Table 1.
Taste stimuli, concentrations, and abbreviations.
Figure 1.
Photomicrographs illustrating assessment of oral stimulation field and histological analysis of recording electrode location.
The top panel shows images of coronal sections (40 μm) through (A) the posterior circumvallate (CV) region and (B) anterior portion of two mouse tongues. Tongues were immediately removed and sectioned following oral delivery of fluorescent dye through our taste presentation system. Fluorescent dye, red under our filter settings, covered the anterior and posterior tongue surface and invaded the posterior tongue CV crypts housing taste receptors. Inset in A is a cross-section through the CV region of a tongue that was not stimulated with dye and shows the tongue does not naturally fluoresce. Inset scale bar is 100 μm. Photomicrographs were adjusted in Adobe Photoshop CS4 software (version 11.0.2; Adobe Systems, Inc., San Jose, CA) using levels, brightness, and contrast. (C) Left, image of a coronal section (40 μm) through mouse brain stem showing an electrolytic lesion made at a recording location (arrow). Schematic on the right (adapted from [37]; with publisher's permission) shows the location of the NTS relative to select landmarks, including the spinal vestibular nucleus (SpVe) and spinal trigeminal tract (sp5).
Figure 2.
Definition of neural clusters.
Groupings of NTS neurons in C3 (A) and C3.SW (B) mice defined by hierarchical clustering of activity to stimuli representative of different taste categories. Y-bars represent mean ± s.e.m. responses (net spikes in 5 s). Dendrograms showing cluster recovery in each line are depicted by insets near the top of each panel. Numbers along the abscissae denote stimuli (legend). In the legend, numbers in stimulus abbreviations indicate concentrations from lowest (e.g., 1) to highest (e.g., 5), where applicable (Table 1).
Figure 3.
Raw response data from taste-sensitive neurons.
Digital oscilloscope sweeps showing electrophysiological activity to all stimuli recorded from two C3.SW cells (A and B) and two C3 neurons (C and D). The C3 neurons were recorded in series from one mouse; C3.SW cells are from different mice. The stimulus tested during each sweep is abbreviated (Table 1) along the left margin. Where applicable, numbers in stimulus abbreviations indicate concentrations from lowest (e.g., 1) to highest (e.g., 5), as in Table 1. Upward and downward arrows at the bottom of each sweep stack indicate stimulus onset and offset, respectively.
Figure 4.
Neural responses to bitter and other stimuli.
Heatmaps showing the net 5 s response to each of 26 taste stimuli (abscissae) across all 36 C3 (A) and 43 C3.SW (B) neurons (ordinates). The heat scale in panel A gives response spike density for panels A and B. Neurons are sorted within mouse line by cluster analysis of activity to all concentrations of quinine, denatonium, cycloheximide, sucrose octaacetate, and propylthiouracil. Pairs of arrowheads of the same color along the base of each dendrogram highlight neurons that were recorded from one mouse and showed differential sensitivity to bitter stimuli (e.g., cells marked by green arrowheads were recorded from one mouse; pair in black from another, etc.). Orange arrowheads along the base of the dendrogram in panel B denote response data from five neurons recorded in series from one C3.SW mouse. Numbers on dendrograms mark neural clusters determined by “scree” plots. Table 1 gives stimulus abbreviations. Numbers above abbreviations for bitter stimuli indicate concentrations from lowest (e.g., 1) to highest (e.g., 5), as in Table 1. Plots of average activity in each cluster are given below dendrograms; numbers color-matched to each plot indicate cluster(s).
Figure 5.
Clustering of taste responses to bitter and other stimuli.
Three-dimensional plots showing the outcome of multidimensional scaling of net responses to all taste stimuli across 36 C3 (A) and 43 C3.SW (B) neurons. Table 1 gives stimulus abbreviations used in each space. Responses to cycloheximide, quinine, denatonium, and SOA are color-coded (legend in panel A), and responses to increasing concentrations of these stimuli (Table 1) are respectively represented by points/circles of increasing diameter. Dimensions of plots represent arbitrary units.
Figure 6.
Modeling time dependencies in bitter coding by C3 neurons.
(A) Plots showing sequential, 500 ms wide windows of taste activity (spike density per half-second, ordinates) across 36 C3 cells (abscissae) to the highest concentrations of quinine, denatonium, cycloheximide, and sucrose octaacetate. The time window of taste activity captured by each plot is indicated. Legend in C gives the stimulus associated with each colored response for all panels in this figure. (B) Three-dimensional plot showing the outcome of principal components (PC) analysis applied to sequential, 500 ms wide windows of activity (cf. panel A) across 36 C3 neurons during taste stimulation with all concentrations of quinine, denatonium, cycloheximide, sucrose octaacetate, and also water. Response windows from stimulus onset to offset (i.e., 0 to 5 s post stimulus) are represented. For each stimulus, PC-mapped points for sequential response windows are connected using color-coded lines, forming “paths” in the space describing time-evolved neural activity to bitter inputs. “Elbows” along each path represent points for response windows. Arrowheads indicate flow and sequencing of contiguous windows. Along each path, the point representing time-sliced activity arising 1 to 1.5 s post stimulus onset is marked by a square. Paths for activity to all low, intermediate and high concentrations (legend, concentrations as in Table 1) of each stimulus are shown; responses to intermediate concentrations are not differentiated. PC1, PC2, and PC3 explain 78% of the total response variance. The general locales of the “start” and “end” points for the trajectories are indicated. (C) Same as panel B, except that activity within each 500 ms response window was averaged over concentrations for each stimulus prior to PC analysis, highlighting global trends in the data. PC1, PC2, and PC3 explain 86% of the total response variance.
Figure 7.
Modeling time dependencies in bitter coding by C3.SW neurons.
(A) Plots showing sequential, 500 ms wide windows of taste activity (spike density per half-second, ordinates) across 43 C3.SW cells (abscissae) to the highest concentrations of quinine, denatonium, cycloheximide, and sucrose octaacetate. The time window of taste activity captured by each plot is indicated. Legend in B gives the stimulus associated with each colored response for all panels in this figure. (B) Three-dimensional plot showing the outcome of principal components (PC) analysis applied to sequential, 500 ms wide windows of activity across 43 C3.SW neurons during taste stimulation with all concentrations of quinine, denatonium, cycloheximide, sucrose octaacetate, and also water. Response windows from stimulus onset to offset (i.e., 0 to 5 s post stimulus) are represented. For each stimulus, PC-mapped points for sequential response windows are connected using color-coded lines, as in Figure 6. Arrows indicate flow of contiguous points/response windows; squares mark points for response windows residing 1 to 1.5 s post stimulus onset. Response “paths” for activity to all low, intermediate and high concentrations (legend and Table 1) of each stimulus are shown; responses to intermediate concentrations are not differentiated. PC1, PC2, and PC3 explain 76% of the total response variance. The general locales of the “start” and “end” points for the trajectories are indicated. (C) Same as panel B, except that activity within each 500 ms window was averaged over concentrations for each stimulus prior to PC analysis, highlighting global trends in the data. PC1, PC2, and PC3 explain 83% of the total response variance.
Figure 8.
Modeling time dependencies in neural coding for bitter and other stimuli.
Three-dimensional plots showing the outcome of principal components (PC) analysis applied to sequential, 500 ms wide windows of activity to bitter tastants, Na+ salts, acidic stimuli, and also water across 36 C3 (A) and 43 C3.SW (B) neurons. Half-second wide response windows from stimulus onset to offset (i.e., 0 to 5 s post stimulus) are represented. For each stimulus, PC-mapped points for sequential response epochs are connected using color-coded lines (legend, Table 1 gives abbreviations), forming “paths” in the space describing time-evolved neural activity to taste inputs. “Elbows” along each path represent points for response windows. Arrowheads indicate flow and sequencing of contiguous windows. The general locale of the “start” for each trajectory in PC space is indicated. Along each path, the point representing time-windowed activity arising 1 to 1.5 s post stimulus onset is marked by a square. Legend in A applies to both panels.
Figure 9.
Similarity and dissimilarity among bitter responses through time.
Correlations (Pearson's r, ordinates) among responses to Na+ salts, acids, and bitter stimuli measured during sequential, half-second wide periods of taste responses (abscissae) for 36 C3 (A) and 43 C3.SW (B) neurons. Legend in B gives the stimulus comparison denoted by each trace and applies to both panels. For bitter stimuli, activity to only the highest concentrations (Table 1) is represented. Analyses/plots involving sucrose octaacetate are not shown for neurons recorded from C3 mice, which are insensitive to this stimulus. Solid black line gives the significance criterion for r as based on the number of cells.