Biohub 用全基因组 CRISPR 筛选结合 AI 发现银屑病新药靶点

Biohub researchers performed what they believe is the first genome-wide CRISPR study of primary human adult skin cells, then used an AI model to mine the results for overlooked drug targets for psoriasis. They found an unlikely candidate: the receptor for oxytocin, a hormone involved in childbirth, and sometimes called the “love hormone” for its additional role in social bonding.

As reported in Nature Communications, when the team formulated topical gels containing compounds that would inhibit the oxytocin receptor, as well as another promising target, both gels reduced psoriasis inflammation in mice as effectively as an injected drug used by many patients.

“A CRISPR screen at this scale, guided by AI, gave us answers the field has never had access to before, as well as a surprisingly effective potential treatment,” said Shana O. Kelley, president of bioengineering and head of Biohub in Chicago, where scientists are decoding the inflammatory processes that drive a wide range of diseases. Kelley co-led the study with Abdalla M. Abdrabou, lead scientist for Biohub’s Functional Immunogenomics group.

“We think of this as a blueprint,” added Abdrabou. “The specific targets we found in psoriasis are exciting, but the method itself is what we are most proud of. This is a generalizable way to find treatments that those of us studying disease wouldn’t even know to look for.”

Psoriasis affects roughly 125 million people worldwide. For mild cases, topical therapies, such as steroid creams, can often manage symptoms effectively. But for the many patients with moderate-to-severe disease, systemic therapies become necessary — and all come with trade-offs. “Biologics” that target immune signaling proteins must be injected, can cost tens of thousands of dollars per year, and carry risks from broad immune suppression. Recently approved therapies in pill form may reduce the burden of injections, but the drugs still circulate throughout the body, affecting immune pathways beyond the skin.

What has been missing is an approach that acts locally in the skin cells where psoriasis begins, targeting disease pathways specific enough to not broadly perturb immune function. Getting there requires understanding the biology of keratinocytes — the outermost skin cells that are the primary site of psoriatic disease — at a level of detail the field has lacked, because these cells have been stubbornly resistant to large-scale genetic screening using CRISPR.

The standard chemicals used in research to deliver genetic cargo into cells are toxic to keratinocytes even at low concentrations, which has made genome-wide CRISPR screening in these cells impractical. The Biohub team solved the problem by developing a method that uses forces generated by a centrifuge, rather than chemicals, to deliver CRISPR into cells.

Using a library of approximately 77,000 guide RNAs, the researchers knocked out roughly 19,000 genes, one by one, in keratinocytes from two human donors, and measured how each gene affected IL17RA, a protein that helps skin cells respond to inflammatory signals involved in psoriasis. Gene sequencing revealed which knockouts changed the abundance of IL17RA in the cells, producing a genome-wide map of IL17RA regulators in skin cells likely to be most relevant to disease.

A genome-wide screen produces thousands of hits, most representing known biology. To find the genuinely novel ones — mechanisms no one has previously linked to the disease — the team used a tool they built called VirtualCRISPR, a large language model trained on the published scientific literature. For each gene, the model was asked whether existing research would have predicted it as a hit. Genes that were strongly implicated in the CRISPR screen but had received little or no attention in the literature were prioritized for further research in the hopes that they might point to novel disease biology for psoriasis.

Two genes for which drug development seemed highly practical cleared this bar. The first was ALOX5, encoding an enzyme targeted by zileuton, an FDA-approved asthma drug. The second was OXTR — encoding the oxytocin receptor — which had no established connection to psoriasis, or to skin inflammation of any kind. A compound known as cligosiban inhibits the actions of OXTR.

The team validated both targets in three-dimensional skin cultures invented at Biohub that replicate the layered architecture of real human skin, and found that zileuton directly reprogrammed keratinocyte metabolism — suppressing inflammatory and proliferative gene programs — without any immune cells present, demonstrating that the drug acts on skin cells themselves rather than through secondary immune effects.

When the scientists formulated zileuton and cligosiban as topical gels and tested them alongside a commonly used, injected anti-IL17RA drug in a mouse model of psoriasis, both gels reduced disease severity in five days. In a week, both had achieved results comparable to the injected drug: the skin had regained its normal thickness, and immune-cell signaling had pivoted toward anti-inflammatory states. Cligosiban additionally produced the broadest restoration of the skin, suggesting that OXTR blockade may address the “barrier dysfunction” underlying chronic psoriasis, not just its inflammatory symptoms.

Because zileuton and cligosiban target keratinocyte-specific pathways with no role in overall immune function, topical delivery should act locally, without the broad immune suppression of current systemic therapies. And the compounds are safe: zileuton’s safety profile is established from decades of asthma use, and OXTR antagonists — including those used to delay preterm labor — also have established human safety records, which could accelerate clinical investigation of OXTR blockade in dermatology.

“This work exemplifies Biohub’s approach to decoding the biology of inflammation in the search for new therapies,” Kelley said. “Combining AI with sophisticated experimental techniques accelerates the entire process of disease research, and will help get new treatments, for psoriasis and other conditions, to patients faster.”

注意转载授权 · 阅读原文