How social evolution theory impacts our understanding of development in the social amoeba Dictyostelium


Author to whom all correspondence should be addressed.


Dictyostelium discoideum has been very useful for elucidating principles of development over the last 50 years, but a key attribute means there is a lot to be learned from a very different intellectual tradition: social evolution. Because Dictyostelium arrives at multicellularity by aggregation instead of through a single-cell bottleneck, the multicellular body could be made up of genetically distinct cells. If they are genetically distinct, natural selection will result in conflict over which cells become fertile spores and which become dead stalk cells. Evidence for this conflict includes unequal representation of two genetically different clones in spores of a chimera, the poison-like differentiation inducing factor (DIF) system that appears to involve some cells forcing others to become stalk, and reduced functionality in migrating chimeras. Understanding how selection operates on chimeras of genetically distinct clones is crucial for a comprehensive view of Dictyostelium multicellularity. In nature, Dictyostelium fruiting bodies are often clonal, or nearly so, meaning development will often be very cooperative. Relatedness levels tell us what benefits must be present for sociality to evolve. Therefore it is important to measure relatedness in nature, show that it has an impact on cooperation in the laboratory, and investigate genes that Dictyostelium uses to discriminate between relatives and non-relatives. Clearly, there is a promising future for research at the interface of development and social evolution in this fascinating group.


Multicellular development in Dictyostelium offers biologists an unusual look at development, because it has evolved independently from the multicellular development of the more traditional animals and plants (Baldauf et al. 2000). Though multicellularity in Dictyostelium is built using some of the same kinds of tools as animals and plants, including signaling, adhesion, and cell death, the details usually differ (Kessin 2001; Li & Purugganan 2010; Williams 2010). Development in Dictyostelium also differs in some fundamental ways. The multicellular stage is strictly for dispersal and cannot feed. There is little cell division during development. The number of tissue types in the multicellular stage is limited to two major categories, spore and stalk. In early years, the separation of development from cell division and the paucity of cell types made Dictyostelium very attractive for studies of development (Kessin 2001).

But there is another difference that early developmental biologists used, perhaps without fully appreciating the evolutionary dilemma it posed – that Dictyostelium comes to multicellularity through aggregation, and not through the mitotic divisions of a single original cell. The evolutionary consequences and ramifications of multicellularity through aggregation are the focus of this paper.

Some might feel that this is a developmental aberration of little consequence, so we point out that the potential impact of this difference is huge (Strassmann et al. 2000). It may be the reason that this lineage has maintained a simple, highly conserved multicellular form, while the animals, for example, diverged into giraffes, corals, squids, lobsters, earthworms, and eagles (Queller & Strassmann 2009; Strassmann & Queller 2010). For reasons we describe below, the importance of a single cell bottleneck for the evolution of somatic variety is profound (Grosberg & Strathmann 1998).

Development without a single cell bottleneck will be less harmonious

When a body develops from clonal division of a single cell, all the daughter cells are genetically identical, except for occasional, rare mutations. These new chimeras are not passed on, since every generation begins again with a single cell (Queller 2000). This makes it easy to evolve differentiation, which is a form of cooperation among cells. Perhaps the most fundamental level of differentiation is the one between germline and somatic cells. Somatic cells die without issue, so one might think that the tendency to become somatic would be removed by natural selection. However, somatic cells do things that aid the reproduction of the genetically identical germline cells. The sterility of somatic cells tends to decrease the number of sterility genes reaching the next generation, but their aid to identical reproductive cells increases the frequency of those genes. The process works only if sterility is not a fixed trait, but instead is a conditional trait based on something not connected to genes, like where they are in the cell cycle, or their relative nutritive condition when the bifurcation towards spore or stalk occurs. Unconditional sterility mutants would all become sterile and have no one of the same type to help, and so would die out.

In the uniclonal case, sterile cells are easily evolved because clonemates are around to be helped. In the extreme opposite, where multicellular bodies are formed by aggregation of a random subset of cells in the population, cells with conditional sterility alleles would not prosper. No matter how much they helped the multicellular body, they could not be favored by selection. They give up their own lives in return for helping a random set of genes in other cells, which does nothing to increase the frequency of the sterility allele. It does nothing because helping random individuals does not alter their relative frequencies. Both sterility alleles and the alternative non-sterility alleles receive the same amount of help. Thus the sterility alleles pay a fitness cost, and everyone receives a benefit, so the alternative alleles do better. Because Dictyostelium bodies form by aggregation, we have to ask both how they got around this problem and evolved differentiated bodies, and whether this issue alters the nature of their development.

The argument we have laid out here may not be familiar to many developmental biologists who are used to working with the standard uniclonal kind of development. However, the argument will be familiar to a different group of biologists, sociobiologists, who have worked over the last several decades on the evolution of social interactions. Their questions go back to Darwin (1859), who worried about how sterility could evolve in the worker caste of social insects. Darwin decided that it could evolve because selection could operate at the level of the family, if families with more sterile workers function better and leave more reproductive offspring.

Kin selection and its role in Dictyostelium sociality and development

Darwin did not know about genetics, and his argument works only under certain conditions. It was W.D. Hamilton who in the 1960s made the argument genetically rigorous and quantitative (Hamilton 1964). He not only solved the problem of social insect sterility, but also provided a framework for understanding all kinds of social interactions including the interactions among cells during development. Hamilton (1964) derived a mathematical approach to social traits using a modeling framework he called inclusive fitness. This approach has been shown to be robust, general, and quite profound in its ability to explain the evolution of social traits (Queller 1992; Frank 1998; West et al. 2007). Another relevant term is kin selection, which can be viewed as a generalization of natural selection that includes passing on genes not just by rearing progeny, but also through other forms of relatives.

According to Hamilton’s Rule, genetic alleles that decrease an individual’s direct fitness (for example, the number of babies she produces) can increase in the population only by increasing her indirect fitness, as would happen when she helps relatives. This is typically expressed as rb − c > 0 where r is genetic relatedness between the altruist and the beneficiary, b is the fitness gain experienced by the beneficiary from the altruist’s aid, and c is the reproduction the altruist relinquished to help the beneficiary.

We can apply this framework to Dictyostelium. The sterility of a somatic cell entails a fitness cost c, but its somatic work gives a net benefit, b, to the reproductive cells in its body. If the cells are genetically identical, r = 1, and Hamilton’s rule tells us that becoming a somatic cell will be favored if c < b. In contrast, if the cells are a random group, then r = 0. Hamilton’s rule then tells us that somatic sterility will be favored when c < 0 which is never satisfied because there is an obvious cost to becoming sterile.

Another way of looking at the problem is to assume that we already have an evolved multicellular body, and then ask what effect low relatedness among cells would have. Continuing with our assumption that the cells are genetically random, imagine a cheater mutant that is better at getting into spores than wild type, though it decreases the fruiting body’s overall success (total spore production). The fbxA knockout mutant is an example (Ennis et al. 2000; Gilbert et al. 2007). Now Hamilton’s rule says that it will be favored when b > rc, where b is the benefit to itself and c is the cost to other cells in the fruiting body. Since r = 0, the selfish cheater will be favored whenever b > 0; therefore, all it has to do is benefit itself, regardless of how much harm it does to the fruiting body. So the altruistic self-sacrifice that we see in development is expected to be unstable if the players are unrelated to each other.

What is genetic relatedness and why is it important?

In general, relatedness measures the degree of non-randomness of groups, or how much genetically similar individuals group together. It does not measure the overall fraction of shared genes, but only those genes more similar than average for the population. Dictyostelium groups fall above random (r = 0) but probably fail to reach perfect assortment (r = 1). This is because a lot of the amoebae in an aggregation territory will be the mitotic progeny of a mother cell only a few generations back, but others could be the daughters of a different mother cell. Unrelated mother cells are likely to share a high fraction of their overall genes, but this is not what matters for determining genetic relatedness, since it is the genes identical by descent above random that matter for favoring an altruistic trait.

From a relatedness level somewhere above 0, but not 1, we might expect intermediate levels of cooperation and also intermediate levels of cheating on that cooperation. Cheating is a term that is defined by an expectation of fairness. In this case the fair outcome would be that representation of each clone in a chimera at the start of aggregation would define the representation in spores at the end of the social stage. Therefore, a cheating clone would be overrepresented in the spores as compared to its representation going into the social stage. Inclusive fitness theory predicts that relatedness within the developmental (also called social) stage aggregations will be high, because of the substantial altruism by stalk cells towards spore cells. But how high is it, and how does it affect cooperation among cells? If chimeric mixtures occur with non-negligible frequency, then development in Dictyostelium might involve a kind of conflict among cells unheard of in the development from a single cell found in animals and plants. It also means that only by taking this conflict into account can we truly understand the multicellular stage of Dictyostelium.

What questions will a social evolution perspective answer?

To pursue these questions, a very different research program from the standard developmental one is required. This other approach explores what is sometimes called ultimate causation, why a certain trait evolved, and why it is adaptive. By contrast, the developmental approach asks what can be called proximate questions, about the functional mechanisms of a trait. We should be very clear that neither approach is better or worse than the other, but their goals are very different (Mayr 1993). For the ultimate approach, it is important to understand how Dictyostelium operates under natural field conditions. It is important to understand the genetic population structure of the species, so we know how likely different clones are to encounter each other. It is important to understand the costs and benefits of the social stage. It is important to understand genetic relatedness in the social stage. Though many of these questions require an approach that would be novel in a developmental laboratory, it is standard in sociobiology. One might argue that the most productive approach of all is to meld the experimental approach of social evolution with the molecular wizardry of cell biology, throwing in a gigabyte or two of genomics for good measure. Then we could know both why a trait evolved and how it works. With this paper, we will attempt to make the discoveries from the social evolution perspective clear, and also explain some of the successes of the collaborative approach.

Lest we lose the developmental biology readership at this point, it is important to make clear that the evolutionary approach actually helps reveal Dictyostelium attributes of crucial importance to the developmental biologist. We can contribute to answering the key question about whether the social, or developmental, stage of Dictyostelium is more like clonal development, or more like a population of genetically distinct competing individuals. Molecular evolution approaches can tell us much about how selection is operating on different genes, and even identify their most important regions. Evolutionary approaches reveal why stalk fate is environmentally determined, add insight to the competitive nature of DIF-1 in spore/stalk allocation, and have opened up a whole new field of the tgr genes involved in recognition (Zahavi & Zahavi 1997; Strassmann et al. 2000; Thompson et al. 2004; Benabentos et al. 2009). Furthermore, gene expression in chimeras versus pure clones will tell us much about which genes are there for social functions. In short, understanding the social perspective as well as the developmental one is likely to be productive.

Early contributors to the sociobiological perspective on Dictyostelium

John Bonner’s many thoughtful books and articles brought Dictyostelium to the attention of many students of social behavior and evolution (Bonner 1944, 1967; Gadagkar & Bonner 1994). Filosa (1962) studied D. mucoroides that had been isolated from giraffe dung, and then kept in culture for as long as 8 years, and found morphological variants that most likely arose in the laboratory. One of the more interesting mutants was a social parasite that could only form fruiting bodies in chimera (Filosa 1962). Buss (1982) published an early report on chimerism and recognized the significance of cheating in D. mucoroides. Several theoretical models have attempted to analyze selection on cooperation and cheating (Armstrong 1984; Matsuda & Harada 1990; Matapurkar & Watve 1997). Taken together, these and other early reports made it clear that a sociobiological perspective to Dictyostelium was likely to be fruitful (Kaushik & Nanjundiah 2003; Queller et al. 2003a; Shaulsky & Kessin 2007; Li & Purugganan 2010). This approach is also proving successful in a wide range of microbial studies (Crespi 2001; Strassmann & Queller 2004; Travisano & Velicer 2004; Foster et al. 2007; West et al. 2007).

Other research has provided a different kind of evolutionary perspective. Raper (1984) published a very valuable systematic treatment of the group that has allowed a phylogenetic perspective, particularly once Baldauf placed the group properly in the tree of life (Raper 1984; Baldauf et al. 2000; Schaap et al. 2006).

Why does Dictyostelium have a social stage: Benefits of a stalk

One of the things we do not know about Dictyostelium discoideum, or any other Dictyostelium species, is how often they go through the social stage in nature. This is frustrating because we can find cells or spores easily in most samples of suitable soils (Cavender 1978; Raper 1984; Cavender et al. 1995; Fortunato et al. 2003), but our own work shows that fruiting bodies are hard to find (Gilbert et al. 2007). They may be rare; they may be short lived; they may protrude into soil interstices rather than on the surface. In the laboratory, by contrast, cell division, nutrient exhausting, and fruiting body formation is easy enough to manipulate for undergraduate laboratory use. In the laboratory with plenty of bacteria that are good to eat, cells divide about every 4 h, and fruiting bodies are formed in a few days (Kessin 2001). If the cells divided without limits in nature, our planet would be nothing but Dictyostelium, but of course, like every other organism, they are limited by food and by predators, parasites, and diseases. And these limits give us an idea as to why they should ever go through the social stage. D. discoideum can only form spores as a consequence of the social process, and these spores can endure for years in a dormant stage.

But the key feature of the social stage for us is that it involves the formation of a stalk of cells that have died to lift the others above the substrate. This stalk has been hypothesized to offer at least two kinds of advantage. It allows for sporulation to proceed above the soil, away from fungi, bacteria, and other organisms that might prey upon the sporulating cell. Possibly more importantly, it facilitates transport because the elevation of the stalk increases the odds that the spores will be transported away. Figuring out what these advantages are will give us more information on the benefits of the social stage. For now, we assume the stalk has an advantage, and spend the rest of this paper investigating how this level of social cell death can evolve.

What is genetic relatedness in naturally occurring fruiting bodies?

Understanding genetic relatedness between interacting individuals is key to understanding the evolution of cooperation. The relatedness that is relevant to the evolved developmental program is that in nature, in naturally occurring fruiting bodies. A number of studies have tackled this important question. At Mountain Lake Biological Station, Virginia, we analyzed relatedness within soil samples of 0.2 g, 6 mm diameter, that we reasoned might be a good dimension for natural encounters among amoebae (Fortunato et al. 2003). In this study we found 102 isolates of D. discoideum that were 46 genetically distinct haplotypes. (A haplotype is a unique combination of alleles.) Genetic relatedness within the soil samples was r = 0.519 ± 0.014. If these mixed freely and equally, this would be the value for genetic relatedness within fruiting bodies.

In another more recent study we measured actual genetic relatedness in wild fruiting bodies from the same location, and found much higher values (Gilbert et al. 2007). In this study some fruiting bodies were collected directly from the field, and others developed from deer dung that was placed on non-nutrient agar plates for moisture, and otherwise unmolested. We then genotyped either the entire fruiting body, or individual clonal isolates from the fruiting bodies. We used three polymorphic microsatellite loci. Of 88 whole fruiting bodies we found that 23% were chimeric, giving a minimum relatedness of 0.86 if the clones mixed equally. If we use the data from 1039 spores from 75 other wild fruiting bodies, we find only 8% were chimeric. In these chimeric fruiting bodies relatedness was 0.684 ± 0.086, and adding in the clonal ones, relatedness was 0.975 ± 0.012 averaging all fruiting bodies in this sample (Gilbert et al. 2007).

In India, researchers studying D. purpureum and D. giganteum found that 15 of 17 fruiting bodies were chimeric (Sathe et al. 2010). They looked at both soil samples, and samples from dung of spotted deer, wild dog, and elephant, and used three to eight random amplification of polymorphic DNA (RAPD) markers and two different analytical techniques. Their second method yielded 1–28 clones per fruiting body (6.55 ± 7.42 SD, n = 11) for D. giganteum, and three to nine clones per fruiting body (5.83 ± 2.04 SD, n= 6) for D. purpureum (Sathe et al. 2010). These numbers yield within-fruiting-body relatednesses considerably lower than were found for D. discoideum. The high values of relatedness within fruiting bodies of D. discoideum mean it will not be hard to understand why cooperation occurs in this species. It is harder to understand for the other species.

In the next sections we discuss how high relatedness values within D. discoideum fruiting bodies are attained, and the consequences of chimerism and low relatedness.

Do Dictyostelium clones favor kin in the social stage?

The discrepancy between high relatedness in wild fruiting bodies, and lower relatedness in 0.2 g soil samples could be explained if amoebae were able to detect clonemates and preferentially associate with them in the social process. We expect that this would happen after the aggregation stage, since aggregates can even include different species that are attracted to cyclic adenosine monophosphate (cAMP) (Jack et al. 2008). We also expect that sorting on the basis of clone identity would occur before any cell had committed to helping a non-relative. In D. discoideum irreversible stalk membership does not occur until the slug has stopped migrating, late in the social stage, something that is unusual in the genus (Raper 1984).

A recent study on D. purpureum discovered a substantial degree of sorting when pairs of clones were mixed in equal proportions on agar (Mehdiabadi et al. 2006). There is also evidence for a lower level of sorting under similar conditions in D. discoideum (West et al. 2007; Ostrowski et al. 2008; Flowers et al. 2010). Ostrowski et al. (2008) found that sorting increased with genetic distance in a study that used clones from Texas to Massachusetts and measured sorting by mixing each clone with green fluorescent protein (GFP)-labeled Ax4. Flowers et al. (2010) also found sorting among a similar set of clones using pyrosequencing of two variable single nucleotide polymorphism (SNP) markers. They did not find a significant decay with distance, but surprisingly found higher levels of sorting when both clones were from Mountain Lake Biological Station than when both clones were from different locations (Flowers et al. 2010). Yet neither of these studies showed the level of sorting on clonal lines required to explain the very high relatednesses found in wild fruiting bodies. Clearly more work is needed to understand natural and laboratory patterns of sorting in D. discoideum. For now, we explore the mechanisms for kin discrimination, and the results of chimerism.

Genes for kin discrimination in D. discoideum: csaA

In microorganisms, kin discrimination is likely to take the form of greenbeard discrimination (Strassmann & Queller 2011). Greenbeard discrimination is based on an idea of Hamilton (1964), that was colorfully described by Dawkins (1976). The idea is that if there is a trait that provides a cue (such as a green beard), recognizes that cue, and then causes a cooperative action, it would spread as a cozy bundle of recognition and altruism. At first investigators postulated that such traits would not occur, because they would set up selection for individuals that carried the cue, but did not cooperate. So such individuals would gain the benefits without paying the cost. This separation would be harder to achieve if all three attributes were the product of a single gene, but that also seemed rather improbable (Dawkins 1976).

Haig (1996) first postulated that adhesion genes might function as greenbeards. We found this to be the case in D. discoideum and the cell adhesion gene csaA (Queller et al. 2003b). This gene, contact site A, is one of the first cell adhesion genes described in any organism. It is a homophilic cell adhesion gene whose protein product is anchored in the cell membrane and migrates to lipid raft regions of the membrane where it adheres in the social stage to other cells. With chimeras of csaA knockouts, and unmanipulated wild type cells, on agar, the csaA knockouts behaved like greenbeard cheaters. They preferentially fell to the back of the slug where they became prespore, taking advantage of the wild type that ended up in the stalk preferentially. In what are potentially more natural conditions on soil, csaA knockouts were underrepresented in the slug and the fruiting body generally, apparently because they were unable to aggregate properly. Thus, wild type csaA is a greenbeard gene. Clones that express the gene recognize, aggregate, adhere, and some die to form stalks that lift the others, thus fulfilling all the attributes of a greenbeard gene.

Genes for kin discrimination in D. discoideum: tgrC and tgrB (formerly lagC and lagB)

Genes that are actively used by an organism for preferentially forming associations should be highly variable, providing active grounds for discrimination. They should produce a product that allows for discrimination, perhaps an odor, an appearance, or, most likely in microbes, a membrane-bound adhesion molecule. The molecules should either recognize self, and have a homophilic binding site, or they should recognize a receptor that is equally variable. The tgrC and tgrB (formerly called lagC and lagB) genes fit these conditions (Benabentos et al. 2009). They are cell adhesion genes that are anchored in the cell membrane. They are highly variable. The protein products of tgrC appear to adhere to the protein products of tgrB on other cells, causing recognition-based adhesion. TgrC knockouts aggregate, but do not progress past this stage, perhaps because no partners are recognized. The very interesting work that has already been done on these genes is likely to be only the beginning of a very fascinating evolutionary story of greenbeard kin discrimination.

Why might this be important to developmental biologists? There are a number of reasons, perhaps most important is that downstream gene expression with matching partners could be different from that in cells that differ at these tgr genes. If cells can distinguish clonemates and non-clonemates, or can tell when they are in a chimera, kin selection theory predicts that they will modulate their cooperation and cheating accordingly. Exploring the multicellular stages with matching and differing tgrC and tgrB genes could reveal unknown D. discoideum capabilities.

What are the consequences of relatedness manipulation in the laboratory?

In the laboratory it is not difficult to generate genetic chimeras and test them to see how they behave (Strassmann et al. 2000; Foster et al. 2002; Buttery et al. 2010). In chimeras it seems clear that selfish behavior increases (Fig. 1). Chimeric slugs move less far than clonal slugs when they are subjected to directional light on a Petri plate (Foster et al. 2002). This can be interpreted as clones competing with each other to become spore and not stalk. Movement towards light is likely to be delayed if clones compete to occupy the prespore zone in the back three quarters of the slug. In addition, chimeras typically produce more spores for a given number of cells (Buttery et al. 2009). This is because there is less selective benefit to being a stalk cell if the main beneficiary is a non-relative.

Figure 1.

 Costs of chimerism. Chimeras invest less in stalk compared to clones. When size is controlled for, chimeric slugs move less far compared to clones. In chimeras some clones may cheat others by winning a larger share of spore cells, relative to stalk cells. Pink color indicates no information on cheating. Benefits of chimerism. Chimeras can become larger than clones because all amoebae in the environment can join the aggregation, not just clonemates. Larger slugs form larger fruiting bodies, than generally invest a lower proportion of cells in stalk. Larger slugs move farther, more than overcoming the tendency of chimeras to move less far than pure clones.Copyright CC 3.0 Joan E. Strassmann.

In microbes, we can manipulate the evolutionary environment in the lab, and allow selection to proceed under either the apparently natural high-relatedness condition or under a novel low-relatedness condition. One such project began with 12 replicates of eight wild clones, and put them through 10 fruiting generations at both high and low relatedness (Saxer et al. 2010). The interesting outcome of this experiment was that in all 12 replicates a few clones predominated. Furthermore, only one clone predominated in every high-relatedness replicate, while three predominated in every low-relatedness replicate. We interpret this to mean that it is hard to both dominate other clones in chimeras and also excel at producing clonal fruiting bodies.

Cheating, and competition in chimeric fruiting bodies

Chimeric slugs of the Dictyostelium multicellular stage behave differently from clonal ones, as discussed above. Now we examine the fates of distinct partners in that multicellular stage. Do some fare better than others? In this case, faring better would mean succeeding in achieving a greater share of spores than might be expected from a fair division, where a fair division means that clones do not differ from each other in their spore/stalk ratios. For example, if cells from two clones are mixed in equal proportions, a fair outcome would mean that the proportions of stalk cells should also be equal, as should the proportions of spore cells. If they are not, we would say the clone with more spore cells is cheating the other clone (Fig. 2).

Figure 2.

 Kinds of cheating. Under fixed cheating, clones maintain in chimera the different spore/stalk allocations they have alone. Under facultative cheating, the cheating clone allocates less to stalk in chimera than it does alone. A social parasite cannot make stalk at all alone, but takes advantage of the stalk made by another clone when in chimera. The social parasite has no fitness except in chimera, and can be viewed as an extreme form of fixed cheating.Copyright CC 3.0 Joan E. Strassmann.

The possibilities for social interactions can be complicated in ways that impact what we call cheating. Imagine two children eating simultaneously from a plate of 10 cookies. If Susie ate seven cookies, and Erica ate only three, we might say that Susie cheated Erica out of two cookies. This is the cheating situation discussed above. But what if we obtain the additional information that Susie always eats more quickly than Erica, even when she is alone? In this case, each child was behaving identically in the social situation as in the nonsocial situation, so some might not call this cheating. On the other hand, if Susie and Erica eat cookies at the same pace when alone, but Susie speeds up when someone else is eating from the plate, we would certainly call her a cheater. These two behaviors can also be called fixed and facultative cheating (Buttery et al. 2009). In many organisms it might not be possible to determine which is occurring, but it is possible in D. discoideum.

To determine whether fixed or facultative cheating, or no cheating is occurring in Dictyostelium chimeras, it is necessary to look at allocation to spores under pure clone conditions and compare that to social conditions. In a careful study Buttery et al. (2009) discovered facultative cheating in chimeras; clones altered their solitary strategies in chimeras. Furthermore, more spores were produced in chimeras than in equally sized pure clones.

Genes that influence cheating in D. discoideum

Earlier we discussed the tgrC and tgrB genes that are involved in kin discrimination in D. discoideum. Though these genes clearly influence sorting, it is not complete, so there is an opportunity for competitive interactions among clones in the multicellular body. Above we presented evidence for competition, and now we explore the genetic basis of competitive interactions.

One way of identifying genes involved in cheating interactions is to make a library of clones that each have a single gene disrupted. In Dictyostelium this is typically done using restriction enzyme mediated integration (REMI). This pool can be subjected to selection that favors certain kinds of mutants, and then these mutants can be identified using the sequence flanking the REMI insertion. The first to use this approach identified a very interesting and powerful gene called cheaterA, or chtA, now called fbxA (Ennis et al. 2000). Using similar procedures, to identify cheater genes, we subjected several pools of REMI mutants to successive rounds of the social stage at low relatednesses. We plated out the cells, allowed them to feed and aggregate, then collected the spores and repeated. Obligate cheaters were removed with a clonal plating step at round 10.

These procedures identified many genes with a wide variety of characteristics (Santorelli et al. 2008). One of the interesting results of this experiment was the large variety of genes whose disruption resulted in cheating. We interpret this as evidence for the importance of pleiotropy for social genes, a subject developed further in the next section. Furthermore, these genes are a rich resource for further investigations of the social side of D. discoideum that we have only begun to explore.

Pleiotropy as a cooperative glue, dimA and csaA

In some cases we predicted that disrupted genes would cause their bearers to cheat wild type clones, but the prediction is not fulfilled, or not fulfilled under apparently more natural conditions. One such gene is the dimA knockout. This gene is involved in the perception of differentiation inducing factor, DIF, an oxyphenone that is produced by prespore cells, and causes other prespore cells to become prestalk O cells. DimA knockouts are unable to respond to DIF, and so we predicted that they would preferentially become spore cells in chimeras (Foster et al. 2004). They did become prespore early in the multicellular stage, but later the dimA knockouts redifferentiated and became stalk cells. We hypothesize that this is because of another function of the disrupted gene that is essential to becoming spore. Thus, the social function of response to DIF seems to be piggy-backed on an unknown essential function (Foster et al. 2004).

In some ways, the csaA knockout is similar. On agar, in equal mixtures of wild type and csaA knockouts, the knockouts slip to the back of the slug, and preferentially become spore. This makes the knockout a cheater. However, on soil, the knockouts do not succeed in even getting into the slug because of their weak adhesion abilities, so they are not cheaters (Queller et al. 2003b).

We suspect that much about social interactions in D. discoideum is mediated through cell adhesion, but it is beyond the scope of this review to go through all the cell adhesion genes and their functions.

Relatedness controls social parasites

The fbxA knockout cheater is a social parasite; though it produces few or no spores on its own, it can make spores by exploiting and cheating other clones (Ennis et al. 2000). Relatedness has a huge effect on how such a mutant would be selected (Gilbert et al. 2007), as shown in Figure 3. At low relatedness, when wild type and the fbxA knockout cheater are well mixed, the knockout wins at all mixture frequencies. This shows that the cheater mutant ought to increase easily under these conditions. At the same time, mixtures with higher frequencies of the cheater do less and less well in terms of overall spore production. Thus, selection for the mutant at low relatedness could drastically reduce or even eliminate spore production in the population. However, at high relatedness, each type experiences primarily its own type, meaning the cheater has no one to cheat, and is selected against because of its own low spore production (Gilbert et al. 2007). Given that our estimates show relatedness to be high in nature, socially parasitic mutants like this should not succeed (Gilbert et al. 2007). This prediction is upheld by data: when we grew up clones from several thousand spores collected from the field, none of them were social parasites that were developmentally incompetent on their own (Gilbert et al. 2007). High relatedness apparently controls the most devastating cheater mutants like the fbxA knockout, but not necessarily cheater mutants that are developmentally competent on their own (Santorelli et al. 2008).

Figure 3.

 Spore production per cell in mixtures of wild type (solid line) and the fbxA knockout social parasite (dashed line). At every mixture, the social parasite does relatively better than wild type, so in well-mixed populations the social parasite will always be favored. But higher frequencies of the social parasite also lower the absolute production of both types. At very high relatedness, the higher productivity of all-wild type groups (closed circle) compared to all-parasite groups (open circle) will select against the fbxA social parasite. Data from Gilbert et al. (2007).Copyright CC 3.0 Joan E. Strassmann.

Power – glucose and other environmental effects

If competition in chimeras has been an important part of selection in D. discoideum development, we would predict that stronger cells become reproductive spores would depend on their power to win the competition. It is better to become a spore than to become a stalk that supports non-relatives, and so cells will evolve to use whatever power they have to become spores. In organisms consisting of a single clone, cells would not be selected to compete strongly to be in the germline (Queller 2000) (Khare & Shaulsky 2006). An allele causing its cell to compete to get into the germline would simply replace other cells with the same allele.

The evidence on this question is somewhat mixed, but there are some strong indications that power plays a greater role in Dictyostelium development than in other multicellular development, the kind of role expected for chimeric organisms. It has long been known that cells deprived of glucose tend to become stalk cells, and this can be viewed as support for the idea that better fed cells exploit their better condition to become spores (Leach et al. 1973). There are at least two caveats, however. First, the effect could be something specific to glucose rather than to overall condition. This is not too likely because other metabolizable sugars have the same effect, and non-metabolizable sugars do not (Takeuchi et al. 1986). Furthermore, harming cells with an acid treatment had a similar effect (Castillo et al. 2011). The stressed cells became stalk. The second caveat concerns whether these are truly competitive effects. We found that, in both cases, the weakened cells also produced fewer spores than the stronger ones when each was grown alone, casting doubt on whether competition alone causes the effect (Castillo et al. 2011).

Power – DIF

However, certain features of D. discoideum development strongly suggest that power is used to become spores. DIF-1 is a chlorinated alkyl phenone produced by prespore cells that is broken down by prestalk O cells, and also cements their fate as stalk cells, often as part of the lower cup or basal disc (Kay 1998; Thompson & Kay 2000; Kessin 2001; Kay & Thompson 2009). It interacts with the glucose effect, with glucose-fed cells being producers of DIF and glucose-starved cells responding to it to become stalk. Of course DIF-1 could simply be a signal with no competitive function. But several features argue against that. It is very unusual for signals to incorporate chlorine atoms, but not so unusual for poisons. DIF-1 is indeed poisonous. Present in the slug at about 62 nmol/L (Kay 1998) it can be lethal at as low as 200 nmol/L (though higher under some conditions) (Masento et al. 1988). Given that no receptor has been found, it is worth considering the possibility that it is more of a weapon than a signal (Atzmony et al. 1997). However, it must be a weapon that is deployed gently because the goal is not to kill other cells but to change their fate and tilt the balance in the direction of making stalk and getting some benefit through helping kin.

Power – first strike

If getting into spores is competitive and is mediated by particular biochemical pathways such as DIF synthesis and breakdown, then an important factor might be when those weapons and defenses are first deployed. Those cells that enter the developmental cycle first would be the first to turn on offensive and defensive pathways, which might give them an advantage (Kuzdzal-Fick et al. 2010). It is true that the first cells to starve become spores, not stalk cells (Huang et al. 1997; Kuzdzal-Fick et al. 2010). This is interesting because a first-strike advantage has to work against the nutritional effect; presumably the first cells to starve are in poorer condition than later ones. That they are nevertheless better at getting into spores is consistent with drawing their “weapons” first.

Though several lines of evidence suggest that cells use power advantages to get into spores and to induce others to make stalk, pure competition is not a sufficient explanation for stalk production. Stronger cells may have the power to weaken or even kill other cells, but it seems implausible that they could force other cells to go through all the complex changes that lead to stalk formation. Instead, it is much more likely that strong cells simply affect the other cells’ optimal decision. It may be best to be a spore, but cells that are stressed and perhaps unable to make the best spores, make the best of a bad job and get some advantage by producing a stalk for relatives,

Conclusions and future prospects

We hope the next decade is as fruitful as the past one for Dictyostelium research. If this proves to be the case, Dictyostelium is likely to become the best-understood social organism from both evolutionary and mechanistic perspectives. Indeed, the most exciting prospects meld these two approaches into a synthetic view of a fascinating social organism.


We thank the members of our laboratory group, and Sandie Baldauf, John Bonner, Rob Kay, Rich Kessin, Adam Kuspa, Pauline Schaap, Gadi Shaulsky, Chris Thompson and many others at the Dictyostelium annual meetings for many helpful discussions. Our research is supported by the US National Science Foundation under Grant Nos. DEB 0816690 and DEB 0918931.