The gene therapy that cannot be stopped

The drug works. You can stop taking it. The body returns to normal. That is not a feature every biotech entrepreneur thinks much about, until you propose making the effect permanent.

Addition Therapeutics, a South San Francisco startup spun out of Professor Kathleen Collins's lab at UC Berkeley, disclosed this week that it is developing a gene therapy to make the body produce GLP-1 continuously, without injections, through a one-time procedure. GLP-1 is the hormone behind Ozempic, Wegovy, and the rest of the obesity drug boom. Addition has been building this quietly since 2021. It has raised $106.5 million, including backing from the Gates Foundation, and now it is ready to talk about the obesity program.

The technology is unlike anything currently in clinical use. Addition uses R2, a retrotransposase protein found naturally in birds, to physically write a therapeutic gene into the human genome. The company delivers R2 and its genetic template together as RNA, packaged in lipid nanoparticles: the same oil-bubble delivery system used in mRNA COVID-19 vaccines. Once inside the cell, R2 converts the RNA template directly into DNA and inserts it into a specific site, the ribosomal DNA. The rDNA is a repetitive stretch of genome containing hundreds of near-identical copies of the genes that build the cell's protein factories.

The elegance of that target site is what Collins, the scientific founder, has described as a "privileged" safe harbor. Disrupt one copy out of several hundred and the cell does not notice. The nucleolus, the cellular compartment housing the rDNA, also maintains a stringent DNA repair environment, which the company argues limits the risk of cancer-causing mutations at the insertion site. "It is multicopy, conserved, and a safe harbor in the sense that you can disrupt one of these copies, and the cell does not care," Collins said. The company calls the platform PRINT: Precise RNA-Mediated Insertion of Transgenes.

There is a second advantage that distinguishes PRINT from the dominant gene therapy architecture. Most gene therapies in development use adeno-associated viruses, or AAVs, as their delivery vehicle: engineered shells that ferry DNA into cells. AAVs have a real problem. A substantial fraction of the human population carries pre-existing immune antibodies that neutralize them before they reach their targets. Manufacturing AAVs is also expensive and complex, which contributes to gene therapy price tags in the hundreds of thousands or millions of dollars per patient. PRINT uses no virus. Because the R2 mechanism inserts by reverse-transcribing directly into the genome, there is no extrachromosomal DNA at any stage, and therefore no innate immune trigger. The all-RNA payload goes in via LNP; the gene goes into the genome; the RNA components are degraded.

What remains is permanent.

That is both the point and the problem.

Fractyl Health, a competing biotech developing AAV-based GLP-1 gene therapies under the Rejuva platform, has thought hard about this. Its lead candidate RJVA-001 targets type 2 diabetes and has a clinical trial authorization filed in Europe, with first-in-human dosing expected in the second half of 2026. Its second candidate, RJVA-002, targets obesity with a dual GIP and GLP-1 hormone approach and demonstrated approximately 30 percent weight loss in preclinical studies after a single administration in mice. Fractyl's design is built around nutrient-responsive gene expression: the gene activates when blood sugar rises after eating, mimicking natural hormone secretion rather than constant production. That design choice is not incidental. It is an attempt to build a self-regulating circuit into a permanent intervention.

Addition has not disclosed its approach to expression control, and the Endpoints reporting that broke the news this week was behind a paywall. What is public, through the company's December 2025 press release and scientific publications from the Collins lab, is that PRINT is a platform with no human data yet. The company planned initial disease-related studies in non-human primates for 2026.

The deeper issue both companies are navigating is what happens when a permanent gene therapy gets the dose wrong in a patient. GLP-1 agonists are already known for side effects, including nausea, vomiting, and gastroparesis, that are dose-dependent and manageable precisely because patients and physicians can titrate the drug or stop it entirely. A therapy that produces GLP-1 continuously, written into the genome, removes that option. Collins herself acknowledged in a 2024 interview with SynBioBeta that cell-specific dynamics in rDNA are not yet fully characterized: "we have to understand a little bit more about the biology of our rDNA." That is a reasonable admission of scientific humility for a preclinical company. It is also the most important unsolved question in the field.

The GLP-1 gene therapy field is forming around two structural bets. The viral camp, led by Fractyl and others using AAV, is building delivery methods that target specific tissues, with expression logic intended to keep the therapy proportionate. The non-viral camp, represented by Addition, is betting that cleaner delivery and a better immune profile clears the path to scale: no immune ceiling, no manufacturing bottleneck, potentially far lower cost. Both camps are pre-human. Both share the same fundamental design problem they haven't fully solved.

Addition was founded in 2021 by Ron Park, a physician and former Roche executive who is chief executive, and Collins, who remains the scientific founder. The company's chief scientific officer is Francine Gregoire, who came from Beam Therapeutics, CRISPR Therapeutics, and Moderna, according to BioPharma Dive. Its investor syndicate includes SR One, Pivotal Life Sciences, Abingworth, Osage University Partners, the Gates Foundation, and BEVC. The Gates Foundation's participation reflects a second program running in parallel: a potential HIV treatment that would use PRINT to insert an HIV-resistance gene. The obesity program is the attention-getter. The HIV program is the one that explains why the Gates Foundation wrote a check.

What to watch: whether Addition's NHP data, expected sometime this year, shows durable expression and acceptable safety in a primate model. That is the minimum evidence needed before anyone should take a human application seriously. Fractyl's first-in-human results from RJVA-001, expected in the second half of 2026, will be the field's first real data point on GLP-1 gene therapy in people. Both companies will be watching what the other one learns.