Download Evolution and Mechanisms of Meiosis: Phylogenetic Variation and Conservation and more Study notes Genetics in PDF only on Docsity!
æ
10. THE CELL LIFE CYCLE
27 December 2020
Having covered the general features of cell growth, we now focus on a number of key mechanistic and temporal details. Just as the soma of multicellular organisms undergo developmental changes, single cells progress through several stages from birth to division, some of which are very tightly defined and regulated. It is, for example, vitally important for cells to have properly duplicated their genomes and oriented each complement to their appropriate destinations at the time of division. This is a particular challenge for eukaryotic cells with multiple chromosomes. During mitosis, each chromosome must be replicated once, and only once, and parallel sets of chromosomes must be transmitted to each daughter cell with a minimum number of errors at the sequence level. Although most eukaryotic (and all prokaryotic) cells reproduce in an effectively clonal manner, indefinite rounds of such propagation in the former are often punc- tuated by phases of sexual reproduction during which pairs of individuals exchange chromosomal segments by recombination. During such sexual phases, eukaryotic cells switch from mitotic to meiotic genome division, wherein a diploid phase is reduced to the haploid life-cycle stage. To return to the diploid state, haploid indi- viduals must locate partners of the appropriate mating type and then undergo fusion with them. Whereas for most multicellular species, the predominant growth stage is diploid, for a wide range of unicellular species, the vegetative phase is haploid, while the diploid stage is simply a transient moment between the initiation of cell fusion and meiosis. Sexual reproduction raises a number of functional and evolutionary issues that will be discussed below. How and why did the complex process of meiosis, which includes organized modes of chromosomal segregation and recombination, evolve out of the already detailed orchestrations of mitosis? How do cells make “decisions” to fuse only with appropriate partners? How are mating types determined, and why is the typical number of mating types within a species just two? Many of the proteins involved in various stages of sexual reproduction appear to diverge at unusually high rates, begging the question as to whether such evolution is the product of genetic drive-like processes associated with the relentless operation of selection for successful gene transmission. A secondary goal here is to introduce a breadth of comparative observations on the molecular basis of cellular features at a deeper level than in previous chapters. Although this initial exploration is restricted to the diversification of life-history mechanisms, many of the underlying themes will reappear in subsequent chapters on other cellular features. Two key issues concern the evolution of complexity at the
2 CHAPTER 10
molecular and network levels. Functional protein molecules often consist of orga- nized multimeric aggregations of subunits. Sometimes the subunits are all encoded by the same locus (homomers), and other times they derive from different genetic loci (heteromers, often derived by gene duplication). Commonly, but not always, eukaryotic proteins take the second route, although there is little evidence (if any) that this increase in molecular complexity is driven by adaptive processes. However, as such transitions involve the sustained, coordinated molecular coevolution at the binding interfaces of interacting partners, in the long-run such molecular remodel- ing can passively lead to species reproductive-isolating barriers, as incompatibilities arise between the component parts residing in different lineages.
Regulation of the cell cycle and the stages of mitosis and meiosis involves cel- lular networks that amount to communication pathways involving gene-product in- teractions. Such systems are often endowed with seemingly excessive and arcane structures, the origins of which raise central evolutionary questions in themselves. Equally significant, however, is the repeated observation that even when a network structure remains constant, changes can occur in the underlying participating pro- teins. Recall that other types of cellular systems with highly conserved functions – ribosome biogenesis and division-time determination – encountered in the previous chapter exhibit high levels of divergence of underlying control mechanisms. Such repatternings recall the legendary Ship of Theseus, whereby over time the Atheni- ans gradually replaced every wooden plank, until none of the original components remained, raising the question as to whether the new construction is still equivalent to the original ship. In cell biology, the replacement planks are sometimes not even made from the original materials.
Both homomer-to-heteromer transitions (Chapter 13) and network rewirings (Chapters 19 and 21) can, in some cases, be promoted by nonorthologous gene replacement, but gene duplication appears to have played a significant evolutionary role in the features discussed below. Many of the gene duplications to be discussed arose prior to the last eukaryotic common ancestor (LECA), raising the question as to whether the FECA-to-LECA transition experienced a partial or complete genome duplication.
The Eukaryotic Cell Cycle
Broadly speaking, the life cycle of all eukaryotic cells can be subdivided into three phases, defined with respect to the genomic state: 1) a growth phase in which all cell contents other than the genome expand in number; 2) a period of genome replication in preparation for division (during which growth might continue); and 3) cell fission (cytokinesis) accompanied by transmission of separate genomes to each daughter cell. In practice, however, most cell biologists partition the eukaryotic cell cycle more finely into four or five genome-focused phases (Figure 10.1), the textbook model being: 1), a brief (and sometimes undetectable) G 0 resting phase immediately following division; 2-4), a prolonged interphase, which is further divided into three phases – the G 1 (gap 1) phase during which cell size increases, the S (synthesis) phase during which the genome is replicated, and the G 2 (gap 2) phase during which the cell continues to grow while containing a duplicated genome; and 5)
4 CHAPTER 10
of some basal fungal lineages harbor both E2F and SBF suggests the presence of a redundant regulatory system in basal fungi, with certain sublineages having lost SBF subsequently. Among the well-studied yeasts Saccharomyces, Schizosaccharomyces, and Candida, there are substantial nonorthologies in additional players in the cell cycle and their downstream regulated genes (Cˆote et al. 2009). Even among species within the genus Saccharomyces, differences exist in the interacting proteins at the G 1 /S checkpoint (Drury and Diffley 2009).
Second, there are numerous examples in which the cell-cycle network topology itself has changed. For example, Saccharomyces deploys just a single CDK, whereas metazoan cells deploy at least four, while those of plants use two (Criqui and Gen- schik 2002). Substantial differences in the numbers of cyclins deployed in the cell cycle also exist among phylogenetic groups, e.g., three to four in yeasts, up to ten in plants and metazoans (Criqui and Genschik 2002; Cross and Umen 2015), and as many as two dozen in diatoms (Huysman et al. 2010) and ciliates (Stover and Rice 2011). In many cases, the additional genes have originated via duplication. On the other hand, a broad phylogenetic survey suggests that many of the cell-cycle proteins observed across the eukaryotic phylogeny (and hence by extrapolation were present in LECA) exhibit lineage-specific losses (Medina et al. 2016). For example, Giar- dia intestinalis, a single-celled parasite with two nuclei, has no anaphase-promoting complex and no checkpoint mechanism for mitotic entry (Gourguechon et al. 2013; Markova et al. 2016). Given that the cell-cycle network has been ascertained in only a few model organisms, and even then generally just partially so, many more variants are likely to be found.
A third key observation about the molecular basis of the cell cycle concerns the frequent redundancy in the underlying mechanisms. For example, during the DNA synthesis phase in yeast, at least three simultaneously acting mechanisms prevent secondary replication events, which would lead to chromosomal imbalance in progeny cells. The first of these involves proteolysis of the replication-initiation proteins; the second involves nuclear exclusion of key proteins; and the third involves direct binding at origins of replication. If deleted singly, none of these lead to inviability, implying that the three systems effectively back each other up (Drury and Diffley 2009).
A potential connection between such redundancy and the phylogenetic turnover of cell-cycle participants noted above can be seen as follows (with a more formal presentation appearing in Chapter 20). Imagine three layers of surveillance with error rates e 1 , e 2 , and e 3 , operating in parallel so that the system fails only if all three layers fail to error-correct. The overall failure rate associated with the first mechanism alone is e 1 , with the first and second is e 1 e 2 , and for all three is e 1 e 2 e 3. Because e 1 , e 2 , and e 3 are all < 1 , this shows how multiple surveillance layers can greatly reduce the overall error rate. However, because natural selection operates on the cumulative error rate, E = e 1 e 2 e 3 , and likely can only reduce it to some level defined by the power of random genetic drift (Chapter 8), there will typically be multiple degrees of freedom by which the overall minimum error rate can be achieved (Lynch 2012), i.e., a low value of e 1 can compensate for a high value of e 2 or e 3. This further implies that provided one or two components can together accomplish the target sum E, one (or even two) components are potentially free to be lost in individual lineages. In the long run, this may lead to a phylogenetic repatterning
CELL LIFE CYCLE 5
of the molecular mechanisms underlying a pathway through evolutionary cycles of emergence of redundancy followed by random loss of individual components (Figure 10.3). Of further relevance to the repatterning of the underlying participants in the cell cycle is the observation that many such proteins can have additional cellular functions, including roles in transcription regulation and development in multicel- lular species. Multifunctional genes may be difficult to completely nonfunctionalize over evolutionary time, while still being subject to loss of individual subfunctions (such as participation in the cell cycle, when other redundant mechanisms remain) (Chapter 6). Under the latter scenario, loss of connectivity of a particular gene with the cell cycle in a phylogenetic lineage might be followed by a regain in connectivity at a later point in time. The key point here is that the cell-cycle, one of the most central features of eukaryotic cells, provides a dramatic example of regulatory rewiring underlying a constant cellular attribute. Many more cases of this nature involving other aspects of cell biology will be encountered in subsequent chapters. Although horizontal transfer (as implicated in yeast) can play a role in such evolutionary repatterning, the combination of gene duplication, transient redundancy, and multifunctionality of underlying participants further facilitates the opportunities for rewiring in an effectively neutral manner. Such neutral evolution apparently extends to key amino- acid residues in the final clients of the CDKs themselves, as phosphorylation sites appear to change locations among closely related taxa, while the regional clustering of sites within proteins is generally preserved (Moses et al. 2007; Holt et al. 2009).
Network complexity. It remains unclear why the cell cycle of most eukaryotes engages such a large number of proteins various promoter and/or inhibitor activities (often on the order of 20 or more), but there is no evidence that such massive genomic investment is essential to an ordered cell-cycle progression. All other things being equal, larger networks of proteins potentially impose a greater energetic burden on the cell, while also providing a larger target for mutational disruption. Prokaryotes do not have the elaborate mitotic cycles (below) that are the hallmarks of eukary- otic genome replication, nor do they harbor any obvious orthologs of cyclins and CDKs. However, bacteria do have loosely defined cell cycles governed by simple kinase-receptor systems that enable a central response protein to cyclically dictate progression through growth, replication, and division stages (Biondi et al. 2006; Garcia-Garcia et al. 2016; Osella et al. 2017; Mann and Shapiro 2018). Notably, the cell cycle of fission yeast (S. pombe) can be engineered to run with an extremely simplified control mechanism relying on just one CDK fused to a single cyclin (Coudreuse and Nurse 2010). If nothing else, this demonstrates the feasibility of an ancestral cell cycle driven by something as simple as a single self-oscillating module, and requiring no differential expression, interaction, and degradation of multiple participants. The evolutionary mechanisms leading to the growth of network complexity, and how this can emerge by effectively neutral processes, have been touched upon in Chapter 6. For now, we simply consider an observation from S. cerevisiae, a member of a yeast lineage that experienced an ancestral genome duplication event, possibly resulting from interspecific hybridization (Wolfe et al. 1997; Marcet-Houben and
CELL LIFE CYCLE 7
from each other and move to opposite poles; and 5) telophase, wherein chromosomes decondense into their new nuclear homes and the cell divides. The evolutionary establishment of these sequential stages introduced numerous innovations not generally found in prokaryotes: 1) enclosure of the genome within a nuclear membrane perforated with nuclear-pore complexes to allow mRNA ex- port to the cytoplasm for translation and import of key proteins into the nuclear environment; 2) the expanded use of nucleosomes (octomers involving four unique histone proteins; whereas archaea use tetrameric homomers) for compacting DNA by spooling; 3) expansion in the numbers of origins of replication per chromosome and their parallel firing, which reduces the time for chromosome duplication; 4) cap- ping of linear chromosomes with repeat-based telomeres and devoting an enzyme (telomerase) to their maintenance to prevent end loss; 5) deployment of molecules for sister-chromatid cohesion prior to anaphase; 6) a switch from a membrane-based to a microtubule spindle-based mechanism for segregating sister chromosomes; 7) establishment of centromeres for spindle attachment; and 8) the insertion of mitotic- checkpoint mechanisms to ensure simultaneous and equitable migration of chromo- somes to daughter cells. Although these features are shared by all of today’s eukaryotes, the order in which they appeared in the origin of eukaryotic mitosis remains unknown. Moreover, as in the case of cell-cycle regulation, the molecular and cellular details of many as- pects of eukaryotic mitosis have diverged so much among phylogenetic lineages that it is difficult to even specify the ancestral state of the underlying machinery. Com- parative phylogenetic analysis implies that at least 43 proteins involved in genome replication were present in LECA, only 23 of which are found in all modern lineages and may be indispensable (Aves et al. 2012). The following paragraphs attempt to highlight the diversity of mitotic mechanisms, with minimum attention given to technical details. A number of the proteins involved in eukaryotic mitosis have orthologous copies in archaea, with many of these experiencing gene duplications followed by diver- gence in function in eukaryotes (Aves et al. 2012; Lind˚as and Berlander 2013). To start the discussion, three key complexes involved in the initiation and progression of eukaryotic chromosome replication merit special attention (Figure 10.6): 1) the PCNA (proliferating cell nuclear antigen), a trimeric ring that serves as a clamp to recruit DNA polymerase to single-stranded DNA; 2) the RFC (replication factor, also known as the clamp loader), a pentamer consisting of a chain of four similar sub- units anchored to a larger component, which together endow the DNA polymerase with processivity; and 3) the MCM (minichromosome maintenance complex), a hex- americ ring that unwinds DNA at the replication fork. All three of these complexes exhibit substantial phylogenetic variation in terms of their underlying components (Chia et al. 2010). For example, the trimeric ar- chaeal PCNA can be a homomer (all three subunits encoded by the same locus) or a heteromer constructed from two or three distinct proteins, whereas it appears to be homomeric in eukaryotes. The RFC chain consists of one or two protein types in archaea, whereas each of the four subunits is encoded by a different gene in eukary- otes. The MCM has one to five subunit types in archaea, whereas all six subunits are encoded by different genes in eukaryotes. In all cases, the divergent components are derived by gene duplication of ancestral components, independent from the du-
8 CHAPTER 10
plication events in archaea, all occurring on the path from FECA (first eukaryotic common ancestor) to LECA (Liu et al. 2009). Finally, another complex (GINS) that interacts with the MCM at origins of replication is generally a homotetramer in archaea but a heterotetramer in eukaryotes (Onesti and MacNeill 2013). This collection of observations provides a first illustration of a recurrent theme of multimeric eukaryotic proteins often (but not always) being more complex (in terms of number of gene products involved) than orthologous prokaryotic complexes (Chapter 13). The stochastic coevolution of interface residues among the partners in heteromeric complexes can initiate and sharpen species boundaries, as the gene products from sister taxa diverge to the point of failing to interact (Zamir et al. 2012). A second set of examples involving variation in the complexity of the compo- nents of the mitotic machinery involves the dimeric SMC (structural maintenance of chromosome) proteins, which are ubiquitous across the Tree of Life. All SMC dimers are coiled-coil proteins, with one end of the two members joining to make a flexible hinge, and the other ends providing an opening that can be closed in certain contexts. In bacteria, the molecules are involved in chromosome maintenance and compaction and are homodimeric, generally with just one gene copy per genome. In eukaryotes, the complexes are heterodimeric, and there are multiple copies with more diverse roles: SMC1/3 dimers form cohesins, which hold sister chromatids together during S phase; SMC2/4 dimers are part of the condensin complex, which condenses chromosomes to their metaphase state; and SMC5/6 dimers are recruited in some forms of DNA repair. Although five gene duplications account for the six SMC proteins in eukaryotes, the components of individual complexes are not con- sistently each others’ closest relatives. For example, SMC1 and 4 are sister genes, as are SMC2 and 3 (Cobbe and Heck 2004). Again, the central point is that all three heterodimers were established prior to LECA, reflecting the deep roots of the components of the eukaryotic mitotic machinery. Finally, although most readers will be familiar with a basic textbook version of mitosis (as illustrated in Figure 10.5), as with the broader cell cycle, the mecha- nisms of eukaryotic chromosome segregation have diversified to an enormous extent from the standard model. For example, in most taxa, the DNA-attachment fac- tors (known as kinetochores) assemble onto the centromeres of sister chromosomes, connecting them to long polymeric proteins (the spindle microtubules) that guide chromosomes into daughter cells. Kinetochore complexes consist of ∼ 50 differ- ent proteins, many of which appear to have arisen by duplication (Tromer et al. 2019), but although the structure is thought to be relatively conserved, there are significant differences in component compositions among mammals, insects, and yeasts (Drinnenberg et al. 2016). In the kinetoplastids (which include the para- sitic trypanosomes), kinetochores are constructed out of 19 lineage-specific proteins (Akiyoshi and Gull 2014). Generally, the ends of spindle microtubules are anchored to cytoplasmic centrosomes during cell division, but the centrosome is absent in some groups such as planarians (Azimzadeh et al. 2012) and replaced by a nonhomologous spindle pole body in budding yeast (Winey and Bloom 2012).
The most remarkable and visually obvious forms of variation in mitosis involve the behavior of the nuclear envelope and the locations of the microtubule organizing centers from which the spindles emerge (Heath 1980; Raikov 1982; Sazer et al. 2014).
10 CHAPTER 10
to each other. The subsequent union of haploid gametes (generally from different parents) into diploids creates still more diversity. Almost certainly an evolutionary derivative of mitosis, the establishment of meiosis required four additional innovations (Figure 10.5): 1) physical pairing of ho- mologous diploid-phase chromosomes during first-division prophase; 2) subsequent recombination between homologs (nonsisters) initiated by enzymatically induced double-strand breaks; 3) suppression of sister-chromatid separation (as defined at centromeres) during the first division; and 4) the absence of chromosome replication during the second division. In effect, the conversion of one-step mitosis to two-step meiosis involves the insertion of the first meiotic division (and its associated peculiarities) into the mi- totic cycle (Gerton and Hawley 2005; Wilkins and Holliday 2009). During the first division, the genome is duplicated and rearranged, with the first two daughter cells being effectively genetically haploid (homozygous) for all DNA residing between the centromere and the proximal crossover, but potentially heterozygous for sites distal to the last crossover. For this reason, the first division is referred to as reductional. Cells enter the second meiotic division in the same way as in mitosis, with replicated chromosomes, which leads to complete haploidy for all chromosomal regions and is referred to as the equational division. Although meiosis provides a reliable mechanism for reducing diploidy to hap- loidy (a key requirement for sexual reproduction), an unresolved issue is why meio- sis universally involves two steps. One can readily envision a one-step meiosis in a diploid cell involving nothing more than a single reductional division to two hap- loid products (without any DNA replication), complete with recombination and segregation. Archetti (2004, 2010) has argued that that a population utilizing one- step meiosis would be subject to invasion by mutant cells reproducing asexually, as the latter’s offspring would be genotypically identical to their parent (regardless as to whether crossing-over occurred). Such asexual derivatives might then displace the sexual population, owing to some intrinsic advantages discussed below. With two-step meiosis, invasion by asexuals is arguably less likely because following re- combination in the first division, homologs will be homozygous for all loci proximal to centromeres; this homozygosity would then be inherited by renegade asexual cells forgoing the second division, reducing fitness by exposing the effects of dele- terious recessives. Under this view, one-step meiosis is not impossible, but simply evolutionarily short lived. One significant caveat with respect to this hypothesis is that the predominant life stage (the so-called vegetative phase) of a large fraction of unicellular eukaryotes is haploid. A second issue is that obligate asexuals almost always suffer the fate of progressive, long-term fitness loss, owing to the accumulation of recurrent deleterious mutations, which cannot be purged (Chapter 23). As a consequence, virtually all asexual eukaryotes appear to be evolutionarily short-lived and generally unknown to drive their sexual ancestors extinct (Bell 1982; Lynch et al. 1993).
Origin and evolutionary modifications of meiosis. There is no shortage of speculation on the order of events by which mitotic processes were gradually coopted and modified into the more complex meiotic program (e.g., Maguire 1992; Kleckner 1996; Solari 2002; Egel and Penny 2007; Niklas et al. 2014). As with mitosis, many of
CELL LIFE CYCLE 11
the molecular components of the meiotic machinery appear to have arisen by gene duplication after the divergence of eukaryotes from prokaryotes. One such pair, Rad51/Dmc1, with the respective copies being used in mitotically and meiotically dividing cells, will be further discussed below. Two pairs of proteins involved in mismatch repair and the processing of recombinant molecules resulting from single- strand invasion (Pms1 and Mlh2, and Mlh1 and Mlh3) also arose by duplication (Ramesh et al. 2005). In addition, Spo11 (which, as described below, is involved in creating double-strand breaks during meiosis) gave rise to two new genes by duplications prior to LECA (Malik et al. 2007); unlike Spo11, these duplicates are incapable of religating DNA.
Although meiosis is consistently associated with sexual reproduction and the production of variation in today’s eukaryotes, it need not follow that the earli- est evolutionary steps towards meiosis had anything to do with generating varia- tion. Assuming a haploid ancestral state, as in prokaryotes, the diploid phase may have started as a simple form of endoreplication without cell division. In this case, the subsequent addition of meiotic mechanisms for restoring haploidy would have evolved prior to sexual reproduction and simply involved closed diploid homozygous lineages (Cleveland 1947). An alternative starting point for diploidy would be the fusion of two compatible haploid cells, a necessary condition for sexual reproduction. Without recombina- tion between homologs, restoration of haploidy would only generate variation by independent segregation of chromosomes (in which case, there would be no effect if the ancestral species had a single chromosome). However, FECA likely had a ca- pacity for recombination, as most prokaryotes harbor systems for repairing broken chromosomes off of homologous sequence in another chromosomal copy or segment (Haldenby et al. 2009). For example, after a double-strand break in bacteria, the RecA protein forms a helical filament that coats single-stranded DNA and plays a central role in searching and transferring the strand to homologous double-stranded sequence. As noted in Chapter 9, multiple chromosomal copies are often present in actively growing bacterial cells, and such transfer processes can also occur when bacteria acquire exogenously derived DNA. Eukaryotic Rad51 is related to bacterial RecA, and like the latter, coats single- stranded DNA and guides the initial stages of repair by homology search in mi- totically dividing cells. Notably, the duplicate version of Rad51, called Dmc1, is specifically involved in inter-homolog pairing during meiosis (Ramesh et al. 2005). Thus, the physical mechanism of recombination is highly conserved, having origi- nated prior to the emergence of meiosis, probably for the repair of DNA damage (Bengtsson 1985; Bernstein et al. 1988; Hurst and Nurse 1991). This further im- plies that although meiosis facilitates recombination via double-strand break repair (through homolog pairing), it has never been a requirement for recombination. The initial causal processes leading to the establishment of meiosis may have even been extrinsic to FECA/LECA. For example, Hickey and Rose (1988) sug- gested a scenario in which cell fusion might have been forced upon an ancestral eukaryote by a selfish DNA element as a means for the latter’s transmission among host cells. Without such transmission, a mobile element is essentially confined to a single host-cell lineage, possibly driving the host to extinction by generating dele- terious insertions if overly aggressive or itself succumbing to mutation load if not
CELL LIFE CYCLE 13
are known to be highly sensitive to temperature variation (and for this reason, are often used as temperature sensors in industrial applications). Basic meiotic pro- cesses tend to be highly sensitive to temperature (Bomblies et al. 2015; Lloyd et al. 2018). Within individual species, the temperature range for the faithful operation of meiosis is often only ∼ 5 ◦C, with the optimum temperature varying substantially among taxa. It remains to be determined whether this high level of evolutionary maleability is, in fact, a consequence of sensitivity of the SC components to amino-acid sub- stitutions. However, just within the genus Drosophila, the amino-acid sequences of several of the component proteins of the SC evolve at rates on the order of at least 40% that expected under neutrality (Anderson et al. 2009; Hemmer and Blumenstiel 2016), with a number of sites putatively being under positive selection for change. Bomblies et al. (2015) have suggested that such rapid evolution is a consequence of a coevolutionary dance between interacting partners – with a slight modifica- tion of one member of the pair being met with a compensatory change in another. However, as touched upon in Chapter 6, numerous factors determine whether co- evolution between interacting molecules accelerate vs. decelerate rates of sequence evolution. One potential reason for high rates of evolution of the meiotic machinery involves the relentless selection that must operate on parental chromosomes competing for successful transmission to gametes (Lindholm et al. 2016). Normally, one expects meiosis to give rise to four gametic products, all of which are free to contribute to the next generation. However, post-meiotic conflict or competition can lead to situations in which one allelic type exhibits superiority with respect to another. Pre-meiotic interactions can have similar consequences, e.g., when one parental haplotype some- how prevents the successful inheritance of another into gametes. Examples of such a meiotic-drive process are the spore-killer genes in a number of fungi, which in- crease their relative rates of transmission by killing haploid products that do not contain them (Turner and Perkins 1991; Vogan et al. 2019). Under such scenarios, parental cells will produce fewer than four haploid gametes, but the driving allele could still be brought to high frequency provided its success during meiosis exceeds the reduced production of successful progeny. Moreover, there is one scenario that naturally invites exploitation by driving chromosomes, while incurring no fertility costs. For reasons that remain unclear, numerous phylogenetic groups exhibit a form of meiosis in females in which only one of the four meiotic products matures to a successful gamete, the other three being discarded. Female meiosis has apparently evolved independently multiple times, being present in metazoans, land plants, ciliates, and a number of diatoms (Chepurnov et al. 2004). This condition naturally sets up a situation in which the four meiotic products compete for transmission to the one successful haploid egg. Such an outcome can arise if centromere variants differ in kinetochore-attachment potential in ways that influence successful navigation to a particular location in the final meiotic tetrad (Figure 10.7). For example, an expansion of centromeric repeats leads to larger centromeres, which in principle can attract more kinetochores and spindle microtubules, and centromere location can also have effects (Chm´atal et al. 2014; Iwata-Otsubo et al. 2017; Bracewell et al. 2019). As discussed in Chapter 8, even a 10 −^5 or so fitness advantage (on a scale of 1.0) would be adequate to drive such
14 CHAPTER 10
a centromere to high frequency, and it is this that motivates the centromere-drive hypothesis (Henikoff et al. 2001; Malik and Henikoff 2001). An anticipated by-product of driving centromeres is that, although there may be few side effects in female meiosis other than determining which homolog is pro- moted to the single egg cell, collateral problems may ensue in male meiosis (or even in mitosis), where there is an expected balanced outcome of cell division. This might then impose counter-selection for modifiers that restore normal meiotic segregation, driving the evolution of other genes involved in meiosis (Figure 10.7). In principle, once a such a modifier is driven to fixation, a new opportunity might arise for an- other driving chromosome of a different nature to emerge, encouraging still another opportunity for the establishment of suppressor mutations. Although the centromere-drive hypothesis provides a potentially simple expla- nation for the rapid evolution of the meiotic machinery, there are several reasons for caution in accepting the validity of the verbal model. First, the key requirement for a coevolutionary drive process is the maintenance of functionally significant polymor- phisms in centromeric regions for a sufficiently long time to enable the centromeric proteins to respond by counter-selection. This is because driving chromosomes only exhibit effects when in heterozygous individuals. If a highly aggressive centromere rapidly achieves high frequency, this will thwart the selective promotion of mod- ifier mutations, as homozygotes for driving centromeres will no longer experience problems with meiotic imbalance. Likewise, if the deleterious effects of a driving centromere on male fitness sufficiently exceed the power of the drive process, the driving centromere will simply be eliminated from the population too rapidly for the arrival of modifier mutations. Although small population size might facilitate stochastic increases in the frequencies of mildly deleterious centromeres, this will also inhibit the population-level rate of mutational origin of modifiers and the ability of natural selection to promote them. A second major concern with the centromere-drive hypothesis is that cen- tromeric proteins must recognize the full set of centromere sequences across all chromosomes – there are no known chromosome-specific centromeric proteins. This means that in order to be successful, any modifier that restores parity at the prob- lematical chromosome would have to do so without generating new difficulties else- where. These caveats raise the concern that the population-genetic environments under which centromere drive (or any other meiotic-drive like process) can lead to long-term acceleration in rates of evolution of the participating genes are rather limited. Finally, whereas the numerous examples noted above suggest a high rate of molecular evolution and turnover of components of the meiotic machinery, it is not entirely clear how remarkable such evolution really is, or whether rapid evolution is actually a consequence of driving centromeres. Of particular interest is the cen- tromeric variant of the histone H3 protein found in nucleosomes (CENP-A), which interacts with kinetochores and has been argued to evolve at an exceptionally high rate (higher than the neutral expectation) in Drosophila (Zwick et al. 1999; Malik and Henikoff 2001). In contrast, very distantly related plant species are able to accept transformations of CENP-A from each other, implying a relatively low level of functional divergence (Rosin and Mellone 2017). Moreover, the yeast Saccha- romyces, which does not have female meiosis, nonetheless exhibits relatively high
16 CHAPTER 10
version if the conjoined single strands are not completely complementary. This occurs as mismatches in the heteroduplex are restored to Watson-Crick (i.e., A:T or G:C) status by the mismatch-repair pathway (Chapter 8), leaving a simple patch of sequence exchange in either the recipient or donor strand. Typically, no more than one double-strand break is resolved as a crossover per chromosome arm during an individual meiotic event (Chapter 4). However, the numbers of noncrossover events are tens to hundreds of times higher (De Muyt et al. 2009; de Massy 2013), and these transiently conjoined chromosomal regions keep the parental chromosomes in parallel during metaphase I. Meiosis requires the use of a meiosis-specific protein in the hinge region of cohesins that hold sister chromatids together (noted above for mitosis), whose removal distal to a crossover allows homologs to separate at meiosis I. Maintenance of cohesion proximal to the centromere keeps sisters joined until meiosis II (Watanabe 2012). Notably, the ciliate Tetrahymena, appears to utilize the same hinge during mitosis and meiosis (Howard-Till et al. 2013), and other species may use an entirely unrelated protein (Watanabe 2005). As discussed in the following section, the typical adaptive view of homolog pairing is that such juxtaposition helps ensure a steady supply of recombinant chro- mosomes, providing useful variation upon which natural selection can act. However, as noted above, it is by no means clear that such an effect drove the origin of meiotic pairing. Moreover, the near-constancy of approximately one crossover per chromo- some arm among all eukaryotes (regardless of chromosome size; Chapter 4) does not inspire confidence in the idea that natural selection favors crossing over. Rather, it raises the possibility that selection reduces the latter to a near absolute mini- mum. A more structural view is that homolog pairing provides a powerful means for minimizing the likelihood of nonhomologous recombination, which would lead to deleterious ectopic insertions, deletions, and chromosomal rearrangements (Wilkins and Holliday 2009). In conjunction with sister-chromatid cohesion (as in mitosis), the physical linkage between homologs also ensures proper orientation of all four chromatids on the spindle in meiosis I, minimizing the chances of aneuploidy. All this being said, however, meiosis is far from a perfect process. For example, in hu- mans on the order of 25% of female meiotic products are aneuploid as a consequence of inappropriate resolution of crossover events (Wang et al. 2017), and separation of sister rather than homologous chromosomes at meiosis I is not uncommon (Ot- tolini et al. 2015). With 23 chromosomes per human genome, this implies an error rate of ∼ 1% per chromosome. The rate of nondisjunction for the X chromosome in Drosophila is estimated to be ' 0 .5% (Zeng et al. 2010).
The evolutionary consequences of sexual reproduction. The myriad of novel cellular features in LECA, as compared to its prokaryotic relatives, opened up nu- merous avenues for further evolutionary elaborations by descent with modification. However, the onset of meiosis was unique in that it dictated a new mechanism for the inheritance of the genetic machinery itself, potentially defining new paths by which general evolutionary genetic processes could proceed. Meiosis combined with conjugation (syngamy) forms the basis of sexual reproduction in eukaryotes. Except in the case of self-fertilization, it ensures that progeny genomes are mixtures of those from two individual parents. This generation of variation my be viewed as beneficial
CELL LIFE CYCLE 17
from the standpoint of natural selection, but it also means that, once obtained, an optimal parental genotype will generally not be perfectly transmitted to offspring. Sexual reproduction is also costly in other ways: cells of different mating types must locate each other and then fuse; cell fusion provides a vehicle for pathogen trans- mission; and in species with separate sexes, females often contribute the bulk of the energetic investment in offspring.
Asexual organisms pay none of these costs. Moreover, obligate asexuality does not appear to be difficult to evolve from sexual reproduction, especially in organ- isms with mixed life cycles where phases of clonal reproduction alternate with sex- ual episodes (the usual situation for unicellular eukaryotes). Nonetheless, although nearly every major phylogenetic group of eukaryotes harbors at least one obligately asexual lineage (Bell 1982), obligate asexuality is rare. This view is largely derived from observations of multicellular organisms, where the mating system is fixed and easily observed relative to the situation in microbes, where induction of the sexual phase often goes ignored. Nonetheless, based on this observation most evolutionary biologists assume that an intrinsic advantage of sexual reproduction – enough to off- set the significant disadvantages just noted – must be associated with the production of genetically variable offspring.
There are numerous ways in which the production of variable offspring by mei- otic segregation and recombination might be advantageous (Maynard Smith 1971; Williams 1975; Kondrashov 1993; Barton and Charlesworth 1998). For example, outcrossing provides a potentially powerful means for promoting beneficial combi- nations of alleles from different genetic loci – instead of waiting for two comple- mentary mutations to sequentially arise in a single asexual lineage, single mutations contained within two different lineages can be combined, potentially reducing the waiting time for the emergence of a complex adaptation (Chapter 6). In addition, sexual reproduction can facilitate the purging of deleterious alleles (which constitute the bulk of spontaneously arising mutations; Chapter 4) – recombination and segre- gation produces not only superior genotypes but also a fraction of offspring loaded with an excess of deleterious mutations, and this expansion in the range of varia- tion provides a more efficient route to reducing harmful mutation load by natural selection.
One concern with all of these arguments for the evolutionary maintenance of sex is their dependence on group-selection arguments – the inferred advantages are viewed through the long-term lens of the population, whereas selection at the individual level is much more efficient. This raises the question as to why, once established in a species with a predominantly diploid phase, sexual reproduction would be resistant to invasion by derived asexuals. One potential mechanism is purely genetic. With no known exceptions, diploid cells are still capable of mitotic recombination, and use this capacity to repair double-strand breaks off a homolog. Because recombination generates local patches of homozygosity via gene conversion, purely asexual lineages can be expected to experience progressive loss of heterozy- gosity, and hence a relentlessly increasing exposure of deleterious recessive alleles carried in the original founder of the asexual lineage (as well as those subsequently arising), eventually leading to extinction. Thus, the capacity for complementation after each round of outcrossing may be a primary factor favoring at least periodic sexual reproduction (Archetti 2004, 2005), although this argument seems of minor
CELL LIFE CYCLE 19
Mating-type determination. The mere existence of mating types raises an evo- lutionary challenge, as any particular individual can only mate with a fraction of the members of the population. The situation is most extreme in the case of two mating types, where only half of the population is available (assuming a 1:1 sex ratio). Yet most sexual eukaryotic species have two self-incompatible mating types. As described more below, most eukaryotic species utilize chemical pheromones to facilitate syngamy with appropriate individuals. Chemical recognition does not impose an absolute need for mating types, as all members of the population could in principle encode for the same signal and receptor proteins, a mutual recognition system defined as bipolar (Figure 10.8). However, an obvious limitation of such a system is the potential for an individual’s receptors to be overwhelmed by its own pheromone molecules, removing the chemical gradients necessary to localize other members of the population. This might then endow a selective advantage to a genotype that loses the ability to either signal. In a sea of bipolar cells, a mutant cell defective for pheromone production might also gain a selective advantage owing to the absence of expenditure on biosynthesis of the attractant. In this sense, a bipolar recognition system is expected to be vulnerable to the emergence of a unipolar system (two unique mating types) by subfunctionalization (Chapter 6) – with one cell lineage retaining the signal-producing gene but losing the receptor, and vice versa for the second cell lineage. Once established, such a system might then be further refined by secondary novel gene acquisitions such that both mating types produce unique pheromones and receptors (Figure 10.8). Maintenance of a unipolar mating system requires that the receptor and signal genes be tightly linked chromosomally, as recombination would assort inappropriate mixes into the same gamete, thereby leading to nonfunctional mating capacities (Nei 1969; Hoekstra 1980). Unfortunately, biology’s descriptive language for mating systems is nonstan- dardized, with different terms often used for functionally equivalent systems in fungi (and other unicellular species), land plants, and animals. Homothallism, equivalent to self-compatibility, refers to situations in which specific genotypes are capable of mating with other members of the same genotype. Heterothallism refers to self- incompatible systems requiring separate mating types (in land plants, systems with separate sexes are denoted as dioecious). These terms get blurred in organisms such as some yeast with internal mechanisms for switching mating types through genetic modifications; such species are homothallic, but could also be termed sequential hermaphrodites. In multicellular organisms, simultaneous hermaphroditism is pos- sible, as in monoecious plants in which individuals produce male and female floral parts, but this is not known for unicellular species. Finally, the terms isogamous and anisogamous are used to refer to situations in which gamete types are mor- phologically indistinguishable vs. distinct (as in eggs and sperm in land plants and animals); even these terms can be a bit misleading, as isogamous species generally have different mating types, which although morphologically identical, have under- lying molecular differences. A broader array of mating systems has been described in the fungi than in any other major eukaryotic lineage, although this could be a simple consequence of the magnitude of research focused on fungi relative to other systems. In S. cere- visiae, S. pombe, and several other yeasts, there are two distinct mating types,
20 CHAPTER 10
each with unique pheromones and receptors, but these are achieved by mating- type switching (Hanson and Wolfe 2017), whereby casettes of genes are swapped into a particular site by recombination. In this sense, mating-type determination involves a single tightly linked region (multigenic, but effectively segregating as a single locus) – individual genotypes are genetically hermaphrodites, but at the phe- noytpic level, individual cells mate in a unipolar manner. Once two complementary types are attracted to each other, the production of mating-type specific agglutinins (coagulants) is induced, and heterodimeric transcription factors constructed from components from each pair member govern the progression to downstream meiotic activities.
In the smuts, a group of plant pathogens within the mushroom family, the mating system sometimes involves two unlinked loci, although again each locus actually consists of linked blocks of genes (Bakkeren et al. 2008). In this case, one locus typically encodes for linked pheromones and receptors, while the second encodes for a dimeric transcription factor that governs downstream cellular events associated with syngamy and meiosis. As in the yeasts, different mating types recognize different pheromones, but in smuts four possible outcomes are possible, from fully compatible to fully incompatible, depending upon the allelic status at the two loci. To further emphasize the diversity of evolved systems, just a few other exam- ples of unicellular mating systems will be noted here. Diatoms are known for their diversity of mating systems and sometimes rapid rates of evolution of underlying components (Armbrust and Galindo 2001; Chepurnov et al. 2004). Some diatoms have homothallic mating systems (capable of selfing), whereas others are heterothal- lic, and among these are cases of both isogamy and anisogamy. The diatom Semi- navis has two mating types whose activities are coordinated by a two-step signaling system, the first involving a chemo-attractant that acts on a global basis, and the second operating only after the perception of a mating partner and stimulating entry into cell-cycle arrest and gametogenesis (Moeys et al. 2016). Not all species use mate-attraction pheromones. Although some ciliates, such as Euplotes (below) do use pheromones, others such as Paramecium simply deploy mating-type specific agglutinins upon contact. In Paramecium tetraurelia, which has a transcriptionally silent germline nucleus (the micronucleus) and a “somatic” macronucleus (Figure 10.9), epigenetic events are involved in the maintenance of the two mating types (E and O, for even and odd), such that the latter are determined entirely by the maternal cytoplasm – in E-type cells, the mating-type gene (mtA, residing in the micronucleus) is passed on intact to the macronucleus, whereas in O-type cells, the promoter region is spliced out, rendering the macronuclear variant nonfunctional (Singh et al. 2014). In the related species, P. septaurelia, mtA is not differentially processed, but instead another gene (mtB, a transcription factor that regulates mtA) experiences a nonfunctionalizing deletion in the macronucleus of O- type cells. Thus, even members of the same genus can have substantially different mechanisms of mating-type determination. In Chlamydomonas, two mating types (+ and −) produce unique agglutinins on their flagella, which cross-react as recognition and adhesion mechanisms, leading to a cascade of events including the formation of a heterodimeric transcription factor composed of subunits derived from each mating type (Goodenough et al. 2007). The
