Systemic Modulation of Serotonergic Synapses via Reuptake Blockade or 5HT1A Receptor Antagonism Does Not Alter Perithreshold Taste Sensitivity in Rats

Abstract

Systemic blockade of serotonin (5HT) reuptake with paroxetine has been shown to increase sensitivity to sucrose and quinine in humans. Here, using a 2-response operant taste detection task, we measured the effect of paroxetine and the 5HT1A receptor antagonist WAY100635 on the ability of rats to discriminate sucrose, NaCl, and citric acid from water. After establishing individual psychometric functions, 5 concentrations of each taste stimulus were chosen to represent the dynamic portion of the concentration–response curve, and the performance of the rats to these stimuli was assessed after vehicle, paroxetine (7mg/kg intraperitoneally), and WAY100635 (0.3mg/kg subcutaneously; 1mg/kg intravenously) administration. Although, at times, overall performance across concentrations dropped, at most, 5% from vehicle to drug conditions, no differences relative to vehicle were seen on the parameters of the psychometric function (asymptote, slope, or EC₅₀) after drug administration. In contrast to findings in humans, our results suggest that modulation of 5HT activity has little impact on sucrose detectability at perithreshold concentrations in rats, at least at the doses used in this task. In the rat model, the purported paracrine/neurocrine action of serotonin in the taste bud may work in a manner that does not impact overt taste detection behavior.

Introduction

The presence of serotonin (5HT) and some of its associated receptors within taste bud cells (Kaya et al. 2004; Tomchik et al. 2007; Huang et al. 2009), as well as its activity in brain areas associated with feeding and reward (Adlersberg et al. 2000; Tarazi et al. 2002; Pratt et al. 2009; Bubar et al. 2011), make it a likely candidate for the modulation of both taste-guided and ingestive behavior. Indeed, taste disturbances are often reported as both a symptom of depression (Amsterdam et al. 1987; Miller and Naylor 1989; Dess and Chapman 1990; Whittemore 1990; Settle and Amsterdam 1991; Steiner et al. 1993; Arbisi et al. 1996; Joiner and Perez 2004—but see Potts et al. 1997; Scinska et al. 2004), a mood disorder in which serotonergic activity is implicated, and as a side effect of the use of antidepressants (Levenson 1985; Deems et al. 1991; Schiffmann et al. 2000), drugs which often impinge on 5HT reuptake and on monoamine receptor activation.

When a chemical stimulus activates taste receptor cells, ATP is released. Application of 5HT or the selective serotonin reuptake inhibitor (SSRI), paroxetine, in a bath solution in which taste bud cells are isolated results in abolition of tastant-stimulated ATP secretion from the cells (Huang et al. 2009), suggesting that serotonin blunts neural responsiveness to taste stimulation. With respect to behavioral responsiveness, though, administration of paroxetine has been shown to decrease the threshold at which sucrose and quinine are recognized in human subjects (Heath et al. 2006), suggesting that increased availability of serotonin enhances taste sensitivity, a result inconsistent with the inhibitory action of the neurotransmitter on ATP release in taste buds.

In the set of experiments reported here, we sought to characterize the effects of systemic paroxetine administration on taste sensitivity in a rat model. Because paroxetine leads to general serotonergic agonism through reuptake blockade, which could theoretically produce offsetting effects, we also focused on the 5HT1A receptor subtype, which is expressed in a subset of taste bud cells and has been postulated to mediate actions of serotonin in taste bud cells (Huang et al. 2009). Furthermore, antagonism of 5HT1A receptors with WAY100635 has been shown to increase taste-stimulated ATP release from Type II cells (Huang et al. 2009). Accordingly, using a rat model, we tested whether systemic administration of the SSRI paroxetine and the selective 5HT1A receptor antagonist WAY100635 would alter sensitivity, especially at perithreshold concentrations, to prototypical taste stimuli using an operant conditioning-based psychophysical signal detection task.

Materials and methods

Subjects

A total of 32 male Sprague-Dawley rats (Charles River) were used in this study. They were housed individually in polycarbonate tubs in a vivarium room in which temperature, humidity, and lighting (on at 0700h and off at 1900h) were automatically controlled. The rats had ad libitum access to maintenance diet (PMI 5001) unless otherwise noted. A water restriction protocol was implemented such that water bottles were removed from the home cages of the rats on Sunday no more than 23h prior to a test or training session on Monday during which they would work for water (described below). The rats worked for water Monday through Friday, and then the bottles were returned to the home cages of the animals after the completion of the session on Friday, and the bottles remained in place until the following Sunday. The body weights of the rats were monitored daily while the animals were water-restricted to ensure that no rat fell below 85% of its ad libitum body weight. During training, the body weights of 2 rats on one occasion each fell below this criterion, and they were given 10mL supplemental water after completion of the day’s session. No animal fell below 85% of its ad libitum body weight during baseline or drug testing. All procedures were performed in accordance with and approved by the Florida State University Animal Care and Use Committee.

Apparatus

The rats were trained and tested in a modified version of a previously described apparatus known as a gustometer (originally presented in Spector et al. 1990; modified as in Blonde et al. 2006). On the front wall of the animal holding chamber of the apparatus is a stimulus access slot centered between 2 stationary horizontally oriented response spouts, above which cue lights are positioned. A third vertically oriented spout from which taste stimuli are delivered can be accessed through the central slot. This sample spout is motorized and can be positioned behind or out of reach from the access slot. The fluid taste stimuli are held in pressurized reservoirs and controlled amounts of the fluids are delivered to the spouts via computer-controlled solenoid valves upon licking. Licks are registered through a contact circuit that passes less than 50 nA of current through the rat. The animal holding chamber, spouts, and motor are held within a sound-attenuating cubicle, and masking noise is produced throughout each session to minimize the impact of extraneous auditory cues.

Trial structure

During each 40-min session, the rats were allowed to initiate as many trials as possible. A trial was initiated when the motorized sample spout rotated behind the central access port and was licked by the rat twice within 250ms. After the rat licked the dry spout twice, a contingency to ensure that the rat was engaged in licking before the stimulus was presented, a solenoid would open for a duration calibrated to fill the shaft of the spout, and then each successive lick resulted in delivery of ~5 µL of fluid through the spout. During sampling, the house lights in the testing chamber were illuminated. After 10 licks were made or 5 s elapsed, the sample spout rotated away from the access slot, the house light turned off, and 2 cue lights positioned over the response spouts illuminated. During this time, the rat had 5–180 s (limited hold) to respond by licking one of the 2 stationary response spouts. The duration of the limited hold started at 180 s and was reduced throughout training to 5 s during testing. If the correct spout was licked (i.e., the response spout that in the past had been associated with the tastant delivered from the sample spout), the cue lights turned off, the house light was illuminated, and the rat has the opportunity to access 20 licks of water (~5 µL/lick) within 10 s directly from the end of the same response spout. If the rat responded incorrectly or did not respond at all, a 10–30 s time-out was imposed during which all of the lights turned off and no water was available. The duration of the time-out started at 10 s and was increased throughout training to 30 s during testing. When the reinforcement or time-out phase ended, the intertrial interval (~6 s) began and the house lights, if on, were turned off, and the sample spout was positioned over a waste funnel and washed with deionized water and dried with pressurized air. An opportunity for the rat to initiate a new trial began with the positioning of the sample spout behind the access port and the illumination of the house lights.

Taste stimuli

The rats were divided into 4 groups and were trained with reverse osmosis deionized water and one of the following chemicals (concentrations presented in Table 1): sucrose (Mallinckrodt), NaCl (Mallinckrodt), citric acid monohydrate (BDH), and quinine hydrochloride dihydrate (Sigma). Solutions were made fresh daily and presented at room temperature.

Table 1.

Concentrations (mM) of taste stimuli used throughout the phases of the study

Stimulus	Training	Baseline detection testing	Vehicle and drug testing
Sucrose	300	300, 150, 75, 37.5, 18.75, 14.1, 9.4, 4.7, 2.3, 1.8, 1.2, 0.6, 0.3	128.9, 21.8, 11.6, 6.2, 1
NaCl	200, 400a	400, 200, 100, 50, 25, 12.5, 6.3, 3.1, 1.6, 0.8, 0.4, 0.2	268.4, 44.8, 9.1, 1.8, 0.31
Citric acid	30	30, 15, 7.5, 3.8, 1.9, 0.9, 0.5, 0.2, 0.1, 0.06, 0.03	30, 11.5, 3.1, 0.8, 0.04
Quinine	0.3, 0.6a	Not applicableb	Not applicableb

^aThe concentrations of these stimuli were increased due to poor performance.

^bDuring training, stimulus control could only be established in 50% of the animals (n = 4), so data from testing are not presented.

Training

The rats were first trained to lick water from each of the spouts in the gustometer. The rats were then trained to lick one response spout after water was sampled and the other response spout after the taste stimulus was delivered as described below. The spout assignments were counterbalanced across rats.

Spout training

Three sessions were performed during which the rats could lick water ad libitum in the gustometer. On the first day, water was dispensed from the sample spout, on the second day, water was dispensed from the right response spout, and on the third day, water was dispensed from the left response spout.

Side training

For a total of 6 sessions, the rats were presented with either their assigned taste stimulus (3 sessions) or water (3 sessions) alone for a given session and, after sampling, were allowed to lick water from the corresponding response spout. The stimulus presentation (water or tastant) was alternated daily. During this phase, the other response spout was recessed from the testing chamber, and the access slot was covered with a stainless steel plate.

Alternation

During each session in this phase, both the taste stimulus assigned to the rat and water were presented from the sample spout, but the rat was required to respond correctly a criterion number of times before the stimulus dispensed from the sample spout changed. The criterion began at 6, dropped to 4, and then to 2, with 2 sessions conducted at each criterion.

Detection training

In this phase, both the taste stimulus and water were presented randomly (without replacement) in blocks of 2 such that each stimulus had a 50% probability of occurring. In the first session of this phase, if a rat responded incorrectly, it was allowed to respond on the other response spout such that time-outs only occurred if the rat did not respond. During the other stages of this training phase, only one opportunity to make the correct response was given, and the limited hold (e.g., the time allowed for the rat to make a response after sampling) decreased while the time-out increased in duration. Once all of the rats across each of the taste stimuli performed with ≥80% accuracy on trials during which they made a response, all of the animals began the taste detection testing phase. After 21 days of this final stage of training, stimulus control was established in only 4 of the 8 rats in the quinine group, precluding the opportunity for statistical analysis after drug testing. Thus, no further data from the quinine group are discussed or presented.

Determination of baseline psychometric detection functions

Testing was conducted using a modified method of limits in which one concentration of the taste stimulus assigned to each rat and water were presented in each session (see Blonde et al. 2006). On Tuesday and Thursday sessions, the tastant concentration was systematically decreased. On Monday, Wednesdays, and Fridays, the lowest concentration that each individual rat performed at ≥80% accuracy was presented to maintain stimulus control custom-tailored to the sensitivity of each rat. Testing continued until each rat performed at a level not significantly different from chance (50%). During establishment of the baseline detection curve for citric acid, a sudden unexplained drop in performance was seen at 0.9mM that recovered at the next lowest concentration (0.5mM). After the testing series was complete, we retested the rats at 0.9mM and improved performance was observed. However, since we could not identify any apparatus or stimulus problems during the first test session, we used the average performance from both testing sessions to establish the data point used in analysis.

Drugs

The SSRI paroxetine maleate (Tocris) was dissolved in dimethyl sulfoxide (DMSO; Tocris) and then distilled water added such that the final concentration of DMSO was 10%. The 5HT1A receptor antagonist WAY100635 (Sigma) was dissolved in phosphate-buffered saline (PBS) for the phase when it was injected subcutaneously (s.c.) and in heparinized saline (Sigma) for the phase when it was administered intravenously (i.v.) All drugs were stored at room temperature in a desiccator, and solutions were prepared fresh on the days they were used.

Implantation and maintenance of jugular catheters

After completion of detection testing with paroxetine and WAY100635 administered intraperitoneally (i.p.) and s.c., respectively, a chronic indwelling jugular catheter was implanted in each rat (procedure adapted from Rowland et al. 2008). The rats were deeply anesthetized using isoflurane (2–5% in O₂ delivered at 1.0L/min), and the skin over the right jugular vein and along the scapular region was shaved and treated with disinfectant. A small incision was made in the skin over the vein, and the jugular isolated through blunt dissection. The jugular was punctured with microscissors, and 3.2cm of Silastic (Dow Corning) was inserted into the vein and tied in place with silk suture. The free end of the catheter was tunneled subcutaneously to a small incision made on the back of the neck of the rat where it was attached to a pedestal port (Plastics One). The incisions were closed with 3-0 nonabsorbable Nylon suture (Johnson & Johnson). The catheter was flushed immediately after surgery with ~0.3mL of heparinized saline (0.2mg/mL), and 0.2mL of the antibiotics streptokinase (200 units) and Timentin (6mg/mL) in heparinized saline were infused. A small piece of Tygon tubing (Saint-Gobain PPL Corp.) was secured to the port and then closed with a blunted stainless steel pin. The analgesic ketoralac tromethamine (2mg/kg s.c.) was also injected immediately after surgery and administered for the next 3 postsurgical days. The rats were allowed to recover for at least 10 days before any experimental manipulations were imposed. Administration of the i.v. antibiotic cocktail of streptokinase and Timentin (doses as above) continued for 10 days after surgery, and the catheters were flushed with heparinized saline (~0.3mL) daily until the end of the experiment or until patency was lost. Patency was determined by the presence of blood in the catheter during flushing or following an increase in blood glucose levels of ≥40% 1–2min following i.v. administration of 2mL/kg of 1.0g/kg glucose in sterile saline (c.f., Frangioudakis et al. 2008).

Determination of psychometric functions following serotonergic drug administration

Five stimulus concentrations spanning the dynamic range of the mean psychometric function (see below) for each taste stimulus produced during baseline testing were chosen for the drug testing phase. The concentrations tested were determined using the equation:

$$\text{Test concentration = }x\text{(}a-\text{\hspace{0.17em}\hspace{0.17em}0.5) + 0.5}$$

in which a = the asymptote of performance for a given taste stimulus and x = 0.995–0.92, 0.8, 0.5, 0.2, and 0.08–0.005. The ranges ensured that all of the tested concentrations were within the span of those used during training (Figure 1 and Table 1). On the Tuesdays and Thursdays across 5 weeks, the rats were tested with one of these stimulus concentrations, presented in descending order, and water. On Tuesdays, vehicle was administered prior to the test session, and, on Thursdays, the drug was administered. This sequence was standardized rather than counterbalanced in order to provide the longest wash-out period possible and minimize carry-over drug effects. On Monday, Wednesday, and Friday sessions, no injections were given and the stimulus concentration at which each rat performed with ≥80% accuracy during baseline detection testing was used in order to maintain stimulus control.

Figure 1.

The mean (±SE) percent correct responding across concentrations (open circles) during baseline taste detection testing for sucrose, NaCl, and citric acid (n = 8/tastant group). Curves were fit to these data using the psychometric function presented in the Materials and methods. Each “x” represents a concentration that was chosen to be tested after vehicle and drug administration in subsequent sessions. Chance = 50%.

First, we assessed the effect of paroxetine on detection performance. Paroxetine (7mg/kg) or its vehicle (10% DMSO, 1mL/kg) was injected i.p. 1h prior to the beginning of the test session. In humans, the half-life of orally administered paroxetine at therapeutic doses (~20–30mg) is 24h (Kaye et al. 1998), and its metabolic pathway is similar in humans and rats (Johnson 1989). Since many drugs themselves have a bitter taste, we chose to bypass the oral cavity and inject paroxetine systemically. Our dose and timing choice was based on prior work that demonstrated an effect of 5–10mg/kg paroxetine on the appetitive portion of taste-guided behavior in a progressive ratio task (Mathes et al. 2013) and a brief-access test (Mathes and Spector 2011). Next, we determined the effect of WAY100635 when injected s.c. WAY100635 (0.3mg/kg) or its vehicle (PBS, 1mL/kg) was injected s.c. 30min prior to the session. Because of the silent antagonist properties of WAY100635 at the 5HT1A receptor, the affinity constant (IC₅₀ dose) and the dose required to reduce neuronal firing to 50% of baseline (ID₅₀) in combination with 5HT1A agonist administration after i.v. administration rather than half-life have been the focus of pharmacologic profiles. Although the EC₅₀ of WAY100635 to attenuate the behavioral 5HT syndrome or hyperphagia in rats or hypothermia in mice induced by 8-OH-DPAT has been reported as 0.01–0.03mg/kg s.c. 0.5–3h after administration (Forster et al. 1995; Fletcher et al. 1996), our dose and timing choice was guided by a previous study in which 0.1mg/kg decreased operant responding for sucrose solution when injected 30min prior to a progressive ratio task (Mathes and Spector 2010). This dose is also consistent with that used in studies demonstrating blunted motivation in the form of trial initiation in a cognitive task after WAY100635 in rats (Fletcher et al. 1996). Finally, we examined the effect of WAY100635 injected i.v. on the accuracy with which rats could discriminate their assigned taste solution from water. The rats were infused with 0.3mL of 1.0mg/kg WAY100635 or its vehicle (heparinized saline) immediately prior to their test session. Doses of WAY100635 as low as 0.1mg/kg i.v. have been reported to block 8-OH-DPAT-induced neuronal inhibition, but doses up to 10mg/kg had no significant independent effect on neuronal activity (Forster et al. 1995) or elicitation of a behavioral 5HT syndrome (Fletcher et al. 1996).

Data analysis

For each taste stimulus during each of the 3 drug conditions, the mean percent correct on trials with a response after vehicle injection and after drug injection were compared via 2-way analysis of variance (ANOVA) with injection condition and taste stimulus concentration as within-subject factors. Curves were fit to the data using the 3-parameter logistic function:

$$f(x)=\left[\frac{a-50}{\left(1+{10}^{(x-c)b}\right)}\right]+50$$

in which x = the log₁₀ of the stimulus concentration tested, a = asymptotic performance, b = slope, and c = the log₁₀ of the tastant concentration at which performance equaled one half of the asymptote (effective concentration₅₀ [EC₅₀]), which we used as an operational definition of the detection threshold. Each of these parameters from the individual curve fits of each rat within a tastant group were analyzed via 1-way ANOVA with injection condition as a within-subject factor. The data from 1 rat in the citric acid group after vehicle administration during the paroxetine testing phase could not be fit to a curve. After testing with WAY100635 s.c., one rat in the sucrose group was euthanized due to ulceration of its hind paws. Only rats that were determined to have patent jugular catheters after testing with all 5 stimulus concentrations were included in analysis of the effect of WAY100635 injected i.v. All of the rats in the sucrose group passed (n = 7) and only one rat in the NaCl group did not pass (n = 6). However, the citric acid group size was reduced to n = 3, and since this precluded statistical assessments we do not present these data.

Results

Paroxetine

Paroxetine at the dose tested had no significant effect on percent correct responses to the 5 concentrations of sucrose or citric acid that we chose (Figure 2 and Table 2), nor was a drug × concentration interaction revealed (all P > 0.05), though, as expected, performance changed as a function of concentration (all P < 0.001). More importantly, there were no significant differences as a function of drug treatment in the analyses of the parameters of the respective curve fits (Figure 2 and Table 3). Indeed, the measures across time were quite stable overall, with a percent change of only ~1% seen on average across the concentrations and threshold values differing only by ~0.1 log unit. While the overall change in percent correct responding from vehicle to paroxetine conditions was, on average, only 5% for NaCl, this decrease was statistically significant (Table 2). However, the effects were only apparent across the concentrations explicitly tested—there were no significant effects of paroxetine on the parameters defining the psychometric function (i.e., asymptote, slope, or EC₅₀; Figure 2 and Table 3). Further, no significant drug × concentration interaction was revealed (P > 0.05) despite percent correct responding decreasing across concentration (P < 0.001). This suggests that a general increase in serotonergic tone via reuptake blockade did not alter the psychophysically assessed threshold at which rats discriminated sucrose, NaCl, and citric acid from water.

Figure 2.

The mean (±SE) percent correct responding across concentrations after administration of vehicle (saline, 1mL/kg i.p.; filled circles with solid curves) or the SSRI paroxetine (7mg/kg i.p.; open circles with hatched curves) during taste detection testing for sucrose, NaCl, and citric acid. Paroxetine had no effect on the asymptote, slope, or EC₅₀ (which was operationally defined as threshold) of the functions fit to the data from any of the tastants (n = 8/group). A statistically significant drug-induced decrease in accuracy during NaCl testing was revealed in analysis of the performance across the concentrations tested, but it was, on average, only a 5% difference. Chance = 50%.

Table 2.

Main effect of injection condition (vehicle vs. drug) revealed by 2-way ANOVA on the percent correct responding during behavioral testing with the chosen taste stimulus concentrations

Taste stimuli	Main effect of drug
	Paroxetine i.p.	WAY100635 s.c.	WAY100635 i.v.
Sucrose	F(1,7) = 0.341, P = 0.578	F(1,7) = 0.918, P = 0.37	F(1,6)=12.457, P = 0.012
NaCl	F(1,7) = 34.119, P = 0.001	F(1,7) = 2.981, P = 0.128	F(1,5)=12.065, P = 0.018
Citric acid	F(1,7a) = 1.121, P = 0.325	F(1,7) = 1.345, P = 0.284	Not applicable

All effects of concentration P ≤ 0.001 and all injection condition × concentration interactions P ≥ 0.05.

^aA curve could not be fit to the data for 1 animal, but its data were still included in the ANOVA for percent correct. However, if that animal is excluded from the ANOVA analysis, there is still no effect (F(1,6) = 2.946, P = 0.137).

Table 3.

The mean ± standard error (M ± SE) values and main effect of injection condition (vehicle vs. paroxetine injected i.p.) on the parameters of the psychometric function during testing as revealed by 1-way ANOVA

Stimulus	Vehicle (M ± SE)	Paroxetine (M ± SE)	1-way ANOVA
Asymptote (a)
Sucrose	96.7±1.2	94.2±1.5	F(1,7) = 1.116, P = 0.326
NaCl	95.6±1.0	91.3±3.3	F(1,7) = 1.907, P = 0.21
Citric acid	94.1±1.5	96.0±0.7	F(1,6) = 1.923, P = 0.215
Slope (b)
Sucrose	−1.9±0.3	−5.7±3.5	F(1,7) = 1.112, P = 0.327
NaCl	−2.7±1.4	−2.4±1.3	F(1,7) = 0.02, P = 0.891
Citric acid	−5.4±2.0	−3.2±0.5	F(1,6) = 0.932, P = 0.372
EC50 (c)
Sucrose	−1.8±0.04	−1.9±0.03	F(1,7) = 0.395, P = 0.55
NaCl	−2.2±0.2	−2.1±0.2	F(1,7) = 1.684, P = 0.235
Citric acid	−2.5±0.08	−2.4±0.08	F(1,6) = 3.102, P = 0.129

WAY100635

When injected s.c., WAY100635 had no significant effect on percent correct responses to the tested concentrations of the taste stimuli (Figure 3 and Table 2), nor was a drug × concentration interaction revealed (all P > 0.05), though, as expected, performance changed as a function of concentration (all P < 0.001). As with paroxetine, WAY100635 s.c. had no effect on the psychometric parameters of the curves derived for rats discriminating sucrose, NaCl, or citric acid from water (Figure 3 and Table 4). When injected i.v., WAY100635 significantly decreased overall performance across the tested concentrations of the rats responding after presentation of sucrose and NaCl, by 2 and 3% on average, respectively, (Figure 4 and Table 2), but there was no concentration × drug interaction (P > 0.05). There was, as expected, an overall effect of taste stimulus concentration (P < 0.001). Importantly, the curve parameters from the psychometric functions of neither of these stimuli were significantly affected by the drug (Figure 4 and Table 5). This provides evidence that antagonism of 5HT1A receptors via systemic administration at these doses has minor, if any, effects on taste detection, at least of sucrose and NaCl, in rats.

Figure 3.

The mean (±SE) percent correct responding across concentrations after administration of vehicle (PBS, 1mL/kg s.c.; filled circles with solid curves) or the 5HT1A receptor antagonist WAY100635 (0.3mg/kg s.c.; open circles with hatched curves) during taste detection testing for sucrose, NaCl, and citric acid. WAY100635 had no effect on either the percent correct performance across concentrations or the asymptote, slope, or EC₅₀ of the functions fit to the data from any of the tastants (n = 8/group). Chance = 50%.

Table 4.

The mean ± standard error (M ± SE) values and main effect of injection condition (vehicle vs. WAY100635 injected s.c. [WAY_s.c.]) on the parameters of the psychometric function during testing as revealed by 1-way ANOVA testing

Stimulus	Vehicle (M ± SE)	WAYs.c. (M ± SE)	1-way ANOVA
Asymptote (a)
Sucrose	98.5±0.5	98.2±0.7	F(1,7) = 0.25, P = 0.632
NaCl	95.5±2.4	96.7±0.7	F(1,7) = 0.267, P = 0.621
Citric acid	95.2±1.4	96.5±2.1	F(1,7) = 1.028, P = 0.344
Slope (b)
Sucrose	−2.6±0.5	−2.2±0.3	F(1,7) = 0.31, P = 0.595
NaCl	−3.5±2.5	−2.6±1.2	F(1,7) = 0.102, P = 0.759
Citric acid	−7.2±2.6	−4.7±1.9	F(1,7) = 0.771, P = 0.409
EC50 (c)
Sucrose	−2.0±0.03	−2.0±0.02	F(1,7) = 0.018, P = 0.896
NaCl	−2.4±0.1	−2.3±0.2	F(1,7) = 2.731, P = 0.142
Citric acid	−2.6±0.06	−2.5±0.04	F(1,7) = 4, P = 0.086

Figure 4.

The mean (±SE) percent correct responding across concentrations after administration of vehicle (0.3mL of 0.2mg/mL heparinized saline i.v.; filled circles with solid curves) or WAY100635 (1.0mg/kg i.v.; open circles with hatched curves) during taste detection testing for sucrose and NaCl. WAY100635 had no effect on the asymptote, slope, or EC₅₀ of the functions fit to the data from sucrose (n = 7) or NaCl (n = 6). A statistically significant drug-induced decrease in accuracy during sucrose and NaCl testing was revealed in analysis of performance across the concentrations tested, but it was, on average, only a 2–3% difference. Chance = 50%.

Table 5.

The mean ± standard error (M ± SE) values and main effect of injection condition (vehicle vs. WAY100635 infused i.v. [WAY_i.v.]) on the parameters of the psychometric function during testing as revealed by 1-way ANOVA

Stimulus	Vehicle (M ± SE)	WAYi.v. (M ± SE)	1-way ANOVA
Asymptote (a)
Sucrose	95.1±1.2	94.3±1.7	F(1,6) = 0.22, P = 0.656
NaCl	92.0±1.9	84.1±3.7	F(1,5) = 3.23, P = 0.132
Slope (b)
Sucrose	−2.3±0.6	−1.7±0.2	F(1,6) = 1.048, P = 0.346
NaCl	−5.6±3.7	−4.2±2.1	F(1,5) = 0.094, P = 0.772
EC50 (c)
Sucrose	−2.0±0.02	−2.0±0.07	F(1,6) = 0.674, P = 0.443
NaCl	−1.9±0.4	−2.2±0.2	F(1,5) = 1.939, P = 0.223

Discussion

In these experiments, modulation of serotonergic tone via systemic administration of the SSRI paroxetine and the 5HT1A receptor antagonist WAY100635 had little to no effect on the accuracy with which rats could discriminate sucrose, NaCl, and citric acid from water. The doses and time courses of the serotonergic agents used have previously been shown to alter hedonic-based taste-guided performance, such as responding in a brief-access test (Mathes and Spector 2011) or under a progressive ratio schedule (Mathes and Spector 2010; Mathes et al. 2013). Although occasionally drug treatment decreased overall performance to the 5 selected concentrations in a statistically significant manner, the effect was typically small (on average between 2 and 5%). The psychometric functions allowed for the characterization of sensitivity in the perithreshold range (EC₅₀ and slope) while assessing general performance (asymptote). At no time did the drugs shift the curves laterally or change their slope significantly. Furthermore, a lack of effect on asymptotic levels of responding suggests that any slight effect of the drug on general performance were not reflected in the nature of the psychometric functions.

These results are somewhat surprising since a single oral administration of a dose of paroxetine that is more than 10 times lower than that used here (~0.3mg/kg) has been shown to decrease the threshold at which healthy humans report tasting a sweet or bitter taste stimulus (i.e., increased sensitivity to sucrose and quinine) and had no effect on recognition of NaCl or citric acid (Heath et al. 2006). Although we were unable to test the effects of paroxetine with quinine, we were able to test its effects with sucrose, and in no way did paroxetine impact performance or detection. Further, the minor effects we did observe were in the opposite direction and were with NaCl, not sucrose (i.e., paroxetine decreased performance across the 5 concentrations of NaCl tested but did not shift the slope or the EC₅₀ of the NaCl function). It is possible that this discrepancy stems from differences in the tasks used to test taste sensitivity (i.e., recognition vs. detection) and the way in which threshold was defined in the 2 studies. During testing trials in the study using humans, only the taste solutions were presented as a stimulus, whereas in our study, trials could either be the taste solution or its absence (i.e., water). This means that the human volunteers were always incorrect if they reported not tasting a test stimulus, whereas our rats had to also correctly report the absence of the stimulus. Accordingly, the percentage correct measure that we used to assess performance accounted for both the false alarm rate (reporting that a stimulus is present when it is not) and the miss rate (reporting that a stimulus is not present when it is) in addition to correctly reporting the presence and absence of a stimulus. It is possible that paroxetine treatment altered the response bias for human subjects to report the presence of the stimulus at weak concentrations as opposed to or in addition to effects of the drug on sensation. Further, in our study, the rats were highly motivated to perform accurately since they were water-deprived and correct responses resulted in water. In the human study, there were no consequences contingent on performance. Moreover, in Heath et al. (2006), the taste stimuli were delivered by the experimenters to the tip of the tongues using a saturated cotton bud, whereas in our study, the rats licked small amounts of the test solutions, allowing the stimulus access to the entire oral receptor field (although the anterior tongue would receive optimal stimulation). Finally, our study used only male rats, whereas 40% of the human participants in Heath et al. (2006) were female. Although sex differences on the effect of paroxetine on sucrose and quinine detection are not discussed in Heath et al. (2006), we may have seen different results had we used female rats, which have been shown to be more sensitive to the anorexigenic effects of serotonergic drugs (Eckel et al. 2005) in ways that influence taste-guided behavior (Atchley et al. 2005).

We expected an effect of paroxetine on taste detection not only in light of the results in humans (Heath et al. 2006), but also because of the suspected role of serotonin in taste bud signaling. Serotonin is released from Type III cells in response to taste stimuli and is hypothesized to activate 5HT1A receptors found on Type II receptor cells, which then blunts ATP release from Type II cells. Blocking 5HT1A receptors with the antagonist WAY100635 nearly doubles taste-evoked ATP release in isolated taste buds, supporting an endogenous mechanism (Huang et al. 2009). However, the functional significance of this activity is unknown, and in our study, the 5HT1A receptor antagonist WAY100635 had little effect on taste sensitivity in rats.

Because these drugs readily cross the blood–brain barrier (Cummings and Gjedde 1993; Kung et al. 1994; Uhr et al. 2003), it is possible that the reason we do not see robust effects on taste detection after either drug is due to central mediation canceling out effects on peripheral taste responses. It should be noted, however, that we cannot verify that the drugs at the doses and time courses we used directly affected taste bud signaling, though at least i.p. paroxetine and WAY100635 when injected s.c. impact other taste-guided behaviors (Mathes and Spector 2010, 2011; Mathes et al. 2013). If activity in the taste bud changes after paroxetine or WAY100635 administration, but brain activity changes in the opposite direction, an overall systemic modulation of serotonergic tone would exert no measurable effect on behaviorally determined taste sensitivity. However, why this would be the case in rats but not in humans is unclear. While it is unfortunate that we could not produce data for quinine, one of the stimuli shown to be affected by paroxetine in humans, it is apparent that stimulus control was well established and maintained during testing of the other taste stimulus altered in the human study—sucrose—and the only time performance for it was impacted was when WAY100635 i.v. decreased it by a margin of 2%.

Our lack of an effect of serotonergic modulation on taste detection does not, however, imply that serotonergic activity, either centrally or peripherally, has no impact on taste-guided behavior. Indeed, we have previously shown that paroxetine decreases the number of trials taken to certain tastants under certain conditions in a brief-access taste test but without affecting the number of licks taken relative to water during trials (Mathes and Spector 2011). We have also shown that paroxetine decreases the amount of work in which a rat will engage in order to receive sucrose (while nondeprived) or quinine (while water-deprived) as a reinforcer under a progressive ratio schedule (Mathes et al. 2013). Others have shown that in a taste reactivity paradigm fenfluramine, a general serotonin receptor agonist and reuptake inhibitor, enhanced aversive reactions and that the 5HT1A agonists, buspirone and gepirone, inhibited both ingestive and aversive oromotor responses to orally infused tastes (Treit and Berridge 1990; Barnfield et al. 1994). Furthermore, it is well documented that SSRIs and agonists of various 5HT receptor subtypes attenuate appetite (see Halford and Blundell 2000; Halford et al. 2007; Garfield and Heisler 2009; Lam et al. 2010; Hayes and Greenshaw 2011), which is, to some extent, influenced by taste. Thus, it appears that systemic modulation of serotonergic activity affects the motivation, or drive, of rats to engage in behaviors guided by taste, but does not impact the level at which rats can detect sucrose, NaCl, or citric acid and discriminate them from water. Studies using centrally injected drugs or 5HT itself, which does not cross the blood–brain barrier (e.g., Fletcher and Burton 1984; Montgomery and Burton 1986), would help unravel the site of taste-mediated effects. The role of serotonin in other functions of the gustatory system, such as the discriminability of one taste stimulus from another, also remains to be determined. However, from these results, we can conclude that basic taste sensibility is not disrupted by systemic blockade of serotonin reuptake or 5HT1A receptors, and, thus, the purported paracrine/neurocrine role of serotonin in the taste bud may work in a manner that does not influence overt behavior involved in a rat model of taste detection.

Funding

This work was supported in part by the National Institute on Deafness and Other Communication Disorders at the National Institutes of Health (1F32DC010517 to C.M.M.).