Long-term protection against SARS-CoV-2 by antibody delivery mediated by an AAV-vector system

I. Abstract The SARS-CoV-2 pandemic led to the rapid development of highly effective vaccines that reduced severe cases of COVID-19. Nevertheless, certain vulnerable patient groups remained that show an impaired response towards vaccination. This includes patients undergoing chemotherapy or recipients of solid organ or cell transplantations or HIV-1 infected people. Here, a long-acting passive immunization strategy could be beneficial. It has been previously demonstrated that the delivery of monoclonal antibodies by Adeno-associated viral vectors leads to protective antibody concentrations for several months and is able to prevent infection. Therefore, a vector construct was designed that encoded for TRES6, an antibody neutralizing SARS-CoV-2 by binding to the spike protein. After optimization of the design, the best vector construct was packaged into the hepatotropic AAV8 and the myotropic AAVMYO capsids. After a dose finding in syngeneic TRIANNI mice, which were originally used for the generation of the TRES antibodies, the dose of 5x1011 vg was chosen for the intravenous application of AAV8-TRES6 and AAVMYO-TRES6 vectors. In a murine challenge study, both vectors were able to protect mice from SARS-CoV-2 infection and disease. After intravenous injection, both vectors induced high TRES6 serum titers over one year, while AAVMYO-TRES6 exceeded AAV8-TRES6 in terms of serum antibody concentration. After AAVMYO injection, the TRES6 serum concentration peaked in week 24 at a concentration of 511 µg/ml. It remained as high as 120 µg/ml 52 weeks after injection and was 3.7 times higher than AAV8-delivered TRES6. The neutralization capacity of AAVMYO-TRES6 serum collected one year after vector application was comparable to the human sera of vaccinated donors. Despite a systemic distribution of vector DNA, the vector encoded RNA showed a restricted pattern. AAV8 showed active transcription only in the liver, while AAVMYO transcription was detectable only in the heart and the quadriceps muscle. This strict separation of TRES6-producing tissues resulted in divergent Fc glycosylation and Fcγ receptor binding, of which the AAVMYO delivery had beneficial properties regarding the protection against SARS-CoV-2. After intramuscular injection of 1x1011 vg TRES6 encoding vectors, AAVMYO, as well as AAV8, induced a stable serum antibody titer for up to one year as well. Here, AAVMYO also induced higher TRES6 concentrations at all examined time points compared to AAV8. Interestingly, the TRES6 concentration was one year after the intramuscular injection similar to the concentration one year after intravenous application, despite administering one-fifths of the i.v. dose. After i.m. injection, vector DNA was also detected in all examined organs and similar transcriptional active tissues was found comparing the capsids. Therefore, also no differential Fcγ receptor binding was observed between AAV8 and AAVMYO delivered TRES6. When dissecting the transduction mechanism in vitro, no differences in intracellular transport and second-strand formation in the nucleus were found for AAVMYO and AAV8 in different cell lines. Additionally, AAMYO had the tendency to have a higher transcriptional activity than AAV8 in all cell lines. Taken together, we successfully designed and optimized a vector construct encoding for the TRES6 antibody. Despite an effective protection of both vectors in a SARS-CoV-2 infection model, the myocyte directed transduction by AAVMYO seems to be superior due to higher TRES6 serum concentrations, a favorable glycosylation profile and beneficial Fcγ receptor binding after i.v. vector injection compared to AAV8.