Prdm9 Incompatibility Controls Oligospermia and Delayed Fertility but No Selfish Transmission in Mouse Intersubspecific Hybrids

PR-domain 9 (Prdm9) is the first hybrid sterility gene identified in mammals. The incompatibility between Prdm9 from Mus musculus domesticus (Mmd; the B6 strain) and the Hstx2 region of chromosome (Chr) X from M. m. musculus (Mmm; the PWD strain) participates in the complete meiotic arrest of mouse intersubspecific (PWD×B6)F1 hybrid males. Other studies suggest that also semisterile intersubspecific hybrids are relevant for mouse speciation, but the genes responsible remain unknown. To investigate the causes of this semisterility, we analyzed the role of Prdm9 and Chr X in hybrids resulting from the crosses of PWK, another Mmm-derived inbred strain. We demonstrate that Prdm9 and Chr X control the partial meiotic arrest and reduced sperm count in (PWK×B6)F1 males. Asynapsis of heterosubspecific chromosomes and semisterility were partially suppressed by removal of the B6 allele of Prdm9. Polymorphisms between PWK and PWD on Chr X but not in the Prdm9 region were responsible for the modification of the outcome of Prdm9 - Chr X F1 hybrid incompatibility. Furthermore, (PWK×B6)F1 hybrid males displayed delayed fertility dependent on the Prdm9 incompatibility. While the Drosophila hybrid sterility gene Overdrive causes both delayed fertility and increased transmission of its own chromosome to the offspring, the segregation of Chr X and the Prdm9 region from the mouse (PWK×B6)F1 males was normal. Our results indicate extended functional consequences of Prdm9 - Chr X intersubspecific incompatibility on the fertility of hybrids and should influence the design of fertility analyses in hybrid zones and of laboratory crosses between Mmm and Mmd strains.


Introduction
When two populations are separated in nature for a sufficient time period, they may fail to produce a fertile offspring when intercrossed again [1]. This condition is called hybrid sterility. Hybrid sterility is one way of reproductive isolation, which deepens the separation of populations by inhibiting the exchange of genetic information. This separation can finally lead to the formation of a new (sub)species. The sterile hybrids are most often of heterogametic sex [1], the male in the case of mammals and flies. Despite great biological interest, only a few animal hybrid sterility genes have been discovered [2,3,4,5], allowing study of the mechanisms of speciation.
Our model of hybrid sterility is based on two closely related subspecies of the house mouse, Mus musculus musculus (Mmm) and M. m. domesticus (Mmd). These subspecies are effectively reproductively isolated in Europe forming a narrow hybrid zone [6]. The Mmm subspecies has been represented in our model by the wild-derived inbred strain PWD, while the classical strain C57BL/ 6J (henceforth B6) has been used as a representative of Mmd.
The Bateson-Dobzhansky-Muller incompatibility (DMI) model suggests that the decreased fitness of inter(sub)specific hybrids is caused by incorrect interaction of diverged alleles that are not adapted to each other due to separate evolution [21,22,23]. The fertility of (PWD6B6)F1 azoospermic males has been rescued by both removal and overexpression of Prdm9 B6 [24], suggesting a dominant-negative DMI. The presence of sperm in the males resulting from the reciprocal cross that uses B6 females, (B66PWD)F1 [4], as well as the absence of sperm in the offspring of B6.PWD-Chr X subconsomic females and PWD males [8] has suggested that one of the partners of Prdm9 B6 in the DMI(s) causing sterility of (PWD6B6)F1 males is located on Chr X. Recently, the locus has been named Hstx2 and mapped to a 4.7 Mb interval of Chr X [10]. The role of Chr X in mouse speciation has been manifested by the limited degree of introgression of subspeciesspecific alleles from Chr X across the hybrid zone [25,26].
Reduced fertility but not always complete sterility has been detected in the offspring of crosses between other Mmm-and Mmd-derived strains [27,28,29,30], as well as in backcrosses [31,32,33,34] and in the hybrid zone [35,36]. Although some of this phenotypic difference may be caused by polymorphisms in Prdm9 or Hstx2, other hybrid sterility genes may be involved.
Therefore, we asked whether the Prdm9 B6 -Hstx2 Mmm DMI plays a role in the phenotypes reducing F1 hybrid fitness other than complete meiotic arrest. To answer this question, we replaced the PWD strain with PWK, derived from Mmm mice caught near the site of origin of PWD, to generate intersubspecific (PWK6B6)F1 males [37]. In contrast to (PWD6B6)F1 males, (PWK6B6)F1 males carry a low amount of sperm. Using a portfolio of genetically modified mice and intersubspecific (sub)consomics, as well as mouse crosses of PWK, PWD, and B6, we addressed the role of Prdm9 and Chr X PWK by measuring multiple quantitative phenotypes associated with various stages of hybrid spermatogenesis. We conclude that the Prdm9 Mmd -Hstx2 Mmm DMI affects meiosis also in (PWK6B6)F1 male hybrids, but it is modulated by polymorphisms between PWD and PWK located on Chr X but not in the Prdm9 region. Because some aged (PWK6B6)F1 males carried functional sperm, we inquired whether mouse hybrid sterility is accompanied by distorted transmission of Chr X or Chr 17.

Ethics Statement
Mice were bred at the Specific Pathogen-Free Facility of the Institute of Molecular Genetics in Prague and at the non-barrier facility of the Institute of Vertebrate Biology in Studenec.

Mice and Genotyping
The PWK/Ph and PWD/Ph strains (Prague Wild K and D, respectively) were developed from two non-overlapping sets of wild Mmm mice caught near Prague [37]. The genome of PWK/PhJ has been sequenced [38]. The STUP strain was derived from Mmm trapped in Studenec, 150 km from Prague [39], but it is now extinct. The C57BL/6J classical strain is mostly of Mmd origin [40].
The Prdm9 tm1Ymat knock-out was made in 129P2/OlaHsd embryonic stem cells by replacement of the first five coding exons with LacZ [17] and carriers backcrossed ten times to B6 background resulting in the B6-Prdm9 KO strain; the differential segment of Chr 17 extends from the position 6.4 to 21.1 Mb (NCBI37 assembly) as confirmed by SNP analysis using the Mega Mouse Universal Genotyping Array (Mega MUGA) chip custom service (Geneseek-Neogen, USA).
The description of crosses and genotypes states the female parent first, e.g., (PWK6B6)F1 indicates the F1 offspring of a PWK female and a B6 male.
PCR primers and conditions used for genotyping have been published [24,41]. Briefly, microsatellite markers for loci located at Chr17 (

Phenotyping and Statistics
Testicular weight of paired testicles (TW) and body weight (BW) were measured in adult males of the indicated age. Sperm was extracted from the entire (SCe) and/or caput (SCc) epididymides. Slides with surface-spread nuclei (chromosomal spreads) were made from adult testicular cells utilizing isotonic [42] or hypotonic [43] treatment. Indirect immunofluorescence was performed with antibodies described in the Results and footnotes of Tables; the antibodies were obtained from the same providers and used in the same way as described earlier [9,24]. Section methods have also been published [10]. The significance of BW, TW, rTW (TW/ BW), and SC data was evaluated using Welsch's t-test and/or Wilcoxon rank sum test, and cellular phenotypes also with x 2 test. ANOVA and Tukey honestly significant difference test as implemented in the JMP program (http://www.jmp.com/) was utilized for the post-hoc age-dependency analysis of fertility parameters of STUP, B6, and their hybrids. The ARRIVE (Animals in Research: Reporting In Vivo Experiments) guidelines checklist and details are included as Text S1.

(PWK6B6)F1 Males are Semisterile
To assess the role of the incompatibility between Prdm9 and ChrX from Mmm in the phenotypes reducing F1 hybrid fitness other than complete meiotic arrest, we focused on the (PWK6B6)F1 intersubspecific hybrids. Unlike (PWD6B6)F1 or (STUS6B6)F1 [37,39], the (PWK6B6)F1 hybrid males carried sperm ( Table 1). The (PWK6B6)F1 males had a lower relative testicular weight (TW) and sperm count in caput epididymis (SCc), and produced less pups than the reciprocal (B66PWK)F1 ( Figure 1, Table 1; p rTW ,0.001, p SCc = 0.036) or other fertile males or parental controls. The (PWK6B6)F1 hybrids yielded only 1.6 offspring per female per month (OFM) on average (Table 1). Moreover, their fertility was dependent on the age at which they could reproduce. Although nine of 13 (PWK6B6)F1 males tested produced pups, five that were mated starting at the age of nine weeks (when intrasubspecific mice are sexually mature) did not sire offspring until the average of 16 weeks of age. Thus the (PWK6B6)F1 males display oligospermia, decreased TW, and delayed fertility.
Partial Age-dependent Meiotic Arrest in (PWK6B6)F1 Hybrid Males The (PWD6B6)F1 males suffer from a complete arrest of meiotic prophase I at the pachytene stage [9,11]. To determine the reason for semisterility of the (PWK6B6)F1 males, indirect immunofluorescence of surface-spread nuclei (chromosomal spreads) from adult (PWK6B6)F1 testes was used to count the relative number of primary spermatocytes (Figure 2A,B, Table 2). This test can reveal the presence of late pachytene meiotic arrest, which causes an increased abundance of leptonemas, zygonemas, and early pachynemas at the expense of late pachynemas and diplonemas. The 9-week-old (PWK6B6)F1 males had the cellular composition intermediate between and significantly different from both the fertile control (B6, p = 0.006) and from the (PWD6B6)F1 male (p,0.001) with completely arrested meiosis. To assess whether the delayed reproduction capability of the (PWK6B6)F1 males correlates with the strength of the meiotic arrest, we analyzed chromosomal spreads of 9-versus 20-week-old males using meiotic markers ( Figure 2A, Table 3). These markers included synaptonemal complex proteins SYCP1 and SYCP3 to stage the cells and phosphorylated histone H2AX to detect the sex body, a round nuclear structure that encompasses the transcription-inactive sex chromosomes [44]. Meiotic arrest in (PWD6B6)F1 hybrids increases the frequency of pachynemas without sex body [4,24] and of pachynemas with an abnormal sex body encompassing unsynapsed autosomes [9]. Meiotic progress determined as the percentage of normal pachynemas (containing normal sex body without autosomal and XY asynapses) improved significantly with the age of (PWK6B6)F1 hybrids (from 42 to 55%, p = 0.037, Table 3), suggesting that the delayed fertility was caused by a factor(s) regulating meiotic arrest.
Fertility Rescue of (PWK6B6)F1 Hybrids by a change of Allele or Dosage of Prdm9 Proximal Chr 17 encompasses the Prdm9 hybrid sterility gene [4]. To directly test the involvement of Prdm9 in the semisterility of (PWK6B6)F1 hybrids, we analyzed the reproductive phenotypes of the F1 male offspring of PWK females and B6 male congenics carrying various Prdm9 alleles and transgenes (Table 1, Figure 1). The fertility of the (PWK6B6)F1 males was rescued by a transgene carrying two copies of the C3H alleles of Prdm9 and flanking genes (BAC5; p rTW ,0.001; p SCc = 0.001), but not by a transgene carrying two copies of the same genes except for truncated Prdm9 (BAC21). An allelic replacement of the Prdm9 B6 allele using the B6-Prdm9 C3H congenic strain also led to improved TW and SCc of (PWK6B6-Prdm9 C3H )F1 animals (p rTW = 0.003; p SCc = 0.010). Removal of the Prdm9 B6 allele utilizing a Prdm9 knock-out (B6-Prdm9 KO ) increased the fertility parameters of the (PWK6B6)F1 hybrids (p rTW ,0.001; p SCc ,0.001). The analyses of chromosomal spreads from adult testes indicated that the partial meiotic arrest was alleviated in the (PWK6B6)F1 males with removed Prdm9 B6 [(PWK6B6)F1-Prdm9 PWK/2 ] compared to their Prdm9 PWK/B6 littermates (Table 3), as the relative number of pachynemas carrying a sex body and all chromosomes synapsed was higher (p,0.001); however, the meiosis in (PWK6B6)F1-Prdm9 PWK/2 was only partially rescued, as it failed to reach the Table 2. Cellular composition of primary spermatocytes from adult testes of F1 males. level seen in the parental control (PWK, p,0.001). The same order applied to the extent of autosomal asynapses ( Figure 2D). In contrast to (PWD6B6)F1 males, pachynemas containing more than four asynapsed autosomes per cell were rare in (PWK6B6)F1, (PWK6B6)F1-Prdm9 PWK/2 , and PWK males. Another way to quantify the meiotic arrest in F1 hybrids is to analyze the ratio of pachytene spermatocytes with completed recombination to all pachynemas ( Figure 2C); this ratio is reduced in some F1 hybrid testes due to meiotic arrest [9,24]. Pachytene nuclei with completed recombination carry over 20 recombination nodules; these nodules can be detected as foci labeled by an antibody against the MutL homolog 1 (MLH1) protein. Meiotic progress determined in this way was near to normal in fertile (PWK6B6.PWD-Chr 17)F1 with Chr 17 PWK/PWD (63% compared to 67% in B6), but remained partially arrested in semisterile (PWK6B6-BAC21)F1 carrying the BAC21 transgene with truncated Prdm9 (25 of 40 versus 12 of 41, p = 0.003). The same conclusion was reached using another criterion of meiotic progress, the relative number of pachynemas carrying a normal sex body and all 19 autosomes synapsed (p,0.001, Table 3).
To explore whether the delayed onset of fertility can be relieved by manipulation of the dosage or allele of Prdm9, we analyzed the adult F1 males resulting from the crosses of PWK and various B6 congenics at two time points (Table 1). Delayed fertility was not detected in any of the four sets of (PWK6B6)F1 males with fertility rescued by removal, addition or replacement of Prdm9 or Chr 17, as the TW and SC of these rescued hybrids were similar in young versus aged adults (Table 1). Moreover, the (PWK6B6)F1-Prdm9 PWK/2 males sired their first litter on average at 10 weeks of age (much earlier than the (PWK6B6)F1 hybrids at 16 weeks; p = 0.017) and their meiotic progress was similar at nine versus 20 weeks of age (p = 0.813, Table 3).
Thus, Prdm9 is one of the causes of partial meiotic arrest and delayed fertility in (PWK6B6)F1 males.

PWK versus PWD Differences on Chr X but not on Chr 17 Modulate Hybrid Sterility
We next investigated the genetic differences underlying the variation of fertility between (PWK6B6)F1 and (PWD6B6)F1 males using males resulting from the crosses ((PWK6PWD)6B6) and ((PWD6PWK)6B6) that were phenotyped at about 11 weeks of age and genotyped for two loci from Chr 17 and four loci from Chr X. Twenty of 75 and six of 13 animals from these crosses were azoospermic ((PWD6B6)F1-like), respectively. Although these crosses generated different proportions of azoospermic males (27% versus 46%), this difference was not statistically significant (p = 0.155) and both populations were combined for analysis. Markers for proximal Chr 17 loci polymorphic between PWK and PWD segregated randomly with the fertility phenotype in the tested offspring of these crosses, suggesting that the difference in sterility of (PWD6B6)F1 versus semisterility of (PWK6B6)F1 cannot be attributed to the Mmm Chr 17 or Prdm9. In other words, the alleles of hybrid sterility genes of PWD and PWK seem to be the same on Chr 17. Indeed, the Prdm9 alleles of the PWD and PWK strains could not be distinguished by sequencing their coding regions [4,19,38].
Only six of 30 tested Chr X microsatellite markers mapping to four distinct loci (positions from 49 to 77 Mb) were polymorphic between PWK and PWD. The mapping of these loci revealed a significant correlation of TW and SCc with Chr X in the males resulting from the ((PWD6PWK)6B6) and ((PWK6PWD)6B6) crosses (p TW ,0.001, Welsch's t-test; p SC ,0.001, Wilcoxon rank sum test; at position 67 Mb). This locus is thus close to or identical with Hstx2 described previously [9,10]. Because two of 42 animals carrying the PWD allele in this region displayed sperm and 16 of 39 males (41%) harboring the PWK allele were azoospermic, there is probably one or more additional loci polymorphic between PWD and PWK affecting fertility of their hybrids.
These data indicate that polymorphisms in the Hstx2 region and additional loci, but not at proximal Chr 17, may explain the variance in hybrid sterility between (PWD6B6)F1 and (PWK6B6)F1.

Multiple Chr X Loci Affecting Semisterility of (PWK6B6)F1 Hybrids
To refine the mapping of the X-controlled semisterility of (PWK6B6)F1 male hybrids, B6.PWD-Chr X# subconsomic strains [41] were used to exchange parts of Chr X in (B6.PWD-Chr X#6PWK)F1 males (Table 4). Only one subconsomic region (X.1), harboring B6 alleles from the position 62.1 Mb to the distal end of Chr X, rescued fertility to a degree similar to (B66PWK)F1 males carrying Chr X B6 (Table 4). In contrast, two (X.1 and X.3) of these three subconsomics crossed with PWD males rescued the resulting F1 males from azoospermia in experiments mapping the locus controlling meiotic arrest in (PWD6B6)F1 [8]  Age, male age (weeks); N, number of counted pachynemas from a total of two or three males; SB, AB, 0B, % pachytene spermatocytes carrying a normal sex body and all chromosomes synapsed (SB), an abnormal sex body (AB, a sex body containing unsynapsed autosomes), and neither abnormal nor sex body (0B; always carried also unsynapsed chromosomes). See Figure 2A for representative phenotypes and Table 1  and the X.3 PWD region causes (PWK6B6)F1-like semisterility when combined with the (PWK6B6)F1 set of autosomes. Thus, at least two loci of Chr X might contribute to the semisterility of (PWK6B6)F1, one of them in the same region as Hstx2, [8,9,10] and the other distal to it. One or both of these loci may differ between PWK and PWD. These data also excluded the interactions of Chr Y, mitochondrial genome, and differentially imprinted autosomal genes as the causes of the asymmetric difference in fertility of the F1 males resulting from the reciprocal crosses of PWK and B6.
Delayed Fertility in PWK Hybrids Carrying the Hstx2 PWD Region of Chr X Hybrids resulting from the cross of a PWD female with a B6 male and of a PWD male with the B6.PWD-Chr X.1s subconsomic female are azoospermic [9,10]. To determine how this differential segment of Chr X PWD carrying the Hstx2 PWD locus can influence hybrid fecundity in (PWK6B6)F1 genetic background regarding delayed fertility, the subconsomic female was crossed to PWK males and the resulting hybrids phenotyped at two time points. At the age of nine weeks, (B6.PWD-Chr X.1s6PWK)F1 male hybrids were azoospermic similarly as (B6.PWD-Chr X.1s6PWD)F1 and (PWD6B6)F1 hybrids (Table 5). However, testicular histology of these 9-week-old (B6.PWD-Chr X.1s6PWK)F1 males revealed immature sperm in some seminiferous tubules ( Figure 3) and 25-week-old hybrids displayed a low number of spermatozoa in epididymides (Table 5), indicating a fertility delay. The later onset of fertility was again caused by meiotic arrest, as confirmed by quantitative analysis of the composition of primary spermatocytes of the 9-versus 25week-old (B6.PWD-Chr X.1s6PWK)F1 males (Table 2; p = 0.045). In contrast, aged (B6.PWD-Chr X.1s6PWD)F1 males remained azoospermic (Table 5) due to arrested meiosis ( Table 2). The delay of sperm production in (B6.PWD-Chr X.1s6PWK)F1 males further points to the role of PWK alleles outside of the Hstx2 region.

Age-dependent Fertility of Adult STUP Hybrids
A possible delay of spermatogenesis has been detected in intersubspecific mouse F1 hybrids involving the wild Mmmderived strain STUP [39]. The STUP strain has been developed from Mmm mice found about 150 kilometers from the site of origin of PWK. To verify the later onset, fertility parameters were measured in both types of STUP hybrids and parental controls at five time points spanning the interval between 60 and 160 days of age ( Figure 4). The 60-days-old (STUP6B6)F1 and (B66STUP)F1 hybrids carried no or a very few spermatozoa. Although SCe increased in hybrids and B6 with age, it reached a plateau by 85 days of age in the B6 males, but by 100 days of age in the hybrids ( Figure 4A and Table 6). In addition, none of the two types of hybrids ever reached the SCe or rTW levels seen in the parental controls ( Figure 4). These results suggest that delayed fertility is a fairly common phenotype of intersubspecific F1 mouse hybrids.

Lack of ''Overdrive'' Effect in (PWK6B6)F1 Hybrids
Young F1 hybrid males resulting from a certain cross of two Drosophila pseudoobscura subspecies are sterile due to incompatibility(ies) of a gene named Overdrive located on D. pseudoobscura Chr X, but some of these males become fertile and produce almost exclusively XX female offspring [5]. To evaluate whether the sterility-ensuring alleles undergo a drive in the aged mouse F1 hybrids, the offspring of (PWK6B6)F1 males were tested for transmission ratio distortions. The (PWK6B6)F1 males produced female and male pups in an approximately equal ratio, as 13 of 25 offspring were males. The segregation of the Prdm9 alleles of proximal Chr 17 from (PWK6B6)F1 males was also similar, because 12 of 24 offspring tested inherited the PWK and 12 the B6 allele. Therefore, there is no evidence that the later onset of fertility in the (PWK6B6)F1 males is accompanied by preferential transmission of chromosomes carrying the major hybrid sterility loci from older males.

Discussion
Although the Prdm9 gene has been shown to participate in the complete sterility of (PWD6B6)F1, (STUS6B6)F1, and (B66STUS)F1, as well as in the semifertility of (B66PWD)F1 and (PWD6B6-Prdm9 C3H )F1 [4,24], its impact on semisterile hybrids carrying Mmm Chr X has remained elusive, despite their great importance for mouse speciation research [29,35,36]. We found a partial age-dependent meiotic arrest in semisterile (PWK6B6)F1 hybrids that was improved by the genetic manipulation of Prdm9, as well as Chr X. Thus the DMI(s) of Prdm9 B6 and a locus (or more likely loci) from Chr X PWK also play a role in the fertility phenotypes of the semisterile (PWK6B6)F1 intersubspecific hybrids.
Our study demonstrated the importance of intersubspecific heterozygosity of proximal Chr 17 for the semisterility of (PWK6B6)F1 males; however, this region was not the cause of the phenotypic difference between (PWK6B6)F1 and (PWD6B6)F1 males. The Hstx2 region had a detectable impact on the reproductive fitness of (PWK6B6)F1 hybrids, and it probably also contains polymorphism(s) between the PWK and PWD Mmm strains modifying the outcome of F1 fertility. Other loci probably affect the difference between fertility of (PWD6B6)F1 and (PWK6B6)F1; one of them could map distal to Hstx2 on Chr X. Mmd displays polymorphisms in Prdm9 affecting hybrid sterility; the B6 allele ensures complete sterility of (PWD6B6)F1 males, whereas the C3H and some other alleles do not [7,11,24]. Thus the DMI of Prdm9 and the Hstx2 region plays the main role in mouse F1 hybrid sterility, whereas their polymorphisms and other loci can mask or magnify the outcome of their interaction, confirming previous findings of incipient speciation between Mmm and Mmd [11,29,45,46]. Our study shows that Mmm is polymorphic in alleles that are incompatible with Prdm9 and/or Hstx2.
Chr X from all Mmm strains thus far investigated in sufficient detail (PWD, PWK, STUP, STUS) contribute to reduced sterility of intersubspecific hybrids. The genetic data from the mouse hybrid zone suggest the presence of multiple speciation gene(s) on Chr X [25,26,47].
The asynapsis of heterosubspecific chromosomes has been suggested to be the primary cause of hybrid sterility in the (PWD6B6)F1 and (PWD6SCHEST)F1 males [9,10]. Asynapses were also detected in the (PWK6B6)F1 hybrids and their number was suppressed by removing Prdm9 B6 ; asynapsis thus participates in multiple mouse hybrid infertility phenotypes. The (PWK6B6)F1 hybrids are not just oligospermic; they also display a narrow window of age in which they can give rise to offspring. This age is delimited by delayed fertility (this report) and earlier cessation of fertility (Aylor DL, Bell TA, Detwiler D, Calaway ME, Crowley J, Pan W, Odet F, McMillan L, O'Brien DA, and Pardo-Manuel de Villena F, personal communication) to only about 3.5 months (approximately 3.5 to 7 months of age). Delayed fertility was also found in males carrying Hstx2 PWD on (PWK6B6)F1 background, suggesting that the difference(s) modifying the (PWD6B6)F1 azoospermia map(s) out of the Hstx2  region. Since a delayed onset of fertility was also demonstrated in (STUP6B6)F1 and (B66STUP)F1 mouse hybrids (Figure 4), it is not an exceptional phenotype. Moreover, the sperm count of (STUP6B6)F1 males decreased after five months of age, supporting the view that the premature cessation of fertility is also a common phenomenon in intersubspecific hybrids.
The F1 male intersubspecific hybrids of the STUP strain did not display a reciprocal fertility bias as it is the case for PWK and PWD. However, STUP is not an exception, as there are other examples of strains lacking fertility differences in reciprocal intersubspecific crosses, including strains derived from the same locality as STUP, named STUS and STUF [39], or the PWB , epididymal sperm count (SCe) versus age (days); (B), relative testis weight (rTW, mg TW per g BW) versus age (days). The letters symbolize values for the time point of the four genotypes named in ABC order from the left (value ''a'' for B6); the same letter used for the value of another genotype (e.g., ''aa'') means that there is no significant difference between the two values (P.0.05, ANOVA) and a different letter used for another genotype (e.g., ''b'' for STUP) indicates a significant difference (P,0.05, ANOVA) between the values at these time points; log(SCe +1) was used for statistical analysis instead of SCe. SCe was significantly lower in young adult (60 to 85 days of age) versus aged adult F1 hybrids (100 to 130 days of age; see Table 6 for statistics), but even the aged hybrids did not reach the parameters of parental controls. doi:10.1371/journal.pone.0095806.g004 Table 6. Statistical evaluation of age dependency of fertility parameters in F1 males of STUP and B6 and their parents. The values at 60, 85, 100, 130, and 160 days of age were named by letters in ABC order from the lowest value; different letters for two time points in the same cell (e.g., a,b) indicate a significant difference (P,0.05, ANOVA) between these two time points. *Only four time points for STUP (no data at 160 days of age). See Figure 4 for SCe and rTW values. doi:10.1371/journal.pone.0095806.t006 strain derived from a site near the origin of PWK and PWD [37]. The STUS and PWB strains produced fully sterile F1 hybrid males with B6 mice regardless of the direction of the cross, while the male offspring resulting from either reciprocal cross of STUF with B6 were fertile [37,39,46]. Thus, the reciprocal fertility bias probably exists independently of the fertility degree or delay. Our findings should affect the design of crosses involving male hybrids of PWK and other wild-derived strains. The age-limited fertility phenotypes can be fully revealed only in the laboratory, not in males collected from the wild, where the age may be advanced or uncertain. Our data should therefore be taken into consideration when designing the studies and interpreting the results of fertility analyses of natural hybrid zones [35,36]. As the PWK strain is the parental strain of the Collaborative Cross [48], the CC strains could be used to dissect the age-restricted fertility.
The hypothesis associating distorted segregation and unisexual reduction of hybrid fitness explains incompatibilities of two genomes as the outcome of the diversification of sex ratio distorters [49,50]. Indeed, a single gene can participate in both transmission ratio distortion and reduced hybrid fertility. A DMI of the Overdrive gene causes hybrid sterility in (Drosophila pseudoobscura Bogota6Drosophila pseudoobscura USA)F1 males; old hybrids can regain fertility but transmit to offspring preferentially the Overdrive-carrying Chr X [5]. However, no skewed segregation of two major sterility alleles was found in the aged mouse (PWK6B6)F1 males. Admittedly, other chromosomes or hybrids may display distorted transmission, and the clarification of the relationship between mouse hybrid fertility and distorted segregation therefore requires additional work.

Conclusions
Our results have several implications. First, the reduced fertility phenotype caused by the Prdm9-Hstx2 incompatibility is restricted neither to full sterility nor to the genetic background on which it was discovered, but can be modified in both cis and trans. Second, asynapsis of heterosubspecific chromosomes during meiosis is dependent on Prdm9 DMI and plays a role in multiple phenotypes reducing reproductive fitness. Third, male age must be considered to properly describe the phenotypes in all studies of hybrid fertility. Fourth, hybrid sterility alleles of the Prdm9 and Hstx2 genes do not undergo transmission ratio distortion in (PWK6B6)F1 males.

Supporting Information
Text S1 The ARRIVE (Animals in Research: Reporting In Vivo Experiments) guidelines checklist and details. (DOC)