Interest in the field of regenerative medicine has greatly advanced in recent years. For example, the Alliance for Regenerative Medicine reported almost US$20 billion investment in 2020 that was divided across cell therapy, gene therapy, cell-based immune-oncology and tissue therapy, with gene therapies alone having over 400 products in clinical trials in 2020 [

]. Gene therapy as an approach treats or prevents a disease by modifying the underlying genetics. This can involve altering, repairing, or replacing a faulty gene to provide a functioning phenotype. Gene therapy has shown promise in preventing or treating certain inherited monogenic disorders (see Glossary) such as hemophilia, sickle cell disease, and Leber congenital amaurosis [

].

Within gene therapy, vectors derived from AAVs have emerged as safe and efficacious for long-lasting transgene expression across a range of tissue types [

,

]. AAVs are small, non-enveloped viruses that are not replication competent unless in the presence of a helper adeno- or herpes virus []. Wild-type AAVs are not associated with any known human disease [

]. It is this lack of pathogenicity, combined with the ability to deliver up to approximately 4.8 kb of single-stranded (ss)DNA into a range of different tissue types that makes them an attractive vector for gene therapy delivery [

,

].

Currently, there are over 260 ongoing clinical trials that use AAV as the transgene vector (Gene Therapy Clinical Trials Worldwide provided by The Journal of Gene Medicine, Wiley; https://www.abedia.com/wiley) (Figure 1, Figure 2). An advantage that AAVs hold over other viral vectors, such as lentivirus, is that they interact minimally with the innate immune system, demonstrating low toxicity and inflammatory responses [

]. Clinically, no acute responses are typically observed.

AAV serotype capsid morphology is highly conserved across recombinant and wild-type capsids; therefore, successful transduction is greatly influenced by previous exposure of the immune system to naturally occurring AAVs [

]. The immune response to AAV therapies is a complex picture and outside the scope of this review. For detailed reviews, please see [

,

,

]. In short, once internalized, the virus is trafficked via endocytosis to the nucleus, where capsid proteins seek entry. The virus is then uncoated and the ssDNA is converted to double-stranded (ds)DNA, and in the case of therapeutic gene therapy, the relevant transgene is then expressed. Following transduction, AAV capsid proteins are broken down in proteasomes, which allows processing of peptides to MHC class I and MHC class II antigen-presenting molecules initiating T cell responses that lead to B cell activation and the production of anti-AAV IgG/neutralizing antibodies (NAbs) (Figure 3); the immune system is then primed to fight reinfection with AAV, essentially acting as a vaccine against re-administration of the therapy if required and reducing the therapeutic transduction efficiency.

Long-term follow-up of one of the first approved AAV vector gene therapies in 2020 demonstrated sustained efficacy in the treatment of hemophilia B in adult males. Fifteen years after administration, no major safety concerns were observed; however, the study highlighted persistent high-titer NAbs for the therapeutically administered AAV2 serotype and, with some cross-reactivity, neutralizing AAV5 and AAV8 [

]. While undoubtedly a success, highlighting AAVs as an attractive choice for gene therapy vectors, the presence of high-titer NAbs remains a concern as it hinders the efficacy of future dosing and drastically reduces the number of patients the therapy can treat effectively.

Although not associated with any clinical symptoms [

], exposure to AAV occurs naturally and often early in life [

]. The seroprevalence of anti-AAV NAbs varies for differing serotypes and geographically between 15% and 60% [

,

,

]. Equally, although AAV vectors demonstrated long-term persistence in a clinical setting [

,

], due to the non-integrative mode of action, it is possible that a further dose is required later in a patient’s life to ensure therapeutic levels of transgene expression. Finding a safe and effective method or technology to overcome and evade AAV neutralization by the immune system will allow efficient and successful curative gene therapy delivery with lifelong therapeutic benefit to the greatest number of patients possible.

The presence of anti-AAV NAbs at even low titers has been demonstrated to reduce successful transgene expression [

,

]. Currently, clinical trials exclude seropositive patients [

], vastly limiting the pool of patients the therapies are able to treat. The ideal successful technology would allow the treatment of seropositive patients while dampening the immune response to the vector and allowing future re-administration. At the same time, ideal immunosuppression methods, recently reviewed in [

] and summarized in Figure 4, require a favorable benefit/risk profile.

Despite being of low immunogenicity compared with other viruses, the immune system, which has evolved to be finely tuned to respond to pathogens, recognizes the AAV vectors as a challenge to the immune system, thus highlighting an important limitation of these gene therapy vectors. There are now increasing data demonstrating that high-vector dosing and doses containing excess empty capsids has shown dose-dependent toxicity [

,

,

], reinforcing the urgency to reduce the interactions between the AAV vector and the immune system and potentially rethink the single-large-dose approach to treatment [

].

In immune-privileged tissues such as the eye, little to no impact of the pre-existing humoral response was observed after vector administration and it was possible to repeat treatment on the contralateral eye [

]. It is clear that this is not applicable for other routes of administration to target tissues, particularly when AAV vectors are dosed systemically. An approach focusing on avoiding AAV contact with NAbs, however, does have benefits. A successful example of this evasion approach was performed through injection of AAV8 into the portal vein branch via balloon catheter that allowed hepatocyte transduction in the presence of low-titer NAbs. Although not applicable in the presence of higher-titer NAbs, and so far not applied in tissues other than liver, it is encouraging that local NAb avoidance can be suitable in some cases of tissue targeting.

An additional approach that allows transduction in seropositive patients is saturation of the NAbs with antigens. A possible method for this would be the administration of empty/decoy capsids or a high vector dose, essentially allowing the NAbs to bind, while a therapeutically efficacious dose of vector remains available to transduce cells. In mice, the administration of a 50- or 100-fold excess of empty capsids is sufficient to obtain full liver transduction in the presence of 1:10 and 1:100 NAb titers. In the presence of very high titers (>1:3000), transduction could not be recovered [

]. This approach greatly increases the dose required and the associated manufacturing costs as well as the immunogenic burden.

One way to increase the treatable patient population is to select an AAV serotype where NAb seroprevalence is low. In screening 888 human serum samples from ten countries, Calcedo and Wilson observed that AAV2 NAbs were the most prevalent, followed by AAV1 and AAV7 and AAV8 [

]. It is clear that, where tissue tropism allows, vector serotype selection is important. However, even for the least immunogenic serotypes, there are many documented cases of pre-existing immune response [

,

,

,

,

,

,

].

Further, as understanding of AAV vector capsid structure and morphology increased, it provided the opportunity to use directed evolution to generate novel, engineered AAV vectors to which there will be no previous exposure in the population [

,

,

]. This approach had varied effectiveness depending on the location of the capsid mutations. It is also possible that these mutations may alter tissue tropism and ultimately reduce transduction [

]. This does not overcome the challenge of vector re-administration unless used in combination with immune suppression [

], as cross-reactivity of NAbs across serotypes can still be observed and serotypes must be engineered to ensure they avoid all NAbs [

].

An alternative to the genetic approach is to modify the capsid chemically. Lysine residues on the AAV surface were covalently coupled with phenyl isothiocyanate anchors to allow coating with specific ligands [

].

Chemical conjugation with polyethylene glycol (PEG) has been shown to reduce AAV neutralization, however, with some interference with tissue tropism and transduction efficiency [

]. A challenge of these modifications is controlling the location of the functionalization reaction. This can be overcome using genetic code expansion tools, allowing the introduction of non-naturally occurring amino acids in the capsid at specific locations [

], which has been utilized to enable site-specific PEGylation showing reduced antibody recognition in vitro and reduced immune response in a rat model [

]. While PEGylation reduces the humoral response, AAV is still subject to cell-mediated immunity; to overcome this, there is the potential for the conjugation of immunosuppressive materials to AAV vectors. As demonstrated with immunosuppressive zwitterionic phosphoserine (PS)-containing polypeptide conjugated to AAV8, maintaining transduction efficiency and original tissue tropism and suppression of anti-AAV production in a mouse model [

] opened an avenue for targeted localized immune suppression, removing the increased infection risk of traditional systemic suppression.

Another possible technique to physically preventing neutralization from NAbs is to associate AAV with microvesicles/exosomes (vexosomes). To perform this, AAV vector production is modified such that the vectors produced are associated with the surface and the interior of microvesicles, thus producing vexosomes [

]. This was performed by pelleting AAV-associated microvesicles by differential centrifugation of the production bioreactor supernatant. Although the mechanism by which microvesicle-associated AAV travels from the cell surface to the nucleus is not known, they were able to successfully transduce cells in the presence of a robust concentration of NAbs (1:2000), much higher than is typically seen in patients. Although higher NAb titers were still able to block the transduction, the resistance to neutralization shown by vesicle-associated AAVs is promising [

].

Hemophilia B patients dosed with AAV2-hFIX have shown expression lasting 12–15 years and counting [

]. Although promising, the full picture of how long transduction may last is unclear and may vary depending on the target tissue and the age of the patient treated. Also, in some diseases capable of drastic reduction of life expectancy when untreated, a treatment early in life may be required [

]. In these instances, re-administration of vector may be needed in the future.

When considering vector re-administration and prevention of NAb generation, perhaps the most clinically ready approach is the use of immune-suppressing strategies similar to those used in antitransplant rejection [

]. A benefit of this is that it uses already-approved therapies and drugs that can be selected based on the immune system components targeted (Table 1) . A combination of ciclosporin A, calcineurin inhibitor lowering T cell activity, and rituximab targeting B cells demonstrated a reduction in NAbs allowing vector re-administration in non-human primate (NHP) model [

] and this has been shown to be partially effective in reducing B cell responses in humans [

]. As rituximab is unable to target plasma cells, which do not express the CD20 antigen directly, it is not applicable to those with pre-existing immunity [

].

Alternatively, rapamycin has been shown to reduce the immune response to AAV in a rat model blocking the humoral response to AAV8 capsid [

]. Further, when formulated in nanoparticles, rapamycin has shown greater efficacy in preclinical models in reducing NAbs and allowing vector re-administration in the primate model [

]. These approaches appear promising, particularly for re-administration to previously seronegative patients. As with rituximab and other pharmacological interventions are ineffective at bringing those previously exposed to wt-AAV into the treatable patient pool. The animal models studied must be seronegative to begin with.

As with all risk–benefit trade-offs, consideration must be given to ways in which the risk can be reduced; in this case, it may be possible to adopt a more targeted approach to immune suppression. For example, Toll-like receptors (TLRs) were identified as playing a role in AAV recognition and subsequent immune response and MyD88 protein was identified as a critical regulator of both T and B cell adaptive immunity against AAV [

]. CRISPR-Cas9 technology enables a platform to temporarily downregulate the MyD88 gene, briefly dampening the immune response, allowing administration of the vector and subsequently reducing the level of NAbs generated in response when observed in the mouse model [

]. As this reduces the immune response, it does not allow the treatment of seropositive patients. However, it may allow re-administration of vector in future and could be used in conjunction with novel serotype vectors or those with a lower prevalence of NAbs.

Temporarily removing NAbs from the target tissue prior to vector dosing is an option for both seropositive patients and vector re-administration. Plasmapheresis is an extracorporeal technique allowing the removal of substances from plasma. Blood is withdrawn and plasma exchanged and filtered to remove free antibodies and circulating immune complexes from the plasma, which can be reinfused. Multiple cycles of plasmapheresis combined with immunosuppression were able to greatly reduce the levels of IgG in serum [

]. In a study of ten patients seropositive for AAV1, 2, 6, and 8, up to five cycles of plasmapheresis reduced NAb titers in some patients to undetectable (

], which would allow treatment of seropositive patients and potential delivery of the vector again if required in the future. Two cycles of plasmapheresis were sufficient to reduce NAb levels in a NHP model, enabling successful transduction comparable with seronegative animals [

], supporting the hypothesis that plasmapheresis may permit successful treatment in patients with pre-existing immunity.

A logical follow-up to this is the development of a NAb-specific immunoadsorption column for plasmapheresis (Figure 5). Initial studies have demonstrated promising selective binding to NAbs from serum by using full AAV particles as ligands, coupled to a chromatography stationary phase. Immobilized AAV reduces the NAb titer and effective liver transduction was achieved in a passive immunized mouse model [

,

]. A significant challenge to the development of these tools is that NAbs are a polyclonal population, which varies serotype to serotype and between different patient populations. The binding efficiencies of some NAbs may differ in free AAV versus immobilized AAV. This approach is, however, highly promising, but requires consideration of ligand coupling density versus capacity, resin reuse cycles, and ligand leakage to ensure feasibility and scaling into larger animal models and then humans.

A promising alternative to IgG depletion by plasmapheresis is use of the endopeptidase imlifidase (IdeS). This drug has been approved as safe and effective and is in use for kidney transplant in HLA-sensitized patients. Cleaving IgG reduces the NAb titer without the need to remove and reinfuse patient plasma [

]. The method sufficiently reduces the NAb titer to allow successful transduction of liver tissue in both mice and NHPs and crucially enabled subsequent successful re-administration of the vector [

,

]. In vitro testing on human plasma has also shown digestion of IgG, reducing anti-AAV8 titers even in the presence of anti-IdeS antibodies that are diffused in the human population [

]. As with nonspecific plasmapheresis, this approach depletes all IgG indiscriminately increasing the risk of infection; however, the safety is already demonstrated in patients with transplants. Therefore, although not without risks, IdeS provides a less invasive approach than plasmapheresis.

It is clear that technologies to overcome humoral immune responses are needed if AAV gene therapy is to be of benefit to as many patients as possible. The current approaches of treating only seronegative patients and targeting immune-privileged patients have demonstrated the potential of these treatments but only for a fraction of possible patients and indications. Evading the initial interactions with NAbs through modified capsids chemically and genetically is promising for initial dosing. However it is likely the vector will be recognized by NAbs in the future, preventing a second dose, and as the field is still in its relative infancy, clinically it remains to be seen whether this may be required or whether lifelong transgene expression is expected. Plasmapheresis combined with immune suppression appears to be a reasonable option for many cases; however, depletion of all IgGs and broad-spectrum immune suppression brings clinical risks. Therefore, it is clear that a method to deplete NAbs specifically via targeted immune suppression, plasmapheresis, or other means is required. For AAV-mediated gene therapy to reach its full potential, further understanding in specific areas is required (see Outstanding questions). One concern is that the NAbs developed may vary in structure from patient to patient, making NAb plasmapheresis column development challenging, and although the use of IdeS and conjugation of AAV to immunosuppressive moieties are promising emerging approaches, previous promising techniques show that it remains to be seen how well these will transfer from animal models to humans. However, the rate at which understanding of the immune response to AAV vectors has increased over recent years alongside vector understanding is promising, allowing more sophisticated methods of overcoming and evading AAV neutralization to develop and ultimately bring life-changing treatments to patients who need them.