Deficiency of DNA double-strand break repair in human preimplantation embryos revealed by CRISPR-Cas9

There has been much interest in the application of CRISPR-Cas9 to human preimplantation embryos for research and clinical purposes. However, understanding of the way in which the double strand DNA breaks (DSB) produced by CRISPR-Cas9 are repaired during this critical developmental phase is limited, and technical challenges have led to controversy over the interpretation of data obtained from the few embryos analyzed thus far. Here we describe development of a novel pipeline for single cell genetic analysis, permitting long-read sequencing, coupled with targeted deep sequencing based on sequence capture, and parallel molecular karyotyping. This strategy allowed systematic evaluation of all DSB repair outcomes following CRISPR-Cas9 editing of human preimplantation embryos, revealing that DSBs are produced with high efficiency and confirming that loss of heterozygosity at targeted sites is common. Contrary to previous hypotheses, we demonstrate that loss of heterozygosity is not a consequence of interhomolog homologous recombination, but instead due to a high frequency of failed or aberrant repair, resulting in large-scale genomic rearrangements. The inappropriate DSB resolution observed is consistent with a sensitivity to DNA damage in cells of human embryos during the earliest stages of development. Time-lapse analysis provided evidence that persistence of DSBs may be due in part to cell cycle checkpoints, required for maintenance of genomic stability, being unusually permissive, allowing mitotic progression in the presence of unresolved DNA damage. We speculated that this preimplantation DNA repair deficiency is transient, and unlikely to extend beyond day-3 of development, when the embryonic genome is activated. To test this hypothesis, we undertook the first application of CRISPR-Cas9 to human preimplantation embryos after activation of the genome. Importantly, aberrant repair was never observed at this stage and 100% frequency of homology directed repair was observed. Altogether, our pipeline allows for novel insights into DNA repair outcomes in single cells, providing results that caution against the clinical use of CRISPR-Cas9 at the time of fertilization, and suggest that application at later stages of preimplantation development may hold promise for achieving editing outcomes of the type required for correction of pathogenic mutations, while minimizing risk of unwanted genomic rearrangements.