Sharp decline in male fertility in F2 hybrids of the female-heterogametic silk moth Bombyx

Abstract

Sexual selection drives rapid evolution of morphological, physiological, and behavioral traits, especially in males, and it may also drive the rapid evolution of hybrid male sterility. Indeed, the faster male theory of speciation was once viewed as a major cause of Haldane's rule in male-heterogametic XY taxa, but is increasingly being replaced by the genetic conflict hypothesis partly because it cannot explain the faster evolution of hybrid female sterility in female-heterogametic ZW taxa. The theory nonetheless predicts that there should be more genes for hybrid male sterility than for hybrid female sterility even in such taxa, but this remains untested. Thus, finding evidence for the faster male theory of reproductive isolation beyond the F1 generation in ZW systems still represents a challenge to studying the impact of sexual selection. In this study, we examined F2 hybrids between the domesticated silkworm Bombyx mori and the wild silk moth Bombyx mandarina, which have ZW sex determination. We found that although only females showed reduced fertility in the F1 generation, the F2 hybrid males had a significant reduction in fertility compared with the parental and F1 males. Importantly, 27% of the F2 males and 15% of the F2 females were completely sterile, suggesting the presence of recessive incompatibilities causing male sterility in female-heterogametic taxa.

Introduction

The faster male theory (Wu and Davis 1993) postulates that sexual selection acts more strongly on male reproductive characters than on female ones, leading to rapid evolution of male characters including genital morphology, sperm physiology, and seminal fluid proteins. The theory was originally proposed to explain the much higher incidence of interspecific hybrid male sterility compared with both hybrid female sterility and inviability in Drosophila and mammals (Wu and Davis 1993; Wu et al. 1996). Indeed, the number of genes involved in hybrid male sterility between Drosophila simulans and Drosophila mauritiana is estimated to be 10 times that involved in either hybrid female sterility or inviability of both sexes (Wu et al. 1996). The theory is therefore considered to be one of the main causes of Haldane's rule (Turelli 1998; Schilthuizen et al. 2011; Cowell 2023), which states that when one sex of the F1 offspring of 2 different animal races is absent, rare, or sterile, that sex is the heterogametic one (Haldane 1922).

Several lines of evidence support the faster male theory as the cause of Haldane's rule. First, mosquitoes of the genus Aedes have 2 functional homomorphic sex chromosomes and still obey Haldane's rule for sterility without exception, as do those of Anopheles, which have heteromorphic (degenerate Y) sex chromosomes (Presgraves and Orr 1998). In addition, Anopheles also obeys Haldane's rule for inviability, but Aedes does not necessarily do so, suggesting a greater importance of hemizygosity for the evolution of hybrid inviability (Presgraves and Orr 1998). Second, introgression analyses have previously found a great excess of hybrid male sterility over hybrid inviability or hybrid female sterility in Drosophila (Haldane 1922; True et al. 1996; Tao and Hartl 2003; Masly and Presgraves 2007) and in hybrid frogs (Dufresnes et al. 2016). Third, Xenopus laevis × Xenopus muelleri hybrids were found to oppose Haldane's rule, in which heterogametic (ZW) females are fertile but homogametic (ZZ) males are sterile (Malone et al. 2007). Furthermore, sex-reversed ZZ hybrid females were fertile and sex-reversed ZW hybrid males were sterile, suggesting that spermatogenesis is more easily perturbed than oogenesis in hybrids (Malone and Michalak 2008). Fourth, genes expressed predominantly in males show higher divergence in both expression and sequence than unbiased or female-biased genes (Cutter and Ward 2005; Torgerson et al. 2002; Ranz et al. 2003; Zhang et al. 2004; Khaitovich et al. 2005; Baines et al. 2008; Harrison et al. 2015; Sánchez-Ramírez et al. 2021), although the faster evolutionary rates of protein-coding genes are not always observed for male-biased genes (Ellegren and Parsch 2007; Dorus et al. 2010).

The faster male theory, however, cannot explain Haldane's rule for hybrid female sterility in female-heterogametic ZW taxa such as birds and butterflies (Presgraves and Orr 1998), in which females are preferentially sterile (Coyne 1992; Wu and Davis 1993; Turelli and Orr 1995; Laurie 1997). The theory nonetheless predicts that there should be more genes for hybrid male sterility than for hybrid female sterility even in such taxa (Wu et al. 1996); however, this prediction remains untested. Hence, finding evidence for the faster male theory beyond the F1 generation in ZW systems still represents a challenge to studying the impact of sexual selection. One way to address this issue is with a genetic analysis of F2 hybrids or recombinant inbred lines. Recessive hybrid incompatibilities that are masked in F1 hybrids may be exposed in F2; indeed, recessive hybrid incompatibilities outnumber dominant ones in Drosophila (True et al. 1996; Presgraves 2003; Tao and Hartl 2003). However, such analyses have hardly been attempted in taxa with heterogametic females and homogametic males.

In this study, we examined F2 hybrids between the domesticated silkworm Bombyx mori and the wild silk moth Bombyx mandarina, which have ZW sex determination. We found that although only females showed reduced fertility in the F1 generation, the F2 hybrid males had significantly reduced fertility compared with the parental and F1 males. Although a quantitative analysis of the relative densities of hybrid male and female sterility factors remains to be performed, the findings herein may imply that hybrid male sterility evolves as fast as hybrid female sterility in this species pair.

Results

Fertility decline in both female and male hybrids

To determine whether recessive hybrid incompatibilities had accumulated in Bombyx, we examined female and male fertility of parental p50 silkworm and F1 and F2 hybrids between B. mori (p50) and B. mandarina (KY09) (Supplementary Tables 1–4). While fertilization rates (i.e. male fertility) did not differ significantly between F1 and parental p50 males, even though the F1 males were crossed to the F1 females (Fig. 1a), the total number of eggs laid per female (i.e. female fertility) significantly decreased in the F1 females when compared with the parental p50 females (Fig. 1b). This decrease was possibly not due to females but due to males because the fertility status of the F2 males significantly affected the total number of eggs laid by the mated females (Supplementary Fig. 2 and Supplementary Table 3). However, most females mated to normal-fertility F2 males laid 250 or more eggs, as did parental p50 females mated to p50 males, and there was no reduction in F1 male fertility. These findings show that the F1 male effect is seemingly not a major factor. By contrast, the fertility of both sexes significantly decreased in the F2 hybrids compared with parental p50 (Fig. 1); 27% (56/210) of F2 hybrid males and 15% (25/164) of the F2 females were sterile, and this difference was statistically significant [G-test of independence with Williams' correction (G′) = 7.2, df = 1, P < 0.01]. Thus, we conclude that recessive incompatibilities reducing the fertility of hybrids between B. mori and B. mandarina have accumulated in both sexes. Finally, it should be added that the reduced fertility of the F2 males may be caused not by sperm quality or number but by other components in the ejaculate, or by cryptic female choice (Eberhard 1996).

Fig. 1.

Male and female fertility of parental p50 silkworm, F1, and F2 hybrids between B. mori (p50) and B. mandarina (KY09) males. a) The fertilization rate (the fraction of fertilized eggs among eggs laid) as an estimate of male fertility. b) The total number of eggs laid per female as an estimate of female fertility. The medians, the number of individuals studied (n), and the Mann–Whitney U test results are given.

Segregation distortion of molecular markers in the F2 males

We genotyped 31 markers in 210 F2 males as given in Supplementary Table 3, in which P and K stand for p50- and KY09-derived alleles, respectively. One marker was generated for each B. mori chromosome, except for chromosomes 13, 21, and 28, where 2 markers separated by about 10 Mb were developed. The frequencies of recombinants between the 2 markers on these 3 chromosomes were 0.26, 0.39, and 0.44, respectively; the markers were therefore largely independent of each other. We first assessed genotypic frequency data of the 31 markers for Mendelian inheritance and found 6 markers that departed significantly from expected Mendelian ratios (Supplementary Table 3; chromosomes 2, 3, 6, 18, 23, and 25). For all 6 of these markers, the frequencies of the p50 alleles were higher than those of the KY09 alleles. This was particularly true of the mitogen-activated protein kinase kinase kinase 4 marker on chromosome 6, where the numbers of PP, PK, and KK F2 males were 125, 76, and 7, respectively, suggesting a partially recessive factor on the KY09 chromosome that reduced hybrid viability, or the expression of a cryptic meiotic drive element in hybrid conditions that caused the p50-biased segregation ratio.

Hybrid male sterility

We divided the F2 males into 3 groups according to their fertility: a normal-fertility group with a fertilization rate of ≥80%, a low-fertility group with a fertilization rate of <10%, and all others. Genotypic frequencies were subsequently compared between the normal- and the low-fertility groups. Among the 31 markers, the flap endonuclease 1 marker on chromosome 20 was the only one whose genotypic frequencies differed significantly between the 2 groups (G′ = 16.4, df = 2, P < 0.0003), where the p50 allele was less frequent in the low-fertility group. Indeed, the fertilization rate of PP was significantly higher than that of both PK (Mann–Whitney U = 957, P < 0.001) and KK (Mann–Whitney U = 713, P < 0.02; Fig. 2a). Because the F1 hybrid males had the same level of fertility as the parental p50 silkworm, we hypothesize that (still unidentified) factors on the KY09 chromosome 20 interacted with homozygous or hemizygous p50 factors on the other chromosomes to reduce male fertility.

Fig. 2.

Significant effects of a) chromosome 20 on hybrid male fertility and b) Z chromosome on hybrid female fertility. In both a) and b), the medians are indicated by the solid bars. a) Male fertility of 3 chromosome 20 genotypes of F2 hybrids. The medians are 92.9 for PP (n = 34), 41.9 for PK (n = 96), and 76.9 for KK (n = 61). b) Female fertility of 2 Z chromosome genotypes of F2 hybrids and F1 hybrids. The medians are 100.5 for PW F2 (n = 54), 69.5 for KW F2 (n = 104), and 142.5 for KW F1 (n = 16). The W chromosome is that of p50 (Supplementary Fig. 1).

Z chromosome effect on segregation ratio and fertility in the F2 females

To assess Z chromosome effects on female fertility, we typed the roquin-1 isoform X2 marker on the chromosome in 164 F2 females, finding that among them 54 individuals carried the p50 Z chromosome (p50Z), 104 individuals carried the KY09 Z chromosome (KY09Z), and 6 were undetermined individuals (Supplementary Table 4). There was a strong segregation distortion against p50Z (G′ = 16.0, df = 1, P < 0.0001). Because all F2 females carried the p50 W chromosome, this distortion seems to stem from Z–autosome incompatibility. Additionally, no segregation distortion was found for this marker in the F2 males and no sex ratio distortion was observed in F1 (68 females and 66 males), suggesting a recessive action of both Z and autosome.

While female fertility did not differ significantly between F2 carrying p50Z and F1 (Fig. 2b; PW F2 vs KW F1; Mann–Whitney U = 308, P = 0.08), fertility was significantly reduced in the F2 females carrying KY09Z when compared with the F1 females of the same sex chromosome constitution (Fig. 2b; KW F2 vs KW F1; Mann–Whitney U = 314.5, P < 0.0001). These results suggest the involvement of an autosomal recessive gene in the fertility reduction of the F2 females.

Discussion

In this study, we found that despite the lack of reduction in male fertility of F1 hybrids, the F2 males showed significantly reduced fertility, while female fertility progressively decreased from parental silkworm to F1 and then to F2. Importantly, 27% of the F2 males and 15% of the F1 females were completely sterile, suggesting the presence of recessive incompatibilities causing male sterility in these female-heterogametic taxa. Before discussing the implications of this finding, we address the following 3 potentially confounding issues that complicate interpretation of the results: chromosome number variation, effects of domestication, and limitations of F2 mapping analysis.

First, while a chromosome number of B. mandarina in China is 28 (the same as B. mori), that of B. mandarina in Japan (including KY09) is 27 (Goldsmith 2010). A fusion of B. mandarina chromosomes 14 and 27 or 28 is thought to be responsible for this variation (Banno et al. 2004). However, heterozygotes for a simple chromosome fusion form a trivalent in meiosis that often segregates normally and have only minimal meiotic problems (Baker and Bickham 1986). Indeed, 26 bivalents and 1 trivalent were observed in primary spermatocytes of F1 hybrids between B. mori and Japanese B. mandarina (Banno et al. 2004). When F1 hybrid females were backcrossed to B. mori males, the F2 males had chromosome numbers of 2n = 55 and 2n = 56 at a 1:1 ratio and their spermatocytes had 26 bivalents + 1 trivalent and 28 bivalents at a ratio of 1:1, implying normal segregation of the trivalent (Banno et al. 2004). In the present study, the F1 hybrid males had no reduction in fertility when compared with the parental B. mori males; none of the markers on chromosomes 14, 27, or 28 departed significantly from the expected Mendelian ratio in F2 hybrid males. Taken together, we concluded that chromosome number variation did not contribute significantly to reduced hybrid male fertility.

Second, presumably strong artificial selection during 4,000–7,500 years of silkworm domestication (Yang et al. 2014) might have biased interspecific divergence and made it different from a natural evolutionary outcome. Because the commercial value of silkworms is determined exclusively by cocoon traits, selection has likely acted more strongly on metabolic traits than on fertility traits. Indeed, a sequence analysis of 137 representative silkworm strains has identified signatures of selection in genes for nitrogen metabolism, amino acid metabolism, circadian rhythm, and disease avoidance, but none of them are linked to male-specific traits (Xiang et al. 2018). Additional mapping of a local B. mori strain from China that has evolved largely independently from local Japanese strains (including p50) after domestication (Xiang et al. 2018) would be very useful to determine whether the changes along the B. mori lineages occurred before or during the domestication process. Currently, the higher reduction in male fertility of F2 hybrids cannot be ascribed simply to artificial selection during domestication.

Third, with the present mapping in the F2 populations, we could not determine the actual numbers of genes involved in hybrid male and female sterility, hybrid inviability, or segregation distortion. To this end, a genome-wide introgression analysis is required as performed previously in Drosophila (True et al. 1996; Tao and Hartl 2003; Masly and Presgraves 2007). Without such an analysis, the numbers and effect sizes of hybrid sterility factors cannot be compared between males and females. This will be left for future study.

The faster male theory of speciation is once viewed as a major cause of Haldane's rule in male-heterogametic XY taxa, especially in Drosophila and mosquitoes (Wu and Davis 1993; Wu et al. 1996; Presgraves and Orr 1998; Turelli 1998; Schilthuizen et al. 2011; Cowell 2023), but it is being replaced by the meiotic drive, or more generally, genetic (genomic) conflict hypotheses (Frank 1991; Hurst and Pomiankowski 1991; Tao et al. 2001; Phadnis and Orr 2009; Johnson 2010; McDermott and Noor 2010; Meiklejohn and Tao 2010; Zhang et al. 2015; Presgraves and Meiklejohn 2021). Because drive systems favor and preferentially accumulate on the sex chromosomes where recombination is largely abolished, the genetic conflict hypothesis can explain both Haldane's rule and the large X-effect, namely a disproportionately large role of the X chromosome in reducing hybrid fitness (Coyne and Orr 1989). More importantly, the hypothesis simultaneously explains the faster evolution of hybrid female sterility in ZW taxa. By contrast, faster male evolution should oppose Haldane's rule in ZW taxa. A major problem for the faster male theory is the lack of evidence for a causal link between sexual selection and hybrid male sterility. In this regard too, the genetic conflict hypothesis gains more support. too much yin in D. simulans (Tao et al. 2001) and Overdrive in Drosophila pseudoobscura (Phadnis and Orr 2009) are involved in both cryptic sex chromosome segregation distortion and hybrid male sterility. In addition, the protein product of the hybrid male sterility gene Odysseus-site homeobox from D. mauritiana (Bayes and Malik 2009) is species-specifically associated with the heterochromatic Y chromosome of D. simulans and possibly causes its decondensation in their hybrid. These findings provide a strong link between sex chromosome conflict and hybrid male sterility in Drosophila species.

Despite these problems, the faster male theory should not be too hastily abandoned because the relative importance of evolutionary forces may vary across taxa and environmental and ecological contexts. The present study found that although only females showed reduced fertility in the F1 generation, the F2 hybrid males showed a significant reduction in fertility compared with the parental and F1 hybrid males. This finding is important because it may give empirical support to the view that hybrid male sterility may evolve as fast as hybrid female sterility in ZW taxa (Wu et al. 1996). A further important challenge is to continue to explore more ZW genomes for recessive hybrid incompatibilities, especially by using species pairs in the early stage of speciation, and to identify the causal genes. Only such an accumulation of data will enable us to discern the relative importance of sexual selection-driven speciation.

Materials and methods

Bombyx strains

We used the following strains of Bombyx: p50, one of the standard strains of B. mori (provided by Silkworm Stock Center, Kyushu University), and KY09, a strain of B. mandarina, which was established from 2 pairs of wild-caught individuals in Kyoto, Japan, in May 2009 and was maintained for 6 generations en masse until use.

Crosses and fertility assay

To obtain F2 hybrids, we crossed 1 B. mori (p50) female with 1 B. mandarina (KY09) male and then intercrossed the F1 hybrids (Supplementary Fig. 1). A total of 164 F2 females and 12 parental p50 females were then crossed to p50 males, and the numbers of fertilized and unfertilized eggs laid by each female were counted. A total of 226 F2 males were individually crossed to p50 females, of which 210 produced more than 10 progeny and were used for subsequent analyses, and 16 males whose fertilization rates could not be accurately determined were excluded. We also counted those numbers for 16 F1 females that were individually crossed to the F1 males. Egg fertilization status was determined by color (Sakai et al. 2019); fertilized eggs were darkly pigmented after incubating at 24°C for a week, but unfertilized ones remained colorless. We took the total number of eggs laid as the estimate of female fertility and the fraction of fertilized eggs among eggs laid (fertilization rate) as that of male fertility.

Molecular marker development and typing

We developed 31 markers throughout the B. mori genome (1 or 2 markers on each of 28 chromosomes; Supplementary Table 5). To this end, we designed a pair of PCR primers in exons flanking a small intron based on the B. mori sequence in KAIKObase (https://kaikobase.dna.affrc.go.jp; Yang et al. 2021) and identified markers that discriminate the p50 and KY09 genomes by PCR product size differences. All differences were confirmed by Sanger sequencing (ABI3500 Genetic Analyzer, Thermo Fisher Scientific Inc., Waltham, MA, USA). We extracted genomic DNAs of p50, KY09, and F2 hybrid individuals from their leg or abdomen tissues according to a standard rapid DNA isolation protocol with SDS and potassium acetate (Sambrook et al. 1989) or by using a GenElute Mammalian Genomic DNA Miniprep Kit (Sigma-Aldrich Inc., St. Louis, MO, USA). We then PCR-amplified DNA fragments of the 31 markers for typing (protocol details are given in Supplementary Table 5). For simplicity, homozygotes for a p50-derived allele are abbreviated as PP, those for a KY09-derived allele as KK, and the heterozygotes as PK.

Statistical analysis

The fertility data shown in Figs. 1 and 2 were analyzed using the Mann–Whitney U tests in GraphPad Prism 10 (GraphPad Software, Boston, MA, USA) and the G-test of independence with Williams' correction (G′; Sokal and Rohlf 1995; Supplementary Table 3). All P-values were raw values.

Data availability

All data for the figures are included in the Supplementary Tables, which are available at GENETICS online. A p50 strain of B. mori is publicly available at the Silkworm Stock Center (NBRP Silkworms) at Kyushu University; a few adults of the KY09 strain of B. mandarina are stored frozen at −80°C, and the genomic DNA is available upon request (to Y.K.).

Supplemental material available at GENETICS online.

Funding

This work was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI grant numbers JP23570123, JP16K07463, JP19K06780, and JP24K09544 (to T.T.-S.-K.).

Author contributions

T.T.-S.-K. conceived the project and wrote the manuscript with input from M.I. M.I. and T.T.-S.-K. designed the experiments. K.M., Y.K., A.Y., H.O., S.N., and M.I. performed the experiments. T.T.-S.-K., K.M., and M.I. analyzed the data.