Integration in Aggregating Cellular Slime Moulds

The post-vegetative amoebae of the cellular slime moulds Dictyostelium and Polysphondylium flow in streams towards collecting centres ; this aggregation is preparatory to building aerial fruiting bodies (Raper, 1940a, b, 1941). Specific chemotaxis is one of the factors involved (Bonner, 1947; Shaffer, 1953, 1956). In the earlier papers in this series (Shaffer, 1957a, b), a distinction has been drawn between integrated and unintegrated cells : the former can adhere strongly to one another and usually secrete the chemotactic agent acrasin ; the latter adhere but poorly and (effectively, at least) do not secrete it. In the present paper the transformations of cells from one state to the other are examined; these have been called integration and disintegration. The development of the first integrated cells, on which the initiation of aggregation depends, is considered in a separate section.

The organisms were cultured and operated on as previously described (Shaffer, 1957a). They were allowed to aggregate on saline agar at a sufficiently low cell-density for the cells to be well separated from one another before aggregation began. In certain specified instances they aggregated on glass under water (Bonner’s technique, 1947).

All the plates are photographs of living cultures.

Frequently an aggregation in P. violaceum began with the cells piling themselves up into a relatively large centre and then grew by the progressive outwards extension from it of continuous streams (Fig. 1) although cells sensitive to acrasin might be found far beyond its edge. This suggested that contact with cells that were already; integrated—those in centres and streams —was the usual stimulus that made unintegrated cells become integrated ones.

An alternative possibility was that the cells were stimulated to change while some distance away from the structure that attracted them, but that they either took as long to develop new properties as to crawl to it or were too far apart for any change that had taken place to be revealed on their way there.

Separate sensitive cells, either unaggregated or de-integrated, were swept with ball-tipped glass needles into two heaps of diameter about 100 to 150μ. One heap served as a control : its cells started to disperse immediately, and usually within 10-15 min it was replaced by a patch of separate unoriented cells. Thus, by itself, extensive contact between sensitive unintegrated cells was insufficient to induce their integration. A young fruiting body, which emitted acrasin, was deposited near the second heap and then carefully dragged away as cells from the heap advanced towards it. (If its slime sheath was not disrupted, few of its own cells were left behind, and these were easily brushed aside.) By this operation the cell heap was induced to transform into a compact stream without its making contact with the fruiting body; and the acrasin secretion of this stream was attested by the reaction of the heap-cells that had been out of range of the original source and had begun to disperse. If the population density was such that the separate sensitive cells were initially close together, a similar result could be obtained without the necessity of sweeping them into heaps ; in this case, the fruiting body was dragged over an area first cleared of cells.

In an experiment designed to find out whether amoebae could be attracted by a centre without being connected to it by a continuous interface, Bonner (1947) mounted two coverslips as shelves side by side under water; one was covered with amoebae that were just starting to aggregate, and on the other, very near the adjacent edge, there was a centre. This did attract the amoebae on the first slip; and as they moved towards it, they joined up into streams, even if they were unable to cross the gap and make contact with it.

This experiment has given a similar result when performed with acrasinsensitive test cells that were not already aggregating, and it has thus provided excellent evidence that a source can induce integration at a distance. More recently (Shaffer, 1956) integration has been induced by acrasin solutions in the absence of a living source.

Acrasin’s induction of acrasin secretion complicated the bio-assay of its emission (Bonner, 1949; Shaffer, 1957a); for though the orientation of cells attracted towards a source did initially record the direction of the maximum gradient of the chemical it produced, subsequently, unless they had already reached it, they were liable to deflect one another when they themselves began to secrete. When the differential in emission along a single source was estimated by recording the extent to which the orientation of acrasin-sensitive indicator cells diverged from the normal to it, the error due to induced secretion could be minimized firstly by placing these cells close to the source and secondly by depositing them not in a continuous streak but in a row of small heaps that were as far apart as was practicable. When the acrasin emission of two sources was compared by observing the partition of sensitive cells between them, induced secretion might lead to the displacement of the zero-gradient zone towards one or other of the sources unless these were very near together ; to reduce the error in this situation the cells were not distributed at all distances between the sources but were grouped in a narrow band distant from one of them by some arbitrary fraction of the gap separating them.

When the preaggregation cells of D. discoideum were not too close together, primary aggregation was marked by the nearly uniform orientation of many of the cells over a wide area before they had made contact with one another. Sometimes small centres were clearly visible before appreciable orientation had occurred (fig. 2 A) ; sometimes they were not. The extent of the field oriented within one system varied with the distribution of the centres and the responding cells and the time intervals over which centres appeared : it might be centimetres across. An aggregation looked as if it had been stippled with short oriented strokes instead of drawn with solid lines; and because of the distances involved, it was clear that orientation could not have been brought about by the secretion of the centre alone, even if the cells, which in reality were in various stages of developing acrasin sensitivity, had all been as sensitive as those taken from the continuous streams that appeared later (the normal indicator cells used in tests). That acrasin was secreted by the periphery was made visibly manifest if the feathery expanses of cells led to the centre not directly but in sweeping curves (fig. 2 B), which reflected the approximate orientation of most of the separate cells within them. Two adjacent expanses within the same aggregation might be almost mirror images of each other, and thus cells on either side of the boundary between them might point in opposite directions. In the extreme case, cells could be guided to a centre through a pathway, formed by other cells, that was a nearly complete circle and first led them straight away from it (fig. 2, c). Evidently the oriented cells, though still separate, were already integrated.

In a stippled aggregation, the cells joined up in twos, threes, and halfdozens to form fine, irregular, and disconnected threads (fig. 2, A), many of which, judged over a small enough area, were roughly parallel, though deviation could readily be found. The bands of cells visible at an early stage condensed into compact streams in much the same positions; and the larger systems fragmented into a number of smaller and more definite ones.

In one part of an aggregation the streams might be broad and almost confluent ribbons, in another only tenuous strings; and as was particularly obvious when the population was comparatively sparse, one area might have the general appearance that another had had 1-2 hours earlier. Such differences in cell density or in progress of association, which were due to local variations in the environment, including the relative amount of bacterial food available earlier, might occur in the same sector: massive streams might feed thin ones, and (figs. 3, A—c) considerable lengths of uninterrupted stream, broad or narrow, be separated from the centre by an area of short discontinuous threads or of cells that showed very little trace of association (but had nonetheless relayed the acrasin stimulus).

The acrasin secretion of the” cells in a stippled aggregation was tested by one of Bonner’s (1949) methods. Quick-moving, highly sensitive cells taken from a more advanced aggregation were deposited in a bare, or bared, sector, close to a radius of oriented cells: they crawled perpendicularly towards it. It was concluded that, as with compact streams, there might be no steady spatial differential in acrasin emission. Yet the separate cells that composed the radius did not turn sharply towards the secreting cells on their flank, but, in the main, preserved their existing orientation. On the other hand, if the cells in a stippled aggregation were swept away early enough from an area that extended radially about 250 μ and was several times as wide (cf. the cutting of a continuous stream), cells in the outer part of the mutilated sector might then turn towards the intact sectors on either side of them, though of course these sources had been there all the time, and form concentric arcs at right angles to their previous alignment. In a variation of this experiment, the cells swept up were not removed but were used to build a radial bridge across the bared swath. The cells farther out might then converge on the outer end of this bridge, as if it were the centre, and in due course cross over it. (If the bridge was built of cells from older aggregations, it was ineffective in conducting the aggregative stimulus, unless replacements were added, because the cells moved so rapidly towards the centre that they soon ceased to span the gap.)

Thus it appeared that the orientation of these organizations was a record of the spread of the aggregative stimulus and also tended to be self-preservative.

Although a centre cannot attract cells that are more than a fraction of a millimetre away if there are no cells in the intervening space, an aggregation may extend for centimetres: the centre’s influence is ‘relayed’ by the peripheral cells, which when exposed to relatively little acrasin begin to secrete it.

Using a method of estimation devised by Bonner (1949), it has been found that a centre and a stream often secrete acrasin at concentrations that are much the same, judged in terms of the effect on the amoebae (Shaffer, 1957a). By another of Bonner’s methods it has been shown that there may be no continuous centrifugal decrement in acrasin emission either along a compact stream or within a stippled aggregation; and indeed it would have been remarkable, in an aggregation of this type, if the organism had been able to establish over the whole distance covered by the relay system a nicely graded differential that was not disturbed by considerable local fluctuations in the density and physiological condition of the cells. Now, if all the separate sensitive cells in an area were simultaneously induced to begin secreting acrasin at about the same concentration, while they were still unoriented and randomly distributed, each would attract all its neighbours, minor irregularities of distribution would be magnified, and a large number of small clumps would be formed, which would then join up into bigger and bigger aggregates. However, secretion does not in fact start simultaneously throughout an area ; and it is to the time and space relationships of the spread of the acrasin relay outwards from the centre that the orderliness and basically radial pattern of aggregation are due.

These relationships depend on a rather complicated set of variables, of which most at the moment are a matter of conjecture. Thus it is possible that if a sensitive unintegrated cell is exposed to an acrasin stimulus adequate to induce its integration, the time lag before it begins to secrete, the course of its secretion, and/or the maximum concentration it produces at its external surface will depend in part on the strength of the stimulus. The precise pattern of stimulation it experiences as it approaches a single source secreting at a constant rate will be determined by the speed at which it orients and advances, which in turn may be affected by the stimulus strength. All the responses of the cell will vary also with its physiological condition, which may change both spontaneously and as a result of exposure to acrasin—that is, it may adapt to the stimulus. And in actual aggregation, instead of a single static source, there may be large numbers of them, distributed irregularly at distances dependent on the average cell density, moving at different and varying speeds, secreting at different and varying rates, and mutually interacting. There is also the question of when, where, and in what concentration the enzyme inactivating acrasin is secreted, of how its secretion is related to the acrasin concentration, and of its subsequent fate. Additionally, if the molecules produced by its extracellular activity are absorbed by the cells, built up into what they were before, and resecreted, or if their structure is sufficiently like what it was before for them to compete for the same component of the cell surface, their distribution may be important too. As so little is known of these matters at present, no more than a preliminary analysis will be attempted here of the relationships responsible for the patterning of aggregation in a field of cells that are initially well separated.

The absence of visible local association of the cells in any given field may be taken as evidence that they are either not secreting enough acrasin to affect one another’s movements or have only just begun to do so. For if they had been secreting for some time, they could have remained separate only if they all had been moving at the same speed in the same direction, whereas, in fact, cells advance at different speeds ; and, in addition, their being initially scattered at random, and their being forced ever closer together if they crawl radially towards a centre, would have enhanced the effect of local attraction. Thus when nothing is to be seen of an aggregation but a single clump, it is clear that there is no significant secretion by the cells outside it (with the possible exception of those moving towards it that are in its immediate vicinity) and no effective propagation by the relay mechanism.

When propagation does occur, streams may develop that from their inception are continuous and grow out rather slowly from the centre. Again, of the separate cells, at most those near the tip of a stream can have started to secrete, and if they have, they are unlikely to do so maximally till they have moved still nearer to it. Once in the stream, they can be guided by the flow of their neighbours, and so it suffices if an adequate gradient is maintained for a relatively short distance outwards from the tip.

Continuous streams may appear even when the cell density is low, and they may in some areas be parallel rather than radial. Their formation is helped by the elongation of the cells in the direction of the gradient and the consequent reduction of the gaps between them, but essentially it depends on the lateral condensation of comparatively wide bands of cells. Differences in speed may be involved too : individual cells pursuing the outer end of a stream may rapidly catch up with it. (Similarly, a length broken off a stream spontaneously may follow and later rejoin the central stump.)

When the ratio of the speed of propagation by the acrasin relay to that of centripetal movement of the separate responding cells is much higher, the aggregations are initially stippled, their streams discontinuous. The orderliness of orientation may be explained as follows. As an integrated area invades an unintegrated one, cells just ahead of its advancing front are attracted towards it. Even if they soon release acrasin at about the same concentration as the cells ahead of them, and those behind them do so too, there will still be a gradient in the same direction as before during the phase in which their secretion is increasing, because of differences in the time of onset of secretion. Further, if by the time all its neighbours are secreting maximally a cell has turned and extended towards the sources that first attracted it, these will tend to exert a somewhat greater influence on it than the newer ones, simply because a cell is guided by the external gradient at its front end. Whether this effect is significant or not will depend on the distances separating the cells (as compared with their lengths).

Sometimes the cells in a stippled aggregation reorient towards another centre, which shows that it is possible for the acrasin relay to be re-excited. This suggests that the original centre may be able to re-excite the relay too, and if it can, and does so periodically in those cases where it retains control over the field around it, it is easier to explain the continued movement of the cells towards it. Instead of there being present constantly or recurrently a differential in secretion finely graded from the centre out to the edge of the aggregation, and controlled in a way not easily imagined, there will be an adequate gradient in comparatively narrow zones that travel out from the centre, and the gradient will result simply from time differences in the activation of the sources. Once the cells in a stippled aggregation have joined up into more or less continuous streams, further pulses will be less necessary, though as the cells may be reoriented even at this stage, presumably they may occur.

What is taken as visible evidence for the repeated excitation of the relay system is the rhythmic movement that may be revealed by time-lapse films, especially those of early aggregation (Arndt, 1937; Bonner, 1944); pulses of rapid centripetal movement are propagated centrifugally; and as they are not halted by gaps between the cells, they can hardly be of mechanical origin.

As yet, almost nothing is known quantitatively about cyclical changes in the properties of the cells. The maximum secretion reached in any given excitation of the relay might be maintained and serve as the base level in the succeeding one; but if, as is more probable, there is an interim decline, the gradient will be reversed after each pulse. Some reasons have already been put forward (Shaffer, 1957a) as to why this need not vitiate the orientation mechanism within a stream. While the cells are still separate, there is an additional factor: changes in velocity can alter the distances between them; so that if the zone in which secretion levels off or actually declines coincides with that in which velocity decreases again, the waning of the influence of a given cell on those behind it may to some extent be offset by their closer approach to it.

After stippled aggregations of D. discoideum had been cooled to about 20 C. for 1-2 days, their cells had slightly disassociated, though it was possible to recognize the positions that had been occupied by the more compact sections of the developing streams. If the cultures were then returned to room temperature, these sections tended to re-form, though they did so somewhat irregularly and not many of them entered the original centre. The renewed association of the rest of the cells was still more irregular: rings and hemispherical clumps, some of them fed at first by short or rather tortuous streams, appeared in large numbers, perhaps several hundred within the area previously covered by a single aggregation (fig. 3, D). Those that were big enough eventually produced miniature fruiting bodies: but most were too small to do this, and except for those that disintegrated or turned into linear streams (Shaffer, 19576) and then joined up into bigger organizations, they persisted long after cells in a neighbouring part of the culture had had time to feed on bacteria, aggregate, and build fruiting bodies. (However, if the cells were released from their sterile configurations early enough by being swept together into larger heaps, their development continued.) On return to room temperature, the cells were largely separate, individually polarized, mutually adhesive, and motile in varying degree; their behaviour was consistent with their secreting acrasin without any regular differential. I believe that aggregation ceased to be radial because the temperature changes upset the temporal sequence of secretion, though just how they acted can only be surmised. By exposing sensitive unintegrated cells uniformly to a solution of acrasin, it should prove possible to disturb the relay system in a way that is more readily explicable.

The relationships of the relay were not constant within a species. But at the densities used, in an aggregation of P. violaceum, streams usually either were lacking or slowly grew out from the centre as virtually continuous structures; whereas in D. discoideum they commonly were discontinuous in their early stages and covered much of their ultimate extent almost as soon as they became recognizable. All three patterns might be seen in D. purpureum.

If unaggregated cells or those from disintegrated streams were swept together with a glass rod into a number of small heaps, these immediately began to dissolve, and within the first minute or two the cells had separated appreciably. (In a small proportion of the heaps a tiny knot of cells was left behind; possibly they had been damaged or had formed an inchoate centre.) On the other hand, if heaps were built of cells obtained from intact streams, only some of them dispersed (Shaffer, 1957a, b). The point to be noted is that even these did not begin to do so for 5-10 min or more, and then their cells separated as fast as those from unintegrated heaps. This time-lag suggested that disintegration, including the loss of adhesiveness, was an active process ; though possibly the disturbance of the cell surface produced by mechanically separating the strongly bonded cells was responsible for it.

As to the causation of disintegration, some of its complexity was revealed by the following observations. A stream sometimes disintegrated if the centre into which it was flowing was removed, but it did not do so invariably. Likewise, if a stream was cut, integration might be maintained, or disintegration might spread upstream from the gap, although this might be no wider than that over which the cells could originally have received the stimulus to become integrated. The spread of disintegration could reasonably be ascribed to a chain reaction in which, as each cell was affected, the one behind it was left exposed and was thereby stimulated to return to the unintegrated state. This reaction might travel along a stream at about an amoeba-length a minute or much more slowly; and just as the initial cutting of the stream might fail to start disintegration, so at some point the cells might prove resistant to the transmitted stimulus, in which case disintegration was halted. Thus, loss of contact with integrated cells was not a sufficient stimulus. But neither was it a necessary one. In the first place, a stream that had never made contact with the acrasin source that induced it sometimes persisted when it was removed, yet at other times it did not. Secondly, disintegration might start spontaneously in an intact stream, first affecting cells whose front ends were touching integrated cells. If disintegration occurred in other intact streams soon afterwards it was found in adjacent streams more frequently than would have been expected had its position been determined entirely by chance. This being so, it could be assumed that once it had begun in a neighbouring area, there was a reciprocal influence on its further progress in the original one.

These results showed that the integrated state was not equally stable in all stream cells. In some of them it was lost following any of a number of visible changes in the aggregation, all of which probably acted by sharply decreasing the acrasin concentration to which they, and especially their front ends, were exposed. In others, which can be said to have stood higher on the ‘social scale’, it was maintained despite such changes; whereas in those at the low end of the scale, its loss was not preceded by any visible changes, and the cells could have experienced at most a minimal external stimulus—one without effect on their neighbours—and possibly none at all. Additionally, at least a fraction and perhaps all of the stream cells must to some extent have adapted to the integrative stimulus. This adaptation was lost if the cells returned to the single state, for they could then reorient to an acrasin source and integration could be reinduced (with the proper procedure) at a distance from it. The concepts ‘variable stability’ and ‘adaptation’ as used here are purely descriptive: for them to acquire an explanatory value they will have to be interpreted in physico-chemical terms.

Disintegration could alter the pattern of cell association in a variety of ways, of which the simplest was the replacement of a minute streamless centre by a small patch where the cells were temporarily rather crowded together (cf. fig. 3, E). In an aggregation with tributary streams, disintegration might appear at one or more points in any or all of them and then spread along them centrifugally, destroying the organization. In some cases it stopped after a short distance; in others it crept through every branch it met right out to the tips of the finest tributaries. If it affected only the outer ends of the streams, the cells involved returned to the condition they were in a short time previously; and the aggregation as a whole still presented a ‘typical’ appearance— it was merely somewhat smaller. Usually, the separated cells later re-formed streams, continuous with the main system, in the same areas as before.

However, if disintegration spread out from the centre along all the streams, the pattern was radically altered (fig. 4, A). What was left was like a fungal fairy ring, which increases in diameter as the mycelium grows outwards in all directions while dying away in the middle. In a symmetrical aggregation of this type, the ‘ring’ was the area of persistent cellular organization ; it was bounded by two concentric circles between which the streams stretched more or less radially. The outer circle represented the site of incorporation of single cells ; the inner, that of their release. Wide variation in the relative rates of expansion of the two circles indicated that the two processes were independent. If the inner circle moved more quickly than the outer, the lengths of the streams decreased ; in the limiting case, the inner caught up the outer and all organization was annihilated. If the inner circle moved more slowly, the streams lengthened. If disintegration stopped, the streams then piled up as solid masses at their inner ends (fig. 4, A 2) or continued onward, guided by acrasin, and so, according to the spacing between them, either looped back on themselves (fig. 4, B) or fused more or less extensively with their neighbours (fig. 4, c). Whatever their precise course, they attracted the recently separated cells from the central area. Thus the original aggregation was replaced by a variable number of discrete organizations, each with streams growing outwards made of cells moving inwards, and others growing inwards of cells moving outwards; this, of course, had not necessitated a reversal of polarity of any bound integrated cells.

If the original centre had persisted (fig. 4, D) and was secreting acrasin, it became the nucleus of a new radial aggregation, which was constructed out of the cells separated from the first one. If this continued to disintegrate, the new one could grow extensively, but it in its turn might soon start to dissolve. It was, however, uncommon for one complete ‘fairy ring’ to form inside another, because actual aggregations were almost always asymmetrical: the streams were not of the same length; they grew into areas of different cell densities and sensitivities and so extended at different rates; and even if disintegration began at the centre, it might be arrested at various distances from it (fig. 4, E), SO that some streams were completely destroyed and others persisted independently. Yet in limited sectors, there might be half a dozen concentric bands alternately containing integrated and unintegrated cells.

When, in an approximately radially symmetrical disintegration, the opposed activities at the inner and outer ends of the streams proceeded at about equal rates, the streams remained the same length, though the cells of which they were composed were constantly changing. The streams moved outwards ; the cells inwards. If the cells within a stream had not moved, if they had merely attached themselves at one end of it and freed themselves at the other, they would have been no more crowded inside the ‘fairy ring’ than outside it. But because they did move they became concentrated inside. The effect was plainly visible (fig. 4, A, C) and persisted, if the ring remained entire, until most of them had been reintegrated in streams; but if some of the streams disintegrated completely, the accumulated cells could leak out through the gap. The cells inside a rapidly expanding ring had the same properties as those outside it: having little intercellular adhesiveness, they were separate and independent, and they were motile but undirected. A ring of streams thus transferred cells of a particular type from one concentration to a higher one ; it acted by imposing a uniform orientation on its polarized cellular components and disciplining their inherent motility. It was interesting that though this polar organization was called into existence by a centre, it did not need one for survival. It was not based on unchanging morphological elements: it was dynamic and self-perpetuating; it depended on a ‘tradition’ of polarity being handed on to the entering cells, which if it was once lost could not be re-established without the intervention of an acrasin source.

Disintegration as it occurred in D. discoideum was much the same as in P. violaceum, the major differences being in the residual adhesiveness of the de-integrated cells, as described already (Shaffer, 1957a), and in its timing: when compared with the speed at which the cells moved, the spread not only of integration but of disintegration too tended to be more rapid in D. discoideum than in P. violaceum ; so that instead of the cells in a stream detaching themselves in an obvious sequence, visible dissolution might start almost simultaneously throughout the whole of the length affected or at least a considerable part of it. Time-lapse photography showed also that the output of acrasin fell abruptly at the same time, as far as could be judged by watching the disorientation of single cells that were being attracted. After multiple sectioning of a single stream or the removal of the centre from an aggregation, a number of the fragments might disintegrate perhaps 5, 10, or 20 min later. Some or even all of them might begin to disperse at almost the same moment or at widely different ones, and a large fragment might break down in several discrete steps.

Much remains to be determined of the causation of disintegration and of the role taken by changes in the acrasin field.

It has been generally agreed that aggregation does not usually occur till almost all the ‘available’ food has been eaten. Potts (1902) cultivated the amoebae in the vegetative phase indefinitely by transferring them periodically to fresh supplies of bacteria. Raper (19406) carefully covered aggregations with a suspension of bacteria and found that only when the cells were disturbed did they return to the solitary state and feed on them. Arndt (1937) reported that some time after a plate of edible bacteria had been inoculated with spores of D. mucoroides at one point, concentric zones were occupied by: fruiting bodies, aggregations, the preplasmodium (a loose association of preaggregation cells), and feeding cells; this showed that the social phase began when the cells experienced a certain degree of starvation. However, in very sparse cultures aggregation was delayed several days, while the cells formed small groups by random collision; but if they were gathered together earlier, they at once developed as true aggregations. Cell contact was thus an additional factor. Working with D. discoideum, Raper (19406) found aggregation to occur sooner in thin cultures on media poor in nutrient than in luxuriant ones on rich media and concluded that contact was not a controlling factor ; it was possible, though, that it was one, but that it acted only after the cells were without food. Raper reported that aggregation began about 42-44 h after nutrient-poor agar had been uniformly inoculated with bacteria and discoideum spores. If the cultures had been grown in daylight instead of darkness, or if, 6 h before aggregation would otherwise have been expected to occur, they had been mildly dried or had had their temperature slightly raised, aggregation started a few hours sooner. Also, under these conditions, the aggregations were smaller and more numerous; but perhaps this was simply because centres had become active more nearly at the same time.

If degree of starvation and possibly cell contact were the main factors involved, a group of replicate cultures started every few hours with amoebae or spores evenly distributed through the bacteria should contain aggregations throughout the day; so should a group of cultures started simultaneously with different amounts of bacteria. In the present work, this was found not to be the case in D. discoideum-. there was a strong tendency for new aggregations to appear in outbursts lasting a few hours and rather well synchronized in different cultures, even in those showing between them a wide range of population densities. A plate of bacteria developed concentric zones of fruiting bodies and migrating aggregates immediately around the point where discoideum spores or amoebae were inoculated, and an expanse of feeding cells farther out. But the area in between did not contain all the intervening stages of development: for periods of perhaps 12 h or more, there was nothing it in but preaggregation cells ; then, in many cases within 2-3 h nearly all of these were marshalled into aggregations, which transformed into migrating aggregates about 8 h later. Beyond them widened a zone of cells that had stopped feeding and were awaiting the next burst of social behaviour.

In contrast, with a strain of D. mucoroides that formed a preplasmodium, point-inoculated cultures showed almost continuously all the zones that Arndt had described; and replicated spread-cultures started at intervals aggregated in succession at about the same intervals. This difference in behaviour from D. discoideum was maintained even when the two species were grown on alternate agar sectors inside the same Petri-dish and were thus exposed to almost the same environmental stimuli.

In an attempt to determine the factor that stimulated D. discoideum, cultures of this species, inoculated in a limited area, were variously exposed (fig. 5) to light and darkness, and hourly counts were made of the new aggregations that appeared (using, in the dark phase, momentary illumination by weak red light). After each change from darkness to light, or ‘dawn’, there was an outburst of aggregation. In one experiment, with cultures kept at 26o C, this began in the first hour, reached its peak in the third or fourth hour, and then declined; at 160 C the peak came 2 h later. At 26° C, bursts followed each dawn when there were alternating 6-h or even 3-h periods of light and darkness. With 12-h periods they apparently anticipated the second and subsequent dawns, beginning a few hours before them. Investigation showed that this was due to the fusion of outbursts in pairs: one induced by light with another that at this particular temperature developed in the dark about 20 h after the previous dawn. An outburst of the second type occurred alone when a culture was placed in continuous darkness after at least one dawn, and it was followed at intervals by other bursts that became progressively less distinct. The probable explanation was that the aggregation stimulated by a single period of light collected all the cells that might otherwise have initiated centres during the ensuing hours, and that there was then a considerable delay before a further crop of cells had developed, in the dark, to the point where they could start aggregation. This again removed any cells on the threshold of social behaviour, and so the process continued ; but as the clearances of a culture were slower and less thorough in darkness, synchrony deteriorated with each burst.

A culture in which the cells were widely separated and one in which they formed a continuous sheet tended to yield simultaneous outbursts when activated by light.

It is not clear that all known cases of synchrony can yet be accounted for even in D. discoideum, and in D. purpureum and P. violaceum the appearance of new aggregations followed a more complicated pattern. Moreover, especially in these species, it sometimes happened that a large proportion and even a majority of the centres spontaneously disintegrated within an interval as short as an hour (fig. 3, E); possibly chemical influences were involved.

The initiation of a primary centre is a process that is still little understood but is clearly rather complex. For example, unaggregated, acrasin-sensitive. cells of P. violaceum could be piled up into heaps after numbers of minute streamless centres had formed spontaneously in the population ; yet even if the heaps were much larger, they did not usually themselves develop into new centres, but dispersed. Later, when the cells were at a considerably lower density and had much less contact with one another, the same population might produce further centres; but again, it might not do so.

Quantitative studies on initiation were hindered by the difficulty of determining just how many primary centres had formed in a culture. Even at the same population density, aggregations might show a considerable size range which varied with the rate at which primary centres developed. Spontaneous breakage and disintegration led the larger aggregations to increase in number by fragmentation, and the smaller ones to disappear without trace. In addition, the cells released by disintegration might form further primary centres, and in some species, unaggregated and de-integrated cells might temporarily or permanently lose their acrasin sensitivity.

Sussman and Noel (1952) found that, at population densities that they claimed did not limit centre formation, the number of centres they counted was proportional to the number of amoebae present, there being, for example, i per 2,110 cells in D. discoideum. As the count at first increased with the population density and then fell off again rather rapidly, there may in fact have been a significant limitation when the maximal number appeared, because this peak could have occurred when some density-dependent inhibitory process outstripped another density-dependent process increasing the number of centres. Thus, even if their method enabled them to make an accurate count of the primary centres, the precise significance of their figures was doubtful.

These authors also examined small, replicate population samples and found that not all of them produced centres; they considered this to be evidence that by the end of the growth period the population was heterogeneous with respect to the ability to initiate aggregation. Perhaps, though, all that could safely be concluded from their data was that initiation involved a change in the cells that occurred with a certain low probability in the conditions used.

Primary centres are founded by cells that become integrated when there are no other integrated cells near them, and they may gain (though not necessarily retain) control of the movements of the surrounding cells simply by being the first to set up acrasin gradients adequate for orientation over even short distances and the first to activate the acrasin relay. However, it has proved extremely difficult to decide whether the initiator of a centre may be a single cell or whether it must be a small group. Direct visual observation tends to detect a centre too late to be sure, and the time-lapse films so far made that record the development of a centre have been taken at too low a magnification or too high a cell density to resolve this point. Besides, if orientation did seem to be initially towards a single cell, this one might just be the only stationary member of a group of acrasin secretors. Even if it were at that moment the sole secretor, some contact between cells might have been essential at a slightly earlier stage.

Sussman (1952) attempted to settle the question by examining the formation of centres in mixed populations of the wild type of D. discoideum and a mutant of it that was unable to aggregate when by itself but was attracted to wild-type centres. When the density of the mutant cells was increased, the number of centres formed in the presence of a given number of wild-type cells rose, but it eventually reached a maximum value. As this maximum varied directly, and not exponentially, with the number (and therefore the density) of the wild type, it was concluded that no collaboration between wild-type cells was necessary and that the initiators were single cells. This argument rested on the assumption that the mutant was unable to contribute to the production of the aggregative stimulus. However, the evidence published was that when the wild type was dispersed to such an extent that, alone, it could not aggregate, the addition of mutant cells permitted the development of centres and normal fruiting bodies, in which only wild-type spores were detectable. The two cell-types must therefore have been able to promote each other’s aggregation; and until the nature of the interaction is known (perhaps there was a transfer of acrasin precursors), the possibility that the initiators contained mutant cells cannot be excluded. (The cells of this mutant placed on a thin agar membrane merely collected opposite wild-type centres that were on the other side and did not form streams of their own (Sussman and Lee, 1955); and one presumes this was because, by themselves, they were unable to operate the acrasin relay. The same agar barrier frustrated the synergistic development of all the pairs of morphogen etically deficient strains that in mixed populations could develop further than could either partner separately (Sussman, 1954).) Furthermore, the wild-type cells were never present in the mixtures at less than a tenth of the density at which they produced the maximum number of centres when by themselves. Were most of them needed only to maintain established centres, or did initiation require some form of interaction between wild-type amoebae, as well as between wild type and mutant? If their density influenced only some steps in the sequence leading to initiation, it may well not have been limiting in the conditions used.

When cells of the mutant fruity were mixed with the wild type, centres formed in excess of the number that would have been produced by the wild type alone; and Sussman (19556) concluded that these were each initiated by a single mutant cell. Here again, his own evidence—that washings of fruity cells increased the number of centres formed by pure wild-type populations— vitiated his conclusion.

The maximum number of centres built by the unmixed wild type was 1 per 2,110 cells, but more appeared in the presence of a very high density of aggregateless cells, the precise increase depending on the particular aggregateless stock used. Sussman (1952) tentatively supposed that there were differences in initiating capacity in the wild type, and that the number of centres actually formed depended on the density of the ‘responder’ cells and on their sensitivity to the initiator’s stimulus. However, on this basis it is hard to explain why no more initiators were revealed in high-density wild-type populations when the possibility of interference between initiators was excluded (Sussman and Noel, 1952; this result was obtained when these authors were still seeking evidence that the initiators were single cells, and was cited as such). In any case, the method of analysis used could not in reality distinguish between a response to the initiators and a response of the initiators. The so-called ‘responder’ cells may, of course, have participated at more than one stage, before, during, or after initiation. (If they did so, the presence in a mixture of wild type and aggregateless of a second aggregateless stock less ‘sensitive’ than the first might well, with some pairs of mutants, have yielded additional centres.)

Recently, Sussman (1956) has presented evidence that the initiators arise not via a stable genetic alteration but by the exploitation of random physiological differences; however, it has still to be determined at just what stage the relevant biochemical heterogeneity appears in the population, whether there is any corresponding visible heterogeneity, and whether a single cell can consummate the initiation process. One can scarcely hope for a real understanding of initiation till it has been described in biochemical terms. For the present, one may tentatively suppose: firstly, that the initiators are distinguished by their ability to release acrasin into the surrounding medium without there being any there already or without their being sufficient to induce the remaining cells to secrete; secondly, that this ability is either the immediate or the indirect consequence of the chance appearance of a peculiar and rather improbable configuration within amoebae that are temporarily in a particular physiological condition; and thirdly, that the environment provided by the rest of the population exerts a considerable influence.

The existence of sexuality in cellular slime moulds was first claimed by Skupienski (1918); he was able to observe syngamy occurring though he found it to be inhibited by the least light. On his findings then, syngamy could not be an essential part of aggregation, for this readily takes place in daylight. Wilson (1952, 1953) believed aggregation to be basically a sexual phenomenon: the centres were started by the first zygotes to form, and as others appeared they joined the streams ; meioses took place during aggregation but were not synchronized. His evidence was cytological and taken from stained culture smears; these were doubly inadequate in that they could show neither the real sequence of development, as a study of the living cells could have done, nor even the actual positions of the fusing and dividing cells in the organization, as would have been revealed by cultures fixed and stained in situ. Many objections to Wilson’s interpretation are discussed in Sussman’s review (1955a), though that based on the quantitative study of initiation is not as serious as he supposed: mention will be made here only of findings in the present work that are not compatible with the view that aggregation is essentially based on syngamy.

1. Visual observation shows that cells may enter streams without first forming zygotes and, in thin streams where observation may be easily continued, may not form them later.

2. Single sensitive cells can be attracted to a natural or an artificial acrasin source and can also be induced to make acrasin. It is therefore hard to see what vital role syngamy could take in the growth of an aggregate.

3. The same cells may undergo integration and disintegration several times. If these changes were associated with a cycle of fusion and meiosis, one might expect there to be a noticeable reduction in cell size : this has not been observed.

4. Amoebae are able to invade and eat a layer of bacteria sandwiched between agar and a coverslip and then aggregate in this position. If a fixative is added, they adhere to the glass. So it is possible to obtain preparations, stained with aceto-orcein, containing feeding cells, preaggregation cells, and entire and largely monolayer aggregations, in their natural positions. Every nucleus in all the streams and in the separate cells just beyond their outer ends may be in interphase.

The possibility that syngamy does sometimes occur cannot of course be excluded. However, as the genetic make-up of the fruiting body depends on that of all its component cells, there may still be some ‘recombination’ even in the absence of a sexual cycle, if there is any naturally occurring, genetically determined variation in those properties of individual cells of importance in fruiting. Selection will favour fruiting bodies with a genetic constitution that best adapts them to the microenvironment.

The work reported in this series of papers was initiated while I held an A.R.C. Studentship; it was completed in 1954 while I was a Dill Fellow at Princeton University, where I enjoyed considerable hospitality from members of the Department of Biology. In particular, Dr. J. T. Bonner was always most interested and helpful, and among many other kindnesses provided me with cultures, and Dr. C. S. Pittendrigh and some of his students generously assisted me with the tedious task of making hourly counts of aggregations. My thanks are due to Dr. B. N. Singh for sending me cultures initially, and I am especially grateful to Dr. M. G. M. Pryor for his invaluable encouragement and criticism throughout.