Acidic Ca²⁺ Stores, Excitability, and Cell Patterning in Dictyostelium discoideum

Dictyostelium discoideum is best known for its self-organizing behavior, as seen in the striking wave patterns formed during cell-to-cell relay of chemotactic cyclic AMP (cAMP) signals and in the stable division of the resulting aggregates into distinct anterior and posterior zones. Amoebae of D. discoideum multiply as separate cells as long as a food source is available. During this time, they are chemotactically responsive to folic acid released by their bacterial food source. The developmental process is initiated by starvation. Over the first few hours, amoebae synthesize the components of the cAMP signaling system (44). Aggregation begins in response to cAMP “pulses” propagated from signaling centers. Cells that have been stimulated by cAMP themselves make a directional movement step lasting about a minute in the direction of the wave source (see reference 23). Thus, cells moving toward an aggregation center are repeatedly excited (at 2- to 8-min intervals) by the successive waves of extracellular cAMP signaling that traverse the aggregation fields. By mimicking such signals with “shots” of cAMP added to aerated suspensions of amoebae that have developed the competence to aggregate, it has been shown that when cAMP binds to cell surface receptors, adenylyl cyclase is transiently activated (in a process referred to as cAMP relay); cAMP accumulates within the cells for a minute or two, and most of it is then released, and both intra- and extracellular cAMP is rapidly hydrolyzed by cAMP phosphodiesterases. Binding of cAMP to its receptors also activates several downstream effectors e.g., phospholipase C and guanylyl cyclase, and therefore leads to transient changes in InsP3, cyclic GMP, and Ca²⁺, H⁺, and K⁺ ions (24, 44).

As developing amoebae move stepwise toward central signal sources on a solid surface, they gather into streams that collect into hemispherical mounds of cells. A protruding tip then forms at the apex of each aggregate. As more amoebae flow into the center, the aggregate elongates upward against gravity to form a finger-like structure referred to as a “first finger.” This structure either falls over to form a migrating slug or, in a process referred to as culmination, immediately forms a fruiting body consisting of a mass of spores held aloft by a cellular stalk. Raper (58) showed that slugs have a constant leading end and that they are divided into an anterior zone consisting of cells (prestalk cells) that give rise to the stalk cells of the fruiting body and a posterior zone of cells (prespore cells) that give rise to the spores. He also showed that slugs have a remarkable ability to regulate their prestalk/prespore proportions when interfered with in various ways by surgical means. Bonner (7) later demonstrated a visible anterior/posterior differentiation corresponding to Raper's fate map: staining of aggregates with neutral red and other weak base dyes revealed that the anterior cells of aggregates contain highly acidic vesicles while those of the posterior cells are much less acidic (see also reference 80) (Fig. 1). Neutral red (a weak base) stains large intracellular acidic structures (autophagic vacuoles), as well as numerous smaller acidic compartments. Subsequent studies using antibodies and reporter constructs revealed that the zones of prestalk- and prespore-specific gene expression correspond closely to the neutral red staining and nonstaining zones, respectively.

FIG. 1.

Neutral red-stained slugs. The prestalk zones are strongly stained. The rear regions of the prespore zones are also stained, weakly, perhaps because ammonia diffuses away from the backs of the slugs and permits some acidification of acidic compartments. (Reproduced from reference 32 with permission.)

In what follows, I shall argue that the properties of the anterior and posterior cells of aggregates (slugs) can be accounted for by the following assumptions. (i) Cytosolic Ca²⁺ is sequestered into a specific type of internal store by an ATP-dependent Ca²⁺/H⁺ exchanger acting in conjunction with a vacuolar H⁺-ATPase that transfers protons into the compartment interior. (ii) cAMP relay by the adenylyl cyclase ACA is dependent, inter alia, on cytosolic Ca²⁺ transients resulting from release of this stored Ca²⁺ in response to the binding of cAMP to its cell surface receptors. (iii) The vacuolar H⁺-ATPase is active in the anterior cells of aggregates but inactive in the posterior cells. The former can therefore fill these stores and experience Ca²⁺ transients, whereas the latter cannot. (iv) The Ca²⁺ transients are responsible for driving prestalk cell-specific (PST) gene expression and inhibiting prespore cell-specific (PSP) gene expression. Hence, anterior cells express PST genes but not PSP genes and posterior cells do not express PST genes. (v) Posterior cells express PSP genes as a result of activation of cAMP-dependent protein kinase A by cAMP generated by a separate, constitutively active, adenylyl cyclase (ACG) present only in the posterior cells.

Evidence that an acidic Ca²⁺ store is involved in cAMP relay.

The idea that acidic vesicles (vesicles possessing a vacuolar-type H⁺-ATPase) are involved in signaling and cell differentiation evolved from studies of the effects of extracellular pH, and weak acids and bases, on cell differentiation in an in vitro system (26). Slime molds produce large quantities of ammonia as a result of protein and RNA turnover during development, and ammonia has many effects on development. The addition of this weak base (and others) to an in vitro cell differentiation system, or elevation of pH_o, was found to inhibit stalk cell formation and favor spore formation, while weak acids had the opposite effect (16, 25, 26, 53). This led to the suggestion that the cell differentiation pathway is influenced by some aspect of intracellular pH (26). However, it subsequently became evident that a difference in cytosolic pH was unlikely to be responsible, since any such difference, if it existed, was very small (27, 31), and so attention shifted to acidifiable intracellular compartments. It was shown that any of a large array of weak bases applied to monolayers of starved cells under conditions of normal development inhibited aggregation in a concentration-dependent manner; they also inhibited tip formation and culmination when added to preformed aggregates (see reference 14). Since weak bases are known to act as protonophores, diffusing into acidic compartments in the uncharged form (e.g., NH₃) and exiting in the protonated form (e.g., NH₄⁺), thereby opposing the acidification of intracellular acidic compartments by vacuolar-type H⁺-ATPases (57), Davies et al. (14) tested whether this was the basis of the inhibitory effect of weak bases on Dictyostelium development. They were indeed able to establish, using ³¹P nuclear magnetic resonance spectroscopy, that continuous incubation with a weak base caused a sustained and dramatic reduction of the acidity of intracellular acidic compartments while having little or no effect on cytosolic pH. This finding provided strong support for the idea that weak bases inhibit signaling by interfering with vesicle acidification, and this view was strengthened by the finding that mutants with known or suspected defects in the V-type H⁺-ATPase, such as Hgr8 (6), are defective in developmental signaling (15), i.e., behave like cells exposed to a weak base (see also 82). These observations thus suggested that weak acids and bases might also affect the choice of cell differentiation pathway by altering vesicle acidification.

In a major contribution, Schindler and Sussman (66) showed that the amount of cAMP synthesized in response to a “shot” of cAMP by suspensions of developing amoebae was greatly reduced in the presence of ammonia. Most importantly, they observed that cAMP accumulation by one of the mutants with defective signaling mentioned above, KY3, was abnormally sensitive to weak base inhibition (66), as would be expected if its ability to acidify intracellular compartments was already compromised. These findings demonstrated a direct link between ammonia action and inhibition of cAMP relay.

Vesicle acidification is coupled to Ca²⁺ sequestration.

All cells contain intracellular compartments that take up Ca²⁺ from the cytosol. Milne and Coukell (49) showed that Dictyostelium vesicle preparations sequestered Ca²⁺ in the presence of ATP. Since proton gradients are often used to energize the antiport or symport of other ions or compounds, it seemed possible that Ca²⁺ sequestration was dependent on vesicle acidification and that interference with this process by weak bases might somehow affect cAMP relay. Rooney and Gross (59) tested this idea and found a substantial reduction in Ca²⁺ uptake in both crude lysates and acid vesicle fractions when the proton pump was inhibited by bafilomycin or when the proton gradient was collapsed by the classical protonophore nigericin. Sequestration was sensitive to vanadate, indicating that it depended on a Ca²⁺-ATPase rather than a simple H⁺/Ca²⁺ antiporter. It was also inhibited by concanamycin and ammonia (60) and was therefore dependent on the transport of H⁺ ions into the vesicles by a vacuolar-type H⁺-ATPase (Fig. 2). Subsequent work in other laboratories has established that the enzyme is a plasma membrane-type (PMCA) Ca²⁺-ATPase designated PAT1 and that the acidic Ca²⁺ stores seem to be of two kinds, so-called contractile vacuoles and acidocalcisomes (42, 45, 50).

FIG. 2.

Schematic diagram of an acidic vesicle with a bafilomycin-sensitive vesicular proton pump and an ATP-dependent Ca²⁺/H⁺ exchanger. Protons are transferred into the vesicle by the proton pump, and the Ca²⁺/H⁺-ATPase pumps Ca²⁺ into the vesicles in exchange for the protons. The thin arrows illustrate how weak bases such as ammonia oppose Ca²⁺ sequestration by diffusing into the vesicles in the uncharged form (e.g., NH₃), picking up protons, and exiting in the charged form (NH₄⁺), thus creating a steady proton leak. Ca²⁺ sequestration is also sensitive to vanadate, indicating that it is dependent on a Ca²⁺-ATPase rather than a simple Ca²⁺/H⁺ antiporter. The Ca²⁺ pump is a PMCA-type ATPase (50) that is inhibited by tBHQ but not thapsigargin (60), and the acidic Ca²⁺ stores seem to be of two kinds, so-called contractile vacuoles and acidocalcisomes (45). The term acidic vesicle is used to refer to vesicles that have an associated vacuolar H⁺-ATPase whether or not their lumens are markedly acidic. In fact, contractile vacuoles are not stained markedly by neutral red, indicating that they are not strongly acidic. (Figure prepared with the assistance of Anthony Morgan.)

Dictyostelium amoebae also possess endoplasmic reticulum (ER) Ca²⁺ stores, but these are clearly distinct from the acidic Ca²⁺ stores since sequestration of Ca²⁺ into the ER is dependent on a Ca²⁺-ATPase that is sensitive to thapsigargin, a well-established sarcoplasmic/endoplasmic-type Ca²⁺-ATPase inhibitor, and resistant to bafilomycin and vanadate (19). Traynor et al. (75) disrupted the only gene likely to encode an IP3 receptor and found that the resulting iplA mutant aggregated and developed normally despite the fact that exposure to cAMP did not give rise to the normal Ca²⁺ influx from the medium into the cell and intracellular Ca²⁺ transients were not detectable with the Ca²⁺ indicator fura-2-dextran. This strongly suggests that the ER Ca²⁺ pool and the associated sarcoplasmic/endoplasmic-type Ca²⁺ ATPase—as well as the normal level of cAMP-induced cellular Ca²⁺ influx—are not essential for the excitable responses.

Filling and release of the acidic vesicle Ca²⁺ pool into the cytosol may control excitability.

The finding that signaling, chemotaxis, and development do not show any substantial dependence on the ER Ca²⁺ pool or on cAMP-induced Ca²⁺ influx (75) could mean that the excitable responses of Dictyostelium are independent of Ca²⁺ signaling. Alternatively, it could mean, as mentioned by Traynor et al. (75) themselves, that highly localized, short-lived, or minor Ca²⁺ transients not presently detectable by imaging still occur in the iplA mutant and in the absence of perceptible Ca²⁺ influx from the medium. If that were the case, it would imply that in the wild-type situation these unobserved transients would occur first and then trigger further Ca²⁺ release from the ER and/or Ca²⁺ influx from the medium, so giving rise to detectable Ca²⁺ transients. It is well known that calcium ions diffuse over only very small distances in the cytosol before being bound by intracellular Ca²⁺ buffers, and detectable Ca²⁺ signals generally depend upon the amplification and propagation of such localized signals by processes such as Ca²⁺-induced Ca²⁺ release (56, 61).

The possibility that Ca²⁺ signaling and acidic Ca²⁺ stores do underlie excitability is supported by the following evidence, most of it from the laboratory of Dieter Malchow. (i) The acidic Ca²⁺ stores are emptied when cAMP binds to its cell surface receptors (65), as required if one supposes that this store participates in excitability. (ii) Transient changes of intracellular and extracellular Ca²⁺ concentration seem to be associated with all forms of Dictyostelium excitable behavior (including chemotactic responses to folic acid). Thus, both folic acid and cAMP induce transient uptake of Ca²⁺ (9) and transient increases in [Ca²⁺]_i (83), and both forms of chemotaxis are dose-dependently inhibited by the calmodulin antagonists trifluoperazine and calmidazolium (22). In addition, spontaneous spike-shaped, as well as sinusoidal, light-scattering oscillations in suspensions of starved amoebae are accompanied by oscillations of the extracellular Ca²⁺ concentration. (iii) Treatments that transiently raise cytosolic Ca²⁺ can delay light-scattering oscillations (and cAMP relay) for up to one unit length of the cycle, suggesting that the elevation of cytosolic Ca²⁺ interferes with a basic oscillator controlling excitable behavior (39, 40, 54). Raising cytosolic Ca²⁺ also blocks cAMP relay in response to an artificial cAMP pulse (40). (iv) The addition of low concentrations of the calmodulin inhibitor W7 to cell suspensions exhibiting early oscillations increases the amplitude of the oscillations (the successive reductions and increases in light scattering) and of the accompanying cAMP transients (41). This appears to be due to an increased amplitude of cAMP-induced Ca²⁺ transients that has been interpreted to be a consequence of partial inhibition of Ca²⁺ reuptake by the vacuolar calcium pump (41); alternatively, uptake might be slowed, and oscillation amplitude increased, by inhibiting the mitochondrial Ca²⁺ uniporter, since this is known to contribute to the termination of Ca²⁺ signals in other organisms (55), and to be inhibited by W7 (51). (v) Ca²⁺-induced Ca²⁺ release liberates only about a third as much Ca²⁺ from preparations of vesicles from the vacuolar H⁺-ATPase mutant Hgr8 (6) as it does from those of the wild type, thus confirming that that the vacuolar H⁺-ATPase influences the extent of in vivo Ca²⁺ sequestration (43). Most importantly, Hgr8 cells did not produce light-scattering oscillations under standard conditions but could be made to do so by adding W7 (43). This appears to support the view that Ca²⁺ transients resulting from release of Ca²⁺ from the acidic store are responsible for light-scattering oscillations and that W7 increases the amplitude of weak transients generated in Hgr8 cells. W7 has been shown to inhibit the Dictyostelium vacuolar H⁺-ATPase (41) but has not been tested on the Dictyostelium mitochondrial Ca²⁺ uniporter. Further study of the effect of W7 on oscillation amplitude in Dictyostelium could provide vital support for the role of Ca²⁺ transients in excitability.

In summary, therefore, I suggest that Ca²⁺ is pumped into acidic stores by an ATP-dependent calcium/proton exchanger in conjunction with the vacuolar H⁺-ATPase, and that binding of agonist to cell surface receptors releases the stored Ca²⁺, generating a localized intracellular Ca²⁺ transient; this, together with other factors (44), is needed to activate cAMP relay and chemotaxis, and the process of release and reuptake of Ca²⁺ is repeated in response to each successive cAMP signal during and after aggregation. Weak bases and other agents that reduce net proton flux via the V-H⁺-ATPase reduce the extent of filling of these stores and hence reduce the magnitude of the cAMP-induced Ca²⁺ transients and of the ensuing cellular responses (Fig. 3).

FIG. 3.

Simplified diagram of the proposed network underlying signaling. Binding of cAMP to G-protein-coupled cell surface receptors (car1) activates a series of components including CRAC, Aimless, Erk2, and Pianissimo, leading to transitory stimulation of the aggregative adenylyl cyclase ACA (44), and much of the cAMP produced is released from the cell to propagate the cAMP signal. This signaling loop has traditionally attracted the most attention. However, according to the model presented here, equally essential for signaling is the generation of a Ca²⁺ transient resulting from the release of Ca²⁺ from acidic vesicles (AV). Ca²⁺ release presumably involves the production of some unidentified agent that binds to the membrane of the acidic vesicles. I propose that the Ca²⁺ transients activate Ca²⁺-dependent proteins, one or more of which are required directly or indirectly to activate ACA, as well as to induce prestalk cell-specific gene expression. Ca²⁺ is returned to the acidic vesicles by the action of the V-type H⁺-ATPase and Ca²⁺/H⁺-ATPase (NB, both transporters are actually in the acidic vesicle membrane), and this sequestration is opposed by the action of weak bases, which therefore reduce the responsiveness of cells to cAMP signals. (Figure prepared with the assistance of Anna Gross.)

Interestingly, this Dictyostelium signaling system appears to be a very early example of a mechanism that is widespread in more recently evolved organisms. Similar Ca²⁺ stores acidified by bafilomycin-sensitive V-type H⁺-ATPases are found in sea urchin and frog eggs and in many different mammalian cell types, including pancreatic acinar and beta cells (81), smooth muscle, and neurons (21). The pancreatic beta cell is an interesting example, in which the vital process of insulin release in response to glucose has been shown to involve a bafilomycin-sensitive Ca²⁺ store (81). In addition, release of Ca²⁺ from acidic stores in rat PC12 cells drives neuronal differentiation (8), rather as I am proposing that it drives PST gene expression in Dictyostelium. In many cases, the stores are thought to be loaded by an ATP-independent H⁺/Ca²⁺ antiporter, but in other cases, an H⁺-countertransporting Ca²⁺-ATPase is clearly responsible (34, 52), as in Dictyostelium. The acidic vacuoles of Saccharomyces cerevisiae possess both a low-affinity H⁺/Ca²⁺ antiporter and a higher-affinity PMCA-type H⁺-countertransporting Ca²⁺-ATPase (13). In many cases, Ca²⁺ appears to be released from acidic stores by the recently identified Ca²⁺-mobilizing agent NAADP, which accumulates in response to the binding of certain agonists to cell surface receptors (21).

Acidic stores and cell patterning.

I turn now to the nature of the difference between prestalk and prespore cells. The proposed role of vacuolar proton transport in cell signaling implies that cells in which this proton transport did not take place would not experience Ca²⁺ transients; they would therefore not relay cAMP signals and would not express any genes whose expression depended upon these Ca²⁺ transients. I have already mentioned that the posterior cells of aggregates fail to stain with neutral red (7, 80); hence, it is reasonable to suppose that they are incapable of transporting protons into their acidic stores. These cells also respond poorly in terms of chemotaxis to cAMP and do not appear to generate Ca²⁺ transients (12). By contrast, the anterior cells are highly responsive to cAMP and direct the movement of aggregates (47, 74); they also store substantial amounts of Ca²⁺ (37, 72) and display striking Ca²⁺ transients (12). All of these differences can be accounted for if anterior cells acidify their vesicles efficiently whereas posterior cells do not. Furthermore, if prestalk gene expression depends upon elevated Ca²⁺ transients (acting via Ca²⁺-dependent transcription factors), only the anterior cells will express prestalk genes. I should mention here that the finding that prestalk cells are high-Ca²⁺ cells and prespore cells are low-Ca²⁺ cells contradicts my earlier prediction that prespore cells were the high-Ca²⁺ cells (59, 72). That prediction was based on viewing the acidic compartments simply as passive stores of Ca²⁺ rather than the active participants in cAMP signal generation that they now appear to be.

What we know of their developmental origin supports the above picture of the nature of anterior and posterior cells. It is well established that the patterning of aggregates into anterior and posterior zones is, at least in large part, due to the sorting of cells already present in the initial population of starved cells that were starved at different stages of the cell cycle (4, 10, 69); these cells display differences in behavior similar to those of the anterior and posterior cells of aggregates: the anterior sorters develop rapidly, are chemotactically more responsive than the posterior sorters, and are much more able to act as initiators of aggregation territories by releasing cAMP. Most importantly, the two preaggregative cell types have high and low vesicular and cytosolic Ca²⁺ levels, respectively (2), like the anterior and posterior cells of aggregates. It is not yet known whether they also differ in the extent of acidification of their acidic compartments, though I would predict this to be the case. If this is so, it would imply that the vesicle-acidifying systems of posterior and anterior cells differ in a relatively stable fashion rather than resulting from some kind of diffusion gradient, since the cells are still unaggregated. Possible mechanisms regulating vesicle acidification include reversible phosphorylation of some component of the V-H⁺-ATPase, dissociation and reassembly of the V-H⁺-ATPase, alterations in the coupling of proton transport to ATP hydrolysis, and effects on the chloride conductance needed for vacuolar acidification (20).

As one might expect if sorting depends ultimately on differences in the ability to acidify intracellular acidic compartments, the mutants, already referred to, that carry mutations which lower their V-type H⁺-ATPase activity form aggregates with smaller prestalk zones and larger prespore zones (35). Cells of these mutants also have a markedly increased preference to populate the prespore zone (i.e., to become prespore cells) when developed in mixtures with wild-type cells, as expected if their less efficient vacuolar acidification “pushes” them toward the prespore fate (35).

Nevertheless, stable structural differences between the V-H⁺-ATPases of anterior and posterior cells cannot be the whole explanation for how the two types of slug cells are formed since it is well known that both prestalk and prespore cells can change their fate and reestablish roughly normal cell proportions when aggregates are bisected surgically. Such regulatory behavior suggests the possibility that gradients of a slowly diffusing weak acid and a fast-diffusing weak base (ammonia) run from the anterior to the posterior of aggregates and slugs (36, 48, 63) (Fig. 4), leading to a step gradient in vesicle acidification. It seems significant in this connection that ammonia has a greater inhibitory effect on the cAMP chemotaxis of the so-called “anterior-like” cells of aggregates than of prestalk cells (18), as one would expect if vacuolar acidification is less efficient in the anterior-like cells than in the prestalk cells (the former being positioned lower in the proposed anterior-posterior gradient than the latter). Low-molecular-weight factors other than ammonia may also be involved in establishing the prestalk/prespore bifurcation (see reference 67).

FIG. 4.

Tipped aggregate and slug. Cells with acidified vesicles are indicated schematically by the red dots, and cells whose acidic vesicles are poorly acidified are indicated by the gray dots. According to the view presented here, scroll waves of cAMP propagate from the tip through the prestalk zone of aggregates, leading to forward movement of the aggregates, but the waves are unable to propagate through the prespore zone because the cells of this zone do not have acidified vesicles. Only prestalk cells, therefore, are thought to experience Ca²⁺ transients, and these transients drive prestalk cell-specific gene expression and inhibit prespore cell-specific gene expression. Prespore cells contain a constitutively active adenylyl cyclase, ACG, responsible for prespore-specific gene expression (1). The highly acidified autophagic vacuoles of prestalk cells break down proteins and RNA, generating substantial amounts of ammonia that diffuses rapidly, and is proposed to form an anterior-posterior gradient that contributes to the inhibition of vesicle acidification in the prespore cells. Prestalk cells also create a local acidic environment consisting of protons—or some kind of weak acid—that protects the anterior cells from the inhibitory effects of the ammonia and promotes vesicle acidification. Prespore cells may move together with the prestalk cells by a follow-my-leader type of mechanism, as well as by being constrained within the aggregate by the slime sheath. (Figure prepared with the assistance of Anthony Morgan.)

Cell patterning and gene expression.

Surprisingly little attention has been paid to the mutually exclusive nature of prestalk cell and prespore cell differentiation: not only is there a clear-cut bifurcation between the two types of cells, but pst cells express only PST genes and psp cells express only PSP genes. I have already suggested that one can account for restriction of PST gene expression to anterior cells if one assumes that it is driven by elevated [Ca²⁺]_i acting directly or indirectly on Ca²⁺-dependent transcription factors. One can then also account for the lack of PSP gene expression in prestalk cells if one supposes that elevation of [Ca²⁺]_i inhibits PSP gene expression, i.e., that the same or different Ca²⁺-dependent transcription factors inhibit PSP gene expression. There is extensive circumstantial evidence for an association between PST gene expression and elevated levels of [Ca²⁺]_i (5, 12, 33), Moreover, Schaap and her collaborators (64), in a key experiment, have shown that incubation of cells in vitro with the Ca²⁺-ATPase inhibitor 2,5-t-butylhydroquinone (tBHQ) leads to a progressive rise in [Ca²⁺]_i and activates PST gene expression (though for some reason it does not appear to activate the early prestalk gene ecmA efficiently). Importantly, the same experimental regimen was found to inhibit the expression of a PSP gene (76), which fits with the idea that PSP gene expression is inhibited by elevated [Ca²⁺]_i.

How, then, do prespore cells make PSP genes? PSP gene expression is known to require elevation of intracellular cAMP acting presumably via cAMP-dependent protein kinase. Now, the assumption that Ca²⁺ transients are absent from prespore cells also means that the aggregative-stage adenylyl cyclase ACA cannot be activated in these cells, since I am proposing that this activation requires Ca²⁺ transients; hence, ACA cannot be the source of the cAMP needed to drive PSP gene expression. Recent work by Alvarez-Curto et al. seems to provide a way out of this dilemma: they have shown that prespore cells—and only prespore cells—contain a dedicated, constitutively active, hence Ca²⁺-independent, adenylyl cyclase, ACG (1).

A further point of note is that if the anterior-posterior patterning of slugs is maintained by some kind of gradient, as suggested above, then the assumption that the amplitudes of Ca²⁺ transients decline to zero at the prestalk/prespore boundary implies that there is a steep gradient in the amplitude of the [Ca²⁺]_i transients within the prestalk zone. Such a gradient, combined with threshold effects on transcription (62), may contribute to the remarkably complex patterns of gene expression observed within this zone (29, 38). For example, pstA genes (expressed at the front of the prestalk zone) could be transcribed by a Ca²⁺-responsive protein with a high Ca²⁺ threshold, while transcription of pstO genes (expressed only in pstO mutant cells at the rear of the prestalk zone and in anterior-like cells) could be inhibited by this or some other Ca²⁺-responsive protein but activated by one with a lower Ca²⁺ threshold. Consistent with this notion, induction of a pstA reporter construct was observed to require a severalfold higher concentration of the stalk cell-inducing factor differentiation-inducing-factor 1 (DIF-1) than induction of a pstO construct (17).

Mode of action of DIF-1.

Mention of DIF-1 leads to the long-standing question of the mode of action of DIF-1. DIF-1 is a membrane-permeable chlorinated hexaphenone that is produced during development and is astonishingly potent at switching isolated amoebae onto the stalk pathway in vitro in the presence of cAMP (30, 73, 78). Mutants lacking DIF-1 fail to form the basal disc that bears the stalk of the fruiting body, and about half the types of transcript normally found in the pstO region of aggregates appear to be missing (38, 70). Nevertheless, developing cells of the mutant form prestalk cells and fairly normal-looking fruiting bodies (70, 71). Interestingly, however, the prestalk regions of the aggregates are roughly half the normal size (46, 70). This implies that DIF-1 increases the proportion of cells that “choose” the prestalk pathway and hence is a component of the dynamic system that controls pathway choice. DIF-1 appears to be made and released by the prespore cells; it may therefore act in concert with the putative acid “activator” (mentioned above and made by prestalk cells) to promote prestalk differentiation.

I believe that a strong case can be made that DIF-1 exerts its effects by increasing [Ca²⁺]_i. Incubation of cells with DIF-1 in vitro led to a progressive rise over several hours in [Ca²⁺]_i that paralleled the effect of the Ca²⁺-ATPase inhibitor tBHQ referred to earlier (64; see also reference 33). Moreover, the effects of both DIF-1 and the inhibitor were antagonized by chelating Ca²⁺ with EGTA or BAPTA. DIF-1 also seems to generate rapid changes in cellular Ca²⁺. Thus, Azhar et al. (3) detected a twofold increase in the fraction of high-Ca²⁺ cells within 2 min of adding DIF-1 to freshly starved amoebae, and Wurster and Kay (79) found that addition of DIF-1 to cell suspensions displaying light-scattering oscillations caused phase shifts and delays—and lowered cyclic GMP levels—just like Ca²⁺ ionophores and other agents that raise [Ca²⁺]_i. An immediate effect of DIF-1 on [Ca²⁺]_i may well account for how it rapidly induces DIF dechlorinase (28) and causes nuclear translocation of the bZip DNA binding proteins DimA and DimB (78). Several mammalian bZip transcription factors are activated by Ca²⁺ (11, 77). Although DIF-1 uncouples mitochondrial respiration (68), the concentrations required are about 2 orders of magnitude greater than those that induce prestalk cell differentiation. It is unlikely, therefore, that this action of DIF-1 is relevant to its role in call patterning.

Concluding remarks.

I have argued above that proton-driven loading of an acidic Ca²⁺ store and subsequent release of the Ca²⁺ into the cytosol, leading to an increase in [Ca²⁺]_i, play a central role in chemotaxis, signal relay, and developmental gene expression in D. discoideum. If I am right, these processes underlie the successive cycles of excitation and recovery during cAMP wave propagation on solid surfaces, as well as light-scattering oscillations in cell suspensions. Moreover, inactivation of this system in prespore cells may be responsible for the stable bifurcation of aggregates into prestalk and prespore cells.

In order to confirm the role of the acidic Ca²⁺ stores in Dictyostelium, it will be important to use improved detection methods such as line scan confocal imaging (61) to detect the cAMP-induced Ca²⁺ transients predicted to occur in the absence of a functional ER Ca²⁺ store, and to establish the identity of the agent that releases Ca²⁺ from the store. It would also be desirable to establish whether the prestalk-sorting and prespore-sorting cells present at the outset of starvation differ, as predicted, in the acidification of their acidic compartments, as well as to confirm that PST genes are induced in response to Ca²⁺ transients and PSP genes are repressed by them. Finally, one of the most interesting challenges will be to establish the mode of action of DIF-1.