Juvenile Hormone receptor Met is essential for ovarian maturation in the Desert Locust, Schistocerca gregaria

Juvenile hormones (JH) are key endocrine regulators produced by the corpora allata (CA) of insects. Together with ecdysteroids, as well as nutritional cues, JH coordinates different aspects of insect postembryonic development and reproduction. The function of the recently characterized JH receptor, Methoprene-tolerant (Met), appears to be conserved in different processes regulated by JH. However, its functional interactions with other hormonal signalling pathways seem highly dependent on the feeding habits and on the developmental and reproductive strategies employed by the insect species investigated. Here we report on the effects of RNA interference (RNAi) mediated SgMet knockdown during the first gonadotrophic cycle in female desert locusts (Schistocerca gregaria). This voracious, phytophagous pest species can form migrating swarms that devastate field crops and harvests in several of the world’s poorest countries. A better knowledge of the JH signalling pathway may contribute to the development of novel, more target-specific insecticides to combat this very harmful swarming pest. Using RNAi, we show that the JH receptor Met is essential for ovarian maturation, vitellogenesis and associated ecdysteroid biosynthesis in adult female S. gregaria. Interestingly, knockdown of SgMet also resulted in a significant decrease of insulin-related peptide (SgIRP) and increase of neuroparsin (SgNP) 3 and 4 transcript levels in the fat body, illustrating the existence of an intricate regulatory interplay between different hormonal factors. In addition, SgMet knockdown in females resulted in delayed display of copulation behaviour with virgin males, when compared with dsGFP injected control animals. Moreover, we observed an incapacity of adult dsSgMet injected female locusts to oviposit during the time of the experimental setup. As such, SgMet is an essential gene playing crucial roles in the endocrine communication necessary for successful reproduction of the desert locust.


Results and Discussion
Description of the relative transcript levels of the genes of interest. SgMet (accession number: MK855050) shows a wide tissue distribution profile. Its expression in 10-day-old adult locusts is mainly observed in the male accessory glands (AGs), CA, fat body, prothoracic glands (PGs) and (flight) muscles ( Supplementary  Fig. S1). SgMet transcript levels remain relatively stable in the fat body of adult female locusts throughout the first gonadotrophic cycle (Fig. 1A), a situation which is similar to what was reported on the expression of Met during the reproductive cycle of the pacific beetle cockroach, Diploptera punctata and the German cockroach, B. germanica 28,33 . Also the relative Met mRNA levels measured in the complete body of the cotton bollworm, Helicoverpa armigera, and the red flour beetle, Tribolium castaneum, remain stable during adult female development 36,38 . On the other hand, the temporal profile of the relative mRNA levels of the downstream transcription factor SgKr-h1 correlate well with the transcript levels of two Vg genes (Fig. 1A). Since SgKr-h1 is a transcription factor acting directly downstream of the JH receptor Met, SgKr-h1 transcript levels can be considered as a measure for Met-mediated JH signalling activity, as was previously described for B. germanica, D. melanogaster and T. castaneum [54][55][56] . The SgVg developmental profile shows a clear increasing trend which starts on day 6, when the basal oocytes enter their vitellogenic stage of maturation, and continues to increase until day 16, when these oocytes almost reach their maximal length. The population of animals from which these samples were taken, completed their first gonadotrophic cycle with oviposition around day 18 (Fig. 1C).
to eliminate the possibility of off-target effects, two separate dsSgMet (SgMet dsRNA) constructs were designed and tested in the same experimental setup, which resulted in similar phenotypic observations.
The knockdown efficiency of the systemic RNAi of SgMet was assessed in different tissues on day 12 after adult eclosion ( Supplementary Fig. S2). A significant downregulation was measured in the CA/CC complex (61%) and fat body (51%). However, SgMet levels were not downregulated in the ovaries. This can be explained by the previously described inefficient uptake of dsRNA into the follicle cells and oocytes of the desert locust 57,58 . In both the CA/CC complexes and the fat body, relative mRNA levels of SgKr-h1 were shown to be significantly reduced with 88% and 73% respectively, upon silencing SgMet ( Fig. 2A,B). Relative SgKr-h1 mRNA levels were not affected in the ovaries as SgMet was not effectively silenced in this tissue (Fig. 2C). These findings confirm the transcription factor SgKr-h1 to be situated downstream of SgMet.
The effect of silencing SgMet was also studied in the CA, the site of JH biosynthesis. Previous work already characterized the genes encoding the enzymes catalysing the penultimate and ultimate steps in the JH biosynthetic pathway of the desert locust 59 . Juvenile hormone acid methyl transferase (JHAMT) methylates farnesoic acid to methylfarnesoate (MF) and is known to act as the rate-limiting enzyme in JH biosynthesis [59][60][61][62] . The ultimate step in the pathway is catalysed by the cytochrome P450 enzyme CYP15A1 which epoxidizes MF to the active hormone, JHIII. Our current results show that silencing the JH receptor causes upregulation of the transcripts that code for the JH biosynthetic enzymes, SgJHAMT and SgCYP15A1 (Fig. 2D), by 550% and 370%, respectively. Upon dissection of the CA, it also became apparent that silencing SgMet resulted in larger sized CA, again suggesting an increase in activity of these JH producing glands ( Supplementary Fig. S3). Our data are in accordance with a recent study in the linden bug, P. apterus, where a rise in the circulating JH levels was observed upon silencing Met 24 . It can thus be hypothesized that in the CA SgMet acts as a JH sensor involved in a negative feedback loop that keeps JH biosynthesis and CA activity under control, a homeostatic mechanism similar to what is often observed in endocrine gland regulation 63 .

sgMet is crucial for initiating ovarian maturation, vitellogenesis and reproductive behaviour.
It should be noted that the locust reproductive cycle is strongly determined by environmental factors, which can cause variation in the exact timing. The first gonadotrophic cycle is therefore usually divided in different stages instead of exact time periods: (i) an early growth period during which the oocytes only reach a length of 0.8-0.9 mm; (ii) the previtellogenic stage, during which oocytes grow until they reach a length of 1.8 mm; (iii) the vitellogenic stage starting from 1.8 mm, characterized by patency, the shrinkage of the follicular epithelial cells, allowing passage of lipids and vitellogenin from the haemolymph into the developing oocytes, causing oocytes to grow until they reach about 7.5 mm; (iv) egg shell formation during the choriogenic stage and (v) ovulation and oviposition 64,65 . It is clear from Fig. 3J that the SgMet knockdown animals never enter the (pre)vitellogenic stage as the oocytes do not appear to have increased in length after 12 days in the adult stage when compared to newly moulted adult females. As such, it became evident that ovarian maturation was arrested by silencing SgMet (Fig. 3A-F). The ovaries of 12-day-old dsSgMet injected locusts remained arrested in a very immature, previtellogenic stage (Fig. 3A,B), still resembling the ovaries of freshly moulted control locusts (D0) (Fig. 3E,F). To further verify whether or not yolk materials were incorporated, transverse sections of oocytes were made (Fig. 3G,H). In oocytes of dsGFP injected control locusts, Vgs (blue matrix) as well as lipid droplets (greyish droplets) were visible. The accumulation of Vgs and lipid droplets was not observed in the SgMet knockdown animals. Moreover, when studying these sections, no patency of the follicular cell layer was observed even though higher JH levels www.nature.com/scientificreports www.nature.com/scientificreports/ might circulate in the haemolymph (due to the higher relative mRNA levels of JHAMT and CYP15A1; Fig. 2D). Our observations are similar to the lack of patency observed in both L. migratoria and D. punctata treated with dsMet which could not be rescued by treatment with methoprene 20,33 . Patency was long thought to be initiated by JH acting on membrane receptors in the ovary [6][7][8][9][10] . The recent report by Seidelmann et al. 11 suggests that PIF, an as yet unidentified oviduct derived factor, is responsible for the initiation of patency in locusts and that its action is boosted by JH 11 . As no Vg was observable in oocytes of locusts where SgMet was knocked down, we analysed SgVg mRNA levels in fat body samples from dsSgMet versus dsGFP injected locusts on day 12 of the first gonadotrophic cycle (Fig. 3I). Silencing SgMet very obviously resulted in much lower relative Vg mRNA levels for both SgVg1 and SgVg2, for which transcript levels during the first gonadotrophic cycle of wild type locusts are shown in Fig. 1C. In control knockdown animals, a large variation in basal oocyte lengths exists. Some of the control locusts were already in their second gonadotrophic cycle, characterised by orange lesions at the base of the new basal oocyte. In the SgMet knockdown locusts, it was evident that ovarian maturation was arrested. Oocytes longer than 1.5 mm were never observed, suggesting oocyte maturation could not be completed, even after 12 days in the adult stage (Fig. 3J). Moreover, dissection of the SgMet females at the end of the copulation experiment 22 days after adult eclosion revealed a continued arrest of oocyte development. It is apparent that JH-Met signalling is a common pathway regulating female vitellogenesis as evidenced by RNAi studies in different insect species, such as T. castaneum, Colaphellus bowringi, Cimex lectularius, B. germanica, D. punctata, Aedes aegypti, H. armigera, R. prolixus, Reticulitermes speratus, P. apterus and L. migratoria 24,25,[27][28][29][30]32,33,36,38,40,66,67 .
Met knockdown in female bed bugs (C. lectularius) also caused a reduced number of eggs deposited 27 . In our study, an effect on copulation behaviour as well as on oviposition was witnessed. Where control females all copulated on day 15, a delay was seen in the dsSgMet injected individuals (Fig. 4A). Moreover, none of these experimental locusts was capable of depositing their eggs within the set experimental timeframe (Fig. 4B), likely the result of incomplete oocyte maturation (Fig. 3J) and no increase in survival was observed in any of the knockdown experiments.
SgMet is also crucial for ovarian ecdysteroid biosynthesis. In vitellogenic female locusts, the developing basal oocytes also incorporate 'free' and conjugated ecdysteroids, while a rise in the circulating ecdysteroid levels can be observed in the haemolymph, as described previously 68,69 . The majority of the ecdysteroids produced by the follicle cell layer are incorporated in the growing oocytes as conjugates. Only a small fraction is stored in the oocytes as 'free' ecdysteroids. An enzyme immunoassay (EIA) was used to determine the effect of SgMet knockdown on ecdysteroid levels in the developing oocytes and the haemolymph. Silencing SgMet resulted in a significantly reduced level of both circulating and ovarian ecdysteroids (expressed in 20E equivalents) compared The data (log transformed) are shown as box plots (min to max) of four independent pools of five animals, run in duplicate and normalized to β-actin and EF1α transcript levels. Statistically significant differences (p) between the measurements were found, after a log transformation, via a t-test (with or without two-sided Welch's correction) and are indicated by asterisks (**p < 0.01; ***p < 0.001****p < 0.0001). (2019) 9:10797 | https://doi.org/10.1038/s41598-019-47253-x www.nature.com/scientificreports www.nature.com/scientificreports/ to the control locusts ( Fig. 5A-C). Ecdysteroid biosynthesis in adult locusts mainly takes place in the follicle cell layer of the developing oocytes 70,71 where the steroid hormone precursor cholesterol is converted to 20E in a series of enzymatic reactions involving the Halloween genes, first characterized in D. melanogaster 72 . Three of these Halloween genes, i.e. Spook (Spo), Phantom (Phm) and Shade (Shd), encoding cytochrome P450 enzymes were already functionally characterized in S. gregaria 68,73 . SgSpo, encoding an enzyme acting in the Black Box of the ecdysteroid biosynthetic pathway, is considered to catalyse the rate-limiting step. SgShd is important in the The data (log transformed) are shown as box plots (min to max) of four independent pools of five animals, run in duplicate and normalized to β-actin and EF1α transcript levels. Statistically significant differences (p) were found, after a log transformation, via a t-test (with or without two-sided Welch's correction) and are indicated by asterisks (**p < 0.01). (J) The oocyte length (mm) of 12-day-old control and dsSgMet treated female locusts was observed. Each point represents the mean length of 3 randomly chosen oocytes (mean ± S.E.M). Statistically significant differences (p) were found via a nonparametric Mann-Whitney test and are indicated by asterisks (****p < 0.0001). The cumulative percentage of copulating females (actual connection between male and female genitalia) and (B) the cumulative percentage of eggs deposited by the females are showed. Statistically significant differences (p) between the two conditions were found via a log-rank (Mantel-Cox) test and are indicated by asterisks (*p< 0.05; ****p< 0.0001).
www.nature.com/scientificreports www.nature.com/scientificreports/ conversion of ecdysone to the active hormone, 20E. In adult locusts, relative mRNA levels of all three are highest in the ovaries and degenerating PGs. The reduced level of circulating and ovarian ecdysteroids in the dsSgMet treated animals is also evidenced in the ovarian transcript levels of both SgSpo and SgPhm (Fig. 5D). Our data indicate that the initiation of ecdysteroid biosynthesis by the reproductive organs is situated downstream of the JH-mediated initiation of oocyte growth and vitellogenesis. The start of ecdysteroidogenesis in locusts may be directed by a nutritional state-dependent factor binding to the Venus Kinase Receptor (VKR), a receptor tyrosine kinase as evidenced by a recent study by Lenaerts et al. 74 . Knockdown of SgVKR in adult females results in reduced oocyte size, ovarian and circulating ecdysteroid levels. Relative transcript levels of several Halloween genes were negatively affected 74 . Downregulation of SgMet resulted in a significant reduction of relative SgVKR transcript levels in the ovary (Supplementary Fig. S4A). In the mosquito, A. aegypti, the ligand for this receptor is the Ovarian Ecdysteroidogenic Hormone (OEH), an 83 AA long peptide released from medial neurosecretory cells in the mosquito brain upon blood feeding which induces the ovaries to produce ecdysteroids. These ecdysteroids, as in other higher insects, will stimulate the fat body to produce yolk protein 75,76 . Interestingly, OEH is considered as a member of the neuroparsin family, which possesses a typical, conserved pattern of Cys residues 48 . Most insects seem to encode a single NP gene, while in some species, such as D. melanogaster, no NP has been discovered 45 . However, in S. gregaria four SgNP precursor transcripts are found, for which transcript profiling studies were performed in the past 77,78 . In gregarious locusts, transcripts for SgNP1 and SgNP2 are predominantly detected in the brain; whereas SgNP3 and SgNP4 are also detected in female fat body and ovaries. As previous research from our lab and current data in Supplementary Fig. S5A,B show, transcripts of SgNP3 and SgNP4 are highly regulated during the first gonadotrophic cycle, suggesting these peptides play a role in reproductive physiology. A decline of both SgNP3 and SgNP4 transcripts was observed approximately at the time when SgVg transcript levels started rising in the fat body. In the dsMet treated animals the relative mRNA levels of both SgNP3 and SgNP4 are higher compared to the control animals in fat body, while the relative transcript level of SgNP4 is also increased in the ovary (see Fig. 6 and Supplementary Fig. S4B). At present, it is still unclear whether the changes in SgNP3 and SgNP4 expression are caused by changes in JH signalling or to indirect effects via other regulatory pathways that have been affected in the dsMet treated locusts. JH injection into the haemocoel of day 5 adult female locusts results in an increase of SgNP3 and SgNP4 transcript levels in both female fat body and ovary 79 . In our current study, we observe that SgMet expression in the ovary is not significantly influenced by the systemic knockdown, which could be due to weaker RNAi sensitivity 57 and/or stronger compensatory effects in this tissue. In the CA, we observe another kind of compensatory effect resulting in significantly increased www.nature.com/scientificreports www.nature.com/scientificreports/ expression of JH biosynthetic enzymes. So, it could be that these compensatory effects may also have resulted in increased ovarian SgNP4 levels. However, this does not explain why both SgNP3 and SgNP4 transcript levels are strongly increased in the fat body, since in this tissue knockdown of SgMet and reduced JH signalling are clearly evidenced by significantly reduced transcript levels of SgMet and SgKrh1, respectively. Another receptor that may be involved in causing the observed changes in SgNP transcript levels is SgVKR. In the current study, we show that SgVKR transcript levels are significantly reduced in dsMet treated desert locusts. The observed effect on SgNP transcript levels might be due to these reduced SgVKR levels, since Lenaerts et al. 74 have shown that knockdown of SgVKR resulted in increased SgNP3 and SgNP4 levels in locusts. Nevertheless, the occurrence of other currently unknown mechanisms cannot be ruled out.

Knockdown of JH receptor Met alters expression of insulin-related peptide and neuroparsin precursor genes in an opposite way. Previous research has shown that IRP has a gonadotropic role,
while NPs have an anti-gonadotropic role in the female reproductive physiology of locusts. Studies in S. gregaria revealed that an increase in SgIRP transcript levels can be observed during specific reproductive events such as vitellogenesis and intensive growth of the oocytes 80 . The effect of this peptide was later evidenced in an RNAi-mediated study in S. gregaria by Badisco et al. 47 showing lower SgVg transcript levels and smaller oocytes compared to control animals upon dsIRP injections 47 . Playing a vital role as nutritional sensor, the ISP has been described in different insect species as having a gonadotropic effect. In dipterans, insulin-like peptides (ILPs) are involved in stimulating egg production and affecting germ line cyst development rate as well as progression through vitellogenesis [81][82][83] . In the yellow fever mosquito, A. aegypti, ILP3 stimulates yolk uptake by the developing oocytes 81 . Moreover, insulin signalling is also required to induce vitellogenin production in the fat body 84 and to stimulate ecdysteroid production in the ovaries 85 . Furthermore, D. melanogaster female ISP mutants generally show reduced vitellogenesis 86 . In addition, mutations in the insulin receptor result in a decrease of JH synthesis suggesting the involvement of the ISP in the regulation of JH biosynthesis 87,88 . In the red flour beetle, T. castaneum, it was shown that JH is able to regulate ILP synthesis, which in turn regulates Vg synthesis 37 . In the adult female beetles, both JH and feeding are crucial in the onset of ILP production in the fat body and brain after which ILPs will trigger the Vg synthesis through removal of a Vg transcriptional repressor 36,37,89 . RNAi of the insulin receptor in B. germanica adult females resulted in a reduction in JH biosynthesis, Vg expression and ovarian development 90 . On the other hand, silencing of SgNPs in the RNAi-mediated study by Badisco et al. 47 resulted in higher SgVg transcript levels and bigger oocytes in 12-day-old female locusts 47 . This anti-gonadotropic action of NPs has also been proven in L. migratoria 80 .
Our current results show that during the locust's first gonadotrophic cycle, a decrease in relative SgNP3 and SgNP4 transcript levels coincides with an increase in relative SgIRP expression in the locust fat body (Supplementary Fig. S5). Also in the dsSgMet treated females, a significant increase in SgNP3 and SgNP4 expression was measured in the fat body. SgIRP expression on the other hand decreased (Fig. 6). This clearly indicates a consistent cross-talk between SgNP and SgIRP signalling downstream of JH.
All the above mentioned effects observed upon SgMet silencing are summarized in Fig. 7.

Conclusions
Our results show that the knockdown of SgMet in female locusts results in a failure to initiate vitellogenin expression in the fat body as well as oocyte growth in the ovaries. These dsSgMet treated animals are still able to copulate, albeit later, but they do not oviposit. As such, strategies ensuring a specific knockdown or inhibition of components of the JH signalling pathway, or of functionally linked pathways, may prove to be excellent methods for specifically targeting locust reproduction and associated swarm formation. Our study also indicates that upon SgMet knockdown some SgNP transcripts are significantly upregulated in the locust fat body, while the SgIRP levels are downregulated. In fact, these observations appear to be very well in line with their previously reported opposing roles in the control of reproduction. Furthermore, we show that SgMet knockdown strongly influences the expression of JH biosynthesis enzymes in the CA, as well as of ecdysteroid biosynthesis enzymes in the ovary, Figure 6. JH receptor knockdown alters the levels of neuroparsin and insulin-related peptide precursor transcripts. Relative SgNP3, SgNP4 and SgIRP transcript levels were measured in the fat body of 12-day-old dsSgMet injected female locusts. The data (log transformed) are shown as box plots (min to max) of four independent pools of five animals, run in duplicate and normalized to β-actin and EF1α transcript levels. Statistically significant differences (p) between the measurements were found, after a log transformation, via a t-test (with or without two-sided Welch's correction) and are indicated by asterisks (***p < 0.001; ****p < 0.0001).
www.nature.com/scientificreports www.nature.com/scientificreports/ as evidenced by lower ovarian concentrations of both free and conjugated ecdysteroids. In conclusion, our study sheds more light on the essential role of the JH receptor SgMet, as well as on the interdependencies between different endocrine pathways involved in the control of reproduction in female desert locusts.

Materials and Methods
Rearing of animals. The desert locusts, S. gregaria, were reared under crowded conditions as previously described by Lenaerts et al. 74 . Sand/turf pots, into which eggs were deposited by mated females, were set apart once a week in clean cages. tissue collection. The tissues of interest were dissected as previously described by Lenaerts et al. 74 . For the tissue and temporal expression profile of the genes of interest, tissues were collected in three pools of ten animals each. Tissues for the RNAi experiment were collected in four pools consisting of five animals each. The collected tissues were used in an RNA extraction protocol or stored at −80 °C until further processing. RNA extraction and cDNA synthesis. Fat body, ovaries, muscles and midgut of female adult locusts as well as the male reproductive system (testes + AGs) were pooled in MagNA Lyser Green Beads Tubes (Roche) and homogenized using a MagNA Lyser instrument (20 s, 5000 rpm; Roche). Subsequently, total RNA was extracted using the RNeasy Lipid Tissue Kit (Qiagen) according to the manufacturer's protocol. To prevent genomic DNA contamination, an on-column DNase digestion (RNase-free DNase set, Qiagen) was performed. Total RNA of the small sized tissues (CA/CC complex, PGs and suboesophageal ganglion (SOG)) was extracted using the RNAqueous-Micro Kit (Ambion) according to the manufacturer's protocol, including the optional DNase step. Quality and concentration of the resulting RNA samples was measured using a Nanodrop spectrophotometer (Nanodrop ND-1000, Thermo Fisher Scientific Inc.). For each RNA sample, equal amounts of RNA (1000 ng for brain, fat body, ovaries, muscles, midgut, testes and AGs and 200 ng for PGs, CA/CC complex and SOG) were transcribed in subsequent cDNA synthesis using a mix of random hexamers and oligo(dT) primers according to the manufacturer's protocol (PrimeScript TM RT Reagent Kit, TaKaRa, Invitrogen Life Technologies). The obtained cDNA was diluted ten-fold with Milli-Q water (Millipore).
Quantitative real-time pCR. Prior to quantitative real-time PCR (qRT-PCR) transcript profiling, optimal housekeeping genes were selected using geNorm software 91 . Therefore, several previously described housekeeping genes 92 were tested for their stability in the designed experiment and β-actin and EF1α appeared to be most stable in the studied samples. Primer Express software (Applied Biosystems) was used to design qRT-PCR primers for the different target genes (Supplementary Table S1). Subsequent primer validation and qRT-PCR reactions were performed as previously described by Lenaerts et al. 74 .
All qRT-PCR results were normalized to both the β-actin and EF1α transcript levels and calculated relative to the transcript level in a calibrator sample according to the comparative Ct method (ΔΔCt) 91 . qRT-PCR was www.nature.com/scientificreports www.nature.com/scientificreports/ used to determine the tissue and temporal distributions of the genes of interest in the adult stage. Therefore, cDNA samples were used from adult female locusts, except for the male AGs and testes. Moreover, statistically significant differences between the dsSgMet injected and control adult female locusts were found via a t-test on the log transformed data (with or without two-sided Welch's correction) using GraphPad Prism 6 (GraphPad Software Inc.). Correlation coefficients between temporal expression profiles of the genes of interest were found via a Pearson correlation calculation using GraphPad Prism 6 (GraphPad Software Inc.).

RNA interference experiments.
DsRNA constructs were prepared using Ambion's MEGAScript RNAi kit following the manufacturer's protocol and as previously described by Lenaerts et al. 74 . In short, a linear transcript with 5′ T7 promotor sequences was used in a high yield RNA transcription reaction. Linear transcripts for the targets of interest were PCR amplified using REDTaq DNA polymerase (Sigma), S. gregaria fat body cDNA and the primers given in Supplementary Table S2. The amplicons were sequenced to confirm target specificity and used in a high-yield RNA transcription reaction using T7 RNA polymerase. ssRNA and DNA were further removed in a nuclease digestion step. The resulting dsRNA was analysed on gel and using the Nanodrop spectrophotometer for integrity, quality and concentration.
Newly moulted adult female locusts, synchronized on the day of ecdysis, were injected with 8 µl dsRNA against SgMet (500 ng dsRNA/locust, diluted in S. gregaria ringer). A second and third boost injection was given on days four and eight after final moult to ensure a lasting knockdown of target gene mRNA levels. A second group of control locusts was injected with dsRNA against GFP following the same injection scheme. From previous observations of our S. gregaria colony, it is known that female adults are fully in the vitellogenic stage of oocyte maturation 12 days after their final moult. Therefore, all the tissues and samples of interest were dissected and collected on this day. In a follow-up experiment, the same injection scheme was followed and copulation behaviour and oviposition were observed. Starting on day 8, the female locusts were transferred to individual cages, supplied with oats and fed daily with fresh cabbage. On day 10 of the females' adult life, one sexually mature virgin male was introduced into the separate cages and mating behaviour and copulation (actual connection between male and female genitalia) was registered. If no copulation was observed within two hours of the male's introduction, the male was removed and another mating was attempted the following day with another virgin male. After mating, pots filled with a humid sand/turf mixture were supplied to the females to allow oviposition. These pots were checked daily for egg pods. Statistically significant differences between the two conditions were found via a log-rank (Mantel-Cox) test using GraphPad Prism 6 (GraphPad Software Inc.).

Measurement of oocyte length.
Terminal oocytes, i.e. oocytes at the base of the ovarioles, were carefully removed from the dissected ovaries and their length was measured using millimetre graph paper. Statistically significant differences between the dsSgMet injected and control adult female locusts were found via a nonparametric Mann-Whitney test using GraphPad Prism 6 (GraphPad Software Inc.).
Microscopy and histological analysis. Images of the dissected ovaries and ovarioles as well as of the oocyte sections of dsSgMet and dsGFP (GFP dsRNA) treated animals were made as previously described by Lenaerts et al. 74 . ecdysteroid measurements using an enzyme immunoassay. Female locusts were pierced behind the hind leg on the dorsal side allowing haemolymph to flow out. 10 µl of haemolymph was collected from each adult female using a glass capillary tube, immediately transferred to 90 µl of 100% ethanol and stored at −20 °C until further processing as described by Marchal et al. 68 .
One ovarium of each female locust was dissected and transferred to 1 ml of 100% ethanol after which it was stored at −20 °C until further processing. These ovarium samples were homogenized using a bar sonicator and subsequently for 10 min at 60 °C. After centrifugation at 10 000 g (10 min), the supernatants were transferred to a fresh Eppendorf tube and the remaining pellets were re-extracted twice by adding 1 ml of 100% ethanol. Combined extracts were dried in a SpeedVac concentrator and redissolved in 1 ml of 70% ethanol and 1 ml of 100% hexane in order to remove the apolar lipids and as such obtain optimal EIA measurements. Both 'free' and total ecdysteroid levels were determined where the 'free' ecdysteroids are non-conjugated. The total ecdysteroid levels were determined after enzymatic conversion of the conjugated ecdysteroids into 'free' ecdysteroids.
Ecdysteroid titers in S. gregaria haemolymph and ovaries were measured using an EIA, modified from Porcheron et al. 93 and discussed by Pascual et al. 94 and Lafont et al. [93][94][95] . In brief a peroxidase conjugate of 20E is used as tracer together with rabbit L2 polyclonal antibodies against ecdysteroids. Both serum and tracer were a kind gift by Prof. J.P. Delbecque (Université de Bordeaux, France).
The resulting absorbance values were compared to the calibration curve obtained with 20E as a standard. Since the data were normally distributed upon log transformation, a t-test (with or without two-sided Welch's correction) was performed using GraphPad Prism 6 (GraphPad Software Inc.) to test for statistically significant differences between the data.