Summary Breakpoint junctions of the chromosomal translocations that occur in human cancers display hallmarks of nonhomologous end-joining (NHEJ). In mouse cells, translocations are suppressed by canonical NHEJ (c-NHEJ) components, which include DNA ligase IV (LIG4), and instead arise from alternative NHEJ (alt-NHEJ). Here we used designer nucleases (ZFNs, TALENs, and CRISPR/Cas9) to introduce DSBs on two chromosomes to study translocation joining mechanisms in human cells. Remarkably, translocations were altered in cells deficient for LIG4 or its interacting protein XRCC4. Translocation junctions had significantly longer deletions and more microhomology, indicative of alt-NHEJ. Thus, unlike mouse cells, translocations in human cells are generated by c-NHEJ. Human cancer translocations induced by paired Cas9 nicks also showed a dependence on c-NHEJ, despite having distinct joining characteristics. These results demonstrate an unexpected and striking species-specific difference for common genomic rearrangements associated with tumorigenesis.

Thermodynamic and kinetic parameters for the triplex-forming reactions between a homopurine-homopyrimidine 22-base-pair duplex (sequence of the purine strand: 5'd[AAAGGAGGAGAAGAAGAAAAAA]3') and the four 22-dN third strands (22 dN: 5'd[TTTCCTCCTCTNCTTCTTTTTT]3', where N = A, C, T, or G) were determined from thermal denaturation and renaturation UV absorbance profiles. Cooling and heating curves were not superimposable and thus allowed us to determine the rate constants of association (k(on)) and dissociation (k(off)) as a function of temperature, assuming a two-state model analogous to that developed for duplex-forming reactions. Experiments were performed in 10 mM cacodylate buffer (pH 6.8) in the presence of NaCl concentrations ranging from 20 to 300 mM. Within experimental accuracy, the main results are the following: (i) The rate constants k(on) and k(off) result in linear Arrhenius plots, consistent with the prediction of two-state association and dissociation (ii) k(on) is independent of the nature of the base N located in the center of the third strand. (iii) k(on) strongly decreases when the NaCl concentration is decreased. (iv) The activation energy, E(on), is always negative and becomes more negative when the NaCl concentration is decreased. (v) k(off) is independent of NaCl concentration but depends on the base N, with its magnitude following the order C greater than G greater than A much greater than T. (vi) The activation energy, E(off), is independent of the base N. All these results are discussed in the light of a nucleation-zipping model similar to that developed for the duplex-coil transitions [Craig, M. E., Crothers, D. M., & Doty, P. (1971) J. Mol. Biol. 62, 383-401; Pörschke, D., Eigen, M. (1971) J. Mol. Biol. 62, 361-381].

Genome editing has now been reported in many systems using TALEN and CRISPR-Cas9 nucleases. Precise mutations can be introduced during homology-directed repair with donor DNA carrying the wanted sequence edit, but efficiency is usually lower than for gene knockout and optimal strategies have not been extensively investigated. Here, we show that using phosphorothioate-modified oligonucleotides strongly enhances genome editing efficiency of single-stranded oligonucleotide donors in cultured cells. In addition, it provides better design flexibility, allowing insertions more than 100 bp long. Despite previous reports of phosphorothioate-modified oligonucleotide toxicity, clones of edited cells are readily isolated and targeted sequence insertions are achieved in rats and mice with very high frequency, allowing for homozygous loxP site insertion at the mouse ROSA locus in particular. Finally, when detected, imprecise knockin events exhibit indels that are asymmetrically positioned, consistent with genome editing taking place by two steps of single-strand annealing.