identification of the collagen-binding domain of

CELL STRUCTURE AND FUNCTION 13, 217-225 (1988)

C by Japan Society for Cell Biology

Identification of the Collagen-Binding Domain of

Vitronectin Using Monoclonal Antibodies

Masako Izumi, Tadashi Shimo-Oka*, Naoki Morishita*, Ichio Ii*, and

Masao Hayashi+

Department of Biology, Ochanomizu University, Bunkyo-ku, Tokyo 112, and *Tissue Culture Laboratory , Iwaki Glass Co., Ltd., Gyoda, Funabashi, Chiba

273, Japan

ABSTRACT. Vitronectin is a 75 kilodalton (kDa) cell-adhesive glycoprotein found in animal blood and connective tissue, also termed serum spreading factor, S-protein, and epibolin. It promotes attachment and spreading of animal cells on tissue culture dishes, and it also binds to collagen. We established four mouse hybridoma lines producing monoclonal antibodies (M1, M2, M4 and M5) to human vitronectin. By immunoblotting, both epitopes recognized by M4 and M5 were suggested to exist in the amino terminal 5 kDa portion of vitronectin, and both Ml and M2 bound to the adjacent 35 kDa portion. Cell spreading on vitronectin-coated dishes was inhibited by M4 = M5 > M1, but not by M2. Collagen binding to vitronectin was inhibited by M2 >M4 =M5, but not by M1. These results indicate that the collagen-binding site is located near the cell-binding site in the amino terminal half of vitronectin. Independent inhibition of vitronectin binding to the cell and to collagen by these monoclonal antibodies will provide a potential tool to dissect the structure and function of vitronectin.

Vitronectin is a cell-adhesive glycoprotein found in animal blood and connective tissue (4, 7, 9, 14), also termed serum spreading factor (3), S-protein (16), and epibolin (18). Since vitronectin promotes the attachment and spreading of a variety of animal cells on tissue culture dishes, it is assumed that vitronectin as well as fibronectin (10, 22) are the major cell-attachment factors in the animal serum used for tissue culture. Furthermore, vitronectin also functions in blood coagulation (5, 11, 17) and complement action (16). In connective tissues, vitronectin exists as one of extracellular matrix proteins, probably interacting with cells directly or associating with heparin-like glycosaminoglycans and collagens, which are reported to bind to vitronectin in vitro (1, 6, 9).

The entire primary structure of vitronectin has been deduced from its cDNA

(13, 19). One amino acid sequence in the cell-binding site was determined to be Arg-Gly-Asp and is located near the amino terminus (19, 20). The heparin-binding site which is cryptic in the native form of vitronectin (2, 8) is located, on the other hand, near the carboxyl terminus (20).

Gebb et al. (6) first reported on the binding of vitronectin to native collagens, but

+ To whom correspondence should be addressed.

Abbreviations used: kDa, kilodalton; ELISA, enzyme-linked immunosorbent assay; SDS, sodium

dodecyl sulfate; BSA, bovine serum albumin; PBS, phosphate-buffered saline.

217

218 M. Izumi et al.

the collagen-binding site was not identified. Using different monoclonal antibodies, we describe in this paper our finding that the collagen-binding site exists near the

cell-binding site in the amino terminal half of vitronectin.

MATERIALS AND METHODS

Vitronectin. Vitronectin was purified from outdated human plasma by heparin affinity

chromatography according to the method of Yatohgo et al. (23) or purchased from Iwaki

Glass (V0001, Iwaki Glass, Tokyo).

Polyclonal antibody. Polyclonal rabbit anti-human vitronectin antibody was prepared as

described previously (1).

Monoclonal antibodies. Monoclonal antibodies were prepared essentially according to

the procedure described by Iwasaki et al. (12). Human vitronectin was mixed with complete

Freund's adjuvant and was intraperitoneally injected into 8-week-old BALB/C female mice

at a dose of 30 ƒÊg vitronectin/mouse. Two weeks later, 40 ƒÊg of vitronectin with incomplete

Freund's adjuvant were injected intraperitoneally. On the 17th day after the 2nd injection,

50 ƒÊg of vitronectin were injected intravenously. Three days later, the spleen cells of the

immunized mice were fused with the mouse myeloma cell line X63-Ag8.653 using polyethylene

glycol to generate primary hybridoma cultures. To select those producting antibodies reacting

with human vitronectin, the hybridoma cultures were screened with an enzyme-linked

immunosorbent assay (ELISA) and by immunoblotting. Cloning was carried out several times

by limiting dilution until monoclonality was evidenced. Finally, 4 hybridoma lines producing

anti-vitronectin antibodies (M1, M2, M4, and M5) were established. Monoclonal antibodies

were obtained by growing the cloned hybridomas in a serum-free culture medium (M 1) or

from ascites fluid produced in syngeneic mice (M1, M2, M4, and M5). The antibodies were

purified with a protein A agarose MAPSTM-II Kit (Nippon Bio-Rad Lab., Tokyo). Isotyping

of antibodies was performed with a Mouse-TyperTM Sub-Isotyping Kit (Nippon Bio-Rad

Lab., Tokyo).

ELISA. An ELISA 96 well plate of polystyrene (MS-3496F, Sumitomo-Bakelite, Tokyo)

was coated with 50 ƒÊl of vitronectin (1 ƒÊg/ml) in phosphate-buffered saline (PBS) at 37•Ž for

1 h. The plate was rinsed and blocked with 250 ƒÊl of 1% (w/v) bovine serum albumin (BSA)

in PBS at room temperature for 1 h. Then the plate was rinsed again and incubated with 50 ƒÊl

of the culture supernatants from the hybridomas or with the monoclonal antibodies in

1% BSA at the indicated concentrations at room temperature for 1 h. The plate was rinsed

again and incubated with 50 ƒÊl of horseradish peroxidase-conjugated goat antibody against

mouse IgG at a 1 to 5,000 dilution in 1% BSA at room temperature for 1 h. After the plate

was rinsed again, the bound enzymatic activity was measured using o-phenylenediamine and

H2O2 as substrates.

Chemical fragmentation and immunoblotting of vitronectin. Vitronectin was cleaved us-

ing 70% formic acid, with or without CNBr, according to the procedure of Suzuki et al. (20).

One hundred ƒÊg of vitronectin in 120 ƒÊl PBS were incubated at 4•Ž or 25•Ž for 16 h with

280 ƒÊl of formic acid containing 50 mg of CNBr. After diluting with 10 volumes of water, the

vitronectin fragments produced at 4•Ž were mixed with those produced at 25•Ž. The mixture

was then lyophilized and solubilized with 200 ƒÊl of sodium dodecyl sulfate (SDS)-sample buffer

containing 5% (v/v) 2-mercaptoethanol. Vitronectin was also fragmented with 70% formic

acid in the absence of CNBr at 37•Ž for 16 h, lyophilized, and solubilized as described above.

SDS-polyacrylamide gel electrophoresis was performed using 1 to 7.5 ƒÊg protein/lane by

the method of Laemmli (15). Immunoblotting was performed essentially according to the

procedure of Towbin et al. (21) using the supernatants from the hybridoma cultures or ascites

Collagen-Binding Domain of Vitronectin 219

fluid diluted 1,000 to 500,000 times. Immunoblotting of human serum was performed using

0.25 ƒÊl human serum/lane.

Radiolabelling. Vitronectin (50 ƒÊg in 100 ƒÊl PBS) was mixed with 10 ƒÊl of carrier free 125I-Nal (100 mCi/ml) and then three 20 ƒÊl aliquots of freshly prepared 1 mg/ml chloramine

T in 0.2 M Na-phosphate buffer (pH 7.2). Labelled vitronectin was separated by Sephadex

G-50 fine column chromatography. The specific activity of the 125I-vitronectin

was 18 mCi/mg (1.7 •~ 107 cpm/ ƒÊg), indicating that the molar iodination ratio was 0.5.

Cell-spreading assay. A 96 well tissue culture plate of polystyrene (1-67008, Nunc,

Denmark) was incubated with 50 ƒÊl of 0.5 ƒÊg/ml vitronectin in Grinnell's adhesion medium

(150 mM NaCl, 1 mM Cal2, 3 mM KCl, 0.5 mM MgCl2, 6 mM Na2HPO4, and 1 mM KH2

PO4; final pH 7.3) at 37•Ž for 1 h (7). After discarding the solution, the plate was blocked

with 250 pl of 1% (w/v) BSA in PBS at 37•Ž for 30 min, and then treated with 50 ƒÊl of

various antibodies in 1% BSA in PBS at 37•Ž for 1 h. After rinsing again, the plate received

2 •~ 104 BHK cells suspended in 0.1 ml adhesion medium and was incubated at 37•Ž for 1.5 h.

The percent of spread cells (number of spread cells per 100 attached cells) was quantitated

microscopically.

Collagen-binding assay. An ELISA 96 well flexible plate of vinyl chloride (MS-7196F,

Sumitomo-Bakelite, Tokyo) was incubated with 100 pl of 10 ƒÊg/ml type I collagen from

porcine skin (Cellmatrix type I-P, Nittagelatin, Osaka) or calf skin (Sigma C-3511, St. Louis,

Mo, USA) in 100 mM Na-carbonate buffer (pH 9.6) at 30•Ž for 1 h. In a control experiment,

the plate was coated under similar conditions with BSA instead of type I collagen. After

rinsing three times with 150 pl of 0.05% (w/v) Tween 20 and 20 mM Na-phosphate buffer

(pH 7.4), the plate was incubated with 50 pl of 125I-vitronectin (0.025 ƒÊg/ml) and 0-25 ƒÊg/ml

anti-vitronectin antibody in 0.05% Tween 20 and 20 mM Na-phosphate buffer (pH 7.4) at

30•Ž for 1 h. After rinsing three times again with 150 ƒÊl of 0.05% Tween 20 and 20 mM

Na-phosphate buffer (pH 7.4), the plate was cut up with scissors. The radioactivity of the

125I-vitronectin bound to the collagen in each well was measured with a gamma counter

(Aloka ARC-500, Tokyo).

RESULTS

Monoclonal antibodies. The spleens of 3 mice immunized with human vitronectin

produced 288 viable cultures after hybridization which reacted with partially

purified vitronectin in ELISA. Forty one of these were selected for cloning. Cultures

cloned by limiting dilution were screened for the production of antibodies which

reacted with vitronectin by ELISA and with the 75 kDa and 65 kDa vitronectin

bands by immunoblotting. From this, 4 hybridoma lines were established and

expanded as ascites tumors in mice or in serum-free cell cultures. These hybridomas

produced antibodies (M1, M2, M4, and M5) that reacted only with the 75 kDa and

65 kDa vitronectin bands of human serum by immunoblotting.

One ml of ascites fluid yielded 1.1 mg of M1, 15.1 mg of M2, 17.4 mg of M4, and

14.6 mg of M5 IgG protein. Since the yield of M1 was very low, M1 was also obtained

from serum-free cell cultures. Subclass analysis revealed that M2, M4, and M5

were of the IgG1 subclass containing K light chains, but that M1 was an IgA type

containing K light chains. The four monoclonal antibodies reacted with vitronectin

by ELISA in the following order, in decreasing strength: M2 •¬ M5 _•¬ M4 > M1

(Fig. 1). The low yield and low reactivity of M1 might be due to its IgA nature.

Immunoblotting of vitronectin fragments. Chemical fragmentation of vitronectin

with formic acid primarily produced 40 kDa, 32 kDa, and 26 kDa fragments

220 M. Izumi et al.

along with a few other molecular weight fragments (Fig. 2A). According to the results of Suzuki et al. (20), the 40 kDa fragment includes the amino terminus of vitronectin, whereas the 32 kDa fragment is located in the carboxyl terminal half of vitronectin and does not share any portion with the above 40 kDa fragment. The entire 26 kDa fragment is included in the 32 kDa fragment. All of monoclonal antibodies, Ml , M2, M4, and M5, selectively reacted with the 40 kDa fragment as well as uncleaved 75 kDa and 65 kDa vitronectin, but they reacted with neither the 32 kDa fragment nor the 26 kDa fragment.

To determine more precisely the location of their epitopes, the binding of the antibodies to CNBr fragments of vitronectin was examined. CNBr treatment of vitronectin produced 53 kDa, 43 kDa, and 35 kDa fragments. According to the results of Suzuki et al. (20), the 53 kDa fragment includes the entire 43 kDa fragment, which in turn includes the entire 35 kDa fragment. Both the 43 kDa and 35 kDa fragments lack the amino terminal 5 kDa portion of vitronectin whereas the 53 kDa fragment retains it. Immunoblotting with the antibodies showed that both M4 and M5 selectively reacted with the 53 kDa fragment, but not with the 43 kDa and 35 kDa fragments. On the other hand, both Ml and M2 reacted not only with the 53 kDa fragment, but also with the 43 kDa and 35 kDa fragments (Fig. 2B). These results suggest that the epitopes for M4 and M5 exist in the 5 kDa portion at the amino terminus of vitronectin, and that M1 and M2 recognize epitopes existing in the adjacent 35 kDa portion.

Fig. 1. ELISA of monoclonal antibodies. An ELISA 96 well plate was coated with 1 ƒÊg/ml of

human vitronectin at 37•Ž for 1 h. The plate was rinsed and blocked with 1% BSA in PBS at room

temperature for 1 h. After rinsing, the plate was incubated with the indicated concentrations of

monoclonal antibodies in 1% BSA in PBS at room temperature for 1 h. After rinsing again, the plate

was incubated at room temperature for 1 h with horseradish peroxidase-conjugated goat anti-mouse IgG

at a 1 to 5,000 dilution in 1% BSA in PBS, and then received 100 ƒÊ of 0.4 mg/ml o-phenylenediamine,

2.5 mM H2O2, 50 mM citric acid, and 0.1 M Na2HPO4. Color was allowed to develop for 10 min. The

absorbance at 492 nm represents the amount of monoclonal antibody bound to vitronectin. Ml, •›; M2,

•¢; M4,• ; M5, •œ.


(A) (B)

Inhibition of cell-spreading by antibodies. The cell-binding site of vitronectin is known to be located in the amino terminal 5 kDa portion (13, 19, 20). The four monoclonal antibodies as well as a polyclonal antibody were tested to see whether

Fig. 2. Immunoblotting of vitronectin fragments with monoclonal antibodies. Vitronectin fragments produced by cleavage with formic acid (A) and with CNBr (B) were stained with Coomassie blue (lane CB) or immunoblotted with monoclonal antibodies M1 (lane 1), M2 (lane 2), M4 (lane 4), and M5 (lane 5) as indicated. The size in kDa is shown at the left.

Fig. 3. Inhibition of vitronectin-induced cell-spreading activity by antibodies. A 96 well tissue

culture plate was coated with vitronectin (0.5 ƒÊg/ml) at 37•Ž for 1 h. The plate was rinsed and incubated

with 1% BSA in PBS at 37•Ž for 30 min, and then with the indicated concentrations of antibodies at

37•Ž for 1 h. After rinsing again, the plate was incubated with 2 •~ 104 BHK cells at 37•Ž for 1.5 h. The

percent of spread cells was quantitated microscopically. Sixty seven % of the cells spread in the absence

of antibody and this was regarded as 100% "cell spreading". M1, •›; M2, •¢; M4, • ; M5, •œ;

polyclonal antibody to vitronectin, •£.

222 M. Izumi et al.

they inhibited cell-spreading activity in a tissue culture plate which had been coated

with vitronectin. Figure 3 shows that BHK cell spreading was strongly inhibited by

M4 and M5, and moderately inhibited by M1 as well as the polyclonal antibody, but

M2 had no effect. These results support the conclusion drawn from the above

immunoblotting experiments that the epitopes for both M4 and M5 exist near the

amino terminal cell-binding site. Furthermore, the results shown in Fig. 3 indicate

that the epitope for M1 exists near the cell-binding site but the epitope for M2 exists

far from it.

Inhibition of collagen-binding by antibodies. Vitronectin is known to bind to

type I collagen (6), but the location of the collagen-binding site is not known. Before

examining the inhibition of the collagen-vitronectin interaction by antibody, the

collagen-binding properties of vitronectin were surveyed. Figure 4A shows that 125I-vitronectin in 0 .05% Tween 20 and 20 mM Na-phosphate buffer (pH 7.4) at 30•Ž

specifically bound to a type I collagen-coated plate within 1 h. The amounts of

vitronectin bound to collagen increased linearly with increasing concentrations of

added vitronectin in the range of from 0 to 0.2 ƒÊg/ml. The relationship remained

linear up to 2 ƒÊg/ml of vitronectin (data not shown). Porcine collagen from

Nittagelatin had a slightly higher binding activity than calf skin collagen from

Sigma. The vitronectin binding to collagen was not inhibited by the addition of

BSA to a final concentration of 100 ƒÊg/ml. Physiological or lower than

physiological ionic strength buffer, however, remarkably inhibited the binding;

0.05 M and 0.1 M NaCl inhibited the binding by 72% and 95%, respectively.

(A) (B)

Fig. 4. Vitronectin binding to collagen and its inhibition by antibodies. (A) Various concentrations

of 125I-vitronectin (8.6 •~ 105 cpm/ƒÊg) were incubated at 30•Ž for 1 h in a flexible plate previously coated

with 10 ƒÊg/ml porcine type I collagen (•›) or BSA (•œ). After rinsing, each well was cut out of the plate.

The 125I-vitronectin bound to each well was measured with a gamma counter. (B) 125I-vitronectin

(25 ng/ml, 8.7 •~ 106 cpm/ƒÊg) was mixed with various concentrations of antibodies in 0.05% Tween 20

and 20 mM Na-phosphate buffer (pH 7.4), and incubated at 30•Ž for 1 h with a flexible plate previously

coated with 10 ƒÊg/ml of porcine type I collagen. Bound 125I-vitronectin was quantitated as described in

(A). VN indicates vitronectin. M1, •›; M2, •¢; M4, • ; M5, •œ; polyclonal antibody to vitronectin, •£.


Therefore, antibody inhibition experiments were performed at low ionic strength;

50 ƒÊl of 0.025 ƒÊg/ml 125I-vitronectin was mixed with antibody to make a final con-

centration of 0 to 25 ƒÊg/ml in a 0.05% Tween 20 and 20 mM Na-phosphate buffer

(pH 7.4) and then incubated with a porcine type I collagen-coated plate. Figure 4B

shows that the binding of 125I-vitronectin to the collagen-coated plate was strongly

inhibited by the polyclonal antibody, moderately by M2, and weakly by M4 and

M5, whereas M1 had no effect. These results indicate that the collagen-binding site

is located near the cell-binding site.

DISCUSSION

From the differential inhibition of vitronectin's binding activities by four

monoclonal antibodies, we have shown that the collagen-binding site is located near

the cell-binding site on the amino terminal half of vitronectin. Inhibition of cell

spreading by the four monoclonal antibodies decreased in the following order:

M4= M5 > M1, while no inhibition by M2 was observed (Fig. 3). In contrast, the

inhibition of collagen-binding activity caused by the antibodies decreased in the

following order: M2 >M4 = M5, while no inhibition was observed using M1 (Fig.

4). The results of the immunoblotting suggest that both M4 and M5 recognize the

epitopes in the amino terminal 5 kDa portion and that both M1 and M2 recognize

the adjacent 35 kDa portion (Fig. 2). The binding activity of M1 was apparently

1,000-100,000-fold lower than those of M2, M4, and M5 (Fig. 1). The binding

activity of M1 might be, however, considerably underestimated because M1, which is

an IgA, was assayed using an anti-IgG antibody. As a matter of fact, an effect on cell

spreading by M1 was observed at 2 ƒÊg/ml although no effect on collagen binding

was observed even at 25 ƒÊg/ml (Fig. 3 and 4). Therefore, the lack of interference by

(A)

(B)

Fig. 5. Functional domain model of vitronectin. Heparin-binding site Hep located near the carboxyl terminus is cryptic in the native form of vitronectin (2, 8). The amino terminal portion of vitronectin which involves the cell-binding site Cell turns back on itself. The collagen-binding site Col exists near the cell-binding site in two possible arrangements; they are close together in the one-dimensional primary structure in model (A), or far from each other in the primary structure but conformationally close due to polypeptide folding in model (B). The proposed positions of epitopes recognized by M1 , M2, M4, and

M5 are mapped with arrows.

224 M. Izumi et al.

M1 in the binding of vitronectin to collagen is suspected to be due to M1 's own properties and not due to the use of lower doses.

Assuming that the monoclonal antibodies do not induce a conformational change in vitronectin, the binding site for the cell and that for collagen cannot be mapped on a one-dimensional linear model of vitronectin. The model needs a two-dimen-sional structure, in which the cell-binding site near the amino terminus is proposed to turn back on itself. To satisfy the above requirements, we have diagrammed two possible two-dimensional models of vitronectin with the arrangement of cell-binding and collagen-binding sites as shown in Fig. 5. In model (A), the proximity between the collagen-binding site and cell-binding site is achieved by placing them close together in the one-dimensional primary structure of vitronectin. In model (B), on the other hand, the two binding sites are distant to each other in the primary structure, but exist close together due to polypeptide folding. To establish which model in Fig. 5 is correct, further experiments including isolation of the collagen-binding domain would be advisable. These models support Preissner's model of vitronectin based on computer analysis of the amino acid sequence, in which the amino terminal portion folds into the vitronectin molecule (17).

We have confirmed the binding of vitronectin to type I collagen originally reported by Gebb et al. (6). They reported that higher than physiological salt concen-trations interfere with the binding of vitronectin to collagen. However, in our hands, binding was remarkably inhibited even by physiological concentrations of

NaCl. Therefore, the interaction between vitronectin and collagen may not be physiological. If it is, the interaction does not seem to be strong in vivo.

M4 and M5 inhibited both the cell spreading and collagen binding activities of vitronectin. However, M1 specifically inhibited the cell spreading activity and M2 specifically inhibited the collagen binding activity. These function-specific monoclonal anti-vitronectin antibodies are described in this paper for the first time, and as specific functional inhibitors, they will provide a strong tool in the clarification of the molecular mechanisms and biological functions of vitronectin in vivo as well as in vitro.

Acknowledgments. We wish to thank Ms. Kazuko Hayashi for her secretarial assistance.

REFERENCES

1. AKAMA, T., K.M. YAMADA, N. SENO, I. MATSUMOTO, I. KONO, H. KASHIWAGI, T. FUNAKI and M.

HAYASHI. Immunological characterization of human vitronectin and its binding to

glycosaminoglycans. J. Biochem. 100, 1343-1351, 1986 2. BARNES, D.W., J.E. REING and B. Amos. Heparin-binding properties of human serum spreading

factor. J. Biol. Chem. 260, 9117-9122, 1985

3. BARNES, D.W. and J. SILNUTZER. Isolation of human serum spreading factor. J. Biol. Chem. 258,

12548-12552, 1983

4. BARNES, D., R. WOLFE, G. SERRERO, D. McCLURE and G. SATO. Effects of a serum spreading

factor on growth and morphology of cells in serum-free medium. J. Supramolec. Struct. 14, 47-63,

1980

5. DAHLBACK, B. and E.R. PODACK. Characterization of human S-protein, an inhibitor of membrane

attack complex of complement. Demonstration of a free reactive thiol group. Biochemistry 24,

2368-2374, 1985

6. GEBB, C., E.G. HAYMAN, E. ENGVALL and E. RUOSLAHTI. Interaction of vitronectin with collagen.

J. Biol. Chem. 261, 16698-16703, 1986

7. GRINNELL, F., D.G. HAYS and D. MINTER. Cell adhesion and spreading factor. Partial purification


and properties. Exp. Cell Res. 110, 175-190, 1977 8. HAYASHI, M., T. AICAMA, I. Korro and H. KASHIWAGI. Activation of vitronectin (serum spreading

factor) binding of heparin by denaturing agents. J. Biochem. 98, 1135-1138, 1985 9. HAYMAN, E.G., M.D. PIERSCHBACHER, Y. OHGREN and E. RUOSLAHTI. Serum spreading factor

(vitronectin ) is present at the cell surface and in tissues. Proc. Natl. Acad. Sci. USA 80, 4003-4007, 1983

10. HYNES, R.O. Fibronectins. Sci. Am. 254, 32-41, 1986 11. ILL, C.R. and E.RUOSLAHTI. Association of thrombin-antithrombin III complex with vitronectin in

serum. J. Biol. Chem. 260, 15610-15615, 1985 12. IwAsmci, T., T. ANDO, K. ICHIKAWA and K. YASUI. Monoclonal Antibody (in Japanese), Kodansha

Scientific, Tokyo, 1982 13. JENNE, D. and K.K. STANLEY. Molecular cloning of S-protein, a link between complement,

coagulation and cell-substrate adhesion. EMBO J. 4, 3153-3157, 1985 14. Krrox, P. and S. GRIFFITHS. The distribution of cell-spreading activities in sera: a quantitative

approach. J. Cell Sci. 46, 97-112, 1980 15. LAEMMLI, U.K. Cleavage of structural proteins during the assembly of the head of bacteriophage

T4. Nature 227, 680-685, 1970 16. PODACK, E.R., W.P. KOLB AND H.J. MULLER-EBERHARD. The C5b-6 complex: formation,

isolation and inhibition of its activity by lipoprotein and the S-protein of human serum. J. Immunol. 120, 1841-1848, 1978

17. PREISSNER, K.T. and G. MULLER-BERGHAUS. Neutralization and binding of heparin by S protein/vitronectin in the inhibition of factor Xa by antithrombin III. J. Biol. Chem. 262, 12247-12253, 1987

18. STENN, K.S. Epibolin: a protein of human plasma that supports epithelial cell movement. Proc. Natl. Acad. Sci. USA 78, 6907-6911, 1981

19. Suzum, S., A. OLDBERG, E.G. HAYMAN, M.D. Pierschbacher and E. Ruoslahti. Complete amino acid sequence of human vitronectin deduced from cDNA. Similarity of cell attachment sites in vitronectin and fibronectin. EMBO J. 4, 2519-2524, 1985

20. SUZUKI, S., M.D. PIERSCHBACHER, E.G. HAYMAN, K. NGUYEN, Y. OHGREN AND E. RUOSLAHTI. Domain structure of vitronectin. Alignment of active sites. J. Biol. Chem. 259, 15307-15314, 1984

21. TOWBIN, H., T. STAEHELIN and J. GORDON. Electrophoretic transfer of proteins from

polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. USA 76, 4350-4354, 1979

22. YAMADA, K.M. Cell surface interactions with extracellular materials. Ann. Rev. Biochem. 52, 761-799, 1983

23. YATOHGO, T., M. IzuMI, H. KASHIWAGI and M. HAYASHI. Novel purification of vitronectin from human plasma by heparin affinity chromatography. Cell Struct. Funct. 13, in press, 1988

(Received for publication, March 13, 1988)

identification of the collagen-binding domain of

Documents