J. Cell Sci. 71, 17-35 (1984) 17 Printed in Great Britain © The Company of Biologists Limited 1984

THE COMPETITIVE EFFECTS OF SERUM ON CELL ADHESION

A. S. G. CURTIS Department of Cell Biology, University of Glasgow, Glasgow GJ2 OTW, U.K. AND J. V. FORRESTER University Department of Ophthalmology, Aberdeen Royal Infirmary, Foresterhill, Aberdeen, U.K.

SUMMARY Binding curves for the adsorption of plasma , alpha-1-antitrypsin, alpha-2- , , and bovine serum to plain and to hydroxylated polystyrene surfaces were measured. These curves were correlated with the adhesion of BHK cells and leucocytes to these adsorbed surfaces in protein-free culture media. Hydroxylated polystyrene adsorbed less of alpha-1-antitrypsin, alpha-2-macroglobulin and albumin than the plain polystyrene. On the other hand the hydroxylated surfaces bound more fibronectin than the plain polystyrene surfaces. Hydroxylated polystyrene surfaces were also more adhesive for both BHK cells and leucocytes than plain polystyrene: a result confirming earlier work. The competition of fibronectin for adsorption to plain polystyrene with alpha-1-antitrypsin, alpha-2-macroglobulin and ceruloplasmin was measured and correlated with effects on cell adhesion. The results suggest that the low adhesiveness of BHK cells and leucocytes on plain polystyrene in sera-containing media is due both to the low binding of fibronectin and to the binding of , alpha-1-antitrypsin and alpha-2-macroglobulin. The relative unimportance of fibronectin in adhesion to these surfaces is shown by the finding that cell attachment will not occur to polystyrene surfaces that have bound high levels of the antiadhesive proteins in the presence of fibronectin, even though attachment will occur in the absence of fibronectin provided that the antiadhesive proteins are lacking.

INTRODUCTION A vast literature has been published on the effects of various sera and sera fractions on cell adhesion. Certain proteins have been identified as being of particular relevance to the adhesion of cells to non-living surfaces and to other cells. Chief amongst these proteins has been fibronectin (Weston & Hendricks, 1972; Hynes, 1976; Yamada & Olden, 1978; Grinnell, 1978), which is present in two forms, one of which circulates in plasma and the other is bound to cell surfaces. Grinnell & Feld (1979) reported that BHK fibroblasts have an absolute requirement for fibronectin for attachment and spreading on polystyrene surfaces, because they would not adhere under conditions in which surface-bound and endogenous fibronectin were totally lacking. Such studies refer primarily to the adhesion of cells to 'tissue culture grade' polystyrene, which has been corona discharge-treated, but adhesion and spreading of cells on the untreated material hardly occur in the presence of serum unless very high levels of fibronectin have been bound (Grinnell & Feld, 1981). Curtis, Forrester, Mclnnes & Lawrie (1983), however, found that BHK cells will adhere to and spread on highly hydroxylated polystyrene in the complete absence of 18 A. S. G. Curtis andjf. V. Forrester fibronectin when the same conditions, as those imposed by Grinnell and Feld, ob- tained. They would not spread, however, on the less highly hydroxylated surfaces provided by 'tissue culture' grade culture dishes. This result argues that fibronectin is not an essential requirement for the adhesion and spreading of fibroblasts or similar cell types. Grinnell & Feld (1981, 1982) and Klebe, Bentley & Schoen (1981) found that the adsorption of fibronectin to polystyrene surfaces is influenced by the wettability of 9uch surfaces; namely, the ratio of their hydrophobicity/hydrophilicity. The first two of these three papers also showed that competition with various other serum proteins occurs when adsorption takes place from a serum medium. Forrester, Lackie & Brown (1983) found that the plasma proteins, alpha-1-antitrypsin and alpha-2- macroglobulin diminished leucocyte adhesion to glass coverslips. Nevertheless, we describe in this paper experiments that show that adsorption of these proteins renders the polystyrene surfaces wettable. Thus the question arises as to whether the dif- ferences in adhesion found in the presence of serum on 'bacteriological grade' and 'tissue culture' grade dishes are due to the differing extents of fibronectin adsorption, influenced perhaps by competition with other plasma proteins, or whether they are due to the differential adsorption of antiadhesive proteins such as alpha-1 -antitrypsin. Finally, the question must be raised again as to the nature of the feature shared by fibronectin and hydroxylated polystyrene, which aids cell spreading and is not found in the plasma antiproteases. This paper addresses these problems. The range of proteins that has been studied has been extended beyond those used earlier (Grinnell & Feld, 1981; Forrester, Lackie & Brown, 1983) to include the other two most common plasma proteins, namely transferrin and ceruloplasmin.

MATERIALS AND METHODS Cells BHK21 C13 cells were obtained from stocks routinely grown in this laboratory, by trypsinization of confluent or near-confluent cultures, using the method of Edwards & Campbell (1971). Poly- morphonuclear leucocytes were obtained from fresh human blood by the technique used by Forres- ter & Wilkinson (1981). The BHK cells were suspended in either Dulbecco's phosphate-buffered saline or Ham's F10 medium plus 3 % (w/v) foetal calf serum and the insulin-transferrin-selenite supplement (ITS) (Collaborative Research, Waltham, Md, U.S.A.) with insulin and transferrin each at S/igml"1 at a cell density of 2 x 105ml~'. Leucocytes were suspended at a density of 1-5 X 10* ml"1 in Hepes-buffered Hanks' balanced salts solution with or without 10% foetal calf serum (purchased from Gibco-Biocult, Paisley, U.K.).

Culture dishes Polystyrene culture dishes, bacteriological and tissue culture grade were purchased from the following makers: Sterilin, Teddington, Middx, U.K., Nunc through Gibco-Biocult, Paisley, U.K., and Corning, Inc through MacFarlane Ribson, Glasgow, U.K. The bacteriological dishes were given the following treatments: chloric acid treatment by adding 3 ml of 70 % perchloric acid and 2 ml saturated aqueous potassium chlorate to the dishes and allowing these to react for 10 min. Control dishes (bacteriological) were stripped of residual mould- release agent by treatment for 30 min with ethanolic sodium hydroxide (5 M in 70 % (v/v) ethanol) Adsorbed proteins and cell adhesion 19 followed by extensive washing with tap water. Subsidiary experiments showed that tap water was no different from distilled water as a wash. Measurement of cell adhesion BHK cells (0-6 X 106 in a final volume of 3 ml) or 1 X 106 leucocytes (in a final volume of 4 ml) were placed in each culture dish, which was then incubated at 37 °C for 15 min (BHK cells) or 30 min (leucocytes). Cell spreading during these incubation periods is inappreciable on control surfaces. Those cells that had not adhered in these culture periods were removed from the dishes by washing with Hepes-buffered Hanks' balanced salts solution three times. The number of adherent cells in each of 10 standard counting areas was then counted using phase-contrast microscopy to detect the cells and a Quantimet 720 image analysing computer to count the cells. The counting areas were 0-0026 cm2. The results are expressed in terms of the number of cells adhering per cm2. Proteins used in binding studies Human transfemn, alpha-1-antitrypsin, ceruloplasmin and bovine serum albumin were pur- chased from Sigma Chemical Co., Poole, U.K. was purchased from Behringwerke, Marburg, West Germany. Alpha-2-macroglobulin was prepared by the method described by Forrester, Wilkinson & Lackie (1983). Fibronectin (bovine) was prepared from bovine serum in this laboratory by affinity chromatography on gelatin-Sepharose columns, using the method described by Engvall & Ruoslahti (1977). The transferrin was converted to the iron- saturated form by reacting it with ferric nitrilotriacetate in stochiometric proportions and the excess and free nitrilotriacetic acid was removed by dialysis. Bovine immunoglobulin G was prepared from bovine serum by column chromatogTaphy on DEAE-cellulose at pH 8-0 taking the fraction eluted by 0-005 M-phosphate buffer. The purity of these proteins waa checked by sodium dodecyl sulphate/polyacrylamide gel electrophoresis (SDS-PAGE) systems, fibronectin degradation products were found at low levels in the fibronectin and very low levels of two other proteins were found in the serum albumin. No impurities could be detected in the alpha-2-macroglobulin, ceruloplasmin or alpha-1-antitrypsin. Radioactive labelling of proteins Transferrin, alpha-2-macroglobulin, ceruloplasmin, albumin and alpha-1-antitrypsin were labelled with I2SI using the iodogen labelling method (Pierce Chemical Co.). Carrier-free sodium iodide was obtained from Amersham International, Amersham, U.K. The labelled proteins were separated using column chromatography on Biogel P10 columns (Biorad, Watford, Herts, U.K.) with 0-1 M-Hepes buffer (pH8-0) as eluant. The labelled proteins were used within 3 weeks of labelling, being stored at — 25 °C until just before use. Binding of proteins to polystyrene surfaces Protein solutions diluted to levels of between 5-500 fig ml"1 with Ham's F10 saline (serum-free) were placed in 3-ml samples in the dishes and allowed to bind by absorption for 30 min at 25 °C. The exact conditions of binding are described separately for each experiment. At the end of the binding period unbound protein was washed off with four washes of Ham's F10 saline. Measurement of protein binding Protein binding was measured by counting the activity of I25I, on a gamma counter (Wilj Interna- tional, Ashford, Kent, England), of known amounts of the protein solutions and then counting the dishes to which radioactive proteins had been bound. In the earlier part of the work the dishes were cut into strips on a hot wire cutter and the pieces of dish were loaded into the counting tubes. Later in the experiments dishes were extracted twice with small volumes of sodium hydroxide (2 M) and the counts extracted were measured. Both methods gave almost identical results. In all those experiments in which the attachment of cells to adsorbed protein surfaces was being measured, the individual dishes in which attachment assays for adhesion had been carried out were then counted to measure the bound protein remaining after cell attachment. No evidence was obtained from subsidiary experiments that the short incubation of the dishes in the presence of cells caused any loss of bound protein into the medium. 20 A. S. G. Curtis andjf. V. Forrester

RESULTS Binding offibronectin to various polystyrene surfaces and cell adhesion thereto We investigated the adhesion of leucocytes and BHK cells to bacteriological grade dishes to which various amounts of fibronectin had been added. The fibronectin was diluted with Ham's F10 saline from the 8M-urea, 0-lM-Hepes buffer (pH7-5)

r 3X104 xO

,c- 1000 - o CD

T3 C •zo .a t5 ID C •o o 500 - c o .Q

- 2X104

0 200 Fibronectin added (j/g)

Fig. 1. Binding of fibronectin to untreated (B) and to chloric acid-treated (C) polystyrene, and adhesion of BHK cells to adsorbed fibronectin. The adhesion of BHK cells to chloric acid-treated surfaces is not shown since this is maximal even in the absence of added fibronectin. The error bars show two standard deviations: for convenience they are shown on one side of each point; this convention is also used in all succeeding figures. Each experimental point for cell adhesion is the mean of 10 separate sampling areas on three or four separate dishes. Each experimental point for protein binding is the mean of counts on the whole area of each of three or four separate dishes. This statement applies to data on all the figures. Adsorbed proteins and cell adhesion 21

Table 1. The binding of fibronectin to polystyrene dishes and cell adhesion Cell adhesion f Dish type BHK cells Leucocytes

Controls: no fibronectin Bacteriological grade 10677(30%) 49292(14%) Chloric acid treated 32485 (22%) 44285 (36%) Tissue culture grade 13435(24%) 44574(14%)

Experimental 12JI bound Fibronectin on bact. grade 292(18%) 17296(25%) 12489(37%) Fibronectin on chloric acid treated 394(24%) 36485(25%) 44423(17%)

Adhesion is expressed as cells bound cm"2. Binding is expressed as disints per min cm"2 per 40/Jg of fibronectin added per dish. Figures in parenthesis are the standard deviations expressed as percentages of the mean. Each measurement is the mean of 10 individual measurements of three dishes for experimental measure- ments; for control results the measurements are the means of at least 10 counts per dish on at least four dishes. The maximum number of leucocytes available for adhesion was 52910 cm"2 and the maximum number of BHK cells 31750cm"2. The labelled fibronectin had a specific activity of 0'3 [tCi mg"1. The measurements of disints refer to the mean of six measurements, one on each dish. The measurements were carried out in serum-free medium. See Fig. 1 for further data. solution in which it had been stored, to 1 M in urea, immediately before addition to the dishes. Adsorption was allowed to continue for 30 min and then excess unbound fibronectin was washed off with Ham's F10 saline, reducing the urea level to very low values. The effect on the adhesion of BHK cells of binding varying amounts to the dishes on cell adhesion is shown in Fig. 1. The data in Fig. 1 established the amount of fibronectin that should be bound in order to increase the adhesion of BHK cells to the dishes appreciably and this level was used in the main experiments thereafter. Table 1 shows that fibronectin on plain polystyrene has the effect of reducing the adhesion of the leucocytes. This result confirms an earlier report by Brown & Lackie (1981). The data presented in Fig. 1 also demonstrate that chloric acid treatment enhances the saturation level and possibly the affinity of polystyrene surfaces for fibronectin. This enhancement of binding by chloric acid treatment appears to be unique for fibronectin amongst the proteins studied in this work. Table 1 and Fig. 1 show that appreciable fibronectin binding occurs on bacterio- logical grade dishes and that this raises the binding of BHK cells. Both results confirm the results of many other groups, for instance Yamada & Olden (1978). Binding of fibronectin to chloric acid-treated surfaces is greater than that seen on bacteriological grade dishes. The binding of fibronectin to these treated dishes does not, however, increase the adhesion of BHK cells, which was of course already very high as a result of the chloric acid treatment (Curtis et al. 1983). In contrast, leucocyte adhesion is 22 A. 5. G. Curtis andj. V. Forrester reduced when fibronectin is bound to an untreated dish surface, but not when it is bound to a chloric acid-treated surface. In a subsequent section we describe the effects that result from competition between other proteins and fibronectin in binding to treated and untreated polystyrene surfaces, and the resulting effects on cell adhesion.

Binding of alpha-1-antitrypsin to various polystyrene surfaces and cell adhesion thereto The data in Table 2 and Fig. 2 show that alpha-1-antitrypsin binds to untreated surfaces at a saturation level of 300 ngcrn"2 while chloric acid treatment of the dishes

r 3xio"

I It It IT - 2-5X104 500 - I \

I I I 450 I I u - 2X104

T3 C s 300 -

- 1-5X104

150 -

I 100 200 AAT added

Fig. 2. The binding of alpha-1-antitrypsin (AAT) to untreated (B) and to chloric acid- treated polystyrene (C) and the adhesion of BHK cells to alpha-1-antitrypsin adsorbed to such surfaces. Note that the adhesion of BHK cells to the protein on untreated surfaces (•-—•) rises at very low levels of adsorbed protein but is reduced at higher levels. (X — X) Adhesion of cells to chloric acid-treated surface. Adsorbed proteins and cell adhesion 23

Table 2. The binding of alpha-1-antitrypsin to polystyrene surfaces and cell adhesion thereto

Cell adhesion

Dish type BHK cells Leucocytes

Controls: no antitrypsin Bact. grade 10677(30%) 49292(14%) Chloric acid-treated 32485 (9%) 34105 (15%)

Experimental 12JI bound Bact. grade +200 fig 3866(1%) 11340(8%) 26666(25%) Bact. grade +100 jig 2662 (3%) 14775 (12%) 32160 (20%) Chloric acid-treated +220 fig 2941(3%) 30811(7%) 34694(10%)

Adhesion is expressed as cells bound cm"2. Binding of protein is expressed disints per min cm"2. Additions of protein, 200 or 100/Jg as stated below. Specific activity of antitrypsin was 4-85 ^Cimg"1. The maximal possible cell adhesion. Leucocytes, 52910cm"2; BHK cells, 31750cm"2. Cell counts, means of 10 areas on each of three dishes. Binding counts, means of three dishes for experimental dishes. Cell counts, means of four or more dishes for controls. For fuller data on the binding of the protein to plain and to hydroxylated polystyrene and BHK cell adhesion see Fig. 2.

Table 3. The binding of alpha-2-macroglobulin to polystyrene and cell adhesion

Cell adhesion ^ BHK cells Dish type Leucocytes Controls: no macroglobulin Bact. grade 10677(41%) 49292(14%) Chloric acid-treated 31485(7%) 44285(5%)

Experimental 12SI bound Bact. grade 2176 (21%) 14664 (43%) 278 (88%) Chloric acid-treated 452(60%) 23758(17%) 45633(17%)

Adhesion is expressed as cells bound cm"2. Binding of protein is expressed as disints per min cm"1 per 100 fig of macroglobulin added per dish. 1251-labelled macroglobulin, sp. act. 19'5 ^Ci mg"1. The maximum number of leucocytes available for adhesion was 52910 cm"2, and the maximum number of BHK cells available was 31750 cm"2. The adhesion assay was in serum-free medium. The cell counts are the means and standard devia- tions (expressed as percentages of the mean) of 10 counts on each of three dishes for the experimental series, and of ten counts on each of ten dishes (control series). The binding counts are the means and standard deviations of counts of three dishes in each group. See Fig. 3 for further data.

reduces the binding to about 70 % of this value. The adhesion of leucocytes is reduced significantly when they attempt to adhere to the alpha-1-antitrypsin surface on untreated polystyrene, but there is no reduction of the adhesion of leucocytes attaching 24 A. S. G. Curtis andj. V. Forrester

- 2x10* 50 -

c o I I ll o 25 - u E O 6

1x10*

200 a-2-M added (/ig dish

Fig. 3. The binding of alpha-2-macroglobulin (a-2-M) to untreated polystyrene surfaces (# •) and the adhesion of BHK cells to the adsorbed protein (O—-O). Note that, as with alpha-1-antitrypsin, very low levels of adsorbed alpha-2-macroglobulin enhance BHK cell adhesion.

Table 4. A. The binding of55Fe-labelled iron-rich transferrin to polystyrene and cell adhesion thereto

Cell adhesion

Dish type 5SFe bound BHK Leucocytes Bact. grade 1033 (5%) 28483 (11%) 46062 (19%) Chloric acid-treated 469 (5%) 27087 (12%) 43895 (30%) Chloric acid-treated No transferrin 32485 (17%) 51562 (40%)

B. The binding of l2sI-labelled apotransferrin and cell adhesion Bact. grade 1152(5%) 27147(16%) 44050(26%) Chloric acid-treated 390 (4%) 26529 (14%) 49147 (27%)

Adhesion of cells is expressed as cells bound cm 2. Binding is expressed as disints per min cm 2 per 300 fig of transferrin added to each dish. This gives approx. 90% saturation. 55 fg 1 55Fe-transferrin ((spp . act. f mg"g). 121255I-labelled apotransferrin (sp. act. 3-0/iCimg"1). Maximal cell binding. Leucocytes, 52910cm"2; BHK cells, 31750cm"2. For other conditions see Table 1. to alpha-1-antitrypsin bound to chloric acid-treated surfaces. The difference is far greater than can be accounted for by the reduced protein binding. The protein when bound to chloric acid-treated surfaces does not reduce the adhesion of BHK cells. Low levels of binding of alpha-1-antitrypsin on plain polystyrene raise the adhesion of BHK cells above the control level found in the absence of the protein but this effect Adsorbed proteins and cell adhesion 25 disappears at higher levels of binding. The difference in the effects of the protein when bound to untreated and to treated surfaces cannot be explained by the difference in binding since this is relatively slight.

Table 5. The binding of serum albumin to polystyrene surfaces and cell adhesion thereto

Cell adhesion

Dish type 125I bound BHK cells Leucocytes

Bact. grade 1874 287 (114%) 19302 (38%) Chloric acid-treated 960 7998 (35%) 52766 (18%)

100jig albumin was added per dish. Adhesion is expressed as cells bound cm 2. Binding is expressed as disints per mincm"2. For controls without absorbed albumin see Tables 1-3. Specific activity of labelled albumin, 3-15/

100 -I

CD _c T3 C 3 o 50 -

CO a a. o

100 200 Protein added (/jgdish)

Fig. 4. The adsorption of ceruloplasmin to untreated (X- -X) and to chloric acid- treated (• •) polystyrene.

CEL71 26 A. S. G. Curtis andj. V. Forrester

Binding of alpha-2-macroglobulin to various polystyrene surfaces and cell adhesion thereto Alpha-2-macroglobulin resembles antitrypsin in its effects on adhesion (see Table 3 and Fig. 3). Its saturation binding to plain hydrophobic polystyrene is about four times greater than to the hydroxylated polystyrene (chloric acid-treated). Alpha-2- macroglobulin reduces leucocyte adhesion when bound to plain untreated polystyrene, to a very marked degree (see Table 3). The level of binding achieved with the addition of 100 fig of protein per dish (3 ml volume) is 52ngcm~2 and this is sufficient to inhibit leucocyte adhesion, but leaves the unhydroxylated surface more adhesive for BHK cells than the control surface (see Table 3). Examination of Fig. 3 shows that, as with alpha-1-antitrypsin, alpha-2-macroglobulin adsorbed to plain polystyrene surfaces enhances BHK adhesion most at very low levels of binding. Alpha-2-macroglobulin has little or no effect on the adhesion of leucocytes to chloric acid-treated surfaces at the rather low binding level of 8• 75 ng cm"2, but reduces BHK adhesion from about 31000 cells cm"2 to about 24000 cells cm"2. Alpha-2- macroglobulin and alpha-1-antitrypsin are alike in that they are most effective in

o CD c

•o

O 2X104 -="

Ceruloplasmin (;tg) 200 Fig. 5. The adhesion of leucocytes (O, A) and BHK cells (X, •) to ceruloplasmin adsorbed to untreated (O, X) or to chloric acid-treated (A, •) polystyrene. The ordinate is calibrated in ng of ceruloplasmin added per dish. See Fig. 4 for details of adsorption of this protein. Adsorbed proteins and cell adhesion 27 reducing adhesion when bound to the hydrophobic untreated surface. They affect leucocyte adhesion to a greater extent than BHK cell adhesion.

Binding of transferrin and of albumin to various polystyrene surfaces and cell ad- hesion thereto Both iron-saturated transferrin and apotransferrin bind more extensively to bac- teriological grade polystyrene than to the chloric acid-treated material (see Table 4). They have little effect on cell adhesion other than raising the adhesion of BHK cells to untreated polystyrene. The results of experiments with bovine serum albumin are shown in Table 5: they indicate that hydroxylated polystyrene surfaces have a reduced binding for albumin when compared with the bacteriological grade dishes. Albumin clearly reduces the adhesion of BHK cells to either type of polystyrene surface even at relatively low bindings, and reduces leucocyte adhesion to bacteriological surfaces, but has no effect when bound to hydroxylated surfaces.

- 3X104

- 2-5x10*

u

CD o

- 2x10*

1-5x104 50 100 150 Fibronectin added per dish (H

Fig. 6. The competitive adsorption of alpha-1-antitrypsin (AAT; 150/ig added) to untreated polystyrene (• •) in the presence of increasing concentrations of fibro- nectin. The effect on BHK cell adhesion (X x) is also shown. Note that fibronectin reduces the binding of alpha-1-antitrypsin. 28 A. S. G. Curtis andj. V. Forrester

The effects of bound ceruloplasmin on cell adhesion Ceruloplasmin binds to a similar level on both chloric acid-treated polystyrene and the untreated material (see Fig. 4). On both types of surface it acts to reduce the adhesion of both leucocytes and BHK cells (see Fig. 5), but the effect with plain polystyrene plates is small presumably because of the low adhesion of these surfaces in the absence of adsorbed protein.

Does fibronectin compete with alpha-1-antitrypsin, alpha-2-macroglobulin or ceruloplasmin for binding with the substrate? The data presented above were obtained from experiments in which serum was absent and in which usually only a single species of protein had been bound to the dish surface. It is pertinent to ask whether the proteins would compete with each other for binding to the treated or untreated polystyrene surfaces. Alpha-1-antitrypsin binds more extensively to hydrophobic than to hydrophilic surfaces and depresses adhesion 400 -T i- 4xio4

300

u 200 - CD C a T3

i .Q 100 - - 1x10"

400 Alpha anti-trypsin added per dish I Fig. 7. The converse of Fig. 6. The adsorption of fibronectin (• •) in the presence of alpha-1-antitrypsin; 80 fig fibronectin was present in the dish during adsorption. Note a slight reduction in the adhesion of BHK cells (X X) to the adsorbed protein on untreated polystyrene. Adsorbed proteins and cell adhesion 29

- 3x10*

- 2X104

4 <3 - 1X10

T 0 50 100 150 Fibronectin added (ftg) Fig. 8. The competitive binding of alpha-2-macroglobulin (100 /ug added) in the presence of fibronectin (O O) to untreated polystyrene. The effect on BHK cell adhesion is also shown (X X).

of leucocytes and/or BHK cells-when so bound (Fig. 2, Table 2). Fibronectin in- creases the adhesion of BHK cells to the surfaces on which it is bound. In consequence we ask whether fibronectin competes with alpha-1-antitrypsin for binding to various polystyrene surfaces and whether cell adhesion thereto is changed according to the relative binding. A constant amount of alpha-1-antitrypsin was used in competition with varying amounts of fibronectin, and the binding of the antitrypsin to a hydrophobic surface and the adhesion of the two cell types were determined: the results are shown in Fig. 6. Clearly, fibronectin reduces alpha-1-antitrypsin binding and the adhesion of BHK cells increases as the fibronectin level rises and alpha-1- antitrypsin decreases. The experiment was repeated in the converse mode, using labelled fibronectin and increasing amounts of antitrypsin (see Fig. 7) with parallel results. The experiment was confined to hydrophobic surfaces since it is only on these that appreciable inhibition of adhesion occurs with bound alpha-1-antitrypsin. A similar pair of experiments were carried out on the competition of alpha-2- macroglobulin with fibronectin (see Figs 8, 9). The results were similar to those with alpha-1-antitrypsin. 30 A. S. G. Curtis andjf. V. Forrester

r 3x10"

-2x104

"D c D u O XI •o c

c 4 o o -1X10 XI

40 60 80 100 ct-2-Macroglobulin added (j/g dish) Fig. 9. The converse of Fig. 8. The adsorption of fibronectin to untreated polystyrene (• •) in the presence of alpha-2-macroglobulin; 80 ng was present in the dish during adsorption. The effect on BHK cell adhesion is also shown (X X).

Fibronectin competes with saturating levels of ceruloplasmin to reduce binding of that protein to about 40% of its saturation level (Figs 10, 11). The adhesion level of BHK cells to such surfaces falls to about a third of the level that would be found on a surface saturated with fibronectin, even though the bound fibronectin level is suf- ficient (see Fig. 1) by itself to maintain an adhesion level of 90 % of maximal values. In other words the depression in adhesion is not solely due to a lack of fibronectin but also to the presence of ceruloplasmin. Clearly, if fibronectin competes to reduce the binding of these antiadhesive proteins to the dishes, cells should adhere well to hydrophobic dishes provided that the proteins could be removed from the serum used to prepare the culture medium. Unfortunately this does not appear to be practical, but it is possible to carry out the experiment in a similar manner. We prepared an artificial serum from albumin, transferrin and immunoglobulin G (for composition see Table 6), and measured cell attachment in this medium with and without the addition' of antitrypsin or macro- . The results (Table 6) show that our prediction is supported and is consistent with the idea that fibronectin reduces the binding of alpha-1-antitrypsin, alpha-2- macroglobulin and ceruloplasmin to a hydrophobic surface and thus maintains its adhesiveness. Adsorbed proteins and cell adhesion 31

-4x10*

-3X104

o I 100-1 i

-2X104 6 •o o c •o c i o .O o. o 3 E -1x10* o soq

0 20 40 Fibronectin added (/jg)

Fig. 10. The competitive binding of ceruloplasmin (200^g per dish), in the presence of fibronectin to untreated polystyrene (•——•) and the effect on BHK cell adhesion (O O).

DISCUSSION The main findings of this paper are as follows. (1) The hydroxyl-rich surfaces formed after chloric acid treatment of polystyrene show reduced binding of those plasma proteins (i.e. alpha-1-antitrypsin, alpha-2- macroglobulin, albumin) that tend to inhibit cell attachment to hydrophobic polystyrene, but increase binding of fibronectin. (2) When fibronectin competes with one of these proteins for adsorption to such surfaces there is a reduction in adsorption of both proteins. Fibronectin binding is greater to the hydroxylated surfaces than to the non-hydroxylated surfaces and this extends to situations in which fibronectin is in competition with a protein that binds 32 A.S.G. Curtis andj. V. Forrester 8OO-1 r-4xio4

-3x10* O I i O

u c O J3

O - 1x104

Ceruloplasmin added (/

Table 6. The binding of cells to proteins adsorbed from synthetic sera

Cells bound cm 2 A \ BHK cells Leucocytes Control, no serum 27989 (43%) 55410 (17%) Albumin, transferrin, IgG 31430 (17%) 45146 (19%) Albumin, transferrin, IgG, 24174 (18%) 24326 (54%) alpha-1-antitrypsin Albumin, transferrin, IgG, 665 (63%) N.D. alpha-2-macroglobulin Albumin, transferrin, IgG, 673 (101%) N.D. alph-1 -antitrypsin, alpha-2-macroglobulin

Bacteriological grade polystyrene. Proteins preadsorbed to dish, no proteins added to culture medium. General conditions as for other experiments. Protein mixture solutions contained 3-5 mgml"' of human serum albumin, 0-35 mgml"1 human transferrin, 017mgml~' immunoglobulin G (IgG) and for experiments with BHK cell 35^gml~' bovine fibronectin. Alpha-1-antitrypsin was at a concentration of 0-15 mg ml"1. Alpha-2-macroglobulin was at a concentration of 0-3 mg ml"1. These concentrations approximate to the levels to be found in a 10% serum medium (see Diem & Lentner (1970) for data on serum composition). Data are the means and standard deviations of 10 measure- ments on three different dishes per treatment. N.D., not determined. Adsorbed proteins and cell adhesion 33 preferentially to a non-hydroxylated surface. The adhesion of BHK cells to such surfaces of two adsorbed proteins is reduced below the level that would be expected for the actual level of fibronectin bound. The effect is more marked on hydrophobic surfaces than on hydrophilic surfaces. (3) The adsorption of alpha-2-macroglobulin or alpha-1-antitrypsin to plain (hydrophobic) polystyrene inhibits leucocyte adhesion but actually increases BHK cell adhesion at low levels of binding. The fall in BHK adhesion at higher levels of binding of these proteins might indicate a conformational change in the adsorbed layer. The results obtained by Grinnell & Feld (1981, 1982) and by Klebe et al. (1981) showed that fibronectin and bovine serum albumin compete in their binding to both bacteriological and tissue culture grade polystyrene as well as to other plastics. Grin- nell & Feld (1982) found that other unidentified proteins from serum also competed in adsorption with fibronectin: we suggest that these proteins may be alpha-1- antitrypsin and alpha-2-macroglobulin. Grinnell & Feld (1982) suggested that fibronectin binding may be aided by albumin adsorption at very low levels of the latter protein. We did not test for this with albumin but found no evidence for such a phenomenon with ceruloplasmin, alpha-2-macroglobulin or alpha-1-antitrypsin. Our results suggest that the failure of cell adhesion, or presumably of cell spreading, on untreated polystyrene surfaces (hydrophobic surfaces) is to be explained, not by an inadequate binding of plasma fibronectin from the serum in the medium, but by the level of binding of the antiadhesive proteins. This interpretation is supported by the experiment with artificial sera reported in Table 6, in which exogenous fibronectin was absent. Addition of these proteins diminished adhesion appreciably. On the other hand the close parallel between declining fibronectin binding in the presence of alpha- 2-macroglobulin and cell adhesion (seen in Figs 7 and 8) argues to the contrary. However, since such very small quantities of alpha-2-macroglobulin (less than 40 ngcm"2) would only occupy a small fraction of the area that might otherwise have been taken up by fibronectin, and since the fibronectin level remains high enough by itself to support a fairly high level of cell adhesion, it seems likely that the effects of the antiadhesive proteins are direct. This conclusion can be reached even more directly by noting the reduction in adhesion once one of the proteins has been adsorbed to a surface (see for example, Figs 2 and 3). We do not know the nature of the binding sites on polystyrene dishes for proteins, but it seems likely that proteins that bind preferentially to hydrophobic surfaces will do so by various hydrophobic interactions whilst those proteins adhering to hydroxylated surfaces may do so by hydrogen bonding. In view of these considera- tions it seems unlikely that the proteins compete for the same sites even though they may compete for space. It is worth noting that the mode of adsorption of proteins to hydrophobic and to hydroxylated surfaces may be vastly different. Binding of a protein to a hydrophobic surface may result in its hydrophilic groupings facing out- wards into the medium whilst binding of a protein to a hydrophilic surface may result in the protein displaying some of its hydrophobic groupings to the medium. Grinnell & Feld (1982) found appreciable adhesion of cells to bacteriological grade 34 A. S. G. Curtis andjf. V. Forrester polystyrene in the absence of fibronectin whereas we did not. The explanation of this difference almost certainly lies in the fact that the BHK cells they used had been exposed to serum components for some hours after trypsinization and before their use in adhesion assays in serum-free media, while in the present work no serum com- ponents were allowed to come in contact with the cells after trypsinization. Subsidiary experiments carried out by us show that even brief exposure of BHK cells to serum media allows sufficient adsorption of components for the cells to become markedly adhesive to plain polystyrene. Various authors (e.g. Van Oss, Gillman & Neumann, 1975) have proposed that wettability of a surface is directly correlated with adhesiveness for cells. It is note- worthy that we observe that adsorption of alpha-1-antitrypsin, alpha-2-macroglobulin or ceruloplasmin to bacteriological dishes renders these surfaces wettable but diminishes cell adhesion. Such observations would seem to destroy the wettability hypothesis, but we remark, in its defence, that the cells may be making local areas of such surfaces unwettable so that they are unable to adhere to the nearby surfaces. One of the questions that we have not resolved in this work is the extent of surface coverage required for an effect on cell adhesion. Grinnell & Feld (1981) suggest that fibronectin coverage must approach 100% for maximal effects on cell adhesion to appear. One important question that needs an answer is whether suboptimal levels of fibronectin (or other protein) adsorption lead to 100% coverage by a more unfolded molecule, or whether the configuration of the molecule is unchanged but appreciable areas of unmodified substrate 'show through' at various places. It is noticeable that both alpha-1-antitrypsin and alpha-2-macroglobulin reduce adhesion at fairly low levels of binding and that adhesion of cells decreases as the amount of protein increases, which suggests that the new surface afforded by the protein is inimical to cell adhesion. In summary, the results suggest that the reduced adhesiveness of bacteriological polystyrene in media containing sera is due, not only to the reduced fibronectin binding to such surfaces, but also to the binding of relatively high levels of anti- adhesive proteins such as albumin, alpha-2-macroglobulin and alpha-1-antitrypsin.

We thank Michael McGrath for excellent technical assistance, and R.G.G., C.K. and R.D. for donations of human blood. The work was supported from the General Funds of the University of Glasgow.

REFERENCES BROWN, A. F. & LACKIE, J. M. (1981). Fibronectin and collagen inhibit cell-substratum adhesion of neutrophil granulocytes. Expl Cell Res. 136, 225-231. CURTIS, A. S. G., FORRESTER, J. V., MCINNES, C. & LAWRIE, F. (1983). Adhesion of cells to polystyrene surf aces. J. Cell Biol. 97, 1500-1506. DIEM, K. & LENTNER, C. (1970). Scientific Tables. Basel: J. R. Geigy Ltd. EDWARDS, J. G. & CAMPBELL, J. A. (1971). The aggregation of trypsinized BHK cells. J. CellSci. 8, 53-72. ENGVALL, E. & RUOSLAHTI, E. (1977). Binding of soluble form of fibroblast surface protein, fibronectin, to collagen. Int.J. Cancer 20, 1-5 FORRESTER, J. V., LACKIE, J. M. & BROWN, A. F. (1983). Neutrophil behaviour in the presence of protease inhibitors. J. Cell Sci. 59, 213-230. Adsorbed proteins and cell adhesion 35 FORRESTER, J. V. & WILKINSON, P. C. (1981). Inhibition of leucocyte locomotion by hyaluronic acid.7. Cell Set. to, 315-331. FORRESTER, J. V., WILKINSON, P. C. & LACKIE, J. M. (1983). Effect of modified macroglobulin on leucocyte locomotion and chemotaxis. Immunology 50, 251-259. GRINNELL, F. (1978). Cellular adhesiveness and extracellular matrix. Int. Rev. Cytol. 53, 65-144. GRINNELL, F. & FELD, M. K. (1979). Initial adhesion of human fibroblasts in serum-free medium: possible role of secreted fibronectin. Cell 17, 117-129. GRINNELL, F. & FELD, M. K. (1981). Adsorption characteristics of plasma fibronectin in relation- ship to biological activity.^, biomed. Mater. Res. 15, 363-381. GRINNELL, F. & FELD, M. K. (1982). Fibronectin adsorption on hydrophilic and hydrophobic surfaces detected by binding and analysed during cell adhesion in serum-containing medium. J. biol. Chem. 257, 4888-4893. HUGHES, R. C, PENA, S. D. J., CLARK, J. & DOURMASHKIN, R. R. (1979). Molecular require- ments for the adhesion and spreading of hamster fibroblasts. Expl Cell Res. 121, 307-314. HYNES, R. O. (1976). Cell surface protein and malignant transformation. Biochim. biophys. Ada 458, 73-107. KLEBE, R. J., BENTLEY, K. L. & SCHOEN, R. L. (1981). Adhesive substrates for fibronectin. J. cell. Physiol. 109, 481-488. VAN OSS, C. J., GILLMAN, C. F. & NEUMANN, A. W. (1975). Phagocytic Engulfment and Cell Adhesiveness as Cellular Surface Phenomena. New York: Dekker. WESTON, J. A. & HENDRICKS, K.I. (1972). Reversible transformation by urea of contact-inhibited fibroblasts. Proc. natn. Acad. Sci. U.SA. 69, 3727-3731. YAMADA, K. M. & OLDEN, K. (1978). Fibronectin: adhesive glycoproteins of cell surface and blood. Nature, Lond. 275, 179-185.

(Received 2 April J984-Accepted 8 May 1984)