We have utilized an engineered mouse allele of Mre11 that functions as wild type until conditionally inactivated via the Cre recombinase (Fig. 1a). Previous work has shown that conversion of conditional (Mre11^cond/−) to null (Mre11^−/−) not only causes depletion of Mre11, but of the entire MRN complex, likely due to instability of the other components⁹. In addition, we have utilized an allele containing a targeted single amino acid change (Mre11^H129N) that eliminates the endo- and exonuclease activities of Mre11 without disrupting the MRN complex, or its ability to sense DSBs and activate ATM (Fig. 1a, b)⁹. Studies of the Mre11^H129N allele revealed that the nuclease activities of Mre11 are required for DSB repair via the HR pathway⁹.

Homozygosity for the Mre11 null allele (Mre11^−/−) or nuclease deficient allele (Mre11^H129N/H129N) causes early embryonic lethality, necessitating use of Mre11^cond and tissue specific expression of the Cre recombinase⁹. CD19-Cre initiates expression in bone marrow B lymphocyte progenitors²¹. Through breeding, we generated the following experimental mice; Mre11^cond/+, Mre11^cond/−, and Mre11^cond/H129N, each containing one allele of CD19-Cre. Therefore, these mice contain B lymphocytes with the following genotypes; Mre11^−/+, Mre11^−/−, and Mre11^−/H129N. Conversion of the Mre11^cond allele to Mre11⁻ was confirmed by PCR in lymphoid organs including bone marrow, lymph node, and spleen (Fig. 1c). Western blot analyses of B cells from spleen demonstrated severe deficiency of all three MRN components in Mre11^−/− cells, whereas they remained at wild type levels in the presence of nuclease deficient Mre11 (Mre11^−/H129N) (Fig. 1d). Thus, we are able to examine the impact of complete MRN deficiency and distinguish this from the impact of an intact MRN complex lacking endo- and exonuclease activities.

Mice lacking MRN or Mre11 nuclease activities in the B lineage produced normal numbers of mature IgM+ lymphocytes in bone marrow and had spleens of normal size and cellularity (Supplementary fig. 1, and not shown). Therefore, early B cell development is not blocked by Mre11 deficiencies generated by CD19-Cre, permitting studies examining switching from IgM surface expression to other antibody isotypes.

Isotype switching was assessed by inducing CSR in B cells isolated from spleens of 6 to 12 week old mice. Exposure to IL-4 and anti-CD40 antibody for 4 days in culture stimulates switching to IgG1 (and IgE) (Fig. 2a). Control (Mre11^−/+), Mre11 nuclease deficient (Mre11^−/H129N) and MRN deficient (Mre11^−/−) cells were induced to switch, and AID protein levels were measured by western blot. Whereas AID was not detected in resting cells, it was present upon stimulation in each of the three Mre11 genotypes, confirming that stimulation was successful in each (Fig. 1d).

Switching from IgM to IgG1 was analyzed by flow cytometric determination of the class of cell surface immunoglobulin receptor in 3 to 5 mice of each genotype. Switching was induced in approximately 30% of control B cells (Mre11^−/+) (Fig. 2b). In the absence of Mre11 nuclease activities (Mre11^−/H129N), switching occurred in 14.5 % of cells, 50% less than control (Fig. 2b,c). This impact is relatively minor, but is significant (p < 0.01 by students T test). Next, we examined the impact of MRN loss (Mre11^−/−), in which the nuclease activities of Mre11 and all other functions of the complex are deficient. In this case, the impact was more severe, with an 80% reduction in switching to IgG1 (p < 0.0001) (Fig. 2b,c). The greater severity caused by MRN deficiency relative to Mre11 nuclease deficiency was statistically significant (p < .01 students two tailed T test). We also quantitated the relative amounts of IgG1 antibody in the media from the experiments described above, since the only source of this isotype is from cells that have successfully switched. Consistent with the cell surface analyses, IgG1 levels were reduced in both Mre11 mutants (p < 0.01), with complete MRN loss having a greater impact than Mre11 nuclease deficiency (p < 0.01) (Fig. 2d).

IL-4 and anti-CD40 also induce switching to IgE. Although positivity cannot be quantitated precisely due to IgE antibody binding to Fc receptors, a trend similar to that of IgG1 was observed (Supplementary fig. 2). Therefore, the roles of MRN in CSR differ from those in DSB repair by HR, in which nuclease deficiency had the same impact as MRN loss⁹.

The isotype switching defect caused by MRN loss could merely reflect reduced activation of the ATM kinase, since MRN controls ATM activation, and mice harboring knockout alleles of ATM display a CSR defect²²,²³. As expected, NBS1 deficiency causes a defect similar to that of ATM^−/−, because NBS1 acts as the primary interface between MRN and ATM²⁴,²⁵. Therefore, we directly compared isotype switching in B cells from MRN and ATM deficiencies. MRN loss clearly caused a more severe defect compared to ATM deficiency (p < 0.001), which was approximately 50% as efficient as control (Fig. 2b,c). The greater impact of MRN loss compared to ATM^−/−, along with the minor defect caused by Mre11 nuclease deficiency (which does not affect ATM activation), collectively argue that MRN plays multiple roles during CSR.

Because CD19-Cre begins expression in early B cell development, we addressed the possibility that reduced isotype switching results from an undetected defect in V(D)J recombination that does not block the development of an IgM+ population, but compromises later attempts to switch. To this end we generated mice with Mre11 genotypes described above, but containing CD21-Cre which expresses the recombinase after cells develop to the IgM+ stage²⁶. We observed an overall trend similar to that observed with CD19-Cre mediated deletion (p< 0.01 by students T test for Mre11^−/+ vs. Mre11^−/H129N, Mre11^−/+ vs. Mre11^−/− and Mre11^−/H129N vs. Mre11^−/−)(Supplementary fig. 3). Western blot analyses of splenic B cell extracts tended to show residual Mre11 in CD21-Cre mice, likely accounting for the less severe Mre11 phenotypes relative to those associated with CD19-Cre (Supplementary fig. 3 and not shown). Therefore, we pursued further experimentation using CD19-Cre.

Isotype class switching involves several rounds of cellular division in addition to the end joining recombination reaction²⁰. Therefore we determined if the quantitative isotype switching defects merely reflect reduced proliferation. We utilized the cell tracking dye CFSE to distinguish sub-populations of cultured B cells having undergone different numbers of cell divisions (Fig. 3a). First, these analyses show that the Mre11 deficiencies do slow proliferation, with MRN loss and Mre11 nuclease deficiency having a similar impact. After 4 days in culture, a higher percentage of mutant cells relative to wild type cells had undergone only 2 or 3 divisions (Fig. 3a). Conversely, a higher percentage of control cells had undergone 4 divisions. Importantly, within the populations at each cell division, the mutants showed lower percentages of cells switched to IgG1 (Fig. 3a). Therefore, the reduced isotype switching in the Mre11 mutants cannot be accounted for by proliferation defects.

Defects in the joining phase of CSR cause DSBs to persist, and a subset can manifest as chromosome breaks at the immunoglobulin heavy chain (IgH) locus on mouse chromosome 12, where CSR occurs²⁷. For example, these breaks are readily detected in the context of C-NHEJ defects such as Lig4 or XRCC4 deficiency⁷. Therefore, we determined if MRN deficiency causes such a phenotype, to directly address whether the complex has a role in end joining.

Standard karyotyping cannot distinguish chromosome breaks at IgH from those arising elsewhere on chromosome 12 caused by general genomic instability. Therefore we employed a two color FISH assay that uses two bacterial artificial chromosomes (BACs) as probes, located 5′ and 3′ of the IgH locus (Fig 3b). Chromosome breaks at IgH are identified as separation of the probes, and previous work has shown that these events are dependent on AID⁷. For each genotype, 100 to 200 total metaphases derived from three mice were examined, and the Fisher Exact Probability test was performed to determine if the percentage of breaks accumulating in the IgH locus was significantly different in the Mre11 mutants compared to control. Loss of MRN (Mre11^−/−) resulted in a significant increase in chromosome breaks (p=0.0003) while loss of Mre11 nuclease activity (Mre11^−/H129N) did not (p=0.3482). This observation strongly supports the notion that MRN is required for joining of ends, and provides further distinction between Mre11 nuclease and full MRN deficiencies.

The established end joining factors Lig4 and XRCC4 operate exclusively in C-NHEJ⁴,⁵,⁷. MRN could potentially be restricted to A-NHEJ, operate with established factors in C-NHEJ, or overlap with both pathways. To distinguish these hypotheses we cloned and sequenced CSR joins from stimulated B lymphocytes. Examination of joins from MRN deficiency, and from Mre11 nuclease deficiency, reveal that both blunt (C-NHEJ) and microhomology mediated (A-NHEJ) joins are represented (Fig. 3c, Supplementary figs. 4^-6). Chi-squared tests were performed using the Mre11 control (Mre11^+/−) percentages as the expected results. The p-value for Mre11^−/H129N joins is .764, and for Mre11^−/− joins is .215, indicating that the Mre11 deficiencies have no significant impact on the ratio of outcomes. Therefore, unlike Lig4 and XRCC4, the MRN complex appears to function in both NHEJ pathways.

H2AX is a specialized core histone which undergoes phosphorylation in megabase regions surrounding DSBs²⁸. Phosphorylated H2AX (γ-H2AX) is bound by MDC1 protein, which in turn interacts with 53BP1, facilitating localization of these proteins to DSB sites²⁹. Mouse knockouts of any member of the H2AX/MDC1/53BP1 axis cause a B lymphocyte development phenotype similar to what we observe for MRN deficiency, i.e. progression of a substantial population of cells through early stages and defective CSR in mature cells²⁷,^30-33. The similarity of these phenotypes raises the possibility that MRN could control functions of these proteins. Alternatively, MRN could provide functions at DSBs independent of γ-H2AX/MDC1/53BP1. We distinguished these hypotheses by determining the impact of MRN loss, or Mre11 nuclease deficiency, on γ-H2AX formation, and localization of MDC1 and 53BP1 to DSBs as marked by immunoflourescent foci. The induction of DSBs must be well synchronized to investigate kinetic differences in DNA damage responses, yet CSR occurs over a time span of days. To circumvent this problem, we induced DSBs with ionizing radiation (IR) in mouse embryonic fibroblasts (MEFs) harboring the same Mre11 alleles⁹. In this case, the Mre11^cond allele was converted to Mre11⁻ prior to IR by delivering the Cre recombinase via adenovirus as previously described⁹.

At 1 and 8 hours after IR, western blots were performed using an antibody specific to phosphorylated H2AX (γ-H2AX). No reduction of γ-H2AX levels were evident in the Mre11 mutants, in contrast to phosphorylated-SMC-1 levels, which are known to be dependent on MRN (Fig. 4a)⁹. Examination of immunoflourescent foci comprised of MDC1 or 53BP1 also revealed no impact of the Mre11 mutants (Fig. 4b, c). While a subtle influence on these proteins cannot be ruled out, these experiments support the notion that CSR defects conferred by Mre11 deficiencies cannot be ascribed to a defective γ-H2AX/MDC1/53BP1 response.

We crossed Mre11^cond/+, Mre11^cond/−, and Mre11^cond/H129N mice with CD19-cre or CD21-cre expressing mice (Jackson Labs). We performed bone marrow analysis on 3 - 5 week old mice, and Class Switch Recombination experiments on 6 - 12 week old mice.

We performed Mre11 genotyping as described⁹.

Cre genotyping was performed using the following PCR primers:

• cre-up1: 5′ CTAGGCCACAGAATTGAAAGATCT 3′
• cre-dn1: 5′ GTAGGTGGAAATTCTAGCATCATCC 3′
• IL-2 up: 5′ GCGGTCTGGCAGTAAAAACCTATC 3′
• IL-2 dn: 5′ GTGAAACAGCATTGCTGTCACTT 3′

Thermocycling conditions were 36 cycles of 94°C for 30sec, 51.7°C for 1min, 72°C 1min.

We performed Western blots for MRN components as described⁹. αGAPDH was used at 1:3000 (Santa Cruz). Rabbit polyclonal αAID was a generous gift from Jayanta Chaudhuri (Sloan-Kettering Institute)⁴⁹.

We enriched mature B lymphocytes from spleen using a B Lymphocyte Enrichment Kit (BD IMag) and cultured B cells in RPMI media supplemented with 10% (v/v) FBS and 1% (v/v) Pen/Strep (10,000 U ml^-1 Pen + 10,000 ug ml^-1 Strep) with or without the following cytokines: 1 μg ml^-1 αCD40 (BD Bioscience) plus 25 ng ml^-1 IL-4 (R&D Systems). For cell tracking experiments, we incubated cells in RPMI media containing 10 μM CFDA SE dye (Vybrant CFDA SE Cell Tracer Kit, Invitrogen) at 37°C for 10 min and washed with PBS prior to plating. B lymphocytes were plated at 1×10⁶ cells ml^-1 and cultured for 4 days.

We analyzed cell surface markers on a Beckman Coulter FC500 Flow Cytometer using Cytomics RXP software. Cells were washed with PBS + 10% (v/v) FBS, and incubated on ice in the dark for 30 - 60 min using various combinations of the following antibodies: αB220 (1:200, eBioscience), αIgG1 (1:100, BD Pharmingen), αIgE (1:100, Southern Biotech), αIgM (1:200, Southern Biotech), αCD25 (1:100, eBioscience). We analyzed data using FlowJo software.

We incubated B lymphocytes in colcemid (KaryoMAX) for 3 h, washed twice with PBS, and incubated in 75 mM KCl for 15 min at 37°C. We fixed cells with 3:1 methanol/acetic acid and dropped onto glass slides which were dried on a 42°C hot plate. 2-color FISH labeling was done as described using BAC 199 (aka 199M11, Invitrogen) and BAC 207 (aka RP22-207123, BAC PAC Resources)⁷. Slides were stained with DAPI (Invitrogen) according to manufacturer's instructions. We acquired images on an Olympus BX61 microscope using a 60× objective, DAPI, FITC, and Texas Red filters, a CCD camera, and FISHview software (Applied Spectral Imaging).

We coated microtiter plates with IgM and IgG1 antibodies (Southern Biotech) in a humid chamber overnight. Plates were washed 4× with PBS + 0.05% (v/v) Tween 20, blocked with PBS + 10% (v/v) FCS for 2 - 5 h, and washed 4× with PBS + 0.05% (v/v) Tween 20. We added serial dilutions (IgG1: 0 - 10^-1, and IgM: 0 - 10^-2) of stimulated B lymphocyte culture supernatants to coated wells. Concentrations of standards were as follows: IgM (0 - 0.1 μg ml^-1, Southern Biotech); IgG1 (0 - 0.25 μg ml^-1, Southern Biotech). We incubated supernatant and standards on plates in a humid chamber at RT for 2 - 5 h. Plates were washed 4× with PBS + 0.05% (v/v) Tween 20, and Alkaline Phosphate reagent (Southern Biotech) was diluted 1:4000 in ELISA Tris Buffer [17.8 g L^-1 Trizma HCl + 10.6 g L^-1 Trizma Base + 10 g L^-1 BSA + 0.2 g L^-1 Magnesium Chloride pH = 8.0], added to the microtiter plates, and incubated overnight. Plates were washed 4× with PBS + 0.05% (v/v) Tween 20. We added 100 μl of 1 mg ml^-1 phosphatase substrate in ELISA Carbonate Buffer [2.33 g L^-1 Sodium carbonate + 2.86 g L^-1 sodium bicarbonate + 0.2 g L^-1 Magnesium Chloride pH = 9.8] to each well and incubated for 30 - 60 min. Readings were taken on an EL808 machine (Biotek Instruments, Inc.) using 405 nm wavelength.

We amplified CSR joins by nested PCR using genomic DNA prepared from day 4 stimulated B lymphocyte cultures. PCR conditions have been described⁵⁰. Sμ primers for Sμ - Sγ1 and Sμ - Sε joins were: SμOut, 5′- AAG TTG AGG ATT CAG CCG AAA CTG GAG AGG -3′; and SμIn, 5′- TTC TTC CCT CTG ATT ATT GGT CTC C -3′. Sγ1 primers for Sμ - Sγ1 joins were: Sγ1Out, 5′- CTG CTC TTC TGT GGT TTT TGA CTG GGT TCC -3′; and Sγ1In, 5′-AAC TAC TAA ACT TGT ACC TGT CCT GGC ACC -3′. Sε primers for Sμ - Sε joins (Sε1 and Sε2) have been described⁷. We cloned PCR products using Topo-TA cloning kit (Invitrogen) and analyzed sequences using Seqbuilder 7.0 software (DNAstar).

For immunofluorescence, we grew MEFs on 4 chamber glass slides, and treated with 10 Gy of IR (¹³⁷Cs source). MEFs were washed with 20 mM HEPES, 50 mM NaCl, 3 mM MgCl₂, 300 mM sucrose, 1% (w/v) triton X-100 and fixed in 4% (w/v) paraformaldehyde. Cells were blocked in PBS with 10% (v/v) fetal bovine serum, incubated overnight at 4°C with polyclonal MDC1 antibody (1:100, abcam, ab41951), or 53BP1 antibody (gift of Xiaochun Yu, University of Michigan) then incubated 1 h with the secondary goat anti-rabbit AlexaFlour 555 antibody (1:1000, Molecular Probes).