A STIM dependent dopamine-neuropeptide axis maintains the larval drive to feed and grow in Drosophila

Appropriate nutritional intake is essential for organismal survival. In holometabolous insects such as Drosophila melanogaster, the quality and quantity of food ingested as larvae determines adult size and fecundity. Here we have identified a subset of dopaminergic neurons (THD’) that maintain the larval motivation to feed. Dopamine release from these neurons requires the ER Ca2+ sensor STIM. Larvae with loss of STIM stop feeding and growing, whereas expression of STIM in THD’ neurons rescues feeding, growth and viability of STIM null mutants to a significant extent. Moreover STIM is essential for maintaining excitability and release of dopamine from THD’ neurons. Optogenetic stimulation of THD’ neurons activated neuropeptidergic cells, including median neuro secretory cells that secrete insulin-like peptides. Loss of STIM in THD’ cells alters the developmental profile of specific insulin-like peptides including ilp3. Loss of ilp3 partially rescues STIM null mutants and inappropriate expression of ilp3 in larvae affects development and growth. In summary we have identified a novel STIM-dependent function of dopamine neurons that modulates developmental changes in larval feeding behaviour and growth.


Author summary
The ability to feed appropriately when hungry is an essential feature for organismal survival and is under complex neuronal control. An array of neurotransmitters and neuropeptides integrate external and internal signalling cues to initiate, maintain and terminate feeding. In adult vertebrates and invertebrates dopamine serves as a reward cue for motor actions, including feeding. Larvae of holometabolous insects, including Drosophila melanogaster, feed and grow constantly followed by gradual cessation of feeding, once sufficient growth is achieved for transition to the next stages of development. Here we identified a subset of larval dopaminergic neurons in Drosophila melanogaster, activity in which maintains continuous feeding in larvae. By analysis of a null mutant we show that these neurons require the Stromal Interaction Molecule (STIM) an ER Ca 2+ sensor, to maintain excitability. In turn they modulate activity of certain neuropeptidergic cells. Among these are the median neurosecretory cells (MNSc) that synthesize and secrete

Introduction
Animal growth occurs primarily during the juvenile stage of development. In holometabolous insects, including Drosophila, larval development is considered equivalent to the juvenile stage [1,2]. Steady and appropriate nutritional intake in larvae, is essential for growth and development, and ultimately determines both survival and fecundity of the animals. In Drosophila, the feeding rate increases in the second instar larval stage and larvae continue to feed voraciously till the wandering stage of third instar larvae [3]. Increase in feeding rate is accompanied by acceleration of cell division and cell growth [4].
Feeding behaviour and its modulation in Drosophila larvae has been studied primarily in the third instar larval stage, where it is regulated by multiple neurotransmitters and neuropeptides that respond to both external cues and the internal metabolic state [5]. Neuropeptide F (NPF; a human NPY homolog) serves as a motivational signal for foraging in larvae in response to appetitive odours. The activity of NPF neurons appears dependent on inputs from two pairs of central dopaminergic neurons that receive tertiary olfactory inputs [6,7]. Mutants for the short neuropeptide F (sNPF), encoded by an independent gene from NPF, affect body size by regulating food intake in larvae [8]. When food is restricted octopaminergic circuits regulate feeding independent of NPF signalling in 3 rd instar larvae [9]. Neurons that secrete the Hugin neuropeptide respond to averse gustatory signals and their activation suppresses larval feeding [10]. Where essential amino acids are imbalanced a subset of dopaminergic neurons are required for rejection of food by larvae [11]. In addition serotonergic neurons from the brain project to the gut where they potentially regulate feeding-related muscle movements [12]. Insulin like peptides (ilps), are secreted by the medial neurosecretory cells (MNSc) that access the internal metabolic state and release ilps into neurohaemal sites for circulation. In adults ilps terminate feeding based on the energy state of the organism [13]. Analysis of the recently concluded larval connectome demonstrates that MNSc receive both direct and indirect inputs from the enteric nervous system found in the larval gut [14]. The larval MNSc also receive synaptic inputs from central neurons that release the Hugin neuropeptide [10]. Ilps released through neurohaemal sites from the MNSc circulate through the body and regulate energy metabolism [15], synthesis and release of the steroid hormone ecdysone from the prothoracic gland which drives larval instar progression [16,17].
A key difference between larval and adult feeding behaviour is that adult Drosophila feed sporadically, driven by hunger and satiety signals [18] whereas Drosophila larvae accelerate feeding as second instar larvae and feed continuously till the wandering stage of third instar larvae to optimise growth. They stop feeding as wandering larvae for a few hours prior to pupariation [19,20]. Despite studies identifying several neurotransmitters and neuropeptides in feeding regulation and their cognate neurons as part of the feeding connectome in larvae [5] mechanisms that initiate and maintain persistent feeding in early larval stages are not fully understood. In this study, whilst characterizing the cellular and molecular phenotypes of null mutants for the ER-Calcium sensor protein STIM (Stromal Interaction Molecule) [21] we identified a novel dopaminergic-neuropeptide connection in the absence of which early larvae feed poorly and grow slowly. Growth deficits and lethality in STIM mutants appeared to have a focus in dopaminergic cells [22]. However, the cellular basis of STIM function and systemic phenotypes arising from loss of STIM in dopaminergic neurons remained to be understood. Here, we show that STIM function is required in a subset of central dopaminergic neurons for their excitability and dopamine release. These dopaminergic neurons impact larval growth by providing the motivation for persistent larval feeding and modulating neuropeptidergic cells, including the MNSc, to regulate levels of insulin-like peptides.

Reduced food intake and growth deficits in STIM mutant larvae
STIM KO larvae appear normal after hatching but their transition from first to second instar stages is slower than wild-type animals (S1A and S1B Fig) [22] and as second instars they die gradually between 86h to 326h after egg laying (AEL; S1B Fig). To identify the precise time window when STIM KO larvae become sickly they were observed over 6h time intervals from 36h to 90h AEL. Whereas, wild type (Canton-S or CS) larvae transition from 1 st to 2 nd instar between 42-54h AEL, the same transition in STIM KO larvae occurs between 60-72h AEL, indicating a delay of 18h (Fig 1A and 1B). The delay is followed by an inability to transition to 3 rd instar (Fig 1C). STIM KO larvae also exhibit retarded growth. At 72h they appear similar to CS larvae of 60h (Fig 1D). After 72h however, there is a complete cessation of growth in STIM KO larvae (Fig 1D and 1E), followed by gradual loss of viability after 80-86h (S1B Fig). From these results, it became evident that cessation of growth precedes loss of viability in STIM KO larvae.
The momentum of larval growth is maintained primarily by cell growth in the endoreplication tissues [23,24]. In a few organs like the brain and the imaginal discs growth is accompanied by constant cell division. Normally, at the end of embryonic development, mitotic cells such as a majority of neuroblasts (NBs) and imaginal disc cells enter a quiescent state [4,25,26]. Postembryonic larval development is initiated in the late first instar and early second instar stages by cell growth and renewed cell proliferation in the brain and imaginal discs, where it is nutrient dependent [26]. Cessation of growth in STIM KO larvae (Fig 1D and 1E), suggested a deficit in cell growth and/or cell division. To investigate the status of cell proliferation in STIM KO larvae we chose the well-characterized system of thoracic neuroblasts [27]. Upon comparison of thoracic segments of WT and STIM KO larvae at 70-74h (Fig 1F, first two columns) it was evident that NBs exited from quiescence and entered the proliferative state in both genotypes. Both the NB marker Deadpan (red) and the post-mitotic cell marker Prospero (blue) appeared normal in STIM KO larvae of 70-74h AEL. Subsequently, at 82-86h, the number of postmitotic cells (Prospero positive) decreased significantly in STIM KO animals as compared to the controls but the number of thoracic neuroblasts remained unchanged (Fig 1F, compare third and fourth columns). Upon quantification, the ratio of dividing neuroblasts (Deadpan surrounded by Prospero positive cells) to non-dividing neuroblasts (Deadpan with either no or few Prospero positive cells) changed significantly in 86h aged STIM KO larvae (Fig 1G and  1H). To identify the cause underlying the reduced number of postmitotic cells we analyzed different phases of the cell cycle of thoracic neuroblasts in STIM KO larval brains. For this, we used the genetically encoded FUCCI system [28]. Here, the G1, S, and G2 phases of interphase are marked by green, red and green+red (yellow) fluorescent tags respectively. At 72-76h, both control and STIM KO showed an asynchronous pattern of division. But at 82-86h, control larvae persisted with the asynchronous pattern, whereas a majority of thoracic neuroblasts in STIM KO animals remained in the G2/M state (S1C Fig).
Among other factors, the exit of quiescence and maintenance of neuroblast proliferation at the early second instar stage depends on nutritional intake [26,29,30]. The slow growth and delayed exit from quiescence suggested that STIM KO larvae may lack adequate nutritional inputs. As a first step staged larvae were placed on yeast, mixed with a blue dye and tested for ingestion of food. Even as early as 40-44h AEL there was a significant reduction of food intake in STIM KO larvae (Fig 1I, quantification in Fig 1J). By 80-84h AEL two classes of STIM KO larvae were evident. One with reduced food intake and others with no food intake (Figs 1I and  S1D Fig). The proportion of STIM KO larvae with no food intake reached~70% by 82-86h AEL (S1E Fig). The ability to feed was further quantified in STIM KO animals by measuring mouth hook contractions through larval development [31]. Control larvae (CS) exhibit a steady increase in mouth hook contractions with age, except prior to and during larval molts, indicating greater nutrient intake with age. In contrast, increase in mouth hook movements of STIM KO larvae follows a slower developmental trajectory, with minimal increase as they progress from first to second instar larvae and a cessation at 74h AEL, that is further retarded at 86h (Fig 1K; S1-S4 Videos). Thus, the acceleration of mouth hook movements observed in CS larvae from first to third instar is retarded in STIM KO larvae before the appearance of growth deficits (Fig 1I-1K) suggesting that the consequent nutritional deficits prevent normal growth.

STIM function is required in a subset of larval dopaminergic neurons
To understand how the loss of STIM might affect larval feeding we identified specific cells that require STIM function for larval growth and viability. From previous work, we know that knock out of STIM in dopaminergic neurons marked by THGAL4 [32] leads to larval lethality [22]. Over-expression of a wildtype UASSTIM + transgene, henceforth referred to as STIM + , in dopaminergic neurons marked by THGAL4 rescued larval lethality of STIM KO animals to a significant extent (S2A Fig). Absence of complete rescue by STIM + expression in dopaminergic cells suggests additional requirement for STIM in non-dopaminergic cells of STIM KO larvae, not investigated further in this study. Further to identify specific dopaminergic neurons that require STIM function for growth and viability we tested rescue by overexpression of STIM + in two non-overlapping subsets of dopaminergic neurons marked by THC'GAL4 and THD'GAL4 [33] and henceforth referred to as THD' and THC'. Rescue of STIM KO larvae from 2 nd to 3 rd instar (~90%) was evident upon over-expression of STIM + in THD' marked neurons (Fig 2A and 2B). Because developmental profiles of the control genotypes STIM KO ; THD' and STIM KO ; STIM + are similar to STIM KO at 80-86h and at 168-174h (S2A Fig) these were not included in the experiment in Fig 2A. Though unlikely, the developmental profile of rescue larvae (STIM KO THD'; STIM + ) between 72h and 84h may thus in part be due to presence of the instar (L2) and 3 rd instar (L3) larvae from CS (grey) and STIM KO (magenta) measured at 6h intervals after egg laying (AEL) at the specified time points (mean ± SEM). Number of sets (N) = 3, number of larvae per replicate (n) = 10. *P < 0.05, Student's t-test with unequal variances. P values are given in S2 Table. (D) Representative images of larvae from CS and STIM KO at the indicated time. Scale bar = 1 mm. (E) Measurement of larval length (mean ± SEM) from CS (grey) and STIM KO (magenta) larvae at the specified time points. Number of larvae per genotype per time point is (n) � 12. *P < 0.05 for all time points, Student's ttest with unequal variances. P values are given in S2 Table. (F) Representative images of thoracic neuroblasts marked with Insc>mCD8GFP (green), a neuroblast marker (anti-Deadpan, red) and a marker for post-mitotic cells (anti-Prospero, blue) from control (Insc>mCD8GFP) and STIM KO ; Insc>mCD8GFP animals at the indicated ages. Similar images were obtained from four or more animals. Scale bar = 20μm. (G) Diagrammatic summary of neuroblast proliferation in control (Insc>mCD8GFP) and STIM KO ; Insc>mCD8GFP animals. (H) Stack bar graph showing number of dividing neuroblasts to non-dividing neuroblasts. *P < 0.05, Student's t-test with unequal variances, n = 4 animals from each genotype. P values are given in S2 Table. (I) Representative images of dye-fed larvae from CS and STIM KO at the indicated times AEL, scale bar = 200μm except for CS (80-84h) where scale = 1mm. (J) Quantification (mean ± SEM) of ingested blue dye in CS and STIM KO larvae at the indicated ages by normalizing optical density (OD) of the dye at 655nm to concentration of protein. Number of feeding plates per time point (N) = 6, number of larvae per plate (n) = 10. *P < 0.05, Student's ttest with unequal variances. P values are given in S2 Table. (K) Line graph with quantification of larval mouth hook contractions per 30 seconds (mean ± SEM) from CS and STIM KO at indicated developmental time points. Number of larvae per genotype per time point is (n) � 10. *P < 0.05 at all time points, Student's t-test with unequal variances. All P values are given in S2 Table. https://doi.org/10.1371/journal.pgen.1010435.g001

PLOS GENETICS
THD' and STIM + transgenes on their own. The rescue by THD' was equivalent to the rescue by THGAL4 (20±2 and 18±1.5 viable adults eclosed respectively from batches of 25 larvae; Next, we analysed THD' driven expression of mCD8GFP and identified two classes of GFP positive cells in the larval brain. All GFP expressing cells in the central brain (three cells of DL1 and two cells of DL2 clusters [11,34,35], appear positive for Tyrosine Hydroxylase (TH) but a pair of THD' cells in the ventral ganglion (VG) appear TH negative (TH -ve ) (Fig 2C). To understand the relative contribution of the VG-localised TH -ve cells to THD'>STIM + rescue of STIM KO animals, we restricted THD'GAL4 expression to VG localised THD' neurons using THGAL80 [36] (Fig 2D). Expression of STIM + in the VG localised THD' neurons (THD'GAL4, THGAL80) reduced the rescue of STIM KO larvae significantly (Figs S2A and 2E) and was absent in adults (Fig 2F). Thus, the rescue of viability in STIM KO animals derives to a significant extent from brain-specific THD' dopaminergic neurons.
Ingestion of food and frequency of mouth hook contractions is also rescued by THD'>STIM + expression in STIM KO larvae (Fig 2G-2I). To further confirm the relevance of STIM function in THD' marked dopaminergic neurons we knocked down STIM using a previously characterized STIM RNAi (dsSTIM) [37]. THD'>dsSTIM animals exhibit delayed larval growth (Figs 2J and S2B) and reduced feeding (Fig 2K and 2L) but no larval lethality. Adults however exhibit reduced body weight (S2C Fig). Taken together these data identified a CNS-specific subgroup of dopaminergic neurons that require STIM function for persistent feeding during the early stages of larval growth. Stack bars with the number of adults (mean ± SEM) that eclosed at 320 to 326h AEL from the indicated genotypes. The genotype of rescue larvae is STIM KO ; THD'>STIM + . Number of sets (N) = 3, number of organisms per set (n) = 25. (C) Representative confocal images of the larval brain from animals of the genotype THD'>mCD8GFP. Anti-GFP (green) indicates the expression of THD'GAL4 and anti-TH (magenta) marks all dopaminergic cells. Asterisks mark TH +ve cells in CNS whereas arrowheads mark non-TH positive cells in ventral ganglia of larval brain (i and ii). DL1 and DL2 clusters in the central brain of three and two dopaminergic cells respectively are marked (iii). Scale bars = 20μm. (D) Representative confocal images of the larval brain from animals of the genotype THGAL80,THD'>mCD8GFP. THD'GAL4 driven GFP expression (green) is suppressed in DL1 and DL2 clusters in the CNS (asterisk) by THGAL80 but not in the ventral ganglia (arrowheads). Scale bar = 20μm. (E) Line graph shows the number (mean ± SEM) of 3 rd instar larvae from CS, STIM KO  In all graphs and box plots, different alphabets represent distinct statistical groups as calculated by one way ANOVA followed by post-hoc Tukey's test. P values for individual panels are given in S2 Table. https://doi.org/10.1371/journal.pgen.1010435.g002

Neuronal excitability and dopamine release requires STIM
The status of THD' marked central brain dopaminergic cell clusters, DL1 and DL2 was investigated next in STIM KO larval brains at 80-84h, when a few viable organisms are still present despite cessation of growth and feeding (Fig 2G). THD' cells were marked with GFP in controls, STIM KO and STIM KO animals with STIM + rescue and the brains were stained with anti-TH sera. THD'>mCD8GFP cells appeared no different in STIM KO as well as STIM + rescued STIM KO animals at 80-84h AEL when compared to controls at either 58-62h or 80-84h AEL (S3A Fig). Moreover, the numbers of THD' GFP cells and TH positive cells in the CNS also appeared identical (S3B Fig). Therefore, the loss of STIM does not lead to the loss of dopaminergic neurons in the larval brain.
In order to test if reduced feeding in STIM KO larvae is indeed due to a loss in dopamine signalling we measured larval feeding with knockdown of a key dopamine synthesising enzyme Tyrosine Hydroxylase (TH), in THD' neurons (THD'>dsTH). Knockdown of TH led to significantly fewer mouth hook contractions in larvae at 80-86h AEL (Fig 3A, S5-S7 Videos), indicating reduced feeding, This was accompanied by slower progression through larval moults and some mortality at each larval stage. Finally out of a total of 25 just 20±1.2 3 rd instar larvae pupated, of which 15 adults emerged (S3C Fig). In agreement with lower nutrient intake during larval stages, third instar larvae were smaller in size (Fig 3B-3C), and gave rise to adults with significantly reduced body weight (Fig 3D).
To understand how loss of STIM in THD' marked neurons might affect their neuronal function, we investigated properties of excitation. For this purpose, Potassium Chloride (KCl, 70mM) evoked cytosolic calcium transients were measured from THD' neurons using the genetically encoded Ca 2+ sensor GCaMP6m in ex vivo preparations of similarly staged control (58-62h AEL) and STIM KO (70-74h AEL) larvae. We chose these time points because at 72h STIM KO larvae appear healthy and developmentally similar to control larvae at 60h (Fig 1D). THD' cells responded with similar changes in GCaMP intensity, in control and STIM KO larvae at these time points (Fig 3E-3G). However, the ability to evoke and maintain cytosolic Ca 2+ transients upon KCl depolarisation was lost in THD' neurons from the DL1 and DL2 clusters of STIM KO larvae at 76-80h AEL (Fig 3E-3G). Overexpression of STIM + in THD' cells of STIM KO larvae rescued the KCl evoked Ca 2+ response in larvae as late as 80-84h AEL (Fig 3E-3G).
STIM requirement for maintaining excitability of THD' neurons was tested further by KCl stimulation of THD' neurons with STIM knockdown (THD'>dsSTIM) from 2 nd instar larvae aged 58-62h. Two classes of responses to depolarisation by KCl were observed. Most cells (70%) responded with normal or greater changes in intensity as compared to control cells, whereas in 30% of cells KCl did not evoke a Ca 2+ transient (S3D- S3F Fig). We attribute this heterogeneous response to differential STIM knockdown by the RNAi in individual THD' cells and a potential effect of STIM knock-down on ER-Ca 2+ homeostasis (see below).
These data suggest that loss of STIM affects membrane excitability properties and the ability of THD' neurons to respond to stimuli. This idea was tested directly by the expression of transgenes that alter membrane potential. Over-expression of an inward rectifier K + channel, Kir2.1 in THD' neurons, that is known to hyperpolarise the plasma membrane [38], resulted in developmental delays followed by the lethality of second and third instar larvae (S4A Fig). Further, overexpression of a bacterial Na + channel NaChBac [39] in THD' neurons of STIM KO larvae evinced a weak rescue of developmental deficits, including the transition to third instar larvae (4.4±0.4) and adult viability (2.4±0.6 from of batches of 15 animals; Fig 3H-3K). Though weak, NaChBac's rescue was consistent. We attribute the variability to a stochastic effect of NaChBac in THD' cells. This is also evident from the variable extent of rescue in growth observed in STIM KO ; THD'>NaChBac larvae (Fig 3I and 3J). Alternatively, in addition to neuronal excitability, STIM might affect other cellular functions such as ER stress, that are not alleviated by expression of NaChBac, resulting in the weak rescue. Control animals with overexpression of NaChBac in THD' neurons exhibit delayed pupariation (S4B Fig).
Neuronal excitability is required for neurotransmitter release at presynaptic terminals. We hypothesized that dopamine release from THD' neurons might be affected in STIM KO larvae. To test this idea, we identified the pre-synaptic (green) and post-synaptic (red) terminal regions of THD' neurons by marking them with Syt-eGFP and Denmark respectively (Fig 4A) [40]. Analysis of pre-synaptic regions (Syt-eGFP expression) identified three distinct areas in the CNS. One at the centre of the CNS corresponding to the mushroom body (Fig 4A; asterisk), the second as a branched form in the basomedial protocerebrum of the CNS (Fig 4A; arrowhead) and the third one consisting of punctae spread across the oesophageal regions where brain-gut interactions take place (Fig 4A; hash). Based on the projection patterns observed we speculate that cells marked by THD' correspond to DL1-2, DL1-5, DL1-6 from the DL1 cluster and DL2-2 and DL2-3 from the DL2 cluster [35].
Next, dopamine release was measured in the most prominent presynaptic areas of THD' neurons, corresponding to the MB and the basomedial protocerebrum, by change in fluorescence of the genetically encoded fluorescent GPCR-activation-based-DopAmine sensor (GRAB DA ) [41]. Dopamine release in THD' neurons of STIM KO larvae at 76-80h is significantly attenuated as compared with controls (Fig 4B and 4C). Importantly, overexpression of STIM + in THD' neurons rescued dopamine release, though with altered dynamics from control animals (Fig 4B and 4C; see discussion). We chose to measure dopamine release in 76-80h STIM KO larvae because THD' neurons in their brains no longer responded to KCl evoked depolarization (Fig 3B) even though the larvae appear normal (Fig 1D). Dopamine release was stimulated by Carbachol (CCh), an agonist for the muscarinic acetylcholine receptor (mAChR), that links to Ca 2+ release from ER-stores through the ER-localised IP 3 receptor [42] and is expressed on THD' neurons [43]. CCh-induced changes in ER-Ca 2+ were tested directly by introducing an ER-Ca 2+ sensor [44] in THD' neurons (Fig 4D and 4E). Though ER-Ca 2+ release, in response to CCh could be measured in just 7 out of 23 cells, the subsequent step of ER-store refilling, presumably after Store-operated Ca 2+ entry into the cytosol through the STIM/Orai pathway, could be observed in all control THD' cells (58-62h AEL), whereas it was absent in THD' neurons from STIM KO brains (76-80h AEL; Fig 4E). The ER-Ca 2+ response was rescued by over-expressing STIM + in THD' neurons ( Fig 4D, and 4E). Taken together our data establish an important role for STIM-dependent ER-Ca 2+ homeostasis in maintaining In all box plots, circles represent single larvae or flies. The box plots span 25 th and 75 th percentiles with the median (bar), and mean (square). Alphabets represent distinct statistical groups as calculated by one way ANOVA followed by post-hoc Tukey's test. P values are given in S2 Table. https://doi.org/10.1371/journal.pgen.1010435.g003

PLOS GENETICS
optimal dopamine release from THD' neurons in turn required for the larval drive to feed constantly.
Because late third instar larvae stop feeding [3,19,20,45] we hypothesized that CCh-stimulated Ca 2+ responses in THD' cells might change in wandering stage third instar larvae. To test this idea we monitored carbachol-stimulated GCaMP activity in THD' neurons from wildtype larvae at 96h AEL (mid 3 rd instar), 120h AEL (early wandering stage) and 124h AEL (late wandering stage). A robust GCaMP response was observed at 96h, whereas at the beginning of the wandering stage (118-122h AEL), the peak response was both reduced and delayed. A further delay in the peak response time was observed in late wandering stage larvae (122-126h AEL) (S4E and S4F Fig). Thus with gradual cessation of feeding in late third instar larvae, the dynamics of CCh-stimulated Ca 2+ responses in THD' neurons also undergo changes (see discussion).

THD' dopaminergic neurons both activate and inhibit specific neuropeptidergic cells
A role for neuropeptides in modulating larval feeding has been identified previously [5][6][7] Based on these findings we postulated that dopamine release from THD' cells might modulate neuropeptidergic cells in the larval brain. This idea was tested by optogenetic stimulation of red-shifted Channelrhodopsin (CsChrimson) [46] expressing in THD' (THD'>UAS-Chrimson) cells followed by Ca 2+ imaging from GCaMP6f expressing neuropeptide cells (C929Lex-A>LexAopGCaMP6f) (Fig 5A). Upon optogenetic activation of THD' neurons, a change in cellular Ca 2+ signals was observed in a total of 64 peptidergic neurons from 9 brains, including some lateral neurosecretory cells (LNCs), median neurosecretory cells (MNSc), and regions of the suboesophageal zone (SEZ) (Fig 5B and 5C). Elevated Ca 2+ signals were observed in a subset of neuropeptidergic cells (Fig 5B-5D, yellow asterisks, n = 24), whereas in some cells Ca 2+ signals were reduced (Fig 5B, 5C and 5E, red asterisks, n = 40). There was no change in cellular Ca 2+ upon optogenetic stimulation in the remaining cells with visible GCaMP6f fluorescence (n = 35). Upon cessation of optogenetic activation of THD' neurons Ca 2+ levels returned to baseline (Fig 5D and 5E). Individual cells exhibit either activatory or inhibitory responses upon repeated optogenetic stimulation with pulses of red light (Fig 5F (i) and (ii)), indicating that THD' cells evoke specific responses of either stimulation or inhibition based on the class of neuropeptide cells. Specificity of the THD'>Chrimson evoked response was further confirmed by testing brains from larvae that were reared without all trans-Retinal (ATR, a cofactor for the function of Channelrhodopsin) [47] and by imaging in the absence of light (Fig 5G).

STIM expression in THD' neurons regulates the expression of insulin-like peptides
Though optogenetic stimulation of THD' cells shows dopaminergic modulation of neuropeptidergic neurons (Fig 5A and 5B) it does not allow identification of specific neuropeptides that function downstream of the THD' neurons. Optogenetic activation of THD' and Ca 2+ responses in neuropeptidergic cells (marked by C929LexA>LexAopmCherry) helped identify three clusters of neuropeptidergic cells that are downstream of THD' neurons and include the well-characterised ilp expressing MNSc cluster [48], as a putative target of THD' neurons ( Fig  5B). Analysis of an RNAseq experiment identified changes in gene expression in brains from second instar STIM KO larvae (72-76h AEL) with developmentally comparable CS brains (58-62h; S5A Fig and S3 Table), found significant changes in expression of ilp2, ilp3 and ilp5. Whereas ilp2 and ilp5 were significantly downregulated, ilp3 was upregulated more than 5 fold (Fig 6A). The differential regulation of ilp2, -3 and -5 expression in STIM KO brains was further validated by qPCRs (Fig 6B). Importantly, expression of ilp3 and ilp5 were restored back to normal in brains from STIM KO larvae, rescued by overexpression of STIM + in THD' neurons ( Fig 6B). During normal larval development ilp3 transcript levels are low in actively feeding larvae (L2 and L3, 12h) and are gradually upregulated in later stages of L3 as larvae stop feeding (Fig 6C; DGET [49]. Based on the up-regulation of ilp3 in STIM KO larvae (Fig 6A and 6B), we hypothesised that knock down of ilp3 in MNSc might rescue STIM KO larvae. Indeed, partial rescue of larval lethality in 2 nd instar larvae followed by their transition to 3 rd instar larvae (5 ±0.5) was observed (Fig 6D-G). A few of the rescued 3 rd instar larvae grew to full size, and pupariated (Fig 6E, L3b type larvae) and some even eclosed as adults (2.3±0.3 from 3 batches of 25 animals; Fig 6D and 6E). The partial rescue observed by ilp3 knockdown may be due to roles of additional dopamine-modulated neuropeptides plus the lower expression of ilp2 and ilp5 in STIM KO animals (Fig 6A and 6B) because both ilps are growth signals in 2 nd and 3 rd instar larvae. A small proportion of STIM KO larvae rescued by ilp3 knockdown appear significantly larger than control larvae (Fig 6E and 6F, L3b), indicating that loss of ilp3 can rescue growth in some animals. This idea is further supported by knock-down of ilp3 in MNSc from wildtype animals that resulted in slower transition to pupariation and larger pupae, and overweight adults (Figs 6H-6J and S5A-S5B). Conversely, over-expression of ilp3 in MNSc resulted in delayed larval transition from L2 to L3 and smaller sized larvae (Fig 6K-6M) similar to delayed larval development in ilp5 knockdown animals (S5A Fig). However, overexpression of ilp3 had no effect on feeding as indicated by measurement of larval mouth hook movements (Fig 6N) even though adults were of reduced body weight (S5B Fig). Taken

PLOS GENETICS
together, these data show that dopamine signals from THD' cells are required to maintain normal expression of growth promoting ILPs (ilp2 and ilp5) and repress expression of ilp3 that appears to function as an anti-growth signal.
A direct synaptic connection between central dopaminergic neurons and the MNSc has not been reported [5,50] and is likely not supported by data that mapped synaptic connections in the larval brain [35,[50][51][52]. However, a neuromodulatory role for dopamine is documented where it can effect a larger subset of neurons, beyond direct synaptic partners, by means of diffusion aided volumetric transmission [53][54][55][56]. To test this possibility we began by measuring larval developmental transitions in animals with knockdown of three dopamine receptors, Dop1R1, Dop2R2 and DopEcR in the MNSc. Among these, knockdown of Dop1R1 delayed development, reduced larval and adult viability (Fig 7A) and negatively impacted growth (Fig 7B and 7C). A weaker phenotype of delayed development and loss of viability was observed with knockdown of DopEcR, whereas larvae with knockdown of Dop2R2 appeared normal (Fig 7A). The response of MNSc to dopamine was tested next. Brains expressing a Ca 2 + sensor in the MNSc (MNSc>GCaMP6m) were stimulated with dopamine in the presence of a Na v blocker Tetrodotoxin [57], to prevent extraneous neuronal inputs. Of the seven targeted MNSc in one hemisphere, we observed consistent activation of one cell whereas dopamine addition inhibited the Ca 2+ response in three to four cells (Fig 7D; quantified in Fig 7E). No changes in the Ca 2+ responses of MNSc were observed in the absence of dopamine (Fig 7D  and 7E).

Discussion
In Drosophila, as in other holometabolous insects, growth is restricted to the larval stages. In early stages of larval development cells exit mitotic quiescence and re-enter mitosis resulting in organismal growth [1,25,58].This change is accompanied by an increase in the feeding rate of the organism so as to provide sufficient nutrition for the accompanying growth in organismal size. In STIM KO larvae we observed a loss of this ability to feed persistently starting from early second instar larvae. The focus of this feeding deficit lies in a subset of central dopaminergic neurons that require STIM function to maintain excitability. Importantly, these dopaminergic  Table. https://doi.org/10.1371/journal.pgen.1010435.g006

PLOS GENETICS
neurons communicate with multiple neuropeptidergic cells in the brain (Fig 5D) to regulate appropriate changes in larval feeding behaviour. The identified dopaminergic cells also communicate with ilp producing neuropeptidergic cells, the MNSc, through which they appear to impact larval growth.

STIM and excitability of dopaminergic neurons
We identified the THD' cells as critical for larval feeding from their inability to function in the absence of the SOCE regulator STIM. Loss of excitability and the absence of dopamine release from THD' cells in STIM KO larvae (Figs 3A-3C and 4B-4D) suggests that voltage-dependent receptor activity is required to maintain growth in early 2 nd instar larvae. Changes in expression of ion channels and presynaptic components have been observed earlier upon knockdown of STIM in Drosophila and mammalian neurons [59,60]. Moreover, loss of STIMdependent SOCE in Drosophila neurons effects their synaptic release properties [61]. Partial rescue of viability in STIM KO organisms by over-expression of a bacterial sodium channel NaChBac (Fig 3D-3G) and restoration of dopamine release upon rescue by STIM + (Fig 4B  and 4C) supports the idea that STIM-dependant SOCE maybe required for appropriate function and/or expression of ion channels and synaptic components in THD' neurons. Changes in ER-Ca 2+ (Figs 4D and 5E) suggest that STIM is also required to maintain neuronal Ca 2+ homeostasis.

Dopaminergic control of larval food seeking
Whilst mechanisms that regulate developmental progression of Drosophila larvae have been extensively studied, neural control of essential changes in feeding behaviour that need to accompany each larval developmental stage have not been identified previously. Artificial manipulation of activity in the central dopaminergic neuron subset studied here (THD'), either by expression of an inward rectifying potassium channel (Kir2.1) (S4A Fig) or the bacterial sodium channel NaChBac (S4B Fig), suggests an important role for THD' neurons during larval development. This idea is supported by the altered dynamics of muscarinic acetylcholine receptor (mAChR) stimulated Ca 2+ release observed in THD' neurons between early, mid and late third instar larvae when larval feeding slows down and ultimately stops (S4C and S4D  Fig) and re-iterates that signaling in and from these neurons drives larval feeding whereas lower carbachol-induced Ca 2+ responses signal cessation of feeding. A weaker rescue of STIM KO larvae is also obtained from STIM + expression in the THC' neuron subset. Taken together these observations suggest a neuromodulatory role for dopamine, where DA release from THD' neurons has a greater influence on feeding than the DA release from THC' neurons, possibly due to the DL1 and DL2 cluster (among THD' marked neurons) receiving more

PLOS GENETICS
feeding and metabolic inputs [7,35,58]. A role for cells other than THD', in maintaining kinetics of dopamine release required for feeding behaviour are also indicated because expression of STIM + in THD' neurons did not revert kinetics of dopamine release to wild type levels (Fig 4C). The prolonged dopamine release observed in wild-type THD' neurons may arise from synaptic/modulatory inputs to THD' neurons from other neurons that require STIM function.
Though the cells that provide cholinergic inputs to THD' cells have not been identified it is possible that such neurons sense the nutritional state. In this context, two pairs of cells in the THD' subset also motivate the search for food in hungry adult Drosophila [62]. Starved flies with knock down of the mAChR on THD' neurons exhibit a decrease in food seeking behaviour [43]. Cholinergic inputs to THD' neurons for sensing nutritional state/hunger may thus be preserved between larval and adult stages.
Interestingly, dopamine is also required for reward-based feeding, initiation, and reinforcement of feeding behaviour in adult mice [63]. These findings parallel past studies where prenatal mice genetically deficient for dopamine (DA -/-), were unable to feed and died from starvation. Feeding could however be initiated upon either enforced supplementation or injection with L-DOPA [64] allowing them to survive. More recent findings show that dopaminergic neurons in the ventral tegmental area (VTA), and not the substantia nigra, drive motivational behaviour and facilitate action initiation for feeding in adult mice [65].

Dopaminergic control of neuropeptide signaling
Both activation and inhibition of specific classes of neuropeptidergic cells by optogenetic activation of THD' cells suggests a dual role for dopamine possibly due to the presence of different classes of DA receptors. The Drosophila genome encodes four DA receptors referred to as Dop1R1, Dop1R2, DD2R and a non-canonical DopEcR [66]. Dop1R1, Dop1R2 and DopEcR activate adenylate cyclase and stimulate cAMP signaling whereas DD2R is inhibitory [66]. Cell specific differences among dopamine receptors have been observed in adults. Down regulation of Dop1R1 on AstA and NPF cells shifted preference towards sweet food whereas down regulation of DopEcR in DH44 cells shifted preference towards bitter food [67]. In third instar larvae a dopaminergic-NPF circuit, arising from central dopaminergic DL2 neurons, two cells of which are marked by THD'GAL4 (Fig 2E), motivates feeding in presence of appetitive odours [6,45]. The dopamine-neuropeptide axis identified here demonstrates a broader role for dopamine in regulating neuropeptide release and/or synthesis, in the context of larval feeding behaviour, perhaps similar to the mammalian circuit described above.
Of specific interest is the untimely upregulation of ilp3 transcripts in STIM KO larvae. Rescue of lethality in STIM KO larvae either by bringing back activity to THD' neurons or by reducing ilp3 levels suggests an interdependence of Dopamine-Insulin signaling that is likely conserved across organisms [68][69][70][71]. Our data suggest that ilp3 expression is suppressed during the feeding and growth stages of larvae (Fig 6H-6M), and once enough nutrition accumulates expression of ilp3 is up-regulated, concurrent with a reduction in carbachol-induced Ca 2+ signals in THD' neurons, possibly followed by upregulation and release of ilp3. The idea of ilp3 as a metabolic signal whose expression is antagonistic to larval growth is supported by the observation that knock-down of ilp3 in the MNSc leads to larger pupae in wild type animals and larger larvae in STIM KO (Fig 6F and 6G). To our knowledge this is the first report of ilp3 as a larval signal that is antagonistic to growth. Given that Drosophila encode a single Insulin receptor for ilp2, ilp3 and ilp5 [72] the cellular mechanism of ilp3 action remains to be elucidated. Possibly, ilps with different affinity for the insulin receptor stimulate different cellular subsets and/or different intracellular signaling mechanisms, including ecdysone signaling that is essential for larval transition to pupae [72]. Interestingly, in STIM KO larval brains there is a significant increase in expression of the Insulin Receptor (S1 Table). Further studies are needed to fully understand ilp3 function in larvae.
Expression of other neuropeptides did not show significant changes in STIM KO larval brains (S1 Table), suggesting that for neuropeptidergic cells in the LNC and SEZ, dopamine signals alter release properties rather than synthesis. However, we were unable to identify specific neuropeptides for cells in the LNC and the SEZ that responded upon activation of THD'.
The importance of dopamine for multiple aspects of feeding behaviour is well documented in juvenile and adult mice [63,64]. Of interest are more recent findings linking dysregulation of dopamine-insulin signaling with the regulation of energy metabolism and the induction of binge eating [73,74]. The identification of a simple neuronal circuit where dopamine-insulin signaling regulates feeding and growth could serve as a useful model for investigating new therapeutic strategies targeted towards the treatment of psychological disorders for obesity and metabolic syndrome [73,75].

Fly rearing and stocks
Drosophila strains were reared on standard cornflour agar media consisting of 80 g corn flour, 20 g glucose, 40 g sugar, 15 g yeast extract, 4 ml propionic acid, 5 ml p-hydroxybenzoic acid methyl ester in ethanol, 5 ml ortho butyric acid in 1l at 25˚C, unless otherwise specified, under a 12:12 hr light: dark cycle. In all studies the Canton S (CS) strain was used as a wild-type control and CRISPR-Cas9 generated deficiency for STIM referred to as STIM KO served as a null mutant for the Drosophila STIM gene [22]. Details of other fly lines used are provided in Table 1 below.

Staging
Synchronized larvae of the appropriate ages as described below were collected and transferred to agar less media containing yeast (4gm), sucrose (8gm), cornflour (16gm), Propionic acid (1ml) and 0.05gm of Benzoic acid in 1ml of absolute alcohol. The number of viable organisms and the developmental stage were scored at specific time points as mentioned below and in the figures and figure legends.
Larval staging experiments were performed to obtain lethality and developmental profiles of the indicated genotypes as described previously [76]. Depending on the experiment, timed and synchronized egg-laying was done either for 6h to allow development profiling at 60-66h, 80-86h, 128-136h, 176-182h and 320-326h after egg laying or for 2h at 35-37h(36h), 41-43h (42h), 47-49h(48h), 53-55h(54h), 59-61h(60h), 65-67h(66h), 71-73h(72h), 83-85h(84h), and 89-91h(90h) after egg laying for identifying a lethality window between 36-90h. Larvae were collected at either 60-66h or 35-37h after egg laying (AEL) in batches of 25 (for developmental profile) or 10 (for lethality window). They were screened and staged subsequently. Heterozygous larvae were identified using dominant markers (FM7iGFP, TM6Tb, and CyOGFP) and removed. Each batch of larvae was placed in a separate vial and minimally three vials containing agar-less media were tested for every genotype at each time point. The larvae were screened at the indicated time points for the number of survivors and stage of development, determined by the morphology of the anterior spiracles and mouth hooks [77]. Experiments to determine the viability of experimental genotypes and their corresponding genetic controls were performed simultaneously in all cases. Larval images were taken on the MVX10 Olympus stereo microscope using an Olympus DP71 camera.

UASFUCCI
Marks different phases of cell cycle with fluorescent markers RRID: BDSC_55100

UASNaChBac
Increases sodium conductance and therefore activates the neuron RRID: BDSC_9468

Feeding assay
Feeding assay was performed at specific developmental time points in larvae (40-44h, 58-62h, 80-84h AEL) of the specified genotypes. Larvae were placed in a 35mm punched dish with coverslip at the base thus creating a small depression in the centre of the coverslip. In this depression a coin sized cotton swab was placed containing 4.5.% of yeast solution with 0.25% eriogluasin dye (blue dye). For scoring the number of larvae that fed, 30 larvae per plate were taken and incubated in the feeding plate for 4hrs at 25˚C. Larvae were removed from the paste, washed, collected and scored for presence of blue dye (Dye +ve ) and absence of blue dye within the gut (Dye -ve ). For quantification of ingested blue dye, 12-15 larvae were incubated for 2h in yeast paste with the blue dye. Larvae were removed from the yeast paste and 10 larvae with blue dye in the gut were washed, and homogenized in 50μl of cold 1xPBS. The homogenate was spun at 5k for 2 minutes in a table top Eppendorf centrifuge. The supernatant (2μl) was taken for quantification of protein using the Thermo scientific Pierce Protein assay kit, Cat#23227. Optical density (OD) at 625nm as a measure of ingested blue dye was measured from 30μl of the homogenate. Due to variation in larval sizes between control and experimental samples, the OD was normalized to whole larval protein concentration (μg/μl). OD was obtained using the SkanIt Software 6.1.1 RE for Microplate Readers RE, ver. 6.1.1.7. Larval imaging and processing was performed on the Olympus MVX10 stereo microscope using FIJI software.

Larval imaging and measurement
Staged larvae from specified genotypes were collected at specific development time points, anesthetized on ice for 1h and mounted with ice cold HL3 buffer. The mounted larvae were imaged immediately using an Olympus MVX10 stereo microscope. For measurement of larval length from mouth hook to tip of the posterior spiracle FIJI software was used. A minimum of 10-15 larvae were taken per genotype for length analysis.

Quantification of larval mouth hook contractions
Mouth hook contractions were measured by placing 1-3 appropriately staged larvae in a drop of 2% yeast solution in a petridish. Videos were taken for 30 seconds on an Olympus MVX10 stereo microscope. For each genotype a minimum of 10 animals were imaged. Mouth hook contractions were counted manually from the visualised videos.

Pupal volume measurement
Pupal volume was measured by obtaining the width and height of each pupal image from the Olympus MVX10 stereo microscope. A formula for obtaining the volume of a cylinder, (πr 2 h) was applied to calculate the volume [78].

Adult fly weight measurement
For weight measurement of adult flies, 10 flies (5 females and 5 males) of the appropriate genotype were taken 6-10 hr post-eclosion and weighed after placing them in a small Eppendorf tube. Thereafter, the weight of the same empty tube was measured. Fly weights were calculated by subtracting the weight of the tube from the total weight of flies + tube. A minimum of five such measurements were performed for each genotype.

Immunohistochemistry
Larval brains were dissected in ice-cold 1xPBS and fixed with 4% paraformaldehyde in 1xPBS on the shaker for 20mins at room temperature. Fixed brains were washed with PBTx (0.3% TritonX-100 in 1XPBS) 3-4 times at 10minutes intervals, blocked with 5% normal goat serum (NGS) in PBTx for 2hrs at room temperature, and incubated with primary antibodies diluted in 5%NGS+PBTx at the appropriate concentration as mentioned below, overnight at 4˚C. Antibody solution was removed and re-used upto three times. Brains were washed with PBTx 3 times at 10 minute intervals followed by incubation with secondary antibodies at the dilutions described below, for 2hrs at room temperature and three washes in PBTx of 10minute intervals each. Brains were mounted in 70% glycerol diluted in 1xPBS. Confocal images were acquired by using FV3000 LSM and the Fluoview imaging software.

Ex-vivo imaging of the larval brain
GCaMP signals were obtained from appropriately aged larval brains dissected from the specified genotypes and dissected in hemolymph like saline (HL3) (70mM NaCl, 5mM KCl, 20mM MgCl2, 10mM NaHCO3, 5mM trehalose, 115mM sucrose, 5 mM HEPES, 1.5mM Ca2+, pH 7.2). Dissected brains were transferred to a 35mm punched dish with a cover slip adhered to the bottom. Brains were embedded in *5μl of 0.8% low melt agarose (Invitrogen, Cat#16520-100) and bathed in 86μl of HL3. Images were acquired as a time series on an XY plane at an interval of 2sec using a 20X-oil objective on an Olympus FV3000 inverted confocal microscope (Olympus Corp., Japan). For KCl stimulation, at the 40 th frame, 7μl of HL3 was added and at the 80 th frame 7μl of 1M KCl was added. The final concentration of KCl in the solution surrounding the brain was 70mM. For stimulation with Carbachol (Sigma Aldrich Cat# C4382), 10μl of HL3 was added at the 40 th frame followed by 10μl of 100mM Carbachol at the 80 th frame. Final carbachol concentration was maintained at 0.5mM Ca 2+ responses were imaged till the 300 th frame (600sec).
Changes in ER-Ca 2+ were measured using an ER-GCaMP-210 strain [44]. The brain sample was prepared as above. Images were acquired as a time series on an XY plane at an interval of 1 sec using a 20X oil objective on an Olympus FV3000 inverted confocal microscope (Olympus Corp., Japan). For Carbachol stimulation, 10μl of HL3 was added at the 50 th frame and 10μl of 100mM of Carbachol was added at the 100 th frame. Final carbachol concentration was maintained at 0.5mM. Images were obtained for 600 frames (600 secs).
Larval brain expressing MNSc>GCaMP is used for Dopamine (DA) (Sigma, Cat#H8502) stimulation. Dissected brains were transferred to a 35mm punched dish with a cover slip adhered to the bottom. Brains were embedded in *5μl of 0.8% low melt agarose (Invitrogen, Cat#16520-100) and bathed in 80μl of HL3 having 1uM of TTX. Images were acquired as a time series on an XY plane at an interval of 1.5sec using a 20X-oil objective on an Olympus FV3000 inverted confocal microscope (Olympus Corp., Japan). At the 30 th frame, 10μl of HL3 was added and at the 60 th frame 10μl of 1mM DA was added. Ca 2+ responses were imaged till the 250 th frame (450sec).

Analysis of Optogenetic signals
Larvae from the specified genotypes were reared in fly media containing 0.2mM ATR (Alltrans retinal Sigma-Aldrich Cat#R2500). Brain samples were prepared as mentioned above. Images were acquired as a time series on an XY plane at an interval of 2 sec/frame using a 20X oil objective on an Olympus FV3000 inverted confocal microscope (Olympus Corp., Japan). For optogenetic stimulation of CsChrimson, a 633nm LED (from Thor labs) was used and GCaMP6f fluorescent images were obtained simultaneously using a 488nm laser line so as to measure changes in cytosolic Ca 2+ upon CsChrimson activation. Image acquired till 200 th frames (400 secs).
A minimum of 6 independent brain preparations were used for all live imaging experiments and the exact number of cells imaged are indicated in the figures. Raw fluorescence data were extracted from the marked ROIs using a time series analyser plugin in Fiji. ΔF/F was calculated using the following formula for each time point (t): ΔF/F = (F t /F 0 )/F 0 , where F 0 is the average basal fluorescence obtained from the first 40 frames.

RNA isolation and library preparation
Larval brains (15-20 per sample) were dissected from animals of appropriate genotypes and age (CS, 58-62h AEL; STIM KO , 72-76h AEL), in ice cold phosphate buffered saline (PBS). Larval brain samples were transferred to tubes containing 300μl TRIzol (Invitrogen-15596018), and vortexed immediately for 10-15 secs. The vortexed samples were stored at −80˚C for further processing for up to one week. RNA isolation was done using Trizol, following manufacturer's protocol. RNA was run on a Bioanalyzer (Agilent) to ensure integrity. For each sample, 10ng of isolated RNA was used for cDNA synthesis using the SMART-Seq v4 Ultra Low Input RNA Kit, following manufacturer's protocol. The kit employs polyA tail complementary primer, template switching and extension by reverse transcriptase. Qubit dsDNA HS kit, following manufacturer's protocol, was used for assessing DNA concentration using 1μL of the cDNA sample. Further, Nextera XT DNA library kit (Illumina-FC-131-1024) was used for library preparation with 1ng of cDNA, following manufacturer's protocol. cDNA libraries were made from four independently isolated sets of brain RNA from each genotype. Libraries were pooled (2nM) at equimolar quantities and subjected to high depth sequencing in Illumina HiSeq 2500 (1 x 50bp).

RNASeq analysis
FastQC and trimmomatic were used for QC of raw reads and adapter removal (if found) respectively. Raw reads were then mapped to Drosophila genome dm6 assembly using hisat2 [79]. The output BAM files were sorted and indexed using Samtools. The BAM files were used as input for htseq-counts. The htseq counts were then used as an input in DESeq2 (Bioconductor-R package) to obtain differentially expressed genes using default thresholds. Ggplot

RNA isolation and quantitative PCR
RNA from 10-15 larval brains was isolated as described above for cDNA library preparation. RNA (1μg) was taken for cDNA synthesis. DNAse treatment and first strand synthesis was performed as previously described [80]. Reverse Transcription followed by PCR (RT-PCR) was performed in a reaction mixture of 25 μl with 1 μl of the cDNA. Quantitative real time PCRs (qPCRs) were performed in a total volume of 10μl with Kapa SYBR Fast qPCR kit (KAPA Biosystems) on a QuantStudio 3 Real-Time PCR system. Duplicates were performed for each qPCR reaction. Minimum of three biological replicates were taken for each qPCR reaction. rp49 was used as the internal control. The fold change of gene expression in any experimental condition relative to wild-type was calculated as 2 −ΔΔCt , where ΔΔCt = (Ct (target gene) − Ct (rp49)) Expt. − (Ct (target gene) − Ct (rp49)) Control.
Sequences of PCR primers used are as follows: R_5'GAGTCGCAGTATGCCCTCAA3'

Quantification and statistical analysis
All bar graphs and line plots show the means and standard error of means. In boxplots, horizontal lines in the box indicate median, box limits are from 25 th -75 th percentiles, and individual data points are represented by closed circles (unless otherwise specified in the figure legends). Unpaired student t-Test (for two genotypes) and one way ANOVA followed by posthoc Tukey's significance test (for data with multiple genotypes) was performed to calculate P values, given for all figures in S2 Supporting information S1 Fig. (A-B). Alphabets indicate different statistical groups. In all panels significant changes between relevant genotypes at the indicated stages were calculated by one way ANOVA followed by posthoc Tukey's test. P values are given in S2 Table. (TIF)  Fig. (A). Upregulated (magenta colour, log2fold � +1; p<0.05) and downregulated (green, log2fold � -1; p<0.05) gene sets in STIM KO larval brains (72-76h AEL) as compared to CS larval brains (58-62h AEL), depicted as a volcano plot. N = 4. Also see S3 Table for gene