Infection pattern of male-killing viruses alters phenotypes in the tea tortrix moth Homona magnanima

Abstract

Male-killing is a microbe-induced reproductive manipulation in invertebrates whereby male hosts are eliminated during development. In the tea tortrix moth Homona magnanima, Osugoroshi viruses 1‒3 (OGVs), belonging to Partitiviridae induce male-killing. The infection patterns of OGVs are diverse; however, how the influence of these patterns of host phenotypes remains largely unknown. Using field-collected larvae, we established a OGV1 and OGV3 double-infection line, in addition to a triple-infection line, and examined the dsRNA segments, purified viral proteins, OGV density, and host phenotypes. PCR analysis demonstrated that the triple-infection line lost one dsRNA segment, whereas the double-infection line lost eight segments, including one RNA-dependent RNA polymerase (RdRp) gene. LC-MS analysis revealed three potential structural proteins in the OGVs. Males died at the larval stage in the triple-infection line and at the embryo-larval stage in the double-infection line of OGV1 and OGV3; the RNA load of female parents did not contribute to the developmental stage at which males died. These findings indicate that the pattern of viral infection, rather than viral RNA load transmitted from female parent, controls the stage of development at which male-killing occurs. Furthermore, the duration of the larval stage of the double-infection line was found to be significantly longer than that of the triple-infection line. The shorter duration of the larval stage of the triple-infection line could be advantageous over the double-infection line in maximizing transmission efficiency.

Introduction

Maternally inherited endosymbionts are common among arthropods (Weinert et al. 2015). Of these, bacterial endosymbionts, such as Wolbachia and Spiroplasma, induce reproductive manipulations, cytoplasmic incompatibility, feminization of genetic males, parthenogenesis, and male-killing (Hurst 1991; Stouthamer et al. 1999; Huigens et al. 2000; Hiroki et al. 2002), whereas viral endosymbionts have only been found to cause male-killing phenotypes (Nakanishi et al. 2008; Kageyama et al. 2017; Nagamine et al. 2023). Depending on how long the victim lives, male-killing can be classified into two groups: “early male-killing,” in which the host male dies during the embryonic stage, whereas “late male-killing,” in which the host male dies while still a larva or pupa (Hurst 1991). Numerous insect species, including Coleoptera, Lepidoptera, Hymenoptera, Diptera, and Hemiptera, have been reported to exhibit early male-killing (Hurst et al. 1999; von der Schulenburg and Habig 2001; Morimoto et al. 2001; Anbutsu and Fukatsu 2003). This strategy indirectly enhances the possibility for vertical transmission of male-killing symbionts by improving female fitness through feeding on nutrient-rich conspecific eggs and increasing their survival, concentrating food resources on females, and reducing inbreeding due to the death of male siblings (Hurst and Jiggins, 2000; Jaenike et al. 2003). Late male-killing, on the other hand, increases the potential for horizontal transmission by killing males after the density of the male-killing symbiont has reached its maximum in the host. In addition to symbiotic species and strains, a number of additional factors influence the male-killing phenotype, including host genetic background, the density of the male-killing agent, and the existence of other symbionts in the same niche (Hornett et al. 2006; Kageyama et al. 2007; Lv et al. 2021).

One or more vertically transmitted symbionts often stably infect invertebrates. Co-infection with distinct symbiont species has occurred, such as Hamiltonella and Arsenophonus infecting greenhouse whitefly Trialeurodes vaporariorum (Hemiptera) and various viruses infecting fruit fly Drosophila melanogaster (Diptera) (Skaljac et al. 2010; Cross et al. 2020). In some circumstances, multiple strains of the same symbiont species co-infect, such as three Wolbachia strains that infect parasitoid wasp Leptopilina heterotoma (Hymenoptera) or the three partitivirus strains co-infecting African armyworm Spodoptera exempta (Lepidoptera) (Mouton et al. 2007; Xu et al. 2020). Although several infections by vertically transmitted symbionts have been documented, little is known regarding how the changes in infection patterns alter host phenotypes.

Homona magnanima (Lepidoptera: Tortricidae) is a major tea pest in East Asia. Male-killing in this species has been attributed to Wolbachia, Spiroplasma, and Osugoroshi viruses 1‒3 (OGVs) (Tsugeno et al. 2017; Arai et al. 2020; Fujita et al. 2021). OGVs belong to the family Partitiviridae and are segmented double-stranded RNA viruses (dsRNA viruses), with each segment individually encapsulated (Vainio et al. 2018). OGVs are composed of three RNA-dependent RNA polymerase (RdRp) genes and 24 genes of unknown function; the genomic composition of each of the 24 OGV segments is still undetermined (Fujita et al. 2021). Partitiviruses are known to infect fungi, plants, and protozoa (Vainio et al. 2018), but recent metagenomic studies (Shi et al. 2016; Wu et al. 2020) have revealed the invertebrate-infecting groups in flies (Cross et al. 2023; Kageyama et al. 2023), mosquitoes (Cross et al. 2020; Parry et al. 2021), and moths (Xu et al. 2020; Fujita et al. 2021). OGVs have been found to kill H. magnanima males at various developmental stages of their fifth instar life cycle (Arai et al. 2020), including the second to fifth instars (Morimoto et al. 2001), fifth instar to the pupa (Hoshino et al. 2008), and the second to third instars (Takamatsu et al. 2021). Molting or pupation is halted in OGV-infected males, resulting in their death (Hoshino et al. 2008; Supplementary Fig. S1). The above-mentioned differences in the lethal time may be related to OGV infection patterns and/or densities of OGVs, but the precise relationship remains uncertain.

To further understand the consequences of virus–virus interactions on the host, we aimed to establish several infectious patterns of OGVs in H. magnanima and compare attributes associated with male-killing phenotypes, such as hatchability, larval development, and viral load. Our findings indicated that the manner of infection of the various OGVs determines the duration of lethality. These results improve our understanding of recently discovered vertically transmitted viral interactions.

Materials and methods

Prevalence of OGVs in field population

To investigate the infection rates of the three OGVs in the field, larvae of H. magnanima from the third to fifth instars were collected in the field at a tea plantation in Ibaraki Prefecture, Japan, in April 2022. The field-collected larvae were reared individually in plastic cups on an artificial diet (INSECTA LFS; Nosan Corporation, Yokohama, Japan). Adult females were allowed to mate with either the field-collected males or males maintained in the laboratory, and the males were confirmed to be OGV-free. Diagnostic RT-PCR was used on both adult males and females to detect the presence of RdRp genes in the OGVs. Siblings whose female parents with OGV infection confirmed by diagnostic RT-PCR were subjected to microinjection to establish a new line.

Insects and lines

Three lines of H. magnanima were maintained in the laboratory. The normal sex ratio (NSR) line was negative for OGVs and Wolbachia, as previously reported (Tsugeno et al. 2017; Arai et al. 2022). The sex ratio (SR) distortion line was infected with OGV 1‒3. The SR line was established from larvae collected from Ibaraki Prefecture in 2021 and has been preserved for over 40 generations through mating with the NSR line. The SRΔ2 line, infected with OGV1 and OGV3, was established by microinjecting the filtrates from adult female abdomens collected in Ibaraki Prefecture, as described in the “Prevalence of OGVs in Field Population” section (Supplementary Fig. S2). In detail, the filtrate was homogenized in 100 µL of PBS per adult abdomen. Then, 1 µL of the filtrate was injected into fourth instar female larvae. Larvae were reared until adult emergence and mated individually. The sex ratio of pupae in each generation was counted. The presence or absence of OGV segment was determined after four generations. To eliminate Wolbachia, the SR and SRΔ2 lines were reared on Silk Mate 2S (Nosan Corp.) mixed with 0.1% (w/w) tetracycline for two generations. These lines were maintained in the laboratory as previously described (Tsugeno et al. 2017). To collect egg masses, 15 adult females from each line and 15 adult males from the NSR line were mated in a plastic container (20 cm × 30 cm × 5 cm) lined with paper. The egg masses were placed in a plastic box (20 cm × 15 cm × 5 cm), and the larvae were fed with an artificial diet (Silk Mate 2S) until pupation. Larvae and adults were reared at 25 °C with a 16 L:8D photoperiod.

Preparation of virus particles

To compare the morphology of viral particles in the SR and SRΔ2 lines, OGVs were purified using sucrose gradient centrifugation. Adult females from the SR and SRΔ2 lines were stored at −80 °C. Ten grams of dissected abdomen were collected, ground in liquid nitrogen, and suspended in 1× PBS buffer (pH 7.4). The suspension was centrifuged at 3000 × g for 20 min at 4 °C. The supernatant was mixed with half a volume of chloroform and centrifuged at 11,000 × g for 20 min at 4 °C. The water layer was collected and centrifuged at 100,000 × g for 2 h at 4 °C using a P45AT rotor (Hitachi, Tokyo, Japan). The resultant supernatant was layered onto a 20–50% sucrose gradient and then centrifuged at 70,000 × g for 2 h at 4 °C using a P28S rotor (Hitachi). Virus particles were collected by fractionator (DGF-U; Hitachi), diluted in PBS buffer, and centrifuged at 100,400 × g for 1 h at 4 °C (P70AT rotor; Hitachi). The pellets were resuspended in PBS buffer. Viral particles were stained with 2% uranyl acetate and examined using a transmission electron microscope (TEM) (JEM-1400Flash; JEOL Ltd., Tokyo, Japan).

SDS-PAGE

The virus particles were combined with an equal volume of EzApply (ATTO Corporation, Tokyo, Japan) and heated at 95 °C for 5 min. The samples were loaded onto a sodium dodecyl sulfate (SDS)-polyacrylamide gel (14%) stained with EzStain Aqua (ATTO Corporation). Protein bands specific to the SR and SRΔ2 lines were excised and subjected to in-gel digestion.

Tryptic digestion of the proteins and mass spectrometry

To identify the OGV structural proteins, proteins extracted from polyacrylamide gels were analyzed using LC-MS. To destain, the excised gels were washed thrice with wash buffer 1 (30% acetonitrile and 25 mM ammonium bicarbonate) and once with wash buffer 2 (50% acetonitrile and 25 mM ammonium bicarbonate). To dehydrate the gels, the wash buffer 2 was removed, 100 µL acetonitrile was added, and the gels were vortexed for 10 min. The gels were dried using a centrifugal evaporator. To reduce alkylation, 100 µL of reducing buffer (25 mM ammonium bicarbonate and 10 mM DTT) was added to the gels and incubated at 37 °C for 60 min. After incubation, the reducing buffer was removed, 100 µL of 25 mM ammonium bicarbonate was added, and the gels were vortexed for 10 min. After adding 100 µL of alkylating buffer (25 mM ammonium bicarbonate and 46 mM iodoacetamide), the gels were vortexed for 45 min in the dark. Following the reaction, the gels were washed once with 25 mM ammonium bicarbonate and twice with wash buffer 2. After removing the buffer, the gels were dried using a centrifugal evaporator. To digest proteins, a 20 µg/mL solution of trypsin in 50 mM ammonium bicarbonate was added to the gels and incubated overnight at 37 °C. Peptides were extracted from gels via sonication in 50 µL of extraction buffers 1 (50% acetonitrile and 0.1% formic acid), 2 (100% acetonitrile and 0.1% formic acid), and 3 (0.1% formic acid). The peptide extracts were concentrated to approximately 75 µL using a centrifugal evaporator.

The purified viral particle suspensions were immediately analyzed using LC-MS to avoid missing any abundant OGV-derived proteins. The protein concentration of the purified viral samples was determined using the Qubit^TM Protein Broad Range Assay Kits (Thermo Fisher Scientific, Waltham, MA, USA). After diluting 5 µg protein in NH₄CO₃ solution (50 mM) and 3 µL of DTT (500 mM), it was incubated at 95 °C for 5 min. The solution was then treated with 6 µL of 500 mM iodoacetamide and incubated at room temperature for 20 min. To digest the proteins in the solution, 1 µL of trypsin solution (100 µg/mL trypsin in 50 mM acetic acid) was added to the solution and incubated at 37 °C for 3 h, 1 µL of trypsin solution was again added, and the solution was incubated overnight at 30°C. The final solution contained 5% trifluoroacetic acid.

The in-gel-digested and protein-digested solutions were stored at −80 °C until further analysis. LC-MS analysis was performed using an LTQ-Orbitrap XL (Thermo Fisher Scientific). Mass spectrum data were processed using an in-house Mascot Server (Matrix Science Inc., Boston, MA, USA) and a custom database containing 27 OGV sequences (taxonomy ID: 2202811–2202814).

Host development (hatchability, survival rate, development, and pupal weight)

To compare the development during the embryonic stage between the three lines, 15 males from the NSR line and 15 females from each line were mated. Egg masses were then randomly collected and scanned using a G6030 scanner (Canon Inc., Tokyo, Japan). The area of each egg mass was measured using ImageJ (Abramoff et al. 2004). The number of eggs in the egg masses of the NSR line was counted, and linear regression was performed to investigate the relationship between egg number (y) and egg mass area [mm²] (x) (Arai et al. 2019, y = 4.192x + 4.298; R² = 0.99; n = 10). To confirm each line’s embryonic development, the numbers of eggs, late embryos, and unhatched late embryos were counted four days after hatching. The survival rate of each period was calculated using the following formulae:

• (i)Early embryonic stage: number of late embryos/estimated number of eggs.
• (ii)Late embryonic stage: (number of late embryos − number of unhatched late embryos)/number of late embryos.
• (iii)Hatchability: (number of late embryos − number of unhatched late embryos)/estimated number of eggs.

To compare the development during the larval and the pupal stages, newly hatched larvae were individually reared in plastic cups on an artificial diet of INSECTA LFS and monitored daily until adult emergence. The survival was determined by the presence or absence of a response to stimuli. The larval development time, larval and pupal mortality, and pupal weight were recorded.

Diagnostic PCR

Genomic DNA was extracted from adult abdomens as described previously (Tsugeno et al. 2017). Total RNA was extracted from adult abdomens using ISOGENII (NIPPON GENE Co. Ltd., Tokyo, Japan) following the manufacturer’s protocol, and from entire larvae using ISOGENII and EconoSpin columns (Epoch Life Science Inc., Missouri City, TX, USA) (Arai et al. 2023). Genetic sex was determined by confirming the presence of the W chromosome. Malpighian tubes were dissected under a stereomicroscope (Olympus-SZ61; Olympus Life Science, Tokyo, Japan) and stained with orcein lactate acetate, and sex was determined using a microscope (Nikon Eclipse CI; Nikon Instruments Inc., Tokyo, Japan) (Kageyama and Traut, 2004; Arai et al. 2022). The extracted RNA was treated with recombinant DNase I (Takara Bio Inc., Shiga, Japan). Complementary DNA was synthesized using AMV Reverse Transcriptase (NIPPON GENE Co. Ltd.) and random hexamers.

Diagnostic PCR was carried out using the Emerald Amp MAX master mix (Takara Bio Inc.) to detect RdRp genes for OGVs (Fujita et al. 2021) and the wsp gene for Wolbachia (Zhou et al. 1998) under the following conditions: initial denaturation at 94 °C for 2 min; 35 cycles of denaturation at 94 °C for 30 s; annealing at 62 °C; and extension at 72 °C for 30 s; followed by a final extension at 72 °C for 3 min. The primers used in this study are listed in Supplementary Table S1.

RT-qPCR

RT-qPCR was performed using StepOnePlus (Thermo Fisher Scientific, Waltham, MA, USA) and KOD SYBR® qPCR Mix (Toyobo Co., Ltd., Osaka, Japan). We used six or seven biological replicates for the larval stage and eight for the adult stage. Three technical replicates were examined for each biological replicate, and the reaction was carried out under the following conditions: 98 °C for 2 min; followed by 40 cycles of 98 °C for 10 sec, 60 °C for 10 sec, and 68 °C for 30 sec; and melting curves were generated. Gene expression was calculated using the ΔΔCq method. For the calculation of ΔCq, the average of technical replicates was used, while for the calculation of ΔΔCq, the average ΔCq values of the reference (0-day males for the larval stage and the SR line for the adult stage) were used. The elongation factor 1-alpha (ef1a) gene served as the reference gene (Arai et al. 2023). The primers used in this study are listed in Supplementary Table S1.

Statistical analysis

All statistical analyses were performed using R ver. 4.2.3 (R Core Team 2023). Log-rank tests were conducted using the Survival (Ver. 3.5, Therneau and Lumley 2015) and Survminer (Ver. 0.4.9, Kassambara et al. 2021) R packages to evaluate survival data, and p-values were adjusted for multiple comparisons using the Holm method. The larval development time and pupal weight were analyzed using the Steel-Dwass test with NSM3 (Ver. 1.18, Schneider et al. 2023) R package. Tukey’s multiple comparison test, or the U test was used to determine the hatchability and relative expression of RdRp genes.

Results

Prevalence of OGVs and established SRΔ2 line

To investigate the prevalence of OGVs, we collected 70 H. magnanima larvae from a tea plantation in the Ibaraki Prefecture. Of the 46 individuals tested, 13 tested positive for OGV1‒3. Five adults were infected with OGV1‒3, four with both OGV1 and OGV3, two with both OGV2 and OGV3, and two with OGV3 alone.

Egg masses were obtained from four adult wild females, one of which was infected with both OGV1 and OGV3. Wolbachia was eliminated by treating the larvae with tetracycline as previously described (Tsugeno et al. 2017), and the removal was validated by PCR. Given this genetic background, we injected filtrates of the OGV1- and OGV3-bearing females into the NSR line (G₀ generation). Two egg masses from the same mother in the injected G₀ generation (G₁ generation) had a slightly female-biased sex ratio (G_{1_a}: 12 males vs 46 females; G_{1_b}: 2 males vs 3 females). Some of the G₂ generations had a female-biased sex ratio (G_{2_f}: 0 males vs 42 females; G_{2_g}: 0 males vs 23 females). The female-biased sex ratio lasted until the fourth generation after injection (G_{4_f}) and the host line is hereafter referred to as the SRΔ2 line (Supplementary Fig. S2).

We looked for the viral dsRNA segments in adult hosts (Fig. 1; Table 1). Of the 27 OGV segments reported by Fujita et al. (2021), the SRΔ2 line was found to be lacking MK25 (RdRp), MK1, MK9, MK14, MK15, MK17, MK18, and MK26, while the remaining segments were positive. G_{4_a–d} was not infected with OGVs, and G_{4_e–g} possessed the same segment as the SRΔ2 line. These eight missing segments in the SRΔ2 line may potentially belong to OGV2. In addition, the SR line was positive for all OGV segments except MK4, whereas the NSR line was negative for all OGV segments. Given that the males in the SRΔ2 line died before the adult stage (Supplementary Fig. S2) and that the SR line maintained the male-killing phenotype in the absence of MK4, we assumed that these nine undetected segments were unnecessary for male-killing.

Fig. 1. Genome composition of OGVs revealed in this study.

Boxes indicated the position of open reading frames and lines indicates untranslated region. Red: RdRp; blue: CP; green: satellite dsRNA of which both the SR and SRΔ2 line possess; purple: satellite dsRNA of which only the SR line possess; yellow: satellite dsRNA of which only the SRΔ2 line possess.

Table 1.

Presence of dsRNA segments in three H. magnanima lines.

Line	RdRp				Uncharacterized segments
	OGV1	OGV2	OGV3		MK1	MK2	MK3	MK4	MK5	MK6	MK9	MK10	MK12	MK13
SR	+	+	+		+	+	+	−	+	+	+	+	+	+
SRΔ2	+	−	+		−	+	+	+	+	+	−	+	+	+
NSR	−	−	−		−	−	−	−	−	−	−	−	−	−

Line	Uncharacterized segments
	MK14	MK15	MK16	MK17	MK18	MK19	MK20	MK22	MK23	MK24	MK26	MK27	MKsp11	MKsp26
SR	+	+	+	+	+	+	+	+	+	+	+	+	+	+
SRΔ2	−	−	+	−	−	+	+	+	+	+	−	+	+	+
NSR	−	−	−	−	−	−	−	−	−	−	−	−	−	−

+ detected, − not detected.

Transmission electron microscopy and mass spectrometry detected putative capsid proteins of OGVs

TEM demonstrated that both the SR and SRΔ2 lines had a similar size, with an estimated size of approximately 20 nm (Fig. 2A, B).

Fig. 2. Virus particles and their proteins purified from the female abdomens.

Virus particles from the SR (A) and SRΔ2 lines (B). The scale is indicated in each panel. Protein bands unique to the SR or SRΔ2 line were excised and analyzed using to LC-MS (C). Identified proteins with protein scores greater than 100 are shown on the right. Arrowheads indicate protein bands unique to the SR and SRΔ2 lines.

The SDS-PAGE analysis of the purified virus particles indicated that both the SR and SRΔ2 lines exhibited a single major band at approximately 20 kDa and double minor bands at approximately 26 and 45 kDa (Fig. 2C). Using LC-MS of proteins extracted from polyacrylamide gels, three distinct OGV-derived proteins were identified in both the SR and SRΔ2 lines: MK6 (band 1) and MK23 and MK24 (bands 2 and 3) (Fig. 2C; Supplementary Fig. S3). We also analyzed non-specific bands, but no viral proteins were detected in the remainder of the bands. Furthermore, purified viral particle suspensions were also subjected to LC-MS analysis, and the aforementioned three virus proteins were identified in both the SR and SRΔ2 lines (Table 2). This suggested that the remaining putative proteins were either not present in the purified viral particles or that their quality was below the detection threshold.

Table 2.

Proteins identified from viral particles of OGVs.

		SR line			SRΔ2 line
Protein	Accession	Score	% Coverage	emPAI	Score	% Coverage	emPAI
MK6	BCP45655.1	3296	72%	2.4	119	17%	0.15
MK23	BCP45671.1	1908	57%	4.58	2207	54%	2.96
MK24	BCP45675.1	1638	40%	3.02	1508	29%	1.53

Score: Protein score of confidence in identification; % Coverage: Percentage of where the peptides that were identified match the indicated sequence; emPAI: Estimation of absolute protein amount.

OGV-induced male-killing differed in the lethal stage between the SR and SRΔ2 lines

The survival rate throughout the larval to adult stages was 89.3% for the NSR line (male: female = 58: 67), 49.3% for the SR line (0: 69), and 71.4% for the SRΔ2 line (0: 100) (n = 140 in each line, Fig. 3A, B). Males in both the SR and SRΔ2 lines died largely during the first and second instar stages (Fig. 3A). The SR and SRΔ2 lines had significantly lower survival rates than the NSR line (Kaplan–Meier log-rank tests corrected by the Holm method: p < 0.001), indicating male-killing in both lines.

Fig. 3. Survivorship curve of H. magnanima lines from the larval to adult stages.

Different letters denote significant differences between lines (Kaplan–Meier log-rank tests corrected using the Holm method: p < 0.05). Each figure indicates the survival curve (A) and sex (female) ratio at the adult stage (B).

Males in both the SR and SRΔ2 lines die in the larval stage. However, the SRΔ2 line had a significantly higher survival rate throughout the larval-adult stages than the SR line (Kaplan-Meier log-rank tests corrected by the Holm method: p < 0.001). If male-killing occurs exclusively during the larval stage, the survival ratio in the adult stage should be approximately 0.5, comparable to that observed in the SR line. The data demonstrated that the sex ratio of hatched larvae in the SRΔ2 line was biased toward females, indicating that a proportion of males died during the embryonic stage. To determine whether male-killing occurred during the embryonic stage, the survival rates of eggs from early to late embryogenesis, and late embryogenesis to egg hatching, were examined. The survival rates during the early embryonic stage did not differ significantly across the three lines (Tukey’s test: p > 0.05, Fig. 4A). Survival rates during late embryonic stage were significantly higher in the NSR and SR lines than in the SRΔ2 line (p < 0.001 for the NSR line and p = 0.002 for the SR line; Fig. 4B). The hatchability rate of the NSR and SR lines was found to be significantly higher than that of the SRΔ2 line (p < 0.001 for the NSR and SR line, Fig. 4C), indicating that male-killing also occurred during the embryonic stage in the SRΔ2 line. Therefore, the stages at which male-killing occurred differed between the SR and SRΔ2 lines.

Fig. 4. Development of three H. magnanima lines.

Different letters denote significant differences between lines (Tukey test: p < 0.05 (A–C), Steel-Dwass test: p < 0.05 (D, E)), and error bars denote standard deviation. Each figure indicates the developmental rate of early to late embryo (A), late to hatchling (B), early embryo to hatchling (C), larval development time (D), and pupal weight (E).

OGV infection patterns affected host development differently

To investigate the relationship between virus infection patterns and host development, we examined larval developmental time and pupal weight. Males in the SR and SRΔ2 lines did not survive into the adult stage (Fig. 3B); hence, only female data from the NSR line were used. The larval developmental time of the SRΔ2 line (26.2 ± 4.8 days) was significantly longer than those of the NSR (23.3 ± 4.7 days, Steel-Dwass test: p < 0.001) and SR lines (22.0 ± 2.8 days, p < 0.001). There was no significant difference in larval developmental time between the NSR and SR lines (p = 0.975, Fig. 4D). The pupal weight of the SR line (87.0 ± 8.6 mg) was significantly heavier than that of the NSR line (82.5 ± 9.9 mg, p = 0.010). However, there was no significant difference in pupal weight between the NSR and SRΔ2 lines (84.6 ± 11.6 mg) or between the SR and SRΔ2 lines (p = 0.247, Fig. 4E).

RNA load of OGVs varied among host lines

The RNA load of OGV RdRps in the SR line was examined during the second instar stage, when most males died. The second instar infected females molted three days after the last molt, whereas the infected males stopped molting and subsequently died (Supplementary Fig. S1). For a comparison, females six days after the first molting (4th instar, day 0) and males six days after the first molting, just prior to death, were selected. The RNA load was found to be significantly higher in males than in females, expect two days in OGV1 (U test: p < 0.05, Fig. 5A‒C). The fold change in RNA load between males and females from zero to two days after the first molting was between 1.5 and 4.0, whereas that at six days was between 4.0 and 8.1. We examined the RNA load of OGV1 and OGV3 RdRps in adult females from the SR and SRΔ2 lines. The RNA load of OGV1 was significantly higher in the SR line than in the SRΔ2 line (U test: p = 0.003, Fig. 6A), whereas there was no significant difference in the RNA load of OGV3 (p = 0.130, Fig. 6B).

Fig. 5. Relative expression of RdRp genes in males and females of the SR line from 0 to 6 days after the first molting.

Relative expression of OGV RdRp. A OGV1 RdRp. B OGV2 RdRp. C OGV3 RdRp. Asterisks indicate statical significance between sexes (U test: p < 0.05).

Fig. 6. Comparison of the expression of RdRp genes in females of the SR and SRΔ2 lines.

Relative expression of OGV1 RdRp (A) and OGV3 RdRp (B). Error bars denote the standard deviation (U test: p < 0.05).

Discussion

In this study, we established a new OGV double-infection line for H. magnanima, designated as the SRΔ2 line by injecting OGV-contained filtrate alongside the triple-infection line, the SR line. A comparative PCR and a series of LC-MS analyses identified potential structural proteins, thereby narrowing down the male-killing candidate. The developmental stage at which male-killing occurred differed between the SR and SRΔ2 lines. The male offspring of the SRΔ2 line died earlier than those of the SR line, but the viral RNA load of the female parents did not contribute to the developmental stage of male-killing. Our results revealed that the female larval development time of the SRΔ2 line was prolonged compared to that of the SR line and that the pupal weight, which is associated with their fecundity, did not exhibit a significant difference.

Partitiviruses typically have two dsRNA segments that encode the RdRp and capsid protein (CP). Some partitiviruses have one or two additional dsRNA segments (Ghabrial et al. 2008), including satellite dsRNAs that lack long open reading frames (Oh and Hillman 1995; Rong et al. 2002; Kim et al. 2005) or that may encode proteins, but their functions remain unknown (Cross et al. 2020). OGVs exhibited a greater number of satellite dsRNAs than plant- or fungi-infected partitiviruses (Table 1; Supplementary Fig. S4). Our findings indicated that the SR and SRΔ2 lines lacked one (MK4) and eight segments, including OGV2 RdRp, respectively. RNA viruses with separately enclosed segments, such as partitivirus, can lose satellite dsRNAs (Chiba et al. 2013; Kageyama et al. 2023). The loss of these genes did not affect male-killing, suggesting that some of the remaining 18 OGV genes could potentially be responsible for the male-killing phenotype. Given that the SRΔ2 line (infected with OGV1 and OGV3) and the OGV1-lacking line (OGV2 and OGV3; Fujita et al. 2021) showed complete male-killing, indicating that OGV3 may be involved in the male-killing phenotype. To investigate the relationship between the male-killing phenotype and viral load, the load of viruses was quantified during the second instar, when most males in the SR line began to die. Although the RNA loads of RdRp in the SR line were consistently higher in males during the second instar, and the difference was more pronounced just before death. This may suggest that increased viral RNA loads may be associated with male-killing. Because the timing of male lethality in the SRΔ2 line varied, it was not possible to pool multiple embryos to extract RNA (Arai et al. 2023), and thus RNA levels in embryonic male-killing could not be verified in this study.

We thought that RNA extraction from multiple individuals will homogenize the virus concentration, making it difficult to accurately quantify the virus. TEM analysis revealed that the virion structure of the SR and SRΔ2 lines was similar, forming particles of approximately 20 nm in diameter, suggesting that the eight segments lost in the SRΔ2 line did not impact the structure of OGV particles. Using the purified viruses from both the SR and SRΔ2 lines, we identified three potential structural proteins of OGVs. This finding is in direct contrast to the fact that partitiviruses are typically observed to encode a single capsid protein. MK23 and MK24 were detected in identical protein bands with molecular masses of approximately 26 and 20 kDa, respectively. The predicted molecular masses of these proteins are 26.0 kDa for MK23 and 25.7 kDa for MK24, suggesting that proteolytic degradation occurs following the expression of intact proteins. The processing of viral structural proteins has been documented in members of the Chrysoviridae, Totiviridae, and Quadriviridae families (Soldevila and Ghabrial, 2000; Urayama et al. 2010; Lin et al. 2012). In the present study, using LC-MS, we could not determine whether the N- or C-terminal region was missing (Supplementary Fig. S3); thus, additional protein sequencing, via Edman degradation, is required to clarify the missing region of the degradation.

The developmental stage of male-killing is influenced by symbiont densities and their infection patterns, as evidenced by the impact of bacterial density in female parents (Kageyama et al. 2007; Charlat et al. 2007; Arai et al. 2020). In H. magnanima, the RNA load of OGV in adult females was not higher in the double-infected SRΔ2 line than in the triple-infected SR line, though the male offspring of the SRΔ2 line showed earlier mortality than the SR line. This finding indicates that the RNA load of OGVs in female parents is not associated with the developmental stage of male-killing in the male offspring. It is possible that the male-killing characteristics of OGVs are influenced to a greater extent by the infection pattern of OGVs than by the RNA load. The infection patterns of symbionts can affect host phenotypes in various ways. For example, the Wolbachia strain wMel has been demonstrated to induce cytoplasmic incompatibility in the Aedes albopictus mosquito. However, when the wMel strain was co-infected with the wAlbA strain, its ability to induce cytoplasmic incompatibility was inhibited (Liang et al. 2020). In other cases, such as the Bemisia tabaci, Cardinium and Wolbachia caused male-killing only when they co-infected the same host (Lv et al. 2021).

It has been postulated that late male-killing presents a strategy for enhancing the potential for horizontal transmission, alongside vertical transmission, as symbiont density may increase in accordance with host development (Skinner 1985; Hurst 1991). In H. magnanima, horizontal transmission to uninfected females can occur via feeding on OGV-contaminated leaves. A mathematical model demonstrated that both vertical and horizontal transmission can achieve higher infection rates in the population than vertical transmission alone (Fujita et al. 2021). Males in the SR line die later than those in the SRΔ2 line, suggesting that triple-infection of OGVs may be more favorable for horizontal transmission. This assumption is supported by the results, which indicate that the highest prevalence of triple-infection of OGVs was observed in the field (Supplementary Table S2). The duration of larval development in females was found to be four days longer in the SRΔ2 line than in the SR line. A longer larval development time increases the risk of predation and may have adverse effect on fitness (Xue et al. 2012). This consequently reduces the possibility of vertical transmission of OGVs. These findings suggest that triple-infection with OGVs is more advantageous than OGV1 and OGV3 double-infection. Besides, additional investigation is required to elucidate the impact of other double-infection patterns, such as OGV1 and OGV2 or OGV2 and OGV3.

This study has demonstrated that viral infection patterns are a critical determinant of host phenotypes, including male-killing. Recent metagenomics studies have detected a diverse range of invertebrate virus genomes (Shi et al. 2016, Parry et al. 2021). A more comprehensive understanding of virus-virus-host interactions can be achieved by examining a broader range of infection combinations, including single infections.

Author contributions

TT conceived the idea of the study, conducted data analysis, and wrote the original manuscripts. TT and HA established the H. magnanima lines. CW, KK, TK, and TT purified the viruses and conducted TEM observations. YI and TT conducted LC-MS analysis. HM contributed to the discussion. MNI supervised all the experiments, prepared the manuscript, and contributed to the discussion.

Data availability

The experimental data that support the findings of this study are available in figshare 10.6084/m9.figshare.27950619.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41437-024-00741-x.