Transduction Profiles of Engineered Adeno-Associated Viral Capsids in Mouse and Marmoset

Abstract

Adeno-associated viruses (AAVs) remain promising vectors for gene therapy due to stable expression in vivo and a strong clinical safety record. Unfortunately, naturally occurring AAV serotypes are inefficient transducers of many gene therapy relevant tissues, requiring high viral doses to achieve therapeutic efficacy for many disease indications. Fortunately, AAVs are amenable to engineering efforts to improve tissue tropism and specificity. Past efforts have produced AAV variants with improved transduction capabilities for clinically relevant cell populations, including crossing the blood-brain-barrier (BBB) in mice (AAV-PHP.eB) [1], but the efficacy of those variants has not translated across all strains and species. With the aim of enhancing viral tropism for refractory targets, libraries of AAV9 were selected for novel characteristics using Multiplexed Cre-recombination-based AAV targeted evolution (M-CREATE) [2]. Viral genomes from capsids that transduced tissues of interest across rodent Cre-lines are selectively amplified and recovered through the M-CREATE method, allowing simultaneous positive and negative selection of AAV variants. Systemic administration of AAV libraries through intravenous injection permits non-invasive transduction of tissues where direct administration is difficult, enabling subsequent variant selection within gene therapy relevant cell populations. We are presenting data on three novel engineered capsids: variant AAV-CAP.A4, which was identified for improved transduction in mouse lung tissue, and variants AAV-CAP.B10 and AAV-CAP.B22 that, when administered systemically, can cross the blood—brain barrier and efficiently transduce neurons in adult mice and marmosets. AAV-CAP.A4 was compared against serotypes AAV5 and AAV9 for lung tissue transduction after systemic injections of 1e11 viral capsids/animal. AAV-CAP.A4 displays a 17-fold higher total lung transduction over AAV9 and a 45-fold improvement over AAV5. In Alveolar Type II pneumocytes, the improvement in transduction over AAV9 and AAV5 reaches 29-fold and 100-fold, respectively. This enhancement in AAV-CAP.A4 transduction is specific to the lung, while the liver and other tissues targeted by AAV9 have similar transduction profiles. A panel of novel AAV9 variants was identified for similar transduction across the BBB as the strongest current neurotropic AAV, PHP.eB [1]. Promising variants were chosen for pooled screening via systemic delivery in adult marmosets, from which AAV-CAP.B10 and AAV-CAP.B22 were chosen for independent screening. Both variants display improved transduction across the brain relative to AAV9. In the marmoset cortex, AAV-CAP.B10 and AAV-CAP. B22 transduce neurons ~15 fold and ~17 fold higher than AAV9 respectively. These novel variants enable robust, non-invasive gene delivery to the adult marmoset brain following IV administration. This work demonstrates that, through the M-CREATE method, novel AAV variants can be developed with sought-after transduction profiles in clinically relevant cell populations. We have also shown that relative improvements to the transduction profile obtained through in vivo selection in mice can be translated to non-human primates.

Additional details