Flies were grown at 25°C in 12h:12h L:D cycle. Normal food: (1L recipe: 80g corn flour, 20g Glucose, 40g Sugar, 15g Yeast Extract, 4mL propionic acid, p-hydroxybenzoic acid methyl ester in ethanol 5mL, 5mL ortho butyric acid) For nutritional stress assay, flies were allowed to lay eggs for 6 hours on normal food. After 88hours, larvae were collected and transferred to either normal or NR (100mM Sucrose) food. Pupae and adults were scored after 10 days of observation.

Canton S was used as the wild type (+) control.

The following strains were obtained from Bloomington Drosophila Stock Centre: AKH-GAL4 (25684), Crz-GAL4 (51976), DSK-GAL4 (51981), sNPF-GAL4 (51981), UAS-Crz^IR25999 (25999), UAS-dicer2 (24651), UAS-TeTxLc (28837), UAS-TeTxLc-IMP (28838), UAS-CaMPARI (58761), UAS-GFP^nls (4776), UAS-mRFP (32218), UAS-TrpA1 (26263), UAS-InR (8248), UAS-amon^IR (28583), UAS-Ral^DN (32094)

The following strains were obtained from Drosophila Genetic Resource Center, Kyoto: sNPF-GAL4 (113901)

The following were from Vienna Drosophila Research Centre stock collection: UAS-IP₃R^IR (106982), UAS-STIM^IR (47073), UAS-Crz^IR (30670), UAS-InR^{I R}(999)

The following was from Exelixis at Harvard Medical School: sNPF⁰⁰⁴⁴⁸ (c00448)

The following were kind gifts: AstA¹-GAL4 (David Anderson), dILP2-GAL4 (Eric Rulifson), hug-GAL4 (Michael Pankratz), NPF-GAL4 (Ping Shen), UAS-sNPR^IR (Kweon Yu), Crz::mCherry (Gábor Juhász), UAS-hid::UAS-rpr (Tina Mukherjee), UAS-Shibire^ts (Toshihiro Kitamoto), tsh-GAL80 (Julie Simpson), UAS-preproANF::GFP (Edwin Levitan), UAS-ChR2XXL (Robert Kittel and Georg Nagel)

The following were previously generated in our laboratory: itpr^ka1091, itpr^ug3, UAS-Orai^E180A, UAS-itpr⁺,UAS-Stim, UAS-Ral^WT.

3^rd instar larvae (10 total) were collected and placed on cotton wool soaked with solution of 4.5% dissolved yeast granules and 0.5% Erioglaucine (Sigma, 861146). Controls contained no dye. Feeding was allowed for 2 hours at 25°C. 5 larvae per tube were crushed in 100μL of double distilled water. Solution was spun at 14000 rpm for 15 minutes and 50μL was withdrawn for absorbance measurement at 625nm in a 96-well plate. 5μL was used to measure protein content using the Pierce BCA Protein Assay kit (#23227).

RNA was isolated from 12–15 larval brains at the specified time points using Trizol. cDNA synthesis was carried out as described [23]. All mRNA levels are reported as fold change normalized to rp49. Primer sequences:

rp49, F:CGGATCGATATGCTAAGCTGT, R:GCGCTTGTTCGATCCGTA.

Crz, F:TCCTTTAACGCCGCATCTCC, R:CGTTGGAGCTGCGATAGACA

CrzR, F:CTGTGCATCCTGTTTGGCAC, R:GGCCTTGTGTATCAGCCTCT

Early third instar larvae were transferred to either normal or NR food. After 24 hours, larvae were recovered and immobilized on double sided tape. UV light from a Hg-arc lamp was focused using the UV filter, on the larvae through a 10X objective on Olympus BX60, for 2 minutes. Larvae were them immediately dissected in ice-cold PBS, mounted in PBS and imaged using Olympus FV-3000 Confocal microscope using a 40X objective and high-sensitivity detectors. Microscope settings for laser intensity, PMT settings and magnification were kept identical for all measurements. Each experiment always had a no UV control, in which larvae were subject to immobilisation but not UV light. Fluorescence intensity was calculated for each cell body using Image J.

For expression patterns, 3rd instar larval brains with RGs attached were dissected in ice-cold PBS and fixed in 3.7% formaldehyde at 4°C for 20mins. The samples were washed 4 times in PBS and mounted in 60% glycerol. Endogenous fluorescence was acquired on Olympus FV-3000 using a 20X, 40X or 60X objective, and processed used ImageJ. For samples requiring antibody staining brains were similarly processed and then subjected to permeabilisation (0.3% Triton X-100 + PBS; PBSTx) for 15 mins, 4 hr blocking in 5% normal goat serum in PBSTx at 4oC, followed by overnight incubation in primary antibody (1:1000 Chicken-GFP, Abcam: ab13970) and secondary with Alexa 488 or Alexa 594 (1:400; Abcam). For corazonin (1:1000; raised in Rabbit; Jan Veenstra, University of Bordeaux), all the above steps remained the step, except that dissected brains were fixed for 1hr at RT in 4% PFA and the secondary was anti-rabbit Alexa 405 (1:300, Abcam). Cell bodies were outlined manually and integrated density was used to calculate CTCF (Corrected Total Cell Fluorescence). For all samples, a similar area was measured for background fluorescence.

Ring glands were dissected in cold HL3.1 and transferred to a MALDI plate as previously described [24]. 0.2 μl of matrix (saturated solution of recrystallized α-cyano-4-hydroxycinnamic acid in MeOH/EtOH/water 30/30/40% v/v/v) was added, containing 10 nM of stable isotope-labeled HUG-pyrokinin (HUG-PK* (Ser–Val[d8]–Pro–Phe–Lys–Pro–Arg–Leu–amide, Mw = 950.1 Da; Biosyntan, Berlin, Germany)) and 10 nM labeled myosuppressin (MS* (Thr–Asp–Val[d8]–Asp–His–Val–Phe–Leu–Arg–Phe–amide, Mw = 1255.4 Da; Biosyntan) MALDI-TOF mass spectra were acquired in positive ion mode on a 4800 Plus MALDI TOF/TOF analyzer (MDS Sciex, Framingham, MA, USA) in a mass range of 900–4000 Da and fixed laser intensity with 20 subspectra and 1000 shots per sample. Data were analyzed with Data Explorer 4.10. Spectra were baseline corrected and de-isotoped. The sum of the resulting relative intensities of the de-isotoped peaks was calculated for the different ion adducts (H⁺, Na⁺, K⁺) of each peptide as well as the labeled peptides*. Then, the ratios sNPF/HUG-PK* and corazonin/MS* were calculated, using the labeled peptide with the most similar molecular weight. For sNPF, all isoforms (1/2-short, 1-long, -3 and -4) variants were totaled.

For thermogenetic (dTrpA1, Shibire^ts) experiments, larvae were matured to 88 hours AEL at 25°C. After transfer to either NR or normal food, vials were placed at 22°C, 25°C or 30°C for either 24 hours (dTrpA1) or till the end of observation time (Shibire^ts). For optogenetic experiments (Chr2-XXL), larvae were matured to 88AEL in the dark. After transfer to either NR or normal food, one set was placed in the dark while another was placed in an incubator with regular white lights that were on continuously till the end of observation time.

Collectively, more than 20 different NPs are known to be made by the neuroendocrine cells in which reducing SOCE components resulted in poor pupariation upon NR [22]. To shortlist specific NPs important for this paradigm, we undertook a curated GAL4-UAS screen. NP-GAL4s were used to drive the knockdown of IP₃R (IP₃R^IR) [25], and pupariation of the resulting larvae were scored on normal vs NR media (S1A Fig). On normal food, a significant reduction of pupariation was seen only with sNPF-GAL4 (S1A Fig), whose expression strongly correlates with neurons producing sNPF [26]. Upon NR, the largest effect was seen with sNPF-GAL4, followed by small but significant pupariation defect with AstA-GAL4 and DSK-GAL4 (S1A Fig). Neurons that secrete NPs may also secrete neurotransmitters, therefore, a role specifically for sNPF was tested. Reducing the level of sNPF (sNPF^IR) or reducing an enzyme required for neuropeptide processing (amontillado; amon^IR) [27] in sNPF-GAL4 expressing cells, as well as a hypomorphic sNPF mutation (sNPF⁰⁰⁴⁴⁸) resulted in impairment of larval development upon NR (S1B Fig). These data indicate that sNPF is required for pupariating under NR conditions.

sNPF-GAL4 expresses in large number of neurons (>300) in the larval brain [26] (S1C Fig), and also expresses in the larval midgut and epidermis. To further refine sNPF⁺ neurons on which we can perform cellular investigations, we tested a Crz-GAL4 driver. This driver expresses in fewer neurons (~22), all of which express the neuropeptide Corazonin (Crz). Importantly, a small subset of these, three bilateral neurons in the brain lobe, make Crz and sNPF. [26] (Fig 1A). Reducing SOCE in Crz neurons, by reducing either IP₃R or STIM (dSTIM^IR) [15,16] or over-expressing a dominant-negative version of Orai (Orai^E180A) [23], resulted in reduced pupariation on NR (Fig 1B). The absence of a developmental defect on normal food suggests that SOCE in these neurons is primarily required to survive NR.

To test if both sNPF and Crz were required, they were specifically reduced (sNPF^IR; Crz^IR) in Crz neurons. Knockdown of either NP resulted in larvae with a pupariation defect on NR media but not, on normal food (Fig 1C; S1D Fig). In Drosophila neurons, enhancing the expression of SOCE regulators leads to increased SOCE [16]. To test the positive effect of SOCE on Crz and sNPF, a genetic compensation experiment was carried out. The SOCE-regulators, IP₃R or dSTIM were over-expressed in Crz neurons which also expressed reduced levels of either sNPF or Crz. NR larvae with this genetic make-up showed a significant improvement in pupariation on NR media, as compared to NR larvae with only reduced NPs (Fig 1C). Interestingly, the compensation was sufficient to also increase the number of adults that emerged (Fig 1D). Notably, over-expression of either of the two SOCE molecules, dSTIM and IP₃R on their own, did not affect pupariation on either normal or NR media (Fig 1C), but unlike on normal food (S1E Fig), did reduce development to adulthood on NR media (Fig 1D). These data underscore the sensitivity of Crz neurons to ER-Ca²⁺ homeostasis during NR.

Loss of IP₃R and sNPF has previously been shown to affect larval feeding [28–30]. Hence, larval intake of dye-colored food in a 2-hour span was measured. Age-synchronized larvae with knockdown of either dSTIM, IP₃R, Crz or sNPF in Crz neurons exhibited no difference in the amount of dye ingested (S1F Fig), suggesting that developmental defects in the NR assay do not arise from a fundamental feeding problem.

Altogether, these genetic experiments helped identify a set of NP expressing neurons, the Crz neurons, a subset of which also express sNPF, in which SOCE plays an important role during development on NR.

In the larval CNS, Crz is expressed in 3 pairs of DLPs (Dorso Lateral Peptidergic neurons) in the pars lateralis region of the brain lobes, 1 pair of neurons in dorso-medial region and 8 pairs of interneurons in the VG (ventral ganglion) [31]. Other than the dorso-medial neurons, the Crz-GAL4 used in this study recapitulates the known expression pattern for Crz. (S2A and S2B Fig; Cartoon: Fig 2A). Additionally, adjacent to DLPs, low expression of Crz-GAL4 was observed in 3–4 neurons that do not express Crz (S2A and S2C Fig). As mentioned previously, Crz⁺ DLPs co-express sNPF [26]. In terms of neuronal architecture, the DLP neurons have two major branches: the anterior branch culminates in a dense nest of neurites at the ring gland (RG), while the posterior branch terminates in the subesophageal zone (SEZ). The VG neurons form a network amongst themselves to ultimately give rise to two parallel bundles that travel anteriorly, and end in the brain lobes. To visualize the overall distribution of NPs in the Crz neurons, we ectopically expressed a rat neuropeptide coupled to GFP (ANF::GFP), a popular tool used to track NP transport and release in Drosophila [32] (S2D Fig). Firstly, within the DLPs, like Crz::mcherry (S2A Fig), ANF::GPF was either in the cell bodies or RG projections, but not in the projections terminating at the subesophageal zone (S2D Fig), suggesting selective NP transport to the RG, which is a major neurohaemal site for systemic release of neuropeptides. Secondly, ANF::GFP intensity was higher in the cell bodies of the DLPs than VG neurons (S2D Fig).

The close proximity of the terminal projections of the Crz⁺ VG neurons and the anterior branch of the Crz⁺ DLP neurons in the brain lobe suggested possible neuromodulation between the two sets of neurons. Therefore, we undertook experiments to distinguish the contribution of DLPs vs VG localized Crz neurons, to the development in NR media. First, we utilized tshGAL80 to restrict Crz-GAL4 expression to the DLPs (S2E Fig). The level of pupariation under NR conditions observed with restricted expression of dSTIM^IR (Fig 2B; Mean: 51%±4.2) was similar to that seen with full expression (Fig 1B; Mean: 40.7%±13.3), suggesting a major contribution of the DLP neurons to the NR phenotype. Furthermore, sNPF>Crz^IR larvae have levels of pupariation of NR larvae (S2F Fig; Mean: 30.9%±7.8) similar to Crz>Crz^IR NR larvae (Fig 1C; Mean: 33.8%±5.9). Because sNPF-GAL4 marks only the Crz⁺ DLP neurons and not the Crz⁺ VG neurons (S1C Fig), this too suggests a major role for the Crz⁺ DLP neurons.

Requirement of SOCE in Crz neurons for pupariation on NR (Fig 1C) suggested that these neurons experience elevated cytosolic Ca²⁺ in NR conditions and are therefore, stimulated by chronic starvation. To test this, the UV light-activated genetically encoded calcium sensor, CaMPARI [33], was utilised. The sensor fluoresces in the GFP range (F₄₈₈) and is converted irreversibly to fluoresce in the RFP range (F₅₆₁), when exposed to UV light and in the presence of Ca²⁺. The level of conversion positively titrates with Ca²⁺ concentrations. Larva expressing CaMPARI in Crz⁺ neurons were placed in either normal or NR media for 24 hours (24h NR). Whole larvae were immobilized, and exposed to UV light for 2mins. Control larva were subject to the same treatment but, without being exposed to UV light (Fig 2C and 2D; no UV). Detection of F₅₆₁ in the fed state suggests that these neurons are active even under normal food conditions (Fig 2C and 2D). Notably, after 24 hours on NR media, chronic starvation caused a ~2-fold increase in average levels F₅₆₁ and therefore, of neuronal activation (Fig 2D). F₅₆₁/F₄₈₈ ratios did not appear to change in the VG neurons (S2G and S2H Fig). While there is a possibility that VG neurons do not exhibit higher F₅₆₁ because of insufficient penetration of UV light, the CaMPARI results together with the genetic experiments (Fig 2B, S2F Fig), formed the basis for selecting the DLP neurons for further analysis on how dSTIM affects Crz and sNPF.

Crz peptide levels were measured in DLP neurons by staining larval brains with an antiserum raised against the mature Crz peptide sequence [34]. Two locations on the DLP neurons were chosen for measurement: neuronal cell body/soma and neurite projections on the RG. In control DLP neurons, 24 hrs of NR caused average levels of Crz levels to increase, in both locations (Fig 3A, S3A and S3B Fig). In comparison, DLP neurons expressing dSTIM^IR displayed increased Crz peptide levels on normal food itself, and this remained unaltered upon NR, for both locations (Fig 3A, S3B and S3C Fig). sNPF levels could not be similarly measured by immunofluorescence because sNPF is expressed in many neurons close to the DLPs (S1C Fig), making measurements specifically from the DLP soma difficult to quantify. Instead, semi-quantitative, direct, mass spectrometric profiling of dissected RGs was employed. This technique can measure peptide levels relative to stable isotopic standards at single neurohaemal release sites [35]. As Crz levels between the cell bodies and projections correlated, and Crz⁺ DLPs are the sole contributors of sNPF on the RG [26], this technique allowed us to infer sNPF levels in DLPs. In controls, 24hrs of NR, increased the average level of sNPF ~5-fold on the RG (Fig 3C). In comparison, RG preparations from larvae where DLPs express dSTIM^IR, displayed increased sNPF levels on normal food itself, and this remained unaltered upon NR (Fig 3C). Although Crz was detected in the RG preparations, it was of much lower intensity. Average Crz levels increased with NR in the control, and in dSTIM^IR condition, but statistically higher levels of Crz were seen only in the NR, dSTIM^IR condition (S3D Fig). Nonetheless, broad agreement in trends, between Crz using immunofluorescence and sNPF using MALDI-MS, suggest that the two peptides are similarly regulated by NR and dSTIM. This is consistent with genetic experiments which showed that over-expression of dSTIM can rescue loss of both, sNPF as well as Crz (Fig 1C and 1D).

Thus, increased activation of DLP neurons by NR (Fig 2D), appears to result in peptide accumulation. Loss of dSTIM increases peptide levels on normal food, and prevents an increase in peptide levels upon NR.

As an ER-Ca²⁺ sensor, dSTIM may potentially regulate several cellular processes that would affect NPs such as their synthesis, processing, trafficking and/or release. As STIM^IR increased peptide levels in the cell body as well as neurite projections on the RG, a major trafficking defect was unlikely (Fig 3A vs S3B and S3C Fig). This does not rule out a role for dSTIM in dense-core vesicle trafficking, but merely indicates that trafficking of Crz is not observably disrupted by STIM^IR. We therefore proceeded to examine systemic and cellular phenotypes when molecules known to reduce overall NP synthesis (InR^IR; Insulin Receptor) [36], peptide processing (amon^IR) [27], and vesicle exocytosis (Ral^DN) [17] were expressed in Crz neurons. All three perturbations caused a pupariation defect on NR media (Fig 3D). However, despite similar systemic outcomes, amon^IR, which reduces the prohormone convertase required for peptide maturation, reduced Crz levels (Fig 3E). Because this is not seen with STIM^IR, a role for dSTIM in peptide processing was not pursued further.

Expression of InR^IR caused a modest increase in Crz peptide levels in DLP neurons in the fed state and peptide levels did not increase as in the control, in NR media (Fig 3F). InR is a global protein synthesis regulator, and its expression scales DIMM⁺ NE cell size, with functional consequences [36]. As the Crz⁺ DLPs are DIMM⁺ [37], we expected InR^IR to reduce, not increase peptide levels. A potential explanation for this observation is that Crz peptide levels are under feedback regulation, which was substantiated when we examined how Crz transcript and peptide levels are connected. Similarities between InR^IR and STIM^IR phenotypes, coupled with a previous observation that IP₃R, another SOCE component, positively regulates protein synthesis in Drosophila neuroendocrine cells [22], prompted us to test if dSTIM too regulates protein synthesis in general. We ectopically expressed a physiologically irrelevant neuropeptide construct (ANF::GFP), that yields a processed peptide in Drosophila neurons [32]. ANF::GFP levels in control and dSTIM^IR DLP neurons were similar (S3E Fig), suggesting that dSTIM does not have generic effects on peptide synthesis. Instead, its effect on Crz and sNPF synthesis may be specific. The lack of Crz elevation upon NR, in DLP neurons where InR^IR is expressed, leads to the speculation that InR signalling is required for the up-regulation of protein synthesis needed for increased peptide synthesis, processing and packaging during NR.

We previously found that Ral expression lies downstream of dSTIM-mediated SOCE in Drosophila pupal brains [38], and in dopaminergic neurons, over-expression of Ral^DN reduces secretion of ANF::GFP [17]. These previous data, coupled with the observation that Ral^DN and dSTIM^IR show similarly high Crz levels in the fed state, suggest that dSTIM affects vesicle secretion through regulation of Ral expression. To independently validate if vesicle release is important in Crz neurons, we over-expressed a temperature sensitive dynamin mutant (Shibire^ts) also shown to reduce NP release [39] and tetanus toxin (TNT) shown to prevent release of eclosion hormone, a neuropeptide [40]. Both manipulations caused a pupariation defect on NR (S3F Fig). It is unclear why Ral^DN causes Crz levels to decrease upon NR. In Drosophila pacemaker neurons, Ral has been shown to bias the sensitivity of a neuropeptide receptor, the Pigment Dispensing Factor Receptor [41]. Perhaps, functions distinct from Ral’s contribution to vesicle exocytosis contribute to this observation. Nonetheless, the lack of increase in Crz levels in DLP neurons upon NR, when InR^IR and Ral^DN are expressed, assumes significance in the context of pupariation of NR larvae. Control larvae subject to 24hrs of NR, display increased Crz (Fig 3A) and sNPF (Fig 3C) levels, and then proceed to successfully complete development to pupae (Fig 1C). Whereas, in dSTIM^IR, InR^IR and Ral^DN conditions, on NR, neither do DLPs display increased Crz levels (Fig 3A and 3F), nor do all larvae pupariate (Figs 1B, 1C and 3D). Thus, an increase in peptide levels on NR correlates with larval ability to pupariate on NR. Taken in context with increased neuronal activation during NR (Fig 2D), and evidence that functional vesicle exocytosis (Fig 3D, S3F Fig) as well as adequate peptide production (Fig 3D) is required for survival on NR, these data suggest that increased production and release of Crz and sNPF during NR, is required for NR larvae to successfully complete development.

To prove that Crz and sNPF are released during NR, the ideal experiment would be to measure levels of secreted NPs. But small size (8–10 amino acids), low hemolymph titres and high complexity of hemolymph, make peptide measurements by biochemical means highly challenging in Drosophila. Moreover, NPs exhibit endocrine as well as paracrine signalling [42]; and the latter will not be reflected in hemolymph measurements. Fortunately, in the Crz signalling system there is feedback compensation between secreted Crz and Corazonin receptor (CrzR) mRNA levels, providing a means to gauge secreted Crz levels indirectly. In adults, expression of Crz^IR in Crz neurons, increased levels of CrzR on the fat body [43]. We thus tested if CrzR, which in larvae appears to be expressed in the salivary glands and CNS [44], was subject to similar feedback. In larval brains, reducing Crz, using two different Crz^IR strains, not only caused a reduction in Crz mRNA levels (Fig 3G), but also a concomitant increase in CrzR mRNA levels (Fig 3H). Conversely, reducing CrzR levels (CrzR^IR) in the larval CNS, results in up-regulation of Crz mRNA (S3G Fig). This confirmed the existence of feedback in the Crz signalling system, and the use of neuronal CrzR transcript levels as a measure of secreted Crz levels. In line with this inference, we observed an up-regulation of CrzR mRNA in larval brains where Crz neurons are expressing either InR^IR or amon^IR or Ral^DN (Fig 3I). Therefore, the observation that in the STIM^IR condition CrzR mRNA levels are high (Fig 3I), supports the idea that dSTIM function is necessary for the secretion of optimal levels of Crz.

Because dSTIM-mediated SOCE is known to induce changes in gene expression [38], we probed if Crz expression is sensitive to NR and dSTIM. In the control, NR did not change Crz mRNA levels (Fig 3J), suggesting that a post-transcriptional mechanism is responsible for increasing Crz peptide levels upon NR (Fig 3A, S3B and S3C Fig). In the STIM^IR condition, Crz transcript levels were up-regulated on normal food conditions (Fed) and no further increase was observed upon NR (Fig 3J). The straightforward explanation for high Crz peptide levels in STIM^IR condition could therefore be attributed to higher gene expression of Crz. However, data from other perturbations in Crz neurons suggested that a linear interpretation between Crz mRNA and peptide levels cannot not be made. When Crz^IR is expressed, Crz mRNA is reduced (Fig 3G), but peptide levels are elevated (S3I Fig); whereas, when amon^IR is expressed, Crz mRNA is increased (S3H Fig), but peptide levels are decreased (Fig 3E). Meanwhile, in three conditions, Crz mRNA as well as peptide levels are higher than controls: CrzR^IR (S3G vs S3I Fig), InR^IR (S3H vs S3F Fig) and Ral^DN (S3H vs S3F Fig). Note also that both higher (dSTIM^IR, InR^IR, Ral^DN) or lower (amon^IR) Crz peptide levels in DLP cell bodies, result in reduced systemic Crz signalling (CrzR mRNA levels; Fig 3I). These data indicate that Crz transcription, translation and release are independently regulated. A simple explanation for elevated levels of Crz transcript as well as peptide levels in STIM^IR is therefore, feedback compensation. Moreover, there is no change in Crz mRNA upon 24hrs of NR, when STIM^IR is expressed (Fig 3J). Together, this argues against a direct role for dSTIM in regulating Crz gene expression.

In summation, these data have been inferred as follows: on normal food, partial loss of dSTIM reduces systemic Crz signalling, indicating a requirement for dSTIM in Crz secretion. On NR media, Crz⁺ DLPs are stimulated to increase peptide synthesis and release, in order for NR larvae to complete development. Peptide up-regulation upon NR is abrogated when dSTIM is reduced. These add up to suggest that dSTIM compromises NE cell function in a manner that affects peptide synthesis and release, with functional consequences for survival on NR.

To validate a role for dSTIM in peptide synthesis and release, we tested genetic perturbations that can compensate for this deficiency, to rescue developmental and cellular phenotypes associated with dSTIM^IR expression in Crz neurons. In the case of NPs, genetic over-expression may not translate to enhanced release, as proteins involved in NP processing as well as the regulated secretory pathway would need to be up-regulated. Furthermore, regulatory feedback from peptides to their transcription may complicate over-expression, as seen for the Crz signalling system (S3I Fig). To get around these issues, and because InR^IR phenocopied STIM^IR (Fig 3A, 3D, 3F and 3I and S3H Fig), we opted to increase protein synthesis by over-expression of the Insulin receptor (InR). Cell size (S4A Fig) as well as Crz levels (S4B and S3C Figs) in DLP neurons scaled with InR over-expression, supporting the effectiveness of InR. To increase release, we over-expressed Ral^WT as Ral over-expression can compensate for vesicle release in dopaminergic neurons expressing STIM^IR [17]. In Crz neurons with reduced dSTIM, over-expression of either InR or Ral^WT rescued pupariation on NR media (Fig 4A); restored peptide up-regulation upon NR (Fig 4A and 4B) and decreased CrzR mRNA back to control levels (Fig 4D). Of note, over-expressing Ral^WT or InR by itself, in Crz neurons, did not alter CrzR mRNA levels (Fig 4D), suggesting that neuronal activation, which happens on NR media (Fig 2D) potentiates their activity.

To increase neuronal activity, we utilised the temperature and voltage-gated cation channel, TrpA1 [45]. Over-expression of dTrpA1 and its activation by raising the temperature to 30°C for 24 hours, in the dSTIM^IR background, rescued pupariation of NR larvae (Fig 4E). It also restored the ability of DLP neurons to increase Crz levels upon NR (Fig 4F) and decreased levels of CrzR mRNA (Fig 4G). In line with the feedback between CrzR mRNA levels and systemic Crz signaling, over-expression of dTrpA1 alone in Crz neurons resulted in a decrease in CrzR mRNA levels (Fig 4G), supporting the role for neuronal activation in secreting Crz. Interestingly, development to adulthood on NR, for dSTIM^IR larvae, was also significantly increased upon over-expression of InR (S4E Fig) and TrpA1 (S4F Fig), but not Ral^WT (S4E Fig).

An optogenetic approach, utilizing the over-expression of Channelrhodopsin (ChR2-XXL), a light-sensitive cation channel, also rescued pupariation but not to the same extent (S4D Fig). Poorer rescue with ChR2-XXL could be because sustained activation of this channel depress synaptic transmission and the channel is less conductive for Ca²⁺ compared to TrpA1 [46]. Similar genetic manipulation, with TrpA1 and Chr2-XXL, in a hypomorphic IP₃R mutant (itpr^ku) resulted in a small but significant rescue of itpr^ku pupariation on NR media (S4G and S4H Fig).

Together, these results strongly suggested that defects arising from dysregulated intracellular Ca²⁺ signalling, may be overcome by increasing vesicle exocytosis (Ral^WT, TrpA1, ChR2-XXL rescue) or protein synthesis (InR rescue). Importantly, the rescues observed with InR, Ral^WT and dTrpA1 are effective at the molecular (CrzR levels), cellular (Crz peptide levels upon NR) as well as systemic (NR larvae) level.