DPP-mediated TGFβ signaling regulates juvenile hormone biosynthesis by activating the expression of juvenile hormone acid methyltransferase

Juvenile hormone (JH) coordinates with 20-hydroxyecdysone (20E) in regulating insect molting and metamorphosis. The molting process is orchestrated by 20E, whereas the nature of the molt is determined by JH during critical JH-sensitive periods. In the presence of JH, 20E induces larva-larva molt, whereas in the absence of JH, 20E promotes larva-pupa or pupa-adult metamorphosis (Gilbert et al., 2000; Riddiford et al., 2003). The recent progress in our understanding of JH molecular action clarifies the function of Methoprene-tolerant (Met) and Krüppel-homolog 1 (Kr-h1) in transducing JH signaling. The Drosophila Met gene encodes a bHLH-PAS protein family member, which is proposed to be a component of the elusive JH receptor (Wilson and Ashok, 1998). Kr-h1 is considered to be a JH signaling component that works at downstream of Met (Minakuchi et al., 2008; Minakuchi et al., 2009).

JH biosynthesis is regulated at three closely linked steps. In the first step, developmental, environmental and physiological cues are received by the central nervous system, which determines the appropriate rate of JH synthesis (Riddiford, 1993). In the second step, the brain transfers these signals to mediate JH biosynthesis in an endocrine gland, the corpus allatum (CA). It has long been thought that JH biosynthesis is regulated primarily by two neuropeptides secreted by brain neurosecretory cells: allatotropin (AT) and allatostatin (AST), which stimulate and inhibit JH synthesis, respectively (Stay, 2000; Weaver and Audsley, 2009). Nevertheless, no AT-like neuropeptides or AT receptor genes have been found in the Drosophila genome thus far (Nassel, 2002; Hauser et al., 2006; Liu et al., 2006; Yamanaka et al., 2008). Although AST-like neuropeptides exist in Drosophila (Lenz et al., 2000), their function of inhibiting JH biosynthesis has not been demonstrated. Alternatively, the brain may directly control JH biosynthesis through neurotransmitters. For example, studies in cockroaches and Drosophila revealed that glutamatergic nerves innervate CA cells and that the N-methyl-D-aspartate subtype of glutamate receptors (NMDAR) are expressed in both the brain and CA. Additionally, glutamate and NMDA were shown to stimulate JH synthesis in vitro (Chiang et al., 2002). In the final step of JH biosynthesis regulation, the brain signals received in the CA should be translated into changes in the expression and/or activity of key regulatory JH biosynthetic enzymes, which directly determine the rate of JH biosynthesis. However, there are major voids in our current understanding of the pathways that lead from brain signals to the activities of JH biosynthetic enzymes.

The evolutionarily conserved TGFβ signaling pathway modulates a wide range of cellular processes, including proliferation, differentiation, migration, apoptosis and cell fate speciation (Kingsley, 1994; Massagué et al., 2000). In addition, studies in C. elegans reveal that the TGFβ signaling pathway controls dauer formation through modulation of dafachronic acid synthesis (for a review, see Hu, 2007). Here, we present a novel, gradient-independent function of Dpp, a TGFβ ligand, in controlling JH biosynthesis. Dpp-Tkv-Mad-mediated TGFβ signaling in the CA serves as a bridge to connect brain-derived neurotransmitter signals to the transcriptional changes of JH acid methyltransferase (JHAMT), a key regulatory enzyme of JH biosynthesis (Shinoda and Itoyama, 2003; Sheng et al., 2008).

All fly strains were grown on standard cornmeal/molasses/agar medium at 25°C. The GAL4-PG12 line is a gift from H.-M. Bourbon (Bourbon et al., 2002). All lethal mutant lines used in the genetic screen were obtained from the Bloomington Drosophila Stock Center. The alleles for dpp, tkv, Mad, Met, Kr-h1 and Nmdar1 used in this study are dpp^s11, dpp^d5, tkv⁷, tkv⁸, tkv^k16713, Mad^1-2, Mad^8-2, Mad^k00237, Mad^kg00581, Met²⁷ (a gift from T. Wilson, Ohio State University, Columbus, OH, USA), Kr-h1^k04411, Kr-h1¹⁰⁶⁴², Nmdar1⁰⁵⁶¹⁶ and Nmdar1^DG23512. RNAi lines of these genes, including UAS-dpp RNAi, UAS-tkv RNAi, UAS-Mad RNAi, UAS-jhamt RNAi and UAS-Nmdar1 RNAi, were also obtained from the Bloomington Drosophila Stock Center. Other fly lines used in this study include hs-GAL4; Dscam-GAL4 (Wang et al., 2004); GAL4-Aug21 (Mirth et al., 2005); UAS-Mad (a gift from S. J. Newfeld, Arizona State University, Tempe, AZ, USA); and UAS-dpp (Bloomington Drosophila Stock Center).

Fly stocks used for the creation of fat body MARCM clones include FRT40; FRT40, tkv⁸/Cyo (a gift from K. Moses, Emory University, Atlanta, GA, USA); FRT40, Mad^8-2/Cyo; FRT40, Kr-h1¹⁰⁶⁴²/Cyo; hs-Flp, UAS-mCD8GFP; and FRT40, tub-GAL80. MARCM clones in the fat body were induced by hs-Flp through heat shock-independent induction as previously described (Britton et al., 2002). To generate jhamt-GAL4 transgenic flies, a 2 kb jhamt promoter was isolated by genomic DNA PCR, fused with GAL4 cDNA and inserted into pCaSpeR4. Transgenic fly lines were generated by P element-mediated germline transformation at Rainbow Transgenic Flies (Camarillo, CA, USA).

Larval fat bodies were dissected from the 2nd or 3rd instar larvae. Immunohistochemistry was performed as described previously (Wang et al., 2002). Florescence signals were captured with a Zeiss LSM510 confocal microscope (Carl Zeiss) and processed using Adobe Photoshop.

The JHA pyriproxifen (Sigma) was dissolved in 95% ethanol to give a 300 ppm stock solution. JHA-containing fly food was prepared by adding JHA stock solution to the standard cornmeal-molasses-yeast food at 50-55°C to a final concentration of 0.1 ppm or as indicated. JHAMT activity in the brain-ring gland complex was measured as previously described (Liu et al., 2009).

Protein extracts isolated from 2nd instar larvae were analyzed by standard SDS-PAGE and western blot. The expression of β-tubulin was used as a loading control. Br mouse monoclonal antibody Br-core (25E9.D7) (Emery et al., 1994) and β-tubulin mouse monoclonal antibody (AA12.1) were obtained from the Developmental Studies Hybridoma Bank at the University of Iowa.

Total RNA samples were prepared from the whole body for 2nd instar larvae or ring glands for 3rd instar larvae and pupae. Primers used are listed in Table S1 in the supplementary material. Quantitative real-time PCR was performed using the LightCycler 480 SYBR Green I Master Kit (Roche). The mRNA levels of different genes were normalized to actin mRNAs, with three replicates for each sample.

The broad (br) gene has been identified as a key regulator in mediating the crosstalk between 20E and JH signaling pathways. Studies in Manduca indicated that the expression of br is directly induced by 20E, but this induction can be prevented by the presence of JH (Zhou et al., 1998; Zhou and Riddiford, 2002). Therefore, in many tissues, br is predominantly expressed during the larval-pupal transition when 20E is high and JH is absent (Huet et al., 1993). To identify genes involved in JH biosynthesis or JH signaling, we developed a Drosophila genetic screen to isolate mutations that de-repress br expression in the 2nd instar larvae. We reasoned that mutations that block JH biosynthesis or disrupt JH action should reduce JH activity and cause precocious br expression.

To monitor br expression in live organisms, we examined the expression patterns of GAL4 enhancer-trap lines inserted near the br gene. One of these lines, GAL4-PG12, closely resembles the temporal and spatial expression pattern of the endogenous br gene in tissues other than the salivary gland. In the genetic screen, GAL4-PG12>UAS-mCD8GFP on the X chromosome was used as a reporter of br expression, and lethal mutations or P-insertions on the 2nd or 3rd chromosome were made homozygous and screened for precocious br expression. Because most of the lethal lines allowed organisms to develop to early larval stages, we were able to examine GFP expression in the 2nd instar under a fluorescent microscope.

From 4400 lethal lines, 55 mutations were isolated based on GFP expression in the 2nd instar larvae. Genes associated with these mutations encode proteins with various molecular functions, including enzymes, signal transduction molecules, transcriptional factors and others. Some of them are known to be involved in JH biosynthesis, such as farnesyl diphosphate synthase (FPPS) (Sen et al., 2007) and NMDAR1 (Chiang et al., 2002).

Among these 55 genes were two main components of TGFβ signaling, thick vein (tkv) and Mothers against Dpp (Mad) (Raftery and Sutherland, 1999). As shown in Fig. S1 in the supplementary material, the expression of GAL4-PG12>UAS-mCD8GFP was restricted to salivary glands in the wild-type 2nd instar larvae, but ubiquitous expression of GAL4-PG12>UAS-mCD8GFP was detected at the same stage of the tkv, Mad and Nmdar1 mutant larvae. Consistently, when assessed with Br-core antibody staining, endogenous Br proteins were not detectable in the fat body (FB) of wild-type 2nd instar larvae but were observed in the FB nuclei of both tkv^k16713 and Mad^k00237 mutant larvae (Fig. 1A). To further test this finding, we examined other tkv and Mad alleles, including tkv⁷, tkv⁸, Mad^1-2, Mad^8-2 and Mad^kg00581. Precocious br expression was detected in all cases. These results suggest that Tkv- and Mad-mediated TGFβ signaling is required to repress br expression at the early larval stages, possibly through regulating JH titer or signaling.

The same genetic screen also isolated genes that are involved in JH signaling, such as Kr-h1 (see Fig. S1 in the supplementary material; Fig. 1A). Another known JH signaling component, Met, was not analyzed by this screen because the Met gene is located on the X chromosome. A reverse genetic study showed that precocious br expression was also detectable in Met mutant larvae (Fig. 1A).

We next asked whether the effects of tkv and Mad mutations on br expression were caused by the decrease in JH titer. Based on this hypothesis, we expected that precocious br expression in the tkv and Mad mutant larvae would be blocked by exogenous JH agonist (JHA). Wild-type, tkv and Mad larvae were reared on a diet containing 0.1 ppm pyriproxifen, an efficient JHA (Riddiford and Ashburner, 1991). The Br-core antibody was used to detect Br proteins in the 2nd instar larvae. Immunohistochemical results revealed that the precocious br expression was suppressed by exogenous JHA in the FB of tkv and Mad mutants (Fig. 1A). However, the precocious br expression in Kr-h1 and Met mutants was not suppressed by exogenous JHA, which demonstrates that Met and Kr-h1 function as the JH signaling components in mediating br expression (Wilson and Ashok, 1998; Minakuchi et al., 2008; Minakuchi et al., 2009) (Fig. 1A).

These observations were further confirmed by western blot analysis (Fig. 1B). Thus, we assume that Tkv/Mad-mediated TGFβ signaling maintains JH titers to inhibit br expression at the early larval stages, thereby blocking precocious metamorphosis.

Proteins that transduce JH signals should cell-autonomously regulate br expression in the JH-affected organs, whereas proteins that affect JH titer should non-cell-autonomously affect br expression. We asked whether the Tkv and Mad proteins were required for the regulation of br expression in the FB, a target organ of JH (Liu et al., 2009). A MARCM analysis of FB cells was conducted for tkv⁸, Mad^8-2 and Kr-h1¹⁰⁶⁴² mutants. Br protein was assessed with Br-core antibody staining and compared between the homozygous mutant cells (GFP-positive) and the surrounding heterozygous or wild-type cells (GFP negative). As expected, br was not expressed in FB of the wild-type 2nd instar larvae (Fig. 2A,A′), but was expressed in the mutant clones homozygous for Kr-h1 (Fig. 2D,D′). However, br expression was not detected in the tkv and Mad mutant FB clones at the same stage (Fig. 2B,B′,C,C′). This result is different from those shown in Fig. 1A, in which br was found to be highly expressed in FB cells of the tkv and Mad mutant larvae at the 2nd instar. Therefore, at the early larval stages, the presence of the Tkv and Mad proteins in the FB cells is not required for JH-mediated br suppression in these cells. Rather, these proteins function in other tissues to control br expression. These results further support the hypothesis that Tkv/Mad-mediated TGFβ signaling inhibits br expression at the early larval stages by maintaining JH titers.

Because JH titer is mainly determined by JH biosynthesis in the CA, which is controlled by the brain, we tested whether the precocious br expression phenotype in the Mad mutant larvae could be suppressed by expressing Mad cDNA specifically in the CA or brain. When hs-GAL4 was used to drive UAS-Mad ubiquitously in the Mad^k00237 larvae, precocious br expression was fully suppressed as expected. By contrast, when we used Dscam-GAL4, a pan-neuronal expression driver (Wang et al., 2004), the precocious br expression phenotype was not affected. However, when we used GAL4-Aug21, a CA-specific GAL4 line (Mirth et al., 2005), precocious br expression was completely suppressed (Fig. 3A). As the expression of Mad in the CA was sufficient to suppress br expression in FB of the Mad mutant larvae, we infer that Tkv/Mad-mediated TGFβ signaling promotes JH biosynthesis in the CA at the early larval stages.

Next, we asked whether TGFβ signaling regulates the expression of genes encoding crucial enzymes of JH biosynthesis. We first compared mRNA levels for these genes between wild-type and Mad mutant larvae at the 2nd instar using quantitative real-time PCR. Six enzymes, including JHAMT, farnesyl diphosphate synthase (FPPS), farnesol oxidase (FARox), farnesol dehydrogenase (FARD), cytochrome P450 6g2 (Cyp 6g2) and farnesoic acid O-methyltransferase (Famet), were chosen because they catalyze the key steps of JH biosynthesis and are predominantly expressed in the CA (Belles et al., 2005; Noriega et al., 2006). We found that in the Mad^k00237 larvae, the mRNA levels of jhamt were decreased to less than half of that in wild type. However, no changes were detected for the mRNA levels of the other five enzymes (Fig. 3B). We also examined the mRNA levels for genes encoding major JH degradative enzymes, including juvenile hormone esterase (JHE), juvenile hormone epoxide hydrolase 1-3 (JHEh1-3) and juvenile hormone esterase duplication (JHEdup) (Goodman and Granger, 2005). Expression of these JH degradative enzymes was not affected in the Mad mutant larvae (see Fig. S2 in the supplementary material).

To test whether the reduced jhamt mRNA expression in Mad mutants resulted in a correspondingly reduced enzymatic activity, we further measured JHAMT activity in the brain-ring gland complex of Tkv- or Mad-deficient larvae. Because tkv and Mad mutants die at early larval stages, we carried out this experiment in the tkv and Mad RNAi larvae. hs-GAL4/UAS-tkv RNAi, hs-GAL4/UAS-Mad RNAi and hs-GAL4/+ control flies were reared under the normal conditions with or without heat-shock treatment. In the hs-GAL4/+ larvae, heat-shock treatment did not affect JHAMT activity. However, JHAMT activity in heat-shocked hs-GAL4/UAS-tkv RNAi and hs-GAL4/UAS-Mad RNAi larvae was reduced to 40-50% of that in the control (Fig. 3C). Additionally, JHAMT activity in non-heat-shocked hs-GAL4/UAS-tkvRNAi and hs-GAL4/UAS-Mad RNAi larvae was also mildly decreased (~80% of control), probably owing to leaky expression of hs-GAL4. The lower jhamt mRNA levels and JHAMT activity in Tkv- and Mad-deficit larvae indicate that Tkv/Mad-mediated TGFβ signaling in the CA promotes JH biosynthesis by upregulating jhamt expression.

The Drosophila genome encodes seven TGFβ superfamily members. Three of them, Decapentaplegic (Dpp), Glass bottom boat (Gbb) and Screw (Scw), belong to the BMP subgroup and are suggested to signal through Tkv as the type I receptor (Raftery and Sutherland, 1999). In an attempt to identify TGFβ ligand(s) participating in the regulation of JH synthesis, we found that, like tkv and Mad mutations, hypomorphic dpp mutants, dpp^s11 and dpp^d5, also caused precocious br expression in the FB (Fig. 4A), suggesting that Dpp may be the TGFβ ligand regulating JH biosynthesis in the CA. Noticeably, although both are hypomorphic alleles, dpp^s11 and dpp^d5 have rearrangements of the disk and shv regulatory regions, respectively (Johnston et al., 1990). It is likely that both regulatory regions are required for the normal dpp expression in the CA.

Next, we asked where the Dpp protein in the CA originates. Dpp is transported via intracellular trafficking initiated by receptor-mediated endocytosis (Entchev et al., 2000). Therefore, we first assessed dpp expression in the ring gland and found that dpp-lacZ was highly expressed in the CA but not in any other parts of the ring gland (Fig. 4B).

Null alleles of dpp mutations are embryonic lethal, whereas hypomorphic alleles, such as dpp^s11, were completely pupal lethal. Interestingly, this pupal lethality could be partially rescued by exogenous JHA. When reared on JHA-containing diet, more than 10% of dpp^s11 larvae developed into adults (Fig. 4C). This finding was further supported by a CA-specific dpp RNAi assay (see Fig. S3 in the supplementary material). These data not only reinforce the importance of Dpp in regulating JH biosynthesis, but also imply that the pupal lethality of dpp^s11 are partially caused by the reduced jhamt expression and JH levels. When jhamt was ectopically expressed in the CA, ~8% of dpp^s11 mutants developed to adulthood (Fig. 4D). In light of all this evidence, we conclude that Dpp expressed in the CA is the TGFβ ligand of Tkv in stimulating JH biosynthesis.

As mutations in dpp, tkv and Mad induce precocious br expression, we asked whether CA-specific knockdown of dpp, tkv, Mad or jhamt affects br expression. When the expression of dpp, tkv, Mad or jhamt was knocked down by CA-specific RNAi, precocious br expression was detected in the FBs of 2nd instar larvae in all cases (Fig. 5). These results further support that Dpp-Tkv-Mad-mediated TGFβ signaling in the CA is required to regulate JH biosynthesis.

We have demonstrated that Dpp and its downstream signaling molecules Tkv and Mad are required for normal jhamt expression and JH biosynthesis in the CA. It is crucial to determine whether TGFβ signaling is an efficient regulation mechanism for jhamt transcription. To address this, we first measured jhamt mRNA levels in the ring glands of Dpp-deficient flies at the wandering larval stage. We found that jhamt mRNA levels in the ring glands of dpp^s11, dpp^d5 and CA-specific dpp RNAi larvae were only 10-40% of that in wild type. By contrast, when dpp was ectopically expressed in the CA, the mRNA level of jhamt increased fourfold (Fig. 6A).

We next compared the developmental profiles of dpp and jhamt expression in the CA of wild-type animals. From the late 3rd instar larva to early pupa, only a single high peak of jhamt mRNA levels was detected in the wandering larval stage, which is consistent both with a previous report (Niwa et al., 2008) and with JH titers in the hemolymph (Riddiford, 1993). As shown in Fig. 6B, jhamt mRNA level in the ring glands of wandering larvae was ~11-fold higher than that of the larvae 10 hours before the wandering stage. The expression pattern of dpp in the CA was similar to that of jhamt, but the increase of dpp mRNA prior to the wandering larval stage occurred hours earlier than that of jhamt mRNA (Fig. 6B). Therefore, despite differences in developmental stages of wild-type animals or in the Dpp-deficient larvae with distinct genetic backgrounds, dpp expression in the CA always correlates with jhamt expression in vivo. These findings further demonstrate that Dpp-mediated TGFβ signaling plays a crucial role in controlling JH biosynthesis through upregulating jhamt expression.

Finally, we asked whether dpp expression in the CA was regulated by brain signals. Drosophila Nmdar1, a glutamate receptor, is expressed in the CA and plays a role in regulating JH biosynthesis (Chiang et al., 2002). Our genetic screen also identified that mutations in Nmdar1 caused precocious br expression (see Fig. S1 in the supplementary material). As shown in Fig. 7, at the wandering larval stage, mRNA levels of both dpp and jhamt in Nmdar1^DG23512 and Nmdar1⁰⁵⁶¹⁶ mutants were reduced to below 30% of those in wild type. In addition, CA-specific knock-down of Nmdar1 expression also induced precocious br expression in the FBs of 2nd instar larvae. These data suggest that dpp expression in the CA is directly controlled by neurotransmitter signals from the brain.

The functions of the TGFβ superfamily and other morphogens in regulating insect metamorphosis are rarely reported. In two independent genetic screens, we discovered that Drosophila TGFβ signaling controls two different aspects of insect metamorphosis. In a previous study, we found that Baboon (Babo) and dSmad2-mediated TGFβ signaling regulates larval neuron remodeling, which is part of the insect central nervous system metamorphosis induced by 20E during the pupal stage. Further investigation revealed that Babo/dSmad2-mediated TGFβ signaling controls larval neuron remodeling through regulating the expression of EcR-B1, a specific isoform of the 20E receptor (Zheng et al., 2003).

In this paper, we report several findings. First, br is precociously expressed in 2nd instar tkv and Mad mutant larvae (see Fig. S1 in the supplementary material). Second, the precocious br expression phenotype in tkv and Mad mutant larvae can be suppressed by exogenous JHA (Fig. 1). Third, Tkv and Mad repressed br expression in a non-cell-autonomous manner (Fig. 2). Fourth, the presence of Mad in the CA is sufficient to repress br expression in the FB (Fig. 3A). Fifth, jhamt mRNA levels and JHAMT activity were significantly reduced in the Mad-deficient larvae (Fig. 3B,C). These results demonstrate that Tkv- and Mad-mediated signaling is required in the CA to activate jhamt expression and thus JH biosynthesis, which in turn controls insect metamorphosis.

The Drosophila genome encodes two TGFβ type II receptors, Punt (Put) and Wishful thinking (Wit) (Raftery and Sutherland, 1999). Our genetic screen failed to identify a role for either of these receptors in the regulation of JH biosynthesis. Put and Wit are most probably functionally redundant in this biological event, as in the case of TGFβ-mediated mushroom body neuron remodeling (Zheng et al., 2003).

Dpp is a key morphogen that controls dorsal/ventral polarity, segmental compartment determination and imaginal disc patterning. Dpp function usually depends on its gradient distribution (Affolter and Basler, 2007). In an attempt to identify the ligand for Tkv/Mad-mediated TGFβ signaling in the CA, we have discovered a novel, gradient-independent role for Dpp that controls JH biosynthesis. We demonstrate that Dpp is the ligand of Tkv, which regulates jhamt transcription. Loss of Dpp, even RNAi reduction of Dpp in the CA specifically, causes precocious br expression at the early larval stages, which phenocopies tkv and Mad mutants (Fig. 1A, Fig. 3A and Fig. 5). Phenotypes of dpp, including precocious br expression and lethality, are at least partially rescued by JHA treatment (Fig. 4C) or ectopic jhamt expression in the CA (Fig. 4D). Notably, dpp-lacZ is strictly expressed in the CA cells, but not in the other two types of endocrine cells in the ring gland: the prothoracic gland and corpus cardiacum cells (Fig. 4B). The developmental expression profile of dpp in the CA is always consistent with that of jhamt (Fig. 6). Finally, dpp expression in the CA may be directly controlled by neurotransmitter signals in the brain, which is supported by reduced dpp and jhamt transcription levels in the Nmdar1 mutant wandering larvae (Fig. 7).

Several lines of evidence suggest that Met is a crucial regulator at or near the top of a JH signaling hierarchy, possibly acting as a JH receptor (Wilson and Ashok, 1998). However, null Met mutants of Drosophila are completely viable, which is unexpected if Met is a JH receptor. A recent investigation indicated that another Drosophila bHLH-PAS protein, Germ cell-expressed (Gce), which has more than 50% homology to Met (Godlewski et al., 2006), may function redundantly to Met in transducing JH signaling (Baumann et al., 2010). Because Met is on the X chromosome in the fly genome, it was not covered by our genetic screen. We tested the br protein in the FBs of a Met null allele, Met²⁷, at the 2nd instar larval stage and observed precocious br expression. Importantly, this precocious br expression phenotype could not be suppressed by exogenous JHA (Fig. 1A). This result not only supports the previous reports regarding the function of Met in transducing JH signaling but also suggests that the precocious br expression is a more sensitive indicator for the reduced JH activity in Drosophila compared with precocious metamorphosis, lethality and other phenotypes.

Kr-h1 was reported to act downstream of Met in mediating JH action. Studies in both Drosophila and Tribolium reveal that, at the pupal stages, exogenous JHA induces Kr-h1 expression, which in turn upregulates br expression (Minakuchi et al., 2008; Minakuchi et al., 2009). Our genetic screen successfully identified that Kr-h1 is cell-autonomously required for the suppression of br expression at young larval stages (see Fig. S1 in the supplementary material; Figs 1 and 2). Precocious br expression occurred in the FBs of Kr-h1 mutants and was not suppressed by JHA treatment (Fig. 1). Therefore, our studies further suggest that Kr-h1 functions as a JH signaling component in mediating insect metamorphosis. However, our finding shows that, at the larval stages of Drosophila, the JH-induced Kr-h1 suppresses, rather than stimulates, br expression. This result is consistent with the facts that Kr-h1 functions to prevent Tribolium metamorphosis (Minakuchi et al., 2009) and Br is a crucial factor in promoting pupa formation (Zhou and Riddiford, 2002).

Taken together, we find a novel function of Dpp, Tkv and Mad-mediated TGFβ signaling in controlling insect metamorphosis. As summarized in our model (Fig. 8), the brain sends neurotransmitters, such as glutamate, to the CA through neuronal axons. Glutamate interacts with its receptor (NMDAR) on the surface of CA cells to induce dpp expression. Dpp protein produced and secreted by CA cells forms a complex with TGFβ type I receptor (Tkv) and type II receptor on the membrane of CA cells, followed by phosphorylation and activation of Tkv. Activated Tkv in turn phosphorylates Mad, which is imported into the nucleus together with co-Smad and stimulates jhamt expression. JHAMT in CA cells transforms JH acid into JH, which is released into hemolymph. The presence of JH in young larvae prevents premature metamorphosis through Met/Gce and Kr-h1 by suppressing the expression of br, a crucial gene in initiating insect metamorphosis.

We thank Drs L. Pick, C. Mitter, W. Bendena, and Q. Song for critical discussion and comments on the manuscript. We thank N. B. Randsholt, K. Moses, S. J. Newfeld, T. Wilson, and D. Harrison for fly strains or antibodies. J.W. is supported by the March of Dimes Foundation (1-FY07-477) and the National Science Foundation (ISO1021767). S.L. is supported by the Natural Science Foundation of China (30770271).