REVIEW OF THE LITERATURE

The first step in the sequence of events leading to the formation of the embryonic vasculature is the induction of the ventral-lateral mesoderm, by members of the fibroblast growth factor (FGF) and transforming growth factor-beta (TGF-β) families (Weinstein, 1999). The ventral-lateral mesoderm gives rise to cardiac, pronephric, hematopoietic and vascular endothelial tissues (Weinstein, 1999). Vasculogenesis begins by a process of in situ differentiation of hemangioblasts from mesoderm and formation of cellular aggregates or blood islands which then interconnect (Carmeliet and Collen, 1999). The cells in the periphery differentiate into early endothelial cells, whereas the central cells become hematopoietic cells (Flamme and Risau, 1992). Vasculogenesis proceeds into a second phase in which endothelial cells invade, fuse and become assembled into a primitive capillary plexus (See Figure 2 and Risau, 1997). The vascularization events intra-embryonally follow those of the yolk sac (See Figure 1). The blood vessels, which differentiate inside the embryo are connected to the yolk sac by the vitelline arteries and veins, which develop within the embryo (Wang et al., 1992). The primitive capillaries are established by intracellular lumen formation which occurs by endothelial vacuolization or by intercellular lumen formation which happens through joining of distal endothelial cells to a pre-existing lumen (Carmeliet and Collen, 1999). Studies of mutant mice reveal the importance of VEGF (Carmeliet et al., 1996; Ferrara et al., 1996), VEGFR-2 (Shalaby et al., 1995), fibronectin (George et al., 1993), the α5 integrin receptor (Yang et al., 1993), and TGF-β1 (Dickson et al., 1995) in the control of vasculogenesis. VEGFR-1 has a primary role in regulation of hemangioblast development (Fong et al., 1999) and VEGFR-1 gene inactivation results in abnormal development of the vasculature (Fong et al., 1995). The larger vessels of the embryo and the primary vascular plexus in the lung, the pancreas, the spleen, the heart and the yolk sac arise by the second phase of vasculogenesis (Wilting and Christ, 1996). Vasculogenesis which is mainly restricted to embryonic development may occur also in adult species, a suggestion supported by the finding that circulating angioblasts in the human peripheral blood are able to in vitro differentiate into endothelial cells (Asahara et al., 1997).

Fig.1.  Mouse embryo within the uterus. The extra-embryonic membranes amnion, yolk sac and Reichert's membrane are shown.

Vasculogenesis and hematopoiesis seem to proceed in concert. These events take place extra-embryonally in the yolk sac as well as intra-embryonally in the trunk intermediate cell mass (ICM) (Al-adhami and Kunz, 1977) and later in the AGM (aorta-gonad-mesonephros), liver and bone marrow (Weinstein, 1999; Medvinsky et al., 1993). The early vascular endothelial cells which have not yet formed a lumen and the hematopoietic cells formed from mesoderm, appear to be intimately associated. This close proximity appears during early ontogeny of the vertebrate blood system and is one of the indications for the existence of the common precursor cell, the "hemangioblast" (Sabin, 1920; Murray, 1932; Wagner, 1980). Other indications are the defects in both the hematopoietic and angioblastic lineages in embryos lacking VEGFR-2 (Shalaby et al., 1995). VEGFR-2 is one of the genes which are co-expressed and/or required during early stages of both blood cell and vessel development (Matthews et al., 1991; Millauer et al., 1993; Yamaguchi et al., 1993). Additionally, angioblasts and hematopoietic stem cells share certain antigenic determinants including Tie-1, Tie-2, and CD34 (Flamme and Risau, 1992; His et al., 1900; Weiss and Orkin, 1996). The existence of the hemangioblast is further supported by the ability of VEGFR-2 positive mesodermal precursors to differentiate into both, angioblastic and hematopoietic lineages in vitro (Eichman et al., 1997; Choi et al., 1998; Nishikawa et al., 1998).

Cell clusters which bud into the lumen of the aorta to start hematopoiesis do not express the endothelial specific molecule VEGFR-2 (Jaffredo et al., 1998) and acquire hematopoietic-specific markers (Vandenbunder et al., 1989; Pardanaud et al., 1996). Hematopoiesis first occurs in the blood islands and at the period of organogenesis, this potential becomes restricted to the aortic region (Pardanaud and Dieterlen-Lievre, 1993; Pardanaud et al., 1989; Jaffredo et al., 1998). Primitive erythropoiesis is suggested to be mediated by an inductive effect of endoderm on associated mesoderm (Pardanaud and Dieterlen-Lievre, 1993; Pardanaud et al., 1989, 1996). At sites with a dual hematopoietic and endothelial potential, such as the yolk sac blood islands, floor of the aorta and allantois (Caprioli et al., 1998), studies indicate that endoderm is in close contact with mesoderm (Kessel and Fabian, 1986; Flamme, 1989). Allantois vascularizes via vasculogenesis rather than by angiogenesis involving the yolk sac or fetus (See Figure 1 and Downs et al., 1998). Unlike in the chick (Caprioli et al., 1998), vasculogenesis in the mouse allantois is not associated with erythropoiesis (Downs et al., 1998). Recent experiments report that the transient contact with endoderm or ectoderm changes the behavior of mesoderm (Pardanaud and Dieterlen-Lievre, 1999). Aorta is formed by distinct endothelial lineages derived from two subpopulations of mesoderm (Pardanaud et al., 1996). The dorsal mesoderm generates endothelial precursors with only angiopoietic potential which form the roof and sides of the aorta. The ventral mesoderm or splanchnopleural mesoderm gives rise to progenitors with a dual hematopoietic and endothelial potential, which make up the floor of the aorta (Pardanaud et al., 1996).

Vascularization proceeds after the primitive embryonic vasculature is established by the budding of endothelial cells from pre-existing vessels (See Figure 2). This process called angiogenesis, is characterized by expansion of the endothelium by proliferation, migration and remodeling. Capillaries are formed by sprouting or by non-sprouting angiogenesis (Risau, 1997). Angiogenesis is distinct from vasculogenesis at the cellular level. The molecular mechanisms controlling both processes may overlap since several genes that are important for angiogenesis are also important for vasculogenesis (Folkman and D'Amore, 1996; Risau, 1997). Sprouting angiogenesis occurs extra-embryonally in the yolk sac and intra-embryonally during brain development (See Figure 1). Non-sprouting angiogenesis occurs in the lung and accounts for splitting of pre-existing vessels by transcapillary pillars. Non-sprouting angiogenesis also occurs by the intercalated growth of blood vessels, during which pre-existing capillaries merge, or additional endothelial cells fuse into existing vessels to increase their diameter and length. This type of angiogenesis takes place in the heart and during wound healing (Patan et al., 1996; Wilting et al., 1996). Yolk sac, lung and heart are sites where both sprouting and non-sprouting angiogenesis can occur (Risau, 1997). Angiogenesis is controlled by VEGF and its receptors as well as VEGFR-3 (Carmeliet et al., 1996; Ferrara et al., 1996; Dumont et al., 1998). Disruption of VEGFR-3 leads to defective large vessel formation and subsequent cardiovascular failure (Dumont et al., 1998). TGF-β signaling molecule as well as its receptor are implicated in vasculogenesis and angiogenesis (Dickson et al., 1995; Oshima et al., 1996). TGF-β signals are mediated by a family of at least nine SMAD proteins, of which SMAD5 plays a role in yolk sac angiogenesis (Yang et al., 1999; Chang et al., 1999). Ephrin-B2 is a member of the ephrin family of transmembrane ligands and it marks arteries whereas its receptor Eph-B4 is expressed in veins. Inactivation of ephrin-B2 gene shows its vital role in angiogenesis preventing the remodeling of veins and arteries into properly branched structures (Wang et al., 1998). After the primary capillary plexus has expanded, it becomes remodeled into a mature system via pruning, fusion and regression of pre-existing vessels with a resulting pattern resembling a tree of arteries, capillaries and veins (Risau, 1997; Carmeliet and Collen, 1999). During maturation the vasculature becomes surrounded by periendothelial cells, the pericytes and smooth muscle cells. These cells need to sprout in concert with endothelial cells and this is mediated by platelet-derived growth factor (PDGF)-B (Lindahl et al., 1998), Tie-2 and angiopoietin-1 (Ang-1) (Carmeliet and Collen, 1999). Recruited periendothelial cells differentiate and provide structural strength to the developing vasculature by deposition of the extracellular matrix (ECM) (Carmeliet and Collen, 1999).

Angiogenesis is a dynamic multistep process, which involves retraction of pericytes from the abluminal surface of the capillary, release of proteases from the activated endothelial cells, degradation of the ECM surrounding the pre-existing vessels, endothelial cell migration toward an angiogenic stimulus and their proliferation, formation of tube-like structures, fusion of the formed vessels and initiation of blood flow (Bicknell, 1997). Unlike vasculogenesis, angiogenesis also occurs during adult life, in the female reproductive system, wound healing, tissue repair and in many pathological conditions. Angiogenesis depends on the adhesion receptor integrin αvβ3, allowing endothelial cells to interact with a variety of extracellular matrix components and leading them to migrate (Leavesley et al., 1993; Brooks et al., 1994; Cheresh et al., 1991). Besides being the most important survival system for nascent vessels by regulating cell adhesion to matrix, αvβ3 participates in the full activation of VEGFR-2 triggered by VEGF (Soldi et al., 1999). Cellular invasion depends on cooperation between adhesion as well as proteolytic modification of the ECM. Evidence suggests that the matrix metalloproteinase (MMP-2) is localized in a proteolytically active form on the surface of invasive cells, based on its ability to bind directly αvβ3 (Brooks et al., 1996). Vascular endothelial VE-cadherin, mediates adhesion between endothelial cells, and participates in their organization to vessel-like structures (Vittet et al., 1997). Vascular defects appear in VE-cadherin negative endothelium when the primitive vascular network expands via sprouting angiogenesis but not during vasculogenesis (Vittet et al., 1997). Deficiency or truncation of VE-cadherin blocks the capacity of endothelial cells to respond to survival signals induced by VEGF (Carmeliet et al., 1999).

Fig. 2. Different phases of embryonic vascular development. Endothelial precursors (angioblasts) differentiate to early endothelial cells (phase 1), which become assembled into a primitive capillary plexus (vasculogenesis) (phase 2). This emerging network expands via intussusceptive growth, intercalated growth and sprouting (angiogenesis) (phase 3), after which it becomes remodelled via pruning, fusion and regression of pre-existing vessels into a tree of arteries, capillaries and veins (phase 4). Endothelial cells further differentiate and acquire specific properties such as the formation of a tight barrier in the brain, or the formation of fenestrations in exocrine glands. Adapted from Carmeliet and Collen (1999).

Angiogenesis is a complex series of interdependent events, controlled by many different factors. The angiogenic process results from a shift in the balance of positive and negative mediators, cytokines and growth factors being the primary inducers (Hanahan and Folkman, 1996). A variety of direct positive regulators of angiogenesis are identified and some of them are mentioned in the following text. Besides, there are factors which indirectly induce angiogenesis for example by the mobilization of macrophages, that could release growth factors (Baird et al., 1985), or chemotactic factors for vascular endothelial cells (Banda et al., 1982) or both. Indirect angiogenic factors may also direct the release of stored angiogenic factors from the ECM or the basement membrane (BM) (Folkman and Klagsbrun, 1987).

The ECM plays an important role in inducing an angiogenic response. BMs and ECMs are the substrates upon which cells migrate, proliferate and differentiate in vivo. Heparan sulfate proteoglycans (HSPGs) are very important for the binding of growth factors and enzymes that need to be immobilized and stored in the ECM for combined action with BM macromolecules like collagen IV, laminin, nidogen/entactin, and proteoglycans during cellular responses to the ECM (Roberts et al., 1988; Ruoslahti and Yamaguchi, 1991; Vlodavsky et al., 1993; Wight et al., 1992). Heparan sulphate (HS) also contributes to the integrity and assembly of the ECM by binding to its fibrillar interstitial collagens, fibronectin, laminin, thrombospondin, and tenascin as well as to different attachment sites on plasma membranes. HSPGs are also components of blood vessels (Jackson et al., 1991), supporting proliferating and migrating endothelial cells as well as stabilizing the capillary wall.

One class of angiogenic factors includes polypeptides like acidic fibroblast growth factor (aFGF) and basic fibroblast growth factor (bFGF). Both, aFGF and bFGF act as direct angiogenesis factors, expressed by many tumors in vivo and tumor cell lines in vitro (Christofori, 1997). Low levels of bFGF are detected in the conditioned medium of endothelial cells (Schweigerer et al., 1987; Sato and Rifkin, 1988) and some tumor cell lines (New and Yeoman, 1992; Yamada et al., 1992; Sato et al., 1989). These two factors have served for a long time as the prototype angiogenic factors, being in vitro mitogenic and chemotactic for endothelial cells and inducing capillary endothelial cells to migrate into three-dimensional collagen matrices to form capillary-like tubes (Burgess and Maciag, 1989; Basilico and Moscatelli, 1992). Both factors induce angiogenesis in vivo in the chick chorioallantoic membrane (CAM) and cornea bioassays (Christofori, 1997). FGFs stimulate endothelial cells to secrete collagenase and plasminogen activator proteases able to degrade BM (Christofori, 1997). Binding of FGFs to the high affinity FGFRs is facilitated by their binding to heparin or heparin-like molecules which then present FGFs to their high affinity receptors (Yayon et al., 1991). FGF is sequestered in the ECM, BM, or on the cell surface upon binding to the heparan sulphate side chains of low-affinity, but high-capacity receptors (Klagsbrun, 1990; D'Amore, 1990). Heparin enhances the mitogenic action of bFGF (Mansukhani et al., 1992) and protects aFGF and bFGF from degradation by heat, acid and proteases (Gospodarowicz and Cheng, 1986; Lobb, 1988; Rosengart et al., 1988). bFGF can be mobilized in a biologically active form from the ECM and BM by displacement with heparin or by degradation with heparanases and proteinases (Folkman et al., 1988; Saksela et al., 1988; Vlodavsky et al., 1987). The role of bFGF in vivo was determined by generation of bFGF knockout mice. These mice are viable, fertile and phenotypically indistinguishable from the normal littermates. However, the mice display abnormalities in cortical neurogenesis with reduction in neuronal density and have delayed wound healing (Ortega et al., 1998).

Another potent endothelial cell mitogen and angiogenesis inducer is VEGF and it is devoid of consistent and appreciable mitogenic activity for other cell types (Leung et al., 1989; Tisher et al., 1991). VEGF promotes angiogenesis in  threedimensional in vitro models, inducing microvascular endothelial cells to invade collagen gels and form capillary-like structures (Pepper et al., 1992). Studies provide evidence for a potent synergism between VEGF and bFGF in the induction of this effect (Pepper et al., 1992). VEGF induces sprouting from rat aortic rings embedded in collagen gel (Nicosia et al., 1994). Furthermore, VEGF elicits a strong angiogenic response in a variety of in vivo models including the CAM (Leung et al., 1989; Plouet et al., 1989) the rabbit cornea (Phillips et al., 1995), the primate iris (Tolentino et al., 1996), and the rabbit bone (Connolly et al., 1989). Over the last few years, several members of the VEGF gene family were identified, including placenta growth factor (PlGF) (Maglione et al., 1991), VEGF-B (Olofsson et al., 1996a), VEGF-C (Joukov et al., 1996; Lee et al., 1996), VEGF-D (Orlandini et al., 1996; Yamada et al., 1997; Achen et al., 1998;) and VEGF-E (Lyttle et al., 1994; Ogawa et al., 1998; Meyer et al., 1999). Three alternative splice isoforms are identified for PlGF, from which the large isoform PlGF-2, differs from the other two PlGF forms by the insertion of a highly basic 21-amino acid stretch at the carboxyl-end of the protein (Hauser and Weich, 1993). PlGF-1 induces neovascularization in the rabbit cornea and the CAM assay either as homodimers or as heterodimers with VEGF₁₆₅ (Ziche et al., 1997; Oh et al., 1998). PlGF-1 induces proliferation and chemotaxis of human umbilical vein endothelial cells (HUVEC) and capillary venule endothelial cells (CVEC) (Ziche et al., 1997). PlGF-2 stimulates mitogenesis of HUVEC (Sawano et al., 1996). VEGF-B stimulates endothelial cell proliferation of HUVE and bovine capillary endothelial (BCE) cells, in one experimental setting but the results are not reproduced in additional trials with pure recombinant factors (Olofsson et al., 1996a). VEGF-C stimulates migration and mitogenesis of cultured endothelial cells (Joukov et al., 1996; Lee et al., 1996; Joukov et al., 1997; Cao et al., 1998). In vivo, VEGF-C induces lymphangiogenesis in the differentiated CAM (Oh et al., 1998), whereas it stimulates angiogenesis in the early CAM (Cao et al., 1998). In addition, VEGF-C promotes angiogenesis in the mouse cornea and in a setting of ischemic hindlimb in rabbits (Cao et al., 1998; Witzenbichler et al., 1998). VEGF-D is mitogenic for microvascular endothelial and bovine aortic endothelial cells (BAEs) (Achen et al., 1998; Orlandini et al., 1996). In addition, treatment with VEGF-D induces the formation of an extensive network of capillary-like cords in a three-dimensional matrix (Marconcini et al., 1999). VEGF-D is an angiogenic factor in rabbit cornea in vivo in a dose-dependent manner (Marconcini et al., 1999). VEGF-E stimulates endothelial cell mitogenesis and migration in vitro (Meyer et al., 1999). VEGF-E induces angiogenesis in vivo, in the rabbit cornea (Meyer et al., 1999), and when injected subcutaneously in matrigel (Ogawa et al., 1998).

Transforming growth factor-alpha (TGF-α) is an angiogenic factor released by tumor associated macrophages (Madtes et al., 1988; Rappolee et al., 1988) and many tumor cells (Todaro et al., 1985). TGF-α is structurally and functionally related to epidermal growth factor (EGF) released by macrophages and both induce endothelial cell DNA synthesis in vitro (Schreiber et al., 1986). TGF-α also stimulates capillary-like tube formation in collagen gels (Sato et al., 1993). The ability of TGF-α  to promote tube formation is dependent on the release by endothelial cells of substances like tissue-type plasminogen activator (tPA) and FGF (Sato et al., 1993). In vivo, TGF-α   is more potent than EGF in promoting angiogenesis in the hamster cheek pouch bioassay (Schreiber et al., 1986). In addition, TGF-α  and EGF are potent promoters of neoplastic cell growth in many carcinomas (Bicknell et al., 1991).

TGF-β which is not structurally related to TGF-α is synthesized by many normal and tumor cells (Anzano, 1982; Derynck, 1985). TGF-β stimulates neovascularization in vivo when subcutaneously injected to mice (Roberts et al., 1986). However, several studies utilizing two-dimensional culture conditions find microvascular endothelial cell proliferation to be inhibited by TGF-β1, TGF-β2 and TGF-β3 raising doubt as to whether TGF-β functions as an angiogenic factor. It is possible that TGF-β promotes angiogenesis by differentiating endothelial cells, or stimulates angiogenesis indirectly (Pertovaara et al., 1994; Mervin et al., 1991; Merwin et al., 1991). Studies also show that low concentrations of TGF-β1 enhance endothelial cell proliferation (Myoken et al., 1990) and in vitro angiogenesis (Pepper et al., 1993), and high concentrations inhibit proliferation and angiogenesis.

TNF-α is a cytokine secreted by macrophages at sites of inflammation and is toxic to tumor cells (Pusztai et al., 1994). TNF-α acts on endothelial cells to exert a procoagulant effect (Fajardo, 1989; van der Poll et al., 1990). TNF-α inhibits proliferation and antagonizes the mitogenic effect of bFGF in HUVE cells (Frater-Schroder et al., 1987). TNF-α  also inhibits the proliferation of cultured capillary endothelial cells (Schweigerer et al., 1987). However, TNF-α does not inhibit mitogenesis of dermal and lung microvascular endothelial cells (Detman et al., 1990; Meyrick et al., 1991). These differences may be due to the different concentrations of cytokines used, or to the difference in the sensitivity of endothelial cells originating from distinct locations. Otherin vitro and in vivo studies reveal that TNF-α stimulates angiogenesis at low doses, whereas it inhibits angiogenesis at high doses (Fajardo et al, 1992). In tissues in vivo the concentration of TNF-α    is low so that the overall effect of TNF-α  is stimulatory to angiogenesis (Klagsbrun and D'Amore, 1991). Additional reports show that TNF-α stimulates endothelial cell migration and tube formation, but inhibits proliferation of these cells. TNF-α stimulates angiogenesis in the rabbit cornea and in the CAM (Leibovich et al., 1987; Frater-Schroder et al., 1987). Hypoxia induces in vitro the release of TNF-α by macrophages (Scannell et al., 1993).

Granulocyte-colony stimulating factor (G-CSF) and granulocyte/macrophage-colony stimulating factor (GM-CSF) are myeloid growth factors required for the survival, growth, and differentiation of hemopoietic precursor cells. Both are produced in many different tumor types (Bottazzi et al., 1985; Fu et al., 1992) and are chemotactic for macrophages in vitro (Wu et al, 1993; Pyke et al., 1991). Recombinant G-CSF and GM-CSF stimulate the migration and proliferation of human endothelial cells (Bussolino et al., 1989).

Insulin-like growth factor I (IGF-I) is secreted by activated macrophages and it is suggested to have a role in inflammatory angiogenesis  (Filkins, 1980; Rom et al., 1988). IGF-I induces endothelial cell mitogenesis, migration and tube formation (Nakao et al., 1992). Interleukin-8 (IL-8) is produced by activated macrophages (Koch et al., 1992; Yoshimura et al., 1987) and is mitogenic in vitro and stimulates angiogenesis in the rat corneal assay (Hu et al., 1993; Strieter et al., 1992). Hepatocyte growth factor/scatter factor (HGF/SF) is a mesenchyme-derived cytokine that stimulates motility and invasiveness of epithelial and cancer cells. HGF/SF is a potent angiogenic molecule in vivo and its angiogenic activity is mediated primarily through direct actions on vascular endothelial cells. These include stimulation of cell migration, proliferation, protease production, invasion, and organization into capillary-like tubes (Rosen et al., 1997). Nonpeptide substances like prostaglandins PGE1 and PGE2, stimulate microvascular endothelial cell proliferation and angiogenesis in vivo (Fajardo et al., 1992). The soluble adhesion molecules E-selectin and VCAM which are involved in inflammatory diapedesis, also induce angiogenesis in vivo in the rabbit cornea (Gasparini, 1994). Additionally E-selectin and VCAM are chemotactic for HUVEC in a Boyden chamber assay (Gasparini, 1994). In the same studies antibodies to the known ligands of E-selectin and VCAM, block HUVEC cell migration.

The angiopoietins have recently joined the members of the VEGF family as the only known growth factors largely specific for the vascular endothelium. Ang-1  is an angiogenic factor that signals through the endothelial cell-specific Tie-2 receptor receptor tyrosine kinase (Davis et al., 1996). Ang-1 is essential for normal vascular development in the mouse. An Ang-1 relative, termed angiopoietin-2 (Ang-2), acts as a naturally occurring antagonist for Ang-1 and Tie-2 (Maisonpierre et al., 1997). Ang-1 has a role during later stages of vascular development. Coming into play after VEGF, it enhances vessel branching and remodeling and promotes maturation and stabilization of vessels (Gale and Yancopoulos, 1999). Ang-1 is potent in inducing sprouting angiogenesis of adrenal-cortex–derived microvascular endothelial cells in vitro in three-dimensional fibrin gels and it is only a weak mitogen for HUVEC (Koblizek et al., 1998). Instead, Ang-2 must act as a destabilizing protein being expressed at sites of vascular remodeling leading to vessel regression in the absence of VEGF or causing vessel outgrowth in the presence of VEGF (Maisonpierre et al., 1997).

Heparin was first found to induce endothelial cell migration and proliferation in vitro (Fraser et al., 1983), but it was then shown that it is not angiogenic in vivo by itself but potentiates the angiogenic activity of tumor cell extracts (Damon et al, 1989). Species of heparin and HS promote dimerization and receptor binding of bFGF (Yayon et al., 1991; Ornitz et al., 1992), as well as bFGF-mediated mitogenesis (Ornitz et al., 1992),  suggesting that bFGF-HSPG complexes serve as the biological active form of this growth factor. Cell surface heparin-like molecules are also involved in binding of VEGF to its high affinity receptor sites (Gitay-Goren et al., 1992). Released FGF can be sequestered from its site of action by binding to HS in the ECM (Vlodavsky et al., 1991; Bashkin et al., 1989) and saved for emergencies, such as wound repair and neovascularization (Vlodavsky et al., 1993).

Abbreviations: (In): indirect angiogenic factor, DNDC: mature form of VEGF-C or VEGF-D.

In vitro angiogenic assays are market by numbers: proliferation=1, migration=2, tube formation=3, sprouting from microcarrier beads in fibrin gel=4, sprouting from rat aortic ring=5. In vivo angiogenic assays are marked by numbers: rabbit bone= 6, CAM=7, cornea pocket=8, limb ischemia model=9, subcutaneous matrigel implantation=10, hamster cheek pouch assay=11, rabbit ear lobe=12.

There are a number of peptide growth regulators, which inhibit vascular endothelial cell proliferation. Many of these peptide growth factors have both stimulatory and inhibitory activity on a single cell, depending on the context of the other signaling molecules present (Sporn and Roberts, 1988). Thrombospondin, an extracellular matrix protein, suppresses neovascularization in vivo in the rat cornea (Rastinejad et al., 1989) and also inhibits endothelial cell proliferation, migration and angiogenesis in vitro (Bagavandoss and Wilks, 1990; Dawson et al., 1997; Good et al., 1990; Iruela-Arispe et al., 1991; Rastinejad et al., 1989; Taraboletti et al., 1990; Tolsma et al., 1997; Tolsma et al., 1993; Volpert et al., 1995). Another factor able to down-regulate or inhibit angiogenesis is INF-α, being potent in regressing hemangiomas (Ezekowitz et al., 1992), and highly vascular Kaposi's sarcomas (Real et al., 1986). INF-γ inhibits aFGF-induced endothelial cell proliferation (Friesel et al., 1987) capillary formation (Tsuruoka et al., 1988) and endothelial cell growth in vitro (Heyns et al., 1985). INF-γ is implicated as an anti-angiogenic agent in vivo (Orchard et al., 1989; White et al., 1989). Platelet factor 4 is a strong anti-angiogenic factor in vitro (Taylor and Folkman, 1982; Walz and Hung, 1985; Jouan et al., 1999), and inhibits growth of solid tumors (Sharpe et al., 1990). TGF-β1 induces cells to secrete thrombospondin, in this way enhancing its own effects (Canfield and Schor, 1995; Penttinen et al., 1988). Protamine is an inhibitor of angiogenesis in the embryo, in tumorigenesis, in inflammation and immune responses (Taylor and Folkman, 1982).

Angiostatin and endostatin inhibit specificly endothelial cell proliferation and are derived from the proteolytic cleavage of native proteins. Angiostatin an internal fragment of plasminogen produced by primary tumors, was identified as an endogenous inhibitor of endothelial cell proliferation and inhibits angiogenesis in vivo. When the primary tumor is present, it in certain cases suppresses metastatic growth by secreting this circulating angiogenesis inhibitor. After tumor removal, metastases neovascularize and grow (O'Reilly et al., 1994; Holmgren et al., 1995). Endostatin, a C-terminal fragment of collagen XVII, is another endogenous inhibitor of angiogenesis. It suppresses endothelial cell proliferation in vitro and in vivo angiogenesis in the CAM (O'Reilly et al., 1997).

Several other endogenous inhibitors of angiogenesis have been identified, such as an interferon-inducible protein. It is able to inhibit angiogenesis in vivo, but has no effect on endothelial cell growth, attachment, and migration in vitro (Angiolillo et al., 1995). Two chemokines, gro-α and gro-β, specifically inhibit growth factor stimulated proliferation of capillary endothelial cells and gro-β is potent in inhibiting in vivo blood vessel formation in CAM (Cao et al., 1995). 16-kDa N-terminal fragment of prolactin is a potent inhibitor of angiogenesis (Clapp et al., 1993). Steroids administered in combination with heparin inhibit angiogenesis in the CAM, and in rabbit corneas (Folkman et al., 1983; Nikolic et al., 1986). The nonanticoagulating derivative of heparin, heparin adipic hydrazide (HAH), covalently linked to the antiangiogenic steroid, cortisol, has an antiproliferative effect on murine pulmonary capillary endothelial cells (Thorpe et al., 1993).

Collagenase inhibitors inhibit angiogenesis, which may depend on the importance of BM synthesis and degradation in the formation of new capillaries (Folkman and Ingber, 1992). Other protease inhibitors like plasminogen activator inhibitor 1 (PAI-2) or 2 (PAI-1), and tissue inhibitors of metalloproteinases (TIMP-1 and TIMP-2) inhibit the degradation of BMs and endothelial cell invasion of the ECM (Fan et al., 1995). Simultaneous expression of enzyme and inhibitor stimulates neovascularization, while protecting newly vascularized tissues. Endothelial cells secrete proteolytic enzymes such as plasminogen activator (PA) enabling them to penetrate the BM and migrate into surrounding tissues (van et al., 1991). Macrophages secrete enzyme inhibitors and produce cytokines that alter enzyme secretion by endothelial cells.

Recently a fragment of MMP-2, which comprises the C-terminal hemopexin-like domain, termed PEX, was shown to prevent the binding of this enzyme to αvβ3 and block cell surface collagenolytic activity (Brooks et al., 1998). PEX blocks MMP-2 activity on the CAM where it disruptsangiogenesis and tumor growth (Brooks et al., 1998).

In vitro angiogenic assays are market by numbers: proliferation=1, migration=2, tube formation=3, sprouting from microcarrier beads in fibrin gel=4, sprouting from rat aortic ring=5. In vivo angiogenic assays are marked by numbers: rabbit bone= 6, CAM=7, cornea pocket=8, limb ischemia model=9, subcutaneous matrigel implantation=10, hamster cheek pouch assay=11, rabbit ear lobe=12.

The lymphatic vessels penetrate most of the tissues as a dense vascular network. Lymphatics control the microcirculation of tissues, by draining fluid from the interstitial spaces which is then being filtered by lymph nodes and returned to the systemic circulation through the thoracic and lymphatic ducts and the lymphaticovenous anastomoses (Leak, 1970). They also take part in the immune functions, inflammation and tumor metastasis. The lymphatic vascular system is not continuous like the blood vascular system, being rather a one-way passage channel. The lymphatic vasculature has evolved in response to high pressure and it bridges the disparity between the intracellular and extracellular environments (Stone and Hugo, 1972). The mechanism of molecular exchange between plasma and interstitial fluid is based on diffusion, rather than hydrodynamic flow like for blood vessels (Pappenheimer, 1953).

Lymphangiogenesis has many similarities but also differences with angiogenesis. The principal difference is that the lymphatic vessels develop much after the formation of veins and arteries. The first lymphatic plexuses accompany veins but later they also grow along arteries (Wilting et al., 1999). Lymphatic endothelial cells seem to derive only from venous endothelium, and new lymphatic endothelial cells form from pre-existing ones (Wilting et al., 1999). The thoracic duct endothelium is able to form new lymphatic-like capillaries, in collagen gel (Clark and Clark, 1932; Nicosia, 1987). Lymphatic vessels are found everywhere in the body except in the central nervous system, the bone marrow, the intralobular portion of the liver, the eye, the internal ear, and the fetal placenta (Bailey, 1964). Similar mechanisms operate in lymphangiogenesis and angiogenesis, such as sprouting, intussusceptive and intercalated growth, fusion and finally regression (Wilting et al., 1999). Studies on the lymphatic system have been fragmentary compared to the blood vascular system. Three theories are presented to explain the origin of lymphatic vessels.

The first theory suggested by Dr Sabin, proposes formation of primary lymphatic structures from sprouts of large central veins in certain locations of the body. These sprouts form the primordial lymph sacs by fusion, which then give rise to the primary lymph nodes. The sinuses of the lymph nodes form by a process resembling intussusceptive growth. In mammals the lymphatic sacs arise from the anterior cardinal veins in the neck or from the inferior vena cava and the adjacent veins of the kidney. The primary lymphatic sacs and the thoracic duct make the primary lymphatic system (Sabin, 1909; Sabin, 1902; Sabin, 1912). Based on this theory of centrifugal spread, the primary sacs enlarge and form sprouts into the periphery of the embryo (Zadvinskis et al., 1992). The lymphatic network seems to further grow by intercalated growth (Wilting et al., 1999). According to the second, so called centripetal theory, lymphatics develop from lymphangioblasts in blind mesenchymal spaces that spring up either around degenerating venous channels or in their vicinity but independently of them. These lymphatic spaces spread from the periphery of the embryo by annexing with other similar spaces and establishing anastomoses with the venous system (Kampmeier, 1960). The third theory combines the two previous theories and supports the venous-mesenchymal origin, suggesting that lymphatics arise from the confluence of small venules and from mesenchymal aggregates (Kutsuna, 1933; van der Jagt, 1932). The fact that the lymph plexuses are located adjacent to veins and are connected to them (Sabin, 1909; van der Putte, 1975), supports the venous origin of the lymphatics. Since lymphatics also develop late in embryogenesis the theory of mesenchymal origin appears less favorable (Wilting et al., 1999).

Endothelial cells lining the blood and lymphatic vessels display unique phenotypes according to the specific functions that they perform. The various regions of the vascular plexus have different intercellular junctions that regulate vascular permeability, leukocyte extravasation and vascular proliferation. Different molecules at junctions are not only important for maintaining adhesion but can also play a role in cell-to-cell signal transduction. On the basis of morphological and functional characteristics at least three types of junctions are described in endothelial cells. Large arteries have a high frequency of tight junctions, which form a very close contact between adjacent cells. In the endothelium of postcapillary venules tight junctions are virtually absent (Dejana et al., 1995). Gap junctions are clusters of transmembrane hydrophilic channels that allow direct exchange of ions and small molecules between adjacent cells (Dejana et al., 1995). Cell-to-cell adherens junctions are cellular membrane contacts formed by cadherins as transmembrane glycoproteins that mediate the physical attachment between cell membrane and an intracellular undercoat network of cytoplasmic proteins and actin microfilaments (Dejana et al., 1995).

Endothelial cell junctions can be modulated in the angiogenic process, possibly by vascular growth factors or inflammatory cytokines affecting the expression or phosphorylation state of the junctional proteins (Dejana et al., 1995). During the formation of vascular sprouts endothelial cells need to detach and invade the surrounding tissues, which is accomplished by the interruption of junctions. VEGF also acts as a permeability factor for the vasculature, possibly decreasing the endothelial junction strength (Connolly et al., 1989).

Morphological criteria divide the microvascular endothelium to continuous, fenestrated and discontinuous, correlating with vascular permeability properties (Bennet et al., 1959; Majno, 1965). Continuous endothelium forms the blood-brain barrier, with "tight" junctions between the capillary endothelial cells that allow only very small molecules to pass into the brain. Discontinuous endothelium is found in the liver, where the clefts between the capillary endothelial cells are wide open, so that almost all dissolved substances of the plasma, including the plasma proteins, pass from the blood into the liver tissues. Fenestrated endothelium is located to the glomerular tufts of the kidney, in which numerous small oval windows called fenestrae penetrate directly through the middle of the endothelial cells, so that tremendous amounts of substances can be filtered through the glomeruli without having to pass through the clefts between the endothelial cells (Risau, 1995).

There are also differences between the endothelia of large vessels and the microvasculature. The endothelium of large vessels regulates vascular tone and blood pressure and the microvasculature is involved in neovascularization participating as well in the exchange of oxygen and nutrients (Kumar et al., 1987). The endocardium, which is a specialized type of large vessel endothelium, is the first endothelium to be formed. It differentiates from a more axial mesoderm, and is the only endothelium to trans-differentiate to mesenchyme during the formation of the endocardial cushion tissue (Sinning et al., 1922).

The lymphatic endothelium differs from the blood vascular endothelium in many morphological aspects. Lymphatics have an irregular and wide lumen, and an attenuated endothelial wall. Except for the collecting lymphatic capillaries lymphatics do not develop a continuous BM (Casley-Smith, 1980; Oh et al., 1997). Lymphatic vessels lack laminin and type IV collagen (Barsky et al., 1983; Ezaki et al., 1990). Their endothelial cells lack tight junctions and they posses anchoring filaments attaching them to the surrounding connective tissue (Leak, 1970). Because of their permeability and their poorly developed BM, lymphatics have been proposed as the main routes for tumor metastasis (Cann et al., 1995).

Altered endothelial permeability and organization of the vasculature is associated with hemangiomas, atherosclerosis and inflammatory diseases (Dejana et al., 1995). Abnormal function of the lymphatic system is implicated in many disease states such as lymphedema, ascites, inflammation, infectious and immune diseases, fibrosis and tumors such as Kaposi's sarcoma and lymphangioma/lymphangiomatosis (Goldberg and Rosen, 1997). In lymphedema, the lymphatic system fails to remove the plasma proteins passing into the tissues, which causes edema (Casley-Smith, 1984). Lymphangioma is a vascular malformation containing lymph with a defect in the deep collecting lymphatics (Cotran et al., 1994).

Most adult tissues have little ongoing angiogenesis. The endothelial cell proliferation is high during embryonic and postnatal development, but adult endothelial cell turnover is very low (Risau, 1997). The endothelium is in a quiescent state during the entire life of a normal adult, yet is able to proliferate in pathological conditions (Risau, 1995). Angiogenesis taking place in the adult may resemble embryonic angiogenesis including remodeling of cell adhesions and junctions, induction of proteolytic processes and neutralization of inhibitors (Risau, 1997). Regression occurs in the capillaries of the primitive vascular plexus during embryogenesis so that few of them persist until adulthood (Risau, 1995). During adult life, physiological angiogenesis occurs in the female reproductive cycle including the development of the ovarian follicle and corpus luteum, and in the endometrium during each menstrual cycle (Hansen-Smith, 1988). Neovascularization plays a major role in processes such as wound healing, tumor growth and metastasis, rheumatoid arthritis, diabetic proliferative retinopathy, atherosclerosis, psoriasis, and many others.

The ovary is one of the adult organs, in which angiogenesis normally occurs during the female reproductive cycle. The development and endocrine function of the ovarian corpus luteum (CL) are dependent on the growth of new capillary vessels. Progesterone release by CL is essential for implantation and maintenance of pregnancy (Bassett et al., 1943). During follicular growth, the theca interna becomes richly vascularized. Following ovulation, the thecal vessels invade the ruptured follicle and form a microvascular plexus that nourishes the developing CL (Bassett et al., 1943). Multiple mediators are involved in CL angiogenesis including VEGF, bFGF (Ferrara et al., 1998; Gospodarowicz et al., 1985), Ang-1 and Ang-2 (Maisonpierre et al., 1997). VEGF or a related factor is essential for CL angiogenesis, since treatment with soluble receptor VEGFR-1 extracellular domain results in complete suppression of CL angiogenesis (Ferrara et al., 1998). At the same time endometrium fails to mature, probably reflecting suppression of ovarian steroid production and an inhibition of locally produced VEGF (Ferrara et al., 1998).

Angiogenesis during wound healing is a good example of pathophysiological neovascularization during adult life, where the capillaries regress after the termination of the process. Angiogenesis is part of the normal wound repair process, where it provides a supply of nutrients, promotes granulation tissue formation and clears debris. Neovascularization in wounds depends on cell-cell interactions, cell-ECM interactions and also the balance between angiogenic agonists and antagonists. Tissue injury is followed by exudation of plasma constituents including fibrinogen. Fibrinogen provides the substrate for the generation of a fibrin-containing matrix, which is subsequently replaced by granulation tissue (Schäfer et al., 1994). Replacement involves proteolysis by plasmin, which is generated from plasminogen present in plasma and the interstitial fluid (Miyashita et al., 1988). PAs convert plasminogen into plasmin by limited proteolysis (Dano et al., 1985). Polymorphonuclear leukocytes (PMNs), monocytes, fibroblasts and capillary endothelial cells are involved in the organization of the primary wound matrix (Dvorak et al., 1988). In vitro, these cells expressPAs (Bajpai and Baker 1985; Granelli-Piperno et al., 1977; Chapman et al., 1982; Hekman et al., 1985), surface receptors for urokinase plasminogen activator (uPA) (Blasi et al., 1986; Bajpai and Baker, 1985; Fibbi et al., 1988) and cell surface binding sites for plasminogen (Miles and Plow, 1987; Hajjar et al., 1986).

Many in vitro and in vivo studies suggest the involvement of bFGF in the angiogenic activity during wound healing. bFGF can be sequestered within cells and these stores can be released during cell injury, lysis or death (Gajdusek and Carbon, 1989; Muthukrishnan et al., 1991). Proteases found in the early wounds are also able to release bFGF stored in the ECM (Bashkin et al., 1989; Saksela et al., 1990; Villaschi and Nicosia, 1993). Platelets at the wound site participate in coagulation and release several growth factors and enzymes. Platelet derived growth factor (PDGF) for example is a powerful chemoattractant for circulating neutrophils, monocytes and tissue macrophages (Ondrick and Samojla, 1992). By releasing bFGF and TNF-α, macrophages are able to exert angiogenic activity in the depths of a wound where oxygen tension is low and lactate is high due to anaerobic metabolism. For example phagocytosis of fibrin increases macrophage angiogenic activity (Dvorak, 1986; Knighton et al., 1983). During the latter stages of wound repair, macrophages stimulated by hypoxia, cytokines, or inflammatory mediators, synthesize new factors like TGF-β and VEGF (Knighton et al., 1983). During the terminal stages of healing the production of angiogenic factors is decreased when granulation tissue is formed and the area is less hypoxic (Constant et al., 1996; Hunt et al., 1978).

Hypoxia is known to increase VEGF levels but whether it stimulates wound angiogenesis is an open question. VEGF, produced by keratinocytes together with other mediators may play a role in wound angiogenesis (Brown et al., 1992; Frank et al., 1995). VEGF is produced in many in vitro and in vivo systems under hypoxic conditions, supporting the view of VEGF production in hypoxic wounds (Shweiki et al., 1992).

In the retina, neovascularization is associated with a number of disease processes, the most common of which is diabetic retinopathy. Occlusion of retinal vessels leading to retinal ischemia is a feature shared by most diseases in which retinal neovascularization occurs. Hypoxia-induced expression of VEGF by retinal cells is a common factor in these diseases (Pe'er et al., 1995). VEGF is a survival factor for the endothelial cells in the embryonic retina and hyperoxia inhibits VEGF expression, leading to regression and death of retinal vessels, which can be rescued by intaocular injection of the factor (Alon et al., 1995). It is not clear, whether hyperoxia induces blood vessel regression under normal conditions, VEGF is instead critical for the survival of immature/remodeling vessels. Independence from VEGF is a sign of maturation when vessels have already acquired a pericyte coating (Benjamin et al., 1998).

2.4. Tumor angiogenesis

Tumors or metastases growing initially as small avascular masses need to gain a blood supply, which permits them to grow beyond a few millimeters in diameter. Tumors implanted into isolated perfused organs fail to develop, but when these tumors are implanted into the ocular chambers of mice they grow. Tumors in the second case obtain oxygen and nutrients from the organism, by establishing a blood supply necessary for their growth (Folkman et al., 1963; Gimbrone et al., 1973; Gimbrone et al., 1972). The induction of angiogenesis is mediated by many angiogenic molecules released by tumors and host cells (Ellis and Fidler, 1995). Activation of endothelial cell proliferation and migration can be achieved by reduction of inhibitor levels or increase of activators (Hanahan and Folkman, 1996).

The most widely accepted theory of initial avascular growth of tumors and metastases may not be valid in all the settings. Avascular growth occurs in tumors that arise in epithelial structures that are separated from the underlying vasculature. However, tumors that arise within or metastasize to vascularized tissue, coopt existing blood vessels. This coopted host vasculature does not immediately undergo angiogenesis to support the tumor but instead regresses, leading to a secondarily avascular tumor and massive tumor cell loss. Finally, the remaining tumor is rescued by angiogenesis at the tumor margin (Holash et al., 1999). The angiogenic antagonist Ang-2 and pro-angiogenic VEGF may be critical regulators of the balance between vascular regression and growth. This is supported by the increased expression of Ang-2 in the endothelium of coopted tumor vessels and induced VEGF expression in the hypoxic tumor periphery (Maisonpierre et al., 1999). Thus, the angiogenic properties of tumor-derived VEGF may be facilitated when vessels are destabilized by Ang-2 (Holash et al., 1999). In established human tumors because of ongoing angiogenesis and remodeling of tumor vessels, there is a significant fraction of vessels that have not yet recruited periendothelial cells. The dependence of these immature vessels on vascular survival factors leads to their regression following VEGF withdrawal (Benjamin et al., 1999). Yet, very recent findings describe a new mechanism by which some aggressive tumors may acquire a blood supply. According to these findings which still need to be verified tumor cells metamorphose into vessels that either carry blood or connect to the host's blood supply (Bissell, 1999). The generation of patterned vascular channels by deregulated aggressive tumor cells is observed in human melanomas in which the establishment of a microcirculation also facilitates metastasis (Maniotis et al., 1999).

Tumors have a complex morphology, differing in regional vasculature, host infiltrates, and connective tissue components (Callahan and Campbell, 1989; Heppner, 1984). Tumor vessels are anatomically heterogeneous structures, being relatively immature. They are lined by a simple endothelium, and have fewer pericytes and smooth muscle cells compared to normal tissue vessels. Tumor vessels are also more permeable to circulating plasma proteins and have differing capacities to leak macromolecules depending on their location (Dvorak et al., 1988).

Neovascularization takes place following different pathways, where tumor cells either synthesize their own angiogenic factors (Klagsbrun et al., 1986), or attract macrophages which release them (Polverini and Leibovich, 1984). Tumors secrete collagenases and heparanases (Kramer et al., 1982) and degrade the ECM, which subsequently releases stored angiogenic factors (Folkman and Klagsbrun, 1987). Tumor cells at the same time are able to release several angiogenic factors like VEGFs (Senger et al., 1983).

Hypoxia-induced VEGF (Shweiki et al., 1995) contributes to the onset of tumor-associated angiogenesis (See Figure 3). In tumors with a significant component of necrosis such as glioblastoma multiforme, VEGF mRNA expression is not uniform but occurs primarily in clusters of tumor cells at the border between viable tumor and necrotic areas (Plate et al., 1992). This localization is consistent with local hypoxia being a major inducer of VEGF gene expression (Shweiki et al., 1992). VEGF seems to play an essential role in tumor and wound angiogenesis, partly by inducing permeability with resulting plasma protein leakage from postcapillary venules. Subsequently, clotting of extravasated fibrinogen results in the formation of a fibrin gel substratum, ideal for endothelial and tumor cell growth (Dvorak, 1986; Dvorak et al., 1987).

Antagonists of VEGF inhibit the growth of many tumors. Antibodies against VEGF have a potent inhibitory effect on the growth of three human tumor cell lines injected subcutaneously in nude mice (Kim et al., 1993). In a nude mouse model of liver metastasis VEGF antibodies inhibit human colon carcinoma cell growth thus decreasing the number of metastases (Warren et al., 1995). Retrovirus-mediated expression of a dominant negative VEGFR-2 mutant suppresses the growth of glioblastoma multiforme as well as other tumor cell lines in vivo (Millauer et al., 1994; Millauer et al., 1996). In an in vivo model of teratocarcinomas VEGF null ES cells were impaired in their ability to form tumors in nude mice (Hillberg et al., 1992).

Fig. 3. The figure depicts schematically the generation of a new capillary sprout from a pre-existing vessel towards the hypoxic region of a tumor, such as a postcapillary venule (on the left). Numbers refer to various, often overlapping stages of the process. Angiogenic factors secreted by the tumor cells activate the vascular endothelium lining the vessel and pericytes (white) retract (1). Proteases degrade the basement membrane beneath endothelial cells (2), which subsequently become less adherent (3). The αvβ3 and αvβ5 integrins are important at these stages (Brooks et al., 1994; Brooks et al., 1994; Friedlander et al., 1995). Vascular permeability is also increased, allowing fibrin deposition into the tissue (4). Endothelial cells migrate toward the angiogenic stimulus (5) and enter the cell cycle (6). Circulating endothelial precursors may also become incorporated to the growing vessel sprout (7). New vessels mature and become established when periendothelial structures are formed (8), but this process is often defective in tumors. Abundant VEGF is secreted by hypoxic regions of the tumor (shown dark). VEGF is involved in most of the steps shown, while Ang-1 may function in concert with VEGF to stimulate vessel sprout invasion. Ang-2, which becomes upregulated in endothelial cells of angiogenic capillary sprouts may disrupt the interactions between endothelial cells and pericytes, thus sensitizing the endothelium to the mitogenic and chemotactic signals secreted by the tumor. Adapted from Veikkola and Alitalo, 1999.

2.5. Tumor lymphangiogenesis

The role played by the lymphatic vascular system in tumorigenesis has gained relatively little attention, despite the potential clinical relevance of lymphangiogenesis. The presence of lymphatic vessels in solid tumors is a matter of debate. According to one report, breast cancers posses lymphatic-like labyrinths (Van Netten et al., 1996). Intra-tumoral lymphatics are identified by others in cancers, but it is not clear whether these lymphatics are residual, newly formed, obstructed, malformed, or labyrinthine (Reichert, 1926; Witte et al., 1997). Tumor microvessels resemble lymphatics more than typical blood vessels, suggesting an indeterminate mix of embryonic vessels in the growing tumor (Witte and Witte, 1997). Furthermore, it has been argued that lymphangiogenesis does not take place in normal adult life or in tumors (Folkman, 1996). Due to the high interstitial pressure in the tumors, generated by leaky blood vessels and rapidly dividing tumor cells, it is thought that the lymphatic vessels collapse (Jain, 1997), or are compressed not having the possibility to extend into most solid tumors (Baish et al., 1997). Due to the lack of functional lymphatics in tumors, the peritumoral host lymphatics in this case drain the fluid and macromolecules coming from the tumors (Baxter and Jain, 1990).

3. Receptor tyrosine kinases

Key signals regulating developmental cell growth and differentiation, as well as remodeling and regeneration of adult tissues, are mediated by polypeptide growth factors and their transmembrane receptors, many of which are tyrosine kinases (RTKs) (Mustonen and Alitalo, 1995; van der Geer et al., 1994). Several families of receptor tyrosine kinases are characterized and some of them are strictly endothelial cell-specific.

Growth factor receptors with protein tyrosine kinase activity have a similar molecular topology. Firstly, they consist of a large glycosylated extracellular domain defining the receptor binding characteristics. Secondly, they consist of the hydrophobic transmembrane region anchoring the receptor in the plane of the plasma membrane. Thirdly, they contain the juxtamembrane domain involved in modulation of receptor functions by e.g protein phosphorylation. Fourthly, they contain the protein tyrosine kinase domain, which is the catalytic domain of the receptor, indispensable for signal transduction and induction of cellular responses. The kinase domain of subclass III RTKs is divided into two halves by insertions of hydrophilic amino acid residues. The role of this kinase insert region is to modulate receptor interactions with certain cellular substrates. Lastly, they contain the carboxy-terminal tail which interacts with the substrate binding sites of the protein tyrosine kinase region, modulating the capacity of the tyrosine kinase (TK) region to interact with exogenous substrates (Ullrich and Schlessinger, 1990).

RTKs are activated by polypeptide ligands commonly known as growth factors or cytokines. Signaling involves ligand binding, which induces a conformational change in the external domain of the receptor resulting in its dimerization (Ullrich and Schlessinger, 1990). This event results in receptor trans-phosphorylation at specific tyrosine residues and activation of the catalytic domains for the phosphorylation of cytoplasmic substrates. These phosphorylated tyrosine residues may serve to control the kinase activity of the receptor, and to create docking sites for the cytoplasmic signaling molecules, which are often substrates for the kinase. These molecules are adapters or enzymes themselves, linking RTKs to different signaling pathways. The interaction of these proteins with the activated RTKs can initiate signaling pathways leading to the nucleus or other cellular targets (Heldin, 1995). Following ligand binding and dimerization, receptors are internalized for degradation or recycling in order to attenuate signaling (Cadena and Gill, 1992).

3.1. Receptors for the VEGF family of growth factors

There are two subfamilies of receptor tyrosine kinases, which are specific to the vascular endothelium, the VEGFR subfamily and the Tie receptor subfamily. To date, three VEGFRs which are members of the class III RTKs are known, VEGFR-1, VEGFR-2 and VEGFR-3 (Shibuya et al., 1990;  Terman et al., 1991; Aprelikova et al., 1992). The other members of the class III receptor family are the receptors for PDGF, colony stimulating factor-1 (CSF-1R/c-fms), stem cell factor (SCFR/c-kit) and the FLT3/FLK2 receptor (Claesson-Welsh et al., 1989; Matsui et al., 1989; Matthews et al., 1991; Rosnet et al., 1991; Sherr, 1990; Witte, 1990). The VEGFRs have seven immunoglobulin homology domains (Ig) whereas the prototype of this subfamily PDGF receptor, has only five Ig repeats, as do CSF-IR, C-KIT and Flt3. Based on this structural similarity, the VEGFRs comprise a separate subfamily. For VEGFRs the second immunoglobulin homology domain is critical for ligand binding, whereas the three first domains are necessary to establish full affinity (Barleon et al., 1997; Davis-Smyth et al., 1996; Davis-Smyth et al., 1998; Wiesmann et al., 1997). Neuropilin (NP-1) is a non tyrosine kinase receptor shown recently to bind VEGF₁₆₅ (Soker et al., 1996; Soker et al., 1998). NP-1 binds also PlGF-2, VEGF-B and VEGF-E_NZ2but not PlGF-1 (Migdal at al., 1998; Makinen et al., 1999; Wise et al., 1999).

Fig. 4. Schematic view of the receptors for the VEGF family of growth factors and interaction of VEGFs with the various receptors.

Growth factors or receptors	Isoforms	Heparin binding	Receptors
VEGF	VEGF121, VEGF165, VEGF189, VEGF206, VEGF-145	VEGF165, VEGF189, VEGF206	VEGFR-1, VEGFR-2 NP-1 (VEGF165)
PlGF	PlGF-1, PlGF-2, PlGF-3	PlGF-2,	VEGFR-1 NP-1 (PlGF-2)
VEGF-B	VEGF-B167, VEGF-B186	VEGF-B167	VEGFR-1 NP-1 (VEGF-B167, VEGF-B186 fully processed)
VEGF-C			VEGFR-2, VEGFR-3
VEGF-D			VEGFR-2, VEGFR-3
VEGF-ENZ2, VEGF-ENZ7 VEGF-ED1701			VEGFR-2, NP-1 (VEGF-ENZ2)
VEGFR-1/FLT-1	VEGFR-1 Soluble VEGFR-1	VEGFR-1 Soluble VEGFR-1
VEGFR-2/KDR	VEGFR-2	VEGFR-2
VEGFR-3/FLT4	VEGFR-3 long VEGFR-3 short

3.1.a. VEGFR-1

Structure and chromosomal localization

VEGFR-1 originally isolated from a human placental cDNA library, is designated flt (fms–like tyrosine kinase), due to its close structural relationship to members of the fms family of receptor tyrosine kinases (Shibuya et al., 1990). Mouse and rat homologues of human VEGFR-1 were also cloned (Choi et al., 1994; Finnerty et al., 1993; Yamane et al., 1994). The human gene is located on chromosome 13q12-13 (Shibuya et al., 1990), and its mouse homologue on chromosome 5 (Rosnet et al., 1993). The promoter region contains a TATA box, a GC-rich region, putative transcription factor binding elements such as a cAMP response element binding the transcription factor (CREB/ATF) and also an ETS binding site (Morishita et al., 1995). VEGFR-1 promoter contains sequences matching the hypoxia inducible factor-1α consensus binding site (Gerber et al., 1997; Ikeda et al., 1996).

VEGFR-1 is a 180 kDa transmembrane glycoprotein, its mRNA occurs mainly as a 7.5-8.0 kb transcript, and alternative splicing produces a shorter soluble protein. The soluble form of VEGFR-1 encodes six of the N-terminal extracellular ligand-binding Ig domains but does not encode the last such domain, the transmembrane-spanning region or the intracellular tyrosine kinase domains (Kendall and Thomas, 1993; Shibuya et al., 1990). The soluble form of VEGFR-1 is a heparin-binding protein (Kendall and Thomas, 1993).

Expression

At E7.5-E12.5 of mouse embryonic development, when the embryonic vasculature is established, VEGFR-1 mRNA expression is detected in most if not all endothelia. During E14.5-E16.5 of fetal development, at a time when endothelial differentiation is complete but blood vessel growth is still ongoing, VEGFR-1 expression is decreased (Peters et al., 1993). VEGFR-1 is also found in populations of embryonic cells from which endothelium derives, such as early yolk sac mesenchyme (Peters et al., 1993). Expression of VEGFR-1 mRNA in the quiescent endothelium of adult organs and in vessels near healing wounds suggests that this receptor may have a role in the regulation of vascular permeability and in vascular repair and maintenance (Peters et al., 1993).

VEGFR-1 is expressed in an endothelial cell-specific manner also in human fetuses (Kaipainen et al., 1993). VEGFR-1 expression has not been extensively studied in lymphatic endothelial cells. However, in a recent study, lymphatic vessels of the human fetal heart are found negative for VEGFR-1 protein expression (Partanen et al., 1999a). According to one report VEGFR-1 is down regulated in the adult brain (Breier et al., 1995). Despite undetectable expression of VEGFR-1 in the normal adult rat brain, the same receptor mRNA is observed in glioma tissues undergoing active angiogenesis (Plate et al., 1993). Besides endothelial cells, monocytes express VEGFR-1 and both VEGF and PlGF stimulate tissue factor production and chemotaxis in monocytes (Clauss et al., 1996).VEGFR-1 mRNA is also expressed in human testis  (Ergun et al., 1997) and in bones (Plouet et al., 1989). Among the tumor cell lines studied, only a few melanomas and leukemias aberrantly express VEGFR-1 (Cohen et al., 1995; Fiedler et al., 1997). VEGFR-1 is upregulatedin the angiogenic vascular endothelium of tumors (Hatva et al., 1995; Shibuya, 1995). Hyperthyroidism, a disease accompanied by increased angiogenesis, also shows upregulated VEGFR-1 mRNA expression (Sato et al., 1995).

VEGFR-1 is upregulated by hypoxia in vivo and in vitro (Brogi et al., 1996; Gerber et al., 1997; Li et al., 1996; Marti and Risau, 1998; Tuder et al., 1995). The VEGFR-1 gene is also transiently expressed in pericytes under hypoxic culture conditions (Nomura et al., 1995). The expression of VEGFR-1 is selectively induced on dermal microvessels in skin explant cultures and in dermal endothelial cell monolayer cultures under hypoxic conditions. VEGFR-2 is downregulated in the same experiment (Detmar et al., 1997). Upregulation of VEGFR-1 but not VEGFR-2 under hypoxic conditions is demonstrated for HUVEC, suggesting differential transcriptional regulation of the two receptors by hypoxia. The VEGFR-1 gene is directly up-regulated by a hypoxia-inducible enhancer element located in the promoter of this receptor (Gerber et al., 1997).

Signal transduction

VEGFR-1 encodes a receptor for VEGF, PlGF and VEGF-B (de Vries et al., 1992; Olofsson et al., 1998; Park et al., 1994). In bovine adrenal cortex-derived capillary endothelial (ACCE) or HUVE cells PlGF is unable to induce tyrosine autophosphorylation (Park et al., 1994). On the contrary, PlGF activates VEGFR-1 autophosphorylation in transfected NIH3T3 cells (Sawano et al., 1996). More recently, PlGF is shown to increase the level of tyrosine phosphorylation in porcine aortic endothelial (PAE)/VEGFR-1 cells to a higher extent than VEGF treatment (Landgren et al., 1998).

Upon stimulation with VEGF VEGFR-1 expressing cells show an increased level of phosphorylation of members of the Src tyrosine kinase family such as Fyn and Yes (Waltenberger et al., 1994). VEGF induces phosphorylation of phospholipase C-gamma (PLC-γ), and GAP complex on tyrosine in VEGFR-1 transfected NIH3T3 cells as well as in endothelial cells (Seetharam et al., 1995). Strong activation of mitogen activated protein kinase (MAPK) is detected only in endothelial cells and tyrosine phosphorylation of SHC protein, an important adaptor for signal transduction for many receptor kinases, is very weak in both cell types mentioned (Seetharam et al., 1995). In contrast, other studies report that VEGFR-1 fails to activate MAPK whereas VEGFR-2 does (Kroll and Waltenberger, 1997). SCK which is one of many SHC homologues, binds to VEGFR-1 via its SH2 domain (Igarashi et al., 1998). MAPK activation is delayed and the mitogenic response is weaker in VEGFR-1 transfected fibroblasts when compared to endothelial cells (Takahashi and Shibuya, 1997). SHP-2, PLC-γ, and GRB2 bind to the VEGFR-1 intracellular domain (Ito et al., 1998). All VEGFRs are shown to be strong activators of signal transducers and activators of transcription -3 and -5 (STAT3 and STAT5), and this suggests that they participate in the regulation of endothelial function (Korpelainen et al., 1999). STATs are latent cytoplasmic transcription factors which can specifically bind to tyrosine phosphorylated receptors through their SH2 domains. VEGFR-1 signal transduction during vascular development appears to be insignificant (See below).

Biological effects

Mutant mice with targeted inactivation of VEGFR-1 gene, show disorganization of blood vessels and die at day 8.5 of embryonic development (Fong et al., 1995). The loss of normal vascular organization is a consequence of an overcrowded population of endothelial cells resulting from an increase in the number of hemangioblasts. The primary cause for this effect is an alteration in cell fate determination among mesenchymal cells (Fong et al., 1999). In contrast, VEGFR-1 tyrosine kinase-deficient homozygous mice develop normal vessels and survive (Hiratsuka et al., 1998). However, VEGF-induced macrophage migration is strongly suppressed in these mice. These results indicate that VEGFR-1 without a tyrosine kinase domain is sufficient to allow embryonic development with normal angiogenesis, and that the main role of this receptor is to act as a ligand binding molecule. VEGFR-1 having a higher affinity for VEGF than VEGFR-2 but little transducing activity of its own (Park et al., 1994), can function as a negative regulator of vasculogenesis and angiogenesis, by binding to free VEGF. Angiogenesis seems to be induced by a combined activation of both VEGFR-1 and VEGFR-2, since PlGF/VEGF heterodimers are angiogenic in contrast to PlGF homodimers (Kurz et al., 1998). However, the use of VEGF-E has demonstrated that VEGFR-2 can potently mediate angiogenesis independently of VEGFR-1 activation (Meyer et al., 1999). Soluble VEGFR-1 is able to inhibit tumor angiogenesis and growth in vivo, supporting the possibility of using VEGFRs as antiangiogenic targets (Goldman et al., 1998; Kong et al., 1998).

3.1.b. VEGFR-2

Structure and chromosomal localization

A second receptor for VEGF, VEGFR-2 first named KDR (kinase insert-domain containing receptor)/flk-1 (fetal liver kinase-1) was cloned from a human endothelial cDNA library (Terman et al., 1991). This was followed by the cloning of the mouse homologue fetal liver kinase 1 (Flk-1) also called NYK from cDNA libraries of mouse hematopoietic stem cells, neuroepithelium and embryonic stem cells (Matthews et al., 1991; Oelrichs et al., 1993; Yamaguchi et al., 1993). The rat, quail and zebrafish homologues have also been isolated (Eichmann et al., 1993; Sarzani et al., 1992; Sumoy et al., 1997).

The human gene encoding VEGFR-2 localizes in chromosomal region 4q11-q13 and the mouse homologue is mapped to chromosome 5, near to the c-kit and PDGFR-A genes (Matthews et al., 1991; Sait et al., 1995; Spritz et al., 1994; Terman et al., 1991). VEGFR-2 shares an overall structural organization with VEGFR-1 (de Vries et al., 1992; Matthews et al., 1991; Shibuya et al., 1990; Terman et al., 1991). Characterization of the human VEGFR-2 promoter reveals an endothelium-specific activating element in the long 5'-untranslated region of the first exon. Negative regulatory elements are located further upstream. The 5'-flanking region is rich in GC residues and lacks typical TATA or CAAT boxes (Ronicke et al., 1996).

VEGFR-2 is a 230 kDa protein. Two VEGFR-2 mRNAs have been reported, a full-length prototypic form, and a short form resulting from alternative splicing (Wen et al., 1998). Soluble VEGFR-1 forms a VEGF-stabilized complex with the extracellular domain of VEGFR-2 in vitro (Kendall et al., 1996).

Expression

VEGFR-2 is a very early marker of endothelial cell precursors, as it is expressed in mesodermal cells prior to any morphological evidence for endothelial cell differentiation (Yamaguchi et al., 1993). During mouse embryonic development VEGFR-2 mRNA distribution is largely restricted to capillaries, blood vessels and the endocardium (Millauer et al., 1993). During mouse brain angiogenesis VEGFR-2 is expressed in the perineural vascular plexus and the invading vascular sprouts (Millauer et al., 1993). In the early postnatal stages, when endothelial cell proliferation occurs, VEGFR-2 expression is evident but in the adult tissues where the vascularization process is completed, the turnover of endothelial cells in parallel with VEGFR-2 expression, are low (Millauer et al., 1993). In most human fetal tissues examined VEGFR-2 is expressed in an endothelial cell-specific manner (Kaipainen et al., 1993). VEGFR-2 is expressed in lymphatic endothelial cells in the quail (Wilting et al., 1997), in the human fetal heart (Partanen et al., 1999a) and in the skin of transgenic mice (Jeltsch et al., 1997).

Additionally, VEGFR-2 is expressed in hematopoietic stem cells, megakaryocytes, and platelets as well as in retinal progenitor cells (Katoh et al., 1995; Yang and Cepko, 1996). Furthermore, VEGFR-2 mRNA is detected in cells lining pancreatic ducts which are considered to contain precursor cells for the endocrine pancreas (Oberg et al., 1994). Some melanoma and leukemia tumor cell lines express VEGFR-2 (Fiedler et al., 1997; Liu et al., 1995). VEGF is expressed by all tumor cells studied whereas its receptor VEGFR-2 localizes at the tumor associated vasculature suggesting a paracrine relationship between them (Ferrara and Davis-Smyth, 1997; Klagsbrun and D'Amore, 1996; Shibuya, 1995). Unlike VEGFR-1, VEGFR-2 is not expressed in pericytes (Detmar et al., 1997).

VEGFR-2 is upregulated by hypoxia in vivo (Li et al., 1996; Marti and Risau, 1998; Tuder et al., 1995), but in vitro this receptor is either downregulated or not affected (Gerber et al., 1997; Takagi et al., 1996). In a recent report, hypoxic treatment of endothelial cells had no effect on VEGF ligand binding, but medium from hypoxic cell cultures indirectly increased ligand binding (Brogi et al., 1996). Postranscriptional regulation seems to be responsible for upregulated VEGFR-2 production under hypoxic conditions (Waltenberger et al., 1996). VEGFR-2 expression is upregulated on endothelial cells after VEGF treatment (Kremer et al., 1997; Shen et al., 1998). Thus, the hypoxia-induced VEGFR-2 expression may be triggered indirectly, since VEGF potentiates VEGFR-2 expression (Wilting et al., 1996).

Signal transduction

VEGFR-2 shows stronger tyrosine autophosphorylation compared to VEGFR-1 (Ito et al., 1998; Landgren et al., 1998; Sawano et al., 1996; Seetharam et al., 1995; Waltenberger et al., 1994). To date, VEGF, VEGF-C, VEGF-D and VEGF-E are shown to bind and induce tyrosine autophosphorylation of VEGFR-2 (Waltenberger et al., 1994; Joukov et al., 1996; Joukov et al., 1997; Achen et al., 1998; Ogawa et al., 1998; Wise et al., 1999; Meyer et al., 1999). Only the final glycosylated form of VEGFR-2 is capable of undergoing autophosphorylation in response to VEGF (Takahashi and Shibuya, 1997). Soluble forms of VEGFR-2 have not been reported so far (Kendall et al., 1996). The identification of a heparin-binding site in the extracellular domain of VEGFR-2 is consistent with the hypothesis that interactions between cell surface HSPGs and the VEGF receptor contribute to maximal VEGF binding (Dougher et al., 1997). Indeed, binding of VEGF to VEGFR-2 is stimulated by heparin (Terman et al., 1994; Roeckl et al., 1998).

VEGFR-2 is able to bind and phosphorylate PLC-γ  (Landgren et al., 1998). VEGF-induced stimulation of VEGFR-2 results in the association and phosphorylation of SHC isoforms and the induction of SHC-GRB2 complex formation (Igarashi et al., 1998). A SHC homologue, SCK binds to VEGFR-2 via its SH2 domain in a tyrosine kinase dependent manner (Igarashi et al., 1998). PKC is implicated in VEGF-induced MAPK activation in VEGFR-2 transfected NIH3T3 fibroblasts (Takahashi and Shibuya, 1997). VEGFR-2 also associates with GRB2 and NCK in a ligand-dependent fashion and VEGFR-2 stimulation is sufficient to activate MAPK (Kroll and Waltenberger, 1997), STAT3 and STAT5 (Korpelainen et al., 1999).

Biological effects

It is possible that VEGFR-2 is the main vascular permeability regulator. VEGFR-1 may not be directly involved in this process since VEGF is a potent inducer while PlGF-1 and PlGF-2 show twenty to forty fold less activity in mediating vascular permeability (Dvorak et al., 1995; Park et al., 1994; Sawano et al., 1996). Furthermore, VEGF-Eis as potent as VEGF₁₆₅ in inducing vascular permeability (Ogawa et al., 1998). In the induction of protease activity such as uPA production, VEGF and PlGF are potent in VEGFR-1 but not in VEGFR-2 expressing cells (Landgren et al., 1998). Although VEGFR-2 is expressed in lymphatic endothelial cells VEGF homo- and heterodimers are not potent in inducing lymphangiogenesis (Wilting et al., 1996; Oh et al., 1997). The formation of VEGFR-2  x VEGFR-3 heterodimers may be needed for the lymphangiogenic effect exerted by VEGF-C.

VEGFR-2 is not necessary for hemangioblast commitment, but is required for subsequent expansion of the lineage (Schuh et al., 1999). VEGFR-2 plays a pivotal role in endothelial development, being essential for yolk-sac blood-island formation and vasculogenesis in the mouse embryo. VEGFR-2-null embryos die between 8.5 and 9.5 days post-coitum, as a result of an early defect in the development of hematopoietic and endothelial cells (Shalaby et al., 1995). VEGFR-2 seems to be involved in the movement of cells from the posterior primitive streak to the yolk sac and hypothetically also to the intra-embryonic sites of early hematopoiesis (Shalaby et al., 1997).

3.1.c. VEGFR-3

Structure and chromosomal localization

VEGFR-3 (FLT4) was cloned from human erythroleukemia (HEL) cell (Aprelikova et al., 1992; Pajusola et al., 1993; Pajusola et al., 1992) and placenta cDNA libraries (Galland et al., 1992; Galland et al., 1993). Mouse (Finnerty et al., 1993) and quail (Eichmann et al., 1996; Eichmann et al., 1993) homologues of VEGFR-3 were cloned from embryonic cDNA libraries. The VEGFR-3 gene maps to chromosomal region 5q33-qter in humans (Aprelikova et al., 1992; Armstrong et al., 1993), close to the fms and PDGFR-β genes (Warrington et al., 1991) and to chromosome 11 in mice (Watkins-Chow et al., 1997). The extracellular domain of VEGFR-3 exhibits 33% and 37% amino acid sequence identity with VEGFR-1 and VEGFR-2 respectively, whereas the tyrosine kinase identity is approximately 80% (Pajusola et al., 1992). VEGFR-3 is expressed as mRNA forms of 4.5 and 5.8 kb, generated by alternative polyadenylation and subsequent alternative splicing during RNA processing (Pajusola et al., 1993). The 3'coding region of the 5.8 kb mRNA is 65 amino acids longer and encodes the long form of VEGFR-3, whereas the 4.5 kb mRNA encodes the short form. Only the longer from of VEGFR-3 is detected in human cell lines endogenously expressing VEGFR-3 proteins (Pajusola et al., 1993). Also only one mRNA form corresponding to the longer VEGFR-3 is found in mouse tissues (Galland et al., 1993). VEGFR-3 polypeptides undergo post-translational processing, which involves glycosylation and proteolytic cleavage to yield a mature receptor composed of two polypeptides held together by disulfide bonds (Borg et al., 1995; Fournier et al., 1995; Pajusola et al., 1994).

Expression

During early mouse development VEGFR-3 mRNA is expressed at E8.5 in the cardinal vein and in the angioblasts of the head mesenchyme. During days 11.5-12.5 of development, the VEGFR-3 signal is more prominent in the developing veins and presumptive lymphatic endothelia, with very weak if any positivity in arterial endothelia. VEGFR-3 expression becomes confined to the lymphatic endothelium during day 14.5 of development (Kaipainen et al., 1995). Additionally, VEGFR-3 mRNA is detected in the endothelial cells of the dorsal aorta in a E9.0 mouse embryos as well as in the yolk sac blood island endothelial cells (Dumont et al., 1998).

VEGFR-3 is also expressed in quail embryos, e.g in the intra-embryonic vascular plexus in a 2-day old embryo. In 3-day old embryos, all endothelial cells are positive but after that arteries become negative and expression is restricted to the lymphatic vasculature. In the quail, also other than endothelial cells express VEGFR-3 such as the cells of the notochord, podocytes of the kidney and some epithelial cells of the gallbladder and extrahepatic bile ducts (Wilting et al., 1997).

Northern blot analysis of human fetal tissues reveals transcript distribution in all the tissues except the thymus and small intestine, the highest levels being found in lung and spleen (Kaipainen et al., 1993; Pajusola et al., 1992). In situ hybridization studies show VEGFR-3 expression in human fetal heart, lung, kidney, mesenterium and gut (Kaipainen et al., 1993). In adults VEGFR-3 is expressed in the lymphatic vessels of the lung, mesenterium and tonsil, while veins and arteries are negative (Kaipainen et al., 1995). By Northern blotting, VEGFR-3 mRNA is detected in the placenta, lung, kidney, heart and liver in a decreasing order (Aprelikova et al., 1992). Surprisingly, VEGFR-3 protein is detected by immunohistochemistry in arterioles and venules as well as endothelial cells of capillaries in the nasal mucosa of human adults (Saaristo et al., 1999).

The two VEGFR-3 mRNA forms are differentially expressed in the HEL and megakaryoblastic DAMI leukemia cell lines. Additionally, a Wilms'tumor cell line, a retinoblastoma cell line and a nondifferentiated teratocarcinoma cell line express VEGFR-3 (Pajusola et al., 1992). VEGFR-3 expression is demonstrated also in cultured Kaposi's sarcoma (KS) cells (Liu et al., 1997).

Increased expression of VEGFR-3 mRNA is observed in the lymphatic sinuses of metastatic lymph nodes and in lymphangioma, andin lymph nodes, it is also detected in high endothelial venules (HEV) (Kaipainen et al., 1995). VEGFR-3 protein is found in the endothelium of lymphatic vessels around lymphomas. In cutaneous nodular AIDS-associated KS the tumor cells as well as the endothelium around the nodules is VEGFR-3 positive (Jussila et al., 1998). VEGFR-3 is expressed in lymphatic endothelia both in normal breast and in breast carcinomas, but it is also weakly expressed in the blood capillary endothelium showing enhanced expression in angiogenic capillaries (Valtola et al., 1999). VEGFR-3 is expressed specifically in lymphatic vessels in normal tissues but not in different benign and malignant vascular tumors where VEGFR-3 is not a specific marker for lymphatic endothelia. Proliferative neoplastic blood vessel endothelia in general acquire VEGFR-3-expression independently of lymphatic vascular differentiation (Partanen et al., 1999c). VEGFR-3 is detected in the fenestrated blood capillary endothelia of tissues with extensive molecular exchange across the blood vessel wall (Partanen et al., 1999b).

Signal transduction

VEGFR-3 binds VEGF-C and VEGF-D. Upon stimulation, not only VEGF-C but also VEGF-D induces strong VEGFR-3 autophosphorylation (Achen et al., 1998; Joukov et al., 1996). In the long form of VEGFR-3, tyrosine 1337 is a potential autophosphorylation site (Fournier et al., 1996), and serves as docking sites for GRB2 and SHC involved in VEGFR-3 signal transduction (Fournier et al., 1995).

In a HEL cell line expressing high levels of VEGFR-3, VEGF-C stimulation results in phosphorylation and activation of the recently identified related adhesion focal tyrosine kinase RAFTK showing an enhanced association of this kinase with the adaptor protein GRB2 (Wang et al., 1997). The tyrosine phosphorylated molecules SHC, GRB2, and SOS have been found to form a signaling complex with the activated VEGFR-3 (Wang et al., 1997). Upon stimulation of VEGFR-3 expressing cells, VEGF-C and VEGF-D induce rapid tyrosine phosphorylation of the SHC protein and activation of the mitogen-activated protein kinases ERK1 and ERK2 (Taipale et al., 1999).

Biological effects

A VEGF-C mutant, which binds to and activates only VEGFR-3 and not VEGFR-2 does not induce vascular permeability in vivo or stimulate migration of BCE cells in culture (Joukov et al., 1998). In contrast, VEGF-C is shown to induce the proliferation of PAE/VEGFR-2 and PAE/VEGFR-3 cells in culture (Cao et al., 1998). Cell shape changes and actin reorganization, as well as chemotactic responses are observed in both VEGFR-3 and VEGFR-2 overexpressing cells upon VEGF-C stimulation (Cao et al., 1998).

During mouse embryonic life VEGFR-3 has an essential role in the development of the cardiovascular system before the emergence of the lymphatic vessels. In VEGFR-3-null embryos there are no major defects in the differentiation of endothelial cells and in the process of vasculogenesis and angiogenesis. However, vascular remodeling and maturation are abnormal in large vessels which have defective lumens. Fluid accumulation occurs in the pericardial cavity, and cardiovascular failure at E9.5 (Dumont et al., 1998). Additionally, the yolk sac vasculature is underdeveloped lacking major blood vessels because of a failure of remodeling of the capillary network into complex vitelline vessels (Dumont et al., 1998).

3.1.d. NEUROPILIN

Structure

NP-1 gene encodes a transmembrane glycoprotein with a short (40 aa) highly conserved cytoplasmic tail (Fujisawa et al., 1989; Takagi et al., 1987; Takagi et al., 1995; Kawakami et al., 1996). Due to the short intracellular domain, lacking obvious docking sites, it is unlikely that NP-1 functions as an independent receptor. NP-1 is expressed on the cell surface as a 130-140 kDa receptor (Soker et al., 1998). Neuropilin-2 (NP-2) is closely related, having 47% homology to NP-1 (Kolodkin et al., 1997).

Ligand binding

NP-1 is a receptor for Collapsin-1/Semaphorin III/D (Sema III) family of proteins (Chen et al., 1997). Semaphorins are a large family of secreted and transmembrane proteins, implicated in repulsive axon guidance. NP-1 binds with high affinity to the three structurally related semaphorins Sema III, Sema E, and Sema IV, whereas NP-2 shows high affinity binding only to Sema E and Sema IV, and not to Sema III (Chen et al., 1997). Additionally, NP-1 binds VEGF₁₆₅, PlGF-2, VEGF-B and VEGF-E_NZ2but not VEGFR-1 or PlGF-1 (Soker et al., 1998; Migdal at al., 1998; Makinen et al., 1999; Wise et al., 1999). Binding of VEGF₁₆₅ to NP-1 occurs via the exon 7 encoded domain of VEGF (Soker et al., 1996; Soker et al., 1997), whereas to VEGFR-2 and VEGFR-1 it involves exons 4 and 3 encoded domains (Keyt et al., 1996). VEGF₁₆₅ and collapsin-1 compete for NP-1-binding sites. Collapsin-1 inhibits the motility of PAE/NP-1 and PAE/NP-1/VEGFR-2 cells, whereas VEGF₁₆₅ stimulates the motility of PAE/NP-1/VEGFR-2 cells (Miao et al., 1999). NP-1 binds PlGF-2 in a heparin-dependent manner (Migdal, et al., 1998), and acts as a VEGFR-2 co-receptor (Soker et al., 1998). PlGF-1 and PlGF-2 have the same effect on the migration of endothelial cells, which may mean that NP-1 does not function as a VEGFR-1 coreceptor. NP-1 and -2 form homo- and heterooligomers (Chen et al., 1998).

Expression

During mouse embryonic development, NP-1 is expressed in mesenchyme, heart, and limb buds (Kitsukawa et al., 1995). Additionally, NP-1 is expressed in endothelial cells of capillaries and blood vessels and in tumor cells (Kitsukawa et al., 1995; Soker et al., 1998). Furthermore, NP-1 is detected in neurons during active axonal growth and is suggested to be involved in axonal fasciculation and guidance (Kawakami et al., 1996). The NP-1 protein expression pattern is similar to that of VEGFR-1 in the human fetal heart. NP-1 is expressed in all the vascular endothelial structures except in the lymphatic vessels (Partanen et al., 1999a). NP-2 mRNA is expressed by developing neurons in a pattern partially overlapping with that of NP-1 (Chen et al., 1997; Soker et al., 1998).

Biological effects

The targeted disruption of the NP-1 gene indicates the important role that NP-sematophorin III/D-mediated chemorepulsive signals play in the guidance of peripheral nervous system (PNS) efferents and in nerve fiber fasciculation (Kawakami, et al., 1996; Takagi, et al., 1995; Fujisawa, et al., 1997; He, et al., 1997; Kolodkin, et al., 1997). Neuropilin-null mutant mice die after E12.5 dpc, and the homozygous mutant embryos show severe abnormalities in the efferent fibers of the PNS. The lack of NP-1 expression primarily affects nerve fiber guidance but not nerve fiber growth itself (Kitsukawa et al., 1997). Embryonic death is probably attributable to anomalies in the cardiovascular system. This evidence supports the importance of VEGF₁₆₅ in NP-1 signaling (Kitsukawa et al., 1997).

Ectopic or excess expression of NP-1 under the control of the β-actin promoter in mice results in morphological abnormalities in the tissues or organs in which this receptor is normally expressed (Kitsukawa et al., 1995). Chimeric embryos overexpressing NP-1 exhibit excess capillaries and blood vessels, ectopic sprouting and axonal defasciculation or extra digits, and these lesions are embryonic lethal. These results reveal the essential role of NP-1 in the normal development of the cardiovascular system, the nervous system and the limbs (Kitsukawa et al., 1995).

4. Vascular endothelial growth factors

VEGF was first described as a protein able to induce vascular permeability and endothelial cell proliferation and it was identified as a major inducer of angiogenesis and vasculogenesis (Ferrara and Davis-Smyth, 1997). VEGF belongs to the PDGF/VEGF family of growth factors which also includes PlGF (Maglione et al., 1991), VEGF-B (Olofsson et al., 1996a), VEGF-C (Joukov et al., 1996; Lee et al., 1996), VEGF-D (Achen et al., 1998) and VEGF-E (Meyer et al., 1999; Ogawa et al., 1998). The members of this family are structurally related growth factors. They are all dimeric glycoproteins sharing a number of biochemical and functional characteristics. Intra- and interchain disulphide bonds among eight characteristically spaced cysteine residues are involved in the formation of the dimeric active proteins (Keck et al., 1989; Bussolino et al., 1997). These factors bind and activate endothelial specific VEGFRs triggering signal transduction (See Figure 4).

Fig. 5. Amino acid sequence alignment of the VEGF and PDGF family members. Small black spots appearing above the sequence indicate (potential) glycosylation sites. Bold letters refer to VEGF homology domain. Gray letters mark the signal sequence and underlined parts indicate the Balbiani ring 3 protein regions. Boxes with gray background mark the invariant amino acids and black boxes indicate the invariant cysteines.

4.1. VEGF

Structure and functional properties

VEGF is a basic, 34-46-kDa homodimeric, heparin-binding glycoprotein, with its amino acid sequence being approximately 20% identical with PDGF A and B chains (Conn et al., 1990; Ferrara et al., 1992; Keck et al., 1989; Leung et al., 1989; Tischer et al., 1989). VEGF is also known as vascular permeability factor (VPF) or vasculotropin, and was originally purified as a tumor cell secreted protein (Connolly et al., 1989; Senger et al., 1987; Senger et al., 1990; Senger et al., 1983; Senger et al., 1986; Senger et al., 1993). Thehuman VEGF gene is located in chromosomal region 6p21.3 (Vincenti et al., 1996) and its coding region spans approximately 14 kilobases (kb). The VEGF gene in humans is organized into eight exons, separated by seven introns (Houck et al., 1991; Tischer et al., 1991). A single VEGF gene is alternatively spliced generating five different molecular species of mRNA, encoding putative proteins of 121, 145, 165, 189 and 206 amino acids (Ferrara et al., 1997; Poltorak et al., 1997). VEGF₁₆₅ is the predominant form produced in many normal and tumor cells, whereas VEGF₂₀₆ is the rarest form. VEGF₁₂₁ and VEGF₁₆₅ are secreted in soluble form, though a significant fraction of VEGF₁₆₅ remains bound to the cell surface or the ECM. The binding of VEGF isoforms to cell surfaces and ECM occurs via interactions with HSPG and reflects the differential heparin-binding capabilities of the VEGF isoforms. VEGF₁₂₁ is a weakly acidic protein which doesn't bind to heparin, whereas VEGF₁₆₅ binds to immobilized heparin. The larger isoforms VEGF₁₈₉ and VEGF₂₀₆ remain cell-associated, or sequestered in the ECM, maybe because of their greater affinity for HSPG but can be released and activated by soluble heparin or heparinase (Houck et al., 1992; Park et al., 1993; Peretz et al., 1992).

Receptor binding

VEGF binds to specific receptor tyrosine kinases, VEGFR-1 and VEGFR-2 (De Vries et al., 1992; Quinn et al., 1993). Cell surface-associated heparin-like molecules regulate the binding of VEGF to its receptors (Gitay-Goren et al., 1992). Recently NP-1 has been shown to bind VEGF_165. When coexpressed in cells with VEGFR-2, NP-1 enhances VEGF₁₆₅ binding to VEGFR-2 and its chemotactic and mitogenic activity (Soker et al., 1998). The binding of VEGF₁₆₅ to NP-1 is strongly enhanced by the addition of exogenous heparin (Soker et al., 1998; Gitay-Goren et al., 1992).

Expression

During mouse embryonic development VEGF mRNA is detected as early as E7.0 in both the extra-embryonic and embryonic endoderm. At E8.0 VEGF mRNA is expressed in the yolk sac endoderm, the mesothelium of the allantois and in the giant cells of the trophoblast layer (Breier et al., 1995; Jakeman et al., 1993). At later developmental stages in the mouse or rat embryos VEGF mRNA is observed in the choroid plexus epithelium, the vertebral column, along the surface of the spinal cord, in the ventricular layers of the brain and in the glomerular epithelium of the kidney (Breier et al., 1992; Jakeman et al., 1993). A certain subtype of endothelial cells, i.e., the endocardium, and cells in the outflow tract of the heart are positive for VEGF (Miquerol, et al., 1999). In the quail embryo, VEGF expression is detected in the endothelium of the aorta close to the heart, suggesting a potential autocrine regulation (Aitkenhead et al., 1998). In the adult rat VEGF mRNA is expressed in the brain, kidney, lung, liver and spleen, suggesting a role in the maintenance of normal vessel physiology (Ferrara et al., 1992). In the human fetus VEGF mRNA expression is detectable in all the tissues and is most abundant in lung, kidney, and spleen (Shifren et al., 1994). VEGF protein is localized in epithelial cells and myocytes, but not vascular endothelial cells (Shifren et al., 1994). VEGF is also produced by macrophages, T cells, smooth muscle cells, kidney cells, mesangial cells, keratinocytes, astrocytes, osteoblasts and tumor cells (Klagsbrun and D'Amore, 1996).

Regulation of expression

VEGF expression is regulated by a variety of hormones, growth factors and cytokines. VEGF mRNA levels are increased in cultured cells by IL-6 (Cohen et al., 1996), PDGF (Finkenzeller et al., 1992), EGF (Goldman et al., 1993), TGF-β (Pertovaara et al., 1994), KGF (Frank et al., 1995), prostaglandin E2 (Harada et al., 1995), Il-1-β (Ben-Av et al., 1995), thyroid stimulating hormone (Soh et al., 1996) and luteotrophic hormone (Garrido et al., 1993). Additionally, phorbol ester (Finkenzeller et al., 1992; Garrido et al., 1993), nitric oxide (Tuder et al., 1995) and the products of v-Has-ras and v-raf oncogenes (Grugel et al., 1995) potently induce VEGF expression.

Hypoxia is a fast and strong inducer of VEGF mRNA expression in a variety of normal and transformed in vitro cell types (Minchenko et al., 1994; Shima et al., 1995). The same effect takes place in vivo under the hypoxic conditions generated around necrotic regions of tumors (Plate et al., 1994; Shweiki et al., 1992), as well as during ischemia caused by occlusion of the coronary artery in pig and rat (Banai et al., 1994; Hashimoto et al., 1994). Hypoxic up-regulation of VEGF mRNA in neuroglial cells, secondary to the onset of neuronal activity, plays an important physiological role in the development of the retinal vasculature (Stone et al., 1995). VEGF gene promoters have cis-acting enhancer elements to which the hypoxia-inducible factor (HIF) binds, mediating hypoxia-induced transcriptional activation (Levy et al., 1995; Liu et al., 1995; Wang et al., 1995; Wang and Semenza, 1995). Increased mRNA stability is identified as a significant post-transcriptional component in upregulating VEGF (Ikeda et al., 1995; Levy et al., 1996).

Bioactivity

VEGF stimulates the migration and proliferation of arterial, venous, and microvascular endothelial cells and induces angiogenesis in vitro and in vivo (Connolly et al., 1989; Ferrara and Davis-Smyth, 1997; Leung et al., 1989; Plouet et al., 1989). Similarly,VEGF stimulates migration of monocytes/macrophages which display VEGFR-1 on their surface (Barleon et al., 1996; Clauss et al., 1990; Clauss et al., 1996). In contrast, PAE cells expressing VEGFR-1 after transfection fail to migrate towards VEGF, whereas PAE/VEGFR-2 cells, proliferate, migrate, show actin reorganization, and changes in cell morphology when stimulated with the same factor (Waltenberger et al., 1994). The angiogenic activity of VEGF is studied in many in vivo models, for example the CAM assay (See table on page 17 and Leung et al., 1989; Plouet et al., 1989). In addition, VEGF acts as a survival factor for serum deprived endothelial cells (Gerber et al., 1998a; Gerber et al., 1998b). Acting as a pro-survival factor, VEGF stimulates the expression of the anti-apoptotic proteins Bcl-2 and A1 in human endothelial cells (Gerber et al., 1998a). Endothelial cell dependence on VEGF is controlled by the presence of pericytes (Benjamin et al., 1998). Furthermore, VEGF induces vascular leakage as indicated in the Miles assay (Dvorak et al., 1995).

Biological role

Inactivation of the VEGF gene in mice results in embryonic lethality in heterozygous embryos, between day 11 and 12. These mice are growth retarded and have developmental anomalies (Carmeliet et al., 1996; Ferrara et al., 1996). In the heart, the outflow region is malformed, the dorsal aorta is rudimentary and the thickness of the ventricular wall is decreased. The yolk sac blood islands have a reduced number of nucleated red blood cells. The vitelline veins fail to fuse with the vascular plexus of the yolk sac. The vasculature shows defects in the placenta and the nervous system. Blood vessel ingrowth fails to the neuroepithelium and the forebrain region appears underdeveloped. VEGF heterozygous mice survive two days longer than VEGFR-1 and VEGFR-2 null embryos due to partial activation of these receptors by VEGF. The homozygous VEGF mice have a strikingly reduced number of vessels, showing a much less complex branching pattern if compared to the VEGF heterozygous mice. At E8.5 the dorsal aorta is absent and the endothelial cell development is delayed resulting into death at E10.5 (Carmeliet et al., 1996). VEGF deficiency impaired most steps of early vascular development, including in situ differentiation of blood islands and angiogenesis. Several factors may contribute to this phenotype, including abnormal accumulation or delayed differentiation of endothelial cells (Carmeliet et al., 1996; Ferrara et al., 1996). In contrast embryos lacking VEGFR-2 die due to the lack of formation of both endothelial cells and hematopoietic cells (Shalaby et al., 1995), as a result of the blocked migration of angioblasts to the initial embryonic sites of vasculogenesis (Shalaby et al., 1997). VEGFR-2 null embryos lack vasculogenesis and fail to develop blood islands (Shalaby et al., 1995). Organized blood vessels fail to develop throughout the embryo and yolk sac, resulting in death between day 8.5 and 9.5 (Shalaby et al., 1995). This difference in the VEGF and VEGFR-2 deficient mice suggests that a VEGF related factor would compensate for VEGF loss. Endothelial cells develop in VEGFR-1 deficient mice but these embryos die as a result of severe defects in the organization of the primitive vascular system (Fong et al., 1995). In these mice the endothelial cells develop in both embryonic and extra-embryonic sites but fail to organize in normal vascular channels (Fong et al., 1995). Changes in the cell fate determination among mesenchymal cells with increased hemangioblast commitment and an overproduction of endothelial cells are the primary defects in these knockout mice which die between days 8.5 and 9.5 (Fong et al., 1999). Thus, VEGF and its receptors are essential for blood island development and angiogenesis. VEGF is the most critical player since reduced concentrations are inadequate to support normal development, suggesting a tight dose-dependent regulation of embryonic vessels development by VEGF (Carmeliet et al., 1996). Embryonic lethality occurs even in the heterozygous state (Ferrara et al., 1996).

There seems to be functional complementarity of the VEGF isoforms in angiogenesis, since mice expressing only the VEGF₁₂₀ isoform show impaired myocardial angiogenesis and ischemic cardiomyopathy (Carmeliet et al., 1999). Specific isoforms may be important for an optimal angiogenic response in vivo. VEGF-C expression in the VEGF₁₂₀ mice is reduced and since this factor is also angiogenic in settings of tissue ischemia (Witzenbichler et al., 1998) the lower VEGF-C levels may have contributed to the impaired postnatal angiogenesis. Overexpression of VEGF during quail development results in hyper-vascularity and hyper-permeability (Flamme, 1995). On the other hand, VEGF administered during the onset of vasculogenesis in quails results in abnormal development and an excessive fusion of vessels (Drake and Little, 1995).

4.2. Placenta growth factor PlGF

Structure and functional properties

PlGF was initially cloned from a human placental cDNA library and is a 149-amino-acid-long protein. It is a N-glycosylated, homodimeric, secreted protein (Maglione et al., 1991). Alternative splicing generates three PlGF forms (Cao et al., 1996;Hauser and Weich, 1993; Cao et al., 1997). The smaller isoform is called PlGF-1 and has a 20-amino acid signal peptide that is cleaved to yield a 129-residue mature protein. The large isoform PlGF-2, differs from the other two PlGF forms by the insertion of a highly basic 21-amino acid stretch at the carboxyl-end of the protein which confers to it its heparin binding ability (Hauser and Weich, 1993). The third isoform, PlGF-3, was cloned by the polymerase chain reaction technique, using a human cDNA library prepared from the placental tissue of term pregnancy (Cao et al., 1997). PlGF-3 contains an in-frame insertion of 72-amino acids near the C-terminal portion of PlGF-1. Transient expression of PlGF-3 cDNA in mammalian cells results in the secretion to the conditioned medium as dimers and monomers (Cao et al., 1997). The PlGF gene is assigned to human chromosome 14q24-q31 (Mattei et al., 1996) and mouse chromosome 12q (DiPalma et al., 1996). Naturally occurring and secreted PlGF/VEGF heterodimers are detected in conditioned media of tumor cell lines (Cao et al., 1996; Park et al., 1994).

Receptor binding

PlGF-1 and PlGF-2 do not interact with VEGFR-2 but are able to bind to VEGFR-1 (Kendall et al., 1994; Park et al., 1994; Terman et al., 1994). Only PlGF-2 binds NP-1 in a heparin-dependent manner (Migdal et al., 1998).

Expression

During mouse embryonic development no mRNA expression of PlGF intra-embryonally is detected (Persico et al., 1999). Extra-embryonally, PlGF is expressed in the villous trophoblasts, whereas VEGF is found in cells of mesenchymal origin within the chorionic plate (Vuorela et al., 1997). Thus, placental cells produce only homodimeric PlGF (Vuorela et al., 1997). Mouse PlGF mRNA is detected at high levels in the placenta and at lower levels in the lung and thyroid (DiPalma et al., 1996).Immunohisto-chemistry shows PlGF expression in the vasculosyncytial membrane and in large blood vessels of the placental villi (Khaliq et al., 1996). Northern blot analysis shows low transcript levels in human thyroid, heart, brain, lung and skeletal muscle (Maglione et al., 1993; Viglietto et al., 1995). PlGF mRNA is abundant in the placenta, cultured human endothelial cells, such as HUVECs, trophoblastic tumors such as choriocarcinoma (JAR and JEG-3), and hepatoma HepG2 cells (Cao et al., 1996; Hauser and Weich, 1993; Maglione et al., 1991; Maglione et al., 1993).

PlGF mRNA levels are not increased during hypoxia (Cao et al., 1996; Viglietto et al., 1996). Under hypoxic conditions there is an increase in PlGF homodimers, but this can be explained by post-transcriptional regulation (Maglione et al., 1993). In the proposed post-transcriptional control mechanism of PlGF expression, the elimination of the 5' untranslated region (UTR) of PlGF mRNA increases PlGF synthesis (Maglione et al., 1993).

Biological activity and role

There seems to be disagreement regarding the biological activities of PlGF. PlGF is mitogenic for calf pulmonary arterial (CPA) endothelial cells in vitro (Maglione et al., 1991). Similarly, PlGF induces mitogenicity but not migration of VEGFR-1 expressing PAE cells (Landgren et al., 1998). In contrast, in ACCE or HUVE cells PlGF isoforms have no direct mitogenic or permeability-enhancing activity on their own but are able to potentiate the action of VEGF in vitro and in vivo (Park et al., 1994). However, still another study demonstrates that PlGF-1 and especially PlGF-2, induce proliferation of endothelial cells, angiogenesis, and permeability (Sawano et al., 1996; Ziche et al., 1997). The heterodimers composed of VEGF and PlGF subunits exert mitogenic activity almost as potent as VEGF homodimers on HUVE cells. In the same study, PlGF homodimers appear inactive and might be mitogenic only at high, possibly non-physiological concentrations (DiSalvo et al., 1995). Furthermore, in separate experimental settings PlGFs induce the migration of cells expressing VEGFR-1, such as monocytes and endothelial cells (Clauss et al., 1996). PlGF elicits an angiogenic response in the avascular rabbit cornea and in the CAM assay (Ziche et al., 1997). Instead, in another report PlGF-1 shows no angiogenic activity in the mouse cornea (Cao et al., 1996). Similarly, PlGF-1 and PlGF-2 have no angiogenic (Birkenhager et al., 1996; Oh et al., 1997) or lymphangiogenic effect (Oh et al., 1997) in the CAM assay.

Inactivation of the PlGF gene does not result in embryonic lethality, even in the homozygous state (Carmeliet and Collen, 1999). PlGF -/- mice are viable and fertile, although they may have some impairment of wound healing (Carmeliet and Collen, 1999).

4.3. VEGF-B

Structure and functional properties

Mouse VEGF-B was isolated from E14.5 embryonic and newborn brain cDNA libraries (Olofsson et al., 1996a; Townson et al., 1996). The human VEGF-B was cloned from a fibrosarcoma and a fetal brain cDNA libraries (Olofsson et al., 1996a; Grimmond et al., 1996). Two VEGF-B isoforms are generated by alternative splicing of mRNA from a single gene. The first one identified, VEGF-B₁₆₇, encodes 167 amino acidswhereas the second isoform VEGF-B₁₈₆ encodes 186 amino acid residues (Olofsson et al., 1996b). The NH2-terminal domains of the two VEGF-B isoforms are identical, whereas the COOH-terminal parts show no resemblance (Olofsson et al., 1996a; Olofsson et al., 1996b).Both isoforms show strong homology to VEGF at their amino termini. Both the human and mouse VEGF-B genes consist of seven coding exons and span about 4 kilobases of DNA (Olofsson et al., 1996b).Human VEGF-B gene maps to the D11S750 marker at chromosomal band 11q13 (Grimmond et al., 1996; Paavonen et al., 1996). VEGF-B₁₈₆ does not contain the highly basic and cysteine-rich COOH-terminal heparin binding domain found in VEGF-B₁₆₇. This results in the free secretion of the longer isoform from cells so that it will not remain bound to cellular or pericellular HSPGs (Olofsson et al., 1996b). Within the PDGF/VEGF family of growth factors, VEGF-B₁₈₆ represents the first isoform, which is modified by O-linked glycosylation. VEGF-B₁₈₆, like VEGF-B₁₆₇ is secreted as a disulfide-linked homodimer able to form disulfide-linked heterodimers with VEGF (Olofsson et al., 1996b). VEGF-B₁₆₇ is also secreted, but due to its strong affinity for HS, it remains cell-surface-associated. Amino acid sequence comparisons reveal that VEGF-B shares approximately 45% identity with VEGF (Olofsson et al., 1996a; Olofsson et al., 1996b). Proteolytic processing of the 60-kDa VEGF-B₁₈₆ dimer results in a 34-kDa dimer containing the receptor-binding epitopes (Olofsson et al., 1998b).

Receptor binding

VEGF-B₁₆₇ and VEGF-B₁₈₆ were transfected into 293-T cells and metabolically labeled conditioned medium from these cells was used to investigate binding to VEGFR Ig fusion proteins. Both VEGF-B isoforms bind to VEGFR-1 selectively. Binding is blocked by excess VEGF, suggesting partially overlapping interaction sites on the receptor (Olofsson et al., 1998). Conditioned medium from mouse VEGF-B₁₈₆ baculovirus-infected cells was used to assay VEGF-B binding to cell-surface-expressed VEGFR-1. VEGF-B competed for ¹²⁵I-VEGF binding to NIH 3T3-VEGFR-1 cells (Olofsson et al., 1998). VEGF-B₁₆₇ and the processed but not the full-length form of VEGF-B₁₈₆ bind also NP-1 (Makinen et al., 1999).

Expression

During mouse embryonic development VEGF-B mRNA is distributed in the heart, spinal cord, cerebral cortex, brown fat and skeletal muscle (Lagercrantz et al., 1996; Olofsson et al., 1996a). Expression in the heart is consistent throughout development (Olofsson et al., 1996a; Olofsson et al., 1996b). Northern blotting analysis reveals VEGF-B expression in mouse heart, brain, skeletal muscle and kidney. In human tissues the VEGF-B transcript is detected in the heart, skeletal muscle, pancreas and prostate. Glioblastomas, breast carcinomas and renal cell carcinomas were assayed for possibly increased VEGF-B levels (Grimmond et al., 1996). Elevated transcription was not found in any of these tumors when compared with their normal counterparts or non-malignant cell lines (Grimmond et al., 1996). Serum and its component growth factors as well as pro-inflammatory cytokines did not stimulate VEGF-B mRNA expression in vitro. Furthermore, hypoxia, Ras oncoprotein and mutant p53 tumor suppressor did not increase VEGF-B mRNA levels (Enholm et al., 1997; Ristimäki et al., 1998).

Biological activity and role

Conditioned medium containing human VEGF-B₁₆₇ collected from human embryo kidney 293EBNA cells transfected with the appropriate expression vector, stimulated proliferation of HUVE and BCE cells (Olofsson et al., 1996a). This stimulation of DNA synthesis might be explained by heterodimer formation with VEGF as, in additional experiments, a mitogenic effect was not obtained by using the purified recombinant human VEGF-B₁₈₆ expressed in Pichia pastoris (personal communication Dr B. Olofsson).

The function of VEGF-B though is still unclear, and even knockout mice show no obvious abnormalities, they are viable and fertile (U. Eriksson and collaborators, unpublished observations). These results imply that VEGF-B does not play a major independent role in vascular development. It is possible that VEGF-B modulates the biological activities of VEGF, either by forming heterodimers or by potentiation of its biological activities, as suggested for PlGF. Both VEGF-B isoforms are able to form homodimers and to heterodimerize with VEGF, and perhaps with other similar growth factors, and this adds to the diversity of their biological roles by allowing a variety of combinations for cellular signal transduction. Similarly, no specific phenotype is detected in transgenic mice overexpressing VEGF-B in the basal layer of the skin or in the heart (unpublished data of Michael Jeltsch and Kari Alitalo).

4.4. VEGF-C

Structure and functional properties

VEGF-C was cloned from the human prostatic adenocarcinoma cell line (PC-3) (Joukov et al., 1996) and human glioma cell line (G61) cDNA libraries (Lee et al., 1996). The avian VEGF-C was from a E4 quail cDNA library (Eichmann et al., 1998). The murine was cloned from a murine embryonic liver cell line (Fitz et al., 1997). The VEGF-C gene is located on chromosome 8 (Chilov et al., 1997; Fitz et al., 1997) and human chromosome 4q34 (Paavonen et al., 1996). Both human and mouse VEGF-C genes comprise over 40 kb of genomic DNA and consist of seven exons, all containing coding sequences (Chilov et al., 1997).Human VEGF-C cDNA encodes a preproprotein of 419 amino acid residues, with a predicted molecular mass of 46.9 kDa. The newly synthesized VEGF-C product is a pre-pro-protein, consisting of an N-terminal signal sequence followed by an N-terminal peptide, the VEGF-homology domain, and a C-terminal pro-peptide (Joukov et al., 1996). VEGF-C is secreted as a disulphide bonded homodimer, and most of it is proteolytically processed from the precursor polypeptide, which contains three putative N-glycosylation sites; two of these remain in mature, fully processed VEGF homology domain (Joukov et al., 1996). VEGF-C is first synthesized as a 58kDa precursor, most of which undergoes dimerization before secretion into the culture medium. It is then further cleaved, forming disulfide-linked 29 and 31kDa polypeptides, and rapidly secreted. These complexes are further cleaved after secretion into 15 and 21kDa polypeptides. Comparison of the obtained sequences with the sequence of the VEGF-C precursor indicates that the 15 and 31kDa polypeptides correspond to the N-terminal region of the secreted VEGF-C after cleavage of the signal peptide. The 29kDa form represents the C-terminal half of the VEGF-C precursor. The 21kDa form is generated by cleavage of the VEGF-C precursor. Mature polypeptides of 21 and 31kDa contain the entire VEGF homology domain with all eight conserved cysteine residues and two putative N-glycosylation sites (Joukov et al., 1997). The VEGF homology domain of VEGF-C is encoded by exons 3 and 4, whereas exons 5 and 7 encode cysteine-rich motifs and exon 6 encodes motifs typical of a silk protein (Chilov et al., 1997).

Receptor binding

The stepwise proteolytic processing of VEGF-C generates several VEGF-C forms with increased binding and activity towards VEGFR-3 and VEGFR-2. Non-processed VEGF-C preferentially binds to and activates VEGFR-3, whereas only the fully processed mature form of VEGF-C is a high affinity ligand and an activator of both VEGFR-3 and VEGFR-2 (Joukov et al., 1997). A VEGF-C site-directed mutant has been generated. It binds and activates VEGFR-3, but not VEGFR-2 (Joukov et al., 1998). The use of a truncated VEGF-C derivative reveals that its receptor-binding capacity resides in the portion of the molecule that is most closely related in primary structure to the other VEGF family members. This corresponds to the mature form of VEGF-C (Achen et al., 1998; Joukov et al., 1996).

Expression

In the quail VEGF-C mRNA is strongly expressed in regions destined to be rich in lymphatic vessels, particularly the mesenteries, in the region surrounding the jugular veins and the kidney (Eichmann et al., 1998). Northern analysis reveals VEGF-C expression most prominently in human heart, placenta, muscle, ovary and small intestine. Very little VEGF-C RNA is seen in the brain, liver or thymus (Joukov et al., 1996; Lee et al., 1996). Additionally, transcripts are detected in human spleen, lymph node, thymus, appendix, bone marrow and fetal liver (Fitz et al., 1997). Tumor cells like HT-1080 fibrosarcoma, PC-3 prostatic adenocarcinoma and G61 glioma cell lines express VEGF-C mRNA (Joukov et al., 1996; Lee et al., 1996). The upstream promoter sequences contain conserved putative binding sites for Sp-1, AP-2, and NF-kB transcription factors but no TATA box. The VEGF-C gene promoter does not contain putative binding sites for hypoxia-regulated factors (Chilov et al., 1997). VEGF-C mRNA levels are increased after stimulation of cultured cells by angiogenic proinflammatory cytokines (Ristimäki et al., 1998). This is mainly due to enhanced transcriptional activation rather than mRNA stability. Hypoxia, Ras oncoprotein and mutant p53 tumor suppressor, which are potent inducers of VEGF mRNA do not increase VEGF-C mRNA levels (Enholm et al., 1997).

Biological activity and role

VEGF-C increases vascular permeability as well as the migration and proliferation of endothelial cells (Joukov et al., 1997). The importance of VEGFR-2-mediated signal transduction appears to be critical for the vascular permeability activity of VEGF-C, and it is strongly suggested that the redundant biological effects of VEGF and VEGF-C depend on binding and activation of VEGFR-2 (Joukov et al., 1998). On the other hand, the fact that the effects of VEGF-C on blood vasculature of the mature CAM are minor (Oh et al., 1997), suggests that the signal for lymphatic angiogenesis is mediated by VEGFR-3, or its heterodimeric complex with VEGFR-2. This is also in good accord with the VEGF-C characteristic of being less potent that VEGF in stimulating the DNA synthesis of capillary endothelial cells (Joukov et al., 1996; Lee et al., 1996).

VEGF-C promotes the mitogenesis of human lung capillary endothelial cells (Lee et al., 1996). These experiments fail to distinguish whether the mitogenic effect is due to activation of VEGFR-3, or VEGFR-2. Evidence comes from a recent experiment (Cao et al., 1998) showing the potential of VEGF-C to stimulate, in a dose-dependent manner, the growth of PAE/VEGFR-3 cells. Recently VEGF-C was shown to exhibit a dose-dependent mitogenic and chemotactic effect on endothelial cells, particularly for microvascular endothelial cells (Witzenbichler et al., 1998).VEGF-C stimulates nitric oxide release from endothelial cells, increased vascular permeability and, in vivo, angiogenesis in the setting of limb ischemia (Witzenbichler et al., 1998). In addition, VEGF-C elicits angiogenic effects in limbal vessels in the mouse cornea and in CAM (Cao et al., 1998). Thus, VEGF-C can be involved in the regulation of physiological and pathological blood vessel growth.

Overexpression of VEGF-C in the skin of transgenic mice results in lymphatic, but not vascular endothelial proliferation and vessel enlargement (Jeltsch et al., 1997).Similar results are obtained with VEGF-C application to the mature CAM, resulting in a robust lymphangiogenic response but only a weak angiogenic response (Oh et al., 1997). It is possible that different responses are obtained depending on whether VEGF-C administration is topical or systemic in the transgenic mice. Depending on the spatial and temporal expression patterns of its receptors, VEGF-C is likely to play a dual role both as an angiogenic and lymphangiogenic growth factor.

VEGF-C potently induces an elongated, spindle-like cell shape change, and actin reorganization in both VEGFR-2 and VEGFR-3 overexpressing endothelial cells (Cao et al., 1998). In the same study both VEGFR-2 and VEGFR-3 mediate proliferative and chemotactic responses in endothelial cells upon VEGF-C stimulation. It is likely that VEGF-C but not VEGF also serves in vivo as a natural chemoattractant for lymphangiogenesis, since it is capable of inducing the migration of VEGFR-3-overexpressing endothelial cells (Cao et al., 1998).

4.5. VEGF-D

Structure and functional properties

VEGF-D is a 358-aa-long secreted protein. The VEGF-D gene was first identified as a c-fos-induced growth factor (FIGF) (Orlandini et al., 1996). VEGF-D was cloned from mouse fibroblast and lung cDNA libraries (Orlandini et al., 1996; Yamada et al., 1997; Achen et al., 1998; Rocchigiani et al., 1998). VEGF-D and VEGF-C share 48% sequence identity and have similar N- and C-terminal extensions, a characteristic which distinguishes them from the other members of the VEGF family. The VEGF homology domain of VEGF-D is much more closely related to that found in VEGF-C than to those of the other family members. The human VEGF-D gene spans about 50 kb and is organized into seven exons and six introns, its promoter contains an AP-1-binding site but no TATA box (Rocchigiani et al., 1998). The human gene is mapped to chromosomal region Xp22.1 (Rocchigiani et al., 1998; Yamada et al., 1997).

Receptor binding

The full length recombinant VEGF-D induces VEGFR-3 tyrosine autophosphorylation, whereas stronger phosphorylation is obtained with a truncated derivative of VEGF-D, containing the VEGF homology domain of VEGF-D corresponding to the so called mature VEGF-C. This shorter form of VEGF-D binds and stimulates tyrosine phosphorylation of VEGFR-2 (Achen et al., 1998, 1999).

Expression

The expression of VEGF-D studied by in situ hybridization during mouse embryonic development is detected in several parts of the body, such as limb buds, acoustic ganglia, developing tooth, heart, anterior pituitary, lung, kidney mesenchyme, liver, skin, and periosteum of the vertebral column (Avantaggiato et al., 1998). In mouse tissues two different transcripts of 2.4 and 3.7 kb are detected, with the highest expression in the lung, and the weakest in the brain, spleen, and liver. In the kidneys and testes, additional transcripts of 2.0 and 3.3 kb are observed (Yamada et al., 1997). In human adult tissues a 2.5-kb VEGF-D mRNA is detected in lung and heart (Rocchigiani et al., 1998). In another study, high expression of a 2.2 kb VEGF-D transcript was reported in the human lung, heart and small intestine, and lower expression in skeletal muscle, colon, and pancreas (Yamada et al., 1997). Expression in cancer cell lines is weak or absent (Yamada et al., 1997).

Biological activity and role

VEGF-D is capable of inducing mitogenic activity in fibroblasts, and also morphological alterations, but the receptors responsible for mediating these effects have not been identified (Orlandini et al., 1996). It is of interest to determine whether the cultured fibroblasts used for these studies expressed VEGFR-2 or VEGFR-3. For example, the intracellular domain of VEGFR-3 is shown to be capable of stimulating mitogenesis in fibroblasts (Pajusola et al., 1994; Borg et al., 1995). VEGF-D is a mitogen also for microvascular and BAE cells but is about fivefold less potent than VEGF₁₆₄ (Achen et al., 1998; Orlandini et al., 1996). VEGF-D differs from the other members of the VEGF family because it is the only one regulated by the nuclear oncogene c-fos (Orlandini et al., 1996). Furthermore, treatment of HUVECs with VEGF-D induces a dose-dependent cell growth, elongation, branching and a formation of an extensive network of capillary-like cords in a three-dimensional matrix (Marconcini et al., 1999). In an immortal cell line derived from Kaposi's sarcoma lesion VEGF-D treatment results in dose-dependent mitogenic and motogenic activities (Marconcini et al., 1999). VEGF-D is also a potent angiogenic factor in rabbit cornea in vivo in a dose-dependent manner (Marconcini et al., 1999).

4.6. VEGF-E

Structure and properties

The genome of three strains of the Orf (OV) virus, NZ2, NZ7 and D1701, encodes polypeptides with significant homology to the VEGFs (Lyttle et al., 1994; Meyer et al., 1999). The viral homologues are collectively called VEGF-E, and they have the characteristic cysteine knot motif present in all mammalian VEGFs. OV is a linear double-stranded DNA virus and a member of the parapoxvirus genus of the Poxvirus family. It causes contagious pustular dermatitis in sheep and goats and occasionally is transmissible to humans. The lesions induced by OV infection show extensive proliferation of the vascular endothelium and blood vessel dilation. These responses are likely to be direct effects of the VEGF-like gene. VEGF-E was expressed as the native protein in mammalian cells or as a recombinant protein in Escherichia coli and found to act as a heat-stable, secreted dimer (Meyer et al., 1999). VEGF-E has 25% amino acid identity with mammalian VEGF and is a dimer of approximately 44 kDa with no basic domain nor affinity for heparin (Ogawa et al., 1998).

Receptor binding and signal transduction

VEGF-E binds VEGFR-2, resulting in receptor autophosphorylation and a biphasic rise in free intracellular Ca2+ (Ogawa et al., 1998). Only VEGF-E_NZ2 binds also to NP-1 (Wise et al., 1999). Another study reports the tyrosine phosphorylation of PLC-γ followed by activation of MAPK, upon induction of VEGF-E (Ogawa et al., 1998).

Biological activity and role

VEGF-E is mitogenic for vascular endothelial cells in vitro and stimulates their chemotactic migration in in vitro angiogenesis assays (Meyer et al., 1999). It also stimulates angiogenesis in vivo, induces new vessel sprouts from pre-existing capillaries in a concentration-dependent manner, and sustains their growth and elongation over time (Meyer et al., 1999). Additionally, VEGF-E induces potent in vivo angiogenic activity in nude mice after subcutaneous Matrigel implantation and is also capable of increasing vascular permeability (Ogawa et al., 1998). Furthermore, VEGF-E promotes neovascularization in rabbit avascular cornea more efficiently than VEGF (Meyer et al., 1999). The angiogenic response appears to be direct as no macroscopic signs of inflammatory reaction are detected (Meyer et al., 1999). VEGF-E induces also an elongated morphology and tissue factor (TF) release in rat sinusoidal endothelial cells (Ogawa et al., 1998; Meyer et al., 1999).

5.1. Tie receptors

Structure and chromosomal localization

Tie-1 and Tie-2 or (Tek) constitute another family of angiogenic endothelial cell surface receptor tyrosine kinases. The human Tie-1 cDNA was isolated from erythroleukemia cells (Partanen et al., 1990; Partanen et al., 1992). The mouse, rat, and bovine homologues were also isolated (Korhonen et al., 1992; Korhonen et al., 1994; Maisonpierre et al., 1993; Sato et al., 1993). The Tie-1 gene is located in chromosomal region 1p33 to 1p34 in humans and in mice it is located on chromosome 4 (Partanen et al., 1992; Laan et al., 1995; Sato et al., 1993). Tie-2, a closely related gene was cloned from different sources i.e. murine embryonic heart, ES cells, brain capillaries, lung and yolk sac libraries (Dumont et al., 1992; Horita et al., 1992; Iwama et al., 1993; Runting et al., 1993; Sato et al., 1993; Schnurch and Risau, 1993). In the mouse Tie-2 is located on chromosome 4, in a region syntenic to human chromosome regions 1p22-23, 9q31-33 and 9p22-13 (Dumont et al., 1992; Sato et al., 1993). Tie-1 and Tie-2 encode 117 kDa and 140-kDa polypeptides, respectively (Dumont et al., 1993; Partanen et al., 1992). Both Tie-1 and Tie-2 are receptors with high sequence conservation in the cytoplasmic region, and difference in the extracellular ligand binding region. Overall aa identity is 46%, with the highest identity, 84% occuring in TK1 and TK2 domains. The extracellular region is composed of three distinct structural motifs that distinguish these receptors from other RTKs. The extracellular region consists of a cluster of three epidermal growth factor homology motifs positioned between one complete and one incomplete immunoglobulin-homology (Ig) domain. The second Ig domain is followed by three domains that show homology to fibronectin type-III domains (Partanen and Dumont, 1999). Their extracellular regions contain putative N-linked glycosylation sites followed by a single hydrophobic transmembrane domain. The cytoplasmic region consists of a tyrosine kinase (TK) domain split by a kinase insert (Sato et al., 1993).

Expression

Both Tie-1 and Tie-2 are expressed in the early embryonic vascular system and in maternal decidual vascular endothelial cells, where the vasculature undergoes active angiogenesis. Tie-2, but not Tie-1, is expressed in the extra-embryonic mesoderm of the amnion (Dumont et al., 1992). Tie-1 expression is downregulated after the fetal period in endothelial cells of many organs, yet the expression of Tie-1 and Tie-2 persists in quiescent adult endothelial cells (Dumont et al., 1992; Korhonen et al., 1994). Tie-1 is induced during physiological and pathological neovascularization (Hatva et al., 1994; Kaipainen et al., 1994; Korhonen et al., 1992), and Tie-2 is as well upregulated in angiogenic capillaries in breast cancer (Peters et al., 1998) and in gliomas (Stratmann et al., 1998). In addition, Tie-1 mRNA and protein levels are increased in endothelial cells of arteriovenous malformations (Hatva et al., 1996). Tie-2 tyrosyl phosphorylation is detected in the normal vasculature, suggesting that this receptor has an active role in the maintenance of blood vessels (Wong et al., 1997). In adults the 4.5 kb Tie-2 mRNA is observed at high levels in the lung and heart (Dumont et al., 1993; Iwama et al., 1993), in the endocardium and the leptomeninges (Dumont et al., 1992). In addition, Tie mRNA is visualized in the walls of medium and large vessels (Partanen et al., 1992).

Tie-1 and Tie-2 are restricted mainly to endothelial cells and their precursors (Korhonen et al., 1994; Dumont et al., 1995) but they are also expressed in hematopoietic progenitors and differentiating megakaryoblasts (Iwama et al., 1993; Sato et al., 1993; Schnurch and Risau 1993; Kukk et al., 1997; Sato et al., 1998). Tie-1 mRNA expression seems to correlate with the differentiation stage of blood cells, being highest in hematopoietic precursor cells, whereas the level is downregulated upon erythroid differentiation (Iwama et al., 1993). The highest levels of Tie-1 mRNA expression in mouse cell lines is found in the 32D and DA-3 cells, which are multipotent myeloid stem cells (Armstrong et al., 1993).

A4.4-kb Tie-1 mRNA is detected in HEL cells, KG-1 myeloid leukemia cells, and Dami megakaryoblastic leukemia cells. Tie-1 mRNA is expressed in endothelial cell lines, such as PAE cells and EA.hy926 a hybrid of human endothelial and lung carcinoma cells (Partanen et al., 1992). Human solid tumor cell lines, such as HT-1080 and 8387 fibrosarcoma as well as one melanoma cell line also expressed Tie mRNA (Armstrong et al., 1993).

Signal transduction

Tie-2 has two ligands, Ang-1 and Ang-2  which bind to it with similar affinity, but neither binds to the related receptor Tie-1 (Davis et al., 1996; Maisonpierre et al., 1997). Ang-1 induces autophosphorylation of Tie-2in cultured endothelial cells (Davis et al., 1996), whereas Ang-2 does not induce receptor phosphorylation (Maisonpierre et al., 1997). In 3T3 fibroblasts ectopically expressing Tie-2 both Ang-1 and Ang-2 induce receptor phosphorylation (Maisonpierre et al., 1997). SHP-2 and the adapter GRB2 are Tie-2 downstream substrates (Huang et al., 1995).

Biological effects

Gene inactivation studies reveal the important role played by Tie-1 and Tie-2 during the later stages of vascular development. Both receptors are required in the microvasculature during late organogenesis and in essentially all blood vessels of the adult (Puri et al., 1999). Tie-2 is involved in establishing the correct vascular patterning, whereas Tie-1 plays a role in maintaining vessel integrity, and preventing hemorrhage. Tie-1 also funtions selectively in various capillary plexuses, for postnatal survival of endothelial cells (Suri and Yancopoulos, 1998; Partanen and Dumont, 1999). Mouse embryos homozygous for a Tie-2-null allele or expressing a dominant negative Tie-2 receptor die around E9.5. These mice display defects in the vascular endothelium of the yolk sac, heart, dorsal aorta and microvasculature (Dumont et al., 1994; Sato et al., 1995). The number of endothelial cells is decreased suggesting a role for Tie-2 in the expansion of this lineage (Dumont et al., 1994). Hemorrhage in the mice is evident in parallel with an underdeveloped heart. Incomplete sprouting and remodeling in addition to vasodilatation characterize the head region of the mice and the yolk sac vessels (Sato et al., 1995). Embryos deficient in Tie-1 fail to establish structural integrity of vascular endothelial cells, resulting to edema and hemorrhage (Sato et al., 1995; Puri et al., 1995). Homozygous Tie-1 mutant mice die between E13.5 and one day after birth (Puri et al., 1995; Sato et al., 1995). Tie-1 is not required for angioblast differentiation but it is needed during organogenesis to promote angiogenic capillary growth (Partanen et al., 1996). Vasculogenesis proceeds normally in embryos lacking both Tie-1 and Tie-2. Tie-2 is essential for the differentiation of the heart endocardium, whereas Tie-2 and Tie-1 are dispensable for the initial assemply of the rest of the vasculature (Puri et al., 1999). Both, Tie-2 and Tie-1 play essential functions in maintaining the integrity of the mature vasculature (Puri et al., 1999).

5.2. Angiopoietins

Structure and properties

Besides the VEGFs, angiopoietins are the only other known growth factors specific to the vascular endothelial cells. The angiopoietin family includes the naturally occurring agonist, Ang-1 (Davis et al., 1996), and the antagonist, Ang-2 (Maisonpierre et al., 1997).  Ang-1 gene encodes 498 amino acid residues, including an amino-terminal secretory signal sequence (Davis et al., 1996). The Ang-2 gene encodes a protein of 496 amino acid residues, with a secretion signal peptide. Ang-2 and Ang-1 are 60% identical (Maisonpierre et al., 1997). More recently two new members have been described, angiopoietin-3 (Ang-3) in mouse which acts as an antagonist and angiopoietin-4 (Ang-4) in humans,  functioning as an agonist (Valenzuela et al., 1999). Although Ang-3 and Ang-4 are structurally more diverged from each other than are the mouse and human versions of Ang-1 and Ang-2, they apparently represent the mouse and human counterparts of the same gene locus (Valenzuela et al., 1999).

Receptor binding

Ang-1 binds and induces the tyrosine phosphorylation of Tie-2 (Davis et al., 1996). Ang-2 binds but does not activate Tie-2 and thereby is able to block Ang-1 activity in endothelial cells (Maisonpierre et al., 1997). In NIH3T3 cells ectopically expressing Tie-2 receptors, Ang-2 is equivalent to Ang-1 in inducing the phosphorylation of the receptor (Maisonpierre et al., 1997). It seems then, that Ang-2 acts as an antagonist for Tie-2 in the context of the endothelial cells, but not in other cell types (Maisonpierre et al., 1997). Ang-4 and Ang-3 bind Tie-2 but only Ang-4 induces phosphorylation of Tie-2 (Valenzuela et al., 1999).

Expression

Early in mouse development, Ang-1 is found to be most prominent in the heart myocardium surrounding the endocardium (Suri et al., 1996). Later in development, as observed in rat embryos, Ang-1 becomes more widely distributed especially in the mesenchyme surrounding developing vessels and in close association with endothelial cells (Davis et al., 1996). Ang-2 is not detected in the heart, but is abundant in the dorsal aorta and its major branches and specifically in the smooth muscle layer (Maisonpierre et al., 1997). In fetal liver, Ang-2 is not present in all vascular structures but is detected in cells at, or close to, the lumen of hepatic vessels, in cells which are probably endothelial cells or pericytes (Hirschi and D'Amore, 1996). In adult human tissues, Ang-2 is detectable only in ovary, placenta, and uterus, which are the three predominant sites of vascular remodeling in the normal adult. Ang-1 is widely expressed although in small amounts in the heart and liver (Maisonpierre et al., 1997). Ang-1 mRNA is found in tumor cells and Ang-2 mRNA is observed in the endothelial cells of a subset of glioblastoma blood vessels. Strong expression of Ang-2 originates from small capillaries with few periendothelial support cells, whereas larger glioblastoma vessels with the same characteristics show little or no expression (Stratmann et al., 1998). Ang-3 mRNA is expressed in multiple mouse tissues whereas Ang-4 transcripts are noted in high levels only in human lung (Valenzuela et al., 1999).

Biological activity and role

The embryonic expression profile of Ang-1, implies that it plays an important role in the heart during early development, and an increasingly widespread role as the rest of the vasculature matures (Davis et al., 1996; Suri et al., 1996). Ang-1 activation of Tie-2 does not induce the typical growth responses seen with other endothelial growth factors (Davis et al., 1996). Ang-1 seems to play a crucial role in mediating reciprocal interactions between the endothelium and surrounding matrix and mesenchyme. Thus Ang-1 may have a role in the maintenance of normal vasculature and Tie-2 activity (Peters et al., 1998). In contrast, it has been suggested that local production of Ang-2 promotes angiogenesis in the presence of VEGF, by inhibition of the stabilizing effect of Ang-1 and the subsequent loosening of the contacts between endothelial and periendothelial cells (Hanahan, 1997). The disruption of the Ang-1 gene leads to distinctive defects in embryonic vascular development, similar although more subtle to those seen in embryos lacking Tie-2. At E9.5 Ang-1 deficient mice have an immature endocardium devoid of trabeculae and loosely associated with the myocardium. A much less complex vascular network is formed composed of an immature primary capillary plexus with dilated vessels that surround the developing forebrain. Remodeling deficit is evident at E11.5 in the yolk sac and in the embryo resulting in a reduced number of large vessels. The endothelial cells in these mutant mice are aberrantly rounded and poorly associated with periendothelial cells and matrix. Mouse embryos homozygous for Ang-1 gene disruption, appear abnormal by E11 and die by E12.5 (Suri et al., 1996). The phenotype of the mutant mice stands as an evidence for the important role that Ang-1 plays in the development of the vasculature. In adult mice and humans, Ang-2 is expressed only at sites of vascular remodeling, and transgenic overexpression of Ang-2 in vascular structures results in disrupted blood vessels (Maisonpierre et al., 1997). In conclusion, the Angs and Tie-2 do not participate in vasculogenesis, but do play an important role in angiogenesis, vessel remodeling and maturation (Dumont et al., 1994; Sato et al., 1995).

6. Ephrins and Eph-receptors

The Eph receptors comprise the largest subfamily or RTKs. The ephrins and their Eph receptors have been mostly studied for their roles in the nervous system. They seem to help guide axons by mediating repulsive cues (Flanagan and Vanderhaeghen, 1998). More recently the interest has been shifted to the role that ephrins and their receptors play in the development of arteries and veins. Arterial and venous endothelial cells are distinct from the earliest stages of angiogenesis. The molecular determinants that distinguish them include the transmembrane ligand, ephrin-B2 (Bennett et al., 1995; Bergemann et al., 1995) and its receptor Eph-B4 (Andres et al., 1994). Ephrin-B2 is expressed on arterial endothelial cells whereas Eph-B4 marks the venous endothelium (Wang et al., 1998). Because both, ephrin-B2 and Eph-B4 are membrane bound, signalling must occur at sites of cell-cell contact. The interactions between arterial and venous endothelial cells are essential for the morphogenesis of the capillary network (Wang et al., 1998).

---

Back to the title page | E-thesis main page