D29-07 Artificial Intelligence Aided Drug Discovery for the Treatment of Idiopathic Pulmonary Fibrosis

Abstract Introduction Idiopathic Pulmonary Fibrosis (IPF) is a progressive lung disease characterized by fibrotic remodeling that thickens alveolar walls and destroys alveolar spaces, with a median survival of about three years after diagnosis. Current FDA-approved therapies, Pirfenidone and Nintedanib, only slow disease progression without reversing fibrosis. We propose a systems-based approach to identify small molecules that reverse profibrotic gene signatures in IPF small airways to uncover key molecular drivers and develop effective therapeutics. Methods We used a single cell-based drug repurposing method for IPF treatment. One recent study conducted scRNA-seq on lung tissues from individuals with IPF, revealing heterogeneity within alveolar macrophages and epithelial cells in subjects with the IPF. A comparison between diseased samples and controls unveiled profound changes in the transcriptional features of all major cell types, including endothelial, epithelial, immune, and mesenchymal cells. We compiled the expression signatures derived from this analysis, as well as signatures obtained by comparing transitional alveolar type 2 (AT2) to AT1 cells and KRT5-/KRT17+ cells to AT1 cells in IPF patients. Additionally, we included signatures from bulk RNA-seq of our mouse models and human small airway tissues. Each signature was subsequently inputted into our drug repurposing pipeline, OCTAD, to predict drugs capable of reversing the specific signature. Results To ensure that we are accounting for all the heterogeneity present in IPF, we decided to compile several candidate lists using different approaches, such as drugs targeting a broad selection of cell types, drugs targeting specific pathological cell types and drugs targeting cell populations that are hypothesized to be involved in pathogenesis. Using our established drug repurposing pipeline, we identified Pyrithyldione and Bergenin that reversed the expression of myofibroblasts and the differentiating ciliated cell signatures. To evaluate the efficacy of the drugs, we assessed their ability to inhibit profibrotic markers in precision cut lung slice (PCLS) obtained from IPF lungs. Pyrithyldione and Bergenin significantly decreased the profibrotic protein expression of collagen (COL1A1), α-smooth muscle actin (α-SMA), fibronectin (FN1) and Collagen triple helix repeat containing 1 (CTHRC1). RNA-seq analysis showed that Pyrithyldione reversed disease gene expression, notably in MUC5B+ epithelial cells and myofibroblasts, highlighting epithelial cells as key therapeutic targets in IPF. In addition, we validated that Pyrithyldione has anti-fibrotic effect in a bleomycin induced lung fibrosis mouse model in vivo. Conclusion Collectively, this study suggests that targeting specific cell types by reversing the disease gene signature could lead to the inhibition of pulmonary fibrosis. This abstract is funded by: NHLBI- 1R61HL177451-01