A New AAV Vector Can Efficiently Target the Brains of Mice Through IV Injection

DOI: https://doi.org/10.1016/j.omtm.2019.10.007

Adeno-associated virus (AAV) vectors are used extensively in gene therapy for their ability to safely deliver therapeutic genes to a wide array of cell targets. Through years of optimization, AAV has finally reached its zenith and proven effective in human trials to treat genetic diseases. Since 2018, two FDA-approved AAV-based medicines have reached the market. Many other clinical trials with AAV-based therapies are under way to treat disorders such as hemophilia, muscular dystrophy, and Parkinson’s disease.

While this progress is exciting, there are still limitations to AAV vectors. Concerns include dose-related toxicities, which have been occasionally reported. Pre-existing immune responses to AAV can also reduce the number of patients eligible to receive therapy. Finally, there are some cell types AAV is inefficient at targeting. Therefore, there has been a concerted effort by AAV vector biologists to design and characterize “better” capsids with improved efficacy at lower doses, thus mitigating many of the issues mentioned above. AAV capsid libraries have led to the identification of additional novel AAVs, by testing millions of capsid variants at once and selecting the most efficient.

In 2016 Deverman et al. reinvigorated the field by using a twist on  conventional AAV9 peptide libraries. They developed the “CREATE” system that allows for the selection of capsids able to enter the nucleus in Cre-expressing transgenic mice. This group identified an extremely efficient capsid, PHP.B, which could transduce almost the entire CNS with a single systemic injection. While the transduction profile of PHP.B in mice did not translate to non-human primates owing to differences in a receptor expressed in some mice but not in primates,  this work demonstrated to the field just how efficient transduction of CNS could be with AAV vectors.

In our study, we developed a novel library selection strategy that allowed us to select AAV-F, an AAV9 variant that can efficiently target the brain when administered intravenously. Here, we improved upon traditional library screens  by directly selecting AAV capsids that can mediate the desired end stage of transduction: transgene expression. We incorporated the recombinase Cre into the  AAV genome encoding random 7 amino acid capsid variants and injected our library into transgenic Ai9 mice. These mice possess a floxed-STOP fluorescent reporter, which will only activate in the presence of Cre. As such, we could inject our library cocktail and isolate capsid variants from fluorescent cells, that is, from cells containing capsids that can drive gene expression (as opposed to any cells positive for vector genome).

After two rounds of selection, we narrowed our library down to only three capsid variants. We tested two of these, which we called AAV-F and AAV-S. AAV-S showed expression levels in the brain that were similar to those of the parent capsid, AAV9. However, AAV-F showed expression levels up to 170-fold better than AAV9. We compared this to AAV9-PHP.B, the efficient capsid mentioned above. AAV-F expressed at levels similar to those of PHP.B. In the cortex AAV-F transduced significantly more astrocytes than PHP.B and this phenotype was reversed for neurons. AAV-F also drove high-level, localized expression when injected directly into the brain. When injected intrathecally, spinal cord was transduced at extremely high levels and widespread expression could even be seen in the brain.

Interestingly, we also observed that AAV-F is effective in another strain of mice, BALB/c. PHP.B targets some strains of mice very well, but do not effectively transduce others at all, as they are inadvertently reliant on a receptor, Ly6a, that is lacking in these mice. The effectiveness of AAV-F in BALB/c and C57BL/6 mice – the two most common strains in biological research – allows researchers to use one capsid for efficient transduction of the brain in their mouse models. It also suggests that AAV-F uses a unique mechanism from PHP.B to transduce the brain from the periphery.

For some diseases, there is currently an unmet clinical need for a vector that can target the entire brain after intravenous administration. Here we have developed a new vector, AAV-F that serves as an excellent new tool for genetically modifying the entire brain of mice. How well AAV-F translates remains unknown, with experiments in non-human primate currently underway. Selection based strategies using AAV libraries are always reliant on chance, particularly whether what emerges can cross the species barrier.

In future studies we will use the selection across multiple species to increase the chances of a universal vector. Further, with this study we have validated iTransduce, and we now intend to employ it to target cell types which may require both biophysical and transcriptional (promoter-based) targeting.

Related Articles