The FOXO transcription factor controls insect growth and development by regulating juvenile hormone degradation in the silkworm, Bombyx mori

Forkhead box O (FOXO) functions as the terminal transcription factor of the insulin signaling pathway and regulates multiple physiological processes in many organisms, including lifespan in insects. However, how FOXO interacts with hormone signaling to modulate insect growth and development is largely unknown. Here, using the transgene-based CRISPR/Cas9 system, we generated and characterized mutants of the silkworm Bombyx mori FOXO (BmFOXO) to elucidate its physiological functions during development of this lepidopteran insect. The BmFOXO mutant (FOXO-M) exhibited growth delays from the first larval stage and showed precocious metamorphosis, pupating at the end of the fourth instar (trimolter) rather than at the end of the fifth instar as in the wild-type (WT) animals. However, different from previous reports on precocious metamorphosis caused by juvenile hormone (JH) deficiency in silkworm mutants, the total developmental time of the larval period in the FOXO-M was comparable with that of the WT. Exogenous application of 20-hydroxyecdysone (20E) or of the JH analog rescued the trimolter phenotype. RNA-seq and gene expression analyses indicated that genes involved in JH degradation but not in JH biosynthesis were up-regulated in the FOXO-M compared with the WT animals. Moreover, we identified several FOXO-binding sites in the promoter of genes coding for JH-degradation enzymes. These results suggest that FOXO regulates JH degradation rather than its biosynthesis, which further modulates hormone homeostasis to control growth and development in B. mori. In conclusion, we have uncovered a pivotal role for FOXO in regulating JH signaling to control insect development.


Forkhead box O (FOXO) functions as the terminal transcription factor of the insulin signaling pathway and regulates multiple physiological processes in many organisms, including lifespan in insects. However, how FOXO interacts with hormone signaling to modulate insect growth and development is largely unknown. Here, using the transgene-based CRISPR/Cas9 system, we generated and characterized mutants of the silkworm Bombyx mori FOXO (BmFOXO) to elucidate its physiological functions during development of this lepidopteran insect. The
BmFOXO mutant (FOXO-M) exhibited growth delays from the first larval stage and showed precocious metamorphosis, pupating at the end of the fourth instar (trimolter) rather than at the end of the fifth instar as in the wild-type (WT) animals. However, different from previous reports on precocious metamorphosis caused by juvenile hormone (JH) deficiency in silkworm mutants, the total developmental time of the larval period in the FOXO-M was comparable with that of the WT. Exogenous application of 20-hydroxyecdysone (20E) or of the JH analog rescued the trimolter phenotype. RNA-seq and gene expression analyses indicated that genes involved in JH degradation but not in JH biosynthesis were up-regulated in the FOXO-M compared with the WT animals. Moreover, we identified several FOXObinding sites in the promoter of genes coding for JH-degradation enzymes. These results suggest that FOXO regulates JH degradation rather than its biosynthesis, which further modulates hormone homeostasis to control growth and development in B. mori. In conclusion, we have uncovered a pivotal role for FOXO in regulating JH signaling to control insect development.
Insect growth and development are intricately modulated by many conserved signaling pathways. The insulin/insulin-like growth factor signaling (IIS) 2 and target of rapamycin pathways regulate growth rate via nutrition and immunity levels (1,2). Growth periods, including molting and metamorphosis, are governed by ecdysteroid and juvenile hormone (JH) (3,4). Previous studies have shown that ecdysone synthesis is regulated by IIS and target of rapamycin signaling, whereas IIS is antagonized by ecdysteroids (5)(6)(7). In Drosophila melanogaster mutations to genes in the IIS not only perturb growth and result in smaller body size (8 -10) but also alter JH biosynthesis (11). The cross-talk between IIS and hormones indicates the complexity of regulation in insect development. However, the detailed mechanism of how IIS interacts with insect developmental hormones remains largely unknown.
As the major terminal transcription factor of IIS and the target of protein kinase B (PKB/Akt), the forkhead box O (FOXO) (12) plays multiple roles in many cellular and physiological processes (13). FOXO has been identified as the key factor in limiting longevity in worms (daf-16) and flies (dFOXO) (14 -17). The overexpression of dFOXO during early larval stages inhibited larval growth (18) and showed reduced body size when expressed in the third instar larvae (19). The ectopic expression of dFOXO in specific tissues (i.e. eye or wing) also led to characteristic phenotypes with reduced cell size and cell number that could partially be rescued by the co-expression of upstream insulin signaling components (19). Furthermore, as IIS-mediated growth control is FOXO-dependent, the loss-offunction of FOXO also results in delayed adult development time and smaller size (20). In contrast to daf-16, the multitude of phenotypes caused by the reduction in IIS are not completely suppressed in the absence of FOXO, and only the increase in lifespan extension phenotype was blocked by the removal of dFOXO in D. melanogaster (20). Although the regulatory networks of IIS and FOXO are well understood, it remains unclear how FOXO interacts with hormone signaling to mediate growth and development.
Most FOXO functional analyses in Drosophila have focused on its roles in the regulation of IIS-mediated growth control (1,2). The effects of JH on growth rate are dependent on FOXO (4). Recent studies have implicated FOXO in the regulation of JH metabolism but did not further explore the underlying mechanisms (22,23). Nevertheless, as Drosophila exhibits relatively short larval development and body size, there are only a few reports concerning the precise function of FOXO in the modulation of the metamorphic transition that could be explored in other insects, such as the lepidopteran model insect Bombyx mori, which exhibits a longer developmental time and a larger body size. In B. mori, BmFOXO is highly expressed in the fat body and could be induced by 20E to promote lipolysis. The down-regulation of BmFOXO during the fourth larval instar using RNAi partially delayed the molting process (24). However, the comprehensive functions of FOXO in B. mori have yet to be revealed. Recent advances in genetic manipulation techniques, such as transgenesis and genome editing tools (25,26), have provided promising approaches for dissecting functions of genes such as BmFOXO.
Here, we elucidated how FOXO interacts with hormone signaling to regulate growth and metamorphosis in B. mori by knocking out BmFOXO using the transgene-based CRISPR/ Cas9 system. BmFOXO mutants showed a significant growth delay from the second larval instar and skipped the last molt, resulting in precocious metamorphosis after only four larval instars (trimolter), although the entire developmental time of the larval instar was comparable with that of WT animals. The hormone titers (both JH and 20E) were lower in mutants, and the phenotypic defects could be rescued by applying either 20E or JH analogs. Intriguingly, the transcriptome analysis revealed that the genes coding for JH metabolizing but not biosynthesis enzymes were up-regulated in mutants. We confirmed that FOXO binds to the response elements present in the promoter of these genes. Our results revealed a novel mechanism by which FOXO interacts with the JH degradation pathway to regulate growth and development in silkworms.

Generation of BmFOXO mutants using the transgenic CRISPR/ Cas9 system
To generate FOXO mutants, we used a binary transgenic CRISPR/Cas9 system that has been described in our previous publication (27). Two guide RNAs (sgRNA) targeting exons 2 and 3 coding for the majority of forkhead domain, were selected (supplemental Fig. S1A). We constructed two piggyBac transposon-based transformation plasmids expressing Cas9 protein or the FOXO-sgRNAs (supplemental Fig. S1B). The nanos promoter was used to specifically express Cas9 protein in the germline (28,29), whereas sgRNAs were expressed under control of the ubiquitous U6 promoter, as described previously (30). Transgenic animals carrying nos-Cas9 or U6-FOXO-sgRNA were fully viable and fertile, indicating that the accumulation of neither Cas9 nor FOXO-sgRNA alone had a deleterious effect on silkworm physiology. We crossed the nos-Cas9 and U6-FOXO-sgRNA lines to obtain founder offspring that specifically express the active Cas9-sgRNA complex in the germline, resulting in mutations at targeted genomic loci of the BmFOXO gene (supplemental Fig. S1B). We investigated the BmFOXO mutation in different tissues (brain, fat body, silk gland, epidermis, mid-gut, ovary, and testis) and found that two CRISPR targeting sites (BmFOXO-s1 and BmFOXO-s2) were disrupted effectively in all tissues tested (supplemental Fig. S1C).
The somatic mutation events were evaluated using polymerase chain reaction (PCR) analysis and subsequent sequencing of the targeted genomic loci. The sequencing results identified various deletion mutations at two sgRNA targets and deletions ranging from 2 to 64 bp, indicating that successful mutations were induced using the transgenic CRISPR system (supplemental Fig. S1D). Western blots showed a significant decrease in the levels of the BmFOXO protein in the fat body (where BmFOXO is dominantly expressed) in the mutants (supplemental Fig.  S1E). These results revealed successful disruption of the BmF-OXO gene in vivo.

Precocious metamorphosis in BmFOXO mutants
BmFOXO mutants showed no deleterious effects during embryonic development. However, unlike WT larvae that molt four times prior to metamorphosis, the mutant larvae grew slower and molted only three times (Fig. 1A). The mutants showed significant growth delay from the first to the third larval stage when compared with the WT animals (Figs. 1A and supplemental Fig. S2). However, both the mutants and WT animals reached similar body sizes before entering the fourth larval stage (L4). Interestingly, unlike the WT animals that entered the final (fifth) instar stage after the fourth molting and ecdysis, 78% of FOXO mutants (n ϭ 80) did not undergo the fourth molt prior to the larval-pupal metamorphosis (Fig. 1A).
The time span of the fourth instar mutants (120 h) is longer than that of the fourth (72 h) and fifth instar (84 h) WT animals (Fig. 1A). The silk glands in the mutants have undergone premature development and the accumulation of carotenoids (which turned the middle silk gland yellow), a character of mature silk glands were detected from the L4D4. The mature silk glands occurred 72 h earlier than in WT animals (supplemental Fig. S3). The mutant body size (including wandering larvae, pupae and adults) was significantly reduced. The mutant silkworms were approximately half the weight of WT animals ( Fig. 1, B and C, supplemental Figs. S4 and S5). However, pupation and eclosion processes were normal (supplemental Fig. S6) and female mutants were fully fertile but laid fewer eggs (supplemental Fig. S7).
Because FOXO is a terminal transcription factor in IIS, expressions of genes known to play critical roles in IIS were determined by quantitative PCR in WT and BmFOXO mutants. The results showed that the expression of the insulin receptor (InR) gene was reduced by 86%, whereas the serine/threonine kinase (Akt) and 4E-binding protein (4EBP) were up-regulated in mutants when compared with their expression in WT ( Fig.  1D and supplemental S2B). These results suggest that the loss of FOXO function perturbed IIS.

Defective ecdysone signaling in BmFOXO mutants
The ecdysteroid titers increase during the larval-larval or larval-pupal transitions to initiate molting or metamorphosis (31). We measured the ecdysteroid levels in hemolymph on the second and third days of the fourth larval stage (L4D2 and L4D3). The results showed that ecdysteroid levels in the mutants decreased ϳ45 and 28% on L4D2 and L4D3, respectively, when compared with their levels in WT animals ( Fig. 2A), suggesting that lower ecdysteroid titers in mutants might be responsible for their failure to undergo a fourth molt. Moreover, injection of 20E into the L4D3 mutant larvae caused 75% of them to molt into the fifth larval instar by 60 h after injection ( Fig. 2B (b and bЉ) and C). Twenty-five percent of the 20E-injected mutant larvae were either partially rescued or died (bٞ in Fig. 2B and supplemental Fig. S8). The successfully rescued larvae went through a longer fifth larval stage and reached a larger pupal body size. These results showed that the application of 20E could turn trimolters back to tetramolters in most FOXO mutants suggesting that lower levels of ecdysteroids during L4 could be a reason for the trimolter phenotype observed in these animals.
We then investigated the expression levels of genes known to be involved in ecdysone action. Significantly lower levels of E75, hormone receptor 3 (HR3), and ecdysone receptor isoform A (EcRA) mRNA were detected on L4D2 and L4D3 in the mutants when compared with their levels in WT (Fig. 3, A-C). Furthermore, we determined the expression pattern of the HR3 gene from L3D1 to wandering stage in WT and mutants and found that HR3 mRNA levels were significantly lower in mutants when compared with its levels in WT during the L4 to L5 transition. These data suggest confirmation of the lower levels of ecdysteroids during this period and the reduction in ecdysteroid titers may be responsible for trimolter phenotype observed (Fig. 3D). No significant expression changes in the broad gene were detected (supplemental Fig. S6D), broad plays an essential role in pupal commitment (3), indicating that loss of FOXO did not interfere with the larval-pupal transition. Our results suggest that the expression of genes from ecdysone signaling was suppressed in FOXO mutants and contributed to the appearance of trimolter phenotype.
We also investigated the changes in expression genes coding for the major key enzymes in ecdysteroid biosynthesis (in the prothoracic glands, PG) and metabolism pathways (in the fat body). The changes in expression of ecdysteroid biosynthesis genes in the PG on L4D3 were variable and gene-dependent. Neverland (Nvd) and Disembodied (Dib) were up-regulated by 3.63-and 1.73-fold, respectively, whereas Spook (Spo), Phantom (Phm), and Shade (Shd) were down-regulated by 1.89-, 4.76-, and 1.92-fold, respectively (Fig. 3F). The expression of ecdysone oxidase (EO) was also significantly reduced (Fig. 3E). The expression pattern of ecdysteroid biosynthesis and metabolism genes suggest that BmFOXO depletion may have disrupted ecdysteroid signaling as well as ecdysteroid biosynthesis and metabolism.

Activation of the JH metabolism pathway in BmFOXO mutants
Because precocious metamorphosis is a typical JH-deficient phenotype in B. mori (32,33), we hypothesized that FOXO depletion might have affected JH signaling. We investigated the JH titers during the L4 of WT and FOXO-M and found that JH I in the hemolymph collected from mutants was reduced significantly to undetectable levels (Fig. 4A). Also, application of the JH analog, methoprene, to the newly molted fourth instar mutant larvae resulted in a rescue and all treated mutants having undergone a fourth molt (Fig. 4, B and C). The acetonetreated mutant larvae skipped the fifth instar and became pupae at 144 h after the third molt, whereas the rescued larvae have undergone the fourth molt at 132 h after methoprene application and began to spin at around 168 h after ecdysis (Fig. 4D). The body size of rescued larvae was similar to WT animals. Taken together, these data suggest that BmFOXO depletion affects titers and action of both JH and 20E, the two most important hormones that regulate molting and metamorphosis. Because of changes in hormone titers and action, the FOXO mutants have undergone three molts, whereas WT animals undergo four molts.
To explore the molecular basis for the phenotypic effects observed, we performed differential gene expression analysis by sequencing the fat body dissected from the mutant and WT animals and detected numerous differentially expressed genes between mutants and WT animals. These genes were primarily involved in cellular and metabolic processes, and binding and catalytic activities according to GO functional classification (supplemental Fig. S9, A and B). There are 4666 genes down-regulated, and only 357 genes were up-regulated in mutants when compared with their expression in WT animals (supplemental Fig. S9, C and D). Interestingly, the primary JH degradation pathway genes, including juvenile hormone esterase (JHE), juvenile hormone epoxidase hydrolase (JHEH), and juvenile hormone diol kinase (JHDK), were highly up-regulated in mutants (supplemental Table S1). We confirmed the activation of these three genes in the fat body using quantitative PCR and found that all three genes were up-regulated on L4D3 (ϳ107fold increase for JHE, 6-fold increase for JHEH, and 3-fold increase for JHDK) (Fig. 5B). In contrast, on L4D2, only JHE showed a significant up-regulation (ϳ19-fold increase). Also, JHE enzymatic activity in the hemolymph was much higher in the mutants than that in WT; the average JHE enzymatic activity was 4.45-and 3.92-fold high in mutants on L4D2 and L4D3, respectively, when compared with that in WT (Fig. 5A). We further explored the JHE expression pattern from the third instar until the wandering stage and observed that JHE mRNA levels are higher during each larval molt, particularly in the later fourth instar stage suggesting a dominant role of JHE overexpression in trimolter phenotype (Fig. 5C).
The lower expression levels of the key genes involved in JH signaling: Kr-h1␣ and Kr-h1␤ during third instar to wandering stage were detected in BmFOXO mutants when compared with their expression in WT (Fig. 5, D and E). In contrast, there is no significant difference in the expression of genes coding for two key enzymes that act at the final steps of the JH biosynthesis pathway, epoxidase (CYP15C1) and JHA methyltransferase (JHAMT) in corpora allata (CA) (Fig. 5F). These data suggest . FOXO-M injected with 20E reached the fifth instar, whereas those injected with DMSO remained in the fourth instar; bЈ, one larva from the DMSO-injected group (fourth instar); bЉ, one larva from 20E-injected group (fifth instar); bٞ, one larva from 20E-injected group, which partially recovered molting but failed to reach the next stage. C, the development of two groups in later stages. L4-P stages are marked in red, yellow, and green, respectively. The curve between different stages shows the transition ratio. The dark filled triangles indicate the time points when 20E or DMSO was injected. D, summary of 20E rescue. Twenty mutant larvae were treated in each group.
that BmFOXO depletion activated the JH degradation pathway but did not inhibit the JH biosynthesis pathway, which is different from the previous studies in Drosophila (20).

Identification of BmFOXO-binding sites in the promoter of genes coding for JH-metabolizing enzymes
As a terminal transcription factor of IIS, FOXO is known to bind to the conserved binding site (GTTTYAA (Y ϭ T or C)) present in the promoters of many genes and regulate their expression (34). We searched 2-kb regions in the promoters of JHE, JHEH, and JHDK to identify potential binding sites for FOXO (Fig. 6A). Two potential FOXO-binding sites in the promoter regions of JHE and JHDK and one binding site in the JHEH promoter were identified (Fig. 6A). To examine whether FOXO can bind to these sites, we performed an electrophoretic mobility shift assay (EMSA) using purified His-tagged FOXO protein (Fig. 6, B-D).
We amplified an ϳ200 bp flanking sequence around the binding sites and labeled it with Cy5, followed by incubation with FOXO protein and resolved by gel electrophoresis. As shown in Fig. 6, FOXO bound to these predicted binding sites (Fig. 6, B-D). The competition experiments confirmed the specificity of FOXO binding. Depletion of BmFOXO induced the expression of genes coding for JH degradation enzymes suggesting that FOXO is likely involved in repression of JH degradation genes, and FOXO mutation eliminated this suppression effect, resulting in activation of these genes to accelerate JH degradation.

Discussion
In the present study, using a transgenic CRISPR/Cas9-derived somatic mutagenesis system, we showed that FOXO depletion affected growth and development as well as ecdysteroid and JH signaling in B. mori. Comparative analyses of growth and development in FOXO mutants and WT animals led to three major findings. First, the depletion of BmFOXO induced growth delay and precocious metamorphosis. Second, BmFOXO directly modulated the JH degradation pathway rather than the biosynthesis and thus perturbed the ecdysone homeostasis. Third, in contrast to typical JH-deficient mutants reported in previous studies (3,32), BmFOXO depletion showed more complex phenotypes, suggesting BmFOXO as the nexus mediating the intimate link between various biological processes (Fig. 7).
Previous studies in B. mori suggested that embryonic growth and morphogenesis are independent of JHs, and precocious metamorphosis was induced after the second larval stage in the absence of JH or JH signaling (32). Also, the nm-g mutant in Bombyx reduced the ecdysteroid titer and induced larval arrest primarily during the first instar, suggesting a role for ecdysteroid during the early larval development (35). The growth delay in BmFOXO mutants initiated during the first larval instar was different from that in JH-or 20E-deprivation mutants, indicating that insulin signaling but not JH or 20E signaling play the dominant role in regulation of the early larval status in B. mori.
An intriguing finding in the present study is the emergence of the trimolter in FOXO mutants. The mutants took 36 h longer to reach the fourth larval stage, and spent 36 h shorter time than  (as a control). b, five individuals from different experimental groups. The group treated with methoprene (FOXO-M ϩ Methoprene) molted into the fifth instar, whereas the other group treated with acetone (FOXO-M ϩ Acetone) remained in the fourth instar. bЈ, one larva from the methoprene-treated group transformed into the fifth instar after 144 h. bЉ, one larva from the acetone-treated group remained in the fourth instar after 144 h. bٞ, one larva in the acetone-treated group reached the wandering stage after 144 h. C, ratio of individuals reaching the fifth instar in two groups (n ϭ 20). The white section of the bar represents individuals that reached the fifth instar (tetramolter), and the black section shows trimolter larvae. All individuals from the methoprene-treated group (100%) transitioned to the fifth instar, whereas only 20% in the acetone-treated group transitioned to this stage. D, the later stages of larval development in the two groups. L4-P stages are indicated in red, yellow, and green. The curve between different stages shows the transition ratio. The dark-filled triangle shows the time points when acetone or methoprene was applied.
the total developmental times of L4 and L5 in WT insects. Although the mutants experienced only four larval stages, the entire larval development period was not significantly different between WT and FOXO-M. The ecdysteroid titer was lower in the mutants, and the expression levels of ecdysone biosynthesis and degradation genes were affected. The phenotypic defects could be partially rescued by 20E application, suggesting the partial involvement of ecdysone action in FOXO mutants. Moreover, the BmFOXO trimolter mutants completely skipped the fourth molt without any deleterious physiological effects, molting defects, or lethality, indicating that reduced ecdysone synthesis and action are not the main reasons for the trimolter phenotype observed.
Furthermore, we detected a significant reduction in JH titers in mutants and fully rescued the trimolter to tetramolter by application of the JH analog, demonstrating the dominant role of JH. RNA-seq analysis, quantitative PCR, and EMSA experiments showed that activation of JH degradation in FOXO mutants may be the primary factor contributing to this phenotype. Previous studies showed that the effects of JH on the growth rate depend on FOXO (4, 22), and our data revealed a direct connection between FOXO, which is involved the IIS pathway and JH metabolism. Furthermore, the BmFOXO depletion trimolter is different from the previously reported JH-deficient mutants in Bombyx (32,33), and the results suggested a more complex mechanisms, including IIS, JH, and ecdysone, are responsible for BmFOXO-induced precocious metamorphosis. The impaired insulin pathway, significant loss of JH, and perturbed ecdysone signaling resulted in skipping of the last molt and the maintenance of feeding until the larvalpupal transition. The major role of JH is to modulate ecdysone action (36), and inappropriate changes in JH titer can produce heterochronic shifts in the developmental program activated by ecdysone. Therefore, the activation of JH degradation by the FOXO mutation reduced the JH titer, which further altered ecdysone homeostasis and induced the loss of the fourth molt-  In FOXO knock-out silkworms, insulin signaling is also reduced, causing growth delay and reduced body size. Taken together, these findings show that FOXO plays a pivotal role in regulating growth and development through interactions with hormone signaling.
ing. Taken together, these results suggest a pivotal role for FOXO, which regulates insulin signaling and interacts with hormone signaling to control insect larval growth and development. With CA ceasing the production of JH during pupal commitment, the effect of FOXO mutation on JH metabolism was decayed and the ecdysone biosynthesis and action recovered, resulting in successful larval-pupal and pupaladult metamorphosis.
The present study provides evidence that BmFOXO affects the JH degradation pathway, rather than the biosynthesis pathway, to regulate insect growth and development. The trimolter phenotype observed as a result of BmFOXO depletion is distinct from silkworm precocious metamorphosis induced by JH deficiency, indicating a more complex mechanism in which JH, IIS, and ecdysone are also involved (Fig. 7). As one of the downstream targets of IIS, FOXO is involved in comprehensive physiological processes, and its biological functions in insect growth and development remains to be fully elucidated.

Insect strains and genetic transformation
The multivoltine and nondiapausing silkworm strain Nistari was used for genetic transformation (33) and phenotypic analysis in the present study. The DNA mixture was microinjected into preblastoderm G 0 embryos within 8 h after oviposition and incubated at 25°C for 10 -11 days until hatching. The larvae were reared with fresh mulberry leaves or artificial diets under standard conditions (33). G 0 adults mated with WT and G 1 offspring were screened under microscopy using RFP or GFP filters (Nikon AZ 100) to obtain positive transgenic animals.

Genotyping of target loci mutations
Genomic DNA were extracted after incubation with proteinase K (Thermo) and purified as previously described (38). The DNA fragments around the designed sgRNA targeting sites were amplified using Phusion Polymerase (New England Biolabs), and the corresponding PCR products were extracted and ligated into the pJET T-vector (Thermo) and subsequently sequenced. The Sanger sequencing results were analyzed and compared with those of the WT to confirm the detailed mutation events in FOXO mutants. For each sgRNA targeting site, at least 10 mono-clones were sequenced, and 3 different types of mutation events were demonstrated. The T7 endonuclease I assay was performed as previously reported (30).

Quantitative PCR
Two-step quantitative PCR was used to assay the related gene transcripts in transgenic BmFOXO mutants. Total RNA was extracted from different tissues at L4D2 or L4D3 (fat body for JH degradation, insulin, and 20E signaling-related genes, corpora allata for JH synthesis-associated genes, prothoracic gland for ecdysone synthesis-related genes) using TRIzol reagent (Invitrogen) and subsequently reverse-transcribed into cDNA using the RevertAid First-Strand cDNA synthesis kit (Thermo Fisher). The transcript levels were assayed using SYBR Green Master Mix (TOYOBO) via a standard curve normalized against BmRP49 expression. The primers are listed in supplemental Table S2.

Electrophoretic mobility shift assay (EMSA)
The entire coding sequence for FOXO was obtained and cloned into the pET-32a expression vector (Thermo). Then, the pET-32a-FOXO-His vector was used to express FOXO protein in the JM110 Escherichia coli strain, followed by introduction with isopropyl 1-thio-␤-D-galactopyranoside at 18°C. Fusion FOXO protein was extracted and purified with nickel-nitrilotriacetic acid beads (Qiagen) according to the manufacturer's instructions. The predicted FOXO-binding sites containing DNA fragments were amplified using the primers listed in supplemental Table S2 comprising sequence-specific and Cy5-universal primer sequences (not labeled). Subsequently, labeled Cy5-tag universal primer was used to amplify the corresponding fragments, and the amplified fragments were collected using a Gel Extraction Kit (OMEGA). EMSA was performed using Cy5-labeled probes (Sangon) as previously described (39). For the competition assay, 100-fold excess of unlabeled DNA and nonspecific DNA were added as competitors. The mixture was electrophoresed at 4°C on a 6% native polyacrylamide gel for 1.5 h at 100 V. The Cy5-labeled DNA on the gel was scanned and analyzed using the Starion FLA-9000 scanner (FujiFilm).

Hormone rescue experiments
For ecdysteroid rescue experiments, 5 l of ␤-ecdysone (Sigma) (2 g/l, 10 g in total) was injected into the first abdominal segment of each L4D3 FOXO-M larvae, and the same amount of 5% DMSO was injected as a control. The JH analog, methoprene (Sigma), was dissolved to 5 g/l in acetone and applied to the dorsum of newly molted fourth instar

FOXO regulates JH metabolism in Bombyx
FOXO mutant larvae (9 -12 h after molting without feeding). Each larva was administered 2 l of methoprene, and the same amount of acetone was used as a control. Twenty individuals were treated for each group.

Hormone titer measurement
Hemolymph (50 -100 l) was collected from L4D2 and L4D3 WT and FOXO mutant silkworms into tubes containing 5-10 l of PTU (phenoloxidase inhibitor). The mixture was centrifuged, and the supernatant was recovered and transferred to a new tube for subsequent experiments.

20E titer measurement
The extracts were vacuum dried, re-dissolved in EIA buffer (Cayman), and subjected to enzyme immunoassay using the EIA kit (Cayman) to estimate the ecdysteroids titer as previously described (40,41). The 20-hydroxyecdysone EIA antiserum, acetylcholinesterase (AchE)-linked 20E, and standard 20E (Sigma) were applied in competitive assays to determine 20E. The AchE activity was quantified using Ellman's reagent and measured at 410 nm using a Multiskan FC microplate photometer (Thermo).

Determination of JH titers
The concentration of JHI in the hemolymph was determined using LC-MS after derivatization as described by Furuta et al. (42). The minimum detection level of JH in the hemolymph was 0.5 pg/l. Briefly, screw-top tubes containing 10 l of internal standard (D 3 -JHIII in toluene (50 pg/ml) was mixed with 1.5 ml of MeOH. To the internal standard solution, 50 l of hemolymph was added and vortexed and mixed with 1.5 ml of 2% NaCl solution. The JHs were extracted three times with 0.5 ml of n-hexane and purified using a mini-column packed with aluminum oxide (activity III) to elute with 50% diethyl ether in n-hexane. After derivatization with MeOH to obtain methoxyhydrins of JH (JH-MHs), further purifications of the derivatives were performed using a aluminum oxide mini-column as describe above to elute with 50% ethyl acetate in n-hexane. The solvent was removed by a nitrogen gas stream of the traction, including the JH-MHs, and dissolved in acetonitrile. The HP1100MSD system was equipped with a 150 ϫ 3-mm C 18 reverse phase column (UG80; Shiseido) protected by a guard column using acetonitrile/distilled water (7:3, v/v) supplemented with 5 M sodium acetate as the mobile phase at a flow rate of 0.4 ml/min. The selected ion groups (JHIII-MHϩNa ϩ as m/z 321.1 or 324.1 (D 3 -internal standard), JHII-MHϩNa ϩ as m/z 335.1 and JHII-MHϩNa ϩ as m/z 349.1) were monitored. All determinations were performed in triplicate.

JHE enzymatic activity determination
JHE activity was determined using the method reported by McCutchen et al. (43) and described in detail by Tsubota et al. (21) using the substrate methyl 1-heptylthioacetothioate. For the assay, 5 l of hemolymph collected from each larvae was added to 293 l of assay buffer, i.e. 0.05 M sodium phosphate (pH 7.4) containing 0.025% of 5,5Ј-dithiobis(2-nitrobenzoic acid) (Nacalai Tesque) and 10% sucrose and preincubated at 25°C for 10 min. Subsequently, 2 l of 30 mM methyl 1-heptyl-thioacetothioate in EtOH was added, and the changes in absorbance at 405 nm were recorded on a microplate reader (Spectra-Max, Molecular Devices, Sunnyvale) for 5 min. The enzymatic activity was determined in triplicate.

RNA-seq analysis
Total RNA from fat bodies at L4D2 in WT and FOXO mutants was extracted and treated with DNase I to avoid genomic DNA contamination followed by purification using phenol/chloroform and dissolved in UltraPure distilled water (Invitrogen). The integrity of total RNA was evaluated using agarose electrophoresis. The RNA samples with high integrity were sent to BGI for RNA seq. The number and annotated information for the genes differentially expressed in WT and FOXO mutants were identified. The KEGG pathway and GO analyses were applied to summarize the pathways and processes showing gene expression changes.

Immunoblot analysis
Fat body homogenates from L4D3 of WT and FOXO-M larvae were dissolved in SDS-lysis buffer and centrifuged, supernatant were collected. The concentration of total protein was measured by BCA Protein Assay Kit (Thermo). Then, 30 g of proteins were resolved by 12.5% SDS-PAGE and transferred to a PVDF membrane. The FOXO protein was detected with rabbit anti-FOXO primary antibody (1:1000 dilution, Youke) and alkaline phosphatase-conjugated goat anti-rabbit IgG (1:10,000 dilution, Youke) was applied as the secondary antibody. The ECL Plus Western blotting Detection Kit (GE Healthcare) was performed to detect the protein signal. The ␤-actin protein was treated as control.