A cinematographical study of the development of vitally stained Dictyostelium discoideum

Much evidence indicates that general principles underly multicellular morphogenesis (Von Baer, 1828; Wolpert, 1971). The evidence suggests that related organisms (possibly all organisms), and different organs within the same embryo, may be constructed according to a relatively small number of common rules for cell behaviour (Wolpert, 1971; Postlethwait & Schneiderman, 1969). Despite recent interest in this aspect of developmental biology, no such rules have yet been identified.

The purpose of this article is to report an investigation of the patterns of cell migration and contact formation underlying sequential events during morphogenesis of the cellular slime mould Dictyostelium discoideum (Dd.). The investigation was motivated by our wish to seek evidence for or against the idea that common rules for cell behaviour underly different morphogenetic events during Dd. development. In particular, we were interested to determine whether 4 cell competences which are known to control cell movement and contact during the first stage of Dd. development (aggregation) also regulate morphogenesis later on. Relevant details of the aggregation control competences and of the Dd. morphogenetic sequence are in Fig. 1, p. 264.

The main method for the investigation was to make and analyse time-lapse films of Dd. development. This method has previously been used successfully to analyse Dd. aggregation as described below. It has not previously been successful for analysis of later development because cell movements are not readily seen in later structures (which are multicellular).

We have used a vital staining method to overcome this difficulty, as described in the first section of Results.

Dd. cells were grown on phosphate-buffered nutrient agar, in association with bacteria at 20 °C (Bonner, 1967). The Dd. strains used were NC4 and its mutant KY3. These were kindly supplied by Professors T. Konijn and J. Ashworth respectively. KY3 makes a very clear neutral-red staining pattern (see below), and was used, in addition to NC4, for observations on slugs and regulating slug pieces. No differences between its behaviour and that of NC-4 were observed except those noted below.

Cells were stained with neutral red in 2 ways: (1) Harvest vegetative cells; wash free of bacteria by differential centrifugation through several washes of phosphate buffer; mix 3:1 with 0·06 % neutral red on the surface of a non-nutrient agar plate (1 % distilled water agar). Or (2) Spread 0·06 % neutral red with an inoculum of spores and bacteria on the surface of a nutrient agar plate.

In either case, remove desired neutral-red-stained structures when they are ready and film them.

The movements of neutral-red-stained granules were recorded on Kodak 16-mm 4 × reversal film, using either a Paillard camera with a Pathex timer or a Bolex camera with a Nikon CFMA autotimer, and Zeiss microscope optics. Structures were filmed on 1·5 % phosphate-buffered agar or on 1 % distilled water agar and illuminated in one of two ways: (1) A suitable mixture of incident and back lighting, supplied by 2 Nikon lamps. This revealed movement of stained granules in the surface of stained structures. Or (2) The structure was squashed, by laying a coverslip on it, so that the tip of a (migrating or culminating) slug or any part of a regulating piece protruded (as in Fig. 5D). This procedure did not necessarily severely perturb development (e.g. slugs continued migration or culmination). It facilitated tracking stained granules, which now remained in the plane of focus in a very thin layer of tissue. It also enabled observation of the movements of unstained cells, as described below.

Cell tracks were recorded manually in cases where they were to be displayed directly. Where further analysis was required, cell coordinates were recorded onto magnetic cassette tape, using a digitizer (Wang X-Y digitizer), connected to a Wang 2200 minicomputer. The tapes were processed further using a BASIC programme run in a Wang 2200 minicomputer.

Time-lapse films have previously revealed much about the control of cell movement during early Dd. aggregation but nothing about cell movement in later Dd. stages (Bonner, 1944; Gerisch, 1968; Durston, 1974: see also Fig. 1). Aggregation is easy to analyse because it occurs via visible movements by separate cells. Later stages are difficult to analyse because they consist of translucent tissues, within which individual cells are not distinguishable.

In the next section (below), we report observations from films made of later Dd. structures stained with the vital dye neutral red (Bonner, 1952). Neutral red stains large intracellular vacuoles (⪕ 5 μm diameter) in a subpopulation of later Dd. Cells (Francis & O’Day, 1969) (Fig. 2A). These vacuoles are visible in intact multicellular structures (Fig. 2B) and individual vacuoles can be tracked in time-lapse films over real time periods of up to about 1 h. Nearly all of our conclusions (below) are based on such records of motion of individual stained vacuoles. They are not affected by the criticism (of cell marking using neutral red), that this dye diffuses from cell to cell (see Loomis, 1975). We have, however, made an observation which demonstrates that a class of later Dd. cells retain neutral-red staining over a long period without much loss to neighbouring cells (see below, p. 269). We have used this finding to enable us to examine the role of cell movement in rearguard zone formation.

The neutral-red-staining pattern is stage dependent. Aggregates and young slugs usually show even staining. Most cells in them contain small stained vacuoles. Mature migrating slugs (> 24 h migration) and developing fruiting bodies typically show an axial pattern of staining. The anterior tip stains well, and so does the posterior part of the slug, but the middle of the slug and posterior part of the young sorus stain little (Fig. 2B). In these structures, most cells in stained regions and a minority of cells in ‘unstained’ regions contain relatively large stained vacuoles. Other cells contain none. Stained vacuoles are particularly large (and thus trackable) in slugs that have migrated > 24 h, and in developing fruiting bodies or regulating pieces derived from these. We have used these routinely for our films. Our films of aggregates were made using aggregates which have delayed development. These, also, contain large stained vacuoles. Delayed aggregates can be found on neutral-red-stained growth plates at a stage when most cells are already in slugs or fruiting bodies.

We have used an electron-microscopical approach to identify the neutral-red-stained organelle (Bluemink, Durston, van Maurik & Vork, in preparation). This showed that the stained organelle is an autophagic vacuole, and that it is virtually prestalk and rearguard cell specific in mature slugs (48 h migration) (but not in young slugs (no migration)). A minority of cells containing autophagic vacuoles (corresponding to the minority of stained cells) occur in the prespore zones of mature slugs, but these contain no prespore vacuoles (i.e. are prestalk or rearguard cells). We conclude that, in the mature slugs and other later structures examined by us, most stained cells that we track are prestalk or rearguard cells.

We filmed neutral-red-stained Dd. structures through their life cycle. The films, and other approaches, showed the following main features.

Previous films of Dd. aggregation, made using transmitted light and unstained cells, show details of cell movement till the stage when cells have entered compact aggregates. After this, individual cells become indistinguishable. We stained compact aggregates with neutral red and filmed them by method (1) above (using incident light). The following statements are based on observation of several hundred aggregates and films of 13.

Marked cells in compact aggregates moved actively (prior to tip formation). Cell velocities were of order 5–20 μm/min, similarly to those of cells entering early aggregates. Cell movements were clearly synchronized in travelling waves in 7 of the 13 aggregates filmed. Wave velocities (∼ 50 μm/min)andfrequencies(0·15–0·4 min⁻¹) resembled those in early aggregation.

We observed some surprising shapes among pre-tip aggregates. Some were hemispherical mounds. Others were sausage-shaped, the sausage commonly being folded back on itself to make an irregularly shaped mound. Others were torus shaped (i.e. a sausage joined in a ring). Films showed that upper cells in hemispherical aggregates often showed rotational movement, that upper cells in sausage-shaped aggregates showed movement across the long axis of the sausage (Fig. 3A) and that upper cells in doughnut aggregates showed approximately radial movement either outward (centripetally) or inward (centrifugally) across the top surface of the torus axis (Fig. 3B, c). In cases showing clear waves of movements, these propagated in the opposite direction to the cell movements themselves (as usual). One explanation which might account for the cell-flow pattern seen in these aggregates -at least, in those showing a clear wave -is that they are organized by various forms of a 3-dimensional spiral wave. This hypothesis is discussed in detail elsewhere (Durston, Vork & Weinberger, in press).

Two films showed details of tip formation. This process appeared identical during aggregation and during regulation of slug pieces. We describe it later, in connexion with regulation.

The following statements are based on films of 42 migrating NC4 slugs and 14 migrating KY3 slugs as well as observations of dissociated NC4 slug cells and histological sections of NC4 slugs.

Twenty-eight of the 56 slugs showed clear waves of cell movement, resembling aggregation waves. These had similar velocities (∼ 50 μm/min) and frequencies (∼ 0·3 min⁻¹) as aggregation waves. Waves were seen in 7/14 intact KY3 slugs filmed by method (1), in 5/26 intact NC4 slugs filmed by method (1), and in 13/16 squashed NC4 slugs filmed by method (2). Waves also occurred in regulating slug pieces and culminating slugs. Examples of synchronized periodic cell movements due to waves are recorded in Fig. 4.

All waves seen in slugs propagated from front to back along the long axis and organized a contradirected (forward) cell movement flow. Their place of occurrence was variable. In 4 intact KY3 slugs, they started in the tip and propagated as far back as could be seen (to the back end of the slug in 2 cases; well into the prespore zone in 2 cases). In 4 intact KY3 slugs and 5 intact NC4 slugs, waves started in the tip, but obvious synchrony was lost within the prespore zone. The squashed NC4 slugs showed the prespore and rearguard zones only. All waves in these were confined to the back part of the squash (the rearguard zone and sometimes the posterior prespore zone). The squash preparations showed that unstained cells as well as stained cells made waves of synchronized movements. Mobile cells in regions not showing clear waves moved intermittently but without obvious synchrony.

In 53/56 migrating slugs, cells flowed more or less directly forward along the slug’s long axis (Fig. 3D); 3/56 migrating slugs (1 NC4, 2 KY3) showed a helical forward flow (Fig. 3E), and 2 slugs rounded up before continuing migration. Cell movement flows in these were exactly as observed during early culmination.

During late aggregation, Dd. cells become connected in files via A-sites. The following observations indicated that files persist in the slug and other later stages.

NC4 slug cells were dissociated to small clumps and dispensed on an agar surface. The clumps began motion within about 60 min (under standard conditions: see Materials and methods). It was then apparent that each consisted of one or more short files of cells (Fig. 5 B). Details of the behaviour of dissociated slug cells will be reported elsewhere (Weinberger & Durston, in preparation).

Rearguard cells may trail at the back of the slug. Individual cells were then visible and were often obviously connected in files (Fig. 5 c).

Squashes of later structures, prepared as described in the Materials and methods may show long files of cells (Fig. 5o). These were seen in 13/16 films of squashed migrating NC4 slugs, 2/4 films of squashed culminating NC4 slugs and 5/6 films of squashed NC4 slug pieces (Fig. 5D). In migrating (and culminating) slugs, all files were parallel and oriented along the slug’s long axis. They were obvious in the posterior prespore and rearguard zones (which commonly consisted almost entirely of parallel files). They were less obvious in the anterior part of the slug. Cells there appear to be strongly connected via lateral as well as end-to-end contacts. The waves of movements described above propagate along files and were most obvious in parts of squash preparations showing files.

Longitudinal sections of slugs may show evidence of files running along the slug’s long axis (Fig. 5E,F).

The following observations indicate that migrating slugs contain subpopulations of cells that differ in their behaviour. (1). In 8/16 squashed NC4 slugs, stained cells in the prespore zone (i.e. the zone containing mainly unstained cells), were concentrated in bifurcating streams (Fig. 6A, B). The streams commonly moved faster than neighbouring (mainly unstained) tissue (Fig. 3 F, G). This difference in rate of motion between the tissue types was most apparent in the anterior prespore zone. It also occurs in regulating slug pieces and in culminating slugs and appears important in morphogenesis and pattern formation. We suspect that it occurs due to paralysis of prespore cells. (2). We made an observation which indicates that certain prestalk cells retain neutral-red staining over a long period (> 24 h), without much loss to neighbouring cells and that these cells consistently move backward and accumulate in the rearguard zone. We grafted small stained prestalk pieces (∼ 1/20 of a slug’s volume) onto the front of large unstained slugs and observed the fate of the neutral-red stain. In all cases (20/20), these grafts gave rise to largely unstained slugs, containing a few, scattered, well stained cells. These stained cells persisted, even when the chimeric slug migrated for several days. (Films also show that some dissociated slug cells retain stained vacuoles for more than 10 h.) In all cases, the stained cells accumulated in the very back end of the slug (Fig. 6C-E). This finding suggests that the (prestalk-like) rearguard zone may originate via cell migration backward from the prestalk zone.

As described in Fig. 1, the embryonic pattern in the slug is regulative. Excised back pieces of slugs, containing only the prespore and rearguard zones, make a new prestalk zone and tip within about 4 h (Bonner, Chiquione & Kolderie, 1955). Regulation is reflected by appropriate changes in the neutral-red-staining pattern (Bonner, 1952). The following statements are based on 50 film sequences of regulation in excised back pieces from slugs (36 NC4 and 14 KY3).

Step a. In 12/17 applicable NC4 sequences and 5/11 applicable KY3 sequences, showing early regulation, the cells in the excised piece were initially motionless, for a period lasting up to 300 min.

Step b. In 15/30 applicable NC4 sequences, and 10/14 applicable KY3 sequences, the piece then showed active cell movement before tip formation. The excised pieces typically rotated. Stained cells in them rotated along more or less complex trajectories. Five of 6 sequences of NC4 pieces squashed during this step showed stained cells concentrated in streams, resembling streams in intact slugs (see Fig. 6A, B). Cells within streams moved faster than neighbouring tissue. Cell movements were never initially synchronized in waves.

Step c. In 8/26 applicable NC4 sequences and 5/8 applicable KY3 sequences, cell movements became synchronized in a (circulating) wave before tip formation. The wave caused a circulating deformation of the surface of the excised piece and periodic movement of it. Presumably it regulated movement of unstained as well as of stained cells. Five of 6 squashes of regulating NC4 slug pieces also eventually showed a circulating wave of movement by stained and unstained cells.

Step d. Between 90 and 400 min after transection, all excised pieces made a tip or tips in their apical part. This step was entered from one of the 3 steps described above. Tip formation was characterized by directional movement and aggregation of stained cells into a focus or foci within the apical part of the regulating piece. Each focus became a tip (Fig. 7, seen in ii/n NC4 and 8/8 KY3 sequences showing tip formation). In 6/11 applicable NC4 sequences and 6/8 applicable KY3 sequences, the stained cells aggregated via obvious waves of synchronized movements. In 1/10 applicable NC4 sequences and 1/8 applicable KY3 sequences, they also entered bifurcating streams of stained cells (Fig. 50).

During tip formation, a variable proportion of the stained cells near to the presumptive tip moved into it. Others moved much less and were sometimes virtually immobile (Fig. 3 G). Tip formation thus commonly involved a decrease in rate of motion for many apical-stained cells in pieces previously in steps b or c. In 2 (NC4) sequences, it was clear that this effect spread progressively from the apex to the base of the piece. In all, the effect increased during the time course of tip formation. In 1/1 NC4 piece, squashed just after tip formation, all unstained (and stained) cells outside the tip were motionless, while the tip itself rotated. These observations indicate the onset of a well defined pattern of cell immobilization during regulation (see Durston, 1976.) Films and observation show that the morphological tip generally corresponds exactly with the well stained region in regulating pieces and other later structures.

Two (KY3) sequences showed tip formation during aggregation. The tip formation in these showed the same features described above.

The following statements are based on 47 films of NC4 culmination.

In 19/20 film sequences showing the beginning of culmination, there was an obvious difference in rate of movement between cells in the upper and lower parts of the culminating slug. Upper cells, in the region immediately behind the tip became very slow moving, or sometimes motionless (Fig. 3H, 1). Lower tissue continued to flow forward and distended the slime sheath beneath, and, eventually, in front of the tip. The slug thus rounded up and the tip moved to the top of it or, in some cases, temporarily disappeared as a morphological entity. In 10/20 sequences, the moving tissue showed antero-posterior waves of movements.

Twenty sequences showed rounding up. In 9 of these, tissue entering the cell mass rotated. In all, stained cells in the body of the cell mass were soon immobile. In 10/20 cases, immobility spread progressively from the upper part of the cell mass to its base.

Eleven sequences showed the apical part of the cell mass and the tip during and just after rounding up. In all cases, cells in the tip remained mobile. In 7, they rotated. In 10, stained cells near the base of the tip moved directionally into it (thus sorting out). In 8/10 sequences, this cell sorting occurred via obvious waves of movements. It is generally true that the neutral-red-staining pattern sharpens during early culmination. This is at least partly due to sorting out.

After rounding up, the fruiting body begins to make a vertical stalk within the cell mass, and elongates this by periodic movements, as described in Fig. 1. Six sequences showed this phase of culmination. In all 6 cases the stained cells in the prestalk zone made periodic waves of movements, which accompanied the length increments of the stalk. This is in line with our previous suggestion that periodic length increments of the stalk occur due to cell movement into the stalk apex (Durston et al. 1976). Stained cells in the main body of the sorocarp invariably remained non-mobile throughout later culmination, and one squash of a later fruiting body also showed that the bulk of the unstained cells are now non mobile. This may be the reason why prespore cells fail to enter the stalk. Migrating slugs sometimes round up temporarily. They then show a similar sequence of events as is observed during early culmination (two cases observed).

The above observations are evidence for a well defined pattern of cell immobilization within the prespore zone during culmination and rounding up in slugs.

Later Dd. structures (late aggregates, migrating slugs, regulating slug pieces and culminating slugs) show non-decremental waves of cell movements, resembling early Dd. aggregation waves. This suggests that the cell competences underlying aggregation waves (competence for chemotaxis to cAMP and for relaying a cAMP signal) may persist in later Dd. stages. We now have direct evidence that cells in an intact slug can chemotact to cAMP (see below). We note that certain regions in later structures as well as some entire later structures show cell movements which are not synchronized in waves (are asynchronous). It is an open question whether these movements are controlled by the same competences underlying waves or by others.

We observed long files of cells, joined end to end, in all later structures. This suggests that A-contact sites persist in later stages. Files are oriented along the long axis of the slug and contain a high proportion of the cells in it. We note that when files first form (during late aggregation), the cells which enter them have rather rigid polarity (Bonner, 1950). We also observe that short files, formed by partly dissociated slug cells, are rather stable and move with fixed polarity (Fig. 5B). These points raise the possibility that the known strong polarity of slug tissue (Bonner, 1950) is contributed to by persistent connexions between permanently polarized cells (thus having a ‘magnetic’ aspect). We cannot rule out that cell polarity and/or cell contact are regulated by an external factor, such as an axial gradient or polarized wave propagation.

Our observations of slug formation and of regulation in slug pieces show that prestalk-zone formation always involves sorting out of stained (prestalk) and unstained (prespore) cells. This finding supports and extends the earlier observation by Takeuchi (1969) that sorting out occurs at the end of aggregation. Our results also show that neutral-red-stained cells migrate from the prestalk zone to the rearguard zone. They suggest that cell migration is also important for rearguard zone formation. We suggest, from these observations, that early Dictyostelium discoideum pattern formation has 2 aspects. (1) Directed cell migration, leading to the spatial pattern (in the slug). And (2) Homeostasis, leading to correct proportionality of the regulative cell types (in the slug). According to our findings, the homeostatic mechanism could be non position-dependent (as also suggested by Garrod & Forman, 1977). It could, alternatively, work via a positional signal of some kind. These possibilities remain to be distinguished.

We note that cell sorting is implicated in maintenance of differently specified regions (compartments) in insects (Lawrence & Morata, 1975) and that cell migration is implicated in setting up the musculature of the vertebrate limb (Chevalier, Kieny & Manger, 1978). We speculate that directed cell migration and resultant sorting may have a universal role in pattern formation and maintenance.

Sorting out of prestalk and prespore cells occurs due to directional movements by prestalk cells through motionless or less-active tissue. We have now shown that this directional movement can be oriented by artificially imposed cAMP gradients, indicating that it occurs via chemotaxis, to cAMP (Matsukuma and Durston, in press. To our knowledge, this is the first case of cell sorting in which there is clear evidence as to the mechanism of the process. We note that previous investigators have not considered chemotaxis a likely component of sorting out (e.g. Steinberg, 1962).

Dd. development involves a well defined spatiotemporal pattern of tissue immobilization. This occurs in the anterior prespore zone, in a migrating slug. It is seen in the apex of a regulating prespore piece, and may spread from the apex downward. It is first seen in the upper anterior prespore zone of a culminating slug, then spreads to the whole anterior prespore zone as the slug rounds up, and eventually spreads downward, through the whole prespore zone as the young fruiting body is lifted from the substrate by the stalk. It appears to affect unstained (prespore) cells and a subpopulation of stained cells. In our view, one possible explanation for these observations would be that prespore cells and a subpopulation of stained cells, but not prestalk cells, are paralysed by a superthreshold concentration of an extracellular substance, which is secreted mainly (or only) by prestalk cells. The substance would be secreted by the slug tip and lost to the substrate. It would thus be present as a monotonically decreasing axial gradient in the migrating slug and as an apico-basal gradient in regenerates or rounded slugs. This hypothesis could account for many Dd. morphogenetic movements as explained in Fig. 8.

We have now shown that prespore cell immobility and cell sorting can be induced by an appropriate concentration of extracellular cAMP (cAMP is secreted by the slug) (Matsukuma and Durston, in press).

Our observations give some information as to the mode of action of a primitive organizer (the tip) and also as to the regulation of organizer formation, (a) We had previously observed that the Dd. aggregate tip is a pacemaker (i.e. initiates waves of cell movements: Durston, 1974). Our new observations reported here, show that the tip, or anterior part, of the migrating and culminating slug may retain this property and may control cell movement in the slug by virtue of it. (b) Our observations of prespore cell paralysis could mean that the tip also controls (inhibits) cell movement via an axial gradient of an extracellular substance, possibly cAMP. (c) Our observations of tip formation show that local aggregation of prestalk cells is an early event in this process, and also that the morphological tip corresponds with the prestalk zone. We suspect that the position of the developing tip and of the future front end of the Dd. axis is normally defined by a local maximum in prestalk cell concentration. This conclusion was also indicated by a previous result: that the slug contains a non-linear gradient in a parameter which activates tip formation, the form of which matches the axial gradient in prestalk cell concentration (Durston & Vork, 1977). We have now also shown that the position of the prestalk cell aggregate (and thus of the future front end of the axis) can be controlled by a local cAMP maximum (Matsukuma & Durston, in press). We suspect that an extracellular cAMP gradient, set up via cAMP secretion and loss to the substrate, also regulates the position of the developing tip during natural development and regulation.

We note that the tip cannot organize a new axis if it is grafted into slug tissue with reverse polarity (Raper, 1940). This may well be because slug tissue is organized in files of polar cells (see above).