Neurobiological and Psychophysical Mechanisms Underlying the Oral Sensation Produced by Carbonated Water

Experiment 1: electrophysiology

Animals. Ten adult male Sprague Dawley rats (Simonsen Inc., Gilroy, CA), weighing between 380–480 gm, were used in the experiments. The animals were housed one per cage in a room maintained on a 12 hr light/dark cycle and an ambient temperature of 21 ± 2°C. Food and water were available ad libitum. To obviate any possible effects of circadian rhythms (Lotsch et al., 1998), the experiments were always started between 12:00 P.M. and 2:00 P.M. All protocols were approved by the University of California (UC) Davis Animal Use and Care Advisory committee.

Surgery. Each animal was anesthetized with thiopental (80 mg/kg, i.p.). Core body temperature was maintained at ∼37°C by placing the animal on a heating pad. A tracheotomy was performed, and a tracheal cannula was implanted. Similarly, a catheter was inserted into the jugular vein so thiopental could be infused intravenously (10 mg · kg⁻¹ · d⁻¹) during the course of the experiment to maintain anesthesia; the rate was increased briefly if the rat showed signs of insufficient anesthesia, such as change in heart rate or reflexive movement in response to a noxious stimulus. The occipital bone and upper cervical spine were visualized via a midline incision, and the base of the cerebellum, lower brainstem, and C1 spinal cord were exposed by removal of the caudal portion of the occipital bone and atlas. Each animal was then placed in a stereotaxic frame (Kopf Instruments, Tujunga, CA) with the head slightly ventroflexed and the upper cervical spine immobilized with a vertebral clamp. The dura mater was removed, and an agar (Difco, Detroit, MI) pool was formed over the brainstem. After the agar hardened, an opening was cut in an area overlying the target recording site and filled with 0.9% saline. Finally, a small clip was placed over the upper and lower incisors in such a way as to keep the mouth open and the tongue easily accessible. Isotonic saline was applied frequently to the tongue to prevent desiccation.

Recording. A Teflon-insulated tungsten recording microelectrode (∼10 MΩ; F. Haer Inc., Brunswick, ME) was advanced into the brainstem in 5 μm steps using a hydraulic microdrive (Kopf Instruments). Extracellular single-unit activity was amplified and displayed by conventional means and fed via an analog-to-digital converter (Microstar Industries, Seattle, WA) to a computer for analysis and storage. Unitary action potentials were discriminated, counted, and displayed in peristimulus-time histogram (PSTH) format (bin width of 1 sec), using software developed in Erlangen, Germany (Forster and Handwerker, 1990).

Recordings were made from single units in the superficial layers of the dorsomedial aspect of Vc that responded to ipsilateral mechanical (touch, pressure, pinch) and heat (∼54°C) stimuli, and carbonated water was applied bilaterally to the dorsoanterior one-third of the tongue. The search for units was restricted to an area previously shown to contain neurons responsive to noxious chemical stimulation of the tongue (Carstens et al., 1995, 1998). Briefly, this region included the area approximately between 0.5 mm rostral and 2 mm caudal to the obex, and 1.5 mm lateral. Tongue units responsive to mechanical stimulation were readily observed at depths ranging from 50 to 500 μm below the medullary surface.

Stimulation. The stimuli used to induce activity in trigeminal neurons included noxious heat [hot (∼ 54°C) deionized water], buffered hydrochloric acid, pH 1 (Fisher Scientific, Pittsburgh, PA), and commercially available carbonated water, pH 3.4 ± 0.1 (Safeway Foods, Inc., Pleasanton, CA).

A syringe was used to apply stimuli bilaterally to the anterodorsal surface of the tongue in ∼0.1 ml volume (except carbonated water, which was applied continuously at the rate of ∼0.1 ml/sec). All chemicals were delivered at room temperature to avoid any confounding effects of cooling or heating. The acid stimulus was applied as a bolus, left on for 30 sec, and then immediately rinsed with ∼2 ml of isotonic saline. Freshly opened carbonated water was applied continuously for 30 sec, followed immediately by a saline rinse.

In cells showing stable responses to carbonated water applied three times in succession at 5 min interstimulus intervals, the carbonic anhydrase inhibitor dorzolamide hydrochloride (22.3 mg/ml; Merck, West Point, PA) was then delivered. Dorzolamide was applied three times (at 0, 5, and 10 min) as a bolus (∼0.1 ml) to the dorsal lingual epithelium. After the last application, the drug was left on the lingual surface for 10 min, after which the response of the unit to carbonated water was recorded again. If the response of the cells to carbonated water was reduced after dorzolamide, we continued testing with carbonated water at ∼10 min intervals to determine whether the response recovered to predorzolamide levels.

Histology. After the completion of each recording session, an electrolytic lesion was made at the recording site by passing current (6 V DC) through the microelectrode for 20 sec. Animals were then killed with a lethal overdose of thiopental (intravenously), and the brains were removed and post-fixed in 10% formalin. The brainstems were frozen, cut into 50 μm sections, collected on glass slides, counterstained with neutral red, and examined under a light microscope. Lesions were identified and collectively plotted onto a representative brainstem section (Fig.1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Histologically recovered recording sites in superficial laminae of the dorsomedial Vc. A, Photomicrograph of lesion site in Vc (arrow).B, Recording sites (●) for nine units, compiled on representative section of brainstem. Lateral dashed lines indicate approximate border of laminae I–II, and medialdashed lines indicate approximate ventral border of Vc.CU, Cuneate nucleus; GR, nucleus gracilis; ION, inferior olivary nucleus;Pyr, pyramid.

Data analysis. Three conditions were used in these experiments (control, control plus dorzolamide, and recovery). Data from each unit under each condition were integrated over the initial 15 sec [spontaneous activity before carbonated water application (0–15 sec)] and again over the next 15 sec period (16–31 sec) during the initial period of carbonated water application and before any significant adaptation occurred. The effects of dorzolamide treatment were analyzed across all units using a Student’s pairedt test to compare responses to carbonated water before and after dorzolamide application; p < 0.05 was taken to be significant.

Experiment 2: immunohistochemistry

Animals. Twenty-four adult male Sprague Dawley rats (Simonsen Inc.), weighing between 400–500 gm, were used in the experiments. The animals were housed in the same manner as described above. Twenty-four hours before the experiment, the animals were brought to the laboratory and acclimated to the experimental environment. The experimental protocol was approved by the UC Davis Animal Use and Care Advisory committee.

Stimulation. Animals were anesthetized with sodium pentobarbital (65 mg/kg, i.p.). Once a proper plane of anesthesia was attained as assessed by areflexia, a clip was placed gently over the upper and lower incisors to hold the mouth open slightly. Parafilm (American National Can, Neenah, WI) was gently placed under the tongue to prevent stimuli from contacting the gingiva and other oral surfaces. Animals received the following chemical stimuli. (1) Carbonated water, pH 3.4 ± 0.1 (Safeway Foods Inc.), which was flowed onto the lingual surface at a rate of ∼10 ml/min for 10 min (n= 6). (2) Dorzolamide hydrochloride (22.3 mg/ml; Merck), applied topically three times at 5 min intervals by bolus application (0.1 ml) to the dorsal surface of the tongue (n = 11). The third application was left on for 10 min, after which carbonated water was applied in the same manner as above. (3) Isotonic saline control, applied in the same manner as dorzolamide (n = 5). This control group allowed us to assess the degree of stimulation potentially caused by dorzolamide application. (4) A flat water control (n = 7). The flat water was made by exposing the same carbonated water as used previously to air for 24 hr and was assessed qualitatively by the investigators. The flat water was flowed at the same rate as the carbonated water; this group controlled for any mechanical activation of lingual fibers by the flow. (5) Unstimulated controls (n = 6). This last control group was perfused 2 hr after induction of anesthesia without any type of stimulation; this allowed us to determine basal levels of c-Fos expression. After the stimulation procedure, the incisor clip was carefully removed, and the animals were allowed to lie quietly on a heating pad until the perfusion.

Staining. Two hours after the onset of stimulation with carbonated water (or saline or flat water), the rats were perfused through the heart with 250 ml of PBS, followed by 500 ml of 4% paraformaldehyde. The brains were removed and post-fixed for 24–48 hr, after which they were placed in a 30% sucrose solution for cryoprotection. One to 2 d later, the brainstems were frozen, cut into 50 μm sections, and processed for c-Fos immunohistochemistry. The sections were first blocked with 3% normal goat serum (in PBS with 0.3% Triton X-100) for 1 hr and then exposed to the primary c-Fos antibody (diluted 1:50,000; Arnel Products Inc., New York, NY) for 24–36 hr. The primary antibody was removed, sections were washed, and the secondary biotinylated goat anti-rabbit antibody (Vector Laboratories, Burlingame, CA) was applied. One hour later, this antibody was removed, and the sections were washed again and then subjected to an avidin–biotin–peroxidase reaction (Vector Laboratories). Finally, cell nuclei expressing Fos-like immunoreactivity (FLI) were stained black by a nickel diaminobenzidine reaction. Brainstem sections were mounted on gelatin-coated slides, air-dried, cleared in alcohol, and coverslipped. The locations of cell nuclei expressing FLI were observed and quantified under the light microscope (E-400; Nikon, Tokyo, Japan).

Data analysis. The numbers of cell nuclei with FLI were counted in five areas of the brainstem shown previously to be responsive to irritant chemicals placed on the tongue (Carstens et al., 1995). Specifically, these areas were (1) the dorsomedial aspect of Vc, (2) the ventrolateral aspect of Vc, (3) the NTS between the level of the pyramidal decussation caudally and area postrema rostrally, (4) a region of the ventolateral medulla near the lateral reticular nucleus, and (5) the AP (see Fig. 5). Brainstem sections were selected for quantification of FLI so that comparisons with sections at corresponding levels of the brainstem could be made between animals. The investigator who selected sections and did the counts was blinded as to the experimental treatment. For illustrations, representative sections were imaged with a color video camera (DC-330; Dage-MTI, Michigan City, IN) using Scion Image software (Scion Corp., Frederick, MD) and imported to commercially available graphical software (Corel Draw; Corel, Ottawa, Ontario, Canada), which allowed locations of FLI to be plotted directly and accurately onto a computer-generated trace of the section. Between-treatment group comparisons of mean bilateral counts of FLI for each region of interest were statistically analyzed by an unpaired t test, with p < 0.05 considered to be significant.

Experiment 3: human psychophysics

Subjects. Twenty-one subjects (10 male, 11 female) ranging in age from 19 to 29 years participated in the experiments; all were students at UC Davis. The protocol was approved by the UC Davis Human Subjects Review committee. Subjects were asked to refrain from eating, drinking, or smoking for at least 1 hr immediately before the experiment (three subjects reported to be smokers). Subjects participated in a single session that lasted <30 min.

Stimuli. The carbonic anhydrase inhibitor dorzolamide hydrochloride (22.3 mg/ml; Merck) was used at full strength. A control solution was prepared that approximately matched the dorzolamide in viscosity [1% methyl cellulose (Sigma, St. Louis, MO) in deionized water] and taste (1.26 mm quinine HCl; BDH Chemicals, Poole, UK). The control solution was carefully chosen so as to minimize bias and to equalize any bitter-induced inhibition of CO₂ pungency as described previously byCometto-Muñiz et al. (1987). The carbonated water stimulus, pH 3.4 ± 0.1, was prepared in our laboratory by pressurizing (50 psi) deionized water at room temperature with CO₂(95%) for 2 d. Pentanoic acid (Sigma) was diluted with deioinized water to a final concentration of 200 mm. To test lingual sensitivity to tactile stimulation, a von Frey monofilament (Stoelting, Chicago, IL) calibrated to 0.229 mN was used.

Stimulus application. The primary purpose of this experiment was to determine whether previous treatment with the carbonic anhydrase inhibitor dorzolamide attenuated the perceived sensation caused by the application of carbonated water to the human lingual surface. To this end, we used a half-tongue protocol similar to one used previously (Dessirier et al., 1997, 1998). Briefly, a filter paper disk (2.5 cm diameter; Whatman International Ltd., Maidstone, UK) was cut in half, saturated with dorzolamide, and applied, with forceps, to one side of the dorsal surface of the tongue. The control solution (1.26 mm quinine HCl in 1% methyl cellulose) was simultaneously applied, in the same manner, to the opposite side of the tongue; the side receiving dorzolamide or control was counterbalanced across subjects. Because the application of dorzolamide increases salivary flow rate, which might cause confounding effects because of the mixing of chemicals on the lingual surface, a suction device (Saliva Ejector, 6 inch clear; Sullivan Dental Products Inc., Sacramento, CA) was used by subjects to remove excess saliva. Subjects were allowed to use the device at any time, except for the 15 sec immediately before performing the two-alternative forced-choice (2-AFC) or rating tests (see below). This device has been used in previous experiments and was demonstrated to be free of any confounding effects (Dessirier et al., 1997). The filter papers were left on the tongue for 5 min, after which time subjects were asked to rinse their mouth with deionized water. After the rinse procedure, lingual sensitivity to CO₂ was tested by flowing carbonated water at a rate of 20 ml/min bilaterally over the dorsal surface of the tongue for 5 sec on one trial and 15 sec on the next. It was noted in pilot studies that increasing the duration of flow diminished the differences between the treated and untreated sides of the tongue. Because we wanted to maximize sensitivity to the carbonated water stimulus, durations were not counterbalanced across subjects; all subjects completed the 5 sec trial before completing the 15 sec trial.

We used a 2-AFC methodology in which we asked subjects to indicate the side of the tongue perceived to have the strongest sensation elicited by carbonated water during a 5 sec application and the last 5 sec of a 15 sec application. In the previous case, subjects were told to concentrate on the sensation elicited by the carbonated water during the entire 5 sec application, whereas, in the latter case, subjects were told to disregard any sensation during the first 10 sec and compare the sensations elicited on either side of the tongue only during the last 5 sec (at which time they were prompted). In addition to the forced-choice, subjects were also asked, at each time point (5 and 15 sec), to independently rate the intensity of the sensation elicited by carbonated water on each side of the tongue using a 0–10 category scale (0, no sensation; 10, intense sensation).

Although we believed that dorzolamide would selectively attenuate CO₂ irritation by blocking carbonic anhydrase activity, it was conceivable that the treatment could have nonspecific effects, thereby blocking the sensations induced by other chemical or tactile stimuli. Therefore, to test this possibility, separate control experiments were conducted using acid (200 mm pentanoic acid) and tactile stimuli (von Frey filament; 0.229 mN). The pentanoic acid was chosen because it had an oil–water partitition coefficient similar to CO₂. The order in which these treatments were given was counterbalanced across subjects. For acidic stimuli, two filter paper disks (1.0 cm diameter; Whatman International Ltd.), each saturated with 15 μl of pentanoic acid, were placed, using forceps, on each side of the tongue in a corresponding area that had been previously treated with dorzolamide or control solution. Subjects were told to close their mouth, and after ∼5 sec, the filter papers were removed and subjects were again asked to perform a 2-AFC and indicate the side of the tongue for which a stronger tingling–burning sensation was perceived. They were asked to disregard sour taste when making this judgment. In addition, they again used the same 0–10 category scale to rate the intensity of the sensation perceived on each side of the tongue.

Tactile sensitivity of the tongue was tested by applying the von Frey filament to the dorzolamide-treated side, the untreated (control) side, or not at all (blank), in randomized order. Each condition was tested 10 times for a total of 30 trials. Subjects responded by indicating whether or not they felt the stimulus and whether or not they were sure of their response. From these data, response matrices for each subject were constructed from which indices representing the tactile sensitivity on each side of the tongue were calculated (R-index) (O’Mahony, 1992).

To show that the effect of dorzolamide was present when subjects performed the acid and tactile control tests, carbonated water was applied again at the end of the session. As before, subjects performed a 2-AFC test and rated the intensity of carbonation on the treated and untreated sides of the tongue at 5 sec.

Finally, it is possible that subjects chose a side as having a stronger sensation elicited by carbonation, not because of the perception encountered, but because of some cue elicited by the dorzolamide or control treatment. Thus, at the end of each experimental session, two filter papers (2.5 cm diameter, cut in half) were again saturated with dorzolamide or control solution, placed onto both sides of the tongue in random order, and left for 5–10 sec. After removal of the papers, subjects were asked to select the side they thought contained the chemical responsible for reducing the intensity of the sensation evoked by carbonated water. If no taste or textural cues were consistently used by subjects, it would be expected that subjects would choose randomly between the dorzolamide or control; this would result in each chemical being selected 50% of the time.

Data analysis. A binomial analysis was used to determine whether a significant majority of subjects chose a particular side (dorzolamide-treated vs nontreated) as having a stronger carbonation- or acid-evoked sensation and a d′ analysis (Ennis, 1993) was also performed to determine the strength and significance of the effect using the method of Bi et al. (1997); p < 0.05 was considered significant in all cases. To determine whether the intensity of the sensation elicited by carbonated water or pentanoic acid varied significantly between the dorzolamide- and control-treated sides of the tongue, mean intensity scores for each side were calculated under each condition, and a paired Student’s t test was used. R-indices for the treated and untreated sides of the tongue were calculated, and a t test was used to determine whether there was a significant difference between the two.