Conserved Amino Acid Sequences in Parvovirus B19 and Adeno-Associated Virus Stimulate Functionally Cross-Reactive CD8 T Cells.

Use of adeno-associated virus (AAV) to achieve liver-targeted gene transfer for the treatment of hemophilia B in human subjects resulted in the expression of therapeutic levels of Factor IX. However, expression of Factor IX was transient and receded to baseline levels over a period of several weeks. The loss of transgene expression was coincident with a rise in CD8+ T cells specific for sequences of the AAV capsid protein, a phenomenon never observed in small and large animal model studies of this treatment. In this study, we tested whether capsid sequences conserved between the pathogenic parvovirus B19 and AAV (~25% identity) could result in cross-reactive activation of capsid specific T-cells. Functional cross-reactivity could lead to an effective vaccination against AAV following exposure to B19 virus, a common infectious virus which was recently shown to elicit a strong CD8 T cell response that can persist for up to one year after exposure to the virus. Furthermore, B19 virus tropism is restricted to humans and this could explain the absence of this outcome in animal models. In order to test our hypothesis, Balb/C mice were immunized against the capsid protein of AAV serotypes 2 or 8 (AAV2 and AAV8). Isolation of splenocytes from mice immunized against AAV2 capsid was followed by ELISpot analysis that compared the response against the immunodominant AAV2 CD8 T cell epitope (VPQYGYLTL) to the response against the homologous B19 capsid epitope (PPQYAYLTV). Indeed cross-reactivity was observed: using ELISpots for IFN-g, incubation with the AAV epitope yielded 1866 spot-forming units (sfu) per 10⁶ splenocytes (average of 3 determinations), and incubation with the B19 epitope yielded 537 sfu/10⁶ splenocytes, while incubation with irrelevant peptide yielded <10 sfu/10⁶ splenocytes (n=5 mice per group). This decreased but significant cross-reactive population of T cells was similarly observed following immunization against AAV8 capsid and ELISpot analysis, with an average of 2137 sfu/10⁶ splenocytes in response to the immunodominant AAV8 CD8 T cell epitope (IPQYGYLTL) and an average of 486 sfu/10⁶ splenocytes in response to the B19 epitope. These results were further substantiated by intracellular cytokine staining, which defined CD8+ T cells as the responding cell population and showed cross-reactivity at levels consistent with ELISpot analysis. To test whether such cross reactivity could be observed in humans, we identified several epitopes through bioinformatics which showed the potential to be cross-reactive in the context of specific HLA haplotypes. Human splenocytes from one B0702 anonymous donor and three B44 anonymous donors were subjected to three rounds of in vitro stimulation with the homologous AAV and B19 capsid peptide sequences. Subsequent ELISpot analysis showed that none of these expansions resulted in the selective expansion of cells specific for these peptides. We conclude from these studies that CD8+ T cells specific for parvovirus capsid sequences can be functionally cross-reactive with AAV capsid, but that further studies of human lymphocytes will be required to determine the significance of this finding for human gene transfer trials.