HGF/NK4, a Four-Kringle Antagonist of Hepatocyte Growth Factor, Is an Angiogenesis Inhibitor that Suppresses Tumor Growth and Metastasis in Mice¹

HGF,³originally identified and cloned as a potent mitogen for hepatocytes(1, 2), plays both a biological and a physiological role in development and in tissue regeneration(3, 4, 5). In malignant tumors, HGF induces invasive,angiogenic, and metastatic responses through the c-Met/HGF receptor tyrosine kinase (6, 7, 8, 9, 10). In many carcinomas,HGF plays a role as a stroma-derived mediator in tumor-stromal interactions that confer invasion and metastatic potentials in cancer cells (10, 11, 12), whereas the generation of an autocrine HGF-Met loop is involved in the development of several types of tumors including sarcoma (13). Missense mutations in the c-met are causative genetic disorders in patients with sporadic and hereditary papillary renal carcinoma(14). These mutations resulted in the constitutive activation of Met and in the enhanced transformation of tumor cells(15), whereas HGF was also suggested to enhance the mutant Met-mediated transformation (16). Thus, blockade of HGF-Met signaling may be one strategy to inhibit tumor invasion and metastasis.

HGF is composed of the α chain, which contains a NH₂-terminal hairpin and four kringle domains and the catalytically inactive serine protease-like β-chain(2). Recently we prepared an antagonist for HGF by proteolytic digestion of HGF (17), and this HGF-antagonist, called HGF/NK4 (or NK4), is composed of the NH₂-terminal hairpin domain and subsequent four kringle domains of the α-subunit of HGF. NK4 binds to the c-Met/HGF receptor, but does not induce tyrosine phosphorylation of c-Met(17). NK4 competitively inhibits biological events driven by HGF-Met receptor coupling, including the invasion of distinct types of tumor cells (17, 18). On the other hand, NK1 and NK2,previously characterized HGF variants, have partial agonist and antagonist activity and elicit motility and the invasion of tumor cells and endothelial cells (19, 20, 21, 22).

Angiogenesis, the formation of new blood vessels from preexisting blood vessels, is a critical process involved in embryonic development,tissue regeneration, and pathological conditions such as tumorigenesis and diabetic retinopathy (23, 24). Many investigators reported the essential role of angiogenesis during tumor progression(23, 25). Studies led to the thesis that angiogenesis is regulated by a balance between angiogenic and angioinhibitory factors(25, 26). In the activated endothelium, angiogenic growth factors predominate, whereas vascular quiescence is achieved by the dominance of angioinhibitory factors including angiostatin, endostatin,thrombospondin, platelet factor IV, the NH₂-terminal fragment of prolactin, etc.(25). Physiological and pathophysiological roles of angioinhibitory factors, as well as mechanisms by which these polypeptides inhibit angiogenesis, are largely unknown; however,several angiogenesis inhibitors have been shown to inhibit tumor growth and metastasis (27, 28, 29, 30), and there are angiogenesis inhibitors under clinical trials for cancer treatment(29, 30, 31, 32).

We now have evidence that NK4 is an angiogenesis inhibitor. NK4 not only antagonizes HGF-induced angiogenesis but also abrogates the angiogenesis induced by other angiogenic inducers. The antiangiogenic activity of NK4 is likely to be exhibited through a mechanism distinct from its initially characterized potential to act as an HGF-antagonist:the blockade of HGF-c-Met coupling. We report here that NK4 suppresses tumor angiogenesis, growth, and metastasis in mice.

Human recombinant HGF was purified from the conditioned medium of Chinese hamster ovary cells transfected with human HGF cDNA (2, 33). Polyclonal anti-human HGF antibody (1 μg/ml) completely neutralized the biological activities of 1 ng/ml HGF (17, 18). Human recombinant bFGF and VEGF165 were obtained from R & D Systems (Minneapolis, MN). Human NK4 was prepared as described elsewhere (17). The level of endotoxin in the purified NK4 was determined to be 0.43–1.02 ng/mg using a Limulusamebocyte lysate kit from BioWhittaker. Accordingly, the level of endotoxin in endothelial growth media containing 1000 nm NK4 (the maximum dose in our system) was<0.067 ng/ml. This amount of endotoxin alone had no cytotoxic effects on endothelial cells [not shown and as described (34)].

Human adult dermal microvascular endothelial cells and human lung-derived microvascular endothelial cells were purchased from Clonetics (San Diego, CA) and grown in EBM-2 supplemented with 5% FBS and endothelial cell growth supplements (Clonetics). Human pulmonary artery endothelial cells, purchased from KURABO (Osaka, Japan),were cultured in endothelial growth medium (EGM; KURABO). Rat coronary endothelial cells were isolated as described (35). Culture plates for endothelial cells were coated with 0.1% gelatin (Difco,Detroit, MI). Human dermal fibroblasts initially proliferated outward from dermal tissue obtained during plastic surgery. These cells were used within passage number 4–8. MDCK epithelial cells were a generous gift from Dr. R. Montesano (University of Geneva,Geneva, Switzerland). NIH3T3 fibroblasts, LLC cells, and Jyg-MC cells were obtained from RIKEN Cell Bank (Tsukuba, Japan). Rat coronary endothelial cells, human fibroblasts, MDCK, and NIH3T3 cells were cultured in DMEM supplemented with streptomycin, penicillin, and 10% FBS.

Endothelial cells were plated at 5 × 10³ cells/cm² onto gelatinized 24-well tissue culture plates and cultured for 24 h. The medium was replaced with 0.5 ml of EBM-2 containing 5% FBS, and cells were cultured in the absence or presence of NK4, 10 ng/ml HGF, 3 ng/ml bFGF, 10 ng/ml VEGF, or their combinations. After 72 h,cells were dispersed by trypsin and counted by Coulter counter. To measure the proliferation of nonendothelial cells, human dermal fibroblasts, LLC cells, and Jyg-MC cells were plated at 5 × 10³ cells/cm²,whereas NIH3T3 and MDCK cells were plated at 2.5 × 10³ cells/cm², on 24-well plates and cultured for 24 h. The medium was replaced with DMEM supplemented with 5% FBS and 10 ng/ml bFGF, test samples were added,and the cells were then cultured for 72 h.

Migration of endothelial cells was evaluated using a modified Boyden chamber assay, as described (12, 36). The cells were serum-starved in EBM-2 medium for 12 h and plated at 12 × 10⁴ cells/cm² onto the polycarbonate filter with 5-μm pores (Costar, Cambridge, MA)coated with 13.4 μg/ml fibronectin (Orgagnon Teknika Corp.,West Chester, PA). Test samples were added to the medium in the outer cup, and the cells were cultured for 5 h. The number of the cells which migrated to the undersurface of the filter was quantified by counting cells in five randomly selected microscopic fields(×200) in each well. In-vitro invasion of carcinoma cells was measured using a Matrigel invasion chamber (Becton Dickinson,Bedford, MA), as described (12).

Tyrosine phosphorylation of c-Met or KDR was analyzed as described elsewhere (18). Briefly, human adult dermal microvascular endothelial cells were grown on 100-mm plates and serum-starved overnight before treatment for 10 min with various concentrations of NK4 with or without bFGF, VEGF, or HGF (10 ng/ml each). Cell lysates were prepared, and equivalent amounts of protein were incubated overnight with a monoclonal antibody against phosphotyrosine (PY99; Santa Cruz Biotechnology, Santa Cruz, CA) and then incubated for 2 h with protein G-Sepharose. The immunoprecipitates were separated by 6% SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes, and the proteins were probed with anti-c-Met antibodies (C-12; Santa Cruz Biotechnology) or anti-KDR/Flk-1 antibodies (C-1158; Santa Cruz Biotechnology). Immunoreactive bands were visualized by the enhanced chemiluminescence system (Amersham). For detection of phosphorylated ERK1/2 (p44/42 MAPK), total cell lysates were separated by 12% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Blots were developed with phosho-specific antibodies against ERK1/2 (New England Biolabs, Beverly, MA). After being“stripped,” the membranes were reprobed with antibody against ERK1/2 (Santa Cruz Biotechnology) as a loading control.

Male nude mice 6- to 8-weeks-old (BALB/c nu/nu; Japan SLC, Inc.,Hamamatsu, Japan) were s.c. implanted with 1 × 10⁶ LLC cells or 5 × 10⁶ Jyg-MC cells. After 4 days, an osmotic pump(Alzet 2002; Alza Corp., Palo Alto, CA) containing NK4 or BSA or saline alone was surgically implanted near the tumor, and NK4 or BSA solution or saline alone was continuously infused for 10 or 14 days into the s.c. region near the tumor mass. The size of tumors in all groups was measured using a dial caliper, and the volume of tumors was determined using the formula width² × length × 0.52. To analyze external lung metastases, mice were infused with NK4 solution or saline alone for 14 days, as described above, then the mice were killed on day 28 after tumor implantation.

Tumor tissues were fixed in Carnoy’s fixative for 4 h or overnight in 70% ethanol at 4°C and embedded in paraffin according to standard histological procedures. For blood vessel staining, tissue sections fixed in Carnoy’s fixative were pretreated with 5 μg/ml Proteinase K at 37°C for 15 min and incubated with antibody against von Willebrand factor (Dako, Glostrup, Denmark). The sections were sequentially incubated with biotin-labeled secondary antibodies and avidin-biotin peroxidase complex, as described (37). The number of blood vessels was counted under a light microscope at ×200 magnification using at least 20 randomly selected fields per section. For detection of proliferating cells or apoptotic cells, tissue sections fixed in 70% ethanol were analyzed as described elsewhere(18). Statistical analyses were performed with unpaired Student’s t test (two-tailed). P < 0.05 was considered to be statistically significant.

Antiangiogenic activity of NK4 on CAM was assayed, as described(29, 38). Briefly, fertilized white Leghorn chicken eggs were incubated at 37°C for 5 days, and a methylcellulose disk containing test material was placed within a sterilized silicon ring on CAM (38). The eggs were incubated at 37°C for 2 days. The white fat emulsion (Intralipos; WelFide Co., Osaka, Japan) was injected into the chorioallantois, and the vascular networks in the CAMs were blindly scored by two independent investigators for the presence or absence of an avascular zone (>3 mm in diameter)surrounding the implant. Statistical analyses were performed with the Fisher’s exact probability test. P < 0.05 were considered to be statistically significant.

The antiangiogenic activity of NK4 was also assayed in the rabbit cornea, as described (39, 40, 41). Slow-releasing pellets were prepared by incorporating HGF, bFGF and/or NK4 into a 20-μl casting solution of an ethylene-vinyl acetate copolymer (EV40) in 5% methylene chloride. A microsurgical pocket (2 × 4 mm) was produced in the lower half of each eye of male albino Japanese rabbits such that a peripheral pocket ended at 2 mm from the limbus. A single pellet was deposited in the bottom of the pocket. Eyes were examined in a blind manner by slit-lamp microscopy every other day. An angiogenic response was scored positive when blood vessels from the limbal plexus occurred and capillaries progressed to reach the implanted pellet within 6–8 days, as described (40). Statistical analysis was performed with the χ² test. P< 0.05 was considered to be statistically significant.

Statistical analyses were performed with unpaired Student’s t test (two-tailed) unless mentioned otherwise in the text. Differences were considered to be statistically significant at P < 0.05.

We demonstrated previously that NK4 inhibits HGF-mediated growth and invasion of human carcinoma cells (18). In the present study, we attempted to elucidate the antagonistic effect of NK4 on metastasis in murine tumor models. For this purpose, we selected two metastatic murine tumors, LLC and Jyg-MC mammary carcinoma,because these cell lines express the c-Met/HGF receptor (not shown). In vitro, HGF, bFGF, and NK4 had no effect on proliferation and survival of these tumor cells (Fig. 1,A, and not shown). On the other hand, both HGF and bFGF stimulated invasion of tumor cells, whereas NK4 specifically antagonized the invasion of these tumor cells induced by HGF but not by bFGF (Fig. 1 B). When LLC and Jyg-MC cells are inoculated s.c. into athymic mice, both tumors formed nodules (20–40 mm³ in volume) 4 days after implantation, and metastatic nodules in lung surface become visible 3–4 weeks after implantation.

To test the effect of NK4 on lung metastasis, NK4 or vehicle (saline alone or with BSA as a protein control) was infused continuously for 2 weeks, using an osmotic minipump, from the 4th day after tumor implantation. The mice were autopsied and the number of metastases in the lung surface was counted on the 28th day after the implantation. Despite the failure of NK4 to suppress tumor cell growth in vitro, continuous infusion of NK4 dose-dependently inhibited the growth of LLC primary tumors. The volume of NK4-treated tumors was inhibited to 62.5% at 10 μg/day (P < 0.05) and 30.1% at 25 μg/day (P < 0.001)on the 28th day as compared with findings in control tumors(n = 4/group; Fig. 2,A). Lung metastasis was strongly suppressed by NK4-treatment in a dose-dependent manner; the number of lung metastases in control mice (12.9 ± 1.35) decreased to 54.8%(P < 0.05) and 9.46%(P < 0.0001) in NK4-treated mice at 10μg/day and 25 μg/day, respectively (Fig. 2,B). Similarly,NK4 (25 μg/day) also suppressed primary tumor growth and lung metastasis of Jyg-MC to 53.6% (P < 0.05)and 35.3% (P < 0.05), respectively(n = 4/group; Fig. 2). Infusion of BSA in saline at 25 μg/day for 2 weeks did not significantly change the tumor volume and number of lung metastases of both tumors (not shown). Although NK4 inhibited tumor metastasis, the result raised the question as to how NK4 suppresses growth of these primary tumors, inasmuch as HGF and NK4 have no direct effect on the proliferation of tumor cells.

To address this question, LLC primary tumors were excised on the 14th day after implantation (10 days after NK4 or saline administration),and the proliferation index of tumor cells in situ was quantified by immunohistochemistry of PCNA. Although the volume of primary tumors in NK4-treated mice (NK4 reached 10.4 ± 1.41 ng/ml in the blood in NK4-treated mice) decreased to 43%of the control (n = 6; P < 0.01), the proliferation index was not significantly changed by NK4-treatment; this was consistent with in vitro results(Fig. 3). Thus, we speculated that NK4 suppresses tumor growth by inhibiting the host compartment, such as angiogenesis, rather than the tumor-cell compartment. To examine this possibility, we measured the microvessel density in tumor tissues by von Willebrand factor immunostaining. The number of von Willebrand factor-positive vessels in NK4-treated tumors decreased to 55.8% of that of control tumors (P < 0.001), and some tumor vessels in NK4-treated mice were disrupted and had a short diameter (Fig. 3). Furthermore, TUNEL assay showed that NK4-treatment led to a 2.25-fold increase in the number of apoptotic cells (P < 0.001; Fig. 3). These findings are consistent with previous studies which showed that an angiogenesis inhibitor suppresses tumor growth by increasing apoptosis of tumor cells (37, 42). Thus NK4 seems to suppress primary tumor growth mainly through inhibition of tumor angiogenesis.

We next investigated the effect of NK4 on proliferation of human dermal microvascular endothelial cells. During a 3-day culture, HGF stimulated cell growth to a 2-fold-higher level over controls (without growth factor), and NK4 dose-dependently inhibited endothelial cell proliferation. Such inhibitory effects of NK4 were likely attributable to HGF-antagonist activity (43). However, NK4 also significantly abrogated the stimulatory effects of both bFGF and VEGF on endothelial cell proliferation, and the increase in cell number was blocked by 1000 nm NK4. Likewise, in the absence of growth factors, the number of cells increased to a 1.5-fold-higher level,presumably by 5% FBS; however, the basal increase in cell number was inhibited by 1000 nm NK4. Heat-treated NK4 (100°C, 10 min) lost inhibitory activity in regard to endothelial growth.

To determine whether NK4 has similar inhibitory effects on other endothelial and nonendothelial cells, we examined the effects of NK4 on the proliferation of several types of cells (Fig. 4,B). Three distinct types of endothelial cells (human lung-derived microvascular endothelial cells, human pulmonary artery endothelial cells, and rat coronary endothelial cells) were cultured in the presence of bFGF with or without NK4. Dose-dependent growth inhibition by NK4 was seen in these three types of endothelial cells,whereas effectiveness differed somewhat for each cell type; the most potent inhibitory effect was seen with human microvascular endothelial cells. A similar inhibitory effect of NK4 was seen when these endothelial cells were cultured in the presence of VEGF (not shown). On the other hand, the proliferation of nonendothelial cells (human dermal fibroblasts, NIH3T3 mouse fibroblasts, MDCK canine renal epithelial cells), which were stimulated by bFGF, was not suppressed by NK4 at all(Fig. 4 B). Likewise, NK4 had no inhibitory effect on proliferation of these nonendothelial cells stimulated by 10% FBS (not shown).

Because migration of endothelial cells, as well as cell proliferation,is an essential component responsible for angiogenesis, we next examined the effects of NK4 on the cell migration of human dermal microvascular endothelial cells (Fig. 4 C). Endothelial cells were seeded on a filter membrane coated with fibronectin, and the number of cells that migrated through the membrane was measured. Migration of endothelial cells was stimulated by bFGF, VEGF, and HGF,whereas migration of the endothelial cells driven by these growth factors was strongly inhibited by NK4, mostly to the basal level at 300 nm NK4. Taken together, NK4 has an inhibitory effect on endothelial cell growth and migration stimulated by bFGF and VEGF as well as by HGF, and the antiproliferative activity is likely to be specific to endothelial cells.

Although NK4 inhibited endothelial growth and migration stimulated by bFGF and VEGF, its antagonizing activity for HGF might be involved in the endothelial-inhibitory action of NK4. To address this issue, we tested the effect of neutralizing antibody on the growth and migration of endothelial cells (Fig. 4 D). In contrast to NK4, anti-HGF antibody did not inhibit bFGF- and VEGF-induced endothelial growth. On the other hand, the stimulatory effect of HGF on endothelial cell growth was almost completely inhibited by anti-HGF antibody, equally to NK4. Similarly, endothelial cell migration mediated by bFGF and VEGF was not affected by anti-HGF antibody (not shown).

To investigate the possibility that the binding of NK4 to the Met receptor may modify ligand-receptor interaction and subsequent activation of signaling events, or that NK4 itself may interfere with ligand-dependent activation of the receptor for angiogenic growth factors, we examined the tyrosine phosphorylation state of c-Met in endothelial cells. HGF, but not bFGF or VEGF, specifically activated c-Met in endothelial cells by increasing the phosphorylation on tyrosine residues, as shown by immunoprecipitation and Western blotting (Fig. 5 A). NK4 inhibited HGF-induced phosphorylation of the Met receptor. On the other hand, NK4 alone, in concentrations up to 1000 nm (67 μg/ml), did not induce tyrosine phosphorylation of c-Met, and NK4 (300 nm) also failed to activate c-Met in the presence of bFGF or VEGF.

Next, to assess whether NK4 modifies the activation of the VEGF receptor, we measured the phosphorylation state of KDR, the activation of which is largely responsible for VEGF-mediated endothelial growth and migration. VEGF, but not HGF or NK4, stimulated tyrosine phosphorylation of KDR (Fig. 5,B). In contrast with the blockade of HGF-Met coupling by NK4, the VEGF-induced phosphorylation of KDR was not inhibited by NK4. Because activation of ERK1/2 (p44/42 MAPK) is closely involved in the process of angiogenesis induced by various angiogenic stimuli, we next analyzed activation of ERK1/2 (Fig. 5 C). HGF, bFGF, and VEGF induced prominent activation of ERK1/2, whereas NK4 did not inhibit the bFGF- or VEGF-induced activation of ERK1/2 but did inhibit HGF-induced ERK1/2 activation. Therefore, NK4 does not inhibit early intracellular signaling events driven by bFGF and VEGF, yet NK4 inhibits bFGF- and VEGF-induced angiogenic responses in endothelial cells.

To study the angiostatic activity of NK4 in vivo, NK4 was tested using CAM (Fig. 6,A). In controls, CAMs with avascular zones were never found. However, NK4 inhibited new blood vessel formation in a dose-dependent manner as determined by the formation of avascular zones(Fig. 6 A); at a dose of 60 μg NK4/disk, avascular zones were seen in 61.5% of embryos (P < 0.01). Importantly, the inhibitory activity of NK4 on angiogenesis in chick CAMs diminished when NK4 was tested after heat-treatment. BSA (60μg/disk) had no inhibitory effect on angiogenesis in the CAM assay. The results indicate that NK4 has antiangiogenic activity in vivo as well as in vitro.

To further confirm the antiangiogenic activity of NK4 in vivo, the effects of NK4 on bFGF-induced neovascularization were tested in the rabbit cornea assay (Fig. 6 B). Application of 100 ng of bFGF induced neovascularization in 70% of the corneas (7 of 10), whereas simultaneous application of NK4 dose-dependently inhibited bFGF-induced neovascularization. NK4 at 1000 ng significantly inhibited bFGF-induced neovascularization to 22%(P < 0.05). Heat-treated NK4 had no effect on bFGF-induced neovascularization, and NK4 alone had no effect on neovascularization in the cornea. Application of HGF in the cornea assay induced neovascularization and NK4 had inhibitory effects on HGF-induced neovascularization (not shown).

Originally we prepared NK4 as a competitive receptor-antagonist devoid of its own HGF-related activities. NK4 inhibits biological activities of HGF, including mitogenic, motogenic, and morphogenic activities (17, 18). Together with our present finding that NK4 functions as a specific angiogenesis inhibitor, NK4 is bifunctional: it acts as an HGF-antagonist and also as an angiogenesis inhibitor. These distinct functions seem to occur through distinct mechanisms, because the inhibitory effects of NK4 on angiogenesis induced by angiogenic factors other than HGF are unlikely to be mediated via its potential to competitively block coupling between HGF and its receptor, c-Met. In fact, the blockade of coupling of HGF and the c-Met receptor by an anti-HGF antibody did not result in inhibition of endothelial proliferation and migration stimulated by other angiogenic growth factors. Neither bFGF- nor VEGF-induced angiogenic response was mediated by the activation of c-Met in endothelial cells. In addition, bFGF- or VEGF-induced angiogenic signals in the early phase are not blocked by NK4, because NK4 allowed VEGF-induced KDR/VEGF receptor tyrosine phosphorylation and bFGF- or VEGF-induced ERK1/2 activation. Thus, a likely explanation is that NK4 may exert angiostatic signals through putative binding molecules on endothelial cells, although we cannot rule out the possibility that the c-Met receptor may participate in bFGF/VEGF-mediated angiogenesis through unknown mechanisms.

With regard to the structure and function of NK4, it is noteworthy that NK4 has a significant structural similarity with angiostatin, a potent angiogenesis inhibitor (29). Angiostatin is an internal fragment of plasminogen that encompasses the first four kringle domains, and the amino acids sequence homology between four kringles of NK4 and angiostatin reaches 47%. Previous studies showed that individual kringle domains of angiostatin, the fifth kringle domain of plasminogen, and the prothrombin kringle-2 domain have antiangiogenic activity (44, 45, 46). Our preliminary results indicated that antiangiogenic activity of NK4 seems to reside within kringle domains.⁴These results suggest the possibility that a structural motif conserved in some kringle domains may be involved in inhibiting angiogenesis. However, it is equally probable that NK4 inhibits angiogenesis through a mechanism distinct from angiostatin, because NK4 inhibits DNA synthesis and induces cell cycle arrest,⁵whereas angiostatin increases endothelial apoptosis without inhibiting DNA synthesis (34, 47).

Of particular importance in the present study is the in vivoinhibition of tumor metastasis by NK4. The metastatic cascade is composed of multiple steps: (a) induction of angiogenesis; (b) dissociation of tumor cells;(c) invasion through the extracellular matrix;(d) intravasation; (e) transport in the circulation; (f) arrest in a distant capillary bed, and(g) extravasation followed by the establishment of secondary foci (48). In our experimental model, inhibition of lung metastasis might be the result of suppression of early events(i.e., angiogenesis and invasion) in the primary tumors,because NK4-treatment was terminated 10 days before evaluation of lung metastasis, and this no-treatment period might have allowed for the growth of secondary tumors (Fig. 2,A). Considering the bifunctionality of NK4 (HGF antagonist and angiogenesis inhibitor), we can raise the possibility that NK4 may have suppressed lung metastasis by both actions: one, the inhibition of tumor angiogenesis, and the other, the blockade of HGF-mediated dissociation and invasion of tumor cells. There are several reports that aggressive angiogenesis in primary tumors correlates well with the high frequency of distant metastasis in cancer patients (49, 50). Reduced angiogenesis by NK4 in the primary tumors may have decreased the incidence of intravasation of tumor cells, the result being inhibition of lung metastasis. On the other hand, HGF potently stimulates dissociation, migration, and invasion of tumor cells (12, 51, 52, 53), and the induction of extracellular protease networks(such as urokinase-type plasminogen activator and a variety of matrix metalloproteases) by HGF is involved in invasion and subsequent metastasis (18, 53, 54). Indeed, HGF potently stimulated in-vitro invasion of LLC and Jyg-MC cells through Matrigel(which mimics basement membrane), whereas NK4 blocked the HGF-induced invasion of these tumor cells (Fig. 1 B). Thus, the blockade of HGF-Met coupling by NK4 might have suppressed the invasion, motility, and subsequent intravasation of tumor cells,leading to the inhibition of lung metastasis, in our model.

Cutting off the vascular supply for malignant tumors by an angiogenesis inhibitor is expected to be a promising form of cancer therapy. On the other hand, an angiogenesis inhibitor will not modify the invasive and metastatic phenotype of cancer cells, though angiogenesis-dependent tumor invasion is suppressed by halting angiogenesis in some cases (55). A number of studies indicated a critical role for HGF in the invasion and metastasis of a wide variety of malignant tumor cells (6, 7, 8, 9, 10), and inhibition of HGF-Met coupling or signal transduction from the Met has been implicated as a therapeutic strategy to prevent cancer invasion and metastasis (18, 56, 57, 58). On the basis of the bifunctional characteristics of NK4 to target both tumor angiogenesis and HGF-mediated invasion, the possibility that NK4 can function as a therapeutic for subjects with cancer warrants ongoing studies.

We thank Dr. S. Mizuno (Department of Oncology, Osaka University, Osaka, Japan) and Dr. Takao Nakamura (Department of Ophthalmology, Osaka University) for technical assistance with the immunohistochemistry and the rabbit cornea assay, respectively. We are grateful to Dr. K. Zaret (Division of Biology and Medicine, Fox Chase Cancer Center, Fox Chase, PA) and to M. Ohara for helpful comments.