Identification of NSC in the Drosophila brain.

(A) Schematic drawing of the different types of NSC and their projections to different release sites within the fly. Based on (Nässel et al., 2013). (B) All NSC projections exit the brain via the nervi corpora cardiaca (NCC). Electron micrograph showcasing a cross section of the NCC. Scale bar = 750nm. (C) Cosine similarity matrix of all NSC in the FlyWire connectome based on their total inputs. The darker the color, the higher the similarity between neurons. Neurons within the clades are colored based on the schematic in (D). (D) Reconstructions of the 80 NSC within the adult brain connectome. (E) Classification of NSC based on their location and neuropeptide expression. Refer to Table 1 for further details. Abbreviations: SEZ, subesophageal zone; CC, corpus cardiacum; CA, corpus allatum.

Classification of Drosophila neurosecretory cells (NSC) based on their cell body position in the central brain.

Quantification of NSC.

(A) On average, there are 16 m-NSCDILP as labelled by DILP3-Gal4 and DILP2 antibody (marked with asterisks). Note that some preparations contain 18 m-NSCDILP, in agreement with the number determined based on the FlyWire connectome. (B) DMS-T2A-Gal4 drives GFP expression in several neuronal populations across the brain. Regions imaged in independent samples in panels (C) and (D) are indicated using a dashed box. (C) DMS-T2A-Gal4 drives nuclear mCherry expression in about six m-NSC in the pars intercerebralis. (D) DMS-T2A-Gal4 drives weak GFP expression in the pair of SEZ-NSCCAPA (indicated with filled arrowheads). (E) CRZ is expressed in 7 pairs of neurons in adult flies, 4 of which co-express Gr64a (empty arrowheads). These smaller Gr64a-expressing CRZ neurons form dense arborizations in the lateral horn. They project contralaterally but do not send projections via the nervii corpora cardiaca (NCC) and are thus not considered neurosecretory.

Synaptic inputs to NSC in the FlyWire connectome.

(A) Postsynaptic sites of different NSC classes. The majority of the dendrites are found in the protocerebrum and SEZ. (B) Input to NSC grouped by the neuronal super classes annotated in the FlyWire connectome. Central neurons are the largest group providing inputs to NSC. (C) Proportion of inputs from various neuronal super classes to different NSC classes. (D) Reconstructions of neurons from different super classes which provide major inputs to NSC. Only the top 10 cell types per super class are shown. (E) Number of strong input connections (greater than 50 synapses) to each NSC class and the total number of synapses constituting these connections. (F) Reconstructions of neurons that provide major inputs (more than 50 synapses per connection) to SEZ-NSCCAPA, l-NSCDH31, m-NSCDH44 and m-NSCDILP. (G) 42 unique cell types (76 neurons) provide inputs to multiple NSC classes. The line width is scaled according to the proportion of synaptic input each of these cell types provides to different NSC classes. m-NSCDH44 and m-NSCDILP receive the majority of these inputs. Out of these 76 neurons, (H) 53 neurons provide inputs to two NSC classes, (I) 22 neurons provide inputs to three classes of NSC and (J) 1 neuron provides input to four NSC classes. Reconstructions of corresponding neurons next to each schematic. For panel (E), bars have been color coded according to the legend in the panel (A).

Sensory inputs to NSC in the FlyWire connectome.

(A) Direct and indirect (disynaptic) sensory inputs to NSC. Interneurons mediating connectivity between sensory neurons and NSC are referred to as sensory interneurons. The donuts represent the proportion of cells and the number in the donut reflects the total number of neurons in that group. NSC receive very minimal direct sensory inputs. Gustatory, mechanosensory and enteric neurons provide the majority of the monosynaptic and disynaptic inputs to NSC. Note that l-NSCITP do not receive any significant synaptic inputs and are thus not represented here. (B) Reconstructions of sensory neurons (separated by class) providing direct inputs to NSC. (C) Reconstructions of sensory neurons providing indirect inputs to NSC. (D) Number of sensory neurons (grouped by sub class) that provide indirect inputs to NSC. (E) Schematic showing the projections of labellar and tarsal gustatory receptor neurons (GRN) from the periphery to the SEZ (adapted from Freeman and Dahanukar, 2015). Reconstructions of four tarsal GRN (colored red; classified as ascending neurons on Codex) that provide indirect inputs to six NSC (also shown). Abbreviations: acc. pharyngeal, accessory pharyngeal; mech, mechanosensory; un, unknown sensory.

Olfactory inputs to NSC.

(A) Schematic showcasing the flow of olfactory information from olfactory receptor neurons (ORN) in the antenna to the higher-order brain centers (e.g. mushroom bodies and lateral horn) via the antennal lobe (adapted from Zhao and McBride, 2020). (B) Number of neurons (grouped by different categories) that comprise the shortest pathway from ORN to NSC. ORN have been grouped based on their behavioral significance (based on Zheng et al., 2022). Antennal lobe associated neurons (AL*) include projection neurons (ALPN) and local interneurons (ALLN). IN represent interneurons that link AL* neurons and NSC. (C) Numbers of each ORN type that provide indirect inputs to NSC. (D) Number of synapses formed by these ORN. Note that the ORN which detect aversive odors followed by those that detect food odors provide the strongest indirect inputs to NSC. (E) Reconstructions of top ten ORN types. (F) Number of AL* in the pathway. v2LN30 is the only ALLN whereas the rest are ALPN. (G) Reconstructions of top four ALPN types and (H) top three NSC classes that are part of this pathway. Bars in (C) and (D) and neurons in (E) and (G) have been colored based on their behavioral significance. (I) Pheromonal and egg-laying associated olfactory information is relayed to m-NSCDILP. (J) ORN belonging to all five behavioral categories provide inputs to l-NSCDH31. (K) SEZ-NSCCAPA primarily receive aversive olfactory inputs. For I-K, the numbers within the circles indicate the number of neurons or the name of that neuron. Arrows have been weighted based on the number of synapses and colored based on the neurotransmitter mediating those connections (see legend). Abbreviations: LN, local interneuron; uni. PN, uniglomerular projection neuron; multi. PN, multiglomerular projection neuron; KC, Kenyon cell; LHON, lateral horn output neuron.

Influence of in silico sensory neuron stimulation on NSC in the FlyWire connectome.

(A) A schematic of the linear dynamical modeling-based approach (Bates et al., 2026) used to calculate the ‘influence’ of sensory source cells on target NSC. (B) Mean influence of different sensory neurons (rows) on NSC classes (columns). Note the strong influence of enteric neurons on NSC, especially m-NSCDMS. (C) Mean influence of different sensory neurons (rows) on major cell types in the brain, including NSC. (D) The mean influence in panel (C) was used to determine the rank of influence on NSC in relation to other cell types. Enteric neurons, especially ENS4, have the strongest influence on medial NSC.

Synaptic output from NSC in the FlyWire connectome.

(A) Presynaptic sites of different NSC classes. (B) Output from NSC grouped by the neuronal super classes annotated in the FlyWire connectome. Central neurons receive inputs from l-NSCunknown and descending neurons receive inputs from l-NSCCRZ. (C) Proportion of outputs from different NSC classes to various neuronal super classes. (D) Reconstructions of l-NSCunknown and all their postsynaptic partners. (E) Reconstructions of l-NSCCRZ and all their postsynaptic partners (DNg27 descending neurons) in the FlyWire connectome. DNg27 reconstructions in the maleCNS connectome show that these neurons innervate the wing tectulum. (F) Weighted connections between l-NSCCRZ and DNg27 descending neurons which innervate wing power motor neurons in the wing tectulum and could thus regulate flight. (G) Individual postsynaptic partners of l-NSCunknown and l-NSCCRZ sorted based on the number of synapses and colored based on their neurotransmitter identity.

Functional characterization of Corazonin (CRZ) and DNg27 descending neurons.

Chemogenetic silencing of CRZ neurons using Kir2.1 results in (A) reduced food intake, (B) increased starvation survival and (C) a reduction in the number of eggs laid. In contrast, silencing DNg27 neurons has no effect on (D) food intake and (E) starvation survival but also results in (F) a reduction in the number of eggs laid. In panels (B) and (E), lines represent mean percent survival curve, with shaded bands representing ± 1 standard error of the mean (SEM). Error bars in panels (C) and (F) indicate SEM. For panels (A) and (D), **** p < 0.0001, as assessed by one-way ANOVA followed by Tukey’s HSD for multiple comparisons. For panels (B) and (E), **** p < 0.0001, as assessed by log-rank (Mantel-Cox) test, followed by Bonferroni correction for multiple comparisons. For panels (C) and (F), *** p < 0.001, as assessed using a linear mixed effects model, followed by Tukey-adjusted pairwise comparisons.

Putative NSC interconnectivity and endocrine output.

(A) Identification of single-cell transcriptomes representing different NSC subsets in the adult brain (Davie et al., 2018). All NSC express genes required for neuropeptide processing and release (amon, svr, Pal2, Phm and Cadps) and were identified primarily based on the neuropeptides that they express. (B) Dot plot showing expression of receptors in NSC. Expression of only those receptors whose corresponding neuropeptides are expressed in NSC are shown. (C) Connectivity diagram (weighted based on neuropeptide and receptor expression) showing putative paracrine connectivity between different classes of NSC. Note that short neuropeptide F (sNPF) and myosuppressin (DMS) are expressed in two different NSC classes. Ion transport peptide and CAPA pathways are not included because their receptors were not detected in these transcriptomes. Leucokinin was excluded because its expression levels were below the threshold used here. Dot plot showing the expression of neuropeptide receptors in (D) adipokinetic hormone cells of the corpus cardiacum and (E) all the tissues in adults. “General” in panel (E) includes cell types that are found across multiple tissues including sensory neuron, visceral muscle and hemocytes amongst others. See Figure 9 Supplement 3 for all the different cell types that are part of this cluster.

Summary of main findings presented in the study in relation to broader context.

(A) Schematic showing the different classes of NSC in the FlyWire (or FAFB) connectome as well as the functions regulated by hormones released from these cells. Filled circle indicates that a hormone regulates that function. The right panel is adapted from Nässel, 2025. (B) Major synaptic inputs regulating NSC. With regards to food-related inputs, enteric neurons have a stronger influence on NSC, followed by peripheral GRN and then ORN. Other novel inputs to NSC include descending neurons which could help maintain homeostasis by coordinating physiology (through NSC) with behavior (through motor neurons). (C) Major synaptic output from NSC, focusing on l-NSCCRZ. l-NSCCRZ target DNg27 and both of these cell types regulate egg-laying. The role of these cells in flight is still unclear. Previous work has shown that l-NSCCRZ regulate feeding and metabolism via multiple targets (Kubrak et al., 2016; Oh et al., 2019). (D) Outstanding questions or future directions include sex-specific differences in connectivity of NSC, functional validation of paracrine interactions between NSC, identification of neuronal messengers in l-NSCunknown and m-NSCunknown, and comprehensive identification of NSC in the ventral nerve cord (VNC). Abbreviations: IN, inter neuron; ALPN, antennal lobe projection neuron; GRN, gustatory receptor neuron; ORN, olfactory receptor neuron; AN, ascending neuron; DN, descending neuron; APC, adipokinetic hormone-producing cells.

NSC in the brain and nerve cord (BANC) and male central nervous system (maleCNS) connectomes.

(A) Reconstructions of the 72 NSC within the BANC connectome. (B) Reconstructions of the 71 NSC within the maleCNS connectome.

Quantification of m-NSCDILP.

Quantification of m-NSCDILP (marked with asterisks) as labelled by (A) DILP2-Gal4 driven nuclear mCherry and DILP2 antibody, (B) DILP2-Gal4 driven nuclear mCherry and DILP3 antibody, and (C) DILP5-Gal4 driven nuclear mCherry and DILP2 antibody.

Differences between m-NSCDH44 and m-NSCDMS.

(A) Retrograde trans-synaptic labelling of m-NSCDH44. m-NSCDH44 are labelled in green and their presynaptic partners are labelled in magenta. Note the ectopic expression in the mushroom body which is also visible in the controls. These representative images are based on at least five independent samples. In silico retrograde tracing of (B) m-NSCDH44 and (C) m-NSCDMS in the FlyWire and maleCNS connectomes. Both of these NSC classes receive the majority of their inputs from neurons in the SEZ which have similar location and morphology. However, m-NSCDMS also receive inputs from a group of central neurons (marked with an arrow) that are not visible in (A) and (B). (D) Reconstruction of myosuppressin (DMS) descending neurons (DNp32 cell type) in the FAFB and maleCNS connectomes. Representative electron micrographs showing a cross section of (E) m-NSCDH44, (F) m-NSCDMS and (G) DMS descending neuron soma. Both types of DMS-expressing cells have darker dense core vesicles (marked by red arrows) compared to those found in m-NSCDH44. Quantification of DCV brightness for the three cell types (m-NSCDH44, m-NSCDMS and DNp32) across the (H) FlyWire, (I) maleCNS and (J) BANC connectomes. For panels (H-J), statistical significance was assessed by one-way ANOVA followed by Tukey’s HSD for multiple comparisons.

Retrograde trans-synaptic labelling of m-NSCDH44.

m-NSCDH44 are labelled in green and their presynaptic partners are labelled in magenta. Note the ectopic expression in the mushroom body. Presynaptic neurons in the subesophageal zone are consistently labelled across five independent samples.

Morphological characteristics of NSC.

(A) cable length, (B) surface area, (C) cell volume and (D) nuclei volume of different NSC classes. (E) Principal component analysis of these four features reveals that the NSC of a given class generally cluster together. Note the high variability for l-NSCCRZ, l-NSCDH31, m-NSCDH44 and m-NSCDILP populations, suggesting that they comprise morphologically heterogenous subpopulations.

Postsynaptic sites of NSC in the FlyWire connectome.

Reconstructions of different NSC classes along with their postsynaptic sites. l-NSCITP are an exception and have very few postsynaptic sites.

Inputs to NSC classes in the FlyWire connectome.

(A) Individual presynaptic partners of different NSC sorted based on the number of synapses. Presynaptic neurons are colored based on the super class they belong to. Only the top 20 neurons are shown. SEZ-NSCCAPA and l-NSCCRZ receive strong sensory inputs whereas l-NSCDH31, m-NSCDH44 and m-NSCunknown mostly receive inputs from central neurons. (B) Number of presynaptic neurons providing inputs to different NSC classes.

Synaptic inputs to NSC in the BANC and maleCNS connectomes.

(A) Input to NSC grouped by the neuronal super classes annotated in the BANC connectome. (B) Proportion of inputs from various neuronal super classes to different NSC classes in the BANC connectome. (C) Input to NSC grouped by the neuronal super classes annotated in the maleCNS connectome. (D) Proportion of inputs from various neuronal super classes to different NSC classes in the maleCNS connectome. Note that central brain intrinsic neurons followed by ascending neurons are the largest groups providing inputs to NSC across both connectomes.

Neurotransmitters providing inputs to NSC classes in the FlyWire connectome.

(A) Individual presynaptic partners of different NSC sorted based on the number of synapses and colored based on their neurotransmitter identity. l-NSCDH31 and m-NSCDH44 receive strong glutamatergic inputs. (B) Input to NSC grouped by the neurotransmitters. Out of the three fast-acting neurotransmitters, GABA provides the least inputs.

Neurons providing input to multiple NSC classes in the FlyWire connectome.

Proportion of inputs (based on the number of synapses) from individual neurons to different NSC classes. Each bar represents an individual neuron and it is filled according to the NSC classes that it provides input to. In total, 76 neurons provide inputs to more than one NSC class, with m-NSCDH44 receiving inputs from most of these neurons.

Classification of enteric neurons in the FlyWire connectome.

(A) Enteric neurons have been classified into five cell types (ENS1 to ENS5) on Codex. Reconstructions of ENS1 to ENS5 show that these are morphologically distinct cell types. (B) Cosine similarity matrix of all enteric neurons in the FlyWire connectome based on their total inputs and outputs. Darker red colors indicate higher similarity between neurons. Neurons within the clades are colored based on the cell types in (A). Note that not all cells belonging to ENS3 and ENS5 cell types cluster together, suggesting that they could represent heterogeneous populations based on synaptic connectivity.

Influence of in silico sensory neuron stimulation on NSC in the BANC connectome.

(A) Mean influence of different sensory neurons (rows) on NSC classes (columns). Note the strong influence of enteric neurons on NSC, especially m-NSCDMS. (B) Mean influence of different sensory neurons (rows) on major cell types in the brain, including NSC. (C) The mean influence in panel (B) was used to determine the rank of influence on NSC in relation to other cell types. Enteric neurons, especially ENS4 and ENS5, have the strongest influence on medial NSC.

Presynaptic sites of NSC in the FlyWire connectome.

Reconstructions of different NSC classes along with their presynaptic sites. l-NSCCRZ have several presynaptic sites in the subesophageal zone.

Synaptic output from NSC in the FlyWire connectome based on a low synaptic threshold.

(A) Proportion of outputs from different NSC classes to various neuronal super classes when the threshold for a significant connection is lowered to 2 synapses. (B) Output from NSC grouped by the neuronal super classes annotated in the FlyWire connectome. (C) Reconstructions of neurons receiving inputs from NSC. Cells belonging to the top four super classes are shown. Note that most of the output from NSC is to partial fragments and non-neuronal cells (undefined), as well as central neurons.

Synaptic output from NSC in the BANC and maleCNS connectomes.

(A) Output from NSC grouped by the neuronal super classes annotated in the BANC connectome. (B) l-NSCCRZ provides all of its output to descending neurons in the BANC connectome. (C) Output from NSC grouped by the neuronal super classes annotated in the maleCNS connectome. (D) Proportion of outputs from different NSC classes to various neuronal super classes in the maleCNS connectome.

Optogenetic activation of CRZ or DNg27 neurons in female flies results in minimal large-scale changes in free-flight kinematics.

(A) Schematic diagram of the wind-tunnel used for free-flight kinematic experiments, annotated with dimensions and the positions of cameras and LEDs. A schematic representation of the trigger volume, which activates the red LEDs upon entrance of a fly, is also indicated. (B) Still images of male empty-Gal4 and CRZ-Gal4 flies taken from videos during optogenetic stimulation. The images compare two related behaviors observed during optogenetic stimulation: (1) an abdominal curl demonstrated by the empty-Gal4 control fly (left), and (2) a copulatory abdominal curl with genital eversion by the experimental CRZ-Gal4 male (right). The extrusion of internal genital structures (yellow arrow) is evident in the zoomed-in portions. (C) Comparison of the number of observed abdominal behaviors in control (empty-Gal4 in white bars; N = 20) and experimental (CRZ-Gal4 in black bars; N = 20) male flies during optogenetic stimulation. Behaviors were counted over n = 5 x 15 s red LED flashes, with an inter-stimulus interval of 120 s of ambient-only illumination. (D) Course direction (radians) versus time relative to the fly entering the trigger volume for flash (top row) and sham (bottom row) trigger events. The shaded region denotes the period of red LED illumination during flash events. N = 60-78 flies for each genotype: CRZ-Gal4, flash n = 412 trajectories, sham n = 198 trajectories; empty-Gal4, flash n = 344 trajectories, sham n = 379 trajectories; DNg27-Gal4, flash n = 519 trajectories, sham n = 491 trajectories; empty-split-Gal4, flash n = 769 trajectories, sham n = 711 trajectories. A course direction of 0 indicates upwind and π/-π is downwind. (E) Horizontal ground speed velocity (m/s) versus time relative to the fly entering the trigger volume for flash (top row) and sham (bottom row) trigger events. The blue line represents the median groundspeed velocity over all trajectories. Fly and trajectory numbers are the same as in panel (D). (F) Altitude (m) versus time relative to the fly entering the trigger volume for flash (top row) and sham (bottom row) trigger events. The blue line represents the median altitude over all trajectories. Fly and trajectory numbers are the same as in panel (D).

Neurons labelled by CRZ-Gal4 and DNg27-Gal4 in males and females.

CRZ-Gal4 drives CsChrimson-mVenus expression in l-NSCCRZ and optic lobe neurons in both males and females. Additionally, it drives expression in neurons in the abdominal ganglion of the ventral nerve cord in males. DNg27-Gal4 (a split-Gal4 driver) drives weak CsChrimson-mVenus expression in a pair of descending neurons in both males and females.

Putative paracrine interconnectivity between NSC.

NSCs classes targeted by (A) myosuppressin (DMS), (B) Hugin, (C) corazonin (CRZ), (D) short neuropeptide F (sNPF), (E) diuretic hormone 31 (DH31), (F) tachykinin (TK), (G) diuretic hormone 44 (DH44) and (H) insulin-like peptides (DILPs). Ion transport peptide and CAPA pathways are not included because their receptors were not detected in these transcriptomes. (I) Dot plot showing the neuropeptides expressed in each NSC class following thresholding. The expression has been scaled and was used to generate the connectivity diagrams in Figure 9C and Figure 9 Supplement 1A-H.

Expression of receptors for hormones released from brain NSC.

t-SNE plots showing expression of hormone receptors across single-cell transcriptomes from all Drosophila tissues (Li et al., 2022). Note that some receptors such as InR and sNPF-R are broadly expressed whereas others such as CapaR and PK2-R1 are sparsely expressed.

Expression of hormone receptors in peripheral tissues.

Dot plots showing expression of hormone receptors in different tissues at single-cell resolution. Expression of only those receptors whose corresponding neuropeptides are expressed in brain NSC are shown.

Expression of hormone receptors in the gut and reproductive tissues.

Dot plots showing expression of hormone receptors in the gut and reproductive tissues at single-cell resolution. Expression of only those receptors whose corresponding neuropeptides are expressed in brain NSC are shown.

Fly strains used in this study.

Antibodies used for immunohistochemistry in this study.