Studies of early stages of differentiation of the cellular slime mould Dictyostelium discoideum

The life-cycle of Dictyostelium discoideum is much studied as a ‘model system’ of development and differentiation (Loomis, 1975). Development is initiated by starvation when the Dictyostelium amoebae, which have remained solitary during growth, aggregate to form multicellular masses that eventually differentiate into fruiting bodies containing essentially only two cell types, i.e. stalk cells and spores.

In standard laboratory conditions (Sussman & Lovgren, 1965), aggregates begin to form after 8 h development and then pass synchronously, and at known times, through successive morphogenetic stages. The pseudoplasmodium stage is attained first as erect ‘finger-like’ structures at 12 h and then continues as slug-like masses when the ‘fingers’ bend over to lie down horizontally on the substratum after 14-15 h development (Treffry & Watts, 1976). The ‘slugs’ migrate towards light and higher temperature and so have an easily recognized anterior-posterior polarity. Migration soon ceases and fruiting bodies, comprising a spore mass surmounting a thin, cellular stalk, are formed at 24 h. The stalk cells are derived from the anterior, and the spores from the posterior, of the slugs (Raper, 1940; Bonner, 1944). Amoebae at the anterior of the slug differ from amoebae at the posterior in ultrastructure (Müller & Hohl, 1973; Gregg & Badman, 1970), enzyme composition (Newell, Ellingson & Sussman, 1969) and in their pattern of protein synthesis (Alton & Brenner, 1979). Hence differentiation has begun by the slug stage of development to give pre-spore and pre-stalk cells. Any attempt to explain how development and differentiation take place in D. discoideum must therefore account for both the formation of pre-spore and pre-stalk cells within 14 –15 h development and for the pattern the two cell types assume in the migrating slug.

Several theories (reviewed by MacWilliams & Bonner, 1979) have been proposed to explain how the population of apparently unspecialized amoebae at the beginning of development can eventually give rise to two populations of specialized cells, but all these theories seem to depend on adoption of one or other of two assumptions. Most commonly, it is assumed that differentiation does not begin until aggregates having anterior-posterior polarity have formed subsequent to aggregation. It is then proposed that, because of this polarity, amoebae within aggregates can respond to some form of ‘positional information’ (Wolpert, 1969) which ensures that cells at the anterior become pre-stalk cells whilst cells at the posterior become pre-spore cells. The alternative assumption suggests that differentiation has to begin before aggregate polarity is established if unspecialized amoebae are to have time to differentiate into pre-spore and pre-stalk cells by the slug stage of development. Hence it is proposed that the two cell types begin to form early in development and that, after aggregation, sorting-out of the two cell types leads to establishment of polarity in slugs as pre-stalk cells collect at the anterior and pre-spore cells at the posterior.

The two accounts of slime mould development differ markedly concerning the time at which they predict that cell differentiation begins, and it would be possible to decide which is the more accurate if it were known whether cell differentiation precedes, or follows, establishment of polarity in aggregates. A sequence of events consistent with the proposal that differentiation precedes establishment of polarity has been observed during D. discoideum development (Forman & Garrod, 1977; Tasaka & Takeuchi, 1981) but only in abnormal conditions, where aggregates were allowed to form from amoebae suspended in phosphate buffer and where aggregates would not eventually develop into fruiting bodies. It has not been possible to determine whether differentiation precedes formation of polar aggregates in the normal conditions of development on a solid substratum because of a lack of criteria for recognizing cell differentiation if it occurs at such early stages of development. However, it seemed that, if two cell types are formed early in development, they must have different surface properties in order to be able to sort out subsequently to give the pre-spore-pre-stalk pattern found in slugs. We have therefore made use of a technique (countercurrent distribution in a polymer, two-phase system), that separates cells having different surface properties (see Walter, 1977; Fisher, 1981), to determine whether two cell types are present early in development of D. discoideum and prior to formation of polar aggregates.

The technique depends on the ability of mixtures of aqueous solutions of dextran and poly (ethylene glycol) to separate, on standing, into two phases, the upper phase being poly (ethylene glycol)-rich and the lower phase dextran-rich (Albertsson, 1971). When mixed with such a phase system, cells with different surface properties have different affinities for the two phases and therefore separate. However, it is usually necessary to repeat such cell partitioning many times in a countercurrent fashion with fresh phases to obtain good separation.

Dextran T 500 (batch 4094) was obtained from Pharmacia Fine Chemicals and poly (ethylene glycol) 4000 (batch 6444220) from BDH Chemicals Limited. Empigen BB was a gift from Albright & Wilson Limited, Marchon Division, Whitehaven, England.

Amoebae of D. discoideum strain Ax-2 were grown at 22 °C in HL 5 medium containing 86 mM glucose (Watts & Ashworth, 1970) and were harvested either during exponential growth at approximately 10⁶ amoebae ml⁻¹ or during stationary phase of growth at densities greater than 10⁷ amoebae ml⁻¹. Amoebae were washed once with distilled water at 5°.

Development was at 22° on Millipore filters (Sussman, 1966). At various times of development, amoebae were washed off the filters with distilled water at 5°. Single-cell suspensions were obtained from aggregates by vigorous mixing on a Vortex mixer.

A mutant resistant to acriflavin was isolated by spreading amoebae of strain Ax-2 and Aerobacter aerogenes NCTC 418 on nutrient agar plates (Sussman, 1966) containing 100μg ml⁻¹ acriflavin. A colony that grew rapidly was recloned before being maintained in HL 5 glucose medium. Frequent checks confirmed that the mutant did not revert to being acriflavin-sensitive. The growth rate in axenic culture and the time-course of development were the same as for the parental Ax-2 wild-type strain.

The phase system comprised 5 ·5% (w/w) dextran and 5 ·5% (w/w) poly (ethylene glycol) to which had been added, per 200 g final mixture, 10 ml 1 ·0 M-NaCl, 10 ml 0 ·2 M-Na₂SO₄ and 1 ml 0 ·2 M phosphate (KH₂PO₄/K₂HPO₄) buffer pH 7 ·8. This gave a ‘zero-potential’ phase system, i.e. one in which there was essentially no potential difference between the two phases (Johansson, 1974). The phases were kept at 4° and were at pH 6 ·8.

Partitioning was at 4° in a thin-layer, countercurrent distribution apparatus similar to that described by Albertsson (1965), and cells were partitioned 59 times between the two phases. Each time, cells were shaken with the phases for 30 sec and the phases were then left for 10 min to separate. After completion of the countercurrent distribution either 1 % (v/v) Empigen or 50 mM phosphate (NaH₂PO₄/K₂HPO₄) buffer pH 6 ·5 was added to convert each fraction into a single phase. Cell density was determined in fractions by counting cells in a haemocytometer or, for samples that had been lysed in Empigen, by measurement of absorbance at 280 nm. It had previously been found that absorbance at 280 nm was linearly related to cell density.

In some experiments, amoebae that were acriflavin-resistant were mixed with wild-type amoebae (see text for further details) and allowed to form fruiting bodies on Millipore filters. Spores were collected and resuspended in water. Samples (0 ·1 ml) of the spore suspension containing about 1000 spores ml⁻¹ were spread with A. aerogenes on ten nutrient agar plates, and on ten other nutrient agar plates containing 100 μg ml⁻¹ acriflavin. The plates were left four days at 22°. The number of spores of the acriflavin-resistant strain was determined by counting colonies growing on the plates containing acriflavin whilst the total number of spores of wild-type and acriflavin-resistant character was given by counting colonies on the plates lacking acriflavin. The viability on nutrient agar plates of spores from the acriflavin-resistant strain was always found to be the same as the viability of spores of the wild-type strain. Spores of the acriflavin-resistant strain had the same viability on nutrient agar plates in the presence or absence of acriflavin.

It has been usual to make use of phase systems containing a high (e.g. 0 ·1 M) phosphate concentration to effect cell separation (Walter, 1977). In such phase systems there is an electrostatic potential difference between the two phases, with the upper phase positive with respect to the lower phase. It is therefore believed that cell separation achieved with these phase systems largely reflects differences in surface charge between the separated cells. Attempts to use charged phase systems to study Dictyostelium discoideum amoebae were unsuccessful, and studies were continued with a ‘zero-potential’ system where there was virtually no potential difference between the two phases. Thus D. discoideum cells were separated on the basis of differences in surface properties that were not dependent on cell surface charge.

Amoebae were found to be between 85 and 95 % viable after partitioning.

Amoebae grown in axenic culture were harvested during exponential growth and allowed to develop on Millipore filters for up to 11 h. Partitioning distributed amoebae harvested at any time during the first six hours of development as a single peak (Fig. 1 A, B, C, D). However, the position of the peak changed with time of development as cell affinity for the upper phase of the two-phase system gradually increased (0 –4 h) and then decreased at 6 h. This would indicate that changes were occurring during development in the surface properties of the amoebae. No change in cell partitioning was detected between 6 h and 7 h development, but marked change was apparent at 8 h (Fig. 2 A) when the amoebae had begun to aggregate. Then the distribution became extremely broad and indicated that the population of amoebae was extremely heterogeneous in cell surface properties. By 9 h development, heterogeneity was so great that the amoebae were divided by partitioning into two populations (Fig. 2 B). A similar distribution was found at 10 h, and the two populations have been designated as peak I and peak II in Fig. 2 C. Two populations were still apparent at 11 h development (Fig. 2 D) but amoebae were distributed between them in the approximate ratio 1:1 (amoebae in peak I : amoebae in peak II), whereas at 9 h the ratio was 1:3 and at 10 h 1:1-5. Thus from 8 h development onwards there was a gradual tendency for amoebae to regain affinity for the lower phase of the two-phase system, and thus to distribute in lower-numbered fractions of the countercurrent distribution (i.e. to the left in Fig. 2).

After 11 h development amoebae were so cohesive that they tended to aggregate during partitioning unless at very low density. Partitioning of amoebae at later stages of development than 11 h has not therefore been studied.

During differentiation D. discoideum amoebae differentiate to give two populations of specialized cells, i.e. the stalk cells and spores. Since partitioning studies indicated that amoebae at 10 h development were also divided into two populations, it seemed possible that one of these populations might eventually differentiate into spores and the other into stalk cells. In order to investigate this possibility an experiment was designed which depended on use of both wild-type amoebae and amoebae of a mutant strain, derived from the wild type and resistant to acriflavin. The time course of development was the same for both strains.

Wild-type amoebae at 10 h development were separated by partitioning into the two populations (peak I and peak II in Fig. 2C). Similarly, the two populations were isolated after amoebae of the mutant strain had been subjected to partitioning after 10 h development. Wild-type amoebae from peak I were then mixed with an equal number (approximately 10⁷) of mutant amoebae from peak II and the mixture was allowed to form fruiting bodies on Millipore filters. It was then possible to identify the origin of the spores of the fruiting bodies by determining whether the spores were of wild-type or mutant character.

The converse experiment was also carried out, in which mutant amoebae from peak I were mixed with wild-type amoebae from peak II.

It was found (Table 1) that the majority of spores were of wild-type character when wild-type amoebae from peak I were mixed with mutant amoebae from peak II. In the converse experiment, where mutant amoebae from peak I were mixed with wild-type amoebae from peak II, it was also found that the majority of spores were formed from amoebae in peak I. Thus it appeared that amoebae from peak I were the precursors of the spores of the fruiting bodies. Similar results (not shown) were obtained in a few experiments where amoebae of mutant strain G8 which is temperature-sensitive for growth (Gingold & Ashworth, 1974) were used in place of the mutant amoebae that were acriflavin-resistant.

Wild-type amoebae from peak I were also allowed to develop on Millipore filters without being mixed with amoebae from peak II. Fruiting bodies were formed in the same time as, and having a similar appearance to, fruiting bodies formed from the mixtures of amoebae from peaks I and II. However, when amoebae from peak II were allowed to develop alone, fruiting body formation was delayed by 24 h and the fruiting bodies had much longer stalks than fruiting bodies formed from peak I amoebae alone or from mixtures of peak I and II amoebae.

Amoebae from peak I also appeared to differ from amoebae from peak II in cohesiveness. Thus pellets of amoebae from peak II were easily resuspended in water to give single-cell suspensions, but amoebae from peak I could be similarly resuspended only after vigorous mixing for several minutes on a Vortex mixer.

Amoebae were isolated from the leading and trailing edges of the broad peak (Fig. 2 A) obtained by partitioning amoebae at 8 h development. It was then possible, in experiments similar to those described in the previous section, to determine the fate of these amoebae when mixed together and allowed to form fruiting bodies. It was found that spores tended to be formed from amoebae from the trailing edge of the peak and not from amoebae from the leading edge (Table 2).

The distribution of amoebae harvested during the stationary phase of growth and allowed to develop for various times is shown in Figs. 3 and 4. Changes in distribution, when related to time of development, were qualitively similar to those previously described for amoebae harvested during exponential growth. Thus a marked broadening in cell distribution by partitioning was first observed at 8 h development for amoebae harvested in both growth phases (compare Fig. 4 A and Fig. 2 A). However, whilst this change was associated with formation of aggregates by amoebae harvested during exponential growth, it was not associated with the beginning of aggregation of amoebae harvested during the stationary phase of growth, since the latter amoebae formed aggregates after only 6 h of development.

Amoebae harvested during exponential growth were distributed by partitioning as a fairly sharp peak during early development. However, the position of the peak changed with time of development so that changes in the cell surface properties of the amoebae were occurring. The nature of these changes is not known, but the changes may be related to the extensive changes in glycoprotein composition of the cell plasma membranes that occur during early development (Toda, Ono & Ochiai, 1980).

After 8 h development the distribution of amoebae by partitioning became extremely broad, and it would appear that the amoebae were then extremely heterogeneous in cell surface properties. Nevertheless, it was possible to discern that amoebae at 9 –11 h development were divided into two, albeit heterogeneous, populations and it was found that during subsequent development amoebae from the two populations had different fates.

When amoebae from the two populations at 10 h (designated as peak I and peak II in Fig. 24′C) were mixed in equal numbers and allowed to form fruiting bodies, about 69% of the spores formed were derived from amoebae from peak I. This would have been possible for fruiting bodies comprising about 75% spores and 25% stalk cells only if the amoebae from peak I had all differentiated into spores. Thus it is concluded that the population designated as peak I contained amoebae that were all the precursors of spores.

The majority of amoebae from peak II did not form spores. Since stalk cells are the only other major cell type formed during differentiation of D. discoideum, it would seem reasonable to conclude that the preferred fate of amoebae from peak II was stalk formation. Consistent with this conclusion was the observation that amoebae from peak II that were allowed to develop in the absence of amoebae from peak I formed fruiting bodies having much longer stalks than the fruiting bodies formed from mixtures of amoebae from peaks I and II.

At 10 h development the two separable populations of amoebae are in aggregates that appear merely to be rounded masses of cells. Differentiation can therefore occur to give the two cell populations with ultimately different developmental fates before aggregates gain apical tips or anterior-posterior polarity. Thus anterior-posterior polarity is not required to initiate cell differentiation during development of D. discoideum and hence it would appear that neither can ‘positional information’ be required. It would now seem more probable that polarity is a consequence of differentiation and, as proposed by Forman & Garrod (1977), is established as a result of sorting-out of the two different cell populations. It was also found that precursors of spore cells are more cohesive than precursors of stalk cells, and Steinberg (1964) has discussed, with reference to vertebrate cells, how differences in cohesiveness may permit cell sorting-out.

Although amoebae that had been harvested during exponential growth were not clearly separable into two populations on the basis of cell surface properties until 9 –10 h development, it was apparent that differential change in amoebal cell surface properties had begun at 8 h development when amoebae began to aggregate. Furthermore, amoebae from the trailing edge of the distribution produced by partitioning amoebae at 8 h development differentiated into spores whilst amoebae from the leading edge tended not to differentiate into spores and, instead, presumably differentiated into stalk cells. Clearly differentiation in D. discoideum begins prior to formation of aggregates, and this proves unequivocally that neither aggregate polarity nor ‘positional information’ is essential to initiate differentiation of amoebae in standard, laboratory conditions of development that allow formation of fruiting bodies. However, it is probable that the correlation between initiation of differential change in cell surface properties and initiation of aggregation was fortuitous. Certainly aggregation is not in itself the stimulus initiating this differentiation, since amoebae harvested during the stationary phase of growth formed aggregates at 6 h development, but differential change in cell surface properties was not detected until two hours later.

We thank Irene Donnelly and Katrina Longmore for technical assistance. We are grateful to the Royal Society (T.E.T.) for a grant to build the counter-current distribution apparatus, to the Wellcome Trust (D. J. W.) for a grant to purchase dextran and to the SRC for a research studentship (P.T.S.).