Drosophila Adiponectin Receptor in Insulin Producing Cells Regulates Glucose and Lipid Metabolism by Controlling Insulin Secretion

Adipokines secreted from adipose tissue are key regulators of metabolism in animals. Adiponectin, one of the adipokines, modulates pancreatic beta cell function to maintain energy homeostasis. Recently, significant conservation between Drosophila melanogaster and mammalian metabolism has been discovered. Drosophila insulin like peptides (Dilps) regulate energy metabolism similarly to mammalian insulin. However, in Drosophila, the regulatory mechanism of insulin producing cells (IPCs) by adipokine signaling is largely unknown. Here, we describe the discovery of the Drosophila adiponectin receptor and its function in IPCs. Drosophila adiponectin receptor (dAdipoR) has high homology with the human adiponectin receptor 1. The dAdipoR antibody staining revealed that dAdipoR was expressed in IPCs of larval and adult brains. IPC- specific dAdipoR inhibition (Dilp2>dAdipoR-Ri) showed the increased sugar level in the hemolymph and the elevated triglyceride level in whole body. Dilps mRNA levels in the Dilp2>dAdipoR-Ri flies were similar with those of controls. However, in the Dilp2>dAdipoR-Ri flies, Dilp2 protein was accumulated in IPCs, the level of circulating Dilp2 was decreased, and insulin signaling was reduced in the fat body. In ex vivo fly brain culture with the human adiponectin, Dilp2 was secreted from IPCs. These results indicate that adiponectin receptor in insulin producing cells regulates insulin secretion and controls glucose and lipid metabolism in Drosophila melanogaster. This study demonstrates a new adipokine signaling in Drosophila and provides insights for the mammalian adiponectin receptor function in pancreatic beta cells, which could be useful for therapeutic application.


Introduction
Mammalian adipokines are produced and secreted from adipose tissue. They play a key role in maintaining energy homeostasis through inter-organ communications. Adiponectin, one of the adipokines, has multiple beneficial roles for regulating energy homeostasis, inflammation, and apoptosis [1,2]. Two adiponectin receptors, AdipoR1 and AdipoR2, are seven transmembrane domain proteins with inverted topology compared to G-protein coupled receptors [3]. AdipoR1 has a higher binding affinity to the globular form of adiponectin whereas AdipoR2 has a higher binding affinity to the full length adiponectin [3]. AdipoR1 and 2 double knockout mice increase the triglyceride level in the liver and exhibit insulin resistance and glucose intolerance, demonstrating that AdipoR1 and 2 regulate lipid and glucose homeostasis [2,4]. In the skeletal muscle and liver, adiponectin receptors activate AMPK (AMP-activated protein kinase), PPARalpha, and p38 MAPK to increase the insulin sensitivity [3]. An adaptor protein APPL1 binds to adiponectin receptors, which activates AMPK and p38 MAPK in the skeletal muscle [5]. However, the mechanism of how adiponectin receptors activate downstream effectors is not made clear and the adiponectin receptor signaling identified in the skeletal muscle is not always applicable in other tissues. A recent study showed that adiponectin receptors are associated with ceramidase activity and regulate cell apoptosis by adjusting the balance between ceramide and sphingosine-1 phosphate levels [6]. Although AdipoR1 and 2 are expressed in pancreatic beta cells [7,8], the function of adiponectin and AdipoRs in IPCs is less studied than in insulin target tissues such as liver and skeletal muscle [1,2]. Adiponectin knockout mice show impaired insulin secretion and intravenous injection of adiponectin to C57BL/6 mice induces insulin secretion [9,10]. These studies indicate that adiponectin regulates insulin secretion but IPC-specific modulation of AdipoR in the animal model has not been demonstrated to show that adiponectin directly regulates insulin secretion through AdipoR.
During the last decade, significant conservation and parallelism were discovered between Drosophila and the mammalian metabolism. For example, Drosophila insulin like peptides (Dilps) regulate growth, energy metabolism, stress response, aging, and reproduction functions similar to that of mammalian Insulin/IGF signaling. Ablation of IPCs or deletion of Dilp genes results in decreased body size, retarded growth, and diabetic phenotypes such as an elevated circulating sugar level and altered stored lipid and carbohydrate levels [11][12][13][14][15]. There are eight Dilp genes in Drosophila genome, and four of them (Dilp 1, 2, 3 and 5) are expressed in IPCs of the brain. Recent studies demonstrate that Dilp production in IPCs is regulated by multiple factors such as neuropeptides, neurotransmitters, microRNA, O-GlcNAc metabolism [16][17][18][19][20][21][22][23]. However, Dilp secretion is not well studied in the Drosophila IPCs. Recently, mammalian leptin like unpaired 2 (upd2) signaling was discovered in Drosophila. When sugar and lipid are fed, Upd2 protein is produced from the fat body and regulates Dilp secretion through GABAnergic neurons in the fly brain [24]. In this report, we present the identification of Drosophila adiponectin receptor and its function on insulin secretion in IPCs of the fly brain.

Drosophila Culture and Stocks
Drosophila melanogaster were cultured at 25uC on standard cornmeal, yeast, sugar, agar diet. The stocks used in this study were UAS-dAdipoR-RNAi (VDRC 40936), wand UAS-AUG-DsRed

Identification of Drosophila Adiponectin Receptor Sequence
To find Drosophila orthologs of human adipokine and adipokine receptors, NCBI standard protein blast program blastp was used (http://blast.ncbi.nlm.nih.gov). Non-redundant protein sequence database of Drosophila melanogaster was blasted with the human adipokine and adipokine receptors protein sequences.

Measurement of Drosophila Body Weight, Size and Wing Size
To synchronize larval growth, the eggs were collected on the grape juice plate for 2 h and after 24 h, 50 hatched 1st instar larvae were transferred to a fly food vial. At 106-108 h after egg laying, the larval weight was measured. Then, the larvae were boiled for 3 minutes to measure the body length. For adult fly weight and wing length, 5 day-old male flies were used. More than 30 flies were used for each measurement.

Measurement of Total Body Triglyceride Level
In each time, 10 larvae or adult flies were ground in PBS solution and centrifuged. The supernatant was used for the analysis. Total glycerol and triglyceride levels were measured using a serum triglyceride determination kit (Sigma). The protein levels were measured in the same samples to normalize the triglyceride level.

Measurement of Trehalose and Glucose Levels in the Hemolymph
Larvae were starved on the water soaked filter paper for 4 h and 7.5% yeast/7.5% sucrose solution was fed for 30 mins. Adult flies were starved on 0.8% PBS-agar for 24 h and refed normal fly food for 2 h. Hemolymphs were collected from ten to fifteen flies. The concentrations of trehalose and glucose were measured as previously described [26].

Dilp2-FLAG ELISA Assay
Dilp2-FLAG ELISA assay was performed as previously described [27] with some modifications. 0.5 ml of hemolymph was collected from feeding third-instar larvae and diluted in PBS. The wells of the Immuno 96-well plate (Maxisorp TM , Nunc International) were coated overnight at 4uC with the diluted hemolymph. The next day, the wells were cleared and processed for ELISA assay. The primary antibody (anti-FLAG M2 antibody, 1:500, Sigma) was added to the wells and incubated at RT for 2 h. The HRP-conjugated secondary antibody (1:1000, SantaCruz Biotechnology) was treated at RT for 1 h. TMB solution (Thermo Scientific; Rochester, NY) was used for color development, and the optical density was measured at 450 nm. The standard curve for quantification was generated with serially diluted 36FLAG peptide (Sigma).

Starvation and High Fat Diet Resistance Assays
The starvation assay was performed as in Broughton et al., 2008. The high fat diet food was made by adding 20% coconut oil (vol/vol) to the normal fly food [28]. 3-5 day old female flies were collected from the normal fly food and transferred to testing media.
Quantitative RT-PCR Analysis cDNA synthesis and quantitative RT-PCR analysis were performed as previously described [16].

Generation of Antiserum and Immunohistochemistry
dAdipoR and Dilp2 antisera were generated by the custom antibody production services from Youngin Frontier Inc. (Seoul, Korea). These antibodies were produced by the immunization of rabbits with synthetic peptides (for anti-dAdipoR, EQAEEFVRKVWEASWK & SLWDKFSEPALRPLR; for anti-Dilp2, SEKLNEVLSMVC & TRQRQGIVERC). Animal care and all experiments in Youngin Frontier Inc. were conducted with the approval of Institutional Animal Care and Use Committee (IACUC) of Youngin Frontier Inc. The animal handling protocol was in accordance with institutional and international guidelines. Anti-dFOXO was gifts from O. Puig (Merck research laboratories). Immunostaining was performed as previously described [16]. Fluorescence images were acquired using a FluoView confocal microscope (Olympus) and an AxioVert 200 M microscope with Apotome (Carl Zeiss). Fluorescence intensity for Dilp2 immunostaining and secGFP was measured as previously described [25] with some modifications. Confocal Z stacks of IPCs (1 mm step size) were obtained with identical laser power and scanning parameters. Using Image J, Z-projected images were generated with the Sum Slices projection type. The raw integrated density was measured encompassing the IPC region in each image. A group of seven IPCs in each brain hemisphere was measured separately, and the measured fluorescence intensity was normalized to the mean fluorescence of starved or brain-only cultured IPCs of Dilp2-Gal4.

Ex vivo Culture
The brains were dissected from the larvae starved on the water for 20 h. The dissected brains were cultured in 20 ml of Schneider's medium with or without human globular adiponectin at the room temperature for 12 h and fixed for Dilp2 immunostaining. The recombinant human globular adiponectin was purchased from R&D Systems.

Statistical Analysis
Each experiment was repeated at least three times, and the data were presented as the mean and error bar (6S.E.M.). Student's ttest was used for the statistical analyses and p,0.05 was accepted as statistically significant.

dAdipoR, an Ortholog of the Mammalian Adiponectin Receptor 1, is Expressed in insulin Producing Cells
To identify Drosophila adipokine signaling, we searched orthologous genes of mammalian adipokines and their receptors in the Drosophila genome. Only dAdipoR (CG5315) was found with obvious homology. dAdipoR showed 66% amino acid sequences similarity to the human AdipoR1 ( Figure 1A), and hydropathy analysis predicted that dAdipoR has seven transmembrane domains ( Figure S1A). According to Flybase database, there are four isoforms of dAdipoR transcripts, which are dAdipoR A, B, C, and D. The isoforms A, C, and D are translated into the same 444 amino acids protein using the same start codon located in the 2 nd exon of the gene, while the isoform B is translated into the 362 amino acids protein using the start codon located in 4 th exon ( Figure S1B). Quantitative RT-PCR analysis revealed that the isoforms A, C, and D are predominant transcripts compared to the isoform B ( Figure S1C). dAdipoR mRNA was expressed throughout all developmental stages from embryo to adult and detected in the central nervous system (CNS), imaginal disc, salivary gland, fat body, gut, and malphigian tubules of the third instar larvae ( Figure  S1D, E). In the Drosophila brain, the immunohistochemical analysis with the dAdipoR antibody revealed that dAdipoR was expressed in IPCs of larval and adult brains ( Figure 1B, F, boxes). IPCs expression of dAdipoR was confirmed by the co-expression of IPCs marker, which is a DsRed reporter driven by Dilp2-Gal4 (Dilp2.DsRed) in the 3 rd instar larval and adult brains (Figure 1C-E, G-I). Beside IPCs expression, dAdipoR expression was additionally detected in neurons of the subesophageal region of larval and adult brains ( Figure 1B, F, arrows) and in lateral neurons of the adult brain ( Figure 1F, arrowheads).

dAdipoR Inhibition in insulin Producing Cells shows Metabolic Phenotypes
Based on the expression pattern of dAdipoR in the brain, we focused on the dAdipoR function in IPCs. To evaluate the function of dAdipoR in IPCs, we inhibited dAdipoR in IPCs by crossing IPCs-specific Dilp2-Gal4 driver and UAS-dAdipoR-RNAi (Dilp2.dAdipoR-Ri). UAS-dAdipoR-RNAi can inhibit all isoforms of dAdipoR transcripts ( Figure S1B). The quantitative RT-PCR analysis confirmed that the mRNA level of dAdipoR in the adult head of Dilp2.dAdipoR-Ri was reduced to 60% of the mRNA level of the Dilp2-Gal4 control ( Figure S2G). Moreover, immunostaining with the dAdipoR antibody showed that the dAdipoR protein level in the IPC of Dilp2.dAdipoR-Ri was reduced to 36% of the protein level of the Dilp2-Gal4 control ( Figure S2A-F). Since loss of Dilps by ablating IPCs results in small body size and metabolic defects [11,12], we examined body size and metabolic phenotypes in Dilp2.dAdipoR-Ri flies. The body size and weight of 3 rd instar feeding larvae (106-108 h AEL) and 5 day-old male flies were not changed compared with those of Dilp2-Gal4 and UAS-dAdipoR-RNAi control flies ( Figure S3A-D). However, hemolymph treha-lose and glucose levels of Dilp2.dAdipoR-Ri larvae and adults were significantly increased in the fed condition in comparison with those of controls and the starved conditions (Figures 2A, B). Triglyceride levels of Dilp2.dAdipoR-Ri larvae and adults also increased by 13-20% ( Figure 2C, D). Since IPC-specific dAdipoR inhibition flies stored excess lipids, we investigated starvation resistance. In the starved condition, Dilp2.dAdipoR-Ri flies survived longer than the Dilp2-Gal4 and UAS-dAdipoR-Ri control flies ( Figure 2E). In contrast, in the high fat diet condition, Dilp2.dAdipoR-Ri flies were more sensitive to the high fat diet than the Dilp2-Gal4 and UAS-dAdipoR-Ri controls. The median lifespan of Dilp2.dAdipoR-Ri flies was shorter compared to those of the control flies ( Figure 2F). After a 5 day high fat diet, the TAG level in Dilp2.dAdipoR-Ri flies increased compared to that in the controls. This suggests that the shorter lifespan of Dilp2.dAdipoR-Ri may be due to increased lipotoxicity ( Figure S3E). These metabolic phenotypes in Dilp2.dAdipoR-Ri flies are similar to those of Dilps inhibition by the ablation of IPCs [11,12].

dAdipoR Regulates Insulin Secretion in IPCs
To identify whether metabolic phenotypes in Dilp2.dAdipoR-Ri flies are due to defects in Dilps mRNA production, we tested expression levels of Dilp2, Dilp3 and Dilp5 which are known to be expressed in IPCs of the fly brain. In Dilp2.AdipoR-Ri, mRNA expression levels of Dilp2, Dilp3 and Dilp5 were similar to those of the Dilp2-Gal4 control in the larval stage, but Dilp3 expression was slightly but significantly decreased in the adult heads relative to that of the Dilp2-Gal4 controls ( Figure 3A, B). Because the reduction of Dilp3 expression in the adult stage does not explain the larval phenotype observed in Dilp2.AdipoR-Ri flies, we examined insulin secretion by the Dilp2 immunostaining in IPCs [25]. After 24 h starvation, IPCs of the Dilp2-Gal4 larval brain were strongly stained with the Dilp2 antibody, reflecting a high accumulation of Dilp2. When the larvae were refed for 2 h, accumulated Dilp2 was secreted from IPCs and the remaining Dilp2 in IPCs was reduced to half ( Figure 4A, B). However, IPCs with dAdipoR inhibition in the refed condition still had a high level of Dilp2 similar to the starved condition ( Figure 4A, B). These data indicate that dAdipoR has a role in the secretion of Dilp2. To confirm the secretion response of IPCs, we used the secretable GFP (secGFP) as a reporter of secretion [25]. Similar to the Dilp2 staining intensity, the secGFP fluorescence intensity of Dilp2-Gal4 IPCs diminished by 80% in the refed condition compared to that of the starved condition. However, the secGFP fluorescence intensity of Dilp2.dAdipoR-Ri IPCs was not reduced in the refed condition ( Figure 4C, D). Because the blocking of secretion and/or the enhanced translation of Dilp2 transcripts in IPCs may have caused the increased staining of Dilp2, we assessed the Dilp2 secretion by measuring the circulating Dilp2 level in the larval hemolymph. We overexpressed FLAG-tagged Dilp2 in IPCs of the control (Dilp2.Dilp2 FLAG ) and the dAdipoR knockdown flies (Dilp2.Dilp2 FLAG , dAdipoR-Ri). Then, we measured the FLAGtagged Dilp2 protein level in the larval hemolymph with the anti-FLAG ELISA assay [27]. When the 3 rd instar larvae were starved for 4 h, the circulating Dilp2 FLAG level of the control was similar to the Dilp2 FLAG level of dAdipoR knockdown larvae ( Figure 4E). The hemolymph Dilp2 FLAG level increased by 1.5-fold in Dilp2.Dilp2 FLAG control larvae after refeeding compared to the level in the starved condition, but the hemolymph Dilp2 FLAG level did not change in the Dilp2.Dilp2 FLAG , dAdipoR-Ri larvae after refeeding ( Figure 4E). In addition, we observed that Dilp2.Dilp2-FLAG , dAdipoR-Ri larvae had a lower level of hemolymph Dilp2 FLAG than that of Dilp2.Dilp2 FLAG larvae by the Western blot analysis ( Figure S4A). This result indicates that dAdipoR regulates Dilp2 secretion in larvae. To test whether dAdipoR also regulates Dilp2 secretion in adult flies, we measured secGFP in the thorax and abdomen from the Dilp2.secGFP control and Dilp2.secGFP, dAdipoR-RNAi flies using the Western blot analysis. The GFP protein level in the body of the refed Dilp2.secGFP control flies increased by 1.4-fold compared to that of the starved flies, whereas the GFP protein level in the body of refed AdipoR inhibition flies was similar to that of the starved condition ( Figure 4F). These results indicate that dAdipoR regulates insulin secretion in IPCs of larvae and adults.

Insulin Signaling is Reduced in the dAdipoR Knockdown Flies
Since dAdipoR positively regulated insulin secretion, the Dilp2.dAdipoR-Ri flies would have reduced insulin signaling in insulin target tissues. To measure the activity of insulin signaling in Dilp2.dAdipoR-Ri flies, we examined the subcellular localization of dFOXO and the activity of dFOXO in peripheral tissues. In insects, the fat body, a homologous organ of mammalian liver and adipocytes, is the major insulin target tissue. In the starved condition, dFOXO was mainly localized in nuclei of the Dilp2-Gal4 control and Dilp2.dAdipoR-Ri fat bodies. After refeeding, dFOXO was relocated to the cytoplasm in the Dilp2-Gal4 control fat body while most dFOXO proteins were still located in the nuclei of the Dilp2.dAdipoR-Ri fat body ( Figure 5A). Then, we measured the expression level of the dFOXO target gene 4E-BP in adult fly bodies. In the starved condition, the expression level of 4E-BP in Dilp2.dAdipoR-Ri flies were similar with those of the controls. In the refed condition, the expression level of 4E-BP in controls decreased to 20% of the starved condition level, but the expression level of 4E-BP in Dilp2.dAdipoR-Ri flies decreased to 60% of the starved condition level ( Figure 5B). These data demonstrate that insulin signaling is reduced in the insulin target tissue of Dilp2.dAdipoR-Ri flies.

Human Adiponectin Activated Dilp2 Secretion from the IPCs of Larval Brains
Due to the structural and functional similarities of dAdipoR with mammalian adiponectin receptors, we speculated that an adiponectin-like protein from the fat body may activate dAdipoR and induce insulin secretion. The yeast adiponectin receptor ligand osmotin can bind and activate human adiponectin receptors, suggesting that an adiponectin from one species can bind and activate adiponectin receptors in another species [31,32].  Moreover, the amino acid residues of dAdipoR predicted to interact with an adiponectin are homologous to those of human AdipoR1 ( Figure 1A) [31]. Therefore, we tested whether human globular adiponectin, which has a high binding affinity to human AdipoR1, binds to dAdipoR and induces Dilp2 secretion in larval IPCs. Three different concentrations of human adiponectin were treated to dissected Dilp2-Gal4 larval brains ( Figure 6A). 10 and 20 mg/ml of adiponectin significantly decreased Dilp2 staining intensity (23% and 16%, respectively) when compared to the untreated control, implying that human adiponectin can stimulate Dilp2 secretion ( Figure 6A). Then, we examined whether inhibition of dAdipoR in IPCs could suppress Dilp2 secretion by the human adiponectin treatment. In Dilp2.dAdipoR-Ri brains, human adiponectin treatments did not change Dilp2 staining intensities in IPCs compared with non-treated controls ( Figure 6B-F). This result suggests that human adiponectin binds to dAdipoR and controls Dilp2 secretion.

Discussion
Adiponectin receptors cloned from yeast to humans play important roles in energy homeostasis across species [3,[32][33][34]. For the regulation of energy metabolism, mammalian AdipoRs are expressed in key metabolic organs such as hypothalamus, insulin target tissues and beta cells. Despite beta cell expression of adiponectin receptors [7,8], their function in the beta cell is ambiguous. In vitro studies with islet cell lines and explanted islets report controversial roles of AdipoRs in the beta cells due to differences in experimental conditions [35]. Thus, beta cell specific disruption of AdipoRs is necessary to clarify AdipoR function. In this study, we found that Drosophila adiponectin receptor was expressed in IPCs and investigated its function by IPC-specific dAdipoR inhibition in flies. Inhibition of dAdipoR in IPCs did not affect development and viability of IPCs (data not shown) but impaired insulin secretion. Together with the secretion defect, a decrease in Dilp3 transcript levels of adult flies was observed ( Figure 3B). The reduced expression of Dilp3 appears to have some correlation with Dilp secretion. When insulin secretion is inhibited by the overexpression of mammalian UCP genes in IPCs, Dilp3 expression is decreased but expression levels of Dilp2 and Dilp5 are not changed [36]. The relationship between Dilp3 expression and Dilp secretion needs further analysis. Our in vivo study for dAdipoR function suggests that mammalian adiponectin receptors in the beta cell may have similar roles in insulin secretion and production.
Unlike mammalian insulin, Drosophila insulin-like peptides regulate larval growth and energy metabolism [11,15,37]. In this study, however, the interference of Dilp secretion by dAdipoR reduction did not change the body size, although it clearly affected energy homeostasis. Partial reduction of the Dilp2 mRNA level changes glucose metabolism and starvation resistance but not growth retardation partly due to the compensatory mechanism among Dilp genes [26]. Therefore, partial inhibition of Dilp secretion by dAdipoR reduction in IPCs may not enough for blocking growth or the compensatory mechanism.
Human globular adiponectin was able to induce Dilp secretion through dAdipoR in the ex vivo culture suggesting dAdipoR is a functional homologue of human adiponectin receptors and globular adiponectin like molecule would be the  ligand of dAdipoR. We could not find Drosophila adiponectin using the amino acid sequence homology search, possibly because the Drosophila adiponectin-like molecule may share only structural similarities to adiponectins of other organisms. Human adiponectin and tobacco osmotin do not share the amino acid sequence homology, but they have an overlapped beta barrel structure and tobacco osmotin that can activate human AdipoR1 [31,32]. Therefore, searching beta barrel structure proteins homologous to human globular adiponectin is one possible approach to uncover the identity of the Drosophila adiponectin-like molecule.
Drosophila adipokine signaling is not well understood yet. The fat body, Drosophila adipose tissue, is a source for the production of adipokines. Depending on nutrient availability, the fat body secretes humoral signals to remotely regulate IPC function [25,[38][39][40]. The amino acid signaling in the fat body releases humoral signals [25]. Recently, cytokine Upd2 was identified as a fat body humoral factor for Dilp secretion in the brain [24]. The expression of upd2 in the fat body is regulated by sugar and lipid not amino acid [24]. These previous studies indicate that the fat body secretes multiple factors to modulate IPC function. Our findings suggest that the unidentified Drosophila adiponectin could be one of the fat body signals to control insulin secretion in IPCs through dAdipoR. These findings can provide an insight for the function of mammalian adiponectin receptor in pancreatic beta cells, which could be useful for therapeutic application.
those in the Dilp2-Gal4 control. (E) Anti-FLAG ELISA assay showed that the level of circulating Dilp FLAG proteins was increased after refeeding in Dilp2.Dilp FLAG larvae but not in Dilp2.Dilp FLAG , dAdipoR-Ri larvae. (F) Western blot analysis showed that the secGFP level was increased after refeeding when compared to the starved condition in adult bodies of the Dilp2.secGFP control but not in Dilp2. secGFP, dAdipoR-Ri adult bodies. Data are presented as means 6SEM; *p,0.05, **p,0.01. Scale bars are 20 mm (A, C). doi:10.1371/journal.pone.0068641.g004