Group transformation: fruiting body and stalk formation

Throughout the eukaryotic tree of life, amoeboid organisms have evolved that aggregate upon starvation and form multicellular fruiting bodies, consisting of a ball of spores atop a stalk. This chapter discusses the remarkable convergent evolution of a stalked fruiting body in these different taxa. It then discusses a well-studied group of aggregative fruiters, the cellular slime molds, in more detail. These organisms exhibit substantial variation in their stalk formation and composition, which allows a better understanding of the evolution, maintenance and possible functions of stalked fruiting bodies, but also points to potential costs and benefits of different types of stalks.

and form a fruiting body. The final fruiting body structure consists of a small, thick stalk composed of secreted material and dead cells that supports a spherical, golden, mucoid sorus (Fig. 1E). The presence of dead stalk cells indicates that this organism shows reproductive division of labor as well.
Finally, while many examples of aggregative multicellular fruiting body formation involve sorocarpic amoebae, it is worth noting that there are additional non-amoeboid taxa that undergo cooperative fruiting.
Myxobacteria are a clade of prokaryotic organisms that undergo aggregative fruiting in response to starvation. Their fruiting bodies can vary dramatically among different species.
For example, some species form clear stalks that lift up the spores, whereas in others, the stalk is reduced or absent (Velicer and Vos 2009). The best-known species is Myxococcus xanthus, which has been used as a model system for cooperation and conflict (see chapter by [Velicer and colleagues]). The life cycle of M. xanthus is similar to that of social amoebae: the soil-dwelling bacteria prey upon other microbes, sometimes cooperatively as a swarm (Mauriello et al. 2010). Upon starvation, the bacteria aggregate into mounds and form fruiting bodies, where only a fraction of the cells differentiates into spores, and others either remain as rod-shaped cells or undergo autolysis ( Fig. 1F) (Nariya and Inouye 2008;Varon, Cohen, and Rosenberg 1984). The percentage of cells that become viable spores is much lower than in some of the other eukaryotic species discussed so far, with a non-spore percentage of up to 90% in M. xanthus, at least under laboratory conditions (Velicer and Vos 2009). The reason for the variation in fruiting body morphology among species, including in the formation of a stalk, is not well understood (Velicer and Vos 2009).
In its unicellular stage, it feeds on other ciliate species. Upon food shortage, however, it aggregates beneath the water surface and forms an aerial fruiting body. The stalk is produced via collective secretion of a mucous material by the entire population. The stalk lifts the population out of the water, after which each of the cells become encysted and together form a sorus (Sugimoto and Endoh 2006).

Fruiting Body Formation in the Amoebozoa
Within the Amoebozoa, the group historically known as Eumycetozoa (true slime molds) consisted of three major classes of organisms: Protostelids, Myxogastrids, and Dictyostelids.
The latter two groups are monophyletic, whereas molecular analyses indicate that protostelids are not. For this reason, they are now sometimes referred to as "protostelioid amoebae" rather than "protostelids", to emphasise common elements of their morphology in lieu of a phylogenetic classification (Shadwick et al. 2009). Protostelioid amoebae undergo sporocarpic development (Spiegel et al. 2017). The amoeba secretes an extracellular matrix, which forms a stalk that lifts the amoeba up. The amoeba then differentiates into a spore, sometimes following cell division (Furtado and Olive 1971;Spiegel et al. 2017;Lahr et al. 2011). Thus, sporocarpy in protostelids results in the production of microscopic fruiting bodies that contain only one or a few cells.
The plasmodial slime molds belong to the monophyletic group Mxyogastrea, another taxon of sporocarpic amoeba. Plasmodial slime molds are named for the slimy structure they produce-called a plasmodium-which forms when amoeboid cells undergo repeated rounds of mitosis without cytokinesis. This process results in a single, massive, multinucleated cell with a continuous cytoplasm, which can reach many meters in size. When conditions turn bad, the plasmodium produces masses of stalked fruiting bodies, a process that occurs not through growth, but through rearrangement of the existing biomass (Stephenson and Schnittler 2017). Thus, they form fruiting bodies, albeit not achieved through aggregation. Spores are mostly wind-dispersed from the fruiting bodies and germinate to form the plasmodium again. Plasmodial slime molds were the likely inspiration for the 1950's horror film The Blob. Like protostelids, the stalks produced by plasmodial slime molds are acellular (i.e., secreted). Unlike the protostelids, however, their fruiting bodies are macroscopic.
Finally, the dictyostelids consists of more than 160 species (Romeralo et al. 2011). Their phylogenetic tree contains many long, unbroken branches, which suggests that they have been undersampled and that the true diversity of the group is even greater (Romeralo et al. 2011). Although the phylogeny has been revised over the years, recent phylogenies based on SSU rRNA and alpha-tubulin sequences group dictyostelids into two major clades, the Dictyosteliales and the Acytosteliales, each of which is composed of two groups-resulting in groups 1-4, referred to below. These groups are then further subdivided (e.g., into groups 2A and 2B) (Romeralo et al. 2011). The model organism D. discoideum, discussed in detail below, belongs to group 4.
Comparative analyses indicate that formation of a stalked fruiting body is conserved within dictyostelids (Schaap et al. 2006;Sucgang et al. 2011;Heidel et al. 2011;Romeralo et al. 2011). Romeralo et al. (2013) combined genetic data from 99 species with the phenotypic data of 24 traits in each of these species. This work suggests that their last common ancestor (~0.6-1.0 billion years ago) formed fruiting body structures that lift spores up in the air. However, as we emphasise below, the species show substantial variation in the formation and appearance of their fruiting bodies (Fig. 1). The evolutionary drivers of such diversity in fruiting structures are still being investigated. Nevertheless, this diversity makes this group suitable to study the function of a stalk, its associated costs and benefits, and the possible functional constraints on this structure. (Bloomfield 2010;Bloomfield 2011; see also Schaap chapter). Meiosis takes place during the formation of the macrocyst, and the amoebae that later emerge are recombinants.
D. discoideum is a model system for cell biology, developmental biology, chemotaxis, and host-pathogen interactions (reviewed in Bozzaro 2019; Williams 2010). It is genetically tractable, has a precise 24-hour development cycle, and terminal differentiation results in a small number of distinct cell types. D. discoideum is also notable for its relatively stable celltype proportions: approximately 80% of cells in the posterior of the slug will form the sporecontaining sorus, whereas ~20% of cells in the anterior die to form the stalk. These cell-type proportions can partially re-establish following perturbations, for example, by ablation of either the anterior (prestalk) or posterior (prespore) sections of the slug (Ràfols et al. 2001;Raper 1940). The robustness of its spore-stalk cell proportions is of interest to developmental biologists interested in how multicellular organisms achieve and maintain specific cell-type proportions, as well as evolutionary biologists interested in whether and how an altruistic stalk can be maintained.
While D. discoideum is by far the best studied of the social amoebae species, it exhibits a variety of traits that are somewhat uncommon among dictyostelids. Below, we focus on three morphological traits, namely stalk composition, stalked migration and clustering and branching patterns, comparing D. discoideum to other species. Although it is difficult to ascertain the adaptive significance of this variation in morphology, we discuss functional implications of the different structures and some of their potential costs and benefits.

Cellular vs Acellular Stalks
Aggregative multicellularity-by virtue of allowing unrelated cells to collaborate to form a multicellular individual-presents opportunities for conflict, especially if there are different fitness costs and benefits associated with adoption of different cell fates. This problem is particularly severe in social amoebae, where cells that form the stalk will die and the remainder will live, providing a large fitness advantage to strains that can avoid the stalk fate. Conflict is thought to emerge over which cells will adopt the dead-cell fate and which will survive into the next generation. The opportunity for different genotypes to co-aggregate means that selection has the opportunity to favour genotypes that behave selfishly (Ostrowski 2019).
Stalk formation does not necessarily require self-sacrifice. The acytostelids (Group 2A), for example, form acellular stalks, consisting of a hollow tube that is made from secreted cellulose, with all cells subsequently forming viable spores atop the stalk (Mohri et al. 2013).
In contrast, cellular stalks are made from an inner layer of hardened vacuolized stalk cells and an outer layer of cellulose (Gezelius 1959). Why some species evolved to use cellular stalks, whose formation depends on the death of a fraction of the population, while others form stalked fruiting bodies without such a sacrifice, is not known.
At present, we can only speculate about why these differences might have evolved. One possibility is that cellular stalks might provide stronger support, allowing larger aggregate sizes and taller stalks that can support more spores. For example, acytostelids have a smaller aggregate size and form smaller structures compared to species with cellular stalk formation (170-1200 mm versus 1200-8200 mm), which might be consistent with a weaker stalk in the former (Raper 1956b;Schaap et al. 2006). Additionally, Kaushik and Nanjundiah (2003) point out that production of an acellular stalk might be energetically costly to the cells and therefore detract from their ability to survive for long periods. The division of labor achieved through formation of a cellular stalk might entail benefits for the survival of the spores.
One possible benefit of acellular stalk formation is the ability to produce a fruiting body with a smaller population. This potential benefit is apparent in Dictyostelium lacteum, the only species known to be capable of producing both cellular and acellular stalks. When food availability is low, the species forms a small, acellular stalk, similar to that produced by acytostelids; only at higher cell numbers is a larger, cellular stalk formed (Bonner and Dodd 1962;Bonner 2006). While D. lacteum is the only species known to be plastic for cellular stalk formation, it is possible that other dictyostelids possess similar plasticity but remain to be discovered, or that their plasticity has simply not been noticed. Finally, although some have speculated that acytostelids could represent an intermediate stage in the evolutionary transition from simple (e.g., single cell type, no division of labor) to complex (multiple cell types with division of labor) multicellularity (Olive 1975;Bonner 2003), a molecular study by Romeralo (2013) concluded that the most recent common ancestor of the dictyostelids likely already displayed cellular stalk formation, suggesting that acellular stalk formation is a derived trait.

Stalked migration
In some species, aggregation of cells is followed by the formation of a slug that migrates away from the point of aggregation (Fig. 2). For example, in D. discoideum, the slug forms approximately 12 hr after the onset of starvation, and it can travel long distances from the site of aggregation, resulting in movement of up to 6 cm in a week (Jack et al. 2015;Jack et al. 2011). In D. discoideum, the slugs are strongly phototactic, moving towards a directional light source. Migration ceases once light is overhead, which triggers culmination to form a fruiting body. The combination of attractants (light and heat) and repellents (high ammonia levels) is thought to direct slugs upwards through the soil, into an open area suitable for fruiting body formation (Raper 1984;Bonner and Lamont 2005).
Slug migration is thought to have evolved several times in the major groups of the dictyostelids (Romeralo et al. 2013). In D. discoideum, the stalk is not formed until after slug migration, during the final stages of development. However, in the majority of dictyostelid species, stalk is continuously produced from the rear of the slug during its migration (Fig.   2B). Ancestral trait reconstruction suggests that the last common ancestor of the bodies would likewise need to be greater. Indeed, while both D. discoideum and D.
purpureum possess mechanisms of kin discrimination, and thus imperfectly separate out during chimeric development (Mehdiabadi et al. 2006;Ostrowski et al. 2008), D. purpureum seems to segregate more completely than D. discoideum. It would be interesting to know whether D. purpureum has a stronger history of selection on the genes that underpin its kin discrimination. These genes have been identified in D. discoideum (Hirose et al. 2011), but are not yet known in D. purpureum.

Clustering and branching patterns
In D. discoideum, each aggregate gives rise to a fruiting body, which consists of a relatively thick, non-branched stalk that holds aloft a single sorus. This morphology is common among species in Group 4, but outside of this group there is a large variety of structures that differ in their degree of clustering and branching (Fig. 3). In some species, secondary tips form after aggregation, giving rise to multiple, closely spaced fruiting bodies (referred to as "gregarious" development) or fruiting body structures with multiple sori emanating from one stalk ("like flowers in a vase", Raper 1956). Fruiting bodies can also be branched and/or consist of whorls (Schaap et al. 2006;Baldauf and Strassmann 2017). The extent of the branching and clustering is also plastic, as it can depend on cell density (Bonner and Dodd 1962;Romeralo et al. 2013). The general pattern was that larger structures-i.e., branched and clusteredtend to form in response to high cell density, whereas unbranched and solitary structures would emerge at low cell density.

Altruism, Stalk Formation, and the Maintenance of Multicellularity in D. discoideum
In previous sections, we discussed numerous examples of aggregative multicellularity that result in the formation of a stalked fruiting body. We emphasized that, in many cases, many or all of the cells in the fruiting body remain viable, although there could be fitness costs associated with exactly which role is adopted by a given cell. We also emphasised that some species form fruiting bodies with secreted, non-cellular stalks, and others form extensive cellular stalks throughout migration, resulting in a potentially large and unpredictable cell sacrifice.
The formation of cellular stalks by dictyostelids is of special interest to evolutionary biologists. Stalk formation is likely altruistic, in that some give up opportunities for direct fitness in order to form a structure that appears to benefit the rest. This differentiation into spore and stalk cells is analogous to the differentiation into soma and germline that is seen in complex multicellularity. Stalk formation is a clear example of reproductive division of labor, where some cells specialise in reproduction, and others specialise in non-reproductive functions.
Certain features of the Dictyostelium life cycle mean that its reproductive division of labor might be evolutionarily fragile. Aggregative multicellularity potentially permits multiple different genotypes to co-aggregate and form chimeric multicellular structures. This genetic diversity in combination with strong fitness consequences for becoming stalk vs spore means that natural selection can operate during this stage of the life cycle. Thus, all else being equal (and it may not be), natural selection should favor genetic variants that can avoid the costly role of the stalk and disproportionately adopt the high-fitness spore fate. The problems posed by chimerism in Dictyostelium and its potential consequences for the evolution of multicellularity are discussed more fully in the chapter by Jahan et al. Given these opportunities for selection to favor stalk-avoiders, one long-standing question is the extent to which opportunities for conflict may have influenced how multicellularity evolves (e.g., whether aggregative multicellularity is successful) and whether it is evolutionarily stable. In Dictyostelium, this possibility has led to interest in whether stalkless forms might evolve in nature.
Under these conditions, well-spaced spores germinate and divide to produce circular plaques-clearings in the bacterial lawn where the amoebae have devoured the prey. At the center of each plaque, where the amoeba cell density is high and food has been depleted, fruiting bodies will form if the strain is capable of multicellularity. Thus, dilution plating of spores to see whether they give rise to fruiting bodies is a way to screen for the presence of strains that have lost multicellularity, yet previously managed to join and form a stalked fruiting body with others, or that exhibit other morphological alterations. However, despite screening >3,300 plaques, Gilbert and colleagues observed no stalkless morphologies.
To our knowledge, the works by Buss (1982) and Gilbert et al. (2007) are the only studies that have attempted to identify and/or quantify the frequency of stalkless strains in nature. In the future, as new methods for single-cell genomics improve, it might be possible to use culture-independent methods to isolate, sequence, and identify each amoeba cell in a soil sample, enabling identification of natural isolates that have lost multicellular development or evolved novel morphologies not currently recognized. For now, however, whether D. discoideum's aggregative fruiting can lead to selfishness that threatens the maintenance of the stalk in nature remains unknown.

Laboratory and Theoretical Studies of Stalklessness
The observation by Buss (1982) of a stalkless morphology motivated laboratory studies of stalk-avoiding mutants as well as theory to address the circumstances under which stalklessness might evolve and persist. For example, two studies examined an insertion mutant (fbxA-mutant) that contributes less to the stalk in chimerae with the wild-type strain (Nelson et al. 2000;Ennis et al. 2000). When developed clonally, the mutant forms aberrant fruiting bodies that contain few to no spores or fail to initiate stalk production altogether (Ennis et al. 2000;Gilbert et al. 2007). However, when co-developed with a wild-type strain, it produces a disproportionate fraction of the spores. Ennis and colleagues (2000) speculated that the fbxA gene takes part in the regulation of a complex involved in the differentiation in spore and stalk cells, where deletion of the gene causes the stalk cell differentiation pathway to be halted. Gilbert and colleagues (2007) subsequently used the fbxA-mutant to examine the extent to which such a strain that does not contribute fairly to the stalk might increase in frequency in a population owing to its advantage in spore production in chimerae. However, such a strain may face a disadvantage at high frequency, if it has displaced the very strain it relies on to sporulate. The impacts of these different frequencies can be quantified as relatedness, which encompasses the degree to which the mutant interacts with self (r = 1) or with the stalk-proficient wild-type (r = 0) to build a fruiting body. The authors found that when r > 0.75, the fbxA-mutant decreased in frequency, indicating net selection against the mutant. This work demonstrated that sufficiently high relatedness could be essential for preventing the invasion and takeover of populations by non-stalk-forming strains.
The above empirical examples suggest that stalk-avoiding strains, provided that the behavior is costly in the absence of a cooperating strain, could be selected against when relatedness is high-but is relatedness in natural populations sufficiently high to accomplish this?
Relatively little is known about relatedness in nature, especially over the small spatial scales in soil where different strains might encounter one another and co-develop to form chimeric fruiting bodies. Fortunato and colleagues genotyped natural isolates from minute soil samples collected using a plastic straw with a diameter of 6 mm (Fortunato et al. 2003). Of 26 soil samples that contained Dictyostelium, 63% yielded more than one genotype, with as many as 9 distinct genotypes from a single soil sample. These results yielded an estimate of average genetic relatedness of 0.52. In addition, Gilbert collected and genotyped 88 individual fruiting bodies that emerged from 25 dung piles incubated in the lab (Gilbert et al. 2007). Seventy-seven percent of the fruiting bodies contained only a single genotype, which yielded a minimum relatedness of r = 0.86-high enough to support their hypothesis that relatedness would be high enough to select against the fbxA-mutation in nature (i.e., 0.86>0.75).
The hypothesis that high genetic relatedness would be sufficient to stop the spread of nonstalked mutants was further supported by Kuzdzal-Fick et al. (2011). They carried out laboratory evolution experiments that involved multicellular development under either high or low relatedness conditions (i.e., in genetically diverse or clonal groups, respectively).
Approximately one-third of populations evolved at low relatedness harbored strains that could not form a stalked fruiting body when developed clonally but were disproportionately represented among the spores when co-developed with their ancestor. Conversely, no losses of multicellularity occurred in the high relatedness experiment. Taken together, these experiments collectively support that high relatedness is an important condition for the maintenance of the stalked fruiting body in this organism. Hudson et al. (2002) developed a mathematical model that sought to address the evolutionary stability of stalk formation. This model made two assumptions about how stalks influence fitness: (1) that the fitness of a stalkless strain would be low and (2) that dispersal success increases with stalk size. They showed that stalk formation could still be maintained in populations founded by genetically unrelated individuals. However, they also showed that the presence of selfish genotypes, which contribute less to the stalk, could drive the evolution of suboptimal stalk sizes-that is, reductions in the allocation to the stalk relative to what would be optimal, assuming that dispersal is an important fitness component. Similarly, a model by Brännström and Dieckmann (2005) supported the potential for coexistence of multiple genotypes (e.g., stalked and stalkless) within a single population. Taken together, these experiments suggest that one way to identify stalkless strains in nature might be to look for them under natural conditions where relatedness is low (i.e., genetic diversity is high) or the importance of dispersal is low. Moreover, while it remains unknown how essential stalk formation is for dispersal per se (as opposed to simply protecting the spores number, GFP reporter strains were used to estimate spore-stalk allocation. They observed variation among strains within both populations in stalk allocation, but also larger size overall for Texas strains compared to those from North Carolina. These results underscore that increases in stalk size can be accomplished through two routes: either by allocating a higher fraction of cells to the stalk or by forming larger fruiting bodies altogether, presumably through production of fewer, larger aggregates from the same starting cell number. In addition, after controlling for differences in overall fruiting body size, within-population variation in relative stalk size was observed in both populations, similar to the findings by Buttery et al. (2009). These studies together indicate that some polymorphism in clonal and chimeric spore-stalk allocation occurs among strains within a given site, but also that the morphology of the fruiting bodies can evolve divergently among locations.
Overall, the studies of altruistic stalk formation by D. discoideum, clonally and in chimerae, illustrate the potential for conflict in organisms that undergo aggregative multicellularity with division of labor. Aggregative multicellularity can lead to genetic diversity within the multicellular organism, providing the fuel for natural selection, whereas division of labor generates the strong competitive advantages to cells that avoid the altruistic fate. Studies in this organism thus help to validate the predictions of evolutionary theory about the problems of aggregative multicellularity.

Conclusions
Aggregative fruiting has independently evolved in five of the six supergroups of the Eukarya, suggesting that the formation of a stalk is a morphological adaptation. Nevertheless, there is a substantial diversity in how these structures are formed-whereas some stalks are composed of dead cells (and thus involve cell sacrifice), others form by secretion, such that all cells potentially survive as spores. Differences in how stalks are formed in organisms that undergo aggregative multicellularity has important implications for the evolutionary maintenance of this trait. Substantial variation in how stalked fruiting bodies form exists within dictyostelids alone, a large clade consisting of more than 150 species within the Amoebozoa. Here we described variation among species in the composition of the stalk (acellular versus cellular), the timing of its production (during or after migration), and its branching morphology (branched or unbranched, whorled or not). Unfortunately, the explanations for the variation in these features are not known, although several studies provide information about some of the potential functions of the stalk. For example, spores that sat atop stalks were more likely to be picked up by an insect vector (smith, Queller, and Strassman 2014), and stalks emanating from the rear of the slug can help in traversing gaps in the soil (Gilbert et al.

2012
)-yet, whether and how these features are used in nature remains to be seen. In addition, while some stalk features may provide a fitness advantage, there may also be functional constraints imposed by development or physics. For example, acellular stalks have the benefit of not necessitating cell sacrifice, but these structures may not support as many spores, or the spores may be of lower quality (Kaushik and Nanjundiah 2003).
In those species that form cellular stalks, death of a fraction of the cells presents opportunities for conflict, as strains that avoid forming the stalk and disproportionately form spores should have a fitness advantage. In the model organism D. discoideum, studies of stalk-avoiding mutants and the behaviors of natural isolates, as well as mathematical models have all contributed to our understanding of the evolutionary maintenance of altruistic stalk formation. These studies confirm an essential role for relatedness, but whether relatedness is high enough in nature to prevent takeover by stalk-avoiding strains remains uncertain.
Future studies would benefit from consideration of how the evolution of morphology has been impacted by variation in relatedness across populations, as well as of other factors that might also promote the evolutionary maintenance of cooperative multicellularity.     Raper (1984), except for P. violaceum, where the value listed is the mean stalk length from Romeralo et al. (2013). Note that fruiting body size can vary substantially depending on plating conditions, so only an approximate range is provided. Approximate fruiting body height is indicated below each species, based on descriptions in Raper (1984), except for P. violaceum, where the value listed is the mean stalk length from Romeralo et al. (2013). Note that fruiting body size can vary substantially depending on plating conditions, so only an approximate range is provided.