The protein shell (capsid) of Adeno-associated viruses (AAV) are presently the most promising delivery vehicles for various in vivo gene therapies. AAVs are non-pathogenic and, through past engineering efforts, have become safe due to their inability to integrate into and damage the genome of target cells. Rather, the delivered DNA containing a therapeutic gene of interest resides outside of the target cell’s genome where it directs expression of the gene to treat a particular dysfunction.
Different subtypes of AAV capsids can deliver DNA to a wide variety of cells and tissues, while the specific cells and tissues vary from subtype to subtype. This is due to the organization of the icosahedron-shaped capsid that forms an envelope around the DNA payload, and that interacts with specific molecules on target cells via protruding spike structures on the capsid surface. A challenge in many gene therapies is the absorption of AAV viruses to cells that are not supposed to be targeted. This can decrease virus efficiency at intended target sites and cause undesired responses in other cell and tissue types. These also include cells of the immune system that can trigger immune responses and an immune memory which later on prevents the therapeutic virus to be provided a second time, should this be necessary.
To overcome these limitations, a Wyss Institute research team led by former Staff Scientist Eric Kelsic and Core Faculty member and Harvard Medical School Professor George Church is taking a systematic unbiased synthetic biology approach to create AAV capsids with different or more selective cell type and tissue-targeting abilities. Using current DNA-synthesis technology, they generated multiple variants of the gene encoding the AAV capsid-forming protein in which all of its amino acid building blocks were individually mutated (Wide Search). Infection of the AAV variants into human cell lines and in vivo into mice led to the identification of amino acids that could be variably tuned without compromising capsid stability, while shifting target cell and tissue-specificities.
Taking the data from this Wide Search as a basis, the team is developing a massively parallel, data-driven rapid evolution approach to identify specific combinations of changes that may further increase target cell specificity and enhance the immune evasion potential of synthetic AAV variants (Deep Search). In this effort, multiple iterations of the design-build-test cycle are performed with machine-learning algorithms to leverage findings from previous cycles, thereby informing the design of a new generation of AAV viruses. These are generated, tested and analyzed in a multiplexed process involving DNA synthesis of synthetic genes for new capsid variants, infection into in vivo models, retrieval from multiple tissues, and DNA sequencing to provide the data for the next optimization step.