SEARCH     advanced search my account e-alerts subscribe register
Journal home
Advance online publication
Current issue
Press releases
For authors
For referees
Contact editorial office
About the journal
For librarians
Contact NPG
Customer services
Site features
NPG Subject areas
Access material from all our publications in your subject area:
Biotechnology Biotechnology
Cancer Cancer
Chemistry Chemistry
Dentistry Dentistry
Development Development
Drug Discovery Drug Discovery
Earth Sciences Earth Sciences
Evolution & Ecology Evolution & Ecology
Genetics Genetics
Immunology Immunology
Materials Materials Science
Medical Research Medical Research
Microbiology Microbiology
Molecular Cell Biology Molecular Cell Biology
Neuroscience Neuroscience
Pharmacology Pharmacology
Physics Physics
Browse all publications
18 December 2000, Volume 19, Number 55, Pages 6443-6454
Table of contents    Previous  Article  Next   [PDF]
Original Paper
Mouse models in the study of the Ets family of transcription factors
Frank O Bartel1,2, Tsukasa Higuchi1,2 and Demetri D Spyropoulos1,2,3

1Center for Molecular and Structural Biology, Medical University of South Carolina, Charleston, South Carolina, SC 29425, USA

2Hollings Cancer Center, Medical University of South Carolina, Charleston, South Carolina, SC 29425, USA

3Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina SC 29425, USA


Correspondence to: D D Spyropoulos, Center for Molecular and Structural Biology, Medical University of South Carolina, Charleston, South Carolina, SC 29425, USA


The Ets family of transcription factors is one of a growing number of master regulators of development. This family was originally defined by the presence of a conserved DNA binding domain, the Ets domain. To date, nearly 30 members of this family have been identified and implicated in a wide range of physiological and pathological processes. Despite the likely importance of Ets-family members, each of their precise roles has not been delineated. Herein, we describe the elucidation of essential functions of a few of these family members in vivo using knockout mouse models. Of the knockouts generated to date, the majority shows important functions in hematopoiesis, ranging from PU.1, a principle regulator of myelo-lymphopoiesis, to Spi-B which regulates the proper function of terminally differentiated cells. Ets1 was shown to be of intermediate importance as a regulator of pan-lymphoid development. Other Ets family members such as Fli1 and TEL1 display distinct and/or overlapping functions in vasculo/angiogenesis, hemostasis and hematopoiesis. The remaining knockouts generated, Ets2 and Er81, show non-hematopoietic defects related to extraembryonic development and neurogenesis, respectively. The pioneering group of knockout models described reveals only the most distinct functions of each of these Ets family members. A better understanding of the roles and hierarchies of Ets family members in cellular differentiation will come with the generation of new null alleles in previously untargeted family members, more mutant alleles in members already disrupted, double knockouts, ES cell differentiation and chimera rescue experiments, and tissue-specific inducible knockouts. Oncogene (2000) 19, 6443-6454.


Ets family; knockout; targeted disruption; hematopoiesis; angiogenesis



A basic understanding of normal and abnormal cellular differentiation requires knowledge of lineage-specific markers and the intrinsic (cell autonomous) and extrinsic (non-cell autonomous) factors that guide the process. Arguably, hematopoiesis is the best studied complex mammalian cellular differentiation process. Now available are extensive repertoires of lineage-specific markers and extrinsic cytokines, which allow the identification of and the enrichment for specific differentiating populations. Lagging behind but gaining ground in this field, is the elucidation of the intrinsic factors that guide hematopoiesis. Gene targeting and knockout mouse technologies have shed some light on which factors are indispensable in hematopoiesis (Orkin, 1998; Orkin et al., 1999). Ets family transcription factors (Ghysdael and Boureux, 1997; Graves and Petersen, 1998; Papas et al., 1997; Watson et al., 2001) constitute a major class of hematopoietic regulators (Bassuk and Leiden, 1997). For each gene knockout herein described, we will discuss the targeted disruption, evidence for the generation of a null allele and resultant phenotypes (Table 1). This review summarizes current knockout mouse models that reveal indispensable roles for Ets family members in guiding hematopoiesis, as well as vasculo/angiogenesis, and other cellular differentiation processes (Figure 1).


Regulators of hematopoiesis

PU.1: The PU.1 gene disruption results in the most profound hematopoietic defects and has been the most extensively analysed Ets knockout to date. Both of the independently derived PU.1 knockout-mouse models demonstrate an absence of morphologically normal B cells and macrophages, disrupted granulopoiesis, and aberrant T-lymphopoiesis (McKercher et al., 1996; Scott et al., 1994). These mutants have normal megakaryopoiesis and normal numbers of erythroid progenitor cells. In both instances, targeting of PU.1 was by insertional disruption of the alpha3 helix of the Ets DNA binding domain (Pio et al., 1996). Also, data are presented that argue for the generation of null alleles. Western blot data demonstrating the absence of PU.1 protein in fetal livers of homozygotes (unspecified day) and the presence of only full length PU.1 protein at reduced levels in spleens of adult heterozygotes (homozygotes could not be assayed at this time due to late embryonic lethality) both suggest that the first targeted mutation is a null (Scott et al., 1997). Further characterization of transcripts from mutant and wild-type fetal livers would have facilitated this interpretation. Regarding the second targeted allele, Northern blot analysis demonstrated the absence of detectable PU.1 transcripts in embryonic day (E) 19 fetal livers (Iwama et al., 1998). Thus, for both targeted alleles data are present that are consistent with the generation of null alleles.

While both of the PU.1 knockout-mouse models were similarly targeted and arguably null alleles, they also exhibited a number of phenotypic differences, including the extent of defects in B-lineage commitment and neutrophil fate, embryonic lethality, and impaired erythropoiesis. Differences were also seen T cell development. Mice homozygous for the first targeted allele presented with the more profound defect in B-lineage commitment as demonstrated by the absence of identifiable pro- (B220+CD43+) or pre- (B220+CD43-) B cells, and the absence of Dmu-Jmu or Vkappa-Jkappa transcripts (Scott et al., 1994). Mice homozygous for the second targeted allele presented a less severe B cell phenotype. B cell progenitors (B220+) were apparent in day-1 bone marrow and spleen, though these cells express a combination of differentiation markers (CD43-, BP-1+, HSA1o/-) not observed in normal B-lymphopoiesis (McKercher et al., 1996). The BP-1+ marker is indicative of immature to mature B cells; however, these cells are normally also CD43+ and HSA+. Furthermore, neither Dmu-Jmu nor Vkappa-Jkappa rearrangements were detected in fetal liver or day-10 bone marrow, suggesting that differentiation is blocked at the pro-B cell stage. Thus, this novel B220+ cell type constitutes a feature of aberrant B-lineage development specific to the second targeted mutant. A comparison of both mutants does not allow a definitive assessment of the role of PU.1 in B- lymphopoiesis (Table 1).

T cell developmental defects were less severe but significant. Initial reports suggested no detectable T cells in PU.1 mutant mice homozygous for the first targeted allele (as evidenced by little or no Thy-1, CD4, or CD8 positive cells) (Scott et al., 1994). Subsequent fetal thymic organ culture indicated the presence of a small population of mature T cells, although the predominant population of thymocytes express cell surface markers corresponding to uncommitted T cell progenitors (Spain et al., 1999). The population of mature T cells was capable of producing IL-2 and proliferating in response to CD3epsilon stimulation demonstrating commitment in vivo. Chimera analysis to rescue the embryonic lethality demonstrated failure of PU.1-/- cells to contribute to any hematopoietic lineages in adults (Scott et al., 1997). In contrast, the demonstration of post-natal T cell development in vivo was possible using the second targeted mutant due to its greater viability (McKercher et al., 1996). While neonates showed no T cell receptor (TCR) alpha/beta+, CD3+, CD4+ or CD8+ thymocytes, day 5 mice were found to have CD4+ and CD8+ thymocytes, albeit at greatly reduced numbers. One to 10% of these cells migrated to the spleen, indicating that PU.1 is dispensable for cell migration/homing to the spleen (McKercher et al., 1996). However, transplantation into lethally irradiated adult recipients allowed quantitation, indicating abnormal homing potential to the bone marrow, spleen, thymus and liver (Fisher et al., 1999). Results of these experiments, using fluorescently-tagged wild-type and PU.1-/- E14.5 immature T lymphocytes (AA4.1+), indicated an 11-fold reduction in homing potential to the bone marrow, twofold reduction to the spleen and thymus and a ninefold accumulation of these cells in the liver within 48 h. Thus, PU.1 is required for normal T cell homing and fetal T-lymphopoiesis, although its role in post-natal T-lymphopoiesis remains unresolved.

Disruption of the PU.1 gene also results in significant defects in myelopoiesis. Monocytes/macrophages were shown to be absent in both PU.1 deficient mice (McKercher et al., 1996; Scott et al., 1994; Tondravi et al., 1997). Mice mutant for the first targeted allele showed no myeloperoxidase staining or lysozyme staining, indicative of the absence of granulocytes, monocytes, and macrophages (Scott et al., 1994). Likewise, neonates to 15 day old pups mutant for the second targeted allele show absence of detectable nonspecific esterase staining (indicative of monocyte/macrophage) (McKercher et al., 1996). Further, immunohistochemical staining of spleen, thymus, and liver detected no macrophages inPU.1-/- mice prior to day 14. A small population of F4/80+ cells (indicative of macrophages) was seen in day 14 mice, but these cells presented with abnormal morphology and altered distribution. Additionally, M-CSF or GM-CSF were unable to induce in vitro differentiation of macrophages (Anderson et al., 1998a; DeKoter et al., 1998; McKercher et al., 1996). However, fetal liver cells bearing the second targeted allele could be induced in vitro to commit to the monocytic lineage (Henkel et al., 1999; Lichanska et al., 1999). Mutants for the second targeted allele also lack osteoclasts, which are of myeloid origin and develop osteopetrotic features (Tondravi et al., 1997). Chimeric mice and stem cell transplantation experiments have been used to show that the monocyte/macrophage/osteoclast defects in PU.1-/- mice result from a cell-autonomous requirement for PU.1, arguing that PU.1 mediated gene regulation is essential for this lineage-specific differentiation (Scott et al., 1997; Tondravi et al., 1997).

Both PU.1-/- mice lack terminally differentiated or functionally competent neutrophils. In the first targeted mutant, which displays the more extreme phenotype, neutrophils were undetectable at E16.5 using Gr-1 expression, lysozyme staining, or by in vitro clonogenic assays, suggesting an essential role for PU.1 in commitment to the neutrophil lineage (Scott et al., 1994). Conversely, mice mutant for the second targeted allele were found to have a small population of cells with neutrophil characteristics (normal morphology, chloracetate esterase staining, and Gr-1+); however, these cells were found to lack functional competence (chemokine response, superoxide generation, bacterial uptake and killing) (Anderson et al., 1998b). Results from this mutant would therefore suggest that PU.1 is essential for proper development of functionally competent neutrophils, and not essential for commitment to the neutrophil fate. The discrepancy in neutrophil fate may be reflective of time of lethality. Finally, thymic dissection and fetal thymic organ culture of mutant embryos for the first targeted allele also revealed that PU.1 is required for development of myeloid-derived dendritic cells, but not lymphoid-derived dendritic cells (Guerriero et al., 2000). Studies with mice carrying the second targeted allele revealed absence of lympoid- and myeloid-derived dendritic cells from 10 day old pups, which is the first example in which the second targeted allele produced mice with a more severe phenotype (Anderson et al., 2000). Thus, mice mutant for both targeted alleles present with a defect in granulo-monocytic progenitors that show the most profound inability to produce monocytes/macrophages with different effects on granulopoiesis.

While mice mutant for both targeted alleles largely showed varying degrees of defects within the same lineages, mice mutant for the first targeted allele also present with embryonic lethality and impaired erythropoiesis (Scott et al., 1994) not observed in mice mutant for the second targeted allele (McKercher et al., 1996). Onset of anemia was observed between E14.5-17.5, and was of variable severity (hematocrit 10-35%; normal is 40%). Histological analysis of the homozygous mutants revealed reduced numbers of reticulocytes and mature erythrocytes. PU.1-/- ES cells were able to contribute to fetal but not adult erythropoiesis in chimeric mice, but PU.1-/- fetal hematopoietic progenitors were able to contribute to adult erythropoiesis in lethally irradiated adult recipients (Scott et al., 1997). The interpretation of this erythroid defect is further complicated by the loss of phenotype observed following back-cross onto a C57BL/6 background (Fisher and Scott, 1998; Simon, 1998).

With the exception of the lymphoid-dendritic cell lineage defect, all other phenotypes were more severe in the first targeted disruption of PU.1. This is interesting considering both targeted disruptions were in the Ets DNA binding domain and presumably null alleles. Although reasons for this paradox are unclear, they could include differences in genetic background, microbial flora in mouse colonies, or the nature of the two mutations. The loss of the anemia phenotype after repeated back-crosses to C57BL/6 suggests that genetic background contributes to phenotypic differences although some argue that genetic background is not a factor (McKercher et al., 1996). It has not been reported whether the viability of these back-crossed mice also improved. Concerning microbial flora, post-natal treatment of pups with antibiotics extended the life-span of homozygotes from 48 h to up to 17 days, suggesting microbial contribution to phenotype (McKercher et al., 1996). Regarding the nature of the two mutations, the presence of an as yet undetected neomorphic or hypomorphic mutant protein could account for the differences observed in the two knockouts. Such proteins would lack DNA binding function but may retain a variety of DNA binding-independent functions. SSFV infection may be a useful approach to enhance the detection of a hypo/neomorphic product, similar to the approach used with the first Fli1 mutant (Melet et al., 1996). It has also been suggested that orientation of the promoter driving transcription of Neo may contribute to the phenotype by activation or repression of upstream or downstream elements (McKercher et al., 1996). We believe that value of these PU.1 knockouts will be realized when a true null PU.1 allele is defined.

Ets1: The analysis of targeted mutants has revealed an essential role for Ets1 in the differentiation of all lymphoid lineages. Two different targeting strategies were used to disrupt the Ets1 gene, both generating null alleles that were phenotypically indistinguishable (Bories et al., 1995; Muthusamy et al., 1995). We will refer to the different mutations as the 3' and 5' targeted alleles. In the 3' targeted allele, a PGK Neo cassette replaced exons VIII and IX, encoding the DNA binding domain (Bories et al., 1995). In the 5' targeted allele, a PGK Neo cassette replaced exon III and most of exon IV, which codes for a protein-protein interacting domain (Muthusamy et al., 1995). The absence of Ets1 transcripts in a Northern blot on total RNA from lipopolysaccharide (LPS) stimulated splenic lymphocytes suggested that the 3' targeted allele was a null allele (Bories et al., 1995). The absence of detectable Ets1 protein in an immunoblot of thymocyte whole-cell extracts using polyclonal anti-Ets1 antibody also suggested that the 5' targeted allele was a null allele (Muthusamy et al., 1995). Since Ets1 protein from wild-type control cells was only faintly visible in the immunoblot, the latter result was less compelling, though concordance of phenotype renders this a moot point.

ES cells homozygous for both targeted alleles were used to generate Rag-2-/-/Ets1-/- chimeras. Rag-2-/- lymphoid cells fail to undergo V(D)J recombination, resulting in lymphoid progenitor apoptosis early in development (Chen et al., 1993). Thus, mature lymphoid cells of Rag-2-/-/Ets1-/- chimeras would be derived entirely from the Ets1-/- cells. Results of chimera analyses demonstrated that both targeted alleles were phenotypically equivalent. Thymuses of chimeric mice showed reduced cellularity by 8 weeks of age. This could be due to low percentage chimerism or defects in Ets1-/- cellular development. Flow cytometric analysis revealed an increased proportion of CD4-/CD8- and CD25+ cells and a statistically significant reduction in the number of immature CD4+/CD8+ cells in chimeric mice. A threefold reduction in the number of splenic T cells and significant reduction in lymph node T cells was also observed in these mice, although a normal proportion of CD4+ and CD8+ cells was maintained. Mature splenic T cells expressed normal levels of CD3 and TCR alpha/beta complex. In vitro stimulation of purified splenic or thymic T cells with anti-CD3 antibody or concanavalin A (conA) showed a severe defect in TCR-mediated proliferation. Unstimulated Ets1-/- splenic and thymic T cells showed higher levels of spontaneous cell death and increased levels of apoptosis in culture. Activation of cultured T cells by conA and phorbol myristate acetate (PMA) did not rescue the survival defect, though early activation markers CD69 and CD25 were induced. Results of these proliferation studies most clearly indicate that T cell defects were related to the loss of Ets1 function. Collectively, these results suggest that Ets1 is required for the maturation of T cells and/or the survival of mature T cells. The induction of early activation markers further suggests that Ets1 is not required for signal recognition, but indispensable for the ability to mount a complete response. Specially, Ets1 is likely necessary for proper response to co-stimulation (signal 2; IL-2, CD28, etc.) rather than for TCR engagement (signal 1).

Ets1 was also found to be important in B lymphopoiesis. While normal numbers of B220+ cells and normal numbers of IgM+ splenic B cells were detected in Rag-2-/-/Ets1-/- chimeras, a larger population of these cells exhibited reduced levels of B220 expression relative to wildtype. This result suggests either the accumulation of a larger number of immature cells or an increased number of IgM+ plasma cells, which is reminiscent of Hyper-IgM syndrome (Muramatsu et al., 2000; Revy et al., 2000). Cytological analysis of splenic lymphocytes revealed that between 19 and 34% of these B220+ cells were plasma cells, suggesting an increased number of IgM+ plasma cells. In addition, serum IgM levels were elevated 5-10-fold compared to wild type (Barton et al., 1998). Splenic B cells were capable of class switching in vitro, however, suggesting that loss of expression of genes such as cytidine deaminase were not involved in this phenotype (Bories et al., 1995). Ets1-/- B cells proliferated normally in response to LPS and anti-CD40, suggesting that this proliferative defect was observed only in T cells. In summary, results of Rag-2-/-/Ets1-/- chimera analysis suggest that Ets1 is essential in different ways for T and B cell maturation: Ets1 is required for the maintenance of resting T cells (as predicted prior to the generation of the Ets1 knockouts; (Bhat et al., 1990) and to prevent precocious maturation of B cells. While similar results were obtained in the Rag-2 complementation system using Ets1-/- ES cells carrying both targeted alleles, some of these results contrast with those obtained in homozygous knockout mice (as described below). Such differences suggest that non-cell autonomous effects involving Rag-2-/- cells contributed to the T and B cell phenotypes described; for example, the high incidence of apoptosis in early Rag-2-/- lymphoid progenitors.

Ets1 targeted ES cells carrying the 5' targeted allele were subsequently used to generate homozygous Ets1 knockout mice (Barton et al., 1998; Walunas et al., 2000). Fifty per cent of the Ets1 homozygotes died by 4 weeks of age. Surprisingly, no details related to the cause of this early post-natal lethality were presented (Barton et al., 1998). This phenotype constitutes the first observed distinction between the Ets1 knockout mice and the Rag-2 complemented chimeras and therefore suggests that lethality is non-lymphoid in origin. Both the Rag-2-/-/Ets1-/- chimeras and the Ets1 knockout mice presented with elevated proportions of IgM+ B cells and with T cell activation defects. In contrast to the Rag-2-/-/Ets1-/- chimeras, there were no differences between wildtype and Ets1 knockout mice with respect to the relative proportions of CD4+CD8+, CD4-CD8-, and single positive (CD4+CD8- or CD4-CD8+) thymocytes. These results argue that Ets1 is essential for T cell activation but not essential for T cell maturation, and that the maturation defects observed in the Rag-2-/-/Ets1-/- chimeras were an artifact of the Rag-2 complementation system.

Changes in NK (Barton et al., 1998) and NK T (Walunas et al., 2000) cell populations were observed in the Ets1 knockout. Flow cytometric analysis of freshly isolated splenocytes, as well as lymph node and bone marrow cells, indicated a marked reduction in DX5+CD3- mature NK cells. NK cytolytic activity in the Ets1 knockout was shown to be diminished or absent against NK susceptible tumor cells, YAC-1 or RMA-S, even when the NK cell population was purified 100-fold by fluorescence activated cell sorting (FACS). Furthermore, loss of NK cell activity in vivo was demonstrated by injecting RMA-S tumor cells into wildtype or Ets1-/- mice. None of the wildtype mice formed tumors, whereas six of seven Ets1-/- mice developed tumors. These results suggest that Ets1 is required for the generation of NK cells with efficient cytolytic activity in vivo. Both RNAse protection assays and flow cytometric analysis reveal that the loss of NK cytolytic activity was not due to the loss of IL-2Ralpha, -beta, -gamma cytokine receptors, as they were detected on the surface of Ets1-/- splenocytes. Further, the addition of supplemental IL-2, IL-15, or both together were unable to restore the defect in NK cell function.

Several lines of evidence also point to a requirement for Ets1 in the normal development of the related NK T cell lineage (Walunas et al., 2000). Flow cytometric analysis of lymphocytes from spleen, thymus, and mesenteric lymph nodes reveals a dramatic reduction in the number of NK1.1+ CD4+ T cells. Additionally, when hepatic lymphocytes expressing an intermediate level of the TCR complex (which is characteristic of NK T cells) are separated by FACS, NK 1.1+ T cells are virtually undetectable. Semi-quantitative PCR analysis of Valpha14 transcripts from both Ets1-/- and CD1-/- livers (which are known to lack NK T cells) showed diminished expression relative to wild type livers. Although there is a significant positive association between the Valpha14 expression and NK T cells, a functional assay for NK T cells was used to further validate the loss of these cells. NK T cells secrete IL-4 in response to stimulation. Anti-CD3 stimulation of thymocytes of Ets1-/- mice produced no detectable amount of IL-4, arguing that Ets1-/- mice lack functional NK T cells. CD1 expression, which is known to be required for the development of NK T cells, is not affected in Ets-/- mice. Thus, analysis of knockout mice demonstrates an absolute requirement for Ets1 function in the development of functional NK and NK T cells.

Both the Ets1 knockout and the Rag-2 complementation system indicate that Ets1 is essential for proper development and function of lymphoid derived cells. While both of these models presented elevated proportions of IgM+ B cells and T cell activation defects, the knockout mice demonstrated early post-natal lethality and lesser defects in T cell maturation. The Ets1 knockout mice further revealed an absence of NK and NK T cells. The Rag-2-/-/Ets1-/- chimeric mice were potentially capable of revealing the NK T cell defect but not the NK cell defect. Whether the failure of affinity maturation in the B cell population is due to an intrinsic B cell defect or is mediated by a defect in T helper cells also remains to be determined. In summary, Ets1 was found to be essential in all lymphoid lineages: for B cell maturation, T cell activation (and to a lesser extent T cell maturation), and for the development of functional NK and NK T cells (Table 1, Figure 1).

Spi-B: Spi-B is closely related to PU.1 through structural homology and by its ability to transactivate PU.1 target genes in vitro (Ray et al., 1992; Ray-Gallet et al., 1995). Despite these similarities, Spi-B-/- mice exhibit a mild phenotype when compared to those of PU.1-/- mice, suggesting there may be functional compensation by PU.1 and/or other Ets family members (Garrett-Sinha et al., 1999). Targeted disruption of the Spi-B gene was accomplished using a targeting vector in which a PGK Neo cassette replaced the Ets domain (Su et al., 1997). Absence of transcripts in a Northern blot of poly (A)+ RNA from Spi-B-/- splenocytes (probed with either full-length Spi-B cDNA or 3' UTR) indicated a true null allele. Spi-B-/- mice are viable and fertile, and generally exhibit normal hematopoietic development. Normal numbers of T cells, B cells, granulocytes, and macrophages were detected in both Spi-B heterozygotes and homozygotes, indicating that Spi-B is not essential for the commitment to myeloid and lymphoid lineages. In addition, B cells of Spi-B-/- mice secrete normal basal levels of all Ig isotypes, undergo V(D)J recombination, and proliferate normally in response to LPS.

Significantly, when challenged with the T cell dependent antigen, keyhole limpet hemocyanin (KLH), a 14-fold increase in IgM levels and significantly lower IgG2a levels were observed in the primary response of Spi-B-/- mice relative to wild type. Also, a 35-fold reduction in IgG2a and IgG2b levels and greater than threefold reduction in IgG1 levels were observed 8 days after secondary challenge. These results suggest a T cell mediated B cell defect in affinity maturation or proliferation of terminally differentiated B cells. Consistent with this phenotype, Spi-B-/- mice developed smaller and more transient germinal centers. This inability to maintain germinal centers after T cell dependent antigen challenge is associated with a sixfold elevation of TUNEL-positive B220+ cells, indicating a hyperapoptotic B cell defect. Cell cycle analysis of IgM cross-linked B cells revealed that Spi-B-/- B cells enter the cell cycle normally, but develop an increased proportion of apoptotic cells over 72 h. This survival defect was only observed in IgM cross-linked cells. In vitro stimulation of T cells with anti-CD3, PMA, ionomycin, or conA revealed no proliferative defects, suggesting that the B cell defect is not T cell-mediated. This view is further supported by the following experiments: IgM cross-linking of Spi-B-/- B cells with or without co-stimulation using anti-CD40, IL-4, or IL-6 resulted in increased proliferation at best 33% that of wild-type levels. This proliferative defect was more pronounced using reduced amounts of antibody. The authors suggest that the proliferation defect may be due to the loss of a critical Spi-B target that lies between surface B cell antigen receptor (BCR) and PKC in signaling pathways. Upon IgM cross-linking, Spi-B-/- B cells normally upregulate activation markers and multiple components (Btk, Blk, mb-1, B29 and tyrosine kinases: syk, fyn and lyn) of the BCR signaling pathway. In spite of this, Spi-B seems to be critical for B cell proliferation in response to BCR-mediated signals. The authors suggest that this BCR-mediated proliferation defect is reminiscent of immature and tolerant B cells that undergo Fas-induced apoptosis. This Spi-B-/- defect is also similar to the T cell defect observed in Ets1-/- mice, whereby the cells show an increased rate of apoptosis in response to TCR stimulation. Taken together, the results argue that Spi-B is important to some extent in the regulation of target gene(s) involved in the BCR signaling pathways and important for B cell proliferation. Adoptive transfer of Spi-B-/- B and T cells into wild type embryos could be used to elucidate intrinsic defects. Clearly, Spi-B target gene-independent (or PU.1 co-dependent, etc.) signalling pathways exist that permit proliferation of Spi-B-/- B cells to some extent.

Fli1: The analysis of recently developed targeted mutants has revealed an essential role for Fli1 in megakaryopoiesis (Hart et al., 2000; Spyropoulos et al., 2000). A defect in megakaryopoiesis is potentially responsible for the dramatic hemorrhagic phenotype associated with embryonic lethality in Fli1 knockout mice. It is instructive to note, however, that the first mice reported to carry a targeted disruption of the Fli1 gene were viable and fertile, with no profound phenotype (Melet et al., 1996). In this first reported gene disruption, a PGK Neo cassette replaced exon II of the Fli1 gene. This targeted disruption introduced a nonsense mutation in transcripts starting at the ATG in exon I. The predicted translational product of the targeted allele was expected to produce only the first 10 amino acids of the Fli1 protein. Preliminary Western blot analysis on total splenic protein from Fli1 homozygous mice failed to detect the wild-type 51 kDa Fli1 protein as would be predicted for a null allele. However, subsequent Western blot analysis revealed low level expression of a 43 kDa protein in Fli1 homozygous mice. Thus, a hypomorph rather than a null allele was produced. Further characterization of this mutant protein indicated that it was generated by alternative translational initiation and splicing around the PGK Neo cassette. Consistent with this, Northern blot analysis revealed a novel 3.4 kb alternative transcript in Fli1 homozygotes that was expressed at levels similar to wild type. The hypomorphic protein produced had an N-terminal substitution but retained all of the known functional domains of Fli1, including the C-terminal activation domain (CTA), pointed domain, Ets domain, and N-terminal transactivation domain. The fact that Fli1 homozygous mice were susceptible to Friend MuLV induced erythroleukemia, and that the Fli1 locus remained a site for proviral integration further emphasized that this was not a null allele.

These Fli1 homozygotes are viable and fertile, and show a 30-50% reduction in thymocyte number. Flow cytometric analysis of subpopulations of thymocytes revealed normal proportions in the mutants. The hypocellularity of the thymus did not appear to be due to an increase in negative selection, as levels of apoptosis were not elevated in homozygotes vs wild types when stimulated with anti-CD3 or other proapoptotic stimuli. To verify that the Fli1 mutant was responsible for the decreased cellularity of the thymus, homozygotes were crossed with transgenic mice that over-expressed Fli1 in the spleen and thymus (under the control of the H2K promoter; H2K-Fli1). Animals homozygous for the Fli1 mutation and expressing the Fli1 transgene produced normal thymocyte numbers. This result and the absence of phenotype in Fli1 heterozygotes argues against the hypomorph being a dominant mutant. In summary, the effect of Fli1 disruption on thymus cellularity and erythroleukemia should be taken with caution due to the generation of a hypomorphic Fli1 protein. Nevertheless, this hypomorph may become useful later in the molecular dissection of Fli1 function.

The Fli1 gene was subsequently targeted by the insertion of a loxP-flanked Neo (loxPneo) cassette into exon IX, between the Ets DNA binding domain and the CTA domain (Spyropoulos et al., 2000). The loxPneo cassette introduced a termination codon downstream of the Ets DNA binding domain which was predicted to generate a truncated Fli1 protein lacking the CTA domain (CTA-less Fli1). Northern blot analysis of total RNA from E10 and E11 homozygous embryos revealed the presence of two distinct transcripts at approximately 10% wild-type levels and Western blot analysis using anti-Fli1 antisera demonstrated even lower levels of CTA-less Fli1 protein. Therefore, this targeted disruption of Fli1 did not produce a true null but an extreme hypomorph. Arguments favoring the idea that this CTA-less hypomorph is a functional null include in vitro data demonstrating that removal of the CTA domain reduces transcriptional activation by 40-50%, and more importantly, that the phenotype of the CTA-less hypomorph is comparable to that of the recently described null allele.

Other targeted alleles of the Fli1 gene were subsequently generated to create true null alleles. In the first of these targeting vectors (FliresZ), an IRES-lacZ reporter cassette replaces the Ets DNA binding domain of Fli1 (Hart et al., 2000). Western blot analysis of total protein of E9.5 embryos shows no detectable Fli1 protein in FliresZ homozygotes. A second targeting vector was also generated and used to produce knockout mice, however no results utilizing this second knockout (PNTFliZ) were described. For this reason, we will limit our discussion to the CTA-less and the FliresZ Fli1 knockout mice.

Mice heterozygous for both the CTA-less and the FliresZ targeted Fli1 alleles are viable, fertile and overtly normal. In contrast to the original Fli1 mutant, homozygotes for these two alleles present with profound cerebrospinal hemorrhaging (E9.5 to E11.5) and ensuing embryonic lethality. The cerebrospinal phenotype occurred at 100% penetrance and was visualized by the pooling of fetal red blood cells in the neural tube and cephalic vessels. A notable distinction between the two mutants was that only intracranial hemorrhaging was reported in FliresZ homozygotes, while both intracranial and intraspinal hemorrhaging were observed in the CTA-less mutant. Fli1 expression, as indicated by lacZ staining in the FliresZ mutant, make it conceivable that hemorrhaging could occur within the spinal cord as well as intracranially. Hemorrhaging was also associated with the absence of red blood cells in embryonic tissues, most notably in yolk sacs and fetal livers (Spyropoulos et al., 2000). Yolk sac vasculature was found to be morphologically normal. Transverse sections of E11 homozygotes revealed a disruption in the columnar neuroepithelium and in the basement membrane at the site of intracranial and intraspinal hemorrhage, coincident with normal Fli1 expression.

Defects in hematopoiesis were also observed in Fli1 homozygous mutants (Spyropoulos et al., 2000). Homozygous embryos show small pale livers with diminished cellularity when compared to wild type embryos. Colony forming assays from homozygous yolk sacs indicate a selective loss of megakaryocytes. The absence of megakaryocyte-derived platelets could be in part responsible for the intracranial and intraspinal hemorrhage. May-Grünwald-Giemsa staining of homozygote livers revealed reductions in the numbers of pronormoblasts and basophilic normoblasts. Clonal culture of E11 homozygous mutant livers showed diminished progenitor numbers relative to heterozygotes and wild types. This reduction may be due to dysregulation of Fli1 target genes, or due to hypoxia and other factors associated with the hemorrhage. We have recently demonstrated that the CTA-less Fli1 knockout has defects in megakarypoiesis as well as erythropoiesis (Kawada et al., manuscript submitted). Cells from the aorta-gonad-mesonephros (AGM) region of E10 embryos and ES cells were cultured in order to determine if the impaired liver hematopoiesis was due to the loss of cells via hemorrhaging, defective migration or the result of an intrinsic defect. There was a striking reduction in the number of megakaryocytes in cultures of mutant AGM cells as compared to cultures of the AGM region from wild type mice. Furthermore, Fli1 mutant ES cells failed to produce megakaryocyte colonies and multilineage colonies containing megakaryocytes. More subtle differences were observed in the erythroid lineage. The percentages of pronormoblasts and basophilic normoblasts were significantly reduced in cultures of mutant AGM embryos which contained primarily polychromatophilic and orthochromatic normoblasts.

Chimeric mice were generated by aggregating both heterozygous and homozygous FliresZ ES cells with wild type CD-1 morula (Hart et al., 2000). Analysis of glucose 6-phosphate isomerase isoforms in these mice indicate that both heterozygous and homozygous targeted cells are able to contribute equally to all cells of embryos up through E10.5. This data contrasts somewhat with this group's previous data suggesting that hemorrhage occurs as early as E9.5. This suggests that non-cell autonomous effects may modulate the onset of hemorrhage. By E12.5, Fli1-/- ES cell contribution to chimeras was diminished in the cerebral meningies and lost by E14.5. Similarly, the contribution of Fli1-/- ES cells to vascular endothelium of major cerebral blood vessels resulted in little or no lacZ staining in E14.5 chimeras. Similar deficiencies of Fli1-/- ES cells were seen in fetal livers and megakaryocytes of adult mice. These results strongly suggest a cell autonomous requirement for Fli1 in either the maintenance or proliferation of these cells. While megakaryopoiesis was disrupted in the FliresZ homozygotes, no difference was seen in the relative numbers of erythroid, macrophage, granulocyte, and mixed progenitors observed in colony forming assays from E11.5 livers. This result contrasts with colony forming assays on E11 livers from the CTA-less homozygotes, which indicate disruption of hematopoiesis across these lineages. Abnormal megakaryocyte progenitors were observed in the FliresZ homozygotes. These abnormal cells were smaller in size with diminished nuclear/cytoplasmic ratios, reduced numbers of alpha-granules and disorganization of the platelet demarcation membranes. Semi-quantitative RT-PCR showed decreased expression of Tie-2 (vasculogenesis), GP-IX and c-mpl (megakaryopoiesis) genes (Hart et al., 2000; Kawada et al., manuscript submitted) in Fli1 homozygotes suggesting that these target genes are mediators of the Fli1 phenotype.

In summary, both knockouts display a hemorrhagic phenotype with slight differences in the site of hemorrhage. Also, both knockouts fail to produce mature megakaryocytes, with slight differences in the extent of progenitor abnormalities. Taken together, these results demonstrate that Fli1 is an essential regulator of megakaryopoiesis, hemostasis, and vascular integrity. These results also argue that both knockouts are comparable null alleles. From this perspective it will be interesting to study the phenotypes of the stabilized (cre-recombined) CTA-less Fli1 allele and the PNTFliZ gene fusion allele.

TEL1: The generation of mice carrying a targeted disruption of the TEL1 gene demonstrated essential roles for TEL1 in early embryonic angiogenesis and adult hematopoiesis (Wang et al., 1997). To date, the TEL1 knockout mouse displays one of the most profound embryonic lethal phenotypes of all Ets family member knockouts. Rescue of embryonic lethality in ES cell-embryo chimeras also revealed a requirement for TEL1 in adult hematopoiesis, specifically erythropoiesis, myelopoiesis, and lymphopoiesis (Wang et al., 1998). For sake of continuity, we will describe the generation of the TEL1 knockout and hematopoietic phenotypes in ES cell-embryo chimeras before describing the defective-angiogenesis phenotype in homozygous embryos.

The TEL1 gene targeted disruption consisted of a PGK-Neo cassette replacing sequences coding for the DNA binding domain (Wang et al., 1997). The absence of TEL1 transcripts in Northern blots on poly(A)+ RNA from homozygous embryos and ES cells suggested that this targeted disruption was a null allele. Subsequent RT-PCR analysis failed to further validate the generation of a null allele and merely reflected genomic structure, since one of the primers used for the analysis recognized sequences that had been deleted in the targeted allele. Still, the authors mention that no transcripts were observed in RT-PCR analysis using other sets of primers corresponding to exons upstream and downstream of the deleted DNA binding domain. These RT-PCR and Northern blot results argue that a null allele of the TEL1 gene had been generated.

TEL1-/- ES cells were injected into wild type and RAG-2-/- blastocysts to generate viable chimeric embryos and adults (Wang et al., 1998). Results of chimera analysis revealed that TEL1 was essential for adult hematopoiesis. TEL1 was shown to be dispensable for fetal liver hematopoiesis by colony forming assays on hematopoietic progenitors from fetal livers of TEL1-/-/wild type embryo chimeras. By 1 week of age, however, TEL1-/- cells (G418 resistant colonies) were not detectable in colony forming assays on bone marrow-derived cells, indicating that TEL1 is required for bone marrow myelopoiesis. Analysis of hemoglobin indicated that adult red cells were derived exclusively from the wild type cells of the chimera, indicating that TEL1 is also required for bone marrow erythropoiesis.

Flow cytometric analysis of TEL1-/-/RAG-2-/- adult chimeras shows a reduction in the frequency and absolute number of B220+ B cells. Progenitor assays for pre-B cells showed an absence of G418 resistant colonies by 1 week of age, suggesting that the deficiency in B cell number is due to a defect at the progenitor level in the bone marrow. A population of B220+/IgM+ cells were found in the spleen despite the drastic reduction in number of B220+ cells in the bone marrow, however, these splenic B cells most likely arose from fetal hematopoiesis and not from adult hematopoiesis. Regarding T cells, flow cytometric analysis of E18 thymic cells of TEL1-/-/Rag-2-/- chimeras showed a comparable number of CD4+ CD8+ T cells relative to the TEL1+/-/Rag-2-/- control. This result suggests that TEL1 is not required for migration of fetal hepatic progenitors into the thymus prior to birth. In the adult, however, the number of CD4+ CD8+ T cells in the thymus was dramatically reduced, and a sixfold reduction in mature peripheral T cells was also observed. The authors suggest that this thymic T cell defect is due to a failure to accumulate TEL1-/- CD25+ prothymocytes. While few in number, the peripheral T and B cells proliferate in response to stimulation, suggesting that TEL1 is dispensable for proliferation and differentiation of these cells. Thus, TEL1 appears to be essential at the earliest stages of adult lymphopoiesis. Collectively, the two chimeric systems studied suggest that TEL1 is required for normal bone marrow hematopoiesis. The absence of detectable TEL1-/- cells in adult hematopoietic tissues (as shown by Southern Blot analysis) is further evidence of the spatially and temporally restricted function of TEL1 in adult bone marrow hematopoiesis. Taken together these results indicate a requirement for TEL1 in either homing of hematopoietic stem cells to the bone marrow or in the creation of a suitable bone marrow microenvironment for hematopoiesis. In a recent review, the latter possibility has been suggested to be likely (Orkin et al., 1999). These experiments place TEL1 as a major player in erythropoiesis, myelopoiesis, and lymphopoiesis in the adult. In contrast to PU.1 which appears to intrinsically drive hematopoietic lineage differentiation, TEL1 may extrinsically drive adult hematopoiesis by controlling the bone marrow microenvironment.


TEL1 in angiogenesis

Adult chimera analysis involving TEL1-/- cells was necessary due to embryonic lethality observed in TEL1-/- embryos (Wang et al., 1997). Vascular defects were apparent in the yolk sacs of TEL1-/- embryos by E9.5, as noted by the absence of branching vitelline vessels (65% Type I mutant). Yolk sacs did develop ample blood islands with lumens, however, these structures failed to develop further to form larger lumens. Some of the embryos had normal vitelline vessels at E9.5 (35% Type II mutant), but these structures were unstable and by E10.5 they resembled Type I mutants. Normal vasculature within the embryo-proper was apparent at E9.5, but further analysis of embryonic vascular development was obscured by lethality initiating around E10.5. These results suggest that TEL1 is not essential for the initiation of yolk sac angiogenesis, but is essential for the subsequent formation and/or maintenance of more complex vascular networks.

TUNEL labeling of TEL1-/- E10 embryos revealed that the highest level of apoptosis correlated to regions in which TEL1 is normally most highly expressed. However, such conclusions about the embryo-proper at this time should be taken with caution considering the generalized embryonic abnormalities resulting from yolk sac defects. Nevertheless, this result suggests that TEL1 may be essential for the survival of mesenchymal cells and neural cells in the embryo-proper. Hematopoietic colony forming assays performed on E9.5 yolk sacs demonstrated normal precursors of erythroid, macrophage and mixed colonies. Similar numbers of primitive and definitive erythroid colonies were observed in wild type and TEL1-/- in vitro differentiated ES cells suggesting that TEL1 is not essential for the differentiation and proliferation of these embryonic lineages. In summary, the analysis of knockout embryos reveals that TEL1 is not essential for the initiation of embryonic angiogenesis within the yolk sac, but is essential for the subsequent development and/or maintenance of more complex vasculature. Analysis of knockout embryos and ES cell differentiation studies demonstrate that TEL1 is not essential for embryonic hematopoiesis. However, chimeric embryo analysis demonstrates that TEL1 is essential for all aspects of adult hematopoiesis.


Regulators of other cellular differentiation processes

Ets2: Ets2 targeted mutant mice present with the earliest embryonic lethal phenotypes a defect in extraembryonic trophectoderm, which is one of the first tissues in which Ets2 is highly expressed (Yamamoto et al., 1998). The Ets2 targeted allele consisted of a MC1Neo cassette replacing most of three exons (including the Ets DNA binding domain and nuclear localization sequences). Indication of a neomorphic allele appeared in a Northern blot analysis on poly (A)+ RNA from Ets2-/- fibroblasts which demonstrated a larger than wild type transcript capable of hybridizing with both neo and 5' Ets2 probes. This mutant transcript was believed to be the result of transcription through neo, extending into the 3' UTR. RNAse protection assay on RNA from Ets2-/- fibroblasts failed to extend these results since the Ets domain probe used constituted sequences deleted in the targeted allele. Immunoprecipitation analysis using anti-Ets2 antisera confirmed the generation of a neomorphic Ets2/neo fusion protein and subcellular fractionation suggested that this fusion protein was present only within the cytoplasmic compartment. The authors suggest that this Ets2/neo fusion protein could be active as an inhibitor, which would obscure interpretations correlating phenotype to loss of Ets2 function.

Lethality occurs in Ets2-/- embryos prior to E 8.5 due to defects in extraembryonic tissues. This phenotype was consistent regardless of genetic background (outbred Swiss B1 and inbred 129/Sv). Whole mount in situ hybridization of E6.0-E7.5 wild-type embryos indicated high level of Ets2 expression in the extraembryonic trophectoderm and extraembryonic endoderm at the time of lethality. While Ets2-/- blastocyst outgrowths were unaltered in their ability to generate trophoblasts in vitro, they did show distinct malformation of the ectoplacental cone (EPC). This defect consisted of reduced EPC size and the presence of an ectopic membrane covering. TUNEL staining of embryos demonstrated significant apoptosis at E7.5. Normal trophoblast migration was also not detected in Ets2-/- embryos. Membrane abnormalities involving a failure to form amnion and chorion were also observed in E7.5 Ets2-/- embryos. These defects resulted in hypocavitation of the embryo and the formation of only one body cavity. Ets2-/- embryos resorption occurred by E8.5.

Tetraploid embryos, generated by electrofusion of diploid two-cell stage embryos, form only extraembryonic trophectodermal lineage (which are themselves polyploid). Aggregation of wild type tetraploid embryos with Ets2-/- embryos was used to rescue the extraembryonic lethality without contributing to the Ets2-/- embryo-proper. Rescued Ets2-/- embryos were found to survive through adulthood and displayed only a mild phenotype which included curly whiskers, wavy hair, and a slightly rounded body shape. Chimeric embryos were also generated by injection of Ets2-/- ES cells into C57B1/6 blastocysts. These studies further demonstrated that Ets2 was not intrinsically required for the development of major organ systems (supplemental data; Yamamoto et al., 1998). Surprisingly, given the expression pattern of Ets2 and results from targeted disruptions of other Ets family members described, no defects were found in lymphoid or myeloid development or function.

Members of our group have generated a phenotypically comparable Ets2 mutant mouse carrying an insertional disruption of the Ets DNA binding domain (A Mjaatvedt, personal communication). A third targeted allele of the Ets2 gene has been generated in ES cells, however, there has been no report of mice generated from these cells (Henkel et al., 1996). This Ets2 targeted allele contains a Neo cassette in place of Ets DNA binding domain sequences. RT-PCR results offered no proof of a null allele since the 3' primer used corresponded to deleted Ets domain sequences. In vitro ES cell differentiation experiments with Ets2-/- cells suggested that wild type Ets2 is dispersible for macrophage commitment and maturation, but this conclusion should be taken with caution as the nature of the mutation has not been characterized. Future analysis of a true Ets2 knockout will be valuable in assessing its essential functions and identifying compensatory genes.

Er81: The phenotype of Er81-/- mice constitutes the most distinct of the Ets knockouts reported to date, indicative of a functional subclass of Ets family members involved in neurogenesis. Knockout mice corresponding to two separate Er81 targeted alleles were generated (Arber et al., 2000). In the Ex11 targeted allele, an IREStaulacZPGKNEO reporter cassette replaced exon XI, which encodes part of the Ets DNA binding domain. In the second targeted allele Ex2nlslacZ, a nuclear localization signal fused to lacZ was inserted inframe with the AUG in exon II. A 5' coding region probe for Er81 detected the presence of an Er81 transcript in Ex11 homozygous mice, proving that this targeted disruption of Er81 does produce a transcript. Antisera generated against the 11 C-terminal amino acids failed to detect the presence of full length Er81 in Ex11 homozygotes. Since these amino acids are deleted in the targeted allele, this result confirms genotypic data without adding proof of a null allele. Embryos homozygous for the Er81nlsLacZ allele were said to have no detectable protein at E13.5. However, by E15, 5% of the dorsal root ganglions (DRGs) normally expressing Er81 were said to express the mutant Er81 protein as defined by Er81 immunoreactivity. While the authors suggest that this is likely an inactive Er81 protein (lacking the activation domain), no evidence was offered in support of this model. The authors state that both targeted mutations produce the same phenotype. Thus, the phenotypes observed are either the result of an effective null or a truncated mutant protein. Mice homozygous for either disruption of Er81 exhibit postnatal lethality by 5 weeks of age that is probably related to neural defects. Overt phenotype of the homozygous mutant mice consisted of marked limb ataxia and abnormal flexor-extensor posturing of the limbs. This overt phenotype was not due to a defect in specification of proprioceptive neurons, since parvalbumin+ (PV+) cells were detected in equal abundance in wild type and homozygous mutant mice. Nevertheless, PV expression was diminished 5-10-fold in the homozygous mutants. The proprioceptive afferents show a mild conductivity defect, with the latency of the fast component of the compound sensory action potential being 1-2 ms longer in homozygous mutants vs. wild-type controls. Low threshold stimulation of dorsal roots of homozygous mutants revealed a tenfold reduction in amplitude of monosynaptic responses, rendering them incapable of generating action potentials. This defect was recapitulated by muscle nerve stimulation, but not by cutaneous nerve stimulation. Proprioceptive afferents of homozygous mutant mice connect only with interneurons and do not form direct functional connections with motor neurons, as evidenced by horseradish peroxidase staining. By E16, the ventral termination zone characteristic of group Ia afferents is almost absent, likely resulting in the reduced conductivity observed in homozygous mutant mice. Perturbations in the peripheral connections of homozygous mutant proprioceptive afferents were evidenced by the loss of Egr3 marker, representing a defect in the differentiation of intrafusal fibers. In addition, deficiency in group Ia fiber stimulation was seen in response to vibratory stimuli. Thus, it is suggestive that Er81 is essential for neural connectivity related to motor neurons, however, such a conclusion awaits more compelling proof that a null allele has been generated or the generation of a true null mutant.


Concluding remarks and future directions

Targeted disruptions for approximately one-fourth of the known Ets family members have been generated and described in this review (PU.1, Ets1, Spi-B, Fli1, TEL1, Ets2 and Er81; Table 1). The majority of these targeted disruptions clearly demonstrate important roles for Ets genes in the regulation of hematopoiesis (PU.1, Ets1, Spi-B, Fli1, and TEL1; Figure 1). At this point we can already place these Ets family members into a hierarchical framework within hematopoiesis on which other new targeted alleles can be superimposed (Figure 1). PU.1 has the greatest impact on the development of multiple hematopoietic lineages; Fli1 and Ets1 impact on the development of specialized compartments within the myeloid and lymphoid lineages, respectively; Spi-B impact is further restricted, apparently affecting only mature B cell function; TEL1 impacts on the development of multiple hematopoietic lineages, but acts later than PU.1 by extrinsic affects on the bone marrow microenvironment. Individual targeted disruptions suggest potentially distinct sub-classifications for Ets family members to also be important in angiogenesis (TEL), neurogenesis (Er81), and the earliest terminally differentiated lineage, the extraembryonic trophectoderm (Ets2). Following the current trend, many more Ets family members are predicted to demonstrate essential distinct functions in hematopoiesis and as yet undefined sub-classifications of functions in the regulation of cellular differentiation. The relative contribution of intrinsic and/or extrinsic functions in the role of the existing mutants and those yet to be generated remain to be clearly defined.

Targeted disruptions of the remaining three-fourths of the Ets family members will delineate their essential functions. A logical first step in the elucidation of gene function in vivo using knockout mouse technology is the generation of null alleles. Achieving such a goal, as shown in this review, is not trivial but defining the targeted mutation in hand as null, hypomorph, or neomorph is obligatory. No mutant allele will lack in value, but may have to wait for the analysis of a null allele before its value is realized. Ets2 is a prime example. It will be interesting to see how the current mutant allele of the Ets2 gene compares to the null allele. If the null allele also demonstrates no defects in lymphoid or myeloid development, then Ets2 will also be a good candidate for the generation of double knockouts with the existing PU.1, Ets1, and Fli1 knockouts. Ets2 may also be found to have redundant function with other Ets family members in neurogenesis.

Subsequent to the initial generation and characterization of null alleles is the elucidation of redundant or combinatorial Ets gene functions within specific lineages. Tissue specific knockouts, adoptive transfer, chimera studies and ES cell differentiation will allow discrimination between intrinsic and extrinsic functions (i.e. cell autonomy). Clearly, double mutants between Ets family members important in hematopoiesis (PU.1, Ets1, Spi-B, Fli1, and TEL1; Figure 1) is warranted. The added complexity of lethal phenotypes will require the generation of tissue-specific and/or inducible knockouts (specific to adult erythropoiesis, myelopoiesis, and/or lymphopoiesis). The cross talk between Ets family members can be dissected through the generation of tissue specific knockouts and transgenics. For example, would a Fli1 transgenic expressed in the PU.1-/- context be able to rescue some of the hematopoietic phenotypes? Conversely, would a tissue specific Fli1 knockout rescue a subset of phenotypes in TEL1-/- mice?

A second avenue of pursuit after the generation and characterization of null alleles is the generation of domain-specific and isoform-specific knockouts. One obvious example is the generation of a stable CTA-less Fli1 mutant allele. Such a mutant could easily be generated via cre-recombination of an existing Fli1 knockout. The resulting cre-recombined Fli1 allele would have a single residual 34 base loxP site (carrying termination codons in all three reading frames) that would be located between the Ets domain and the CTA domain. Isoform-specific gene disruption can be used to better understanding the distinct lymphopoietic functions of the p42 and p51 isoforms of Ets1, for example. Other subtle mutant alleles for each Ets family member can be similarly envisioned. A third avenue of pursuit after the generation and characterization of null alleles is the elucidation of differentially regulated target genes. Poweful new expression array and proteomic analysis techniques are now available that can be used to rapidly identify molecular differences in particular tissues between mutant and wild type individuals for the identification of differentially regulated target genes. Such approaches will take the study from the level of master transcriptional regulator to effector, defining what gene product(s) are directly involved in the phenotype observed. For example, what Spi-B activated target genes lying between surface BCR and PKC in signaling pathways are important in regulating B cell proliferation vs. apoptosis? Once identified, a target gene could be disrupted in a tissue specific fashion reflecting the expression pattern of its regulating transcription factor to determine whether it is responsible for the observed phenotype. For example, would disruption of Tie-2, GP-IX and c-mpl specifically within Fli1 expressed tissues recapitulate the Fli1 phenotype? Alternatively, considering a repressed target gene, would tissue-specific activation of that target gene recapitulate the Ets family member knockout? The possible avenues of pursuit are so great in number they can be considered virtually limitless, however, following the leads provided by the study of Ets family member genes will surely be one of the fastest routes to understanding hematopoiesis and other cellular developmental processes.



This article is dedicated to the memory of Dr Takis S Papas, a mentor, scientific colleague and friend from the beginning. We thank Dennis K Watson and Mark P Rubinstein for helpful suggestions. This work was supported in part by National Institutes of Health Grant P01CA78582.


Anderson KL, Perkin H, Surh CD, Venturini S, Maki RA and Torbett BE. (2000). J. Immunol. 164, 1855-1861. MEDLINE

Anderson KL, Smith KA, Conners K, McKercher SR, Maki RA and Torbett BE. (1998a). Blood 91, 3702-3710. MEDLINE

Anderson KL, Smith KA, Pio F, Torbett BE and Maki RA. (1998b). Blood 92, 1576-1585. MEDLINE

Arber S, Ladle DR, Lin JH, Frank E and Jessell TM. (2000). Cell 101, 485-498. MEDLINE

Barton K, Muthusamy N, Fischer C, Ting CN, Walunas TL, Lanier LL and Leiden JM. (1998). Immunity 9, 555-563. MEDLINE

Bassuk AG and Leiden JM. (1997). Adv. Immunol. 64, 65-104. MEDLINE

Bhat NK, Thompson CB, Lindsten T, June CH, Fujiwara S, Koizumi S, Fisher RJ and Papas TS. (1990). Proc. Natl. Acad. Sci. USA 87, 3723-3727. MEDLINE

Bories J-C, Willerford DM, Grevin D, Davidson L, Camus A, Martin P, Stehelin D and Alt FW. (1995). Nature 377, 635-638. MEDLINE

Chen J, Lansford R, Stewart V, Young F and Alt FW. (1993). Proc. Natl. Acad. Sci. USA 90, 4528-4532. MEDLINE

DeKoter RP, Walsh JC and Singh H. (1998). EMBO J. 17, 4456-4468. Article MEDLINE

Fisher RC, Lovelock JD and Scott EW. (1999). Blood 94, 1283-1290. MEDLINE

Fisher RC and Scott EW. (1998). Stem Cells 16, 25-37. MEDLINE

Garrett-Sinha LA, Su GH, Rao S, Kabak S, Hao Z, Clark MR and Simon MC. (1999). Immunity 10, 399-408. MEDLINE

Ghysdael J and Boureux A. (1997). Oncogenes as Transcriptional Regulators, Vol. 1: Progress in Gene Expression. Yaniv M and Ghysdael J. (eds). Birkhauser Verlag: Basel, pp. 29-88.

Graves BJ and Petersen JM. (1998). Adv. Cancer Res. 75, 1-55. MEDLINE

Guerriero A, Langmuir PB, Spain LM and Scott EW. (2000). Blood 95, 879-885. MEDLINE

Hart A, Melet F, Grossfeld P, Chien K, Jones C, Tunnacliffe A, Favier R and Bernstein A. (2000). Immunity 13, 167-177. MEDLINE

Henkel GW, McKercher SR, Leenen PJ and Maki RA. (1999). Blood 93, 2849-2858. MEDLINE

Henkel GW, McKercher SR, Yamamoto H, Anderson KL, Oshima RG and Maki RA. (1996). Blood 88, 2917-2926. MEDLINE

Iwama A, Zhang P, Darlington GJ, McKercher SR, Maki R and Tenen DG. (1998). Nucl. Acids Res. 26, 3034-3043.

Lichanska AM, Browne CM, Henkel GW, Murphy KM, Ostrowski MC, McKercher SR, Maki RA and Hume DA. (1999). Blood 94, 127-138. MEDLINE

McKercher SR, Torbett BE, Anderson KL, Henkel GW, Vestal DJ, Baribault H, Klemsz M, Feeney AJ, Wu GE, Paige CJ and Maki RA. (1996). EMBO J. 15, 5647-5658. MEDLINE

Melet F, Motro B, Rossi DJ, Zhang L and Bernstein A. (1996). Molec. Cell. Biol. 16, 2708-2718. MEDLINE

Muramatsu M, Kinoshita K, Fagarasan S, Yamada S, Shinkai Y and Honjo T. (2000). Cell 102, 553-563. MEDLINE

Muthusamy N, Barton K and Leiden JM. (1995). Nature 377, 639-642. MEDLINE

Orkin SH. (1998). Int. J. Dev. Biol. 42, 927-934. MEDLINE

Orkin SH, Porcher C, Fujiwara Y, Visvader J and Wang LC. (1999). Cancer Res. 59, 1784s-1787s; discussion 1788s. MEDLINE

Papas TS, Bhat NK, Spyropoulos DD, Mjaatvedt AE, Vournakis J, Seth A and Watson DK. (1997). Leukemia 11, 557-566. MEDLINE

Pio F, Kodandapani R, Ni CZ, Shepard W, Klemsz M, McKercher SR, Maki RA and Ely KR. (1996). J. Biol. Chem. 271, 23329-23337. MEDLINE

Ray D, Bosselut R, Ghysdael J, Mattei MG, Tavitian A and Moreau-Gachelin F. (1992). Mol. Cell. Biol. 12, 4297-4304. MEDLINE

Ray-Gallet D, Mao C, Tavitian A and Moreau-Gachelin F. (1995). Oncogene 11, 303-313. MEDLINE

Revy P, Muto T, Levy Y, Geissmann F, Plebani A, Sanal O, Catalan N, Forveille M, Dufourcq-Labelouse R, Gennery A, Tezcan I, Ersoy F, Kayserili H, Ugazio AG, Brousse N, Muramatsu M, Notarangelo LD, Kinoshita K, Honjo T, Fischer A and Durandy A. (2000). Cell 102, 565-575. MEDLINE

Scott EW, Fisher RC, Olson MC, Kehrli EW, Simon MC and Singh H. (1997). Immunity 6, 437-447. MEDLINE

Scott EW, Simon MC, Anastasi J and Singh H. (1994). Science 265, 1573-1577. MEDLINE

Simon MC. (1998). Semin Immunol 10, 111-118. MEDLINE

Spain LM, Guerriero A, Kunjibettu S and Scott EW. (1999). J. Immunol. 163, 2681-2687. MEDLINE

Spyropoulos DD, Pharr PN, Lavenburg KR, Jackers P, Papas TS, Ogawa M and Watson DK. (2000). Mol. Cell. Biol. 20, 5643-5652. MEDLINE

Su GH, Chen HM, Muthusamy N, Garrett-Sinha LA, Baunoch D, Tenen DG and Simon MC. (1997). EMBO J. 16, 7118-7129. MEDLINE

Tondravi MM, McKercher SR, Anderson K, Erdmann JM, Quiroz M, Maki R and Teitelbaum SL. (1997). Nature 386, 81-84. MEDLINE

Walunas TL, Wang B, Wang CR and Leiden JM. (2000). J. Immunol. 164, 2857-2860. MEDLINE

Wang LC, Kuo F, Fujiwara Y, Gilliland DG, Golub TR and Orkin SH. (1997). EMBO J. 16, 4374-4383. Article MEDLINE

Wang LC, Swat W, Fujiwara Y, Davidson L, Visvader J, Kuo F, Alt FW, Gilliland DG, Golub TR and Orkin SH. (1998). Genes Dev. 12, 2392-2402. MEDLINE

Watson DK, Li R, Sementchenko VI, Mavrothalassitis G and Seth A. (2001). Encyclopedia of Cancer. Bertino JR. (ed.). Academic Press: San Diego.,

Yamamoto H, Flannery ML, Kupriyanov S, Pearce J, McKercher SR, Henkel GW, Maki RA, Werb Z and Oshima RG. (1998). Genes Dev. 12, 1315-1326. MEDLINE


Figure 1 A schematic diagram depicting the roles of Ets family members in hematopoiesis and angiogenesis. Solid green line - transcription factor is indispensable for development. Dashed green line - transcription factor affects function and/or cell number. *Discrepancies in the literature


Table 1 Summary of Ets family knockout mice

18 December 2000, Volume 19, Number 55, Pages 6443-6454
Table of contents    Previous  Article  Next    [PDF]