ARPA-H bets $4.4M that mRNA can be turned into a controllable dial

ARPA-H, the U.S. health agency modeled on DARPA that funds high-risk biomedical bets, has awarded up to $4.4 million to test whether an mRNA molecule can be dialed up or down with a small-molecule drug. The mechanism borrows a piece of cell biology that bacteria already use and that, in human cells, is still mostly uncharted.

The pilot sits inside PROPEL, Programmable RNA for Optimal Precision in Therapeutic Efficacy and Localization. The company on the team is RNAV8 Bio (pronounced Renovate Bio), a small firm whose founder and CEO, Devan Shah, has been the named spokesperson in the trade press. ARPA-H's program manager is Shannon Greene, PhD.

Every mRNA molecule has parts that don't code for protein, called untranslated regions or UTRs, that fold into three-dimensional shapes. Those shapes decide how much protein the cell actually makes from the message. In bacteria, certain UTR folds double as on-switches: when a small molecule binds one, the fold rearranges and protein output changes. These are called riboswitches, and they let a bacterium tune its own genes with chemistry it already has on hand. The PROPEL team's hypothesis is that human mRNA can be engineered the same way: a clinician delivers the mRNA once, then uses a separate small-molecule drug to turn protein output up, down, or off at will. The trade-press write-up of the award captures the framing plainly as a "dose becomes a dial" on protein expression, per the company's release.

The Rouskin Lab at Harvard, run by Silvi Rouskin, has spent years measuring how RNA actually folds inside cells, the half of the work that determines whether a small molecule can find a target. The Weissman Lab at MIT, run out of the broader Weissman orbit now that Melissa Moore has moved on, is one of the most cited groups in the world on mRNA delivery, including the lipid nanoparticle chemistry behind the COVID vaccines, the half that gets the mRNA into the cell in the first place. RNAV8 Bio's job is to screen protein output from pools of transfected mRNA and engineer UTRs in the delivery format. The company's named validation targets are cell-type-selective expression of gene-editing enzymes and engineered immune-cell receptors used in some cancer therapies, the CAR constructs familiar from CAR-T cell therapy.

Most approved genetic medicines, including gene replacement, CRISPR-based editing, and even base editing, change DNA, and that change is permanent. A controllable RNA switch leaves the genome alone, is transient by design (the mRNA is degraded like any other), and would let a clinician adjust dose after delivery, the way an insulin pump adjusts insulin. The HMS announcement relayed by Newswise frames PROPEL explicitly as a "different kind of control."

The team itself flags the bacterial-to-human-cell translation as "largely uncharted." The riboswitch logic is well-mapped in bacteria and has only begun to be tested in human cells. Independent peer-reviewed validation of drug-tunable UTR control in human cells is not in the public record cited by the award, and what works at bench scale often fails when scaled up or moved into animals. The award is structured as a one-year pilot specifically to find out whether the hypothesis survives contact with human cells, not to deliver a drug.

The pilot is also small, and that is the point. ARPA-H's broader RNA portfolio is heavier on delivery and manufacturing. The agency is running a separate, larger program, covered by BioSpace, that could pay out up to $125 million for personalized RNA manufacturing, a different problem with a different budget. The $4.4 million here is not meant to build a manufacturing line. It is meant to test whether a single mechanistic idea, that human mRNA can be made drug-tunable, is even sound. The team's next milestone is a one-year readout.