Ectopic TLX1 Expression Accelerates Malignancies in Mice Deficient in DNA-PK

The noncluster homeobox gene HOX11/TLX1 (TLX1) is detected at the breakpoint of the t(10;14)(q24;q11) chromosome translocation in patients with T cell acute lymphoblastic leukemia (T-ALL). This translocation results in the inappropriate expression of TLX1 in T cells. The oncogenic potential of TLX1 was demonstrated in IgHμ-TLX1Tg mice which develop mature B cell lymphoma after a long latency period, suggesting the requirement of additional mutations to initiate malignancy. To determine whether dysregulation of genes involved in the DNA damage response contributed to tumor progression, we crossed IgHμ-TLX1Tg mice with mice deficient in the DNA repair enzyme DNA-PK (PrkdcScid/Scid mice). IgHµ-TLX1TgPrkdcScid/Scid mice developed T-ALL and acute myeloid leukemia (AML) with reduced latency relative to control PrkdcScid/Scid mice. Further analysis of thymi from premalignant mice revealed greater thymic cellularity concomitant with increased thymocyte proliferation and decreased apoptotic index. Moreover, premalignant and malignant thymocytes exhibited impaired spindle checkpoint function, in association with aneuploid karyotypes. Gene expression profiling of premalignant IgHµ-TLX1TgPrkdcScid/Scid thymocytes revealed dysregulated expression of cell cycle, apoptotic and mitotic spindle checkpoint genes in double negative 2 (DN2) and DN3 stage thymocytes. Collectively, these findings reveal a novel synergy between TLX1 and impaired DNA repair pathway in leukemogenesis.


Introduction
The homeobox gene TLX1 encodes a member of the nonclustered subclass of homeodomain-containing transcription factors. TLX1 was initially identified at the breakpoint of the t(10;14)(q24;q11) reciprocal chromosome translocation in patients with T cell acute lymphoblastic leukemia (T-ALL) [1,2,3]. This translocation places the entire TLX1 coding region under the transcriptional control of the T cell receptor d (TCRd) promoter resulting in inappropriate expression of TLX1 in T cells, and is found in ,5% of pediatric T-ALL and 30% of adult T-ALL cases. TLX1-initiated T-ALL exhibits a block at the stage of b-selection [4]. This maturation arrest might be due to down regulation of the BCL11B tumor suppressor gene [5,6] or as a result of recruitment of TLX1 by the transcription factor ETS1 to the enhancer of the TCR-a locus leading to repression of Va-Ja rearrangements [7]. TLX1 is not typically expressed in adult tissues but is critical for the development of the spleen during embryogenesis [8,9,10].
Transgenic mice have played a central role in defining the molecular mechanisms of TLX1-initiated hematopoietic malignancies. Transgenic mice in which the TLX1 gene is placed under the control of the immunoglobulin heavy chain (IgH) promoter and enhancer, thus directing its expression predominantly to the B cell compartment (IgHm-TLX1 Tg mice), develop mature marginal zone B-cell lymphomas after an extended latency [11]. Surprisingly, although IgHm-TLX1 Tg mice express elevated levels of TLX1 in thymocytes during the early stages of T lymphopoiesis, no cases of T-ALL have been detected in these mice. More recently, p56 Lck -TLX1 and doxycyline repressible tet-TLX1 transgenic mice with Tcell specific TLX1 expression have been reported [5,12]. TLX1 expression in T cells induces a block in thymocyte development at the DN2 stage and, similar to human TLX1-initiated T-ALL, TLX1 transgenic mice develop cortical stage tumors with a heterogeneous CD4 and CD8 immunophenotype. Additionally, the doxycyline repressible tet-TLX1 transgenic mice subsequently acquire activating NOTCH1 mutations, consistent with reports that more than 50% of T-ALL patients carry NOTCH1 activating mutations [13] and that NOTCH1 and TLX1 can coregulate transcription in T-ALL cells [14].
An association of TLX1 with aberrant cell cycle checkpoint regulation has been known for many years. In the first such report, TLX1 was shown to interact with protein phosphatases PP2A and PP1 to disrupt a G2/M cell cycle checkpoint [15]. Subsequently, high throughput comparison of gene expression profiles of two cell lines established from patients with TLX1-positive T-ALLs revealed modulated expression of numerous genes associated with G1/S progression [16]. More recently, multiple groups reported that ectopic expression of TLX1 results in dysregulation of the spindle checkpoint [5,17], attributable, in part, to down-regulated expression of the mitotic checkpoint regulator CHEK1 [5]. One possible manifestation of such widespread subversion of cell cycle checkpoint control would be a loss of genomic integrity. Indeed, TLX1-transgenic mice exhibit a heightened predisposition for the development of whole chromosome copy changes [5,17,18].
Prkdc encodes the catalytic subunit of the DNA-dependent protein kinase (DNA-PKcs) which plays a critical role in the nonhomologous end joining (NHEJ) pathway of DNA repair. It is also critical for V(D)J recombination, a process which relies on NHEJ to promote immune system diversity at the Ig or TCR loci. Mice harboring homozygous germline inactivating mutations in the Prkdc gene (Prkdc Scid/Scid ) are viable but have severe combined immunodeficiency due to an inability to undergo V(D)J recombination. B cell development in these mice is arrested at the B220 + CD19 + IgMstage, whereas thymocyte development is arrested at the DN3 stage [19]. Critically, DNA-PK deficiency in mice contributes to genomic instability [20,21,22,23] and Prkdc Scid/Scid mice are prone to the spontaneous development of T-ALL, AML and nonthymic tumors [24,25,26].
In this report, we tested whether increased genomic instability associated with DNA-PK loss collaborates with TLX1 to accelerate T-ALL. To this end, we generated double mutant IgHm-TLX1 Tg Prkdc Scid/Scid mice and showed that these mice exhibit statistically significant accelerated onset of leukemia relative to Prkdc Scid/Scid mice. Moreover, we identify numerous genetic pathways that are perturbed in association with TLX1 overexpression, including those involved in chromosome segregation, cell cycle checkpoints and apoptosis.

Similar Premalignant Phenotypes in Young
IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid Mice To determine whether dysregulation of a DNA repair pathway collaborated with ectopic expression of TLX1 in disease progression, we crossed IgHm-TLX1 Tg mice with CB17 ICR-Prkdc scid mice (referred to as Prkdc Scid/Scid mice). We chose to use the ICR background as it most resembles the outbred background of the CD1-TLX1 mice.
The thymi and spleens of control Prkdc Scid/Scid mice and double mutant IgHm-TLX1 Tg Prkdc Scid/Scid mice were initially examined at 6 weeks of age, prior to the development of any overt leukemia. Thymi from IgHm-TLX1 Tg Prkdc Scid/Scid mice exhibited significantly reduced cellularity compared to wildtype mice, but was increased relative to those of Prkdc Scid/Scid mice ( Table 1). Spleens of IgHm-TLX1 Tg Prkdc Scid/Scid mice were also reduced in cellularity as compared to wild type mice, but were also increased as compared to spleens of Prkdc Scid/Scid littermates (Table 1). Due to the lack of lymphoid cells, Peyer's patches and lymph nodes of IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice were small and difficult to detect. Histologically, thymi of IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice lacked cortico-medullary delineation and consisted of predominantly epithelial cells and immature thymocytes. In the spleen, lymphoid follicles and germinal centers were devoid of lymphoid cells and populated with fibroblasts and plasma cells. Bone marrow histology appeared normal.
Expression of the TLX1 Transgene in Thymocytes of IgHm-TLX1 Tg Prkdc Scid/Scid Mice The TLX1 transgene in the double mutant mice was driven by the IgHm promoter. Although the activity of the IgHm promoter is typically restricted to B cells, leaky expression in thymocytes and myeloid progenitors has been reported [27,28]. Therefore, to quantitate IgHm-directed TLX1 expression during T lymphopoiesis, thymocytes were derived from wild type, Prkdc Scid/Scid and IgHm-TLX1 Tg Prkdc Scid/Scid mice by co-culture of murine fetal liver cells-derived hematopoietic stem cells (HSCs) on OP9-DL1 stromal cells and TLX1 expression was analyzed by RT-PCR ( Figure 1A). Weak TLX1 transgene expression was detected in DN1, DN2 and DN3 thymocytes and in CD44 -CD25cells ( Figure 1B). Expression was down regulated as cells progressed from DN1 to the DN3 stage and was significantly decreased in CD44 -CD25cells, consistent with declining activity of the IgHm promoter during thymocyte maturation as previously reported [27,28]. Expression of TLX1 was not detected in DN1, DN2, DN3 or CD44 -CD25thymocytes derived from Prkdc Scid/Scid or wild type mice indicating that the TLX1 detected originated from the IgHm-TLX1 transgene and not the endogenous mouse Tlx1 gene. To confirm that the TLX1 transgene was also expressed in vivo, quantitative RT-PCR analysis of thymocytes obtained from six week old IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid age-and sexmatched littermates was performed ( Figure 1C). Expression of the TLX1 transgene was detected in DN1, DN2, DN3 and CD44 -CD25thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid mice ( Figure 1D). Of note, the level of expression of the IgHm-TLX1 transgene in thymocytes was significantly lower than that detected in purified B220 + splenocytes from IgHm-TLX1 Tg mice.

Accelerated Leukemogenesis in IgHm-TLX1 Tg Prkdc Scid/Scid Mice
To determine whether ectopic expression of TLX1 collaborated with DNA-PK deficiency in the pathogenesis of hematopoietic malignancies, a cohort of 60 IgHm-TLX1 Tg Prkdc Scid/Scid mice and 50 age and sex matched Prkdc Scid/Scid mice were followed for 24 months. At the end of this period, all mice had developed a malignancy with histological features and flow cytometric profiles consistent with either T-ALL or acute myeloid leukemia (AML) (Figure 2A). However, IgHm-TLX1 Tg Prkdc Scid/Scid mice succumbed to disease at earlier time points relative to Prkdc Scid/Scid littermates. The median survival of the complete cohort of IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice, irrespective of disease diagnosis, was 6.6 and 10.25 months, respectively ( Figure 2D). The one-year survival probability estimate of IgHm-TLX1 Tg Prkdc Scid/Scid mice was also decreased relative to those of Prkdc Scid/Scid littermates developing either T-ALL or AML ( Figure 2E).
In total, 59% of IgHm-TLX1 Tg Prkdc Scid/Scid mice and 54% of Prkdc Scid/Scid mice developed an immature T-cell leukemia. The median survival of IgHm-TLX1 Tg Prkdc Scid/Scid mice with T-ALL was 6.5 months, whereas Prkdc Scid/Scid mice exhibited a protracted latency with a median survival of 8.25 months ( Figure 2B, D). Histological examination of tissues of Prkdc Scid/Scid and IgHm-TLX1 Tg Prkdc Scid/Scid mice exhibiting T-ALL indicated thymic involvement in 100% of animals, while splenic and bone marrow infiltration of tumor cells was detected in 67% and 52% of mice, respectively. This suggests that the thymus was the primary site for disease initiation. The architecture of the thymus was disrupted with indistinguishable cortico-medullary delineation. The spleens had disrupted follicles, with patchy areas infiltrated with a large, relatively uniform population of lymphocytes having polylobulated pleomorphic nuclei with diffuse, loose, speckled, open chromatin and prominent nucleoli ( Figure 3A).
Immunohistological analysis was performed using frozen sections of thymi and spleens isolated from age-and sex-matched healthy and terminal IgHm-TLX1 Tg Prkdc Scid/Scid mice. Staining of sections with the pan T-cell antibody, Thy1.2, revealed numerous Thy1.2 + cells in the subcortical area of the thymi while Thy1.2 + cells were absent in the medullar area of thymi and were not detected in the spleens of healthy mice. In contrast, a homogeneous population of Thy1.2 + tumor cells was detected throughout the thymi and spleens of terminal mice ( Figure 3B).
Flow cytometric analysis revealed heterogeneity in the tumor phenotype with respect to expression of CD4 and CD8 ( Figure 3C) with tumor cells isolated from the thymus, spleen and bone marrow expressing both CD4 and CD8 or only CD4 with decreased levels of expression of CD44 and CD25. Other tumors consisted of CD44 and CD25 expressing cells with profiles typical of DN1, DN2 and DN3 cells or only DN3 thymocytes. Tumor cells did not expressed abor cd-T cell receptors, indicative of an immature thymocyte phenotype ( Figure 3C). The majority (60%) of thymic tumors had a mixed phenotype expressing low-levels of  CD4 and CD8 along with a combination of CD44 and CD25 ( Figure 4). Forty one percent of IgHm-TLX1 Tg Prkdc Scid/Scid mice developed AML, with a median survival of 7.0 months and 46% of Prkdc Scid/Scid control mice developed AML with a median survival of 14.0 months ( Figure 2C, D). In these cases, histological analysis of bone marrow and spleens from terminally ill mice revealed the presence of leukemic cells with a primitive myeloid phenotype ( Figure 5A). Histological analysis indicated the absence of tumor cells in the thymus while flow cytometry analysis using CD44 and CD25 antibodies indicated that thymi contained a normal profile of immature DN thymocytes typical for mice with the Prkdc Scid/Scid mutation ( Figure 5B). Moreover, thymocytes did not express TCRb, CD3, CD4 and CD8 receptors and thus, had not progressed

Leukemic Thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid mice Exhibit Molecular Features Reminiscent of Human T-ALL
To test whether the T-ALL-like disease seen in the IgHm-TLX1 Tg Prkdc Scid/Scid mice resembled human T-ALL at a molecular level, we assessed the expression of several genes which had previously been shown to be associated with the progression of human T-ALL such as NOTCH1, PTEN and BCL11B [5,13,29,30] in ten tumors isolated from IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice. Consistent with gene expression patterns in human T-ALLs, these analyses revealed a statistically significant (p,0.05) reduction in the expression of Bcl11b and Pten in T-ALL arising in IgHm-TLX1 Tg Prkdc Scid/Scid mice relative to tumors isolated from Prkdc Scid/Scid mice ( Figure 6). Furthermore, elevated expression of Notch1 was detected in six of ten tumors isolated from IgHm-TLX1 Tg Prkdc Scid/Scid mice relative to Prkdc Scid/Scid mice (p,0.05). These data indicated that malignant thymocytes derived from IgHm-TLX1 Tg Prkdc Scid/Scid mice exhibited a similar pattern of expression of a subset of genes that has previously been implicated in TLX1-initiated human T-ALL.

Gene Expression Profiling in Premalignant Thymocytes Reveals Dysregulated Expression of Genes Involved in Cell Cycle, Apoptosis and Chromosome Segregation
Next, we sought to interrogate the molecular pathways dysregulated in the earliest stages of TLX1-induced transformation in the context of Prkdc deficiency. Therefore, we purified premalignant DN1, DN2 and DN3 thyomocytes from four cohorts of IgHm-TLX1 Tg Prkdc Scid/Scid mice and four cohorts of control Prkdc Scid/Scid mice, and compared their gene expression profiles using Affymetrix Gene Chip microarrays ( Figure S1A-B). Each cohort consisted of 2-5 sex-and age-matched mice. From this analysis, we identified 103 TLX1-associated differentially expressed genes (DEGs) in DN1 thymocytes, (89 up regulated genes, 14 down regulated genes), 151 DEGs in DN2 thymocytes (139 up regulated, 12 genes down regulated) and 522 DEGs in DN3 thymocytes (495 up regulated, 27 genes down regulated) ( Figure 7A, Tables S1-3). The over-abundance of DEGs in the DN3 fraction suggested that the effects of TLX1 on target gene expression were more profound at more mature stages of thymocyte development.
We next validated dysregulated expression of a subset of DEGs in thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid control mice by qRT-PCR ( Figure 8). Analysis of DN1, DN2 and DN3 thymocytes confirmed elevated expression of Aurka, Bub1 and Anaps5 in TLX1-expressing thymocytes. Furthermore, we demonstrate increased expression of the apoptosis genes, Brca1 and Birc5 in DN2 thymocytes and decreased expression of Chek1 in DN1 thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid mice. Finally, expression of additional genes involved in cell cycle regulation known to be associated with human T-ALL initiation and progression were also assessed. qRT-PCR analysis confirmed upregulated expression of the cell cycle regulators, cyclin B1 and cyclin A and upregulated expression of the oncogenes, c-myc and cmyb, in IgHm-TLX1 Tg Prkdc Scid/Scid thymocytes relative to Prkdc Scid/Scid thymocytes. Combined, these results suggest that TLX1 in collaboration with loss of DNA-PK dysregulates the expression of genes involved in the regulation of the cell cycle, the spindle checkpoint and chromosome segregation.

Increased Cellularity of Thymi from
IgHm-TLX1 Tg Prkdc Scid/Scid Mice We predicted that the effect of TLX1 on cell cycle and apoptotic control may cause aberrant thymocyte proliferation. Therefore, to explore the effects of TLX1 expression on thymus cellularity, assessments of absolute numbers, apoptosis and proliferation of thymocytes were performed using thymi isolated from premalignant IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice. Cell counts revealed a statistically significant increase in the absolute number of thymocytes isolated from thymi of IgHm-TLX1 Tg Prkdc Scid/Scid mice relative to Prkdc Scid/Scid control littermates ( Figure 9A). This increased cellularity was observed in all three DN fractions ( Figure 9B). Cell viability, as assessed by Annexin V and Propidium Iodide (PI) staining ( Figure 9C), revealed an increase in the proportion of viable thymocytes and a reduction in apoptotic and dead thymocytes obtained from IgHm-TLX1 Tg Prkdc Scid/Scid thymi relative to Prkdc Scid/Scid controls ( Figure 9D). Moreover, the proportion of proliferating thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid mice was also significantly increased, as established by in vivo BrdU incorporation ( Figure 9E-F). Combined, these data suggested that the increased cellularity of thymi of premalignant IgHm-TLX1 Tg Prkdc Scid/Scid mice results from

Aneuploidy and Spindle Checkpoint Dysfunction in Premalignant Thymocytes and Thymic Tumors
Previous studies revealed an association of TLX1 overexpression with dysregulated chromosome segregation [5,17]. To determine whether a similar phenomenon was also a feature of malignant TLX1 Tg Prkdc Scid/Scid -derived T cells, we compared chromosome numbers in metaphase spreads prepared from thymocytes and tumor cells obtained from premalignant and terminally ill IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice. Premalignant thymo-cytes were analyzed directly after isolation of thymi and after in vitro culture in the OP9-DL1 co-culture system. These analyses indicated that premalignant thymocytes and tumor cells of IgHm-TLX1 Tg Prkdc Scid/Scid mice had a statistically significant increase in the numbers of aneuploid cells relative to Prkdc Scid/Scid samples ( Figure 10A-B).
Spectral karyotype (SKY) analysis of three tumors isolated from IgHm-TLX1 Tg Prkdc Scid/Scid mice confirmed these results. In addition, these studies revealed a recurrent loss of chromosome 12 and gain of 17 in the cells of two independently arising tumors ( Figure 10C). Moreover, two of three tumors showed an unbalanced trisomy 15 with tumor cells containing two normal chromosome 15 and a  Figure 10D).
We speculated that the abnormal karyotypes detected in thymocytes of IgHm-TLX1 Tg Prkdc Scid/Scid mice were due to dysregulation of the spindle cell cycle checkpoint. To address this possibility, we compared the frequency at which premalignant thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice bypassed the spindle cell cycle checkpoint. Thymocytes derived using the OP9-DL1 co-culture system were arrested with colchicine, a mitotic spindle poison, and the proportion of cells in G1, S or G2/M phase of the cycle determined at eight-hour intervals by analysis of DNA content. In two independently performed experiments, IgHm-TLX1 Tg Prkdc Scid/Scid thymocytes showed a reduced proportion of thymocytes in G2/M at all-time  points assessed, relative to Prkdc Scid/Scid thymocytes, suggesting an increased ability to bypass this checkpoint ( Figure 11A). To further verify the ability of IgHm-TLX1 Tg Prkdc Scid/Scid premalignant thymocytes to bypass the spindle cell cycle checkpoint, cells were subjected to colchicine-induced spindle checkpoint arrest and then cultured in the presence of BrdU. Incorporation of BrdU into the nucleus was used as an indicator of cell division. These analyses indicated a statistically significant increase in BrdU positive thymocytes for IgHm-TLX1 Tg Prkdc Scid/Scid mice relative to thymocytes from Prkdc Scid/Scid mice (30% 64.6 versus 16.3% 65.1; p, 0.05) ( Figure 11B). Taken together, these results indicated that the aneuploidy observed in Igm-TLX1 Tg Prkdc Scid/Scid thymocytes resulted from dysregulated control of the spindle checkpoint in TLX1 overexpressing cells.

Discussion
The first transgenic mouse model confirming the oncogenic potential of TLX1 in vivo was reported by Hough et al [11]. These mice expressed elevated levels of TLX1 throughout all stages of B cell development, during the early stages of thymocyte development and in myeloid progenitors. A premalignant stage involving mature B cells was detected in mice as early as eight weeks of age. Most mice died of a mature B cell lymphoma although ,20% died of myeloid hyperplasia. The average age of death was 14 months. More recently, two transgenic mouse models were described in which ectopic TLX1 expression was restricted to the T cell compartment. These mice develop a T-ALL-like disease with a disease latency of 35 weeks months [5,12] In both reports, there was an extended latency prior to the development of malignant disease, suggesting that ectopic expression of TLX1, by itself, was insufficient to initiate malignancy and that additional mutations were required for premalignant cells to progress to full malignancy.
In this report, we shed light on an additional process, namely the loss of the DNA repair enzyme DNA-PK, which collaborates with TLX1 in neoplastic transformation. IgHm-TLX1 Tg Prkdc Scid/Scid mice rapidly died of T-ALL or AML at a significantly reduced latency compared to Prkdc Scid/Scid control mice. That a proportion of mice developed myeloid malignancies was consistent with our previous report that expression of TLX1 in myeloid progenitors predisposed mice to myeloid hyperplasia [11]. Furthermore, these studies indicated that the dysregulated expression of TLX1 did not influence disease phenotype, but rather, accelerated malignant qRT-PCR analysis of selected genes whose protein products are involved in chromosome segregation (Chek1, Aurka and Bub1), cell cycle progression (Cyclin A, Cyclin B1, Anapc5, c-myc and c-myb) and apoptosis (Birc5, Brca1). Samples are presented as pairs with the first bar representing the level of expression of cells isolated from Prkdc Scid/Scid mice and the second from IgHm-TLX1 Tg Prkdc Scid/Scid mice. Data were normalized relative to b-actin. Red bars: DN1; brown: DN2; green: DN3 cells; black; total unsorted thymocytes. Error bars represent SD. Statistical testing was performed using the student's T-test. Statistically significant differences are indicated by asterisks (* depicts p,0.05, ** depicts p,0.01, *** depicts p,0.001). doi:10.1371/journal.pone.0089649.g008 transformation of not only T-ALL, but also AML in IgHm-TLX1 Tg Prkdc Scid/Scid mice.
Our report aso provides some insights into possible molecular mechanisms by which DNA-PK insufficiency synergizes with TLX1 expression to accelerate T-ALL in mice. Given its role as a key DNA repair enzyme in the NHEJ pathway, we were somewhat surprised that malignant thymocytes from IgHm-TLX1 Tg Prkdc Scid/Scid mice did not exhibit increased incidences of translocations, amplifications or end-to-end telomere fusions. Rather, these cells instead exhibited increased incidences of aneuploidy. It remains unclear whether this reflects a direct or indirect effect of DNA-PK insufficiency. Of note, in addition to its canonical role in the NHEJ pathway, DNA-PK has also been implicated as a regulator of chromosome segregation [22,23], and its loss in IgHm-TLX1 Tg Prkdc Scid/Scid mice may directly compromise spindle checkpoint function. Alternatively, the aneuploidy in IgHm- which is associated with spindle checkpoint dysfunction and a predisposition towards aneuoploidization [5]. Interestingly, in that study, decreased expression of spindle checkpoint factors Aurka and Bub1 was observed, whereas increased expression was noted in the IgHm-TLX1 Tg Prkdc Scid/Scid mice. This discrepancy may be linked to variations in TLX1 expression in the different mouse models, differences in the genetic backgrounds of the mice and a gene expression modulatory effect of DNA-PK deficiency. For several of these spindle checkpoint proteins, however, both increased and decreased expression can been correlated with checkpoint dysfunction [31,32].
It is worth noting that, despite vastly different methodological approaches, the karyotypic abnormalities seen in IgHm-TLX1 Tg Prkdc Scid/Scid mice (+13, +17, -12) recapitulated those reported in other murine models of T-ALL [5]. It is likely that these chromosomes harbor oncogenes and tumor suppressors that may be especially relevant in T-ALL pathogenesis. For example, the putative tumor suppressor gene Bcl11b resides on mouse chromosome 12 and BCL11B is deleted in ,16% of human T-ALL [5]. These overlaps indicate that the pathways which serve to maintain proper ploidy is particularly important in T-ALL biology, and may be a frequent node for disruption for the nascent T-cell malignancy. While, to our knowledge, mutations in the Prkdc gene have not been identified in whole exome sequencing efforts of two cohorts of primary T-ALL patients [33,34], further data mining and sequencing efforts to identify lesions in other components of the NHEJ pathway as well as other DNA repair pathways in T-ALL may be warranted.

Mice
Mice were bred and maintained under pathogen-free conditions and monitored daily for signs of illness. Mice were euthanized and necropsied when moribund. All manipulations were performed under Canadian Animal Care protocols and approved by the Sunnybrook Research Animal Care Committee (Protocol number: 12-005). Double mutant IgHm-TLX1 Tg Prkdc Scid/Scid mice were generated by mating previously described IgHm-TLX1 Tg mice (original name IgHm-HOX11 transgenic mice [12,13]), bred onto the CD1 genetic background, with CB17 ICR-Prkdc Scid/Scid mice bred onto an ICR genetic background (Charles River Laboratories (Wilmington, MA, USA). IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid mice were mated and progeny were genotyped and used to generate cohorts.

Reverse Transcriptase-PCR (RT-PCR) and Quantitative RT-PCR (qRT-PCR)
Genomic DNA was purified using DNAzol (Invitrogen) and total RNA was isolated using an RNeasy kit (Qiagen, Valencia, CA, USA). Total RNA was amplified with MessageAmp TM II aRNA kit (Invitrogen) according to the manufacturer specifications. The IgHm-TLX1 transgene was detected with the following primers: hTLX1-1560F (59-TGGGCATCTATGGGAGAGTG-39) and hTLX1-1799R (59-CTGGATTGGGCTGGGATGT-39). The following PCR conditions were used: initial denaturation at 95uC for 3 minutes; 34 cycles of denaturation (95uC) for 30 seconds, annealing (55uC) for 30 seconds, extension (72uC) for 30 seconds and a final cycle of 7 minutes at 72uC. All reactions were carried out in a 25 ml volume containing 50-100 ng genomic DNA for PCR or 10-50 ng total RNA for reverse transcribed cDNA, and 0.2 mmol of each primer, 10x buffer, 1.5 U AmpliTaq DNA Polymerase and 2.5 mM MgCl 2 (Invitrogen) using the PCR C1000 Thermal Cycler (BioRad). The identity of PCR products was verified by DNA sequencing (The Center for Applied Genomics, The Hospital for Sick Children, Toronto, ON).
First-strand cDNA synthesis was performed using the Super-Script III First Strand cDNA reverse transcriptase kit and random primers (Invitrogen). qRT-PCR was performed with the Quanti-Tect SYBR Green PCR kit (Qiagen) according to the manufacturer's protocol with 1 ml of the reverse transcription reaction product (from 10-50 ng total RNA) amplified for 39 cycles with an ABI PRISM 7000 Applied Biosystems Sequence Detection System and analyzed with ABI Prism 7000 SDS software (Applied Biosystems, Carlsbad, CA, USA). qRT-PCR primers are listed in Table 2.  Co-Culture of Thymocytes on OP9-DL1 Stromal Cells GFP-expressing OP9-DL1 cells (kindly provided by Dr. Juan Carlos Zúñ iga-Pflücker) were cultured in 6-well dishes (BD Falcon) to 80% confluence in aMEM (Gibco) medium containing 20% FBS (Hyclone) and 100 U/ml Penicillin/Streptomycin (Gibco). Cultures were incubated in a humidified atmosphere at 37uC with 5% CO 2 . Fetal livers were dissected from E13.5 embryos and dispersed to single cell suspensions in HBSS supplemented with 10% FBS by passage through a 40 mm nylon cell strainer (BD Falcon). Ten percent of cells were used for genotyping with the remaining cells depleted of lineage-restricted cells using biotinlabeled antibodies specific for mouse lineage specific antigens (B220, CD3e, Gr-1, Mac-1, Ter119 and F4/80; BD Pharmingen). Lineage-positive cells were bound to Miltenyi Biotec streptavidin magnetic beads and depleted using a MACS column according to manufacturer's instructions (Miltenyi Biotech, Auburn, CA USA). Lineage negative (Lin -) HSCs were sorted for expression of the CD117/c-kit (APC) and Sca-1 (PE) antigens on a FACS ARIA flow sorter using II FACS DIVA software (BD Biosciences). Ckit + Sca + Lincells (10 3 ) were resuspended in 3 ml of complete DMEM medium supplemented with murine Flt-3 Ligand (5 mg/ ml), IL-7 (5 mg/ml) and SCF (10 mg/ml; all from R&D Systems, Minneapolis MN, USA) and plated onto an OP9-DL1 feeder layer. Cells were harvested on day-7, passed through a 70 mm cell strainer, stained with CD44 (PE) and CD25 (APC) antibodies and thymocyte subpopulations sorted as follows: CD44 + CD25 -(DN1), CD44 + CD25 + (DN2), CD44 -CD25 + (DN3) and CD44 -CD25 -. OP9-DL1 cells were distinguished from thymocytes based on GFP expression.

Bromodeoxyuridine (BrdU) Labeling and Bypass of the Spindle Cell Cycle Checkpoint
To induce mitotic arrest, day-4 IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid c-kit + Sca + Lin -HSC-derived thymocytes co-cultured on OP9-DL1 were treated with 100 ng/ml colchicine overnight. BrdU was added to a final concentration of 20 mM and incubated at 37uC for 2 hours. T-lymphocytes were collected by centrifugation at 1,300 rpm for 5 minutes and fixed in 80% ethanol, followed by DNA denaturation in 2 N HCl for 1 hour at 37uC. Cells were pretreated by incubation with 1X D-PBS supplemented with 2% BSA for 1 hour at 37uC. Immunostaining was performed using an anti-BrdU antibody for 2 hours at RT, followed by staining with a FITC-conjugated anti-mouse antibody for 1 hour at RT. Counterstaining for detection of nuclei was performed with VECTASHIELD DAPI mounting medium (Vector Laboratories, Inc. Burlingame, CA). Slides were evaluated using fluorescence microscopy (Axiovert 200 M, Zeizz, GmbH, Germany). In all, 20 random fields were scored for the frequency of BrdU positive cells.
To assess spindle checkpoint bypass following mitotic arrest, day 4 IgHm-TLX1 Tg Prkdc Scid/Scid and Prkdc Scid/Scid HSC-derived thymocytes, co-cultured in the OP9-DL1 system, were subjected to a double-thymidine block, treated with 100 ng/ml colchicine then harvested at 8 hour intervals, stained with PI and analyzed by flow cytometry for DNA content as previously described [35]. For each sample, 10,000 events were acquired using a FACS Calibur and cell cycle analysis was performed using FlowJo software (BD Biosciences).

Cytogenetic and Ploidy Analysis of Tumors
Primary leukemic cells were dissected from lymphoid tissues of moribund IgHm-TLX1 Tg Prkdc Scid/Scid mice and immediately processed to single cell suspensions. Fresh tumor samples were placed in RPMI 1640 medium (Wisent; Montreal, Quebec, Canada) containing 10% FBS. Cells were arrested in metaphase by treatment with colcemid (0.03 mg/mL; Invitrogen). Metaphase chromosomes were prepared according to standard cytogenetic protocols, using 0.075 M KCl and Carnoy's fixative. Slides were aged overnight at 55uC then G-banded using 4X USP Pancreatin (diluted to 0.4X) followed by staining with Leishmann/Giemsa stain. 20 metaphase spreads were analyzed and imaged with a CCD camera (VDS Vosskühler GmbH) using an Olympus microscope set in bright-field mode with a 100X/N.A. 1.40 oil immersion objective. Co-ordinates were recorded for metaphase relocation following Spectral karyotyping (SKY). Prior to SKY, slides were destained with Carnoy's fixative, rehydrated with an ethanol series, pretreated with 2X SSC at 37uC, then post-fixed with 1% formaldehyde/PBS/MgCl 2 and washed in PBS. Sky-Paint TM probes were used according to the manufacturer's instructions (Applied Spectral Imaging, Carlsbad, CA). Metaphases were viewed with an Olympus BX61 microscope (Olympus, Center Valley, USA) equipped with a SpectraCube SD300 (Applied Spectral Imaging, Migdal HaEmek, Israel). SKY images were analyzed using SkyView Version 2.1.1 (Applied Spectral Imaging).
Abnormalities in chromosome number of mitotic cells were estimated by calculating the proportion of aneuploid thymocytes and tumor cells relative to cells displaying a normal karyotype. 40 and 100 random metaphases of thymocytes and tumor cells, respectively, were photographed and chromosomes were enumerated using light microscopy (LEICA DM LB2, Bannockburn, IL, USA). Metaphase chromosome spreads were prepared using standard Geimsa staining techniques.

Statistics
Disease-free survival during a 25-month observation period was calculated from the date of birth to the date when mice became terminally ill using the Kaplan-Meier estimator. Column comparison analysis was performed using the logrank test with the GraphPad Prism 4 software. Student's t-test was used to detect differences in cycling thymocytes between IgHm-TLX1 Tg Prkdc Scid/ Scid and Prkdc Scid/Scid cultures. All tests with p,0.05 were accepted as statistically significant.

Sample Isolation and High-density Oligonucleotide Microarray Screening
Double mutant IgHm-TLX1 Tg Prkdc Scid/Scid and control Prkdc Scid/ Scid mice were sacrificed at 4-8 weeks of age. DN1, DN2, DN3 and CD44 -CD25subpopulations were sorted based on expression of CD44 and CD25. To minimize sample variability caused by individual differences among animals, individual thymocyte subpopulations from sex-matched mice were pooled (total 10 4 -10 5 cells). Each pool consisted of between 2-5 age-and sexmatched IgHm-TLX1 Tg Prkdc Scid/Scid or control Prkdc Scid/Scid mice. The pooled cell pellets were homogenized in TRIzol reagent (Invitrogen) and stored at -80uC for total RNA isolation. Total RNA was isolated and purified with RNeasy kits according to the manufacturer's instructions. TLX1 genotypes were confirmed prior to microarray experiments. Microarray processing was performed on the Affymetric platform according to manufacturer's protocols.
Raw microarray CEL data files were imported to GeneSpring GX11 (Silicon Genetics, Redwood City, CA). Data was subjected to background reduction, filtered for genes flagged as Present or Marginal in at least 12.5% of samples, prior to standardization by normalization to the mean of all samples. Statistically significant differentially expressed genes (DEGs) were determined by a twotailed, unpaired t-test, with DEGs defined as those with a p#0.05 and a minimum average fold change of 1.4. Gene annotation queries were conducted through Affymetrix NetAffx Analysis Center, and the Database for Annotation, Visualization and Integrated Discovery (DAVID). Gene Set Enrichment Analysis (GSEA) was performed to analyze enrichment of gene sets following the developer's protocol (http://www.broadinstitute. org/gsea/index.jsp). Analysis was done using 1,000 gene set permutations.

Accession Code
Microarray data are available in the Gene Expression Omnibus (GEO) under accession number GSE47421.