The social amoebae possess a sexual cycle that involves transient mutlicellularity: first a zygote attracts surrounding haploid amoebae to form a walled aggregate around it, and then cannibalizes these peripheral cells, eventually forming a dormant single-celled macrocyst. Self-fertile homothallic isolates occur as well as breeding groups of self-infertile heterothallic cells, which commonly have more than two mating types. The mating-type locus of the widely studied model organism Dictyostelium discoideum, which has three mating types, has recently been identified. Two of the three mating types are determined by single putative regulatory genes bearing no mutual similarity, while the third is specified by homologues of both of these genes. This is the first sex-determining locus of an Amoebozoan to be described and, since none of the key regulators show homology to known proteins, may be a first glimpse of a novel mode of regulation used in these organisms. The sexual cycle of dictyostelids has been relatively neglected, but continues to yield much interesting biology as well as having the potential to add to the genetic tools available for the study of these organisms.
Unicellular eukaryotes commonly couple their sexual cycles with a dormant resting phase. Many dinoflagellates (Figueroa et al. 2010), apicomplexans (Belli et al. 2006), algae (Cavalier-Smith 1976), cercozoans (Roepstorf et al. 1993) and amoebae (Mignot & Raikov 1992) form sexual structures described variously as cysts, oocysts, and zygospores. Their widespread occurrence suggests that similar structures may have been an early feature of eukaryotic sex (Cavalier-Smith 2002), and also that they are most likely in part a response to recurrent selective pressures. The social amoebae also possess a resting cyst during their sexual cycle, in this case called the macrocyst (Fig. 1; Blaskovics & Raper 1957). However, being facultatively multicellular organisms, these amoebae present some unusual characteristics when compared to more strictly unicellular protists.
As at the beginning of asexual development, the early stages of macrocyst production involve the collective movement of previously dispersed amoebae into tightly coherent cell masses (O’Day 1979). In this case though at the centre of each aggregation is a zygote, formed by the fusion of two compatible haploid amoebae (Szabo et al. 1982; Saga et al. 1983). The role of ethylene in cell fusion is reviewed elsewhere in this issue (Amagai 2011). The zygote secretes cyclic AMP to attract surrounding cells (Abe et al. 1984), as well as other factors to repress their fusion into competing diploids (O’Day et al. 1981), and it most likely also signals to the cells captured into its aggregate in order to orchestrate their subsequent behavior. The outer cells first cooperate to lay down an outer cellulose-containing protective wall around the entire aggregate (Blaskovics & Raper 1957). Meanwhile, the zygote begins ingesting the cells surrounding it, progressively eating its way out from the center until it fills the whole structure, retaining its partially digested cannibalized “prey” in vacuoles within its cytoplasm (Filosa & Dengler 1972). During this process, further walls are produced inside the first, primary wall (Blaskovics & Raper 1957; Erdos et al. 1972) This completes the macrocyst, which then remains apparently dormant for several weeks or more, with the only visible change being a slow, progressive degradation of the ingested amoebae into small granules (Blaskovics & Raper 1957). During this quiescent period meiosis is also thought to occur, perhaps in the first days after cyst formation (Erdos et al. 1972; Okada et al. 1986), but whether meiosis proceeds uninterrupted or if it arrests at any stage is not known. The progeny divide multiple times by mitosis and finally break out of the cyst to disperse or resume growth if conditions allow (Nickerson & Raper 1973b).
A number of studies have begun to characterize the molecular events underlying the early events of the Dictyostelium sexual cycle, and have been discussed elsewhere (Urushihara & Muramoto 2006). This review focuses on the genetics of sex determination, which remains very poorly understood in many protists, and until recently was an almost-complete unknown in the dictyostelids.
Homothallism and heterothallism
The first studies of macrocysts involved species that possess self-fertile homothallic isolates. It was not until ultrastructural studies revealed in detail the early stages of the cycle that it was suspected that macrocysts are sexual structures (Erdos et al. 1972). Subsequently, pairings of self-infertile isolates revealed the presence of sexes, hereafter called “mating types” (Clark et al. 1973; Erdos et al. 1973). Each mating type within a species is compatible to produce macrocysts in pairings with each of the others, but not with itself. These isolates are called heterothallic. Different species have different numbers of mating-types, and the social amoebae are unusual in commonly having more than two sexes (Hurst & Hamilton 1992; Hurst 1996). The species in which macrocysts have been observed are listed in Table 1, and the presence of heterothallic and homothallic isolates recorded in each case.
Table 1. Dictyostelid species known to produce macrocysts
Each species for which macrocysts have been reported is listed and, where known, whether only homothallic (hom), only heterothallic (het), or both kinds of isolates are found. A question mark is given if there is evidence of macrocyst formation but further details are required. The taxonomy of dictyostelids is difficult because some species are very similar in morphology, and several species named in this table may be composed of multiple syngens (groups that can interbreed only among themselves) that would be better categorized as separate species, including P. violaceum (Clark et al. 1973) and D. purpureum (Hagiwara et al. 2005). There are also reasons to suppose that heterothallic and homothallic isolates of a given species may also represent separate breeding groups. It is not clear whether the homothallic “P. pallidum” isolate PP28S studied by Robson & Williams (1980) is P. pallidum or P. album (or another Polysphondylium species) according to the revised taxonomy of Kawakami & Hagiwara (2008a). D. brefeldianum and D. mucoroides are perhaps best treated as synonymous (Romeralo et al. 2010), and the sexual behavior of the morphologically very similar D. sphaerocephalum should be further examined. The case of D. polycephalum is unclear both because no specific citation is given by Raper (1984), and also since it is not certain that the report by Reddy et al. (2010) shows evidence of macrocysts as well as microcysts; clearly, however, this species needs further investigation.
One of the first reports of mating types of Dictyostelium discoideum noted the presence of three apparent mating types (Erdos et al. 1973). However, most subsequent studies have treated the “third” sex as a special class, known either as “bisexual” or “ambisexual.” This class was believed to represent abnormal clones possessing both of the other presumptive alleles of the mating-type locus (Robson & Williams 1980). We and others have more recently argued that they should be treated as a third true sex (Urushihara & Muramoto 2006; Bloomfield et al. 2010). Molecular evidence supports this view, and the three mating types show equivalent abilities to fuse with each other and to prevent self-fusion. We propose to name the mating types I, II, and III, reverting to and extending the earliest nomenclature for this species (Clark et al. 1973).
Sexual genetics of dictyostelids
Early genetic studies gave evidence of meiosis in the homothallic Dictyostelium mucoroides (Macinnes & Francis 1974) and the heterothallic Dictyostelium giganteum (Erdos et al. 1975), but most subsequent work has been on the widely used model organism, D. discoideum. Much effort was expended in trying to establish sexual genetics in this species, with frustratingly little success (Wallace & Raper 1979). As well as the long periods required for macrocysts to mature (several weeks in most cases), germination rates were, with few exceptions, extremely low and unpredictable. The development of molecular genetic techniques shortly afterwards proved to be much more fruitful, so the sexual cycle was almost entirely abandoned as an experimental tool. Although this work failed in its main objective, it provided the first, and until recently the only, characterization of the genetic basis of sex determination in this organism. The mating behaviors of progeny from sexual as well as parasexual crosses (where cells of the same mating-type are fused to produce vegetative diploids, not macrocysts) were consistent with the existence of a single genetic locus with two or more alleles that each stably determined one of the mating-types (Robson & Williams 1979; Wallace & Raper 1979). It was further suggested that one of the three D. discoideum mating-types, and homothallic isolates of this species, might possess the two alleles of this locus from the other two mating-types (Robson & Williams 1980). Later work found different antigens on the surface of the different mating-types under conditions that promote the initial cell fusion to form zygotes (Urushihara et al. 1988; Aiba et al. 1992). Furthermore, limited protease treatment of one mating-type results in the removal of a block to self-fusion (Urushihara & Aiba 1996). Clearly cell surface properties are important in permitting or excluding cell interaction and fusion; however, the fundamental determinant of the differences between the sexes remained elusive.
Discovery of the mating type locus of D. discoideum
Recent work using a comparative genomic hybridization approach succeeded in identifying the mating-type locus of D. discoideum (Bloomfield et al. 2010). Using microarrays designed from the already sequenced genome of this species (the strain sequenced was AX4, which is mating type I), the genomic DNA of 10 isolates of types I and II were compared. Of more than 8000 genes a single candidate was present in all type I strains but not in any of the type II strains tested. This is the expected result if the two mating types are specified by sequences that are substantially diverged or not homologous. If the two sequences were closely related or if, as in the yeasts, multiple versions are present in each genome (Hicks et al. 1979; Egel & Gutz 1981), the strategy would have failed. The candidate locus was cloned from representative type II and III strains, and found to differ both in length and gene content between them (Fig. 2). In the type I genome, the locus contains a single small gene, matA, flanked by two genes that do not vary between mating types. The type III version of this locus is longer and completely different in sequence: it contains two genes, matS and matT, which are not homologous to each other, nor to matA. The type II version is larger still: it contains a homologue of matA, called matB, situated between homologues of matS and matT, named matC and matD, respectively.
Genetic dissection of the different versions of the mating-type locus demonstrated that in types I and III a single gene is responsible for determining the mating orientation of haploid cells. First, deletion of matA from a type I strain was found to result in a block to macrocyst formation with both the other mating types. Reintroduction of the matA coding sequence restored mating ability. The introduction of only matS into this null mutant strain was sufficient to reorient it to mate with type I and type II cells, but not with type III. It was necessary to express both matB and matC in the null strain to recapitulate the type II phenotype. Strains possessing matB alone are able to mate only with a type III partner strain, while strains expressing only matC can mate with type I but not type III (nor type II). The two other mat genes, matD and matT, are not required for mating-type determination. Figure 3 shows a summary of these results, which overall show that matA is sufficient to specify mating type I, matS alone specifies type III, but matB and matC together are necessary to give the type II phenotype. For a productive cross to take place, a matA-type and a matS-type gene must be present in the two prospective partner strains. However, matB and matC are not compatible despite being homologues of matA and matS, respectively: the strains expressing singly matB and matC do not form macrocysts when paired with each other, and the strains expressing both matB and matC are not self-fertile.
The composite nature of mating type II
Mating type II thus appears to be the result of an anomalous fusion between type I and type III ancestors. It is possible that it originated in an abortive cross between cells with at least one mutation making them not fully compatible; after nuclear fusion, one can speculate that recombination between the two versions of the mating-type locus resulted in an insertion of the ancestral matB sequence between the ancestral matC and matD. If there are roles for these genes after cell fusion, a parsimonious explanation would be that the incipient matB and matC were already incompatible in the same way as they are in extant type II cells, resulting in a diploid arrested at an early stage of the sexual cycle. Subsequent haploidization by random loss of chromosomes, as happens during the parasexual cycle (Brody & Williams 1974), could then have resulted in a haploid cell resembling the extant type II isolates. However this class of cell arose, the functional result is that type II cells are equivalent to self-infertile hermaphrodites.
It should be noted that there have been reports of type II cells “selfing,” forming macrocysts without fusing with cells of another mating type, although after being primed either by a small number of type I cells (Bozzone & Bonner 1982), or by nearby type I cells physically prevented from contacting them (MacHac & Bonner 1975; O’Day & Lewis 1975; Lewis & O’Day 1977); this effect is possibly mediated by ethylene (Amagai 2011). We have also generated a mutant containing the type II version of the locus that under some conditions produces abnormal cyst-like structures when plated in isolation (Bloomfield et al. 2010). These are regarded as exceptions, because under standard assays type II cells are not found to be self-compatible in the same way as truly homothallic isolates. Nevertheless, this phenomenon might reflect an interaction between matB and matC that is only obtained under a limited set of conditions.
The nature of homothallism
Homothallic dictyostelids appear to be truly autogamous (proceeding through the sexual cycle in the same way as heterothallic cells, with fusion followed ultimately by meiosis), given the evidence of high-frequency fusion (Amagai 1989; Urushihara et al. 1990) and physical and genetic evidence of meiosis in some species (Erdos et al. 1972; Macinnes & Francis 1974). The mating-type locus has also been characterized in two homothallic isolates (Bloomfield et al. 2010). The status of these isolates as D. discoideum has been questioned (Briscoe et al. 1987; Evans et al. 1988), but since the more diverged isolate, AC4, groups very closely with this species in ribosomal small subunit (SSU) trees (Schaap et al. 2006) we continue to associate both isolates with this species. Both isolates show the same pattern as type III heterothallic strains, with genes allelic to matS and matT, although diverged (77% and 92%, respectively, in AC4). The similarity of the homothallic and type III versions of the locus is striking, and the molecular and evolutionary reasons for this relationship remain unclear. We are currently investigating the genetic basis for homothallism: in principle it could be caused by changes within the mating locus or at one or more loci elsewhere in the genome. Further work is also needed to establish whether homothallism or heterothallism is ancestral in D. discoideum (and in other species). Both mating strategies can be advantageous: heterothallism enforces a degree of out-breeding, while homothallism permits formation of the protective cyst structure without the need for a compatible partner to be present.
Dictyostelium rosarium has three known mating types, and so might follow the genetic pattern seen in D. discoideum. Two species, D. giganteum and D. purpureum, possess more than three mating types (Erdos et al. 1975; Hagiwara et al. 2004, 2005; Mehdiabadi et al. 2009, 2010), and the genetic basis of these expansions to the number of sexes will be of interest. Clear homologues of matC/matS and matD/matT are present in the genome of the sequenced D. purpureum clone, and Acytostelium subglobosum possesses a definite matD/matT homologue and a small adjacent gene very weakly similar to matC/matS. Both species therefore appear to follow the type III (and homothallic) pattern. The sequenced isolate of D. purpureum is heterothallic (Mehdiabadi et al. 2009), but the mating behavior of the sequenced A. subglobosum strain is not known. No homologues of the mat genes have been discerned in the Polysphondylium album or Dictystelium fasciculatum genomes. The sequenced P. album isolate is known to be heterothallic (Kawakami & Hagiwara 2008a), and might thus be the equivalent of type I with a short matA/matB homologue that is too divergent to be recognizable with our current tools; alternatively this species might use an entirely different system. Macrocysts have not been observed in D. fasciculatum, so its status is unknown. The degree of homology of the matC/matS-related sequences, albeit from a very small sample, suggests they may be relatively rapidly evolving, which is a widespread property of genes involved in sex (Swanson & Vacquier 2002). It is possible that homologous sequences are present but undetectable in the genomes of other Amoebozoa.
Function and localization of mat proteins?
The key sex determining genes, matA and matS, have no homology to known proteins except matB and matC. Their polypeptide sequences do not contain any predicted domains or informative motifs, and their functions remain unknown. They encode small, relatively hydrophilic proteins, and appear to be cytosolic when tagged and overexpressed (G. Bloomfield, unpubl. data, 2010). The MatD and MatT proteins are larger, and possess predicted signal peptides and glycophosphatidylinositol-anchor attachment sites, implying that they are associated with the outer leaflet of the plasma membrane (or internal membranous compartment). Furthermore, they are weakly similar in sequence to the HAP2/GCS1 family of membrane proteins, which have a conserved role in gamete fusion (Wong & Johnson 2010), to another conserved dictyostelid putative integral membrane protein (MrhA), and to three Naegleria gruberi predicted proteins (the N-terminal regions of XP_002675063, XP_002677524, and XP_002682216). Naegleria is a free-living Heterolobosean not closely related to the Amoebozoa, and it is possible that sequencing of other protists that are not obligatory parasites will reveal further membrane proteins distantly related to HAP2 and MatD. The short HAP2/GCS1 domain is also weakly conserved, with three cysteine residues invariant among HAP2 homologues all retained in every MatD/MatT sequence. This is suggestive of a role for MatD and MatT during the fusion process; although they are not necessary for sex determination, the presence of matD or matT in quantitative crosses tends to result in higher yields of macrocysts (Bloomfield et al. 2010).
The sequenced dictyostelids also each have a distinct HAP2 homologue elsewhere in the genome, which is much more closely related to the canonical HAP2/GCS1 proteins. The mRNA of the D. discoideum gene hapA is enriched in cells competent for mating (Muramoto et al. 2003), so it is likely to have retained a function in gamete fusion. In plants, algae, and Plasmodium, the function of HAP2 is sex-specific, required in pollen (von Besser et al. 2006; Mori et al. 2006), the minus mating type in Chlamydomonas (Liu et al. 2008), and male gametocytes in Plasmodium (Hirai et al. 2008; Liu et al. 2008). It is possible that dictyostelids expanded the existing HAP2 system, giving rise to matD and matT (as well as perhaps mrhA), as part of the process that enabled more than two sexes to occur in an ancestral species.
Comparison with other sex-determining systems
In ascomycete fungi, sex determination systems often consist of a pair of relatively short, dissimilar sequences encoding unrelated proteins. In these cases, the two forms of the locus are described as idiomorphs rather than alleles (Metzenberg & Glass 1990). The D. discoideum system conforms to this pattern, albeit with the complication of the “composite” third sex. In the unicellular algae this pattern is followed on a larger scale, with expanded regions again containing different genes (Goodenough et al. 2007). In ciliates, which can also accommodate more than two mating types, different kinds of systems appear to operate (Phadke & Zufall 2009). In Euplotes raikovi, a single polymorphic gene distinguishes between mating types, acting both as a growth factor and mediator of mating interactions (Luporini et al. 2005). In Tetrahymena pyriformis (Nanney et al. 1955), different alleles of an unknown system shift the probability distribution of which of seven possible mating-types will be expressed. In basidiomycetes (Raper et al. 1958; Casselton 2002) and the plasmodial slime mold Physarum (Kawano et al. 1987), two or more different loci result in multiple mating types (or “incompatibility types”), so again contrast with the case of D. discoideum, which requires only one locus. Evidently many different modes of regulation have evolved in different organisms to fulfill the centrally important function of dividing their species into mutually compatible groups. To our knowledge, the organizational pattern used by D. discoideum to specify its three sexes is unique.
The discovery of the D. discoideum mating-type locus should open several lines of analysis. The lack of homology of the key sex-determining genes to other known proteins is of interest in itself, and investigation into their mode of action may lead to further insight into the machinery of cell–cell recognition and fusion during the sexual cycle as well as potentially novel mechanisms that they may use in their direct functions. It also seems likely that the mating-type locus plays a role after fusion because genetic evidence implicates it in a block to mitotic growth of sexual diploids (Robson & Williams 1979; G. Bloomfield, unpubl. data, 2010). This, and the ability to engineer strains differing only at the mating-type locus may help to address some of these questions, and might also lead to the development of generally useful genetic tools. Research into broader issues related to the sexual cycle might also benefit from this progress: the coercive nature of macrocyst formation, with the zygote’s progeny surviving at the expense of cannibalized peripheral cells raises similar questions regarding altruism and the potential for “cheating” as the sacrifice of the cells that form the stalk in the asexual fruiting bodies of Dictyostelium and Polysphondylium (Shaulsky & Kessin 2007; Li & Purugganan 2011). Finally, investigation into the environmental regulation of the earliest-induced sexual genes may also shed light on the ecological context and significance of the dictyostelid sexual cycle, which remain almost completely mysterious.
I would like to thank the Medical Research Council for support, and Rob Kay for his interest in and encouragement of my research. Grateful thanks are also due to Yoshimasa Tanaka, James Cavender, David Traynor, Francisco Velazquez-Duarte, Roberto Zanchi, John Nichols, Louise Fets, Evgeny Zatulovskiy, and Marwah Hassan for advice and discussions. Comparisons of the D. discoideum sequences with those from other dictyostelids would not have been possible without the generous release of genome sequence data prior to publication by the US Department of Energy Joint Genome Institute (http://www.jgi.doe.gov/) and co-workers, the Actyostelium Genome Consortium, and the Franz Lipmann Institute and co-workers. The curators at dictyBase have also provided invaluable help during the course of our work.