Centromere-Associated Female Meiotic Drive Entails Male Fitness Costs in Monkeyflowers

See allHide authors and affiliations

Science  05 Dec 2008:
Vol. 322, Issue 5907, pp. 1559-1562
DOI: 10.1126/science.1161406


Female meiotic drive, in which paired chromosomes compete for access to the egg, is a potentially powerful but rarely documented evolutionary force. In interspecific monkeyflower (Mimulus) hybrids, a driving M. guttatus allele (D) exhibits a 98:2 transmission advantage via female meiosis. We show that extreme interspecific drive is most likely caused by divergence in centromere-associated repeat domains and document cytogenetic and functional polymorphism for drive within a population of M. guttatus. In conspecific crosses, D had a 58:42 transmission advantage over nondriving alternative alleles. However, individuals homozygous for the driving allele suffered reduced pollen viability. These fitness effects and molecular population genetic data suggest that balancing selection prevents the fixation or loss of D and that selfish chromosomal transmission may affect both individual fitness and population genetic load.

In the female meioses of both plants and animals, all but one of the meiotic products generally degenerate (1). This asymmetry of cell fate can allow homologous chromosomes to compete for inclusion in the single surviving egg or megaspore, a process termed “female meiotic drive” (14). Female meiotic drive may explain the rapid diversification of centromeres, the DNA-protein complexes that mediate chromosomal segregation (5), and may promote speciation through the evolution of hybrid incompatibilities (5) and karyotypic rearrangements (6). Because nondisjunction during chromosomal competition can cause infertility (2, 5), female meiotic drive may also contribute to genetic variation for reproductive fitness within populations (7), a central issue in evolutionary biology (812) and human health. Despite its potential importance as an evolutionary force, little is known about female meiotic drive in natural populations.

The female meiotic-drive locus in Mimulus (D) exhibits extreme non-Mendelian segregation through female meiosis in hybrids between M. guttatus (IM62 inbred line) and its close relative M. nasutus (SF inbred line), which is predominantly self-fertilizing (13, 14). As seed parents, interspecific heterozygotes transmit >98% M. guttatus (IM62) alleles at markers tightly linked to D, and there is no evidence of postmeiotic mechanisms of transmission ratio distortion (13). Near-complete transmission bias via female meiosis suggests that D is the functional centromere of the chromosome corresponding to the linkage group [linkage group 11 (LG11)] on which it is located (13), because only the centromere (and linked loci) can attain >83.3% transmission via female drive (15). To test this inference, we cytogenetically mapped D in M. guttatus, M. nasutus, and interspecific hybrids (Fig. 1) (SOM text). Because plant centromeres generally consist of megabases of tandemly repetitive DNA with individual repeats 150 to 1000 base pairs (bp) in length (16, 17), we searched the M. guttatus (IM62 line) 6× draft whole-genome sequence [Mimulus Genome Project, U.S. Department of Energy (DOE) Joint Genome Institute] for repeats with those features. A probe for the most common class of repeat found, 728 bp in length (Cent728; fig. S1), hybridized to a single narrow band near the center of each IM62 metaphase chromosome (Fig. 1A and fig. S2A). However, a single pair of homologous chromosomes exhibited two unusually large regions of hybridization (arrows; Fig. 1A and fig. S2A). A probe for the CycA genetic marker tightly linked to D (13) localized between the large Cent728 arrays on this chromosome (Fig. 1B and fig. S2B), demonstrating that this distinctive chromosomal structure (henceforth, C11.2) corresponds to the driving region of IM62 LG11. The region of Cent728 hybridization on each non-C11.2 chromosome was flanked by arrays of typically pericentromeric retrotransposons (Fig. 1C and fig. S2C) (18). This pattern suggests that Cent728 is, if not the centromere-specifying DNA repeat, a marker for centromeric chromosomal regions. Although we cannot yet determine whether the molecular mechanism of Mimulus drive is strictly centromeric (5) and whether the duplication and expansion of Cent728 arrays is causal, this association is consistent with the genetic evidence for centromeric drive (13, 15).

Fig. 1.

Fluorescence in situ hybridization to M. guttatus lines and M. nasutus × M. guttatus hybrids (2N = 28). (A) IM62 M. guttatus metaphase karyotype showing a single band of Cent728 hybridization (green) on each chromosome and two large regions of Cent728 hybridization on one pair of chromosomes (arrows). (B) Colocalization of a genetic marker for drive (CycA; red; above) with C11.2 Cent728 arrays (merged; below). (C) Pachytene IM62 chromosomes with Cent728 (green; above) and flanking Mg_Copia69.2 retrotransposon arrays (red; overlay with Cent728 below). (D) C11.2 (arrow) from IM62 introgressed into M. nasutus genetic background. (E) IM767, an independent inbred line derived from the Iron Mountain M. guttatus population. Each image includes merged false-colored images of DNA-bound DAPI (4′,6-diamidino-2-phenylindole) (blue) with images of additional probes labeled in Alexa-Fluor (green) and/or Texas Red (red); component images are in fig. S2. Scale bar: 2 μm.

We examined metaphase chromosomes from nearly isogenic lines (NILs) containing heterozygous introgressions of M. guttatus D in a largely M. nasutus genetic background (13). Both strong Cent728 arrays from IM62 C11.2 appear present in the NILs (Fig. 1D and fig. S2D), which indicates that they were transmitted as a single genetic unit over five generations of recombination (there is a single weak Cent728 array on LG11 in the SF M. nasutus parent; fig. S3). Thus, the two IM62 C11.2 arrays are inherited as a single genetic locus. The large physical size of the D locus is similar to that of the best-known female meiotic drive system, Ab10-knob in maize (19), but in that case the drive elements are DNA arrays (knobs) that segregate as genetic loci unlinked from the centromeric regions (20). In contrast, our data suggest that female meiotic drive in Mimulus results from competition between chromosomal homologs divergent in centromere-associated repetitive DNA arrays. Regardless of its molecular mechanism, Mimulus drive differs from maize drive in its genomic location relative to centromeres (1921) and thus provides a comparative model for understanding selfish chromosomal evolution.

To investigate drive within M. guttatus, we examined Cent728 hybridization to chromosomes from inbred lines from the Iron Mountain M. guttatus population (10). Some lines completely lacked the distinct C11.2 found in IM62 (Fig. 1E and fig. S2E), suggesting that the driving chromosome is structurally divergent from homologs within the same species. The Iron Mountain M. guttatus population is also polymorphic for heterospecific female meiotic drive, exhibiting discrete variation in segregation patterns at drive-linked markers (Fig. 2). Of the eight independently derived inbred lines test-crossed to M. nasutus, four exhibited strongly distorted segregation (M. guttatus allele transmitted at ∼95% via F1 female meiosis) similar to SF × IM62 hybrids (13, 14) and four exhibited Mendelian segregation. Thus, the driving allele (D) is at intermediate frequency in this M. guttatus population, along with nondriving alternative alleles (henceforth, D). In addition, all three nondriving lines that we examined cytogenetically (including IM767; Fig. 1E and fig. S2E) lack the C11.2 arrays, supporting the inference that drive is associated with chromosomal divergence. This analysis also revealed that allelic variation at the microsatellite marker aat356 (13, 14) was diagnostic for the drive genotype, because all five D lines shared the most common 180-bp allele [overall frequency = 37 out of 113 lines tested (33%)], whereas the four D lines were diverse and carried other alleles (P < 0.008; Fisher's exact test).

Fig. 2.

Cumulative frequencies of M. nasutus homozygotes (NN; black), M. guttatus homozygotes (GG: white), and heterozygotes (NG; gray) for eight F2 testcross families. By χ2 tests (df = 2), four families (D) differed significantly (all P < 0.0001) from Mendelian segregation (1:2:1; NN: NG:GG), but not from the IM62 heterospecific drive expectation of 2:49:49 (P range: 0.12 to 0.40), whereas four families (D) did not differ from Mendelian (P range: 0.34 to 0.97).

Because the driving D allele shows a near-complete transmission advantage over M. nasutus alleles via female meiosis (13), we tested whether it exhibits female drive against alternative conspecific genotypes by examining segregation in within-population test-crosses (SOM text). On average, D displayed a 58:42 conspecific transmission advantage via female meiosis (χ 2 = 4.45, P = 0.035), but was transmitted in a Mendelian fashion via male meiosis (χ2 = 0.51, P = 0.475), resulting in 16% excess transmission of D via female function. Thus, although weaker than heterospecific drive, chromosomal competition appears to be a potent selective force within this interbreeding population. The difference in strength between heterospecific and conspecific drive suggests that suppression of drive has evolved within M. guttatus or that genomic divergence between M. guttatus and M. nasutus intensifies chromosomal competition.

In the absence of countervailing selection, we would expect conspecific female drive to rapidly fix the driving D allele. Therefore, the observed polymorphism within the Iron Mountain M. guttatus population predicts that female meiotic drive has deleterious effects on individual fitness. We characterized the male fitness effects of heterospecific drive in the segregating progeny of a NIL with a heterozygous introgression at D (Dd, where d is the nondriving M. nasutus allele) (13). DD homozygotes had significantly reduced pollen viability relative to Dd heterozygotes and dd homozygotes (Fig. 3A). Because female meiotic drive in heterospecific (Dd) heterozygotes was near 100%, this suggests that even strongly biased chromosomal segregation may impose little direct cost via nondisjunction. This result supports the assumption of costs in the female meiotic drive model of centromere evolution (5), but rejects nonrandom chromosomal segregation in heterozygotes as the mechanism of drive costs in this system.

Fig. 3.

Mean pollen viability (±SEM) of drive genotypes in (A) controlled heterospecific genetic background and environment and (B) wild M. guttatus plants. In both cases, the overall effect of D genotype was highly significant (oneway analysis of variance: P < 0.007) and DD homozygotes had significantly lower pollen viability than the other genotypes, which did not differ from each other. (A) Least squares means (LSMs) contrasts. DD versus other: P = 0.0003; dd versus Dd: P = 0.82. (B) LSMs contrasts. DD versus other: P = 0.0018; DD versus DD: P = 0.59.

We assessed the effects of D on male fertility in wild M. guttatus plants at the Iron Mountain population (SOM text). The inferred genotype at the D locus strongly affected pollen viability in the field (Fig. 3B). DD homozygotes suffered a 20% reduction in pollen viability relative to other genotypic classes. Thus, deleterious recessive effects of D contribute to male fitness variation under natural conditions. Pollen inviability may be a pleiotropic effect of meiotic interactions between paired C11.2 homologs that cause nondisjunction or may reflect hitchhiking at a locus linked to D, because the C11.2 region contains expressed genes (SOM text).

The opposition of female meiotic drive and associated male fertility costs may produce a true balanced polymorphism in M. guttatus. In a random mating population, selection against a female meiotic drive allele in homozygous males must be approximately greater than twice its advantage in heterozygous females to prevent its fixation (Eq. S1). Given the measured male fertility cost of ∼0.20 to D homozygotes, any female-specific transmission ratio distortion below 60:40 should result in a protected polymorphism. Our observation of a 58:42 transmission advantage for D is within this range, indicating that D cannot be lost and is unlikely to fix under current conditions.

Patterns of molecular polymorphism and linkage disequilibrium (LD) also suggest short-term balancing selection or an ongoing selective sweep by D (22). We estimated LD between the drive-diagnostic marker aat356 and three highly polymorphic microsatellites (6 to 19 alleles each; table S2A) that are physically associated with CycA and located at least 45 kb from aat356 (SOM text). Like aat356, these markers each had one or two alleles only found in D lines, leading to high pairwise LD with aat356 across all lines analyzed (N = 74; Fig. 4A). This was due to the low polymorphism of the inferred D lines, because inferred D lines (N = 45) were diverse and exhibited no significant LD when analyzed separately. Sequencing of the nine lines of known drive phenotype confirmed the uniqueness of the driving haplotype and revealed that drive-specific LD extends up to 2 cM (Fig. 4B and fig. S4). Thus, the driving allele D is a single physically and genetically extensive haplotype associated with the C11.2 chromosomal structure. As in maize (21), structural differences between driving and nondriving haplotypes (which may cause variation in recombination rate) complicate the interpretation of molecular population genetic data. However, the extent and uniformity of the D haplotype are consistent with recent selfish spread via female meiotic drive.

Fig. 4.

(A) LD between CycA BAC microsatellites and aat356 for all D and D lines (N = 74) and D lines only (N = 45). Asterisks indicate significant LD (P < 0.05). (B) Haplotype structure in the region flanking D (map location underlined) (13). Polymorphic sites (singletons excluded) are shown (IM62 reference sequence: yellow; alternative alleles: blue, pink, green, and red; missing data: gray). The nine lines of known heterospecific drive phenotype (D and D, respectively) are linked by vertical lines. See SOM text for methods. Full alignments are in fig. S4.

We have shown that selfish chromosomal drive has brought an allele with unconditionally deleterious effects on individual fitness to high frequency in a primarily outcrossing wildflower population. This high frequency contrasts with the generally low frequency of male drive elements (23, 24), which generally achieve excess transmission via the postmeiotic disabling of gametes with alternative genotypes, often entailing high fitness costs in heterozygotes. Female meiotic drivers such as D, which take advantage of the intrinsic asymmetry of female meiosis and may have primarily recessive costs, may spread to high frequency despite biologically significant effects. Biometric tests (10) have found more standing variation for pollen viability at Iron Mountain than predicted under mutation-selection balance models, and female meiotic drive by D may account for this unexpectedly high genetic load. By contributing to inbreeding depression for male fertility, D may play an important role in mating system and floral trait evolution in monkeyflowers.

Untangling the molecular mechanism and evolutionary origins of Mimulus drive remains a challenge, as we do not yet know whether D biases transmission by using the machinery of normal centromere function or via an alternative mechanism. Regardless of mechanism, however, it is clear that selfish chromosomal drive can be an important determinant of fitness variation within natural populations.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Tables S1 and S2

Equation S1


References and Notes

Stay Connected to Science

Navigate This Article