Propolypeptide of von Willebrand factor is a novel ligand for very late antigen-4 integrin.

We have previously reported that propolypeptide of von Willebrand factor (pp-vWF) promotes melanoma cell adhesion in a β1 integrin-dependent manner. In this report, we identified the α subunit of the cell adhesion receptor for pp-vWF as α4. Human leukemia cell lines that express α4β1 integrin (very late antigen-4, VLA-4), but not cell lines which lack VLA-4, attached well to pp-vWF substrate and these adhesions were completely inhibited by anti-α4 integrin monoclonal antibody HP2/1. Adhesion of mouse melanoma expressing α4 integrin was also inhibited by anti-mouse α4 mAb PS/2. Furthermore, transfection of human α4 cDNA into α4− Chinese hamster ovary cells resulted in an acquisition of adhesive activity to pp-vWF, indicating that pp-vWF is a ligand for VLA-4 integrin. Using a recombinant fragment of pp-vWF, the cell attachment site was shown to be located within amino acid residues 376-455 of pp-vWF. A series of synthetic peptides covering this region were tested for the ability to promote cell attachment and a 15-residue peptide designated T2-15 (DCQDHSFSIVIETVQ, residues numbered 395-409) promoted VLA-4 dependent cell adhesion. The peptide was also capable of inhibiting cell adhesion to pp-vWF, suggesting that this sequence represents the cell attachment site. By affinity chromatography using peptide T2-15-Sepharose, it was found that α4β1 integrin complex from extracts of surface iodinated B16 cells specifically bound to the peptide. These results strongly suggest that pp-vWF is a novel physiological ligand for VLA-4.

Propolypeptide of von Willebrand factor (pp-vWF), 1 which is also called von Willebrand antigen II (1), is an unusually large propolypeptide (ϳ100 kDa) produced only in endothelial cells and megakaryocytes together with blood coagulation protein von Willebrand factor (2). It is processed from a large precursor of vWF (prepro-vWF) during biosynthesis and stored in the granule of both endothelial cells and platelets independent from mature vWF (3,4). We have been investigating the biological functions of pp-vWF and found that pp-vWF bound to collagen and inhibited collagen-induced platelet aggregation in contrast to the mature vWF (5,6). Furthermore, we have found that pp-vWF serves as a substrate for transglutaminase and is specifically cross-linked to laminin (7), suggesting a possibility that it acts as transient matrix protein upon secretion from platelets and endothelial cells at the site of vascular injury. In a previous paper (8), we reported that pp-vWF promoted the attachment and spreading of melanoma cells. The receptor responsible for this adhesion was the ␤1 class of integrin but the corresponding ␣ subunit could not be identified.
Integrins are heterodimeric transmembrane proteins consisting of ␣ and ␤ subunits and mediate cell adhesion to extracellular matrix proteins as well as cell-cell interactions (9 -12). To date more than 15 ␣ subunits and 8 ␤ subunits have been identified and combination of ␣ and ␤ subunits determines the ligand specificity of individual integrins. Integrin-mediated cell adhesion plays crucial roles in regulating the morphology, proliferation, migration, and differentiation of cells. Ligands for integrins are quite diverse and include many extracellular matrix proteins as well as cell surface molecules. VLA-4 (␣4␤1) is an integrin complex that recognizes both alternatively spliced segment III (CS1) of fibronectin (13), and vascular cell adhesion molecule-1 (VCAM-1) (14). Lymphocyte adhesion to endothelial cells is primarily mediated by interaction of lymphocyte VLA-4 with VCAM-1 expressed on cytokine-activated endothelial cells and this pathway is thought to be central to the lymphocyte recruitment to the site of inflammation (15). Moreover, interactions of VLA-4 and its counter-receptors have also been implicated in a number of physiologic and pathogenic processes including CD3-dependent T cell activation (16), lymphohemopoiesis (17)(18)(19), myogenesis (20), and melanoma cell metastasis (21,22). Therefore, VLA-4 is attracting broad attention from those who are investigating the molecular mechanisms underlining these processes.
In the present paper, we found that pp-vWF is the novel ligand for the VLA-4. Furthermore, we identified the cell attachment site as a 15-residue linear sequence present in the midregion of pp-vWF molecule. A synthetic peptide with this sequence could promote cell attachment in an ␣4␤1 integrindependent manner and directly bound to VLA-4 complex. Grant-in-Aid 06454644 from the Ministry of Education, Science, and Culture of Japan and by grants from the Ito Memorial Foundation, the Nissan Science Foundation, and the Kanagawa Academy of Science and Technology. 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.
Construction of Human Integrin Expression Vectors and Transfection-Wild-type human ␤1 integrin cDNA (24) was subcloned into pBJ-1 vector and transfected into Chinese hamster ovary (CHO) cells together with pFneo DNA containing the neomycin resistance gene by electroporation as described previously (25). Stable clones were selected using 700 g/ml G418 (Life Technologies, Inc.), and cells expressing human ␤1 integrin most abundantly were selected by single-cell sorting using an anti-␤1 mAb. For obtaining CHO cells expressing both human ␣4 and ␤1 integrins, wild-type human ␣4 cDNA (26) was transfected into ␤1 integrin-expressing clone together with CDhygro DNA containing the hygromycin-resistance gene and maintained in the medium containing 400 g/ml hygromycin. Stable CHO cell line clones expressing high levels of human ␣4 were again selected by single-cell sorting using an anti-␣4 mAb. Expression levels of each human integrin subunit were verified by fluorescence-activated cell sorting analysis using FACScan (Becton Dickinson, Mountain View, CA) according to the method described previously (24).
Cloning and Expression of Recombinant 8-kDa Fragment Variants-A cDNA clone encoding the 8-kDa portion of human pp-vWF was synthesized using reverse transcriptase-polymerase chain reaction amplification of human placental mRNA. Briefly, total mRNA was prepared using the guanidinium isothiocyanate/cesium chloride method (27). First strand cDNA was then generated by reverse transcription using a 3Ј-oligonucleotide primer complementary to the C terminus of the 8-kDa fragment (Lys 455 ) with a terminal EcoRI restriction sequence (5Ј-GAATTCTTTCAGGAGGGGGAGCTGGA-3Ј). The reverse transcribed cDNA mixture was subjected to 30 amplification cycles of polymerase chain reaction using a 5Ј primer with an EcoRI restriction sequence (5Ј-GAATTCACTTCAAGAGCTTTGACAACAGATA-3Ј). The polymerase chain reaction products were ligated into the pBluescript and transformed into Escherichia coli JMl09. The cloned cDNA was identified by restriction analysis and sequenced using the dideoxy chain termination method. The insert was then subcloned into the EcoRI site of a modified version of the pGEX-2T expression vector (Pharmacia, Milton Keynes, United Kingdom), which included additional cloning sites in its polylinker, and were used to transform JMl09. Recombinant clone (pGEX-r8k1A) was checked for its insert orientation by restriction mapping. Using the same polymerase chain reaction product as a starting material, two additional clones containing an insert of the truncated version of the 8-kDa fragment (r8k1B, corresponds to Ser 373 -Asp 415 , and r8k2A, corresponds to Gln 409 -Lys 455 ) were also obtained. Primers used in the amplification of r8k1B were antisense primer 5Ј-GAATTCTGC-AGTCAGTCGCGGTCATCAGCACACT-3Ј and sense primer 5Ј-GAAT-TCACTTCAAGAGCTTTGACAACAGATA-3Ј; primers for amplification of r8k2A were antisense primer 5Ј-GAATTCTTTCAGGAGGGGGAGC-TGGA-3Ј and sense primer 5Ј-GAATTCTGCAGAAACAGTGTGCTGA-TGACCGCGA-3Ј.
Glutathione S-transferase (GST)-r8k1A fusion protein was induced and isolated as follows. Briefly, 5 ml of overnight cultures of JM109 transformed with recombinant plasmids were diluted 1:100 with fresh LB medium containing 50 g/ml ampicillin and cultured for about 4 h at 37°C. Isopropyl-␤-D-thiogalactoside was then added to 0.5 mM and the culture continued for an additional 2 h. Cells were then centrifuged, suspended in 1/30 volume of phosphate-buffered saline containing 1 mM phenylmethylsulfonyl fluoride, and sonicated. As the fusion protein was completely recovered in the insoluble fraction, extracts were centrifuged and the pellet solubilized in 20 mM Tris-HCl, pH 8.0, containing 8 M urea and 0.2 mM ␤-mercaptoethanol, separated on a gel filtration column of Sephacryl S-200 equilibrated with the same buffer, and dialyzed against 20 mM Tris-HCl, pH 8.0, containing 1 M urea and 0.2 mM ␤-mercaptoethanol. The GST portion of the fusion protein was cleaved by the addition of 5 units/ml bovine thrombin (Sigma) for 14 h at room temperature. The cleavage mixture was then applied to a gel filtration column of Sephacryl S-200, and the cleaved 8-kDa fragment was further purified by reverse phase high performance liquid chromatography on a RESOURCE RPC column (Pharmacia). Fusion proteins of the truncated version of the 8-kDa fragment (r8k1B and r8k2A) were recovered in the soluble fraction and purified by glutathione-agarose affinity chromatography. Those proteins were pooled and dialyzed against 150 mM NaCl, 10 mM Tris-HCl, pH 7.4, and stored at Ϫ70°C. Protein concentrations were measured using the BCA assay (Pierce).
Cell Adhesion Assay-B16 murine melanoma, human monocytic cell lines U937 and THP-1, human lymphoma cell lines MOLT-3 and Jurkat, and human erythroleukemic cell line K562 were provided from the Japanese Cancer Research Resources Bank (Tokyo, Japan) and cultured in either Eagle's minimum essential medium containing 10% fetal calf serum or RPMI 1640 medium containing 10% fetal calf serum and non-essential amino acids. Adherent cells were grown to near confluency and harvested by incubation with phosphate-buffered saline, pH 7.2, containing 2.5 mM EDTA and 2 mg/ml bovine serum albumin for 30 min at 37°C. Detached cells as well as suspension cells were washed three times and suspended in serum-free Eagle's minimum essential medium or RPMI 1640 prior to the adhesion assay. In the case of assay using human leukemia cells and CHO cells transfected with human integrins, the cells were pretreated with 1:1500 dilution of anti-human ␤1 activating mAb TS2/16 ascites to activate human ␤1 integrin. Cell adhesion assay was performed according to the method described previously (8) with a slight modification. In brief, 6-mm square chips cut from bacteriologic plastic dishes (Falcon 1029) were coated for 16 h at 4°C with 50 l of bovine pp-vWF (5 g/ml), CS1-rat serum albumin (0.1 g/ml), and bovine fibronectin (10 g/ml). Some peptides were coated by mounting peptide solution on the chips and drying up at room temperature. It was confirmed by the protein assay that more than 80% of ligands were absorbed on the surface under these conditions. For VCAM-1, chips were first coated with anti-mouse C chain (2 g/ml, Caltag Laboratories, South San Francisco, CA) and then incubated with recombinant VCAM-1-mouse C chain (2 nmol/chip). After blocking nonspecific protein-binding sites by incubation with 10 mM Tris-HCl, 150 mM NaCl, pH 7.4, containing 1% bovine serum albumin and 100 g/ml mouse IgG at room temperature for 1 h, the chips were placed at the bottom of 48-well tissue culture dishes (Costar 3548) and overlaid with 2-5 ϫ 10 5 cells in 50 l of serum-free medium. After incubation at 37°C for 90 min, the chips were picked up and rinsed in cold phosphate-buffered saline to remove nonadherent cells, and fixed with 1% glutaraldehyde in phosphate-buffered saline. Adherent cells were either photographed or counted using a light microscope with a calibrated grid marked on the ocular lens. An adhesion assay using peptide-coupled Sepharose was conducted as follows. Cell suspension was added to 96-well microtiter plate containing monolayer of peptide-Sepharose beads and incubated at 37°C for 90 min. The beads bearing adherent cells were fixed with 1% glutaraldehyde in phosphatebuffered saline and separated from nonadherent cells by differential centrifugation, and stained with Giemsa prior to observation.
Affinity Chromatography-Confluent B16 melanoma cells grown in a 100-mm dish were detached as described above and iodinated by the lactoperoxidase/glucose oxidase method (29). Cells were then lysed with 1 ml of 100 mM octyl-␤-D-glucopyranoside in 10 mM Tris-HCl, pH 7.4, containing 150 mM NaCl, 2.5 mM MgCl 2 , 1 mM MnCl 2 , 1 mM phenyl-methylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml pepstatin A at 4°C for 15 min. After removal of insoluble material by centrifugation, the extract was mixed with an equal volume of peptide-Sepharose, gently shaken at room temperature for 3 h, and packed into a column. After washing out the unbound materials with 8 column volumes of a washing buffer (50 mM octyl-␤-D-glucopyranoside, 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2.5 mM MgCl 2 , 1 mM MnCl 2 ) followed by 1 column volume of a pre-elution buffer (same as washing buffer without MgCl 2 and MnCl 2 ), the bound materials were eluted with an elution buffer which contains 5 mM EDTA in the pre-elution buffer. Eluted materials were collected at 1 ml/fraction and analyzed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) on 7.5% polyacrylamide gel under reducing conditions as described by Laemmli (30) followed by autoradiography using Bioimaging Analyzer BAS 2000 system (Fuji Film Co., Tokyo, Japan).
Immunoprecipitation-Eluates from T2-15-Sepharose were subjected to immunoprecipitation assay using anti-integrin antibodies. Samples were incubated either with 3 l of antisera or 1 g of IgG. Protein G/Protein A-agarose (Oncogene Science, Inc.) was then added to the mixture and incubated at 4°C for 4 h with shaking. The immunocomplexes were centrifuged and the beads were washed four times with washing buffer as described above. The samples were then boiled in reducing Laemmli sample buffer and analyzed by SDS-PAGE on a 7.5% polyacrylamide gel as described above.

pp-vWF Serves as an Adhesion Substrate for ␣4␤1
Integrinexpressing Leukemia Cells-In a previous study (8), we have found that only a limited number of cell lines are capable of adhering to pp-vWF, i.e. only two cell lines of melanoma origin out of more than 15 cell lines of both normal and tumor tissue origin adhered well to pp-vWF. Therefore, we thought at first that the receptor for pp-vWF is rather specifically expressed on melanoma cells. However, this was not the case because several cells of leukemic origin adhered to pp-vWF (Table I). Two monocytic cell lines, THP-1 and U937, as well as two lymphocyte-like cells, Jurkat and MOLT-3, are capable of attaching to the pp-vWF substrate in the presence of a human ␤1 integrinactivating monoclonal antibody (mAb) TS2/16. In contrast, an erythroleukemic cell line K562, which also expresses ␤1 integrin, did not adhere to pp-vWF even in the presence of both TS2/16 and Mn 2ϩ ion. It is therefore very likely that the former four cell lines express integrin ␣ subunit responsible for adhesion to pp-vWF while K562 does not. When the expression of various ␣ subunit of integrin was analyzed by fluorescenceactivated cell sorting analysis, the adhesive property of these cells corresponded well to the expression level of ␣4 integrin ( Table I). As it is known that several melanoma cells also express high levels of ␣4␤1 integrin, it is suggested that the ␣ subunit of the ␤1 integrin receptor for pp-vWF is ␣4. To confirm this, the effect of anti-␣4 integrin mAbs on the cell adhesion to pp-vWF was assessed. As depicted in Table II, adhesion of human leukemia cells to pp-vWF was completely inhibited by an anti-human ␣4 mAb HP2/1, as well as an anti-human ␤1 mAb 4B4, but not by an anti-human ␣5 mAb BIIG2. None of the function-blocking antibodies against ␣2, ␣3, and ␣6 subunit affected this adhesion (data not shown). The adhesion of mouse melanoma B16 was also inhibited by an anti-mouse ␣4 mAb PS/2, indicating that VLA-4 integrin is the responsible adhesion receptor for pp-vWF on both melanoma and hematopoietic cells.
Adhesion of ␣4␤1 Integrin Transfectant to pp-vWF-To confirm that pp-vWF is a ligand for VLA-4 integrin, CHO cells were transfected with cDNA coding for human integrin subunits and the resultant stable transfectants were assessed for the ability to adhere to pp-vWF. As shown in Fig. 1A, the transfectant clones (␤1-CHO and ␣4␤1-CHO) express high levels of respective human integrin subunits. Both transfectants adhered well to plasma fibronectin (Fig. 1B), probably by using intrinsic ␣5 subunit complexed with both intrinsic and transfected ␤1 subunit. On the other hand, only ␣4␤1-CHO cells adhered to CS1 and VCAM-1, the known ligands for VLA-4, indicating that they express functional VLA-4 integrin. Furthermore, ␣4␤1-CHO but not ␤1-CHO adhered to pp-vWF. This adhesion of ␣4␤1-CHO to pp-vWF was completely inhibited by an anti-␣4 mAb HP2/1 (data not shown). These results strongly indicate that pp-vWF is a ligand for VLA-4 integrin.
Cell Adhesion Activity of Recombinant 8-kDa Fragment-As pp-vWF does not contain any homologous sequence to the known VLA-4 ligand sequences (CS1 region in fibronectin and first and fourth Ig domain in VCAM-1), we were interested in knowing what sequence in the pp-vWF primary structure is involved in cell adhesion. In a previous paper (8), we have already suggested that the cell adhesion site in the pp-vWF molecule resides within the central region of about 8 kDa. We have shown that an 8-kDa fragment generated by the lysylendopeptidase digest of bovine pp-vWF promotes melanoma cell adhesion in a dose-dependent manner. Furthermore, a mAb reactive with this fragment blocked the adhesive activity of pp-vWF. To determine the VLA-4 ligand sequence in the pp-vWF, we decided to construct a recombinant protein corresponding to this region. As amino acid sequence analysis of the isolated fragment suggests that it corresponds to a fragment having Ser 373 as the N terminus and extends to Lys 435 or Lys 455 , we prepared a recombinant fragment containing sequence 373-455 in human pp-vWF (r8k1A, Fig. 2) using a conventional bacterial fusion protein expression system. When coated on plastic surface, this recombinant fragment did promote attachment and spreading of B16 melanoma (Fig. 3). When compared at molar basis, the fragment showed similar coating concentration dependence as intact pp-vWF protein, suggesting that this fragment contains the authentic cell attachment site. To narrow down the location of the active site, we prepared other clones expressing the N-terminal (r8k1B) and C-terminal (r8k2A) half of the 8-kDa fragment (Fig. 2) and Ϫ ϩ Ϫ Ϫ ϩ Ϯ Ϯ a Expression of each integrin subunit was analyzed by FACS and scored according to the relative fluorescence intensity. Ϫ, no detectable fluorescence above background; Ϯ, less than 2-fold increase in mean fluorescence channel; ϩ, more than 2-fold but less than 20-fold increase in mean fluorescence channel; ϩϩ, more than 20-fold increase in mean fluorescence channel. The antibodies used are 6F1(␣2), P1B5(␣3), HP2/ 1(␣4), BIIG2(␣5), GoH3(␣6), and LM609(␣v␤3). Mouse ␣4 0 Ϯ 0 a Adhesions of the cells to pp-vWF were expressed as percent of the control that was conducted in the absence of antibodies. Data are mean Ϯ S.E. from three independent experiments in which quadruplicate determinations were made. The concentrations of the antibodies were 5 g/ml for purified antibody (4B4, HP2/1, and BIIG2) and 1:1000 dilution for ascites (PS/2). assessed the cell adhesion activity of GST fusion protein of these peptides. As depicted in Fig. 4, B16 melanoma cells adhered to GST-r8k1B but not to GST-r8k2A. Again the extent of cell adhesion to the recombinant fragment (r8k1B) was comparable to that of intact pp-vWF (more than 80% of the input cells were adhered) indicating that the active site was solely located in the N-terminal half 42-residue portion.
Cell Adhesion Activity of Synthetic Peptides Derived from the 8-kDa Fragment-To further narrow down the cell attachment site in the 8-kDa region, we synthesized a series of peptides corresponding to this region. Five 20-residue peptides with overlapping sequence covering the entire length of the 8-kDa fragment were synthesized (Table III) and tested for their ability to support B16 mouse melanoma cell adhesion. Among the 5 peptides, 3 were active in supporting cell attachment (Fig.  5). However, the adhesions to peptides designated T1 and T4 were completely reversed by the addition of heparin. On the other hand, the adhesion to peptide T2 was not affected by heparin at all and, furthermore, was blocked by treatment of cells with an anti-mouse ␣4 integrin mAb, which was also observed with the intact pp-vWF. It can be concluded that the integrin recognition sequence is contained in this 20-residue portion of pp-vWF (391-410). As adhesion to intact pp-vWF was not affected by heparin at all, the cell attachment to T1 and T4 peptides could be a result of nonspecific electrostatic interaction between the cell surface and peptides, which was not exhibited when these sequences are included in the intact protein.
To obtain shorter peptides with the cell adhesion activity, we synthesized truncated versions of T2 peptide. As it was expected that shorter peptides might have problem in coating on the plastic surface by ordinary protocol, we immobilized these peptides covalently on Sepharose beads and performed a cell adhesion assay using the peptide-conjugated beads. As clearly depicted in Fig. 6, both 15-and 10-residue truncated peptides (T2-15 and T2-10, respectively) were active in supporting adhesion of B16 murine melanoma. A peptide derived from the central portion of T1 (T1-8) did not show any adhesive property under this condition. Although the extent of adhesion (average number of adherent cells on beads) was almost the same for T2-15 and T2-10, cell spreading on T2-10-Sepharose was less obvious than T2-15, suggesting that the 15-residue extension is necessary for full activity.
Inhibition of the Cell Adhesion to pp-vWF by Soluble Peptides-When the effect of soluble peptides on the B16 cell adhesion to pp-vWF was assessed, the CS1 peptide strongly inhibited the adhesion (Fig. 7). Complete inhibition was achieved at 30 M. In contrast, a peptide containing RGD sequence did not affect adhesion at all, even at concentrations as high as 1 mM. Both peptide T2-15 and T2-10 had an inhib- itory effect on the B16 adhesion to pp-vWF. When added at 300 M, T2-15 inhibited cell adhesion by 75%. T2-10 had lower inhibitory activity compared with T2-15; the inhibition at 300 M was only 40%, suggesting that the affinity of T2-10 is significantly lower than that of T2-15. Two control peptides, T1-8 and BP-5 (derived from a distant region of pp-vWF sequence containing similar net charge and hydrophilicity value), had essentially no inhibitory effect. Similar inhibition by CS1, T2-15, and T2-10 was also observed with the B16 cell adhesion to CS1-rat serum albumin but not with adhesions to collagen and laminin (data not shown), suggesting that the effect of these peptides is specific to VLA-4-mediated adhesion.
Binding of VLA-4 Integrin to T2-15 Peptide Immobilized on Sepharose Beads-The results presented above strongly indicate that the T2-15 sequence (DCQDHSFSIVIETVQ) represents the VLA-4-binding site in the pp-vWF molecule. To confirm this, binding of VLA-4 integrin to the T2-15 peptide was evaluated. As B16 murine melanoma cells adhered on T2-15-Sepharose (Fig. 6), it is obvious that the peptide retains active conformation to recognize its receptor even on the Sepharose beads. A control peptide derived from the central portion of T1 (T1-8) did not show any adhesive property under this condition. These peptide-Sepharose conjugates were then used as affinity matrices to isolate the cell surface receptor. Octyl-␤-D-glucopyranoside extract of surface-iodinated B16 mouse melanoma was applied to an affinity column of T2-15-Sepharose, unbound materials were removed by washing, and the bound materials were eluted with EDTA. As shown in Fig. 8A, EDTA eluted two polypeptides of approximate molecular sizes of 150 and 125 kDa in the early step of elution from T2-15-Sepharose, followed by the elution of a 210-kDa protein in a rather retarded fraction. This 210-kDa band was also seen in the eluted fraction from a control column, T1-8-Sepharose, suggesting that it is a nonspecifically bound protein. The apparent molecular mass of the two polypeptides shifted from 150/125 kDa to 145/115 kDa when analyzed under nonreducing conditions (data not shown). These values are quite similar to that of the murine VLA-4 complex. To confirm that the bound material is a Net charge is calculated by assuming a ϩ1 net charge for Lys and Arg residues, and a net Ϫ1 charge for Glu and Asp at neutral pH. His is assumed to be uncharged at this pH.
b Calculated by the method of Kyte and Doolittle (45). According to this method, more hydrophilic peptides corresponds to the more negative values.

FIG. 5. Properties of cell adhesion activity of synthetic peptides derived from the 8-kDa portion of human pp-vWF.
Five synthetic peptides were subjected to cell adhesion assay using B16 mouse melanoma in the absence (closed column) or presence of 1:1000 dilution of anti-mouse ␣4 integrin mAb PS/2 ascites (open column) or 100 g/ml heparin (hatched column). Note that only the peptide designated T2 shows ␣4 integrin-dependent, heparin-independent adhesion like intact pp-vWF, while peptides T1 and T4 exhibit an ␣4 integrinindependent, heparin-inhibitable one. Data are mean Ϯ S.E. of one of the representative experiments in which triplicate determinations were made. BSA, bovine serum albumin. VLA-4 integrin, its reactivity toward available immunological probes for the various integrin ␣ subunits was examined. The EDTA-eluted proteins were subjected to immunoprecipitation using antisera against the cytoplasmic tail of integrin subunits ␣3, ␣4, ␣5, ␣7, ␣v, and an anti-human ␣6 rat mAb which can also recognize mouse ␣6 (GoH3). As clearly shown in Fig. 8B, only anti-␣4 antibody could immunoprecipitate radiolabeled polypeptide from the material eluted from T2-15-Sepharose. The relative molecular mass of the polypeptide is about 150 kDa, suggesting that the 125-kDa band (␤1 subunit) was lost during the immunoprecipitation procedure. This is consistent with a notion that ␣4␤1 integrin complex tends to be dissociated during immunoprecipitation experiments conducted in the presence of EDTA (14,31). It is concluded that T2-15 peptide can directly bind to VLA-4 complex in the presence of divalent cation and contains a novel ligand sequence for the VLA-4 integrin. DISCUSSION We report herein that the receptor responsible for cell adhesion to pp-vWF is VLA-4 (␣4␤1 integrin). At first we tested more than 20 cell lines, which grow on culture dishes, for their ability to adhere to pp-vWF and found that only the cultured tumor cell lines with melanoma origin, including mouse melanoma B16, human melanoma G-361 and MeWo, and hamster melanoma RPMI 1846, adhered on pp-vWF substrate (data not shown). We then thought that the receptor for pp-vWF might be a melanoma-specific molecule. When we tested cells of hematopoietic origin, however, most of those floating cells attached well to pp-vWF. It became clear that the receptor responsible for adhesion to pp-vWF was rather widely distributed among hematopoietic cells but its expression on adherent cells was limited to several melanoma cells. In a previous paper (8), we reported that the cell adhesion to pp-vWF was mediated by the ␤1 class of integrin using a function-blocking anti-␤1 mAb, but could not identify the corresponding ␣ subunit. Today 10 ␣ subunits (␣1-9 and ␣v) are known to be associated with ␤1 integrin. Among those, ␣4␤1 integrin (VLA-4) is known as a rather common antigen on hematopoietic cells (32) and it is also known that several melanoma cells express high levels of VLA-4 (33). As the expression of ␣4 but not other ␣ subunits correlated with the adhesive activity of hematopoietic cells to pp-vWF, we tested the effect of anti-␣4 mAbs on the cell adhesion to pp-vWF and found they completely inhibited adhesion. This strongly indicates that VLA-4 is the receptor for pp-vWF. Other mAbs with function blocking ability, including anti-␣2(6F1), ␣3(P1B5), ␣5(BIIG2), ␣6(GoH3), and ␣v(LM609), did not affect adhesion (data not shown), ruling out these subunits as receptor candidates. Finally, transfection of ␣4 integrin cDNA into CHO cells resulted in acquisition of adhesion activity toward pp-vWF as well as other well known ligands; CS1 and VCAM-1. It is concluded at this point that the cell adhesion receptor to pp-vWF is ␣4␤1 integrin.
VLA-4 is an integrin complex that recognizes both CS1 region in fibronectin and VCAM-1. Lymphocyte adhesion to endothelial cells is primarily mediated by interaction of lymphocyte VLA-4 with VCAM-1 expressed on cytokine-activated endothelial cells and this pathway is thought to be central to the lymphocyte recruitment to the site of inflammation (34,35). In addition, VLA-4/VCAM-1 interaction is involved in lymphocyte homing to high endothelial venule in peripheral lymph nodes where VCAM-1 is constitutively expressed (36). Leukocyte extravasation at the site of inflammation as well as the vascular wall injury is also thought to be mediated by leukocyte integrins, including VLA-4 although other adhesion receptors such as selectins play some part (10). These adhesions need to occur immediately at the right place, which means the adhesive ligands must be distributed spatially and timely. Fibronectin is a molecule abundantly found in almost all tissues and its distribution is hard to be controlled by immediate inflammatory response. VCAM-1 is an inducible cell surface molecule on endothelial cells and is a desirable docking molecule for leukocyte recruitment. However, VCAM-1 must be newly synthesized before expression on the cells and it is only detected after 2 h treatment of cultured endothelial cells with tumor necrosis factor-␣ (37). This means that VCAM-1 cannot mediate leukocyte adhesion in the very early steps in the inflammation. VCAM-1/VLA-4 pathway also cannot work when the endothelial cells are seriously injured or removed from the vascular wall. If there is another molecule that can mediate leukocyte adhesion in the very early step of inflammatory response, the whole system would be more effective. Vonderheide and Springer (38) have suggested that there is an unknown ligand for VLA-4 on endothelium because VLA-4-dependent adhesion of lymphoid cells to cultured endothelial cells is not completely inhibited by antibodies against VCAM-1 and fibronectin. Hahne et al. (39) have also suggested that there is an unknown leukocyte adhesion mechanism on mouse endothelioma cells that is not mediated by VCAM-1, ICAM-1, E-selectin, or Pselectin. Our results that VLA-4 also recognizes pp-vWF as an adhesion substrate may suggest that the adhesion to pp-vWF functions as a physiologically important pathway of leukocyte recruitment to the inflammatory/vascular injury sites. Biosynthesis of vWF precursor is limited in megakaryocytes and endothelial cells (4). Only platelets and endothelial cells are places that pp-vWF protein exists in the whole body and pp-vWF is immediately released from these cells upon activation by agonists such as thrombin (40). As pp-vWF has the ability to bind to both collagen and laminin (5,7), it may be deposited at the exposed subendothelial matrix, thus presenting an adhesion site for VLA-4-bearing leukocytes. In preliminary experiments, we have observed that pp-vWF rapidly accumulates at injured sites of glomerular vasculature during the experimental glomerulonephritis while mature vWF does not. 2 It is possible that pp-vWF functions as an emergency tag for leukocytes/lymphocytes targeting and mediates successful recruitment and infiltration of these inflammatory cells together with VCAM-1.
We found that an amino acid sequence, DCQDHSFSIVI-ETVQ, derived from the midregion of human pp-vWF, was a novel ligand sequence of VLA-4. A peptide with this sequence by itself could promote cell adhesion in an ␣4␤1 integrin-dependent manner, and, moreover, direct binding of the VLA-4 integrin complex to this peptide is observed. Although the shorter version of the peptide, T2-10 (DCQDHSFSIV), also had cell adhesion activity, the inhibitory activity of this peptide on cell adhesion to pp-vWF was significantly lower than that of T2-15. It suggests that the 10-residue portion is essential for the recognition by VLA-4, but the C-terminal extension increases the affinity and/or stabilizes the favorable conformation of the peptide. CS1 and T2-15 peptide both can inhibit the cell adhesion to pp-vWF as well as to CS1-bovine serum albumin, although these peptides share no homologous motif in their sequences. It is possible that both peptides can bind to the ligand binding pocket of VLA-4 in a similar way, thus inhibiting the interaction of each other. Although the tertiary structure for the pp-vWF protein has not been resolved, a computeraided prediction of secondary structure of the region containing this sequence by using the algorithm of Robson (41) suggests that the segment Asp 395 -Phe 401 form a ␤-turn structure (data not shown). It is possible that this segment forms loop structures facing outward and is thus available to interact with the cell surface. As this sequence contains neither the essential VLA-4-recognition motif identified in CS1 (EILDV) (42) nor that in VCAM-1 (QIDSPL) (43), it seems that this sequence represents a novel integrin-binding motif. However, it contains a sequence, VIET, which is similar to the sequence around the essential glutamate residue (Glu 34 ) in ICAM-1 (GIET) (44). As these sequences all contain some oxygenated residues (Asp, Glu, Thr, or Ser) as well as isoleucine residue, these different but rather similar sequences might be utilized as general integrin binding motifs in each molecule together with other structures that define the specificity. Determination of minimum essential residues for the binding to the integrin within the T2-15 sequence is underway.
In conclusion, our finding that pp-vWF serves as a ligand for VLA-4 integrin provides new insights on the physiological significance of this vWF gene product which was originally thought as a mere propeptide without any particular functions. Furthermore, elucidation of mechanisms of the cell adhesion to pp-vWF will help in our understanding of mech-anisms underlying the melanoma metastasis as well as vascular inflammation.