Cell-based assays and comparative genomics revealed the conserved and hidden effects of Wolbachia on insect sex determination

Abstract It is advantageous for maternally transmitted endosymbionts to skew the sex ratio of their hosts toward females. Some endosymbiotic bacteria, such as Wolbachia, cause their insect hosts to exclusively produce female offspring through male killing (MK) or feminization. In some lepidopteran insects, MK is achieved by affecting the sex-determining process in males, and a unique mechanism of MK and its functional link with feminization have been implicated. However, comparative analysis of these phenotypes is often difficult because they have been analyzed in different host–symbiont systems, and transinfection of Wolbachia across different hosts is often challenging. In this study, we demonstrated the effects of nine Wolbachia strains on the splicing of sex-determining genes in Lepidoptera by fixing the host genetic background using a cell culture system. Cell transinfection assays confirmed that three MK-inducing Wolbachia strains and one feminization-inducing Wolbachia strain increased the female-type splicing products of the core sex-determining genes doublesex, masculinizer, and zinc finger protein 2. Regarding Wolbachia strains that do not induce MK/feminization, three had no effect on these sex-determining genes, whereas two strains induced female-type splicing of masculinizer and doublesex but not zinc finger protein 2. Comparative genomics confirmed that homologs of oscar, the Wolbachia gene responsible for MK in Ostrinia, were encoded by four MK/feminizing Wolbachia strains, but not by five non-MK/nonfeminizing strains. These results support the conserved effects underlying MK and feminization induced by oscar-bearing Wolbachia and suggested other potential mechanisms that Wolbachia might employ to manipulate host sex.

It has been hypothesized that the mechanisms underlying MK and FM are associated with sex-determination cascades in insects (18).Indeed, some MK Wolbachia strains directly interact with sex-determination systems in lepidopteran insects (19)(20)(21)(22).The lepidopteran sex-determination system consists of multiple transcriptional regulators, some of which exhibit sex-linked expression and/or splicing isoforms.Lepidopterans generally have a female heterogametic sex chromosome system (e.g.WZ/ZZ) and employ the dosage compensation system, which equalizes the sex-linked (Z-chromosome-linked) gene dose between males and females (20,21).The dosage compensation system is regulated by a Z-linked CCCH-type zinc finger motif encoding gene, masculinizer (masc), which is critical for development as a male.The masc gene also regulates the downstream master regulator of sex determination and differentiation, doublesex (dsx), which exhibits sexdependent splicing isoforms (dsxF in females and dsxM in males) (23,24).Ostrinia and Homona male moths use these factors in their sex-determination cascades, and MK Wolbachia strains disrupt the expression of the masc gene and induce the "female" isoform of dsx (dsxF), leading to a "mismatch" between genetic sex (male: ZZ sex chromosome constitution) and phenotypic sex (female: based on dsxF) and male death (19)(20)(21)(22).In addition to dsx, masc also exhibits splicing variation in the untranslated region between males and females of Ostrinia and Homona moths, and MK Wolbachia suppresses the expression levels of the male-type masc isoforms in male embryos (22,25).Although the extent to which it is involved in sex-determination remains unclear, Ostrinia moths exhibit male-and female-type splicing variants of the autosomal CCCH-type zinc finger motif encoding gene zinc finger protein 2 (znf2) (25), which is an upstream regulator of dsx in the silkworm Bombyx mori (26).In male embryos, MK Wolbachia (wSca) induces a female-specific transcript variant and reduces the male-type isoforms of znf2.Similarly, the FM Wolbachia strain wFem induces dsxF production in Eurema butterflies (27).
More recently, a Wolbachia protein Oscar (abbreviation of "Osugoroshi [i.e.MK, in Japanese] protein containing CifB C-terminus-like domain and many Ankyrin Repeats") was shown to recapitulate the Wolbachia-induced MK phenotype in Ostrinia (23).The Oscar protein degrades and interacts Masc, which leads to the failure of dosage compensation and the production of female-type dsx isoforms in Ostrinia male moths (23,24).Furthermore, an Oscar homolog was found in the prophage region of an MK Wolbachia strain in Homona moths (28).These findings imply a general role for Oscar in MK in lepidopteran insects, which have not been demonstrated.Based on the Wolbachia strains sequenced to date, Oscar is not widely conserved among MK Wolbachia strains (28).In addition, transgenically overexpressed Oscar does not function in Drosophila flies, which do not employ masc in sex determination.Instead, another Wolbachia factor, Wmk, induces male lethality (28,29).In contrast to Oscar, Wmk homologs are widely conserved among Wolbachia strains regardless of their phenotypes (28,29).These findings imply that Wolbachia uses multiple mechanisms to induce MK in insects, but their similarity/diversity and functional link with FM remain elusive.
The induction of Wolbachia-induced reproductive manipulation is deeply associated with the host genetic background (30)(31)(32)(33).MK and FM can significantly distort the sex ratio of the host population toward females (13,31,(34)(35)(36).Under extremely female-biased sex ratios, male production is strongly selected, and host suppressor mutations have evolved to mask the endosymbiont-induced MK in some insects, such as butterflies, ladybugs, lacewings, and planthoppers (36)(37)(38)(39)(40). Therefore, the absence of obvious Wolbachia-induced phenotypes in the natural host does not necessarily indicate the absence of Wolbachia MK capabilities.Thus, investigations of the inherent MK/FM abilities of Wolbachia strains are warranted to elucidate the genetic basis of these phenotypes.
In this study, we reveal a conserved and hidden feminizing effect induced by Wolbachia strains in Lepidoptera.Transinfection assays using a recently established cell culture system (25) revealed that MK/FM Wolbachia strains altered the splicing patterns of genes involved in the moth sex-determination cascade.Genomic analyses confirmed the presence of Oscar in MK/FM Wolbachia strains.Furthermore, an inherent feminizing ability was observed in Oscar-deficient Wolbachia strains that do not induce MK or FM in their native hosts.Based on these findings, we discuss the underlying mechanisms of MK and FM in Lepidoptera and highlight the diversity of causative factors.

Oscar is conserved among MK and FM Wolbachia strains in Lepidoptera
A series of genome assembly and polishing processes were used to construct closed genomes of the Wolbachia strains wSca, wCauA, wCauB, wKue, wCI, wNi1, and wHm-c, which were used for the transactivation assays (Table 2).Comparative genomics using seven newly sequenced strains and the wHm-t strain (28) identified 788 protein clusters shared by the Wolbachia strains, including Cifs (CifB and CifA) and Wmk homologs (Fig. 2A and Table S1).Oscar homologs were the only protein clusters shared by MK Wolbachia (wSca, wCauA, and wHm-t) but are absent in non-MK Wolbachia strains (wCauB, wKue, wCI, wNi1, and wHm-c).As observed in the Oscar proteins of wFur and wSca (23), the Oscar homologs identified in this study encoded ankyrin repeats and the CifB C-terminus-like domains (papain-like cysteine proteinase domains; Fig. 2A and B).An Oscar homolog was also identified in the sequencing data derived from cells coinfected with wFem and wCI.Although we could not reconstruct the genome of the wFem strain due to technical difficulties in isolating wFem from the coinfecting wCI, the Oscar homolog is presumed to be encoded by wFem because the wCI strains used in this study did not carry Oscar homologs in their genomes (Fig. 2C and Table S1).Further diagnostic PCR confirmed that the Oscar homolog was consistently present in wild-caught Eurema mandarina butterflies infected with wFem and wCI (n = 37) but not in those infected with wCI alone (n = 77), suggesting that the Oscar homolog is encoded by wFem.Moreover, wFem-infected females (n = 3) of Eurema hecabe, a close relative of Eu. mandarina, were Oscar positive.
To assess the phylogenetic associations of the Oscar homologs, we further searched Oscar homologs using the GenBank database, which identified Oscar-like proteins in four Wolbachia strains (wGro in Epagoge grotiana [ 2C and S1).These Oscar homologs were highly diverse in sequences, and notably, Oscar in the newly sequenced wSca strain was not identical to that in the previously sequenced wSca strain (similarity: 90.4% in 1,797 amino acids; bit score: 3,241, BLASTp) and wFur (similarity: 94.2% in 1,830 amino acids; bit score: 3,432) (45) because of deletions in ankyrin repeats (Table S1).Intriguingly, the type II Oscar homologs Hm-Oscar and Em-Oscar exhibited higher homology than the wSca strains (similarity: 98.5% in 1,181 amino acids; bit score: 2,362).The MLST and genome-based phylogeny of the Wolbachia strains used in our analyses did not match the phylogeny of Oscar, suggesting horizontal transfer and/or rapid divergence of Oscar (Figs. 3 and S1).

Oscar is absent in wHm-c and wNi that interact with insect sex-determination systems
The wNi1 and wHm-c strains, which induced female-type splicing of masc (OsmascF) and dsx (OsdsxF) in Ostrinia male cells, did not carry Oscar homologs (Fig. 3 and Table S1).In addition, comparative genomics did not identify any proteins that were exclusively present in Wolbachia strains that affected the host sexdetermination system (wHm-t, wSca, wCauA, wNi1, and wHm-c) but not in other strains (wKue, wCauB, and wCI; Fig. 2A).A protein cluster specific to wHm-c and wNi1 (i.e.absent in the other Wolbachia strains) consisted of three hypothetical proteins (63 amino acids: wHmcTK_11260, wNi_10160, and wNi_12630) that did not encode obvious domains and exhibited no homology to known MK genes (Tables S1 and S2).Genes carried by wHm-c or wNi1 that were present in some MK strains (wSca, wCauA, and wHm-t) but absent in non-MK strains (wCauB, wKue, and wCI; Table S2) included a wmk homolog (wNi_11830), whereas others displayed extremely low or no homology to known MK genes (i.e.Oscar (23), Spaid (46), and PVMKp1 (47)).

Discussion
Our cell transinfection assays proved effective in assessing the feminizing ability of Wolbachia strains.Comparative genomics further revealed the presence of Oscar homologs in these strains, which induced female-specific splicing of three sex-determining genes (dsx, masc, and znf2) in male Ostrinia cells.As previously argued (19,27), the present study highlights a common mechanism underlying Wolbachia-induced MK and FM in lepidopteran insects: Oscar-induced suppression of Masc, leading to female-type sex determination and disruption of the dosage compensation system.Among the Wolbachia strains that feminized sex determination, wSca was the most efficient at inducing the female-type isoforms of the dsx and znf2 genes.This efficiency likely reflects its long-term adaptation to its native host, O. scapulalis, from which the OsM1 cell is derived.Notably, consistent with previous findings (25), the expression ratio of dsx (OsdsxF vs. OsdsxM) was lower than that of masc (OsmascM vs. OsmascF) and znf2 (Osznf2M vs. Osznf2F) in the MK/FM Wolbachia-transinfected cells.These patterns may indicate a rapid effect of Wolbachia on the upstream sex determinants, with a subsequent time lag before influencing the expression levels of downstream factors.
Unexpectedly, our study also revealed that two Oscar-deficient Wolbachia strains (wNi1 and wHm-c) affect the sex-determination system of Ostrinia, albeit likely through a different mechanism from Oscar-bearing Wolbachia strains: Oscar-bearing Wolbachia strains (i.e.wCauA, wSca, wFem, and wHm-t) affected Osdsx, Osmasc, and Osznf2, whereas Oscar-deficient wNi1 and wHm-c affected Osdsx and Osmasc but not Osznf2.These differences probably arose from the different machineries caused by different Wolbachia genes, i.e. oscar and other unknown genes.In B. mori, Znf2 is an upstream factor in the sex-determination cascade that regulates dsx sex-specific splicing (26).However, as the sexdetermining gene cascade of Ostrinia (e.g. the hierarchical relationships among Osmasc, Osznf2, and Osdsx) is not fully understood, it remains unclear why only Osznf2 is not affected by wNi1 or wHm-c.One possibility is that Oscar can suppress the functions of Znf2 (or the cascade containing Znf2) as well as Masc, whereas the factors carried by wHm-c and wNi1 cannot act on Znf2 or its upstream components.Our study illustrated that closely related bacteria in the genus Wolbachia can activate different functions to manipulate sex determination in a single insect species.Further investigation of the causes is warranted.This study provides compelling evidence that Wolbachia strains (wCauA, wNi1, and wHm-c) that do not induce MK or FM in their native hosts (43,44,48) retain the inherent ability to manipulate the sex-determination systems of Ostrinia.Despite its ability to alter the sex-determination cascade in Ostrinia cells, wHm-c does not affect the sex-determination cascade in the natural host Homona magnanima (22).These findings suggest that the abilities of Wolbachia to manipulate insect reproduction have been suppressed by the natural hosts but have been maintained through evolution.Wolbachia-induced reproductive manipulation can be influenced by the genetic background of the host (30)(31)(32)(33).Under the condition of a female-biased sex ratio, mutations in the host genome that rescue males can be favored by selection.Indeed, the spread of suppressors of symbiont-induced MK has been observed in nature (18,31,(37)(38)(39)(40).For example, the spread of an MK suppressor in the butterfly Hypolimnas bolina was almost complete within 5 years (31,36).In Cadra cautella, an MK phenotype, presumably induced by the Oscar-bearing wCauA, was recorded in the 1970s (49).In contrast wCauA did not induce MK in C. cautella collected around 2000 (48,50), although it did induce MK when transferred to the closely related host Ephestia kuehniella (30).These findings suggest that C. cautella may have evolved a suppressor against wCauA-induced MK between the 1970s and 2000.
Since wHm-c relatives are found in many species of tortrix moths (43,44), it is possible that a suppressor against wHm-c relatives evolved and became fixed in the ancestral species of tortrix moths.This suppressor cannot suppress the feminizing ability of Oscar-bearing wHm-t, and thus, the mechanistic basis of this suppressor would be different from the suppressor of C. cautella against Oscar-bearing wCauA.Consequently, a new suppressor against wHm-t may arise in H. magnanima in the future, leading to an evolutionary arms race between Wolbachia and the host.Recent studies have shown that MK microbes, including Wolbachia, have evolved diverse MK genes/mechanisms (22,28,46,47).In response to these diverse MK mechanisms, host insects may have evolved various suppressors or suppression mechanisms, potentially masking the Wolbachia phenotypes.
In summary, our study demonstrates that the feminizing effect of Wolbachia strains on the sex-determining gene cascade is most likely the mechanistic basis of MK and FM in lepidopteran insects.This study highlights the effectiveness of combining cell culture systems and genomic analyses to uncover the inherent ability of Wolbachia to manipulate sex.These approaches could also prove effective in elucidating the mechanistic interplay between the host and other selfish reproductive manipulators.A molecular understanding of the commonality and diversity of microbial reproductive manipulations will contribute to a better understanding of the evolutionary interactions between selfish elements and their hosts.

Wolbachia strains and insects
Ephestia kuehniella lines transinfected with wKue, wCauA, and wCauB in each were maintained in the laboratory, as described by Sasaki et al. (30).Females infected with wCauA were mated with uninfected males (50).The insects were reared on a diet consisting of wheat bran, dried yeast, and glycerol (20:1:2) at 25 °C under a 16-h/8-h light/dark photoperiod.Homona magnanima infected with wHm-t (Taoyuan, Taiwan ( 42)) or wHm-c (Takao, Japan ( 28)) was maintained using Silkmate 2S (Nosan Co., Yokohama, Japan).The MK host line W T12 was maintained by crossing it with a Wolbachia-free normal sex ratio line, as described by Arai et al. (42).For Eu. mandarina, all-female blood coinfected with wFem and wCI and a normal sex ratio line singly infected with wCI were collected on Tanegashima Island, Japan, and maintained on an artificial diet, as reported by Kageyama et al. (27).Trichoplusia ni harboring wNi1 was collected in Matsudo, Chiba, Japan.The MK wSca strain, maintained in BmM2 cells as described by Herran et al. (25), was used for the following assays.

Reverse-transcription polymerase chain reaction
Total RNA (200-500 ng) extracted from harvested cells using Isogen II (Nippon Gene, Tokyo, Japan) or TRIzol RNA Isolation Reagent (Thermo Fisher Scientific, Waltham, MA, USA) was reverse-transcribed using a PrimeScript II 1st strand cDNA Synthesis Kit (Takara Bio, Shiga, Japan), according to the manufacturer's protocol.qPCR was performed with 5 µL of KOD SYBR (Toyobo), 0.4 µL each of the forward and reverse primers (10 pmol/µL), 2.2 µL of water, and 2.0 µL of the cDNA template.qPCR was performed in a LightCycler 96 System (Roche, Basel, Switzerland) with a temperature profile of 180 s at 95 °C; 40 cycles of 8 s at 98 °C, 10 s at 60 °C, and 10 s at 68 °C; and heating to 90 °C for melting curve analysis.The primers used in this study are   listed in Table S3.The relative expression (Osmasc, Osdsx, and Osznf2 vs. the control gene Osef1a) and the ratio of male-to-female splice variants were estimated.

PCR detection of Em/Eh-oscar gene
Eurema mandarina collected from Tanegashima Island (Kagoshima, Japan) (35) and E. hecabe collected from Kohama Island (Okinawa, Japan) were subjected to DNA extraction using a DNeasy kit (Qiagen, Hilden, Germany), according to the manufacturer's protocol.The DNA concentration was adjusted to 10 ng/µL and subjected to PCR using primer sets amplifying the complete sequences of Em/Eh-oscar and Hm-oscar (Table S3).Em/ Eh-oscar was amplified using Emerald Amp Max Master Mix (TaKaRa) at 94 °C for 3 min; 35 cycles of 94 °C for 30 s, 62 °C for 30 s, and 72 °C for 3 min; and a final extension at 72 °C for 7 min, as described by Arai et al. (28).

Statistical analysis
For RT-qPCR assays, we used the average cycle threshold value (Ct) for each sample and estimated the relative expression, as described by Herran et al. (25) and Sugimoto et al. (20).Elongation factor a (Ef1a) was used as the control gene.The 2 −ΔCt (i.e.Ct Ave target gene −Ct Ave ef1a) and 2 −ΔCtMal 2 −ΔCtFem (i.e.relative expression in males vs. females) values were calculated.The expression ratios of the splice variants (i.e.DsxM/DsxF, MascM/MascF, and Znf2M/Znf2F) were analyzed using the Steel-Dwass test in R software v4.0 (62).

Fig. 2 .
Fig. 2. Phylogenies and structures of the Oscar and Cif proteins.A) Protein clusters conserved among the Wolbachia strains used in this study.The number of clusters is shown on the left, and the number of proteins contained in the clusters is indicated by black ellipses.B) Structure of Oscar and its homologs.aa, amino acids.The structure of Oscar (1,830 aa), derived from the MK wFur strain, is based on Katsuma et al. (23).C) Phylogeny of Oscar and CifB homologs.Accession and gene numbers are given in parentheses.

Fig. 3 .
Fig.3.Phylogenies and phenotypes of Wolbachia strains and their effects on the sex-determination system in Ostrinia male cells.wsp and MLST genes were used to construct the phylogenetic tree.The classification of Oscar proteins was based on the phylogeny presented in Fig.2C.The effects on splicing of each sex-determining gene analyzed in this study are presented as either Yes (feminized) or No (not affected).MK, male killing; NMK, nonmale killing; Fem, feminization; CI, cytoplasmic incompatibility; NA, not assessed.

Table 1 .
Characteristics of the Wolbachia strains used in this study.

Table 2 .
Genomic features of the Wolbachia strains used in this study.