Insulin and 20-hydroxyecdysone oppose each other in the regulation of phosphoinositide-dependent kinase-1 expression during insect pupation

Insulin promotes larval growth of insects by stimulating the synthesis of the steroid hormone 20-hydroxyecdysone (20E), which induces pupation and apoptosis. However, the mechanism underlying the coordinate regulation of insect pupation and apoptosis by these two functionally opposing hormones is still unclear. Here, using the lepidopteran insect and serious agricultural pest Helicoverpa armigera (cotton bollworm) as a model, we report that phosphoinositide-dependent kinase-1 (PDK1) and forkhead box O (FoxO) play key roles in these processes. We found that the transcript levels of the PDK1 gene are increased during the larval feeding stages. Moreover, PDK1 expression was increased by insulin, but repressed by 20E. dsRNA-mediated PDK1 knockdown in the H. armigera larvae delayed pupation and resulted in small pupae and also decreased Akt/protein kinase B expression and increased FoxO expression. Furthermore, the PDK1 knockdown blocked midgut remodeling and decreased 20E levels in the larvae. Of note, injecting larvae with 20E overcame the effect of the PDK1 knockdown and restored midgut remodeling. FoxO overexpression in an H. armigera epidermal cell line (HaEpi) did not induce apoptosis, but promoted autophagy and repressed cell proliferation. These results reveal cross-talk between insulin and 20E and that both hormones oppose each other's activities in the regulation of insect pupation and apoptosis by controlling PDK1 expression and, in turn, FoxO expression. We conclude that sufficiently high 20E levels are a key factor for inducing apoptosis during insect pupation.

The insulin/insulin-like growth factor (IGF) 3 signaling (IIS) pathway is a conserved pathway in animals (1). Mammalian insulin is produced in the islets of Langerhans in the pancreas, and insect IGFs are produced in the brain and various tissues. Although the production organs for mammalian insulin and insect IGFs are different, the structures of insulin and IGFs are similar and consist of two peptide chains, namely the A and B chains (2). Bombyxin is the first insulin-like peptide identified in Bombyx mori (3). In insects, up to 38 insulin-like peptides have been identified in each species (4). The function of the IIS pathway is to promote growth (5,6). In mammals, the IIS pathway regulates the metabolism of carbohydrates, fats, and proteins by promoting the absorption of glucose from the blood into the liver and fat and skeletal muscle cells (7). In Drosophila, the IIS pathway promotes cell growth and proliferation (5) (8). The effect of the IIS on metabolism of insects is little known.
Phosphoinositide-dependent kinase-1 (PDK1) is a central mediator in various pathways, including the insulin pathway, by phosphorylating many AGC kinases, such as protein kinase B (Akt/PKB), p70 ribosomal S6 kinase, p90 ribosomal S6 kinase, protein kinase N, and all protein kinase Cs (9). In the insulin pathway, PDK1 phosphorylates Akt/PKB (10), which then phosphorylates the transcription factor forkhead box O (FoxO). The phosphorylated FoxO is located in the cytoplasm and loses its transcription activity in the nucleus (11,12). Thus, insulin can promote cell proliferation and repress apoptosis (13). Repression of the IIS pathway influences the morphology of Drosophila (14). Repression of insulin/TOR signaling in Drosophila suppresses larval growth and delays pupation time (15). Overexpression of PDK1 in Drosophila increases cell and organ size (9,14). However, there is limited information on the involvement and mechanism of PDK1 during pupation in lepidopteran insects.
In insects, pupation is promoted by 20-hydroxyecdysone (20E), which is a steroid hormone that promotes insect molting and metamorphosis (16). 20E combines with its nuclear receptor EcRB1 to form the EcRB1-USP1 heterodimeric transcription complex (17,18), which regulates the expression of genes in the 20E pathway, including hormone receptor 3 (HR3) (19) and broad (Br) (20), that initiate metamorphosis and midgut remodeling. Midgut remodeling is controlled by 20E, and it includes imaginal midgut formation after the proliferation of regenerative cells and larval midgut programmed cell death (PCD). During larval midgut PCD, the color of the larval midgut changes to red, and it is finally degraded through PCD (21)(22)(23).
In B. mori, autophagy and apoptosis successively occur during larval midgut PCD (24). Insulin and 20E cross-talk to regulate insect development. Insulin promotes 20E production by promoting insect growth to critical body weight, and the prothoracic glands (PGs) grow larger to produce more 20E for metamorphosis (25,26). In Drosophila melanogaster, overexpression of FoxO in the PGs delays ecdysone biosynthesis and critical weight (27). Human insulin can induce Akt phosphorylation in Helicoverpa armigera, thereby inducing FoxO phosphorylation and cytoplasm localization. In contrast, 20E represses Akt phosphorylation, thereby repressing FoxO phosphorylation and inducing FoxO nuclear localization (28). Because PDK is a key protein kinase in the insulin pathway, we hypothesized that PDK is involved in the cross-talk between insulin and 20E.
Using H. armigera, the cotton bollworm, as a model, we identified PDK1 as a key factor for the coordinated regulation of insect midgut remodeling and pupation by insulin and 20E. Our results showed that PDK1 plays key roles in insect pupation by participating in the insulin pathway. Insulin and 20E cross-talk by counteractively regulating PDK1 mRNA levels to regulate insect pupation.

Insulin promoted PDK1 expression and 20E repressed PDK1 expression
To examine the involvement of PDK1 in insect development, the expression profile of PDK1 was examined (Fig. 1). The qRT-PCR results showed that PDK1 was expressed in the epidermis, fat body, and midgut. The transcript levels of PDK1 increased during the feeding stages of the sixth instar at 6 h to sixth instar at 24 h (6th-6h to 6th-24h) in the fat body, epidermis, and midgut and then decreased from the sixth instar at 48 h and were low during the metamorphic stages (6th-72h to 6th-120h) (Fig. 1A), indicating that PDK1 transcript levels are correlated with feeding.
HaEpi, a cell line established in our laboratory from the fifth instar larval epidermis (29), was used to examine the regulation of PDK1 mRNA levels by insulin and 20E. After insulin stimulation for 3 h, PDK1 transcripts were up-regulated in a dose-de-  (3-24 h). DMSO was used as the solvent control. F-I, hormonal regulation of the expression of PDK1 in the larval midgut. All of the experiments were performed in triplicate, and statistical analysis was conducted using Student's t test. Bars, mean Ϯ S.D. The asterisks indicate the significant differences when compared with PBS or DMSO: *, p Ͻ 0.05; **, p Ͻ 0.01.

Insulin and 20E cross-talk to regulate insect pupation
pendent manner from 2 to 5 g/ml (Fig. 1B). By stimulation with 2 g/ml insulin, PDK1 transcripts were up-regulated from 1 to 3 h gradually, and they then recovered to the base levels (Fig. 1C). These results suggested that insulin increases the mRNA levels of PDK1 in a dose-dependent manner in a short period.
Because the 20E titer is higher at metamorphic molting in lepidopteran insects (16), HaEpi cells were treated with exogenous 20E to examine the effects of regulation of 20E on PDK1 expression. We found that 5 M 20E decreased PDK1 transcript levels significantly in 6 h (Fig. 1D), and 5 M 20E consistently decreased PDK1 transcript levels significantly from 3 to 24 h (Fig. 1E). The in vivo results of the larval midgut also confirmed that insulin promoted PDK1 expression and 20E repressed PDK1 expression (Fig. 1, F-I). These results suggested that 20E consistently decreases the expression of PDK1 in a dose-dependent manner.

PDK1 determined pupation time and pupa weight
To study the functions of PDK1 during H. armigera growth and pupation, PDK1 was knocked down by injecting dsRNA against PDK1 into sixth-instar 6-h larvae. The larvae formed small pupae ( Fig. 2A). Statistical analysis showed no differences in the pupation rate after injection with dsPDK1 or dsGFP, with 75% survival and 25% death (Fig. 2B); however, the dsPDK1 injection prolonged pupal duration and delayed the initiation time of pupation for 34 h on average (recorded from sixth instar 0 h) (Fig. 2C). Moreover, the pupal weight decreased to an average of 0.16 g, compared with 0.32 g of the dsGFP injection control (Fig. 2D). These results suggested that PDK1 is critical for pupation.
To understand how PDK1 knockdown delayed pupation, the downstream genes in the insulin pathway were detected. The qRT-PCR assay showed that the expression of Akt decreased and that of FoxO increased after knockdown of PDK1 in the midgut (Fig. 2E). Therefore, PDK1 affects the transcript levels of Akt and FoxO.

Knockdown of PDK1 in the larvae repressed midgut remodeling and decreased 20E titer
To examine the effects of PDK1 knockdown on midgut remodeling, we dissected the larvae 90 h after dsRNA injection of sixth-instar 6 h larvae (6th-6h larvae). All midguts of the dsGFP-injected larvae appeared in red color (23), which implies that the midgut was entering PCD (30); in contrast, all midguts of the dsPDK1-injected larvae appeared yellow, indicating that the midgut did not enter PCD (Fig. 3A). Hematoxylin-eosin   (HE) staining showed that the larval midgut separated from the imaginal midgut after dsGFP injection, indicating the occurrence of midgut remodeling. In contrast, the imaginal midgut did not form after dsPDK1 injection (Fig. 3B), confirming that PDK1 knockdown repressed midgut remodeling.

Insulin and 20E cross-talk to regulate insect pupation
The detailed structure of the midgut was observed using transmission EM (TEM) after PDK1 knockdown. The number of typical autophagosomes, which are two-membraned and contain degenerating cytoplasmic organelles or cytosol (31), increased in the midgut after PDK1 knockdown in the midgut when compared with the dsGFP control (Fig. 3, C and D), suggesting that PDK1 knockdown arrested the midgut at the autophagic stage.
Because 20E is the key factor for promotion of midgut remodeling, the 20E titer was suspected to be insufficient on the basis of decreased body weight after PDK1 knockdown. Therefore, we determined the 20E titer in the whole larvae 90 h after injection with dsRNA and found that the 20E titer had decreased significantly after injection with dsPDK1 (Fig. 3E). These results suggested that knockdown of PDK1 decreases the 20E titer, thereby repressing midgut remodeling.

Complementation of 20E rescued the effect of PDK1 knockdown
To prove that knockdown of PDK1 repressed apoptosis because of the low titer of 20E, 20E (500 ng/larva) was injected into sixth-instar 6-h larvae after injection with dsPDK1 for 48 h. When compared with the DMSO control, the larval midgut appeared red (23, 32) (Fig. 4A), and the larval midgut separated from the imaginal midgut ( Fig. 4B), which indicated that complementation of 20E recovered midgut remodeling when PDK1 was knocked down.
PDK1 was knocked down in HaEpi to confirm that it caused autophagy, as observed in the midgut. The autophagy-related protein, microtubule-associated protein 1 light chain 3 (LC3, one isoform of ATG8) (33), was overexpressed as a fusion protein with the cell-penetrating TAT peptide and red fluorescent protein (RFP) (His-TAT-RFP-LC3-His) to detect autophagosomes in HaEpi (34). The TAT peptide helped the fusion protein His-TAT-RFP-LC3-His to enter the cells (35), RFP showed red fluorescence, and LC3 protein indicated the puncta of autophagosomes by its location in the autophagosome membrane (31). His-TAT-RFP-LC3-His indicated that LC3 formed puncta after PDK1 was knocked down in dsPDK1 ϩ DMSOtreated cells, but 5 M 20E blocked the formation of the LC3 puncta significantly in dsPDK1 ϩ 20E-treated cells (Fig. 4, C and D).
In contrast, caspase-3 activity was detected in few dsPDK1 ϩ DMSO-treated cells, but caspase-3 activity was detected in 85% of dsPDK1 ϩ 20E-treated cells (Fig. 4, E and F). The efficiency of interference of PDK1 in HaEpi was significant (Fig.  4G). These results suggested that PDK1 interference indeed induces autophagy but could not induce apoptosis, and 20E is the key factor for inducing apoptosis.

Overexpression of FoxO induced autophagy but not apoptosis
We observed that PDK1 knockdown caused an increase in FoxO mRNA levels in the larval midgut; however, the midgut was arrested at the autophagic stage, which implied that FoxO could induce autophagy but not apoptosis without sufficient 20E. To confirm that FoxO induced autophagy, we detected the transformation of LC3-I to LC3-II (phosphatidylethanolamine form) by using Western blotting, which indicates autophagy (36). FoxO-GFP was confirmed to be overexpressed in the HaEpi cells (Fig. 5A). Overexpression of FoxO-GFP increased the LC3-II level, whereas the addition of 5 M 20E for 72 h decreased the LC3-II level significantly (Fig. 5, B and C).
RFP-LC3 was overexpressed in HaEpi to indicate the autophagosomes to further confirm that overexpression of FoxO induces autophagy. RFP-LC3 is a fluorescent protein that can reveal autophagosomes (34,37,38). The number of autophagosomes increased after overexpression of FoxO-GFP-His; however, the addition of 20E for 72 h decreased the autophagosomes (Fig. 5, D and E). The negative control by the RFP tag had fewer autophagosomes (Fig. 5F). These results confirmed that FoxO promotes autophagy, but treatment with 5 M 20E for 72 h caused the disappearance of autophagosomes.

Insulin and 20E cross-talk to regulate insect pupation
FoxO-His was overexpressed in HaEpi ( Fig. 6A) to evaluate the role of FoxO in apoptosis. Caspase-3 activity was detected in 18% of His tag overexpression cells after the addition of 20E, with DMSO treatment as a solvent control, suggesting 20Einduced apoptosis. Similarly, caspase-3 activity was detected in 18% cells in FoxO-His-overexpressing cells after the addition of 20E (Fig. 6, B and C). Flow cytometry analysis further confirmed that His tag overexpression and FoxO-His overexpression induced 3-4% cell apoptosis, but 20E supplementation induced 18% cell apoptosis (Fig. 6, D and E). These results confirmed that overexpression of FoxO alone cannot induce apoptosis.

Overexpression of FoxO repressed cell proliferation
PDK1 knockdown in the larvae produced small pupae and repressed the formation of imaginal midgut cells, which was accompanied by an increase in FoxO mRNA levels. Thus, FoxO was hypothesized to be involved in the repression of cell proliferation. To verify the assumption, cell proliferation was detected by 5-ethynyl-2Ј-deoxyuridine (EdU) staining after overexpression of FoxO-GFP by using the pIEx-FoxO-GFP-His plasmid in HaEpi. The results showed that the proliferative sig-nal EdU was detected in the nucleus of GFP-overexpressing cells when compared with the 20E-treated cells (5 M for 72 h), suggesting that the GFP-overexpressing cells proliferate normally and 20E represses cell proliferation. However, a very low EdU signal was detected in the FoxO-GFP-overexpressing cells, regardless of DMSO treatment or 5 M 20E treatment (Fig. 7, A and B); this suggested that FoxO overexpression results in an effect similar to that after 20E treatment to repress cell proliferation.

PDK1 knockdown inhibits insulin-induced Akt and FoxO phosphorylation
To verify the function of PDK1 in the insulin pathway in H. armigera, we analyzed the involvement of PDK1 in insulininduced Akt and FoxO phosphorylation in HaEpi by overexpression of Akt fused with RFP-His and FoxO fused with GFP-His in the cells, respectively. Western blotting showed insulin-induced phosphorylation of both Akt and FoxO. However, knockdown of PDK1 blocked insulin-induced Akt and FoxO phosphorylation significantly (Fig. 8, A (a) and B (b)). These results confirmed that insulin, via PDK1, induced Akt and FoxO phosphorylation in H. armigera. Moreover, insulin

Insulin and 20E cross-talk to regulate insect pupation
induced FoxO translocation to the cytosol from the nucleus by GFP co-fluorescence indication. However, knockdown of PDK1 blocked insulin-induced FoxO translocation to the cytosol (Fig. 8C). These results confirmed that PDK1 is involved in the insulin pathway in H. armigera.

Discussion
IGF intensifies the cell proliferation and inhibition of cell apoptosis via the PI3K/Akt signaling pathway (39), whereas the steroid hormone 20E promotes apoptosis during insect metamorphosis (40). It is intriguing how these functionally counteractive hormones regulate insect pupation and apoptosis coordinately. Our study showed that PDK1 plays a critical role in the cross-talk between insulin and the steroid hormone 20E. PDK1 is necessary for 20E synthesis. PDK1 functions as a repressor of FoxO. 20E represses PDK1 expression to release FoxO expression for autophagy, and 20E finally promotes apoptosis.

PDK1 is necessary for 20E synthesis to regulate insect metamorphosis and apoptosis
Repression of insulin and reduced ecdysone levels have similar effects on larval pupation (41). Insulin promotes insect body growth and PG growth, thereby promoting 20E synthesis (25). The 20E titer is lower than 0.5 g/ml during the feeding stage, but it increases to about 5 g/ml in the hemolymph during the metamorphic stage of Antheraea mylitta (42). In this study, we found that repression of the insulin pathway via knockdown of PDK1 in the larvae resulted in a decrease in the 20E titer from 1.5 to 0.1 g/g, suggesting that insulin via PDK1 regulates the increase in 20E levels.
Larval midgut PCD has been proven to be apoptosismediated during metamorphosis in the lepidopteran insects Bombyx mori (43) and H. armigera (30). Autophagy was observed before apoptosis in the midgut of B. mori (44,45). In H. armigera, a high concentration of 20E promotes apoptosis, but a low concentration of 20E induces autophagy; autophagy is switched to apoptosis when a high concentration of 20E triggers Ca 2ϩ influx (34). We found that the decreased 20E titer after PDK1 knockdown could not trigger midgut remodeling and pupation. By complementation of 20E in the larvae, midgut remodeling was recovered, confirming that 20E is the critical factor that promotes midgut remodeling. Our study confirmed that the insulin and 20E interplay for regulating metamorphosis is a common mechanism in insects.

PDK1 plays a significant role in repressing FoxO mRNA levels
FoxO is a transcription factor that regulates diverse processes, such as energy homeostasis, apoptosis, and cellular dif-

Insulin and 20E cross-talk to regulate insect pupation
ferentiation, and life span. FoxO is phosphorylated by Akt and arrested in the cytoplasm to lose its transcriptional function under insulin regulation (13). The phosphorylation of Akt and FoxO in H. armigera is induced by insulin and repressed by 20E in HaEpi, and the nonphosphorylated FoxO is localized in the nucleus (28). However, little is known about the transcriptional regulation of FoxO, compared with its post-translational modifications (46). Previous studies have reported that 20E represses FoxO phosphorylation and promotes FoxO nuclear translocation in Drosophila (41) and Bombyx (47). In H. armigera, 20E via its nuclear receptor EcRB1 promotes FoxO expression (28). In this study, we found that PDK1 knockdown caused an increase in FoxO mRNA, which suggests that PDK1 is a repressor of FoxO. PDK1 plays a key role in the insulin pathway by regulating Akt phosphorylation (10); thus, FoxO transcription is possibly repressed by the insulin pathway.
FoxO regulates the life span of Drosophila (48,49). We found that PDK1 knockdown increased FoxO mRNA levels, elongated the larval stage, and delayed pupation time, suggesting that FoxO plays a role in the elongated larval life span. Further analyses of the HaEpi cells showed that the overexpressed FoxO induced autophagy. This phenomenon is correlated to the finding that FoxO promotes autophagy-related gene expression in animals (50). However, overexpressed FoxO did not induce apoptosis. In addition, we found that, although PDK1 knockdown decreased Akt expression and increased FoxO expression, midgut PCD was not induced when 20E was insufficient. Therefore, apoptosis is controlled by 20E in a FoxO-independent manner, which is inconsistent with the finding that FoxO induces apoptosis in humans (51, 52). This may be because 20E is necessary for inducing high cellular Ca 2ϩ levels for apoptosis (53) and the transformation of autophagy to apoptosis in insects (34).

20E via repressing PDK1 mRNA levels antagonizes the insulin pathway
Insulin and 20E had an antagonistic effect on the regulation of glycogen mobilization and showed a similar effect on carbo-  for the regulation of pupation by insulin and 20E via counteractive regulation of PDK1 expression. 1, insulin up-regulates PDK1 expression, which induces Akt phosphorylation. Akt induces FoxO phosphorylation and cytosol localization to allow cell proliferation and high titer of 20E production. 2, high 20E titer represses PDK1 mRNA levels by an unknown negative feedback mechanism and therefore represses Akt and FoxO phosphorylation, resulting in FoxO nuclear localization. 3, FoxO in the nucleus induces autophagy and represses cell proliferation. 4, 20E promotes autophagy transformed to apoptosis in the midgut during metamorphosis by increasing cellular calcium (34,53,66). High 20E titer functions as a switch between growth and metamorphosis. Akt-P, phosphorylated Akt. FoxO-P, phosphorylated FoxO. PCD, programmed cell death. *, p Ͻ 0.05; **, p Ͻ 0.01.

Insulin and 20E cross-talk to regulate insect pupation
hydrate reserves in B. mori (54). A lower 20E concentration could not repress insulin function (30), and a higher 20E concentration is critical for antagonizing insulin function in H. armigera (55). The ecdysone counteracts the growth-promoting action of insulin by mediating nuclear localization of dFOXO in Drosophila (41). 20E promotes FoxO expression and represses FoxO phosphorylation by up-regulating PTEN (phosphatase and tensin homolog) expression, which represses Akt and FoxO phosphorylation in H. armigera (28). Here, we showed that 20E counteracts the insulin pathway by repressing PDK1 expression in a dose-dependent manner. 20E repressed the insulin pathway via a negative feedback mechanism when it was increased to a critical titer by insulin. This is consistent with the finding that increased 20E levels negatively repress insulin function (56). However, the mechanism that 20E represses PDK1 expression needs further study in future work.

Conclusions
PDK1 was highly expressed during the larval feeding stage to promote cell proliferation and larval growth under insulin regulation. 20E titer was increased along with larval growth and therefore repressed PDK1 expression in a dose-dependent manner. PDK1 repressed FoxO expression. FoxO induced autophagy but not apoptosis under insufficient 20E titer conditions. A sufficient 20E titer is the key factor for triggering metamorphosis and midgut apoptosis (Fig. 8D).

Experimental animals
H. armigera specimens were raised on an artificial diet, according to a previous study (57), in our laboratory. The larvae were maintained at 27°C with 60 -70% relative humidity and conditions of 14-h light and 10-h dark.

HaEpi culture
HaEpi was established in our laboratory from the fifth-instar larval epidermis (29). The cell line has been used to study the hormonal pathways of 20E, JH, and insulin by RNAi or overexpression (28). The cells were cultured in cell culture flasks with a monolayer at the bottom and maintained at 27°C. In every tissue culture flask, 4 ml of Grace's medium supplemented with 10% heat-inactivated FBS was added, and subculture was performed once a week.

RNAi in the larvae and HaEpi cell line
dsRNA was prepared using PDK1 cDNA as the template (1719 bp) with the MEGAscript TM RNAi kit (Ambion, Austin, TX) and primers (PDK1RNAiF and PDK1RNAiR sequences in Table 1). A single long dsRNA was used for H. armigera and produced good results in our previous study (58,59) and other lepidopteran insects (60). Thus, a long dsRNA was used (870 bp from bp 352 to 1222) for the experiments. The long dsRNA was cleaved into multiple siRNAs in the cells by intracellular RNase to target multiple sites of the gene (61,62). We checked the RNAi ratio by using the primers PDK1-RTF and PDK1-RTR (Table 1). For the RNAi experiment using the larvae, dsRNA was diluted to 400 ng/l with PBS (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 ), and 5 l was injected into the hemocoel of the sixth-instar 6-h larvae. The control larvae were injected with 2 g of dsGFP. In every experiment, 30 larvae were injected with dsRNA, and the injection was repeated twice over a 24-h interval. HaEpi was cultured at 27°C in Grace's medium supplemented with 10% FBS to almost 80% density. Then the cells were incubated in 2 g of dsRNA and 10 l of Lipofectamine in 1 ml of Grace's medium supplemented with 10% FBS after 6 h.

Hormone treatment for the larvae and HaEpi cell line
The stock solution for 20E (10 mg/ml, about 20 mM in DMSO) was diluted 100-fold with PBS (10 mM Na 2 HPO 4 , 1.8

Insulin and 20E cross-talk to regulate insect pupation
mM KH 2 PO 4 , 140 mM NaCl, and 2.7 mM KCl, pH 7.4) and injected into sixth-instar 6-h larvae (100 -500 ng/larva). The stock solution for human insulin (10 mg/ml) was diluted 10-fold with PBS and injected into sixth-instar 6-h (6th-6h) larvae (0.5-5 g/larva). HaEpi was cultured to ϳ80% density, and 20E or stock solution of human insulin (10 mg/ml in PBS) was added at various concentrations or intervals. The control larvae or HaEpi cells were treated with the same amount of DMSO as the solvent control.

Detection of 20E levels
The larvae were frozen in liquid nitrogen for 5 min and homogenized sufficiently in 1 ml of 80% methyl alcohol and centrifuged at 12,000 ϫ g and 4°C for 15 min. The supernatant was air-dried. Then the 20-hydroxyecdysone enzyme immunoassay kit (Bertin Pharma, Montigny-le-Bretonneux, France) was used to measure the 20E titer, according to the manufacturer's instructions.

qRT-PCR
Total RNA was acquired from the larvae using TransZol reagent (Transgen, Beijing, China). Total RNA was transcribed into first-strand cDNA as the midgut template for qRT-PCR. Then 10 l of the reaction mixture, containing 5 l of SsoFast EvaGreen Supermix (Bio-Rad), 2 l of 1 mol/liter forward primer, 2 l of 1 mol/liter reverse primer, and 1 l of cDNA (1:6 dilution), was used. ␤-Actin was used for quality control (59). The data were analyzed using the formula, r ϭ 2 Ϫ(⌬Ct sample Ϫ ⌬Ct control) , where R is the relative transcriptional level, ⌬Ct sample is the difference between the Ct of the gene and average ␤-actin in the experimental sample, and ⌬Ct control is the difference between the Ct of the gene and average ␤-actin in the control sample (63).

Hematoxylin-eosin (HE) staining and TEM
The waxing and dewaxing of the samples and HE staining were performed as described previously (30). The section was washed with running water for 1 min and stained with Scott's liquid for 1 min. Then the section was washed with hydrochloric acid ethanol differentiation medium (70% ethanol in 1% hydrochloric acid) for 20 s, stained with Scott's liquid for 1 min, incubated with 0.5% water-soluble eosin dye solution for 30 s, and then washed with running water. Finally, the sections were sealed with 80% glycerin and observed using an Olympus BX51 fluorescence microscope (Tokyo, Japan). TEM was performed by the Servicebio Company (Wuhan, China) after the larvae were treated in the various experiments.

Overexpression of proteins in the HaEpi cells
FoxO-GFP-His, FoxO-His, GFP-His, or His tag proteins constructed in our laboratory (28) were overexpressed with 4 g of pIEx-4 plasmid (Novagen) and 10 l of QuickShuttle-enhanced transfection reagent (Biodragon Immunotechnologies Co., Ltd., Beijing, China) for 48 h in 1 ml of Grace's medium, according to the manufacturer's instructions and our previous study (34).

Western blotting
The cell homogenates were separated using 12.5% SDS-PAGE. Equal amounts of each sample (50 g) was subjected to SDS-PAGE. The proteins were transferred to the nitrocellulose membrane (0.45 m) by electrotransfer. The membrane was blocked with the blocking buffer (2% nonfat milk in PBS) for 1 h at room temperature. The membrane was then incubated with the primary antibody (diluted 1:5000 in the blocking buffer) against GFP or His tags (Zhong Shan Jin Qiao, Beijing, China) overnight at 4°C and with the secondary antibody for alkaline phosphatase (Zhong Shan Jin Qiao) conjugated with horse antimouse IgG diluted to 1:10,000 in the blocking buffer. The protein signal was visualized using 45 l of nitro blue tetrazolium (75 mg/ml) and 35 l of 5-bromo-4-chloro-3-indolylphosphate (50 mg/ml; Sigma) in 10 ml of TBS in the dark at room temperature. The protein bands on the membrane were analyzed using Quantity One software (Bio-Rad).

Detection of apoptosis
An aspartic acid proteinase called caspase-3 is a highly conserved protein that can form active caspase-3 during apoptosis (65). Thus, caspase-3 activity can be used to indicate apoptosis. The NucView TM caspase-3 assay kit (catalog no. 30029, Biotium, Fremont, CA) was used to detect caspase-3 activity in the HaEpi cells, according to the manufacturer's instructions. Annexin V and propidium iodide (PI) were used to detect apoptosis by using the Annexin V-FITC apoptosis detection kit (GK3603, Genview) and flow cytometry (Amnis). Annexin V-FITC stained the earlier apoptotic cells by binding to the membrane phosphatidylserine, and PI indicated the later apoptotic cells and dead cells by entering the cells.

Detection of cell proliferation
The EdU kit (Ribobio, Guangzhou, China) was used to detect cell proliferation levels, according to the manufacturer's protocol. FoxO-GFP and GFP tags were overexpressed by the pIEx-4-FoxO-GFP-His and pIEx-4-GFP-His plasmids, respectively, in the HaEpi cells for 24 h. Then the cells were treated with 5 M 20E for 72 h. The same volume of DMSO was used as the control.

Statistical analysis
Student's t test was used for statistical analysis of the paired data. The bars represent mean Ϯ S.D. values of three independent biological experiments. Statistical significance was determined using two-tailed paired Student's t test. Differences were considered statistically significant at p Ͻ 0.05 (*) and p Ͻ 0.01 Insulin and 20E cross-talk to regulate insect pupation (**). GraphPad Prism 5 (version 5.01) was used to produce the figures.