The kinetics of spreading of trypsinized FL cells on plastic or glass substrata covalently or passively coated with various proteins to make simplified model extracellular matrices have been measured. Kinetics have also been obtained in the presence and absence of serum and amniotic fluid. Data from such experiments are shown to be sigmoid and have been computer-fitted with great accuracy to 12 mathematical models discussed in the accompanying paper. A log normal distribution function is shown to give the best fit over 18 different types of experiment. For the first time, therefore, such data can be characterized quantitatively and compared. We obtain the simple parameters μ(mean time to spread) and a (standard error of the mean) and show that the cells spread most rapidly in amniotic fluid or on fibronectin ‘carpets’. They also spread rapidly on a fibronectin-fibrinogen complex extracted from placenta. Spreading is slower in serum, in amniotic fluid lacking fibronectin and on type I collagen or cellular microexudate from FL cells or fibroblasts. On concanavalin A, spreading is rapid but the distribution of times to spread (as expressed by cr) is relatively wide.

The basement membrane upon which epithelial cells rest is likely to play an important part in the government of cell behaviour (Grant, Heathcote & Orkin, 1981 ; Hay, 1981). Local tissue architecture is determined by the complex interplay between epithelial cells, their extracellular matrix (Alitalo, Kurkinen, Vaheri, Kreig & Timpl, 1980) and the underlying stroma containing fibroblasts, which migrate within (Schor, Schor & Bazill, 1981), and deposit (Vaheri, Kurkinen, Lehto, Linder & Timpl, 1978) a three-dimensional collagenous extracellular meshwork. Recent experiments have elucidated the behaviour of fibroblasts over short periods, using model substrata prepared by covalent (Aplin & Hughes, 1981a,b) or non-covalent (Butters, Devalia, Aplin & Hughes, 1980) coating of plastic or glass with components of the extracellular matrix such as fibronectin (Butters et al. 1980), or molecules known to interact with the cell surface such as antibodies or lectins (Aplin & Hughes, 19816). In this way the influence of substratum on cell morphology and the cell surface requirements for cell spreading have been determined. The possibility has been raised that adhesion and motility are controlled in part by the interplay between positive and negative extracellular modulators whose relative concentrations may affect the time course of the interaction (Abatangelo, Cortivo, Martelli & Vecchia, 1982; Atherly, Barnhart & Kraemer, 1977).

By chemical cross-linking from such artificial substrata, cellular components closely associated with adhesion plaques - including a glycoprotein of molecular weight approximately 47 000 (Aplin, Hughes, Jaffe & Sharon, 1981) - have been isolated, and antibodies raised against such components inhibit or reverse adhesion (Hughes, Butters & Aplin, 1981). Experiments in two dimensions, however, may not accurately reflect the behaviour of fibroblasts in three-dimensional stroma.

Such a limitation does not apply to epithelial cells. In addition, much less is known about their surface, extracellular matrix and adhesion properties. Some epithelial-type cells have been shown to adhere to substrata containing the basement membrane components type IV collagen and laminin (Terranova, Rohrbach & Martin, 1980; Carlsson, Engvall, Freeman & Ruoslahti, 1981); adhesion may also occur to fibronectin ‘carpets’ (Hughes, Mills & Courtois, 1979). Epithelial cells in the absence of a prepared adhesive substratum have been observed to adhere only slowly, suggesting the requirement for synthesis and secretion of molecules to form a suitable extracellular environment for spreading.

Human amniotic epithelium provides a ready source of pure preparations of epithelial cells as well as extracellular matrix. Cell lines of epithelioid character from amnion are also available and here we characterize qualitatively some of the substratum spreading requirements of one such cell line as well as their spreading behaviour in amniotic fluid, their milieu in vivo. Preliminary results suggest that these properties are similar to those of primary amniotic epithelial cells (S. Campbell, unpublished). We then go on to examine quantitatively the transformation of these cells from rounded to spread out as a function of time and of substratum composition, applying the kinetic analysis described in the accompanying paper. Using numerical parameters obtained by computer-assisted curve-fitting we compare the efficacy of various substratum-associated proteins in the promotion of cell spreading.

Cell culture

FL amnion epithelial cells (Flow) were grown in plastic tissue-culture flasks (Flow) at 37 °C in a humidified atmosphere of 5 % CO2 :95 % air. The growth medium was Eagle’s minimal essential medium (MEM) supplemented with Earle’s salts, 10% foetal calf serum (FCS), 2 mM-glutamine, 100units/ml benzylpenicillin (Glaxo), 100/tg/ml streptomycin sulphate (Glaxo), 2·5 μg/ml amphotericin B deoxycholate (Squibb), and 1 % non-essential amino acids (Flow). pH was maintained by the presence of 25 mM-Hepes. Cells used in adhesion assays were between pass numbers 500 and 510. Murine L cells were cultured under the same conditions.

Proteins

Fibronectin (FN) was prepared from pooled human plasma or amniotic fluid by affinity chromatography on gelatin-Sepharose (Crouch, Balian, Holbrook, Duksin & Bornstein, 1978; Pena, Mills, Hughes & Aplin, 1980). To remove all the fibronectin (AF —FN), fluid (AF) was passed twice through a gelatin-Sepharose affinity column. Placental spreading factor (PSF) is a high molecular weight aggregate derived from placental extracellular matrix which contains fibronectin and fibrinogen. Its preparation and properties are described elsewhere (Aplin & Foden, 1982).

Concanavalin A (ConA) and wheat germ agglutinin (WGA) were from Sigma. Soybean agglutinin (SBA) and peanut agglutinin (PNA) were gifts from Dr Charles Jaffe.

Substrata

All adhesion assays were performed in flat-bottomed multiwell trays of 16 mm diameter tissueculture plastic wells (Flow), either on coated plastic or on coated 13 mm diameter glass coverslips as indicated for each set of data.

Plastic

A total of 200 μl of ligand solution was added to each tissue-culture plastic well, to cover the bottom completely. After incubation at 37 °C for at least 1 h, the ligand solution was removed and washed five times with phosphate-buffered saline (PBS) immediately prior to incubation of the cell suspension.

Glass

(1) Passive physical coating: 150 μl of ligand solution was added as a meniscus to each coverslip and the preparations incubated at 37 °C for at least 1 h. Washing the substrata was as for plastic substrata. (2) Covalent coating: the method of covalently binding ligands to the surface of glass coverslips has been described previously (Aplin & Hughes, 1981a).

Collagen and gelatin

A total of 150 μl of 2·5 mg/ml collagen type I (Sigma) was prepared in 0·1 M-acetic acid. This was diluted to 1500 μl with water and 100 μl of this solution added to each well; 200/11 of 20 mg/ml gelatin solution in water was added to each well of the assay. In each case the solvent was allowed to evaporate in air to dryness and the substratum rinsed with PBS before use.

Substrate-attached material (Culp, Murray & Rollins, 1979)

Sterile 13 mm diameter glass coverslips were placed in multiwell trays and cells suspended in MEM + 10% FCS added to each well, to a final volume of 1·0 ml. The trays were sealed and incubated at 37 °C for 2 days until each coverslip bore a confluent monolayer of cells. To remove the cells, the monolayers were washed with PBS lacking divalent cations (PBS A) and 0·5 ml. of 2mM-EGTA in PBS A was added to each culture. After 30 min incubation at 37 °C, approximately 50% of the cells had detached. The solution was then replaced with fresh EGTA and the preparation incubated overnight at 37 °C to remove all remaining cells. The prepared substrate-attached material was then stored in a sealed tray in sterile PBS at 4°C until use.

Cell suspensions

Cultures in tissue-culture flasks were taken for assays when semi-confluent. The monolayers were washed briefly with PBS A and the cells were dispersed by treatment with 0·01 % trypsin/0·004 % EDTA solution in PBS A for 5 min at 37 °C. An equal volume of MEM + 10 % FCS was then added. The cells were washed twice with MEM alone and resuspended in sufficient MEM to allow 250 μl cell suspension or 500/d cell suspension per assay point.

Cell spreading assays

A total of 250 μl or 500 μl of cell suspension in MEM was added to each previously coated well, the number of wells dependent on the number of assay points required. Where supplements were added, these were present as 20 % (v/v). In control experiments it was shown that the time required for cells to settle and attach from this volume of suspension could be neglected in measurements of the kinetics of cell spreading. In each case, the assays were incubated at 37°C and fixed at specified time points.

For assays performed on any one day, two control experiments were performed. The positive control involved adding a cell suspension in MEM + 20% FCS into a well previously coated with FCS, and the negative control involved addition of a cell suspension in MEM onto untreated substratum. The controls were fixed at the same time as the last time point of the assay. Only where the negative control gave less than 10 % cell spreading at fixation were the assays included in this data set. Less than 10% of all assays performed were discarded. No cell flattening was evident in phase contrast in any assay before 10 min of incubation. Therefore first assay points were taken at least 10 min after the start of incubation. Also, it was observed that after prolonged incubation in culture in the absence of serum, the proportion of spread cells on protein-coated substrata decreased. For this reason, assays were terminated after between 120 and 180 min of incubation. In each assay, 5–6 time points were taken.

At given times, the supernatant medium was removed from the wells, the monolayer rinsed carefully with 0–5 ml PBS and then fixed in 0-·5 ml 0·25 % glutaraldehyde in PBS. The time required for fixation was approximately 1 min.

Assays were scored by counting cells in eight separate microscope fields for each time point so that the total cell count was at least 250 cells per well. By observation using phase-contrast microscopy, any cell possessing a highly refractile cell body and lacking cytoplasmic protrusions was defined as round, and any cell lacking retractility and/or showing cytoplasmic extensions or flattening of the peripheral cytoplasm was defined as spread.

For each time point, the data were used to calculate the mean proportion of cells spread and the standard deviation of this value from the observed proportions in the eight counts.

Occasionally, as a more rapid semi-quantitative alternative, assays were simply scored by inspection as follows: 80–100% spread, 4+; 60–80% spread, 3+; 40–60%, 2+; 20–40% spread, +; 5–20% spread, (+) ; 0–5% spread, −.

Computation

Data were written into data files containing the times, estimate of mean fractions spread and estimate of standard error of the means. Then the data were fitted to equations (1) to (6) of the previous paper with and without a factor λ, i.e. equation (7) was also fitted. The models for equations (1), (2) and (3) involved numerical integration and Simpson’s rule was used with 100 divisions. The denominator for equation (1) was estimated as:
formula
using 100 steps over each of the two intervals −25 to 0 and −525 to −25. An independent program was written to estimate the accuracy of the integrations and the results were within 1 % of the values calculated using 1000 steps for μ, and σ values in the ranges found experimentally.
The regression program has been described elsewhere (Waight, Leff & Bardsley, 1977; Bardsley, Leff, Kavanagh & Waight, 1980) and involved a Monte Carlo search with 1000 steps for equations (1), (2) and (3) but 10000 steps for equations (4), (5) and (6). The best values from this coarse search of allowed parameter space were used as starting estimates for a Simplex routine followed by a Migrad minimization. The estimated covariance matrix was tested for positive definiteness and printed along with the convergence criteria, agreement between calculated and experimental points, weighted and unweighted sums of squares and changes in these and step sizes following each of the fitting routines. In general, the minimization was satisfactorily accomplished in the Monte Carlo search and the Simplex and Migrad routines were not always required. The same data were fitted using several differing starting estimates but in all cases the same minimum was found. The most useful starting parameters are given in the order: name, equation, start value, starting stepsize, lowest allowed value, and highest allowed value:
formula
A program was written that accepted Q1, Q2, m2, m1 and n of equation (9) of the preceding paper, calculated the F-test statistic and recorded the confidence limit for accepting the model as either not significant, 95 % significant or 99 % significant. Best-fit parameters obtained for pairs of data sets were compared in some cases by computing the increase or decrease in a given parameter (Xi, Xj) as expressed by:
formula

Substratum and medium requirements for FL cell spreading

FL amnion cells attach and spread on a variety of protein-coated substrata in the absence of serum in 1–2 h (Table 1). Full adhesion is supported by fibronectin, the fibronectin-derived factor (PSF; Aplin & Foden, 1982), cellular microexudates (SAM) and the lectins ConA and SBA. Type I collagen and gelatin substrates allow about half of the cells to spread. WGA and PNA are less effective promoters of spreading, while cells do not adhere to BSA-coated surfaces. Essentially similar results are obtained in these experiments using glass and plastic supports, and using covalent and non-covalent or passive means of coating the support. Passive coating on plastic or glass is efficient for FN, but possible complications arising from desorption of surface-binding lectins from substrata led to the choice of an irreversible covalent glass-attachment procedure for them.

Table 1.

FL cell adhesion in the absence of serum or amniotic fluid on coated substrata

FL cell adhesion in the absence of serum or amniotic fluid on coated substrata
FL cell adhesion in the absence of serum or amniotic fluid on coated substrata

It is likely that a minimum or ‘threshold’ substratum density of the extracellular moiety may be required before sufficient anchorage is obtained to allow cells to spread (Aplin & Hughes, 19816 ; Hughes, Pena, Clark & Dourmashkin, 1979). The extent of FL cell spreading after 2 h on plastic supports coated with various concentrations of plasma FN is shown in Fig. 1, which supports the threshold hypothesis. In assays of spreading kinetics, it was necessary to use concentrations of coating components that could exceed the threshold; we used 30 μg/ml of fibronectin, 100 μg/ml ConA and other lectins, and 60 μg/ml PSF protein.

Fig. 1.

Extent of spreading of FL cells 2 h after plating onto plastic surfaces coated with various concentrations of human plasma fibronectin. Note that the horizontal axis is logic.

Fig. 1.

Extent of spreading of FL cells 2 h after plating onto plastic surfaces coated with various concentrations of human plasma fibronectin. Note that the horizontal axis is logic.

The cells also attach and spread on plastic or glass substrata in the presence of supernatant medium containing amniotic fluid or serum, as well as on surfaces coated with amniotic fluid (Table 2). Fibronectin is a constituent of amniotic fluid (Crouch etal. 1978) and adsorbs rapidly and avidly to surfaces (Hugheset al. 1979; Klebe, Bentley & Schoen, 1981 ; Grinnell & Feld, 1982). Amniotic fluid from which the fibronectin had been removed was therefore tested for the ability to support FL cell attachment and spreading (Table 2).’ In the presence of fibronectin-depleted fluid added as a medium supplement (20 %), adhesion was inefficient. Surfaces coated with the depleted fluid did not support full cell spreading. These results suggest that fibronectin is the major (though probably not the only) adhesion-promoting component of amniotic fluid.

Table 2.

FL cell spreading in the presence of amniotic fluid and serum: effect of experi-mental design

FL cell spreading in the presence of amniotic fluid and serum: effect of experi-mental design
FL cell spreading in the presence of amniotic fluid and serum: effect of experi-mental design

Kinetics of cell spreading: comparison of mathematical models

The rate of transformation from rounded to spread morphology of cells on a variety of prepared substrata (Table 1) and in the presence of serum or amniotic fluid (Table 2) was measured as described in Materials and Methods. Initially we wished to determine which of the mathematical models described in the accompanying paper would be most suitable in this range of experimental situations. Eighteen data sets were therefore collected and curve-fitted using the six two-parameter models, and the corresponding six three-parameter models, the latter incorporating a Scaling factor to attempt to take into account the small percentage of non-viable cells in each experiment. Fig. 2 shows the six pairs of models used to fit a single typical data set.

Fig. 2.

A typical data set (cells spreading on placental spreading factor) with computergenerated curves using each of the six two- and three-parameter models. The best fit to these data was given by the log normal model, and an improvement significant at 95 % confidence level obtained by adding the third parameter.

In each frame, a continuous line shows the fit obtained using the three-parameter model, while the broken line shows the computer-generated fit for the corresponding two-parameter model. Abbreviations are as in Fig. 3.

Fig. 2.

A typical data set (cells spreading on placental spreading factor) with computergenerated curves using each of the six two- and three-parameter models. The best fit to these data was given by the log normal model, and an improvement significant at 95 % confidence level obtained by adding the third parameter.

In each frame, a continuous line shows the fit obtained using the three-parameter model, while the broken line shows the computer-generated fit for the corresponding two-parameter model. Abbreviations are as in Fig. 3.

Quality-of-fit was compared in two ways. In the first place, sums of squares of weighted residuals (Q) obtained for each model using each of the 18 data sets were added, the smallest total thus indicating the model that gave the best fit across all the data sets (Fig. 3). Secondly, the Q values for each model were placed in order of increasing magnitude for each data set and assigned a corresponding rank : 1–6 for the two parameter models, and 1–6 for the three-parameter models. The positions of each model in the 18 data sets were then added to give the total. Again the greater this number, the less accurate the fit across the range of experiments (Fig. 4). According to both these criteria the log normal model provides the best fit, with the empirical Hill equation close behind.

Fig. 3.

Histogram of Q-sums over 18 data sets for the six two-parameter and six three-parameter models. The smaller the error function Q, the better the fit. In, log normal; h, Hill; tn, truncated normal; sn, symmetrical normal; de, double exponential; me, monoexponential; 2, two-parameter model; 3, three-parameter model.

Fig. 3.

Histogram of Q-sums over 18 data sets for the six two-parameter and six three-parameter models. The smaller the error function Q, the better the fit. In, log normal; h, Hill; tn, truncated normal; sn, symmetrical normal; de, double exponential; me, monoexponential; 2, two-parameter model; 3, three-parameter model.

Fig. 4.

Histogram of ranking-sums for 18 data sets fitted to the six two-parameter models (A) and to the six three-parameter models (B). The smaller the sum, the better the fit. Abbreviations as in Fig. 3.

Fig. 4.

Histogram of ranking-sums for 18 data sets fitted to the six two-parameter models (A) and to the six three-parameter models (B). The smaller the sum, the better the fit. Abbreviations as in Fig. 3.

In addition, the three-parameter models fit the data better than the two-parameter models in each case. The extent of this improvement was assessed using the F-test as described in the accompanying paper. The results are given in Table 3, which shows that 30% of the fits were improved at ⩾95 % confidence limit by introducing the scaling factor λ. The normal and log normal models showed the greatest improvements, while the Hill model was least affected. On the basis of these findings, it was decided to utilize the three-parameter log normal model for quantitative comparison of the kinetics of cell spreading in different experiments.

Table 3.

Improvement in quality-of-fit obtained by introducing a third parameter (scaling factor λ) into each of the six models.

Improvement in quality-of-fit obtained by introducing a third parameter (scaling factor λ) into each of the six models.
Improvement in quality-of-fit obtained by introducing a third parameter (scaling factor λ) into each of the six models.

Effect of substratum composition on cell spreading kinetics

Table 4 shows the μLand μL values obtained using the three-parameter log normal model to fit 21 assays of cell spreading kinetics, along with the values for the scaling parameter and the error in the fits expressed as Q. It must be noted that the values for standard errors in μL and σL. express the precision of the curve-fitting procedure as well as the experimental error involved in obtaining the raw data. In order to assess the latter, we repeated certain experiments exactly, on different days. The results led us to the belief that variations of greater than ±10% in μL or ±25% in σL may be regarded as significant. Some experiments shown in the table represent variations in method of coating or type of surface that were performed to assess the effect of alterations in methodology on observed cell behaviour. Small differences were observed in kinetics of cell spreading on substrata coated with the same material but using different methods; for carpets of either FN or PSF, adhesion was slightly more rapid after covalent attachment to glass than after non-covalent coating onto plastic. This is not a protein density effect, and so may reflect the influence of positive charges introduced onto the glass during coupling (Aplin & Hughes, 1981a). However, control experiments demonstrated that FL cells do not adhere well to glass treated only with the coupling reagent. Differences in rates of spreading were also observed in the presence of complex mixtures such as amniotic fluid or serum added as medium supplements, precoated onto substrata, or both. Thus the results show that valid comparisons of μL and λL in different experiments can be made only when identical methodology is used.

Table 4.

Log normal fit parameters for cell spreading kinetics

Log normal fit parameters for cell spreading kinetics
Log normal fit parameters for cell spreading kinetics

On this basis, several observations can be made about the rate of spreading of FL cells. The cells spread with a mean time of 36–40 min on fibronectin obtained either from plasma or amniotic fluid. Removal of FN from amniotic fluid results in considerable increases in mean spreading time (18% in μL.) and in the distribution of times (25 % in λL.) (Fig. 5) as expected from Table 2. Perhaps surprisingly, adhesion is more rapid in amniotic fluid (μL= 30·6min) than in serum (μL = 55’1 min) despite the fibronectin content of the latter. This observation, together with the finding that some cell spreading occurs in amniotic fluid lacking FN (μL = 139 min; Fig. 5) suggest that other components present in the fluid may influence cell adhesion. Slow spreading in serum may simply reflect competition between FN and other protein components for adsorption sites on the substratum (Grinnell & Feld, 1982) although other serum components are known to influence cell morphology actively (Knox & Griffith, 1982).

Fig. 5.

Kinetics of cell spreading on carpets made from amniotic fluid fibronectin (+), and in the presence of complete amniotic fluid (○) and amniotic fluid lacking fibronectin▫). Data points (± S.E.) with computer three-parameter log normal curves.

Fig. 5.

Kinetics of cell spreading on carpets made from amniotic fluid fibronectin (+), and in the presence of complete amniotic fluid (○) and amniotic fluid lacking fibronectin▫). Data points (± S.E.) with computer three-parameter log normal curves.

Providing the PSF concentration remains in excess of the threshold value for promotion of spreading, dilution of the coating solution by more than 10-fold leaves the kinetics of spreading (μL or λL) essentially unaffected. We have confirmed this observation using other cell lines spreading on carpets of FN (J. D. Aplin, unpublished). Only at or below the threshold do the kinetics begin to vary. These findings have implications for the mechanism of spreading, which will be developed elsewhere.

Mean times for spreading on FN, PSF and ConA-coated glass are similar ; however, the distribution is markedly broader for ConA (μL = 1·035) than for FN (μL = 0·675) or PSF (λL = 0·414) (Fig. 6). This may result from a requirement to sort out ‘relevant’ ConA-binding components in the cell membrane (Aplin & Hughes, 19816), the initial orientation at cell attachment being random.

Fig. 6.

Kinetics of spreading on carpets of ConA (+), plasma FN (○) and placental FN-fibrinogen (PSF) (▫). Data points with three-parameter log normal curves.

Fig. 6.

Kinetics of spreading on carpets of ConA (+), plasma FN (○) and placental FN-fibrinogen (PSF) (▫). Data points with three-parameter log normal curves.

Cells spread rather slowly on type I collagen (μL = 80·2 min), and more slowly still on gelatin (μL = 142·6min); however, these processes occur in the absence of other promoters (Fig. 7). Adhesion on cellular microexudate is also considerably slower than on ConA, PSF or FN (Figs 6, 7), and although the pi_ value is similar for adhesion on FL to that on fibroblast SAM, the λL value is 43 % greater in the former case (Table 4, Fig. 7). This may indicate considerable differences in fibroblast and epithelial cell extracellular matrices and deserves further investigation.

Fig. 7.

Kinetics of spreading on gelatin (▫), type I collagen (○), and substratum-associated microexudate (SAM) from FL (•) and L (+) cells. Data points with computer three-parameter log normal curves.

Fig. 7.

Kinetics of spreading on gelatin (▫), type I collagen (○), and substratum-associated microexudate (SAM) from FL (•) and L (+) cells. Data points with computer three-parameter log normal curves.

The demonstration that the kinetics of cell spreading are well described by a log normal distribution carries several advantages. It is reasonable to suppose that the principal differences in cell populations at plating arise from variation in position in the cell cycle at harvesting. These might lead to variations in time taken to achieve altered morphology. The reason for the log time dependence of spreading behaviour is now open to experimental testing. Secondly, spreading kinetics may now be characterized by two parameters: μL, the mean spreading time, and λL, the distribution about the mean in log time. We have previously noted that fibroblasts require a greater substratum ‘threshold’ density of ConA and ricin than of FN for spreading (Aplin & Hughes, 19816) ; here we also show that the distribution of spreading times is greater on ConA than on FN, a second aspect of the decreased efficiency on an ‘unnatural’ substrate. Slower spreading kinetics on collagen and cellular microexudates may suggest the need for cells to secrete additional promoters before spreading, or (in the case of SAM) the presence of negative modulators of adhesion. We have observed that FL cells migrate more rapidly on SAM than on bare plastic in the presence of serum (V. M. Niven, unpublished). The present work provides an experimental basis for further study of the molecular interactions occurring between cells and their growth surfaces.

This work was supported by a grant from the Medical Research Council awarded to J.D.A. We thank L. J. Foden for assistance in cell culture.

Abatangelo
,
G.
,
Cortivo
,
R.
,
Martelli
,
M.
&
Vecchia
,
P.
(
1982
).
Cell detachment mediated by hyaluronic acid
.
Expl Cell Res
.
137
,
73
78
.
Alitalo
,
K.
,
Kurkinen
,
M.
,
Vaheri
,
A.
,
Kreig
,
T.
&
Timpl
,
R.
(
1980
).
Extracellular matrix components synthesized by human amniotic epithelial cells in culture
.
Cell
19
,
1053
1062
.
Aplin
,
J. D.
&
Foden
,
L. J.
(
1982
).
A cell spreading factor, abundant in human placenta, contains fibronectin and fibrinogen
.
J. Cell Sci
.
58
,
287
302
.
Aplin
,
J. D.
&
Hughes
,
R. C.
(
1981a
).
Protein-derivatised glass coverslips for the study of cell-to-substratum adhesion
.
Analyt. Biochem
.
113
,
144
148
.
Aplin
,
J. D.
&
Hughes
,
R. C.
(
1981b
).
Cell adhesion on model substrata: threshold effects and receptor modulation
.
J. Cell Sci
.
50
,
89
103
.
Aplin
,
J. D.
,
Hughes
,
R. C.
,
Jaffe
,
C. L.
&
Sharon
,
N.
(
1981
).
Reversible cross-linking of cellular components of adherent fibroblasts to fibronectin- and lectin-coated substrata
.
Expl Cell Res
.
134
,
488
494
.
Atherly
,
A. G.
,
Barnhart
,
R. J.
&
Kraemer
,
P. M.
(
1977
).
Growth and biochemical characteristics of a detachment variant of CHO cells
.
J. cell. Physiol
.
90
,
375
386
.
Bardsley
,
W. G.
,
Leff
,
P.
,
Kavanagh
,
J. P.
&
Waight
,
R. D.
(
1980
).
Deviations from Michaelis-Menten kinetics
.
Biochem. J
.
187
,
739
765
.
Butters
,
T. D.
,
Devalia
,
V.
,
Aplin
,
J. D.
&
Hughes
,
R. C.
(
1980
).
Inhibition of fibronectin-mediated adhesion of hamster fibroblasts to substratum: effects of tunicamycin and some cell surface modifying reagents
.
J. Cell Sci
.
44
,
33
58
.
Carlsson
,
R.
,
Engvall
,
E.
,
Freeman
,
A.
&
Ruoslahti
,
E.
(
1981
).
Laminin and fibronectin in cell adhesion: enhanced adhesion of cells from regeneratory liver to laminin
.
Proc. natn. Acad. Sci. U.SA
.
78
,
2403
2406
.
Crouch
,
E.
,
Balian
,
G.
,
Holbrook
,
K.
,
Duksin
,
D.
&
Bornstein
,
P.
(
1978
).
Amniotic fluid fibronectin. Characterisation and synthesis in culture
.
J. Cell Biol
.
78
,
701
715
.
Culp
,
L. A.
,
Murray
,
B. A.
&
Rollins
,
B. J.
(
1979
).
Fibronectin and proteoglycans as determinants of cell-substratum adhesion
.
J. supramolec. Struct
.
11
,
401
427
.
Grant
,
M. E.
,
Heathcote
,
J. G.
&
Orkin
,
R. W.
(
1981
).
Current concepts of basement membrane structure and function
.
Biosci. Rep
.
1
,
819
842
.
Grinnell
,
F.
&
Feld
,
M. K.
(
1982
).
Fibronectin adsorption on hydrophilic and hydrophobic surfaces detected by antibody binding and analysed during cell adhesion in serum-containing medium
.
J. biol. Chem
.
257
,
4888
4893
.
Hay
,
E. D.
(
1981
).
Extracellular matrix
,
J. Cell Biol
.
91
,
205s
223s
.
Hughes
,
R. C.
,
Butters
,
T. D.
&
Aplin
,
J. D.
(
1981
).
Cell surface molecules involved in fibronectin-mediated adhesion. A study using specific antisera
.
Eur.J. Cell Biol
.
26
,
198
207
.
Hughes
,
R. C.
,
Mills
,
G.
&
Courtois
,
Y.
(
1979
).
Role of fibronectin in the adhesiveness of bovine lens epithelial cells
.
Biologie cell
.
36
,
321
330
.
Hughes
,
R. C.
,
Pena
,
S. D. J.
,
Clark
,
J.
&
Dourmashkin
,
R. R.
(
1979
).
Molecular requirements for the adhesion and spreading of hamster fibroblasts
.
Expl Cell Res
.
121
,
307
314
.
Klebe
,
R. J.
,
Bentley
,
K. C.
&
Schoen
,
R. C.
(
1981
).
Adhesive substrates for fibronectin
.
J. cell. Physiol
.
109
,
481
488
.
Knox
,
P.
&
Griffith
,
S.
(
1982
).
The abnormal morphology of polyoma-transformed BHK cells is due to a failure to respond to 70K spreading factor
.
J. Cell Sci
.
55
,
301
316
.
Pena
,
S. D. J.
,
Mills
,
G.
,
Hughes
,
R. C.
&
Aplin
,
J. D.
(
1980
).
Polypeptide heterogeneity of hamster and calf fibronectins
.
Biochem. J
.
189
,
337
347
.
Schor
,
S. L.
,
Schor
,
A. M.
&
Bazill
,
G. W.
(
1981
).
The effects of fibronectin on the migration of human foreskin fibroblasts and Syrian hamster melanoma cells into three-dimensional gels of native collagen fibres
.
J. Cell Sci
.
48
,
301
314
.
Terranova
,
V. P.
,
Rohrbach
,
D. H.
&
Martin
,
G. R.
(
1980
).
Role of laminin in the attachment of PAM 212 (epithelial) cells to basement membrane collagen
.
Cell
22
,
719
726
.
Vaheri
,
A.
,
Kurkinen
,
M.
,
Lehto
,
V. P.
,
Linder
,
E.
&
Timpl
,
R.
(
1978
).
Codistribution of pericellular matrix proteins in cultured fibroblasts and loss in transformation: fibronectin and procollagen. Proc
.
natn. Acad. Sci. U.S.A
.
75
,
4944
4948
.
Waight
,
R. D.
,
Leff
,
P.
&
Bardsley
,
W. G.
(
1977
).
Steady-state kinetic studies of the negative cooperativity and flip-flip mechanism for Escherichia coli alkaline phosphatase
.
Biochem. J
.
167
,
787
798
.