Dietary Amino Acids Promote Glucagon-like Hormone Release to Generate Novel Calcium Waves in Adipose Tissues

Nutrient sensing and the subsequent metabolic responses are fundamental functions of animals, closely linked to diseases such as type 2 diabetes and various obesity-related morbidities. Among different metabolic regulatory signals, cytosolic Ca2+ plays pivotal roles in metabolic regulation, including glycolysis, gluconeogenesis, and lipolysis. Recently, intercellular calcium waves (ICWs), the propagation of Ca2+ signaling through tissues, have been found in different systems to coordinate multicellular responses. Nevertheless, our understanding of how ICWs are modulated and operate within living organisms remains limited. In this study, we explore the real-time dynamics, both in organ culture and free-behaving animals, of ICWs in Drosophila larval and adult adipose tissues. We identified Adipokinetic hormone (AKH), the fly functional homolog of mammalian glucagon, as the key factor driving Ca2+ activities in adipose tissue. Interestingly, we found that AKH, which is released in a pulsatile manner into the circulating hemolymph from the AKH-producing neurosecretory cells (APCs) in the brain, stimulates ICWs in the larval fat by a previously unrecognized gap-junction-independent mechanism to promote lipolysis. In the adult fat body, however, gap-junction-dependent random ICWs are triggered by a presumably uniformly diffused AKH. This highlights the stage-specific interplay of hormone secretion, extracellular diffusion, and intercellular communication in the regulation of Ca2+ dynamics. Additionally, we discovered that specific dietary amino acids activate the APCs, leading to increased intracellular Ca2+ and subsequent AKH secretion. Altogether, our findings identify that dietary amino acids regulate the release of AKH peptides from the APCs, which subsequently stimulates novel gap-junction-independent ICWs in adipose tissues, thereby enhancing lipid metabolism.


Introduction
Adipose tissue is a primary organ of fat storage that coordinates energy metabolism pathways under various nutritional states.Such metabolic processes are under the hormonal control of insulin and glucagon, the two key hormones playing a central role in energy storage and homeostasis regulation.
Notably, glucagon promotes lipid breakdown in the adipose tissue by elevating cytoplasmic Ca 2+ levels through activation of its GPCR receptor 1 .Intriguingly, these Ca 2+ activities are spatially organized into intercellular Calcium waves (ICWs) across the mammalian liver 2 .However, the mechanism and biological signi cance of this collective ICWs in vivo remains poorly understood.Fruit ies have a functional homolog of glucagon, namely Adipokinetic hormone (AKH), which plays similar roles to mobilize adipose lipids upon starvation [3][4][5] .In this study, we demonstrated that AKH triggers signi cant ICWs in the y adipose tissue, presenting a unique opportunity to investigate the mechanics and regulations of ICWs in vivo.
ICWs are complex tissue-level biological events that exist in a wide array of cell types in multicellular organisms and that impact many biological processes relevant to cellular life (e.g., muscle contraction 6,7 , gene expression 8,9 , cellular proliferation 10 , differentiation 11 , neuronal ring and excitability 12 , and metabolism 13 ).Dysregulated Ca 2+ signaling can result in various disease states ranging from various neurological disorders to cardiovascular diseases and cancer 14,15 .ICWs are generated by an increase of cytoplasmic Ca 2+ ions concentration from organelle reserves as well as propagation of the signal from one cell to its neighbors, which consequently coordinate concerted action at the tissue level.The propagation of ICWs has been reported to be governed by two mechanisms: (1) the direct diffusion of the secondary messengers i.e.Ca 2+ and IP 3 through gap junctions that stimulate Ca 2+ e ux in neighboring cells 16,17 ; and (2) paracrine signaling, where the secretion of extracellular ATP stimulates nearby cells to release Ca 2+ from the endoplasmic reticulum (ER) 18 .Despite these insights, the underlying biophysical mechanisms and the biological signi cance of the emergent pattern and signal propagation of ICWs in different biological systems have largely remained unidenti ed.
Here, using the Drosophila larval fat body/adipose tissue as a model to study ICWs in vivo, we analyzed the real-time dynamics of ICWs by live-cell imaging.We show that novel global ICWs are triggered by AKH in the larval fat body, reminiscent of glucagon-induced Ca 2+ waves observed in the human liver 1 .More interestingly, genetic perturbation of gap junctions signi cantly abrogates ICWs triggered by AKH in cultured fat bodies, indicating that these cellular junctions are required to propagate the ICWs.However, disruption of gap junctions did not affect the global propagation of waves at the tissue level, leading us to propose the existence of a gap junction-independent mechanism that helps propagate ICWs across the tissue.To characterize this mechanism, we used a combination of experimental and computational modeling approaches.Speci cally, we determined that these gap-junction-independent ICWs are regulated through facilitated extracellular diffusion of AKH.Indeed, global wave propagation requires a constant circulation of hemolymph which depends on cardiac pumping, as disrupting larval lymph circulation completely abrogates the tissue-level wave propagation.Meanwhile, the global ICWs were triggered by the periodic release of AKH from the brain corpora cardiaca AKH-producing neurosecretory cells (APCs), as silencing the APCs immediately stops the ICWs in the fat body.This nding is further supported by computational simulation of the ICWs that produces similar patterns and outcomes resembling our in vivo observations.
Finally, we utilized this system to identify the dietary factor regulating the AKH-dependent ICWs.Previous studies have demonstrated that AKH, like its functional mammalian homolog Glucagon, plays a pivotal role in maintaining blood sugar homeostasis and modulating starvation-induced hyperactivity in adult.However, compared to insulin, the regulation of AKH remains less understood.Various dietary, hormonal and neuronal factors regulating insulin signaling have been identi ed in the y 19 .Notably, it has been recently demonstrated that y insulin-producing cells directly sense the dietary amino acid leucine, which consequently regulates the secretion of dILPs 20 .In contrast, whether speci c amino acids exert regulatory control over AKH secretion remains unknown.In this study, we found that secretion of AKH from the APCs is not only repressed by sugar, but also promptly stimulated within minutes after consumption of particular amino acids and that amino acids consumption promotes fat loss in both larvae and adult ies through AKH-dependent signaling.Collectively, our study suggests that speci c amino acids regulate AKH/Glucagon signaling, which in turn activates ICWs throughout adipose tissues, coordinating lipid metabolism.

Materials and Methods
Fly husbandry.Flies were raised on standard food (2000 ml water, 12.8 g agar, 80 g yeast, 112 g cornmeal, 176 g glucose, 2.5 g methylparaben, 20 ml propionic acid, total 2 l of y food) at 25°C with 12 h:12 h light: dark cycles.Fly strains used in this study are listed in Supplementary Table 1.
Ex vivo GCaMP imaging in APCs.Early 3rd instar larval brains were dissected in modi ed basal hemolymph-like solution devoid of any amino acids (modi ed HL6(AA-) buffer) (74.2 mM NaCl, 2.0 mM MgCl 2 , 10.0 mM NaHCO 3 , 24.8 mM KCl, 0.5 mM CaCl 2 , 80 mM Trehalose, 5 mM BES, pH 7.2).Dissected brains were immobilized using a holder in the perfusion chamber.The samples in the HL6 (AA-) buffer were recorded for 1 min to generate a baseline.Next, the solutions were changed to HL6 (AA-) buffer + AA (5 mM) with the pH adjusted back to 7.2 by gentle perfusion for 10 min.All imaging studies were performed with a Leica M205 FCA high-resolution stereo uorescence microscopy (Leica).
Ex vivo imaging of larval fat bodies.The ex vivo imaging chambers were assembled on a live cell imaging dish (Nest, 801001), a metal ring with an inner diameter of 10 mm, and a cellulose lm from a tea bag modi ed to match the chamber.Brie y, the fat bodies of 3rd instar larvae were dissected in Schneider's medium (Sigma) and placed at the center of the imaging dish in 20 µl of Schneider's medium.Next, the modi ed lm was placed on top of the fat body and the metal ring was gently placed on the lm to trap the tissue underneath.Finally, 200 µl of Schneider's medium was added inside the insert.The samples in Schneider's medium were recorded for 5 min (time interval:5s) to generate a baseline.Then, Schneider's medium was replaced by Schneider's medium containing 100 ng/mL of different synthesized Drosophila neuropeptides (Genscript, the sequences of the peptides are shown in Supplementary Table 2), and recording was done for 20 min (time interval:5s).Time-lapse recording was performed on a Leica DMi 8 equipped with a Leica DFC9000 sCMOS camera and a 1.25x HC PL Fluotar objective (Leica).For higher resolution of the gap junction knockdown experiments, a 5x N Plan objective (Leica) was used.GCaMP5G was excited with a 475 laser.Imaging was performed in a dark room at 18°C.We used MATLAB to remove background noise and highlight the edges of the sample.
Ex vivo imaging of adult fat bodies.For adult ies, as most of the fat bodies adhere to the inside of the abdominal cavity, we dissected the dorsal shell together with the fat bodies.The dorsal shell was then adhered to a live cell imaging dish by Vaseline, such that the fat bodies face upwards, and bathed in 200 uL Schneider's medium.Time-lapse recording was performed on a Leica M205 FCA high-resolution stereo uorescence microscope (Leica) equipped with a Leica DFC7000 GT camera.The remaining steps and parameters are similar to those in the larval experiment.
In vivo imaging of immobilized larvae and adults.For larvae, we attached two coverslips to a microscope slide with double-sided tape to form a thin slit, then used plasticine to plug both sides of the slit so that the gap in the slit is the same width as a 3rd instar larva.Next, a coverslip was added on the top of the chamber to immobilize the larva.For adults, the six legs of an adult female y were cut off, and the wings adhered to a live cell imaging dish with Vaseline so that the abdomen faced upward.For the carbon dioxide and chloroform treatment, the chamber was covered with a transparent petri dish to prevent gas leakage.Time-lapse recording was performed on a Leica M205 FCA high-resolution stereo uorescence microscope (Leica) equipped with a Leica DFC7000 GT camera.
In vivo imaging of fat bodies of free-behaving larvae.We used a 3D printed mold (Wenext) to produce an agarose gel containing different nutrients.The center of the gel has a 11mm*13mm*0.66mmchamber, which can provide a free-behaving arena for more than 10 early 3rd instar larvae.Finally, a coverslip was added on top of the chamber to prevent the larvae from escaping.After being allowed to acclimate for 10 min, larvae were recorded for 20 min (time interval:5s), then all larvae were transferred to an agarose gel chamber containing another type of food for 20-min recording.To ensure that there was no food residue in and on the body of the 3rd instar larvae, larvae were starved for 9 hours before imaging and washed with ddH 2 0 during each transfer process.Time-lapse recording was performed on a Leica M205 FCA high-resolution stereo uorescence microscope (Leica) equipped with a Leica DFC7000 GT camera.
In image processing, we used connected component analysis to approximate each connectome as a larva.Fluorescent signal changes were normalized using the following formula: where F(t) is the uorescence at time t, F 0 is the average baseline (before transfer).
In vivo imaging of APCs of free-behaving larvae.To perform high-resolution neuronal imaging of freebehaving 1st instar larvae, we used an extended-depth-of-eld microscope with two modules: 1.A dark eld imaging module equipped with a 4X NA 0.2 air objective (Nikon, Japan) and a high-speed nearinfrared camera (acA2000-340kmNIR, Basler ace), used to track and record a free-behaving 1st instar larva.2. A uorescence imaging module equipped with a 10X NA 0.3 air objective and a sCMOS camera (Zyla 4.2, Andor Inc., UK), with an imaging surface split in two by Optosplit II (Cairn, UK), which allows simultaneous recording of two uorescent signals (calcium-sensitive GCaMP and calcium-insensitive RFP used as reference).Through extended-depth-of-eld technology, the effective depth of eld was extended by about 5 times, avoiding errors on the Z-axis caused by motion.Before imaging, the 1st instar larvae were starved for 6 hours, and then the larvae were gently picked with a brush into an agarose gel chamber (Φ20 mm*0.15mm) made with a 3D printed mold (Wenext) for 15-min recording.
All image analyses were conducted using ImageJ and MATLAB.The uorescent signal changes were normalized using the following formula: ΔF/F 0 = (F(t) -F 0 )/F 0 , where F(t) is the uorescence at time t, F 0 is the average baseline (before transfer).
AKH secretion assay.To measure AKH retention in PACs, early 3rd instar larvae were picked out from standard food and washed with ddH 2 0, and after feeding for a period of time under different dietary conditions, the brains were dissected in PBS (1.86We used an antifade agent to mount the samples.All images were acquired using a Leica M205 FCA high-resolution stereo uorescence microscope (Leica).
Wave direction analysis.We set the direction of the larvae from head to tail as columns, took the average of each row, and derived a one-dimensional vector for each frame.Next, we correlated these measurements with time to obtain a two-dimensional graph to show the direction of calcium wave movement.
Ca 2+ proportion calculation.We use the baseline uorescence of GCaMP5G as the threshold to calculate the proportion of the area where calcium activation occurs.Each calculation requires at least 20 minutes of time-lapse data.This parameter also represents the probability of a calcium activation event occurring per unit time per unit area.
Wave velocity calculation.We rst de ned two lines parallel to the wavefront.Then, we measured the time required by the wavefront to cross the distance between the two lines to manually calculate wave velocity using ImageJ.
Ca 2+ diffusion calculation.To characterize calcium wave propagation, we arti cially set a parameter "Ca 2+ diffusion area".According to the concept of connected components in image processing, we calculated the area of each connected region where calcium activation occurred in the 20-min time-lapse data and calculated the average value.
TAG assay.5 ies from each group were homogenized with 100 µl of isopropyl alcohol (BBI, A600918-0500), centrifuged at 10,000g for 10 minutes at 4°C, and the supernatant was collected.2 µl of sample solution was mixed with 200 µl of assay reagent (Elabscience, E-BC-K261-M), and incubated at 37°C for 10 minutes.We measured the absorbance at 492 nm in a microplate reader (Thermo Scienti c Multiskan FC).
Heart rate assay.The early 3rd instar larvae were attached to a glass slide with light-curing glue, ensuring that their dorsal sides were facing up.Recordings were captured on a Leica M205 FCA highresolution stereo uorescence microscope (Leica) equipped with a Leica DFC7000 GT camera.The tracheal movements can readily be seen moving with each heartbeat.Kymographs were generated by drawing a single-pixel straight line perpendicular to the trachea in each frame.All image analyses were conducted using ImageJ.
Hemolymph ow assay.To detect the direction of the hemolymph ow, ies anesthetized with carbon dioxide were injected with PBS + 0.1%BSA containing 5 µm diameter uorescent beads (ex/em: 535/610 nm, Hugebio).Before the injection, the beads were blocked in PBS + 10% yeast extract overnight to prevent adhesion to the tissue in vivo.For larvae, we chose to inject from the tail, and for adults, we chose to inject from the abdomen.After injection, the ies were attached to a glass slide with lightcuring glue.The recording was performed on a Leica M205 FCA microscope.All image analyses were conducted using ImageJ and MATLAB.
Statistics.Sample sizes were determined through preliminary experiments and previous studies to achieve the required statistical power.All assays were repeated more than three times.

Results
Periodic ICWs in y larval adipose tissues are induced by a brain-derived factor The y fat body, which is the functional equivalent of the mammalian liver and adipose tissue, plays central roles in metabolic regulation and nutrient sensing 21 .Ca 2+ levels in the y fat body are essential in lipolysis 22,23 , and in vivo Ca 2+ waves in the larval fat body have been mentioned without in-depth study 24 .
Similarly, Ca 2+ waves have been reported in the mammalian liver 25 .Despite these ndings, the functional consequences of what triggers such waves have not been investigated in vivo in the y fat body.To address these questions, we expressed the genetically encoded Ca 2+ indicator GCaMP5G speci cally in the larval fat body, and immobilized larvae within a glass channel for observation (Supplementary Fig. 1A).To our surprise, we noticed global intercellular Calcium waves (ICWs) that emanate in a periodic fashion from the larval head to tail (Fig. 1A-B, Supplementary video 1).To ensure that these waves were not an artefact due to immobilization of the larvae, we examined Ca 2+ activity in free-behaving 3rd instar larvae.Similarly, prominent global ICWs were observed, indicating that these waves were not a consequence of immobilization (Supplementary Fig. 1B-C).We examined whether the global ICWs could be caused by the contractions of skeletal muscles which may mechanically stimulate the fat body.
Paralyzing the larva by feeding with neurotoxin Tetrodotoxin (TTX) completely stopped muscle movements, but the global ICWs persisted, suggesting the ICWs are not caused by muscle contractions (Supplementary Fig. 2).
Previous studies in y imaginal discs suggested that ICWs are triggered autonomously or in response to self-secreted epithelial-derived factors 26,27 .To discern whether the Ca 2+ in the fat body is autonomously regulated or under the peripheral control of other organs, we dissected and cultured fat bodies in Schneider's Drosophila medium.Interestingly, the isolated larval fat bodies displayed no ICWs except for a few cells that sustained damage during dissection and maintained an abnormally high Ca 2+ level (Fig. 1D).These data suggest that fat body ICWs are triggered by signals from other organs.To identify the speci c organ responsible for this regulation, we co-cultured fat bodies with different larval organs including muscle/cuticle, intestine, and brain (Fig. 1C).ICWs were only triggered when fat bodies were co-cultured with brains (Fig. 1.C-E, Supplementary video 2), suggesting that these waves are initiated by a brain-derived factor.
To identify the putative factor, we generated a brain-conditioned medium by incubating Schneider's Drosophila medium together with dissected larval brains (10 dissected 3rd instar larval brains in 150 µl medium), and applied the conditioned medium to isolated fat bodies.The brain-conditioned medium induced robust ICWs (Fig. 1F-G).To determine the nature of this factor, we ltered the brain-conditioned medium through a Pierce 10K MWCO concentrator that removed molecules larger than 10 kDa.The ltrate evoked robust fat body ICWs, suggesting that the factor has a relatively small molecular weight (Fig. 1G).We then treated the conditioned medium with proteinase K (0.1 mg/mL), DNAse I (1 U/mL) or RNAse A (0.1 µg/mL), followed by removal of added enzymes with the 10 kDa lter.Among these treatments, only proteinase K signi cantly reduced Ca 2+ activities, suggesting that the factor is a small peptide (Fig. 1H).

AKH released from APCs is responsible for inducing ICWs in the fat body
To identify the relevant peptide(s), we screened 32 major y peptides on isolated fat bodies (10 ng/mL each).Among them, only AKH evoked signi cant Ca 2+ waves in the fat body (Fig. 2A).Washing away the added AKH resulted in an immediate abrogation of ICWs, indicating that the Ca 2+ waves require a sustained supply of AKH (Fig. 2B).The observation that the fat body responds within several seconds to the addition or removal of AKH, and the fact that there was no activity reduction in response to prolonged activation (up to 4 hours), suggests that the AKH receptor in the fat body has fast binding and dissociation kinetics with no signal adaptation.Further, we found that the fat body is sensitive to AKH at doses as low as 0.1 ng/mL, which is within the physical range of neuropeptides.In addition, both the amplitude and frequency of Ca 2+ oscillations generally increased with AKH concentration (Fig. 2C, Supplementary Fig. 3A-D).Finally, knocking down the only y-encoded AKH receptor, AkhR, also reduced the Ca 2+ oscillation frequency, suggesting that the frequency is regulated by the concentration of both ligand and receptor (Supplementary Fig. 3E-F).
To test whether AKH is indeed the fat-stimulating factor in the brain-conditioned medium, we added an anti-AKH antibody in the medium to sequester free AKH.Strikingly, the ICWs were blocked immediately after adding the anti-AKH antibody (Fig. 2D-E).Next, as AKH is secreted from the APCs of the ring gland, which is associated with the larval brain, we knocked down Akh in the APCs (Akh-Gal4 > UAS-Akh-RNAi) or inhibited AKH release by expressing the neuronal silencing Kir2.1 potassium channel (Akh-Gal4 > UAS-Kir2.1).This treatment suppressed the release of AKH resulting in a signi cant decrease in fat body Ca 2+ waves in the co-culture experiment (Fig. 2F-G).Finally, fat bodies from AkhR mutant larvae no longer responded to the applied AKH peptide (Fig. 2H-I, Supplementary Video 3) and the Ca 2+ waves were completely blocked in AkhR mutant larvae in vivo (Fig. 2J-K, Supplementary Video 4).
AkhR is a GPCR receptor that has been found to trigger both cAMP and Ca 2 + 21 .AKH is known to regulate Ca 2+ increase by activating phospholipase C (PLC) and subsequent generation of IP3 and DAG 28 .To block the AKH-induced Ca 2+ increase, we knocked down Gαq, which is responsible for the GPCR-dependent PLC activation.Knocking down Gαq in fat bodies blocked the AKH-triggered ICWs, suggesting that Gαq is an AKH-downstream effector for Ca 2+ activity (Fig. 2H-I).Consistent with this observation, Gαq overexpression generated active Ca 2+ waves both in isolated fat bodies without the addition of AKH and in free-behaving larvae on a 5% m/v sucrose diet (Supplementary Fig. 4).In addition, knocking down SERCA, an important ER-located Ca 2+ pump, led to a sustained elevation of Ca 2+ in the fat body (Supplementary Fig. 4).Collectively, these ndings show that Ca 2+ activity in the larval fat body is triggered by AKH secreted from the APCs, and relies on Gαq-mediated e ux of Ca 2+ from the ER downstream of the AKH receptor.

Global and local ICWs form through distinct mechanisms
The AKH-induced ICWs observed in the y fat body resemble the glucagon-triggered Ca 2+ waves observed in the mammalian liver 1 .However, the biological signi cance of the ICW propagation in both mammalian liver and y fat body has not been explored.Moreover, AKH triggers random local Ca 2+ waves in the cultured fat body, but induces directional global ICWs propagating from the larval head to tail.The mechanism underlying the different behaviors of ICWs in vivo and ex vivo is unknown (Fig. 3A).
To investigate the biological mechanism of ICWs propagation, we rst genetically blocked the intercellular signal transduction in the fat body.Based on previous studies showing that the ICWs are mediated either by gap junctions or extracellular ATP 18,29 , we added ATP to the cultured fat body and found that it did not trigger ICWs (Supplementary Fig. 5), ruling out ATP as a trigger for ICWs in this system.We then screened all major gap junction proteins by RNAi and found that knockdown of Inx2 or Inx3 signi cantly reduced the intercellular Ca 2+ waves in ex vivo fat bodies (Fig. 3B-C, Supplementary Fig. 6).Our observation is consistent with previous studies suggesting that Inx2 is predominately required for gap junction function in the y epithelium 26,29,30 and that Inx3 probably functions together with Inx2 to form a heterohexamer 31 .Both RNAi lines for Inx2 and Inx3 have been validated in multiple studies 30,32-34 and showed similar phenotypes in our study.Since Inx2 knockdown exhibited the strongest phenotype, we used Inx2-RNAi to disrupt gap junctions in all subsequent experiments.
In vitro experiments demonstrated that Inx2 knockdown effectively blocked the local propagation of Ca 2+ activities (Fig. 3B-C, Supplementary Video 5).However, in vivo global ICWs were surprisingly unaffected by Inx2 knockdown.Both propagation speed and oscillation period of global ICWs in the Inx2-RNAi fat body are similar to the control, while random local ICWs were blocked by the Inx2-RNAi (Fig. 3D-E, Supplementary Fig. 7A).In fact, Ca 2+ oscillations in Inx2-RNAi ies exhibited an increase in amplitude and a decrease in frequency (Supplementary Fig. 7B-D).The overall effect of Inx2 knockdown led to an unexpected elevation of average Ca 2+ activity in the fat body of free-behaving larvae (Fig. 3F-G).To rule out the possibility of non-speci c effects of Inx2-RNAi, we tested the effects of the gap junction inhibitor carbenoxolone.Consistent with Inx2-RNAi, carbenoxolone treatment showed a similar increase of Ca 2+ amplitude and decrease of frequency in the cultured fat body (Supplementary Fig. 7E-G).
Despite these changes, our data clearly demonstrate that the spreading of the global ICWs in vivo does not require intercellular signal diffusion.Meanwhile, both the directional global and the random local Ca 2+ waves were completely abrogated in AkhR mutants, suggesting that these Ca 2+ waves in the fat body are triggered by extracellular AKH (Fig. 3D, Supplementary Video 6).Thus, we propose that local Ca 2+ waves are generated by diffusion of Ca 2+ from cells stochastically activated by AKH, but that the global Ca 2+ waves are generated in response to the pulsed secretion and diffusion of a high level of extracellular AKH from the APCs near the larval head region (Fig. 3A).Supporting this hypothesis, the directional propagation of the ICWs were disrupted into a random pattern after overexpression of AKH in the fat body (Fig. 3D, Supplementary Video 6).In addition, we noticed that the in vivo global ICWs travel approximately 10 times faster than the local Ca 2+ waves both in vivo and ex vivo (Fig. 3E).These data further support the notion that global ICWs are triggered by fast traveling extracellular AKH signals rather than by slow diffusion of Ca 2+ spikes facilitated by gap junctions.
As most previous studies showed that ICWs in y imaginal discs are signi cantly reduced after gap junction knockdown 24,26 , we wondered if the gap junction independent ICWs are stage-speci c.Thus, we analyzed the Ca 2+ activity in the adult fat body and discovered that the cultured adult fat body exhibited a similar gap junction-dependent local ICWs in response to the administered AKH like the larval fat body (Fig. 3H-I, Supplementary Video 7).However, contrary to the larval fat body, no global ICWs were observed in the adult fat body in vivo, and the local ICWs in the adult fat body were signi cantly reduced in adult ies with Inx2 knockdown (Fig. 3J-K, Supplementary Video 8).These data indicate that the ICWs in the adult fat body behave as the previously reported Ca 2+ waves in the imaginal disc epithelium, suggesting that global ICWs are unique to the larval stage.
As the increase of AKH secretion and cytosolic Ca 2+ levels have been linked with TAG lipolysis in adipose tissues [21][22][23]35 , we assessed the effect of gap junction disruption on TAG catabolism under both normal feeding and starvation conditions. TA levels in Inx2 knockdown larvae remained unaffected under standard feeding conditions, as AKH secretion is not expected to be induced (Fig. 3L).However, under starvation conditions, TAG levels in gap junction knockdown larvae were signi cantly lower than in control larvae (Fig. 3L).This is consistent with our observation that gap junction blockage increases AKH-induced Ca 2+ activities in larvae.In contrast, adult ies with disrupted gap junctions in the fat bodies displayed a signi cant increase in TAG accumulation under normal diet and starvation conditions (Fig. 3M), consistent with the reduction of Ca 2+ activities in the adult fat body.Thus, the strikingly different responses of larval and adult fat bodies to gap junction disruption likely re ect the presence of global ICWs in larvae.
Global ICWs are triggered by the periodic release of AKH in the circulating hemolymph To test whether global ICWs are triggered by the fast transport of extracellular AKH released from the APCs, we used chloroform to temporarily stop the heartbeats of 3rd instar larvae (Fig. 4A-B).Stopping the heartbeat completely prevented the propagation of global ICWs, and only the head region of the larvae showed periodic increase of Ca 2+ activities (Fig. 4C).Notably, the global Ca 2+ waves also showed a much longer period (~ 400 sec) than the intrinsic Ca 2+ oscillation period in the fat body (~ 200 sec), supporting that the period of global waves is controlled by the pulsatile release of AKH from the APC cells rather than an intrinsic property of fat body.
We also noticed that the propagation speed of the global ICWs is around 40 µm/sec, which is more than 10 times faster than the intrinsic speed of the triggering wave of about 2.5µm/s (i.e. the speed of the local waves, which is generated presumably by gap-junction mediated intercellular signaling).This speed also surpasses the free diffusion velocity of a 1 kDa molecule (equivalent to the size of AKH) in water by approximately two orders of magnitude 36 (Fig. 3E), which further supports that AKH released from APCs are transported by the circulating hemolymph.However, the ow characteristics of the y hemolymph have never been measured before.Thus, we injected red uorescent polystyrene beads with a 5 µm diameter into the 3rd instar larvae and estimated the ow speed of the hemolymph by tracing the beads (Fig. 4D).After tracing multiple beads, we found that the anterograde (head to tail) ow speed of hemolymph is about 250 µm/sec and that the retrograde (tail to head) ow speed in the heart tube is nearly 10,000 µm/sec (Fig. 4E).This high-speed hemolymph circulation is fast enough to facilitate the transport of secreted AKH to the head region.Interestingly, the hemolymph ow in adult ies was primarily retrograde outside the adult heart tube, with an occasional reverse ow direction as previously reported 37 (Supplementary Fig. 8).As the APCs are located in the anterior region of the adult thorax 21 , we speculate that the change in circulation direction of the adult hemolymph may make the transport of secreted AKH less e cient and thus be responsible for the absence of global ICWs.
Moreover, the circulating AKH model also implies that the release of AKH must have a short half-life compared with the 300 sec period of the global ICW, otherwise, AKH will accumulate in the hemolymph and trigger continuously random ICWs as observed in the ex vivo tissue culture.One way to test this model is to acutely block the release of AKH in APCs and monitor the decrease of Ca 2+ in the fat body as a read-out of circulating AKH in vivo.Unfortunately, the available neuron-silencing optogenetic tools are incompatible with GCaMP imaging.As an alternative, we found that anesthetizing the larvae with carbon dioxide (CO 2 ) rapidly suppressed the Ca 2+ activity of the APC neurons, while the whole brain activity was largely unaffected (Fig. 4F-H, Supplementary Fig. 9).As our model predicted, the ICWs in the larval fat body rapidly returned to baseline within ~ 70 seconds after CO 2 administration, implying that the in vivo functional half-life of AKH is ~ 35 seconds (Fig. 4I-L).Meanwhile, in larvae with AKH overexpression in the fat body, the signi cant level of ICWs remained after CO 2 treatment, suggesting that CO 2 does not block the fat body response to AKH (Fig. 4I-L).

Computational modeling of the global and local ICWs
With the measured dynamic parameters of ICWs and AKH, we applied a receptor-operator calcium model to gain insights into the distinct dynamics of the ICWs in both larval and adult ies.The model incorporates three key elements: intracellular Ca 2+ dynamics, intercellular signaling via gap junctions, and tissue-level AKH transport.The cytoplasmic Ca 2+ level is governed by a receptor-operator calcium channel model 24,38 .When AkhR binds to AKH, it activates a downstream signaling pathway involving inositol trisphosphate (IP3), triggering a rapid Ca 2+ e ux from the endoplasmic reticulum (ER) to cytoplasm via a positive feedback loop.The SERCA pump on the ER membrane acts as a slow negative feedback loop, restoring Ca 2+ balance by pumping it back into the ER from the cytoplasm.Intercellular Ca 2+ signaling through gap junctions is described by diffusion proportional to the cytoplasmic Ca 2+ concentration difference between neighboring connected cells 24 .Either AKH binding to AkhR or the Ca 2+ ow through gap junctions can trigger the intracellular fast-activation-slow-inhibition Ca 2+ signaling cycles.In larvae, AKH transport and diffusion are described by a reaction-diffusion-advection equation.This equation accounts for the pulsatile AKH secretion from APCs and its uniform degradation.The advection re ects the transport of AKH via the circulating hemolymph.In adult ies, the global AKH pulses are absent, and the model only considers AKH diffusion with uctuations.This distinction accounts for potential differences in extracellular AKH concentration dynamics between larvae and adult ies.Model details can be found in Supplementary Methods 1.The parameters used in the model are listed in Supplementary Table 3.
The simulations successfully replicated the observed differences in wave speed between larvae and adult ies, as well as the impact of gap junction knockdown on Ca 2+ signaling (Fig. 5A-D).Both computational simulation and experimental data support that directional transport of AKH through the lymphatic circulation is responsible for the global ICWs with wave speed around 40 µm/sec, while the trigger wave mechanism through gap junctions is responsible for the local ICWs with wave speed around 3-4 µm/sec in WT and Inx2-RNAi ies 39 (Fig. 5A-D, Supplementary Video 11,12).Meanwhile, the modeling also showed similar changes in average Ca 2+ intensity and peak width in individual fat cells after gap junction knockdown in larval fat bodies (Fig. 5E-F), supporting that the modeling can precisely recapitulate the dynamics of Ca 2+ activities.
Interestingly, our modeling results implied that the excitable Ca 2+ signaling cycles are sensitive to both external and internal noises, such as uctuations in AKH concentration or intracellular signaling molecules.These uctuations have the potential to cause spontaneous Ca 2+ signaling randomly, leading to asynchronous ring between individual cells.However, gap junctions help coordinate these excitation processes, resulting in spatially synchronized oscillations.In our simulations, we noticed that the Ca 2+ wavefront is more synchronized in the WT larvae compared to larvae with the Inx2 knockdown, despite similar wave speeds.To quantify the spatial coordination of Ca 2+ activities, we calculated the normalized variance of Ca 2+ intensity in the wavefront region (Fig. 5G-H).As the global ICWs propagate from head to tail, the normalized variance remains low for WT larvae but increases sharply in larvae with gap junction knockdown.Notably, the experimental data also exhibited a similar increase of variance in the tail region of the Inx2-RNAi larvae but not in the WT larvae (Fig. 5G-H).This nding reinforces the relevance of our modeling results, although the biological signi cance of the variance increase requires further exploration.

Regulation of AKH release from APCs by amino acids
Our ndings demonstrate that the fat body exhibits a speci c and rapid response to extracellular AKH.This property provides a clear and direct readout for monitoring real-time secretion of AKH in freebehaving larvae under different nutrient conditions.As most previous live-imaging setups used immobilized or even anesthetized larvae, which may cause undesirable distress and artefacts, we decided to study metabolic signaling in free-behaving animals.In addition, previous studies have suggested that AKH, a functional homolog of mammalian glucagon, is released upon starvation 3,5,35,40 .Early 3rd instar larvae were rst starved for 9 hours to cleanse their digestive systems, and then fed on 2% sucrose for 20 min, allowing the fat body Ca 2+ activities and AKH secretion to return to a basal level (Fig. 6A).Subsequently, these conditioned larvae were transferred onto 2% agarose plate with 2% sucrose, 2% sucrose plus 10% Tryptone (a protein-rich diet equivalent to approximately 5% protein), or 2% agarose with no nutrient (starvation diet).Larvae fed on a starvation diet displayed a signi cant increase in fat body Ca 2+ waves after ~ 10 minutes of feeding, indicating a swift response to starvation, which agrees with the established function of AKH as a starvation-induced hormone.Importantly, consumption of the protein-rich diet also signi cantly increased Ca 2+ activities in the fat body within just 10-15 minutes (Fig. 6B-C, Supplementary Video 13), suggesting that AKH secretion is also triggered by amino acids.To con rm whether the protein-induced fat body Ca 2+ waves are indeed AKH-dependent, we compared WT larvae with larvae carrying a AkhR mutation or fat-body speci c AkhR knockdown using Lpp-Gal4.In both cases, we observed a signi cant reduction of the Ca 2+ waves under both starvation and protein-feeding conditions, supporting the idea that both processes depend on AKH signaling (Fig. 6D-E).
The in vivo Ca 2+ activity of the fat body suggests that AKH release from APCs is triggered by amino acids.To test this directly, we examined whether APC activity is regulated by a protein diet in freebehaving larvae.We used Akh-Gal4 > GCaMP5G-T2A-mRuby3 to visualize and quantify Ca 2+ activity in the APCs.However, as the thick cuticle of the 3rd instar larvae caused a strong blurring of the signal from the APCs when they moved, we decided to use 1st instar larvae, which are smaller and more transparent.We used an Extended-Depth-of-Field (EDoF) microscope, which turns slow 3D imaging into a quick 2D acquisition (Supplementary Fig. 10A-D).In addition, because the Ca 2+ signals from the APCs are much dimmer than those of the fat body, we used a starvation diet as the initial condition to achieve a reliable visualization of Ca 2+ activity.1st instar larvae were starved for 6 hours on 2% agarose plate, then transferred to new agarose plates containing 5% sucrose (sugar diet), 10% Tryptone (protein-rich diet), or 2% agarose only (starvation) (Fig. 7A).As expected, Ca 2+ levels in the APCs signi cantly decreased after being fed on 5% sucrose (Fig. 7B-G, Supplementary Video 6).Interestingly, Ca 2+ levels in the insulin-producing cells (IPCs) were reduced when we performed simultaneous dual labeling of both the APCs and IPCs (Supplementary Fig. 10E-G).This real-time observation of the counter activities of the IPCs and APCs strongly supports the reliability of this live-imaging system.Next, we tested the effect of a 10% tryptone diet and found that protein consumption triggers a Ca 2+ increase in the APCs after ~ 10 minutes of feeding, which is consistent with the quick Ca 2+ response observed in the fat body (Fig. 7B-G).We used 5% sucrose instead of 2% because 5% sucrose triggers a quicker and stronger AKH suppression.However, for the experiment in 3rd instar larvae, the secretion of AKH secretion is too severely suppressed with 5% sucrose, making the Tryptone diet less effective.
Studies in mammals have found that only some amino acids trigger glucagon release, with branchchained amino acids failing to induce such secretion 41 .Thus, we tested which particular amino acids may activate AKH-mediated Ca 2+ waves in ex-vivo fat body tissues using our conditioned-medium system.As the complete Schneider's Drosophila medium contains all amino acids and triggers the release of AKH from dissected larval brains, we prepared a basal Drosophila HL6 buffer devoid of any amino acids (referred to as HL6(AA-) buffer).Subsequently, each amino acid (5 mM) was individually added to the HL6(AA-) buffer to assess its capacity to induce AKH release (Fig. 8A).Brain-conditioned HL6(AA-) buffer does not activate the cultured fat body; however, the addition of most small polar amino acids triggered AKH secretion and subsequent elevation of fat body Ca 2+ (Fig. 8B-C).Notably, the large branch-chained amino acids leucine (Leu) and isoleucine (Ile) failed to trigger AKH release, akin to their effects on the mammalian glucagon system.We further monitored Ca 2+ activities in APCs in isolated larval brains.APCs responded to threonine (Thr) within two minutes and reached an activation plateau at around six minutes, consistent with the response speed observed in vivo.In contrast, APCs showed little response to Leu (Fig. 8E-G).Finally, the release of AKH from APCs was con rmed by AKH staining of APCs in 3rd instar larvae.Starvation, known to trigger AKH release, served as a positive control (Supplementary Fig. 11).Next, we tested 36 hours of feeding on 5% sucrose, 5% sucrose plus 40 mM methionine (Met), or 5% sucrose plus 40 mM Leu (Fig. 8H-J).Met feeding signi cantly reduced the AKH signal in the APCs compared to the sugar control and Leu, suggesting that Met triggers the release of AKH in vivo.Previous studies have shown that increased AKH levels lead to lipolysis in fat body tissues, especially in adult ies 35 .Hence, we examined whether amino acid feeding could reduce triacylglycerol (TAG) content.Indeed, 40 mM Met feeding reduced the TAG content in adult ies, and this reduction was entirely abrogated in AkhR mutant animals (Fig. 8K).Altogether, our results show that speci c dietary amino acids, i.e., Methionine and Threonine, are sensed by APCs to trigger AKH release, which in turn activates ICWs in the fat body.

Discussion
In this study, we demonstrate that AKH secreted from the APCs stimulates ICWs in the Drosophila fat body to promote lipid metabolism.Furthermore, we discovered that the global and local ICWs in the fat body are generated through different molecular mechanisms: global Ca 2+ waves are generated by extracellular circulation of AKH secreted from APCs in a pulsatile manner, whereas local Ca 2+ waves are formed through intercellular signal propagation mediated by gap junctions.Finally, we found that speci c dietary amino acids activate the APCs, leading to increased intracellular Ca 2+ and subsequent AKH secretion.
ICWs in the y fat body are controlled by AKH-AkhR signaling Previous studies have established Ca 2+ as a key regulator of lipolysis in the y fat body 22,23 , and AKH has been found to trigger Ca 2+ increase in the fat body under ex vivo conditions 28 .Our current study demonstrates that the primary in vivo driver of Ca 2+ activity in the y fat body is the AKH-AkhR pathway and its downstream effector Gαq.Similar to mammalian glucagon, AKH is a central hormone that has been found to integrate diverse biological processes, including mobilization of lipid storage, stimulation of locomotion, oxidative stress protection, and immune response 21 .Therefore, it is important to study the release of AKH at high temporal resolution.Our ndings suggest that Ca 2+ signaling within the y larval fat body serves as a reliable real-time indicator for AKH signaling in free-behaving animals.
AKH-mediated ICWs in the larvae and adult fat body are driven by distinct Ca 2+ has been found to generate a variety of inter and intra-cellular activities, including ashes, sparkles, oscillations and ICWs, which have been implicated in diverse biological processes 29 .In this work, we focus on ICWs in the Drosophila fat tissue.We demonstrate that these fat ICWs are speci cally under the control of the AKH hormone produced from the brain-associated neuroendocrine cells, the APCs.
Canonically, ICWs are considered to spread through tissues via gap-junctions 17,26,29,42 .However, we delineate a new gap-junction independent mechanism in which AKH release from the brain actively diffuses through the circulating lymph of the animal, which in turn results in the organ-level ICWs.These gap junction-independent Ca 2+ waves uncovered in our study present an intriguing model illustrating how a hormone can function as an extracellular orchestrator, creating a collectively moving pattern across a large epithelial tissue.Our ndings reveal that these gap-junction-independent ICWs in the larval fat body bypass the need for intercellular Ca 2+ propagation and are self-su cient in maintaining tissue-level Ca 2+ activities.These tissue-level ICWs also suggest that AKH is secreted in a strong pulsatile manner in larvae, a phenomenon similar to the pulsatile release of mammalian glucagon and insulin, which are disrupted in patients with type-2 diabetes, potentially contributing to hyperglucagonemia 43,44 .However, the biological signi cance of this pulsatile hormone release compared to continuous release is not clear.Our study suggests that in the y larva, a pulsatile secretion of AKH creates a strong increase of hormone "shock" that collectively activates fat body cells, which renders the intercellular Ca 2+ spreading between neighboring cells less important.
Intriguingly, the adult fat body depends on gap junctions to uphold a functional Ca 2+ level under starvation or amino acid feeding conditions.Meanwhile, no organ-level global Ca 2+ waves were observed -the Ca 2+ waves detected in vivo in the adult fat body appear completely random, suggesting that the circulating AKH in adult hemolymph may not be strong enough to collectively activate the fat body cells.Thus, in the absence of a strong extracellular AKH pulse, cells activated by AKH become sparsely distributed as in the adult fat body, requiring intercellular spreading of Ca 2+ .The biological signi cance behind these differences between larval and adult tissue remains an open question.Understanding these differences might help us understand the difference in response to AKH-AkhR signaling in larva vs. adult y, i.e., AKH-AkhR signaling activation of lipolysis in the adult y fat body but not in the larval fat body.
Through mathematical modeling of these waves in the fat body, we also noticed that inclusion of random uctuation is essential to recapitulate the Ca 2+ wave properties in fat bodies with Inx2-RNAi.The increase of Ca 2+ activities after gap junction knockdown can be observed only when the random uctuation of the Ca 2+ concentration is considered in our model.Within the parameter range that ensures oscillating/wave behavior, an increase in Ca 2+ uctuation intensity will amplify the magnitude difference between WT and Inx2-RNAi in the simulation of the larval fat body.The precise reason why random uctuations serve as a determining factor for wave properties requires further investigation, yet our results suggest that the random effect should not be overlooked in the research of biochemical waves.

Dietary amino acid-mediated activation of APCs
The in vitro kinetics of Ca 2+ activities also reveal intriguing aspects of amino and AKH-AkhR signaling.We found that cultured APCs do not respond immediately to the applied amino acids but instead exhibit a gradual increase in activity over a 5-10 minutes period.This suggests that amino acids may not function through a rapid response mechanism such as ligand-gated channels, but via a comparatively slower metabolic process, possibly involving an increase in cytosolic ATP following amino acid breakdown.However, it remains to be elucidated whether the APCs sense amino acids directly or indirectly through other neurons or a secondary metabolite within the brain.Furthermore, the fast and continuous response of the fat body to extracellular AKH without adaptation implies that AkhR probably does not undergo activation-induced inactivation.This immediate and persistent response to extracellular AKH could provide ies with an advantage in promptly adapting their metabolic state to environmental changes.Meanwhile, the in vivo half-life of AKH is remarkably short due to an unknown mechanism, probably involving a speci c secreted proteinase.Identi cation of this unknown serum factor may provide key insights into the dynamic regulation of hormonal signaling.
By tracing Ca 2+ activities in the y fat body and AKH-producing cells in response to different nutrients, we found that AKH secretion is regulated by certain dietary amino acids in the y hemolymph, which subsequently increases the mobilization of fat body lipids through AkhR-Gαq signaling.Interestingly, previous studies have identi ed other amino acids sensing mechanisms in different organs: amino acid triggers the release of GBP1/2 and Stunted from the fat body to stimulate the insulin-like peptides secretion and promote larval growth 45,46 , essential amino acids promote the release of CNMa from intestine cells to regulate feeding behavior 47 , FMRFa secretion from brain neurons is triggered by amino acid consumption to mobilize lipid stores 48 , and insulin-producing cells (IPCs) sense amino acids to increase insulin-like peptides production 20 .It will be interesting to explore how and where these amino acid-dependent signals interact or integrate to achieve metabolic homeostasis.
Although we have observed that amino acids regulate the secretion of AKH, the precise biological signi cance of this phenomenon is still not fully understood.Our study primarily focused on the regulation of AKH secretion and its effect on fat body Ca 2+ increase and subsequent TAG lipolysis.However, it is conceivable that the activation of AkhR has a more extensive role than facilitating neutral lipid mobilization.Recently, AKH has been found to activate extracellular signal-regulated kinase (ERK), which in turn increases amino acid catabolism and gluconeogenesis in the y fat body 49 .Together with our observations, it seems plausible that amino acid-induced AKH secretion serves as a mechanism for ies to process excessive amino acids ingested from the diet.AKH is believed to be a functional homolog of mammalian glucagon, which is speci cally produced under starved or low-energy conditions to promote lipolysis in peripheral organs.However, we found that AKH is not only stimulated under energy challenges conditions but also by a high protein diet.Similarly, a high-protein diet has also been reported to trigger mammalian glucagon.A more comprehensive study of the downstream signaling pathway of AKH is needed to fully understand the consequence of this AKH-mediated amino acid sensing axis.In addition, our and previous work suggest that different dietary amino acids can trigger different neuronal centers in the y brain, i.e., leucine and isoleucine speci cally trigger insulin-producing cells (IPCs) in the larval brain 20 , whereas we found that these two amino acids do not trigger APCs to release AKH.Further research is required to decipher how activation of different neuronal centers may take place following a high protein diet consisting of different groups of amino acids.
Lastly, the increase of amino acid uptake and usage by AKH signaling in the fat body resembles the recently discovered mammalian Liver-α-Cell axis, whereby an increase in glucagon levels upregulates the expression of speci c amino acid transporters such as Slc38a4 and Slc38a5 in the liver, thereby enhancing amino acid uptake and promoting gluconeogenesis as well as urea production 50 .It will be interesting to investigate whether certain amino acid transporters are similarly upregulated by AKH in the y fat body.Moreover, elevated amino acid levels in the bloodstream not only stimulate glucagon secretion but also contribute to α-cell proliferation, leading to pancreatic α-cell hyperplasia in mice, creating a lasting endocrine feedback 50 .Whether the function of APCs is modulated by amino acid consumption in Drosophila is still unknown.Experiments examining whether high protein intake induces lasting effects on APCs could provide evidence for the conservation of the Liver-α-cell axis across species.

Figure 2 Brain
Figure 2

Figure 3 Global
Figure 3

Figure 4 Global
Figure 4

Figure 8 AKH
Figure 8 Quantitative and statistical parameters, including statistical methods, error bars, n numbers, and p-values, are indicated in each gure.Error bars shown in all results are from biological replicates.Differences were assessed using a two-tailed unpaired Student's t-test, unless stated otherwise in the gure legends.P < 0.05 was considered statistically signi cant.Signi cance was noted as *p < 0.05, **p < 0.01, ***p < 0.001.Image processing and quanti cation were performed in ImageJ and MATLAB.Plotting of graphs and statistical analyses were conducted with GraphPad Prism 8.4.2.