A 'Stealth' DNA Circle Lets Researchers Insert Whole Genes Without Triggering Immune Havoc

Mass General Brigham

Gene therapy has a delivery problem. Not the kind that better couriers can fix, but a fundamental biological conflict: the very molecules needed to carry corrective genes into cells look, to the immune system, like an invasion. Double-stranded DNA (dsDNA), the standard cargo format for inserting large gene sequences, triggers potent innate immune responses that can cause severe toxicity, particularly when delivered directly into the body. This immune reaction limits dosing, reduces efficacy, and has been a persistent obstacle to scaling gene therapies beyond single-mutation corrections.

A study published in Nature by researchers at Mass General Brigham and collaborators describes a workaround that borrows a trick from bacteria: wrap the DNA cargo in a form the immune system largely ignores.

Why inserting whole genes matters

Most genome editing therapies target individual mutations. That works when a disease is caused by a single, well-characterized genetic error. But many genetic disorders arise from dozens or even thousands of unique mutations scattered across the same gene. Cystic fibrosis, for instance, involves more than 2,000 known variants in the CFTR gene. Developing a custom therapy for each mutation is impractical at scale.

A more universal approach would insert a complete, functional copy of the gene at a specific location in the genome, providing a single treatment applicable to all patients regardless of their particular mutation. The enzymes that can perform this insertion, called recombinases, require double-stranded DNA to function. But dsDNA is precisely what triggers the immune alarm.

The single-strand solution and its catch

The research team, led by senior author Benjamin Kleinstiver of the Center for Genomic Medicine at Mass General Brigham, recognized that circles made of single-stranded DNA (ssDNA) could largely evade innate immune detection. The immune system's DNA-sensing pathways are tuned primarily to detect dsDNA, the signature of most invading pathogens. A single-stranded circle slips past these sensors.

But here is the catch: recombinase enzymes evolved to recognize and cut dsDNA. They cannot work with single-stranded molecules. The team needed a cargo that was invisible to the immune system but visible to the insertion machinery.

The solution came from studying how bacteria and bacteriophages handle a similar problem in nature. These organisms have evolved mechanisms to insert single-stranded DNA into double-stranded host genomes using recombinases. The researchers adapted this principle for human cells.

The INSTALL design

The method, called INSTALL, uses a DNA circle that is mostly single-stranded but contains a short double-stranded region. The dsDNA segment is long enough for the recombinase to bind and function, but short enough to avoid detection by the cell's immune sensors. The bulk of the circle, carrying the actual gene cargo, remains in stealth-mode single-stranded form.

Lead author Connor Tou, a postdoctoral fellow in the Kleinstiver lab, demonstrated that INSTALL could integrate DNA in multiple human cell types without triggering the toxic immune responses seen with conventional dsDNA delivery. The real test came in mice.

Mouse livers, lipid nanoparticles, and a stark contrast

The team delivered INSTALL cargo to mouse livers using lipid nanoparticles (LNPs), the same delivery technology used in mRNA COVID-19 vaccines. When conventional dsDNA molecules were delivered via LNPs, the mice suffered fatal immune reactions. When INSTALL circles were used instead, the mice showed successful gene insertion in their liver cells and looked similar to untreated controls, with no signs of immune toxicity.

The contrast was dramatic enough to be visible without sophisticated analysis. The INSTALL-treated animals behaved normally; the dsDNA-treated animals did not survive.

This result matters because it demonstrates that large-scale genome writing, inserting gene-sized sequences rather than single-base corrections, can be done without viral vectors and without triggering the immune reactions that have limited in vivo gene therapy approaches. Viral vectors, while effective at shuttling DNA into cells, carry their own limitations in terms of manufacturing cost, payload size, and immunogenicity with repeated dosing.

What INSTALL does not yet prove

The mouse liver results are promising but preliminary. Liver cells are among the easiest targets for LNP-based delivery because LNPs naturally accumulate in the liver after intravenous injection. Whether INSTALL can achieve similar efficiency and safety in other organs, including the lungs, brain, or muscle tissue where many genetic diseases manifest, has not been tested.

The study also does not report long-term follow-up data. Gene insertion efficiency, durability of expression, and potential off-target integration events over months or years remain open questions. Recombinase-mediated integration is designed to be site-specific, but confirming this specificity across the full genome at scale requires extensive sequencing that this initial study does not include.

The efficiency of insertion, while sufficient for proof of concept, will likely need improvement before clinical application. The authors acknowledge that continued optimization of both the DNA cargo design and the recombinase enzymes themselves will be necessary to advance toward human trials.

There is also the manufacturing question. Single-stranded DNA circles are produced by Full Circles Therapeutics, a company whose co-authors contributed to this study. Scaling production of these specialized molecules to clinical-grade quantities and maintaining quality control at that scale is a challenge the field has not yet solved.

Still, INSTALL addresses what has been one of the most stubborn barriers in gene therapy: how to get large DNA cargo into cells without the immune system destroying the patient in the process. By splitting the cargo's identity, making it look like ssDNA to the immune system but like dsDNA to the insertion machinery, the method creates a path that was not previously available.