Endothelial signaling during development


	Journal home

	Advance online publication

	Current issue

	Archive

	Press releases



	Supplements

	Focuses

Guide to authors

Reprints and permissions

About this site

For librarians


	NPG Resources

	Nature

	Nature Reviews

	Nature Immunology

	Nature Cell Biology

	Nature Genetics

	news@nature.com

	Nature Conferences

	Dissect Medicine

Molecular Cell Biology

Neuroscience

Pharmacology

Physics

Browse all publications

Review

Endothelial signaling during development

Ondine Cleaver & Douglas A Melton

Howard Hughes Medical Institute and Department of Cellular and Molecular Biology, Harvard University, 7 Divinity Ave, Cambridge, Massachusetts 02138, USA.

Correspondence should be addressed to Ondine Cleaver ocleaver@fas.harvard.edu

Blood vessels perfuse all tissues in the body and mediate vital metabolic exchange between tissues and blood. Increasing evidence, however, points to a direct role for paracrine signaling between blood vessel cells and surrounding target organ cells, during embryonic development and cell differentiation. Understanding the nature of this signaling and its heterogeneity, both in the embryo and in adult tissues, may not only provide insights into mechanisms for normal developmental cell fate decisions, but could also lead to novel targeted therapeutic approaches for a variety of diseases such as heart disease, diabetes or cancer.

The 'active' endothelium
The endothelium of blood vessels has traditionally been regarded as rather inert 'plumbing', which enables metabolic exchange between blood and tissues by virtue of its permeability and proximity to all cells. In recent decades, however, there has been a reassessment of the role of the endothelium, suggesting that it is "more than a sheet of nucleated cellophane"¹. Endothelial signals are now implicated in the regulation of many processes, and endothelial cells (ECs) are known to communicate directly with adjacent cells.

There is also emerging evidence that the endothelium provides developmental signals to organs and differentiating cells. Although it is well known that tissues signal to ECs, providing cues for the patterning of vessels, recent studies have identified reciprocal signals from ECs back to surrounding tissues. The possibility of EC signaling is not entirely surprising, given that ECs express numerous secreted and cell-surface signaling molecules and are intimately associated with developing tissues, growing and changing with them in a coordinated manner. Immediate physical contact between endothelial and tissue cells occurs both in the early embryo, before the recruitment of cells that surround and sheath larger vessels (vascular wall or mural cells), and later in capillary beds, where individual mural cells (pericytes) sparsely cover the endothelium (Fig. 1). It is therefore possible that mutual signaling between endothelium and tissues during development forms the molecular basis for interdependent physical and physiological relationships that last into adulthood.

Figure 1. Morphology of mural cells associated with large and small blood vessels.

	Heterogeneity in vessel wall composition is evident between vessels of different sizes, and between arterial and venous vessels. Larger vessels (top) have multiple layers of cellular and extracellular materials, whereas capillaries (bottom) have a loose covering of pericytes. Large vessel vascular walls are composed of the tunica intima, which consists of the endothelium, the basement membrane and an internal elastic layer; the tunica media, which consists of a thick layer of smooth muscle with reticular fibers (collagen III), elastin and proteoglycans; and the tunica adventitia, which consists of connective tissue with both elastic and collagenous fibers. Veins have valves for preventing backflow of blood, whereas arteries have strong elastic vessel walls for withstanding the high blood pressures downstream of the heart. Full Figure and legend (104K)

This review will examine the potential role of endothelial paracrine signals during organogenesis and cell differentiation, with consideration given to the ramifications of endothelial diversity and reciprocal signals from tissues to ECs.

Morphological and functional EC heterogeneity
Upon cursory examination, one might presume that the endothelium consists of a rather homogeneous population of cells that have a shared propensity for assembling into tubes. There is considerable evidence, however, that although ECs display many common features, they also exhibit remarkable heterogeneity². In fact, endothelial diversity is reflected by vessel size-specific, organ-specific and even age-specific differences. This diversity of cellular properties makes the endothelium uniquely adapted to communicate regionally with specific underlying organ tissues.

Endothelial heterogeneity is most evident at the morphological level. EC vessel phenotype is classified as continuous, fenestrated or discontinuous³ (Fig. 2). Other distinguishing criteria include cell size, shape, orientation with respect to the direction of blood flow, complexity of interendothelial adhesions, presence or absence of diaphragms on fenestrations and of plasmalemmal bodies, and composition of the recruited vessel wall⁴. For instance, ECs lining microvessels are generally flattened, elongated and often fenestrated, whereas those lining large arteries are polygonal, nonfenestrated and thicker⁴. Even among capillary beds there are organ-specific differences. Capillaries in skeletal muscle, heart, lung and brain have continuous endothelium, whereas capillaries in endocrine and exocrine glands, choroids plexus and intestinal villi have fenestrations⁵. In fact, "there are almost as many varieties of capillaries as there are organs and tissues"⁶.

Figure 2. Capillary wall morphology.

	Capillaries in different tissues exhibit different cellular morphology. (a) Continuous capillaries have no openings in their walls and are lined continuously with the endothelial cell body. (b) Fenestrated capillaries have small openings, called fenestrae, of about 80−100 nm in diameter. Fenestrae are covered by a small, nonmembranous, permeable diaphragm, and allow the rapid passage of macromolecules. The basement membrane of endothelial cells is continuous over the fenestrae. (c) Discontinuous capillaries, also called sinusoids, have a large lumen, many fenestrations with no diaphragm and a discontinuous or absent basal lamina. Full Figure and legend (29K)

In addition to morphological heterogeneity, there is functional heterogeneity of ECs, including roles in the control of vasoconstriction and vasodilation, blood coagulation and anticoagulation, fibrinolysis, leukocyte homing and diapedesis, acute inflammation and wound healing, atherogenesis, antigen presentation and catabolism of lipoproteins⁷. ECs in different organs are also required to perform different organ-appropriate functions. Brain ECs, for instance, interact with astrocyte endfeet to produce the blood-brain barrier by forming a continuous endothelium with complex tight junctions and highly regulated polarized endocytosis and transcytosis⁸. Liver ECs, in contrast, are highly fenestrated and discontinuous, and they line the hepatic sinusoids mediating the exchange of metabolites and the processing of toxins between the portal blood, Kupffer cells and hepatocytes⁹.

Transcriptional and antigenic EC heterogeneity
Structural and functional diversity of ECs is inevitably a result of molecular differences between EC populations; characterizing these differences will be crucial to understanding organ-appropriate endothelial signals. Over a decade ago, it was reported that endothelial cells in different organs display organ-specific antigens¹⁰. More recently, these have been identified using in vivo phage display, a 'biopanning' technique in which peptide libraries, which are expressed on the surface of bacteriophages, bind to EC surface molecules⁵. For example, lung-specific, breast-specific and prostate gland-specific ECs were identified by the binding of phage-presented peptides to membrane dipeptidase, aminopeptidase P and interleukin-11, respectively^5,^11,¹². Aminopeptidase N, notably, was found to be tumor EC−specific¹³. The screening of peptide libraries has revealed additional organ-specific EC populations in the brain, kidney, pancreas and many other locations^14,^15,¹⁶. In vivo phage display therefore creates a map of vascular 'addresses' and provides a promising basis for targeting therapeutic agents to tissues affected by cancer and other diseases.

Transcriptional differences have already been identified between various populations of ECs, such as those of arteries and veins^17,^18,¹⁹, large and small vessels²⁰ and normal and tumor vessels^21,²². A striking example of differential gene expression in mouse endothelium was noted when Wang and colleagues described the arterial, but not venous, expression of Ephrin-B2 (Efnb2), an Eph family transmembrane ligand¹⁷. In contrast, Eph-B4 (Efnb4), a receptor tyrosine kinase that binds Efnb2, is found at much higher levels in veins than in arteries. Similarly, the Notch and Gridlock genes are expressed in precursors of arteries, but not of veins, in the zebrafish embryo^18,¹⁹. Since then, the expression of several artery- and vein-specific genes has been described in a number of organisms²³. Genetic studies in zebrafish reveal that the arterial fate of ECs is the result of VEGF signaling through Notch²⁴. It will be interesting to see whether analogous mechanisms for arterial and venous cell fates will be identified in other organisms, such as mice.

In addition, Brown, Chi and colleagues at Stanford identified differential EC gene expression when they compared ECs derived from different blood vessels, organs and tissues using DNA microarrays (personal communication). Fifty-three different EC populations showed prominent distinctions between large and small vessel ECs, with thousands of genes consistently varied between them. Emerging technologies such as microarray screening will therefore allow the characterization of signaling pathways in different types of ECs. Identifying the important players in endothelial signaling, however, will require in vivo confirmation and further analysis.

Transcriptional profiling has also identified tumor-specific endothelial genes. Kinzler, St.Croix and colleagues compared the gene expression patterns of tumor and normal vessels^21,²². They found that 46 transcripts were specifically elevated in the tumor endothelium of adult tissues. Further analysis of a number of these genes, however, indicated that they were also expressed during embryonic vascular development and adult angiogenic remodeling events such as wound healing. Thus, analogous molecular mechanisms are likely to direct both tumor angiogenesis and normal physiological neovascularization. These analogies underscore the critical importance of studying embryonic vasculogenesis and angiogenesis and endothelial signaling as a foundation for the development of therapeutic applications that target tumors through their ECs.

Intercellular crosstalk: tissues 'speak' to endothelial cells
The obvious questions arise: what is the underlying cause of endothelial diversity and what is the impact of EC diversity on EC signaling and tissue development? How early does EC heterogeneity arise and why? Initially, ECs of the first embryonic vessels seem largely homogeneous, both morphologically and by the expression of early vascular markers²⁵. However, molecular distinctions between ECs with arterial or venous fates do exist at very early stages of vascular development¹⁸. Are there additional subpopulations of ECs in the early embryo that could be molecularly defined? How much plasticity do ECs retain after their local differentiation? These questions demand further investigation.

It is clear that later during development, capillary-bed heterogeneity is the result of tissue signals to ECs, or what has been called 'microenvironments'. Heterotopic transplantation experiments show that brain blood vessels growing into grafts of peripheral tissue acquire peripheral vascular morphology with fewer tight junctions, whereas peripheral vessels growing out of the graft into brain tissue assumed tight junctions, characteristic of brain vessels²⁶. In other experiments, heart tissue placed in the ear resulted in auricular vessel growth into the graft. These new vessels began to express von Willebrand factor, which is normally expressed in heart capillaries but not in ear vessels²⁷. These experiments demonstrate that vessels take on attributes associated with their microenvironment and that endothelial cells retain some level of cell fate plasticity.

In addition, it is overwhelmingly clear that tissues regulate vascular architecture by signaling to ECs through potent local angiogenic agents. In particular, the family of vascular endothelial growth factor (VEGF) proteins modulates a range of endothelial cell behaviors, ranging from their initial patterning in the embryo, to their recruitment during wound healing or tumor angiogenesis, to their maintenance in normal tissues²⁸. There are many specific examples of endothelial development being driven by locally secreted VEGF signals²⁹. For example, VEGF-expressing peripheral sensory nerves provide a template for branching and arteriogenesis in mouse embryonic skin, alveolar VEGF patterns the formation of the pulmonary vasculature, and VEGF produced by astrocytes in response to physiological hypoxia drives vascularization of the retina^30,^31,³². It is likely that paracrine regulation of vessel formation is a universal phenomenon regulated by the VEGF family of proteins.

A further level of complexity has recently been added by the discovery of a tissue-specific vascular growth factor. Ferrara, LeCouter and colleagues identified endocrine gland VEGF (EG-VEGF), which is functionally similar to but structurally distinct from VEGF-A, and is likely to function through different receptors. EG-VEGF is expressed only in steroidogenic glands and is mitogenic, motogenic and morphogenic for adrenal cortex endothelial cells³³. In contrast to VEGF-A, EG-VEGF did not affect ECs from aorta, umbilical vein or human dermal microvasculature. It will be interesting to see whether EG-VEGF proves uniquely specific to endocrine ECs, such as adrenal cortex endothelial cells, as its function is assayed on additional organ-specific ECs, and to see how many additional organ-specific angiogenic molecules will be identified. The discovery of EG-VEGF suggests the possible existence of a generalized mechanism whereby tissue-specific regulation of angiogenesis modulates and complements the more ubiquitous VEGF-A system. The identification of tissue-specific angiogenic molecules presents a clear challenge for future investigation, as does the development of specific antagonists that could provide targeted clinical applications.

Do endothelial cells 'speak' to tissues?
It is therefore evident that tissues signal to ECs. But what are the extent and the impact of EC signaling back to adjacent tissue cells during development and tissue maintenance in the adult? EC communication with surrounding cells is clearly suggested by EC production of a variety of humoral and growth factors, cytokines and cell-surface molecules (Supplementary Table 1 online)³⁴. The role of many of these EC signaling molecules during development or tissue maintenance remains unclear, however. In addition, it remains unknown whether any of these secreted or cell-surface factors are downstream of key regulators of embryonic endothelial development and neovascularization, such as the endothelial receptor tyrosine kinases VEGFR2 (Flk-1) or Tie-2.

There are, however, tantalizing clues regarding EC signaling from studies that disrupt blood vessel architecture by altering VEGF expression. This often results in a corresponding disruption of specific cell types, presumably as a result of loss or alteration of EC signals. For instance, disruption of VEGF signaling in kidney glomerular epithelium causes capillary tuft depletion and a dramatic decrease in the number of nephrons^35,³⁶. In contrast, when VEGF is overexpressed in lung mesenchyme, central pulmonary vessels increase in size, while the number of terminal buds is decreased and type I alveolar cell differentiation is inhibited³⁷.

Although these experiments suggest reciprocal EC signals, the effects of altered nutrition and gas exchange cannot be ruled out. Below, we examine a number of EC signaling examples, including studies that make use of in vitro assays such as tissue explants, coculture systems and endothelial cell lines, precluding the problem of blood flow. The list of examples below is by no means exhaustive, and additional instances may become available as investigators begin to look for such interactions.

Endothelial signals during blood vessel wall development
Soon after the endothelium forms primitive tubes, mesenchymal cells are recruited to its surface and organize into layers, generating mature, functional blood vessels³⁸. The sequence of these events led to the suggestion that EC signals might govern vessel wall assembly. Early work showed that platelet-derived growth factor (PDGF) is secreted by the basal surface of the endothelium, making it an excellent candidate for mediating vessel wall recruitment³⁹. In addition, PDGF receptors are present in the vascular wall smooth muscle cell progenitors⁴⁰. In vitro coculture experiments demonstrated that mesodermal cells migrate towards endothelial cells and that this migration was indeed mediated by PDGF-B⁴¹. Ablation of either PDGF-B or its receptor leads to hemorrhaging resulting from reduced pericyte coverage of microvessels^42,⁴³. Thus, PDGF is one of the first identified endothelial-derived, paracrine signals required for communication with surrounding mural cells (Fig. 3).

Figure 3. Vessel wall assembly.

	Angioblasts begin to differentiate into endothelial cells and assemble into tubes, most likely in response to VEGF signals from surrounding tissues. Once endothelial cells form patent tubes, pericytes and smooth muscle cells are recruited to form the vascular wall. In microvessels, PDGF signals are involved in the recruitment of pericytes. In large vessels, the Tie-2 and Ang-1 receptor-ligand pair is involved in the recruitment of smooth muscle cells. (Adapted from ref. 38) Full Figure and legend (32K)

Another signaling pathway important during vessel wall formation lies downstream of the angiopoietins and their receptors. The endothelial receptor tyrosine kinase Tie-2 is expressed on ECs and is stimulated by the angiopoietin ligand angiopoietin (Ang)-1, which is secreted by surrounding tissues⁴⁴. This activation leads to the endothelial production of an unidentified paracrine factor, which then recruits primordial smooth muscle. Absence of Ang-1 leads to vascular defects characterized by the failure to recruit both smooth muscle cell and pericyte precursors⁴⁵.

Endothelial signals during neural development
The ability of astrocytes to modulate structural and functional changes in the endothelium has been well characterized, although there have been a few studies that suggest reciprocal signaling. For instance, coculture of cortical astrocytes and brain capillary ECs results in the aggregation of astrocytes into "networks of elongated multicellular columns" and in enhanced astrocyte Ca⁺ responsiveness to bradykinin and glutamate⁴⁶. In similar assays, EC signaling stimulates astrocyte proliferation, and enhances aquaporin-4 localization on endfeet and astrocyte secretion of laminin-5 (refs. 47,48). In addition, rat optic nerve ECs induce differentiation of astrocyte precursors, specifically through endothelium-produced leukemia inhibitory factor⁴⁹.

Endothelial influence has also been shown to be important during neural cell fate determination. Bone morphogenetic protein (BMP)-2 secreted from the mouse dorsal aorta drives the differentiation of neural crest−derived neurons⁵⁰. In vitro clonal cultures of neural crest stem cells grown in the presence of BMP-2 result in rapid neurogenesis, as assayed both by neuronal markers such as MASH-1 and by neurite morphology. In contrast, transforming growth factor- beta

1 exclusively promotes smooth muscle differentiation of neural crest stem cells. In similar experiments, chick dorsal aorta BMPs enhance the development of adrenergic sympathetic neurons, both in vitro and when ectopically expressed in the embryo⁵¹. These experiments indicate that endothelial signals can act instructively to determine alternative neural cell fates.

Endothelial signals have also been implicated in adult hippocampal neurogenesis⁵². Neural and endothelial precursors proliferate together in the subgranule zone, an area known to generate new neurons throughout adult life. Tight clusters of these cells, consisting of proliferative aggregates of neural precursors, committed neuroblasts, glia and angioblasts, are commonly found at branch points or termini of fine capillaries and have been said to represent a 'vascular niche' for neurogenesis. Palmer and colleagues characterize how neurons and glia are generated over the course of approximately 1 month, in immediate juxtaposition to vascular endothelium (Fig. 4a). This study is limited to correlative inferences, but it does suggest a connection that should be investigated further.

Figure 4. Capillary endothelium and neurogenesis.

	(a) Clusters of neuronal, glial and endothelial precursors proliferate coordinately in 'neuroangiogenic' foci. These clusters are often located at branches or termini of fine capillaries. (b) Testosterone-driven neurogenesis functions through endothelial cues. Testosterone reaches HVC neural tissues through capillary blood. It activates androgen receptors on neural progenitor cells and upregulates neuron secretion of VEGF. This, in turn, promotes angiogenesis and endothelial secretion of BDNF. BDNF promotes migration and recruitment of HVC ventricular neurons. IGF, insulin-like growth factor. Full Figure and legend (64K)

Additional evidence of endothelial support of neurogenesis is revealed in cocultures of rat forebrain explants and ECs⁵³. Explants of the subependymal zone showed significant neuron maturation, survival, neurite outgrowth and migration when cultured with ECs. The EC signal responsible was identified as brain-derived neurotrophic factor (BDNF), and its neurotrophic effects could be blocked using soluble receptor molecules. Building on these results, another study characterized the coordination of neurogenesis and angiogenesis in the adult songbird brain⁵⁴. Goldman, Louissaint and colleagues showed that testosterone upregulates the expression of both VEGF and its receptor VEGFR2 (Flk-1) specifically in the higher vocal center (HVC) of songbirds, and induces the HVC endothelium to secrete BDNF. Additionally, they show that BDNF, in turn, acts as a neural chemotactic and survival factor for neurons in the ventricular zone (Fig. 4b).

Endothelial signals during tissue development
Endothelial signals have also been identified during the development of various tissues, such as adipose and bone tissue. In the embryo, an established vascular network precedes the formation of adipose tissue and both tissues grow coordinately throughout life. In vitro experiments have shown that both soluble and insoluble vascular extracellular matrix can drive the proliferation and differentiation of adipocyte precursors^55,⁵⁶. ECs from both subcutaneous and omental sites possessed the same paracrine ability, but control dermal ECs did not, which is consistent with the expected tissue-specific characteristics of ECs.

EC signaling is also evident during endochondral ossification. Bone development occurs through a series of steps dependent on the precise coupling of cartilage production and regression (chondrogenesis) and bone formation (osteogenesis). Resting cartilage is a distinctly avascular tissue that produces strongly antiangiogenic factors. During bone formation, chondrocytes first begin to proliferate, then quickly switch to expressing angiogenic molecules. This stimulates the growth of subchondral bone vessels into the cartilage, chondrocyte apoptosis and invasion of osteoclasts and osteoblasts. Blood vessel angiogenesis of the cartilage is thus coupled with its replacement by bone⁵⁷.

Coculture experiments with chick resting chondrocytes and porcine ECs showed that endothelial soluble factors cause rapid hypertrophy and late differentiation of chondrocytes in vitro⁵⁸. In another study, injection of soluble VEGF receptor proteins indicated that VEGF, originating from hypertrophic chondrocytes, is essential for correct directional angiogenesis into cartilage⁵⁹. In addition, this experiment showed that blood vessel invasion is absolutely required for the apoptosis of chondrocytes and the recruitment of osteoclasts. A third study showed that EC signaling occurs in vitro between human umbilical vein endothelial cells and osteoblast progenitor cells. Specifically, these two cell types communicate through gap junctions involving connexin-43, and osteoblast marker expression induction is dependent on this communication⁶⁰. Therefore, it seems likely that EC signals will prove crucial to the progression of cartilage replacement by bone.

Endothelial signals during organ development
Endothelial signals also have been identified during organogenesis. For instance, ECs exert multiple regulatory influences in the highly vascularized adrenal gland. The adrenal gland is composed of an inner medulla, which produces catecholamines, and an outer cortex, which secretes steroid hormones. The primary medullary cell type is the chromaffin cell, which is closely associated with endothelium. Cocultures of an adrenal medulla−derived cell line (PC12) with bovine adrenal medullary endothelial cells resulted in chromaffin-like differentiation of the PC12 cells⁶¹. Similarly, in the adrenal cortex, dense capillaries called sinusoids surround clusters of steroidogenic cells. Coculture studies of adrenal capillary ECs with steroidogenic zona glomerulosa cells show that different EC signals modulate aldosterone release from the zone glomerulosa cells^62,⁶³.

Reciprocal endothelial signaling is also evident during heart development, between the endocardium (endothelium) and the myocardium. Transforming growth factor- beta

and tight regulation of VEGF in the myocardium are required for the proper endothelial-to-mesenchymal transition that precedes endocardial cushion formation^64,^65,⁶⁶. In addition, endothelial signals are also required for myocardial development. Mouse or zebrafish mutants lacking the endocardium experience a failure of myocardial maturation^67,⁶⁸. One endocardial signal, neuregulin, binds its receptors ErbB2 (HER-2) and ErbB4 (HER-4) on adjacent cardiomyocytes and is required for trabeculation of the myocardium⁶⁹. Paracrine regulatory roles for the endocardium and the myocardial capillaries have also been identified in the modulation of cardiac muscle growth, contractile performance and rhythmicity^70,⁷¹.

Another example of EC signaling can be found in the renal collecting ducts and tubules that are immediately adjacent to microvessels. Coculture of ECs with either collecting duct cells or proximal tubule epithelial cells showed that regulation of transcellular sodium ion transport was dependent on endothelial mitric oxide^72,⁷³. In kidney organ cultures, VEGF was shown to induce EC proliferation with a marked coordinate increase in tubule epithelium proliferation⁷⁴. In contrast, inhibition of VEGF function in the kidney with soluble VEGF receptor proteins significantly disrupts renal morphogenesis and function³⁵. Finally, zebrafish mutants show that when ECs are absent or when vascular flow is inhibited, glomerular assembly is severely disrupted^75,⁷⁶. In the absence of ECs, podocytes differentiate normally but their foot processes are effaced, the glomerular basement membrane is discontinuous, and renal mesangial cells are absent⁷⁵. In the absence of blood flow, podocyte differentiation again seems normal, but endothelial cells do not invade podocyte clusters and the glomerular primordia do not migrate and fuse at the midline⁷⁶.

Endothelial signals are also important during the development of the pancreas (Fig. 5a,b)⁷⁷. Histological observations place key events of pancreatic differentiation in close association with endothelial cells. Pancreatic organogenesis is initiated when dorsal and ventral buds develop from the gut endoderm. These buds form precisely at locations that maintain close contact with the endothelium of the large vessels, the aorta dorsally and the vitelline veins ventrally. In addition, endocrine differentiation of the first insulin-and glucagon-expressing cells occurs in immediate contact with the overlying endothelium, as pancreatic cells delaminate from the budding endodermal epithelium.

Figure 5. Endothelial signals during pancreas and liver development.

	(a) Endothelial signals are required during the differentiation of the primitive endoderm into pancreatic and liver organs. Endothelial cells immediately contact the embryonic endoderm before any morphological signs of organogenesis, and this association continues as the organ matures into adulthood. (b) Endothelial cells are closely associated with endocrine cells in the mature pancreas. (c) Endothelial cells in the mature liver continue to be closely associated with hepatocytes. Full Figure and legend (102K)

Endothelial signals during pancreatic development have been shown in three sets of experiments. In vivo embryonic manipulation of frog embryos to block the formation of the dorsal aorta endothelium leads to failure of pancreatic gene expression in the underlying endoderm. This indicated that EC signals are required for pancreatic differentiation. Conversely, when VEGF-A is overexpressed in transgenic mice using a pancreatic promoter, hypervascularization of the pancreas leads to hyperplasia of the pancreatic islets. Notably, ectopic insulin-expressing cells and bud-like structures were observed in hypervascularized regions of the stomach and duodenum that also express the transgene. In vitro recombination experiments were also conducted to exclude the possible effects of altered nutrition and gas exchange. Dorsal endoderm and aortic endothelium from early embryos were cultured together, resulting in pancreatic gene induction. In contrast, recombination of the same endoderm with other tissues did not. These coculture experiments showed that endothelial signals were sufficient for the pancreatic program, at least in this assay with a relatively prepatterned endoderm.

Finally, endothelium-derived signals are also involved in hepatic development (Fig. 5a,c)⁷⁸. Endothelial-endoderm interactions can be observed before overt liver morphogenesis as angioblasts aggregate between the thickening hepatic epithelium and the septum transversum mesenchyme. These interactions presage liver budding from the ventral foregut endoderm. Liver cells then delaminate and migrate into the septum transversum mesenchyme, intermingling with angioblasts before the formation of functional blood vessels.

Three sets of experiments were carried out to investigate the role of ECs during liver development. First, liver budding was analyzed in mutant mice that lack most endothelial cells. Hepatic endoderm was observed to thicken normally and liver gene expression was normal. However, hepatic cells did not delaminate and proliferate into the septum transversum in VEGFR2 (Flk-1) mutant embryos. To exclude the possibility of secondary effects resulting from impaired embryonic growth or lack of blood flow, liver buds from wild-type and mutant embryos were cultured in vitro. The results clearly show that hepatic differentiation and growth was impaired in mutant explants, underlining the importance of endothelial signals during liver morphogenesis. Finally, explant experiments showed that the continued presence of vascular tissues was required, as inhibition of endothelial development using an angiogenic inhibitor (NK4) also inhibited the expression of liver albumin.

An important contradictory observation regarding the role of endothelial signals during hepatic development comes from the study of cloche zebrafish mutants, which are reported to lack most endothelial cells⁷⁹. Surprisingly, a recent report investigating liver morphogenesis in cloche embryos shows that liver (and pancreas) budding seems to proceed normally despite the absence of endothelium⁸⁰. The discrepancy might be explained by species-specific differences in the sources of required inductive signals, or in the morphological steps required for organ outgrowth. Early endoderm development proceeds by different cellular mechanisms in zebrafish and in vertebrates such as mice, with gut tube formation in fish emerging through the anastomoses of endoderm-lined cavities in fish, and gut formation in mice emerging through the transformation of an endodermal sheet into a tube⁸¹. Alternatively, it is conceivable that some form of EC still provides signals to precursors of liver and pancreas in the cloche mutants, given that cloche mutants still possess trunk vessels that are "lined by morphologically normal endothelial cells" and that cells expressing the vascular markers fli1, flk1 and flt4 are still present in the tail in a proliferative population of cells^82,⁸³. Yet another interesting interpretation of the divergent observations in fish and mice might be based on fundamental differences in the liver function of fish and higher vertebrates. In mammals, the embryonic liver is a site of hematopoiesis and may have evolved to require signals from the endothelial vessels that permit the invasion of hematopoietic cells. In zebrafish, in contrast, hematopoiesis occurs in the kidney and the liver may therefore not have evolved a developmental requirement for signals from endothelial cells⁸⁴.

A recent study, however, has confirmed a role for EC signals in the liver. LeCouter, Ferrara and colleagues demonstrate that VEGF-A induces liver endothelial cells to communicate with neighboring hepatocytes by promoting the secretion of growth and survival factors⁸⁵. First, administration of circulating VEGF-A caused a substantial enlargement of the liver because of increased proliferation of hepatocytes and other liver cell types. Surprisingly, this effect was not a direct result of increased liver EC proliferation, but rather a result of the secretion of hepatocyte growth factor and interleukin-6 from the sinusoidal endothelium through activation of VEGFR1 (Flt-1). In both in vitro and in vivo assays, a VEGFR1 (Flt-1)-specific ligand stimulated hepatocyte proliferation in the absence of endothelial proliferation, whereas a VEGFR2 (Flk-1)-specific ligand resulted in increased EC growth in the absence of hepatocyte proliferation. Finally, signaling through VEGFR1 (Flt-1) protects liver from hepatoxin-induced damage (by CCl₄), implying the therapeutic potential of VEGFR1 (Flt-1) agonists in liver disorders.

Conclusions and perspectives
The emerging likelihood that paracrine endothelial cues have a role in tissue differentiation and organ formation underlines the importance of investigating basic vascular biology. The reports outlined in this review enumerate examples of crosstalk between ECs and adjacent tissues, although these represent only the beginning of our understanding of a potentially global role for endothelial cell signaling. At the moment, however, we understand little about the regulatory mechanisms that direct EC paracrine signaling, either in the developing vasculature or in adult vessels, and we are only beginning to identify downstream events in responding cells. There is a real need to characterize endothelial signaling at the cellular level, by identifying, for example, relevant intracellular signaling pathways and studying how this translates into signals outside the cell. We will also need to examine closely the endothelial-tissue interface and understand the role of the endothelial basal lamina. Characterizing the nature of this signaling and the added complexity of tissue-specific heterogeneity of ECs will undoubtedly have many applications. It may allow the exploitation of tissue-specific ECs as a basis for the development and delivery of targeted therapies⁵. It may also provide a basis for identifying key signaling events during cancer progression, given that tumor growth involves reciprocal signaling events between ECs and cancer cells⁸⁶. Understanding EC biology will also provide a foundation for learning to build new vascular networks, in ischemic or engineered tissues, for example⁸⁷. Finally, we predict that endothelial signaling may also have an impact on our understanding of stem cell differentiation, which may require the contribution of inducing factors secreted from ECs.

Supplementary information is available on the Nature Medicine website.

	Top

REFERENCES

Florey, L. The endothelial cell. Br. Med. J. 5512, 487-490 (1966). | PubMed |
Garlanda, C. & Dejana, E. Heterogeneity of endothelial cells. Specific markers. Arterioscler. Thromb. Vasc. Biol. 17, 1193-1202 (1997). | PubMed | ChemPort |
Majno, G. & Joris, I. Endothelium 1977: a review. Adv. Exp. Med. Biol. 104, 169-225, 481-526 (1978).
Thorin, E. & Shreeve, S.M. Heterogeneity of vascular endothelial cells in normal and disease states. Pharmacol. Ther. 78, 155-166 (1998). | Article | PubMed | ChemPort |
Pasqualini, R., Arap, W. & McDonald, D.M. Probing the structural and molecular diversity of tumor vasculature. Trends Mol. Med. 8, 563-571 (2002). | Article | PubMed | ChemPort |
Suter, E.R. & Majno, G. Passage of lipid across vascular endothelium in newborn rats. An electron microscopic study. J. Cell Biol. 27, 163-177 (1965). | PubMed | ChemPort |
Gerritsen, M.E. Functional heterogeneity of vascular endothelial cells. Biochem. Pharmacol. 36, 2701-2711 (1987). | Article | PubMed | ChemPort |
Abbott, N.J. Astrocyte-endothelial interactions and blood-brain barrier permeability. J. Anat. 200, 629-638 (2002). | Article | PubMed | ChemPort |
Kmiec, Z. Cooperation of liver cells in health and disease. Adv. Anat. Embryol. Cell Biol. 161, III-XIII, 1-151 (2001).
Auerbach, R. Vascular endothelial cell differentiation: organ-specificity and selective affinities as the basis for developing anti-cancer strategies. Int. J. Radiat. Biol. 60, 1-10 (1991). | PubMed | ChemPort |
Rajotte, D. & Ruoslahti, E. Membrane dipeptidase is the receptor for a lung-targeting peptide identified by in vivo phage display. J. Biol. Chem. 274, 11593-11598 (1999). | Article | PubMed | ChemPort |
Essler, M. & Ruoslahti, E. Molecular specialization of breast vasculature: a breast-homing phage-displayed peptide binds to aminopeptidase P in breast vasculature. Proc. Natl. Acad. Sci. USA 99, 2252-2257 (2002). | Article | PubMed | ChemPort |
Pasqualini, R. et al. Aminopeptidase N is a receptor for tumor-homing peptides and a target for inhibiting angiogenesis. Cancer Res. 60, 722-727 (2000). | PubMed | ChemPort |
Pasqualini, R. & Ruoslahti, E. Organ targeting in vivo using phage display peptide libraries. Nature 380, 364-366 (1996). | PubMed | ChemPort |
Rajotte, D. et al. Molecular heterogeneity of the vascular endothelium revealed by in vivo phage display. J. Clin. Invest. 102, 430-437 (1998). | PubMed | ChemPort |
Samoylova, T.I. & Smith, B.F. Elucidation of muscle-binding peptides by phage display screening. Muscle Nerve 22, 460-466 (1999). | Article | PubMed | ChemPort |
Wang, H.U., Chen, Z.F. & Anderson, D.J. Molecular distinction and angiogenic interaction between embryonic arteries and veins revealed by ephrin-B2 and its receptor Eph-B4. Cell 93, 741-753 (1998). | PubMed | ChemPort |
Lawson, N.D. et al. Notch signaling is required for arterial-venous differentiation during embryonic vascular development. Development 128, 3675-3683 (2001). | PubMed | ChemPort |
Zhong, T.P., Rosenberg, M., Mohideen, M.A., Weinstein, B. & Fishman, M.C. gridlock, an HLH gene required for assembly of the aorta in zebrafish. Science 287, 1820-1824 (2000). | Article | PubMed | ChemPort |
Muller, A.M., Hermanns, M.I., Cronen, C. & Kirkpatrick, C.J. Comparative study of adhesion molecule expression in cultured human macro- and microvascular endothelial cells. Exp. Mol. Pathol. 73, 171-180 (2002). | Article | PubMed | ChemPort |
St Croix, B. et al. Genes expressed in human tumor endothelium. Science 289, 1197-1202 (2000). | Article | PubMed | ChemPort |
Carson-Walter, E.B. et al. Cell surface tumor endothelial markers are conserved in mice and humans. Cancer Res. 61, 6649-6655 (2001). | PubMed | ChemPort |
Lawson, N.D. & Weinstein, B.M. Arteries and veins: making a difference with zebrafish. Nat. Rev. Genet. 3, 674-682 (2002). | Article | PubMed | ChemPort |
Lawson, N.D., Vogel, A.M. & Weinstein, B.M. Sonic hedgehog and vascular endothelial growth factor act upstream of the Notch pathway during arterial endothelial differentiation. Dev. Cell 3, 127-136 (2002). | PubMed | ChemPort |
Drake, C.J. & Fleming, P.A. Vasculogenesis in the day 6.5 to 9.5 mouse embryo. Blood 95, 1671-1679 (2000). | PubMed | ChemPort |
Stewart, P.A. & Wiley, M.J. Developing nervous tissue induces formation of blood-brain barrier characteristics in invading endothelial cells: a study using quail--chick transplantation chimeras. Dev. Biol. 84, 183-192 (1981). | PubMed | ChemPort |
Aird, W.C. et al. Vascular bed-specific expression of an endothelial cell gene is programmed by the tissue microenvironment. J. Cell Biol. 138, 1117-1124 (1997). | Article | PubMed | ChemPort |
Ferrara, N. & Alitalo, K. Clinical applications of angiogenic growth factors and their inhibitors. Nat. Med. 5, 1359-1364 (1999). | Article | PubMed | ChemPort |
D'Amore, P.A. & Ng, Y.S. Won't you be my neighbor? Local induction of arteriogenesis. Cell 110, 289-292 (2002). | PubMed | ChemPort |
Mukouyama, Y.S., Shin, D., Britsch, S., Taniguchi, M. & Anderson, D.J. Sensory nerves determine the pattern of arterial differentiation and blood vessel branching in the skin. Cell 109, 693-705 (2002). | PubMed | ChemPort |
Ng, Y.S., Rohan, R., Sunday, M.E., Demello, D.E. & D'Amore, P.A. Differential expression of VEGF isoforms in mouse during development and in the adult. Dev. Dyn. 220, 112-121 (2001). | Article | PubMed | ChemPort |
Stone, J. et al. Development of retinal vasculature is mediated by hypoxia-induced vascular endothelial growth factor (VEGF) expression by neuroglia. J. Neurosci. 15, 4738-4747 (1995). | PubMed | ChemPort |
LeCouter, J. et al. Identification of an angiogenic mitogen selective for endocrine gland endothelium. Nature 412, 877-884 (2001). | Article | PubMed | ChemPort |
Vapaatalo, H. & Mervaala, E. Clinically important factors influencing endothelial function. Med. Sci. Monit. 7, 1075-1085 (2001). | PubMed | ChemPort |
Kitamoto, Y., Tokunaga, H. & Tomita, K. Vascular endothelial growth factor is an essential molecule for mouse kidney development: glomerulogenesis and nephrogenesis. J. Clin. Invest. 99, 2351-2357 (1997). | PubMed | ChemPort |
Gerber, H.P. et al. VEGF is required for growth and survival in neonatal mice. Development 126, 1149-1159 (1999). | PubMed | ChemPort |
Zeng, X., Wert, S.E., Federici, R., Peters, K.G. & Whitsett, J.A. VEGF enhances pulmonary vasculogenesis and disrupts lung morphogenesis in vivo. Dev. Dyn. 211, 215-227 (1998). | Article | PubMed | ChemPort |
Hungerford, J.E. & Little, C.D. Developmental biology of the vascular smooth muscle cell: building a multilayered vessel wall. J. Vasc. Res. 36, 2-27 (1999). | Article | PubMed | ChemPort |
Zerwes, H.G. & Risau, W. Polarized secretion of a platelet-derived growth factor-like chemotactic factor by endothelial cells in vitro. J. Cell Biol. 105, 2037-2041 (1987). | PubMed | ChemPort |
Shinbrot, E., Peters, K.G. & Williams, L.T. Expression of the platelet-derived growth factor receptor during organogenesis and tissue differentiation in the mouse embryo. Dev. Dyn. 199, 169-175 (1994). | PubMed | ChemPort |
Hirschi, K.K., Rohovsky, S.A. & D'Amore, P.A. PDGF, TGF-, and heterotypic cell-cell interactions mediate endothelial cell-induced recruitment of 10T1/2 cells and their differentiation to a smooth muscle fate. J. Cell Biol. 141, 805-814 (1998). | Article | PubMed | ChemPort |
Leveen, P. et al. Mice deficient for PDGF B show renal, cardiovascular, and hematological abnormalities. Genes Dev. 8, 1875-1887 (1994). | PubMed | ChemPort |
Lindahl, P., Johansson, B.R., Leveen, P. & Betsholtz, C. Pericyte loss and microaneurysm formation in PDGF-B-deficient mice. Science 277, 242-245 (1997). | Article | PubMed | ChemPort |
Davis, S. et al. Isolation of angiopoietin-1, a ligand for the TIE2 receptor, by secretion-trap expression cloning. Cell 87, 1161-1169 (1996). | PubMed | ChemPort |
Suri, C. et al. Requisite role of angiopoietin-1, a ligand for the TIE2 receptor, during embryonic angiogenesis. Cell 87, 1171-1180 (1996). | PubMed | ChemPort |
Yoder, E.J. Modifications in astrocyte morphology and calcium signaling induced by a brain capillary endothelial cell line. Glia 38, 137-145 (2002). | Article | PubMed |
Estrada, C., Bready, J.V., Berliner, J.A., Pardridge, W.M. & Cancilla, P.A. Astrocyte growth stimulation by a soluble factor produced by cerebral endothelial cells in vitro. J. Neuropathol. Exp. Neurol. 49, 539-549 (1990). | PubMed | ChemPort |
Rash, J.E., Yasumura, T., Hudson, C.S., Agre, P. & Nielsen, S. Direct immunogold labeling of aquaporin-4 in square arrays of astrocyte and ependymocyte plasma membranes in rat brain and spinal cord. Proc. Natl. Acad. Sci. USA 95, 11981-11986 (1998). | Article | PubMed | ChemPort |
Mi, H., Haeberle, H. & Barres, B.A. Induction of astrocyte differentiation by endothelial cells. J. Neurosci. 21, 1538-1547 (2001). | PubMed | ChemPort |
Shah, N.M., Groves, A.K. & Anderson, D.J. Alternative neural crest cell fates are instructively promoted by TGF superfamily members. Cell 85, 331-343 (1996). | PubMed | ChemPort |
Reissmann, E. et al. Involvement of bone morphogenetic protein-4 and bone morphogenetic protein-7 in the differentiation of the adrenergic phenotype in developing sympathetic neurons. Development 122, 2079-2088 (1996). | PubMed | ChemPort |
Palmer, T.D., Willhoite, A.R. & Gage, F.H. Vascular niche for adult hippocampal neurogenesis. J. Comp. Neurol. 425, 479-494 (2000). | Article | PubMed | ChemPort |
Leventhal, C., Rafii, S., Rafii, D., Shahar, A. & Goldman, S.A. Endothelial trophic support of neuronal production and recruitment from the adult mammalian subependyma. Mol. Cell. Neurosci. 13, 450-464 (1999). | Article | PubMed | ChemPort |
Louissaint, A., Jr., Rao, S., Leventhal, C. & Goldman, S.A. Coordinated interaction of neurogenesis and angiogenesis in the adult songbird brain. Neuron 34, 945-960 (2002). |