T Cell Responses to AAV Vector Capsid Limit the Duration of Transgene Expression in Humans after Liver-Directed Gene Therapy.

In a Phase I/II study of gene transfer for hemophilia B, an adeno-associated viral vector serotype 2 (AAV2) was introduced into the liver of human subjects and therapeutic levels of transgene expression (up to 11% of normal) were reached in one subject. However, the duration of gene expression was limited; four weeks after infusion, levels of circulating factor IX (F.IX) began to decline, and returned to baseline by week 10. This phenomenon was accompanied by a mild, self-limited, increase in liver enzymes (transaminitis). After a similar phenomenon was observed in another subject, the clinical trial was halted. One hypothesis for the loss of transgene expression is that immune mediated destruction of transduced hepatocytes caused the delayed rise in transaminases and loss of F.IX expression. IFNγ ELISpot analysis of peripheral blood mononuclear cells (PBMCs) from subjects in the clinical study and from normal donors, and bioinformatics tools for the prediction of MHC class I binders, were used to define two possible MHC Class I-restricted T cell epitopes for two HLA types common in the general population (B*0702 and B*0801). We then designed MHC class I epitope-specific pentamers for the detection and study of CD8+ T cells reacting to the predicted AAV2 capsid epitopes. Indeed, in the two subjects who developed transaminitis after vector infusion, AAV-specific CD8+ T cells were detected by pentamer staining of PBMCs up to two years later. In contrast, normal donors rarely had detectable AAV-specific CD8+ T cells in peripheral blood; even after several rounds of in vitro stimulation with vector capsid antigens, we often did not find AAV-specific CD8+ T cells by pentamer staining. We also performed functional assays, including intracellular cytokine staining (ICCS) and cytotoxic T lymphocyte (CTL) assays, on expanded AAV-specific T cells from one of the subjects in the clinical trial. An Epstein-Barr-virus-transformed lymphoblastoid cell line (LCL) derived from an HLA-matched donor was incubated with an AAV2 epitope, with the homologous peptide from the AAV-8 capsid sequence, or with a known HIV gag that is also HLA B*0702 restricted. Similarly to what observed in mice (Sabatino et al., unpublished), we found that IFNγ was specifically produced when the expanded T cells were stimulated with AAV antigens of both serotype 2 and serotype 8, but not with the HIV gag epitope, in the context of HLA B*0702. Peptide-loaded LCLs were used as targets in a CTL assay and incubated with in vitro expanded T cells (effectors). Specific lysis was observed at an effector-target ratio of 100:1 for the AAV2 epitope and for the homologous peptide from the AAV8 capsid. Together, these data provide direct evidence that humans can mount a cytotoxic immune response to AAV capsid proteins. This T cell response may account for the limited duration of transgene expression that we have observed in our clinical study. Moreover, the use of alternate serotypes may not easily avoid this immune response, as T cells from subjects infused with AAV2 showed functional responses to AAV8-derived peptides by both ICCS and CTL assays. We conclude that the use of immunomodulatory therapy may be a better approach to ensure durable transgene expression in the setting of liver-directed gene therapy.