An insulin-like growth factor-like peptide promotes ovarian development in the silkmoth Bombyx mori

Insulin family peptides are known to be key regulators of growth and metabolism in insects and vertebrates. Insects have two types of insulin family peptides: insulin-like peptides and insulin-like growth factor (IGF)-like peptides (IGFLPs). We recently demonstrated that an IGFLP in the silkmoth, Bombyx mori (BIGFLP) promotes the growth of the genital imaginal disc ex vivo. However, the role of BIGFLP in the regulation of insect growth remains unclear because no in vivo study has been performed. Therefore, we analysed the functions of BIGFLP in vivo by constructing BIGFLP knock-out (KO) B. mori using the clustered regularly interspaced palindromic repeats (CRISPR) and CRISPR-associated protein 9 (CRISPR-Cas9) system. The KO moths exhibited decreased body weights and size of the appendages compared wild-type (wt) moths. Interestingly, KO females also had drastically lower ovary weights and number of eggs than wt females. However, mutant ovaries that were transplanted into wt host pupae reached a similar weight to wt ovaries that were transplanted into the wt hosts, suggesting that IGFLP in the haemolymph promotes ovarian development. These findings show that BIGFLP regulates the growth and development of adult organs, particularly the ovaries, in B. mori.

conserved in ILPs within its amino acid sequence and IPCs must have proprotein convertases as evidenced by the production of bombyxins. Ex vivo studies in B. mori have shown that purified BIGFLP promotes the growth of adult organs, such as the male genital disc, via the IIS pathway 13,17 . In D. melanogaster, it has been shown that Dilp6 knock-out (KO) flies exhibited a decreased body mass and a smaller number of wing cells than wild-type (wt) flies, suggesting that Dilp6 promotes organ growth and systemic growth in flies 14,15 . Furthermore, the inhibition of Dilp6 expression after pupation by RNA interference reduces the consumption of triacylglycerol and glycogen 15 . Thus, it appears that Dilp6 regulates the metabolism and growth of adult organs during development.
So far, in vivo analyses of the physiological functions of ILPs and IGFLPs have been performed mainly in D. melanogaster using genetic approaches, whereas functional analyses of these hormones in B. mori and other Lepidoptera have been limited to those ex vivo until very recently, due to the difficulties in applying genetic approaches. However, it is evident that both in vivo and ex vivo studies are necessary to establish the functions of hormones and to reveal molecular and cellular mechanisms of hormone actions. Furthermore, it is important to study a variety of insects to reveal general and species-specific features of hormone functions.
An innovative approach to analyse B. mori ILP (bombyxin) functions in vivo has recently been reported in which secretion of bombyxins from IPCs was inhibited by using an improved Gal4-UAS system and tetanus toxin that blocks synaptic transmission 18 . This study revealed that genetic silencing of the IPCs resulted in a decrease in the growth rate of B. mori larvae, confirming the involvement of ILPs in the regulation of larval growth.
In the present study, we investigated the functions of BIGFLP in vivo by constructing BIGFLP KO B. mori using the clustered regularly interspaced short palindromic repeat (CRISPR) and CRISPR-associated protein 9 (CRISPR-Cas9) system 19 . Phenotypic analyses of the CRISPR-Cas9-based BIGFLP KO mutants clearly showed that BIGFLP is responsible for organ growth and particularly development of the ovaries during pupa-adult development in B. mori.

Results
Genome editing. BIGFLP is encoded on chromosome 1 (the Z chromosome). Therefore, since B. mori is a female-heterogametic insect (ZZ in the male and ZW in the female), males have two BIGFLP loci and females have one. The precursor of BIGFLP consists of a secretory signal peptide and a B-domain, C-domain and A-domain 13 (Fig. 1a). To induce mutation in the BIGFLP locus, CRISPR-Cas9 system-based genome editing was performed using the Kosetsu strain of B. mori. Two different sgRNAs were designed, both of which targeted the locus encoding the secretory signal peptide of preproBIGFLP (Fig. 1a). The synthesised sgRNA and Cas9 mRNA were injected into B. mori eggs and two mutant alleles were obtained in the next generation: one containing a 14-base deletion and 3-base insertion (#1-8) and one containing 1-base deletion (#2-24) (Fig. 1b). Both of these were considered to represent BIGFLP KO strains because of a frame-shift in the BIGFLP mRNA. BIGFLP was not detected in the pupal haemolymph of either strain (Fig. 1c), confirming that both were null mutants.
The BIGFLP titres in the haemolymph of wt males, heterozygous males (+/#1-8) and wt females were determined using time-resolved fluoroimmunoassay (TR-FIA) (Fig. 1d,e). (Note: No heterozygous mutant exists in females because BIGFLP is encoded on the Z chromosome.) The peak concentrations in wt males and females were similar to those observed for the Kinshu × Showa strain in our previous study 17 . However, two peaks were observed in wt individuals: at P1-P2 (1-2 days after pupation) and P4 in males and at P0 and P4 in females. The +/#1-8 males had lower titres than the wt males. phenotypes of the Ko strains. Adults of both KO strains were slightly smaller than wt adults, although no specific morphological differences were observed ( Fig. 2a-f). For both sexes, the KO strains had significantly lower body weights than the wt strain from 1 day before the onset of wandering (Fig. 2g,h), suggesting that the BIGFLP gene product affects the growth of larvae towards the end of the feeding stage. No significant difference in body weight was detected between the KO and wt strains in adult males, whereas the difference was maintained in adult females ( Fig. 2g-j). Neither of the KO strains exhibited any developmental delay or mortality.
Previous ex vivo analyses have demonstrated that BIGFLP promotes the growth of the genital imaginal discs 13,17 . Therefore, we first compared the length of the long axis and weight of the male genitalia in adults. Because male genitalia could not be isolated without any loss of content, the genitalia-testes complex was weighed instead. The length of the male genitalia was not different between wt and #1-8 males, while the weight of the genitalia-testes complex of #1-8 males was significantly lower than that of wt males (Fig. 2k,l), consistent with the previous results obtained ex vivo 17 .
We next investigated the effects of BIGFLP on the growth of other organs by measuring the size of the appendages ( Fig. 2m-o). The forelegs (tibia and tarsus) and antennae were significantly shorter in both KO strains than in the wt strain (Fig. 2m,n). By contrast, no significant difference in the area of forewings was detected between the KO strains and wt strain (Fig. 2o). Heterozygous males did not show any significant decrease in the size of their organs. Together, these results demonstrate that the BIGFLP gene product is responsible for the growth of adult organs, although its effect varies among different organs.

Gonadal development.
There was a significant difference in the body weight of adult females but not adult males between the wt and KO strains, suggesting that a difference in the weight of some sex-specific tissues may contribute to the observed difference in body weight. Therefore, the weight of the testes and ovaries in KO and wt adults was compared (Fig. 3a,b). There was no significant difference in the weight of the testes between KO and wt males, whereas the weight of the ovaries was remarkably lower in KO females than in wt females (39% decrease in #1-8; 42% decrease in #2-24).
To determine the critical time at which BIGFLP affects gonadal development, the testes and ovaries of the wt and #1-8 strains were weighed daily throughout post-wandering stages (Fig. 3c,d). No significant difference in testis weight was observed between the two strains at any time (Fig. 3c). However, the ovaries of the KO strain had a lower weight at as early as P0, following which the difference gradually increased to reach a plateau at around P6 (Fig. 3d,e). This observation suggests that the BIGFLP gene product promotes ovarian development during the first half of adult development.
To identify the cause of the observed decrease in the ovary weight, the number and size of laid eggs were examined. It was found that females of both KO strains laid significantly fewer eggs than wt females (Fig. 3f). The genotypes of the mated males did not affect the number of eggs, suggesting that BIGFLP affects neither the mating behaviour nor reproductive potential of males. Furthermore, there was no difference in the size of the laid eggs between the KO and wt strains (Fig. 3g) and most eggs developed normally even in the KO strains. Together, these results suggest that the BIGFLP gene product increases the number of female gametes but does not affect the fertilisation of these.

Decrement of iiS in the mutant ovaries.
We recently reported that BIGFLP promotes the growth of the male genital disc ex vivo via the IIS pathway 17 . To investigate the mechanisms of BIGFLP action on the ovary in vivo, we examined the phosphorylation status of one of the components of the IIS pathway, Akt, in the ovaries of  www.nature.com/scientificreports www.nature.com/scientificreports/ P0 females by Western blotting. The signal intensity of phosphorylated Akt was significantly lower in the #1-8 strain compared with the wt strain ( Fig. 4a,b). To confirm that IIS is stimulated by circulating BIGFLP, we injected purified BIGFLP (0.75 µg) into #1-8 pupae, which caused the signal intensity of phosphorylated Akt in the ovaries to significantly increase (Fig. 4c,d). Thus, it appears that BIGFLP activates the IIS pathway in the ovaries.
timing of BiGfLp action. To determine when BIGFLP promotes ovarian development, we injected BIGFLP into KO females at various times after pupation and measured the weight of the ovaries 24 h after the injection (Fig. 5). Contrary to our expectation, the administration of BIGFLP did not increase the weight of the ovaries at any stages. Therefore, we were unsuccessful in rescuing the decrease in ovary weight in KO females by BIGFLP injection. We considered that the injected BIGFLP was rapidly degraded in the haemolymph, preventing it from affecting the ovaries.

Rescue of restricted ovarian development by ovary transplantation.
To overcome the difficulty in rescuing a restricted development of the ovaries by BIGFLP injection, we next performed an ovary transplantation experiment in which one of the pair of ovaries of KO (mutant) females were exchanged with those of wt females at P0, because this treatment should enable the mutant ovaries to be continuously exposed to the BIGFLP gene product in vivo. As a control, ovaries were also exchanged between the same strains of pupae. This experiment also had another purpose, which is to determine the source of the BIGFLP gene product affecting ovarian development. A large amount of BIGFLP is present in the haemolymph of wt pupae (Fig. 1e). However, BIGFLP is also expressed in the testis sheaths and ovarioles during adult development 16 , raising the possibility that it is the BIGFLP that is produced in the gonad that promotes gonadal development in a paracrine manner. Therefore, it was necessary to test this possibility. We measured the weight of the transplanted ovaries after eclosion, predicting that if the circulating BIGFLP gene product is responsible for ovarian development, the implanted mutant ovary would gain a greater weight in the wt host, whereas if the BIGFLP produced by the ovary itself is responsible, development of the implanted mutant ovary would not be promoted in the wt host. Values are means ± SD (n = 6). Bars with different letters and asterisks indicate significant differences between groups [Tukey-Kramer's test (a,b,f,g) or Student's t-test (c,d) ** p < 0.01; N.S., not significant]. Vn: n days after ecdysis to the fifth instar; Pn: n days after pupation; A0: the day of adult emergence. (2019) 9:18446 | https://doi.org/10.1038/s41598-019-54962-w www.nature.com/scientificreports www.nature.com/scientificreports/ We found that the mutant ovaries that were transplanted into the wt hosts reached a similar weight to the wt ovaries that were transplanted into the wt hosts. By contrast, the wt ovaries that were transplanted into the mutant hosts were much smaller than the wt ovaries that were transplanted into the wt hosts and were comparable in  www.nature.com/scientificreports www.nature.com/scientificreports/ weight to the mutant ovaries that were transplanted into the mutant hosts (Fig. 6a-e). The sizes of the eggs in the implanted ovaries were similar in all donor and host combinations (Fig. 6a-d). These results indicate that BIGFLP is undoubtedly responsible for ovarian development, and it is the BIGFLP gene product in the haemolymph that promotes ovarian development after pupation, with the BIGFLP that is produced by the ovary itself having little, if any, effect.
Effect of brain removal in BIGFLP Ko strains. The ovary transplantation experiment has shown that the gene product of BIGFLP in the haemolymph promotes ovarian development. However, it is unclear whether BIGFLP is the only factor regulating ovarian development because BIGFLP is also expressed in the brain, and its product may be an ILP but not an IGFLP. To address this question, we removed the brain at P0 from both wt and #1-8 females and weighted their ovaries after eclosion (Fig. 7). Although the ovary weights of brain-removed females seemed slightly lower than those of intact or sham-operated females, no significant differences in the weights were observed between the intact and brain-removed animals in both wt and #1-8 strains. These results indicate not only that the fat body-derived IGFLP is the major factor regulating ovarian development but that ILPs produced by the IPCs in the brain, including the gene product of BIGFLP, have little, if any, effect on ovarian development.

Discussion
The present analysis of CRISPR-Cas9-based gene KO B. mori clearly showed that the BIGFLP gene product regulates the growth and development of the adult organs. The size of appendages such as the antennae and legs was significantly lower in BIGFLP-deficient mutants, suggesting that the BIGFLP gene product plays a role in promoting the growth of these organs. However, the most remarkable effect of BIGFLP KO was observed in the ovaries during pupa-adult development, with the weight of the ovaries and the number of eggs laid being reduced by approximately half in the KO mutants but the transplantation of mutant ovaries into wt pupae that expressed BIGFLP resulting in a nearly 50% increase in ovary weight. The decerebration experiments demonstrated that the main BIGFLP gene product involved in the regulation of ovarian development is the BIGFLP produced by the fat body.
In some insects, including B. mori and D. melanogaster, brain-derived ILPs are released in response to a feeding signal and activate the IIS pathway in the peripheral tissues to promote their growth during the feeding stage 9,10 . However, no feeding signal is available during pupa-adult development when the adult organs exhibit www.nature.com/scientificreports www.nature.com/scientificreports/ drastic growth and development. Therefore, another hormonal signal is likely required to activate the IIS pathway and promote tissue growth during this stage of insect development, and it is conceivable that IGFLP plays such a role.
The CRISPR-Cas9 system is a convenient method for constructing gene KO organisms. In the present study, two KO strains with two different sgRNAs were independently constructed (Fig. 1a) and three generations of backcrossing were performed to remove any off-target mutations. In addition, there were no differences in any of phenotypic characteristics between the #1-8 and #2-24 strains, indicating that the decreases in body weight and organ size that were observed in the KO strains did not result from off-target effects of the genome editing but rather were a direct result of the BIGFLP defect.
The KO larvae had a lower body weight than the wt larvae on day 4 of the fifth instar ( Fig. 2g-j), which is a long time before BIGFLP is expressed in the fat body. It has previously been reported that BIGFLP is also expressed in the IPCs in the brains of fourth and fifth instar larvae 16 . The BIGFLP that is produced by the fat body is released as a single-chain peptide, like IGFs, despite the existence of potential cleavage sites in its molecule at the conserved positions that have been identified in various ILPs 13 , presumably because the fat body cells lack processing enzymes for cleaving this peptide. By contrast, the BIGFLP gene product in the IPCs is expected to be processed to generate a two-chain insulin-like structure in these cells. Thus, the IPC-derived BIGFLP gene product may not be an IGFLP but rather an ILP and may be released into the haemolymph together with other ILPs (bombyxins) from the IPCs and function as a member of the bombyxins. It has recently been reported that genetic blockade of bombyxin neural transmission using a Gal4-UAS system in B. mori led to a decrease in the growth rate of fifth instar larvae, demonstrating a role of bombyxins in the regulation of larval growth 18 . Therefore, it is possible that the observed difference in body weight between the KO and wt larvae at the mid-fifth instar (Fig. 2g-j) reflects the total amount of bombyxins in the haemolymph.
This then raises the question why the body weight decline in the KO strain is only manifested after several days in the fifth instar despite the BIGFLP gene being constantly expressed in this instar. It is assumed that the relative importance of this peptide among the bombyxins increases at the mid-fifth instar stage due to a decrease in the expression levels of other bombyxin genes or a decrease in the total bombyxin concentration in the haemolymph caused by the dramatic increase in haemolymph volume that is associated with rapid larval growth in the fifth instar.
After the onset of wandering, there was no change in the difference in body weight between the KO and wt strains in either sexes until eclosion (Fig. 2i,j). This is reasonable because gut-purged larvae and pupae do not eat or excrete anything. However, the difference in males disappeared after eclosion. Soon after eclosion, adult moths excrete the waste materials that accumulated during pupa-adult development in the form of meconium. Therefore, the observed disappearance of the difference in body weight between KO and wt males after eclosion suggests the possibility that wt animals had a larger volume of meconium. This, in turn, might reflect a higher metabolic activity of wt animals and thus indicate that BIGFLP enhances the metabolic activity of the pupae.
Satake et al. 20 previously reported that the injection of bombyxin into the isolated abdomens of B. mori larvae led to a decrease in the contents of trehalose in the haemolymph and glycogen in the fat body, both of which are major storage carbohydrates in insects, and it has also recently been suggested that these carbohydrates are used for energy production 21 . Since only a single insulin/IGF-like receptor gene exists in the B. mori genome 8 , it is considered that bombyxins and BIGFLP act through the same receptor, with their functions compensating each other. Therefore, BIGFLP might support the growth of adult organs by activating energy metabolism. To test this hypothesis, the amounts of unused energy and material stocks for adult organ development must be compared between KO and wt males in the future. In females, the differences in the body weight between the strains did not disappear. This is possibly due to a large reduction in the ovary weights of the KO females. Figure 7. Effects of brain removal on ovarian development. The brain was removed from KO and wt females within 6 h of pupation, and the ovaries were weighed after eclosion. Values are means ± SD (n = 6). There were no significant differences between groups (Tukey-Kramer's test: N. S., not significant). www.nature.com/scientificreports www.nature.com/scientificreports/ In D. melanogaster, the downregulation of Dilp6 expression after pupation results in a decrease in the consumption of triacylglycerol and glycogen 15 , suggesting that one of the roles of Dilp6 during metamorphosis is the activation of energy metabolism. Thus, the activation of energy metabolism may be a common action of insect IGFLPs during adult development, although these peptides may also have other functions.
A notable phenotype of the Dilp6 mutant in D. melanogaster was the decrease in the body weights of post-feeding larvae, pupae and adults 14,15 . This phenotype is similar to that in the BIGFLP mutant of B. mori but is somewhat more severe, especially in adults: mutant adults had obviously lower body weights than wt even in males. This difference may be attributed to the difference in the number of ILP/IGFLP genes expressed during post-feeding stages between B. mori and D. melanogaster. In the genome of D. melanogaster, six ILP/IGFLP genes have been identified (Dilp1-6; note that Dilp7 and Dilp8 are considered relaxin-like genes) 10 , and only one of them, Dilp6, is strongly expressed during the pupal stage 14 . In contrast, B. mori has approximately 40 such genes in its genome, and three genes, bombyxin-X1, -Y1 (identical to BIGFLP) and -Z1, are strongly expressed during the pupal stage in the fat body 7,22 . Therefore, it is conceivable that the gene products of bombyxin-X1 and -Z1 may also promote the growth of adult organs like BIGFLP does. The gene products of these genes are possibly IGFLPs rather than ILPs (bombyxins) because the fat body is not likely to possess proprotein convertases. It is unknown at present whether these peptides are actually released into the haemolymph. However, if the concentrations of these peptides are high enough to support the growth of adult organs, the presence of these IGFLPs in the haemolymph of BIGFLP KO pupae may explain the milder-than-expected phenotype of this mutant.
The KO moths had drastically lower ovary weights than the wt moths (Fig. 3d). Furthermore, although the KO moths had similar egg sizes to the wt moths, they laid approximately 50% fewer eggs (Fig. 3f,g), suggesting that the BIGFLP gene product increases the number of mature eggs. The mechanism by which BIGFLP regulates the number of mature eggs remains unclear but one possibility is that it is involved in the accumulation of energy and material stores for vitellogenesis. Since the body weight of wandering larvae was significantly lower in the BIGFLP KO strains (Fig. 2h), the mutant pupae may not have sufficient nutrients to support the production of a normal number of mature eggs.
However, since a large amount of BIGFLP is secreted only after pupal ecdysis, this peptide may exert its major effect(s) on ovarian development during pupa-adult development, although the effect of the brain-derived BIGFLP gene product on mid-fifth instar larvae may also contribute to egg production to some extent. The finding that the IIS signal level in the ovary as measured by Akt phosphorylation greatly increased in a BIGFLP-dependent manner (Fig. 4) indicates a possible mechanism by which BIGFLP regulates ovarian development. In D. melanogaster, the ovaries develop after eclosion and the number of eggs laid decreases under starving conditions [23][24][25] . Furthermore, it has been shown that the starvation-induced decrement of IIS in the ovaries induces programmed cell death of the nurse cells in the ovarian follicles 23 . In a eutrophic environment, the increased secretion of ILPs activates the IIS pathway in the ovaries, whereas under starving conditions, ILP secretion is inhibited and IIS decreases in the ovaries. These observations suggest that continuous activation of the IIS pathway in the ovary is necessary for normal follicle maturation. In the present study, BIGFLP injection into the KO pupae was unable to rescue the decrease in ovary weight (Fig. 5). However, mutant ovaries that had been transplanted into wt pupae attained a similar weight to control wt ovaries (Fig. 6), also suggesting that continuous activation of IIS is necessary for maintaining a normal number of follicles in the ovary. It is therefore possible that BIGFLP also activates the IIS pathway to prevent programmed cell death of the nurse cells in the ovaries.
Another possible mechanism by which BIGFLP regulates the number of eggs may be related to its effect on energy metabolism. As discussed above, this peptide may activate energy metabolism to support the growth of adult tissues. A substantial amount of energy is required to produce mature eggs and pupae do not incorporate any nutrition, meaning that the ovaries must develop using accumulated nutritional stocks. Therefore, it is conceivable that the KO female pupae were unable to efficiently use their stored nutritional resources to develop ovaries. To test this hypothesis, the amount of unused energy stores should be compared between KO and wt adult females in the future.
It has previously been demonstrated that ecdysteroids are involved in the regulation of ovarian development in B. mori 26 . For example, it was found that ovarian development did not occur when the abdomens of female Bombyx pupae were isolated from the anterior parts of the bodies by ligature immediately after pupal ecdysis but that the ovaries developed to reach an almost comparable size to normal mature ovaries when 20-hydroxyecdysone (20E) was injected into the isolated abdomens 27,28 , indicating that 20E is essential for inducing ovarian development. Therefore, we considered it possible that the reduced ovary weight in the KO females was caused by a decline in the ecdysteroid titre in the haemolymph. However, the ecdysteroid titre in the KO female pupae was the same as or even higher than that in the wt pupae ( Supplementary Fig. S4). Therefore, the observed reduction in the ovary weight of the KO females was not caused by a reduced ecdysteroid signal but rather by a lack of BIGFLP itself, although ecdysteroids may be essential in the initial stage of the induction of ovarian development. Thus, we speculate that the increase in ecdysteroids after pupal ecdysis initiates ovarian development and, at the same time, induces BIGFLP secretion from the fat body, thereby promoting the growth and development of the ovaries.

Animals.
A bivoltine strain of B. mori (Kosetsu) was reared at 25 ± 1.0 °C under a 12-h light and 12-h dark photoperiod and a relative humidity of 50-70% on an artificial diet 'Silkmate' (Nihon Nosan Kogyo, Yokohama, Japan). Most larvae (>95%) started wandering on day 5 of the final instar and ecdysed to pupae on day 8. Most pupae (>90%) eclosed at 8 days (males) or 9 days (females) after pupation. During screening, larvae of the KO strains were reared at 25 ± 1.0 °C under a 20-h light and 4-h dark photoperiod to induce non-diapause eggs as described previously 29 .
Haemolymph collection and hormone detection. The haemolymph was collected from day-0 pupae for BIGFLP detection by Western blotting. Western blotting was carried out as previously described by Okamoto et al. 13 , with the exception that the anti-BIGFLP antibody D7H3 was used in place of the anti-bombyxin antibody M7H2 as the primary antibody. The haemolymph was also collected daily at 6 h after lights-on for 8 days after pupation (P0 to P7) and the titres of BIGFLP and ecdysteroids were determined by TR-FIA, as described previously 17,32 . phenotypic analyses. The body weight of each individual was measured daily at the same time of day (2.5 h after lights-on) from day 0 of the fifth instar to day 0 of the adult stage. Images of the male genitalia and adult appendages (antennae, forelegs and forewings) were obtained with a Nikon D5000 (Nikon, Tokyo, Japan) or Finepix S9900W (Fujifilm, Tokyo, Japan) digital camera. The lengths of the male genitalia, antennae and forelegs (tibia and tarsus) and the area of the forewings were then measured from the images. The ovaries and testes were dissected daily from pupae at 4-8 h after lights-on and weighed. egg laying. Adults were mated for 2 h at 25 °C under light conditions. The lights were then switched off for 24 h, during which time the females laid their eggs on egg boards at 25 °C. The numbers and sizes of the laid eggs were measured using images obtained using a Finepix S9900W digital camera.
BiGfLp injection. BIGFLP was purified from the pupal haemolymph as described previously 17 . From day 0-6 female pupae were anesthetized by submerging them in water for 30 min. BIGFLP was diluted with deionized water containing 0.1% bovine serum albumin, and 0.75 or 1 µg of the hormone (25 µl) were injected into the dorsal intersomitic region of the pupae. Control pupae were injected with the vehicle.
Detection of phosphorylated Akt. Western blotting was performed to detect phosphorylated Akt, Akt and alpha-tubulin, as described previously 17 . The signal intensities of phosphorylated Akt and Akt were measured using Image Lab ™ version 4.1 software (Bio-Rad, Hercules, CA, USA). ovary transplantation and brain removal.  and wt females were water-anesthetized for 30 min within 6 h of pupation. For ovary transplantation, one ovary was removed from each individual. A wt or #1-8 ovary was then implanted in different individuals. For brain removal, the tip of the head was cut open, and the brain was removed. The wound was sealed with paraffin. After eclosion, the implanted ovaries were dissected, photographed using a Finepix S9900W digital camera and weighted.
Statistical analyses. All data are presented as means ± standard deviations (SD). Student's t-test was used for the comparison of two groups and Dunnett's or Tukey-Kramer's tests were used for multiple comparisons.

Data availability
The datasets in the current study are available from the corresponding author on reasonable request.