Researchers have developed a drug-screening method that uses strands of DNA as tiny switches to identify molecules that change how proteins interact. The platform, called FIND, is designed to search large libraries for a desired biological effect rather than merely detect whether a candidate sticks to one target.
In one test, the team started with a library containing 400 million DNA-encoded small molecules and identified compounds that promoted a particular protein interaction. These are research leads, not approved medicines. The study does not show that a new treatment is safe or effective in patients.
Researchers at Harvard’s Wyss Institute, Harvard Medical School and Boston Children’s Hospital described the work in an announcement on November 5, 2026. The work was published in Science. Its most interesting feature is the way it keeps a candidate molecule linked to a record of what happened during its test.
Why search for protein interactions?
Proteins carry out many jobs by interacting with other molecules, including other proteins. A signal can begin when a molecule binds to a receptor, and that receptor can then influence another protein inside a cell. Changes in these interactions are involved in many normal processes and some diseases.
Drug discovery often begins with a large number of possible compounds and a question about what each one does. A molecule that binds to a target can be a useful starting point, but binding alone does not establish whether the interaction increases, decreases or meaningfully changes the target’s activity.
Imagine finding a key that fits into a lock. That observation matters, but it does not tell you whether the key turns the lock, jams it or has an effect under the conditions that matter. The analogy is simplified, yet it highlights why researchers may want to measure an effect on a protein interaction, not just attachment.
FIND stands for Functional Interaction-based Nanoswitch Discovery. It is intended to help make that distinction in a particular experimental setup. It does not remove the need for later biological testing, because activity in a controlled screen is only one part of the evidence needed for a treatment.
What is a DNA nanoswitch?
The team’s nanoswitch is made from a long strand of DNA arranged so that two potential protein partners are attached at different points. When the proteins do not bind, the switch remains in a more linear shape. When they bind and bring the attachment points together, the DNA forms a loop.
The shape is a readable signal of the interaction being studied. A candidate molecule that promotes binding can increase the proportion of looped switches. In the researchers’ setup, the candidate remains associated with its own switch and an identifying nucleic acid barcode.
This arrangement allows many candidates to be tested together while preserving a way to identify which candidate was associated with each result. The barcode is a label used by the researchers; it is not a claim that the candidate has already become a medicine or that its effects are known throughout the body.
How do researchers read the result?
The team used gel electrophoresis, a laboratory technique that separates molecules as they move through a gel under an electric field. In this experiment, looped nanoswitches moved differently from linear ones. Researchers could collect the relevant group and read its barcodes to identify associated candidates.
The benefit is organizational as well as chemical. Instead of running every possible compound as an entirely separate experiment, the method can bring many tests together and later recover information about which molecules were associated with the selected outcome.
A large pooled screen still needs controls. A signal can arise for reasons other than the desired effect, and a candidate that looks promising in one assay may fail in another. The team also checked its approach using a known receptor activator added to a mixture of barcoded controls.
What did the 400-million-molecule test show?
One demonstration involved PPAR-gamma, a receptor involved in processes including insulin sensitivity and fat storage. Researchers screened an encoded small-molecule library and identified compounds that promoted an interaction between the receptor and a fragment of a partner protein called CBP.
The researchers also observed increased lipid accumulation in cultured cells exposed to the compounds, consistent with activation of the receptor. That follow-up provides additional biological context beyond the initial switch result. Cultured cells, however, are not a person with diabetes, and the result does not establish a clinical benefit.
The size of the starting library shows the scale of the search, not the number of medicines discovered. Screening hundreds of millions of encoded candidates does not mean hundreds of millions were separately validated in patients, or that every candidate had an equal chance of becoming useful.
What about the coronavirus example?
The researchers also identified a nanobody that affected an interaction between the SARS-CoV-2 spike protein and its receptor on human cells. Nanobodies are small antibody-like proteins that can be useful as experimental tools.
This example shows that the platform can be applied to a different kind of protein interaction. It does not establish that the identified nanobody prevents infection in people, works against every viral variant or is ready to be used as a COVID-19 treatment.
That distinction is especially important with drug-discovery headlines. A molecule can be valuable because it helps scientists understand a mechanism even if it never becomes a product. Conversely, an eventual medicine would have to demonstrate far more than a measurable interaction in a laboratory screen.
What happens after a promising screen?
Researchers typically need to confirm an initial result using independent tests. They may ask whether a candidate acts on the intended target, whether it affects other targets, whether it reaches the right tissue and whether its chemistry can be developed into a usable formulation.
Safety is a separate problem. A molecule that changes a desired interaction in a test tube may be unstable, poorly absorbed or harmful at a useful dose. Studies in cells, appropriate animal models and carefully designed human trials each answer different questions, and many candidates do not progress through every stage.
The Wyss Institute says the technology has been licensed to a new startup co-founded by members of the research team. That commercial connection provides context for the development pathway and the researchers’ interest in the method. It does not independently establish performance or clinical value.
Where could this approach be useful?
The platform may be useful when researchers have a clearly defined protein interaction and want to look for molecules that change it. Its potential value depends on whether the nanoswitch faithfully represents the interaction and whether selected candidates work in more complex biological environments.
Some targets may be difficult to attach or study in the required configuration. Different molecules can interfere with a screen in unexpected ways. Practical adoption would also depend on cost, preparation, validation and the types of libraries that can be used.
The central insight is that a screening platform can encode both an experiment and the identity of its candidate in the same molecular setup. FIND is an early tool for making the search more informative. Whether particular leads become treatments remains a question for future research.
Explore the book: How to Do Hard Things: The Discomfort Trap
Comments are closed.