Yolk sac angiogenic defect and intra‐embryonic apoptosis in mice lacking the Ets‐related factor TEL

Li Chun Wang, Frank Kuo, Yuko Fujiwara, D. Gary Gilliland, Todd R. Golub, Stuart H. Orkin

Author Affiliations

  1. Li Chun Wang1,2,
  2. Frank Kuo3,
  3. Yuko Fujiwara1,2,
  4. D. Gary Gilliland2,4,
  5. Todd R. Golub1 and
  6. Stuart H. Orkin1,2
  1. 1 Department of Pediatrics, Harvard Medical School, Boston, MA, 02115, USA
  2. 2 Howard Hughes Medical Institute, Boston, MA, 02115, USA
  3. 3 Department of Pathology, Harvard Medical School, Boston, MA, 02115, USA
  4. 4 Department of Medicine, Harvard Medical School, Boston, MA, 02115, USA


The TEL gene, which is frequently rearranged in human leukemias of both myeloid and lymphoid origin, encodes a member of the Ets family of transcription factors. The TEL gene is widely expressed throughout embryonic development and in the adult. To determine the requirement for the TEL gene product in development we generated TEL knockout mice (TEL−/−) by gene targeting in embryonic stem cells. TEL−/− mice are embryonic lethal and die between E10.5–11.5 with defective yolk sac angiogenesis and intra‐embryonic apoptosis of mesenchymal and neural cells. Two‐thirds of TEL‐deficient yolk sacs at E9.5 lack vitelline vessels, yet possess capillaries, indicative of normal vasculogenesis. Vitelline vessels regress by E10.5 in the remaining TEL−/− yolk sacs. Hematopoiesis at the yolk sac stage, however, appears unaffected in TEL−/− embryos. Our findings demonstrate that TEL is required for maintenance of the developing vascular network in the yolk sac and for survival of selected cell types within the embryo proper.


The TEL (ETV6) gene, which encodes a member of the Ets family of transcription factors, was first identified by virtue of its rearrangement in human chronic myelomonocytic leukemia associated with a t(5;12) chromosomal translocation (Golub et al., 1994). Subsequently, the TEL gene has been shown to be rearranged in a variety of human leukemias, each subtype being associated with the formation of a chimeric protein product. Fusion partners for TEL include the platelet‐derived growth factor receptor‐β (PDGFRβ) and Abelson (ABL) tyrosine kinases (Golub et al., 1994, 1996; Papadopoulos et al., 1995), the transcription factors AML‐1 (Golub et al., 1995; Romana et al., 1995a), EVI1 (Raynaud et al., 1996a) and proteins of unknown function (Buijs et al., 1995). The TEL/AML‐1 fusion is common in childhood acute lymphoblastic leukemia, where this gene rearrangement accounts for 25–30% of cases (Romana et al., 1995b; Shurtleff et al., 1995; McLean et al., 1996; Raynaud et al., 1996b). However, the contribution of TEL to the properties of chimeric proteins is not fully understood. In the TEL/PDGFRβ, TEL/ABL and TEL/AML1 fusions, an amino‐terminal conserved region referred to as the helix‐loop‐helix (HLH) or pointed domain, is contributed to the fusion protein. This domain, which is able to serve as a self‐association motif, is critical to the biological activity of the fusion protein (Carrol et al., 1996; Golub et al., 1996; Hiebert et al., 1996; Jousset et al., 1997). Conversely, the MN1/TEL fusion, the consequence of a t(12;22) translocation in myeloid leukemias, results in the aberrant expression of the TEL DNA‐binding Ets domain, suggesting that this fusion protein may initiate leukemic transformation through the dysregulation of genes normally regulated by TEL (Buijs et al., 1995). Finally, it has been observed that loss of heterozygosity at the TEL gene locus is associated with the TEL/AML‐1 gene rearrangement in childhood leukemia (Cave et al., 1995; Golub et al., 1995; Romana et al., 1995b; Stegmaier et al., 1995; Raynaud et al., 1996b; Takeuchi et al., 1996;). These findings have been interpreted to suggest that TEL loss of function may also contribute to the process of leukemogenesis.

TEL is a widely expressed nuclear protein which recognizes DNA sequences containing a GGAA core motif through a highly conserved 85 amino acid ETS domain (T.R.Golub et al., unpublished data; Poirel et al., 1997). The amino‐terminal HLH or pointed domain is also highly conserved among a subset of Ets proteins, including Ets‐1, Ets‐2, Erg, Fli‐1, GABPα and the Drosophila proteins Yan and Pointed (Wasylyk et al., 1993; O'Neill et al., 1994). While this domain facilitates self‐association of TEL, such oligomerization properties have not been observed for other members of the family, where the function of this domain remains obscure. Nonetheless, it has been established recently that the corresponding region of Ets‐1 and Ets‐2 is required for full transactivation and particularly for synergy with the Ras pathway (Galang et al., 1994; Yang et al., 1996). The transcriptional properties of TEL are not fully characterized. Studies suggest that the fusion of TEL to AML‐1 results in conversion of AML‐1, a transcriptional activator, to a transcriptional repressor (Hiebert et al., 1996). Preliminary studies of the normal TEL protein similarly suggest that TEL may function as a transcriptional repressor (T.R.Golub et al., unpublished data). In vivo targets for TEL, however, are not known.

Members of the ETS family of transcription factors are important in diverse developmental processes. For example, the Drosophila Yan protein acts as a negative regulator of photoreceptor cell development in the eye (Lai and Rubin, 1992) and is itself negatively regulated by the Ras/MAPK pathway (O'Neill et al., 1994). In mice, the expression patterns of Ets‐1 and Ets‐2 during mouse development suggest potential roles in cell proliferation and differentiation (Maroulakou et al., 1994). Indeed, studies in chimeric mice have shown that Ets‐1 is essential for the survival of T lymphoid cells and for the maintenance of a normal pool of B lymphoid cells (Bories et al., 1995; Muthusamy et al., 1995). Ets‐2 is required for early embryonic development (R.Oshima et al., personal communication). Other Ets‐related proteins, including PU.1 and Fli‐1 are essential for the commitment or differentiation of hematopoietic lineages (Scott et al., 1994; McKercher et al., 1996; Melet et al., 1996).

To pursue the normal role of the TEL protein we have examined the expression of the TEL gene in mouse development and generated knockout mice by gene targeting in embryonic stem (ES) cells. We report here that TEL is widely expressed with increased expression in neural tissues, developing kidney, lung and liver in the embryo. Analysis of loss of TEL function demonstrates that TEL is essential for normal development. While not strictly required for yolk sac hematopoiesis, the TEL protein is essential for maintaining integrity of the developing vascular network in the yolk sac and for survival of neural and mesenchymal cells within the embryo. Thus, these studies establish TEL as a critical regulator in the survival of multiple cell types during early embryonic development.


Widespread expression of TEL in the embryo and adult mouse

First we examined the pattern of TEL expression in the developing embryo and in adult tissues. By Northern blot analysis and in situ hybridization, TEL mRNA is detected in the embryo as early as E7.0; expression is markedly increased at E17 (Figure 1A). At E8.5 and E9.5, TEL is expressed throughout tissues of the embryo proper and the yolk sac (Figure 2A–D and data not shown). At E12.5, higher transcript levels are seen in multiple tissues and organs, including developing lung, kidney and liver. Notably, higher TEL expression is also detected in the cranial nerve ganglia, the dorsal root ganglia, and the ventral region of the caudal neural tube (Figure 2E–G). Finally, TEL mRNA is expressed in several adult tissues (Figure 1B).

Figure 1.

Northern blot analysis of TEL mRNA on staged mouse embryos and adult tissues. (A) Poly‐A mRNA from embryonic (E) day 7, 11, 15, 17 and (B) adult mouse tissues (MTN blots, Clontech) were hybridized with TEL cDNA probes as described in Materials and methods. Two principal transcripts were detected in these tissues. The large transcript represents full‐length TEL mRNA; the nature of the small transcript, however, has not been analyzed. H, heart; B, brain; S, spleen; L, lung; Lv, liver; M, skeletal muscle; K, kidney; T, testis.

Figure 2.

In situ hybridization of TEL mRNAs on mouse embryos. Frozen embryo sections from E 8.5 (A–D) and E 12.5 (E–G) were hybridized with digoxigenin‐11‐UTP labeled sense (A, C and E) and antisense (B, D, F and G) TEL probes as described in Materials and methods. (E) and (F) represent higher magnification of the caudal neural tube region depicted in (G). YS, yolk sac; Mb, midbrain; Lv, liver; Kd, kidney; Ln, lung. Arrowhead indicates trigeminal ganglia; arrows indicate dorsal root ganglia; asterisks indicate ventral region of the caudal neural tube.

Targeted disruption of TEL results in early embryonic lethality

To inactivate the TEL gene, two exons comprising the TEL DNA‐binding domain and an additional 26 bp of the immediate 3′ adjacent exon were replaced by a PGK‐neo cassette (Figure 3A). Of 287 G418‐ and gancyclovir‐resistant ES clones analyzed, five contained an appropriately targeted TEL locus (Figure 3B). Two clones were injected into C57BL/6 blastocysts and led to germline transmission of the mutation. In addition, three independently derived double knockout TEL ES clones (TEL−/−) were obtained by targeting the second allele with a replacement vector containing a PGK‐hygromycin resistance cassette (Figure 3B)

Figure 3.Figure 3.
Figure 3.

Gene targeting of the murine TEL locus. (A) Targeting strategy. The homologous recombination event replaces the TEL DNA‐binding domain (DBD) exon 5, 6 and portion of exon 7 with a PGK‐neo cassette. X, XhoI; S, SalI; Sp, SpeI; B, BamHI; EV, EcoRV; C, ClaI. (B) Generation of targeted ES cells. G418‐ and gancyclovir‐resistant ES cell clones were screened by Southern blot analysis with both 5′ (left panel) and 3′ (middle panel) external probes using EV/Xho and EV/Spe digests respectively. Expected fragments hybridize with both probes as depicted in (A). TEL−/− ES cell clones were obtained by targeting the second wild‐type TEL allele with a PGK‐hygromycin cassette. Hygromycin‐resistant clones were screened as described above. Three TEL−/− ES cell clones were obtained and one representative clone is shown (right panel) (C and D) Absence of TEL expression in TEL−/− ES cell clones and TEL−/− embryos. Northern blot (C) and RT–PCR analyses (D) were performed on TEL−/− ES cell clones (C) and E 9.5 embryos (D) respectively. Upper panel in (C) shows TEL mRNA, lower panel, β‐actin: upper panel in (D) shows amplified 304 bp TEL PCR product, lower panel, HPRT. Note the absence of TEL mRNA or PCR products in TEL−/− ES and embryos.

To establish that the targeting event resulted in a null mutation, Northern analysis was performed on three TEL−/− ES cell clones. As shown in Figure 3C, no TEL mRNA transcripts were detected in these clones, whereas TEL mRNA was readily detected in wild‐type ES cell samples. In addition, when sets of primers specific for the TEL HLH domain, DNA‐binding domain, and exons downstream of these domains were used in an RT–PCR assay to amplify TEL mRNA, no TEL transcripts were detected in TEL−/− embryos; products of the predicted size were seen in control or heterozygous littermates (Figure 3D, and data not shown). Thus, the engineered mutation produces a null allele of the TEL locus.

F1 mice heterozygous for disruption of the TEL gene appeared normal in size, fertility, and overall development (data not shown). Heterozygous mice were intercrossed to generate homozygous mutants. Of 168 neonates genotyped, no homozygotes (TEL−/−) were identified (Table I). Likewise, no TEL−/− embryos were found after E13.5. Although TEL−/− embryos could be retrieved at E11.5, they were under‐represented and grossly retarded in their development (Table I and data not shown). These findings indicate that loss of TEL function results in early embryonic lethality.

View this table:
Table 1. TEL homozygotes die before E 11.5 of gestation

Failure to maintain yolk sac blood vessel formation in TEL−/− mice

Analysis of E8.5 embryos from TEL+/− F1 intercrosses revealed no discernible differences between TEL−/− and control embryos (Table I and data not shown). However, at E9.5, 65% of TEL−/− embryos (Table I, type I mutant) exhibited yolk sacs that lack the branching vitelline vessels normally present in controls (Figure 4A and B). Instead, as highlighted by anti‐PECAM antibody staining, a marker for endothelial cells (Baldwin et al., 1994), a honeycomb‐like network of interconnecting sinusoids was observed (Figure 4C and D). Histologically, branching vitelline vessels of wild‐type yolk sacs appear in cross‐section as large luminal spaces attached to the undersurface of the mesodermal layer (Figure 4E). Although TEL−/− mutant yolk sacs develop ample sinusoidal spaces or blood islands with lumens, similar in diameter to wild‐type and containing pooled primitive red cells, larger lumens are absent (Figure 4F). Endodermal and endothelial cells lining the sinusoids in mutant yolk sacs appear normal (Figure 4E and F). This yolk sac blood vessel defect is seen as early as E9.0 (data not shown). At E9.0–9.5, TEL−/− embryos are grossly appropriate in appearance (Figure 4G and H) with regularly beating hearts and visible blood in the circulation. In addition, the number of somites and appearance of the cranial prominence are similar in controls and TEL−/− embryos (Figure 4G and H). Placento‐allantoid fusion also occurs normally (data not shown). Interestingly, in contrast to the abnormal appearance of yolk sac blood vessels, the dorsal aorta, intersomitic vessels and branching head veins of TEL−/− embryos proper are evident by whole mount staining with anti‐PECAM antibody, and appear histologically normal (data not shown). Yolk sac‐derived embryonic red blood cells are present in the blood vessels of TEL −/− embryos, suggesting that yolk sac sinusoidal spaces communicate with the embryonic circulation through small anastomosing channels rather than the vitelline vessels.

Figure 4.

Yolk sac angiogenic defects in TEL−/− embryos. (A and B) E9.5 TEL+/− and TEL−/− yolk sacs. Note the lack of branching vitelline vessels in TEL−/− yolk sac. (C and D) Whole mount anti‐PECAM antibody staining of E9.5 TEL+/− and TEL−/− yolk sacs shows presence of honeycomb‐like vasculature in both TEL+/− and TEL−/− yolk sacs, but lack of branching vitelline vessels in TEL−/− yolk sac. (E and F) Hematoxylin and eosin staining on paraffin‐embedded TEL+/− and TEL−/− yolk sacs. Arrow in (E) identifies a large lumen indicative of large vitelline vessel in (A) and (C) that is observed in TEL+/+ and TEL+/−, but not TEL−/− yolk sacs. En, extra‐embryonic endoderm; Me, extra‐embryonic mesoderm; Am, Amnion. (G and H) E 9.5 TEL+/− and TEL−/− embryos. Note the stage appropriate development of the TEL−/− embryo. (I) Two types of TEL−/− phenotypes at E 9.5. Two‐thirds of TEL−/− yolk sacs (type I) exhibit yolk sac phenotype as described in (B), (D) and (F), one‐third of TEL−/− yolk sacs (type II) exhibit branching vitelline vessels indistinguishable from that of the control littermates as in (A). (J) Type I and Type II TEL−/− yolk sacs at E 10. Note the disintegration of the vitelline vessels of the type II mutant as indicated by the arrow pointing to a residual vessel. (K) Embryos from the same litter of E 10 shown in (J). Asterisk indicates a type II embryo; note that it is developmentally less retarded as compared with the other TEL−/− embryo (type I). (L) TEL−/− embryos at E 10.5. Both type I and type II mutant embryos are grossly retarded; approximately two‐thirds of these mutants exhibit enlarged pericardial sac.

Some 35% of TEL−/− embryos (Table I, type II mutant) displayed normal‐appearing yolk sac vitelline vessels at E9.5 (Figure 4I). However, by E10, all mutants (type I and type II) lack normal vitelline vessels. Occasionally, some residual remnants of vessels formed earlier in type II embryos can be seen as streak‐like structures (Figure 4J). At this stage (E10), type II embryos are also developmentally less severely retarded than type I embryos (Figure 4K) correlating with the later onset of the yolk sac blood vessel defect.

By E10.5, all TEL−/− embryos are grossly abnormal. The majority of embryos (about two‐thirds) are markedly growth‐retarded and exhibit an enlarged pericardial sac (Figure 4L). This most likely reflects the fraction of type I mutants that failed to develop normal yolk sac vasculature at E9.5. Histologically, these embryos are largely necrotic (data not shown). The remaining one‐third of embryos, presumably type II mutants—those that had developed normal yolk sac vasculature earlier—lose the vitelline vessels completely and exhibit the same honeycomb‐like vasculature as type I mutants (data not shown). Therefore, the phenotype of type II mutant embryo indicates that vascular cells of the yolk sac develop, but that they can not be maintained after initial remodeling of primary vessels into a more complicated vascular network. Thus, absence of TEL leads to a defect in angiogenesis in the developing yolk sac.

Apoptotic cell death in TEL−/− embryos

The embryo proper of both type I and type II TEL−/− mutants appears grossly normal with respect to overall development and vascular structure. However, histological examination reveals cell death restricted to particular regions. This cell death can be attributed to apoptosis as revealed by labeling 3′‐OH DNA ends (ApoTag; Gavrieli et al., 1992). Regions of apoptosis include the developing neural tube (Figure 5A–C), the mesenchymal tissues immediately adjacent to the primitive gut (Figure 5D–F) and along the entire body length (Figure 5G and data not shown). In addition, neural crest‐derived cranial nerve ganglia also contain numerous apoptotic cells (Figure 5H–L). In general, those areas with evident apoptosis correlate with regions in which TEL is most highly expressed. Specifically, apoptosis is prominent in the cranial nerve ganglia V trigeminal and VII–VIII facial–acoustic regions (Figure 2E). Thus, in addition to the requirement of TEL in maintaining the integrity of yolk sac vascular network, TEL is also essential for the survival of mesenchymal cells and neural tissues.

Figure 5.

Apoptosis in E10 TEL−/− embryos. (A, D and H) Hematoxylin and eosin staining of the TEL−/− embryos. (A) Transverse section of the neural tube; (D) sagittal sections of the primitive gut; and (H) cranial nerve ganglia regions. Arrows in (A) and (D) point to pyknotic nuclei indicative of cell death; magnification ×400. The single asterisk in (H) and (J) indicates V trigeminal region and double asterisk indicates VII–VIII facial–acoustic nerve regions; ×200. OV, otic vesicle. (B and C) TUNEL assay was performed on the TEL+/+ (B) and TEL−/− (C) neural tube tissue sections; ×1000. (E and F) TUNEL assay on the TEL+/+ (E) and TEL−/− (F) embryo sections of the primitive gut; ×1000. (G) Representative TUNEL assays show mesenchymal cell death observed along the entire body length of the TEL −/− embryos; magnification, ×1000. (I and J) TUNEL assay on the TEL+/+ (I) and TEL −/− (J) tissue sections of the cranial nerve ganglia regions; ×400. (K) TUNEL assay on the V trigeminal region; ×1000. (L) TUNEL assay on the VII–VIII facial–acoustic nerve region; ×1000. Note the high degree of apoptosis, as indicated by brownish nuclei staining, only observed in TEL−/− embryos.

Normal myeloerythroid hematopoiesis in TEL−/− embryos at the yolk sac stage

As the TEL gene was first identified through a chromosomal translocation in human leukemia (Golub et al., 1994), we have assessed the requirement for TEL in hematopoiesis in the embryo. Several other transcription factors identified in a similar fashion have been shown to be essential for normal hematopoiesis. For example, AML‐1 (CBFαA2) protein is required for fetal liver, but not yolk sac, hematopoiesis (Okuda et al., 1996; Wang et al., 1996), whereas SCL/tal‐1 is essential for development of all hematopoietic lineages (Porcher et al., 1996; Robb et al., 1996). Hematopoietic colony‐forming assays were performed from E9.5 yolk sacs by plating cells onto methylcellulose‐containing media supplemented with appropriate cytokines (KL/Epo for erythroid colonies and IL‐1/IL‐3/GM‐CSF/G‐CSF for macrophage colonies). In addition, mixed erythroid–myeloid colonies were obtained by growth in a combination of these growth factors. As shown in Figure 6A, the total number and content of the erythroid, macrophage and mixed colonies derived from precursors present in TEL−/− yolk sacs do not differ from control littermates. Moreover, similar numbers of primitive and definitive erythroid colonies are observed upon in vitro differentiation of wild‐type and TEL−/− ES cells (Figure 6B). Thus, TEL is not strictly required for the differentiation and maturation of erythroid and macrophage cell lineages at the yolk sac stage.

Figure 6.

Normal myeloerythroid hematopoiesis in TEL −/− mice at the yolk sac stage. (A) Yolk sac progenitor assays were performed as described in Materials and methods. The data shown were obtained from one litter at E9.5. Three independent litters were assayed at E9.5 with similar results. (B) In vitro differentiation of ES cells was performed on three independently targeted TEL−/− ES cell clones and two TEL+/− ES cell clones. Solid bars indicate the numbers of progenitors from TEL−/− yolk sacs or ES cell clones. Colonies were counted and harvested for cytospin and May–Grunwald–Giemsa staining after 4–7 days of in vitro culture with the indicated growth factors.


The TEL gene is frequently rearranged by chromosomal translocation in human leukemias of both myeloid and lymphoid origins. In chronic myelomonocytic leukemia (CMML), one TEL allele is disrupted and fused in‐frame to the tyrosine kinase domain of the platelet‐derived growth factor beta (PDGFβ) receptor to generate a dominantly‐acting oncogene (Golub et al., 1994; Carroll et al., 1996; Jousset et al., 1997). Loss of the normal TEL allele in association with translocation of the TEL gene to the AML‐1 (CBFαA2) locus in childhood acute lymphoblastic leukemia suggests that TEL loss of function may also contribute to pathogenesis of leukemia (Cave et al., 1995; Golub et al., 1995; Romana et al., 1995b; Stegmaier et al., 1995; Raynaud et al., 1996; Takeuchi et al., 1996). To initiate study of its in vivo role, we have used gene targeting in ES cells to inactivate the TEL gene in mice. Our results demonstrate that TEL is essential for normal development and is specifically required for maintaining blood vessel integrity within the developing yolk sac and for survival of different cell types in the developing embryo.

The formation of blood vessels involves two distinct cellular processes: (i) vasculogenesis, the in situ differentiation of angioblast and the subsequent assembly into primary vascular channels (Risau, 1995; Risau and Flamme, 1995); and (ii) angiogenesis, the proliferation of pre‐existing endothelial cells to expand and remodel the vascular network (Pardanaud et al., 1989). The latter process is believed to include the formation of the vascular wall by recruitment of pericytes/smooth muscle cells from mesenchymal progenitor cells and neural crest cells (Nakamura, 1988; Kirby and Waldo, 1995). In the absence of TEL, vasculogenesis in the yolk sac and embryo appears to occur normally, suggesting that TEL is not required for the proliferation or differentiation of endothelial cells per se. In contrast, a defect in maintenance of the vascular network in the yolk sac is seen. In two‐thirds of the TEL−/− yolk sacs, branching vitelline vessels are not observed at E9.5. However, one‐third of the yolk sacs exhibit normal branching vessels at E9.5 which are not present at E10.5. These observations indicate that, while initiation of yolk sac angiogenesis occurs in the absence of TEL, the integrity of a more complex vascular network cannot be maintained. This phenotype is remarkably similar to that of embryos lacking tissue factor (TF) (Carmeliet et al., 1996b). In contrast to TF−/− yolk sacs, we observe an apparently normal, rather than reduced, number of mesenchymal cells expressing α‐actin in the TEL−/− yolk sacs (data not shown). Such differences suggest diverse mechanisms by which yolk sac angiogenesis is regulated and the integrity of the vascular network is maintained.

It has been hypothesized that organs of ectodermal or mesenchymal origin, such as brain and kidney, are vascularized by angiogenic mechanisms (Bar, 1980; Sariola et al., 1983). Similar to TF−/− embryos, TEL−/− embryos display normal vasculature within the embryo proper E9.5, at which time the yolk sac vascular defect is first apparent. It is likely that a complex vascular network within the embryo has yet to develop at this stage and the early death of these mutant embryos precludes the appreciation of an intra‐embryonic angiogenic defect, if present. A particularly striking and specific feature of the TEL−/− embryos is prominent mesenchymal cell apoptosis. Whether some of these apoptotic mesenchymal cells are progenitors of pericyte/smooth muscle cells is unknown. It is attractive to speculate that the failure to maintain a complex yolk sac vascular network is related to a function for TEL in preventing apoptosis in a critical cell population, either of endothelial or mesenchymal origin. In addition to mesenchymal cell apoptosis, TEL−/− embryos also exhibit apoptosis in regions of the neural tube and neural crest‐derived ganglia which normally display the highest levels of TEL mRNA transcripts. We speculate that apoptosis in these cell populations may lead to neural defects at later stages of the mouse development.

Several genes encoding receptors, their ligands, or a transcription factor involved in hypoxia response have recently been shown to be essential for proper yolk sac vascular development. These include receptor tyrosine kinases, such as Flk‐1 (Shalaby et al., 1995), Flt‐1 (Fong et al., 1995), Tie‐1 (Puri et al., 1995; Sato et al., 1995), and Tie‐2 (Dumont et al., 1994; Sato et al., 1995); the vascular endothelial growth factor VEGF (Carmeliet et al., 1996a; Ferrara et al., 1996), the Tie‐2 ligand, angiopoietin‐1 (Suri et al., 1996), TF (Carmeliet et al., 1996b); and arylhydrocarbon receptor nuclear translocator, ARNT (Maltepe et al., 1997). Recent study of the role of angiopoietin‐1 during embryonic angiogenesis (Suri et al., 1996) suggests that interactions between Tie‐2‐expressing endothelial cells and angiopoietin‐1‐producing mesenchymal cells underlies blood vessel remodeling (Folkman and D'Amore, 1996; Vikkula et al., 1996). We envision that loss of TEL function in an as yet unknown manner disrupts maintenance of these critical cellular interactions.

It is worth noting that Ets family of transcription factors, including Ets‐1, Ets‐2, yan and pointed, have been shown to be downstream targets of the Ras signaling pathway (O'Neill et al., 1994; Yang et al., 1996). Biochemical studies suggest that molecules involved in Ras signaling pathway, including rasGAP complex, GRB2 and SH‐PTP may be substrates for Flk‐1, Flt‐1 and Tie‐2 tyrosine kinase activity (Guo et al., 1995; Huang et al., 1995; Seetharam et al., 1995). Whether Ras signaling events result from interaction between these endothelial cell‐specific receptor tyrosine kinases and ligands lead to activation of an Ets‐like transcription factors, such as TEL, remains to be determined. Interestingly, recent knockout studies on mutants lacking the Ras signaling molecules GTPase‐activating protein (GAP) and GTP‐binding protein α13 subunit demonstrate yolk sac angiogenic defects (Henkemeyer et al., 1995; Offermanns et al., 1997) and neuronal apoptosis (Henkemeyer et al., 1995) similar to those in TEL−/− embryos. In addition, expression of Tie‐1, Tie‐2, Flk‐1, Flt‐1 and GAP mRNAs is not appreciably affected in the TEL−/− embryos (data not shown), suggesting that TEL might lie downstream of a receptor tyrosine kinase signal transduction pathway in yolk sac angiogenesis.

Members of the Ets‐family of transcription factors, especially Ets‐1, have been indirectly implicated previously in blood vessel formation based on the pattern of expression (Pardanaud and Dieterlen‐Lièvre, 1993) and the presence of putative ets‐binding sites in relevant genes (Risau and Flamme, 1995). Although our results establish TEL as essential for normal yolk sac vascular development, Ets‐1 appears to be dispensable at this stage (J.M.Leiden, personal communication), though it is required for the survival of T lymphoid cells (Bories et al., 1995; Muthusamy et al., 1995). Whether Ets‐family members other than TEL are critical for blood vessel formation remains to be determined. Our data add to the increasing evidence that Ets‐proteins serve to control apoptosis in diverse cell types.

Chromosomal translocations in human leukemia often activate expression of transcription factors or generate chimeric proteins containing a portion of a transcription factor. In several instances, notably SCL/tal‐1(Porcher et al., 1996), Rbtn‐2/LMO2 (Warren et al., 1994), and AML‐1/CBFαA2 (Okuda et al., 1996; Wang et al., 1996), the relevant factors are also essential for normal hematopoietic development. Our analysis, however, demonstrates that TEL is not strictly required for the proliferation or differentiation of erythroid and myeloid lineages at the yolk sac stage. However, these findings do not preclude roles for TEL in later fetal liver or adult hematopoiesis, or in lymphopoiesis. Analysis of chimeric mice made with TEL−/− ES cells injected into wild‐type or RAG‐2−/− blastocysts may permit more direct study of blood cell development in the absence of TEL.

In summary, by gene targeting in ES cells, we have demonstrated that the widely expressed Ets‐family member TEL is essential for the integrity of remodeled blood vessels in the developing yolk sac and is also required to prevent apoptosis in a variety of cell types within the embryo. Identification of target genes for TEL and the creation of developmentally regulated or tissue‐restricted knockouts should allow for further characterization of its function in blood vessel formation, neural development and hematopoiesis.

Materials and methods

Construction of the targeting vector

Murine TEL genomic sequences were isolated from a 129/sv strain λFIXII library (Strategene) with a 3.5 kb TEL cDNA probe which encompassed the TEL DNA‐binding domain. Inserts from positive phage were isolated as SalI fragments, subcloned into pUC18 plasmid, and subjected to restriction enzyme mapping. A 4.8 kb XhoI–SpeI fragment spanning two exons of the TEL DNA‐binding domain were replaced with a 1.8 kb PGK‐neo cassette. The resulting SalI fragment was then fused with XhoI‐digested BlueScript‐SK vector (Strategene). A 2 kb HSV‐TK cassette was then inserted into SalI site of this vector to generate the final targeting construct.

To obtain TEL−/− ES clones, a similar construct was assembled in which a PGK‐hygromycin‐resistant cassette replaced the PGK‐neo cassette.

Gene targeting in ES cells and generation of mutant mice

The TEL targeting construct was linearized with NotI and electroporated into J1 ES cells as described previously (Shivdasani et al., 1995). 5′ flanking 0.7 kb PstI fragment and 3′ flanking 0.7 kb XbaI–ClaI fragments were used to identify appropriately targeted ES cell clones by Southern blot analysis. TEL+/− ES cell clones were subjected to karyotyping to check for chromosomal integrity. Two ES clones with a normal karyotype were injected into C57BL/6 blastocysts to generate chimeras which contributed to the germline. Chimeras exhibited >90% contribution from ES cells on the basis of agouti coat color were used to mate with C57BL/6 mice. Genotyping of pups was performed by Southern blot analysis. Breeding of the TEL+/− mice and subsequent analysis of the homozygous mutant mice generated from two independently targeted ES cell clones were performed and yielded the same phenotype.

To confirm the null mutation in TEL−/− ES clones and embryos, total RNAs from the ES cells and embryos were prepared by extraction with RNAzol B (Tel‐Test Inc.) and subjected to either Northern analysis or RT–PCR. A combination of two probes: 0.63 kb and 0.56 kb of XhoI fragments from the mouse TEL‐7 cDNA plasmid were used for Northern analysis. For RT–PCRs, primers amplifying 304 bp of the TEL DNA‐binding domain (TEL‐1184:ACAAACATG‐ACCTATGAGAAA; TEL‐1487R:AGAAGTGTCCCTGCTATTCCC and 249 bp of constitutively expressed HPRT (Keller et al., 1993) were used under the following conditions: 94°C, 1 min; 56°C, 2 min; 72°C, 1 min for 30 cycles in a reaction mix containing 10% DMSO. The samples were then subjected to Southern analysis using a fragment spanning the DNA‐binding domain as probe. Control experiments without reverse transcriptase in the cDNA synthesis reactions did not show specific PCR products (data not shown).

Whole mount and in situ hybridization

Whole mount immunohistochemistry using anti‐PECAM antibody (Pharmigen) was performed as described previously (Schlaeger et al., 1995). In situ hybridization was performed according to the protocol of Wilkinson (1992). Frozen sections (12 μm) of staged embryos and yolk sac were prepared and hybridized at 65°C with digoxigenin‐11‐UTP (Boehringer‐Mannheim) labeled probes. A 350 bp BamHI–SmaI fragment representing the 3′ end of the TEL coding region and a 550 bp EcoRI–Eco47III fragment spanning the 5′ untranslated region were subcloned into BlueScript vectors. Both sense and antisense RNA were transcribed using T3 or T7 primers. Both antisense probes yielded similar results.

Histological analysis and TUNEL assay

Mouse embryos were fixed overnight in 10% buffered formalin and embedded in paraffin. Some sections (12 μm) were used for the TUNEL assay, while others were stained with hematoxylin and eosin for histological examination. The TUNEL assay was performed according to the manufacturer's protocol (Oncor) with some modification. Briefly, paraffin‐embedded tissue sections were de‐waxed and rehydrated through series of decreased concentration of ethanol. Following proteinase K reaction (20 μg/ml, 15 min) and quenching (2% H2O2, 10 min) treatments, sections were incubated with reaction mix as suggested by the manufacturer, except that 1 μl of TdT was used to reduce non‐specific labeling. After washing, the sections were incubated with peroxidase conjugated anti‐digoxigenin antibody for 30 min at RT and developed with DAB substrate (Vector). A DNase I‐treated section was included in each experiment as a positive control. Sections were then counterstained with methyl green according the manufacturer's suggestion (Oncor). Controls without added TdT enzyme did not show specific staining (data not shown).

Yolk sac progenitor and in vitro ES cell differentiation assay

E9.5 embryos were dissected under sterile conditions and were used for genotyping. Yolk sacs from embryos were incubated in Ca2+/Mg2+‐free PBS with 20% fetal calf serum (FCS) and 0.1% collagenase (Sigma) at 37°C for 1 h. Cells were then disaggregated by passage through a syringe and 22‐gauge needle (Wong et al., 1986). The yield from each yolk sac was 1–5×104 cells. Subsequently, 5×103 cells of each yolk sac were plated in α‐minimal essential medium supplemented with 0.9% α‐methylcellulose, 30% FCS, 1% BSA and growth factors. For erythroid colonies: 2 units/ml erythropoietin (Epo), 50 ng/ml recombinant c‐kit ligand (KL) were added; for macrophage colonies: IL‐1 (103 U/ml), IL‐3 (10 ng/ml), granulocyte colony‐stimulating factor (G‐CSF, 1 ng/ml), granulocyte‐macrophage CSF (GM‐CSF, 5 ng/ml) were added; for mixed erythroid–myeloid colonies: IL‐1, IL‐3, IL‐11 (5 ng/ml), KL, Epo, GM‐CSF, G‐CSF were added (combination). In vitro ES cell differentiation was performed as described (Keller et al., 1993; Porcher et al., 1996). Briefly, ES cell clones (TEL+/−, TEL−/− and wild‐type) were permitted to form embryo bodies. At day 9 of culture, embryo bodies were disaggregated and 2×104/ml of cells were replated onto methycellulose medium supplemented with Epo (2 U/ml) for differentiation of primitive erythroid colonies and KL (50 ng/ml) + Epo for differentiation of definitive erythroid colonies.


We thank Carol Browne and Kerrianne Cunniff for technical assistance, and Drs David Rowitch and Judah Folkman for helpful discussion. T.R.G. is a recipient of a Burroughs Wellcome Fund Career Award in the biomedical sciences.