We have determined the specificity profile of the homing endonuclease I-AniI and compared it to the conservation of its host gene. Homing endonucleases are encoded within intervening sequences such as group I introns. They initiate the transfer of such elements by cleaving cognate alleles lacking the intron, leading to their transfer via homologous recombination. Each structural homing endonuclease family has arrived at an appropriate balance of specificity and fidelity that avoids toxicity while maximizing target recognition and invasiveness. I-AniI recognizes a strongly conserved target sequence in a host gene encoding apocytochrome B and has fine-tuned its specificity to correlate with wobble versus nonwobble positions across that sequence and to the amount of degeneracy inherent in individual codons. The physiological target site in the host gene is not the optimal substrate for recognition and cleavage: at least one target variant identified during a screen is bound more tightly and cleaved more rapidly. This is a result of the periodic cycle of intron homing, which at any time can present nonoptimal combinations of endonuclease specificity and insertion site sequences in a biological host.

Gene targeting by homologous recombination (HR) can be induced by double-strand breaks (DSBs), however these breaks can be toxic and potentially mutagenic. We investigated the I-AniI homing endonuclease engineered to produce only nicks, and found that nicks induce HR with both plasmid and adeno-associated virus (AAV) vector templates. The rates of nick-induced HR were lower than with DSBs (24-fold lower for plasmid transfection and 4- to 6-fold lower for AAV vector infection), but they still represented a significant increase over background (240- and 30-fold, respectively). We observed severe toxicity with the I-AniI ‘cleavase’, but no evidence of toxicity with the I-AniI ‘nickase.’ Additionally, the frequency of nickase-induced mutations at the I-AniI site was at least 150-fold lower than that induced by the cleavase. These results, and the observation that the surrounding sequence context of a target site affects nick-induced HR but not DSB-induced HR, strongly argue that nicks induce HR through a different mechanism than DSBs, allowing for gene correction without the toxicity and mutagenic activity of DSBs.