Effect of His-Gly-Lys motif derived from domain 5 of high molecular weight kininogen on suppression of cancer metastasis both in vitro and in vivo.

We have demonstrated previously that kinin-free high molecular weight kininogen, its domain 5 (D5H, Gly402-Lys502), and peptides derived from D5H inhibited vitronectin-mediated migration and invasion of cancer cells in vitro (Kamiyama, F., Maeda, T., Yamane, T., Li, Y. H., Ogikubo, O., Otsuka, T., and Ohkubo, I. (2001) Biochem. Biophys. Res. Commun. 288, 975-980). In this study, we found that the amino acid sequence His-Gly-Lys (HGK) in D5H is the core motif for inhibition of adhesion and invasion of MDA-MB-231 cells in vitro. P-5m (484GHGKHKNK491, Gly484-Lys491), an octapeptide including the HGK motif derived from D5H, and HGK, a tripeptide, inhibited both cell adhesion and invasion in vitro. However, an octapeptide designated P-5m (K487R), in which Lys487 was changed to Arg, did not inhibit either cell adhesion or invasion, and peptides HGR and HGG also had no inhibitory effect. Recombinant GST-D5H expressed in Escherichia coli had a stronger inhibitory effect on cell adhesion and invasion in vitro than did GST-D5H (K487R) in which Lys487 was changed to Arg. Furthermore, P-5m (Gly484-Lys491) peptide clearly suppressed lung metastasis in mice experimentally induced by using B16-F10 cells, but P-5m (G487R) had no effect. These data strongly indicate that both the HGK motif and lysine residue (Lys487) play essential roles in inhibition of cell adhesion and invasion in vitro and in prevention of metastasis of cancer cells in vivo. We tried to identify the HGK motif binding protein on the surface of cancer cells. A 95-kDa surface biotin-labeled membrane protein was specifically detached from GST-D5H by P-5 (His479-Lys493) peptide but not by P-1 (Gly402-Lys420) peptide originating from the N-terminal region of D5H.


In this study, we found that the amino acid sequence His-Gly-Lys (HGK) in D5 H is the core motif for inhibition of adhesion and invasion of MDA-MB-231 cells in vitro.
P-5m ( 484 GHGKHKNK 491 , Gly 484 -Lys 491 ), an octapeptide including the HGK motif derived from D5 H , and HGK, a tripeptide, inhibited both cell adhesion and invasion in vitro. However, an octapeptide designated P-5m (K487R), in which Lys 487 was changed to Arg, did not inhibit either cell adhesion or invasion, and peptides HGR and HGG also had no inhibitory effect. Recombinant GST-D5 H expressed in Escherichia coli had a stronger inhibitory effect on cell adhesion and invasion in vitro than did GST-D5 H (K487R) in which Lys 487 was changed to Arg. Furthermore, P-5m (Gly 484 -Lys 491 ) peptide clearly suppressed lung metastasis in mice experimentally induced by using B16-F10 cells, but P-5m (G487R) had no effect. These data strongly indicate that both the HGK motif and lysine residue (Lys 487 ) play essential roles in inhibition of cell adhesion and invasion in vitro and in prevention of metastasis of cancer cells in vivo. We tried to identify the HGK motif binding protein on the surface of cancer cells. A 95-kDa surface biotin-labeled membrane protein was specifically detached from GST-D5 H by P-5 (His 479 -Lys 493 ) peptide but not by P-1 (Gly 402 -Lys 420 ) peptide originating from the N-terminal region of D5 H .
High molecular weight kininogen (HK) 1 is well known to be a cofactor in the intrinsic pathway of blood coagulation cascade (1). The protein assists conversion of factor XII to factor XIIa with kallikrein and a negatively charged surface (1). Plasma kininogens are categorized into two types, high molecular weight kininogen (HK) and low molecular weight kininogen. Both types of kininogen have been shown to be cysteine protease inhibitors (2). HK consists of six domains. After cleavage by kallikrein, factor XIIa, or plasmin and release of kinin from HK molecules, HK changes into kinin-free high molecular weight kininogen (HKa). It has been reported that HKa possesses anti-adhesive properties (3,4). HKa consists of two chains, a heavy chain comprising domains 1-3 and a light chain comprising domains 5 and 6 (1,5). Each domain has been reported to possess specific functions. Domain 1 has a low affinity calcium-binding site (6); domains 2 and 3 have specific sites functioning as cysteine protease inhibitors (7), and domain 3 has platelet and endothelial cell binding activity (8). Domain 4 is the bradykinin sequence. Domain 5 (D5 H ) has cell-binding sites (9 -11) and a negatively charged surface-binding site, and domain 6 (D6 H ) has prekallikrein-and factor XI-binding sites (12,13). Thus, HK is a typical multifunctional protein.
Cancer cells have not been used in most previous experiments on HK, and it is not clear what kinds of signal transduction occur inside tumor cells after binding of D5 H or its derived peptides. We reported previously (26,27) that HKa or P-5 (His 479 -Lys 493 ) peptide acts as an inhibitor of cell adhesion to and invasion of vitronectin or Zn-␣ 2 -glycoprotein using established cancer cell lines such as MC3T3-E1 and MG-63. However, the core amino acid sequence responsible for and the mechanism underlying the inhibitory activities have not been elucidated, and it has also not been determined whether these peptides have the same effect in vivo. * This work was supported in part by the Ministry of Education, Science, and Culture of Japan Research Grant 20179823 (to T. T.). 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.
¶ To whom correspondence should be addressed: Dept. of Medical Biochemistry, Shiga University of Medical Science, Seta, Otsu 520-2192 Japan. Tel.: 81-77-548-2161; Fax: 81-77-548-2164; E-mail: ionfox@ belle.shiga-med.ac.jp. 1 The abbreviations used are: HK, high molecular weight kininogen (single chain); BSA, bovine serum albumin; D5 H , domain 5 of high molecular weight kininogen; ECL, enhanced chemiluminescence; HKa, kinin-free high molecular weight kininogen (two-chain); PBS, phosphate-buffered saline; RGD, Arg-Gly-Asp; VN, vitronectin; HUVEC, Here we report that the HGK amino acid sequence in D5 H of HK is the core motif for inhibition of VN-mediated cancer cell adhesion and invasion in vitro and also that an HGK-containing peptide suppressed experimental lung metastasis in vivo. We also report that a cancer cell surface membrane protein with a molecular mass of 95 kDa binds to the D5 H molecule.

EXPERIMENTAL PROCEDURES
Materials-Growth factor-reduced matrix Matrigel and 24-and 96well tissue culture plates were obtained from BD Biosciences. Polyvinylpyrrolidone-free 8-m pore membranes as basement membranes were obtained from Millipore (Bedford, MA). Sulfo-NHS-LC-biotin was obtained from Pierce. Glutathione-Sepharose 4B, a glutathione-Sepharose column, pGEX-4T-1 plasmid, and ECL plus were obtained from Amersham Biosciences. Biotinylated SDS-PAGE standards (high range) and polyvinylidene difluoride membranes were obtained from Bio-Rad.
Cell Culture-MDA-MB-231 is an estrogen and progesterone receptor-negative human breast cancer cell line. This cell line and the mouse malignant melanoma cell line B16-F10 were obtained from American Type Culture Collection (Manassas, VA). These cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with L-glutamine and 10% fetal bovine serum at 37°C in a 5% CO 2 and 95% air-humidified atmosphere. Human umbilical vein endothelial cells (HUVEC) were obtained from Kurabo (Osaka, Japan). The cells were maintained in HuMedia-EG2 medium containing 2% fetal bovine serum, 10 g/ml heparin, and 1 g/ml hydrocortisone as well as 5 ng/ml human basic fibroblast growth factor and 10 ng/ml human vascular endothelial cell growth factor at 37°C in a 5% CO 2 and 95% air-humidified atmosphere.
GST Fusion Proteins-A human D5 H cDNA fragment (from the codon for Gly 402 to the codon for Lys 502 ) was amplified by PCR using forward and reverse primers tagged with EcoRI and SalI sites, respectively. The primers were 5Ј-GATCGAATTCGGAAAAGAACAAGGGC-ATAC-3Ј (forward; F1) and 5Ј-GATCGTCGACTCATTTCCAACCATT-GTGCT-3Ј (reverse; R1), boldface letters and underlined letters indicate restriction sites and the stop codon, respectively. The PCR-amplified fragment was cloned between the EcoRI and SalI sites of pGEX-4T-1. The plasmid was expressed in Escherichia coli (BL21), and the expressed GST fusion protein was designated GST-D5 H . A recombinant protein was purified on a glutathione-Sepharose column as described previously (29).
A point-mutated recombinant protein with a substitution of Lys 487 to Arg in GST-D5 H , designated GST-D5 H (K487R), was prepared in the same manner except that the cDNA fragment was amplified by human D5 H cDNA using standard two-step site-directed mutagenesis strategies. Briefly, as the first step, two PCR fragments were obtained, one pair of forward (F1) and reverse (5Ј-TATTTTTATGCCGTCCGTGGC-CATGACCAT-3Ј) primers and one pair of forward (5Ј-TGGCCACG-GACGGCATAAAAATAAAGGCAA-3Ј) and reverse (R1) primers. Mutated codons are indicated by underlines. As the next step, PCR was performed using those two PCR products as templates with the primers F1 (forward) and R1 (reverse).
Cell Adhesion Assay-Cell adhesion to VN was performed as described previously with minor modifications (30,31). Ninety six-well microtiter plates (Maxisorp; Nunc, Roskille, Denmark) were coated overnight with 100 l per well of VN (10 g/ml) at 4°C. The plates were washed with PBS, and nonspecific binding sites were blocked with 1% BSA in PBS for 1 h at room temperature. One hundred l of a suspension of 2 ϫ 10 4 cells was placed in each well with or without GST-D5 H (2 M), GST-D5 H (K487R) (2 M), and each peptide (400 M). After 2 h of incubation at 37°C in 5% CO 2 , non-adherent cells were removed by rinsing three times with PBS, and adherent cells were fixed for 15 min with 10% formaldehyde and then stained with Giemsa staining solution. Cells in four non-overlapping microscopic fields (ϫ100) inside each well were counted. These experiments were performed independently at least three times.
Cell Invasion Assay-The effects of peptides derived from HK on invasion of cancer cells were assessed using the modified Boyden chamber assay (32)(33)(34)(35). As preparation, the lower surfaces of polyvinyl 8-m pore membranes were each coated with 25 l of 160 nmol/liter VN and dried for 2 h at 37°C. The upper surfaces of the membranes were each coated with 100 l of Matrigel (15 g/100 l), dried overnight at 37°C, and reconstituted in a standard medium to give an even layer over the membrane surface. The lower chamber was filled with DMEM containing 0.1% BSA. One hundred l of cell suspension containing 1 ϫ 10 5 cells in DMEM with 0.1% BSA was added to each of the upper wells with or without various concentrations of GST-D5 H , GST-D5 H (K487R), and peptides. After incubation for 7 h at 37°C in 5% CO 2 , the cells on the upper surface were removed by wiping with a cotton swab. Cells that had traversed into the Matrigel and attached to the lower surface of the membrane were fixed in 10% formaldehyde, stained with Giemsa staining solution, and counted in 5 randomly selected microscopic fields (ϫ100) per filter. These experiments were performed independently at least three times.
Assay for Experimental Metastasis in Mice-Female mice of the C57BL/6J strain, 6 -7 weeks old, were purchased from Nippon CREA (Tokyo, Japan). The mice were maintained in the Laboratory of Animal Experiments under laminar airflow conditions. All animal experiments were carried out according to the institutional guidelines for animal use.
To produce experimental lung metastasis, 2 ϫ 10 5 B16-F10 malignant melanoma cells were injected into the lateral tail vein with or without 500 g of P-5m (Gly 484 -Lys 491 ) and P-5m (K487R). At 2 weeks after the injection, the mice were sacrificed under anesthesia; lungs were removed and fixed in Bouin's solution, and visible metastatic colonies were counted using a dissecting microscope.
Endothelial Cell Tube Formation Assay-The wells of a 96-well tissue culture plate were coated with 90 l of ice-cold 5 mg/ml Matrigel and incubated overnight at 37°C. One hundred l of a suspension of 2 ϫ 10 4 HUVEC in HuMedia-EG2 medium with 10 M ZnCl 2 (5) was layered on top of the gel with or without 2 M of GST, GST-D5 H , and GST-D5 H (K487R) or 400 M of peptides P-5m (Gly 484 -Lys 491 ) and P-5m (K487R). After incubation for 16 h at 37°C in 5% CO 2 , the cells were fixed in 10% formaldehyde, and pictures were taken.
Labeling of Cell Surface Membrane Proteins with Biotin-Four ϫ 10 7 MDA-MB-231 cells were detached in 0.05% trypsin, 1 mM EDTA, washed three times with ice-cold PBS (pH 8.0), and incubated in 2 ml of PBS containing 1 mg/ml sulfo-NHS-LC-biotin at 4°C for 1 h (36). Thereafter, cells were washed three times with ice-cold PBS.
Detection of D5 H -binding Protein-Biotinylated cells were lysed in 1 ml of lysis buffer consisting of 50 mM Tris-HCl buffer (pH 7.4) containing 1% Triton X-100, 100 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 10 mg/ml leupeptin, 10 mg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride (38). The lysate was clarified by centrifugation at 15,000 ϫ g at 4°C. The supernatant obtained was incubated with 30 g of GST-D5 H and 30 l of glutathione-Sepharose 4B for 1 h at 4°C to produce a labeled D5 H -binding protein-GST-D5 H -glutathione-Sepharose 4B complex. After precipitation by centrifugation at 3,000 ϫ g, the complex was washed three times with the above-described lysis buffer. The bound proteins were then detached from the complex by the addition of 15 l of elution buffer containing 0.1 mg/ml peptide (P-1 or P-5), 100 mM Tris-HCl (pH 7.6), 0.1% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride (37). The labeled sample was subjected to 10% SDS-PAGE and blotted onto a polyvinylidene difluoride membrane. The membrane was blocked with 1% low fat milk for 30 min and incubated with streptavidin conjugated to horseradish peroxidase (1:500) in PBS for 30 min to visualize the resolved proteins. Then the membrane was washed three times with PBS and developed using ECL reagent (38).
Statistics-Error bars in all figures represent standard deviations. The significance of differences in numbers of cells was determined using the Student's two-tailed t test compared with the control.

Inhibition of Cell Adhesion by Peptides Derived from D5 H in
Vitro-We have shown previously that peptide P-5 (His 479 -Lys 493 ) derived from D5 H inhibited VN-mediated haptotaxis and haptoinvasion of MG-63 cells (27). On the basis of those findings, we synthesized three additional overlapped peptides derived from P-5, P-5n (Lys 480 -Lys 487 ), P-5m (Gly 484 -Lys 491 ), and P-5c (His 488 -Asn 495 ), and we attempted to identify the core sequence on D5 H for inhibition of cell adhesion and invasion in vitro (Table I). We first carried out VN-mediated MDA-MB-231 cell adhesion assays using GST-D5 H , P-5, P-5n, P-5m, and P-5c. The doses of protein and peptides used were based on doses of GST-D5 H and P-5 that resulted in about 50% inhibition in preliminary studies. Cell adhesion activity was expressed as percent of the control level. GST-D5 H , P-5, P-5n, and P-5m inhibited cell adhesion by 57, 48, 31, and 40% of the control level at concentrations of 2 M GST-D5 H and 400 M peptides (Fig. 1). However, P-5c did not show inhibitory activity. We assumed that the site of inhibitory activity for VN-mediated cell adhesion of MDA-MB-231 cells is in the Gly 484 -His-Gly-Lys 487 sequence, because both P-5n (Lys 480 -Lys 487 ) and P-5m (Gly 484 -Lys 491 ) peptides also showed inhibitory activity.
Inhibition of Cell Invasion by Peptides Derived from D5 H in Vitro-VN-mediated haptoinvasion assays were performed using MDA-MB-231 cells and tested GST-D5 H , P-5, P-5n, P-5m, and P-5c. GST-D5 H and peptides were added at concentrations of 0.25-2.0 and 100 -400 M, respectively. The maximum dose used was based on results of the cell adhesion assay. GST-D5 H , P-5, P-5n, and P-5m all inhibited cell invasion in a dose-dependent manner (Fig. 2, data for GST-D5 H not shown). At concentrations of 2 M GST-D5 H and 400 M peptides, they inhibited cell invasion by 66, 54, 40, and 50% of the control level, respectively. However, P-5c did not inhibit the invasion at any dose as was observed in the cell adhesion assay (Fig. 2, data for GST-D5 H not shown). In the haptoinvasion experiments, we assumed that the GHGK sequence is the core sequence for expression of both inhibitory activities.
We further constructed a new peptide with a scrambled P-5n (His 480 -Lys 487 ) sequence, designated P-5nS (KGHHGKHG). At a concentration of 400 M, this peptide inhibited both cell adhesion and invasion by 11 and 34% of the control level, respectively (data not shown). Based on this finding, we hypothesize that HGK is the minimum active sequence because HGK exists as a common amino acid sequence in both peptides P-5n and P-5nS.
Determination of Active Amino Acid Residues in Peptides Containing HGK and HGR for Both Cell Adhesion and Invasion in Vitro-We synthesized peptide P-5m (K487R) in which Lys 487 was changed to Arg (Gly 484 -Lys 491 ), and we compared the inhibitory effects on MDA-MB-231 cell adhesion and invasion of P-5m (Gly 484 -Lys 491 ) and P-5m (K487R). Their inhibitory activities were expressed as percent of the control level. P-5m (K487R) had no effect on cell adhesion, but P-5m (Gly 484 -Lys 491 ) inhibited cell adhesion by 40% of the control level at 400 M (Fig. 3A). In haptoinvasion assays, P-5m (K487R) had no effect at any dose, but P-5m (Gly 484 -Lys 491 ) inhibited in a dose-dependent manner (Fig. 3B). These results also indicate that HGK is the core sequence for inhibition of cell adhesion and haptoinvasion.
To confirm that the HGK amino acid sequence is the minimum core sequence for inhibition of cell adhesion and haptoinvasion, we synthesized three tripeptides, HGK, HGR, and HGG. The peptide HGK inhibited VN-mediated cell adhesion of MDA-MB-231 cells by 20% of the control level at the concentration of 1.6 mM, but HGR and HGG showed no effect at the same concentration (Fig. 4A). In haptoinvasion assays, the peptide HGK inhibited VN-mediated haptoinvasion in a dosedependent manner by 25% of the control level at the concentration of 1.6 mM, but neither HGR nor HGG showed an inhibitory effect at any dose as was observed in the cell adhesion assay (Fig. 4B).
Effects of D5 H (K487R) on Cell Adhesion and Invasion-To determine whether the HGK sequence also worked as the core site of inhibitory effects on MDA-MB-231 cell adhesion and a The one-letter code for amino acids residues is used. b The positions of the amino acids in HK protein are shown. c This is a previously reported peptide that has inhibitory effects on cell migration and invasion in vitro (27). d This peptide is scrambled P-5n but the HGK sequence remains. e Lys 487 is changed to Arg in this peptide.  (Fig. 5B). Peptide P-5m (K487R) had no effect on cell adhesion or invasion at any dose (Fig. 3, A and B), but GST-D5 H (K487R) had an inhibitory effect on VN-mediated cell invasion. These results indicate that the local conformations of HGK in the peptide and in the domain structure are different.
Inhibition of Experimental Lung Metastasis by Peptide P-5m (Gly 484 -Lys 491 ) in Vivo-On the basis of the possibility that cell invasion in Matrigel is closely related to tumor metastasis, we examined the effect of P-5m (Gly 484 -Lys 491 ) on metastasis in vivo. Because MDA-MB-231 cells have been reported to be cells with low metastatic potential in vivo (39, 40), we used highly metastatic tumor cells, B16-F10 cells, for this in vivo assay. Before this experimental metastasis assay was performed, we confirmed that P-5m (Gly 484 -Lys 491 ) and P-5m (K487R) had the same effects on B16-F10 cells and MDA-MB-231 cells in vitro. P-5m (Gly 484 -Lys 491 ) inhibited cell adhesion and invasion by 35 and 40% of the control level at the concentration of 400 M, respectively, but P-5m (K487R) had no effect on B16-F10 cell adhesion or invasion (data not shown). These results indicate that both peptides had the same effects on different cell lines, B16-F10 and MDA-MB-231. Accordingly, we used B16-F10 cells for experimental lung metastasis. Five hundred g of each peptide and 2 ϫ 10 5 B16-F10 cells were co-injected into the tail vein of each mouse (41,42). After 14 days, the presence of P-5m (Gly 484 -Lys 491 ) had resulted in a reduction of colony formation on the surface of the lung, but the mutant P-5m (K487R) did not inhibit lung metastasis (Table II and Fig.  6).
Effects of HGK Motif on Endothelial Cell Tube Formation-To investigate the possible anti-angiogenic activity of the HGK motif, we performed a tube formation assay (43). Microscopically, HUVEC treated with GST showed a well formed tubular branching network with increased cellularity. HUVEC treated with GST-D5 H or GST-D5 H (K487R) showed fewer and thinner tubules and a smaller network of tubes, regardless of whether or not protein contained an HGK motif. On the other hand, HUVEC treated with P-5m (Gly 484 -Lys 491 ) or P-5m (K487R) showed well formed tubules as those of treated with GST did, regardless of whether or not peptide contained an HGK motif (Fig. 7).
Detection of D5 H -binding Protein-We tried to clarify the process by which HK-derived peptides inhibit VN-mediated cell adhesion and invasion, and we tried to identify receptors that act as intermediaries in the process. To detect very small amounts of protein, the surfaces of MDA-MB-231 cells were labeled with sulfo-NHS-LC-biotin. This technique specifically labels surface membrane proteins, whereas cytoplasmic proteins are left intact (44). The cells were homogenized, and the biotinylated surface proteins were solubilized and then incubated with GST-D5 H and glutathione-Sepharose 4B to produce the complex D5 H -binding protein-GST-D5 H -glutathione-Sepharose 4B. Biotinylated D5 H -binding proteins were eluted with P-5 (His 479 -Lys 493 ) from the complex. The eluted proteins were detected by using ECL systems. Fig. 8 shows a major D5 H -binding protein with an apparent molecular mass of 95 kDa. Even trace amounts of the 95-kDa protein were not detected upon elution with P-1 (Gly 402 -Lys 420 ) peptide originating from the N-terminal region of D5 H .

TABLE II
Effects of P-5m (Gly 484 -Lys 491 ) and P-5m (K487R) on experimental lung metastasis B16-F10 cells (2 ϫ 10 5 cells/mouse) were intravenously co-injected with PBS, P-5m (Gly 484 -Lys 491 ) and P-5m (K487R) into C57BL/6J mice (n ϭ 4). The amounts of peptides per mouse were 500 g. On day 14, mice were sacrificed under anesthesia, and the metastatic colonies in lungs were counted.  6. Inhibition of lung colonization by P-5m (Gly 484 -Lys 491 ) and P-5m (K487R). A, control lung; B, lung from animal treated with 500 g of P-5m (Gly 484 -Lys 491 ); C, lung from animal treated with 500 g of P-5m (K487R). The experimental lung metastasis assay was carried out as described under "Experimental Procedures."  (24,26,27,45). The His-Gly-Lys-rich region (His 475 -Lys 502 ) of D5 H is thought to bind to a negatively charged region of the cell surface, and this region might be responsible for inhibition of cell adhesion. We found in the present study that the HGK sequence is the core sequence for anti-adhesive activity in the His-Gly-Lys-rich region of D5 H . HGR-containing peptides, in which Lys of HGK was changed to Arg, do not have antiadhesive activity (Figs. 3A and 4A). Both Lys and Arg are positively charged amino acids, but there is a difference between the side chains of these amino acids. Our results indicate that the anti-cell adhesion activity depends on the amino acid sequence, not on the positively charged amino acid cluster, in the His-Gly-Lys-rich region. Peptides containing HGK, depending on their amino acid sequence, also have anti-cell invasion activity (Figs. 3B and 4B). With regard to anti-cell adhesive activity at the domain level and not the peptide level of HK, it was found that GST-D5 H (K487R) had no effect on VN-mediated cell adhesion (Fig. 5A). However, in the cell invasion assay, GST-D5 H (K487R) inhibited VN-mediated haptoinvasion by 30% of the control level at the concentration of 2 M (Fig. 5B). This result suggests that other sites for anti-invasive activity such as cationic amino acid cluster sites or an unknown signaling region are present within the D5 H domain.
Chavakis et al. (45) reported that D5 H binds directly to VN, resulting in competition with urokinase plasminogen activator receptor in the case of leukocyte adhesion to VN. We could not confirm the validity of their hypothesis that P-5 or other P-5derived peptides bind to VN and inhibit both cell adhesion and invasion, because we obtained similar results showing that fibronectin-mediated MDA-MB-231 cell adhesion and invasion were inhibited by 30 and 15% of the control level, respectively, by 100 M P-5 (data not shown).
In experimental lung metastasis in vivo, P-5m also inhibited melanoma cell metastasis, but P-5m (K487R) had no effect (Table II). The high metastatic potential of melanoma cells is prevented by the HGK motif in the P-5m peptide as like as the RGD motif. The RGD motif has a well known cell adhesion signal that serves as a recognition signal for interaction with cell surface receptor integrins (46,47). Peptides containing RGD inhibited cell adhesion mediated by cell adhesion proteins such as VN, fibronectin, von Willebrand factor, and fibrinogen, but an RGE-containing peptide in which the third amino acid Asp was changed to Glu had no effect (48). RGD peptides directly bind to integrins and control cell adhesion, invasion, and metastasis (41,49). Tetrapeptide RGDS is the most effective small peptide, and tripeptide RGD causes a nearly complete loss of activity (51). However, we found that the tripeptide HGK inhibited cell adhesion and invasion by about 20 and 25% of the control level at the concentration of 1.6 mM, respectively (Fig. 4). Thus, the tripeptide HGK may be the most effective small peptide. An HGK motif-containing peptide or a material mimicking the HGK structure may be capable of suppressing metastasis by blockade of binding to a receptor present on the tumor cell surface.
D5 H has also been reported to inhibit angiogenesis (5,52). We carried out tube formation assays to determine whether the HGK motif also has anti-angiogenic properties. In the tube formation assays, 2 M of either GST-D5 H (D5 H , Gly 402 -Gly 502 ) or GST-D5 H (K487R) inhibited tubule formation of HUVEC (Fig. 7). However, 400 M of either P-5m (Gly 484 -Lys 491 ) or P-5m (K487R) had no inhibitory effect (Fig. 7). These findings indicate that the HGK motif may not be responsible for inhibition of angiogenesis. Guo et al. (52) reported that D5 H could induce apoptosis of endothelial cells, which might represent a critical part of the anti-angiogenic activity of D5 H . Zhang et al. (5) reported that 50 M of the peptide H5-14 ( 488 HKNKGKK-NGKHNGWKT 503 ) caused near-complete inhibition (93%) of endothelial cell proliferation. The peptide H5-14 does not contain an HGK ( 485 HGK 487 ) sequence. Accordingly, the C-terminal sequence (His 488 -Lys 502 ) of D5 H might be responsible for the inhibition of angiogenesis. Further study is needed to determine the core sequence of D5 H for inhibition of angiogenesis.
Furthermore, we detected a 95-kDa D5 H -binding protein on breast cancer cells, MDA-MB-231 cells, which was detached from D5 H by the HGK-containing peptide P-5 (His 479 -Lys 493 ) (Fig. 8). Integrin ␣ v ␤ 3 is a heterodimer of ␣ v -and ␤ 3 -subunits, and the molecular weights of the ␣ v -and ␤ 3 -subunits have been estimated by reduced SDS-PAGE to be 130 -125 and 105 kDa, respectively (53). It is possible that the 95-kDa protein is a ␤ 3 -subunit of integrin ␣ v ␤ 3 , which is called vitronectin receptor, and that an HGK-containing peptide directly inhibits VN-mediated cell adhesion by blocking integrin ␣ v ␤ 3 activity. However, integrin ␣ v ␤ 3 generally appears as two bands on SDS-PAGE when the integrin is collected from the solubilized cell fraction by an immunoprecipitation method or another method using anti-integrin ␣ v ␤ 3 antibodies, RGD-Sepharose, or VN-Sepharose. This means that the 95-kDa protein is not identical to the integrin ␤ 3 -subunit, because the 95-kDa protein always appeared as one band on reduced SDS-PAGE (Fig. 7). Urokinase plasminogen activator receptor (a 45-kDa protein), p33/ gC1qR, cytokeratin 1 (a 54-kDa protein), and tropomyosin (36and 39-kDa proteins) were identified previously as high affinity specific binding proteins for D5 H or His-Gly-Lys-rich peptides derived from D5 H (17,22,45,50,54). Herwald et al. (14) reported that three proteins of 33, 60, and 100 kDa were produced from an endothelial cell line, EA.hy926, as HK-binding proteins by using an HKH20 (His 479 -His 498 ) peptide-immobilized column. They eluted HKH20 peptide-binding proteins by changing pH and finally eluted a 33-kDa protein (gC1qR) from the HKH20 column at pH 3. The 100-kDa protein was eluted at pH 4 or at pH 5, because the conformation of the 100-kDa protein may be more sensitive to pH than that of gC1qR/33-kDa protein. However, we specifically eluted the 95-kDa pro- tein using P-5 (His 479 -Lys 493 ), and the molecular weights of the proteins were different. Accordingly, these proteins are thought to be independent proteins. It has recently been reported that gC1qR/33-kDa protein interacts with integrin ␤ 1 and regulates endothelial cell adhesion and spreading (54). Our 95-kDa cell surface protein may control adhesion and invasion of breast carcinoma cells via interaction with not only integrin ␤ 1 but also integrin ␤ 3 , because an HGK-containing peptide inhibited both VN-and fibronectin-mediated cell adhesion. Further study is needed to identify the 95-kDa protein and to determine whether the protein interacts or inhibits signal transduction of cell adhesion by using other extracellular matrices such as fibronectin, von Willebrand factor, collagens, or fibrinogen.