Complementary actions of VEGF and Angiopoietin-1 on blood vessel growth and leakage

Abstract

Vascular endothelial growth factor (VEGF) and Angiopoietins are families of vascular-specific growth factors that regulate blood vessel growth, maturation and function. To learn more about the effects of these factors in vivo, we have overexpressed VEGF-A or Angiopoietin-1 (Ang1) in two systems in mice, and examined the effects on blood vessel growth and function. In one set of studies, VEGF, Ang1, or both factors, were transgenically overexpressed in the skin under the keratin-14 (K14) promoter. The skin of mice overexpressing VEGF (K14-VEGF) had numerous tortuous, capillary-sized vessels which were leaky to the plasma tracer Evans blue under baseline conditions. In contrast, the skin of mice overexpressing Ang1 (K14-Ang1) had enlarged dermal vessels without a significant increase in vessel number. These enlarged vessels were less leaky than those of wild-type mice in response to inflammatory stimuli. In double transgenic mice overexpressing VEGF and Ang1, the size and number of skin vessels were both increased; however, the vessels were not leaky. In a second set of studies, VEGF or Ang1 was systemically delivered using an adenoviral approach. Intravenous injection of adenovirus encoding VEGF (Adeno-VEGF) resulted in widespread tissue oedema within 1–2 days after administration, whereas injection of Adeno-Ang1 resulted in the skin vessels becoming less leaky in response to topical inflammatory stimuli or local injection of VEGF. The decreased leakage was not accompanied by morphological changes. Thus, overexpressing VEGF appears to promote growth of new vessels accompanied by plasma leakage, whereas overexpressing Ang1 promotes the enlargement of existing vessels and a resistance to leakage. Further understanding of the interrelationship of these factors during normal development could lead to their application in the treatment of ischaemic diseases.

Introduction

Two families of endothelial-specific growth factors, vascular endothelial growth factors and Angiopoietins, are necessary for the formation of blood vessels. These factors seem to act in co-ordinated and complementary ways to produce mature blood vessels. Vascular endothelial growth factor A (VEGF-A, here called VEGF), the initial member of the VEGF family to be identified (Ferrara et al. 1991; Dvorak et al. 1992), is essential for early blood vessel formation and angiogenesis. Mice deficient in even one allele for VEGF die in embryogenesis due to a decrease in endothelial cell number and severe defects in blood vessel formation (Carmeliet et al. 1996; Ferrara et al. 1996). Angiopoietin-1, the first member of a second family of endothelial-specific factors (Davis et al. 1996; Maisonpierre et al. 1997; Valenzuela et al. 1999), is essential for a later stage of blood vessel formation. Mice deficient for Ang1 die by embryonic day 12.5 (E12.5) due to defects in vessel remodelling and maturation (Suri et al. 1996). VEGF and Ang1 act via distinct endothelial-cell-specific tyrosine kinase receptors, which are also essential for embryogenesis and blood vessel formation. Deletion of the VEGF receptor VEGF-R2 (Fong et al. 1995; Shalaby et al. 1995) or the Angiopoietin receptor (Tie-2) (Dumont et al. 1994; Sato et al. 1995) is also lethal to the embryo, with phenotypes broadly similar to those in the corresponding ligand-deficient mice.

Thus a model has evolved to describe the role of these factors in developmental angiogenesis. In particular, VEGF-A and its endothelial cell receptor VEGF-R2 are believed to play a role in vasculogenesis and early angiogenesis, whereas Ang1 and its receptor are believed to be involved in blood vessel remodelling and maturation (Hanahan, 1997; Yancopoulos et al. 2000). However, because of the early lethality of the gene-targeted embryos, it has been difficult to fully define the role of these factors in adult and pathological angiogenesis.

To gain further insight into the function of VEGF and Ang1 in vivo, we have used two approaches to overexpress these factors in mice. In particular, we have used tissue-specific transgenic mice and adenoviral vectors. In the transgenic approach, VEGF or Ang1 was overexpressed continuously in the mouse skin from mid-embryonic stages into adulthood. The factors, and their effects, were localized to the skin (Fig. 1A). In the adenoviral approach, VEGF or Ang1 was expressed systemically in otherwise normal adult mice, and the factors acted for a defined duration on blood vessels throughout the mouse (Fig. 1B). Comparing and contrasting these different approaches and the different factors has helped reveal the distinct actions of VEGF and Ang1 on blood vessel growth and leakage.

Fig. 1.

Outline of Evans blue leakage experiments in (A) K14 transgenic mice and (B) adenovirus treated mice. (A) In transgenic mice, secreted factor (VEGF or Ang1 – green) is overexpressed in embryonic skin and throughout the lifetime of the mouse. Evans blue dye (blue) was injected into adult mice, and then inflammatory stimuli (red) were applied topically to ear skin on one side (other ear served as control). (B) In adenovirus experiments, adenovirus encoding factor (VEGF or Ang1 – green hexagons) was injected intravenously into normal adult mice (white). Factor was overexpressed in liver and secreted into circulation. Inflammatory stimuli were applied topically to the distal site: the ear skin on one side.

Transgenic overexpression – K14-VEGF and K14-Ang1 mice

Several groups have generated mice which overexpress VEGF-A in the skin, under either K14 or K5 promoters (Detmar et al. 1998; Larcher et al. 1998; Thurston et al. 1999). We generated K14-VEGF₁₆₄ transgenic mice (Thurston et al. 1999), which appeared normal but had some redness in the skin of the ears and snout. The epidermis of the K14-VEGF mice was thickened, and the dermis contained infiltrating leucocytes. Lesions appeared in the ear skin of older mice (Thurston et al. 1999). Upon examination of tissue sections stained for endothelial cells with antibodies to platelet endothelial cell adhesion molecule (PECAM, CD31) or whole mounts stained with intravascular lectin, the skin of K14-VEGF mice showed increased blood vessel density characterized by increased numbers of small capillary-sized vessels near the epidermis and surrounding the hair follicles (Fig. 2). The tortuous skin vessels of K14-VEGF mice showed leakage of the plasma tracer Evans blue under baseline conditions (Thurston et al. 1999). In addition, the basement membrane of these vessels could be labelled by intravascular perfusion of Ricin lectin, indicating defects in the endothelial barrier. Application of inflammatory stimuli to the ear skin resulted in even larger amounts of plasma leakage. The VEGF transgenic mice highlight the role of VEGF as a potent angiogenic factor, but also emphasize that the resultant vessels can be leaky and inflamed.

Fig. 2.

Morphology of blood vessels in ear skin of control, K14-VEGF, K14-Ang1, and Adeno-Ang1 mice. Vessels were stained by intravascular perfusion of biotinylated Lycopersicon esculentum lectin and DAB-peroxidase reaction, and examined in whole mounts (Thurston et al. 1999, 2000). Lectin binds to the luminal surface of endothelium and reveals blood vessels

Mice overexpressing Ang1 under the K14 promoter were also produced (Suri et al. 1998). The skin of K14-Ang1 mice was notably reddened, but the epidermis was normal in thickness and the dermis did not contain infiltrating cells. The dermal vessels were dramatically increased in diameter compared to control mice (Fig. 2), but only moderately increased in number. The enlarged vessels were in the position of capillaries subjacent to the epidermis and surrounding the hair follicles. The enlarged vessels had an increased number of endothelial cells, indicating that Ang1 increased endothelial cell proliferation or survival (Thurston et al. 1999). Unlike K14-VEGF mice, the vessels in K14-Ang1 mice were not leaky under baseline conditions and, remarkably, seemed to be resistant to plasma leakage induced by inflammatory mediators such as histamine, serotonin and mustard oil.

K14-Ang1 mice were bred to K14-VEGF mice (Thurston et al. 1999). The skin of the resultant double transgenic K14-Ang1/VEGF mice was dramatically reddened, and the vascularity of the skin was higher than either K14-VEGF or K14-Ang1 mice. The morphology of the dermal vessels appeared to be a combination of the Ang1 and VEGF effects. In particular, numerous small vessels and enlarged vessels were both present (Thurston et al. 1999). The dermis of K14-Ang1/VEGF mice was normal in thickness and did not contain infiltrating leucocytes. Furthermore, the skin vessels in K14-Ang1/VEGF mice were not leaky to Evans blue or Ricin lectin under baseline conditions (Thurston et al. 1999). Thus, Ang1 seems to inhibit some of the inflammatory actions of VEGF, but Ang1 and VEGF appear to act on distinct signalling pathways for vessel growth.

Adenoviral overexpression – Adeno-VEGF and Adeno-Ang1

Adenoviruses encoding VEGF₁₆₄, Ang1, or green fluorescent protein (GFP) as a control, driven by the cytomegalous viral (CMV) promoter were injected intravenously into mice (age 8–10 weeks) (Thurston et al. 2000). GFP was localized in the liver, confirming that most (> 95%) of the adenoviral gene expression is in hepatocytes (Yao et al. 1996; Michou et al. 1997). High levels of VEGF or Ang1 (10 µg mL⁻¹) were detected in the serum within 1 day after injection of adenovirus and, depending upon the strain of mouse, the levels remained above 500 ng mL⁻¹ for 10 days or longer.

Following injection of adeno-VEGF (1 × 10⁸ pfu or more), mice became lethargic and died within 2–3 days. Histological examination of various organs in the mice given adeno-VEGF revealed evidence of widespread oedema (Thurston et al. 2000). In contrast, mice injected with adeno-Ang1 (1 × 10⁹ pfu) appeared healthy and remained active. Similar to the K14-Ang1 mice, the blood vessels in the skin of mice given adeno-Ang1 became resistant to the plasma leakage normally induced by local injection of VEGF or topical application of mustard oil (Thurston et al. 2000). The resistance to leakage was found at 1 day after intravenous injection of adenovirus. However, unlike the K14-Ang1 mice, adult mice given adeno-Ang1 did not appear reddened, and the morphology of the skin blood vessels was normal for at least 7 days after adenoviral injection (Fig. 2)(Thurston et al. 2000). Subsequent experiments have shown that the antileakage action can be duplicated by intraperitoneal injection of Ang1 proteins (E. Joffe et al. unpublished results); thus this effect is not due to the adenoviral production of Ang1.

Discussion

Our experiments, using two approaches to overexpress VEGF and Ang1, demonstrate that these factors act separately and distinctly on blood vessels. In both overexpression systems, Ang1 resulted in vessel enlargement and resistance to leak, whereas VEGF resulted in leakiness and, if expressed locally, in vessel sprouting. The actions on blood vessel growth appear to be able to take place independently because, when given together, VEGF caused vessel sprouting and Ang1 caused vessel enlargement. However, at least in the situation where the two factors are overexpressed transgenically, the antileakage action of Ang1 appears to predominate.

In contrast to the transgenic mice in which Ang1 was overexpressed throughout development, Ang1 given to adults did not cause vessel enlargement in the skin, even when given for 50 days. Why did adenoviral expression of Ang1 not cause vessel enlargement? One possibility is that the vasculature in adults is less plastic than in embryonic and neonatal mice, and thus does not enlarge in response to Ang1. Another possibility is that Ang1 acts differently depending on whether it is delivered via the circulation or locally in the interstitium. Studies in which neonatal mice are treated with Ang1, and those in which Ang1 is inducibly overexpressed in adult mice using inducible transgenic approaches, may help shed light on this question.

Our findings using overexpression systems support and extend the studies of gene-targeted mice. Both sets of studies show clearly that VEGF and Ang1 play distinct and complementary roles in blood vessel development. Our overexpression studies help to pinpoint the distinctions, in particular highlighting the vessel enlargement and maturation actions of Ang1, and the vessel sprouting and leakiness actions of VEGF. However, it is not known whether the actions of these factors that were highlighted in our experimental overexpression systems are indeed analogous to their roles in normal development. For example, does VEGF induce leakage during normal blood vessel development? Does Ang1 have an antileakage action in development? Does coexpression of Ang1 normally prevent VEGF-induced leakage? Does Ang1 help vessels to enlarge in development? It must be borne in mind that the readouts from the assays used in experimental studies may not have a direct correlate in normal blood vessel growth. Nevertheless, our studies help define the range of actions that can be induced by endothelial-specific factors.

One clinical application of Ang1 may be to use it in combination with VEGF to grow normal, non-leaky vessels. Therapeutic application of VEGF has been tested in a number of situations to induce the growth of new blood vessels (therapeutic angiogenesis –Isner & Losordo, 1999). However, a potential side-effect of large amounts of VEGF may be the growth of leaky vessels. Indeed, some reports have suggested this may be the case (Baumgartner et al. 1998; Springer et al. 1998; Lee et al. 2000). Co-administration of Ang1 may help reduce the leak-inducing actions of VEGF without suppressing the vessel growth actions. Further studies comparing the actions of VEGF and Ang1 will undoubtedly help define the discrete steps of normal blood vessel development.