Introduction

Sleep is a fundamental state with significant impacts on many aspects of cognition and metabolism (Blumberg et al., 2005; Davis et al., 2004; Sorribes et al., 2013). While sleep remains essential throughout life, its regulation unfolds dynamically over the course of development. For instance, the pace of sleep-wakefulness cycle is relatively fast in infants, but then gradually slows down and stabilizes in adults (Blumberg et al., 2005; Davis et al., 2004; Sorribes et al., 2013). Furthermore, sleep-wake cycle in infants is independent of circadian rhythm, but it gradually comes under the circadian regulation (Davis et al., 2004; Frank et al., 2017; Poe et al., 2023). Such developmental evolution of sleep-wakefulness cycles appears to be conserved across flies and mammals, including humans (Blumberg et al., 2005; Davis et al., 2004; Poe et al., 2023; Szuperak et al., 2018). In contrast to the deep insight into the molecular and neural mechanisms underlying adult sleep regulation (Barlow & Rihel, 2017; Scammell et al., 2017; Shafer & Keene, 2021), little is known about regulatory mechanisms underlying infant sleep, partially due to a lack of convenient models to study developmental sleep.

Drosophila larvae have recently emerged as a suitable model for studying sleep regulation mechanisms. Recent studies have reported that the second instar larvae show short periods (< 6 seconds) of an “inactive state” during locomotion. Importantly, this “inactive state” is consistent with the general definition of sleep: reduced responsiveness to noxious stimuli, homeostatic responses to sleep deprivation, and rapid reversibility upon stimulation (Szuperak et al., 2018), suggesting that the “inactive state” might correspond to the sleep state. Interestingly, loss-of-function mutations in the clock genes clock and cyc fail to impact larval sleep, while significantly altering the sleep patterns in adult flies (Dubowy & Sehgal, 2017; Hendricks et al., 2003; Szuperak et al., 2018). Likewise, mutations in the dopamine transporter DAT do not influence larval sleep, while reducing sleep amounts in adults (Kume et al., 2005; Szuperak et al., 2018). These observations imply that sleep regulation mechanisms may be, at least in part, distinct between larvae and adults, yet the neural mechanism of larval sleep remains vastly understudied compared to that of adult sleep.

In this study, we performed an unbiased genetic screen in second-instar larvae, and identified the neuropeptide Hugin and its receptor PK2-R1 as a pair critical for larval sleep control. At the circuit level, Hugin-producing HugPC neurons directly stimulate the insulin-producing cells (IPCs) via PK2-R1 to regulate sleep. Surprisingly, we found that regulators of larval sleep are either irrelevant or exert the opposite effects on adult sleep. Overall, our findings uncover the neuropeptidergic circuitry that regulates larval sleep, and highlight the mechanistic differences between larval and adult sleep.

Results

PK2-R1 is required for sleep control in Drosophila larvae

To understand the molecular and neural mechanisms of larval sleep, we modified a previously reported system (Churgin et al., 2019; Szuperak et al., 2018) to automatically quantify the sleep amount of second instar larvae (Figure 1—figure supplement 1) (see Materials & Methods for details). Our system automatically annotates larvae in each frame as either “inactive” or “active” states on the basis of pixel changes, with ∼90% accuracy compared to manual annotation (Figure 1—figure supplement 1C). We found that bouts of ≥ 12 inactive frames are consistent with the general criteria for sleep: reduced responsiveness to noxious stimuli, homeostatic responses to sleep loss, and rapid reversibility upon stimulation (Raizen et al., 2008; Szuperak et al., 2018) (Figure 1—figure supplement 2. Therefore, the present study defines sleep as an inactive state of ≥ 12 consecutive frames (Figure 1—figure supplement 1A).

Using this system, we next carried out an unbiased screen for genes involved in sleep regulation. To this end, we tested 47 CRISPR knock-out lines of genes encoding enzymes or receptors for neuropeptides and monoamines (Deng et al., 2019). As illustrated in Figure 1—figure supplement 3, 13 out of 47 homozygous null mutants exhibited significant changes in sleep amounts compared to the control group. To further test the functions of these candidate genes, we next took advantage of the CRISPR knock-in drivers in which GAL4 sequence is inserted into each gene locus. With these drivers, we silenced neurons expressing each candidate gene by expressing the inward-rectifying potassium channel Kir2.1 (Baines et al., 2001). Silencing neurons expressing Oamb and PK2-R1 resulted in a significant sleep decrease, phenocopying the knock-out of these genes (Figure 1A, 1B, and Figure 1—figure supplement 4D). Oamb encodes the octopamine receptor and has been reported to be involved in larval sleep (Szuperak et al., 2018). The role of PK2-R1 in larval sleep, on the other hand, has been unknown to date, prompting us to focus on this gene.

Figure 1.

Insulin-producing cells express PK2-R1 and regulate sleep in developing larvae

Since the PK2-R1 driver broadly labeled neurons (∼1000 cells) throughout the CNS (Figure 1C), we next attempted to identify a subpopulation of these neurons relevant for sleep regulation. Notably, IPCs in the pars intercerebralis (PI), the neurosecretory center of Drosophila, have been reported to express PK2-R1 (Schlegel et al., 2016). Consistently, the PK2-R1^2A-LexA knock-in driver labeled all IPCs (Figure 2A and 2B). To assess whether IPCs are involved in larval sleep, we silenced IPCs with Kir2.1 expression using the IPC-specific drivers Dilp3- and Dilp5-GAL4 (Figure 2C). We found that silencing of IPCs significantly increased sleep amounts compared to the control, phenocopying PK2-R1 neuron silencing and PK2-R1 knock-out (Figure 1A, 1B, and 2C). Consistently, both Dilp3 and Dilp5 null mutants exhibited larger sleep amounts compared to the control (Figure 2D). These phenotypes of Dilp mutants IPC silencing are unlikely to be attributed to locomotion defects, as the travel distances during wake periods were unaffected (Figure 2—figure supplement 1B and 1C). Collectively, these findings suggest that PK2-R1 in IPCs regulates larval sleep.

Figure 2.

Hugin-expressing neurons control larval sleep

Given that PK2-R1 is a receptor for the neuropeptide Hugin (Rosenkilde et al., 2003), we next examined whether Hugin is involved in larval sleep. We found that Hug mutant larvae exhibited significantly larger sleep amounts compared to the control (Figure 3A), consistent with the PK2-R1 knock-out (Figure 1A). When we silenced Hug-expressing neurons by driving the expression of Kir2.1 with the Hug CRISPR-knock-in driver Hug^2A-GAL4 (Deng et al., 2019), it resulted in a significant increase in sleep amounts without detectable changes in locomotion speed (Figure 3B and Figure 3—figure supplement 1B). Conversely, activation of Hug-expressing neurons with the red-shifted channel rhodopsin ReaChR (Inagaki et al., 2014) or the heat-sensitive ion channel TrpA1 (Hamada et al., 2008) both caused a significant reduction in sleep amount (Figure 3C and 3D). These data together indicate that Hug-expressing neurons are responsible for downregulating larval sleep.

Figure 3.

Hug-expressing neurons are comprised of 20 cells in the subesophageal zone (SEZ) of the brain, and are classified into four subpopulations based on their projection patterns (Bader et al., 2007). Of these, one subpopulation called HugPC neurons (Bader et al., 2007) has been reported to form synaptic connections with IPCs (Hückesfeld et al., 2016). We thus suspected that HugPC neurons may be responsible for sleep regulation. In line with this notion, axons of HugPC neurons labeled by a newly generated HugPC-LexA driver (Figure 3—figure supplement 2) and IPCs labeled by Dilp3-GAL4 are located in close proximity to each other (Figure 3E). Indeed, silencing of HugPC neurons by Kir2.1 significantly increased larval sleep (Figure 3F), phenocopying those observed in Hug and PK2-R1 mutants. These results hint at the possibility that HugPC neurons activate IPCs via Hugin/PK2-R1 signaling to regulate sleep.

Hugin triggers Ca²⁺ elevation and Dilp secretion in larval IPCs via PK2-R1

A previous report showed that IPCs exhibit increased firing rates and intracellular calcium (Ca²⁺) levels upon stimulation (Kréneisz et al., 2010). This motivated us to test whether Hugin activates IPCs. To this end, we utilized the recently developed calcium indicator CRTC::GFP, which can report Ca²⁺ dynamics in the timescale of several minutes (Bonheur et al., 2023). As glucose reportedly triggers Ca²⁺ dynamics in IPCs in a range of minutes (Oh et al., 2019), we wondered if CRTC::GFP can detect these dynamics as a positive control. To this end, we applied D-glucose to isolated larval brains and found a significant cytosol-to-nuclear translocation of CRTC::GFP in IPCs within ∼5 min, confirming that this reporter can indeed detect Ca²⁺ responses in IPCs upon stimulation (Figure 5A and 5B). We then examined whether Hug neurons act upstream of PK2-R1-expressing IPCs, by measuring Ca²⁺ responses of IPCs while thermogenetically activating Hug neurons with TrpA1 (Figure 4A). In both isolated brains (ex vivo) and living larvae (in vivo), we found that activation of Hug neurons significantly increased Ca²⁺ levels in IPCs (Figure 4B and 4C), suggesting that Hug neurons act upstream of IPCs (Figure 4D). Next, we tested whether Hug peptides can induce Ca²⁺ responses in IPCs using ex vivo preparations. The Hug gene encodes a precursor that potentially produces at least two distinct peptides, Hug-γ and PK-2, both of which can bind PK2-R1 to induce Ca²⁺ responses (Rosenkilde et al., 2003). We thus incubated the CRTC::GFP-expressing larval brain with chemically synthesized Hug-γ or PK-2 (Figure 5A). We found that bath application of either Hug-γ or PK-2 induced significant Ca²⁺ responses in IPCs (Figure 5B–D). In agreement with the idea that Hug peptides activate IPCs via PK2-R1, knock-out of PK2-R1 blocked the Hug peptide-evoked Ca²⁺ elevation in IPCs (Figure 5G). Similar results were obtained when we bath-applied the synthetic Neuromedin U (NMU), the mammalian homlog of Hugin (Figure 5A, 5C, and 5G), suggesting that the role of Hugin/PK2-R1 signaling may be conserved across species. Furthermore, Dilp3 accumulated in IPCs was significantly reduced after applying Hug peptides, suggesting that Hugin causes IPCs to release Dilp3 (Figure 5E). Collectively, these data suggest that Hug neurons activate IPCs via Hugin/PK2-R1 signaling to regulate sleep (Figure 5H).

Figure 4.

Figure 5.

Hugin/PK2-R1 axis has distinct impacts on sleep control in larvae and adults

Given the differences between larval versus adult sleep in temporal pattern and circadian influence, we next wondered if the Hugin/PK2-R1/Dilps axis functions in adult sleep as well. As in larvae, PK2-R1 knock-out adults showed increased sleep amount compared to the control (Figure 6—figure supplement 1B). In contrast, Hug knock-out or silencing of HugPC neurons failed to influence adult sleep (Figure 6A and 6B). This suggests that, unlike in larvae, Hug is dispensable for adult sleep.

Figure 6.

Interestingly, the expression patterns of PK2-R1 and Hug, as well as the morphology of HugPC neurons and IPCs, were almost comparable between larvae and adults (Figure 6—figure supplement 2). This implies that the differential roles of Hug in larvae versus adults are likely due to physiological differences in HugPC neurons and/or IPCs. We thus examined how Hug peptides evoke Ca²⁺ responses in adult IPCs and found that neither Hug-γ nor PK-2 could induce Ca²⁺ responses in adult IPCs (Figure 6D and 6E), unlike in larvae. Consistently, Hug peptides failed to reduce Dilp3 levels in adult IPCs (Figure 6F). These data indicate that Hug induces Ca²⁺ responses and Dilp secretion in larval IPCs, but not in adult IPCs. Surprisingly, we further found that Dilp3 or Dilp5 null mutations reduced sleep amounts in adults (Figure 6C), while in larvae, these mutations in fact increased the sleep amounts. These data unexpectedly reveal how the same set of neuronal circuits and molecular signaling therein can evolve over development in a complex manner.

Discussion

In this study, we have identified the neuropeptide Hugin and its receptor PK2-R1 as a ligand/receptor pair critical for sleep regulation in Drosophila larvae. We have further shown that larval IPCs express PK2-R1, respond to and release Dilp3 upon Hugin stimulation. Surprisingly, the Hugin/PK2-R1 axis is dispensable for sleep regulation in adults, even though gene expressions as well as the circuit structure appeared to be conserved between larval and adult brains. Furthermore, Dilps appear to modulate sleep in opposite directions in larvae and adults. Our findings thus uncover the neuropeptidergic modulation circuitry that specifically regulates developmental sleep, and suggest divergent usage of the same molecules and circuitry for sleep modulation in larvae and adults.

Hugin/PK2-R1/Dilps axis negatively regulates sleep in larvae

The present study shows that the neuropeptide Hugin and its receptor PK2-R1 are required for sleep regulation in the early larval stage. This notion is supported by the following lines of evidence. First, knock-out mutations in Hug or PK2-R1 increased sleep amounts compared to the control (Figure 1A and 3A). Consistently, silencing a subpopulation of Hug-expressing neurons, called HugPC neurons, or PK2-R1 neurons increased sleep amounts (Figure 1B and 3F). Second, thermogenetic activation of Hug neurons decreased sleep amounts (Figure 3D). Third, HugPC neurons project their axons to IPCs in the brain that express PK2-R1 receptors (Figure 3E). Fourth, Hug peptides induced Ca²⁺ elevation and promoted Dilp3 release from IPCs in a PK2-R1- dependent manner (Figure 5). Last, Dilp3 null mutations increased sleep amounts (Figure 2D). Based on these data, we propose a model in which the Hugin/PK2- R1/Dilps axis along the HugPC-IPC circuitry modulates larval sleep (Figure 7).

Figure 7.

Examples of Hugin acting through PK2-R1 are prominent in the context of feeding regulation in larvae and adults (Schoofs et al., 2014). For example, a recent study has suggested that Hugin regulates the timing of pupariation and body wall contraction during pupariation through PK2-R1 in PTTH neurons (Ohhara et al., 2024). In the present study, we showed that HugPC in the larval brain modulates sleep (Figure 3F). A previous electron microscopic analysis suggested that larval HugPC neurons likely form synaptic connections to downstream neurons, including IPCs (Hückesfeld et al., 2021; Schlegel et al., 2016). We here showed that larval sleep control requires PK2-R1, and that thermogenetic activation of Hug neurons causes Ca²⁺ responses in PK2-R1- expressing IPCs (Figure 4). Importantly, our ex vivo data indicated that Hugin peptide- induced Ca²⁺ response in IPCs was largely abolished by PK2-R1 mutation (Figure 5), indicating the critical role of the Hugin/PK2-R1 axis for the HugPC-mediated IPC activation. Together, we propose that the HugPC-IPCs circuity mobilizes Hugin/PK2-R1 signaling to selectively modulate larval sleep. Interestingly, our ex vivo data indicated that Neuromedin U (NMU), a mammalian ortholog of Hugin, can trigger Ca²⁺ responses in larval IPCs in a PK2-R1-dependent fashion (Figure 5). Given that NMU has been implicated in sleep regulation in fish and mammals (Ahnaou & Drinkenburg, 2011; Chiu et al., 2016), an analogous ligand/receptor pair might be involved in sleep modulation in mammals as well.

Divergent roles of insulin in regulation of larval sleep and feeding

IPCs in larvae responded to exogenously applied Hug peptides, both by exhibiting elevated Ca²⁺ levels and decreases in intracellular Dilp3 storage (Figure 5C–E). In contrast to Dilp3, however, the amount of Dilp2 accumulated in IPCs did not decrease following Hug peptide application (Figure 5—figure supplement 1B). On the other hand, Dilp2 accumulation in IPCs decreased after glucose treatment (Figure 5—figure supplement 1A) (Oh et al., 2019). These results unexpectedly suggest that larval IPCs form parallel channels inside, one for Hug-Dilp3 and the other for glucose-Dilp2.

Mechanistically, Hug acts directly on IPCs via the PK2-R1 receptor. As for glucose, on the other hand, glucose-sensing CN neurons in the brain take up glucose and subsequently secrete another neuropeptide to activate IPCs (Oh et al., 2019). Such parallelism inside IPCs might provide insights into why insect species broadly express multiple insulin-like peptide genes, in some cases even as many as 40 genes in the silkworm Bombyx mori (Okamoto, 2021). Drosophila genome harbors eight Dilp genes (Nässel et al., 2015), of which Dilp2, Dilp3, and Dilp5 are considered nutrient-related.

Nevertheless, mRNA expression or release patterns of these Dilps are not identical under different nutrient conditions (Géminard et al., 2024; Nässel et al., 2015; Oh et al., 2019; Ikeya et al., 2002). Furthermore, these Dilps exhibit different binding affinities for the insulin receptor (InR), the secreted decoy of InR (SDR), and the ecdysone-inducible gene L2 (Imp-L2) (Okamoto et al., 2013). These reports hint at the differential roles of different insulin-like peptides; however, mechanistic insights remain largely lacking. Our data on larval IPCs forming parallel channels for Hug-Dilp3 and glucose-Dilp2 may provide a point of entry for tackling this mystery.

Differential impacts of the same neuropeptidergic modulations on larval and adult sleep

Our data show that the Hugin/PK2-R1/Dilps axis likely downregulates sleep amounts in larvae. Despite the significant increase in sleep amount in Hug mutant larvae, Hug mutant adults showed sleep amounts comparable to wild-type control (Figure 6A), consistent with a previous report that Hugin is dispensable for control of baseline sleep levels (Schwarz et al., 2021). According to our histological analyses, the Hugin/PK2-R1 expression patterns, as well as the HugPC-IPCs circuitry, appear to be largely conserved in adults (Figure 3E and Figure 6—figure supplement 2). Unlike larval brains, however, Hug peptides failed to trigger Ca²⁺ responses in adults IPCs (Figure 6E). It is thus possible that Hugin/PK2-R1 signaling along the HugPC-IPCs circuitry is suppressed in adults. IPCs in adults receive multiple positive and negative modulatory inputs through GPCRs including the metabotropic GABA_B receptors (Enell et al., 2010), which suppress IPC activity and Dilp release in adult IPCs (Enell et al., 2010). It is thus plausible that such negative modulatory inputs to IPCs in adults might counteract the Hugin/PK2-R1 axis to suppress Dilp release. In addition, our data suggest that Dilps modulate sleep amount in the opposite directions in larvae and adults (Figure 7).

Comparing the expression levels and activities of GPCRs in larval and adult IPCs would be essential to better understand how same modulatory signals over the course of development come to exert differential impacts on sleep. Interestingly, Hugin in adults appears irrelevant for the baseline sleep amount but is required for homeostatic regulation of sleep (Schwarz et al., 2021). Thus, testing whether the Hugin/PK2-R1 axis is involved in the homeostatic regulation of larval sleep, and how this system compares to its adult counterpart, may further provide mechanistic insights into how homeostatic sleep regulation maturates over development.

In summary, we identified the Hugin/PK2-R1/Dilps signaling along the HugPC/IPCs circuit as the molecular and circuitry basis for larval sleep regulation. Our study begins to fill the knowledge gap in larval sleep regulation, and sheds light on mechanistic differences in sleep regulation between early developmental stages and adults.

Materials and methods

Key resource table

Fly strains

The following strains of Drosophila melanogaster were obtained from the Bloomington Drosophila Stock Center: iso31 (an isogenic w¹¹¹⁸ strain; BDSC# 5905), PK2-R1^attP (BDSC# 84563), PK2-R1^2A-GAL4 (BDSC# 84686), PK2-R1^2A-LexA(BDSC# 84431), Dilp3¹(BDSC# 30882), Dilp5¹ (BDSC# 30884), Dilp3-GAL4 (BDSC# 52660), Dilp5-GAL4 (BDSC# 66007), Hug^attP (BDSC# 84514), Hug^2A-GAL4 (BDSC# 84646), Hug^2A-LexA (BDSC# 84397), UAS-Kir2.1::EGFP (BDSC# 6596), UAS-Stinger, LexAop-tdTomato.nls (BDSC# 66680), UAS-mCD8::mCherry, UAS-CRTC::GFP (BDSC# 99657), UAS-ReaChR (BDSC# 53741), UAS-TrpA1 (BDSC# 26263), LexAop-ReaChR (BDSC# 53746), LexAop-rCD2::RFP-p10.UAS-mCD8::GFP-p10 (BDSC# 67093). w¹¹¹⁸ was from Y. Rao (Peking University); HugPC-GAL4 (Hückesfeld et al., 2016) was from M. Pankratz (University of Bonn); LexAop-TrpA1 (Burke et al., 2012) was from S. Waddel (University of Oxford). HugPC-LexA was generated in this study (see the following section, “Generation of transgenic lines”). Detailed lists of fly genotypes can be found in the key resource table and Source Data files.

General fly maintenance

Flies were maintained on conventional fly food under a 9AM:9PM light/ dark cycle at 25°C and 60 +/- 5 % humidity.

Egg collection

Agar substrates for egg collection were prepared by microwaving a mixture of 2.0 g sucrose (Fujifilm Wako Pure Chemicals, cat# 196-00015), 3.0 g agar (BD, cat#214010), 2.5 mL apple juice (Dole), and Milli-Q water up to 100 mL. Melted agar was poured into Petri dishes with a diameter of 4 cm (approximately 3 mL per dish). Solidified agar was scratched with tweezers to make the surface rough. Agar plates were stored inside a sealed container at 4°C. Upon egg collection, yeast paste (dried yeast (Nippon Beet Sugar Co., Ltd.) kneaded with Milli-Q water at 40% (w/w)) was placed in the center of the agar plates to further promote oviposition.

Larval staging

First-instar larvae were collected two days after egg collection. Among them, second-instar larvae that molted within 2 h were used for video recording. The distinction between first- and second-instar larvae was based on morphological characteristics: while anterior spiracles were not evident in the first-instar, fist-shaped spiracles were observed in the second-instar.

Larval sleep analysis

Twenty-four-well silicone chambers for sleep analysis, described in Figure 1—figure supplement 1B, were made by pouring dimethylpolysiloxane (PDMS) into molds based on a previous study (Churgin et al., 2019; Szuperak et al., 2018). As a substrate, 120 µL of a mixture of 3% agar and 2% sucrose was poured into each well. After the gel solidified, 10 µL of 60 mg/mL yeast suspension was applied to the surface of the gel in each well. One larva was transferred into each well, and a glass plate (Azone, cat# 1-4540-01, 5 mm × 200 mm × 200 mm) was placed on top of the chamber to prevent the larvae from escaping. The chamber was placed inside a dark box equipped with a ring-type infrared LED (CCS, cat# LDR2-132IR940-LA) and a camera (The Imaging Source, cat# DMK27BUP031), and recording was started immediately without acclimation. Images were captured through the IC Capture software (The Imaging Source, version 2.5) placed above the chamber under the following conditions: MJPEG Compressor for codec, a frame rate of 0.87 fps, and an image size of 2592 × 1944. For each well and frame, the image of larva was blurred and binarized to automatically locate the larva. Consecutive frames were then compared so that changes in values were summed to estimate the larval motion. This motion estimate was then used to classify every consecutive pair of frames as representing “active” or “inactive.” We found that inactivity across 12 consecutive frames meets the sleep criteria, as shown in Figure 1—figure supplement 2.

Sleep deprivation

Blue light irradiation was performed using a 455-nm LED (Thorlabs, cat# M455L4) with a collimating adapter (Thorlabs, cat# SM1U25-A). The irradiation intensity was set to the maximum scale using a T-cube LED driver (Thorlabs, cat# LEDD1B). Temporal control of the LED illumination was achieved using an Arduino UNO microcontroller board (Arduino, RRID:SCR_017284). For sleep deprivation experiments, the first 1 h was recorded without LED illumination, followed by another 1 h consisting of repeated cycles of “LED ON for 90 s and OFF for 30 s.”

Immunohistochemistry

The following steps were carried out in succession: dissection in phosphate-buffered saline (PBS); fixation in 4% paraformaldehyde/PBS at room temperature for either 90 min in CRTC::GFP experiments or 30 min in the others; washing with 0.3% Triton X-100/PBS (hereafter referred to as PBST) for 10 min × 3; blocking in 5% normal goat serum (NGS)/PBST at room temperature for 30 min; primary antibody treatment at 4°C overnight; washing with PBST for 10 min × 3; secondary antibody treatment at 4°C overnight; washing with PBST for 10 min × 3; and mounting in VECTASHIELD medium (Vector Laboratories, cat# H-1000, RRID:AB_2336789). The above steps were performed either in Terasaki 96-well plates (stem Corp., cat# P96R37N) for adults or in 48-well culture plates (Corning, cat# 3548) for larvae. The following primary antibodies were diluted in 5% NGS/PBST: chicken polyclonal anti-GFP (1:1000, Aves Labs, cat# GFP-1020, RRID:AB_10000240), mouse monoclonal anti-mCherry (1:1000, Takara Bio, cat# 632543, RRID:AB_2307319), rabbit polyclonal anti-Dilp3 (1:250, (Veenstra et al., 2008)), rabbit polyclonal anti-Dilp2 (1:2000, (Okamoto et al., 2012)), mouse monoclonal anti-BRP (1:100, Developmental Studies Hybridoma Bank, clone nc82; RRID:AB_2314866), and rat monoclonal anti-histone H3 phospho S28 (1:1000, Abcam, cat# ab10543; RRID:AB_2295065). The following secondary antibodies were diluted in 5% NGS/PBST: goat anti-chicken Alexa Fluor 488 (1:200, Thermo Fisher Scientific, cat# A-11039, RRID:AB_2534096), goat anti-mouse Alexa Fluor 555 (1:200, Thermo Fisher Scientific, cat# A-21424, RRID:AB_141780), goat anti-rabbit Alexa Fluor 633 (1:200, Thermo Fisher Scientific, cat# A-21071, RRID:AB_2535732), and goat anti-rat Alexa Fluor 633 (1:200, Thermo Fisher Scientific, cat# A-21094; RRID:AB_2535749). Fluorescence images were obtained using a confocal microscope (Leica, TCS SP8) with 20× (Leica, HC PL APO CS2 20×/0.75 IMM) and 63× (Leica, HC PL APO CS2 63×/1.40 OIL) objectives. Images were captured at 512 × 512 pixel resolution in 1 µm-steps along the z-axis. Z-stack images of the maximum projections were generated using the ImageJ v1.53t software ((Schneider et al., 2012); RRID:SCR_003070).

Generation of transgenic lines

HugPC-LexA lines were generated by ΦC31 integrase-mediated transgenesis, as previously described (Groth et al., 2004). The 544-bp enhancer region of Hug, identical region to that used for HugPC-GAL4 (Hückesfeld et al., 2016), was PCR-amplified from fly genomic DNA. The gel-purified amplicon was subjected to another round of PCR to add hangout sequences. The backbone suicide-vector pBPnlsLexA::p65Uw (RRID:Addgene_26230) was initially amplified by transforming the ccdB-resistant E. coli strain, One Shot™ ccdB Survival™ 2 T1^R competent cells (Thermo Fisher Scientific, cat# A10460). The PCR product and the vector were digested by Aat II (NEB, cat# R0117S) and Fse I (NEB, cat# R0588S), followed by ligation using In-Fusion HD Cloning Kit w/Cloning Enhancer (Takara Bio, cat# 639635). The integrity of the resulting plasmid was confirmed by DNA sequencing. Plasmid insertion was targeted to four attP sites (attP2, attP40, VK00005, and su(Hw)attP5) individually through ΦC31 integrase-mediated transgenesis by BestGene Inc. (Chino Hills, CA).

Transformants were selected based on the eye color marker. Among the generated strains harboring HugPC-LexA in each insertion site, labeling patterns in the larval brain were checked by driving fluorescent marker expression under the control of LexAop sequence (Figure 3—figure supplement 2). As the HugPC-LexA inserted in attP2 closely resembled the labeling pattern of the original HugPC-GAL4, we chose this driver in further experiments.

The following primers were used for cloning:

Forward primer for the HugPC enhancer; 1^st round PCR against genomic DNA:

5’-AAGGGTTTGGTTTAATTTATTTATGTCATA-3’

Reverse primer for the HugPC enhancer; 1^st round PCR against the genomic DNA:

5’-GGACAACTGATGCCAGCAGC-3’

Forward primer for the HugPC enhancer with hangout sequences (underlined); 2^nd round PCR against the 1^st round amplicon:

5’-gaaaagtgccacctgacgtAAGGGTTTGGTTTAATTTATTTATGTCATA-3’

Reverse primer for the HugPC enhancer with hangout sequences (underlined); 2^nd round PCR against the 1^st round amplicon:

5’-cccgggcgagctcggccggGGACAACTGATGCCAGCAGC-3’

Measurement of neuronal activity and Dilp3 accumulation in IPCs

The GFP-tagged CRTC (CRTC::GFP) reporter was applied to quantify the activities of larval IPCs, essentially according to the previous report in adult flies (Bonheur et al., 2023). Parental Dilp3-GAL4 and the reporter line harboring UAS-CRTC::GFP and UAS-mCD8::mCherry (BDSC# 99657) were crossed, and eggs were collected within an 8–10 h time window. Offspring larvae were reared for 4 days after egg laying (AEL) at 25°C until the early third instar. Collected larvae were washed with water and starved on water-soaked paper for 12–14 h. For in vivo thermogenetic activation, larva-containing Petri dishes were warmed at 30°C for 60 min. For ex vivo thermogenetic experiments, larvae were dissected in calcium- and sugar-free HL3.1 buffer (Nakamizo-Dojo et al., 2023) and each brain was incubated in the same buffer at 30°C for 5 min.

For ex vivo glucose or peptide application assays, larvae were first dissected in the same buffer as above at room temperature, and brain samples were then treated with either 20 mM D-glucose or 1 mM of chemically synthesized peptides dissolved in HL3.1 buffer for 5 min. After incubation, samples were subjected to immunohistochemical analyses to visualize GFP, mCherry, and anti-Dilp3 signals (see the section “Immunohistochemistry”).

Image processing was performed using ImageJ v1.53t ((Schneider et al., 2012); RRID:SCR_003070). Nuclear ROIs and cell body outlines were determined based on the mCD8::mCherry fluorescence insensitive to neuronal activity, and the mean intensities of CRTC::GFP signals within the nucleus (F_nuc) and the cytoplasm (F_cyto = F_{whole cell} - F_nuc) were quantified. NLI was calculated using the following formula: (F_nuc - F_cyto)/(F_nuc + F_cyto). To assess Dilp3 accumulation in IPCs, the anti-Dilp3 signal intensity was measured within the cytosol of each cell, excluding the nucleus. Data from 4–14 Dilp3-positive IPCs per brain, collected from 3–7 larvae, were pooled for each genotype/treatment group.

Chemical synthesis of Hug peptides

Peptides with the following amino acid sequences were chemically synthesized using the Fmoc solid phase method, and their purity was assessed by HPLC to be over 95%. All procedures were carried out by Eurofins Genomics K.K. (Tokyo, Japan). The amino acid sequences of the peptides used were as follows:

• Drosophila Hug-γ:

  • Acetyl-QLQSNGEPAYRVRTPRL-CONH2 (17 a.a.)

• Drosophila PK-2:

  • Acetyl-SVPFKPRL-CONH2 (8 a.a.)

• Human NMU:

  • Acetyl-FRVDEEFQSPFASQSRGYFLFRPRN-CONH2 (25 a.a.)

Sleep quantification in adult flies

The following parts of the Drosophila activity monitor (DAM) system were purchased from TriKinetics Inc. (Waltham, MA): the Drosophila Activity Monitor (cat# DAM2), the power supply interface unit (cat# PSIU9), monitor tubes with an outside diameter of 5 mm and a length of 65 mm (cat# PGT5x65), and tube caps (cat# CAP5-Black). Newly eclosed adult flies were collected under brief CO₂ anesthesia and kept as groups of 10– 15 in standard food vials. Males and females were kept separately to avoid mating. To minimize disturbance by noise and vibration, flies were placed inside a fan-less Peltier-type incubator (Mitsubishi Electric Engineering Co., Ltd., cat# SLC-25) set to 25°C, externally connected with 5-mm diameter LED lights (MY-CRAFT, Ltd., cat# 5W0601) and a timer device (Panasonic Corp., cat# WH3311WP) to generate a 9AM:9PM light/dark cycle. After 3 days of eclosion, each fly was gently aspirated into a Pyrex glass monitor tube, one end of which was stuffed with agar food (2% (w/w) agar and 5% (w/w) sucrose), and the other was plugged with a cotton string. The tubes were set to the monitor device and placed inside the same fan-less incubator. The DAM system was connected to an external PC running DAMSystem3 Data Collection Software (RRID:SCR_021809; available at https://trikinetics.com/), which counted beam-crossings by each fly for two successive days. Time windows longer than 5 min without beam-crossing were judged as sleep bouts, following the conventional criteria for adult fly sleep (Paul et al., 2000).

Statistical analysis

Statistical analyses by Kruskal–Wallis one-way ANOVA, Mann–Whitney U-test, or chi-squared test were performed using Prism 9.5.1 (GraphPad, RRID:SCR_002798). Bonferroni correction was applied for multiple comparisons. Asterisks (*) represent p-values as indicated within each figure legend. All data necessary to reproduce the figure panels and statistical analyses are available as Source Data files.

Figure supplements

Figure 1—figure supplement 1.

Figure 1—figure supplement 2.

Figure 1—figure supplement 3.

Figure 1—figure supplement 4.

Figure 2—figure supplement 1.

Figure 3—figure supplement 1.

Figure 3—figure supplement 2.

Figure 5—figure supplement 1.

Figure 6—figure supplement 1.

Figure 6—figure supplement 2.

Acknowledgements

We thank Y. Rao, M. Pankratz, S. Waddel, and the Bloomington Drosophila Stock Center for fly stocks; J. A. Veenstra for the anti-Dilp3 antibody; T. Nishimura for the anti-Dilp2 antibody; Y. Nishizuka for valuable suggestions on writing codes for automated larval detection and sleep quantification; H. Ito, N. Yamaji, M. Maetani, A. Ogasawara, E. Kato, T. Rikiishi, S. Ando, M. Hayashi, and S. Miyazaki for technical assistance and fly maintenance; and the members of Emoto Lab for critical comments and discussion. This work is supported by MEXT Grants-in-Aid for Scientific Research on Innovative Areas “Dynamic regulation of brain function by Scrap and Build system” (KAKENHI 16H06456), JSPS (KAKENHI 16H02504), WPI-IRCN, AMED-CREST (JP22gm310010), JSTCREST, Toray Foundation, Naito Foundation, Takeda Science Foundation, and Uehara Memorial Foundation to K.E.; the Leading Initiative for Excellent Young Researchers (LEADER) from MEXT, JSPS (KAKENHI 22K06309), and AMED-PRIME (JP22gm6510011) to K.I.; Grant-in-Aid for JSPS Fellows (24KJ0911) to M.M.; and Grant-in-Aid for JSPS Fellows (22J21096; 22KJ1042) to C.H.

Additional information

Author contributions

C.H., K.I., and M.M. performed experiments and data analyses; M.T. wrote the codes for automated larval detection and sleep quantification; C.H., K.I., and K.E. are responsible for study conception, experimental design, and data interpretation; C.H., K.I., and K.E. wrote the original manuscript, and M.T. edited the paper.