Mechanisms of cellular adhesion: IV. role of serum glycoproteins in fibroblast spreading on glass

Anchorage to a suitable substrate is a fundamental condition for growth of normal fibroblasts in culture and the response of suspended cells to such a surface provides (Rees, Lloyd & Thom, 1977; Rees et al. 1978) an appropriate system for elucidating the molecular mechanisms by which external perturbations are transduced into specific responses within the cell. After initial contact, cellular protrusions ‘explore’ the surrounding surface (Witkowski & Brighton, 1971, 1972; Albrecht-Buehler, 1976) but in the absence of added serum the subsequent spreading is passive; the characteristic cell shape and ultrastructure do not develop (Witkowski & Brighton, 1971; Pegrum & Maroudas, 1975) and growth does not occur. When active spreading is promoted by serum, arrays of discrete adhesion zones are formed between cell and substrate (Abercrombie, Heaysman & Pegrum, 1971; Revel & Wolken, 1973; Revel, Hoch & Ho, 1974; Abercrombie & Dunn, 1975) and a highly organized cytoskeleton develops in which at least some of the fibrillar protein systems are organized in relation to the focal contacts (Rees et al. 1978; Izzard & Lochner, 1976; Heath & Dunn, 1978). Previous evidence suggested (Lieberman & Ove, 1958; Fisher, Puck & Sato, 1958; Lieberman, Lamy & Ove, 1959; Klebe, 1974; Grinnell, 1976a, Grinnell, Hays & Minter, 1977) that the transition from the rounded shape of suspended cells to the fully spread shape (which is an expression of cytoskeletal development and often distinguishes cell lines) requires specific serum macromolecules adsorbed to the substrate. They need not be present in the solution phase at the same time.

There is some published evidence (Grinnell, 1978; Chiquet, Puri, Turner & Eppenburger, 1977), which we now confirm, that the functional serum components are closely related to the ‘cold insoluble globulin’ (CIG), a well known β-macroglobulin from serum (Morrison, Edsall & Miller, 1948; Mosesson & Umfleet, 1970; Mosher, 1975; Mosesson, Chen & Huseby, 1975). This is antigenically closely related (Ruoslahti & Vaheri, 1975) to a surface glycoprotein from normal fibroblasts and other cell lines, variously named LETS (large external transformation sensitive; Hynes, 1973), CSP (cell surface protein; Yamada & Weston, 1974), galactoprotein a (Gahmberg & Hakomori, 1973) and fibronectin (Keski-Oja, Mosher & Vaheri, 1977) and indeed impure preparations of this cell-derived protein have similar spreading activity (Pearlstein, 1976). This class of glycoproteins aroused widespread interest when it was found that there is a considerable decrease in the pericellular form following malignant transformation (Hynes, 1976; Olden & Yamada, 1977) which can generally be correlated with tumorigenicity (Chen, Gallimore & McDougall, 1976) which, in turn, correlates well with a loss of anchorage dependency and density-dependent inhibition of growth. Readdition of cell-derived material to fibroblasts naturally deficient in it (Yamada, Yamada & Pastan, 1976) has shown that the glycoprotein has no direct influence on growth but the decreased nuclear overlap together with the restoration of normal adhesion and morphology suggest a structural role which involves enhanced development of actomyosin stress fibres (Yamada et al. 1976; Willingham et al. 1977; Ali, Mautner, Lanza & Hynes, 1977).

We now report the isolation of 2 biologically active glycoprotein preparations from chicken plasma which are closely related to CIG/LETS protein. They have been characterized by a variety of biochemical, immunological and physical-chemical techniques. We confirm that prior adsorption of the preparations to glass is sufficient for active spreading of cells in the absence of other serum components. Antibodies against the glycoproteins block this activity and whole chick serum loses activity in this assay when passed through an affinity column carrying such antibodies. We distinguish between the flattening of the cell into a characteristic shape for that cell type and the formation of specialized adhesive zones visualized by interference reflexion microscopy and demonstrate that although pretreatment of glass with CIG is a simple condition for the former response the latter requires the stimulus of additional components in serum.

A cold-insoluble globulin preparation (CIG) was prepared from chicken plasma cryoprecipitate essentially by the solvent extraction procedures of Mosesson & Umfleet (1970). The ethanol-precipitated material was boiled for 10 min in 0–25 M Tris/0·25 M sodium phosphate buffer pH 70 containing 4 4% (w/v) sodium dodecyl sulphate (SDS) and fractionated on a column, 2 · 5 × 95 cm, of Sepharose 4B (Pharmacia, London) equilibrated in 10 mM Tris/1 % (w/v) SDS buffer pH 73.

SDS-polyacrylamide gel electrophoresis was carried out using the method of Fairbanks, Steck & Wallach (1971) at pH 7 · 1. Samples in borate buffer were adjusted to pH 7 · 1 by adding 10 × concentrated electrode buffer. Non-reduced samples were made 2 % (w/v) in SDS and boiled for 3 min; reduced samples were first treated with SDS, then made 1 % (v/v) in 2-mercaptoethanol and boiled for 3 min. Slab gels, 15 × 10 cm, 4% (w/v) acrylamide monomer, were used routinely. They were stained for protein using Coomassie Brilliant Blue (Fairbanks et al. 1971).

Rabbits were injected with 0·5 mg protein in 10 mM Tris/1 % (w/v) SDS in Freund’s complete adjuvent. Booster injections of the same weight of protein were given after 15 and 23 days and the rabbits were bled out at 38 days.

Rabbit antisera were first adsorbed by stirring overnight at +2 °C with heat-denatured (56 °C, 3 min) human fibrinogen, prepared from human plasma cryoprecipitate and purified by repeated glycine precipitation. The fibrinogen was removed by centrifugation and the antibody (IgG) fraction prepared by twice precipitating with sodium sulphate (16 % w/v).

The antibodies (50 mg) were coupled to cyanogen bromide-activated Sepharose 4B (10 ml) by the method of Cuatracasas (1970). For the routine preparation of cold-insoluble globulin preparations, aliquots (25-ml) of chicken plasma were treated for 60 min at room temperature with the protease inhibitor phenylmethylsulphonyl-fluoride (PMSF; Sigma, London) at a concentration of 0 · 2 mM. In one experiment chicken blood was drawn directly into trisodium citrate anticoagulant solution containing PMSF and Aprotinin (Sigma) to give final concentrations of 0 · 38% (w/v) citrate, 0 · 2 mM PMSF and 500 Kallikrein Inactivator Units/ml Aprotinin. When the final preparations were to be used for cell-spreading experiments, PMSF was omitted from all buffers. Neither change from our standard procedure had any effect, either quantitatively or qualitatively, on the final product as judged by electrophoresis.

The plasma was then applied to a column, 1×13 cm, of antibody-coupled Sepharose 4B equilibrated in phosphate-buffered saline (PBS; containing, in g per 1., 8 ·0 NaCl, 0 · 2 KC1, 0 · 2 KH, PO₄, 0 · 3 Na₂HPO₄.12H₁O) pH 7 · 4 containing 0 · 2 mM PMSF, at +4 °C. The column was washed with PBS until the optical density (O.D. 280 nm) returned to background level. Specifically bound protein was then eluted with 50 mM borate/150 mM NaCl buffer pH 10 · 4, containing 0 · 2 mM PMSF. Appropriate fractions were pooled and concentrated using an Amicon ultrafiltration cell (Amicon Corp., Lexington, Mass. U.S.A.).

Pooled preparations were further.fractionated on a column, 1 · 6 × 67 cm of Biogel Ai.jm (Biorad Laboratories, Watford) equilibrated with borate/NaCl buffer containing PMSF at + 4 °C. A routine second chromatography step on this column gave improved separation. About 20 mg of the 225/215000 Dalton preparation and 12 mg of the 140000 Dalton preparation were obtained from 75 ml plasma as determined by Lowry protein assay (Lowry, Rosebrough, Farr & Randall, 1951).

Cells (chick embryo fibroblasts) were seeded at 100000 per glass coverslip (22-mm diameter) and grown for 2 days. After extensive washing with PBS, cells were fixed for 1 h in paraformaldehyde (3·5% w/v in PBS) and then washed first in PBS containing 0 · 1 M NH₄C1 (30 min) and then in PBS alone. The coverslips were drained, covered with 100 μl of antibody solution (rabbit IgG against chicken 140000 Dalton glycoprotein or 225/215000 Dalton preparation at 100 μg/ml or rabbit antiserum against human CIG diluted 1 in 20) and incubated for i h in a wet box. After repeated washing with PBS, the coverslips were incubated under a 1 in 30 dilution of FITC conjugated goat anti-rabbit antiserum (Nordic Diagnostics, London) and finally extensively washed with PBS. Fluorescence was observed using a Leitz Ortholux II microscope fitted with a Ploem vertical illuminator and photographed on Ilford FP4 film.

Sedimentation analyses were performed at 20 °C in a Beckman Model E analytical ultracentrifuge. Observed sedimentation values were corrected to standard conditions in water at 20 °C by the method of Schachman (1957).

Amino acid analyses were carried out after hydrolysis of the glycoproteins for 24 h in refluxing 6 N HC1 in a nitrogen atmosphere. Aliquots were analysed by an automated high performance ion-exchange chromatography system (‘Chromaspeck ‘; Rank Hilger, Margate) as described by Marks & Bailey (1971).

Carbohydrate analyses were carried out using gas-liquid chromatography after methanolysis, N-acetylation and trimethylsilylation essentially as previously published (Chambers & Clamp, 1971; Bhatti, Chambers & Clamp, 1970).

Circular dichroism spectra were recorded on a Cary 61 c.d. spectropolarimeter using 0·5- and o-i-mm path length cells and a 10-s integration period. Sample temperature was maintained using a Haake thermocirculator and a thermostattable cellholder. The observed spectra were fitted by an iterative least-squares curve fitting process using the model spectra of Greenfield & Fasman (1969) and Chen, Yang & Chau (1974).

Infrared spectra were recorded on a Perkin-Elmer 257 grating spectrophotometer. The sample was studied in a Teflon-spaced cell with calcium fluoride windows. The spacing was 25 μm. Sample absorption was matched against the absorption of radiation by pure 50 mM sodium borate/150 mM NaCl buffer in D, O, pD 10 · 4, in a variable path length cell adjusted to give a suitable base line in the infra red spectrum around 1800 cm⁻¹. The air space between samples and detectors was purged with argon to exclude water vapour. Spectra were recorded at slow scan.

The spectra of proteins of known secondary structure were recorded under identical conditions; proteins used were bovine serum albumin, ribonuclease and chymotrypsin. The spectrum of the 140-K glycoprotein corresponded most closely to that of ribonuclease (40% β-sheet).

An established line of rat dermal fibroblasts (16C, Colworth strain) and a commercial preparation of chick embryo fibroblasts (Gibco-Biocult, Paisley, Scotland) were maintained and grown in 200-ml glass medical flat bottles in Dulbecco’s modification of Eagle’s medium (Gibco-Biocult, special formulation), containing 10 % foetal calf serum and in the case of chick embryo fibroblasts, Kanamycin (Flow Laboratories, Irvine, Scotland) 500 μg/ml, in a gas phase of 10 % CO, in air.

Glass coverslips (22 mm diameter) were incubated overnight under sterile growth medium containing Kanamycin and supplemented with the appropriate serum. After washing with serum-free medium, coverslips were transferred to clean plastic Petri dishes. 16C cells were removed from the culture bottle by first washing with Ca²⁺/Mg²⁺-free PBS and then incubating the cells with ethylene-glycol bis (β-amino ethyl ether) N, N’-tetra-acetic acid (EGTA), 0 05 % (w/v) in Ca^,+/Mg²⁺-free PBS, 15 min at room temperature, washing with serum-free medium (3 times) and suspending in supplemented medium containing Kanamycin. Approximately 60000 cells were seeded onto each coverslip which was then incubated at 37 °C. At appropriate intervals unattached cells were washed off the coverslip by gently rolling PBS over the cells, essentially as described by Johnson & Pastan (1972) and counted in a Coulter counter (Coulter Electronics, Dunstable). Attached cells were removed by incubating the coverslip in Tetracaine (Sigma; 4 mM, 20 min) and counted.

Glass coverslips were pretreated as described above, with medium supplemented with either serum or glycoprotein preparation. Bovine serum albumin (Sigma) was used at a concentration of 500 μg/ml. Solutions of the chicken glycoprotein preparations in borate buffer were added to the medium to give final concentrations between 6·25 and 200 μg/ml. In the case of the 225/215000 Dalton preparation, this resulted in the precipitation of the glycoprotein which was found to be only sparingly soluble at neutral pH, particularly when neutralized from high pH. The 140000 Dalton preparation was completely soluble at both high and neutral pH. Protein-coated coverslips to be treated with antibody were incubated with antibodies (IgG) against chicken 140000 Dalton glycoprotein (200μg/ml; 1 h) and thoroughly washed before seeding cells. ‘CIG-depleted ‘chick seium was prepared by 2 passages through an anti-chick CIG antibody column. Cells were dissociated with EGTA, washed and suspended in medium, with or without serum, seeded at 100000 per coverslip and incubated at 37 °C in an atmosphere of 10% CO₂ in air for the appropriate length of time. After washing with PBS, fixing for 1 h with glutaraldehyde (3%, v/v, in PBS) and further washing, coverslips were examined on a Leitz Ortholux II microscope using interference-contrast, interference-reflexion or phase optics, and photographed using Ilford FP4 film.

Glass coverslips (64 × 76 mm) were pretreated in 90-mm plastic Petri dishes. Cells were EGTA dissociated, washed 3 times, and suspended in the appropriate supplemented medium; 1 × 10⁶ cells in a volume of 7 ·5 ml supplemented medium were placed on each coverslip, and 35 μl U-¹⁴C protein hydrolysate (The Radiochemical Centre, Amersham; code CFB 25), 50 μCi/ml, added to each dish. After incubation in an atmosphere of 10% CO₂ in air for an appropriate length of time, the medium was removed, the coverslip washed with Ca*⁺/Mg²⁺-free PBS and transferred to a clean dish. Cells were removed from the coverslip by EGTA treatment and an aliquot counted using a haemocytometer. A further aliquot was made 5 % (w/v) in trichloroacetic acid (TCA) and left overnight at +2 °C. Precipitated material was collected by centrifugation (2000 g, 10 min), washed with 5% TCA twice, and dissolved by boiling for 5 min in 0 · 2 N NaOH. Radioactivity was determined by counting duplicate samples in a Philips liquid-scintillation analyser using NE 260 scintillant (Nuclear Enterprises, Edinburgh).

Initially a crude CIG preparation was isolated from the cryoprecipitate from chicken plasma essentially as described for human CIG by Mosesson & Umfleet (1970). Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) showed 2 major fractions (Fig. 1) a closely spaced doublet with apparent molecular weights of 225000 and 215000 Daltons and a protein band of 140000 Daltons, together with minor contaminants amounting to < 10% of the total protein. Our preliminary studies indicated that the major components could not be separated by DEAE-cellulose chromatography (Mosesson & Umfleet, 1970) and therefore they were fractionated on a column of Sepharose 4B in the presence of SDS. As shown (Fig. 1) by reanalysis of appropriate concentrated fractions by SDS-PAGE, this gave purified preparations of (i) the protein mixture giving the doublet at 225000 and 215000 Daltons and (ii) the protein of molecular weight 140000 Daltons. They will be referred to as the 225/215-K preparation and the 140-K glycoprotein respectively.

Reduction with 2-mercaptoethanol produced at least 5 protein bands from the doublet (Fig. 1). Their intensity varied with individual preparations but the major polypeptides were 155000 and 100000 Daltons (by SDS-PAGE) and the most conspicuous minor ones 130000, 70000 and 30000 Daltons with some material consistently remaining at 225000 Daltons. Similar reduction of the 140000 Dalton protein consistently produced 2 polypeptides, one at 65 000 Daltons and one running close to the marker dye with an apparent molecular weight of 30000 Daltons.

Routine isolation of the macromolecules from chick serum and plasma in biologic ally active form was achieved using affinity columns of Sepharose (4B) beads carrying antibody (IgG) fractions from antisera against either of the purified SDS-denatured preparations. The serine protease inhibitor, phenylmethylsulphonylfluoride, was routinely added to plasma and buffers of all preparations for biochemical analyses and we also demonstrated that carefully drawing the blood directly into citrate containing this inhibitor and aprotinin (see experimental procedures) did not alter the final products. A typical elution profile is shown in Fig. 2A. SDS-PAGE analysis confirmed (Fig. 3) that the unretarded fraction (‘CIG-depleted serum’) contained the other serum proteins. Specifically bound material was eluted with 50 mM borate/ 0 · 15 M NaCl buffer, pH 10·4 and contained only a mixture of the glycoproteins used to raise the antisera. They were then fractionated by gel chromatography (Biogel A 1.5; Fig. 2B).

SDS-PAGE (Fig. 3) confirmed the purity of the 140000 Dalton material prepared in this way, consistent with the sedimentation velocity analysis (Fig. 4) which showed predominantly a single symmetrical peak ⁠. Reduction with 2-mercapto-ethanol gave the expected pattern (c.f. Figs. 1, 3), namely 2 bands with molecular weights of 65000 and 30000 Daltons. When prepared from the affinity column the 225/215-K preparation was invariably contaminated (by SDS-PAGE; Fig. 3) with residual traces of material of molecular weight 140000 and its sedimentation behaviour was more complicated. A double peak (S_{20 w} = 14 · 5 and 18 · 5), was observed which may correspond to the 225/215-K preparation (14 · 5 s) and an aggregated form containing the 140-K glycoprotein as well (18 · 5 s), but residual contamination with faster sedimenting material and material with behaviour similar to the purified 140000 Dalton preparation was also observed. Reduction produced the 5–6 prominent lower-molecular-weight bands obtained previously from the doublet (Figs. 1, 3) plus contaminating bands from 140000 Dalton material.

Several types of immunological test confirmed the close relationship of the 225/ 215-K and 140-K glycoproteins to each other and to CIG/LETS protein, (a) SDS-PAGE (not shown) of material eluted with strong denaturing buffers (containing 6 M urea or 2% SDS) from Sepharose beads carrying antibodies against either preparation contained both 225/215-K and 140-K material and this was confirmed by our preparative affinity chromatography. (6) In immunodiffusion tests (Fig. 5 A) antisera to either preparation gave precipitin lines against both preparations. The distinct spurs and markedly stronger reaction of each antiserum for its homologous antigen suggested that each determinant contained distinctive backbone features. In contrast the definite arc formed against an antiserum to native human CIG emphasized the close relationship of the native serum proteins, (c) Indirect immunofluorescence using antibodies against either preparation showed the fibrillar distribution characteristic of LETS protein on chick embryo fibroblasts (Fig. 5).

The amino acid compositions of the 2 preparations (Table 1) show no unusual concentration of either apolar or hydrophilic amino acids and are distinguished only by a striking similarity to each other and to previously reported CIG/LETS preparations (Mosher, 1975; Mosesson et al. 1975; Carter & Hakomori, 1977). Carbohydrate analyses confirmed the glycoprotein nature of the preparations (Table 2) and although the carbohydrate content is low (< 5 %) it shows more distinctive features. In contrast to human CIG (Mosesson et al. 1975) but consistent with the high levels of asialoglycoproteins in avian circulation (Lunney & Ashwell, 1976), we were unable to detect N-acetylneuraminic acid in either preparation. The absence of N-acetyl-galactosamine, the residue characteristic of carbohydrate chains O-glycosidically linked to serine/threonine, indicates that the side chains are likely to be N-glycosidically linked and the relatively high proportion of mannose relative to galactose suggests that both ‘high mannose’ and ‘complex’ side chains (Robbins, Hubbard, Turco & Wirth, 1977) may be present.

The circular dichroism (c.d.) spectra of chick CIG preparations are shown in Fig. 6. Both have 2 distinct bands closely analogous in shape, although not in magnitude, to those of human preparations (Mosesson et al. 1975); a negative trough which is centred at 217 · 5 nm and a strong positive contribution below 200 nm, beyond the accessible range of the spectrometer. Neither show the small positive contribution which is seen in the spectrum of human material around 230 nm. Matching the observed spectra by an iterative computer method using linear combinations of model spectra derived from poly-L-lysine (Greenfield & Fasman, 1969) or by linear regression from the c.d. of proteins of known secondary structures (Chen et al. 1974) provided similar information about the secondary structure of both glycoproteins, (a) The strong positive band at low wavelength has no corresponding contribution in the usual α-helix, β-sheet and disordered model spectra and the observed c.d. cannot be matched below 215 nm. (b) Curve fitting to 215 nm reveals that any attempt to include a significant contribution from the characteristic negative doublet of the a-helix results in marked deviation of calculated shapes from those observed, (c) For both preparations the optimal overall match has 35-40% β-sheet conformation with the remainder in disordered segments and no α-helix. This agrees with the recent re-interpretation of the c.d. spectrum of human CIG (M. W. Mosesson, personal communication).

The infrared spectrum of the amide vibrational band of the 140000 Dalton glycoprotein in deuterated buffer (Fig. 6) shows a sharp amide I absorption at 1640 cm⁻¹with the amide II band at 1450 cm⁻¹. This gives support to our c.d. conclusions because analysis of well defined globular proteins (Susi, Timasheff & Stevens, 1967; Timasheff, Susi & Stevens, 1967) and protein-gelling systems (A. Clark, personal communication) have shown that the frequency of amide I bands is sensitive to secondary structure with α-helix, β-sheet and disordered conformations providing contributions at 1656, 1632 and 1643 cm⁻¹ respectively. Detailed comparison with proteins of known secondary structure (experimental procedures) confirmed that the 140-K glycoprotein contains little (< 10%) α-helix and again suggests a structure containing ⩽ 40% β-sheet with the remainder having a disordered conformation.

In agreement with observations for other cell lines (Grinnell, 1976a, b) rat dermal fibroblasts (16C line, Colworth strain) in serum-free medium attached but did not spread on bovine serum albumin-treated or plain glass. As shown in Fig. 7 normal spreading (as visualized by interference-contrast optics) to give the characteristic fibroblast morphology only occurred using glass which had been pretreated overnight with 10% (v/v) foetal calf or chick serum in Dulbecco’s medium. The spreading activity on coverslips coated with chick serum was completely inhibited by subsequent treatment with antibodies (200 μg/ml, 1 h, 22 °C) against our purified chick glycoprotein preparations. Overnight preadsorption of coverslips with 140-K glycoprotein at concentrations of 200 μg/ml gave (Fig. 7) complete spreading of fibroblasts 4 h after seeding in serum-free medium and successive dilution showed that full spreading activity persisted down to 25 μg/ml. Similar spreading activity was also observed for the 225/215-K preparation (but see experimental procedures). Partial spreading of the seeded cell population was observed below 25 μg/ml although, in our 4-h assay, few individual cells were captured at intermediate stages in spreading. Successive dilution from this concentration increased the percentage of round cells observed among those fully spread which suggested that fibroblast spreading may be an ‘all or none’ response with a critical initiating stimulus. ‘CIG-depleted’ chick serum prepared by passage (twice) through the affinity column completely loses this ability to promote spreading in serum-free medium when adsorbed to glass.

Detailed comparisons using phase-contrast and interference-reflexion microscopy showed that in complete medium (10% v/v foetal calf or chick serum) 2 major processes can be distinguished in the transition of 16C fibroblasts from the rounded shape in suspension to their fully organized, anchored state. In stage 1 the cells make the recognized transition (Witkowski & Brighton, 1971, 1972) from the attached, minimally spread state after 1 h, through polygonal shapes at 2–3 h, to the fully developed fibroblast shapes after 4 h. Although the area in contact with the glass increased, interference-reflexion microscopy showed (Fig. 8) little development of adhesive organization beyond mottled, grey close contacts (∼ 30 nm separation) (Abercrombie & Dunn, 1975; Izzard & Lochner, 1976). Even at 4 h few cells (< 15 %) showed specialized adhesive zones (∼ 10 nm separation). In stage 2 the organization of the adhesions in the interference-reflexion image developed further (Fig. 8) but without apparent change in the spreading or morphology. By 24 h arrays of dark focal contacts were seen surrounded by white, more distant areas (> 30 nm separation) (Abercrombie & Dunn, 1975; Izzard & Lochner, 1976).

Similar comparisons show (Fig. 9) that for periods up to 4 h with serum-free medium and glass pretreated with 140-K glycoprotein, spreading and organization occurred in a manner which was directly analogous to that observed (stage 1) in complete medium. However, in the absence of additional serum components in the solution phase there was little development beyond this and the cells often began to retract and round up. Certainly the cells never attained the extent of organization of specialized contacts (stage 2) shown by those cultured for 24–48 h in complete medium. Again no evidence was observed of any spreading or organized adhesion in serum-free medium on glass pretreated with CIG-depleted chick serum. Attachment assays (Table 3) confirmed that initial cell behaviour is similar in complete medium, medium containing CIG-depleted serum and serum-free medium on glass coated with 140-K glycoprotein or CIG-depleted serum. Although there was considerable variation from run to run, under all conditions attachment was always significant after 30 min and reached a maximum (> 60% attached) within 2 h. This ‘simple adhesion’ is therefore insensitive to the protein present on the glass or in the medium and under conditions facilitating subsequent spreading it is the nature and quality of the adhesive interaction which changes and not the extent of adhesion.

In contrast to the results in serum-free medium rat fibroblasts allowed to spread on glass pretreated with 140000 Dalton glycoprotein in the presence of medium supplemented with CIG-depleted serum (10% v/v) did show (Fig. 10) highly developed arrays of specialized adhesions comparable with those of cells cultured in complete medium. The adhesion zones (∼ 1 μ m² in area) occurred mainly at the leading edge of cells or lined up along the cell underside, almost certainly along linear tracts of actin containing stress fibres (Rees et al. 1978; Izzard & Lochner, 1976; Heath & Dunn, 1978). Our interference-reflexion evidence therefore suggests that components in the depleted serum allowed further cellular development to form focal contacts and the concomitant cytoskeletal organization. Precoating coverslips with CIG-depleted serum did not inhibit the spreading of cells seeded in normal medium (10% v/v foetal calf or chick serum). In addition, and in further contrast to the results in serum-free medium, highly organized adhesions could be observed (Fig. 10) for cells on glass pre-treated with CIG-depleted serum provided that the medium also contained this material. Thus, given the stimulus of the essential components in the depleted serum, rat fibroblasts are perfectly capable of producing all the components necessary for fully organized focal contacts.

Fibroblast-derived LETS protein has no direct effect on cell growth or protein biosynthesis (Yamada et al. 1976; Willingham et al. 1977) and we now show that the serum-derived glycoprotein is not essential for the initiation of these cellular activities; cells multiply and incorporate radiolabelled amino acids whether 140-K glycoprotein is present or absent (Fig. 11). Growth and incorporation of labelled amino acids occurs at similar rates in foetal calf serum and chick serum. In medium containing CIG-depleted serum fibroblasts grow at their normal rates when glass is precoated with 140-K glycoprotein although their metabolic activity, as reflected by incorporation of labelled amino acids, is considerably reduced (Fig. 11). When the substrate is coated with CIG-depleted serum cells do show distinct evidence of a lag in growth and incorporation of radioactivity up to 24 h which suggested that appropriate adhesion may be essential for optimum growth. However, detailed phasecontrast and interference-reflexion microscopy studies showed no conclusive evidence of a corresponding change in either the pattern or the time course of spreading when compared with cells cultured in complete medium. After 24 h cell growth and incorporation of labelled amino acids were at rates comparable with those of fibroblasts on glass coated with 140000 Dalton glycoprotein.

CIG from other sources has been shown (Mosher, 1975; Mosesson et al. 1975; Chen et al. 1977; Jilek & Hermann, 1977) to be a protease-sensitive, disulphide-bridged dimer of molecular weight 400-450000 Daltons (S₂o_>w = 12-14’5). The initial proteolysis product is characteristically a doublet with molecular weights of 200-250000 Daltons and our SDS-PAGE evidence therefore suggests that chick 225/215000 Dalton preparation is such a product of proteolytic cleavage. However, it seems unlikely that this proteolysis could have occurred during collection and processing of chick blood. Drawing blood directly into citrate containing proteolytic inhibitors and including inhibitors in all buffers did not alter final products quantitatively or qualitatively. We conclude that the degradation probably occurred in the natural avian circulation, presumably as a result of high protease levels. Affinity chromatography using insolubilized gelatin provides similar evidence for chicken plasma (E. Ruoslahti, personal communication) but the fact that successful preparations of native 450-K dimer can sometimes be obtained also indicates variability between individual birds. As yet we cannot determine whether our 140-K glycoprotein is also a proteolytically derived fragment of CIG, although plasmin digestion of human CIG (Chen et al. 1977; Jilek & Hormann, 1977) produces little material in this molecular weight range.

Our 225/215-K preparation and 140-K glycoprotein are remarkably similar to each other and to previously reported material in amino acid and carbohydrate residue composition and in biological activity. In aqueous solution both have little α-helix and 30-40% β-sheet, with the remaining segments in disordered conformations. This similarity in biological activity, composition and secondary structure is very surprising in view of the differences in molecular weight (by SDS-PAGE) and sedimentation behaviour, and shows that the molecule can retain its function in a variety of forms.

We confirm that our CIG-related preparations are the sole components of chick serum that can facilitate fibroblast spreading when bound to glass since (a) complete serum loses this ability when this material is specifically removed, and (6) serum-treated glass is inactivatedby antibodies to our 225/215-K or 140-K glycoprotein preparations. Similar preparations from foetal calf serum do not bind to cells in suspension (Grinnell, 1978) or competitively inhibit their binding to substrates (Grinnell, 19766) and either of 2 hypotheses satisfactorily explain the initiation of the active spreading process, (a) The molecule becomes ‘functional’ by binding to substrate (Grinnell, 1976 a, b). (b) Isolated molecules bind too weakly and binding is a cooperative phenomenon requiring patches of fixed glycoprotein. Either hypothesis is consistent with our view of fibroblast spreading as a specific recognition event coupled to a response of the whole cell. We favour hypothesis (b) because, (1) our analysis of the concentration-dependent spreading activity of the 140-K glycoprotein indicated that fibroblast spreading is an all or none response with a critical threshold for its initiation. (2) In a closely related system the threshold for stimulation of the cellular cytoskeleton can be correlated with the cooperative binding of an exogenous agglutinin (Rees et al. 1978; D. Thom, R. Safford, D. S. Williams & D. A. Rees unpublished results).

A summary of the observed behaviour of fibroblasts seeded onto pre-coated glass in different supplemented media is given in Table 4. Interference reflexion microscopy distinguishes 2 distinct phases in fibroblast spreading: (a) an active spreading like cell growth, appears to be a cellular activity which requires additional components in serum. Indirect immunofluorescence staining of isolated focal adhesions using antibodies to CIG has shown that a form of the glycoprotein is present at the specialized adhesions (Rees et al. 1978; Badley et al. 1978; Mautner & Hynes, 1977). Electron micrographs of such preparations and of cells captured in the act of detachment (Rees et al. 1978) provide direct evidence of (a) a distinct external organization of surface macromolecules with ‘side-to-side’ periodicity mediating this interaction with CIG related protein(s), and (Z>) a tightly associated and similarly ordered cytoplasmic plaque associated with actin stress fibres. Since we know that cells synthesize LETS protein we presume that the distribution of related protein(s) in the fully developed focal adhesions is also controlled by cellular activities stimulated by the additional serum components. The formation of ordered adhesions under some circumstances when exogenous CIG-related material is completely absent, further suggests that the cells secrete the necessary LETS glycoprotein(s). An increased ability to secrete endogenous LETS glycoprotein(s) can also explain the active spreading of particular cell lines in the absence of serum (Evans et al. 1956; Rappaport, 1956; Hayashi & Sato, 1976). This secretion need not imply complete biosyntheses in the early stages (Olden & Yamada, 1977; Hynes & Wyke, 1975; Kurkinen, Wartio-vaara & Vaheri, 1978).

We conclude that although adsorption of serum cold-insoluble globulin to glass facilitates fibroblast spreading in serum-free medium, it is not essential when other serum components are present to maintain the cell in a condition in which it can secrete endogenous material. In any case, even in the presence of CIG additional serum components are necessary to develop a fully organized adhesive state.

We thank Miss H. M. Webb and Mts. T. P. Ogden for skilled technical assistance, Mr A. J. Marks and Mr C. Allcock for the amino acid and carbohydrate residue analyses, respectively, and Mrs. L. E. Luddington and Miss C. B. Evans for processing the micrographs. The antiserum to human CIG was kindly provided by Professor M. W. Mosesson, and the antisera to chick 225/215K preparation and 140K glycoprotein were raised by Dr J. Chidlow. We also thank Dr A. Clark, Mr M. A. F. Davis and Dr E. R. Morris for their help with the infra-red spectroscopy, sedimentation analyses and circular dichroism analyses, respectively.

Since preparing this manuscript, we have seen a publication (F. Grinnell (1978). Int. Rev. Cytol., 53, 65-T44) which points out that there is good evidence in the literature that different fibroblast cells differ in their ability to secrete CIG/LETS protein. While our observation that 16C cells require the stimulus of other soluble serum components is also likely to be true for BHK21-Cr3 and BALB 3T3 cells, WI38 and MRC-5 cells can apparently achieve this secretion in serum-free medium. This publication also confirms that preparations of CIG from foetal calf serum do not bind to cells in suspension and it discusses the possibility that the interactions between the cell surface and adsorbed CIG is a cooperative event.