TEL-AML1, Expressed from t(12;21) in Human Acute Lymphocytic Leukemia, Induces Acute Leukemia in Mice¹

The TEL-AML1 fusion protein is expressed from the t(12;21)(p13;q22) chromosome in 25% of pediatric and 3% of adult B-ALL⁴ cases (1, 2). TEL is an Ets family transcription factor with a COOH-terminal DNA-binding domain. In TEL-AML1, the majority of TEL, with the exception of its DNA-binding domain, is positioned NH₂-terminal to AML1. TEL-AML1 binds AML1 cis DNA elements and represses linked reporter genes (3, 4). The AML1 gene itself harbors point mutations in the AML1 DNA-binding domain in 5% of AMLs, and the genes encoding AML1 or its dimerization partner, CBFβ, are involved in several other chromosomal abnormalities in an additional 20% of AML cases. Mutation of the AML1 gene and expression of the AML1-ETO, CBFβ-SMMHC, or TEL-AML1 fusion proteins apparently have in common the ability to inhibit the activity of endogenous AML1. In this regard, mice lacking either AML1 or CBFβ and knockin mice expressing AML1-ETO or CBFβ-SMMHC each fail to develop definitive hematopoiesis (5). Whereas TEL-AML1 might contribute to leukemogenesis directly, by repressing AML1 target genes or by binding other Ets proteins via its pointed domain, it is possible that disruption of one copy of the AML1 gene by the t(12;21) translocation is the more critical abnormality. Similarly, disruption of one TEL locus might facilitate transformation because the second copy of TEL and expression of TEL protein are lost in the majority of ALLs harboring TEL-AML1 and because a small number of B-ALLs without TEL-AML1 lose both TEL alleles (6, 7, 8). We therefore sought to determine whether TEL-AML1 induces acute leukemia in mice, expressed either alone or together with loss of the overlapping p16^INK4ap19^ARF genes.

The MING retroviral vector encodes NGFR-GFP, residues 1–277 of the human low-affinity NGFR (p75), containing its extracellular and transmembrane domain, linked to the NH₂ terminus of full-length enhanced GFP, residues 1–238. The TEL-AML1 cDNA was subcloned upstream of the IRES to generate MING-TEL-AML1.

NIH 3T3 and ecotropic CRE packaging cells (9) were cultured in DMEM with 10% heat-inactivated calf serum. CRE cells were transfected by calcium phosphate precipitation with 20 μg of MING or MING-TEL-AML1 and 2 μg of pSV40-puro. Pooled transfectants were selected using 2 μg/ml puromycin. Cells (10⁶) were then immunoaffinity purified with biotin-conjugated anti-NGFR monoclonal antibody (HB8737; American Type Culture Collection) and streptavidin-coated magnetic beads (Stem Cell Technologies, Vancouver, British Columbia, Canada). After expansion in culture, selected cells were reselected one to three additional times to obtain populations uniformly expressing NGFR. The viral titers generated by the two packaging lines were assessed using a viral RNA slot blot analysis. In brief, viral particles were precipitated using 10% polyethylene glycol 8000 in 0.5 m NaCl. RNA was then extracted using phenol and loaded onto a nylon filter using a slot blot apparatus. A 1.7-kb SphI/EcoRI fragment containing the 5′ LTR and adjacent retroviral sequences was then used as probe using a Northern blot procedure (10). As a second approach, NIH 3T3 cells were transduced for 48 h using filtered supernatants from the CRE cell lines and 4 μg/ml Polybrene. NGFR expression in the NIH 3T3 cells was assessed 48 h later by fluorescence-activated cell sorting. Assessment of TEL-AML1 expression by Western blotting was carried out as described previously (10). AML1 antiserum was kindly provided by H. Drabkin, and TEL antiserum was kindly provided by J. R. Downing.

Donor C57BL/6 or C57BL/6 p16^INK4a/p19^ARF(−/−) mice (11) were treated with 150 mg/kg 5-fluorouracil i.v., and 3 days later, marrow cells were flushed from the femurs and tibias using QBSF-58 medium (generously provided by Quality Biological Inc.). After red cell lysis with ammonium chloride, the cells were resuspended in QBSF-58 supplemented with 2 mm glutamine, 100 ng/ml murine interleukin 6, 100 ng/ml murine interleukin 3, and 50 ng/ml murine stem cell factor (PeproTech) and cultured for 48 h at 37°C in 5% CO₂. Confluent CRE packaging lines were irradiated with 40 Gy before coculture with 8 × 10⁶ marrow cells/100-mm dish with 4 μg/ml Polybrene for an additional 2 days. Transduced marrow cells (2 × 10⁶) were then injected into the tail veins of recipient wild-type C57BL/6 mice that had received 6.0 Gy of γ-irradiation the day before the injection and 3.5 Gy of γ-irradiation on the day of the injection. Leukemia transplantation to secondary recipients was by i.p. injection of 2 × 10⁶ cells.

Cytospins of marrow or spleen cells were subjected to Wright’s-Giemsa staining. Murine marrow or spleen cells (10⁶) were stained with biotin-conjugated antibodies at 1 μg/10⁶ cells for 30 min at 4°C, washed, incubated with streptavidin-conjugated R-phycoerythrin (BD PharMingen, San Diego, CA) for 30 min at 4°C, and fixed with 1.0% paraformaldehyde and 0.01 m HEPES (pH 7.4). Antibodies used recognized Sca-1 (Ly6A/E; stem cell), TCRβ (H57-597; T lineage), B-220 (RA3-6B2; B lineage), Mac-1 (M1/70; macrophage), Gr-1 (RB6-8C5; granulocyte), Ter-119 (erythroid), CD4 (GK1.5; T lineage), and CD8α (53-6.7; T lineage; all from BD PharMingen). Immunophenotypes were analyzed with a FACScan flow cytometer (Becton Dickinson).

The end point of this study was time to leukemia. Event time distributions were estimated by the method of Kaplan and Meier. HRs were obtained using the Cox model. When the Cox model did not converge, the log-rank P is presented. Statistical computations were performed using the SAS or EGRET PC packages (12, 13).

The control vector MING, which expresses only NGFR-GFP, and MING-TEL-AML1 are diagrammed in Fig. 1,A. These vectors were transfected into CRE packaging cells along with pSV40-puro. After puromycin selection, NGFR-GFP-expressing cells were selected using NGFR antibody and magnetic microbeads. Two rounds of selection yielded a population of CRE-MING cells homogenously expressing NGFR, whereas four rounds of selection were required to purify CRE-MING-TEL-AML1 cells. TEL-AML1 was detected at the expected molecular weight in these cells, using either an AML1 or a TEL antiserum (Fig. 1,B). GFP fluorescence was present uniformly in the CRE-MING cells but absent in the TEL-AML1 cells, a phenomenon observed previously with NGFR antibody selection of MING producer lines⁵ and likely reflecting outgrowth of a clonal population carrying a mutation in the GFP segment of the integrated vector. The viral titers generated by the resulting two producer populations were compared by analyzing retroviral RNA in the CRE supernatants (Fig. 1 C) and by transduction of NIH 3T3 cells (data not shown). Both assays indicate that the titer of the MING viral stock was approximately four times that of the TEL-AML1 vector. Based on quantification of 3T3 transduction, the titers of the MING and TEL-AML1 vectors were approximately 8 × 10⁴/ml and 2 × 10⁴/ml, respectively. CRE-MING cells transferred GFP expression to 27% of Sca-1+ marrow stem/progenitor cells in an in vitro assay, and MING vectors expressing CBFβ-SMMHC or E7 and having a 2-fold lower titer transferred GFP expression to 10% or 14% of Sca-1+ cells (14). Inactivity of the GFP segment in the CRE-TEL-AML1 producer line and lack of detectable surface NGFR in marrow cells transduced with the MING vector or any of its derivatives (14) precluded assessment of marrow transduction by CRE-TEL-AML1 cells. However, based on the titers of the TEL-AML1 and MING viral stocks, we estimate that these cells transduced approximately 7% of Sca-1+ marrow cells.

The MING or TEL-AML1 vector was used to transduce marrow cells isolated from C57BL/6 mice, and unselected marrow was then transplanted into syngeneic recipients. Leukemia incidence for these two cohorts is shown in Fig. 2. Two leukemias developed in the nine mice transduced with the TEL-AML1 vector. Because the MING group had no leukemias, its HR versus the TEL-AML1 cohort could not be calculated, but the log-rank test shows a trend toward a significant difference between these groups (P = 0.07). As we will describe, no leukemias developed in a cohort of mice transplanted with p16^INK4ap19^ARF(−/−) marrow transduced with the MING vector. The incidence of leukemia in the TEL-AML1 group is significant compared with the combined MING groups (P = 0.01).

The morphologies of the two TEL-AML1 leukemias, compared with normal spleen or bone marrow cells, are shown in Fig. 3. One leukemia, T/A-1 (T/A, TEL-AML1), was best detected in the spleen and had prominent nucleoli and visible cytoplasm. The second leukemia, T/A-2, homogenously replaced the bone marrow and demonstrated an open chromatin pattern and only a rim of dark cytoplasm. Both blast populations were significantly larger than the normal lymphocytes visible as the major population in the top left panel. Fluorescence-activated cell-sorting analysis demonstrated that T/A-1, which developed at 6.5 months, was TCRβ⁺CD4⁺CD8⁻ and negative for other antigens evaluated, a T-lineage ALL (T-ALL). T/A-2 was only B220⁺, a B-ALL. T/A-2 was transplanted into a secondary, nonirradiated recipient, and overt leukemia developed 3 months later. TEL-AML1 expression was detected by Western blotting in marrow cells isolated from the secondary recipient (Fig. 4, top panel). A small amount of expression was also detected in bone marrow cells from a TEL-AML1 mouse that died on day 490 without leukemia, but expression in the B-ALL was significantly higher. Endogenous TEL was not detected (data not shown). Fast Green staining (Fig. 4, bottom panel) indicates that the B-ALL sample was underloaded severalfold relative to the other samples.

The MING or TEL-AML1 vectors were also introduced into marrow cells isolated from C57BL/6 mice lacking the overlapping p16^INK4a and p19^ARF genes, and the transduced cells were then injected into irradiated, wild-type, syngeneic recipients. p16^INK4a is a cyclin-dependent kinase inhibitor. Loss of murine p19^ARF or human p14^ARF is expected to increase Mdm2 and thereby decrease p53 activity. Deletion of the overlapping p16^INK4ap19^ARF genes occurs in approximately 20% of B-ALL and 60% of T-ALL (15). The p16^INK4a gene is also specifically inactivated by promoter methylation in 80% of high-grade lymphomas, but not in ALL cases (16). Leukemia incidence for these two cohorts is shown in Fig. 2. Leukemia developed in 6 of 8 TEL-AML1/p16p19 mice and in 0 of 17 of the MING/p16p19 cohort (P < 0.0001). Compared with the TEL-AML1 group, the TEL-AML1/p16p19 mice developed leukemia more rapidly and at a higher incidence (log-rank P = 0.01). The HR between these groups is also significant (HR = 6.3; 95% confidence interval, 1.3–32; P = 0.03). One of the TEL-AML1/p16p19 leukemias was likely a T-ALL because the thymus was grossly enlarged, one was a Mac-1⁺Gr-1⁺ AML, and the phenotypes of the four other leukemias in this group could not be determined because the mice were found dead between observation periods with grossly enlarged but necrotic spleens (190, 300, 440, and 580 mg; normal spleen, 80 mg). Mice with T-ALL invariably have enlarged thymuses, whereas this organ was not enlarged in these four mice.

Using two different approaches, expression in wild-type or in p16^INK4ap19^ARF(−/−) marrow cells, we have provided evidence indicating that TEL-AML1 contributes to leukemic transformation. This finding is of particular significance because loss of TEL expression or mutation in one or both AML1 alleles occurs in leukemias lacking TEL-AML1. Whereas our results indicate that TEL-AML1 participates in transformation, we cannot rule out the possibility that t(12;21) also contributes to the formation of B-ALL by disrupting a TEL and an AML1 locus. It is possible that the murine leukemias that developed in this study harbor alterations in the endogenous TEL or AML1 genes. TEL-AML1 but not endogenous TEL was detected in the one murine TEL-AML1 leukemia analyzed, mimicking human leukemias with t(12;21) (6, 7, 8). In future experiments, we will analyze additional TEL-AML1 leukemias for genetic alterations that may cooperate in leukemogenesis. It will also be of interest to determine whether leukemias obtained are monoclonal, especially when TEL-AML1 is combined with loss of the p16p19 genes.

Our findings also indicate that TEL-AML1 and loss of the p16^INK4a/p19^ARF genes functionally cooperate during leukemogenesis. Deletion of the p16^INK4a/p19^ARF genes in pediatric B-ALL is common (15). Inhibition of CBF blocks myeloid and lymphoid differentiation in normal marrow and in 32D cl3 myeloblastic cells and slows apoptosis of Ba/F3 cells in response to DNA damage (10, 17, 18). On the other hand, inhibition of CBF also slows G₁ progression in myeloid or lymphoid cells (10). Exogenous cyclin-dependent kinase 4 or cyclin D2 overcomes delayed cell cycle progression of 32D cl3 and Ba/F3 cells resulting from blockade of CBF transactivation (10), and c-Myc acts similarly (19). These findings suggest that hematopoietic cells expressing CBF oncoproteins might be blocked at an early stage of differentiation but are incapable of proliferation. We previously demonstrated that loss of the p16p19 genes or coexpression of E7 cooperates with CBFβ-SMMHC to induce acute leukemia (14), and we now find that deletion of the p16p19 genes similarly cooperates with TEL-AML1. This result supports our proposal that mutations that accelerate G₁ potentiate the ability of CBF oncoproteins or AML1 mutations to block differentiation or apoptosis (5, 10, 14). Deletion of p16p19 might allow lymphoid stem cells to proliferate despite the presence of TEL-AML1 and might even enable increased levels of TEL-AML1, allowing more effective blockade of differentiation or apoptosis. In the future, transduction of marrow isolated from recently developed mice lacking either p16^INK4a or p19^ARF, but not both, will allow assessment of whether TEL-AML1 cooperated with alteration of the Rb or p53 pathways in our TEL-AML1/p16p19 cohort.

TEL-AML1 contributed to the formation of a B-ALL but also to other leukemia phenotypes in this study. Perhaps this diversity reflects the expression profile of the murine stem cell virus retroviral promoter together with a difference between murine and human progenitor/stem cells with regard to lineage-specific susceptibility to additional mutations capable of cooperating with TEL-AML. We expect that future studies in which TEL-AML1 is expressed specifically in immature B cells will yield a more restricted pattern of leukemia phenotypes. Nevertheless, our findings support the conclusion that TEL-AML1 contributes to the formation of human B-ALLs harboring t(12;21).

We thank D. Bodine for advice regarding marrow transduction, W. Wang for technical assistance, and N. E. Sharpless and R. A. DePinho for the p16^INK4a/p19^ARF(−/−) mice.