ABSTRACT
The behaviour, during the multicellular phase of the life-cycle, of amoebae of Dictyostelium discoideum grown indifferent media is described. Amoebal populations were marked by growthtemperature-sensitive genetic lesions which do not interfere with developmental phenomena. The fate of cell populations was determined by measuring the relative number of mutant and wild-type cells at various stages in the life-cycle. Cells sort out during development in such a way that they may be ordered in a sequence in which those given early in the following list preferentially appear in the spore population when mixed with those given later in the list: cells grown in axenic medium + 86 mM glucose and harvested when in the exponential phase of growth ; cells grown in axenic medium and harvested when in the exponential phase of growth ; cells grown on bacteria and harvested when in the exponential phase of growth; cells grown in axenic medium+ 86 mM glucose and harvested when in the stationary phase of growth. Chemotactic aggregation and grex migration are not essential for sorting-out to occur but, in the normal life-cycle, the cells of a grex formed from amoebae grown in different media have sorted out anteroposteriorly. The relationship between this sorting out behaviour and the mechanism of pattern formation in fruiting-body morphogenesis is discussed. Differences in density of the amoebae cannot account for the sorting out predispositions we observe.
INTRODUCTION
The development of Dictyostelium discoideum is a seemingly simple example of pattern formation. Vegetative amoebae from an apparently homogeneous population aggregate and eventually form an elongate, migrating grex in which pre-spore cells are positioned posteriorly and pre-stalk cells anteriorly (see Bonner, 1967). These two cell types may be distinguished on the basis of cytological, density and enzyme differences (see Garrod & Ashworth, 1973). However, since cut grexes regulate, cells in the grex are not yet determined in the embryological sense. It seems clear that determination only occurs when the stalk cells and the spores are finally formed in fruiting body and that there is therefore no clear temporal distinction in this system between cell determination and cell differentiation.
Bonner (1959) suggested that pre-spore and pre-stalk tendencies existed amongst the amoebae and that the amoebae sorted out during aggregation according to these tendencies. This suggestion was based on studies of the behaviour of cells marked with vital dyes, such as Nile blue and cresyl violet (Bonner, 1957), and on the behaviour of mixtures of cells carrying morphologically distinguishable genetic lesions. Farnsworth & Wolpert (1971) have questioned the reliability, as cell marking agents, of the dyes used by Bonner (1957) and they have reported that there is no mass relative movement of cells in the migrating grex. It is also possible to argue that the morphological markers used by Bonner (1959) affected the sorting out behaviour of the cells carrying them since this behaviour is postulated as forming an integral part of the morphogenesis. More recently Takeuchi (1969) has shown that vegetative amoebae differ in density and that the denser cells of a population sort out to the anterior position in the grex. Bonner, Sieja & Hall (1971) have confirmed this observation using cells derived from strains containing genetic markers (such as spore size) which are unlikely to affect the behaviour of the cells in the grex.
It thus seems established that amongst populations of amoebae there exist variations in a cell property (which may be related to density) as a consequence of which certain cells sort out to the anterior and others to the posterior end of the grex. Thus certain cells are more likely to become spores and others stalk cells. However, since the developmental fate of every cell in the grex is reversible until fruiting body formation commences (the culmination stage) it is clear that we must be dealing with two pattern-forming processes. The first of these is associated with the sorting-out behaviour of the cells and the second is imposed on the cells of the grex during culmination and results in the correct morphogenetic relationship between the stalk and the spore mass. These two patternforming processes could, but need not necessarily, be related.
In this paper we confirm that sorting out of cells can occur and that the developmental fate of an amoeba can thus be predicted. We also show that the density per se of a cell is unlikely to be the determining parameter and that sorting out can occur in the absence of chemotactic aggregation and grex migration.
MATERIALS AND METHODS
Organisms and growth conditions
Dictyostelium discoideum strain Ax-2 was isolated, and has been described, by Watts & Ashworth (1970). Two mutants of this strain, designated G-2 and G-8, have been isolated by our colleague, Dr E. Gingold, after treatment of strain Ax-2 with the mutagen N-methyl-N-nitroso-N’-nitroguanidine and as a result of selecting for organisms unable to grow at 27 °C. Dr Gingold has designated the presumed single mutations carried by these strains as gts-2 and gts-8 respectively. The amoebae of both mutant strains will grow at 22 °C (the optimal temperature for growth of D. discoideum) but G-2 has a longer generation time than either the wild type or G-8. Both G-2 and G-8 strains differentiate in the same time and their differentiation follows the same morphological sequence as does the wild type at 22 or 27 °C. Loomis (1969) has described similar mutants in which growth but not development is temperature-sensitive.
D. discoideum amoebae were grown on bacterial lawns as described by Sussman (1966). Plaques formed from Ax-2 and G-8 amoebae were scored after 4 days incubation at 22 °C and from G-2 amoebae after 5 days incubation at 22 °C. Only strain Ax-2 amoebae grow at 27 °C and plaques formed by these amoebae were scored after 7 days incubation at this temperature. The reversion rates of the gts markers to the gts+ condition are less than 1 in 105.
Two different kinds of axenic medium have been used in this study. One (NS medium) contains yeast extract, peptone and inorganic salts and the other (glucose medium) contains, in addition, 86 mM glucose. Cells grown in these media (which have been described in detail with the exact growth conditions by Garrod & Ashworth, 1972) are described as NS cells and glucose cells respectively. Exponential phase amoebae were harvested when at a cell density of less than 6 × 106 amoebae/ml and stationary phase cells grown in glucose medium were harvested when the cell density had risen by less than 10% over a 24 h period (this corresponds to a cell density of 1–2 × 107 amoebae/ml).
Differentiation
Amoebae were harvested and washed in cold distilled water and resuspended in cold distilled water at a final cell density of 108 amoebae/ml. Portions (0·5 ml) of this suspension were placed on Millipore filters resting on filter paper pads as described in detail by Garrod & Ashworth (1972). Fruiting bodies are formed after 25 h incubation in a moist environment at 22 °C but the filters were usually left for 36 h before the spores were harvested and plated out clonally.
When it was necessary to perform, manipulations on intermediate stages of the life-cycle, amoebae were spread over the surface of a 2% (w/v) solution of ionagar in distilled water (non-nutrient agar) at a cell density of approximately 7 × 104 amoebae/cm2 and incubated in a moist atmosphere at 22 °C. Examination of the plates after overnight incubation enabled the appropriate stages to be identified and picked for experimental manipulation.
Aggregation of amoebae in suspensions of 0·0167 M phosphate buffer were carried out as described by Garrod (1972).
Determination of the density of amoebae
Density gradients were made using the commercially available colloidal silica preparation Ludox (available as a 40 % suspension from Du Pont Chemical Co.) essentially as recommended by Pertoft (1966). A 10 % aqueous suspension of Ludox was adjusted to 1 mM final concentration in EDTA and pH 6·5. The resulting solution was centrifuged at 20000 rev/min in the 3 × 25 ml swing-out rotor of an MSE 50 centrifuge for 90 min to create a very stable density gradient. Portions (1-0 ml) of cell suspension were then layered over the gradient and the tubes centrifuged for a further 10 min at 2500 rev/min. After such centrifugation the cells are found in a very narrow band and are 100% viable.
RESULTS
Characterization of mutant strains
The viability of the wild-type strain Ax-2 and the two growth-temperature-sensitive mutants G-2 and G-8 derived from it varied slightly from experiment to experiment but these viabilities were always consistent within any one experiment. In Table 1 the highest and the lowest viabilities that we have obtained are recorded for the permissive (22 °C) and the discriminating (27 °C) temperature. The viabilities obtained were independent of the medium used to grow the amoebae.
Nature of the spores formed by mixtures of cells grown in various media
Exponentially growing amoebae were grown in axenic media in the presence (glucose medium) or absence (NS medium) of glucose. The amoebae were harvested, washed in cold distilled water, mixed in various proportions and allowed to form fruiting bodies on Millipore filters. The spores from such fruiting bodies were collected and plated clonally and in duplicate with A. aerogenes and the number of clones which grew at 22 °C was compared to the number which grew at 27 °C. After correction for the variable viabilities at these two temperatures (Table 1) the results were expressed as the percentage of mutant spores formed as a function of the percentage of mutant amoebae present in the initial mixture. In Fig. 1 it can be seen that growth of cells in glucose containing medium leads to the preferential appearance of those cells in the spore population. Similar results were obtained when G-8 amoebae were used in mixtures with Ax-2 instead of G-2 amoebae.
Spores formed by mixtures of amoebae grown in different media. ○, G-2 glucose grown + Ax-2 NS grown; •, G-2 NS grown + Ax-2 glucose grown; △, G-2 and Ax-2 glucose grown; ▴, G-2 and Ax-2 NS grown. All amoebae were harvested when in the exponential phase of growth.
Similar experiments were done with mixtures of cells grown on bacteria on solid media and harvested before complete lysis of the bacterial lawn had occurred and cells grown in axenic medium. Fig. 2 gives the results of such experiments using G-8 amoebae. Here it can be seen that cells grown in axenic media preferentially appear in the spore population when mixed with cells grown on bacteria whether the axenic medium contained glucose or not.
Spores formed by mixtures of amoebae grown in different media, (d) ○, G-8 glucose grown + Ax-2 bacterial grown; ◼, G-8 bacterial grown + Ax-2 glucose grown; ◻, G-8 and Ax-2 bacterial grown. (6) •, G-8 NS grown + Ax-2 bacterial grown; ◼, G-8 bacterial grown + Ax-2 NS grown. All amoebae were harvested when in the exponential phase of growth.
Spores formed by mixtures of amoebae grown in different media, (d) ○, G-8 glucose grown + Ax-2 bacterial grown; ◼, G-8 bacterial grown + Ax-2 glucose grown; ◻, G-8 and Ax-2 bacterial grown. (6) •, G-8 NS grown + Ax-2 bacterial grown; ◼, G-8 bacterial grown + Ax-2 NS grown. All amoebae were harvested when in the exponential phase of growth.
The experiments described in Figs. 1 and 2 were done with amoebae harvested when in the exponential phase of their growth cycle. When amoebae grown in glucose medium were allowed to go into the stationary phase of the growth cycle then the results shown in Figs. 3 and 4 were obtained. Similar results were obtained when G-2 amoebae were used instead of G-8 amoebae. Thus cells harvested in exponential growth preferentially appear in the spore population whether they were grown on bacteria or in axenic medium when mixed with stationary phase cells. Experiments similar to those of Figs. 3 and 4 are difficult to carry out with cells harvested in the stationary phase of growth in NS medium since in these circumstances the amoebae have little glycogen and cell lysis occurs quite rapidly (Weeks & Ashworth, 1972). Similarly it is difficult to obtain bacterial grown cells in the stationary phase of growth using solid media since such cells aggregate and begin differentiation.
Spores formed by mixtures of amoebae grown in different media. •, G-8 exponential amoebae grown in NS medium + Ax-2 stationary phase amoebae grown in glucose medium; +, Ax-2 exponential amoebae grown in NS medium-+ G-8 stationary amoebae grown in glucose medium.
Spores formed by mixtures of amoebae grown in different media. ◼, G-8 exponential amoebae grown on bacteria + Ax-2 stationary phase amoebae grown in glucose medium; +,G-8 stationary phase amoebae grown on bacteria + Ax-2 exponential amoebae grown in glucose medium.
The results shown in Figs. 1–4 demonstrate that cells may be arranged in a sequence so that those given early in the following list preferentially appear in the spore population when mixed with those given later in the list: exponential phase cells grown in glucose medium; exponential phase cells grown in NS medium; exponential phase cells grown on bacteria; stationary phase cells grown in glucose medium.
The results shown in Figs. 1–4 were obtained from an analysis of pooled spores from many fruiting bodies and are thus open to two alternative interpretations; either cells grown in different media form separate fruiting bodies and the spore analyses merely reflect the proportions of different fruiting body types or, cells grown in different media form composite fruiting bodies and the spore analyses therefore reflect the behaviour of cells within individual grexes. Analysis of individual fruiting bodies (Table 2) shows that the latter alternative is correct and that each fruiting body is composed of both types of cell.
Analysis of migrating grexes
This demonstration of sorting out of the two cell populations in the fruiting body poses the question of whether this sorting out is manifest in the migrating grex, as implied by the studies of Takeuchi (1969) and Bonner et al. (1971).
Amoebae were grown in different media and allowed to aggregate together on non-nutrient agar. After 16 h one large migrating grex from each mixture was cut at right angles to its longitudinal axis into three sections. The front and back sections were dissociated by trituration in cold distilled water and the cells plated clonally with A. aerogenes to determine the percentage of mutant cells in each section (Fig. 5). From the results of such experiments (Table 3) it is clear that the sorting out of the amoebae had occurred at or before the migrating grex stage. Hence the question arises of which stage in the life-cycle is essential for sorting out.
Analysis of aggregates made in suspension
In order to see if the chemotactic aggregation stage was essential for sorting out Ax-2 and G-2 or G-8, amoebae were grown in various media and then mixed in 0-0167 M phosphate buffer, pH 6-0, in stoppered tubes. These tubes were then rotated about their long axis at 18 rev/min at 22 °C overnight (Gerisch, 1968 ; Garrod, 1972) so that spherical balls of cells formed. When these were placed on a solid surface they gave rise to fruiting bodies whose spores were analysed for mutant and wild-type cells (Fig. 6). As can be seen in Table 4 the cells in such aggregates sorted out during fruiting body construction.
Sorting out of cells in aggregates made in suspension without any chemotactic aggregation stage

Mixing of cells at the migrating grex stage
It is known that when cut grexes regulate the regulation is optimal if the cut tip is allowed to migrate before it constructs a fruiting body (Raper, 1940). This suggests that the migration of the grex might be in some way necessary for sorting out. In order to test this hypothesis amoebae grown in various conditions were allowed to form migrating grexes on non-nutrient agar. Grexes of different provenances were then placed side by side on the agar and stirred with a fine needle so that the component cells of the grexes were mixed thoroughly. It is not possible to determine accurately the number of cells of each type mixed in this experiment. However, care was taken to select grexes of similar size and shape so that there were approximately equal numbers of glucose and NS cells present after mixing of the two grexes. From the resulting mass of cells a number (usually 3–6) of small fruiting bodies are formed directly without any further migration (Fig. 7). The spores of these fruiting bodies were collected and analysed for mutant and wild-type spores. The results described in Table 5 show that considerable sorting out occurs and hence migration of the grex is not essential for sorting out to occur.
Grafting experiments
Raper (1940) showed that when the tip of one grex was placed on the back of another whose tip had been removed the two pieces of grex fused to form a composite grex in which the grafted cells largely retained their relative positions. We have shown that it is possible to predispose cells to sort out to the back of a grex (Table 3) and so it is of interest to place the tip of a grex composed of cells which are known to sort out posteriorly on to a grex (whose original tip has been removed) composed of cells known to sort out anteriorly (Fig. 8). In this way it is possible to determine if cells ‘remember’ their position or their sorting out predisposition when these two are in conflict. In all cases examined (Table 6) the sorting out predisposition is dominant and grafted tip cells can be found preferentially amongst the spore population.
Determination of the density of cells
Density gradients made from Ludox are much easier to make, less toxic, more reproducible and more discriminating than those we have used previously (Miller, Quance & Ashworth, 1969). Amoebae grown in various media were analysed and found to have densities in the order: stationary-phase cells grown in glucose medium > exponential-phase cells grown in glucose medium > exponential-phase cells grown on bacteria > exponential-phase cells grown in NS medium. Dissociated grex cells formed two distinct bands of different density as found before (Miller et al. 1969) and both these bands were at higher densities than that characteristic of the stationary phase cells grown in glucose medium.
DISCUSSION
In this paper we have studied the behaviour, during the multicellular phase of the life-cycle, of mixed cell populations the component cells of which have been grown in different media. In each mixture one of the cell populations has been genetically marked such that its cells were incapable of growth at 27 °C. It is important to establish that the genetic markers we have used have no effect on the observed sorting out behaviour. We believe that there is no such effect because (a) Ax-2, G-2 and G-8 cells do not sort out from one another when they are grown in the same media (Figs. 1, 2), (b) G-2 and G-8 were isolated as a consequence of independent mutagenic events, are phenotypically distinguishable, yet have identical sorting out behaviour, (c) development of strains Ax-2, G-2 and G-8 is in all respects identical at 22 and 27 °C, (d) genetic lesions which specifically affect growth processes have been isolated before (Loomis, 1969) and in this organism growth and development are mutually exclusive phenomena, and (e) our results are consistent with previous work using other methods of marking cell populations (Takeuchi, 1969; Bonner et al. 1971).
It is not easy to estimate the degree of precision shown by the sorting out process which we report here. The sharpness of the points of inflexion in Figs. 1–4 (particularly in Fig. 1 and Fig. 4, crosses) suggests that there is a very high degree of precision. However, this is difficult to express quantitatively since it is known that whereas the spore : stalk cell ratio is 3-95:1 for cells grown in glucose media it is only 2·7:1 for cells grown in NS medium (Garrod & Ashworth, 1972). We do not know what the spore : stalk cell ratio is for mixtures of cells of the type we have been studying here but it is unlikely to be linearly related to the cell composition of the mixture. Further, it is technically difficult to measure accurately small numbers of one cell type in the presence of a great excess of the other. However, we feel that it is possible to postulate that all, to a first approximation, of the ‘vacancies’ for spore cells are filled by those amoebae with the appropriate sorting out predisposition, and only then when all these vacancies have been filled, are amoebae predisposed by their sorting out behaviour to become spore cells forced, by the necessities of the morphogenetic pattern, to become stalk cells.
We must therefore now examine the question of whether there is any mechanistic connexion between sorting out and pattern formation. There are essentially two possibilities: (1) either glucose cells have a greater predisposition to form spores than NS cells, even in the vegetative stage, and once incorporated into a composite grex such cells sort out according to their predisposition, thus giving rise to the spore:stalk pattern, or (2) vegetative amoebae are merely predisposed by their growth conditions to sort out from each other in such a way that glucose cells preferentially adopt the posterior position in the grex and NS cells the anterior. In the latter case the cellular spore:stalk pattern would be imposed on the grex cells by some pattern-specifying mechanism within the grex which is distinct from the sorting out predispositions of its component cells. In the first case there would be a causal connexion between the pattern and the sorting out whereas in the latter case the connexion would be merely fortuitous. We have presented arguments against believing that there is any causal connexion between sorting out and pattern formation previously (Garrod & Ashworth, 1973) and the work we report here provides no reason for changing this view. Whilst we cannot discount entirely a causal connexion between sorting out and pattern formation, at present there seems no evidence which would lead us to conclude that the connexion is other than fortuitous.
Our results show that during the normal life-cycle sorting out of cell populations has taken place by the migrating grex stage (Table 3). This does not enable us to say at which stage sorting out normally occurs. However, we can say with some confidence that certain stages in the life-cycle are not essential for sorting out to be manifest. Two experiments, making mixed aggregates in suspension and grex grafting (Tables 4, 6), allow the development of heterogeneous cell populations without a chemotactic aggregation stage. In both cases sorting out of the two cell populations took place, demonstrating that the aggregation stage is not essential for sorting out. When two homogeneous grexes each composed of a different cell type were stirred together fruiting bodies were formed in which sorting out had occurred without the intervention of a migrating grex stage, showing that a migration stage itself is also not essential for sorting out. Thus, provided only that a culmination stage intervenes between mixing of the cell populations and fruiting body formation, sorting out can occur.
Our grex grafting and mixing experiments (Figs. 7, 8) require special discussion in relation to a conflict which has arisen between the results of various workers. On the one hand Bonner (1952) and Takeuchi (1969) suggest that cells of the migrating grex ‘remember’ their positions, and if displaced therefrom will return, whilst on the other hand Farnsworth & Wolpert (1971) have demonstrated that cells displaced within a grex (e.g. tip cells placed at the back of a grex) remain where they are put and they conclude ‘that there is no cell sorting out in the migrating grex ‘. Our experiments show that grex cells sort out according to a hierarchy determined by the medium used for their growth rather than according to the position in the grex from which they were taken. We cannot, though, say whether this sorting out takes place during the migration period following the graft or during the subsequent culmination stage. It appears that the medium in which a cell is grown is more important than the position from which it was taken when these two influences are in opposition in grex grafts.
The fact that tip cells move to the back of the grex (Table 6) in our grafts makes it most unlikely that the tip of the grex can usefully be said to act in a manner analogous to the autonomous ‘organizer’ region postulated to regulate pattern formation during early vertebrate embryogenesis (Spemann, 1938). Rather we would suggest that the importance of the tip is a morphogenetic one in that it leads and directs the movement of the cell mass (Raper, 1940; Garrod, 1969).
The physiological basis for the sorting out process remains obscure. Takeuchi (1969) suggested that the cells sort out on the basis of their density such that the lightest cells tend to become spores, but we find that the amoebae of strain Ax-2 have densities in the following order: stationary-phase cells grown in glucose medium > exponential-phase cells grown in glucose medium > exponentialphase cells grown on bacteria > exponential-phase cells grown in NS medium. This order is the same as that of the glycogen content of the amoebae (Weeks & Ashworth, 1972; Ashworth & Watts, 1970) but is quite different from the sorting out order of preferential appearance in the spore population of cells : exponential-phase cells grown in glucose medium; exponential-phase cells grown in NS medium; exponential-phase cells grown on bacteria; stationary-phase cells grown in glucose medium.
From a physiological point of view the most revealing experiment is that shown in Fig. 4, where it is clear that the predisposition of cells grown in glucose medium alters as the time at which they were harvested alters. It is known that as such cells leave the exponential phase of the growth curve and enter the stationary phase they synthesize and excrete into the medium relatively large amounts of adenosine 3’,5’-cyclic monophosphate (cAMP) (Malkinson & Ashworth, 1972). Preliminary experiments have shown that the rate of change in the predisposition of the glucose-grown cells can be markedly accelerated by incubating exponential-phase cells with cAMP and it is known that high concentrations of cAMP can cause isolated amoebae to become similar in appearance to stalk cells (Bonner, 1970). If these observations can be confirmed then the most likely explanation for the sorting out phenomena that we have reported here is that it is due to changes in the cell membrane catalysed by enzymes released from amoebae as a consequence of an increase in the extracellular concentration of cAMP (Ashworth, 1971).
Finally, the emergence of an order of preferential appearance of physiologically distinct populations of cells in the spore population is reminiscent of the finding of an order for preference for the internal position in the sorting out of embryonic chick cells (Steinberg, 1970). We have no indication, of course, of whether this apparent similarity is trivial or significant.
Acknowlegment
We thank Mrs Janet Kwasniak and Miss Julia Johnson for technical, and the Science Research Council for financial, assistance. One of us (C. K.L.) thanks the Medical Research Council for the award of a fellowship.