Homology-directed repair (HDR) of double-strand breaks can be used to introduce precise edits in the genome, but is thought to be an inefficient process requiring long stretches of homology between donor and target sequences. We have found that HDR of cuts made by CRISPR/Cas9 is surprisingly efficient and requires only short homology arms, even for long, gene-sized edits (i.e., whole-gene deletions and GFP insertions).

Zhao et al. (2014) reported earlier this year that single-strand DNA oligos (ssODNs) can be used to introduce base-size edits near Cas9 sites without selection. Inspired by that report, we tested whether ssODNs could also be used to introduce small insertions, as shown for TALEN sites by Lo et al., (2013). We designed ssODNs to contain a restriction site (7 bases) or protein tags (18 to 66 bases: Tetra-Cys,V5, Myc, OLLAS, and FLAG) flanked by short homology arms (50-60 bases) and screened for edits in the first generation after injection (F1), using PCR and restriction digestion as in Zhao et al., (2014). We identified edits at frequencies ranging from 0.9 to 7% of F1s. Edit frequency was affected more by sgRNA efficiency than by edit size. We also found that ssODNs that bridge two sgRNA cut sites spaced far apart can be used to generate a precise, whole-gene deletion (1.7 kb, 3.8% of F1s). (Imprecise gene-size deletions can also be generated using two spaced sgRNAs and no repair template).

Emboldened by these results, we tried the same approach to insert GFP. We used PCR fragments made by amplifying GFP (864 bases) with primers containing ~60 bases flanking the Cas9/sgRNA site, inserting the GFP right in the sgRNA site or 27 bases away. We obtained 4% and 0.9% GFP-positive F1s overall, with 15% of injected mothers giving 20-48% edited F1s on the second day after injection (“jackpot broods”). The edits can be identified directly among the F1s by PCR or, most efficiently, by direct inspection for GFP expression. In this way, it is possible to go from injection to beautiful GFP pattern in just four days!

Based on these results, we developed a systematic protocol to mutate, delete, or tag any gene using ssODNs or PCR fragments. Using this protocol, the lab has already modified 13 genes. We are currently testing whether our approach could be streamlined even further using Co-CRISPR schemes to identify jackpot broods (Kim et al., 2014; Zhang and Glotzer, 2014; Arribere et al., 2014).

Update: Paper (including protocol) is now in press at GENETICS (http://www.genetics.org/content/early/2014/09/30/genetics.114.170423).