Introduction

Many types of tastants are beneficial at low concentrations, and harmful at high levels. Examples include minerals such as Na⁺ and Ca²⁺ (Zhang, Ni et al. 2013, Lee, Poudel et al. 2018), and fatty acid such as hexanoic acid (Ahn, Chen et al. 2017, Pradhan, Shrestha et al. 2023). Organic molecules such as cholesterol are a crucial sterol that plays essential roles in cellular membrane integrity, signaling functions, and steroid hormone synthesis. This vital molecule supports numerous biological processes in animals, including reproduction, nutrient transport, and cellular activation (Igarashi, Ogihara et al. 2018). However, excessive cholesterol consumption can lead to a host of poor health consequences, including cardiovascular disease, and type 2 diabetes (Soliman 2018). Due to the bivalent impact of cholesterol on human health, it stands to reason that there may be mechanisms that exist to promote or repress the taste of cholesterol. However, it is not clear whether cholesterol is sensed by the mammalian taste system. Mice and humans express several dozen taste receptors, most of which function in bitter taste (referred to as either T2Rs or TAS2Rs). The activities of two of these bitter receptors, T2R4 and T2R14 have been shown to be modulated by cholesterol. However, it is unclear if they contribute to the taste of cholesterol (Pydi, Jafurulla et al. 2016, Shaik, Jaggupilli et al. 2019, Kim, Gumpper et al. 2024).

It is plausible that insects such as Drosophila melanogaster might display a gustatory attraction to low levels of cholesterol since unlike vertebrates, which can synthesize cholesterol internally, fruit flies must obtain sterols through their diet (Clark and Bloch 1959, Niwa and Niwa 2011, Shaheen 2020). Insects acquire cholesterol primarily from plant-derived phytosterols or pre-existing cholesterol (Jing and Behmer 2020). For instance, Manduca sexta and Bombyx mori convert plant sterols to cholesterol through dealkylation in their gut, which is essential for producing hormones such as ecdysone (Igarashi, Ogihara et al. 2018). Drosophila acquires cholesterol directly from dietary sources such as phytosterols (e.g., sitosterol, stigmasterol) and fungal sterols (e.g., ergosterol) found in yeast (Niwa and Niwa 2011). Given that consumption of high cholesterol is harmful (Soliman 2018, Schade, Shey et al. 2020), fruit flies might display a gustatory aversion to high levels, while finding low levels attractive. Such a bivalent response would be similar to the flies’ taste attraction to low concentrations of Na⁺ and their repulsion to high Na⁺ (Zhang, Ni et al. 2013, Jaeger, Stanley et al. 2018, Taruno and Gordon 2023, Sang, Dhakal et al. 2024). Ca²⁺ is also required at low levels and deleterious at high concentrations. However, we have previously shown that fruit flies are indifferent to modest levels of Ca²⁺ and avoid high Ca²⁺ (Lee, Poudel et al. 2018). Thus, if flies are endowed with the capacity to taste cholesterol, it is open question as to whether they would have a bivalent gustatory response depending on concentration, similar to Na⁺ (Zhang, Ni et al. 2013, Jaeger, Stanley et al. 2018, Taruno and Gordon 2023, Sang, Dhakal et al. 2024), or be indifferent to low cholesterol and reject high cholesterol, similar to the flies’ differential reaction to Ca²⁺ depending on concentration (Lee, Poudel et al. 2018).

In Drosophila, gustatory organs are distributed on multiple body parts, including the labellum at the end of the proboscis, which represents the largest taste organ. The labellum consists of two halves, each of which is decorated with 31 external bristles. These sensilla house either two or four gustatory receptor neurons (GRNs), which allow them to respond to external chemical stimulation and modulate behavioral responses (Dahanukar, Foster et al. 2001, Larsson, Domingos et al. 2004, Suh, Wong et al. 2004, Benton, Vannice et al. 2009, Cameron, Hiroi et al. 2010, Chen, Wang et al. 2010, Kim, Lee et al. 2010, Kwon, Kim et al. 2010, Rimal and Lee 2018). This is accomplished through expression of a diverse repertoire of receptor classes, including gustatory receptors (GRs), ionotropic receptors (IRs), pickpocket (PPK) ion channels, and transient receptor potential (TRP) channels.

In this work, we reveal that flies taste cholesterol. Reminiscent of their reaction to Ca²⁺ (Lee, Poudel et al. 2018), they are indifferent to low cholesterol and reject high cholesterol. Using a combination of behavioral and electrophysiological assays, we demonstrate that a subset of the same class of GRNs that respond to bitter chemicals is also required in adults for avoiding the taste of higher cholesterol levels. In addition, we found that multiple members of the IR family are involved in cholesterol taste perception, and that there are two overlapping sets of IRs that are sufficient to confer cholesterol sensitivity to GRNs that normally do not respond to cholesterol. This work establishes that flies can taste cholesterol, and defines the underlying cellular and molecular mechanisms involved in rejection of high cholesterol.

Results

Flies taste cholesterol through a subset of bitter GRNs

To address whether fruit flies can taste cholesterol, we investigated whether cholesterol triggers action potentials in GRNs associated with taste bristles in the labella. The 31 sensilla present in each hemisphere of the labellum are categorized into long (L), intermediate (I), and short (S) subtypes (Figure 1A) (Hiroi, Marion-Poll et al. 2002). To examine cholesterol-induced action potentials in response to a range of cholesterol concentrations, we focused on S7, I8, and L6 sensilla and performed tip recordings. We detected action potentials in S7 once the cholesterol concentration reached 10^-3 %, whereas the I8 and L6 were nearly unresponsive even at 0.1% (Figures 1B and C). The methyl-β-cyclodextrin (MβCD) used to dissolve cholesterol did not evoke spikes (Figure 1—figure supplement 1A and B). We then analyzed all 31 sensilla using 0.1% cholesterol. We found that the S-type sensilla, especially the S6 and S7 sensilla, were most responsive, while very few spikes were induced from the I-type or L-type sensilla (Figure 1D).

Figure 1.

To determine which GRN type contributes to cholesterol-induced action potentials, we selectively inactivated different classes of GRNs by expressing a transgene encoding the inwardly rectifying potassium channel, Kir2.1 (Baines, Uhler et al. 2001). The bristles on the labellum harbor GRNs that fall into four main classes, each of which expresses a gene driver unique to that class (Montell 2021). These include A GRNs, which respond to sugars, low Na⁺ and other attractive compounds (Gr64f-GAL4), B GRNs, which are stimulated by bitter compounds, high Na⁺ and other aversive chemicals (Gr33a-GAL4), C GRNs, which are activated by H₂O and hypo-osmolarity (ppk28-GAL4), and D GRNs, which respond to Ca²⁺ and high concentrations of other cations (ppk23-GAL4) (Thorne, Chromey et al. 2004, Dahanukar, Lei et al. 2007, Moon, Lee et al. 2009, Cameron, Hiroi et al. 2010, Lee, Poudel et al. 2018). We found that silencing B GRNs reduced neuronal responses to cholesterol, whereas inhibition of other GRN types exhibited normal neuronal firing (Figures 1E and F). These data demonstrate that cholesterol is sensed by B GRNs.

A cluster of IRs is required to sense cholesterol in adult Drosophila

To pinpoint the molecular sensors for detecting cholesterol, we first investigated requirements for the largest family of taste receptors–the GRs. Six GRs are broadly expressed in GRNs and three of them serve as co-receptors (Montell 2021, Shrestha and Lee 2023), including GR32a (Miyamoto and Amrein 2008), GR33a (Lee, Moon et al. 2009, Moon, Lee et al. 2009), GR39a.a (Dweck and Carlson 2020), GR66a (Moon, Köttgen et al. 2006), GR89a (Shrestha and Lee 2021), and GR93a (Lee, Moon et al. 2009). We performed tip recordings, demonstrating that mutations disrupting any of these co-receptors had no impact on cholesterol-induced action potentials (Figure 2— figure supplement 1A). Drosophila encodes 13 TRP channels, several of which function in taste (Al-Anzi, Tracey et al. 2006, Kang, Pulver et al. 2010, Kim, Lee et al. 2010, Zhang, Raghuwanshi et al. 2013, Zhang, Raghuwanshi et al. 2013, Mandel, Shoaf et al. 2018, Leung, Thakur et al. 2020, Montell 2021). We analyzed mutant lines disrupting most of these channels and found that the neuronal responses were normal (Figure 2— figure supplement 1B).

IRs comprise another large family of receptors that function in taste, as well as in other sensory processes (Rytz, Croset et al. 2013, Rimal and Lee 2018). To address whether any IR is required for cholesterol taste, we screened the 32 available Ir mutants by performing tip recording on S7 sensilla using 0.1% cholesterol. Most mutants displayed normal responses (Figure 2A), including those with previously identified gustatory functions such as Ir7a¹(acetic acid sensor) (Rimal, Sang et al. 2019), Ir7c^GAL4 and Ir60b³ (high Na⁺) (McDowell, Stanley et al. 2022, Sang, Dhakal et al. 2024), Ir56b¹ (low Na⁺) (Dweck, Talross et al. 2022), Ir62a¹ (Ca²⁺) (Lee, Poudel et al. 2018), Ir94f¹ (cantharidin) (Pradhan, Shrestha et al. 2024) as well as Ir20a¹, Ir47a¹, Ir52a¹, and Ir92a¹ (alkali) (Pandey, Shrestha et al. 2023). In contrast, our survey revealed that five mutants (Ir7g¹, Ir25a², Ir51b¹, Ir56d¹, and Ir76b¹) exhibited strong defects in firing in response to cholesterol (Figure 2A). Two of the mutations, Ir25a² and Ir76b¹, disrupt co-receptors that are necessary for sensing most attractive and aversive tastants (Ganguly, Pang et al. 2017, Lee, Poudel et al. 2018, Dhakal, . 2021, Shrestha and Lee 2021, Stanley, Ghosh et al. 2021, Aryal, Dhakal et al. 2022, Xiao, Baik et al. 2022, Li, Sun et al. 2023, Pandey, Shrestha et al. 2023, Pradhan, Shrestha et al. 2024). In further support of the roles of these five IRs for detecting cholesterol, we observed similar phenotypes resulting from mutation of additional alleles (Ir7g², Ir51b², Ir56d², and Ir76b²) (Zhang, Ni et al. 2013, Sánchez-Alcañiz, Silbering et al. 2018, Dhakal, Sang et al. 2021, Pradhan, Shrestha et al. 2024), or due to placing the mutation (Ir25a²) in trans with a deficiency (Df) spanning the locus (Figure 2B). Furthermore, using phytosterol stigmasterol to stimulate S6, S7, and S10 sensilla, we confirmed that the five mutants exhibited consistent phenotypes, underscoring the specificity of these IRs in sterol detection (Figure 2—figure supplement 1C—E).

Figure 2.

To provide additional verification that the phenotypes exhibited by the Ir25a², Ir56d¹, and Ir76b¹mutants were attributed to the loss of Ir25a, Ir56d, and Ir76b, we conducted rescue experiments. To do so, we used GAL4 lines specific to each gene to drive the respective wild-type UAS-cDNAs in the corresponding mutant backgrounds. We found that the responses to cholesterol were fully restored in S7 sensilla stimulated with 0.1% cholesterol (Figures 2D and F).

To evaluate the dose-dependent defects exhibited by the mutants, we performed tip recordings to examine the neuronal responses of S7 sensilla to a spectrum of cholesterol percentages (10^-5 to 10^-1). All five mutants exhibited significantly reduced neuronal firing in response to cholesterol percentages over a 100-fold range (10^-3 to 10^-1; Figure 2E). However, at lower percentages of cholesterol (10^-5 and 10^-4), all five IR mutants did not vary significantly from the control (Figure 2E).

IRs required in B GRNs for cholesterol-induced neuronal firing

To test whether the IRs function in B GRNs, we used two approaches: RNA interference (RNAi) and gene rescue experiments. To knock down gene expression in B GRNs, we took advantage of the Gr33a-GAL4 and found that targeting any of the five genes dramatically reduced action potentials in S7 sensilla in response to 0.1% cholesterol (Figure 2C). In contrast, when we used a D GRN driver (ppk23-GAL4) in combination with the same UAS-RNAi lines, there was no decrease in neuronal firing. To perform gene rescue experiments, we used the Gr33a-GAL4 to express each wild-type cDNA transgene in the corresponding mutant background and performed tip recordings. In all cases, we rescued the mutant phenotypes (Figures 2D and F). Thus, we conclude that the IRs function in B GRNs.

Given that the five IRs are required in B GRNs, it stands to reason that they are expressed in these neurons. Indeed, Ir7g, Ir25a, Ir51b, and Ir76b have been shown previously to be expressed in B GRNs (Lee, Poudel et al. 2018, Dhakal, Sang et al. 2021, Pradhan, Shrestha et al. 2024). However, Ir56d, which has a role in sweet-sensing A GRNs, has not. To explore this possibility, we performed double-labeling experiments. We expressed UAS-dsRed under the control of the Ir56d-GAL4 and did so in files that included a B GRN reporter (Gr66a-I-GFP). Each of the two bilaterally symmetrical labella contain 11 S-type sensilla, 11 I-type sensilla and 9 L-type sensilla. We found that 10.7 ± 1.4 cells co-expressed both the dsRed and GFP markers (Figure 2— figure supplement 1F). By tracing dendrites from individual GFP-expressing cells, we identified the specific sensilla innervated by each marker. Most S2, S3, S4, S6, and S7 sensilla that expressed the Ir56d reporter were co-labeled with the B GRN reporter. Thus, the B GRNs in the two sensilla that elicited the highest frequency of cholesterol-induced action potentials (S6 and S7) were labeled by the Ir56d reporter.

IRs required for avoiding the taste of cholesterol

The requirement for the five IRs for cholesterol-induced action potentials in B GRNs suggests that cholesterol is an aversive taste. To explore this question, we used the well-established binary choice assay in which we allowed flies to choose between 2 mM sucrose alone or 2 mM sucrose mixed with various percentages of cholesterol. We mixed the two food alternatives with either blue or red food dye so we could inspect the flies’ abdomens to assess which option they consumed (Aryal, Dhakal et al. 2022). At the lowest percentage tested (10^-5%), flies showed only a slight aversion to cholesterol-containing food (Figure 3A). As cholesterol concentration increased, they showed a dose-dependent aversion, with a very strong aversion at 0.1% (PI=-0.72 ±0.03; Figure 3A). Both male and female flies showed comparable avoidance responses to 0.1% cholesterol, indicating the behavior is not sex-specific (Figure 3B). The aversion was not due to the MβCD used to dissolve the cholesterol since the flies were indifferent to sucrose alone versus sucrose plus MβCD (Figure 3—figure supplement 1A). Moreover, the flies showed similar levels of aversion to sucrose plus cholesterol versus either sucrose alone (Figure 3A) or sucrose plus MβCD (Figure 3—figure supplement 1B). The dye used in the study also did not alter the behavioral response (Figure 3—figure supplement 1C).

Figure 3.

To determine the impact of inhibiting B GRNs on gustatory behavior, we used the Gr33a-GAL4 to drive an expression of UAS-Kir2.1. As a control, we also inactivated other classes of GRNs and found that expression of kir2.1 in A GRNs (Gr64f-GAL4), C GRNs (ppk23-GAL4), and D GRNs (ppk28-GAL4) had no impact on cholesterol avoidance (Figure 3C). Surprisingly, inhibiting B GRNs (Gr33a-GAL4) not only eliminated cholesterol avoidance, it caused the flies to exhibit a preference for cholesterol-containing food, thereby unmasking some unknown attractive mechanism.

We also set out to assess the requirements for the five IRs for cholesterol taste. Therefore, we performed two-way choice assays. All mutants showed defects in avoiding the sucrose containing 0.1% cholesterol over sucrose alone (Figures 3D and E). We then tested the behavior of the mutants across a range of cholesterol percentages (10^-5% to 10^-1%). We found that the Ir25a², Ir51b¹, Ir56d¹, and Ir76b¹ mutants exhibited reduced aversion across all cholesterol concentrations tested (Figure 3F). However, Ir7g¹showed a deficit in behavioral avoidance only at higher cholesterol percentages (10^-2 and 10^-1; Figure 3F). To test for rescue, we expressed the wild-type cDNA of each IR using its respective GAL4. The avoidance deficiencies in the five IR mutants were fully restored (Figure 3G).

To determine whether the IRs function in B GRNs, we performed rescue experiments and RNAi. In all cases, the mutant phenotypes were rescued using the B GRN driver (Gr33a-GAL4) in combination with the corresponding UAS-cDNA (Figure 3G). To perform RNAi, we also took advantage of the Gr33a-GAL4. We found that knockdown of each Ir in B GRNs eliminated cholesterol avoidance (Figure 3—figure supplement 1D). However, RNAi knockdown in D GRNs (ppk23-GAL4) had no impact on the repulsion to cholesterol (Figure 3—figure supplement 1E). Thus, we conclude that the Irs function in B GRNs.

We also conducted binary food choice assays to address whether olfaction contributed to the avoidance of cholesterol. We found that the orco null mutant (orco¹), which disrupts the olfactory co-receptor (ORCO) broadly required for olfaction, exhibited cholesterol repulsion similar to control flies (Figure 3—figure supplement 1F). Consistent with these findings, surgically ablating the antennae and maxillary palps, which are the main olfactory organs, did not diminish cholesterol avoidance (Figure 3—figure supplement 1G).

Ectopic co-expression of two sets of IRs confers responses to cholesterol

To explore whether IR7g, IR25a, IR51b, IR56d, and IR76b are sufficient to confer cholesterol sensitivity to GRNs that are normally unresponsive to cholesterol, we conducted ectopic expression experiments. We expressed the five Irs in all B GRNs (Gr33a-GAL4; Figure 4A) or in all A GRNs (Gr5a-GAL4; Figure 4C) and then characterized cholesterol-induced action potentials by focusing on cholesterol-insensitive I-type sensilla. Introducing all five Irs into cholesterol-insensitive, I9 sensilla elicited strong responses (Figure 4B). Misexpression of just Ir7g, Ir51b, and Ir56d also replicated cholesterol-induced responses (Figure 4B), presumably because Ir25a and Ir76b are endogenously expressed in GRNs in these sensilla (Lee, Poudel et al. 2018). Expression of any one of these Irs (Ir7g, Ir51b, and Ir56d) or combining Ir7g and Ir51b was insufficient to induce cholesterol sensitivity (Figure 4B). Of significance, we found that combining Ir56d with either Ir51b or Ir7g conferred cholesterol sensitivity to I9 sensilla (Figure 4B).

Figure 4.

L-type sensilla are missing B GRNs and are unresponsive to cholesterol. Therefore, we misexpressed all five Irs in A GRNs using the Gr5a–GAL4 (Figure 4C) and characterized action potentials in L6 sensilla. Ectopic expression of the Irs in the A GRNs also bestowed responsiveness to cholesterol (Figure 4D). Consistent with the results in B GRNs, co-expression of IR51b and IR56d, or IR7g and IR56d, was sufficient to confer cholesterol sensitivity (Figure 4D). This indicates that either of two groups of four IRs (IR7g, IR25a, IR56d, IR76b or IR25a, IR51b, IR56d, IR76b) is sufficient to comprise a functional cholesterol receptor.

Inducing attraction to cholesterol

Activation of A GRNs by sugars and several other attractive chemicals promotes feeding. Given that ectopic expression of Ir7g and Ir56d, or Ir51b and Ir56d alone was sufficient to induce a response to cholesterol in A GRNs, we investigated whether this would elicit attraction toward cholesterol. Control flies exhibited a preference for 2 mM sucrose alone over 2 mM sucrose laced with cholesterol (Figure 4E). Ir56d¹and Ir7g¹ mutant flies showed a slight avoidance of cholesterol-laced food. Flies carrying the Ir56d¹ or the Ir7g¹mutation and expressing both UAS-Ir7g and UAS-Ir51b in A GRNs exhibited similar behavior as Ir56d¹ or Ir7g¹ flies. However, when we introduced Ir7g and Ir56d, or Ir51b and Ir56d in the mutants, the flies exhibited attraction to the cholesterol-laced food (Figure 4E).

Discussion

The impact of cholesterol on an animal’s health depends on the concentration of cholesterol that is consumed (Huang, Hsu et al. 2024). While low levels are crucial, animals must avoid consuming excessive cholesterol (Beynen 1988, Soliman 2018, Zhang, Yuan et al. 2019). This differential effect of cholesterol is reminiscent of the impact of Na⁺ and Ca²⁺ on health, depending on their concentration (Zhang, Ni et al. 2013, Lee, Poudel et al. 2018). Flies have a bivalent reaction to Na⁺ depending on concentration but only avoid high Ca²⁺ and are indifferent to low Ca²⁺. Therefore, it was an open question as to whether flies have the capacity to taste cholesterol, and if so, whether they are endowed with the capacity to respond differentially to low and high concentrations or only avoid high cholesterol.

Several biochemical and biophysical studies focusing on the mammalian taste receptors T2R4 and T2R14 demonstrate that these receptors can bind and be activated or modulated by cholesterol (Pydi, Jafurulla et al. 2016, Shaik, Jaggupilli et al. 2019, Kim, Gumpper et al. 2024). While T2R4 and T2R14 are expressed in the taste system, they are expressed at high levels extraorally, such as in airway epithelial cells, pulmonary artery smooth muscle cells, and breast epithelial cells (Hariri, McMahon et al. 2017, Jaggupilli, Singh et al. 2017, Singh, Shaik et al. 2020). Therefore, it has been thought that these receptors may function in interoception, enabling the body to sense and respond to internal levels of cholesterol. Currently, evidence that these or other receptors function in cholesterol taste is lacking in mammals or any other animal.

Our research highlights the discovery that flies reject higher levels of cholesterol, but do not show attraction to low cholesterol even though flies cannot synthesize cholesterol and therefore must meet their needs for cholesterol from their diet. Nevertheless, the repulsion to a required substance is reminiscent of the fly’s response to Ca²⁺, which is aversive even though Ca²⁺ is required for life (Lee, Poudel et al. 2018). Moreover, we discovered that the repulsion to high cholesterol is mediated through the taste system since cholesterol stimulates action potentials in a subset of GRNs. The class of GRN that is activated by cholesterol is the B class, which also responds to bitter chemicals and other aversive tastants. However, cholesterol stimulates only a subset of the bitter responsive GRNs.

An unexpected observation is that the inhibition of B GRNs with Kir2.1 not only eliminates cholesterol repulsion but causes cholesterol to become highly attractive. This unmasking of an attractive mechanism for cholesterol when B GRNs are inhibited raises questions about the underlying neural circuitry and molecular mechanisms. Notably, this effect is unlikely to be due to cholesterol-induced activation of GRNs that promote feeding, since cholesterol does not activate any GRN in L-type sensilla, which are devoid of B GRNs, but include three types of attractive GRNs: A GRNs, C GRNs and other class of GRNs (E) (Montell 2021). Thus, we suggest that the attraction to cholesterol, which is unmasked by inhibition of B GRNs, occurs through a mechanism postsynaptic to the GRNs. Understanding the attractive mechanisms could provide valuable insights into how Drosophila regulates cholesterol intake based on internal nutritional states. This is particularly relevant given that Drosophila, like other insects, are cholesterol auxotrophs and must obtain sterols from their diet (Carvalho, Schwudke et al. 2010) Furthermore, elucidating the neural and molecular basis of cholesterol attraction and its potential modulation by internal metabolic states in Drosophila has the potential to reveal evolutionarily conserved mechanisms.

A key issue concerns the molecular identity of the Drosophila cholesterol taste receptor. We addressed this question using both electrophysiological and behavioral approaches to assess the impact of mutating genes encoding receptors belonging to the major families of fly taste receptors. We found that five IRs are required for cholesterol taste. These include two broadly required co-receptors (IR25a and IR76b) and three other receptors (IR7g, IR51b and IR56d). The contribution of five IRs to cholesterol taste was unanticipated since IRs are thought to be tetramers (Wicher and Miazzi 2021). Therefore, we conducted a series of ectopic expression experiments to determine whether all five IRs were necessary to confer cholesterol sensitivity to GRNs that do not normally respond to cholesterol. We found that either of two combinations of four IRs was sufficient to endow cholesterol responsiveness to GRNs. Three of the IRs were common to both receptors (IR25a, IR56d, IR76b). However, addition of either IR7g or IR51b as the fourth IR was necessary to generate a cholesterol receptor. It remains to be determined as to why there are two cholesterol receptors, why both are required for cholesterol taste, and why four rather three IR subunits comprise each receptor.

The findings reported here raise the question as to whether mammals such as mice and humans perceive cholesterol through the sense of taste. It is notable that in flies cholesterol taste depends on B GRNs, which also sense bitter compounds, and that in mammals T2Rs are activated by cholesterol, since this family of receptors also respond to bitter compounds. Therefore, it is intriguing to speculate that cholesterol taste may be aversive in humans.

Materials and Methods

Chemical reagents

The following chemicals and reagents were purchased from Sigma-Aldrich: cholesterol (catalog no. C4951), MβCD (catalog no. 332615), stigmasterol (catalog no. S2424), propionic acid (catalog no. 402907), butyric acid (catalog no. B103500), acetic acid (catalog no. A8976), caffeine (catalog no. C0750), denatonium benzoate (catalog no. D5765), sulforhodamine B (catalog no. 230162), tricholine citrate (TCC; catalog no. T0252), umbelliferone (catalog no. 24003), berberine sulfate hydrate (catalog no. B0451), cholesteroloroquine diphosphate salt (catalog no. C6628), lobeline hydrocholesteroloride (catalog no. 141879), quinine hydrocholesteroloride dihydrate (catalog no. Q1125), papaverine hydrocholesteroloride (catalog no. P3510), strychnine hydrocholesteroloride (catalog no. S8753), coumarin (catalog no. C4261), and sucrose (catalog no. S9378). Brilliant Blue FCF (catalog no. 027-12842) was purchased from Wako Pure Chemical Industry Ltd. The following antibodies were purchased from respective sources: mouse anti-GFP antibody (Molecular Probes, catalog no. A11120), rabbit anti-DsRed (TaKaRa Bio, catalog no. 632496), goat anti-mouse Alexa Fluor 488 (Thermo Fisher Scientific, catalog no. A11029), and goat anti-rabbit Alexa Fluor 568 (catalog no. A11011, Thermo Fisher Scientific/Invitrogen).

Binary food choice assay

In accordance with a previous study, we carried out experiments involving binary food choice tests (Aryal, Dhakal et al. 2022). To initiate the experiment, a group of 50−70 flies (aged 3−6 days, consisting of both males and females) were subjected to an 18-h period of fasting in a controlled humidity chamber. The subsequent procedures included the preparation of two distinct food sources, both incorporating 1% agarose as the base. The first food source was enriched with 2 mM sucrose, and the second source contained different concentrations of cholesterol in addition to 2 mM sucrose. To distinguish between these two food sources, we introduced blue food coloring dye (0.125 mg/mL brilliant blue FCF) to one and red food coloring dye (0.1 mg/mL sulforhodamine B) to the other. We evenly distributed these prepared solutions into the wells of a 72-well microtiter dish (Thermo Fisher Scientific, catalog no. 438733), alternating between the two options. Approximately 50−70 starved flies were introduced to the plate within approximately 30 min of food preparation. The flies were allowed to feed at room temperature (25°C) for 90 min, which occurred in a dark, humid environment to maintain consistent conditions. Afterward, the tested flies were carefully frozen at −20°C for further analysis. With the aid of a stereomicroscope, we observed and categorized the coloration of their abdomens as either blue (N_B), red (N_R), or purple (N_P). For each fly, we calculated the PI, a value derived from the combinations of dye and tastant, as follows: (N_B – N_R) / (N_R + N_B + N_P) or (N_R – N_B) / (N_R + N_B + N_P). A PI of either 1.0 or −1.0 indicated a significant preference toward one of the food alternatives, and a PI of 0.0 signified no bias among the flies toward either option.

Tip recording assay

The electrophysiological procedure, specifically the tip recording assay, was conducted in strict adherence to previously established protocols (Moon, Köttgen et al. 2006, Shrestha, Nhuchhen Pradhan et al. 2022). Flies of both sexes, aged between 4 and 7 days, were gently anesthetized on a bed of ice to facilitate the procedure. A reference glass electrode containing Ringer’s solution was meticulously inserted into the thoracic region of the flies. Subsequently, the electrode was incrementally advanced toward the proboscis of each fly. This precise process was repeated over multiple days to ensure the reliability and consistency of the results. To stimulate the sensilla, a recording pipette with a tip diameter ranging from 10 to 20 μm was connected to a preamplifier. The pipette was filled with a blend of chemical stimulants dissolved in a 30 mM TCC solution, which served as the electrolyte solution. Signal amplification was achieved through a Syntech signal connection interface box and a band-pass filter spanning a range of 100−3000 Hz. These amplified signals were recorded at a sampling rate of 12 kHz and subsequently analyzed using AutoSpike 3.1 software (Syntech). To ensure the integrity of the recorded signals, all recordings were carried out at regular 1-min intervals. This approach was meticulously enforced to guarantee the acquisition of precise and dependable data throughout the experiments.

Immunohistochemistry

Immunohistochemistry analysis was executed following established procedures (Lee, Kang et al. 2012). The labellum or brain of the flies were meticulously dissected and fixed using a 4% paraformaldehyde solution (Electron Microscopy Sciences, catalog no. 15710) in PBS-T (1X phosphate-buffered saline containing 0.2% Triton X-100) for 25 min at 4°C. The fixed tissues were thoroughly rinsed three times with PBS-T for 15 min each, precisely bisected using a razor blade, and then incubated for 30 min at room temperature in a blocking buffer composed of 0.5% goat serum in 1X PBS-T. Primary antibodies (1:1000 dilution; mouse anti-GFP [Molecular Probes, catalog no. A11120] and rabbit anti-DsRed [TaKaRa Bio, catalog no. 632496]) were added to freshly prepared blocking buffer and incubated with the samples overnight at 4°C. After overnight incubation, the samples underwent an additional round of thorough washing with PBS-T at 4°C before being exposed to secondary antibodies (1:200 dilution in blocking buffer; goat anti-mouse Alexa Fluor 488 [Thermo Fisher Scientific, catalog no. A11029] and goat anti-rabbit Alexa Fluor 568 [Thermo Fisher Scientific/Invitrogen, catalog no. A11011]) for 4 h at 4°C. After another three rounds of washing with PBS-T, the tissues were immersed in 1.25X PDA mounting buffer (37.5% glycerol, 187.5 mM NaCl, and 62.5 mM Tris, pH 8.8) and examined using an inverted Leica LASX confocal microscope for visualization and analysis.

Quantification and statistical analyses

We processed and conducted data analysis using GraphPad Prism version 8.0 (RRID: SCR 002798). Each experiment was independently replicated on different days, and the number of trials for each experiment is indicated as data points on the graphs. Error bars on the graphs represent the standard error of the mean (SEM). Single-factor ANOVA was combined with Scheffe’s post hoc analysis to compare multiple datasets. All statistical analyses were carried out using Origin (Origin Lab Corporation, RRID: SCR 002815). In the figures, asterisks are used to indicate statistical significance, with denotations of *P < 0.05 and **P < 0.01.

Additional files

Data availability

Source data for all figures contained in the manuscript have been deposited in ‘figshare’ (https://doi.org/10.6084/m9.figshare.28293062).

Supplementary figures

Figure 1—figure supplement 1.

Figure 2—figure supplement 1.

Figure 3—figure supplement 1.

Acknowledgements

This work was supported by grants to Y.L. from the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (RS-2021-NR058319) and the Korea Environmental Industry and Technology Institute (KEITI) grant funded by the Ministry of Environment of Korea. R.N.P was supported by the Global Scholarship Program for Foreign Graduate Students at Kookmin University in Korea. C.M is supported by grants from the National Institute on Deafness and other Communication Disorders (DC007864 and DC016278).

Additional information

Author contributions

Roshani Nhuchhen Pradhan: Conceptualization, Investigation, Methodology, Analysis, Writing – Original draft. Craig Montell: Conceptualization, Funding acquisition, Writing – Review & Editing. Youngseok Lee: Conceptualization, Funding acquisition, Supervision, Writing – Review & Editing.