Harvard and MGH’s Engineered AAV Capsids Targeted Gene Therapy Delivery with Lower Doses and Fewer Immune Reactions

Adeno-associated viral (AAV) vectors are the leading platform for delivering gene therapies directly into human cells. However, the doses required to achieve sufficient gene expression in target tissues can trigger dangerous immune responses. Researchers at Harvard University and Massachusetts General Hospital have developed a method to discover and engineer novel AAV capsids that can transduce target cells far more efficiently, enabling therapeutic effects at lower and safer doses.

The Problem

AAV9, the most widely used serotype for central nervous system gene therapy, has demonstrated remarkable clinical success, most notably in the treatment of spinal muscular atrophy in infants. However, the systemic delivery of AAV vectors at the doses required for adequate transgene expression carries significant safety risks. High systemic doses have been associated with T-cell responses that eliminate transduced cells, severe inflammatory reactions in primate models, and patient deaths. A recent clinical trial for Duchenne muscular dystrophy was placed on hold by the US Food and Drug Administration after a patient experienced a serious immune reaction.

The fundamental problem is efficiency. Standard AAV vectors transduce cells relatively inefficiently on a per-genome-copy basis. Achieving therapeutic levels of transgene expression therefore requires large total doses, and large doses trigger immune toxicity. Developing capsid variants that transduce target cells more efficiently – delivering the same therapeutic effect at lower total vector doses – would simultaneously improve efficacy and reduce the risk of immune complications.

The challenge in developing new capsids is the vast sequence space of possible capsid protein variants. A systematic method for identifying which variants achieve efficient transduction in specific target cell types is needed.

What This Invention Does

The patent describes a selection-based approach to identifying novel AAV capsids with enhanced cell-type-specific transduction efficiency. The method uses a two-part expression cassette inserted into the viral genome. The first part is a Cre recombinase cassette driven by a promoter active in the cell type of interest. The second part is an engineered capsid gene cloned in-cis, containing a random peptide insert – a heptamer of seven amino acids – displayed on the capsid surface. A library of AAV particles, each displaying a different heptamer sequence, is used to transduce target cells. Successful transduction in the target cell type activates the Cre cassette, which recombines a loxP-flanked stop sequence upstream of a reporter gene, marking successfully transduced cells and allowing the responsible capsid sequences to be recovered and identified.

This selection strategy creates a feedback loop: capsid sequences recovered from efficiently transduced cells in one round can be used as the basis for a second round of selection, progressively enriching for variants with superior properties. The two capsid peptide sequences STTLYSP (SEQ ID NO:1) and FVVGQSY (SEQ ID NO:2) are among the engineered variants identified through this process, shown to mediate enhanced transgene expression in specified cell types.

The resulting engineered AAVs, which do not contain wild-type VP1, VP2, or VP3 capsid proteins, can carry therapeutic transgenes and are designed for delivery to the central nervous system, peripheral nervous system, inner ear, heart, retina, and other target tissues.

Key Features

Directed capsid evolution with Cre-dependent selection. A two-part library construct enables in vivo selection of capsid variants based on actual transduction efficiency in the target cell type, rather than binding affinity alone, producing capsids with genuine functional superiority.

Random heptamer surface display. A random seven-amino-acid peptide insert is displayed on the capsid surface, with STTLYSP and FVVGQSY identified as sequences that improve transduction in specific neuronal, cardiac, and inner ear cell types.

Iterative enrichment. Recovered capsid sequences from a successful selection round can be fed into subsequent rounds of selection, enabling progressive optimisation of transduction efficiency for specific tissue targets.

Broad target tissue applicability. The engineered capsids are described for use in neurons (including spiral ganglion neurons and dorsal root ganglion neurons), astrocytes, cardiomyocytes, inner and outer hair cells of the cochlea, retinal cells, and photoreceptors.

Multiple delivery routes. The application describes parenteral, intracerebral, intrathecal (including lumbar injection and cisternal magna injection), and cochlear delivery routes, with cochlear access through the round window membrane, cochleostomy, or oval window fenestra.

Disease applications. The vectors are described for potential use in Alzheimer’s disease, Parkinson’s disease, spinal muscular atrophy, Huntington’s disease, Duchenne muscular dystrophy, Usher syndrome, Leber congenital amaurosis, haemophilia, and multiple forms of inherited deafness.

Who Is Behind It?

The applicants are President and Fellows of Harvard College and The General Hospital Corporation (the research arm of Massachusetts General Hospital). The named inventors are Casey A. Maguire, Eloise Marie Hudry, and Killian S. Hanlon. The work was supported by US National Institutes of Health grants AG047336 and DC017117. This divisional application was filed on 26 February 2026, derived from parent application AU 2020248116, and traces to PCT/US2020/025720 with a US provisional priority date of 28 March 2019. The application is managed by RnB IP Pty Ltd in Deakin, Australian Capital Territory.

Why It Matters

The immunotoxicity problem with high-dose AAV gene therapy has slowed clinical progress and caused serious patient harm in recent trials. A solution that delivers equivalent therapeutic gene expression at lower total vector doses by using more efficient capsids would represent a fundamental improvement in the safety profile of the entire AAV gene therapy platform. Given that AAV therapies are being developed for dozens of genetic diseases affecting the nervous system, sensory organs, heart, and muscles, the impact of a broadly applicable capsid engineering approach is significant.

The Harvard/MGH approach is particularly valuable because it selects for transduction efficiency in the actual target cell type, in vivo, under physiologically realistic conditions – rather than optimising for surrogate measures such as receptor binding that may not predict in vivo performance. This functional selection principle could be applied to identify optimised capsids for virtually any target cell type where a cell-type-specific promoter is available.

---

AU 2026201434 was published in the Australian Official Journal of Patents on 19 March 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.

Related Concepts

Adeno-associated viruses (AAVs) are small, non-integrating DNA viruses widely used as gene therapy vectors because of their low pathogenicity and broad tissue tropism. Different naturally occurring serotypes, such as AAV9, differ in which cell types they preferentially transduce.

Engineering novel capsid proteins with enhanced cell-type specificity and transduction efficiency is a central challenge in gene therapy development. Higher-efficiency capsids allow therapeutic transgene expression at lower total vector doses, directly reducing the immunotoxicity risk that has caused serious adverse events – including patient deaths – in trials for Duchenne muscular dystrophy and other diseases.