A 2010 report that Adeno-associated virus 2 (AAV2) encodes an additional protein in the second open reading frame of the cap(sid) gene significantly changed our perception of the biology of this virus. This so-called assembly-activating protein or AAP appears to interact with the major viral capsid protein 3 (VP3) and to chaperone its assembly into virus-like particles, via mechanisms that remain unknown. Moreover, it was shown that deliberate mutation of the AAP gene diminishes AAV particle production.As the AAP open reading frame fully overlaps with that of the AAV capsid proteins VP1/VP2 (and partially VP3), it is likely that the AAP gene is inadvertently disrupted or mutated during molecular AAV evolution protocols that involve capsid gene shuffling (homology-directed recombination of cap gene fragments). To study this possibility comprehensively and to provide solutions, if needed, we first investigated the role of AAP for 9 further AAV serotypes other than AAV2. Therefore, we knocked out AAP expression in serotypes 1, 3-9 and rh10 while leaving the rep and cap genes intact. Indeed, all mutants were markedly impaired in particle production, confirming that functional AAP is critically required for at least 10 different AAV serotypes.Notably, all vectors were fully restored upon co-delivery of the corresponding AAP via an extra plasmid during particle production. Furthermore, when testing the cross-reactivity of all AAPs with the AAV2-AAP mutant (AAV2mut), we observed a broad interchangeability except for AAP4 and 5 which predominantly rescued their cognate serotype.These results were confirmed independently in a baculovirus AAV2 vector production system where we also observed a dramatic reduction in particle yields upon AAP mutation. Interestingly, in both, mammalian and insect cells, we consistently noted that absence of AAP reduced the steady-state levels of VP proteins, suggesting a similar biological function of AAP in the two heterologous cellular systems. As we could rule out an effect on the transcriptional level, we postulate that AAP-mediated chaperoning of AAV virion assembly indirectly stabilizes VP proteins in cells of different species.To explicitly dissect the role of AAP in the context of molecular AAV evolution, we produced five different AAV capsid libraries with increasing complexities, subcloned 46 randomly selected shuffled AAPs from these and tested them with the AAV-AAP mutants. Out of these 46 shuffled AAPs, 37 could rescue AAV2mut, and two more efficiently complemented an AAV5-AAP mutant. In total, 84.8 % of all tested AAPs were functional.Of note, the highest proportion of defect AAPs was found in libraries that contained the diverse capsids of AAV4 and 5. We thus shuffled serotypes 2, 4, 5, 8 and 9 and produced viral libraries with or without a cocktail of excess AAP from all five serotypes. Strikingly, albeit up to 40% of all AAP genes were predicted to be non-functional as a result of inadvertent shuffling, AAP trans-complementation had no significant effect on particle yields.Collectively, our data suggest that albeit AAP is disrupted to some extent during AAV capsid gene shuffling, this is likely not a major restriction for the efficiency and success of this technology.
Read full abstract