Taste sensitivity to a mixture of monosodium glutamate and inosine 5′-monophosphate by mice lacking both subunits of the T1R1+T1R3 amino acid receptor

Abstract

The taste of l-glutamate and its synergism with 5′-ribonucleotides is thought to be primarily mediated through the T1R1+T1R3 heterodimer in some mammals, including rodents and humans. While knockout (KO) mice lacking either receptor subunit show impaired sensitivity to a range of monosodium glutamate (MSG) concentrations mixed with 2.5 mM inosine 5′-monophosphate (IMP) in amiloride, wild-type (WT) controls can detect this IMP concentration, hindering direct comparison between genotypes. Moreover, some residual sensitivity persists in the KO group, suggesting that the remaining subunit could maintain a limited degree of function. Here, C57BL/6J, 129X1/SvJ, and T1R1+T1R3 double KO mice (n = 16 each to start the experiment) were trained in a two-response operant task in gustometers and then tested for their ability to discriminate 100 µM amiloride from MSG (starting with 0.6 M) and IMP (starting with 2.5 mM) in amiloride (MSG+I+A). Testing continued with successive dilutions of both MSG and IMP (in amiloride). The two WT strains were similarly sensitive to MSG+I+A (P > 0.8). KO mice, however, were significantly impaired relative to either WT strain (P < 0.01), although they were able to detect the highest concentrations. Thus, normal detectability of MSG+I+A requires an intact T1R1+T1R3 receptor, without regard for allelic variation in the T1R3 gene between the WT strains. Nevertheless, residual sensitivity by the T1R1+T1R3 KO mice demonstrates that a T1R-independent mechanism can contribute to the detectability of high concentrations of this prototypical umami compound stimulus.

INTRODUCTION

The presence of l-amino acids in ingested foods and fluids is thought to stimulate the gustatory system and ultimately promote feeding. In particular, monosodium glutamate (MSG), the salt form of l-glutamic acid, a nonessential amino acid, is used as a food additive and elicits a “savory” quality in humans referred to as “umami” that is amplified when l-glutamate is combined with 5′-ribonucleotides, such as inosine 5′-monophosphate (IMP).

In some mammals, including rodents and humans, the taste of l-glutamate is thought to be primarily mediated through the heterodimer T1R1+T1R3 (21, 25). This G protein-coupled receptor is mostly expressed in taste buds of the anterior tongue, innervated by the chorda tympani nerve (CT). In rats, the CT responds to MSG and also shows synergistic amplification of this signal when mixed with IMP (40). The existence of knockout (KO) mice lacking either receptor subunit has allowed testing for the necessity of this receptor to signal the presence of l-glutamate, and indeed, knockout mice display reduced or abolished CT responses to l-glutamate and an absence of the hallmark synergism between MSG and IMP (e.g., 8, 19, 20, 25, 41).

Behavioral work has likewise shown that mice lacking either T1R1 or T1R3 show impairment in some tasks once the sodium taste component of MSG is reduced with amiloride (6, 13, 33), which is an epithelial sodium channel blocker tasteless to rodents (12, 23, 24). Mice lacking either subunit show no enhancement of unconditioned licking responses to MSG by IMP (in amiloride) in brief-access taste tests designed to minimize the impact of postingestive influences on behavior (43); double-KO mice lacking both subunits T1R1 and T1R3 show no concentration-dependent licking responses to even higher concentrations (5). However, mice lacking T1R1 are able to learn and express a conditioned taste aversion to MSG + IMP mixtures and mice lacking either T1R1 or T1R3 can detect high concentrations of MSG + IMP (in amiloride) in a taste sensitivity task, albeit impaired relative to wild-type mice (35; but see Ref. 9). Double-KO mice show similar levels of appetitive (approach) behavior toward MSG+IMP mixtures in brief-access and progressive ratio tests, relative to wild-type mice (5). Together, these studies indicate that receptors other than the T1Rs are capable of supporting some taste-related behavioral functions for these stimuli.

Indeed, other receptors in the oral cavity, in particular, taste-variants of metabotropic glutamate receptors subtypes 1 and 4 (mGluR1, mGluR4; e.g., Refs. 7 and 32), have been identified and found to respond to MSG. Calcium-imaging studies in isolated taste receptor cells have revealed that mGluR1 and mGluR4 respond to monopotassium glutamate (MPG) and IMP (29, 30). The receptors mGluR1 and mGluR4 are primarily expressed in taste buds of the posterior tongue, innervated by the glossopharyngeal nerve (GL); the GL does not show a synergistic response to l-glutamate mixed with a 5′-ribonucleotide, although it does respond to both compounds individually (7, 8, 20; but see Ref. 27). Mice lacking mGluR4 show reduced CT and GL responses to MPG, with and without IMP (42). Thus, it may be that these receptors are capable of supporting the residual behavior seen in other studies.

Besides the possible role of other receptor types, it is important to note that the psychophysical work demonstrating residual sensitivity to MSG mixed with IMP and amiloride has so far been conducted in mice lacking only one subunit of the T1R1+T1R3 receptor. The possibility that the remaining subunit may be forming a functional homodimer cannot be dismissed. Indeed, some taste receptor cells in mice have been found to express each T1R subunit alone (18). In addition, although there are limited data available using amino acids, there are reports of partial functionality of the T1R3 subunit in response to sugars (37, 43). Thus, while not explicitly tested in prior behavioral studies involving single-subunit KO mice, it may be that some of the glutamate responses are also being signaled through the remaining T1R1 or T1R3 subunit.

Accordingly, we tested mice lacking both subunits of the T1R1+T1R3 receptor in a two-response operant taste discrimination task to determine the necessity of the receptor for normal sensitivity to MSG+IMP mixtures in amiloride. In addition, it is significant that these mice have a mixed C57BL/6J (B6) and 129X1/SvJ (129) background. These strains exhibit allelic variation of the T1R3 receptor subunit. This variation reduces sugar sensitivity and preference in the 129 strain (e.g., 3, 11, 15, 22, 26). Moreover, there is some evidence for a reduced preference for MSG, although this is not apparent in all behavioral tests (3, 15, 16, 24). Little information is available using 129 mice in a task separating discriminative sensitivity to the stimulus from its hedonic value. Thus, in the current study, the performance of B6 and 129 wild-type mice, the background strains used to generate the double-KO mice (25), was also evaluated.

MATERIALS AND METHODS

Subjects

Mice used in this study were C57BL/6J (B6, n = 16; 8 males and 8 females, ages 12–14 wk), 129X1/SvJ (129, n = 16; 8 males and 8 females, ages 11–13 wk), knockout mice missing both subunits of the T1R1+T1R3 receptor (T1R1+T1R3, n = 16; 8 males and 8 females, ages 11–20 wk), and knockout mice missing both subunits of the T1R2+T1R3 receptor (T1R2+T1R3, n = 16; 9 males and 7 females ages 11–22 wk). B6 and 129 mice were purchased from Jackson Laboratories (Bar Harbor, ME).

To generate the knockout mice, male and female breeding pairs of mice that were homozygous null for the Tas1r1, Tas1r2, or Tas1r3 gene (initially derived from 129 backcrossed with B6 mice) were generously provided by Dr. Charles Zuker (University of California, San Diego; now at Columbia University). An additional backcross with B6 mice (Jackson Laboratory) was done with the donated breeders. From the resulting mice, homozygous-null mice for the Tas1r1 or Tas1r3 were paired together to generate mice heterozygous for both Tas1r1 and Tas1r3. This subsequent generation was, in turn, paired to eventually generate mice heterozygous, homozygous null, and wild-type for Tas1r1 and/or Tas1r3. Mice that were homozygous-null for both Tas1r1 and Tas1r3 were paired to generate more animals that were homozygous-null for both genes. A similar breeding strategy was used to generate mice lacking both Tas1r2 and Tas1r3. SNP genome scanning analysis of known polymorphisms between B6 and 129 strains (completed by Jackson Laboratories) indicated that the double-KO mice have ~20–30% contribution from the 129 strain, with the remaining contribution from the B6 strain. Genotypes for the KO mice were independently confirmed after the experiment by Transnetyx, using qPCR to detect the presence of the WT allele or neocassette for both genes.

Animals were housed singly throughout the experiment in polycarbonate shoebox cages in a room with computer-controlled temperature, humidity, and a 12:12-h light-dark cycle. A cotton-fiber nestlet (Ancare, Bellmore, NY) was provided for environmental enrichment. Testing occurred during the light phase. Mice were provided ad libitum chow (Rodent Laboratory Chow 5001; Nestlé Purina Petcare, St. Louis, MO). Deionized reverse-osmosis water was available on weekends and as reinforcement in the detection task, as described below.

Body mass and body condition score (39) were measured when water restriction occurred and on all training and test days. Sessions were typically run Monday to Friday. Water bottles were removed the afternoon before the first training or test day of the week and returned following the last session of the week. Any animal falling below 85% of its ad libitum body mass or with a decrease in body condition score was provided 1–2 ml supplemental water at least 30 min after the end of the daily session. When water repletion occurred, water bottles were returned to the cage at least 30 min after the last daily session. All procedures were approved by the Animal Care and Use Committee at the Florida State University.

Taste Stimuli

All solutions were prepared daily with deionized reverse-osmosis water and presented at room temperature. Test stimuli consisted of reagent-grade NaCl (Macron), Maltrin-580 (Grain Processing, Muscatine, IA), monosodium glutamate (MSG; Sigma-Aldrich), inosine 5′-monophosphate disodium (I; Sigma-Aldrich), and amiloride hydrochloride (A; Sigma-Aldrich). When amiloride was used, it was dissolved in deionized water and left in a covered flask to stir overnight before use. All other chemicals were prepared daily, immediately before the first session.

Apparatus

Training and testing took place in a gustometer, as described in detail elsewhere (36). A mouse was placed in the rectangular testing chamber that consisted of three Plexiglas sides and a stainless-steel front panel with three access slots. The centrally positioned access slot allowed the mouse to access fluid deposited upon a borosilicate sample ball, which spins around a horizontal axis. Licks were registered by a force transducer connected to the sample ball; fluid was deposited onto the sample ball via Teflon tubing connected to a syringe that was mounted to a stepper motor. Each stimulus presented in a session was delivered by a different syringe through isolated tubing that was threaded through a circular turret that rotated to position the appropriate tubing before the start of a trial. A fixed polyoxymethylene reinforcement ball was stationed behind the remaining access slots, one on either side of the sample ball. Fluid was delivered by a pump via PTFE (polytetrafluoroethylene) tubing connected to a stainless-steel stub threaded through the ball, and licks were registered via force transducer in the same way as for the sample ball. The availability of any of these balls could be blocked with a stainless-steel shutter, as was done during some parts of training (described below in NaCl detection training.).

The testing chamber was housed within a sound attenuation enclosure, and masking noise was presented during all sessions. Air was drawn away from the sample ball via ductwork connected to an exhaust fan to reduce olfactory cues.

Two-Response Operant Detection Task

Trial structure.

We used a two-response operant task to determine the detectability of stimuli. This task requires a “thirsty” animal to sample a small volume (~1 µl) of fluid and determine how to respond on the basis of which stimulus has been presented to receive a water reward.

At the start of the trial, the clean sample ball was positioned in front of its access slot. The trial began when the mouse licked twice within 250 ms to ensure it was prepared for active sampling, following which a preload (~10 µl) was deposited onto the sample ball. Each subsequent lick delivered an additional small sample (~1–1.6 µl, depending on phase; see below in Maltrin detection training). After the mouse had taken all sample licks (5–7, depending on phase), or 5 s had passed, the sample ball was retracted and the mouse had to respond on one of the reinforcement balls within 5 s (referred to as limited hold). If the animal responded correctly, it received a water reward (~1 µl/lick, 15–20 licks depending on phase). If it responded incorrectly or it failed to respond before the end of the limited hold, it did not receive fluid and instead was given a time-out of 30 s before the start of the next trial. After the reward or time-out was an intertrial interval (~7 s) during which the sample ball was washed with deionized water, dried with pressurized air, and then positioned for the animal to start the next trial.

NaCl detection training.

The first stimulus tested was NaCl. This served as a control to allow the animals to learn the task before being tested with a stimulus more difficult to detect. A summary of experimental phases is provided in Table 1.

Table 1.

Schedule of training and testing phases

Phase	Days	Stimuli	Schedule
NaCl Detection
Stationary training	4	Water	Constant
Side training	6	0.6 M NaCl or water	Constant
Alternation	4	0.6 M NaCl and water	Alternated after x correct
Random training I	24	0.6 M NaCl and water	Semirandom*
Random training II	4	0.6 M NaCl and water	Semirandom
Random training III	3	0.6, 0.2, 0.1, and 0.05 M NaCl and water	Semirandom
NaCl testing	20	4 concentrations NaCl and water	Semirandom
Maltrin Training
Side	2	32% Maltrin or water	Constant
Random	10/12†	32% Maltrin and water	Semirandom
MSG+I+A Training‡
Side	2	0.6 M MSG+I+A or water	Constant
Random training I	38	0.6 M MSG+I+A and water	Semirandom
Random training II	6	0.6, 0.5, 0.4, 0.3 M MSG+I+A and water	Semirandom
MSG+I+A testing	21	4 concentrations MSG+I+A and water	Semirandom

MSG+I+A, monosodium glutamate (MSG) + inosine 5′-monophosphate (IMP) in amiloride (A).

^*Stimuli were presented in randomized blocks without replacement, using three or four tubes each of the stimulus and water.

^†After 10 sessions, the sample was increased to 7 licks at ~1.6 µl/lick, and the reinforcement was increased to 20 licks at ~1.0 µl/lick.

^‡The sample was returned to 5 licks at ~1.0 µl/lick, and the reinforcement was decreased to 15 licks at ~1. 0 µl/lick.

Mice were first familiarized to the apparatus and the sources of fluid with stationary training. In each session, one ball was accessible through the slot, while the other two were blocked with a stainless-steel shutter. For the entire 25-min session, mice could freely lick the available ball and receive fluid (~1 µl/lick). After each ball had been available for one session, the sample ball was available a second time, on the fourth day of stationary training. The next phase of training was side training, during which, the taste sample solutions were paired with one of the reinforcement spouts. During each session, mice received either 0.6 M NaCl or water from the stimulus ball and then had to respond to the reinforcement ball associated with that stimulus. The other stimulus and the associated reinforcement ball were inaccessible during the session. Side assignments were counterbalanced within genotype; approximately half of the mice had NaCl assigned to the left reinforcement ball, while the other half had NaCl assigned to the right. The limited hold (time allowed to respond) was 180 s, following a sample of 5 licks of ~1 µl/lick. Upon responding to the reinforcement ball, mice received up to 15 licks (~1 µl/lick) within 10 s. It was not possible to respond incorrectly in this phase, and the mice were not given a time-out for failing to respond. Each stimulus/side pairing was presented for three sessions each, after which alternation training began.

During alternation training, both stimuli were presented during the same session, and mice had to respond correctly to the associated reinforcement ball to receive fluid. The same stimulus was presented repeatedly until the mouse reached the criterion of nonconsecutive correct responses. The criterion decreased across four sessions (criteria: 8, 6, 4, 2). Once the criterion was met, the mouse would receive the other stimulus until reaching the same criterion. Stimulus presentations alternated in this way for the duration of the session. The limited hold was reduced to 15 s, and a time-out of 5 s was introduced for incorrect responses.

During the final phases of training, stimuli were presented in randomized blocks of 6–8 trials (3 or 4 each of NaCl and water). The probability of receiving NaCl as the stimulus was P = 0.5. In random training I, the training parameters remained the same as for alternation, except that stimuli were randomly presented. However, some mice displayed side biases or stopped responding after taking a sample; the limited hold was increased to 180 s for these mice, and they were given additional side training for one session with each reinforcement ball before being returned to random training I. This occurred up to three times. The remaining mice were given random training throughout this phase. In random training II, the time-out was increased to 30 s, and the limited hold was decreased to 5 s. In random training III, mice were trained using the same session parameters as for random training II, but additional NaCl concentrations were tested within a single session. The concentrations presented were 0.6, 0.2, 0.1, and 0.05 M. These concentrations were used as the standard array in NaCl testing and were tested with four tubes of water in randomized order without replacement (block size: eight tubes). The probability of receiving NaCl as a stimulus remained P = 0.5.

NaCl detection testing.

Once all mice were performing ≥80% correct overall on trials with their responses, NaCl detection testing commenced. Typically, testing occurred Monday through Friday. On the first day of testing each week, the standard array was presented. During the remaining days of testing in that week, four concentrations of NaCl and water were presented as stimuli. A clearly detectable concentration was always included in the array, but the concentrations tested each week systematically decreased until all animals reached a sufficiently low performance level (as described in Data Analysis). The concentrations of NaCl tested ranged from 0.025 to 0.6 M.

Maltrin detection training.

After the end of NaCl testing, the animals were trained to detect Maltrin-580 in water. Maltrin-580 is a maltodextrin that is similar to Polycose, which has been found to be detectable by mice lacking subunits of the T1R2+T1R3 receptor (35a, 38) and is thought to be signaled via an independent, currently unknown, receptor (see 34).

An abbreviated training schedule was followed to familiarize the animals with the new stimulus. Session parameters were not changed from NaCl testing, since the animals were already familiar with the task. The training concentration was 32% Maltrin. The side assignments for stimuli remained the same, with Maltrin replacing NaCl in the session. Two sessions of side training, one for each reinforcement ball were given first. Random training then commenced. Some mice had low performance, so on day 11 of random training, the total sample and reinforcement volumes were increased (see Table 1). This has been shown to slightly increase performance to high concentrations of Polycose (38). After an additional 12 sessions, a number of the T1R2+T1R3 group failed to reach a performance above chance (50%) with no evidence of improvement, so Maltrin training was ended. It is important to note, also, that some of these animals were showing some difficulty with NaCl, indicating that the same animals were struggling with the task, in general.

MSG+I+A detection training.

The stimulus was changed to a mixture of MSG and IMP in amiloride, an epithelial sodium channel blocker that reduces the sodium signals arising from MSG without itself being detectable to rodents (10, 12). The taste of sodium seems an important part of the taste quality of MSG for mice, and reducing its contribution is important when investigating the role of the other components of the stimulus (24, 28). This mixture of the salt form of l-glutamate and the 5′-ribonucleotide IMP is a stimulus thought to stimulate multiple receptor types in the oral cavity, in particular, the T1R1+T1R3 receptor, and elicits a taste quality referred to as “umami” in humans. The training concentrations of this mixture were 0.6 M MSG + 2.5 mM IMP, in 100 µM amiloride (MSG+I+A), with 100 µM amiloride replacing water as both a comparison stimulus and for reinforcement. An abbreviated training schedule commenced (Table 1) using the same session parameters as for NaCl testing. MSG+I+A was assigned to the same reinforcement ball as for NaCl testing. The schedule began with two sessions of side training, one for each reinforcement ball, using the original sampling parameters of five licks at ~1 µl per lick and 15 reinforcement licks. The animals were then moved to MSG+I+A random training. During this phase, if animals were displaying side biases, they were given additional side training and then returned to random training.

After 38 training sessions, mice that were not performing above chance levels (50%) were removed from the study. Because of the low number of T1R2+T1R3 mice under stimulus control at this point, along with the results of the NaCl testing and the difficulties with training Maltrin, the entire group was removed from study. Remaining mice were moved to MSG+I+A random training II with multiple concentrations of the stimulus. To reduce the stimulus intensity, the stock concentration (0.6 M MSG + 2.5 mM IMP in 100 µM amiloride) was diluted with 100 µM amiloride, allowing for identical dilution of both MSG and IMP for each testing concentration. The standard array resulted in MSG concentrations of 0.6, 0.5, 0.4, and 0.3 M MSG with similarly diluted concentrations of IMP from the stock concentration of 2.5 mM. These were tested along with four tubes of 100 µM amiloride, in randomized order without replacement in blocks of eight tubes. The diluent was always 100 µM amiloride. For the sake of simplicity, only the MSG concentration is reported in text.

MSG+I+A testing.

After six training sessions with the standard array, MSG+I+A testing began. As with NaCl, the standard array was always presented during the first session of the week, and the remaining test days for the week included concentrations that were subsequently lowered across weeks. This continued until all animals had reached a sufficiently low performance level to generate psychometric functions from the data. The concentrations tested ranged from 0.0125 M to 0.6 M MSG+I+A.

After the last testing day for MSG+I+A, a water control test was performed in which eight tubes were filled with water. Four tubes were assigned to the MSG+I+A reinforcement ball, while four tubes were assigned to the water/amiloride reinforcement ball. This test is designed to confirm that the mice were using chemical cues to guide performance, as it is expected that mice will perform at chance levels. After the water test, mice were euthanized and tail samples were obtained to send to Transnetyx to confirm genotype.

Data Analysis

For NaCl and MSG+I+A, the proportion of correct responses was calculated from trials with a response for each concentration and was adjusted for the false alarm rate using the following equation: P(Hit)_c = [P(Hit) – P(FA)]/[1 – P(FA)], where P(Hit)_c is the corrected hit rate (CHR), P(Hit) is the proportion of correct responses for the stimulus, and P(FA) is the proportion of incorrect responses to water (or amiloride). Separate corrections were made for each concentration of each stimulus. Data from all mice completing the study were used in mixed-design ANOVAs with repeated measures for one factor (concentration) to compare performance across genotypes.

Psychometric functions were derived from the CHR values of mice performing above-chance levels in the standard array using the following logistic function:

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

where a is  asymptotic performance, b is slope, x is molar concentration, and c is the log₁₀ molar concentration at ½ asymptote, or the EC₅₀, operationally defined here as the detection threshold. For three T1R1+T1R3 mice, curves could not be fit to their performance, and they were not included in analyses of curve parameters (but were included in the ANOVAs of performance across concentrations). Curve parameters were compared using one-way ANOVAs and Studentʼs t-tests. Performance on the water control test was compared with chance (50% overall) using a binomial probability distribution. For all statistical tests, significance was considered to be P < 0.05.

Only mice completing the entire study were included in analyses. Five mice died during testing (three B6 and two 129). Four mice from the 129 strain, two mice from the B6 group, and three T1R1+T1R3 KO mice did not perform above chance levels to the highest MSG+I+A concentrations after 38 training sessions and were removed from study before the start of testing. All T1R2+T1R3 mice were removed from the study due to low numbers (five of the 16 starting the experiment, see above) learning the discrimination. This resulted in the following group sizes: B6, n = 10; 129, n = 12; T1R1+T1R3: n = 13 (n = 10 for curve analyses).

RESULTS

NaCl Detection and Maltrin Training

Two-way ANOVAs indicated an effect of concentration but no significant effect of genotype or genotype × concentration interaction for the corrected hit rates to NaCl (Fig. 1; Table 2). Thus, T1R1+T1R3 KO mice were able to perform in this task with similar competence and sensitivity as B6 and 129 mice. One-way ANOVAs comparing curve parameters also indicated no differences between genotypes for NaCl detectability (Table 3). These data add support that T1R1+T1R3 KO mice are capable of performing this detection task at similar levels to WT mice when presented with a stimulus independent of T1R receptor mechanisms.

Fig. 1.

Mean (±SE) corrected hit rates to NaCl in water for 129X1/SvJ (129), C57BL/6J (B6), and T1R1+T1R3 knockout (R1/3) mice and the associated psychometric functions. Data included are for mice completing the monosodium glutamate (MSG) + inosine 5′-monophosphate (IMP), in amiloride (A) (MSG+I+A) testing phase. Statistics for these data are reported in Table 2.

Table 2.

Two-way ANOVAs of corrected hit rate for NaCl and MSG+I+A detection testing

	NaCl	MSG+I+A
Genotype	F (2,32) = 1.21; ;P = 0.31	F (2,32) = 15.07; P < 0.01
Concentration	F (7,224) = 166.37; P < 0.01	F (8,256) = 62.59; P < 0.01
Genotype × concentration	F (14,224) = 1.59; P = 0.08	F (16,256) = 2.40; P < 0.01

MSG+I+A, monosodium glutamate (MSG) + inosine 5′-monophosphate (IMP) in amiloride (A). Bolded values indicate statistical significance (P < 0.05).

Table 3.

Mean (±SE) curve parameters (a, b, and c) and one-way ANOVAs comparing genotypes for NaCl and MSG+I+A detection

	B6	129	T1R1+T1R3	ANOVA
NaCl
a	0.94 (0.03)	0.97 (0.01)	0.93 (0.02)	F (2,29) = 0.90; P = 0.42
b	−5.05 (3.42)	−1.25 (0.16)	−4.03 (2.53)	F (2,29) = 0.63; P = 0.54
c	−1.42 (0.04)	−1.55 (0.13)	−1.69 (0.07)	F (2,29) = 2.38; P = 0.11
MSG+I+A
a	0.84 (0.04)	0.88 (0.36)	0.91 (0.04)	F (2,29) = 0.60; P = 0.56
b	−7.38 (3.76)	−2.57 (1.09)	−6.18 (4.47)	F (2,29) = 0.48; P = 0.62
c	−1.57 (0.08)	−1.55 (0.09)	−0.62 (0.13)	F (2,29) = 22.99; P < 0.01

B6, C57BL/6J mice;129, 129X1/SvJ mice. Bolded values indicate statistical significance (P < 0.05).

Similarly, the T1R1+T1R3 KO mice were able to learn the Maltrin discrimination to similar levels as either WT strain. An ANOVA comparing performance to 32% Maltrin on the last training day revealed no effect of genotype [F(2,29) = 0.48, P > 0.62] among the T1R1+T1R3 KO, B6, and 129 mice, and all animals in those strains were performing above 80% overall [B6: 90.19% (±1.70); 129: 91.93% (±2.32); T1R1+T1R3 KO: 87.19%(±2.94)]. However, since multiple concentrations of this stimulus were not tested, it is not possible to report on liminal sensitivity of these groups to the maltodextrin.

It is unclear why so many T1R2+T1R3 mice, unlike the other strains, struggled with detecting Maltrin [average performance on the last training day with 32% Maltrin: 67.34% (±6.45)], when proportionally more were able to perform the task with Polycose in previous work (35a). As noted in that study, there seems to be some variability in maltodextrin sensitivity unmasked by the removal of T1R2+T1R3 input. It is possible that most of the mice chosen for this experiment were carrying alleles that result in subsensitivity to maltodextrins. The receptor responsible for signaling the presence of such taste compounds remains to be identified, however, making it impossible to determine. Also, the KO mice were generated on a mixed background of B6 and 129 (26) and perhaps a particular genetic combination from these two sets of mice are interacting to negatively impact performance on this task. Another possible explanation would be that the stimulus used here, Maltrin-580, may not be as effective a stimulus for the unidentified polysaccharide receptor; Polycose was used previously (35, 35a).

MSG+I+A Detection

Two-way ANOVAs of corrected hit rates for all animals completing the study showed a significant effect of genotype, concentration, and a significant interaction (Table 2). Further analyses revealed no difference between B6 and 129 mice but a significant effect of concentration (Table 4). The T1R1+T1R3 KO mice performed significantly poorer to MSG+I+A than both B6 and 129 strains, with significant effects of genotype, concentration, and significant interactions in comparing the KO group to either WT strain (Table 4).

Table 4.

Two-way ANOVAs between genotypes for MSG+I+A

	B6/129	B6/T1R1+T1R3	129/T1R1+T1R3
Genotype	F (1,20) = 0.03, P = 0.86	F (1,21) = 29.43, P < 0.01	F (1,23) = 22.3, P < 0.01
Concentration	F (8,160) = 49.42, P < 0.01	F (8,168) = 45.54, P < 0.01	F (8,184) = 35.59, P < 0.01
Genotype × concentration	F (8,160) = 0.54, P = 0.83	F (8,168) = 3.70, P < 0.01	F (8,164) = 3.13, P < 0.01

B6, C57BL/6J mice;129, 129X1/SvJ mice; MSG+I+A, monosodium glutamate (MSG) + inosine 5′-monophosphate (IMP) in amiloride (A). Bolded values indicate statistical significance (P < 0.05).

Psychometric functions to MSG+I+A performance could not be fit to three mice in the T1R1+T1R3-KO group, despite meeting the performance criterion (i.e., >50% on the training array) to be included in the testing phase. One-way ANOVAs comparing curve parameters of the remaining 10 KO mice with those of WT mice demonstrated no effect on asymptote or slope of psychometric functions between genotypes (Fig. 2). In contrast, there was a significant effect on the EC₅₀, with KO mice being impaired relative to either WT strain (c-value; Fig. 3; Table 3). The comparison of genotypes via Studentʼs t-tests indicated no difference between c-value for B6 and 129 mice; T1R1+T1R3 KO mice, however, had significantly higher detection thresholds than both wild-type strains, indicating a lower sensitivity to the umami mixture (Fig. 3).

Fig. 2.

Mean (±SE) corrected hit rates to monosodium glutamate (MSG) + inosine 5′-monophosphate (IMP), in amiloride (A) (MSG+I+A) for 129X1/SvJ (129), C57BL/6J (B6), and T1R1+T1R3 knockout (R1+3) mice and the associated psychometric functions. Data included are of mice that completed the MSG+I+A testing phase and for which curves could be fit. Corrected hit rates are plotted relative to MSG concentration (M). IMP concentrations were diluted from 2.5 mM IMP to match the dilution of MSG concentration from 0.6 M MSG. Statistics for these data are reported in Tables 2 and 4.

Fig. 3.

Individual c-values by phase. The c-value of each animal completing the monosodium glutamate (MSG) + inosine 5′-monophosphate (IMP), in amiloride (A) (MSG+I+A) phase and for which curves could be fit are plotted by genotype for NaCl and MSG+I+A. Mean values for each genotype are indicated by horizontal lines. Mean (±SE) values are reported with statistics in Table 3.

Water Control Test

Water control test results did not differ statistically from chance (50%) as compared with a binomial probability distribution for any mouse, indicating that performance during detection testing was under the control of chemical cues from the stimuli. Mean (±SE) performance for each group was as follows: B6, 52.7% (1.6); 129, 50.6% (1.9); and T1R1+T1R3, 51.8% (1.4).

DISCUSSION

Overall, the clear difference between the T1R1+T1R3 KO group compared with either B6 or 129 mice (Fig. 2) demonstrates that the heterodimer is necessary for normal detectability of the amino acid MSG mixed with the 5′-ribonucleotide IMP, a prototypical umami stimulus, when the sodium component is minimized by amiloride. These results extend previous work showing differences between single-KO and WT mice of a similar mixed background (35). Our results here demonstrate that the residual responding to the MSG mixture by the single KO mice in that study was not dependent on the presence of the remaining T1R subunit. Although it has not been demonstrated that any T1R receptor subunit, isolated from the other T1R proteins, provides a signal when stimulated by amino acids, some responses to high concentrations of sucrose have been observed for T1R3 in a heterologous expression system when neither T1R1 nor T1R2 are also expressed in the cell (43). However, the double-KO mice used here lacked both receptor subunits, ruling out the necessity of either T1R1 or T1R3 forming a homodimer or a heteromer with some other protein in the detection of the higher concentrations.

The previous study with single-KO mice focused on the detectability of l-glutamate in the umami mixture by varying MSG concentration in a constant concentration (2.5 mM) of IMP and amiloride. Although the single-KO mice could not detect the IMP concentration used, WT mice could (35). Thus, an MSG concentration-dependent psychometric function could not be derived for the WT mice, precluding an explicit comparison of the sensitivity of the respective genotypes to l-glutamate in the presence of IMP. Here, both MSG and IMP decreased across stimuli tested, resulting in concentration-dependent responses for all three strains (Fig. 2) and, thus, the necessity of the T1R1+T1R3 receptor in the detection of the mixture could be assessed relative to WT mice. Still, the double-KO mice were demonstrably impaired relative to either WT strain (Fig. 2), providing more direct evidence that the T1R1+T1R3 receptor is necessary for the normal detectability of umami stimuli.

What is still unclear, however, is what aspect of the stimulus drives responding to MSG+I+A. Smith and Spector (35) found that WT mice could not discriminate MSG (+amiloride) from amiloride without the addition of IMP. However, because WT mice were also clearly able to discriminate 2.5 mM IMP (+amiloride) from amiloride, it was possible that the WT responses to the MSG+I+A mixture in that study were due to the IMP component. It is possible that the psychophysical curve derived for the WT mice in this study was a function of the IMP rather than the MSG concentration. A separate psychometric function would need to be derived with IMP alone to determine to what extent the mixture of MSG and IMP differs from the single component. On the other hand, Smith and Spector (35) found that T1R1 and T1R3 single-KO mice were unable to discriminate either component of the mixture from amiloride. Indeed, the lack of sensitivity to either MSG or IMP alone (in amiloride) when missing an intact T1R1+T1R3 receptor is why those stimuli were not included for comparison in this study. That is, the KO mice do not detect the presence of either component alone but can detect the presence of the mixture at sufficiently high concentrations, making the detectability of the mixture and the concentration dependence seen here and previously (35) all the more extraordinary.

The mechanism by which some functionality is maintained in T1R1+T1R3 KO mice remains to be explained. Cellular and peripheral nerve responses in mice missing either T1R1 or T1R3 show reduced or abolished signals to MSG and 5′-ribonucleotides and do not display any synergistic effect when the two types of compounds are mixed (8, 19, 20, 25). Thus, it is unlikely that the two components are combining synergistically (a hallmark characteristic of an umami stimulus) to generate a detectable taste perception. However, as noted, there are some weak taste nerve responses that have been reported for both MSG and IMP, albeit without synergy (8, 20, 25; but see 27). Indeed, other receptors that bind with l-glutamate are expressed in taste buds, particularly, mGluR1 and mGluR4 (7, 29). The effectiveness of mGluR1 and mGluR4 antagonists to attenuate responses to l-glutamate in T1R1- or T1R3-KO mice suggests a role for these receptors in signaling the oral presence of this amino acid (20, 30, 42). It is possible that the total activity arising from stimulation of mGluRs by the highest concentrations of MSG and IMP tested here provides a supraliminal signal for the T1R1+T1R3 KO mice, despite neither component seeming to be detectable alone (35). However, the mGluR antagonists that have been tested do not abolish all responding to the umami stimuli, indicating the possibility of yet other receptor types contributing to the perception of MSG+I+A.

Although it is possible to include the antagonists for mGluRs as part of the stimulus in the detection task here to determine the role of the receptors in the residual sensitivity by KO mice, it would first be necessary to determine whether the drugs are, themselves, detectable. For example, amiloride was confirmed to be virtually tasteless to both rats and mice (12, 23, 24), making it a perfect pharmacological tool for the behavioral assessment of the contribution of the epithelial sodium channel in salt taste. If, unlike amiloride, the antagonists are detectable by mice, the task would become a discrimination of the umami stimulus in a chemical background, which is a considerably more difficult task than detecting a compound in water (or the tasteless background of amiloride).

One caveat of note with regard to the use of amiloride is that it only partially reduces the detectability of sodium (e.g., 6, 10, 13). This could mean that some of the sensitivity of the WT mice, as well as residual responding by KO mice, could be related to the amiloride-insensitive sodium signaling pathway. However, as noted previously, mice were unable to detect MSG in amiloride at similar concentrations to those used here (35), meaning that the simple presence of the sodium cation in solution with l-glutamate and IMP is not sufficient to drive responding in this task. The more parsimonious explanation is that l-glutamate, IMP, or the combination of the two chemicals is activating another receptor mechanism. The detectability of the mixture by KO mice might not be entirely based on the signals generated from a single type of peripheral taste receptor, however. It is also possible that a central circuit integrates the MSG and IMP signals arising from other receptor types, to allow at least some degree of sensitivity in the absence of the T1R1+T1R3 receptor. Another possibility is that mice are using a nontaste cue, such as smell, to detect the stimulus. Although the gustometer is designed to minimize olfactory cues via an exhaust system that draws air away from the animal (36), this does not rule out the contribution of retronasal stimulation. However, the fact that single KO mice can neither detect MSG+amiloride nor IMP at the concentrations used here (35), at least, suggests that the smell of either these compounds alone or a simple volatile contaminant is not responsible for the performance observed.

While the presence of a functional T1R1+T1R3 receptor is necessary for the normal detection of MSG+I+A, both B6 and 129 mice displayed similar sensitivity to the mixture (Figs. 2 and 3; Table 4). These strains of mice differ in allelic variation of the Tas1r3 gene that ultimately impacts sensitivity to sugars (11, 15, 22, 26, 31). However, although the T1R3 protein is a component of the T1R1+T1R3 heterodimer for l-amino acids, the limited data available suggest that taste-related behavioral responses to amino acids are not influenced by which isoform of the T1R3 protein is present. Whole nerve responses from the CT and GL of mouse substrains (C57BL/6BL/J and 129P3/J) different than those used here were similar for MSG, although the sodium component was not diminished for those preparations (14). Although B6 mice will display a higher preference for MSG with and without IMP in long-term intake tests (3, 4), work with F₂ hybrids of B6 and 129 mice suggests that MSG and IMP intake (in water) is independent of allelic variation of the Tas1r3 gene (15, 16). That said, a difference between these two strains in the detectability of the achiral amino acid glycine has been demonstrated in a task similar to that employed here (11). Some evidence suggests that some of the differences in glycine detectability between the strains are not due to the sweet “taster” status of the animal (i.e., T1R2+T1R3 receptor variant; 11, 15). As such, some of the strain difference in detectability may be related to the T1R1+T1R3 receptor, given that glycine activates both T1R2+T1R3 and T1R1+T1R3 receptors (21, 25), and so a difference in detectability of MSG+I+A would not have been wholly unexpected. However, at least as assessed in this task, sensitivity to MSG and IMP mixtures seems independent of the T1R3 receptor variant (as well as other genetic loci known to differ between the two strains).

Still, although both B6 and 129 show similar detectability for this mixture, it cannot be said that the two strains experience identical perceptions arising from the stimulus mixture or its components. Despite similar CT nerve responses to MSG, B6, and 129 mice differ in their responsiveness to the 5′-ribonucleotides IMP and GMP (14). When conditioned to avoid the taste of IMP and then tested for a generalization to other compounds, these mice do not generalize those aversions to the same chemicals, with 129 mice avoiding d-tryptophan and saccharin more than B6 mice. On the other hand, when conditioned to avoid MSG and tested with the same array of stimuli, B6 and 129 mice do generalize the aversion to the same compounds (24). Thus, the taste of IMP may influence behavior separately from MSG. As such, the mixed stimulus tested here is complex and possibly perceived differently by the two mouse strains, despite being similarly detectable. Indeed, a difference in taste quality perceived by these mice might explain the difference in preference and intake sometimes reported between these strains (1, 2, 14). However, it is important to note that these long-term intake tests are also potentially influenced by differences in postingestive events. For instance, B6 and 129 mice have been shown to metabolize both MSG and glycine differently, and such processes may also influence long-term behavioral outcomes (17). Other taste-guided behaviors in response to umami stimuli, separate from sensitivity, may differ between these strains just as other behaviors involving umami stimuli can be influenced by other receptor types. Indeed, while sensitivity was severely impaired with the loss of the T1R1+T1R3 receptor in this study, other work has suggested that remaining receptors are sufficient to maintain some types of taste-guided behavior related to the motivational properties of the amino acid and 5′-ribonucleotides (5, 20).

Taken together, these results provide further evidence that the taste of l-glutamate by rodents relies heavily on the T1R1+T1R3 receptor, but that other receptor mechanisms are capable of supporting at least some degree of responsiveness.

Perspectives and Significance

Many questions remain regarding how the gustatory system signals the presence of amino acids in general, and l-glutamate, in particular. With the identification of multiple receptor types that respond to this class of compounds, it is important to rigorously investigate the contribution of each receptor to taste-guided behaviors. Although the present results and past data supply ample evidence to suggest that the T1R1+T1R3 heterodimer is highly important for signaling the presence of l-glutamate and associated compounds such as 5′-ribonucleotides, the current findings also add to a growing literature that implicates other receptor mechanisms in supporting some degree of taste responsivity. In particular, the current findings extend previous work to demonstrate that although allelic variation in Tas1r3 does not have an impact, an intact T1R1+T1R3 receptor is necessary for normal concentration-dependent detectability of this “umami” compound in a task that is independent of the hedonic value of the stimulus. In contrast, other tasks that focus on the quality and hedonic value of the stimulus suggest that the T1R1+T1R3 receptor is not necessary to maintain all normal behavioral responses to l-glutamate and/or IMP. In a brief-access test, appetitive (approach) behavior seems unaffected in the KO mice, while consummatory behaviors are virtually eliminated (5). Also, a conditioned taste aversion task that focuses on the quality of the stimulus (and potentially also its hedonic value) was not severely impaired in mice lacking one receptor subunit (20). Thus, while the T1R1+T1R3 receptor provides critical input with regard to certain taste-guided behaviors, others are left intact following genetic deletion of the receptor. Future work will need to focus on the necessity of additional receptor types, such as mGluRs, in these and other taste-guided behaviors. Taken together with the results presented here, these studies indicate the possibility that signals arising from T1R-independent receptor mechanisms may be destined for divergent neural circuitries that subserve different taste-guided behaviors, an intriguing possibility that warrants further investigation.

GRANTS

This project was funded by the National Institute on Deafness and Other Communication Disorders Grant R01-DC004574 (to A. C. Spector).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.D.B., S.P.T., and A.C.S. conceived and designed research; G.D.B. performed experiments; G.D.B. analyzed data; G.D.B., S.P.T., and A.C.S. interpreted results of experiments; G.D.B. prepared figures; G.D.B. drafted manuscript; G.D.B., S.P.T., and A.C.S. edited and revised manuscript; G.D.B., S.P.T., and A.C.S. approved final version of manuscript.