Defects in coding joint formation in vivo in developing ATM-deficient B and T lymphocytes

Deficiencies in proteins required for normal coding or signal joint formation should result in an increased accumulation of unrepaired CEs or SEs, respectively, in developing lymphocytes. In Atm^−/− mice, the most profound block in lymphocyte development is at the CD4⁺:CD8⁺ (double positive [DP]) stage of thymocyte development, where TCRα chain genes are assembled and expressed (12). To determine whether ATM functions during TCRα chain gene coding and/or signal joint formation, wild-type (Atm^+/+) and Atm^−/− DP thymocytes were purified by flow cytometric cell sorting, and genomic DNA from these cells was assayed for Jα56 CEs and SEs by LMPCR (Fig. 1, A and B). To optimize for CE detection, genomic DNA was treated with a DNA polymerase to blunt DNA ends before linker ligation. Whereas the analyses of Atm^+/+ and Atm^−/− DP thymocytes revealed nearly equivalent levels of Jα56 SEs, considerably higher levels of Jα56 CEs were detected in Atm^−/− DP thymocytes (Fig. 1 B). Thus, in agreement with our findings in AMuLV pre–B cells, ATM deficiency leads to coding joint, but not signal joint, defects during TCRα chain gene assembly, as indicated by the increased accumulation of unrepaired Jα56 CEs in ATM-deficient DP thymocytes.

To determine whether ATM deficiency leads to coding joint defects at other TCR loci, we assayed for TCRβ and δ SEs and CEs in Atm^+/+ and Atm^−/− thymocytes. TCRβ and δ chain genes are assembled in CD4⁻:CD8⁻ (double negative [DN]) thymocytes that express CD25 (29). CD25⁺ DN thymocytes were purified from Atm^+/+ and Atm^−/− mice by flow cytometric cell sorting, and genomic DNA from these cells was assayed for Jβ1.1, Jβ1.2, and Jδ1 CEs and SEs by LMPCR (Fig. 1, C–E). As both the TCRβ and δ loci have D gene segments, these analyses will also detect CEs generated by cleavage at DJβ1.1, DJβ1.2, and DJδ1 rearrangements (Fig. 1, C, D, F, and G).

Atm^+/+ and Atm^−/− CD25⁺ DN thymocytes had nearly equivalent levels of Jδ1 SEs (Fig. 1 E). These cells also had near equivalent levels of Jβ1.1, Jβ1.2, and 3′Dβ1 SEs (detected by the LMPCR for Jβ1 CEs; Fig. 1, C–E). However, as was observed for the TCRα locus, there were higher levels of unrepaired Jβ1.1, Jβ1.2, and Jδ1 CEs in Atm^−/− DN thymocytes as compared with Atm^+/+ DN thymocytes (Fig. 1 E). Sequence analyses revealed that the Jβ1.1 and Jδ1 LMPCR products were heterogeneous, and several included Dβ and Dδ gene segment nucleotides, respectively (Fig. 1, F and G). Collectively, these data demonstrate that ATM deficiency leads to broad defects in coding joint formation during different steps (V to J, D to J, and V to DJ) of TCR gene assembly at multiple loci (TCRα, β, and δ) in developing thymocytes.

To determine whether ATM functions during coding and/or signal joint formation at Ig loci in developing B cells, IgH and IgLκ chain gene CEs and SEs were assayed by LMPCR of genomic DNA from Atm^+/+ and Atm^−/− bone marrow (Fig. 2).

As was observed for TCR gene rearrangement, analyses of Atm^+/+ and Atm^−/− bone marrow revealed similar levels of JH1 SEs at the IgH locus and Jκ1 SEs at the IgLκ locus (Fig. 2 C). However, considerably higher levels of JH1 and Jκ1 CEs were detected in Atm^−/− bone marrow (Fig. 2 C). Sequence analyses demonstrated that the LMPCR products were heterogeneous and were generated by linker ligation to DJH1, JH1, and Jκ1 CEs (unpublished data). Collectively, these data demonstrate that developing ATM-deficient B cells also have defects in coding joint formation, as indicated by an increased accumulation of unrepaired CEs during DH to JH and VH to DJH rearrangements at the IgH locus and Vκ to Jκ rearrangements at the IgLκ locus.

The LMPCR analyses demonstrate that unrepaired CEs exist at higher levels in developing Atm^−/− lymphocytes. However, this type of analysis cannot be used to determine the fraction of alleles that have unrepaired CEs, which is needed to assess the potential developmental impact of coding joint defects in ATM-deficient mice. As most antigen receptor loci contain many gene segments, it would be difficult to quantitatively assay the total accumulation of unrepaired CEs at a specific locus. Accordingly, we generated Atm^+/+ and Atm^−/− mice that were homozygous for the TCRα^sJ allele (TCRα^sJ/sJ:Atm^+/+ and TCRα^sJ/sJ:Atm^−/− mice, respectively; Fig. 3 A) (30). The wild-type TCRα locus (TCRα⁺) has ∼100 Vα and 61 Jα gene segments, and all Vα to Jα rearrangements occur by deletion (31). The TCRα^sJ allele is identical to the wild-type TCRα locus except that the 61 Jα gene segments have been replaced with 2 Jα gene segments (Jα61 and Jα56) through a multistep gene-targeting approach (Fig. 3 A) (30). Thus, rearrangement of any of the ∼100 Vα gene segments on the TCRα^sJ allele must occur to 1 of the 2 closely linked (0.4 kb) Jα gene segments.

Southern blot analyses of TCRα signal and coding joints in TCRα^sJ/sJ thymocytes should reveal heterogeneously sized fragments, as any 1 of the ∼100 Vα gene segments can rearrange to either of the 2 Jα gene segments in each individual cell. However, as all of these rearrangements will involve the generation of a SE and CE at one of the two closely linked Jα gene segments, we reasoned that these DNA ends may be detected as discrete fragments by Southern blotting.

Southern blot analyses of TCRα^sJ/sJ:Atm^+/+ and TCRα^sJ/sJ:Atm^−/− thymocyte DNA digested with two restriction enzymes and hybridized to probe 8 (P8; Fig. 3 B) upstream of Jα61 revealed fragments of the expected size for germline TCRα^sJ alleles and many different-sized fragments generated by heterogeneous signal joint formation (Fig. 3 C). Importantly, prominent fragments of the expected size for Jα61 SEs were also observed at equal intensities in TCRα^sJ/sJ:Atm^+/+ and TCRα^sJ/sJ:Atm^−/− thymocyte DNA (Fig. 3 C). These fragments were sensitive to exonuclease V digestion, further demonstrating that they likely represent unrepaired SEs (Fig. 3, D and E). Notably, prominent fragments of the expected size for Jα56 SEs were not detected (Fig. 3 C). This would be expected if, like the wild-type TCRα locus, rearrangement of the TCRα^sJ allele is ordered, with the more 5′ Jα gene segment (Jα61) being used before the 3′ Jα gene segment (Jα56) (32–34). In this regard, P8 would not detect Jα56 SEs on TCRα^sJ alleles that had undergone a Vα to Jα61 rearrangement, which deletes the P8-hybridizing region from the chromosome (Fig. 3 B). LMPCR analyses, however, confirmed that Jα56 SEs are present at nearly equivalent levels in TCRα^sJ/sJ:Atm^+/+ and TCRα^sJ/sJ:Atm^−/− thymocytes (Fig. 3 F).

Jα56 and Jα61 CEs were assayed by Southern blot analysis using two different restriction enzymes and a probe (CαI; Fig. 4 A) downstream of the Jα56 gene segment (Fig. 4 A). In addition to fragments expected for heterogeneous VJα coding joints and germline TCRα^sJ alleles, these analyses revealed a novel fragment of expected size for unrepaired Jα56 CEs in TCRα^sJ/sJ:Atm^−/− thymocyte DNA (Fig. 4, A and B). This fragment was not present in TCRα^sJ/sJ:Atm^+/+ thymocyte DNA (Fig. 4 B). Furthermore, hybridization with SJA probe 6 (PS6; Fig. 4 A) to a region between the Jα56 and Jα61 gene segments revealed a fragment of expected size for Jα61 CEs only in TCRα^sJ/sJ:Atm^−/− thymocytes (Fig. 4, A and B). Quantification of these novel fragments, as described in Materials and methods, revealed that 17% (8% Jα56 and 9% Jα61) of TCRα^sJ alleles in TCRα^sJ/sJ:Atm^−/− thymocytes have unrepaired CEs. LMPCR analyses confirmed the increased accumulation of Jα56 and Jα61 CEs in TCRα^sJ/sJ:Atm^−/− thymocytes (Fig. 4 C).

Treatment of genomic DNA with exonuclease V led to a 40% loss in hybridization of the Jα56 CE fragment (Fig. 4, D and E). Treatment of genomic DNA with mung bean nuclease, which opens hairpin-sealed CEs, before exonuclease V treatment led to a nearly complete loss of hybridization (Fig. 4, F and G). Thus, approximately half of the unrepaired CEs in ATM-deficient thymocytes are hairpin sealed, raising the possibility that ATM may have some function in the hairpin opening process. Collectively, these data demonstrate that a substantial fraction of TCRα^sJ/sJ:Atm^−/− thymocytes have unrepaired Jα CEs that exist in either hairpin-sealed or opened configurations.

ATM-deficient mice have increased numbers of nonmalignant αβ T cells with karyotypic abnormalities involving chromosome 14, which contains the TCRα/δ locus (11, 18). Our analyses of V(D)J recombination in Atm^−/− AMuLV pre–B cells demonstrated that unrepaired CEs are frequently resolved aberrantly as translocations or large chromosomal deletions or inversions (28). In DP thymocytes, each TCRα allele usually undergoes several Vα to Jα rearrangements (see Discussion). Thus, the increased frequency of chromosome 14 aberrations observed in Atm^−/− αβ T cells could be due, in part, to the cumulative risk that any one of these multiple rearrangements would be aberrantly resolved.

To investigate this possibility, cytogenetic analyses were performed on metaphases from proliferating TCRα^+/+:Atm^−/− and TCRα^sJ/sJ:Atm^−/− αβ T cells, as described in Materials and methods. As discussed above, the TCRα^sJ allele can undergo only two Vα to Jα rearrangements, whereas the wild-type TCRα allele has the potential to undergo many more Vα to Jα rearrangements. Metaphases were hybridized with red whole chromosome paint (WCP) for chromosome 14 and green WCP for chromosome 15, which serves as a control as it does not contain antigen receptor loci. Chromosome 14 or 15 aberrations were not observed in TCRα^+/+:Atm^+/+ or TCRα^sJ/sJ:Atm^+/+ αβ T cells (Fig. 5). In close agreement with published studies, 8% of TCRα^+/+:Atm^−/− αβ T cells had chromosome 14 aberrations (Fig. 5) (18). In contrast, only 4% of TCRα^sJ/sJ:Atm^−/− αβ T cells had chromosome 14 aberrations (Fig. 5). Notably, the chromosome 14 aberrations in TCRα^+/+:Atm^−/− and TCRα^sJ/sJ:Atm^−/− αβ T cells were equally divided between translocations and replicated chromosomal breaks (Fig. 5 C). From these analyses, it is not possible to determine whether the observed chromosome 14 aberrations are derived from TCRα or TCRδ chain gene rearrangements. However, the TCRδ locus is unaltered on the TCRα^sJ allele, making it unlikely that the change in the frequency of chromosome 14 aberrations is caused by differences in the TCRδ locus–derived translocations. Collectively, these findings demonstrate that ATM-deficient mice with TCRα alleles that undergo fewer Vα to Jα rearrangements have fewer peripheral αβ T cells with chromosome 14 aberrations. In addition, they suggest that RAG-induced DSBs generated in ATM-deficient thymocytes can persist unrepaired, as indicated by the chromosome 14 breaks in mature ATM-deficient αβ T cells.

In this paper, we show that ATM deficiency leads to broad defects in coding joint formation in vivo, as indicated by the increased accumulation of unrepaired CEs at Ig and TCR loci in developing ATM-deficient B and T lymphocytes. In contrast, unrepaired SEs were found at nearly equivalent levels when comparing wild-type and ATM-deficient lymphocytes, demonstrating that ATM does not perform an essential nonredundant function during signal joint formation.

The structure of most antigen receptor loci prohibits the simple quantitative assessment of the fraction of loci with unrepaired CEs or SEs. However, through Southern blot analysis of ATM-deficient thymocytes with a modified TCRα locus (TCRα^sJ), we show that 17% of alleles have unrepaired Jα CEs (Fig. 4 B). That 2% of peripheral TCRα^sJ/sJ:Atm^−/− αβ T cells have chromosome 14 breaks suggests that at least some Jα CEs generated in ATM-deficient DP thymocytes can persist unrepaired for the period of time it takes these cells to be positively selected and released into the periphery (Fig. 5). These findings are in remarkable agreement with analyses of ATM-deficient AMuLV pre–B cells, which revealed that 10–20% of retroviral recombination substrates develop persistent unrepaired CEs after induction of V(D)J recombination (28). Several lines of evidence, including interphase fluorescent in situ hybridization, revealed that unrepaired CEs accumulate in these cells due, at least in part, to their loss from postcleavage complexes (28). Although it is not possible to perform similar types of analyses on developing lymphocytes, we suspect that the loss of CEs from postcleavage complexes also contributes to the coding joint defect observed in developing ATM-deficient lymphocytes in vivo.

Cytogenetic analyses revealed that 2% of proliferating TCRα^sJ/sJ:Atm^−/− αβ T cells have chromosome 14 translocations, suggesting that some Jα CEs are aberrantly resolved (Fig. 5). In agreement, cytogenetic analyses of ATM-deficient AMuLV pre–B cells that had undergone recombination revealed that 1.5% had chromosomal translocations involving the retroviral recombination substrate CEs (28). However, Southern blot analysis of clones revealed that 14% of the retroviral substrates had aberrantly resolved CEs (28). This difference (1.5% by cytogenetic and 14% by clonal analysis) is caused by the inability of cytogenetic approaches to detect many chromosomal deletions and inversions involving the retroviral substrate (28). By analogy, we would expect that the true fraction of TCRα^sJ/sJ:Atm^−/− αβ T cells with aberrant TCRα^sJ allele rearrangements is likely greater than the 2% with translocations detected using cytogenetic approaches (Fig. 5).

Defects in coding joint formation were found at both Ig and TCR loci and during D to J and V to DJ rearrangements in developing B and T cells, yet the lymphopenia of ATM deficiency is more profound in the T cell than the B cell compartment (15). It is possible that distinct features of TCR gene assembly contribute to the more pronounced T lymphopenia of ATM deficiency. The generation of translocations or large chromosomal deletions or inversions involving an antigen receptor locus would likely prevent subsequent rearrangements. Inactivation of a single allele in this manner would require the cell to generate an in-frame rearrangement on the alternate allele if it is to continue development, whereas inactivation of both alleles would preclude any further development. The developmental impact of such a defect should be greatest for loci that must undergo multiple rearrangements, as each additional rearrangement increases the possibility of inactivating the allele. In this regard, it is notable that in ATM-deficient mice the most profound block in T cell development occurs at the DP thymocyte stage, where TCRα chain genes are being assembled and expressed (12).

Developing T cells must assemble TCRα chain genes that are in frame and encode TCRα chains that form nonautoreactive αβ TCRs capable of positive selection. Most randomly generated αβ TCRs are not capable of mediating positive selection (35–37). The large Vα and Jα gene segment clusters are positioned such that all Vα to Jα rearrangements occur by deletion (31). Thus, VJα rearrangements that are not in frame, or that are in frame but do not encode a TCRα chain that forms a selectable αβ TCR, can be replaced through rearrangement of an upstream Vα to a downstream Jα gene segment on the same allele. In addition, TCRα chains that form autoreactive αβ TCRs can, in some instances, be similarly replaced, rescuing the cell from negative selection (38–40). These revision rearrangements proceed in an orderly fashion using progressively more 3′ Jα gene segments in the locus (32, 33). That most developing DP thymocytes must undergo multiple Vα to Jα rearrangements on each allele, as they attempt to generate a selectable αβ TCR, is supported by the defect in positive selection observed in TCRα^sJ/sJ mice (30).

A recent modeling study predicted that, on average, each TCRα allele undergoes five Vα to Jα rearrangements in developing thymocytes (41). In contrast, the TCRα^sJ allele can undergo only two Vα to Jα rearrangements. In this regard, it is notable that TCRα^+/+:Atm^−/− αβ T cells have twice as many chromosome 14 breaks and translocations as compared with TCRα^sJ/sJ:Atm^−/− αβ T cells (Fig. 5). Thus, there is a close correlation between the frequency of chromosome 14 breaks and translocations in ATM-deficient αβ T cells and the potential number of Vα to Jα rearrangements that these cells can make during thymocyte development. Collectively, these findings suggest that, in ATM-deficient cells, loci that undergo multiple rearrangements have an increased probability of generating a persistent unrepaired DSB or of sustaining an aberrant rearrangement, both of which should inactivate the allele and prevent further rearrangements.

What is the developmental impact of TCRα locus inactivation in ATM-deficient thymocytes? Although it is not possible to answer this question with certainty, an estimate of the potential developmental impact can be made. In ATM-deficient AMuLV pre–B cells, ∼14% of rearrangements lead to CEs that are aberrantly resolved in ways (translocations, deletions, and inversions) that would inactivate an antigen receptor locus (28). If 14% of TCRα gene rearrangements are also resolved aberrantly and an average of five Vα to Jα rearrangements occur on each TCRα allele, then on average there would be a 53% (1 − 0.86⁵) possibility that a single TCRα allele would be inactivated in developing ATM-deficient DP thymocytes. Furthermore, approximately one third (0.53 × 0.53 = 0.28) of ATM-deficient DP thymocytes would be expected to inactivate both TCRα alleles, precluding further development of these cells.

In addition to unrepaired TCRα CEs, unrepaired TCRβ chain gene CEs are also found in ATM-deficient thymocytes (Fig. 1 E). Thus, defects in TCRβ chain gene assembly may also contribute to the T lymphopenia of ATM deficiency. However, the only constraint on TCRβ chain gene assembly is that it must be in frame and encode a TCRβ chain that can form a pre-TCR (42). As a result, it is possible that developing thymocytes may undergo fewer TCRβ chain gene rearrangements than TCRα chain gene rearrangements. Moreover, the developmental impact of defects in TCRβ chain gene assembly may be blunted by the cellular expansion that occurs during the DN to DP transition (42). In contrast, positive selection is not accompanied by a substantial cellular expansion that could compensate for defects in TCRα chain gene assembly in ATM-deficient thymocytes.

Defects in coding joint formation during IgH and IgLκ chain gene assembly were also found in developing ATM-deficient B cells (Fig. 2 C), yet B cell development is minimally compromised in ATM-deficient mice (15). Like TCRβ chain gene assembly, IgH chain genes need only to be in frame and encode an IgH chain that can form a pre–B cell receptor (BCR) (43, 44). Furthermore, the pro– to pre–B cell transition is accompanied by a cellular expansion that could compensate for defects in IgH chain gene assembly (43, 44). IgL chain genes are assembled in pre–B cells, and they must encode an IgL chain that forms a nonautoreactive BCR (43, 44). As is the case for TCRα chain gene assembly, IgL chain gene assembly is not followed by a substantial cellular expansion that could compensate for defects in this process (43, 44). Like the TCRα locus, the structure of the IgLκ locus permits VJκ rearrangements to be replaced with new VJκ rearrangements, on the same allele, as pre–B cells attempt to generate a nonautoreactive BCR (45). However, unlike the αβ TCR, there is no requirement for positive selection of the BCR. Thus, the additional constraint of positive selection may mandate more TCRα chain gene rearrangements in DP thymocytes than IgL chain gene rearrangements in pre–B cells and, as such, contribute to the more profound T cell lymphopenia of ATM deficiency.

It is also possible that developing T cells may have unique requirements for ATM function in addition to its role in antigen receptor gene assembly. For example, recent studies in macrophages have demonstrated that ATM deficiency leads to constitutive c-Jun N-terminal kinase (JNK) activation (46). JNK signals are involved in negative selection; therefore, if JNK activity is also perturbed in ATM-deficient thymocytes, this may contribute to the T lymphopenia of ATM deficiency (47). In addition, like ATM-deficient mice, mice that are deficient in both ATM and RAG eventually succumb to thymic lymphomas (21). However, these lymphomas do not have translocations involving the TCRα/δ locus (21). This demonstrates that ATM deficiency can promote thymocyte transformation through defects in processes other than the assembly of antigen receptor genes. Therefore, although we demonstrate unequivocally that ATM-deficient lymphocytes have defects in V(D)J recombination, the relative contribution of these and other defects to the lymphoid phenotypes of A-T remains to be determined.

All mice were bred and maintained under specific pathogen-free conditions at the Washington University School of Medicine and were handled in accordance with the guidelines set forth by the Division of Comparative Medicine of Washington University.

Southern blot analyses of genomic DNA and PCR products were performed as previously described (30, 48). P8, the CαI probe, and the RAG-2 probe (PR2) have been previously described (49). PS6 is a 0.4-kb PCR fragment amplified from TCRα^sJ/sJ genomic DNA using the PS6-1 and PS6-2 oligonucleotides (Table S1, available). For exonuclease V digestion, 15 μg of thymocyte DNA was treated with increasing concentrations of exonuclease V, using the manufacturer's buffer conditions (USB Corporation), for 1 h at 37°C before restriction enzyme digestion and Southern blot analyses. For mung bean nuclease assays, 15 μg of thymocyte DNA was digested with 12.5 U of mung bean nuclease (New England Biolabs, Inc.) in 200 μl using the manufacturer's buffer supplemented with 0.4 mM ZnSO₄ at 25°C for 1 h before treatment with exonuclease V, as described in this section.

Quantifications of Southern blot fragments were performed on a PhosphorImager (Storm 840; GE Healthcare) using ImageQuant software (GE Healthcare). The percentage of TCRα^sJ alleles with unrepaired Jα56 CEs was calculated from the Southern blot of StuI-digested DNA probed with the CαI probe (Fig. 4 B) or PR2 as a DNA loading control (Fig. 4 B).

The following formula was used to determine the percentage of TCRα^sJ allele Jα56 CEs:

The hybridization intensity of the CαI probe hybridizing the Jα56 CE fragment (Jα56CE^Thy) and PR2 hybridizing the RAG-2 gene fragment (R2^Thy) from TCRα^sJ/sJ:Atm^−/− thymus DNA is used. In addition, the hybridization intensity of the CαI probe hybridizing the germline TCRα^sJ allele fragment (CαI^Kid) and PR2 hybridizing the RAG-2 gene fragment (R2^Kid) from TCRα^sJ/sJ:Atm^−/− kidney DNA is used.

The percentage of TCRα^sJ alleles with unrepaired Jα61 CEs was similarly calculated from Southern blotting of StuI-digested DNA probed with PS6 (Fig. 4 B) using the following formula:

The hybridization intensity of PS6 hybridizing the Jα61 CE fragment (Jα61CE^Thy) from TCRα^sJ/sJ:Atm^−/− thymus DNA and the hybridization intensity of PS6 hybridizing the germline TCRα^sJ allele fragment (PS6^Kid) from TCRα^sJ/sJ:Atm^−/− kidney DNA are used. R2^Thy and R2^Kid are as described in this section.

The flowing formulas were used for quantifying the retention of Jα61 SEs and Jα56 CEs, respectively, after exonuclease V or mung bean nuclease and exonuclease V treatment:

and

The intensities of the Jα56 CE (Jα56CE), Jα61 SE (Jα61SE), or TCRα^sJ germline fragments in lanes that were either not treated (^−ExoV) or treated with different concentrations of exonuclease V (^{+ ExoV}) are used (Fig. 3 D; and Fig. 4, D and F).

LMPCR was performed as previously described (48). In brief, 4 μg of thymus DNA that was treated with the Klenow DNA polymerase (New England Biolabs, Inc.) in the presence of dNTP before ligation to the BW or ANC linker (48, 50). Heminested PCR was performed with Taq polymerase (Roche Molecular Systems, Inc.), according to the manufacturer's recommendation, using 1 mM MgCl₂. All primary PCRs were performed at 94°C for 5 min, followed by 17 cycles of 94°C for 30 s, 57°C for 30 s, and 72°C for 60 s. This was followed by a final incubation at 72°C for 7 min. All secondary PCRs were performed at 94°C for 5 min, followed by 23–36 cycles (Table S1) at 94°C for 30 s, 58°C for 30 s, and 72°C for 60 s. This was followed by a final incubation at 72°C for 7 min. Primer sequences are provided in Table S1. Oligonucleotides used as probes for PCR products are also listed in Table S1. IL-2 control PCR was performed as previously described (32).

To purify CD25⁺ DN thymocytes, total thymocytes were depleted with CD4- or CD8-expressing cells using magnetic beads, according to the manufacturer's instruction (Dynal), followed by flow cytometric purification of CD25⁺ DN thymocytes (FACSDiva; Beckton Dickinson) using PE-Cy7–conjugated anti-CD25 (BD Biosciences). The resulting cells were >95% CD25⁺ DN thymocytes. DP thymocytes were purified to >95% purity by flow cytometric cell sorting using FITC-conjugated anti-CD8 and PE-Cy7–conjugated anti-CD4 (BD Biosciences).

Metaphases were prepared from ConA-stimulated peripheral T cells, as previously described (28). Fluorescent in situ hybridization of metaphase chromosomes using the mouse WCPs (Applied Spectral Imaging) for chromosomes 14 (red) and 15 (green) were performed using the manufacturer's recommended procedure. In brief, denatured probes were allowed to reanneal at 37°C for 20 min and were hybridized to denatured chromosomes for 16 h in a humidified chamber at 37°C. Stringency washing was done in 0.4× SSC at 70°C for 4 min, followed by 4× SSC/0.1% Tween 20. DAPI-counterstained slides were mounted in antiquench mount (Vectashield; Vector Laboratories) and analyzed on an epifluorescence microscope (Axioplan 2ie; Carl Zeiss MicroImaging, Inc.). Image acquisition and processing was done using ISIS software (Metasystems).

Table S1 presents a list of oligonucleotides used for LMPCR and probes. For each SE or CE LMPCR, the allele-specific oligonucleotides used as PCR primers or Southern probes are grouped with their names and sequences indicated. For TCRα locus LMPCR, the primers used for detecting SEs and CEs in the wild-type TCRα locus (TCRα⁺) or TCRα^sJ locus (TCRα^sJ) are grouped, respectively. The cycle numbers for the primary and secondary reactions of the heminested PCR are also indicated alongside the allele-specific primers that were used. Primers A, B, and C are linker-specific primers. PS6-1 and PS6-2 are used to amplify PS6 for thymocyte DNA Southern analyses.

We thank Drs. Andrea Bredemeyer and Michael Krangel, and Beth Helmink for critical review of the manuscript.

B.P. Sleckman is supported by National Institutes of Health (NIH) grants AI47829 and AI49934 and American Cancer Society grant RSG-05-070-01-LIB. T.K. Pandita is supported by NIH grants NS34746 and CA10445. C.H. Bassing is a Pew Scholar. C.-Y. Huang is supported by a postdoctoral training grant from the NIH. Mice were housed in a facility supported by National Center for Research Resources grant RR012466.

The authors have no conflicting financial interests.

Note added in proof. Since the acceptance of this manuscript, another paper has appeared that also reports an increase in TCRα CEs in ATM-deficient mice (Vacchio, M.S., A. Olaru, F. Livak, and R.J. Hodes. 2007. Proc. Natl. Acad. Sci. USA. 104:6323–6328).