What the Study Found
- A single injection turned mouse muscle into a long-running factory for a weight-loss drug, with useful levels still circulating 70 days later.
- The treated mice showed better blood sugar control and less weight gain over the months the hormone kept producing.
- The approach points toward gene therapies that work for months from one dose, instead of weekly injections.
Inject a loop of DNA into a mouse’s leg, pulse the muscle with a quick jolt of electricity, and the cells start doing something they were never built to do. They begin reading the foreign instructions, stitching together a GLP-1 weight-loss drug, and pumping it into the bloodstream. Not for a day or two. For weeks. In the mice that got the shot, a hormone mimic was still circulating at useful levels 70 days later, no top-up required.
That’s the headline result from a team at The Wistar Institute in Philadelphia, published in Trends in Biotechnology. The idea is to stop handing the body a drug and start handing it the recipe.
Drugs like semaglutide, the molecule inside Ozempic and Wegovy, work by imitating incretins, the gut hormones GLP-1 and GIP that tell the pancreas to release insulin and tell the brain you’ve eaten enough. Native incretins are gone in minutes; the enzyme DPP-IV chews them up almost as fast as the gut makes them. The pharmaceutical fix was to redesign the peptide so it survives for days instead of minutes, which is why a weekly injection works at all. The Wistar approach sidesteps the half-life problem entirely. Instead of a peptide that decays, you get a gene that keeps expressing.
Why Once Beats Weekly
“We want to deliver a drug once and have it work for a really long time,” says Ebony Gary, the study’s first author. The DNA platform, she argues, has shown it can do exactly that.
Here is roughly how it goes. The researchers wrote synthetic plasmids, small circular pieces of DNA, encoding GLP-1 or GIP with a tweak at the second amino acid so DPP-IV can no longer cleave them, then fused each hormone to a fragment of an antibody (the Fc region) to make the protein bigger, longer-lived, and harder to flush out. They call the constructs pLincretins. A shot into the muscle, a burst from an electroporation device that briefly opens pores in the cell membranes, and the genetic instructions slip inside. After that the muscle does the manufacturing.
The mouse data are the interesting part. A single dose drove down body weight, food intake, and blood glucose, and those effects held after the observation window closed. Run head-to-head against semaglutide, the contrast was stark: the semaglutide mice started regaining weight the moment dosing stopped, while the DNA-treated animals kept their losses. Gary’s framing is that instead of giving a drug the body clears, you give cells the instructions and they keep making it.
Designer Hormones, Drawn by an AI
Then the team did something the platform makes unusually easy. Having a gene you can rewrite, instead of a peptide you have to synthesise from scratch, means you can design proteins that don’t exist in nature and just ask the muscle to build them. Using AlphaFold 3 and a technique called synthetic consensus design, which lines up GLP-1, GIP, and existing incretin drugs and keeps whichever amino acid shows up most often at each position, they assembled a single molecule meant to hit both the GLP-1 and GIP receptors at once. This is the same dual-action trick behind tirzepatide, the drug in Mounjaro. The new construct, pSynCretin, bound both receptors, showed higher avidity for GLP-1R than the plain GLP-1 version, and produced sustained weight loss in obese mice from one dose. A bespoke hormone, modelled on a computer and then grown inside a living animal.
None of this is close to a clinic yet. By the authors’ own technology-readiness reckoning, the incretin work sits at roughly level 4: candidate optimisation in animals, a long way short of a product.
The mice were also immune-suppressed for the experiments, their T cells depleted so they wouldn’t reject the partly human proteins, which is standard for this kind of mouse study but not how a human patient would be treated. What carries the optimism is the platform underneath. The same Weiner lab method has already expressed two COVID-19 antibodies continuously in people for more than 72 weeks in a Phase 1 trial, with no anti-drug responses reported, so the underlying delivery system isn’t hypothetical even if this particular use is.
Gary keeps returning to a puzzle the clinic has handed researchers: patients on incretin drugs sometimes find their arthritis or psoriasis easing, which hints these hormones are doing something to the immune system that has little to do with appetite. She talks about how much we still don’t know about what these molecules are doing beyond weight loss and blood glucose. A platform that can produce any engineered protein on demand, for months, from a single jab is a way to start asking.
“Once we have this toolkit, we can think about making novel proteins that didn’t exist before,” Gary says.
Whether the body proves a reliable enough factory to trust with chronic disease is the question the next decade gets to answer.
- Study type: Preclinical, T cell–depleted diet-induced obese mouse models (BALB/c and C57BL/6); peer-reviewed, Trends in Biotechnology (June 2026)
- Intervention: Single intramuscular injection of plasmid DNA (pGLP1-mFc, pGIP-mFc, or AI-designed pSynCretin-mFc) followed by Cellectra 3P electroporation; 50–100 μg dose
- Comparator: Daily semaglutide biosimilar (5 μg) head-to-head; untreated and empty-plasmid controls; single-dose vs. two-dose pLincretin regimen
- Sample size: Small cohorts of DIO mice per arm (typical n≈5–10; exact n varies by figure), group-housed 5 per cage across separate cohorts for expression kinetics, weight/glucose efficacy, and lipidomic profiling
- Duration observed: Sustained hormone expression tracked through day 70; efficacy studies ran 2–6 weeks; single- vs. two-dose comparison extended to day 42
- Funding / conflicts of interest: Funded in part by Inovio Pharmaceuticals (funder reported no role in study design or analysis); senior author D.B. Weiner sits on Inovio’s board of directors and consults for multiple pharma companies including AstraZeneca, Sanofi, and Pfizer
- Main limitation: Mice required T cell depletion to tolerate human-Fc constructs; no in vivo head-to-head comparison against the commercial dual agonist tirzepatide; incretin-specific application sits at Technology Readiness Level 4. No human trial yet
Reference
Gary, E. N., Gao, Y., Hojecki, C. E., Tursi, N. J., Zhang, W., Goldman, A. R., Laenger, N., Tomirotti, M., Khan, M. S., Sahai, R., Oblander, J., Huang, J., Lin, Z. J., Cui, J., Liu, X., Humeau, L., Pallesen, J., Parzych, E. M., & Weiner, D. B. (2026). Engineering single-dose plasmid DNA for sustained in vivo delivery of designer incretins. Trends in Biotechnology. https://doi.org/10.1016/j.tibtech.2026.05.023
Frequently Asked Questions
How can one injection keep working for months when Ozempic needs a weekly shot?
One injection can keep working for months because it delivers DNA instructions rather than the drug itself. The muscle cells read those instructions and manufacture the hormone mimic continuously, so there is nothing to wear off the way an injected peptide does. In mice, useful drug levels were still detectable around 70 days after a single dose.
Is this the same as the gene therapy people worry about changing your DNA?
This is not the kind of gene therapy that rewrites your DNA. The plasmids used here are small circular pieces of DNA that stay separate from the cell’s own chromosomes and are not designed to integrate into the genome, which is one of the safety arguments the researchers make for the approach. The cell simply reads them as instructions to make a protein.
Could this DNA platform be used for diseases other than diabetes and obesity?
This DNA platform could in principle be used well beyond diabetes and obesity, because the same delivery system can encode almost any therapeutic protein. The lab has already used it to make COVID-19 antibodies in people, and the researchers are now interested in incretins’ effects on immune conditions like arthritis and psoriasis. The bottleneck is safety and dosing control, not the range of possible targets.
Why did the researchers build a brand-new molecule with AI instead of just using existing drugs?
The researchers built a new molecule with AI because the DNA approach lets them design proteins that do not exist in nature and have the body produce them directly. Using AlphaFold and synthetic consensus design, they created a single hormone that activates both the GLP-1 and GIP receptors at once, the dual action behind drugs like Mounjaro, demonstrating that the platform can prototype custom therapeutics rather than only copy existing ones.
How close is this to being available for people?
This is not close to being available for people yet. The work was done in mice and sits at an early technology-readiness stage, with key steps like toxicology, large-animal studies, and human dosing trials still ahead. The encouraging part is that the underlying delivery method has already been shown to work safely in a human antibody trial for over 72 weeks.
Cite This Page
