Homology Based Capsid Shuffling

Introduction

Our main goal was to produce a library of Adeno-associated viruses that infect cells with different specificities and efficiencies, due to differing capsid composition, and select from this library the best AAVs that are highly specific. Since it is not the case that viruses that work well in vitro will work as good in vivo, we had the aim of testing our candidates in mice as well. To achieve this, the capsid genes from AAV serotypes 1,2,5,6,8 and 9 were shuffled between each other in a primerless PCR reaction that relies on the close homology between the different cap genes. Those capsid genes were first digested with DNase I, then pooled together and allowed to anneal to each other. A second PCR was done then to amplify the shuffled cap gene fragments that were generated in the first PCR while introducing AscI and PacI restriction sites, which are used to clone the cap genes into a helper vector. AAVs that were then produced in HEK293 cells, which were transfected with the cap-gene-ITR construct and another Adeno helper construct, made up the viral library that was subjected to selection pressure in target cells. In this manner, only AAVs have the most fitting caspsids will successfully infect target cells and produce more viral particles with the help of Adeno virus. By taking the viruses that survive one selection round and applying them to the next, we would end up with a group of highly efficient and specific viruses that overcome many of the issues with the wild-type AAVs. In parallel, we picked 48 randomly selected clones from the AAV library, packaged them with a YFP construct and tested them using flow cytometry on HepG2, Huh-7 and mouse primary hepatocytes. One clone of those was able to infect both HepG2 and Huh-7 cells, and was thus packaged with a luciferase construct and injected into mice. Mice experiment showed a specific targeting into the liver.

It has been shown that for example viruses isolated after selection on a breast cancer cell line (MDA-MB231) are able to more efficiently transducer breast cancer cells than other AAV2-based virus hybrids (Koerber et al., 2008).