Delineation of the Role of the Mre11 Complex in Class Switch Recombination*

Class switch recombination (CSR) is a region-specific, transcriptionally regulated, nonhomologous recombina-tional process that is initiated by activation-induced cytidine deaminase (AID). The initial lesions in the switch (S) regions are processed and resolved, leading to a recombination of the two S regions involved. The mechanism involved in the repair and ligation of the broken DNA ends is however still unclear. Here, we describe that switching is less efficient in cells from patients with Mre11 deficiency (Ataxia-Telangiectasia-like disorder, ATLD) and, more importantly, that the switch recombination junctions resulting from the in vivo switching events are aberrant. There was a trend toward an increased usage of microhomology ( > 4 bp) at the switch junctions in both ATLD and Nijmegen breakage syndrome (NBS) patients. However, the DNA ends were not joined as “perfectly” as those from Ataxia-Tel-angiectasia (A-T) patients and 1–2 bp mutations or insertions were often observed. In switch junctions from ATLD patients, there were fewer

The repair of DNA double-strand breaks (DSB) 1 is crucial for the maintenance of genome stability and defects in the cellular response to DSBs have been linked to a number of inherited human cancer-prone syndromes (1,2). There are two general types of repair: homologous recombination (HR) and nonhomologous end-joining (NHEJ). The former includes gene conversion, break-induced replication, and single-strand annealing. The latter, which is also composed of several pathways, requires a set of proteins, including the DNA end-binding proteins Ku70 and Ku80, DNA ligase IV and its associated XRCC4 protein, and possibly, the Mre11-Rad50-Xrs2 complex.
Programmed changes in the genomic structure are essential for development of the immune system. During the early stages of T and B lymphocyte differentiation, V(D)J recombination takes place in order to assemble variable (V) exons of the T cell receptor and immunoglobulin genes respectively, giving rise to a large repertoire of lymphocytes. The V(D)J recombination is a site-specific event and is initiated by the lymphoid-specific RAG1 and RAG2 proteins (3,4). After activation of the B cells, their Ig genes undergo two types of DNA modification, class switch recombination (CSR) and somatic hypermutation (SHM), which further diversify the immune response. In CSR, the constant (C) region encoding gene of the heavy chain is replaced by a downstream CH gene, resulting in a change from IgM to IgG, IgE, or IgA production, without changing the specificity of the antibody. SHM, on the other hand, leads to accumulation of mutations in the V genes and, when coupled with selection, results in an increased affinity for the antigen.
Recent studies have shown that in parallel to the crucial role of RAG proteins in V(D)J recombination, both CSR and SHM are initiated by a single differentiation-specific factor, activation-induced cytidine deaminase (AID) (5,6), probably by its deamination of dC residues in the immunoglobulin locus (7)(8)(9)(10). Depending on which way the initial dU/dG mismatch is resolved, it will result in introduction of mutations in the V region genes (SHM) or recombination of the two switch (S) regions (CSR). These later stages, which include repair and ligation of the broken DNA ends, are still not well understood. However, like V(D)J recombination, some factors that are required in the general DSB repair process might be involved in CSR and/or SHM. These include Ku70 (11), Ku80 (12), DNAP-Kcs (13), mismatch repair enzymes (14 -16), ␥-H2AX (17,18), ATM (19,20) and error-prone DNA polymerases (21)(22)(23)(24).
The Mre11 complex is a multisubunit nuclease composed of Mre11, Rad50, and Nbs1(p95)/Xrs2 (Nbs1 is the mammalian equivalent of Xrs2 in yeast). This complex is required for telomere maintenance, cell cycle checkpoint signaling, meiotic recombination and efficient repair of DSBs by HR and/or NHEJ in lower vertebrates (25,26), whereas its involvement in DNA repair in higher vertebrates remains controversial (27,28).
Another important issue, which remains unresolved, is the role of this complex in NHEJ during gene rearrangement of the immunoglobulin locus. In human, no mutations in Rad50 have been observed to date. However, mutations in genes encoding Nbs1 and Mre11 results in two related disorders, Nijmegen breakage syndrome (NBS) (1,29) and Ataxia-Telangiectasialike disorder (ATLD) (30), where both exhibit features that are characteristic for Ataxia-Telangiectasia (A-T). Biochemical studies have also provided a functional link between the Mre11 complex and ATM (31)(32)(33)(34)(35), suggesting that these factors may be required in a common pathway(s) in DNA repair. We have previously shown that in cells from A-T patients, the switch recombination junctions are aberrant, with a significantly increased usage of microhomologies at the junctions and an altered pattern of SHM-like mutations in the S region, where mutations are mainly due to transitions and biased to A or T nucleotides. The frequency and pattern of mutations in the V region is however largely normal, suggesting a role for ATM in CSR, but not in SHM (19,20). Nbs1 and ␥-H2AX have been shown to form nuclear foci at the CH region in cells undergoing CSR (17) and analysis of switch junctions from cells from NBS patients also suggests a potential role of Nbs1 in CSR (19). We therefore performed a detailed analysis of breakpoints, resulting from in vivo switching from to ␣ in cells from ATLD patients. We also studied SHM-like mutations in the S regions in both ATLD and NBS patients. These data may help delineate the DSB repair pathways in CSR and SHM.

Patient Material and Analysis of Serum Immunoglobulin Levels-
The study included 4 of 6 ATLD patients identified in the world to date. The clinical details have been described previously (30,36,37). ATLD1 and 2 are first cousins and part of a large family. ATLD3 and 4 were from another family, with nonconsanguineous parents. The mutations in Mre11 in these patients (30,38) are shown in Table I. The serum levels of immunoglobulin classes and IgG subclasses were measured using nephelometry. The genetic characterization and immunoglobulin levels in the A-T and NBS patients have been described previously (19,39,40). Patients with "idiopathic" IgA deficiency (IgAD) have also been described previously (19).
Amplification of S-S␣ and S-S␥ Fragments and the Germline S Region-Genomic DNA was purified from peripheral blood cells from patients and healthy blood donors with normal serum levels of immunoglobulins. The amplification of S-S␣ fragments from in vivo switched cells was performed as described before (19,41). Briefly, two pairs of S-and S␣-specific primers were used in a nested PCR assay. The number of S-S␣ fragments was determined from 10 reactions run in parallel using DNA from each individual and represents random amplification of in vivo switched clones. Patient samples were always run at the same time as the control samples. In controls, only strong and distinct bands were counted as the weak bands might be due to nonspecific amplification. In patients, the amplification normally resulted in fewer and weaker bands, and all the distinct bands were counted. A human IgA1-producing cell line (313) was used to assess the fidelity of the nested PCR reaction and the PCR error rate was 0.9/1,000 nucleotides (41). The S-S␥ fragments were amplified as described previously (42,43).
A long PCR kit (Expand TM Long Template PCR System kit, Roche Applied Science) was employed to amplify the full-length germline S region. This system utilizes an enzyme mixture containing TaqDNA polymerase and a proofreading Pwo DNA polymerase. The PCR error rate was estimated to be 0.2/1,000 nucleotides (20). S-specific primers Hu-SLs (5Ј-GGGGACCTGCTCATTTTTATCACA) and Hu-SLas (5Ј-GAGGACCCGCAGGACAAAAGAGAA) were chosen from the region flanking the S repetitive sequences. Amplification was first performed in 10 cycles, where each cycle consisted of 92°C 10 s, 64°C 30 s, and 68°C 4 min, and subsequently in 20 cycles, with the same conditions except that the elongation time was extended for 10 s for each new cycle. PCR products (4.4 kb) were gel purified and subsequently used for direct sequencing. The sequencing primer S5 is located 235 bp downstream of the Hu-SLs primer. The first 500 bp of the sequence was analyzed, covering the region where most of the S-S␣ breakpoints were located.
Analysis of the S-S␣ Junctions-The PCR amplified S-S␣ fragments were gel purified, cloned and sequenced as described previously (19). The S-S␣ breakpoints were determined by aligning the S fragment sequences with both the S (44) and S␣1 (45) or S␣2 (41) sequences. Perfectly matched short homology was defined as successive nucleotides that were shared by both the S and S␣ regions at the S junction (without mismatches). The term imperfect repeat was used when mismatches were allowed on either side of the breakpoint. Microhomology includes both the perfectly matched sequence homology and imperfect repeat. Insertion was defined as a nucleotide at the breakpoints that was not identical to either of the S regions. Mutation was defined as a nucleotide change within 15 bp on either side of the S junction. Polymorphisms in the switch regions were excluded from the mutation analysis.

Reduced Frequency of Switching in Cells from ATLD Patients-
The serum IgM, IgA, IgG, and IgG subclass levels were measured in the four ATLD patients, and no major abnormality was found (Table I and data not shown). However, the level of IgG3 was low in ATLD3 and 4 (0.17 g/liter and Ͻ0.05 g/liter, respectively), and the level of IgG2 was in the lower, agematched, normal control range in ATLD1 and 2. The level of IgG4 was below the detection limit in both ATLD1 and 2.
Genomic DNA was purified from peripheral blood cells from ATLD patients and healthy blood donors (n ϭ 17, where 7 were run in parallel with the ATLD patient samples and 10 were run in parallel with the A-T and NBS patient samples). The DNA samples were subjected to nested PCR in order to amplify in vivo generated -␣ switch fragments. The numbers of -␣ switch fragments was determined from ten PCR reactions run in parallel using DNA from each individual and can be used as an estimation of the in vivo switch frequency (41). Due to the polyclonal rearrangement at the S/S␣ region, different sizes of switch fragments are amplified and visualized. In controls, single or multiple bands were visualized and the intensity of the bands were usually rather strong, whereas in patients, switch fragments could be amplified only in some of the lanes and some of the bands were rather weak. A typical run is shown in Fig. 1A. At least 11 distinct S-S␣ fragments were amplified from the control, whereas only few fragments were generated from the ATLD patients (1, 5, 2, and 3, respectively, for ATLD1, -2, -3, and -4). The experiments were repeated independently for additional three times in order to confirm the above finding and also to obtain a sufficient number of switch junctions for further analysis. As shown in Table I and Fig. 1B, the average numbers of -␣ switch fragments generated from each ATLD patient from four experiments ranged from 4.0 to 5.3 with an average of 4.4, which is significantly lower than the corresponding numbers in controls (7 to 15, average 11.8) (p Ͻ 0.001, Mann-Whitney test). In the NBS patients, the corresponding numbers were 0 -8, with an average of 1.8 ( Fig. 1B and Ref. 19). Thus, the number of clones switching to IgA appears to be reduced in ATLD patients, although not as markedly as in the NBS patients. As the level of IgG3 was lower in ATLD3 and 4, we also estimated the level of switching to IgG3 in all four patients. The number of S-S␥3 fragments in ATLD1 was comparatively normal (10 versus 12.7 in controls), however the numbers in ATLD2 and -3 were lower than the controls (3 and 6, respectively). We could only amplify one S-S␥3 fragment from ATLD4 in ten PCR reactions, which correlated to the very low level of IgG3 in this patient. Based on the sequences of the S-S␥3 fragments from these patients, ATLD3 and 4 carry the G3m(g) allotype markers in their S␥3 regions, which is known to be associated with a reduced frequency of switching to IgG3 (42, 46).
S-S␣ Recombination Junctions in ATLD Patients-We subsequently cloned and sequenced 48 switch fragments (47 S-S␣ and 1 S-S␥-S␣) from the ATLD patients and 39 switch fragments (all S-S␣) from normal controls, generated in the above PCR reactions. All the switch fragment sequences were unique and therefore represent independent switch recombination events. The sequences of all the S-S␣ junctions are available at www.biosci.ki.se/users/qipa/switch-junctions2.pdf. The switch junctions from the 7 controls included are similar to those from our previous experiments (19) in all respects analyzed. The new control data (n ϭ 39) were therefore summarized together with the previous data (n ϭ 115) and used for comparison with the different patient groups. The junctional sequences from the previously investigated A-T and NBS patients were also re-analyzed to ensure that the same standard was applied in all patient groups.
Switching to ␣1 occurred more often than to ␣2 both in ATLD patients and normal controls. However, the percentage of ␣1 switching was slightly lower in ATLD patients (57 versus 66%). Nine percent of S-S␣ junctions from ATLD patients showed additional intra-switch region recombination in the S sequences. This number was higher than the NBS (7%) and A-T patients (2%) and was lower than the controls (14%); however, there was no significant difference between the respective patient groups and controls. Similar frequency of intra-S␣ region recombination was found in S-S␣ fragments from ATLD, NBS patients and controls (13,13, and 11%, respectively), however a much lower rate was again observed in S-S␣ fragments from the A-T patients (2%). Thus, ATM, but not the Mre11 complex, is probably involved in CSR-induced internal recombination within the S and S␣ regions.
The distribution of S-S␣ breakpoints in ATLD patients is shifted toward the more 3Ј part of S and the 5Ј-part of S␣1 or S␣2 (Fig. 2); this may reflect a shift toward increased donor/ acceptor homology. Indeed, significantly more S breakpoints from ATLD patients (57 versus 33% in controls) were located in the part of S region that shows the highest degree of homology with S␣1 or S␣2 (position 275-760 in S; position 275 was marked with a vertical line in Fig. 2) ( 2 test, p Ͻ 0.001). Breakpoints from NBS and A-T patients showed a similar shift toward this part of S region ( Fig. 2; 2 test, p Ͻ 0.01 and p Ͻ 0.001, respectively). The distribution of the S-S␣ breakpoints therefore suggests an increased dependence on donor/acceptor homology in the switch junctions from all three groups of patients; however, the actual usage of microhomology still depends on the local sequences at the switch junctions.
We next analyzed the microhomology usage at the switch junctions. There was a clear trend toward an increased usage of perfectly-matched short sequence homologies (Ն4 bp) in both FIG. 1. Amplification of S-S␣ breakpoints. A, typical runs for control and ATLD patients. The numbers of -␣ switch fragments was determined from ten PCR reactions run in parallel (lanes 1-10) using DNA from each individual. Eleven switch fragments were amplified from the control (lane 9 has two distinct bands), whereas only 1 (ATLD1, lane 6), 5 (ATLD2, lanes 4 -6 and 8), 2 (ATLD3, lanes 1 and 6), and 3 (ATLD4, lanes 3 and 5) switch fragments were amplified from the four patients. M, molecular weight marker. B, estimation of switch frequency by counting the number of switch fragments in controls and patients. In controls (n ϭ 17), A-T (n ϭ 12), and NBS (n ϭ 9) patients, the numbers of S-S␣ fragments obtained from ten PCR reactions from each individual were plotted. In ATLD patients (n ϭ 4), the average numbers of switch fragments from four independent experiments (total number of switch fragments in 40 PCR reactions divided by 4) were plotted.
NBS and ATLD patients, although not to a statistically significant degree (Table II and Fig. 3A). However, the DNA ends with sequence homologies from these patients were not joined as "perfectly" as those from A-T patients and 1-2-bp mutations or insertions were often observed at the switch junctions. Thus, when more relaxed criteria of microhomology were used, i.e. allowing one mismatch at either side of the switch junction, there were actually significantly more S-S␣ junctions flanked by long imperfect repeats in both NBS and ATLD patients as compared with controls (Table II and Fig. 3B). The proportion of switch junctions that used different sizes of perfectly matched microhomologies from different patient groups are given in Fig. 3A. In addition to the trend for an increased usage of microhomology of Ն 4bp in ATLD and NBS patients, there was also a trend toward decreased usage of microhomology of 1-3 bp in both ATLD and NBS patients. The details of using imperfect repeats are presented in Fig. 3B. The data clearly show that the order for the dependence of microhomology at the switch junction is A-TϾNBSϾATLDϾcontrol.
Mutations at the S-S␣ Junctions from ATLD and NBS Patients-Lack of mutations or insertions was one of the characteristics of the S-S␣ junctions from A-T patients (19). On the contrary, mutations or insertions occurred frequently at, or close to, the S-S␣ junctions (Ϯ15 bp) derived from ATLD patients, with a frequency of 29.8/1,000 bp, which is significantly higher than those from A-T patients ( 2 test, p Ͻ 0.001), but similar to those from NBS patients (25.9/1,000 bp) and controls (22.9/1,000 bp). Notably, the pattern of nucleotide substitutions was markedly different between the ATLD patients and controls, with less base substitutions due to transitions (Table III). The most striking difference was that transition at C nucleotides (C 3 T), which occurred most often in the controls, was never observed in the ATLD patient junctions ( 2 test, p Ͻ 0.01). The pattern of nucleotide substitution in NBS patients was different from both ATLD patients and controls, with 100% G/C mutations and 57% transitions; however this difference did not reach a statistical significance due to the low number of switch fragments we could generate.

Mutations in the S Region (Upstream of the Switch Junctions) in ATLD and NBS Patients-SHM-like mutations have
been previously identified in the S region (up to several hundred base pairs upstream of the S-S␣ breakpoints) in normal in vivo switched human B cells (20). With the new set of controls (n ϭ 39), we could confirm our previous finding that mutation patterns in the S region (upstream of the breakpoints) are clearly different from those at, or close to, the switch breakpoints. The latter showed a strong bias toward mutation of G/C nucleotides (83 versus 63% for upstream mutations; 2 test, p Ͻ 0.01) and transversions were slightly favored (55% versus 41% for upstream mutations; 2 test, p Ͼ 0.05). Furthermore, they occurred at a much higher frequency (22.9/1,000 bp versus 6.5/1,000 bp, Tables III and IV; 2 test, p Ͻ 0.001)). These data suggest that different mechanisms might be involved in generating these mutations. Those at, or close to, the switch breakpoints are mainly generated during the final repair step(s) in CSR, whereas those away from the breakpoints may be occurring at an earlier step during CSR (17), which may share some of the molecular pathways/components with SHM.
In A-T patients, SHM-like mutations were observed at lower frequency (3.0/1,000 bp) (20) as compared with those from controls (6.5/1,000 bp, 2 test, p Ͻ 0.01). Furthermore, the general pattern of base substitutions was different from that in controls, with more occurring at A/T sites (56 versus 34%; 2 test, p Ͻ 0.05) and a strong preference for transitions 86 versus 58%; 2 test, p Ͻ 0.01) (20). In ATLD and NBS patients, mutations were observed at a frequency of 4.1/1,000 bp or 5.4/1,000 bp respectively, which is slightly lower than those from controls, but not to a significant degree. The mutation patterns in the S region (15-bp upstream of the breakpoints) are different from those at, or close to, the switch breakpoints in both ATLD and NBS patients (Tables III and IV). Mutations were often associated with the previously described hotspot consensus for SHM, the RGYW/WRCY (r ϭ A or G, Y ϭ C or T, W ϭ A or T) motif (47), in both patients (77 and 59% for ATLD and NBS patients, respectively) and controls (69%). The order for the frequency of mutations occurring at these motifs were ATLDϾcontrolϾA-TϾNBS. In contrast to the results from A-T patients, the pattern for base pair substitutions from ATLD and NBS patients were largely similar to those from the controls, except that G 3 A substitutions occurred more often in ATLD patients than the controls ( 2 test, p Ͻ 0.05) ( Table IV). The Mre11 complex may therefore mainly be involved in generating mutations close to, or at, the switch junctions whereas ATM may also influence the process of generation of mutations in the S region, away from the switch breakpoints. DISCUSSION In yeast, the Mre11 complex is one of the most critical biochemical components involved in DNA DSBs and repair (26,27,48). In vertebrates however, the exact function of this complex in this process remains elusive. One of the difficulties encountered is that disruption of Mre11, Rad5 , and Nbs1 in mice results in embryonic lethality (49 -51). In human cells, mutations of Mre11 and Nbs1 genes are responsible for the rare radiation-sensitivity disorders ATLD and NBS, respectively, demonstrating that the complex plays an important role in the response of human cells to DSBs. However, no gross deficiency of DNA repair has been found in cells from these patients (30,52). As the mutations in these patients are most likely hypomorphic, it is difficult to delineate the potential role of the Mre11 complex in DNA repair in humans. In this study, we analyzed CSR, a process that also utilizes the NHEJ pathway, in B cells from ATLD patients, to explore the potential role of the Mre11 complex in DNA repair.
The serum concentrations of IgA and IgG were largely normal in the ATLD patients. The serum level of IgA or IgG is however not a sensitive monitor of the efficiency of CSR. Skewing by antigen selection and the long half-life of plasma cells allows accumulation of serum IgG even if the switching process itself is substantially compromised (e.g. uracil-DNA glycosylase (UNG)-deficient or MSH2-deficient mice) (9,15,16). Indeed, using our previously described PCR strategy (41), we could show that the number of S-S␣ clones were significantly reduced in ATLD patients as compared with controls, suggesting a reduced level of switching to IgA in B cells from these patients. Similarly, switching to IgG3 was also reduced in ATLD2, 3 and 4, although ATLD3 and 4 may have additional defects in IgG3 switching as they carried the G3m(g) allotype markers in their S␥3 regions. In addition, ATLD1 and 2 also had extremely low levels of anti-Pneumococcal antibodies after immunization, 2 which is most likely due to a less efficient switching to the IgG2 subclass in these two patients. The CSR 2 M. R. Taylor, unpublished data. process is therefore not abolished in ATLD patients, but shows a reduced efficiency. The effects of the Mre11 complex on CSR was further supported by the altered pattern of in vivo recombination at the switch junctions, where an increased usage of perfectly matched short homology or imperfect repeat was noted in both ATLD and NBS patients. Furthermore, the mutation pattern close to, or at the switch junctions, was also different between patients and controls. These data strongly suggest that the Mre11 complex is involved, either directly or indirectly, in the final steps of switch recombination, which include DNA end modification, repair, and joining. The reduced efficiency of CSR and altered pattern of switch recombination junctions could possibly be explained by a number of the known functions of the Mre11 complex. First, this complex has DNA binding and bridging activities. Structural and biochemical studies have shown that the complex binds to both single-stranded (ss) and double-stranded (ds) DNA (53,54), and the role of the complex in NHEJ may thus be to tether the DNA ends together, facilitating the binding of other DNA repair factors (55,56). In the absence of Mre11, donor-acceptor homology (Ն4 bp) is probably needed to stabilize the DNA ends and to allow an efficient repair process. Second, this complex also displays a variety of enzymatic activities and Mre11 shows a 3Ј 3 5Ј exonuclease activity on (ds) DNA as well as an endonuclease activity on (ss) DNA and hairpin structures (27,(57)(58)(59). The lack of any of the above enzymatic activities may theoretically reduce the efficiency of end joining and might also be responsible for the altered mutation pattern at the switch junctions. Third, the Mre11 complex may interact with factors that are involved in end joining in CSR. In yeast, the Mre11 complex interacts directly with the Dnl4/Lif1 (the yeast counterpart of mammalian DNA ligase IV/XRCC4) complex and promotes its DNA ligation activity (55). The role of DNA ligase IV in CSR is still unclear. However, our preliminary data suggest that the switch junctions are also altered in cells from DNA ligase IV-deficient patients, in a similar way as those from A-T patients. 3 The absence of the Mre11 complex may thus reduce the efficiency of the ATM and DNA ligase IV dependent repair pathway in CSR. Finally, although the mechanism involved is not clear, it has been suggested that the Mre11 complex also plays a role in DNA replication (27). Based on the DNA deamination model, AID initiates SHM and CSR by deaminating dC nucleotides in DNA to create a U:G mismatch (10). The mismatch is processed by a number of pathways, one of which is through replication, which does not require further modification by uracil-DNA glycosylase (UNG) or other nucleases that produce a C 3 T transition. It is thus theoretically possible that the lack of the C 3 T transition is due to impairment of replication at the DNA breaks in the absence of Mre11.
A role for the 3Ј 3 5Ј exonuclease activity of Mre11 has previously been suggested in microhomology-based end joining, where it degrades the mismatched DNA ends until sequence identity is revealed, thus stabilizing the junction at a site of microhomology (60). If this activity were essential for microhomology-based end joining, absence of Mre11 would result in a lack of sequence homologies in the junctions. However, although the general activity of switching is much lower in NBS and ATLD patients, an increased level of microhomology was observed in the examined switch junctions. One explanation could be that the mutated Mre11 and Nbs1 alleles are hypomorphic, and that some of the enzymatic activities remain. Indeed, a recent study by Paull and co-workers (61), has shown that the mutated Mre11 complex expressed from the ATLD1/2 (1897 TϾC) and ATLD3/4 (350 AϾG) alleles, still exhibit a normal level of exonuclease activity, although the initial rate and the endonuclease activity were reduced. Alternatively, there are several pathways for microhomology-based end joining in CSR, some of which do not require the Mre11 complex. We hypothesize that the dominant pathway in CSR, i.e. the error-prone end joining pathway, using non-complementary DNA ends with a short homology (1-3 bp), is probably affected by the mutated Mre11 complex and accounts for the reduced efficiency of CSR in both ATLD and NBS patients. The alternative, microhomology (Ն4 bp)-dependent, error-free end joining pathway that is frequently used in A-T patients, may also be affected by the mutated Mre11 complex. This is probably related to the strand annealing activity of the latter (53). Thus, in the absence of the Mre11 complex, rejoining of the complementary ends is not precise and small insertions are intro-3 M. Ehrenstein and Q. Pan-Hammarström, unpublished data.  duced, resulting from misalignment and filling-in of the ends. When the sequence homology exceeds 7 bp or even 10 bp, a pathway, independent of both the Mre11 complex and ATM, appears to be functional. In fact, the increased dependence of sequence homology at switch junctions in both ATLD and NBS patients were mainly due to the proportionally high number of junctions showing a microhomology or imperfect repeat of Ն7/8 bp or even Ն10/11 bp. Although the switch junctions in cells from ATLD and NBS patients share most of their characteristics, they are still different in some respects. The Nbs1 mutation results in a more severe impairment of switching and, in addition to the quantitative difference, the base substitution pattern at the switch junctions was also different between NBS and ATLD patients ( Table III). The C 3 T transition could be readily observed in the switch junctions from NBS patients but not from ATLD patients. The NBS patients included in this study all carry the 657del5 allele, which results in a 70-kDa Nbs1 protein (p70) lacking the N terminus (62). This truncated protein still forms a complex with Mre11/Rad50 in vitro and supports the exonuclease activity of Mre11, indicating that Nbs1 does not need to be fully functional within the Mre11 complex (61,62). The proteins derived from the Mre11 alleles from ATLD1/2 and ATLD3/4 could still bind to wild-type Nbs1, although the latter showed a reduced affinity (61). There was however no major difference observed at the switch junctions between the ATLD1/2 and ATLD3/4 patients, except that the rate of mutations/insertions at the switch junctions was higher in the former (35.7/1,000 bp versus 21.0/1,000 bp). Both Mre11 and Nbs1 may therefore have functional role(s) in CSR that are independent of each other.
Honjo and co-workers (63) have previously reported that murine B cells stimulated with LPS/IL-4 accumulate SHM-like point mutations in the germline (GL) S region, suggesting that a similar cellular machinery, involving AID, may initiate both SHM and CSR. Schrader et al. (64) have recently shown that the pattern of mutations in GL and recombined S segments differ; those in recombined S regions show a preferentially targets G/C base pairs and WRCY/RGYW hotspots, whereas mutations introduced into the GL S do not. They have however not analyzed the pattern of mutations close to and distant from the switch junctions separately, in the recombined S regions, probably due to the limited number of mutations analyzed (32 mutations for the wild-type mice) (64). We have previously shown that SHM-like point mutations can be observed in human recombined S region, away from (starting 15-bp upstream) the switch junctions and that the pattern of these mutations are clearly different from those at, or closed to, the switch junctions (20). In this study we confirmed our previous finding both in additional normal controls and in ATLD and NBS patients. The mutations observed in S, away from the switch junctions, show a similar spectrum as those in V and GL S regions and often occur in the predicted SHM hotspots, supporting the notion that a similar cellular machinery, involving AID, may initiate both SHM and CSR by producing lesions in the V and S regions. However, at least part of the mutations in the S region (away from the switch junctions) seem to be generated by a pathway that is not used in SHM, as deficiency of ATM (20) or DNA-PKcs (65-67) resulting altered patterns of mutations in S, but not in V regions. Furthermore, the pattern of mutations at, or close to, the switch junctions is clearly distinguished from those in the S (away from the switch junctions) and in the V regions, with a strong bias toward mutation of G/C nucleotides and furthermore, transversions are favored. The repair pathway(s) of break resolution in the final step(s) in CSR is therefore most likely different from those in SHM. DNA-PKcs (65,66), ␥-H2AX (18), ATM (20), and Mre11 2 appear to be functional only in the repair pathway(s) of CSR, whereas other factors such as mismatch repair enzymes (14,15,68,69) appear to impact both processes.
In summary, CSR utilizes several end-joining pathways and multiple protein complexes are involved in the switch recombination machinery. In this paper, we have provided the first evidence linking Mre11 to the NHEJ repair pathway(s) in CSR. These findings may shed light on the mechanism of CSR and may also help us to understand the role of this complex in DNA repair in human cells.