Thermal transitions in the adhesiveness of hela cells: effects of cell growth, trypsin treatment and calcium

As far as we know the effect of temperature on adhesion invariably has been investigated using cells from tissues or from monolayer cultures. The cells either were dissociated by trypsinization, or by mechanical means in calcium- and magnesium-free medium, and the adhesiveness subsequently was studied by examining cell re-aggregation in suspension at various temperatures. The work of Kolodny (1972) may be an intermediate case in that he studied the influence of temperature on the reattachment to a substratum of trypsinized monolayer cells.

There appears to be agreement on the observations that (a) cells dissociated by trypsinization are less adhesive at 6 and at 37 °C, than cells dissociated mechanically; and that (b) the adhesiveness of trypsinized cells diminishes on lowering the temperature (Moscona, 1952, Moscona, 1961; Steinberg, 1962; Moscona & Moscona, 1966; Roth, 1968; Jones & Morrison, 1969; Edwards & Campbell, 1971; Kolodny, 1972).

For the influence of temperature on untrypsinized cells, conflicting results have been reported (Curtis, 1963; Curtis & Greaves, 1965; Steinberg, Armstrong & Granger, 1973; George & Rao, 1975). Some of the differences probably can be ascribed to the addition of serum to the resuspended cells (Curtis, 1973; Curtis, Campbell & Shaw, 1975a).

Little attention has been paid to the growth state of the cells. In all studies mentioned cells were used which normally can grow only when attached to a substratum, but measurement of the adhesiveness was performed on suspended cells. A priori it cannot be excluded that detachment from the substratum, which so drastically affects growth, also might affect intercellular adhesion.

In previous publications we demonstrated the existence of a marked difference between the adhesiveness of HeLa cells from fast-growing (F.G.) suspension cultures and HeLa cells harvested from density-inhibited (D.I.) suspension cultures. Measurements were performed at 37 °C (Deman & Bruyneel, 1974, 1975). In this study we measured the adhesiveness of both kinds of cells at temperatures between 6 and 37 °C. In order to allow comparison with work of others we also examined the effect of trypsinization of the cells and of the presence of calcium ions in the buffer medium during the measurement of adhesion.

HeLa cells were grown asynchronously in spinner cultures. Growth became density-inhibited above cell densities of 1·8 × 10⁶ cells/ml. Fast-growing cells were harvested at densities of approximately 1·0 × 10⁵ cells/ml. More explicit conditions together with the preparation of the cells before measurement of the adhesion were described in a previous paper (Deman, Vakaet & Bruyneel, 1976).

Intercellular adhesion was measured at different temperatures in a Couette-viscometer. The method is based on the effect of particle size on the rate of sedimentation in a rotating suspension (Deman & Bruyneel, 1973). The results were expressed in R.C.C. values (relative cell concentrations). These are inversely related to the rate of cell adhesion i.e. an increase in adhesiveness is accompanied by a decrease in R.C.C. value. Before the measurement, the suspended cells were placed in a waterbath for 10 min, at the required temperature.

Cell viability was assessed by the trypan blue exclusion test. Immediately after harvest the viability varied from 90 to 93 %. After measurement of the adhesion the viability of the cells in the Couette-viscometer varied from 80 to 86 %.

During the measurement of the adhesiveness the HeLa cells were suspended in a Tris Saline Phosphate (TSP) buffer of following composition: 8·0 g NaCl, 0 3 g KC1, 0 05 NaH,PO₄. H₂O, 0·025 g KH₂PO₄, 1·0 g NaHCO₃ and 1 g Tris in 1 litre of distilled water. The pH of this solution was adjusted to pH 7·5 by addition of sodium acetate buffer, 0·5 M, pH 4·0. The osmolality of 310 was reduced to 288 by addition of distilled water. In some experiments explicitly mentioned CaCl₂.2H₂O, MgCl₂.6H₂O or disodium-EDTA.₂H₂O were added. All products were analytical grade.

In order to investigate the effect of trypsin on the adhesiveness, harvested cells after one washing with 0· 9 % NaCl were resuspended in the TSP-buffer mentioned above (Ca²⁺- and Mg²⁺-free), to which trypsin had been added in a concentration of 100μg/ml (crystalline trypsin, Sigma Chemical Co, St Louis, Mo, U.S.A., from bovine pancrease, type I). The cell density in the suspension was about 12 × 10. cells/ml. The suspension was warmed to 37 °C and then incubated at 37 °C for 20 min with gentle shaking. The cells then were washed twice with 0 9 % NaCl and were resuspended in TSP-buffer medium.

Cell diameters were measured with an ocular micrometer following the procedure described previously (Deman et al. 1976). Measurements of diameter at different temperatures were achieved simply by placing the microscope and the samples in a chamber which was kept at the appropriate temperature by an air thermostat.

Shinitzky & Inbar (1974) have introduced a method for the measurement of plasma membrane viscosity (reciprocal of fluidity). The method is based on the fluorescence polarization properties of the fluorescent probe DPH (1,6-diphenyl 1,3,5-hexatriene) (supplied by Elscint, Rehovoth, Israel), which is incorporated into the lipid regions of the plasma membrane. In this work we proceeded exactly as prescribed by the above authors. Fluorescence polarization was measured by a microviscometer Model MV-1 (Elscint, Rehovoth, Israel). Temperature was measured by a thermistor immersed in the sample. The microviscosity was calculated using the expression η = 2P/(0 η 46-P), where η is the viscosity in poise (1 poise = 10 ⁻¹ N 8 m⁻²) and P is the degree of fluorescence polarization. The latter value is obtained directly in the apparatus.

In a first series of experiments HeLa cells after harvest and washing were resuspended in TSP-buffer solution which contained 3 mM EDTA.

From the figure we notice a marked difference between cells harvested from density-inhibited suspension cultures (D.I. cells) and cells from fast-growing cultures (F.G. cells). The R.C.C. values of D.I. cells remain approximately constant between 37 and 20 °C and display a slight increase at the lower temperatures. In contrast the R.C.C. values of F.G. cells show a considerable decrease between 30 and 10°C. We also remark that the results confirm previous findings (Deman & Bruyneel, 1975) in which F.G. HeLa cells at 37 °C were found to be more adhesive than D.I. HeLa cells at the same temperature.

We next compared the effect of different calcium concentrations in the buffer medium. In the figure it is seen that at 37 °C addition of Ca²⁺ lowers the R.C.C. values of D.I. cells but has no appreciable effect on F.G. cells, in agreement with previous reports (Deman, Bruyneel & Mareel, 1974; Deman & Bruyneel, 1977).

With D.I. cells the R.C.C. values decrease at temperatures below 20 °C, a fact which was not observed for measurements in presence of EDTA (Fig. 1). With F.G. cells Ca²⁺ induces a narrowing of the thermal transition region.

Above mentioned phenomena are observed even in TSP-buffer solution to which no Ca²+ ions were added. This raises a question on the mode of action of EDTA, the curves of Fig. 1 being most significantly different from those which in Fig. 2 are measured in TSP-buffer alone. Therefore we did an experiment in which we added 3η0 mM EDTA + 6η0 mM Ca²⁺ to the TSP-buffer. As can be seen from Fig. 3 the curves resemble those obtained on addition of 3 mM Ca²⁺ alone (Fig. 2), which might indicate that EDTA, apart from its action as Ca²⁺- and Mg²⁺-complexans, has no effect leading to a change in R.C.C. values.

The results therefore are most readily explained by assuming that the thermal transitions are profoundly influenced by very low Ca²⁺concentrations. Small amounts of Ca²⁺ might leak out of the cells. Also small amounts of Ca²+ might remain bound on the plasma membrane after washing and resuspension in TSP-buffer. Addition of EDTA removes these Ca²⁺ ions.

Addition of 3 mM Mg²⁺ to the TSP-buffer gives rise to a curve which is similar to that obtained in presence of 3 mM Ca²⁺, except that the R.C.C. value of D.I. cells at 37 °C is higher in presence of Mg²⁺.

The R.C.C. values at 37 °C of both D.I. and F.G. cells are increased by previous trypsinization of the cells. Temperature has no significant effect on the R.C.C. values of D.I. cells whereas on F.G. cells there is a slight increase in R.C.C. with lower temperature. The latter result can be compared to the behaviour of intact F.G. cells in presence of EDTA (Fig. 1). It demonstrates that trypsinization alters the physicochemical properties of the plasma membrane. At the lower temperatures the curves of D.I. and F.G. cells are very similar.

The behaviour of trypsinized F.G. cells in the presence of Ca²⁺ ressembles that of intact F.G. cells (Fig. 2), with the difference that the R.C.C. values of trypsinized F.G. cells in the lower temperature range lie significantly higher than those of intact F.G. cells. As with intact F.G. cells, the greatest change in adhesiveness is found when the cells are suspended in TSP-buffer alone. As observed above this might indicate that very low Ca²⁺ concentrations produce profound effects on the R.C.C. values at low temperature.

As has been pointed out in Materials and methods, R.C.C. values are considered to be inversely related to the intercellular adhesiveness. The method is based on Stokes law which predicts that larger particles (e.g. aggregates) will sediment at a higher velocity than smaller particles (e.g. single cells). The relation between aggregation rate and R.C.C. value was confirmed by microscopic examination of the suspension at 37 °C (Deman & Bruyneel, 1973).

For other temperatures such confirmation is lacking. We therefore did experiments in which cells were examined at 6, 20 and 37 °C. As can be inferred from Table 1 there is a clear relation between aggregation rate and R.C.C. value.

(v = sedimentational velocity, g = gravitational constant, r = diameter of the particle, Δ ρ = difference in buoyant density between particle and medium, η = viscosity of medium in poise (10⁻¹ N s m⁻²)), it is apparent that alterations of the buoyant density and the viscosity of the medium will affect the sedimentation rate. However since the change of temperature applied in our experiments can give rise to only minor alterations in both parameters which moreover will affect the sedimentation rate of F.G. and D.I. cells alike, they can be neglected as factors which can explain our results.

Another factor is variation in the size of single cells with temperature. At constant density difference Δ ρ, large spherical cells will sediment faster than small spherical cells. In order to test the influence of temperature on cell size we divided cells from the same harvest into 3 portions which were examined at 6, 20 and 37 °C, respectively, on the same day and using the same microscope and micrometer. As can be judged from Table 2 there is no significant effect of temperature on cell diameter. In subsequent experiments not represented here, we compared the effects of EDTA and Ca²⁺ addition on the cell diameter of intact and trypsinized cells at different temperatures and for different cell densities at harvest. No significant effects were found.

It was found earlier in this study that low concentrations of Ca²⁺ ions produce considerable changes in the adhesiveness of intact F.G. cells. In artificial phospholipid membranes low concentrations of Ca²⁺ (10⁻⁴–10⁻⁸ M) are known to produce shifts in the solid-liquid crystalline thermal transitions. Phase separation in liquid membranes can be induced by Ca²⁺ ions but not by Mg²+ ions (Papahadjopoulos & Poste, 1975). Recent work of Curtis and co-workers (Curtis et al. 1975 a, b) indicated that plasma membrane lipids might affect adhesion.

For these reasons and because we suspected the R.C.C. curves of Figs. 1, 2 and 6 of being indicative of a phase transition (see Discussion), we decided to measure the microviscosity of the plasma membranes of F.G. and D.I. HeLa cells. In the method of Shinitzky & Inbar (1974), (see Material and methods) which we adopted, the dependency of log η (microviscosity) on (I/T) (absolute temperature) yields a straight line when there is no change of phase of the membrane lipids.

In Fig. 7, no phase transition can be detected on intact or trypsinized cells. In this respect our measurements confirm those of Shinitzky & Inbar (1974, 1976) on normal and malignant fibroblasts and lymphocytes. We observe that trypsinization increases the microviscosity of both D.I. and F.G. HeLa cells. The microviscosity is higher with D.I. cells than with F.G. cells.

The concept of membrane fluidity which now is widely accepted is a complex one. It can be taken in general sense to indicate lateral movement of molecules in the plane of the membrane. However, the plasma membrane is composed of different kinds of molecules and it has been observed by Shinitzky & Inbar (1976) that the fluidity of membrane lipids is to be distinguished from that of membrane (glyco) proteins. This difference cannot be ascribed to a different methodology for the measurement of the respective fluidities.

Phase transitions from a rigid to a more fluid structure upon raising the temperature are commonly observed with liposomes. No thermal phase transitions could be detected in the cell surface lipids (Shinitzky & Inbar, 1974, 1976). As concerns the cell surface (glyco)proteins the situation may be otherwise. The lateral (Nicolson, 1972) movement of receptor sites for concanavalin A is temperature dependent (Nicolson, 1973; Rosenblith, Ukena, Yin, Berlin & Karnovsky, 1973). Adam & Adam (1975) observed thermal transitions in the surface charge density. They interpreted this as evidence for the occurrence of a phase transition at the plasma membrane.

According to Williams (1975), to speak about the existence of different phases, all that is required is a discontinuity in a thermodynamic function. A phase change is defined only by a co-operative change of heat content over a narrow temperature range. The adhesion curves of intact F.G. cells therefore are most readily interpreted as the expression of a phase transition at the plasma membrane. The higher temperatures being correlated with the more fluid phase, it follows that higher R.C.C. values generally stand for higher fluidity. D.I. cells therefore can be considered to be more fluid than F.G. cells. Trypsinization increases the fluidity whereas Ca²⁺ induces higher viscosity.

In absence of Ca²⁺, there is a broad transition region for intact F.G. cells extending over nearly 20 °C. Presence of Ca²⁺ (Fig. 2) causes narrowing of the transition region. Phase transitions are less obvious on intact D.I. cells (Figs. 1, 2) and are definitely absent on trypsinized D.I. cells (Figs. 5, 6). As concerns intact D.I. cells the results can be explained by assuming that over the temperature range of the experiments, there is no (Fig. 2) or only one (Fig. 1) stable phase.

Trypsinized F.G. cells and intact D.I. cells give rise to similar curves in EDTA-containing medium (Figs. 1, 5) which indicates that trypsin-removable material might contribute to the stability of the rigid phase of F.G. cells. Trypsin is known to release glycopeptides from mammalian cell surfaces (Cook, Heard & Seaman, 1960; Miller, Sullivan & Katz, 1963; Langley & Ambrose, 1967; Buck, Glick & Warren, 1970, 1971; Santer, Cone & Marchalonis, 1973; Kapeller, Gal-Oz, Grover & Doljanski, 1973). Cell surface glycolipids are not degraded or released by trypsin treatment (Hakomori, Teather & Andrews, 1968). Huang, Tsai & Caneflakis (1973) demonstrated that in HeLa cells the proteolytic action of trypsin is almost exclusively directed towards glycoproteins exposed on the cell surface (i.e. those glycoproteins which can be iodinated in the presence of lactoperoxidase). In order to allow comparison with the present study we mention that the authors used a trypsin concentration of 50 μg/ml and that incubation took place at 37 °C for 20 min. The suspension contained 6×10⁵ HeLa cells/ml. As was shown by Philips (1972), also on human platelet membranes the primary target for the action of trypsin are glycoproteins. The author presented evidence that trypsinization might lead to a conformational change of the partially degraded glycoproteins in the platelet membrane.

Trypsin is also known to cause redistribution of some cell surface components. The difference in agglutinability by plant lectins between untreated and trypsin-treated cells (Burger, 1969; Sela, Lis, Sharon & Sachs, 1970) has been explained by a different topographic distribution of receptor sites on the plasma membrane after trypsinization (Nicolson, 1972; Nicolson & Baulstein, 1972; Rosenblith et al. 1973). Also an increased lateral mobility of these receptor sites after protease-treatment has been postulated (Rosenblith et al. 1973). Specific monosaccharides are known to function as binding sites for the lectins. Glycopeptides which act as receptors for concanavalin A and wheat germ agglutinin have been isolated from the plasma membrane (Wray & Walborg, 1971). In some cases workers have succeeded in isolating the membrane glycoproteins which contain the specific receptors (Allen, Auger & Crumpton, 1972; Codington, Cooper, Brown & Jeanloz, 1975). Taken together, this strongly indicates that glycoprotein topography might be affected by trypsin treatment.

Other evidence for the involvement of glycoproteins in the phase phenomena under discussion, is based on the circumstance that our measurements reveal only those phase transitions which lead to alterations in intercellular adhesiveness. There are strong indications that adhesiveness is conferred upon the cells by cell surface glycoproteins (Oppenheimer, Edidin, Orr & Roseman, 1969; Oppenheimer, 1975; Yamada, Yamada & Pastan, 1975; Stockert, Morell & Scheinberg, 1974; Deman et al. 1974).

It seems reasonable to assume that the phase changes under question might be those which are accompanied by conformational changes (Adam & Adam, 1975) of the adhesive glycoproteins or by a modified topographical distribution of these glycoproteins in the lipid continuum of the plasma membrane. Trypsin is assumed to remove most readily the exposed part of the glycoprotein molecules which might be that part of the molecule which contributes most to intercellular adhesion. Partial degradation of the glycoproteins by trypsin might result in a conformational change in these macromolecules and in an increase in their lateral mobility which might prevent the attainment of the glycoprotein distribution characteristic of the rigid phase, upon lowering the temperature. We realize that much in this hypothesis is speculative and that several of its points need further confirmation.

Absence of a phase transition in the membrane lipids might point to the circumstance that although the organization of (glyco)proteins and lipids in the plasma membrane is mutually dependent (Kleemann & McConnel, 1976; Nicolson, 1976), phase transitions of the diverse components may take place at different temperatures.

The fluidity characteristics of the membrane components are believed to play a major role in the cellular control mechanisms and to determine normal and malignant cell growth (Adam & Adam, 1975; Shinitzky & Inbar, 1976). The present study demonstrates that release from density inhibition of cell growth in suspension is accompanied by profound alterations at the plasma membrane, the exact nature of which is unknown but which most readily are interpretable in terms of fluidities of glycoproteins and lipids.

This study was supported by a grant from the Fund for Medical Research (F.G.W.O.), Belgium. We thank Mrs R. Van Cauwenberghe for skilled technical assistance. We also are grateful to Dr M. Y. Rosseneu-Motreff and Mr Dutoit for the use of the Elscint microviscosimeter.