Functional Characterization of a Juvenile Hormone Esterase Related Gene in the Moth Sesamia nonagrioides through RNA Interference

Juvenile hormone esterase (JHE) is a carboxylesterase that has attracted great interest because of its critical role in regulating larval to adult transition in insects and other arthropods. Previously, we characterized an ecdysteroid sensitive and juvenile hormone non-susceptible juvenile hormone esterase related gene (SnJHER) in the corn stalk borer, Sesamia nonagrioides. SnJHER was rhythmically up-regulated close to each molt during the corn stalk borer’s larval development. In this paper we attempted to functionally characterize SnJHER using several reverse genetics techniques. To functionally characterize SnJHER, we experimented with different dsRNA administration methods, including hemolymph, bacterial or baculovirus-mediated RNA interference, (RNAi). Our findings indicate the potential implication of SnJHER in the developmental programming of Sesamia nonagrioides. It is still unclear whether SnJHER is closely related to the authentic JHE gene, with different or similar biological functions.


Introduction
Carboxylesterases (COEs) are a multifunctional superfamily ubiquitous in all living organisms [1]. Insect COEs have been the subject of intense research, in terms of their catalytic mechanism, molecular evolution and developmental regulation [2]. Based on sequence similarity and substrate specificity, insect COE genes can be subdivided into eight subfamilies: a-esterases, b-esterases, juvenile hormone esterases, gliotactins, acetylcholinesterases, neurotactins, neuroligins and glutactin class [1]. Lepidopteran insects have been known to possess a high number of COEs in their genome, sometimes making extremely difficult to distinguish them in terms of substrate specificity and biological function. For instance, in the cotton bollworm, Helicoverpa armigera, 39 putative carboxyl/cholinesterases (CCEs) sequences have been found scattering in its genome [3], while in the silkworm Bombyx mori the total amount of the putative CCEs genes is equivalent to 69 [4].
Juvenile hormone esterase (JHE) is a COE that has attracted great interest for its critical role in regulating larval to adult transition in insects and other arthropods. JHE hydrolyzes the key developmental and reproductive hormone, juvenile hormone (JH) and partially regulates its titer [5,6,7]. Juvenile hormone (JH) plays a major role in the control of growth, development, metamorphosis, diapause and reproduction in insects [8,9]. The onset of metamorphosis is preceded by a decrease in the biosynthesis of JH and an increase in JHE activity [7]. This then sets the stage for the elevation of ecdysteroid titer [10]. JH is normally present at the time of increase in ecdysteroid titers for larval molts and ensures that larvae molt to the next larval stage. However, at the time of the final larval molt, JH disappears allowing ecdysone to induce metamorphosis [11]. JHE is crucial for JH hemolymph titer reduction and therefore the initiation of metamorphosis in diverse insects. Strong inhibition of JHE activity in S. nonagrioides larvae has no effect on the onset of metamorphosis [12]. The transcripts of JHE-encoding genes that have already been described in insects are strongly induced by JH, e.g. Trichoplusia ni [13], Heliothis virescens [14], Leptinotarsa decemlineata [15], Choristineura fumiferana [16], Bombyx mori [17], Drosophila melanogaster [18] and Nilaparvata lugens [19].
In previous studies we characterized an ecdysteroid sensitive and juvenile hormone non-susceptible juvenile hormone esteraserelated gene (SnJHER) in the moth Sesamia nonagrioides (Lefebvre) (Lepidoptera: Noctuidae) [20]. SnJHER has all the typical motifs of JHEs (RF, DQ, E, GxxHxxD/E). The primary structure of the deduced amino acid sequence of the cDNA showed that the catalytic site of SnJHER has a cysteine residue next to the catalytic serine (GQSCG), while most described juvenile hormone esterases have alanine at this position (GQSAG). The JH analog methoprene did not affect SnJHER gene expression, whereas ecdysteroids and xenobiotics induced it. SnJHER mRNAs reached higher expression levels on the days close to each larval molt.
The corn stalk borer, S. nonagrioides, is a multivoltine species that causes significant damage on maize throughout the Mediterranean basin. Larvae that develop under long-day conditions invariably pupate at the end of the 6 th larval instar, while those grown under a short-day photoperiod enter diapause and undergo several supernumerary larval molts. Corn borer larvae programmed for diapause increase their body weights continuously until the 9 th instar [20].
RNA interference (RNAi) is a valuable tool for reverse functional genomics. In genetically transformable species, RNAi can be triggered by expressing long double stranded hairpin RNAs in the transformed cells and tissues [21] while in non-model insect species, RNAi can be triggered by delivering in vitro synthesized dsRNAs to a chosen stage (from egg to adult) and then examining the resulting phenotype [21]. Moreover in insects, RNAi can be induced via the oral route, either by feeding them directly with in vitro synthesized dsRNAs or with bacteria expressing the dsRNAs in vivo [22]. In comparison with producing dsRNAs in vitro, bacterially expressed dsRNAs is a low cost method and is more easily used in large scale gene function analysis [22]. In addition to the above RNAi techniques, RNAi can be triggered either by infecting insects with recombinant baculoviruses [23] or other viruses [24] that express the dsRNAs in the infected cells.
In this study we examined the functional role of SnJHER in the regulation of the corn stalk borer's larval, pupal and adult development, using several reverse genetics approaches. The dsRNAs were delivered indirectly by using either baculovirus or bacterial vectors or directly after hemolymph administration. Moreover, we investigated the relative capacity of each one of these techniques to induce a SnJHER RNAi (SnJHERi) specific phenotype. We conclude that SnJHER is implicated in the developmental programming of S. nonagrioides, however the exact mechanism of this regulation it is still unknown. Further biochemical and molecular data, are required in order to further elucidate the function of this particular esterase gene with key roles in the developmental regulation of S. nonagrioides.

RNA Silencing
Hemolymph dsRNA administration. For larval and prepupal stages we injected animals with specific dsRNAs which target three different regions of SnJHER cDNA, a 472 bp part of its 59-translated region (Fig. 1A), a 1276 bp part of its central translated, 39-translated and part of its 39 -untranslated region (Fig. 1A) and a 1725 bp part encompassing both of the above regions, which spans 94% of the total cDNA (Fig. 1A). The experiments were performed in independent triplicates (three trials) of a total amount of 100 insects. Each trial consisted of at least 30 insects either for control and experimental groups (Table  S2, S3, S4 and S5 in File S1). For RT-PCR analyses, we randomly selected 15 insects of each treatment and replicate, 3 days post injection. These were analyzed as individuals and subjected in semi-quantitative RT-PCR analyses in order to measure the SnJHER mRNA levels (Table S1 in File S1). Some examples of SnJHERi analyzed individuals are presented in Fig. 1B-1G.
For larval stage we injected animals of 5 th instar d3, in which the SnJHER mRNAs were higher comparing to the other larval stages [20]. Targeting the 472 bp part of the 59-translated region of SnJHER resulted in a decrease of SnJHER mRNA levels (Fig. 1B). In contrast to the transcriptional effect, no phenotypic impact associated with the decrease in gene expression, was observed in the injected population of 100 insects (3 independent trials). The same phenomenon was observed when we targeted the 1276 bp part of SnJHER (Fig. 1C). However, when we used the dsJHER 1725 , a SnJHERi specific phenotype (described in the next session) was obvious in an average of 5% of the total injected animals (N = 100, three independent trials), (Fig. 1D, Table S2 in File S1). In order to check whether there is a correlation between the used constructs and the efficiency of the RNAi after intrahemolymph administration in L5d3 larvae, we analyzed the semiquantitative RT-PCR data in terms of % silenced animals in the total population of the 15 randomly selected individuals (3 independent trials). Our results showed that, the percentage of the total silenced individuals is increased from dsJHER 472 to dsJHER 1725 . This increase was statistically significant (student's ttest, p,0.05, 3 biological replicates) comparing the dsJHER 1725 with the dsJHER 472 and dsJHER 1276 constructs, (Table S1 in File S1).
For prepupal stage we injected animals of 6 th instar d9. After prepupal JHERi, an average of .90% of the total (N = 100, three independent trials) dsJHER 472 bp/ or 1276 bp/ or 1725 bpinjected animals failed to ecdyse to the next pupal instar and died as larval-pupal intermediates, (Table S3, S4 and S5 in File S1; see next session). Randomly selected individuals 3 days post injection, were subjected to semiquantitative RT-PCR analysis in order to measure the SnJHER mRNA levels. All of the randomly selected larval-pupal intermediates were found to contain low SnJHER mRNA levels, when compared to the control-untreated ones (Fig. 1E, 1F and 1G). We note that in other lepidopteran insects, maximum silencing effects were observed between 2-4 days postinjection [25,26].
Bacterial administration. For bacterial dsJHER administration we used the same part of the 59-translated region of SnJHER that previously resulted no phenotypic effects (dsJHER 472 ) (Fig. 1A). This was expressed in the RNAse III deficient HT115 (DE3) E. coli strain. The bacterial administration would be more persistent and prolonged comparing with the hemolymph administration method.
The insects were fed in artificial diet that was supplemented with the dsJHER expressing bacteria. Three bioassays were subsequently performed. In the first bioassay the IPTG induced bacteria were supplied to S. nonagrioides larvae from the d0 of the first instar till the d0 of the 5 th instar d0 (N = 100, three independent trials). After this period of continuous feeding, the insects were placed to their normal diets and sampled for RT-PCR analysis 6 and 15 days post recovery. Pools of ten randomly selected insects were subjected to RT-PCR analyses. The RT-PCR showed a decrease of SnJHER mRNA levels 6 days post recovery (5 th instar d6) while the mRNA levels returned to a normal accumulation 15 days post recovery (6 th instar d5) (Fig. 2B). Since no phenotypic effect was observed in the first biossay, a second one was performed, in which the induced bacteria were applied to S. nonagrioides larvae for a more extended time, representing the entire larval life of the insect (L1d0RL6d9, N = 100, three independent trials). Also in this case a SnJHERi specific phenotype was not observed (data not shown). In a third biossay, the insects were fed with dsRNA expressing bacteria for a shortened period of 7 days, from 5 th instar d0 until 5 th instar d7 (N = 100, three independent trials). Ten randomly selected individuals were RNA extracted, mixed in one RNA pool and subjected in RT-PCR analyses for semi-quantifying the SnJHER, mRNA levels (Fig. 2C). In addition, expression of other developmental genes, SnHSPs and SnEcR, which were expected to be affected by SnJHER knockdown, was also quantified. The sampling day was chosen according to previous reports in which maximum silencing is observed after 7 days of continuous feeding with the dsRNA producing bacteria [22,27]. It was indeed observed that continuous feeding with dsRNA expressing bacteria resulted in significant knockdown of SnJHER mRNA and that silencing of SnJHER affected expression of the other developmental genes ( Fig. 2C; see further below). Also in this case a SnJHERi specific phenotype was not observed.
Baculovirus-mediated RNAi. For baculovirus-mediated RNAi we used the same part of the 59-translated region of SnJHER that previously resulted into no phenotypic effects, either with hemolymph or bacterial mediated administration (dsJHER 472 ) (Fig. 1A). The appropriate baculovirus strain was selected according to the supplementary text (Text S1 in File S1). Insects of several developmental stages were injected with 10 7 pfu/ ml of the BmNPV-BmA::GFP/BmA::JHER loop virus. As control we used a virus expressing double stranded molecules of the luciferase gene (BmNPV/BmA::GFP-BmA::dsLuciferase virus). The infections were carried out at two different developmental stages, 5 th instar d3 (larval stage) and 6 th instar d9 (prepupal stage) larvae.
The experiments were performed in independent triplicates (three independent trials) of a total amount of 200 insects. Each trial was consisted of at least 60 insects either for control and experimental groups, (Table S6, S7 in File S1). Following infections, the insects were allowed to complete their development recording daily for phenotypic effects suggestive of developmental abnormalities. For RT-PCR analyses insects were sampled 7 days post infection, the day in which the maximum infectivity of the BmNPV virus in terms of GFP expression was observed (Fig. S3).
Randomly selected SnJHERi positive animals were subjected to semi-quantitative RT-PCR analysis in order to measure the SnJHER, GFP and construct specific mRNA levels. In Fig. 3B we present a particular case of an analyzed individual. This analysis showed that the SnJHER mRNA levels were specifically decreased, 7 days post infection in the BmNPV-BmA::GFP/BmA::JHER loop infected animals, while the GFP mRNA levels were similar in both BmNPV-BmA::GFP/BmA::dsLuciferase and BmNPV-BmA::GFP/BmA::JHER loop infected larvae (Fig. 3B). The JHER hairpin mRNA was only expressed in the BmNPV-BmA::GFP/ BmA::JHER loop infected larvae (Fig. 3B). Moreover, 14 SnJHERi phenotype positive BmNPV-BmA::GFP/BmA::JHER loop infected animals and their BmA::GFP/BmA::dsLuciferase infected controls (5 from trial 1, 4 from trial 2 and 5 from trial 3, Table S6 in File S1) were subjected to quantitative RT-PCR analysis, in order to measure the average silencing levels (Fig. 3C). This analysis showed that the SnJHER mRNA levels were decreased by 20% in the BmNPV-BmA::GFP/BmA::JHER loop infected larvae 7 days post infection, when compared to the control BmA::GFP/ BmA::dsLuciferase infected ones (Fig. 3C). Fig. 3C represents the relative expression levels of SnJHER, normalized to these of b-

Phenotypic Analysis
Hemolymph administration. Figure 4B shows the phenotypic impact of prepupal hemolymph administration of dsJHER 472 , dsJHER 1276 and dsJHER 1725 dsRNAs. Prepupal hemolymph administration of each one of these constructs resulted in larval-pupal intermediates in an average of .90% (three independent trials; Table S3, S4 and S5 in File S1) of the total injected population of N = 100 insects (Fig. 4BI, 4BII and 4BIII). These differences were statistically significant since none of these phenotypic abnormalities were presented in dsL4440 MCS injected controls (Student's t-test, p,0.05; Table  S3, S4 and S5 in File S1). On the contrary, in larval hemolymph administration, none of the dsJHER 472 and dsJHER 1276 constructs resulted in a SnJHERi specific phenotype, despite the efficient transcriptional silencing of the gene (data not shown). However, administration of SnJHER 1725 resulted in a SnJHERi specific phenotype in an average of 5% of the total injected (N = 100, three independent trials, Table S2 in File S1) animals (Fig. 4C). These insects were incapable to shed their old cuticles, which were fused with the newest synthesized epidermis dying during the molting procedure (Fig. 4C). This phenotype was obvious in all three independent replicates, and never was presented in the dsL4440 MCS injected controls (Student's t-test, p,0.05; Table S2 in File S1).
The surviving pupae were allowed to complete their development. Between the two viruses no significant change in the pupal-adult transition was observed when insects were infected in the 5 th instar (Table 1). On the contrary, when insects were infected in the prepupal stage, 26% of the BmNPV-BmA::GFP/BmA::JHER loop infected animals failed to emerge to the adult stage as compared to the 10% of the BmA::GFP/BmA::dsLuciferase infected animals ( Table 1, three independent trials, Student's t-test p,0.05). Moreover, the emerged adults of both cases shared several developmental abnormalities, which could be distinguished in several categories; e.g. normal adults, curly wings/pupal head, scale-less (Fig. 6). The scale-less category was presented only to the BmNPV-BmA::GFP/BmA::JHER loop infected animals in a percentage of 9% of the emerged adults and only on them which were infected in the last larval instar ( Fig. 6; Table S8 in File S1, three independent trials, Student's t-test p,0.05).

SnJHERi Influences the mRNA Synthesis of S. nonagrioides Ecdysone Receptor, Heat Shock Cognate 70 and Heat Shock Protein 70 Genes
In order to shed light on the potential molecular communication between SnJHER and several components of the basal endocrine system of S. nonagrioides, including the ecdysone receptor (GenBank: JN572102), heat shock cognate 70 (Gen-Bank: DQ004584) and heat shock protein 70 (GenBank: EU430480) genes, we selected to silence SnJHER through bacterial RNAi. This method should cause less stress effects in treated insects, in terms of technical handling (injections versus infections). In that way, any transcriptional alteration in HSP gene expression, will has been caused by the RNAi effect and not the technical stress. Semiquantitative RT-PCR analysis of SnJHER mRNA in RNA pools of 10 randomly selected individuals seven days post feeding, showed a decrease in SnJHER mRNA levels, suggesting efficient JHER silencing (Fig. 2C). Moreover, the mRNA accumulation of SnHsp70 is increased while the SnHsc70 and SnEcR mRNA accumulation is decreased after SnJHER silencing (Fig. 2C).

Summary of the Results
As we can see in Table 2, hemolymph administration of dsJHER in prepupal stage resulted in a SnJHERi specific phenotype in an average of .90% of the total injected prepupae (N = 100 for each construct used, three independent trials) by any construct used. In contrast to the prepupal, in larval RNAi only injection with the longest dsRNA (dsJHER 1725 ) resulted in a SnJHERi specific phenotype in the 5% of the total injected (N = 100 animals, three trials). Bacterial RNAi, on the other hand, resulted in no phenotypic effects, although efficient knockdown was achieved. On the contrary, baculovirus-mediated dsJHER administration resulted in a SnJHERi specific phenotype in an average of 14% of the total infected instead of the 5% of the total dsJHER 1725 injected animals (three independent trials). Moreover, baculovirus-medi- ated RNAi complemented the prepupal RNAi through hemolymph administration resulting into useful information regarding the pupal-adult transition.

Discussion
This paper describes the functional characterization of a JHE related gene in the moth S. nonagrioides. For this insect, standard reverse or forward genetics techniques have not been established, hindering any attempt for an accurate functional characterization of an experimental gene. For non-model lepidopteran insects, several reverse genetics techniques have been used in the past, with variable efficiencies and successes regarding the obtained results. In order to cover the whole spectrum of the technical difficulties that could result into erroneous experimental conclusions, we selected to use different RNAi approaches. These differed in terms of the dsRNA delivery method including hemolymph, bacterial or baculovirus-mediated. Our combined results shed light on the functional characterization of SnJHER and the efficacy of the techniques used for this characterization.

RNAi Efficiency
In larval RNAi through hemolymph injection, the incapability of dsJHER 472 and dsJHER 1276 to develop a SnJHERi specific phenotype, despite efficient silencing of the gene (Fig. 1A, 1B, 1C), could be explained by several reasons related to the efficiency of this method. In a publication, which described the efficiency of RNAi after hemolymph administration in the model organism Tribolium castaneum, Sherry et al. [28] found that longer dsRNAs were more effective than shorter ones with respect to both the initial knockdown and the duration of the RNAi effect. This was not due to differences in length per se, but to the fact that the longer dsRNA produce a greater variety of siRNAs some of which could be more effective at silencing level than the limited number of siRNAs produced by the shorter dsRNA fragment. Moreover, the size of dsRNA seemed to have a more drastic affect on the duration of RNAi than on the initial RNAi efficiency [28]. Here, we observed almost the same phenomenon, in terms of the efficiency of the hemolymph dsRNA administration to produce a phenotype associated with the SnJHER RNAi. While dsJHER 1725   injection resulted in abnormal larvae in the 5% of the total injected animals (Fig. 4C), dsJHER 472 and dsJHER 1276 only caused a reduction of SnJHER mRNA levels without producing any SnJHERi specific phenotype (Fig. 1B, 1C). We speculate that dsJHER 1725 was more effective regarding the duration of the RNAi effect and cell penetration efficiency as happened with T. castaneum. In contrast to the larval injections, in the prepupae all constructs were capable of producing a SnJHERi specific phenotype (Fig. 4B). The stage dependent effectiveness of RNAi in terms of phenotype production in other Lepidopteran species has previously reported [29]. In the silkworm, B. mori it has been claimed that the stage of early wandering (EW) larvae (prepupal stage) is more sensitive to RNAi [30]. Furthermore, in Drosophila melanogaster, larval RNAi through hemolymph dsRNA administration is not effective for many tissues, despite the success of adult injection. This may suggest different tissue specificity at different developmental stages; the basis of the difference between larval and adult tissues is still unknown, but may be due to fundamental developmental differences between tissue types, such as cell ploidy, or due to differences in gene expression required for the uptake and transport of dsRNA [31]. Bacterial administration of dsJHER 472 had no developmental consequence in S. nonagrioides larvae, despite the efficient silencing of the gene, at any treatment tested (Fig. 2B, 2C). This could be explained as previously, by factors that have to deal with the RNAi ineffectiveness, with respect to the initial knockdown, duration of the RNAi effect and cell penetration efficiency.
The phenotype produced by infection with the BmNPV/ dsJHER virus was similar with this of the hemolymph administration of dsJHER 1725 in the 5 th larval instar of the insect (Fig. 4C,  5A). The Autographa californica nuclear polyhedrosis virus (AcMNPV) and Bombyx mori nuclear polyhedrosis virus (BmNPV) encode an ecdysteroid UDP-glucosyltransferase (EGT) gene, which inactivates ecdysone by conjugating the hydroxyl group at C-22 with a sugar [32,33,34,35]. Insects infected with a virus containing the gene encoding EGT do not molt because of a lack of active ecdysone [32,33,35]. BmNPV's infections showed a high blockage of larval-pupal and pupal-adult molt of S. nonagrioides in contrast to the AcMNPV's infections in which the molting arrest was observed in larval instars as well (see Text S1 in File S1). We speculate that BmNPV in contrast to AcMNPV's EGT is not effective enough to block molting procedure during the larvallarval molts of the corn stalk borer, probably due to its gene expression dynamics or enzyme's biochemical efficiency; maybe an indirect consequence of the BmNPV's host incompatibility. Baculovirus administration method may be more effective in terms of the systematic distribution of the expressed dsRNAs. Consequently, baculovirus-mediated RNAi could not be applied directly to functionally characterize genes of the lepidopteran insects since in pupal and adult stages the infection effects mask the potentially produced phenotypes after the loss of function of the experimental gene (Fig. 5B, 6).

SnJHER Function
We have previously shown that SnJHER was rhythmically upregulated close to each molt during S. nonagrioides larval development [20]. Additionally, we demonstrated that while the JH analog methoprene does not affect SnJHER gene expression, ecdysteroids induce the SnJHER mRNA synthesis. Combining these two results, we speculate that SnJHER is following the ecdysteroid titer of S. nonagrioides in contrast to other conventional JHE genes of several insects in which the JHE mRNA levels are following the JH titer. The findings of the current work complete our previous results suggesting that SnJHER is implicated in larvallarval molt of S. nonagrioides larvae. Insects of 5 th instar d3 injected with 4 mg of dsJHER 1725 or infected with 50 ml of 10 7 pfu/ml of the BmNPV-BmA::GFP/BmA::JHER loop virus presented several developmental abnormalities including a total failure to complete the molting process (Fig. 4C, 5A). Searching the global contemporary literature, there are no similar works in other insect species, in order to compare them with our results. On the contrary, there are many publications describing the application effects of several JH analogs or JHE inhibitors in the development of many insect species including the lepidopterans. In the neuropteran, Chrysoperla  carnea larvae treated with the juvenile hormone analog fenoxycarb showed two major alterations of pre-imaginal development: i. the inhibition of metamorphosis and ii. the inhibition of cocoon spinning [36]. In this species, metamorphosis was strongly affected by fenoxycarb. Aside from the presence or not of a complete cocoon, a high percentage of larvae did not succeed in metamorphosing to adults. Insects were considered to be affected by inhibition of metamorphosis when: a. they were still alive when the non-affected larvae were already inside a cocoon and b. they continued as larvae, prepupae, pupae or pharate adults, never becoming adults. When mortality occurred in the period when the non-affected larvae had already metamorphosed, this was considered to be a consequence of metamorphosis inhibition, although the exact cause of mortality was unknown. Should JHER be a JHE conventional gene, we would expect that larval SnJHERi in the corn stalk borer would cause an extension in larval life rather than a blockage in the molting procedure. We conclude that SnJHER could be implicated in the ecdysteroid instead of the JH signalling of S. nonagrioides by interfering with the molting process. Our data demonstrated that SnJHERi through hemolymph administration, resulted in a total failure of larval-pupal metamorphosis in .90% of the total injected prepupae by all used constructs (dsJHER 472 , dsJHER 1276 and dsJHER 1725, Fig. 4A). In addition, BmNPV-BmA::GFP/BmA::JHER loop infected prepupae shared several categories of developmental abnormalities of larval-pupal intermediates with the control infected ones, but only one of them could be considered as SnJHERi specific (Fig. 5B, case 3). Previously we have shown that SnJHER mRNAs were low in the beginnings of the 5 th instar and increased gradually until L5d3, just before the larval ecdysis [20]. In the 6 th (last) larval instar, SnJHER mRNAs were lower than those of the 5 th instar and increased gradually from L6d4 to L6d5, when they peaked; on the next days, the transcripts declined and disappeared [20]. The total physiological impact of SnJHERi in the final larval instar of the corn stalk borer suggests that this gene is important for the larvalpupal transition despite its low expression during the L6. Moreover the low SnJHER mRNA levels in L6d9 compared with the high mRNA levels in L5d3 could be the reason for the total observed abnormality in this particular instar after SnJHERi.
Infection of prepupae with the BmNPV-BmA::GFP/BmA::J-HER loop virus resulted in a decrease of adult emergence compared with the control treated ones (Table 1). Moreover, some of the BmNPV-BmA::GFP/BmA::JHER loop emerged adults presented a scale-less wing morphology (Fig. 6). The deficiency in pupal-adult transition was also observed in Spodoptera exigua EcR silenced animals. In this species silencing of SeEcR resulted in a total adult malformation and emergence complications. In Sesamia the adult abnormalities which were observed after the silencing of SnJHER in the last larval instar indicate that they may be a possible consequence of the indirect downregulation of SnEcR, caused by the SnJHERi. However due to the common deficiencies between the control and experimental groups, an indirect effect of the virus infection, it is still extremely difficult to distinguish the JHERi effects in this particular instar with baculovirus-mediated dsRNA administration [37].
Here we also showed that two major components of the ecdysteroid pathway, the SnEcR and the SnHsc70 genes are downregulated after SnJHERi (Fig. 2C). There are numerous studies showing the importance of the ecdysone receptor gene in the control of the molting process. In D. melanogaster mutants of the EcR-B isoform present predominant time of death between the 1 st and 2 nd larval stages [38]. Many of the dead EcR-B mutants carry a duplicated larval cuticle suggesting that they have arrested during the process of larval molting. The Drosophila Hsc70 is required for activation of the EcR/USP heterodimer in vivo [39]. The USP polypeptide folds appropriately into a relatively stable configuration that is not further stabilized by chaperones. By contrast, the EcR polypeptide folds into an unstable configuration easily subject to irreversible unfolding or protease degradation. The unstable EcR interacts with appropriate Drosophila chaperones including Hsp90 and Hsc70, which stabilize EcR in a configuration appropriate for formation of EcR/USP heterodimers capable of binding EcRE DNA sequences [39]. The phenotypic results of the SnJHERi experiments showed similar molting deficiencies with those of the EcR-B mutants of Drosophila. Considering the importance of DmHsc70 in the stabilization of the EcR/USP heterodimer in vivo, we conclude that there may be a relation between the downregulation of the S. nonagrioides EcR and Hsc70 genes and the developmental abnormalities observed in the JHER silenced animals. The developmental deficiencies observed after the silencing of SnJHER indicate that they may be a possible consequence of the indirect downregulation of SnEcR and SnHsc70 genes, caused by the reduction of the SnJHER mRNA levels.

Conclusion
Our results showed that SnJHER presents important biological functions regulating the larval, pupal and adult development. With respect to the high diversity of insect COEs in either substrate specificity or developmental gene expression we speculate that SnJHER may possess distinct molecular functions than the conventional JHE gene. Further biochemical studies are needed, in terms of substrate specificity and enzyme's selectivity, in order to shed light on the functional role of this gene in the regulation of the corn stalk borer's development.

Materials and Methods
Insect rearing and staging of larvae. The insects were obtained from an established laboratory colony of S. nonagrioides, maintained at 25 6 1uC, 55 6 5% RH and reared on an artificial diet [20], under long day (LD) conditions (16:8, light:dark). Larvae which were reared under LD conditions completed their larval stage in 6 instars. The age of analyzed larvae within each instar was measured in days after the preceding ecdysis, in respect to physiological markers such as body mass and head capsule width. The nomenclature of stages follows the pattern of designation of the instar followed by the day of the stadium (e.g. L5d2 denotes larvae of the 5 th instar, 2 days after ecdysis). Larvae were checked daily for molting. The age of the analyzed larvae within each instar was measured in days after the preceding ecdysis and in respect to physiological markers such as body mass and head capsule width. To obtain synchronously growing animals, newly molted larvae were removed from the colony everyday during the 6 th -8 th hour of photophase. The selected larvae had mean weight and mean head capsule width as follows: 101.3 mg and 1.74 mm (L5d0); 160.4 mg and 2.32 mm (L6d0). In the 9 th day of the last instar, larvae transform into prepupae (L6d9) and begin all the necessary physiological and morphological changes in order the metamorphosis to occur.
Insect cell growth and maintenance. Bombyx mori Bm5 cells [40] were grown in IPL-41 insect cell culture medium, supplemented with 10% fetal bovine serum (Life Technologies), were maintained at 28uC and subcultured weekly.
RNA isolation and cDNA synthesis. Total RNA was isolated from larvae and insect cells using TRIzolH reagent (Sigma) according to the supplier's instructions and stored at 280uC. The isolated RNA was treated with the RNase-free DNAse I (Promega) and 1.5 mg of it, was used as template in first strand cDNA synthesis. The cDNAs were synthesized by priming with the universal poly-thymine primer Oligodt (Table 3), using as reverse transcription enzyme, the Superscript TM II RNase H-Reverse Transcriptase (Invitrogen). In all experiments the RNA was extracted from the whole body tissue of the analyzed animals.
Bright field and UV field microscopy. All fluorescence observations were conducted directly on living cells or tissues using a Zeiss Axiovert 25 inverted microscope equipped with a HBO 50 illuminator for incident-light fluorescence excitation and a Zeiss filter set 09 (450-490 nm excitation filter, 510 nm barrier filter).
Quantitative and Semiquantitative RT-PCR analysis For semiquantitative and quantitative RT-PCR analysis of SnJHER (GenBank: EU178813) and semiquantitative RT-PCR analysis of SnEcR (GenBank: JN572102), SnHsc70 (GenBank: DQ004584), SnHsp70 (GenBank: EU430480), GFP and construct specific mRNA levels we used the primer sets, 39F/39R, ECRF/ ECRR, Hsc70F/Hsc70R, Hsp70F/Hsp70R, GFPF/GFPR and Wf/39F respectively ( Table 3). As control, part of the coding region of S. nonagrioides b-tubulin gene (GenBank: DQ147771) was amplified by using the primer set TubF/TubR, ( Table 3). The RT-PCR products were separated on 1.5% agarose gels. Incorporation of the fluorescent dye SYBR Green Brilliant (Stratagene) into double-stranded PCR products was used to determine the mRNA copy number of SnJHER. Standard plasmids were constructed by inserting a fragment from the coding region of SnJHER (using the primer set 39F/39R, Table 3) or Sesamia nonagrioides b-tubulin (using the primer set TubF/TubR, Table 3) into pGEM T-easy vector (Promega). These plasmids were used as template DNA to produce standard curves. Each sample was analyzed in technical triplicates and the means were calculated. The quantity of mRNA levels was normalized with those of b-tubulin.
Hemolymph Administration of dsJHER dsRNA quantity/ Control treatments. For all experiments we used 4 mg of the in vitro synthesized dsRNAs. For control injections we selected an in vitro synthesized dsRNA produced by the multiple cloning site of L4440 vector (Addgene), flanked by the T7 promoter sequences. To our knowledge, GFP-based or other ''neutral'' constructs or just ddH 2 O, which we had previously used as controls in RNAi experiments, resulted in the same effects in terms of semiquantitative/quantitative RT-PCR or phenotypic analysis when compared with these of the L4440's multiple cloning site (Table S9 in File S1). Previous studies were also underlining that [25]. The probe for RNA synthesis was isolated from L4440's multiple cloning site of ,250 bp, by amplifying with the universal T7 primer ( Table 3). The amplified fragment flanked by the T7 promoter sequences was used as a template for dsRNA synthesis; T7 RNA polymerase (Fermentas) was allowed to RUN off overnight at 37uC. DNA was removed by DNase treatment (Promega). The dsRNA was then phenol/chloroform extracted, alcohol precipitated overnight and quantified.
Targeting the 472, 1276 and 1725 bp part. PCR was performed using cDNA isolated from S. nonagrioides larvae fat tissue, by priming with the Wf/59R and Wf/39R primer sets, which amplify 472 and 1725 bp respectively, ( Table 3). The PCR products were gel extracted and T/A cloning was performed in pGEM T-easy vector (Promega). For pGEM Teasy/JHER 472 two clones were selected one with SP6T7 and the other with T7SP6 orientation (Fig. S1A). Both clones (sense and antisense) were linearized with SalI (New England Biolabs) and used as templates for RNA synthesis. The 1725 bp fragment was excised from pGEM T-easy/JHER 1725 with EcoRI (New England Biolabs) and ligated in the EcoRI position of pBIISK-vector (Agilent Technologies). The clone with T3RT7 orientation (pBIISK-/SnJHER 1725T3T7 ) was double digested with XhoI/NcoI (New England Biolabs) and the resulting 1276 bp fragment was force ligated into the L4440 vector (L4440/ SnJHER 1276 , Fig. S1B). The L4440/ SnJHER 1276 plasmid was then linearized with either XhoI or NcoI (New England Biolabs) and used as template for RNA synthesis.
For dsJHER 1725 synthesis the pBIISK-/ SnJHER 1725T3T7 (Fig. S1C) was linearized with either XhoI (New England Biolabs) or XbaI (New England Biolabs). RNA synthesis was performed with T7 or T3 RNA polymerase (Fermentas). Sense and antisense RNA strands were quantified and equal amounts of RNA were mixed and annealed at boiling water for 10 minutes. Hybridization was performed by gradient cooling the boiled mix overnight. Plasmid DNA removed by DNase treatment. The dsRNA was phenol/chloroform extracted, alcohol precipitated overnight and quantified.

Bacterial Administration of dsJHER
Control treatments. For control treatments, we transformed bacteria with the empty L4440 vector. This vector produces dsRNA molecules of ,250 bp from its multiple cloning site which is surrounded by the T7 promoter sequences. The empty L4440 vector resulted in the same effects, in terms of semiquantitative/ quantitative RT-PCR or phenotypic analysis, when compared with these of GFP-based constructs (Table S1 in File S1). Previous studies were also underlining that [22]. Targeting the 472 bp part. PCR was performed using cDNA isolated from S. nonagrioides larval fat tissue, by priming with the Wf/59R and Wf/39R primer sets. The PCR products were gel extracted and T/A cloning was performed in pGEM T-easy vector. The 1725 bp fragment was excised from pGEM T-easy/JHER 1725 with EcoRI and ligated in the EcoRI position of pBIISK-vector. The clone with T3RT7 orientation was named as pBIISK-/SnJHERa (Fig. S2A). Furthermore, the pGEM T-easy/JHER 472 with T7RSP6 orientation was named as pGEM T-easy/SnJHERs (Fig. S2B). Both pBIISK-/ SnJHERa and pGEM T-easy/SnJHERs plasmids were double digested with SalI/SacI (New England Biolabs). The SalI/SacI fragment excised from pBIISK-/SnJHERa and was re-cloned to the SalI/SacI digested pGEM T-easy/SnJHERs plasmid. Positive clones were selected after digestion with EcoRI and NotI RE. The resulting plasmid was named as pGEM T-easy/ SnJHER loop (Fig. S2C).
HT115 (DE3) competent cells lacking RNase III were prepared using standard CaCl 2 methodology and were transformed with the pGEM T-easy/SnJHER loop plasmid DNA (Fig. S2C). Single colonies of HT115/pGEM T-easy/SnJHER loop cells were cultured in LB at 37uC with shaking at 220 rpm overnight. The culture was diluted 50-fold in 100 ml LB supplemented with 100 mg/ml ampicillin (Sigma) plus 15 mg/ml tetracycline (Sigma) and cultured at 37uC to OD 600 = 0.5. Synthesis of T7 polymerase was induced with 0.4 mM IPTG and the bacteria were incubated with shaking for an additional 4 h at 37uC. For feeding experiments, bacteria were centrifuged at 5,000 g for 10 min and resuspended in 0.5 ml of ddH 2 O. In order to check efficient dsRNA expression, total RNAs from the bacterial cells were isolated. Total RNAs from bacterial cells were extracted using TRIzolH reagent (Sigma) according to the supplier's instructions. The RNA pellets were dissolved in 20 ml of ddH 2 O. In order to remove ssRNAs from the RNA samples 1 ml of 1 mg/ml of RNase-A (Ribonuclease-A from bovine pancreas, Sigma) and NaCl to 0.3 M was added (RNase-A in high salinity buffers selectively digests ssRNAs leaving undigested the dsRNAs, [41]). The reaction occurred for 10 minutes at 37uC. The length and the quality of the produced dsRNAs were confirmed by electrophoresis on 1 % agarose gel ( Fig. 2A).
After the confirmation of the dsRNA synthesis, feeding bioassays were followed. 100 ml of the IPTG induced cultures was centrifuged and pellets resuspended in 0.5 ml of ddH 2 O. The artificial diet was cut into different sizes of pellets depending on the instar and the number of feeding larvae. For each 100 S. nonagrioides neonates or 1 st or 2 nd instar larvae, a 10610610 mm 3 pellet was used on which 100 ml of fresh IPTG induced bacteria were applied every 12 hours. For each 100 3 rd or 4 th instar larvae, 200 ml of fresh IPTG induced bacteria were applied on a 20620620 mm 3 pellet, while for each 50 5 th or 6 th instar larvae 300 ml of fresh IPTG induced bacteria were applied every 12 hours on a 30630630 mm 3 pellet. The pellets were replaced every 2 days, depending on the remaining undigested material. As control we used IPTG induced HT115 bacteria transformed with the empty L4440 vector.

Baculovirus-mediated dsRNA Administration
Control treatments. As control we used the BmNPV/ BmA::GFP-BmA::dsLuciferase virus, a virus expressing double stranded molecules of the reference luciferase gene.
dsJHER loop construction. The pGEM T-easy/SnJHER loop plasmid (Fig. S2C) was partially digested by incubating 1 mg of it with 1/10 U of the NotI restriction enzyme (New England Biolabs) for 5 minutes at 37uC. The pFastBac Actin-BGH transfer plasmid (Fig. S2D) was digested with NotI and after dephosphorylation (0.5 U of Alkaline Phosphatase in 50 ml of restriction reaction for 30 min) was ligated with the NotI digested SnJHER loop construct (Fig. S2C). The recombinant plasmid pFastBac Actin-BGH /SnJHER loop , was transformed into competent DH10Bac/BmNPV-BmA::GFP cells. Transformed bacteria were selected in LB plates containing 50 mg/ ml kanamycin, 7 mg/ml gentamicin, 10 mg/ml tetracycline, 100 mg/ml X-a-gal and 40 mg/ml IPTG (Sigma) after O/N incubation at 37uC. 7 single colonies were picked up and grown in liquid LB with 50 mg/ml kanamycin and 7 mg/ml gentamicin. After O/N incubation, bacmid DNA was extracted and each colony was analyzed in PCR reactions using the Wf primer (Table 3) which amplifies approximately 2.300 bp of the construct. All bacmids were SnJHER loop positive. Bacmids 1, 3, 6 and 7 were used for transfection of Bm5 cells with Escort IV transfection Reagent (Sigma). 7 th day post transfection the cells were checked for GFP. Few cells were GFP positive 7 th day post transfection (Fig. S4). The supernatants were collected and stored as viral stocks (viral stock 1). 20 ml of each viral stock 1 of BmNPV-BmA::GFP/ BmA::SnJHER loop 1, 3 and 7 were used for infection of Bm5 cells. 7 days post infection the supernatants were collected and stored (viral stock 2). The infected cells were observed for GFP expression. The previous step was repeated for one more time (Fig. S4) and the viral stock 3 was collected. In order to ensure positive transposition of the SnJHER loop construct, PCR was performed either in DNA or cDNA of infected Bm5 cells. The DNA or the cDNA was primed-off with the Wf/39F primer set (Table 3) which amplifies approximately 750 bp of the construct. All viruses were SnJHER loop positive (data not shown). For in vivo assays we selected to use the BmNPV-BmA::GFP/ BmA::SnJHER loop 7 virus (Fig. S4). The viral stock 2 of this virus and a viral stock of the BmNPV-BmA::GFP/ BmA::dsLuciferase virus were used for titration, in order to proceed to the in vivo assays. Both viruses were measured to have a titer of approximately 10 7 pfu/ ml.
Biological assays. 50 ml of each virus were used for infections. Two different developmental stages were selected in order to perform the infections, the 5 th instar d3 (larval stage) and the 6 th instar d9 larvae (prepupal stage). Following infections, the insects were allowed to complete their development recording daily the potential developmental abnormalities and phenotypic effects. For RT-PCR analyses we sampled insects 7 days post infection, the day in which we observed the maximum infectivity of the BmNPV virus in terms of GFP expression (Fig. S3). File S1 Text S1. Virus strain selection/ Virus infectivity and localization; Table S1. RNAi efficiency, after hemolymph administration in L5d3 larvae; Table S2. Hemolymph administration of dsJHER 1725 in L5d3 larvae; Table S3. Hemolymph administration of dsJHER 472 in L6d9 larvae; Table S4. Hemolymph administration of dsJHER 1276 in L6d9 larvae; Table S5. Hemolymph administration of dsJHER 1725 in L6d9 larvae; Table  S6. Baculovirus-mediated administration of dsJHER 472 in L5d3 larvae; Table S7. Baculovirus-mediated administration of dsJHER 472 in L6d9 larvae; Table S8. Baculovirus-mediated administration of dsJHER 472 in L5d3 and L6d9 larvae; Table  S9. Selecting the appropriate control treatment for hemolymph dsRNA administration. (DOC)