Research Article

A brain circuit that synchronizes growth and maturation revealed through Dilp8 binding to Lgr3

See allHide authors and affiliations

Science  13 Nov 2015:
Vol. 350, Issue 6262, aac6767
DOI: 10.1126/science.aac6767

Brain keeps body size and shape in check

Animal systems show amazing left-right symmetry—think of how our legs or arms, or the legs or wings of an insect, are matched in size and shape. Environmental insults and growth defects can challenge these developmental programs. In order to limit the resultant variation, juvenile organisms buffer variability through homeostatic mechanisms, so that the correct final size is attained. Vallejo et al. report that the Drosophila brain mediates such homeostatic control via an insulin-like peptide Dilp8 binding to the relaxin hormone receptor Lgr3. Lgr3 neurons distribute this information to other neuronal populations to adjust the hormones ecdysone, insulin, and juvenile hormone in a manner that stabilizes body and organ size.

Science, this issue p. 10.1126/science.aac6767

Structured Abstract


Animals have a remarkable capacity to maintain a constant size, even in the face of genetic and environmental perturbations. Size imperfections and asymmetries have an effect on fitness, potentially decreasing competitiveness, survival, and reproductive success. Therefore, immature animals must employ homeostatic mechanisms to counteract substantial size variations and withstand developmental growth perturbations caused by genetic errors, disease, environmental factors, or injury. Such mechanisms ensure that, despite inevitable variations, the appropriate final body size is attained. A better understanding of homeostatic size maintenance will afford insights into normal organ and organismal size control, as well as the developmental origin of anomalous random left-right asymmetries.


The Drosophila insulin-like peptide Dilp8 has been shown to mediate homeostatic regulation. When growth is disturbed, Dilp8 is strongly activated and sexual maturation is postponed until the affected elements are recomposed; simultaneously, the growth of other organs is retarded during this process. This compensatory mechanism allows the growth of the affected tissues to catch up. It maintains the synchrony between organs so that the animals achieve the correct size, preserving proportionality and bilateral symmetry. However, the Dilp8 receptor and its site of action remain uncharacterized.


We found that Dilp8 binds to andactivates the relaxin leucine-rich repeat–containing G protein–coupled receptor Lgr3 to mediate homeostatic control through a pathway dependent on adenosine 3′,5′-monophosphate. Larvae that lack lgr3 in neurons alone do not respond to Dilp8, indicating that the homeostatic system is centered in the brain. Dilp8 delays reproductive maturation by suppressing the neurons releasing the prothoracicotropic hormone (PTTH), which projects to the prothoracic gland and regulates ecdysone production for growth termination. However, this modulation alone is insufficient to adjust growth and stabilize body size. We show that Dilp8-Lgr3 balances growth against the extended growth period by dampening the production of dilp3 and dilp5 by insulin-producing cells (IPCs) in the brain and inhibiting synthesis of the juvenile hormone (JH).

We also identify two pairs of dorsomedial neurons in the pars intercerebralis that are necessary and sufficient to mediate the effects of Dilp8. Simultaneous detection of pre- and postsynaptic markers revealed that the Lgr3 neurons mediating this homeostatic control have extensive axonal arborizations. Genetic and GRASP (GFP reconstitution across synaptic partners) analyses demonstrate that these neurons are connected to both the IPCs and PTTH neurons critical for adjusting growth and maturation rate, respectively. Thus, through their extensive axonal arborizations, Lgr3 neurons function like a “neuronal hub”: They route peripheral information about growth status to other neuronal populations, thereby synchronizing damaged tissues and other (undamaged) ones and allocating additional development time so that each organ attains the correct size and maintains proportionality and symmetry.


We identified the relaxin receptor Lgr3 as a Dilp8 receptor and defined a brain circuit for homeostatic control of organismal and organ size in the face of perturbations. Lgr3 neurons that respond to Dilp8 signals directly input on the insulin-producing cells and the PTTH-producing neurons. As Lgr3 outputs, the modulation of these neuronal populations according to Dilp8 levels is critical to delay maturation and promote growth compensation in a manner that stabilizes body size. Without adequate Dilp8-Lgr3 signaling, the brain is incapable of stabilizing size between the distinct body parts, and we see left-right asymmetries and size variations that are greater than usual, reflecting developmental instability.

Dilp8-Lgr3 neural circuit and outputs for body-size homeostasis.

The brain detects growth status and anomalies via Dilp8 activation of the Lgr3 receptor in a pair of symmetric neurons. These neurons distribute this information to IPCs and PTTH neurons, which then trigger the hormonal responses that regulate size. Without Dilp8-Lgr3 homeostasis, the brain cannot correct variation, and identical body parts can display imperfect symmetry and size.


Body-size constancy and symmetry are signs of developmental stability. Yet, it is unclear exactly how developing animals buffer size variation. Drosophila insulin-like peptide Dilp8 is responsive to growth perturbations and controls homeostatic mechanisms that coordinately adjust growth and maturation to maintain size within the normal range. Here we show that Lgr3 is a Dilp8 receptor. Through the use of functional and adenosine 3′,5′-monophosphate assays, we defined a pair of Lgr3 neurons that mediate homeostatic regulation. These neurons have extensive axonal arborizations, and genetic and green fluorescent protein reconstitution across synaptic partners show that these neurons connect with the insulin-producing cells and prothoracicotropic hormone–producing neurons to attenuate growth and maturation. This previously unrecognized circuit suggests how growth and maturation rate are matched and co-regulated according to Dilp8 signals to stabilize organismal size.

The impressive consistency and fidelity in the size of developing organisms (13) reflect both the robustness of genetic programs and the developmental plasticity necessary to counteract the variations in size arising from genetic noise, erroneous morphogenesis, disease, or injury (4, 5). To counterbalance growth abnormalities, systemic homeostatic mechanisms are implemented that delay the onset of the reproductive stage of adulthood until the correct size of the individual and its body parts has been reached (69). Most animals initiate a pubertal transition only after the critical size and body mass have been achieved and, generally, in the absence of tissue damage or growth abnormalities (5, 811). However, the mechanisms underlying such homeostatic regulation have yet to be fully defined.

Recently, the secreted peptide Dilp8, a member of the insulin/relaxin-like family, has been identified as a factor that mediates homeostatic control in Drosophila melanogaster. During the larval (growth) stage, the expression of dilp8 declines as maturation proceeds, whereas its expression is activated when growth is disturbed (12, 13). Hence, fluctuating Dilp8 levels provide a reliable read-out of overall growth status (e.g., deficit) and the time needed to complete growth. In addition, Dilp8 orchestrates hormonal responses that stabilize body size. This includes (i) inhibiting the production of the steroid hormone ecdysone by the prothoracic gland (PG) until the elements or organs affected are recomposed and also (ii) slowing down growth rates of undamaged tissues to ensure that affected organs catch up with normal tissues so that the adult flies reach a normal body size and maintain body proportions and symmetry. Accordingly, in the absence of dilp8, mutant flies are incapable of maintaining such strict control over their size, as reflected by the exaggerated variation in terms of overall proportionality and imperfect bilateral symmetry (12). However, the receptor that transduces Dilp8 signals and its site of action remained unknown.

Two models can be envisioned to establish such homeostatic regulation: (i) a central mechanism that dictates coordinated adjustments in both the duration and rate of growth and (ii) an endocrine mechanism that involves sensing and processing Dilp8 signals directly by hormone-producing cells (Fig. 1A) (14). In Drosophila, several anatomically separate neural populations regulate growth and maturation time by impinging directly on the ring gland [which is made up of the PG and the juvenile hormone-producing corpus allatum (CA)] (1, 2, 4). Thus, the receptors that transduce the Dilp8 signals of growth status may act directly or may communicate with neurons that produce the prothoracicotropic hormone (PTTH) (15) and/or the neurons of the pars intercerebralis, including the insulin-producing cells (IPCs), which synthesize and release insulin-like peptides Dilp2, Dilp3, and Dilp5 (9, 16). Insect PTTH neurons, which are analogous to the gonadotropin-releasing hormone (GnRH) neurons in mammals (5, 10), signal the commitment to sexual reproduction by stimulating the production of ecdysone in the PG to terminate growth (14). The IPCs in the pars intercerebralis, a functional equivalent of the mammalian hypothalamus (10, 15), integrate nutritional signals and modulate tissue growth accordingly (1620). Manipulation of IPCs by genetic ablation, starvation, or mutations in the single insulin receptor (17, 18, 2022) leads to the generation of animals with smaller size. Similarly, manipulations of the PTTH neuropeptide and neurons result in adult fly size variations, leading to flies that are larger or smaller than normal due to an extension or acceleration of the larval period (15). The insulin receptor also directly activates synthesis of the juvenile hormone (JH) (a hormone that promotes growth and juvenile development) in the CA (23) and production of the steroid prohormone ecdysone in the PG (14), again augmenting the variation in normal adult size. These observations may explain how environmental and internal influences operate through individual IPCs or PTTH neurons to enable body-size variation and plasticity in developmental timing that can be vital for survival in changing environments. However, the origin of developmental stability and invariant body size may require different or more complex neural mechanisms from those involved in adaptive size regulation.

Fig. 1 Orphan relaxin Lgr3 mediates Dilp8 behavior and functions in neurons.

(A) Model of Dilp8 sensing and homeostatic size regulation. The insulin-like Dilp8 elicits diverse yet coordinated responses that prolong the larval stage (by inhibiting ecdysone production in the PG) and slow down the growth rate (by dampening insulin signaling in the imaginal discs and/or other as-yet-uncharacterized signals). The underlying mechanism may involve direct Dilp8 sensing in neurons expressing PTTH or IPCs, the two prominent yet separate neural circuits that regulate ecdysone production in the PG and/or overall growth rates during larval development. Alternatively, the receptor may transduce Dilp8 signals in a novel neuronal population or directly in endocrine cells. (B) Knockdown of lgr3, but not of lgr4, prevents the pupariation delay induced by dilp8 overexpression. tub> indicates tubulin-Gal4. Error bars indicate SD. (C) Average puparion time of the indicated genotypes, exposing acceleration or delay relative to their controls. Error bars (SD) are invisible when the three replicates coincide. Approximately 60 pupae per genotype were scored, and the graph shows data pooled from three independent experiments. ***P < 0.001 (two-tailed unpaired t test). n.s., not significant. (D) Knockdown of lgr3 prevents excess body weight induced by dilp8 (tub-dilp8 da>lgr3-IR). ***P < 0.001; significant difference from all controls (two-tailed unpaired t test). Data are mean ± SD. n = 25 age-synchronized adult males in each genotype. (E) Tissue-specific knockdown of lgr3, using UAS-lgr3-IR and the indicated Gal4 lines. The graph shows data pooled from three independent experiments, and each data point is mean ± SD. A total of 60 pupae were scored per genotype. ***P < 0.001 (two-tailed unpaired t test). (F) Fluctuating asymmetry index of left-right wings of males of the indicated genotypes and rescued animals. Numbers indicate pairs of wings scored. **P < 0.01; ***P < 0.001 (F test).

By employing a candidate approach and biochemical assays, we demonstrate that the orphan relaxin receptor Lgr3 acts as a Dilp8 receptor. We identify the neuronal population molecularly defined by the lgr3 enhancer fragment R19B09 (24) and show that it is necessary and sufficient to mediate such homeostatic regulation. Using tools for circuit mapping and an adenosine 3′,5′-monophosphate (cAMP) sensor as an indicator of Lgr3 receptor activation in vivo, we determined that a pair of these Lgr3 neurons is highly sensitive to Dilp8. These neurons display extensive axonal arborizations and appear to connect with IPCs and PTTH neurons to form a brain circuit for homeostatic body-size regulation. Our data identify the insulin genes, dilp3 and dilp5, the JH, and the ecdysone hormone as central for developmental size stability. Collectively, these findings unveil a homeostatic circuit that forms a framework for studying how the brain stabilizes body size without constraining the adaptability of the system to reset body size in response to changing needs.


The relaxin receptor Lgr3 mediates Dilp8-induced homeostatic control

Dilp8 bears homology to the human relaxin peptides (12, 25). Therefore, we investigated the role for the two fly relaxin receptors encoded by the orphan leucine-rich repeat–containing G protein–coupled receptors (LGRs) Lgr3 (CG31096) and Lgr4 (CG34411) (26, 27). We used RNA interference (RNAi) transgene expression (28) in animals overexpressing a UAS-dilp8 transgene under the control of the yeast transcription factor Gal4 that is driven by the ubiquitous tubulin-Gal4 promoter (tub-Gal4 UAS-dilp8 UAS-receptor-RNAi) (Fig. 1B). Ubiquitous expression of UAS-RNAi transgenes (several lines were tested) against each of the fly relaxin receptors reveals whether they are required for developmental delay resulting from dilp8 overexpression. When the lgr3, but not lgr4, gene was silenced (tub-Gal4 UAS-dilp8 UAS-lgr3-RNAi, hereafter tub>dilp8>lgr3-IR), Dilp8-induced developmental delay was fully suppressed (Fig. 1, B and C). Depletion of lgr3 accelerated pupariation by ~8 hours (Fig. 1C), as in dilp8 mutants (13). We verified the efficiency of the lgr3-RNAi transgene by quantitative real-time polymerase chain reaction (qRT-PCR) (fig. S1A).

To investigate the tissue- and cell-specific requirement for Lgr3, we constructed transgenic lines in which the coding sequence of dilp8 was under the direct control of the ubiquitous tubulinα1 promoter (tub-dilp8) (see materials and methods), and we used a Gal4/UAS system to drive tissue-specific expression of the UAS-lgr3-RNAi transgene. The delay in pupariation induced by Dilp8 resulted in normal-sized adults, owing to Dilp8-induced growth compensation. Yet, the extra time the tub-dilp8 larvae spent in the feeding period led to overweight adults (12, 13). Knockdown of lgr3 also prevented the dilp8-overexpressing animals [tub-dilp8, daughterless (da)-Gal4 UAS-lgr3-RNAi] from being overweight (Fig. 1D).

Tissue-specific knockdown of lgr3 further showed that Lgr3 is required in the nervous system (tub-dilp8 elav>lgr3-IR) (Fig. 1E) and not in the ring gland (using retnR9F04-Gal4) (fig. S1, B to E). Knockdown of lgr3 in neurons, but not in the ring gland, also prevented the Dilp8-induced reduction of growth rate (fig. S1, F and G). Further, neuronal depletion of lgr3 using elav-Gal4 (elav>lgr3-IR) produced adults that displayed greater fluctuations in asymmetry, as evidenced by significantly larger left-right variations in the size of adult wings (Fig. 1F). Expression of a transgenic Lgr3 cDNA (UAS-lgr3) (materials and methods) prevented this defect by the RNAi against lgr3 (elav>lgr3-IR>lgr3) (Fig. 1F), excluding potential off-target effects of the RNAi (28). Thus, similarities in the phenotypes of dilp8 (12) and lgr3 loss—as well as the prevention of Dilp8-induced developmental delay, growth-rate reduction, and excess body weight through the loss of lgr3—strongly suggest that Lgr3 acts as a Dilp8 receptor. These data also suggest a central mechanism for systemic homeostatic size regulation although other lgr3-expressing peripheral tissues, such as the larval fat body (27), could also contribute.

Lgr3 is a Dilp8 receptor

Next, we used biochemical assays to investigate the interaction of Dilp8 and Lgr3. Human relaxin receptors largely activate cytosolic cAMP (25); thus, we tested whether the response of Drosophila Kc cells transiently expressing lgr3 to synthetic Drosophila Dilp8 peptides (materials and methods) was coupled to cAMP. To control for specificity, we also transfected Kc cells with constructs encoding the structurally related Lgr4 (26, 27), as well as Lgr2, which is known to provoke a well-characterized cAMP-mediated response upon binding its respective cognate ligand (29). Only cells transfected with the lgr3-expressing plasmid responded to a 30-min exposure to Dilp8 (50 nM) with an increase in cAMP levels, from 213.8 ± 67.94 fmol/5 × 104 cells to 1.612.36 ± 302.6 fmol/5 × 104 cells (Fig. 2A and materials and methods). As a reference, the cAMP levels in Kc cells transfected with the empty vector alone were 132.69 ± 66.71 fmol/5 × 104 cells and 127.73 ± 77.19 fmol/5 × 104 cells in the presence and absence of synthetic Dilp8 (50 nM), respectively (Fig. 2A). This response to Dilp8 is highly specific, because we did not detect comparable changes in cAMP when cells expressing the Lgr4 and Lgr2 receptors were exposed to Dilp8 (Fig. 2A).

Fig. 2 Lgr3 is a Dilp8 receptor.

(A) cAMP measurement in untreated Drosophila Kc cells (5 × 104 cells per culture) transiently transfected with the indicated Lgr plasmids and the empty plasmid or treated with either 5 or 50 nM Dilp8 peptide for 30 min. Data are shown as mean ± SD (n = 3 independent repeats), and the asterisks indicate that the cAMP level was statistically different from untreated controls, ***P < 0.001 (t test). (B) Dilp8-stimulated dose-dependent cAMP production by Kc cells expressing Lgr3. The concentration of Dilp8 ranged from 0 to 250 nM. Kc cells were transiently transfected with the lgr3, lgr4, or lgr2 plasmids, and an EC50 value of 6.31 ± 0.1277 nM was obtained for Lgr3. Exposure of the Kc cells expressing the related receptors Lgr4 or Lgr2 to Dilp8 did not affect cAMP production. A sigmoid fit to the lgr3 data is shown. Total cAMP production was measured in triplicate (materials and methods). Each data point is mean ± SEM (n = 3 independent repeats). (C) Dilp8 and Lgr3 colocalization assessed by confocal immunofluorescence. Kc cells expressing the extracellular domain of Lgr3-ECD::3xHA were incubated with medium containing Dilp8-Flag (materials and methods). The cells were fixed without permeabilizing and were then stained with anti-FLAG (red) and anti-HA (green) antibodies. The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Images are also presented for Kc cells transiently transfected with the empty::3xHA vector and exposed to Dilp8-Flag. Representative images of three repeats are shown. (D) Binding of Dilp8 to Lgr3 assessed by coimmunoprecipitation. Rabbit anti-Myc or rabbit anti-IgG antibodies were used to pull down Lgr3-ECD::3xHA complexed to Dilp8-Myc. Input controls are also shown. IP, immunoprecipitation; WB, western blot.

A dose-response curve indicated that Dilp8 peptides activate Lgr3 to produce a median effective concentration (EC50) of 6.31 ± 0.12 nM, whereas Lgr4 and Lgr2 did not stimulate cAMP production in response to Dilp8 at any of the doses assayed (Fig. 2B). As the full receptor could not be solubilized, we used a strategy previously employed for the identification of LGR7 and LGR8 as receptors of human relaxin (30) and LGR4 and LGR5 of R-spondins (31). We cloned the ectodomain of Lgr3, fused it to the epitope 3x hemagglutinin (3xHA), and designated this as Lgr3-ECD::3xHA. On the basis of structural homology of LGRs to glycoprotein hormone receptors (25, 31), the extracellular domain of the Lgr3 is expected to be soluble and to bind cognate ligands. Indeed, we detected a strong colocalization of Dilp8 at the surface of Lgr3-expressing cells (Fig. 2C and fig. S2). Furthermore, we coimmunoprecipitated a Myc-tagged Dilp8 with the extracellular domain of Lgr3 (Lgr3-ECD::3xHA) (Fig. 2D, materials and methods, and fig. S2). Collectively, these data suggest that Lgr3 encodes a functionally relevant Dilp8 receptor that is coupled to cAMP signaling like the human relaxin receptors RXPF1-2 (25).

The Lgr3 receptor acts in a small set of central brain neurons

When the endogenous expression of the lgr3 gene was quantified (fig. S1A) (27), it appeared to be expressed only very weakly. Not surprisingly, attempts to map lgr3-expressing neurons by conventional immunological approaches using antibodies against the Lgr3 protein (materials and methods) were unsuccessful. For example, the Lgr3719-733 antiserum readily detected the Lgr3 protein ectopically expressed using the GAL4/UAS system (fig. S2, G to G′′), confirming the specificity of our antisera, yet it could not detect endogenous lgr3 expression, supporting the weak expression of the Lgr3 protein.

Using the Gal4/UAS system to coarsely map functionally relevant neurons, we found that Lgr3 is not required within the IPCs themselves (dilp3-Gal4), the neuropeptide F-expressing cells (npf-Gal4), the circadian clock neurons (pdf-Gal4 and per-Gal4), the PTTH neurons (ptth-Gal4), or the ventral nerve cord (VNC) [teashirt (tsh)-Gal4], all of which have been previously established to regulate the larval-pupal transition and/or body size in response to nutrition, sensory inputs, and developmental cues (14, 15, 17, 20, 32, 33). When UAS-lgr3-RNAi was expressed in these specific brain regions using these Gal4 lines in tub-dilp8 animals, the animals entered pupation at times similar to those of tub-dilp8 animals carrying Gal4 or UAS-lgr3-RNAi alone (Fig. 3A).

Fig. 3 Lgr3 acts in a set of central neurons molecularly defined by R19B09-Gal4.

(A) Puparion time of animals with brain-region–specific knockdown of lgr3, using RNAi and the indicated Gal4. (B) Puparion time of animals with knockdown of lgr3, using lgr3 brain-enhancer fragment-Gal4 lines. Organization of the lgr3 genomic region and the intervals of each of the Lgr3 enhancers (24) are presented in the top image. In (A) and (B), data are mean ± SD and are pooled from three independent experiments; 60 pupae were scored per genotype. Error bars are invisible when the three replicates coincide. ***P < 0.001 (two-tailed unpaired t test). (C) Imaginal disc growth rate in the indicated genotypes assayed by expression of the FoxO target gene, Thor/4E-BP, analyzed by qRT-PCR. mRNA was isolated from imaginal discs from 15 age-synchronized larvae (100 hours AEL) for each genotype (n = 3 biological repeats, mean ± SD). **P < 0.01 (two-tailed unpaired t test). (D and E) Cumulative puparion time of animals overexpressing UAS-lgr3, using R19B09-Gal4 (D), or with electrical hyperexcitation of neurons labeled by R19B09-Gal4, using the UAS-NaChBac ion channel (E). Approximately 60 pupae were scored per genotype. Each data point is mean ± SD (n = 3 independent repeats).

We next took advantage of the available lines expressing Gal4 under control of genomic fragments from the lgr3 locus (24) (Fig. 3B). We found that using the R19B09-Gal4 enhancer to deplete lgr3 (Fig. 3B) fully suppressed the Dilp8-induced delay and did so with the same magnitude as when lgr3 was ubiquitously depleted by tub-Gal4 (Fig. 1B). No other Gal4 enhancer lines prevented the Dilp8-induced delay (Fig. 3B). Depleting lgr3 in neurons labeled by R19B09-Gal4 also prevented the slow growth rate of imaginal discs induced by Dilp8, as reflected by the restoration of almost-normal transcript levels for the Thor/4E-BP gene, a direct target of the growth inhibitor FoxO, and a diagnosis for imaginal disc growth rates (12, 18) (Fig. 3C). Thus, Dilp8 influences growth and maturation through Lgr3 activation in neurons molecularly defined by the R19B09 enhancer.

We also found that overexpression of the UAS-lgr3 transgene in R19B09-labeled neurons was sufficient to evoke a ~12-hour delay in pupariation (Fig. 3D), but this process was not delayed when expressed under control of the other lgr3 genomic fragments (fig. S3). Because most G protein–coupled receptors (GPCRs) display some level of constitutive activity [two-state model of GPCR function (34)] in the absence of agonist ligands, delayed pupariation due to increased levels of endogenous lgr3 via overexpression in R19B09-labeled neurons might reflect an increased response to a low concentration of endogenous Dilp8 or constitutive activity. However, Lgr3 displays high levels of constitutive activity only when expressed in a heterologous system [human embryonic kidney 293 cells (27)]; this high constitutive activity is not observed in Drosophila cells (Fig. 2A) or in vivo in neurons (see below). Moreover, Lgr3 activity is greatly increased in the presence of Dilp8 (Fig. 2, A and B, and below). Activation of R19B09 neurons by expressing the UAS-NaChBac ion channel transgene (35) was sufficient to trigger a delay of ~18 hours (R19B09>NaChBac) (Fig. 3E), which suggests that Dilp8-stimulated Lgr3 activation excites these neurons electrically.

R19B09-Gal4 labels cells in the central brain (CB) and the VNC (Fig. 4A; see figs. S4 and S5A for expression of other lgr3 enhancer fragments). Together with the observations made for the tsh-Gal4 line (Fig. 3A), which typically labels all neurons in the VNC (33), we conclude that a set of ~12 central neurons per hemisphere, molecularly defined by R19B09-Gal4, reflects the Lgr3 neurons that are necessary and sufficient to control size and developmental timing in response to Dilp8. These include neuronal clusters in the dorsomedial region and in the supraesophageous ganglion (SOG) region, as well as individual cells in the dorsal and ventral protocerebrum (Fig. 4, A and B).

Fig. 4 A targeted cAMP biosensor reveals a pair of neurons responding acutely to Dilp8.

(A) Expression of UAS-DsRed in neurons defined by the R19B09-Gal4 enhancer (see fig. S4 for the expression of the other lgr3-Gal4). (B) CRE-F-luc reporter activity is Flp-dependent. (Left) The transgene contains stop sequences flanked by a pair of FRTs (FRT-cassette) inserted between the CREs and the luciferase construct. R19B09-Gal4 activates the UAS-Flp transgene, and the FLP protein (purple) excises the FRT cassette in the CRE-F-luc transgene, only in R19B09 cells. (Right) Central neurons (red dots) labeled by R19B09-Gal4 and cells that respond to Dilp8 signals (outlined in green). The color code denotes the intensity of the signal; the numbers indicate the quantification in tub-dilp8/wt/tub-dilp8 lgr3-IR brains (n > 10 brains scored for each genotype). (C and C′) Higher-magnification views of central neurons in the dorsomedial region stained with anti-Luc (red) and the neuroblast marker anti-Mira (green). The brain is counterstained with anti-DE-Cad (blue). The luciferase response in neuroblasts (Mira-positive cells) in the dorsal region (designated as [3]) is not reproducible and can occur unilaterally. (C′) Single-channel confocal image of anti-Luc staining. Genotype: UAS-Flp/+; tub-dilp8/+; CRE-F-luc/R19B09. (D) Brain of WT control (UAS-Flp/+; +/+; CRE-F-luc/R19B09). (E) Knockdown of lgr3 (elav-Gal4/UAS-Flp; tub-dilp8/+; CRE-F-luc/UAS-lgr3-IR) inhibits the Dilp8-induced cAMP response detected by luciferase driven by the CRE-F-luc construct, reflecting Lgr3 activation and probing specificity of the UAS-lgr3-IR transgene. (F and G) Confocal sections of the control brains (F) elav-Gal4/UAS-Flp; tub-dilp8/+; CRE-F-luc/+ and (G) elav-Gal4/UAS-Flp; +/+; CRE-F-luc/+. Scale bars, 75 μm in (A) and (C) [also applies to (C′) and (D)]; 40 μm in (E) [also applies to (F) and (G)].

A pair of dorsomedial neurons acutely responds to Dilp8

The Lgr3 receptor response to Dilp8 is strongly coupled to cAMP stimulation (Fig. 2, A and B), enabling us to precisely determine the Lgr3-responding neurons via a cAMP biosensor. We used the CRE-F-luciferase (luc) construct (CRE, cAMP response element) (Fig. 4B) that has already been characterized in vivo (36). Thus, by combining the CRE-F-luc construct with UAS-Flp and R19B09-Gal4, we could assay specific cAMP responses in a physiological context (R19B09 neurons and tub-dilp8 background) (Fig. 4, C and D). To test whether the depletion of lgr3 via UAS-lgr3-RNAi rendered the sensor insensitive to Dilp8, we used elav-Gal4 on the X chromosome.

Typically, two neurons with their soma in the dorsomedial region of the pars intercerebralis were bilaterally and strongly labeled in all tub-dilp8 brains (Fig. 4, C and C′, cells designated as type 1) but not in wild-type (WT) brains (Fig. 4D) or in tub-dilp8 elav-Gal4>lgr3-RNAi brains (Fig. 4E). Two to three weakly labeled cells in the dorsolateral region of the CB (designated as type 2) (Fig. 4, B, C′, and G) were also consistently labeled in tub-dilp8 brains (Fig. 4, C, C′, and F). Moreover, unilaterally labeled dorsal cells (type 3) were occasionally seen in the three genotypes, and these luciferase-positive cells were identified as neuroblasts and not neurons by simultaneously colabeling with the Miranda (Mira) protein (Fig. 4C).

Other lgr3 genomic fragments (e.g., R17G11-Gal4) (fig. S5) that did not suppress the Dilp8-induced delay failed to produce levels of Dilp8-induced cAMP comparable to those found in R19B09-labeled neurons (fig. S5, A and B), in agreement with their inability to prevent Dilp8-induced delay (Fig. 3B). The number and position of cells that activated de novo luciferase in response to tub-dilp8 in CRE-F-luc brains expressing the UAS-Flp pan-neuronally (Fig. 4F) matched the cells identified using the R19B09-Gal4 line, which suggests that the neurons labeled by this intronic lgr3 enhancer represent the majority of cells sensitive to Dilp8 signals. The intensity of luciferase in other Lgr3-independent cells in the elav-Gal4 brains was not generally altered (fig. S5D), indicating that the loss of CRE-F-luc signal in the dorsomedial and dorsolateral neurons was not due to nonspecific effects of the lgr3-RNAi.

Lgr3 neurons are connected to PTTH neurons and IPCs

We used R19B09-Gal4 and -LexA constructs (Fig. 5, A to G), presynaptic (syt::GFP; GFP, green fluorescent protein) and postsynaptic (DenMark, dendritic marker) markers (37) (Fig. 5, B and C), and brainbow tools (Fig. 5, D to D′′) (38) to more precisely define the connectivity of possible synaptic interactions of the distinct Lgr3 neuronal populations defined by R19B09-Gal4. Lgr3 neurons with their soma in the dorsomedial region and with a prominent response to Dilp8 (Fig. 4C) display extensive axonal arborizations reminiscent of hub neurons (39). These axonal arborizations of the dorsomedial Lgr3 neurons cover the dendritic fields and axons of PTTH neurons extensively [Fig. 5, A and B, blue denotes antibody to PTTH (anti-PTTH); and movie S1]. Note that the dendrites of PTTH neurons extend in the same direction as their axons (15). Lgr3 axons and dendritic fields (revealed by Syt::GFP and DenMark) (Fig. 5C) are also in close apposition to the IPCs revealed by anti-Dilp2 (Fig. 5, A and C) and by dilp3-Gal4 (movie S2).

Fig. 5 Lgr3-responding neurons acts as hubs connecting distinct neuron subpopulations.

(A) Lgr3 neurons with soma in the dorsomedial region and a prominent response to Dilp8 detected by DsRed (R19B09-Gal4>DsRed) ensheathed PTTH dendritic fields and axons (anti-PTTH, blue) and densely innervated IPCs (anti-Dilp2, green). Image represents a z-projection of confocal optical sections (29 μm thick) from the brain of a larva in the late third-instar (L3) stage. (B and C) Single confocal optical sections of larval brains of R19B09-Gal4 driven presynaptic (UAS-syt::GFP, green) and postsynaptic (UAS-DenMark, red) markers. PTTH-producing neurons are labeled by anti-PTTH [blue (B)] and IPCs by anti-Dilp2 [blue (C)]. Arrowheads point to Lgr3 axonal projections close to PTTH (B) and IPC neural projections (C). Lgr3 axons also project contralaterally (asterisk). Insets show single channels of PTTH (B) and Dilp2 (C). (D to D′′) UAS-dBrainbow reveals that projections from distinct neural subpopulations labeled by R19B09-Gal4 converge on the dorsomedial Lgr3 neurons (green neurons, arrowheads). Single-channel images of Lgr3 neurons in the SOG (red) and dorsomedial (green) regions are shown in (D′) and (D′′). Image is a 52-μm-thickness reconstruction of confocal sections. (E) Positive, robust signals of GRASP revealed extensive connections between Lgr3 (R19B09-LexA>spGFP11) and IPCs (dilp3-Gal4>spGFP1-10>mCD8::RFP). Brains were counterstained with anti-DE-Cad (blue). A 21-μm-thick reconstruction is shown. The inset shows GRASP signals (gray). (F) GRASP signals (arrowheads) between Lgr3 neurons (R19B09-LexA>spGFP11) and PTTH-producing neurons (ptth-Gal4>spGFP1-10) are detected with immunofluorescence (green) (fig. S6, C and D). Brains were costained with anti-PTTH (red) and anti-DE-Cad (blue). The image represents a 26-μm-thick reconstruction. The inset shows GRASP signals (gray). (G) Brains stained with anti-PTTH (blue) could detect potential contact sites (asterisks) of the circuit. GRASP signals (green) were contributed by connections between Lgr3 (R19B09-LexA>spGFP11) and IPCs (red) (dilp3-Gal4>spGFP1-10>mCD8::RFP). The inset shows single-channel staining (anti-PTTH, gray). The image is a single confocal section (1 μm thick). (H) Dilp8-induced delay in puparion formation is prevented by the constitutive active PTTH receptor torsoRL3 mutation. (I) Electrical silencing of PTTH neurons (ptth-Gal4 UAS-mKir2.1) delays pupariation, as compared with controls. Data in (H) and (I) are pooled from three independent experiments, and each data point is mean ± SD. Approximately 60 pupae were scored per genotype. (J) PTTH neuronal silencing does not evoke growth compensation and results in larger pupae, as compared with controls. Data are mean ± SD (n = 3 independent repeats), and a total of 35 pupae were measured per genotype. *P < 0.05 (unpaired t test). AU, arbitrary units. (K) Expression of Eip75B at 100 hours AEL in control larvae (ptth-Gal4) and larvae with electrically silenced PTTH neurons (ptth>mKir2.1). ***P < 0.001 (unpaired t test). mRNA was obtained from seven larvae per genotype, and the experiment was repeated three times. Scale bars, 50 μm in (A) [also applies to (C)]; 60 μm in (D) [also applies to (D′) and (D′′)]; 40 μm in (E) and (G); and 30 μm in (F).

Brainbow-assisted analysis and the pre- and postsynaptic markers revealed that Lgr3 neurons in the dorsomedial region extend both ipsilateral and contralateral axon projections (Fig. 5, D to D′′), with thin dendrites descending into the VNC (Fig. 5, B and C, and movie S3). Brainbow analysis also suggests that a dialogue is maintained between the distinct cell subpopulations defined by R19B09-Gal4 and that this converges on the synaptic sites of the Lgr3 dorsomedial neurons (green neurons in brainbow image; Fig. 5, D and D′′).

Spatial overlap between axonal and dendritic arborization is a prerequisite for potential connectivity between defined neurons and their potential targets. In this sense, the dense presynaptic sites of Lgr3 neurons indicate strong connectivity between these neurons and the PTTH neurons and IPCs. Thus, to detect direct connections, we used GRASP (GFP reconstitution across synaptic partners) analysis, which is based on the expression of two nonfluorescent split-GFP fragments (spGFP1-10 and spGFP11) tethered to the membrane in two neuronal populations (40). We used R19B09-LexA (24) to drive expression of LexAop-spGFP11 and dilp3-Gal4 or ptth-Gal4 to drive expression of UAS-spGFP1-10 in IPCs and PTTH neurons, respectively (Fig. 5, E and F). When paired with R19B09-LexA, strong, specific GRASP signals were observed for IPCs (Fig. 5E and fig. S6, A and B). GRASP signals also suggest possible connections between Lgr3 and PTTH neurons, as detected by immunofluorescence (anti-GFP, Invitrogen) (Fig. 5F) as in (40). This punctate staining was lacking in control brains (fig. S6C). We detected unreconstituted GFP, using immunofluorescence resulting from expression of the spGFP1-10 fragment at the PTTH soma and axons (compare panels C and F in fig. S6). Immunofluorescence staining of PTTH neurons and signals of GRASP between IPCs and Lgr3 revealed probable synaptic contact sites in the circuit (Fig. 5G, asterisks, and fig. S6, D to H). These data suggest that Lgr3 neurons link Dilp8 input to IPCs and/or PTTH neurons to form a homeostatic circuit for synchronizing growth with maturation timing for body-size regulation.

Inhibition of PTTH neurons and mechanism for puparium delay

We reasoned that activation of Lgr3 neurons would delay pupariation by suppressing PTTH synaptic targets, so we used electrical silencing and genetic tools to investigate functional communication between Lgr3 and PTTH neurons. A PTTH receptor mutation (torsoRL3) that produces a constitutively active receptor has previously been shown to accelerate puparion formation in heterozygosis by 9.2 hours (41). Introducing the torRL3 allelic mutation in tub-dilp8 animals prevented pupariation delay (Fig. 5H) to the same extent as depleting lgr3 (Fig. 1B) or feeding larvae with the active form of ecdysone, 20E ecdysone (12). These observations, coupled with the anatomical and genetic data, establish that the Dilp8-Lgr3 axis acts upstream of the PTTH-torso network, probably by suppressing PTTH neuron activity.

We wanted to probe the sufficiency of electrically silencing the PTTH neurons to delay the timing of the larval-pupal transition. Thus, we tested the effect of hyperpolarization of the membrane of PTTH neurons by expressing the potassium channel mKir2.1, which has proven to be a highly effective approach for shunting neuronal activity in excitable neurons (42). Expression of UAS-mKir2.1 in the PTTH neurons using ptth-Gal4 produced a larval-pupal transition delay of 12 hours compared with WT controls (Fig. 5I and see Fig. 5K for measurement of ecdysone signaling). This is similar to the effect of genetic ablation of PTTH neurons or genetic inactivation of the Ptth gene by RNAi reported previously (14). Hence, and as predicted (14), the release of PTTH that triggers the larval-pupal transition is related to PTTH neuron activity. However, electrical hyperexcitation by expressing the UAS-NaChBac ion channel by ptth-Gal4 could neither accelerate pupariation nor prevent Dilp8-induced delay (fig. S7). It is possible that the release of PTTH at the larval-pupal transition might additionally require disinhibition of inhibitory input(s), as proposed for the secretion of GnRH from hypothalamic neurons at the onset of puberty (5, 10). Electrical silencing of PTTH neurons did not trigger a compensatory growth response (Fig. 5J), and therefore, the animals bred after the extended larval period were larger than normal. Thus, the coupled control of the growth rate probably involves the other branch (the IPCs) of the Lgr3 neuronal circuit.

The IPCs as an Lgr3 output pathway and the role of JH in growth compensation

Our previous study showed that dilp8 overexpression reduces the growth rate associated with a reduction in insulin-like peptide dilp3 (12). Hence, we tested the possibility that this transcriptional modulation in the postsynaptic target (IPCs) may be a consequence of the inhibitory input to IPCs from the Lgr3 neurons. Ablation or electrical silencing of IPCs produces adults that are much smaller than normal (9, 16), suggesting that size compensation via Dilp8 is unlikely to affect insulin signaling globally or completely. IPCs modulate growth systemically via circulating insulin-like peptides (such as Dilp2, -3, and -5) and via endocrine mechanisms, such as direct regulation of JH synthesis in the CA (23, 43), which was also recently shown to instructively regulate larval growth in Drosophila (44, 45). We therefore examined the expression of candidate output pathways as a read-out of the physiological dialogue between Lgr3 neurons that directly contact the IPCs and the regulation of JH.

Because JH titer is normally determined by its rate of biosynthesis by the larval CA gland, as well as its rate of degradation, we used qRT-PCR to measure the expression of a gene encoding a key biosynthetic enzyme [juvenile hormone acid methyltransferase (JHAMT)] (43) and the direct target of JH that encodes a transcription factor that transduces the actions of JH [kruppel-homolog-1 (kr-h1)] (46, 47). Together, these elements should allow us to detect the effective JH signaling in tub-dilp8 animals compared with age-synchronized and population-controlled WT animals, as well as tub-dilp8 animals with depleted lgr3 in the neurons labeled by R19B09-Gal4. We also measured the transcriptional levels of Eip75B, a direct target of the ecdysone receptor, as a read-out for ecdysone signaling (14).

Control larvae experience a steep increase in Eip75B level at 100 hours after egg laying (AEL), which reflects the surge of ecdysone levels at the time of pupariation in our experimental conditions. As expected, no such accumulation was observed in tub-dilp8 larvae at 100 hours AEL, but expression of Eip75B was restored to almost normal levels in tub-dilp8 in which the lgr3 receptor was knocked down in neurons labeled by R19B09-Gal4 (Fig. 6A). The levels of dilp3 and dilp5, which are known to respond to nutrition and stress (16) (Fig. 6B), and of JHAMT and kr-h1 in JH biosynthesis and signaling (Fig. 6C) were also significantly down-regulated in tub-dilp8 larvae, and their expression was restored to almost-normal levels by specific knockdown of lgr3 in R19B09 neurons. This non–cell-autonomous effect was specific because the expression of dilp2 was not altered (fig. S8).

Fig. 6 Probing functional connection between Lgr3 neurons and IPCs neurons.

(A to C) Expression of Eip75B (A), dilp3 and dilp5 (B), JHAMT [(C), left] and kr-h1 [(C), right] genes analyzed by qRT-PCR in mRNA isolated from ~10 larvae for each genotype and age [~90 hours AEL (white bar) and 100 hours AEL (all other bars)]. Overexpression of dilp8 by the tubulin promoter (tub-dilp8) sustainably decreased dilp3, dilp5, JHAMT, and kr-h1 transcripts in the extended third-instar larval period, and this regulation was abrogated by specific knockdown of lgr3 in R19B09 neurons. Data are mean ± SD (n = 3 repeats). *P < 0.05; **P < 0.01; ***P < 0.001 (two-tailed unpaired t test). IIS, insulin/IGF-like signaling. (D) Treatment of JHA abrogates the compensatory growth response of tub-dilp8 animals. Box-and-whisker plots of the pupal volume of control animals (tub>dilp8C150A) (left) and animals overexpressing dilp8 (tub>dilp8) (right). Any differences between tub>dilp8C150A animals fed with methoprene [Met] or without (control) are not significant. Met-fed tub>dilp8 animals produced noticeably larger pupae. Plotted data are pooled from three biological repeats (50 pupae per genotype and treatment). ***P < 0.001 (two-tailed unpaired t test). (E) Model for plastic and homeostatic regulation of body size. If the growth rate is fixed, an extension of the developmental time results in larger adults (plastic regulation). If developmentally delayed animals also experience a proportional decrease in growth rate, they will reach a normal adult size (homeostatic regulation). Developmental time (days) represents age at the larval-pupal transition. (F) Model of Lgr3 circuit and output pathways. Dilp8 produced by peripheral tissues conveys information about overall growth status. Circulating Dilp8 enters the brain through the blood-brain barrier, in an unknown manner, and binds and activates Lgr3 in a dose-dependent manner. With their high connectivity, Lgr3-responding neurons distribute this growth-status information to IPCs (gray) and PTTH neurons (red), which ensures that rates of growth and maturation are matched and co-regulated according to the intensity of the Dilp8 signals.

Although the exact mechanism by which Lgr3 neurons influence JH synthesis and signaling is not known, we attempted to establish a causal role for the observed reduction in JH signaling by pharmacological means. To control for the genetic background, we used UAS-dilp8 and tub-Gal4 (tub>dilp8) to treat animals overexpressing dilp8; the biologically inactive UAS-dilp8C150A peptide hormone served as a control (12). The tub>dilp8 larvae reach the correct pupal size and adult size [as assessed by measuring pupal volume and adult wing size and shape (12)]. In contrast, tub>dilp8 larvae fed with the JH analog (JHA) methoprene produced significantly larger pupae (~25%) than did control animals (Fig. 6D). The onset of pupariation was slightly delayed (~6 hours) and resulted in 100% lethality, which consequently prevented us from measuring adult size. Treatment of control tub>dilp8C150A animals that display normal pupation time (12) did not increase their size above that of their untreated siblings (Fig. 6D, right) (44). Thus, we conclude that reduced JH signaling diminishes larval growth, contributing to ensuing normal-sized tub>dilp8 animals (Fig. 6E).


Our data provide strong evidence that Dilp8 signals for organismal and organ homeostatic size regulation are transduced via the orphan relaxin receptor Lgr3 and that activation of Lgr3 in molecularly defined neurons mediates the necessary hormonal adjustments for such homeostasis. Human insulin/relaxin-like peptides are transduced through four GPCRs: RXFP1 to -4. RXFP1 and -2 are characterized by large extracellular domains containing leucine-rich repeats, similar to fly Lgr3 and Lgr4 receptors (25, 26). Additionally, as for Lgr3 (this study), activation of RXFP1 and -2 by their cognate ligand binding stimulates increased cAMP production (25). RXFP3 is distinctly different in structure from fly Lgr3 (25), and its biochemical properties are also distinct, but RXPF3 is analogous to fly Lgr3 in the sense that it is found in highest abundance in the brain, suggesting important central functions for relaxin 3/RXFP3 (48, 49). However, a function in pubertal development and/or growth control for vertebrate relaxin receptors is presently unknown.

The neuronal populations that regulate body size and, in particular, the mechanisms by which their regulation generates size variations (plasticity) in response to internal and environmental cues (such as nutrition) have been investigated thoroughly (4, 9, 14, 16, 45, 46). Less is known about how the brain stabilizes body size to ensure that developing organisms reach the correct, genetically determined size. We also do not know how limbs grow to precisely match the size of their contralateral limbs, nor do we understand how limbs maintain proportion with other body parts, even when faced with perturbations (this statement is also applicable to other bilaterally symmetric traits). Paired organs are controlled by an identical genetic program and grow in the same hormonal environment, yet small deviations in size can occur as a result of developmental stress, genetic noise, or injury. Imperfections in symmetry thus reflect the inability of an individual to counterbalance variations and growth abnormalities.

Our study shows that without lgr3, the brain is unable to detect growth disturbances and, more importantly, cannot adjust the internal hormonal environment to allocate additional developmental time for restoring affected parts or catching up on growth. Without lgr3, the brain is also not capable of retarding the growth rate to compensate for the extra time so that unaffected and affected tissues can develop with normal size, proportionality, and symmetry. Our study also identifies the Lgr3-expressing neurons necessary and sufficient to respond to Dilp8. Moreover, using a cAMP sensor, we have identified a pair of neurons that are highly sensitive to Dilp8.

Communication in neuronal networks is essential for synchronization and efficient performance. Notably, although most neurons have only one axon, Lgr3-responding neurons display extensive axonal arborizations reminiscent of hub neurons (39). GRASP analyses show that Lgr3 neurons are broadly connected with the IPCs and, to a lesser extent, with PTTH neurons, linking (Dilp8) inputs to the neuronal populations that regulate the key hormonal outputs modulating larval and imaginal disc growth. Furthermore, the information flow from Lgr3 neurons to IPCs and PTTH may explain how the brain matches growth with maturation in response to Dilp8 (Fig. 6F). This brain circuit provides the basis for studying how the brain copes with genetic and environmental perturbations to stabilize body size, proportions, and symmetry, all of which are vital for survival.

Materials and methods

Drosophila husbandry

The five lgr3 enhancer Gal4 lines (R17G11-Gal4, R17H01-Gal4, R18A01-Gal4, R18C07-Gal4, and R19B09-Gal4); the R17G11-LexA, R19B09-LexA, and 13xLexAop2-IVS-myr::GFP lines; and the retnR9F04-Gal4 line are from the Janelia Farm Collection (Howard Hughes Medical Institute, Ashburn, VA). The da-Gal4, dilp3-Gal4, dpp-Gal4, elav-Gal4, tub-Gal4, P0206-Gal4, NPF-Gal4, pdf-Gal4, per-Gal4, ptth-Gal4, tsh-Gal4, LexAop-CD4::spGFP11, UAS-CD4::spGFP1-10, UAS-dcr2, UAS-Denmark, UAS-DsRed, UAS-Flp, UAS-lgr3-TRiP.GL01056-RNAi, UAS-mCD8::GFP, UAS-mCD8::RFP, UAS-mKir2.1, UAS-NaChBac, and UAS-Syt::GFP lines are from the Bloomington Stock Center at Indiana University (Bloomington, IN). UAS-dilp8 and UAS-dilp8C150A are described in (12). CRE-F-luc and torRL3 were gifts from J. C. P. Yin and J. Casanova, respectively (36, 50).

Flies were reared in standard “Iberian” fly food at 25°C (except when indicated) on a 14:10-hour light:dark cycle (surrogate of laboratory summer time). Standard Iberian fly food consisted of 15 liters of water, 0.75 kg of wheat flour, 1 kg of brown sugar, 0.5 kg of yeast, 0.17 kg of agar, 130 ml of a 5% nipagin solution in ethanol, and 130 ml of propionic acid.

G-TRACE analysis

G-TRACE (Gal4 technique for real-time and clonal expression) analysis was performed by crossing UAS-Flp, UAS-RedStinger, and ubip63-FRT-stop-FRT-StingerGFP stocks (51) with R19B09-Gal4, retnR9F04-Gal4, and P0206-Gal4 lines.

Brainbow clones

We built hs-Cre; R19B09-Gal4 stocks and crossed them with UAS-dBrainbow (38) virgin females. We reared fly crosses at 25°C and did not heat-shock them. The following primary antibodies were used: rabbit anti-HA (1/500; Abcam) and mouse anti-V5 (1/500, Invitrogen).

GRASP analysis

We built R19B09-LexA; LexAop-CD4::spGFP11, UAS-CD4::spGFP1-10/TM6B stocks and crossed them with dilp3-Gal4/CyO-GFP; UAS-mCD8::RFP (n = 14 larval brains were analyzed) or ptth-Gal4/CyO-GFP (n = 43 larval brains were analyzed). Control experiments were performed by staining larval brains of the following genotypes (n = 10 larval brains per genotype were analyzed): R19B09-LexA; LexAop-CD4::spGFP11, UAS-CD4::spGFP1-10/TM6B, dilp3-Gal4/+; LexAop-CD4::spGFP11, UAS-CD4::spGFP1-10/+, or ptth-Gal4/+; LexAop-CD4::spGFP11, UAS-CD4::spGFP1-10/+. The following primary antibodies were used: rabbit anti-GFP (1/2000; Invitrogen) to detect GRASP signal between PTTH and Lgr3 neurons, guinea pig anti-PTTH [1/500 (52)], and rat anti–Drosophila E-Cadherin (anti-DE-Cad) [1/50, Developmental Studies Hybridoma Bank (DSHB)] to counterstain larval brains.

Confocal imaging and immunohistochemistry in brains

Brains were dissected in cold phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde (PFA) for 20 min (53), and stained with the following primary antibodies: guinea pig anti-PTTH [1/500 (52)], mouse anti-luciferase (1/200, Thermo Fisher Scientific), rabbit anti-Dilp2 [1/500 (54)], rabbit anti-Mira [1/2000 (55)], rabbit anti-Pdp1 [1/1000 (56)], and rat anti-DE-Cad (1/50, DSHB). Secondary antibodies were purchased from Invitrogen and Jackson ImmunoResearch. The brains were mounted in Vectashield (Vector Labs), maintaining their three-dimensional (3D) configuration (53), and images were obtained on a Leica TCS SP2 confocal microscope. Z stacks were recorded at 1-μm intervals. 3D reconstructions of individual WT Drosophila larval brains were created using Imaris software (Bitplane, Zurich, Switzerland). To assess changes in cAMP levels in the larval brain, we used the in vivo CRE-F-luc reporter system (36). Dissected brains were stained using mouse anti-luciferase (1/200, Thermo Fisher Scientific).

Generation of DNA constructs and transgenic lines

For the tub-dilp8::FLAG construct, dilp8 cDNA was C-terminally fused in frame to the 3xFLAG coding sequence (12) and cloned into the pCasper-tubulin promoter plasmid at the KpnI/NotI sites.

The lgr3 WT cDNA sequence was based on the WT amino acid sequence corresponding to GenBank accession number AAF56490, codon-optimized using GeneOptimizer (GENEART), and cloned into the pMK-RQ plasmid (SfiI/SfiI sites) (GENEART). The obtained construct was verified by sequencing and then cloned into the pUASt plasmid at the EcoRI/NotI sites.

Constructs were injected in w1118 embryos following standard P-element–mediated transformation procedures (BestGene).

The entire open reading frame (ORF) sequence of lgr4 was PCR-amplified from total mRNA obtained from w1118 larvae using primers containing attB1 and attB2 Gateway recombination sites whose sequences were as follows:

lgr4 forward primer: 5′-ATGTGTATAGCTCACCTGCCTATCAC-3′

lgr4 reverse primer: 5′-CTACAGATAGCTCATCTGCCGGTGTG-3′

The amplified product was cloned into a pDON/Zeo entry vector (Life Technologies), according to the Gateway technology manual (Life Technologies), and verified by sequencing. Verified entry clones were used to introduce full-length ORFs into the pUASt Gateway plasmid [Drosophila Genomics Resource Center (DGRC), stock no. 1129] by LR recombination (Gateway technology, Life Technologies).

The Lgr3 extracellular domain (hereafter Lrg3-ECD) (amino acid residues 1 to 433) was PCR amplified from the pUASt-lgr3 full-length plasmid described above using primers containing attB1 and attB2 Gateway recombination sites with the following sequences:


lgr3-ECD reverse primer: 5′-CAGCACGGGCTTGCTCAGCAGGTC-3′

The PCR product was cloned into pDON/Zeo entry vector (Life Technologies) following the Gateway technology manual instructions (Life Technologies) and verified by sequencing. Entry Lrg3-ECD construct was used to introduce the insert into the pUASt-C-terminal 3xHA Gateway plasmid (DGRC, stock no. 1100) by LR recombination (Gateway technology, Life Technologies).

Cell culture

Drosophila S2 and Kc cells (Invitrogen) were cultured in Schneider’s Drosophila medium (Invitrogen) supplemented with 10% fetal bovine serum (Gibco, Thermo Fisher Scientific) at 25°C in a nonhumidified, ambient-air–regulated incubator.

cAMP measurement

Concentrations of cAMP were measured using the cAMP Enzymeimmunoassay (EIA) System (Amersham, catalog no. RPN2251). Drosophila Kc cells (Invitrogen) were seeded (50.000 per well) in a 96-well plate and transfected with the plasmid DNA indicated, using Fugene-HD (Promega). After 36 hours of transfection, cells were exposed for 30 min to the Dilp8 peptide (at 0, 5, or 50 nM; Phoenix Pharmaceuticals, catalog no. 035-79) and 3-isobutyl-1-methylxanthine (IBMX) (100 μM; Sigma, catalog no. I5879). After treatment, the cells were processed according to the manufacturer’s protocol, and the cAMP concentrations are presented as femtomoles per well.

EC50 determination

The cAMP concentrations were determined using the Direct cAMP ELISA Kit (Enzo Lifescience, catalog no. ADI-900-066) according to the manufacturer’s protocol. 50.000 Drosophila Kc cells (Invitrogen) were seeded per well in a 96-well plate and transfected with the indicated plasmid DNA, using Fugene-HD (Promega). After 36 hours of transfection, cells were exposed for 30 min to IBMX (100 μM) and Dilp8 peptide (12 serial dilutions of Dilp8 starting from 250 nM to 0 nM). Total intracellular cAMP concentration was determined using the nonacetylation cAMP enzyme immunoassay from 100 μl of 0.1 M HCl in all experiments. cAMP levels were calculated using the 4 Parameter Logistic Curve (4PL) online data analysis tool (MyAssays). Results are expressed in picomoles per milliliter of cAMP. The EC50 analysis was calculated with GraphPad Prism software (version 6, for Mac), using a sigmoidal dose-response (variable slope) equation.

Coimmunoprecipitation assays

Drosophila S2 cells (3 × 106 cells per 10-cm dish) were transiently cotransfected with pActin-Gal4 and either UAS-lgr3-ECD::3XHA or empty vector UAS::3XHA, using Fugene-HD (Promega). To obtain secreted tagged Dilp8, 6 × 106 S2 cells per 10-cm dish were transiently transfected using the pActin-Gal4 and UAS-dilp8::Myc plasmid. Thirty-six hours after transfection, supernatant containing Dilp8::Myc was collected, filtered, and used to replace the medium of UAS-lgr3-ECD::3XHA and UAS::3XHA dishes. After 2 hours of incubation, cells were PBS-washed and cross-linked for 30 min using DTSSP [3,3′-dithiobis(sulfosuccinimidyl propionate)] (Thermo Fisher Scientific). Cells were PBS-washed and then lysed using modified RIPA buffer containing proteinase inhibitors. Precleared extracts were incubated at 4°C with 1 μg of rabbit anti-Myc (Abcam, ab9606) or rabbit anti–immunoglobulin G (anti-IgG) (Sigma, I8140). After 2 hours, 25 μl of equilibrated protein A magnetic beads (Millipore, catalog no. 16-661) was added to each extract and incubated over night at 4°C. After three washes using the modified RIPA buffer containing proteinase inhibitors, proteins bound to beads were recovered by boiling for 10 min in 25 μl of 3x sample buffer and were separated by SDS–polyacrylamide gel electrophoresis. After blotting, membranes were incubated in rat anti-HA–horseradish peroxidase (clone 3F10, Roche) and analyzed.

Immunostaining of Drosophila cultured cells

Drosophila Kc cells (8 × 105 cells per well) were cotransfected in six-well plates with pActin-Gal4 and either UAS-lgr3-ECD::3XHA or empty vector UAS::3XHA and, in parallel, were cotransfected with tub-dilp8::FLAG using Fugene-HD (Promega). Thirty-six hours after transfection, supernatant containing Dilp8::FLAG was collected, filtered, and used to replace medium from UAS-lgr3-ECD::3XHA and UAS::3XHA. After a 2-hour incubation period, cells were washed twice with PBS, fixed using 4% PFA, and immunostained (cells were not permeabilized). Antibodies used: rabbit anti-HA (1/200, Abcam ab9110) and mouse anti-FLAG-M2 (1/200, Sigma).

Lgr3 antibody

To generate specific antiserum for Lgr3, two peptides corresponding to amino acids 719 to 733 (C+ GWKKITSRKRAEAGN) and 487 to 501 (C+ GVQDYRYRNEYYKVV) (57) were synthesized by Eurogentec (Seraing, Belgium) and used to immunize rabbits according to an 87-day polyclonal antibody program.

Quantitative RT-PCR

To assess mRNA levels, total RNA was extracted from Drosophila larvae using the RNeasy-Mini Kit (Qiagen). To remove contaminating DNA, RNA was treated with Turbo DNA-free (Ambion, Life Technologies). cDNA was synthesized with SuperScript III First-Strand Synthesis System for RT-PCR (Life Technologies) using oligo-dT primers. qRT-PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems), with gene-specific primers, on an ABI7500 apparatus (Applied Biosystems). Rp49 primers were used for mRNA normalization. Comparative qRT-PCRs were performed in triplicates, and relative expression was calculated using the comparative Ct method.

Primer sequences:




























Measurement of the developmental timing of pupariation

Females and males (20 to 30 of each) were crossed and, after 24 to 48 hours, flies were transferred to grape juice agar plates with yeast paste and left 4 hours for egg deposition. Parental flies were removed, and laid eggs were incubated 48 hours at 26.5°C. Second-instar larvae were transferred onto 5 ml of Drosophila standard Iberian food (20 larvae per tube) and reared at 26.5°C. A survey of the pupae was performed at 8-hour intervals, with “time 0” designated as 4 hours after the initiation of egg laying.

Weight and size measurements

For weighing adult flies, 20 to 30 females and 20 to 30 males were crossed and left 24 hours for egg deposition. Parental flies were transferred every 24 hours to fresh tubes, and laid eggs were reared at 26.5°C. Eclosed adult males of each genotype were collected (five groups of five individuals) and weighed after 12 to 24 hours, using a precision scale.

For pupae volume determination, 20 to 30 females and 20 to 30 males were crossed and left 24 hours for egg deposition. Parental flies were transferred every 24 hours to fresh tubes, and laid eggs were reared at 26.5°C. Pupae were collected and photographed with their dorsal side up. Length and width were measured using ImageJ; volume was calculated according to the following formula: v = 4/3π(L/2)(l/2)2 (L, length; l, width).

For adult wing measurements, 20 to 30 females and 20 to 30 males were crossed and left 24 hours for egg deposition. Parental flies were transferred every 24 hours to fresh tubes, and laid eggs were reared at 26.5°C. Adults were collected and left, and the right wings of each individual were excised and rinsed thoroughly with ethanol and mounted in a glycerol-ethanol solution. Wing areas were measured using ImageJ. Intraindividual variation of wing areas was calculated using fluctuating asymmetry index (FAi) as in (12), employing the formula FAi = Var(Ai), where Ai are the differences between left and right wing areas of each individual.

Juvenile hormone analog (methoprene) treatment

Males and females (20 to 30 of each) were crossed, and after 24 to 48 hours, flies were transferred to grape juice agar plates with yeast paste and left 4 hours for egg deposition. Parental flies were removed, and laid eggs were incubated 48 hours at 26.5°C. Second-instar larvae were transferred onto 5 ml of Drosophila standard Iberian food (20 larvae per tube) and incubated at 26.5°C. Larvae were transferred 24 hours later (72 hours AEL) to 3 ml of Drosophila standard Iberian food supplemented with a liquid solution of pure methoprene (Sigma, catalog no. 33375) at a Met:food ratio of 1 μm:1000 μm. An equivalent volume of water was added to the control.

Supplementary Materials

References and Notes

  1. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  2. Acknowledgments: We thank P. Leopold for reagents and the PTTH antibody; M. Vidal for the lgr2 plasmid; E. Hafen and H. Stocker for Dilp antibodies; A. Garelli for stock construct and initial work on this project; A. Gontijo for plasmid design; and J. Casanova, J. C. P. Yin, B. J. Dickson, J. Blau, B. Hassan, and F. Matsuzaki for fly stocks or antibodies. We also thank the Bloomington Stock Center (NIH grant P40OD018537), the DGRC (NIH grant OD010949-10), and the Transgenic RNAi Project at Harvard Medical School (NIH National Institute of General Medical Sciences grant R01-GM084947) for providing the fly stocks and plasmids used in this study and the Developmental Studies Hybridoma Bank at the University of Iowa for antibodies. This work was supported by a Ramon y Cajal grant (RyC-2010-07155) and a Ministerio de Economia y Competitividad grant (SAF2012-31467) to J.M.; a Junta de Andalucia PAIDI group CTS-569 to J.B.; and Spanish national grants (SAF2012-35181 and SEV-2013-0317) and a Generalitat Valenciana grant (PROMETEO II/2013/001) to M.D. M.D. was also supported by the Botin Foundation. S.J.-C. is a fellow from Formación del Personal Investigador (grant BES-2013-064947) from the Spanish Ministerio de Economia y Competitividad.

Stay Connected to Science

Navigate This Article