Purpose: 
Adeno-associated virus (AAV) vectors are promising delivery tools for human in vivo gene therapy. However, broad tissue tropism and pre-existing immunity against natural serotypes limit their clinical use. In addition, a large proportion of the vectors are often lost in off-target tissues, especially in the liver. Thus, to identify more specific and safer AAV vectors for cardiac gene therapy, we performed an in vivo AAV2 peptide display library screening in a murine model of pressure overload-induced cardiac hypertrophy.

Methods and Results: 
An in vivo peptide display library screening based on AAV2 was performed in murine hypertrophic cardiomyocytes. The initial AAV2 library contained approximately 4 million different capsid variants, which differ in 7mer peptide insertions that are displayed at the surface of the virus capsid. Three rounds of in vivo selection in mice with pressure overload-induced cardiac hypertrophy identified two AAV capsid variants, AAV2-THG and AAV2-NLP. Both variants showed significantly improved specificity for cardiomyocytes compared to the parental serotypes, reaching transduction levels comparable to AAV9 (the gold standard for cardiomyocyte transduction in vivo). Importantly from the clinical and economic perspective, in contrast to AAV9, significantly fewer vector genomes accumulated in the liver (the main off-target organ). The variants also showed reduced transduction efficiency of non-cardiac cells as well as heparan sulfate proteoglycan (HSPG)-independent cell entry, while efficient transduction of human induced pluripotent stem cell-derived cardiomyocytes indicated cross-species activity. Notably, the re-targeted tropism of AAV2-THG and AAV2-NLP allowed for efficient and consecutive in vivo targeting of cardiomyocytes with AAV9 and the novel variants, a promising feature in light of possible vector (re-)application scenarios. Superiority of our novel variants compared to AAV9 was further confirmed in the pressure overload-induced cardiac hypertrophy mouse model, where delivery of the anti-hypertrophic long non-coding RNA H19 already mediated significantly stronger therapeutic effects at low viral vector doses compared to AAV9.

Conclusion: 
AAV2-THG and AAV2-NLP are promising novel tools for cardiac-directed gene therapy as both capsid variants clearly outperform AAV9 with regard to specificity and therapeutic efficiency of in vivo cardiomyocyte transduction.