Combinatorial regulation of endothelial gene expression by ets and forkhead transcription factors.

Vascular development begins when mesodermal cells differentiate into endothelial cells, which then form primitive vessels. It has been hypothesized that endothelial-specific gene expression may be regulated combinatorially, but the transcriptional mechanisms governing specificity in vascular gene expression remain incompletely understood. Here, we identify a 44 bp transcriptional enhancer that is sufficient to direct expression specifically and exclusively to the developing vascular endothelium. This enhancer is regulated by a composite cis-acting element, the FOX:ETS motif, which is bound and synergistically activated by Forkhead and Ets transcription factors. We demonstrate that coexpression of the Forkhead protein FoxC2 and the Ets protein Etv2 induces ectopic expression of vascular genes in Xenopus embryos, and that combinatorial knockdown of the orthologous genes in zebrafish embryos disrupts vascular development. Finally, we show that FOX:ETS motifs are present in many known endothelial-specific enhancers and that this motif is an efficient predictor of endothelial enhancers in the human genome.

Cooperative activation of cardiac transcription through myocardin bridging of paired MEF2 sites.

Enhancers frequently contain multiple binding sites for the same transcription factor. These homotypic binding sites often exhibit synergy, whereby the transcriptional output from two or more binding sites is greater than the sum of the contributions of the individual binding sites alone. Although this phenomenon is frequently observed, the mechanistic basis for homotypic binding site synergy is poorly understood. Here, we identify a bona fide cardiac-specific Prkaa2 enhancer that is synergistically activated by homotypic MEF2 binding sites. We show that two MEF2 sites in the enhancer function cooperatively due to bridging of the MEF2C-bound sites by the SAP domain-containing co-activator protein myocardin, and we show that paired sites buffer the enhancer from integration site-dependent effects on transcription in vivo Paired MEF2 sites are prevalent in cardiac enhancers, suggesting that this might be a common mechanism underlying synergy in the control of cardiac gene expression in vivo.

Endothelin signaling activates Mef2c expression in the neural crest through a MEF2C-dependent positive-feedback transcriptional pathway.

Endothelin signaling is essential for neural crest development, and dysregulated Endothelin signaling is associated with several neural crest-related disorders, including Waardenburg and other syndromes. However, despite the crucial roles of this pathway in neural crest development and disease, the transcriptional effectors directly activated by Endothelin signaling during neural crest development remain incompletely elucidated. Here, we establish that the MADS box transcription factor MEF2C is an immediate downstream transcriptional target and effector of Endothelin signaling in the neural crest. We show that Endothelin signaling activates Mef2c expression in the neural crest through a conserved enhancer in the Mef2c locus and that CRISPR-mediated deletion of this Mef2c neural crest enhancer from the mouse genome abolishes Endothelin induction of Mef2c expression. Moreover, we demonstrate that Endothelin signaling activates neural crest expression of Mef2c by de-repressing MEF2C activity through a Calmodulin-CamKII-histone deacetylase signaling cascade. Thus, these findings identify a MEF2C-dependent, positive-feedback mechanism for Endothelin induction and establish MEF2C as an immediate transcriptional effector and target of Endothelin signaling in the neural crest.

Mammalian Fused is essential for sperm head shaping and periaxonemal structure formation during spermatogenesis.

During mammalian spermatogenesis, the diploid spermatogonia mature into haploid spermatozoa through a highly controlled process of mitosis, meiosis and post-meiotic morphological remodeling (spermiogenesis). Despite important progress made in this area, the molecular mechanisms underpinning this transformation are poorly understood. Our analysis of the expression and function of the putative serine-threonine kinase Fused (Fu) provides critical insight into key steps in spermatogenesis. In this report, we demonstrate that conditional inactivation of Fu in male germ cells results in infertility due to diminished sperm count, abnormal head shaping, decapitation and motility defects of the sperm. Interestingly, mutant flagellar axonemes are intact but exhibit altered periaxonemal structures that affect motility. These data suggest that Fu plays a central role in shaping the sperm head and controlling the organization of the periaxonemal structures in the flagellum. We show that Fu localizes to multiple tubulin-containing or microtubule-organizing structures, including the manchette and the acrosome-acroplaxome complex that are involved in spermatid head shaping. In addition, Fu interacts with the outer dense fiber protein Odf1, a major component of the periaxonemal structures in the sperm flagellum, and Kif27, which is detected in the manchette. We propose that disrupted Fu function in these structures underlies the head and flagellar defects in Fu-deficient sperm. Since a majority of human male infertility syndromes stem from reduced sperm motility and structural defects, uncovering Fu׳s role in spermiogenesis provides new insight into the causes of sterility and the biology of reproduction.

An arterial-specific enhancer of the human endothelin converting enzyme 1 (ECE1) gene is synergistically activated by Sox17, FoxC2, and Etv2.

Endothelin-converting enzyme-1 (Ece-1), a crucial component of the Endothelin signaling pathway, is required for embryonic development and is an important regulator of vascular tone, yet the transcriptional regulation of the ECE1 gene has remained largely unknown. Here, we define the activity and regulation of an enhancer from the human ECE1 locus in vivo. The enhancer identified here becomes active in endothelial progenitor cells shortly after their initial specification and is dependent on a conserved FOX:ETS motif, a composite binding site for Forkhead transcription factors and the Ets transcription factor Etv2, for activity in vivo. The ECE1 FOX:ETS motif is bound and cooperatively activated by FoxC2 and Etv2, but unlike other described FOX:ETS-dependent enhancers, ECE1 enhancer activity becomes restricted to arterial endothelium and endocardium by embryonic day 9.5 in transgenic mouse embryos. The ECE1 endothelial enhancer also contains an evolutionarily-conserved, consensus SOX binding site, which is required for activity in transgenic mouse embryos. Importantly, the ECE1 SOX site is bound and activated by Sox17, a transcription factor involved in endothelial cell differentiation and an important regulator of arterial identity. Moreover, the ECE1 enhancer is cooperatively activated by the combinatorial action of FoxC2, Etv2, and Sox17. Although Sox17 is required for arterial identity, few direct transcriptional targets have been identified in endothelial cells. Thus, this work has important implications for our understanding of endothelial specification and arterial subspecification.

The MADS box transcription factor MEF2C regulates melanocyte development and is a direct transcriptional target and partner of SOX10.

Waardenburg syndromes are characterized by pigmentation and autosensory hearing defects, and mutations in genes encoding transcription factors that control neural crest specification and differentiation are often associated with Waardenburg and related disorders. For example, mutations in SOX10 result in a severe form of Waardenburg syndrome, Type IV, also known as Waardenburg-Hirschsprung disease, characterized by pigmentation and other neural crest defects, including defective innervation of the gut. SOX10 controls neural crest development through interactions with other transcription factors. The MADS box transcription factor MEF2C is an important regulator of brain, skeleton, lymphocyte and cardiovascular development and is required in the neural crest for craniofacial development. Here, we establish a novel role for MEF2C in melanocyte development. Inactivation of Mef2c in the neural crest of mice results in reduced expression of melanocyte genes during development and a significant loss of pigmentation at birth due to defective differentiation and reduced abundance of melanocytes. We identify a transcriptional enhancer of Mef2c that directs expression to the neural crest and its derivatives, including melanocytes, in transgenic mouse embryos. This novel Mef2c neural crest enhancer contains three functional SOX binding sites and a single essential MEF2 site. We demonstrate that Mef2c is a direct transcriptional target of SOX10 and MEF2 via this evolutionarily conserved enhancer. Furthermore, we show that SOX10 and MEF2C physically interact and function cooperatively to activate the Mef2c gene in a feed-forward transcriptional circuit, suggesting that MEF2C might serve as a potentiator of the transcriptional pathways affected in Waardenburg syndromes.

ETS-dependent regulation of a distal Gata4 cardiac enhancer.

The developing heart contains an inner tube of specialized endothelium known as endocardium, which performs multiple essential functions. In spite of the essential role of the endocardium in heart development and function, the transcriptional pathways that regulate its development remain largely undefined. GATA4 is a zinc finger transcription factor that is expressed in multiple cardiovascular lineages and is required for endocardial cushion development and embryonic viability, but the transcriptional pathways upstream of Gata4 in the endocardium and its derivatives in the endocardial cushions are unknown. Here, we describe a distal enhancer from the mouse Gata4 gene that is briefly active in multiple cardiac lineages early in cardiac development but restricts to the endocardium where it remains active through cardiogenesis. The activity of this Gata4 cardiac enhancer in transgenic embryos and in cultured aortic endothelial cells is dependent on four ETS sites. To identify which ETS transcription factors might be involved in Gata4 regulation via the ETS sites in the enhancer, we determined the expression profile of 24 distinct ETS factors in embryonic mouse hearts. Among multiple ETS transcripts present, ETS1, FLI1, ETV1, ETV5, ERG, and ETV6 were the most abundant in the early embryonic heart. We found that ETS1, FLI1, and ERG were strongly expressed in the heart at embryonic day 8.5 and that ETS1 and ERG bound to the endogenous Gata4 enhancer in cultured endothelial cells. Thus, these studies define the ETS expression profile in the early embryonic heart and identify an ETS-dependent enhancer from the Gata4 locus.

An injury-responsive mmp14b enhancer is required for heart regeneration.

Mammals have limited capacity for heart regeneration, whereas zebrafish have extraordinary regeneration abilities. During zebrafish heart regeneration, endothelial cells promote cardiomyocyte cell cycle reentry and myocardial repair, but the mechanisms responsible for promoting an injury microenvironment conducive to regeneration remain incompletely defined. Here, we identify the matrix metalloproteinase Mmp14b as an essential regulator of heart regeneration. We identify a TEAD-dependent mmp14b endothelial enhancer induced by heart injury in zebrafish and mice, and we show that the enhancer is required for regeneration, supporting a role for Hippo signaling upstream of mmp14b. Last, we show that MMP-14 function in mice is important for the accumulation of Agrin, an essential regulator of neonatal mouse heart regeneration. These findings reveal mechanisms for extracellular matrix remodeling that promote heart regeneration.

Differential Etv2 threshold requirement for endothelial and erythropoietic development.

Endothelial and erythropoietic lineages arise from a common developmental progenitor. Etv2 is a master transcriptional regulator required for the development of both lineages. However, the mechanisms through which Etv2 initiates the gene-regulatory networks (GRNs) for endothelial and erythropoietic specification and how the two GRNs diverge downstream of Etv2 remain incompletely understood. Here, by analyzing a hypomorphic Etv2 mutant, we demonstrate different threshold requirements for initiation of the downstream GRNs for endothelial and erythropoietic development. We show that Etv2 functions directly in a coherent feedforward transcriptional network for vascular endothelial development, and a low level of Etv2 expression is sufficient to induce and sustain the endothelial GRN. In contrast, Etv2 induces the erythropoietic GRN indirectly via activation of Tal1, which requires a significantly higher threshold of Etv2 to initiate and sustain erythropoietic development. These results provide important mechanistic insight into the divergence of the endothelial and erythropoietic lineages.

Direct transcriptional regulation of Gata4 during early endoderm specification is controlled by FoxA2 binding to an intronic enhancer.

The embryonic endoderm is a multipotent progenitor cell population that gives rise to the epithelia of the digestive and respiratory tracts, the liver and the pancreas. Among the transcription factors that have been shown to be important for endoderm development and gut morphogenesis is GATA4. Despite the important role of GATA4 in endoderm development, its transcriptional regulation is not well understood. In this study, we identified an intronic enhancer from the mouse Gata4 gene that directs expression to the definitive endoderm in the early embryo. The activity of this enhancer is initially broad in all endodermal progenitors, as demonstrated by fate mapping analysis using the Cre/loxP system, but becomes restricted to the dorsal foregut and midgut, and associated organs such as dorsal pancreas and stomach. The function of the intronic Gata4 enhancer is dependent upon a conserved Forkhead transcription factor-binding site, which is bound by recombinant FoxA2 in vitro. These studies identify Gata4 as a direct transcriptional target of FoxA2 in the hierarchy of the transcriptional regulatory network that controls the development of the definitive endoderm.

Isl1 is a direct transcriptional target of Forkhead transcription factors in second-heart-field-derived mesoderm.

The cells of the second heart field (SHF) contribute to the outflow tract and right ventricle, as well as to parts of the left ventricle and atria. Isl1, a member of the LIM-homeodomain transcription factor family, is expressed early in this cardiac progenitor population and functions near the top of a transcriptional pathway essential for heart development. Isl1 is required for the survival and migration of SHF-derived cells into the early developing heart at the inflow and outflow poles. Despite this important role for Isl1 in early heart formation, the transcriptional regulation of Isl1 has remained largely undefined. Therefore, to identify transcription factors that regulate Isl1 expression in vivo, we screened the conserved noncoding sequences from the mouse Isl1 locus for enhancer activity in transgenic mouse embryos. Here, we report the identification of an enhancer from the mouse Isl1 gene that is sufficient to direct expression to the SHF and its derivatives. The Isl1 SHF enhancer contains three consensus Forkhead transcription factor binding sites that are efficiently and specifically bound by Forkhead transcription factors. Importantly, the activity of the enhancer is dependent on these three Forkhead binding sites in transgenic mouse embryos. Thus, these studies demonstrate that Isl1 is a direct transcriptional target of Forkhead transcription factors in the SHF and establish a transcriptional pathway upstream of Isl1 in the SHF.

An endoderm-specific transcriptional enhancer from the mouse Gata4 gene requires GATA and homeodomain protein-binding sites for function in vivo.

Several transcription factors function in the specification and differentiation of the endoderm, including the zinc finger transcription factor GATA4. Despite its essential role in endoderm development, the transcriptional control of the Gata4 gene in the developing endoderm and its derivatives remains incompletely understood. Here, we identify a distal enhancer from the Gata4 gene, which directs expression exclusively to the visceral and definitive endoderm of transgenic mouse embryos. The activity of this enhancer is initially broad within the definitive endoderm but later restricts to developing endoderm-derived tissues, including pancreas, glandular stomach, and duodenum. The activity of this enhancer in vivo is dependent on evolutionarily-conserved HOX- and GATA-binding sites, which are bound by PDX-1 and GATA4, respectively. These studies establish Gata4 as a direct transcriptional target of homeodomain and GATA transcription factors in the endoderm and support a model in which GATA4 functions in the transcriptional network for pancreas formation.

mef2c is activated directly by myogenic basic helix-loop-helix proteins during skeletal muscle development in vivo.

Skeletal muscle development requires the coordinated expression of numerous transcription factors to control the specification of mesodermal progenitor cells to a muscle fate and the differentiation of those committed myoblasts into functional, contractile muscle. Two families of transcription factors play key roles in these processes. The myogenic basic helix-loop-helix (bHLH) proteins, MyoD and Myf5, are required for myoblast specification, while two members of the same family, myogenin and MRF4, play key roles in myoblast differentiation in vivo. All four members of the myogenic bHLH family are sufficient to dominantly induce myogenesis when introduced into a variety of non-muscle cells in culture, however this function requires the activity of a second family of transcriptional regulators, the myocyte enhancer factor 2 (MEF2) family. MEF2 factors are essential for muscle differentiation, and previous studies have shown that MyoD and MEF2 family members function combinatorially to activate transcription and myogenesis. Consistent with these observations, the majority of skeletal muscle genes require both MyoD and MEF2 family members to activate their transcription. A possible exception to this combinatorial model for activation is suggested by the observation that myogenic bHLH factors may be able to independently activate the expression of MEF2. This raises the question as to how mef2 gene transcription is induced by MyoD factors without cooperative activation by MEF2. During skeletal muscle development, mef2c is the first member of the MEF2 family to be expressed. In this study, we have investigated the regulation of a skeletal muscle-specific enhancer from the mouse mef2c gene using a transgenic approach. We show that mef2c is a direct transcriptional target of the MyoD family in vivo via an essential E box in the skeletal muscle enhancer of mef2c, and we show that mef2c is not a direct target for autoregulation by MEF2.

Alveolar type II cells possess the capability of initiating lung tumor development.

Identifying cells of tumor origin is a fundamental question in tumor biology. Answers to this central question will not only advance our understanding of tumor initiation and progression but also have important therapeutic implications. In this study, we aimed to uncover the cells of origin of lung adenocarcinoma, a major subtype of non-small cell lung cancer. To this end, we developed new mouse models of lung adenocarcinoma that enabled selective manipulation of gene activity in surfactant associated protein C (SPC)-expressing cells, including alveolar type II cells and bronchioalveolar stem cells (BASCs) that reside at the bronchioalveolar duct junction (BADJ). Our findings showed that activation of oncogenic Kras alone or in combination with the removal of the tumor suppressor p53 in SPC⁺ cells resulted in development of alveolar tumors. Similarly, sustained EGF signaling in SPC⁺ cells led to alveolar tumors. By contrast, BASCs failed to proliferate or produce tumors under these conditions. Importantly, in a mouse strain in which Kras/p53 activity was selectively altered in type II cells but not BASCs, alveolar tumors developed while BADJs retained normal architecture. These results confirm and extend previous findings and support a model in which lung adenocarcinoma can initiate in alveolar type II cells. Our results establish the foundation for elucidating the molecular mechanisms by which lung cancer initiates and progresses in a specific lung cell type.

Local protease signaling contributes to neural tube closure in the mouse embryo.

We report an unexpected role for protease signaling in neural tube closure and the formation of the central nervous system. Mouse embryos lacking protease-activated receptors 1 and 2 showed defective hindbrain and posterior neuropore closure and developed exencephaly and spina bifida, important human congenital anomalies. Par1 and Par2 were expressed in surface ectoderm, and Par2 was expressed selectively along the line of closure. Ablation of G(i/z) and Rac1 function in these Par2-expressing cells disrupted neural tube closure, further implicating G protein-coupled receptors and identifying a likely effector pathway. Cluster analysis of protease and Par2 expression patterns revealed a group of membrane-tethered proteases often coexpressed with Par2. Among these, matriptase activated Par2 with picomolar potency, and hepsin and prostasin activated matriptase. Together, our results suggest a role for protease-activated receptor signaling in neural tube closure and identify a local protease network that may trigger Par2 signaling and monitor and regulate epithelial integrity in this context.

Gata4 expression in lateral mesoderm is downstream of BMP4 and is activated directly by Forkhead and GATA transcription factors through a distal enhancer element.

The GATA family of zinc-finger transcription factors plays key roles in the specification and differentiation of multiple cell types during development. GATA4 is an early regulator of gene expression during the development of endoderm and mesoderm, and genetic studies in mice have demonstrated that GATA4 is required for embryonic development. Despite the importance of GATA4 in tissue specification and differentiation, the mechanisms by which Gata4 expression is activated and the transcription factor pathways upstream of GATA4 remain largely undefined. To identify transcriptional regulators of Gata4 in the mouse, we screened conserved noncoding sequences from the mouse Gata4 gene for enhancer activity in transgenic embryos. Here, we define the regulation of a distal enhancer element from Gata4 that is sufficient to direct expression throughout the lateral mesoderm, beginning at 7.5 days of mouse embryonic development. The activity of this enhancer is initially broad but eventually becomes restricted to the mesenchyme surrounding the liver. We demonstrate that the function of this enhancer in transgenic embryos is dependent upon highly conserved Forkhead and GATA transcription factor binding sites, which are bound by FOXF1 and GATA4, respectively. Furthermore, the activity of the Gata4 lateral mesoderm enhancer is attenuated by the BMP antagonist Noggin, and the enhancer is not activated in Bmp4-null embryos. Thus, these studies establish that Gata4 is a direct transcriptional target of Forkhead and GATA transcription factors in the lateral mesoderm, and demonstrate that Gata4 lateral mesoderm enhancer activation requires BMP4, supporting a model in which GATA4 serves as a downstream effector of BMP signaling in the lateral mesoderm.

Essential role for Galpha13 in endothelial cells during embryonic development.

Toward identifying the roles of protease-activated receptor-1 (PAR1) and other G protein-coupled receptors important for vascular development, we investigated the role of Galpha13 in endothelial cells in the mouse embryo. LacZ inserted into Galpha13 exon 1 was highly expressed in endothelial cells at midgestation. Endothelial-specific Galpha13 knockout embryos died at embryonic days 9.5-11.5 and resembled the PAR1 knockout. Restoration of Galpha13 expression in endothelial cells by use of a Tie2 promoter-driven Galpha13 transgene rescued development of endothelial-specific Galpha13 knockout embryos as well the embryonic day 9.5 vascular phenotype in Galpha13 conventional knockouts; transgene-positive Galpha13-/- embryos developed for several days beyond their transgene-negative Galpha13-/- littermates and then manifested a previously uncharacterized phenotype that included intracranial bleeding and exencephaly. Taken together, our results suggest a critical role for Galpha13 in endothelial cells during vascular development, place Galpha13 as a candidate mediator of PAR1 signaling in this process, and reveal roles for Galpha13 in other cell types in the mammalian embryo.

Palmitoylation is required for the production of a soluble multimeric Hedgehog protein complex and long-range signaling in vertebrates.

Hedgehog (Hh) signaling plays a major role in multiple aspects of embryonic development. A key issue in Hh signaling is to elucidate the molecular mechanism by which a Hh protein morphogen gradient is formed despite its membrane association. In this study, we used a combination of genetic, cellular, and biochemical approaches to address the role of lipid modifications in long-range vertebrate Hh signaling. Our molecular analysis of knockout mice deficient in Skn, the murine homolog of the Drosophila ski gene, which catalyzes Hh palmitoylation, and gene-targeted mice producing a nonpalmitoylated form of Shh indicates that Hh palmitoylation is essential for its activity as well as the generation of a protein gradient in the developing embryos. Furthermore, our biochemical data show that Hh lipid modifications are required for producing a soluble multimeric protein complex, which constitutes the major active component for Hh signaling. These results suggest that soluble Hh multimeric complex travels in the morphogenetic field to activate Hh signaling in distant Hh-responsive cells.

Mef2c is a direct transcriptional target of ISL1 and GATA factors in the anterior heart field during mouse embryonic development.

The vertebrate heart forms initially as a linear tube derived from a primary heart field in the lateral mesoderm. Recent studies in mouse and chick have demonstrated that the outflow tract and right ventricle originate from a separate source of mesoderm that is anterior to the primary heart field. The discovery of this anterior, or secondary, heart field has led to a greater understanding of the morphogenetic events involved in heart formation; however, many of the underlying molecular events controlling these processes remain to be determined. The MADS domain transcription factor MEF2C is required for proper formation of the cardiac outflow tract and right ventricle, suggesting a key role in anterior heart field development. Therefore, as a first step toward identifying the transcriptional pathways upstream of MEF2C, we introduced a lacZ reporter gene into a bacterial artificial chromosome (BAC) encompassing the murine Mef2c locus and used this recombinant to generate transgenic mice. This BAC transgene was sufficient to recapitulate endogenous Mef2c expression, and comparative sequence analyses revealed multiple regions of significant conservation in the noncoding regions of the BAC. We show that one of these conserved noncoding regions represents a transcriptional enhancer that is sufficient to direct expression of lacZ exclusively to the anterior heart field throughout embryonic development. This conserved enhancer contains two consensus GATA binding sites that are efficiently bound by the zinc finger transcription factor GATA4 and are completely required for enhancer function in vivo. This enhancer also contains two perfect consensus sites for the LIM-homeodomain protein ISL1. We show that these elements are specifically bound by ISL1 and are essential for enhancer function in transgenic embryos. Thus, these findings establish Mef2c as the first direct transcriptional target of ISL1 in the anterior heart field and support a model in which GATA factors and ISL1 serve as the earliest transcriptional regulators controlling outflow tract and right ventricle development.

Mef2c is activated directly by Ets transcription factors through an evolutionarily conserved endothelial cell-specific enhancer.

Members of the Myocyte Enhancer Factor 2 (MEF2) family of transcription factors play key roles in the development and differentiation of numerous cell types during mammalian development, including the vascular endothelium. Mef2c is expressed very early in the development of the endothelium, and genetic studies in mice have demonstrated that mef2c is required for vascular development. However, the transcriptional pathways involving MEF2C during endothelial cell development have not been defined. As a first step towards identifying the transcriptional factors upstream of MEF2C in the vascular endothelium, we screened for transcriptional enhancers from the mouse mef2c gene that regulate vascular expression in vivo. In this study, we identified a transcriptional enhancer from the mouse mef2c gene sufficient to direct expression to the vascular endothelium in transgenic embryos. This enhancer is active in endothelial cells within the developing vascular system from very early stages in vasculogenesis, and the enhancer remains robustly active in the vascular endothelium during embryogenesis and in adulthood. This mef2c endothelial cell enhancer contains four perfect consensus Ets transcription factor binding sites that are efficiently bound by Ets-1 protein in vitro and are required for enhancer function in transgenic embryos. Thus, these studies identify mef2c as a direct transcriptional target of Ets factors via an evolutionarily conserved transcriptional enhancer and establish a direct link between these two early regulators of vascular gene expression during endothelial cell development in vivo.