RAG-1 and ATM coordinate monoallelic recombination and nuclear positioning of immunoglobulin loci.

Coordinated recombination of homologous antigen receptor loci is thought to be important for allelic exclusion. Here we show that homologous immunoglobulin alleles pair in a stage-specific way that mirrors the recombination patterns of these loci. The frequency of homologous immunoglobulin pairing was much lower in the absence of the RAG-1-RAG-2 recombinase and was restored in Rag1-/- developing B cells with a transgene expressing a RAG-1 active-site mutant that supported DNA binding but not cleavage. The introduction of DNA breaks on one immunoglobulin allele induced ATM-dependent repositioning of the other allele to pericentromeric heterochromatin. ATM activated by the cleaved allele acts in trans on the uncleaved allele to prevent biallelic recombination and chromosome breaks or translocations.

INTRODUCTION

Immunoglobulin (Ig) and T cell receptor (Tcr) loci are assembled through the rearrangement of variable (V), diversity (D), and joining (J) gene segments. V(D)J recombination involves synapsis of compatible conserved recombination signal sequences (RSSs) that flank the V, D and J gene segments of these genes1-3. The lymphocyte-specific V(D)J recombinase enzyme consisting of RAG1 [http://www.signaling-gateway.org/molecule/query?afcsid=A002009] and RAG2 [http://www.signaling-gateway.org/molecule/query?afcsid=A002010] mediates synapse formation and introduces DNA double-strand breaks (DSBs) between the two participating gene segments and their flanking RSSs. The RAG recombinase (unless specifically referring to the individual proteins the recombinase will hereafter be referred to as RAG) additionally stabilize recombination intermediates by holding the broken DNA ends in place and shuttling them into the non-homologous end-joining (NHEJ) DSB repair pathway4,5. In addition, the Ataxia Telangiectasia mutated (ATM) [http://www.signaling-gateway.org/molecule/query?afcsid=A000349] repair-checkpoint protein stabilizes post-synaptic cleavage complexes and coordinates DSB repair with the G1-S cell cycle checkpoint to suppress chromosomal translocations and lymphomas that arise from aberrant V(D)J recombination events6-10.

Ig recombination takes place during B cell development in the bone marrow with each developmental stage delineated by distinct recombination events. Temporally ordered rearrangement of the immunoglobulin heavy chain (Igh) locus initiates with DH-to-JH recombination in pre-pro-B cells prior to commencement of VH-to-DJH rearrangement in pro-B cells. At the pro-B cell stage of development, up-regulation of the B cell specific transcription factor, Pax5, induces locus contraction on both alleles11-14 enabling VH-to-DJH rearrangement of mid and distal VH gene families. The mechanisms that coordinate the individual Ig recombination steps have not been elucidated. Based upon the rearrangement status of Igh alleles in Abelson-murine leukemia virus-transformed developing B cells and mature B cell hybridomas15 it has been postulated that both Igh alleles undergo concurrent DH-to-JH rearrangement within a single cell and then one allele undergoes VH-to-DJH rearrangement. However, to date there are few studies that have investigated the targeting of individual Igh alleles during V(D)J recombination in primary cells and, thus, no direct evidence to support simultaneous biallelic DH-to-JH rearrangement.

Allelic exclusion is initiated, established and maintained during B cell development and enforces clonality and monospecific recognition by the B cell receptor expressed on individual B lymphocytes. Allelic exclusion of the Igh locus is initiated at the pro-B cell stage of development as only one allele undergoes functional rearrangement. This implies a mechanism for ensuring that both alleles do not undergo simultaneous VH-to-DJH recombination, but to date none has been identified. Establishment of allelic exclusion at the pro-B to pre-B cell transition inhibits VH-to-DJH recombination of the second DJH-rearranged allele during Igk light chain rearrangement. In pre-B cells, rearrangement at the Igk locus is preceded by biallelic locus contraction and, in contrast to the Igh, one Igk allele is positioned at pericentromeric heterochromatin prior to the onset of rearrangement13,16. This process is thought to contribute to allelic exclusion by reducing accessibility to one allele during recombination17. It is not known what mechanism regulates the differential treatment of Igk alleles.

We have investigated potential mechanisms by which V(D)J recombination is coordinated on the two alleles of the Igh and Igk loci during B cell development. Here we show that the V(D)J recombinase and the cellular DNA damage response machinery coordinate DNA cleavage and nuclear positioning of Ig loci to initiate allelic exclusion and preserve genomic integrity during antigen receptor rearrangement.

RESULTS

Homologous pairing of Igh and Igk alleles

It is known that pairing of X chromosomes has an important role in X inactivation during early development of female mammals18-20. To determine whether ‘pairing’ might have a role in regulating Ig loci, we analyzed the frequency of paired homologous Ig alleles in sorted bone marrow primary B lymphocytes of different developmental stages (Supplementary Fig. 1 online) and in mouse embryonic fibroblasts (MEFs). For these experiments, we performed two-color three-dimensional (3-D) DNA fluorescent in situ hybridization (FISH) experiments and subsequent confocal microscopy analysis as described previously13 (Fig. 1a—d and Supplementary Tables 1,2 online). The Igh DNA BAC probes used, CT7-526A21 and CT7-34H6, map to the distal VH gene region located at the 5′ end of the Igh locus and the 3′ constant (CH) region, respectively (Fig. 1b). The Igk BAC DNA probes used, RP23-101G13 and RP24-387E13, map to the distal Vκ24 gene region located at the 5′ end of the Igk locus and the 3′ constant (Cκ) region, respectively (Fig. 1d). Homologous pairing of Ig alleles was scored if at least one end of the homologous Igh and Igk alleles were separated by less than 1 μm. At the Igh locus we observed homologous pairing of the two alleles in 24% of CD19- pre-pro-B cells and in 21% of CD19+ pro-B cells. (Fig. 1a,b and Supplementary Table 1). Similarly at the Igk locus we observed interallelic pairing in 18–20% of pre-B cells, early immature B cells and late immature B cells (Fig. 1c,d and Supplementary Table 2). Pairing of homologous Igh and Igk alleles was seen in 8–10% of B cells of developmental stages where recombination does not occur but where the locus is being transcribed. This result is significantly different to the low frequency of interallelic pairing (2%) we observed for both Igh and Igk in MEFs where the loci are transcriptionally inactive (Fig. 1a,b,c and Supplementary Table 1,2). These data indicate that pairing of homologous Igh and Igk alleles occurs in a lineage and developmental stage specific manner that mirrors recombination of these loci

Figure 1

Homologous pairing of Igh and Igk alleles occurs during recombination.

(a) Graph indicating the frequency of inter-allelic Igh pairing. 3-D DNA FISH analysis was carried out on ex vivo sorted cells of the indicated lineage and developmental stage. Pre-pro-B cells were sorted as lineage marker negative (Lin-), B220+, CD19-, cKit+, CD25-, IgM-. Pro-B cells were sorted as CD19+, cKit+, CD25-, IgM-, pre-B cells as cKit-, CD19+, CD25+, IgM-, early immature B cells as CD19+, IgMlo, IgD- and late immature B cells as CD19+, IgMhi, IgD-. (Supplementary Fig.1 and Methods online). Murine embryonic fibroblasts (MEFs) were obtained from E13.5-15.5 embryos and cultured for four days in vitro. DNA probes were generated from two bacterial artificial chromosomes, BACs CT7-526A21 (red signal) and CT7-34H6 (green signal), which map to the distal V gene region located at the 5′ end of the Igh locus and the 3′ constant (C) region, respectively. Pairing was determined by measuring the distance separating the two alleles. A cut-off separation of 1μm was used to define pairing. Data are representative of more than three independent experiments. The rearrangement status of the cells analyzed is indicated below each bar of the graph. Selected statistical significances according to the χ2 test are shown between biologically relevant pairs. See Supplementary Table 1 for complete statistical results. (b) Confocal sections representative of paired and separated Igh alleles at the indicated stage of development. Decontracted Igh alleles are shown in pre-pro-B cells and contracted Igh alleles are shown in pro-B cells. A schematic representation of the position of probes used for detecting the different regions of Igh in the 3-D DNA FISH analyses is shown. (c) Graph indicating the frequency of inter-allelic Igk pairing. 3-D DNA FISH analysis was carried out on ex vivo sorted cells of the indicated lineage and developmental stage. DNA probes used were generated from two bacterial artificial chromosomes, RP23-101G13 (red signal) and RP24-387E13 (green signal), which map to the distal Vκ24 gene region located at the 5′ end of the Igk locus and the 3′ constant (Cκ) region, respectively. (d) Confocal sections representative of paired and separated Igk alleles at the indicated stage of development. A schematic representation of the position of probes used for detecting the different regions of Igk in the 3-D DNA FISH analyses is shown. (e) Graph showing the frequency of inter-allelic Igh pairing in wild-type and Pax5–/– cultured pro B cells. The rearrangement status of the cells analyzed is indicated below the appropriate bar of the graph. (f) Confocal sections representative of paired contracted Igh alleles in wild-type pro-B cells and separated decontracted alleles in Pax5–/– pro-B cells. For all graphs selected statistical results are shown, see Supplementary Tables 1 and 2 online for complete statistical results. All data are representative of more than three independent experiments.

Antigen receptor locus rearrangement correlates with germline transcription, open chromatin, and contraction of Ig and Tcr loci12,13,21,22. In the absence of Pax5, VH germline transcription, histone acetylation of VH chromatin and expression of Rag1 and Rag2 are normal, but both Igh contraction and VH-to-DJH rearrangement of mid and distal VH genes are substantially impaired11,23,24. To determine whether Ig pairing might be influenced by locus conformation, we analyzed the frequency of pairing of homologous Igh alleles in short-term cultured wild-type and Pax5-/- pro-B cells. Homologous pairing was substantially reduced in Pax5-/- pro-B cells (Fig. 1e,f and Supplementary Table 3 online). These data indicate that homologous Igh pairing occurs when the locus is present in distinct conformations at two phases of B cell development during DH-JH and VH-DJH recombination. Pairing of different Igh regions or pairing of the same region mediated by different factors could provide a mechanism to prevent the initiation of VH-DJH rearrangement prior to completion of DH-JH rearrangement on both alleles.

RAG1 mediates homologous pairing of Ig alleles

The identical developmental stage specific patterns of Ig pairing and rearrangement suggests that there may be a mechanistic link between these two phenomena. To test this we next asked whether the V(D)J recombinase contributes to interallelic pairing of Igh and Igk alleles. For this we analyzed the frequency of pairing of homologous Ig alleles in primary bone marrow derived B lymphocytes of different developmental stages sorted from wild-type, Rag1-/- or Rag1-/- mice expressing a functionally rearranged B1.8 Igh knock-in25 and/or Rag1D708A transgene26. Expression of the B1.8 rearranged Igh knock-in drives B cell development to the pre-B cell stage25 enabling analysis of the Igk locus in Rag1-/- mice. The D708A active site mutation of the RAG1 protein prevents DNA cleavage but still enables binding and synapsis of RSSs by the RAG complex26-28. A significantly higher percentage of wild-type CD19- pre-pro-B cells (24%) and CD19+ pro-B cells (22%) contained paired Igh alleles as compared to Rag1-/- CD19- pre-pro-B cells (12%) or CD19+ pro-B cells (12%) (Fig. 2a). Likewise, Igk allele pairing was significantly higher in B1.8 pre-B cells (18%) compared to Rag1-/- B1.8 pre-B cells (9%) (Fig. 2b). Notably, the frequencies at which Igh and Igk alleles were paired in Rag1-/- Rag1D708A CD19+ pro-B cells (28%) and Rag1-/- Rag1D708A B1.8 pre-B cells (17%), respectively, were increased (Igh) or comparable (Igk) to those observed in wild-type pro-B cells or B1.8 pre-B cells, respectively. Collectively, these data demonstrate that high frequencies of interallelic Ig pairing are dependent upon RAG1 expression, but not on RAG DNA cleavage activity.

Figure 2

RAG1 contributes to homologous pairing of Igh and Igk alleles.

(a) Graph indicating the frequency of inter-allelic Igh pairing in wild-type, Rag1–/– and in Rag1–/– pro-B cells containing the Rag1 transgene with the active site mutation D708A. DNA FISH experiments were carried out as described for Fig. 1. (b) Graph indicating the frequency of inter-allelic Igk pairing in B1.8, B1.8 Rag1–/– and in B1.8 Rag1–/– pre-B cells containing the Rag1 transgene with the active site mutation D708A. Selected statistical results are shown. See Supplementary Tables 1 and 2 for complete statistical results. Data are representative of at least three independent experiments.

RAG1 differentially marks paired Ig alleles

X inactivation takes place in developing female (XX) cells and seems to be dependent on pairing of X-inactivation centers present on the two individual X chromosomes18-20. To determine whether the two processes share a common function we next asked whether Ig loci pairing occurs prior to the initiation of V(D)J recombination on one allele and, if so, whether this pairing might silence or reduce the accessibility of the unrearranged Ig allele upon separation of paired Ig alleles. We first analyzed by confocal microscopy the position of paired Igh alleles relative to pericentromeric heterochromatin by performing three-color 3-D DNA FISH using the two BACs CT7-526A21 and CT7-34H6 in combination with a γ-satellite probe, which hybridizes to major satellite centromeric repeats. Wild-type pre-pro-B and pro-B cells had paired Igh alleles that were predominantly positioned close to pericentromeric heterochromatin with one allele juxtaposed to this repressive compartment and the other allele located in a euchromatic region of the nucleus (Fig. 3a,b and Supplementary Table 4 online). These data indicate that homologously paired Igh alleles are differentially marked as defined by their nuclear locations.

Figure 3

RAG1 differentially marks paired Ig alleles as defined by their location within euchromatic and heterochromatic regions of the nucleus.

(a) Graph showing the frequency with which paired Igh alleles are positioned at pericentromeric heterochromatin in wild-type and Rag1–/– pre-pro-B and pro-B cells in addition to Rag1–/– pro-B cells containing a Rag1-D708A transgene. Alleles that are not located at pericentromeric heterochromatin are positioned within euchromatic regions of the nucleus. 3-D DNA FISH analysis was carried out on ex vivo sorted cells of the indicated lineage and developmental stage as described for Fig.1. DNA probes used were CT7-526A21 (red signal) and CT7-34H6 (green signal), which map to the distal VH gene region located at the 5′ end of the Igh locus and the 3′ constant (CH) region, respectively. A γ-satellite probe was used to detect pericentromeric heterochromatin. (b) Left panels — confocal sections representative of wild-type pre-pro-B and pro-B cells in which paired Igh alleles are equivalently located in euchromatin or differentially marked with one allele juxtaposed at pericentromeric heterochromatin. Right panel — confocal section representative of paired Igh alleles relative to pericentromeric heterochromatin in Rag1–/– pro-B cells containing the Rag1-D708A transgene. A schematic representation of the position of probes used for detecting the different regions of Igh in the 3-D DNA FISH analyses is shown. A γ-satellite probe was used to detect pericentromeric heterochromatin (blue signal). (c) Mono and biallelic pericentromeric recruitment of all Igh alleles in wild-type, Rag1–/– and Rag1–/– pre-pro-B and pro-B cells and pro-B cells containing the Rag1-D708A transgene. (d) Confocal sections representative of unpaired and paired Igh alleles relative to pericentromeric heterochromatin in cells with one rearranged VDJH allele. Unrearranged alleles are indicated by the presence of a BAC signal for the VH to DH region RP24-275L15 (red signal), alongside a CH CT7-34H6 probe (green signal). A schematic representation of the position of probes used for detecting the different regions of Igh in the 3-D DNA FISH analyses is shown. A γ-satellite probe was used to detect pericentromeric heterochromatin (white signal). Rearranged alleles are indicated by the presence of a single BAC signal, CT7-34H6 (green signal). (e) Graph showing the frequency with which rearranged or unrearranged Igh alleles are positioned at pericentromeric heterochromatin in cells with one VH-to-DJH rearranged allele. The analysis was performed on cells with unpaired or paired alleles, as indicated. The rearrangement status of the cells analyzed is indicated below the graph. (f) Graph showing the frequency with which paired Igh alleles are positioned at pericentromeric heterochromatin in wild-type pro-B cells irrespective of rearrangement status and in comparison to paired unrearranged alleles containing the intergenic VH-DH region of the Igh locus (as judged by the presence of both BAC RP24-275L15 signals). Probes used are described in c and the rearrangement status of the cells analyzed is indicated below the appropriate bar of the graph. For all data, selected statistical results are shown. See Supplementary Tables 4, 5 and 6 for complete statistical results. All data are representative of three independent experiments.

Figure 3 part 2 (g) Graph showing the frequency with which paired Igk alleles are positioned at pericentromeric heterochromatin in B1.8, B1.8 Rag1–/– and B1.8 Rag1–/– pre-B cells containing the Rag1 transgene with the active site mutation D708A. Probes used were RP23-101G13 (red signal) and RP24-387E13 (green signal), which map to the distal Vκ24 gene region located at the 5′ end of the Igk locus and the and the 3′ constant (Cκ) region, respectively. (h) Upper panels — Confocal sections representative of differentially marked paired alleles in B1.8 and B1.8 Rag1–/– pre-B cells containing the Rag1 transgene with the active site mutation D708A and paired alleles located equivalently in euchromatic regions in Rag1–/– pre-B cells. A γ-satellite probe (white signal) was used to detect pericentromeric heterochromatin. Lower panels — Confocal sections showing the position of unpaired Igk alleles relative to pericentromeric heterochromatin in B1.8 Rag1–/– and B1.8 Rag1–/– pre-B cells containing the Rag1 transgene with the active site mutation D708A. (i). Mono and biallelic pericentromeric recruitment of all Igk alleles in B1.8, B1.8 Rag1–/– and B1.8 Rag1–/– pre-B cells containing the Rag1 transgene with the active site mutation D708A. For all graphs selected statistical results are shown. See Supplementary Tables 7 and 8 for complete statistical results. Results are representative of three independent experiments.

To determine whether recombination could be a prerequisite for differential marking, we next analyzed the position of paired Igh alleles relative to pericentromeric heterochromatin in sorted Rag1-/- pre-pro-B and pro-B cells in the presence or absence of the Rag1D708A transgene (Fig. 3a,b and Supplementary Table 4). We observed that paired Igh alleles in Rag1-/- pre-pro-B and pro-B cells had one Igh allele associated with pericentromeric heterochromatin at a significantly lower frequency (19% and 9%, respectively) than in comparable wild-type populations (45% and 39%, respectively). This difference could not be rescued by the presence of the Rag1D708A transgene. Hence, differential nuclear positioning of Igh alleles is induced by RAG-mediated cleavage. Recombination also affected the relative distribution of Igh in total pre-pro-B and pro-B populations (Fig. 3c and Supplementary Table 5 online) indicating that RAG-mediated DNA cleavage was required for monoallelic Igh association with pericentromeric heterochromatin in cells where alleles were paired or unpaired.

We next analyzed the VH-to-DJH rearrangement status of Igh alleles relative to pericentromeric heterochromatin by performing three-color 3-D DNA FISH using a γ-satellite probe in combination with the CH BAC and a BAC probe, RP24-275L15, which hybridizes to the intergenic region between the VH and DH gene segments (Fig. 3d). Any VH-to-DJH recombination event will result in deletion of this intergenic region and loss of this signal. We analyzed cells containing one germline (275L15+) and one VHDJH rearranged (275L15-) Igh allele and observed the germline (275L15+) Igh allele associated with pericentromeric heterochromatin in 67% of pro-B cells with unpaired Igh alleles and in 81% of pro-B cells with paired Igh alleles (Fig. 3e and Supplementary Table 6 online). We additionally looked at the position of paired unrearranged Igh alleles in cells in which both 275L15 signals were present (Fig. 3f and Supplementary Table 4). Our results indicate that 78% of unrearranged paired alleles were positioned in euchromatic regions of the cell. Taken together our data suggest that unrearranged alleles pair up in a recombinase-dependent manner in euchromatic regions of the nucleus. Differential marking of paired alleles, following RAG1-mediated cleavage, directs the unrecombined allele to pericentromeric heterochromatin while the rearranged allele remains in a euchromatic region of the nucleus.

We next analyzed the position of Igk alleles relative to pericentromeric heterochromatin by performing DNA FISH using the Vκ and Cκ BACs in combination with a γ-satellite probe. We compared the nuclear location of paired Igk alleles in control B1.8 pre-B cells with paired Igk alleles in Rag1-/- B1.8 and Rag1-/- Rag1D708A B1.8 pre-B cells. Substantially less pericentromeric association of paired Igk occurred in Rag1-/- cells (27%) compared to 80% of the control cells (Fig. 3g,h and Supplementary Table 7 online), and, in contrast to Igh, expression of the Rag1D708A mutant rescued this effect (78%). Furthermore RAG1 expression, independent of RAG-mediated DNA cleavage, was required for monoallelic Igk association with pericentromeric heterochromatin in overall pre-B populations (Fig. 3i and Supplementary Table 8 online).

ATM directs allelic repositioning

We have recently demonstrated that RAG-mediated Ig cleavage activates a multi-functional genetic program that is, in part, dependent upon the ATM repair-checkpoint protein29. We next analyzed the relative positions of Igh loci in pre-pro-B and pro-B cells from Atm-/- mice to determine whether ATM dependent signals were required for RAG-cleavage induced association of unrearranged Igh alleles with pericentromeric heterochromatin. Although we observed that homologous pairing of Igh alleles occurs at the same high frequency in Atm-/- and wild-type pre-pro-B and pro-B cells, paired Igh alleles remain predominantly euchromatic (72% and 75%, respectively) in contrast to what was observed in the equivalent wild-type populations (44% and 41%, respectively; Fig. 4a,b and Supplementary Table 4). In addition the overall frequency of mono- and biallelic pericentromeric recruitment of Igh alleles was reduced in Atm-/- pre-pro-B and pro-B cells compared to that seen in the equivalent wild-type populations (41% and 34%, respectively; Fig. 4c, d and Supplementary Table 5).

Figure 4

ATM directs repositioning of one Ig allele to pericentromeric heterochromatin.

(a) Graph indicating the frequency of interallelic Igh pairing in wild-type and Atm–/– pre-pro-B and pro-B cells. The developmental profile of Atm–/– compared to wild-type sorted bone marrow primary B lymphocytes is described in Supplementary Fig. 2 (online). (b) Graph showing the frequency with which paired Igh alleles are positioned at pericentromeric heterochromatin in wild-type and Atm–/– pre-pro-B and pro-B cells. (c) Mono and biallelic pericentromeric recruitment of total Igh alleles in wild-type and Atm–/– pre-pro-B and pro-B cells. (d) Confocal sections showing the position of undifferentially marked paired Igh and Igk alleles at the indicated stages of development. Probes used in these experiments are indicated. (e) Graph indicating the frequency of interallelic Igk pairing in wild-type and Atm-/- pre-B cells. (f) Graph showing the frequency with which paired Igk alleles are positioned at pericentromeric heterochromatin in wild-type and Atm–/– pre-B cells. (g) Mono and biallelic pericentromeric recruitment of Igk in wild-type and Atm–/– pre-B cells. For all graphs selected statistical results are shown. See Supplementary Tables 1, 2, 5, 6, 7 and 8 for complete statistical results. All results shown are representative of three independent experiments.

At the Igk locus, again comparable frequencies of homologous pairing occurred in Atm-/- and wild-type pre-B cells (Fig. 4e and Supplementary Table 2) but differential marking of paired Igk alleles was reduced (50% in Atm-/- pre-B cells compared to 68% in wild-type pre-B cells) and we additionally observed significantly reduced monoallelic pericentromeric recruitment of Igk in the overall population of Atm-/- pre-B cells (39% in Atm-/- pre-B cells compared to 71% in wild-type pre-B cells) (Fig. 4d,f,g and Supplementary Tables 7,8). Collectively our data suggest that the activation of ATM by RAG-mediated cleavage during Ig rearrangement provides signals by which the unrearranged allele is associated with pericentromeric heterochromatin.

ATM prevents biallelic RAG-mediated cleavage

ATM rapidly phosphorylates the H2AX histone variant to form γ-H2AX in chromatin around DSBs, including those induced by RAG during Tcra rearrangements in thymocytes8. To determine whether initiation of V(D)J recombination is coordinated between homologous Ig alleles, we investigated whether RAG-dependent H2AX phosphorylation along Ig loci occurred in a monoallelic or biallelic manner in developing B cells. Importantly, the localization of γ-H2AX with Igh was absolutely dependent on recombinase enzymes since these foci are never associated with Igh loci in Rag1-/- cells (Supplementary Fig. 3 online). In cells containing γ-H2AX-associated Igh signals, 97.2% of pre-pro-B and 99.2% of pro-B cells had monoallelic γ-H2AX association with Igh whereas 2.8% of pre-pro-B and 0.8% of pro-B cells had biallelic association. Similarly in pre-B cells containing Igk alleles coincident with γ-H2AX 94.8% had monoallelic association with Igk while 5.2% were biallelically associated (Supplementary Fig. 4a,b online and Supplementary Tables 9, 10 online). Localization of γ-H2AX foci on both alleles of Igh or Igk could represent either simultaneous biallelic RAG-generated DSBs or a lag in the resolution of γ-H2AX foci after repair of monoallelic DSBs. Regardless, these data indicate that RAG-mediated cleavage occurred on only one allele at a time in the majority of primary bone marrow cells during DH-to-JH, VH-to-DJH and Vκ-to-Jκ recombination.

Monoallelic targeting of the V(D)J recombinase could reflect either an inefficient or regulated process. To distinguish between these possibilities, we calculated the number of cells in which we would expect to find biallelic association of γ-H2AX (as judged by the presence of γ-H2AX foci) if recombinase targeting were not regulated and compared it with the actual number observed. If γ-H2AX foci are observed in x% of a given population, the expected frequency of biallelic recombination would be x% of x%. Notably, the percentage of cells in which we actually observed biallelic association of γ-H2AX with Igh and Igk in cells undergoing recombination was significantly lower than the predicted frequency (Fig. 5a—e and Supplementary Table 9, 10). These observations support the notion that DH-to-JH, VH-to-DJH and Vκ-to-Jκ rearrangement each occur on one allele at a time and that monoallelic V(D)J recombination occurs as a result of regulated or restricted targeting of RAG-mediated cleavage and is not simply a result of inefficient recombination.

Figure 5

ATM prevents bi-allelic RAG-mediated cleavage during Ig V(D)J rearrangement.

(a) Graph showing the predicted and actual percentages of cells with biallelic colocalization of γ-H2AX on Igh in wild-type and Atm–/– pre-pro-B cells. This percentage is calculated based on the total percentage of pre-pro-B cells with mono or biallelic colocalization of γ-H2AX on Igh alleles. (b) Graph showing the predicted and actual percentages of cells with biallelic colocalization of γ-H2AX on Igh in wild-type and Atm–/– pro-B cells. This percentage is calculated based on the total percentage of pro-B cells with mono or biallelic colocalization of γ-H2AX on Igh alleles. (c) Graph showing the predicted and actual percentages of cells with biallelic colocalization of γ-H2AX on Igk in wild-type and Atm–/– pre-B cells. This percentage is calculated based on the total percentage of pre-B cells with mono or biallelic colocalization of γ-H2AX on Igk alleles. (d) Individual and merged confocal microscopy sections indicate the localization of γ-H2AX foci on Igh alleles in sorted wild-type and Atm–/– pre-pro-B cells. Immunofluorescence staining indicates γ-H2AX foci (red). The DNA probe CT7-526A21 (white signal) indicates the position of the 3′ constant (CH) region. Representative cells show paired Igh alleles with γ-H2AX localized mono-allelically in wild-type pre-pro-B cells and bi-allelically in Atm–/– pre-pro-B cells. Right hand panels show enlarged Igh alleles. (e) Individual and merged confocal microscopy sections indicate the localization of γ-H2AX foci on Igk alleles in sorted wild-type and Atm–/– pre-B cells. Immunofluorescence staining indicates γ-H2AX foci (green). DNA probes RP23-101G13 (white signal) and RP24-387E13 (red signal) indicate the position of the distal Vκ gene region located at the 5′ end of the Igk locus and the 3′ constant (Cκ) region, respectively. Representative cells show paired Igk alleles with γ-H2AX localized mono-allelically in wild-type pre-B cells and bi-allelically in Atm–/– pre-B cells. Right hand panels show enlarged Igk alleles. For all graphs selected statistical results are shown. See Supplementary Tables 9 and 10 for complete statistical analysis. All results shown are representative of three independent experiments.

Since activation of ATM by RAG-mediated cleavage provides signals by which the unrearranged allele associated with pericentromeric heterochromatin, we next asked whether ATM restricted RAG-mediated cleavage to one allele at a time. For this we carried out γ-H2AX-Ig immuno-DNA FISH on developing B lymphocytes sorted from the bone marrow of Atm-/- mice. Compared to the equivalent wild-type cells, the average number of γ-H2AX foci was reduced but not absent in Atm-/- pro-B cells (5.97 compared to 5.65, respectively) and pre-B cells compared (5.05 and 3.18, respectively) indicating that many γ-H2AX foci in developing B cells were generated by another H2AX kinase such as DNA-PKcs or ATR30. In Atm-/- cells containing Igh alleles coincident with γ-H2AX, 10.3% of pre-pro-B and 3.1% of pro-B cells had biallelic association and in pre-B cells containing Igk alleles coincident with γ-H2AX 18.8% had biallelic association with Igk (Fig. 5d,e and Supplementary Fig. 4a,b and Supplementary Table 9,10). We next calculated the predicted frequency at which we would expect to see biallelic DSBs introduced on Igh and Igk in cells where these loci are undergoing rearrangement and compared these values to the actual frequency at which we observed biallelic localization of γ-H2AX on Ig loci in these cells. No significant differences were found between predicted frequencies for biallelic localization of γ-H2AX on Igh and Igk in pre-pro-B cells and pre-B cells, respectively, and the frequencies that we actually observed (Fig. 5a,c). Although the introduction of DSBs on both alleles was also elevated in Atm-/- pro-B cells, the increase was less marked than what we observed in Atm-/- pre-pro-B cells. However, as has been previously reported by others31, we observed broken and/or missing Igh alleles in a significant percentage of Atm-/- pro-B cells. In this context, deregulated targeting of the recombinase machinery during DH-to-JH rearrangement in Atm-/- pre-pro-B cells could interfere with VH-to-DJH recombination at the later pro-B cell stage because of a reduced number of intact Igh loci. Collectively, our observations suggest that ATM-mediated repositioning of unrearranged Ig alleles with pericentromeric heterochromatin reduced accessibility and inhibited biallelic RAG-mediated cleavage.

ATM prevents biallelic Igk chromosome breaks

Treatment of NHEJ-deficient v-Abl transformed (Abl) pre-B cell lines with the Abl kinase inhibitor STI571 leads to G1 arrest, Rag1 and Rag2 expression and accumulation of unrepaired RAG-generated Igk coding ends (CEs), which activate ATM-dependent signals29,32. To directly evaluate whether ATM prevents biallelic RAG-mediated cleavage, we treated Rag2-/-, Artemis-/- and Artemis-/-Atm-/- cells with STI571 for 1–7 days and performed Jκ locus Southern blot analysis on DNA isolated from these cells. Artemis is an NHEJ factor required for opening CEs so that RAG-generated DSBs are repaired33,34. Retention of the germline Jκ band was observed in the control Rag2-/- cells (Fig. 6a, while in Artemis-/- cells, we found a 30–50% reduction in the germline Jκ band and a corresponding increase in Jκ CE bands (Fig. 6a and Supplementary Fig. 5 online). These data demonstrate that RAG-mediated cleavage occurred on half of the Igk alleles within these cells. In contrast, we found an 80% reduction in the germline Jκ band and a corresponding increase in Jκ CE bands in Artemis-/-Atm-/- cells (Fig. 6a and Supplementary Fig. 5), indicating that RAG-mediated cleavage occurred on more than half of the Igk alleles within these cells. Similar findings were observed in independently derived cell lines (Supplementary Fig. 5). These data suggest that Jκ CEs activate ATM signals that prevent RAG-mediated cleavage of the other Igk loci until NHEJ-mediated formation of VκJκ coding joints.

Figure 6

ATM prevents bi-allelic Igk chromosome breaks and translocations.

(a) Southern blot analysis of the Igk locus was performed on BamHI-digested genomic DNA isolated from Rag2–/–, Artemis–/–, and Artemis–/–Atm–/– Abelson pre-B cell lines treated with STI571 for the indicated number of days. The top panel depicts blots hybridized with a 3′Jκ probe. Bands corresponding to germline Igk (Jκ GL) and un-repaired Jκ coding ends (Jκ CE) are indicated. The same blots were stripped and re-hybridized with a Tcrb locus probe to account for DNA amounts. The intensity of the germline Jκ band was normalized to that of the Tcrb band to represent RAG-mediated Igk cutting within each population of cells. (b) Quantification of RAG-induced Igk genomic instability during V(D)J recombination in pre-B cell lines. Left, representative light microscopy images of two-color FISH analysis conducted on G1 phase nuclei of STI571 treated Artemis–/–p53–/– and Artemis–/–Atm–/– pre-B cells using a 5′Vκ BAC (red signal) and 3′Cκ BAC (green signal) and DAPI to visualize DNA. Right, graph showing the number and percentage of Artemis–/–p53–/– and Artemis–/–Atm–/– pre-B cells with coincident (C) and non-coincident (NC) hybridization of the 5′Vκ and 3′Cκ signals. The percentages of cells with separated signals on both alleles are statistically different (P = 0.005) between Artemis–/–p53–/– and Artemis–/–Atm–/– pre-B cells. (c) Quantification of RAG-induced Igk chromosome breaks or translocations during V(D)J recombination in pre-B cell lines. Left, representative light microscopy images of whole chromosome 6 (red) paints and FISH analysis using the 5′Vκ and 3′Cκ BACs (green signals) and DAPI (blue) to visualize DNA on metaphases prepared from STI571-treated and released Artemis–/–p53–/– and Artemis–/– Atm–/– pre-B cells. Right, graph showing the percentage of STI571 treated and released pre-B cell lines with Igk chromosome breaks or translocations on one or both copies of chromosome 6. All data shown are representative of at least three independent experiments.

During V(D)J recombination, ATM maintains CEs within repair complexes and prevents RAG-initiated genomic instability31,32. To investigate whether ATM prevents biallelic dissociation ofIgk CEs and biallelic chromosome breaks or translocations involving Igk we performed two-color DNA FISH using a 5′ Vκ BAC RP24-243E11 and a 3′ Cκ BAC RP23-341D5 probe on G1 nuclei prepared from STI57-treated Artemis-/-p53-/- and Artemis-/-Atm-/- pre-B cells. We used Artemis-/-p53-/- cells rather than Artemis-/- cells so that the ATM-p53-dependent G1-S checkpoint was impaired as in Artemis-/-Atm-/- cells. We observed coincident (<1 μm) probe signals on both alleles in 85% of Artemis-/-p53-/- nuclei, with non-coincident probe signals on one allele in 15% and both alleles in less than 1% of nuclei (Fig. 6b). We found coincident probe signals on both alleles in 50% of Artemis-/-Atm-/- nuclei, with non-coincident probe signals on one allele in 41% and on both alleles in 10% of nuclei (Fig. 6b). The difference between non-coincident signals in 1% of Artemis-/-p53-/- nuclei and 10% of Artemis-/-Atm-/- nuclei was statistically significant (P = 0.005). Next, we performed FISH using the Vκ and Cκ BACs and a chromosome 6 ‘paint’ on metaphase spreads prepared from untreated cells or cells that had been released back into cycle by removing STI571. We observed chromosome 6 translocations in less than 3% of metaphases prepared from untreated cells (data not shown). We found RAG-induced Igk chromosome breaks or translocations in 4% of Artemis-/- p53-/- cells, whereby lesions arose from a single Igk allele (Fig. 6c). In contrast, we observed Igk chromosome breaks or translocations in 66% of Artemis-/-Atm-/- cells, with 30% of cells containing lesions arising from both Igk alleles (Fig. 6c). The difference between the absence of biallelic Igk abnormalities in Artemis-/-p53-/- nuclei and 30% of Artemis-/-Atm-/- nuclei was statistically significant (P = 0.005). Collectively, these data demonstrate that ATM regulates monoallelic RAG-mediated cleavage of homologous Igk loci to prevent biallelic V(D)J recombination errors.

Igh pairing can occur beyond the pro-B cell stage

A prediction of our findings is that in Atm-/- mice allelic exclusion could be violated. However flow cytometry analyses for surface and cytoplasmic IgH expression of Atm-/- mice containing allotypically marked Igh alleles indicate that Igh allelic exclusion was maintained (Fig. 7a and Supplementary Fig. 6 online). We next asked whether maintaining accessibility of the Igh locus at the pre-B cell stage could maintain a high frequency of homologous association leading to silencing of one allele in cells that are expressing two functionally rearranged Igh alleles. Interleukin 7 receptor (IL-7R) signaling mediated by the transcription factor STAT5 is thought to be responsible for histone acetylation and transcription of distal VH genes at the pro-B cell stage35. Following functional rearrangement of one Igh allele, pre-BCR signaling results in attenuation of IL-7R signaling, which leads to histone deacetylation and reduced VH gene accessibility36. The expression of a constitutively active Stat5A (caStat5) transgene37 may therefore prevent the loss of Igh accessibility at the pre-B cell stage. To determine this, we used mice that expressed either caStat5A or caStat5B transgenes37,38. The results of the experiments with the two transgenic mice strains were equivalent, and data are shown for B cells derived from the caStat5B mice. We performed DNA FISH with sorted developing B cells from caStat5 and wild-type mice as described in Fig. 4. In contrast to wild-type pre-B cells, recruitment of one Igh allele to pericentromeric clusters did not occur in caStat5 pre-B cells and was delayed until the immature B cell stage (Fig. 7b and Supplementary Table 5). These data indicate that inactivation of STAT5 was pivotal in changing accessibility at the Igh locus as defined here by nuclear location and that STAT5 was important in translating signals from the pre-BCR into alterations of Igh position within the nucleus. To determine whether accessibility could influence homologous Igh pairing beyond the pro-B cell stage, we examined the frequency of association of the two Igh alleles in pre-B and immature B cells from wild-type and caStat5 mice. In contrast to wild-type cells, the association of Igh alleles remained high in caStat5 pre-B cells and started to decline only at the subsequent immature B cell stage (Fig. 7c,d and Supplementary Table 1), when one Igh allele also became repositioned to pericentromeric heterochromatin (Fig. 7b and Supplementary Table 5). In summary, the frequency of Igh pairing depended on accessibility and was reduced when one Igh allele repositioned to a repressive nuclear compartment in pre-B cells.

Figure 7

Igh pairing can occur beyond the pro-B cell stage if locus accessibility is maintained.

(a) Flow cytometric analysis of bone marrow cells from wild-type and Atm–/– mice heterozygous for two allotypically marked Igh alleles. (b) Graph showing mono- and biallelic recruitment of Igh at different developmental stages in ex vivo sorted wild-type (grey bars) and caStat5 (red bars) B lymphocytes. 3-D DNA FISH was carried out as described for Fig. 1. (c) Graph showing frequency of association of Igh alleles in sorted wild-type and caStat5 pre-B and immature B cells. (d) Confocal sections showing tightly paired Igh alleles in a caStat5 pre-B cell. Representative confocal sections indicate the position and conformation of non-associating Igh alleles in a wild-type pre-B cell. DNA probes used in the FISH experiments are as indicated. (e) Confocal sections showing cells with (right) and without (left) deletion of one CT7-526A21 VHJ558 signal. Probes used in this experiment are indicated in d. (f) The frequency of deletion of the CT7-526A21 VHJ558 signal in wild-type and caStat5 pre-B cells is shown in the graph. (g) Graph showing mono- and biallelic recruitment of Igh at in ex vivo sorted wild-type (grey bars) and Atm–/– (red bars) pre-B cells. (h) Graph showing frequency of association of Igh alleles in sorted wild-type (grey bars) and Atm–/– (red bars) pre-B cells. For all graphs, selected statistical results are shown. See Supplementary Table 6 and 8 for complete statistical results. All results shown are representative of three independent experiments.

Association of Igh with the Igk locus occurs at the pre-B cell stage. Analysis of sorted wild-type and caStat5 B lymphocytes at different developmental stages revealed that the frequency of association of Igh and Igk alleles was substantially reduced in pre-B and early immature B cells of caStat5 mice compared to wild-type mice (Supplementary Fig. 7a online and Supplementary Table 11 online). As association of the Igh locus with an Igk allele is a prerequisite for decontraction of the Igh locus in pre-B cells39 we analyzed the contraction state of the Igh locus by 3-D FISH in developing B lymphocytes of caStat5 mice. Consistent with the low frequency of association between Igh and Igk alleles, the Igh locus remained contracted in pre-B and immature B cells of caStat5 mice (Supplementary Fig. 7b and Supplementary Table 3). We have previously shown that ongoing accessibility of the Igh locus in 3′ Igk enhancer-deficient mice (3′Eκ-/-) pre-B cells results in an increase of rearrangements involving distal VHJ558 gene segments as judged by an absence of CT7-526A21 FISH signals, which detects more proximally located VHJ558 gene segments39. A similar FISH analysis of caStat5 pre-B cells revealed that these CT7-526A21 VHJ558 gene segments were also deleted more frequently as compared with wild-type pre-B cells (Fig. 7e,f and Supplementary Table 12 online). Ongoing VH-DJH recombination was confirmed by ligation-mediated PCR (LM-PCR) to assay for the presence of recombinase mediated DSBs at Igh locus RSSs. DNA breaks represent reaction intermediates of rearrangement (Supplementary Fig. 7c). Flow cytometric analysis of mice heterozygous for two allotypically marked Igh alleles indicated, however, that allelic exclusion was not violated in splenic B cells of caStat5 mice (Supplementary Fig 8 online) similar to 3′Eκ-/- mice39,40.

In Atm-/- mice accessibility of the Igh locus was prolonged beyond the pro-B cell stage as it was in the 3′Eκ-/- and caStat5 mice. In all three genotypes ongoing accessibility of Igh alleles at the pre-B cell stage in Atm mice was correlated with a high frequency of association of Igh alleles (Fig. 7c,g,h and Supplementary Table 1 (data for the 3′Eκ-/- mice is not shown)). Furthermore in all the genotypes we have examined allelic exclusion remained intact despite ongoing accessibility leading to increased VHJ558 gene rearrangement or biallelic RAG-mediated cleavage. These data support the idea that there is a second checkpoint for establishing allelic exclusion at the pre-B to immature B transition41, which could be mediated by homologous pairing and differential marking of Igh alleles in pre-B cells.

DISCUSSION

Generation of a monospecific BCR on individual B cells is thought to be important for preventing autoimmunity but the mechanisms by which this is achieved remain unclear. Based upon our collective observations, we propose a model in which homologous Igh and Igk alleles proceed through cycles of pairing and separation with V(D)J recombination initiating on a single paired Ig allele. RAG-mediated cleavage activates ATM to mediate differential marking of the two alleles, which in turn leads to preferential association of the non-cleaved allele with pericentromeric heterochromatin. We propose that the ATM driven repositioning of the undamaged allele to these repressive nuclear regions enables a mechanism for transiently inhibiting accessibility to the recombinase and provides a time-window in which the RAG-cleaved allele can be repaired, transcribed, translated and tested for functionality. Assembly and expression of a productive VHDJH or VκJκ rearrangement on the first allele leads to expression of the pre-BCR or BCR and initiation of classical feedback inhibition of further VH or Vκ rearrangement. The ATM driven marking of the uncleaved allele could be reversed either because DSB repair leads to cessation of ATM-dependent signals, and/or by restoring accessibility. As the accessibility of Ig loci is maintained at their respective stage of V(D)J recombination, it is conceivable that changes induced by differential marking of the unrearranged allele could be reversed. This would result in disassociation of the unrearranged allele from pericentromeric heterochromatin and enable another round of pairing, cleavage and repositioning to occur. Such cycles could occur until either a functional VH-to-DJH or Vκ-Jκ rearranged allele is produced or all possible V(D)J recombination events are exhausted and the cell dies. Although we hypothesize that V(D)J recombination initiates on paired Ig alleles our data cannot rule out the possibility that RAG-mediated cleavage also occurs on unpaired alleles. However if ATM that is localized on the rearranging allele acts in trans to inhibit recombination of the other allele this is more likely to occur if the two Ig alleles are closely paired than for the pool of activated ATM to diffuse over a large distance if the two alleles are separated.

In addition to allelic exclusion of antigen receptor loci, monoallelic expression of maternal or paternal genes in mammalian cells includes X-inactivated and autosomal genes. It is known that pairing of X chromosomes has an important role in X inactivation, but the mechanisms leading to monoallelic expression of autosomal genes remain poorly understood. Transient pairing of the two X chromosomes in developing female cells requires transcription of both alleles and ensures that only one chromosome is chosen and targeted for silencing prior to differential epigenetic marking and separation of the homologous alleles18,19. Our data suggest that in parallel with X inactivation, homologous pairing of Ig loci contributes to allelic exclusion by ensuring that only one allele is targeted for recombination at any time. Interallelic pairing could therefore be a general mechanism for establishing monoallelic gene expression.

From the ‘selfish gene’ perspective, selective pressure may have forced the adaptive immune system to evolve in higher organisms to ensure the propagation of genetic material from generation to generation. In this context, it has been proposed that antigen receptor allelic exclusion evolved under selective pressure to prevent autoimmunity by ensuring proper positive or negative selection and receptor editing. Here, we propose that allelic exclusion may additionally have evolved from necessity to preserve cellular survival, maintain genomic integrity, and suppress oncogenic translocations during antigen receptor gene rearrangements.

Methods

Mice

The following mice were maintained on the C57BL/6 background and genotyped as described: Rag1-/-42 and Rag2-/-43. The mouse Rag1-D708A BAC transgene was derived from the HG BAC (≈170 kb), kindly provided by M. Nussenzweig, which spans the Rag locus and encodes GFP instead of Rag244. The GAT codon for Rag1 aspartate 708 was changed to GCG (encoding alanine) by homologous recombination in bacteria. Mice containing the full-length Rag1-D708A BAC transgene (copy number 2-3) were bred with Rag1-/- mice on the C57BL/6 background to generate Rag1-/- D708A mice. B1.8mice contain a functionally rearranged Igμ knock-in25. The Atm-/- mice were generated through the interbreeding of 129SvEv Atm+/- mice and genotyped as described45. The animal care was approved by NYU SoM Institutional Animal Care and Use Committee (IACUC) Protocol #060509-03, Yale IACUC Protocol# 2006-07610 and Children’s Hospital of Philadephia IACUC Protocol # 2007-9-711.

Flow cytometry sorting and analysis

Bone marrow cells were isolated, non-specific antibody staining was suppressed by pre-incubation of cells with CD16/CD32 Fc-block (clone 2.4G2) and cells were purified by flow cytometry on a MoFlo (Dako). All antibody staining was carried out at 4°C for 20 minutes. For sorting of pre-pro-B, pro-B and pre-B populations, cells were stained with lineage markers and streptavidin PE-Cy5.5, IgM FITC, CD25 PE, B220 PE-Cy7, cKit APC, CD19 APC-Cy7 and sorted as lineage marker negative (Lin-), CD25-, IgM-, B220+, cKit+ and then either CD19- or CD19+ for pre-pro-B cells and pro-B cells respectively. Pre-B cells were sorted as Lin-, IgM-, B220+, cKit- CD25+, and CD19+. Alternatively, pro-B, pre-B, early immature and late immature B cells were stained with CD19 APC-Cy7, cKit APC, CD25 PE, IgM PE-Cy7, IgD FITC and pro-B cells sorted as CD19+, cKit+, CD25- and IgM-, pre-B cells as cKit-, CD19+, CD25+, IgM--, early immature B cells as CD19+, IgMlo, IgD- and late immature B cells as CD19+, IgMhi, IgD-. The purity of the sorted cells was verified by reanalysis. Antibodies were from BD Biosciences, except where mentioned: CD19 allophycocyanin-Cy7 (APC-Cy7, clone 1D3), CD117 allophycocyanin (cKit APC, 2B8), CD25 R-phycoerythrin (PE or FITC, PC61), IgM R-phycoerythrin-Cy7 (PE-Cy7 or FITC, R6-60.2), IgD fluorescein isothiocyanate (FITC, 11-26c.2a), B220 PE-Cy7 (RA3-6B2). The following lineage (Lin) marker antibodies were used conjugated to biotin and detected with streptavidin-PE-Cy5.5, in order to eliminate lineage-positive cells (Lin+) by electronic gating: CD3e (145-2C11), CD8a (53-6.7), NK1.1 (PK136), Ly6G and 6C (Gr-1, RB6-8C5), CD11b (Mac-1, M1/70), CD11c (eBioscience, N418), TER-119/Erythroid cells. Purification of resting splenic B cells was carried out as described previously46.

Three-dimensional DNA FISH

Cells sorted by flow cytometry were washed in PBS and then were fixed on poly-L-lysine-coated slides for two-color and three-color three-dimensional DNA-FISH analysis as described11. Probes were directly labeled by nick translation with ChromaTide Alexa Fluor 488-5-dUTP, ChromaTide Alexa Fluor 594-5-dUTP (Molecular Probes) or dUTP-indodicarbocyanine (GE Healthcare). The γ-satellite probe was prepared from a plasmid containing eight copies of the γ-satellite repeat sequence46 and was directly labeled with dUTP—fluorescein isothiocyanate (Roche; Enzo Biochem) or dUTP-indodicarbocyanine.

Immunofluorescence and DNA FISH-immunofluorescence

Detection of γ-H2AX alone was carried out on cells adhered to poly-L lysine coated coverslips, fixed with 2% paraformaldehyde in PBS for 10 min at 22°C and permeabilized for 5 min with 0.4% Triton in PBS. Non-specific antibody binding was blocked with 2.5% BSA, 10% normal goat serum and 0.1% Tween-20 in PBS for 30 min. γ-H2AX staining was carried out using an antibody against phosphorylated serine-139 of H2AX (clone JBW301, Millipore) diluted 1:500 in blocking solution for 1 h at 22°C. Cells were rinsed with 0.2% BSA, 0.1% Tween-20 in PBS and stained with goat-anti-mouse IgG Alexa Fluor 488 or 633 (Invitrogen) for 1 h. Cells were rinsed with 0.1% Tween-20 in PBS and mounted in Vectashield (Vector Laboratories) containing DAPI to counterstain total DNA.

Combined detection of γ-H2AX and Igh was carried out on cells fixed and stained for γ-H2AX detection as above. Following the final 0.1% Tween-20 in PBS rinse cells were post fixed in 3% paraformaldehyde for 10 min at 22°C, permeabilized in 0.7% Triton-X-100 in 0.1M HCl for 15 min at 0°C and treated with 0.1 mg/ml RNaseA for 30 min at 37°C. Cells were then denatured with 1.9 M HCl for 30 min at 22°C and rinsed with cold PBS. DNA probes were denatured for 5 min at 95°C and applied to coverslips which were sealed onto slides with rubber cement and incubated overnight at 37°C. Cells were then rinsed for 30 min with 2×SSC at 37°C, 2× SSC for 30 min at 22°C, 1×SSC for 30 min at 22°C and rinsed in PBS. Cells were mounted and counterstained as above.

Confocal microscopy and analysis

Cells were analyzed by confocal microscopy on a Leica SP2 or Leica SP5 AOBS system (Acousto-Optical Beam Splitter). Optical sections separated by 0.3 μm were collected, and only cells with signals from both alleles (typically over 95%) were analyzed using Leica software. Alleles were measured in 3-dimensions and pairing was defined as less than 1 μm distance. Alleles were defined as pericentromerically localized if the γ-satellite signal was overlapping or immediately juxtaposed. Alleles were defined as co-localized with γ-H2AX if the signals overlapped. Samples sizes were typically 200 cells minimum for association of alleles and upwards of 1000 cells for γ-H2AX co-localization. Statistical significances were calculated using χ2 analysis as described47 in a pairwise analysis where each genotype and cell type was paired with the most biologically relevant differentiation stage or cell type. To avoid observer bias experiments were analyzed by more than one person.

Cell culture

Pro-B cell cultures were established using bone marrow cells enriched for CD19+ cells by magnetic activated cell sorting (MACS) and cultured on OP9 stromal cells, in Iscove’s modified Dulbecco’s medium containing 2% serum, 0.03 % primatone RL (Mediatech), 4 mM glutamine, 50 μM β-mercaptoethanol, penicillin/streptomycin and 0.5 ng/ml IL-7 supernatant harvested from J558L IL-7 secreting cells. Murine embryonic fibroblasts were obtained from E13.5-15.5 embryos, head and visceral organs were removed, trypsinised and cultured for 4 days in Dulbecco’s modified Eagle’s medium with 10% serum, 1 mM glutamine, 50 μM β-mercaptoethanol and penicillin/streptomycin.

Analysis of Igk rearrangements in Abl pre-B cells

The generation and characterization of the Atm-/-, Artemis-/-, and Artemis-/-Atm-/- Abelson pre-B cell lines were previously described29,32. The Artemis-/-p53-/- Abelson pre-B cell lines were generated as described32 from Artemis-/- and p53 mice48 and then subject to Tat-Cre mediated deletion of p53. Induction and Southern blot analysis of Igk rearrangement in these cells also was previously described29,32. Nuclei of G1-arrested cells following STI571 treatment were isolated by hypotonic treatment, fixed in methanol:acetic acid (3:1 in volume), and subjected to interphase 2-color FISH using biotin- or digoxin-labeled BAC probes. After Igk break induction and release from STI571, cells were allowed to proliferate and re-enter the cell cycle, before Colcemid (Invitrogen) treatment and metaphase preparation according to standard protocols. The 5′ Vκ BAC RP24-243E11 and the 3′ Cκ BAC RP23-341D5 were purchased from Children’s Hospital of Oakland Research Institute. Mouse whole chromosome paints were purchased from Applied Spectral Imaging (ASI). Images were captured using an Olympus BX61 microscope and a COOL-1300QS camera, and analyzed through Case Data Manager Version 5.5, installed and configured by ASI.

Analysis of allotypically marked Igh alleles in Atm-/- cells by flow cytometry

Lymphocytes from bone marrow, spleen and lymph nodes were stained with the following antibodies: FITC anti-mouse IgMa (BD Biosciences, clone DS-1), PE anti-mouse IgMb (Pharmingen, AF6-78), PE-Cy anti-B220 (Pharmingen, 552772), FITC mouse IgG1 isotype control (Pharmingen, MOPC-31C), and PE mouse IgG1 isotype control (Pharmingen, MOPC-31C). Samples were RBC-depleted using NH4Cl lysis buffer and stained in PBS containing 0.5% BSA. Live cells were gated on the basis of forward/side scatter and DAPI (Invitrogen, D1306) exclusion. Data was collected on an LSRII and analyzed with FlowJo.

Analysis of RSS breaks in sorted pro- and pre-B cells

Ligation Mediated PCR (LM-PCR) analysis of dsDNA breaks originating from the Igh locus was performed on sorted pro- and pre-B cell DNA samples as previously described 49,50.