Stage-specific requirement for myosin II during Dictyostelium development

The soil amoeba Dictyostelium discoideum responds to starvation by undergoing a well-characterized transition from unicellularity to multicellularity as it sporulates. It spends the vegetative portion of its life cycle in the form of single, motile amoeboid cells. Upon starvation, pulses of cyclic AMP induce the cells to aggregate into mounds consisting of approximately 105 cells each (Loomis, 1975). These mounds undergo a process of multicellular development that leads to the formation of fruiting bodies consisting of a spore head (sorus) supported by a several millimeter-tall stalk. The relative simplicity of this process in comparison with that of higher eukaryotes, along with recent evidence for cell sorting within mounds and the existence of gradients of diffusible morphogenetic regulators, makes Dictyostelium an attractive model system for the study of pattern formation and its regulation (Meinhardt, 1983; Williams et al., 1987; Esch and Firtel, 1991; Howard et al., 1992; Kay, 1992).

Dictyostelium has also proven to be a fruitful system with which to study the involvement of molecular motors such as myosin II in development (Spudich, 1989; in the remainder of this report, myosin II is referred to simply as myosin). Cells in which the single myosin heavy chain gene (mhcA) is truncated (De Lozanne and Spudich, 1987), completely deleted (Manstein et al., 1989), or inactivated by the expression of anti-sense RNA (Knecht and Loomis, 1987) are still motile but show several defects (Peters et al., 1988; Wessels et al., 1988). In addition to being unable to undergo the myosin-mediated processes of cytokinesis and capping, myosin null cells fail to develop past the mound stage (De Lozanne and Spudich, 1987; Knecht and Loomis, 1987; Pasternak et al., 1989). The cells aggregate, but their movements are more chaotic and limited than those of cells expressing myosin, and cell sorting occurs aberrantly within the resulting myosin null mounds (Eliott et al., 1993; Traynor et al., 1994; J. McNally, personal communication; D. Knecht, personal communication). Similarly, cells that lack two F-actin crosslinking proteins, α-actinin and gelation factor, aggregate to form terminal mounds (Witke et al., 1992).

Because null cells are blocked at the mound stage, their behavior tells us nothing about the requirements for myosin at later stages of fruiting body formation (culmination). However, aggregation is only the beginning of the impressively choreographed cell movements that comprise the rest of development. Therefore, in this study, we have used a recently-generated Dictyostelium mutant (strain HS80) carrying a cold-sensitive myosin heavy chain allele (B. P. and J. S., in preparation). At its permissive temperature of 26°C the HS80 mutant exhibits essentially wild-type behavior; whereas at its non-permissive temperature of 13°C, it exhibits all of the hallmarks of the absence of functional myosin: it does not undergo cytokinesis and therefore does not grow in suspension, it does not lose substrate adherence in response to treatment with sodium azide, and, most importantly, it does not develop into fruiting bodies. The sequence of the HS80 mhcA gene shows an arg→his change at amino acid position 562. Repair of this region by homologous recombination in vivo with the corresponding fragment of the wild-type gene restores the wild-type phenotype, thus confirming that the HS80 defect is a result of the mutant myosin heavy chain (B. P. and J. S., unpublished data). We now show that the loss of myosin function is rapid and reversible, which allows us to ‘turn off’ and ‘turn on’ myosin function at will during development by shifting cells between the permissive and non-permissive temperatures. In doing so, we have been able to examine the requirement for myosin during the remaining developmental stages, which until now have not been accessible. These experiments have led us to propose that there are two stages of Dictyostelium development during which myosin function is essential, and have allowed us to examine more carefully the role of cell sorting during the mound stage.

Dictyostelium strains Ax-2 (Watts and Ashworth, 1970), Del16-11 (Kalpaxis et al., 1991; kindly provided by T. Dingermann), and HS80 (B. P. and J. S., in preparation) were propagated in growth medium HL5 (Cocucci and Sussman, 1970) by growth either on the bottom of plastic Petri dishes at 21°C or in shaken culture at 21°C or 26°C. The myosin null strains HS2 (K. Ruppel and B. P., unpublished) and HS801 (HS80 with truncated mhcA gene; B. P. and J. S., unpublished data) were grown in HL5 in Petri dishes at 21°C. All strains were sometimes grown on lawns of Klebsiella aerogenes on SM/5 agar plates (Sussman, 1987).

The plasmids p63NeoGal and p56NeoGal, containing the coding sequence of E. coli LacZ fused to the promoters of the prestalk genes ecmA and ecmB, respectively, were kindly provided by J. Williams. The extrachromosomal plasmids pBIGA-gal and pBIGB-gal were derived from p63NeoGal and p56NeoGal, respectively, by excising the BamHI-XhoI fragment containing the fusion gene without terminator, adding the Dictyostelium actin 15 terminator (Knecht et al., 1986), and cloning into the plasmid pBIG (B. P. and J. S., in preparation) that had been cut with BamHI and XbaI. Plasmids were transfected into Dictyostelium cells using previously published methods for electroporation and calcium phosphate-mediated transformation (Howard et al., 1988; Nellen et al., 1984).

For development on agar, 10⁸ cells were harvested, pelleted at low speed in a clinical centrifuge for 2 minutes, washed in LPS (Sussman, 1987), and resuspended in 750 μl of LPS. The cell slurry was evenly plated on 2% agar LPS plates that had been dried for 1 day at room temperature (otherwise they were too damp and development was impaired), and the plates were incubated at 26°C, 21°C or 13°C. For photomicroscopy, the plates were placed in a clamp without their lids so that they were almost vertical, viewed with transmitted light using a Zeiss SV 6 stereomicroscope, and photographed using Kodak T-MAX 100 ASA black and white film.

Unfixed fruiting bodies and aggregates were stained to visualize cellulose by transferring them with fine forceps to a drop of 100 μg/ml Calcofluor (Sigma) in 10 mM sodium phosphate, pH 6.5, with 50 mM sucrose. Several short bursts in a bath sonicator were required to break up myosin null mounds. After 5 minutes, the unrinsed samples were viewed with fluorescence microscopy using a Zeiss Axiophot compound microscope equipped with a UV excitation filter set, and photographed using Kodak Ektachrome 100 ASA daylight color film.

Filter development was based on the method described by Sussman (1987). 5×10⁷ cells were harvested, pelleted and washed in LPS as described above, and resuspended in 500 μl of LPS. For each sample, a white Millipore 45 mm filter (type HA, catalogue no. HAWP047S0) was placed upon two 45 mm absorbent pads (Millipore, no. AP10047S1) in a 47 mm plastic Petri dish, and the filter assembly was saturated with LPS. After several minutes, excess LPS was withdrawn with a pipet and the cell slurry was plated on the filter. Newly excess LPS was removed and the dish was incubated over saturated paper towels inside a closed humid container at the desired temperature.

Fruiting bodies were fixed and stained with X-gal using previously published methods (Dingermann et al., 1989; Howard et al., 1992) with the modification that filters supporting fruiting bodies were first floated for 30 minutes on top of 5 ml of fixative solution consisting of 1% glutaraldehyde and 0.05% Triton X-100 in Z buffer (Dingermann et al., 1989), covered dropwise with fix solution until submerged and fixed for an additional 30 minutes. The fixation was performed in a 47 mm plastic Petri dish. In our hands, the extended fixation time improved the durability of the spore heads. The fruiting bodies were rinsed in Z buffer, stained overnight and rinsed in Z buffer. Fixed and stained fruiting bodies were picked off of the filter with fine forceps, placed in Z buffer on a glass slide between two coverslip supports and viewed with Nomarski optics with a Zeiss Axiophot compound microscope. Samples were photographed using Kodak Ektachrome 160 ASA tungsten color film.

HS80 cells were allowed to develop on Millipore filters as described above at 13°C. They were fixed and stained with X-gal as described above, without the floating prefixation step. After they had been stained and rinsed, the filters were put through an ethanol series consisting of 30%, 50%, 70%, 95% and 100% ethanol for 15 minutes per step. The series was followed by an additional 100% ethanol step for 30 minutes. At this point, the filters, which were composed of mixed esters of nitrocellulose and cellulose acetate, were quite soft, and small pieces were cut with scissors and transferred with a spatula to a 50%-50% mixture of ethanol and Epon-Araldite (Electron Microscopy Sciences). After approximately 1 minute, the mixture was replaced by 100% Epon-Araldite. The samples were allowed to infiltrate in a vacuum for 24 hours and were then polymerized for 2 days at 65°C. The filters dissolved in the resin, but the aggregates held their approximate positions within the resin and were therefore easily cut out in blocks after the plastic had polymerized. 5 μm sections were cut with a microtome, mounted with Permount (Fisher), and viewed and photographed as described above for fruiting bodies.

Strain HS80 was chosen for this study because out of the coldsensitive (cs) myosin mutants that were initially isolated (B. P. and J. S., unpublished data), it exhibited the largest difference in phenotypes at permissive and non-permissive temperature. At 26°C, development is essentially normal. Cells aggregate and develop through the normal intermediates into fruiting bodies that are somewhat shorter than those of a strain expressing wild-type myosin (Ax-2), but are nonetheless clearly recognizable as fruiting bodies (Fig. 1A). The difference between the mutant and wild-type strains is more noticeable when they are developed directly on starvation agar or Millipore filters than when they are grown and developed on bacterial lawns. At 21°C, the standard growth temperature for Dictyostelium, fruiting bodies are still formed; but these structures are markedly shorter than those formed at 26°C, and have a thicker stalk and a less prominent spore head (Fig. 1B). Frequently, but not always, the bottom of the stalk is surrounded by a mass of cells that seem to have been left behind by the ascending spore head (not shown).

At the non-permissive temperature of 13°C, normal fruiting bodies are not formed. Rather, many aggregates develop into balls such as those shown in Fig. 1C. Occasionally, a miniature version of a fruiting body can be observed in HS80 cells developing at non-permissive temperature; these are typically approximately one tenth the size of normal fruiting bodies. Some mounds do not undergo any further morphogenesis, but remain as mounds. For comparison, terminal mounds formed by the myosin null strain HS2 (K. Ruppel and B. P., unpublished data) are shown in Fig. 1D. The progression from mound to ball in HS80 at non-permissive temperature occurs through a tipped mound intermediate resembling those in normal development (Fig. 1E). However, this intermediate occurs one day after aggregation, whereas wild-type cells spend only several hours in the mound stage and then rapidly progress through the tipped mound stage and the rest of development.

The ability of HS80 to form fruiting bodies at 13°C is restored upon repair of the mutation by homologous recombination with the corresponding fragment of the wild-type mhcA gene (B. P. and J. S., unpublished data). Therefore, the HS80 defect is a result of the mutation in mhcA. However, cell aggregates of myosin null strains such as HS2 remain as mounds that do not develop into balls. In order to confirm that the ball phenotype is not caused by another mutation in strain HS80, we observed development at 13°C of an HS80-derived strain in which the myosin heavy chain gene has been disrupted by truncation (strain HS801; B. P. and J. S., unpublished data), effectively making it analogous to the original myosin disrup-tant strain isolated by De Lozanne et al. (1987) but with a mutant myosin head. The HS801 cells aggregated to form terminal mounds that did not develop into balls (not shown). Therefore, this effect is myosin heavy chain specific and is not caused by another mutation in the strain.

As controls (not shown), development of HS80 at 26°C, 21°C and 13°C was compared to development of the standard laboratory strain Ax-2 and to that of the parent strain of both HS80 and the myosin null strain HS2, strain Del16-11 (Kalpaxis et al., 1991). At 26°C and 21°C, both of these strains produced normal fruiting bodies. At 13°C, the fruiting bodies formed by both strains were for the most part normal, except for an occasional spore head from each strain that would be larger than normal and too heavy for the stalk to support. This confirms that the mound/ball morphology at 13°C is specific to HS80 and is not a characteristic of the parent strain as well.

The validity of our strategy to use HS80 to turn myosin function ‘on’ and ‘off’ in vivo during development depends on the following properties: the loss of function must occur rapidly and function must be rapidly restored when the cells are returned to the permissive temperature. We assayed for the loss and gain of myosin function using the same azide assay with which the HS80 mutant was originally isolated (B. P. and J. S., unpublished data). This assay takes advantage of the fact that wild-type cells grown on a surface respond to treatment with 2 mM sodium azide by increasing their cortical tension, rounding up and detaching from the dish, while myosin null cells do not respond in this fashion (Pasternak et al., 1989; B. P. and J. S., unpublished data).

The results of these experiments are summarized in Table 1. When HS80 cells were grown at the permissive temperature of 26°C and subsequently incubated at 26°C with a solution of azide in growth medium (HL5) that had been pre-equilibrated to the same temperature, most of the cells detached from the dish and floated away within 5 minutes, a behavior correlated with the presence of functional myosin. Likewise, when the cells were both grown and azide-treated at 13°C, most of the cells behaved like myosin null cells and remained attached to the surface in their normal flattened state. Both of these results were expected, because the mutant was isolated in precisely this way. When cells were grown at permissive temperature and then incubated at non-permissive temperature with preequilibrated azide/HL5, the vast majority of the cells remained adhered to the surface after 5 minutes. This indicates that myosin function was lost rapidly enough that the cells did not have a chance to detach before they lost their ability to do so. Conversely, when the cells were grown at non-permissive temperature and then treated at permissive temperature, most of them detached within 5 minutes, revealing that they had regained the ability to detach and had therefore presumably regained myosin function. This implies that the loss of function is reversible as well. For further confirmation, cells that had been grown at permissive temperature and treated at non-per-missive temperature, and were therefore adhering to the dish, were subsequently shifted back to permissive temperature. The cells detached from the dish, which suggests that we were able to turn myosin function ‘off’ and ‘on’ again in the same cells all within a period of 10 minutes.

Having demonstrated that, at least with regard to the azide assay, myosin function could be manipulated in HS80 cells by varying temperature, we attempted to reproduce the effect during development by allowing cells to develop at permissive temperature until after aggregation was complete and then turning off myosin function to see if it played a role in later developmental stages. Cells were induced to develop synchro-nously by plating them as a lawn on starvation agar. The agar in each Petri plate was cut in half with a sterile spatula and one half was transferred to a new empty plate, and each pair of plates was incubated at permissive temperature for various times. Every two hours, one half-plate was shifted to non-per-missive temperature and its sister half-plate was photographed using a dissecting microscope to chronicle the stage at which the cells had been shifted. After one day at the non-permissive temperature, the other half-plate was photographed to record the final phenotype that had been achieved while myosin function was compromised. The results of this experiment are shown in Fig. 2 and summarized schematically in Fig. 3.

Fig. 2 shows the time and developmental stage at which each plate was shifted to the non-permissive temperature, and the corresponding final stage that each cell lawn reached after one day without functional myosin. When cells were shifted during aggregation or during the first half of the mound stage (hours 0–10), the final structures formed at 13°C were identical to the mounds/balls made by cells developed entirely at that temperature, which is consistent with the observation that the first developmental block in the absence of functional myosin is at the mound stage (Knecht and Loomis, 1987; De Lozanne and Spudich, 1987). However, if the cells were allowed to develop at the permissive temperature until the second half of the mound stage (hour 12) and then shifted, the final structures made at 13°C were very short, stumpy versions of fruiting bodies, averaging approximately 100 μm in height. The structures had small but distinct spore heads on thick neck-like miniature stalks approximately 25 μm thick. These cells developed at non-permissive temperature through the normal developmental intermediates until the stage immediately preceding final culmination (not shown). Consistent with this observation is the fact that if development was allowed to proceed at permissive temperature until any stage between the late mound stage and the stage right before final culmination (hours 12–18), the final structures formed after the temperature shift were the same stumpy mini-fruiting bodies. Only if development was allowed to proceed at permissive temperature all the way through the final culmination step (hours 20–24) did HS80 cells make fruiting bodies of normal size and shape, averaging 300 μm in height with 10 μm thick stalks. It should be noted that, because perfect synchrony of development is difficult to obtain, there were always a few aggregates that developed slower than the rest at both temperatures. However, the majority of the aggregates were able to develop to the stages described. These results are summarized in Fig. 3, in which the late aggregation/early mound stage and the final culmination stage are referred to as the two myosin-essential periods.

When cells began development at non-permissive temperature but were shifted to permissive temperature before the beginning of the first myosin-essential period, they were able to develop into mature fruiting bodies. Likewise, when cells were allowed to develop at permissive temperature past the first apparent myosin-essential period, shifted to non-permissive temperature and allowed to develop until the beginning of the second apparent myosin-essential period, and then shifted back to permissive temperature, normal fruiting bodies were formed. This implies that the loss of myosin function is reversible during development. Thus, mirroring the azide assay results, we were able to turn myosin function off when it was not essential and then on again when it was essential, resulting in the formation of fruiting bodies identical to those formed in cells possessing functional myosin throughout development (Fig. 3).

Cells that are devoid of functional myosin aggregate into mounds that undergo no further morphological changes. However, HS80 cells at non-permissive temperature form mounds that develop further into balls. The reason for this difference is unclear (see discussion). In order to examine the nature of these balls, they were stained with the fluorescent dye Calcofluor to visualize any spores or stalk material that may have formed. Calcofluor binds to cellulose, which is both a major component of the spore coat and the main structural reinforcement of the stalk tube (Harrington and Raper, 1968). As controls, fruiting bodies from both Ax-2 and HS80 that were developed at permissive temperature were stained as well. Unfixed fruiting bodies and mounds/balls were placed in a drop of Calcofluor and viewed with fluorescence microscopy.

Fig. 4A and B show fruiting bodies of Ax-2 and HS80, respectively. The spore heads dispersed in the solution, but many fluorescently labeled spores were still visible surrounding the brightly staining stalks. The stalks formed by HS80 at 26°C were again noticeably shorter and thicker than those formed by Ax-2. The Ax-2 spores were of the normal elliptical shape; however, the HS80 26°C spores were round. This effect, which may be a result of the HS80 myosin being slightly defective even at permissive temperature, was less pronounced when the cells were grown and developed on bacterial lawns than when they were starved and developed on filters or agar. The base of an HS80 26°C stalk tube is shown in Fig. 4C.

When we stained and viewed HS80 13°C balls, they were squashed slightly by the coverslip and released a mass of stained cells that were indistinguishable from spores found in HS80 26°C fruiting bodies (Fig. 4D,E). There was also always a subset of unstained cells that was usually in a cluster (most visible in Fig. 4D). A brightly stained tangle of material was invariably found in these balls. This material showed no internal organization and did not contain the patchy cellular pattern characteristic of stalks.

We also stained mounds made by the myosin null strain HS2 and by strain HS801, the HS80 myosin truncation strain discussed earlier. The properties and staining patterns of mounds formed by each strain were identical; only HS801 is shown (Fig. 4F). These mounds were much harder to dissociate than were the balls made by HS80 and were unaffected by vortexing. This was most likely due to a stained sheath that appeared to hold most of the cells together in the mounds. Only after several seconds in a bath sonicator did the mounds open to release stained round cells identical to those found in the HS80 balls and fruiting bodies.

The presence of stained round cells in HS80 balls and myosin null mounds does not necessarily mean that normal spores are produced by these structures. We examined the stained round cells further to test whether they were surrounded by a cellulose spore coat, osmotically stable in water, and able to remain viable after a 30 minute incubation at 45°C, all of which are properties of normal spores (Cotter and Raper, 1966). We did not attempt to use spore-specific gene expression as a criterion because it is already known that myosin null mounds express spore genes in the absence of fruiting body formation (Knecht and Loomis, 1988). The issue of a Calcofluor-stainable spore coat has already been addressed in Fig. 4. Examination by microscopy confirmed that the cells are stable in distilled water, whereas vegetative amoebae swell and can ultimately lyse. However, the round cells were killed by heat treatment that had no significant effect on Ax-2 spores or spores obtained by development of HS80 cells at permissive temperature (Table 2). As expected, heating killed 100% of the vegetative amoebae. Therefore, we conclude that, while the round cells found in HS80 13°C balls are osmotically stable and possess a cellulose-containing cell coat, they have not completed the processes necessary for the acquisition of heat resistance, and thus are not normal spores.

We did not attempt similarly to examine cells derived from myosin null mounds, because it was very difficult to separate or disperse these cells. However, we did not deem this experiment necessary, because we felt that it was highly unlikely that these cells would possess heat resistance abilities that spores from the further-developed HS80 13°C balls lacked.

Within a single synchronized population of developing HS80 cells at non-permissive temperature, some aggregates remain mounds while others develop further into balls. Since recent work by others has suggested a role for myosin in cell sorting during or after aggregation (Eliott et al., 1993; Traynor et al., 1994), we set out to ascertain whether cell sorting could play a role during the first myosin essential period or in the decision of HS80 at 13°C to make mounds or balls. Towards this end, we transformed HS80 with plasmids containing the structural gene for E. coli β-galactosidase fused to the promoters of the prestalkspecific ecmA and ecmB genes (p63NeoGal and p56NeoGal; kind gifts of J. Williams). These reporter gene fusions have proven to be useful markers of cell sorting in both Dictyostelium (Williams et al., 1989; Jermyn and Williams, 1991) and Polysphondylium pallidum (Vocke and Cox, 1992). Using these gene fusions, it has been observed that, while prestalk cells differentiate at random positions within both Ax-2 and myosin null aggregates, they sort to the tips of Ax-2 mounds and to the central core of myosin null mounds (Traynor et al., 1994).

When HS80 cells containing either type of plasmid were allowed to develop at permissive temperature, the expression patterns of ecmA and ecmB were similar to those exhibited by Ax-2 (Jermyn and Williams, 1991). Examples of the expression patterns of both the integrating and the extrachromosomal ecmB constructs are shown in Fig. 5A and B. These fruiting bodies had been fixed in glutaraldehyde and stained with X-gal. The stalks are heavily stained, as well as the cells at the top and bottom of the spore head (upper cup cells and lower cup cells). The lighter staining in Fig. 5B makes it possible to see the highly vacuolated nature of the stalk.

When HS80 cells containing these plasmids were allowed to develop at non-permissive temperature, they formed a mixed population of mounds and balls. Representative aggregates of all types were fixed, stained, embedded in resin and completely sectioned at 5 μm per section. Some mounds were quite flat, resembling myosin null mounds. These mounds possessed a central core of stained but non-differentiated cells (Fig. 5C) that was identical to the cores that have been observed in myosin null mounds (Traynor et al., 1994). It is likely that these flat mounds correspond to the terminal mounds that do not develop further. Other mounds developed further, to the tipped mound stage and sometimes beyond. In these cases, the upper region was stained and there was no stained central core (Fig. 5D). The sorting of prestalk cells to the top or center of the aggregates appeared to correlate with the ability of the aggregates to develop further.

Taller mounds were also observed that possessed cores of darkly stained cells that were highly vacuolated and resembled stalk cells (Fig. 5E, F). The positions of these cores were variable; they were observed both in the center of some mounds and closer to the base of others. Regardless of the position of the core, these taller mounds also possessed a separate cluster of lightly stained, non-differentiated cells. Some of these taller mounds (Fig. 5E) also contained large numbers of non-stained, round, refractile cells that were most likely the defective spore cells that had been observed with Calcofluor staining; other mounds in this class did not appear to contain any spores (Fig. 5F). The presence of a central core of stalk cells suggests that these tall mounds may have been derived from mounds in which prestalk cells sorted to the center (such as that in Fig. 5C) and then partially differentiated, along with spore cells, in the absence of morphological development.

Some partially rounded aggregates exhibited dark staining at their top and contained spores below (Fig. 5G). It is not clear whether or not these structures had completed their development, or how they relate to the other structures observed. However, the balls that constituted the majority of the structures resulting from development at 13°C all displayed a prominent cylinder of darkly stained, highly vacuolated cells that usually ran up one side of the ball and curled inwards (Fig. 5H,I). The upper end of the cylinder was in contact with a cluster of cells that were not differentiated according to any visible criteria, and were not stained except for those cells on the surface of the ball. The rest of the mass of each ball was inhabited by spore cells. The HS80 13°C balls, therefore, appear to be the end result of attempted culmination that was hindered by lack of functional myosin.

The studies of De Lozanne and Spudich (1987) and of Knecht and Loomis (1987) provided a link between a known molecular motor and morphogenesis on a multicellular level. Subsequently, Kiehart and colleagues (Young et al., 1993) demonstrated that myosin II plays an essential role in Drosophila embryonic development. The idea that motor proteins such as the myosins should play a role in multicellular development is not surprising. Morphogenesis is based on movement, and myosin II plays well-characterized roles in intracellular changes in morphology such as cytokinesis and muscle contraction. It is less clear, however, why motor proteins that are not essential for the motility of specific cells can nevertheless be essential for the correct movements of those cells during development. Our results suggest that, in Dictyostelium fruiting body development, myosin II can play both a role in multicellular pattern formation and a stage-specific role in aiding the cell movements that comprise culmination.

The tool that we have chosen to use to address this question is the conditional myosin mutant, HS80. The nature of the mutation in HS80 is still unclear. The mutation is a single amino acid replacement of a histidine for an arginine that normally occupies position 562 of the Dictyostelium myosin heavy chain amino acid sequence (B. P. and J. S., unpublished data). This position is usually occupied by lysine or arginine in various species (Warrick and Spudich, 1987). According to three-dimensional reconstructions from electron micrographs of the actomyosin complex, the mutation is in a region that has been implicated in forming a potential secondary actomyosin interaction (Rayment et al., 1993a,b; Schröder et al., 1993), although it is not yet known whether the mutant myosin is impaired with respect to actin binding. The rapidly reversible nature of the cold-sensitive phenotype rules out cold-induced irreversible denaturation or improper folding during synthesis as possible explanations for the defective phenotype. This is a crucial property of the mutation, as it allows the switching on and off of myosin function that is the basis for the temperature-shift experiments that we have conducted.

Our observations of the various patterns of prestalk cell sorting in strain HS80 suggest a role for myosin II in the positioning of cells within the aggregate. Such a supposition is supported by findings from several groups. Eliott et al. (1993) observed that, while aggregating cells expressing wild-type myosin take on a specific elongated shape and pack closely in parallel within their aggregation streams, aggregating myosin null cells have a more random shape and follow each other in a less organized fashion. Myosin null cell movements within the resulting mounds are chaotic and limited compared to those of wild-type cells (Eliott et al., 1993; J. McNally, personal communication), unlike the large-scale spiral movements associated with wild-type aggregation (Clark and Steck, 1979). Experiments using fusions of the lacZ reporter gene to prestalk promoters have revealed that prestalk cells differentiate at random positions in aggregates of both Ax-2 and myosin null cells. These cells still sort in myosin null aggregates, but to the center of the mound instead of to the tip (Traynor et al., 1994). The picture that is emerging is that, potentially by exerting an effect on cell shape, myosin may be responsible for the establishment of cell-cell contacts during aggregation that permit the cells to aggregate correctly. Mounds in which aggregation has occurred in the absence of functional myosin would therefore not be able to continue with development because the cells did not end up in the right positions. The failure to set up the correct pattern during aggregation may be linked to the inability of prestalk cells to sort correctly within the mound.

Several lines of evidence support this idea. Myosin null cells exhibit decreased cortical tension (Pasternak et al., 1989), which should affect cell shape as well. Dictyostelium cells that lack either the myosin essential light chain or a pair of actin cross-linking proteins also fail to develop past the mound stage (Pollenz et al., 1992; Witke et al., 1992). All of these essential proteins would be expected to mediate cortical tension. Another piece of supporting evidence is the observation that the destination of prestalk cell sorting correlates with the ability of mounds to undergo further development (Fig. 5). This is consistent with the hypothesis that development only proceeds if cells have ended up in the correct positions. Furthermore, the temperature-shift experiments presented in Figs 2 and 3 demonstrated that when myosin function is lost during the 2nd half of the mound stage, development proceeds unimpeded until the last developmental stage. Thus full myosin function is not essential during the second half of the mound stage, by which time the cells have already sorted to the correct positions such that development can proceed further.

We propose that during the first myosin essential period, myosin plays a role in directing cell sorting within the aggregate in such a way as to allow further development to occur. Its role during the second myosin essential period is much less clear, owing to the lack of data from cells lacking functional myosin. The temperature-shift experiments with HS80 demonstrate that myosin is essential for the final stage of culmination, in which the stalk elongates with the sorus of spores riding to the top.

This manoeuver is accomplished by the formation of a stalk tube that pushes down through the center of the cell mass. By following what has been described as a reverse fountain movement (Bonner, 1967), the cell mass both contributes to the growing stalk and climbs it to remain at the top. As the cells ascend, fibers have been observed in the region of the cytoplasm closest to the stalk (George et al., 1972). It is possible that this fibrous region consists of an actin-based cortical network. Myosin may act as a motor that drives the process of culmination, or it may continue to mediate cell shapes and contacts that are critical for correct pattern formation, as appears to be the case during the first myosin essential period. When myosin function is absent during the final stage of culmination, structures are formed that are miniature versions of fruiting bodies; therefore, the pattern of development has been correctly, but weakly, implemented. This would argue against myosin playing a role in directing cells to their correct positions, in that the positions appear to be reached in its absence. If myosin then plays a motor role, cells lacking the motor must still be able to undergo limited movements sufficient to set up a stalk/spore head pattern, but only on a small scale.

There is one caveat that deserves consideration. The fact that HS80 at non-permissive temperature can form both mounds and balls within the same developing culture implies that there may well be a small degree of residual myosin function very close to the threshold that is necessary to promote near-correct sorting of prestalk cells during the first myosin essential period. Since the ball phenotype is specific to the myosin heavy chain mutation and not to another mutation in HS80, we can speculate in the following manner based on decreasing ability of cells to undergo sorting: (1) mounds that have undergone correct sorting at permissive temperature may, upon removal of myosin function, give rise to small but recognizable fruiting bodies; (2) those mounds that have undergone near-correct sorting at non-permissive temperature are only able to develop into balls surrounding miniature stalks in their attempt to make fruiting bodies, and (3) those mounds in which correct sorting has not occurred at all at non-permissive temperature remain as mounds.

One consequence of the aberrant development that leads to ball formation is that prespore cells do not differentiate into normal spores as defined by the ability to resist heat treatment. While it proved to be impractical to apply the same test to cells embedded in myosin null mounds, the idea that myosin null mounds contain spore cells that are incompletely differentiated is backed up by several observations in the literature. Knecht and Loomis (1988) showed that prespore and spore genes are expressed in myosin null mounds; and Elliot et al. (1993) report that the prespore-specific antibody MUD1 stains many cells in myosin null mounds, while the spore coat-specific antibody MUD3, which should stain only mature spores, does not stain cells in these mounds. It is interesting to note that the SP96 spore coat protein, which is the antigen that MUD3 recognizes, is present in myosin null mounds according to protein immunoblot analysis (Knecht and Loomis, 1988) but is not visualized by MUD3 immunofluorescence (Eliott et al., 1993), suggesting that the protein is made but not incorporated into the spore coat in the absence of myosin and proper morphogenesis.

Our data suggest that myosin function is required at least twice during Dictyostelium development, first during aggregation to mediate cell sorting, and again during final culmination to aid in the drastic cell movements that occur at that stage. Furthermore, a small degree of myosin function appears to be sufficient to cause enough cell sorting during aggregation to allow the formation of a miniature stalk, but the resulting structure is not capable of raising a sorus and the stalk remains embedded within the cell mass. Finally, we suggest that aggregates that either lack prestalk cells or possess them in the wrong positions due to defective myosin can still produce prespore cells that differentiate along the spore development pathway, but are incapable of completing differentiation into mature spores.

We are grateful to Jeffrey Williams for supplying lacZ fusion constructs, Theodor Dingermann for supplying the Del16-11 strain, and Jim McNally, David Knecht and Jeffrey Williams for communicating results before publication and critically reviewing this manuscript. This investigation was supported by PHS grant number 5T32CA09151-18 and PHS grant number 5T32CA09302-15 awarded by the National Cancer Institute, DHHS, to M. L. S.; grant number DRG1002 from the Damon Runyon-Walter Winchell Cancer Fund and PHS grant number 5F32AR08235-02 from the National Institutes of Health to B. P.; and grant GM-40509 from the National Institutes of Health to J. A. S.