INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The rat ventroposteromedial nucleus pars parvicellularis (VPMpc) is an obligatory gustatory relay between the parabrachial nucleus (PBN) and the gustatory cortex (Cx). It is located at the medial tip of the VPM nucleus of the thalamus, centered approximately 3.6 mm posterior to bregma, 1.0 mm lateral to the midline, and 6.5 mm ventral to the surface of the dura (Paxinos and Watson 1986).

The gustatory code at the level of VPMpc in rat is only poorly defined. Exploratory studies that date from the 1950s and 1960s used a limited stimulus array and multiunit recordings that revealed little about coding (Emmers et al. 1960, 1962; Frommer 1961; Norgren 1970; Pfaffmann et al. 1959). These early experiments generally reported an increase in integrated multiunit responses to NaCl and cold stimuli on the tongue with less activity evoked by the other three basic tastes. Warm water decreased multiunit integrated responses. Data indicated that most medially in the VPM responses were evoked only by sapid stimuli, whereas progressively more lateral penetrations revealed oral temperature and touch-elicited responses. It was also reported that the receptive fields of neurons from the anterior part of the VPMpc were served bilaterally by the chorda tympani nerve (CT), whereas those of posterior cells were innervated bilaterally by the glossopharyngeal nerve (N. IX) (Emmers et al. 1962).

In the past three decades, four single-unit studies on rat VPMpc responses have been published (Nomura and Ogawa 1985; Ogawa and Nomura 1988; Scott and Erickson 1971; Scott and Yalowitz 1978), allowing for some insight into the neural code at this relay. The most recent of these (Nomura and Ogawa 1985; Ogawa and Nomura 1988) focused on the receptive fields of both taste and mechanosensitive neurons rather than on gustatory coding.

Studies reported in the 1970s (Scott and Erickson 1971; Scott and Yalowitz 1978) were based on stimulation of only the anterior tongue, which was isolated by a rubber dam with its attended risk of interrupted circulation. Evoked responses across an extended array of taste stimuli were compressed into a small range, and concentration-response functions were nearly flat.

Thus information on gustatory coding in the VPMpc lags that from lower-order gustatory relays. The present study was designed to address this by exploring responses of single neurons in the VPMpc of rat while stimulating the entire oral cavity and using an extensive array of gustatory, thermal, and tactile stimuli in an acute preparation of the rat.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The subjects were 13 female and 21 male Wistar rats weighing 485 ± 107 (SD) g. They were maintained on pellet food and tap water, which were available ad libitum in their home cages. Room lights were on from 0800 to 2200 (14 h).

Surgery

Each rat was anesthetized to a surgical level with urethan (1.2 g/kg ip) and -chloralose (20 mg/kg ip). The depth of anesthesia was monitored by frequent testing for the presence of leg flexion reflexes, which if found, warranted a supplemental dose of urethan (usually 0.1 g/kg every 1.5 h). Lidocaine was applied to all wound margins. Body temperature was maintained at 36-38°C, and heart rate was continuously monitored. All surgery was performed using aseptic techniques. The trachea was cannulated, and an esophageal fistula was inserted to prevent postingestive effects of oral stimulation. The head was mounted in a nontraumatic headholder (Erickson 1966), and the skull was exposed. It was leveled by taking measurements 4 mm on either side of the midsagittal suture (M-L) and at bregma and lambda (A-P). A square region of the skull (4 × 4 mm) was removed and centered on the following coordinates: 3.6 mm posterior to bregma and 1.0 mm lateral to midline. The underlying dura mater was removed. Sterile saline was used to keep the exposed cortical surface moist.

Electrophysiological recording

Borosilicate micropipettes, filled with 2.5 M potassium citrate, were used to isolate the activity of individual cells in the VPMpc. The inner diameter of the electrode tips was about 1 µm and impedances were 1-4 M at 1 kHz. Conventional electrophysiological recording techniques were used for differential amplification, display, and recording of the neural signal. Action potentials of a single cell were identified by consistency of amplitude and waveform and an interspike interval of at least 1.5 ms. Neural data, voice commentary, and onset marker signal for stimulus delivery were stored on a four-channel TEAC tape recorder and analyzed off-line.

Stimuli and stimulus delivery

A total of 22 stimuli was used. These comprised 16 room-temperature (23°C) sapid stimuli, plus distilled water and 0.1 M NaCl, each at 0, 23, and 37°C. These chemicals, their concentrations, abbreviations used to identify them in this paper, their molecular weights, and chemical groups are listed in Table 1. All solutions were prepared with distilled water, except that 20% tap water was added to nonionic solutions to activate the stimulus onset marker.

Five milliliters of solution was delivered at a rate of 2 ml/s by way of perforated PE-20 tubing that was inserted into the rat's open mouth. The spray was released from perforations along the final 2 cm of the tube and contacted nearly the entire receptor surface, including nasoincisor ducts. The moment of stimulus contact with the tongue was marked by a TTL logic device that passed 11-nA current through the rat (Chang and Scott 1984), an amount two orders of magnitude below the threshold for electric taste (Bujas 1971). Each stimulus was followed by a 20-ml DH₂O rinse and by a minimum rest period of 30 s. Additional rinses and rest periods were occasionally required for baseline activity levels to be reestablished. Stimuli were presented in quasi-random order with the stipulation that chemicals representing similar taste qualities, or stimuli having temperatures different from room temperature, not be applied consecutively. Tactile stimulation was represented by the application of 23°C DH₂O.

Recording

Initial coordinates to locate the VPMpc were 3.6 mm posterior to bregma and 1.0 mm lateral to the midline. During exploration of the VPMpc, the electrode was advanced in 30-µm steps from a depth of 6.0-7.2 mm from the cortical surface. The functional area of the VPMpc was defined as the distance from the first to the last taste cell in each penetration.

Responses of taste cells were recorded as mean impulses per second over a 5-s poststimulus period. This rate was compared with the spontaneous activity distribution of the neuron. An evoked response was defined by a change of at least 1.65 SD (P = 0.05) from mean spontaneous rate, sustained over 5 s. Sensitivity to thermal and tactile stimuli was determined by the same criterion. For a neuron to be classified as gustatory, it had to give a defined response to at least three sapid stimuli at room temperature.

Data analysis

Action potentials were counted using a window discriminator for off-line analysis. A personal computer was used to perform basic statistical analyses. Spontaneous rate was measured for 3 s before stimulus delivery. The minimum number of readings taken on any cell to determine its spontaneous activity distribution was 17; the mean was 22. The net evoked firing rate (gross evoked minus spontaneous rate) during 5 s provided the basis for the derived analyses, which included breadth-of-tuning measures, calculations of interneuronal and interstimulus Pearson product-moment correlation coefficients, multidimensional scaling, ANOVAs, and cluster analyses using the program Systat 7.0 and SPSS 9.0. t-tests were performed using Microsoft Excel 7.0. A criterion of P < 0.05 was considered statistically significant and was Bonferroni corrected where appropriate. Means are presented with their associated SEs unless indicated otherwise.

Data from male and female rats were analyzed separately. The only significant difference was that males had higher spontaneous activity (P = 0.008). This was deemed a minor difference, and so the data from both sexes were treated together in the remaining analyses.

Histology

The locations of the recorded taste cells were estimated by bringing the tip of a 5-µm pipette containing pontamine sky blue to the same coordinates as those of the final taste cell in each rat. The dye was iontophoretically applied (10 µA 10 min, 10 s ON-OFF). The rat was than killed by administration of 250 mg urethan followed by decapitation, and its brain was frozen at 40°C in iso-pentane. Coronal sections (20 µm) were obtained and alternately stained with cresyl violet or left unstained. Brain sections of four rats were stained for acetylcholinesterase (Karnovsky and Roots 1964), with the modification that all compounds were dissolved in 0.013 M potassium phosphate monobasic-sodium hydroxide buffer in place of Tris maleate buffer (final pH = 6.0-6.5).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Histology

Figure 1 shows an example of an acetylcholinesterase stained coronal section and dye spot in the dorsolateral area of the VPMpc (cells 23-25 were recorded from this animal).

View larger version (109K): [in this window] [in a new window]   Fig. 1. Coronal acetylcholinesterase stained slide at the level of the ventroposteromedial nucleus (VPMpc). , iontophoretically deposited dye at location where a gustatory neuron was found.

Sensory modalities

The activity of 151 single neurons in VPMpc was isolated. Of these, 54 (36%) gave no response to any of the three modalities tested (Fig. 2). Thirty-five cells (23%) were responsive to only one of the modalities: 14 (9%) to taste, 12 (8%) to temperature, and 9 (6%) to touch in the oral cavity. The remaining 62 neurons (41%) were multimodal. Fifty (33%) responded to both taste and temperature (primarily cold). Three others (2%) responded to temperature and touch, whereas none responded only to taste and touch. Finally, nine cells (6%) gave responses to all three modalities. Touch responses were nearly equally divided between excitatory (55%) and inhibitory (45%). Temperature responses were more skewed toward excitation: of the 19 responses to 37°C water, 15 (79%) were excitatory; of the 55 responses to 0°C water, 53 (96%) were excitatory. The mean spontaneous rate of temperature-sensitive cells was 6.9 ± 6.4 spikes/s; that of touch cells was 4.7 ± 6.6 spikes/s.

View larger version (31K): [in this window] [in a new window]   Fig. 2. Distribution of sensory modalities of neurons in the VPMpc. G, gustatory; T, temperature sensitive; S , somatosensory; N, no response.

The mean coordinates at which the taste cells were found were as follows: 3.56 ± 0.18 (SD) mm posterior to bregma, 1.03 ± 0.10 mm M-L from the midline suture, and 6.64 ± 0.28 mm D-V. More lateral cells were found to be more dorsal. The regression line across the nucleus was 0.72 × M-L coordinate +5.9 mm (P = 0.03, R² = 0.065). No relationship was found between the location of the cells and the body weights of the rats, nor between the location of a taste cell and its sensitivity profile.

Response characteristics of taste cells

The mean spontaneous activity level of taste cells in this study was 6.3 ± 0.7 spikes/s (range 0.0-29.9; n = 73). Of the 17 × 73 = 1,241 stimulus-neuron interactions, 803 (65%) generated excitatory responses and 23 (2%) inhibitory. Fourteen of these inhibitory responses were associated with two cells that responded only with inhibition. When a neuron responded to only a small subset of the stimulus array, the subset was always associated with a particular taste quality, implying a nonrandom profile.

Mean evoked discharge rates across all taste cells are depicted in Fig. 3. Responses evoked by the basic tastants were 13.8 ± 1.6 spikes/s for 0.1 M NaCl (N), 9.3 ± 1.4 spikes/s for 0.01 M HCl (H), 5.1 ± 0.9 spikes/s for 0.5 M sucrose (S), and 4.3 ± 0.6 spikes/s for 0.01 M quinine HCl (Q). Cold, room-temperature, and warm water evoked mean response rates of 9.9 ± 1.5, 0.6 ± 0.4, and 1.3 ± 0.9 spikes/s, respectively. The exponent of the power function that defined the NaCl concentration-response curve was 0.39 (significance of curve fit P = 0.02, R² = 0.961). Of the 61 cells that were tested for cold and warm NaCl, the mean firing rate evoked by 0°C NaCl was 17.0 ± 2.8 spikes/s and by 37°C NaCl was 15.0 ± 2.4 spikes/s.

View larger version (11K): [in this window] [in a new window]   Fig. 3. Profile of the mean net evoked responses across all 73 taste cells to the 17 gustatory stimuli that were delivered at 23°C.

Thalamic taste cells were broadly tuned. The breadth-of-tuning coefficient is the standard metric for evaluating the range of a neuron's response across the four prototypical taste stimuli (Smith and Travers 1979). The proportional distribution of a cell's response across these stimuli is converted to a coefficient that may range from 0.00 (total specificity to one of the stimuli) to 1.00 (equal responsiveness to all four). The mean coefficient for these 73 gustatory cells was 0.79 ± 0.02 (range: 0.04-1.00).

Gustatory neuron types

Taste cells may be classified based simply on their "best" responses. Thirty-seven neurons (51%) responded best to NaCl among the four basic stimuli, 19 (26%) to HCl, 12 (16%) to sucrose, and 5 (7%) to quinine HCl. However, a more comprehensive categorization derives from an analysis of the similarity of their response profiles across all four basic stimuli. We generated a similarity matrix based on correlation coefficients between each pair of activity profiles, then subjected the coefficients of this matrix to a cluster analysis. The results appear as a dendrogram in Fig. 4. Cells were numbered in the order in which they were isolated. Distances are defined as 1 minus the correlation coefficient. Pairs of cells were interconnected at the level of correlation between their response profiles; groups were fused at the mean level between their constituent members in an iterative process that was complete when all neurons in the sample were interconnected.

View larger version (20K): [in this window] [in a new window]   Fig. 4. A dendrogram in which is indicated the degree of similarity among the response profiles of the 73 taste-responsive neurons of this study. The basic stimulus that evoked the largest response from each cell is indicated on the left of the dendrogram, while any other basic stimuli that elicited 80% of this maximum is indicated next to it. Numbers indicate the order of recording of the neurons.

Cells are labeled according to their best stimulus (first), followed by any stimulus that evoked a response 80% of the best response. Three main groups were defined and subsequently verified (cf. following paragraph) as being statistically independent: a sodium-oriented group (n = 40) from cells 67 to 22; an acid-oriented group (n = 12) from cells 46 to 52; and a sugar-oriented group (n = 17) from cells 12 to 44. There were also four quinine-oriented neurons (58 to 8) that were not sufficiently similar to be identified as a statistically separate group. Note that neurons 8 and 37 were inhibitory cells, explaining their apparently anomalous placement among the quinine quartet and the sugar group, respectively (the cluster analysis was based in their absolute firing rates). The salt-oriented group was fully intercorrelated at a level of 0.68, the acid-oriented group at 0.63, the quinine-oriented neurons at 0.55, and the sugar-oriented group at 0.39. Thus the grouping in VPMpc is not as well-defined for neurons most responsive to sugar and quinine as for those more oriented to salt and acid. This reflects the fact that higher response rates lead to more extreme (in this case positive) correlation coefficients.

Responses of these groups to the 17 taste stimuli were analyzed independently to evaluate the significance of their differences. Repeated-measures ANOVA on three groups plus quinine cells indicated a significant interaction between taste and group membership [F(48,168) = 3.00; P < 0.001] and a marginal difference between groups [F(3,69) = 2.62; P = 0.058]. Bonferroni-corrected post hoc t-tests (k = 24, adjusted  = 0.002) suggest that the three groups were significantly different from one another by the responses to their defining stimulus (comparisons only on the 4 prototypical stimuli) (cf. McCaughey and Scott 2000 for a more comprehensive description of the division of taste cells into subgroups). The acid and sodium groups, for example, gave significantly different responses to HCl (P < 0.001). The sodium group responded significantly differently from the sugar group to NaCl (P = 0.002) and to sucrose (P < 0.001). The acid and sugar cells differed significantly in their responses to sucrose (P = 0.001) and HCl (P < 0.001). The difference between responses to the prototypical tastants of the quinine cells (n = 4) and those of the three groups did not reach significance.

Figure 5 shows the mean response profiles of cells in each of the three defined groups. The NaCl-oriented group was most sensitive to NaCl and least to sucrose. The acid-oriented group had a relatively high response to 0.3 M NaCl, HCl and citric acid and to bitter salts. The sucrose-oriented cells tended to show smaller responses, especially to the acids, quinine and bitter salts. There were no significant differences in breadths of tuning among the three groups.

View larger version (19K): [in this window] [in a new window]   Fig. 5. Profile of the mean net evoked responses of each taste neuron group across all 17 gustatory stimuli and water.

We found no evidence for any differences among the A-P, M-L, and D-V coordinates of cells of the different groups. Hence we conclude that there is no apparent chemotopy at the level of the VPMpc.

Next each group was analyzed for the possible existence of subgroups. Separate analyses of variance indicated that only the NaCl-oriented cells could be statistically subdivided. Subgroup one (n = 5, from cells 67 to 57, Fig. 4) had a significantly lower response rate than the other two subgroups to sucrose (P = 0.007 and P = 0.005). Subgroup two (n = 21, from cells 27 to 51the core of the N group) had a significantly lower response rate than subgroup three (n = 14, from cells 53 to 22) to HCl (P = 0.003). The breadth of tuning of subgroup two (0.72 ± 0.04) was nearly lower than that of subgroup three (0.81 ± 0.03, P = 0.06). NaCl specificity was calculated by determining the fraction of the sum of responses evoked by the four basic stimuli that was due to 0.1 M NaCl. The sodium specificity of subgroup two (63 ± 3%) indeed was higher than that of subgroup three (47 ± 4%; P = 0.002) but not significantly different from subgroup one (52 ± 8%; P = 0.09). Therefore the sodium group may be seen as being composed of a relatively salt-specific subgroup of 21 neurons, plus a subgroup that also responded well to HCl (n = 14) and one that is also oriented to quinine (n = 5).

Taste qualities: across-neurons analysis

A three-dimensional representation (R² = 0.97) of the relative similarity among the profiles generated by all sapid stimuli appears in Fig. 6. The sodium and lithium salts (left), the acids and bitter salts (right), and sucrose-fructose (low) occupy three distinct regions of the space. Glucose is separated from sucrose and fructose in the x dimension. The correlation between sucrose and fructose was 0.65, whereas that between sucrose and glucose was 0.47 and between fructose and glucose only 0.37. Overall there is a stratification of stimuli on the z axis, with sugars low, Na-Li salts intermediate but rising with concentration, and acids and bitter salts high. This suggests that this dimension represents a hedonic tone, a correlate of stimulus toxicity (Scott and Mark 1987).

View larger version (24K): [in this window] [in a new window]   Fig. 6. Three-dimensional space showing the distribution of taste qualities represented by this stimulus array.

A stimulus dendrogram, based on hierarchical cluster analysis of the relative similarity among the profiles generated by all stimuli, appears in Fig. 7. The seven sodium and lithium tastants are grouped together (r = 0.66), the six acids, quinine, and bitter-salts form another group (r = 0.53), as does sucrose and fructose (r = 0.65) and polycose and glucose (r = 0.70). The polycose-glucose pair clusters with the sodium group (r = 0.46), which new group clusters with the acid-bitter group (r = 0.41), and finally with the sucrose-fructose pair (r = 0.20).

View larger version (18K): [in this window] [in a new window]   Fig. 7. A dendrogram indicating the degree of similarity among the neural profiles elicited by each stimulus.

Temporal analysis

The time course of a response carries information relevant to stimulus quality (DiLorenzo and Schwartzbaum 1982; Scott and Mark 1987). Figure 8 shows PSTHs for responses to the four prototypical tastants based on a 300-ms moving average (the present 100-ms bin plus the one on either side of it) obtained by averaging across all neurons. Because the stimulation of different receptive fields can have an impact on the time course of a response, it is important that the stimulus delivery procedure used here has been shown to contact the primary body of receptors for each prototypical stimulus (Chang and Scott 1984). The responses to 0.1 M NaCl reached a peak around 550 ms with a discharge rate of 3.5 spikes/100 ms and gradually declined to a plateau of about 1.5 spikes/100 ms. Responses to HCl achieved a maximum at 450 ms of 2.2 spikes/100 ms, then gradually declined with continued small fluctuations. Responses to NaCl and HCl were very similar from about 2 s onward. The responses to quinine HCl initially followed those to HCl (peak at 250 ms) but diverged at about 1 s and were lower thereafter. The most gradual onset was in response to sucrose, with a peak response around 1.1 s at 1.5 spikes/100 ms followed by a gradual decline to 1 spike/100 ms at 5 s. The response to quinine HCl appeared very similar to that of sucrose from about 1 s onward. Repeated-measures ANOVA and Bonferroni-protected t-tests (adjusted  = 0.0025) on these data recalculated at five periods of 1-s duration confirmed these pattern similarities. A significant interaction was found between time and tastant [F(12,864) = 11.93; P < 0.001]. There was a difference between responses elicited by NaCl and HCl over the first second (P < 0.001), but no difference thereafter. The time course evoked by sucrose was different from that of HCl for the periods between 0-1 (P < 0.001) and 3-4 s (P < 0.001). Responses to HCl and quinine diverged only after 2 s and remained separate from then on. Only the first second was significantly different between the responses evoked by sucrose and quinine HCl (P = 0.001). Hence despite distinct features of the responses to each of the prototypical stimuli, responses to HCl showed early pattern similarities with quinine HCl and late pattern similarities with NaCl; responses to sucrose and quinine HCl were distinguishable only by their phasic (0-1 s) component. Given the differences among qualities, which place sucrose most distinct from the other three, and HCl and QHCl most similar, it appears that the phasic portion of the response is most critical for defining taste quality.

View larger version (13K): [in this window] [in a new window]   Fig. 8. Poststimulus time histogram of the prototypical tastants over 5 s. A moving average of 300 ms was used to increase stability. The total gross evoked activity is shown.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Morphology

Acetylcholinergic histology (Fig. 1) yielded the typical absence of staining in the VPMpc [see, for example, plate 32 of Paxinos and Watson (1986) delineating this nucleus and indicating the relative absence of cholinergic activity]. The neurons have been described as being uniformly small and of higher density than in the VPM (Norgren 1984). The low rate at which taste neurons were encountered, however, suggests that they are sparsely distributed among cells in the nucleus. This implies that many neurons in the VPMpc are not taste sensitive, an implication supported by the distribution data of Fig. 2.

The mediolateral axis of the VPMpc lies in a horizontal plane (our histology) (see also Paxinos and Watson 1986). Gustatory neurons, however, were found to lie at an angle: lateral neurons were located dorsally to those that were medial. The same observation was made by Emmers (1964) in the cat shortly after the VPMpc had been identified as the thalamic taste relay.

General response characteristics

The mean spontaneous rate for VPMpc taste cells was 6.3 ± 0.7 spikes/s, similar to those reported earlier by Scott and Erickson (1971) (6.0 spikes/s) and Scott and Yalowitz (1978) (5.3 spikes/s, see Table 2 for references). It was somewhat higher than reported by Nomura and Ogawa (1985) (2.8 ± 2.6 spikes/s] and Ogawa and Nomura (1988) (3.7 spikes/s; Table 2). Spontaneous activity in VPMpc is intermediate between the rate calculated from six studies on the rat PBN [6.6 ± 3.8 (SD) spikes/s] that projects to the VPMpc and that from three studies on the rat primary taste cortex to which VPMpc projects (3.1 ± 1.1 spikes/s; Table 2).

Responses evoked by the four prototypical tastants (13.8 spikes/s to 0.1 M NaCl, 9.3 spikes/s to 0.01 M HCl, 5.1 spikes/s to 0.5 M sucrose, and 4.3 spikes/s to 0.01 M quinine HCl) were of similar magnitude to those reported in the previous three single-unit studies on VPMpc (8.8, 7.6, 6.0, and 6.4 spikes/s, respectively). They were generally intermediate between responses to the same stimuli in PBN and cortex (Table 2).

Gustatory neurons in the nucleus of the solitary tract (NTS) are the most active in the rat's taste system with both spontaneous (8.9 spikes/s) and evoked activity well above those reported in PBN, VPMpc, and cortex (Table 2). Despite the higher overall responses in the NTS, the pattern of evoked activity across tastants was strikingly similar to that of the VPMpc: the correlation between the mean responses to 15 identical stimuli in NTS (Giza and Scott 1991) and VPMpc (this study) was +0.94 (P < 0.001, R² = 0.88).

More generally across the taste system, gustatory neurons have characteristic discharge rates to the basic stimuli at each synaptic relay. Those in NTS are most responsive. If we take mean responses to the four basic stimuli in NTS as 1.0 (Giza and Scott 1991), then the corresponding responses in the CT are 0.23 (Yamamoto et al., 1984), in PBN 0.73 (Scott and Perrotto 1980), in VPMpc 0.32 (current study), and in cortex 0.07 (Yamamoto et al., 1989). These are remarkably linear transforms. Responses to the four basic stimuli form nearly parallel lines as they are plotted across the five synaptic levels from CT to cortex (Fig. 9). Hence, the relative response rates across tastants is preserved throughout the gustatory system. Both spontaneous and evoked response rates progressively decline from NTS to cortex. The activity in VPMpc fits this trend. However, the variability of responses evoked by the four basic tastants decreases even more than the decline in mean evoked firing rate at higher levels. The SDs of responses to the basic stimuli divided by the mean responses yield ratios of 0.75 (CT), 0.53 (NTS), 0.52 (PBN), 0.27 (VPMpc) ,and 0.20 (Cx; references from Table 2). Thus taste responses stabilize as they progress through the neuraxis.

View larger version (12K): [in this window] [in a new window]   Fig. 9. Net evoked discharge rates to each of the 4 basic stimuli at five synaptic levels of the rat's taste system.

We defined signal-to-noise ratio as mean net evoked rate across the basic stimuli divided by mean spontaneous rate and found a gradual decrease from the periphery to the VPMpc: 4.1 (CT), 3.2 (NTS), 2.0 (PBN), and 1.6 (VPMpc). The signal-to-noise ratio in the cortex was similar to that in VPMpc: 1.7.

The BOT showed a downward trend from NTS to cortex, neurons of the latter having a similar BOT as those in the CT (Table 2).

Inhibition has been reported to constitute 12% of the responses in gustatory cortex (cf. Table 2) yet plays only a minor role in hindbrain taste relays. Only 1.9% of stimulus-neuron applications elicited inhibitory responses in VPMpc versus 65% excitatory responses. This is in accord with reports from hindbrain studies and establishes the taste cortex as the first synaptic level at which inhibition plays a significant role.

Taste, touch, and temperature coding

Some 68% of taste cells also responded to cold or warm water and 12% were sensitive to both temperature and touch, nearly always with excitation. The sensitivity of gustatory neurons to thermal stimulation has been tested only sporadically. Ogawa et al. (1968) reported that 40% of the investigated CT fibers responded to cooling the tongue from 40 to 20°C, whereas only 6% were activated by warming from 20 to 40°C (Table 2). Travers and Norgren (1995) reported that 33% of NTS neurons responded to switching a flow of water over the tongue back and forth between 25.6 and 38°C (Table 2). Perrotto and Scott (1976) reported that 10°C water "consistently excited" taste neurons recorded in the PBN and that none responded to 25 and 43°C water (Table 2). Work in our lab has shown that 42 of 57 PBN cells (73.7%) responded to 0 and/or 37°C water, whereas 14.0% were activated by both temperature and touch (Q. Yu, personal communication). Yamamoto and his colleagues have looked more consistently for temperature-induced responses in the rat gustatory cortex. Across three studies, 39.8% of the cells responded to thermal stimulation (Table 2). Thus despite the variability introduced by the use of different temperature stimuli and stimulus volumes, flow rates and response criteria, it is consistently found that a considerable fraction of gustatory cells is also sensitive to thermal stimulation of the oral cavity with an emphasis on responses to cold stimuli.

The interaction between temperature and taste was also investigated. The sum of the responses to 0°C water (9.9 spikes/s) and 23°C 0.1 M NaCl (13.8 spikes/s) is 23.7 spikes/s. The response to 0°C 0.1 M NaCl17.0 spikes/swas significantly less than a linear addition of the activity evoked by the separate modalities of temperature and taste (P = 0.002) as if cold and taste were treated as a mixture and subject to mixture suppression (Bartoshuk 1977). An additional exercise supports this interpretation: we determined from the concentration-response function, the concentration of NaCl (0.046 M) that would evoke the same firing rate as 0°C water; adding this to 0.1 M NaCl (0.146 M NaCl) and interpolating the firing rate evoked by this hypothetical concentration yielded a discharge of 15.6 spikes/s. This is similar to that evoked by cold 0.1 M NaCl (17.0 ± 2.8), suggesting again that the neurons display mixture suppression and follow an interpolated cross-modal Stevens power law (Moskowitz 1971).

Neuron types

The distribution of best responses across all cells found here is remarkably similar to that reported by Nomura and Ogawa (1985) in the VPMpc (Table 2). The accumulated data suggest that from the PBN to VPMpc to cortex the proportion of quinine- and acid-best neurons increases at the expense of the NaCl-best cells. From four studies in the PBN, the mean proportion of N-best cells is 57%, H-best 11%, S-best 27%, and Q-best 5% (Table 2). In the thalamus, the proportions are 49% N-best cells, 25% H-best, 19% S-best, and 7% Q-best cells. In taste cortex, 31% N-best, 29% H-best, 18% S-best, and 23% Q-best (Table 2). This trend is likely to result from the fact that response magnitudes evoked by the basic tastants are all lowerhence more similarat higher levels. The VPM_pc is intermediate in this progression.

We report here the statistically probable existence of three neuron groups: NaCl- (55%), HCl- (16%), and sucrose-oriented (23%) plus 5% quinine-oriented cells. At the level of the NTS, using the same statistical criteria, similar groupings have been reported. When defining each group by the most effective of the four basic stimuli McCaughey and Scott (2000) suggest the same grouping as we report here at the VPMpc: 51% sodium cells, 23% acid cells, 21% sugar cells, and 5% outlying quinine cells. Furthermore, in both studies, the sodium group was first clustered with the acid group, the new cluster formed being clustered with the sugar group, and this at similar levels of correlation.

Three subgroups were identified within the NaCl-group, the central and largest being relatively NaCl specific. McCaughey and Scott (2000) reported the existence of two functionally distinct sodium subgroups in the NTS and found that the effects of rapid induction of sodium appetite were selective for those neurons that were most responsive to NaCl. Thus there may be a functional implication to the subdivision of the N group.

Stimulus similarities

Multidimensional scaling of the VPMpc taste responses showed grouping among sodium and lithium salts, between fructose and sucrose, and among acids and bitter salts. A larger distance was found between glucose and the sucrose and fructose positions in accord with that reported previously in NTS (Giza et al. 1991). These interpretations are in keeping with hierarchical cluster analysis of the responses to the tastants across all neurons (Fig. 7): the sodium and lithium tastants are grouped together; the acids, quinine, and bitter salts form another group; sucrose and fructose, and then polycose and glucose do the same. The polycose-glucose pair clusters with the sodium group, which new group clusters with the acid-bitter group and finally with the sucrose-fructose pair. This analysis suggests, however, that the grouping is not crisp because the correlations within each of the four groups range from 0.53 to 0.70, nearly continuous to those between the groups (0.20-0.46).

It has been reported that one dimension along which the taste system is organized is that of physiological welfare, bounded by nutrients at one extreme and toxins at the other (Scott and Mark 1987). An objective measure of this dimension is the toxicity of a chemical as indexed by its oral LD₅₀ in the rat. In the present data set, a significant regression was obtained between rat oral LD₅₀ and the coordinates of the corresponding stimuli on the z axis (P = 0.004, R² = 0.48), a relationship that is dependent on the presence of the three sugars. Thus the concept of taste as an arbiter of physiological well-being, established at the level of the NTS, is preserved through the thalamus. No LD₅₀ measures were available for polycose and sodium saccharin, and they were excluded from this analysis. Polycose, a readily accepted stimulus, is located anomalously in this respect as has also been reported based on data from the NTS (Giza et al. 1991).

Temporal aspects

Scott and Mark (1987) reported a significant relationship between the phasic-tonic ratios of neural responses in NTS and the locations of the stimuli that elicited them along one dimension of a temporally derived taste space. These positions were closely related to stimulus toxicity: the higher the phasic-tonic ratio, the more toxic the stimulus. At a psychophysical level, this may be related to the sharp onset of tastes evoked by toxins versus the smooth onset that typify carbohydrates. The phasic-tonic ratio (the ratio of evoked activity during the first second to that during the next 4 s) ranged from a mean of 0.64 for three sugars to 2.34 for five alkaloids (Scott and Mark 1987). We did a similar analysis for the 73 thalamic neurons. The phasic portion of each response approached or reached its peak by 300 ms and returned to a tonic state during the following second. Therefore we divided the mean firing rate evoked during the first 300 ms by that evoked between 300 and 1,300 ms after stimulus onset. We found a similar relationship to that reported earlier: quinine HCl 1.45 ± 0.18, HCl 1.34 ± 0.17, NaCl 1.12 ± 0.18, and sucrose 0.98 ± 0.16. Next, we correlated these values to the avidity with which stimuli are preferred to water by rats, as measured by the Hedonic Index (Yamamoto et al., 1985). Reported values are 1.17 for 0.5 M sucrose, 0.43 for 0.1 M NaCl, 0.49 for 0.03 M HCl, and 0.89 for 0.01 M quinine HCl. The phasic-tonic ratio correlated highly with hedonic index (P = 0.002, R² = 0.995). Thus temporal information may be involved in coding for the hedonic value of stimuli.

Overall conclusions

Gustatory neurons of the VPM_pc have response characteristics similar to those of neurons at neighboring levels of the taste system. With respect to polymodal sensitivity, spontaneous rate, evoked responses, variation in activity evoked by basic stimuli, signal-to-noise ratio, proportions of cells responding best to basic tastants, neuron groups, and temporal aspects, VPM_pc neurons appear to fill their roles along the functional continua from NTS to cortex.