Durable gene delivery is becoming a designable, portable platform

What the work claims

This is a grant, an NIGMS Maximizing Investigators' Research Award (R35) to Liujiang Song at Ohio State University, funded at $386,105 for fiscal year 2026, running from 2026-08-01 to 2031-03-31.¹ Its central observation is that after AAV delivery, most vector DNA is lost from cells within weeks, but a small fraction persists long-term as episomes, separate DNA molecules that survive without integrating into the host genome. The project calls this "Selective Persistence" and states plainly that its mechanism is poorly understood more than sixty years into the study of AAV.

The preliminary evidence is real but murine. The abstract reports that the lab identified a novel inverted terminal repeat (ITR) design that attenuates toxicity, found genetic "signatures" associated with long-term episomal persistence in mice, and showed proof-of-concept that grafting those signatures onto synthetic substrates prolongs gene expression in mice. Whether the signatures exist in humans is explicitly unknown.¹ The project asks three questions: what happens to AAV genomes across human tissues and what molecular features the persisting episomes have; whether ITR engineering can promote stable episome formation and cut the required vector dose; and how far AAV persistence mechanisms can be grafted onto non-viral vectors to make synthetic delivery durable.

To get at the human question the lab proposes something unusual: a "human decedent H2H platform" that examines AAV genome fate directly across tissues from deceased donors, paired with long-read sequencing, spatial biology, mass spectrometry, and organ-on-chip systems.¹

How it works

AAV is the workhorse delivery vehicle of modern gene therapy and of a large fraction of laboratory cell engineering: a small, non-enveloped virus whose single-stranded DNA cargo is delivered into the cell nucleus. Its inverted terminal repeats, short palindromic sequences at each end of the genome, are the control elements: they are required for genome packaging, for conversion of the delivered single strand into a double-stranded form, and, as this project and prior work argue, for persistence and safety. What the field has lacked is an explanation for why persistence is selective, meaning why a minority of genomes survives while the rest disappears, and what molecular marks distinguish the survivors.

The project logic is engineering logic applied to a biology problem. Step one, characterize the survivors: find the episomes in human tissues and read their molecular signatures. Step two, design the input: if the signatures can be attributed in part to ITR sequence, then ITRs become an engineering variable, and better ITRs could mean stable expression at lower vector dose, which is both a safety improvement (less viral load) and a cost improvement. Step three, decouple the mechanism from the vehicle: if persistence is a property of sequence features rather than of the viral particle itself, the same features can be built into plasmid or other synthetic delivery systems, which are cheaper to manufacture and carry none of AAV's packaging constraints.¹

Where a skeptic should push

The load-bearing assumption is cross-species conservation. Every persistence signature in hand comes from mice. The entire human relevance of the program rests on the decedent platform showing that the same signatures mark persisting episomes in human tissue, and the abstract itself flags this as an open question, not a finding. Mouse liver and human liver are different enough in AAV tropism and immune context that conservation is a hypothesis worth funding, not a fact worth assuming.

Second, the H2H platform has a selection problem hiding inside it. Decedents available for tissue donation are typically older and were typically treated, often with high-dose AAV under medical indications. Episome fate measured in such tissue reflects the biology of sick patients receiving therapeutic doses, which is the population gene therapy actually serves, but it is not a clean window into persistence in healthy tissue, and it will not reveal what low-dose, engineered-ITR vectors do in people. Dose reduction is a goal stated in the abstract; the platform as described cannot demonstrate it in living patients.

Third, the dose-reduction and synthetic-grafting claims are aims, not results. The proof-of-concept for grafting persistence signatures onto synthetic substrates is stated to prolong expression in mice. Nothing in the record demonstrates durable synthetic delivery in humans, and the step from episomal persistence in a tissue to stable, predictable expression in a manufactured cell product is a long one.

Substrate portability outruns vector regulation

For platform access and vendor capability, the third research question is the one to watch. Today, durable gene delivery is concentrated in the hands of a small number of viral-vector manufacturers because durability has lived inside the virus: capsid engineering, packaging cell lines, and pharmacovigilance infrastructure are all built around AAV. If persistence becomes a property of portable sequence signatures that can be grafted onto synthetic substrates, the capability migrates. Any competent synthetic-biology shop could in principle build durable modification into a plasmid system, and the manufacturing moat of the vector oligopoly deflates. That is the opportunity: cheaper, more distributed durable cell engineering.

The threat is that regulation is keyed to the vehicle, not the cargo's behaviour. Oversight categories for gene delivery, from manufacturing standards to long-term follow-up obligations, attach to viral vectors as a class. A synthetic episome that achieves the same biological end state through a different physical substrate does not obviously inherit those categories. If the mechanism ports while the rules do not, durable human cell engineering escapes the regime built to watch it, not through anyone's intent but through a classification lag. Anyone planning engineered organoid lines, including neural organoid lines for research or for computing, should read the third question as a substrate-independence announcement: the durability layer is being deliberately decoupled from the delivery layer.

There is also a quieter governance fact underneath the science. The project's human evidence base comes from decedents, and under the Common Rule a "human subject" is defined as a living individual, which places research confined to decedent-derived material outside the rule's human-subjects protections.² That makes the H2H platform scientifically powerful, because it yields human data no trial could ethically produce. It also means the reference data set that will define what "human-relevant" persistence looks like, and which will likely sit inside commercial ITR design, is assembled outside the consent architecture that governs the living. The abstract makes no mention of consent or governance for the decedent platform.

For computing on living neural tissue specifically, the link is conditional but direct in structure. Engineered neural organoid lines would want exactly what this project is trying to build: durable, non-integrating genetic modification, because integration carries mutagenesis risk while transient transfection does not survive passaging. An episomal line avoids the integration hazard and creates a new one instead: a heritable-in-culture state that is invisible to phenotype-based quality control. A cell vendor can certify identity, potency, and sterility of an organoid line and say nothing about whether it carries persistent vector-derived episomes, because nobody standardly assays for them and the persistence state is not written on the cells' appearance.

The bottom line

Established: selective episomal persistence after AAV delivery is real, ITRs are implicated in persistence and safety, and persistence signatures can be grafted onto synthetic substrates to extend expression, at least in mice. Hypothesis: the signatures are conserved in humans, ITR engineering will lower required doses, and synthetic substrates can be made durably expressive. What would confirm it: the decedent platform finding the murine persistence signatures in human episomes, plus engineered-ITR vectors maintaining expression at meaningfully lower dose in human-relevant models. What would break it: human episomes showing a different persistence architecture, which would strand the synthetic-grafting strategy on a murine peculiarity.