CRISPR-Cas9 is a promising technology for genome editing. Here we use Cas9 nuclease-induced double-strand break DNA (DSB) at the UROS locus to model and correct congenital erythropoietic porphyria. We demonstrate that homology-directed repair is rare compared with NHEJ pathway leading to on-target indels and causing unwanted dysfunctional protein. Moreover, we describe unexpected chromosomal truncations resulting from only one Cas9 nuclease-induced DSB in cell lines and primary cells by a p53-dependent mechanism. Altogether, these side effects may limit the promising perspectives of the CRISPR-Cas9 nuclease system for disease modeling and gene therapy. We show that the single nickase approach could be safer since it prevents on- and off-target indels and chromosomal truncations. These results demonstrate that the single nickase and not the nuclease approach is preferable, not only for modeling disease but also and more importantly for the safe management of future CRISPR-Cas9-mediated gene therapies.

The discovery of TATA-binding protein–related factors (TRFs) has suggested alternative mechanisms for gene-specific transcriptional regulation and raised interest in their biological functions. In contrast to recent observations of an embryonic lethal phenotype for TRF2 inactivation in Caenorhabditis elegans and Xenopus laevis , we found that Trf2 -deficient mice are viable. However, Trf2 ^–/– mice are sterile because of a severe defect in spermiogenesis. Postmeiotic round spermatids advance at most to step 7 of differentiation but fail to progress to the elongated form, and gene-specific transcription deficiencies were identified. We speculate that mammals may have evolved more specialized TRF2 functions in the testis that involve transcriptional regulation of genes essential for spermiogenesis.