ABSTRACT
The ability to sequence, interpret, and make changes to the human genome has transformed 21st century
biosciences. Historically, diseases caused by genetic mutations could be at best recognized and treated, but
rarely cured. However, the rapidly developing field of genome engineering has promised permanent, curative
options for a multitude of genetic conditions such as metabolic liver diseases, epidermolysis bullosa, and sickle
cell disease. The ability to manipulate the genome has also led to better disease models, more robust control
over cellular fate, and high-resolution maps of cellular dynamics during embryonic development. Indeed, nearly
all bioscientific and biomedical fields have benefited greatly from advances in genome engineering. However,
the tools used to modify the genome are imperfect and still in development. There is room to increase editing
efficiency, decrease off-target effects, and improve on-target fidelity.
The combination of Cas9 and adeno-associated virus-6 (AAV6) has proven to be highly efficient for site-specific
genome editing. Cas9 induces a double-stranded break at a target genomic site, while AAV6 delivers single-
stranded DNA repair templates into the nucleus. Since AAV is a virus, it has evolved to deliver DNA into cell
nuclei in a manner more efficient than most other transfection protocols. The cell then employs its endogenous
homology-directed repair machinery to fix the Cas9-induced break, using the AAV6-delivered DNA as a repair
template. This approach has been used to make both small changes and large insertions in the genome of cells
in vitro and in vivo. For these reasons, AAV is the vector-of-choice in over 100 clinical trials worldwide.
We recently generated data that questions the fidelity of target-site genome modifications when using AAV6 to
deliver the repair template. Using a comprehensive and sensitive assay for detecting regions of DNA, we found
that nearly half of the edited cells had additional, unexpected genomic inserts of the template. Further analysis
revealed that these insertions are on-target and concatemeric in nature. Shockingly, the frequency of this
unintended genotype has not been reported in the literature. Common techniques researchers and clinicians
use to analyze AAV6-induced knockins would fail to detect these concatemeric insertions. However, there is
evidence in some of the publications that, unknown to the authors, supports our finding.
Unintended concatemeric insertions during targeted genome-editing that occur at such high frequencies could
have disastrous consequences. If genomic modifications are unknowingly incorrect, researchers will report
unreliable and incorrect results, while clinicians may be disrupting the genes in which they are trying to repair.
Therefore, in this proposal we aim to (1) identify the variation and extent of Cas9/AAV6-induced concatemeric
insertions in regards to cell-type and genomic location, and (2) develop strategies to prevent, attenuate, and
exploit these unintended concatemeric insertions.