DISSECTING THE GENETIC ARCHITECTURE OF F1 HYBRID STERILITY IN HOUSE MICE

Authors


Abstract

Hybrid sterility as a postzygotic reproductive isolation mechanism has been studied for over 80 years, yet the first identifications of hybrid sterility genes in Drosophila and mouse are quite recent. To study the genetic architecture of F1 hybrid sterility between young subspecies of house mouse Mus m. domesticus and M. m. musculus, we conducted QTL analysis of a backcross between inbred strains representing these two subspecies and probed the role of individual chromosomes in hybrid sterility using the intersubspecific chromosome substitution strains. We provide direct evidence that the asymmetry in male infertility between reciprocal crosses is conferred by the middle region of M. m. musculus Chr X, thus excluding other potential candidates such as Y, imprinted genes, and mitochondrial DNA. QTL analysis identified strong hybrid sterility loci on Chr 17 and Chr X and predicted a set of interchangeable autosomal loci, a subset of which is sufficient to activate the Dobzhansky–Muller incompatibility of the strong loci. Overall, our results indicate the oligogenic nature of F1 hybrid sterility, which should be amenable to reconstruction by proper combination of chromosome substitution strains. Such a prefabricated model system should help to uncover the gene networks and molecular mechanisms underlying hybrid sterility.

Hybrid sterility belongs to postzygotic reproductive isolation mechanisms that contribute to speciation by restricting gene flow between closely related taxa. It is defined as a situation where two parental forms, each of which is fertile inter se, produce a hybrid that is sterile (Dobzhansky 1951). Hybrid sterility is a universal phenomenon described in such diverse organisms as yeast, plants, insects, birds, or mammals. Mus m. musculus and M. m. domesticus are relatively young subspecies of house mouse, which diverged from their common ancestor ∼350,000 years ago (She et al. 1990; Geraldes et al. 2011). Thus they can be considered species in statu nascendi, a true model of choice in studies of mammalian speciation. In Europe, M. m. domesticus occupies the western part and M. m. musculus the eastern part of the continent. They form a narrow hybrid zone in the regions of secondary contacts with a limited exchange of genes in both directions (Payseur et al. 2004; Macholan et al. 2007; Teeter et al. 2008; Dufkova et al. 2011). To study the molecular mechanisms underlying hybrid sterility in mice, the existing extensive set of genetic and genomic tools primarily developed for biomedical studies can be used. The genomic sequence of C57BL/6J inbred strain became the reference mouse genome (Waterston et al. 2002) and more recently, the genomic sequences of additional 17 inbred strains including wild-derived strains from M. m. musculus and M. m. domesticus subspecies were reported (Keane et al. 2011). Null alleles of more than half of the mouse protein coding genes were induced by homologous recombination in embryonic stem cells, mostly of C57BL/6 origin, and have been made available for functional studies (Skarnes et al. 2011). The house mouse is therefore a highly suitable species to answer such basic questions as: (1) How many genetic loci control F1 hybrid sterility? (2) Where in the genome are hybrid sterility loci located? (3) Do all hybrid sterility loci harbor coding genes and what is their physiological function within the species? (4) What molecular mechanisms underlie the infertility of hybrids? (Dobzhansky 1951; Wu and Hollocher 1998; Coyne and Orr 2004; Maheshwari and Barbash 2011).

Genetic rules of reproductive isolation can be inferred from wild mice sampled across a hybrid zone by correlating their fertility traits with species–specific genomic sequences (Payseur et al. 2004; Vyskocilova et al. 2005; Macholan et al. 2007; Teeter et al. 2008; Teeter et al. 2010). Such studies give an overall picture of true genomic and phenotypic variations, but their genetic resolution is usually low. Alternatively, the inbred strains derived in the laboratory from the extant (sub)species can be used, which enable higher resolution of genetic and molecular studies, but at a cost: instead of analyzing the dynamic genetic structure of real populations, the inbred strains provide a single genomic snapshot, not necessarily the most typical of a given species. In this study, and with this caution in mind, we are using genetically defined inbred strains for genetic dissection of the mechanisms underlying hybrid sterility in the house mouse.

In classical studies, hybrid sterility is defined as infertility of F1 hybrids between the studied taxa. However, introgression experiments and F2 crosses clearly showed that many hybrid sterility genes can be recessive, hence by definition cannot participate in F1 sterility (Masly and Presgraves 2007; White et al. 2011). Our present study was deliberately limited to a single model of F1 hybrid sterility defined by crosses of PWD/Ph and C57BL/6J (henceforth PWD and B6) inbred strains representing M. m. musculus and M. m. domesticus subspecies. Our long-term goal is to use these genetically defined inbred strains and their derivatives to decipher complete genomic architecture of this F1 hybrid sterility model and the underlying regulatory networks.

The first hybrid sterility gene in the mouse was mapped in crosses between wild mice of M. m. musculus and laboratory inbred strains C3H/J and C57BL/10ScSnPh (B10), predominantly of M. m. domesticus origin. The mapping was possible because C3H and B10 strains differ in regard to hybrid sterility phenotype in only one gene, Hst1. The Hst1s allele (of B10 origin) but not Hst1f allele (of C3H origin) interacts with the Hstws allele of M. m. musculus in F1 hybrid background to ensure sterility of hybrid males showing early meiotic arrest, low testes weight (TW), and no sperm in ductus epidydimis (Forejt and Ivanyi 1974). We used high-resolution genetic and haplotype mapping and PWD/Ph inbred strain as a representative of M. m. musculus (Gregorova and Forejt 2000) to locate the Hst1 gene to a 250 kb interval on Chr 17 harboring seven protein coding genes (Forejt et al. 1991; Gregorova et al. 1996; Trachtulec et al. 1997, 2008). In the subsequent BAC transgenic rescue experiment we identified the Hst1 locus with Prdm9 gene encoding meiotic histone H3K4 trimethyl transferase (Mihola et al. 2009).

Here we identify the X-chromosomal region responsible for asymmetry of reciprocal F1 hybrid sterility and investigate the number, location, and interaction of hybrid sterility loci engaged in early meiotic arrest of F1 hybrid males between PWD and B6 inbred strains representing M. m. musculus and M. m. domesticus subspecies.

Materials and Methods

ANIMALS AND EXPERIMENTAL DESIGN

The PWD/Ph (abbreviated PWD) inbred strain was created within 1972–1980 period from a pair of wild mice of M. m. musculus origin trapped in locality Kunratice, near Prague, Czech Republic (Gregorova and Forejt 2000). The C57BL/6J inbred strain (B6) was imported from the Jackson Laboratory in 1998. Recently, 93% and 95% of genomic sequence of PWD and B6 was confirmed to be of M. m. musculus and M. m. domesticus origin, respectively (Yang et al. 2011). Each chromosome substitution strain C57BL/6J-Chr #PWD (here abbreviated B6.PWD-Chr#) carries a single PWD chromosome that replaced its B6 homolog on otherwise B6 genetic background (Gregorova et al. 2008). All mice are maintained in the Specific Pathogen Barrier facility of the Institute of Molecular Genetics, Academy of Sciences of the Czech Republic. The principles of laboratory animal care obeyed the Czech Republic Act for Experimental Work with Animals (Decree No. 207/2004 Sb., and Acts Nos. 246/92 Sb. and 77/2004 Sb.), fully compatible with the corresponding EU regulations and standards, namely, Council Directive 806/609/EEC and Appendix A of the Council of Europe Convention ETS123.

To estimate the number and location of hybrid sterility genes we setup a backcross of (PWD × B6)F1 females mated to B6 males and analyzed fertility parameters in 254 N2 males. The information from QTL analysis of N2 males was complemented by crosses involving the B6.PWD-Chr# chromosome substitution strains. To test the role in hybrid sterility of each autosome separately, we checked the fertility parameters of the male progeny of PWD females and B6.PWD-Chr# males. The crosses of B6.PWD-Chr X1, B6.PWD-Chr X2, and B6.PWD-Chr X3 females, carrying proximal middle or distal part of Chr X, with PWD males localized the genetic locus responsible for asymmetry of infertility of reciprocal F1 hybrids between B6 and PWD strains.

Phenotyping and Genotyping

Mice were sacrificed at 9 weeks of age and wet weight of both testes was determined to the nearest milligram immediately after dissection. For sperm count, whole epididymis including caput, corpus, and cauda were removed and cut in cold phosphate-buffered saline. The released sperm cells were counted in a Bürker chamber. Three quantitative phenotypes were collected: body weight (BW), TW, and sperm count. Log2(1 +x) transformation of sperm count, denoted as logSC, was applied to achieve normality and used for QTL analysis. Genomic DNA was extracted from liver by standard phenol/chloroform isolation method combined with Phase Lock Gel (5 Prime) to provide greater yield and higher purity. A set of 86 simple sequence length polymorphisms (SSLP) (MIT) markers were used to genotype backcross males (3–6 per chromosome, equidistantly spaced). Additional 14 MIT markers were added in the regions of interest. Identifiers, positions of markers (NCBI build 37 July 2007) and genotyping data are provided in Table S1. PCR amplification was done as described (Storchova et al. 2004).

The marker order was checked by pairwise calculation of recombination fractions and positioning markers along the chromosome. D4Mit51 and D10Mit75 were removed because they appeared to be incorrectly positioned (data not shown). GeneBank accession numbers of primers for markers D17Zt634, DXSr62 are JQ685512 and JQ685513. Genotypes were also pairwise compared to search for possible duplicated samples but no errors were identified. The dataset was submitted to QTL Archive at The Jackson Laboratory (http://www.qtlarchive.org).

QTL Analysis

R 12.1 (R_Development_Core_Team 2008) and its qtl package (Broman et al. 2003; Broman and Sen 2009) were used to perform statistical analysis. Marker positions were taken from MGI mouse genetic map (Bult et al. 2008). Standard interval mapping was implemented using scanone function. TW and logSC were modeled as continuous variables, fertility/subfertility group as a binary variable. Genotype probabilities between markers were calculated at a grid size of 5 cM and with genotyping error rate of 0.01%. Genome-wide significance was calculated by 1000 permutations and compared to α= 5% threshold. For two-dimensional scans to detect epistasis we used the scantwo function to identify combinations of loci where the full model LOD score (logarithm (base 10) of odds) exceeded its permutation thresholds and either the conditional-interactive model LOD or the interaction model LOD exceeded their threshold.

Results

ASYMMETRY OF RECIPROCAL F1 HYBRID MALE STERILITY IS CONTROLLED FROM THE MIDDLE PART OF CHR X

Asymmetry of male infertility is a general phenomenon in reciprocal interspecific hybrids of various species. It has been reported in Drosophila (Zeng and Singh 1993; Turelli and Moyle 2007; Reed et al. 2008) as well as in mouse intersubspecific crosses (Good et al. 2008b; Pialek et al. 2008; Mihola et al. 2009). Hybrid males (PWD × B6)F1 (PWD female crossed with B6 male) had low weight of paired testes (66.8 ± 3.4 mg), no sperm in ductus epidydimis and were sterile, while males from the reciprocal cross (B6 × PWD)F1 were affected by only a partial spermatogenic arrest compatible with fertility (TW 107 ± 8 mg and almost 3 × 106 sperm cell count, see Fig. 1, Tables S2 and S3). The asymmetry in F1 sterility could be explained by any of the following five mechanisms: (1) X-autosomal interaction, (2) Y-autosomal interaction, (3) X-Y incompatibility, (4) mitochondrion-nuclear incompatibility, or (5) by incompatibility of imprinted autosomal gene(s). To discriminate between these options, we crossed B6.PWD-Chr X.1, X.2, or X.3 consomic females identical with B6 but carrying the proximal, middle, or distal part of Chr (Chromosome) X of PWD origin (Gregorova et al. 2008) with PWD males. In all three crosses the offspring received B6 and PWD autosomal sets from the opposite parents compared to original (PWD × B6)F1 males, moreover the Chr Y was of PWD origin (Table S2). The male progeny of B6.PWD-Chr X.2 females showed typical F1 hybrid sterility phenotype with small testes (TW = 60.9 ± 4.3 mg) and no sperm, while crosses of B6.PWD-Chr X.1 or X.3 consomic females yielded fertile males. Based on the known distal PWD border of B6.PWD-Chr X.1 and proximal border of B6.PWD-Chr X.3, the region carrying the PWD-specific incompatibility locus is delimited to 61.0–94.3 Mb interval. The X-Y interaction as a possible cause of asymmetry was independently tested in B6.PWD-Chr Y consomic strain hybrids. The (PWD × B6.PWD-Chr Y)F1 males carried both sex chromosomes of PWD origin but were sterile. Moreover, the effect of PWD mitochondria was excluded because hybrids of B6.PWD-Mit conplastic females and PWD males were indistinguishable from fertile B6 × PWD hybrids (Table S2). In contrast to the present data from genetic crosses, number of studies of the European hybrid zone of M. m. musculus and M. m. domesticus indicated a differential inhibition of Y chromosome introgression suggestive for reproductive isolation (Vanlerberghe et al. 1986; Tucker et al. 1992, Macholan et al. 2008). Recently Ellis et al. (2011) described a model of X-Y incompatibility based on genomic conflict between Sly and spermatid-expressed genes. Such interaction would not function in our model of F1 hybrid sterility because the spermatogenic block occurs at primary spermatocytes. However, it could play a role in postmeiotic hybrid sterility caused by introgression of the X chromosome (musculus or molossinus) into M. m. domesticus genome (Storchova et al. 2004; Oka et al. 2004).

Figure 1.

Box plot ranges for male fertility parameters of reciprocal F1 hybrids and hybrids of selected chromosome substitution strains. TW = testes weight; logSC = log sperm count. Chr # is an abbreviated designation for a B6.PWD-Chr# chromosome substitution strain.

In conclusion, we excluded the X-Y interaction, the autosome-Y or mitochondrial incompatibility and monoallelic expression of imprinted autosomal genes as the cause of asymmetry of the F1 hybrid fertility, leaving the Dobzhansky–Muller (D-M) incompatibility (Dobzhansky 1951) of the middle part (61.0–94.3 Mb) of the XPWD (M. m. musculus) chromosome with the heterospecific hybrid autosomal genome as the sole cause of the asymmetry in reciprocal F1 hybrids.

ESTIMATING THE NUMBER OF F1 HYBRID STERILITY GENES

The number and location of genetic factors involved in the control of M. m. musculus×M. m. domesticus hybrid sterility was estimated by QTL analysis of male progeny from (B6 × PWD) × B6 backcross (N2 generation). First, we asked how many out of the total 254 N2 males recapitulate the F1 hybrid sterility phenotype. The absence of sperm in ductus epididymis or drastic reduction of sperm count was considered as the most reliable proxy for the sterility phenotype. Twenty-one out of 254 males (8.3%) were azoospermic or oligospermic (sperm count 0.0–0.9 × 106) with small testes (average wet weight of paired testes 64.3 mg, range 48–105 mg). Such frequency could correspond to four independently segregating hybrid sterility loci (but see section Discussion) provided that simultaneous PWD/B6 heterozygosity at all hybrid sterility loci is the necessary condition for the spermatogenic arrest. Further requirements are that the same form of meiotic arrest should not occur in N2 males by allelic combination of genes unrelated to F1 hybrid sterility genes and that the B6/B6 allelic combination of any unrelated gene should not suppress the sterilizing effect of F1 hybrid sterility genes. Under these premises the frequency of sterile males should follow the formula: F= 100/2n, where F is the percentage of sterile males in the population of N2 males and n is the number of independently segregating F1 hybrid sterility genes.

MAJOR HYBRID STERILITY QTLs ARE LOCATED ON CHR 17 AND CHR X

The fertility parameters were followed in backcross males as quantitative traits by scoring wet weight of paired testes (TW) and log-transformed sperm count (logSC see section Material and Methods). BW was followed to check for possible correlation with the fertility traits. Histograms and kernel density estimators (Venables and Ripley 2002) of TW and logSC suggested that the distributions are mixtures of values corresponding to fertile, subfertile, and sterile males (Fig. S1). TW and logSC were positively correlated (Spearman's ρ= 0.67, P < 0.001) but neither of them showed significant correlation to BW (Spearman's ρ= 0.11, P > 0.10). Therefore, we used absolute values of the weight of paired testes in milligrams. TW is highly heritable (Le Roy et al. 2001) but it is influenced not only by early arrest of spermatogenesis but also by physiological variation in testes size between parental B6 and PWD strains. Sperm count reflects better the spermatogenic breakdown, but the technical variation is higher than in TW. Thus, we modeled the joint distribution of TW and logSC as a mixture of two Gaussian two-dimensional distributions (flexmix function, Leisch 2004), which allowed us to assort the animals into sterile/subfertile and fertile groups (Fig. S2) comprising 31 and 223 mice, respectively.

For genotyping, 23 most sterile/subfertile and 17 most fertile (highest SC) males were selected from the extreme ends of the fertility parameter distribution (Silver 1995). Single QTL interval mapping with 74 informative SSLPs (see section Materials and Methods) identified two highly significant QTLs for both TW and SC, one on Chr 17 and one on Chr X (Fig. 2, Table 1). These QTLs mapped to chromosome intervals with already known hybrid sterility loci defined by another approach; QTL on Chr 17 with the highest LOD score 5.42 (1.5 LOD interval 0–34.9 Mb) overlaps Hst1/Prdm9 on Chr 17 (Forejt and Ivanyi 1974; Forejt 1996; Mihola et al. 2009) and QTL on Chr X with the highest LOD score 8.85 (1.5 LOD interval 35.3–88.5 Mb) overlaps Hstx1 responsible for defect in spermiogenesis on Chr X (Storchova et al. 2004), see also (Oka et al. 2004; Oka et al. 2007; White et al. 2011). All 23 sterile/subfertile animals carried both, PWD allele of the DXSr62 marker on Chr X and PWD/B6 heterozygosity of the D17Zt634 locus on Chr 17. A clear epistatic interaction was observed between QTLs on Chr 17 and Chr X because the simultaneous presence of incompatible genotypes at both chromosomes was necessary condition for hybrid sterility (Fig. 3).

Figure 2.

Single QTL scan for hybrid sterility gene in the backcross (B6 × PWD) × B6 identifies loci on Chr 17 and Chr X. LOD scores plotted at 5-cM intervals separately for testes weight (TW) and sperm count (logSC). Genome-wide significance threshold (α= 0.05) is indicated by the dashed lines and was derived independently for the autosomes and Chr X from 1000 permutations.

Table 1.  Hybrid sterility QTLs in (B6 × PWD) × B6 backcross males.
PhenotypeChrLOD scoreMaximum LOD attained atPosition (cM)1.5 LOD interval (cM)Position (Mb)Interval (Mb) 1.5 LODCandidate loci/genes
  1. 1Group refers to binary distribution of N2 generation into “fertile” and “subfertile” males, as shown in Figure S2.

Group1175.42 D17Mit19, D17Zt634 3.0, 8.20.0–18.84.8, 15.70–34.9 Hst1/Prdm9
TW 17 4.86 D17Mit19, D17Zt634 3.0, 8.2 0.0–18.8 4.8, 15.7 0–34.9 Hst1/Prdm9
logSC175.67 D17Mit19, D17Zt634 3.0, 8.20.0–18.84.8, 15.70–34.9 Hst1/Prdm9
Group X 8.82 DXMit140, DXMit76, DXSr62 19.0, 20.00, 26.7 11.4–36.4 56.9, 62.0, 65.5 35.3–88.5 Hstx1
TWX10.75 DXMit140, DXMit76, DXSr62 19.0, 20.00, 26.76.4–36.456.9, 62.0, 65.521.1–88.5 Hstx1
logSC X 14.63 DXMit140, DXMit76, DXSr62 19.0, 20.00, 26.7 6.4–36.4 56.9, 62.0, 65.5 21.1–88.5 Hstx1
Group144.58 D14Mit258, D14Mit183 17.0, 19.011.5–40.050.0, 52.634.1–60.3?
TW 14 3.36 D14Mit258, D14Mit183 17.0, 19.0 0.7–40.0 50.0, 52.6 12.9–60.3  
logSC146.37 D14Mit258, D14Mit183 17.0, 19.015.0–22.550.0, 52.647.7–57.2?
TW 13 3.03 Between D13Mit202 D13Mit35 61.0 36.0–75.0 111.2 60.1–120.1  
Figure 3.

Epistatic interaction between QTLs on Chr17 (D17Zt634) and ChrX (DXSr62). Only combinations of PWD (P) allele of DXSr62 and B6/PWD (BP) heterozygosity at D17Zt634 can be found in sterile-subfertile males. Incompatible (sterility) genotype at only one of these two chromosomes had no detectable effect o fertility. All 254 N2 males are shown, phenotypic means ± SE are indicated for each genotype.

CHR X AND CHR17 ARE NECESSARY BUT NOT SUFFICIENT: SEARCH FOR ADDITIONAL HYBRID STERILITY LOCI

From 40 males genotyped for QTL analysis, all 23 sterile males were PWD at DXSr62 locus and PWD/B6 at D17Zt634. This D-M incompatibility was not seen in any of the 17 genotyped fertile males. Therefore, we genotyped all remaining 214 animals at the same two loci and found an additional 47 males with the same D-M incompatibility. Taking all 254 N2 males together, 70 of them displayed incompatible allelic combination on Chrs 17 and X, from which 31 (44%) were sterile or subfertile. None of the males with compatible genotypes on Chr17 (B6/B6) and Chr X (B6/-) was sterile (Fig. 3).

In the next step, we completed genome-wide genotyping of these 70 animals and search for the missing hybrid sterility loci. Two QTL loci reaching or exceeding the level of significance were found, one on Chr 14 (D14Mit258-D14Mit183) for both fertility traits and the other on Chr 13 (D13Mit202-D13Mit35) for TW only (Fig. 4). At this step the subfertile phenotype did not completely segregate with PWD/B6 heterozygosity of the QTL marker. Eight out of 39 males with “fertility” B6/B6 genotype at D14Mit183 were sterile/subfertile (Fig. 5). Moreover, the QTL for TW on Chr 13 conferred the opposite effect, the PWD/B6 heterozygotes had larger testes than B6B6 homozygotes. This locus obviously does not belong to the hybrid sterility network and could reflect the genetic component of physiological variation (Le Roy et al. 2001). Our result could be explained by assuming a set of multiple and weak hybrid sterility loci (including the QTL on Chr 14), of which a subset is sufficient for interaction with the two strong loci (Hst1/Prdm9 and Hstx1) to ensure full manifestation of hybrid sterility (Fig. 5). No other interacting genes were identified by two-locus scan. Thus, no other strong loci comparable to those on Chr 17 and Chr X were found. An alternative, not mutually exclusive explanation presumes that a certain threshold level of PWD/B6 heterospecific heterozygosity and/or sequence micro-nonhomology in the genetic background is a conditio sine qua non for the sterilizing effect of Hst1/Prdm9Hstx1 incompatibility. The idea gets some support from the observed correlation between sterility and total heterozygosity of the genome in the 70 males selected for incompatible genotype at Chr 17 and Chr X, even if the markers of Chr 17 and Chr X are removed from the comparison (Spearman's rank correlation rho =−0.5, P < 0.001 in both cases, see Fig. 6). The correlation still remained significant when the markers of Chr 14 were additionally removed (not shown).

Figure 4.

Single QTL scan of a selection of 70 mice with incompatible “sterility“ allelic genotypes on Chr17 and ChrX. LOD scores plotted at 5-cM intervals separately for testes weight (TW) and sperm count (logSC). Genome-wide significance threshold (α= 0.05) is indicated by the dashed lines and was derived independently for the autosomes and Chr X from 1000 permutations.

Figure 5.

Recursive partitioning of phenotypic traits according to identified QTLs. Means ± SEMs (TW, logSC) and counts (sterile/subfertile vs. fertile groups) are reported in the initial and terminal nodes. Segregation of sterility/subfertility is absolute in the case of DXSr62 (Chr X) and D17Zt634 (Chr 17). QTL on Chr 14 (D14Mit183) behaves as PWD dominant (compare with Fig. 7), while QTL on Chr 13 reflects most probably a genetic variation in size of fertile testes (see Le Roy et al. 2001).

Figure 6.

Correlation between the percentage of markers with heterozygous allele combination and phenotypic traits: testes weight (left) and log-sperm count (right). The dashed lines show the linear regression. The P-values for the Spearman rank correlation test are given in top right corner.

We confirmed the indispensable role of these minor hybrid sterility loci by reconstructing the F1 sterility genotype of Chr X and Chr 17 in males with B6 genetic background. The reconstruction was achieved by crosses of consomic females B6.PWD-Chr X and B6.PWD-Chr 17 males. The male progeny carried Chr XPWD and a PWD/B6 heterosomic pair of Chr 17, yet produced sperm (TW 141.6 ± 19.6 mg, SC 19.7 ± 8.2 million). Limited data on males with additional Chr 14PWD/B6 indicate that even the proper genotype of these three chromosomes was not sufficient to reconstitute the F1 sterility (data not shown).

In conclusion, the backcross to the B6 (M. m. domesticus) parent identified two strong hybrid sterility loci on Chr 17 and Chr X, whose epistatic interaction is obligatory but not sufficient for meiotic arrest in F1 hybrid sterility. The dataset was apparently underpowered to unambiguously identify additional hybrid sterility loci, which we predict to be weak, multiple, and mutually replaceable.

EPISTATIC AND ALLELIC INTERACTIONS OF HYBRID STERILITY GENES

Hybrid sterility is most often based on incompatible epistatic interactions between independently evolving hybrid sterility genes. These incompatibilities can be modified by allelic interactions of autosomal hybrid sterility genes, which can vary from recessive to underdominant or dominant. Only dominant or underdominant genes can act in F1 hybrid sterility, but in the N2 generation or in the case of heterospecific introgessions also recessive incompatibility genes can be involved.

The “stand-alone” dominant and recessive PWD hybrid sterility genes were not detected in any of 23 consecutive introgressions of individual PWD autosomes (or their parts) into B6 genome (see Fig. 2 in Gregorova et al. 2008), even though the overall fecundity of the homosomic mice was significantly compromised for the majority of autosomes. The finding is in agreement with the proposed multiple epistatic interactions in the F1 hybrid sterility, the conclusion based on the backcross analysis. In contrast to autosomal introgressions, the XPWD chromosome on B6 genetic background results in male-limited sterility associated with incomplete postmeiotic arrest and production of abnormal sperm unable to fertilize eggs. The Hstx1 hybrid sterility locus responsible for the X-linked male sterility was localized to the middle region of the XPWD. Its action depends on epistatic cis-interaction with at least one proximal and one distal region on Chr X (Storchova et al. 2004). It is possible to speculate that the Hstx1 locus, which affects spermiogenesis when standing alone, functions as a component of the multigenic F1 hybrid sterility causing the breakdown at primary spermatocyte stage.

The systematic analysis of the effect of PWD/PWD homosomy of individual autosomes on F1 hybrid genetic background revealed underdominance of the Chr 17 hybrid sterility locus. The presence of two copies of Chr 17PWD in (PWD × B6.PWD-Chr 17)F1 males resulted in full fertility rescue. The action of hybrid sterility gene at Chr 17, most probably Hst1/Prdm9, was clearly underdominant because both its homozygous forms, PWD/PWD and B6/B6, rescued hybrid sterility when situated on F1 hybrid background or in N2 hybrids. All other homosomic PWD/PWD autosomes, when present individually on F1 hybrid background, did not interfere with F1 male sterility with the exception of Chr 19. Pachytene arrest was released in (PWD × B6.PWD-Chr 19)F1 males, permitting production of a limited number of nonfunctional and deformed sperm cells (Fig. 7). The effect of Chr 19PWD homosomy can be interpreted as the action of a dominant B6 hybrid sterility locus, which could not be mapped in (PWD × B6) × B6 backcross. Similarly, the QTL on Chr 14 can be viewed as PWD dominant because (PWD × B6.PWD-Chr 14) males remained sterile (Fig. 7). The allelic interactions deduced from chromosome substitution strains and their hybrids are summarized in Figure 8.

Figure 7.

Fertility parameters of F1 hybrids between PWD and chromosome substitution strains. Bar plots represent average testes weight (mg) and sperm count across individual crosses. Chr # is an abbreviated designation for a B6.PWD-Chr# chromosome substitution strain.

Figure 8.

Impact of chromosomal constitution on male fertility in chromosome substitution strains and the hybrids. P and B stands for PWD (Mus m. musculus) and B6 (M. m. domesticus) chromosome, mat and pat for maternal and paternal origin of the chromosome. (A) Standard F1 hybrid genotype with maternal set of P chromosomes and paternal B chromosomes. Chromosomes known to be engaged in hybrid sterility are in bold letters, P but not B copy of the Chr X is the essential component of hybrid sterility incompatibility network. (B) Chr 17 is involved in underdominant interaction because BB (see Figs. 3 and 5) and PP homozygous forms support fertility, only PB heterozygotes are found among sterile males). (C) Chr 19 behaves as B dominant, because only PP homozygosity rescues intrameiotic arrest. (D) Individual chromosomes were tested for their capacity to rescue F1 sterility when PP homozygous. Autosomes (Chr #) other than Chr 17 and Chr 19 do not rescue F1 meiotic arrest. Only genotype of PWD × B6.PWD-Chr 1 hybrid is depicted. (E) Introgression of individual P autosomes into B background did not arrest spermatogenesis. Introgression of Chr 10 and Chr 11 resulted in very low fecundity of breeding pairs (compare White et al. 2011), not associated with spermatogenic breakdown. Fertility was improved by introgression of each chromosome as three overlapping segments (see Gregorova et al. 2008). Introgression of PWD Chr X results in spermiogenesis abnormalities and male limited sterility (see Storchova et al. 2004). (F) Reconstruction of incompatibility genotype at Chr17 and Chr X does not recapitulate hybrid sterility phenotype. The requirement of additional interacting loci indicates complex epistasis behind our model of reproductive isolation incompatibility. (G) Reconstruction of PB heterozygosity of Chr 2, Chr 14, Chr 17 in the presence of P Chr X still does not result in meiotic arrest.

Discussion

The basic principles of genetic regulation of hybrid sterility were derived from Drosophila species and their universality was confirmed in a wide range of other species including mammals (Dobzhansky 1951; Coyne and Orr 2004; Presgraves 2010; Maheshwari and Barbash 2011). They include the Haldane's rule stating that heterogametic sex (XY or ZW) is more often affected by hybrid sterility or inviability (Haldane 1922) or the large X-effect, which posits that genes with large effect on postzygotic isolation reside often on the X chromosome (Coyne and Orr 2004). A caveat already articulated by Lewontin (Lewontin 1974) refers to the fact that reproductive isolation genes can represent either the cause or the consequence of speciation. Gene variants leading to D-M incompatibilities can arise even faster once the barrier between the evolving species has been established, the phenomenon known as snowball effect (Orr and Turelli 2001; Moyle and Nakazato 2008; Matute et al. 2010). These a posteriori hybrid sterility genes, often found in Drosophila hybrids, are obviously less important in the process of speciation than those acting in emerging species.

Mus m. musculus and M. m. domesticus are young house mouse subspecies particularly suitable for hybrid sterility studies because of their recent origin and availability of extensive genomic resources (Forejt 1996; Geraldes et al. 2011; Keane et al. 2011). To study the genetic architecture of hybrid sterility, we used inbred strains PWD and B6 representing M. m. musculus and M. m. domesticus subspecies (see section Materials and Methods) and chromosome substitution (consomic) strains carrying individual M. m. musculus chromosomes introgressed in the genome of M. m. domesticus (Gregorova and Forejt 2000; Gregorova et al. 2008). By combining the QTL analysis of segregating N2 generation with the results from F1 hybrids between consomic strains and PWD mice we estimated the nature of asymmetry of F1 hybrids, the number of hybrid sterility genes, their location in the genome, and their mode of interaction. The role of mitochondrial genome, autosomal imprinted genes, and Chr Y was also evaluated.

ASYMMETRY OF F1 HYBRIDS

Asymmetry of hybrid sterility can result from D-M incompatibilities involving uniparental inheritance, such as sex chromosomes, imprinted genes, or mitochondrion (Coyne and Orr 2004; Turelli and Moyle 2007). The analysis of male hybrids between PWD and consomics provided direct evidence that asymmetry in male infertility is controlled from the central region of Chr XPWD, excluding the role of the Chr Y, mitochondrial genome, or imprinted autosomal genes. A similar conclusion was reached previously to explain asymmetric infertility in PWK × LEWES (M. m. musculus×M. m. domesticus) male hybrids (Good et al. 2008a,b). However, the study could not exclude the role of imprinted genes or mitochondrial incompatibility because the intrameiotic arrest of F1 hybrids and postmeiotic breakdown in Chr X congenics could originate from different sets of incompatibilities.

NUMBER OF HYBRID STERILITY LOCI AND THEIR INTERACTION

Twelve percent of sterile/subfertile males in the N2 generation could be explained by epistatic interaction of —three to four hybrid sterility genes recreating the F1 male sterility genotype. The genome-wide mapping of hybrid sterility genes revealed two strong QTL loci for TW and for sperm count, one overlapping Hst1/Prdm9 on Chr 17 and the other spanning Hstx1 region on Chr X. Proper genotypes at these two strong loci was an absolute requirement for male sterility because all 31 sterile/subfertile males of N2 generation shared the PWD allelic form of the DXSr62 region on Chr X and PWD/B6 heterozygosity at the Hst1/Prdm9 interval on Chr 17. On the other hand, not all males of this genotype were actually sterile (see Fig. 6), showing that both loci are necessary but not sufficient for sterility. A reconstruction experiment with chromosome substitution strains confirmed this conclusion. The QTL analysis of sterile and fertile males carrying the sterility genotype at Chr 17 and Chr X indicated additional involvement of QTLs on Chr 14 and, in case of TW, also Chr 13. Correlation of the degree of genome-wide heterozygosity (Moehring 2011) with sterility parameters also suggested additional loci of weaker effect as a requirement for manifestation of Chr 17–Chr X epistatic incompatibility. Altogether, the QTL analysis argues in favor of a model of two strong hybrid sterility loci, on Chr 17 and Chr X, interacting with a set of weak and interchangeable loci, a subset of which is necessary to activate the D-M incompatibility between the strong loci. The data can be explained by three weak hybrid sterility loci, from which at least two have to be in PWD/B6 heterozygous state, as a prerequisite to ensure the F1-like male sterility in N2 mice. A simulation study (data not shown) suggests that under such a scenario the sample size should be increased from 70 to at least 150 to have 90% chance to reveal all QTLs. Thus additional studies will be required to validate the proposed model of minor hybrid sterility loci.

Mapping of hybrid sterility genes in backcrosses has some inherent limitations, namely it cannot detect recessive hybrid sterility genes of the parental strain to which the F1 hybrid females have been crossed. Thus the rescue of intrameiotic arrest by PWD homozygosity of Chr 19 can be interpreted by the presence of a B6 recessive hybrid sterility gene or it can be the effect of a recessive B6 rescue gene, which overcomes meiotic checkpoint but has no active role in the F1 hybrid sterility. Backcross to the PWD parent will distinguish between both explanations.

Chr 17 and Chr X showed significant association with TW and fertility in a QTL analysis of the backcross (C57BL/6J ×M. macedonicus) F1× C57BL10 (Elliott et al. 2004). While the QTL interval on Chr X seems to overlap Hstx1, the peak on Chr 17 is distal to Hst1/Prdm9. The phenotype of F1 hybrid sterility was mostly represented by premeiotic block, which differed from the intrameiotic arrest seen in (PWD × B6)F1 males. It is of interest that another mouse species, M. spretus, carries a hybrid sterility gene in the same region of Chr X as observed in M. m. musculus (PWD) and M. macedonicus hybrids (Elliott et al. 2001). The major role of Chr X in reproductive isolation of Japanese wild mice M. m. molossinus was reported by Oka et al. (2004) using MSM/Ms and B6 inbred strains as models for M. m. molossinus and M. m. musculus subspecies. All these examples point to the disproportionate effect of Chr X genes on hybrid sterility known as large X-effect from Drosophila studies (Coyne and Orr 2004).

Recently, Payseur and co-workers (White et al. 2011) analyzed the genetic architecture of hybrid sterility of M. m. musculus×M. m. domesticus hybrids using F2 crosses of PWD and WSB inbred strains. They found several QTLs on autosomes and X chromosome. Using multiple QTL mapping, they detected QTL overlapping Prdm9 for relative testis weight and sperm density. Several QTLs on X chromosome controlled sperm abnormalities but not TW. Comparison of B6 and WSB strains is interesting in this context because both genomes are mostly of M. m. domesticus origin, but they differ in the zinc finger domain of the Prdm9 gene (Parvanov et al. 2010). The WSB allele (Dom3, see Fig. S2 in Parvanov et al. 2010) is more similar to the C3H “fertility” allele than to the B6 (Dom2) “sterility” allele. This intraspecific polymorphism corresponds to the meiotic phenotype of hybrids, because (PWD × WSB)F1 do not display intrameiotic block, produce a limited number of sperm cells, and rarely even offspring (White et al. 2011). Thus, spermatogenic impairment of (PWD × WSB)F1 males is more similar to (PWD × C3H)F1 males than to (PWD × B6)F1 hybrids and indicates that Prdm9 gene could control hybrid sterility not only at primary spermatocytes but also at later stages of spermatogenesis.

Intraspecific polymorphism of compatible (fertility) and incompatible (sterility) alleles of Hst1/Prdm9, was documented in both subspecies (Forejt and Ivanyi 1974; Forejt 1981; Pialek et al. 2008; Vyskocilova et al. 2009) and indicates recent origin of their reproductive isolation.

Search for X-linked genes engaged in reproductive isolation of M. m. musculus and M. m. domesticus has been conduced in wild mice from several transects across the European hybrid zone (Hunt and Selander 1973; Tucker et al. 1992; Dod et al. 1993; Payseur et al. 2004; Macholan et al. 2007). Payseur et al. (2004) studied introgression of 13 X-linked markers and localized a candidate interval spanning 35.7 Mb centered around polymerase (DNA directed), alpha 1 (Pola1). It is tempting to speculate that the restricted gene flow in this region was caused by the hybrid sterility locus mapped in the present study. However, any extrapolations of our data to the situation in the hybrid zone is premature because the overlapping regions are too large and mechanisms of reproductive isolation other than hybrid sterility can be involved.

A mechanistically similar model of F1 hybrid sterility has been described between incipient species of D. pseudoobscura Bogota and D. pseudoobscura USA (Prakash 1972; Orr and Irving 2001; Phadnis and Orr 2009; Phadnis 2011). As in the M. musculus subspecies, both D. pseudoobscura subspecies diverged quite recently and their reproductive isolation is incomplete. Hybrid sterility is asymmetric, as the male progeny of Bogota females and USA males are sterile, while reciprocal hybrids are fertile in both sexes. QTL analysis distinguished seven hybrid sterility loci, three strong on the Chr X and four autosomal with a weaker effect. As in the PWD × B6 mouse model, hybrid sterility loci in D. pseudoobscura form a single D-M incompatibility, where replacement of a “sterility” allele by a compatible allele at a single gene resulted in the restoration of fertility. The absence of any effect of the Chr Y (Phadnis 2011) is an additional feature common to the mouse and Drosophila models.

FUTURE PROSPECTS

Analysis of recently diverged “young” subspecies, M. m. musculus and M. m. domesticus, with currently available genomic tools gives a real chance to identify and characterize all genes involved in this model system of reproductive isolation. We are aware that the postzygotic reproductive isolation between M. m. musculus and M. m. domesticus is far more complex than this model system (Vyskocilova et al. 2009; Oka et al. 2010; Turner et al. 2012) and includes interaction of recessive hybrid sterility genes that can operate in mosaic genomes analogous to experimental N2 or F2 generations (White et al. 2011). However, to understand the molecular mechanisms behind discontinued gene flow, the simplified but genetically well-defined model system of hybrid sterility may be useful. Hybrid sterility 1 gene, originally mapped by its intraspecific polymorphism, was recently identified as Prdm9 (Mihola et al. 2009). The physiological function of this gene is to methylate histone 3 at lysine 4, a modification associated with gene activation but also with meiotic recombination hotspots (Baudat et al. 2010; Parvanov et al. 2010; Ponting 2011). Once all loci necessary for the F1 hybrid sterility are mapped to chromosomes, a reproducible reconstruction of the hybrid sterility genotype will be possible by proper combination of chromosome substitution strains. Using these strains as a “jigsaw puzzle,” identification of partner genes of Prdm9 in D-M incompatibility will become easier and may help to elucidate not only the molecular mechanism of meiotic arrest in sterile hybrids, but also their normal function in parental species. Such analyses will require increased resolution of hybrid sterility phenotypes. Because the critical arrest of (PWD × B6)F1 hybrids occurs during the first meiotic division (Forejt 1985; Mihola et al. 2009), studies of the subcellular and molecular phenotypes, such as chromosome pairing, histone modifications, or meiotic sex chromosome inactivation may be necessary to understand the functions of individual hybrid sterility genes.


Associate Editor: C. A. Buerkle

ACKNOWLEDGMENT

This work was supported by the Academy of Sciences of the Czech Republic (Praemium Academiae to JF) and by the Ministry of Education, Youth and Sports of the Czech Republic COST LD11079 and AVOZ50520514. MD-G and TB are PhD students supported in part by the Faculty of Science, Charles University, Prague. We thank R. Mott and B. Harr for valuable comments, R. Reifova for her help in the initial phase of the work, Z. Trachtulec and members of our laboratory for helpful discussions, and S. Takacova for critically reading the manuscript.

Ancillary