1. Neuroscience
Download icon

Taste sensing and sugar detection mechanisms in Drosophila larval primary taste center

  1. G Larisa Maier
  2. Nikita Komarov
  3. Felix Meyenhofer
  4. Jae Young Kwon
  5. Simon G Sprecher  Is a corresponding author
  1. Department of Biology, University of Fribourg, Switzerland
  2. Department of Biological Sciences, Sungkyunkwan University, Republic of Korea
Research Article
  • Cited 0
  • Views 635
  • Annotations
Cite this article as: eLife 2021;10:e67844 doi: 10.7554/eLife.67844

Abstract

Despite the small number of gustatory sense neurons, Drosophila larvae are able to sense a wide range of chemicals. Although evidence for taste multimodality has been provided in single neurons, an overview of gustatory responses at the periphery is missing and hereby we explore whole-organ calcium imaging of the external taste center. We find that neurons can be activated by different combinations of taste modalities, including opposite hedonic valence and identify distinct temporal dynamics of response. Although sweet sensing has not been fully characterized so far in the external larval gustatory organ, we recorded responses elicited by sugar. Previous findings established that larval sugar sensing relies on the Gr43a pharyngeal receptor, but the question remains if external neurons contribute to this taste. Here, we postulate that external and internal gustation use distinct and complementary mechanisms in sugar sensing and we identify external sucrose sensing neurons.

Editor's evaluation

This paper provides an alternative theory about taste coding in larval insects. The authors use a volumetric imaging method to capture neural activity from an entire taste sensory organ in Drosophila larvae. Their results suggest that taste coding in Drosophila larvae might be different than adult flies, and use multimodal neurons and neural dynamics to represent taste information in the larval brain. Understanding taste coding in Drosophila larvae might provide further understanding to how food is encoded in the larval insect brain.

https://doi.org/10.7554/eLife.67844.sa0

Introduction

The sense of taste provides an innate ability among animals to avoid toxic or deleterious food and to choose a nutritious diet. Despite comprising only few chemosensory neurons at the periphery (Stocker, 1994; Stocker, 2008), Drosophila larvae are able to sense a wide range of chemicals but the principles of how a small neuronal population encodes a large sensory space entail unresolved questions (Komarov and Sprecher, 2022). They display behaviors to different gustatory cues (Gerber and Stocker, 2007), being repelled by bitter tastants (El-Keredy et al., 2012; Kim et al., 2016) as possible signals of toxic food, and attracted by sweet, certain amino acids (Schipanski et al., 2008; Croset et al., 2016; Kudow et al., 2017), or ribonucleosides (Mishra et al., 2018), indicators of a nutritious diet. For Drosophila larvae, an animal that increases body size up to 200–250-fold in 5 days (Robertson, 2009) and spends the bulk of its time searching for food (Green et al., 1983), the most important nutritional sources are sugar, such as sucrose and fructose found in fermenting fruits – their natural diet, as well as proteins and fatty acids, taken up from yeast growing on this substrate. As one of the most important macronutrients in the larval diet, sugar is associated with high nutritional values for survival and with proper function of metabolic processes. Nevertheless, larval external taste neurons have not been associated thus far to a role in sugar taste preference.

In the fruit fly, taste encoding at the periphery has been described in line with segregated hedonic valence logic, attractive and aversive taste being detected through separate cell populations. Accordingly, specific members of the Gustatory receptors (GRs) family mediate sweet or bitter taste, namely cells expressing Gr5a, Gr61a, or Gr64a-f detect sugars and do not overlap with bitter-taste cells expressing Gr66a or Gr33a (Dahanukar et al., 2007; Fujii et al., 2015; Thorne et al., 2004; Liman et al., 2014; French et al., 2015). Additional taste modalities such as salt, fatty acid, carbonation, polyamines, or amino acids partially overlap onto these by means of added-on expression of Ionotropic receptors (IRs) or pickpocket receptors (Ppk) (Zhang et al., 2013; Hussain et al., 2016; Lee et al., 2017; Tauber et al., 2017; Ahn et al., 2017; Ganguly et al., 2017; Sánchez-Alcañiz et al., 2018; Jaeger et al., 2018), but maintaining a segregated attraction/aversion quality of taste in the separate cell populations. Contrary to described adult fly taste sensing logic, in Drosophila larva taste of opposed valence can overlap, shown by a single gustatory receptor neuron (GRN) activated by both sugar and bitter tastants (van Giesen et al., 2016a) and thereby indicating a higher level of taste integration than in the adult fly.

Chemosensory organs are located at the tip of larval head, and comprise bipolar sensory neurons that extend axonal projections into defined regions of the brain (Figure 1A; Python and Stocker, 2002). The main external sense organs are the terminal organ ganglion (TOG) with taste primary function and the dorsal organ ganglion (DOG) as olfactory organ (Stocker, 1994; Klein et al., 2015). Out of more than 30 GRNs in the TOG only seven neuronal identities (C1–C7) have been mapped and are traceable by means of individual GR-Gal4 lines (Figure 1B; van Giesen et al., 2016a; Kwon et al., 2011; Rist and Thum, 2017). Because of experimental restrictions in probing the system due to the limited availability of individually mapped TOG Gal4 lines, physiological characterization has been thus far tackled in few single GRNs, but a global approach for measuring responses in larval taste sense neurons could prove useful to an overview of peripheral gustatory responses.

Figure 1 with 2 supplements see all
Peripheral taste neurons of the fruit fly larva.

(A) The chemosensory system of the larva. External sense organs extend dendrites to the periphery and have main olfactory function—the DOG (dorsal organ ganglion), or gustatory function—the TOG (terminal organ ganglion) and the VOG (ventral organ ganglion). Dendrites of the internal dorsal, ventral, and posterior pharyngeal sense organs (DPS, VPS, and PPS) innervate the pharynx, thus involved in taste sensing during food ingestion. All chemosensory neurons project axons to the brain in the subesophageal ganglion (SOG)—first central taste integration relay—or to the antennal lobe (AL) for central olfactory processing. (A) Has been adapted from (C) from Gerber and Stocker, 2007. (B) External chemosensory organs. Seven GSNs previously identified and named C1–C7 are represented here by the color code they were first described with (van Giesen et al., 2016a; Kwon et al., 2011). Used Gal4 lines: Gr22e-Gal4 (C1), Gr94a-Gal4 (C2), Gr66a-Gal4 (C1, C2, C3, and C4), Gr59e-Gal4 (C5), Gr21a-Gal4 (C6), and GMR57BO4-Gal4 (C7). (C) UAS-myrGFP reporter was expressed in individual GSNs using corresponding Gal4 lines (B). We observed a relatively stereotypic position of specific neurons within the organ across animals (n≥3)—exemplified on C6 (Gr21a-Gal4) and C1 (Gr22e-Gal4). (D) Illustrative 3D map of TOG segmented cells. Position of neurons with known identities (white dots) is approximated based on separate immunostainings.

© 2007, Gerber and Stocker. Figure 1A has been adapted from Figure 1C from Gerber and Stocker, 2007.

Therewithal, larval repertoire of taste receptors differs from expression described in adult flies, with many GRs showing developmental stage-specificity. Gr66a/Gr33a-expressing neurons, mediators of bitter perception in the adult fly, are also present in the larva and have been associated with roles in bitter taste avoidance (Kim et al., 2016; Apostolopoulou et al., 2014; Choi et al., 2020). However, sweet sensing neurons or receptors have not been identified in the larval peripheral taste system (Kwon et al., 2011), although larvae can detect and elicit behaviors towards most sugars (Rohwedder et al., 2012). Larval sugar perception has been allotted to Gr43a fructose receptor (Mishra et al., 2013), which, however, is expressed in pharyngeal sensory organs, foregut and in the brain, but absent from external taste neurons, similar to its expression in the adult fly where it acts as main hemolymph fructose sensor (Miyamoto et al., 2012; Miyamoto and Amrein, 2014). As one of the most important macronutrients in larval diet, and therefore it seems surprising that external taste organs at direct contact with animal’s environment would not be involved in sugar detection and output behavior, all the more so as sucrose can elicit physiological response in TOG neurons (van Giesen et al., 2016a).

In this study, we probe response profiles of TOG taste neurons by means of whole-organ in-vivo calcium imaging and subsequently pursue to investigate sugar peripheral taste in the larva. Our experimental approach allows an assessment of general tuning of larval GRNs and identification of distinct response dynamics, while revealing a high fraction of calcium response to sucrose. We test larval behavior and tackle the described central role of Gr43a receptor in sugar taste by separating external taste sensing from internal taste. Segregation of internal and external larval taste neurons in behavior has also been recently established for bitter taste (Choi et al., 2020). We equally explore the contribution of individual TOG neurons in sugar taste and identify C2 as one neuron participating in peripheral sucrose detection in the larva. Our study reveals a high complexity in taste response at the level of external taste neurons in the fruit fly larva, and we propose that distinct neurons are employed in sugar sensing depending on the dietary substrate.

Results

Whole organ larval taste recordings with cellular resolution

At the level of the TOG, main larval external taste organ containing around 30 sense neurons, 7 GRNs have been identified based on receptor-Gal4 expression (van Giesen et al., 2016a; Kwon et al., 2011). To assess their spatial location, we performed immunostainings of the seven individual TOG neurons (Figure 1C and D). Although cell positions vary slightly between stainings (n≥3) (Figure 1C), 3D locations can be approximated on an illustrative map (Figure 1D). Four individual TOG neurons have previously been characterized physiologically, revealing taste multimodality as a characteristic in the larval sensory system (van Giesen et al., 2016a). Considering that molecular mapping for all TOG neurons through single driver lines has not yet been compiled, physiological characterization of the larval external taste organ through labeling individual neurons is not readily achievable. To circumvent this constraint in direction for a response profile overview of larval GRNs in the TOG we implemented a whole-organ imaging approach.

We performed in-vivo whole-organ calcium imaging recordings by expressing cytoplasmic GCaMP6m (Chen et al., 2013) and nuclear RFP (Egger et al., 2007) reporters in all neurons (Figure 2A, Figure 1—figure supplement 1, Figure 1—video 1). The use of a nuclear reporter was chosen for efficient cellular segmentation and for establishing a reproducible data processing pipeline for whole organ recordings (Figure 1—figure supplement 1—source data 1). As previously described (van Giesen et al., 2016b), we made use of a customized microfluidic chamber allowing for simultaneous imaging and chemical stimulation. We used a series of chemical compounds belonging to different canonical taste categories in concentrations that have been previously used in similar studies in the fruit fly (van Giesen et al., 2016a; Park and Carlson, 2018; Ling et al., 2014; Table 1, Figure 2, see Materials and methods).

Figure 2 with 2 supplements see all
Whole-organ calcium imaging shows an overview of response tuning of GRNs.

(A) Representative recording of the larval primary taste organ stimulated with citric acid 100 mM. Cytoplasmic expression of GCaMP6m and nuclear expression of RFP in all neurons using nsyb-Gal4 driver line (upper left panel) and cell segmentation (white spots, lower left panel). Responding neurons are indicated by blue, red, and magenta arrows (upper right panel) and corresponding dots and fluorescence traces in the same colors (lower right panels). (B) Stimulation with groups of tastants by taste modality, in order: two sugar categories (mono and disaccharides), low and high concentration salts, and four amino acid groups. A response intensity threshold of 20% DF/F was considered and neuronal responses across different preparations were pooled together. Among all neurons responding to sugar group stimulation, a comparable ratio of cells showed activation to either one or the other category and 25.4% to both groups of mono and disaccharides. For salt-responding neurons, a higher proportion of neurons showed activation to high salt than to low salt concentration, and 20% neurons responded to both categories. Amino acids split into four groups gave a rounded 62% of uni-group responses and the rest of 38% neurons responded to two or more amino acid groups per animal. (C) Responses in animals stimulated with two groups of bitter compounds or with both sugar and bitter test groups. Within the total number of bitter-responsive neurons, 55% were uniquely activated by the DSoTC group and respectively 30% by QLSC group. In animals stimulated with both modality groups, integration between sweet and bitter taste has been calculated and we observed similar proportions of taste cells responding only to bitter groups (30.3%), cells responding only to sugar (39.4%) and cells responding to groups of both modalities (30.3%). (D) GRN responses to series of individual tastants. Taste stimulation was done using a series of five tastants per animal, each belonging to one of the five canonical taste modalities represented by sucrose 500 mM (sweet/red color), denatonium 10 mM or quinine 5 mM (bitter/blue), NaCl 1 M for high salt or NaCl 100 mM for low salt (salt/purple), valine 100 mM or arginine 100 mM (amino acids/brown) and citric acid (CiA) 100 mM (sour/yellow). Seven and respectively eight separate organs were recorded within the two series of stimulation (N=15) and a total of 197 neurons showed responses above 20% DF/F0 and were further analyzed. The table shows illustrative traces for cells responding to one or more tastants in two separate representations, each for one of the two stimulation sets. (D’) Total percentages of taste integration in pooled data are represented on the pie chart: 68% of total responding neurons were activated by only one tastant per organ and up to 32% of neurons responded to more than one tastant. Most unimodal responses (uni-taste/cell, left extended pie) were recorded to sucrose and to high salt. Conversely, numerous neurons were activated by different combinations of taste categories, with different frequency of occurrence (multi-taste/cell, right extended pie). Among the most frequent combinations of taste modalities activating the same cell we noted sucrose (sweet)+ any other taste category. Some of these co-modalities involve a presumed positive valence taste (sucrose/sweet) together with a negative valence taste (bitter or high salt) sensed by the same neuron.

Figure 2—source data 1

Referring to graphs B & C.

Excel files comprising neuronal calcium imaging data: sheet 1 represents a summary of cell and response numbers to different modalities, sheet 2 contains the list of neurons activated by taste groups of sugar, salts, and amino acids, and sheet 3 contains the list of neurons activated by taste groups of sugar and bitter in Figure 2C. Each cell in the list received a cell name (cell_name_in_prep) as identifier for the given biological replicate, not valid for cell mapping.

https://cdn.elifesciences.org/articles/67844/elife-67844-fig2-data1-v2.xlsx
Figure 2—source data 2

Excel sheet one refers to data in Figure 2D’, comprising the list of neuronal responses above 20% DF/F0.

Signals below this threshold have not been used for figure generation; cell_name_in_prep is the cell identifier in a given biological preparation; map_cell_name refers to cell response mapping presented in Figure 3 and Figure 3—figure supplement 1: all activation signals were considered comprising both ON and OFF responses and map_cell_name is the id for cell mapping across different biological preparations.

https://cdn.elifesciences.org/articles/67844/elife-67844-fig2-data2-v2.xlsx
Table 1
Chemicals used for taste stimulation in calcium imaging.
Tastant or taste groupCompoundAssociated taste (humans)ConcentrationGroup final Concentration
Tastant series 1SucroseSweet500 mM
Denatonium benzoateBitter10 mM
ValineAmino acid/Bitter100 mM
NaClSalty1 M
Citric acidSour100 mM
Tastant series 2SucroseSweet500 mM
QuinineBitter5 mM
ArginineAmino acid/Bitter100 mM
NaClSalty100 mM
Citric acidSour100 mM
Group sugars MonosaccharidesFructoseSweet100 mM500 mM
GlucoseSweet100 mM
ArabinoseSweet100 mM
MannoseSweet100 mM
GalactoseSweet100 mM
Group sugars DisaccharidesSucroseSweet100 mM500 mM
TrehaloseSweet100 mM
MaltoseSweet100 mM
LactoseSweet100 mM
CellobioseSweet100 mM
Group bitter DSoTCDenatonium benz.Bitter1 mM22 mM
Sucrose octaacetateBitter1 mM
TheophyllineBitter10 mM
CoumarinBitter10 mM
Group bitter QLSCQuinineBitter1 mM13 mM
LobelineBitter1 mM
StrychnineBitter1 mM
CaffeineBitter10 mM
Group Amino acids AValineBitter10 mM60 mM
LeucineBitter10 mM
IsoleucineBitter10 mM
MethionineBitter10 mM
TryptophanBitter10 mM
CysteineBitter10 mM
Group Amino acids BAlanineSweet10 mM41 mM
PhenylalanineBitter10 mM
GlycineSweet10 mM
ProlineSweet10 mM
TyrosineBitter1 mM
Group Amino acids CArginineBitter10 mM50 mM
LysineSalty-Bitter10 mM
Aspartic acidUmami10 mM
Glutamic acidUmami10 mM
HistidineBitter10 mM
Group Amino acids DSerineSweet10 mM40 mM
ThreonineSweet10 mM
AsparagineNeutral10 mM
GlutamineSweet10 mM
Group high saltNaClSalty500 mM1M
KClSalty500 mM
Group low saltNaClSalty25 mM50 mM
KClSalty25 mM

To start with, we grouped gustatory cues used for GRN stimulation in groups by taste category for sweet, bitter, amino acid, and salt taste modalities (Table 1) and pooled all neuronal responses for the ratio of taste modality integration per cell (Figure 2B and C). In the sweet taste category, we tested two sugar groups chosen as monosaccharides (fructose, glucose, arabinose, mannose, and galactose) and disaccharides (sucrose, trehalose, maltose, lactose, and cellobiose) and calculated approximately 74% of sugar-responsive neurons to be activated by either monosaccharides (41%) or disaccharides (33%), and approx. 25% by both groups (Figure 2B, left panel). For salt taste stimulation, we mixed NaCl and KCl at final concentration of 50 mM for low salt group, and at 1 M for high salt group. We observed a low percentage of cell activation overlap among salt-responding cells with 20% activated by both high and low concentrations (Figure 2B, middle panel) and we recorded more responses to high salt (56%) than to low salt. The 20 proteinogenic amino acids were compiled in four groups (A, B, C, and D) as previously described by Park and Carlson, 2018 (Table 1, see Materials and methods). 38% amino acid-responding neurons showed activation to two or more amino acid groups (Figure 2B, right panel). Bitter tastants were randomized in two groups of four substances each (Table 1, see Materials and methods): DSoTC group (denatonium benzoate, sucrose octaacetate, theophylline, and coumarin), and respectively QLSC group (quinine, lobeline, strychnine, and caffeine) chosen at previously used concentrations (Ling et al., 2014). Among cells responsive to bitter stimulation, we observed a bigger fraction of cells activated by DSoTC group (55%). In animals tested for sweet and bitter taste response, we identified cells activated by both sugar and bitter groups (30%, Figure 2C, right panel), in agreement with cell integration of opposite valence tastants.

Sucrose elicits neuronal responses in the TOG

Further, we focused GRN stimulation on few specific compounds representative of the five canonical taste categories: sweet, bitter, salt, amino acids, and sour. Each animal was stimulated with a representative substance from each category, either with sucrose (sweet), denatonium (bitter), NaCl 1 M (high salt), valine (amino acid), and citric acid (sour) or with sucrose (sweet), quinine (bitter), NaCl 100 mM (low salt), arginine (amino acid), and citric acid (sour) (Figure 2D). We recorded a 15 total of TOG organs and we selected activation responses calculated above a threshold of 20% DF/F0. Out of all responding neurons (N=197, 15 organs), 68% showed activation to only one tastant per animal and most unimodal responses were recorded to sucrose and to high salt (Figure 2D’, left extended pie chart). Up to 32% responding neurons were activated by more than one test substance used for stimulation of the animal. The most frequent taste modalities combinations eliciting response in the same neuron within a given preparation (Figure 2D’, right extended pie chart) were: sucrose+ citric acid, sucrose and valine/arginine (sweet+ amino acid, 17%), sucrose and high salt (sweet+ high salt, 19%), sucrose and low salt (10%), sucrose and denatonium/quinine (sweet+ bitter, 9%), sucrose and citric acid (9%), citric acid and high salt (sour+ high salt, 12%), and finally amino acid and low salt (10%). Interestingly, therefore, sucrose was the tastant corresponding to several uni-taste/cell responses but also to most frequent taste combinations that activate the same neuron.

The above fractions observed across pooled data recapitulate the responses recorded per each organ, as represented in Figure 2—figure supplement 1, with an average of taste modality integration per cell per organ of 67% multimodality/33% unimodality, for a mean of 19 responses per organ, 13 responding neurons, and 5 sucrose-activated neurons per organ detected in average.

To test if order of tastant presentation would generally alter neuronal responses within a series of stimulations, we looked into responses of C7 to sucrose 500 mM before or after the denatonium stimulation. C7 is a TOG neuron previously shown to respond to both sugar and bitter substances (van Giesen et al., 2016a), but we observed no significant difference in sucrose-elicited response in this neuron based on order of sweet and bitter presentation (Figure 2—figure supplement 2).

Distinct response temporal dynamics

The observations described above for grouped or individual series of tastant stimulations refer to canonical recorded calcium signals characterized by a rise in GCaMP fluorescence during stimulus application. However, at a closer look, tastants elicit different types of signals within the TOG organ, besides canonical ON response, neurons also showing OFF responses and late activation or even fluorescence intensity drop upon stimulation, as exemplified in Figure 3—figure supplement 1A for sucrose 500 mM, citric acid 100 mM, or NaCl 100 mM. We proceeded to manually map responses to organ across different larvae and for this purpose, we compiled all activation signals of early/ON responses or late onset/OFF responses with a fluorescence rise above 20% response threshold. Four neurons localized anterior-laterally to the other cells (named Anterior Lateral Neurons [ALNs], Figure 3 insets) served as landmark in determining the relative spatial location of a cell. We mapped a tastant to a cell location if it associated a response at the respective identity in more than four organs (n≥4, Figure 3A, Figure 3—figure supplement 1B, C), except for bitter-evoked responses for which n≥3 (Figure 3—figure supplement 1B, C). Mapped cells were named based on their spatial location (e.g., PM1=posterior-medial 1, Figure 3B).

Figure 3 with 1 supplement see all
Taste-elicited responses mapped to neurons across animals.

(A) Illustrative traces for a given tastant stimulation from three different organs are shown for manually mapped neurons. (B) Representative map used for matching neuronal identities across different recorded organs. Names were chosen as reference for difference in neighboring cell positions, suggested with respect to the sagittal plane, anterior-posterior, and ventral-dorsal axis (C=central/center, V=ventral, D=dorsal, P=posterior, A=anterior, M=medial/n, L=lateral).

Figure 3—source data 1

Referring to Figure 3 and Figure 3—figure supplement 1C; excel file comprising responses associated to mapped cells.

Data were considered for responses mapping to a given cell (map_cell_name) at least in four different preparations, N≥4 (exception for bitter stimulation, N≥3). CDL2 neuron associates deactivation responses as represented in Figure 3—figure supplement 1B.

https://cdn.elifesciences.org/articles/67844/elife-67844-fig3-data1-v2.xlsx

In line with unmapped responses summarized in Figure 2D, by mapping cellular calcium signals, we found neurons with different tuning profiles and sucrose responses were mapped to different TOG cells (Figure 3—figure supplement 1C, Figure 3—source data 1).

Tastant-evoked deactivation (fluorescence drop) was identified for different test substances (Figure 3—figure supplement 1A) but mapping this type of signals was evident especially in the case of one neuron in the organ named CDL2 (central-dorsal-lateral neuron 2). In particular for citric acid, this neuron was often recorded to respond with a decrease of fluorescence while its neighbor, CDL1, typically showed canonical calcium activation to citric acid (Figure 3A, Figure 3—figure supplement 1B).

Sugar sensing in Drosophila larva

TOG neurons can detect and contribute to sucrose preference behavior

Larval sugar taste has been associated with the pharyngeal and brain expression of Gr43a receptor (Mishra et al., 2013). Yet, as we describe above and as previously shown (van Giesen et al., 2016a), sucrose elicits calcium responses in external chemosensory neurons, raising the question of a potential additional role of TOG neurons in sugar detection, which we set to explore next in this study.

To start with, by means of cell silencing using tetanus toxin reporter UAS-TNTE (Sweeney et al., 1995), we looked into the role of individual TOG neurons that have previously been shown calcium imaging responses to sucrose, respectively C2, C6, and C7 (Figure 4). Behavioral preference of L3, 4-day-old larvae were tested as described before (van Giesen et al., 2016a) by way of two-choice test after 5 min on 2.5% agarose petri-dishes (see Materials and methods). Among the three probed neurons, only C2-silencing showed a preference index different from controls (Figure 4A, one-way ANOVA p<0.05), while C6 or C7 silencing did not show a notable behavioral defect. We equally tested Gal4 lines yielding expression in several neurons that also include C2 (Figure 4—figure supplement 1A) and observed a defect for silencing these lines, although more diminished than for C2 silencing alone and not significantly different from controls. Silencing C5 neuron, TOG neuron that does not express expressed Gr66a/Gr33a bitter co-receptors, also showed a trend for a reduced preference (Figure 4—figure supplement 1A).

Figure 4 with 1 supplement see all
Mechanism of larval sugar taste at the periphery.

(A) Behavioral tests on C2, C6, or C7 neuron silenced larvae. Although all three neurons have previously been associated with physiological responses to sucrose 500 mM (van Giesen et al., 2016a), a defect in preference index to Sucrose 500 mM compared to control larvae is found only for C2 (p-value=0.038), indicating a direct contribution of this neuron in sucrose sensing. Driver lines: Gr94a-Gal4 (C2), Gr63a-Gal4 (C6), and GMR57BO4-Gal4 (C7). N=10 for all genotypes; one-way ANOVA; *p<0.05, **p<0.005, ***p<0.001; plot as mean with SEM. (B) The role of the C2 neuron in preference to a range of sucrose concentrations on 1% versus 2.5% agarose. Larvae with C2 silenced show a greater defect for sugar preference on 2.5% agarose compared to 1% for all concentrations. This suggests larvae rely at least partly on C2 in sucrose sensing on a denser substrate. At 1%, C2 silencing has a lesser effect on sugar preference, suggesting that other mechanisms come into play for behavior on a softer substrate. N=10 for all genotypes; one-way ANOVA, *p<0.05, **p<0.005, ***p<0.001; plot as mean with SEM. (C) Food ingestion measurement using blue dye. Larvae were tested at different timepoints to compare eating on agarose 1% versus agarose 2.5%, both containing Sucrose 500 mM. Blue coloring of the intestine was observed as early as 2 min on 1% concentration, while on 2.5% animals didn’t seem to be able to eat as easily as indicated by the lack of tint of their abdomen at 2 min and 5 min testing. For longer time points, larvae also start showing food ingestion on 2.5% agarose, but the intestinal blue dye is less prominent than on 1% concentration. This suggests that animals easily eat the softer 1% food substrate but that on denser 2.5% concentration ingestion is slower and evidently impaired especially for early test time points. (D) Temporally restricted inhibition of C2 in sucrose 500 mM sensing. Alternative silencing of C2 using G. theta anion channelrhodopsin 1 (GtACR1) confirms the observed defect for sugar preference on 2.5% agarose across the timelapse. In dark conditions (no light activation), a defect is not noted. N=7–8; one-way ANOVA, *p<0.05, **p<0.005, ***p<0.001; plot as mean with SEM.

Figure 4—source data 1

Referring to Figure 4.

The excel file contains larval preference indexes and p-values Mann-Whitney test comparisons shown in the figure.

https://cdn.elifesciences.org/articles/67844/elife-67844-fig4-data1-v2.xlsx

Although it expressed Gr66a/Gr33a bitter co-receptors, a possible role of C2 in detecting an attractive tastant is not conflicting with previous observations. Behavioral taste screening studies have not linked C2 with bitter taste detection by contrast with other Gr66a-expressing neurons (El-Keredy et al., 2012; Kim et al., 2016) and, furthermore, optogenetic activation of C2 actually associates a ‘weakly attractive’ larval response (Hernandez-Nunez et al., 2015).

C2 silencing shows a defect in sucrose preference at different concentrations—100 mM, 500 mM, and 1 M, measured at different time points (2, 5, and 15 min Figure 4B— upper panels).

Given that Gr43a receptor has been described to carry a predominant role in larval sugar detection, we noted agarose concentration differences between our tests and the behavioral assay described by Mishra et al., 2013. Our two-choice assays have been developed on 2.5% agarose substrate (see Materials and methods), while Mishra et al. performed tests using 1% concentration. Interestingly, when we tested sucrose preference on 1% substrate, we observed a diminishment of the defect in C2-silenced larvae compared to control (Figure 4B, lower panels).

Larvae can detect sugar with or without food ingestion

To test if the agarose concentration difference reflects a divergent larval capacity to eat, we compared the two substrates by adding sucrose 500 mM and Brilliant Blue 2% and checked for ingested agarose in the larval intestine (Figure 4C) at different time points. Excitingly, we detected blue dye in the animal intestine after as early as 2 min in larvae probed on the 1% agarose substrate but no blue coloration in the abdomen on larvae from 2.5% agarose. At 5 min preference—representative time point we generally employed in two-choice experiments as, for example, in Figure 4A —we could equally see the difference in ingested agarose, reflected by more blue dye intake in larvae wandering on 1% substrate compared to larvae on 2.5%. After 15 and respectively 30 min, larvae probed on 2.5% agarose also appeared blue colored, but visibly less than larvae tested on 1% concentration (Figure 4C). These observations suggest that on 1% agarose larvae can more readily ingest and therefore likely employ the internal taste organs for navigating the environment, while on a more solid consistency of 2.5% agarose larvae cannot immediately ingest the substrate and probably rely on external chemosensory organs for sampling their food substrate.

Complementary taste sensing mechanism for larval sugar preference

To test this hypothesis, we assessed the role of Gr43a receptor in the detection of sucrose 100 mM and 500 mM, as well as to fructose 100 mM and 500 mM on both 1% and 2.5% agarose (Figure 4—figure supplement 1B, B'). We have employed the same genetics and fly lines as previously described by Mishra et al., 2013, testing Gr43a mutant, Gr43a rescue, and ‘Gr43a brain-only’ larvae, which, as described in the referenced study, lack Gr43a peripheral expression but keep brain receptor expression intact.

The Gr43a mutant showed diminished preference at different time points to both sucrose concentrations on 1% agarose (Figure 4—figure supplement 1B, upper panels), whereas on 2.5% mutant preference was more similar to rescue control larvae (Figure 4—figure supplement 1B, lower panels). Similar observations could be made for the ‘Gr43a brain-only’ larvae tested on sucrose 500 mM that showed a defect at immediate preference on 1% but no defect on 2.5%.

When preference to fructose 100 mM was tested on 1% agarose, Gr43a mutant and ‘Gr43a brain-only’ showed a completely abolished immediate preference (2 min) (Figure 4—figure supplement 1B', upper left panel), as expected from previously reported results (Mishra et al., 2013). Moreover, equally reproducing observations from the referenced study for Gr43a, the ‘brain-only’ larvae gradually rescue the defect, in line with the idea that brain receptors compensate fructose sensing at delayed preference (15 min) (Mishra et al., 2013). Interestingly, the mutant defect appears to be lesser at higher fructose concentrations and especially on 2.5% substrate concentration (Figure 4—figure supplement 1B'). These observations suggest that larvae deficient of Gr43a receptor in pharyngeal taste neurons rely on other taste sensing mechanisms when unable to eat, such as via the C2 neuron in the TOG, and also that Gr43a in the larva, as in the adult (Miyamoto et al., 2012), is more attuned to fructose sensing than to sucrose.

We also tested Gr43a brain silenced larvae (Figure 4—figure supplement 1C) and although the reference study (Mishra et al., 2013) described no defect for brain-silenced receptor, we found an impaired immediate preference on 1% but not on 2.5%.

We further proceeded to reassess sucrose detection of C2 by using an optogenetics approach as alternative to TNTE neuronal silencing. Optogenetic neuronal silencing of C2 using anion channelrhodopsin (GtACR1) expression (Mauss et al., 2017) (see Materials and methods) led to an altered preference to sucrose 500 mM in larvae tested on 2.5% (Figure 4D).

All in all, our observations subscribe to the hypothesis that C2 and possibly other TOG neurons complement internal taste in sugar detection and that larvae can use distinct sensing paths—pharyngeal neurons being employed for sugar taste during food ingestion, whereas external taste neurons are participating in sugar detection especially when eating is not accessible (Figure 5). Thus far, our observations reveal partial roles in sucrose and fructose sensing associated to C2 or C5 neurons, and therefore it is probable that several TOG neurons might contribute to the output behavior in sugar response, either through a direct role or indirectly by computing a mixture of sucrose and other tastants as previously shown for C7.

Peripheral taste sensing in Drosophila larvae—contribution to sugar and bitter taste of GRNs at different anatomical levels.

(A) Schematic representation of the larval head and the chemosensory system. Bitter tastants detected at the level of external sense neurons are quinine in C3 neuron (Apostolopoulou et al., 2014) and denatonium in C7 and C1 (van Giesen et al., 2016a; Choi et al., 2020) but not caffeine, the latter being sensed through DP1 pharyngeal neuron (Choi et al., 2016; Apostolopoulou et al., 2016). Similarly, sugar taste seems to be also segregated at different levels of detection—brain Gr43a receptor neurons detect hemolymph sugar (Mishra et al., 2013), Gr43a expression in the pharyngeal organs (DP4 neuron) is central in sensing ingested sugar, while neurons like C2 and C5 contribute to sucrose preference at external level especially when food ingestion is less accessible (Figure 4 and Figure 4—figure supplement 1). As the first gate for food evaluation, it is important to note the capacity of TOG neurons for integration of opposed valence taste wither within the same neuron as in C7 (van Giesen et al., 2016a), or by co-activation of a sugar sensing with a bitter sensing neuron as suggested for C2 and C1 (Hernandez-Nunez et al., 2015). (B) Schematic representation of the proposed model in larval sugar sensing. Upper panel: When unable to ingest as readily, larvae utilize C2 and/or other TOG neurons to guide sugar preference via a C2/TOG pathway. Gr43a is the main internal nutrient sensor, expressed in the pharyngeal taste neurons, proventriculus (gut), and brain. While the lack of ingestion prevents pharyngeal-expressing Gr43a neurons from contributing to immediate sugar preference, internal metabolic state signals may alter behavioral output. Lower panel: upon ingestion, internal sensing mechanisms take over sugar-associated preference most specifically for fructose, with a diminished role of external neurons (C2/TOG).

Discussion

In this study, we explore physiological and molecular characteristics of Drosophila larval taste neurons using a whole-organ imaging approach. We designed an experimental framework for calcium imaging in all GRNs comprised by the larval primary taste-sensing organ, the TOG (Figure 1—figure supplement 1, Figure 1—video 1). Our approach complements previous strategies based on targeting single or discrete subpopulations of neurons (Croset et al., 2016; Sánchez-Alcañiz et al., 2018; van Giesen et al., 2016a; Kwon et al., 2011; Stewart et al., 2015), allowing for an overview of activation patterns within the organ. As previously shown in Drosophila larvae, concurrent activation of two neurons can differ from their separate activation readout (Hernandez-Nunez et al., 2015) as specific sense neurons firing together might dictate the relevance of a stimulus and ultimately its behavioral output. Coherently, each tastant used here typically elicits response in several GRNs in support of a combinatorial sensing, while also raising the question about the redundancy in contribution of distinct neurons recruited in response to a given taste quality, to be elucidated by future studies.

For the tested substances at the concentrations chosen within this study, we observed roughly a 60–70% versus 30–40% division of unimodal versus multimodal responding GRNs (Figure 2 and Figure 2—figure supplement 1), comparable with tuning proportions reported for mouse fungiform taste sensing cells (Jisoo Han, 2018). Taste stimulation with substance-modality groups showed a mixed tuning of responding neurons and, in particular, 30% of neurons stimulated with bitters and sugars were activated by both taste groups, in accord with the integration of opposed valence cues in single larval neurons. As previously shown for one GRN (van Giesen et al., 2016a), such types of sensory integration might allow the animal to evaluate mixed food sources at the level of the input neuron.

Interestingly, in animals stimulated with individual substances representative of different taste modalities (Figure 2D), up to 68% of responsive neurons were activated by only one tastant per animal. This could be due to the selected substance concentrations that could elicit weak responses undetectable at the chosen intensity threshold of response or at the imaging resolution used in this study. By varying the number and concentration of tested substances, we expect neuron tuning to be altered, as described on afferent taste neurons in mammals (Wu et al., 2015).

An interesting feature in larval taste physiology revealed by our observations was related to distinct response dynamics elicited by tastants in larval GRNs. Apart from canonical GCaMP fluorescence increase during stimulus, we also identified late neuronal activation that could be characterized as OFF-response with onset after stimulus application, but also GCaMP fluorescence drop during taste stimulus as perhaps deactivation response. Most notable deactivation signals were recorded in CDL2 neuron and most stereotypically to citric acid (Figure 3 and Figure 3—figure supplement 1B). Interestingly, citric acid elicited concomitant activation in a neighboring cell named CDL1, indicating response patterns of activation-deactivation. The synchronous activation coupled with deactivation responses in adjacent neurons resemble the description of ephaptic signaling which implies that the electrical field of a responding cell generates hyperpolarization in the partner cell (Jefferys, 1995; Faber and Pereda, 2018). While ephaptic transmission has been described in the olfactory system of the adult fly (Su et al., 2012; Zhang et al., 2019), it remains to be determined whether such lateral inhibition mechanism also occurs in Drosophila taste. Regardless of the underlying molecular mechanisms, deactivation events together with delayed and OFF-responses most probably enter in assembly of spatio-temporal codes as described for taste discrimination in the moth (Reiter et al., 2015), and remain an open fascinating question in the physiology of larval sensory neurons.

A large proportion of neurons responsive to one taste substance per animal were activated by sucrose and, interestingly, sucrose also elicited numerous responses in neurons activated by two or more tastants, in combinations such as sucrose+ citric acid, sucrose+ amino acid, sucrose+ high salt, and sucrose+ bitter (denatonium/quinine). It is possible that, at least in part, these neurons integrate sucrose response with other modalities for purposes of evaluating mixed dietary substrates for the purpose of avoiding dangerous food mixtures rather than having a direct role in sucrose sensing and output behavior, as previously indicated for C7 neuron (van Giesen et al., 2016a).

Nevertheless, since a direct participation of TOG neurons in sugar taste behavior has thus far not been shown, we here explored the roles of individual external gustatory cells. We started by testing C2, C6, and C7—neurons that have associated calcium responses to sucrose (van Giesen et al., 2016a). As expected, C7 has been linked to taste integration of opposed valence of sucrose and denatonium or quinine and we have not observed a defect in direct sucrose behavior in C7-silenced larvae (Figure 4A). Similarly, silencing C6 also did not result in behavioral defects. Known to have a role in CO2 sensing (Kwon et al., 2007; Jones et al., 2007; Faucher et al., 2006), C6 is likely not involved in sucrose behavioral response per se, but prospectively important for integration of this taste in detecting fermenting substrates, and therefore would also have a role in computing mixed tastes, a hypothesis to be confirmed by future work.

We identified C2 to contribute in the attraction to sucrose. Interestingly, Hernandez-Nunez et al., 2015 described a ‘weak attractive response’ in larvae when C2 was optogenetically activated alone but when co-activated together with C1 neuron, the C1-dependent repellent behavior was potentiated. This suggests that C2 neuron might be involved in sensing of an attractive cue while also enhancing the aversive behavior in association with a bitter neuron, as a possible mechanism for avoiding risky food mixtures. As silencing C2 does not result in a complete abolishment of sucrose attraction, additional TOG or external sense neurons might participate to sugar behavior. In support of this idea, we tested C5 neuron as a likely sugar-sensing candidate not expressing bitter-implicated receptors Gr66a/Gr33a, whose silencing also resulted in a trend for preference decrease compared to control larvae (Figure 4—figure supplement 1A).

The Gr43a receptor expressed in pharyngeal sense neurons and in the brain has a central role in sugar sensing in larval taste (Mishra et al., 2013). To fit in this context our observations on TOG calcium responses to sugar and a role of C2 neuron in sucrose sensing, we make distinction between external and pharyngeal taste. We compared larval behavior on easily ingestible 1% agarose concentration, as tested by Mishra et al., and on 2.5% substrate that is less accessible for eating (Figure 4B, Figure 4—figure supplement 1B, B'). Our results support the hypothesis that C2 neuron in the TOG senses sucrose and contributes to behavior prior to and in absence of eating, whereas Gr43a is the main sugar receptor but requires food ingestion. The overrun of Gr43a guided behavior during food ingestion may be due to the lesser need for external sugar sensing when it has already been located and ingested, thus switching to an alternative internally guided pathway for sugar preference behavior (Figure 5B). Also, the reminiscent behavioral defects of Gr43a mutant to fructose sensing on 2.5% substrate (impaired ingestion) could be due to Gr43a expression in several internal tissues (pharynx, proventriculus, and brain) and to its role as hemolymph fructose sensor. It has been shown that hemolymph carbohydrates influence larval feeding behavior through insulin signaling and SLC5A11 (Ugrankar et al., 2018), but it remains to be determined if and how does a Gr43a mutant correlate with metabolic pathways and possibly interfere with sugar preference output behavior.

In the adult, Gr43a acts as the main fructose hemolymph sensor but also associates a role in feeding suppression in satiated animals (Miyamoto et al., 2012). Other moderators, such as the Ir60b neurons, carry a role in limiting sugar consumption (Joseph et al., 2017). Our study adds a new piece of evidence to deconvolute sugar sensing mechanisms in the larva while setting down a path for future investigation of sugar detection at this developmental stage.

The observations reported here in relation to taste sensing and cellular segregation of behavior can be paired with previous data for a depiction of sugar/bitter taste detection at the periphery (Mishra et al., 2013; Apostolopoulou et al., 2014; Kim et al., 2016; van Giesen et al., 2016a; Choi et al., 2016; Choi et al., 2020; Figure 5): in the TOG larvae detect quinine in C3 neuron, denatonium in C1, sucrose in C2 and C5, but also mix of sucrose and denatonium/quinine in C7 and also possibly a mix with sucrose in CO2-detector C6; at the level of pharyngeal neurons dietary sugar is sensed in DP4 (expressing Gr43a) and caffeine in DP1; in the brain, hemolymph sugar is detected by Gr43a receptor neurons. Based on this summary, taste sensing in Drosophila larva would entail cell multimodality for food mixture processing, combinatorial taste through co-activation of neurons of distinct valence, and anatomical or molecular segregation of specific tastant detection. Evidently, despite its numerical simplicity, the study of taste in the fruit fly larva reveals numerous layers of complexity.

The present study offers a sample overview for physiological taste responses in the peripheral taste organ of the Drosophila larva. Our data illustrates the sensing capacity and complexity of responses at this level as revealed by multimodality of response in a fraction of neurons and by distinct temporal dynamics in activation. We also determine for the first time roles in sugar sensing behavior of larval external taste neurons, by segregating external and pharyngeal taste. We thus identify C2 as contributor to sucrose taste and open directions for further investigations on additional external sugar sense neurons and the query of their redundancy, role in computing substance mixtures, as well as specific molecular mechanisms and receptors involved in sugar detection.

Materials and methods

Fly stocks

Request a detailed protocol

We used the following fly strains: nSyb-Gal4 (gift from Shanhaz Lone), UAS-GCaMP6m (Bloomington #42748), UAS-H2B:RFP (gift from Boris Egger), GMR57BO4-Gal4 (Bl. #46355), Gr94a-gal4 (Bl. #57686), Gr21a-Gal4 (Bl. #23890), Gr22e-Gal4 (Bl. #57608), Gr66a-Gal4 (Bl. #57670), Gr33a-Gal4 (Bl. #57623), Gr59d-Gal4 (Bl.57653), Gr32a-Gal4 (Bl. #57622), Gr97a-Gal4 (Bl. #57687), Gr59e-Gal4 (Bl. #57655), UAS-myr:GFP (Bl. #32198), and UAS-TNTE (Bl. #28837). Gr43a lines: Gr43a-KI-Gal4 mutant, Gr43a-KI-GAL4; UAS-Gr43a and Gr43a-KI-GAL4; UAS-Cha7.4kb were kindly offered by Hubert Amrein.

Chemicals

Chemicals of highest purity available were chosen for chemicals: L-Alanine (Sigma-Aldrich, 56-41-7), L-Glutamine (Sigma-Aldrich, 56-85-9), L-Aspartic acid (Sigma-Aldrich, 56-84-8), L-Glutamic acid (Sigma-Aldrich, 56-86-0), L-Asparagine (Sigma-Aldrich, 70-47-3), L-Proline (Sigma-Aldrich, 147-85-3), L-Tyrosine (Sigma-Aldrich, 60-17-4), L-Threonine (Sigma-Aldrich, 72-19-5), L-Phenylalanine (Sigma-Aldrich, 63-91-2), L-Lysine (Sigma-Aldrich, 56-87-1), L-Arginine (Sigma-Aldrich, 74-79-3), L-Histidine (Sigma-Aldrich, 71-00-1), L-Serine (Sigma-Aldrich, 56-45-1), L-Valine (Sigma-Aldrich, 72-18-4), L-Cysteine (Sigma-Aldrich, 52-90-4), L-Methionine (Sigma-Aldrich, 63-68-3), L-Leucine (Sigma-Aldrich, 61-90-5), L-Isoleucine (Sigma-Aldrich, 73-32-5), L-Tryptophan (Sigma-Aldrich, 73-22-3), L-Glycine (Roth, 3187.3), D-Sucrose (Sigma-Aldrich, 57-50-1), D-Glucose (Sigma-Aldrich, 50-99-7), G-Fructose (Fluka, 57-48-7), D-Trehalose dihydrate (Roth, 9286.1), D-Arabinose (Sigma-Aldrich, 10323-20-3), D-Maltose monohydrate (Sigma-Aldrich, 6363-53-7), D-Mannose (Sigma-Aldrich, 3458-28-4), D-Galactose (Roth, 4979.1), D-Cellobiose (Roth, 6840.3), Lactose monohydrate (Roth, 8921.1), Quinine hemisulfate salt monohydrate (Sigma-Aldrich, 6119-70-6), Denatonium benzoate (Sigma-Aldrich, 3734-33-6), Caffeine anhydrous (Fluka, 58-08-2), Coumarin (Sigma-Aldrich, 91-64-5), Sucrose octaacetate (Sigma-Aldrich, 126-17-7), Lobeline hydrochloride (Sigma-Aldrich, 134-63-4), Theophyline (Sigma-Aldrich, 58-55-9), Strychnine (Roth, 4843.1), Potassium chloride (Merck, 7447-40-7), Sodium chloride (Sigma-Aldrich, 7647-14-5), Citric acid (Sigma-Aldrich, 77-92-9), Agarose standard (Roth, 3810.4), and Brilliant Blue FCF (Wako, 42090).

Immunostainings

Request a detailed protocol

Third instar larvae (4 days after egg laying) were washed and dissected in ice-cold PBS. Tissue containing chemosensory neurons was fixed in 3.7% formaldehyde for 20 min at room temperature. After fixation, incubation in primary antibodies was performed at 4°C overnight, followed by washing steps and incubation in secondary antibodies at 4°C overnight, and finally, in Vectashield antifade medium (Vector Laboratories) for at least 1 hr before mounting the samples on microscope slides. After each step described above, PBST (PBS 0.3% Triton X-100) was employed for three consecutive plus three 30-min washes at room temperature. Primary antibodies: chicken anti-GFP (1:1000, Abcam, ab13970), rabbit anti-DsRed (1:1000, Clontech, No. 632496), and rat anti-ELAV (1:30, DSHB, No. 7E8A10). Secondary antibodies conjugated with Alexa Fluor fluorescent proteins (488, 568, and 647) were used in dilution of 1:200 (Molecular Probes nos. A-11008, A-11039, A-21244, A-21247, and A-11011).

Confocal images were acquired on a Leica TCS SPE-5 confocal, using the 40× oil immersion objective, at 0.8–1 µm slice thickness. Images were assembled using Fiji (Schindelin et al., 2012), Adobe Illustrator, and Adobe Photoshop.

Calcium imaging

Request a detailed protocol

The semi-intact sample for calcium-imaging recordings was prepared as previously described (van Giesen et al., 2016b). Third instar larval heads comprising chemosensory neurons, the brain and the connecting nerves were dissected in AHL (adult hemolymph-like) saline solution. After mounting into the microfluidic chip, the sample was connected to the tubing and the micropump setup. For whole-organ TOG recordings, UAS-GCaMP6m (Chen et al., 2013) was driven in all neurons and RFP was expressed in nuclei using UAS-H2B:RFP (Egger et al., 2007).

Microscopy settings

Request a detailed protocol

Calcium imaging was performed using a Visitron VisiScope CSU-W1 and Nikon Ti-E inverted spinning disk confocal microscope. The images were captured using a Photometrics Evolve 512 Electron Multiplying Charge-Coupled Device (EM-CCD) of 16×16 µm2 pixel size, pixel binning 2×. The total field of view (FOV) diameter was 55 µm. Laser power for both channels was set at 150 mW, and the exposure time for each Z-section was set at 80 ms. Imaging parameters: 25 slices/stack; 1.5 μm each slice for a total depth size of 37.5 μm; time acquisition speed 0.5 time points per second (7.5 fr/s) or 2 seconds per stack.

For each recording, an RFP stack (one time point) was acquired separately from the GCaMP time series consisting in 60 time points per taste stimulation. Recordings were processed in Fiji/ImageJ (Figure 1—figure supplement 1) and transferred to Imaris 9.6.0 software for visualization and cell segmentation (Figure 1—video 1).

Taste stimulation

Request a detailed protocol

Each tastant was administered using a micropump (mp6, Bartels Mikrotechnik) controlled in VisiView software through built-in macros. For each taste stimulation, macros were designed for 2-min length recording: 60-s pre-wash followed by 30-s stimulation and a final 30-s wash. Millipore water was used as solvent for taste solutions and as washing control.

Taste stimulation with groups of tastants

Request a detailed protocol

We compiled two groups of sugars, each containing five sugars at a concentration of 100 mM each: monosaccharides (fructose, glucose, arabinose, mannose, and galactose), and disaccharides (sucrose, trehalose, maltose, lactose, and cellobiose). Amino acids were prepared as previously described by Park and Carlson, 2018 (Table 1) in four groups each at 10 mM concentration: group A (valine, leucine, isoleucine, methionine, tryptophan, and cysteine), group B (alanine, phenylalanine, glycine, and proline), group C (arginine, lysine, aspartic acid glutamic acid, and histidine), group D (serine, threonine, asparagine, and glutamine), and tyrosine was included in group B at 1 mM concentration, limited by its solubility. Salt taste groups contained NaCl and KCl in equal proportions at a total concentration of 50 mM (low salt) and respectively 1 M (high salt). Bitter tastants were randomized in two groups of four substances each, with a final concentration of 22 mM and respectively 13 mM, restricted by the solubility of certain substances such as lobeline or strychnine reduced to 1 mM instead of 10 mM (Table 1): DSoTC group (denatonium benzoate, sucrose octaacetate, theophylline, and coumarin), and respectively QLSC group (quinine, lobeline, strychnine, and caffeine). The order of chemicals was randomized. The response threshold was chosen at 20% and the number of responding neurons within each trial of stimulation is specified in sheet 1 of the Figure 2—source data 1.

Taste stimulation with series of individual tastants

Request a detailed protocol

We used two series of tastants (Table 1), each containing five substances from five different taste categories: sucrose 500 mM for sweet taste, denatonium benzoate 10 mM (series 1), or quinine hemisulfate 5 mM (series 2) for bitter taste, NaCl at 1 M for high salt (series 1) or at 100 mM for low salt (series 2), valine 100 mM (series 1), or arginine 100 mM (series 2) as amino acid and citric acid for sour taste. Each recorded animal was stimulated with one tastant from each taste category and a total of 15 TOG organs were recorded, 7 within the first series of stimulation and eight within the second series. The order of chemicals was randomized. Neurons responding above the amplitude threshold of 20% are listed in Figure 2—source data 2.

Data processing and analysis of whole-organ physiological recordings

Request a detailed protocol

RFP stacks and GCaMP recordings were treated for deconvolution using the third-party program Huygens (Scientific Volume Imaging). A Fiji/ImageJ plugin was written to duplicate the RFP signal onto all time points of the GCaMP recording (Figure 1—figure supplement 1; Figure 1—figure supplement 1—source data 1). For correction of drift or animal movement, a 3D drift correction plugin (Parslow et al., 2014) was run on the GFP signal before merging the two channels. This software code allows for decent amendment of z-drift of hyper-stacks. Subsequently, the drift-corrected GFP and the original duplicated RFP stack were merged and the misalignment between the two signals was adjusted using an in-house Fiji/ImageJ macro for manually guided realignment on the x-y axis (Figure 1—figure supplement 1; Figure 1—figure supplement 1—source data 1). A final 3D drift correction was performed on the aligned two-channel hyperstack for better stabilization. If at this step the drift or animal movement rectification results were not satisfying, the recording was discarded. The resulting recording was transferred in Imaris 9.6.0 for automatic segmentation with spots of 3.5 µm diameter (Figure 2A, Figure 1—video 1). Each uniquely identified neuron through spot detection received a random identity number and a corresponding fluorescence trace over time (Figure 1—figure supplement 1).

Downstream analysis was performed with R (R Development Core Team, 2017; Wickham et al., 2018). Normalization of fluorescence values followed the formula (Ft−F0)/F0, where Ft is the intensity at time point t and F0 the baseline calculated as average intensity of 10 frames prior to stimulation. Response amplitude was calculated as the difference between the peak intensity after stimulation (average of 5 frames near the maximal fluorescence value) and F0. All traces were manually validated to exclude false positive responses. Any of the following criteria was sufficient to exclude traces as possibly false positives: the baseline signal was unstable or the baseline variance was higher than the calculated response amplitude; the fluorescence rise upon stimulation could have been caused by an artifact; a delocalized fluorescence signal could have been captured by the cell spot, compromising the measured response. For further analysis, a response threshold was set at 20% DF/F0. To calculate taste integration percentages (Figure 2), for each data set, we pooled together all neurons eliciting at least one response above 20% in amplitude and determined the fraction activated by one or by multiple tastants/taste groups per animal.

Cellular mapping

Request a detailed protocol

This step was performed entirely manually for recordings with individual tastant stimulation (Figure 2D and D’). Responsive neurons were compared to their neighbors and to the reference four neurons localized anterior-laterally named ALN group (Anterior-lateral-neurons, Figure 3B). For each recorded response above the 20% threshold, the activated cell was thus compared with the neighboring cells for determining an approximate location within the organ and then compared across animals with responding neurons with similar approximated position. If one of these comparison steps was inconclusive, the cell response was not considered for the mapping. In Figure 3C, we plotted responses that were each manually mapped in at least four separate recordings (exception for responses to bitter substances where plotting was considered for at least three separate recordings).

Figures were generated in RStudio (R Core Team, 2015) using ‘ggplot2,’ ‘reshape2’, and ‘scales” packages (Wickham, 2007; Wickham, 2009; Wickham, 2019) or in GraphPad Prism 7 and finally adjusted in Adobe Illustrator and Adobe Photoshop.

Larval behavior

Request a detailed protocol

L3 4-day-old larvae grown on stable conditions of 25°C and 12-hr light/dark cycle, were picked from the food (feeding larval stage) with a brush, washed in tap water, and tested for taste preference in a behavior dark room to exclude additional stimulation of the animal.

Two-choice assay

Request a detailed protocol

Petri dishes (94 mm) were prepared on the day of testing as follows: plain agarose at 2.5% concentration was poured in the dish; after hardening, a middle straight line was used to cut the substrate in two halves; one half was discarded and replaced with agarose of the same concentration containing a test substance. Agarose plates were left to reach ambient temperature for at least 1 hr before testing. For comparison with previously reported results for Gr43a sugar sensing role, agarose plates were prepared identically at 1% concentration. For each test 30 larvae: after washing, larvae were placed on a middle line region of the plate and left to wander. After 5 min, animals were counted on each side of the plate and the preference index was calculated as follows:

PI=(nlarvaeontestsubstancenlarvaeonplainagarose)/ntotallarvae

For experiments consisting in several time points, counting was repeated at 2 min, 5 min, and 15 min. At the time of counting, animals that were found on the middle region defined as a 1-cm width-line between the two halves were not considered on either choice side and therefore taken as neutral. For example, if among 30 larvae, 19 were found on the sucrose side, 7 on the empty half, and 4 on the middle stripe region, the index is calculated as PI=(19−7)/30=0.4.

For each experiment, tests were repeated for N≥7 (typically N=10—see Figure 4, Figure 4—figure supplement 1 and respective source data files for details). GraphPad Prism 7 was used for data visualization and for statistics. Mann-Whitney test was applied to statistically compare the test genotype with each control line (p-value<0.05; **p<0.01; ***p<0.001).

Optogenetics

Request a detailed protocol

The optogenetics experiments were performed as described in Mauss et al., 2017. Briefly, larvae were assayed in petri dishes half coated with 2.5% agarose and half with 2.5% agarose containing 500 mM sucrose. Tubes containing larvae were exposed to 527 nM color light emitted from a projector (Epson LCD Projector model H763B, default settings) situated 70 cm above the experimental space for 2 hr prior to experimentation for stimulated experiments, or conversely kept in darkness for 2 hr for unstimulated (control) tests (Figure 4D). The number of larvae on each substrate was counted for the preference index calculation at 2, 5, and 15 min after placement.

Food ingestion test

Request a detailed protocol

Petri dishes were filled with 1% or 2.5% agarose mixed with sucrose 500 mM and Brilliant Blue dye 2%. Larvae were left to wander on the plate for 2, 5, 15, or 30 min, before being removed and washed in tap water. The blue dye present in their abdomen tracks the agarose larvae were able to ingest on the two different concentrations.

Data availability

All data is available as part of the submitted manuscript.

References

Decision letter

  1. Sonia Sen
    Reviewing Editor; Tata Institute for Genetics and Society, India
  2. K VijayRaghavan
    Senior Editor; National Centre for Biological Sciences, Tata Institute of Fundamental Research, India
  3. Sonia Sen
    Reviewer; Tata Institute for Genetics and Society, India

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "Taste sensing and sugar detection mechanisms in Drosophila larval primary taste center" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Sonia Sen as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by K VijayRaghavan as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The methods used in this manuscript are useful and the findings extremely interesting. There were a few points that concerned us, and we list here experiments that will help support the claims made in manuscript better.

– Functional Imaging

1. We would like to get a sense of the variability in responses to the tastants – both inter- and intra- individual variability. To illustrate this, we are asking the authors to (1) represent the data as per neuron, per animal, and (2) show more traces (as in 2D) for individual cells from individual animals across trials with different tastant stimulations.

2. To irrefutably assign taste response profiles to each of the neurons, we're also suggesting (1) dose responses to each tastant, and (2) swapping the order of tastant presentation while recording responses.

3. Central to better describing inter-individual variability will be the unambiguous identification of individual GRNs across individuals. This needs to be shown more definitively and we recommend a uniformed naming system to avoid confusion. One possible way they could do this is by labelling known GRNs in the whole organ imaging. If only a small number of GRNs can be so identified, perhaps the authors could focus on just these for their analyses.

4. The methods and criteria used for image acquisition and processing needs to be described better.

– Behaviour

The authors' central thesis – peripheral sweet neurons>hard food; internal sweet neurons>soft food – hinges on the result that on hard food, despite the absence of 43a from the pharyngeal neurons, larvae still show a preference for sugar. Presumably larvae are not eating during the 2-5 minute window (4B) and so the brain 43a receptors cannot driving this preference for sugar. They reason, it must be peripheral neurons that are responsible for this. We have three issues with this central finding and their interpretations:

1. Sugar or Texture? There's a possibility that mechanosensation (response to texture of the food) might be playing a role in this observation. The authors could demonstrate the specificity of this response by performing a dose response curve with sucrose (we recommend sticking to sucrose as it shows more robust PIs). So, we recommend that they perform the experimental conditions in 4B(bottom) with sucrose at different concentrations.

2. The 43a null: If the peripheral neurons were driving this behaviour, the 43a null mutants should show at least a residual sugar preference, which they don't on either soft or hard food. It's possible that the authors are missing this in the low PIs of 4B. The above experiment should uncover this.

3. Peripheral neurons on soft food only? What keeps the peripheral neurons from driving sugar preference on soft food in the absence of 43a? If the authors are suggesting that the peripheral sugar neurons are gated by both sugar and texture, they would need to demonstrate this through functional imaging. And in that case, activating the peripheral-GRNs in the brain-only 43a animal should induce sugar feeding on soft food.

In the individual reviews below, the authors will find more specific points that pertain to the broad points mentioned here. Could the authors please pay attention to them as they address these points?

Reviewer #1:

The authors use a very nice microfluidic set-up to monitor neuronal activity in an ex-vivo larval TOG while simultaneously bathing it in a variety of tastants. They show a surprisingly high level of multimodality in the GRNs – >30% respond to 2-4 tastants. On occasion, these overlaps were of opposing hedonic value and so the authors propose a high level of signal integration in the larval taste organ. They seem to suggest that this could occur via temporally distinct responses in the GRNs.

Subsequently, the authors focus on sugar sensing in this peripheral organ. Interestingly in larvae, the 43a receptor that senses sugars is not expressed in the periphery but in the pharynx (where it can detect sugar in the ingested food) and in the brain (where it can detect haemolymph sugars). Here the authors demonstrate that many of the peripheral GRNs can respond to sugars. They silence 3 GRN classes to find that one of them is necessary/involved in sugar sensing.

They next turn to addressing the issue of peripheral vs internal sensing of sugar. They perform behavioural experiments to calculate preference for sugars while silencing the peripheral GRN or mutationally inactivating GR43a. They perform these experiments in soft and hard food and suggest that in soft food (when the animal can easily ingest) the internal sugar sensors are utilised. On hard food however, they suggest that the internal sensors are not as important, but that the peripheral GRNs involved in sugar sensing are necessary. This is an interesting idea.

My primary concern with this manuscript is the section dealing with the internal vs peripheral sensing of sugars. It's possible that some of my concerns arise out of the manner in which these results are reported and discussed, nevertheless, I will list them out here:

– The preference indices in figure 4B seem very low: 0.2. At such low PIs I wonder if the authors might mistakenly interpret a lower π for a 'no-preference'. In general I did not understand the reasons for using the various sugars at differing concentrations in this section.

– The whole animal 43a mutant on both hard and soft food shows a complete inability to detect sugar. If this is true, then it suggests that the pheripheral neurons are NOT involved in sugar sensing. Do the authors agree with this interpretation?

– The brain-only 43a mutant data makes the whole animal mutant data more difficult to interpret. If the difference between soft and hard food is a difference in time taken to ingest the food (as shown in Figure 4B), then there should be no difference in the trends of sugar responsiveness except in the axis of time. I have not understood why the authors think that on soft food the peripheral GRNs don't compensate for the lack of pharynx-43a to drive sugar feeding. Can the authors please comment on this?

Reviewer #2:

Maier et al., investigates neural basis of sugar perception/coding in Drosophila larvae using in vivo calcium imaging and behavior. Authors developed a volumetric imaging method to capture taste responses in all of the TOG neurons in larvae. The whole organ imaging method developed here will be a very useful technique for Drosophila larval chemosensory imaging. However, the imaging method and data analysis pipeline to identify cellular identity is not clearly explained in the manuscript. It is important to provide a metric for how effective and reliable cellular identity classifications are and the variations among individual larvae. The volumetric imaging results indicate that TOG neurons can respond to one modality or multiple modalities. The neural response dynamics are also variable. Some neurons show ON responses while some show OFF-responses or inhibition upon tastant stimulation. Furthermore, the authors test the function of TOG neurons in regulating sugar preference and found that the C2- TOG neuron is important for discriminating sugar when food is more viscous and harder to ingest. Overall, I think the volumetric imaging method is promising however, the imaging method and data analysis pipeline is not clearly explained and no metric has been provided to show how effective 3D cellular identity analysis is. The behavioral experiments regarding C2 neurons are not consistent with the previously published functional imaging analysis from the same group. Further functional imaging experiments are necessary to support the claims in the paper.

1. The whole organ imaging method developed here will be a very useful technique for Drosophila larval chemosensory imaging. However, the details of image capture and processing is somewhat vaguely explained in the manuscript. Multiple details and metrics are missing about how imaging is done. What is the laser power used in these experiments? What is the frame rate for each volume scanned? What is the size of each volume, field of view etc. The movement and the drift correction for calcium imaging makes sense, however more detailed explanation about how they process imaging data is necessary. For example, in Figure Supplement 1, the authors are correcting GFP channels using the position information from the RFP channel and their method is able to handle the xy-plane. However, it is not clear how they handle z-drift. Did the post-processing also apply for the movement and drift correction in the Z-axis? What is the error rate for x-y or z correction? It would be good to know how accurate their methods are in identifying the same cell identities across animals. The authors need to expand the methods section and explain clearly how volumetric imaging and data processing was done in the paper. This is key for accurate interpretation of cell identification and the calcium imaging results.

2. Is figure 2B and 2C showing the percent responsive cells recorded across multiple animals or the percent responsive cells within an animal? I think it is important to clarify this point. What is the variance between individuals? I would like to understand how reliable cell identification and functional imaging analysis is across animals. This can be achieved if authors show data per animal rather than lumping all the data across the animals together. Another reason to visualize individual animal responses is to investigate whether the order of tastant stimulation is a critical factor for larval taste perception: For a neuron that can respond to bitter and sugar, would the sequence of stimulation matter? Will the neuron change activity if it was stimulated first with a bitter compound then a sugar compound? What about the opposite? I strongly encourage authors to show more traces for individual animals across trials with different tastant stimulations as done in Figure 2D. This will help the reader understand what is the variation in taste responses across animals and it will also increase confidence for the consistency of authors' volumetric imaging method.

3. In figure 3A, the authors show different calcium dynamics upon tastant stimulation in different TOG neurons. They have observed, off responses and inhibition of neural activity in certain neurons upon stimulation. I would like to see what is the percentage of OFF-responses and inhibition observed in these experiments. Are these events frequent or rare? Are they seen in all animals or some of the animals used in the calcium recordings? It is important to understand whether these are biologically relevant neural responses or artifacts seen in some of the animals used in the experiments.

4. In Figure 3, authors replot the results from Figure2 this time by identifying cellular identity using anatomical position. This is something similar to what I was trying to get to in my previous comment (see comment 2). But even in plots shown in Figure 3C, it is hard to understand how successful authors' methods are in classifying each neuron type in each animal. For example, how many larvae in total have been tested in Figure 3C and in what percentage of the animals they were able to identify each neuron type and what is the variation in neural responses among each animal. I think this is a very critical point that needs to be clarified in the paper variation of taste responses among animals and a metric to show how successful cellular identification is across all the animals tested. If my understanding is correct, the total number of tested organs/animals is 15, however the data plotted in Figure 3C does not correspond to this number. Please explain what is the success rate to define a cell type across animals.

5. The authors mention that the only known sugar receptor gene Gr43a is not expressed in TOG but TOG neurons can respond to sugars. This is a puzzling finding. Is the expression analysis of Gr43a receptor based on transgenic reporters or single cell sequencing? In other words, is there other evidence that TOG does not express any known sugar receptors in larvae?

6. In figure 4, authors performed behavioral experiments to test the function TOG sugar sensing in larvae. Inhibiting one of the TOG neurons, C2 causes defects in sucrose preference when agarose ss 2.5 % but not 1%. They claim the external sugar sensing via TOG sensory neurons is critical when food is not soft enough to ingest. These are interesting hypothesis however, in their previous study, authors have shown C2 neurons respond weakly to 500mM sucrose but not to any other sugar. However, in this manuscript the stronger behavior phenotype is seen in 250mM fructose preference upon silencing of C2 neurons. It is not clearly to me what is the mechanism here. Please explain.

7. In general, I think it will help to use one nomenclature for the paper or provide a table matching C-X identity to cellular identities in Figure 3C and list which compound(s) each neuron responds to and indicate the type of response; ON, OFF or inhibition.

8. The manipulations for larval GRNs in the manuscript are done chronically by expressing TNT-E. This might impact the development and wiring of the larval chemosensory system and cause problems in sugar detection unrelated to GRN function. I think using a conditional inhibitor of neural activity such as shits or newly developed ontogenetic inhibitors, GtACRs, might be better to use here to show conditional inactivity of the neurons result in similar defects in sugar sensing.

Reviewer #3:Drosophila larvae have been put forth as a good model to study taste because of the simplicity of the system (in terms of the number of organs and neurons) and efficient behavior assays. However, an understanding of taste coding in the larval system lags behind that described for the adult stage, in which most taste sensilla/neurons are accessible for recordings or imaging. Maier et al., seek to address this gap by developing a whole organ calcium imaging approach to assess responses of neurons in the larval terminal organ to sweet, bitter, salt, amino acid, and acid tastants. From imaging 15 organs, the authors describe responses to a set of sweet, bitter, salt, amino acid, and acid tastants, collectively represented in neurons that appear to be tuned to one, two or multiple categories of tastants.Although the new approach has promise, the current study makes a limited conceptual advance and the number of positive responses used to ascribe responses to neurons is small. Overlap of tastant sensitivities has now been described, and more rigorously so, in the adult stage and in the authors' own previous work focusing on genetically labeled subsets of taste neurons. The identification of responsiveness to sucrose, which may be somewhat unexpected given the absence of sweet Gustatory receptors (Grs) in larvae, has also been previously described for a few neurons, one of which is further studied in behavior assays.

The imaging analysis is followed by investigation of the potential roles of sugar sensing neurons in the periphery. While this analysis provides further support for a previously identified peripheral neuron (C2), it does not convincingly parse contributions of internal and external neurons.

Specific weaknesses are elaborated via points below and could be addressed either with additional experiments or textual modifications.

Calcium imaging:

– Neuron categories were determined by imaging responses to only a single concentration of each tastant; more careful dose response analysis may alter the conclusions. This possibility has not been considered.

– There is no effort to reconcile the imaging results with predictions from previous studies. For example, the number of bitter-responsive neurons is 1-2 per larva. However, bitter-GAL4 lines have been mapped to many more neurons in the TO.

– The correspondence between neurons named from previous GAL4 mapping experiments to those named by position in this study is not at all clear. A better comparison between the two could be presented.

–The criteria that were used to manually exclude false positive responses in calcium imaging experiments (Line 475-476) have not been described.

– A response was ascribed to an identified neuron if the position was the same in >3 organs. This threshold seems low, considering that 7-8 organs were tested with each stimulus. In the table for Figure 3, for instance, several tastants have n=3. Is this from a higher number of tastant applications? If so, what was observed in the other cases – that a response could not be ascribed to the identified neuron or that there was no response at all?

Behavior:

– There are no indications of statistically significant differences between the genotypes in Figure 4B; the legend only mentions that the preference of Gr43a mutants is not significantly different from zero. Finally, data in Figure 4C are compared with Mann-Whitney tests, but ANOVA should be used instead.

– Whether sucrose response of internal neurons is affected in Gr43a mutants has not been tested (relevant to Figure 4C).

– In Figure 4B right, assuming that Gr43a mutant controls are not significantly different from zero, whereas Gr43a "brain only" are, I don't know how to interpret a potential role for Gr43a brain neurons in preference for 2.5% agarose food, at 2 minutes when the larvae are not consuming anything.

– The calcium imaging analysis is at the cellular level, not looking into contributions of receptors. Manipulating cells rather than receptors for behavior experiments would allow for better comparison between the two aims of the study. If the Gr43a mutant is the only way to go, suggested experiments are:

1. Test if the complementary brain ablated genotype (Gr43a internal taste only) yields the expected phenotypes on 1% and 2.5% agarose.

2. Test if peripheral sugar responses are affected in Gr43a mutants.

https://doi.org/10.7554/eLife.67844.sa1

Author response

Essential revisions:

The methods used in this manuscript are useful and the findings extremely interesting. There were a few points that concerned us, and we list here experiments that will help support the claims made in manuscript better.

– Functional Imaging

1. We would like to get a sense of the variability in responses to the tastants – both inter- and intra- individual variability. To illustrate this, we are asking the authors to (1) represent the data as per neuron, per animal, and (2) show more traces (as in 2D) for individual cells from individual animals across trials with different tastant stimulations.

Thank you for your observation, we indeed agree that showing more traces helps in better grasping the data and the potential entailed by imaging tools described here.

We have therefore added more traces in Figure 2D per neuron to different tastants within the same organ in 2 separate representations. Also, in Figure 3 we are including traces per mapped neuron to a given tastant from three different organs. The previous data representation of Figure 3 was moved to Figure 3-figure supplement 1 with a more stringent threshold for plotted tastant responses per cell. All responses recorded as above 20% in amplitude included in the paper are available in the supporting information (source data excel tables).

2. To irrefutably assign taste response profiles to each of the neurons, we're also suggesting (1) dose responses to each tastant, and (2) swapping the order of tastant presentation while recording responses.

1. This is an excellent suggestion, as proved by the well documented example in Drosophila of low vs. high salt concentrations that activate distinct taste cell populations for divergent output behaviors. Nevertheless, manually mapped neurons in response to a given tastant across separate organs are displayed in Figure 3 and Figure 3-Supplement1 as proof of concept and we believe that more ambitious attempts for neuronal mapping and to a range of tastant concentrations should be assisted by tissue registration techniques. We propose that irrefutably assigning response profiles to each neuron does not make the main object of this study, but we present a valuable whole organ analysis tool in assisting future efforts using complementary necessary techniques for neuron labeling towards this purpose. Instead, our data aims at providing a grasp of the variety of recorded responses to 5 tastants from distinct modalities in average in the larval external taste organ.

We hope that reviewers accept our reasoning, as a dose response analysis on a manually mapping approach of cell identities would be extremely challenging.

2. The tastant order swapping is an important point and this is included in the protocol design of whole organ recordings, as the order of tastant presentation was randomized for each organ. Also, we have now additionally performed single neuron calcium imaging in C7 – neuron known to respond to both sugar and bitter tastants (van Giesen et al., 2016a) – to exemplify the influence of order presentation on sucrose-elicited responses. We have found no significant difference between sucrose response before or after stimulation with denatonium in this neuron, as shown in Figure 2-Supplement 2.

3. Central to better describing inter-individual variability will be the unambiguous identification of individual GRNs across individuals. This needs to be shown more definitively and we recommend a uniformed naming system to avoid confusion. One possible way they could do this is by labelling known GRNs in the whole organ imaging. If only a small number of GRNs can be so identified, perhaps the authors could focus on just these for their analyses.

This valid point constitutes one main challenge for manual cell annotation in whole organ imaging, as touched upon under the previous reviewer comment. The system employed here for neuronal annotation as proof of concept relies on an estimated neuron location relative to the sagittal plane, anterior-posterior axis and ventral-dorsal axis (CPV = centroposterior-ventral). In Figure 3-Figure Suppl.1 we have plotted mapped neurons using more narrow constraints than previously in Figure 3, in line with the reviewer’s comment for more focused analyses. However, in order to retackle these identities for further characterization and for building a reference retraceable naming system, our tools constitute only one building block and should be complemented by:

– A working model for cell stamping of known GRNs within a whole organ labeling in physiological recordings,

– A tissue registration approach of whole organ imaging for automatic cell mapping circumventing manual cell annotation and

– A higher number of Gal4 transgenic lines mapped alongside the 7 known identities of GR-expressing neurons, such as co-mapping GR lines with IR and Ppk Gal4 lines.

Also, considering that individual physiological characterizations for the few known GRNs of the larval TOG organ have been tackled previously for 5 out of 7 mapped neurons, the main aim of our study was rather towards an overview of the range of response types per organ, attempted for the first time in larval peripheral taste sensing, to our knowledge. We thus could identify an average cell response multimodality ratio per organ, cell-specific deactivation events and sucrose-evoked responses in the larval TOG. This informed the second part of this manuscript for sugar sensing in the larva and could similarly direct further dissections of larval taste coding.

4. The methods and criteria used for image acquisition and processing needs to be described better.

We appreciate this feedback and have clarified the methods description, all textual changes marked with track change.

– Behaviour

The authors' central thesis – peripheral sweet neurons>hard food; internal sweet neurons>soft food – hinges on the result that on hard food, despite the absence of 43a from the pharyngeal neurons, larvae still show a preference for sugar. Presumably larvae are not eating during the 2-5 minute window (4B) and so the brain 43a receptors cannot driving this preference for sugar. They reason, it must be peripheral neurons that are responsible for this. We have three issues with this central finding and their interpretations:

1. Sugar or Texture? There's a possibility that mechanosensation (response to texture of the food) might be playing a role in this observation. The authors could demonstrate the specificity of this response by performing a dose response curve with sucrose (we recommend sticking to sucrose as it shows more robust PIs). So, we recommend that they perform the experimental conditions in 4B(bottom) with sucrose at different concentrations.

Thank you for your suggestion. We have modified the figure 4 to include 3 different sucrose concentrations for C2 on 1% versus 2.5% agarose, and have included in Figure 4-Supplemental1 two sucrose and fructose concentrations for Gr43a mutant/rescue genotypes.

Overall, these results indicate that sugar sensing in TOG neurons, specifically in C2, can detect sucrose on a hard substrate when food ingestion is impaired or slowed down. We propose that pharyngeal Gr43a neurons come into play mainly on ingestible substrate, at lower concentration, as also schematized in Figure 5. This hypothesis is supported by the expression of Gr43a receptors, which are present at the level of pharyngeal neurons, brain and other peripheral tissues such as in the foregut, but absent from the TOG external neurons (Kwon et al., 2011, Mishra et al., 2013).

2. The 43a null: If the peripheral neurons were driving this behaviour, the 43a null mutants should show at least a residual sugar preference, which they don't on either soft or hard food. It's possible that the authors are missing this in the low PIs of 4B. The above experiment should uncover this.

Indeed, as the reviewer correctly suggested, the above experiment shows residual preference of the Gr43a mutant to sucrose/fructose at higher concentrations.

3. Peripheral neurons on soft food only? What keeps the peripheral neurons from driving sugar preference on soft food in the absence of 43a? If the authors are suggesting that the peripheral sugar neurons are gated by both sugar and texture, they would need to demonstrate this through functional imaging. And in that case, activating the peripheral-GRNs in the brain-only 43a animal should induce sugar feeding on soft food.

We believe the above experiment equally addresses this point, as Gr43a mutant larvae do not show a complete defect on higher concentrations, including on 1% agarose. This indicates that sugar sensing via C2/TOG neurons contribute to output preference including during food ingestion but that it is overran by Gr43a-dependent pathways.

All in all, we do not believe C2/TOG neurons would be inactivated on soft food, but rather they play a role in a multicomponent system with different leverages for sugar sensing in the larva, as we are illustrating in Figure 5. Also worth noting is that the natural environment the fruit fly larva feeds on often implies the animal is immersed in a soft substrate, with a lesser need for clear taste dissociation between external and internal taste sensing.

In the individual reviews below, the authors will find more specific points that pertain to the broad points mentioned here. Could the authors please pay attention to them as they address these points?

Reviewer #1:

[…]

My primary concern with this manuscript is the section dealing with the internal vs peripheral sensing of sugars. It's possible that some of my concerns arise out of the manner in which these results are reported and discussed, nevertheless, I will list them out here:

– The preference indices in figure 4B seem very low: 0.2. At such low PIs I wonder if the authors might mistakenly interpret a lower π for a 'no-preference'.

Indeed, the 100mM Fructose concentration tests yield low preferences in the Gr43a mutant. We have therefore, at the reviewer’s suggestion, included higher tastant concentration in what is now Figure 4 – Figure Suppl.1.

In general I did not understand the reasons for using the various sugars at differing concentrations in this section.

To briefly describe our reasoning:

1. We have initially focused on sucrose 500mM because this was previously employed in other larval taste studies and in the whole organ imaging in the first part of the manuscript, which informed the behavioral testing in the second part.

2. Then, having formulated our hypothesis that larvae can sense sucrose despite not being able to readily ingest the food, we wanted to connect our observations with the previous reports for Gr43a sensing. As Mishra et al., 2013, have mainly described Gr43a sensing on fructose 100mM at 1% agarose concentration, we have added this sugar to our analysis to reproduce observations from this study and to compare between agarose concentrations.

3. Other sucrose and fructose concentrations have been added at the reviewer’s suggestion showing a lesser role of Gr43a at higher concentrations and also, as expected from literature reports, lesser to sucrose and more specific to fructose.

– The whole animal 43a mutant on both hard and soft food shows a complete inability to detect sugar. If this is true, then it suggests that the pheripheral neurons are NOT involved in sugar sensing. Do the authors agree with this interpretation?

At the reviewer’s suggestion we added behavior data in Figure 4-Figure Suppl.1 which show Gr43a mutant does not yield a complete defect for fructose 500mM preference, and neither is this the case for sucrose 100mM or 500mM.

However, we were generally not taken aback by an inability of the Gr43a mutant in sugar sensing, be it on soft or hard substrate. Since Gr43a is the main internal fructose sensor in Drosophila, expressed not only in the DPS, but also in the brain and in the proventriculus of the gastrointestinal system, we propose that a null mutant presents with a more systemic defect, which may be due to metabolic signaling, as shown for hemolymph carbohydrates influencing larval feeding behavior though insulin signaling and SLC5A11 (Ugrankar et al., 2018) or through the inability to gauge the state of satiety. Sugar sensing in the adult also consists of a multicomponent sensing system, with brad sugar sensing receptors on the proboscis distinct from Gr43a expressed internally that acts both as internal nutrient sensor and as feeding suppresor in satiated animals, while other receptors also carry roles in limiting sugar intake (Miyamoto et al., 2012; Joseph et al., 2012).

Altogether, we propose a model whereby C2 / TOG neurons act to signal the presence of sugar, but Gr43a expressed elsewhere can outweigh the behavioral output, either directly triggered by food ingestion, but also potentially indirectly through internal signaling (Figure 5B). This model for larval sugar-sensing mechanism aligns with, and complements, reports in the adult.

– The brain-only 43a mutant data makes the whole animal mutant data more difficult to interpret. If the difference between soft and hard food is a difference in time taken to ingest the food (as shown in Figure 4B), then there should be no difference in the trends of sugar responsiveness except in the axis of time. I have not understood why the authors think that on soft food the peripheral GRNs don't compensate for the lack of pharynx-43a to drive sugar feeding. Can the authors please comment on this?

Thank you. These are very interesting points.

Firstly, does the difference between soft and hard food preference for sugar resume to only time axis? We believe this is not the case because behaviors driven by C2 or Gr43a-neurons seem to not be merely additive in output but rather Gr43a internally guided preference would overrun, as proposed and schematized in Figure 5 according to our current understanding.

If Gr43a-neuron sensing outweighs sugar detection via C2 as we believe to be the case, this could also explain why peripheral GRNs don’t compensate for the lack of Gr43a. Gr43a is also expressed in other peripheral tissues such as the gut and the brain, acting as an internal nutrient sensor in both the larva and the adult (Miyamoto and Amrein 2013, Miyamoto et al., 2012, Mishra et al., 2013). Additionally, Gr43a seems to be more specifically tuned to fructose sensing, as Gr43a mutant is less impacting on the preference to sucrose (disaccharide of fructose and glucose), which is likely better compensated for in lack of Gr43a by external neurons (Figure 4 B). It’s worth mentioning that in the adult fly Gr43a receptor in the pharynx is also more specifically tuned for fructose sensing while external GR-expressing sweet sensing neurons are more broadly detecting sugars.

Reviewer #2:

Maier et al., investigates neural basis of sugar perception/coding in Drosophila larvae using in vivo calcium imaging and behavior. Authors developed a volumetric imaging method to capture taste responses in all of the TOG neurons in larvae. The whole organ imaging method developed here will be a very useful technique for Drosophila larval chemosensory imaging. However, the imaging method and data analysis pipeline to identify cellular identity is not clearly explained in the manuscript. It is important to provide a metric for how effective and reliable cellular identity classifications are and the variations among individual larvae. The volumetric imaging results indicate that TOG neurons can respond to one modality or multiple modalities. The neural response dynamics are also variable. Some neurons show ON responses while some show OFF-responses or inhibition upon tastant stimulation. Furthermore, the authors test the function of TOG neurons in regulating sugar preference and found that the C2- TOG neuron is important for discriminating sugar when food is more viscous and harder to ingest. Overall, I think the volumetric imaging method is promising however, the imaging method and data analysis pipeline is not clearly explained and no metric has been provided to show how effective 3D cellular identity analysis is. The behavioral experiments regarding C2 neurons are not consistent with the previously published functional imaging analysis from the same group. Further functional imaging experiments are necessary to support the claims in the paper.

1. The whole organ imaging method developed here will be a very useful technique for Drosophila larval chemosensory imaging. However, the details of image capture and processing is somewhat vaguely explained in the manuscript. Multiple details and metrics are missing about how imaging is done. What is the laser power used in these experiments? What is the frame rate for each volume scanned? What is the size of each volume, field of view etc. The movement and the drift correction for calcium imaging makes sense, however more detailed explanation about how they process imaging data is necessary. For example, in Figure Supplement 1, the authors are correcting GFP channels using the position information from the RFP channel and their method is able to handle the xy-plane. However, it is not clear how they handle z-drift. Did the post-processing also apply for the movement and drift correction in the Z-axis? What is the error rate for x-y or z correction? It would be good to know how accurate their methods are in identifying the same cell identities across animals. The authors need to expand the methods section and explain clearly how volumetric imaging and data processing was done in the paper. This is key for accurate interpretation of cell identification and the calcium imaging results.

Thank you for your observation, we now added to the description in the Methods section Microscopy settings, lines 619-698.

3D drift correction explanations and corresponding reference for the drift correction plugin employed are included in Methods and we added additional specifications under Data processing and analysis of whole-organ physiological recordings (from line 744).

2. Is figure 2B and 2C showing the percent responsive cells recorded across multiple animals or the percent responsive cells within an animal? I think it is important to clarify this point.

Thank you for pointing this out, we recognize the importance for this elucidation and have attempted to describe the data more clearly. Figure 2 shows data pooled across multiple recordings, responding neurons were pooled together and the multimodality percentage was thus calculated over the entire dataset. Figure 2 B and C represent all responsive cells from organs stimulated with mixtures of tastants, while Figure 2 D and D’ represent all responsive cells from organs stimulated with single tastants.

In Figure 2-figure supplement 1 we have now also illustrated data per animal, as alternative representation of Figure 2 D’. This newly added figure shows the unimodality vs multimodality percentage calculated within each of the 15-recorded organs, giving a total average of 67% unimodality vs 33% multimodality. These numbers are very closely mirroring the 68%/32% fraction from Fig2D’ for all pooled responses.

What is the variance between individuals? I would like to understand how reliable cell identification and functional imaging analysis is across animals. This can be achieved if authors show data per animal rather than lumping all the data across the animals together. Another reason to visualize individual animal responses is to investigate whether the order of tastant stimulation is a critical factor for larval taste perception: For a neuron that can respond to bitter and sugar, would the sequence of stimulation matter? Will the neuron change activity if it was stimulated first with a bitter compound then a sugar compound? What about the opposite? I strongly encourage authors to show more traces for individual animals across trials with different tastant stimulations as done in Figure 2D. This will help the reader understand what is the variation in taste responses across animals and it will also increase confidence for the consistency of authors' volumetric imaging method.

Thank you, we agree with your observations and with the validity of the raised points.

We have included more traces in Figure 2 D for 2 separate representations of traces per organ. We have also added in Figure 2-Figure Suppl1 an alternative representation per each recorded organ of the multimodality computations done on pooled data in Figure 2D’ as mentioned under the previous reviewer comment. This will give a feel of how variable the captured neuronal responses were across organs, without taking into account cell identity mapping.

Regarding order of tastants: this aspect was incorporated in the experimental design through a randomized order of stimulation. We considered through this approach an eventual effect caused by the order of application would be averaged out as noise. To exemplify this, we have now added as Figure 2-Supplement 2 single neuron calcium imaging recordings in C7 (TOG neuron activated by both sugar and bitter modalities – van Giesen, 2016a). The order of presentation of two substances, sucrose and denatonium, yields no significant difference in sucrose responses recorded in this neuron.

We also replaced Figure 3 to add illustrative traces for some of the mapped neurons. To your comment we would like to add that ease in identifying cell identities equally depends on the nature of recorded tissue. For TOG cells we have observed relatively high variability of cell localization in live imaging as well as in immunostainings due to a supple architecture of the organ. On the other hand, we have found the DOG organ, which does not make the subject of this work, to be more compact with fewer variables in cell position. Similarly, the adult proboscis has a fairly rigid conformation and hair bristles that aid for referencing in cell identification. All in all, consistency of our imaging tools in the TOG can be improved through future efforts by using complementary methods for cell labeling and/or tissue registration.

3. In figure 3A, the authors show different calcium dynamics upon tastant stimulation in different TOG neurons. They have observed, off responses and inhibition of neural activity in certain neurons upon stimulation. I would like to see what is the percentage of OFF-responses and inhibition observed in these experiments. Are these events frequent or rare? Are they seen in all animals or some of the animals used in the calcium recordings? It is important to understand whether these are biologically relevant neural responses or artifacts seen in some of the animals used in the experiments.

OFF-responses have been assigned when a neuron showed activation after the stimulation, and a deactivation response was considered when a neuron responded with a decrease in florescence during stimulation. It remains to be determined if the 2 types of events are related to distinct cellular mechanisms or if they are coupled, but it is often the case that a neuron responds with both deactivation and a subsequent OFF-response to a tastant.

In particular, such events have been systematically assigned to the CDL2 neuron as presented in Figure 3-figure supplemental 1 B. Citric acid, employed for stimulation in all 15 organs subjected to individual tastants, yielded a deactivation in an approximated cell location named by us CDL2 in 9 out of 15 recorded organs (plotted in Figure 3-Figure Suppl.1 B). An OFF-response to citric acid in the same neuron was actually captured in 11 out of 15 organs. The two numbers don’t coincide because the amplitude for deactivation or OFF-response didn’t always simultaneously meet the criteria of 20% threshold, but all in all we consider a deactivation/OFF-response to citric acid in CDL2 neuron to be more than merely an artifact.

Other tastants than citric acid have elicited OFF-responses but less robustly, especially in CDL2 but also in other neuronal locations. Generally, we believe such events are neuron and tastant/concentration specific and that a larger range of tested tastants and concentrations would reveal further such neuron-deactivation pairings. However, CDL2 associated most deactivation signals in our dataset (included in source data supplementary files), indicating this neuron could comport distinctive tuning properties.

4. In Figure 3, authors replot the results from Figure2 this time by identifying cellular identity using anatomical position. This is something similar to what I was trying to get to in my previous comment (see comment 2). But even in plots shown in Figure 3C, it is hard to understand how successful authors' methods are in classifying each neuron type in each animal. For example, how many larvae in total have been tested in Figure 3C and in what percentage of the animals they were able to identify each neuron type and what is the variation in neural responses among each animal. I think this is a very critical point that needs to be clarified in the paper variation of taste responses among animals and a metric to show how successful cellular identification is across all the animals tested. If my understanding is correct, the total number of tested organs/animals is 15, however the data plotted in Figure 3C does not correspond to this number. Please explain what is the success rate to define a cell type across animals.

We believe the whole-organ approach described here can serve for future characterizations with more precise neuronal annotation results by using complementary tissue registration or double neuronal labeling techniques. Within the purpose for this study we have manually mapped cells for which a tastant was found to yield a response in at least 3 different instances. Therefore, for some data plotted in the Figure 3-Figure Suppl1 (previously Figure 3) the N was 3, while a given tastant was presented in at least 7 recordings, except for sucrose and citric acid that were used for stimulation in all 15 organs. We have now restricted the threshold to N>=4, being at least half of stimulated organs with a given tastant, except for bitter stimulations for which N>=3 as these responses were more fade and more difficult to map. This threshold does not necessarily mean a given neuron only responded in 3-4 organs out of 7-8 stimulated with a given tastant but it can also mean one of the following:

1. The response was not picked up above 20% DF/F0 amplitude or

2. the fluorescent trace was filtered out due to unstable baseline or high variance of signal before stimulation or

3. The response could not be unambiguously assigned to the given cell location or

4. the experimenter was not confident about the cell identity even if a response to the given tastant was picked up.

Therefore we would like to propose Figure 3-Figure Suppl.1 should be interpreted as illustration of cell tuning types that can be picked up in the TOG, rather than a complete cell mapping of the organ.

Nevertheless, the reproducible mono/multimodality percentages (illustrated per organ in Figure 2-Figure Suppl.1) and the captured deactivation/OFF-responses add to our understanding of the taste sensing range in the larval external organ.

5. The authors mention that the only known sugar receptor gene Gr43a is not expressed in TOG but TOG neurons can respond to sugars. This is a puzzling finding. Is the expression analysis of Gr43a receptor based on transgenic reporters or single cell sequencing? In other words, is there other evidence that TOG does not express any known sugar receptors in larvae?

Yes, indeed, thus far a sugar receptor expression in the larval TOG has not been reported in previous efforts for transgenic lines mapping. Except for a newly identified appetitive taste sensing to ribonucleosides via Gr28a (Mishra et al., 2018) and to certain amino acids via IRs (Croset et al., 2016), not much has been described for attractive behaviors attributed to neurons/receptors in the larval TOG.

Known sugar receptors (Gr5a, Gr64e-f) rely on description in the adult fly, but these have not been detected in the larval TOG. Conversely, certain Gr receptors are larval specific and incompletely characterized and therefore might cover sugar sensing at this developmental stage.

According to studies using transgenic reporters, larval Gr43a – similar to adult Gr43a – is expressed in internal tissues (pharyngeal taste neurons, gut, brain) but not in external sensing neurons.

We are presenting an enticing opportunity to investigate novel sugar receptors based on these results.

6. In figure 4, authors performed behavioral experiments to test the function TOG sugar sensing in larvae. Inhibiting one of the TOG neurons, C2 causes defects in sucrose preference when agarose ss 2.5 % but not 1%. They claim the external sugar sensing via TOG sensory neurons is critical when food is not soft enough to ingest. These are interesting hypothesis however, in their previous study, authors have shown C2 neurons respond weakly to 500mM sucrose but not to any other sugar. However, in this manuscript the stronger behavior phenotype is seen in 250mM fructose preference upon silencing of C2 neurons. It is not clearly to me what is the mechanism here. Please explain.

Indeed, we found C2 to respond physiologically to fructose 100mM with higher amplitudes than to fructose 500mM (data not shown), therefore this neuron’s role in fructose detection might depend to a large extent on concentration. Given that sucrose is a disaccharide constituted by fructose and glucose, we would expect C2 to also respond at least partially to fructose. On the other hand, it seems the Gr43a role in sugar detection is more specific for fructose than for sucrose, as also reported by others (Miyamoto and Amrein 2013, Miyamoto et al., 2012) and as we show in Figure 4-Figure Suppl.1 B.

Since the phenotype yielded by C2 in sucrose preference is more pronounced, we chose to focus on this sugar and to not include fructose preference indexes or physiological activation responses in C2, for simplicity and overall readability of the data.

7. In general, I think it will help to use one nomenclature for the paper or provide a table matching C-X identity to cellular identities in Figure 3C and list which compound(s) each neuron responds to and indicate the type of response; ON, OFF or inhibition.

Thank you for your suggestion, a table is included in the supplementary information (corresponding source data excel table).

8. The manipulations for larval GRNs in the manuscript are done chronically by expressing TNT-E. This might impact the development and wiring of the larval chemosensory system and cause problems in sugar detection unrelated to GRN function. I think using a conditional inhibitor of neural activity such as shits or newly developed ontogenetic inhibitors, GtACRs, might be better to use here to show conditional inactivity of the neurons result in similar defects in sugar sensing.

This is an excellent suggestion and we now incorporated in Figure 4 D new data for C2 neuron behavior for temporally controlled silencing with GtACR1. This experiment reproduces a defect to sucrose 500mM on 2.5% substrate, as observed with TNTE silencing.

Reviewer #3:

Drosophila larvae have been put forth as a good model to study taste because of the simplicity of the system (in terms of the number of organs and neurons) and efficient behavior assays. However, an understanding of taste coding in the larval system lags behind that described for the adult stage, in which most taste sensilla/neurons are accessible for recordings or imaging. Maier et al., seek to address this gap by developing a whole organ calcium imaging approach to assess responses of neurons in the larval terminal organ to sweet, bitter, salt, amino acid, and acid tastants. From imaging 15 organs, the authors describe responses to a set of sweet, bitter, salt, amino acid, and acid tastants, collectively represented in neurons that appear to be tuned to one, two or multiple categories of tastants.

Although the new approach has promise, the current study makes a limited conceptual advance and the number of positive responses used to ascribe responses to neurons is small. Overlap of tastant sensitivities has now been described, and more rigorously so, in the adult stage and in the authors' own previous work focusing on genetically labeled subsets of taste neurons. The identification of responsiveness to sucrose, which may be somewhat unexpected given the absence of sweet Gustatory receptors (Grs) in larvae, has also been previously described for a few neurons, one of which is further studied in behavior assays.

The imaging analysis is followed by investigation of the potential roles of sugar sensing neurons in the periphery. While this analysis provides further support for a previously identified peripheral neuron (C2), it does not convincingly parse contributions of internal and external neurons.

Specific weaknesses are elaborated via points below and could be addressed either with additional experiments or textual modifications.

Calcium imaging:

– Neuron categories were determined by imaging responses to only a single concentration of each tastant; more careful dose response analysis may alter the conclusions. This possibility has not been considered.

Indeed, this is a valid observation, and we have in fact noticed an important difference in responses to different concentrations of the same tastants, in fact as also exemplified by distinct responding cells and divergent behaviors elicited by low and high salt concentrations. This possibility is also discussed in the Discussion section.

Given the high level of complexity we faced in our analysis, particularly as the cell identity mapping was approached manually in this study, dose response would exponentially add to the complexity rendering manual response annotations impossible, and therefore we have focused on unique concentrations. We judge our tools valuable for future efforts in elucidating this complexity in tuning using various tastant concentrations with complementary methodology for tissue registration or double labeling of neurons that would make room for computerized cell response annotations.

– There is no effort to reconcile the imaging results with predictions from previous studies. For example, the number of bitter-responsive neurons is 1-2 per larva. However, bitter-GAL4 lines have been mapped to many more neurons in the TO.

In effect, our approach comports limitations in cell identification, as mentioned above. However, physiological characterization of bitter neurons in the TOG is relatively limited. Namely, neurons described to comport aversive behavior in the TOG are more or less 3: C1 to a diversity of bitter tastants (Kim et al., 2016, Choi et al., 2020), C3 to quinine (Apostolopoulou et al., 2014) and C7 to denatonium and quinine (van Giesen 2016a) and these are also summarized in Figure 5A. Any other contributions to bitter taste in the larva is inferred based on transgenic expression of GRs characterized as bitter sensing in the adult but not physiologically associated to bitter sensing in the corresponding larval neurons. All considering, bitter responding neurons per organ identified in our study is conform in number to previous physiological characterizations.

– The correspondence between neurons named from previous GAL4 mapping experiments to those named by position in this study is not at all clear. A better comparison between the two could be presented.

As restated in several other places above, the manual cell identification based on position did not allow for a high confidence association with previously Gal4-mapped neurons. We believe the tools presented by our manuscript can serve future studies in bridging the gap by using double labeling of neurons.

–The criteria that were used to manually exclude false positive responses in calcium imaging experiments (Line 475-476) have not been described.

Thank you for raising this valid point, a description is added under the corresponding Methods section Data processing and analysis of whole-organ physiological recordings, from line 744.

– A response was ascribed to an identified neuron if the position was the same in >3 organs. This threshold seems low, considering that 7-8 organs were tested with each stimulus. In the table for Figure 3, for instance, several tastants have n=3. Is this from a higher number of tastant applications? If so, what was observed in the other cases – that a response could not be ascribed to the identified neuron or that there was no response at all?

In other cases (instances where a response could not be ascribed to a given cell) the response was either not picked up / not above the set threshold of 20% amplitude, or the responding cell could not be confidently associated to a given identity.

Therefore we would like to propose Figure 3-Suppl.1 should be interpreted as illustration of cell tuning types that can be picked up in the TOG, rather than a complete cell map of the organ. The TOG tissue is flexible and shows a fluid structure, much more than its DOG counterpart that is more compactly organized. Therefore we anticipate the above question will be addressed by other studies by using a computerized tissue registration.

Nevertheless, the reproducible mono/multimodality percentages (illustrated per organ in Figure 2-Suppl1), the sucrose responses in the TOG, as well as the captured deactivation/OFF-responses, all add to our understanding of the taste sensing range in the larval external organ.

Behavior:

– There are no indications of statistically significant differences between the genotypes in Figure 4B; the legend only mentions that the preference of Gr43a mutants is not significantly different from zero. Finally, data in Figure 4C are compared with Mann-Whitney tests, but ANOVA should be used instead.

Thank you for your very useful comment; we have now replaced Mann-Whitney by ANOVA tests, except for specific comparisons where 2 genotypes instead of 3 were measured in Figure 4-Supplement1. All statistical comparisons are indicated in the Supporting information (source data excel files).

– Whether sucrose response of internal neurons is affected in Gr43a mutants has not been tested (relevant to Figure 4C).

Thank you for raising this point – we have tested and now included this experiment in Figure 4-Figure Suppl.1 B. In our hands the Gr43a mutant shows a mild defect in sucrose preference (Figure 4 – Figure Suppl.1 B), which is confirmed from observations in the adult system where Gr43a has been described as an internal nutrient sensor (Miyamoto and Amrein 2013, Miyamoto et al., 2012) more specifically tuned to fructose than to other sugars, including sucrose (disaccharide of fructose and glucose).

– In Figure 4B right, assuming that Gr43a mutant controls are not significantly different from zero, whereas Gr43a "brain only" are, I don't know how to interpret a potential role for Gr43a brain neurons in preference for 2.5% agarose food, at 2 minutes when the larvae are not consuming anything.

We have now included more data points and proposed a supporting model in Figure 5. Indeed, we suggest that even for immediate preference at 2 min on 2.5% in absence of food ingestion a Gr43a mutant can still weigh on the output behavior. This could be due to Gr43a expression in several internal tissues (pharynx, proventriculus, brain) and to its role as hemolymph fructose sensor. It has been shown that hemolymph carbohydrates influence larval feeding behavior though insulin signaling and SLC5A11 (Ugrankar et al., 2018), but it remains to be determined if and how does a Gr43a mutant correlate with metabolic pathways and possibly interfere with sugar preference output behavior.

Furthermore, the data added for Gr43a mutant behavior in Figure 4-Figure Suppl.1 shows a lesser defect at higher concentrations and lesser in sucrose preference than fructose, implying a possible systemic effect of Gr43a more specifically to fructose, as mentioned under reviewer’s previous comment.

– The calcium imaging analysis is at the cellular level, not looking into contributions of receptors. Manipulating cells rather than receptors for behavior experiments would allow for better comparison between the two aims of the study. If the Gr43a mutant is the only way to go, suggested experiments are:

1. Test if the complementary brain ablated genotype (Gr43a internal taste only) yields the expected phenotypes on 1% and 2.5% agarose.

Thank you very much for your suggestions. We have now added the complementary brain silencing genotype by reproducing the corresponding crosses described in the original paper by Mishra et al., 2013. We tested therefore Gr43a-Gal4-ki; Cha-Gal80 x UAS-TNT (Gr43a brain receptor silenced) on 1% and 2.5% (Figure 4-Figure Suppl.1 C). In the reference study for this test, the brain ablated Gr43a larvae don’t show a defect compared to control but on our hands a deficient preference was calculated for immediate preference on 1% and no defect on 2.5% percent. This difference with the reference paper might be due to laboratory dissimilarities in experimental manipulation, in fly food differences linked to subliminal metabolic state variations between animals or perhaps due to distinct potency of the UAS-TNTE line employed. A lack of defect of brain Gr43a silencing suggests that internal expression from pharynx and gut tissues is sufficient for internal sugar sensing.

2. Test if peripheral sugar responses are affected in Gr43a mutants.

Thank you, we followed your suggestion and found that Gr43a mutants associate some defect in sucrose preference but lesser than to fructose on both 1% and 2.5% (Figure 4-Figure Suppl.1 B and B’). This confirms adult fly reports describing Gr43a to have the highest specificity for fructose among sugars (Miyamoto and Amrein 2013, Miyamoto et al., 2012), including compared to sucrose (disaccharide of fructose and glucose).

https://doi.org/10.7554/eLife.67844.sa2

Article and author information

Author details

  1. G Larisa Maier

    Department of Biology, University of Fribourg, Fribourg, Switzerland
    Contribution
    Conceptualization, Data curation, Methodology, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Nikita Komarov
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4486-8296
  2. Nikita Komarov

    Department of Biology, University of Fribourg, Fribourg, Switzerland
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Contributed equally with
    G Larisa Maier
    Competing interests
    No competing interests declared
  3. Felix Meyenhofer

    Department of Biology, University of Fribourg, Fribourg, Switzerland
    Contribution
    Software
    Competing interests
    No competing interests declared
  4. Jae Young Kwon

    Department of Biological Sciences, Sungkyunkwan University, Suwon, Republic of Korea
    Contribution
    Conceptualization, Supervision
    Competing interests
    No competing interests declared
  5. Simon G Sprecher

    Department of Biology, University of Fribourg, Fribourg, Switzerland
    Contribution
    Conceptualization, Funding acquisition, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    simon.sprecher@unifr.ch
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9060-3750

Funding

Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (310030_188471)

  • Simon G Sprecher

Novartis Stiftung für Medizinisch-Biologische Forschung (18A017)

  • Simon G Sprecher

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors thank the Sprecher lab for constructive discussions, and especially to Cornelia Fritsch for lab management support. The authors are grateful to Boris Egger for guidance on microscopy and data analysis. Many thanks for fly stocks to Hubert Amrein, Shanhaz Lone, Bloomington Drosophila Stock Center.

Senior Editor

  1. K VijayRaghavan, National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

Reviewing Editor

  1. Sonia Sen, Tata Institute for Genetics and Society, India

Reviewer

  1. Sonia Sen, Tata Institute for Genetics and Society, India

Publication history

  1. Received: February 24, 2021
  2. Accepted: November 23, 2021
  3. Accepted Manuscript published: December 3, 2021 (version 1)
  4. Version of Record published: December 24, 2021 (version 2)

Copyright

© 2021, Maier et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 635
    Page views
  • 174
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Neuroscience
    Christian Brodbeck et al.
    Research Article

    Speech processing is highly incremental. It is widely accepted that human listeners continuously use the linguistic context to anticipate upcoming concepts, words, and phonemes. However, previous evidence supports two seemingly contradictory models of how a predictive context is integrated with the bottom-up sensory input: Classic psycholinguistic paradigms suggest a two-stage process, in which acoustic input initially leads to local, context-independent representations, which are then quickly integrated with contextual constraints. This contrasts with the view that the brain constructs a single coherent, unified interpretation of the input, which fully integrates available information across representational hierarchies, and thus uses contextual constraints to modulate even the earliest sensory representations. To distinguish these hypotheses, we tested magnetoencephalography responses to continuous narrative speech for signatures of local and unified predictive models. Results provide evidence that listeners employ both types of models in parallel. Two local context models uniquely predict some part of early neural responses, one based on sublexical phoneme sequences, and one based on the phonemes in the current word alone; at the same time, even early responses to phonemes also reflect a unified model that incorporates sentence level constraints to predict upcoming phonemes. Neural source localization places the anatomical origins of the different predictive models in non-identical parts of the superior temporal lobes bilaterally, with the right hemisphere showing a relative preference for more local models. These results suggest that speech processing recruits both local and unified predictive models in parallel, reconciling previous disparate findings. Parallel models might make the perceptual system more robust, facilitate processing of unexpected inputs, and serve a function in language acquisition.

    1. Neuroscience
    Travis A Hage et al.
    Research Article

    Understanding cortical microcircuits requires thorough measurement of physiological properties of synaptic connections formed within and between diverse subclasses of neurons. Towards this goal, we combined spatially precise optogenetic stimulation with multicellular recording to deeply characterize intralaminar and translaminar monosynaptic connections to supragranular (L2/3) neurons in the mouse visual cortex. The reliability and specificity of multiphoton optogenetic stimulation were measured across multiple Cre lines and measurements of connectivity were verified by comparison to paired recordings and targeted patching of optically identified presynaptic cells. With a focus on translaminar pathways, excitatory and inhibitory synaptic connections from genetically defined presynaptic populations were characterized by their relative abundance, spatial profiles, strength, and short-term dynamics. Consistent with the canonical cortical microcircuit, layer 4 excitatory neurons and interneurons within L2/3 represented the most common sources of input to L2/3 pyramidal cells. More surprisingly, we also observed strong excitatory connections from layer 5 intratelencephalic neurons and potent translaminar inhibition from multiple interneuron subclasses. The hybrid approach revealed convergence to and divergence from excitatory and inhibitory neurons within and across cortical layers. Divergent excitatory connections often spanned hundreds of microns of horizontal space. In contrast, divergent inhibitory connections were more frequently measured from postsynaptic targets near each other.