Driving lessons: a brief (personal) history of centromere drive

“Individuals are not stable things, they are fleeting. Chromosomes too are shuffled into oblivion, like hands of cards soon after they are dealt. But the cards themselves survive the shuffling. The cards are the genes. The genes are not destroyed by crossing-over, they merely change partners and march on. Of course they march on. That is their business.”

      –Richard Dawkins, The Selfish Gene (Dawkins 1976)

Meiosis is an important specialized cell division in many eukaryotic species, including fungi, plants, and animals. Meiosis results in the production of haploid gametes starting from a diploid cell via 1 round of replication and 2 rounds of cell division. In an influential article published in 1957, Sandler and Novitski first pointed out that meiosis is also an intense battleground, in which gametes vie for evolutionary supremacy with each other, often poisoning their competition to gain a fratricidal advantage (Sandler and Novitski 1957). This competition, which they termed “meiotic drive,” operates as an evolutionary force that can cause an increase in frequency of the allele that is favored during meiotic transmission. Unlike alleles that rise in frequency because they confer a fitness advantage to their carriers, meiotic drivers can rise in frequency even while conferring significant fitness disadvantages on their carriers. Thus, meiotic drivers can be viewed as the quintessential selfish genes; it is the best interest of the rest of the genome to counteract their action to restore organismal fitness.

With their proposal and subsequent work, Sandler and Novitski laid the foundations of the meiotic drive field. Thus, it is a special honor for me to receive the 2022 Novitski Prize for postulating that essential components of the cell division apparatus have been shaped by relentless evolutionary battles during meiosis. This essay is not intended to be a comprehensive summary of meiotic drive mechanisms but rather a personal summary of both the preceding discoveries that inspired our hypothesis and the subsequent studies that have validated and extended its key postulates. I have tried to be as historically accurate as possible while relying on an unreliable memory.

In their simplest forms, meiotic drivers can be classified into 2 types (Sandler and Novitski 1957; Burt and Trivers 2006; McLaughlin and Malik 2017). The first of these acts during or after male meiosis, which normally produces 4 haploid gametes (e.g. pollen in plants, spores in fungi, sperm in animals). Meiotic drive alleles typically act by encoding toxins that prevent the survival or maturation of chromosomes that do not carry them. They protect themselves either by encoding antidotes that prevent their own destruction or by specifically targeting chromosomal “susceptibility” loci that are only carried by competing chromosomes (McLaughlin and Malik 2017). Many independent meiotic drive genes with distinct mechanisms of toxicity have been discovered in fungi, flies, plants, and mice (Olds-Clarke and Johnson 1993; Larracuente and Presgraves 2012; Lindholm et al. 2016; Saupe and Johannesson 2022). In some cases, meiotic drivers arise as neomorphic alleles that target chromosome condensation or sperm flagella functions (Merrill et al. 1999; Herrmann et al. 1999). However, the evolutionary origins of most meiotic drive genes or their toxicity mechanisms are not well understood.

The second form of meiotic drive occurs in female meiosis. Unlike male meiosis, female meiosis in plants and animals is asymmetric, i.e. it produces only 1 haploid gamete per meiosis that is passed onto future generations, whereas the other 3 gametes are “evolutionary dead ends” (e.g. polar bodies in animals). This introduces a different type of competition during female meiosis. In contrast to the rampant fratricide mediated by male meiotic drivers, female meiotic drivers simply must ensure a higher-than-random probability of inclusion in the next generation. Thus, female meiotic drive incurs a much less significant cost to their carriers in terms of gametic loss. However, this also makes female meiotic drive much harder to detect even in laboratory experiments, except for large-scale experiments designed to detect deviation from Mendelian inheritance patterns.

Female meiotic drivers originally emerged from studies of chromosomal elements that were clearly cytologically visible. The first of these are “knobs,” which are blocks of satellite DNA repeat-rich heterochromatin found distal to maize centromere by Barbara McClintock. Marcus Rhoades discovered that knobs preferentially segregate in a pattern that favors their own transmission during female, but not male, meiosis (Rhoades 1942; Rhoades and Dempsey 1966). Rather fortuitously, the knob elements were genetically linked to an anthocyanin pigment marker, which allowed for visual scoring of a large number of progeny derived from heterozygotes to detect this non-Mendelian segregation (Birchler et al. 2003; Birchler 2016; Dawe 2022). Subsequent studies over many decades culminated in the discovery that this preferential segregation arises due to its neocentromere activity mediated by recruitment of specialized cytoskeletal motors which bind satellite repeats in knob heterochromatin and preferentially orients it during meiosis II (Dawe et al. 2018; Swentowsky et al. 2020).

A second significant discovery on the cell biological basis of female meiotic drive emerged from studies in the grasshopper Myrmeleotettix maculatus, which carries a variable number of supernumerary B chromosomes that are readily visible cytologically (Hewitt 1976). Although transmission of B chromosomal number followed Mendelian expectation through male meiosis, there was a preferential increase in B chromosomes through transmission via female meiosis. Using cytological observations of primary oocytes, Hewitt (1976) discovered that unlike regular chromosomes, B chromosomes were preferentially misaligned on the metaphase plate, orienting preferentially toward the egg side, which also correlated with their preferential non-Mendelian inheritance of B chromosomes through female meiosis (Hewitt 1976). This study provided the first direct link between cytological asymmetry of the oocyte spindle and female meiotic drive.

These and other emerging discoveries were synthesized in an influential article by Pardo-Manuel de Villena and Sapienza (2001a, 2001b, 2001c), who identified 3 basic requirements for nonrandom segregation: (1) asymmetrical meiotic division(s); (2) functional asymmetry of the meiotic spindle poles; and (3) functional heterozygosity at a locus that mediates attachment of a chromosome to the spindle (Pardo-Manuel de Villena and Sapienza 2001b). They further identified that Robertsonian fusion chromosomes displayed a segregation advantage in female but not male meiosis (Pardo-Manuel de Villena and Sapienza 2001c, 2001d). Robertsonian chromosomes arise when 2 acrocentric chromosomes (which have centromeres toward 1 end) fuse at those centromeres to form a single metacentric chromosome (which has a centromere in the middle). Intriguingly, this segregation bias was not universal; whereas Robertsonian chromosomes were preferentially inherited in humans and chickens, they were selected against in mice. Indeed, most mammalian karyotypes are closer to mostly metacentric or mostly acrocentric, likely reflecting the change in segregation bias toward or away from the preferential inheritance of Robertsonian chromosomes (Pardo-Manuel de Villena and Sapienza 2001a). These studies were especially instructive to us as we contemporaneously considered centromeres acting as selfish female meiotic drivers.

Most of the previous work on female meiotic drive had been focused on the chromosomes or chromosomal loci that acted as meiotic drivers during female meiosis. However, an influential study by Zwick et al. (1999) focused attention on chromosomal proteins that may bind and mediate the biased chromosome segregation required for meiotic drive. This study arose from an observation that 2 natural populations of Drosophila melanogaster show dramatically different levels of nondisjunction frequency among X chromosomes. This led the authors to discover that 2 alleles of the nod gene, which encodes a chromokinesin or cytoskeletal motor protein required for chromosome segregation in female meiosis, are found at intermediate frequencies across many populations of this species. They argued that this high frequency was the likely result of ootid competition, where different chromosomes had competed for inclusion into the egg cell. Such competition could simultaneously explain both the high levels of nondisjunction as well as the high frequency of variants of chromosomal proteins like NOD being maintained in natural populations.

I am especially grateful to the Zwick and Langley study because I read this article on the flight to my interview for a postdoctoral position in the lab of Steve Henikoff, ostensibly to try and work on remote homology detection. Two things happened during my initial meeting with Steve. First, he told me that he was discontinuing his lab’s work on protein homology detection. Second, he told me about his lab’s work on centromeric H3 (cenH3) histones and showed me a tree that suggested they were more rapidly evolving than canonical H3 histones (Henikoff et al. 2000). We both realized that this could either be the result of relaxed selection (after all, canonical histones are among the most slowly evolving proteins) or positive selection. Jumping ahead to consider the second possibility, we both realized that the essential cenH3 (called Cid in Drosophila) would be expected to be subject to the same evolutionary pressures as the nod-encoded chromokinesin. My disappointment at the unexpected loss of my proposed postdoctoral project evaporated in my excitement at the prospect of studying centromere biology in the Henikoff lab. I still remember the nearly three-hour-long dinner conversation with Steve and Jorja Henikoff; we fleshed out the broad strokes of the (initially wrong but eventually right) centromere-drive model at a Middle Eastern restaurant, and we were only interrupted by the proprietor informing us that we had to leave because they were closing the restaurant for the night!

Indeed, using sequencing of multiple strains of Drosophila melanogaster and its close relative D. simulans, we were able to show a clear signature of positive selection in the essential Cid gene, one of the first essential genes shown to be subject to such strong evolutionary constraints (Malik and Henikoff 2001). Realizing that the Cid gene was not closely linked to any centromeres, we hypothesized that Cid evolution must therefore be acting to oppose rather than support the preferential inheritance of “selfish” centromeres (Hartl 1975), which are in the perfect position to subvert the first meiotic division to orient themselves to a preferred position. We hypothesized a 2-step model, in which centromeric DNA underwent changes to over-recruit centromeric proteins and improve its own transmission during female meiosis, followed by cenH3 and other centromeric proteins evolving to suppress this “selfish” behavior or deleterious consequences associated with this centromere drive (Henikoff et al. 2001; Henikoff and Malik 2002). The heretical nature of this idea was that 2 essential components of the chromosome segregation apparatus—centromeres and centromeric proteins—were combatants because of meiotic drive, even though they are normally collaborators that ensure high fidelity segregation. Despite being controversial, this idea elegantly explained why satellite repeat-enriched centromeric DNA was among the most rapidly evolving component of most animal and plant genomes (Henikoff et al. 2001). Further validation of the idea came from evolutionary analyses of cenH3 and other centromeric proteins in a variety of taxa, finding rapid evolution of many centromeric proteins in taxa that are subject to female and male meiosis (e.g. animals, plants) (Talbert et al. 2004; Schueler et al. 2010), but not in taxa with only male meiosis (e.g. fungi like Saccharomyces cerevisiae) (Baker and Rogers 2006) or only female meiosis (e.g. ciliated protozoans like Tetrahymena thermophila) (Elde et al. 2011).

Shortly afterwards, elegant studies in Mimulus monkeyflowers provided a satisfying validation of the centromere-drive model. First, Lila Fishman and John Willis showed compelling evidence of ∼100% female meiotic drive in interspecies crosses between Mimulus nasutus and Mimulus guttatus (Fishman and Willis 2005). The magnitude of this drive strongly implicated a centromere-mediated drive mechanism that occurred in the first meiotic division since drive in the second meiotic division can never exceed 83% (Fishman and Willis 2005; Malik 2005). Subsequent studies showed that the driving centromere locus had undergone a substantial expansion of a centromeric satellite, consistent with the first step of the centromere-drive model (Fishman and Saunders 2008). Finally, more recent work implicated that a novel variant of cenH3 might be associated with suppressing drive, providing an elegant demonstration of the second step of the centromere-drive model (Finseth et al. 2021).

Use of Robertsonian chromosome polymorphisms in mice provided the most insight into the molecular and cell biological basis of centromere drive. Studies in the labs of Michael Lampson, Richard Schultz, and Ben Black showed that centromeric satellite DNA expansions recruit more centromeric proteins (Chmátal et al. 2014) and this endows them with greater success in female meiosis (Iwata-Otsubo et al. 2017). Moreover, cell biological studies in mouse oocytes showed that “cheating” centromeres exploit an intrinsic asymmetry of the spindle cytoskeletal apparatus in female meiosis (Akera et al. 2017), like the original postulate by Godfrey Hewitt in grasshoppers (Hewitt 1976). Centromeric success was shown to be the direct result of recruitment of microtubule-destabilizing proteins like mitotic centromere-associated kinesin (Akera et al. 2019). Preferential recruitment of microtubule-destabilizing proteins leads to the detachment of selfish centromeres from spindle microtubules that would otherwise direct them to the polar body. This attachment–detachment dynamic allows selfish centromeres one more turn at the meiotic roulette wheel and ensures a higher-than-Mendelian inheritance of these centromeres to the oocyte (Schroeder and Malik 2019). It is likely that all the preferential recruitment of most centromeric proteins to “winning” centromeres is tied to the recruitment of microtubule-destabilizing proteins that ultimately dictates female meiotic success (Akera et al. 2019). These studies have provided an elegant demonstration of the “cheating” stages of the centromere-drive model.

Although the idea of centromere drive was inspired by the finding of positive selection of cenH3 (Malik and Henikoff 2001), the fitness cost that spurred this positive selection is still unknown. Implicit is the assumption that centromere drive must be deleterious. However, unlike male meiotic drive that leads to gametic dysfunction, it is not immediately obvious what this deleterious consequence would be in the case of centromere drive. Early results suggested that human carriers of Robertsonian fusion chromosomes might be subject to higher levels of male meiotic dysfunction (Daniel 2002). Similarly, work on the driving centromere in Mimulus monkeyflowers suggested that homozygosity of the driving centromere is associated with lower male fertility and seed survival, which can be attributed in large part due to deleterious mutations that have accumulated because of their genetic linkage to the driving centromere (Fishman and Saunders 2008; Fishman and Kelly 2015). An idea has begun to emerge in the field that the most immediate deleterious consequences of centromere drive may not be in female meiosis at all. Indeed, cenH3 replacement experiments in Arabidopsis thaliana show normal meiosis, but an epigenetic mismatch of parental centromeres packaged in different cenH3 proteins lead to catastrophic loss of centromere function and genome elimination during early mitosis (Marimuthu et al. 2021). Understanding these deleterious consequences and how they are suppressed by adaptive evolution of centromeric proteins would elegantly explain the second stage of centromere drive. Several labs, including mine, are actively engaged in these studies.

One of the exciting implications of the centromere-drive model was its potential to explain reproductive isolation, i.e. speciation. The foundation for this idea predates the proposal of centromere drive. Two influential articles (Frank 1991; Hurst and Pomiankowski 1991) had argued that rapid origins and battles between dueling meiotic drivers and genetic suppressors of drive could rapidly trigger reproductive isolation between populations that were previously compatible with each other. We argued that a similar dynamic at the essential centromeric protein–DNA interface would similarly lend itself to the initial stages of postzygotic reproductive isolation (Henikoff et al. 2001; Henikoff and Malik 2002). We reasoned that rapid changes in either satellite DNAs or satellite DNA-binding centromeric or heterochromatin proteins could result in hybrid sterility or inviability and begin to drive a wedge between incipient species. Despite initial resistance to this idea, the identification of heterochromatin proteins or satellite DNAs whose loss leads to rescue of otherwise inviable or sterile hybrids has provided some support to this model (Sawamura et al. 1995; Ting et al. 1998; Barbash et al. 2003; Brideau et al. 2006; Phadnis et al. 2015). Additional work has begun to provide insight into how such protein–DNA incompatibilities might mechanistically lead to sterility or inviability (Bayes and Malik 2009; Ferree and Barbash 2009; Jagannathan and Yamashita 2021). This is likely to be an exciting area of research in the coming years because of advances in genetically manipulating many closely related species.

Lessons from meiotic drive

“Let us try to teach generosity and altruism, because we are born selfish. Let us understand what our own selfish genes are up to, because we may then at least have the chance to upset their designs, something that no other species has ever aspired to do.”

      –Richard Dawkins, The Selfish Gene (Dawkins 1976)

Having a career-long obsession with selfish genes provides a unique perspective of viewing many other aspects of life in science through that same lens. For example, one could make an analogy between scientific fields and an ecosystem of genes vying for evolutionary success. Meiotic drivers bestow upon their descendants the ability to constantly thrive by outcompeting the competition, often by brutal means. This would seem to be an admirable Darwinian trait, worth emulating in real life. Indeed, science funding today can provide similar incentives for increasing lab sizes by writing for more grant funding and attracting more trainees, whose individual success can no longer be guaranteed. The hypersuccess of one research group at the expense of others is often justified as Darwinian selection for “the best science,” but is that really the best outcome for the scientific enterprise at large? Previous studies have shown “diminishing returns” on scientific investment beyond a certain lab size (Alberts 1985; Cook et al. 2015), although there is vociferous disagreement about what this optimal lab size ought to be. Meiotic drive might also present as Darwinian selection for the fittest alleles, but it comes at the expense of organismal or population fitness. Moreover, selfish genes often create the conditions to benefit themselves. Incentivizing highly established investigators to earmark large grants for their own labs or institutions poses a similar risk and will require constant efforts by the larger scientific community to resist such efforts, much like suppressors that must arise in natural populations to curb selfish genes from completely taking over. I have been fortunate to have spent nearly my entire career at the Basic Sciences Division at Fred Hutchinson Cancer Center, which prides itself on its egalitarian model of doing science, with modest-sized labs that do not sacrifice an expectation of high-quality science. Yet this model of science finds itself under constant threat (Alberts 2012).

So, how do we maximize the evolutionary fitness of a scientific field or the scientific enterprise at large? Having benefited from the largesse of many mentors and collaborators, to me the answer is personal and simple: support junior faculty. I can make a simple Darwinian argument for this. Just as people compete within a field, fields compete to attract and retain the best talent. Ultimately, fields that prove to be inhospitable and hypercompetitive will attract fewer and fewer talented junior researchers and drive themselves to the brink of extinction, much like selfish genes often do to host populations. Although we live in an unprecedented era with amazing tools for doing creative science at our disposal, the costs and salaries associated with doing science have risen precipitously. No single group has faced the brunt of these challenges as much as young investigators starting their labs, and these difficulties have only been exacerbated by the effects of the COVID-19 pandemic that unequally stifled momentum of junior labs. It is ultimately in our best self-interest to acknowledge and account for these difficulties when we evaluate and support our junior colleagues, and colleagues in countries where scientific investment has plummeted. Recognizing that all labs have a finite life, it is not a successful evolutionary strategy to throttle the success of our “progeny.”

A final lesson we can learn from meiotic drivers and other selfish genes is that we must celebrate and strive for scientific diversity. Like genetic diversity, scientific diversity can often provide the best defense against selfishness. In an era of ever-dwindling resources, earmarking resources for just few kinds of research or a few organisms because they are viewed as translational and utilitarian, however admirable the goal might appear to be, could be ultimately counterproductive for the scientific enterprise. For example, this short essay recounts studies from many organisms over several decades that each provided 1 critical piece of the jigsaw puzzle of centromere drive. Moreover, the study of genetic elements that subvert meiosis have revealed unprecedented insight into the process of meiosis, just like the study of pathogens has provided the best insight into the cell biology of the host.

The maxim that most scientific discoveries benefiting human health started off as basic science discoveries is only worn out because it happens to be true.