Biotechnology · us
Searching 400 Million Compounds for Drug Leads: DNA Nanoswitches Read Out Protein Interactions
A molecule that binds to a target does not necessarily change its function. The FIND platform converts protein interactions into changes in DNA shape to identify candidate compounds with cellular activity, opening a new path for early drug screening.
When searching for new drugs, finding a molecule that binds to a disease-related protein is only the starting point. The harder question comes next: can this molecule change how the protein functions? The FIND platform, developed by a Harvard University research team, translates whether two proteins bind more readily into a separable, traceable signal based on DNA shape, bringing large-scale screening closer to the biological effects researchers seek.
Published in *Science* on November 5, 2026, the study was a collaboration among Harvard University's Wyss Institute, Boston Children's Hospital, Harvard Medical School, and other institutions. Built around DNA nanoswitches, FIND places a candidate molecule and a pair of proteins on the same device and records the candidate molecule's identity with a nucleic acid barcode, allowing many tests to run in parallel in a pooled sample.
The switch is read by its shape. When two proteins attached at different positions on the DNA bind to each other, they pull the DNA into a loop; if a candidate molecule promotes this interaction, the proportion of looped DNA increases. Researchers then use gel electrophoresis to separate looped and linear DNA, collect the desired form, and sequence the barcodes to identify the corresponding molecules. This keeps the screening results linked to the candidate molecules responsible for the changes.
Using the diabetes drug target PPARγ as a demonstration, the team screened a DNA-encoded small-molecule library containing 400 million members and identified compounds that promote binding between PPARγ and a fragment of its partner protein CBP. These molecules also increased lipid accumulation in cultured cells, consistent with the effects of PPARγ activation. However, this cellular result does not equate to improved blood glucose, nor has it established that the candidate molecules offer therapeutic benefits to patients.
The study also demonstrated other applications: the team identified nanobodies that modulate interactions between the SARS-CoV-2 spike protein and a human receptor, and recovered a known β2-adrenergic receptor agonist that had been added to a sample mixed with barcoded controls. These experiments provide evidence of feasibility in different systems; the viral protein demonstration cannot be interpreted directly as an antiviral treatment outcome.
Commercialization has also begun. According to the Wyss Institute announcement and the version published by Brightsurf, the technology has been licensed to a new company co-founded by members of the research team, but neither the company's name nor the terms of the deal were disclosed; Northpond Labs supported the related program through its collaboration with the Wyss Institute. Both Phys.org and Brightsurf published institution-supplied news, representing dissemination of the same announcement rather than two independent verifications.
FIND's advance lies in bringing changes in specific protein interactions into large-library screening, identifying leads worth further investigation at an earlier stage. However, multiple layers of validation still separate interactions in a simplified device, responses in cultured cells, and therapeutic effects in humans. The candidate molecules' selectivity, safety, and effects in vivo will determine whether these leads can continue toward drug development.