|Home | About | Journals | Submit | Contact Us | Français|
Recent work has highlighted the importance of alternative, error-prone mechanisms for joining DNA double-strand breaks (DSB) in mammalian cells. These noncanonical, non-homologous end joining (NHEJ) pathways threaten genomic stability but remain poorly characterized. The RAG post-cleavage complex normally prevents V(D)J recombination-associated DSBs from accessing alternative NHEJ. Because the MRE11/RAD50/NBS1 complex localizes to RAG-mediated DSBs and possesses DNA end tethering, processing and joining activities, we asked whether it plays a role in the mechanism of alternative NHEJ, or participates in regulating access of DSBs to alternative repair pathways. We find that NBS1 is required for alternative NHEJ of hairpin coding ends, suppresses alternative NHEJ of signal ends, and promotes proper resolution of inversional recombination intermediates. These data demonstrate that the MRE11 complex functions at two distinct levels, regulating repair pathway choice (likely through enhancing the stability of DNA-end complexes) and participating in alternative NHEJ of coding ends.
DNA double strand breaks (DSBs) must be rapidly detected and repaired to preserve genomic stability. The cell’s response to DNA damage must be exquisitely sensitive, and it must lead to very specific repair: each broken DNA end must be correctly identified and rejoined with its proper partner. Despite the complexity of the DNA damage response, the mechanisms and proteins involved in the two major DSB repair pathways – homologous recombination and classical nonhomologous end joining (NHEJ)—have been well characterized over the past few decades by a combination of genetic and biochemical approaches (Weterings and Chen, 2008). This picture is complicated, however, by recent studies highlighting the activity of a poorly understood, error-prone pathway (or set of pathways), termed alternative NHEJ (Corneo et al., 2007; Soulas-Sprauel et al., 2007; Yan et al., 2007). The normal physiological function(s) of alternative NHEJ remain unknown, but it is characterized by junctions bearing frequent microhomologies, excessive deletions (Roth and Wilson, 1986), and chromosome translocations (Bennardo et al., 2008; Guirouilh-Barbat J, 2004; Guirouilh-Barbat et al., 2007; Haber, 2008; Zhu et al., 2002).
We have used a physiologic system of NHEJ-mediated genomic rearrangements to investigate the nature of alternative NHEJ and to begin to understand how this error-prone pathway is kept in check. V(D)J recombination employs site-specific DSBs generated by the products of the recombination activating genes (RAG-1 and RAG-2) to assemble antigen receptor genes in lymphocyte precursors via NHEJ (Weterings and Chen, 2008) and provides a tractable system to study the fates of physiological DSBs. The RAG complex cleaves DNA at specific sites known as recombination signal sequences (RSS), generating pairs of covalently sealed (hairpin) coding ends and blunt signal ends. The RAG proteins remain associated with these broken DNA ends in a post-cleavage complex (Agrawal and Schatz, 1997; Hiom and Gellert, 1997) and direct them specifically to the classical NHEJ machinery (Corneo et al., 2007; Lee et al., 2004), which includes Ku70/80, DNA-PKcs, Artemis, XRCC4, Cernunnos/XLF, and DNA ligase IV (Weterings and Chen, 2008). Whereas the blunt signal ends are simply joined to form precise signal joints, the hairpin coding ends must be opened and processed before ligation can take place. Hairpin opening is carried out by Artemis, a protein whose hairpin endonuclease activity is stimulated by DNA-PKcs (Goodarzi et al., 2006; Ma et al., 2002). Mice deficient in Artemis or DNA-PKcs are defective for hairpin opening and accumulate coding ends (Gao et al., 1998; Rooney et al., 2002; Roth et al., 1992). Interestingly, mice deficient in DNA-PKcs (scid, or Prkdcscid/scid mice) are also somewhat defective for signal joint formation (reviewed in (Bogue et al., 1998)).
We recently developed a RAG mutant that reveals abundant alternative NHEJ activity in wild-type and classical NHEJ-deficient cells. This RAG2 C-terminal truncation mutant, RAG2FS361, enables robust bypass of the coding and signal joint defects in DNA-PKcs-deficient and XRCC4-deficient cells (Corneo et al., 2007). These results signify the existence of alternative hairpin opening and end joining functions. Further support for the presence of one or more alternative hairpin opening nucleases is provided by the residual coding joint formation observed in Prkdcscid/scid (Carroll and Bosma, 1991; Petrini et al., 1990) and Artemis-deficient (Artemis−/−) (Li et al., 2005; Rooney et al., 2002) mice. The identity of the nuclease(s), however, remains unknown. Both RAG and Mre11 have been put forward as candidates, as they can open hairpins in vitro (Besmer et al., 1998; Paull and Gellert, 1998; Shockett and Schatz, 1999), and Mre11 has been implicated in processing hairpin DNA intermediates in yeast (Lobachev et al., 2002). The Mre11 complex is also involved in microhomology-mediated end joining in yeast (Lee and Lee, 2007; Ma et al., 2003), although it remains unclear whether this corresponds to alternative NHEJ observed in mammalian cells.
Although NBS1 localizes to rearranging T cell receptor (TCR) loci in mouse thymocytes (Chen et al., 2000), V(D)J recombination occurs with apparently normal efficiency in patients with Nijmegen breakage syndrome (NBS) or Ataxia-telangiectasia-like disease (ATLD), which are caused by hypomorphic mutations in the Nbs1 and Mre11 genes, respectively (Carney et al., 1998; Harfst et al., 2000; Stewart et al., 1999; Yeo et al., 2000). Nevertheless, in mouse models, genetic deficiencies in MRE11 complex members result in genomic instability, increased levels of TCR-β-γ trans-rearrangements and chromosome translocations involving the TCRα locus (Kang et al., 2002; Theunissen et al., 2003). Similarly, NBS patients suffer from an increased incidence of malignancies, especially lymphomas (The International Nijmegen Breakage Syndrome Study Group, 2000). These data suggest that although the MRE11 complex is not required for V(D)J recombination, it might play a regulatory role, preventing aberrant handling of broken DNA ends through an undefined mechanism.
Could the MRE11 complex also function more directly in alternative NHEJ? Biochemical analyses have been suggestive: in addition to hairpin-opening and nuclease activities, the MRE11 complex has DNA-end binding, DNA-end tethering, and DNA-unwinding functions, and it is important for many aspects of chromosome break metabolism (de Jager et al., 2001; Hopfner et al., 2002; Hopfner et al., 2001; Stracker et al., 2004). Indeed, based on its biochemical and structural properties, the MRE11 complex has been proposed to direct DNA joining at microhomologies (Paull and Gellert, 2000; Zhang and Paull, 2005).
We considered several potential roles for the MRE11 complex in executing or regulating alternative NHEJ of V(D)J recombination intermediates, including 1) providing hairpin opening nuclease activity, 2) providing end tethering and/or end processing activities critical for alternative NHEJ, and 3) suppressing alternative NHEJ via an undefined regulatory activity. To test these hypotheses, we employed a variety of strategies to examine alternative NHEJ of both coding and signal ends in a variety of defined genetic settings: in wild-type cells, in the absence of functional DNA-PKcs (Prkdcscid/scid)(Bosma et al., 1983), in the absence of Artemis (Artemis−/−) (Rooney et al., 2002), and in the presence of a hypomorphic mutant form of NBS1 (Nbs1ΔB/ΔB) (Williams et al., 2002) alone or in combination with the NHEJ mutants. In these studies, we examined the effects of both wild-type RAG1/2 and the RAG2FS361 mutant.
We find that the NBS1 ΔB mutant ablates joining of coding ends by alternative NHEJ in a variety of experimental situations, including endogenous T cell receptor gene rearrangements in Artemis-deficient mice. The NBS1 ΔB mutant does not, however, diminish signal joint formation via alternative NHEJ, indicating that NBS1 is not required for the ligation step. These data localize the role of NBS1 in alternative coding joint formation to reactions specific for coding ends, such as hairpin opening or subsequent end processing steps. We also find that the NBS1 ΔB mutant promotes alternative NHEJ of signal ends and facilitates formation of aberrant reaction products involving joining of coding ends to signal ends (hybrid joints). Thus, the MRE11 complex functions at two distinct levels in V(D)J recombination, participating in alternative NHEJ of coding ends (perhaps by assisting hairpin opening and/or tethering coding ends) and suppressing alternative signal joint and hybrid joint formation (likely by stabilizing DNA end complexes).
Homozygous Nbs1ΔB/ΔB cells and mice are viable (Williams et al., 2002), and like cells from NBS patients (Carney et al., 1998), Nbs1ΔB/ΔB cells are unable to form irradiation-induced MRE11 and NBS1 foci, and the MRE11 complex is mislocalized in unirradiated cells (Williams et al., 2002). We tested the effect of the Nbs1ΔB mutant on alternative NHEJ in a variety of cell lines deficient for classical NHEJ, NHEJ-proficient cells, and Artemis-deficient mice, in the presence of either wild-type RAG proteins or the RAG2FS361 mutant.
We first tested the ability of RAG2FS361 to bypass the severe coding joint defects in Prkdcscid/scid and Artemis−/− mouse fibroblasts (Blunt et al., 1995; Rooney et al., 2002) using transiently transfected substrates that report recombination in mammalian cells as drug-resistant colonies in a bacterial transformation assay (Hesse et al., 1987). As expected, recombination catalyzed by wild-type, full-length RAG proteins in Prkdcscid/scid and in Artemis−/− cells was identical to background levels observed with a catalytically inactive RAG1DDE mutant (Landree et al., 1999) (Fig. 1, Supplementary Table 1). RAG2FS361 substantially rescued coding joint formation in DNA-PKcs-deficient cells, up to almost 50% of levels formed by wild-type RAG proteins in wild-type mouse fibroblasts (Fig. 1A, Supplementary Table 1, experiments I–III; Supplementary Fig. 1A,B) [P=0.006; Student’s two-tailed t-test], in agreement with our previous results (Corneo et al., 2007). These results were confirmed by PCR analysis of DNA from transfected cells (Supplementary Fig. 1A,B). Similar results were obtained in Artemis−/− mouse fibroblasts: again, RAG2FS361 rescued coding joint formation to ~40% of levels observed with wild-type RAG proteins in wild-type fibroblasts (Fig. 1B, Supplementary Table 1, experiments IV–VI) [P=0.049, Student’s two-tailed t-test].
We next analyzed the nucleotide sequences of coding joints. As expected, numerous coding joints obtained from DNA-PKcs- and Artemis-deficient cells transfected with RAG2FS361 showed excessive deletions (average deletion in wild-type cells expressing wild-type RAG proteins is 5.2 bp versus 11 bp in DNA-PKcs- or Artemis- deficient cells expressing RAG2FS361, P<0.05, Student’s two-tailed t-test), many with microhomologies (average length of microhomologies in wild-type cells expressing wild-type RAG proteins is 0.6 versus 1.7 in DNA-PKcs or Artemis deficient cells expressing RAG2FS361, P<0.05, Student’s two-tailed t-test) (Supplementary Fig. 2A,B; Supplementary Fig. 3). A large proportion of junctions (43%) exhibited excessively long P-nucleotides (palindromic sequences derived from hairpin opening, which are generally less than 3 bp (Lafaille et al., 1989)) (average P-nucleotide length in wild-type cells expressing wild-type RAG proteins is 0.2 versus 2.1 nucleotides in DNA-PKcs or Artemis deficient cells expressing RAG2FS361, P<0.0005, Student’s two-tailed t-test) (Supplementary Fig. 2C; Supplementary Fig. 3), indicative of abnormal hairpin opening. RAG2FS361 thus clearly reveals the existence of robust alternative pathways for joining hairpin ends in the absence of DNA-PKcs or Artemis.
To test the effect of the Nbs1ΔB mutation on the ability of RAG2FS361 to permit alternative NHEJ, we used Prkdcscid/scid Nbs1ΔB/ΔB and Artemis−/− Nbs1ΔB/ΔB double mutant fibroblasts. Concurrently, we tested wild-type RAG2 and RAG2FS361 in Nbs1ΔB/ΔB fibroblasts. In agreement with previous results (Williams et al., 2002), Nbs1 hypomorphism alone had little effect on coding joint formation (Fig. 1, Suppl Table 1; Supplementary Fig. 1A,B). In Artemis- and DNA-PKcs-deficient cells, however, Nbs1 hypomorphism abolished RAG2FS361-mediated coding joint formation (Fig. 1, Supplementary Table 1; P=0.06 and P=0.001 respectively, Student’s two-tailed t-test, Supplementary Fig. 1A,B). These results indicate that RAG2FS361-mediated bypass of the defects conferred by Artemis- or DNA-PKcs-deficiency requires functional NBS1.
To determine whether NBS1 is required for alternative NHEJ of coding ends in the context of wild-type RAG proteins, we took advantage of the fact that NHEJ-deficient animals, including Prkdcscid/scid and Artemis−/− mice, form coding joints at the T cell receptor (TCR) δ locus (Bogue et al., 1998; Bogue et al., 1997; Carroll and Bosma, 1991; Rooney et al., 2002). Therefore, we analyzed TCR Dδ2-Jδ1 rearrangements in thymocytes from Artemis−/− and Artemis−/− Nbs1ΔB/ΔB mice. (This analysis was not conducted in Prkdcscid/scid Nbs1ΔB/ΔB mice because of the very low birth rate of double mutant pups (Stracker et al., 2009)).
As expected, TCR Dδ2-Jδ1 coding joints were readily detected in Artemis−/− thymocytes, reaching ~60% of wild-type levels (Fig. 2A–C; Supplementary Fig. 4). NBS1 hypomorphism dramatically reduced coding joint formation: Artemis−/− Nbs1ΔB/ΔB thymocytes formed only low levels of Dδ2–Jδ1 coding joints (9%; significantly below levels detected in Artemis−/−; P= 2×10−5, Student’s two-tailed t-test), and Artemis−/− Nbs1ΔB/+ thymocytes formed intermediate levels of coding joints (47%) (Fig. 2. B,C). The Nbs1ΔB/ΔB mutation alone did not affect TCR Dδ2-Jδ1 rearrangements (data not shown). As expected, signal joints were readily detectable in Artemis−/−, Artemis−/− Nbs1ΔB/+ and Artemis−/− Nbs1ΔB/ΔB thymocytes and cell lines (Supplementary Table 2, experiments V–VII; Fig. 2D). These data show that NBS1 is required for coding joint formation by alternative NHEJ in vivo in the context of wild-type RAG1/2.
We have established that, in the context of NHEJ deficiency, alternative NHEJ requires NBS1— whether with the RAG2FS361 mutant or wild-type RAG2. But because alternative NHEJ could reflect an adaptation in cells deficient for classical NHEJ, it is important to examine the role of NBS1 in cells proficient for classical NHEJ. We therefore employed a substrate specifically designed to measure alternative NHEJ of coding ends (Corneo et al., 2007). This substrate encodes a functional green fluorescent protein only upon formation of a unique coding joint sequence within a 9-nucleotide microhomology, necessitating aberrant deletion of 20 nucleotides from the coding ends (Fig. 3A) (Corneo et al., 2007). Consistent with previous work (Corneo et al., 2007), only RAG2FS361 allowed robust alternative NHEJ with this substrate, in wild-type, Artemis−/, and Prkdcscid/scid fibroblasts (Fig. 3B,C). Coding joint formation was completely dependent upon NBS1, as shown in the double mutants (Prkdcscid/scid Nbs1ΔB/ΔB and Artemis−/− Nbs1ΔB/ΔB) and, importantly, in Nbs1ΔB/ΔB cells which are proficient for classical NHEJ (Fig. 3B,C). Thus, NBS1 hypomorphism prevents RAG2FS361-mediated alternative end joining in both NHEJ-deficient and NHEJ-proficient cells. Together, these results demonstrate that NBS1, and by extension, a functional MRE11 complex, is necessary for alternative NHEJ-mediated coding joint formation. Although recent work shows that the nuclease activities of MRE11 are not required for coding or signal joint formation during V(D)J recombination (Buis et al., 2008), we considered the possibility that they could be responsible for alternative hairpin opening. We therefore asked whether the frequency of coding joint formation mediated by alternative NHEJ is diminished in the presence of a MRE11 nuclease-deficient mutant, MRE11H129N (Bressan et al., 1998; Buis et al., 2008). We used Mre11Cond/H129N mouse embryonic fibroblasts bearing a nuclease-deficient allele and a floxed wild-type allele because Mre11H129/H129N causes early embryonic lethality (Buis et al., 2008)). We infected immortalized Mre11Cond/H129N mouse embryonic fibroblast lines with MSCV-CRE-IRES-Thy1.1 retroviral vector, which allows simultaneous expression of the cre recombinase and surface expression of a Thy1.1 marker (Supplementary Fig. 5A,B). Flow cytometry for Thy1.1 indicated that >95% of the Mre11Cond/H129N cells had been infected (Supplementary Fig. 5A). One passage after infection, we observed almost complete deletion of the floxed wild-type Mre11 allele, along with appearance of the deletion product (Supplementary Fig. 5B) but we detected no decrease in RAG2FS361-mediated alternative NHEJ in multiple experiments (Supplementary Fig. 5C). These data suggest that MRE11 does not provide a significant source of alternative hairpin opening activity. Supporting this interpretation, hairpin accumulation is not detected in mice harboring a Rad50 allele deficient for hairpin opening in S. cerevisiae (Bender et al., 2002). However, the possibility remains open that minimal wild-type MRE11 might provide residual nuclease activity in some cells.
We next asked whether the NBS1 mutation might also affect signal joint formation, a “simple” blunt end ligation reaction. We tested the ability of wild-type, Nbs1ΔB/ΔB, Prkdcscid/scid, and double mutant mouse fibroblasts to undergo V(D)J recombination using transiently transfected substrates designed to test signal joint formation, which is impaired in DNA-PKcs-deficient cells (Fig. 4, Supplementary Table 2, experiments I–VI) (Bogue et al., 1998). In agreement with our previous results (Corneo et al. 2007), RAG2FS361 fully rescued signal joint formation in Prkdcscid/scid cells (Fig. 4, Supplementary Table 2, experiments I–VI). Interestingly, Nbs1ΔB/ΔB and Prkdcscid/scid Nbs1ΔB/ΔB cells showed robust signal joint formation, demonstrating that the MRE11 complex is not required for joining blunt signal ends (Fig. 4, Supplementary Table 2; Supplementary Fig. 1A,C). These results are consistent with plasmid-based and chromosomal end joining analyses in these cells, which revealed no defects in end joining (Stracker et al., 2009). Notably, signal joint formation in Prkdcscid/scid Nbs1ΔB/ΔB cells was rescued to wild-type levels by the Nbs1ΔB/ΔB mutation in conjunction with wild-type RAG1/2, an effect that we observed consistently in 6 experiments (Fig. 4, Supplementary Table 2, experiments I–VI; P=0.04, Student’s two-tailed t-test). This observation suggests that NBS1 helps to prevent alternative NHEJ of signal joints.
If the absence of functional NBS1 permits alternative NHEJ of signal ends, we would expect the structures of the resulting signal joints to bear hallmarks of alternative end joining. Indeed, whereas >95% of signal joints recovered wild-type cells transfected with either wild-type RAG proteins or RAG2FS361 contained minimally deleted signal ends (<10bp, with less than 10% microhomologies) (Supplementary Fig. 6), more than 25% of signal joints obtained from Nbs1ΔB/ΔB and Prkdcscid/scid cells harbored deletions of >10 bp, and more than 20% exhibited microhomologies (Supplementary Fig. 6). These data are consistent with previous work reporting aberrant signal joint formation in DNA-PKcs-deficient or NBS1-deficient cells (Bogue et al., 1998; Donahue et al., 2007). The situation was even more striking in the double mutant cells: 50% of signal joints from Prkdcscid/scid Nbs1ΔB/ΔB cells showed deletions of >10 bp, most with microhomologies (Supplementary Fig. 6), suggesting that the Nbs1ΔB mutation allows signal joint formation to access alternative NHEJ pathways.
The experiments described above show that whereas a functional MRE11 complex is required for alternative NHEJ of coding ends, it also suppresses alternative joining of signal ends. These apparently paradoxical findings led us to hypothesize that the MRE11 complex might act at two distinct points: in the alternative NHEJ pathway itself and also upstream, at the level of the RAG post-cleavage complex, perhaps stabilizing DNA ends through its end tethering activity. Differences in the effect of the Nbs1ΔB/ΔB mutation on the processing of coding and signal ends could reflect the fact that coding and signal ends are handled differently by the V(D)J recombination machinery: RAG proteins bind avidly to signal end pairs, forming stable signal end complexes after cleavage, but much more poorly to the coding ends (Agrawal and Schatz, 1997; Hiom and Gellert, 1998). Furthermore, processing and joining of the coding and signal ends requires different DNA repair factors (Roth, 2003).
With these considerations in mind, we entertained the possibility that NBS1 might help to stabilize the coding ends, signal ends, or both in the post-cleavage complex, as has been observed in the case of ATM (Bredemeyer et al., 2006).
To test this hypothesis, we studied inversional recombination, a reaction which requires coordination of all four broken DNA ends in the post-cleavage complex to form a signal joint and a coding joint on the same DNA molecule. We transduced wild-type, Nbs1ΔB/ΔB, Prkdcscid/scid and Prkdcscid/scid Nbs1ΔB/ΔB SV40-immortalized mouse embryonic fibroblasts with the pMX-RSS-GFP/IRES-hCD4 (pMX-INV) retroviral inversional recombination substrate (Fig. 5A, upper panel) (Bredemeyer et al., 2006; Liang et al., 2002). pMX-INV has a single pair of RSSs that flank an anti-sense green fluorescent protein (GFP) cDNA and mediate recombination by inversion. To monitor transduction efficiency, we examined cell lines by flow cytometry for the retrovirally-encoded hCD4 and also by Southern blot hybridization (Supplementary Fig. 7). Flow cytometry and PCR analyses of RAG-transfected pMX-INV cell lines showed that, as expected, RAG2FS361 rescued the V(D)J recombination defects in Prkdcscid/scid cells bearing the chromosomally integrated substrate (Fig. 5B,C). Also, DNA-PKcs-deficiency combined with NBS1 hypomorphism (Prkdcscid/scid Nbs1ΔB/ΔB) abrogated RAG2FS361-mediated inversional V(D)J recombination (Fig. 5B,C), consistent with the failure of RAG2FS361 to rescue coding joint formation by alternative NHEJ in Prkdcscid/scid Nbs1ΔB/ΔB cells with extrachromosomal substrates (Fig. 1).
Interestingly, both wild-type RAG- and RAG2FS361-mediated inversional rearrangements in Nbs1ΔB/ΔB cells were significantly reduced (to 20% of wild-type levels; Fig. 5B, P=0.02 and P=0.03 respectively, Student’s two-tailed t-test), as shown by flow cytometry (Fig. 5B) and confirmed by PCR analysis (Fig. 5C). NBS1 hypomorphism thus reduces levels of inversion, even in the context of wild-type RAG proteins in NHEJ-proficient cells. These data suggest that this effect is not due to defects in hairpin opening or end joining per se, but rather reflects difficulty in coordinating the four-ended inversion event.
We reasoned that the deficit in inversions observed in the Nbs1ΔB/ΔB mutants might be accompanied by a corresponding increase in two-ended deletion events formed by joining of a coding end to a signal end, forming a so-called hybrid joint (Lewis et al., 1988), which provides a means for restoring chromosomal integrity when four-ended joining events cannot be accomplished (Fig. 5A, lower panel). Sleckman and colleagues found that ATM deficiency diminishes the occurrence of inversions and increases levels of hybrid joints, leading them to conclude that ATM stabilizes the post-cleavage complex (Bredemeyer et al., 2006). In agreement with our prediction, PCR analysis showed a clear increase in hybrid joint formation from the pMX-INV substrate in Nbs1ΔB/ΔB fibroblasts compared with wild-type cells (Fig. 5C). We confirmed these results using an extrachromosomal inversional substrate, pJH299 (Hesse et al., 1987) (Supplementary Fig. 8A). Our results confirmed that Nbs1ΔB/ΔB reduces levels of inversional recombination and increases levels of hybrid joint formation by wild-type RAG proteins or RAG2FS361 (Supplementary Fig. 8B).
To quantify more precisely the relative proportions of inversions and hybrid joints in wild-type and Nbs1ΔB/ΔB cells, we transformed the recovered extrachromosomal DNA into bacterial cells and selected individual chloramphenicol-resistant colonies, which result from either inversional recombination or deletional hybrid joint formation (Supplementary Fig. 8A) (Lewis et al., 1988; Sekiguchi et al., 2001). We then determined whether each individual recombination product was an inversion or a hybrid joint by PCR analysis (Supplementary Fig. 8A). This approach allowed us to determine the ratio of inversions to hybrid joints. In comparison with wild-type cells, Nbs1ΔB/ΔB mutant cells expressing either wild-type RAG proteins or RAG2FS361 show a >5-fold increase in hybrid joint formation (p<0.0005; Fisher’s exact test) (Fig. 5D). We conclude that NBS1 hypomorphism both decreases inversional recombination and increases hybrid joint formation during attempted inversion events.
To extend these observations to lymphocytes expressing wild-type RAG proteins, we utilized the pMX-RSS-GFP/IRES-hCD4 (pMX-INV) retroviral recombination vector. The proviral LTR drives the production of a bicistronic transcript, allowing simultaneous assessment of transduction efficiency (surface expression of hCD4 marker) and recombination (inversion of the RSS-GFP cassette with GFP expression) by flow cytometry (Fig. 5A). We used these reporter viruses to infect either adult CD4−CD8− double negative (DN) thymocytes (Fig. 6A, upper panel) or total fetal thymocytes (Fig. 6A, lower panel) from wild-type, Nbs1ΔB/+ and Nbs1ΔB/ΔB mice and calculated a RAG-INV activity index as the percentage of GFP+ cells divided by the percentage of total hCD4+ cells (Fig. 6A). Nbs1ΔB/ΔB cells displayed a statistically significant (2- to 3-fold; P<0.01, Student’s two-tailed t-test) reduction in inversional recombination relative to wild-type and Nbs1ΔB/+ cells (Fig. 6B). Similar results were obtained in both adult DN thymocytes and unfractionated fetal thymocytes (Fig. 6A). PCR analysis also revealed increased hybrid joint formation in Nbs1ΔB/ΔB thymocytes compared with wild-type and Nbs1ΔB/+ thymocytes (Fig. 6C), consistent with our results (shown above) obtained in fibroblasts expressing exogenous RAG proteins.
Finally, we examined hybrid joint formation at endogenous antigen receptor loci in lymphocytes. In agreement with previous results (Bredemeyer et al., 2006), we observed increased hybrid joint formation at the Igκ locus (Vκ6-23 to Jκ1) in Atm−/− lymphocytes (Figure 6E). We also observed an increase in Vκ6-23 to Jκ1 hybrid joints in Nbs1ΔB/ΔB splenic B cells, as compared to their wild-type and Nbs1ΔB/+ counterparts (Fig. 6D,E). Increased hybrid joint formation and decreased inversional recombination are thus general features in Nbs1ΔB/ΔB B and T cells. Together, these results indicate a role for NBS1, and by extension the MRE11 complex, in the handling of DNA intermediates during V(D)J recombination.
Our data demonstrate that NBS1 is required for alternative NHEJ-mediated coding, but not signal, joint formation. We considered three potential explanations for this finding. First, the nuclease activity of the MRE11 complex could be responsible for alternative hairpin opening activity, which could be mislocalized in the presence of the Nbs1ΔB mutation. Although this fits with biochemical data (Paull and Gellert, 1998) and with experiments showing that Mre11 is important for processing hairpins in yeast (Lobachev et al., 2002), the failure of the Mre11 nuclease-deficient mutant to diminish alternative NHEJ of coding ends argues against this interpretation. Second, other nucleases associated with the MRE11 complex, such as CtIP (Sartori et al., 2007), could be responsible for alternative hairpin opening, with a functional MRE11 complex required for their recruitment to the broken ends. Third, the MRE11 complex could facilitate joining through its end-bridging activity (de Jager et al., 2001; Wiltzius et al., 2005). This hypothesis is consistent with the other effects of the Nbs1ΔB mutation on V(D)J recombination, as discussed below.
The second role for NBS1 revealed by our results is in preventing alternative NHEJ of signal ends. We found that the Nbs1ΔB/ΔB mutation substantially increased signal joint formation in Prkdcscid/scid cells. Furthermore, the rescued signal joints showed a high incidence of aberrant deletions and microhomologies, demonstrating that joining occurred via alternative NHEJ. Finally, we observed aberrant signal joints (bearing deletions and microhomologies) even in NHEJ-proficient cells bearing the Nbs1ΔB/ΔB mutation, revealing an increased availability of the signal ends to alternative NHEJ. One potential explanation for these results is that NBS1, and the MRE11 complex, plays a role in stabilizing the signal end complex prior to joining, preventing the signal ends from accessing alternative NHEJ pathways. This model, which could reflect end-tethering activities of the MRE11 complex, fits with the effects of NBS1 mutation on inversional recombination, as described below. Given the modest effects of the DNA-PKcs mutation on signal joint formation, it would be informative to study signal joint formation in NBS1 mutant cells bearing mutations in XRCC4, DNA ligase IV, or Ku; unfortunately, such double mutants are not yet available.
The third role we find for NBS1 in V(D)J joining is in coordinating four-ended inversional recombination events. We observed significant decreases in inversional recombination in Nbs1ΔB/ΔB cells, with corresponding increases in hybrid joint formation, using both extrachromosomal and integrated substrates. These data reflect altered handling of the coding and signal ends, and suggest that the four ended complexes required for inversional recombination are much more likely to “lose” DNA ends in the absence of functional NBS1, giving rise to aberrant, two-ended joining events (hybrid joints). Similar results have been obtained by Sleckman and colleagues in ATM-deficient cells (Bredemeyer et al., 2006) and, while this manuscript was being prepared, in NBS1 and MRE11 mutants (Helmink et al., 2009), leading these workers to conclude that ATM and the MRE11 complex stabilize the DNA end complex during V(D)J recombination.
Taken together, these data firmly implicate the MRE11 complex in the handling of V(D)J recombination intermediates. We suggest a molecular hypothesis to explain how the complex might fulfill a surveillance role in end joining, preventing aberrant joining reactions by stabilizing a complex containing multiple broken DNA ends. Such complexes could serve to promote inversions, discourage deletional hybrid joint formation (by promoting retention of the intervening DNA segment), and suppress accessibility of the broken DNA ends to alternative NHEJ pathways. This hypothesis is consistent with analysis of mice bearing mutations in members of the MRE11 complex, which show increased levels of TCR-β-γ trans-rearrangements and chromosome translocations involving the TCRα locus (Kang et al., 2002; Theunissen et al., 2003).
Sleckman and colleagues have suggested that, directly or through the activation of downstream targets, ATM helps stabilize coding ends and promotes inversional recombination, suppressing deletional hybrid joint formation (Bredemeyer et al., 2006). Our data and recent work from Sleckman’s group (Helmink et al., 2009) have now implicated NBS1 in stabilizing such complexes, and we have shown that NBS1 discourages error-prone alternative NHEJ of signal ends. Given that ATM is activated by RAG-induced DSBs in the MRE11 and NBS1 mutants (Helmink et al., 2009), it appears that the MRE11 complex plays a direct role in stabilizing the DNA end complex, rather than indirectly by activating ATM.
How might we reconcile the paradoxical effects of the MRE11 complex, which on one hand promotes inversional recombination, discourages deletional hybrid joint formation, promotes joining of coding ends by classical NHEJ (Helmink et al., 2009), and suppresses alternative NHEJ of signal ends, but on the other hand is essential for alternative NHEJ of hairpin coding ends? We suggest that the former effects reflect the end-tethering activity of the MRE11 complex, whereas the latter observation may reveal a noncatalytic role for the MRE11 complex in hairpin opening, perhaps recruiting alternative hairpin opening nucleases such as CTiP. Alternatively, end-tethering could be especially important in alternative NHEJ of coding ends, given their weak association with the post-cleavage complex (Agrawal and Schatz, 1997; Hiom and Gellert, 1998). Although abundance evidence indicates that signal and coding ends are normally processed along different pathways (Ramsden and Gellert, 1995; Roth, 2003), we have not yet defined the precise requirements for alternative NHEJ of these two kinds of ends.
We do not yet know whether the MRE11 complex stabilizes the DNA ends in the context of the RAG post-cleavage complex, or after release of ends from the RAG post-cleavage complex. In the latter case, the MRE11 complex could continue to “shepherd” the ends into appropriate joining pathways or encourage “escaped” ends to re-join the RAG post-cleavage complex. Both scenarios are supported by the striking parallels between the behavior of certain RAG mutants and the NBS1 mutant studied here. First, formation of aberrant signal joints through alternative NHEJ pathways is stimulated by a variety of RAG mutants (Corneo et al., 2007; Lee et al., 2004; Talukder et al., 2004) and also by NBS1 mutants (this work) (Donahue et al., 2007). Second, the stability of the RAG post-cleavage complex is diminished by NBS1 mutations (this work; (Helmink et al., 2009) and by RAG2 mutations that increase alternative NHEJ (R. Wendland and D.B.R., unpublished observations). It is therefore tantalizing to speculate that the MRE11 complex and the RAG proteins collaborate in some way to tightly restrict joining of RAG-mediated DSBs to classical NHEJ. Mutations affecting NBS1 clearly contribute to genomic instability and lymphomagenesis in humans (The International Nijmegen Breakage Syndrome Study Group, 2000), and may do so, at least in part, through the effects described here.
Artemis−/− (Rooney et al., 2002), Nbs1ΔB/ΔB (Rooney et al., 2002; Williams et al., 2002), Rag2−/− (Taconic), and wild-type mice were maintained on a mixed 129/SvEv and C57BL/6 background and analyzed at 2–3 months of age.
DNA from single-cell thymocyte suspensions (Roth et al., 1992) was amplified by semi-quantitative PCR for the analysis of TCRδ recombination products as described (Bogue et al., 1996; Zhu et al., 1996). DNA from single-cell splenocyte suspensions was assayed for Vκ6-23 hybrid joints by sequential nested PCR amplification (Bredemeyer et al., 2006). See also Supplementary Methods.
WT, NBS1ΔB/ΔB, Prkdcscid/scid and Prkdcscid/scid Nbs1ΔB/ΔB murine embryonic fibroblasts and WT, NBS1ΔB/ΔB, Artemis−/− and Artemis−/− Nbs1ΔB/ΔB ear fibroblasts were generated as described (Theunissen and Petrini, 2006). Mre11Δ/+ and Mre11Cond/H129N murine embryonic fibroblasts were generated as described (Buis et al., 2008). Cre recombinase retroviruses were generated by calcium phosphate transfection of MSCV-Cre-Thy1.1 vector into Phoenix ecotropic packaging cell line (ATCC). 5×105 murine embryonic fibroblasts were seeded in 10cm plates and transduced with viral supernatant and 5μg/ml polybrene three times at 12-hour intervals. Bulk MEF populations were characterized by flow cytometry after incubation with phytoerythrin-conjugated anti-Thy1.1. PCR based genotyping following Cre recombinase introduction was assayed as described (Buis et al., 2008). Cell lines were maintained in DMEM/10% fetal bovine serum (FBS)/L-Glutamine (Gibco) and split at 1:5.
FACS analysis employed a BD LSRII flow cytometer (BD Biosciences) equipped with FacsDiVa and FlowJo as described (Corneo et al., 2007).
We thank F. Alt for Artemis−/− mice, D. Ferguson and J. Buis for the generous gift of Mre11 cell lines and protocols, M. Schlissel and B. Sleckman for the pMX-INV substrate, T. Egawa and D. Littman for the mscv-cre-thy1.1 vector, and G. Celli for technical assistance, discussions, and comments on the manuscript. J. P. was funded by grants from the NIH and the Joel and Jean Smilow Initiative; T. S. was supported by a National Research Service Award from the NIH and is a Leukemia and Lymphoma Society Special Fellow. D.R. was supported by NIH grants and the Irene Diamond Fund. L. D. is a Fellow of The Leukemia and Lymphoma Society.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.