ECs (endothelial cells) in the developing vasculature are heterogeneous in morphology, function and gene expression. Inter-endothelial signalling via Dll4 (Delta-like 4) and Notch has recently emerged as a key regulator of endothelial heterogeneity, controlling arterial cell specification and tip versus stalk cell selection. During sprouting angiogenesis, tip cell formation is the default response to VEGF (vascular endothelial growth factor), whereas the stalk cell phenotype is acquired through Dll4/Notch-mediated lateral inhibition. Precisely how Notch signalling represses stalk cells from becoming tip cells remains unclear. Multiple components of the VEGFR (VEGF receptor) system are regulated by Notch, suggesting that quantitative differences in protein expression between adjacent ECs may provide key features in the formation of a functional vasculature. Computational modelling of this selection process in iterations, with experimental observation and validation greatly facilitates our understanding of the integrated processes at the systems level. We anticipate that the study of mosaic vascular beds of genetically modified ECs in dynamic interactions with wild-type ECs will provide a powerful tool for the investigation of the molecular control and cellular mechanisms of EC specification.
- Delta-like 4 (Dll4)
- vascular endothelial growth factor receptor (VEGFR)
Most chordates and higher organisms rely on a blood vascular system for transport of oxygen and nutrients. As cells require oxygen and nutrients to divide, blood vessel formation (angiogenesis) is a prerequisite for tissues to expand beyond the size limits imposed by oxygen diffusion. In a variety of pathologies such as cancer, diabetic retinopathy and rheumatoid arthritis, deregulated overproduction of vessels contributes to the progression of the disease. Increasing insights into the basic regulatory mechanisms of angiogenesis greatly support the development of successful therapeutic approaches aimed at modulating the vasculature in disease. Recent advances highlight the importance of functional EC (endothelial cell) specification and co-ordinated cell behaviour to pattern growing sprouts. Each vascular sprout is headed by an explorative EC (the tip cell), tightly connected to following endothelial stalk cells that build the new vascular tube [1,2]. The family of VEGFs (vascular endothelial growth factors) and their receptors are major regulators of sprouting angiogenesis (reviewed in [3,4]). The receptors VEGFR1 (VEGF receptor-1), VEGFR2 and VEGFR3 are transmembrane tyrosine kinases that dimerize and become activated on ligand binding. All three receptors are expressed by ECs during development and are all required for proper angiogenesis, since deletion of any of the receptors in mice leads to vascular abnormalities and death at E (embryonic day) 8.5–10.5 [5–7].
In addition to the receptor tyrosine kinases and their ligands, several recent studies illustrate the importance of Notch signalling components including Dll4 (Delta-like ligand 4), Jagged-1 and Notch1 in EC specification during formation of a functional vascular network [8–11]. Multiple connections between the VEGFR system and the Dll4/Notch signalling cascade have been described previously [11–14], indicating the existence of intricate regulatory feedback loops that ensure adaptive and robust patterning of vascular networks.
To dynamically visualize and modify the molecular regulation and cellular behaviour in vivo poses major technical challenges. Computational modelling in reiteration with experimental observation allows us to investigate emergent behaviour of regulatory systems and cell–cell communication. An agent-based computational model comprising ECs in a vessel responding to environmental VEGF and to each other, through Dll4/Notch signalling, simulates molecular profiles and their impact on morphology (filopodia extension/retraction) and hence the patterning of tip and stalk cells .
In the present brief review, we will discuss emerging concepts, derived from experimentation and computational modelling, of the dynamic interplay between VEGF and Notch signalling during tip/stalk cell establishment and vascular patterning.
Most evidence points to VEGFR1 functioning as an effective decoy for VEGFA in ECs, thereby limiting signalling via VEGFR2; importantly, vegfr1-knockout mice die at E8.5–9 due to vascular overgrowth, whereas targeting of the intracellular domain has no vascular phenotype [6,16]. Furthermore, the VEGFA–VEGFR1 complex has a roughly 10-fold higher affinity when compared with the VEGFA–VEGFR2 complex . Precisely what VEGFR1 does to the EC behaviour is unclear. Interestingly, vegfr1-null cells contribute to a functional vasculature when ‘diluted’ with wild-type cells in chimaeric mice . A detailed description of the effect of VEGFR1 deficiency in this context is still lacking however. Under certain circumstances VEGFR1 seems to have a positive effect on VEGFR2 signalling by retaining VEGFA to the plasma membrane . VEGFA bound to VEGFR1 may in turn be released on competition with the VEGFR1-specific ligand PlGF (placenta growth factor) [17,19]. The balance between the membrane-bound and soluble forms of VEGFR1 adds to this complexity. Taken together, these results suggest that ECs may use cell-bound or soluble VEGFR1 to modulate VEGFR2 activity in both a cell-autonomous and a non-cell-autonomous manner [20,21].
VEGFR2 is expressed by all ECs throughout development. Signalling via its receptor tyrosine kinase domains is a prerequisite for vascular development and angiogenesis. This has been clearly demonstrated by the early lethality in mice as a result of either targeted deletion of the complete receptor or by mutation of Tyr-1173 to phenylalanine in the C-terminus [7,22,23]. The receptor is mainly activated by the hypoxia-inducible dimeric ligand VEGFA. VEGF signalling through VEGFR2 induces permeability, proliferation, migration and survival of ECs depending on the context (reviewed in ).
VEGFR3 [alternatively denoted as Flt-4 (Fms-like tyrosine kinase-4) in mouse] is predominantly expressed in embryonic blood endothelium, lymphatic ECs, monocytes and macrophages. Mice with deleted VEGFR3 show blood vessel malformations and die at E9.5–10.5 due to cardiac failure . VEGFR3 is the major receptor inducing lymphangiogenesis , but recent publications demonstrate additional expression and important function during sprouting angiogenesis. Intra-vitreous injections of VEGFR3-blocking antibodies led to reduced vessel density in the postnatal mouse retina, suggesting that VEGFR3 acts as a positive regulator of sprouting angiogenesis . In contrast, Matsumura et al.  suggested that VEGFR3 negatively affects VEGFR2 activity as blockage of VEGFR3 with antibodies resulted in the same cellular morphological response as did treatment with VEGFA, in vitro. Further studies are warranted to clarify the precise role of VEGFR3, the relationship to VEGFR2 including possible heterodimerization, as well as the localized expression and function of the various ligand combinations affecting the sprouting process. VEGF-C and -D are the main ligands for VEGFR3; however, when proteolytically processed, they can also bind (and activate) VEGFR2 (reviewed in ). VEGF-C is expressed by macrophages and subsets of ECs throughout the developing retinal vasculature , suggesting potential paracrine and autocrine functions in sprouting angiogenesis. The impact of differential VEGFR3 expression on EC behaviour is not known.
Notch in angiogenesis
Like the VEGFR system, many of the Notch receptors and their ligands are required during vascular development and morphogenesis. In mammals, the transmembrane Notch receptors 1, 2, 3 and 4 interact with five single-pass transmembrane ligands, Jagged-1 and -2 and Dll-1, -3 and -4. On ligand binding, the extracellular domain is cleaved by proteases [ADAM (a disintegrin and metalloproteinase)]. It is suggested that this induces a conformational change allowing γ-secretase to cleave Notch to release the NICD (Notch intracellular domain), enabling its transition to the nucleus. In the nucleus, NICD binds to the RBP-Jκ (recombination signal-binding protein 1 for Jκ), which normally functions as a transcriptional repressor but, when in complex with NICD, turns into a transcriptional activator . Most of the Notch targets are repressors involved in a negative feedback loop. The downstream targets Hes1 and Hes7 repress their own expression by interactions with their own promoters, whereas the Nrarp (Notch-regulated ankyrin repeat protein) represses Notch signalling activity by promoting NICD degradation . A number of different modifications of the Notch receptors and their ligands in the developing as well as tumour vasculature dramatically affect angiogenesis (reviewed in [27,29]). Taken together, these results indicate that Dll–Notch signalling restricts EC proliferation, branching and angiogenesis in development as well as in a tumour context [27,30]. Endothelial Jagged-1, however, appears to promote sprouting and branching by opposing the Dll4-mediated activation of Notch1 . Furthermore, Notch signalling plays a central role in the initiation of venous to arterial differentiation [27,29].
VEGFRs and Notch: the interplay
VEGFR1, VEGFR2 and VEGFR3 are differentially expressed in the endothelium of growing vessels [25,31]. The underlying mechanisms and the importance of such restricted receptor expression in patterning of the blood vasculature have not been clarified. The previously reported regulatory functions of Notch activation on the expression of the receptors, together with the distinctive expression pattern of Dll4 , suggest a tight connection of these pathways during angiogenesis (reviewed in ). In the absence of Dll4 or Notch, all ECs will respond to VEGFR2 activation by extensions of highly dynamic protrusions of thin actin-rich filopodia [1,33]. Computer simulations indicate that the proposed VEGF–Notch feedback loop is sufficient to dynamically select tip and stalk cells under VEGF stimulation . In vessels comprising Dll4+/− ECs, Notch activity still occurs, but the levels are not sufficient to support adequate lateral inhibition; thus all cells remain in the default tip cell state .
The initial migratory sprouting response is induced by VEGF-A gradients. However, not all cells in a pre-existing vessel will adopt the migratory behaviour in response to VEGF. Adjacent cells communicate via Dll4 and Notch1 to pattern out explorative tip cells and tube-forming stalk cells [9,34]. Most of the tip cells express higher mRNA levels of Dll4 than their stalk cell neighbours. Several mechanisms could account for the restricted expression of Dll4 in angiogenic tissues. For example, Dll4 is up-regulated in response to hypoxia . However, whereas this could explain high levels in the leading tips, the strong arterial Dll4 expression occurs in a highly oxygenated context. In addition the near on/off expression of Dll4 mRNA between adjacent cells indicates that other mechanisms must be involved. VEGF is able to induce Dll4 expression via the phosphoinositide 3-kinase/Akt (also called protein kinase B) pathway downstream of VEGFR2 . Moreover, regions where ECs strongly express Dll4, such as the angiogenic regions in the retinal vasculature, show intense VEGFR2 expression . Whether high VEGFR2 and Dll4 expression levels correlate in single cells in situ has not been resolved. Expression of all three VEGFRs is in turn regulated by Notch signalling (see Figure 1). VEGFR1 is induced upon Notch activation , but the underlying molecular mechanism is unknown. In contrast, VEGFR2 is repressed by Dll4 and Notch. This is at least in part mediated by the basic helix–loop–helix transcription factor HESR1 (also called CHF2) [11–14]. The regulation of VEGFR3 is less clear. Siekmann and Lawson  describe up-regulation of VEGFR3 in Notch-deficient cells (i.e. Notch restricts VEGFR3) in the zebrafish . Also, Tammela et al.  report induced expression of VEGFR3 in Dll4+/− retinas and upon inhibition of γ-secretase. In contrast, Shawber et al.  reported up-regulation of VEGFR3 in vivo and in vitro in ECs overexpressing NICD. They also demonstrated that the NICD/CSL (CBF1/suppressor of Hairless/Lag-1) complex bound directly to the VEGFR3 promoter to initiate its transcription . This discrepancy may indicate differential regulation in time and space.
Also the expression of the VEGFR1-specific ligand PlGF is reduced by Notch activation . The results from various organs and species found a conserved regulatory connection between the VEGFRs, their activity and Notch; however, the precise functional role of this connection remains poorly understood.
Several studies of mosaic vascular beds in zebrafish and mice illustrated that Notch signalling-deficient cells preferentially adopt the tip cell position [9,10]. In contrast, cells with constitutive Notch activation avoid this position, together suggesting that Notch activity cell-autonomously gauges the potential of cells to compete for the tip cell position. Also ECs deficient in Jagged-1 expression in a mosaic context are highly represented as tip cells, indicating a negative role for Jagged-1 in Notch signalling . To what extent this may be a consequence of a changed VEGFR expression profile and activity remains to be resolved. Furthermore, the cell-autonomous role of Dll and VEGFR levels has not been investigated.
Current concepts favour the idea of a constant battle for the tip cell position, which is fought at the level of Dll4 and VEGFR2 signalling output. Imaging of the zebrafish vasculature illustrates a dynamic migratory behaviour of ECs that clearly adds to the complexity of cellular competition, as cells would constantly change their Notch input/output depending on new neighbours. The stability of tip versus stalk cell selection and how long an individual EC maintains a given position in the sprout remain unclear.
The dynamics of Notch signalling and downstream components has been studied in detail during somitogenesis in mouse, chicken and zebrafish. Differences in protein stability, timing of transcription and timing of translation mediate an oscillating expression of several signalling components. Spatial patterning of somitogenesis depends on synchronization of these oscillations between neighbouring cells through reciprocal Notch signalling (reviewed in [27,38]). In the vasculature, however, direct evidence for possible oscillations is still lacking. The expression pattern and dynamic behaviour of ECs during angiogenesis do not exclude this possibility. Indeed our simulation studies predict that selection requires an initial oscillation in protein levels, caused by adjacent cells competing. This oscillation frequency was predicted to increase with VEGF concentration such that normal patterning eventually could not be achieved; cells instead synchronously oscillate between tip and stalk phenotypes . We anticipate that a deeper understanding of the dynamics of VEGF–Notch signalling under physiological and pathological conditions will greatly advance our concepts of the mechanisms controlling vascular patterning.
This work was supported by Cancer Research UK, the Lister Institute of Preventive Medicine, the European Molecular Biology Organization Young Investigator Programme (H.G.) and a European Molecular Biology Organization long-term postdoctoral fellowship (L.J.).
Molecular and Cellular Mechanisms of Angiogenesis: Biochemical Society Focused Meeting held at University of Chester, Chester, U.K., 15–17 July 2009. Organized and Edited by Ian Zachary (University College London, U.K.) and Sreenivasan Ponnambalam (Leeds, U.K.).
Abbreviations: D114, Delta-like ligand 4; E, embryonic day; EC, endothelial cell; NICD, Notch intracellular domain; PlGF, placenta growth factor; VEGF, vascular endothelial growth factor; VEGFR, VEGF receptor
- © The Authors Journal compilation © 2009 Biochemical Society