The molecular foundations of zygosis

Abstract

Zygosis is the generation of new biological individuals by the sexual fusion of gamete cells. Our current understanding of eukaryotic phylogeny indicates that sex is ancestral to all extant eukaryotes. Although sexual development is extremely diverse, common molecular elements have been retained. HAP2-GCS1, a protein that promotes the fusion of gamete cell membranes that is related in structure to certain viral fusogens, is conserved in many eukaryotic lineages, even though gametes vary considerably in form and behaviour between species. Similarly, although zygotes have dramatically different forms and fates in different organisms, diverse eukaryotes share a common developmental programme in which homeodomain-containing transcription factors play a central role. These common mechanistic elements suggest possible common evolutionary histories that, if correct, would have profound implications for our understanding of eukaryogenesis.

Introduction

Sex is a cyclical process that produces new eukaryotic individuals in two ways. In one phase of sex, zygosis (also known as syngamy and amphimixis [1]), two individuals are combined: two cells fuse, and then their nuclei fuse, forming a new individual with a doubled genetic content. In the other phase, meiosis, cells of this higher genetic content are divided to produce cells that have reverted to the lower ploidy level. At its simplest, this sexual cycle is restricted to single-celled individuals, but it always involves another, asexual replicative mode of individuation, in which cellular ploidy is changed only insofar as a new genome copy is produced before each division, so that each daughter cell reproduces its parent. However, even the simplest sexual cycles necessitate an elaborate sequence of events that must be carefully coordinated. Sex was ancestral to all extant eukaryotes (as far as is known), and understanding its conserved molecular foundations can help to shed light on some of the central structural and regulatory requirements that shaped the earliest eukaryotic cells, which must have been decisive in shaping the tremendous diversity of eukaryotic organisms that have evolved as their descendants [2–5].

Zygosis involves the pairing and fusion of gametes (Fig. 1, left), which are normally monoploid (possessing one genome copy), forming a prozygote cell [6]. The two nuclei then fuse to form the zygote (Fig. 1, right). The zygote then progresses to meiosis without dividing in haplontic organisms, or enters the mitotic cell cycle in diplontic and haplo-diplontic eukaryotes. Variations on this basic pattern of behaviour are seen in ciliates and certain fungi, in which cell fusions are transient and nuclei are exchanged without mixture of cytoplasms [7, 8]; and in social amoebae, in which prozygotes have a transient syncytial stage, where many gametes fuse simultaneously, mixing their cytoplasms thoroughly then dividing gradually to uni- and binucleate cells before nuclear fusion occurs [9–11]. Despite these elaborations of the basic pattern and the diversity of sexual cycles in general, recent research has illustrated common mechanisms that are widely conserved in very different eukaryotes and that, therefore, very likely reflect ancestral mechanisms governing zygosis. This review will highlight certain of these recent advances, mostly in unicellular eukaryotes, and discuss how they might deepen, and even transform our understanding of fundamental aspects of eukaryotic cell biology and evolution.

Fig. 1.

The rudiments of zygosis. Zygosis involves the fusion of two gametes (shown here immediately after the initial fusion of their membranes), followed by nuclear fusion of their nuclei to form a zygote, a new individual with twice the original genetic content (2n). Zygotes can either enter the mitotic cell cycle (in diplontic and haplo-diplontic organisms) or progress immediately to meiosis to regenerate haploid progeny (in haplontic organisms). Typically, but not always, gametes are monoploid and zygotes diploid

Conserved mechanisms of gamete fusion

Sex is advantageous and involves multiple cells, and so is a competitive process. Gametes might be slow or otherwise inefficient in pairing and fusing with compatible partners, and many will fail to fuse at all. Genetic variants that affect gamete fusion can, therefore, be expected to be under strong selection. Relatedly, we might also expect that gamete recognition genes might be relatively fast-evolving: positive selection on these genes could lead to frequent fixation of effective new variants. Variants that introduce new proteins into the recognition and fusion processes to increase the success rate of gamete fusion can also be selected for, which might promote the loss of ancestral genes if those can be completely replaced. Accordingly, gamete fusion and recognition mechanisms vary considerably between different organisms, even otherwise closely related species [12].

The most highly conserved gamete fusion gene, as far as is known, encodes the fusogen protein HAP2-GCS1. This gene is present in at least some species in almost all major eukaryotic lineages, though it appears to have been lost multiple times, notably being absent in the entire vertebrates and fungal lineages [5, 13]. HAP2-GCS1 is structurally related to viral class II fusion proteins [14–16]; these proteins, together classed as fusexins [14, 17], undergo multimerisation and substantial conformational changes to bring lipid bilayers together, and promote membrane fusion [18]. In the case of the viral fusogens, one of the two membranes is the envelope bilayer surrounding the infective virion, in which the fusogen protein is embedded, and the other is the endolysosomal membrane within target cells. Fusion of the membranes in this case allows the virion and its contents to enter the host cytoplasm. The viral fusexins are often activated by acidification of the endosome lumen after endocytosis [19, 20]. In the case of HAP2-GCS1, the two membranes to be fused are the plasma membranes of two gametes. In this fusexin, a hydrophobic or amphipathic polypeptide region comprising one or more loops at one extremity of the fusion structure inserts into the target membrane during the initial contact, allowing the subsequent merger of both membranes [21]. Substantial variation in the sequence and structure of the HAP2-GCS1 fusion loops exists between lineages [22, 23], presumably reflecting differences in target membrane composition combined with the long-term effects of sexual competition. Variable regions away from the fusion interface likely reflect different protein–protein interactions undergone by HAP2-GCS1 during its activation cycle [23].

HAP2-GCS1 function has been characterised in several organisms. In plants, where the gene encoding the protein was first discovered, HAP2-GCS1 is expressed in male gametes and is necessary for fusion with the female gamete [24, 25]. This male-specific requirement for HAP2-GCS1 has also been demonstrated in the apicomplexan Plasmodium [26, 27], and seems likely to be conserved in metazoa, as well [28, 29]. In the isogamous green alga Chlamydomonas, the HAP2-GCS1 orthologue is again only required to function in the minus gamete, not the plus gamete [26]. In contrast, in the ciliate Tetrahymena, which has several mating types, HAP2-GCS1 is required in both of the paired gametes [30], in a deviation from the “virus-like” unidirectional function that occurs in plants, green algae, and others. The social amoeba Dictyostelium discoideum appears, puzzlingly, to be an intermediate case: this species has three mating types, and two of them express HAP2-GCS1 and require its function during fusion, but, in the third mating type, the protein is not necessary for fusion [31]. Although biochemical data are required to confirm it, this finding suggests that, in two of the three D. discoideum mating configurations, HAP2-GCS1 function is ‘conventionally’ unidirectional, while, in the third, it is bidirectional as in Tetrahymena.

The trigger of the fusogen activity of HAP2-GCS1 is not known and might vary between different organisms. In the flowering plant, Arabidopsis membrane localisation of the protein is a regulated step: HAP2-GCS1 is only delivered to the plasma membrane of male gametes after stimulation by proteins secreted by the female gamete [32]. In contrast, in the green alga, Chlamydomonas HAP2-GCS1 is constitutively present in a small region of the differentiated minus gamete plasma membrane [33]. The cytoplasmic C-terminal domain is important for the function of the protein: a cluster of cysteine residues therein, often found but not always in corresponding positions in different HAP2-GCS1 orthologues, is important for fusion in Chlamydomonas, and another mutation affecting the C-terminal domain interferes with targeting of the protein to the fusion site [33]. In Arabidopsis, positively charged residues in the C-terminal domain are required for efficient fusion [34] (but see also [35]).

As noted above, several eukaryotic lineages appear to have lost the HAP2-GCS1 gene, very likely after its function in gamete function was made redundant after the emergence of novel fusion-promoting mechanisms. In fungi, the membrane proteins Prm1 and Fig1 have been implicated as important regulators of fusion [36–40], but the precise mechanism of membrane fusion in this lineage remains elusive. In vertebrates, mechanisms of gamete recognition, and perhaps fusion, appear to be diverse, and perhaps fast-evolving [41–43]. It seems likely that further diversity will be discovered when other sexual lineages that lack HAP2-GCS1 homologues are examined.

Prevention of fusion—how do gametes fuse in twos, not threes, and more?

Sex is a cycle of doubling then halving the ploidy level, and so must normally involve fusions of pairs of gametes to form zygotes: uncontrolled ploidy increases through fusion of multiple cells are not commonly found (most likely because of costs to polyploidy [44]). If gametes are rare, this pairwise fusion will occur almost automatically, but since sex is competitive (as mentioned above), in many organisms, there is scope for multiple fusions to occur. Consequently, mechanisms have evolved to promote biparental sex, and triploid zygotes are rare, aberrant occurrences [45–47]. Remarkably, in Chlamydomonas, the HAP2-GCS1 protein is rapidly degraded after gamete fusion, along with another membrane protein called FUS1 [45]. The destruction of HAP2-GCS1 prevents fusion of the prozygote with plus gametes, and destruction of FUS1, which is expressed in plus gametes and important for gamete recognition [48], prevents fusion with minus gametes [45]. A recent study in the ascomycete yeast Schizosaccharomyces pombe demonstrated a rapid post-fusion transcriptional response, involving a homeodomain transcription factor in limiting the potential for polyploidy [49]; a pair of homeoproteins has also been implicated in preventing supernumerary fusion in the basidiomycete fungus Cryptococcus [50], and more generally, the frequent involvement of related transcription factors during zygosis (see below) suggests that this might be a common function in evolution.

In animals, the fusion of multiple sperm cells with a single egg occurs frequently in some species, but not others [51, 52]; the physical properties of the egg can alter after the initial fusion, limiting further entry, or degradation of supernumerary sperm-derived nuclei or pronuclei can occur in the egg cytoplasm, depending on the species [51]. Zygosis in flowering plants occurs through an orchestrated series of events including the repulsion of supernumerary pollen tubes from each ovule, as well as blocks to polyspermy; mechanisms underlying the former ‘polytubey’ are known [53, 54], but those limiting polyploidy remain unclear [55]. Finally, as noted above, social amoebae are unusual in forming frequent syncytia before gamete nuclei fuse [9–11]. Puzzlingly, haploid nuclei appear to coexist in these syncytia for up to 8 h as the syncytia gradually break apart to form binucleate cells, in which nuclear fusion occurs [10, 56]. How nuclei of different mating types might recognise each other in these cells, and how fusion between them might be controlled, are not known. Nuclear fusion during zygosis is another complex process, involving fusion of outer and inner nuclear membranes, along with their associated endoplasmic reticulum. These fusion events involve the KAR5-GEX1 protein in budding yeast, Chlamydomonas, and zebrafish, and its wide conservation across eukaryotes suggests that, like HAP2-GCS1, KAR5-GEX1 is ancestral to all eukaryotes [40, 57–59]. Complex regulation of nuclear fusion is important in certain fungi as well (apparently) in social amoebae [8, 11, 40]; it is conceivable that nuclear fusion was more complex ancestrally than in most extant organisms if, as some have hypothesised [60], early eukaryotes were multinucleate.

Zygote differentiation—a conserved ancestral function for homeoproteins

Across eukaryotes, zygotes have diverse features and fates: as mentioned earlier, they can immediately commit to meiosis or can enter a prolonged diploid (or polyploid) phase of the organism’s lifecycle, depending on the species. Whether this involves the entry into the diploid (or polyploid) mitotic cell cycle, or an immediate commitment to meiotic division, zygote-specific genes must be induced and then further changes in gene expression must occur after fusion of the gamete nuclei. Despite this diversity of zygotic fates in different lineages, it has become clear that a common element in the initial transition from the haploid to the diploid phase is present in several eukaryotic supergroups: an involvement of homeodomain-containing transcription factors [61].

The genetic control of the haploid to diploid transition was first dissected in detail in the budding yeast Saccharomyces cerevisiae, in which proteins encoded at the mating-type locus were found to govern both haploid and diploid functions [62]. Two of these proteins, MATa1 and MATα2, contain homeodomains [63]. The genes encoding these proteins form part of different idiomorphs of the mating-type locus (that is, part of the different versions of the locus that determines mating type of haploid budding yeast cells), so they function independently in haploids. Upon gamete fusion, MATa1 and MATα2 bind each other to form a heterodimer that functions as a diploid-specific transcription factor [64]. As described above, a similar heterodimerisation involving a homeoprotein occurs at the same lifecycle stage in the distantly related ascomycete Schizosaccharomyces pombe; in this case, a homeoprotein named Pi binds a much shorter polypeptide, Mi [49, 65]; Mi is not recognisably homologous to homeodomain proteins, and its tertiary structure is not known, but it seems possible that the gene encoding it is a degenerate and truncated descendant of an ancestral homeodomain gene. In basidiomycete fungi, a similar process occurs to that founding in budding yeast: homeoproteins encoded by genes within different alleles of one mating-type locus (basidiomycetes typically have two such loci) are able to heterodimerise after cell fusion and activate dikaryon functions [66, 67].

Homeoproteins, which can be divided into two classes, TALE (three amino acid loop extension) and non-TALE [68, 69], also have important functions in establishing zygotic identity in plants and green algae. In Chlamydomonas, gametes fuse in a HAP2-GCS1-dependent process as described above, and then, zygotes differentiate into zygospores (Fig. 2, top). Zygotes do not divide by mitosis: sex in these algae is induced by nitrogen starvation, and zygospores can remain dormant for a period before undergoing meiotic divisions [70]. Like budding yeast and basidiomycetes, Chlamydomonas gametes of each mating type express a specific homeodomain transcription factor, one of the BELL and one of the KNOX class, although the genes encoding them are not located at the mating-type locus [71, 72]. Again, upon gamete fusion, the two homeoproteins, called GSM1 and GSP1, bind to each other to form a dimer that can only form in diploids heterozygous at the mating-type locus, and then activate zygotic gene expression [72]. Mutants lacking GSP1 fail to form zygospores after gamete fusion, instead remaining in the mitotic cell cycle (Fig. 2, bottom) [73]. The kinetics of degradation of HAP2-GCS1 and FUS1 in Chlamydomonas zygotes (mentioned above) are consistent with a process activated by fresh transcription after fusion, at least in part [45]. Remarkably, although land plants have dramatically elaborated lifecycles compared to their green algal ancestors, this ancestral function of homeoproteins in regulating zygote differentiation has been conserved in bryophytes. Physcomitrella mutants lacking two KNOX homeoproteins normally expressed in egg cells develop abnormally after zygosis, forming gametophyte-like (haploid-like) morphologies instead of diploid sporophytes [74]. Overexpressing a different, BELL class, homeoprotein in haploid cells of the same species produces sporophyte-like differentiation [75].

Fig. 2.

a Zygosis in Chlamydomonas. Two gametes, one of the plus mating type (upper) and one of the minus mating type (lower) fuse, ultimately forming a zygospore. bGSP1 mutants fail to complete zygosis. After fusion with a minus gamete, GSP1 mutant cells fail to limit mitochondrial and chloroplast inheritance in the normal fashion and do not differentiate into zygospores, resulting in vegetative diploid cells containing organelle genomes from both parents

The brown alga Ectocarpus, a member of the eukaryotic supergroup Heterokonta (also known as Stramenopiles), has a complex lifecycle involving multicellular haploid and diploid phases [76]. As in the very distantly related bryophytes and green algae, mutations affecting homeoproteins cause defects during zygosis: homozygous mutants in either the oro or sam genes cause diploid organisms to develop similarly to gametophytes (haploids) instead of sporophytes [77, 78]. The two homeoproteins heterodimerise, are TALE class like those with similar functions in plants and green algae, underscoring the conserved pattern of behaviours, although both proteins are expressed in both gamete classes unlike their fungal and green algal counterparts [78].

Finally, in the social amoeba Dictyostelium, like the fungi, genes located within the mating-type locus are essential for normal zygote development as well as mating [79, 80]. One pair of gametolgues in Dictyostelium discoideum is necessary for zygotic function in diploids formed from two of this species three mating types, and encodes proteins that have a homeodomain-like fold, but extremely divergent in sequence [80]. Their evolutionary history is not clear, nor their biochemical interactions, but they may also share ancestry with the homeoproteins with roles in zygosis in other eukaryotic lineages.

Inheritance of mitochondria and plastids during zygosis

In many cases, mitochondria and chloroplasts are inherited uniparentally during sex, usually through females in both animals and plants [81, 82]. Even in unicellular eukaryotes with gametes indistinguishable in size, more or less strict uniparental inheritance of these organelles is maintained. For instance, in Chlamydomonas, mitochondrial genomes are inherited only from the minus parent, while chloroplast genomes are inherited from the plus parent [83, 84], in a process depend on the aforementioned homeoprotein GSP1 (Fig. 2) [73]. Similarly, in the basidiomycete yeast Cryptococcus, two mating-type-specific homeodomain proteins are required for normal uniparental mitochondrial inheritance [85, 86]. In ascomycete yeasts, mitochondrial inheritance into the zygote is often biparental, but meiotic progeny revert quickly to homoplasmy (possessing a single mitotype) due to spatial segregation of mtDNA nucleoids. This segregation occurs immediately during meiosis in Schizosaccharomyces pombe and during the first mitotic divisions of meiotic progeny in Saccharomyces cerevisiae [87, 88]. In Dictyostelium, strict uniparental inheritance of mitochondria is not maintained, and because of the syncytial phase during gamete fusion mentioned earlier, mitochondrial genomes can be inherited laterally, so that meiotic progeny have three parents, their nuclear chromosomes recombined from two parents, and their mtDNA inherited from a third [11]. This unusual mode of inheritance may be related to the cannibalistic sexual development of zygotes in these amoeba [89] leading to selection for cytoplasmic genes that promote survival into progeny. It is possible that this feature of social amoebae, in which the mitochondrial genome seems to meet a strict definition of a selfish genetic element [90], is atavistic, resembling patterns of (proto-)mitochondrial inheritance early in eukaryotic evolution before strict controls on gamete fusion and organelle inheritance first emerged [11].

The evolution of zygosis and lifecycle regulators

Sexual lifecycles could have evolved originally in two ways: either meiosis (or another reductional mode of division) arose first as a mechanism for ploidy reduction in cells that had undergone serial reduplication of their genomes [91–93], or zygosis preceded it as a way to mask disadvantageous genetic variants [94]. One plausible hypothesis for the origin of zygosis suggests that cell fusion could have arisen as a by-product of a conjugative infectious process by which symbionts spread from cell to cell [11, 95, 96]. The realisation that HAP2-GCS1, the gamete fusogen very likely to be ancestral to all eukaryotes, is structurally related to viral fusion proteins provides a possible molecular basis for a parasitic origin of sex: co-option of a viral fusion protein [97]. It is, of course, also possible that this co-option could have occurred in the other direction; viruses frequently take possession of host genes [98]. Eukaryotes likely evolved from stem archaeal cells that acquired a bacterial symbiont that became the mitochondrion [99–101], though this is still a matter of contention (see [102–104]). This raises the possibility that elements specifically functioning in zygosis in eukaryotes, including HAP2-GCS1, could have archaeal (or archaeoviral) ancestry, as do components of meiosis like Spo11 [105, 106].

A related question concerns the origin of mating types: did unisexual, self-fertile individuals precede the evolution of genetically determined, distinct mating types [107]? The apparent ancestral role of HAP2-GCS1 does not settle this question, because it is not clear whether it originally functioned unilaterally or bilaterally in promoting fusion (both are found in extant eukaryotes, as described above). Mating types have been proposed to have evolved primarily as a way to cleanly trigger diploid-specific functions: [108] gamete fusion mixes cytoplasmic components from two differentiated haploid cell types, allowing fusion to be used as a logical AND gate. The roles of homeoproteins during zygosis in plants, green algae, brown algae, and fungi (and apparently divergent homeoproteins in social amoebae) might seem to support this idea, since the apparent conservation of this function again suggests that it could be ancestral to all eukaryotes. However, these homeoproteins are TALE class in plants, green algae, brown algae, and some fungal proteins (MATα2-like and Pi proteins in ascomycetes, HD1-type proteins in basidiomycetes), and non-TALE in other fungal proteins (MATa1-like proteins in ascomycetes and HD2-type proteins in basidiomycetes); the relationship of the social amoebae proteins is unclear. These differences make it difficult to exclude the possibility of convergent evolution. Again, if functions in lifecycle transitions could be ascribed to archaeal proteins related in structure to homeodomains [80], or if homeoproteins were found to be involved in zygosis in other eukaryotic lineages, and any widely conserved downstream targets of such proteins identified, the deep evolutionary picture could become clearer. The unclear evolutionary origins of fungal gamete fusion mechanisms, involving the apparent loss of ancestral HAP2-GCS1 function, and homeoprotein functions during zygosis might be elucidated by a broader examination including diverse fungal lineages, in the same way as a recent study demonstrating how a viral protein was co-opted to rewire ancestral mitotic cell cycle regulation [109].

Advances in our understanding of the molecular mechanisms of zygosis in diverse eukaryotes in recent years have transformed our understanding of the evolution of sex. It is to be hoped that further exploration of the biochemistry and cytology of the key components, ideally in even more eukaryotic lineages, along with the phylogenetic data that will result from ongoing genome and transcriptome sequencing efforts will allow us to address the important questions that so far remain unanswered and, perhaps, to identify further conserved ancestral genes.