ATM deficiency augments constitutively nuclear cyclin D1-driven genomic instability and lymphomagenesis

Abstract

Cyclin D1 deregulation is implicated in the genesis of multiple human cancers. Importantly, nuclear cyclin D1 retention during S-phase promotes DNA re-replication and subsequent genomic instability, providing a direct correlation between aberrant cyclin D1/CDK4 activity, transcriptional regulation and double strand DNA break (DSB) induction. Together, these molecular events catalyze the genomic instability necessary for neoplastic transformation. Given that replication-associated DNA damage is central to cyclin D1-driven neoplasia, inactivation of critical checkpoint mediators should augment cyclin D1-dependent tumorigenesis in vivo. To interrogate potential synergy between constitutively nuclear cyclin D1 expression and impaired DSB-induced checkpoint integrity, Ataxia Telangiectasia Mutated (ATM)-deficient mice harboring the Eμ-D1T286A transgene were generated and evaluated for tumor onset. Eμ-D1T286A/ATM −/− mice exhibit dramatically accelerated incidence of both B- and T-cell lymphomas relative to Eμ-D1T286A or ATM −/− control cohorts. Lymphomas exhibit clonal chromosomal alterations distinct from ATM −/− mice, which typically acquire translocations involving the Tcrα/δ locus during V(D)J recombination, and instead harbor alterations at the c-Myc locus. Collectively, these findings reveal an intricate relationship wherein nuclear cyclin D1/CDK4 drives genomic instability in the absence of ATM function and clonal selection of cells harboring alterations within the murine c-Myc locus, ultimately facilitating transformation and tumor formation.

Introduction

Cyclin D1 deregulation is a hallmark of mantle cell lymphoma (MCL), a pathological signature associated with aggressive disease and poor response to chemotherapeutic treatment approaches.¹ Although MCL is characterized by the t(11:14) chromosomal translocation, which juxtaposes cyclin D1 with the immunoglobulin (Ig) H locus,^2,3 additional chromosomal abnormalities are also observed at a reduced yet significant frequency including deletion of 11q22-q23.^4–6 Of note, the latter chromosomal region harbors the Ataxia Telangiectasia Mutated (ATM) gene, encoding a key factor in the DNA damage response. Following genotoxic insult, ATM is activated and subsequently phosphorylates key DNA damage response effectors, including Chk2 and H2AX, thereby signaling DNA repair as well as either p53-dependent cell cycle arrest or apoptosis.^7,8 Of note, p53 loss is relatively infrequent in MCL, occurring in fewer than 30% of cases, suggesting the existence of alternative mechanisms that bypass p53-dependent tumor suppression.^9,10

Although frequently overexpressed in cancer, wild-type cyclin D1 itself is weakly oncogenic.^11,12 However, its potency as a driver oncogene is increased dramatically upon deregulation of its nuclear export-dependent degradation during S-phase.^13,14 Constitutively nuclear cyclin D1 mutants such as D1T286A, where the phospho-acceptor site for GSK3β is absent, is a potent oncogene that promotes neoplastic growth in vitro and in vivo.^12,13,15,16 Nuclear stabilization and accumulation of cyclin D1-dependent kinase during S-phase is associated with marked genomic instability, which reflects re-licensing of DNA replication origins, DNA re-replication and double strand DNA break (DSB) checkpoint activation.¹⁷ Molecular analysis of cells expressing D1T286A suggest a model where D1T286A/CDK4 represses Cul4 gene expression, resulting in the subsequent stabilization of the Cul4 substrate Cdt1, a key component of the DNA replication licensing machinery.^18–21 Stabilized and overexpressed Cdt1 then contributes to iterative rounds of replication initiation, which culminates in DSB and activation of the DSB checkpoint.¹⁷

Loss of genomic integrity provides a mechanism for the acquisition of additional pro-tumorigenic mutations or ‘second hits’ that permit neoplastic growth. Expression of cyclin D1T286A in the hematopoietic compartment of mice triggers high-grade diffuse large B-cell lymphomas, originating in an early mature B-cell population. DSB checkpoint activation in response to cyclin D1T286A expression triggers p53-dependent apoptosis, contributing to the extended latency preceding overt tumor formation.¹² The second hit generally impacts p53, either through direct mutation or inactivation of a key p53 activator or effector. Indeed, tumorigenesis can be dramatically accelerated by deletion of a single p53 allele, and characterization of pre-malignant tissue revealed activation of ATM and Chk2 and phosphorylation of both p53 and H2AX, suggesting that activation of the DSB checkpoint functions to limit tumor progression¹⁷. Consistent with this notion, ATM signals accelerated cyclin D1 proteolysis following DNA damage at the G1/S transition, thereby maintaining genome stability. In contrast, degradation-refractory cyclin D1T286A expression sustains DNA synthesis in the presence of genotoxic insult, leading to chromosomal alterations and genomic instability conducive to transformation.²²

Although several parallels have been drawn between checkpoint activation and the neoplastic potential of cyclin D1T286A, whether the DSB checkpoint response is engaged as a tumor suppressive mechanism to limit the neoplastic potential of cyclin D1 remains to be tested. Herein, we directly addressed the contribution of the DSB checkpoint and ATM in the suppression of cyclin D1-driven hematopoietic malignancy through generation of Eμ-D1T286A/ATM −/− mice. Deletion of ATM significantly accelerates cyclin D1T286A-dependent lymphomagenesis and alleviates selective pressure for p53 mutation. Our data suggest that loss of the ATM-dependent checkpoint response and subsequent persistence of unrepaired DSBs facilitates chromosomal translocations in the presence of constitutively nuclear cyclin D1, supporting the notion that the ATM-dependent checkpoint response is a critical suppressor of nuclear cyclin D1-driven neoplastic transformation.

Results and Discussion

Loss of ATM accelerates cyclin D1T286A-driven lymphomagenesis To directly assess the contribution of the ATM to the suppression of cyclin D1T286A-driven tumorigenesis, we intercrossed transgenic Eμ-D1T286A mice with ATM +/− mice.^12,23 Wild-type, cyclin D1T286A/ATM +/+, cyclin D1T286A/ATM +/−, cyclin D1T286A/ATM −/−, ATM +/−, and ATM −/− progeny were generated, and cohorts were monitored over time for onset of lymphomagenesis. Strikingly, the combined deletion of ATM with expression of the D1T286A transgene substantially accelerated the onset of lymphomagenesis relative to either single transgenic cohort (Figure 1a). The double transgenic cyclin D1T286A/ATM −/− mice exhibited a significant tumor burden with a mean onset of 6 months as compared with either ATM −/− or single Eμ-D1T286A mice, which exhibited pathology with a mean onset of 9 and 13 months, respectively. ATM heterozygosity did not impact tumor latency, suggesting that the remaining copy of the ATM gene remains intact in this model system. Tumors derived from cyclin D1T286A/ATM +/+ and cyclin D1T286A/ATM −/− mice each retain ectopic Flag-D1T286A protein, consistent with the notion that its continued expression is important not only for tumor initiation, but also tumor progression and survival (Figure 1b). All cyclin D1T286A/ATM −/− mice developed lymphoid malignancies within the spleen or thymus, with occasional disseminated tumor in peripheral lymphoid tissue including the cervical and mesenteric lymph nodes and liver (data not shown). Primary tumors arise with similar frequency in the spleen and the thymus, compared with cyclin D1T286A and ATM −/− controls, which typically present in the spleen and thymus, respectively (Figure 1c).

Figure 1.

ATM deficiency significantly accelerates B- and T-cell lymphomagenesis in Eμ-D1T286A transgenic mice. (a) Kaplan — Meier percent tumor-free survival curve of D1T286A/ATM −/− double-mutant mice compared with control cohorts including non-transgenic (non-Tg), D1T286A, D1T286A/ATM +/−, and ATM −/− mice (up to 18 months of age). The green line denotes Eμ-D1T286A/ATM^−/− mice, which develop tumors with a significantly decreased latency (mean survival = 6 months) compared with controls. Lymphoma-free survival analysis was carried out using GraphPad Prism; statistical evaluation of survival data (as determined by log-rank test) is shown. Sample size of each genotype cohort is noted in the legend. (b) Eμ-D1T286A transgene expression is sustained in D1T286A/ATM −/− tumors. Representative immunoblot analysis of D1T286A (TA) and of D1T286A/ATM −/− (TA/ATM −/−) splenic tumor lysate. (c) Distribution of primary tumor presentation sites in D1T286A/ATM −/− and control cohorts.

Previous work demonstrated that Eμ-D1T286A mice develop diffuse large B-cell lymphomas, originating in early mature B-cells positive for markers B220 and CD19.^12,17 In contrast, ATM −/− mice develop immature CD4 +/CD8+ T-cell thymic lymphomas, which commonly arise during recombination-activating gene-mediated V(D)J recombination at the murine Tcr α/δ (T-cell receptor) locus.^23-25 We performed immunological and histological analysis of D1T286A/ATM −/− tumors to address their cellular and molecular characteristics. A subset of cyclin D1T286A/ATM −/− splenic tumors exhibited characteristics similar to those in single transgenic Eμ-D1T286A mice (Table 1, Supplementary Figures 1B, D and data not shown). These tumors were B220+ and negative for markers of T-cells (CD4, CD8, CD3). The architecture of the spleen was largely lost and in many cases there was evidence of tumor dissemination to the liver. In contrast, the cyclin D1T286A/ATM −/− thymic lymphomas are of T-cell origin. This subset of tumors exhibited characteristics of various stages of T-cell development, from immature CD4+/CD8+ to single CD4+ or CD8+ phenotypes (Table 1, Supplementary Figures 1A, C). These results reveal significant synergism between D1T286A expression and ATM deletion. Although lymphomas of both B- and T-cell origin occur, neither exhibits characteristics of those arising on an ATM −/− background, suggesting that impaired recombination-activating gene-dependent Tcr rearrangement may only be a minor contributor to neoplastic transformation in the presence of cyclin D1 as the driver oncogene.

Table 1. D1T286A/ATM ^−/− mice develop B- and T-cell lymphomas with clonal chromosome alterations.

Mouse Number	Survival	Tumor	Immunophenotype	Immunohistochemistry	SKY (clonal alterations)
124	7 months	Thymus	CD4+CD8+	B220−	t(4;X), reciprocal t(11;3), t(3;11)
298	4 months	Spleen	CD4+	B220−	t(15;1); trisomy15
310	5 months	Thymus	CD4+CD8+	B220−	t(12;15)
316	6 months	Pre-malignant Spleen	CD19+B220+	B220+	non-clonal
378	5.5 months	Thymus	CD8+ CD410	B220−	t(12;15)
390	8 months	Spleen	CD19+ B220+	B220+	—
435	5 months	Thymus	CD8+	B220−	t(15;2)
441	7 months	Spleen	CD19+ B220+	B220+	t(19;14),t(8;15)

Abbreviation: Spectral karyotyping, SKY. Representative summary of tumors observed in D1T286A/ATM −/− mice. SKY analysis was performed for a cohort of 8 D1T286A/ATM −/− mice to determine whether tumors harbor clonal chromosome alterations. The immunophenotype of each tumor, as determined by immunohistochemical staining and flow cytometric assays, is also shown.

Cyclin D1T286A/ATM −/− lymphomas harbor distinct clonal chromosome alterations

Constitutively nuclear cyclin D1/CDK4 activity has been associated with the loss of genomic integrity including non-clonal chromosomal translocations.¹⁷ ATM deletion typically results in the development of immature T-cell lymphomas harboring clonal translocations that commonly join the Tcrδ locus to telomeric regions of chromosome 12 (t(12;14)), due to aberrant V(D)J recombination and DSB repair.²⁵ To determine whether transgenic cyclin D1T286A expression in the absence of ATM results in an increased propensity of clonal chromosomal translocations, we performed spectral karyotyping (SKY) on a panel of tumors of either B- or T-cell origin, as well as one pre-malignant spleen. SKY analysis revealed clonal cytogenetic alterations in cyclin D1T286A/ATM −/− tumors that are distinct from those commonly observed in Eμ-D1T286A or ATM-deficient mice (Table 1). Tumors arising in cyclin D1T286A/ATM −/− mice largely maintain normal chromosome number and exhibit clonal chromosomal translocations (Figures 2a and b); this is in contrast to Eμ-D1T286A single transgenic tumors, which exhibit a high degree of polyploidy and exhibit non-clonal translocations.¹⁷ A majority of tumors exhibit clonal alterations involving chromosome 15, including t(15;1), t(12;15), t(15;2), and t(8;15) (Table 1, Figure 2a). Furthermore, pre-malignant cyclin D1T286A/ATM −/− splenocytes display both normal metaphases, as well as metaphases harboring an array of chromosomal translocations, including reciprocal t(8;15) and t(12;15), suggesting that productive translocations involving chromosome 15 that result in deregulation of oncogene expression or loss of a tumor suppressor are clonally selected (Figure 2c). The absence of t(12;14) translocations suggest that expression of D1T286A bypasses the previously observed aberrant V(D)J/class-switch recombination-driven genomic instability associated with ATM loss, and thus by definition, utilizes an alternative mechanism to drive neoplastic transformation.

Figure 2.

D1T286A/ATM −/− lymphomas exhibit clonal chromosome alterations commonly involving chromosome 15. (a) Representative D1T286A/ATM −/− metaphases from two tumors. Top panel: t(12;15), t(6;15). Bottom panel: t(12;15), t(2;15). White arrowheads denote translocations. (b) Chromosome number and translocations shown for one tumor harboring the t(12;15) translocation (complete metaphase shown in top panel of a). (c) Pre-malignant splenocytes isolated fromD1T286A/ATM −/− spleens harbor multiple, non-clonal chromosome alterations, typically involving chromosome 15. (d) Elevated Cdt1 protein expression and histone arginine methylation marks in D1T286A/ATM −/− tumors supports a role for D1T286A-driven re-replication in driving genomic instability in the absence of ATM. Non-Tg control or lymphoid tumor tissue was snap frozen, followed by lysis in Tween 20-containing buffer, homogenization, SDS — polyacrylamide gel electrophoresis and immunoblot analysis as indicated.

c-Myc deregulation and translocation in cyclin D1T286A/ATM −/− lymphomas

The unique cytogenetic signature observed in cyclin D1T286A/ATM −/− tumors suggests that synergy between nuclear cyclin D1 function and an impaired checkpoint response contributes to tumor formation. Provided this hypothesis, we anticipated overexpression of Cdt1 should be apparent as has been previously noted in Eμ-D1T286A tumors.^17,26 Consistent with this notion, Cdt1 protein expression is elevated and Cul4B accumulation is reduced in cyclin D1T286A/ATM −/− tumors compared with non-transgenic controls, suggesting that cyclin D1T286A-dependent Cul4A/B repression facilitates Cdt1 stabilization in the resulting tumors (Figure 2d). Interestingly, c-Myc protein expression is also elevated consistent with translocation events occurring in proximity of the c-Myc locus (Figure 2d; Figure 3). Typically, p53-dependent apoptosis counters neoplastic transformation in Eμ-D1T286A lymphocytes and subsequent p53 mutation or loss of heterozygosity in Eμ-D1T286A/p53 +/− mice permits progression to overt neoplasia.¹⁷ The activation of p53 in these tumors could reflect either mitogenic or oncogenic stress triggered by expression of D1T286A, analogous to expression of oncogenic Ras mutants,^27,28 or in these tumors p53 activation may reflect DNA damage that accrues through deregulated DNA replication. In the case of the latter, we expected that deletion of ATM would obviate the selection for p53 loss of function mutations. Sequencing of p53 from a panel of cyclin D1T286A/ATM −/− tumors revealed that no mutations are acquired in the ‘hotspot’ p53 DNA-binding domain²⁹ (Supplementary Figures 2A–D). Taken together, these data suggest that the DSB checkpoint functions as the critical tumor suppressor in cells expressing deregulated cyclin D1; in the absence of ATM, unrepaired DNA damage leading to c-Myc translocation and overexpression then provides oncogenic insult sufficient for transformation.

Figure 3.

D1T286A/ATM −/− tumors exhibit chromosomal alterations involving the c-Myc locus. (a) Duplication of the c-Myc locus in a representative D1T286A/ATM −/− tumor (tumor #435, Table 1). (a) FISH probe encompassing the c-Myc locus was utilized to visualize c-Myc copy number in tumors harboring alterations involving chromosome 15. Bottom panel: schematic of the murine c-Myc locus and FISH probe. Note that locus organization is conserved in humans on chromosome 8. (b) c-Myc FISH in combination with chromosome 15 paint demonstrating duplication of centromeric chromosome 15 sequence and translocation to chromosome 1, t(15;1), tumor #298 (Table 1).

In order to determine whether clonal chromosome 15 translocations impact the c-Myc locus, we performed fluorescence in situ hybridization (FISH) on a panel of cyclin D1T286A/ATM −/− tumors using the c-Myc internal BAC probe 307D14. FISH analysis revealed several tumors with c-Myc duplication or a translocation consistent with SKY results. For instance, the tumor #435 (harboring t(15;2)), displays two intact c-Myc signals, along with duplicated c-Myc sequence that could correspond to the chromosome 15 translocation (Figure 3a). Chromosome 15 painting in combination with c-Myc FISH on tumor #298 (harboring t(15;1)) revealed two intact chromosome 15 signals and duplication of chromosome 15 sequence encompassing the c-Myc locus translocated to chromosome 1 (Figure 3b), suggesting that the c-Myc gene was not only duplicated, but DNA damage within this region resulted in chromosomal translocation as well. Although DNA damage upstream of c-Myc leading to translocation events has been documented in developing lymphocytes,^30–33 it is intriguing to note the presence of a putative origin of replication mapped ∼1 kb upstream of the c-Myc locus conserved in human, mouse and xenopus.^34,35 Thus, it is possible that nuclear cyclin D1-dependent Cul4 repression/Cdt1 stabilization promotes DNA re-replication in developing B- and T-cells, and frequent replication fork collision/collapse leads to DSB induction. In the absence of ATM, such breaks would not be efficiently sensed and repaired, resulting in chromosomal translocations. Although this proposed mechanism could occur at any replication origin, productive chromosome alterations that facilitate loss of tumor suppressor proteins or activation of oncogenes should be clonally selected for tumor growth.

The strong selection for translocations involving c-Myc apparent in cyclin D1T286A/ATM −/− tumors may reflect the presence of a replication origin at this locus; furthermore, the concurrent occurrence of clonal t(12;15) translocations in many of these tumors also suggests that DNA re-replication could promote breaks downstream of the IgH locus in an AID (Activation-Induced cytidine Deaminase)-independent manner, as AID is primarily expressed in germinal center B-cells and is involved in class-switch recombination and somatic hypermutation processes during B-cell development.^33,36 Interestingly, a putative replication origin was mapped between the murine 3′IgH regulatory region and the downstream gene, Crip; further analysis revealed that sequences downstream of this region are replicated bi-directionally.^37,38 It is therefore plausible that bi-directional fork movement in the context of DNA re-replication could generate significant DNA damage, permitting chromosome translocations involving IgH, independent of AID function.

The role of DNA damage per se and the DSB checkpoint as a tumor suppressive mechanism has gained significant credence in recent years. Our results demonstrate that the ATM kinase functions as a key sensor of DNA damage in lymphocytes harboring oncogenic cyclin D1, and its activation is critical for p53-dependent tumor suppression. Given the associated loss of genomic material containing the ATM gene in human MCL, our work supports a model wherein ATM loss is not simply a rider mutation. Rather, deletion of ATM functionally contributes to neoplastic growth and in tumors where this occurs, bypasses a need for loss of function p53 mutations.

Materials and Methods

Generation of Eμ-D1T286A/ATM −/− Mice

Transgenic Eμ-D1T286A mice were bred with ATM +/− mice to generate Eμ-D1T286A/ATM +/− offspring. Transgenic breeders were then crossed with ATM +/− mice to yield pups of the desired Eμ-D1T286A/ATM −/− genotype. Importantly, Eμ-D1T286A was maintained as a single-copy transgene. Genomic DNA isolated from tail snips was used to genotype pups. Mice were examined daily for signs of distress or palpable tumor mass according to Institutional Animal Care and Use Committee (IACUC) guidelines, and tumor-free survival was assessed by Kaplan — Meier analysis. Prism GraphPad Software (Graph Pad Software, La Jolla, CA, USA) was utilized for generating Kaplan — Meier mouse tumor-free survival plots and statistical analysis of survival and tumor onset (log-rank test). Please refer to the Supplementary Methods for complete details on cytogenetic and immunoblot experiments performed on Eμ-D1T286A/ATM −/− tumors.