Neutralizing the Neutralizers in AAV Gene Therapy

Main Text

Pre-existing immunity has long been a thorn in the side of developers of gene therapy drugs based on adeno-associated viral (AAV) vectors. Except for young children, neutralizing antibodies (NAbs) are frequently encountered in the human population, with a seroprevalence of 15%–50%, depending on capsid and geographical location. NAbs can have a substantial impact on the efficacy of gene transfer, particularly in protocols that rely on vascular routes of vector administration.¹ Consequently, many patients with genetic disease will not be able to benefit from these gene medicines. Immune suppression, while showing promise as a tool to eliminate B cell responses and thus enabling re-administration of vector, is unlikely to solve this problem. Methods to remove antibodies from plasma are established but require specialized medical procedures and have limitations. Now there is help from an unexpected source: bacteria. Work recently presented at the American Society of Gene and Cell Therapy (ASGCT) 2020 annual meeting utilized Mycoplasma protein M, which binds to immunoglobulins with broad reactivity, thereby blocking the interaction with antigen.^2,3 Hence, administration of protein M prevents all existing antibodies from binding antigen, including NAbs to AAV capsid so that they cannot neutralize the virus. In a different approach, immunoglobulin G (IgG)-degrading enzymes of Streptococcus (IdeS), cysteine proteinases that cleave IgG at a single site below the hinge region (yielding F(ab’)₂ and Fc fragments), have been adapted to “inactivate” immunoglobulins prior to AAV administration. Here, the hope was that F(ab’)₂ fragments, which still bind antigen, will fail to neutralize the vector. Among these endopeptidases, Imlifidase derived from Streptococcus pyogenes is already in clinical trial to eliminate donor-specific antibodies against human leukocyte antigen (HLA) in order to prevent rejection of kidney transplants in highly HLA-sensitized patients. The protease is also in clinical development for the elimination of pathogenic autoantibodies in Guillain-Barré syndrome. An engineered form of IdeS is recombinantly produced in E. coli and, importantly, cleaves all 4 subclasses of human IgG with high efficiency.

Leborgne et al.⁴ now report the usefulness of IdeS for successful gene transfer with AAV vectors in the face of NAbs, a strategy that enables gene therapy despite pre-existing immunity stemming from natural infection or prior vector administration. IdeS fails to cleave murine IgG and can therefore not be tested in mice immunized against capsid using traditional methods. Therefore, the authors tested pre-existing immunity using a passive immunization model based on transfer of human IgG to mice. After intravenous administration of IdeS on 2 consecutive days, neutralizing activity against AAV8 was eliminated, and hepatic gene transfer was successful when vector was given 1 day after the second dose of the protease. The authors then tested this protocol in two non-human primates (NHPs) with similar levels of pre-existing NAbs against AAV8. One animal received the protease, while the other served as control. The IdeS pre-treated animal achieved 2–4-fold higher levels of transgene expression (factor IX) from the AAV8 vector, and total IgG levels recovered within 1 week. Next, eight NHPs received hepatic gene transfer with AAV-LK03 expressing a lysosomal storage enzyme. As expected, vector administration resulted in the formation of high-titer NAbs against LK03 (a hepatotropic capsid containing sequences from AAV3 and other serotypes). Seven months later, the animals continued to show very high titers of NAb (≥1:1,000). Five of these primates were then treated with IdeS, followed by re-administration 1 day later with a vector that was identical to the first vector in capsid and dose (2 × 10¹² vg/kg) but contained a different transgene (human factor VIII [hFVIII]). The 3 remaining animals were also given this second vector, but without IdeS treatment, and thus failed to show transgene expression. In contrast, the five IdeS-treated animals reached therapeutic circulating hFVIII levels of up to 50% of normal (but lost expression 1 month later due to antibody formation against the immunogenic hFVIII). While the study did not include naive animals to see if the maximum possible levels of transgene expression was reached, the summary of the data from the two primate experiments strongly support the usefulness of the method to achieve gene therapy despite the presence of NAbs.

As opposed to using decoy empty capsids to soak up pre-existing antibodies to capsid, the IdeS approach avoids the introduction of additional capsid antigen. It is also not capsid-specific and therefore can be applied to any AAV serotype. This is particularly relevant for patients who have cross-reacting NAbs to multiple serotypes. For instance, NAbs formed by humans who received AAV8 cross-react with AAV5, which is in contrast to the experience in NHP studies.¹ One question about the approach is whether immunity to IdeS matters. The authors have shown a high prevalence of IdeS antibodies in the sera from 52 healthy donors. They also present in vitro data showing that the proteolytic activity of IdeS inactivates NAbs against itself, thus suggesting that antibodies to IdeS do not appear to interfere with its proteolytic domain, although the outcome may be different upon repeated in vivo exposure to the bacterial enzyme. Interestingly, NAbs to LK03 were still considerably high after IdeS treatment, as measured by in vitro assay. Normally, one would have anticipated that such titers would prevent re-administration. However, one has to keep in mind that IdeS generates Fab fragments rather than entirely removing the antibody molecules, which may account for differences in neutralization between in vivo and in vitro settings. Additionally, proteolytic cleavage of NHP IgGs also does not appear to be as complete as for human IgG, and residual levels of single-chain IgG (scIgG) with potential neutralization activity was still detected in these animals. IdeS activity diminished within 24–48 h, which explains why antibodies against the transgene product were able to emerge several weeks after gene transfer. It would have been interesting to test for the ability of further IdeS administration to restore hFVIII expression.

It would also be interesting to test whether IdeS treatment could improve safety of high-dose systemic AAV gene transfer for neuromuscular and storage disorders, where antibodies against capsid are suspected to cause complement activation and other immunotoxicities. However, this may not be the case if the problem is primarily caused by IgM. In general, inactivation of IgG may not entirely eliminate the need for immune modulatory/immune suppression regimens in clinical protocol design because the approach does not target B or T cell responses to transgene and vector, including memory responses. Nonetheless, the field now has a powerful tool in the kit to deal with a major obstacle for wider use of in vivo gene therapy and for re-administration of vector when needed.