A nutrient-specific gut hormone arbitrates between courtship and feeding

Abstract

Animals must set behavioural priority in a context-dependent manner and switch from one behaviour to another at the appropriate moment^1-3. Here we probe the molecular and neuronal mechanisms that orchestrate the transition from feeding to courtship in Drosophila melanogaster. We find that feeding is prioritized over courtship in starved males, and the consumption of protein-rich food rapidly reverses this order within a few minutes. At the molecular level, a gut-derived, nutrient-specific neuropeptide hormone-Diuretic hormone 31 (Dh31)-propels a switch from feeding to courtship. We further address the underlying kinetics with calcium imaging experiments. Amino acids from food acutely activate Dh31⁺ enteroendocrine cells in the gut, increasing Dh31 levels in the circulation. In addition, three-photon functional imaging of intact flies shows that optogenetic stimulation of Dh31⁺ enteroendocrine cells rapidly excites a subset of brain neurons that express Dh31 receptor (Dh31R). Gut-derived Dh31 excites the brain neurons through the circulatory system within a few minutes, in line with the speed of the feeding-courtship behavioural switch. At the circuit level, there are two distinct populations of Dh31R⁺ neurons in the brain, with one population inhibiting feeding through allatostatin-C and the other promoting courtship through corazonin. Together, our findings illustrate a mechanism by which the consumption of protein-rich food triggers the release of a gut hormone, which in turn prioritizes courtship over feeding through two parallel pathways.

Extended Data Figure 1.. Schematic illustration of gene structures and knock-in constructs

(a) A T2A-GAL4.2 cassette was inserted at the C-terminus of Dh31 by Cas9-mediated knock-in strategy. 3XP3-RFP was used as a positive selection marker. (b) An nls-LexA∷p65 cassette was inserted after the start codon of Dh31R by Cas9-mediated knock-in strategy. (c) An nls-LexA∷p65 cassette was inserted after the start codon of Crz by Cas9-mediated knock-in strategy.

Extended Data Figure 2.. Dh31 expression outside the brain is required for the effect of ingested amino acids on courtship

(a) Validation of the Dh31^T2A-GAL4 driver line by a Dh31 antiserum. Representative confocal images show the expression pattern of Dh31^T2A-GAL4 (Blue) and Dh31 (magenta) in the brain, ventral nerve cord (VNC) and midgut of male flies carrying Dh31^T2A-GAL4 and UAS-Redstinger. Schematic indicates the organization of Dh31 cells in the brain, VNC, and GI tract, drawn from the confocal stacks of the sample shown in the left panel. Magenta dots show cell positive for Dh31 antibody and Redstinger; black dots show cells positive for Dh31 antibody and negative for Redstinger. The average number (± SEM) of cells in each region is from 5 different male flies. Scale bar, 50 μm. (b-d) The courtship effect of region-specific Dh31 knockdown. Dh31 knockdown in all Dh31^T2A-GAL4 cells (b); Dh31 knockdown in brain Dh31^T2A-GAL4 cells (c); Dh31 knockdown in Dh31^T2A-GAL4 cells outside the brain (d). The total number of matches (n) is shown, with 5 experiments per condition and 12–24 successful matches per experiment. p-value was determined by Chi-square test to indicate whether males of a given genetic manipulation respond to amino acids. Blue bar indicates average copulation percentage, and the dashed line indicates chance level.

Extended Data Figure 3.. Dh31 expression in enteroendocrine cells is required for the effect of ingested amino acids

(a) Expression of Redstinger (blue) and Dh31 (anti-Dh31, magenta) in the brain, VNC and midgut of a male containing Dh31-GAL4 and UAS-Redstinger. Scale bar, 50 μm. Schematic indicates the organization of cell clusters in the brain, VNC and GI tract, drawn from the confocal stacks of the sample shown in the left panel. Magenta dots: Dh31 antibody⁺ and GAL4⁺ cells; black dots: Dh31 antibody⁺ and GAL4⁻ cells. The average number (± SEM) of cells in each region is from 5 different male flies. (b) Validation of Dh31 knockdown. Knockdown of Dh31 in the gut is achieved by using Dh31-GAL4 and Tsh-GAL80 to drive UAS-Dh31-RNAi. Immunostaining with an antiserum to Dh31 (magenta) shows that knockdown is effective in the gut. Scale bar, 50 μm. (c) The courtship effect of gut-specific Dh31 knockdown (left) or chemo-activation of gut Dh31⁺ cells (right). Chemo-activation is achieved by using Dh31-GAL4 and Tsh-GAL80 to drive UAS-VR1. The total number of matches (n) is shown, with 5–6 experiments per condition and 12–28 successful matches per experiment. p-value was determined by Chi-square test to indicate whether males of a given genetic manipulation respond to amino acids (left) or capsaicin (right). Blue bar indicates average copulation percentage, and the dashed line indicates chance level.

Extended Data Figure 4.. Expression of Dh31R in brain Crz⁺ neurons is required for the aphrodisiac effect of amino acids

(a) Validation of Dh31R knockdown. qRT-PCR results show that Dh31R expression level is reduced in flies having the Actin-GAL4 and UAS-Dh31R-RNAi transgenes. RNA was extracted from 5 male flies per experiment, with 6 experiments for each genotype. (b-d) Schematic indicates genetic strategies to perturb Crz⁺ neurons in the brain and VNC (b), only in the brain (c), and only in the VNC (d). Representative confocal images show the different intersections between Crz-GAL4 and otd-nls∷flp (corresponding genotypes are shown below the images). Samples were immuostained with anti-GFP (green), nc82 (blue). Scale bar, 50 μm. The courtship effect of Dh31R knockdown in both brain and VNC, in brain only or in VNC only. The total number of matches (n) is shown, with 5–6 experiments per condition and 8–26 successful matches per experiment. p-value was determined by Chi-square test to indicate whether males of a given genetic manipulation respond to amino acids. Blue bar indicates average copulation percentage, and the dashed line indicates chance level.

Extended Data Figure 5.. Expression of Crz in brain Crz⁺ neurons is required for the aphrodisiac effect of amino acids

(a) Validation of Crz knockdown. Immunostaining of anti-Crz (red) showed Crz is undetectable in brain Crz⁺ neurons in Crz-Gal4, UAS-Crz-RNAi flies. Scale bar, 50 μm. (b and c) The courtship effect of Crz knockdown in brain Crz⁺ neurons (b) or in VNC Crz⁺ neurons (c). The total number of matches (n) is shown, with 5 experiments per condition and 11–23 successful matches per experiment. p-value was determined by Chi-square test to indicate whether males of a given genetic manipulation respond to amino acids. Blue bar indicates average copulation percentage, and the dashed line indicates chance level.

Extended Data Figure 6.. Dh31 in the gut and Dh31R in brain Crz⁺ neurons are required for males to transition from feeding to courtship

(a, b) The effect of gut-specific Dh31 knockdown on feeding and courtship in fed (a) or starved (b) flies. The knockdown increased feeding duration and eliminated courtship when food contained amino acids, but had no effect when food contained only sucrose. GAL4: Dh31-GAL4, Tsh-GAL80. UAS: UAS-Dh31-RNAi. Gut Dh31 KD flies had all three transgenes. The fraction of time individual males spent on feeding and courtship in (a). n = 7 for each condition. ****, p < 0.0001, ANOVA followed by Tukey’s test. (c) Chemo-activation of gut Dh31-GAL4 cells. The ingestion of capsaicin-containing food induced courtship but food containing only sucrose had no effect. GAL4: Dh31-GAL4, Tsh-GAL80. UAS: UAS-VR1. Gut Dh31 VR1 flies had all three transgenes. (d) Dh31R knockdown in brain Crz⁺ neurons eliminated the effect of amino acids on courtship, but not feeding. GAL4: Crz-GAL4, tub>GAL80>stop. UAS: otd-nls∷flp, UAS-Dh31R-RNAi. Brain Dh31R KD flies had all four transgenes. The Raster Plots show the second-by-second behaviors of 5 representative males. Behaviors: feeding (blue), courtship (orange) and copulation (black).

Extended Data Figure 7.. Dh31 enteroendocrine cells respond to amino acids

(a) Response of Dh31⁺ cells to different nutrients at the indicated concentrations. Suc: sucrose, Fru + Glu: mixture of fructose and glucose, EAA: essential amino acids, NEAA: nonessential amino acids. The average calcium response (ΔF/F) is shown in black with SEM as gray shaded areas (n=6). (b) The response of individual Dh31⁺ cells to amino acids at different concentrations. Gray lines below the traces indicate the stimulation period of amino acids or saline control.

Extended Data Figure 8.. Effect of CO₂ exposure on heartbeat and the response of brain Crz⁺ neurons to the activation of gut Dh31⁺ cells

(a) Images show the changing diameter of the heart-tube before, during and after CO₂ exposure. Heartbeat frequency was extracted from the changing diameter in male flies expressing tdTomato in cardiomyocytes. The light and dark blue lines show the heartbeat frequency of individual flies and the average, respectively (n=6). (b) The effect of CO₂-exposure on the peak ΔF/F of Crz⁺ neurons in response to optogenetic activation of Dh31⁺ enteroendocrine cells. Stimulation intensity: 1.75 mW/mm². n.s., not significant, t-test (n = 5).

Extended Data Figure 9.. Identification of brain Dh31R⁺ neurons that regulate the intake of protein-containing food

(a) Expression pattern of neuropeptides, neuropeptidergic driver lines and Dh31R in the brain. Top row: arrowheads indicate neurons that co-express Dh31R with AstA-GAL4, Lk-GAL4, or Dsk-GAL4. Bottom row: arrowheads indicate neurons that co-express AstC with Dh31R or R67F03-GAL4. Scale bar, 50 μm. (b) Food intake of flies with Dh31R knockdown in different neurons. Gal4 lines (Itp, AstA, LK, R67F03) were crossed to UAS-Dh31R-RNAi and tested for amino acid food consumption. Male flies with Dh31R knockdown in R67F03-GAL4 cells consumed more amino acids than controls. (c and d) Genetic strategies to label R67F03-GAL4 neurons in the brain (c) and outside the brain (d). Representative confocal images show the different intersections between R67F03-GAL4 and otd-nls∷flp (corresponding genotypes are shown). Samples were immunostained with anti-GFP (green), nc82 (blue). Scale bar, 50 μm. (e) Knockdown of Dh31R (left panel) or AstC in R67F03-GAL4 cells outside the brain. Knockdown male flies consumed the same amount of amino acids and sucrose food compared with control flies. GAL4: R67F03-GAL4, otd-nls∷flp; UAS: UAS-Dh31R-RNAi, tub>stop>GAL80; Dh31R KD flies have all four transgenes (left panel). GAL4: R67F03-GAL4, otd-nls∷flp; UAS: UAS-AstC-RNAi, tub>stop>GAL80; AstC KD flies have all four transgenes (right panel). 8–11 experiments for each condition/genotype. **, p < 0.01, ANOVA followed by Tukey’s test.

Figure 1.. Consumption of amino acids suppresses feeding and promotes male courtship

(a) Simultaneous monitoring of feeding and courtship in individual males. One male was placed in a chamber with one virgin female and food. Behaviors were recorded for 30 min. (b-d) Effect of starvation duration and nutritional content on feeding and courtship behaviors. For each condition, the raster plots show the second-by-second behaviors of 5 representative males (top) and the fraction of time each male spent on feeding and courtship (bottom). n = 7 for each condition. n.s., not significant; *, p < 0.05; ***, p < 0.001; t-test (b and c) or different letters (p<0.05) as determined by ANOVA followed by Tukey’s test (d). (e) A courtship competition assay with two naïve males and one WT virgin female. (f) Dh31 signaling is required for the effect of ingested amino acids on courtship. Two genetically identical males that were refed with different nutrients competed to copulate with one virgin female. Copulation percentages are shown for males fed on the indicated food source. Food sources: AAs, a mixture of all amino acids (total biologically available nitrogen, 200 mM); sucrose, 50 mM; and agar. 5–6 experiments per condition and 18–25 successful matches per experiment. p-value was determined by Chi-square test. The dashed line indicates chance level. ***, p < 0.001.

Figure 2.. Gut Dh31 and brain Dh31R are required for the switch from feeding to courtship

(a) Dh31 enteroendocrine cells in the gut are specifically labeled by the combination of Dh31-GAL4 and Tsh-GAL80. Images show the brain, VNC and gut of a fly containing Dh31-GAL4, Tsh-GAL80 and UAS-CD8-GFP. Anti-GFP (green), nc82 in the brain and VNC (blue), phalloidin in the gut (blue). (b) Gut-specific Dh31 knockdown increased feeding duration and eliminated courtship when food contained amino acids (left) but had no effect when food contained only sucrose (right). GAL4: Dh31-GAL4, Tsh-GAL80. UAS: UAS-Dh31-RNAi. Gut Dh31 KD flies had all three transgenes. (c) Chemogenetic activation of gut Dh31-GAL4 cells increased courtship duration. GAL4: Dh31-GAL4, Tsh-GAL80. UAS: UAS-VR1. Gut Dh31 VR1 flies had all three transgenes. (d) Confocal images show the expression pattern of Crz-Gal4. Samples were immuostained with anti-GFP (red) and anti-Dh31R (blue). Arrowheads indicate neurons that express both Dh31R and Crz. (e) Dh31R knockdown in brain Crz-GAL4 neurons eliminated the effect of amino acids on courtship. GAL4: Crz-GAL4, tub>GAL80>stop. UAS: otd-nls∷flp, UAS-Dh31R-RNAi. Brain Dh31R KD flies had all four transgenes. In (b, c, e), n = 7 for each genotype/condition. *, p < 0.05; **, p < 0.01; ANOVA followed by Tukey’s test. Scale bar, 50 μm.

Figure 3.. Dh31 released from the gut by amino acids activates Dh31R⁺ neurons in the brain through circulation

(a) Amino acids evoked calcium response in Dh31⁺ enteroendocrine cells. Grayscale image shows KCl-responsive Dh31⁺ cells in posterior midgut and pseudocolored image shows amino acids-evoked responses in a representative sample. Scale bar, 50 μm. (b) Response of individual Dh31⁺ cells in a representative sample. Each row represents ΔF/F of one cell over time in response to different concentrations of amino acids. The vertical dash lines mark the beginning and end of stimulation. (c) The average calcium response (ΔF/F) is shown in black with SEM as gray shaded areas (n=11). Gray lines below the traces indicate the stimulation period of amino acids or saline control. (d) The percentage of responsive cells at different concentrations of amino acids. (e) Dose-response curve of peak ΔF/F in response to amino acids from 2.5 to 150 mM. (f) Hemolymph Dh31 levels in flies of various genotypes. Bar graphs show the relative Dh31 level normalized to control. Flies were starved for 24 hrs and then refed with different foods. Genotypes: WT; Dh31 mutant; Dh31 KD (Dh31-GAL4, tsh-GAL80, UAS-Dh31 RNAi); VR1 (Dh31-GAL4, tsh-GAL80, UAS-VR1). Foods: sucrose (50 mM), amino acids (200 mM) or capsaicin (50 μM mixed with 50 mM sucrose). Total protein was used as the loading control. 7 experiments per condition; 60 male flies per experiment. *, p < 0.05; **, p < 0.01. t-test or ANOVA followed by Tukey’s test. (g-i) Imaging calcium response of brain Crz⁺ neurons with three-photon microscopy. Response of brain Crz⁺ neurons to synthetic Dh31 peptide injected into the heart-tube of the fly. Image: three-photon fluorescence signals in green and third-harmonic signals in red. Scale bar, 25 μm. White arrow: cuticle; yellow arrows: Crz⁺ neurons. Grayscale image shows GCaMP7s basal signal of Crz⁺ neurons and pseudocolored image shows injection of Dh31-evoked response in a representative sample. The average calcium response (ΔF/F) is shown in black line with calcium response of individual flies shown in light red (n=6). (h) Response of brain Crz⁺ neurons to the activation of gut Dh31⁺ cells. Latency and ΔF/F of Crz⁺ neurons in response to optogenetic activation of gut Dh31⁺ cells at stimulation intensity ranging from 0.25 to 1.5 mW/mm². Inset shows average ΔF/F traces aligned by the rising phase of each trace. n=7. Significant differences (p < 0.05) are indicated by different letters, ANOVA followed by Tukey’s test. (i) The effect of CO₂-induced cardiac arrest on the response latency of Crz⁺ neurons in response to optogenetic activation of gut Dh31⁺ cells. Stimulation intensity: 1.75 mW/mm². *, p < 0.05, n = 5, t-test. (j) Expression of Dh31R in brain Crz⁺ neurons is required for their increased calcium levels in response to ingested amino acids. Foods: sucrose (50 mM), AAs (200 mM). Flies: control (Crz-GAL4, UAS-CaLexA), Dh31R KD (Crz-GAL4, UAS-CaLexA, UAS-Dh31R RNAi). Confocal images show representative samples. Scale bar, 50 μm. Box plots show quantification of fluorescence intensity from Crz⁺ neurons. 7 neurons per sample, 5–7 samples per condition. ****, p < 0.0001, Wilcoxon rank test. Error bars indicate SEM in all panels.

Figure 4.. AstC release from a population of Dh31R⁺ neurons suppresses protein feeding

(a) Male flies with Dh31 knockdown in the gut consumed more amino acids, but not sucrose, than controls. (b) Dh31R knockdown in brain Crz⁺ neurons did not affect food intake. (c) Confocal images show expression of GFP (green) in the brain of a male fly containing R67F03-GAL4 and UAS-CD8∷GFP. Anti-Dh31R (red) and anti-Crz (blue). Scale bar, 50 μm. Arrowheads indicate neurons positive for both Dh31R and AstC. (d-e) Male flies with Dh31R (d) or AstC (e) knockdown in brain R67F03 neurons consumed more amino acids, but not sucrose, than control flies. Brain AstC-GAL4: R67F03-GAL4, otd-nls∷flp. UAS-Dh31R RNAi: UAS-Dh31R-RNAi, tub>GAL80>stop. Brain Dh31R KD flies have all four transgenes (d). Brain AstC-GAL4: R67F03-GAL4, otd-nls∷flp. UAS-AstC RNAi: UAS-AstC-RNAi, tub>GAL80>stop. Brain AstC KD flies have all four transgenes (e). (f) Model: gut derived neuropeptide Dh31 acts on brain neurons expressing Dh31R to regulate feeding and courtship behaviors through AstC and Crz, respectively.