ab, and rabbit anti-goat-HRP as the second Ab, then sections were stained by DAB reagent.

12
After that, all the sections were counterstained with hematoxylin and reviewed using a microscope 13 (Nikon Eclipse 80i, Nikon).

5
Human and mouse gastric tissues from specimens were collected, homogenized in 1 ml sterile 6 Protein Extraction Reagent, and centrifuged. Tissue supernatants were collected for ELISA.

7
Concentrations of IFN-γ in the tissue supernatants; concentrations of IFN-γ or IL-12 in T cell culture 8 supernatants or macrophage culture supernatants were determined using ELISA kits according to 9 the manufacturer's instructions. Results are expressed as mean ± SEM. Student t test was generally used to analyze the 20 1 differences between two groups, but when the variances differed, the Mann-Whitney U test was 2 used. Inflammation score data were analyzed by the Mann-Whitney U test. Correlations between 3 parameters were assessed using Pearson correlation analysis and linear regression analysis, as 4 appropriate. SPSS statistical software (version 13.0) was used for all statistical analysis. All data 5 were analyzed using two-tailed tests, and P<0.05 was considered statistically significant.   Each dot in panels A, B, C, D and E represents 1 patient or mouse. **P<0.01 for groups connected 7 by horizontal lines compared.