Expression of BCR/ABL and BCL-2 in myeloid progenitors leads to myeloid leukemias

Abstract

Chronic myelogenous leukemia is a myeloproliferative disorder (MPD) that, over time, progresses to acute leukemia. Both processes are closely associated with the t(9;22) chromosomal translocation that creates the BCR/ABL fusion gene in hematopoietic stem cells (HSCs) and their progeny. Chronic myelogenous leukemia is therefore classified as an HSC disorder in which a clone of multipotent HSCs is likely to be malignantly transformed, although direct evidence for malignant t(9;22)⁺ HSCs is lacking. To test whether HSC malignancy is required, we generated hMRP8p210^BCR/ABL transgenic mice in which expression of BCR/ABL is absent in HSCs and targeted exclusively to myeloid progenitors and their myelomonocytic progeny. Four of 13 BCR/ABL transgenic founders developed a chronic MPD, but only one progressed to blast crisis. To address whether additional oncogenic events are required for progression to acute disease, we crossed hMRP8p210^BCR/ABL mice to apoptosis-resistant hMRP8BCL-2 mice. Of 18 double-transgenic animals, 9 developed acute myeloid leukemias that were transplantable to wild-type recipients. Taken together, these data indicate that a MPD can arise in mice without expression of BCR/ABL in HSCs and that additional mutations inhibiting programmed cell death may be critical in the transition of this disease to blast-crisis leukemia.

Human chronic myelogenous leukemia (CML) is a biphasic myeloproliferative disorder (MPD) that comprises ≈15% of adult leukemias (1). The chronic phase is characterized by high white blood cell (WBC) counts, extramedullary hematopoiesis, and splenomegaly. After a period of 3–5 years, the number of immature myeloid cells rapidly expands as the disease progresses to a fatal acute phase characterized by an overwhelming number of immature blast cells.

Over 90% of CML patients possess the Philadelphia (9;22) chromosomal translocation (2, 3). This translocation results in the joining of the BCR and C-ABL genes (4), forming the fusion gene BCR/ABL that produces a chimeric 210-kDa protein (p210^BCR/ABL) with deregulated tyrosine kinase activity (5). The t(9;22) translocation can be found in clonal patterns in granulocytic, monocytic, megakaryocytic, and erythroid populations (6) and in populations containing hematopoietic stem cells (HSCs) (7–9). This has led to CML being classified as a stem cell disorder. Whereas these results strongly suggest that the t(9;22) translocation occurs in HSCs, expression of the p210^BCR/ABL fusion protein and mRNA transcripts is found primarily in mature cells of the myelomonocytic lineage and is reported to be absent in primitive CD34⁺Lin^– progenitors (10). The p210^BCR/ABL transcript is undetectable in the CD34⁺Thy1⁺Lin^– HSC (11) population (7), suggesting that BCR/ABL may transform cells downstream of HSCs.

Several attempts to model CML in mice have targeted the p210^BCR/ABL fusion gene to hematopoietic cells by retroviral transduction or germ-line transgenesis. Transplantation of mouse bone marrow retrovirally transduced with p210^BCR/ABL (12–16) into irradiated hosts can produce animals with a CML-like disease. Although this approach can show a CML-like disease, it often shows a lack of chronic phase, a preponderance of lymphoid malignancies, and variable phenotypes apparently due to integration effects, cell type(s) infected, and/or the genetic background of the mice. Several lines of p210^BCR/ABL transgenic mice have been created by using promoter elements from the metallothionein (MT) (17, 18) and Tec (19) genes. Whereas hematopoietic neoplasms arising in MT p210^BCR/ABL mice were exclusively lymphoid, offspring from Tecp210^BCR/ABL transgenic mice appear to most faithfully model human CML. Although the Tec gene is preferentially expressed in hematopoietic cells, its expression in stem and progenitor cells, to our knowledge, has not been addressed.

Here we present a mouse model of myeloid malignancy whereby expression of p210^BCR/ABL is targeted to myeloid progenitors (20) and their myelomonocytic progeny alone or in cooperation with the protooncogenes BCL-2, RAS, or MYC through the use of the human MRP8 promoter element (21). In this model, hMRP8 p210^BCR/ABL transgenic mice develop a chronic MPD in the absence of HSC expression. Additionally, coexpression of Bcl-2 but not Ras or Myc under the same promoter is synergistic in promoting acute myeloid leukemia (AML).

Materials and Methods

Mice. cDNA encoding for the 210-kDa BCR/ABL fusion protein was excised from the SRα p210^BCR/ABL TKneo vector (kindly provided by Owen Witte, University of California, Los Angeles) by digestion with XbaI and subcloned into the hMRP8 cassette as described (21). Transgenic mice were generated from (C3H × C57BL/6) F₁ hybrid mice according to standard procedures and backcrossed onto C57BL/Ka-Thy1.1 mice (referred to as wild type in this report) for at least three generations to minimize strain differences.

To study oncogene collaboration, double-transgenic mice were generated by crossing hMRP8p210^BCR/ABL F₄ mice to previously generated hMRP8BCL-2/2 animals (C57BL/Ka-Thy1.1) (21), hMRP8C-MYC mice (D.T., E.L., and I.L.W., unpublished results), and hMRP8N-RAS mice (kindly provided by S. Kogan and J. M. Bishop, University of California, San Francisco). All animals were maintained in Stanford University's Research Animal Facility in accordance with Stanford guidelines.

Screening. hMRP8p210^BCR/ABL mice were screened by using PCR. All reactions were performed in a volume of 50 μl containing 5 μl of 10× PCR buffer (50 mM KCl/10 mM Tris/2 mM MgCl₂/0.01% gelatin), 1 μl of 10mM dNTPs, 2 units of Taq polymerase (GIBCO), 1 μl of 10 pM primer for forward (5′-CAGGGTGCACAGCCGCAACGGCAA-3′) and reverse (5′-GTCCAGCGAGAAGGTTTTCCTTGGA-3′) primers, and 100–200 ng of DNA (obtained from mouse tails by using Proteinase K digestion followed by alkaline lysis). PCR conditions for amplification of the p210^BCR/ABL transgene were 40 cycles of denaturation (30 sec at 94°C), annealing (30 sec at 60°C), and extension (1 min at 72°C). The PCR for p210^BCR/ABL generated a fragment length of 379 bp. PCR primers were synthesized at the Protein and Nucleic Acid Facility of Stanford University. hMRP8BCL-2 mice were screened by flow cytometry of peripheral blood as described (21, 22).

Flow Cytometry. Cell suspensions were prepared from hematopoietic tissues and stained for fluorescence-activated cell sorter (FACS) analyses by using monoclonal antibodies to the Mac-1, Gr-1, B220, CD3, CD45.1, CD45.2, c-Kit, CD47, Sca-1, CD34, FcγRIII, and BCL-2 proteins. Staining for neutrophils, B lymphocytes, and T lymphocytes (20) as well as myeloid-restricted progenitors (20) and HSCs (23) was performed as reported, and the cells were sorted by using four- or five-color staining on a highly modified Becton Dickinson FACSVantage at the Stanford University Shared FACS Facility. All antibodies except anti-human BCL-2 (FITC-coupled, DAKO) were produced in our laboratory and directly conjugated to phycoerythrin, FITC, allophycocyanin, Texas red, or biotin. Data were analyzed with flowjo software developed by Tree Star (San Carlos, CA) as described (24).

RT-PCR. HSCs, common myeloid progenitors (CMPs), B cells, T cells, and neutrophils were triple-sorted (sorted, resorted, and resorted again) by using the antibodies discussed above and the methods described (20). For each reaction, RNA was extracted from a starting population of 1,000 cells by using MS2 phage carrier RNA (Roche Molecular Biochemicals) as reported (25). Reverse transcription and subsequent semiquantitative, nested PCR for p210^BCR/ABL (26) and murine Mrp8 (27) were performed as described.

Histology and Statistical Analyses. Marrow, spleen, and blood morphological analyses and counts were performed as described (24). Results are shown as arithmetic means ± SDs.

Cell Transfers. A mixture of 5 × 10⁶ marrow cells and 1 × 10⁷ splenocytes from leukemic animals (C57BL/Ka-Thy1.1, CD45.2) were injected into the retroorbital sinusoids of lethally irradiated (920 rad) congenic recipient animals (C57BL/Ka-Thy1.1, CD45.1). All irradiated mice used in this study were between 8 and 10 weeks of age.

Results

To determine whether BCR/ABL expression in cells downstream of HSCs could lead to myeloid malignancies, we developed lines of transgenic mice that expressed human p210^BCR/ABL cDNA under the control of the hMRP8 promoter (Fig. 1A). The hMRP8 promoter has been shown to direct expression exclusively in myelomonocytic cells, initiating with low-level expression in CMPs and their granulocyte/monocyte progenitor (GMP) progeny (27), and is highly expressed in mature monocytes and neutrophils (21). Injections of this construct into C3H × C57BL/6 F₁ hybrid mice yielded 13 transgenic founders A–M.

Fig. 1.

(A) hMRP8p210^BCR/ABL construction. (B) RT-PCR analysis of BCR/ABL transcript expression from whole bone marrow (BM), HSCs, CMPs, neutrophils (PMN), B lymphocytes (B), and T lymphocytes (T). HPRT, hypoxanthine phosphoribosyltransferase.

All these potential founders were bred with C57BL/Ka mice to propagate heterozygous lines. From successfully breeding founder animals, whole bone marrow from F₂ and F₃ progeny was assessed for BCR/ABL expression. Expression of p210^BCR/ABL was undetectable by Western or Northern analyses, but transcripts of the BCR/ABL transgene were readily detected in whole bone marrow by RT-PCR in all lines assayed (data not shown). To confirm that transgene expression was limited to cells of the myeloid lineage, semiquantitative RTPCR was performed on sorted populations of cells from an F₂ M-line p210^BCR/ABL mouse. Expression of BCR/ABL was seen in CMPs and neutrophils but not in HSCs, B cells, or T cells (Fig. 1B).

Of the 13 transgenic founders, 4 developed myelomonocytic malignancies within 12 weeks of birth. Founders C, K, L, and M displayed symptoms similar to human chronic phase CML: increased WBC counts with neutrophilia (Fig. 2A) and splenomegaly. Founders C, K, and M had WBC counts of 170 × 10³, 34 × 10³, and 6.1 × 10³ cells per μl, respectively (littermate controls had a mean WBC count of 4.4 ± 1.3 × 10³ cells per μl, n = 3). Founders C, L, and M had spleen masses of 0.33, 1.04, and 0.21 g, respectively (controls had a mean spleen weight of 0.10 ± 0.03 g, n = 3). Founder M displayed a marked neutrophilia in the blood (62.0% of nucleated cells, control was 14.5 ± 4.1%, n = 5) and spleen (24%, control was 3.2 ± 1.6%, n = 5). In addition to the CML phenotype, founders C and L also developed thymomas and showed large numbers of CD3⁺ cells in the spleen and bone marrow. Because the MRP8 promoter is not normally expressed in HSCs, common lymphoid progenitors, or their lymphoid progeny, it is conceivable that these 2 of the 13 founders had aberrant expression in cells with lymphoid potential due to position effects of the transgene or activated an endogenous leukemogenic retrovirus. Founder K had an over-abundance of myeloid blasts in all hematopoietic tissues, as well as leukemic infiltration of the lymph nodes and liver, resembling blast-crisis CML (Fig. 2 A). Cells from this mouse were able to transfer the disease to lethally irradiated congenic recipients (data not shown).

Fig. 2.

hMRP8p210^BCR/ABL animals develop a CML-like MPD. (A) Morphology of cells from leukemic and control animals. (B) Differential cell counts in chronic phase (n = 6), accelerated phase (n = 3), and wild-type (n = 5) mice. B, myeloblasts; P, promyelocytes; M, myelocytes; N, neutrophils; Mo, monocytes; L, lymphocytes. (C) Incidence of CML-like MPD in hMRP8 transgenic mice decreased after each backcross onto the C57BL/Ka genetic background. (D) Nonleukemic hMRP8p210^BCR/ABL mice (n = 4) have higher WBC counts throughout life than littermate controls (n = 5). Error bars represent ±SD.

Of the original 13 founders, the E–K and M mice were bred successfully with wild-type mice. Of a total of 39 F₁ mice generated, 5 (4 from the F line and 1 from the M line) developed a CML-like disease. Two of these mice (both from the F line) progressed to accelerated phase and blast crisis. In F₁ matings, the F–J and M lines bred successfully. Of a total of 51 F₂ mice generated, 4 (2 each from the H and M lines) developed a CML-like disease, and none progressed to accelerated phase. Only the M line produced F₃ mice. One of 23 total F₃ mice developed a CML-like disease. All subsequent experiments were performed by using mice from the M line. In this study, the latency period to onset of leukemia was highly variable, ranging from 12 weeks to 10 months, and none of the mice that screened negative for hMRP8p210^BCR/ABL developed myeloid malignancies.

Fig. 2B shows a summary of differential cell counts for a cohort of mice from the founder, F₁, and F₂ generations displaying chronic phase (n = 6) and accelerated or acute phase (n = 3) leukemia. Similar to the human disease, there was neutrophilia in the blood (49.1 ± 13.3%, 4.1 ± 1.1 × 10³ cells per μl) and spleen (20.2 ± 9.0%) of chronic phase mice compared with wild-type mice (14.5 ± 4.1%, 0.76 ± 0.21 × 10³ cells per μl in blood and 3.2 ± 1.6% in spleen, n = 3). Chronic phase mice had a modest increase in WBC counts (9.3 ± 4.4 × 10³ cells per μl) compared with wild-type mice (5.3 ± 0.4 × 10³ cells per μl, n = 3). Increases in progenitor numbers were seen in both chronic and acute phase bone marrow. Chronic phase animals had higher numbers of blasts in the marrow (9.6 ± 3.4%) and spleen (4.2 ± 2.6%) compared with control (2.4 ± 1.6% in marrow and 0% in spleen). In the blood, blasts (1.07 ± 0.96%) and promyelocytes and myelocytes (0.97 ± 0.87%) were found in chronic phase but not wild-type (0%) mice. These chronic phase mice showed an absolute increase in myeloid series cells in the blood with a neutrophil predominance and a marked shift to the left, criteria which are consistent with CML or myelodysplasias. The acute phase mice showed high numbers of blasts in the blood (11.3 ± 4.8%), marrow (22.1 ± 2.7%), and spleen (23.3 ± 7.6%).

The incidence of the CML-like MPD dropped in mice possessing the hMRP8 p210^BCR/ABL transgene with each backcross of the original transgenic strain to C57BL/Ka-Thy1.1 mice (Fig. 2C). Expression of p210^BCR/ABL was still detectable by RT-PCR in F₅ and F₆ transgenic mice at levels similar to those assayed in earlier generations (data not shown), suggesting that transgene inactivation was not occurring (although we cannot rule out the possibility that expression levels had decreased in the later generations) but not to undetectable levels. Accordingly, we sought a discernable phenotype in these later generations of hMRP8p210^BCR/ABL mice that maintained expression of the transgene. Nonleukemic F₅ hMRP8p210^BCR/ABL mice had modestly higher WBC counts than their nontransgenic littermates from age 10 weeks onward (Fig. 2D). By 23 weeks of age, nonleukemic hMRP8p210^BCR/ABL mice had average blood counts of 15.4 ± 2.1 × 10⁶ cells per ml compared with wild-type counts of 10.4 ± 0.8 × 10⁶ cells per ml. In addition, hematopoietic organs from nonleukemic F₅ and F₆ transgenic animals showed a shift toward immature myeloid elements (Table 1). These results are consistent with the hypothesis that modifier genes expressed in the original transgenic background led to the higher disease incidence.

Table 1. Differential cell counts in leukemic hMRP8p210^BCR/ABL × hMRP8BCL-2 mice and controls (shown as percentage of total nucleated cells).

	n	Blast	Promyelocyte myelocyte	Metamyelocyte	Granulocyte	Promonocyte monocyte	Lymphocyte	Nucleated erythrocyte
Tissue
Genotype
Marrow
Wild type	5	2.4 ± 1.6	5.3 ± 1.8	4.0 ± 1.5	47.4 ± 7.9	7.3 ± 2.1	18.3 ± 2.4	15.4 ± 7.0
hMRP8p210BCR/ABL	4	4.2 ± 1.4	7.6 ± 1.6	4.4 ± 0.8	38.4 ± 1.4	9.2 ± 2.6	27.3 ± 2.3	9.1 ± 2.1
hMRP8BCL-2	5	3.0 ± 1.5	6.0 ± 2.0	8.0 ± 3.9	32.2 ± 7.0	15.7 ± 3.5	18.8 ± 4.5	16.3 ± 5.3
hMRP8p210BCR/ABL × hMRP8BCL-2	3	6.6 ± 3.3	7.4 ± 2.4	4.8 ± 1.2	36.2 ± 4.9	15.6 ± 2.0	21.2 ± 4.9	8.1 ± 2.3
Leukemic hMRP8p210BCR/ABL × hMRP8BCL-2	5	30.0 ± 8.8	27.5 ± 3.7	12.0 ± 2.8	7.5 ± 3.3	4.1 ± 1.0	12.6 ± 4.7	6.4 ± 1.6
Spleen
Wild type	5	<1	<1	<1	3.2 ± 1.6	4.3 ± 2.6	87.5 ± 1.9	4.0 ± 2.4
hMRP8p210BCR/ABL	4	1.5 ± 1.2	2.1 ± 0.3	1.7 ± 0.6	2.6 ± 0.7	4.8 ± 1.3	81.5 ± 4.6	5.8 ± 1.8
hMRP8BCL-2	5	1.6 ± 0.7	1.6 ± 1.0	3.0 ± 2.7	4.9 ± 2.3	10.0 ± 5.1	70.6 ± 8.2	9.1 ± 4.6
hMRP8p210BCR/ABL × hMRP8BCL-2	3	1.2 ± 0.9	3.1 ± 0.3	2.4 ± 1.6	4.7 ± 2.3	6.8 ± 1.8	71.8 ± 7.6	9.9 ± 5.3
Leukemic hMRP8p210BCR/ABL × hMRP8BCL-2	5	14.9 ± 8.5	10.2 ± 2.4	7.2 ± 2.9	2.6 ± 7.2	7.2 ± 4.9	48.4 ± 9.1	9.5 ± 4.0
Blood
Wild type	5	<1	<1	<1	14.5 ± 4.1	5.4 ± 2.3	79.8 ± 5.5	<1
hMRP8p210BCR/ABL	4	<1	1.1 ± 1.3	2.0 ± 1.5	17.0 ± 6.3	6.9 ± 2.3	71.9 ± 7.8	1.2 ± 0.9
hMRP8BCL-2	5	<1	<1	1.2 ± 0.5	7.4 ± 3.3	17.1 ± 3.0	64.0 ± 9.3	1.3 ± 1.3
hMRP8p210BCR/ABL × hMRP8BCL-2	3	<1	2.1 ± 2.1	2.5 ± 2.3	8.1 ± 6.6	17.2 ± 1.7	68.0 ± 8.1	1.4 ± 1.0
Leukemic hMRP8p210BCR/ABL × hMRP8BCL-2	5	10.0 ± 4.4	8.5 ± 3.8	6.2 ± 2.7	5.1 ± 1.4	17.7 ± 13.6	50.2 ± 16.7	2.2 ± 1.1

Differential cell counts were performed by counting at least 300 cells per marrow and spleen cytospin and at least 100 cells per blood smear. Cells were stained with May—Grünwald/Giemsa. Numbers are presented as means ± SD.

To determine whether other protooncogenes could act synergistically with p210^BCR/ABL in myeloid leukemogenesis, we crossed F₅ and F₆ hMRP8p210^BCR/ABL mice to transgenic animals expressing human C-MYC, N-RAS, and BCL-2 proteins under control of the hMRP8 promoter. Neither hMRP8p210^BCR/ABL × hMRP8C-MYC (n = 12) nor hMRP8p210^BCR/ABL × hMRP8N-RAS (n = 12) double-transgenic animals developed myeloid malignancies with frequencies higher than single-transgenic hMRP8p210^BCR/ABL mice over an observation period of 1 year (data not shown), but coexpression of hMRP8BCL-2 proved synergistic in the development of myeloid leukemia.

Six different breeding pairs of hMRP8p210^BCR/ABL heterozygotes crossed to hMRP8BCL2 heterozygotes generated 72 offspring in this time period. Eighteen of 72 mice expressed both BCR/ABL and BCL-2, and 9 of these 18 double-transgenic animals developed AML over the course of 1 year. Animals developing AML did so without observation of a chronic phase and became moribund within 1 week after showing visible signs of illness. No hMRP8BCL-2 or wild-type mice developed overt leukemia over this time period. Of the nine double-transgenic mice that developed AML, eight became moribund by 20 weeks of age (Fig. 3C). All leukemic mice had splenomegaly and visibly pale marrow. Morphological examination of hematopoietic tissues showed a large proliferation of undifferentiated myeloblasts (Fig. 3A).

Fig. 3.

Double-transgenic hMRP8p210^BCR/ABL × hMRP8BCL-2 mice develop AML. (A) Morphology of cells from leukemic and control mice. NCAE, α-naphthyl chloroacetate esterase. (B) Morphology of leukemic blasts in bone marrow (Left) and spleen (Right) of lethally irradiated recipients transplanted with leukemic cells from double-transgenic mice. (C) Incidence of AML-like disease in hMRP8p210^BCR/ABL × hMRP8BCL-2 mice (n = 18) and lethally irradiated congenic recipients transplanted with leukemic cells (n = 10). Db Tg, double transgenic. (D) FACS profile of bone marrow cells from leukemic animals. (E and F) Staining for myeloid progenitors in healthy (E) versus leukemic (F) hMRP8p210^BCR/ABL × hMRP8BCL-2 mice. Note the substantial increase in the GMP subpopulation in leukemic animals. FSC, forward scatter.

Eight of nine leukemic mice displayed morphological characteristics similar to the human disease acute myeloblastic leukemia with maturation (AML-M2), characterized by >30% myeloblasts, >10% maturing granulocytic elements, and <20% monocytic cells in the nonerythroid compartment of the bone marrow. Hematologic parameters in double-transgenic mice are consistent with this diagnosis (Table 1), as is lightly positive myeloblast staining for α-naphthyl chloroacetate esterase, a marker of granulocyte differentiation (Fig. 3A). In addition, the majority of cells in leukemic bone marrow are CD47⁺, FcγRII/III⁺, Mac-1^lo, Gr-1^lo, and negative for the lymphoid markers B220 and CD3 (Fig. 3D). Similar plots are obtained from FACS analyses of leukemic spleens (data not shown). Overexpression of CD47 was originally found by differential microarray analyses comparing cells from leukemic and nonleukemic Fas^lpr/lpr × hMRP8BCL-2 mice (ref. 24 and D.T., J. S. Hu, J. T. Ma, D. Lockhard, I.L.W., and E.L., unpublished results). The remaining leukemic animal displayed monocytosis with apparent blast crisis, resembling human acute myelomonocytic leukemia (AML-M4). Patients with the t(9;22) translocation often show abnormalities in erythrocytes and/or platelets and occasionally progress to acute erythroleukemias or megakaryocytic leukemias, although we did not observe abnormalities in either of these lineages. Although we did not assay BCR/ABL expression in megakaryocyte/erythrocyte-restricted progenitors (MEPs), previous work from our laboratory has shown that expression from hMRP8-driven transgenes is absent in MEPs and their progeny (27).

To confirm that frank transformation was an intrinsic property of cells from leukemic animals, hematopoietic cells from three leukemic mice were injected into 10 lethally irradiated congenic recipient animals. Eight of the transplanted mice died within 30 weeks of transplant (Fig. 3C). Examination of marrow and spleen tissues from these recipients confirmed that there was a preponderance of myeloblasts (Fig. 3B). FACS analysis confirmed that the donor-derived cells in these irradiated animals showed a high forward scatter, which is characteristic of large, blastic cells (data not shown) and stained positively with markers identical to the primary leukemias.

We examined bone marrow by flow cytometry from leukemic and healthy double-transgenic animals to analyze the profiles of HSCs, CMPs, GMPs, and MEPs (20). Among all populations analyzed, we found that leukemic animals contained greatly increased numbers of GMPs (4.0% of bone marrow nucleated cells) (Fig. 3E) compared with healthy double-transgenic (0.89%) (Fig. 3F) and wild-type (0.4%) mice. Whereas frequency of Lin^–c-Kit⁺Sca-1⁺ cells highly enriched for HSC were approximately equal (0.1% of marrow in leukemic and healthy double-transgenic mice), CMP percentages approximately doubled in leukemic animals (0.4–0.6% of marrow) compared with healthy double-transgenic (0.3%) and wild-type (0.2%) controls. The proportion of MEPs in leukemic animals was approximately equivalent to those in wild-type (0.2% of marrow) but less than healthy double-transgenic (0.4–0.6%) animals.

Discussion

Previous mouse models using p210^BCR/ABL have generally targeted expression to HSCs or whole bone marrow either through retroviral transduction (12–16) or creation of transgenic animals (17–19). Although CML associated with t(9;22) is presumed to be a stem cell disease, the lack of consistent BCR/ABL transcripts in HSCs in patient samples (10) could be interpreted to mean that the presence of translocation-positive HSCs is not direct evidence that HSCs are part of the leukemic clone rather than myeloid progenitors. Here we present a model in which expression of the hMRP8p210^BCR/ABL transgene is targeted to be absent in HSCs and instead targeted to the earliest myeloid committed progenitors, the CMP, GMP, and their myelomonocytic progeny. Most importantly, we show that enforced expression of Bcr/Abl synergizes with expression of Bcl-2 in myeloid progenitors to generate AML.

hMRP8p210^BCR/ABL animals develop a myeloid disease that shares many characteristics with human CML: an absolute increase in myelomonocytic blood cells, an expansion of peripheral granulocytes with left shift of blood cell types, splenomegaly, and, in some cases, a progression to accelerated phase and blast crisis. However, the leukemia reported here differs from other mouse models in that disease penetrance is lower than many models and the time until disease onset is variable and may take up to 12 months. In addition, the disease incidence decreased with each backcross of original founder lines onto C57BL/Ka. Low disease incidence and long latency in our model are likely to result from lower expression of the p210^BCR/ABL transgene when compared with models using the robust but nonspecific retroviral and metallothionein promoters. Although an increase in the dosage of p210^BCR/ABL has been shown to be crucial for both antiapoptotic effects (28, 29) and resistance of the disease to treatment (30), perhaps the expression levels found in our transgenic mice more closely mimic those found in preleukemic and chronic phase patients (10, 31, 32). The strain-specific decrease in CML-like disease also suggests that heritable factors, in particular genetic backgrounds, can either enhance or suppress the physiologic effects of BCR/ABL.

Early experiments modeling BCR/ABL in mouse systems suggested that the t(9;22) chromosomal translocation gene product may be sufficient for leukemic transformation and the development of frank leukemia (33). Our results and those of other investigators (34–36) instead suggest that additional mutations or epigenetic modifications are necessary in BCR/ABL-positive cells for full transformation. Two studies have reported that 30% (37) and 75% (38) of normal, healthy individuals possessed the t(9;22) translocation in a minority of peripheral blood leukocytes, strongly supporting the notion that BCR/ABL alone is insufficient for disease development. In this study we have demonstrated that the enforced expression of BCR/ABL in myelomonocytic cells but not HSCs synergizes with enforced expression of BCL2 to induce blast crisis in ≈50% of the mice. Here BCR/ABL collaborates with BCL2 but not C-MYC or N-RAS; the latter finding suggesting that RAS or MYC are either not involved in leukemogenic progression or, more likely, that their roles are redundant with BCR/ABL function. The potent synergy between enforced, high-level expression of BCL-2 and low-level BCR/ABL expression suggests that blockade of programmed cell death may be a key event in the progression of CML to blast crisis. This hypothesis correlates well with clinical evidence showing that the majority of patients with myeloid blast crisis have leukemic cells with higher levels of BCL-2 expression compared with cells from chronic phase CML patients (39, 40). Thus, BCL-2 may accelerate the onset of acute leukemia by promoting the survival of BCR/ABL-expressing cells, enabling them to accumulate additional genetic lesions.

We propose here that the results in human CML and this mouse model of CML can be interpreted in a consistent model of disease pathogenesis: Blast-crisis CML results from a progressive acquisition of heritable alterations within either HSCs directly or a downstream cell population that gains self-renewal potential (41–44). In mice (and likely humans) the only cells in the myeloid lineage that self-renew perpetually are long-term HSCs (20, 23, 45). Thus HSCs are the only logical cells in the lineage that can accumulate the rare mutations and translocations that are necessary events in myeloid leukemogenesis. These events include elimination of internal control of programmed cell death, resistance to externally induced death and phagocytosis pathways mediated by cells of the innate (e.g., macrophages and natural killer cells) and acquired (T and B cells) immune systems, acquisition of telomere-maintenance pathways, and increased proliferative capacity that results in extensive self-renewal (reviewed in ref. 44). In our mouse model of CML blast crisis, however, we provide the elements of leukemic progression by transgenesis such that progression to the ultimate transforming events can occur in myeloid progenitors rather than HSCs.

Data supporting the chromosomal rearrangement of the BCR and ABL genes in HSCs include the detection of t(9;22) in myelomonocytic, megakaryocytic, erythroid, and lymphoid cells as well as in sorted populations enriched for HSCs and progenitor cells (7–9). Although it is generally agreed that the founding t(9;22) chromosomal translocation arises in an HSC, it is unclear whether BCR/ABL is consistently expressed in HSCs or other very primitive hematopoietic populations (10) or whether it exerts its effects on more mature myeloid progeny. We have shown in AML patients with AML1/ETO translocations that translocation-positive HSCs (CD34⁺38^–Thy1⁺Lin^–) are not malignant, whereas among their progeny are CD34⁺38^–Thy1^–Lin^– leukemic stem cells (25). Thus, the issue is difficult to discern because clones of HSCs may exist for years (46–48) and thus may carry genetic lesions without effect. However, the fact that CML is a clonal disorder (49) and the BCR/ABL translocation is consistently present in HSCs has led to the proposition that CML is a stem cell disease (43), and failure to clear the patient of translocations is a failure of therapy (50–52).

Because our mice develop both chronic leukemias and AMLs without detectable HSC expression of BCR/ABL, we believe that a cell downstream of the HSC can be the ultimate target of direct leukemic transformation. It shall be important to test all defined hematopoietic progenitors with myeloid differentiation potential from leukemic animals for their ability to confer leukemia to transplanted hosts. In this assay, putative leukemogenic progenitor cells would be required to self-renew and proliferate indefinitely to generate leukemia in a wild-type recipient, thus having acquired the fundamental properties of stem cells. Intriguingly, double-transgenic hMRP8p210^BCR/ABL × hMRP8BCL-2 mice show a large increase in the phenotypic GMP population, suggesting that this hematopoietic subset may contain leukemia-repopulating cells. Accurately determining the cellular targets of transformation should allow us to purge patient samples of leukemogenic subsets more effectively and potentially allow the development of therapies targeting the root of disease.