Article Figures & Data
Figures
- Figure 1.
Monitoring the responses of GRNs to timed pulses of tastant. a, Scanning electron micrographs of the proboscis. Left, Distal third (a portion of which is shown here) contains types of sensilla associated with chemoreception and mechanoreception. Insets (clockwise from top), Examples of sunken basiconic, styloconic, chaeticonic, and spire-shaped basiconic sensilla. Right, Middle and proximal portions of the proboscis contain primarily mechanoreceptive sensilla. Insets, Examples of chaeticonic sensilla (Faucheux, 2013). Scale bar, 50 μm. b, A schematic of the apparatus. Dye-colored tastants were pressure injected into a water stream passing over the proboscis. The instantaneous concentration of tastant was monitored by a color sensor. Top inset, Example spiking responses from GRNs recorded by tetrode. Calibrations: vertical = 40× channel-wise median absolute deviation, horizontal = 1 s. Bottom inset, Example color sensor trace showing tastant delivery (upward deflection, sucrose). Time axis as above. c, Frontal view of the moth brain. Magenta represents left maxillary nerve, filled with rhodamine-dextran. Scale bar, 100 μm. All recordings of GRNs were made from the exclusively afferent maxillary nerve which extends processes throughout the ipsilateral SEZ. AL, Antennal lobe; E, esophagus .
- Figure 2.
Tastant-elicited responses of GRNs are diverse and specific. a, Raster plots represent responses of all 83 tested neurons (rows) to 12 tastants (columns, 4 trials/tastant shown, of 4 trials for water, 7 trials of all other tastants), ordered approximately by response characteristics. Yellow shading represents tastant presentation, 1 s. For examples of spike sorting, see Figure 6. b–d, Summary of GRN responses shown in a. All plots represent binary response classifications for a broad range of response-detection thresholds (1–12; see Materials and Methods). b, Percentage of GRNs responding to different numbers of tastants. c, Percentage of GRNs responding to each tastant. d, Percentage of GRNs responding exclusively to each tastant.
- Figure 3.
Temporal response properties of GRNs. a, Histograms (trial-averaged spiking times smoothed with a Gaussian filter, SD 60 ms). Timing of responses of GRNs varied with the tastant (color code at right). Bottom, red trace, Trial-averaged color sensor voltage indicates stimulus timing. b, Histograms, Seven GRNs responded with different timing to the same stimulus (sucrose, 1000 mM). Scale bars, 10 Hz. GRN numbers refer to Figure 2a.
- Figure 4.
Quantifying information content of the spatiotemporal responses of GRNs. a, Classification of tastant identity using the GRN population. Blue represents cross-validated success of 12-way classification of tastants using variable vector lengths of GRN responses (n = 83 neurons), beginning at the time of tastant delivery. Pink represents control, as blue, but including data taken before (2 s) tastant delivery. b, Hierarchical clustering of the GRN population activity shows that their responses do not cluster by basic taste category. Distance was normalized to show the multiple of the average intertrial distance across tastants. Colored squares represent traditional basic taste categories from which different tastants were drawn (n = 83, see Materials and Methods).
- Figure 5.
Increasing tastant concentration leads to stronger GRN responses. a, Raster plots of three simultaneously recorded GRNs responding to pulses of water, 6 concentrations of sucrose, and NaCl. Top, All extracellular recordings (6 wires) for each GRN together show the waveforms are well separated. Calibration: vertical = 20 SD above noise level, horizontal = 1 s. b, Response magnitude of a population of GRNs increases with the concentrations of sucrose delivered in a. Responses could be excitatory or inhibitory (see Materials and Methods). c, Strength of GRN responses to two concentrations of sucrose: NaCl, and caffeine. Black lines connect responses of single GRNs. Blue bars represent mean response strength across GRNs. For comparison, the response strengths are scaled so the strongest response for every pair of concentrations is equated. For all tastants, the higher concentration elicited stronger responses (1-tailed Wilcoxon signed rank tests; sucrose: n = 75, z = 5.77 p = 3.86e-9; NaCl: n = 51, z = 5.77, p = 0.03; caffeine: n = 44, z = 1.77, p = 0.04.).
- Figure 6.
SONs transform the representations of tastants arising in GRNs. a, Example STA EPSP provides physiological verification that a GRN-SON pair is monosynaptically connected. Calibration: vertical = 0.5 mV, horizontal = 25 ms. Dashed line indicates GRN spike time (2.13 ms delay before EPSP onset). b, Frontal view of the moth brain showing a single neurobiotin-filled SON in the anterior SEZ. c, Raster plots represent responses of an example simultaneously recorded, monosynaptically connected GRN and SON pair to different tastants. Bottom left, red trace, Average color sensor voltage. Gray shading represents tastant delivery. Bottom right, blue, Voltage record of the SON, last trial of the raster plot. The STA shown in a was generated from these two neurons. Calibration: vertical = 20 mV, horizontal = 1 s. d–f, Summary of SON responses (n = 13, nonpaired recordings). d, Percentage of SONs responding to different numbers of tastants. e, Percentage of SONs responding to each tastant. f, Percentage of SONs responding exclusively to each tastant.
- Figure 7.
Spatiotemporal response properties of SONs. a, Raster plots of 8 example SONs responding with temporally structured firing patterns to various tastants. Blue and red traces and gray shading as in Figure 6c. Calibration: vertical = 10 mV, horizontal = 1 s. b, Histogram represents the average spiking response of GRNs (green, n = 83) and SONs (red, n = 13) elicited by 1 s delivery of the 7 tastants shown in a. Spiking responses were smoothed with a Gaussian filter (SD 60 ms) and normalized to show equal maximum average spike rates. Red trace represents average color sensor voltage. Calibration: 1 s. c, Given the same tastant set, SONs provided faster and more accurate tastant classification with fewer cells than GRNs. Comparison of 7-way classification using the GRN (n = 83, green) and SON responses (n = 13, red). Plots represent mean ± SEM. d, Schematic summary of our findings. Various tastant chemicals in the environment (24 examples are illustrated) are detected in the proboscis by GRNs that have overlapping sensitivities that range from narrow (GRN c) to broad (GRN d) (see Fig. 2a). Multiple GRNs converge upon each SON in the SEZ (see Fig. 6). We hypothesize that excitatory and inhibitory SONs (dotted arrows) may interact with each other, leading to elaborately structured firing patterns (a).
- Figure 8.
Proboscis extension is elicited by specific tastants. Probability of eliciting proboscis extension varied significantly for individual tastants (n = 95 moths/tastant). Error bars indicate SEM. Letters indicate initials of tastants eliciting significantly different response probabilities (p < 0.05, omnibus χ2 test followed by individual Bonferroni-corrected χ2 comparisons). Inset, Proboscis extension. Moths restrained in tubes were tested with metered drops of tastant onto the proboscis (see Materials and Methods).
