Insulin receptor signaling in gustatory cells suppresses sugar taste sensitivity and feeding in Drosophila

Summary

Dynamic sensory processing helps animals adjust behaviors in response to changing environments and internal states. Hunger is a salient internal cue that can directly modify chemosensory signaling to promote homeostatic behavior. In Drosophila melanogaster, hunger enhances the sensitivity of sweet gustatory receptor neurons (GRNs) to encourage feeding, and here we investigated candidate hunger/satiety hormone receptors to identify mechanisms of metabolic regulation in primary taste cells. The conserved insulin receptor (InR) is expressed in sweet GRNs, and targeted InR knockdown and inactivation within these cells elevated sucrose sensitivity, increased sucrose consumption, and altered feeding behavior under sated conditions. Additionally, sweet cellular responses to sucrose were reciprocally affected by inactive and overactive InR signaling. These findings reveal that InR signaling can modulate sweet sensitivity at the level of primary taste cells, suppressing feeding by reducing taste responsiveness under specific metabolic conditions.

Graphical abstract

Highlights

• •Drosophila insulin receptor (InR) is expressed in labellar sweet taste cells
• •InR inactivation elevates sweet taste sensitivity and sugar feeding in the fed state
• •Overactive and inactive InR reciprocally impact cellular taste responses to sugar

Neuroscience; Behavioral neuroscience; Systems neuroscience

Introduction

Behavioral plasticity allows organisms to appropriately respond to shifting external and internal stimuli.^1,2,3,4,5 Hunger is a conserved internal cue that regulates widespread behaviors, including locomotion,^6,7,8 reproduction,^9,10,11,12 and decision making^13,14 to prioritize feeding. A key mechanism of hunger modulation involves direct modification of chemosensory processing within the nervous system. Food deprivation heightens both olfactory and gustatory sensitivity to certain cues in humans and other animals.^{15,16,17,18,19} In particular, sweet taste elicits strong neural and behavioral responses across species,^{20,21,22,23,24} making the control of sweet sensation and perception especially important for flexible feeding.²⁵ For example, gustatory alliesthesia²⁶ describes how hunger enhances the hedonic expression of sweet stimuli to encourage feeding and maintain energy homeostasis.²⁵ Despite neuroimaging evidence for this phenomenon,^27,28,29 the precise mechanisms underlying the integration of metabolic information within gustatory circuits remain unclear.

Previous research on the interactions between taste and hunger has primarily focused on central processing^27,28,29 with less known about how metabolic signals may influence peripheral taste responses. The fruit fly, Drosophila melanogaster, is an advantageous model organism for studying gustatory neurobiology³⁰ because primary taste responses can be recorded in awake animals with intact internal states.^{31,32,33,34,35,36,37,38} The robust genetic tools³⁹ combined with this powerful in vivo imaging approach^38,40,41 in Drosophila have greatly improved our understanding of taste encoding and how gustatory circuits coordinate feeding behavior.^{31,32,33,34,35,36,37,41,42} Flies detect diverse chemical tastants using gustatory receptor neurons (GRNs) located in various structures, most prominently within taste sensilla populating the surface of the labellum—the fly’s primary peripheral taste organ.^43,44,45 Of the five recognized labellar GRN classes,³¹ sugar-sensing “sweet” GRNs are the most abundant and are characterized by expression of nine sugar-specialized gustatory receptors (GRs)^46,47 that function as tetrameric ligand-gated cation channels.⁴⁸ Besides sugar, these taste cells respond to many attractive chemical compounds to ultimately drive appetitive feeding responses.^{31,32,49,50,51,52,53}

Importantly, recording from mammalian taste cells requires anesthesia, a treatment that significantly alters internal status and endocrine signaling.^54,55 In flies, sweet GRNs are amenable to in vivo calcium imaging in awake animals, which has revealed that starvation enhances taste sensitivity at the level of primary sweet cells.^56,57,58 Dopamine receptor signaling is one mechanism of this starvation-induced increase in sweet sensitivity,^56,57 and conversely, factors that suppress sweet GRN responses also alter taste sensitivity and impact feeding.^34,35,59 Efforts to find other neuromodulators in sweet GRNs have identified CCKLR-17D3, a receptor for the Drosophila cholecystokinin analog: drosulfakinin (DSK),^60,61 but the role of canonical hunger/satiety hormones in maintaining dynamic gustatory processing has not been fully explored.

Metabolic hormones maintain energy homeostasis through various mechanisms, including hunger/satiety balance in the brain.^62,63,64 In mice, glucagon signaling affects sweet taste responsiveness⁶⁵ and mammalian type 2 taste receptor cells (TRCs) express insulin receptors,⁶⁶ indicating the possibility of endocrine signaling in primary gustatory cells. In flies, Drosophila-insulin-like-peptides (Dilps) are a group of hormones with diverse roles in metabolism and reproduction that act through a conserved insulin receptor (InR),⁶⁷ and AKHR is a receptor for the Drosophila glucagon analog, adipokinetic hormone (AKH). Akin to mammals, Dilp and AKH signaling changes with metabolic status to help regulate feeding.^{68,69,70,71,72,73,74,75} Furthermore, InR signaling has been found to impact olfactory sensitivity during food-deprivation,^76,77 but its potential function in primary gustatory cells remains unclear. Our work aims to characterize the role of hunger/satiety hormone receptors in sweet GRN modulation, sweet taste sensitivity, and overall feeding behavior under different metabolic conditions. Using single-cell transcriptomic evidence, we found that InR and AKHR were expressed in many sweet GRNs and that InR contributes to satiety-state feeding regulation at the level of primary taste cells. Specifically, we found that InR signaling in sated conditions dampens sweet GRN cellular responses to reduce taste sensitivity and feeding. These findings broaden our understanding of gustatory-metabolic integration and demonstrate how endocrine signaling directly alters chemosensory input to modify behavior.

Results

Transcriptomic analysis of candidate modulatory receptors in sweet GRNs

To explore how internal metabolic state may impact peripheral taste processing, we first searched through available transcript databases. A microarray dataset from the proboscis and maxillary palps of fed and starved flies (GEO: GSE27927) revealed that InR and AKHR levels fluctuated significantly in a state-dependent manner: in flies starved one or two days, InR levels increased (4.06 and 5.55-fold change, respectively) and AKHR levels decreased (−4.27 and −5.35-fold change, respectively).⁷⁸ These reciprocal changes in InR and AKHR expression could potentially reflect transcriptional feedback mechanisms that help maintain homeostatic signaling during transitions in satiety state as Dilp and AKH levels rapidly fluctuate.^{68,75,79,80,81} To search specifically for hunger/satiety hormone receptor expression in sweet GRNs, we used the single-cell transcriptomic database, Fly Cell Atlas.⁸² Putative sweet GRNs were identified via expression of Gr64f, a sugar-specific GR and marker of each sweet GRN on the labellum.^31,46,47 33 total Gr64f + cells were isolated from a database cluster previously annotated “Gustatory receptor neuron of the labellum” (ArrayExpress: E-MTAB-10519) (Figures 1A and S1). Within this subset, we screened for receptors with the potential to modulate sweet GRN activity.

Figure 1.

Insulin receptor knockdown in sweet GRNs affects sucrose sensitivity and feeding

(A) Expression heatmap of Gr64f+ sweet GRNs on the labellum and several candidate modulatory receptors via single-nucleus transcriptomic analysis from Fly Cell Atlas.⁸² Cells ordered from highest to lowest relative Gr64f expression level. Warmer colors represent increased relative expression values.

(B–D) Labellar PER with a sucrose concentration series (10–1,000 mM) in flies expressing RNAi against DopECR (B), AKHR (C), InR (D), or empty RNAi controls in sweet GRNs under both fed (top) and 24-h starved (bottom) conditions. n = 8 groups of 6–10 flies per genotype. All mated females. All data plotted as mean ± SEM. p values represent main effects of genotype by two-way ANOVA. Multiple comparisons by Šidák’s post-test at each individual concentration are depicted. ∗p < 0.05, ∗∗p < 0.01.

(E–G) One-choice FLIC (fly liquid interaction counter) assay with 10 mM sucrose to quantify feeding interactions (left) and mean feeding event duration (right). Data from fed flies expressing RNAi against DopECR (E), AKHR (F), InR (G), or empty RNAi controls in sweet GRNs. n = 13–30 flies per genotype. All mated females. All data plotted as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01 by unpaired t test.

As a positive control, we first searched for expression of DopECR, a receptor previously shown to be expressed in sweet GRNs and mediate dopamine’s hunger-dependent enhancement of sweet taste sensitivity.⁵⁶ Consistent with previous results, we found DopECR expression in the majority of Gr64f+ cells (Figures 1A and S1B). Expanding our search to metabolic receptors, AKHR (55%) and InR (42%) also showed high levels of expression in this subset (Figures 1A and S1B), a finding that aligns with a previous study that used TRAP-seq (translating ribosome affinity purification sequencing) of Gr5a+ sweet GRNs to detect a high abundance of active InR mRNA.^83,84 Conversely, we found no expression of two receptors previously linked to sweet GRN modulation: CCKLR-17D3⁶¹ and NPFR⁸⁵ (Figure S1A). In contrast to mammalian work demonstrating a role for leptin signaling in TRCs,⁸⁶ we did not find expression of the fly leptin receptor analog: dome (Figure S1A). Despite the apparent prevalance of other neurotransmitter receptors in sweet GRNs (Figure S1A), we decided to focus our subsequent analyses on InR and AKHR, as both endocrine receptors exhibit high Gr64f + co-expression and have been strongly associated with hunger/satiety regulation.^67,68,69,70 We conducted additional analyses using other sugar-sensing GRs to demonstrate that expression of our candidate receptors is not enriched in any particular subset of sweet GRNs (Figure S2A). Moreover, we found lower relative expression of InR and AKHR in Gr66a+ bitter GRNs^40,87 (21%) (Figure S2B) compared to sweet GRNs.

InRs expressed in sweet cells impact sucrose taste sensitivity and feeding

We were next interested in how these candidate modulatory receptors modify sweet taste sensitivity in the context of hunger or satiety. As a comparison to previous findings, we first selectively knocked-down DopECR via targeted RNAi only in Gr64f+ cells. We assessed how this manipulation altered taste sensitivity to sucrose on the labellum via the proboscis extension response (PER) assay⁸⁸ under both fed and 24 h-starved conditions. In the sated state, DopECR-RNAi flies displayed lower PER rates than Empty-RNAi controls on average; however these differences were statistically insignificant (Figure 1B, top). Under 24 h starved conditions, DopECR knockdown did not change sucrose sensitivity (Figure 1B, bottom), replicating previous results.⁵⁶ We then repeated this approach for both AKHR and InR knockdown. AKHR knockdown had no effect on sucrose sensitivity, regardless of internal state (Figure 1C). However, InR knockdown resulted in a significant enhancement of sucrose sensitivity in the fed state, particularly at low concentrations (Figure 1D). These results imply that InR in sweet GRNs helps to regulate sucrose sensitivity, particularly under sated conditions.

While PER assesses acute taste detection, we were also interested in how sweet GRN-specific candidate receptor knockdown would impact overall feeding behavior. To quantify nuanced feeding metrics, we utilized the fly liquid interaction counter (FLIC).^34,89 Fed flies were individually placed into feeding chambers and given access to a 10 mM sucrose food source for 3 h (1-choice FLIC assay). Food interactions, feeding events, and average feeding event durations were quantified for each fly. Consistent with the dynamic nature of feeding behavior, we observed varying levels of feeding across all three metrics. DopECR knockdown in sweet GRNs resulted in a trending suppression of sucrose interactions (Figure 1E) and a strong reduction in feeding events (Figure S1C). Interestingly, these flies also had a significant increase in the average feeding event duration (Figure 1E), potentially suggesting the involvement of compensatory post-ingestive mechanisms that lengthen feeding bouts and maintain sufficient caloric intake during this assay. In contrast to the strong effects of DopECR knockdown, flies expressing AKHR-RNAi did not exhibit shifts in any of the tested FLIC feeding metrics (Figures 1F and S1D). Aligning with our observed PER phenotype, InR knockdown in sweet GRNs also significantly elevated feeding interactions and increased average feeding event duration (Figure 1E), but did not impact the number of feeding events (Figure S1E). Importantly, recent work comparing FLIC and other feeding assays showed that feeding bout duration is strongly correlated with food intake.⁹⁰ Together, these findings indicate that InR’s impacts on sweet GRNs and sucrose sensitivity also meaningfully affect prolonged feeding behavior.

Since InR knockdown in sweet GRNs resulted in robust behavioral phenotypes compared to the other candidate receptors, we decided to focus our subsequent analysis on the impacts of InR-mediated modulation. Accordingly, we next validated InR expression in sweet GRNs by co-staining labellar tissue from Gr64f>GFP flies with a human insulin receptor (INSR) antibody previously shown to successfully label InR in larval olfactory neurons.⁷⁷ There was prominent immunoreactivity throughout the labellum, including clear overlap with GFP+ sweet GRN cell bodies in the sensilla and along the labial nerve (Figure S3A). This fluoresence was absent in control tissue without the primary antibody (Figure S3B). This finding is consistent with our analysis of the labellar transcriptome and provides protein-level confirmation for InR’s presence in sweet GRNs.

InR signaling in sweet cells modulates taste sensitivity and sucrose feeding in a state-dependent manner

To thoroughly investigate how InR signaling influences sweet GRN sensitivity, we expressed either a dominant-negative (InR[DN]) or constitutively active (InR[CA]) InR transgene in Gr64f+ cells to induce inactive or overactive insulin signaling, respectively.^68,91 We first tested how expression of InR[DN] in sweet GRNs shifted sucrose sensitivity in mated female flies using sucrose PER under both fed and 24-h starved conditions. We hypothesized that inactive InR signaling would have the strongest effects on taste sensitivity in the fed state, when circulating insulin is relatively high.⁹² Indeed, sated InR[DN] flies had significantly elevated PER rates compared to InR[WT] controls, at all tested sucrose concentrations (Figure 2A, top). Under starvation conditions, PER rates between InR[WT] and InR[DN] were not statistically significant (Figure 2A, bottom).

Figure 2.

Inactive insulin receptor signaling in sweet GRNs enhances sucrose sensitivity

(A–C) Labellar PER to a sucrose concentration series (10–1,000 mM) in mated female (A and B) and male (C) flies expressing dominant negative (DN) (A–C), constitutively active (CA) (B), or wild-type (WT) InR in sweet GRNs. Experiments conducted under both fed (top) and 24-h starved (bottom) conditions. n = 8 groups of 9–14 flies per genotype. All data plotted as mean ± SEM. p values represent main genotype effects by two-way ANOVA. Multiple comparisons by Šidák’s post-test at each individual concentration are depicted. ∗∗p < 0.01, ∗∗∗p < 0.001.

Next, we predicted that overactive insulin signaling via InR[CA] expression would most likely reduce PER in the starved state, when Dilp signaling is relatively low.⁶⁷ However, we found no differences between InR[CA] flies and InR[WT] controls in either the fed or starved experiment (Figure 2B). We also tested whether the behavioral impacts of overactive InR signaling would manifest when flies were moderately hungry instead of fully starved; however, sucrose PER in flies food-deprived for 6 h was also unaffected by InR[CA] expression (Figure S5A). Due to well-documented sex differences in Drosophila insulin sensitivity and signaling,^93,94,95 we also modified InR signaling in male flies and assessed changes in sucrose sensitivity using the same PER paradigm. Like females, InR[DN] expression in male flies resulted in higher sucrose PER, especially at lower concentrations, but only in the fed state (Figure 2C). Importantly, Gr64f is only expressed in adulthood,⁴⁶ thus the Gr64f-gal4 driver should not be active throughout development to have indirect effects on adult flies. To directly determine if prolonged loss of InR signaling in adult sweet GRNs elicits gross metabolic dysregulation that could otherwise confound our behavioral phenotypes, we quantified starvation sensitivity and found that InR[DN] expression did not impact starvation resistance for either female or male flies (Figure S4A). To determine if inactive InR signaling also affects sweet GRNs in the tarsi,^96,97 we first identified Gr64f+ sweet GRNs in the leg-specific dataset from Fly Cell Atlas and searched for InR co-expression. While fewer tarsal cells exhibited Gr64f expression, the proportion that co-expresses InR was similar to that observed in the labellum (Figure S4B). Accordingly, we repeated our sucrose PER experiments but instead stimulated flies’ legs. Tarsal PER rates were not affected by InR[DN] expression in either the fed or starved state (Figure S4C) suggesting InR’s modulatory effects are specific to labellar taste cells. Together, these results demonstrate that inactive InR signaling in labellar sweet GRNs, for both sexes, causes a fed-state specific elevation in sucrose PER, suggesting that InR typically acts to temper sweet sensitivity with satiety.

We were also interested in how InR signaling within sweet GRNs regulates overall feeding behavior and consumption. We first returned to our 1-choice FLIC assay to determine how inactive InR signaling in sweet GRNs would impact prolonged feeding behaviors. In mated females, InR[DN] and InR[WT] flies shared a similar number of feeding interactions and events (Figures 3A and S4D), but InR[DN] flies exhibited longer feeding event durations at both 10 and 1,000 mM sucrose (Figure 3A). In males, InR[DN] flies had significantly more feeding interactions at several sucrose concentrations and showed longer feeding event durations at 10 mM (Figure 3B) but no strong changes in the number of feeding events (Figure S4E). Recent work comparing FLIC and other feeding assays showed that feeding bout duration is strongly correlated with food intake.⁹⁰ Aligning with our PER findings, overactive InR signaling via InR[CA] expression did not impact FLIC feeding metrics in flies pre-starved for 6 or 24 h (Figures S5B and S5C).

Figure 3.

Inactive insulin signaling in sweet GRNs increases sucrose intake behaviors

(A and B) One-choice FLIC (fly liquid interaction counter) assay at increasing sucrose concentrations (10–1,000 mM) to quantify feeding interactions (left) and mean feeding event duration (right). Data from mated female (A) and male (B) flies expressing dominant negative (DN) or wild-type (WT) InR in sweet GRNs. n = 15–24 flies per condition. All data plotted as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01 by unpaired t test.

(C and D) Dye-based binary feeding choice assay to measure threshold-level (1 vs. 5 mM) sucrose preference (left) and consumption (right) at increasing starvation time points (0–24 h). Data from mated female (C) and male (D) flies expressing DN or WT InR in “sweet” GRNs. n = 19–21 groups of 10 flies per condition. All data plotted as mean ± SEM. p values represent main genotype effects by two-way ANOVA. Multiple comparisons by Šidák’s post-test at each individual starvation level are depicted. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

To further characterize the behavioral effects of InR signaling, we used a dye-based binary feeding choice assay^31,32,33 to measure preferential consumption of a low concentration of sucrose. For 1 h, flies were allowed to freely choose between and consume two solutions: 1 or 5 mM sucrose, the approximate Drosophila threshold preference level for sucrose.⁹⁸ We conducted these experiments at five different starvation time points (0, 2, 6, 12, and 24 h) to determine if the effect of InR would change with increasing hunger. Both female and male InR[DN] flies had elevated preference and consumption of this low concentration of sucrose compared to InR[WT] controls, particularly at low/moderate hunger levels (Figures 3C and 3D). Consistent with the rest of our behavioral results, preferential consumption was not impacted by InR[CA] expression, regardless of starvation level (Figure S5D). Collectively, these findings suggest that InR signaling modulates sweet GRNs to reduce sucrose consumption.

Insulin signaling modifies sweet taste cell calcium responses

Previous investigation of sweet GRN regulation revealed that hunger enhances cellular sucrose sensitivity via increased calcium influx.^56,57 To determine if InR signaling regulates cellular sucrose responses in a similar manner, we performed in vivo calcium imaging of sweet GRNs during labellar sucrose stimulation³⁸ while recording changes in GCaMP6f fluorescence.⁹⁹ Specifically, we imaged sweet GRN axon terminals in the sub-esophageal zone of the brain, a region where GRNs synapse with second-order taste neurons.⁴⁵ This technique has been used to describe nuances about Drosophila gustatory processing in the context of internal state modulation.^2,31,56,57 We quantified calcium responses from the SEZ under both fed and 24-h starved conditions, recording from flies expressing wild-type (WT), inactive, or overactive InR. Between InR[DN] and InR[WT] flies, there was no overall effect of genotype for either satiety state (Figures 4A and S6B). However, under fed conditions, InR[DN] flies exhibited a significant increase in peak calcium responses to 10 mM sucrose (Figures 4A and S6A). Conversely, constitutively active insulin signaling during starvation significantly suppressed calcium responses to sucrose. InR[CA] flies showed a significant decrease in peak fluorescence compared to InR[WT] controls, most prominently to 10 and 1,000 mM sucrose (Figure 4B). Expression of InR[CA] did not alter any cellular calcium responses in the fed state (Figure S6C). These results demonstrate that inactive and overactive InR lead to reciprocal and state-dependent effects on sweet GRN calcium responses, further confirming the suppression of neuronal taste responses by InR, at least at lower concentrations.

Figure 4.

Insulin receptor signaling in sweet GRNs influences cellular responses to sucrose

(A and B) In vivo calcium imaging of GCaMP6f in sweet GRNs during labellar sucrose stimulation. Data from flies expressing dominant negative (DN) (A), constitutively active (CA) (B), or wild-type (WT) InR in sweet GRNs under fed (A) or 24-h starved (B) conditions. Calcium responses measured as ΔF/F over time at each concentration (left) and peak ΔF/F (right). Yellow lines under each curve indicate when the stimulus was on the labellum. n = 11–16 flies per genotype. All mated females. All data plotted as mean ± SEM. p values represent main genotype effects by two-way ANOVA. Multiple comparisons by Šidák’s post-test at each concentration are depicted. ∗p < 0.05.

(C) Model for satiety state-dependent modulation of sweet GRNs by InR. In the fed state, InR signaling (red) in sweet GRNs dampens cellular sensitivity to sucrose and reduces feeding. In the starved state, hunger signals like dopamine (DA) via DopECR (purple) plus downstream circuit modulation can impact sucrose sensitivity and promote feeding.

Discussion

Describing the molecular and cellular basis of metabolic-gustatory interactions is essential for understanding the pathological consequences of their dysregulation. For example, sweet taste sensitivity in healthy individuals fluctuates throughout the day,¹⁰⁰ but this variation is absent in those with insulin resistance.¹⁰¹ In addition, chemosensory disruption is a well-documented symptom of various metabolic disorders,^102,103,104 but we are only beginning to elucidate the impact of sensory nutrition¹⁰⁵ on overall health. From our basic investigation in Drosophila, we propose that high circulating insulin/Dilps act on InRs expressed in sweet GRNs to suppress cellular responses to sucrose, reduce sweet sensitivity, and deter feeding in a fed state, but in a starved state, any signal from low circulating insulin/Dilps is potentially overridden by other hunger signals that enhance sweet sensitivity (Figure 4C.) These hunger signals may include dopamine released onto sweet GRNs via DopECR^56,57 or additional downstream modulatory mechanisms in the brain^37,106 that increase sweet salience and encourage feeding (Figure 4C). These results offer insight into potentially conserved mechanisms of sensory modulation by metabolic signals.

Negative regulation of sweet taste sensitivity by InRs

Mammalian investigations of gustatory regulation at the level of taste bud have implicated several neuropeptide and endocrine signals, including: neuropeptide Y,¹⁰⁷ cholecystokinin,¹⁰⁸ leptin,⁸⁶ and glucagon.⁶⁵ InR expression has also been observed in type 2 TRCs,⁶⁶ but their role in taste sensitivity was not evaluated. Aligning with our single-cell transcriptomic analysis, previous studies implementing TRAP-seq of Gr5a+ sweet GRNs detected a high abundance of active InR mRNA^83,84 and recent work using the same Fly Cell Atlas database also reported InR expression in primary GRNs.¹⁰⁹ We also analyzed candidate receptor expression from Gr66a + putative bitter GRNs and found lower relative expression levels compared to sweet GRNs (Figure S2B). However, we currently do not have sufficient evidence to determine whether InR-mediated modulation is unique to sweet cells. InR is broadly expressed in the proboscis, and future research could explore how other GRNs, mechanosensory cells, or peripheral glial cells are potentially affected by insulin signaling. Importantly, the internal conditions of the flies used for Fly Cell Atlas could impact expression levels, since at least InR and AKHR fluctuate with metabolic status.⁷⁸ Additionally, the Fly Cell Atlas transcriptomes were generated using single-nucleus sequencing, an approach that captures less genetic information per-cell compared to single-cell techniques.^{82,110,111,112} Relatively low sampling in these datasets could limit expression levels for certain genes, especially if a proportion of their mRNA is cytoplasmic.^113,114

Previously, one study found no changes in sucrose PER when Dilp3-expressing cells were acutely silenced,⁵⁸ implying that insulin signaling would have limited influence on sweet taste responsiveness. However, this study used a global silencing approach and performed PER under 24-h starved conditions. Our InR manipulation was specific to sweet GRNs, and the resulting phenotype only emerged in the fed state, when circulating Dilp levels are relatively high.⁹² Furthermore, cholecystokinin-like DSK acts as a prominent satiety signal in the fly brain and is co-released by insulin-producing cells (IPCs) to reduce food intake.⁶⁰ Another study provided evidence that DSK suppresses sweet taste within GRNs following refeeding, but targeted knockdown of CCKLR-17D3 implied that DSK receptor signaling actually had the opposite effect—enhanced sweet sensitivity.⁶¹ In addition to the lack of expression in our Fly Cell Atlas search (Figure S1A), this discrepancy complicates the current interpretation of DSK and CCKLR-17D3 involvement in sweet GRNs.

Prior investigation found that starvation-dependent sucrose sensitivity required DopECR in sweet GRNs, but their knockdown of DopECR only reduced sucrose PER following 6 h of starvation and, consistent with our results, this effect was not seen at 24 h.⁵⁶ Other previous studies have identified mechanisms of sweet GRN regulation by internal state in response to dietary manipulation. A high sugar diet has been shown to dampen sweet GRN sensitivity through the activation of the intracellular sugar sensor: O-linked N-Acetylglucosamine transferase (OGT)³⁴ and through a gut-taste axis mediated by Hedgehog signaling.¹¹⁵ Importantly, our tarsal PER results indicates that InR does not significantly modulate sweet GRNs in the legs (Figure S4C). Previous work using tarsal PER found that prolonged exposure to dietary sucralose enhances sweet GRN sensitivity via NPF, a Drosophila neuropeptide Y analog, and its receptor, NPFR.⁸⁵ Thus, it is possible that modulation occurs through distinct pathways on different sensory organs. Other studies have instead proposed that NPFR more generally contributes to hunger-enhanced sweet sensitivity via upstream excitation of dopaminergic interneurons in the brain,^57,58 and we were unable to find NPFR expression in labellar Gr64f + cells (Figure S1A). Consistent with previous studies, we found strong and specific expression of AKHR in sweet GRNs (Figure 1A).^8,116 However, we did not find any impact of this receptor on sweet sensitivity or feeding behavior (Figures 1C and S1D). While the exact function of this glucagon-like metabolic receptor within sweet taste cells remains unclear, past work found that AKHR expressed in the brain mediates starvation-induced hyperactivity.⁸ This result could suggest that AKHR signaling in sweet GRNs could contribute to other hunger-dependent behaviors.

Modification of feeding behavior from insulin signaling in taste cells

In addition to its metabolic functions, insulin helps control energy homeostasis by signaling satiety in the brain.¹¹⁷ In mammals, this regulation is characterized by anorexigenic circuitry in the hypothalamus,¹¹⁷ but insulin can also interact with dopaminergic reward pathways to affect sweet valence.¹¹⁸ In Drosophila, global InR inactivation in the brain leads to increased preference for caloric sugar, an effect that mimics a food-deprivation phenotype and further supports the anorexigenic nature of insulin signaling in the fly nervous system.¹¹⁹ In the current study, our dye-based binary feeding choice assays showed that InR signaling in sweet cells diminishes sucrose detection thresholds and consumption. Additional experiments using FLIC revealed that these same signals shorten the duration of sucrose feeding bouts, a metric that is representative of food intake.⁹⁰ Importantly, inactive InR signaling in sweet GRNs did not impact starvation survival (Figure S4A), suggesting these changes in feeding output were likely not driven by metabolic dysfunction. The phenotypes we observed in our dye-based assays were most prominent at low/moderate starvation levels and the FLIC results were in fed flies, further confirming the state-dependent nature of insulin signaling in sweet GRNs.

While both feeding assays have been used to uncover connections between taste processing and behavior,^31,32,33,34 these experiments measure sucrose intake over multiple hours. This raises the possibility that our feeding phenotypes were impacted by post-ingestive mechanisms within the brain¹²⁰ or gut¹²¹ that regulate food intake over longer periods of time. Despite this concern, we saw significant enhancement in feeding behavior when only primary GRNs were manipulated. A potential explanation is that increased consumption caused by the loss of satiety signaling in sweet cells increases secretion of Dilps, which are still unable to modulate taste responses, leading to even more consumption. Moreover, sweet GRNs in the pharynx express Gr64f and are known to regulate prolonged sugar feeding,¹²² suggesting these feeding phenotypes could also reflect modulation of pharyngeal sweet GRNs. Additionally, studies in humans,¹²³ rodent models,¹²⁴ and Drosophila^93,94,95 highlight significant sex differences in insulin signaling and sensitivity. Despite these important differences, our data indicate that InR similarly modulates sweet GRNs in both sexes. InR inactivation significantly enhanced sucrose sensitivity (Figures 2A and 2C), feeding behavior (Figures 3A and 3B), or preferential consumption (Figures 3C and 3D) in both male and female flies. The one subtle dimorphism we observed was InR[DN] expression having a stronger enhancing effect on sucrose interactions in male flies compared to females of the same genotype (Figure 3B). Along with previous evidence showing that insulin pathway activity is elevated in males on a low-sugar diet⁹⁴ our finding could suggest that the male gustatory system is more sensitive to InR modulation under certain conditions. However, future work is necessary to fully characterize how these potential sex differences manifest at a behavioral level.

Direct modulation of cellular taste responses by InR signaling

Since our InR manipulation impacted sucrose PER and feeding, we anticipated that cellular changes in primary taste neurons may drive this behavioral shift. Measuring changes in neuronal activity with in vivo calcium imaging is strongly indicative of action potential firing patterns¹²⁵ and this approach has been used to study complex aspects of gustatory processing.^{31,32,33,34,35,36,37,38} For example, previous calcium imaging analysis showed that during starvation, sucrose elicits stronger responses in sweet GRNs.^56,57 In our control flies, we replicated this starvation effect and found that food-deprivation alone enhances calcium responses in sweet GRN axon terminals (Figure 4A). Similar approaches have also revealed that sweet GRN activity can be affected by high dietary sugar^34,35 or parallel bitter circuitry via sweet cell-specific GABA_BR signaling.⁵⁹ Here, we propose that InR in sweet GRNs modulates neuronal activity in a state-dependent manner, adding another mechanism by which primary taste cell responses can be dynamically regulated.

Expression of InR[CA] in sweet GRNs significantly suppressed cellular responses to sucrose during starvation (Figure 4B), a finding that is consistent with our initial prediction that the effects of overactive InR signaling would most likely manifest under low Dilp conditions. However, this same manipulation did not elicit behavioral phenotypes. This discrepancy could reflect redundancy in hunger signals acting downstream to preserve feeding drive even when sweet input is diminished. For example, modulatory DopECR signaling in sweet GRNs does not fully explain hunger-induced enhancement of sweet taste sensitivity,^56,57 and higher order gustatory neurons within the PER or related circuits are also directly modulated by starvation.^37,106

In contrast, expression of InR[DN] in sweet GRNs strongly enhanced sucrose PER and feeding in the fed state, but these behavioral phenotypes were associated with an increase in neuronal calcium responses only at 10 mM sucrose (Figure 4A). This concentration likely lies near the sensory threshold for sweet GRN activation, particularly under fed conditions, and subtle elevations in calcium influx caused by inactive InR signaling may lift suppression just enough to generate a detectable increase in fluorescence. At higher sucrose concentrations, when InR[WT] sweet GRNs are already strongly activated, our assay may lack the sensitivity required to detect further modulatory increases in calcium signal, masking potential physiologically relevant changes in sweet GRN activity that may contribute to behavioral modification. Alternatively, mild differences in calcium responses could also indicate the prevalence of other mechanisms underlying InR-mediated sweet GRN suppression that would not be associated with changes in calcium transients, including potential rapid structural changes. In Drosophila, acute presynaptic homeostatic potentiation (PHP) in phasic motor neurons can occur within just 10 min and is not associated with changes in calcium influx, but has been shown to promote homeostatic signaling by rapidly adjusting presynaptic active zone nanostructure.^126,127 In mice, leptin induces presynaptic structural changes within hypothalamic feeding circuits after only 6 h.¹²⁸ Moreover, evidence from other invertebrate model systems has revealed that insulin and other neuromodulators can drive presynaptic remodeling in sensory neurons after 12–24 h.^129,130 Future investigation is required to explore the precise intracellular or circuit-level mechanisms in which InR signaling regulates sweet GRN output and the resulting behavioral consequences.

In conclusion, we describe a mechanism of direct communication between endocrine signaling and taste processing that regulates sweet sensitivity at a primary cellular and behavioral level. Future work can explore the intracellular or circuit-level mechanisms by which insulin elicits taste suppression and how higher order taste neurons may also integrate internal state signaling to further control feeding.

Limitations of the study

While quantifying InR expression or signaling via qPCR or western blot would help confirm the effects of InR knockdown or inactivation, detecting these minor changes would be extremely technically challenging as we are altering InR in only one cell type across a complex sensory organ (labellum). Despite the effects elicited by our InR manipulations being consistent, we acknowledge that the magnitude of these changes were minor, especially at the cellular level. In the discussion, we have included potential explanations for the discrepencies between our behavioral and cellular findings.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Molly Stanley, (molly.stanley@uvm.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All original data generated in this study can be found in a Mendeley dataset at https://data.mendeley.com/datasets/b3y8dtwcz3/1, https://doi.org/10.17632/b3y8dtwcz3.1
• •This paper does not report original code. Previously generated code was used to analyze FLIC data and can be found at https://github.com/MStanleyLab/and in the Mendeley dataset at https://data.mendeley.com/datasets/b3y8dtwcz3/1, https://doi.org/10.17632/b3y8dtwcz3.1
• •Any additional information required to reanalyze the data reported in this paper are available from the lead contact upon request.

Acknowledgments

This project was supported by a grant from the National Institute on Deafness and Other Communication Disorders (NIDCD, award number 1R01DC022843-01A1) and new lab start-up funds from the University of Vermont. We thank the Bloomington Stock Center (BDSC) for fly stocks. UAS-InR[WT], UAS-InR[DN], and UAS-InR[CA] were provided by Exelxis, Inc. to BDSC. We thank Dr. Dennis Matthew at the University of Nevada, Reno for the human insulin receptor (INSR) antibody recommendation and evidence for its successful cross-reactivity in Drosophila. We also thank Michael Gordon and Viktoriya Li in the Gordon Lab at the University of British Columbia, where pilot studies were conducted. Graphics were generated with BioRender.com and we thank Kayla Audette for contributing to the graphics.

Author contributions

Conceptualization, M.S.; methodology, C.A. and M.S.; investigation, C.A., L.E., M.I., and M.S.; writing – original draft, C.A. and M.S.; writing – review and editing, C.A., L.E., M.I., and M.S.; visualization, C.A. and M.S.; resources, M.S.; supervision, M.S.; funding acquisition, M.S.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken anti-GFP	AbCam	Cat#: 13970 RRID: AB_300798
Rabbit anti-INSR	Cloud-Clone Corp.	Cat#: PAA895Hu02
Goat anti-chicken 488	AbCam	Cat#: 150169 RRID: AB_2636803
Goat anti- rabbit 647	Invitrogen	Cat# A21245; RRID: AB_2535813
Chemicals, peptides, and recombinant proteins
Sucrose	Sigma-Aldrich	S7903
Agar	Sigma-Aldrich	A1296
Amaranth, FD&C Red #2	Sigma-Aldrich	A1016
Erioglaucine, FD&C Blue #1	Sigma-Aldrich	861146
Deposited data and code
Mendeley Dataset	https://data.mendeley.com/datasets/b3y8dtwcz3/1	https://doi.org/10.17632/b3y8dtwcz3.1
FLIC Analysis Code	Guillemin et al.33 Pletcher Lab	https://github.com/MStanleyLab/
Experimental models: Organisms/strains
D. melanogaster: Gr64f-Gal4	Dahanukar et al.131	Flybase: FBtp0057275
D. melanogaster: UAS-Empty. RNAi	Bloomington Drosophila Stock Center	RRID: BDSC_36304 Flybase: FBti0114379
D. melanogaster: UAS-Empty. RNAi	Bloomington Drosophila Stock Center	RRID: BDSC_36303 Flybase: FBti0040535
D. melanogaster: UAS-DopECR.RNAi	Bloomington Drosophila Stock Center	RRID: BDSC_93710 Flybase: FBti0217551
D. melanogaster: UAS-AKHR.RNAi	Bloomington Drosophila Stock Center	RRID: BDSC_51710 Flybase: FBti0157647
D. melanogaster: UAS-InR.RNAi	Bloomington Drosophila Stock Center	RRID: BDSC_51518 Flybase: FBti0158797
D. melanogaster: UAS-InR[WT]	Exelixis Inc. - Bloomington Drosophila Stock Center	RRID: BDSC_8262 Flybase: FBti0040676
D. melanogaster: UAS-InR[DN]	Exelixis Inc. - Bloomington Drosophila Stock Center	RRID: BDSC_8252 Flybase: FBti0040688
D. melanogaster: UAS-InR[CA]	Exelixis Inc. - Bloomington Drosophila Stock Center	RRID: BDSC_8263 Flybase: FBti0040693
D. melanogaster: UAS-CsChrimson.mVenus	Bloomington Drosophila Stock Center	RRID: BDSC_55136 Flybase: FBti0160803
D.melanogaster: UAS-GCaMP6f	Bloomington Drosophila Stock Center	RRID: BDSC_52869 Flybase: FBti0156888
Software and algorithms
ImageJ	Schneider et al.132	RRID: SCR_002285, SCR_003070 https://imagej.nih.gov/ij
R Studio (4.3.2)	R Studio Team	RRID: SCR_000432 https://www.rstudio.com/
Prism 10 (10.3.1)	Graphpad	RRID: SCR_002798https://www.graphpad.com/scientificsoftware/prism/
Fly Cell Atlas - ASAP	Li et al.82	https://asap.epfl.ch/fca
Illustrator	Adobe	RRID: SCR_010279 https://www.adobe.com
BioRender	BioRender	RRID: SCR_018361 https://app.biorender.com/

Experimental model and study participant details

Fly strains and maintenance

All flies were kept on standard cornmeal food at 25°C and 60% humidity before testing. For all experiments, flies were 2–10 days old. Mated females and males were used where indicated. Information for each genotype can be found in the key resources table. All fly stocks are available from the BDSC. Stocks were not back-crossed, rather all comparisons were made with genetic background-matched controls. Two separate UAS-Empty-RNAi lines were used to generate the control genotypes used in Figure 1. To match their respective experimental alleles, BDSC #36303 (chromosome 3) was used for the DopECR knockdown experiment and BDSC #36304 (chromosome 2) was used for the AKHR and InR knockdown experiments.

Method details

Chemicals

A full list of chemicals with source information can be found in the key resources table. Sucrose was made up in water at 1 M and diluted further in water to the indicated concentrations.

Fly Cell Atlas analysis

Using ASAP (https://asap.epfl.ch/fca), we analyzed Fly Cell Atlas’ 10x proboscis and maxillary palps (stringent) or 10x leg (stringent) single-nucleus transcriptomes (E-MTAB-10519). We first identified the previously annotated cell clusters “gustatory receptor neuron of the labellum” from the proboscis and “gustatory receptor neuron” from the leg dataset via HVG-UMAP visualization.⁸² A subset of cells in each cluster that exhibited expression of Gr64f were identified as sweet GRNs. In each putative sweet GRN, relative expression values for Gr64f and each candidate receptor were recorded from ASAP. Raw UMI reads for each cell were normalized (10,000 counts/cell) and log-transformed (log(x+1)) to generate relative expression levels for the genes of interest.⁸²

PER

Proboscis extension response (PER)⁸⁸ was used to assess appetitive feeding responses and taste sensitivity by quantifying the proportion of proboscis extensions in response to test stimulations. Flies were collected into vials containing standard cornmeal food (fed experiments) or into food deprivation vials containing 1% agar (starved experiments) 24 h prior to testing and kept at 25°C. For labellar PER, flies were individually mounted under a dissection scope using a mouth pipette to transfer each fly into a 200 μL pipette tip that was cut so only the head and labellum were exposed. For tarsal PER, flies were mounted on slides using quick-dry nail polish to immobilize the flies while exposing their tarsi. For all PER experiments, flies recovered in a humidity chamber for ∼1 h after mounting and then manually water satiated. Water also served as the first stimulus to ensure subsequent PER responses were not driven by thirst or dehydration. The test solutions consisted of increasing sucrose concentrations, including 10 mM, 100 mM, 250 mM, and 1000 mM sucrose. Labellar sensilla or tarsi were manually stimulated three times for each concentration. A positive response was recorded if the fly extended to at least one stimulation. Water stimulations in between each sucrose solution were used to verify continued water satiety.

Immunohistochemistry

Immunofluorescence on labella was conducted as previously described.^31,33 Labella from female, Gr64f>CsChrimson.mVenus flies (shorthand: Gr64f>GFP) were dissected, split, and fixed in 4% paraformaldehyde in PBS +0.2% Triton (PBST) for 25 min before washing. Tissues were then blocked in 5% normal goat serum (NGS) for 1 h before adding primary antibodies (chick anti-GFP at 1:1000, rabbit anti-InR at 1:40) overnight. A human polyclonal insulin receptor (INSR) antibody (PAA895Hu02) was chosen due to its immunogen aligning closely with the conserved region of Drosophila InR and previous evidence showing reactivity to InR protein in larval olfactory neurons.⁷⁷ The anti-InR primary antibody was excluded for negative control tissue, but the rest of the protocol remained the same. After washing with 0.2% PBST, secondary antibodies (goat anti-chicken Alexa 488, goat anti-rabbit Alexa 647, both at 1:200) were incubated for 3 h. After washing in PBST, samples were placed on slides in SlowFade gold with #1 coverslips as spacers. Images were acquired with a 3i Spinning disc Confocal station (Zeiss upright microscope, 2 K × 2 K 40 fps sCMOS camera, CSU-W1 T1 50 μm spinning disc) with a 40x water immersion objective. A z stack with 1 micron steps were captured with the following settings: laser power 40, 488 channel 20 ms exposure, 647 channel 100 ms exposure, brightfield 100 ms exposure. FIJI (ImageJ) was used to z-project 20 microns of labellar tissue (maximum projection for fluorescent channels, minimum projection for brightfield), and images were exported separately and merged with the 647 INSR blue channel pseudocolored to red for clearer visualization of overlap. All scale bars = 50 microns.

Starvation survival assay

For starvation survival experiments, groups of 10 mated female or male flies were transferred into food deprivation vials (1% agar) at 2 days old and kept at 25°C for the duration of the experiment. Every 12 h, the number of flies that were alive in each vial was recorded until all the flies had died. Female and male vials were run in parallel.

Fly liquid interaction counter (FLIC)

FLIC experiments were performed as previously described.^33,89 For each experiment, flies were taken directly off standard cornmeal food and individually transferred into behavioral feeding chambers via mouth pipette. Capacitance sensors in each chamber measured the number of interactions between a fly and a single sucrose food source (1-choice assay). Flies from each genotype were loaded into chambers on opposite sides of a Drosophila Feeding Monitor (DFM) and interactions with the sucrose solution were recorded over 3 h. This raw output was analyzed via a custom R script modified from the Pletcher Lab (full code on GitHub, see key resources table). Measurements from the loading process were excluded to avoid potential artifacts caused by adding flies into the chambers. Data were analyzed with R code like previous publications to quantify feeding interactions, feeding events, and feeding event durations. Minimum signal threshold was set to 10, feeding threshold was set to 20, and measurements were taken every 200 ms. A feeding interaction was defined as any individual reading above the feeding threshold. A feeding event was defined as 10 consecutive interactions above the minimum threshold, as long as one of these readings reached the feeding threshold. Sequential periods of continuous interactions were combined into one single event as long as the gaps of inactivity between them lasted less than 1 s. Feeding event duration was measured in seconds. Data from chambers that exhibited disrupted signal detection were removed, including flies with 0 recorded interactions or raw values that were exceedingly high (>1000 discrepancy between raw interaction count and R analysis output). Flies that failed to interact with the food source (<15 interactions) or exhibited extremely high mean event duration measurements (>12 s) were also excluded.

Dye-based feeding assays

Binary choice experiments were conducted as previously described.^31,32,33 Vials containing 10 flies were flipped into 1% agar food-deprivation vials for the indicated number of hours before test. Flies assessed at 0-h starvation were flipped into vials containing standard cornmeal food. At test, flies were transferred into vials that contained six dots of alternating red (0.5 mg/mL Amaranth, FD and C Red#2) and blue (0.125 mg/mL Erioglaucine, FD and C Blue#1) colored solutions. Each 10uL dot contained dye, 2% agar, and either 1 mM sucrose or 5 mM sucrose. The color associated with each sucrose concentration was counterbalanced so that half of the replicates had red 5 mM sucrose and half had blue 5 mM sucrose. Flies were allowed to freely choose between either food option for 1 h in the dark in an incubator with ∼50% humidity at 29°C. After being frozen at −20°C, a dissection microscope was used to score each fly’s postmortem abdomen color as red, blue, purple, or no color. Preference index for each vial was calculated as: ((# of flies labeled with X color)-(# of flies labeled with Y color))/(total # of flies with color). The number of flies in each vial that consumed either food choice (%Eating) was calculated as: ((# of flies labeled blue, red, or purple/total # flies in vial) ∗100).

In vivo calcium imaging

GCaMP6f fluorescence of Gr64f + GRN terminals was performed as previously described.^31,32,38 Flies were briefly anesthetized with CO₂ and mounted in a custom imaging chamber with the proboscis extended and held in place with a dental waxer. Mounted flies recovered for 1 h in a humidity chamber, then a small area of cuticle was removed and the esophagus cut to expose the SEZ. Adult hemolymph-like (AHL) solution (108 mM NaCl, 5 mM KCl, 4 mM NaHCO₃, 1 mM NaH2PO₄, 5 mM HEPES, 15 mM ribose, 2 mM Ca²⁺, 8.2 mM Mg²⁺, pH 7.5) was continuously applied to the area during the dissection and the same solution was used for immersion during imaging. A Leica SP5 II Confocal microscope with a 25x immersion objective captured the GCaMP6f signal over time. The SEZ was imaged at 4–6x zoom with an 8000Hz line speed, line accumulation of 2, and a resolution of 512 x 512 pixels with the pinhole opened to 2.86 AU. A micromanipulator with a glass capillary delivered the solutions (water or sucrose) over the labellum while the GCaMP6f signal was recorded. Each stimulation included 5 s of baseline, 2 s with the stimulus over the labellum, and 7 s while the signal returned to baseline after stimulus removal. Each fly received water first, followed by increasing concentrations of sucrose. Fed flies were kept on food while starved flies were kept on 1% agar for 24 h prior to mounting. Experiments comparing fed and starved flies of the indicated genotypes were run in parallel on the same day. FIJI (ImageJ) was used to draw a tight ROI around sweet GRN axonal projections in the SEZ via baseline GCaMP6f signal.³⁸ Flourescence values from the ROI was then extracted over time. For each GCaMP6f recording, ten timepoints in a row were used as a baseline and the signal converted to ΔF/F (%). The peak ΔF/F was calculated as the maximum change in fluorescence using an average of three timepoints around the highest value.

Quantification and statistical analysis

All statistical analyses were conducted using GraphPad Prism 10 software. Sample sizes were chosen based on estimated effect sizes and power analyses from previous experiments using similar assays. Genotype controls were run in parallel for each experiment. Specific statistical tests are included in the figure legends. FLIC experiments were conducted separately at each sucrose concentration and genotypes were compared using unpaired t-tests. For all other experiments, sucrose concentrations or starvation time points were run in parallel and genotype differences were quantified by two-way ANOVA (genotype main-effect) with Šidák’s post-test. All data plotted as mean ± SEM. Asterisks indicate ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p. < 0.001, ∗∗∗∗p < 0.0001.