Contact-mediated triggering of lamella formation by Dictyostelium amoebae on solid surfaces

When Dictyostelium amoebae are allowed to settle on solid surfaces, cell spreading is often followed by the transient appearance of a thin cytoplasmic sheet or lamella (Gingell & Vince, 1982). Such ultrathin lamellae are formed by centripetal cytoplasmic retraction, beginning at the outer boundary of cell-substratum contact. The ultrathin lamellae are easily seen by interference reflection microscopy (IRM), where they appear much darker than the rest of the cell. Although a dark IRM image can indicate intimate contact between the cell membrane and the substratum (as in focal contacts, see Izzard & Lochner, 1980; Bailey & Gingell, 1988), ultrathin lamellae (=100 nm) also give very dark images that depend but little on the cell-substratum separation (Gingell, 1981). Using total internal reflection aqueous fluorescence (TIRAF), Todd et al. (1988) have proved that these dark IRM images are due to such ultrathin lamellae and that the cell-substratum gap is indeed constant beneath the whole cell.

That circumferential formation of ultrathin lamellae is not merely an arcane phenomenon in an obscure proto-zoon is demonstrated by the fact that such lamellae can be Printed in Great Britain (E) The Company of Biologists Limited 1988 formed by human monocytes. When these are allowed to settle on glass in the absence of proteins the cells rapidly develop peripheral dark edges, as seen by IRM. When viewed by TIRAF under appropriate optical conditions, the strikingly uniform image shows that the ‘fried-egg’ IRM image, like that of the amoebae, is due to peripheral ultrathin lamellae, rather than contact specializations (see Fig. 6, below). Neutrophils on glass can evidently form dark peripheral ultrathin lamellae, and these are clearly seen in the IRM images of Keller et al. (1979), although the authors do not interpret them as such. Furthermore, a recent paper by Curtis & McMurray (1986) shows BHK fibroblasts with broad dark decussated peripheral zones that look remarkably like the retracting ultrathin lamellar regions of amoebae (compare fig. 5 of Curtis & McMurray, 1986, with figs 5, 6 of Gingell & Vince, 1982). However, Curtis & McMurray (1986) do not discuss the possibility that the images of their cells are due to ultrathin lamellae. Vertebrate cells can form ultrathin lamellae in more normal situations; for example, they are seen when human macrophages engulf Leishmauia parasites (Zenian et al. 1979). It is well known that fibroblasts develop a transient and highly dynamic (=100 nm thick) lamellipodium at the tip of the leading lamella (Aber-crombie et al. 1971), and this has been examined by IRM by Izzard & Lochner (1980). However, one should exercise caution before assuming a common mode of formation of ultrathin lamellae in cases where they involve protrusion versus centripetal cytoplasmic retraction. Dictyostelium amoebae form ultrathin lamellae only by centripetal retraction, and this happens only after cells have fully spread. Furthermore, on non-triggering surfaces, fully spread amoebae do not form ultrathin lamellae, so that in these cells spreading and ultrathin lamellar formation are totally distinct processes. In contrast, fibroblast lamellipodia are perhaps formed out of contact with the substratum, and are apparently protruded. In the case of phagocytosis studied by Zenian et al. (1979), the ultrathin lamellae that envelop the parasite also seem to be protruded, following contact stimulation. Finally, we should distinguish between a mode of cell spreading that involves ultrathin lamella formation versus a mode that does not. The response of amoebae investigated here is quite distinct from the latter, but might be related to the former, in terms of either the mechanics of ultrathin lamella formation or possibly its triggering.

The nature of the transductive process that triggers the cytoplasmic response is of great interest. Since amoebae respond on some types of surface but not on others, we decided to search for common physicochemical characteristics among the diverse surfaces that promote the response. To this end, the response of Dictyostelium discoideum Ax2 amoebae on covalently aminated or methylated glass coverslips was studied as a function of pH and ionic concentration, and compared with behaviour on a clean glass control surface. To assess a possible relationship between the ultrathin lamella response and surface electrical charge density, the streaming potentials generated by a laminar flow of electrolyte across the glass surfaces were measured as functions of pH and ionic concentration, using a parallel plate flow chamber. Surface charge and zeta potentials were calculated from streaming potentials by standard methods.

3-Aminopropyl triethoxysilane and octadecyldimethyl chlorosilane were purchased from Sigma Chemical Co., Poole, UK. The former was used undiluted whereas the latter was dissolved in re-distilled, dry chloroform to give a 2% (w/v) solution. Poly-L-lysine hydrobromide from Sigma Chemical Co. (.’M_r = 40000) was dissolved in distilled water to give a 0 ·1% (w/v) solution. Agarose (type II-A, from Sigma Chemical Co.) was dissolved in phosphate buffer, pH6 ·8, to give a 2% (w/v) gel. Concanavalin A conjugated with rhodamine B tetramethyl isothiocyanate (ConA-Rh) from Sigma Chemical Co. was dissolved in 0 ·3% (w/v) Tris buffer (pH 7 ·4) to give a stock solution of 0’5 mg ml⁻¹. Samples (200 μl) were stored at —40°C in capped Eppendorf tubes until required.

Methods for the culture and isolation of Dictyostelium discoideum Ax2 strain amoebae from shaken suspension in axenic medium supplemented with glucose and the isolation of human red blood cells have been described (Owens et al. 1987).

Glass coverslips (22 mm × 22 mm × 0 ·17 mm and 76 mm × 26mm × 0’17mm, from Chance Propper Ltd, Smethwick, UK) were cleaned by brief immersion in hydrofluoric/nitric acid solution as described by Owens et al. (1987) and then derivatized using one of the following methods: (1) 8h immersion in 2% (w/v) octadecyldimethyl chlorosilane (ODMS).-The derivatized coverslips (octadecyl glass) were well washed with, and then stored in, dry chloroform until required. (2) Overnight immersion in 3-aminopropyl triethoxysilane (APTS). Just before the APTS coverslips were used, the APTS was decanted off and the coverslips were thoroughly washed in distilled water. Sialation of -OH groups on the glass surface with APTS yields covalently linked amino groups on the surface (Aplin & Hughes, 1981; Rauvala & Hakomori, 1981). (3) By immersion in an aqueous 0 ·1% (w/v) solution of poly-L-lysine hydrobromide for 10 min. The coverslips were removed from the poly-L-lysine solution, well rinsed and kept in distilled water until required.

The presence of adsorbed poly-L-lysine and covalent amination was routinely checked using a dilute suspension of glutaraldehyde-fixed red blood cells (RBC) in distilled water (pH5 ·5). Black interference images, indicating intimate adhesion due to electrostatic attraction, indicated satisfactorily derivatized surfaces. In contrast, RBCs were not adherent to control glass, due to the electrostatic repulsion between the cell and the anionic surface (Gingell & Todd, 1980). Aminated surfaces yielding a negative RBC test were discarded.

For cell behaviour studies, NaCl solutions (5-200 mM) containing 2 ·5 mM-phosphate buffer were used. Electrokinetic measurements were made in NaCl solutions over a lower concentration range (l-20mM). Except where indicated, solutions were buffered with 2 ·5 mM-Sorenson’s phosphate.

The equipment used for IRM is essentially that described by Gingell et al. (1982), except that the universal microscope was inverted. For epifluorescence microscopy, the interference reflector insert was replaced with a Zeiss fluorescence filter assembly (fluorescein: exciter 450/490, mirror FT510 and barrier 515/565; rhodamine: exciter BP456, mirror FT580 and barrier LP590). Photographs were either taken directly with a 35 mm SLR camera (Nikon F3) using Ilford Pan F4 b/w film or from videotape recordings made on a JVC-Umatic videorecorder from a Falcon SIT camera (LTC 1160 Custom Camera Designs, Wells, Somerset, UK).

Cell behaviour was observed under IRM by mounting the glass coverslip as the floor of a simple polytetrafluoroethylene (PTFE) well (Fig. 1) on the stage of the microscope. Dictyostelium amoebae (0 ·2ml at a density of 8 ×10⁴cells ml⁻¹) in buffered saline were introduced and then, after allowing 5 min for settling, 60 amoebae were scored for the presence or absence of ultrathin lamellae. Each experiment was done two or three times and the mean percentage of responding cells was recorded.

Protein adsorption was sometimes a problem on cationic glass, which then gave a false negative in failing to elicit the ultrathin lamellar response. The main source of this contamination was cell rupture during isolation. Routine checks were therefore made by adding RBCs with the amoebae, especially if false negatives were suspected. The frequency and IRM appearance of RBC adhesion, particularly in dilute solutions where electrostatic forces are maximized, provided a check for consistency of the preparations.

Fig. 2 is an exploded diagram of the parallel plate flow system, based on a design by Van Wagenen & Andrade (1980). The flow chamber consists of two perspex sections F,H (140 mm ×63 mm × 20 mm) clamped

Fig. 3 . Exploded detail of the parallel plate flow conduit within the chamber. together through a gasket G with stainless steel bolts. A flat channel (76 mm × 26 mm × 4 mm) in FI leads into a rectangular chamber at each end. The upper face of the channel was formed by the under-surface of F, which also contained end chambers as counterparts to those in H. The gasket was made by coating the mating face of the lower section with liquid silicone rubber. Steel shims 100 pm thick were placed at intervals on uncoated parts of the face and the two sections were clamped together for 48 h. The gasket was then washed in distilled water for 7h to remove acetic acid formed in the curing process. The end chambers served as mixing reservoirs for the electrolyte and also housed the Ag/AgCl measuring electrodes E,E′, which entered through ground-glass joints set in polystyrene pillars. Fluid connections were made via a cylindrical pipe in the end wall of each mixing chamber. To reduce turbulence at entry and exit the corners of the chambers were machined to a 60° radius.

The flow conduit (75 8 mm × 15-0 mm × 0 ·134 mm) shown in Fig. 3 was formed between the upper and lower faces of two parallel glass coverslips C,C′ and the edges of two PTFE spacing gaskets D (Pampus Fluoroplast Ltd, Newcastle, Staffordshire, UK). The coverslips (CM5 glass, 76mm × 26mm × 0 ·17 mm, Chance Propper Ltd) were backed by 2 mm thick glass microslides B,B′ to minimize bending under hydraulic pressure. Leakage of electrolyte between the contacting glass faces was prevented by rendering the backing plates hydrophobic with ODMS. The channel was lined with a 250 qm thick Silescol rubber membrane A,A′ (Esco Rubber Ltd, Tedd-ington, UK), which ensured a leak-free fit of the backing plate. The gaskets (88 mm × 5 mm × 0 ·134 mm) were secured in the conduit with platinum pins. Electrolyte from an elevated reservoir passed through the flow conduit via washed silicone rubber tubing and exited via a glass capillary. The pressure difference was calculated from the mean level in the reservoir before and after flow. The chamber was enclosed in an earthed Faraday box to minimize electrostatic interference.

For R_e < 20, L_e constitutes less than 0 ·003% of the conduit length and thus has negligible influence on the measured streaming potentials.

Under a fully developed laminar flow, charges in the mobile part of the electrical double layer near the wall of a capillary are transported to one end of the flow path. A streaming current (I_S) is thereby established and an electric field set up by the accumulation of charges. This causes a conduction current (I_c) to flow in the opposite direction through the bulk liquid. When a steady state prevails I_c = I_S and the resulting electrostatic potential difference measured by the Ag/AgCl electrodes is the streaming potential (E_s).

Electrodes were made from equal lengths of silver wire (1 mm diam., Good fellow Metals, London, UK) sealed with silicone rubber into the glass joints such that 5 mm (the terminal) protruded at the upper end and 40 mm (the electrode) extended from the base. They were cleaned by immersion in nitric acid (50%, v/v) for 30 min and then electroplated with a silver chloride layer from 0 ·1 M-NH₄CI at a current density of 380 μA per electrode for 6h. A bright platinum plate (80 mm²) was used as the anode in a 12Vd.c. circuit. Electrode asymmetry was checked by measuring potentials developed in l ·40mM-NaCl solutions using a reference calomel electrode. Asymmetry never exceeded 1 ·4%. Agarose salt bridges were required for measurements with NaCl-free electrolytes. Electrodes were connected by screened leads through the Faraday box to a digital electrometer (model 616, Keithley, Reading, Berkshire, UK) with input resistance >10¹⁴ Ω. The asymmetry potential E₀ of the electrodes was measured for the cell in a static condition before determining the potential E_s, developed at the conduit wall during electrolyte flow. For a given fluid pressure difference P, the streaming potential E_s is given by (E_S — E₀). Similarly, the streaming current l_s was measured. Specific conductivities of the electrolyte solutions were measured at 1000 Hz with a standard platinum black conductivity cell connected to a high-impedance capacitance-resistance bridge (model B221 by Wayne-Kerr Laboratory Ltd, Bognor Regis, Sussex, UK). All measurements were made at 20(±l)°C.

A critical review by Ball & Fuerstenau (1973) indicates that much of the earlier published work on streaming potentials suffers from an unpredictable departure of the intercept of E_s from zero when plotted against P, so our data were examined carefully. A plot of (—)E_Sversus P for control glass in Fig. 5 is linear and passes through the origin. A very small deviation from zero was consistently observed for hydrophobic glass over the same range. The important result from all of this is that calculation of ζ_E potentials from E_s data is independent of hydrostatic pressure, as required.

The percentage of cells responding by forming ultrathin lamellae is recorded as a function of pH and electrolyte concentration in Table 1. Subscripts in the tables give the ζ_E potential values of the surfaces (in mV) calculated from streaming potential measurements using identical electrolytes. Bold numbers in Table 1 indicate where the generally low level of response exceeds 40%, but in all other tables the criterion is 70%.

The ultrathin lamellar response was virtually confined to the highest salt concentrations (100-200 mM), though at pH 4-0 and 4-5 some cells produced ultrathin lamellae at 20 and 50 mM. Throughout the entire concentration range there was little response at pH 5·0, but this is not due to any suppressive ionic effect on the cells, since they responded vigorously to aminated glass and octadecyl glass at pH 5·0 (see below).

On aminated glass the response was quite distinct (Table 2). Between 63 and 99% of amoebae gave the ultrathin lamellar response at all pH values and ionic strengths, except at pH 8·0. The lack of response on aminated glass at this pH was not due to the inability of cells to respond, as they did so on control glass at high ionic strength (Table 1). Where the response was seen on the aminated surface, it began as soon as the rapid spreading was complete. After several minutes the thin cytoplasmic sheet was completely withdrawn into the cell, leaving a reticular outline of the IRM ‘black’ area remaining on the substratum (Fig. 7), which appeared to be composed of lamellar fragments. The contact areas observed when cells were fully spread varied greatly from cell to cell, but one with a small initial substratum contact area was just as likely to produce a lamellar response as one which was extensively spread. In contrast to the actively motile response on poly-L-lysine and methylated surfaces (see below), amoebae showed no tendency whatever to locomote on aminated glass, even after an interval of 45-60 min. Amoebae on poly-L-lysine-treated glass also failed to make ultrathin lamellae at pH 8, as shown in Table 3. This indicates that the lack of response on covalently aminated glass is not due to hydrolysis of the silane linkages of amino groups. The observation that RBC adhesion on the aminated surface drops almost fivefold between pH 7·5 and 8 0 is consistent with these results (actual values: in distilled water alone 23% (pH 7·5) versus 6% (pH 8·0); in 200mM-NaCl 92% (pH7·5) versus 20% (pH8·0)). Detailed consideration of these findings will be given below in relation to streaming potential data and surface charge density.

A high percentage of amoebae formed ultrathin lamellae on the strongly hydrophobic surface of octadecyl glass (Table 4), although as on aminated glass there was practically no response at pH 8’0. On octadecyl glass the ultrathin lamellae formed far more slowly than on aminated glass, often taking 2·3 min to appear and 10-15 min to develop fully. Subsequent cell locomotion left a well-defined, dark granular trail of lamellar fragments, shown under IRM in Fig. 8. Evidence that such trails contain cellular debris was obtained from fluorescence labelling experiments. Brightly labelled trails were left by motile amoebae that had been incubated in suspension with ConA-Rh at a concentration of l×10⁻³mgml⁻¹ for 10 min prior to settling on the octadecyl glass substratum. This finding contrasted markedly with the fact that trails left by initially unlabelled amoebae did not bind subsequently added ConA-Rh. In this case trails were distinguished as black footprints against the uniform pink background of the ConA-Rh-labelled octadecyl substratum. These observations may indicate that relatively immobilized membrane carbohydrates cannot bind lectins; a similar lack of binding has been observed for deposited monolayers of glycosides (Owens & Gingell, unpublished data).

Streaming potentials, E_s, for aminated glass were measured in 2·5 mM-buffer alone, 1 mM-NaCl alone, and 5 mM-NaCl containing 2·5 mM-buffer. Values from the means of at least three determinations were transformed into zeta potentials using equation (3). Curves of ζ_Eversus pH for these cases are depicted in Fig. 9.

Fig. 10 shows values of ζ_E calculated from streaming potential data for 1 mM-NaCl on control glass (curve A_g) and hydrophobic octadecyl glass (curve O_g). Although these surfaces are fundamentally different in their chemistry (advancing contact angle (θ_A) for water was zero on clean glass and 112° on octadecyl glass (Mingins & Owens, 1975)) there was remarkably little difference in the ζ_E potentials. Similar puzzling findings were reported by Laskowski & Kitchener (1969) for hydrophilic and hydrophobic silica particles.

The relation between ζ_E and pH (in the range 3-9) for poly-L-lysine adsorbed on glass, in the presence of 2·5 mM-phosphate buffer, is shown in Fig. 11.

General properties of the substrata that might trigger the lamellar response include chemical specificity, substratum wettability, the state of electrostatic charge of the surface and the more complex property of adhesive energy. Since ultrathin lamellae form at high ionic strength on -SiOH (control glass), as well as -CH3 and -NH2-derivatized glass surfaces, the response cannot be chemically specific. Substratum wettability as a unifying factor can also be immediately dismissed owing to the positive response found on highly hydrophobic (-CH₃), and strongly hydrophilic (-NH₂) surfaces. The hypotheses of an electrostatic or an adhesive trigger are far more difficult to assess, and need a detailed consideration of the data, which is developed in the following sections. The reader is advised in advance that an electrostatic consideration does not seem to provide a satisfactory explanation of the biological phenomenon and an explanation in terms of adhesive strength remains as a likely but unproved hypothesis.

Nature of control glass and octadecyl glass su/faces

After amination the glass surface behaves as a charge mosaic consisting of NH₃⁺ protonated amino groups and SiO⁻ groups of silicic acid. This is shown in Fig. 9 by the fact that, in all cases presented, the surface is net positive at low pH due to NH₃⁺ groups, but at high pH it becomes net negative, indicating residual SiO⁻ groups. In 1 mM-NaCl the isoelectric point, pI, is at pH 5-6 (curve a). In 2·5 mM-phosphate buffer alone (curve b) phosphate ion adsorption occurs, as shown by the shift in pI to 4·0. On adding 5 mM-NaCl, pI rises again to 5·6 as seen in curve c, showing that phosphate ion adsorption is abolished by’ NaCl. (Thus streaming potential measurements can provide an elegant alternative to radiotracer methods for phosphate adsorption.)

From these data it is possible to make a minimalestimate of the positive charge contribution from NH3⁺ as a function of pH. Direct calculation from streaming potentials shows that the charge density on the aminated mosaic surface becomes less negative with decreasing pH, passing through the isoelectric point and then becoming positive (Fig. 13, curve a). It is possible to estimate the negative component of the charge mosaic and obtain the positive part by algebraic addition. The argument has several steps. The negative component σ₀⁻ of the mosaic surface is obtained as follows. Assume that the mosaic bears no positively charged groups at pH 8·0 and transfer point •from curve a of Fig. 13 to Fig. 12 and construct the rest of curve b in arithmetic proportion to curve a, since it refers to the same ionogenic species. The resulting curve represents the residual negative charge on the glass surface after derivatization with amino-silane. The positive amine charge (σ⁺), shown in curve b of Fig. 13, is obtained as the difference between the negative component σ⁻ (Fig. 12, curve b) and the total mosaic charge (Fig. 13, curve a). Examination of the curves in Fig. 13 near pH 3·0 shows the density of protonatable N groups is near 5’4 per nm². This procedure gives a minimal estimate of the amine charge, since some NH₃⁺ evidently remains at pH 8’0, as shown by the steeper slope of curve a in Fig. 13 compared with curve a in Fig. 12 at this pH.

On both the control and octadecyl glass surfaces, a disparity was observed between ζ₁and ζ_E (the zeta potentials obtained from the measured streaming current and potential, respectively). This was also noted by Voigt et al. (1983), who attributed it to surface conductivity (additional current flow behind the zone of shear), which decreases ζ₁in comparison to ζ_E. Closer correspondence was observed between these potentials for aminated and poly-L-lysine-treated glass. In the light of these facts, positive and negative components of the charge on the mosaic surface were re-calculated on the basis of ζ₁ but the resulting density of NH₃⁺ (Fig. 13, ‘broken line) differs little from that calculated via ζ_E.

Table 2 shows that amoebae responded to aminated glass by forming ultrathin lamellae in the range 4 < pH < 7 · 5 over the entire concentration range 0 · 200 mM-NaCl. Within this matrix of conditions it was possible to measure zeta potentials of aminated glass only at electto-lyte concentrations <15 mM. Despite this, limits can be put on the likely potential range, since potentials at higher concentrations of NaCl will be closer to zero due to counterion screening. The largest negative potential that still gives the lamellar response is found to be —27 mV at pH 7-5 in 2 · 5 mM-phosphate buffer alone. Thus the response on aminated glass is not confined to conditions where the net zeta potential is positive. Furthermore, since the change in (σ σ _N⁺) versus pH will not be dependent on NaCl at concentrations exceeding 1 mM (since we found that phosphate adsorption is inhibited), examination of Table 2 in relation to curve b in Fig. 13 shows that a high lamellar response is maintained inde-pendent of the value of (σ _N⁺) ^over nearly the whole pH range.

These results on covalently bound amine are also broadly in agreement with observations on poly-L-lysine-treated glass (Table 3). Although stimulation occurs over a narrower pH range (4 · 0—6 · 5), the response nevertheless falls off rapidly where σ₁ s more negative than —35 mV (see subscript values, Table 3). The possibility that a potential near —30 mV is ‘critical’ on hydrophilic surfaces (see next section) is supported by these facts.

If, as the previous analysis suggests, a potential around — 30 mV or less-negative values (i.e. including positive values) is a trigger, the electrostatic conditions on control glass conducive to ultrathin lamella formation should be similar to those on amines. Thus, the zeta potential corresponding to any entry in Table 1 or 2 where cells do not respond should be a larger negative number than that corresponding to any entry where they do, and responses should only occur where σ_Eis less negative than —30 mV. The latter is evidently true of the data on aminated glass in Table 2. Unfortunately no σ_E potential values are available on control glass (Table 1) for any case where cells responded with a high frequency (bold numbers) but the fact that potentials become less negative as ionic strength increases and as pH falls makes it very likely that the response is only significant when is less negative than —30 mV. A more complete assessment could not be made, since streaming potential measurements are unreliable at electrolyte concentrations exceeding 20 mM.

A puzzling feature of Tables 2 and 3 is the complete absence of an ultrathin lamella response at pH 8 · 0 on amino surfaces. Possible explanations include: (1) the cells are directly inhibited at pH 8 · 0; (2) the derivatized surface has been denatured at this pH; and (3) the electrostatic potential of the derivatized substratum has reached some critical value that prevents the stimulation of ultrathin lamellae. Since amoebae form them at pH 8-0 on underivatized glass (Table 1) and octadecyl glass (Table 4), the lack of response on amine surfaces cannot be due to the inability of cells to respond at this pH. Removal of amino groups by alkaline hydrolysis can be dismissed since aminated surfaces exposed to pH 8-0 subsequently stimulate —90% of cells to form ultrathin lamellae at pH 5 · 0. Moreover, streaming potentials measured on aminated glass at pH 4 · 0, before and after exposure to pH 8’0, were found to be identical. These results rule out permanent surface modification and the absence of any abrupt potential change near pH 8-0 argues against any reversible effect. At the present time the significance of these findings remains an enigma.

The analysis so far permits the hypothesis that suggests that on hydrophilic surfaces the lamellar response is triggered when the potential of the substratum is either positive or not too negative (up to —30mV). However, on hydrophobic glass (Table 4) in 2 · 5 mM-phosphate buffer, the calculated ζ_E potential is much more negative than —27 mV at all values exceeding pH 4 · 0, yet ultrathin lamellae are frequently formed at all pH values. Thus the simple electrostatic rationale cannot be extended to this surface.

We are therefore forced by default to conclude that adhesive strength is probably the triggering factor. No single physical property that we have measured can account for the ability of a surface to initiate the lamellar response. It is not due to chemical specificity, wettability or electrostatic potential or sign of the charge. Elimination of these factors makes it likely that sufficiently strong adhesion, operating via an unknown transduction pathway, acts as the trigger. Since the net attractive force pulling a cell onto a hydrophilic or a hydrophobic surface depends on the size of the entropie hydration force (Parsegian & Rau, 1984) at any given set of ionic conditions, there should indeed be no common ‘critical’ electrostatic potential for a lamellar response to adhesive strength. Proposed improvements to our hydrodynamic flow apparatus (cf. Owens et al. 1987) or our recent direct detachment method (Francis et al. 1987) should make it possible to test this hypothesis. Should adhesive energy prove to be the triggering factor, the action of the membrane as a mechanochemical transducer will remain to be understood. The recently reported adhesion-induced redistribution of membrane proteins during HeLa cell spreading (Mason et al. 1987) may provide an intriguing pointer in this direction.

We thank Dr Len Fisher (CSIRO) for a critical reading of our manuscript.