AAV9, a small virus repurposed as a delivery vehicle for therapeutic genes, has become one of the most important tools in gene therapy, powering the first FDA-approved treatment for spinal muscular atrophy and driving dozens of clinical trials for diseases of the brain, heart, and muscle. Its appeal comes from a rare combination of traits: it can cross the blood-brain barrier, it efficiently enters muscle and cardiac tissue, and it does not cause disease in humans. But the story of AAV9 is far from a simple success narrative. High-dose toxicity, immune barriers that lock out many patients, and a price tag that strains health systems are all shaping where this technology goes next.
How AAV9 Finds Its Way Into Cells
Adeno-associated virus serotype 9 is a tiny, non-enveloped virus whose outer shell, the capsid, determines which cells it can enter. AAV9 uses the sugar galactose on cell surfaces as its primary receptor. Researchers have mapped the specific amino acids on the AAV9 capsid that form a pocket for grabbing galactose, identifying five residues clustered around the capsid’s three-fold axis of symmetry.1PubMed Central. Identification of the galactose binding domain of the adeno-associated virus serotype 9 capsid Because galactose is found on many different cell types, AAV9 has broad tropism, meaning it can enter a wide variety of tissues. But its standout feature is what happens at the blood-brain barrier.
The blood-brain barrier is a tightly sealed layer of endothelial cells lining the brain’s blood vessels, designed to keep most molecules out. AAV9 penetrates this barrier more effectively than older serotypes like AAV2, using an active, cell-mediated process called transcytosis rather than simply slipping between cells. In laboratory models, AAV9 crossed endothelial barriers without disrupting their integrity, leaving tight junction proteins and electrical resistance intact.2PubMed Central. Trafficking of adeno-associated virus vectors across a model of the blood-brain barrier; a comparative study of transcytosis and transduction using primary human brain endothelial cells This ability to reach the central nervous system after a simple intravenous injection is what made AAV9 the vector of choice for treating neurological diseases.
Spinal Muscular Atrophy and the Landmark Clinical Results
The clearest proof that AAV9 gene therapy works in humans came from spinal muscular atrophy type 1, a devastating genetic condition in which infants lack a functional SMN1 gene and progressively lose motor neurons. Without treatment, most children with the severest form cannot sit independently and many do not survive past age two. Onasemnogene abeparvovec (sold as Zolgensma) delivers a working copy of the SMN1 gene using an AAV9 capsid, given as a single intravenous infusion.
The phase 3 STR1VE-EU trial treated 22 infants and found that about 59% achieved functional independent sitting by 18 months of age, compared with none in a matched untreated group. Roughly 91% survived free from permanent ventilation at 14 months, versus about a quarter of untreated children.3The Lancet Neurology. Onasemnogene abeparvovec for presymptomatic infantile-onset spinal muscular atrophy (STR1VE-EU): an open-label, single-arm, multicentre, phase 3 trial Long-term follow-up data from the original phase 1 START trial, extending beyond five years after dosing, showed that all ten patients in the therapeutic-dose group were alive and free from permanent ventilation, with motor milestones maintained and no regression. Two patients even achieved a new milestone, standing with assistance, years after the single infusion.4JAMA Neurology. Five-Year Extension Results of the Phase 1 START Trial of Onasemnogene Abeparvovec in Spinal Muscular Atrophy
The durability of that single dose is remarkable. Because the AAV9 genome persists in non-dividing cells as an episome rather than integrating into the host chromosome, neurons that receive the gene can continue to express the therapeutic protein for years. The five-year data suggest the effect holds. Hepatotoxicity, meaning liver inflammation reflected in elevated liver enzymes, remains the primary drug-related safety concern, and patients receive corticosteroids to manage it.5PubMed Central. Gene Therapy for Spinal Muscular Atrophy (SMA): A Review of Current Challenges and Safety Considerations for Onasemnogene Abeparvovec (Zolgensma)
Muscular Dystrophies and Expanding Disease Targets
AAV9’s affinity for muscle and heart tissue has made it a natural candidate for muscular dystrophies well beyond SMA. In Duchenne muscular dystrophy, where the dystrophin gene is too large to fit inside an AAV capsid, researchers have engineered smaller “mini-dystrophin” genes that retain enough function to protect muscle fibers. Preclinical studies using AAV9 to deliver mini-dystrophin in rat and mouse models of Duchenne showed dose-dependent increases in mini-dystrophin-positive muscle fibers across skeletal muscle, the diaphragm, and the heart.6Molecular Therapy Methods & Clinical Development. Nonclinical evaluation of fordadistrogene movaparvovec, a recombinant AAV9 vector carrying a human mini-dystrophin transgene, in the DMD-mdx rat model Several of these programs have advanced into human trials.
A rarer form, limb-girdle muscular dystrophy type R25, has also seen promising preclinical results. AAV9 carrying the BVES gene under a muscle-specific promoter improved body weight, muscle strength, exercise performance, and heart rhythm abnormalities in knockout mice, with benefit seen even when treatment started after disease onset in adult animals.7PubMed Central. Systemic AAV9.BVES delivery ameliorates muscular dystrophy in a mouse model of LGMDR25 These results, though still in animals, illustrate the breadth of diseases AAV9 could address wherever efficient muscle transduction is needed.
The Safety Picture at High Doses
AAV9 was long considered a mild vector with a favorable safety profile, and at low doses it generally is. But as clinical programs pushed into high systemic doses, particularly above 100 trillion viral genomes per kilogram of body weight, a more sobering picture emerged. Toxicities at these doses can include severe immune responses, liver damage, and a dangerous condition called thrombotic microangiopathy, in which small blood clots form in capillaries and damage organs.8PubMed. Systemic Toxicity of Recombinant Adeno-Associated Virus Gene Therapy Vectors
Research in nonhuman primates has shown that high-dose AAV9 triggers acute liver and endothelial injury characterized by rising liver enzymes, a drop in platelet counts, and activation of the complement system, all peaking around three days after administration. Platelet microthrombi were found in the liver’s sinusoidal blood vessels, alongside endothelial injury. This toxicity appeared at doses of 100 trillion viral genomes per kilogram or above, regardless of the transgene being carried, suggesting the capsid load itself is the driver.9Molecular Therapy. Acute liver and endothelial injury following high-dose systemic AAV administration in nonhuman primates Thrombotic microangiopathy in particular has been linked to anti-capsid antibodies that trigger complement activation, turning the immune system against the vector itself.10JCI Insight. Thrombotic microangiopathy following systemic AAV administration is dependent on anti-capsid antibodies
A separate concern involves dorsal root ganglia, clusters of sensory nerve cell bodies along the spine. In studies involving 260 cynomolgus macaques given AAV9 vectors by various routes, about 78% showed microscopic injury to dorsal root ganglia neurons within the first 2 to 12 weeks, though the damage was asymptomatic and at most moderate in severity.11Molecular Therapy – Methods & Clinical Development. Neurofilament light chain and dorsal root ganglia injury after adeno-associated virus 9 gene therapy in nonhuman primates Encouragingly, when animals were followed for a full year, the incidence dropped to about 42% and severity decreased to minimal, suggesting these changes can resolve rather than worsen over time.12Molecular Therapy Methods & Clinical Development. Soluble biomarker assessment of dorsal root ganglion injury and resolution in nonhuman primates following AAV9 gene therapy Still, the dorsal root ganglia finding has prompted monitoring requirements in clinical trials and a search for vectors that avoid sensory neurons.
Pre-Existing Antibodies and the Re-Dosing Problem
One of the biggest practical limitations of AAV9 gene therapy is that many people already carry antibodies against it from natural exposure to wild-type adeno-associated viruses earlier in life. These neutralizing antibodies can intercept the vector before it reaches target cells, rendering the treatment useless. A large study of the Chinese population found that roughly 59% of people aged 0 to 90 tested positive for anti-AAV9 neutralizing antibodies, with adults showing substantially higher rates than children (about 75% versus 34%). The lowest prevalence was in children between six months and three years old, at under 8%, which aligns with the window in which SMA patients are typically treated.13PubMed Central. Pre-existing Anti-AAV9 antibodies in the Chinese healthy and rare disease populations: Implications for gene therapy
Even for diseases where patients lack antibodies at the time of first treatment, re-dosing creates a separate hurdle. A single AAV9 infusion triggers a robust immune response, generating high-titer antibodies that persist. This means a second dose would likely be neutralized. Researchers are exploring several strategies to get around this. One approach involves physically removing antibodies from a patient’s blood before dosing. In a rat model, hemapheresis combined with immunoadsorption using AAV9-coated beads restored transduction in the liver and dramatically increased it in the heart.14Molecular Therapy: Methods & Clinical Development. Successful Transduction with AAV Vectors after Selective Depletion of Anti-AAV Antibodies by Immunoadsorption
Immunosuppressive drug regimens are another option. Combinations of drugs that dampen T-cell and B-cell activity, such as ciclosporin A with rituximab, have reduced neutralizing antibodies enough to permit vector re-administration in primate models. Rapamycin, especially when packaged in nanoparticles, has also shown promise at blunting the immune response to the AAV capsid.15Trends in Biotechnology. Evading and overcoming AAV neutralization in gene therapy None of these strategies has become routine in clinical practice yet, but they represent the most active frontier for making AAV therapies repeatable.
Engineering Capsids for Better Targeting
AAV9’s broad tropism is both its greatest strength and a liability. When you inject a high dose intravenously to reach the brain, much of the vector ends up in the liver instead, wasting therapeutic payload and contributing to hepatotoxicity. This has driven a wave of capsid engineering aimed at creating AAV9 variants that go more precisely where they are needed.
One strategy uses directed evolution, essentially mutating the capsid gene, packaging the variants into libraries, and selecting the ones that best transduce the target tissue in living animals. This approach yielded a family of capsid variants containing an RGD peptide motif that transduces muscle with superior efficiency and selectivity after intravenous injection, performing well in both mice and nonhuman primates. The lead variant, named MyoAAV 1A, emerged when all top-performing capsids from stringent muscle-selection rounds converged on the same RGD motif.16PubMed Central. Directed evolution of a family of AAV capsid variants enabling potent muscle-directed gene delivery across species For brain targeting, a similar mRNA-based evolution strategy in mice and macaques has identified engineered vectors with increased brain potency and decreased liver tropism.17PubMed Central. Systemic administration of novel engineered AAV capsids facilitates enhanced transgene expression in the macaque CNS
Another line of work has generated galactose-binding-deficient AAV9 capsids as starting platforms for peptide display libraries, producing new variants with enhanced brain targeting in both mice and marmosets.18Cell Press. AAV9: Current Progress and Breakthroughs in Gene Therapy The practical payoff of these engineered capsids is the possibility of achieving the same or greater therapeutic effect at much lower doses, which would reduce both toxicity and manufacturing cost.
Smarter Vector Design Beyond the Capsid
Capsid engineering addresses where the vector goes. But what happens once it arrives in a cell depends on the DNA payload inside. Two innovations have significantly improved AAV9 vector performance.
The first is the self-complementary AAV genome. Standard AAV genomes are single-stranded, meaning the host cell must synthesize a second strand of DNA before the gene can be expressed, a slow step that limits how quickly and how completely transduction occurs. Self-complementary AAV vectors fold back on themselves to form double-stranded DNA immediately upon entering the nucleus, bypassing that bottleneck. The trade-off is a halved packaging capacity, roughly 2.4 kilobases instead of the usual 4.7, but for small transgenes the speed advantage is substantial.19PubMed Central. Self-Complementary AAV (scAAV): A Promising Tool in Gene Therapy Self-complementary AAV9 has been shown to outperform its single-stranded counterpart in transducing adult motor neurons, likely by achieving greater saturation of cells within the tissue it reaches.20Molecular Therapy. Intravenous Administration of Self-complementary AAV9 Enables Transgene Delivery to Adult Motor Neurons
The second innovation is microRNA-regulated expression. Even with a tissue-specific promoter, AAV9 vectors administered intravenously tend to produce some transgene expression in off-target organs, especially the liver. By engineering binding sites for tissue-specific microRNAs into the vector’s untranslated region, researchers can silence expression in unwanted tissues while leaving the target tissue unaffected. For CNS-targeted therapies, incorporating microRNA target sites that are highly expressed in the liver and other peripheral organs allowed stable brain expression with virtually no detectable peripheral expression.21Molecular Therapy. MicroRNA-regulated, Systemically Delivered rAAV9: A Step Closer to CNS-restricted Transgene Expression A similar approach using liver-specific microRNA-122 target sites completely eliminated liver expression of AAV9-delivered reporter genes in mice while preserving cardiac expression at normal levels, outperforming even a cardiac-specific promoter at liver de-targeting.22PubMed Central. microRNA122-regulated transgene expression increases specificity of cardiac gene transfer upon intravenous delivery of AAV9 vectors
Choosing the Right Delivery Route
How and where you inject an AAV9 vector changes the outcome dramatically. Intravenous delivery is the simplest and most broadly distributed, but it requires the highest doses because most of the vector gets absorbed by the liver before it can reach the brain or spinal cord. Direct injections into the cerebrospinal fluid, whether through the lumbar spine or directly into the brain’s ventricles or cisterna magna, achieve higher transgene expression in the central nervous system at lower total doses.
A pharmacokinetic study in rats comparing four routes for an AAV9 vector found that brain expression ranked from highest to lowest as follows: direct injection into the brain tissue (intrastriatal), injection into the cisterna magna, injection into the brain ventricles, and finally intravenous delivery. The cisterna magna route favored the hindbrain and spinal cord, while the ventricular route produced more even distribution across brain regions.23PubMed. Pharmacokinetics of AAV9 Mediated Trastuzumab Expression in Rat Brain Following Systemic and Local Administration A separate mouse study confirmed that direct cerebrospinal fluid injections achieved higher CNS expression than intravenous delivery, though an important caveat emerged: cerebrospinal fluid routes did not eliminate peripheral tissue expression the way many researchers had hoped.24PubMed Central. A Head-to-Head Comparison of AAV9 Biodistribution in Mice: Routes of Administration and Age Dependence Combining lumbar and ventricular injections in that study did not meaningfully boost expression beyond what either route achieved alone, suggesting there is a ceiling for cerebrospinal fluid delivery.
Manufacturing Bottlenecks
Making enough AAV9 vector to treat even a small patient population is a formidable challenge. The doses required for systemic therapy are enormous, often on the order of trillions of viral genomes per kilogram of body weight, and producing that much virus at clinical-grade purity pushes current manufacturing platforms to their limits.
One persistent headache is empty capsids, viral shells that assemble correctly but fail to package the therapeutic DNA. These empties can make up a large fraction of the total and are immunogenic without being therapeutic. Traditional separation by ultracentrifugation works but is slow and hard to scale. More scalable chromatography-based methods, such as anion exchange chromatography optimized with histidine-containing buffers and sodium acetate elution, have achieved comparable empty capsid reduction (down to about 20% of total particles) while increasing yield by about 14% over ultracentrifugation.25PubMed. Removal of empty capsids from high-dose adeno-associated virus 9 gene therapies Another purification strategy reported over 96% full capsids in the final product using a scalable, ultracentrifugation-free protocol.26Molecular Therapy Methods & Clinical Development. Scalable and Ultracentrifugation-Free Purification of Recombinant Adeno-Associated Virus Serotype 9
Beyond empty capsids, residual host cell proteins from the production cell line present a subtler quality concern. Proteomic analysis of AAV products has identified hundreds of host cell proteins that co-purify with the vector, some of which are considered high-risk for causing aggregation, modifying the drug product, or triggering immune responses.27Molecular Therapy – Methods & Clinical Development. Quantitative proteomic analysis of residual host cell protein retention across adeno-associated virus affinity chromatography As AAV therapies move toward larger patient populations and higher doses, tightening control over these impurities will matter more than ever.
AAV9 as a Delivery Vehicle for Gene Editing
AAV9 is not limited to delivering replacement genes. It can also carry gene-editing machinery, including base editors that make precise single-letter changes to DNA without cutting both strands. A compact adenine base editor packaged into a single AAV9 vector achieved dose-dependent editing across tissues in mice, reaching roughly 55 to 64% editing in liver, about 13 to 23% in heart, and 5 to 8% in skeletal muscle, matching or exceeding what previously required two separate AAV vectors working in tandem.28Nature Biomedical Engineering. Efficient in vivo base editing via single adeno-associated viruses with size-optimized genomes encoding compact adenine base editors
This approach has already been tested against a real disease model. In humanized mice carrying a mutation that causes autosomal dominant polycystic kidney disease, a single dose of dual AAV9 vectors delivering an adenine base editor corrected the pathogenic variant in kidneys, hearts, and livers. The treatment delayed cyst growth, decreased heart enlargement, and improved liver function.29Nature Communications. In vivo base editing rescues ADPKD in a humanized mouse model These results point toward a future where AAV9 is not just delivering genes but permanently correcting mutations in place.
The Cost and Access Problem
Zolgensma launched with a price tag exceeding two million dollars per dose, making it one of the most expensive medicines ever sold. That figure reflects the reality of gene therapies that are given once to treat rare diseases with small patient pools. Research, development, manufacturing, and the cost of treating complications all feed into the price, and existing health-technology assessment frameworks were not built to evaluate a one-time treatment that replaces years of chronic therapy.30PubMed Central. Gene Therapy for Neuromuscular Diseases: Health Economic Challenges and Future Perspectives Some health systems have adopted outcomes-based payment models that tie reimbursement to whether the treatment actually works over time, but these are still experimental. Until manufacturing costs come down and the pipeline broadens to include more common diseases, the tension between the transformative potential of AAV9 gene therapy and equitable access to it will remain unresolved.