We investigated the mechanisms of vascularization in a brain metastases model of malignant melanoma. Parenchymal metastases expressing little vascular endothelial growth factor-A (VEGF-A) co-opted the preexistent brain vasculature, leading to an infiltrative phenotype. Metastases of the human melanoma cell line Mel57, engineered to express recombinant VEGF-A165, showed accelerated growth in a combined expansive and infiltrative pattern with marked central necrosis. This difference in growth profile was accompanied by dilation of co-opted intra- and peritumoral vessels with concomitant induction of vascular permeability. Our data show that modulation of preexistent vasculature can contribute to malignant progression without induction of sprouting angiogenesis.

Brain metastases of malignant melanoma are a life-threatening complication of this disease. Primary and metastatic tumors of the CNS4 are generally highly vascularized; therefore, antiangiogenic therapy has been considered promising (1). However, the brain itself has a dense vascular bed, and primary CNS tumors often show a highly infiltrative growth pattern. This raises the question of whether the vessels in such tumors are formed by neoangiogenesis or are preexistent ones (2). Indeed, the use of preexistent vessels in early metastatic outgrowths in the brain, a process referred to as co-option, has been reported (3, 4, 5). This was followed by a switch to an angiogenic phenotype. Vessel co-option as a means of tumor vascularization might have important consequences for diagnosis and therapy. For example, in contrast-enhanced magnetic resonance imaging, tumor lesions are visualized by virtue of tumor-induced vascular changes such as hyperpermeability, leading to extravasation of contrast agent. Although vascular hyperpermeability is considered to be a characteristic of tumor neovasculature, it is unclear whether co-opted vessels in CNS malignancies are modified in this respect. In addition, vessel co-option obviously will have important consequences for application of tumor therapies using angiogenesis inhibitors. In this regard, it is important to realize that most studies with antiangiogenic compounds have been carried out in animal models where tumors are grown s.c. Especially for CNS tumor biology, s.c. models have only limited clinical relevance because the s.c. space is essentially avascular. s.c. tumors will therefore be artificially selected to grow in an angiogenesis-dependent fashion, whereas the microenvironmental conditions that exist in highly vascularized tissues such as the brain are bypassed.

In previous work, we showed that the human melanoma cell line Mel57 metastasizes to mouse brain parenchyma (6) after injection into the internal carotid artery (6, 7, 8). In vitro, Mel57 expresses very low amounts of VEGF-A (9). We report here on the mechanisms of tumor vascularization in this model and on the effects of VEGF165 expression on tumor growth and vascular parameters. We show that expression of VEGF165 induced significant progression of tumor growth that was not associated with classical (i.e., sprouting) angiogenesis but was caused by architectural and functional changes of the co-opted preexistent brain vasculature.

Mice.

Specific pathogen-free, male BALB/c nu/nu mice, 6–8 weeks of age and weighing 18–25 g, were purchased from the university central animal facility. Five mice/cage were housed under specific pathogen-free conditions (temperature, 20–24°C; relative humidity, 50–60%; 15 air changes/hour; light/dark periods, 14 h/10 h). Water and food (RMH, Hope Farms, the Netherlands) were available to the animals ad libitum. Experiments were carried out in accordance with the national animal protection laws.

Cell Lines, Transfections, and Microsurgical Injections.

Human melanoma cell lines Mel57, M14, and 530 were cultured in DMEM (Life Technologies, Inc., Breda, the Netherlands), supplemented with 10% FCS, streptomycin, and penicillin at 37°C. Mel57 cells were transfected using Fugene (Roche, Mannheim, Germany) with plasmid pVEGF-IRESneo or pEGFP-IRESneo (enhanced green fluorescent protein) as a control, according to the manufacturer’s guidelines. pVEGF-IRESneo carries the cDNA for human VEGF165, slightly modified at 3′ primed end for cloning purposes, cloned in the EcoRI-BamHI sites of vector pIRESneo (Clontech, Palo Alto, CA) under control of the cytomegalovirus promoter. Two days after transfection, selection of transfectants was started using 400 μg/ml G418. The use of an IRES to generate neomycin resistance led in our hands to >95% positivity for the recombinant protein in G418-resistant cells. Therefore, after 2 weeks of selection, colonies of transfected cells were pooled, expanded, and frozen. Levels of recombinant VEGF in conditioned medium were determined using Western blot analysis as described (9). Metastasis was induced as described previously (6) by microsurgical injection of tumor cell suspensions into the right internal carotid artery of anesthetized nude mice

Histological and Immunohistochemical Analysis.

Mice were sacrificed after development of severe cachexia or acute neurological deficits. For immunohistochemistry, material was snap-frozen in liquid nitrogen or fixed in formalin and embedded in paraffin. Sections of 4 μm underwent conventional H&E staining. The brains of animals bearing lesions of the Mel57 cell lines, sacrificed 20–22 days after tumor cell injection, were cut semiserially, and lesion sizes were determined using measurement oculars. Antibodies used were antimurine CD31 (Mec 7.46; Hycult Biotechnology, Uden, the Netherlands), antimouse tight junction protein ZO1 (mAB1520; Chemicon, Temecula, CA), rabbit antimouse Ki-67 (Dianova, Hamburg, Germany), mouse antihuman α-smooth muscle actin (α-Sm1; Sigma Chemical Co., Zwijndrecht, the Netherlands), anti-Glut-1 (Dako, Glostrup, Denmark), rabbit anti-KDR (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-angiopoietin I (Santa Cruz Biotechnology), and rabbit anti-angiopoietin II (Santa Cruz Biotechnology). Frozen 4-μm sections were fixed in acetone for 10 min, dried, and incubated with antibody for 1 h at room temperature in PBS containing 1% BSA (PBS/BSA). After washing with PBS, bound antibodies were detected with a peroxidase-conjugated secondary antibody (Vector, Burlingame, CA) using the Vectastain elite ABC kit (Vector). Sections were counterstained with hematoxylin. Integrity of the BBB was investigated by staining for extravasated mouse immunoglobulins as described previously (10). In all stainings, a negative control was included in which primary antibodies were omitted. These controls were always negative.

Tumor Cell Proliferation Assays.

The S-phase marker BrdUrd (Sigma Chemical Co.) was administered i.p. at a dose of 100 mg/kg in 0.5 ml 0.9% NaCl 15 min prior to sacrifice. The proliferation index of individual Mel57-lesions (n = 3) and Mel57-VEGF-lesions (n = 6) in different mice was determined in frozen brain sections by quantification of incorporated BrdUrd as described previously (11). Tumor cell proliferation in vitro was determined as follows. Cells (5 × 103) were seeded in six-well culture plates and cultured in serum-free medium. After 2, 3, and 4 days, cells were trypsinized and counted using a Coulter counter. All experiments were performed in triplicate.

Growth of Mel57 Lesions in Brain Parenchyma.

Three to 5 weeks after injection of Mel57 cells in the right internal carotid artery, animals started to suffer from cachexia. Morphological analysis showed that numerous parenchymal lesions with diameters up to 3 mm were present in each brain, with preferential localization in the right side of the brain (cerebrum, cerebellum, and brain stem). These tumors exhibited an infiltrative growth pattern in the parenchyma (Fig. 1,A). Tumor cells exploited the preexistent brain vessels by growing in the space of Virchow-Robin (forming lesions containing remnants of preexistent brain parenchyma (Fig. 1,A, inset). Tumor growth in the space of Virchow-Robin was also seen in metastases derived from the human melanoma cell lines M14 and 530 (Fig. 1, E and F). The cell line Mel57 was chosen for further studies. We did not detect differences between the vasculature within the lesions and that in the surrounding normal brain tissue with regard to vessel diameter (Fig. 1,C), activation status of the endothelium (assessed by KDR/Flk-1, Fig. 2,G; and CD105 expression, not shown), and maturation status (assessed by pericyte coverage, Fig. 2,I). Also, intratumoral vessels still had all characteristics of an intact BBB, as demonstrated by unaltered morphology (Fig. 1,C), lack of mouse IgG extravasation (Fig. 2,A), and the presence of endothelial Glut-1 (Fig. 2,C) and ZO-1 (Fig. 2,E) at levels similar to those found in normal brain parenchyma. Vessel density within the lesions was slightly lower than in the surrounding normal brain parenchyma (Fig. 1 C), indicating that there was no angiogenic compensation for the tissue volume increase at the lesion site. This lack of angiogenesis was in line with the overall absence of vascular changes. Despite the absence of neovascularization, necrosis was scarce in these lesions, implying that the blood supply from the preexistent vascular bed sufficed to provide the tumor cells with oxygen and nutrients. Although injection of higher numbers of cells led to a proportionally higher number of lesions, the morphology of the individual lesions was not affected (not shown). All together, the infiltrative character, the low intratumoral vessel density, and the intact BBB strongly suggested that these lesions grew by mere vessel co-option without induction of an angiogenic switch.

There were neither qualitative nor quantitative differences in tumor growth patterns between parental Mel57 and control EGFP transfectants (data not shown).

Effects of VEGF165 on Growth of Mel57 Lesions.

Because Mel57 produces low levels of VEGF in vitro (∼30 pg/106 cells/24 h; data not shown) and sprouting angiogenesis was absent in Mel57 CNS lesions, we were interested in the effects of expression of this potent angiogenic factor in this model. To this end, we created stable transfectants of Mel57, which produced high levels of VEGF165 (∼30 ng/106 cells/24 h) and analyzed the growth profile. Mel57-VEGF165 developed a growth pattern that was completely different from parental tumor cells. These lesions had a more solid and expansive, rather than infiltrative, growth pattern (Fig. 1,B), although at the tumor rim infiltration into the parenchyma, again along preexisting vessels, was still present (Fig. 1,B, inset). Similar to Mel57 lesions, the intratumoral vessel density was lower than in the surrounding brain parenchyma (Fig. 1,D). However, now there were marked differences between intra/peritumoral vessels and extratumoral vessels. (Peri-)tumoral vessels were irregularly dilated (Fig. 1,D) and showed up-regulation of KDR (Fig. 2,H) and CD105 (not shown) expression, indicating that VEGF had induced an activated state of the endothelium. Vessel dilation gradually decreased with increasing distance from the lesion (see Fig. 1,D, inset), pointing at tumor-derived VEGF as the causative factor. Staining for mouse immunoglobulins indicated high permeability of blood vessels in the tumor and at the tumor periphery (Fig. 2, compare B with A). The dilated vessels still stained positive for the BBB markers Glut-1 and ZO-1 (Fig. 2, D and F), indicating that these vessels are truly preexistent and not neoangiogenic. Interestingly, Glut-1 expression on the endothelium in intratumoral vessels was markedly decreased as compared with that in normal brain vessels (Fig. 2, compare D with C and note the Glut-1-negative vessel indicated by the arrow in D).

Staining for α-smooth muscle actin revealed a high grade of pericyte coverage of tumor vessels (Fig. 2,J), indicating the presence of a mature phenotype. Despite the lack of sprouting angiogenesis, both endothelial cells and pericytes responded to VEGF by proliferation, because these cells frequently stained positive for (murine) nuclear antigen Ki-67 (MIB-1; Fig. 2,L). This was clearly a VEGF effect because vascular cells in Mel57 parental lesions were quiescent (Fig. 2 K). Despite all of the vascular changes observed, we did not see induction of angiogenesis in terms of sprouting and branching of new capillaries in Mel57-VEGF165 lesions.

Average diameters of Mel57-VEGF165 lesions were increased ∼2-fold as compared with parental Mel57 lesions (Fig. 3,A), whereas the proliferation index of the Mel57-VEGF165 lesions was >4-fold higher than that of control tumors (Fig. 3,B). Proliferation was boosted by factors from the tumor environment because there was no difference in growth rate between the Mel57 transfectants and the parental line in vitro (results not shown). The high proliferation rate, in combination with the lack of sprouting angiogenesis in the Mel57-VEGF165 lesions, led to evident hypoxia as shown by up-regulation of Glut-1 in tumor cells themselves (Fig. 2,D and Ref. 12) and subsequently to necrosis, even in small lesions (Fig. 1 B). In immunohistochemistry, we detected no differences in expression of both angiopoietins I and II between tumors from Mel57 and Mel57-VEGF165 cells; both angiopoietins were expressed (not shown).

Tumors are considered to start as avascular masses that become vascularized by sprouting angiogenesis, a process that is induced by tumor-derived factors such as VEGF. However, several groups have reported that tumors can also use the preexistent vasculature in the host tissue. Holash et al.(3) described initial growth of a brain tumor by vessel co-option, which was followed by vessel regression and induction of classical angiogenesis via up-regulation of VEGF and angiopoietin-2 (3, 4). Li et al.(5) described a s.c. model of early tumor outgrowth where vasodilatation was followed by sprouting angiogenesis in lesions as small as a few thousand cells. Here we show in a brain microenvironment that tumors consisting of several hundreds of thousands of cells can grow without induction of sprouting angiogenesis, even in the presence of high levels of VEGF. Obviously, the capillary network of the brain parenchyma is one of the densest in the mammalian body. Thus, in such an environment tumors can thrive in an angiogenesis-independent manner and produce lesions that in the murine brain can reach diameters of up to 3 mm (∼14 mm3), far beyond Folkman’s angiogenic switch threshold of 2 mm3(13).

In vitro, Mel57 produces minimal amounts of angiogenic factors such as VEGF, basic fibroblast growth factor, interleukin-8, and platelet-derived growth factor (9). The absence of vascular changes in Mel57 and Mel57-EGFP lesions demonstrates that VEGF was not up-regulated in vivo as well. In a previous paper, we reported low expression levels of VEGF in melanoma lines M14 and 530 as well (9). Brain metastases derived from these cell lines behaved comparably with those derived from Mel57, showing that the phenomenon of vascular co-option is not restricted to one cell line.

On the basis of several criteria, we conclude that there were neither morphological nor functional differences between intratumoral and distant vessels in Mel57 or Mel57-EGFP lesions. The BBB was intact, as demonstrated by the absence of extravasated immunoglobulins and by the presence of Glut-1 and ZO-1. From these results, we conclude that VEGF production was not essential for the formation of brain metastases. Clearly, the capacity to metastasize to the brain in the absence of VEGF may be tumor type dependent, because Yano et al.(14) showed that antisense VEGF cDNA transfection of human carcinoma cell lines decreased the metastatic capacity to the brain of these cells.

In contrast with the prevailing idea that VEGF165 induces sprouting angiogenesis, brain metastases of stable VEGF165 transfectants of Mel57 did not have a classical angiogenic phenotype; despite a high proliferation rate of endothelial cells and pericytes, there was no branching or sprouting of the otherwise extensively dilated capillaries in these tumors. This is in accordance with recent reports on VEGF effects in nontumorous settings (10, 15). Another remarkable observation was that VEGF caused severe deterioration of BBB function, leading to vascular hyperpermeability, which is in line with previous reports (16, 17, 18).

Despite the lack of classical (i.e., sprouting) angiogenesis, VEGF165 expression did lead to tumor progression, which was reflected by increased tumor cell proliferation rates and larger lesion sizes. This resulted in a more solid, expansive growth of the lesions with lack of blood supply in the center of the lesions, often causing central necrosis, even in relatively small lesions.

Because the growth rates of the different tumor cell lines in vitro were equal, the overall increased tumor growth of the Mel57-VEGF165 tumors must have been caused by modulation of the preexistent vascular bed. The dilation of the blood vessels may have led to elevated perfusion, thereby providing the lesions with a better blood supply. This, however, occurred predominantly in the peritumoral zone. The inability to induce sprouting angiogenesis caused a lack of neovasculature within the tumor lesions and subsequent local necrosis. Our data therefore suggest that tumor-derived factors, other than VEGF165, are required for sprouting angiogenesis to occur. Angiopoietins have been reported to play an important role in angiogenesis as well (3, 4, 19). Immunohistochemical staining for angiopoietins I and II demonstrated that these factors were produced by both parental Mel57 and Mel57-VEGF165 cells, implying that the angiopoietin/Tie2 system probably is not of significance in this model. It has also been reported that expression of the larger VEGF isoforms by tumors correlates with poor prognosis (20). We are currently investigating the effects of expression of these isoforms in our model.

Most data on tumor growth, angiogenesis, and angiogenesis inhibition are derived from experimental settings in which tumors are grown in the largely avascular s.c. space and therefore are selected for angiogenic capacity. In our model, we closely mimic the human situation of hematogeneous CNS metastasis formation (8). We show that mere vessel co-option can account for providing tumor blood supply in highly vascularized organs. Constitutive VEGF expression per se does not lead to classical angiogenesis but may promote tumor growth by functional modulation of the co-opted vessels. Thus, the irregularly shaped and dilated vessels that are often found in human CNS malignancies may not always represent newly formed vessels but may represent morphologically and functionally altered preexistent ones.

Our results might have clinical relevance with regard to diagnosis and therapy. Because tumors detected by contrast-enhanced magnetic resonance imaging are detected on the basis of vascular changes (e.g., hyperpermeability), the absence thereof, such as we observed in the Mel57 lesions, will hamper tumor detection. With regard to therapy, our results show that antiangiogenic treatment of brain tumors, such as glioblastoma multiforme, high-grade astrocytomas, and metastases with vascular changes, may provide a benefit in that tumor progression might be slowed down. However, an important unwanted side effect might be that blockade of VEGF leads to a nonpermeable tumor vasculature, potentially with a concomitant resistance to chemotherapeutic treatment. These issues are currently under investigation in our laboratory.

Fig. 1.

Growth patterns of Mel57 (A and C), Mel57-VEGF165 (B and D), M14 (E), and 530 (F) melanoma brain metastases. H&E staining (A and B) and CD31 staining (C and D) of brains containing Mel57 lesions (A and C) and Mel57-VEGF165 lesions (B and D) are shown. The insets in A and B show that tumor cells invade the brain parenchyma along the brain vessels in the space of Virchow-Robin. The inset in D shows a decrease of vessel dilation with distance to a small sized lesion. Note the vessel dilation in Mel57-VEGF165 lesions and the presence of central necrosis in Mel57-VEGF165 lesions (B) but not in Mel57 lesions (A). The cell lines M14 and 530, producing low amounts of VEGF in vitro, also display an infiltrative growth pattern by vascular co-option in brain metastases. Note the Glut-1 BBB marker on co-opted vessels in E. T, tumor; N, necrosis; V, vessel.

Fig. 1.

Growth patterns of Mel57 (A and C), Mel57-VEGF165 (B and D), M14 (E), and 530 (F) melanoma brain metastases. H&E staining (A and B) and CD31 staining (C and D) of brains containing Mel57 lesions (A and C) and Mel57-VEGF165 lesions (B and D) are shown. The insets in A and B show that tumor cells invade the brain parenchyma along the brain vessels in the space of Virchow-Robin. The inset in D shows a decrease of vessel dilation with distance to a small sized lesion. Note the vessel dilation in Mel57-VEGF165 lesions and the presence of central necrosis in Mel57-VEGF165 lesions (B) but not in Mel57 lesions (A). The cell lines M14 and 530, producing low amounts of VEGF in vitro, also display an infiltrative growth pattern by vascular co-option in brain metastases. Note the Glut-1 BBB marker on co-opted vessels in E. T, tumor; N, necrosis; V, vessel.

Close modal
Fig. 2.

Effect of VEGF165 on blood-brain barrier function. Brains containing Mel57 lesions (A, C, and E) or Mel57-VEGF165 lesions (B, D, and F) were stained for mouse IgG (A, B), Glut-1 (C, D), or ZO-1 (E, F) as described in the text. Only vessels in and around Mel57-VEGF165 lesions are leaky, as demonstrated by the presence of extravasated IgG (B). Dilated vessels in Mel57-VEGF165 lesions express the BBB markers ZO-1 (B) and Glut-1 (D), although expression on vessels within the lesions is clearly down-regulated. The arrow in D points at a Glut-1-negative vessel. Glut-1 is expressed in tumor cells adjacent to necrosis, indicating hypoxia (D, *). Activation of vascular endothelium by VEGF165. Immunostaining of Mel57 (G, I, K) and Mel57-VEGF165 (H, J, L) lesions for VEGFR-2/Flk-1 (G, H), a smooth muscle actin (α-SMA) (I, J), and Ki-67 (K, L). VEGFR-2/Flk-1 is up-regulated on vessels in and around the Mel57-VEGF165 lesions, indicating an activated state of the endothelium. These vessels display a mature phenotype as indicated by the high pericyte coverage (J) compared with vessels in lesions from the parental cell line (I). Although VEGF-A165 in these tumors induces proliferation of endothelial cells, as demonstrated by Ki-67 staining (L), this is not accompanied by vascular sprouting. V, vessel; T, tumor.

Fig. 2.

Effect of VEGF165 on blood-brain barrier function. Brains containing Mel57 lesions (A, C, and E) or Mel57-VEGF165 lesions (B, D, and F) were stained for mouse IgG (A, B), Glut-1 (C, D), or ZO-1 (E, F) as described in the text. Only vessels in and around Mel57-VEGF165 lesions are leaky, as demonstrated by the presence of extravasated IgG (B). Dilated vessels in Mel57-VEGF165 lesions express the BBB markers ZO-1 (B) and Glut-1 (D), although expression on vessels within the lesions is clearly down-regulated. The arrow in D points at a Glut-1-negative vessel. Glut-1 is expressed in tumor cells adjacent to necrosis, indicating hypoxia (D, *). Activation of vascular endothelium by VEGF165. Immunostaining of Mel57 (G, I, K) and Mel57-VEGF165 (H, J, L) lesions for VEGFR-2/Flk-1 (G, H), a smooth muscle actin (α-SMA) (I, J), and Ki-67 (K, L). VEGFR-2/Flk-1 is up-regulated on vessels in and around the Mel57-VEGF165 lesions, indicating an activated state of the endothelium. These vessels display a mature phenotype as indicated by the high pericyte coverage (J) compared with vessels in lesions from the parental cell line (I). Although VEGF-A165 in these tumors induces proliferation of endothelial cells, as demonstrated by Ki-67 staining (L), this is not accompanied by vascular sprouting. V, vessel; T, tumor.

Close modal
Fig. 3.

Effects of VEGF165 expression on lesion size and proliferation. A, size of Mel 57 and Mel57-VEGF165 lesions, expressed as diameter (mm). Lesions in five mice brains were used in this analysis. Mice were sacrificed 21 ± 1 days after intracarotid injection of tumor cells. Lesions were measured as described in the text. Sizes differed significantly between the groups (P < 0.001, Student’s t test). B, tumor cell proliferation index in vivo as determined by BrdUrd incorporation (see text). Data are expressed as fraction of labeled tumor cell nuclei (P < 0.001, Student’s t test). Bars, SD.

Fig. 3.

Effects of VEGF165 expression on lesion size and proliferation. A, size of Mel 57 and Mel57-VEGF165 lesions, expressed as diameter (mm). Lesions in five mice brains were used in this analysis. Mice were sacrificed 21 ± 1 days after intracarotid injection of tumor cells. Lesions were measured as described in the text. Sizes differed significantly between the groups (P < 0.001, Student’s t test). B, tumor cell proliferation index in vivo as determined by BrdUrd incorporation (see text). Data are expressed as fraction of labeled tumor cell nuclei (P < 0.001, Student’s t test). Bars, SD.

Close modal

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This study was supported by Grants KUN 2000-2302 and KUN 2001-2399 from the Dutch Cancer Society. B. K. is a recipient of the Research Fellowship Grant 920-03-149 from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO, Dutch Organization for Scientific Research).

4

The abbreviations used are: CNS, central nervous system; VEGF-A, vascular endothelial growth factor-A; IRES, internal ribosome entry site; BBB, blood-brain barrier; BrdUrd, bromodeoxyuridine; Glut, glucose transporter; EGFP, enhanced green fluorescent protein.

1
Plate K. H., Risau W. Angiogenesis in malignant gliomas.
Glia
,
15
:
339
-347,  
1995
.
2
Wesseling P., Ruiter D. J., Burger P. C. Angiogenesis in brain tumors: pathobiological and clinical aspects.
J. Neurooncol.
,
32
:
253
-265,  
1997
.
3
Holash J., Maisonpierre P. C., Compton D., Boland P., Alexander C. R., Zagzag D., Yancopoulos G. D., Wiegand S. J. Vessel cooption, regression, and growth in tumors mediated by angiopoietins and VEGF.
Science (Wash. DC)
,
284
:
1994
-1998,  
1999
.
4
Holash J., Wiegand S. J., Yancopoulos G. D. New model of tumor angiogenesis: dynamic balance between vessel regression and growth mediated by angiopoietins and VEGF.
Oncogene
,
18
:
5356
-5362,  
1999
.
5
Li C. Y., Shan S., Huang Q., Braun R. D., Lanzen J., Hu K., Lin P., Dewhirst M. W. Initial stages of tumor cell-induced angiogenesis: evaluation via skin window chambers in rodent models.
J. Natl. Cancer Inst. (Bethesda)
,
92
:
143
-147,  
2000
.
6
Küsters B., Westphal J. R., Smits D., Ruiter D. J., Wesseling P., Keilholz U., de Waal R. M. W. The pattern of metastasis of human melanoma to the central nervous system is not influenced by integrin αvβ3 expression.
Int. J. Cancer
,
92
:
176
-180,  
2001
.
7
Schackert G., Fidler I. J. Development of in vivo models for studies of brain metastasis.
Int. J. Cancer
,
41
:
589
-594,  
1988
.
8
Fidler I. J., Schackert G., Zhang R. D., Radinsky R., Fujimaki T. The biology of melanoma brain metastasis.
Cancer Metastasis Rev.
,
18
:
387
-400,  
1999
.
9
Westphal J. R., Van’t Hullenaar R., Peek R., Willems R. W., Crickard K., Crickard U., Askaa J., Clemmensen I., Ruiter D. J., de Waal R. M. Angiogenic balance in human melanoma: expression of VEGF, bFGF, IL-8, PDGF, and angiostatin in relation to vascular density of xenografts in vivo.
Int. J. Cancer
,
86
:
768
-776,  
2000
.
10
Proescholdt M. A., Heiss J. D., Walbridge S., Muhlhauser J., Capogrossi M. C., Oldfield E. H., Merrill M. J. Vascular endothelial growth factor (VEGF) modulates vascular permeability and inflammation in rat brain.
J. Neuropathol. Exp. Neurol.
,
58
:
613
-627,  
1999
.
11
Bussink J., Kaanders J. H., Rijken P. F., Martindale C. A., van der Kogel A. J. Multiparameter analysis of vasculature, perfusion, and proliferation in human tumour xenografts.
Br. J. Cancer
,
77
:
57
-64,  
1998
.
12
Clavo A. C., Brown R. S., Wahl R. L. Fluorodeoxyglucose uptake in human cancer cell lines is increased by hypoxia.
J. Nucl. Med.
,
36
:
1625
-1632,  
1995
.
13
Folkman J. What is the evidence that tumors are angiogenesis dependent?.
J. Natl. Cancer Inst. (Bethesda)
,
82
:
4
-6,  
1990
.
14
Yano S., Shinohara H., Herbst R. S., Kuniyasu H., Bucana C. D., Ellis L. M., Davis D. W., McConkey D. J., Fidler I. J. Expression of vascular endothelial growth factor is necessary but not sufficient for production and growth of brain metastasis.
Cancer Res.
,
60
:
4959
-4967,  
2000
.
15
Rosenstein J. M., Mani N., Silverman W. F., Krum J. M. Patterns of brain angiogenesis after vascular endothelial growth factor administration in vitro and in vivo.
Proc. Natl. Acad. Sci. USA
,
95
:
7086
-7091,  
1998
.
16
Dvorak H. F., Brown L. F., Detmar M., Dvorak A. M. Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis.
Am. J. Pathol.
,
146
:
1029
-1039,  
1995
.
17
Berkman R. A., Merrill M. J., Reinhold W. C., Monacci W. T., Saxena A., Clark W. C., Robertson J. T., Ali I. U., Oldfield E. H. Expression of the vascular permeability factor/vascular endothelial growth factor gene in central nervous system neoplasms.
J. Clin. Investig.
,
91
:
153
-159,  
1993
.
18
Strugar J. G., Criscuolo G. R., Rothbart D., Harrington W. N. Vascular endothelial growth/permeability factor expression in human glioma specimens: correlation with vasogenic brain edema and tumor-associated cysts.
J. Neurosurg.
,
83
:
682
-689,  
1995
.
19
Jones N., Iljin K., Dumont D. J., Alitalo K. Tie receptors: new modulators of angiogenic and lymphangiogenic responses.
Nat. Rev. Mol. Cell. Biol.
,
2
:
257
-267,  
2001
.
20
Cheng S. Y., Nagane M., Huang H. S., Cavenee W. K. Intracerebral tumor-associated hemorrhage caused by overexpression of the vascular endothelial growth factor isoforms VEGF121 and VEGF165 but not VEGF189.
Proc. Natl. Acad. Sci. USA
,
94
:
12081
-12087,  
1997
.