A Mouse Speciation Gene Encodes a Meiotic Histone H3 Methyltransferase

See allHide authors and affiliations

Science  16 Jan 2009:
Vol. 323, Issue 5912, pp. 373-375
DOI: 10.1126/science.1163601


Speciation genes restrict gene flow between the incipient species and related taxa. Three decades ago, we mapped a mammalian speciation gene, hybrid sterility 1 (Hst1), in the intersubspecific hybrids of house mouse. Here, we identify this gene as Prdm9, encoding a histone H3 lysine 4 trimethyltransferase. We rescued infertility in male hybrids with bacterial artificial chromosomes carrying Prdm9 from a strain with the “fertility” Hst1f allele. Sterile hybrids display down-regulated microrchidia 2B (Morc2b) and fail to compartmentalize γH2AX into the pachynema sex (XY) body. These defects, seen also in Prdm9-null mutants, are rescued by the Prdm9 transgene. Identification of a vertebrate hybrid sterility gene reveals a role for epigenetics in speciation and opens a window to a hybrid sterility gene network.

Hybrid sterility is one of the postzygotic reproduction isolating mechanisms that play an important role in speciation. Hybrid sterility is defined as a situation where parental forms, each fertile inter se, produce infertile offspring (1, 2). Hybrid sterility follows Haldane's rule by affecting predominantly the heterogametic sex (XY or ZW) in crosses where one sex of the progeny is sterile or missing (3). Identification of speciation genes has not been particularly successful. Despite decades of effort, only two hybrid sterility genes have been isolated, both from Drosophila species (4, 5).

Here, we report identification of a hybrid sterility gene in a vertebrate species. Hybrid sterility 1 (Hst1) is one of several genes responsible for spermatogenic failure in Mus m. musculus–Mus m. domesticus (Mmm-Mmd) hybrids (6, 7). It was genetically mapped to mouse chromosome 17 (Chr17) in hybrids between the Mmm-derived PWD/Ph inbred strain (8) and several classical laboratory strains, mostly of Mmd origin (9). Whereas most laboratory inbred strains, including C57BL/6J (B6), carry the Hst1s (sterility) allele, a few strains, such as C3H/DiSnPh (C3H) or P/J, carry the Hst1f (fertility) allele (table S1) (10). In sterile male hybrids, the Hst1 interacts, among other genes, with Hstws locus on Chr17 of Mmm subspecies. However, it remains to be determined whether Hst1 and Hstws are identical genes.

A series of high-resolution genetic mapping experiments (1113) and haplotype analyses (14, 15) localized Hst1 to a 255-kb single-copy candidate region on Chr17, harboring six protein-coding genes (Dll1, Pgcc1, Psmb1, Tbp, Pdcd2, and Prdm9) and six pseudogenes (Fig. 1A). To narrow the Hst1 critical region, we attempted rescue of the hybrid sterility phenotype by transgenesis with bacterial artificial chromosomes (BACs) derived from the C3H/HeJ strain carrying the “fertile” Hst1f allele. Four overlapping BAC clones (CHORI-34-45F17; hereafter BAC5, CHORI-34-255E14 -BAC19, CHORI-34-289M8 -BAC21, and CHORI-34-331G23-BAC24) (16, 17) were transfected into embryonic stem (ES) cells of (129 × B6)F1, predominantly of Mmd origin. The mice with BAC19 did not transmit the BAC to progeny and were not studied further. The other three BACs were transmitted, and as expected, none of them interfered with fertility after outcrossing to the B6 laboratory strain. However, in crosses to PWD (Mmm) females, BAC5 and BAC24 fully restored male fertility of the F1 hybrid males, whereas BAC21 transgenic F1 hybrid males remained sterile (Table 1). Psmb1, Tbp, and Pdcd2 were excluded as Hst1 candidates based on their expression in BAC5 (rescuing) as well as in BAC21 (nonrescuing) transgenic lines (see below). Pgcc1 was excluded because of its absence in rescuing BAC24. Previous sequencing of Dll1, Pgcc1, Psmb1, Tbp, and Pdcd2 alleles suggested that they are unlikely candidates for Hst1 (13, 14). Thus, the newly defined Hst1 critical region was restricted to the 15.9-kb interval [Chr17-15689705-15705634, National Center for Biotechnology Information (NCBI) Build 37.1] shared by rescuing BAC24 and BAC5 but absent in BAC21. This region is occupied by the 5′ end of the PR-domain 9 (Prdm9) gene and the Mrps21-rs pseudogene (Fig. 1B).

Fig. 1.

The Prdm9 gene encodes Hst1. (A) The cosegregating Hst1 region is defined by the markers CR212 and M33 (table S2). The arrows point in the direction of gene transcription; the boxes denote pseudogenes. The C3H BAC clones used for transgenesis are shown as horizontal lines with their sizes on the left. The BAC19 chimeras did not transmit the transgene (red line). The blue lines show the BACs rescuing hybrid sterility, whereas BAC21 did not rescue sterility; the C region is necessary for the rescue. (B) The Hst1 critical region. Dark blue boxes: coding exons; light blue box: untranslated region; red boxes: alternative exons (marked 5a, 5u, B2, A, S1, and S2); gray boxes: putative pseudogenes; empty boxes or vertical black lines: repetitive sequences; and asterisks: polyadenylation sites. The vertical arrow points to the site of insertion of a zinc-finger in the last exon of Prdm9 in the C3H mouse strain. The numbers at the top indicate the positions on Chr17 (in kb, NCBI m37 assembly).

Table 1.

The effect of BAC transgenes on male fertility phenotypes. TW, wet weight of paired testes; OFM, offspring per female per month; ND, not determined; F, fertile; S, sterile.

View this table:

To exclude the possibility that BAC21 did not rescue hybrid sterility because its genes were silenced in the BAC integration site, we analyzed the BAC transgenics for the expression of the genes within the Hst1 candidate region. The C3H allele-specific transcript of Tbp was found in adult testis, proving its activity in the BAC transgene (fig. S1). Psmb1 and Pdcd2 could not be tested owing to the lack of suitable polymorphism, but the dosage-dependent increase of their testicular expression (Fig. 2A) suggested that the genes within the BAC21 were active but unable to rescue the meiotic arrest. A dosage-dependent increase in Prdm9 expression was seen in the BAC5 and BAC24 hybrids but was absent in BAC21 hybrids, confirming that the Prdm9 transcript from the Hst1f allele was not present in the latter (Fig. 2A). The analysis of prepubertal fertile and sterile hybrids revealed no significant differences in mRNA expression in any of the six Hst1 candidate genes (fig. S2). These results further corroborated Prdm9 as the only candidate gene. Previously unknown testicular mRNA isoforms of Prdm9 were found (fig. S3); however, none of them exhibited reproducible differential expression between prospectively fertile and sterile prepubertal hybrid testis (fig. S4).

Fig. 2.

The effect of BAC transgenes on gene expression in testes of hybrid males. (A) Fold change of gene expression in the Hst1 critical region in BAC carriers versus their wild-type littermates. Stars indicate the presence of an intact gene within the BAC. Psmb1 and Pdcd2, predominantly expressed in spermatogonia and pachynema, show dosage-dependent increase in expression (BAC21, two copies; BAC5, two copies; BAC24, six copies). (B) Indirect immunofluorescence of γH2AX (red) and SCP1 (synaptonemal complexes, green) in pachytene spermatocytes. Most pachytene spermatocytes of fertile hybrids (Hstws/Hst1f, right) display γH2AX within the sex body. In the sterile hybrids (Hstws/Hst1s, left), the patches of γH2AX are scattered over the autosomes in most of the examined cells. Three hundred pachytene spermatocytes of each genotype were analyzed. Congenic strain B10.P carries the Hst1f allele on C57BL/10 background. (C) Transcription of Morc2b is induced in prepubertal fertile (PWD × B6.C3H-Hst1f) testes but is missing in (PWD × B6)F1 sterile hybrids. The expression at day 17.5 is rescued in a dosage-dependent manner in (PWD × BAC) hybrids by BAC5 and BAC24, but not by BAC21. The fertility status of F1 hybrids of various combinations of inbred mouse strains is shown in table S1.

Next, we sequenced the 25-kb region containing the C3H allele of Prdm9 including the 5′-flanking region (GeneBank EU719625) and found 57 differences between the B6 and C3H strains: 35 microsatellite length polymorphisms, 21 single-nucleotide polymorphisms, and one zinc-finger–encoding repeat variant. All except the zinc-finger variant were in noncoding regions. Whereas PRDM9B6 contained 13 C2H2 zinc fingers, we found that PRDM9C3H contains 14 of them (fig. S5). The pseudogene Mrps21-rs is not polymorphic between C3H and B6. Thus, Prdm9 remains the only candidate for Hst1.

The Prdm9 gene, also known as Meisetz, is expressed in testis and ovaries (18). It encodes a histone 3 lysine 4 (H3K4) trimethyltransferase. Trimethylation of histone H3K4 at promoters leads to the transcriptional activation of genes. The Prdm9-null mice show arrest of spermatogenesis and oogenesis at pachynema, impairment of double-strand break repair, chromosome asynapsis, and disrupted sex-body formation (18). To further verify Prdm9 as Hst1, we compared the phenotypes of the sterile Hstws/Hst1s (PWD × B6)F1 males with the published phenotypes of Prdm9–/– mutants. Sterile hybrids have small testes with spermatogenic arrest predominantly during pachytene and no sperm in the seminiferous tubules (6, 19, 20) (Table 1). Rare surviving primary spermatocytes at diakinesis–metaphase I manifest two to six univalents and frequent X-Y dissociation (6), resembling the impairment of synapsis between homologous chromosomes in the Prdm9–/– mutants. Both sterile hybrids and Prdm9–/– mice display abnormal sex-body formation in pachytene spermatocytes (Fig. 2B). Prdm9–/– pachytene spermatocytes lack a sex body and exhibit patches of γH2AX over the synaptonemal complexes (18). We observed a comparable failure of sex-body formation with scattered γH2AX in 60% of pachytene spermatocytes in sterile (PWD × B6)F1 hybrids versus 7% in fertile hybrid controls (Fig. 2B).

The microrchidia 2b gene, Morc2b, or 4932411A10Rik, encoding a gonad-specific protein, is directly induced by Prdm9. Similar to Prdm9–/– testis (18), we found that Morc2b mRNA is barely detectable in sterile hybrids. The Morc2b expression in hybrid males was restored by the Prdm9-containing BACs (Fig. 2C). Transcription from Morc2b corresponds to the amounts of histone H3K4 trimethylation controlled by the enzymatic activity of PRDM9 (18). Chromatin immunoprecipitation revealed decreased H3K4 trimethylation of Morc2b in sterile hybrid testis (Fig. 3).

Fig. 3.

Histone 3 K4 trimethylation in prepubertal hybrid testis by chromatin immunoprecipitation. Gray and white columns show values for 17-day-old sterile and fertile hybrids, respectively. The quantified regions are indicated at the bottom. The values of immunoprecipitated DNA on the y axis are normalized by input DNA (nonimmunoprecipitated positive control). The Psmb1 promoter and the Tbp-Pdcd2 3′-intergenic region served as positive and negative control for H3K4 trimethylation, respectively. The Psmb1 gene was selected as a control because it is expressed at the same levels in prepubertal hybrid testis (fig. S3). The probability (P) of the difference was determined by the Welsch's t test; *P = 0.02.

Two differences were observed between sterile hybrids and Prdm9-null mutants. Following Haldane's rule (3), hybrid sterility is male-limited, yet meiotic arrest of Prdm9–/– mice affects both sexes. This discordance could be explained by incompatible domesticus-musculus epistatic interaction(s) of the Prdm9 gene in sterile hybrids in contrast to the complete silencing of Prdm9 in the knockout. Similar meiotic effects, sterility of both sexes or dominant male-limited sterility, have been described for the null and missense mutations of the Dmc1 gene (21), respectively. Second, the Prdm9-null mutation acts independently, whereas meiotic arrest in F1 hybrids results from the epistatic interaction of the Hst1 gene (Hstws/Hst1s) with several independently segregating genes (10). Hybrids between the consomic strain B6. PWD-Chr17 and B6 (22), as well as hybrids (B6 female × PWD male, Table 1), carry the “sterile” Hstws/Hst1s genotype but are fertile due to their lack of an interaction of Hst1 with other hybrid sterility genes (10, 23).

The parallel between the role of Hst1 in mouse hybrid sterility and the role of Lhr in hybrid male inviability of Drosophila is striking. In both cases, a variant form able to rescue hybrid incompatibility was found within a species. It behaved as an autosomal locus, Hst1f within Mmd and Lhr in the case of D. simulans, and interacted with an X-linked genetic factor (7, 23, 24). Finally, both in the mouse and Drosophila, the two loci were necessary but not sufficient to reconstitute the hybrid incompatibility phenotype.

Our data show that Prdm9, known to activate genes essential for meiosis by methylation of histone H3 at lysine 4, is the only candidate for Hst1. It is the only known gene located within the newly defined 15.9-kb hybrid sterility 1 critical region, expressed at the right tissue and at the right time of germ cell differentiation (primary spermatocytes). The pertubation of Prdm9 function observed in sterile hybrid males corresponds to the phenotype of the Prdm9–/– mutants.

The genes that reduce hybrid fitness because of their divergent evolution can be the cause or the consequence of speciation, depending on whether they evolved before or after the complete reproductive isolation of the studied taxa (25, 26). The advantage of our mouse model is that reproductive isolation of Mmm and Mmd is still incomplete. Thus, Prdm9 may be an essential component of a Dobzhansky-Muller incompatibility that is part of an incipient speciation event. It can lead us to the Dobzhansky-Muller incompatibility gene(s) that interferes with the normal meiotic function of histone H3K4 methyltransferase. The meiosis-specific function of Prdm9 can explain the breakdown of meiotic cells with no effect on somatic tissues in intersubspecific hybrids. Uncovering Prdm9 as a hybrid sterility gene will permit us to search for the epigenetically regulated downstream genes and their role in the hybrid sterility gene network.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article