AGA Abstracts Table 1 Helminth naïve Helminth exposed p value MLN MLN % CD4+ YFP+ cells 8.9 ± 1.3 % 4.5 ± 0.3 % < 0.05 % CD4+ YFP+ cells 3.3 ± 1.4 % 0.8 ± 0.2 % < 0.05 expressing IFNγ % CD4+ YFP+ cells 3.1 ± 1.6 % 8.8 ± 1.5 % < 0.05 expressing IL10 %CD4+ YFP+ cells 3.6 ± 0.5 % 8.1 ± 1.7 % < 0.05 expressing Foxp3 % proliferation on co-culture with 19.4 ± 3.2 % 7.7 ± 2.1 < 0.05 YFP+ cells from % suppression of in 29.0 ± 4.2 % 66.7 ± 3.3 % < 0.01 vitro proliferation * T cell transfer coli- tis score** with co- 3.6 ± 0.5 0.5 ± 0.2 < 0.05 administered YFP + from Results are mean + SE from at least three separate experiments. *compared to mitogen- stimulated spleen cells without any co-cultured YFP+ cells, ** 0-4 histology score; 0 = normal, 4 = consistent ulceration, transmural inflammation, and mucus depletion 376 INTERLEUKIN-33 PROMOTES REG3G EXPRESSION IN INTESTINAL EPITHELIAL CELLS AND REGULATES GUT MICROBIOTA Wenjing Yang, Yi Xiao, Xiangsheng Huang, Suxia Yao, Zhanju Liu, Yingzi Cong Background: Regenerating islet-derived protein 3-gamma (REG3G), an antimicrobial pep- tide, plays crucial roles in intestinal homeostasis and controlling gut microbiota. However, the mechanisms which regulate IEC expression of REG3G are still not completely understood. In this study, we investigated whether and how IL-33, an alarmin produced by IEC in response to injury, regulates REG3G expression in IEC, thus, contributing to the maintenance of intestinal homeostasis. Methods: IECs were isolated from wild-type (WT) and IL-33 -/- mice to determine expression of REG3G and other antimicrobial peptides by qRT-PCR and Western blot. Mouse and human IEC cell lines and mouse small intestinal enteroids were used for mechanistic studies. CRISPR/Cas9 was used to knock out genes in IEC cell lines. Gut microbiota was measured by 16S rRNA pyrosequencing. Citrobacter rodentium was used both in vitro and in vivo for functional assay of IL-33 REG3G axis. Results: We demonstrated that the expression of REG3G, but not b-defensins, in IECs of IL-33 -/- mice was significantly lower than WT mice, either under normal condition or after Citrobacter rodentium infection. IL-33 promoted REG3G expression in MSIE cells, an immortalized mouse small intestine epithelial cell line, mouse intestinal enteroids, and human HT-29 cells, in a dose and time dependent manner. Furthermore, IL-33 induces IECs REG3G synergistically with IL-17 and IL-22 both in vitro and in vivo. Consistent with in vivo data, REG3G expression was greatly decreased in IL-33 knockout (IL-33KO) MSIE cells compared to WT cells. IL-33 induced REG3G production in IL-33KO MSIE cells but at a lower level compared to WT cells, indicating that IL-33 stimulates IEC production of IL-33 to promote REG3G production. Mechanistically, IL-33 activated STAT3, mTOR, and ERK1/2 in IECs, and inhibition of these pathways abrogated IL-33 induction of REG3G. Moreover, IL-33 -/- mice demonstrated lower total bacteria load, and altered microbiota composition in feces, with increased levels of Bacteroidetes, and Firmicutes, and decreased levels of Proteobacteria and Deferribacteres. Inter- estingly, IL-33 did not directly inhibited bacterial growth, but promoted WT, not REG3GKO, IECs to kill bacteria, and inhibit bacterial growth in vitro. Consistently, Citrobacter rodentium infection induced IEC IL-33 expression, and IL-33 -/- mice demonstrated an impaired bacterial clearance when mice were infected with Citrobacter rodentium. Conclusions: Our study demonstrated that IL-33, which is produced by IECs in response to injury and inflammatory stimulation, in turn, promotes IECs expression of REG3G, which contributes to the mainte- nance of intestinal homeostasis and regulates gut microbiota. 377 PROTECTIVE AND POTENTIALLY AGGRESSIVE BACTERIAL SPECIES OCCUR SIMULTANEOUSLY IN HEALTHY HOSTS Bo Liu, Akihiko Oka, Jisun Jang, Lisa Holt, Graham J. Britton, Jeremiah Faith, R. Balfour Sartor Introduction: Resident microbiota include a spectrum of bacterial species with diverse biological activities. In normal hosts inflammation is regulated by balanced diverse microbial community structure and homeostatic mucosal responses induced by bacterial components and metabolic products, but chronic inflammation occurs when microbial community struc- ture and immune responses are disturbed in genetically susceptible hosts. Hypothesis: Healthy human gut microbiota contain subsets of aggressive and protective bacterial species. Methods: We screened 19 bacterial strains isolated from a healthy human subject in gnotobi- otic il10 egfp reporter and il10 -/- mice to identify high and low IL-10/IFNγ ratio-inducing bacterial strains. We compared in vivo functions of the 3 highest (HI) and 4 lowest (LO) IL-10/IFNγ ratio strains in selectively- colonized gnotobiotic il10 egfp reporter mice and 2 chronic colitis models (ex-GF il10 -/- or il10 egfp CD4 + /CD19 + cells transferred to il10 -/- ;rag2 -/- mice ). We measured colonic lamina propria (LP) regulatory T and B cell subsets by FACS, colonic immunoregulatory gene expression and colitis activity by blinded histologic scoring, spontaneous secretion of cytokines from cultured colonic strips and colonic mRNA expres- sion. We measured relative luminal bacterial strain concentrations in cecal contents using strain- specific qPCR. Results: The 3 HI strains failed to induce colitis, while the 4 LO strains induced moderate to severe colitis in both models; the combined 3 HI and 4 LO S-78 AGA Abstracts bacterial strains exhibited protection and reversed active colitis (Fig.1A, B). In il10 egfp mice the 3 HI strains increased frequency of IL-10 + producing cells, including CD4 + Foxp3 + , CD4 + Rorγ + Foxp3 + , Tr1 cells and CD19 + B cells (Fig. 2), and increased mRNA expression of il10, tgfb, apoe, stab1 and klrg regulatory immune genes in colonic LP B and T cells. In contrast, the LO bacteria strains increased ifng, il12p40, il17 and tnfa expression in selectively colonized il10 -/- mice (Fig.1B), induced fewer regulatory cells (Fig. 2) and decreased protective immune gene expression in the LP of il10 egfp mice. We replicated results with 3 different high and 3 low ratio bacterial strains isolated from a separate normal individual. qPCR luminal bacterial profiling showed dominance of 1 HI strain in the HI, prevention and treatment groups, and consistent preferential colonization of the 3 HI strains during competi- tive colonization with the LO strains in both prevention and treatment groups (Fig. 1C). Conclusion: Normal human microbiota contain small subsets of functionally active, predomi- nantly protective and aggressive bacterial species identified by induction of high and low IL-10/IFNg ratios by combined in vivo / ex vivo screening in gnotobiotic mice. Manipulating the balance of these pivotal functional strains modulates mucosal homeostasis and colitis activity. 400a EFFECT OF INITIATING ASPIRIN ON CANCER EVENTS IN THE HEALTHY ELDERLY: PRIMARY RESULTS FROM THE ASPREE RANDOMIZED CONTROLLED TRIAL Andrew T. Chan, Peter Gibbs, Suzanne Orchard, Jessica Lockery, Andrew Haydon, Rory Wolfe, Robyn Woods, Finlay A. Macrae, John Zalcberg, Joanne Ryan, Galina Polekhina, Wendy Bernstein, Yin Cao, Jeanne Tie, Josie van London, Mark Nelson, Christopher Reid, Brenda Kirpach, Raj Shah, Anne Murray, Ellen Richmond, Leslie Ford, Asad Umar, John McNeil Background: Compelling evidence largely derived from middle-aged adults (<70 years old) demonstrates that aspirin reduces risk of cancer, particularly colorectal, over long-term follow-up. Recently, the ASPirin in Reducing Events in the Elderly (ASPREE) study, a placebo-controlled randomized trial of daily low dose aspirin (100 mg) in older adults, showed a surprising increase in the secondary endpoint of all-cause mortality, primarily due to excess cancer deaths (McNeil, NEJM 2018). For the first time, we report the effect of aspirin treatment on primary incident cancer events, a prespecified secondary endpoint.