AI-Guided CRISPR Screen Accelerates Discovery of New Drug Targets

Abstract

Psoriasis affects over 125 million people worldwide, yet the mechanistic understanding of keratinocyte-driven inflammation remains incomplete, limiting therapeutic innovation beyond costly systemic biologics that are prone to side effects. Here, we performed the first genome-wide CRISPR knockout screen in primary human adult epidermal keratinocytes to systematically identify regulators of IL-17 receptor A (IL17RA), a central node in psoriatic inflammation. To prioritize therapeutically tractable targets from over 19,000 screened genes, we integrated a large language model - VirtualCRISPR - trained on functional genomics data, identifying arachidonate 5-lipoxygenase (ALOX5) and oxytocin receptor (OXTR) as high-confidence novel hits with minimal prior association with psoriasis. Multi-omics validation revealed that ALOX5 and OXTR regulate IL17RA expression through distinct signaling pathways - ALOX5 through lipid mediators that stabilize the receptor at the cell surface, and OXTR through calcium signaling that reprograms cellular metabolism. Topical delivery of their inhibitors Zileuton (ALOX5) and Cligosiban (OXTR) exhibited therapeutic efficacy comparable to systemic anti-IL17RA antibody in the imiquimod-induced psoriasis model, suppressing pathogenic Th17/Tc17 responses, polarizing macrophages toward anti-inflammatory phenotypes, and normalizing epidermal hyperproliferation. Proteomic profiling in human 3D organotypic skin and murine models confirmed on-target pharmacology and revealed convergent suppression of neutrophil-keratinocyte inflammatory circuits. The use of VirtualCRISPR significantly shortened the timescale from screen to the identification of druggable hits with robust validation, and this work establishes a blueprint for integrating AI-driven target prioritization with functional genomics to accelerate therapeutic discovery.