## Abstract

The ataxia telangiectasia mutated (ATM) kinase and H2AX histone tumor suppressor proteins are each critical for maintenance of cellular genomic stability and suppression of lymphomas harboring clonal translocations. ATM is the predominant kinase that phosphorylates H2AX in chromatin around DNA double-strand breaks, including along lymphocyte Ag receptor loci cleaved during V(D)J recombination. However, combined germline inactivation of *Atm* and *H2ax* in mice causes early embryonic lethality associated with substantial cellular genomic instability, indicating that ATM and H2AX exhibit nonredundant functions in embryonic cells. To evaluate potential nonredundant roles of ATM and H2AX in somatic cells, we generated and analyzed *Atm*-deficient mice with conditional deletion of *H2ax* in αβ T-lineage lymphocytes. Combined *Atm*/*H2ax* inactivation starting in early-stage CD4^{–}/CD8^{–} thymocytes resulted in lower numbers of later-stage CD4^{+}/CD8^{+} thymocytes, but led to no discernible V(D)J recombination defect in G1 phase cells beyond that observed in *Atm*-deficient cells. *H2ax* deletion in *Atm*-deficient thymocytes also did not affect the incidence or mortality of mice from thymic lymphomas with clonal chromosome 14 (TCRα/δ) translocations. Yet, in vitro-stimulated *Atm/H2ax*-deficient splenic αβ T cells exhibited a higher frequency of genomic instability, including radial chromosome translocations and TCRβ translocations, compared with cells lacking *Atm* or *H2ax*. Collectively, our data demonstrate that both redundant and nonredundant functions of ATM and H2AX are required for normal recombination of TCR loci, proliferative expansion of developing thymocytes, and maintenance of genomic stability in cycling αβ T-lineage cells.

## Introduction

Conserved features of the cellular DNA damage response (DDR) to double-strand breaks (DSBs) are activation of the ataxia telangiectasia mutated (ATM) kinase and its phosphorylation of H2AX in chromatin around breakage sites. ATM-mediated phosphorylation of H2AX enables subsequent histone modifications that nucleate assembly of other DDR proteins into complexes around DSBs (1). ATM-mediated phosphorylation of these and other substrates including the p53 tumor suppressor activates cell cycle checkpoints, promotes DSB repair, and induces apoptosis if DNA damage is too severe (2). Cellular ATM deficiency leads to increased genomic instability, impaired cell cycle checkpoints, and defective p53-mediated apoptosis (3, 4), whereas germline ATM inactivation causes highly elevated predisposition to lymphoid cancers with clonal translocations (5–10). Cellular H2AX deficiency leads to increased genomic instability and a defect in the G2/M checkpoint but does not alter p53-mediated apoptosis (11–15), whereas germline H2AX inactivation only causes a slightly increased cancer predisposition (12). These findings indicate that ATM preserves genomic stability and suppresses transformation through H2AX-dependent and H2AX-independent mechanisms.

The ATM-related kinases DNA-PKcs and ATR also phosphorylate H2AX (16, 17), suggesting that H2AX may exhibit ATM-independent DDR functions. Consistent with this notion, combined germline inactivation of *Atm* and *H2ax* caused embryonic lethality and increased genomic instability in embryonic cells compared with *Atm* or *H2ax* deficiency (18). This elevated genomic instability was associated with a requirement for ATM-dependent H2AX functions to repair oxidative DNA damage (18). These *Atm/H2ax*-deficient cells exhibited a slight increase in chromosome breaks compared with *Atm*-deficient cells and a substantial increase in chromatid breaks compared with *H2ax*-deficient cells (18). Considering that chromosome breaks predominantly arise from G1 phase DSBs and chromatid breaks generally derive from S/G2 phase DSBs, the genomic instability pattern of *Atm/H2ax*-deficient embryonic cells suggests that ATM and H2AX have redundant functions in DSB repair through non-homologous end-joining (NHEJ) yet nonredundant functions in homologous recombination (HR). Because DDR mechanisms exhibit tissue-specific differences (19), additional studies are required to assess potential redundant and nonredundant functions of ATM and H2AX in somatic cells.

In mice and humans, αβ and γδ T cells develop in the thymus from common progenitor cells through programs that link the assembly of TCR genes with differentiation (20). TCRβ, TCRγ, and TCRδ genes are assembled from variable (V), diversity (D), and joining (J) gene segments in CD4^{−}/CD8^{−} “double-negative” (DN) thymocytes (20). V(D)J recombination is initiated by the RAG1/RAG2 (RAG) endonuclease, which induces DSBs adjacent to participating segments, and is completed by NHEJ proteins, which repair RAG DSBs to form V(D)J coding exons upstream of constant (C) exons (21). Assembly and expression of functional TCRγ and TCRδ genes generates γδ TCRs that promote differentiation into mature γδ T cells (22). In contrast, assembly and expression of functional TCRβ genes leads to TCRβ/pre-Tα (pre-TCR) complexes and β-selection signals that drive proliferative expansion as cells differentiate into CD4^{+}/CD8^{+} “double-positive” (DP) thymocytes (20). This TCRβ-dependent proliferation is associated with spontaneous replication-associated DSBs that must be repaired by HR for normal DN-to-DP thymocyte expansion (23). Assembly and expression of functional TCRα genes in DP cells leads to formation of αβ TCRs, which upon their selection signal differentiation into CD4^{+} or CD8^{+} “single-positive” (SP) thymocytes that exit the thymus as CD4^{+} or CD8^{+} mature αβ T cells (20).

In developing αβ T cells, ATM and H2AX each suppress oncogenic translocations arising from aberrant repair of DSBs during NHEJ and HR. Germline Atm-deficient (*Atm*^{–}^{/}^{–}) and H2ax-deficient (*H2ax*^{–}^{/}^{–}) mice each exhibit fewer numbers of thymocytes and αβ T cells due to impaired proliferation of cells lacking Atm or H2ax (6, 9, 10, 12–14, 24). *Atm*^{–}^{/}^{–}, but not *H2ax*^{–}^{/}^{–}, mice exhibit a severe defect in DP-to-SP development due to H2AX-independent functions of ATM during coding join formation in G1 phase cells (6, 9, 10, 13, 14, 24–26). Almost all *Atm*^{–}^{/}^{–} mice succumb to thymic lymphomas with clonal chromosome 14 translocations created through aberrant repair of DSBs induced by RAG during TCRδ recombination and by DNA replication errors during thymocyte expansion (7, 8). In contrast, despite H2AX phosphorylation by ATM and DNA-PKcs along RAG-cleaved Ag receptor loci (27), *H2ax*^{–}^{/}^{–} mice rarely succumb to these or other tumors due to ATM/p53-dependent activation of checkpoints and apoptosis (12–14).

To investigate potential redundant and nonredundant DDR functions of ATM and H2AX in somatic cells, we generated and analyzed *Atm*^{–}^{/}^{–} mice with conditional deletion of *H2ax* in developing αβ T lymphocytes. We show that combined *Atm/H2ax* deletion starting in DN thymocytes results in lower numbers of DP thymocytes without causing a V(D)J recombination defect beyond that observed in *Atm*-deficient cells. We also show that *H2ax* deletion in *Atm*-deficient thymocytes does not affect the incidence or mortality of mice from thymic lymphomas with clonal chromosome 14 translocations. Yet, we find that proliferating *Atm/H2ax*-deficient αβ T cells exhibit a higher frequency of genomic instability, including radial chromosome translocations and TCRβ translocations, relative to cells lacking *Atm* or *H2ax*. Collectively, our data demonstrate that both redundant and nonredundant functions of ATM and H2AX are required for normal TCR recombination, proliferative expansion of developing thymocytes, and maintenance of genomic stability in cycling αβ T-lineage cells.

## Materials and Methods

### Mice

*Lck-*cre transgenic (28), *Atm*^{–}^{/}^{–} (24), and *H2ax ^{F/F}* (12) mice were bred to generate the animals in this study. The background strain of these mice was mixed 129SvEv and C57BL/6, with the 129SvEv background predominant. PCR analyses of

*H2ax*deletion were performed as described (12), demonstrating complete deletion of

*H2ax*in

*Lck-*cre

*Atm*

^{–}

^{/}^{–}and

*Lck-*cre

*Atm*

^{–}

^{/}^{–}

*H2ax*αβ T-lineage cells. Experiments were conducted on 4- to 6-wk-old mice of each genotype, performed in accordance with national guidelines, and approved by the Institutional Animal Care and Use Committee of the Children's Hospital of Philadelphia.

^{flox/flox}### Flow cytometry

Single-cell suspensions from thymuses and spleens of 4- to 6-wk-old mice or tumors were stained with Abs in PBS with 2% FBS. For analysis of CD4 and CD8 expression, single-cell suspensions of thymocytes and splenocytes were stained with FITC-conjugated anti-TCRβ, PE-conjugated anti-CD8, and allophycocyanin-conjugated anti-CD4 (BD Pharmingen). Data were collected using a FACSCalibur (BD Biosciences, San Jose, CA) and CellQuest software (BD Biosciences) and analyzed using FlowJo software (Tree Star).

### Chromosomal V(D)J recombination assay

Bone marrow from 3- to 5-wk-old *Atm*^{−/−}*H2ax ^{F/F}* mice harboring the

*Eμ*-Bcl-2 transgene were cultured and infected with the pMSCV v-abl retrovirus to generate Abelson transformed (abl) pre-B cell lines as previously described (25). Generation of the

*H2ax*

^{−/−}abl pre-B cells was previously described (29).

*H2ax*

^{−/−}and

*Atm*

^{−/−}

*H2ax*abl pre-B cells containing pMX-DEL

^{F/F}^{CJ}retroviral recombination substrates were made as described (25).

*Atm*

^{−/−}

*H2ax*

^{−/−}abl pre-B cells containing pMX-DEL

^{CJ}were created through Tat-Cre induction of

*Atm*

^{−/−}

*H2ax*:pMX-DEL

^{F/F}^{CJ}abl pre-B cells as previously described (29). Southern blot analysis was performed upon abl pre-B cells after treatment with 3 μM STI571 for the indicated times at a density of 10

^{6}cells/ml, as previously described (25).

### Southern blotting

Genomic DNA (20–30 μg) of thymocytes, tumors, or kidneys was digested with 100 U of the indicated enzymes (New England Biolabs), separated on a 1.0% TAE agarose gel, transferred onto ζ-probe membrane (Bio-Rad), and hybridized with ^{32}P-labeled DNA probes as described (12).

### Cytogenetics

Metaphase spreads were prepared as described. Spectral karyotyping and chromosome painting were performed according to the manufacturer’s instructions (Applied Spectral Imaging). BAC fluorescence in situ hybridization (FISH) probes were labeled with biotin (Biotin-Nick Translation Mix; Roche). The TCR-Cβ-164G11 BAC used for FISH has been described (30). Images were captured and analyzed using Case Data Manager version 5.5 configured by Applied Spectral Imaging.

### Statistical analysis

The Kaplan–Meier curves were generated in SAS version 9 (SAS Institute). Statistical analyses were performed with two-tailed unpaired Student *t* tests in Excel (Microsoft) or Prism (AMPL Software). We considered *p* < 0.05 to be statistically significant.

## Results

### Both ATM and H2AX are required for normal TCRβ-mediated thymocyte expansion

Because the *Lck-cre* transgene promotes deletion of “floxed” *H2ax* (*H2ax ^{F}*) genes in DN cells prior to V(D)J recombination (29), we sought to generate and analyze

*Lck-cre*,

^{+/−}Atm^{−/−}*Lck-cre*, and

^{+/−}H2ax^{F/F}*Lck-cre*mice.

^{+/−}Atm^{−/−}H2ax^{F/F}*Atm*and

*H2ax*are closely linked on chromosome 9, therefore we first bred together

*Atm*

^{+}

^{/}^{–}and

*H2ax*mice to generate

^{F/F}*Atm*

^{+}

^{/}^{–}

*H2ax*males containing the

^{+/F}*Atm*and

^{−}*H2ax*alleles on different chromosomes. These

^{F}*Atm*

^{+}

^{/}^{–}

*H2ax*males were crossed with wild-type females to select for meiotic crossover events that created

^{+/F}*Atm*

^{+}

^{/}^{–}

*H2ax*mice with the

^{F/+}*Atm*

^{–}and

*H2ax*alleles linked on the same chromosome. We frequently observe deletion of

^{F}*H2ax*genes in nonlymphoid cells when

^{F}*Lck-cre*is transmitted maternally, but not when transmitted paternally (data not shown). Thus, to avoid nonspecific

*H2ax*deletion, as well as potential complications due to homozygous

*Lck-cre*transgene integration, we bred heterozygous

*Lck-cre*males with

*Atm*

^{+}

^{/}^{–}

*H2ax*females to generate

^{F/+}*Lck-cre*

^{+/−}Atm^{+}

^{/}^{–}

*H2ax*mice. Because

^{F/+}*Atm*

^{–}

^{/}^{–}mice are infertile (6, 9, 10, 24), we bred

*Lck-cre*

^{+}

^{/}^{–}

*Atm*

^{+}

^{/}^{–}

*H2ax*males with

^{F/+}*Atm*

^{+}

^{/}^{–}

*H2ax*females to generate experimental

^{F/+}*Lck-cre*

^{+/−}Atm^{–}

^{/}^{–}

*H2ax*mice, hereafter referred to as

^{F/F}*LAH*mice. We used a similar breeding strategy to generate control

*Lck-cre*

^{+/−}Atm^{–}

^{/}^{–}and

*Lck-cre*mice, hereafter referred to as

^{+/−}H2ax^{F/F}*LA*and

*LH*mice, respectively.

To assess potential redundant and nonredundant functions of ATM and H2AX in αβ T-lineage cells, we first analyzed the thymuses and spleens of *LA*, *LH*, and *LAH* mice by cell counting and flow cytometry (FACS) analysis with anti-CD4 and anti-CD8 Abs. We found that *LH* and wild-type mice exhibited comparable numbers of total thymocytes, cells within each thymocyte developmental stage, and splenic αβ T cells (data not shown). We detected ∼2-fold lower numbers of total thymocytes and splenic αβ T cells in *LA* mice compared with *LH* mice (Fig. 1). The fewer numbers of *LA* thymocytes reflected an ∼2-fold reduction in DP cells and ∼5-fold reductions in CD4^{+} SP and CD8^{+} SP cells (Fig. 1A, 1B). These data are consistent with the phenotypes of *Atm*^{–}^{/}^{–} and *H2ax*^{–}^{/}^{–} mice (6, 9, 10, 12–14, 24), indicating that *Lck-cre* expression has negligible effects upon the development of αβ T cells lacking Atm or H2ax. We found that the average numbers of thymocytes and splenic αβ T cells in *LAH* mice were each reduced ∼2-fold compared with *LA* mice and ∼4-fold compared with *LH* mice (Fig. 1), although the difference in numbers of *LA* and *LAH* splenic αβ T cells was not significant from the numbers of mice analyzed. Notably, as compared with *LA* mice, *LAH* mice contained an ∼2-fold reduction in DP cell numbers, yet showed no significant differences in DN, CD4^{+} SP, or CD8^{+} SP cell numbers (Fig. 1A, 1B). These observations indicate that TCRβ-dependent DN-to-DP thymocyte expansion is more impaired in *Atm/H2ax*-deficient cells than in *Atm*^{–}^{/}^{–} or *H2ax*^{–}^{/}^{–} cells. This difference may reflect nonredundant ATM and H2AX functions in coding join formation during TCRβ recombination and/or repair of replication-associated DSBs during TCRβ-driven cellular proliferation.

The similar defect in the DP-to-SP thymocyte differentiation step in *LAH* and *LA* mice suggests that combined *Atm*/*H2ax* deletion does not impair coding join formation during TCRα recombination beyond that observed in *Atm*^{–}^{/}^{–} thymocytes. Analysis of chromosomal V(D)J recombination in abl pre-B cells enables distinction between NHEJ defects in G1 phase cells and impaired cellular proliferation after the assembly, expression, and selection of functional Ag receptor genes (25). Therefore, to obtain direct evidence that *H2ax* deletion does not impair coding join formation beyond that observed in *Atm*^{–}^{/}^{–} cells, we sought to analyze chromosomal V(D)J recombination among *H2ax*^{–}^{/}^{–}, *Atm*^{–}^{/}^{–}, and *Atm*^{–}^{/}^{–}*H2ax*^{–}^{/}^{–} abl pre-B cells. For this purpose, we established abl pre-B cell lines from *Atm*^{–}^{/}^{–}*H2ax ^{F/F}* mice and incubated these with TAT-Cre protein to delete

*H2ax*genes, generating otherwise identical

^{F}*Atm*

^{–}

^{/}^{–}

*H2ax*

^{–}

^{/}^{–}abl pre-B cells. We infected the parental

*Atm*

^{–}

^{/}^{–}

*H2ax*and derivative

^{F/F}*Atm*

^{–}

^{/}^{–}

*H2ax*

^{–}

^{/}^{–}abl pre-B cells, as well as previously established

*H2ax*

^{–}

^{/}^{–}abl pre-B cells (29), with the pMX-DEL

^{CJ}retroviral V(D)J recombination substrate (Fig. 2A) (25). Rearrangement of pMX-DEL

^{CJ}in wild-type cells results in formation of a chromosomal coding join (CJ) through coding end (CE) intermediates (Fig. 2A). We treated populations of

*H2ax*

^{–}

^{/}^{–},

*Atm*

^{–}

^{/}^{–}

*H2ax*, and

^{F/F}*Atm*

^{–}

^{/}^{–}

*H2ax*

^{–}

^{/}^{–}abl pre-B cells containing chromosomally integrated pMX-DEL

^{CJ}with STI571 for 48 or 96 h. We then conducted Southern blot analysis to identify and quantify substrates that rearranged to form CJs, accumulated unrepaired CEs, or were unrearranged. Consistent with our previous observations (29), we found that CJ formation was normal in G1 phase

*H2ax*

^{–}

^{/}^{–}cells but defective in G1 phase

*Atm*

^{–}

^{/}^{–}cells due to accumulation of unrepaired CEs (Fig. 2B). Notably, we detected no obvious differences in the ratio of unrepaired CEs to repaired CJs in G1 phase

*Atm*

^{–}

^{/}^{–}

*H2ax*

^{–}

^{/}^{–}cells compared with G1 phase

*Atm*

^{–}

^{/}^{–}cells (Fig. 2B). These results confirm that combined

*Atm*/

*H2ax*inactivation does not impair chromosomal CJ formation in G1 phase cells substantially beyond that detected in

*Atm*

^{–}

^{/}^{–}cells. Collectively, our data indicate that ATM and H2AX do not exhibit redundant functions during NHEJ-mediated repair of RAG DSBs in G1 phase lymphocytes, yet they exhibit overlapping functions during proliferative expansion of TCRβ-selected thymocytes.

*Atm*^{−/−}H2ax^{−/−} and *Atm*^{−/−} thymocytes exhibit similar predisposition to transformation

^{−/−}H2ax

^{−/−}

^{−/−}

Almost all *Atm*^{–}^{/}^{–} mice succumb by 6 mo of age to thymic lymphomas (6, 9, 10, 24), whereas *H2ax*^{–}^{/}^{–} mice rarely succumb to cancer within the first year of life (12–14). Thus, to evaluate potential nonredundant functions of ATM and H2AX in suppression of αβ T-lineage lymphoma, we generated and analyzed parallel cohorts of 20 *LA* and 27 *LAH* mice. Although we did not assess a parallel cohort of *LH* mice, we have never observed any tumors in *LH* mice (data not shown). We found that *LA* and *LAH* mice survived tumor-free between 75 and 145 d with both genotypes exhibiting a median age of mortality around 85 d (Fig. 3A). All 20 *LA* cohort mice succumbed to thymic lymphomas that showed no dissemination to peripheral lymphoid organs (Supplemental Table I), similar to the tumor phenotype of *Atm*^{–}^{/}^{–} mice (6, 9, 10, 24). Of the 27 *LAH* cohort mice, 25 succumbed to thymic lymphomas, one succumbed to a peripheral lymphoma, and another succumbed to a sarcoma (Supplemental Table II). These observations demonstrate that *Atm*^{–}^{/}^{–} mice and *Atm*^{–}^{/}^{–} mice with *H2ax* deletion initiating in DN thymocytes exhibit similar predisposition to spontaneous thymic lymphomas.

Thymic lymphomas of *Atm*^{–}^{/}^{–} mice are clonal malignancies that contain TCRβ rearrangements, lack surface TCRβ expression, and consist of CD4^{+}/CD8^{+}, CD4^{–}/CD8^{+}, and/or CD4^{+}/CD8^{–} cells (6–10, 24). To evaluate whether *H2ax*-deletion affects the phenotype of *Atm*^{–}^{/}^{–} thymic lymphomas, we determined the clonality and developmental stage of *LAH* and *LA* tumors. Most of the 19 *LAH* and 19 *LA* lymphomas analyzed by Southern blotting contained two TCRβ alleles with rearrangements involving Dβ1-Jβ1 segments (Fig. 3B; data not shown). The others contained another rearrangement involving Dβ2-Jβ2 segments and/or deleted TCRβ locus sequences (Fig. 3B; data not shown). Southern blot analysis also confirmed *H2ax* deletion in all *LAH* tumors analyzed (data not shown), demonstrating these cells lacked both Atm and H2ax. Of the 17 *LA* and 21 *LAH* thymic lymphomas analyzed by flow cytometry, most were TCRβ^{–} and similarly composed of CD4^{+}/CD8^{+}, CD4^{–}/CD8^{+}, and/or CD4^{+}/CD8^{–} cells (Supplemental Tables I, II). Collectively, these data reveal that *Atm*^{–}^{/}^{–} mice and *Atm*^{–}^{/}^{–} mice with *H2ax* deletion DN cells succumb to clonal αβ T-lineage tumors of similar developmental stages.

Most *Atm*^{–}^{/}^{–} thymic lymphomas harbor clonal chromosome 14 translocations formed by aberrant repair of RAG DSBs induced during TCRδ recombination in DN thymocytes and additional clonal and nonclonal translocations generated through aberrant repair of replication-associated DSBs (7, 8). To assess whether deletion of *H2ax* in *Atm*^{–}^{/}^{–} DN thymocytes affects the translocation pattern of *Atm*^{–}^{/}^{–} thymic lymphomas, we conducted spectral karyotyping (SKY) on metaphases prepared from four *LA* and five *LAH* tumors. Three *LA* thymic lymphomas contained clonal chromosome 14 translocations, as well as other clonal or nonclonal translocations (Table I); the other harbored clonal t (15, 6), t (12, 7), and t (4, 15) translocations and several nonclonal translocations (Table I). This pattern of translocations is similar to that observed in *Atm*^{–}^{/}^{–} tumors (7, 8), indicating that *Lck-cre* expression does not substantially affect the type or frequency of translocations that arise in *Atm*^{–}^{/}^{–} thymic lymphomas. Four *LAH* tumors harbored clonal chromosome 14 translocations and many nonclonal translocations (Fig. 3C, Table II). The remaining *LAH* thymic lymphoma lacked clonal translocations but contained numerous nonclonal translocations (Fig. 3D, Table II). These findings indicate that *LA* and *LAH* tumors exhibit similar patterns of translocations compared with each other and with *Atm ^{−/−}* thymic lymphomas. Thus, we conclude that conditional

*H2ax*deletion in

*Atm*thymocytes does not substantially affect the incidence or mortality of mice from thymic lymphomas with clonal chromosome 14 translocations.

^{−/−}Tumor . | Clonal Translocations . | Nonclonal Translocations . |
---|---|---|

728 | t (12, 14) | t (12, 7) |

873 | t (12, 14), t (8, 6), and t (X;16) | t (4, 5), t (17, 16), t (9, 4), t (1, 15), t (1, 6), t (7, 14), t (2, 11), t (15, 11), t (17, 14), t (4, 7), t (14, 6), t (17, 9), t (8, 15, 16) |

26 | t (12, 14) and t (14, 18) | t (12, 2), t (14, 1), t (3, 8), t (2, 12), t (9, 15), t (2, 14), t (16, 15), t(X;15), t (8, 14), t (14, 19), t (3, 16), t (3, 7), t (5, 9), t (15, 9), t(X;5), t (19, 3), reciprocal t (1, 12) and t (12, 1) |

560 | t (15, 6), t (4, 15), and t (12, 7) | t (12, 4), t (12, 15), t (17, 15), t (4;X), t (14, 16), t (12, 15), t (15, 8), t (6, 12), t (6, 14), t (7, 15), t (1, 6), t (13, 1), t (15, 15, 6) |

Tumor . | Clonal Translocations . | Nonclonal Translocations . |
---|---|---|

728 | t (12, 14) | t (12, 7) |

873 | t (12, 14), t (8, 6), and t (X;16) | t (4, 5), t (17, 16), t (9, 4), t (1, 15), t (1, 6), t (7, 14), t (2, 11), t (15, 11), t (17, 14), t (4, 7), t (14, 6), t (17, 9), t (8, 15, 16) |

26 | t (12, 14) and t (14, 18) | t (12, 2), t (14, 1), t (3, 8), t (2, 12), t (9, 15), t (2, 14), t (16, 15), t(X;15), t (8, 14), t (14, 19), t (3, 16), t (3, 7), t (5, 9), t (15, 9), t(X;5), t (19, 3), reciprocal t (1, 12) and t (12, 1) |

560 | t (15, 6), t (4, 15), and t (12, 7) | t (12, 4), t (12, 15), t (17, 15), t (4;X), t (14, 16), t (12, 15), t (15, 8), t (6, 12), t (6, 14), t (7, 15), t (1, 6), t (13, 1), t (15, 15, 6) |

Translocations classified as clonal were detected in at least 80% of metaphases.

Tumor . | Clonal Translocations . | Nonclonal Translocations . |
---|---|---|

798 | t (12, 14) and t (14, 15) | t (2, 8), t (2, 15), t (12, 2), t (2, 12), t (2, 15), t (2, 14), reciprocal t (8, 2) and t (2, 8) |

629 | t (12, 14) and t (14, 15) | t (17, 14), t (12, 8), t (5, 14), t (1, 2), t (2, 14), t (2, 1), t (17, 13), t (18, 11), t (13, 19), t (19, 11), reciprocal t (1, 2) and t (2, 1) |

884 | t (12, 14) and t (14, 15) | t (12, 2), t (14, 12), t (1, 2), t (1, 7), t (1, 17), t (8, 7), t (7, 15), reciprocal t (6, 7) and t (7, 6), reciprocal t (4, 10) and t (10, 4) |

590 | t (4, 14), t (14, 16), and t (16, 4, 14) | t (14, 4), t (13, 9), t (16, 4, 14), t (13, 2), t (14, 3), t (4, 8), t (2, 17), t (12, 6), t (12, 8), t (16, 12), t (2, 6), t (19;Y), t (8, 16), t (8, 19), t (8, 14), t (12, 2), t (12, 15), t (12, 14), t (2, 16), t (2, 12), t (8;X) |

589 | None | t (2, 1), t (4, 5), t (12, 2), t (14, 15), t (15, 4), t (4, 15), t (4, 6), t (12, 14), t (5, 4), t (4, 15, 16), t (8, 19), t (1, 17), t (7:15;14), t (17, 19), t (15, 7), t (2, 12), t (12, 2), t (1, 8), t (1, 2), t (8, 1), t (18, 12), t (14;15;Y), t (5, 1), reciprocal t (4, 15) and t (15, 4), reciprocal t (5, 17) and t (17, 5), reciprocal t (6, 18) and t (18, 6), reciprocal t (3, 4) and t (4, 3) |

Tumor . | Clonal Translocations . | Nonclonal Translocations . |
---|---|---|

798 | t (12, 14) and t (14, 15) | t (2, 8), t (2, 15), t (12, 2), t (2, 12), t (2, 15), t (2, 14), reciprocal t (8, 2) and t (2, 8) |

629 | t (12, 14) and t (14, 15) | t (17, 14), t (12, 8), t (5, 14), t (1, 2), t (2, 14), t (2, 1), t (17, 13), t (18, 11), t (13, 19), t (19, 11), reciprocal t (1, 2) and t (2, 1) |

884 | t (12, 14) and t (14, 15) | t (12, 2), t (14, 12), t (1, 2), t (1, 7), t (1, 17), t (8, 7), t (7, 15), reciprocal t (6, 7) and t (7, 6), reciprocal t (4, 10) and t (10, 4) |

590 | t (4, 14), t (14, 16), and t (16, 4, 14) | t (14, 4), t (13, 9), t (16, 4, 14), t (13, 2), t (14, 3), t (4, 8), t (2, 17), t (12, 6), t (12, 8), t (16, 12), t (2, 6), t (19;Y), t (8, 16), t (8, 19), t (8, 14), t (12, 2), t (12, 15), t (12, 14), t (2, 16), t (2, 12), t (8;X) |

589 | None | t (2, 1), t (4, 5), t (12, 2), t (14, 15), t (15, 4), t (4, 15), t (4, 6), t (12, 14), t (5, 4), t (4, 15, 16), t (8, 19), t (1, 17), t (7:15;14), t (17, 19), t (15, 7), t (2, 12), t (12, 2), t (1, 8), t (1, 2), t (8, 1), t (18, 12), t (14;15;Y), t (5, 1), reciprocal t (4, 15) and t (15, 4), reciprocal t (5, 17) and t (17, 5), reciprocal t (6, 18) and t (18, 6), reciprocal t (3, 4) and t (4, 3) |

Translocations classified as clonal were detected in at least 80% of metaphases.

### Nonredundant ATM and H2AX functions maintain genomic stability in αβ T-lineage cells

ATM-independent H2AX functions protect *Atm ^{−/−}* embryonic cells from genomic stability arising from DSBs induced before, during, and after DNA replication (18). To evaluate whether ATM-independent H2AX functions similarly maintain genomic stability in somatic cells, we conducted SKY on large numbers of metaphase spreads prepared from in vitro-stimulated

*LA*,

*LH*, or

*LAH*splenic αβ T cells. We quantified chromosome breaks, chromatid breaks, detached centromeres, chromosome fusions, and translocations, as these types of chromosomal abnormalities arise from unrepaired or misrepaired DSBs (Table III). We found that a higher percentage of

*LAH*metaphases contained such chromosome abnormalities relative to

*LH*(41 versus 23%,

*p*= 0.0001) or

*LA*(41 versus 23%,

*p*< 0.0001) metaphases (Fig. 4A). The average number of such abnormalities per metaphase also was higher for

*LAH*cells compared with

*LH*(0.68 versus 0.34,

*p*= 0.0001) or

*LA*(0.68 versus 0.33,

*p*< 0.0001) metaphases (Table III). In addition, the frequency of each abnormality was greater in

*LAH*metaphases relative to

*LA*and

*LH*metaphases (Supplemental Table III). Consequently, this cytogenetic analysis demonstrates that nonredundant functions of ATM and H2AX maintain genomic stability in proliferating mature αβ T cells.

Genotype . | Metaphases Analyzed . | Metaphases with Genomic Instability
. ^{a} | Total Number of Lesions . | Lesions per Metaphase . |
---|---|---|---|---|

LH | 330 | 75 (23%) | 109 | 0.33 |

LA | 254 | 66 (26%) | 97 | 0.38 |

LAH | 237 | 98 (41%) | 164 | 0.69 |

Genotype . | Metaphases Analyzed . | Metaphases with Genomic Instability
. ^{a} | Total Number of Lesions . | Lesions per Metaphase . |
---|---|---|---|---|

LH | 330 | 75 (23%) | 109 | 0.33 |

LA | 254 | 66 (26%) | 97 | 0.38 |

LAH | 237 | 98 (41%) | 164 | 0.69 |

^{a}Indicated are the total numbers of abnormalities and the percentage of metaphases with each type of abnormality.

Although most abnormalities were nonrecurrent, we observed higher frequencies of two types of chromosomal lesions in *LAH* metaphases compared with *LA* and *LH* metaphases. We detected radial chromosome translocations involving aberrant fusion and cohesion of heterologous chromatids in 5.5% of *LAH* metaphases but not in any *LA* or *LH* metaphases (5.5% versus 0, *p* = 0.0005) (Fig. 4B, 4C). These radial translocations arise in HR-defective backgrounds and form when chromatid breaks are not repaired via HR but instead aberrantly joined via NHEJ before G2/M (31). Thus, our detection of radial translocations in *LAH*, but not *LA* or *LH*, metaphases reveals that ATM and H2AX have nonredundant functions during HR in proliferating αβ T cells. The Ag receptor loci that recombine in developing αβ T cells are located on chromosomes 6 (Tcrβ), 13 (Tcrγ), 14 (Tcrα/δ), and 12 (Igh). We detected similar frequencies of translocations involving chromosomes 13, 14, and 12 among LH, LA, and LAH metaphases (Supplemental Table IV). Notably, we observed a higher frequency of chromosome 6 translocations in *LAH* metaphases than in either *LA* or *LH* metaphases (Supplemental Table IV). To identify and quantify potential TCRβ translocations, we conducted FISH with chromosome 6 paints and a genomic DNA probe that binds downstream of Dβ-Jβ segments (3′TCRβ). We scored TCRβ translocations when 3′TCRβ FISH signals were located at breakpoints of chromosome 6 translocations (Fig. 4D). We found a greater frequency of TCRβ translocations in *LAH* metaphases compared with *LA* (3.5 versus 0.46%, *p* = 0.0045) or *LH* metaphases (3.5 versus 0.63%, *p* = 0.014) (Fig. 4E). Because TCRβ-selection is required for αβ T cell development, the elevated frequency of TCRβ translocations in *LAH* cells versus *LA* and *LH* cells demonstrates that nonredundant functions of ATM and H2AX suppress the aberrant repair of RAG DSBs induced on nonselected TCRβ alleles. For reasons explained in the 13*Discussion*, we conclude from our data that nonredundant functions of ATM and H2AX protect proliferating αβ T-lineage cells from genomic stability initiated by spontaneous and programmed DSBs.

## Discussion

In this study, we generated and analyzed *Atm*-deficient mice and cells with conditional deletion of *H2ax* to assess potential redundant and nonredundant functions of ATM and H2AX in developing lymphocytes. We had previously demonstrated that G1 phase *Atm ^{−/−}*, but not H2ax

^{−/−}, abl pre-B cells exhibit defects in chromosomal CJ formation (25, 29). In the current study, we showed that

*Atm*and

^{−/−}H2ax^{−/−}*Atm*abl pre-B cells display similar defects in NHEJ-mediated CJ formation in G1 phase cells. Analogous to the situation for H2AX, inactivation of the XLF DDR has no obvious effect upon CJ formation in G1 phase abl pre-B cells (19). However, combined inactivation of

^{−/−}*H2ax*and

*Xlf*leads to a major defect in CJ formation that is associated with degradation of RAG-generated chromosomal CEs (32). In addition, combined inactivation of

*Atm*and

*Xlf*leads to a near complete block in chromosomal CJ formation in G1-arrested abl pre-B cells (32). These XLF studies revealed that XLF has essential nonredundant functions with ATM and H2AX during chromosomal CJ formation in G1 phase cells. Because the identical assay was used in these experiments and our current study, our results provide strong evidence that ATM promotes NHEJ-mediated CJ formation in G1 phase lymphocytes through mechanisms that are not critically dependent upon H2AX.

Despite showing that G1 phase *Atm ^{−/−}H2ax^{−/−}* and

*Atm*lymphocytes exhibit similar defects in CJ formation, we discovered a 7-fold higher frequency of TCRβ translocations in splenic αβ T cells from

^{−/−}*LAH*mice compared with

*LA*mice. TCRβ genes are assembled in G1 phase DN cells through an ordered process where Dβ-to-Jβ recombination is followed by Vβ rearrangement to DβJβ complexes on one allele at a time (33). TCRβ gene expression from one allele selects DN thymocytes for further differentiation and drives G1-to-S progression and multiple cell cycles as cells differentiate into DP thymocytes (33). Although ATM activates the G1/S checkpoint to prevent DN cells with RAG DSBs from entering into S phase (34), a fraction of TCRβ-selected thymocytes enters S phase with unrepaired RAG DSBs induced during Vβ recombination on nonselected alleles (35). We previously showed that H2AX stabilizes RAG-cleaved DNA strands to prevent DSBs that persist into S phase from progressing into chromosome breaks and translocations during cellular proliferation (29). We also recently found that H2AX phosphorylation blocks resection of CEs to prevent RAG DSBs from entering homology-mediated repair pathways that promote translocations (36). Consequently, we conclude that, at least in ATM-deficient cells, H2AX functions downstream of DNA-PK to suppress TCRβ translocations that arise from RAG DSBs induced on nonselected TCRβ alleles in DN cells experiencing TCRβ-proliferation signals. Notably, although ∼5% of splenic αβ T cells from

*Atm*mice harbor TCRδ translocations that arise in DN thymocytes (8, 37), we found a similar frequency of chromosome 14 translocations in

^{−/−}*LAH*and

*LA*αβ T cells. These data suggest that ATM-independent H2AX functions may have a more prominent role in preventing aberrant repair of RAG DSBs induced in TCRβ loci than in TCRδ loci.

Our finding that *LAH* mice exhibit a greater reduction of TCRβ-mediated thymocyte expansion than *LA* mice indicates that ATM and H2AX serve nonredundant functions in developing αβ T cells. Because ATM and H2AX have redundant functions during CJ formation in G1 phase cells, the impaired proliferative expansion of *LAH* thymocytes likely reflects nonredundant functions of H2AX and ATM in response to replication-associated DSBs. Our detection of radial translocations in in vitro-stimulated αβ T cells of *LAH* mice, but neither *LA* nor *LH* mice, indicates that ATM and H2AX exhibit nonredundant functions during HR in proliferating lymphocytes. H2AX phosphorylation by ATM and DNA-PK is required for HR repair of DNA replication-associated DSBs and restart of DNA replication (38). Thus, we conclude that the more substantial defect in TCRβ-mediated thymocyte expansion in *LAH* mice, compared with *LA* mice, is due to loss of DNA-PK–mediated H2AX functions that suppress translocations from replication-associated DSBs. Such H2AX functions could be the recruitment of HR proteins to promote rapid DNA repair and/or the cohesion of sister chromatids to prevent separation of broken DNA ends.

Our finding that *LAH* and *LA* mice exhibit similar mortality from thymic lymphomas with clonal chromosome 14 translocations has implications for treatment of human cancers with *ATM* inactivation. The clonal t (12, 14) translocations of *Atm ^{−/−}* thymic lymphomas delete one allelic copy of the

*Bcl11b*haplo-insufficient tumor suppresser gene that encodes a protein required for activation of the DDR during DNA replication stress (7, 39, 40). Mutation or deletion of

*ATM*and

*BCL11B*often occur in human T cell acute lymphoblastic leukemia (T-ALL) (5, 41–45), whereas inactivation of only

*ATM*is frequently observed in human T cell prolymphocytic leukemia (T-PLL) (5, 46–48).

*ATM*inactivation in T-ALL is associated with therapy resistance and relapse (42), whereas no cure exists for T-PLL with

*ATM*mutation (49). Clearly, the development of more effective therapies for these malignancies is warranted. Targeted inhibition of tumor cell intrinsic nonredundant DDR mechanisms holds great promise for more effective and less toxic cancer treatments (50–52). This approach leads to increased genomic instability that causes apoptosis in tumor cells while minimally affecting normal cells (50–52). Although combined inactivation of

*Atm*and

*H2ax*impairs cellular proliferation and causes substantial genomic instability, our data suggest that targeted inhibition of ATM-independent H2AX functions would not be an effective treatment for T-ALL tumors with

*ATM*and

*BCL11B*inactivation. However, additional studies are warranted to evaluate whether this strategy would be effective for other T-ALL subtypes, T-PLLs, and/or other cancers with recurrent

*ATM*inactivation, such as mantle cell lymphoma (53).

## Footnotes

This work was supported by a Cancer Research Institute Predoctoral Emphasis Pathway in Tumor Immunology training grant awarded to the University of Pennsylvania (to B.Y.), the Department of Pathology and Laboratory Medicine and the Center for Childhood Cancer Research of the Children’s Hospital of Philadelphia Research Institute, a grant from the Pennsylvania Department of Health, and National Institutes of Health Grants R01 CA125195 (to C.H.B.) and R01 CA136470 (to C.H.B. and B.P.S.).

The online version of this article contains supplemental material.

Abbreviations used in this article:

- abl
Abelson transformed

- ATM
ataxia telangiectasia mutated

- CE
coding end

- CJ
coding join

- DDR
DNA damage response

- DN
double-negative

- DP
double-positive

- DSB
double-strand break

- FISH
fluorescence in situ hybridization

- HR
homologous recombination

*LA**Lck-cre*^{+/−}Atm^{–}^{/}^{–}*LAH**Lck-cre*^{+/−}Atm^{–}^{/}^{–}*H2ax*^{F/F}*LH**Lck-cre*^{+/−}H2ax^{F/F}- NHEJ
nonhomologous end-joining

- RAG
RAG1/RAG2

- SKY
spectral karyotyping

- SP
single-positive

- T-ALL
T cell acute lymphoblastic leukemia

- T-PLL
T cell prolymphocytic leukemia.

## References

## Disclosures

The authors have no financial conflicts of interest.