A new general type of morphogenetic process has been revealed by experiments on the phenomenon of non-coalescence between different strain types (alpha and delta) in the sponge Ephydatia fluviatilis. The question investigated was whether any process of cell adhesion was responsible for the phenomenon. Preliminary results suggested that the cells might show specific adhesion but further results indicated that a more complex system existed. Each strain produces a soluble factor that increases the adhesiveness of homologous cells but decreases that of cells of heterologous strains. The adhesion of cells, even in the presence of these factors, is non-specific but the factors specifically control adhesion and determine its quantitative value. The adsorption of the factors to the cells was tested for with inconclusive results. Heterologous factors may irreversibly alter a cell’s adhesiveness. It is shown that this system, particularly by reason of its negative effect on adhesiveness, (a) accounts adequately for the phenomenon of non-coalescence, (b) provides a model system for many forms of morphogenesis and (c) allows many apparently contradictory results obtained by other workers to be reconciled.

The phenomenon of non-coalescence of the tissues of differing species or of differing strains of the same species of sponge discovered by van de Vyver (1970) appears to form a remarkable system of morphogenesis. When the tissues of two individuals of the sponge Ephydatia fluviatilis are brought into contact by experimental means or when they come into contact through natural processes, they form an adhesion in the zone of contact. If the tissues are of the same strain type the pinacoderms become confluent, and later the canal systems of the two sponges join up and the sponges become effectively one organism. If, however, the two sponges are of different strain type the initial adhesion persists for a time, after which the two sponge bodies separate from one another leaving a gap between them. Generally the two sponge bodies move apart. This behaviour has been termed non-coalescence by van de Vyver (1970). She has suggested that this behaviour may be due to a specificity of adhesion amongst the cells such that cells from different strains cannot stick to one another. The first aim of the work described here was to test whether or not those strains which show non-coalescence also show specificity of adhesion. This test was carried out using the collision efficiency method (Curtis, 1970a, b). During this work results were obtained and independently further information reported by van de Vyver (1971 b), which now suggest to us an alternative mechanism as the basis of non-coalescence. Consequently further experiments to test this second mechanism are described here.

Sponges of the alpha and delta strains of Ephydatia fluviatilis were grown from gemmules supplied by Professor Rasmont’s laboratory (Université Libre de Bruxelles). The sponges were cultivated in Rasmont medium (Rasmont, 1961) for 1–2 weeks before use, at ca. 20 °C. Each sponge was grown from ten gemmules and contained about 1 × 105 cells. Cell suspensions were prepared by scraping the sponge bodies off the culture-dish surface and then transferring the sponges to the dispersion medium. Each 10−3m3 (approx. 1 litre) of this medium contained 2·5×10−4 mol. EDTA, 0·034 mol. NaCl, 0·00134 mol. KC1, 0·00138 mol. glucose, 0·00107 mol. NaHCO3, 7×10−4mol. K2HPO4 and was buffered to pH 7·8 with 0·006 mol. 2-amino-2-hydroxymethyl, 1·3 propanediol (Tris). After exposing the sponges to this medium at ca. 20 °C for 7 min they were mechanically dispersed in the same medium by pipetting. From this point two different methods of cell dispersion were used, namely the ‘washed cell’ technique and a second technique in which an appreciable portion of the disaggregation medium is included in the final cell suspension. In the washed cell method the cell suspensions were then centrifuged at ca. 200 g for 5 min, the supernatant discarded and the pellet resuspended in Rasmont medium. In the second technique the cells were dispersed in a very small volume of EDTA medium and the suspension was then made up with a large volume of Rasmont medium. Since a small amount of EDTA would be carried over into the cell suspension by this procedure and because of the possible existence of soluble factors affecting cell adhesion (see later) the volumes of medium used and removed at each stage were made standard for all but the preliminary experiments, in which they were roughly standard. The EDTA level in the final cell suspension was ca. 1 × 10−4mol. (10−3 m3)-1 compared with a Ca + Mg concentration of 1·5 ×10−3 mol. (10−3 m3)-1. The dispersion technique used in each separate experiment is stated in the appropriate part of the text. The adhesivenesses of the cells (collision efficiencies) were measured with a Couette viscometer, using the method described by Curtis (1969, 1970a, b, c). Measurements were made at shear rates between 8 and 9 sec-1 and at room temperature (20–25 °C). Each value for a collision efficiency quoted in the Results section is the mean of three sets of measurements each made for five different periods of aggregation. Counts of cell and aggregate population densities were made using haemocytometers.

Alpha and delta factors were prepared from alpha and delta sponges respectively by a modification of the method described by van de Vyver (1971 b). A precise number of gemmules (usually 30 sponges from 300 gemmules) were dispersed in the EDTA medium used for cell dispersal after soaking in this medium for 7 min. The cell suspensions were then exposed to ultrasound (100 W at 20 kHz for 2 min). The gemmule cases were removed, and the suspension of cells and broken cells concentrated and dialysed by vacuum dialysis against normal Rasmont medium. The dialysis membrane used had an effective permeation limit for molecules of mass 1000 daltons. The dialysed material was filtered through a Millipore HA filter and made up in a volume of 3 ml. The factors were stored under refrigeration and used within 2 days of preparation. A subsidiary experiment showed that this method of preparation of the factors yielded a product apparently identical with those prepared by the original method (van de Vyver, 1971 a).

Specificity of adhesion was tested by the method described by Curtis (1970a, b). In this test the collision efficiency, E (written as a in Curtis, 1969), of the two cell types is measured for each type separately and then for mixtures of various proportions of the two types. If there is no specificity of adhesion the collision efficiency for any proportion of the two cell types will be the average of the efficiencies for both types (measured separately) weighted for the proportion of each cell type in the mixture. If, however, there is complete specificity of adhesion such that cells derived from different strains will not adhere, then the measured collision efficiency will be reduced because all collisions between unlike cells will fail to result in adhesions. The full theory of this test is given by Curtis (1970b). In summary, if there is no specificity of adhesion, the collision efficiency for a mixture of two types of cells of adhesivenesses E1 and E2 mixed in proportion and n1 is given by n2E2+ n2E2. If there is complete specificity of adhesion the equivalent relationship is . If specificity is incomplete the plots of collision efficiency against proportion of a cell type in the mixture will lie between those for complete specificity and complete lack of specific adhesion. The difference in collision efficiency for any two degrees of specificity is at a maximum for a 50:50 mixture of the two cell types. Appropriate statistical tests for the treatment of experimental data have been published (Curtis, 1970 a, b) and these are used in this work.

Tests were made, using the collision efficiency method, to discover whether specificity of adhesion occurred for cells of the alpha and delta strains. The results are given in Table 1 and Fig. 1. The results suggest that specific adhesion is found. The second method of cell dispersion was used.

Table 1.

Test for specificity of adhesion of alpha and delta strain cells (see also Fig. 1)

Test for specificity of adhesion of alpha and delta strain cells (see also Fig. 1)
Test for specificity of adhesion of alpha and delta strain cells (see also Fig. 1)
Fig. 1.

Test of specificity of adhesion for mixtures of alpha and delta strain cells. Experimental results are shown as means (○) with standard deviations. The straight line represents results that will be expected on the hypothesis of non-specific adhesion. The continuous curve represents the results that will be obtained on the hypothesis of specific adhesion.

Fig. 1.

Test of specificity of adhesion for mixtures of alpha and delta strain cells. Experimental results are shown as means (○) with standard deviations. The straight line represents results that will be expected on the hypothesis of non-specific adhesion. The continuous curve represents the results that will be obtained on the hypothesis of specific adhesion.

However, the discovery of factors which promote aggregation, made by van de Vyver (1971 a), suggests another explanation of these results. An alter native hypothesis which would account for the reduced adhesiveness of mixtures of the cells of two strain types is that these factors promote the adhesiveness of their own strain but diminish the adhesiveness of other strains. The method of preparation of these factors is very similar to that used for the preparation of cell suspensions; indeed van de Vyver used exactly the same method for factor preparation in her work as used here for cell suspension preparation. Therefore it is probable that these factors are present in the cell suspensions unless the cells are very thoroughly washed. If this hypothesis is correct the especial circumstances of the test for specific adhesion would lead to its simulation when two equal aliquots of suspensions of cells of unlike type are mixed. Before mixing, each suspension separately contained a concentration of the aggregation factor sufficient to maintain a value for adhesiveness E. On mixing with an equal aliquot of the other cell suspension the concentrations of both factors would be halved so that the adhesiveness of each cell type would be reduced (the exact extent of reduction depends upon the nature of the doseresponse curve); moreover each cell type would experience a great increase in the concentration of the factor derived from its opposite cell type, further reducing its adhesiveness. In consequence, although the cells do not show specificity of adhesion the collision efficiency of the mixture is reduced and thus the existence of specific adhesion might be deduced from the results incorrectly. The general hypothesis will be termed the factor-specificity theory. It is of course possible that the above system might exist with specific adhesion occurring as well. Either in such a case or if the dose-response curves for the factors are very steep it would be possible to obtain a greater reduction of adhesiveness for the mixed suspensions than is predicted by the simple theory.

We felt that in order to be sure that the actual mechanism of adhesion is specific it was necessary to show that the above explanation of specificity in terms of these factors was incorrect. One simple postulate of the factor specificity hypothesis is that the delta factor, for example, would increase the adhesiveness of delta cells and reduce that of alpha cells as the concentration of factor was increased. An alpha factor would operate in a converse manner. A second postulate is that the dilution of a cell suspension with a bland medium will reduce the adhesiveness of the cells because the concentration of the aggregation promoting factor will be reduced. The result of this test may show also that the aggregation factor is normally present in a cell suspension. These two tests are described in the next section. However, the results of these tests, even if positive, will only show that the necessary components of a system of factor-specificity exist, not that the system operates amongst the cells. A third, more complex, test was designed to test this latter point, but it is described later because it is best understood in the light of the results of the first two tests.

Effect of aggregation factors on cell adhesiveness

In the first test alpha and delta factors were added to ‘washed’ suspensions of alpha and delta cells within 1 min of the start of reaggregation. In the absence of any knowledge of the chemical nature of these factors an arbitrary concentration scale was used. Concentrations are expressed in terms of the number of sponge cells disrupted per cc (10−6 m3) of medium ×10−5. Thus a 30-unit solution would contain the yield of 3 × 106 cells/cc. The effect of alpha and delta factors on the aggregation and adhesiveness of both alpha and delta cells is shown in Table 2 and Fig. 2.

Table 2.

Dose-response curve for the adhesiveness (collision efficiency) means E and standard deviations σ of alpha and delta strain cells in the presence of homologous and heterologous factors

Dose-response curve for the adhesiveness (collision efficiency) means E and standard deviations σ of alpha and delta strain cells in the presence of homologous and heterologous factors
Dose-response curve for the adhesiveness (collision efficiency) means E and standard deviations σ of alpha and delta strain cells in the presence of homologous and heterologous factors
Fig. 2.

Dose-response curves for the adhesiveness of (a) alpha strain and (b) delta strain cells in the presence of alpha and delta factors. The points are mean values from experimental observations. See also Table 2.

Fig. 2.

Dose-response curves for the adhesiveness of (a) alpha strain and (b) delta strain cells in the presence of alpha and delta factors. The points are mean values from experimental observations. See also Table 2.

These results confirm the existence of the alpha and delta factors postulated by van de Vyver and additionally show that the factors diminish the adhesiveness of cells of the opposite strain type. It should be appreciated that this result with this test depends upon the accidental or intelligent choice of a suitable level of factors in terms of the dose-response curve.

The second test examines whether dilution of a cell suspension has any effect on its adhesiveness. If an effect is found it can be concluded that an aggregation promoting factor normally present in the cell suspension has been diluted. This test of course depends upon the accidental or intelligent choice of two suitably placed points on a dose-response curve. Cell suspensions of both alpha and delta cells were separately made up at high population densities (l·7x106 cells/10−6m3) by method 2. An aliquot of each suspension was then placed in the Couette viscometer for collision efficiency measurement; another aliquot was diluted with an equal volume of Rasmont medium and then the collision efficiency was measured. The adhesivenesses of concentrated and diluted suspensions are shown in Table 3. It is clear that dilution of both alpha and delta cell suspensions diminish their adhesiveness.

Table 3.

Dilution experiment

Dilution experiment
Dilution experiment

These results suggest that it is at least possible that the appearance of specific adhesion reported in the ‘Preliminary Results’ is due to the presence of the alpha and delta factors. In those experiments where a delta strain cell suspension is mixed with an alpha suspension, the adhesiveness of the delta cells will fall because of the dilution of delta factor, and because of the presence of the alpha factor; similarly the adhesiveness of the alpha cells will fall because of the presence of the delta and dilution of the alpha factors. The main question to be decided is whether the extent of the reduction of adhesiveness observed in the preliminary results can be entirely accounted for by the presence of the two factors together in mixed suspensions. If the reduction in adhesiveness is greater than can be ascribed to the presence of the factors, it is possible that the hypothesis of specific cell adhesiveness should be used to explain the results. Examination of the dose-response curves and the data given in the section on Preliminary Results suggest that the alpha and delta suspensions in those experiments contained respectively 0-5 units/cc and 0-8 units/cc of their factors. Mixing equal volumes of these two suspensions would halve these concentrations in the mixed suspension. Assuming that the collision efficiency of cells in the simultaneous presence of both factors is the mean of the efficiencies measured in the separate presence of each factor, the collision efficiency of each strain type in the mixture can be calculated. In turn the collision efficiency of the mixed cells can be determined and the question of the specificity of adhesion resolved.

The effect of the simultaneous presence of both factors on the adhesiveness of each strain type was measured using well-washed cells (method 1), to which varying concentrations of alpha or delta factors were added. The results are shown in Table 4. They indicate that the collision efficiency of the cells in the presence of both factors is slightly lower than a value obtained by taking the mean of the values found for the separate presence of each factor at the same concentrations.

Table 4.

Collision efficiency in the simultaneous presence of homo- and heterologous factors (means E and standard deviations σ)

Collision efficiency in the simultaneous presence of homo- and heterologous factors (means E and standard deviations σ)
Collision efficiency in the simultaneous presence of homo- and heterologous factors (means E and standard deviations σ)

We can now calculate the collision efficiencies that would be expected in the mixed systems described in the Preliminary Results from the operation of the alpha-delta factor system. The two pure cell strain suspensions contained respectively 0·5 units/cc and 0·8 units/cc of their respective factors. Taking into account the results shown in Table 4 and the dose-response curves, the collision efficiency of the alpha cells in a 50:50 mixed suspension would be 1·35 % and that of the delta cells 1·40%. This would give a mean value of 1·37 % in the mixed suspension assuming that there is no specificity of adhesion. This is close to the value determined experimentally of 1·69 % for a mixed suspension. If there was complete specificity of adhesion in the presence of the alpha-delta factor system the collision efficiency would be 0·68 % because specificity of adhesion would further lower the apparent adhesiveness of the cells in the mixture. Since this value lies outside the 99 % confidence limit (lower value 1·13 % ; 1-tailed test) of the measured value there is no reason for concluding that specific adhesion occurs in this system.

It might be thought at first sight that the alpha-delta factor system is in effect a mechanism for producing specific adhesion amongst cells which do not possess it when they are extensively washed. However, it is clear that much of the effectiveness of the system arises from its ability to diminish the adhesiveness of cells of unlike strain types. This is not a characteristic of any system of specific adhesion. In addition it is clear that from the experimental data that the kinetics of aggregation are such that adhesions must form indiscriminately between cells of different strain type. The alpha-delta factor system is, however, a system in which there is a strain-specificity in the control of adhesion. In the Discussion we shall consider other experiments which can be reinterpreted in the knowledge we now have of the alpha-delta factor system. In some of those experiments evidence has been put forward for specific mechanisms of cell adhesion involving the existence of a specific cementing factor. One very simple experiment to test for the action of a cementing factor is to show that the factor is bound to cells prior to or during its action in promoting adhesion. Curiously this test does not appear to have been carried out on any system of cell adhesion by any previous worker. Since we are interested in the manner in which the alpha-delta factor system operates and in establishing any similarities with or differences from systems of specific adhesion, we carried out tests to establish whether these factors are adsorbed to the cells. It is clear from the dose-response curves that it is possible to choose concentrations of factor that are below levels that might saturate any binding sites that may exist on the cells. Cell suspensions prepared by method 1 (well-washed cells) were treated with non-saturating levels of heterologous factor. After aggregation the suspensions were centrifuged and the medium recovered. This medium was used to treat a second batch of cells of the heterologous strain during their aggregation. (It is undesirable to carry out the test on cells of a homologous strain type because homologous factor may be added to the system by the cells.) If the apparent activity of the factor, as shown by its effect on the adhesiveness of the second batch of cells, is reduced by exposure to the first batch of cells it would seem probable that the factor is adsorbed to the cells. The results are given in Table 5. They give no support to the hypothesis that the factors act by adsorption to the cells, though this cannot be excluded (see Discussion).

Table 5.

Factor absorption experiment

Factor absorption experiment
Factor absorption experiment

Such a result suggests that the factors are not adsorbed to the cells, and hence cannot form a part of a cell binding material. The results are consistent with the factors being enzymic in their action. Related to this point is the question of whether the effect of each factor is reversed by the opposite factor or otherwise.

The reversal test was carried out by two methods : in the first, well-washed cells were treated with factor of the opposite type, and their adhesiveness was measured over a 30-min period, then the cells were centrifuged out of the medium and resuspended in a medium containing homologous factor, before measurement of their adhesiveness. The second method of performing this experiment is to treat well-washed cells with homologous factor first and then with heterologous factor, after removal of the homologous factor by centrifugation. This method of carrying out the experiment is less satisfactory than the first technique because there is appreciable cell adhesion during the treatment with homologous factor. In consequence, the measurement of collision efficiencies during the second treatment is less accurate because the method of measurement presumes that at the start of aggregation the cell population is monodisperse. However, the second method of carrying out this experiment was used because there may be differences in the result depending on whether the cells are exposed first to homologous or first to heterologous factors. Results are given in Table 6. The values that would be expected in the absence of a pretreatment step can be read in Table 2. The results show that the effect of treatment by a homologous factor is reversible by a factor of the opposite type, but that the reverse is not the case. This suggests that an active site involved in cell adhesion is irreversibly altered by the heterologous factor so that a subsequent treatment with homologous factor is ineffective.

Table 6.

‘Reversed treatments’ - sequential treatment with homologous and heterologous factors

‘Reversed treatments’ - sequential treatment with homologous and heterologous factors
‘Reversed treatments’ - sequential treatment with homologous and heterologous factors

The experiments described in this paper start with the apparent demonstration of the specific adhesion of the cells of the alpha and delta strains of Ephydatia fluviatilis. The earlier experiments were carried out before the discovery by van de Vyver (1971 b) of the soluble factors, released during dispersion of the sponge tissues, which promote the aggregation of cells of the same strain type as the factor. The existence of these factors suggested that the preliminary experiments could be reinterpreted as evidence for a system in which strain differences in the response of the cells to the factors would lead to results simulating specific adhesion. The experiments reported in the main section of results show that homologous factors increase and heterologous factors decrease the adhesiveness of the cells. They also show that the adhesiveness of cells in the presence of both homo- and heterologous factors is the mean of the values that would be expected in the separate presence of each factor. Most important of all, the results show that the extent of adhesion reduction in mixed alpha-delta suspensions is quantitatively predicted with accuracy from the concentrations of alpha and delta factors in the mixed suspension, on the assumption that the cells show no specificity of adhesion. In other words the factor specificity system accounts for the behaviour, reported in the Preliminary Results, which simulates specific adhesion.

At this point it is perhaps appropriate to consider the criticism of the test system for specific adhesion, used in this work, made by Humphreys (1970). Humphreys suggested that Curtis (1970 c) had not shown that the rate of aggregation has a complete dependence on cell population density and therefore that the interpretation of the experiments might be incorrect. However, it is implicit in the Swift & Friedlander (1964) treatment, which is the basis of the method of measuring cell adhesiveness introduced by Curtis (1969), that the kinetics of aggregation of any two size classes will follow second order kinetics. (The aggregation rate integrated over all size classes will show first order kinetics.) If the cells did not aggregate in this way very large standard deviations would be found in the measurements of collision efficiency over a time course during which the total particle concentration falls to about a half of the initial value. The actual measured standard deviations are small; this suggests that the aggregation kinetics for one or two size classes are second order. It might at first sight be felt that the behaviour shown in Table 3, where dilution of a cell suspension diminishes the collision efficiency, is in fact direct proof that second-order kinetics apply for any two size classes, but it should be borne in mind that in calculating the collision efficiency by the Swift and Friedlander treatment a dimensionless value is obtained, in other words, differences in cell concentration are removed. Humphreys also suggested that the aggregation rate actually measured might not be a true measure of adhesiveness, because the rate-limiting step might be the rate of recovery of the cells from dispersion procedures. If the cells recover their adhesiveness slowly the rate of aggregation would be determined by this process. Again it is clear that if this occurred, collision efficiencies would rise during the course of aggregation and in consequence large standard deviations would be found in the values derived from measurements over the time course of the experiments. The constancy of collision efficiency values found for any defined set of conditions in this work argues against the hypothesis that the measured aggregation rate is limited by a recovery of cellular adhesiveness. This criticism cannot of course be applied to those experiments where heterologous factors diminish cell adhesiveness.

The next question to be considered is whether the system revealed in this work can be used to account for the non-coalescence and the sorting out described by van de Vyver (1971 a, b). It appears from her work that when two sponges of unlike type make contact the pinacoderms of each sponge form an adhesion. The cells of each individual do not interpenetrate. Some 15-30 h later (depending on the strain types involved) the adhesion of the sponges breaks down and they separate. It is possible that contractions in each sponge body help to pull the sponges apart. When two homologous sponges make contact an adhesion develops between pinacoderms. Later the pinacoderm cells migrate to the surface of the fused sponge and the cells from each individual interpenetrate to a considerable extent. It should be appreciated that if this phenomenon is explained on the hypothesis that the cells show complete specificity of adhesion (cf. Humphreys’ results on Microciona and Haliclona - Humphreys, 1963) it would be expected that two sponges of unlike type would not even form a temporary zone of adhesion.

A more satisfactory explanation of the major features of non-coalescence can be developed in terms of the idea that the alpha and delta factors control this phenomenon. The concentration of a sponge’s own factor will tend to be maximal at the centre of the sponge body if it is assumed that all cells of the sponge produce the substance. The concentration will be minimal at the periphery of the sponge (under a wide variety of assumptions about the origin, diffusion and destruction of the factor). Therefore the adhesiveness of the cells of the surface of the sponge body will be low but not negligible. When two sponges of like type make contact an adhesion will form. Since the factors are of diffusible nature and since they may act without being irreversibly bound to the cells, the concentration of the factors will rise in the region of contact between the sponge bodies. As the homologous factors increase in concentration the cells will become more adhesive and a single permanent sponge body will be established.

When two sponges of unlike type make contact an adhesion will form initially because the concentrations of the two factors will still be low at the original surface of the sponges along the zone of contact. However, because of the changed geometry of the sponges the concentrations of the two factors will rise along the region of contact. Though the measurements of the simultaneous effect of hetero- and homologous factors on the adhesiveness of cells show a tendency for the two factors to cancel out (Table 4), it is clear from the sequential treatment experiments (Table 6) that the heterologous factor will irreversibly alter the adhesiveness of the cells (unlike the homologous factor) so that eventually the cells exposed to heterologous factor will become non-adhesive. We can view the process as a competition between homo- and heterologous factors, the heterologous factor being a competitive inhibitor. During a short-term experiment (30 min) the progressive inactivation of adhesion is not obvious. The diminution in adhesion will be maximal for the cells lying at the zone of contact since they are of opposite type on either side. Thus the sponges will cease to adhere to each other some time after the initial contact is made.

It is possible that the cells in the original zone of contact might become so non-adhesive that they would separate from the sponge bodies and migrate or drift to other regions ; in this way a gap would as it were be eroded between the sponges. Though the phenomenon of non-coalescence of unlike strain types can be adequately explained on the hypothesis advanced in the last paragraph it is clear from van de Vyver’s results (van de Vyver, 1971 b) that there are more complex and perhaps secondary features of the phenomenon that are less easy to explain. When contact is made between homologous sponges the pinacoderm cells in the region of contact eventually migrate to the free surface of the sponge, thus allowing the inner regions of the two sponges to unite. In heterologous contacts the pinacoderms remain intact in the region of contact.

The phenomenon of non-coalescence has certain similarities to that of contact inhibition of movement of cells (Abercrombie & Heaysman, 1954). In both phenomena cells make contact and form adhesions with each other and subsequently re-separate. However, contact inhibition of movement is displayed between homologous and heterologous cells whereas non-coalescence only occurs between unlike cells. Moreover Abercrombie & Gitlin (1965) have shown that it is unlikely that any diffusible agent is involved in contact inhibition of movement. There remains the considerable, if superficial, resemblance between the two phenomena in that an outgrowth of cells stops movement on contact with another body of cells. Contact inhibition of movement has been suggested as being the phenomenon which prevents cells of normal tissues from invading one another (Abercrombie, 1967). However, it can now be seen that the strain-specific adhesion factors discovered in this work would, if they acted as tissue specific adhesion factors, prevent tissues from invading one another. Moreover such a system would have two further capabilities of great importance in morphogenesis. First, during morphogenesis, tissue factors of this type could interact to diminish the adhesiveness of cells in the region of contact of two cell types. As a result, gaps and cavities between tissues could be formed. Second, such systems for the control of adhesion would act to ensure that cavities and gaps between tissues would persist in adult life. It should be remembered that the vertebrate body develops many cavities within it during development and that many organs - for example, the liver - lie with the majority of their surface free and unadherent to the surrounding peritoneum even though there may be prolonged contact between the two tissues. In a special sense these factors can be regarded as being the morphogens postulated by Edelstein (1970), their action being, however, not chemotactic or chemokinetic but as controllers of adhesion. It is important to emphasise that it is the soluble and therefore presumably diffusible nature of these factors that makes them of particular utility in explaining morphogenesis.

These factors, if they diffuse from the cells that produce them, could act as a means of providing positional information and determination of pattern. In general terms, such a system would be analogous to that proposed by Wolpert (1969), with the distinction that these factors would control the positioning of cells rather than the differentiation of cells in fixed positions.

When the results of this work are compared with those obtained on the reaggregation of other species of sponge, in particular by Humphreys (1963), Sara (1965), Sara, Liaci & Melone (1966), Humphreys (1970) and Curtis ( 1970a, c), it is clear that it is now possible to reconcile a number of differing interpretations. Humphreys (1963) identified a factor which specifically promoted the adhesion of cells of the species type from which the factor was derived. He appears to have carried out no test to discover whether the factor diminished the adhesiveness of other species types. Such factors were discovered for the species Haliclona occulata, Microciona proliféra, Halichondria panicea and Clione celata. Humphreys suggested that these factors acted as cements which would attach to binding sites on two apposed cell surfaces, thus binding the cells together. No evidence was put forward that these factors were adsorbed by cells. In order to demonstrate the activity of the factors the cells had to be very thoroughly washed to reduce their adhesiveness for control experiments. On adding the factors, specific promotion of adhesion could be detected. Curtis (1970c) repeated part of Humphreys’ work using the two species Haliclona occulata and Halichondria panicea. The cells were very thoroughly washed and no evidence for specific adhesion could be detected from measurements of the collision efficiency of mixed suspensions. It can now be seen that Curtis’s experiments can be reinterpreted as showing that the cells, in the absence of any factor, show non-specific adhesion. Sara (1965) could find no evidence for specific adhesion in quite a wide range of marine sponges. Unfortunately his papers give few details of how the cell suspensions were prepared. Curtis (1970cz) also investigated whether a variety of other marine sponge cells showed specific adhesion. In these experiments the cell suspensions were well washed and no evidence for specific cell adhesion was found. Curtis’s and Humphreys’ separate results can be reinterpreted in the following way as a consequence of the hypothesis advanced above. We shall assume that the factors discovered in E. fluvial His strains have species-specific counterparts in marine sponges. Curtis’s failure to detect specificity of adhesion or specific control of adhesion would arise from the fact that his cell suspensions were washed clean of the factors. Humphreys diminished the adhesiveness of the cells by the method of cell dispersion and by aggregating the cells at low temperatures. If we assume that the factors he isolated were identical in general behaviour with those discovered in E. fluviatilis they would act not as agents of specific adhesion but as substances specifically controlling the appearance or loss of an adhesion whose direct mechanism would be unspecific. In the experimental design used by Humphreys such factors of the Ephydatia type would give exactly the same results as he obtained. In detail the aggregation of cells of a heterologous type would be prevented and adhesion of cells of a homologous type promoted in the presence of a single factor. When two species types are aggregated in the presence of both factors it would be expected that the first adhesions would be random with respect to the species types joining together. Later as small groups of cells of one species type began to produce appreciable additional amounts of their factor they would make their local environment less favourable for interspecific adhesions and more favourable for adhesions with their own species type; in this way sorting out of the species would occur. This would account for the fact that it is frequently seen that aggregates involved in sorting-out ‘expel’ cells (Curtis, 1967). Thus apparently contradictory results on cell adhesion can be reconciled by the hypothesis that there are diffusible factors which specifically control the adhesiveness of cells whose mechanism of adhesion is itself unspecific.

The mode of action of these factors has considerable implications for the explanation of the mechanism of adhesion of Ephydatia cells. If the factors are adsorbed by the cells it is, at least in principle, possible that they may act as binding agents (cements). If they form binding materials they must have at least two binding sites, one to attach to each cell surface. An alternative explanation of the activity of the factors is that they modify a cell surface component involved in adhesion by enzymic action. The failure of the adsorption test to detect adsorption of these factors cannot be regarded as anything more than negative evidence because if the number of binding sites were small and the binding constant (stability constant) high a very small quantity of factor would be removed from a given initial concentration of the factor on each successive exposure to cells. Consequently, since it is improbable that the test would detect a small reduction in the activity of the factor from one exposure to cells to another, no clear evidence about adsorption would be obtained. Interestingly enough other workers (e.g. Humphreys, 1963 ; Moscona, 1968) who have claimed to isolate specific cell-binding materials, do not appear to have tested whether their factors are bound by cell surfaces. The results obtained when cells of either strain are treated sequentially with both factors are explicable both on the theory that the factors attack a site involved in adhesion enzymically or that they adsorb irreversibly to a cell surface binding site. However, if the factors are adsorbed to the cell surface it is improbable that they act as cements for the following reason. It is clear that heterologous factors cannot act as binding agents because they decrease cellular adhesiveness, though they might block binding sites. The homologous factors increase cell adhesion but do not do so specifically, yet if they acted as binding agents they would produce specific adhesion, because they clearly could not bind to heterologous cells. Hence it seems unlikely that these factors act as specific binding agents, though we cannot yet conclude whether their action is enzymic or whether they modify some general method of cell adhesion after surface adsorption.

Une nouvelle forme de processus morphogénétique a été mise en evidence par l’étude du phénomène de non-confluence entre deux souches (Alpha et Delta) de l’éponge d’eau douce E. fluviatilis.

Ce travail a pour but de verifier si le phénomène de non-confluence est en relation avec les mécanismes de l’adhérence cellulaire.

Les résultats préliminaires suggéraient une spécificité de l’adhérence cellulaire mais des experiences ultérieures ont montré qu’il existait un système plus complexe. Chaque souche produit un facteur soluble qui augmente l’adhérence des cellules homologues et réduit celle des cellules heterologues. L’adhérence des cellules, même en présence des facteurs n’est pas spécifique, mais les facteurs contrôlent spécifiquement l’adhérence et determine sa valeur quantitative.

L’absorption des facteurs par les cellules a été testée mais sans résultats. Les facteurs heterologues peuvent altérer l’adhérence cellularie de manière irreversible.

Le système mis en evidence, notamment par son action negative sur l’adhérence cellulaire (a) explique le phénomène de non-confluence, (b) fournit un modèle pour diverses formes de morphogénèse et (c) permet de concilier des résultats apparemment contradictaires obtenus par différents auteurs.

We should like to express our thanks to Professor R. Rasmont for the supply of sponge gemmules, to Miss Rose McKinney for her expert assistance and to Mr Graeme Ferguson for ‘planting’ gemmules. We are most grateful to the Science Research Council for a grant to one of us (A. S. G. Curtis), which provided much of the support for this project, and to the British Council and Le Ministère de 1‘Education Nationale et de la Culture de Belge who provided travel funds. Additional support was provided by the University of Glasgow and by the Université Libre de Bruxelles.

Abercrombie
,
M.
(
1967
).
Contact inhibition: the phenomenon and its biological implications
.
Natn. Cancer. Inst. Monog.
26
,
249
277
.
Abercrombie
,
M.
&
Gitlin
,
G.
(
1965
).
The locomotory behaviour of small groups of fibroblasts
.
Proc. Roy. Soc. Lond. B
162
,
289
302
.
Abercrombie
,
M.
&
Heaysman
,
J. E. M.
(
1954
).
Observations on the social behaviour of cells in tissue culture. H. ‘Monolayering’ of fibroblasts
.
Expl Cell Res.
6
,
293
306
.
Curtis
,
A. S. G.
(
1967
).
The Cell Surface.
London
:
Logos Press, Academic Press
.
Curtis
,
A. S. G.
(
1969
).
The measurement of cell adhesiveness by an absolute method
.
J. Embryol. exp. Morph.
22
,
305
325
.
Curtis
,
A. S. G.
(
1970a
).
Problems and some solutions in the study of cellular aggregation
.
Symp. Zool. Soc. Lond.
25
,
335
352
.
Curtis
,
A. S. G.
(
1970b
).
On the occurrence of specific adhesion between cells
.
J. Embryol. exp. Morph.
23
,
253
272
.
Curtis
,
A. S. G.
(
1970c
).
Re-examination of a supposed case of specific cell adhesion
.
Nature, Lond.
226
,
260
261
.
Edelstein
,
B. B.
(
1970
).
Cell specific diffusion model of morphogensis
.
J. theor. Biol.
30
,
515
532
.
Humphreys
,
T.
(
1963
).
Chemical dissolution and in vitro reconstruction of sponge cell adhesion. I. Isolation and functional demonstration of the components involved
.
Devl Biol.
8
,
27
47
.
Humphreys
,
T.
(
1970
).
Species specific aggregation of dissociated sponge cells
.
Nature, Lond.
228
,
685
686
.
Moscona
,
A. A.
(
1968
).
Aggregation of sponge cells: cell-linking macromolecules and their role in the formation of multicellular systems
.
From Jn vitro, vol. III. Publ. Tissue Culture Association
.
Rasmont
,
R.
(
1961
).
Une technique de culture des éponges d’eau douce en milieu contrôlé
.
Annls Soc. r. zool. Belg.
91
,
147
156
.
Sara
,
M.
(
1965
).
Aggregazione cellulare interspecifica fra specie diverse di Poriferi e fra Poriferi ed Anemonia sulcata.
Boll. Zool.
33
,
1067
1077
.
Sara
,
M.
,
Liaci
,
L.
&
Melone
,
N.
(
1966
).
Bispecific cell aggregation in sponges
.
Nature, Lond.
210
,
1167
1168
.
Swift
,
D. L.
&
Friedlander
,
S. K.
(
1964
).
The coagulation of hydrosols by Brownian motion and laminar shear flow
.
J. Colloid Sci.
19
,
621
647
.
Van De Vyver
,
G.
(
1970
).
La non confluence intraspécifique chez les spongiaires et la notion d’individu
.
Ann. Embryol. Morphogén.
3
,
251
252
.
van de Vyver
,
G.
(
1971a
).
Analyse de quelques phénomènes d’histoincompatibilite intraspécifique chez l’éponge d’eau douce Ephydatia fluviatilis (Linné)
.
Archs Zool. exp. gén. (in the Press)
.
van de Vyver
,
G.
(
1971b
).
Mise en évidence d’un facteur d’agrégation spécifique chez l’éponge d’eau douce Ephydatia fluviatilis (Linné)
.
Ann. Embryol. Morphogen, (in the Press)
.
Wolpert
,
L.
(
1969
).
Positional information and the spatial pattern of cellular differentiation
.
J. theor. Biol.
25
,
1
47
.