Early T cell precursor acute lymphoblastic leukemia (ETP-ALL) exhibits lymphoid, myeloid, and stem cell features and is associated with a poor prognosis. Whole genome sequencing of human ETP-ALL cases has identified recurrent mutations in signaling, histone modification, and hematopoietic development genes but it remains to be determined which of these abnormalities are sufficient to initiate leukemia. We show that activating mutations in the interleukin-7 receptor identified in human pediatric ETP-ALL cases are sufficient to generate ETP-ALL in mice transplanted with primitive transduced thymocytes from p19Arf−/− mice. The cellular mechanism by which these mutant receptors induce ETP-ALL is the block of thymocyte differentiation at the double negative 2 stage at which myeloid lineage and T lymphocyte developmental potential coexist. Analyses of samples from pediatric ETP-ALL cases and our murine ETP-ALL model show uniformly high levels of LMO2 expression, very low to undetectable levels of BCL11B expression, and a relative lack of activating NOTCH1 mutations. We report that pharmacological blockade of Jak–Stat signaling with ruxolitinib has significant antileukemic activity in this ETP-ALL model. This new murine model recapitulates several important cellular and molecular features of ETP-ALL and should be useful to further define novel therapeutic approaches for this aggressive leukemia.
Early T cell precursor acute lymphoblastic leukemia (ETP-ALL) is a recently described subtype of acute lymphoblastic leukemia that occurs in both adults and children and has a relatively poor survival rate with current therapies (Coustan-Smith et al., 2009). The leukemic blasts in ETP-ALL have a unique phenotype characterized by cytoplasmic expression of CD3, a lack of expression of mature T cell markers such as CD4 and CD8, and aberrant expression of myeloid and stem cell markers. There has been no animal model for ETP-ALL, so the biology of the disease and identification of new therapeutics remains relatively unexplored.
The prevailing hypothesis is that ETP-ALL is caused by transformation of a primitive hematopoietic cell that retains the capacity to differentiate into both T cells and myeloid cells. The thymus is seeded by primitive thymic immigrants derived from the BM that then proceed through a series of maturational steps, ultimately generating CD4 and CD8 single-positive T cells (Rothenberg et al., 2010). The initial stages of thymocyte development are characterized by differentiation of cells that lack expression of CD4 or CD8. As these double negative (DN) cells differentiate, at least four distinct differentiation stages can be distinguished by differential expression of CD44 and CD25 (DN1, DN2, DN3, and DN4). The potential for myeloid, dendritic, and natural killer cell differentiation is retained at the DN1 stage and at the early DN2 stage (Bell and Bhandoola, 2008). The ability to adopt non–T cell fates is lost by the DN3 stage and most likely by the latter half of DN2 progression (Yui et al., 2010). Therefore, it seems plausible that the tumor-initiating cell in ETP-ALL could be derived from DN1 and/or DN2 thymocytes.
Whole-genome sequencing studies in ETP-ALL have discovered several recurrent mutations involving genes that participate in cytokine signaling, epigenetic control of gene expression, and hematopoietic transcriptional regulation. In particular, activating mutations in the IL7R pathway were seen in five of 54 cases of pediatric ETP-ALL (Zhang et al., 2012). Most of these IL7R mutants can confer cytokine independent proliferation in various cell lines (Shochat et al., 2011; Zenatti et al., 2011; Zhang et al., 2012); however, there is no proof that these signaling mutants are sufficient to initiate ETP leukemia in primary cells. Furthermore, it is unclear how these mutations can lead the specific phenotypic features of ETP-ALL and what other collaborative mutations may be required. Therefore, we tested Il7r mutant alleles homologous to those that have been identified in human ETP cases (Zhang et al., 2012) in a mouse thymocyte transplant assay to determine if they were sufficient to generate ETP-ALL.
Our experimental system is based on transducing CD4−CD8− thymocytes from Arf−/− mice with retroviral vectors expressing either of two constitutively active Il7r mutant receptors. These transduced thymocytes were then transplanted into sublethally irradiated recipients by tail vein injection. The use of Arf−/− thymocytes allows efficient thymic engraftment due to enhanced self-renewal associated with p19Arf loss (Treanor et al., 2011). Transplanted mice were followed over time for the development of leukemia and the resulting malignancies were characterized at a cellular and molecular level. Relevant molecular abnormalities seen in the murine model system were then evaluated in human ETP-ALL samples from pediatric cases identified at St. Jude Children’s Research Hospital to determine if they were present in both human and murine ETP-ALL.
Constitutively active, Il7r mutants block thymocyte differentiation at the DN2 stage and induce ETP-like ALL in transplanted mice
A significant number of human pediatric ETP-ALL cases contain mutations in cell signaling pathways with 20% of cases harboring activating mutations in the IL7R or in the downstream Janus kinases JAK1 and JAK3 (Zhang et al., 2012). These IL7R mutations are most commonly in-frame insertions that introduce a cysteine in the transmembrane domain and result in ligand-independent receptor dimerization and cytokine-independent proliferation in cell lines and primary murine BM cells (Zhang et al., 2012). To determine whether these IL7R mutations are oncogenic drivers in ETP-ALL, DN thymocytes from Arf−/− mice were transduced with MSCV vectors that expressed murine homologues of either of two IL7R mutations identified in human ETP-ALL, IL7R-241-242TC and IL7R-GCinsL243 (Zhang et al., 2012). When these transduced thymocytes were cultured on OP9-DL1 stromal cells, the Il7r mutants caused a partial block at the DN2 stage (Fig. 1 A), similar to the block seen with enforced expression of the Lmo2 transcription factor (Larson et al., 1995; Neale et al., 1995; Treanor et al., 2011). These results show that these mutant Il7r receptors induce a block in differentiation early in thymocyte development and at a stage before irreversible commitment to the T cell lineage. Given that a DN2 block was also seen with enforced Lmo2 expression, we measured Lmo2 mRNA expression in cultured thymocytes transduced with the Il7r-241-241TC vector. Increasing accumulation of endogenous Lmo2 mRNA was observed as the Il7r-241-241TC–transduced cells were cultured for 20 d (Fig. 1 B). These results suggest that signaling from the mutant Il7r receptor specifically induces Lmo2 expression and thereby leads to the block at the DN2 stage of maturation.
We next tested whether these transduced DN2 thymocytes would induce leukemia in transplanted recipient mice. Thymocytes transduced with the WT or mutant Il7r vectors or empty GFP-expressing vector (GFP) were isolated and transplanted into sublethally irradiated, Rag2−/−, γc−/− recipient mice. Mice transplanted with thymocytes transduced with the mutant Il7r vectors died of leukemia with a median latency of 71 d (Fig. 1 C). By 100 d after transplant, all 11 mice that received cells transduced with either of the two Il7r mutants had died from acute leukemia (Fig. 1 C). In contrast, none of the mice transplanted with Arf−/− thymocytes transduced with the MIG or the Il7r WT vector developed leukemia over this period of time.
Each Il7r mutant receptor induces a distinct ETP-ALL phenotype
The Il7r-241-242TC Il7r mutant resulted in an aggressive acute leukemia in six mice, with marked elevation of the peripheral leukocyte count, significant splenomegaly, and myeloid morphology of leukemic cells (Fig. 2 A). Leukemic cells expressed the vector-encoded GFP gene, Gr1, Mac1, and myeloperoxidase myeloid markers (Fig. 2, B and C). Leukemic infiltration of the kidney and other solid organs was noted and the tumor cells were uniformly positive for expression of the vector-encoded GFP gene (Fig. 2, B and C). In most of these cases, the myeloid tumor cells also expressed intracellular CD3, a hallmark for the ETP-ALL phenotype (Table 1). Monoclonal rearrangement of the TCRβ loci was demonstrated via PCR (King et al., 2002) in tumor cell DNA derived from the spleen of two independent cases, proving that these tumors originated from an early thymocyte and not a contaminating myeloid cell (Fig. 2 D).
A stem cell leukemia with erythroid features was observed in five of six mice transplanted with Il7r-GCinsL243 expressing thymocytes (Fig. 3 and Table 1). The peripheral blood of leukemic mice contained numerous immature erythroid precursors and blast cells (Fig. 3 A). Tumor cells were present at high levels in the spleen, thymus, BM, and peripheral blood and expressed the vector-encoded GFP gene (Fig. 3 B). Leukemic cells also expressed high levels of the erythroid marker CD71 but did not express the Ter119 erythrocyte marker, as has been described for other murine erythroleukemias (Torchia et al., 2007). Leukemia cells also expressed Gata1 and Runx1 by immunohistochemistry, consistent with an early erythroid/stem cell phenotype, and also expressed intracellular CD3 as would be expected for ETP-ALL (Fig. 3 C). Monoclonal rearrangement of the TCRβ loci was verified in three cases, indicating the thymocyte origin of these tumors (Fig. 3 D). This phenotype suggests a previously unrecognized erythroid potential of early DN thymocytes, perhaps manifest only in the setting of Il7r mutant-induced leukemic transformation.
Tumor cells from both a myeloid and a stem/erythroid ETP-like ALL case were transplanted into secondary recipients and resulted in aggressive leukemias at ∼24 d post-transplant. The tumor-initiating cell frequency in leukemic cells was ∼1/1,500 for the Il7r-241-242TC–induced cases (Table 2). This relatively high tumor-initiating cell frequency is consistent with what has been previously noted in Notch1-induced tumors originating from Arf−/− thymocytes (Volanakis et al., 2009). Altogether, these results demonstrate that overexpression of activating Il7r mutants in early Arf−/− thymocytes are potent oncogenic drivers and function in part by blocking thymocyte development in a manner similar to that seen with Lmo2 overexpression. Interestingly, each mutant displayed a unique and specific phenotype, perhaps reflecting quantitative and/or qualitative differences in constitutive Il7r signaling with each mutant.
LMO2 overexpression is seen in Il7r mutant ETP tumors and is sufficient to generate ETP-ALL
Because the Il7r mutant receptors induce a DN2 block in transduced thymocytes that recapitulates that observed with enforced LMO2 expression, we measured endogenous Lmo2 protein expression in murine ETP-ALL tumor cells by Western blot analysis. Relatively high levels of Lmo2 protein expression were seen in splenic tumor samples arising from both the 241-242TC mutant receptor and from the GCins243 receptor (Fig. 4 A) and approximated that seen in the retroviral producer clone for the MSCV-Lmo2-GFP vector (+ve).
To determine if enforced Lmo2 expression was sufficient to induce murine ETP, we transduced DN thymocytes with an Lmo2 expressing retroviral vector and transplanted the cells into irradiated recipient mice. Thymocytes from reconstituted primary recipients were then transplanted into secondary recipients. ETP-ALL leukemias arose in three independent cases in transplanted secondary recipients that showed a myeloid phenotype and CD3 coexpression (Fig. 4 B) and rearrangement of the TCR-β chain (Fig. 4 C). Note that thymic granulocytes do have faint TCR rearrangements and are known to arise from early thymocyte precursors. Immature thymocyte populations were sorted from normal mice (DN1-DN3) and showed polyclonal TCR rearrangements, as expected. Enforced Lmo2 expression in Arf-null thymocytes blocked development at the DN2a stage (unpublished data) and results in an expanding thymocyte population with significant myeloid differentiation capacity demonstrated in myeloid CFU-C assays (Fig. 4 D). The accumulation of DN2 cells with myeloid potential appears to be caused by a block in differentiation rather than a developmental reversion arising in mature thymocytes (unpublished data).
Alterations in Bcl11b and Notch1 signaling in murine ETP-ALL
BCL11B is a transcription factor that is required for thymocytes to progress beyond the DN2 checkpoint and is required for irreversible commitment to the T cell lineage (Li et al., 2010). Mutations in BCL11B have also been associated with T-ALL and may play a role in leukemic transformation (Gutierrez et al., 2011). BCL11B expression is also known to be negatively regulated by IL-7 signaling during normal thymocyte development (Ikawa et al., 2010), suggesting the possibility that the constitutive signal from Il7r mutants could repress Bcl11b expression in ETP-ALL. To test this possibility, splenic leukemia cells derived from either the Il7r-241-242TC or GCinsL243 vectors were analyzed by Western blot analysis and shown to have little or no Bcl11b expression (Fig. 5 A). Specifically, no Bcl11b was detected in nuclear extracts from three independent splenic tumors initiated with Il7r-241-242TC or in one tumor from the GCins243 mutant. In contrast, classical CD4+ CD8+ murine T-ALLs derived either from LMO2 overexpression in hematopoietic stem cells (HSCs; Treanor et al., 2011) or from the one case arising from Il7r GCinsL243 mutant (Table 1) had significantly elevated levels of Bcl11b expression (Fig. 5 A, T cell). Likewise, no Bcl11b expression was detectable in nuclear extracts from four independent splenic ETP-ALL tumors initiated with the Lmo2 vector (Fig. 5 B). These results show that repression of Bcl11b is characteristic of murine ETP-ALL and may distinguish ETP-ALL from classic T-ALL cases. Furthermore, it is possible that repression of Bcl11b is necessary for the developmental block in DN thymocyte progression seen with either the Il7r mutants or the Lmo2 vector.
NOTCH1 mutations leading to constitutively increased signaling are frequently present in classical human T-ALL cases (Weng et al., 2004), but are significantly less frequent in human ETP-ALL (Zhang et al., 2012). To determine if this was also true in the murine ETP-ALL tumor samples, the level of stabilized intracellular Notch1 (ICN) was evaluated in several murine ETP-ALL samples derived from each of the two Il7r mutants. ICN was not expressed in any of five independent murine ETP-ALL samples derived from either of the two Il7r mutations (Fig. 5 C). Notch activation was detected in a control case of Lmo2-derived classical T-ALL as previously described (Treanor et al., 2011), and in the one case of CD4+, CD8+ T-ALL that was derived from the Il7r-GCinsL243 vector (Fig. 5 C). These studies show that murine ETP-ALL, like human ETP-ALL, lacks Notch 1 activation and this feature distinguishes ETP-ALL from classical T-ALL. We also detected activation of p38 Map kinase in ETP-ALL tumor samples arising from all three vectors in contrast to very low levels of activation in murine CD4+ CD8+ T-ALL (Fig. 5 D).
Gene expression patterns in human pediatric ETP-ALL cases parallel that seen in the murine ETP-ALL model
Having established that our mouse model phenotypically resembled human ETP-ALL, we next tested whether the murine model recapitulated the molecular gene expression profile of human ETP-ALL using gene set enrichment analysis (GSEA) from expression array data. Murine ETP-ALL shared statistically significant positive correlations with several gene expression sets that were highly correlated with human ETP-ALL (Zhang et al., 2012), including the profile from human hematopoietic stem cells and from human granulocyte-macrophage precursors and negative correlation with human early thymic precursor cells (Fig. 6 A). Furthermore, the murine ETP-ALL cases highly resembled a GSEA expression profile derived from the top 500 most overexpressed genes in human ETP-ALL (Zhang et al., 2012). These results show that the molecular characteristics of the murine model parallel those seen in human ETP-ALL.
Also paralleling murine ETP-ALL, human pediatric ETP-ALL specifically show high levels of LMO2 mRNA expression and low levels of BCL11B expression. Analysis of microarray (Fig. 6 B) and quantitative RT-PCR (Fig. 6 C) data of 12 ETP cases versus 40 non-ETP T-ALL cases showed higher LMO2 expression in ETP versus non-ETP T-ALL and lower levels of Bcl11b expression. All ETP cases had LMO2 mRNA levels that were equivalent to a Jurkat T cell clone (G-25) with engineered retroviral vector insertions in the LMO2 locus (Ryu et al., 2008) and K562 cells, both of which express high levels of LMO2 protein on Western blot analysis. In contrast, the classical T-ALL cases showed variable and, on average, lower levels of LMO2 mRNA. A highly significant decrease in BCL11B expression was noted in the 12 pediatric ETP cases versus classical T-ALL in the expression array studies (Fig. 6 B). Together, these results demonstrate that high levels of LMO2 expression, repression of Bcl11b, and lack of Notch1 mutations are hallmarks of ETP-ALL in both human pediatric cases and in our mouse model and suggest that these alterations may be functionally important for the development of ETP-ALL.
Targeting JAK–STAT signaling in murine ETP-ALL cells with ruxolitinib
Mutations in the α chain of the IL7R receptor that lead to ligand independent activation function via activation of JAK1 kinase and the STAT5 transcription factor but do not require common gamma chain or JAK3 signaling (Zenatti et al., 2011), suggesting that pharmacologic inhibition of this pathway could inhibit growth of ETP-ALL tumor cells. Phospho-flow analysis for activated Stat5 was performed on ETP-ALL cells derived from the Il7r-241-241TC vector and showed Stat5 phosphorylation from three independent ETP-ALL lines (Fig. 7 A). Stat5 phosphorylation was detected only in cells containing the Il7r mutant vectors (GFP+) but not in GFP− nontransduced control cells (Fig. 7 A). The pStat5 signaling was augmented by the phosphatase inhibitor pervanadate and treatment with exogenous IL7.
Ruxolitinib is a selective drug inhibitor of Jak1 and 2 that is approved for the treatment of myelofibrosis due to activating JAK2 mutations (Harrison et al., 2012). To determine whether Lmo2 expression was linked to Jak–Stat activation, thymocytes transduced with the Il7r-241-241TC vector were cultured in the presence of ruxolitinib and Lmo2 mRNA expression was measured at 96 h of exposure. Endogenous Lmo2 mRNA expression was decreased by ruxolitinib treatment (Fig. 7 B) relative to DMSO controls demonstrating that Lmo2 expression is at least in part controlled by the mutant Il7r signal.
To determine whether ruxolitinib would have an antiproliferative effect on ETP-ALL cells, we cultured two independent Il7R-241-241TC ETP-ALL cell lines in physiologically achievable concentrations of drug and measured cell proliferation relative to control cultures. Both cell lines showed a significant decrease in cell growth after 96 h of culture in 0.5 µM ruxolitinib relative to DMSO control cultures (Fig. 7 C). Finally, we tested whether ruxolitinib treatment of mice inoculated with these two ETP-ALL cell lines would have antileukemic effects in vivo. Mice were treated with 60 mg/kg of ruxolitinib via oral gavage twice daily beginning 2 d after transplant of ETP-ALL cells. In the S801.2 group, all of the animals treated with vehicle succumbed to leukemia by 25 d after transplant (Fig. 7 D), whereas animals receiving ruxolitinib survived until days 28 and 29 (P < 0.0001). In the second ETP-ALL line, S1520.2, all of the animals treated with vehicle died of leukemia by 39 d after transplant, whereas the ruxolitinib-treated group showed a significantly prolonged survival, with some mice living until day 44 (P < 0.0001; Fig. 7 D). These results demonstrate that ruxolitinib has significant antileukemic activity in vivo as a single agent. Further studies will be required to understand the mechanisms of leukemic breakthrough in this model and whether other agents can be added to achieve longer disease-free intervals.
Several recent studies have suggested that constitutively active IL7R mutants may be oncogenic drivers in ETP-ALL (Shochat et al., 2011; Zenatti et al., 2011), but did not test transformation capacity in primary hematopoietic cells or define how these mutant receptors contribute to the ETP-ALL phenotype. We describe the first de novo mouse model of ETP-ALL showing that two different Il7r mutant alleles, homologous to those identified in human ETP-ALL, are sufficient to induce murine ETP-ALL when expressed in primitive, Arf-null thymocytes, proving that mutant IL7R receptors can be leukemia drivers in vivo and cause an ETP-ALL like disease in mice. This murine disease highly resembles human ETP-ALL pathologically, immunophenotypically, and molecularly with regard to gene expression profiles. These activating mutations in Il7r result in activation of Stat5 in primary tumor cells, suggesting that inhibition of Jak activation could be an effective therapy. We indeed show that ruxolitinib has significant antileukemic activity in vivo, indicated by a significant prolongation of survival in treated mice with ETP-ALL. This result indicates that the ETP-ALL may require continual Il7r signaling for maintenance and growth. This mouse model should also be useful for studying the activity of other therapeutics and for studying mechanisms of treatment resistance.
In addition to providing information about therapy for ETP-ALL, this work illustrates several biological insights into the mechanism of ETP-ALL. All oncogenes that cause ETP-ALL in our model lead to a block in thymocyte development at the DN2 stage, resulting in transformation at a stage in T cell development at which significant stem cell and myeloid potential is retained. At the DN2 stage, thymocytes retain the ability to generate myeloid, NK, and dendritic cells (Bell and Bhandoola, 2008; Wada et al., 2008); however, the physiological significance of this multipotentiality has been questioned (Schlenner et al., 2010; Schlenner and Rodewald, 2010; Richie Ehrlich et al., 2011). Recent work has shown that thymic granulocytes do arise from early thymocyte progenitor cells (De Obaldia et al., 2013). Our results further establish the significance of this developmental plasticity by showing that the cellular mechanism of ETP-ALL results from transformation of an early thymocyte that has retained myeloid differentiation potential.
The expression patterns of LMO2, BCL11B, and NOTCH1/ICN1 transcription factors clearly distinguished ETP-ALL cases from non-ETP T-lineage ALL in both human and murine tumors and suggest that these alterations are characteristic of this leukemia subtype. The cellular stage at which LMO2 overexpression and/or IL7R mutations are acquired could determine whether ETP versus non-ETP-ALL arises; e.g., if the mutations are acquired after DN2 progression, classical T-ALL would likely be favored. In this regard, it is notable that activating IL7R mutants have been identified in classical T-ALL cases as well as in ETP-ALL (Zenatti et al., 2011; Zhang et al., 2012). The absolute expression level of LMO2 may also determine whether ETP T-ALL versus non-ETP-ALL will arise. In previous studies in transgenic mice using nonviral promoters to express LMO2, a partial DN3 block was noted before development of classical T-ALL (Larson et al., 1995; Neale et al., 1995, 1997). Our experiments use the stronger MSCV retroviral promoter to express Lmo2, resulting in an earlier, partial DN2 block. We suggest that the critical DN2-DN3 checkpoint may be sensitive to the absolute level of Lmo2 expression in transitioning thymocytes and that differentiation block before the DN2 stage is required for ETP-ALL. Consistent with this idea, high levels of Lmo2 are strongly selected against in mature CD4+/CD8+ T-ALLs (Treanor et al., 2011). Another possibility, that is not mutually exclusive, is that additional cooperating mutations may play a dominant role in determining the leukemia phenotype.
We propose a model for ETP-ALL in which transformation is initiated at a cell stage at or before DN2 because of altered expression of key transcription factors such as LMO2 and/or BCL11b. During normal T cell development, LMO2 is expressed at high levels in hematopoietic stem cells, in the earliest thymic immigrants from the BM, and in DN1 cells. Expression of LMO2 is then sharply down-regulated as cells pass from DN2a to DN2b (Herblot et al., 2000). At this critical transition, IL7R signaling decreases and Bcl11b is induced, permitting commitment to the T cell lineage (Ikawa et al., 2010). For instance, deletion of Bcl11b results in a DN2 differentiation block and loss of development of mature T cells (Wakabayashi et al., 2003). Thymocytes that are blocked at this stage retain both myeloid and T cell differentiation potential and the phenotype and biological behavior of these leukemic blasts reflect this bipotentiality. The exact mechanisms by which Lmo2, Bcl11b, and Il7r signaling are related are not fully understood, however we do show that pharmacologic inhibition of Il7r signaling results in decreased expression of Lmo2, suggesting that the pathways are linked.
ETP-ALL tumors in mice do not express activated Notch1, and Notch1 mutations are relatively infrequent in human cases. Notch 1 signaling normally increases after the DN2 checkpoint and is necessary for development of double- and single-positive thymocytes (Tanigaki and Honjo, 2007). Activating Notch1 mutations are also commonly seen in classical T-ALL and presumably reflect a requirement for this signal in the transformation process. The lack of Notch1 signaling in our murine ETP-ALLs most likely reflects the stage-specific independence from Notch1 signals at this early stage of thymocyte development.
Currently, only ∼25% of pediatric patients with ETP-ALL are cured of their disease and new therapeutic approaches need to be identified. Drug sensitivity studies on cultured human leukemic blasts can be used to screen for new therapeutic approaches (Holleman et al., 2004); however, primary leukemia cells often require complex culture conditions to grow and proliferate (Nishigaki et al., 1997) and this assay may not represent drug sensitivity in the actual leukemia-initiating/stem cells. Another approach is to establish xenografts in immunodeficient mice using human leukemia cells, as has been done for acute myeloid leukemia (Eppert et al., 2011); however, primary ETP-ALL samples are not widely available for this purpose. Therefore, our mouse model for ETP-ALL has value in providing an approach to screen and test candidate drugs for antileukemic activity, such as we have done with ruxolitinib. Although we do detect significant antileukemic activity, it is not surprising that this single drug therapy does not result in curative eradication of leukemia, at least not in our initial studies. Newer drugs and combination therapy will likely be required for more durable responses. Other possible targets are the characteristic alterations in transcription factor expression in ETP-ALL that we describe. Prior studies have shown that tumor growth can depend upon deregulated expression of oncogenic transcription factors (Felsher and Bishop, 1999; Felsher, 2003; Zuber et al., 2011), but drugs for targeting transcription factors have been more difficult to identify than those associated with oncogene addiction based on signaling proteins, such as RAS (Singh et al., 2009) and BCR-ABL1 kinases (Shah et al., 2008). However, there has been recent success in targeting specific oncogenic transcription factors in the Myc family, so that targeting certain oncogenic transcription factors may be possible for ETP-ALL.
MATERIALS AND METHODS
Viral vectors and production.
A cDNA encoding mouse Lmo2 (National Center for Biotechnology Information reference sequence BC057880.1) was inserted into a mouse stem cell vector containing an internal ribosome entry site–GFP gene reporter (MSCV-IRES-GFP) as previously described (Shou et al., 2006). To facilitate experiments in which p19Arfgfp/gfp cells were used (Sherr, 2004), analogous mCherry vectors were constructed by removing the GFP cassette and replacing it with a cDNA encoding mCherry. A polyclonal population of ecotropic retroviral producer GPE-86 cells was generated for each vector and freshly collected supernatants containing high-titer vector were generated.
The Il7R IL241-242TC and Il7R GCinsL243 vectors have been described previously (Zhang et al., 2012). For production of ecotropic retroviral supernatants, vectors were prepared by transient transfection of vector plasmids and the pMD gagpol and pCAG4-Eco packaging plasmids.
Arf−/− (Kamijo et al., 1997)and Arfgf/gfp mice (Sherr, 2004) are both p19Arf null strains, were all on a backcrossed C57BL/6J background and were provided by C. Sherr (St. Jude Children’s Research Hospital, Memphis, TN). They were used between 8 and 12 wk of age. The experiments in Figs. 1 and 2 used the Arfgf/gfp mice, all the other experiments were done using Arf−/− mice. Female C57BL/6J and Rag2−/− γc−/− mice (Taconic) were used as transplantation recipients. Transgenic mice on a B6 background carrying an enhanced GFP gene whose expression is controlled by the proximal lck promoter (plck-GFP mice) were a gift from T. Nakayama (Chiba University, Chiba, Japan).
All experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee of St. Jude Children’s Research Hospital.
Thymi were explanted from 4–8-wk-old animals. Whole thymocyte populations were stained with CD4-phycoerythrin (PE) and CD8-PE antibodies (BD), incubated at 4°C, washed with MACS buffer (PBS, 0.5% BSA [Sigma-Aldrich], and 2 mM EDTA), and incubated with anti-PE microbeads (Miltenyi Biotec) for 15 min. Cells washed with MACS buffer were centrifuged and placed on a LS column (Miltenyi Biotec), and unattached CD4−/CD8− cells were collected. An aliquot was analyzed by flow cytometry to ascertain the purity of the “double-negative” thymocyte population.
For transduction with the Lmo2 vector the CD4/CD8-depleted thymocytes were co-cultured for 48 h at a density of 0.5–0.7 × 106 cells/well with irradiated (30 Gy) vector producer cells in α-MEM (Invitrogen) containing 20% FBS, penicillin/streptomycin, glutamine, sodium pyruvate (Invitrogen), and polybrene. After 48 h, thymocytes were removed and plated onto confluent OP9-DL1 stromal cells (gift from J.C. Zuniga-Pflucker, University of Toronto, Toronto, Canada) in 12-well plates in complete medium containing 5 ng/ml each of FLT-3 (R&D Systems) and IL-7 (PeproTech). For the MSCV-IRES-GFP, Il7R-WT, and mutant vectors the cells were plated on OP9-DL1 for 24 h. After 24 h, the cells were transduced with 1 ml of vector supernatant collected from 293T cells, spun at 2,000 RPM for 1 h at room temperature. This was repeated 24 h later. Every 3–5 d, thymocytes were replated onto fresh OP9-DL1 cells. Cells harvested at indicated times were immunophenotyped by flow cytometric analysis.
Thymocyte transplantation and colony forming unit assay.
Vector-transduced thymocytes were removed from OP9-DL1 cultures on day 21 and stained for expression of CD4, CD8, CD44, and CD25 (antibodies from BD). Fluorescence-activated cell sorting was used to isolate vector-positive (mCherry+) CD4/CD8 double-negative, CD44/CD25 DN cells, 105 to 3 × 105 of which were transplanted by tail vein injection into sublethally irradiated (6Gy) female γc−/−Rag2−/− mice.
Lmo2, Arf−/− DN2-sorted thymocytes were plated at 33,000 cells/plate in MethoCult GF M3434 (StemCell Technologies) for myeloid progenitor CFU-C assays. 7 d after plating, primary colonies were counted, pooled, and either DNA was extracted or they were analyzed by flow cytometry.
Immunoblotting and immunohistochemistry.
The M-PER and NE-PER mammalian protein extraction reagents (Thermo Fisher Scientific) were used to prepare lysates from tissues for LMO2, Notch1, and Bcl11b immunoblots together with the Halt protease inhibitor cocktail (Thermo Fisher Scientific). Protein concentration was quantified by the Bradford method (Bio-Rad Laboratories). Samples (30–40 µg protein per lane) were electrophoretically separated on 4–12% Bis-Tris NuPAGE gels (Invitrogen), transferred to polyvinylidene fluoride membranes (Invitrogen), and probed using a goat polyclonal antiserum to LMO2 (R&D Systems) at a 1/500 dilution, a rabbit monoclonal antibody to Notch1 cleaved at val1744 (D3B8; Cell Signaling Technology) at a 1/250 dilution, and a rabbit polyclonal antiserum to Bcl11b at ½,000. The secondary antibodies were either a HRP-conjugated rabbit anti–goat (EMD Millipore) or donkey anti–rabbit (GE Healthcare) used at a dilution of 1/1,000. Antibodies to β-actin (Abcam) were used to control for protein loading.
Immunohistochemistry was done using standard protocols. In brief, slides of 4–6-µm sections were cut from formalin-fixed paraffin-embedded tissues, deparaffinized, and rehydrated, and then epitope retrieval was performed. Slides were probed with a rabbit polyclonal antiserum to GFP (Invitrogen) at a 1/200 dilution, a goat polyclonal antiserum to CD3 (Santa Cruz Biotechnology, Inc.) at a 1/350 dilution, a rabbit monoclonal to RUNX1 (Abcam) at a 1/600 dilution, a goat polyclonal antiserum to GATA-1 at a 1/500 dilution, and a rabbit polyclonal antiserum to Myeloperoxidase (MPO; DAKO) at a 1/1,500 dilution.
TCR-β receptor analysis.
The method was done as previously described (King et al., 2002). In brief, genomic DNA was isolated and amplified by a PCR touchdown method (10 s at 94°C, 30 s at 68–63°C, 2 min at 72°C), followed by a 15-cycle PCR (10 s at 94°C, 30 s at 63°C, 2 min at 72°C) with primers Dβ1.1ext and Jβ1.7ext. For analysis of TCR-β rearrangement in DN1-DN4 populations, colonies and tumor samples the PCR products were separated on a 1.2% gel and 0.2–2 kb was excised to exclude germline bands. The DNA was purified by spin column (QIAquick gel extraction kit; QIAGEN). 0.5–1-µl aliquot from the first amplification was subjected to the second PCR with nested primers Dβ1.1int and Jβ1.7 for 24–30 cycles.
Tumor samples were obtained from diagnostic BM aspirates or peripheral blood, and comprised at least 90% tumor cells from 12 children with ETP-ALL and 40 children with non-ETP-T-ALL treated at St. Jude Children’s Research Hospital. All cases fulfilled pathological and immunophenotypic criteria for ETP-ALL. The study was approved by the Institutional Review Boards of St. Jude Children‘s Research Hospital.
Gene expression profiling and real-time quantitative RT-PCR.
We examined LMO2 expression in the published U133 Plus 2 data from 12 ETP and 40 non-ETP-T-lineage ALL samples that have been previously reported (Coustan-Smith et al., 2009). Expression signals were normalized in log2 scale by the RMA algorithm.
For qRT-PCR assays, reverse transcription was done with SuperScript VILO cDNA Synthesis kit (Invitrogen). qPCR was done for human with the LMO2 assay (Hs00153473_m1) or the mouse Lmo2 assay (Mm01281680; Applied Biosystems), and HPRT1 (Applied Biosystems) was used for the human or GAPDH (Applied Biosystems) for the mouse endogenous control in a step one plus instrument. Relative LMO2 expression level was calculated as the ratio of LMO2/HPRT1 or Lmo2/GAPDH using standard curves for each of the two transcripts.
50,000 sorted GFP+ Il7r-GCinsL243 and Il7r241-242TC ETP leukemic cells were plated in triplicate in a 96-well plate. The cells were cultured in DMEM with 15% FBS with Il-3, IL-6, and stem cell factor containing either 0.5 µM ruxolitinib (LC laboratories) or DMSO (vehicle control; Thermo Fisher Scientific). Cell viability was measured by incubating the cells with Cell Titer-Blue (Promega) for 4 h and reading the plate on a FlUOstar Omega plate reader (BMG Labtech).
Thymocytes were cultured as above on OP9-DL1 and either 1 µM ruxolitinib (LC Laboratories) or DMSO (vehicle control; Thermo Fisher Scientific) were added and the cells were analyzed for cell surface markers by flow cytometry as indicated. For RNA analysis, the cells were sorted for GFP− and GFP+ populations and RNA was extracted using the RNAqueous-micro kit (Ambion). Lmo2 RNA levels were measured using the same qRT-PCR assay as above.
For the in vivo studies, animals were sublethally irradiated and transplanted with 5,000 cells from 2 independent ETP-ALL cell lines, S801.2 and S1520.2. 2 d after transplant, treatment was initiated with ruxolitinib (Chemitek) at a dose of 60 mg/kg twice a day by oral gavage or vehicle (5% methylcellulose; Sigma-Aldrich) twice daily for 27 d. Animals were monitored for signs of illness or hematological abnormalities and the endpoint was survival.
ETP leukemic cells were placed in serum and cytokine-free RPMI for 4 h at 37°C. After 4 h, 8 × 105 ETP leukemic cells were treated with pervanadate as a positive control. Upon removal from culture the ETP leukemic cells were fixed and permeabilized (eBioscience). Nonspecific binding was eliminated using 50 µg/ml rabbit IgG and, after blocking, the cells were incubated with a rabbit monoclonal antibody to pStat5 conjugated with Alexa Fluor 647 (Cell Signaling Technology). The cells were resuspended in PBS and their fluorescence was analyzed on a LSR flow cytometer (BD).
The authors thank J.C. Zuniga-Pflucker for providing the OP9-DL1 cells. We thank the St. Jude Children’s Research Hospital flow cytometry core facility for assistance, Nicole Lantz for animal husbandry, David Finkelstein for computational support, and members of the Sorrentino laboratory for constructive suggestions and criticisms throughout the course of these studies.
This work was supported by the National Heart, Lung, and Blood Institute (grant P01 HL 53749), the National Cancer Center (support grant P30 CA 21765), the Assisi Foundation of Memphis, and the American Lebanese Syrian Associated Charities.
The authors declare no competing financial interests.
B cell lymphoma/leukemia 11B
colony forming unit in culture
early thymic precursor leukemia acute lymphoblastic leukemia
gene set enrichment analysis
intracellular notch 1
mitogen activated protein kinase 38 kD
signal transducer and activator of transcription 5
T cell acute lymphoblastic leukemia