ABSTRACT
The aggregation of isolated cells into coherent multicellular bodies is widely thought to be due mainly if not entirely to the adhesiveness of the cells for one another, according to Moscona (1961a, b), Curtis (1962) and Steinberg (1962a) amongst others. In consequence the aggregation of cells from dispersed (disaggregated) tissues has been widely used as a test for the degree of adhesiveness shown by the cells, and conditions affecting aggregation have been interpreted as affecting cell adhesion. Using this type of test Moscona (1961a, b) found that embryonic chick cells would not aggregate at temperatures below 14° C. It was also discovered that aggregation was inhibited at 37° C. by puromycin and actinomycin D (Moscona & Moscona, 1963), by glucosamine-HCl (Garber, 1963), and by chloramphenicol (Nakanishi et al., 1963). Moscona concluded from the failure of aggregation at low temperatures that the metabolic synthesis of an adhesive substance was being prevented under such conditions and this interpretation was reinforced by the evidence of chemical inhibition of aggregation. Moscona (1962) suggested that the adhesive substance was responsible for producing cell to cell adhesion. In all these experiments cells were aggregated in media containing serum, usually horse serum.
INTRODUCTION
However, Curtis (1963) found that chick embryonic cells could be reaggregated at 1°C. provided that a serum-free medium was used, the presence of horse serum tending to inhibit or limit aggregation at such low temperatures. This result confirms the brief observation by Steinberg (1962a) that some aggregation was possible in serum-free media at 6° C. Curtis suggested that this finding might indicate that serum contains a factor which inhibits aggregation at low temperatures. In view of Moscona’s findings it might be expected that such a factor would be inactivated by cellular metabolism at higher temperatures. This paper describes further evidence for the existence of this factor, its isolation, purification and identification, and some of its biological effects.
METHODS
(a) Biological
Five-, seven-and nine-day chick embryos (White Leghorn) were dissected to give isolated liver, heart and limb-bud. These tissues were disaggregated by treatment with 0·001M ethylene-diamine tetra-acetate (EDTA) made up in calcium and magnesium-free Hank’s solution buffered to pH 8·0 with tris-HC 1 buffer 0·005M, followed by washing with Ca-and Mg-free Hanks saline and mechanical disaggregation. By this means suspensions containing up to 12×106 cells/ml. were prepared. In nearly all experiments in which low temperature aggregation was attempted all manipulations of cell suspensions after disaggregation were carried out below 6°C. and all media used were precooled to below 6°C. The suspensions were then centrifuged at ca. 25 g for 1 min. in order to sediment all cell clumps and erythrocytes leaving a supernatant composed of single cells which was then diluted with medium 1 (see below) to give a single cell density of ca. 6 × 106/ml. Cell population densities were measured on a haemocytometer, which was also used to check the suspensions for the presence of cell clumps (two or more cells in adhesion). More than 95 per cent, of all suspensions were composed of single cells.
The following media were used for aggregation.
Hanks saline, 50 per cent.; Medium 199 (Glaxo), 50 per cent.
Hanks saline, 45 per cent.; Medium 199, 45 per cent.; Horse serum (Burroughs-Wellcome, No. 2 pooled from recalcified plasma), 10 per cent.
Additional media are described in later sections of this paper, and were made by adding treated sera or serum components to medium 1. All culture and disaggregation media were sterilised by filtration (Millepore HA) which also removed dust particles from the media.
All cell suspensions for aggregation were made up to a density of 1·4 × 106/ml. by adding 1 ml. of suspension to 3 ml. of the appropriate medium. The made-up suspensions were then placed in 10-ml. flasks for reaggregation, the gas phase in these flasks being air, and were shaken at 92 strokes/min. at either 37° C. or 1−2°C. In all experiments controls were run using 5-day limb bud cells, in media 1 at 1−2°C., and in many experiments additional controls were provided by aggregating in medium 1 at 37° C. or medium 2 at either temperature. Unless stated to the contrary all experiments were done with 5-day limb bud cells. The measurement of aggregation is discussed and the method adopted is described later.
(b) Chemical
Fractionation of serum was carried out by the following techniques. Whole serum was buffered with phosphate buffer to 0·05 M at pH 7·0 before ammonium sulphate precipitation at 3°C. After (NH4)2 SO4 precipitation fractions were redissolved in 0·005 M phosphate buffer pH 8·0 and desalted on a Sephadex G25 column using the same phosphate buffer as eluant. Further fractionation of the serum was carried out using a DEAE-cellulose column and 0·005M phosphate buffer pH 8·0 as eluant, followed by a CM-cellulose column and 0·02 M acetate-HCl buffer pH 5·5 as eluant with an NaCl concentration gradient. All columns were run at 1°C. Protein concentraton was measured by optical density at 2800 Å using a Uvispek spectrophotometer. Concentration and desalting of protein fractions after the DEAE and CM-cellulose column fractionation was performed by dialysis under reduced pressure at 1°C. Protein solutions were cleaned and sterilised by Millepore filtration. Molecular weight and purity of the serum component active in inhibiting aggregation were determined on the analytical ultracentrifuge by Dr B. Banks.
The measurement of aggregation: the kinetics of aggregation
Moscona (1961a) introduced a method of assessing aggregation produced in shaker systems by measuring the diameter of aggregates produced under given conditions. He suggested that with increasing cellular adhesiveness the larger would the aggregates be. This would be expected because with increasing size the aggregates would be subjected to greater shearing forces tending to break them up ; consequently those aggregates composed of the more adhesive cells would be the larger. There are three main criticisms of this method. First, the smallest aggregate detected by it contains some fifty cells. Aggregates containing fewer cells are scored as being composed either of non-adhesive cells or of cells of low adhesiveness. But, obviously, if a population of isolated cells aggregates into, say, aggregates composed of ten cells apiece, the cells must be adhesive. Only if all of the cell population fails to aggregate can the cells be described as nonadhesive. Second, measuring the size of the completed aggregates cannot be done till some 12 to 18 hr. after the aggregation system has been set up, during which time there may have been changes in cellular adhesiveness. Third, the final size of aggregates does not allow any prediction about adhesion mechanisms.
We felt that these problems might be solved by treating aggregation as an orthokinetic flocculation system and examining the flocculation kinetics of the cells. Flocculation occurs because of the adhesiveness of the particles concerned, and will be either of the fast type if all collisions between particles result in adhesion, or of the slow type if only a proportion of collisions result in adhesions being formed. The probability of an adhesion forming at any one collision will increase with increasing cellular adhesiveness. The most important measurements in flocculation kinetics are (i) the rate of disappearance of single particles as they are built up into aggregates, and (ii) the rate of increase and disappearance of 2, 3−4, 5−8, etc. celled aggregates. If the adhesiveness of the particles remains the same throughout the process the number of single particles and aggregates of various sizes at any given time should be related in a manner predictable from ideal kinetic equations (Overbeek, 1952). These measurements immediately detect formation of the smallest aggregates by the decline in single cell number and the rise in 2-celled aggregates etc. Provided that cellular adhesiveness does not change appreciably during aggregation the flocculation kinetics over the early stages of aggregation allow the prediction of the final aggregate size in rough terms; if the adhesiveness does change during aggregation the flocculation kinetics allow these changes to be followed and thus a clearer picture of cell adhesiveness obtained. Finally, if the rate of particle movement is known the effectiveness of particle collision in promoting adhesion can be calculated, following the treatment given by Overbeek (1952) which should allow a quantitative measurement of adhesion and test of whether the adhesion is with cell surfaces in molecular contact or is a secondary minimum type of adhesion (Curtis 1962).
If normal flocculation kinetics apply in aggregation systems the number of single cells left after a certain aggregation time should give a measure of the adhesiveness of cells. For these reasons we decided to examine the flocculation kinetics of 5-day limb bud cells in aggregation systems grown in media and 2 at both 37° C. and 1−2° C. The populations were sampled at frequent intervals and the number of single cells, 2-cell 3−4, 5−8, and larger than 9-cell aggregates counted. In Text-fig. 1 and 2 the number of cells in the first two of the above classes and the remainder summed together are plotted for various stages of aggregation, together with the theoretical curves from flocculation kinetics for decline in number of single cells, fitted to actual values at zero time and when the population of single cells has declined to a quarter of its original value.
Flocculation curves for the aggregation of cells in medium 1 at (a) 37° C., and (b) 1−2° C. The ordinates of both graphs are calibrated in cell population density (× 106) per ml. and the abscissae in minutes, x points are for measurements of single cell population densities and + points for the number of cells in the two-celled aggregate class at any time. The continuous line curves in both (a) and (b) which closely follow the points are the curves calculated for single cell and two celled aggregate cell number from flocculation kinetics. These curves are derived from the initial density of single cells (1·4 × 106) per ml. and the time taken to reduce the single cell population to a quarter of its initial value. Note the closeness of fit of both curves to experimental points. The broken line curve is an eye fit to experimental points (·) for the number of cells in aggregates larger than 2 celled ones for GJ both (a) and (b).
Flocculation curves for the aggregation of cells in medium 1 at (a) 37° C., and (b) 1−2° C. The ordinates of both graphs are calibrated in cell population density (× 106) per ml. and the abscissae in minutes, x points are for measurements of single cell population densities and + points for the number of cells in the two-celled aggregate class at any time. The continuous line curves in both (a) and (b) which closely follow the points are the curves calculated for single cell and two celled aggregate cell number from flocculation kinetics. These curves are derived from the initial density of single cells (1·4 × 106) per ml. and the time taken to reduce the single cell population to a quarter of its initial value. Note the closeness of fit of both curves to experimental points. The broken line curve is an eye fit to experimental points (·) for the number of cells in aggregates larger than 2 celled ones for GJ both (a) and (b).
Flocculation curves for the aggregation of cells in a medium containing 40 per cent, horse serum at 37° C. and 1−2°C. Ordinate as in Text-fig. 1 but the abscissa is calibrated in hours. The flocculation curves are fitted to the initial single cell density at zero time and to the time for the single cell population to fall to a quarter of its initial value for 37° C. Experimental curve for reduction in single cell number at 1−2° C. shown as broken line marked 1−2°C. It can be seen that the rate of aggregation at 1−2°C. is very low in consequence of which no flocculation curves have been fitted to it. Note that the experimental points for aggregation at 37° C. which are joined by broken line do not fall on the ideal curve. It can be seen that there is a 2-hr. lag period in aggregation at 37° C. which is not found in serum-free media (see Text-fig. 1) and that because the fit is poor it is probable that the cells change their properties of adhesion in these conditions whereas they do not in serum-free media. Note also the slowness of aggregation in this medium compared with that in serum-free media.
Flocculation curves for the aggregation of cells in a medium containing 40 per cent, horse serum at 37° C. and 1−2°C. Ordinate as in Text-fig. 1 but the abscissa is calibrated in hours. The flocculation curves are fitted to the initial single cell density at zero time and to the time for the single cell population to fall to a quarter of its initial value for 37° C. Experimental curve for reduction in single cell number at 1−2° C. shown as broken line marked 1−2°C. It can be seen that the rate of aggregation at 1−2°C. is very low in consequence of which no flocculation curves have been fitted to it. Note that the experimental points for aggregation at 37° C. which are joined by broken line do not fall on the ideal curve. It can be seen that there is a 2-hr. lag period in aggregation at 37° C. which is not found in serum-free media (see Text-fig. 1) and that because the fit is poor it is probable that the cells change their properties of adhesion in these conditions whereas they do not in serum-free media. Note also the slowness of aggregation in this medium compared with that in serum-free media.
It can be seen that the actual aggregation kinetics in serum-free medium (medium 1), both at 37° C. and at 1−2° C., follow ideal flocculation kinetics fairly closely. The number of single cells left after 16 hr. is fairly close to the expected number, though it has been invariably found to be rather greater than the expected value, probably because a slight increase in total cell number due to mitosis occurs at 37° C. and possibly because a small proportion of the cell population has been damaged in disaggregation so that it is unaggregable. It is clear from Textfig. 1 that the single cell population declines as the aggregates build up into successively larger and larger bodies. In other words a measure of the decline in single cell population is a reliable indication of the formation of large aggregates and not merely of two celled aggregates. For these reasons the extent of aggregation was measured by counting the number of single cells after 16 hr. and comparing these with the original number. In all cases when the single cell population had declined appreciably visible formation of large aggregates had occurred.
Aggregation in the presence of serum at 37° C. (see Fig. 2) did not follow ideal flocculation kinetics. At high serum concentrations there is a marked delay period before aggregation starts at 37° C, and the shape of the curve for single cell population after aggregation starts suggests that cell adhesiveness increases with time. However, as in aggregation in serum-free media the fall in un-aggregated cell number is reproducibly related to the rise in aggregate size and the appearance of visible aggregates. In serum media at 1-2° C. there is some slight aggregation (see Table 1) even in the 2-cell aggregate class, unlike the findings of Moscona (1961a) who used other methods of measurement. For these reasons it was decided to use the decline in single cell number between 0 and 16 hr. aggregation as a measure of the aggregability and adhesiveness of the cells. The reproducibility of this technique of measurement can be appreciated from the data for control systems in media 1 and 2 (see Table 1). Similar though not identical flocculation curves were obtained for 5-and 6-day liver and heart cells. It is hoped to publish a full report on these kinetic measurements elsewhere.
Preliminary evidence for the existence of an aggregation-inhibiting factor
The finding that chick limb-bud cells aggregate at 1−2° C. in the absence of serum (Curtis 1963), confirmed here, suggests the existence of a serum factor inhibiting aggregation. The delay period (see Text-fig. 2) before aggregation starts in the presence of serum at 37° C. is not matched in serum free aggregation systems at the same temperature, in which aggregation starts immediately. This difference is compatible with the hypothesis that a serum factor which inhibits aggregation has to be destroyed by cell metabolism before aggregation starts at 37°C.
If the factor is destroyed by cell metabolism at 37° C. then serum containing medium, conditioned by incubation in the presence of cells at 37°C., should not inhibit reaggregation of a second batch of cells at low temperature. On five occasions cells were aggregated at an initial density of l·45× 106 cells/ml. at 37°C. for 20 hr.; the cells were then centrifuged down, the supernatant filtered through a Millipore HA filter, cooled to 3°C. and then used to reaggregate freshly disaggregated cells at 1−2° C. Although the aggregation was not as complete as the control in serum-free medium, it was significantly more complete than that in serum-medium 2 (see Table 1). This result might be due to the absorption or adsorption of a factor from the serum by the first cells aggregated in the medium.
Incubating serum at 37° C. in the absence of cells had no effect on aggregation at low temperature. A similar experiment was carried out by centrifuging a cell suspension (1·45 × 106/ml.) in serum medium at 3° C., taking the supernatant and filtering it through an HA filter, and using it for the aggregation of a second set of cells at 1−2° C. These cells aggregated well in this treated medium (see Table 1). These results provide further reason for accepting the hypothesis that serum contains a factor which inhibits aggregation until it is destroyed by cell action.
Additional evidence in favour of the hypothesis that serum contains an aggregation inhibitor was obtained by pre-treating serum with rabbit anti-horse serum precipitating antibody (Burroughs-Wellcome, Titre 1/2000 0·5 ml. antiserum was added to 20 ml. serum at 20° C. Allowed to stand 1 hr. and then centrifuged). The serum was then used to make up a medium (10 per cent, antibody treated horse serum, 45 per cent. 199, 45 per cent. Hanks aline) in which cells were aggregated at 1−2° C. It can be seen from Table 1 that this antibody-treated serum allowed extensive aggregation at 1−2° C.
These results all provide confirmatory evidence for the idea that horse serum contains a factor inhibiting aggregation which can however be destroyed by cellular metabolism. The absorption experiments suggest that the factor is rapidly bound to the cells. On these results we proceeded to the isolation of the factor.
Isolation and characterisation of the adhesion-inhibiting factor
The test system used in isolation of the factor was the extent of inhibition of aggregation of 5-day limb bud cells at 1−2°C. in Hanks: 199 medium to which a given serum fraction had been added. By volume these media consisted of 45 per cent. Medium 199, 45 per cent. Hanks saline, 10 per cent, serum fraction in 0·005M phosphate buffer pH 7·0; the protein concentration was usually 1 mg./ml. (compared with ca. 15 mg./ml. in medium 2) and was always recorded for the purpose of calculating the purification factor achieved in that stage of fractionation. Aggregations in media 1 and 2 at 1−2° C. were carried out as controls.
At this point dose-response curves were worked out for the inhibition of aggregation at 1−2° C. in terms of single cell number after 16 hr. aggregation, in order to complete calculation of purification in terms of aggregation inhibiting activity and recovery factors at all stages of the isolation. Response was measured by the reduction in number of single cells after 16 hr. aggregation. Dose response curves were constructed for (a) whole serum in terms of percentage of serum in medium, and (b) 32-38 per cent. (NH4)2SO4 saturation precipitated fraction after re-solution and desalting in terms of mg./protein per ml. The combined curve from these two curves is shown in Text-fig. 3. This curve is for aggregation set up at an initial cell density of 1·4 × 106/ml., curves for other initial cell densities have not yet been worked out. It can be seen that the dose-response curve is reproducible. However, the curve is slightly unsatisfactory in that measurements are only possible over a small dose range. Consequently, in practice several dilutions of any one fraction were assayed in order to obtain at least one measurement of aggregation inhibiting activity which lay on the ‘measuring’ part of the curve. It was decided to define a unit of activity as one hundredth of that amount of factor which leads to 45 per cent, inhibition of the aggregation of 1·4 × 106 cells after 16 hr. as compared with aggregation in medium 1.
Dose response curve for the effect of whole serum, serum fractions and the purified aggregation inhibiting protein on aggregation at 1-2° C. Abscissa calibrated in reduction in single cell population per ml. after 16 hr. aggregation. Ordinate calibrated in per cent, of the medium as serum for whole serum, in mg./ml. for the 32-38 per cent, ammonium sulphate precipitate fractions and in units/ml. for the pure protein. It should be remembered that in a population of 1·4 × 106 cells put only ca. 1·2 × 106/ml aggregate in the absence of any inhibitor.
Dose response curve for the effect of whole serum, serum fractions and the purified aggregation inhibiting protein on aggregation at 1-2° C. Abscissa calibrated in reduction in single cell population per ml. after 16 hr. aggregation. Ordinate calibrated in per cent, of the medium as serum for whole serum, in mg./ml. for the 32-38 per cent, ammonium sulphate precipitate fractions and in units/ml. for the pure protein. It should be remembered that in a population of 1·4 × 106 cells put only ca. 1·2 × 106/ml aggregate in the absence of any inhibitor.
The final isolation procedure adopted is shown in Table 2, with recovery percentages and purification values for each stage. Although the ammonium sulphate precipitation properties (see below) suggested that the fraction was a protein this was confirmed by the absorption spectrum between 2400 Â and 3200 Â and by amino-acid analysis (see Curtis, in preparation). The purity of the isolated protein was established by two techniques. First, the specific activity (units/mg. protein) and absolute activity (units/ml.) of the supernatant was measured for (NH4)2SO4 precipitation at successive 2 per cent, saturation steps between 30 and 44 per cent. (NH4)2SO4 saturation. The specific activity of the supernatant remained constant during precipitation over the whole saturation range. The result of this test indicated that the preparation probably contained a single protein (see Taylor, 1962). Second, the protein was examined on the analytical ultracentrifuge. This was carried out for us by Dr B. Banks, to whom we are most grateful for the following report.
The purified protein has been subjected to ultracentrifugal analysis in a Spinco, Model E Ultracentrifuge. The best samples showed a single peak over the entire period of centrifugation (60−90 min. at 59,700 r.p.m.), both at pH 6-0 (0 ·1M acetate buffer) and at pH 7·4 (0 ·1M phosphate buffer) (see Text-fig. 4). The sedimentation constants are given in Table 3 below. Approximate values of the diffusion constant, under the same conditions of protein concentration and solvent, have been determined from the rate of spread of a protein solutionsolvent boundary, formed in a capillary-type synthetic boundary cell, in the course of centrifugation at low speeds (6,166 r.p.m.). The constance (within 2 per cent.) of the area under the Schlieren curve representing the protein-solvent boundary, strongly supports the apparent homogeneity of the protein sample.
Schlieren diagrams obtained during centrifugation in 0 ·1M phosphate, pH 7·4, t = 20,7°C., speed 59,780 r.p.m. Plates at 9-min. intervals : a, after 28 min., b, after 52 min.
The molecular weight of the protein has been calculated, from the ratio of S/20D20 to be of the order of 140,000—assuming a partial specific volume of 0·745. The protein is very soluble, and has no absorption bands in the visible spectrum. At 2800 Å the protein has an optical density of 1·51.
It can be seen from the results of these tests that a pure protein has been isolated from horse serum which carries the aggregation inhibiting potency for limb-bud cells found in whole serum. Further tests (see also data in Table 2) show that this protein is a potent inhibitor of aggregation at low temperatures (see Table 4), not only for limb-bud cells but also for heart and liver cells from 5-, 7-and 9-day embryos. The inhibition increases with amount of protein (see Text-fig. 3 and Table 4). In addition the protein inhibits the aggregation at 1−2° C. of quail, Coturnix sp., limb bud cells (see Table 4). Ammonium sulphate precipitates of pooled human, calf and chicken sera were prepared according to details given in Table 2. After desalting these fractions were tested for their effects on aggregation of limb-bud cells at 1−2° C. It can be appreciated from Table 4 that these sera fractions are all inhibitors of aggregation. Although these results are suggestive that a protein identical or very similar to that in horse serum occurs in chick, calf and human sera the identity of these factors in all four species has not yet been established. Serum reconstituted from ammonium sulphate precipitates of horse serum below 32 per cent, and above 38 per cent, saturation, which lacks the aggregation inhibiting protein does not inhibit cell aggregation at 1−2° C. (see Table 4). This result suggests that there is only one aggregation inhibiting protein.
Aggregation in presence of aggregation inhibiting protein and with fractions from various sera. The effect of inhibitors on aggregation

These results demonstrate that a single specific protein occurs in horse serum which causes the inhibition of aggregation found at low temperatures. In view of the kinetics of aggregation in the presence of serum at 37° C. it might be expected that the protein is slowly destroyed by the cells before aggregation starts. Moscona and others (Moscona & Moscona, 1963; Garber, 1963; Nakanishi et al., 1963) had shown that a variety of inhibitors prevented aggregation in serum medium at 37° C. We felt that these inhibitors might be acting to prevent the destruction of the aggregation inhibiting protein and the following experiments were set up to test this hypothesis.
Inhibition of the cellular destruction of the aggregation-inhibiting protein
Moscona & Moscona (1963), Garber (1963) and Nakanishi et al. (1963) had shown that puromycin, actinomycin D, glucosamine and chloramphenicol inhibit aggregation of chick embryo cells in the presence of serum. If these inhibitors act on the destruction of the aggregation inhibiting protein these experiments should be repeatable, using serum-free medium with the addition of the aggregation-inhibiting protein. In addition the inhibitors should not prevent aggregation in medium free from the aggregation inhibiting protein. These two tests were carried out using the inhibitors in the same concentrations as previous workers had tried; namely puromycin 5 μg./ml., actinomycin D, 5 μg./ml., glucosamine-HCl 5000 μg./ml., chloramphenicol 2500 μg./ml., both D-threo and L-threo-chloramphenicol were used. These two isomers of chloramphenicol were used in view of the report (Ellis, 1963) that only the D-isomer affects protein synthesis, both D-and L-isomers altering cell permeability. The isomer used by Nakanishi et al. v/as not stated. These inhibitors were made up in medium 1 and in medium 1 with the addition of 12 or 25 μg./ml. of aggregation inhibiting protein. Chick limb bud cells were suspended in these media containing one or other of the inhibitors They were shaken at 1-2° C. or at 37° C. Control aggregation was carried out with medium 1 and medium 1 with the addition of aggregation-inhibiting protein at both temperatures. Results are shown in Table 4. They show that cell aggregation fails in the presence of the aggregation-inhibiting protein at both 1−2°C. and 37° C. provided that any one of the inhibitors is present, but that the inhibitors do not affect aggregation at either temperature in the absence of the aggregation-inhibiting protein. The inhibition of aggregation at 37° C. in the presence of both the aggregationinhibiting protein and any one of the inhibitors exactly parallels the results of previous workers using whole serum and the inhibitors. This provides further evidence that the aggregation-inhibiting protein is responsible for all the features of aggregation previously observed in the presence of serum. But the finding that puromycin etc. do not prevent aggregation in the absence of the aggregationinhibiting protein shows that aggregation is not basically dependent on any system susceptible to these inhibitors. It appears that the inhibitors act by preventing the destruction of the aggregation-inhibiting protein which would normally occur at 37° C. allowing aggregation to start after its destruction.
DISCUSSION
The conclusions to be drawn from this work are clear and fall under four headings (i) the mechanism of adhesion in embryonic cells, (ii) the nature of this and other macromolecular factors which affect cell adhesion, (iii) the so-called medium ‘conditioning’ reported in certain types of tissue culture, and (iv) aggregation mechanism.
The kinetics of aggregation closely agree with flocculation kinetics for an ideal system. This finding provides further reason for believing that the aggregation of cells is due entirely to their adhesiveness, for flocculation kinetics are a direct expression of particle adhesiveness (Overbeek, 1952). For this reason we interpret the degree of aggregation shown in the various experiments to be a measure of cellular adhesiveness as other workers have done previously (Moscona, 1961 tz, b;Steinberg, 1962a, b). In this work we have confirmed every experimental result on conditions affecting aggregation obtained by Moscona (1961a, b), Moscona & Moscona (1963), Garber (1963) and Nankanishi et al. (1963), but our additional results show that the correct interpretation of these experiments is the converse of that made by Moscona and others. They claimed that the prevention of aggregation by low temperatures and chemical inhibitors showed that a synthesis of an adhesive substance was essential for adhesion of cells. Our isolation of an aggregation-inhibiting protein, which is of persistent action at low temperatures, the finding that the chemical inhibitors did not prevent aggregation in the absence of serum or the isolated protein though they did in their presence, the observation that cells would aggregate in the absence of serum or protein at low temperatures, and the disovery of a lag period before aggregation starts in the presence of serum at 37° C, all show that the action of low temperatures and chemical inhibitors is to prevent the destruction of the aggregation-inhibiting protein rather than the synthesis of an adhesive substance. Moscona (1961b) and others used their evidence from aggregation studies to support the hypothesis that cells are adherent to one another because of the presence of an intercellular cement (ECM). Our results show that there is no longer any evidence from aggregation studies in support of the cement hypothesis of cell adhesion. Although our results do not provide at present any positive support for the lyophobic colloid theory of cell adhesion (Curtis, 1962) they do not contradict this theory. The five inhibitors of destruction of the aggregation-inhibiting protein do not, as far as appears at present, in view of their known specificities act at the same point on the destruction of the protein.
The aggregation inhibiting protein is, we believe, the first protein to be isolated in a pure state which affects cell adhesion. Although fetuin (Fisher et al., 1958, 1959) appeared to affect cell adhesion it was never obtained in pure state, and its effect in promoting cell attachment to glass appeared to increase stretching of the cell which may not be an adhesive function. Lieberman & Ove (1959) described an a-globulin which affected cell adhesion but their preparation contained at least two proteins.
The destruction of the aggregation inhibiting protein by cells cultured at 37° C. suggests that experiments to test the adhesive properties of whole serum, serum fractions and purified serum proteins should be carried out rapidly or at low temperatures in order to prevent the destruction of this or any other protein which may be labile. In addition, many workers (e.g. Weiss, 1959) who have investigated the effect of sera and serum proteins on adhesion of cells have tended to use whole serum as a control medium. Obviously if an adhesion-inhibiting protein is removed during fractionation other fractions will appear to promote cell adhesion by comparison with whole serum. For these reasons previous work in which measurements of cell adhesion lasted many hours or in which whole serum was used as a control medium must be called in question.
The observation that adhesion will not commence until the aggregation-inhibiting protein is destroyed suggests that destruction of this protein is a feature of medium ‘conditioning’ ; such ‘conditioning’ of the medium by cellular activity has been described (Wilde, 1961) as a prerequisite for cell growth and adhesion but previously no specific change has been identified nor has it been suspected that, as appears in this work, conditioning is at least in part a removal rather than an addition of components to the culture medium by the cells.
The close parallel between the measured kinetics of aggregation in serum-free medium and ideal flocculation kinetics indicates that the adhesiveness of the cells remains constant during the process of adhesion. The slight difference between the curves for aggregation at 1−2°C. and 37°C. in serum-free medium might suggest that the cells are slightly less adhesive at the lower temperature but it should be remembered that the rate of flocculation at a given shaking rate also depends upon the viscosity of the medium which is higher at lower temperatures. It is hoped to solve this question in subsequent work. The flocculation curves for aggregation in serum medium at 37°C. strongly suggest that cellular adhesiveness increases during aggregation because the curves show a delay by comparison with the ideal curves at the start of aggregation but then show an acceleration of aggregation by comparison with the ideal curve. The chemical properties of the protein and further biological effects will be described in two separate papers (Curtis, in press and in preparation).
CONCLUSIONS
The aggregation of embryonic chick and quail limb bud, heart and liver cells in a shaker system has been investigated in order to elucidate the mechanism of cell adhesion.
A new method of assessing aggregation is described and evaluated; this method involved measurements of aggregation kinetics. It was found that the measured kinetics agree closely with flocculation kinetics, which provides a direct quantitative measurement of adhesiveness in terms of the probability of an adhesion forming on contact between two cells.
Using this technique it was confirmed that aggregation is inhibited at 1−2°C. in the presence of serum but not in its absence. The kinetics of aggregation in the presence of serum at 37°C., the lack of inhibition of aggregation at 1−2°C. by serum which has been ‘conditioned’ by the presence of cells, and the similar lack of inhibitory power of serum which has been pre-treated with antiserum antibodies, all suggest that serum contains a factor which inhibits aggregation until destroyed by cellular metabolism.
The factor was isolated from horse serum as a pure protein, M.Wt. 140,000. This protein is an inhibitor of aggregation at 1−2°C. and carries all the inhibitory activity shown by whole serum. Similar proteins have been detected in human, calf and chick sera.
Puromycin, Actinomycin D, glucosamine hydrochloride and D-threo-and L-threo-chloramphenicols, which had been found by other workers to inhibit aggregation at 37°C. in the presence of serum, also show inhibitory action when the aggregation-inhibiting protein is added to a serum-free aggregation medium, but show no inhibitory power in the absence of serum or the protein at either 37°C. or 1−2°C.
It is concluded from these results that whole serum contains a protein which inhibits cell aggregation until it is destroyed by cell metabolism. This destruction is prevented by the inhibitors or low temperatures as described above. Previous work on the inhibition of aggregation suggested that aggregation will only take place when cells are able to synthesize an adhesive intercellular cement, but these results demonstrate that the inhibition is due to prevention of the breakdown of the aggregation inhibiting protein. The results are also discussed in relation to other identifications of proteins from serum which are believed to affect cell adhesion and to medium ‘conditioning’.
RÉSUMÉ
Inhibition de Vagrégation cellulaire par une protéine sérique pure
On a étudié l’agrégation de cellules embryonnaires de poulet et de caille appartenant au bourgeon de membre, au coeur et au foie, dans un système agité, pour élucider le mécanisme de l’adhésion des cellules.
On décrit et on évalue une nouvelle méthode d’estimation de l’agrégation; cette méthode implique des mesures sur la cinétique de l’agrégation. On a trouvé que les valeurs cinétiques mesurées sont en accord étroit avec la cinétique de la floculation, qui fournit une mesure quantitative directe de l’adhésivité en termes de probabilité d’une adhésion se formant par contact entre deux cellules.
A l’aide de cette technique, on a confirmé que l’agrégation est inhibée à 1−2°C. en présence de sérum mais pas en son absence. La cinétique de l’agrégation en présence de sérum à 37°C., l’absence d’inhibition de l’agrégation à 1−2°C. en présence de sérum ‘conditionné’ préalablement par des cellules, et l’absence similaire de pouvoir inhibiteur dans le sérum pré-traité par des anticorps sériques, suggèrent tous que le sérum renferme un facteur qui inhibe l’agrégation jusqu’à ce qu’il soit détruit par le métabolisme cellulaire.
Ce facteur a été isolé du sérum de cheval sous forme d’une protéine pure de PM = 140.000. Cette protéine est un inhibiteur de l’agrégation à 1−2°C. et est dépositaire de toute l’activité inhibitrice du sérum complet. Des protéines similaires ont été décelées dans le sérum humain et les sérums de veau et de poulet.
La puromycine, l’actinomycine D, le chlorhydrate de glucosamine, les D-thréo et L-thréo-chloramphénicols, dont d’autres auteurs avaient montré qu’ils inhibent l’agrégation à 37°C. en présence de sérum, présentent aussi une action inhibitrice quand on ajoute la protéine inhibitrice de l’agrégation à un milieu d’agrégation dépourvu de sérum, mais n’ont pas de pouvoir inhibiteur en l’absence de sérum ou de protéine, soit à 37°C., soit à 1−2°C.
On conclut de ces résultats que le sérum complet renferme une protéine qui inhibe l’agrégation cellulaire jusqu’à ce qu’elle soit détruite par le métabolisme cellulaire. Cette destruction est empêchée par les inhibiteurs ou les basses températures comme on le décrit ci-dessus. Des travaux antérieurs sur l’inhibition de l’agrégation ont suggéré l’idée que l’agrégation ne se produit que si les cellules sont aptes à synthétiser un ciment intercellulaire adhésif, mais ces résultats démontrent que l’inhibition est due à l’empêchement de la destruction de la protéine inhibitrice de l’agrégation. On discute aussi ces résultats en rapport avec d’autres identifications de protéines sériques, dont on croit qu’elles affectent l’adhésion cellulaire, et avec le ‘conditionnement’ du milieu.
ACKNOWLEDGEMENTS
We are very grateful to Professor M. Abercrombie for his encouragement and to Dr C. Vernon for his advice. We also greatly appreciate the help and advice given by Dr B. Banks, in particular for the ultracentrifugation. We would also like to thank Dr J. Hanson, Dr S. Doonan for their help and Mr R. Aldridge for his technical assistance. The work was supported by a grant from D.S.I.R.