BCR-ABL-transformed GMP as myeloid leukemic stem cells

Abstract

During blast crisis of chronic myelogenous leukemia (CML), abnormal granulocyte macrophage progenitors (GMP) with nuclear β-catenin acquire self-renewal potential and may function as leukemic stem cells (Jamieson et al. N Engl J Med, 2004). To develop a mouse model for CML-initiating GMP, we expressed p210^BCR-ABL in an established line of E2A-knockout mouse BM cells that retain pluripotency in ex vivo culture. Expression of BCR-ABL in these cells reproducibly stimulated myeloid expansion in culture and generated leukemia-initiating cells specifically in the GMP compartment. The leukemogenic GMP displayed higher levels of β-catenin activity than either the nontransformed GMP or the transformed nonGMP, both in culture and in transplanted mouse BM. Although E2A-deficiency may have contributed to the formation of leukemogenic GMP, restoration of E2A-function did not reverse BCR-ABL-induced transformation. These results provide further evidence that BCR-ABL-transformed GMP with abnormal β-catenin activity can function as leukemic stem cells.

Chronic myelogenous leukemia (CML) is caused by the expression of a constitutively active BCR-ABL tyrosine kinase from the abnormal Philadelphia chromosome, which is present in the hematopoietic stem cells and progenitor cells of CML patients (1, 2). The clinical success of the ABL kinase inhibitor imatinib (Gleevec) in treating chronic phase CML serves as a model for molecular targeted therapy of cancer (3–6). However, CML patients in accelerated phase or blast crisis are refractory to imatinib because of the rapid emergence of drug-resistant BCR-ABL mutant clones (7, 8). In CML blast crisis patients, granulocyte macrophage progenitors (GMP) with an aberrant potential for self-renewal were detected (9). This finding suggests that BCR-ABL-transformed GMP may function as the leukemic stem cells during blast crisis (9).

Retroviral transduction of human BCR-ABL into 5-fluorouracil-activated primary mouse bone marrow (BM) cells has been shown to generate a CML-like disease (2), with the leukemia-initiating cells found in the hematopoietic stem cell compartment (10). By contrast, transduction of BCR-ABL into primary GMP isolated from the mouse BM did not lead to the formation of leukemic stem cells (11). To maximize the potential of transforming the myeloid precursors, and to achieve ex vivo propagation of leukemic stem cells, we expressed BCR-ABL in an established long-term culture of murine hematopoietic progenitor cells (mHPC) derived from the BM of E2A-knockout mice, which are defective in B-cell development (12). Previous studies have shown that these cultured mHPC can repopulate the myeloid, erythroid and T/NK compartments when they are cotransplanted with normal BM cells into lethally irradiated congenic mice (12). After retroviral transduction of mHPC with p210^BCR-ABL, we observed an immediate expansion of myeloid cells, including GMP, MEP (megakaryocyte erythroid progenitors) and the myeloid lineage-positive (CD11b⁺) cells in culture. Using this experimental system, we were able to identify, isolate and propagate the cells that cause myeloid leukemia upon transplantation into congenic mice. Similar to what has been reported for blast crisis patients, these leukemia-initiating cells were found to reside in the GMP compartment and display elevated levels of β-catenin activity.

Results

Transformation of Cultured Hematopoietic Progenitor Cells by BCR-ABL.

We infected the mHPC, cultured on top of the S17 stromal cells (12), with an ecotropic retrovirus expressing either GFP alone or p210^BCR-ABL-Internal Ribosome Entry Sequence (IRES)-Green Fluorescence Protein (GFP) (13). The GFP and p210 proteins were detected in GFP⁺ populations isolated by fluorescence-activated cell sorting (FACS) 3 days after infection (Fig. 1 A and B). The levels of p210 protein and tyrosine phosphorylation (Y245) in mHPC/p210 cells were comparable to those in the human blast crisis CML cell line K562 (Fig. 1B). We then expanded the GFP⁺ cells in culture and transplanted 1 × 10⁶ mHPC/GFP or mHPC/p210, along with 2 × 10⁵ freshly isolated normal BM cells, into lethally irradiated congenic mice (12, 14). Mice were killed weekly after transplantation, and the fraction of GFP⁺ cells in the BM, peripheral blood (PB) and spleen was determined. Beginning at 6 weeks posttransplantation, a dramatic expansion of GFP⁺ cells in the BM, PB and spleen of mice transplanted with mHPC/p210 was observed (supporting information (SI) Fig. S1A). At week 7, the mHPC/p210-transplanted mice displayed an increase in the number of white blood cells (WBC) in the periphery (Fig. 1C). By week 8, the mHPC/p210-transplanted mice displayed phenotypes of BM hypercellularity (Fig. 1D and Fig. S1C), splenomegaly (Fig. 1E, Fig. S1 B and C), and infiltration of leukemic cells into the liver (Fig. S1C). Of the 20 mice transplanted with mHPC/p210, 18 had to be killed at week 8, whereas none of the mHPC/GFP-transplanted mice (n = 6) developed leukemia (Fig. 1F).

Fig. 1.

Transformation of murine pluripotent hematopoietic progenitor cells (mHPC) by BCR-ABL. (A) Retroviral-mediated expression of GFP. Three days after infection with the MSCV-IRES-GFP or the MSCV-p210-IRES-GFP retrovirus, GFP-expression was determined by FACS, with mock-infected mHPC as negative control. (B) Expression of BCR-ABL and GFP. Lysates (50 μg of protein) of the indicated cells were probed with anti-pTyr245-Abl, anti-Abl (8E9), and anti-GFP antibodies by immunoblotting. The levels of the endogenous ABL protein were used as loading control. (C) Peripheral blood analyses. PB samples from the indicated mice were collected at 7 weeks post transplantation. The white blood cell (WBC) counts, the levels of hemoglobin (Hb), and the platelet (PLT) counts are shown as mean ± SEM. from the indicated number (n) of mice. A representative Wright-Giemsa staining (original magnification, ×1,000) of PB from an mHPC/p210-transplanted mouse is shown. (D) Expansion of GFP⁺ cells in mHPC/p210-transplanted mice. C57BL/6J-Tyr^c-2J/J mice (CD45.1, 8 weeks old) were transplanted as described in Experimental Procedures. Percentages of GFP⁺ cells in the BMs from mice killed at 8 weeks post transplantation are shown. The data shown are mean ± SEM. from the indicated number (n) of mice. (E) Splenomegaly in mHPC/p210-leukemic mice. The spleen weights are shown as mean ± SEM. determined from the indicated number (n) of mice. (F) Summary of transplantation experiments. Leukemia was scored by two criteria: splenomegaly (> 400 mg) and dominant GFP-positive cells (> 70%) in BMs by FACS. NBM: normal BM.

Transplantation of total BM cells from the leukemic mice into secondary recipients propagated the leukemic phenotypes. The disease latency was reduced in secondary and tertiary recipients as BM hypercellularity and splenomegaly were observed 4 weeks after transplantation with either 1 × 10⁶ or 1 × 10⁵ BM cells from primary or secondary leukemic mice (Fig. 1F and data not shown). BM cells from mHPC/GFP-transplanted primary recipients did not cause leukemia in secondary recipients (Fig. 1F). In primary and secondary leukemic mice derived from one batch of mHPC/p210, Southern blot analysis of splenic DNA revealed two EcoRI fragments suggesting an oligoclonal, transplantable disease (Fig. S2B, left panels). In a separate set of secondary leukemic mice, derived from mHPC/p210 generated in an independent retroviral transduction experiment, we observed one predominant EcoRI band, suggesting a monoclonal disease (Fig. S2B, right panel). Thus, expression of p210^BCR-ABL in mHPC reproducibly generates transplantable leukemic cells that can be monoclonal or oligoclonal in origin.

BCR-ABL-Transformed Leukemia-Initiating Cells Reside in the GMP Compartment.

Previous studies have shown that the mHPC express low levels of multiple lineage markers at the level of mRNA and a predominant expression of the B-lineage marker B220 at the cell surface (12). Similarly, we found that cells in the mHPC/GFP culture were mostly positive for B220 (Fig. 2A). By contrast, the majority (≈80%) of the mHPC/p210 culture was positive for CD11b, a myeloid lineage marker (Fig. 2A). The shift from B220⁺ to CD11b⁺ in the mHPC/p210 cultures occurred within 7 days of retroviral infection (Fig. 2A). Microarray-based gene profiling results also indicated an up-regulation of myeloid lineage markers in the mHPC/p210 cultures (Fig. S3 A and B). The mHPC/p210 cells readily acquired cytokine and stromal independent growth (Fig. S3C). Interestingly, however, transplantation of mHPC/p210 grown in the absence of cytokines and S17 stromal cells did not induce leukemia in congenic mice (data not shown), suggesting that cytokine- and stromal-independent proliferation is not sufficient to confer leukemogenic potential.

Fig. 2.

BCR-ABL-transformed leukemia-initiating cells occur in the GMP population. (A) Expansion of CD11b⁺ cells in mHPC/p210 cultures. The expression of GFP and two lineage markers: B220 and CD11b was determined by FACS at 7 days post infection with MSCV-IRES-GFP (Vector), MSCV-p210-IRES-GFP (p210) or MSCV-p210KD-IRES-GFP. (B) Analyses of myeloid progenitors in ex vivo-cultures. Infected cultures within 2 passages (7 days) after GFP-sorting were used for the analyses of GMP (Lin⁻, IL-7R⁻, Sca-1⁻, c-Kit⁺, CD34⁺, FcγR⁺) and MEP (Lin⁻, IL-7R⁻, Sca-1⁻, c-Kit⁺, CD34⁻, FcγR⁻) as percentages of total GFP⁺ cells. The data shown are mean ± SEM. from three independently derived mHPC/GFP and mHPC/p210 cultures. (C) Kaplan–Meier survival curves. The indicated numbers of GFP⁺-GMP and GFP⁺-nonGMP sorted from the mHPC/p210 cultures were mixed with normal BM cells (2 × 10⁵) and injected into lethally irradiated recipient mice. The disease free survival of the indicated number of mice was monitored for 90 days. (D) Analyses of progenitors in BM cells. BM samples from mice transplanted with mHPC/GFP or mHPC/p210 were collected at 8 weeks post transplantation. BM samples from untreated healthy mice were also collected to detect distribution of normal myeloid progenitors in the BM cells. Percentages show each population of total GFP-positive cells. The data are shown as mean ± SEM. of the indicated number (n) of mice. (E) Kaplan–Meier survival curves. The indicated number (5 to 50,000) of GFP⁺-GMP from primary mHPC/p210-leukemic mice (8 weeks post transplantation) were mixed with normal BM cells (2 × 10⁵) and injected into the indicated number (n) of lethally irradiated recipient mice. The disease free survival of the indicated number of mice was monitored for 80 days.

Expression of p210^BCR-ABL also affected the lineage-negative population (i.e., CD3⁻, CD5⁻, CD8⁻, CD11b⁻, Gr-1⁻, B220⁻ and Ter119⁻). In particular, the GFP⁺-GMP fraction (Lin⁻, IL-7R⁻, Sca-1⁻, c-Kit⁺, CD34⁺, FcγR⁺) in the mHPC/p210 culture was 0.3%, which was 10-fold higher than the GFP⁺-GMP fraction in the mHPC/GFP culture (Fig. 2B). We isolated the GFP⁺-GMP population by stepwise sorting of Lin⁻/Sca-1⁻/IL-7R⁻, GFP⁺/c-Kit⁺ and FcγR⁺/CD34⁺ (Fig. S4B) and transplanted them along with normal BM cells into lethally irradiated congenic mice (Fig. 2C). We found that injection of as few as 50 such cells from the mHPC/p210 culture induced leukemia in three of six mice (Fig. 2C and Fig. S4A). We also transplanted cells from the GMP-depleted fraction (GFP⁺-nonGMP) for comparison. Whereas transplantation of 5,000 GFP⁺-GMP induced leukemia in five of five mice, only one of seven mice transplanted with GFP⁺-nonGMP developed leukemia (Fig. 2C and Fig. S4A).

We next sorted the GFP⁺-GMP population from the leukemic mouse BM to determine its leukemia-initiating potential (Fig. 2D and Fig. S4B). The GMP and MEP fractions in BM of nontransplanted mice were similar to those previously reported for C57BL/6 mice (Fig. 2D) (15, 16, 17). Eight weeks after transplantation with mHPC/p210, the GFP⁺-GMP fraction in leukemic BM was 10–15-fold higher than that in mice receiving mHPC/GFP (Fig. 2D). By contrast, the GFP⁺-MEP fractions of mHPC/p210 and mHPC/GFP transplanted mice were similar (Fig. 2D). Thus, the expansion of the p210-transformed GMP population occurred both ex vivo and in vivo.

The GFP⁺-GMP and GFP⁺-MEP sorted from leukemic mice were then injected along with normal BM helper cells into lethally irradiated secondary recipients (Fig. 2E and Fig. S4C). Again, as few as 50 cells from the GFP⁺-GMP population isolated from the leukemic BM were sufficient to transfer the disease to secondary recipients (Fig. 2E and Fig. S4C). In contrast, the sorted GFP⁺-MEP did not induce leukemia, nor did the GFP⁺-GMP isolated from the BM of mHPC/GFP-transplanted mice (Fig. S4C).

To ascertain that the GFP⁺-GMP from mHPC/p210 cultures were functional myeloid progenitors, we plated the isolated cells in culture, with or without the S17 stromal cells, and analyzed the formation of CD11b⁺ (a myeloid lineage marker) cells with time (Fig. S5). Upon replating, the GFP⁺-GMP expanded to form CD11b⁺ cells and the conversion to these differentiated myeloid cells was slower in the presence of S17 stromal cells (Fig. S5), indicating S17 may promote the maintenance of the leukemic GMP. Taken together, the above results show that the BCR-ABL transformed GFP⁺-GMP could generate myeloid lineage cells in culture and myeloid leukemia in mice.

BCR-ABL Kinase Activity Is Required to Initiate the Transformation of mHPC.

The results of numerous previous studies have demonstrated that BCR-ABL tyrosine kinase activity is essential for its transforming function (1–6). We expressed a kinase-defective p210 (p210KD) (18) in mHPC and found that this mutant did not cause an expansion of GFP⁺ or CD11b⁺ cells (Fig. 2A). Expression of p210KD also failed to cause the expansion of myeloid progenitors (GMP and MEP) (Fig. 2B). Furthermore, transplantation of mHPC/p210KD into congenic mice failed to induce leukemia or the expansion of GMP in the BM (Fig. 1 D and E, and Fig. 2D). Thus, BCR-ABL tyrosine kinase activity is required for the transformation of mHPC.

Restoration of E2A-Function Does Not Suppress the Transforming Function of p210.

As the mHPC culture was derived from the BM of E2A-knockout mice (12), we sought to determine whether restoration of E2A function could reverse the transformation of GMP. Therefore, we stably infected mHPC/p210 cultures with a retrovirus expressing an E2A-Estrogen Receptor (ER) fusion protein and the cell surface marker CD25 (Fig. 3A). We selected CD25/GFP-double positive cells, verified the expression of E2A-ER (Fig. 3B), and demonstrated 4-hydroxytamoxifen (4-OHT)-dependent activation of E2A-target genes: Id2 and Hes-1 (Fig. 3C) (19). The stable expression and activation of E2A-ER did not alter the fraction of CD11b⁺ cells, which remained at the ≈90% level (Fig. 3D). Similarly, expression of E2A-ER and treatment with 4-OHT did not affect the fraction of GMP, which remained at the 0.3% level (Fig. 3D). Thus, E2A restoration did not affect the lineage distribution of mHPC/p210.

Fig. 3.

Restoration of E2A-function does not eliminate p210-transformed GMP. (A) The pCSretTAC-E2A-ER-IRES-hCD25 retroviral construct. LTR: retroviral long terminal repeat; E2A-ER: murine E47-human estrogen receptor fusion protein; IRES: internal ribosome entry sequence; hCD25: human CD25. (B) Expression of E2A-ER. The indicated cells were treated with 4-OHT (1 μM, 24 h) on S17 stromal cells plus cytokines, and total lysates (50 μg of protein) were immunoblotted with anti-E2A and anti-tubulin antibodies. (C) Activation of E2A-regulated genes after 4-hydroxytamoxifen (4-OHT)-treatment. E2A-ER-expressing mHPC/p210 cells were treated with 4-OHT (1 μM, 24 h) and the levels of Id2, Hes-1 and GAPDH RNA were determined by quantitative real-time PCR as described in Methods. The ΔCT values were standardized by actin and the standardized mean values without 4-OHT-treatment were set as 1. The data shown are mean ± SEM from three independent experiments. (D) Analysis of GFP⁺-CD11b⁺ and GFP⁺-GMP cells. The mHPC/p210 and mHPC/p210/E2A-ER cells were cultured under the indicated conditions with or without 4-OHT (1 μM) for 24 h. The percentages of CD11b⁺ and GMP cells among the total GFP⁺ cells were determined by FACS. The data shown are mean ± SEM. from three independent experiments.

Higher Levels of β-Catenin Activity in BCR-ABL-Transformed GMP.

It was reported that the self-renewal potential of GMP from blast crisis CML patients was a result of the aberrant activation of β-catenin (9). To measure β-catenin activity, we infected mHPC/GFP and mHPC/p210 cultures with a lentiviral virus carrying a “Top” reporter in which 7 TCF/LEF sites lie upstream of the luciferase gene (Fig. 4 A and B). The resultant cells, mHPC/GFP-Top or mHPC/p210-Top, were cultured ex vivo, sorted into GMP and nonGMP, and the β-catenin activity in the four populations assessed by quantification of the luciferase mRNA. The GMP fraction of mHPC/p210-Top cells exhibited a significant increase in β-catenin activity relative to the other three populations (Fig. 4C). We also injected mHPC/GFP-Top or mHPC/p210-Top cells into mice and found that only mice transplanted with mHPC/p210-Top developed leukemia (data not shown). We sorted GMP and nonGMP from the mouse BM at 3 weeks after transplantation, measured the luciferase mRNA and found that whereas the GFP-Top/GMP and p210-Top/nonGMP had slightly elevated levels of luciferase mRNA relative to the GFP-Top/nonGMP population, the p210-Top/GMP cells had a nearly 20-fold increase in luciferase expression (Fig. 4D). Thus, cells in the p210-transformed GMP compartment exhibited an increase in β-catenin activity that was further enhanced by the in vivo BM microenvironment. Taken together, these results suggest that leukemia induced by mHPC/p210 is propagated by GMP cells with abnormal β-catenin activity.

Fig. 4.

BCR-ABL-transformed GMP displayed higher levels of β-catenin activity than the nonGMP cells. (A) Schematic of lenti-Topflash. Seven TCF/LEF binding sites upstream of a minimal promoter (P_TA) drive expression of firefly luciferase. The pgk-puro sequence is inserted downstream of the luciferase gene and allows infected cells to be selected with puromycin. (B) Infection of mHPC cells with lenti-Topflash. Total RNA was collected from control or puromycin-selected cells after infection with lenti-Topflash and reverse transcribed into cDNA. Infection was monitored by PCR using primers specific for the luciferase gene. (C) β-Catenin activity from ex vivo-cultured mHPC/p210-Top. Three batches of ex vivo-cultured mHPC/GFP-Top or mHPC/p210-Top cells were sorted into GFP⁺-GMP and GFP⁺-nonGMP fractions. cDNA was made from total RNA and transcript levels of the luciferase reporter were determined by quantitative RT-PCR. The ΔCT values were standardized by those of actin and the standardized mean values of GFP⁺-nonGMP from the mHPC/GFP cultures were set as 1. The data shown are mean ± SEM. from three biological repeats and triplicate PCRs per sample. (D) β-Catenin activity from mouse BM. 1 × 10⁶ mHPC/GFP-Top or mHPC/p210-Top cells were injected into each of 3 mice along with 2 × 10⁵ freshly isolated normal BM cells. Three weeks after injection, BM cells were collected and sorted into GFP⁺-GMP and GFP⁺-nonGMP fractions, and quantitative RT-PCR was carried out and analyses were done as described in Fig. 4C. The standardized mean values of GFP⁺-nonGMP from mHPC/GFP-Top-transplanted mice were set as 1. The data shown are mean ± SEM from three independently transplanted mice with triplicate PCRs per sample.

Discussion

By expressing BCR-ABL in an ex vivo culture of E2A-knockout pluripotent progenitor cells (mHPC), we have generated BCR-ABL-transformed GMP that exhibit higher β-catenin activity and acquire the ability to initiate leukemia in mice. Our finding is consistent with the report that BCR-ABL-positive GMP isolated from CML blast crisis patients acquire self-renewal potential through activation of the β-catenin pathway (9). A recent study using β-catenin-deficient mice has also confirmed the importance of this self-renewal pathway in BCR-ABL-dependent myeloid leukemogenesis (20). It has been shown that BCR-ABL cannot convert primary GMP isolated from the mouse BM into leukemic stem cells (11). We show here that BCR-ABL can transform ex vivo cultured, E2A-deficient, mHPC into leukemia initiating cells, which are found in the GMP compartment. Together, these results suggest that BCR-ABL-induced formation of leukemogenic GMP may require other genetic and/or epigenetic alterations that have occurred in the E2A-deficient mHPC during ex vivo culture. We have examined the levels of β-catenin protein in the GMP and nonGMP fractions isolated from the BM of wild type and E2A-knockout mice, as well as from ex vivo cultures of mHPC/p210 (Fig. S6). We did not observe any significant differences in the levels of total β-catenin between the wild type and E2A-deficient GMP, thus ruling out abnormal β-catenin levels as the permissive factor for transformation of E2A-deficient GMP by BCR-ABL. We did observe a twofold higher β-catenin level in ex vivo-cultured p210/GMP (Fig. S6), consistent with the increase in reporter activity (Fig. 4C). Because the restoration of E2A function did not suppress the propagation of GMP in mHPC/p210 cultures, it appears that E2A itself is unable to reverse the transforming activity of BCR-ABL. The factors that collaborate with BCR-ABL to convert GMP into leukemic stem cells remain to be identified.

A unique feature of the experimental system developed here is the ex vivo propagation of BCR-ABL-transformed GMP and nonGMP in the same cultures (Fig. 5). At steady state, the mHPC/p210 cultures contain a very low level of GMP (≈0.3%) and a high level of CD11b⁺ myeloid cells (≈80%). The CD11b⁺ cells are likely to be derived from the GMP as demonstrated by the replating experiments (Fig. S5). This mixed culture provides an experimental system to compare the pathobiological properties of BCR-ABL-transformed leukemic stem cells (GMP) and BCR-ABL-transformed leukemic cells (nonGMP). It has been shown that BCR-ABL kinase can directly activate β-catenin (21, 22). Those studies observed a 2 to 3-fold activation of the Topflash reporter in transient cotransfection experiments employing established cell lines. Using a stably integrated Topflash in the mHPC/p210 experimental system, we also detected a twofold increase in the nonGMP population of mHPC/p210 relative to mHPC/GFP. However, we observed a further increase in the reporter activity in the GMP population of mHPC/p210. This increase was even more dramatic with the GMP population isolated from the BM of mHPC/210-transplanted mice. These results suggest that the activation of β-catenin by BCR-ABL is modulated by the cell context (GMP versus nonGMP) and possibly also the microenvironment (S17 stromal cells versus BM). Gene expression profiling experiments have suggested that deregulation of the β-catenin pathway is a hallmark of CML disease progression (23). If the activation of β-catenin by BCR-ABL is more efficient in the leukemic stem cells, the overall activity of this pathway detected by gene profiling experiments may represent the relative levels of leukemic stem cells in blast crisis CML patient samples. The mHPC/p210 experimental system developed in this study may be useful in testing the efficacy of anti-β-catenin therapies in the treatment of blast crisis CML.

Fig. 5.

Summary of mHPC transformation by p210^BCR-ABL. Expansion of GMP (minority) and CD11b⁺ cells (majority) in mHPC cultures requires p210^BCR-ABL kinase activity. Once established, the CD11b⁺ and GMP cells coexist in the same cultures. The leukemic stem cells reside in the GMP population with high β-catenin activity. The differentiated myeloid leukemic cells with low β-catenin activity develop cytokine and stromal independence but do not cause leukemia in mice.

Methods

Cell Culture and Retroviral Infection.

Pluripotent murine hematopoietic progenitor cells (mHPC) derived from E2A-deficient mouse (C57BL/6; CD45.2) were cocultured with the S17 stromal cells plus stem cell factor (SCF), Flt-3-ligand (FL) and interleukin-7 (IL-7) (R&D Systems) as previously described (12). The MSCV-IRES-GFP, the MSCV-p210-IRES-GFP and the MSCV-p210KD-IRES-GFP plasmids were kindly provided by R. Ren (Brandeis University, Waltham, MA) (13). The pCSretTAC-E2A-ER-IRES-hCD25 vector was previously described (24). Retrovirus production was performed by using the Phoenix Retroviral Packaging System (Orbigen) and FuGENE 6 (Roche Applied Science) for DNA transfection. Viral supernatants (500 μl per well, 24-well plates) with polybrene (Sigma-Aldrich, final concentration; 2 μg/ml) were added to the mHPC (at 10⁶/100 μl per well), the plates were centrifuged at 2,500 rpm for 1.5 h at 30°C. After centrifugation, the viral supernatants were removed and complete media (2 ml per well) was added. After another 6-hour incubation, infected cells were transferred onto 6-well plates coated with S17 stromal cells for expansion using the established mHPC culture conditions (12).

Mouse Transplantation and Analysis for Leukemic Phenotype.

Recipient C57BL/6J-Tyr^c-2J/J mice (CD45.1, 8 weeks old) were irradiated (1,000 rads) between 6 to 8 h before transplantation. GFP⁺ cells (1 × 10⁶) from the mHPC/GFP or mHPC/p210 cultures in 100 μl of IMDM media with 1% FBS were mixed with 2 × 10⁵ normal BM cells from C57BL/6J-Tyr^c-2J/J mice (CD45.1, 8 weeks old) in 100 μl of media and injected into the tail veins of recipient mice (12, 14). Mice were monitored daily for cachexia, lethargy and ruff coats and the distressed animals were killed (25). Peripheral blood collection and blood examinations were performed by the UCSD Murine Hematology and Coagulation Core Facility, and histopathological analyses were performed by the UCSD Histology and Immunohistochemistry Core Facility.

Statistical Analysis.

Kaplan–Meier analysis and statistical analysis were performed by using PRISM 4 (2003) from GraphPad as previously described (25). Data were expressed as mean ± SEM. determined from the indicated number. Values of P < 0.05 was considered statistically significant.

Flow-Cytometric Analysis and Cell Sorting.

Flow-cytometric analysis and sorting were performed with FACS Calibur and FACS Aria (Becton Dickinson), using established methods (9, 16, 26). Biotin-conjugated antibodies against seven lineage markers (CD3, CD5, CD8, CD11b, Gr-1, B220 and Ter119), Sca-1 and IL-7R, as well as APC-anti-CD11b, PE-anti-B220, APC-anti-cKit, PE-anti-FcγR, PerCP-Cy5.5-anti-CD34 and PE-anti-Sca-1 antibodies were from eBioscience. PE-Cy7-anti-streptavidin and APC-anti-human CD25 antibodies were from CALTAG Laboratories. Staining with propidium iodide (PI) to measure PI-uptake (intact cells) and DNA contents were performed with standard procedures as previously described (27).

Immunoblotting.

Anti-Abl antibody (8E9) and anti-E2A antibody were from BD Biosciences. Anti-phospho-c-Abl (Tyr-245) and anti-β-catenin were from Cell Signaling Technology. Anti-Tublin antibody and anti-GFP antibody were from Sigma-Aldrich. To measure protein concentrations of cell lysates, Bio-Rad Protein Assay Kit was used. (Bio-Rad Laboratories). Immunoblotting were performed with standard protocols as previously described (28, 29), and developed with enhanced chemiluminescence (Amersham).

Restoration of E2A-Function.

The E2A-ER-IRES-hCD25-infected mHPC/p210 cells were selected by hCD25-expression and PI-negativity using FACS Aria. The levels of E2A-ER protein were determined by immunoblotting with anti-E2A antibody. The activity of E2A-ER was induced by treating cells with 4-hydroxytamoxifen (4-OHT) (Sigma-Aldrich) at 1 μM for 24–48 h as previously described (19).

Real-Time Quantitative PCR.

Total RNA was prepared by using the RNeasy kit (Qiagen), and cDNA prepared with SuperScript II reverse transcriptase (Invitrogen) or the High-Capacity cDNA reverse transcription kit (Applied Biosystems). Real-time quantitative PCR using the Power SYBR Green PCR master mix (Applied Biosystems) was conducted by using the 7900 HT Fast Real-time PCR System (Applied Biosystems). The primers used were: Id2 (forward) CGGAAGGAAAACTAAGGATG and (reverse) TGTAGAAAGGGCACTGAAAG; Hes1 (forward) TTGGCTGAAAGTTACTGTGG and (reverse) ACTATTCCAGGACCAAGGAG (19); Luciferase (forward) TGCACATATCGAGGTGGACATC and (reverse) GCCAACCGAACGGACATTT; Actin (forward) CGAGAAGATGACCCAGATCATGTT and (reverse) CCTCGTAGATGGGCACAGTGT.

β-Catenin Reporter.

The Topflash sequence was amplified from the Super8XTopflash plasmid by PCR and cloned into a self-inactivating (SIN) lentiviral vector. The pgk-puro sequence was amplified from pMSCV-puro (Clontech) and inserted downstream of the Topflash sequence. The lenti-Topflash plasmid and lentiviral packaging plasmids were then transfected into 293FT cells (Invitrogen) using calcium phosphate. Media was changed 18 h after transfection and virus from 12 15-cm dishes was harvested 48 h later and concentrated into a final volume of 400 μl. mHPC/GFP and mHPC/p210 were then infected with 50 μl viral supernatant containing 8 μg/ml polybrene. After 48 h, cells were replated in 10-cm dishes and infected cells were selected by addition of 0.5 mg/ml puromycin (Sigma-Aldrich) to the media. For in vivo-experiments, mice were injected with mHPC/GFP-Top or mHPC/p210-Top as described above. Three weeks after injection, mice were killed and GFP-positive cells were sorted into GMP and nonGMP populations and luciferase expression was determined by real-time quantitative PCR.