USING AN ARTIFICIAL INTELLIGENCE (AI) AND MACHINE LEARNING (ML) PLATFORM TO IDENTIFY MAST CELL FOCUSED THERAPEUTIC TARGETS AND ASSOCIATED GUT-LIVER-BRAIN AXIS INDICATIONS

Background: Abdominal inflammatory pain is a common and persistent symptom in inflammatory bowel disease (IBD, up to 60% of patients experience chronic abdominal pain regardless of disease activity). Pain can arise from different mechanisms and persists due to changes in afferents sensory neurons activating threshold. A key role in this process is played by a protease activated receptor 2 (PAR2), a G-Protein-Coupled Receptors (GPCRs), which upon activation on the plasma membrane internalize in endosomes where it continues to signal. AIM: To test the hypothesis that blocking PAR2 endosomal signaling can reduce inflammation and pain in preclinical rodent models of IBD. Methods: T-84 and HEK293 cells were used to assess PAR2 internalization, activation and signaling blockage in intracellular compartments. DSS cand TNBS models of colitis were used to characterize colonocytes PAR2 trafficking from plasma membrane to endosome and to study the role of endosomal PAR2 persistent signaling in inflammatory pain in the colon of C57BL/6 and PAR2-GFP mice. Immunofluorescence was used to localize the receptor on the plasma membrane, and following DSS and TNBS, the trafficking of the receptor to endosome. Nanotechnology delivering cargo (PAMAM_G3-Az3451, a par2 antagonist) was used to specifically target the receptor in the endosome and inhibit endosomal signaling to study the effect of persistent PAR2 endosomal signaling in preclinical rodents models of IBD.
Results: In vitro: in T-84 and HEK293 cells, upon stimulation with PAR2 agonist, the receptor internalize in endosomes and other intracellular compartments, PAMAM_G3-AZ3451 inhibit receptor to signal from endosome, while the "free" AZ3451 drug inhibit plasma membrane signal but failed to inhibit endosomal PAR2 signal. In vivo: PAR2 is activated and internalized into endosome in a DSS and TNBS preclinical models of IBD. PAMAM_G3-AZ3451 nanoparticles, injected by enema into the colonic lumen are able to inhibit sustained PAR2 endosomal signaling, abdominal pain (measured using abdominal Von Frey filament stimulation) and to reduce proinflammatory cytokine levels (measured using qRT-PCR), free drug instead has little or no effect. Summary and Conclusion: in IBD, PAR2 receptor is activated and internalized to endosome, where it continues to signal persistent inflammation and pain. inhibition of endosomal signaling using smart delivery system, blocked PAR2 endosomal signaling. Inflammation and pain were greatly reduced. Drugs should target the receptor not only on the plasma membrane, where the signal is quickly desensitized, but also in intracellular compartments where the signal is sustained. Many GPCRs FDA approved drugs to treat pain, fail to cure chronic pain. A better understanding of GPCRs endosomal signaling is may needed to improve management of chronic pain and to discover new therapies.

Background: The circuitous network of anatomical, cellular, and molecular connections between the central and enteric nervous systems, gastrointestinal (GI) tract, and hepatobiliary (HB) system, is known as the gut-liver-brain axis, or simply the gut-brain axis (GBA). GBA dysregulation and the resulting neuroimmunoendocrine imbalance lead to symptoms such as abdominal pain, diarrhea, swallowing problems, and itching, which are associated with disorders including inflammatory bowel disease (IBD) and certain fibrotic conditions. These symptoms are also consistent with increased mast cell activity. Mast Cells (MCs) are emerging as a key component of the GBA. Triggered by external stressors, MCs can amplify and advance inflammatory and fibrotic responses through the release and activation of proteases, which in turn stimulate the production of inflammatory cytokines, alter epithelial barrier integrity, promote infiltration of immune cells, and activate fibrotic factors. The complexity of the GBA and its impact on GI disorders provide a clear rationale to deploy AI to develop precision therapeutics. Methods: We utilized a proprietary AI/ML platform that uses natural language processing (NLP) algorithms to identify therapeutic targets downstream of MC activation, as well as candidate GI and HB indications for a MC modulator. This platform was trained on hundreds of thousands of sentences and extracts biological entities, including cell types, genes, and diseases, and scores both directional and inferred relationships between entities from the literature to build testable hypotheses. Results: Our platform identified high-scoring associations between MC activation and downstream inflammatory/fibrotic targets including TNF, IL6, IL1β, TGFβ, and chymase. We selected chymase for indication prioritization based on its preferential expression by MCs and NLP scores. By combining the NLP scores for MC activation to GI and HB indications with unmet medical need analysis, we identified IBD and certain fibrotic conditions as candidate indications. Using a chymase inhibitor, we validated our in-silico results in vivo in mouse models. In both cases we observed significant symptom amelioration and reduction in markers of inflammation for certain fibrotic conditions, suggesting chymase inhibition as an attractive therapeutic avenue for these chronic conditions. Conclusions: Using our AI platform, we identified chymase as a candidate therapeutic target for selected GBA-related disorders. Our preliminary preclinical data support a significant role for chymase in IBD.