Patterns in Dictyostelium discoideum: the role of myosin II in the transition from the unicellular to the multicellular phase

Animal morphogenesis involves precisely timed cell differentiation and cell migrations. For co-ordinated cell movements to occur during morphogenesis, cells must be able to respond to extracellular signals and alter their shape and movements accordingly. There is increasing evidence that the cytoskeleton is involved in many aspects of morphogenesis including cell migration, cytodifferentiation and developmental gene expression (Ben-Ze’ev, 1991, for review). We are interested specifically in the role of myosin II in animal morphogenesis. Actin and myosin II form a contractile unit that is thought to be involved in the folding of epithelial cell sheets during gastrulation in chick (Lee et al., 1983) and Drososphila embryos (Young et al., 1991; see also Odell et al., 1981) and the compaction of the mouse morula (Sobel, 1983, 1984).

We use the cellular slime mould, Dictyostelium discoideum, to study the cytoskeleton and cell-cell interactions during development of a simple three-dimensional tissue (Williams et al., 1986; De Lozanne and Spudich, 1987; Eliott et al., 1991). D. discoideum is a soil amoeba that can exist as single cells or as a structured multicellular aggregate, depending on the stage of the asexual lifecycle. Molecular genetic techniques involving the use of homologous recombination have made it possible to construct myosin II null cells from which the myosin heavy chain gene has been excised (Manstein et al., 1989). Though the mutant amoebae are able to migrate on a surface and aggregate to form rudimentary mounds, further development is blocked; these cells are unable to form the organized multicellular ‘slug’ or mature fruiting body (Knecht and Loomis, 1987; De Lozanne and Spudich, 1987; Wessels et al., 1988; Peters et al., 1988). This implies an important role for myosin II in tissue formation. The timing of developmental gene expression in the mutant is apparently normal (Knecht and Loomis, 1988), indicating that the block in morphogenesis is due to a spatial or mechanical abnormality.

Using immunofluorescence techniques, it has been shown that the myosin II protein is distributed unevenly in cells of D. discoideum wild-type slugs, being concentrated in the cortex of peripheral and anterior cells, and evenly through-out the cytoplasm in the inner posterior cells (Eliott et al., 1991). This distribution corresponds with the prestalk-prespore pattern, a finding that is consistent with prestalk cells providing the major motive force for slug movement (Inouye and Takeuchi, 1980; Williams et al., 1986; Voet et al., 1984; Yumura, 1992). Cells at the periphery of the slug are polarised with respect to myosin II, resembling individual motile amoebae. We suggest that these cells have a specialised role in the morphogenesis and migration of the whole organism.

Here we study the aggregation and development of two myosin null mutant strains, HS2205 and HS2206, in order to elucidate the role of myosin II in normal D. discoideum development. We show that cells lacking myosin II are able to form aggregation streams that appear similar to wildtype strains on casual observation. However, cell shape is abnormal in both streams and mounds. Myosin null cells do not undergo the coordinated spiralling movements observed for wild-type cells as they approach the aggregation centre. Although coordinated movement is absent in mutant mounds, a distinctive prespore pattern is established. Mutant mounds are unable to form tips and, when mixed with wild-type cells, are absent from the tips of chimeric slugs and preculminates, suggesting that myosin II may be involved in the process of tip formation (a factor that may be crucial in the inability of myosin null cells alone to progress morphologically beyond the mound stage).

For light, immunofluorescence and scanning electron microscopy (SEM) and synergy experiments, the D. discoideum myosin null mutants, HS2205 and HS2206 (Manstein et al., 1989) and two wild-type axenic strains, AX3 (Loomis, 1969) and AX2 (Watts and Ashworth, 1970), were used. Amoebae were grown on slime mould nutrient agar (SM) with Klebsiella aerogenes . To develop aggregates, cells were collected from a lawn of K. aerogenes, washed three times in Bonner’s salts solution and 2-25 μl drops at a density of between 2 × 10⁶ and 4 × 10⁶ cells/ml were pipetted onto water agar (1% Noble agar in distilled water with 250 μg/ml dihydrostreptomycin sulphate) which formed thin (i.e. 2.75 ml agar: for filming and photography) or thick (i.e. 30 ml agar: for scanning electron and immunofluorescence microscopy) layers in 9 cm petri dishes. Drops were air dried for 5-10 min and cells allowed to develop in an illuminated room at 21(±1)°C and 70-80% relative humidity. In the case of thin water agar plates, agar within each Petri dish was kept moist with circular strips of filter paper soaked in Bonner’s salt solution.

Aggregating amoebae of strains AX3 and HS2206 were developed on thin water-agar plates as described above and photographed with an Olympus camera (PM10.ADS) attached to a Zeiss Universal microscope.

Myosin null cells take much longer to aggregate than the parental wild-type strain (see Manstein et al., 1989; Knecht and Loomis, 1987, 1988; De Lozanne and Spudich, 1987). HS2205 mounds were fixed after all streams had either collected into the mounds or begun to break up (between 17 and 19 h depending on initial cell density). For comparison, tipless aggregates of AX2 cells were developed as described above and fixed after 7 h development. All mounds were vapour fixed in 1% (v/v) glutaraldehyde and 2.5% (w/v) paraformaldehyde in starvation buffer (25 mM MES, 2 mM MgSO₄, 0.2 mM CaCl₂, pH 6.8) for approximately 2 h. After fixation, plastic replicas were made of the mounds using Exaflex hydrophobic dental impression material (Kerr) as described by Williams and Green (1988) for plant meristems. This technique requires minimal fixation and handling of tissue and thus reduces artefacts due to conventional preparation for SEM. The replicas were coated with gold and viewed under a Philips 505 scanning electron microscope.

For video observation and analysis of cell movements during aggregation, cells were filmed using a low-light-sensitive camera (Mintron, CCD, MTV-1801CB) attached to a Zeiss OPM1 microscope fitted with a f100 or f150 mm lens and recorded on a National time-lapse cassette recorder, NV-8051. Myosin null cell streams and aggregates were filmed for 2 to 8 h periods within a time frame of 12 to 24 h after plating. AX2 and AX3 streams and aggregates were filmed for 30 min to 5 h periods within a time frame of 5 to 12 h after plating.

Speed of spiral motion (min per revolution) in aggregation mounds was determined from time lapse video segments by measuring the amount of time taken for a number of cells in the outer third of a rotating aggregate to move 1/4 or 1/2 way around the circumference of the mound.

For computer analysis, images taken at 10 s intervals using the above microscope and camera were stored directly in a 386 computer using a Matrox PIP(1024) image board. Cell movements were tracked using a program which maps individually selected points through consecutive frames (Breen et al., unpublished data). This program measures flow of identifiable features rather than delineating the cells and tracking them individually. Cell-cell boundaries and intracellular vesicles were chosen as points to be tracked because they could be easily seen and tracked for several minutes.

HS2206 amoebae (prepared as described above) were allowed to develop for 19, 23, 30 and 65 h at 21(±1)°C before fixing in 2.5% (w/v) paraformaldehyde in PBS (0.15 M potassium phosphate; 0.9% (w/v) NaCl, pH 7.2) for 15 h at room temperature. Mounds were then encased in molten water agar and cubes bearing 1 to 4 mounds were washed in PBS overnight at 4°C. Cubes were infiltrated with OCT embedding compound (Miles-Tissue Tek II) for 2 days at 4°C, and 3-4 μm sections perpendicular to the agar surface on which the mounds had developed were cut at −22(±2)°C and picked up onto slides treated with chrome alum-gelatine (Krefft et al., 1984). Slides bearing frozen sections were washed in PBS for approximately 2 h and blocked with 5% (w/v) skim milk for 30 min. After three 5-min washes in PBS, sections were labelled with MUD3 supernatant (a mouse monoclonal antibody that recognises a spore coat protein (SP96; Voet et al., 1985) or MUD1 (a monoclonal antibody that recognises glycoprotein PsA on the surface of prespore cells; Krefft et al., 1983) for 1 h and washed in PBS. Sections were blocked a second time for 10 min, washed again in PBS (2 × 10 min) and labelled with FITC-conjugated sheep anti-mouse IgG (Sigma), diluted 1:25 with PBS, for 45 min. After washing in PBS (3 × 20 min), sections were mounted with Aquamount (Gurr). Control slides were treated as described above except that they were incubated with the second antibody only.

AX2 and HS2205 amoebae were collected from bacterial lawns and washed as described above. AX2 or HS2205 cells were stained with a fluorescent membrane dye, DiIC16 (Molecular Probes; 20 μg/ml) in starvation buffer on a shaker for 15 min at a total cell concentration of approximately 10⁶ cells/ml. Stained and unstained cells were washed four times in starvation buffer. Unstained AX2 cells were mixed with stained HS2205 cells at a ratio of 8:1. Stained AX2 cells were mixed with unstained AX2 cells at a ratio of 1:1 as a control. Mixed cell suspensions were centrifuged (2000 revs/min) and the cell pellets vortexed and pipetted onto water agar plates. Petri dishes were enclosed in square black polyvinyl chloride containers in an illuminated room at 21(±1)°C and 70-80% relative humidity. Slugs developed and migrated towards light entering a 3 mm hole in the side of each container.

Chimeric slugs and pre-culminates were fixed in 3% paraformaldehyde and 0.02% glutaraldehyde in starvation buffer for 2 to 6 hours, washed in PBS and whole mounts observed under a fluorescence microscope.

To check for the stability of DiIC16 over time, stained AX2 cells were observed under a fluorescence microscope at regular intervals over a period of 30 hours. After approximately 12 hours much of the dye had been internalised and was seen as vesicles inside the cell, which nevertheless remained strongly fluorescent. After 24 hours, however, a large proportion of spores appeared to have lost the dye. To check for transfer of dye to unstained cells, stained cells were mixed with unstained cells in known proportions and cells were counted at regular intervals. Only very slight staining of some unstained cells was observed after prolonged periods (this could have been due to phagocytosis by unstained cells of fluorescent cell debri and other particles).

With low power observation it could be seen that myosinnull cells were able to form long continuous aggregation streams (Fig. 1B) similar to AX3 (Fig. 1A). At high power, however, it was apparent that the mutant cells (Fig. 1D) did not have the elongated morphology and parallel orientation of cells in wild-type aggregation streams (Fig. 1C). HS2206 cells in streams were more irregular in their shape and size and more flattened (Fig. 1D) when compared to AX3 cells, which exhibited a more three-dimensional cylindrical appearance (Fig. 1C). During later aggregation, wild-type cell streams collected into the central aggregate towards which they were migrating, while many HS2206 streams broke up, eventually forming many smaller aggregates. The beginning of this breakup is apparent in Fig. 1B.

It has often been observed that cells in aggregation streams form spirals as they approach the aggregation centre (Clark and Steck, 1979). Spiralling motion was observed in 19 of the 20 AX3 and AX2 aggregates filmed by time-lapse photography. Cells within the AX3 mound that did not spiral moved towards the aggregation centre in concentric circles. Whether or not spiralling commenced later in aggregation in this exceptional aggregate was not followed. On the other hand, only 2 of the 18 HS2206 aggregates filmed underwent significant spiralling motion. In these two cases, spiralling was observed during early aggregation (between 12 and 15 h development) of HS2206 cells but was short-term, seldom lasting more than 12 min (or one quarter of a revolution), and the cells frequently changed direction. The speed of spiral motion was determined for one of the mutant aggregates and a wild-type aggregate with the same diameter (200 μm). Spiralling motion in the HS2206 aggregate was slower (49 min per revolution) than in the AX3 aggregate (21 min per revolution).

Spiralling motion was observed to continue in AX3 and AX2 aggregates as the wild-type mounds underwent further morphogenesis, extending upwards into standing fingers (also observed by Clark and Steck, 1979). Spiralling was never observed to cease during the period of filming of such aggregates. Any co-ordinated movements observed in early HS2206 aggregates, on the other hand, gave rise to chaotic movements that finally ceased altogether after approximately 19 h; at this stage only membrane ruffling and intracellular particle movement could be seen. As cell migration stopped within a mutant mound, any incoming streams still attached to the mound also ceased moving towards the aggregation centre and began to break up and form separate smaller aggregates.

Movement in wild-type and mutant strains was analysed in detail using a computer tracking program and representative data is shown in Fig. 2. The HS2206 aggregate shown in Fig. 2B,D was filmed 12 h after plating and was representative of the majority of mutant aggregates, which did not show any spiral motion. HS2206 cells moved chaotically once within the mound and no apparent coordination of cell movement was indicated during the period of analysis (Fig. 2D). Spiralling motion was clearly observed in the AX3 aggregate analysed (Fig. 2A,C).

Once aggregation is complete, wild-type aggregates appear as smooth hemispherical mounds which then form tips and undergo further morphogenetic movements that give rise to the slug and fruiting body. Scanning electron micrographs of plastic replicas indicated that mature myosin null mutant aggregates are irregularly shaped cellular masses (Fig. 3B,D) without the smooth surface or ordered alignment of cells apparent in their wild-type counterparts (Fig. 3A,C). This lack of organisation was apparent during the whole period of mound formation and at no stage was a tip observed to form on mutant aggregates. The example illustrated for HS2205 had a layer of slime sheath covering the mound, indicated by cells with shrunken sheath stretched over them (Fig. 3D). This was confirmed by immunofluorescence on cryosections of HS2206 mounds using an antibody (MUD62) that labels sheath material (data not shown). Low-temperature scanning electron microscopy was also employed to confirm that the dental impression material used did not significantly alter the shape of aggregates. In this technique aggregation mounds were plunged into liquid nitrogen, transferred to a cryotransfer system and examined in a Cambridge 3600 Stereo scanning electron microscope. The results confirmed those observed with the replica technique, but they are not shown here, due to the formation of ice crystals, which partially obscured the image.

Disruption of the myosin heavy chain gene in D. discoideum apparently does not significantly affect the expression of a range of developmentally regulated genes, despite the delay in aggregation and the morphogenetic aberrations of these cells (Knecht and Loomis, 1988). Thus spore and stalk cells are expected to be formed eventually within mutant mounds. In wild-type strains, terminal differentiation of spore and stalk cells is normally complete within 24 h from the onset of starvation (Knecht and Loomis, 1988). To determine the pattern and timing of prespore and spore cell differentiation in myosin null aggregates, frozen sections of 19-, 23-, 30- and 65-h-old mounds were labelled with a prespore-specific antibody, MUD1 and a spore coat-specific antibody MUD3 (identifying the SP96 antigen). At least 6 mounds from each stage were viewed and representative examples are shown in Fig. 4. After 19 h development, negligible MUD1 and no MUD3 labelling was apparent in mutant mounds (Fig. 4A,B). By 23 h MUD1 labelling indicated the presence of prespore cells in the outer regions of mounds (Fig. 4C). At this stage MUD3 labelling was still absent. At 30 and 65 h, MUD1 labelled the entire centre of aggregates but was absent from the base and a single cell layer across the top of mutant mounds (Fig. 4E,G,I,K). MUD3 labels prespore vesicles inside prespore cells and the surface of spores in wild-type fruiting bodies (Voet et al., 1985). In myosin null mutant mounds, however, labelling of prespore vesicles was weak after 30 h (Fig. 4F) and surface labelling of mature spores was not often observed, even after 65 h (Fig. 4F,H). Cells with MUD3 labelling of intracellular vesicles were labelled strongly by MUD1, indicating that most prespore cells failed to differentiate normally into mature spores, resulting in a low spore to nonspore cell ratio. When mature spores were apparent, they varied considerably in their size and shape (Fig. 4J,L). Control sections incubated with second antibody alone showed no labelling.

Multicellular development of myosin null cells can be rescued if they are mixed with sufficient wild-type cells at the onset of starvation (Knecht and Loomis, 1988). We synergised AX2 cells with stained HS2205 cells in order to observe where the myosin null cells go within chimeric slugs and pre-culminates. Since much of the fluorescent dye appears to be lost from spores during sporulation, we did not check mature fruiting bodies.

Slugs made from mixtures of HS2205 and AX2 cells were very long, with tails that extended back to the site of aggregation by as much as 1 cm. A gradation of stained null cells was observed along the length of chimeric slugs, being absent from the extreme tip and more numerous towards the rear (Fig. 5A,B). The boundary between unstained cells in the tip and stained cells at the base of the tip was distinct (Fig. 5B). In some cases, however, very few mutant cells were observed in the entire front half of the slug. Many slugs were seen to leave behind trails and clumps of cells, some of which underwent varying degrees of further development but rarely formed slugs or fruiting bodies.

During culmination of synergised slugs, the stained HS2205 cells were still absent from the tip and some appeared to collect into a kind of girdle at the base of the tip (Fig. 5E,F). The absence of mutant cells from the tips of slugs and early culminates was seen in all aggregates observed from three separate experiments. In control experiments in which stained AX2 cells were mixed with unstained AX2 cells, stained cells were distributed along the entire length of chimeric slugs (Fig. 5D).

Although myosin null cells are generally inefficient in their chemotactic response, they are sufficiently coordinated in their movements to form two-dimensional streams. Essentially all coordination ceases, however, once cells are within the multicellular aggregate. This implies that other molecules can be involved in cell interactions in two-dimensions, but that myosin II becomes essential for proper cellcell interactions in a three-dimensional array. The inability of the mutant cells to undergo coordinated movement and progress beyond the mound stage may be due to the cells having insufficient strength to generate sufficient force for the coordinated movement of large groups of cells, and/or could relate to confused chemotactic signalling within the cell mass.

Myosin null cells form mounds that are roughly hemispherical but lack the smooth surface and ordered appearance of wild-type mounds at an equivalent stage of development. The irregular surface of mutant mounds probably reflects the lack of cortical tension within myosin null cells as described by Pasternak et al. (1989). Myosin II is concentrated in the posterior and outer lateral cortex of wildtype cells at the periphery of the early aggregate, suggesting that, in normal development, these cells could exert a centripetal force on the aggregation centre (Yumura et al., 1984); further morphogenetic events may rely on the strength (i.e. cortical tension) of these outer cells. This is consistent with recent evidence from D. discoideum double mutants that lack two F-actin cross-linking proteins, alpha actinin and gelation factor, and thus have greatly reduced cortical viscoelasticity (Witke et al., 1992). These cells aggregate normally but rarely develop beyond the tipped aggregate stage.

Individual myosin null cells have a less polarised shape than wild-type cells, and their movements are consequently less directed (Peters et al., 1988; Wessels et al., 1988). This lack of directed movement was also observed in this study in both aggregation streams and mounds. Another striking feature of myosin null cells in streams was their flattened appearance. Normal migrating amoebae have a threedimensional, elongated appearance and myosin II is con-centrated in the posterior cortex (Yumura et al., 1984). In wild-type aggregates cell movements are coordinated, with streams of cells invariably forming spirals as they join the aggregation centre. Although spiralling in D. discoideum aggregation is well documented (Durston, 1973; Clark and Steck, 1979; Robertson and Grutsch, 1981) its importance in development has not been thoroughly investigated. It has been proposed that orbital motions drive morphogenetic movements and slug migration and give the multicellular stages their characteristic circular axis (Clark and Steck, 1979). Spiralling was rarely seen in myosin null cell aggregation and when it did occur did not last long and quickly gave way to locally chaotic cell movements. In all mutant mounds observed by time lapse filming, cell migrations appeared to cease after 17 to 19 h development.

The lack of spiral motion in myosin null cell aggregation may be due to the mutant cells being unable to respond appropriately to morphogenetic signals once within the multicellular array. In aggregation streams, cells move as a monolayer or as a layer only a few cells thick; waves of cAMP are unidirectional and the cells move in one direction towards the aggregation centre (Durston, 1973; Robertson and Grutsch, 1981). Aggregation-competent amoebae in which the myosin heavy chain gene has been truncated are essentially normal in their response to cAMP (Peters et al., 1988). In the three-dimensional structure, however, most cells are surrounded on all sides by other cells and signalling must become more complex. The polarised shape and myosin II distribution of wild-type cells may be essential for cells to receive and respond appropriately to extracellular signals within a three-dimensional mass.

Responding to signals within three dimensions may also require cells to be polarised with respect to their membrane signal receptors. Unlike wild-type cells, myosin null cells are unable to cap cell surface proteins cross-linked by concanavalin A (Pasternak et al., 1989; Fukui et al., 1990). Analysis of particle transport in the plasma membrane of myosin null mutant amoebae indicates that myosin II causes mobile particles to be drawn into focal points (Jay and Elson, 1992). These results imply that the actomyosin meshwork in wild-type cells is linked to proteins in the plasma membrane and thus could be involved in regulating their distribution, as has been suggested for a number of mammalian cells (Gingell and Owens, 1992, for a review). During the formation of the tight aggregate, almost continuous secretion of cAMP causes a down-regulation of the cAMP receptor and a concomitant redistribution of the receptor into patches in the plasma membrane (Wang et al., 1988). It is not clear, however, to what extent this redistribution of cAMP receptors is important in D. discoideum development. It will be interesting to see whether myosin null cells in mature mounds show a different distribution of cAMP receptors to that observed in cells in wild-type aggregates.

Myosin null cell aggregates fail to form tips. In most wild-type strains, a tip forms after all cells have collected into the mound and acts as an organising centre that emits cAMP and directs further morphogenetic movements (Rubin and Robertson, 1975); tip formation is thus an essential part of D. discoideum development. How tips are formed is not well understood. It has been proposed that a subgroup of prestalk cells (pstA cells), initially distributed throughout the aggregate, sort out, migrating upwards to form a tip at the top of the mound (Williams et al., 1989; Esch and Firtel, 1991). In our immunofluorescence experiments no cohesive group of prestalk (i.e. unlabelled) cells was seen at or near the top of mutant mounds (apart from a single cell layer that covered the entire upper surface of the mound after 30 h) where a tip would be expected to form. If tip formation does involve sorting out, then our results suggest that myosin null cells destined to form the tip are unable to sort out within the aggregate.

It is interesting that, despite the apparent chaos of the myosin null cell aggregates, a distinctive pattern was observed within mature mutant mounds; prespore and spore cells localised to the centre of the aggregate while a layer of unlabelled cells (presumably prestalk cells) was observed along the base and a thinner layer along the top of mutant mounds. Since this pattern was established well after cell migrations within the mounds appeared to cease, it is likely that the observed pattern was generated by positional cues rather than through sorting out of differentiated cells. Evidence for positional information in the establishment of the prepore/prestalk pattern has been presented previously by Krefft et al. (1984) and Williams et al. (1989).

The inability of myosin II null cells to form tips was also observed in synergy experiments in which null cells stained with a fluorescent membrane dye were mixed with unstained wild-type cells. Null cells were absent from the extreme tips of mixed slugs and early culminates, suggesting that wild-type cells might be able to rescue null cell development by providing a tip for the cellular mass. Further experiments are required, however, to establish whether null cells are absent from the tips of mixed slugs because they lack an essential feature for tip formation (such as cell shape or proper signalling etc.) or whether they simply join the aggregate too late to form the tip (myosin-deficient cells move at less than half the speed of the parent strain; Wessels et al., 1988).

This research was supported by a Macquarie University Research grant and an ARC program grant to Keith Williams, and a National Institute of Health grant GM40509 to Jim Spudich. We thank Nick Rasmussen for help with the replica technique and Fran Thomas for help with the scanning electron microscopy. We also thank James Spudich, Malcolm Jones, Manuela Fuchs and Ines Carrin for helpful discussions, and Ron Oldfield for time and patience with the light microscopy.