Meiotic Drive in House Mice: Mechanisms, Consequences, and Insights for Human Biology

Abstract

Meiotic drive occurs when one allele at a heterozygous site cheats its way into a disproportionate share of functional gametes, violating Mendel’s law of equal segregation. This genetic conflict typically imposes a fitness cost to individuals, often by disrupting the process of gametogenesis. The evolutionary impact of meiotic drive is substantial, and the phenomenon has been associated with infertility and reproductive isolation in a wide range of organisms. However, cases of meiotic drive in humans remain elusive, a finding that likely reflects the inherent challenges of detecting drive in our species rather than unique features of human genome biology. Here, we make the case that house mice (Mus musculus) present a powerful model system to investigate the mechanisms and consequences of meiotic drive and facilitate translational inferences about the scope and potential mechanisms of drive in humans. We first detail how different house mouse resources have been harnessed to identify cases of meiotic drive and the underlying mechanisms utilized to override Mendel’s rules of inheritance. We then summarize the current state of knowledge of meiotic drive in the mouse genome. We profile known mechanisms leading to transmission bias at several established drive elements. We discuss how a detailed understanding of meiotic drive in mice can steer the search for drive elements in our own species. Lastly, we conclude with a prospective look into how new technologies and molecular tools can help resolve lingering mysteries about the prevalence and mechanisms of selfish DNA transmission in mammals.

Introduction

Mendel’s Law of segregation asserts that alleles at a heterozygous site have equal probability of being transmitted to the next generation. This law provides a useful null model for predicting trends in genetic inheritance. However numerous exceptions to this simple rule have been identified in the natural world. Such cases of transmission ratio distortion (TRD) are characterized by the over-transmission of one allele at the expense of the other. TRD can arise from multiple distinct mechanisms, including genotype-dependent embryonic lethality, gene conversion, and meiotic drive. In this article, we focus on meiotic drive, largely to the exclusion of other types of TRD.

While TRD is fascinating from a basic science perspective, the phenomenon also bears relevance for both evolutionary biology and the genetic mechanisms of infertility. The over-transmission of one allele mimics the action of natural selection and, if unchecked, can lead to the allele’s rapid fixation, even when linked to deleterious variants. Furthermore, the evolution of a selfish element in a population imposes an intense selection pressure for the emergence of a suppressor to restore the allelic transmission ratio to parity. The decoupling of a selfish element from its co-evolved suppressor(s) via outcrossing can expose incompatible allelic interactions that may pose reproductive barriers between individuals. Such phenomena can actively contribute to speciation (Crespi and Nosil 2013) and, potentially, infertility within populations (Zanders and Unckless 2019). Moreover, the co-evolutionary arms race between drive elements and their suppressors may fuel the rapid evolution of genes expressed during gametogenesis in mammals (Malik and Bayes 2006; Helleu et al. 2016; Vedanayagam et al. 2021). Finally, some selfish loci can bias their transmission by killing non-resident gametes, effectively reducing the number of gametes produced and impacting reproductive fitness (Zanders and Unckless 2019).

Rates of infertility and subfertility are alarmingly high in humans, and in many cases, the underlying etiological basis is not known. Given the broad association between meiotic drive and reduced hybrid fertility in nature, it is tempting to speculate that drive may be a meaningful contributor to infertility in our own species (Zanders and Unckless 2019). However, this intriguing prospect has received limited recognition and very few cases of drive have been documented in humans. This latter point may owe to the intrinsic challenge of identifying weak TRD signals in small human families, the confounding effects of genotyping error, the difficulty of distinguishing TRD from more ubiquitous signals of natural selection, and the frequent absence of a clear phenotypic signal to herald the presence of meiotic drive (Meyer et al. 2012). Thus, the paucity of documented cases of meiotic drive in humans could reflect biases in ascertainment and detection, rather than the genuine absence of this phenomenon from our genomes.

While the scope and scale of meiotic drive in humans is difficult to evaluate, the challenges of detecting selfish transmission are far more readily overcome in experimental model systems. Here, we dedicate special focus to the house mouse (Mus musculus), the premiere biomedical model system. Hundreds of inbred mouse strains are commercially available, enabling the study of genetically controlled crosses with balanced allele frequencies that maximize power to find meiotic drive. In addition, purpose-bred mouse populations often harbor modest numbers of accumulated crossover events, ensuring that any drive elements present reside on large haplotypes with many linked markers that share the expected signal of biased transmission. Laboratory mouse populations can also be precisely controlled to minimize the potential confounding effects of natural selection, population structure, and environmental factors, further improving detection power. House mice are also equipped with a rich toolbox of cutting-edge technologies and resources that set the stage for functional investigations into how drive elements cheat the rules of Mendelian inheritance. Beyond these technical advantages, the broad convergence of human and mouse biology foreshadows consilience in the mechanisms and genomic impact of TRD in both species (Wang et al. 2016; Mungall et al. 2017; Saul et al. 2019).

In this article, we outline how different house mouse resources — from inbred strains to complex mapping populations — can be used to identify and characterize loci that defy Mendelian segregation. We then profile several paradigms for meiotic drive in the mouse genome, detailing current molecular models for drive action. We then suggest approaches for leveraging insights about meiotic drive in mice to make inferences about the extent of this phenomenon in the human genome. We end with a forward look into how new sequencing technologies and tools for genome editing stand to transform our understanding of the prevalence and mechanisms of meiotic drive in mammals.

Meiotic Drive: Definition and General Principles

Meiotic drive is defined by the overrepresentation of one allele among functional gametes. Consequently, drive elements are universally expressed in the germline where they either perturb the normally unbiased process of chromosome transmission into gametes or fuel competition between gametes with distinct genotypes. Owing to sex differences in the developmental mechanisms of gamete formation, meiotic drive manifests via distinct mechanisms in mammalian males and females. Meiotic drive loci that are active in the female germline function by hijacking the inherent asymmetry of oogenesis, in which one viable oocyte emerges from a single round of meiosis. Thus, selfish female drive elements vie for representation in the oocyte (and exclusion from the residual polar bodies) (Pardo-Manuel de Villena and Sapienza 2001a). In males, drive elements typically function by immobilizing or killing non-self sperm, ensuring their own fertilization advantage (Lindholm et al. 2016). Within a species, a given drive element generally functions in males or females, but not in both sexes.

We note that some authors employ a stricter definition of meiotic drive that only applies to the molecular manipulation of chromosome segregation during asymmetric female meiosis (Didion et al. 2015; Wei et al. 2017). Throughout this review, we adopt a less stringent definition of meiotic drive that also encompasses mechanisms for biased transmission of male gametes. Such loci are also often referred to as “segregation distorters”.

Meiotic drive loci employ diverse mechanisms to cheat meiosis, but share several features in common. Most meiotic drive systems are composed of at least two elements: a responder locus and one or more distorters. When heterozygous, a responder locus exhibits unequal transmission of one allele at the expense of the other. Distorters (which are sometimes alternatively referred to as drivers) promote transmission bias at the responder. Responders and their associated distorter(s) are always genetically linked, ensuring the evolutionary success of the drive system. Throughout, we use the term “drive element” to refer to a single responder and its linked distorter(s). The strength of distortion can be modified through the action of linked enhancers that amplify the strength of the distorter(s) or unlinked suppressors that act to restore Mendelian segregation (Crow 1991). Thus, a given drive element may be widespread across populations of a given species, but due to the segregation of multiple suppressors, biased transmission may only manifest in certain individuals with permissive genetic backgrounds.

Meiotic drive elements tend to reside in structurally complex loci, including copy number variable loci, inversions, and satellite-rich regions. While satellites were historically regarded as “junk DNA”, work in the last decade has highlighted their important roles in heterochromatin establishment, dosage compensation, reproductive isolation, genome stability, development, and chromosome segregation (Biscotti et al. 2015; Shatskikh et al. 2020). Indeed, two of the most essential loci for chromosome segregation — centromeres and telomeres — are satellite-rich chromatin domains. Centromeres orchestrate the assembly of the kinetochore complex, which in turn tethers chromosomes to the microtubules of actively dividing cells (Cheeseman and Desai 2008). Mammalian telomeres are composed of a 6-mer satellite repeat unit and play essential roles in directing the chromosome movements that culminate in homologous chromosome pairing during meiosis (Saint-Leandre and Levine 2020). Given their specialized biological functions in chromosome segregation, both telomeres and centromeres are ripe for exploitation by selfish drive elements (Zwick et al. 1999). Further, the intrinsically high mutation rate of satellite DNA can lead to rapid expansion/contraction in satellite array sizes and high rates of satellite sequence turnover (Lower et al. 2018). In turn, these phenomena may allow distorters to rapidly evade detection by their suppressor(s), creating intense selection pressures for rapid, cognate changes in the suppressor(s) to silence the distorter and restore a balanced transmission ratio.

Drive elements are also enriched in genomic regions with low recombination. Selfish drive systems that emerge in low recombination regions face lower likelihood of demise due to recombination between the responder and distorter(s) (Schwander et al. 2014). At the same time, satellites and structurally complex loci are hotspots for the accumulation of drive elements and are broadly characterized by low recombination rates (George and Alani 2012). Furthermore, a drive element may induce selection for reduced recombination to maintain linkage with drive enhancers (Charlesworth and Hartl 1978). Thus, it remains unclear whether the association between meiotic drive systems and regions of low recombination rates is causal or consequential. Regardless, due to their presence in regions of reduced recombination, drive systems often serve as hubs for the emergence of hitchhiking deleterious alleles.

This latter consideration implies that meiotic drive elements should often be associated with low fitness alleles that impose evolutionary costs. Indeed, virtually all meiotic drive elements discovered to date are associated with reduced fertility or sterility in hybrids (Fishman and Saunders 2008; Didion et al. 2015; Zhang et al. 2015; Zanders and Unckless 2019). By subverting the rules of Mendelian transmission, selfish drive elements can overwhelm the removal of linked deleterious alleles by natural selection, allowing such alleles to persist within populations (Hartl 1970; Schimenti et al. 2005). In other instances, the very mechanism of meiotic drive may impose fitness costs. For example, meiotic drive rendered via targeted killing or immobilization of gametes from one genotype class will result in the production of fewer functional gametes.

Leveraging Mouse Strain Resources to Identify Meiotic Drive Loci

As the premiere biomedical mammalian model system, house mice are equipped with numerous strain resources and genetic tools that can be harnessed to (i) identify loci that are selfishly transmitted and (ii) dissect the molecular details of how these loci break the usual rules of genetic transmission. In the next subsections, we outline how different mouse strain resources — from inbred strains to complex mapping populations — have led to the discovery of meiotic drive elements. We also suggest novel approaches for using these resources that could spur the discovery of new loci subject to TRD.

Inbred strains:

For drive to manifest, there must be heterozygosity at the underlying responder locus, such that two alleles can compete for transmission to the next generation. Many drive responders are rapidly evolving due to the instability of their underlying repeat-rich sequences (Henikoff et al. 2001; Larracuente 2014). This rapid evolution may give rise to appreciable rates of heterozygosity within inbred strains, enabling drive to manifest. Thus, the assumption that inbred strains are static, immortal, and reproducible genetic resources may not be valid, especially at rapidly evolving selfish elements (Chesler et al. 2016; Chebib et al. 2021). This point, coupled with standard maintenance of inbred strains via strict brother-sister matings, could set the stage for the rapid fixation of newly emerged variants that confer a transmission advantage. Indeed, we have previously shown that mouse inbred strain centromeres are markedly larger than those of wild-caught mice (Arora et al. 2021). These findings could suggest that reduced selection against weakly deleterious centromere-associated drive elements in bottlenecked inbred laboratory populations has promoted their rapid fixation (whereas purging of centromere-associated drive elements in outbred wild populations may be more efficient). Periodic genomic surveillance of inbred strain genomes and/or comparisons between closely related sub-strains of mice could facilitate the discovery of such rapidly changing loci and motivate detailed follow-up investigations (Flynn et al. 2021).

Experimental Crosses between Strains:

In nature, the emergence of a drive element often imposes an immediate, strong selection pressure for the evolution of an unlinked suppressor that restores the Mendelian transmission ratio. Due to their independent evolutionary histories, isolated populations and subspecies may commonly harbor distinct suites of drive and suppressor elements. Outcrossing between divergent populations stands to re-expose ancestrally active drive systems by decoupling drive elements from their co-evolved suppressors (Mercot et al. 1995; Phadnis and Orr 2009; Haines et al. 2021). These considerations suggest that crosses between divergent mouse strains could be ripe for the discovery of selfish elements.

Progeny from experimental crosses should yield predictable Mendelian segregation ratios. The presence of a locus leading to drive (or other forms of TRD) yields a tell-tale signature in progeny genotype data: a departure from the expected genotype frequencies across a contiguous stretch of the genome. The size of the genomic region exhibiting this distorted signal is dependent on the local recombination rate. In a typical F2 or backcross design, such regions are expected to stretch over many megabases of physical sequence and implicate numerous surveyed markers. This approach has previously led to the identification of several TRD loci in mice (Siracusa et al. 1991; Rowe et al. 1994; Montagutelli et al. 1996; Dumont et al. 2011).

F2 and backcross paradigms are commonly used by mouse geneticists as screening approaches to find genetic loci contributing to variation in traits of interest. Databases such as the Mouse Phenome Database (MPD; RRID: SCR_003212) have been developed to curate data from mapping crosses (Bogue et al. 2020), allowing for ready access to decades of cross genotype data that can be retroactively exploited to identify regions of potential TRD. Although population sizes may be too small to permit the discovery of weak TRD signals, this data mining strategy is likely to be effective for picking up loci associated with strong distortion. Indeed, we scanned ~50 published F2 and backcross genotype datasets available on the MPD for genomic regions that depart from Mendelian expectations using Chi-square statistics. Even though genotype cleaning protocols often eliminate loci that depart from Mendelian ratio expectations, and despite the modest number of surveyed markers in many studies, we observe potential instances of TRD in 50% of surveyed backcrosses (7/14) and 42.6% of surveyed intercrosses (20/47; Table S1).

One limitation of this approach is that drive cannot be readily teased apart from genotyping error or other mechanisms leading to TRD, including embryonic lethality and differential survival (Adams et al. 2013). Thus, this strategy is useful as a screening tool for finding candidate meiotic driver elements, but subsequent follow up work is necessary to establish the biological basis of observed signals and, if confirmed, address the underlying mechanisms leading to non-Mendelian segregation.

Many inbred strain genomes have now been fully sequenced, including several strain-specific de novo assemblies (Keane et al. 2011; Adams et al. 2015). There is considerable unrealized potential to interrogate these high-quality genomic sequences for loci with exceptional variability between strains that align with common features of drive elements. Experimental crosses between such divergent strains can then be established to evaluate concrete hypotheses about locus-specific effects on TRD. Such an approach has recently met with success in Drosophila (Brand and Levine 2021). In our own recent work, we employed k-mer based approaches to survey the abundance and diversity of centromere satellites in inbred strain genomes. We identified large differences in overall centromere satellite abundance between strains, which may promote differences in centromere drive potential in F1 hybrids (Arora et al. 2021). Such hypotheses can be tested using pooled sequencing in the F2 or backcross progeny of F1 animals (Wei et al. 2017) or via direct cytological evaluation of segregating chromosomes in F1 oocytes (Chmátal et al. 2014). Consistent with predictions based on our findings, hybrids between C57BL/6J, a strain with intermediate centromere satellite copy number, and ZALENDE/EiJ, a strain with low centromere satellite copy number, exhibit preferential segregation of telocentric C57BL/6J chromosomes into oocytes (Iwata-Otsubo et al. 2017). While our prior work was restricted to centromeres, additional investigations into other satellite-rich domains, including telomeres and heterochromatic regions, could prompt additional hypotheses.

Multiparent mouse mapping populations:

Multiparent mouse mapping populations (MMPs), such as the Diversity Outbred (DO) population (Churchill et al. 2012; Ashbrook et al. 2021), the BXD recombinant inbred strain panel (Ashbrook et al. 2021), and the Collaborative Cross (CC) (Churchill et al. 2004), have already proven their value as powerful resources for the discovery of loci that violate Mendel’s Laws. All MMPs feature multiple founder strains mated through several generations of organized outcrossing to scramble the strain haplotype origins of loci across the genome. In the case of the BXD and CC, animals were subsequently inbred to yield large numbers of independent strains, each a reproducible and distinct genomic patchwork of the founder genomes. In contrast, the DO population is continually maintained by random outcrossing among mice at each generation.

MMPs boast notable advantages over inbred strain analyses and simple crosses for the discovery of drive elements and loci influencing transmission ratios. MMPs feature pedigreed mice maintained over many serial generations, allowing for allele frequency changes to be monitored over prolonged periods of time. This advantage is key because weak drivers may not leave a perceptible footprint in allele proportions after just one (or even a few) generations. The ability to contrast differences between expected haplotype frequencies (determined by founder haplotype proportions) and current-day haplotype frequencies provides a significant boost in statistical detection power.

These advantages helped to spur the recent discovery of the R2d2 drive system in the DO population (Didion et al. 2015). Through routine genetic monitoring, it was noted that one founder haplotype (WSB/EiJ) was steadily increasing in frequency across a region on chromosome 2 (Chesler et al. 2016). Females heterozygous for the WSB/EiJ haplotype over-transmit this portion of chromosome 2 to their progeny. The underlying drive element was subsequently mapped to a 9.3 Mb region harboring a copy number gain of a 127kb interval spanning a single gene, Cwc22 (chr2: 76.9–86.2). In most DO founder strains, Cwc22 is present at a single locus (R2d1) at the expected diploid copy number (CN) state. However, in WSB/EiJ, Cwc22 has been massively amplified at a second locus (cleverly termed R2d2) and is present at CN=30. Serendipitously, a spontaneous mutant with a deletion of most Cwc22 copies on the WSB/EiJ R2d2 haplotype was discovered in a DO breeding colony. Transmission patterns in females carrying this WSB/EiJ deletion allele obey the predictions of Mendelian segregation, providing direct evidence for Cwc22 copy number expansion as the responder in this selfish drive system. Intriguingly, the strength of R2d2-mediated drive is variable depending on genetic background, arguing for the presence of multiple as-yet unmapped modifiers of drive strength. Cwc22 is not functionally well-characterized, but it is known to play a role in alternative splicing of transcripts in oocytes. It has been hypothesized that the R2d2 locus may function as a neocentromere, although specific evidence in support of this hypothesis is currently lacking (Didion et al. 2015). Thus, the mechanisms through which R2d2 triggers meiotic drive are not currently understood.

In mammals with single locus sex determination, the presence of a sex-linked drive locus will lead to departures from the Mendelian expectation of sex ratio parity. Recently, we screened breeding records from the CC to explore the incidence of sex ratio distortion in this MMP (Haines et al. 2021). Strikingly, we found that more than one-third of CC lines exhibit significant departures from 50:50 expected sex ratio. We showed that this distortion is stable across environments and time and is not mediated by maternal effects, arguing for its genetic basis. Further, we demonstrated that sex differences in embryonic and neonatal lethality are unlikely to account for the observed sex biases. As the inbred CC founder strains yield sex-balanced litters, we speculate that the genetic shuffling of eight diverse parental genomes during the early CC breeding generations led to the decoupling of sex-linked distorters from their co-evolved suppressors, unleashing complex, multiallelic systems of sex chromosome drive (Haines et al. 2021). While our work has established the CC as a valuable model for studying sex chromosome conflict, future investigations are needed to unlock the identity of any drive elements involved and the mechanisms through which they enable biased chromosome transmission.

The discovery of the R2d2 element and pervasive sex ratio distortion in the CC underscore a critical reality for MMPs: through the accumulation of new mutations and recombination of driver alleles onto permissive genetic backgrounds, MMPs encapsulate genomes evolving in real time. Routine genetic monitoring of populations is essential for ensuring balanced representation of alleles and enabling swift responses to the emergence of suspected drive elements, thereby preserving population genetic integrity (Chesler et al. 2016).

Insight into the molecular mechanisms of meiotic drive in house mice

Beyond their utility for the discovery of drive loci, house mice present powerful models for understanding the molecular mechanisms by which selfish elements hijack the processes of chromosome transmission to bias their own passage to the next generation. Indeed, the t-haplotype on mouse chromosome 17 has been the focus of investigation for nearly a century (Schimenti 2000), providing a textbook example of a segregation distorter functioning in the male germline. More recently, cutting-edge imaging and genomic approaches have fostered new insights into the molecular mechanisms of centromere drive and sex ratio distortion. We summarize this progress and highlight lingering knowledge gaps below.

The mouse t-haplotype:

The mouse t-haplotype is a male meiotic drive element and one of the first-discovered and most intensively investigated transmission ratio distorters in nature (Silver 1985). The t-haplotype extends over ~40Mb on chr17 (from approximately 3,000,000–42,000,000 bp), spanning a remarkable 1.5% of the mouse genome (Figure 1A; (Kelemen and Vicoso 2018)). Multiple t-haplotype alleles segregate in nature (Morita et al. 1992; Ardlie and Silver 1998; Dod et al. 2003), but all feature a core common architecture. Each t-haplotype harbors a responder locus (Tcr) that exhibits biased transmission, along with multiple trans-acting t-haplotype distorters (Tcd) that are physically linked to Tcr through a series of inversions (Figure 1A; (Lyon 1984; Hammer et al. 1989)). The inversions effectively limit recombination between the t-haplotype and standard haplotype (Silver and Artzt 1981; Artzt et al. 1982), rendering t-haplotypes susceptible to the accumulation of deleterious mutations via Muller’s Ratchet. Consequently, many t-haplotypes contain recessive lethal alleles, with t/t homozygous individuals dying at various stages of embryonic development (Safronova 2009). However, t-haplotypes from different wild populations may harbor distinct recessive lethal variants that can often complement one another, allowing some t/t hybrids to survive (Artzt et al. 1982; Pla and Condamine 1984). In such cases, t/t homozygotes are infertile, presumably due to the recessive action of one or more of the distorter loci.

Fig 1.

(A) Genomic position and architecture of the mouse t-haplotype. The relative positions and sizes of four inversions are depicted as arrows. The approximate positions of known distorter loci Tagap1^Tcd1 and Fgd2^Tcd2 are shown, along with the position of the TRD responder locus, Smok^Tcr and the wild-type Smok genes. The positions of additional distorter (D) and sterility (S) loci are also designated. (B) Schematic of the molecular mechanism by which the t-haplotype induces cis-encoded differences in sperm motility. Distorter loci and their wild-type protein counterparts are freely exchanged between adjacent t- and +- round spermatids via syncytial bridges. In contrast, the t-haplotype responder locus, Smok^Tcr, is restricted to cells with the t-haplotype. Distorter loci increase signaling through wild-type Smok, which in turn leads to increased phosphorylation of the axoneme and decreased sperm motility. In cells that carry the t-haplotype, Smok^Tcr inhibits Smok signaling in the presence of distorters, leading to normal levels of axoneme phosphorylation and sperm motility

Historically, the inverted architecture of the t-haplotype posed a barrier to mapping the identity of Tcr and Tcd. However, taking advantage of rare recombination events facilitated by duplicated sequences shared between the t-haplotype and standard haplotype, Hermann et al. successfully localized Tcr to a 155-kb region (Herrmann et al. 1999). Using gene knockouts and sequence comparisons between the standard and t-haplotypes, the authors identified a novel fusion gene between the 3’ end of Rsk3, a serine/threonine kinase, and the 5’ end of a novel protein kinase termed sperm motility kinase, or Smok. As expected, this novel fusion gene, Smok^Tcr, is only present on the t-haplotype, is expressed in late spermatogenesis, and promotes TRD of its resident chromosome when transgenically expressed elsewhere in the genome. Both wild-type and t-haplotypes also harbor multiple non-mutant copies of Smok, which function as kinases that regulate sperm flagellar movement via phosphorylation of the sperm flagellar axoneme (Herrmann et al. 1999).

Efforts to map t-haplotype distorters have been similarly challenged by the presence of large inversions spanning the t-haplotype. Nonetheless, heroic positional cloning efforts coupled with use of gene knockouts and transgenic validation experiments have led to the successful identification of several t-haplotype distorters and sterility loci (Huw et al. 1995; Braidotti and Barlow 1997; Fraser et al. 1997; Bauer et al. 2005, 2007, 2012; Charron et al. 2019). These loci act as signaling molecules that function upstream of Smok to regulate its role in sperm flagellar motility.

In t/+ heterozygotes, males produce equal numbers of t and +-bearing sperm but virtually always transmit the t-haplotype to live offspring (>90%; (Silver and Olds-Clarke 1984)). In contrast, there is balanced transmission of both alleles through the germline of female carriers. How does the t-haplotype lead to such strong, male-limited drive? In a nutshell, the t-haplotype appears to function as a poison-antidote system (Figure 1B). T-haplotype distorters encode proteins that additively impact sperm flagellar movement via increased Smok signaling. While only expressed from the t-haplotype, these distorters are transported to haploid spermatids carrying the standard haplotype via cytoplasmic bridges that link developing spermatids in a syncytium, effectively transporting the “poison” to all cells (Schimenti 2000). The hyperactivation of Smok in sperm with the standard haplotype triggers increased phosphorylation of Smok targets, including the sperm flagellar axoneme. T-haplotype bearing sperm harbor a mutant Smok^Tcr allele that inhibits Smok signaling, leading to near-baseline levels of sperm axoneme phosphorylation in the presence of distorters and rescuing the impaired sperm flagellar phenotype (Herrmann et al. 1999). Smok and Smok^Tcr avoid diffusion across the syncytium and are uniquely retained in their parent cells (i.e., are expressed in cis). The mechanism behind this parent-cell retention is not clear, but may be related to the late expression of Smok (perhaps just before the dissolution of the cytoplasmic bridges) or the active sequestration of Smok transcripts or protein products in specialized subcellular compartments (Herrmann et al. 1999; Schimenti 2000). Single-cell RNA-seq on heterozygous t/+ male testes could provide firm evidence for the cis-retention of Smok and Smok^Tcr transcripts. Additionally, immunoprecipitation coupled with mass spectrometry could reveal interacting cytoplasmic proteins that facilitate sequestration of Smok and Smok^Tcr within their parent cells.

Beyond its impact on sperm motility, the t-haplotype exerts broader phenotypic and genomic effects on its carriers. For example, recent work has demonstrated broad transcriptional differences between t-haplotype carriers and mice with the standard haplotype. Intriguingly, these differences are not limited to genes within the t-haplotype locus and are broadly observed across diverse tissues (Lindholm et al. 2019). Emerging evidence suggests that t-haplotype carriers have a higher propensity to emigrate from their parent population, implying that the locus can influence animal behavior, potentially in ways that benefit its long-term survival (Runge and Lindholm 2018). Additionally, female t-haplotype carriers have increased longevity (Manser et al. 2011) and decreased activity (Auclair et al. 2013). Finally, the t-haplotype harbors 78% of all MHC protein complex genes and 82% of all pheromone activity genes in the mouse genome (Lindholm et al. 2019). While to our knowledge, no studies have yet demonstrated differences in immune function or the chemosensory repertoire associated with the t-haplotype, it is tempting to speculate the presence of such functional links.

The t-haplotype is present in all M. musculus subspecies, but absent from M. spretus, an outgroup species that diverged from M. musculus ~2 million years ago (Silver 1985). Evolutionary analyses suggest that the t-haplotype was present in the common ancestor of all M. musculus, implying its persistence over >0.5 million years of evolution, despite significant costs to male carriers. Indeed, male carriers of the t-haplotype have reduced sperm competitive ability (Sutter and Lindholm 2015; Manser et al. 2020), impaired sperm motility (Olds-Clarke and Johnson 1993), and sire smaller litters (Lindholm et al. 2013). Furthermore, t/t homozygotes are often non-viable (Safronova 2009). Although the persistence of the t-haplotype remains an evolutionary enigma, its continued survival may be enabled by occasional recombination events with the standard haplotype that purge recessive deleterious variants and reinstate the function of essential genes (Kelemen and Vicoso 2018). The frequency of gene conversion between the t- and standard haplotype is not known, but it is also possible that moderate rates of gene conversion could override the gradual erosion of key genes embedded in the t-haplotype via mutation accumulation. Demographic considerations may also facilitate the long-term survival of the locus. T-haplotype frequency is inversely related to effective population size (Ardlie and Silver 1998), implying a relative t-haplotype survival advantage when selection against deleterious alleles is weak and sperm competition between males is minimized (Dean et al. 2006). The demic structure of wild house mouse populations may provide an optimal context for t-haplotypes to take root.

The precise origins of the t-haplotype also remain debated, with some investigators arguing that it was introgressed from a distinct (and potentially now extinct) mouse species, and others contending that the t-haplotype arose in an ancestral M. musculus population (reviewed in (Kelemen and Vicoso 2018)). With increasing genomic resources for wild mice (Harr et al. 2016) and the continued development of ever-more sophisticated population genomic methodologies, disentangling the complex evolutionary history of this locus may be within close reach.

Molecular mechanisms of centromere drive in house mice:

Mouse centromeres are capable of instigating whole chromosome TRD during female meiosis (Chmátal et al. 2014). Broadly speaking, centromere drive is enabled by the selfish exploitation of centromere satellite repeats by kinetochore proteins and the asymmetric nature of female meiosis. The “stronger” centromere associates with more kinetochore proteins and undergoes non-Mendelian segregation into the single functional meiotic product of female meiosis – the oocyte. To suppress this biased segregation, kinetochore proteins adaptively evolve to counter the cheating centromere and equalize transmission ratios. The ensuing evolutionary arms race between centromere satellites and protein components of the kinetochore fuels their rapid coevolution (Malik and Henikoff 2001; Henikoff et al. 2001).

The architecture of mouse centromeres is important in the context of centromere drive. The majority of inbred lab mice have telocentric chromosomes defined by distally positioned centromeres at the end of the chromosome, proximal to the telomere (Kipling et al. 1991; Kalitsis et al. 2006). Each mouse centromere can be broadly subdivided into two domains. The core domain is composed of minor satellite DNA and associates with kinetochore proteins, including the centromere-defining histone variant, CENP-A. The pericentromeric heterochromatin domain is made up of major satellite DNA and immediately flanks the core domain. These two satellite domains perform distinct roles in chromosome segregation. The minor satellite domain is involved in regulating kinetochore complex assembly, while the major satellite array is involved in sister chromatid cohesion and kinetochore-microtubule attachment signaling (Cheeseman and Desai 2008; Fukagawa and Earnshaw 2014; Musacchio and Desai 2017).

To date, multiple centromere drive and suppressor mechanisms have been discovered in mice. Below, we touch upon these diverse mechanisms of centromere drive and suppression and highlight key outstanding questions in the context of centromere evolution and function.

Centromere drive between karyotypically identical mice

The inbred mouse strains C57BL/6 and SJL have the same karyotype but divergent minor satellite array sizes. F1 hybrids exhibit biased chromosome transmission through female meiosis, with the “stronger” centromere harboring a larger minor satellite array, reduced density of tubulin, weakened microtubule organizing center (MTOC) force, smaller centromere-kinetochore distance, and lower SPC24 levels (Figure 2A). SPC24 is a marker of the outer kinetochore NDC80 complex; in turn, NDC80 complex size is thought to be directly proportional to minor satellite array size (Wu et al. 2018). Why “stronger” chromosomes with larger minor satellite arrays have lower SPC24 levels and apparently smaller NDC80 complex size is not yet understood.

Fig 2.

(A) An oocyte in metaphase of meiosis I from a F1 hybrid of C57BL/6J (dark gray chromosome) and SJL (light gray chromosome). Chromosome flipping occurs before spindle migration and favors C57BL/6J chromosomes to preferentially segregate to the oocyte. (B) Models of outer kinetochore protein recruitment. In the top schematic, outer kinetochore protein abundance is proportional to the abundance of inner kinetochore proteins and minor satellite array size. In the bottom schematic, outer kinetochore protein abundance is not proportional to inner kinetochore protein abundance and minor satellite array size. (C) An oocyte in metaphase of meiosis I from a F1 hybrid of C57BL/6J (dark gray chromosomes) and ZALENDE/EiJ (light gray chromosomes). Chromosome flipping occurs after spindle migration and favors preferential segregation of C57BL/6J telocentric chromosomes and ZALENDE/EiJ metacentric chromosomes to the oocyte. (D) Schematic representing the relationships between centromere satellite DNA arrays, CENP-A nucleosome density and stability, and kinetochore protein association. (E) An oocyte in metaphase during meiosis I from a F1 hybrid of C57BL/6J (dark gray chromosome) and SPRET/EiJ (light gray chromosome). In this model, chromosome flipping cannot occur after spindle migration because of immediate anaphase onset

Current models of kinetochore assembly on centromere DNA assume a stoichiometric relationship, with the density of kinetochore complex components proportional to the number of CENP-A nucleosomes and minor satellite array size (Westhorpe and Straight 2016; Walstein et al. 2021). The discordance between SPC24 levels and minor satellite array size suggest that either (1) the size of the minor satellite array is not always proportional to the number of associated CENP-A nucleosomes or (2) SPC24 levels do not faithfully reflect variation in minor satellite array size or CENP-A nucleosome abundance. It is also possible that there is a relationship between the amount of major satellite DNA and SPC24. Regardless, the relationship between centromere drive, centromere size, and kinetochore protein abundance is evidently complex.

An additional interpretation is that the recruitment of outer kinetochore proteins is not proportional to the abundance of centromere-associated inner kinetochore proteins (Figure 2B). In this case, the abundance of outer kinetochore proteins may be dynamic and regulated by microtubule proximal regulatory factors (Dhatchinamoorthy et al. 2019). Indeed, the lack of correlation between SPC24 levels with minor satellite array size in C57BL/6 × SJL F1s supports this possibility. Future studies investigating the relationship between outer kinetochore protein levels and minor satellite array size in diverse mice are needed to tease apart overarching trends, as well as their mechanistic relevance for drive potential.

Centromere drive between karyotypically divergent mice

Many isolated populations of Mus musculus domesticus harbor derived karyotypes formed by the fusion of two telocentric chromosomes yielding a single metacentric chromosome. When these so-called Robertsonian chromosome fusions first appear in a population, they are, of necessity, present in a heterozygous state (Kalitsis et al. 2006). Male mice heterozygous for Rb chromosomes exhibit chromosome pairing defects at meiosis and reduced fitness (Eaker et al. 2001), suggesting that they should be rapidly purged from populations. The survival and fixation of numerous Robertsonian translocations in diverse mouse populations is therefore paradoxical.

Preferential transmission of Rb chromosomes via centromere drive provides one potential mechanism for their pervasiveness in wild mouse populations. Indeed, prior work has demonstrated strong transmission advantages for metacentric chromosome fusions in some, but not all, genomic backgrounds (Chmátal et al. 2014). The breakpoints of Rb translocations most often occur in minor satellite arrays (Garagna et al. 2001). Consequently, Robertsonian fusions can result in increases or decreases in centromere size relative to telocentric chromosomes, depending on the positioning of the breakpoints leading to the fusion and the centromere background on which the centromere fusion arises (Kalitsis et al. 2006). Thus, Rb metacentrics can either have “strong” or “weak” centromeres. Rb metacentrics with strong centromeres are more likely to be transmitted to offspring and, ultimately, fix within populations. These considerations likely account for variation in the probability that a metacentric chromosome increases in frequency or fixes in a given population, as well as background-dependent differences in drive strength.

How, precisely, do strong centromeres lead to meiotic drive in Rb heterozygotes? To address this question, Akera et al. (2017) applied powerful optogenetic techniques to the oocytes of F1 hybrids between ZALENDE/EiJ (2n = 22) and CF-1 or C57BL/6J (2n = 40) mice. ZALENDE/EiJ metacentric chromosomes were shown to harbor strong centromeres that exhibit preferential segregation in F1 hybrids with either C57BL/6 or CF-1 (Chmátal et al. 2014). Akera et al. (2017) showed that centromere drive in this system is mediated via differences in the level of CDC42-regulated tyrosination of alpha-tubulin between the spindle poles oriented on the egg and cortical sides of the dividing cell (Akera et al. 2017). This functional asymmetry leads to unstable chromosome attachments between stronger centromeres that are positioned toward the cortical (non-egg) spindle (Figure 2C). By comparison, chromosome attachments with the weaker centromere are comparatively more stable. Thus, reorientation of the chromosomes about the metaphase plate favorably resolves this instability and is energetically favored.

Following the finding that stronger centromeres formed more unstable attachments with microtubules on the cortical side, Akera et al (2019) looked more closely at the mechanisms of microtubule destabilizer recruitment by stronger centromeres. They observed asymmetric localization of MCAK, Survivin, and phosphorylated INCENP, proteins that are all known to be involved in kinetochore-microtubule attachment signaling. Stronger centromeres recruited more microtubule destabilizer proteins, resulting in their detachment from the cortical side and selfish reorientation to the oocyte.

Pericentromeric heterochromatin is known to recruit microtubule destabilizers, leading to the possibility that differences in the major satellite arrays between strains could foster differences in the abundance of microtubule destabilizers. However, the ZALENDE/EiJ, C57BL/6J, and CF-1 strains have approximately equal amounts of pericentromeric heterochromatin, ruling out this potential explanation. Instead, the authors propose that differences in microtubule destabilizer recruitment are established via strain differences in BUB1 activity (Akera et al. 2019). Although BUB1 is most well-known for its role in the spindle assembly checkpoint, its function in microtubule destabilizer recruitment appears to be independent of the spindle assembly checkpoint and critical for the stability of kinetochore-microtubule attachments (Perera et al. 2007; Chen et al. 2021). However, the potential contribution, if any, of the major satellite to centromere drive remains an open question.

Centromere drive between Mus species

M. musculus (C57BL/6 or CF-1) and M. spretus (SPRET/EiJ) are ~2MY divergent but share a common 2N=40 karyotype (She et al. 1990; Chevret et al. 2005). F1 hybrids between M. spretus and M. musculus yield infertile, but viable, male offspring, although females are generally fertile (Bonhomme et al. 1978; Matsuda and Chapman 1992). M. spretus harbors significantly more minor satellite DNA than most M. musculus strains (Miyanari et al. 2013). However, M. spretus has reduced pericentromeric heterochromatin repeat content compared to M. musculus. Interestingly, M. spretus centromeres exhibit increased abundance of the microtubule destabilizer MCAK compared to M. musculus centromeres suggesting that factors other than pericentromeric heterochromatin are responsible for microtubule destabilizer recruitment to centromeres (Akera et al. 2019). Given the increased abundance of microtubule destabilizers, M. spretus centromeres should be more likely to engage in chromosome flipping and prone to preferential segregation into the oocyte. However, in M. musculus × M. spretus F1 oocytes, anaphase I initiates right after spindle migration, leaving no time for chromosome reorientation about the metaphase plate (Figure 2E; (Sebestova et al. 2012)). These observations underscore the importance of subtle variation in the timing of meiotic progression and its potential consequences for centromere drive. Given the complex network of genes responsible for cell cycle control, it is very likely that the phenomenon of centromere drive is a complex trait influenced by many genetic loci, with certain drive mechanisms only manifesting on specific genetic backgrounds.

The potential for CENP-A associated mechanisms of centromere drive in house mice

Centromere function is dependent on the localization of CENP-A, a specialized histone H3 variant, to the minor satellite array. CENP-A localization is required for kinetochore complex formation and subsequent chromosome attachment to microtubules for segregation (Musacchio and Desai 2017).

Diverse eukaryotic species harbor differences in their centromere satellite repeat unit as well as CENP-A amino acid sequence. Unlike other histone proteins, which are subject to strong purifying selection and exhibit minimal sequence divergence between species, the N-terminal DNA associating domain of CENP-A exhibits clear signals of adaptive evolution in many species (Malik et al. 2002; Talbert et al. 2002; Schueler et al. 2010). This trend is presumably an evolutionary response to ensure functional compatibility with rapidly evolving centromere satellite sequences (Henikoff et al. 2001), a theory that rests on the assumption that CENP-A harbors sequence-specific binding preferences.

Our recent work profiled centromere architecture across diverse inbred mouse strains and uncovered considerable variation in centromere satellite copy number and satellite sequence heterogeneity (Arora et al. 2021). Remarkably, all surveyed strains in our analyses are characterized by an identical CENP-A amino acid sequence, suggesting that the relationship between CENP-A binding and properties of centromeric satellite DNA is complex. The question of how centromere satellite heterogeneity influences CENP-A association requires further investigation (Figure 2D). If strain differences in centromere architecture lead to strain differences in the CENP-A association landscape, it will be fascinating to explore the role of this variation on kinetochore protein association and centromere drive potential in house mice.

Suppression of Centromere Drive

Although the fitness costs associated with centromere drive are not always evident, it is theorized that mechanisms should evolve to suppress the action of a selfish centromere variant. There are two prevailing mechanistic theories for how suppression might emerge. The most widely invoked theory advances that centromere DNA-associated kinetochore proteins evolve counter-adaptations to selfish centromeres by adjusting their association with minor satellite repeats, thereby modifying centromere strength and restoring equity of the transmission ratio (Malik and Henikoff 2001; Henikoff et al. 2001). A second, recently posited theory asserts that the microtubule destabilizers that associate with pericentromeric repeats can acquire adaptations to counter the effects of the selfish centromere via altered kinetochore-microtubule signaling dynamics (Kumon et al. 2021). Sequence-based analyses of adaptive evolution of kinetochore proteins across a range of taxa may provide clues into the relative evolutionary importance of these alternative mechanisms of drive suppression.

Whether the evolution of kinetochore proteins as suppressors of centromere drive during meiosis could impact chromosome segregation dynamics in mitosis is not known. The disruption of mitotic chromosome segregation can lead to aneuploidy, an established trigger for oncogenesis. Thus, understanding the mechanisms and costs of centromere drive and its associated modes of suppression may yield novel insights into the origins of cancer as well as new therapeutic strategies for cancer treatment.

Inter-gametic conflict in the male mouse germline:

Evolutionary theory predicts that sex chromosomes should be enriched for drive elements (Hurst and Pomiankowski 1991). In species with single locus X/Y sex determination, a Y- (X)-linked driver will skew the population sex ratio toward males (females). The departure from the idealized 1:1 sex ratio will impose a strong selection pressure for the emergence of a suppressor on the other sex chromosome or autosomes to restore the sex ratio to parity (Hamilton 1967). In turn, new sex-linked drivers capable of overcoming the suppressive mechanism can emerge, leading to an evolutionary arms race between sex-linked selfish elements and their supressors. This evolutionary feud can profoundly shape the architecture and gene content of the sex chromosomes (Mueller et al. 2008; Soh et al. 2014; Bachtrog 2020) and contribute to the emergence of species barriers (Phadnis and Orr 2009; Crespi and Nosil 2013).

The house mouse sex chromosomes are enriched for co-amplified, spermatid expressed ampliconic gene families that are hypothesized to reflect recurrent bouts of driver/suppressor evolution (Figure 3A; (Mueller et al. 2008; Soh et al. 2014)). The most well-characterized of these sex-linked amplonic gene families is the Sycp3-like gene family members Slx, Slxl1, and Sly which are each present at upward of 50–100 copies in the mouse genome (Scavetta and Tautz 2010; Morgan and Pardo-Manuel de Villena 2017). X-linked Slx/Slxl1 and Y-linked Sly are neofunctionalized copies of Sycp3, a critical structural component of the meiotic synaptonemal complex, that have been co-opted for new roles in gene regulation in round spermatids (Kruger et al. 2019).

Fig 3.

(A) Genomic architecture of homologous ampliconic genes on the mouse X and Y chromosome. Much of the mouse Y is composed of an oligogenic “cassette” harboring copies of Sly, Srsy, and Ssty1/2 repeated many times in tandem. The genic organization and structure of a single cassette is shown. The homologous X-linked counterparts of these Y-linked ampliconic genes are organized into tandem clusters, the relative positions of which are denoted by lines of the same color as their Y-linked homologs. Images are adapted from (Morgan and Pardo-Manuel de Villena 2017). (B) Schema depicting the current working model for Slx/Slxl1 and Sly impact on gene regulation in postmeiotic round spermatids. In wild-type animals with balanced Slx/Slxl1 and Sly gene copy number ratios, most SLX/SLXL1 protein is cytoplasmic and not capable of competing with SLY for binding to SSTY1 (and potentially other spindlin proteins) at gene promoters. In this situation, SLY outcompetes SLX/SLXL1 and elicits repression of target genes via association with the SMRT/Ncor complex. When there is a relative excess of SLX/SLXL1 compared to SLY, SLX/SLXL1 relocates to the nucleus where it out-competes SLY for binding to SSTY1 at gene promoters. SLX/SLXL1 does not interact with the repressive SMRT/Ncor complex, and upstream target genes are expressed. Images are adapted from (Moretti et al. 2020)

SLX/SLXL1 and SLY are transcriptional regulators that exhibit antagonistic effects on gene expression in round spermatids (Moretti et al. 2017, 2020). SLY associates with gene promoters where it recruits the SMRT/Ncor complex to repress transcription of its targets, including Slx/Slxl1. SLX/SLXL1 associates with a subset of SLY targets, but fails to recruit the SMRT/Ncor repressive complex (Cocquet et al. 2012; Moretti et al. 2020). As a result, SLX/SLXL1 induce upregulation of target genes, although the mechanism of this upregulation remains unknown (Moretti et al. 2020). In the absence of SLY, SLX/SLXL1 associate with a wider array of SLY targets and exert a more severe dysregulatory effect. Taken together, these observations reveal a genetic conflict between X-linked and Y-linked members of the Sycp3 gene family that bears out in the program of gene regulation in round spermatids (Cocquet et al. 2012; Moretti et al. 2020).

To ensure the appropriate tension between gene silencing and up-regulation during round spermatid development, Slx/Slxl1 and Sly gene copy number have been maintained in stoichiometric balance over house mouse evolution (Cocquet et al. 2012; Morgan and Pardo-Manuel de Villena 2017). Relative overexpression of SLY by targeted deletion or knockdown of Slx/Slxl1 results in broad perturbations of the meiotic program that ultimately manifest as reduced fertility and male-biased litters (Cocquet et al. 2010). Conversely, overexpression of SLX/SLXL1 results in female-biased litters and reduced fertility (Cocquet et al. 2012; Kruger et al. 2019). Investigation of sperm from mice harboring a spontaneous deletion of the Yq chromosome (and, therefore, exhibiting relative overexpression of SLX/SLXL1) reveal differences in motility and morphology between X- and Y-bearing sperm (Rathje et al. 2019). While sperm with both sex chromosome genotypes are equally capable of fertilizing oocytes in intracytoplasmic sperm injection experiments, Y-bearing sperm have decreased motility and a higher proportion of morphological abnormalities that give X-bearing sperm a fertilization advantage in the context of natural mating (Rathje et al. 2019).

Many of the regulatory targets of SLX/SLXL1 and SLY are members of ampliconic gene families, including other sex-linked gene families (Moretti et al. 2017, 2020). Notable among these targets are the Y-linked spindlins, Ssty1/2. SSTY1/2 are H3K4me3 readers that can directly interact with both SLX/SLXL1 and SLY (Kruger et al. 2019; Moretti et al. 2020). Collectively, these observations have informed a current working model for SLX/SLXL1 and SLY competition (Figure 3B). Briefly, SSTY1 (and potentially SSTY2) recognizes H3K4me3 at the promoters of spermatid-expressed genes. SLY and SLX/SLXL1 then compete for SSTY1 binding, with SLY typically winning out in mice with balanced Slx/Slxl1 and Sly copy number ratios. When Sly is knocked down, Slx/Slxl1 expression is increased and SLX/SLXL1 occupy a comparatively larger fraction of SSTY1 binding sites. The repressive SMRT/NCor complex, which normally associates with SLY but not SLX/SLXL1, is not recruited to these SLX/SLXL1-bound sites and, consequently, the associated target genes are upregulated (Moretti et al. 2020).

The molecular mechanism(s) by which the broad program of gene dysregulation in Sly-deficient round spermatids leads to sex ratio distortion, sperm motility defects, and male infertility is not known. The cast of players in this drive system is potentially quite large, complicating efforts to unravel causal mechanisms. For one, SLX/SLXL1 and SLY target thousands of genes, including multiple sex-linked ampliconic genes that are poorly understood and studied (Moretti et al. 2020). Several of these sex-linked ampliconic gene families have been co-amplified with Slx/Slxl1 and Sly, suggestive of potential functional connections between their encoded products (Morgan and Pardo-Manuel de Villena 2017; Moretti et al. 2020). However, the challenges of generating robust and reproducible mouse models for ampliconic genes have posed notable impediments to elucidating the functions of other sex-linked ampliconic gene families (Kruger et al. 2019; Arlt et al. 2020). Recent advances in gene editing technologies and continued improvements to the mouse reference assembly in these structurally complex regions are now poised to enable the development of needed animal resources, allowing investigators to situate the potential roles of other sex-linked ampliconic genes into the broad network of Slx/Slxl1-Sly-mediated genetic conflict.

Intriguingly, Smok genes are also among the targets of SLX/SLX1 and SSTY1 (Moretti et al. 2020), raising the possibility that the t-haplotype and Slx/Slxl1 and Sly drive systems are functionally interrelated. There are several striking phenotypic similarities between t-haplotype carriers and Slx/Slxl1 or Sly knockdown mice, including male infertility with cis-acting effects on sperm motility. Future investigations are needed to expose possible molecular links between these two established drive systems.

From mice to men: Meiotic drive in humans

Perhaps the most compelling motivation for studying meiotic drive in the mouse is its potential to illuminate the scope and scale of non-Mendelian transmission in our own species. Due to the convergence of our biology, and most notably, the broad conservation of the processes of spermatogenesis and oogenesis – we may expect some commonalities in mechanisms and overall genomic impact of meiotic drive. However, as drive elements (and their suppressors) are rapidly evolving, it is unlikely that specific drive-associated loci are conserved between mice and humans. Instead, properties of drive elements identified in house mice may provide “templates” for the identification of analogous drive loci in humans. For example, the Slx/Slxl1 and Sly sex-linked ampliconic selfish genes in mice are not conserved outside of Mus. However, the human sex chromosomes harbor a family of sex-linked ampliconic genes — VCX and VCY — that play roles in male reproduction and have been hypothesized to similarly act as segregation distorters (Lahn and Page 2000).

Meiotic drive is associated with reduced fertility and sterility in hybrids from many species (Zanders and Unckless 2019), leading to the sensible assumption that it could also play a role in the etiology of human infertility. Indeed, the incidence of infertility and subfertility are puzzlingly high in humans, impacting approximately 1 in 8 couples trying to conceive (Louis et al. 2013). Despite the plausibility of this hypothesis, efforts to identify meiotic drive elements in humans have met with little success. Heterozygous female carriers for Robertsonian translocations preferentially transmit the Rb chromosome and suffer from sub-fertility (Pardo-Manuel de Villena and Sapienza 2001b), but there are effectively no other meiotic drive elements that have been identified in humans to date. Genome-wide scans using human pedigrees (Meyer et al. 2012) and single-cell analyses of large pools of sperm (Carioscia et al. 2021) have systematically failed to uncover compelling evidence for transmission biases.

The absence of evidence for meiotic drive in humans could reflect the exceptional rarity of drive in our species. However, such a conclusion evokes discredited ideas of human genetic exceptionalism and is at apparent odds with our shared evolutionary origins with organisms (such as mice) where drive elements are evidently quite common. Alternatively, the paucity of drive signals in the human genome could be a manifestation of ascertainment challenges. Indeed, the small size of human families severely limit statistical power to find drive elements within pedigrees. Drive elements may often segregate at low frequency in human populations if they exert deleterious effects on fitness, rendering them especially difficult to detect in population genomic data. Furthermore, evidence from mice suggest that drive may only manifest on certain conducive genetic backgrounds (Didion et al. 2015). The genetic heterogeneity of human populations may impose additional practical challenges to the detection of drive signals. Nearly all mapped drive loci in the mouse genome are entrenched in structurally complex or repeat-rich regions. Such regions are poorly tagged by commercial human SNP arrays, potentially barring the discovery of selfish elements in these regions of the human genome. Furthermore, genotyping and sequencing errors can often masquerade as drive signals, giving rise to frequent false positives that further challenge the detection of true drivers (Meyer et al. 2012). Such technical errors are potentially most abundant in the structurally complex and repetitive regions of the genome where selfish drive elements tend to localize, further compounding their detection.

The strength or presence of drive may also be conditional on certain non-genetic factors, layering added complexity to the discovery of drive elements in humans with heterogeneous environmental exposures. Indeed, meiotic drive systems have been shown to be influenced by age and temperature in Drosophila (Courret et al. 2019). To our knowledge, the possibility of meiotic drive-by-environment interactions has not been explored in mammals. Intriguingly, many of the processes involved in centromere drive in mice and other taxa – including spindle assembly checkpoint dysregulation, cohesin dysregulation, abnormalities in tubulin, and the stability of kinetochore-microtubule attachment (Mogessie et al. 2018) – are also associated with the rapid decline of oocyte quality with maternal age (Mikwar et al. 2020). These observations raise the intriguing prospect that maternal age could influence the potential for centromere drive. This possibility may bear especially strong repercussions for human fertility in light of contemporary trends toward increasing maternal age at birth.

Despite the near absence of described selfish elements, unique aspects of human biology hint at the potential legacy of drive elements in our genomes. Relative to other species, human female meiosis is characterized by exceptionally high rates of nondisjunction. Although speculative, this high nondisjunction rate could have emerged as a potential trade-off to counter the action of a selfish centromere drive element. Indeed, one plausible mode of centromere drive suppression is to shorten the time frame between metaphase and anaphase, minimizing the opportunity for stronger chromosomes to reorient on the spindle and optimally position themselves for segregation into the oocyte. Such a mechanism has been hypothesized to explain the absence of drive in M. musculus × M. spretus F1 females with differences in their microtubule destabilizer abundance (Akera et al. 2019). However, abbreviating this time window would also minimize the active time for the spindle checkpoint to operate, coming at the possible cost of reduced genome surveillance and higher rates of nondisjunction. Given that many genes that function in meiosis have homologous, pleiotropic functions in mitosis, suppressors that evolve to counter meiotic centromere drive could fuel the uniquely high rates of mitotic chromosome instability and oncogenesis in humans. Thus, the human health consequences of now-extinct drive systems could be substantial, extending beyond impacts on fertility.

Future Prospects and Conclusions

While the past decades have seen considerable advances in dissecting the genetic and molecular mechanisms of meiotic drive systems in mice, there has been comparatively minimal progress toward understanding the extent to which selfish drive elements pervade our own genomes. We contend that insights from house mice, combined with recent technological advances and the availability of massive human genomic datasets, are now primed to empower new insights into the presence, prevalence, mechanisms, and impact of meiotic drive in the human genome.

The availability of genomic sequence data from large-scale biobanks and populations has transformed understanding of the genetic basis of human disease, but these resources have been under-utilized as tools for the discovery of potential drive elements in the human genome. Drive elements are expected to rise in frequency quickly over evolutionary time, leaving a telltale signature that mimics signals of natural selection in genomic data. However, in contrast to natural selection, which is associated with the increase (or fixation) of alleles that confer an adaptive advantage, drive elements almost universally impose a fitness cost. As approaches for allelic effect prediction continue to improve (Kellis et al. 2014; McLaren et al. 2016; Kim et al. 2021), genome-wide scans for putatively deleterious alleles that have rapidly increased in frequency or fixed could be carried out as preliminary screens for drive element discovery.

While it is also common practice to eliminate markers with genotypes that defy Mendelian expectations as a normal part of QC prior to mapping in experimental crosses or pedigrees, such signals are the very hallmark of meiotic drivers. This practice therefore may lead to the elimination of many possible drive elements before they can be discovered. Re-visiting prior large human pedigree analyses with this point in mind may expose previously overlooked drive loci.

The growing use of long-read sequencing technologies is also poised to resolve many lingering questions about the potentially selfish nature of our genomes. The recent telomere-to-telomere assembly of the human genome closed long-standing centromere gaps, marking a major milestone in human genetics (Nurk et al. 2021). Importantly, this achievement also set the stage for the future discovery and cataloging of human centromere diversity. While the relationship between centromere diversity and drive potential is not well-understood, functional centromere heterozygosity established by allelic variation is a prerequisite for this form of TRD. Cataloging human centromere diversity therefore represents a critical first step toward the eventual goal of linking centromere variation with functional differences in chromosome transmission in humans (Langley et al. 2019).

Long-read sequence technologies are also well-suited for the discovery and resolution of polymorphic inversions in human populations. Inversions present attractive regions to focus the search for drive elements due to reduced rates of recombination that maintain responders, drivers and enhancers as linked, co-adapted complexes. Indeed, the mouse t-haplotype is composed of multiple inverted regions that irreversibly link distorter loci to the drive responder locus (Figure 1A). Many large polymorphic inversions have been identified in humans, including several that have been linked to differences in reproductive fitness (Stefansson et al. 2005; Muthuvel et al. 2016). However, it remains unknown whether any of these fitness-associated inversions are capable of drive.

New CRISPR-based approaches for targeted editing and visualization of repetitive and satellite DNA sequences will allow us to address standing knowledge gaps about the molecular mechanisms by which drive elements circumvent the normal rules of genetic transmission (Anton et al. 2014; Smith et al. 2020). Engineered deletion or expansion of satellite arrays can enable explicit testing of hypotheses about relative satellite DNA abundance and the strength of selfish transmission. Existing diverse human cell line resources provide powerful platforms for such investigations (e.g., (Fairley et al. 2020)). Further, allele swap experiments involving putative distorter or responder loci between individuals with and without drive could help establish their causality and unravel the evolution of these complex multiallelic systems (Brand and Levine 2021).

Single-cell methods also hold considerable promise for exposing the molecular mode of action of meiotic drivers, particularly those that function in the male germline. Like mice, human male round spermatids develop in a syncytium. Understanding which gene products are shared between genotypically distinct haploid cells and which are uniquely retained within their cell of origin may help reconcile puzzling observations of phenotype difference between X- and Y- bearing sperm in the mouse t-haplotype and Slx/Slxl1-Sly drive systems. Applied to humans, these technologies could reveal functionally related or interacting proteins that show distinct patterns of cellular localization, hinting at potential engagement in germline genetic conflict (Bhutani et al. 2021).

Finally, the discovery and characterization of new drive systems in mice and other mammals will help uncover the breadth of molecular mechanisms that may contribute to drive in humans. Toward this end, crosses invoking the newly developed Nachman wild mouse strain panel (Phifer-Rixey et al. 2018; Bittner et al. 2021) and diverse strain holdings in repositories like the Czech Wild Mouse Repository (Piálek et al. 2008; Chang et al. 2017) present untested arenas for the discovery of drive elements. The establishment of new MMPs and the continued genetic tracking of established MMPs also present fertile ground for the unveiling and emergence of drive systems.