The neuronal and molecular basis of quinine-dependent bitter taste signaling in Drosophila larvae

Abstract

The sensation of bitter substances can alert an animal that a specific type of food is harmful and should not be consumed. However, not all bitter compounds are equally toxic and some may even be beneficial in certain contexts. Thus, taste systems in general may have a broader range of functions than just in alerting the animal. In this study we investigate bitter sensing and processing in Drosophila larvae using quinine, a substance perceived by humans as bitter. We show that behavioral choice, feeding, survival, and associative olfactory learning are all directly affected by quinine. On the cellular level, we show that 12 gustatory sensory receptor neurons that express both GR66a and GR33a are required for quinine-dependent choice and feeding behavior. Interestingly, these neurons are not necessary for quinine-dependent survival or associative learning. On the molecular receptor gene level, the GR33a receptor, but not GR66a, is required for quinine-dependent choice behavior. A screen for gustatory sensory receptor neurons that trigger quinine-dependent choice behavior revealed that a single GR97a receptor gene expressing neuron located in the peripheral terminal sense organ is partially necessary and sufficient. For the first time, we show that the elementary chemosensory system of the Drosophila larva can serve as a simple model to understand the neuronal basis of taste information processing on the single cell level with respect to different behavioral outputs.

Keywords: Drosophila larvae; bitter; feeding; gustation; gustatory receptors; learning and memory; single cell.

Figure 1

Quinine affects larval choice behavior, feeding, survival, and associative olfactory learning. (A) Shows the dose-response curve for larval choice behavior for quinine concentrations between 0 (green) and 6 mM (red). Larvae significantly avoid quinine concentrations from 1 to 6 mM. (B) Depicts relative feeding on quinine. Only if 6 mM (red) is mixed into the substrate, larvae reduce feeding significantly with respect to baseline feeding at 0 mM quinine (indicated by the line). A lower concentration of 3 mM, quinine (orange) does not significantly reduce larval food intake. (C) Larval survival on agarose-quinine mixture diet depends on the quinine concentration (Kaplan–Meier survival curves). 3 mM quinine reduces larval survival compared to a pure agarose diet. The effect was even stronger for 6 mM quinine as more larvae died during the experiment. (D) 6 mM (red) quinine reinforces immediate negative olfactory associative learning when paired with a particular odor. Sample size for each box plot is n > 12. Differences against random distribution are given at the bottom of each panel. Differences between groups are presented above the related box-plots (n.s., non-significant p > 0.05, ^*p < 0.05, ^**p < 0.01, or ^***p < 0.001). Small circles indicate outliers.

Figure 2

GR66a and GR33a neuronal signaling is required for quinine-dependent choice behavior and feeding, but dispensable for survival and associative olfactory learning. (A,B) Show frontal views of CNS projections of GR66a-GAL4 and GR33a-GAL4 lines crossed with UAS-mCD8::GFP. To visualize the expression patterns, the CNS was stained with anti-GFP (green) and anti-ChAT; anti-FasII (magenta). In both lines, GRNs innervate the SOG in a similar way (arrows); sog, subesophageal ganglion; vnc, ventral nerve cord; hemi, brain hemisphere. (C,D) Genetically ablating the GRNs covered by GR66a-GAL4 or GR33a-GAL4 expression completely diminishes larval choice behavior for 6 mM quinine. Both experimental larvae behave significantly different from genetic controls [p < 0.001 in (C) and p < 0.01 and p < 0.001 in (D)] and do not show any aversive choice behavior (n.s.). (E,F) Genetically ablating GRNs covered by GR66a-GAL4 or GR33a-GAL4 expression partially increases feeding behavior on a substrate containing 6 mM quinine. GR66a/UAS-hid,rpr (E) and GR33a/UAS-hid,rpr (F) experimental larvae feed more than control larvae on a substrate containing 6 mM quinine [n.s. and p < 0.01 in (E); p < 0.01 and p < 0.001 in (F)]. Both experimental groups do not feed on baseline level, i.e., wild type control larvae on pure agarose substrate [p < 0.01 in (E); p < 0.001 in (F)]. (G,H) Larval survival is plotted as Kaplan–Meier survival curves; genetically ablating the GRNs covered by GR66a-GAL4 or GR33a-GAL4 expression does not reduce larval survival on a 6 mM quinine agarose mixture diet. Neither GR66a/UAS-hid,rpr (G) nor GR33a/UAS-hid,rpr (H) experimental larvae live longer than both control groups [log-rank test, n.s. for the GR66a-GAL4 control, but log-rank test, p < 0.001 for the UAS-hid,rpr control in (G); log-rank test, n.s. for the UAS-hid,rpr control in (H), but log-rank test p < 0.001 for the GR66a-GAL4 control also in (H)]. (I,J) Genetically ablating GRNs covered by GR66a-GAL4 or GR33a-GAL4 expression does not affect associative olfactory learning reinforced by 6 mM of quinine. GR66a/UAS-hid,rpr (I) and GR33a/UAS-hid,rpr (J) experimental larvae perform on the same level as respective control groups (n.s.) in an assay testing quinine-reinforced associative olfactory learning. Sample size for each box plot is n > 12. Differences against zero are given at the bottom of each panel. Differences between experimental groups are depicted above the respective box plots (n.s., non-significant p > 0.05, ^*p < 0.05, ^**p < 0.01, or ^***p < 0.001). Scale bars: 50 μm. Small circles indicate outliers.

Figure 3

Gr33a receptors but not Gr66a receptors are required for quinine avoidance. (A,B) GR66a-GAL4 and GR33a-GAL4 are expressed in the same set of six GRNs of the TO, schematically represented for GR66a-GAL4 (A) and GR33a-GAL4 (B). Two of the neurons belong to the dorsolateral group, called B1 and B2, and the other four to the distal group, called C1–C4. This suggests that both receptors are co-expressed in the same set of bitter sensing GRNs. (C,D) The GFP expression for GR66a-GAL4 (C) and GR33a-GAL4 (D) crossed to UAS-mCD8::GFP is shown in the terminal organ (to). The two GRNs B1 and B2 of the dorsolateral group have their cell bodies located within the dorsal organ ganglion (dog; arrow). The four GRNs C1–C4 of the distal group have their cell bodies in the terminal organ ganglion (tog; arrowhead). Anti-GFP (green) and anti-elav (magenta) antibodies were used to visualize the specimens. (E,F) In a frontal view part of the central projection of GR66a-GAL4 (E) and GR33a-GAL4 (F) is shown in the SOG (sog). The projections were traced by crossing the two GAL4 lines with UAS-mCD8::GFP. GFP is visualized by anti-GFP antibody staining (green) and the neuropil by anti-ChAT; anti-FasII antibody staining (magenta). (G) When testing different mutants for the GR33a and GR66a receptor, only mutants affecting GR33a receptor function [GR33a(1) and GR33aGAL4] show a reduced choice behavior in the presence of 6 mM of quinine, compared to an appropriate w1118 control group (p < 0.001). A mutant that abolishes GR66a receptor function (GR66aex83) does not change choice behavior (n.s.) compared to the appropriate w1118 control. Sample size for each box plot is n > 12. Differences against zero are given at the bottom of each panel. Differences between experimental groups are depicted above the respective box plots (n.s., non-significant p > 0.05 and ^***p < 0.001). mb, mushroom body; scale bars: 25 μm in (C) and (D); 20 μm in (E) and (F). Small circles indicate outliers.

Figure 4

Single TO neurons—behavior function correlation for quinine-dependent choice behavior. The first column shows a schematic overview of the respective GRNs included in the expression pattern of each GAL4 line, based on Kwon et al. (2011). The second column illustrates the expression pattern for each GAL4 line crossed with UAS-mCD8::GFP at the level of the TO (arrowheads mark the indicate the innervation of the terminal organ and the position of the respective cell body in the dorsal organ ganglion or terminal organ ganglion). The third column shows in frontal views part of the GRN projections in the SOG for each GAL4 line. These were crossed with UAS-mCD8::GFP and stained by anti-GFP (green) and anti-ChAT, anti-FasII (magenta) antibody staining. The last column shows the quinine-dependent choice behavior for each GAL4 line when crossed to UAS-hid,rpr in order to specifically induce cell death in small sets of GRNs or single GRNs. The analysis includes GR10a-GAL4 (A), GR36c-GAL4 (B), GR47b-GAL4 (C), GR94a-GAL4 (D), GR97a-GAL4 (E),GR57a-GAL4 (F), GR39ab-GAL4 (G), and GR59d-GAL4 (H). Ablation of the B2, C1, C2 neurons alone does not alter quinine-induced choice behavior [n.s. for each experimental genotype compared to both control groups in (A–D); except for the UAS-hid,rpr control in (D); p < 0.05]. However, ablation of C3 only (E; p < 0.05 to respective controls) or in combination with C2 (F; p < 0.05 and p < 0.001 to respective controls) reduces choice behavior significantly. Ablation of C1, C2, and C4 in combination (in H) significantly reduces choice behavior as well (p < 0.05 and p < 0.001 to respective controls); expression within GRNs other than in the TO is not analyzed here. For further details see also Kwon et al. (2011). The identity of the C4 neuron (in G,H) as proposed by Kwon et al. (2011) was not verifiable in our specimens (indicated by a “?”). Sample size for each box plot is n > 12. Differences between experimental groups are depicted above the respective box plots (n.s., non-significant p > 0.05, ^*p < 0.05, and ^***p < 0.001; Small circles indicate outliers). mb, mushroom body; to, terminal organ; dog, dorsal organ ganglion; tog, terminal organ ganglion; sog, subeosophageal ganglion; scale bars: 20 μm.

Figure 5

Artificial activation of small set of GRNs or of the single C3 neuron is sufficient to induce an aversive choice behavior. Expression of the mammalian vanilloid receptor protein (VR1) allowed to quantify the behavioral relevance of GR66a-GAL4- and GR97a-GAL4-positive neurons after capsaicin-dependent activation. (A) Experimental GR66a/UAS-VR1 larvae avoided a 50 μM capsaicin agarose mixture against pure agarose (p < 0.001). The behavioral response was significantly different than the appropriate genetic controls (p < 0.001 each) that did not avoid capsaicin (n.s.). (B) Activation of a single GRN C3 in the TO in GR97a/UAS-VR1 experimental larvae induced an aversive capsaicin-dependent choice behavior (p < 0.001) that was significantly different from both genetic controls (p < 0.01 for both controls). Thus, activation of the single GRN C3 is sufficient to elicit gustatory-guided choice behavior. Sample size for each box plot is n > 12. Differences against zero are given at the bottom of each panel. Differences between experimental groups are depicted above the respective box plots (n.s., non-significant p > 0.05, ^**p < 0.01, and ^***p < 0.001). Small circles indicate outliers.

Figure 6

Schematic overview of the neurons signaling quinine-dependent bitter taste in Drosophila larvae. According to Kwon et al. (2011), GRs are expressed in only 10 neurons of the two major chemosensory organs of the larva, the dorsal organ (DO) and the terminal organ (TO). The two GRNs of the DO are called A1 and A2 (DO group). Additionally, B1 and B2, which are located in the dorsolateral group of sensilla of the TO, have their cell bodies sitting in the DO ganglion (TO dorsolateral group). C1–C6 located in the distal group of sensilla of the TO have their cell bodies located in the TO ganglion (TO distal group). Bitter quinine taste information affecting larval choice behavior is mediated mainly by the TO neurons C1–C4 (pink) and especially by the single TO neuron C3 (red). Signals reach the subesophageal ganglion via the maxillary nerve. Output of the single C3 neuron is necessary for quinine-dependent avoidance and artificial activation of the C3 neuron is sufficient to elicit aversive choice behavior.