Endocrine Reviews 18 (1): 4-25
Copyright © 1997 by The Endocrine Society
The Biology of Vascular Endothelial Growth Factor
Napoleone Ferrara and
Terri Davis-Smyth
Department of Cardiovascular Research, Genentech, Inc., South San
Francisco, California 94080
 |
Abstract
|
|---|
- I. Introduction
- II. Biological Activities of VEGF
- III. Organization of the VEGF Gene
- IV. Properties of the VEGF Isoforms
- V. Regulation of VEGF Gene Expression
- A. Hypoxia
- B. Cytokines
- C. Differentiation and transformation
- VI. The VEGF Receptors
- A. Characterization and distribution of VEGF-binding sites
- B. The Flt-1 and Flk-1/KDR tyrosine kinases
- 1. Binding characteristics
- 2. Signal transduction
- 3. Regulation
- 4. Structural requirements for ligand binding in Flt-1 and Flk-1/KDR
- 5. VEGF determinants for binding Flt-1 and Flk-1/KDR
- VII. VEGF-Related Molecules
- VIII. Role of VEGF and Its Receptors in Physiological Angiogenesis
- A. Distribution of VEGF, Flk-1/KDR and Flt-1 mRNA
- B. Analysis of Flk-1/KDR, Flt-1 and VEGF gene knockouts
- IX. Role of VEGF in Pathological Angiogenesis
- A. Tumor angiogenesis
- 1. Expression of VEGF in human tumors
- 2. Inhibition of VEGF action in vivo
- B. Intraocular neovascular syndromes
- C. Other pathological conditions
- X. Therapeutic Applications of VEGF-Induced Angiogenesis
- XI. Perspectives
|
I. Introduction
|
|---|
THE development of a vascular supply is a fundamental
requirement for organ development and differentiation during
embryogenesis (1, 2) as well as for wound healing and reproductive
functions in the adult (3, 4). Angiogenesis is also implicated in the
pathogenesis of a variety of disorders: proliferative retinopathies,
age-related macular degeneration (AMD), tumors, rheumatoid arthritis,
psoriasis, etc. (3, 4). In the case of proliferative retinopathies and
AMD, the new blood vessels are directly responsible for many of the
destructive events characteristic of these conditions. Leakage and
bleeding, followed by organization of the clot and fibrosis, may
ultimately lead to retinal detachment or irreversible damage to the
macula (5). Conversely, tumor-associated neovascularization, by
establishing continuity with the systemic circulation, allows the tumor
cells to express their critical growth advantage and also facilitates
metastatic spreading (3, 4). Accordingly, a correlation has been
observed between density of microvessels in primary breast carcinoma
sections, nodal metastases, and survival (6, 7, 8). Similarly, a
correlation has been reported between vascularity and invasive behavior
in a variety of other tumors (9, 10, 11, 12). These findings led several
investigators to conclude that the number of vessels in tumor sections
is an independent predictor of outcome in cancer patients (9, 10, 11, 12).
The search for potential regulators of angiogenesis has yielded
numerous candidates: acidic fibroblast growth factor (aFGF), basic
fibroblast growth factor (bFGF), transforming growth factor-
(TGF-), TGF-ß, hepatocyte growth factor, tumor necrosis factor-
(TNF-), angiogenin, interleukin-8 (IL-8), etc. (13, 14). Although
these molecules are able to promote angiogenesis, at least in certain
model systems, it has been difficult to correlate such activity with
the physiological or pathological regulation of blood vessel growth.
Work done by several laboratories over the last few years has
elucidated the pivotal role of vascular endothelial growth factor
(VEGF) in the regulation of normal and abnormal angiogenesis (15). In
particular, the recent finding that the loss of even a single VEGF
allele results in embryonic lethality points to an irreplaceable role
played by this factor in the development and differentiation of the
vascular system (16, 17). Furthermore, VEGF-induced angiogenesis has
been shown to result in a therapeutic effect in animal models of
coronary (18, 19, 20) or limb (21, 22, 23) ischemia and, most recently, in a
human patient affected by critical leg ischemia (24).
|
II. Biological Activities of VEGF
|
|---|
VEGF is a potent mitogen (ED50 2–10 pM)
for micro- and macrovascular endothelial cells derived from arteries,
veins, and lymphatics, but it is devoid of consistent and appreciable
mitogenic activity for other cell types (25, 26, 27, 28, 29, 30, 31). The denomination of
VEGF was proposed to emphasize such narrow target cell specificity (25, 26). VEGF promotes angiogenesis in tridimensional in vitro
models, inducing confluent microvascular endothelial cells to invade
collagen gels and form capillary-like structures (32). These studies
provided evidence for a potent synergism between VEGF and bFGF in the
induction of this effect (32). Also, VEGF induced sprouting from rat
aortic rings embedded in a collagen gel (33). This model emphasizes the
specificity of VEGF, as the proliferation induced by this growth factor
consisted almost exclusively of vascular endothelial cells. In
contrast, insulin-like growth factor-I (IGF-I) or platelet-derived
growth factor (PDGF) induced endothelial cell sprouting accompanied by
extensive fibroblastic proliferation (33). VEGF also elicits a strong
angiogenic response in a variety of in vivo models including
the chick chorioallantoic membrane (26, 29), the rabbit cornea (34),
the primate iris (35), the rabbit bone (27), etc.
VEGF induces expression of the serine proteases uro-kinase-type and
tissue-type plasminogen activators (PA) and also PA inhibitor 1 (PAI-1)
in cultured bovine microvascular endothelial cells (36). Moreover, VEGF
increases expression of the metalloproteinase interstitial collagenase
in human umbilical vein endothelial cells but not in dermal fibroblasts
(37). The coinduction of PA and collagenase by VEGF is consistent with
a prodegradative environment that facilitates migration and sprouting
of endothelial cells. Pepper and Montesano (38) proposed that PAI-1
provides a negative regulatory step that serves to balance the
proteolytic process. Other studies have shown that VEGF promotes
expression of urokinase receptor (uPAR) in vascular endothelial cells
(39). Considering that the PA-plasmin system and in particular the
interaction of uPA with uPAR is an important element in the chain of
cellular processes that mediate cellular invasion and tissue remodeling
(40), these findings are consistent with the proangiogenic activities
of VEGF.
VEGF is known also as vascular permeability factor (VPF) based on its
ability to induce vascular leakage in the guinea pig skin (41, 42).
Dvorak and colleagues (43, 44) proposed that an increase in
microvascular permeability is a crucial step in angiogenesis associated
with tumors and wounds. According to this hypothesis, a major function
of VPF/VEGF in the angiogenic process is the induction of plasma
protein leakage. This effect would result in the formation of an
extravascular fibrin gel, a substrate for endothelial and tumor cell
growth. Recent studies have also suggested that VEGF may be a factor
that induces fenestrations in endothelial cells (45). Topical
administration of VEGF acutely resulted in the development of
fenestrations in the endothelium of small venules and capillaries, even
in regions where endothelial cells are not normally fenestrated, and
was associated with increased vascular permeability (45).
Interestingly, Dellian et al. (46) have described the
quantification and long-term physiological characterization of
microvessels induced by gels containing either VEGF or bFGF in
transparent chambers in the dorsal skin or in the cranium of mice.
These studies indicate that VEGF- or bFGF-induced vessels have similar
diameter, permeability to albumin, and red cell velocities. However,
permeability and red cell velocities were higher in the cranium than in
the dorsal skin. These findings led to the conclusion that the
steady-state physiological properties of blood vessels, including
permeability, are primarily determined by the local microenvironment,
rather than the initial angiogenic stimulus (46).
An additional effect of VEGF on the vascular endothelium is the
stimulation of hexose transport (47). Exposure of bovine aortic
endothelial cells to VEGF or TNF- resulted in a significant increase
in the rate of hexose transport. The combination of factors had an
additive effect. This action may have relevance for increased energy
demands during endothelial cell proliferation or inflammation.
Recently, Melder et al. (48) have shown that VEGF promotes
expression of VCAM-1 and ICAM-1 in endothelial cells. This induction
may result in the adhesion of activated natural killer (NK) cells to
endothelial cells, mediated by specific interaction of endothelial
VCAM-1 and ICAM-1 with CD18 and VLA-4 on the surface of NK cells (48).
It has been suggested that these effects may provide an explanation for
the previously observed preferential adhesion of IL-2-activated NK
cells to the tumor vasculature (49).
VEGF has been reported to have regulatory effects on certain blood
cells. Clauss et al. (50) reported that VEGF may promote
monocyte chemotaxis. More recently, Broxmeyer et al. (51)
have shown that VEGF induces colony formation by mature subsets of
granulocyte-macrophage progenitor cells that had been stimulated with a
colony stimulating factor. These findings may be explained by the
common origin of endothelial cells and hematopoietic cells and the
presence of VEGF receptors in progenitor cells as early as
hemangioblasts in blood islands in the yolk sac (see Section
VIII). Furthermore, Gabrilovich et al. (52) have
reported that VEGF may have an inhibitory effect on the maturation of
host professional antigen-presenting cells such as dendritic cells.
VEGF was found to inhibit immature dendritic cells, without having a
significant effect on the function of mature cells. These findings led
to the provocative hypothesis that VEGF may facilitate tumor growth
also by allowing the tumor to avoid the induction of an immune response
(52).
VEGF induces vasodilatation in vitro in a dose-dependent
fashion (53) and produces transient tachycardia, hypotension, and a
decrease in cardiac output when injected intravenously in conscious,
instrumented rats (54). Such effects appear to be caused by a decrease
in venous return, mediated primarily by endothelial cell-derived nitric
oxide, as assessed by the requirement for an intact endothelium and the
prevention of the effects by N-methyl-arginine (53, 54).
Accordingly, VEGF has no direct effect on contractility or rate in
isolated rat heart in vitro (54). These hemodynamic effects,
however, are not unique to VEGF: other angiogenic factors such as aFGF
and bFGF may also induce nitric oxide-mediated vasodilatation and
hypotension (55).
|
III. Organization of the VEGF Gene
|
|---|
The human VEGF gene is organized in eight exons, separated by
seven introns, and its coding region spans approximately 14 kb (56, 57). The human VEGF gene has been assigned to chromosome 6p21.3 (58).
cDNA sequence analysis of a variety of human VEGF clones had indicated
that VEGF may exist as one of four different molecular species, having,
respectively, 121, 165, 189, and 206 amino acids (VEGF121,
VEGF165, VEGF189, VEGF206) (26, 28, 56, 57). It is now well established that alternative exon splicing of a
single VEGF gene is the basis for this molecular heterogeneity.
VEGF165 lacks the residues encoded by exon 6, while
VEGF121 lacks the residues encoded by exons 6 and 7.
Compared with VEGF165, VEGF121 lacks 44 amino
acids; VEGF189 has an insertion of 24 amino acids highly
enriched in basic residues, and VEGF 206 has an additional
insertion of 17 amino acids. Interestingly, there is no intron between
the coding sequence of the 24-amino acid insertion in
VEGF189 and the additional 17-amino acid insertion found in
VEGF206. The 5'-end of the 51-bp insertion of
VEGF206 begins with GT, the consensus sequence for the
5'-splice donor necessary for mRNA processing. Therefore, the
definition of the 5'-splice donor site for removal of a 1-kb intron
sequence is variable (57). Analysis of the VEGF gene promoter region
reveals a single major transcription start that lies near a cluster of
potential Sp1 factor-binding sites. Also, several potential binding
sites for the transcription factors AP-1 and AP-2 are present in the
promoter region (56). VEGF165 is the predominant molecular
species produced by a variety of normal and transformed cells.
Transcripts encoding VEGF121 and VEGF189 are
detected in the majority of cells and tissues expressing the VEGF gene
(56, 57). In contrast, VEGF206 is a very rare form, so far
identified only in a human fetal liver cDNA library (57). The
organization of the murine VEGF gene has been also described (59).
Similarly to the human gene, the coding region of the murine VEGF gene
encompasses approximately 14 kb and is comprised of eight exons
interrupted by seven introns. Analysis of exons suggests the generation
of three isoforms, VEGF120, VEGF164 and
VEGF188. Therefore, murine VEGFs are shorter than human
VEGF by one amino acid. A fourth isoform comparable to
VEGF206 is not predicted, since an in-frame stop codon is
present in the region corresponding to the human VEGF206
open reading frame. Analysis of the 3'-untranslated region of the rat
VEGF mRNA has revealed the presence of four potential polyadenylation
sites (60). A frequently used site is about 1.9 kb further downstream
from the previously reported translation termination codon (30). The
sequence within this 3'-untranslated region reveals a number of motifs
that are known to be involved in the regulation of mRNA stability (60).
(See also Section V. A.)
|
IV. Properties of the VEGF Isoforms
|
|---|
Native VEGF is a basic, heparin-binding, homodimeric
glycoprotein of 45,000 daltons (61). These properties correspond to
those of VEGF165, the major isoform. VEGF121 is
a weakly acidic polypeptide that fails to bind to heparin (62).
VEGF189 and VEGF206 are more basic and bind to
heparin with greater affinity than VEGF165 (62). Previous
studies demonstrated that such differences in the isoelectric point and
in affinity for heparin may profoundly affect the bioavailability of
VEGF (62, 63). VEGF121 is a freely soluble protein;
VEGF165 is also secreted although a significant fraction
remains bound to the cell surface and the extracellular matrix. In
contrast, VEGF189 and VEGF206 are almost
completely sequestered in the extracellular matrix (63). However, these
isoforms may be released in a soluble form by heparin or heparinase,
suggesting that their binding site is represented by proteoglycans
containing heparin-like moieties. Interestingly, the long forms may be
released by plasmin (62, 63) after cleavage at the COOH terminus. This
action generates a bioactive proteolytic fragment having a molecular
mass of 
34,000 daltons (62, 63). Plasminogen activation and
generation of plasmin have been shown to play an important role in the
angiogenesis cascade. Thus, proteolysis of VEGF is likely to occur also
in vivo. Generation of bioactive VEGF by proteolytic
cleavage may be especially important in the microenvironment of a tumor
where increased expression of proteases, including PA, is well
documented (64, 65). Keyt et al. (66) have shown that the
bioactive product of plasmin action is comprised of the first 110
NH2-terminal amino acids of VEGF. These findings suggest
that the VEGF proteins may become available to endothelial cells by at
least two different mechanisms: as freely diffusible proteins
(VEGF121, VEGF165) or after protease activation
and cleavage of the longer isoforms. However, loss of heparin binding,
whether it is due to alternative splicing of RNA or plasmin cleavage,
results in a substantial loss of mitogenic activity for vascular
endothelial cells: compared with VEGF165,
VEGF121 or VEGF110 demonstrate 50-fold reduced
potency when tested in endothelial cell growth assay; the
VEGF165/110 heterodimer resulting from limited proteolysis
of VEGF165 demonstrated a 5–10-fold loss in potency when
compared with wild type VEGF165 (66). It has been suggested
that the stability of VEGF-heparan sulfate-receptor complexes
contributes to effective signal transduction and stimulation of
endothelial cell proliferation (66). Thus, VEGF has the potential to
express structural and functional heterogeneity to yield a graded and
controlled biological response. Figure 1
illustrates some of
the actions of the VEGF isoforms on the vascular endothelium and
possible regulatory mechanisms.
View larger version (55K):
[in this window]
[in a new window]
|
Figure 1. Schematic representation of the actions of VEGF
isoforms on the vascular endothelium. Several stimuli may result in the
release of the diffusible alternatively spliced VEGF isoforms
(VEGF165, VEGF121) from a variety of cell
types. These proteins may induce a complex series of effects on the
vascular endothelium, including cell sprouting, induction of
interstitial collagenase, plasminogen activators (PA), and plasminogen
activator inhibitor I-1 (PAI-1), as well as extravasation of plasma
proteins. Plasminogen activation results in generation of plasmin,
which may cleave extracellular matrix-bound VEGF (VEGF189
or VEGF206) to release a diffusible proteolytic fragment
(VEGF110). Plasmin may also activate procollagenase.
Activation of PAI-1 may constitute a negative regulatory step, by
inhibiting the action of PA.
|
|
|
V. Regulation of VEGF Gene Expression
|
|---|
A. Hypoxia
Several mechanisms have been shown to participate in the
regulation of VEGF gene expression. Among these, oxygen tension plays a
major role, both in vitro and in vivo. VEGF mRNA
expression is rapidly and reversibly induced by exposure to low
pO2 in a variety of normal and transformed cultured cell
types (67, 68, 69, 70). In glioblastoma multiforme and other tumors with
significant component of necrosis, the VEGF mRNA is highly expressed in
ischemic tumor cells that are juxtaposed to areas of necrosis (68),
suggesting that local hypoxia is a major inducer of VEGF gene
expression in the microenvironment of a tumor. Also, ischemia caused by
occlusion of the left anterior descending coronary artery results in a
dramatic increase in VEGF RNA levels in the pig myocardium, suggesting
the possibility that VEGF may mediate the spontaneous revascularization
that follows myocardial ischemia (71). Furthermore, Stone et
al. (72) proposed that 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.
Similarities exist between the mechanisms leading to hypoxic regulation
of VEGF and erythropoietin (Epo) (73). Hypoxia inducibility is
conferred on both genes by homologous sequences. By deletion and
mutation analysis, a 28-base sequence has been identified in the
5'-promoter of the rat and human VEGF gene that mediated
hypoxia-induced transcription in transient assays (60, 74). Such
sequence reveals a high degree of homology and similar protein- binding
characteristics as the hypoxia-inducible factor 1 (HIF-1) binding site
within the Epo gene, which behaves like a classic transcriptional
enhancer (75). HIF-1 has been purified and cloned as a mediator of
transcriptional responses to hypoxia and is a basic, heterodimeric,
helix-loop-helix protein (76, 77). Forsythe et al. (78)
presented more direct evidence that HIF-1 is indeed implicated in the
activation of the VEGF gene transcription during hypoxia. When reporter
constructs containing the VEGF sequences that mediate hypoxia
inducibility were cotransfected with expression vectors encoding HIF-1
subunits, reporter gene transcription was much greater than that
observed in cells transfected with the reporter alone, both in hypoxic
and normoxic conditions (78).
It has been shown that accumulation of adenosine, which occurs under
hypoxic conditions, is involved in the induction of the VEGF gene
during hypoxia (79). According to these studies, adenosine, by
activating adenosine A2 receptors, results in elevated cAMP
concentrations that in turn increase VEGF mRNA levels, possibly through
a protein kinase A-mediated pathway (79). Activation of c-Srcalso has been shown to participate in the hypoxic up-regulation of
the VEGF gene (80). Hypoxia increases the kinase activity of
pp60c-src and its phosphorylation on tyrosine 416.
Expression of a negative dominant mutant of c-Srcsignificantly reduced the hypoxic induction of VEGF (80).
It is noteworthy that several studies have shown that transcriptional
activation is not the only mechanism leading to VEGF up-regulation in
response to hypoxia. Increased mRNA stability has been identified as an
important posttranscriptional component (81, 82, 83). Sequences that
mediate increased stability were identified in the 3'-untranslated
region of the VEGF mRNA (see also Section III). Also, a
hypoxia-induced protein that bound to such sequences was identified
(81).
B. Cytokines
Several cytokines or growth factors up-regulate VEGF mRNA
expression and/or induce release of VEGF protein. Exposure of quiescent
human keratinocytes to serum, epidermal growth factor (EGF), TGF-ß,
or keratinocyte growth factor results in a marked induction of VEGF
mRNA expression (84). Also, primary, nontransformed, keratinocytes show
VEGF up-regulation in response to TGF- (85, 86). EGF also stimulates
VEGF release by cultured glioblastoma cells (87). In addition,
treatment of quiescent cultures of several epithelial and fibroblastic
cell lines with TGF-ß resulted in induction of VEGF mRNA and release
of VEGF protein in the medium (88). Based on these findings, it has
been proposed that VEGF may function as a paracrine mediator for
indirect-acting angiogenic agents such as TGF-ß (88). Furthermore,
IL-1ß induces VEGF expression in aortic smooth muscle cells (89).
Both IL-1 and PGE2 have been shown to induce expression
of VEGF in cultured synovial fibroblasts, suggesting the participation
of such inductive mechanisms in inflammatory angiogenesis (90). IL-6
has been also shown to significantly induce VEGF expression in several
cell lines (91). Not only promoter elements, but also motifs in the
5'-untranslated region of the VEGF mRNA were found to be involved in
such up-regulation (91). IGF-I, a mitogen implicated in the growth of
several malignancies, has also been shown to induce VEGF mRNA and
protein in cultured colorectal carcinoma cells (92). The induction was
mediated by a combined increase in transcriptional rate of the VEGF
gene and in the stability of the mRNA. Thus, IGF-I, in addition to its
direct mitogenic effects on malignant cells, may facilitate tumor
growth via an increase in the vascular supply, mediated by VEGF.
C. Differentiation and transformation
Cell differentiation has been shown to play an important role in
the regulation of VEGF gene expression (93). The VEGF mRNA is
up-regulated during the conversion of 3T3 preadipocytes into adipocytes
or during the myogenic differentiation of C2C12 cells. Conversely, VEGF
gene expression is repressed during the differentiation of the
pheochromocytoma cell line PC12 into nonmalignant, neuron-like cells.
These studies also indicate that induction of VEGF mRNA expression in
preadipocytes requires pathways mediated by both protein kinase C and
protein kinase A activation (93). Consistent with the presence of AP-1
and AP-2 sites in the VEGF gene promoter, phorbol esters and forskolin,
a potent activator of adenylate cyclase, induce VEGF mRNA expression
(94). Accordingly, luteotrophic hormone, a known activator of adenylate
cyclase, has been shown to induce expression of VEGF mRNA in cultured
bovine ovarian granulosa cells (94).
Specific transforming events also result in induction of VEGF gene
expression. A mutated form of the murine p53 tumor suppressor gene
(Ala135 > Val) has been shown to induce VEGF mRNA
expression and potentiate phorbol ester-stimulated VEGF mRNA expression
in NIH 3T3 cells in transient transfection assays (95). Likewise,
oncogenic mutations or amplification of ras lead to VEGF
up-regulation (96, 97). This effect is blocked by treatment with
inhibitors of ras farnesyl transferase. Interestingly,
expression of oncogenic ras, either constitutive or
transient, potentiated the induction of VEGF by hypoxia (98). Also,
overexpression of v-raf (97) or v-Src (99) lead
to VEGF up-regulation. Moreover, the von Hippel-Lindau (VHL) tumor
suppressor gene has been recently implicated in the regulation of VEGF
gene expression (100). Human renal cell carcinoma cells either lacking
endogenous wild type VHL gene or expressing an inactive mutant
demonstrated altered regulation of VEGF gene expression, which was
corrected by introduction of wild type VHL gene. Essentially all of the
endothelial cells mitogenic activity released by tumor cells expressing
mutant VHL gene was neutralized by anti-VEGF antibodies (100). These
findings suggest that VEGF is a key mediator of the abnormal vascular
proliferations and solid tumors characteristic of VHL syndrome (101).
Most recently, Iliopulos et al. (102) have shown that a
function of the VHL protein is to provide a negative regulation of a
series of hypoxia-inducible genes, including the VEGF, platelet-derived
growth factor B chain, and the glucose transporter GLUT1 genes. In the
presence of a mutant VHL, mRNAs for such genes were produced both under
normoxic and hypoxic conditions. Reintroduction of wild type VHL cDNA
resulted in inhibition of mRNA production under normoxic conditions and
restored the characteristic hypoxia inducibility of those genes (102).
Taken together, these findings indicate that several, unrelated,
alterations in cellular regulatory pathways result in VEGF
up-regulation. Therefore, this event may be a final common pathway
necessary for uncontrolled proliferation in vivo.
|
VI. The VEGF Receptors
|
|---|
A. Characterization and distribution of VEGF-binding sites
Two classes of high-affinity VEGF-binding sites were initially
described in the surface of bovine endothelial cells, with dissociation
constant (Kd) values of 10 pM and 100
pM, respectively, and molecular mass in the range of
180–220 kDa (103, 104). Lower affinity binding sites on mononuclear
phagocytes were subsequently described (105). It has been suggested
that such binding sites are involved in mediating chemotactic effects
for monocytes by VEGF (50). Recently, it has been suggested that
low-affinity, low molecular mass (120–130 kDa), receptors exist on
endothelial and tumor cells (106, 107). Such receptors cross-link
VEGF165 but not VEGF121. Thus, certain tumor
and endothelial cells express lower affinity sites that bind
selectively exon 7-encoded sequences. The molecular nature and
biological significance of these receptors remain to be elucidated.
Ligand autoradiography studies on fetal and adult rat tissue sections
demonstrated that high-affinity VEGF-binding sites are localized to the
vascular endothelium of large or small vessels in situ (108, 109). These findings represented direct evidence for the hypothesis
that the vascular endothelium is the major target of VEGF action.
Interestingly, VEGF binding was apparent not only on proliferating but
also on quiescent endothelial cells (108, 109). Also, the earliest
developmental identification of high-affinity VEGF binding was in the
hemangioblasts in the blood islands in the yolk sac, suggesting that
expression of VEGF receptors is one of the earliest events in
endothelial cell differentiation (109).
B. The Flt-1 and Flk-1/KDR tyrosine kinases
1. Binding characteristics. Two VEGF receptor tyrosine kinases
(RTKs) have been identified (110, 111, 112, 113, 114, 115, 116). The Flt-1
(fms-like-tyrosine kinase) and KDR (kinase domain region)
receptors bind VEGF with high affinity. Flk-1 (fetal liver kinase-1),
the murine homolog of KDR, shares 85% sequence identity with human KDR
(114). Both Flt-1 and KDR/Flk-1 have seven immunoglobulin (Ig)-like
domains in the extracellular domain (ECD), a single transmembrane
region and a consensus tyrosine kinase sequence that is interrupted by
a kinase-insert domain (110, 111, 112, 113, 114, 115, 116). Figure 2 shows the
alignment of the amino acid sequences of the ECD of Flt-1 and KDR.
Flt-1 has the highest affinity for rhVEGF165, with a
Kd of approximately 10–20 pM (110). KDR has a
somewhat lower affinity for VEGF: the Kd has been estimated
to be approximately 75–125 pM (111).
View larger version (53K):
[in this window]
[in a new window]
|
Figure 2. Alignment of the extracellular domains of human
Flt-1 and KDR. The seven immunoglobulin (Ig)-like domains are shown as
individual boxed areas.
|
|
A cDNA coding an alternatively spliced soluble form of Flt-1 (sFlt-1),
lacking the seventh Ig-like domain, transmembrane sequence, and the
cytoplasmic domain, has been identified in human umbilical vein
endothelial cells (117, 118). This sFlt-1 receptor binds VEGF with high
affinity (Kd 10–20 pM) and is able to inhibit
VEGF-induced mitogenesis, suggesting that it may be a physiological
negative regulator of VEGF action (117, 118).
An additional member of the family of RTKs with seven Ig-like domains
in the ECD is Flt-4 (119, 120, 121, 122) which, however, is not a receptor for
VEGF but rather binds a newly identified ligand called VEGF-C or
VEGF-related peptide (VRP) (see Section VII).
2. Signal transduction. Our understanding of the signal
transduction properties of the VEGF receptors is still incomplete. VEGF
has been shown to induce the phosphorylation of at least 11 proteins in
bovine aortic endothelial cells (113). PLC-
and two proteins that
associate with PLC- were phosphorylated in response to VEGF (123).
Furthermore, immunoblot analysis for mediators of signal transduction
that contain SH2 domains demonstrated that VEGF induces phosphorylation
of phosphatidylinositol 3-kinase, ras GTPase activating
protein, and several others. These findings suggest that VEGF promotes
the formation of multimeric aggregates of VEGF receptors with proteins
that contain SH2 domains. These studies, however, did not identify
which VEGF receptor(s) are involved in these events.
Several studies have indicated that Flt-1 and KDR have different signal
transduction properties (124, 125). Porcine aortic endothelial cells
lacking endogenous VEGF receptors display chemotaxis and mitogenesis in
response to VEGF when transfected with a plasmid coding for KDR (124).
In contrast, transfected cells expressing Flt-1 lack such responses
(124). Flk-1/KDR undergoes strong ligand-dependent tyrosine
phosphorylation in intact cells, while Flt-1 reveals a weak or
undetectable response (110, 124, 125). Also, VEGF stimulation results
in weak tyrosine phosphorylation that does not generate any mitogenic
signal in transfected NIH 3T3 cells expressing Flt-1 (125). These
findings agree with other studies showing that placenta growth factor
(PlGF), which binds with high affinity to Flt-1 but not to Flk-1/KDR,
lacks direct mitogenic or permeability-enhancing properties or the
ability to effectively stimulate tyrosine phosphorylation in
endothelial cells (126) (see Section VII). Therefore,
interaction with Flk-1/KDR is a critical requirement to induce the full
spectrum of VEGF biological responses. In further support of this
conclusion, VEGF mutants that bind selectively to Flk-1/KDR are fully
active endothelial cell mitogens (see Section VI.B.5) (127).
Furthermore, Kendall et al. (118) suggested that sFlt-1 may
form heterodimeric complexes with KDR, which could potentially exert a
dominant-negative effect on KDR signal transduction. These findings
contributed to cast doubt on the role of Flt-1 as a truly signaling
receptor. However, more recent evidence indicates that Flt-1 indeed
signals, although our understanding of these processes is clearly
fragmentary. Cunningham et al. (128), using the yeast
two-hybrid system, have demonstrated an interaction between Flt-1 and
the p85 subunit of phosphatidylinositol 3-kinase. Mutagenesis analysis
revealed that change of a tyrosine residue at position 1213 to
phenylalanine completely abolished such interaction. These data suggest
that p85 couples Flt-1 to intracellular signal transduction systems and
implicate elevated levels of PtdIns(3, 4, 5)P3 levels in this process
(128). Also, members of the Src family, such as Fynand Yes, show an increased level of phosphorylation
after VEGF stimulation in transfected cells expressing Flt-1 but not
KDR (124). Furthermore, Barleon et al. (129) have shown that
a specific biological response, the migration of monocytes in response
to VEGF (or PlGF), is mediated by Flt-1. However, the most compelling
evidence so far for an important biological role played by the Flt-1
receptor has been provided by gene knockout studies (see Section
VIII. B).
3. Regulation. The expression of Flt-1 and Flk-1/KDR genes is
largely restricted to the vascular endothelium (see Section
VIII.A). The promoter region of Flt-1 has been cloned and
characterized and a 1-kb fragment of the 5'-flanking region essential
for endothelial-specific expression was identified (130). Likewise, a
4-kb 5'-flanking sequence has been identified in the promoter of KDR
that confers endothelial cell-specific activation (131).
Similarly to VEGF, hypoxia has been proposed to play an important role
in the regulation of VEGF receptor gene expression. Exposure of rats to
acute or chronic hypoxia led to pronounced up-regulation of both Flt-1
and Flk-1/KDR genes in the lung vasculature (132). Also, Flk-1/KDR and
Flt-1 mRNAs were substantially up-regulated throughout the heart after
myocardial infarction in the rat (133). However, in vitro
studies have yielded unexpected results. Even though Thieme et
al. (134) have shown that hypoxia increases VEGF receptor number
by 50% in cultured bovine retinal capillary endothelial cells, the
expression of KDR is not induced but paradoxically shows an initial
down-regulation (135). Brogi et al. (136) have proposed that
the hypoxic up-regulation of KDR observed in vivo is not
direct but requires the release of an unidentified paracrine mediator
from ischemic tissues. Also, recent studies have shown that both
TNF- (137) and TGF-ß (138) are able to inhibit the expression of
the KDR gene in cultured endothelial cells.
4. Structural requirements for ligand binding in Flt-1 and
KDR. As noted above, the VEGF receptors have seven Ig-like domains
in the ECD. Until now, the significance and function of these domains
for ligand binding and receptor activation were unknown. Recently, the
domains in the ECD of Flt-1 and KDR responsible for specific ligand
recognition were identified by constructing and analyzing a variety of
receptor variants (139). These included individual Ig-like domain (140)
deletions, as well as chimeras in which domains of either KDR or Flt-4
were exchanged for the homologous sequences from Flt-1. Deletion of the
second Ig-like domain of human Flt-1 completely abolishes VEGF binding
(Fig. 3). Introduction of the second domain of KDR into an
Flt-1 mutant lacking the homologous domain restored VEGF binding.
However, PlGF was unable to displace VEGF bound to such mutant, a
pattern characteristic of the KDR but not the Flt-1 receptor (Fig. 3).
Also, "swap" experiments in which the second Ig-like domain of
Flt-1 replaced the corresponding domain in Flt-4 demonstrated that such
a chimeric receptor had the ability to bind VEGF with affinity nearly
identical to that of wild type Flt-1. Furthermore, transfected cells
expressing this chimeric Flt-4 receptor exhibited increased DNA
synthesis in response to both VEGF and PlGF (139). Thus, VEGF binding
to domain 2 of Flt-1 is able to initiate a signal transduction cascade,
even in the context of the ECD of a foreign receptor. Further studies
are required to elucidate the significance of the remaining Ig-like
domains in receptor dimerization (141) and in coupling binding with
signal transduction.
View larger version (28K):
[in this window]
[in a new window]
|
Figure 3. The second Ig-like domain contains the major
determinants for binding and ligand specificity in the VEGF receptors.
In panel A, Flt-1-IgG individual domain deletion variants (5 ng per
reaction) were tested for their ability to bind
[125I]VEGF165 in the absence (striped
bars) or presence of 50 ng cold VEGF165
(solid bars). Deletion of the second Ig-like domain
completely abolishes the binding of VEGF. The second Ig-like domain of
KDR was cloned into the Flt-1 domain 2 deletion construct to produce
"swap" mutants (panel B). Replacement of the second domain of Flt-1
with the homologous domain of KDR reestablished VEGF-binding. However,
PlGF152 (open bars) could not displace
VEGF165 bound to Flt.K2, a pattern characteristic of the
KDR but not the Flt-1 receptor.
|
|
5. VEGF determinants for binding Flt-1 and KDR. Site-directed
mutagenesis has been used to localize the determinants on VEGF that
mediate binding to the KDR and Flt-1 receptors. Alanine-scanning
analysis was performed to identify a positively charged surface in VEGF
that mediates receptor binding (127, 142). A model based on the crystal
structure of PDGF-BB was used (143). Arg82,
Lys84, and His86, located in a hairpin loop,
were found to be critical for binding KDR, while negatively charged
residues, Asp63, Glu64, and Glu67,
were primarily responsible for Flt-1 binding. The single mutations
R82A, K84A, and H86A were found to display modestly decreased KDR
binding. The triple mutants involving alanine replacement or
neo-glycosylation sites, R82A, K84A, H86A VEGF and R82N, I83L, K84S
VEGF, exhibited minimal binding to KDR receptor. The half-maximally
effective concentrations (EC50) to stimulate bovine
capillary endothelial cell growth for most of the VEGF mutants were
similar to those observed for wild type VEGF. The most significant
effect on endothelial cell proliferation was observed with mutations in
the 82–86 region. The EC50 of R82A, K84A, H86A VEGF
increased 20-fold such that mitogenic potency of this mutant was
decreased to 5% of wild type VEGF. In contrast, the mutants that
failed to bind Flt-1 were fully active endothelial cell mitogens (127, 142).
|
VII. VEGF-Related Molecules
|
|---|
Over the last few years, three VEGF-related genes have been
identified from mammalian sources. The encoded factors are known as
PlGF, VEGF-B, and VEGF-C/VRP. In addition, two sequences in the genome
of the parapoxvirus orf virus show homology to VEGF. Figure 4 shows the alignment of the amino acid sequences of these
molecules with the sequence of VEGF165. Although the
biological role of these factors is still largely unclear, their
structural homology to VEGF suggests that they may play a role in the
regulation of blood vessel growth. The first VEGF-related factor
identified is PlGF. This molecule shares a 53% identity with the
PDGF-like region of VEGF. The encoded protein was expected to have 149
amino acids, including the signal peptide (144). Subsequently, a longer
form characterized by a 21-amino acid insertion was identified (145).
Similar to the 24-amino acid insertion in the longer forms of VEGF,
this insertion is highly enriched in basic residues. These two
isoforms, which arise from alternative splicing of mRNA, are known as
PlGF-1 and PlGF-2 or PlGF131 and PlGF152,
respectively. Similar to VEGF, these molecules are dimeric
glycoproteins. Park et al. (126) have shown that PlGF binds
with high affinity (Kd 250 pM) Flt-1 but not
KDR. Purified PlGF demonstrated minimal activity in vascular
endothelial cell growth and vascular permeability assays, suggesting
that binding to KDR is a requirement for both activities. However, PlGF
was able to potentiate the bioactivity of low, marginally efficacious,
concentrations of VEGF, both on endothelial cell growth and on vascular
permeability (126). The molecular basis of this effect remains to be
fully elucidated. Interestingly, naturally occurring heterodimers
between VEGF and PlGF have been identified in the conditioned medium of
a rat glioma cell line (146). In agreement with previous studies, the
PlGF homodimer demonstrated minimal mitogenic activity on endothelial
cells. However, the VEGF:PlGF heterodimer was active, although its
potency was approximately 7-fold lower than the VEGF homodimer. It has
been suggested that the formation of heterodimers with PlGF constitutes
a mechanism of negative regulation of VEGF bioactivity, by shifting the
balance toward less potent molecules (147).
View larger version (42K):
[in this window]
[in a new window]
|
Figure 4. Amino acid sequence of VEGF165 and
VEGF-related molecules: VEGF-B, VEGF-C, PlGF152, and a
VEGF-like sequence identified in the genome of the parapoxvirus orf
virus. The conserved cysteine residues are boxed.
|
|
As previously noted, similarly to the VEGF receptors, Flt-4 is a
RTK with seven Ig-like domains in the ECD (119, 120, 121, 122). Interestingly,
the expression of Flt-4 mRNA, which is initially localized to
angioblasts and venules in the early embryo, becomes restricted to
lymphatic endothelium at later stages of development (148). This
expression pattern suggested that Flt-4 may play a role in the
regulation of lymphangiogenesis (122, 148). A ligand selective for
Flt-4 has been recently identified by two groups and has been named
VEGF-C (149) or VRP (150). VEGF-C/VRP is a secreted protein with 399
amino acid residues and has a 32% identity to VEGF. Its COOH-terminal
half contains a 180-amino acid region that is not found in VEGF. This
region contains cysteine-rich motifs similar to a protein component of
silk produced by the larval salivary gland of the midge C.
tentanus (149, 150). VEGF-C/VRP has been reported to stimulate the
growth of human lung endothelial cells, albeit at 100-fold less potency
than VEGF165 (150). It is still unclear whether VEGF-C/VRP
is capable of high-affinity interaction with KDR (149, 150).
A newly identified member of the VEGF gene family is VEGF-B (151, 152).
This molecule consists of 188 amino acids, including the signal
peptide. VEGF-B has been reported to stimulate the growth of human and
bovine vascular endothelial cells (151). Interestingly, VEGF-B is
distributed primarily in the skeletal muscle and myocardium and is
coexpressed with VEGF (151). Similar to the long forms of VEGF, VEGF-B
is expressed as a membrane-bound protein that can be released in a
soluble form after addition of heparin. VEGF-B and VEGF are also able
to form heterodimers, when coexpressed (151). These findings led to the
hypothesis that VEGF-B may participate in the regulation of
angiogenesis, particularly in muscle (151). Figure 5
schematizes the interaction of VEGF and VEGF-related factors with their
tyrosine kinase receptors.
View larger version (39K):
[in this window]
[in a new window]
|
Figure 5. The diagram illustrates the interaction of VEGF
and VEGF-related molecules with the three known members of the family
of RTKs with seven Ig-like domains in the ECD. VEGF interacts with
Flt-1 and KDR; PlGF binds only Flt-1 and VEGF-C/VRP binds with high
affinity to Flt-4. It is unknown at the present time whether VEGF-B
binds to any of these receptors.
|
|
Intriguingly, two sequences having a significant homology to VEGF have
been identified in the genome of two different strains of
orf virus, a parapoxvirus that affects goats, sheeps, and
occasionally humans (153). This suggests that the viral VEGF-like gene
has been acquired from a mammalian host and is undergoing genetic
drift. Interestingly, the lesions of goats and humans after
orf virus infection are characterized by extensive
microvascular proliferation in the skin, raising the possibility that
the product of the viral VEGF-like gene is responsible for such
lesions.
|
VIII. Role of VEGF and Its Receptors in Physiological Angiogenesis
|
|---|
A. Distribution of VEGF, Flk-1/KDR, and Flt-1 mRNA
The proliferation of blood vessels is crucial for a wide variety
of physiological processes such as embryonic development, normal growth
and differentiation, wound healing, and reproductive functions.
Previous studies have indicated that the VEGF mRNA is temporally and
spatially related to the proliferation of blood vessels in the rat,
mouse, and primate ovary and in the rat uterus, suggesting that VEGF is
a mediator of the cyclical growth of blood vessels that occurs in the
female reproductive tract (154, 155, 156, 157). In fact, in situ
hybridization studies in the rat ovary provided the first evidence that
VEGF may be a regulator of physiological angiogenesis (154).
During embryonic development, VEGF expression is first detected within
the first few days after implantation in the giant cells of the
trophoblast (109, 158), suggesting a role for this factor in the
induction of vascular growth in the decidua, placenta, and vascular
membranes. At later developmental stages in the mouse or rat embryos,
the VEGF mRNA is expressed in several organs, including heart,
vertebral column, kidney, and along the surface of the spinal cord and
brain (109, 158). In the developing mouse brain, the highest levels of
mRNA expression are associated with the choroid plexus and the
ventricular epithelium (158). In the human fetus (16–22 weeks), VEGF
mRNA expression is detectable in virtually all tissues and is most
abundant in lung, kidney, and spleen (159). VEGF protein, as assessed
by immunocytochemistry, is expressed in epithelial cells and myocytes,
but not vascular endothelial cells (159).
In situ hybridization studies have shown that the Flk-1 mRNA
is expressed in the yolk sac and intraembryonic mesoderm and later on
in angioblasts, endocardium, and small and large vessel endothelium
(115, 116). There is evidence that the Flk-1 mRNA is down-regulated in
adult endothelial cells as compared with fetal endothelial cells (115, 116). These findings strongly suggested a role for Flk-1 in the
regulation of vasculogenesis and angiogenesis. Other studies have
demonstrated that expression of Flk-1 mRNA is first detected in the
proximal-lateral embryonic mesoderm, which gives rise to the heart
(160). Flk-1 is then detectable in endocardial cells of heart primordia
and subsequently in the major embryonic and extraembryonic vessels
(160). These studies have indicated that Flk-1 may be the earliest
marker of endothelial cell precursors (160). The Flt-1 mRNA is
selectively expressed in vascular endothelial cells, both in fetal and
adult mouse tissues (161). Similar to the high-affinity VEGF binding
(108, 109), the Flt-1 mRNA is expressed in both proliferating and
quiescent endothelial cells (161), suggesting a role for Flt-1 in the
maintenance of endothelial cells.
Interestingly, VEGF expression is also detectable around microvessels
in areas where endothelial cells are normally quiescent, such as kidney
glomerulus, pituitary, heart, lung, and brain (61, 162, 163). These
findings raised the possibility that VEGF may be required not only to
induce active vascular proliferation but, at least in some
circumstances, also for the maintenance of the differentiated state of
blood vessels (61). In agreement with this hypothesis, Alon et
al. (164) have shown that VEGF acts as a survival factor, at least
for the developing retinal vessels. They propose that hyperoxia-induced
vascular regression in the retina of neonatal animals is a consequence
of inhibition of VEGF production by glial cells. Accordingly,
intraocular administration of VEGF to newborn rats at the onset of
hyperoxia was able to prevent cell apoptosis and regression of the
retinal vasculature (164).
It has been suggested that VEGF is also involved in a major
pathophysiological process such as wound healing (84, 85, 86).
Keratinocytes in a healing wound express VEGF mRNA. Interestingly, a
decreased expression of VEGF mRNA has been observed in the skin of
genetically diabetic db/db mice (84), suggesting that an altered
regulation of VEGF gene expression contributes to defective
angiogenesis and impaired wound healing characteristic of this
disorder.
B. Analysis of Flk-1/KDR, Flt-1, and VEGF gene knockouts
Recent studies have demonstrated that both Flt-1 and Flk-1/KDR are
essential for normal development of embryonic vasculature. However,
their respective roles in endothelial cell proliferation and
differentiation appear to be distinct (165, 166). Mouse embryos
homozygous for a targeted mutation in the Flt-1 locus died in
utero between day 8.5 and 9.5 (165). Endothelial cells developed
in both embryonic and extraembryonic sites but failed to organize in
normal vascular channels. Mice in which the Flk-1 gene had been
inactivated lacked vasculogenesis and also failed to develop blood
islands (166). Hematopoietic precursors were severely disrupted and
organized blood vessels failed to develop throughout the embryo or the
yolk sac, resulting in death in utero between day 8.5 and
9.5 (166).
However, these findings do not necessarily imply VEGF as being equally
essential, since other ligands might potentially activate the Flt-1 and
Flk-1/KDR receptors and thus substitute VEGF action. Very recent
studies (16, 17) have generated direct evidence for the role played by
VEGF in embryonic vasculogenesis and angiogenesis. Unexpectedly,
inactivation of the VEGF gene in mice resulted in embryonic lethality
in heterozygous embryos, between day 11 and 12. The VEGF+/- embryos
were growth retarded and also exhibited a number of developmental
anomalies (167). The forebrain region appeared significantly
underdeveloped. In the heart region, the outflow region was grossly
malformed; the dorsal aortas were rudimentary, and the thickness of the
ventricular wall was markedly decreased. The yolk sac revealed a
substantially reduced number of nucleated red blood cells within the
blood islands. Also, the vitelline veins failed to fuse with the
vascular plexus of the yolk sac. Significant defects in the vasculature
of other tissues and organs, including placenta and nervous system,
were evidenced. For example, in the nervous system of heterozygous
embryos at day 10.5, vascular elements could be demonstrated in the
mesenchyme but not in the neuroepithelium (17) (Fig. 6).
This failure of blood vessel ingrowth was accompanied by apoptosis and
disorganization of neuroepithelial cells (Fig. 6). The VEGF+/- embryos
survive approximately 2 days longer than the Flt-1 or Flk-1/KDR null
embryos, presumably reflecting a partial activation of these tyrosine
kinases by VEGF. In situ hybridization confirmed expression
of VEGF mRNA in heterozygous embryos. Thus, the VEGF+/- phenotype is
due to gene dosage and not to maternal imprinting.
View larger version (114K):
[in this window]
[in a new window]
|
Figure 6. Hematoxylin and eosin staining (upper
panels) and CD34 immunostaining (lower panels)
on sections of neuroepithelium (ne) from wild type
(left) and VEGF+/- (right) E 10.5 mouse
embryos. Arrows indicate blood vessels. Blood vessel
lumina can be identified in the mesenchyme adjacent to the ne in both
groups. However, they are absent within the ne of the heterozygous
embryos. Note also the presence of apoptotic cells in the ne of the
heterozygous embryos. This contrasts with the well differentiated and
vascularized ne in the wild type. [Reproduced with permission from N.
Ferrara et al.: Nature 380:439–442, 1996
(17). ©1996 Macmillan Magazines Limited.]
|
|
Although several heterozygous phenotypes have been described (168),
this may be the first report that the loss of a single allele of a gene
that is not maternally imprinted can be lethal. Therefore, VEGF and its
receptors are essential for blood island formation and angiogenesis
such that even reduced concentrations of VEGF are inadequate to support
a normal pattern of development. These findings also indicate that, in
the VEGF+/- mutant, the mechanisms that normally up-regulate VEGF gene
expression, such as hypoxia, are unable to provide an effective
compensatory response. It is tempting to speculate that, as VEGF
concentrations and angiogenic gradients fall below a threshold during
critical periods, this can cause irreversible disruption of normal
organogenesis.
|
IX. Role of VEGF in Pathological Angiogenesis
|
|---|
A. Tumor angiogenesis
1. Expression of VEGF in human tumors. In
situ hybridization studies have demonstrated that the VEGF mRNA is
markedly up-regulated in the vast majority of human tumors so far
examined. These include: lung (169), thyroid (170), breast (171, 172),
gastrointestinal tract (173, 174), kidney and bladder (175), ovary
(176), and uterine cervix (177) carcinomas, angiosarcoma (178), germ
cell tumors (179), and several intracranial tumors including
glioblastoma multiforme (68, 180, 181) and sporadic, as well as VHL
syndrome-associated, capillary hemangioblastoma (182, 183) (Table 1). Only sections of lobular carcinoma of the breast
and papillary carcinoma of the bladder failed to show significant VEGF
mRNA expression (184). As already indicated in Section V.B,
the expression of VEGF in glioblastoma multiforme and other tumors with
significant necrosis is highest in hypoxic tumor cells adjacent to
necrotic areas (68, 180, 181). A correlation has been noted between
VEGF mRNA expression and vascularity of the tumor (169, 174, 177, 182, 183). In the tumors where VEGF and PlGF were coexpressed, only VEGF
expression correlated with the degree of malignancy and vascularity
(170, 179). In virtually all specimens examined, the VEGF mRNA was
expressed in tumor cells but not in endothelial cells. In contrast, the
mRNAs for Flt-1 and KDR were up-regulated in the endothelial cells
associated with the tumor (173, 180, 185). These findings are
consistent with the hypothesis that VEGF is primarily a paracrine
mediator (186). An interesting exception may be angiosarcoma, where
VEGF and Flt-1 mRNA were found to be coexpressed in angiosarcoma cells,
raising the possibility that in this malignancy VEGF may play a role as
an autocrine factor (178). Angiosarcoma cells, however, arise from the
endothelium. Recently, Freeman et al. (187) have suggested
that lymphocytes infiltrating the tumor may constitute an additional
source of VEGF, which contributes to tumor angiogenesis.
Immunohistochemical studies have localized the VEGF protein not
only to the tumor cells but also to the vasculature (173, 180, 185).
This finding indicates that tumor-secreted VEGF accumulates in the
target cells. Ultrastructural studies have localized VEGF bound to
tumor endothelial cells to the abluminal plasma membrane and to the
recently described vesiculovacular organelles, cytoplasmic structures
that are thought to be involved in macromolecular transport across the
tumor endothelium (188).
Elevations in VEGF levels have been detected in the serum of some
cancer patients (189). Also, a correlation has been observed between
VEGF expression and microvessel density in primary breast cancer
sections (190). A postoperative survey indicated that the relapse-free
survival rate of patients with VEGF-rich tumors was significantly worse
than that of VEGF-poor tumors, suggesting that expression of VEGF is
associated with stimulation of angiogenesis and with early relapse in
primary breast cancer (190). A similar correlation has been described
in gastric carcinoma patients (191). VEGF positivity in tumor sections
was correlated with vessel involvement, lymph node metastasis, and
liver metastasis. Furthermore, patients with VEGF-positive tumors had a
worse prognosis than those with VEGF-negative tumors (191).
2. Inhibition of VEGF action in vivo. The
availability of specific monoclonal antibodies capable of inhibiting
VEGF-induced angiogenesis in vivo and in vitro
(192) made it possible to generate direct evidence for a role of VEGF
in tumorigenesis. In a study published by Kim et al. in 1993
(193) , such antibodies were found to exert a potent inhibitory effect
on the growth of three human tumor cell lines injected subcutaneously
in nude mice, the SK-LMS-1 leiomyosarcoma, the G55 glioblastoma
multiforme, and the A673 rhabdomyosarcoma. The growth inhibition ranged
between 70% and more than 95%. Figure 7 illustrates the
effects of the anti-VEGF-neutralizing antibody on the in
vivo growth of such cell lines. These findings provided the first
direct demonstration that inhibition of the action of an endogenous
endothelial cell mitogen may result in suppression of tumor growth
in vivo. Subsequently, other tumor cell lines were found to
be inhibited in vivo by this treatment (194, 195, 196, 197, 198) (Table 2).
View larger version (20K):
[in this window]
[in a new window]
|
Figure 7. Effects of anti-VEGF monoclonal antibody on tumor
size (A, B) and weight (C). A673, G55, and SK-LMS-1 cells were injected
subcutaneously in nude mice. Animals were then treated with anti-VEGF
neutralizing antibody (A.4.6.1) or a control antibody (5B6) twice
weekly intraperitoneally, at the indicated doses. In A and B, tumor
size was measured weekly. Panel C illustrates the weight of the tumors
at the end of the experiment. Data shown reflect the response to 100
µg (5 mg/kg) of antibody twice weekly. Plus and minus signs denote
the presence or absence of antibody treatment. A673 and G55 tumors were
collected 4 weeks after tumor cell injection. SK-LMS-1 tumors were
harvested after 10 weeks. [Reproduced with permission from J. Kim
et al.: Nature 362:841–844, 1993 (193).
© 1993 Macmillan Magazines Limited.]
|
|
In agreement with the hypothesis that inhibition of neovascularization
is the mechanism of tumor suppression, the density of blood vessels was
significantly lower in sections of tumors from antibody-treated animals
as compared with controls (193, 194). Furthermore, neither the
antibodies nor VEGF had any effect on the in vitro growth of
the tumor cells (193, 194). Intravital videomicroscopy techniques have
allowed a more direct verification of the hypothesis that anti-VEGF
antibodies indeed block tumor angiogenesis (195). Tumor spheroids of
A673 cells were implanted in dorsal skinfold chambers inserted in nude
mice. Noninvasive imaging of the vasculature revealed a nearly complete
suppression of tumor angiogenesis in anti-VEGF treated animals as
compared with controls, at all time points examined. These findings
were corroborated by histological analysis that showed a dramatic
difference in the density of CD34-positive vascular elements between
the two groups (195). Thus, inhibition of VEGF-induced angiogenesis
caused a dramatic change in growth characteristics of this cell line,
from a rapidly growing malignancy to a dormant tumor seedling. Very
similar findings were subsequently obtained with other tumor cell lines
(P. Borgström et al., submitted).
Warren et al. (194) have demonstrated that VEGF is a
mediator of the in vivo growth of human colon carcinoma HM7
cells in an orthotopic nude mouse model of liver metastasis. Similar to
human tumors, in this murine model the expression of Flk-1 mRNA was
markedly up-regulated in the vasculature associated with liver
metastases. Treatment with anti-VEGF monoclonal antibodies resulted in
a dramatic decrease in the number and size of metastases. Most of the
tumors in the treated group were less than 1 mm in diameter and all
were less than 3 mm. Also, neither blood vessels nor Flk-1 mRNA
expression could be demonstrated in such metastases. Also,
administration of anti-VEGF-neutralizing antibodies inhibited primary
tumor growth and metastasis of A431 human epidermoid carcinoma cells in
scid mice (196) or HT-1080 fibrosarcoma cells implanted in
BALB/c nude mice (197).
Recently, Borgström et al. (submitted) have shown that
a combination treatment that includes anti-VEGF monoclonal antibody and
doxorubicin results in a significant enhancement of the efficacy of
either agent alone and led in some cases to complete regression of
tumors derived from MCF-7 breast carcinoma cells in nude mice.
Combination treatments that include anti-VEGF monoclonal antibody and
cisplatin have resulted in similar enhancement of the efficacy of each
agent (our unpublished observations).
Intravital fluorescence microscopy and video imaging analysis have been
also applied to address the important issue of the effects of VEGF on
permeability and other properties of tumor vessels (198). Three
different human tumor cell lines (U87, P-MEL, and LS174T) were
implanted in two locations in immunodeficient mice, the cranium and the
dorsal skinfold. Treatment with an anti-VEGF monoclonal antibody (192)
was initiated when the tumor xenografts were already established and
vascularized and resulted in time-dependent reductions in vascular
permeability (198). These effects were accompanied by striking changes
in the morphology of vessels, with dramatic reduction in diameter and
tortuosity (198). This reduction in diameter is expected to block the
passage of blood elements and eventually stop the flow in the tumor
vascular network. Accordingly, a regression of blood vessels was
observed after repeated administrations of anti-VEGF antibody. These
findings led to the intriguing conclusion that tumor vessels require
constant stimulation with VEGF to maintain not only their proliferative
properties but also some key morphological features (198).
An additional verification of the hypothesis that VEGF action is
required for tumor angiogenesis has been provided by the finding that
retrovirus-mediated expression of a dominant negative Flk-1 mutant,
which inhibits signal transduction through wild type Flk-1/KDR
receptor, suppresses the growth of glioblastoma multiforme as well as
other tumor cell lines in vivo (
Top
Abstract
I. Introduction
