Central metabolic sensing remotely controls nutrient-sensitive endocrine response in Drosophila via Sir2/Sirt1–upd2–IIS axis

Metabolic homeostasis is indispensable for all organisms and involves the coupling of metabolic sensing with adaptive responses. In multicellular organisms, metabolic homeostasis depends on efficient communication across diverse organ systems, predominantly by endocrine mechanisms (Frühbeck et al., 2001; Pedersen, 2011; Stefan and Häring, 2013; Unger et al., 1978). Pancreatic hormones such as insulin and glucagon couple organismal nutrient status with nutrient uptake across organ systems (Unger et al., 1978). A large body of work has facilitated our understanding of the mechanisms within insulinogenic cells (specifically, pancreatic β-islets in vertebrates) that mediate the integration of organismal nutrient status with insulin secretion (Rorsman and Braun, 2013). However, emerging findings from evolutionarily diverse organisms highlight the importance of such an endocrine control from distant tissues (Song et al., 2014; Géminard et al., 2009). In this context, relatively less is known about molecular factors that mediate distant regulation of insulin secretion across species.

The evolutionarily conserved NAD⁺ sensor Sir2/Sirt1 plays critical roles in controlling insulin secretion from the β-islets in the pancreas (Ramachandran et al., 2011; Bordone et al., 2006). Previous reports from our lab and others have indicated a possible role for Sir2/Sirt1 in distant tissues in regulating insulin production and secretion (Palu and Thummel, 2016; Schenk et al., 2011; Wang et al., 2011; Purushotham et al., 2009; Banerjee et al., 2013). However, the mechanistic details and the physiological understanding of this endocrine control are currently lacking.

Drosophila melanogaster has been extensively used to investigate the physiological and genetic bases of metabolic homeostasis. Under conditions of nutrient excess, the insulin-producing cells (IPCs) located in the median neurosecretory cluster (mNSC) in the brain secrete Drosophila insulin-like peptides (dILPs) (Ikeya et al., 2002; Kim and Rulifson, 2004). Interestingly, IPCs lack the ability to sense organismal metabolic status and depend on signals from the fat body (Ikeya et al., 2002; Kim and Rulifson, 2004). Genetic evidence implicates the metabolic transcription factor FOXO (dFOXO) (Hwangbo et al., 2004), target of rapamycin (TOR) signaling (Géminard et al., 2009) and a secretory cytokine upd2 (Rajan and Perrimon, 2012), among others, in mediating the distant control of dILP secretion by the fat body. However, the physiological relevance of these factors and a role for a master metabolic sensor within the fat body in establishing systemic metabolic homeostasis by controlling an inter-organ communication network has not been addressed.

Here, we have investigated the molecular underpinnings of the distant control of a metabolic response by addressing the role of the master metabolic sensor Sir2/Sirt1 within the fat body in regulating an endocrine response to nutrient fluctuations.

S₁106 (P{Switch 1}106 Gal4), Sir2/Sirt1^EP2300 (w¹¹¹⁸; P{w[+mC]=EP}Sirt1[EP2300] DnaJ-H[EP2300]/CyO), chico (cn¹ P{ry11}chico¹/CyO; ry⁵⁰⁶) and InR (InR^E19/TM2) stocks were obtained from Bloomington Stock Center (Indiana University). SIR2^RNAi (23201) and upd2^RNAi (6513) lines were obtained from Vienna Drosophila RNAi Center (VDRC).

Flies were grown on standard CM/S/Y media (8.6% cornmeal, 5% sucrose, 2.5% yeast) under non-crowding conditions at 25°C with a 12 h:12 h light:dark cycle. Age-matched 3–5 day old female flies were used for all analyses.

Gene-switch Gal4 S₁106 was crossed with relevant transgene lines. Gal4 was activated to drive UAS-RNAi or EP-Sir2/Sirt1 lines, by rearing the appropriate lines on diets supplemented with 200 μmol l⁻¹ (or 400 μmol l⁻¹, in the case of RNAi combinations) RU486 (Mifeprestone, Sigma-Aldrich; hereafter, RU) dissolved in 95% ethanol. Flies that were reared on diets containing only ethanol served as controls.

Flies grown on standard diet were fasted for 16 h on 2% agar. Glucose stimulation was performed by shifting these flies to vials containing 10% glucose media for 1 h. Immediately after glucose stimulation, the flies were collected for hemolymph extraction.

Hemolymph was isolated from the head capsule of 50 flies by capillary action as described earlier (Banerjee et al., 2013). For western blot analysis, protein lysates were prepared by incubating the hemolymph extracts in lysis buffer. The protein content of the hemolymph lysates was estimated by loading 25 μl of the extract on an SDS-PAGE gel and normalized to 25 μg of bovine serum albumin (BSA) by Coomassie Brilliant Blue (R250) staining. Following this, 1 mg of total hemolymph protein extract was loaded on a separate gel and proteins below 25 kDa were probed with anti-dILP5 antibody (ab128220, Abcam) to quantify circulating dILP5. The upper part of the same gel was stained with Coomassie Brilliant Blue and used as a loading control.

Adult flies were anesthetized using CO₂ and the intestines/fat body were dissected in ice-cold Tris-buffered saline (TBS; pH 7.4). Fat body/intestines were pooled for RNA isolation (in 1 ml Trizol) or protein lysates (in 200 μl RIPA), respectively, to generate individual replicates.

RNA was isolated using Trizol (15596018, Life Technologies) as per the manufacturer's instructions. For cDNA synthesis, 2 μg of RNA was reverse transcribed using SuperScript-III reverse transcriptase kit (18080-044, Life technologies).

Real-time qPCR (qRT-PCR) reactions to quantify relative mRNA expression were performed using Kapa SYBR green (KK4600, KAPA Biosystems) on Eppendorf Realplex. Data were analyzed by the ΔΔCT method. Triplicate samples (30 flies each) from at least two separate experiments were used. rp49 and actin5c were used as normalization controls. The primer pairs used for the qRT-PCR are as follows; upd2: forward CGGAACATCACGATGAGCGAAT and reverse TCGGCAGGAACTTGTACTCG, actin5c: forward GAGGCTTGCGGCATCCACGAGACCAC and reverse GACAGAGTACTTGCGCTCTGGCG and rp49: forward TACAGGCCCAAGATCGTGAA and reverse TCTCCTTGCGCTTCTTGGA. The cycling conditions were as prescribed by the manufacturers.

Protein lysates were prepared by incubating tissues on ice for 15 min in RIPA buffer (50 mmol l⁻¹ Tris-HCl pH 7.5, 150 mmol l⁻¹ NaCl, 1 mmol l⁻¹ EDTA, 6 mmol l⁻¹ EGTA, 20 mmol l⁻¹ NaF, 1% Triton X-100) with protease inhibitors (05-056-489001, Roche) and phosphatase inhibitor (PhosSTOP, 04-906-837-001, Roche). Anti-pAKT (1:750, 9271, Cell Signaling Technology), anti-AKT (1:750, 9272, Cell Signaling Technology) and anti-actin (1:3000, A1978, Sigma Aldrich) antibodies were used for the western blots. Chemiluminescence detection (1859023/185022, Thermo Scientific) was used to visualize the bands.

SigmaPlot12.0 was used for statistical analyses. Student's t-test and ANOVA were used to determine statistical significance of the data.

Clinical evidence implicates impaired metabolic sensing by central metabolic organs, such as the liver and adipocytes, in metabolic syndromes such as type 2 diabetes (T2DM) and insulin resistance (Song et al., 2014; Roden, 2006; Guilherme et al., 2008). In this regard, hepatic and adipocyte-specific disruption of Sirt1 results in hyperinsulinemia in mammals (Purushotham et al., 2009; Wang et al., 2011; Gillum et al., 2011), but the mechanisms are poorly understood. Although a role for Sir2/Sirt1 in controlling dILP2 secretion was identified (Palu and Thummel, 2016), the tissue of origin of Sir2/Sirt1-dependent dILP control is still unknown.

To address the link between central metabolic sensing and an endocrine metabolic response, we investigated the role of fat body Sir2/Sirt1 in controlling the secretion of dILP from the IPCs. Knockdown of Sir2/Sirt1 in the fat body (S₁106;Sir2^RNAi with and without RU) (Osterwalder et al., 2001) significantly increased hemolymph dILP5 levels (Fig. 1A), while over-expression (S₁106/Sir2^OE with and without RU) reduced circulating dILP5 to undetectable levels (Fig. 1B). Acute glucose administration to control flies led to an increase in dILP5 secretion in the hemolymph (Fig. 1C). Furthermore, fat body-specific Sir2/Sirt1 knockdown resulted in heightened secretion of dILP5 and prevented a further increase in dILP5 secretion upon administration of glucose (Fig. 1C).

These results clearly demonstrate a role for Sir2/Sirt1 in the Drosophila fat body in the secretion of dILP5 from IPCs. More importantly, these findings provided a premise to interrogate the mechanistic underpinnings of the role of Sir2/Sirt1 in mediating inter-tissue communication between central metabolic tissue and insulinogenic tissue.

Fat body, liver and adipocytes have well-established and evolutionarily conserved endocrine functions (Frühbeck et al., 2001; Ouchi et al., 2011; Stefan and Häring, 2013). However, insights into molecular mechanisms within these tissues that couple metabolic inputs to the regulation of cytokines (adipokines/hepatokines) are limited.

The JAK/STAT ligand upd2 is a humoral link between the fat body and IPCs, which controls dILP secretion, specifically in response to glucose feeding (Rajan and Perrimon, 2012). To gain insights into the mechanisms by which fat body Sir2/Sirt1 might remotely regulate dILP5 secretion from the IPCs, we investigated its genetic interaction with upd2. As reported earlier (Rajan and Perrimon, 2012), knockdown of upd2 in the fat body (S₁106;upd2^RNAi with and without RU) reduced circulating dILP5 levels (Fig. 2A). Furthermore, simultaneous knockdown of upd2 and Sir2/Sirt1 within the fat body (S₁106;Sir2^RNAi/upd2^RNAi with and without RU; Fig. 2B) resulted in undetectable circulatory dILP5, mimicking S₁106;upd2^RNAi flies. These results pointed towards an epistatic association between Sir2/Sir1 and upd2 in regulating dILP5. Physiologically, the ability to regulate dILP5 secretion suggested a role for upd2 in controlling a response to nutrient inputs. In this regard, we knocked down upd2 in the fat body and assayed for glucose sensitivity. We found that knockdown of upd2 in the fat body blunted the glucose-stimulated increase in hemolymph dILP5 levels (Fig. 2C). Importantly, the combined knockdown of upd2 and Sir2/Sirt1 in the fat body subdued the glucose-dependent increase in dILP5 secretion (Fig. 2D). The effects of combinatorial knockdown of upd2 and Sir2/Sirt1 in the fat body on dILP5 secretion under basal as well as glucose-stimulated conditions indicated an epistatic interaction between these two factors in the regulation of glucose sensitivity (Fig. 2). Motivated by our results, we investigated the nature of the interaction between Sir2/Sirt1 and upd2.

As shown in Fig. 2, overexpression of Sir2/Sirt1 significantly reduced upd2 expression (Fig. 2E) and knockdown of Sir2/Sirt1 in the fat body led to a robust increase in the expression of upd2 (Fig. 2F). These results indicated that in addition to exhibiting an epistatic association in physiologically controlling dILP5 secretion in response to glucose inputs, Sir2/Sirt1 acts as an upstream regulator of upd2 expression in the fat body (Fig. 2E,F). The regulation of upd2 by Sir2/Sirt1 in the fat body provides comprehensive evidence for the control of a JAK/STAT ligand by Sir2/Sirt1 in metabolic signaling. While Sirt1 has previously been shown to regulate Stat3 in the control of hepatic gluconeogenesis (Nie et al., 2009), our results show for the first time the importance of Sir2/Sirt1-dependent control of JAK/STAT signaling in insulin secretion.

Clinical reports correlate dysfunctional central metabolic sensing by liver and brain with metabolic maladaptation in the intestine (Badman and Flier, 2005). However, whether metabolic sensing within central metabolic organs can influence metabolic signaling pathways, such as insulin/IGF signaling (IIS) in the intestine remains elusive.

Inspired by our results, we sought to investigate the role of the interplay between Sir2/Sirt1 and upd2 in controlling intestinal IIS. Knockdown of Sir2/Sirt1 in the fat body led to an increase in intestinal insulin signaling (Fig. 3A), while overexpression of Sir2/Sirt1 in the fat body resulted in a strong reduction in insulin signaling within the intestine (Fig. 3B). These results provide the first evidence to implicate the role of a metabolic sensor in liver/adipocytes in distantly controlling intestinal IIS.

In vivo and in vitro studies have elucidated that systemic insulin signaling is controlled by ligand (insulin)-dependent and -independent mechanisms (Pessin and Saltiel, 2000). In keeping with this, we wanted to ascertain whether the distant control of intestinal insulin signaling by fat body Sir2/Sirt1 involved a ligand (dILP5, in this case)-independent or -dependent mechanism. In this regard, we reasoned that if fat body Sir2/Sirt1-dependent control of intestinal insulin signaling is ligand (dILP5) independent, then a reduction in insulin signaling via genetic disruption would obviate this coupling. Conversely, retention of the sensitivity of intestinal insulin signaling to changes in fat body Sir2/Sirt1 levels despite the genetic ablation of insulin signaling would suggest a ligand-dependent mechanism. We employed two models of reduced insulin signaling: (1) heterozygosity of InR (Tatar et al., 2001) and (2) heterozygosity of chico (Clancy et al., 2001). Heterozygosity of InR (S₁106/+;Sir2^RNAi/InR without RU) and chico (S₁106/chico;Sir2^RNAi/+ without RU) attenuated intestinal insulin signaling (Fig. 3C,D, lanes 7–9) basally. Interestingly, simultaneous knockdown of Sir2/Sirt1 in the fat body of the InR (S₁106/+;Sir2^RNAi/InR with RU) and chico (S₁106/chico;Sir2^RNAi/+ with RU) heterozygotes resulted in an increase in pAKT levels (Fig. 3C,D, lanes 10–12) when compared with those exhibited by only the heterozygote background (Fig. 3C,D, lanes 7–9).

Given the complex interplay between dILPs (Grönke et al., 2010; Okamoto and Nishimura, 2015), we knocked down the upstream regulator upd2 simultaneously with Sir2/Sirt1 in the fat body to gain mechanistic insights into intestinal control of IIS. Knockdown of upd2 alone in the fat body (S₁106;upd2^RNAi with and without RU) or with a simultaneous knockdown of Sir2/Sirt1 (S₁106;Sir2^RNAi/upd2^RNAi with and without RU) was sufficient to bring about a strong reduction in intestinal insulin signaling (Fig. 3G, lanes 7–12) and override the increase observed in response to fat body Sir2/Sirt1 knockdown (Fig. 3G, lanes 1–6).

In summary, we have provided mechanistic and physiological insights into the regulation of dILP5 secretion by the fat body involving the NAD⁺-dependent metabolic sensor Sir2/Sirt1 and the JAK/STAT ligand upd2. A recent study reported that Sir2/Sirt1 is a negative regulator of dILP2 secretion (Palu and Thummel, 2016). However, whether this is mediated in a tissue-autonomous or -non-autonomous manner was unclear. A genetic or molecular interaction of Sir2/Sirt1 with previously established cytokine communicators that may bring about such a control has also not been demonstrated. In this context, we emphasize that this is the first report of an inter-tissue communication between the fat body, IPCs and the intestine, which is controlled by the NAD⁺-dependent Sirt1/Sir2. Specifically, our findings provide conclusive mechanistic underpinnings of a fat-body Sir2/Sirt1-mediated distant regulation of dILP5 via upd2. These results support the role of a central metabolic organ in controlling intestinal metabolic signaling and highlight the importance of a master metabolic sensor in this distant control of organismal physiology (Fig. 3I). Given the recent advances towards understanding a distant control of insulin secretion in vertebrates and invertebrates, our study identifies an axis between nutrient sensing by the fat body and humoral inputs from the neuroendocrine system.

We thank Ms Sweta Parik and Ms Namrata Shukla for help in preparing the manuscript and for their critical inputs.