We describe the effect of cyclic AMP on regulation of the proportion of prespore and prestalk cells in Dictyostelium discoideum. Prespore and prestalk cells from slugs were enriched on Percoll density gradients and allowed to regulate in suspension culture under 100 % oxygen. The transition of prespore to prestalk cells is blocked by cAMP, while cAMP phosphodiesterase and caffeine cause a decrease in the number of prespore cells. This suggests that extracellular cAMP plays a role in cell type proportioning by inhibiting the conversion of prespore to prestalk cells.
Low concentrations of cAMP prevent the conversion of prestalk to prespore cells; the same effect is seen with hydrolysis products of cAMP, 5 AMP, adenosine and also adenine. We suggest that, when low concentrations of cAMP are added to regulating cells, the cAMP itself is quickly broken down and the breakdown products thereafter inhibit the prestalk-to-prespore conversion.
The relevance of these findings is discussed in the context of an non-positional double-negative feedback model for cell type homeostasis.
The life cycle of the cellular slime mould Dictyostelium discoideum (Dd) involves a simple differentiation process in which cells of one type, vegetative amoebae, differentiate into at least two cell types, stalk cells and spores. The ratio of the two cell types is almost constant over a large size range, under specified environmental conditions (Bonner & Slifkin, 1949; Stenhouse & Williams, 1977).
Prestalk and prespore cells are first recognizable at the late aggregate stage and are organized in a one-dimensional pattern in the slug stage. At this stage the prestalk cells are localized in the front quarter of the slug and the prespore cells in the back three quarters. Under normal circumstances prestalk and prespore cells convert quantitatively into stalk and spores respectively (Tsang & Bradbury, 1981). The prespore–prestalk pattern at the slug stage is still regulative, i.e., when a slug is cut into two pieces, separating the prestalk and prespore zones, both pieces, have the ability to regulate and reform the missing parts and finally give rise to normally proportioned fruiting bodies (Raper, 1940; Sakai 1973). This indicates the existence of a well-developed cell type homeostatic mechanism. There is good evidence that the pattern-forming mechanism in Dd consists of a largely positionindependent cell type homeostasis mechanism and another mechanism, chemotactic cell sorting, that puts the cells into the right place (Takeuchi, 1969; Forman & Garrod, 1977; Durston & Vork, 1979; Tasaka & Takeuchi, 1981; Sternfeld & David, 1981; Weijer, McDonald & Durston, 1984a; Weijer, Duschl & David, 1984b).
One possible way to investigate the cell type homeostasis mechanism is to study the cell type transitions under conditions of regulation, a method first used by Sakai to quantitatively study the regulation of prespore and prestalk pieces of slugs (Sakai, 1973). The normal proportion of prespore and prestalk cells is perturbed and the kinetics with which both cell types return to their equilibrium proportion is followed.
We have modified this procedure by taking gradient-purified cell types (Weijer et al. 1984a) instead of prestalk and prespore pieces cut out of slugs. We let the cell types regulate in oxygenated suspensions (Sternfeld & Bonner, 1977), instead of on water agar, which allows better control over the extracellular environment. Under these conditions both the purified prestalk and prespore cells rapidly reaggregate, i.e. they form submerged aggregates, and regulate their proportions. At various times aliquots of the aggregates were collected, dissociated and the percentage of prespore cells and prestalk cells determined.
This procedure allowed us to follow the regulation of the prestalk and prespore cells with and without the addition of putative morphogens. In this study we have concentrated on the role of cyclic AMP and metabolites in cell type regulation.
MATERIALS AND METHODS
Strains and developmental conditions
All experiments were performed with Dictyostelium discoideum strain Ax2, which was grown in axenic medium as described earlier (Watts & Ashworth, 1970). Cells were harvested when they had reached a density between 2−7 × 106 cells. Ml−1 and collected by centrifugation (2 min 400g). Cells were washed three times in 20mM-potassium phosphate buffer pH 6·8 (KK2) and resuspended in the same buffer to a density of 2 × 107 cells. ml−1. Plates of synchronous slugs were prepared by pouring 5 ml of the cell suspension onto levelled agar plates (1 % Difco bacto agar in KK2) 9 cm in diameter. The cells were allowed to settle for 15 min after which the excess Equid was carefully decanted and the plates air dried for 10 min. This resulted in an evenly distributed multilayer of cells which gave rise to synchronous standing slugs after 16 h of development at 22 °C in the dark.
Gradient separation of prespore and prestalk cells
The gradient-separation procedure used is described in detail elsewhere (Weijer et al. 1984a). The relevant points from this procedure are described below. After 16–18 h slugs were scraped off the plates with an microscope slide and resuspended in KK2 buffer. The slugs were separated from the unaggregated cells by filtration through a fine nylon sieve (mesh size 40 μm). The slugs were dissociated, either in 0·6% NaCl containing 2mM-EDTA or in a pronase/BAL mixture according to published procedures (Takeuchi & Yabuno, 1970). After this the cells were washed twice in KK2 buffer and once in a salt solution containing 0·6 % NaCl, 2 MM-EDTA and 20 MM-KK2. The density of the cells was adjusted to 5 × 107 ml and 2 ml were layered on top of a 10 ml Percoll gradient (linear 30-90%, preformed) containing the same salt mixture as the solution used for the last cell wash. The gradients were centrifuged for 10 min at 2000g in a Hettich K2S table-top centrifuge at 10 °C. The gradients were fractionated and the fractions containing the majority of the prespore (bottom 1 ml) and the prestalk cells (2nd to the 6th ml from the top) were collected by hand with a Pasteur pipette. Fractions from three such gradients were pooled, washed three times in KK2 buffer and resuspended to a density of 5 × 106 cells ml−1.
The purity of the fractions was determined by staining cells with prespore antiserum. The prestalk (lightest) fraction normally contained 10 % of total cells put on the gradient and consisted of 10–25 % prespore cells and 75-85 % neutral-red-staining cells. This fraction also showed the prestalk-specific acid phosphatase isoenzyme. Based on these criteria this fraction was enriched about fourfold for prestalk cells. The dense fraction normally contained about 50 % of total cells and consisted of 90–95 % prespore cells, 2–7 % neutral-red-staining cells and was negative for the prestalk-specific acid phosphatase. This fraction therefore consisted almost entirely of prespore cells.
Roller tube cultures
3 ml of cell suspension were added to a 15 ml screwcap test tube with or without added test substances and the tubes were flushed out with ten tube volumes of pure oxygen and sealed. The test tubes were put on a test tube rotator where the tubes were rotated around their own long axis at 20r.p.m. at 22°C.
Aggregate dissociation and prespore cell staining
The cells in the roller tubes aggregated rapidly and the aggregates developed to tight aggregates within 4–6 h. At this stage they were very difficult to dissociate with protease/BAL, therefore we tried a number of different enzymes and found Cellulase Onozuka R10 (10 mg. ml−1) in KK2 pH 6·0 containing 2 mM-EDTA and 0·6 % NaCl to be very effective. Incubation of the aggregates for 10 min at room temperature and then pipetting the suspension two to three times through an Eppendorf pipette resulted essentially in a single-cell suspension. Cells were pelleted by centrifugation for 8 sec in an Eppendorf microcentrifuge and the pellet was resuspended in 50 μl KK2 and fixed in 60 % methanol. 5–10 μl of the fixed cell suspension was put on a microscope slide and air dried. The cells were postfixed with 25 μl 100 % methanol and air dried again. Then the cells were stained for 30 min with 10/d of a rabbit anti-spore antiserum, which had been adsorbed with aggregation competent cells (Takeuchi, 1963). The final dilution of the antiserum was 500-fold in KK2 buffer pH 6·8 and the staining was done in a moist chamber. After this the slides were rinsed by immersing in KK2 buffer and then stained for another 30 min with a 100-fold dilution of a FITC-labelled goat anti-rabbit IgG. (Nordic Immunological Laboratories, Tilburg, the Netherlands). After this the cells were rinsed again in KK2 and mounted in a 50 % dilution of glycerin in 100 mM-Tris/HCl pH 8·0. Prespore cells are defined as those cells that show staining of more than two prespore vacuoles per cell. There did not seem to be an appreciable loss of cells during the staining procedure. At least 400 cells/time point were counted, resulting in counting errors less than 4 %.
Neutral red staining
Dissociated cells were stained for 5 min in 0·02 % neutral red in KK2. The fraction of neutral-red-stained cells was determined by counting strongly stained cells with large vacuoles under Nomarski optics. In all cases at least 400 cells/sample were counted.
Acid phosphatase gel electrophoresis
Acid phosphatase isoenzymes were separated on non-denaturing polyacrylamide gradient gels modified according to existing procedures (Oohata, 1983 and Loomis & Kuspa, 1984). Linear gradient gels were made with a gradient mixer from 3 % acrylamide (3 % acrylamide, 0·08 % bis acrylamide, 0·06% ammonium persulphate and 0·03% N,N,N,N:-tetramethyl ethylenediamine) solution in 100 mM-imidizole/HCl pH 6·8 and 6% acryl/bis acrylamide in 100mM-imidizole/HCl pH 7·8. The running buffer consisted of 4g.l−1 imidizole and 4g.l−1 sodium acetate pH 7·0. The samples were lysed in 0·2 % Triton ×100 in imidizole/HCl pH 6·8 for 30 min at 4 °C. The samples were made up to 15 % sucrose and 20 μl were layered onto the gels. The gels were run for about 4 h at 4 °C at a constant voltage of 100 V after which time the tracking dye bromphenolblue had reached the bottom. To visualize the isoenzymes the gels were incubated with 0·5 mg. ml−1 α-naphtyl acid phosphate sodium salt and 0-5 mg. ml-1 Fast Garnet GBC in 100 mM-citrate buffer pH4-8. The gels were stained for 60min at 25 °C.
Protein was determined using the Coomassie Blue method according to Bradford (1976), with bovine serum albumin as a standard.
The regulation of prespore and prestalk cells in roller tubes
Under our culture conditions cell type regulation occurred from prestalk to prespore cells and from prespore to prestalk cells. At equilibrium about 50 % of the cells were prespore cells, in contrast to about 80 % prespore cells in slugs. This ratio shift is a property of our in vitro culture system, and is in agreement with the observation of an enlarged neutral-red-stained area in submerged aggregates (Sternfeld & David, 1981).
The regulation of prespore cells isolated on density gradients gives the following picture. Over the first 2–6 h a significant proportion of the prespore cells lose their prespore vacuoles and a new equilibrium concentration of about 50 % prespore cells is established. This condition is then stable (Fig. 1A). In order to investigate the differentiation state of the newly formed non-prespore cells we determined the fraction of cells containing large neutral-red-staining vacuoles, a marker for prestalk and anterior-like cells at the slug stage (Bonner, 1959; Durston & Vork, 1979; Sternfeld & David, 1981; Tasaka & Takeuchi, 1981). Figure 1A indicates that during the regulation of prespore cells neutral-red-stained cells are formed at a rate comparable with the disappearance of prespore cells.
During regulation of prestalk cells isolated from density gradients prespore cells are newly formed. At the same time the fraction of neutral-red-staining cells diminishes (Fig. 1B).
In both regulation experiments (Fig. 1A,B) the change in neutral-red-stained cells roughly equals the change in prespore cells since the sum of the percentages of prespore cells and neutral-red-stained cells is approximately 100 % during regulation. Thus the changes in cell type proportions must be due to conversion of one cell type into the other. As is the case for slug cells, the neutral-red and presporestaining properties of cells appear to be mutually exclusive states (Yamamoto & Takeuchi, 1983).
In order to further characterize the neutral-red-stained cells we investigated another prestalk-specific marker, the prestalk-specific acid phosphatase isoenzyme (Oohata, 1983; Loomis & Kuspa, 1984). In contrast to previous observations we found that under our conditions there are at least three distinct acid phosphatase isoenzymes (Fig. 2). The band with the highest mobility is present in all cells from the vegetative stage onwards. Slug-stage cells normally contain three bands of which the one with the lowest mobility is most strongly represented in old and culminating slugs. Prestalk cells contain all three bands, while prespore cells contain only the two fastest bands.
Figure 3 shows the changes in acid phosphatase isoenzymes during prestalk and prespore regulation experiments. During regulation of prestalk cells all three bands remain present at about the same intensity from the beginning of the experiment onwards (Fig. 3A), despite the decrease in the number of neutral-red-stained cells (Fig. IB), thus the isoenzymes are relatively stable under our conditions. During regulation of prespore cells the slowly migrating prestalkspecific band is only formed after 20 h of regulation (Fig. 3B), while neutral-redpositive cells are formed during the first 4–6 h in the regulation process (Fig. 1). Thus, under these conditions, the differentiation of prespore cells and neutral-red-staining cells appears to be uncoupled from the expression of the prestalkspecific acid phosphatase isoenzyme. This uncoupling is also found during normal development of vegetative cells in our in vitro system. Although 50 % of the cells differentiate to prespore cells after 20 h of development, the prestalk-specific acid phosphatase does not appear until 24–28 h of development (Fig. 4). Although the expression of the acid phosphatase during regulation experiments confirms the identification of neutral-red-stained cells as prestalk cells, it is an inconvenient marker due to the delay of its expression (in prespore-to-prestalk regulation) and its long-term stability (in prestalk-to-prespore regulation). Hence we have used neutral-red-staining and prespore staining to identify prestalk and prespore cells in most experiments.
The effect of cAMP on prespore regulation
We first investigated the influence of cyclic AMP (cAMP) on prespore regulation. Several cell differentiation systems have been described in which cAMP is needed to obtain differentiation of prespore cells (Kay, Garrod & Tilly, 1978; Okamoto, 1981; Abe, Saga, Okada & Yanagisawa, 1981; Chung et al. 1981; Kay, 1982; Mehdy, Ratner & Firtel, 1983). From earlier experiments it is known that when slug cells are shaken in suspension under conditions where they are not able to make cell contact, they dedifferentiate, but that dedifferentiation can be inhibited by the inclusion of cAMP in the medium (Takeuchi & Sakai, 1971; Okamoto & Takeuchi, 1976; Tasaka et al. 1983). We found that in our system cAMP blocks the regulation of prespore cells, irrespective of the initial prespore cell ‘concentration’: 95 % prespore cells remain 95 % prespore cells under the influence of cAMP (Fig. 5), while untreated prespore cells regulate to around 50 % prespore cells (Figs 1,5). The effect of cAMP is concentration dependent, higher concentrations have a stronger effect (Fig. 5). When cAMP is added at millimolar concentrations it blocks the regulation of prespore cells almost completely for at least 6–8 h and during this time there are no appreciable numbers of neutral-red-staining cells formed.
The effect of cAMP on the regulation of prestalk to prespore cells is more complicated. At high concentrations of cAMP there is little effect on the conversion of prestalk to prespore cells (Fig. 5), but at lower concentrations there is a pronounced inhibition of prespore cell formation and a stabilization of neutral-red-staining cells. This rather surprising result is further characterized below.
The effect of cAMP hydrolysis products on cell type regulation
Slug-stage cells contain appreciable extracellular activities of phosphodiesterase and 5′nucleotidase (Brown & Rutherford, 1980; Armant & Rutherford, 1980; Tsang & Bradbury, 1981; Weijer et al. 1984a). Therefore, when cAMP is added to a slug cell suspension, it will be degraded at least partially to 5′AMP and adenosine and possibly to adenine. Therefore we tested the effect of cAMP breakdown products on the cell type transitions.
The effect of two possible breakdown products of cAMP, adenosine and adenine, on cell type regulation is shown in Fig. 6. At concentrations up to 10−5M there is no effect upon the conversion of prespore to prestalk cells. High concentrations of adenosine however, increase the rate of disappearance of prespore cells and lead to a shift in the equilibrium proportions, an effect also seen with caffeine (see below). The conversion of prestalk to prespore cells is inhibited by 10−5 M-adenosine and adenine (Fig. 6). The results in Fig. 7 confirm the stabilization of prestalk cells by 10−5 M-adenosine, scored as neutral-red-stained cells. The effect of 5′AMP is similar to that of adenosine (data not shown).
The effect of the dissociation procedure on cell type regulation and its inhibition by cAMP
We used pronase dissociation of slug tissue in order to achieve good cell type separation on Percoll gradients. It has however, been shown that pronase treatment of slug-stage cells of certain mutants directs those cells into the stalk pathway (Peacy & Gross, 1981). Therefore we compared the effect of three different dissociation procedures on cell type regulation in our culture system. Slugs were dissociated either with pronase or with cellulase or dissociated mechanically by tituration in a salt/EDTA solution. The results in Fig. 8 show that the dissociation method does not greatly influence the results, although pronase dissociation may slightly increase the sensitivity of prespore cells to cAMP.
Evidence for a prespore-stabilizing action of the cAMP produced endogenously by the regulating cells
The results above indicate that cAMP blocks the prespore-to-prestalk conversion in our culture system. To test the possibility that cAMP produced endogenously by regulating cells stabilizes prespore cells under our culture conditions we added cAMP phosphodiesterase in order to lower the concentration of extracellular cAMP. Under these conditions the prespore cells disappeared faster than normal and the equilibrium of the cell types shifted to fewer prespore cells (Fig. 9). This result is consistent with the idea that, under conditions of regulation, some cells secrete cAMP into the extracellular space, where it stabilizes prespore cells. The added phosphodiesterase hydrolyses this external cAMP and therefore destabilizes prespore cells resulting in a more rapid decrease in the proportion of prespore cells.
An alternative method for reducing endogenously produced cAMP is to treat cells with caffeine. It has been shown that caffeine blocks the relaying response of aggregating cells by inhibiting the activation of adenylate cyclase and thus the production of cAMP (Theibert & Devreotes, 1983; Brenner & Thoms, 1984). We therefore examined the effect of caffeine on cell type regulation. Caffeine blocks the conversion of prestalk to prespore cells (Fig. 10A) and accelerates the conversion of prespore to neutral-red-stained cells (Fig. 10B). Thus caffeine also leads to a shift in the ratio of the cell types in the direction of prestalk-like cells. Millimolar concentrations of adenosine and adenine have the same destabilizing effect on prespore cells as caffeine. The effect of caffeine can only be partially antagonized by the addition of millimolar concentrations of cAMP. Thus the activation of the adenyl cyclase and therefore elevated levels of cAMP appear to be involved in the stabilization of prespore cells.
Cell type regulation
We have shown that, in suspension culture, regulation from prestalk to prespore and from prespore to prestalk occurs on the time scale of a few hours to an equilibrium situation of 50 % prespore cells and 50 % prestalk cells (Fig. 1). Since essentially all cells can be accounted for as prespore (antigen-positive) or prestalk (neutral-red-stained) cells, these results indicate that cell type conversion is occurring under regulation conditions.
Expression of the prestalk-specific acid phosphatase isoenzyme however is delayed until 20 h during regulation (Fig. 3). A delay in timing of differentiation in suspension culture can also be seen at the level of prestalk/prespore sorting. Although prespore cells form at 16 h of development in suspension culture, sorting only occurs at 24 h (Forman & Garrod, 1978; Tasaka & Takeuchi, 1981). Together these results suggest that the neutral-red-stained cells initially formed in regulation experiments are anterior-like cells, which do not sort (Sternfeld & David, 1981). Thus the delay in suspension cell culture seems to be at the level of the conversion of anterior-like (early prestalk) cells to prestalk cells. This suggests that the acid phosphatase isoenzyme may be a marker which distinguishes prestalk cells from anterior-like cells.
The reason for this difference in the timing of differentiation in suspension culture and on agar or filters remains to be elucidated. One possibility is that the timing of prestalk differentiation is markedly different due to the inability of the cells to condition the medium with factors required for later stages of differentiation (Town, Gross & Kay, 1976; Kopachick et al. 1983). Preliminary experiments using conditioned media of various types, however, have not shown any substantial effect on the rate of appearance of the prestalk specific isoenzyme during in vitro development.
Effect of cAMP and hydrolysis products on cell type regulation
We have found that cAMP added to roller tube cultures of prespore cells blocks the regulatory process in which about half of these cells convert to neutral-red-staining cells. The effect is concentration dependent; higher concentrations have a stronger effect. Treatments with caffeine or cAMP phosphodiesterase, which lower endogenous cAMP, destabilize prespore cells. It is difficult to determine the effective extracellular cAMP concentrations since there is no simple way to estimate the extracellular cAMP concentration in the aggregates. However micromolar concentrations of cAMP show a short-term stabilization of prespore cells and it is therefore likely that the in vivo concentrations lie well below this value. It thus appears that extracellular cAMP stabilizes prespore cells.
Millimolar concentrations of cAMP do not affect the conversion of prestalk to prespore cells, although lower concentrations (10−5−10−7M) do inhibit this process. Furthermore low concentrations of cAMP hydrolysis products block conversion of prestalk to prespore cells. These seemingly self-contradictory findings can be explained by the following assumptions:
Added cAMP is rapidly broken down and the hydrolysis products inhibit the prestalk-to-prespore conversion.
cAMP itself has no effect on prestalk-to-prespore conversion.
High concentrations of cAMP antagonize the inhibitory action of cAMP hydrolyses products, for instance by competition for binding sites and therefore relieve the inhibition of the prestalk-to-prespore conversion.
It is at present not known whether cAMP hydrolysis products normally play a role in proportion regulation in slugs. However, it has recently been shown that aggregation stage cells, at least, have both low- and high-affinity adenosine receptors which are distinct from their receptors for cAMP (Newell, 1982; Newell & Ross, 1982; van Haastert, 1983) and that cAMP when added in a 100-fold excess suppresses adenosine binding effectively (Newell & Ross, 1982). In our in vitro regulation system adenine was also found to inhibit the prestalk-to-prespore conversion as effectively as adenosine and 5’AMP. The possibility that the adenosine receptors also bind adenine or that there are special adenine receptors has to our knowledge, not yet been investigated.
The mechanism by which added cAMP stabilizes prespore cells remains to be elucidated. The finding however that external cAMP is only partially effective in antagonizing the caffeine effect on prespore cell differentiation would indicate that the activation of adenylate cyclase is important for the stabilization of prespore cells.
Effect of high and low cAMP concentrations on cell type differentiation in a mutant
Ishida (1980) has described a mutant in which differentiation of single cells depends on the concentration of added cAMP; at low cAMP levels the mutant forms stalk cells, while at higher levels spores are formed. This effect can be explained under the assumption that, in the presence of cAMP, any prespore cells formed are stable, but that cells, in the absence of added cAMP, cannot form sufficient cAMP to stabilize prespore cells but can produce sufficient cAMP or hydrolysis products to form stalk cells. This result indicates a clear concentrationdependent cell-type-specific function for cAMP in cell differentiation and qualitatively agrees with our findings.
Timing differences in prestalk-to-prespore regulation in vitro and in vivo
Sakai (1973) found that there is a timing difference in the regulation of prespore and prestalk pieces of slugs: in prestalk-to-prespore regulation there is lag of at least 3 h before any cell type conversion occurs, while in the opposite direction cell type conversion begins immediately. We observe no such difference under our regulation conditions. This suggests that the cell type switch from prestalk to prespore is not the limiting step in prestalk-to-prespore regulation in slugs. Possibly the delay seen in slugs is a feature of the dynamics of the cell-type-regulating signal, which could be quantitatively different in vivo and in vitro.
A model for cell type regulation
To place our findings in a conceptual framework we note that it is possible to have a pattern-forming mechanism that consists of two processes, a cell type homeostasis mechanism and a sorting process that puts cells in the right place. The easiest way to imagine a cell type homeostasis process that regulates the proportions of two cell types is to assume that each cell has at any given time a certain probability to convert into the other cell type, i.e. cycles through two differentiation states (Durston & Weijer, 1980). Analogous to cells traversing different cell cycle phases, where the entry in the DNA synthesis phase can be described as regulated by a random transition probability (Smith & Martin, 1973).
If both cell types can switch, this will lead to an equilibrium situation where the ratio of the cell types is given by the ratio of the transition probabilities. Disturbance of this equilibrium will lead to an exponential return to this equilibrium state. When one applies this idea to the regulation of the prespore and prestalk cell types one is confronted with the fact that prestalk and prespore cells are spatially separated. To maintain this separation in the presence of continuous cell type switching there would have to be continuous sorting out of the cell types. This does not seem to be in agreement with the observation that during slug migration there is not much cell sorting of neutral-red-stained cells going on; sorting mainly takes place during slug formation and regulation of slug pieces (Durston & Vork, 1979) . It also does not agree with the finding that it is possible to label subpopulations of cells at the vegetative stage and find the labelled cells back with a non-random distribution at the slug stage (Takeuchi, 1969; Weijer et al. 1984).
To solve this problem we propose that each cell type produces an inhibitor that inhibits its own formation from cells of the other type (Fig. 11). This is equivalent to saying that the transition probability is a function of the inhibitor (cell type) concentration, i.e. at higher inhibitor concentrations the transition probability gets smaller leading to a drastic reduction of the cell type transitions as soon as the inhibitor concentration increases. This results in a situation where the cells are stable once in equilibrium and therefore no sorting is required to maintain the pattern in the slug once it is formed. The possibility that cell type conversion continues at a low rate (Durston & Vork, 1979) is not excluded. Cells will start to convert to the other cell type only when the equilibrium is disturbed (i.e. the inhibitor level falls below effective inhibition) and the population will regulate until equilibrium is reached again.
An essential feature of this model is that the production of the substance that stabilizes the prespore cells is dependent on prestalk cells. The concentration of the inhibitor is expected to be highest close to the main prestalk mass and to fall off towards the end of the slug. We predict therefore that prespore cells are likely to be most stable at the boundary between the prespore and prestalk zone and most unstable at the posterior end of the slug, where the inhibitor concentration is lowest. This prediction is contrary to the one made by positional information models which would say that cells are most unstable at the border between two cell type domains (MacWilliams & Bonner, 1979). Our proposal that prespore cells are most unstable in the rear of the prespore zone appears to be supported by the observation that the anterior-like cells, which are scattered in the prespore zone and most likely are a differentiation state between prespore and prestalk cells, are more frequent in the posterior part of the prespore zone than in the anterior part (Durston & Vork, 1976; Voet & Williams, personal communication). The anterior-like and rearguard cells might form preferentially at the distal end of the slug due to a lower extracellular concentration of cAMP, whose synthesis is dependent on prestalk cells and is therefore expected to be the highest near the prestalk zone and lowest in the posterior prespore zone.
Possible molecular realizations of the model
We have shown that cAMP inhibits the conversion of prespore to prestalk cells and we thus suggest that cAMP is the prespore stabilizing factor (whose synthesis is dependent on prestalk cells). It can now easily be seen why it was difficult to obtain evidence for a function for cAMP in cell type regulation. When cAMP is added from the beginning of development onwards it does not result in a noticeable shift in the prestalk-to-prespore ratio as might be expected for an activator (Kay et al. 1978; Abe et al. 1981, own unpublished observations). Any prespore cells formed are stabilized by the added cAMP, but as soon as enough prespore cells are formed, the conversion of prestalk to prespore cells will be blocked by the prespore-controlled inhibitor of the prestalk-to-prespore-cell conversion (Fig. 11). The effect can only be seen under conditions where due to low cell density no cell interactions can occur (Ishida, 1980; Kay, 1982), or where prespore cells are studied under regulation conditions (see above).
It is clear that the transition from prestalk to prespore cells has to be regulated as well. In our model it is suggested that this transition is also regulated by an inhibitor. Our experiments indicate that cAMP hydrolysis products at low concentrations inhibit the transition of prestalk to prespore cells, while they do not appreciably affect the prespore-to-prestalk conversion. Hence these hydrolysis products are potential candidates for the inhibitors of the prestalk-to-prespore conversion.
Another factor that has been shown to influence cell type regulation is DIF (Town, Gross & Kay, 1976; Kay et al. 1978). It has recently been shown that, in the presence of cAMP, DIF can convert isolated prespore cells to stalk cells (Kay & Jermyn, 1983). One possible way to explain the action of DIF under those conditions is to suppose that it desensitizes prespore cells to cAMP for instance by inhibition of the activation of adenylate cyclase (like caffeine) and therefore results in the loss of prespore cells in the presence of cAMP. At the moment we are investigating the effect of DIF on prespore and prestalk regulation in our in vitro system.
The finding that under certain conditions cells can form stalk cells in the presence of cAMP (Bonner, 1970; Town & Stanford, 1978; Kay & Jermyn, 1984) seems at first sight to contradict our findings that cAMP stabilizes prespore cells. We however think it is possible that there is more than one signal involved in prespore and spore differentiation. Cells are only sensitive to cAMP for a limited period of time and that during this limited period of time further signals are required (Sternfeld & David, 1979; Wilkinson, Wilson & Hames, 1984) for spore maturation. Such a sequence of stimuli has recently been shown to be necessary for the differentiation of stalk cells (Sobolewski, Neave & Weeks, 1983).
It has recently been shown that by treatment of isolated cells of sporogenous mutants with weak acids or bases cell differentiation can be shifted into the stalk or spore pathway (Gross, Bradbury, Kay & Peacy, 1983). It therefore would be interesting to investigate whether these treatments affect the internal cAMP concentration by altering the internal pH or altering the associated membrane potential.
We thank Dr S. K. Brahma for the preparation of the prespore-specific antiserum and Charles N. David and Harry K. MacWilliams for helpful suggestions in the preparation of the manuscript. Rob Bleumink for expert technical assistance in the initial experiments and Gerdi Duschl for making the phosphatase gels work. We also wish to thank the referees for constructive criticism.
This work was supported by the Foundation for Fundamental Biological Research (BION), which is subsidized by the Netherlands Organization for the Advancement of Pure Research (ZWO).