Diverse Molecular Mechanisms Underlying Microbe-Inducing Male Killing in the Moth Homona magnanima

ABSTRACT Male killing (MK) is a type of reproductive manipulation induced by microbes, where sons of infected mothers are killed during development. MK is a strategy that enhances the fitness of the microbes, and the underlying mechanisms and the process of their evolution have attracted substantial attention. Homona magnanima, a moth, harbors two embryonic MK bacteria, namely, Wolbachia (Alphaproteobacteria) and Spiroplasma (Mollicutes), and a larval MK virus, Osugoroshi virus (OGV; Partitiviridae). However, whether the three distantly related male killers employ similar or different mechanisms to accomplish MK remains unknown. Here, we clarified the differential effects of the three male killers on the sex-determination cascades and development of H. magnanima males. Reverse transcription-PCR demonstrated that Wolbachia and Spiroplasma, but not OGVs, disrupted the sex-determination cascade of males by inducing female-type splice variants of doublesex (dsx), a downstream regulator of the sex-determining gene cascade. We also found that MK microbes altered host transcriptomes in different manners; Wolbachia impaired the host dosage compensation system, whereas Spiroplasma and OGVs did not. Moreover, Wolbachia and Spiroplasma, but not OGVs, triggered abnormal apoptosis in male embryos. These findings suggest that distantly related microbes employ distinct machineries to kill males of the identical host species, which would be the outcome of the convergent evolution. IMPORTANCE Many microbes induce male killing (MK) in various insect species. However, it is not well understood whether microbes adopt similar or different MK mechanisms. This gap in our knowledge is partly because different insect models have been examined for each MK microbe. Here, we compared three taxonomically distinct male killers (i.e., Wolbachia, Spiroplasma, and a partiti-like virus) that infect the same host. We provided evidence that microbes can cause MK through distinct mechanisms that differ in the expression of genes involved in sex determination, dosage compensation, and apoptosis. These results imply independent evolutionary scenarios for the acquisition of their MK ability.

similar DNA damage in Drosophila males, they use different genes, with S. poulsonii using the MK toxin called Spaid and Wolbachia using an MK candidate gene called wmk (18,22,23). However, it remains unclear to what extent male killers share their MK mechanisms.
Here, we report that distantly related three male killers exert MK via different mechanisms in the tea tortrix moth Homona magnanima (Tortricidae, Lepidoptera). Previously, we demonstrated that Wolbachia strain wHm-t (23,32) and Spiroplasma ixodetis sHm (33,34) cause early MK (embryonic MK); however, the partiti-like Osugoroshi virus (OGV) causes late MK (larval MK) in H. magnanima (11,12,35,36). By comparing the effects of microbes at critical time points where MK occurs (Fig. 1b), we confirmed that the previously mentioned MK microbes affect the dosage compensation system, sex-determination cascades, and host development via different mechanisms. In addition, we discuss the origin and evolution of MK mechanisms induced by various microbes.

RESULTS AND DISCUSSION
MK Wolbachia and Spiroplasma, but not OGVs, altered the splicing patterns of dsx in males. To evaluate the effects of the three male killers on sex determination in H. magnanima, we first determined the dsx transcript sequences. Insects frequently exhibit sex-specific dsx splicing variants, referred to as dsx-M (male type) and dsx-F (female type) (15,(37)(38)(39). According to rapid-amplification of cDNA ends (RACE) assays using an H. magnanima normal-sex ratio line (NSR), males showed a variant dsx-M encoding the protein DSX-M, whereas females had eight splicing variants (dsx-F1 to 8) encoding three proteins (DSX-F1 to 3) (Fig. 2a). We then designed a primer pair to distinguish the H. magnanima dsx splicing variants dsx-M and dsx-F (type 3 to 8, Fig. 2a) using a diagnostic PCR (forward primer in coding sequence and reverse primer in noncoding region Exon E). Male embryos (at 108 h postoviposition [hpo] when early MK occurs [32,34] [ Fig. 1]) infected with either early MK wHm-t (W T12 , W T24 , and W TN10 ) or Spiroplasma sHm (S1) exhibited dsx-F, whereas male embryos of the NSR line, L line harboring OGVs, and the W c line harboring non-MK Wolbachia wHm-c (40) showed only dsx-M ( Fig. 2b and c). In addition, S1 and W T24 males exhibited both dsx-M and dsx-F ( Fig. 2b and c). Bacterial density is an important factor determining the expression of MK induced by Wolbachia and Spiroplasma (32,34,41). Previously, we have shown that W T24 lines exhibit higher hatchability and lower wHm-t amounts than the W T12 lines (32). The appearance of both dsx-M and dsx-F in embryos suggests that bacterial densities were not enough to completely alter the dsx splicing. Moreover, OGVs did not alter dsx splicing in embryos as well as in moribund male larvae (5th instar) showing symptoms of OGV infection (i.e., carcinoma-like tissue) (see Fig. S1 in the supplemental material).
Previous studies have shown that MK Wolbachia strains wFur and wSca alter dsx splicing patterns in male embryos of Ostrinia moths (14)(15)(16). In the present study, we highlighted that MK Wolbachia wHm-t and Spiroplasma sHm, but not OGVs, induced femaletype dsx splicing in males of H. magnanima. We also confirmed that non-MK strain wHm-c, which is closely related to the MK strain wHm-t (32), did not alter dsx splicing. We recently discovered an MK-associated prophage region that was present in wHm-t but absent in wHm-c by comparing their genomes (23). The prophage region encodes a homolog of the protein Oscar that recapitulates wFur-induced MK and alters dsx splicing in Ostrinia moths (42). The Wolbachia strain wHm-t carries phage genes that may disrupt the sex-determination cascades of H. magnanima males. We speculate that MK Wolbachia generally interferes with the male's sex-determination cascades in Lepidoptera. In contrast, to our knowledge, it has not been reported that MK Spiroplasma interferes with the host's sex-determination cascades, including Drosophila (17)(18)(19)(20). The presence or absence of the altered dsx splicing in H. magnanima and Drosophila could be due to the difference in host genetic backgrounds (i.e., Diptera and Lepidoptera) and/or the difference in bacterial species (i.e., S. poulsonii and S. ixodetis). In addition, no homologous genes were found in the S. ixodetis sHm genome and the MK-associated prophage region of wHm-t (23,34), suggesting that sHm and wHm-t alter dsx splicing in males via different mechanisms.
Early MK Wolbachia impaired dosage compensation in males. The dosage compensation system adjusts expression levels of sex chromosome genes between males and females (24)(25)(26)(27)(28)(29). In the family Tortricidae (Lepidoptera), males have two Z chromosomes (ZZ) and show equivalent expression levels of Z-linked genes as those in females (ZW) (29,(43)(44)(45)(46)(47). In contrast to B. mori and Ostrinia moths, Tortricidae moths, including Homona, generally contain a large Z chromosome consisting of homologs of B. mori chromosome 1 (Z chromosome) and chromosome 15 (autosome) (29,(43)(44)(45)(46)(47). Considering that Wolbachia infection results in male-specific embryonic lethality due to a failure of dosage compensation in Ostrinia moths (14,42), we hypothesized that microbes in H. magnanima accomplish MK by causing a failure of dosage compensation via Z-linked gene overexpression in males. We evaluated the effects of male killers on dosage compensation in H. magnanima embryos (108 hpo) (Fig. 1) by RNAsequencing (RNA-seq). Among the H. magnanima de novo assembled data (293,111 contigs; mean length, 868.3 bp; total length, 254,508,797 bp), 54,071 contigs were annotated to the B. mori genes. The fold change in the expression (transcripts per million [TPM]) of each annotated contig between males and females was calculated and plotted on the corresponding B. mori chromosome. As reported previously in Wolbachia-infected Ostrinia (14), putative Z-linked genes (i.e., H. magnanima genes corresponding to genes on B. mori chromosome 1 and 15) were expressed at higher levels in wHm-t-infected H. magnanima males than those in females. In contrast, expression levels of putative Z-linked genes were equivalent between males and females in the NSR, sHm-infected, and OGVs-infected lines. These results suggest that only wHm-t affects the dosage compensation system of H. magnanima during the embryogenesis stage.
To confirm whether MK microbes affect the male dosage compensation system differently, we further quantified the expression of conserved Z-linked genes triosephosphate isomerase (HmTpi) and kettin (HmKettin), which are homologs of the genes on B. mori chromosome 1, using quantitative PCR (qPCR) assays. The HmTpi gene dose was 2fold higher in males than that in females, regardless of Wolbachia infection, confirming that males had homogametic sex chromosomes (ZZ), whereas females had heterogametic sex chromosomes (ZW) (Fig. 3e). Moreover, the Z-linked genes HmTpi and HmKettin were expressed at 2-fold higher levels in male embryos (108 hpo) harboring wHm-t than those in females (Steel-Dwass test, P , 0.005). In contrast, Spiroplasma and OGV infection did not affect gene expression levels between the sexes (Spiroplasma, P = 0.65 for HmTpi and P = 1.00 for HmKettin; OGV, P = 0.99 for HmTpi and P = 0.99 for HmKettin) ( Fig. 3f and g). These results confirmed that only wHm-t disrupts the male dosage compensation system during the embryogenesis stage. We also assessed whether OGVs affect dosage compensation during the larval stages. Although the samples were limited in number, the OGV-infected moribund male larvae (5th instar) did not show apparent fold changes in the Z-linked genes compared with females ( Fig. S1) (Steel-Dwass test, P = 0.89 for HmTpi and P = 0.29 for HmKettin). However, in contrast to the effect of Wolbachia, the expression of HmTpi, but not HmKettin, was slightly downregulated in males rather than in females. Although this finding may be an artifact of our experiments, we still cannot exclude the possibility that OGVs somehow overactivated dosage compensation in males resulting in the reduced expression of Z-linked Tpi genes; hence, further assessment is warranted.
Only early MK Wolbachia affected expression patterns of the masc gene. The dosage compensation system is tightly associated with the sex-determination cascade in several insects (14,(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42). The masculinizer gene (masc) is considered an important factor regulating the dosage compensation as well as dsx splicing in lepidopteran insects (41). In B. mori males (ZZ), masc activates the male-type BmDSX protein that promotes male development, and the degraded masc expression leads the femaletype BmDSX protein in females (ZW) (41). MK Wolbachia in Ostrinia furnacalis disrupts dosage compensation and dsx splicing by degrading Masc in males (14,42). We therefore tested whether the MK microbes affect the expression levels of the masc gene in H. magnanima with different manners by RNA-seq using sex-determined mature embryos (108 hpo, 2 replicates under each condition). Although expression levels were low in 108-hpo embryos, two masc isoforms (HmMasc v1 and v2) were expressed in H. magnanima (Fig. 4a), which is similar to the results reported by Herran et al. (48), where Ostrinia scapulalis showed sex-specific alternative splicing of the masc gene altered by MK Wolbachia infection. In the Spiroplasma sHm-infected, OGV-infected, and NSR lines, HmMasc v1 was expressed in females but showed little expression in males, while HmMasc v2 was more abundant in males than in females (Fig. 4a). Conversely, wHm-tinfected male embryos tended to express higher levels of HmMasc v1 and lower levels of HmMasc v2 (Fig. 4a). To further clarify the effects of wHm-t on the expression of masc, we compared time-dependent RNA-seq data of both wHm-t-infected and uninfected H. magnanima pooled embryos (consisted of approximately 100 to 150 males and females) at 12, 36, 60, 84, and 108 hpo. The expression levels of HmMasc v1 were highest at the early embryogenesis stage (12 hpo) (Fig. 4b) and decreased as embryogenesis proceeded. At late embryogenesis stages (108 hpo), expression levels of HmMasc v1 were similar to the expression levels shown in Fig. 4a. Notably, wHm-t infection reduced the abundance of HmMasc v1 at 12 hpo H. magnanima embryos and altered the expression dynamics of the HmMasc v2 through embryogenesis stages ( Fig. 4b and c).
In H. magnanima, early MK Wolbachia and Spiroplasma impaired sex-determination cascades by altering dsx splicing, but only Wolbachia affected the dosage compensation and expression levels of two masc variants. Therefore, our findings suggest that MK Wolbachia and Spiroplasma alter dsx splicing in H. magnanima through different machinery. We speculate that MK Wolbachia generally alters dsx splicing by targeting the masc gene in lepidopteran insects, whereas MK Spiroplasma alters dsx splicing by targeting another factor, such as downstream components of the pathway.
MK microbes alter the expression patterns of genes involved in metabolism, endocrinology, detoxification, and stress response in different manners. Through pairwise comparisons among eight groups of pooled 108-hpo embryos (uninfected, wHmt-infected, Spiroplasma-infected, and OGV-infected males or females; two replicates), we RNA-seq data (108 hpo) were used to make the following comparisons: W T12 males versus W T12 females (a), S 1 males versus S 1 females (b), L males versus L females (c), and NSR males versus NSR females (d). The chromosome number for each H. magnanima transcript-derived contig was assigned based on B. mori gene models. The boxes in the box-and-whisker diagrams represent the median and 25 to 75 percentile ranges of the expression ratios. (e to g) Quantification of Z-linked genes in H. magnanima embryos (108 hpo). The Tpi gene dose (e) was quantified to assess whether males have two Z chromosomes. Kettin (f) and Tpi (g) expression levels were normalized to that of the autosomal Ef1a gene. Error bars represent the standard error. The y axis indicates relative abundances adjusted to those of NSR females. Different letters indicate significant differences determined by the Steel-Dwass test (P , 0.05). The numbers inside the bars indicate replicates. Wol, Wolbachia wHm-t, WT12 line; Spi, Spiroplasma sHm, S 1 line; OGV, L line; NSR, NSR line.
found that the three male killers altered gene expression patterns in different manners at the late embryogenesis stage compared with those of the uninfected NSR males ( Fig. 4; see Table S1 and S2 in the supplemental material). For example, wHm-t specifically upregulated more Z-linked genes (accounting for 33 out of 140 differentially expressed genes [DEGs]), such as autophagy-protein 5 and multidrug resistance-associated protein, than Spiroplasma (8 out of 71) and OGV (5 out of 102) (Table S2) in males, which was probably due to the impaired dosage compensation. In addition, Spiroplasma specifically upregulated more numbers of ribosome-associated genes (n = 8) in males, whereas OGVs downregulated 10 ribosome-associated genes, suggesting impacts on the protein synthesis (Table S1 and S2; Fig. 4e). Principal-component analysis (PCA) revealed that the expression patterns of H. magnanima differed largely depending on microbe infections (Fig. 4d). Intriguingly, the expression pattern of Spiroplasma-infected females was similar to that of   Table S2). As mentioned prior, we identified that S. ixodetis sHm and Wolbachia wHm-t both induced female-type dsx splicing but affected the dosage compensation system differently in H. magnanima. The observed gene expression patterns of H. magnanima may reflect similarities and differences in the effects of Wolbachia and Spiroplasma on the host. In addition, males harboring each microbe shared DEGs involved in stress responses (e.g., glutathione S-transferase 1), endocrine systems (e.g., Kruppel homolog 2), morphogenesis (e.g., cuticle proteins and chitinase), metabolism (e.g., aminopeptidase-N) (Table S1), and signal responses (e.g., Ras suppressor protein) ( Fig. 4e; Table  S2). These results indicate common and specific host responses to each MK microbe. MK Wolbachia and Spiroplasma, but not OGVs, caused abnormal DNA damage during embryogenesis. Male embryos infected with wHm-t or Spiroplasma (132 hpo) were fragile and exhibited nuclear condensation ( Fig. 5a and b). We therefore aimed to determine whether male killers underwent abnormal apoptosis during embryogenesis. Nuclear degradation was observed specifically in males (but not females) infected with either wHm-t or Spiroplasma (132 hpo) (Fig. 5c). The NSR and L lines harboring OGVs did not show DNA fragmentation. The activities of caspase-3 (an apoptosis effector) were higher (Steel-Dwass test, P , 0.05) in wHm-t-infected males (73,539.9 6 3,625.8, relative fluorescence units [RFUs], mean 6 SD) and Spiroplasma-infected males (84,434.8 6 3,773.8) than those in noninfected males (48,995 6 3,773.8), wHm-t-infected females (46,697.9 6 4,357.6), and Spiroplasma-infected females (52,128 6 6,536.4) (132 hpo) (Fig. 5d). Moreover, terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) assays confirmed that males infected with wHm-t or Spiroplasma exhibited abnormal nuclear segmentation (108 and 132 hpo) ( Fig. 5e; see Fig. S2 in the supplemental material).
How do microbes kill H. magnanima male insects? It has been hypothesized that microbes accomplish MK by targeting any molecular mechanisms involved in sex determination and differentiation (5,(13)(14)(15)(16)(17)(18)(19)(20). One question is whether the mechanisms are shared or specific to each microbe. In this study, we revealed the effects of three male killers on sex determination, the dosage compensation system, and the development of H. magnanima (Fig. 6). Wolbachia strain wHm-t specifically impaired the male dosage compensation system. Moreover, Wolbachia strain wHm-t and Spiroplasma strain sHm altered dsx splicing and triggered abnormal apoptosis in males. In contrast, late-MK OGVs did not impair host sex-determination cascades during embryogenesis and larval stages.
Improper dosage compensation is considered a direct cause of Wolbachia-induced embryonic MK in Ostrinia moths (14). Our finding suggests that MK Wolbachia generally affects the dosage compensation in lepidopteran insects. The failure of dosage compensation triggers a differential expression of the Z chromosomal genes, which probably affects the viability of Wolbachia-infected H. magnanima males. However, considering that Spiroplasma did not alter the expression of Z-linked genes (and masc gene), a failure of dosage compensation may not be the only cause of MK. In Drosophila, MK S. poulsonii does not impair the dosage compensation itself but induces X chromosome-specific DNA damage by targeting the dosage compensation complex (DCC), which controls the dosage compensation machinery (17)(18)(19)(20). Although it is unclear whether H. magnanima has some factor(s) corresponding to Drosophila DCC, S. ixodetis sHm may induce MK by damaging the dosage-compensated Z chromosome without affecting the dosage compensation itself. In contrast to the early MK Wolbachia and Spiroplasma, OGVs killed larval or pupal males (12,35,36) and did not affect the sex-determination cascades. Charlat et al. (49) suggested that the sex-determination cascade is not the sole target of MK in insects. The present work supports that prediction and indicates that OGVs have a distinct MK mechanism. OGVs may kill males by utilizing the differences in susceptibility between males and females toward viral factors (e.g., genes or viral loads), damaging male-specific organs (e.g., testes), or impairing factors involved in sex dimorphism (downstream sex determination or maintenance systems, such as hormone synthesis). Our study highlights that microbes achieve MK by a variety of mechanisms even in the same insect species. Considering that MK Wolbachia wHm-t, S. ixodetis sHm, and OGVs do not share any genes (11,23,34), our results strongly suggest that microbes have acquired MK abilities independently through different evolutionary processes.
Previous studies reported similarities and differences in the effects induced by MK microbes in different Drosophila species. For instance, S. poulsonii inhibits neurogenesis and induces male-specific abnormal apoptosis in D. melanogaster (17, 18, 50). Wolbachia strain wBif also induces abnormal apoptosis but does not affect neurogenesis in the male

Distinct Machineries of Male-Killing Microbes
Applied and Environmental Microbiology hosts of D. bifasciata (19). Our study also showed that Wolbachia and Spiroplasma induced female-type dsx and abnormal apoptosis in H. magnanima males, while only Wolbachia specifically affected the dosage compensation system. Kageyama and Traut (13) predicted that mismatches between the genetic sex (ZZ, male) and phenotypic sex (i.e., dsx and subsequent gene expression levels) would affect the viability of males. The dsx gene controls subsequent gene expression and sex-dependent characteristics in insects (37)(38)(39). In Homona, defects in sex-determination cascades caused either by Wolbachia or Spiroplasma likely lead to mismatches between the genetic sex and phenotypic sex. Our transcriptome analyses suggested that Wolbachia and Spiroplasma affected the motor functions, endocrine systems, and antioxidative/antiaging activities of males, which may elicit severe adverse effects on early male development. We speculate that Wolbachia-or Spiroplasma-induced stresses result in a similar outcome of abnormal DNA damage and the death of H. magnanima male embryos.

MATERIALS AND METHODS
Insects. In this study, we used four Taiwanese H. magnanima lines (W T12 , W T24 , W TN10 , and NSR) and three Japanese H. magnanima lines (S 1 , L, and W c ). The W T12 , W T24 , and W TN10 lines are all-female matrilines harboring the MK Wolbachia (wHm-t) strain. The NSR line is a normal sex ratio line (female:male = 1:1) and is free of intracellular bacteria and OGVs (32). The S 1 and L lines are all-female lines and harbor MK S. ixodetis (sHm) (12,33,34) and OGVs (11,12), respectively. The W c line is a normal sex ratio line (female:male = 1:1) and harbors a non-MK wHm-c strain (40), which is closely related to the MK strain wHm-t but lacks an MK-associated prophage region (23). The Taiwanese H. magnanima insects were collected from the Tea Research and Extension Station (Taoyuan City, Taiwan) with permission from the Ministry of Agriculture, Forestry, and Fisheries (permission number 27-Yokohama Shokubou 891 and permission number 29-Yokohama Shokubou 1326) in 2015 and 2017. The insects were reared as described previously (51). The nuclear genetic backgrounds of the host lines were homogenized by mating them with the males of the NSR line for at least 10 generations prior to the subsequent experiments. Wolbachia possibly affects the dosage-compensation system and sex-determination cascade separately via different mechanisms or by targeting only the Masc gene and its upstream cascades, as predicted by Fukui et al. (14). In contrast, Spiroplasma utilizes a distinct but unknown mechanism that affects the sex-determination cascade. (b) Summary of the effects of male killers.

Distinct Machineries of Male-Killing Microbes Applied and Environmental Microbiology
Nucleic acid extraction and sex chromatin observations. To extract RNA and DNA from the sexdetermined mature embryos (108 hpo) (Fig. 1), the genetic sexes of mature embryos were determined by observing W chromosomes as described previously (13,51). Briefly, mature embryos (108 hpo) were dissected on glass slides with forceps. Malpighian tubules were fixed with methanol-acetic acid (50%, vol/vol) and stained with lactic acetic orcein for W chromosome observations. The remaining tissues not used for sexing were stored in the Isogen II reagent (Nippon Gene; for RNA extraction) or a cell lysis solution (10 mM Tris-HCl, 100 mM EDTA, and 1% SDS (pH 8.0) for DNA extraction) at 280°C until subsequent extraction. In total, 12 male or female mature embryos (108 hpo) were pooled and homogenized in the cell lysis solution or Isogen II reagent. The DNA extraction procedure with cell lysis solution was performed as described by Arai et al. (40,51). To extract RNA, samples homogenized in 600 mL Isogen II reagent were mixed with 240 mL UltraPure distilled water (Invitrogen) and centrifuged at 12,000 Â g and 4°C for 15 min. Six hundred microliters of each supernatant was mixed with the same volume of isopropanol to precipitate the RNA; then, the resulting solutions were transferred to EconoSpin columns (Epoch Life Science) and centrifuged at 17,900 Â g and 4°C for 2 min. The RNAs captured in the column were washed twice with 80% ethanol and eluted in 20 mL UltraPure distilled water (Invitrogen).
Total RNA and DNA were also extracted from adults, egg masses (12 to 108 hpo), and OGV-infected 5th instar larvae, as described by Arai et al. (32). The extracted DNA and RNA were quantified using a Qubit v4.0 fluorometer (Invitrogen) and NanoPhotometer NP80 instrument (Implen) and stored at 280°C until subsequent analysis.
RACE and detection of the H. magnanima dsx gene. To determine sex-specific splicing variants of the dsx gene, 39 RACE experiments were performed according to the method described by Sugimoto and Ishikawa (15), with several modifications. The RNA samples extracted from sex-determined adults and mature embryos (108 hpo) were reverse transcribed via avian myeloblastosis virus (AMV) reverse transcriptase XL (TaKaRa) using the oligo(dT) adapter primer ( Table 1). The resulting cDNA samples were amplified with Hmdsx_long2F targeting the conserved dsx sequences (coding sequence [CDS]) ( Fig. 2a; Table 1) and adapter primers using KOD-Plus-Ver.2 (Toyobo Co., Ltd.) under the following PCR conditions: 2 min at 94°C, followed by 35 cycles of 10 s at 98°C, 30 s at 68°C, and 30 s at 72°C. Because the 1st PCR did not yield clear band patterns, 1 mL of product of the 1st PCR was subjected to nested PCR with Hmdsx_long4F (designed on the 39 side from Hmdsx_long2F) ( Table 1) and adapter primers using Emerald Amp Max master mix under the following conditions: 2 min at 94°C, followed by 20 cycles of 10 s at 98°C, 30 s at 68°C, and 30 s at 72°C. The band observed in the nested PCR products was purified using the Qiaquick PCR/Gel purification kit (Qiagen). Purified DNA was ligated into the pGEM-T easy vector (Promega, WI) and used to transform Escherichia coli JM109 competent cells. Plasmids extracted from E. coli colonies formed on the Luria broth (LB) agarose plates were sequenced using the 3100 genetic analyzer (Applied Biosystems) according to the method described by Arai et al. (40). Primer sets specific for T7 and SP6 promoters of the pGEM-T easy vector (Table 1) were used for sequencing reactions.
RNA sequencing, de novo assembly, and transcript quantification. We used 1.0 mg of the total RNA extracted from W T12 , S 1 , L, and NSR mature embryos (108 hpo) or W T12 and NSR egg masses (12,36,60,84, and 108 hpo) to prepare mRNA-seq libraries. Two biological replicates were prepared for each treatment. First, mRNA was extracted from total RNA using the NEBNext poly(A) mRNA magnetic isolation module (New England BioLabs). The mRNA libraries were constructed with the NEBNext ultra II RNA library prep kit for Illumina (New England BioLabs) following the manufacturer's protocol to prepare 300-bp RNA fragments for 150-bp paired-end (PE150) analysis. The generated sequence data from each library were trimmed by eliminating (i) adaptor sequences, (ii) reads harboring nondetermined sequences exceeding 10%, and (iii) sequences harboring low-quality nucleotides (Qscore, ,5) spanning .50% of the read length by Novogen (Beijing, China). Furthermore, all reads showing the average quality below 30 were removed with Trimmomatic (52). The trimmed reads were assembled de novo to generate a transcriptome database for H. magnanima using Trinity (53) and NAAC Galaxy with the default parameters. All contigs, showing high homologies to genes of Wolbachia, Spiroplasma, and OGV based on BLASTn analysis, were removed manually. For quantifying transcript abundances from RNA-seq data, the trimmed reads were aligned to the de novo-assembled transcriptome database using Kallisto (54) that generated the normalized read count data (transcripts per million [TPM]) for all H. magnanima contigs with approximately 60 to 70% map ratios.
Analyzing dosage compensation and quantifying genes on the Z chromosome. The effects of male killers on dosage compensation were verified by measuring gene expression differences, as described by Fukui et al. (14) and Gu et al. (29). To assess fold changes in gene expression levels between males and females, the binary logarithms of TPM differences between males and females of each H. magnanima line were calculated. H. magnanima contigs were then annotated using the B. mori gene sets (55) obtained from KAIKObase (https://kaikobase.dna.affrc.go.jp). The binary logarithms of TPM differences between males and females on B. mori chromosomes 1 to 28 were plotted.
Z chromosomal gene expression in H. magnanima was quantified using reverse transcription-qPCR. Homologs of the B. mori Z chromosomal genes kettin and tpi were extracted from H. magnanima de novo-assembled data by performing a BLASTx search. The primer sequences used to quantify these  Table 1. Gene doses were quantified using the DNA extracted from male and female embryos. Next, 100 ng of RNA extracted from 12 male or female mature embryos (108 hpo) was reverse transcribed using PrimeScript II reverse transcriptase (TaKaRa) at 50°C for 30 min, followed by denaturation at 95°C for 5 min. The cDNA was used to quantify relative gene expression levels, with normalization to the control gene elongation factor 1a (ef1a) ( Table 1). The mean cycle threshold (C T ) values of dual samples were calculated for at least eight replicates, and both DC T (C T Ave Z gene -C T Ave ef1a) and DDC T (DC T male -DC T fem) values were calculated. The dosage (number of Z chromosomes) was estimated based on the 2 -DDCT method, as described by Sugimoto et al. (16). Z chromosomal gene expression levels were analyzed using the Steel-Dwass test in JMP software v9 (SAS, Cary, NC). Quantification of the expression levels of HmIMP, HmPSI, and HmMasc. Homologs of imp, psi, and masc genes were extracted from the H. magnanima de novo-assembled transcriptome database using BLASTx (bit-score, .200) with the B. mori protein data sets. The time-dependent gene expression levels of the HmIMP, HmPSI, and HmMasc genes in males and females during embryogenesis (12 to 108 hpo) were plotted using the TPM values calculated using Kallisto (54) as mentioned above.
DEG analysis. To identify DEGs, TPM values of each transcript calculated as mentioned above were analyzed on iDep9 (http://bioinformatics.sdstate.edu/idep/). The DEGs of each comparison between MK microbe-infected and the uninfected (NSR) H. magnanima males were called using DEseq2 (false discovery rate [FDR] cutoff, ,0.1; minimum fold change, .2) and were annotated using blast2go (56) and B. mori protein data sets (55). In addition, we confirmed that the DEGs were not derived from the microbeassociated transcripts by BLAST searches against Wolbachia, Spiroplasma, and OGV genome data. The functions of each gene were annotated based on the B. mori gene ontology (GO) data on iDep9. GO enrichment analysis was conducted using all annotated genes as the background.
Observing apoptosis in mature embryos. DNA segmentation was visualized, as described by Staley et al. (57), using DNA extracted from the sex-determined W T12 , S 1 , L, and NSR nearly hatched mature embryos (132 hpo). Briefly, 100 ng of DNA, 1 mL of 24-bp linker (1 nM), 1 mL of 12-bp linker (1 nM) ( Table 1), and 5 mL of 2Â buffer for T4 DNA ligase (Promega) were mixed and incubated at 55°C for 10 min, cooled down gently to 10°C for 55 min, and then incubated at 10°C for 10 min. The reaction mixture was incubated with 1 mL of T4 DNA ligase (3 U/mL) for 15 min at 25°C. Then, 1.5 mL of ligated reactant (15 ng of DNA) was amplified using Ex Taq (TaKaRa) with the following conditions: 72°C for 5 min, 25 cycles of 94°C for 1 min, and 72°C for 3 min. The genomic DNA and PCR amplicons were electrophoresed on 1.5% TBE agarose gels to visualize the ladders.
To quantify caspase-3 activity, the sex-determined W T12 , S 1 , and NSR nearly hatched mature embryos (132 hpo) were homogenized in 120 mL of 1Â lysis buffer (10 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM EDTA, and 0.01% Triton X-100). After centrifugation at 3,000 Âg for 5 min, 100 mL of each supernatant or 1Â lysis buffer was used as a sample or background control, respectively, for the following assays. Caspase-3 activity was quantified using EnzChek Caspase-3 assay kit number 1 (Invitrogen) following the manufacturer's protocol. The fluorescence intensity (excitation/emission at ;342 and 441 nm, respectively) was quantified using the 1420 ARVO MX-fla multilabel counter (Perkin Elmer). Caspase activities were analyzed using the Steel-Dwass test in JMP software v9 (SAS, Cary, NC).
Benefit sharing. H. magnanima was collected from tea plantations at the Tea Research and Extension Station (Taoyuan City, Taiwan) and imported with permission from the Ministry of Agriculture, Forestry and Fisheries (no. 27-Yokohama Shokubou 891 and no. 297-Yokohama Shokubou 1326). Taiwanese H. magnanima was maintained only at Tokyo University of Agriculture and Technology. A research collaboration was developed with scientists from the countries providing genetic samples, all collaborators are included as coauthors, and the results of the research have been shared with the provider communities and the broader scientific community. More broadly, our group is committed to international scientific partnerships, as well as institutional capacity building.
Data availability. The Masc, Imp, PSI, and dsx sequences of H. magnanima were deposited in GenBank under accession numbers LC701633 to LC701646. High-throughput sequencing data are available under accession numbers DRA013555 and PRJDB13118 (BioProject). Assembled data are accessible under PRJDB13118.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, XLSX file, 0.04 MB. SUPPLEMENTAL FILE 2, PDF file, 0.7 MB.