Yolk Vitronectin PURIFICATION AND DIFFERENCES FROM ITS BLOOD HOMOLOGUE IN MOLECULAR SIZE, HEPARIN BINDING, COLLAGEN BINDING, AND BOUND CARBOHYDRATE*

This is the first report on a unique vitronectin molecule, yolk vitronectin, which is similar to its blood homologue in cell spreading activity but different in molecular size, bound carbohydrate, and heparin and collagen binding activity. Yolk vitronectin was purified 2,600-fold from chick egg yolk by a combination of hydroxylapatite, DEAE-cellulose, and anti-vitro-nectin-Sepharose column chromatographies. In SDS- polyacrylamide gel electrophoresis under reducing conditions, yolk vitronectin was separated into 64- and 46-kDa bands, which are 16 and 26 kDa smaller, respectively, than the 70-kDa major band of chick blood vitronectin. The 64-kDa band shares the same NHa-terminal sequence as chick blood vitronectin. In contrast, the NHa-terminal sequence of the 45-kDa band is somewhat homologous with the internal sequences of mammalian vitronectins beginning at the 50th amino acid from the NHa terminus. The bound carbohydrate of the 64- and 46-kDa species of yolk vitronectin is similar to, but distinct from, that of blood vitronectin. Unlike blood vitronectin, yolk vitronectin cannot bind to either heparin or collagen. is multifunctional glycoprotein in


6357.
$ To whom correspondence should be addressed. Fax: 81-3-3943- (McGuire et al., 1988;Korc-Grodzicki et al., 1988), sulfated (Jenne et al., 1989), and cross-linked by transglutaminase (Sane et al., 1988). Its physiological functions are based on the above interactions and/or modifications. Vitronectin allows cells to adhere to a substrate, promotes haptotaxis of cells (Basara et al., 1985;Naito et al., 1991), modulates thrombin and plasmin activity in fibrinolysis, and prevents the cell lytic action of the membrane attack complex. But, the role, if any, of vitronectin in animal development and even the existence of vitronectin in embryos had not been established.
Fibronectin, a plasma glycoprotein similar to but distinct from vitronectin, also promotes cell spreading and cell migration in vitro (Yamada et al., 1976;Ali and Hynes, 1978;Rovasio et al., 1983). Extending this observation, Thiery and his colleagues have presented evidence that one of the major functions of fibronectin in uiuo is related to migration of neural crest cells and gastrulation in the early development of embryos (for reviews, see Thiery et al. (1985,1989)). Many other cell adhesion proteins, such as cadherin, laminin, and tenascin, play a role in morphogenesis in the early stages of animal development (for reviews, see Takeichi (1988), Ekblom et al. (1986), and Erickson and Bourdon (1989)). Particular combinations of these adhesion proteins and their cell surface receptors may be the key to the mechanisms by which cells are arranged to build the fine architecture of tissues during development.
We have, therefore, started to study the role of vitronectin in early development using chick embryos and anti-chicken blood plasma vitronectin. In an early stage of the experiments, surprisingly, we found that one of the egg yolk proteins reacted with anti-vitronectin. Yolk proteins are expected to influence the development of the embryo. So, we set out to characterize the yolk protein with the aim of obtaining new insight into vitronectin function in development. The yolk protein reacting with anti-vitronectin is called yolk vitronectin. In this paper, we describe its purification and a comparison of its heparin binding, collagen binding, bound carbohydrate, and molecular size with those of blood vitronectin of the same species.

EXPERIMENTAL PROCEDURES
Materials-Chick eggs were used on the day of laying or purchased commercially. Chicken blood serum and plasma were obtained from the same hen with or without the addition of a 1/7 volume of 3.18% sodium citrate as an anticoagulant and centrifuged immediately at 3,000 rpm for 10 min. For preparation of chicken blood vitronectin, pooled chicken blood plasma was obtained from Ichirei Inc. (Saitama, Japan) and stored at -20 "C until use. Chicken blood vitronectin was purified from pooled chicken blood plasma as described previously (Yatohgo et al., 1988;Kitagaki-Ogawa et al., 1990). Antibody to chicken blood vitronectin was raised in a rabbit, purified by ammonium sulfate precipitation followed by DEAE-cellulose column chro-matography, and conjugated with horseradish peroxidase by a conventional procedure (Harlow and Lane, 1988). Antibody to chicken blood fibronectin was similarly prepared. Anti-vitronectin-Sepharose 4B was prepared by the coupling of 13 mg of anti-chicken blood vitronectin with 1.3 g of CNBr-activated Sepharose 4B (Pharmacia LKB Biotechnology Inc.) according to the manufacturer's manual.
Fractionatwn of Yolk Protein and Purification of Yolk Vitronectin-Chick eggs were divided into egg yolk and egg white. Egg yolk (18-22 g each) was suspended in an equal volume of cold 0.16 M NaCl, 2 mM phenylmethanesulfonyl fluoride, and 10 mM sodium phosphate (pH 7.4) and centrifuged at 12,000 rpm at 4 "C for 20 min (Belitz and Grosch, 1987). The precipitate (yolk granules) was washed twice with the above solution. The supernatant (yolk plasma) was dialyzed extensively at 4 "C against 1 mM sodium phosphate (pH 7.4) containing 5 mM @-mercaptoethanol and centrifuged at 12,000 rpm at 4 "C for 20 min. Yolk plasma was separated into an upper solid layer (low density lipoprotein (LDL)' fraction) and a lower soluble layer (livetin fraction). The LDL fraction was solubilized with 15 ml of 0.16 M NaCl and 10 mM sodium phosphate (pH 7.4).
Yolk vitronectin was purified from the LDL fraction by means of three chromatographic procedures on columns of hydroxylapatite, DEAE-cellulose, and anti-vitronectin-Sepharose at 4 "C as follows. The LDL fraction was applied to a hydroxylapatite column (10-ml bed volume) that had been pre-equilibrated with 0.5 M NaCl and 10 mM sodium phosphate (pH 7.4). The column was washed with 100 ml of 0.5 M NaCl and 10 mM sodium phosphate (pH 7.4) followed by 50 ml of 10 mM sodium phosphate (pH 7.4). Yolk vitronectin was eluted with 200 mM sodium phosphate (pH 7.4) from the hydroxylapatite column. The eluate was diluted with an equal volume of distilled water to decrease the ionic strength in the eluate and applied directly to a DEAE-cellulose column (2-ml bed volume). The column was washed with 20 ml of 0.15 M NaCl, 5 mM 0-mercaptoethanol, and 10 mM sodium phosphate (pH 7.4). Yolk vitronectin was eluted with 0.25 M NaCl, 5 mM 0-mercaptoethanol, and 10 mM sodium phosphate (pH 7.4). The eluate from the DEAE-cellulose column was mixed and incubated with a 2-ml slurry of anti-vitronectin-Sepharose 4B at room temperature for 1 h. The slurry was packed in a column and washed with 20 ml of 0.25 M NaCl, 5 mM P-mercaptoethanol, and 10 mM sodium phosphate (pH 7.4) followed by 10 ml of 0.5 M NaC1, 5 mM P-mercaptoethanol, and 10 mM sodium phosphate (pH 7.4) and then 10 ml of 0.1 M sodium acetate (pH 4.4) containing 0.25 M NaCl and 5 mM p-mercaptoethanol. Yolk vitronectin was eluted with 0.25 M glycine HCI (pH 2.5) containing 0.25 M NaCl and 5 mM B-mercaptoethanol. The pH of the eluate was immediately adjusted to neutrality by adding an appropriate amount of 0.5 M sodium phosphate (pH 7.7). The antibody affinity chromatography described above was repeated several times.
Quuntitatwn of Vitronectin and Fibronectin-Vitronectin and fibronectin in crude preparations were determined by using a sandwich ELISA with their specific antibodies. For the determination of vitronectin, the wells of 96-well microtiter plates were coated with rabbit anti-chicken blood vitronectin (10 pg/ml) in 0.01 M sodium carbonate (pH 9.6) at 37 "C for 30 min, and then the plates were washed four times with phosphate-buffered saline (PBS) followed by 0.025% Tween 20 in PBS. As egg yolk and egg yolk granules contained insoluble materials, a final concentration of 0.01% SDS was added to all samples to solubilize them. This amount of SDS was confirmed not to interfere with the ELISA system. These samples (50 pl each) were incubated on the antibody-coated plates at 37 "C for 1 h. The plates were washed four times with PBS, and then the second antibody, horseradish peroxidase-conjugated anti-vitronectin, in PBS containing 0.025% Tween 20 was added. After seven cycles of washing with PBS, the bound enzyme was measured using o-phenylenediamine and HzOz as substrated. The estimated concentrations of yolk vitronectin were relative values since purified chicken blood vitronectin was used as a standard protein. The relative values seem to reflect a possible difference in the antibody reactivity toward these moleused with rabbit anti-chicken blood fibronectin as the antibody and cules. For the determination of fibronectin, the same procedure was trations of pure blood vitronectin and fibronectin were estimated by means of absorbance measurements at 280 nm with 1-cm path length cells, using absorption coefficients of 1.38 for vitronectin (Dahlback and Podack, 1985) and 1.28 for fibronectin (Mosesson and Umfleet, 1970) at 1 mg/ml.
Determination of Amino-terminal Sequence-A mixture of polypeptides was separated by SDS-polyacrylamide gel electrophoresis (PAGE) using polyacrylamide gels that had been polymerized the day before and pre-electrophoresed with an electrode buffer containing 0.1 mM sodium thioglycolate to scavenge radicals (Moos et al., 1988). Proteins separated on the gels were transferred onto a polyvinylidene difluoride membrane (Bio-Rad). The amino-terminal sequence of each polypeptide on the membrane pieces was determined with a protein sequenator model 477A (Applied Biosystems) (Matsudaira, 1987). Cysteines were not identified.
SDS-PAGE, Western Blotting, Cell Blotting, and Lectin Staining-Proteins were separated by SDS-PAGE according to Laemmli (1970) and stained with Coomassie Blue. For Western blotting, cell blotting, and lectin staining, proteins separated by SDS-PAGE were transferred to nitrocellulose sheets (Schleicher and Schuell) essentially according to Towbin et al. (1979). Protein bands on the nitrocellulose sheets were observed by staining with 0.1% Amido Black 10B, 45% methanol, and 10% acetic acid for 30 s followed by washing with 90% methanol and 2% acetic acid for 30 s.
For Western blotting, the sheet was incubated with 0.2% skim milk in PBS for 30 min and then allowed to react with anti-chicken blood vitronectin antiserum at a 1/1,500 dilution for 1 h. Bound antibody was visualized by means of sequential incubations with horseradish peroxidase-goat antibody against rabbit IgG at a l/2,000 dilution for 1 h followed by 25 pg/ml o-dianisidine and 0.01% H202 for 20 min. Densitometry of the stained bands was performed as described by Kubota et al. (1988).
Cell blotting was performed according to the original report by Hayman et ai. (1982) except that 0.1-0.03 pg of vitronectin/lane was used instead of 30-60 pg. BHK cells were attached to the nitrocellulose sheet at a concentration of 5 X lo6 cells/ml in Grinnell's adhesion medium at 37 "C for 90 min.
Ligand Binding Assay-Collagen binding activity was determined by ELISA. Polystyrene microtiter plates (Sumitomo Bakelite, MS-3496F) were coated with 50 pl of native type I collagen from porcine skin (Cellmatrix I-P, Nitta Gelatin Co., Osaka, Japan) or gelatin at 10 pg/ml in 0.1 M sodium carbonate (pH 9.6) at 37 "C for 1 h. Gelatin was prepared by boiling 0.3 mg of type I collagen/ml in phosphatebuffered saline for 5 min. After being blocked with 0.2% skim milk and 10 mM sodium phosphate (pH 7.4), the wells were washed four times with a washing solution of 0.05% Tween 20 and 10 mM sodium phosphate (pH 7.4). Various concentrations of vitronectin in 50 pl of washing solution were incubated in the wells at 37 "C for 1 h. The plates were washed with the washing solution, and vitronectin bound to immobilized protein in the wells was allowed to react with horseradish peroxidase antibody against chicken blood vitronectin diluted to 1/500 in 0.2% skim milk and 10 mM sodium phosphate (pH 7.4) at 37 "C for 1 h. The wells were washed and incubated with 100 pl of 0.4 mg/ml o-phenylenediamine, 2.5 mM H202, 0.1 M citric acid, and was stopped by adding 50 p1 of 4 N H2S04, and the absorbance at 492 0.2 M Na2HP04 at room temperature for 10 min. Color development nm was measured with a microtiter plate reader, Corona MTP-32 (Corona Electric Co. Ltd., Katsuta, Japan).
Heparin binding activity was examined with a heparin-Sepharose column in the presence of 8 M urea. Pure vitronectin (10 pg) was applied to a heparin-Sepharose column (50-pl bed volume) in 0.13 M NaC1,S M urea, 5 mM EDTA, and 10 mM sodium phosphate (pH 7.7), and then the column was washed with the same solution and eluted with 0.5 M NaCI, 8 M urea, 5 mM EDTA, and 10 mM sodium phosphate (pH 7.7). Each fraction was diluted to 1/4 with 0.025% Tween 20 in phosphate-buffered saline to decrease the concentration of urea, and vitronectin in the fraction was quantitated by a sandwich ELISA.
Cell-spreading Assay-Microtiter plates of 96 wells (Nunc) were coated with increasing concentrations of vitronectin up to 10 pg/ml at 37 "C for 1 h. BHK cell suspension (lo4 cells in 0.1 ml) in a serumfree medium was incubated on the protein-coated wells with or without a synthetic peptide, GRGDSP or GRGESP, at the indicated concentrations. After 90 min at 37 "C, BHK cells were fixed and observed under a microscope. Cell-spreading activity was expressed as the number of spread cells/100 attached cells.

Identification of Vitronectin in Yolk
Fractionation-A rabbit antibody to chicken blood vitronectin specifically reacted with vitronectin from chicken blood plasma in an Ouchterlony double diffusion test, ELISA, and Western immunoblotting (data not shown). Anti-chicken blood fibronectin also reacted with only fibronectin in the same tests. No cross-reaction was observed. A sandwich ELISA using these specific antibodies revealed that yolk plasma contained a fairly high concentration (0.16 mg/ml) of yolk vitronectin and a low concentration (0.03 mg/ml) of fibronectin (Table I). The vitronectin concentration in yolk was almost the same as that in the blood plasma of the same hen. Egg white and egg yolk granules contained essentially no vitronectin or fibronectin. All the vitronectin in yolk plasma was fractionated into an upper solid layer (low density lipoprotein fraction) after extensive dialysis against 1 mM sodium phosphate (pH 7.4) followed by centrifugation at 12,000 rpm at 4 "C for 20 min. The lower soluble layer after the centrifugation, called the livetin fraction, did not contain vitronectin.
Western immunoblotting showed that vitronectin in yolk plasma migrated under reducing conditions as two bands of 54 and 45 kDa (Fig. 1, lane 6), which are 11-25 kDa smaller than those of blood vitronectin (70 and 65 kDa) (Fig. 1, lanes   1 and 4 ) . Blood plasma also contained small amounts of vitronectin migrating at 56 and 45 kDa (Fig. 1, lane 5 ) , which were similar in size to yolk plasma vitronectin, but they were not detected in a pure preparation of blood vitronectin ( Fig.  1, lanes 1 and 4 ) . Under nonreducing conditions, pure blood vitronectin migrated as a single band at 72 kDa ( Fig. 1, lanes  7 and l o ) , the same position to which the major vitronectin band in blood plasma migrated (Fig. 1, lane 1 I), suggesting that a small polypeptide of 5 kDa may be disulfide-bonded to  Lanes I (6 pg), 4 (20 ng), 7 (6 pg), and 10 (20 ng), pure chicken blood vitronectin; lanes 2, 5, 8, and 1 1 , chicken blood plasma (6 pg each); lanes 3, 6, 9, and 12, chick egg yolk plasma (60 pg each). Molecular mass (in kDa) is indicated at the left. the 65-kDa species of blood vitronectin. On the other hand, yolk plasma vitronectin migrated as several bands at positions 54, 68, and 116 kDa (Fig. 1, lane 12), none of which comigrated with blood plasma vitronectin of 72 kDa. These results suggest that the size of whole yolk vitronectin is different from that of blood vitronectin even if the possible existence of small disulfide-bonded polypeptides is taken into account. Yolk vitronectin may contain intramolecular disulfide bond(s) since its migration became slower under nonreducing conditions. Western immunoblotting also revealed a vitronectin aggregate at 180 kDa in blood plasma (Fig. 1, lane  11 ) and at 116 kDa in yolk plasma (Fig. 1, lane 12), suggesting the presence of intermolecular disulfide bond(s). The pattern of yolk vitronectin bands on Western immunoblotting was the same for chicken eggs stored at 4 "C for several days and eggs within 3 h after being laid, even when the yolk plasma from the latter eggs was prepared at 4 "C in the presence of a mixture of protease inhibitors (1,000 units/ml aprotinin, 20 pg/ml leupeptin, 1 mM phenylmethanesulfonyl fluoride, 30 pg/ml soybean trypsin inhibitor, and 20 mM EDTA). Thus, the small molecular size seems to be an intrinsic property of yolk vitronectin and not an artifact resulting from degradation by yolk proteases during preparation.
Purification of Yo& Vitronectin-Yolk vitronectin was present in the solid LDL fraction of yolk plasma after centrifugation. The LDL fraction was solubilized in 0.16 M NaCl and 10 mM sodium phosphate (pH 7.4), and yolk vitronectin was purified from the soluble LDL fraction (Fig. 2, lane 3) by a sequence of three types of column chromatography: hydroxylapatite, DEAE-cellulose, and anti-vitronectin-Sepharose ( Fig. 2; Table 11). Yolk vitronectin was 77-fold enriched in the fraction bound to the hydroxylapatite column (Table 11; Fig. 2, lane 4 ) , and this fraction was further applied to the DEAE-cellulose column. The fraction bound (Fig. 2, lane 5 ) to the DEAE-cellulose column contained most of the vitronectin, giving 410-fold purification (Table 11). The vitronectin preparation after the final step of antibody column chromatography was purified 2,500-fold from yolk plasma (Table 11). In SDS-PAGE, pure yolk vitronectin was detected by staining with Coomassie Blue as two major bands at 54 and 45 kDa with slight contamination by small polypeptides at around 10 kDa (Fig. 2, lane 6). Only the 54-and 45-kDa bands reacted with anti-blood vitronectin on Western blotting at all stages from the starting egg yolk to the purified yolk vitronectin. The contaminating small polypeptides did not react (Fig. 2,  1 2 3 Table I1 and "Experimental Procedures" for details) and subjected to protein composition analysis (lanes 1-6)

FIG. 3. Spreading of BHK cells on vitronectin-coated wells.
BHK cells were incubated a t 37 "C for 90 min on wells precoated with 3.4 pg/ml chick blood vitronectin ( A ) , chick egg yolk vitronectin ( B ) , or bovine serum albumin (C).
lunes 7-22). The relative amount of 45-kDa band/54-kDa band was estimated from the Western blots by densitometry. The 45/54 kDa ratio ranged from 0.3 to 1, depending on the preparations, from crude egg yolk to pure vitronectin. It tended to be high in older or purer preparations, suggesting some conversion to the 45-kDa polypeptide. Cell-spreading Activity of Yolk Vitronectin-BHK cells spread on yolk vitronectin-coated microtiter plates (Fig. 323).
The shape of the spread cells was similar to that on blood vitronectin-coated plates (Fig. 3A). The dose-response curve of the cell-spreading activity was identical for blood and yolk vitronectins, giving a half-maximal concentration of 0.03-0.1 pg/ml (Fig. 4A) (lanes 2 and 3 ) and blood vitronectin (lanes 2 and 4 ) were subjected to SDS-PAGE and transferred from the gel onto nitrocellulose sheets. One sheet was stained with Amido Black 10B for protein (lanes I and 2). Another sheet was examined by cell blotting using BHK cells at 37 "C for 90 min (lanes 3 and 4 ) .

Molecular mass (in kDa) is indicated at the left.
GRGESP, completely inhibited spreading of BHK cells on a yolk vitronectin-coated plate as well as on a blood vitronectincoated plate (Fig. 4B). These results indicate that the cellspreading properties of yolk vitronectin are essentially the same as those of blood vitronectin.
Yolk vitronectin is a mixture of 54-and 45-kDa proteins. We examined the cell attachment activity of each yolk vitronectin band using a so-called "cell-blotting" analysis developed by Hayman et al. (1982). On nitrocellulose sheets, BHK cells were attached to the 54-kDa yolk vitronectin (Fig. 5,  lane 3 ) as well as 70-and 65-kDa bands of chicken blood vitronectin (Fig. 5, lane 4 ) . BHK cells, however, were not attached to 45-kDa yolk vitronectin (Fig. 5, lane 3). Protein staining of the nitrocellulose sheet (Fig. 5, lanes 2 and 2) indicates that the failure of cell attachment was not caused by a lower efficiency of transfer of the 45-kDa band to the nitrocellulose sheet. The contaminating small polypeptides a t around 10 kDa also lacked cell attachment activity (Fig. 5,  lane 3 ) .
Amino-terminal Sequence of Yolk Vitronectin-Yolk vitronectin was separated into 54-and 45-kDa bands by SDS-PAGE under reducing conditions. These bands were transferred onto a polyvinylidene difluoride membrane, and their amino-terminal sequences were determined. We compared the results with those reported previously (Table 111). Only the NH2-terminal sequence of chicken blood vitronectin is available (Nakashima et al., 1992), whereas the whole sequences  * Suzuki et at., 1985;and Jenne and Stanley, 1985. Sat0 et al., 1990. of human, rabbit, and mouse vitronectins are known (Suzuki et al., 1985;Jenne and Stanley, 1985;Sato et al., 1990;. The NHz-terminal sequence of the 54-kDa band is the same as that from chicken blood vitronectin. In contrast, the NHz-terminal sequence of the 45-kDa band is completely different, though it does have some homology with the internal sequences of mammalian vitronectins beginning at the 50th amino acid from the NH2 terminus. These results suggest that the 45-kDa band may be derived from the 54-kDa band by cleavage of the NHz-terminal 49-amino acid peptide. Heparin and Collagen Binding Activity of Yolk Vitronectin- Fig. 6 shows that blood vitronectin binds to native type I collagen and heat-denatured gelatin. In contrast, yolk vitronectin was bound to neither collagen nor gelatin. Neither of the vitronectins bound to bovine serum albumin, which was used as a negative control protein (data not shown).
Heparin binding activity was assayed after treatment of vitronectin with 8 M urea, since 8 M urea appears to prevent nonspecific aggregation of vitronectin and also strongly acti-

FIG. 7. Heparin binding of vitronectin. Blood vitronectin (0)
or yolk vitronectin (0) amounting to 10 pg was applied to a small column of heparin-Sepharose (100-pl bed volume) in 0.13 M NaCl, 8 M urea, 5 mM EDTA, and 10 mM sodium phosphate (pH 7.7). The column was washed with the same solution, and bound proteins were eluted with 0.5 M NaCl in the same solution from the fraction indicated by the arrow. Vitronectin in the fractions was measured in terms of absorbance at 492 nm by means of a sandwich ELISA assay using anti-vitronectin.
vates the heparin binding of pure vitronectin  as well as endogenous vitronectin in blood serum (Yatohgo et al., 1988). Fig. 7 shows that pure blood vitronectin was bound to heparin-Sepharose in 8 M urea, 0.13 M NaCl, 5 mM EDTA, and 10 mM sodium phosphate (pH 7.7) and eluted with 0.5 M NaC1. In contrast, pure yolk vitronectin was not bound to heparin-Sepharose. Further, none of the vitronectin in a crude sample of yolk plasma was bound to heparin-Sepharose in the presence of 8 M urea (data not shown). Thus, the yolk vitronectin molecule seems to lack a heparin-binding site as well as a collagen-binding site.
Carbohydrate of Yolk Vitronectin-Through chemical analysis and examination of the reactivity to several kinds of horseradish peroxidase lectins, we previously showed that chicken blood vitronectin contains both 0and N-linked saccharides with sialic acids (Kitagaki-Ogawa et al., 1990). In agreement with our previous results, chicken blood vitronectin of 70 and 65 kDa reacted with ConA, WGA, A110 A, UEA-I, PHA-L, and PNA, but not with LCA (Fig. 8, lane B). Yolk vitronectins of 54 and 45 kDa were stained similarly to blood vitronectin (Fig. 8 lane Y ) . PNA reacted more strongly with '9'9-6 8 -

FIG. 8. Lectin binding of vitronectin.
Blood vitronectin ( B ) or yolk vitronectin (Y) amounting to 6 pg was subjected to SDS-PAGE and transferred from the gel onto nitrocellulose sheets, and then stained with several kinds of horseradish peroxidase-conjugated lectins as indicated on the top. The amount of horseradish peroxidase lectins was sufficient to strongly stain 6 pg of porcine vitronectin as a positive control. CB, the same gel stained for protein with Coomassie Blue before the transfer. The asialo-type vitronectins (aB, aY) were compared with intact vitronectin for PNA staining. Molecular mass (in kDa) is indicated at the left.
both bands of both blood and yolk vitronectins after treatment with neuraminidase, suggesting the existence of terminal sialic acids in at least some of their 0-linked saccharides. None of the small polypeptides contaminating yolk vitronectin reacted with any of the lectins, suggesting that they have no bound carbohydrates. On the basis of the above lectin reactivities and the previous carbohydrate analysis ( Kitagaki-Ogawa et al., 1990), the 0-linked saccharides of both chicken vitronectins are considered to include a (SAa2-3)GalPl-3GalNAc-Ser/Thr structure, a PNA receptor (where SA is sialic acid). The N-linked saccharides probably have a core structure of Man~1-4GlcNAc~l-4GlcNAc-Asn with many

Man and
GlcNAc residues, as well as a Fucl-2BGall-4pGalNAc sequence.
Quantitatively, ConA, A110 A, and PNA stained yolk vitronectin similarly to blood vitronectin, but WGA stained yolk vitronectin more strongly and UEA-I and PHA-L stained it less strongly than blood vitronectin. These differences suggest that yolk vitronectin contains more NeuNAc, less Fuc, and less terminal Gal than blood vitronectin.
To examine the amount of bound carbohydrates, yolk and blood vitronectins were sequentially deglycosylated sialic acids by neuraminidase, 0-linked asialosaccharides by endoa-N-acetylgalactosaminidase, and finally N-linked saccharides by glycosidase F according to a procedure for sequential deglycosylation of mammalian and avian blood vitronectins (Nakashima et al., 1992). The mass of the two bands of yolk vitronectin decreased during sequential deglycosylation from 54/45 (Fig. 9, lane 1) to 52/43 (Fig. 9, lane 2), 47/38 (Fig. 9,  lane 3), and finally 44/35 kDa (Fig. 9,  lane 4 ) . The high molecular mass band in lanes 3, 4, 7, and 8 is endo-a-Nacetylgalactosaminidase. The difference in mass between the two vitronectin bands did not vary, suggesting that the two bands contain the same amount and the same composition of bound carbohydrates. Thus, yolk vitronectin contained an approximately 10-kDa mass of carbohydrate: a 2-kDa mass of sialic acids, a 5-kDa mass of 0-linked asialosaccharides, and a 3-kDa mass of N-linked saccharides. Similarly, the molecular mass of the major band of blood vitronectin decreased from 70 (Fig. 9, lane 5) to 70 (Fig. 9, lane 6 ) , 65 (Fig.  9, lane 7), and finally 63 kDa (Fig. 9, lane 8) during sequential deglycosylation. Blood vitronectin thus contained an approximately 7-kDa mass of carbohydrate, a less than 1-kDa mass of sialic acids, a 5-kDa mass of 0-linked asialosaccharides, and a 2-kDa mass of N-linked saccharides. These results suggest that yolk vitronectin contains more sialic acids and similar amounts of Nand 0-linked saccharides compared with blood vitronectin.  (lanes 2 and 6), 0-linked asialosaccharides (lanes 3 and  7 ) , and N-linked saccharides (lanes 4 and 8 ) were sequentially removed with neuraminidase, endo-a-N-acetylgalactosaminidase, and glycopeptidase-F (see "Experimental Procedures" for details). Deglycosylation was confirmed by an increment or complete loss of lectin binding activity. The band of high molecular mass in lanes 3, 4, 7, and 8 is endo-a-N-acetylgalactosaminidase. Molecular mass (in kDa) is indicated at the left.

DISCUSSION
This is the first report of the existence and biochemical characterization of a distinct vitronectin molecule in chick egg yolk. The yolk vitronectin is composed of 54-and 45-kDa glycoproteins, incorporating an approximately 10-kDa mass of carbohydrate. Yolk vitronectin has cell-spreading activity but lacks heparin and collagen binding activity.
Since vitronectin was first isolated from human plasma (Hayman et al., 1983;Barnes and Silnutzer, 1983), almost all structural and functional studies of the vitronectin molecule have been concerned with human plasma vitronectin (for reviews, see Preissner (1991) and Tomasini and Mosher (1990)). Human plasma vitronectin separates into two bands of 75 and 65 kDa in SDS-PAGE under reducing conditions, and it has a heparin-binding site toward the COOH terminus, a collagen-binding site possibly toward the NH, terminus, and an RGD-dependent cell-spreading site near the NH2 terminus. It contains Nbut not 0-linked saccharides in an amount of 10% (w/w). Its functions include modulation of the activity of membrane attack complement and hemostatic enzymes as well as promotion of cell spreading. Vitronectins from human placenta (Hayman et al., 1983), HepG2 human hepatoma cells (Barnes and Reing, 1985;Nakashima et al., 1992), human yolk sac carcinoma cells (Cooper and Pera, 1988), and human platelets (Preissner et al., 1989) seem to be similar to plasma vitronectin. Blood plasma vitronectins from 13 other animal species have similar properties, except for some variation in apparent molecular mass, number of bands in SDS-PAGE, and carbohydrate composition (Hayman et al., 1983;Kitagaki-Ogawa et al., 1990;Nakashima et al., 1992). Therefore, the binding activity to heparin and collagen and the cell-spreading activity are considered to be common properties of the vitronectin molecule. Vitronectin-like proteins have recently been reported to exist in a flowering plant, Physarum, brown algae, and a variety of invertebrates (Sanders et al., 1991;Nakashima et al., 1992;Miyazaki et al., 1992;Wagner et al., 1992). Among them, the vitronectin-like proteins from Physarum and brown algae have been examined and have heparin binding activity Wagner et al., 1992). However, we have found in this study that yolk vitronectin, surprisingly, lacks binding activity to heparin and collagen (Figs. 6 and 7). This makes it unique among the vitronectins so far isolated.
The 54-kDa molecule of yolk vitronectin shares the same NHz-terminal sequence as the 70-kDa blood vitronectin molecule. The NHz-terminal sequence of the 45-kDa molecule of yolk vitronectin is possibly homologous with the intramolecular sequence beginning at the 50th amino acid from the NHz terminus (Table 111). These results suggest that the 45-kDa band is an NHz-terminally truncated product of the 54-kDa vitronectin molecule. This interpretation is supported by the fact that 45-kDa yolk vitronectin lacks cell-spreading activity (Fig. 5). The site required for cell-spreading activity in all vitronectins sequenced so far (Suzuki et al., 1985;Jenne and Stanley, 1985;Sato et al., 1990; is the NHz-terminal Arg45-Gly46-Asp47 sequence, which should be located in the missing 49-amino acid segment of 45-kDa yolk vitronectin. Similarity in the carbohydrate compositions of the 54-and 45-kDa bands (Figs. 8 and 9) also supports this interpretation and indicates that the NHz-terminal 49-amino acid peptide of the 54-kDa molecule does not contain any carbohydrate. The difference of 9 kDa between the 54-and 45-kDa molecules is seemingly larger than would be expected for 49 amino acids, which would correspond to roughly a 6-kDa mass. Vitronectin molecules, however, are known to behave abnormally in SDS-PAGE (Nakashima et al., 1992), and therefore the sizes of 54 and 45 kDa were possibly overestimated. Thus, the apparent difference of 9 kDa is probably derived only from the NHz-terminal truncation, not from additional COOH-terminal truncation or from intramolecular deletion in the 54-kDa molecule.
In comparison with the major 70-kDa molecule of blood vitronectin, 54-kDa yolk vitronectin is 16 kDa smaller. Considering the carbohydrate masses of 10 and 7 kDa, the polypeptide portion would be 19 kDa smaller. Human blood vitronectin of 75 kDa cleaves to 10-and 65-kDa polypeptides at Ala380, the two-chain cleavage site. The domain adjacent to this site, spanning 32 amino acids toward the NHz terminus, L~S~*~-A~$~' , is a heparin-binding site. The 54-kDa yolk vitronectin shares the same NHz-terminal sequence as chicken blood vitronectin of 70 kDa (Table 111), suggesting that 54-kDa yolk vitronectin lacks the COOH-terminal 19-kDa polypeptide of the 70-kDa blood vitronectin molecule, which contains the heparin-binding site and the two-chain cleavage site. These interpretations are summarized in a tentative structural model (Fig. lo), which compares yolk vitronectin of 54 and 45 kDa with blood vitronectin of 70 kDa.
Yolk vitronectin is present in the solid LDL fraction at low ionic strength during the purification procedure. Because purified yolk vitronectin becomes soluble in physiological salt solutions, the insolubility seems to be caused by complex formation through ionic bonding with some components in the LDL fraction. This association may depend on the lack of the heparin-binding domain, a highly charged domain.
The collagen-binding site has been suggested to be located in the NHz-terminal half of the human blood vitronectin molecule . The yolk vitronectin 54-kDa molecule seems to span this domain, judging from its molecular mass (Fig. 10). This interpretation seems to be inconsistent with the lack of collagen binding activity in yolk vitronectin, unless there is an inhibitory modification of the yolk vitronectin molecule. Further characterization of the collagen binding property should enable us to elucidate this discrepancy in the future.
Our research on yolk vitronectin was initiated to study the role of vitronectin in the early development of the chick embryo, spurred by the work of Thiery and his colleagues on the role of fibronectin, a similar cell-spreading protein, in early development (Thiery et al., 1985(Thiery et al., , 1989. In egg-laying species, the developing embryo depends completely on the egg components for its physiological and nutritional requirements. Cell adhesion is important during early development, and the abundance of vitronectin but not fibronectin suggests that yolk vitronectin may serve as a main cell adhesion protein in early embryogenesis of the chick. There are other reports on cell adhesion proteins in yolk including 30-and 108-kDa proteins in the newt (Komazaki, 1987) and a 160-kDa protein in the sea urchin (No11 et al., 1985). However, they are not yolk vitronectins. The newt 30-kDa protein was reported to be composed of lipovitellin 2 and phosvitin by Komazaki (1987). Phosvitin from chick egg yolk purchased commercially was examined and found to have cell-spreading activity.' Lysozyme from chick egg white has also been reported to have cell adhesion activity (Satta et al., 1980). Thus, the early development of the chick embryo probably requires the interplay of several kinds of cell adhesion proteins.