Gustatory receptors required for avoiding the insecticide, l-canavanine

Abstract

Insect survival depends on contact chemosensation to sense and avoid consuming plant-derived insecticides, such as l-canavanine. Members of a family of ~60 gustatory receptors (GRs) comprise the main peripheral receptors responsible for taste sensation in Drosophila. However, the roles of most Drosophila GRs are unknown. In addition to GRs, a G-protein coupled receptor (GPCR), DmXR, has been reported to be required for detecting l-canavanine. Here, we showed that GRs are essential for responding to l-canavanine, and that flies missing DmXR displayed normal l-canavanine avoidance and l-canavanine evoked action potentials. Mutations disrupting either Gr8a or Gr66a resulted in an inability to detect l-canavanine. We found that l-canavanine stimulated action potentials in S-type sensilla, which are where Gr8a and Gr66a were both expressed, but not in Gr66a-expressing sensilla that did not express Gr8a. l-canavanine-induced action potentials were also abolished in the Gr8a and Gr66a mutant animals. Gr8a was narrowly required for responding to l-canavanine, in contrast to Gr66a, which was broadly required for responding to other noxious tastants. Our data suggest that GR8a and GR66a are subunits of an l-canavanine receptor, and that GR8a contributes to the specificity for l-canavanine.

Introduction

Contact chemosensation is crucial for animals to discriminate safe, nutritious foods from toxic substances in the environment. In insects such as the fruit fly, Drosophila melanogaster, this ability is essential for survival since many plants produce compounds with insecticidal properties. One such botanically derived insecticide is l-canavanine, which is an analog of the amino acid l-arginine (Dahlman and Rosenthal, 1975; Rosenthal and Dahlman, 1986; Rosenthal, 2001). Consumption of l-canavanine causes lethality to fruit flies and many other insects since this amino acid derivative is incorporated into proteins in place of l-arginine, thereby hindering protein function.

In fruit flies, tastants are detected in gustatory receptor neurons (GRNs), which are housed in gustatory hairs (sensilla) (Vosshall and Stocker, 2007; Montell, 2009). The main class of taste receptors consists of ~60 gustatory receptors (GRs), and these multipass transmembrane proteins are unrelated to G-protein coupled receptors (Clyne et al., 2000; Robertson et al., 2003; Montell, 2009). Currently, only four GRs (GR32a, GR33a, GR66a and GR93a) are known to be required for sensing aversive tastants, ranging from caffeine to quinine, strychnine and the synthetic repellent, DEET (Moon et al., 2006; Lee et al., 2009; Moon et al., 2009; Lee et al., 2010). In contrast to the fructose receptor (GR43a) (Sato et al., 2011) and possibly the glycerol receptor (GR64e) (Wisotsky et al., 2011), the subunit composition of GRs that respond to sugars (Dahanukar et al., 2007; Jiao et al., 2007; Jiao et al., 2008) and to aversive chemicals is complex and appears to consist of at least three subunits (Lee et al., 2009; Moon et al., 2009). The TRPA1 channel as well as GRs are expressed in GRNs and respond to noxious compounds (Kang et al., 2010; Kim et al., 2010). In addition to GRs and TRP channels, a G-protein coupled receptor (GPCR), DmXR, encoded by the mangetout (mtt) gene has been reported to be required for the behavioral avoidance to l-canavanine (Mitri et al., 2009).

Here, we found that both the behavioral and electrophysiological responses to l-canavanine were indistinguishable between mtt mutant and control flies. Rather, Gr8a and Gr66a were indispensible for sensing l-canavanine. While Gr66a was required for detecting multiple aversive tastants (Moon et al., 2006; Lee et al., 2010), Gr8a mutant flies responded normally to all other repellent compounds tested. We propose that GRs that function in the sensation of aversive tastants are comprised of assemblies consisting of a narrowly tuned subunit, as well as at least one broadly required co-receptor.

Materials and Methods

Generation of the Gr8a¹ mutant and transgenic flies

The P-element line, G17495, which was inserted in the Gr8a coding region, was donated by Genexel (Korea). To generate the Gr8a¹ mutant, the P-element was mobilized by genetically introducing transposase using the P[ry(Δ2-3)] line. The Δ2-3 transgene results in a high level of transposase expression as a result of deleting an intron that is normally spliced out inefficiently (Laski et al., 1986). To identify a deletion that removed Gr8a, we conducted a PCR-based screen using the following primers: P1, TCCGAAACATACAGTGGCGTCGAT and P2, TGTTCTTCGTGCTGACTGGCTACA (Fig. 2A,B). We recovered one P-element excision line out of 200 lines screened. The deletion was confirmed by genomic PCR using the following primers: P2 and P3, GCTCAAATGTTTCCAGGGTGCCTT (Fig. 2A,B). To determine the ends of the deletion, we amplified the genomic DNA and obtained the DNA sequence. The deletion was 934 base pairs and removed the sequences encoding the amino acid residues 15 to 305.

Figure 2.

GRs required for l-canavanine avoidance. A, Physical map of the Gr8a genomic region. The P-element (G17495) that inserted in the Gr8a coding region is indicated by the inverted triangle. The deletion and precise excision were confirmed using the indicated PCR primers (arrows) to amplify genomic DNA (P1, P2 and P3), followed by DNA sequencing. B, PCR analyses of genomic DNA. The PCR products were generated using the indicated fly stocks and PCR primer pairs, and fractionated on an agarose gel. C, Screen for Grs required for l-canavanine avoidance. The two-way choice tests were performed using the indicated chemicals and fly stocks. n=5-15. D, Dose-avoidance relationship determined using the two-way choice behavioral assay. Flies of the indicated genotypes were tested using l-canavanine over a range of concentrations (1-30 mM). n=4-11. E, Two-way choice assays using 5 mM sucrose plus 5 mg/ml erioglaucine (in the presence or absence of 30 mM l-canavanine) versus 5 mM sucrose plus 20 mg/ml sulforhodamine B. F, Rescue of l-canavanine avoidance defects by expressing wild-type Gr transgenes in GRNs using the GAL4/UAS system. n=6-13. The error bars indicate SEMs. *P<0.05, **P<0.01.

To obtain P[UAS-Gr8a] and P[UAS-Gr66a] transgenic flies, we prepared RNA from the labellum of wild-type flies and amplified the Gr8a and Gr66a cDNA by RT-PCR. We subcloned the cDNA into the pUAST vector (Brand and Perrimon, 1993). The transformation vector was injected into w¹¹¹⁸ embryos.

Fly stocks

We reported the following mutants previously, and deposited them with the Bloomington Stock Center: Gr33a^GAL4 (Moon et al., 2009), Gr66^ex83 (Moon et al., 2006) and Gr93a³ (Lee et al., 2009). mtt^f06268 and Df(2R)Exel7096 were obtained from Y. Grau (Mitri et al., 2009). H. Amrein provided the Gr32a¹ (Miyamoto and Amrein, 2008) and the P[Gr66a-GAL4] flies (Thorne et al., 2004). The P[Gr8a-GAL4] flies and P[Gr66a-I-GFP] were from J. Carlson (Weiss et al., 2011) and from K. Scott (Wang et al., 2004), respectively. The fly stocks were outcrossed for at least five generations into a w¹¹¹⁸ background, and w¹¹¹⁸ was used as the “wild-type” control.

Immunohistochemistry

The labella of GR8a-GAL4 /+; UAS-DsRed/GR66a-I-GFP flies were dissected and fixed using 4% paraformaldehyde with 0.2% Triton X-100 for 15 min at room temperature. The labella were washed three times with PBST (1X PBS and 0.2% Triton X-100), cut in half with a razor blade and blocked with 0.5% goat serum in PBST for 30 min at room temperature. The labella were transferred to new blocking buffer containing the primary antibodies and incubated overnight at 4°C: mouse anti-GFP (Molecular Probe, 1:1,000) or rabbit anti-DsRed (Clontech, 1:1,000). The labella were washed three times with PBST at room temperature and incubated with secondary antibodies (goat anti-mouse Alexa488 and goat anti-rabbit Alexa568) for 1 hour at room temperature. The tissues were washed three times with PBST and mounted in VECTASHELD (Vector laboratories). The samples were viewed using a Carl Zeiss LSM510 confocal microscope. The sensilla expressing Gr8a were identified by examining the dendrites of the GRNs in the fluorescent image and the bristles by DIC imaging.

Chemicals

Sucrose, caffeine, l-canavanine, DEET, denatonium, lobeline, papaverine, quinine, strychnine, l-arginine and N-methyl-l-arginine (NMA), sulforhodamine B and KCl were purchased from Sigma-Aldrich Co. Berberine sulfate trihydrate and brilliant blue FCF were obtained from Wako Pure Chemical Industries, Ltd.

Two-way choice behavioral assays

With the exception of the data shown in Fig. 1G and Fig. 2E, the two-way choice assays were conducted as described previously (Meunier et al., 2003; Moon et al., 2006). Briefly, we starved 50 flies (3-6 days old) for 24 hrs and then introduced the animals into 72-well microtiter dishes. Alternating well was filled with 1% agarose combined with one of two types of test mixtures: 1 mM sucrose, or 5 mM sucrose plus an avoidance chemical. To monitor food intake, one test mixture contained a blue dye (brilliant blue FCF, 0.125 mg/ml) while the other contained a red dye (sulforhodamine B, 0.2 mg/ml). After allowing the flies to feed for 90 min at room temperature in the dark, the animals were frozen at –20°C. The numbers of flies that were blue (N^B), red (N^R), or purple (N^MIX) were determined in a blind fashion based on the colors of the abdomen, and the preference indexes (P.I.) values were calculated according to one of the two following equations: (N^B+0.5N^mix)/(N^R+N^B+N^mix) or (N^R+0.5N^mix)/(N^R+N^B+N^mix), depending on the dye/tastant combinations. P.I.s = 1.0 and 0 indicated complete preferences for either the 1 mM or 5 mM sucrose, respectively. A P.I. = 0.5 indicated no preference between the two food alternatives.

Figure 1.

mtt was not required for l-canavanine-induced action potentials or for behavioral avoidance. A, Representative traces of l-canavanine triggered nerve firings from w¹¹¹⁸ and mtt^f06268 flies. The tip recordings were performed in S3 sensilla using 25 mM l-canavanine. B, Mean number of action potentials in response to 25 mM l-canavanine recorded from S3, S5, and S10 sensilla. The genotypes are indicated. Df stands for Df(2R)Exel7096. n=10-12. C, Tip recordings surveying nerve firings in GRNs from w¹¹¹⁸ labellar sensilla in response to 25 mM l-canavanine. Shown are the mean action potentials. n=10-18. D, Effects of red and blue food dyes on two-way choice assays. The flies were given a choice between 1 mM sucrose and 5 mM sucrose (plus or minus 30 mM l-canavanine). The red and blue bars indicate the color of the food dye added to the 1 mM or 5 mM sucrose. The dashed line indicates a P.I.=0.5, which results if there is no preference. n=5. Unless indicated otherwise, the two-way choice tests were performed with the 0.125 mg/ml brilliant blue FCF and 0.2 mg/ml sulforhodamine B. E, Effects of different concentrations of l-canavanine on the selection of 5 mM versus 1 mM sucrose. n=4-10 for each concentration. Shown are means ±SEMs. F, Two-way choice tests in the presence or absence of 30 mM NMA. G, Two-way choice tests using 5 mM sucrose plus 5 mg/ml erioglaucine (with or without 30 mM l-canavanine) versus 5 mM sucrose plus 20 mg/ml sulforhodamine B.

To test further a possible role for mtt in l-canavanine avoidance, we also performed two-way choice behavioral assays as described (Mitri et al., 2009). In these assays (Fig. 1G and 2E), we added 5 mM sucrose to all wells in the microtiter dishes. Alternating wells also contained either a different blue dye than described above (5 mg/ml erioglaucine), or 100-fold more of the red dye than used in our standard assays (20 mg/ml sulforhodamine B). Flies were allowed to feed in the dark for 2 hours. As reported, the high concentration of red dye was repulsive, causing the flies to prefer the 5 mM sucrose plus erioglaucine (Mitri et al., 2009). To test for aversion to l-canavanine, we added 30 mM l-canavanine to the 5 mM sucrose that contained the 5 mg/ml erioglaucine.

Electrophysiology

Tip recordings (Hodgson et al., 1955; Wieczorek and Wolff, 1989) were performed as described (Moon et al., 2006) using either 25 mM l-canavanine or 25 mM l-arginine. Briefly, newly eclosed flies were immobilized by inserting a glass capillary filled with Ringer's solution into the abdomen all the way to the head. This electrode also served as the indifferent electrode. The labellar hairs were stimulated with tastants dissolved in the buffer solution of the recording pipette (10-20 μm tip diameter). 1 mM KCl was used as the electrolyte for all the recordings. The recording electrode was connected to a preamplifier (TastePROBE, Syntech, Hilversum, The Netherlands), and the signals were collected and amplified (10x) using a signal connection interface box (Syntech) in conjunction with a 100-3000 Hz band-pass filter. Recordings of action potentials were acquired using a 12-kHz sampling rate and analyzed using Autospike 3.1 software (Syntech).

Data analyses

All error bars represent SEMs. Unpaired Student's t-tests were used to compare two sets of data. ANOVA with the Tukey post hoc tests were used to compare multiple sets of data. Asterisks indicate statistical significance (*P<0.05, **P<0.01).

Results

mtt was dispensable for l-canavanine induced action potentials and avoidance

Mutation of the mtt locus, which disrupts a GPCR (DmXR), has been reported to be necessary for the behavioral avoidance to l-canavanine (Mitri et al., 2009). In this prior study, the electrophysiological responses of gustatory receptor neurons (GRNs) were not tested. Therefore, we performed tip recordings to assess l-canavanine-induced action potentials. GRNs are housed in gustatory hairs (sensilla) situated on several body parts, including the main taste organ, the labellum, and are broadly grouped into three size classes (small, S; intermediate, I; large, L) (Vosshall and Stocker, 2007; Montell, 2009). We recorded from S3 sensilla, which respond to most aversive tastants (Weiss et al., 2011). In control flies, addition of l-canavanine stimulated action potentials (10.6±1.9), immediately upon contact with the recording pipet (Fig. 1A,B). Surprisingly, we found that mtt^f06268 mutant and wild-type flies displayed similar frequencies of l-canavanine-induced action potentials (Fig. 1A,B). Because mtt^f06268 produced normal l-canavanine action potentials, we confirmed the reported transposon insertion in mtt^f06268 by genomic PCR and analyzed mtt^f06268 in trans with a deletion that completely removed mtt, Df(2R)Exel7096. These latter flies also showed a wild-type electrophysiological response to l-canavanine (Fig. 1B).

Since the mtt^f06268 S3 sensilla responded normally to l-canavanine, we wondered whether other mutant sensilla might be impaired in l-canavanine-induced action potentials. To conduct this analysis, we first surveyed wild-type sensilla. Most I- and S-type sensilla, but none of the L-type sensilla respond to aversive compounds (Weiss et al., 2011). Therefore, we recorded l-canavanine-induced action potentials from the 10 S-type sensilla (out of 12 total) that were most accessible for recordings. In addition, we sampled two I-type and two L-type sensilla, the latter of which we did not expect to be stimulated by l-canavanine. The S3, S5 and S10 sensilla were most responsive to l-canavanine (>10 spikes/500 ms), while the S1, S6, S7 and S11 displayed a low level activity (Fig. 1C). The remaining three S-type sensilla tested, S2, S4 and S8, were insensitive to l-canavanine (Fig. 1C). None of the L- or I-type sensilla examined were stimulated by l-canavanine (Fig. 1C). Given the high responsiveness of wild-type S5 and S10, we recorded from these latter sensilla from mtt^f06268 and mtt^f06268/Df flies, and found that the responses were also indistinguishable from wild-type (Fig. 1B).

Due to the preceding results, we tested mtt^f06268 and mtt^f06268/Df for behavioral avoidance to l-canavanine using a two-way choice behavioral assay (Meunier et al., 2003; Moon et al., 2006). We starved young flies for 24 hours, and allowed them to feed for 1.5 hours in the dark in a microtiter dish that contained alternating wells consisting of 1 mM sucrose or 5 mM sucrose plus 30 mM l-canavanine. Each of the two tastants was mixed with red or blue food coloring. We then counted the number of flies with red, blue, or purple abdomens. Complete preferences for the 1 or 5 mM sucrose yield P.I.s of 1.0 and 0 respectively, while a P.I. of 0.5 results if there were a lack of bias for the two concentrations of sucrose. In the absence of l-canavanine, the control flies (w¹¹¹⁸) selected the higher concentration of sucrose, and this preference was unaffected by the red or blue food coloring (P.I.=0.07 or 0.04: Fig. 1D). When given a choice between 1 mM sucrose and 5 mM sucrose plus 30 mM l-canavanine, the w¹¹¹⁸ flies decreased their preference for 5 mM sucrose (Fig. 1D). We assayed the behavioral responses to a range of l-canavanine concentrations. In control flies, 1 mM l-canavanine did not affect the preference for 5 mM sucrose over 1 mM sucrose, while 20 and 30 mM l-canavanine induced large avoidance responses (Fig. 1E). Consistent with the electrophysiological results, the mtt^f06268 mutant displayed normal behavioral avoidance to l-canavanine. Wild-type and mtt^f06268 flies also showed indistinguishable l-canavanine avoidance using a modified two-way choice test employed previously to characterize mtt^f06268 (Fig. 1G, see Materials and Methods for detailed description). Based on the combination of electrophysiological and behavioral assays, we conclude that DmXR, which is encoded by mtt, was not required for detecting l-canavanine.

Requirement for GR8a and GR66a for avoiding l-canavanine

To address whether a Drosophila GR was required for avoiding l-canavanine, we tested the available mutants that disrupt GRs that function in sensing aversive chemicals. These include Gr32a, Gr33a, Gr66a, Gr93a (Moon et al., 2006; Lee et al., 2009; Moon et al., 2009; Lee et al., 2010) and Gr47a (Lee, Moon and Montell, unpublished). In addition, we generated a null mutation in Gr8a by mobilizing a P-element that inserted in the second exon, thereby deleting most of the coding region (Fig. 2A,B). The Gr8a¹ flies were homozygous viable and fertile.

We found that mutations disrupting Gr8a¹ and Gr66a^ex83 severely impaired l-canavanine avoidance (Fig. 2C). These mutant flies displayed the same preference for 5 mM sucrose, even at the highest concentration of l-canavanine (30 mM; Fig. 2D). Loss of the remaining four GRs that are known to affect the responses to aversive compounds had no impact on the repulsion to l-canavanine (Fig. 2C). These include Gr33a^GAL4, which is required for sensing all other noxious tastants tested (Moon et al., 2009), and Gr32a¹ (Lee et al., 2010), which affected the responses to all avoidance tastants examined, except caffeine. We tested Gr8a¹ using the previously described two-way choice test (Mitri et al., 2009), and found that the mutant was also impaired in l-canavanine repulsion using this alternative assay (Fig. 2E).

To confirm that the deficits in l-canavanine avoidance were due to the mutations affecting Gr8a and Gr66a, we tested for rescue of the mutant phenotypes with wild-type transgenes. Expression of UAS-Gr8a under control of either the Gr66a promoter (Gr66a-GAL4) or the Gr8a promoter (Gr8a-GAL4) rescued the Gr8a¹ phenotype (Fig. 2F). Similarly, we rescued the Gr66a^ex83 phenotype using UAS-Gr66a and either the Gr66a-GAL4 or the Gr8a-GAL4 (Fig. 2F). We detected expression of the Gr8a-GAL4 reporter in S-type sensilla (S1, S3, S5, S6, S7, S10 and S11) and one of I-type (I1) sensillum, but not other I- or L-type sensilla (Fig. 3), similar to that reported previously (Weiss et al., 2011).

Figure 3.

Expression of the Gr8a reporter in subsets of Gr66a positive GRNs. A-C, Confocal images. A, Gr66a-I-GFP labeled all bitter responsive GRNs (anti-GFP, green). B, UAS-DsRed was expressed under control of the Gr8a-GAL4 (anti-DsRed, red). Most but not all Gr8a-postive cells were detected in single confocal images C, Merged image of A and B. D, Schematic illustration of a fly labellum showing gustatory sensilla that expressed the Gr8a reporter (red). The nomenclature for the sensilla is as described (Hiroi et al., 2002). This differs somewhat from another nomenclature (Weiss et al., 2011).

Elimination of l-canavanine-induced action potentials

To test whether GR8a and GR66a were required for the electrophysiological responses to l-canavanine, we recorded from S3, S5 and S10 sensilla, which in control flies elicit the largest responses to l-canavanine. We found that l-canavanine-induced action potentials were eliminated in either the Gr8a¹ or Gr66a^ex83 mutants, and were restored upon expression of the wild-type transgenes that were driven under control of the Gr8a-GAL4 or Gr66a-GAL4 (Fig. 4A,B). The S3, S5 and S10 sensilla also responded to l-arginine, which is a structural analog of l-canavanine (Fig. 4C). l-arginine evoked similar frequencies of action potentials in Gr8a¹ as in wild-type (Fig. 4C). L-type sensilla, which did not express the Gr8a reporter (Weiss et al., 2011), also responded to l-arginine (Fig. 4D).

Figure 4.

GR8a and GR66a are indispensible for l-canavanine induced nerve firings. A, Representative traces of l-canavanine evoked action potentials from control flies (w¹¹¹⁸), Gr8a¹, Gr66a^ex83, and mutants with rescue transgenes (Gr8a¹;Gr8a-GAL4/UAS-Gr8a and Gr8a-Gal4/UAS-Gr66a;Gr66a^ex83). The recordings were performed from S5 sensilla. B, Tip recordings showing that GR8a and GR66a are required for l-canavanine induced nerve firings. The assays were performed from S3, S5, and S10 sensilla using animals of the indicted genotypes. n=10-12. The error bars represent SEMs. **P<0.01. C, Tip recordings showing l-arginine-induced action potentials in S3, S5 and S10 sensilla. n=10-11. The error bars represent SEMs. D, l-arginine induced action potentials in L4 and L6 sensilla. n=10-11. Shown are means ±SEMs.

Distinct requirements for GR8a and GR66a

Loss of GR66a diminishes the repulsion to caffeine, lobeline and papaverine and the synthetic insect repellent DEET (Moon et al., 2006; Lee et al., 2010). To investigate whether Gr8a¹ displayed impaired avoidance to other noxious chemicals, or was narrowly required for l-canavanine, we performed additional behavioral assays. Gr8a¹ displayed normal responses to all other repulsive tastants tested: caffeine, quinine, berberine, denatonium, lobeline, papaverine, strychnine and DEET (Fig. 5A). The frequencies of action potentials induced by application of these bitter chemicals were also similar between the mutant and the control flies (Fig. 5B). Thus, GR8a was more narrowly tuned to l-canavanine than GR66a.

Figure 5.

Behavioral and electrophysiological responses of Gr8a¹ to multiple noxious tastants. A, Behavioral responses of w¹¹¹⁸ and Gr8a¹ using two-way choice assays and one of the following concentrations of chemicals: 6 mM caffeine, 1 mM quinine, 0.05 mM berberine, 0.1 mM denatonium, 0.1 mM lobeline, 0.5 mM papaverine, 0.5 mM strychnine or 0.2% DEET. n=5-12. The error bars indicate SEMs. B, Mean nerve firing responses from w¹¹¹⁸ and Gr8a¹ by performing tip recordings from S6 sensilla using the indicated concentrations of chemicals: 10 mM caffeine, 1 mM quinine, 0.1 mM berberine, 1 mM denatonium; 1 mM lobeline; 1 mM papaverine; 1 mM strychnine; 0.2% DEET. n=11-18. Shown are means ±SEMs.

Discussion

GRs but not DmXR are required for avoiding l-canavanine

We conclude that GRs rather than the GPCR (DmXR) were essential for detecting the plant-derived toxin, l-canavanine. Gr8a and Gr66a were required since mutations disrupting each of these genes eliminated the behavioral repulsion to the insecticide. These deficits were more dramatic than those associated with mutations disrupting the sensation of other aversive tastants since avoidance to l-canavanine was abolished even at the highest concentrations examined. Both the Gr8a and Gr66a mutants displayed the same strong preference for 5 mM sucrose over 1 mM sucrose, even when the 5 mM sucrose was laced with 30 mM l-canavanine. Thus, in contrast to other aversive tastants, which inhibit the sucrose response thereby causing a reduced preference for 5 mM sucrose (Meunier et al., 2003), l-canavanine did not appear to inhibit the sugar-response. In further support of the findings that Gr8a and Gr66a were required for sensing l-canavanine, we found that the electrophysiological responses to l-canavanine were greatly reduced in the mutant flies (≥7-10 fold). Confirmation that Gr8a and Gr66a were essential for l-canavanine repulsion was that we rescued the mutant phenotypes by introduction of wild-type transgenes.

In contrast to the requirements for Gr8a and Gr66a, the GPCR (DmXR) encoded by mtt was dispensable for the repellent action of l-canavanine, since the behavior and electrophysiological responses of the mtt mutant were indistinguishable from control flies. This finding was not due to contamination of the mtt^f06268 stock with wild-type flies, since we confirmed the presence of the mtt mutation by PCR. The lack of requirement for DmXR described here was not due to differences in behavioral assays, since DmXR was also dispensable when we employed the two-way choice test used previously to assay mtt^f06268 mutant flies. Furthermore, we found that l-canavanine avoidance was not blocked with NMA, which has been reported to be a DmXR antagonist. Finally, we performed tip recordings and found that l-canavanine-induced action potentials depended on GR8a and GR66a and not DmXR. Thus, while DmXR appears to be activated in vitro by l-canavanine, we conclude that GRs are the critical receptors for avoiding this plant derived insecticide in vivo.

Gr8a and Gr66a are co-required in S-type sensilla

While Gr8a and Gr66a were both essential for avoiding l-canavanine, these two Grs displayed different expression patterns. Gr66a is widely expressed in most sensilla that respond to aversive compounds, which include the I-type sensilla, such as I7 and I8, which do not respond to l-canavanine (Hiroi et al., 2002; Thorne et al., 2004; Wang et al., 2004; Marella et al., 2006; Moon et al., 2006; Weiss et al., 2011). However, there was a strong correlation between sensilla that expressed Gr8a and displayed l-canavanine induced activity. Sensilla with undetectable expression of the Gr8a reporter were unresponsive to l-canavanine (L-type, I-type, S2, S4 and S8), while the S-type sensilla with relatively high expression displayed l-canavanine-evoked action potentials. We were not able to record from the one I-type sensillum that expressed Gr8a (I1), due to poor accessibility of this sensillum. Gr66a and Gr8a were necessary for sensing l-canavanine in Gr8a-expressing cells since the loss of the l- canavanine response in the mutant flies was reversed by expression of the wild type transgenes under control of the Gr8a promoter.

The requirements for GR8a and GR66a were also distinct. GR8a functioned narrowly in sensing l-canavanine. In contrast, and consistent with the wider expression pattern of Gr66a reporter, this latter GR is necessary for the repulsion to most (six out of nine) noxious tastants examined (Moon et al., 2006; Lee et al., 2010). The other available Gr mutations did not affect l-canavanine repulsion. We suggest that GR8a and GR66a may form a multi-subunit receptor that functions in a subset of S-type sensilla for l-canavanine repulsion. GR8a may contribute to the specificity for sensing l-canavanine, and GR66a may be a co-receptor for multiple GRs. However, GR8a and GR66a were not sufficient for responding to l-canavanine since co-expression of the two GRs in tissue culture cells, or ectopic co-expression of Gr8a and Gr66a in sugar-responsive GRNs, or Gr8a in Gr66a-expressing GRNs that did not express Gr8a (e.g. I7 or I8) did not confer l-canavanine sensitivity (data not shown).

Aversive GRs comprised of subunits with narrow and broad specificities

The widely differing specificities for the two GRs that act in sensing l-canavanine is reminiscent of our previous findings concerning the repertoire of GRs that function in caffeine repulsion. GR93a is required for avoidance of caffeine only (Lee et al., 2009), while mutations disrupting two other GRs that impair caffeine aversion (GR33a and GR66a), disrupt the repulsion to most deterrent tastants (Moon et al., 2006; Lee et al., 2009; Moon et al., 2009; Lee et al., 2010). Thus, while a single GR appears to be sufficient for the respond to fructose (Sato et al., 2011), we propose that the GRs that participate in sensing noxious compounds are complexes comprised of at least one subunit with a narrow specificity and others that are broadly required co-receptors.