RNA-X: Modeling RNA interactions to design binder RNA and simultaneously target multiple molecules of different types

RNA-X, an RNA interaction foundation model, designs binder RNAs that target proteins, other RNAs, and DNA, including a guide RNA with predicted binding energy comparable to wild type.

• Why it matters: Selective, strong RNA binding is hard because sequence space is vast and current experimental and computational tools are limited. Modeling RNA interactions is needed for both understanding biology and creating RNA therapeutics.
• What they did: RNA-X uses masked language modeling for representation learning, conditional RNA design, and optimization, enabling design against multiple targets at once. It was tested in vitro on prostate cancer cells with an RNAX-designed RNA targeting the 5'UTR of androgen receptor mRNA, and further evaluated with affinity predictors and molecular dynamics simulations.
• The result: The RNAX-designed RNA reduced both AR transcript and protein levels in prostate cancer cells, and simulations showed designs targeting p53 and thrombin proteins and binding PHF6 without prior interaction data. As a proof of concept, a guide RNA designed from scratch simultaneously binds bacterial DNA and Cas9 protein with predicted binding energy comparable to wild-type guide RNA in the Protein Data Bank.

AI summary of the abstract · may contain errors · not medical advice · read the paper.