Cooperative Interaction between α- and β-Chains of Hepatocyte Growth Factor on c-Met Receptor Confers Ligand-induced Receptor Tyrosine Phosphorylation and Multiple Biological Responses*

Hepatocyte growth factor (HGF) is a heterodimeric molecule composed of the α-chain containing the N-terminal hairpin domain, four kringle domains, and the serine protease-like β-chain. We prepared HGF/NK4 and HGF/β from the entire HGF after single-cut digestion with elastase. HGF/NK4 contains the N-terminal hairpin and four kringle domains, while HGF/β is composed of the C-terminal 16 amino acids of the α-chain and the entire β-chain, linked by a disulfide bridge. HGF/NK4 competitively inhibited the binding of125I-HGF to the receptor, and affinity cross-linking analysis indicated that HGF/NK4 alone can bind to the c-Met receptor. In contrast, HGF/β alone did not competitively inhibit the binding of125I-HGF to the receptor and did not bind to the c-Met/HGF receptor. Scatchard analysis and affinity cross-linking experiments indicated that HGF/β specifically binds to c-Met in the presence of HGF/NK4 but not HGF/NK2. Neither HGF/NK4 nor HGF/β alone induced mitogenic, motogenic (cell scattering), and morphogenic (induction of branching tubulogenesis) responses; however, HGF/β did induce these biological responses in the presence of HGF/NK4. Consistent with these results, although neither HGF/NK4 alone nor HGF/β alone induced tyrosine phosphorylation of the c-Met/HGF receptor, HGF/β induced tyrosine phosphorylation of the receptor when c-Met/HGF receptor was occupied by HGF/NK4. These results indicate that HGF/β binds to the c-Met/HGF receptor that is occupied by HGF/NK4 and induces receptor tyrosine phosphorylation and the subsequent biological activities of HGF. We propose that there exists a unique cooperative interaction between α- and β-chains, this interaction leading to β-chain-dependent receptor tyrosine phosphorylation and subsequent biological responses.

Coupling between hepatocyte growth factor (HGF) 1 and its receptor, c-met protooncogene product of heterodimeric tyrosine kinase integrates mitogenic, motogenic, and morphogenic activities in a wide variety of cells (1)(2)(3). Most epithelial cells and several types of mesenchymal cells are target cells of HGF (1)(2)(3). Physiologically, HGF is a potent "trophic" factor for regeneration of organs (3), and it possesses potent angiogenic activity (4,5). In addition to roles in epithelial and endothelial tissues, HGF is an important regulator in the maintenance of stromal tissues and cells, including bone formation (6), chondrogenesis (7), and hematopoiesis (8,9). The particular importance of HGF and c-Met/HGF receptor in developmental processes was demonstrated by targeted mutation of HGF or c-Met/HGF receptor gene (10 -13). HGF is essential for the development of the liver and placenta, and it supports migration of myogenic precursor cells. In vitro analysis also showed that HGF supports morphogenic events during development of the kidney, mammary gland, lung, and tooth (1-3, 14 -16). Together with a preferential expression of HGF in mesenchymal (or stromal) tissue (17,18), HGF is considered to be a mesenchymal-derived mediator in epithelial-mesenchymal (or -stromal) interactions during organogenesis and organ regeneration.
Biologically active HGF is a disulfide-linked heterodimer composed of a 69-kDa ␣-chain and a 34-kDa ␤-chain (19 -21). The ␣-chain contains the N-terminal hairpin domain and subsequently four homologous kringle domains, while the ␤-chain contains a serine protease-like domain (22). To address the specific function of each subunit in the HGF molecule, variously mutated variant HGFs were tested for biological activities and receptor binding. A small molecule consisting of the N-terminal hairpin domain, the first kringle domain (K1), and the second kringle domain (K2), designated HGF/NK2 exists as a naturally biosynthesized variant form, and HGF/NK2 can bind the c-Met/HGF receptor (23)(24)(25)(26). Importantly, HGF/NK2 shows motogenic activity, i.e. enhancement of cell motility, but lacks mitogenic activity (23,(25)(26)(27). Thus, HGF/NK2, capable of receptor binding, is an antagonist for the mitogenic activity of HGF (23), yet retains selective agonistic activity in terms of cell motility (25,27). Subsequently, Lokker and Godowski (28) showed that HGF/NK1, composed of the N-terminal hairpin domain and K1, can bind to the c-Met/HGF receptor, while Cioce et al. (29) reported that HGF/NK1 is a naturally occurring variant with partial agonistic or antagonistic activity in a different assay system. On the other hand, the ␤-chain alone cannot bind to the c-Met/HGF receptor and has none of the biological activities of HGF (25,26,30,31). Nevertheless, deletion of the ␤-chain in HGF results in loss of biological activities of HGF, even though the ␣-chain alone can bind to the c-Met/HGF receptor (25,30,31).
Previous studies clarified the specific function of the N-terminal half of the ␣-chain (HGF/NK2), as a receptor-binding motif, as well as a partial agonist in terms of motogenic activity. Thus, biological function of the ␤-chain remained to be specified. We recently obtained the antagonist for HGF, termed "HGF/NK4" (32). HGF/NK4 contains the N-terminal hairpin structure and four kringle domains. We have now obtained evidence which supports the proposal that the ␤-chain can bind to the c-Met/HGF receptor which is specifically occupied with HGF/NK4, and that this cooperative binding induces receptor tyrosine phosphorylation of c-Met, leading to mitogenic, motogenic, and morphogenic responses. The ␣-chain of HGF can bind to the c-Met/HGF receptor, but the optimum activation of c-Met/HGF receptor for the transduction of multiple biological activities of HGF depends on the ␤-chain.

EXPERIMENTAL PROCEDURES
Materials-Human recombinant HGF was purified from the conditioned medium of Chinese hamster ovary cells transfected with human HGF cDNA (22,33). HGF used in the present study was of the 5-amino acid-deleted type (33). The purity of HGF exceed 98%, as determined by SDS-PAGE and protein staining. Anti-phosphotyrosine monoclonal antibody (PY-20) was obtained from Chemicon International Inc. (Temecula, CA) and anti-c-Met polyclonal antibody (C-12) from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Pyroglutamate aminopeptidase was obtained from TAKARA Co. Ltd. (Otsu, Japan). Monoclonal antibody against the ␣-chain of HGF was prepared as described elsewhere (34).
Preparation of HGF/NK4, HGF/␤, and HGF/NK2-Human recombinant HGF was digested with pancreatic elastase (Sigma), and the digested material was applied onto Bondapak C4 reverse-phase HPLC column and adsorbed materials were eluted with a gradient of acetonitrile containing 0.05% (v/v) trifluoroacetic acid (32). HGF/␤ was further purified using a Hi-Trap heparin column (Amersham Pharmacia Biotech, Uppsala). The preparation of HGF/␤ was rechromatographed on C4 reverse-phase HPLC, as described above. The final preparation of HGF/␤ used in this study contained 0.003% (in molar ratio) of intact HGF, as detected by enzyme-linked immunosorbent assay (32). Biochemical analysis indicated that HGF/NK4 is composed of the N-terminal hairpin and subsequent four kringle domains, which corresponds to the ␣-chain deleted with C-terminal 16 amino acids, while HGF/␤ is composed of the C-terminal 16 amino acids of the ␣-chain and the entire ␤-chain, linked by a disulfide bridge between Cys 487 and Cys 604 (Fig.  1A).
HGF/NK2 was produced by transient expression in COS-7 cells, using the expression vector pCDM containing cDNA, which corresponds to the sequence of human HGF/NK2, as described elsewhere (23). HGF/NK2 was purified from the serum-free conditioned medium using a Hi-Trap heparin column (Amersham Pharmacia Biotech). The recombinant HGF/NK2 showed an apparent molecular mass of 28 kDa in SDS-PAGE and following Western immunoblotting, under reducing condition (not shown). SDS-PAGE was done using a 4 -20% or 2-15% gradient gel, and proteins were visualized by silver staining (Wako Pure Chemical, Osaka).
Cell Culture and Measurement of DNA Synthesis-MDCK (clone 3B) renal epithelial cells, a kind gift from Dr. R. Montesano (University of Geneva) were cultured in DMEM containing 10% fetal calf serum. HuCC-T1 human cholangiocarcinoma and A549 human lung adenocarcinoma cells were obtained from the Japanese Cancer Research Resources Bank and cultured in DMEM containing 10% fetal calf serum. For migration assay, MDCK cells were seeded on a 48-well plate at a density of 2.5 ϫ 10 3 cells/well in DMEM containing 10% fetal calf serum, with or without test samples. The cells were cultured for 20 h, then photographed. For three-dimensional culture in collagen gels, MDCK cells were harvested using trypsin-EDTA solution, suspended in ice-cold 0.2% collagen solution (Nitta Gelatin, Tokyo) at a density of 10 4 cells/ml, and 500-l aliquots were added to wells of a 16-mm width (Nunc, Roskilde, Denmark). After the collagen solution gelled, 500 l of culture medium containing HGF, HGF/NK4, and/or HGF/␤ were added. Culture medium was changed every 3 days.
Mitogenic activity of HGF, HGF/NK4, HGF/␤, or their combinations was measured using adult rat hepatocytes in primary culture, as described elsewhere (30). HGF, HGF/NK4, HGF/␤, HGF/NK2, or their combinations were added to cultures of hepatocytes, the culture was run for 20 h, and then pulse-labeled with 0.3 Ci/ml 125 I-deoxyuridine for 6 h. The cells were washed twice with phosphate-buffered saline and once with trichloroacetic acid, then solubilized with 1 M NaOH. Radioactivity of 125 I-deoxyuridine incorporated into nuclei was measured using a ␥-counter.
Radiolabeled Ligand Binding Assay to the c-Met/HGF Receptor-HGF, HGF/NK4, and HGF/␤ were respectively radioiodinated by the chloramine-T method, as described elsewhere (32). The competitive binding assay was performed by incubating 60 pM 125 I-HGF and various concentrations of unlabeled HGF, HGF/NK4, or HGF/␤, simultaneously with 50 g of plasma membranes from rat livers at 12°C for 1 h, in 0.1 ml of binding buffer (Hanks' solution containing 20 mM HEPES and 2 mg/ml bovine serum albumin, pH 7.0). Membranes were centrifuged at 12,000 ϫ g for 10 min at 4°C, resuspended with 10 l of binding buffer and transferred to fresh tubes. 125 I-HGF specifically bound to membranes was measured using a ␥-counter.
Concentration-dependent binding of radiolabeled ligand and Scatchard analysis were performed using HuCC-T1 cells, as described elsewhere (32). Briefly, the cells were cultured on a 24-well plate, and the cultures were washed once with the binding buffer and equilibrated in the same buffer for 30 min at 10°C. Ice-cold binding buffer containing increasing concentrations of 125 I-HGF or 125 I-HGF/␤, with or without 100 times excess molar unlabeled HGF or HGF/␤, was added, and the preparation was incubated at 12°C for 1 h. Cultures were washed three times with ice-cold binding buffer, and radiolabeled ligand bound to cells was measured. All binding experiments were done in quadruplicate.
Detection of Receptor Tyrosine Phosphorylation-Subconfluent A549 cells were cultured in serum-free DMEM supplemented with 0.2% (w/v) bovine serum albumin for 20 h. The cells were treated with HGF, HGF/NK4, and/or HGF/␤, washed with phosphate-buffered saline containing 1 mM Na 3 VO 4 , and the cell lysate was centrifuged at 12,000 ϫ g for 10 min, as described previously (32). The resultant supernatant was preadsorbed with protein A-Sepharose (Amersham Pharmacia Biotech) and centrifuged at 12,000 ϫ g for 10 min. The supernatant was treated with anti-human c-Met antibody and protein A-Sepharose. Immunoprecipitated materials were washed with lysis buffer and solubilized with sample buffer for SDS-PAGE. The immunoprecipitates were separated by SDS-PAGE, electroblotted onto a polyvinylidene difluoride membrane (Bio-Rad, Hercules, CA), and probed with anti-phosphotyrosine monoclonal antibody or anti-c-Met antibody. Proteins reacting with these antibodies were detected using ECL enhanced chemiluminescence (Amersham, Little Chalfont, UK).
Affinity Cross-linking-HuCC-T1 cells cultured in a 90 mm dish were washed twice with ice-cold binding buffer consisting of Hanks' balanced salt solution containing 20 mM HEPES-NaOH (pH 7.0) and 0.2% (w/v) bovine serum albumin and incubated in the binding buffer for 30 min at 10°C. The binding buffer was changed to fresh binding buffer, and the radiolabeled ligand was added. After 1 h of incubation at 10°C, bis(sulfosuccinimidyl) suberate (Pierce) was added at the final concentration of 0.5 M, and the cells were incubated 1 h at 4°C. After cells had been washed twice with phosphate-buffered saline, the cells were lyzed and scraped into buffer composed of 20 mM Tris-HCl (pH 7.4), 10 mM EDTA, 150 mM NaCl, 5 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 0.5% Triton X-100. The cell suspension in a siliconized tube was centrifuged at 15,000 rpm for 10 min, and the supernatant was preadsorbed with 25 l of protein A-Sepharose (Amersham Pharmacia Biotech) for 1 h at 4°C. After centrifugation at 5,000 rpm for 5 min, the supernatant was treated with rabbit anti-c-Met antibody for 2 h at 4°C, then 7.5 l of protein A-Sepharose was added. The preparation was centrifuged, and the precipitated material was washed four times with lysis buffer and solubilized in the sample buffer for SDS-PAGE. The solubilized proteins were subjected to SDS-PAGE, using 2-15% gradient gel, and the gel was subjected to autoradiography.

RESULTS
Receptor Binding Ability of HGF/NK4 and HGF/␤-To determine whether HGF/NK4 and HGF/␤ can bind to the cell surface receptor, competitive binding analysis was carried out using 125 I-HGF (Fig. 1B). Liver plasma membranes were incubated in the presence of 125 I-HGF alone, or 125 I-HGF plus various concentrations of unlabeled HGF, HGF/NK4, or HGF/␤. Addition of unlabeled HGF inhibited the specific binding of 125 I-HGF to the plasma membranes, and 50% inhibition by unlabeled HGF was seen with 60 pM HGF; the dose being approximately equimolar to that of 125 I-HGF.
Addition of HGF/NK4 also inhibited the binding of 125 I-HGF, and the inhibition by 50% was seen with 600 pM HGF/NK4; the concentration was 10-fold higher than that of HGF. In contrast, the addition of HGF/␤ did not inhibit the binding of 125 I-HGF, at least up to 30 nM, the maximal concentration tested here. Therefore, HGF/NK4 seems to bind to the c-Met/HGF receptor with a 10-fold lower affinity than that of the native HGF; however, HGF/␤ does not bind to the receptor.
Mitogenic Activity of HGF/NK4 and HGF/␤-We next examined the mitogenic activity of HGF/NK4 and HGF/␤, using adult rat hepatocytes in primary culture ( Fig. 2A). Addition of 100 pM HGF potently stimulated DNA synthesis to over 10-fold higher levels. In contrast, addition of HGF/␤ alone up to 30 nM had no apparent effect on DNA synthesis. This result is consistent with the finding that HGF/␤ alone cannot bind to the c-Met/HGF receptor. On the other hand, 1.5 nM HGF/NK4 alone had no evident effect on DNA synthesis, even though HGF/NK4 at this concentration seems to occupy most of the c-Met/HGF receptor. However, it is noteworthy that the addition of HGF/␤ in the presence of 1.5 nM HGF/NK4 dose-dependently induced mitogenic responses in hepatocytes, and the maximal stimulatory effect was seen at 30 nM HGF/␤ and was 50 -60% of that of HGF.
Previous studies showed that the two-kringle variant form, HGF/NK2 can bind to the c-Met/HGF receptor and has motogenic (enhancement of cell motility) but not mitogenic activity (23,(25)(26)(27). We then asked whether the combination of HGF/NK2 and HGF/␤ would elicit mitogenic activity, as seen in the combination of HGF/NK4 and HGF/␤ (Fig. 2B). Although HGF/NK4 plus HGF/␤, as well as native HGF, stimulated DNA synthesis in hepatocytes, HGF/NK2 alone or 30 nM HGF/␤ plus HGF/NK2 did not stimulate DNA synthesis.
Motogenic and Morphogenic Activities of HGF/NK4 and HGF/␤-When HGF was added to the monolayer culture of MDCK cells, HGF enhanced their motility and induced scattering of the cells (Fig. 3). Neither HGF/NK4 (1.5 nM) alone nor HGF/␤ (30 nM) alone induced scattering of the MDCK cells.
However, when HGF/␤ was added to the culture in the presence of HGF/NK4, HGF/␤ dose-dependently induced scattering of the cells. The cell scattering seen with 1.5 nM HGF/NK4 plus 30 nM HGF/␤ was comparable to that seen with 100 pM HGF. Consistent with previous reports (23,(25)(26)(27), the addition of HGF/NK2 alone (3.3 nM) induced scattering of MDCK cells (not shown). On the other hand, Hartmann et al. (25) previously showed that the ␣-chain of HGF has weak motogenic activity. A discrepancy regarding the biological activity between HGF/ NK4 and the ␣-chain may be attributable to the structural difference between HGF/NK4 and the ␣-chain; HGF/NK4 lacks C-terminal 16 amino acids of the ␣-chain (Fig. 1A). Importantly, the C-terminal 16-amino acid fragment contains a cysteine residue involved in a disulfide bond between the ␣and ␤-chain of HGF. When the entire ␣-chain alone was expressed in mammalian cells, the ␣-chain formed, at least, a homodimer, presumably through a disulfide bond between C-terminal-free cysteines. Moreover, the recombinant ␣-chain preparation showed weak motogenic activity. 2 Covalently dimerized ␣-chain may induce receptor dimerization and thus allow low level signaling, leading to cell scattering.
We also asked whether HGF/NK4 and HGF/␤ have morphogenic activity (Fig. 4). When MDCK cells were grown in a collagen gel matrix, they form spherical cysts, but when grown in the presence of HGF, branching tubulogenesis occurred. Neither HGF/NK4 alone, nor HGF/␤ induced branching tubular structures in the MDCK cells; however, the addition of HGF/␤ in the presence of 1.5 nM HGF/NK4 did induce branching tubulogenesis, as seen with the native HGF. These results indicate that the combination of HGF/NK4 and HGF/␤ elicits mitogenic, motogenic, and morphogenic activities, all typical for multiple biological activities of HGF, although neither HGF/NK4 alone nor HGF/␤ alone has biological activities.
Induction of c-Met Tyrosine Phosphorylation-Multiple biological activities of HGF depend on tyrosine phosphorylation of c-Met/HGF receptor upon HGF binding. We next analyzed tyrosine phosphorylation of c-Met/HGF receptor in A549 cells (Fig. 5). Tyrosine phosphorylation of c-Met/HGF receptor was not seen in nonstimulated cells, but addition of HGF induced tyrosine phosphorylation of the c-Met/HGF receptor. Neither 1.5 nM HGF/NK4 alone nor 30 nM HGF/␤ alone induced tyrosine phosphorylation; however, a combination of HGF/NK4 plus HGF/␤ dose-dependently induced tyrosine phosphorylation of the receptor. The tyrosine phosphorylation seen with 1.5 nM HGF/NK4 plus 30 nM HGF/␤ was slightly lower than that seen with 100 pM HGF. Taken together, biological activities of HGF/NK4, HGF/␤, and their combination seem to depend on their potential to induce tyrosine phosphorylation of the c-Met/ HGF receptor.
Specific Binding of HGF/␤ to the Receptor-Based on above results, we considered that although HGF/␤ alone does not bind to the c-Met/HGF receptor, HGF/␤ might specifically bind to the receptor in the presence of HGF/NK4. To test this hypothesis, we analyzed concentration-dependent binding of radiolabeled HGF and HGF/␤ to HuCC-T1 cells (Fig. 6). Scatchard analysis of concentration-dependent binding of 125 I-HGF up to 60 pM resulted in a rectilinear plot (Fig. 6A, inset). The K d value and the number of HGF receptors were calculated to be 36 pM and 2728 sites/cell, respectively. Our previous study demonstrated that 125 I-HGF and 125 I-HGF/NK4, respectively, bind to the receptor on rat liver plasma membranes with a K d values of 64.5 pM and 486 pM (32), indicating that HGF/NK4 binds to the receptor with 8-fold lower affinity than that of HGF. The value seems to be fairly consistent with the result of the competitive binding (Fig. 1). On the other hand, Scatchard analysis on 125 I-HGF/␤ binding indicated that the K d value and the number of binding sites were 14455 pM and 18042 sites/cell, respectively (Fig. 6B). The abundant binding sites and the very low affinity suggest that 125 I-HGF/␤ alone seems to bind nonspecifically to sites clearly distinct from the c-Met/HGF receptor. However, in the presence of 1.5 nM HGF/NK4, 125 I-HGF/␤ specifically bound to the cells with a higher affinity than that without HGF/NK4; the K d value and the number of binding sites were 2449 pM and 3394 sites/cell, respectively (Fig. 6C). Although the affinity of HGF/␤ to the binding sites in the presence of HGF/NK4 was still 68-fold lower than that of HGF, the K d value of 2449 pM seems to coincide with a biologically effective concentration for its half-maximal activity (approxi- To test our hypothesis, we carried out affinity cross-linking experiments using HuCC-T1 cells (Fig. 7). The cells were incubated with radiolabeled ligand and a cross-linker was added. Cell lysate was immunoprecipitated with anti-c-Met antibody and the immunoprecipitate was subjected to SDS-PAGE. When 125 I-HGF was cross-linked, radiolabeled cross-linked complexes with molecular masses of 300 -500 kDa were immunoprecipitated with anti-c-Met antibody. Radiolabeled complexes with a molecular mass ϳ300 kDa may be a cross-linked product between HGF (85 kDa) and c-Met/HGF receptor (200 kDa), while complexes with a molecular mass over 400 kDa may be attributable to complexes between HGF and the dimerized c-Met/HGF receptor. Radiolabeled cross-linked products were not formed in the presence of an excess amount of unlabeled HGF, but the formation of cross-linked products was not af- These results indicate that the cross-linked products were specifically formed between c-Met/HGF receptor and 125 I-HGF.
When 125 I-HGF/NK4 was used, radiolabeled complexes with molecular masses of 300 -500 kDa were specifically immunoprecipitated by anti-c-Met antibody. The formation of crosslinked complexes was competitively inhibited by the addition of excess amount of unlabeled HGF/NK4 or HGF, but was not affected by TGF-␣. Thus, HGF/NK4 alone can specifically bind to the c-Met/HGF receptor. In contrast, when 125 I-HGF/␤ was added alone, it was not cross-linked with the c-Met/HGF receptor, indicating that HGF/␤ alone does not bind to c-Met/HGF receptor. However, when 125 I-HGF/␤ was added in the presence of 0.5 nM HGF/NK4, 125 I-HGF/␤ was specifically immunoprecipitated by anti-c-Met antibody, as a complex with a molecular mass of over 300 -500 kDa. The formation of cross-linked complexes was competitively inhibited by the addition of an excess amount of unlabeled HGF, as well as HGF/␤, indicating that HGF/␤ binds to the c-Met/HGF receptor in the presence of HGF/NK4, but it is competitively inhibited HGF. On the other hand, 125 I-HGF/␤ scarcely cross-linked with the c-Met/HGF receptor in the presence of HGF/NK2. Therefore, HGF/␤ forms a complex with c-Met/HGF receptor in the presence of HGF/ NK4, but not HGF/NK2. HGF/␤ Binds and Activates c-Met/HGF Receptor Occupied by HGF/NK4 -Based on these results, we hypothesized that HGF/NK4 binds and occupies the c-Met/HGF receptor, and subsequently, HGF/␤ binds to the c-Met/HGF receptor that is occupied with HGF/NK4. To examine this hypothesis, we analyzed tyrosine phosphorylation of c-Met/HGF receptor, under the following conditions: 1) A549 cells were first incubated with 1.5 nM HGF/NK4 at 4°C for 1 h and washed three times with culture medium to remove unbound free HGF/NK4, and then HGF/␤ was added and the cells were incubated at 37°C for 20 min; and 2) inversely, the cells were first incubated with 30 nM HGF/␤, washed three times, and then HGF/NK4 was added (Fig. 8).
When the cells were first pretreated with HGF/␤ and subsequently with HGF/NK4, the c-Met/HGF receptor was not tyrosine-phosphorylated (Fig. 8). Likewise, neither pretreatment nor treatment with HGF/NK4 alone induced tyrosine phosphorylation of the receptor. In contrast, when the cells were pretreated with HGF/NK4 and subsequently treated with HGF/␤, this sequential treatment induced tyrosine phosphorylation of the c-Met/HGF receptor (Fig. 8). The addition of HGF in pretreatment or subsequent treatment induced tyrosine phosphorylation of the receptor. Taken together, we conclude that HGF/␤ binds and activates the c-Met/HGF receptor occupied by HGF/NK4, rather than that HGF/␤ complexes with HGF/NK4 and subsequently binds and activates the c-Met/HGF receptor. DISCUSSION The unique structural characteristics of HGF as a multipotent growth factor suggested to us that a specific domain is likely responsible for binding to the c-Met/HGF receptor and/or selective biological activity, i.e. mitogenic, motogenic, or morphogenic activity of HGF. On the other hand, biological activities of HGF are mediated through the c-Met/HGF receptor, which integrates complex intracellular signal transduction pathways. The ligand binding to the c-Met/HGF receptor evokes phosphorylation of tyrosine residues located in the kinase domain, the event up-regulates tyrosine kinase activity, and tyrosine phosphorylation in C-terminal tyrosine residues, so-called multiple docking sites, gathers intracellular signaling molecules (35)(36)(37). How each domain or subunit in the HGF molecule is involved in activation of the c-Met/HGF receptor upon HGF binding is virtually unknown; however, a specific domain or subunit is likely to regulate a distinct event, such as receptor binding, receptor oligomerization, and tyrosine phosphorylation. We obtained evidence that supports a cooperative mechanism of ␣and ␤-chains of HGF for receptor binding and the subsequent activation, as follows: 1) HGF/NK4 alone binds to the c-Met/HGF receptor, but does not induce tyrosine-phosphorylation of the receptor or elicit biological activities; 2) HGF/␤ alone does not bind to the receptor but does bind to the receptor when the receptor is occupied by HGF/NK4, induces tyrosine phosphorylation of the receptor, and exerts mitogenic, motogenic and morphogenic activities; 3) K3 and/or K4 in HGF/ NK4 are required for the binding of HGF/␤ to the complex of HGF/NK4 and c-Met/HGF receptor.
Previous studies indicate that HGF/NK2 binds to the c-Met/ HGF receptor, weakly induces tyrosine phosphorylation, and exerts motogenic activity (23,(25)(26)(27)38). HGF/NK2 has no apparent mitogenic activity in several types of cells, including hepatocytes and endothelial cells (23,25,26,38) (this study), while it has an apparent mitogenic activity in some types of cells (39). Silvagno et al. (38) reported that HGF/NK3 also has no mitogenic activity but has motogenic activity in endothelial cells. In contrast, HGF/NK4 does not induce tyrosine phosphorylation and has no apparent biological activities, even though it does bind to the receptor. We consider that K4 in HGF/NK4 does not block the specific binding between HGF/NK2 or HGF/ NK3 and the c-Met/HGF receptor, but may suppress the motogenic activity of HGF/NK2 or HGF/NK3 by suppressing weak tyrosine phosphorylation of the receptor. One possible explanation for the discrepancy in biological activity of HGF/NK2 or HGF/NK3 and HGF/NK4 is that HGF/NK2 or HGF/NK3 may allow for partial or residual activation of the c-Met/HGF receptor and thus partial or low level signaling, leading to cell scattering, but it may not allow for efficient or optimal signaling, leading to cell proliferation and tubule formation. In contrast, the existence of K4 in HGF/NK4 may induce a conformational change in the c-Met/HGF receptor or inhibit receptor dimerization, such that tyrosine autophosphorylation would be mostly impaired. Our results also demonstrated the importance of K3 and/or K4 in the cooperative interaction among HGF/NK4, HGF/␤, and the c-Met/HGF receptor, which occurs on the cell surface c-Met/HGF receptor. HGF/␤ did not induce the mitogenic response nor did it form a cross-linked complex with the c-Met/HGF receptor in combination with HGF/NK2. Therefore, the N-terminal hairpin structure and subsequent two kringle domains (K1 and K2) are a specific motif for the high affinity binding of HGF and HGF/NK4 to c-Met/HGF receptor, while K3 and/or K4 in HGF/NK4 are essential for exposing the specific binding site of HGF/␤ on the c-Met/HGF receptor occupied by HGF/NK4. In this context, it is noteworthy that the ␤-chain alone of HGF-like protein/macrophagestimulating protein, a family molecule of HGF, directly binds to its receptor, Ron (40). It is interesting to assume that HGF/␤ also has a putative binding motif to c-Met/HGF receptor, but in the case of HGF/␤, the binding of HGF/␤ depends on the preoccupation of c-Met/HGF receptor with HGF/NK4. HGF/NK2 elicits motogenic activity, but competitively antagonizes the mitogenic activity of HGF (23,25,26). Together with our earlier finding that HGF with K3 or K4 deleted still sustains significant mitogenic activity (30,31), the importance of the ␤-chain for the optimal activation of c-Met/HGF receptor has been equivocally considered. Of particular importance in It should be emphasized that a combination of pretreatment with HGF/ NK4 and subsequent treatment with HGF/␤ induced tyrosine phosphorylation of the c-Met/HGF receptor, but a combination of pretreatment with HGF/␤ and subsequent treatment with HGF/NK4 did not. NK4, HGF/NK4; ␤, HGF/␤. our present results is that, although HGF/␤ itself does not play a role in a specific recognition processes between HGF and the c-Met/HGF receptor, HGF/␤ is an indispensable domain for the optimum activation and subsequent activation of intracellular signal transduction pathways that lead to mitogenic, motogenic, and morphogenic responses. How does HGF/␤ induce tyrosine phosphorylation of the c-Met/HGF receptor and activate mitogenic, motogenic, and morphogenic responses? Schwall et al. (39) reported that heparin dimerizes HGF/NK1 and confers mitogenic activity, suggesting that heparin-induced dimerization of HGF/NK1 in turn may facilitate dimerization and activation of c-Met/HGF receptor. One possible role of HGF/␤ is likely to be that HGF/␤ may facilitate dimerization of the receptor, through inducing dimerization and/or stabilization of the ligand. Donate et al. (41) implicated the possibility of HGF to form a noncovalently linked homodimer, through putative interactions between K2 and K3 and/or HGF/␤ of each HGF molecule. However, we could not detect HGF/␤-dependent dimerization or oligomerization of c-Met/HGF receptor in our affinity cross-linking experiment (Fig. 7). On the other hand, given that the c-Met/HGF receptor exists in a preassociated form, one possible explanation for the role of HGF/␤ is that the binding of HGF/␤ to c-Met/HGF receptor occupied with HGF/ NK4 can induce the allosteric conformational change required for activation of tyrosine kinase. Which mechanism is involved in activation of the c-Met/HGF receptor should be further analyzed; however, the separation of the two distinct biochemical events, i.e. receptor binding and receptor activation, through utilizing HGF/NK4 and HGF/␤ would provide insights into the initial mechanism involved in activation of c-Met/HGF receptor.
In conclusion, we here show that, although HGF/␤ alone cannot bind to c-Met/HGF receptor, HGF/␤ can bind to the c-Met/HGF receptor occupied with HGF/NK4 and that the binding of HGF/␤ induces tyrosine phosphorylation of the receptor and subsequent mitogenic, motogenic, and morphogenic responses in cells. Clearly, the ␣and ␤-chains of HGF have distinct functions. The N-terminal hairpin-and kringle-containing ␣-chain is a motif which specifies high affinity binding to the c-Met/HGF receptor, while the ␤-chain seems to play a role in optimal activation of the c-Met/HGF receptor, which enables mitogenic, motogenic, and morphogenic actions of HGF.