Aggregative behaviour of embryonic chick cells in the presence of anti-bodies directed against actomyosins

On the basis of an actomyosin-adenosine triphosphate (ATP) system being present at the surfaces of rat liver cells and human platelets and of the ability of cations and exogenous nucleotides to modulate the adhesiveness of platelets and tissue cells, it was previously suggested that a surface-localized actomyosin-like protein may well function in controlling cell adhesion (Jones, 1966, 1967; Jones & Kemp, 1970). Support for this view appeared to be strengthened by the finding that rabbit antiserum against platelet actomyosin (thrombosthenin) inhibited theadenosine triphosphatase(ATPase) activity of this protein at the platelet surface and induced platelet aggregation (Chambers, Salzman & Neri, 1967). However, there was undoubtedly room for questioning the specificity of this antiserum in that it may have contained contaminant antibodies against, for example, tropomyosin, glycoproteins and ATPase systems other than myosin ATPase. There also remained the possibility that the anti-thrombosthenin serum acted as an agglutinant.

Rabbit antisera against human actomyosins have been found to be type-specific (Gröschel-Stewart & Gigli, 1968; Gröschel-Stewart, 1968) but not species-specific (Jones, Kemp & Gröschel-Stewart, 1970). The anti-uterine smooth muscle actomyosin (anti-UAM) serum and anti-pectoralis striated muscle actomyosin (anti-PAM) serum were capable of blocking the Ca²⁺-dependent myosin ATPase of their respective anti gens (Gröschel-Stewart, 1968) but did not react with the Na⁺, K⁺, Mg²⁺-dependent ATPase of human erythrocyte ghosts (Gröschel-Stewart, 1969). Preliminary studies indicated that anti-UAM serum markedly inhibited the aggregation of embryonic chick muscle and liver cells (Jones et al. 1970). Anti-PAM serum, calf serum and rabbit non-immunized serum failed to produce this inhibitory effect. It can be inferred from the finding that anti-UAM labelled with fluorescent dye specifically stained embryonic chick cells (Gröschel-Stewart, Jones & Kemp, 1970) that this antiserum accomplished its aggregation-inhibitory effect (Jones et al. 1970) by reacting with a surface-localized actomyosin-like protein.

However, several doubts remained. The possibility that the inhibitory effect was due to impurities in the anti-UAM serum could not be dismissed. It was not known whether or not the effect on aggregation was restricted to antibodies against human smooth muscle actomyosin.

The aims of this investigation have therefore been 4-fold: first, to prepare antisera against chicken actomyosins and to test their equivalence to those prepared against human actomyosin proteins; secondly, to confirm the effects of human anti-actomyosin sera on cell aggregation and to compare these effects with those of chicken anti actomyosin sera; thirdly, to eliminate possible antibodies against impurities in the smooth muscle preparations as causal factors in the aggregation-inhibitory effect; and, fourthly, to test the ability of fluoresced antibodies against both smooth- and striated muscle actomyosins to react with various non-muscle cell systems. The results of these studies have been discussed in relation to the formation of antibody–actomyosin complexes at the cell periphery and to the view that cell adhesion may involve the participation of surface-localized actomyosin-like protein.

Skeletal muscle and liver tissue from 9-day-old chick embryos at the 35th stage of develop ment (Hamburger & Hamilton, 1951) were dissociated into separate cells by treatment with 0·25% (w/v) crude trypsin (Burroughs Wellcome, 1:300) in Hanks’s balanced salts solution (BSS) for 10 min at 37 °C (Kemp, Jones, Cunningham & James, 1967). The dissociated cells were washed twice in Hanks’s BSS. The mixture of fibroblasts and myoblasts obtained from the embryonic muscle tissue are referred to in the text as muscle cells.

The turbidimetric method (Born, 1962) modified for tissue cells by Jones (1965) was used to estimate quantitatively the aggregation of the muscle cells which were rotated in Hanks’s BSS without phenol red at a concentration of 7·0 × 10⁶ cells/ml (Cunningham & Hirst, 1967). The gyratory shaker method of Moscona (1961) was employed to estimate the aggregative com petence of both muscle cells and liver cells. In this method cells prepared under sterile condi tions were suspended in Eagle’s minimum essential medium (MEM) at a concentration of 2·0 × 10⁶ cells/ml (Kemp, 1970). The cells were rotated in sealed flasks at 70 rev/min for 24 h at 37 °C.

Cell viability was checked by using a 2 % (w/v) aqueous solution of lissamine green (Goldacre & Sylvèn, 1959), which is readily taken up by non-viable cells but not by viable cells (Kemp et al. 1967).

Striated muscle actomyosin was extracted from chicken and human pectoralis muscle (PAM) using a Weber-Edsall solution (0·6 M KC1, 0·01 M Na₂CO₃, 0·04 M NaHCO₃; pH 8·2) and purified according to the method advocated by Grdschel-Stewart & Turba (1963). The resultant preparations were free of haeme impurities. Aggregated protein and particulate im purities were then removed by centrifugation at 10⁶g for 1 h at 0°C in the presence of 10^–3 M sodium pyrophosphate, pH 7·0 (Gergely, Martonosi & Gouvea, 1959).

Smooth-muscle actomyosin was isolated from the chicken gizzard (GAM) using a Weber–Edsall solution, and from the myometrium of non-pregnant human uteri (UAM) using sodium phosphate buffer, pH 7·3, at low ionic strength, μ 0·075 (Huys, 1963). The methods of Gröschel–Stewart & Turba (1963) and Gergely et al. (1959) were employed to purify the smooth-muscle actomyosin. Tropomyosin was extracted from smooth muscle according to the procedure of Baily (1948).

The protein content of the preparations was estimated by the micro-Kjeldahl method. The actomyosin solutions were stored at –20 °C in 50% (v/v) glycerol.

Antisera against actomyosin were prepared by injecting purified actomyosin in gel form into rabbits. Approximately 1 mg of protein in complete Freund adjuvant (Difco) containing 1 mg of killed Mycobacterium tuberculosi was injected directly into the popliteal lymph node. On the 21st day and on 2 further alternate days, 10 mg of protein in 3–4 ml of 0·155 M NaCl was injected into the lateral vein of the ear (Finck, 1965). The course of injections was repeated 21 days later. The antiserum was collected 12 days after the last dosage by catheterization of the arteria carotis extema and stored at —20 °C. When required for use in aggregation experiments the antiserum was heated to 56 °C for 10 min to destroy complement factor.

Globulin-enriched fractions of rabbit antisera and of non-immunized serum were obtained by 3-fold precipitation with a saturated ammonium sulphate solution at a final concentration of 1·35 M, pH 7·4 (Campbell, Garvey, Cremer & Sussdorf, 1964). The fractions were then lyo-philized after thorough dialysis against saline buffered to pH 8. Before use the globulin fraction was dissolved in Dulbecco physiological saline and the protein concentration determined from the optical density ratio at 215 and 225 nm (Waddell, 1956).

A double-diffusion technique was used for demonstrating antibody-antigen reactions. Glass slides were coated with 3 ml of a warm 1 % (w/v) solution of agar (Difco Noble) in 0·6 M KC1-0·02 M tris-HCl buffer, pH 7·4. The distance between the ‘antigen’ holes, 4 mm in diameter, and the ‘antibody’ groove was 3 mm. The antiserum was used as prepared; it was not diluted. The concentration of the antigenic substance to be tested was adjusted to 1–2 mg/ml. Neither the groove nor holes were refilled. Test preparations were kept for 2–6 days in a moist container at room temperature. On the appearance of precipitin bands, the slide preparations were washed consecutively in 0·6 M KC1 and water, and then dried.

Prior to testing their effects on rotated cell suspensions, antisera and globulin fractions were sterilized by filtration through a Millipore Swinnex adapter attached to a syringe. Rabbit antihuman serum and specific rabbit antisera directed against human α₁-acid glycoprotein, β_2-glycoprotein 1, 19-sγ-globulin, γ₁A-globulin and 7-sγ-globulin (all obtained from Behringwerke AG, Marburg-Lahn, Germany) were also tested on rotated cell suspensions to determine whether or not they affected the aggregative competence of the cells.

Antisera were conjugated with fluorescein isothiocyanate (FITC) (Sylvana Co., Milburn, N.J.) in the ratio of 20 μg FITC/mg protein (Roitt & Doniach, 1967). Unconjugated dye was removed by molecular sieve chromatography on a column (20 × 2 cm) of Sephadex G-25. The substances were eluted from the column with phosphate-buffered saline, pH 7·4. The FITC/protein ratio of the conjugate was determined by an absorbance ratio method (Goldman, 1968); the ratio was usually found to be in the region of 3–4:I. Prior to use the conjugates were ab sorbed for 60 min on lyophilized chicken liver powder (100 mg/ml) which had been moistened with a few drops of phosphate-buffered saline (Gröschel-Stewart & Doniach, 1969).

Suspensions of human platelets and cryostat tissue sections, 5 μm thick, were treated for 30 min with FITC conjugates by the direct method (Coons, 1956). The material was then given 3 30-min washes in veronal buffer and mounted in a mixture of 70 % (v/v) glycerol-glycine buffer, pH 8·6 (Roitt & Doniach, 1967).

Cryostat sections through the human chorda umbilicalis revealed that the fibrocytes reacted specifically with FITC-labelled anti-uterine actomyosin, anti-UAM (Fig. 7). Fibrocytes which had migrated into Wharton’s gel were particularly well stained. FITC-labelled anti-human pectoralis striated muscle actomyosin (anti-PAM) and similarly labelled rabbit non-immunized serum (NIS) did not react with the cells. When FITC-labelled anti-UAM was adsorbed on UAM gel (2 mg/ml) before use, its reaction with the tissue was abolished.

Plasma-free platelets, separated by centrifugation from citrated human blood, were stained by FITC-labelled anti-UAM but not by the labelled anti-PAM or NIS (Fig. 8).

Rabbit antisera against chicken gizzard smooth-muscle actomyosin (anti-GAM) and against chicken pectoralis muscle actomyosin (anti-PAM) both gave single pre-cipitin bands with their respective actomyosins on diffusion through agar (Figs. 9, 10). The anti-GAM serum coded AS 2 and the anti-PAM serum coded AS 1 were used in the experiments described later, because they produced the better precipitin reaction. Anti-GAM and anti-PAM sera failed to give a reaction with gizzard tropomyosin. The antibodies against the 2 contractile protein types only reacted with the homologous actomyosin; that is, they were type-specific.

When serial cryostat sections of human cervix were treated with FITC-labelled rabbit anti-UAM or anti-GAM precisely the same pattern of staining was displayed (Figs. 11, 12). There was thus no evidence of a species specificity in smooth-muscle actomyosins derived from the two sources.

It will be seen from Figs. 1 and 2 that when embryonic chick muscle or liver cells were rotated in Eagle’s MEM containing either anti-UAM or anti-PAM at a concen tration of 10% (v/v), progress in cell aggregation was inhibited by anti-UAM, but not by anti-PAM. The inhibitory effect was clearly expressed when the cells had been rotated for 2 h and was still exerted at the end of the 24-h experimental period. The rate of aggregation in the presence of anti-PAM was comparable with that of the con-trols rotated in medium containing 10% (v/v) of either calf serum or rabbit NIS.

Visual examination of aggregates formed in 24 h by liver cells in the presence of anti PAM (Fig. 15) showed that they were comparable in size and shape to those in the controls (Figs. 13, 14). However, liver cells treated with anti-UAM did not form large aggregates, though the many small aggregates present were rounded and compact (Fig. 16). A similar pattern of results was obtained using muscle cells, those treated with anti-PAM producing aggregates like those of the controls. The small aggregates formed in the presence of anti-GAM were, however, irregularly shaped and loosely constructed (compare Fig. 16). The pattern of results indicating the inhibitory effect of anti-UAM on progress in aggregation was qualitatively reproducible in pre-parations of both muscle and liver cells. In all experiments approximately 95 % of the cells were still viable at the end of the 24-h period, showing that the inhibitory effect of anti-UAM could not be correlated with reduced viability.

The inhibitory effect of anti-GAM at a concentration of 10% (v/v) on the aggregation rate of both embryonic chick liver (Fig. 3) and muscle cells was of the same order as that produced by anti-UAM (Figs. 1, 2). Antiserum directed against chicken PAM did not inhibit progress in aggregation. The aggregates produced at 24 h by the cells in the presence of anti-GAM were smaller than the calf serum- or NIS-treated controls and comparable in size to those formed by cells treated with anti-UAM serum (Fig. 16). Aggregates produced in the presence of anti-chicken PAM were similar in size and shape to the controls.

The level of efficiency of anti-GAM in inhibiting cell aggregation was retained when the concentration was decreased serially from 10% to 1 % (v/v) in Eagle’s MEM. At concentrations below 1 % the aggregation-inhibitory effect of the anti-GAM was markedly reduced. Specific rabbit antisera against α₁-acid glycoprotein, β₂-glyco protein 1, 7-sγ-globulin, 19-sγ-globulin and γ₁A-globulin at both 10% and 1% concentrations did not produce an inhibitory effect on the aggregation of liver cells. Rabbit anti-human serum generally failed to inhibit aggregation. However, 2 batches of serum at a concentration of 10% appeared to inhibit aggregation. It was estimated that in these cases about 20% of the cells were non-viable, which could explain the apparent inhibitory effect.

Globulin-enriched fractions precipitated with ammonium sulphate from rabbit antisera against chicken GAM and PAM and from NIS were added to cell suspensions to give a final concentration of 500 μg per 2 × 10⁸ cells/ml Eagle’s MEM, equivalent to 10% (v/v) antiserum. Anti-PAM globulins did not alter the rate of aggregation of embryonic chick liver or muscle cells compared with controls rotated in Eagle’s MEM containing globulin-enriched rabbit NIS (Figs. 4, 5). In these preparations the aggre gates produced by the end of the 24-h period were large in size and compact in form (Figs. 17–19, 21).

In contrast to these results the globulin-enriched fraction from anti-GAM serum markedly retarded the rate of aggregation of both liver and muscle cells (Figs. 4, 5). The inhibitory effect was evident within 2 h and persisted to the end of the experiment-tal period. The aggregates produced by this time were considerably smaller than the controls (Figs. 20, 22). Muscle-cell preparations treated with anti-GAM globulins differed from similarly treated liver cells in that the aggregates were irregularly shaped and loosely-formed whereas the liver aggregates were rounded and compact. The anti-GAM globulin fraction exerted the same degree of inhibition on aggregation when used at 50 μg/ml instead of 500 μg/ml. At concentrations below 50 μg/ml the aggregation-inhibitory effect of the globulin-enriched fractions was considerably reduced. It was estimated that cell viability in all preparations containing globulin fractions did not fall below 95 %

When muscle cells were rotated by the turbidimetric method in Hanks’s BSS containing 250μg/ml anti-GAM globulin fraction the decrease in optical density was markedly less than for preparations of the same cells containing anti-PAM globulin fraction, signifying that the anti-GAM reduced the rate of aggregation (Fig. 6). This effect was evident within 30 min of the start of the experiment and continued to the end of the 4-h experimental period. Aggregates formed at this time were comparatively small compared with the controls rotated in the presence of anti-PAM or NIS globu-lins. When anti-GAM globulins were introduced into preparations in which the cells had already been aggregating in Hanks’s BSS for either 1 or 2 h, they immediately reduced the rate of aggregation (Fig. 6). The size of the aggregates produced at the end of the 4-h period reflected the extent of the inhibition of aggregation.

The results of all these aggregation experiments with globulin fractions were quali-tatively reproducible.

The results of this investigation confirmed the preliminary finding that rabbit anti human UAM serum inhibits the aggregation of embryonic chick muscle and liver cells (Jones et al. 1970) and showed that anti-chicken gizzard smooth-muscle actomyosin (anti-GAM) is equally effective in inhibiting the aggregation of these same cells. Antisera against human or chicken striated muscle actomyosins and control sera did not produce this effect. In view of the possibility that the platelet-aggregation effect of anti-thrombosthenin serum (Chambers et al. 1967) was due to the presence of contaminants, it was important to dismiss the chance of similar substances being responsible for the inhibitory effect of anti-smooth-muscle actomyosin sera. The use of antibodies against actomyosin of both human and chicken origin rules out the possibility that the inhibitory effect of anti-GAM and anti-UAM was due to either contaminants specific to the chicken or those occurring only in humans. The inhibitory effect could not be due to antibodies formed from impurities common to all actomyosin preparations, because presumably these would have been present in human and chicken PAM as well as in UAM and GAM. It is also apparent from the results showing that antibodies against human serum proteins did not affect aggregation that the inhibition of aggregation is not a general property of antibodies. It is most unlikely that anti tropomyosin antibodies were present in anti-smooth-muscle actomyosin sera, because tropomyosin did not form a precipitin band on immunodiffusion against anti-GAM. The presence of anti-collagen and anti-glycoprotein antibodies in anti-GAM and anti UAM sera is also improbable because preliminary immunodiffusion studies showed that ground substance and anti-UAM or the reciprocal cross do not react. The foregoing considerations eliminate antibodies against impurities in the actomyosin preparations as causal factors in the aggregation-inhibitory effect of anti-GAM and anti-UAM.

It was previously demonstrated that anti-UAM and anti-PAM could react only with human smooth and striated muscle respectively, thus displaying type specificity and showing that the immune properties of the actomyosins from human smooth and striated muscles were different (Gröschel-Stewart & Gigli, 1968; Gröschel-Stewart, 1969). The present studies have shown that this same conclusion applies to chicken smooth- and striated-muscle actomyosins. The fact that anti-UAM and anti-GAM succeeded in inhibiting cell aggregation while anti-PAM did not provides a further distinct difference in the properties of antibodies prepared against smooth- and striated muscle actomyosins. The same finding also dictates that if the aggregation-inhibitory property of anti-smooth-muscle actomyosin sera was a manifestation of the formation of an antibody-antigen complex, then, because anti-PAM failed to produce the in-hibitory effect, the antigenic site could not have been an actomyosin of the type found in striated muscle.

The immunofluorescent results indicate that the anti-GAM and anti-UAM reacted only at the cell surface. The surfaces of human fibrocytes and platelets were stained specifically with FITC-labelled anti-UAM (Figs. 7, 8). Gröschel-Stewart et al. (1970) have detected fluorescent staining with labelled anti-GAM at the surfaces of embryonic chick muscle and liver cells. Support for these findings can be derived from similar studies showing FITC-labelled anti-actomyosin serum at the surfaces of adult rat liver cells (Neifakh & Vasilets, 1964, 1965; Vasilets, 1964; Vasilets & Zubzhitskii, 1966; Neifakh et al. 1965) and human platelets (Becker & Nachman, 1969). Thus it is likely that the antigenic site with which the anti-UAM and anti-GAM reacted to inhibit aggregation is at the cell surface. Since anti-UAM serum will react with the Ca²⁺-dependent myosin ATPase of smooth-muscle actomyosin (Gröschel-Stewart, 1968), it follows that the antigenic site is the heavy meromyosin of a smooth muscle-type actomyosin.

There is good evidence for believing that the reaction of anti-GAM and anti-UAM with the Ca²⁺-ATPase of smooth muscle myosin is highly specific and not part of general reaction with ATPases. First, Gröschel-Stewart (1968) showed that anti-UAM does not inhibit the activity of the Na+, K+, Mg²⁺-ATPase system of human erythron-cyte ghosts. Secondly, it has been found that anti-human UAM does not block the Ca²+-ATPase of striated-muscle actomyosin (Gröschel-Stewart, 1968). Thirdly, there are strong indications that smooth-muscle antibodies do not react with the calcium transport Ca²⁺, Mg²⁺-ATPase. It has been shown that although erythrocyte ghosts possess a Ca²⁺-transport ATPase (Dunham & Glynn, 1961; Schatzmann, 1966, 1967; Wins & Schoffeniels, 1966; Wins, 1969, 1970), anti-UAM does not react with the erythrocyte membrane (Gröschel-Stewart, 1968).

It is not easy to provide a reliable estimate of the number of antigenic actomyosin sites at the surface of dissociated embryonic chick cells because actomyosins are soluble at high ionic strength (μ 0·6), which hampers determining antibody titre by conven-tional methods. However, knowing that 50 μg globulin-enriched protein maximally inhibited the aggregation of 2 × 10⁸ liver cells, and supposing that 20 % of the globulin enriched fraction was immunoglobulin, it is possible to provide an approximate figure of 10⁶ to 10⁷ antigenic sites/cell surface.

The fact that anti-UAM blocked the activity of the Ca²⁺-dependent ATPase of smooth-muscle myosin (Gröschel-Stewart, 1968) and inhibits the aggregation of embryonic chick cells is consistent with the view that cell adhesion involves the partici-pation of surface-localized actomyosin (Jones, 1966, 1967). A functional actomyosin system requires an energy supply derived from ATP exclusively hydrolysed by the myosin ATPase (Weber, 1958). Specific inhibition of the activity of this ATPase by anti-UAM or anti-GAM would prevent the protein performing its contractile function, thereby rendering the adhesion mechanism inoperative. If the actomyosin like protein controlled the breaking of adhesive bonds in the same manner as their establishment (Jones & Morrison, 1969), antibodies capable of reacting with this protein system should not disperse aggregates. This reasoning was shown to be valid, for cell aggregates did not break up when treated with globulin-enriched anti-GAM.

Although the nature of the adhesive mechanism has not been established, the pro posal that it involves an actomyosin-like protein capable of controlling cell adhesion seems to provide a reasonable explanation of the aggregation-inhibitory effect of anti-UAM and anti-GAM sera.

We wish to thank Mrs Barbara Jones, Miss Barbara Morris and Mrs Annette Ashby for technical assistance and Mr R. Moore for processing the photomicrographs. This investigation has been aided by generous grants provided by the Science Research Council and Deutsche Forschungsgemeinschaft.