Abstract
The pathogenicity of two granuloviruses (GVs), Xestia c-nigrum GV (XecnGV) and Pseudaletia unipuncta GV (PsunGV), was examined in Mythimna separata. Partial sequencing of the genome of PsunGV indicated that it is related closely to XecnGV, but considered to be a different species. PsunGV and XecnGV showed similar pathogenicity in terms of dose–mortality response and pattern of host mass changes following infection. Both GVs killed infected larvae in 2–3 weeks. Temporal changes in the concentrations of GV-specific DNA in the larval haemolymph were measured by using a real-time quantitative PCR. Viral DNA concentration increased quickly and reached a plateau at 60–72 h post-inoculation. Rates of budded virus (BV) production of each GV were estimated on the basis of viral DNA concentrations by a modified Gompertz model. The slopes of the estimated BV growth curves of both XecnGV and PsunGV in M. separata larvae were equivalent to that of Mamestra brassicae nucleopolyhedrovirus (NPV) in its original host, reported in our previous study. This suggested that BV production is not a major factor in the slower killing speed of GVs in comparison to NPVs. The GV-infected larvae survived for an additional 10 days or more after reaching a maximum level of BV concentration, and kept growing without pupation. These findings also suggested that the GVs have a unique mechanism to regulate the growth of host larvae.
-
The GenBank/EMBL/DDBJ accession numbers for the granulin, lef-8 and lef-9 gene sequences of PsunGV are AB290316–AB290318, respectively.
-
A supplementary figure showing a comparison of partial lef-8 and lef-9 genes of TnGV M10-5, PsunGV and XecnGV is available with the online version of this paper.
INTRODUCTION
Baculoviruses are characterized by double-stranded, circular DNA genomes and rod-shaped virus particles. Recently, a revision has been proposed for the family Baculoviridae based on molecular phylogeny and host insects (Jehle et al., 2006a). According to this revision, Granulovirus (GV), one of two genera of lepidopteran-specific baculoviruses, has been given a new genus name, Betabaculovirus. Three types of GV have been recognized based on their tissue tropism (Federici, 1997). Type 1 includes several noctuid GVs, such as Trichoplusia ni GV (TnGV). Type 1 GVs invade the host through the midgut epithelium and subsequently only infect the fat body tissues, resulting in a slow speed to kill relative to lepidopteran members of Nucleopolyhedrovirus (NPV), another genus within the family Baculoviridae. Type 2 GVs show broad tissue tropism similar to that of NPVs, and kill the infected host at a similar speed to NPVs. Type 3 contains only one member, Harrisina brillians GV, a virus that only infects the midgut epithelium. The molecular biology and genetics of GVs have been less well studied than those of NPVs, in part because of the difficulty of establishing cell lines that are permissive for GV infection (Winstanley & Crook, 1993).
The genome sequence of Xestia c-nigrum GV (XecnGV) has been determined and shown to contain 181 putative open reading frames (Hayakawa et al., 1999). High similarity in a large part of the genomic sequences between XecnGV and Pseudaletia unipuncta GV (PsunGV) is indicated by Southern blot hybridization, despite the marked difference in restriction-endonuclease profile between these two GVs (Goto et al., 1992). Some lepidopteran baculoviruses, including XecnGV and PsunGV, encode a unique gene family named enhancin, which promotes viral infection (Liu et al., 2006). A nucleotide sequence of PsunGV enhancin has been determined (Roelvink et al., 1995), although other regions of the PsunGV genome have yet to be identified. The deduced amino acid sequences of PsunGV enhancin and XecnGV enhancin-3 share 80.3 % identity (Hayakawa et al., 1999), whilst the identity between PsunGV and TnGV enhancins is 98 % (Roelvink et al., 1995). PsunGV has been studied as a synergist of NPV infection since the mid-1950s (Tanada, 1985). There are numerous reports on the enhancing effects of PsunGV and other noctuid GVs for NPV infection (reviewed by Corsaro et al., 1993; Liu et al., 2006). In contrast, few reports have been published on the pathogenicity of PsunGV and XecnGV (Goto et al., 1985; Tanada, 1959; Tanada & Hukuhara, 1968). These reports have suggested the possibility that PsunGV and XecnGV belong to the type 1 GVs.
The baculovirus produces two viral forms, occlusion-derived virus (ODV) and budded virus (BV), during its infectious cycle. The ODVs that are released from occlusion bodies (OBs) after ingestion by a host initiate primary infection in midgut cells. The BVs released from the primary-infected cells cause secondary infections in other host tissues. OBs are produced in the infected tissues, especially in fat body for type 1 GVs, during a late phase of infection. Microscopic observations of the infection process of GVs in host tissues have been performed with TnGV (Summers, 1969, 1971; Tanada & Leutenegger, 1970), Cydia pomonella GV (Hess & Falcon, 1987), Plodia interpunctella GV (Begon et al., 1993) and Epinotia aporema GV (Goldberg et al., 2002). However, an initially infected midgut cell has not been detected by microscopy, even by observation at short time intervals (Begon et al., 1993; Goldberg et al., 2002).
Analysis of the multiplication process of BVs in vivo will provide a key to understanding the slow killing speed of the GVs. We have reported that the BV titre of NPVs in the host haemolymph is quantifiable by using real-time quantitative PCR (RTQ-PCR) and that BV production follows a modified Gompertz model (Mukawa & Goto, 2006). The oriental armyworm, Mythimna separata (Walker), a close relative of Pseudaletia (Mythimna) unipuncta (Haworth) of North America, is an important insect pest of graminaceous crops in Asia and Australia. In this study, we examined the pathogenicity of XecnGV and PsunGV against their common host, M. separata, and characterized the multiplication of BVs by measuring viral DNA concentration in the larval haemolymph.
METHODS
Insects and viruses.
M. separata, a culture obtained from the Center for Ecological Research, Kyoto University, Japan, was maintained on an artificial diet, Insecta LFS (Nihon Nosan-Kogyo Co.), for more than 4 years at the National Agricultural Research Center. All experiments were conducted in our laboratory at 25 °C under a 16 h light/8 h dark photoperiod.
Two granuloviruses: XecnGV α-4 clone (Goto et al., 1985, 1992; Hayakawa et al., 1999) and PsunGV Hawaiian strain (Tanada, 1959; Tanada & Hukuhara, 1968) were used. These viruses were propagated in M. separata larvae and purified as described previously (Mukawa & Goto, 2006) with some modification. Centrifugal acceleration was increased to 15 000 g to pellet the OBs of the GVs (granules). To remove contaminants, the suspension of GV granules was mixed with an equal volume of glycerol and centrifuged at 15 000 g for 60 min. The resultant pellet of GV granules was washed three times at 11 000 g for 10 min with sterilized distilled water and desiccated at 4 °C. The dried GV granules were stored at 4 °C until use. The concentration of the GV granule suspension was quantified mainly by dry mass and secondly by phase-contrast microscopy using a Petroff–Hausser bacterium counter with a 0.02 mm depth.
PCR amplification and sequencing of the granulin (gran), lef-8 and lef-9 genes of PsunGV.
DNA from granules of PsunGV was extracted with phenol/chloroform as described previously (Goto et al., 1992). PCR amplification of portions of the lef-8 and lef-9 genes was performed by using degenerate primer sets (prL8-1B, prL8-2, prL9-1 and prL9-2; Jehle et al., 2006b). For amplification and sequencing of the gran gene, degenerate primers were synthesized as GSF001 (5′-ATGGGATAYAAYARAKCATTRAGATA-3′) and PER001 (Mukawa & Goto, 2006), in addition to the prPH1 and prPH2 primer set of Jehle et al. (2006b). Amplification reactions were performed by touchdown PCR as described by Herniou et al. (2004). The amplification products were purified by using Microcon-100 (Millipore). Direct cycle sequencing of entire PCR fragments was performed in both directions by using T7, M13(−21) and M13R universal primers with a BigDye v1.1 cycle sequencing kit (Applied Biosystems). The sequences were run on a capillary sequencer (ABI PRISM 3100 Genetic Analyzer; Applied Biosystems). Similarity searches were carried out by using the updated GenBank/EMBL/DDBJ database via the NCBI website and the blast algorithm (Altschul et al., 1990). Nucleotide sequences of three GVs, XecnGV α-4 (GenBank no. AF162221), TnGV M10-5 (GenBank no. AY519201, AY519202 and AY519203 for gran, lef-8 and lef-9, respectively) and another TnGV isolate named TnGV Indiana in this report (GenBank no. K02910 for gran), were used. Pairwise and multiple sequence alignments were performed by using clustal w (Thompson et al., 1994) implemented in the BioEdit program v. 7.0.1 (Hall, 1999). Nucleotide sequences of the gran, lef-8 and lef-9 fragments were concatenated to a single dataset for PsunGV, TnGV M10-5 and XecnGV as described by Lange et al. (2004). Pairwise distance of the aligned nucleotide sequences between the three GVs was calculated by applying the Kimura two-parameter model (Kimura, 1980) using mega v. 3.1 (Kumar et al., 2004).
Droplet-feeding bioassay.
Larvae of M. separata were inoculated with XecnGV or PsunGV by a modified droplet-feeding method as described by Mukawa & Goto (2006). The concentrations of both GVs used in the experiments were 101, 101.5, 102, 102.5, 103, 103.5 and 104 ng dry granules per larva. Larvae inoculated with a droplet without virus were used as a control. Experiments were replicated three times with 34–36 larvae per treatment. Larvae were observed daily for mortality until death or pupation. In order to collect haemolymph, larvae were inoculated with 10 μg dry granules of GV per larva as described above at 4–5 h after the beginning of the light period. This period was chosen because preliminary observations revealed that the timing of larval ecdysis of M. separata was affected by the photoperiod. Cell-free haemolymph was prepared as described previously (Mukawa & Goto, 2006). Haemolymph was collected individually from six larvae at 6 h intervals from 12 to 144 h post-inoculation (p.i.), and at 24 h intervals from 144 to 240 h p.i. Replicate inoculations were performed in order to determine the host body mass at 3, 4, 5, 6 and 10 days p.i. Twenty insects were weighed for control and each GV treatment at each time point p.i. (total of 100 insects for each treatment).
Quantification of viral DNA by RTQ-PCR.
Viral DNAs were extracted from BVs in the haemolymph of infected larvae as described previously (Mukawa & Goto, 2006). The primers [forward, 5′-AGGCGGCTATTCAAGCACTA-3′; reverse, 5′-ATGTTACGCAGCGTGTCAAG-3′ for the enhancin-3 (orf154) gene of XecnGV; forward, 5′-ATCAAGGACATCGCCAACCA-3′; reverse, 5′-TGCCGCTCCAGTTACACACA-3′ for the gran gene of PsunGV] were designed to amplify 124 and 139 bp long products, respectively. RTQ-PCRs were carried out under the same conditions: 95 °C for 10 min, followed by 40 cycles of 95 °C for 10 s and 65 °C for 50 s. Known dilutions of viral DNA were used as internal standards for each RTQ-PCR. Agarose-gel electrophoresis and thermal denaturation (melting-curve analysis) were performed to confirm specific replicon formation.
The concentration of BVs in larval haemolymph was estimated from measurement of viral genomic DNA by using RTQ-PCR, as the BV of baculovirus is generally composed of a single nucleocapsid (Williams & Faulkner, 1997) and RTQ-PCR quantification of viral DNA is highly correlated with BV titre (Lo & Chao, 2004). One copy of viral genomic DNA of XecnGV was calculated to be 1.96×10−7 ng on the basis of the XecnGV genome size, 178 733 bp (Hayakawa et al., 1999). One copy of PsunGV DNA was calculated to be 1.92×10−7 ng on the basis of the TnGV Indiana genome size, 175.6 kbp (Hashimoto et al., 1996).
Data analysis.
Probit analysis (Finney, 1978) using the computer program spss v. 11.5.1 (SPSS Inc.) was applied to the mortality data. The lethal times of larvae inoculated with 10 μg dry granules of GV per larva were used for survival analysis. A log-normal distribution was assumed for the data and parametric survival analysis was used to determine the significance of differences among the lethal times. Data of larval body mass from 3 to 6 days p.i. were compared by using two-factor ANOVA to evaluate the effect of virus treatment and time after inoculation. The growth curve of the BVs was estimated from the concentration of viral DNA in the haemolymph of infected larvae by a modified Gompertz model (Zwietering et al., 1990). The Gompertz parameters (A, maximum concentration of virus; μm, maximum rate of viral increase; λ, time that virus appeared in the haemolymph), which determine the shape of the BV growth curve, were estimated as described previously (Mukawa & Goto, 2006). jmp software v. 5.0.1 (SAS Institute) was used for the survival analysis, ANOVA and parameter estimation of the Gompertz equation.
RESULTS
Phylogenetic analysis of PsunGV and the relationship of PsunGV and XecnGV
PCR-amplified fragments of the gran, lef-8 and lef-9 genes of PsunGV were sequenced. The PsunGV gran sequence (582 bp; GenBank accession no. AB290316) that was obtained from fragments amplified with the four primers (GSF001, PER001, prPH1 and prPH2) was 72 bp longer at the 5′ end than the corresponding sequence of TnGV M10-5. The nucleotide sequence of PsunGV gran showed 89, 100 and 100 % identities to the corresponding regions of XecnGV (582 bp), TnGV Indiana (582 bp) and TnGV M10-5 (510 bp), respectively, whilst the deduced amino acid sequence of PsunGV granulin showed 100 % identities to the granulin sequences of XecnGV and two TnGV isolates. The obtained PsunGV lef-8 sequence (665 bp; GenBank no. AB290317) was 12 and 4 bp shorter at the 5′ and 3′ ends, respectively, than the corresponding sequence of TnGV M10-5. The obtained lef-9 sequence of PsunGV (269 bp; GenBank no. AB290318) was 2 bp longer at the 5′ end than the corresponding sequence of TnGV M10-5. Both lef-8 and lef-9 nucleotide sequences of PsunGV showed 99 % identities to the corresponding genes of TnGV M10-5, whilst the deduced amino acid sequences of these genes showed 99 and 100 % identities, respectively, to the LEF-8 and LEF-9 proteins of TnGV M10-5 (see Supplementary Fig. S1, available in JGV Online). The deduced amino acid sequences of PsunGV lef-8 and lef-9 showed 89 and 92 % identities, respectively, to the corresponding sequences of XecnGV. Distance matrices for the aligned gran, lef-8 and lef-9 nucleotide sequences of PsunGV, TnGV M10-5 and XecnGV are shown for single gene sequences and for the concatenated sequences in Table 1⇓. The pairwise distances between PsunGV and TnGV M10-5 were <0.010, whereas those between PsunGV and XecnGV were >0.100.
Pairwise distances for the nucleotide sequences of (top) lef-9 and lef-8 fragments and of (bottom) gran and concatenated gran/lef-8/lef-9 fragments of PsunGV, TnGV and XecnGV
Distances were calculated by using mega (Kimura two-parameter model). Bold values indicate that the two GVs are considered as the same species.
Pathogenicity of XecnGV and PsunGV
The LD50 values of XecnGV and PsunGV were 0.42, 0.24 and 0.58 μg dry granules per larva (4.9×106, 2.8×106 and 6.7×106 granules per larva) and 0.64, 0.48 and 0.95 μg dry granules per larva (5.9×106, 4.4×106 and 8.8×106 granules per larva), respectively, in the first to third trials (Table 2⇓). The 95 % confidence interval (CI) of the LD50 of XecnGV in the first and second trials did not overlap with that of PsunGV in the third trial, but the 95 % CIs of the LD50 values between the other combinations overlapped each other. The LD95 values of XecnGV and PsunGV were 1.7, 1.2 and 1.5 μg dry granules per larva (2.0×107, 1.4×107 and 1.7×107 granules per larva) and 2.4, 2.6 and 5.5 μg dry granules per larva (2.2×107, 2.4×107 and 5.1×107 granules per larva), respectively, in the three trials (Table 2⇓). The slopes of the probit mortality lines were not significantly different among all trials of XecnGV and PsunGV (parallelism test: χ2=8.12, d.f.=5, P=0.150).
Log-dose-probit parameters for XecnGV and PsunGV against M. separata fifth-instar larvae
All of the control larvae metamorphosed into pupae at the end of the sixth instar at 9–11 days post-mock infection. Following inoculation with 10 μg dry XecnGV granules per larva (1.16×108 granules per larva), all of the 108 larvae tested were infected and died during the sixth instar. The median times to death of the larvae inoculated with XecnGV in the three trials were estimated to be 15, 16 and 20 days p.i., respectively, and larval death occurred at ranges of 12–22, 12–24 and 13–23 days p.i., respectively (Fig. 1a⇓). Following inoculation with 10 μg dry PsunGV granules per larva (9.26×107 granules per larva), 104 of 105 larvae were infected in the three trials, 92.3 % of infected larvae died during the sixth instar and the remainder died after moulting to an extra-seventh instar. The median times to death of the larvae inoculated with PsunGV in the first to third trials were estimated to be 18, 19 and 19 days p.i., respectively, and larval death occurred at ranges of 15–29, 11–27 and 14–23 days p.i., respectively (Fig. 1b⇓). By likelihood-ratio test of the parametric survival analysis, the progression rates to death differed significantly between the two GVs (d.f.=1, χ2=19.61, P<0.001) but did not differ significantly among the three trials (d.f.=2, χ2=4.31, P=0.116). However, an interaction was detected (d.f.=2, χ2=7.32, P=0.026).
Proportion of surviving larvae of M. separata following inoculation with granules of XecnGV (a) and PsunGV (b) at a dose of 10 μg per larva.
Mean body masses of larvae inoculated with XecnGV or PsunGV are shown in Fig. 2⇓. Body mass was affected significantly by both treatment and day, and an interaction between treatment and day was observed (Table 3⇓). The effect of XecnGV treatment was not significantly different from that of PsunGV treatment; however, a significant difference in the body masses between control and virus-treated larvae was detected (Table 3⇓). All control larvae pupated at 10 days post-mock infection and the mean±sem pupal mass was 420±4.3 mg. Mean±sem body masses of larvae inoculated with XecnGV or PsunGV at 10 days p.i. were 968±23.6 and 1021±41.0 mg, respectively.
Mean body mass of larvae of M. separata at 3, 4, 5 and 6 days after inoculation with granules of XecnGV and PsunGV at a dose of 10 μg per larva. Bars indicate sem.
Results of two-factor ANOVA for larval body mass of M. separata inoculated with the two GVs
Multiplication of XecnGV and PsunGV in larval haemolymph
Viral DNA of both XecnGV and PsunGV was first detected in the haemolymph at 18 h p.i. The mean DNA concentration of XecnGV and PsunGV reached 105.36 and 104.64 copies μl−1, respectively, at 66 h p.i. The viral concentration of XecnGV subsequently decreased to 104.07 copies μl−1 at 114 h p.i., and then increased to 105 copies μl−1 after 120 h p.i. The viral concentration of PsunGV decreased to 103.73 copies μl−1 at 120 h p.i. and recovered to 104.86 copies μl−1 at 138 h p.i. Following inoculation with either XecnGV or PsunGV, the larvae were observed as pharate sixth instar at 54, 60 and 66 h p.i., and completed ecdysis by 72 h p.i. The viral DNA concentration data were fitted to a modified Gompertz model in order to generate viral growth curves (Fig. 3⇓). The three parameters (A, μm and λ) that determine the shape of the curves were estimated as shown in Table 4⇓.
Temporal changes in the mean concentration of viral DNA of XecnGV (a) and PsunGV (b) in the cell-free haemolymph of larvae of M. separata after inoculation with 10 μg granules per larva. Bars indicate sem. The solid line indicates the fitting curve estimated from the modified Gompertz model. The dotted lines indicate 95 % confidence limits of the regression curves.
Growth parameters of viral DNA of two GVs in M. separata larval haemolymph, fitted to the modified Gompertz model
DISCUSSION
Our analysis of the partial gran, lef-8 and lef-9 genes suggested that PsunGV is related more closely to TnGV M10-5 than to XecnGV. Jehle et al. (2006b) proposed species criteria for baculoviruses as follows: two (or more) isolates belong to the same species if pairwise Kimura two-parameter distances of single and/or concatenated gran, lef-8 and lef-9 genes are <0.015, and they belong to different species if the distances are >0.050. According to this classification, PsunGV and TnGV M10-5 belong to the same virus species, whereas PsunGV and XecnGV are different. Our results showed that the dose–mortality response of PsunGV was similar to that of XecnGV against larvae of M. separata. The patterns of mass gain of the larvae infected with XecnGV and PsunGV were also similar. The lethal times of these GVs were 2–3 weeks, which were longer than those of general NPVs and type 2 GVs by about 1 week (Federici, 1997). Thus, we conclude that XecnGV and PsunGV belong to the type 1 GVs.
Hackett et al. (2000) suggested that Helicoverpa armigera GV (HaGV) spreads through host tissues more quickly than Helicoverpa zea NPV (HzSNPV). This was hypothesized because, in larvae of H. zea that are co-infected with HaGV and HzSNPV, HaGV produces more granules in comparison to the number of polyhedra produced by HzSNPV, and the larvae survive longer than those infected only with HzSNPV. We have also observed a similar superiority of XecnGV over an NPV in fourth-instar X. c-nigrum (Goto, 1990). These reports suggest that GVs may outcompete NPVs during the early phase of infection. The BV concentrations of both XecnGV and PsunGV in the host haemolymph reached maximum within 72 h p.i. (Fig. 3⇑). The rate of BV multiplication in vivo was estimated by a modified Gompertz model on the basis of the concentration of viral genomic DNAs in the haemolymph. The slope of the viral growth curve during exponential growth (i.e. the μm parameter) was estimated to be 0.160 or 0.116 following inoculation with XecnGV or PsunGV, respectively. The μm value of Mamestra brassicae NPV (MabrNPV) is 0.145 with a 95 % CI of 0.124–0.171 when larvae of M. brassicae are inoculated with a minimum dose of MabrNPV, using a fluorescent brightener as a viral enhancer (Mukawa & Goto, 2006). The overlap of the 95 % CI of the μm values of XecnGV, PsunGV and MabrNPV indicates that the initial speeds of BV production of the two slow-killing GVs are equivalent to that of MabrNPV.
Electron-microscopic analysis has shown that TnGV progeny virions are found in the midgut cells of larvae of T. ni at 22–24 h p.i. (Summers, 1971). In the case of type 2 GVs, C. pomonella GV and E. aporema GV, infection is detected in the larval fat body of the homologous host at 24 and 36 h p.i., respectively (Hess & Falcon, 1987; Goldberg et al., 2002). In our study, the time lag between the onset of primary and secondary infection (i.e. the λ parameter of the modified Gompertz model) was estimated to be 7.6 and 11.2 h p.i. following inoculation with XecnGV and PsunGV, respectively. However, viral DNA of both XecnGV and PsunGV was detected for the first time in the haemolymph at 18 h p.i. Measurements of viral DNA suggest that the onset of secondary infection is later than 12 h p.i. The early detection of BVs in our study indicates that the sensitivity of viral detection by the RTQ-PCR method is higher than that by electron-microscopic observation. Further examination with much shorter intervals will clarify the actual time for initial BV appearance in the haemolymph.
The viral concentration of XecnGV was maintained at a level of 105 virions μl−1 after 120 h p.i. The maximum concentration of XecnGV in the haemolymph was estimated from the A parameter of the modified Gompertz model to be 105.04 virions μl−1, with a 95 % CI of 104.91–105.17 virions μl−1. The concentration of virions in the haemolymph estimated from the mean measured DNA concentration after 120 h p.i. was always higher than that obtained from the A parameter. The difference appeared to result from the decrease in the viral DNA concentration from 96 to 114 h p.i. This decrease may be associated with an increase in the volume of haemolymph resulting from larval growth rather than an absolute decrease in total viral DNA, because a linear increase in larval body mass was found following inoculation with the GVs. In the case of PsunGV, the A parameter gave an estimated virus concentration of 104.62 virions μl−1, which was also lower than the value calculated from the mean measured DNA concentration (i.e. 105.23 copies μl−1 at 240 h p.i.).
We found that XecnGV took approximately 60–66 h to reach a maximum concentration of virus (108 virions ml−1) in the host haemocoel, whilst the progress of PsunGV multiplication was slightly slower than that of XecnGV. Surprisingly, the rates of BV release into the haemolymph of M. separata larvae infected with these slow-killing GVs were similar to that in M. brassicae infected with the fast-killing MabrNPV. In the former case, larvae survived for an additional 10 days or more after the BV concentrations in the larval haemolymph reached a maximum level, whilst in the latter case, infected larvae died about 4 days after reaching a plateau concentration of the BV (Mukawa & Goto, 2006). Therefore, we conclude that the multiplication speed of the BV does not influence the survival time of the host larvae. Federici (1997) suggested that the long survival time of a host following infection with type 1 GVs is a consequence of tissue tropism (i.e. type 1 GVs may not attack important tissues other than the fat body). In our observations, both XecnGV and PsunGV prevented the pupation of infected hosts, but allowed them to moult. A supernumerary moult was observed following inoculation with PsunGV. This phenomenon was also observed in XecnGV-exposed larvae at a dose of 103.5 ng granules per larva (data not shown). The larval period was extended by GV infection and the body mass of the infected larvae reached roughly 1000 mg, a mass that was heavier than the maximum mass of the control larvae. This suggests that XecnGV and PsunGV regulate the growth of the host larvae to maximize production of the progeny virus. Prevention of pupation has been reported in Adoxophyes honmai infected with A. honmai GV, whilst a baculovirus ecdysteroid UDP-glucosyltransferase (egt) gene inactivating host ecdysteroid plays an important role in preventing pupation (Nakai et al., 2004). Interestingly, there is no homologue of the egt gene in the genome of XecnGV (Hayakawa et al., 1999). This suggests that XecnGV (and probably PsunGV) prevents host pupation by some unique mechanism without interfering with larval ecdysis. Further work is needed to investigate viral gene expression related to the alteration of normal host growth at a late stage of infection and the slow pathogenesis of GVs.
Acknowledgments
We thank Johannes A. Jehle for providing information about the PCR method and Takayuki Mitsunaga for valuable suggestions on the statistical analysis. We are grateful to Satoshi Nakamura (Japan International Research Center for Agricultural Sciences, Ohwashi, Tsukuba, Ibaraki, Japan) for supplying us with the culture of M. separata. We are grateful to Shizuo George Kamita and Yoshito Suzuki for critical reading of the manuscript. This research was partly supported by a grant from the National Institute of Agrobiological Sciences, Japan.