e., the P.I. approached 0. For some bitter tastants (e.g., azadirachtin [AZA] and umbelliferone [UMB]), testing was limited by the low solubility of the tastant, but near-maximal avoidance was observed at the highest concentrations available. Some bitter compounds were more aversive than

others (Figures 2B and 2C). To quantify the sensitivity of the fly to each compound we calculated the concentration of bitter tastant that is required to render 5 mM sucrose equally attractive, or “isoattractive,” to 1 mM sucrose. We defined the isoattractive concentration as the concentration at which the P.I. is 0.36, which is the arithmetic mean of the control P.I. (0.71) and the minimal P.I. (0). Thus the isoattractive concentration for denatonium benzoate (DEN), illustrated in Figure 2B, Selleckchem LBH589 lies between 10−4.5 M and 10−5

M. Among our panel of tastants, DEN elicits the strongest avoidance (Figure 2C). Interestingly, DEN has also been identified as the tastant that is perceived as most bitter by humans in psychophysical studies PR-171 cost (Hansen et al., 1993 and Keast et al., 2003). The isoattractive concentrations of our bitter panel ranged over more than two orders of magnitude, with the weakest avoidance elicited by escin (ESC) (Figure 2C). These results confirmed that all members of the tastant panel are aversive or bitter to Drosophila ( Figure S1). The results also identified a concentration range over which each bitter compound is behaviorally active in this paradigm. Together these results established a foundation for a detailed physiological analysis of the cellular basis of bitter coding. As a first step toward understanding the coding of bitter stimuli, we systematically examined the electrophysiological responses (Hodgson et al., 1955) elicited by all 14 bitter substances from all 31 labellar taste sensilla (Figure 1A).

These tastants were tested at 1 mM or 10 mM, or 1% in one case, concentrations at which they were active in our behavioral paradigm. We also tested two additional compounds, aristolochic acid (ARI) and gossypol (GOS), described as bitter in other insect species, yielding a total of 16 × 31 = 496 sensillum-tastant combinations, each tested n ≥ 10 times. All 16 compounds elicited action potentials from at least some sensilla. The action potentials were of a large amplitude characteristic of the bitter neuron (Figure 1B). In a few cases those we observed a small number of additional action potentials of smaller amplitude, presumably generated by the water neuron, particularly in the initial period of the recording (e.g., see ARI trace in Figure 1B). Three of the 31 sensilla, S3, S5, and S9, generated a second, high-frequency and low-amplitude spike train of unknown source that appeared to be independent of stimulus identity and concentration (Figure 1C). However, in all cases the large-amplitude action potentials of the bitter neuron could easily be distinguished and are the basis of the analysis that follows.