Masked AAV Vectors Switch On Gene Delivery Only Where Disease Signals Strike

Adeno-associated viruses, or AAVs, have become the workhorses of modern gene therapy, ferrying therapeutic genes into cells with remarkable efficiency and a long record of clinical safety. Yet one stubborn problem has shadowed the field for decades: once an AAV vector is injected into the bloodstream, it tends to transduce whatever tissue it encounters, most often the liver, rather than the diseased cells that actually need the therapeutic payload. Researchers have engineered capsids, deployed antibodies and nanobodies, and screened vast libraries of variants in search of tissue specificity, but a truly controllable system that keeps the virus inert until it reaches its destination has remained elusive. Now a team at Peking University, working with collaborators at Zhejiang University, reports in Nature Materials a chemical strategy that effectively puts a molecular lock on AAV particles and hands clinicians the key in the form of disease-associated enzymes, light, or inflammatory chemistry.

The core of the approach, described by Zhiying Zeng, Liying Chang and colleagues under the supervision of Tao Liu, is a masking system built on genetic code expansion. The researchers engineered AAV capsid proteins to carry a non-canonical amino acid bearing a tetrazine group at a precisely chosen position on the viral surface. Tetrazines react rapidly and selectively with trans-cyclooctene, a pairing that has become one of the most reliable tools in bio-orthogonal chemistry. By attaching trans-cyclooctene-modified molecules to this tetrazine handle, the team could decorate the capsid at a single defined site rather than scattering modifications across the particle. This site-specificity matters because AAV infectivity depends on a small set of surface loops that engage cellular receptors; modifying the wrong residues would cripple the virus permanently rather than merely silencing it temporarily.

Two kinds of masking groups were tested. In one configuration, the researchers tethered a truncated fragment of the AAV receptor, the cellular protein that AAV must bind to enter cells, onto the capsid. This decoy fragment occupies the receptor-binding domains from the outside, physically blocking the real receptor on target cells from docking. In the other configuration, the team attached strands of polyethylene glycol, a hydrophilic polymer long used to shield biologics from immune surveillance. Both strategies achieved the same result: masked AAVs injected into mice showed essentially no transduction, their infectivity completely suppressed until the mask was deliberately removed. The PEGylated version carried an added benefit, since the polymer cloak also reduced recognition of the capsid by anti-AAV antibodies, a significant concern given that many patients carry pre-existing neutralizing antibodies against common AAV serotypes.

The unlock step relies on cleavable linkers, and the study demonstrates three distinct activation chemistries, each tailored to a different clinical scenario. The first is a protease-activated mask designed around matriptase-2, a serine protease whose expression is largely restricted to the liver. The masking group was connected to the capsid through a peptide linker carrying an matriptase-2 cleavage sequence. When the masked vector circulated through hepatic sinusoids, the liver enzyme clipped the linker, shed the mask, and restored the capsid’s ability to bind its receptor. In mice, systemic administration of these protease-activated particles produced gene expression concentrated in the liver, while a non-cleavable control linker kept the virus silent throughout the body. The result effectively converts liver tropism from a passive side effect of capsid biology into a programmable, switch-like property.

The second activation mode uses near-infrared light, offering something no endogenous trigger can: an external dial that an operator controls in space and time. Here the masking group was attached through a linker cleavable by near-infrared irradiation. After intravenous injection of the masked AAV9 vectors, the researchers shone near-infrared light on selected regions and observed transduction only where the beam had passed. In mice, this produced localized gene expression in the gastrocnemius muscle and, notably, in the brain, where light delivered through the skull activated vectors that had circulated inertly through the bloodstream. Because near-infrared light penetrates tissue more deeply than visible wavelengths, this scheme suggests a path toward optically demarcated gene delivery, in which a clinician traces the treatment zone with a light source and the vector follows that contour, leaving unilluminated tissue untouched.

The third chemistry exploits the inflammatory microenvironment itself. Diseased and injured tissues, from fibrotic liver to reperfused heart muscle, are rich in reactive oxygen species such as hydrogen peroxide, which healthy tissue maintains at far lower concentrations. The team designed a boronate-based linker that is stable in blood but cleaved by reactive oxygen species, so the mask falls off preferentially where oxidative stress marks the pathology. In a model of liver fibrosis, ROS-activated vectors showed controlled transduction consistent with the oxidative environment of the diseased organ. The strategy effectively lets the disease decrypt the vector: only tissue emitting the chemical signature of injury gains access to the therapeutic gene.

The most therapeutically compelling demonstration came in the heart. In a mouse model of myocardial ischaemia-reperfusion injury, the oxidative burst that accompanies restored blood flow to damaged tissue provided the activation signal. Intravenously administered ROS-activated masked AAV9 vectors unmasked preferentially in the injured myocardium and delivered a gene encoding VEGF-A165, a vascular endothelial growth factor isoform that promotes angiogenesis. Localized expression of the growth factor in the damaged heart improved cardiac function relative to controls, pointing toward a regenerative application in which a healing factor is produced exactly where the heart is remodeling, rather than systemically where it could promote unwanted vascular growth elsewhere.

Behind these demonstrations lies a substantial body of characterization work. The team quantified the pharmacokinetics of masked versus unmasked vectors, showing that PEGylation altered circulation behavior in measurable ways, and mapped biodistribution of vector genomes across organs to confirm that masking suppressed off-target accumulation of expression. Immunogenicity experiments showed that PEG-shielded capsids elicited lower anti-AAV antibody responses and resisted neutralization by human intravenous immunoglobulin better than unshielded particles, an encouraging signal for patients with pre-existing immunity. Structural data from the Protein Data Bank guided the choice of the capsid residue used for tetrazine incorporation, and the plasmids underlying the platform, including constructs for AAV-DJ, AAV9 and AAV8 capsids, have been deposited with Addgene, making the system available to other laboratories.

The conceptual lineage of this work runs through the prodrug field, in which pharmacologically inactive precursors are converted into active drugs by enzymes or chemistry specific to diseased tissue. Antibody probodies carrying tumor-protease-cleavable masks have already entered clinical trials, and earlier studies had produced protease-activatable AAVs and light-controlled viral transduction through other means. What distinguishes the new platform is the combination of genetically encoded, site-specific chemical handles with a modular family of cleavable linkers, allowing the same masked capsid chassis to be retargeted to liver, to an optical field, or to an oxidative injury zone simply by swapping the linker chemistry. That modularity, the authors suggest, makes the platform a general-purpose architecture rather than a single-purpose vector.

Cautions remain before such systems reach patients. The studies were conducted in mice, and the sensitivity, kinetics and completeness of unmasking in larger animals with different protease expression patterns, tissue optics and immune backgrounds will need to be established. Near-infrared activation requires a light source capable of reaching the target tissue, which constrains the approach to accessible sites or to applications where fiber-optic or transdermal illumination is feasible. ROS-triggered unmasking depends on the disease producing enough oxidative chemistry to cleave the linker within a useful window, and the specificity of that trigger across diverse pathologies will require careful validation. Even so, the demonstration that a single chemical masking principle can gate AAV infectivity behind three orthogonal keys, and that one of those keys can direct reparative gene expression to an injured heart, marks a meaningful advance in the quest to make gene therapy as much a matter of placement as of payload.

Subject of Research: Chemically masked adeno-associated viral vectors activated by disease-related signals for controlled gene delivery

Article Title: Controllable gene delivery via masked adeno-associated viral vectors

Article References: Zeng, Z., Chang, L., Wang, J., Liu, Y., Wang, X., Sun, Y., Zuo, Y., Hou, X., Wang, Y., Gu, Z., & Liu, T. (2026). Controllable gene delivery via masked adeno-associated viral vectors. Nature Materials. https://doi.org/10.1038/s41563-026-02724-1

Image Credits: AI Generated

DOI: 10.1038/s41563-026-02724-1

Keywords: adeno-associated virus, gene therapy, gene delivery, genetic code expansion, non-canonical amino acids, bio-orthogonal chemistry, matriptase-2, near-infrared light, reactive oxygen species, myocardial ischaemia-reperfusion, VEGF-A165, PEGylation