Two Independent Mechanisms Essential for Tumor Angiogenesis: Inhibition of Human Melanoma Xenograft Growth by Interfering with either the Vascular Endothelial Growth Factor Receptor Pathway or the Tie-2 Pathway

Development of the vasculature involves two successive processes: (a) vasculogenesis, the establishment of a primitive vascular network during embryogenesis from newly differentiated endothelial cells; and (b) angiogenesis, the sprouting of new capillaries from preexisting vessels. In the adult, neovascularization by angiogenesis occurs in normal settings such as ovulation, placental development, and wound healing, as well as in pathological settings such as retinopathy, rheumatoid arthritis, psoriasis, and tumor growth. The establishment and remodelling of the vascular system involves paracrine signaling between stromal cells and endothelial cells. Protein ligands that specifically regulate endothelial cell function, and their cognate endothelium-specific receptor tyrosine kinases have been discovered as main players in angiogenesis, as well as in vasculogenesis (for review, see Refs. 1, 2, 3, 4, 5, 6).

Two families of receptor tyrosine kinases have been identified that are expressed primarily on endothelial cells. The members of the VEGF² receptor family, which includes Flt-1 (VEGF-R1), Flk-1/KDR (VEGF-R2), and Flt-4 (VEGF-R3), are characterized by having seven immunoglobulin-like domains in their extracellular regions. These receptors are recognized by members of the growing family of VEGF-related growth factors in that the ligands of Flt-1 are VEGF and placenta growth factor, whereas Flk-1/KDR binds VEGF, VEGF-C, and VEGF-D, and the ligands of Flt-4 are VEGF-C and VEGF-D (6). The members of the Tie receptor family, which includes Tie-1 and Tie-2 (Tek), are characterized by a complex extracellular region consisting of two immunoglobulin-like domains, three EGF-like domains, and three fibronectin type III domains. Whereas Tie-1 still remains an orphan receptor, recently two ligands of Tie-2—Ang-1 and Ang-2— have been discovered (7, 8).

The pivotal roles of VEGF and of its receptors during vascular development were exemplified in studies on targeted gene inactivation. Even the heterozygous disruption of the VEGF gene resulted in fatal deficiencies in vascularization (9, 10). Mice carrying homozygous disruptions in either the Flt-1 or the Flk-1/KDR gene die in mid-gestation of acute vascular defects. However, the phenotypes are distinct in that Flk-1/KDR knockout mice lack both endothelial cells and a developing hematopoietic system (11), whereas Flt-1 deficient mice have normal hematopoietic progenitors and endothelial cells that fail to assemble into functional vessels (12). Disruption of the Flt-4 gene, the embryonic expression of which becomes restricted to lymphatic vessels in adults, revealed an essential role of Flt-4 for the remodeling and maturation of the primary vascular networks into larger blood vessels during early development of the cardiovascular system (13). Consistent with the lymphatic expression of Flt-4 in adults, overexpression of VEGF-C in the skin of transgenic mice resulted in lymphatic, but not vascular, endothelial proliferation and vessel enlargement (14). Moreover, VEGF-C was reported to induce neovascularization in mouse cornea and chicken embryo chorioallantoic membrane models of angiogenesis (15). The second class of endothelial cell-specific receptor tyrosine kinases has also been found to be critically involved in the formation of vasculature. Mice deficient in Tie-1 die of edema and hemorrhage resulting from poor structural integrity of the endothelial cells (16). The Tie-2 knockout phenotype is characterized by immature vessels that lack branching networks and periendothelial support cells (16, 17). Targeted inactivation of the Tie-2 ligand Ang-1 as well as overexpression of Ang-2, an inhibitory ligand, resulted in phenotypes similar to the Tie-2 knockout (8, 18). Taken together, the results of gene targeting experiments revealed indispensable but probably distinct roles of each of the endothelial receptor/ligand systems in embryonic vasculogenesis and/or angiogenesis.

In pathological settings associated with aberrant neovascularization, elevated expression of angiogenic growth factors and of their receptors has been observed. Most solid tumors express high levels of VEGF, and the VEGF receptors appear predominantly in endothelial cells of vessels surrounding or penetrating the malignant tissue (19). Interference with the VEGF/VEGF receptor system by means of VEGF-neutralizing antibodies (20), retroviral expression of dominant-negative VEGF receptor variants (21), or recombinant VEGF-neutralizing receptor variants (22) resulted in reduced tumor growth and tumor vascularization. However, although many tumors were inhibited by interference with the VEGF/VEGF receptor system, others were unaffected, which suggests the existence of alternative pathways driving tumor angiogenesis (23). The observation of elevated expression of Tie receptors in the endothelium of metastatic melanomas (24), in breast carcinomas (25), and in tumor xenografts grown in the presence of dominant-negative VEGF receptors (23)—as well as elevated expression of Flt-4 receptors in the endothelium of lymphatic vessels surrounding lymphomas and breast carcinomas (26) and elevated expression of VEGF-C in various human tumor samples (27)—suggested these endothelium-specific growth factors and receptors as candidate alternative pathways driving tumor neovascularization.

The present study was undertaken to evaluate the contribution of the endothelium-specific ligand/receptor tyrosine kinase systems to tumor angiogenesis. A human melanoma xenograft model was chosen in which the tumor cells express at least the angiogenic growth factors VEGF, VEGF-C, and Ang-1. Paracrine expression of neutralizing extracellular receptor domains was used to interfere with the respective growth factor/receptor systems in vivo. Using this strategy, we addressed the question of whether the blockade of a single angiogenic pathway in the presence of other angiogenic growth factors may result in the inhibition of tumor angiogenesis and tumor xenograft growth.

Tie-1 and Tie-2 receptors were PCR-cloned from a human placenta cDNA library (Clontech, Heidelberg, Germany) by using the following primer pairs (the nucleotide sequences of the PCR primers used are given in Table 1): (a) Tie-20 and Tie-952c; (b) Tie-926 and Tie-2863c; (c) Tie-2395 and Tie-3439c; (d) Tek-131 and Tek-1051c; (e) Tek-1017 and Tek-2674c; and (f) Tek-2465 and Tek-3537c. Individual PCR fragments were purified by agarose gel electrophoresis, and cDNAs containing complete reading frames for Tie-1 and Tie-2 (Tek) were generated by ligation of PCR fragments via internal restriction sites (HindIII and XhoI for Tie-1; SphI and BssHII for Tie-2). cDNAs encoding human Flt-1 (clone 3–7; Ref. 28) and Flt-4 (clone SHP18; Ref. 29) were generously provided by Dr. M. Shibuya (University of Tokyo, Tokyo, Japan) and Dr. D. Birnbaum (INSERM, Marseille, France), respectively. cDNAs encoding soluble extracellular domains of Tie-1 (sTie-1), Tie-2 (sTie-2), and Flt-4 (sFlt-4) were generated by PCR amplification using primer pairs Tie-20 and Tie-2290c; Tek-131 and Tek-2345c; and Flt-4-Bam and Flt-4-Eco. The fragments were initially cloned into pVL1393 baculovirus transfer vector (PharMingen, San Diego, USA) via EcoRI/BglII sites (sTie-1), EcoRI/NotI sites (sTie-2), or BamHI/EcoRI sites (sFlt-4). Construction of a cDNA encoding the five extracellular immunoglobulin-like domains of Flt-1 (sFlt-1; Ref. 5) has been described previously (30).

Vectors for stable expression of sFlt-1 (5), sFlt-4, sTie-1, and sTie-2 in mammalian cells were generated by the subcloning of the cDNA fragments into pCEP4 vector (Invitrogen, Groningen, the Netherlands), which provides a cytomegalovirus promoter and enhancer. cDNAs encoding sFlt-1 (5), sTie-1, and sTie-2 were restriction-digested using EcoRI, filled in by Klenow polymerase, and subsequently digested by BamHI, BglII, or NotI, respectively. cDNA encoding sFlt-4 was restriction-digested using BamHI, filled in by Klenow polymerase, and subsequently digested using NotI. Fragments were purified by agarose gel electrophoresis and ligated into either HindIII/BamHI- or HindIII/NotI-linearized pCEP4 vector DNA, in which the HindIII overhangs had been filled in by Klenow polymerase. Because of the cloning strategy, the expression vector pCEP4/sFlt-1 encodes amino acids M₁ to N₅₆₂ of human Flt-1; pCEP4/sTie-1 encodes amino acids M₁ to E₇₄₉ of human Tie-1 plus 3 COOH-terminal vector-encoded amino acids (IQT); and pCEP4/sTie-2 encodes amino acids M₁ to L₇₂₉ of human Tie-2 plus 16 COOH-terminal vector-encoded amino acids (RPLEAGKAGSEHDKNH), and pCEP4/sFlt-4 encodes amino acids M₁ to E₇₇₅ of human Flt-4 plus a COOH-terminal 6xHis-tag. All of the constructs were verified by nucleotide sequencing.

Cell line A375v is a subclone of A375 human malignant melanoma (American Type Culture Collection CRL 1619) that spontaneously aquired a high expression level of VEGF. A375v cells were stably transfected with pCEP4/sFlt-1, pCEP4/sFlt-4, pCEP4/sTie-1, pCEP4/sTie-2, and pCEP4 vector control, respectively, by the standard calcium phosphate precipitation method and subsequent selection in hygromycin-containing (1 mg/ml) medium (DMEM-10% FCS). From pCEP4 vector control-transfected cells, a pool-clone was selected, whereas from soluble receptor cDNAs-transfected cells, individual clones were further characterized.

For the analysis of soluble receptor mRNA expression, total RNA was prepared from cell clones using an RNeasy kit (Quiagen, Hilden, Germany). RT-PCR was performed by converting 1 μg of total RNA to cDNA using a first-strand cDNA synthesis kit (Amersham Pharmacia Biotech, Freiburg, Germany) with random hexanucleotide primers followed by PCR amplification of the respective cDNA fragments. The primers (Table 1) Flt-1–5up and Flt-1-II; Flt-4-Bam and Flt-4–366R; Tie-20 and Tie-640cHind; and Tek-131 and Tek-1051c were used for amplification of 757-bp Flt-1, 378-bp Flt-4, 621-bp Tie-1, and 941-bp Tie-2 cDNA fragments, respectively.

Secreted sFlt-1 protein was detected by immunoprecipitation of the protein from 1 ml of conditioned medium of A375v/sFlt-1 clones using 2 μg of monoclonal antibody 4C8–10 raised against human Flt-1.³ Immunoprecipitates were electrophoresed on 10% polyacrylamide gels, electrotransferred to Hybond-P membranes (Amersham, Braunschweig, Germany), probed with monoclonal antihuman Flt-1 antibody 7A6 (30), and detected using an enhanced chemoluminescence detection system (Amersham, Braunschweig, Germany). For the detection of secreted sTie-2 protein, aliquots (40-μl) of conditioned media were electrophoresed on 10% polyacrylamide gels, electrotransferred to Hybond-P membranes, and probed with monoclonal antibody Tek-15 raised against human Tie-2.³

RNA was isolated from cultured cells, separated in formaldehyde-containing agarose gels, transferred to nylon membranes, and hybridized to ³²P-DNA fragments by using standard methods (31). The DNA probes used were human Ang-1 cDNA and human VEGF-C cDNA.

The VEGF protein concentration in conditioned media of A375v cells was determined using a commercial VEGF-ELISA assay following the instructions of the manufacturer (R&D Systems, Wiesbaden, Germany).

Wild-type and transfected A375v cells were seeded at a density of 4 × 10≥ cells/well on 48-well plates in 250 μl of DMEM (without indicator dye)-10% FCS. After incubation for 48, 72, 96, and 120 h, respectively, tetrazolium salt (MTT) was added to a final 1.3 mm, and the cells were incubated for an additional 3 h at 37°C. The formazan salt formed was quantitated on solubilization in 250-μl/well solution of 0.1 n HCl in 2-propanol by measurement of absorbance at 550 nm.

Cells (3 × 10⁴)/well in top-agarose (0.4% agarose in DMEM-4% FCS) were layered onto bottom-agarose (0.5% agarose in DMEM, 5% FCS, and 5% newborn calf serum) in six-well plates. Cells were incubated for 14 days at 37°C, and the colonies formed were counted.

Swiss nu/nu mice were s.c. injected with 1 × 10⁶ tumor cells, and tumor growth was determined by caliper measurement of the largest diameter and its perpendicular. Tumor volume (mm³) was calculated as the product of scaling factor 0.5, tumor diameter (mm), and the square of its perpendicular (mm²).

Four xenograft tumors from each group were randomly selected, and four cryostat sections (5 μm) of each tumor were stained for CD31 using a rat monoclonal antibody to mouse CD31 (PharMingen, San Diego, CA). Specific binding of the primary antibody was visualized with a alkaline phosphatase-labeled antirat IgG antibody (Jackson Immunoresearch Laboratories, West Grove, PA) and p-nitrophenyl phosphate (Sigma, Deisenhofen, Germany) as substrate for alkaline phosphatase. The microvessel density was determined by the method of Weidner et al. (32). After scanning the sections, three fields of the tumor stroma that showed the highest vascularity (hot spots) were chosen. The number of discrete microvessels stained for CD31 was counted at ×200. This was performed by two independent pathologists (M. S. and G. S.). The final count per group represents the average (mean ± SE) of 48 fields.

cDNAs encoding the extracellular ligand-binding domains of human endothelial receptors Flt-1, Flt-4, Tie-1, and Tie-2 were cloned into the cytomegalovirus promoter-driven episomal expression vector pCEP4 generating the plasmids pCEP4/sFlt-1, pCEP4/sFlt-4, pCEP4/sTie-1, and pCEP4/sTie-2, respectively. The constructs covered almost the complete extracellular domains of the receptors with the exception of pCEP4/sFlt-1, which encodes immunoglobulin-like domains I–V of Flt-1; these are sufficient for efficient VEGF binding and neutralization (30). The constructs were stably transfected into human malignant melanoma cell line A375v, a subline of A375, which has spontaneously aquired a high level of VEGF secretion. A375v cells, cultured in medium containing 10% serum, secrete about 1 ng of VEGF per 10⁶ cells in 24 h, whereas VEGF secretion of parental A375 cells remained below the detection limit of the VEGF-ELISA used. In addition to VEGF, A375v cells express the angiogenic ligands VEGF-C and Ang-1 as revealed by Northern blot analysis (data not shown). RT-PCR analysis showed that A375v cells do not express detectable levels of Flt-1, Flt-4, Tie-1, or Tie-2 mRNA by themselves (Fig. 1,A). Transfected clones were initially screened for expression of soluble receptor mRNA using RT-PCR. Clones that were positive for sFlt-1, sFlt-4, sTie-1, and sTie-2 mRNA were detected by amplification of 757-bp, 378-bp, 621-bp, and 941-bp cDNA fragments, respectively, which were absent from nontransfected A375v cDNA (Fig. 1,A). sFlt-1 and sTie-2 were detectable as proteins of M_r 90,000 and 116,000 in conditioned media of A375v/sFlt-1 and A375v/sTie-2 clones, respectively, by immunoprecipitation and Western-blot methods using antibodies raised against the extracellular domains of human Flt-1 and Tie-2 (Fig. 1 B). Transfected A375v clones showing expression of soluble receptors on mRNA and/or protein level were selected for additional characterization.

Determination of in vitro proliferation of soluble receptor cDNA-transfected A375v cells showed a slightly increased proliferation of sTie-1-, sTie-2-, and vector-transfected cells as compared with A375v wild-type cells, as well as sFlt-1-, and sFlt-4-transfected cells (Fig. 2). No inhibition of the proliferation of transfected cells compared with wild-type cells was observed. Anchorage-independent growth was determined as an in vitro indicator for the maintenance of the transformed phenotype of the soluble receptor-expressing cell clones. Colonies formed on incubation of 3 × 10⁴ cells/well for 14 days in soft agar were counted (Fig. 3). Soluble receptor-transfected cell lines showed a slightly increased colony formation rate as compared with A375v wild-type and vector-transfected cells. This observation was most pronounced with sTie-1-expressing cells. Measurement (using an VEGF-ELISA) of VEGF secreted by the transfected cells revealed that transfection of the cells did not affect the amount of VEGF that was secreted by the cells (data not shown). Taken together, these results confirmed that, during the selection procedure of stable transfected soluble receptor-expressing A375v cells, neither in vitro proliferation nor the ability for anchorage-independent growth was compromised.

To determine the effect of soluble endothelial receptor expression on tumor growth and tumor vascularization, two soluble receptor expressing clones were randomly selected from each cDNA transfection, and 1 × 10⁶ cells of each clone were s.c. injected into Swiss nu/nu mice. Tumor growth was monitored by caliper measurement over a period of 28 or 29 days, when it was necessary to kill the animals that had been injected with pCEP4 vector control-transfected cells (A375v/pCEP4). Xenografts raised from A375v/sFlt-1 or A375v/sTie-2 tumor cells grew to significantly (P < 0.05) smaller tumors (Fig. 4,A). sFlt-1-expressing clones reached 18 and 43%, respectively, of the volume of the control tumors, whereas sTie-2-expressing clones reached only 4 and 8%, respectively, of the volume of the controls. A similar degree of inhibition was achieved with xenografts raised from s.c. injection of a mixture of three clones (clones 5, 6, and 12 for A375v/sFlt-1, and clones 11, 15, and 16 for A375/sTie-2) adjusted to 10⁶ cells per aliquot (data not shown). In contrast, tumors derived from A375v/sTie-1 or A375v/sFlt-4 cells showed no significant reduction of tumor volume as compared with the control (Fig. 4 B). No additional increase of tumor growth inhibition was determined when primary tumors were raised from a 50:50 mixture of A375v cells that expressed either sFlt-1 or sTie-2 (data not shown). Variations in the efficacy of tumor-growth inhibition between individual transfected cell clones are most probably due to differences in the amount of secreted soluble receptor proteins and/or to differences in the stability of recombinant protein expression on release from selection pressure in in vivo experiments. Such interclonal variance was also observed by others in similar experiments (22).

The effect of soluble receptor expression on tumor vascularization was evaluated by the determination of microvessel density. Cryostat sections were stained for CD31, and the number of discrete microvessels was counted in areas showing the highest vascularity (hot spots) of the tumor stroma. Vascularization of tumors derived from sFlt-1- or sTie-2-expressing A375v cells was dramatically decreased as compared with the highly vascularized A375v/pCEP4 control (Fig. 5). Quantification of microvessels (Fig. 6) revealed a statistically significant (P < 0.001) 4- to 5-fold reduction of microvessel density in sFlt-1- and sTie-2-expressing tumors, whereas no reduction was observed in sTie-1-expressing tumors. Tumors derived from A375v/sFlt-4 cells showed a slightly increased microvessel density that was not statistically significant.

Recently, several reports (21, 22, 23) were published showing the inhibition of tumor growth by interference with the VEGF receptor pathway by means of soluble or dominant-negative receptor domains. Others reported on the inhibition of tumor growth by interference with the Tie-2 pathway using similar approaches (34, 35). Goldman et al. (22) ectopically expressed the Flt-1 splice-variant—encoding a natural sFlt-1 protein—that consisted of the extracellular immunoglobulin-like loops I–VI (33) in HT1080 human fibrosarcoma cells. In xenograft experiments, they achieved a reduction of tumor size to 3–14% of the size of the controls. Another report (23) indicated that interference with the VEGF receptor pathway by means of retroviral mediated-expression dominant-negative VEGF receptors inhibited the growth of several experimental tumors, whereas others remained unaffected. Lin et al. (34) showed a 75% reduction of tumor growth and vascularization of a rat mammary tumor grown in a rat cutaneous window chamber by the addition of a recombinantly produced extracellular domain of Tie-2 into the cutaneous window chamber. Furthermore, very recently, the inhibition of growth of a murine mammary carcinoma (64% reduction) and of a murine melanoma (47% reduction), as well as the inhibition of experimental metastasis, was demonstrated by adenoviral mediated expression of the extracellular Tie-2 domain (35). However, each of these reports focused on only a single angiogenic pathway. Therefore, in the present study, we carefully chose a tumor system in which several known angiogenic growth factors are present and evaluated the effects of interference with the angiogenic pathways in an identical cellular background. In addition, we considered the Tie-1 and Flt-4 pathways, which had not been addressed by previous tumor-growth inhibition studies.

The contribution of the individual endothelial receptor tyrosine kinase pathways to tumor angiogenesis in vivo was analyzed using a human melanoma xenograft model. Notably, the A375v melanoma cells used express at least the angiogenic growth factors VEGF, VEGF-C, and Ang-1 and do not express detectable mRNA levels of the endothelial receptors by themselves. The A375v cells were transfected to overexpress the extracellular ligand-binding domains of the endothelium-specific receptor tyrosine kinases Flt-1, Flt-4, Tie-1, and Tie-2, respectively. In vitro proliferation and colony formation assays excluded inhibitory effects of stable transfection of cDNAs encoding these soluble receptor domains on mitogenicity and transformed phenotype of the transfected cells. Xenograft tumors derived from tumor cells expressing sFlt-1 or sTie-2 reached only 18–43% and 4–8%, respectively, of the size of the controls. The microvessel density within these tumors was statistically significant reduced by a factor of 4–5. These results indicate that both the VEGF receptor pathway and the Tie-2 pathway are essential for the vascularization and tumor growth of A375v xenografts. Neither of the two pathways is able to functionally compensate blockade of the other pathway. This observation is in agreement with the results of gene-targeting experiments of VEGF receptor Flk-1/KDR and of Tie-2 (11, 16, 17), which revealed indispensable but distinct functions of the receptor pathways. Flk-1/KDR knockout mice lacked endothelial cells indicating that VEGF signaling via Flk-1/KDR is essential for differentiation and survival of endothelial cells. In Tie-2-deficient mice, endothelial cells lack the support of underlying periendothelial mesenchymal cells, a finding that implicates the predominant function of Tie-2 in vessel maturation and maintenance. As a consequence, the endothelial cells collapse into the lumen of vessels that were preformed by tissue fold, which leads to vessel occlusion (36).

Expression of sTie-1 or sFlt-4 in A375v melanoma cells failed to inhibit tumor growth and vascularization of the xenografts. Surprisingly, the A375v/sFlt-4-derived tumors appeared even more vascularized compared with the controls. Expression of sTie-1 and of sFlt-4 in the transfected A375v cells was demonstrated on the level of mRNA. Because of the lack of appropriate tools for the detection of these proteins, we have no direct evidence for proper expression and secretion of sTie-1 and sFlt-4. However, the failure of inhibition of tumor growth by interference with Tie-1 function in the present study is in agreement with a very recently published examination of Tie-1-deficient mice (36). From the observation of an approximately 2-fold increased vessel density in the Tie-1 knockout embryos and of endothelial cells that appeared to be hyperactive and possessing a large number of extensions and filopodia as well as numerous transcellular holes, it was concluded that one main function of Tie-1 is the inhibition of vessel division and intussusceptive microvascular growth.

In summary, we have shown that, in the A375v human melanoma xenograft model, tumor angiogenesis and tumor growth was inhibited by blockade of either the VEGF receptor pathway or the Tie-2 pathway. The observation that the blockade of a single angiogenic growth factor pathway in the presence of other angiogenic growth factors was sufficient to inhibit xenograft tumor growth suggests that using antiangiogenic drugs that interfere with either the VEGF receptor pathway or the Tie-2 pathway would be therapeuticly successful. Furthermore, one would expect additive effects on tumor growth by simultaneous interference with both pathways.

We thank Drs. M. Shibuya and D. Birnbaum for generously providing human Flt-1 and Flt-4 cDNAs. We thank Katja Mohrs, Steffanie Koidl, Karola Henschel, and Sigrun Wellershoff for excellent technical assistance.