Introduction

Much attention has been increasingly paid to selfish genetic elements such as Wolbachia (eg, Werren et al, 1988; O’Neill et al, 1997). Wolbachia are a group of cytoplasmically transmitted bacteria prevailing widely among arthropods (eg Stouthamer et al, 1999). Wolbachia are primarily of interest because they can cause alterations to the reproduction of their hosts; male-killing (eg, Hurst et al, 1999), thelytokous parthenogenesis (eg, Stouthamer et al, 1993), feminization of genetic males (eg, Rousset et al, 1992), hybrid breakdown (Vala et al, 2000) and cytoplasmic incompatibility (eg, O’Neill et al, 1992). Wolbachia may have promoted rapid speciation in arthropods (Bordenstein et al, 2001).

The Asian corn borer, Ostrinia furnacalis (Lepidoptera: Crambidae) is a major pest of Zea mays in eastern and southeastern Asia. Miyahara (1984) first reported the occurrence of female-biased sex ratio (thelygeny) in Japanese populations of O. furnacalis, but its mechanism and causal agent were not identified. Kageyama et al (1998) found that thelygeny was inherited in a matriline of O. furnacalis and that feminization of genetic males was caused by a bacterial infection. However, the causal bacteria of the feminization in O. furnacalis remained unidentified.

In the present study, we reveal that the feminization of genetic males in O. furnacalis is caused by infection with a Wolbachia strain, and estimate the phylogenetic position of the Wolbachia strain based on the wsp gene sequence. This is the first study to report the occurrence of a feminizing Wolbachia in insects.

Materials and methods

Insects

We collected female adults of O. furnacalis from Matsudo (Chiba pref., Japan) in the summers of 1996–2000. Some of the findings of the present study, for insects collected in 1996, has previously been published by Kageyama et al (1998).

The captured females were individually allowed to oviposit in the laboratory. Most of the collected females laid fertile eggs within a few days. Larvae were reared by broods on an artificial diet (Silkmate 2M, Nihon-Nosan, Yokohama, Japan). At the pupal stage, insects were sexed based on the abdominal tip morphology. A piece of cotton soaked with 3% sucrose was provided for adult moths. Insects were reared under the conditions of 23°C and 15L/9D. Twenty females and 20 males were put in a mating cage, and 2 days later, females were separately allowed to oviposit. After oviposition, ovaries of females were dissected and stored in STE buffer (O’Neill et al, 1992) at −20°C until extraction of DNA.

Maternal inheritance of female-biased sex ratio

When a wild female produced a thelygenic family (P < 0.01, chi-squared test), daughters were used to found a matriline. Matrilines were maintained by crossing females with males from normal lines. A normal line was a pool of matrilines with the parental sex ratio not significantly distorted from 1:1.

Tetracycline treatment

Tetracycline hydrochloride was mixed into the larval diet at 0.06% wet weight and fed to the larvae from the neonate stage.

The tetracycline treatment was used to check whether the thelygeny was due to bacteria-induced feminization. Lepidopteran insects have been suggested to have either ZW/ZZ or ZO/ZZ sex chromosomes (Traut and Marec, 1996), and this was also assumed for O. furnacalis. If the production of all-male offspring occurs after antibiotic elimination of the feminizer, it indicates feminization of a genetic male, since the genotype of their mother should be ZZ to produce ZZ eggs exclusively. As another possibility, if a male-killing bacterium is causing the thelygeny, a 1:1 sex ratio should result from the tetracycline treatment.

PCR assay of Wolbachia infection

One of the pair of ovaries in a female adult was ground in 100 μl of STE buffer (O’Neill et al, 1992) with 2 μl of proteinase K (20 mg/ml), 1 μl of 2-mercaptoethanol and 10 μl of 10% SDS. The homogenate was incubated at 37°C for at least 30 min and at 95°C for 5 min. The lysate was extracted with phenol-chloroform and chloroform once each, and DNA was precipitated with ethanol. The DNA pellet was dissolved in 50 μl of TE buffer.

Polymerase chain reactions (PCR) specific to the Wolbachia 16S rDNA gene were conducted in 10 μl reaction volumes including 1 μl of DNA samples following O’Neill et al. (1992).

Sequencing and phylogenetic analysis

A PCR specific to the wsp gene of Wolbachia was performed (Zhou et al, 1998). The PCR products were purified using the GENE CLEAN III kit (Bio101, La Jolla, CA, USA), and both DNA strands were directly sequenced using the ABI 377 DNA sequencer with the BigDye terminator cycle sequencing kit (Perkin Elmer).

The wsp sequence of Wolbachia infecting O. furnacalis was initially aligned to 29 wsp sequences of other strains of B-group Wolbachia and two outgroup sequences from A-group Wolbachia (the Wolbachia strains from Drosophila melanogaster and D. sechellia) using the program package CLUSTAL W ver1.5 (Thompson et al, 1994), and the alignment was manually modified based on the inferred amino acid sequences using the sequence alignment editor BioEdit (Hall, 1999). The third hypervariable region (positions 518–565) was excluded from the analyses (Zhou et al, 1998). The molecular phylogenetic tree was estimated by the maximum likelihood method using the program package PAUP* (Swofford, 1996). We employed the general time reversible (GTR) model with discrete-gamma approximation (four rate categories). All model parameters were estimated from the data using a starting tree topology that had previously been inferred with unweighted maximum parsimony. These parameter estimates and the starting tree topology were then employed for a heuristic search based on tree bisection and reconnection (TBRs). Bootstrap analysis was done with 100 replications.

Results

Wolbachia infection and female-biased sex ratio

Three of 13 females collected in the field in 1996 were thelygenic (Figure 1). We found Wolbachia infection in two thelygenic females (M9 and M11, Table 1), but we did not examine infection in the other 11 wild females (Kageyama et al, 1998).

Figure 1
figure 1

Distribution of sex ratio in progenies produced by wild females of Ostrinia furnacalis collected at Matsudo in (a) 1996, (b) 1997, (c) 1998, (d) 1999 and (e) 2000. Boxes with circles and minus signs indicate Wolbachia-infected and uninfected broods, respectively. Broods without these signs were not examined for Wolbachia infection. Black and grey boxes indicate strong and weak SR broods respectively. Two broods with size less than 10 in (c) and (d) each were excluded from the figure.

Full size image
Table 1 Wolbachia infection and sex ratios of thelygenic matrilines (strongly and weakly) of Ostrinia furnacalis with egg hatch rates in parental families
Full size table

Wolbachia infection was found in 11 of 79 females collected in the field during 1997–2000 (Figure 1). Throughout the 4 years, the frequencies of infected females were rather low (0–43%). Ten of the 11 infected females produced all-female offspring (strong thelygeny), while no uninfected females produced the strongly thelygenic offspring (Table 1, Figure 1). One infected female (MD771) produced a brood with a sex ratio not strongly distorted (22 females and 10 males, P = 0.03 by chi-squared test). Six wild females that were not infected produced families with significantly female-biased sex ratios (P < 0.01 by chi-squared test), but these families included males at more than 20% (defined as weak thelygeny, Table 1, Figure 1) in contrast to the strong thelygeny seen in the Wolbachia-infected lines. In addition, two wild uninfected females produced families that appeared to be biased to female (10 females and two males, P < 0.05; 13 females and three males, P < 0.05; Figure 1).

The strong thelygeny was maternally inherited for all of the seven matrilines examined albeit with four exceptional cases (Table 1). Two F2 broods (three females and 10 males, and two females and two males) in matriline MD910 were not biased towards females. One F1 brood (22 females and seven males) in matriline M9 and the F2 brood (37 females and eight males) in matriline M11 showed rather weak thelygeny, although they are significantly female-biased (P < 0.01 by chi-squared test). On the other hand, the weak thelygeny disappeared in subsequent generations for both of the two matrilines examined (Table 1). We could not obtain offspring from the MD771 female.

Thus, Wolbachia infection was strongly associated with the all-female production in O. furnacalis females.

The mechanism of strong thelygeny

The average egg hatch rate for four families of strong thelygeny was 0.923 (Table 1), indicating that the Wolbachia-associated thelygeny was not due to early male-killing. All the larvae fed with tetracycline developed into females in all of seven matrilines of strong thelygeny, but these female adults produced all-male progenies (Table 2). The PCR assay confirmed that Wolbachia was eliminated from ovaries of tetracycline-treated females. These findings strongly suggest that the Wolbachia infection caused the strong thelygeny through feminization of genetic males in O. furnacalis. Namely, Wolbachia feminizes individuals carrying ZZ sex chromosomes (male genome), and such feminized individuals solely produce ZZ eggs that develop into male adults in the absence of the Wolbachia.

Table 2 Sex ratios of broods produced from tetracycline-treated strongly thelygenic females
Full size table

If Wolbachia-infected genetic females (ZW*, an asterisk indicates Wolbachia infection) were included among the strongly thelygenic females in addition to the feminized genetic males (ZZ*), they would have produced broods with normal sex ratio after the tetracycline treatment. In the present study, however, no such female was found (Table 2). The present finding can be explained if we assume that the infection of Wolbachia is old; Since ZZ* females produce solely ZZ* females while ZW* females produce ZZ* and ZW* females, the proportion of ZW* females will decrease sharply, and eventually almost all of the Wolbachia-bearing females will be ZZ*, provided that the fitness of ZZ* and ZW* females is similar.

Molecular phylogenetic affiliation of the Wolbachia strain in O. furnacalis

Nucleotide sequences of a portion (555 bp) of the wsp gene of Wolbachia in the 13 infected O. furnacalis females were determined. All the 13 sequences were identical (DDBJ/EMBL/GenBank nucleotide sequence databases with the accession number AB056664), suggesting infection with an identical strain of Wolbachia.

Molecular phylogenetic analyses have revealed the occurrence of two major clades (A and B) in Wolbachia strains infected with arthropods (Werren et al, 1995; van Meer et al, 1999; see also Vandekerckhove et al, 1999). The present analysis of wsp sequences showed that the Wolbachia strain in O. furnacalis is a member of the B group (Figure 2). Within the B group, the Wolbachia in O. furnacalis did not have a sister relationship to the feminizing Wolbachia strains in isopods, Armadillidium vulgare, Oniscus asellus and Porcellio scaber.

Figure 2
figure 2

Phylogenetic tree of B-group Wolbachia based on wsp gene sequence data with two outgroups of A-group Wolbachia from Drosophila sechellia and D. melanogaster. Wolbachia strains are given as their host species names. The tree was constructed by the maximum likelihood methods using PAUP* (Swofford, 1996). The tree has log likelihood of −4136.75. The results of 100 bootstrap replicates are shown above the branches. GenBank accession numbers of included published sequences are as follows: AB035514, AB038326, AB039284, AB045314, AB046720, AF020060, AF020065, AF020073, AF020076, AF020080, AF020083, AF020084, AF020085, AF071915, AF071916, AF071917, AF071927, AF217720, AF217725, AF237884, AF243436, AF295346, AF295347, AF295348, AJ130714, AJ130716, AJ269474, AJ269475, AJ269476, AJ271194, AJ271197.

Full size image

Discussion

In lepidopteran insects, thelygenies due to early (embryonic) male-killing have been known, some of which are caused by Wolbachia infection (Hurst et al, 1999; Jiggins et al, 2000, and references therein). In O. furnacalis, however, two lines of evidence indicated that Wolbachia infection causes feminization of genetic males. First, Wolbachia infection was strongly correlated to the strong thelygeny: this trait was found in 12 of 13 wild Wolbachia-infected females, but not in 68 uninfected females. Second, the result of antibiotic elimination of Wolbachia infection strongly suggested that feminization of genetic males underlay the strong thelygeny, although an alternative interpretation of the result is possible, as discussed below. This is the first study to report the occurrence of Wolbachia-induced feminization in insects, which has been known only from several isopod species (Rigaud et al, 1997).

It is possible to interpret the all-male production after antibiotic elimination of Wolbachia in a more complicated way; First, Wolbachia-infected females of O. furnacalis are assumed to be not viable without infection. This ‘obligatory infection’ hypothesis, in combination with a further hypothesis that Wolbachia causes male-killing during the larval stage (late male-killing), can explain the strong thelygeny in O. furnacalis. However, although a case of obligatory infection was recently reported in a parasitic wasp (Dedeine et al, 2001), Wolbachia have never been demonstrated to be obligatory in any other host species. Moreover, late male-killing due to Wolbachia infection has not been found in any animal to date. Taken together, it is very unlikely that the Wolbachia-induced strong thelygeny in O. furnacalis is caused by ‘obligatory infection and late male-killing’.

Among the 13 Wolbachia-infected females of O. furnacalis, one female (MD771) produced non-thelygenic progeny (22 females and 10 males, P > 0.01). Three explanations are possible for this exceptional case. First, the MD771 female may have harboured a Wolbachia strain different from the other 12 infections, although it was not distinguishable in terms of the wsp sequence. Secondly, the MD771 female may have failed to transmit the infection to some of the eggs. Lastly, the density of Wolbachia in MD771 females may have been low, leading to failure in feminizing some of the progeny. In addition, the latter two explanations might be relevant to the two non-thelygenic broods in matriline MD910 and the two weakly thelygenic broods in matrilines M9 and M11 (Table 1).

Wolbachia-induced feminization is known in isopods such as Armadillidium vulgare (Rigaud et al, 1997 for a review). In comparison, two phenotypic differences were found between the feminizations in A. vulgare and O. furnacalis. First, Wolbachia-infected lines of A. vulgare occasionally produce intersexes, which has not been found in O. furnacalis. Second, young females of Wol- bachia-infected A. vulgare, when reared at 30°C for eliminating the infection, progressively acquired the male phenotype within the treated generation. In O. furnacalis, however, tetracycline did not change the sex of individuals in the treated generation, and all-male offspring was produced in the subsequent generation.

These distinct phenotypic characteristics of feminization are most likely to be relevant to differences in sex determination and/or differentiation processes between A. vulgare and O. furnacalis. In A. vulgare, a ‘male gene’ has been suggested to control development of the androgenic gland that produces androgenic hormone. The androgenic hormone triggers the male differentiation after the fourth moult. Wolbachia may affect the activity of the male gene (Rigaud et al, 1997).

In insects, the sex determination process has been well elucidated only in D. melanogaster (for a recent review, Schütt and Nöthiger, 2000). The sex in Drosophila is determined at the embryonic stage, and is not affected by diffusing substances such as sex hormones in the later developmental stages. This mechanism is also supported for insects other than Diptera (eg, Hoy, 1994; but see also De Loof and Huybrechts, 1998). In congruence with this widely accepted notion, tetracycline-treated larvae of infected O. furnacalis developed into female adults, suggesting that the feminizing action of Wolbachia operates at the embryonic stage. The target(s) of the feminizing action may be some molecule(s) that play a role in sex determination during the embryonic stage of O. furnacalis, although the molecular mechanism of sex determination in Lepidoptera has not been elucidated. The silkworm Bombyx mori possesses a homolog of dsx, which is one of the Drosophila sex determination genes (Ohbayashi et al, 2001). The dsx homolog in B. mori, as well as dsx in Drosophila, is subjected to sex-specific splicing.

On the maximum likelihood tree of Wolbachia strains based on wsp sequences, the feminizing Wolbachia in Ostrinia does not have a sister relationship with that in A. vulgare, while the Wolbachia strains in isopods were monophyletic. This suggests that the evolutionary origins of feminization were independent in A. vulgare and O. furnacalis, provided that there has been no recombination between Wolbachia strains (but see Werren and Bartos, 2001).

Feminization mediated by a cytoplasmic parasite such as Wolbachia can increase the frequency of the parasite in the host population (O’Neill et al, 1997). However, our observation of Wolbachia infection in the Matsudo population of O. furnacalis during 5 years (Figure 1) did not show that the frequency of infected moths were increasing, but rather strongly suggested that the infection was maintained at a low prevalence. In isopods, the presence of nuclear gene(s) that resist to the vertical transmission or feminizing activity of Wolbachia has been argued (Rigaud and Juchault, 1992; Juchault et al, 1994). A similar nuclear resistance gene may explain the low prevalence of Wolbachia infection in O. furnacalis. However, such a resistance factor in O. furnacalis, if present, appears not to be functioning, at least under the laboratory conditions, because the rates of vertical transmission of Wolbachia were high and Wolbachia infection induced strong, not weak, thelgeny in O. furnacalis (Table 1, Figure 1). As another possibility, sexual selection against the feminized genetic males may be preventing the spread of Wolbachia infection in the O. furnacalis population, as has been argued in the case of A. vulgare (Moreau et al, 2001); however, this hypothesis must be critically tested through future studies.