Regulation of insulin-producing cells in the adult Drosophila brain via the tachykinin peptide receptor DTKR

The insulin signaling pathway plays a prominent role in the regulation of metabolism, growth, stress resistance and lifespan both in mammals and in Drosophila (Baker and Thummel, 2007; Birse et al., 2010; Broughton et al., 2005; Clancy et al., 2001; Goberdhan and Wilson, 2003; Rulifson et al., 2002; Taguchi and White, 2008; Tatar et al., 2003; Zhang et al., 2009) and is also largely conserved over evolution (Brogiolo et al., 2001; Garofalo, 2002; Giannakou and Partridge, 2007; Grönke et al., 2010). Together, Drosophila insulin-like peptides (DILPs) and adipokinetic hormone (AKH) seem to mirror the functions of mammalian insulin and glucagon in Drosophila with respect to metabolic homeostasis. DILPs stimulate the uptake of carbohydrate and thus diminishes the circulating blood sugar levels whereas AKH augments the circulating trehalose levels by stimulating glycolysis in the fat body; both hormones also affect lipid homeostasis (Broughton et al., 2005; Géminard et al., 2006; Grönke et al., 2007; Isabel et al., 2005; Kim and Rulifson, 2004; Lee and Park, 2004; Rulifson et al., 2002). Ablation of the insulin-producing cells (IPCs) in the brain results in retarded growth, increased glucose levels in the circulation, increased storage of lipid, reduced fecundity and increased resistance to stress (Broughton et al., 2005; Rulifson et al., 2002). A recent paper demonstrated that DILP signaling to olfactory sensory neurons in the Drosophila antennae decreases the sensitivity to odors and thus food search in fed flies (Root et al., 2011). Additionally there is evidence that DILPs, or at least signaling from the IPCs, may participate in the regulation of sleep-wakefulness (Crocker et al., 2010) and ethanol sensitivity (Corl et al., 2005).

In Drosophila seven different DILPs have been identified (DILP1–7), some of which resemble mammalian insulin, one that is relaxin-like and one that is similar to insulin-like growth factor (Brogiolo et al., 2001; Grönke et al., 2010; Ikeya et al., 2002; Miguel-Aliaga et al., 2008; Okamoto et al., 2009; Slaidina et al., 2009; Yang et al., 2008). However, only one DILP receptor (dInR) is known in Drosophila so far. Extensive work has addressed the mechanisms and roles of signaling downstream of the dInR in Drosophila, but the hormonal or neuronal mechanisms that control insulin production and release are not yet understood. The regulation of DILP release in response to nutritional state of the Drosophila larva is based on activation of IPCs by nutrient-mediated signals released into the circulation from the fat body where nutrient sensors are located (Geminard et al., 2009). It cannot be excluded that the IPCs also have autonomous nutrient sensors, like the AKH-producing cells (Kim and Rulifson, 2004; Kreneisz et al., 2010), but it is far from clear how DILP production and release is regulated in the fly.

The IPCs of the brain are located in the median neurosecretory cell group of the protocerebrum. These IPCs produce three of the insulin-like peptides, DILP2, DILP3 and DILP5, and are presumed to release these into the circulation from axon terminations in the neurohemal organ corpora cardiaca as well as in the aorta and anterior midgut (Brogiolo et al., 2001; Cao and Brown, 2001; Ikeya et al., 2002; Rulifson et al., 2002). Importantly, the three DILPs are colocalized in each of the IPCs (Geminard et al., 2009; Ikeya et al., 2002). Some recent studies have suggested neuronal regulation of the brain IPCs in Drosophila, and a few neurotransmitters have been suggested as regulators in these neuronal systems: short neuropeptide F (sNPF) (Lee et al., 2008; Lee et al., 2004), octopamine (Crocker et al., 2010), γ-amino butyric acid (GABA) (Enell et al., 2010), and serotonin (Kaplan et al., 2008; Luo et al., 2011). Of these, sNPF and octopamine seem to stimulate insulin signaling, GABA is inhibitory, and serotonin is not yet clear. Thus, the IPCs appear to be under tight neuronal control by multiple neurotransmitters, probably released from distinct neuronal systems. Here we present evidence for yet another type of neuronal regulator of the IPCs, a set of neuropeptides, designated Drosophila tachykinins (DTKs), derived from a single precursor and related to mammalian tachykinins of the preprotachykinin A encoding gene (Siviter et al., 2000).

The six Drosophila tachykinins, DTK1–6, originate from the gene Dtk (Siviter et al., 2000) and can activate two G protein-coupled receptors designated DTKR and NKD (Birse et al., 2006; Poels et al., 2009). The DTKs appear to have pleiotropic functions in the Drosophila nervous system, including neuromodulation in olfactory circuits of the antennal lobes and in locomotor circuits of the central complex (Ignell et al., 2009; Kahsai et al., 2010; Nässel and Winther, 2010; Winther et al., 2006). We show here that signaling through DTKR seems to have an inhibitory effect on brain IPCs whereas the other receptor NKD appears not to be expressed in IPCs and thus plays no role in this regulation. Interestingly, mammalian tachykinins, such as substance P, have been shown to increase insulin secretion from the pancreas, indicating a role in regulation of metabolism (Adeghate et al., 2001; Crocker et al., 2010; Schmidt et al., 2000).

All flies [Drosophila melanogaster (Meigen)] were reared at 25°C on a standard yeast, corn meal and agar medium, under 12 h:12 h light:dark conditions. The following fly lines were used: w¹¹¹⁸ and UAS-cd8GFP were from the Bloomington Drosophila Stock Center (BDSC; Indiana University, Bloomington, IN, USA). A Dilp2-Gal4 line was donated by Ping Shen [University of Georgia, Athens, GA, USA (Wu et al., 2005)]. The UAS-Dtkr-RNAi, UAS-Dtkr, UAS-Dtkr-gfp and UAS-Nkd constructs and fly lines for manipulation of DTKR and NKD receptor levels were described previously (Ignell et al., 2009; Söderberg et al., 2011). Finally, we also used a UAS-Nkd-RNAi line from the National Institute of Genetics RNAi Center (NIG-fly) in Kyoto, Japan.

Rabbit antisera to the amino acids 488–506 of the C-terminus of the Drosophila tachykinin receptor DTKR (Birse et al., 2006) and to the full-length cockroach tachykinin LemTRP1, to detect peptides of the DTK precursor (Winther and Nässel, 2001), were used at dilutions of 1:1000–1:2000. These antisera have previously been characterized on Drosophila tissues (Birse et al., 2006; Söderberg et al., 2011; Winther et al., 2003). A rabbit antiserum to DILP2 [a gift from M. R. Brown, Athens, GA, USA (Cao and Brown, 2001)] was used at a dilution of 1:4000. To enhance the green fluorescent protein (GFP) signal, we used either a mouse monoclonal antibody to GFP (mAb 3E6; code A-11120; Molecular Probes, Leiden, The Netherlands) or a rabbit anti-GFP (code#A-6455; Invitrogen, Carlsbad, CA, USA).

For immunocytochemistry adult Drosophila heads or central nervous systems (CNS) of third instar larvae were dissected in 0.01 mol l^–1 phosphate buffered saline with 0.25% Triton X-100, pH 7.2 (PBS-TX) and fixed in ice-cold 4% paraformaldehyde in 0.1 mol l^–1 sodium phosphate buffer pH 7.4 (PB) for 2–4 h or in Bouin's fixative for 4 h (for the DTKR antiserum applied to adult brains). After rinsing with 0.1 mol l^–1 PB adult brains or larval CNS were either dissected out for whole-mount immunocytochemistry or whole heads were incubated overnight in 20% sucrose in 0.1 mol l^–1 PB at 4°C as cryoprotection. Cryostat sections (50 μm thick) of the heads were cut on a cryostat at –23°C. Bouin-fixed tissues were embedded in paraffin and sectioned on a microtome at 12 μm thickness. Incubation with primary antisera for whole-mount tissues was performed for 72 h whereas sections were incubated for 48 h, both at 4°C. For detection of primary antisera Cy3-tagged goat anti-rabbit antiserum (Jackson ImmunoResearch, West Grove, PA, USA) was used at a dilution of 1:1000. Tissues or sections were rinsed thoroughly with PBS-TX, followed by a final wash in PBS and then mounted in 80% glycerol in PBS.

Microscopic analysis was performed on a Zeiss LSM 510 laser scanning confocal microscope (Zeiss, Jena, Germany) and edited in LSM software and Adobe Photoshop CS3 Extended version 10.0 (Adobe Systems, Mountain View, CA, USA).

Immunocytochemistry with DILP2 antiserum was performed on brain IPCs of different genotypes for quantification of immunofluorescence. Confocal images of the brains from different genotypes were obtained with identical laser intensity and scan settings and monitored for immunofluorescence intensity as described by Broughton et al. and Kaplan et al. (Broughton et al., 2010; Kaplan et al., 2008). Immunofluorescence of both IPC cell bodies and background fluorescence in unlabeled adjacent brain neuropil was quantified in a set of regions of interest, using Image J 1.40 from NIH, Bethesda, MD, USA (http://rsb.info.nih.gov/ij/). The fluorescence of each cell body was measured four times in different regions, and the mean value was obtained. The final immunofluorescence intensity of cell bodies was calculated by subtracting the intensity of tissue background. For each genotype, 40–45 cell bodies in total from eight brains were measured. The data were analyzed in Prism GraphPad 6.0 (San Diego, CA, USA) with Student's t-tests.

Male flies, 4–8 days old, were anesthetized with carbon dioxide and placed individually in 2 ml glass vials with 500 μl of 0.5% aqueous agarose at 25°C in an incubator with 12 h:12 h light:dark conditions and controlled humidity. The vials were checked for dead flies every 12 h. These starvation experiments were run in three replicates with at least 40 flies of each genotype per replicate.

Whole-body trehalose was measured according to Isabel et al. (Isabel et al., 2005) with a few minor alterations. Male flies (4–8 days old) were put in groups of 5 in Eppendorf tubes. Flies were weighed (wet mass), then incubated for 1 h in 500 μl of 70% ethanol. Each tube of flies was sonicated (Sonics and Materials Inc., Danbury, CT, USA) for 20 s. 1 ml of 70% ethanol was added, and the tubes were incubated at room temperature for 1 h. A trehalose standard was made with a 2-fold dilution series starting at 200 μg ml^–1. The samples were centrifuged for 5 min at 16,000 g, and 1 ml of the samples and 500 μl of the standards were placed in 2 ml Eppendorf tubes and then dried in a vacuum centrifuge (Savant Speed Vac; Speed Vac Plus Sc110A, Farmingdale, NY, USA). 200 μl of 2% NaOH and 1.5 ml of fresh Anthrone reagent (Sigma Catalogue #A 1631, Sigma Chemical Co., St Louis, MO, USA) were added and the tubes were briefly vortexed. Samples were incubated in a water bath at 90°C for 10 min. 100 μl of each sample was placed in a 96-well ELISA plate and measured in triplicate on an ELISA plate reader (Labsystems Multiscan Plus, Stockholm, Sweden).

Lipid content was determined according to the method of Service et al. (Service et al., 1987). Groups of 5 male flies were weighed (wet mass; Mettler MTS, Stockholm, Sweden) and subsequently dried at 65°C for 24 h. Flies were then weighed again to obtain dry mass. Lipids were extracted by placing intact dry flies in glass vials containing 9 ml of diethyl ether for 24 h with gentle agitation at room temperature. The diethyl ether was removed and flies were dried for a further 24 h. Flies were weighed to obtain lean dry mass. The difference between dry mass and lean dry mass was considered the total lipid content of the flies. Lipid content was measured at 0 h and 24 h of starvation.

Total RNA was extracted from whole heads by using TRIzol (GIBCO, Carlsbad, CA, USA), according to the manufacturer's protocol. The RNA was treated with DNase to remove any remaining residual genomic DNA (Turbo DNA-freeTM, Ambion). Treated mRNA was reverse transcribed to cDNA using Quantitect Reverse Transcription Kit (Qiagen, Düsseldorf, Germany), according to the manufacturer's instructions.

The dilp primers were as follows: dilp2F (forward), TCTGCAGTGAAAAGCTCAACGA; dilp2R (reverse), TCGGCACCGGGCATG; dilp3F, AGAGAACTTTGGACCCCGTGAA; dilp3R, TGAACCGAACTATCACTCAACAGTCT; dilp5F, GAGGCACCTTGGGCCTATTC; and dilp5R, CATGTGGTGAGATTCGGAGCTA.

PCR was carried out using Taqman Universal PCR Master Mix, according to the manufacturer's instructions (Applied Biosystems, Foster City, CA, USA), with the exception that 25 ml reaction volumes were used, on an ABI Prism 7000 (Applied Biosystems). Endogenous control (rp49) primers were as follows: rp49F, CACACCAAATCTTACAAAATGTGTGA; and rp49R, AATCCGGCCTTGCACATG. The samples were analyzed in triplicate, and the measured concentration of mRNA was normalized relative to the rp49 control values. Experiments were made in three replicates, starting from new RNA collection. The normalized data were used to quantify the relative levels of a given mRNA according to the DCt analysis.

We also measured levels of Dtkr and Nkd transcripts in wild-type flies that were normally fed or starved for 24 h. All flies were 4–8 days old. Approximately 50 flies per tube were placed on 0.5% agarose for 24 h, after which 30 flies per tube were frozen at –80°C. These experiments were made with a different PCR protocol described here. All samples were homogenized in liquid nitrogen and total RNA was extracted using the recommended protocol from the Qiagen RNeasy mini kit (Qiagen catalog no. 74124). RNA was quantified by spectrophotometry and transcribed using the Superscript II First-Strand cDNA synthesis kit (Invitrogen catalog no. 18064). Reverse transcription products were quantified using an ABI PRISM 7700 (Applied Biosystems). PCR reactions were set up as triplicates using SYBER Green Master Mix (Applied Biosystems). The following primers were used: for Dtkr 5′GAGTAAGCGAAGGGTGGTGAAG and 3′GAACGGCAGCCAGCAGAT; for Nkd 5′GCATCAATGAGGCCAAAGACTT and 3′TGGCAACTGGTACGGTCTTG; and for rp49 5′AGCGCACCAAGGACTTCGT and 3′GCCATTTGTGCGACAGCTT.

Experiments were made in four replicates. All calculations were made using software from Applied Biosystems and statistics performed with Prism Graphpad 5.0.

Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) was used for collecting and analyzing data. For statistical analysis Prism GraphPad 5.0 was used. Log rank tests (Mantel–Cox) were preformed to analyze trends in lifespan in the starvation assays. One-way and two-way ANOVAs were used for the trehalose and lipid assays to test for any relationships between the different genotypes and starvation times.

Antiserum to cockroach (Leucophaea maderae) tachykinin-related peptide 1 (LemTRP-1) is known to recognize DTKs (Siviter et al., 2000; Winther et al., 2003). Immunolabeling with this antiserum revealed neuronal processes with varicose terminations adjacent to the presumed dendrites of IPCs, revealed by Dilp2-Gal4-driven GFP (Fig. 1A,B). The DTK-immunolabeled terminations do not make many direct contacts with the IPC branches. However, neuropeptides are known to signal over some distance in the nervous system, so called volume transmission or paracrine signaling (see Maywood et al., 2011; Nässel, 2009; Zupanc, 1996). We could not identify the individual neurons giving rise to these DTK-expressing processes, but an earlier study (Winther et al., 2003) suggests that the candidate neurons have their cell bodies in the tritocerebrum. Immunocytochemistry with an antiserum to a sequence of the DTK receptor, designated DTKR, labels a set of cell bodies in the pars intercerebralis reminiscent of the IPCs, both in adult (Fig. 1C) and larval brains (Fig. 1D). Unfortunately the DTKR antiserum works only on Bouin-fixed tissues where GFP fluorescence is quenched and also the GFP seems denatured because anti-GFP does not work either. Thus, we were unable to perform DTKR immunolabeling of Dilp2-Gal4-driven GFP for proper identification. The specificity of this receptor antiserum was tested by running immunocytochemistry on fly brains expressing DTKR in IPCs by means of cross Dilp2-Gal4/UAS-Dtkr-GFP. A strong DTKR immunolabeling was seen in the IPCs of these flies (supplementary material Fig. S1).

As DTK-expressing neuron processes seem to target the brain IPCs, we tested whether DTK signaling affects levels of Dilp transcripts in these cells. Knockdown of DTKR on IPCs was accomplished by crossing the Dilp2-Gal4 driver to a UAS-DTKR-RNAi line. The efficacy of the UAS-DTKR-RNAi in knocking down receptor transcripts has been determined previously by quantitative real-time PCR (qPCR) after pan-neuronal RNAi (using an Elav-Gal4 driver); the DTKR transcript was reduced by approximately 40% (Ignell et al., 2009). We sampled RNA from this fly cross at 0 h (fed flies) and 24 h starvation, and performed qPCR to determine levels of Dilp2, Dilp3 and Dilp5 transcripts after DTKR knockdown in IPCs.

In control flies relative Dilp2 and Dilp3 transcripts were not significantly affected by 24 h starvation (Fig. 2A,B) whereas Dilp5 transcript significantly decreased (Fig. 2C). Knockdown of DTKR in IPCs affected the three Dilp transcripts differentially, both in fed and starved flies (Fig. 2). Both Dilp2 and Dilp3 transcripts increased significantly in fed flies after DTKR knockdown in IPCs (Fig. 2A,B), whereas levels of Dilp5 were not affected compared with controls (Fig. 2C). Furthermore, only the Dilp2 and Dilp3 levels were affected by 24 h starvation in the DTKR-knockdown flies. Dilp2 significantly increases compared with fed flies of the same genotype and compared with controls (Fig. 2A) whereas Dilp3 drastically decreases after starvation (Fig. 2B). Thus, it appears as if transcription of both Dilp2 and Dilp3 increases after reduction of DTK signaling via DTKR in normally fed flies and that 24 h starvation drastically decreases Dilp3 transcription and increases Dilp2 when DTKR signaling is reduced in IPCs.

We also measured relative DILP2 immunofluorescence in cell bodies of IPCs in control flies and flies with DTKR diminished in the IPCs to obtain an estimate of peptide levels, using a protocol similar to previous studies (Broughton et al., 2010; Kaplan et al., 2008). There is a significantly higher level of DILP2 immunofluorescence after receptor knockdown in the IPCs (Fig. 3).

To determine whether DTK signaling via its two receptors is modulated by nutritional status we tested the levels of NKD and DTKR receptor transcripts by qPCR in wild-type flies that were fed normally or starved for 24 h. We found no change in relative levels of the two receptor transcripts due to starvation (supplementary material Fig. S2). Thus, it is likely that the strength of DTK signaling to the IPCs is mainly determined by the amount of DTK peptide released during starvation and not by changes in receptor expression levels.

Insulin signaling from brain IPCs affects the lifespan of flies exposed to starvation and dietary restriction (Broughton et al., 2005; Broughton et al., 2010). To study the influence of DTK signaling to IPCs on survival during starvation, experimental flies were allowed access to water but no food. We determined the effects of starvation on total and median lifespan (when 50% of the flies have succumbed) of different genotypes.

Knockdown of DTKR in the IPCs by Dilp2-Gal4/DTKR-RNAi leads to a shortening of total and median lifespan at starvation by approximately 20% compared with controls (P<0.001; log rank test, Fig. 4A). In contrast, overexpression of DTKR in the IPCs did not result in a significant change of lifespan (Fig. 4A).

To test whether the other DTK receptor NKD plays a role in IPCs at starvation, we monitored the effect of expressing UAS-NKD-RNAi or UAS-NKD with the Dilp2-Gal4 driver. Neither of these crosses led to a change in lifespan at starvation compared with controls (Fig. 4B). These results suggest that NKD is not expressed in the IPCs or at least it plays no role in the regulation of starvation responses.

Trehalose is a non-reducing disaccharide and is the major carbohydrate that is used as an energy source by most insects. Insulin signaling from the brain IPCs is known to regulate trehalose levels (Broughton et al., 2005; Rulifson et al., 2002). Specifically, knockdown of Dilp2 is known to increase the total trehalose content (Broughton et al., 2008). Thus, we tested total trehalose levels in flies with DTKR levels diminished in IPCs. Whole-body trehalose levels in flies of different genotypes were tested at 0 h, 5 h and 12 h of starvation. The wild-type response to starvation is a gradual decrease in trehalose levels over the 12 h starvation period. Here, the 0 h value is a measure of trehalose levels under normal fed conditions.

Knocking down DTKR in the IPCs did not affect trehalose levels in fed flies compared with controls (Fig. 5A). However, after 5 h of starvation the flies with diminished DTKR levels displayed a significantly larger drop in trehalose than controls (P<0.001, one-way ANOVA), but after 12 h levels were the same in all genotypes (Fig. 5A). Thus, the trehalose levels in DTKR-knockdown flies fell more rapidly at onset of starvation than in control flies. At starvation the trehalose is likely to be metabolized, partly due to hunger-induced hyperlocomotion, and with increased DILP signaling carbohydrates are depleted more rapidly.

Insects, including Drosophila, utilize lipids as fuel for prolonged exercise or in other situations of high-energy demands, as well as at times of restricted availability of nutrients (Baker and Thummel, 2007; Gäde et al., 1997; Van der Horst et al., 2001). Lipid levels in Drosophila are regulated in part by insulin signaling from brain IPCs (Broughton et al., 2005; Grönke et al., 2010), but also by AKH (Baker and Thummel, 2007; Grönke et al., 2007). Thus, we investigated whole-body lipid levels in flies with DTKR knockdown targeted to IPCs in fed conditions (0 h starvation) and after 24 h of starvation. In fed animals there is no significant difference in lipid levels between controls and DTKR-knockdown flies (Fig. 5B). After 24 h starvation there is a significant drop in lipids both in experimental and control flies (Fig. 5B). This suggests that DTK signaling to IPCs does not affect lipid levels in fed or starved flies or that such an action is masked by compensatory DILP or AKH signaling (see Grönke et al., 2010; Grönke et al., 2007).

We have investigated the effects of DTK signaling to IPCs in the Drosophila brain by monitoring Dilp transcript levels and survival at starvation, as well as trehalose and lipid levels in fed and starved flies. The brain IPCs are presumed to release DILP2, DILP3 and DILP5, orthologs of mammalian insulins (Cao and Brown, 2001; Ikeya et al., 2002). Since these insulin-like peptides have been shown to play a significant role in lifespan, in nutritional stress responses and in metabolic regulation (Baker and Thummel, 2007; Broughton et al., 2005; Giannakou and Partridge, 2007; Grönke et al., 2010; Rulifson et al., 2002; Tatar et al., 2001), the DTK signaling onto these cells may be of significance for the regulation of vital physiological functions. However, the IPCs are also known to regulate feeding behavior, locomotor activity, sleep-wakefulness and ethanol sensitivity, and they may do so independent of the insulin signaling pathway (Cognigni et al., 2011; Corl et al., 2005; Crocker et al., 2010; Mattaliano et al., 2007; Wu et al., 2005). Thus, activation or inhibition of signaling in the IPCs may result in actions that are non-insulin mediated, either via other messengers released by the same cells or indirectly by the action of DILPs on specific neurons (Root et al., 2011; Wu et al., 2005).

The most direct evidence that DTK signaling affects the brain IPCs is that knock down of the receptor DTKR in IPCs leads to altered expression levels of Dilp2 and Dilp3 transcripts in fed flies and that Dilp3 RNA drops drastically in knockdown flies after 24 h starvation whereas Dilp2 levels increase. Interestingly, the Dilp5 transcript is not affected by DTKR-RNAi, but is the only one that seems affected by starvation both in controls and knockdown flies. It is known that restricted diet conditions in adult (control) flies alter the transcript level of Dilp5, but not Dilp2 and Dilp3 (Broughton et al., 2010), corroborating our findings. To our knowledge our data here are the first to quantify Dilp transcripts at complete starvation in adults, but an earlier report monitored Dilp transcripts by in situ hybridization of fed and starved third instar larvae (Ikeya et al., 2002). These authors noted decreased Dilp3 and Dilp5 transcripts but unaffected levels of Dilp2. This difference could be either dependent on a difference in larval and adult functions of the IPCs or could be due to the difference in techniques used for monitoring transcript levels. Certainly the feeding behavior and metabolism differs greatly between larvae and adults in Drosophila (Baker and Thummel, 2007). It should be noted here that insulin expression/signaling also involves autocrine or paracrine feedbacks so that DILP3 may act in stimulatory regulation of expression of DILP2 and DILP5 in the IPCs (Broughton et al., 2008; Grönke et al., 2010) whereas DILP6 released from the fat body may negatively regulate the IPCs (Grönke et al., 2010).

Unfortunately there are no reports that unequivocally demonstrate the release of DILPs into the circulation of Drosophila in a quantitative fashion (but see Geminard et al., 2009). Thus, indirect measurements, such as Dilp transcript or DILP peptide-immunofluorescence levels in cell bodies of IPCs, have to be matched against the physiological effects seen after manipulations of IPCs. A few indicators of altered insulin signaling have been used here: levels of carbohydrate and lipids as well as effects on lifespan at starvation. As one of the functions of DILPs is to stimulate uptake of circulating blood sugar and thereby decreasing trehalose levels in the circulation (Giannakou and Partridge, 2007; Rulifson et al., 2002; Tatar et al., 2003), we monitored whole-body trehalose levels in fed and starved flies after DTKR knockdown in IPCs. Knockdown of DTKR in IPCs had no effect on trehalose in fed flies, but induced an acute drop in trehalose after 5 h starvation, compared with controls, suggesting an increase in insulin signaling. Analysis of Dilp mutants or knockdown indicated that trehalose levels are regulated by DILP2 (Broughton et al., 2008; Grönke et al., 2010); one of the peptides whose transcripts was indeed altered by DTKR knockdown at starvation. In our experiments lipid levels were not affected by manipulations of DTKR on IPCs in fed or starved flies. Lipid metabolism may be regulated by multiple DILPs, including Dilp6 (Grönke et al., 2010), or by compensatory AKH signaling (Baker and Thummel, 2007; Grönke et al., 2007), and this may explain our lack of effect after manipulating only IPC activity.

Diminishment of DTKR expression on IPCs results in flies that display a shortened lifespan at starvation. This would also indicate increased insulin signaling, as deletion of IPCs or knocking down combinations of DILPs produce the opposite phenotype (Broughton et al., 2005; Buch et al., 2008; Grönke et al., 2010), and overexpression of Dilp2 in IPCs resulted in decreased resistance to starvation (Enell et al., 2010). A similar reduction of lifespan at starvation was seen after knock down of the inhibitory GABA_B receptor on IPCs (Enell et al., 2010). It is not clear which of the DILPs regulates the lifespan at dietary restriction or starvation, but DILP2 has been suggested as a candidate (Bauer et al., 2007; Grönke et al., 2010; Hwangbo et al., 2004; Wang et al., 2005).

The second known DTK receptor, designated NKD (Li et al., 1991; Poels et al., 2009), does not seem to play a role in the regulation of IPCs. NKD can be activated only by one of the DTKs, the N-terminally extended DTK-6 (Poels et al., 2009), which has not been detected in the Drosophila brain, in contrast to the DTK-1-5 known to activate DTKR (Winther et al., 2003; Yew et al., 2009).

It can be mentioned that an earlier report shows that DTKR is expressed in renal (Malpighian) tubules where it regulates DILP5 signaling (Söderberg et al., 2011). It is proposed that this regulation is mediated by DTKs released hormonally from endocrine cells of the midgut. DTKs circulating locally act on DTKR expressed in principal cells of the renal tubules, resulting in a local activation of DILP5 signaling. The DTKR-regulated DILP5 signaling in renal tubules does not affect trehalose levels in fed or starved flies, but seems to be part of the defense against oxidative stress (Söderberg et al., 2011). Thus, this DTK-controlled DILP5 signaling in the tubules is probably independent of the paracrine DTK-mediated IPC regulation in the brain, but further studies of gut-derived DTK action are required to confirm this.

In summary our results indicate that in wild-type flies the activated DTKR inhibits insulin signaling in the brain IPCs, and knockdown of the receptor therefore leads to increased insulin signaling. This can be seen in the decreased lifespan and a considerable decrease in trehalose levels during short-term starvation compared with controls, similar to what is expected at increased DILP signaling. The most direct evidence that DTKR is involved in IPC regulation is the effect on Dilp2 and Dilp3 transcript levels seen after receptor knockdown in the IPCs in fed and starved flies. However, it is important for the future to develop a sensitive assay for quantifying hemolymph levels of individual DILPs to monitor how their release is affected by the DTKR signaling to IPCs.

Farideh Rezaei provided technical assistance in an early phase of this project. We thank Paul H. Taghert (University of Washington, St Louis, MO, USA), Ping Shen, (University of Georgia, Athens, GA, USA), the Bloomington Drosophila Stock Center (University of Indiana, Bloomington, IN, USA) and the National Institute of Genetics RNAi Center (Kyoto, Japan) for flies and reagents.