Accelerated evolution of the Prdm9 speciation gene across diverse metazoan taxa

Abstract

The onset of prezygotic and postzygotic barriers to gene flow between populations is a hallmark of speciation. One of the earliest postzygotic isolating barriers to arise between incipient species is the sterility of the heterogametic sex in interspecies' hybrids. Four genes that underlie hybrid sterility have been identified in animals: Odysseus, JYalpha, and Overdrive in Drosophila and Prdm9 (Meisetz) in mice. Mouse Prdm9 encodes a protein with a KRAB motif, a histone methyltransferase domain and several zinc fingers. The difference of a single zinc finger distinguishes Prdm9 alleles that cause hybrid sterility from those that do not. We find that concerted evolution and positive selection have rapidly altered the number and sequence of Prdm9 zinc fingers across 13 rodent genomes. The patterns of positive selection in Prdm9 zinc fingers imply that rapid evolution has acted on the interface between the Prdm9 protein and the DNA sequences to which it binds. Similar patterns are apparent for Prdm9 zinc fingers for diverse metazoans, including primates. Indeed, allelic variation at the DNA-binding positions of human PRDM9 zinc fingers show significant association with decreased risk of infertility. Prdm9 thus plays a role in determining male sterility both between species (mouse) and within species (human). The recurrent episodes of positive selection acting on Prdm9 suggest that the DNA sequences to which it binds must also be evolving rapidly. Our findings do not identify the nature of the underlying DNA sequences, but argue against the proposed role of Prdm9 as an essential transcription factor in mouse meiosis. We propose a hypothetical model in which incompatibilities between Prdm9-binding specificity and satellite DNAs provide the molecular basis for Prdm9-mediated hybrid sterility. We suggest that Prdm9 should be investigated as a candidate gene in other instances of hybrid sterility in metazoans.

Figure 1. Schematic of Prdm9 protein encoded by M. musculus.

Schematic of the domain architecture for the long protein isoform encoded by the M. musculus Prdm9 gene. The Prdm9 protein contains KRAB, SSXRD, and SET domains and a single zinc finger in its N-terminal half, while the C-terminal half consists of an array of zinc finger domains . The shorter Prdm9 protein isoforms lack the C-terminal zinc fingers and apparently do not localize to the nucleus. Sterility and fertility associated alleles of Prdm9 in M. m. musculus differ only in one extra zinc finger (red triangle) .

Figure 2. Concerted evolution among rodent Prdm9 genes.

(A) Prdm9 C-terminal zinc fingers for 13 rodent species are shown as pink rectangles. Zinc fingers whose nucleotide sequences are identical are joined by solid lines. Zinc fingers with identical sequences from the same species are consistent with gene conversion and/or intra-exon duplication. A phylogeny of these species is also shown with estimated divergence dates (indicated at nodes) given in millions of years (my) ,. Common names to species are listed in the legend to Figure 6. (B) The proportion of pairwise cDNA comparisons between aligned zinc fingers from the same gene (see Materials and Methods) which show greater than 90% identity. All mouse Prdm9 zinc fingers are more than 90% identical to all other C-terminal zinc fingers in the same protein (indicated in red), a much higher fraction than for any other zinc finger protein encoded by the mouse genome.

Figure 3. Positive selection of zinc fingers encoded by rodent Prdm9 genes.

(A) A multiple alignment of the zinc finger sequences from M. musculus C57BL/6J highlights the invariant Cys₂His₂ Zn²⁺-coordinating residues as well as positions −1, 3, and 6 that dictate the DNA-binding specificity of individual zinc fingers. Deviations from the consensus amino acid at each position are shown in boldface. In this species, positions −1 and 3 meet the criteria for positive selection (highlighted in yellow and with red crosses). (B) Predicted positively selected sites in Prdm9 from diverse rodent lineages. Positive selection was inferred for each species from intra-species Prdm9 zinc finger sequence alignments. Positively selected sites (P<0.05 after multiple testing correction) are shown mapped to the third mouse Prdm9 zinc finger sequence (MMM3). The majority of positively selected sites fall at positions −1, 3, and 6. (C) Estimated d_N/d_S values at four zinc finger positions (namely, −2, −1, 3, and 6) in a comparison of zinc fingers from all rodents (in contrast to the analyses of species-specific zinc fingers in (B)) for which there exists strong evidence of positive selection . The P-values shown have been corrected for multiple testing. Common names to species are listed in the legend to Figure 6.

Figure 4. Concerted evolution and positive selection among primate PRDM9 genes.

(A) PRDM9 C-terminal zinc fingers for 6 primate species are shown as pink rectangles. Zinc fingers whose nucleotide sequences are identical are joined by solid lines. Zinc fingers with identical sequences from the same species are consistent with gene conversion and/or intra-exon duplication. (B) Predicted positively selected sites in Prdm9 from divergent primate lineages. Positive selection was inferred for each species from intra-species Prdm9 zinc finger sequence alignments. Positively selected sites (P<0.05 after multiple testing correction) are shown mapped to the third mouse Prdm9 zinc finger sequence (MMM3) as shown in Figure 3A. (C) Estimated d_N/d_S values at three zinc finger positions (namely −1, 3, and 6) in a comparison of zinc fingers from all primates for which there exists strong evidence of positive selection . The P-values shown have been corrected for multiple testing. Common names to species are listed in the legend to Figure 6.

Figure 5. Sequence divergence and diversity among human and chimpanzee PRDM9 zinc finger sequences.

(A) Multiple sequence alignment of human (Homo sapiens) PRDM9 zinc finger sequences, with positively selected positions (P<0.05, after multiple testing correction) indicated by red asterisks interspersed among a consensus amino acid sequence. Positions −1, 3, and 6 (numbered relative to the start of the zinc finger α–helix) that represent sequence-variable positions frequently involved in DNA binding are also indicated. Codons highlighted in green are not found at the same position in any chimpanzee PRDM9 zinc finger. (B) Multiple sequence alignment of chimpanzee (Pan troglodytes) PRDM9 zinc finger sequences with a predicted positively selected site indicated as in panel (A). Note that several chimpanzee PRDM9 zinc finger codons (highlighted in green) at positions −1 and 3 are unique to this species, relative to humans (A). (C) Numbered and boxed codons in panel (A) contain human nonsynonymous SNPs. SNPs numbered 1–7 were identified in this study among 50 Chinese individuals whilst heterozygous SNPs numbers 3, 6, 8, and 9 are significantly enriched among fertile, as opposed to infertile, males in the study by Irie et al. .

Figure 6. Predicted positively selected sites in Prdm9 from divergent metazoan lineages.

Results for previously presented rodent and primate lineages are also shown here for comparison (blue shading). Positive selection was inferred for each species from intra-species Prdm9 zinc finger sequence alignments. Positively selected sites (P<0.05 after multiple testing correction) are shown mapped to the third mouse Prdm9 zinc finger sequence (MMM3). The majority of positively selected sites, across 700 million years of divergence from sea anemone to mammals, fall at positions −1, 3, and 6. The inferences of positive selection for Capitella were made on the basis of three sequences on separate unassembled genomic scaffolds. Despite their high sequence similarity, multiple uncorrelated point substitutions, especially among the zinc fingers, suggest that they may represent allelic copies or rapidly diverging paralogues.

Figure 7. A novel satellite–DNA binding model for Prdm9 in hybrid sterility.

(A) Prdm9 could serve as a satellite-DNA binding protein that facilitates its heterochromatinization. Hybrid sterility ensues when the sterility-associated Prdm9 protein (blue) cannot bind to “newly expanded” satellite DNA repeats (red or green) potentially at pericentric regions of chromosomes that arose due to centromere-drive . (B) Under this model, in isolated populations, satellite–DNAs diverge quickly by “centromere-drive” and Prdm9 DNA–binding specificity evolves rapidly to suppress this drive. However, in hybrid males, inappropriate localization of Prdm9 to diverged DNA–binding satellites from other species would result in either inappropriate chromosome condensation (as shown) or compromised centromere function (not shown), either of which would result in male sterility.