Nutritional control of body size through FoxO-Ultraspiracle mediated ecdysone biosynthesis

1Environmental conditions mould the developmental programs of many organisms to produce dramatic differences in body size and shape, in developmental time and in pigmentation patterns. In insects, environmental cues often mediate their effects by regulating the timing and amount of hormone biosynthesis at specific points in development. These changes in hormone production have been associated with a wide variety of environmentally induced changes in morphology, including the dramatic reshaping of the body in honeybee castes and seasonal wing pattern polyphenisms in butterflies. Understanding the molecular underpinnings through which environmental conditions modify hormone production would provide valuable insight into our understanding of developmental plasticity.

2Larvae of the fruit fly, Drosophila melanogaster, provide a tractable model to address this question. Drosophila larvae regulate their body size and developmental timing in response to nutritional conditions, similar to many other animals. Early in the third (final) larval instar (L3), a small peak of the steroid hormone ecdysone has been proposed to induce a developmental transition known as critical weight. The critical weight ecdysone peak responds to both environmental cues and internal developmental processes. Environmental cues including nutrition, temperature and oxygen levels affect the timing of the critical weight ecdysone peak. In addition, the neuropeptide important for inducing all ecdysone peaks, prothoracicotropic hormone (PTTH), stimulates ecdysone biosynthesis at critical weight. The combination of environmental and developmental regulation of this ecdysone peak ensures that developmental timing can be altered with changes in environmental conditions.

3Critical weight itself determines the duration of the growth period, and therefore final body size, in response to environmental conditions including nutrition. Before larvae reach critical weight, starvation delays the onset of metamorphosis. After critical weight, larvae initiate metamorphosis without any developmental delay even when starved. Feeding ecdysone to larvae with genetically-induced delays in critical weight rescues the timing of the onset of metamorphosis.

4Nutrition regulates size in organisms ranging from flies to humans via the insulin/insulin-like growth factor signaling (IIS)/Target of Rapamycin (TOR) signaling pathway. The IIS/TOR pathway controls body size by regulating growth rate, and also by regulating the timing of critical weight to determine the duration of the growth period. At critical weight, the IIS/TOR pathway acts directly on the glands that synthesize ecdysone, the prothoracic glands (PGs), to alter the timing of the expression of several cytochrome P450 (CYP450) genes necessary for ecdysone biosynthesis. Increasing IIS/TOR activity in the PGs causes precocious ecdysone biosynthesis, precocious critical weight transitions, precocious metamorphosis and dramatic reductions in body size. Reducing IIS/TOR activity in the PGs induces the opposite effects. However, the mechanisms through which the IIS/TOR pathway mediates these effects have been unclear.

5Under well-fed conditions, insulin-like peptides (ILPs) are secreted into the insect blood or hemolymph. By binding to the Insulin Receptor (InR), ILPs activate IIS/TOR signaling in the target tissues. Activating IIS/TOR signaling regulates a series of phosphokinases, including Akt. Akt, in turn, phosphorylates a negative regulator of growth, Forkhead Box class O transcription factor (FoxO), displacing it from the nucleus to the cytoplasm. In starved larvae, FoxO localizes in the nucleus where it acts on its targets, such as 4E-binding protein (4E-BP, also known as Thor), to suppress cell growth and division.

6In mammalian cells, FoxO binds to several nuclear hormone receptors (NHRs), such as constitutive androstane receptor (CAR) and pregnane X receptor (PXR), to regulate CYP450 expression. The functional ecdysone receptor is composed of two NHRs, Ecdysone Receptor (EcR) and Ultraspiracle (Usp). Since many CYP450 enzymes are involved in ecdysone biosynthesis, this led us to the hypothesis that the effect of IIS/TOR signaling on ecdysone biosynthesis is mediated by the interaction between FoxO and either EcR or Usp.

7Here, we provide definitive evidence that critical weight results from the small nutrition-sensitive ecdysone peak early in the L3. Further, we report that IIS/TOR regulates the timing of ecdysone biosynthesis at critical weight via a novel mechanism, the direct association of FoxO and Usp. With these findings, we have constructed a detailed model of the molecular mechanisms underlying environmentally-sensitive ecdysone biosynthesis during critical weight, an event that ultimately determines the duration of the growth period and accordingly final body size.

8Previous studies have shown that activating IIS/TOR signaling in the PGs induces early critical weight transitions, precocious ecdysone biosynthesis at wandering, and precocious metamorphosis. This has led authors to propose that the small pulse of ecdysone early in the L3 is nutrition-sensitive and induces critical weight in Drosophila. However, these studies have not measured ecdysone concentrations with sufficient resolution early in the instar to show that ecdysone biosynthesis was delayed in starved pre-critical weight larvae. Therefore, we first examined whether this early ecdysone peak is delayed in starved larvae. In accordance with our hypothesis, we found that the small ecdysone peak that occurs around 10 hr after L3 ecdysis (AL3E) in well-fed larvae is suppressed in starved larvae, at least until 18 hr AL3E (Figure 1A). Thus, the timing of this early peak is indeed sensitive to nutrition.

9Next we reasoned that if this early peak of ecdysone induced critical weight, feeding ecdysone to starved, pre-critical weight larvae should eliminate the delay in their development. To determine when wild type larvae reach critical weight, we starved carefully staged larvae of defined age classes on non-nutritive agar and measured the time it takes for them to reach pupariation from the onset of starvation. A hallmark of critical weight is that before it is attained starvation delays the onset of metamorphosis, whereas after critical weight larvae metamorphose early when starved. We estimate the age at critical weight using breakpoint analysis, which fits a bi-segmental linear regression to the relationship between age at starvation and time to pupariation, and calculates the age at critical weight as the inflection point where this relationship changes. We then use the linear relationship between larval weight and larval age to convert the age at which larvae reach critical weight to the size at which larvae reach critical weight (Figure 1—figure supplement 1B). Finally, we repeated the analysis on 1000 bootstrap datasets to generate 95% confidence intervals for the age and size of larvae when they reach critical weight. Data and scripts for the analysis of size and age at critical weight, including the growth rate data, for all genotypes and treatments are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.75940.

10Wild type larvae reached critical weight at 8.66 hr AL3E (Figure 1B, Supplementary file 1), correlating with the time when well-fed, wild type larvae show a peak of ecdysone (Figure 1A). When we added the active form of ecdysone, 20-hydroxyecdysone (20E), to the medium even the youngest larvae no longer delayed their onset of metamorphosis when starved (Figure 1B). Instead, larvae starved on 20E-supplemented agar between the ages 0–8 hr AL3E pupariated 32 hr after the onset of starvation (Supplementary file 1). Finally, larvae fed 20E-supplemented fly medium throughout the L3 were more than 25% smaller than control larvae (Figure 1—figure supplement 1A). These results demonstrate that this early peak of ecdysone is nutrition sensitive and that it induces critical weight.

11We next sought to understand how nutrition regulated the timing of the critical weight ecdysone peak. We hypothesized that IIS/TOR signaling controlled the timing of this ecdysone peak, and therefore critical weight, via FoxO. We reasoned that if FoxO was involved in regulating ecdysone biosynthesis, FoxO would be present in the PG nuclei immediately after the molt to the L3 and would become progressively excluded from the nucleus as the larvae fed and approached critical weight. We found that FoxO was localized primarily in the nuclei of the PG cells of newly ecdysed L3 larvae (0 hr AL3E) (Figure 2A). As the larvae fed, FoxO was gradually transported out of the nuclei into the cytoplasm. At 5 hr AL3E, FoxO appeared evenly dispersed inside the PG cells (Figure 2B). By 10 hr AL3E, immediately after critical weight (Figure 1B), FoxO was mostly localized in cytoplasm of fed larvae (Figure 2C,D). Thus, FoxO appears to be progressively transported out of the nucleus as larvae approached critical weight.

12For FoxO to regulate ecdysone biosynthesis in a nutrition-dependent manner, we would expect that it would remain in the nucleus in starved, pre-critical weight larvae. In larvae starved from 0–15 hr AL3E, FoxO remained in the nuclei of the PG cells (Figure 2E). In contrast, FoxO was found primarily in the cytoplasm in fed controls. Taken together, the localization of FoxO suggests that it could be involved in regulating ecdysone biosynthesis at critical weight.

13Since FoxO associates with NHRs to regulate CYP450 gene expression in mammalian cells, we hypothesized that FoxO could associate with either EcR or Usp to regulate the nutrition-sensitive ecdysone peak by regulating the expression of CYP450 ecdysone biosynthesis genes. Using GST-pulldown assays, we found that FoxO bound to Usp but not to EcR in vitro (Figure 2F). Co-immunoprecipitation experiments using larval extracts showed that FoxO bound to Usp only in pre-critical weight or starved larvae, but not in well-fed post-critical weight larvae (Figure 2G,H). FoxO neither bound to EcR, nor did it impede EcR/Usp binding in starved larvae (Figure 2G,H). This suggests that FoxO could interact with Usp to regulate the critical weight ecdysone peak, and further that this interaction is unlikely to interfere with EcR/Usp function.

14Because in vertebrates FoxO-NHR interactions sometimes change in the presence of hormones, we tested whether 20E altered FoxO/Usp binding or induced FoxO/EcR binding. The presence of 20E neither changed the binding properties of the FoxO–Usp interaction nor induced a FoxO–EcR association (Figure 2I).

15If FoxO/Usp interactions regulate ecdysone biosynthesis at critical weight, we would expect that altering the expression of FoxO or Usp in the PGs would change both the size and the age at which larvae reach critical weight. We used Phantom (Phm)-Gal4, a Gal4 driver specific for the PG cells, to overexpress FoxO in the PGs. These larvae attained critical weight at larger sizes and 10 hr later than in controls (Figure 3A). Overexpressing Usp in the PGs did not produce any significant difference in either the size or the age at which critical weight was achieved (Figure 3B). Overexpressing both FoxO and Usp in the PGs resulted in larvae that reached critical weight more than 13 hr later and about 1 mg larger than control larvae (Figure 3C). Further, the size of these larvae at critical weight was significantly larger than when either FoxO or Usp was overexpressed in the PGs alone (Figure 3D). These data suggest that both FoxO and Usp regulate the timing of critical weight.

16In contrast, knocking down either FoxO or Usp alone in the PGs reduced the size but not the age at critical weight (Figure 3E or Figure 3F, respectively). When we simultaneously knocked down both FoxO and Usp in the PGs, larvae reached critical weight significantly earlier at smaller sizes (Figure 3G) than knocking down either FoxO or Usp alone (Figure 3H). These knock down experiments corroborate our results from our FoxO and Usp overexpression experiments and provide further evidence that both FoxO and Usp suppress ecdysone biosynthesis.

17Because critical weight is a key determinant of final body size, we also weighed pharate adults as a proxy of final adult size. Overexpressing either FoxO or Usp in the PGs significantly increased body size compared to parental controls (Figure 3—figure supplement 1A). In addition, females that overexpressed both FoxO and Usp together in the PGs had significantly larger body sizes than those overexpressing either FoxO or Usp alone (Figure 3—figure supplement 1A, ANOVA interaction term, p = 0.045). Knocking down Usp resulted in a significant decrease in body size (Figure 3—figure supplement 1B). Knocking down FoxO caused a slight, but significant decrease in body size in males but not in females. However, knocking down both Usp and FoxO in the PGs dramatically reduced body size (Figure 3—figure supplement 1B,C). These results demonstrate that altering the size and timing of critical weight, by manipulating expression of FoxO and Usp, has definitive effects on final adult body size.

18Our data show that the earliest ecdysone peak in the L3 regulates critical weight and that FoxO and Usp alter the timing of this transition. To confirm that FoxO and Usp regulate critical weight by controlling the timing of ecdysone biosynthesis, we examined the expression of two CYP450 ecdysone biosynthetic genes, phm and disembodied (dib), known to be sensitive to IIS/TOR signaling, in larvae with altered FoxO and Usp expression. In addition, we quantified the expression of an ecdysone response gene, e74B (eip74ef isoform B), which tracks the early effects of ecdysone signaling in these larvae. In the parental controls, both phm and dib increased in expression around 8 hr AL3E, shortly before the critical weight ecdysone peak (Figure 4A,B,D,E). E74B expression peaks around 12 hr in parental controls, after the critical weight ecdysone peak (Figure 4C,F). When both FoxO and Usp were overexpressed in the PGs, the increase in phm and dib expression was delayed (Figure 4A,B) and e74B expression remained low up to 20 hr AL3E (Figure 4C). In contrast, when we knocked down FoxO and Usp, both phm and dib expression levels were high immediately after the molt to the L3 (Figure 4D,E) and e74B expression was nearly undetectable at ecdysis but increased rapidly thereafter (Figure 4F). Taken together, these results suggest that alterations in FoxO and Usp affect the timing of ecdysone biosynthesis at critical weight.

19Although our results suggest that both FoxO and Usp act in the PGs to regulate the timing of critical weight ecdysone peak, thereby mediating the timing of critical weight, they do not allow us to distinguish whether FoxO and Usp regulate ecdysone biosynthesis independently or together via the FoxO/Usp complex. To discern between these two possibilities, we developed a genetic tool to manipulate the FoxO–Usp interaction.

20First, we identified the Usp binding site in the FoxO protein using GST-pulldown assays. We created overlapping GST-tagged FoxO fragments and, using increasingly smaller overlapping fragments, we narrowed down the Usp binding region to a 35 amino acid region overlapping with 5 amino acids in the C-terminal end of the forkhead (DNA binding) domain (Figure 5A). This motif is well conserved across arthropod species including ticks and water fleas, but is not conserved in FoxO proteins in other ecdysozoans or vertebrates (Figure 5—figure supplement 1). Interestingly, this Usp binding motif is different from the well-known ‘LXXLL’-type NHR binding motif identified in vertebrates. Next, we identified eighteen candidate amino acids by comparing the crystal structure of mammalian FoxO3a to the Drosophila FoxO sequence and selecting residues that occupied positions permissive for protein–protein interactions. We mutated each of these to alanine. At least 4 of the 18 amino acid residues appeared to be involved in FoxO–Usp binding (residues W172, N175, R202 and K204). When we introduced these single point mutations into the full length FoxO protein, they showed only mild reductions in FoxO–Usp binding (Figure 5A). We then tested four double–mutant combinations (W172-R202, W172-K204, N175-R202, and N175-K204) all of which were sufficient to dramatically reduce FoxO–Usp interactions (Figure 5A). Two of these double–mutant combinations partially reduced the FoxO activity (W172-R202 and W172-K204) (Figure 5—figure supplement 2C,D), as determined by the expression of known FoxO targets InR and 4E-BP. Because our aim was to disrupt FoxO–Usp binding, but not FoxO function, these were excluded from further analyses. The remaining two double–mutant combinations (N175-R202 or FoxO NR, and N175-K204 or FoxO NK) showed normal translocation to the nucleus (Figure 5—figure supplement 2A,B) and did not affect FoxO's ability to regulate InR and 4E-BP promoter activities (Figure 5—figure supplement 2C,D, respectively).

21Finally, we tested whether FoxO NR and FoxO NK retained the ability to suppress tissue growth. We placed the wild type FoxO, FoxO NR and FoxO NK constructs under the control of a UAS promoter and inserted them into the fly genome, using targeted integration to control for positional effects of the transgenes. All three constructs have the same genetic background and differ only in the two amino acids mutated to interfere with FoxO–Usp binding. We then drove expression of the FoxO, FoxO NR and FoxO NK in the wings, using the C765- and MS1096-Gal4 drivers, and in the eyes, using the eyeless (Ey)- and GMR-Gal4 drivers. For all drivers, overexpression of FoxO and FoxO NK reduced organ size to a similar degree (Figure 5—figure supplement 3). FoxO NR showed milder reductions in tissue size (Figure 5—figure supplement 3), therefore in subsequent experiments we used FoxO NK. We confirmed that both FoxO and FoxO NK expressed FoxO protein at the same level (Figure 5B). Taken together, FoxO NK showed reduced FoxO–Usp affinity, but maintained Usp-independent FoxO function.

22To explore whether FoxO and Usp regulate critical weight independently or together as a complex, we drove expression of FoxO NK in the PGs of developing larvae and compared them with larvae expressing wild type FoxO in the PGs. Larvae that overexpressed FoxO NK in their PGs reached critical weight earlier and at smaller sizes than those that overexpressed wild type FoxO, albeit later than the parental controls (Figure 6A). Thus, impeding FoxO–Usp binding reduced the delay in critical weight induced by FoxO overexpression. Similarly, pupae in which FoxO NK was overexpressed in the PGs were significantly smaller than pupae that overexpressed wild type FoxO, although they were still larger than pupae from the parental controls (Figure 6—figure supplement 1A).

23Because FoxO alters developmental timing via its effects in other tissues, like the fat body, we eliminated effects of the endogenous gene by overexpressing FoxO in the PGs of FoxO null mutant larvae (FoxO Δ94/Df(3R)Exel8159). When we used the Phm-Gal4 driver to overexpress FoxO in the PGs of FoxO null animals, most larvae did not survive to the L3. To circumvent this problem, we used P0206-Gal4, which expresses Gal4 moderately in the PGs. Since P0206-Gal4 driver expresses Gal4 in other tissues such as the oenocytes and corpora allata, we tested the effects of overexpressing FoxO in these other tissues. To do this, we compared the duration of the L3 and final body size when overexpressing FoxO in the ring gland, oenocytes and corpora allata, using P0206-Gal4, in the oenocytes alone, using PromE(800)-Gal4, and in the corpora allata alone, using Aug21-Gal4 all in the FoxO null mutant background. We found that overexpressing FoxO in the oenocytes and corpora allata does not affect the duration of the L3. In contrast, overexpressing FoxO in the ring glands using P0206-Gal4 prolonged the duration of the L3 compared to parental controls (Figure 6—figure supplement 2A). Further, overexpressing FoxO in the oenocytes and corpora allata did not affect final body size, whereas overexpressing FoxO using P0206-Gal4 increased body size (Figure 6—figure supplement 2B). Taken together, our data suggest that FoxO overexpression in the oenocytes and corpora allata had no measurable effect on growth rate or the duration of the L3. Thus, we conclude that developmental delay and size increase induced by overexpressing FoxO using P0206-Gal4 is due to the functions of FoxO in the PGs.

24Similar to what we found in the FoxO wild type background, FoxO null larvae that overexpressed FoxO NK in their PGs reached critical weight earlier and at smaller sizes than larvae that overexpressed wild type FoxO in their PGs (Figure 6B). Further, we confirmed that overexpressing both FoxO NK and Usp in the PGs of FoxO null mutant larvae did not alter the age and size at critical weight (Figure 6—figure supplement 2C) nor did it alter final body size when compared to overexpressing FoxO NK alone (Figure 6—figure supplement 2D). These results suggest that Usp does not affect the timing of critical weight on its own and that critical weight is regulated, at least in part, by FoxO/Usp.

25We next tested whether exogenous ecdysone could rescue the delay in critical weight induced by the FoxO/Usp complex. To do this, we assessed critical weight in P0206>FoxO, FoxO null larvae and the P0206>+, FoxO null parental controls on medium supplemented with 20E. We found that adding 20E to the medium altered the relationship between age at starvation and time to the onset of metamorphosis in both genotypes (Figure 6D). In the parental controls, larvae starved before 12 hr AL3E on ecdysone-supplemented agar did not delay the onset of metamorphosis, pupariating 36 hr after starvation (Figure 6C, Supplementary file 1). When we starved P0206>FoxO, FoxO null larvae on 20E supplemented agar, time to pupariation increased with age of starvation from 0–12 hr AL3E, with larvae showing a maximum time to pupariation of 57 hr (Supplementary file 1) at 12 hr AL3E, then decreased thereafter. This suggests that (1) 20E administration eliminated the strong delays in time to metamorphosis seen in pre-critical weight P0206>FoxO, FoxO null larvae and (2) FoxO overexpression in the PGs has additional stage-specific effects on the time to pupariation after the critical weight ecdysone peak and this effect is nutrition-sensitive.

26In addition to the effects we observed on developmental time, we found that both P0206>FoxO, FoxO null and P0206, FoxO null parental control animals were significantly smaller when they were continuously fed 20E-supplemented normal fly medium when compared to animals reared on normal fly medium alone (Figure 6—figure supplement 1C,D). These data demonstrate that altering the timing of critical weight, by exogenous ecdysone administration, impacts final adult size.

27The goal of this study was to uncover the molecular mechanism through which nutrition regulates ecdysone synthesis at critical weight. Our results show that FoxO and Usp interact to regulate critical weight and suggest that this interaction alters the timing of the ecdysone peak. To definitively test whether the FoxO/Usp complex regulates ecdysone biosynthesis at critical weight, we examined whether the FoxO/Usp complex altered the timing of phm and dib mRNA expression, the timing of e74B expression and finally the timing of the ecdysone peak itself.

28The expression of both phm and dib mRNA peaked shortly before critical weight in the parental controls (Figure 7A,B and Figure 7—figure supplement 1A–C). However, P0206>FoxO, FoxO null larvae showed significant delays in this peak (Figure 7B,D). Overexpressing FoxO NK in the PGs reduced the delays induced by FoxO overexpression (Figure 7B,D). Similarly in the wild type background, phm and dib expression was upregulated significantly earlier when FoxO NK was expressed in the PGs than when FoxO was overexpressed in this tissue (Figure 7—figure supplement 1A,B).

29Even if we observe alterations in ecdysone biosynthesis gene expression, this does not necessarily mean that ecdysone biosynthesis is affected when we manipulate FoxO expression in the PGs. To assess if overexpressing FoxO in the PGs affected ecdysone signaling, we examined the expression of e74B. In the parental control larvae, e74B mRNA expression was up-regulated around 12 hr AL3E, shortly after critical weight (Figure 7C, Figure 7—figure supplement 1C). Overexpressing FoxO in the PGs delayed the up-regulation of e74B in both the FoxO mutant and wild type backgrounds (Figure 7C, Figure 7—figure supplement 1C). Finally, interfering with FoxO/Usp complexes, by overexpressing FoxO NK, in the PGs reduced this delay in both the FoxO mutant and wild type backgrounds (Figure 7C, Figure 7—figure supplement 1C). Thus, FoxO/Usp complex plays a role in regulating the dynamics of ecdysone signaling at critical weight.

30Finally, to show that the FoxO/Usp complex regulates ecdysone biosynthesis at critical weight, we measured ecdysone concentrations in larvae that expressed either FoxO or FoxO NK in their PGs from 6 hr AL3E until the nutrition-dependent critical weight ecdysone peak. Overexpressing FoxO in the PGs of FoxO mutant larvae induced a significant delay in the critical weight ecdysone peak (Figure 7D). In addition, the maximum concentration of this peak was approximately 50% lower than the critical weight ecdysone peak in parental controls. The critical weight ecdysone peak occurred significantly earlier in P0206>FoxO NK larvae than in P0206>FoxO larvae. This difference in the timing of the critical weight ecdysone peak was not due to differences in the effects between FoxO and FoxO NK on PG size. Overexpressing either FoxO or FoxO NK induced indistinguishable reductions in PG size (Figure 7E). Taken together our results show that FoxO acts to control the timing of ecdysone biosynthesis via the FoxO/Usp complex, but also via Usp-independent mechanisms.

31Environmental conditions influence developmental processes by affecting hormone synthesis in many organisms. These interactions form the basis of developmental plasticity, and can act to resize and reshape the whole animal. Although environmental effects on hormone synthesis have been identified as a mechanism underlying plasticity in many insects, what causes hormones to become environmentally-sensitive was poorly understood. Here, we demonstrated that FoxO associates with Usp to regulate nutrition-sensitive ecdysone biosynthesis. Our work uncovers a novel mechanism that allows hormone biosynthesis to become environmentally-sensitive at key developmental events, in this case to control plasticity in body size.

32Increasing IIS/TOR activity in the PGs induces precocious critical weight and reducing its activity in the PGs prolongs this transition. Because IIS/TOR signaling positively regulates the expression of CYP450 ecdysone biosynthetic genes, phm and dib, we previously hypothesized that IIS/TOR exerted these effects by regulating the timing of the small peak of ecdysone that coincides with critical weight.

33Our data both tested this hypothesis and identified a novel interaction between the IIS/TOR and ecdysone signaling pathways. We have found that interactions between FoxO and Usp regulate ecdysone biosynthesis, critical weight and body size. This allows us to propose a model for nutrition-sensitive ecdysone biosynthesis during critical weight (Figure 8). During the molt to the L3, larvae undergo a period of starvation while they expel their mouthparts. As a consequence, IIS/TOR signaling activity in the PGs is reduced and FoxO remains in the nucleus and forms a complex with Usp. The FoxO/Usp complex suppresses ecdysone biosynthesis at least in part by repressing transcription of phm and dib, although we do not know whether this repression is direct. Once larvae start feeding, increasing IIS/TOR activity in the PGs results in the phosphorylation of FoxO, causing the dissociation FoxO/Usp complexes as FoxO moves out of the nucleus. This progressive dissociation of FoxO/Usp complexes results in a gradual rise in ecdysone biosynthesis. Once ecdysone reaches a threshold, it triggers critical weight. Afterwards, the time to metamorphosis is set and can no longer be delayed by starvation. For other ecdysone peaks, a negative feedback loop induced by ecdysone signaling itself down-regulates ecdysone biosynthesis. We expect that negative feedback by ecdysone results in the decline in ecdysone biosynthesis after the critical weight peak.

34In contrast, when larvae are starved before attaining critical weight, FoxO remains in the nucleus. In these larvae, the FoxO/Usp complex suppresses ecdysone biosynthesis and delays critical weight. Consequentially, the onset of metamorphosis is delayed. This work uncovers a mechanism that allows IIS/TOR signaling to control ecdysone biosynthesis, providing an elegant means for nutrition to regulate body size.

35Although the ecdysone peak at critical weight is environmentally-sensitive, many other peaks that occur throughout the larval period show less plasticity in response to environmental cues. Ecdysone biosynthesis is also regulated by a developmental neuropeptide, prothoracicotropic hormone (PTTH). Several extrinsic and intrinsic stimuli affect PTTH secretion, such as photoperiod, oxygen concentrations, signals released from damaged imaginal discs, and the sesquiterpenoid ‘status quo’ hormone juvenile hormone. Activating downstream targets of PTTH signaling in the PGs accelerates the onset of metamorphosis and ablating the PTTH-producing cells delays critical weight. Further, without PTTH the ecdysone peak that stimulates wandering behavior, where the larvae emerge from the food to begin metamorphosis, is dramatically delayed. Thus in contrast to IIS/TOR signaling whose major effects are to control the critical weight ecdysone peak, PTTH regulates all ecdysone peaks. Why particular ecdysone peaks are more sensitive to IIS/TOR signaling is unclear, however understanding the mechanisms underlying this differential sensitivity may be key to understanding developmental plasticity.

36FoxO also regulates the critical weight ecdysone peak independently of Usp; overexpressing FoxO NK in the PGs still induces delays in ecdysone biosynthesis and critical weight, even if these delays are more moderate than those induced by wild type FoxO. Thus, our data suggest that FoxO plays additional roles in regulating ecdysone biosynthesis, either on its own or through interaction with other binding partners.

37The effects of starvation on ecdysone biosynthesis do not appear to be the same for all stages of development. Even though starvation causes a delay in development before attaining critical weight, once they reach critical weight, starvation induces moderate acceleration in the time to metamorphosis. This suggests that reducing IIS/TOR signaling induces a mild acceleration of ecdysone biosynthesis at later stages of the L3 development. How IIS/TOR activity regulates ecdysone biosynthesis differently depending on the stage of development is unclear, but it may result from interaction of alternate FoxO binding partners.

38Our findings have broad implications for our understanding of the mechanisms of size regulation and the development of other environmentally-sensitive traits. In other insects, traits such as seasonal wing morphs in butterflies or horn length in male dung beetles arise from differential regulation of ecdysone biosynthesis. Horn length in dung beetles is highly nutrition-dependent, with small, poorly-fed males bearing small horns and large, well-fed males having disproportionately larger horns. Small-horned males have a characteristic peak of ecdysone in their final instar absent in their well-fed, larger conspecifics. Our data propose a mechanism through which nutrition, via FoxO–Usp interactions, might regulate this peak.

39FoxO is also known to form complexes with many vertebrate NHRs, including thyroid hormone, androgen and estrogen receptors. The steroid sex hormones, such as testosterone and estrogen, are important for initiating puberty and the development of adult characters in humans. In girls, reaching a body mass of 48 kg determines the timing of first menses. Obese girls reach this mass faster, resulting in earlier onset of puberty possibly due to higher levels of insulin signaling. These findings suggest that IIS/TOR activity regulates the production of the steroid sex hormones to regulate developmental timing in vertebrates. Furthermore, two mammalian NHRs, CAR and PXR, associate with FoxO1 to regulate the expression of the CYP450 enzymes. The similarity in the roles of FoxO/NHR complexes between mammals and insects provides a testable model that FoxO-NHR complexes regulate environmentally-sensitive development in a wide range of organisms.

40Wild type FoxO and FoxO NK were amplified by RT-PCR using cDNA made from whole body extract of post-feeding (wandering) w[1118] larvae. After sequencing, the constructs were inserted into pUAST attB vector using EcoRI and KpnI whose recognition sites are included on the primers, then integrated on the second chromosome by site-directed insertion using the phiC31 integrase and an attP landing site carrying recipient line, y[1] w[1118]; PBac{y[+]-attP-9A} VK00018 (Bloomington Drosophila Stock Center #9736). w; UAS Usp 26A3 line was a gift from Dr Michael O'Connor (University of Minnesota). We used FoxO Δ94, a gift from Dr Linda Partridge (University College London), with the deficiency line, w[1118]; Df(3R)Exel8159/TM6B, Tb[1] (#7976; Bloomington) as our FoxO null mutant. We obtained the PromE(800)-Gal4 (also known as Oe-Gal4) line from Dr Carlos Ribeiro (Champalimaud Centre for the Unknown). For Usp and FoxO knock down experiments, we used y[1] v[1]; P{y[+t7.7] v[+t1.8] = TRiP.JF02546}attP2 (#27258; Bloomington) as UAS double-stranded (ds) Usp and Vienna Drosophila RNAi Center 107786 as UAS dsFoxO.

41Egg collections were performed on normal food plates and larvae were reared at controlled densities without additional yeast (about 200 eggs/60 mm diameter normal fly medium plate). Newly molted L3 larvae were collected every 2 hr. Collected larvae were raised in a normal cornmeal/molasses medium without additional yeast until the appropriate time point. For starvation treatments, we used 1% non-nutritional agar. To determine the duration of the L3, pupariation time was observed every 2 hr until all treated larvae pupariated or died. We defined pupariation as cessation of movement with evaginated spiracles. All treatments were performed at 25°C under constant light to avoid the effect of circadian rhythm on PTTH secretion. To analyze critical weight, we used a breakpoint analyses as previously described. We constructed growth curves by weighing larvae across a range of defined ages. We then starved larvae of different age classes on non-nutritive agar media and measured the time it took for them to reach pupariation from the onset of starvation, checking for pupariation every 2 hr. We converted the time at which larvae reached critical weight to size using linear regression models from the growth curves.

42For ecdysone feeding experiments, we added 0.15 mg of 20E (SciTech Chemicals, Dejvice-Hanspaulka, Czech Republic) to 1 g of normal fly medium or starvation medium (1% non-nutritive agar). After 20E was added, the media were well mixed and spun down a day before use. To measure time to pupariation from the onset of starvation, larvae were collected as above in 2 hr intervals from the molt. They were then transferred to 20E-supplemented fly medium until they reached the desired age for transfer to 20E-supplemented agar.

43As a proxy for adult body size, we individually weighed pharate adults. Pharate adults, which were about 6–14 hr before eclosion, were collected from vials, carefully cleaned off using distilled water and a paint brush, and then dried for 15 min on paper towels. Once dry, pharate adults were individually weighed on an ultra-microbalance (Sartorius, SE2). We observed the presence or absence of male specific sex combs through pupal cases under a stereoscope to distinguish males from females.

44Concentrations of 20E were quantified using a 20-Hydroxyecdysone EIA kit (Cayman Chemicals). Carefully staged larvae were washed in distilled water twice, briefly dried on paper towels, weighed and flash-frozen on dry ice. Larvae were preserved in three-times their volume of ice cold methanol and kept at −80°C until use. Ecdysone extraction was performed as previously described. Concentrations of 20E were quantified according to the manufacturer's instructions.

45Entire coding regions of Drosophila foxo, ecr-A, ecr-B1 and usp cDNAs were isolated by RT-PCR using cDNA made from w[1118] wandering larvae. For ecr-A, ecr-B1 and usp RT-PCR, forward primers were designed for gene specific sequences with a Flag-tag sequence on the 5′-end and reverse primers were designed for gene specific sequences including native stop codons. To create point mutation constructs, we designed primers containing point mutation(s) and performed standard site-directed mutagenesis methods with minor modifications. GST-tagged protein was purified by Glutathione Sepharose 4B (GE Healthcare). Flag-tagged protein was detected by the anti-Flag M2 monoclonal antibody (1:1000, Sigma). For co-immunoprecipitation assays, we used 500 µg of larval protein or cell extract, the AB11 (anti-Usp monoclonal antibody) [gifts from Drs Sho Sakurai (Kanazawa University) and Lynn M Riddiford (Janelia Farm Research Campus, HHMI)] and DDA2.7 (anti-EcR monoclonal antibody, Developmental Studies Hybridoma Bank). For western blots, the antibodies we used were: anti-Usp (1:1000, AB11), anti-EcR (1:5000, DDA2.7), anti-FoxO (1:1000) and anti-Histone H3 (1:1000, Cell Signaling).

46Immunocytochemistry was performed using standard methods as described previously. The antibodies we used were: anti-Usp (1:100, AB11), anti-FoxO (1:1000) and anti-HA (1:100, Covance). For nuclei and actin staining, we used DAPI (Invitrogen) and Phalloidin (Sigma), respectively.

47Total RNA was extracted from entire larval bodies using TRIzol (Invitrogen). After DNase treatment, total RNA concentration was quantified and 1 µg total RNA was converted to cDNA using oligo dT primers and reverse transcriptase. qPCR was performed using SYBR Green PCR Master Mix (Applied Biosystems) and ABI 7900HT (Applied Biosystems). Primers are listed in Supplementary file 2.

48The Dmel cell line was used for all cell culture experiments. Cells were cultured in the Express Five SMF medium (Gibco) without any serum, insulin or additives, unless mentioned. Transfection was performed using FuGENE HD Transfection Reagent (Roche), according to the manufacturer's instructions. For insulin treatment, transfected cells were re-suspended 66 hr after transfection, and split into two groups. 10 µg/ml bovine insulin (Sigma) was added into the medium of one of these groups. Cells were kept for additional 6 hr at 25°C. Luciferase assays were performed using the Luciferase Assay System (Promega), according to the manufacturer's instructions. To transfect equal amount of plasmid between all treatments, we used bacterial ampicillin resistance gene (ampr). InR- and 4E-BP-luciferase constructs were made according to previous study. Briefly, we designed restriction enzyme site-attached primers (Supplementary file 3) and amplified these promoter regions by PCR using w[1118] genomic DNA. After sequencing, we digested these fragments by NotI and BamHI and inserted into modified pAc5-V5-His B vector (Invitrogen).

49Data for pharate adult weight for males and females, critical weight, growth rates, qPCR and ecdysone quantifications are deposited in Dryad (doi:10.5061/dryad.75940). In addition, we have uploaded the scripts to generate the breakpoint plots, calculate critical size from the growth curves and to perform the permutation tests.