Expression of an endogenous galactose-binding lectin in the early chick embryo

Carbohydrate-binding proteins or lectins are widely distributed in animal organisms (Lis & Sharon, 1986). Changes in lectin activity occur during embryogenesis (Harris & Zalik, 1985) as well as in the development of organ systems (Barondes, 1984; Zalik & Milos, 1986). In the early chick embryo a galactose-specific lectin is already present in extracts of preincubated blastoderms (Cook et al. 1979); this lectin has been isolated from primitive streak blastoderms (Zalik et al. 1983). When examined by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) using 10% gels, this lectin separates into two bands. One of these (lectin I) is a polypeptide that migrates with the dye front. The second lectin (lectin II) has a relative molecular mass of 70(±2)×10³, and appears sometimes as a doublet (Zalik et al. 1983).

Cells obtained by mechanical dissociation from different areas of the primitive streak embryos have galactose-bearing receptors at their surfaces, as shown by the agglutinability of these cells by plant lectins (Phillips & Zalik, 1982). From the area opaca of these embryos we have isolated pure populations of endodermal cells. These cells are the progenitors of the extra-embryonic epithelium of the yolk sac and in this paper they will be referred to as extra-embryonic endoderm cells (EEC). In vitro, EEC form aggregates that subsequently undergo cavitation (Milos et al. 1979). The endogenous lectin decreases the adhesion of suspensions of EEC (Milos & Zalik, 1982, 1983). Also during the cellular reorganization that occurs during aggregate cavitation, lectin is released into the vesicular fluid contents of EEC aggregates (Milos & Zalik, 1986). In this paper we describe one of the antisera that we have obtained against preparations of the endogenous blastoderm lectin. From this antiserum we have affinity-purified antibodies specific for lectin I (L₁), and have used these to study the cellular localization of L₁ during gastrulation and early neurulation. Our results indicate that lectin is expressed in cellular populations that are undergoing cell migration and cell spreading. Using affinity-purified antibodies combined with immunoblot transfers of gradient gels, an estimate of the molecular weight of L₁ has been obtained.

Crude lectin extracts were obtained from 23-h incubated eggs (primitive streak blastoderms stages 4–5; Hamburger & Hamilton, 1951), as reported (Zalik et al. 1983). Essentially this procedure consists of homogenization of the blastoderms in a lectin extraction solution MEPBS (0·15 M-NaCl, 0·005 M-NaHPO₄-KH₂PO₄, pH 7·2, 0·004 M-mercaptoethanol) containing 0·3M-lactose and 0·25 mM-phenvlmethylsulphonyl fluoride (PMSF; Sigma) using a TenBroek glass homogenizer. Following centrifugation (100000 g-, 1 h) extracts were dialysed exhaustively against MEPBS and concentrated. The affinity-purified lectin preparations used for immunoblot analysis were isolated using Affi-Gel 10 (BioRad) coupled, according to the manufacturer’s instructions, to p-amino-phenyl-β-D-lactoside (APL) (Vega-Fox Biochemicals). The lectin preparations used for preparation of the antisera were obtained from extracts from approximately 3000 blastoderms and purified by affinity chromatography on APL-Sepharose as reported (Zalik et al. 1983). Lectin activity was assessed using stabilized trypsinized rabbit erythrocytes (Zalik et al. 1983). Protein content was determined according to Bradford (1976) using gamma globulin as a standard.

The procedure used by Nowak et al. (1977) for the preparation of antibodies against the chick pectoral muscle lectin was used with some modifications-. Nine-month-old rabbits were injected intradermally at multiple sites in the back. A total of 131 μg of purified lectin in Freund’s complete adjuvant was given in three approximately equal injections at 2-week intervals over a 6-week period. Two weeks later 3T5pg of lectin in saline was injected intravenously. Since the serum collected 2 weeks after this injection gave no clear precipitin reaction with crude lectin extracts, additional intravenous injections of purified lectin were administered. A second injection (47 μg) was given 2 weeks after the first intravenous injection and the third injection (37 μg) was given three and a half months later. When serum collected 10 days after the last injection was tested in double gel diffusion against crude lectin extracts, it gave rise to two bands, while no bands were present in control serum (Fig. 1A). Agar diffusion was performed in 35 mm Petri dishes using 0·3% agarose (BioRad) in borate-buffered saline according to Garvey et al. (1977), except that 0·3M-lactose was incorporated into the borate-buffered saline prior to addition and solubilization of the agarose. Antibody concentration in the antiserum and in the affinity-purified antibodies was assessed by radial immunodiffusion (Mancini et al. 1965). For this purpose goat anti-rabbit immunoglobulin (Miles) was incorporated into the agar. Standards consisted of rabbit immunoglobulin (Miles). The precipitin ring was enhanced using tannic acid (Simmons, 1971) and the area encompassed by the ring halo was measured with a measuring microscope (Gaertner). The immunoglobulin concentration was calculated by regression analysis (Fig. 1B,C).

Crude lectin extracts we separated by SDS–PAGE using 10% or 10% to 17·5% gradient gels, the latter were stabilized with a 1% to 10% sucrose gradient. A 3% stacking gel was used in both gel types. Crude lectin extracts, purified lectin samples or protein standards (Sigma, B.R.L.) were heated to boiling for 5 min in sample buffer (0·05M-Tris-HC1, pH 6·8, 2% SDS, 7% glycerol 4·3% β-mercaptoethanol, 5 M-urea). The electrophoresis buffer (25 mM-Tris-HC1, 192 mM-glycine, 0·1% SDS) was prepared according to Laemmli (1970). Following electrophoresis gels were transferred to nitrocellulose membranes (BioRad 0·22μm pore size) according to Towbin et al. (1979), using 25 mM-Tris HC1, 192 mM-glycine with or without 20% methanol on a Hoefer Transfer or a Bio-Rad Trans-Blot apparatus. Gels to be transferred without methanol were allowed to swell in the respective transfer buffer for 20—30 min before transfer. Transfer of L, in the presence of methanol was near completion following 2h at 0·3 A. To permit representative transfers of the higher molecular weight proteins, gels were transferred in the absence of methanol for 3h at 1·5–2·0 A. Coomassie Brilliant Blue R-250 (BioRad) staining of vertical gel slices taken before and after transfer, and Amido Black staining of vertical slices of nitrocellulose transfers were used to monitor the efficiency of the transfer. The Amido Black stain consisted of a 5-min stain in 0·1% (w/v) Amido Black in 5% methanol and 10% acetic acid followed by a water destain. Proteins and standards in the transfers were localized by staining representative vertical slices with Amido Black.

To determine the specificity of the antisera and the affinity-purified antibodies, nitrocellulose strips were blocked with 5% bovine serum albumin (BSA; 98–99% albumin, Sigma) in Tris-buffered saline (TBS) containing 0·05 M-Tris-HC1, 0·15 M-NaCl, pH 7-4, for 2h. All probing steps were carried out at room temperature. The blots were then incubated with rabbit antiserum diluted 1:400 in TBS containing 3% BSA, for 14 h followed by four washes, of 20 min each, with TBS containing 0·05% Tween 20 (BioRad). Blots were probed for 2h with goat anti-rabbit immunoglobulin coupled to horseradish peroxidase (GAR-HRP) diluted 1:3000 (BioRad) and washed in TBS as outlined above. Peroxidase activity was detected with 4-chloro-l-naphthol substrate (BioRad), 30 mg dissolved in 10 ml methanol and diluted to 50 ml with TBS and made to 0·01% H₂O₂ (Hawkes et al. 1982). Positive bands gave a blue colour within 1–10 min.

Antibodies specific for lectin I were purified according to Olmsted (1981) and Talian et al. (1983). Crude lectin fractions were separated on 10% SDS–PAGE slab gels (Laemmli, 1970) as described above using a comb containing two end reference wells interspaced by a wide sample well. Gels were transferred to 0·22μm nitrocellulose and L] was initially localized on a representative vertical nitrocellulose strip using the immunoperoxidase staining procedure outlined above. In subsequent experiments L] was localized by staining representative nitrocellulose strips from slab transfers with 0·1% Amido Black. For affinity purification of the antibodies 0·5 cm wide horizontal strips of nitrocellulose containing localized L₁ were cut out, blocked with 5% BSA in TBS (TBS–BSA) for 2h, incubated with the crude antiserum diluted 1:400 in TBS–BSA for 14 h at room temperature and washed four times as outlined for the immunoblotting staining. Antibodies were eluted by washing the nitrocellulose strip with 200 mM-glycine HC1, pH 2·8, for 3–4 min with the aid of a pipette. The strips were then washed with Dulbecco’s phosphate buffered saline (170mM-NaCl, 3mM-KCl, 10mM-Na₂HPO₄, 2mM-KH₂PO₄), with a ratio of 3:1 (v/v) saline/glycine HC1. The pooled glycine-saline mixture was brought to pH 7-0 with 1 M-NaOH within an overall extraction time of 5 min. The extraction was repeated three times and the pooled extracts were concentrated over a ym 10 membrane (Amicon).

A blot-dot procedure (Hawkes et al. 1982) was used to monitor antibody activity following the various elution procedures. Samples of 1–5 μl of the antibody extract or rabbit serum (positive control) were spotted on nitrocellulose sheets marked in square grid patterns. The nitrocellulose was then blocked with BSA and probed with GAR–HRP as described above.

Blastoderms at stages 3–7 (Hamburger & Hamilton, 1951) were fixed at room temperature, either for 2h in freshly prepared 3·7% paraformaldehyde in Pannett & Compton’s (1924) saline (PCS), pH7·4, or for 1 h in absolute ethanol. The embryos were dehydrated with and embedded in polyethylene glycol (PEG) by a modification of the procedure of Drews (1975). Embryos were rinsed briefly in PCS followed by distilled water and dehydrated sequentially in 50%, 75% and 100% aqueous PEG-400 (Sigma) ; blastoderms remained in each PEG solution for 30 min. Blastoderms were then transferred to a 1:1 (v/v) mixture of PEG 400/PEG 1000 at 45°C for 30min, followed by PEG 1000 and by PEG 1500 at 45°C each for 1 h. Embryos were then embedded in PEG 1500 and sectioned shortly after solidification of the PEG. Sections were mounted on slides coated with rubber cement (Lepage’s) thinned with ethyl acetate (Drews, 1975) and kept at 4°C until stained. Staining for immunofluorescence was performed at room temperature. After treatment for 1 min in acetone sections were washed with three changes, 5 min each, of phosphate-buffered saline (PBS: 150mM-NaCl in 5 mM-Na/K phosphate buffer, pH7·1). Sections were incubated for 1h with affinity-purified antibody (8μg m1 ⁻¹ immunoglobulin) or in antiserum diluted 1:400 (15 μg ml ^{− 1} immunoglobulin) both in PBS. Following a PBS wash as outlined above, sections were incubated for 1 h in fluorescein-labelled goat anti-rabbit immunoglobulin (Miles) diluted 1:50 in PBS. After three washes for 5 min each in PBS, slides were mounted in 90% glycerol in PBS with DABCO (Johnson et al. 1982) (Sigma). Controls consisted of sister sections treated with preimmune rabbit serum or with commercial rabbit immunoglobulin (Miles). Sections were observed with a Zeiss Photomicroscope HI with epifluorescence and photographs were taken with Kodak Ektachrome 200 film.

The antiserum obtained following the injection schedule outlined previously, gave rise to two bands when reacted against crude lectin extracts in agar diffusion gels (Fig. 1A). When this antiserum was used to probe nitrocellulose transfers of affinity-purified lectin preparations separated on gradient gels two bands were obtained (Fig. 2a). One band had an approximate M_r of 6·5 × 10³; the second band migrated close to the BSA standard with a M_r of 70× 10³. When this antiserum was tested against crude extracts of chick blastoderms, it reacted with about 10 different bands (Fig. 3c), while no bands were observed when transfers were probed with control preimmune serum. This indicates that the rabbit produced antibodies to additional proteins that were probably present in undetectable trace amounts in the affinity-purified lectin preparations. The relative staining intensities obtained with the rabbit serum probe suggested that L₁ is present in higher concentrations in preparations of affinity-purified lectin (Fig. 2a). The molecular weight of L₁ is close in range to that reported for other galactose-binding lectins of tissues from chick embryos (Bar-ondes, 1984). Thus, it was decided to use it as a probe to affinity-purify antibodies using the methodology described by Olmsted (1981) and Talian el al. (1983). Good antibody recovery was obtained with two successive extractions with glycine HC1; residual antibody could also be eluted by a third extraction (Fig. 2c). Other elution protocols were studied. The inclusion of Tween 20 (Smith & Fisher, 1984) gave reduced resolution due to the fuzziness of the peroxidase colour reaction. Extraction with 2% NH₄OH in 0-5M-NaCl gave results similar to the glycine HC1. Thus three successive glycine-HO extractions were used. The affinity-purified antibodies, reacted with only a single band when tested against crude blastoderm extracts and affinity-purified lectin preparations (Figs 3d,e,i, 2b). This was true for nitrocellulose transfers prepared from 10% gels where L, migrates close to the tracking dye and from gradient gels transferred in the presence or absence of methanol (Fig. 3A,B). When estimated from transfers of gradient gels the average M_r of L₁; is 6·5(±0·3)× 10’³.

Having established the identity of blastoderm protein with which the affinity-purified antibody reacted, we proceeded to determine the localization of L₁ in sections of chick blastoderms undergoing gastrulation and early neurulation. The staining patterns for both area pellucida and area opaca at different developmental stages are described separately.

In the area pellucida (presumptive embryonic region), the localization of the staining varied with the developmental stage of the embryo. At stage 3 during early gastrulation, staining was present in some of the cells of the epiblast anterior and lateral to the nascent immature primitive streak as well as in some cells of this streak. Staining was mainly present in intracellular inclusions (Fig. 4B). At stages 4–5 (definite streak) the fluorescent staining was localized in the lowest portions of the definite primitive groove as well as in the cells emerging from this region and migrating laterally to form the endoderm (Fig. 4C,D). In many of these cells, within the limits of resolution of the technique, staining could be observed at the cell periphery, suggesting an extracellular location. This staining pattern was specially noticeable in the endodermal cells emerging laterally from the streak (Fig. 4D). By stage 7 (one-somite stage), staining was restricted to some endodermal cells (data not shown) and to large round cells located in the mid-anterior region of the embryo, below the endoderm (Fig. 4E). These cells probably correspond to the primordial germ cells migrating away from the germinal crescent (Clawson & Domm, 1969; England, 1982). At this stage no staining was observed in the neural plate, epiblast and mesoderm. No staining was observed in controls using adjacent sister sections reacted with immunoglobulin from non-immunized rabbits (Fig. 4A).

In the area opaca the pattern of immunofluorescent staining with the affinity-purified antibodies persisted throughout the stages examined. At all stages staining was absent from the ectodermal cells, while endodermal cells showed intense staining (Fig. 5A,B,C,D). Here staining appeared to be mainly cytoplasmic and located homogeneously between the numerous yolk platelets. Intense staining was also present in many of the short filopodial extensions of the endodermal cells, as well as in some areas of apposition between these cells and the basal surface of the ectoderm (Fig. 5B); the latter staining pattern was a frequent occurrence in sections of this area. It was not possible to assess the degree of extracellular staining on the endodermal cells because of the intense intracellular staining and the resolution limit of the immunofluorescence technique used. In the anterio-lateral area opaca, staining was also present in some regions of the leading sheet of the mesoderm that was beginning to penetrate between the ectoderm and endoderm; here staining was prominent at the cell periphery, suggesting an extracellular location (Fig. 5A). Also in the lateral region of the area opaca (proximal to the area pellucida) distributed among the endodermal cells, were large round cells that displayed intense intracellular labelling (Fig. 5B,D). These cells probably correspond to the primordial germ cells of the germinal crescent (Clawson & Domm, 1969; England, 1983). In all of the studies reported here these staining patterns in the area pellucida and area opaca were consistent between sister sections stained in different experiments as well as between embryos fixed with ethanol or paraformaldehyde. However, a better preservation of cellular structure was observed in embryos fixed with paraformaldehyde. A total of 13 embryos were examined in these studies.

Owing to the limited supply of affinity-purified antibody, representative sections of different segments of the embryo, rather than the complete serial sections of the embryo, were examined. In all cases no staining was observed when control adjacent sister sections were allowed to react with commercial immunoglobulin from non-immunized rabbits. The same lectin distribution was observed when sister sections were stained with the lectin antiserum, although background staining was noticeable.

Our results using gradient gels indicate that the blastoderm L₁ has a lower relative molecular mass than other galactose-binding lectins from differentiating tissues and organs of the chick embryo and from other animal tissues (Barondes, 1984; Zalik & Milos, 1986). Lectins with small subunits (M_r between 5571 and 6000) have, however, been reported in Vicia faba and Pisuni sativum (Hemperly et al. 1979; Meehan et al. 1982; Quiocho, 1986). We believe L₁ is a natural polypeptide and not a breakdown product, since antibodies specific for L₁ do not react with other proteins on nitrocellulose transfers of crude blastoderm extracts separated on SDS-PAGE gels. Partial breakdown products reacting with L₁ antibodies would be expected as a result of proteolytic activity. In addition, when the same crude extracts were run on native gels and transferred to nitrocellulose, the L, antibodies reacted with one distinct band of high molecular weight (Thomson & Zalik, unpublished data).

The immunofluorescence studies reported here indicate that during early embryogenesis in the chick, L₁ is expressed differentially within the cell populations of the blastoderm. In the area opaca no appreciable amount of lectin is associated with the ectoderm, while the cells of the EEC as well as the primordial germ cells display relatively large concentrations of cytoplasmic lectin that persist throughout the stages studied. In the area pellucida lectin is not detectable in the ectoderm lateral to the streak; however, lectin expression occurs as cells migrate inwards into the streak and emerge laterally to form the endoderm.

The question arises as to the biological significance of this lectin in the developmental events that occur during embryogenesis. Previous studies from this laboratory have shown that cell suspensions prepared from the area pellucida, as well as from EEC, are highly agglutinable with Ricinus communis agglutinin, a lectin specific for β-D-galactose (Phillips & Zalik, 1982). This indicates the glycoconjugates bearing β -galactoside groups are present at the surfaces of these cells. It is conceivable, therefore, that the endogenous blastoderm lectin could interact with these glycoconjugates and affect a surface-mediated function, i.e. cell adhesion. The only cell population in which the effect of the endogenous blastoderm lectin on adhesion has been investigated is the EEC. Here, lectin exerts an inhibitory effect on cell adhesion that can be overcome by galactose-bearing compounds (Milos & Zalik, 1982).

The facts that inhibitory effects of the lectin are transitory (Milos & Zalik, 1983) and that the spontaneous release of lectin by EEC is associated with decreased adhesion (Milos & Zalik, 1982) have led us to suggest that localized release of this molecule could be involved in modulating the transitory adhesions and de-adhesions that occur as cells relocate in the embryo (Zalik & Milos, 1986). In the EEC these rearrangements involve the epibolic spreading of cells to form the yolk sac. Studies from other investigators also indicate that EEC from primitive-streak embryos have a ready ability to spread and separate from each other (Sanders et al. 1978; Bellairs, 1982). It is also known that lectin activity increases significantly during the spreading of the yolk sac (Mbamalu & Zalik, 1987). A role in transitory adhesive bond formation has also been postulated for the galactose-binding lectin of some sponges (Müller & Müller, 1980).

The primitive streak is a structure formed by cells ‘in transit’ (Bellairs, 1986) from the epiblast to form the mesoderm and the embryonic endoderm. Cells in the streak lose their epithelial appearance and assume a mesenchyme-like shape via the process of de-epithelialization (Bellairs, 1986; Sanders, 1986). The present studies show that when cells move into the streak they display the lectin as they change from epithelial-like to fibroblast-like in shape; this lectin becomes extracellular as cells emerge from the streak and migrate laterally to form the endoderm. The endogenous lectin could also modulate cell shape during the formation of transitory adhesions and deadhesions that take place in cell ingression. When EEC are maintained in stationary culture the blastoderm lectin induces the acquisition of a fibroblast-like morphology in the epitheloid EEC (Milos & Zalik, 1981). A similar phenomenon could occur during epiblast cell ingression and emergence from the streak. As the endodermal cells migrate laterally from the streak they do so in the absence of a visible substratum. The blastoderm lectin has a strong tendency to aggregate (Zalik et al. 1983; Thomson & Zalik, unpublished data). Extracellular aggregates of this molecule could provide a transitory substratum over the subgerminal cavity that could be used as scaffold by the spreading endoderm. A role in cell spreading during metastasis has also been suggested for the endogenous galactose-binding lectins present at the surface of some tumour cell lines (Raz et al. 1986). The lectin could also bind to other adhesion molecules expressed at this stage, such as CAMs or cadherins (Edelman, 1985; Hatta & Takeichi, 1986). In the experiments reported here, lectin was found to be expressed transiently in the embryonic endoderm. As mentioned before, owing to the low amounts of affinity-purified antibody available only selected sections of representative regions of the embryo were studied. Detailed studies with serial sections of complete embryos will be needed to determine if the lectin disappears in all of the regions of the embryonic endoderm.

The primordial germ cells also display abundant lectin in their cytoplasm. It is not clear whether these epiblast and subsequently migrate to the endoderm (England, 1983; Eyal-Giladi et al. 1981; Ginsburg & Eyal-Giladi, 1986). Regardless of their site of origin, the primordial germ cells will migrate into the area pellucida, penetrate into the developing blood vessels and gain access to the circulatory svstem. Subsequently, in development the primordial germ cells migrate out of the blood vessels to colonize the gonad (England, 1983). The endogenous lectin in these cells could be associated with the cvtoplasmic inclusions that stain with periodic acid—Schiff (Clawson & Domm, 1969; Fujimoto et al. 1976) and with Ricinits communis agglutinin (Didier et al. 1981). The primordial germ cells could externalize the lectin and use it during their migration to and from the blood vessels to colonize the gonad.

Because of the very high concentration of intracellular lectin present in the EEC and primordial germ cells, it is not possible to ascribe with certainty a cell surface location for this lectin. Studies at the clcctron-micro-scope level are needed to determine the localization of this lectin in the above cells as well as in those of the primitive streak.

This work was supported by the Medical Research Council of Canada and bv the National Science and Engineering Research Council of Canada. We thank Miss Julie Schemas for typing the manuscript.