Zuckerberg-Chan Biohub uses CRISPR and AI to find new psoriasis drug targets

On 8 July 2026 Biohub, the science organisation created by Mark Zuckerberg and Priscilla Chan through the Chan Zuckerberg Initiative, announced results published in Nature Communications that point to unexpected new ways to treat psoriasis. The chronic inflammatory skin disease affects roughly 125 million people worldwide.

Keratinocytes, the outer skin cells where psoriasis begins, have long resisted large-scale genetic screening because the chemicals normally used to slip CRISPR gene-editing tools into cells are toxic to them. The Biohub team in Chicago, led by Shana Kelley and Abdalla Abdrabou, got around this by using centrifugal force instead. They then switched off about 19,000 genes one at a time in skin cells from two human donors and measured how each change affected IL17RA, a receptor that helps skin cells respond to the inflammatory signals behind psoriasis.

A screen of that size throws up thousands of hits, most of them already known. To find the surprising ones, the researchers built an AI tool they call VirtualCRISPR, a language model trained on the scientific literature, and asked it which genes earlier research would have predicted. Genes with a strong effect in the experiment but little prior attention were flagged as promising. Two stood out as practical drug targets: ALOX5, an enzyme already blocked by the approved asthma drug zileuton, and OXTR, the receptor for oxytocin, a hormone best known for its roles in childbirth and social bonding and never before linked to skin inflammation.

After confirming the findings in lab-grown three-dimensional skin, the team made topical gels containing compounds that block each target. In mice, both gels reduced psoriasis-like inflammation about as well as an injected antibody therapy widely used by patients. The researchers stress that the work is preclinical, and they describe the method as a general blueprint for finding treatments for other diseases.