The evolution of sex and two sexes
The evolution of sex and two sexes
Abstract and Keywords
Sexual reproduction has proven so formidable a challenge for evolutionary biologists that it is commonly spoken of as “the paradox of sex.” Stated simply, individuals forsake one-half of their genetic representation in offspring by engaging in sexual versus asexual reproduction. There must be substantial benefits to compensate for so great a cost. Recent theory proposes that the primary benefit of sex is the tremendous diversity of genotypes produced via recombination during sexual reproduction that provides the raw material necessary to compensate for mutational erosion of mitochondrial genes. Another line of new thinking proposes that the reason that virtually all eukaryotes have two mating types rather than multiple mating types is that the existence of two mating types enables single mitochondrial genotypes to be vetted for compatibility with nuclear genotype. This chapter considers the implications and evidence for these new mitonuclear-based theories of key evolutionary ideas.
When driving toward the rim of a basket, LaBron James is phenotypic perfection. He is huge and powerful, bigger than nearly all other men, but still blazing fast, remarkably coordinated, and even nimble. The magnificence of his phenotype was recognized before he was 15 years old, and the adult LaBron James has not disappointed those who saw future greatness in the boy from Akron, Ohio. As I write this book, he has dominated professional basketball for a decade, but he would dominate most sports that are based on size, strength, and eye–hand coordination. He may be the greatest athlete among the world’s 7.5 billion people. He may be the greatest athlete who has ever lived. And yet, there is no guarantee that his children will be standouts in basketball or any other sport. They almost surely will not match the athletic prowess of their father. Albert Einstein’s children did not excel in math or physics; Winston Churchill’s children were not great statesmen; Marie Curie’s children contributed nothing to science. Sexual reproduction scrambles and recombines sets of genes every generation, and in so doing, sex makes phenotypes like LaBron James’ unique and ephemeral.
Explaining the evolution of sexual reproduction is among the most significant and long-standing problems in evolutionary biology (Williams, 1975; Maynard Smith, 1978). Sexual reproduction has proven so formidable a challenge for evolutionary biologists that it is commonly spoken of as “the paradox of sex” (Otto and Lenormand, 2002). Aside from breaking up adaptive phenotypes, sex dilutes genetic representation in offspring. If an individual reproduces simply by asexually duplicating its genotype, it gains twice the genetic representation in offspring compared with a sexually reproducing individual. This loss of genetic representation in offspring is termed the “two-fold cost of sex,” and there must be substantial benefits to compensate for so great a penalty associated with sexual reproduction.
In a related but fundamentally distinct line of investigation, researchers have long pondered why the existence of two mating types—males and females—is a nearly universal condition for eukaryotes (Fisher, 1930). Theoretically, mating types are not necessary for sex, and yet essentially all eukaryotes have mating types. Two mating types necessitates that half of the individuals in a population are inappropriate as mates. On first consideration, it would seem that, if you must have mating types, the most beneficial strategy would be to have numerous mating types so that most individuals encountered would be a suitable mate. A system with two mating types seems like a losing strategy for sexual reproduction, and yet two mating types is the nearly (p.97) universal state across eukaryotes. As with sexual reproduction, two mating types is an evolutionary outcome that begs for an explanation (Lane, 2005).
New thinking places the interactions of mt and N genomes at the center of explanations for the evolution of both sexual reproduction and two mating types (Lane, 2005). Incorporating a consideration of the fundamental necessity of mitonuclear coadaptation into existing theory has re-invigorated research into both the evolution of sex and the evolution of mating types. In this chapter, I will consider the implications and evidence for these new mitonuclear-based theories of key evolutionary ideas.
The evolution of sex
The necessity of recombination
Sexual reproduction is a specific form of biotic replication, and it is unique to eukaryotes. Prokaryotes certainly replicate their genetic material and engage in various forms of gene exchange, but no prokaryotic lineage engages in reciprocal exchange of genetic material (Xu, 2004; Narra and Ochman, 2006). Sexual reproduction in eukaryotes involves assortment and recombination of chromosomes in a process that entails the fusion of haploid gametes to produce a diploid individual (Goodenough and Heitman, 2014) (Figure 5.1). With very few exceptions (Box 5.1), all eukaryotes engage in sexual reproduction (Vrijenhoek, 1998; Whitton et al., 2008). When judged on an evolutionary time scale, the few eukaryotic lineages that exist as obligate asexuals rarely persist for long (Neiman et al., 2009).
Sexual reproduction with recombination appears to be a primitive trait for eukaryotes. Despite approximately 2 billion years of divergent evolution in lineages such as amoeba and elephants, the proteins required for recombination have been conserved across eukaryotes (Ramesh et al., 2005; Speijer et al., 2015). Conserved protein structure indicates that there has been persistent and consistent selective pressure for recombination across vast evolutionary time and that recombination has been a necessity for eukaryotes throughout their history (Garg and Martin, 2016). The simultaneous evolution of sexual reproduction with recombination and a two-genome cellular structure suggests that something about genomic architecture of eukaryotes made sexual reproduction a necessity (Lane, 2005; Garg and Martin, 2016). It is rather surprising, therefore, that it was only after a hundred years of contemplating the evolution of sex and fifty years after the discovery of the mt genome that mitonuclear coevolution was invoked as an explanation for sexual reproduction.
Theories that have been proposed to explain sexual reproduction do not focus specifically on the exchange of genes between individuals—that can be accomplished with lateral gene transfer as in prokaryotes (Figure 5.1). Rather, the key benefit of sexual reproduction is invariably proposed to be the recombination of sets of N genes within chromosomes (Otto, 2009). Reciprocal recombination of genotypes is only possible through sexual reproduction. The “recombination” in which prokaryotes (p.98) engage is a one-way lateral transfer of genes and is fundamentally different than the reciprocal recombination of eukaryotes (Ochman et al., 2000; Lang et al., 2012) (Figure 5.1). To enable reciprocal recombination, individuals of at least one sex must exist as diploids for at least part of a life cycle so that haploid gametes of one individual join with the haploid gametes of another individual to produce an offspring with a diploid genotype drawn from two individuals. Reciprocal recombination is the exchange of genes between the two parental genotypes (Figure 5.1). The simple process of recombination creates the opportunity for separating linked genes and thus for isolating specific genes for elimination via negative selection or for proliferation via positive selection (Maynard Smith, 1978).
Two specific benefits of recombination have been the focus of hypotheses for why sexual reproduction is so prominent across eukaryotes. The first and most frequently stated hypothesis for the evolution of sexual reproduction is that there is an urgent and perpetual need for genetic variation on which selection can act (Otto, 2009). Unstable and changing environments is frequently stated as the reason that genetic diversity is essential, and pathogens are often presented as the most significant environmental factor that create a need for diverse gene combinations on which natural selection can act (Hamilton et al., 1990; Howard and Lively, 1994). However, (p.99) as pointed out by Otto (2009) there are two fundamental problems with the assertion that genetic diversity is beneficial enough to make nearly all eukaryotes sexual. First, recombination is not necessary to generate genetic diversity. Prokaryotes derive tremendous genetic diversity via horizontal gene transfer without sexual reproduction (Figure 5.1). Second, there is a significant cost to genetic recombination that should counteract and potentially outweigh the benefits of genetic diversity. Recombination not only generates new gene combinations that present the opportunity for novel adaptation; recombination also shuffles fit gene combinations making it harder to maintain coadaptive gene complexes. The fact that most genome replication in (p.100) eukaryotes is asexual replication (Box 5.2) supports the assumption of significant costs associated with sexual reproduction.
The second set of hypotheses to explain the evolution of sexual reproduction focuses on the accumulation of deleterious genes within a genome (Maynard Smith, 1978; Kondrashov, 1988). As I discussed in previous chapters, without recombination, Hill–Robertson effects arising from absence of genetic recombination can lead to the fixation of slightly deleterious mutations within the genome (Figure 4.1). In addition, once a deleterious allele is fixed within a non-recombining genome, selection has no means to remove it without eliminating the entire genome (Muller, 1964; Felsenstein, 1974). Thus, there is a perpetual accumulation of deleterious alleles via the process known as Muller’s ratchet. The result can be mutational erosion and loss of fitness over time (Maynard Smith, 1978). Because sexual reproduction scrambles gene sets, it enables the selective elimination of bad alleles without the elimination of beneficial alleles. The hypothesis that sexual reproduction and recombination are necessary to combat Muller’s ratchet is decades old, but new thinking recasts this hypothesis in terms of compensatory coevolution between the N and mt genomes.
The evolution of sex in light of mitochondrial evolution
In the first two chapters of this book, I promoted the hypothesis that the evolution of eukaryotes and the evolution of mitochondria were the same event; in other words, eukaryotes were born of the chimeric fusion of two prokaryotes (Koonin, 2010; Speijer, 2015). I’ll now extend this origin theory a step further: the evolution of eukaryotes, the evolution of mitochondria, and the evolution of sexual reproduction occurred simultaneously (Lane, 2015a). Essentially all modern eukaryotes engage in sexual reproduction with recombination and evidence indicates that sexual reproduction is a primitive (p.101) trait in eukaryotes (Goodenough and Heitman, 2014; Garg and Martin, 2016). It would seem that either sexual reproduction was a key to a successful fusion of the genomes of two prokaryotes, or else the two-genome architecture of the proto-eukaryote made sexual reproduction an indispensable reproductive strategy.
The ideas that I will summarize regarding the evolution of the N genome and sexual reproduction are new and, at present, these new hypotheses are largely speculative. However, these new explanations for the evolution and maintenance of sexual reproduction are consistent with genomic data, and I find them compelling. For simplicity in the following discussions, I’ll refer to the cell that gave rise to the nucleus as the “host” and the cell that gave rise to the mitochondrion as the “endosymbiont.” As I’ve stated repeatedly throughout this book, I advocate the hypothesis that these cells were equal partners in the origin of eukaryotes, but “host” and “endosymbiont” labels make the explanations for the evolution of sex much simpler.
The conditions and circumstances that are proposed to have given rise to the evolution of sexual reproduction emerge at the very origin of eukaryotes. When an archaeon and a bacterium fused to form a single organism with two genomes, there would have been immediate problems for the archaeon host partner, whose genome existed in one copy surrounded by multiple copies of the bacterial endosymbiont’s genome. Individual endosymbionts inevitably died within host cells without the host cell itself dying, and these deceased endosymbionts would have released DNA into the cytosol. As a consequence, host DNA would have been bombarded by pieces of endosymbiont DNA. These segments of the mt genome would have inserted themselves into the host genome and disrupted coding for essential genes (Lane, 2005, 2015a; Rogozin et al., 2012) (Figure 5.2). Such corruptions of the host genome by insertion of endosymbiont genes created what was essentially a very high mutation rate—an unprecedented rapid and massive alteration of the nucleotide sequence of the host (Timmis et al., 2004). It would also have led to a potentially unstable genome as segments of endosymbiont DNA were inserted into the host genome (Lane, 2011a). It is easy to imagine that bombardment of the host genome by endosymbiont DNA would have put a quick end to the proto-eukaryote before it had much of a chance to get started. One can imagine that bombardment of the host DNA by endosymbiont DNA did put a quick end to a multitude of false starts at eukaryotic evolution. After what likely was an unimaginably large number of failed attempts, a series of highly improbable events fell in sequence and a chimeric cell survived.
There is substantial evidence to support the hypothesis that endosymbiont DNA was inserted into host DNA very early in eukaryote evolution. To begin with, the N genome of all eukaryotes is structured as coding sequences disrupted by non-coding sequences—“genes in pieces” (Koonin, 2009). The coding nucleotide sequences are called exons, and the non-coding sequences that sit between the exons are called introns. Most introns are of mitochondrial origin (Martin and Koonin, 2006), and numerous introns are conserved among diverse eukaryotic lineages, indicating that the introns in universally important and conserved genes (like citrate synthase) are of ancient origin, dating to the origin of eukaryotes (Rogozin et al., 2003). Incorporation of mt DNA into the N genome continues commonly to the present day because the (p.102) problem of endosymbionts deconstructing and releasing DNA into the cytosol has not gone away (Hazkani-Covo et al., 2010). Indeed distinguishing between true mt genes located in the mt genome and copies of mt genes that have inserted into the N genome is an ongoing challenge for modern molecular biologists (Sorenson and Quinn, 1998).
With the random insertion of endosymbiont DNA into the host genome, mutational meltdown would have occurred at an unprecedented rate, jeopardizing the survival of a proto-eukaryotic lineage (Garg and Martin, 2016). There certainly would have been urgent need for mechanisms to edit out inserted DNA as well as to select against deleterious changes to genotype (Lane, 2015a). The need for gene editing is proposed to have given rise to the eukaryotic spliceosome and the nuclear membrane, two adaptations that enabled post-transcriptional editing of the chunks of endosymbiont DNA that had been inserted into functional sequences of the host genome (Lane, 2015a; Garg and Martin, 2016). According to these hypotheses, insertion of endosymbiont DNA into the host DNA in the early evolution of eukaryotes gave rise to introns as well as the potential for alternate splicing—combining exons (p.103) in different combinations to produce different gene products. Alternate splicing, in turn, created enormous opportunities for evolutionary novelty (Gilbert, 1978). The evolution of the spliceosome, the nuclear membrane, introns, and alternate splicing are each seminal events in the evolution of eukaryotes (Lane, 2015a; Garg and Martin, 2016). Their origins are the subject of current investigation and debate, but these topics are somewhat peripheral to the evolution of sex. Interested readers can easily pick up the threads of these investigations from the literature I’ve cited. The focus in this chapter is on the evolution of sexual reproduction.
Gene editing is proposed to be a key adaptation that enabled a proto-eukaryotic nucleus to deal with endosymbiont DNA insertion. But in the face of mutational meltdown of the host genome resulting from rampant insertion of endosymbiont DNA, it is reasonable to propose that there was also strong selective pressure for the evolution of recombination, which is the mechanism currently recognized as the best protection against the accumulation of deleterious alleles (Barton and Charlesworth, 1998; Otto and Lenormand, 2002). Even if introns could generally be accommodated through spliceosome editing of DNA transcripts, there would still be important variation among individuals in the position of particular introns (Lane, 2015a). Moreover, with many novel genes being generated through the lateral transfer of genes, there would be a critical need for natural selection to both remove deleterious variants and promote useful novelties. These factors are proposed to have led to the evolution of sexual reproduction involving recombination (Lane, 2015a).
This hypothesis for the origin of sexual reproduction explains why the mechanism of gene exchange that had served prokaryotes for 2 billion years was not adequate for the proto-eukaryote. The massive mutational load that came with a second genome within the proto-eukaryotic cell would have created a heretofore unprecedented destabilizing force on the host genome (Lane, 2015a). The eukaryotic cell required a new form of gene exchange that provided a means to avoid Hill–Robertson effects and Muller’s ratchet. Sexual reproduction with recombination restored genome stability, perhaps rescuing the failing lineage that was being overwhelmed by rapid genomic alterations. The incorporation of mt genes into the N genome would have begun immediately following chimeric fusion, so selection for recombination would have arisen during the initial formation of the first eukaryote, explaining why sexual reproduction is a shared and unique feature of eukaryotes. Presumably, because it remains an ongoing issue for all eukaryotes, lateral transfer of mt DNA to the N genome remains a reason why sexual reproduction is maintained in essentially all eukaryotic lineages (Timmis et al., 2004). All other benefits of sexual reproduction—such as the need for genetic diversity to adapt to changing environments—that have been widely discussed as explanations for the evolution of sexual reproduction would only reinforce the need for sexual reproduction in eukaryotes driven by a need to deal with insertions of mt DNA into the N genome.
Another key consideration regarding the evolution of the N and mt genomes of eukaryotes is that sharing genes via a metagenome, which was the mechanism for generating genetic diversity in both of the prokaryotic lineages that gave rise to eukaryotes, was no longer workable. As soon as the core respiratory machinery of mitochondria became dependent on products of both the N and mt genomes, which (p.104) likely occurred during the early evolution of eukaryotes, coadaptation and compatibility became key factors in individual fitness. N gene sets and mt gene sets had to be transmitted vertically so that, each generation, compatible sets of genes could be matched correctly and the two sets of genes could coevolve. Pulling genes from a common metagenome could not accommodate tight coevolution toward coadaptation and thus it ceased. Recombination of diploid chromosomes became the mechanism for reshuffling of genes without endangering mitonuclear compatibility. Choice for compatible genomes via mate choice prior to sexual reproduction, the topic of Chapter 8, ensures that coevolved and coadapted gene sets are transmitted together across generations. In this way, exchange of genes via sexual reproduction was limited to individuals with a shared coadapted mitonuclear genotype, and thus, this new form of reproduction was also the origin of species boundaries (see Chapter 7).
Avoiding mutational meltdown
Among the many seminal changes that occurred following the chimeric fusion of two prokaryotes to form one eukaryotic organism, aerobic respiration on a heretofore unachievable scale was undoubtedly the most significant (Lane, 2015b). The flow of energy provided by aerobic respiration in many individual mitochondria, which existed as compartmentalized and duplicated power centers within a eukaryotic cell, enabled a massive expansion of the eukaryotic genome (Lane and Martin, 2010). This expansion included an increase in the number of proteins produced by eukaryotes by an order of magnitude or more compared with prokaryotic ancestors, and it trivialized the cost of the transfer of genetic elements from the endosymbiont to the host in terms of the energetic cost of DNA replication (Lane, 2014, 2015a). The clear benefit of an oxygen-fueled respiratory furnace is massive energy production. But, as the manager of any public utility can attest, energy production is a messy business, and the messy part of aerobic respiration via an electron transport system (ETS) in mitochondria is the production of free radicals. These free radical products are unstable and react with whatever molecules they contact, causing a host of cell damage that includes mutating DNA. Indeed, ionizing radiation is mutagenic largely because it creates free radicals through the dislodging of electrons from water molecules within biological systems, thereby creating free radicals that damage DNA as well as other cellular components (Lane, 2002).
The mt genome is positioned immediately adjacent to the ETS, which is the primary source of reactive oxygen species (ROS) production in the cell, so mt DNA exists on the front lines of the redox danger zone. It has long been proposed that exposure to free radicals produced during aerobic respiration is the reason that mt DNA has a much higher mutation rate than N DNA in many eukaryotes (Figure 1.7). However, new evidence indicates that copy error and not exposure to free radicals is the source of most mt DNA mutations (Lagouge and Larsson, 2013; Itsara et al., 2014). I’ll take up the discussion of the source of mt DNA mutation in detail in Chapter 6. For now, we need only recognize that mutations of mt DNA lead to fixation of deleterious alleles because, in a haploid and non-recombining genome, Hill–Robertson effects cannot be (p.105) avoided, and once deleterious mutations are fixed, lack of recombination shields slightly bad genes from natural selection. Accumulation of deleterious alleles in the mt genome is fundamentally bad because mt genes code for core respiratory processes. Hence, it is proposed that N genes perpetually evolve to compensate for deleterious mt genes—these considerations were the topic of Chapter 4. What I did not consider in detail in Chapter 4 is the mechanism that enables the N genome to generate sufficient genetic diversity to plausibly compensate for deleterious genes in the mt genome.
Havird et al. (2015a) proposed that compensatory coevolution by the N genome (Chapter 3) is only effective if the N genome can evolve adaptive changes at a pace that keeps up accumulation of deleterious alleles in the mt genome. Therefore, they proposed that sexual reproduction and recombination evolved in eukaryotes in direct response to the need for substantial and perpetual genetic variation in the N genome to enable compensatory coevolution to keep pace with the high mutation rate of mt genes (Figure 5.2). Sexual reproduction with recombination generates the needed variation in N genotypes. This hypothesis proposes that the benefits of compensation for mutational erosion of the mt genome through the perpetual generation of novel gene sets via sexual reproduction with recombination more than offset the substantial cost of scrambling of successful gene combinations.
Havird et al. (2015a) recognize two critical assumptions of their hypothesis. First, their model assumes that early in eukaryotic evolution and continuing in the lineages of most extant eukaryotes, bi-parental transmission of mt genomes was not an evolutionary option. If eukaryotes could evolve bi-parental transmission of mitochondria then there would be no need for compensatory coevolution by N genes. Bi-parental inheritance of mt genomes creates opportunities for recombination of mt genes and, as I’ve discussed extensively, recombination stops Muller’s ratchet and mutational erosion in the mt genome. The evolution of bi-parental versus uniparental inheritance of mt DNA is the topic of the second half of this chapter, but here I will note that uniparental inheritance of mt DNA is the common pattern of mitochondrial transmission observed in eukaryotes.
Second, the hypothesis proposed by Havird et al. (2015a) requires that a high mt DNA mutation rate was the ancestral condition for eukaryotes. In support of this assumption, these authors point out that a high mt DNA mutation rate is the common state among eukaryotes (Figure 1.7) and that mt DNA is replicated much more than N DNA, providing more opportunity for mutations arising from replication error (Melvin and Ballard, 2017; Szczepanowska and Trifunovic, 2017). Moreover, the DNA of endosymbionts living in eukaryotic cells shows elevated rates of mutation compared with N genomes (Itoh et al., 2002). The rates of substitution are faster in the bacterial lineage that gave rise to mitochondria than in the archaeon lineage that gave rise to the nucleus (Sung et al., 2012). mt DNA is highly derived relative to bacterial DNA suggesting a rapid increase in mutation rate early in eukaryotic evolution (Gray et al., 1989). And finally, as previously noted, mt DNA sits next to the redox furnace with massive exposure to mutagenic free radicals. So, it is certainly plausible that from the early phases of eukaryotic evolution there was a fundamental need for compensatory coevolution by the N genome to bail out the mt genome.
(p.106) To this point, I’ve presented two hypotheses for the evolution of sexual reproduction that involve mitochondria: endosymbiont bombardment of the host genome, and the need for N genomic diversity to enable compensatory coevolution (Figure 5.2). Speijer (2016) proposed a third hypothesis for the evolution of sexual reproduction in eukaryotes that involves mitochondria. This hypothesis focuses on the mutagenic effects of the free radicals that came along with aerobic respiration at the origin of eukaryotes in the evolution of sex. In contrast to the Havird et al. (2015a) hypothesis that focused on mutations in the mt genome, however, Speijer (2016) considered the integrity of the N genome. The basic premise is the same as in the hypothesis put forward by Havird et al. (2015a): the evolution of massive respiration via mitochondria created more free radicals and more DNA damage in eukaryotic cells (Lane, 2011b). Where Havird et al. (2015a) focus on mutational meltdown in the mt genome and propose that a need for nuclear compensation drove the evolution of sexual reproduction, Speijer (2016) focuses on maintaining the integrity of the N genome itself as the driving force in the evolution of sexual reproduction with recombination. According to this hypothesis, the high mutation rate of N genes caused by free radical damage resulted in selection for sex and recombination to maintain the integrity of the N genome (Figure 5.2). These two free radical-based hypotheses—the Havird et al. (2015a) hypothesis focused on mt mutations and the Speijer (2016) hypothesis focused on mutations in the N genome—are not mutually exclusive hypotheses. As Speijer (2016) explains, compensatory coevolution to reverse problems in the mt genome would be an added benefit to advantages attained through better editing via sexual reproduction of the N genome. Both of these hypotheses that are founded on the premise that mt DNA mutations arise primarily from the effects of free radicals are challenged by new data showing that, in modern eukaryotes, free radicals contribute little to mt DNA mutation rate, and that mt DNA mutations arise primarily from copy error (Melvin and Ballard, 2017; Szczepanowska and Trifunovic, 2017). However, free radical-induced mutation of both mt and N DNA may have been more of a problem in early eukaryotic evolution before more sophisticated DNA editing processes evolved. Also, the rate of mt DNA replication would have increased exponentially to accommodate the energy needs of the larger and more complex eukaryotic cell, so copy error of mt DNA would have increased mt DNA mutation rate even if free radicals did not induce significant mutation.
At present, all hypotheses for the evolution of sex that invoke mitonuclear coevolution and coadaptation are untested. Going forward, a consideration of mitonuclear dynamics will certainly be a part of any discussion of the evolution of sex.
The evolution of two sexes
The evolution of anisogamy
While the paradox of sex is known to all evolutionary biologists, the paradox of two sexes is a major puzzle in eukaryotic evolution that attracts much less attention (Billiard et al., 2011). For many biologists, the question of why there are two sexes (p.107) was settled more than 40 years ago by evolutionary theorist Geoffrey Parker and his colleagues when they proposed a game theory model of the likely selective forces at work on a primitive population of sexually reproducing organisms with one gamete type (Parker et al., 1972) (Figure 5.3). In this hypothetical population, all individuals produce gametes that are acceptable as partners for gametes from all other individuals in the population, and there is scramble competition to find a mating partner. It is assumed that there will be variation among individuals in the size of gametes they produce, just as there is variation in expression of any trait in a population. Moreover, it is assumed that there would be a necessary tradeoff between gamete size, the likelihood of gamete survival, and gamete numbers. Large gametes require lots of resources that increase the survival prospects of both the gamete and the zygote that results, but only a few large gametes can be produced because they require substantial investment in resources. Small gametes, on the other hand, are less costly to produce enabling the production of more, but because such small gametes carry few resources, their prospects for survival and the survival prospects of the zygote they produce are low. Thus, some individuals in the hypothetical ancestral population produce fewer but larger gametes and some produce more but smaller gametes.
Initially, there would be mostly gametes of intermediate size, motility, and vitality, but Parker et al. (1972) showed in simulations based on game theory models that this scenario sets the stage for disruptive selection for the two extreme morphologies. The population was predicted to rapidly evolve to be anisogamous—to have two types of individuals that each produce one extreme form of gamete: egg or sperm (Figure 5.3). (p.108) The reason that extreme morphologies win out is that small gametes gain the huge benefit of being mobile. They are more likely to find a mating partner, but they pay the cost of low survival. Large gametes, in contrast, gain the huge benefit of high survival, but they pay a cost of low motility and low chance of encountering another immobile mating partner. The small gametes ensure that large gametes encounter a sexual partner and the large gametes ensure that sufficient resources are available for the zygote to survive (Parker et al., 1972).
The model created by Parker et al. (1972) is so compelling that it is easy to look past what it does not explain. One glaring omission in the model is the transmission of cytoplasmic elements such as mitochondria. Perhaps it goes without saying that mitochondria are transmitted by the bigger gamete because they have more room to haul around mitochondria, but in real-world anisogamous eukaryotes, sperm also carry mitochondria. Even more fundamentally, the Parker et al. (1972) model does not explain the evolution of mating types. The endpoint of the model is simply two gamete types: small and mobile and big and immobile (Charlesworth and Charlesworth, 2010). In the Parker et al. (1972) model, big gametes can fuse with big gametes and small can fuse with small. As a matter of fact, Parker et al. (1972) considered the likelihood that big gametes would benefit by fusing with other big gametes and eliminate the benefits of small gametes, but an isogamous population of large gametes is subject to invasion by small, mobile gametes and, in these models, anisogamy inevitably evolves.
Thus, at the endpoint of the game theory model put forth by Parker et al. (1972), we are still left with one mating type with two morphological strategies. In the eukaryotic world in which we live, however, there are always mating types, and for nearly all eukaryotes there are exactly two mating types (Billiard et al., 2011). The preponderance of eukaryotes with two mating types presents an evolutionary puzzle potentially as great as the paradox of sex (Hurst and Hamilton, 1992; Lane, 2005).
The paradox of two mating types is easiest to understand if we consider the limitations that are imposed by such a system. When there are two mating types, half of the individuals in the population are unsuitable sexual partners. This is not a trivial consideration. For essentially all organisms, there are real costs involved in searching for a suitable sexual partner (Parker, 1978), and so it is easy to imagine the benefits to be had if a third mating type evolved in a population with two mating types. Assuming that all mating types except one’s own mating type are suitable sexual partners, a third mating type would be expected to spread in the population until three mating types existed at equal frequency. With the addition of a third mating type, two out of every three randomly encountered individuals would be a suitable mate, odds that are considerably better than the one-out-of-two success rate when there are two mating types. Addition of a fourth mating type would, in turn, further reduce the cost of mate choice—with three out of four rather than two out of three suitable. Based on this simple assessment, we might expect the evolution of many mating types (Lane, 2005). The other potential winning strategy would be no mating types, whereby every randomly encountered individual would be a suitable mate. Two mating types is the worst system with regard to the cost of mate searching (Box 5.3). It is therefore remarkable that, with few exceptions, eukaryotes have two mating types (Billiard et al., 2011). (p.109)
The connection between uniparental inheritance of mitochondria and two mating types is fundamental (Figure 5.4). If there are no mating types such that any individual in a population is a suitable sexual partner, then there is no mechanism to ensure that, during sexual reproduction, one individual will transfer a mt genome and one will not. Uniparental inheritance of mitochondria becomes a possibility only if there are two mating types and transmission of mitochondria is restricted to just one mating type. This argument also comes back to the evolution of isogamy/anisogamy: uniparental inheritance of mitochondria is most commonly achieved in eukaryotes when one mating type (female/egg) receives N DNA from the other mating type (male/sperm) and the DNA-receiving gamete does not permit the transfer of mitochondria. It is not a coincidence that nearly without exception across eukaryotes, the DNA-receiving sex is also the sex that exclusively transmits mt DNA to the next generation. Anisogamy is not essential for uniparental inheritance of mitochondria—there are diverse mechanisms for ensuring uniparental inheritance of mt DNA (Sato and Sato, 2013)—but anisogamy provides a ready mechanism for transmission of mt DNA by only one parent. Moreover, it is certainly possible to have two mating types and to still allow for bi-parental transmission of mitochondria—all that is required is weak gatekeeping by the egg—and indeed such bi-parental transmission in systems with two mating types is not rare in eukaryotes (Birky, 2001). Most eukaryotes, however, have two mating types and uniparental inheritance of mitochondria that is enforced by the sex with larger gametes.
Genomic conflict within an individual
For decades, explanations for the evolution of two mating types in eukaryotes focused on genomic conflict among cytoplasmic elements, including particularly mitochondria (Cosmides and Tooby, 1981; Hoekstra, 1987; Hurst and Hamilton, 1992; Hutson and Law, 1993). These hypotheses are all founded on the importance of avoiding heteroplasmy, which is having multiple mt DNA types within a single individual. Heteroplasmy is proposed to be a problem because, with multiple mt genotypes replicating within a single eukaryotic individual, there can be selection for traits that promote a mt genotype at the expense of the N genome and the eukaryotic organism as a whole. While there is strong theoretical support for the problems that can be caused by heteroplasmy, only a few empirical studies have directly confirmed a loss of organism fitness due to competition among mitochondria (e.g. Sharpley et al., 2012).
The basic argument for the potential dangers of heteroplasmy can be explained most easily with a simple verbal model. Begin with an ancestral population of a simple eukaryote with no mating types and bi-parental transmission of mitochondria, chloroplasts, and other cytoplasmic elements. During sexual reproduction, the nuclear material from the two parents would combine to form one N genotype in the offspring, but mt genomes would be transmitted as multiple copies with no fusion or recombination. Thus, divergent mt genotypes would exist within a single organism. (p.111) In such a situation, genomic conflict is theoretically inevitable (Cosmides and Tooby, 1981; Hurst and Hamilton, 1992). Any variant mt genotype that promoted its own replication at the expense of the replication of any other mt genotype would rapidly go to fixation. Such self-serving strategies would be good for that particular mitochondrial lineage, but they would likely be bad for the N genome and for the organism as a whole. This concept can be recast in mathematical models, and the conclusions are the same: bi-parental inheritance of mitochondria inevitably leads to conflict among mitochondria within an individual (Hurst and Hamilton, 1992; Christie et al., 2015).
How does a lineage solve the problem of conflict among mt genomes within an individual? The most basic answer is to allow only one parent to transmit mitochondria; in other words, to have uniparental inheritance of mitochondria (Figure 5.4) (Hurst and Hamilton, 1992). The mitochondria within an individual will inevitably be more genetically similar than mitochondria between individuals. Limiting transmission of mitochondria to one parent goes a long way toward eliminating conflict among mitochondria within an individual. The most familiar manifestation of uniparental transmission is to allow only females to transmit mitochondria (Hutson and (p.112) Law, 1993). There can still be genetic variation among mitochondria even within the germ line of a single individual (a topic that will be taken up in Chapter 6), but restricting between-generation transmission of mitochondria to a single mating type vastly reduces the problem of conflict within the population of mitochondria in a zygote. Thus, a widely accepted explanation for why a system of two mating types predominates in eukaryotes is that this provides a mechanism to enable uniparental inheritance of mitochondria and to avoid genomic conflict among mitochondria (Cosmides and Tooby, 1981; Hoekstra, 1987; Hurst and Hamilton, 1992; Hutson and Law, 1993).
Selection against heteroplasmy and selection for mitonuclear coadaptation
Because the mt genome is subject to Muller’s ratchet and mutational erosion, theory predicts that mt DNA will be prone to the accumulation of deleterious changes to coding sequences. Paradoxically, however, the coding sequences of mt genes tend to be more conserved than comparable coding sequences of N genes (Popadin et al., 2013) (see Chapter 3). Two factors are proposed to explain how a genome predicted to be eroded by deleterious mutations is observed to undergo little functional change over time. The first is a mitochondrial genetic bottleneck via a sequestered germ line each generation. I’ll come back to mitochondrial genetic bottlenecks when I discuss the evolution of senescence in Chapter 6. Here I focus on the second factor that is hypothesized to counteract mutational erosion of mt DNA: the evolution of uniparental inheritance of mt genomes.
As I reviewed in Chapter 3, a prominent theory for how the ETS remains functional in the face of mutational erosion of mt DNA is compensatory coevolution: N genes are proposed to evolve functional changes that compensate for deleterious mutations in mt genes. Even the most ardent supporter of the compensatory coevolution hypothesis, however, acknowledges that compensatory coevolution with a N genome can at best be an emergency bailout of the mt genome once deleterious alleles become fixed. The first line of defense against mutation erosion must be strong and perpetual natural selection against deleterious mitochondrial mutation each generation. A serious limitation for natural selection acting on mt genotypes, however, is that unlike a N genome that always exists as one genotype, a single individual can carry multiple and divergent mt genotypes (Chinnery et al., 2000).
The optimal situation for efficient purging of deleterious mitochondrial alleles is to have little variation in the mt genotypes within an individual and substantial variation in mt genotypes between individuals. In this way, selection against poorly performing individuals will effectively remove mitochondrial alleles that bestow low fitness. Conversely, if there is heteroplasmy and thus multiple mt genotypes within each individual, then variation in mt genotypes within an individual can be as great as variation in mt genotypes among individuals and natural selection on any one genotype is less effective (Figure 5.5). As a consequence, when there is heteroplasmy, the purging of deleterious alleles becomes less effective and Muller’s ratchet ensues (p.113) (Bergstrom and Pritchard, 1998; Roze et al., 2005; Christie and Beekman, 2016). Christie and colleagues (Christie et al., 2015; Christie and Beekman, 2016) used simulation models to study the importance of uniparental versus bi-parental inheritance for natural selection on mt genotypes. They found that even when the mutation rate was relatively low and with or without mating types, the benefits of homoplasy led to the evolution of uniparental inheritance of mitochondria.
The simulations by Christie and colleagues focused on the purging of deleterious mt genes. But there is potentially more to selection on mt genotypes than simply eliminating bad and promoting good variants. For mitochondrial variants, good versus bad must always be considered in light of the N genes to which they are paired. Each generation, new N genotypes are produced through sexual reproduction, and these new N genotypes are matched to mt genotypes that are transmitted to offspring either from the mother, from the father, or from both. It is essential that combinations of mt and N genes with poor compatibility are revealed to selection so they can be eliminated. Without perpetual selection for mitochondria that are properly coadapted to N genes, such that they enable full respiratory function, then there will be a loss of mitonuclear coadaptation and a decline in the respiratory efficiency and fitness of a lineage (Lane, 2011c). It is in consideration of the need to purge deleterious mitochondrial alleles and to maintain mitonuclear compatibility that Lane (2005) and (p.114) Hadjivasiliou et al. (2012) proposed a new hypothesis for the evolution of two mating types and uniparental transmission of mitochondria. According to this new idea, uniparental inheritance of mitochondria via two mating types evolved specifically for the maintenance of mitonuclear coadaptation that can be eroded when there is heteroplasmy.
As with hypotheses focused exclusively on mutational erosion of mitochondria, this coadaptation hypothesis is centered on the effectiveness of selection when there is too much variation in mt genotypes within individuals. When mitonuclear function arises from diverse mt genotypes within an individual, selection does a poor job of either promoting good genotypes or eliminating bad genotypes. This is because natural selection is necessarily limited to assessment of average mitochondrial function, and bad mt genotypes can be carried along by good mt genotypes when there is heteroplasmy (Figure 5.5). As a result, when there is heteroplasmy the coevolution of co-functioning mt and N-mt genes falters and coadaptation declines (Lane, 2005). In contrast, when a single mt genotype is matched to a single N genotype within cells, selection is much more effective at eliminating the poorly functioning, poorly coadapted combinations and at promoting highly functioning, fully coadapted combinations. These authors proposed that the necessity for tight mitonuclear coadaptation might be an even more important selective force in the evolution of mating types and uniparental inheritance of mitochondria than genomic conflict among mitochondria with different genotypes (Hadjivasiliou et al., 2012).
Is this advantage of better mitonuclear coadaptation enough to drive the evolution of two mating types and uniparental inheritance of mitochondria? This question is trickier than it seems on first consideration. Mitochondrial mutation rate is a big factor. Mutation rate determines how rapidly variant mt genotypes will be generated and hence how valuable it is, each generation, to test the compatibility of the N genotype and single mt genotypes. Hadjivasiliou et al. (2012, 2013) used simulation models to study the conditions under which two mating types and uniparental inheritance evolve. They found that, under diverse conditions, uniparental inheritance of mitochondria enhances mitonuclear coadaptation and leads to higher population fitness over time. However, the model really only worked after uniparental inheritance of mitochondria was already established. Moreover, when the model started with a population with bi-parental inheritance of mitochondria, uniparental inheritance would only spread through a minority of the population no matter how high they made the mutation rate. The costs of mate searching were simply too high. The results of these models could explain why the degree of heteroplasmy is so variable in many groups of eukaryotes (Hadjivasiliou et al., 2013).
Thus, it seems that avoidance of Muller’s ratchet is the primary benefit of uniparental inheritance of mitochondria (Christie et al., 2015; Christie and Beekman, 2016), but that promotion of mitonuclear coadaptation can play an important role (Hadjivasiliou et al., 2012, 2013). This is an area of active research and more empirical studies to validate the modeling outcomes are needed.
A theme that runs through the chapters of this book is the dichotomy from which evolutionary biologists view the interactions of mt and N genes. Many evolutionary biologists focus on the conflicts that can arise when independently replicating genomes exist within a single individual. Genes that are beneficial to one genome might proliferate even if they detrimental to other genomes and to the organism as a whole (Burt and Trivers, 2008). As discussed above, resolving genomic conflict is the leading explanation for the evolution of two mating types and uniparental transmission of mitochondria. But even as there is a potential for genomic conflict, there remains the fundamental necessity for cooperation among mt genomes and between the mt and N genomes. The products of mt genes and N-mt genes must be able to function together with high efficiency; otherwise, cellular respiration is compromised and there is loss of fitness. Models suggest that the benefits for maintaining mitonuclear coadaptation can indeed be a major force in the evolution of two sexes. But conflict and cooperation are not alternative explanations; they are complementary explanations and both are perpetually in play during eukaryotic evolution. In the end, the necessity for genomic cooperation should perpetually reel in a tendency toward genomic conflict (Connallon et al., 2018).
For many decades, evolutionary biologists pondered the evolution of sexual reproduction from the perspective of extant eukaryotes with highly integrated systems. But sexual reproduction seems to have evolved in the very early stages of eukaryotic evolution not only before genomic systems were fully integrated but when the state of the organism was probably closer to genomic chaos. The fusion of two formerly independent prokaryotic organisms to form a proto-eukaryote with two genomes necessitated rapid and dramatic restructuring of both genomes. This process of genomic restructuring appears to have coincided with the evolution of sexual reproduction with recombination, and it seems likely that these events were causally linked. The bombardment of the N genome with genes and gene fragments from the mt genome drove selection for mechanisms that enabled cutting and splicing DNA sequences as well as for recombining sets of genes to better enable selection against deleterious changes to the genome. According to this hypothesis, the origin of a eukaryotic cell with mitochondria necessitated the evolution of sexual reproduction with recombination, a nuclear membrane, and the spliceosome. The prokaryotic genome that was evolving into the mt genome also faced challenges even as it became streamlined. Because the symbiont genome existed as multiple copies, competition among symbiont genomes within an individual posed a grave risk to the emerging eukaryote. In addition, because the mt genome remained a single-copy genome that was asexually transmitted, Muller’s ratchet leading to mutational meltdown (p.116) posed an ever-present risk to the eukaryote. And, as a backdrop to all of these interactions, there was a perpetual need to maintain coadaptation between the N and mt genomes. Two mating types promoted the transmission of a single mt genotype to offspring enabling stronger selection for mitonuclear coadaptation. According to these hypotheses, both conflict and cooperation played key roles in the evolution of two mating types.