Drosophila insulin release is triggered by adipose Stunted ligand to brain Methuselah receptor

See allHide authors and affiliations

Science  30 Sep 2016:
Vol. 353, Issue 6307, pp. 1553-1556
DOI: 10.1126/science.aaf8430


Animals adapt their growth rate and body size to available nutrients by a general modulation of insulin–insulin-like growth factor signaling. In Drosophila, dietary amino acids promote the release in the hemolymph of brain insulin-like peptides (Dilps), which in turn activate systemic organ growth. Dilp secretion by insulin-producing cells involves a relay through unknown cytokines produced by fat cells. Here, we identify Methuselah (Mth) as a secretin-incretin receptor subfamily member required in the insulin-producing cells for proper nutrient coupling. We further show, using genetic and ex vivo organ culture experiments, that the Mth ligand Stunted (Sun) is a circulating insulinotropic peptide produced by fat cells. Therefore, Sun and Mth define a new cross-organ circuitry that modulates physiological insulin levels in response to nutrients.

Environmental cues, such as dietary products, alter animal physiology by acting on developmental and metabolic parameters like growth, longevity, feeding, and energy storage or expenditure (1). The systemic action of this control suggests that intermediate sensor tissues evaluate dietary nutrients and trigger hormonal responses. Previous work in Drosophila melanogaster established that a specific organ called the fat body translates nutritional information into systemic growth-promoting signals (24). The leptinlike Janus kinase–signal transducers and activators of transcription (JAK-STAT) ligand unpaired 2 and the CCHamid2 peptide are produced by fat cells in response to both sugar and fat and trigger a metabolic response (5, 6). Dietary amino acids activate TORC1 signaling in fat cells and induce the production of relay signals that promote the release of insulin-like peptides (Dilps) by brain insulin-producing cells (IPCs) (3, 7). Two fat-derived peptides (GBP1 and GBP2) activate insulin secretion in response to a protein diet, although their receptor and neural targets remain uncharacterized (8). To identify critical components of this organ crosstalk, we conducted a genetic screen in Drosophila larvae (fig. S1A). The gene methuselah (mth), which encodes a heterotrimeric GTP-binding protein (G protein)–coupled receptor belonging to the subfamily of the secretin-incretin receptor subfamily (912) came out as a strong hit. Impairing mth function in the IPCs reduces larval body growth (Fig. 1A), whereas silencing mth in a distinct set of neurons or in the larval fat body had no impact on pupal volume (fig. S1B). Larvae in which expression of the mth gene is reduced by RNA interference (RNAi), specifically in the IPCs (hereafter, dilp2>mth-Ri), present an accumulation of Dilp2 (Fig. 1B) and Dilp5 (fig. S1G) in the IPCs, whereas dilp2 gene expression remains unchanged (fig. S1H), a phenotype previously described as impaired Dilp secretion (13). Indeed, forced depolarization of the IPCs rescues pupal volume and Dilp2 accumulation upon IPC-specific mth depletion (knockdown) (fig. S1, J and K). Therefore, Mth is required for Dilps secretion and larval body growth.

Fig. 1 Mth and Sun are required for systemic growth.

(A) Three different mth RNAi constructs (KK 102303, TRIP 36823, and BA3) decrease pupal size when driven in the IPC (dilp2>) (n > 20). (B) Silencing mth in the IPC induces accumulation of Dilp2 (n > 20). Fluorescence intensity measured as fold change (f.c.). (C) Silencing sun in the fat body (lpp-Gal4>sun-Ri) reduces pupal volume. No defect is observed when silencing in other tissues (myo1d-Gal4, gut; MHC-Gal4, muscle; elav-Gal4, CNS; or dilp2-Gal4, IPC). (D) sun silencing in the fat body causes Dilp2 accumulation in the IPC (n > 60). (E) sun overexpression in the fat body partially rescues the pupal size reduction observed in larvae fed a low-protein diet compared with those fed a high-protein diet (n > 20). Amino acid, aa. Red dashed line here and below shows the level of controls. (F) Overgrowth observed upon forced fat body expression of sun in larvae fed a low–amino acid diet is observed in mth1/+ but not in mth1/1 homozygous flies (n > 40). In graphs, means are shown, and error bars represent ±SEM; **P < 0.01.

Two peptides encoded by the stunted (sun) gene, SunA and SunB, serve as bona fide ligands for Mth and activate a Mth-dependent intracellular calcium response (14, 15) (see fig. S3E for peptide map). Silencing sun in fat cells, but no other larval tissue, of well-fed larvae mimics the mth loss-of-function phenotype (Fig. 1, C and D, and fig. S1I) with no effect on the developmental timing (fig. S1L). Conversely, overexpression of sun in the larval fat body (lpp>sun) partially rescues the systemic growth inhibition observed upon feeding larvae a diet low in amino acids (Fig. 1E and fig. S1M) or upon “genetic starvation” [silencing of the slimfast (slif) gene in fat cells (3)] (fig. S1N). This growth rescue is abolished in mth1 homozygous mutants (Fig. 1F). This shows that Sun requires Mth to control growth. However, sun overexpression has no effect in animals fed a normal diet (Fig. 1E). A modification of sun expression does not prevent fat body cells from responding to amino acid deprivation as seen by the level of TORC1 signaling, general morphology, and lipid droplet accumulation (fig. S2, A and B) but affects the ability of larvae to resist to starvation (fig. S2C).

Dilp2-containing secretion granules accumulate in the IPCs following starvation and are rapidly released upon refeeding (7) (fig. S3A). Mth is required in the IPCs to promote Dilp secretion after refeeding (Fig. 2A and fig. S3B), and forced membrane depolarization of IPCs using a bacterial sodium channel (dilp2>NaChBac) is dominant over the blockade of Dilp2 secretion in dilp2>mth-Ri animals (Fig. 2A). This dominance indicates that Mth acts upstream of the secretion machinery. In addition, Dilp2 secretion after refeeding is abrogated in lpp>sun-Ri animals (Fig. 2B), and overexpression of sun in fat cells prevents Dilp2 accumulation upon starvation (Fig. 2B). Altogether, these findings indicate that Mth and its ligand Sun are two components of the systemic nutrient response controlling Dilp secretion.

Fig. 2 Mth and Sun control Dilp2 secretion from brain IPCs.

(A) Representative pictures of Dilp2 staining in IPCs (dashed outline) showing the kinetics of Dilp2 accumulation upon refeeding after prolonged starvation. Genotypes are as indicated. (B) sun silencing (lpp>sun-Ri) in the fat body prevents Dilp2 secretion upon refeeding, whereas sun overexpression (lpp>sun OE) reduces Dilp2 accumulation upon starvation. Graphs represent quantifications of ΔDilp2 fluorescence relative to lpp>w control after 90 min of feeding (means ± SEM; n > 20).

Hemolymph from fed animals triggers Dilp2 secretion when applied to brains dissected from starved larvae (7) (Fig. 3A). This insulinotropic activity requires the function of Mth in the IPCs (Fig. 3A and fig. S3C) and the production of Sun by fat body cells (Fig. 3B). Conversely, overexpressing sun in the fat body (lpp>sun) is sufficient to restore insulinotropic activity to the hemolymph of starved larvae (fig. S3D). A 2-hour incubation with a synthetic peptide corresponding to the Sun isoform A (Sun-A) is also sufficient to induce Dilp secretion from starved brains (Fig. 3C). A similar effect is observed with an N-terminal fragment of Sun (N-SUN) that contains the Mth-binding domain (14, 15) but not with a C-terminal fragment (C-SUN) that does not bind Mth (fig. S3, E and F). The insulinotropic effect of N-SUN is no longer observed in brains from larvae of the mth allele, mth1 (fig. S3F). This absence of effect indicates that N-SUN action requires Mth in the brain. In addition, preincubation of control hemolymph with antiserum containing Sun antibodies specifically suppresses its insulinotropic function (Fig. 3D). These results indicate that Sun is both sufficient and necessary for insulinotropic activity in the hemolymph of protein-fed animals.

Fig. 3 Sun is a fat body–derived insulinotropic signal.

(A) Hemolymph collected from fed, but not starved, larvae activates Dilp2 secretion when incubated on dissected brains from starved control (dilp2>w), but not mth-deficient (dilp2>mth-Ri), larvae. (B) Hemolymph from fed larvae deficient for adipose sun (lpp>sun-Ri) does not induce Dilp2 secretion. (C) Incubation of brains from starved larvae with various concentrations of SUN-A stimulates Dilp2 secretion. DMSO, dimethyl sulfoxide. (D) Hemolymph collected from fed larvae preincubated with preimmune serum (fed+PPI), but not from larvae treated with antibodies against SUN (αSUN), induces Dilp2 secretion. Titration of αSUN with blocking peptides allows reactivation of Dilp2 secretion. Diluted (dil.) blocking peptides do not block αSun action [fed+SUNpep(dil.) + αSUN]. Graphs represent quantifications of ΔDilp2 fluorescence relative to control brains (brains from starved larvae incubated with hemolymph, from fed larvae, or with DMSO) (means ± SEM; n > 20); **P < 0.01.

To directly quantify the amount of circulating Sun protein, we performed Western blot experiments on hemolymph using antibodies against Sun. A 6-kD band was detected in hemolymph collected from fed larvae (Fig. 4A), and size was confirmed using Schneider 2 (S2) cell extracts (Fig. 4C). The band intensity was reduced upon sun knockdown in fat body cells but not in gut cells (Fig. 4C). Therefore, circulating Sun peptide appears to be mostly contributed by fat cells, as suggested by functional experiments (see Fig. 1C). The levels of circulating Sun are strongly reduced upon starvation (Fig. 4A). In line with this, sun transcripts are drastically reduced after 4 hours of protein starvation and start increasing after 1 hour of refeeding (Fig. 4B), whereas expression of the sun homolog CG31477 is not modified (fig. S4A). sun transcription is not affected by blocking TORC1, the main sensor for amino acids in fat body cells (3) (lpp>TSC1/2 in fig. S4B). However, adipose-specific TORC1 inhibition induces a dramatic reduction of circulating Sun (Fig. 4C), indicating that TORC1 signaling controls Sun peptide translation or secretion from fat cells. PGC1-Spargel is a transcription activator, the expression of which relies on nutritional input (fig. S4D) (16). We find that PGC1 is required for sun transcription (Fig. 4D) and that fat body silencing of PGC1 and sun induce identical larval phenotypes (Fig. 4E and fig. S4C). Although PGC1 expression is strongly suppressed upon starvation, blocking TORC1 activity in fat cells does not reduce PGC1 expression (fig. S4E). Conversely, knocking down PGC1 does not inhibit TORC1 activity (fig. S4F). This finding suggests that PGC1 and TORC1 act in parallel. Therefore, Sun production by fat cells in response to nutrition is controlled at two distinct levels by PGC1 and TORC1.

Fig. 4 The level of circulating Sun relies on TOR and PGC1 in fed animals.

(A) Circulating Sun peptide is detected by Western blotting of hemolymph from fed larvae (lanes 1 and 2) but not starved larvae (lanes 3 and 4). Antibodies against Crossveinless d lipoprotein (αCv-d) are used as a loading control. Quantification in arbitrary units (a.u.) of the normalized circulating Sun detected in the hemolymph according to nutritional conditions (means ± SEM; n = 4); **P < 0.01. (B) The sun transcript levels in the fat body decrease upon starvation and increase upon refeeding (measured by quantitative reverse transcription polymerase chain reaction) (means ± SEM; n = 4); **P < 0.01. (C) Circulating Sun levels decrease upon sun silencing in fat body cells (lpp>sun-Ri) but not in gut cells (myo1d>sun-Ri). Blocking TORC1 in fat cells (lpp>TSC1/2) strongly decreases circulating Sun. Quantification of normalized circulating Sun. (D) sun expression is severely reduced when PGC1 is silenced in fat body cells (means ± SEM; n = 3); **P < 0.01. (E) Silencing PGC1 in the fat body (lpp>PGC1-Ri) decreases pupal size (means ± SEM; n > 40); **P < 0.01. (F) Forced fat body expression of GFP-Sun (lpp>GFP-Sun) or Sun-GFP (lpp>Sun-GFP) rescues pupal size reduction observed from larvae fed the low–amino acid diet (means ± SEM; n > 40); **P < 0.01.

The Sun peptide is identical to the ε subunit of the mitochondrial F1F0-adenosine triphosphatase (F1F0-ATPase) synthase (complex V) (14, 17). Indeed, both endogenous Sun and Sun labeled with a hemagglutinin tag (Sun-HA) (fig. S5A) colocalize with mitochondrial markers in fat cells (fig. S5B), and the Sun peptide cofractionates with mitochondrial complex V in blue native polyacrylamide gel electrophoresis (fig. S5C). In addition, silencing sun in fat cells decreases mitochondrial Sun staining (fig. S5B) and the amounts of adenosine triphosphate (ATP) (fig. S5D). However, recent evidence indicates that an ectopic (ecto) form of the F1F0-ATP synthase is found associated with the plasma membrane in mammalian and insect cells (1821). In addition, coupling factor 6, a subunit of complex V, is found in the plasma (22). Therefore, Stunted could participate in two separate functions carried by distinct molecular pools. To address this possibility, we used a modified form of Stunted carrying a green fluorescent protein (GFP) tag at its N terminus (GFP-Sun), next to the mitochondria-targeting signal (MTS) (fig. S5A). When expressed in fat cells, GFP-Sun does not localize to the mitochondria (fig. S6A), contrarily to a Sun peptide tagged at its C-terminal end (Sun-GFP) (fig. S6C). This suggests that addition of the N-terminal tag interferes with the MTS and prevents mitochondrial transport of Sun. However, both GFP-Sun and Sun-GFP are found in the hemolymph (fig. S6B) and rescue pupal size and Dilp2 accumulation in larvae fed a low–amino acid diet as efficiently as wild-type Sun (wt-Sun) (Figs. 4F and 1E and fig. S6E) and do so in a mth-dependent manner (fig. S6D). This indicates that the growth-promoting function of Sun requires its secretion but not its mitochondrial localization and suggests the existence of one pool of Sun peptide located in the mitochondria devoted to F1F0-ATP synthase activity and ATP production and another pool released in the hemolymph for coupling nutrient and growth control. In this line, although fat body levels of Sun are decreased upon starvation (fig. S6F), its mitochondrial localization is not reduced (fig. S6G). This finding indicates that starvation affects a nonmitochondrial pool of Sun. In support of this, starved fat bodies contain normal levels of ATP and lactate (fig. S6, H and I), indicating that mitochondrial oxidative phosphorylation is preserved in fat cells in poor nutrient conditions. Last, other subunits from complex V (ATP5a) or complex I (NdufS3) were not detected in circulating hemolymph (fig. S6J). Therefore, the release of Sun in the hemolymph relies on a specific mechanism.

In conclusion, we provide evidence for a molecular cross-talk between fat cells and brain IPCs involving the ligand Stunted and its receptor Methuselah. Stunted is a moonlighting peptide present both in the mitochondria as part of the F1F0-ATP synthase complex and as an insulinotropic ligand circulating in the hemolymph. The mechanism of Stunted release remains to be clarified. The beta subunit of the ectopic form of F1F0-ATP synthase is a receptor for lipoproteins (1821), which serve as cargos for proteins and peptides. In addition, Drosophila lipid transfer particle–containing lipoproteins were shown to act on the larval brain to control systemic insulin signaling in response to nutrition (23). This suggests that Sun could be loaded on lipoproteins for its transport. Given the role of insulin–insulin-like growth factor (IGF) signaling in aging, our findings could help in understanding the role of Sun/Mth in aging adult flies (911, 13, 14).

The same genetic screen previously identified the fly tumor necrosis factor α Eiger (Egr) as an adipokine necessary for long-term adaptation to protein starvation (24), and recent work pointed to other adipose factors (5, 6, 8), illustrating the key role of the larval fat body in orchestrating nutrient response. The multiplicity of adipose factors and their possible redundancy could explain the relatively mild starvation-like phenotype obtained after removal of only one of them. Overall, these findings suggest a model whereby partially redundant fat-derived signals account for differential response to positive and negative valence of various diet components, as well as acute versus long-term adaptive responses.

Supplementary Materials

Materials and Methods

Fig. S1 to S6

References (2530)

References and Notes

Acknowledgments: We thank G. Jarretou for technical assistance and L. Martinez and laboratory members for discussions and comments on the manuscript. We thank B. Ja, the Vienna Drosophila RNAi Centers, the Drosophila Genetics Resource Center, and the Bloomington stock center for providing Drosophila lines. This work was supported by the CNRS; INSERM; European Research Council (advanced grant 268813); the French Foundation for Cancer Research, ARC (grant PGA120150202355); and the Labex SIGNALIFE program (grant ANR-11-LABX-0028-01). Funding for H.M. was provided by a Canadian Institutes of Health Research (CIHR) Foundation Grant. H.M. is a Tier 1 Canada Research Chair.

Stay Connected to Science

Navigate This Article