During V(D)J recombination, the repair of RAG-generated DSBs by the classical NHEJ pathway contributes much of the antigen receptor diversity necessary for immune system function (6
). We have determined that chromosomal breaks generated by the RAG recombinase can be repaired by pathways other than NHEJ in mouse ES cells, although repair by alternative pathways occurs at a significantly lower frequency than NHEJ. HDR was estimated to be utilized at a ≥40-fold-lower frequency than NHEJ for both the coding end and signal end reporters. SSA was more difficult to precisely quantify, but it was estimated to occur at a comparable or lower frequency than HDR. As expected, coding joint formation was impaired in cells deficient for the classical NHEJ components Ku70, XRCC4, and DNA-PKcs. Concomitant with this, RAG-induced HDR was increased in each of the NHEJ-deficient cell lines, including cells lacking DNA-PKcs. Among the NHEJ mutants, the largest increase in HDR occurred in Ku70−/−
cells. Thus, RAG-generated DSBs are typically channeled into the NHEJ pathway and the absence of NHEJ components allows for repair by the alternative HDR pathway.
The RAG recombinase has recently been postulated to directly “shepherd” DSBs into NHEJ, since joining-deficient RAG proteins display enhanced HDR (17
). Given the low frequency of HDR in wild-type cells, our results are consistent with a role for the RAG proteins in affecting repair pathway choice. However, our results also demonstrate a striking suppression of RAG-induced HDR by NHEJ pathway components. Loss of the Ku70 protein, in particular, caused a substantial shift toward HDR, such that the level of HDR in Ku70−/−
cells approached the estimated level of coding joint formation in wild-type cells (~1% versus 3%, respectively), and as well approached the level of HDR found in wild-type cells after I-SceI cleavage (~1% versus 2%, respectively).
NHEJ pathway components also affect pathway choice in the repair of I-SceI-generated DSBs, with Ku70 having a stronger suppressive effect on I-SceI-induced HDR (fivefold) compared with other NHEJ components (35
). It has been postulated that the Ku70/80 heterodimer prevents access of DNA ends to processing factors, such as the nuclease responsible for strand resection (19
), and that the resulting increased access of ends to these factors in the Ku70
mutant leads to a greater increase in HDR compared with loss of other NHEJ factors (35
That RAG-induced HDR was also increased in DNA-PKcs−/−
cells indicates that DNA-PKcs is not essential for hairpin opening at coding ends prior to HDR. Biochemical studies have indicated that DNA-PKcs forms a complex with and phosphorylates the Artemis protein, allowing Artemis to acquire the hairpin-opening activity which is essential for V(D)J recombination (27
). Consistent with this model, hairpin coding ends accumulate in both DNA-PKcs-deficient and Artemis-deficient thymocytes (9
). A low level of coding joints is recovered from these cells (3
), however, implying the existence of alternative nicking activities, such as from the Mre11 complex. Direct analysis of hairpin metabolism in wild-type and scid
cells also supports the existence of robust alternative pathways (20
We expected that the regulation of the stability of RAG2, which is modulated by cell cycle-dependent phosphorylation, might also play a major role in restricting repair to NHEJ, by confining DSB formation to the G1
stage of the cell cycle (18
), when HDR frequency is generally considered to be low (47
). Substitution of a phosphorylation-defective RAG2 (RAG2-T490A) did not, however, affect the level of HDR (data not shown). Phosphorylation has been shown to enhance ubiquitin-dependent degradation of RAG2 by promoting the transit of the protein from the nucleus to the cytoplasm (30
). The requirement for T490 phosphorylation in ubiquitination is not absolute, however, implying that it is not the sole arbiter of RAG2 stability. Moreover, the ability of RAG-generated breaks to be efficiently channeled into the HDR pathway in Ku70−/−
cells indicates that RAG-generated breaks have the potential to access the HDR machinery such that the regulation of RAG2 protein levels does not by itself regulate repair pathway choice.
Our results demonstrate that, like I-SceI-generated DSBs (22
), RAG-generated breaks can be repaired by multiple pathways. Nevertheless, the results also suggest that the frequency of usage of the various pathways differs between the two types of breaks. In wild-type ES cells containing the reporter DR-GFP, we previously observed that I-SceI-generated chromosomal breaks are repaired by each of the three pathways (NHEJ, HDR, and SSA) at similar frequencies, with NHEJ being only a fewfold more frequent than either HDR or SSA (35
), rather than the ≥40-fold preference for NHEJ that we observed for RAG-generated DSBs. Direct comparison of the repair of I-SceI- and RAG-generated DSBs is complicated, however, because I-SceI-generated DSBs have the potential to be precisely rejoined (26
), which allows for subsequent recleavage.
RAG-induced HDR was reported by Lee and colleagues (17
), with remarkably similar levels of HDR observed with their hamster cell system as with our ES cell system. In this previous report, RAG-generated nicks rather than DSBs were postulated to be the instigators of most of the observed HDR events. The results we obtained are consistent with the induction of HDR by RAG-generated DSBs. For example, enhanced HDR was observed in the three different NHEJ mutants, each lacking proteins with distinct roles in DSB repair (24
). In addition, RAG-induced SSA was detected, which presumably reflects a DSB intermediate. Finally, we obtained a substantially higher frequency of HDR from the DRGFP-CE reporter, in which the two RSS elements are separated by a spacer, than with the DRGFP-NICE reporter, in which the two RSS elements are adjacent to each other. RAG-generated nicks would be expected to lead to the same level of HDR in the DRGFP-NICE reporter or even enhanced HDR, since substantially less nonhomologous sequence would need to be removed from the broken end prior to HDR than with the DRGFP-CE reporter. Nonetheless, we cannot rule out the possibility that a portion of the HDR events that we detect are induced by RAG-generated nicks or nicks converted to DSBs, rather than frank DSBs.
In summary, RAG-generated chromosomal breaks can be repaired by multiple DSB repair pathways, although repair by V(D)J recombination using classical NHEJ components predominates. The biological significance of the alternative repair pathways is not certain, although they may provide a backup mechanism for the repair of RAG-generated breaks under some circumstances. The use of alternative pathways highlights the potential for DNA ends to “escape” from the normal control of the classical NHEJ components, which can lead to the misrepair of DSBs, genomic rearrangements, and tumorigenesis (7