Engineering adeno-associated viral vectors with improved in vivo targeting of cardiomyocytes

Adeno-associated viral (AAV) vectors are promising tools for long-term gene delivery in vivo as they are non-pathogenic, non-integrating and low immunogenic. While their high efficacy and good safety profile has already resulted in seven market approvals to date, a heart-specific AAV vector is still missing. AAV serotype 9 (AAV9) is commonly used for in vivo transduction of cardiomyocytes in rodents and considered to have a cardiac tropism. However, the high off-target expression in the liver limits the application of AAV9 for cardiac gene therapy in clinical settings due to the risk of hepatotoxicity. Here, we performed an in vivo screening of an AAV2 peptide display library in order to identify novel capsid variants with an improved specificity for cardiomyocytes. The initial library contained approximately 4 x 10^6 different 7-mer peptide insertions at position 587 of the capsid protein, which is located on the second highest protrusion of the viral capsid. Three selection rounds were performed in a disease context in mice with established cardiac hypertrophy after transverse aortic constriction (TAC)-induced pressure overload. By performing next generation sequencing, we identified 53 novel cardiomyocyte-enriched capsid variants. For further in depth analysis, we combined these top 53 cardiomyocyte-specific candidates to an eGFP-expressing barcoded sub-library that was again screened in vivo. By this approach, we were able to identify two capsid variants that showed a higher transduction efficiency as compared to previously described gold-standard variants including AAV9, while leading to a minor transduction of liver tissue. To confirm the results from the library screening, individual AAV vector variants expressing eGFP were systemically injected into mice. The two novel capsid variants showed an improved cardiomyocyte-to-liver transduction ratio as compared to AAV9 (2-fold and 4.9-fold, respectively), thereby confirming their cardiac-specific tropism. In vitro assays demonstrated that the retargeting of the vectors is accompanied by a reduced binding affinity to the AAV2 primary receptor heparan sulfate proteoglycan. Cross-neutralizing antibodies between wild type AAV9 and our capsid variants were not detected. Therefore, our novel variants could be used in combination with AAV9 for repeated injections. Interestingly, one of the capsid variants showed a disease-specific tropism and was inefficient in transducing cardiomyocytes of healthy mice. Further animal experiments in mice after TAC surgery as well as in a Mybpc3-mutant mouse line with hypertrophic cardiomyopathy will be performed to validate the specificity of this variant for hypertrophic cardiomyocytes. In ongoing studies, we are also evaluating the potential superiority of the novel variants to deliver therapeutic circular RNAs in a doxorubicin mouse model of cardiotoxicity. Lastly, we also plan to test our novel capsid variants in human living myocardial slices and cardiac organoids to investigate their cross-species activity. Discovering novel AAV vectors that target cardiomyocytes, particularly those affected by disease, will improve the safety and feasibility of cardiac gene therapies, thereby enhancing the treatment options for many patients with inherited and non-inherited cardiac diseases.