hugin+ neurons were a novel glucose sensor in the fly brain.

(A) Experimental procedure schematic. Newly eclosed flies were gathered and provided with regular food for 5 days, followed by a 12-hour period of starvation. Proboscis Extension Response (PER) assays were conducted. (B) Fraction of flies showing PER to different concentrations of sucrose (two-way ANOVA; *,p=0.0349;***,p=0.002;****,p<0.0001; n=4 groups, each of 10 flies). (C-D) Fraction of flies of the indicated genotypes under 30 °C showing PER to 400 mM of sucrose (one-way ANOVA; *,p=0.0233;**,p=0.0054;***,p=0.001;****,p<0.0001; n=4-5 groups, each of 10 flies). (E) Fraction of indicated flies showing PER to different concentrations of sucrose (ND represent standard fly food) (two-way ANOVA; ****,p<0.0001; n=4-5 groups, each of 10 flies). (F) hugin expression in the brain, illustrated by mCD8::GFP expression driven by huginGAL4. Scale bar represents 100 μm. (G) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons during the perfusion of glucose with or without TTX (one-way ANOVA; **,p=0.002; n=6-7). Horizontal black bar represents the duration indicated glucose solution stimulation. (H) AstA expression in the brain, illustrated by mCD8::GFP expression driven by AstAGAL4. Scale bar represents 100 μm. (I) Representative traces and quantification of ex vivo calcium responses of AstA+ neurons during the perfusion of glucose with or without TTX (one-way ANOVA; *,p=0.0187 or 0.022; n=6). (J) Representative images of pre-photoconversion (pre-PC) and post-photoconversion (post-PC) CaMPARI signal in hugin-expressing neurons (upper). The Red:Green ratio represents intracellular Ca2+ concentrations (lower). Scale bar represents 100 μm (one-way ANOVA; ****,p<0.0001; n=18-23). ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Student’s t-test, one-way and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

Glucose activated hugin+ neurons via Gltu1 and KATP channel.

(A) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons during the perfusion of glucose with or without phlorizin (1mM),(t-test;**,p=0.0052; n=6-7). Horizontal black bar represents the duration indicated glucose solution stimulation. (B) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons in indicated flies during the perfusion of glucose (t-test;*,p=0.022; n=6). (C) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA; ****,p<0.0001;n=4 groups, each of 10 flies). (D-E) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons during the perfusion of glibenclamide (100 μM, D), glucose with alloxan (10 μM, E) (t-test;**,p=0.0091; n=6). (F) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons in indicated flies during the perfusion of glucose (t-test;**,p=0.0011; n=6). (G) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons during the perfusion of pyruvate (50 mM), (t-test;ns,p>0.05; n=6). (H) Representative traces and quantification of ex vivo calcium responses of hugin+ neurons during the perfusion of D-glucose and L-glucose (50 mM), (t-test;***,p=0.0045; n=6) ns, P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001. Student’s t-test and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

hugin+ neurons suppressed sweet taste through PK2-R1.

(A-B) Fraction of flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA; *,p=0.0153 or 0.0237; ****,p<0.0001;n=4-5 groups, each of 10 flies). (C) Fraction of flies showing PER to different concentrations of sucrose (two-way ANOVA; *,p=0.0486 or 0.0438; ***,p=0.0007 or 0.0001; ****,p<0.0001; n=4 groups, each of 10 flies). Flies were injected with saline or synthetic hugin for 30 minutes before the assay. (D) Representative traces (upper) and quantification (lower) of peak calcium transients of Gr5a+ neurons in indicated flies upon 5% sucrose after injection of synthetic hugin (t-test;***,p=0.0047; n=6). Horizontal black bar represents the duration sucrose stimulation. (E-F) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA; **,p=0.0055 or 0.0016 or 0.003; ***,p=0.0002 or 0.0005; ****,p<0.0001;n=4 groups, each of 10 flies)ns, P > 0.05; *P < 0.05; ***P < 0.001; ****P < 0.0001. Two-way ANOVA followed by post hoc test with Bonferroni correction was used for multiple comparisons when applicable.

hugin+ neurons activated AstA+ neurons through PK2-R1.

(A-B) Fraction of flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA; *,p=0.043; ***,p=0.0007 or 0.0001; ****,p<0.0001;n=4-5 groups, each of 10 flies). (C) Co-localization (dashed box) of PK2-R1+ neurons (green) and AstA+ neurons (red) in the SEZ region. Scale bar represents 100 μm. (D) Representative traces (upper) and quantification (lower) of peak calcium transients of AstA+ neurons after the photo-activation of hugin+ neurons (t-test;*,p=0.0114; n=6) from in vivo calcium imaging. Horizontal black bar represents the duration of red-light stimulation. (E-F) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA; *,p=0.0486 or 0.033 or 0.0224;**,p=0.0043 or 0.0011; ***,p=0.0001 or 0.0008; ****,p<0.0001; n=4 groups, each of 10 flies). ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Student’s t-test and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

AstA+ neurons inhibited Gr5a+ neurons through AstA-R1.

(A) Fraction of flies showing PER to different concentrations of sucrose (two-way ANOVA; *,p=0.024;**,p=0.0072 or 0.002; ***,p=0.0002; ****,p<0.0001; n=4 groups, each of 10 flies). Flies were injected with saline or synthetic AstA for 30 minutes before the assay. (B) Representative traces (left) and quantification (right) of peak calcium transients of Gr5a+ neurons in indicated flies upon 5% sucrose after injection of synthetic AstA (t-test;*,p=0.0476; n=6). Horizontal black bar represents the duration sucrose stimulation. (C) Localization of Gr5a+ neurons (green) and AstA+ neurons (red) in the brain. Scale bar represents 100 μm. (D) Co-localization of Gr5a+ neurons (green) and AstA-R1+ neurons (red). Arrows point to the neurons co-expressing the two receptors. Scale bar represents 50 μm. (E-F) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA **,p=0.0088; ***,p=0.0004; ****,p<0.0001; n=4 groups, each of 10 flies). (G-H) Representative traces (left) and quantification (right) of peak calcium transients of Gr5a+ neurons in indicated flies upon 5% sucrose during the process of photo-activation of AstA+ neurons (G, n=6; t-test;**,p=0.0058) and hugin+ neurons (H, n=7-8;t-test;*,p=0.0351). Shadows represents the duration of red-light stimulation. Horizontal black bar represents the duration of sucrose stimulation. ns, P > 0.05; *P < 0.05; ***P < 0.001; ****P < 0.0001. Student’s t-test and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

hugin-AstA circuitry inhibited feeding behavior.

(A) The illustration of experimental design in this Figure. Note that all flies were maintained at 20°C before the feeding assay to prevent neuronal silencing by Shibirets1 during these procedures. (B) Fraction of flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA; ****,p<0.0001; n=4 groups, each of 10 flies). (C-D) Food consumption of flies of the indicated genotypes and environmental temperatures (one-way ANOVA; **,p=0.004; ****,p<0.0001; n=8 groups, each of 4 flies). Food consumption is presented as fold change relative to the corresponding genetic control.

NMU+ neurons were the central energy sensor in mouse.

(A) Blood NMU levels of starved mice re-fed with 20% sucrose (t-test; *, p=0.0373; n=6). (B) Two-bottle preference tests for starved mice re-fed with sucrose (t-test; ***, p=0.0002; n=7). (C) Two-bottle preference tests for starved mice with or without intraperitoneal injection of NMU peptide (YFLFRPRN-NH2, 4.5μm/kg) (t-test; *, p=0.0293; n=7). (D) Two-bottle preference tests for indicated starved mice (t-test; **, p=0.0073; n=6). (E) Fiber photometry to record the calcium dynamics of NMU+ neurons in the VMH with GCaMP6m (left) and representative IHC image of expressing GCaMP6min VMH (right). (F-G) Representative trace (left) and heatmaps (right, n=5) showing calcium dynamics of NMU+ neurons in the VMH in response to gastric glucose infusion (20% sucrose, 200 μl), which elevates circulating glucose levels independently of oral sensory stimulation.. (H) Experimental approach to assess calcium signaling in NMU+ neurons in the VMH in vitro (left) and representative IHC image of expressing GCaMP6min VMH (right). (I-J) Representative traces and quantification of ex vivo calcium responses of NMU+ neurons during the perfusion of glucose with or without TTX (I), Alloxan (J) and Phlorizin (J) (one-way ANOVA; **, p=0.0049;***, p=0.0001 or 0.0006; n=6-7). Horizontal black bar represents the duration indicated glucose solution stimulation. (K) Anterograde trans-synaptic tracing of downstream targets of NMU neurons. NMU neurons were labeled by GFP expression following injection of a Cre-dependent AAV2/1-DIO-GFP into NMU-Cre mice. This virus undergoes anterograde trans-synaptic transfer to postsynaptic neurons. To enable GFP expression specifically in downstream target regions, an AAV-Cre virus was locally injected into the rNST. As a result, postsynaptic neurons in the rNST receiving input from NMU neurons were labeled by GFP delivered anterogradely from upstream NMU neurons. Representative images show that GFP-labeled downstream neurons in the rNST colocalize with Calb2 immunoreactivity, indicating that NMU neurons preferentially target Calb2 rNST neurons. (L) Fiber photometry to record the calcium dynamics of Calb2+ neurons in the rNST with GCaMP6m (left) and representative IHC image of expressing GCaMP6m in rNST (right). (M) Representative traces and quantification of calcium responses of Calb2+ neurons during glucose licking under physiological feeding conditions (500 mM sucrose), with or without NMU administration, assessing downstream modulation of sweet-responsive brainstem circuits(t-test; ****, p<0.0001; n=6). ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Student’s t-test and one-way ANOVA followed by post hoc test with Bonferroni correction was used for multiple comparisons when applicable.

Satiety suppressed sweet sensitivity in a dopamine-independent manner.

(A) Representative traces (left) and quantification (right) of peak calcium transients of Gr5a+ neurons in indicated flies (t-test; *, p=0.0373; n=6). Horizontal black bar represents the duration 5% sucrose stimulation. (B-D) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA; *, p=0.0498;****, p<0.0001; B-C, n=4 groups, each of 10 flies). The Area Under the Curve (AUC, D) represents sweet sensitivity in C. *P < 0.05; ****P < 0.0001. Student’s t-test and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

The effect of satiety signals on taste modulation.

(A) Fraction of fed flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA; *, p=0.012; ***, p=0.0006;****, p<0.0001; n=4 groups, each of 10 flies). (B) Fraction of starved flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA; *, p=0.0154; **, p=0.0081 or 0.0011; ***,p=0.0005 or 0.0001; ****, p<0.0001; n=4-5 groups, each of 10 flies). ns, P > 0.05; *P < 0.05; ***P < 0.001; ****P < 0.0001. Two-way ANOVA followed by post hoc test with Bonferroni correction was used for multiple comparisons when applicable.

Activation of IPCs decreased the circulating sugar levels and increased sweet sensitivity.

(A) Hemolymph glucose levels from the indicated genotypes and treatment temperatures (t-test; ****, p<0.0001; n=8-9, 1 hour at 30 °C). (B-C) Fraction of fed flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA;**, p=0.0048; ****, p<0.0001; n=4 groups, each of 10 flies). ns, P > 0.05; ****P < 0.0001. Student’s t-test and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

Sugar intake suppressed sweet sensation.

(A) Fraction of indicated flies which refed with different sugar with energy showing PER to different concentrations of sucrose (two-way ANOVA; ****, p<0.0001; n=4 groups, each of 10 flies). (B) Fraction of indicated flies which refed with different sugar without energy showing PER to different concentrations of sucrose (two-way ANOVA; ****, p<0.0001; n=4 groups, each of 10 flies). (C) Fraction of indicated flies showing PER to different concentrations of fructose (two-way ANOVA; *, p=0.0255; ***, p=0.0002; ****, p<0.0001; n=4 groups, each of 10 flies). (D) Fraction of indicated flies showing PER to different concentrations of trehalose (two-way ANOVA;**, p=0.0081 or 0.0013; ***, p=0.0002; ****, p<0.0001; n=4 groups, each of 10 flies). ****P < 0.0001. two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

Starvation led to a decrease in the circulating sugar levels.

Hemolymph glucose levels from indicated flies (t-test; ***, p=0.00017; ****, p<0.0001; n=6-8). ***P < 0.001. Student’s t-test was used for comparisons.

The response of different neuropeptide-expressing neurons to glucose.

(A) LK expression in the brain, illustrated by mCD8::GFP expression driven by LKGAL4. Scale bar represents 100 μm. (B) Representative traces and quantification of ex vivo calcium responses of LK+ neurons during the perfusion of glucose (n=6). (C-D) The distribution (C) and calcium responses (D) of TK+ neurons (n=6). (E-F) The distribution (E) and calcium responses (F) of CCHa1+ neurons (n=6). (G-H) The distribution (G) and calcium responses (H) of CCHa2+ neurons (n=6).

The calcium response of hugin neurons upon glucose stimuli.

(A) Representative image of hugin neurons in SEZ. Scale bar represents 50 μm. (B) Calcium responses of different cluster of hugin neurons during the prefusion of glucose (t-test, *, p=0.0151; n=6).*P < 0.05; ***P < 0.001. Student’s t-test used for comparisons.

The calcium response of AstA+ neurons upon glucose treatment.

(A) Representative image of different part of AstA neurons. Scale bar represents 50 μm. (B) Calcium responses of different cluster of hugin neurons during the prefusion of glucose (n=6)

Knockout of PK2-R1 enhanced sweet sensation.

(A-C) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA; *, p=0.0464 or 0.0153; **, p=0.0015 or 0.0038 or 0.0035; ***, p= 0.0002; ****, p<0.0001; n=4 groups, each of 10 flies). ***P < 0.001; ****P < 0.0001. Two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

The expressing pattern of PK2-R1 in the proboscis

Representative image of PK2-R1 in the proboscis. Scale bar represents 50 μm.

hugin+ neurons in the brain activated AstA+ neurons.

(A-B) Fraction of flies of the indicated genotypes and environmental temperatures showing PER to different concentrations of sucrose (two-way ANOVA; **, p=0.0082; ***, p= 0.0007 or 0.0001; ****, p<0.0001; n=4 groups, each of 10 flies). Prior to the behavioral assay, the neural connection between the brain and the ventral nerve cord (VNC) was surgically severed, while the flies remained physically intact and retained normal proboscis musculature and motor output. PER was measured using the same protocol as in intact flies. (C) Representative traces (upper) and quantification (lower) of peak calcium transients of AstA+ neurons after the photo-activation of hugin+ neurons (t-test;*, p=0.0337; n=6) from ex vivo calcium imaging. Horizontal black bar represents the duration of red-light stimulation.

Manipulating gene AstA-R1 enhance PER.

(A-C) Fraction of flies of the indicated genotypes showing PER to different concentrations of sucrose (two-way ANOVA; *, p=0.0404; **, p=0.009 or 0.0022 or 0.0019;***, p=0.0005; ****, p < 0.0001; n=4 groups, each of 10 flies). ***P < 0.001; ****P < 0.0001. Two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

The expressing pattern of AstA-R2 in the proboscis

Representative image of AstA-R2 in the proboscis. Scale bar represents 50 μm.

hugin-AstA-Gr5a circuitry inhibited feeding behavior.

(A) Food consumption of flies of the indicated genotypes (one-way ANOVA; ***, p=0.0009; ****, p < 0.0001; n=8 groups, each of 4 flies). (B) The TG content of indicated genotypes (one-way ANOVA; ****,p<0.0001; n=8 groups, each of 2 flies). (C) Food consumption of flies of the indicated genotypes (one-way ANOVA; ****,p<0.0001; n=8 groups, each of 4 flies). ns, P > 0.05; ***P < 0.001; ****P < 0.0001. Student’s t-test and two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

NMU peptide suppressed sweet taste in fly.

Fraction of flies showing PER to different concentrations of sucrose (two-way ANOVA; *, p=0.106; **, p=0.0015 or 0.0029 or 0.0025; ****, p < 0.0001; n=4 groups, each of 10 flies). Flies were injected with saline or synthetic NMU for 30 mins before the assay.**P < 0.01; ****P < 0.0001. Two-way ANOVA followed by post hoc test with Bonferroni correction were used for multiple comparisons when applicable.

The expressing pattern of NMU neurons and Calb2 neurons

(A) Co-labeling of NMU+ neurons using a Cre-dependent mCherry reporter (red) and in situ hybridization (green).Scale bar represents 200 μm. (B) NMU arbors was extended near Calb2rNST neurons. Scale bar represents 200 μm. (C) Co-labeling of Calb2+ neurons with a Cre-dependent mCherry reporter (red) and NMU antiboy staining (green). Scale bar represents 200 μm.

A working model

Elevated circulating glucose following feeding directly activates hugin+ neurons via cell-autonomous glucose sensing. Activated hugin neurons release Hugin neuropeptide, which engages PK2-R1 receptors on AstA neurons, promoting AstA release. AstA then suppresses the sensitivity of Gr5a-expressing gustatory sensory neurons through AstA-R1 signaling, thereby reducing sweet-driven feeding behavior. Under starvation, reduced circulating glucose decreases endogenous hugin neuronal activity, partially relieving this inhibitory tone and permitting enhanced sweet sensitivity. However, experimental manipulation of the circuit demonstrates that the hugin–AstA axis retains the capacity to modulate feeding behavior across feeding states, indicating that it functions as a glucose-responsive, state-modulated inhibitory system rather than a strictly satiety-exclusive brake. The modest phenotype observed upon PK2-R2 reduction suggests the potential existence of parallel or complementary hugin-dependent pathways, possibly arising from functional heterogeneity within hugin neuronal subpopulations. These additional branches remain to be defined.