Directing oncogenic fusion genes into stem cells via an SCL enhancer

Abstract

TEL-TRKC is a fusion gene generated by chromosomal translocation and encodes an activated tyrosine kinase. Uniquely, it is found in both solid tumors and leukemia. However, a single exon difference (in TEL) in TEL-TRKC fusions is associated with the two sets of cancer phenotypes. We expressed the two TEL-TRKC variants in vivo by using the 3′ regulatory element of SCL that is selectively active in a subset of mesodermal cell lineages, including endothelial and hematopoietic stem cells and progenitors. The leukemia form of TEL-TRKC (–exon 5 of TEL) enhanced hematopoietic stem cell renewal and initiated leukemia. In contrast, the TEL-TRKC solid tumor variant (+ TEL exon 5) elicited an embryonic lethal phenotype with impairment of both angiogenesis and hematopoiesis indicative of an effect at the level of the hemangioblasts. The ability of TEL-TRKC to repress expression of Flk1, a critical regulator of early endothelial and hematopoietic cells, depended on TEL exon 5. These data indicate that related oncogenic fusion proteins similarly expressed in a hierarchy of early stem cells can have selective, cell type-specific developmental impacts.

Chimeric fusion genes generated in multiple combinations are a consistent and specific feature of leukemias and some solid tumors, including sarcomas, and may arise in stem cells (1–3).

The vulnerability of stem cells and the selective association of individual fusion genes with particular subtypes of leukemia or cancer are incompletely understood but may be best explained by selection. A plausible explanation for the former is that the self-renewal competence of stem cells, coupled with their longevity, provides an opportunity for the acquisition of both initiating and necessary secondary mutations; i.e., the mutations arise at any developmental level but have selective potency in stem cells. The selectivity of fusion genes for particular cancers could reflect a restricted origin, as an illegitimate recombinant, in different stem cells but, perhaps more plausibly, by selective transforming impacts on particular stem cells or progeny. Evidence supporting the latter interpretation derives from studies on transgenic fusion genes expressed with common promoters that recapitulate in vivo the leukemic lineage phenotype (4) and by nonobese diabetic/severe combined immunodeficient mice transplantation experiments in which different subtypes of acute myeloid leukemia (5) were found to be driven by a common population of early stem cells.

We have sought direct evidence that oncogenic fusion proteins expressed in a hierarchy of early stem cells can have selective, cell-type-specific developmental impacts that depend on their intrinsic molecular properties. To achieve this objective, we adopted a strategy of directing the expression in vivo of fusion oncogenes with a regulatory element that is itself selectively active in stem cells. We also elected to conduct these studies with TEL-TRKC (ETV6-NTRK3) fusions, which, uniquely, are associated with both solid tumors and leukemias. TEL-TRKC encodes activated TRKC kinase (6, 7); one variant of TEL-TRKC, “CFS,” is a consistent genetic abnormality in congenital fibrosarcoma (CFS) (8) and congenital mesoblastic nephroma (9, 10), tumors of possible embryonic mesodermal, or mesenchymal stem cell origin. Another variant, “L,” is a rare fusion gene in acute myeloid leukemia (11). The L differs from the CFS fusion gene solely by the lack of TEL exon 5.

We made transgenic constructs in which the two cancer variants of TEL-TRKC were under the control of the 3′ regulatory elements of the SCL gene as defined in ref. 12 (Fig. 1). These constructs as transgenes would then be expected to be expressed selectively during development in hemangioblastic bipotential precursors of endothelial cells and blood cells (13–15), in hematopoietic stem cells (12, 16), and in some other/mesodermal progenitors (13).

Fig. 1.

Structure of the TEL-TRKC transgene. (A) Both L- and CFS-type of TEL-TRKC fusion gene were inserted into murine SCL exon 4 within +6E5 SCL promoter fragment, replacing first ATG codon of SCL with ATG of TEL-TRKC. The 3′ terminal of +6E5 fragment was directly connected to the 3′ enhancer fragment of SCL (3′Enh). (B) The structure of L- and CFS-type TEL-TRKC chimeric protein is indicated. Hatched area of TEL indicates pointed domain, and shaded area of TRKC indicates protein tyrosin kinase domain. (C) Detection of TEL-TRKC fusion protein expression in 293T cells by Western blot analysis.

Materials and Methods

Vectors. Mouse SCL promoter (+6E5) and 3′ enhancer fragment (3′Enh), hemangioblast/hematopoietic-specific regulatory elements of the mouse SCL gene, are described in ref. 12. The two variants of TEL-TRKC fusion were cloned from acute myeloid leukemia (11) and congenital fibrosarcoma (8) and were inserted within the +6E5 SCL fragment as in Fig. 1A. TEL-TRKC coding sequence was inserted into exon 4 within +6E5 fragment, replacing the first ATG codon of SCL exon 4 with the first ATG of TEL-TRKC. The 3′ terminal of +6E5 fragment (within exon 5 of SCL) was directly linked to the 3′ enhancer fragment (3′ UTR of SCL exon 6) (Fig. 1). Further details of construction of these transgenic vectors are available upon request. The mouse Flk1 promoter (17) and enhancer (18), which, in combination, confer endothelial cell-specific expression, were amplified by PCR and cloned into pGL3-Basic vector. For transient transfection of TEL-TRKC fusion in the luciferase assay, coding sequences of both types of TEL-TRKC fusion gene were subcloned into a pcDNA3 expression vector. All constructs were checked by direct sequencing to exclude mutations introduced by PCR.

Generation of TEL-TRKC Transgenic Mice. Transgenic mice were generated by microinjection of fertilized mouse oocytes isolated from superovulated B6/CBAF1 mice. Genomic DNA was prepared from the tails of liveborn mice or placenta of embryos, and genotyping was performed by PCR analysis with the primers specific for TEL-TRKC transgenic constructs. Expression of TEL-TRKC transgene was confirmed by RT-PCR by using peripheral blood or bone marrow cells.

N-Ethyl-N-Nitrosourea (ENU) Mutagenesis. Four- to 16-week-old L-type TEL-TRKC transgenic mice were injected intraperitoneally with a single sublethal dose of 100 mg/kg ENU (Sigma) as described in ref. 19.

Whole-Mount Immunostaining. The embryos were fixed overnight in 4% paraformaldehyde, and immunohistochemical analysis on sections or on whole embryos was performed as described in ref. 20 by using anti-PECAM1 (CD31) antibody (Pharmingen).

Flow Cytometry and Cell Sorting. Single-cell suspensions were prepared from fetal liver, bone marrow, or cultured cells and then analyzed by flow cytometry by using antibodies against c-kit, CD34, Flk1, Mac-1, Gr-1, B220, TER119, and anti-CD3 (Pharmingen). The stained cells were analyzed by FACScan or FACSVantage (Becton Dickinson).

In Vitro Methylcellulose Hematopoietic Colony Assays and Colony Replating Assay. Fetal liver from 12.5 days post coitum (dpc) embryos was dispersed into single-cell suspensions, and the number of viable cells was counted by trypan blue staining and plated in triplicate in Methocult M3231 (StemCell Technologies, Vancouver), supplemented with 1.8 ng/ml IL-3/10 ng/ml stem cell factor/10 ng/ml granulocyte–macrophage colony-stimulating factor/2 units/ml erythropoietin. The numbers of colonies comprising >40 cells were scored after 7 days, and myeloid, erythroid, and mixed colonies were defined based on their morphology. Fetal liver cells were also sorted for CD34⁺ and c-kit⁺ and plated into methylcellulose cultures as described above. Colonies were plated in triplicate, and at least five embryos were analyzed for each group.

Clonogenic growth of the hematopoietic progenitor cells with L-type TEL-TRKC transgene was assayed by methylcellulose colony replating assay as described in ref. 21 with some modifications. Bone marrow cells were harvested from femurs of mice, and 1 × 10⁴ sorted CD34⁺/c-kit⁺ cells were plated in triplicate in Methocult M3231, supplemented with 10 ng/ml murine IL-3, IL-6, granulocyte–macrophage colony-stimulating factor, and 20 ng/ml stem cell factor. After 7 days of culture, colonies were counted and 1 × 10⁴ recovered cells were replated into methylcellulose with the same growth factors.

In Vitro Culture of Aorta–Gonad–Mesonephros (AGM) Cells. The method described for in vitro AGM culture in ref. 22 was used to expand AGM cells. Cells were obtained from 12.5 dpc transgenic or wild-type (WT) embryos and cultured in DMEM supplemented with 15% FCS in the presence of 100 ng/ml stem cell factor/1 ng/ml basic fibroblast growth factor/10 ng/ml murine oncostatin M on gelatin-coated plates and incubated at 37°C, 5% CO₂ in air. After 7 days of culture, cells were trypsinized and subjected to FACS analysis for Flk1. At least five embryos from each group were analyzed.

In Vitro Angiogenesis Assay of AGM. In vitro culture of AGM explant was performed as described in refs. 23 and 24 with some modifications. The stromal cell line OP9 (25) was maintained in α-MEM (GIBCO/BRL) supplemented with 20% FCS. Explants of AGM of 12.5 dpc embryos with and without transgene were cultured on OP9 stromal cells in a slide chamber in 10% FCS and 10^–5 M 2-mercaptoethanol (Sigma) containing RPMI medium 1640 (GIBCO/BRL) and supplemented with 20 ng/ml IL-6/20 ng/ml IL-7/50 ng/ml stem cell factor/2 units/ml Epo at 37°C in a humidified 5% CO₂ air. After 12 days of culture, cells were fixed and stained with anti-CD31 antibody.

Fragments of yolk sac from 12.5 dpc embryos were plated in Matrigel basement membrane matrix (Becton Dickinson) according to the manufacturer's instructions. Tube formation was analyzed after 12 days. More than three embryos were analyzed for each genotype. ECV304 stably expressing L- or CFS-type TEL-TRKC were also plated in Matrigel to assess the effect of the fusion gene for angiogenesis.

Reporter Gene Assay (Luciferase Assay). Transient transfection of human umbilical vein endothelial cells was performed with Lipofectamine 2000 (Invitrogen). The pcDNA3 expression vector with each type of TEL-TRKC (0.3 μg) was cotransfected with a pGL3 luciferase reporter construct (0.1 μg), which contained the luciferase gene driven by murine Flk1 promoter/enhancer. Cytomegalovirus immediate early Renilla plasmids were included as internal controls for transfection efficiency. At 36 h after transfection, firefly and Renilla luciferase activities were analyzed by using dual luciferase assay kit (Promega) according to the manufacturer's instructions. Firefly luciferase activities were normalized based on Renilla firefly levels. Each cotransfection experiment was performed at least three times.

Western Blot Analysis. Western blot analysis was performed as described in ref. 11. Antibodies used were anti-TRK (Santa Cruz Biotechnology) and anti-actin antibodies (Santa Cruz Biotechnology).

RT-PCR. Total RNA was prepared from cultured cells by using the RNeasy kit (Qiagen, Crawley, U.K.) with DNase treatment and reverse transcribed with random hexamer by using Moloney murine leukemia virus reverse transcriptase (Stratagene). Reverse transcription products were amplified by PCR with specific primers by using standard procedures. The details of primer sequences and PCR condition are available on request. The products were electrophoresed on 2% agarose gels and stained by ethidium bromide.

Statistical Analysis. A two-tailed Student t test was used to determine the difference between control (WT or mock) and CFS or L.

Results

Structure and Activity of the TEL-TRKC Transgene. The L and CFS full-length TEL-TRKC fusion genes were inserted into mouse SCL exon 4 within +6E5, replacing the first ATG codon of SCL with the ATG of TEL-TRKC. The 5′ +6E5 fragment was connected to the 3′ enhancer (3′ Enh) of SCL (Fig. 1 A). The resultant two variant TEL-TRKC constructs were as shown schematically in Fig. 1B. Expression of the two alternative constructs in 293T cells resulted in similar levels of TEL-TRKC protein of the anticipated sizes (Fig. 1C).

Distinct Phenotypes of L- and CFS-Type TEL-TRKC Transgenic Mice. CFS-type TEL-TRKC was embryonically lethal when present, and no offspring were born that were positive for the CFS transgene (Table 1). Viable offspring were all negative for the transgene. Detailed analysis showed that no CFS transgenic embryos survived beyond stage of 14.5 dpc. The embryos that did survive to this stage were transgene-negative (Table 1). Examination of earlier transgenic embryos at 12.5 dpc revealed extensive vascular leakage on the head and back along the intersomite vessels (Fig. 2B). This phenotype was observed in 12 CFS embryos (Table 1). Earlier embryos (11.5 dpc) with the CFS transgene appeared normal (Table 1). In contrast, 11.5–12.5 dpc L-type transgenic mice appeared identical to WT embryos in size and developmental stage and showed no bleeding phenotype (Table 1).

Table 1. CFS transgenic embryo die of hemorrhage at 12.5 dpc.

Stage	No. of alive embryos (litters)	Transgenic offspring	Hemorrhage
	L-type transgenic
12.5 dpc	137 (30)	23	0
Postnatal	27 (5)	4	—
	CFS-type transgenic
11.5 dpc	20 (5)	8	0
12.5 dpc	99 (26)	22*	12
14.5 dpc	30 (7)	0	0
Postnatal	29 (7)	0	—

^*Abnormal structure of yolk sac vessels was observed in three and retardation was observed in two embryos that did not show apparent bleeding.

Fig. 2.

Disorganization of the vascular system in CFS-type TEL-TRKC transgenic embryos. Phenotypes of 12.5-dpc WT (A, C, E, G, and I) and CFS (B, D, F, H, and J) embryos are shown. The CFS mutant embryo is distinguished by the presence of hemorrhage in the intervertebral space and in peripheral areas (B). Vitelline vessels are thin and irregular in the CFS yolk sac (D). Whole-mount anti-PECAM1 (CD31) antibody staining shows irregular shape, less branching of the vitelline vessels (arrowhead), and lack of fine honeycomb-like vascular structure in the CFS yolk sac (F). Whole-mount anti-CD31 antibody staining shows abnormal, irregular vessel distribution in the tail (H) and body (J) of CFS embryos.

Disorganization of Vascular System in CFS-Type TEL-TRKC Transgenic Embryo and Its Yolk Sac. Examination of the yolk sacs and bodies of 12.5-dpc embryos transgenic for CFS TEL-TRKC indicated considerable architectural disruption in peripheral vessels. The appearance of yolk sacs of 12.5-dpc CFS embryos was quite distinct from WT and L-type. These yolk sacs were pale and “avascular” in appearance, and vitelline vessels could not be observed clearly (Fig. 2 C and D). When stained with CD31, the vasculature of the yolk sac of a 12.5-dpc CFS embryo showed few large vitelline vessels, and the regular branching from large to thin vessels forming fine honeycomb-like structures observed in WT yolk sacs was not present (Fig. 2 E and F).

Whole-mount staining with an anti-endothelial CD31 antibody showed disturbances of the vascular network structure in the body of CFS transgenic embryos as well as in the yolk sac (Fig. 2 H and J). The vascular network of the tail of the 12.5-dpc WT embryo showed a fine regular network of vessels consisting of many thin vessels connecting to each other (Fig. 2G). In CFS embryos, at the same developmental stage, the vascular network was very irregular, loose, and randomly formed (Fig. 2H). The vascular network on the back of the embryo body also showed significant differences between CFS and WT. In WT embryos, the diameter of surface vessels were similar and regular in shape forming a uniform vasculature (Fig. 2I), whereas vessels in CFS transgenics were irregular with some narrow in diameter and some dilated (Fig. 2J).

In contrast to CFS embryos, all L-type transgenic embryos examined showed no abnormality in vascular networks.

Defect of Endothelial Progenitor Function in AGM and Yolk Sacs of CFS-Type TEL-TRKC Transgenic Embryos. Disruption of the vascular architecture in CFS embryos indicated negative effects of the CFS-type transgene on vascular network formation (or angiogenesis) and, possibly, on differentiation/proliferation of endothelial progenitors (angioblasts). To clarify these possibilities, in vitro angiogenesis assays were performed by using the yolk sac and AGM region as sources of angioblasts. Yolk sac cells from WT embryos displayed vessel-like structures after 12 days of culture on Matrigel basement membrane matrix, whereas the yolk sacs from CFS embryos failed to form similar structures (Fig. 3 A and B). Vascular networks composed of CD31-positive cells developed successfully when AGM from WT embryos was cultured on OP9 cells, whereas only vascular beds without proper sprouting were generated from the AGM from CFS embryos (Fig. 3 C and D). A similar result was reproduced with the endothelial cell line ECV304, known to form vessel-like structures in Matrigel, which failed to form such structures when CFS TEL-TRKC was expressed (data not shown). These results suggest that there is a major defect in endothelial progenitor function to form vascular networks in CFS-type TEL-TRKC embryos.

Fig. 3.

Defective angiogenesis in the yolk sac and AGM of CFS-type TEL-TRKC transgenic embryos. Fragments of yolk sacs from 12.5-dpc WT (A) and CFS embryos (B) were plated in Matrigel basement membrane matrix. Yolk sacs from CFS embryos failed to form complex vessel-like structures as observed in yolk sacs from WT littermates after 12 days of culture. AGM from 12.5-dpc WT (C) and CFS embryos (D) were plated on OP9 cells to promote vasculoangiogenesis and stained with anti-CD31 antibody after 12 days of culture. AGM from CFS embryos failed to form a vascular network. vb, vascular bed; vn, vascular network. (Original magnifications: ×50–100.)

CFS-Type TEL-TRKC Transgene Represses Flk1 Expression. One possible interpretation of the transgenic CFS TEL-TRKC phenotype is that it reflects failure of endothelial progeny of hemangioblastic stem cells to differentiate appropriately in response to developmental cues. To assess this possibility, we isolated AGM cells from 12.5 dpc transgenic or control embryos and analyzed the expression of Flk1, a major extracellular regulator of endothelial cell development (26). Because it is difficult to assess the number of Flk1-positive cells directly from each embryo, AGM cells were put in culture for 7 days as described in ref. 22 in the presence of oncostatin M, a growth factor for hemangioblasts. AGM cells from CFS transgenic embryos showed significantly lower Flk1 positivity compared with AGM cells from WT embryos or L transgenic embryos (Fig. 4A).

Fig. 4.

CFS TEL-TRKC targets Flk1-positive cells. (A) Flk1-positive cells after 7 days of culture of AGM cells were significantly reduced in CFS-type transgenic mice compared with WT littermates. (B Upper) Selective down-regulation of Flk1 promotor/enhancer activity by CFS TEL-TRKC chimeric protein (luciferase assay). The normalized reporter gene activity of the control transfections (Mock) was arbitrarily set at 1. A schematic of luciferase vector used in the experiment is shown in B Lower.

We next tested this potential level of developmental impairment more directly by evaluating the ability of CFS and L forms of TEL-TRKC to repress expression of a luciferase gene under the control of the Flk1 promoter and enhancer in endothelial human umbilical vein endothelial cells. When cotransfected with L-type TEL-TRKC fusion gene, the luciferase activity in human umbilical vein endothelial cells was similar to mock-transfected cells. On the other hand, when CFS-type TEL-TRKC was introduced, there was a >3-fold suppression in luciferase activity (Fig. 4B). These data again indicate that CFS TEL-TRKC may repress Flk1 expression.

Defects of Hematopoietic Progenitors in the Fetal Liver of CFS-Type TEL-TRKC Transgenic Embryos. Flk1 expression is a marker of hemangioblasts and vascular progenitors (27) but is also present in a population of early embryonic hematopoietic precursors (28, 29). A null mutation of Flk1 can affect both vasculogenesis and hematopoiesis (26, 30). We therefore evaluated the hematopoietic status of CFS embryos, anticipating a defect here also. Fetal liver cells from 12.5-dpc CFS embryos were assayed for hematopoietic colonies. As shown in Fig. 4, 12.5-dpc CFS embryos had a 6-fold deficit of progenitor cells measured by myeloid colony formation and a 3-fold deficit of CD34⁺/c-kit⁺ cells in the fetal liver (Fig. 5 A and B). On the other hand, fetal liver from CFS embryos had the capacity to form similar numbers of colonies compared with fetal liver cells from the WT or L-type transgenic embryos when CD34⁺/c-kit⁺ cells were sorted and same number of cells were plated on methylcellulose (Fig. 5C). Morphological analysis of colony forming cells showed no differentiation arrest of the hematopoietic cells from CFS fetal liver (data not shown). These data suggest that CFS TEL-TRKC impedes hematopoiesis at the level of stem/progenitor cell commitment and/or expansion rather than the differentiation competence of progeny cells.

Fig. 5.

Reduced hematopoietic progenitors in CFS embryos. Fetal liver cells from WT, L, and CFS embryos were analyzed to assess the status of hematopoietic progenitors. Hematopoietic colonies (A) or percentage of c-kit⁺/CD34⁺ cells (B) were significantly lower in fetal liver of CFS embryos compared with that of WT or L-type transgenic embryos. (C) When c-kit⁺/CD34⁺ cells were sorted, the number of colonies formed were similar between WT and L or CFS transgenic embryos, indicating that defects of hematopoiesis reside at the level of expansion of stem/progenitor cells rather than in downstream differentiation pathways. Results are presented as means ± SD. NS, not significant.

Increased Proliferation and Self-Renewal of Hematopoietic Stem Cells with L-Type TEL-TRKC Transgene. In view of the cancer subtype associations of the TEL-TRKC variants, we anticipated that the L form might have an impact on self-renewal capacity of hematopoietic stem cells. CD34⁺/c-kit⁺ sorted cells from offspring transgenic for L TEL-TRKC were harvested from bone marrow and assessed for self-renewal potential as described in refs. 21, 31, and 32 (Fig. 6A). In contrast to progenitors from nontransgenic littermates, whose replicative potential was exhausted by two rounds of serial plating in methylcellulose, L transgenic bone marrow stem cells maintained high level clonogenicity indicative of a transformed or preleukemic status (Fig. 6B).

Fig. 6.

Enhanced self-renewal and leukemia in L-type transgenic mice. (A) Scheme of the clonogenic replating assay. (B) c-kit⁺/CD34⁺ bone marrow cells from three of four L-type transgenic mice showed enhanced self-renewal capacity showing proliferating cells after the third round of methylcellulose colony replating. Cells from age- and sex-matched WT mice showed very few or no remaining colonies after the third-round plate. Numbers below the x axis represent individual mice. (C) Cell morphology (May-Grünwald/Giemsa staining) and flow cytometric profiles of leukemic cells [acute myeloid leukemia (AML) and acute lymphocytic leukemia (ALL)] in L-type transgenic mice.

L-Type TEL-TRKC Transgenic Mice Develop Leukemia with Protracted Latency, but ENU Can Accelerate the Leukemogenic Process. Four of 27 liveborn mice had L-type TEL-TRKC transgene, and these transgene-positive founders provided germ-line transmission of the transgene to offspring. Two lines expressing L TEL-TRKC transgene (by RT-PCR) were subjected to further analysis.

L transgenic mice were born healthy and survived normally without any symptoms of malignancy up to 17 months. However, a sizeable fraction (16%) of transgenic mice developed acute leukemia, mostly myeloid, after a latency of ≈20 months (Fig. 6C and Table 2). The long latency and modest level of penetrance of leukemia in L-type transgenic mice suggests that TEL-TRKC, although initiating enhanced self-renewal of stem cells, is insufficient to cause overt leukemia, and additional genetic change(s) are required. To endorse this interpretation, we injected 4- to 16-week-old transgenic mice with a single sublethal dose of ENU, a potent DNA alkylating mutagen. This treatment both increased the numbers of L TEL-TRKC mice with leukemia and reduced latency (Table 2), as in other described models of preleukemia (19, 33).

Table 2. High frequency of leukemia in L transgenic mice.

	No. of mice	Leukemic death (%)	Latency, m	Phenotype (n)
Control	25	0 (0)	—	—
L type	25	4 (16)	19.8 ± 4.2	AML(3), T-ALL(1)
Control (ENU)	19	0 (0)	—	—
L type (ENU)	19	5 (26.3)	12.7 ± 2.0	AML(4), T-ALL(1)

m, months.

Discussion

Our data provide evidence that directing the expression of two different but closely related oncogenic fusion gene-encoded kinases into the same hierarchical population of intramesodermal stem cells and progenitors produces distinctive phenotypes. L TEL-TRKC lacks TEL exon 5 and was discovered in the clinical context of acute myeloid leukemia (11). As shown here, a similar phenotype can be replicated when L TEL-TRKC is expressed as a stem cell-active transgene in mice. The CFS variant of TEL-TRKC shares with L TEL-TRKC the same functional consequence of gene chimerism, constitutive activation of TRKC kinase, and the mitogen-activated protein kinase pathway (refs. 6 and 7 and unpublished data) but has a strikingly different disease affiliation. CFS TEL-TRKC is associated with rare subsets of nonhematopoietic cancers in newborns that includes CFS (8) and congenital mesoblastic nephroma (9, 10). The developmental or stem cell origins of these cancers are unknown. CFS may have ambiguous or mixed cellular phenotypes (34), but the expression of vimentin and other markers (35, 36) is believed to reflect a mesenchymal phenotype. In this context, it will be of interest to assess whether CFS TEL-TRKC can elicit any increased self-renewal in hemangioblasts and/or their antecedent mesodermal precursors. More recently, CFS TEL-TRKC has been found to be associated also with secretory breast carcinoma (SBC) (37). SBC is a very rare form of breast cancer occurring in very young females and some older women. Its cellular origins are unknown, but CFS TEL-TRKC can transform an epithelial cell line and the carcinoma cells in vitro, and, after transfer into severe combined immunodeficient mice, express epithelial markers. This finding indicates that expression of this fusion oncogene is at least compatible with an epithelial phenotype but leaves unanswered tumor origin in vivo. Based upon these data, however, Tognon et al. (37) have challenged the concept that fusion genes have marked lineage specificities or are subject to lineage-specific constraints.

The clinical, histopathological data indicate that, in fact, TEL-TRKC fusions are highly restricted to subsets of cancer cells in terms of lineage, albeit perhaps not as uniquely affiliated as some chimeric fusion genes. There is no basis for supposing that TEL-TRKC fusions arise as illegitimate recombinants exclusively in a highly restricted set of stem cells. The alternative explanation we favor is that individual fusion genes can arise in a wide variety of cell types but have oncogenic impact only in particular stem cells (38) and induce transformed phenotypes dictated or restrained by particular cell type, developmental, or lineage context. This interpretation is in accord with prior observations on the phenotype of leukemias initiated in vivo by “lymphoid” or myeloid fusion genes (4, 39) and analysis of the stem cell origins of different acute myeloid leukemias (5). The data we present here provide evidence that even very closely related fusion genes with the same enforced expression pattern can have differential impacts. This selectivity is most readily explained by TEL-TRKC variants having their selective impacts at different levels within the developmental hierarchy of stem cells and precursors, i.e., hemangioblasts or hematopoietic stem cells (Fig. 7). The SCL enhancer-driven construct used constrains the expression patterns and phenotypes that are possible, and we have no “transformed” readout for CFS TEL-TRKC. Nevertheless, we speculate that the nonleukemic cancers with CFS-type TEL-TRKC fusions might originate from a subset of mesodermal stem cells “upstream” of the hemangioblasts (Fig. 7), generating divergent mesenchymal phenotypes that depend on the tissue site of transformation.

Fig. 7.

A speculative model for the possible stages of cell transformation by CFS and L TEL-TRKC fusion proteins. Because some mesodermal stem cells appear to express Flk1 also (44), the model postulates that some tumors with CFS-type TEL-TRKC may arise at this level (i.e., congenital fibrosarcoma and congenital mesoblastic nephroma) because of a block in further differentiation. On the other hand, the L form of TEL-TRKC enables cells to differentiate into hematopoietic stem cells (HSC), where constitutive kinase activity of the fusion protein transforms cells into leukemic cells. Cells expressing SCL are indicated by gray circles, and Flk1 expressing cells are in the gray box. Me, mesoderm; HA, hemangioblast; End, endothelial cell. Two parallel lines represent blocks in differentiation, and circular arrows represent enhanced self-renewal.

The selective impacts of activated TRKC kinase on either endothelial and hematopoietic differentiation versus hematopoietic stem cell renewal can be entirely ascribed to activity residing in exon 5 of the TEL fusion partner of TRKC, because these two chimeric genes have equivalent TRKC kinase activation (refs. 6 and 7 and unpublished data).

The finding that the CFS variant of TEL-TRKC but not the L form interferes with the expression of Flk1 provides a lead in how this selectivity might arise, as does the known activity of central region of TEL (encoded by exon 5) to recruit transcriptional corepressors (40–43). CFS TEL-TRKC protein mainly localizes in the cytoplasm (data not shown), and, therefore, any transcriptional deregulating activity would presumably be indirect, e.g., by sequestering components in the cytoplasm.

In summary, we have presented a transgenic model for targeting oncogenes into stem cells that may provide a useful tactic to explore other aspects of stem cell regulation and fate. The model provides the evidence that oncogenic fusion proteins expressed in a hierarchy of early stem cells can have selective, cell-type-specific developmental impacts that depend on their intrinsic molecular properties. This result in turn provides a rationale for the striking associations between different chromosome translocations, unique fusion genes, and cancer or leukemia phenotypes.