Fig. 1: Study design.
The premise of a 4-step drug discovery pipeline is summarized on the top (Step 0) is a recently published9 Boolean implication network-based computational model of disease continuum states in inflammatory bowel disease (IBD map). The map, comprised of 6 gene clusters, was created and validated database containing 1497 gene–expression data (1263 human and 234 mouse samples). Paths, clusters and a list of genes in the network-based model were prioritized to discover one clinically actionable drug target (PRKAB1)9. Steps 1-4 outline the AI-guided identification and validation of another target pair, PPARA, and PPARG. Step 1: Dual agonists of PPARα/γ were predicted to—(i) modulate epithelial and macrophage processes; (ii) Citrobacter and chemical models of colitis were predicted as most optimal models; (iii) have high therapeutic index indicative of likelihood to succeed in Phase III clinical trials. Step 2: A combination of inhibitor and agonist studies helped establish that dual agonists reduce inflammation (PPARγ) while ensuring the induction of adequate immune response (PPARα). Step 3: Dural agonists ameliorated colitis in two preclinical models of colitis, and reversed the patterns of disease-associated gene expression that were altered in the IBD map. Step 4: In phase ‘0’ human pre-clinical trials, PBMCs from CD, but not UC or healthy showed defective microbe clearance; this defect was reversed with a dual agonist of PPARα/γ.
