Can the immunogenicity of AAV vectors be mitigated?

Immunogenicity of biological products is a big obstacle to their clinical development. Immune responses against these products can hinder their safety and efficacy profiles and, in some cases, pre-existing immunogenicity can prevent patient eligibility to receive therapies for rare disorders that do not have any alternative therapy. The immunogenicity challenge is especially prominent in gene therapy products. Due to the complex mechanism of action and the use of natural viruses to deliver the genes, the immune response to viral vectors that are used in gene therapies is multifactorial and involves activation of multiple arms of the immune system. These immune reactions often result in infusion related reactions, immune related liver toxicities, neutralization of the therapeutic effect, and importantly, development of neutralizing titers that prevent repeated dosing, essentially limiting these life-saving therapies to a single dose.    

Adeno-associated virus (AAV) is currently the most commonly used and developed viral vector for gene therapy because it is non-pathogenic to humans with minimal integration of the viral genetic information into the human genome, thereby reducing the risk for DNA damage and unpredictable consequence of vector sequence integration. Immunogenicity of AAV vectors is a significant safety and efficacy issue. The immune response to the AAV capsid causes neutralization and accelerated clearance, limiting the therapy to a single dose. In some cases, cytotoxic immune responses against the patient's transfected cells can happen, reversing the gene therapy effect. T cells (both CD4 and CD8 T cells) play an important role in most of these immune responses and are monitored on a regular basis during AAV clinical trials.

Immunodominance is a phenomenon in which T cell responses are restricted to a small number of peptide epitopes derived from a small set of antigenic precursors. In this study, we successfully identified T-cell-reactive epitopes in the AAV9 capsid (Fig 1a), including one promiscuous and immunodominant epitope that is highly conserved within most natural AAV serotypes (Fig 1b). The immunodominant epitope in AAV9 was identified between amino acids 307 and 327 and immune response to this epitope was found in 23% (12 of 52) of donor samples. In addition, this epitope is not located on the surface of the AAV9 capsid and is recognized by CD4 T cells through HLA-DP presentation. The binding of this epitope was also predicted using the Immune Epitope Database’s (IEDB) binding algorithms and confirmed that the epitope 307-327 is HLA-DP restricted and promiscuous.

Importantly, we noted that this epitope is conserved between AAV serotypes (conservation rate: 86%) with the exception of amino acids R312, L313, N314, F315, and L317 (Fig 1b). Non-conserved amino acids were found in the AAV5 serotype (R299, S300, R312, V313, and I315), which is among the most distantly related to AAV9 in the AAV phylogeny tree. Based on this observation, we engineered the epitope 307-327 through rational swapping of individual residues. First, we predicted the binding affinity of the AAV5 sequence to candidate amino acids responsible for the immunosilencing of the AAV9 immunodominant epitope (Fig 1c), and then chose two AAV9 variants (AAV9_VI and AAV9_RSRVI) that were predicted to bind to fewer HLA class II molecules and therefore are expected to be less promiscuous. Both new AAV9 variants showed a diminished T cell response (Fig 1d) in human PBMC from 4 donors after expansion and restimulation. To reduce or eliminate immunogenicity, it is vital that the method that eliminates MHC-binding epitopes does not disrupt vector structure and function. The two chimeric AAV9/5 capsids resulting from this engineering effort showed similar performance features in all assays including in vitro and in vivo transduction (Fig 1e), implying that the induced chimerism has not perturbed essential functions of the AAV9 capsid.

Such rational designs of a chimeric AAV vector can result in safer and more efficacious gene delivery vectors by reducing T cell mediated toxicities and by preventing T cell mediated death of transduced cells, potentially resulting in longer persistence of transgene expression. We envision that similar rational immunosilencing could be applied to other AAV vectors or epitope regions.