Taste does not determine daily intake of dilute sugar solutions in mice

Abstract

When a rodent licks a sweet-tasting solution, taste circuits in the central nervous system that facilitate stimulus identification, motivate intake, and prepare the body for digestion are activated. Here, we asked whether taste also determines daily intake of sugar solutions in C57BL/6 mice. We tested several dilute concentrations of glucose (167, 250, and 333 mM) and fructose (167, 250, and 333 mM). In addition, we tested saccharin (38 mM), alone and in binary mixture with each of the sugar concentrations, to manipulate sweet taste intensity while holding caloric value constant. In experiment 1, we measured taste responsiveness to the sweetener solutions in two ways: chorda tympani nerve responses and short-term lick tests. For both measures, the mice exhibited the following relative magnitude of responsiveness: binary mixtures > saccharin > individual sugars. In experiment 2, we asked whether the taste measures reliably predicted daily intake of the sweetener solutions. No such relationship was observed. The glucose solutions elicited weak taste responses but high daily intakes, whereas the fructose solutions elicited weak taste responses and low daily intakes. On the other hand, the saccharin + glucose solutions elicited strong taste responses and high daily intakes, while the saccharin + fructose solutions elicited strong taste responses but low daily intakes. Overall, we found that 1) daily intake of the sweetener solutions varied independently of the magnitude of the taste responses and 2) the solutions containing glucose stimulated substantially higher daily intakes than did the solutions containing isomolar concentrations of fructose. Given prior work demonstrating greater postoral stimulation of feeding by glucose than fructose, we propose that the magnitude of postoral nutritive stimulation plays a more important role than does taste in determining daily intake of dilute sugar solutions.

carbohydrates are among the most abundant sources of energy in nature. They occur primarily in plant tissues, as polymers of glucose (e.g., starch) and as mono- and disaccharides (e.g., sugars). Even though starch is more common than sugars, most mammals preferentially seek out sugars (47). In humans, the sweet taste of sugar has been described as a supernormal reinforcer that is difficult to resist (11). Most other species of mammals also display a strong motivation to ingest sweet-tasting substances (i.e., have a “sweet tooth”) (47). For instance, rats will voluntarily leave a warm nest box containing food and water and run down a 16-m-long runway kept at −15°C to obtain a sweetened food or solution (10).

The sweet tooth is adaptive in the wild, because it motivates animals to locate and ingest energetically rich foods, which tend to occur sparsely and ephemerally. In modern human society, however, the sweet tooth is less adaptive, because there is a superabundance of calorically rich foods. This superabundance, together with the ability of these highly palatable foods to override the normal homeostatic controls of feeding, is thought to be an important factor contributing to the growing incidence of obesity (6, 34). Because the obesity epidemic is associated with increased intake of sugar-sweetened beverages (5, 9, 50), there is renewed interest in the factors that control sugar intake.

When a rodent licks a sugar solution, taste circuits in the central nervous system are activated. These circuits facilitate stimulus identification, prepare the body for digestion (66), and activate dopaminergic and endorphinergic reward systems in the brain that motivate further intake (21, 30, 43, 45). Once the sugar solution is swallowed, it stimulates viscerosensory mechanisms in the duodenum and jejunum (2, 22, 51, 69) that activate dopaminergic reward systems in the nucleus accumbens (19, 70) and cause potent reinforcement of feeding (2, 53, 58, 59). The relative contribution of taste vs. viscerosensation to daily sugar intake, however, is unclear.

Here, we examined the specific contribution of taste to daily intake of dilute sweet-tasting solutions. We limited the study to dilute sugar solutions (i.e., ≤333 mM), because high sugar concentrations inhibit intake by activating postoral satiety mechanisms (16, 18), thereby confounding attempts to draw causal connections between taste input and daily intake. We added a noncaloric sweetener (saccharin) to some of the experimental solutions to manipulate sweet taste intensity while holding caloric value constant. The hypothesis that taste determines daily intake of sweeteners is supported by the observations that 1) taste-mediated ingestive responses to caloric and noncaloric sweeteners increase with concentration (28, 42, 63) and 2) disruption of peripheral taste responses by surgical ablation of taste nerves (67, 72) or genetic ablation of taste transduction proteins (e.g., T1R3, α-gustducin, Trpm5, P2X₂, and P2X₃) (14, 15, 23, 61, 76) severely attenuates intake of and preference for caloric and noncaloric sweeteners.

If taste is the primary determinant of sweetener intake, then daily intake should increase with magnitude of the taste-mediated response. We describe two experiments that tested this prediction in C57BL/6J mice, a strain that consumes sugars avidly (4). In experiment 1, we assessed taste responsiveness to a battery of sweetener solutions in two ways: 1) peripheral taste responses from the chorda tympani (CT) nerve and 2) initial licking responses. In experiment 2, we asked whether the two measures of taste responsiveness reliably predicted daily intake of the sweetener solutions.

METHODS

Animals

We used C57BL/6J (B6) mice from Jackson Laboratories (Bar Harbor, ME). The animals were 7–10 wk old at the onset of testing; they were maintained in a temperature- and humidity-controlled vivarium, with a fixed 12:12-h light-dark cycle. Mice were housed in standard, polycarbonate shoebox cages (27.5 × 17 × 12.5 cm) with Bed-O'Cobs bedding (Andersons, Maumee, OH) and Nestlets (Ancare, Bellmore, NY). Prior to each experiment, the animals had ad libitum access to laboratory chow (Rat Diet 5012, PMI Nutrition, Brentwood, MO) and water. All animals were naïve to the sweetener solutions prior to testing. Separate sets of mice were used in each of the experiments. All procedures were approved by the Institutional Animal Care and Use Committee of Columbia University and were performed in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals.

Test Solutions

We tested three concentrations of glucose [167, 250, and 333 mM (G₁, G₂, and G₃, respectively)], three concentrations of fructose [167, 250, and 333 mM (F₁, F₂ and F₃, respectively)], one concentration of sodium saccharin [38 mM (S)], and binary mixtures of saccharin + each of the glucose and fructose concentrations (e.g., S + 167 mM glucose = SG₁; Sigma-Aldrich, St. Louis, MO). (See Table 1 for abbreviations for each sweetener solution.) We selected the three dilute concentrations of glucose and fructose (i.e., ≤333 mM), because B6 mice show concentration-dependent increases in daily intake of these solutions (3). We selected 38 mM saccharin, because it is highly preferred by B6 mice (55) and was reported to have synergistic effects on daily intake when mixed with G₁ in B6 mice (12). We dissolved all sweeteners in deionized water and tested them at room temperature.

Table 1.

Composition of the sweetener solutions and their abbreviations

Type of Solution	Sweetener 1	Sweetener 2	Abbreviation
Single-component	167mMfructose		F1
	250mMfructose		F2
	333mMfructose		F3
	167mMglucose		G1
	250mMglucose		G2
	333mMglucose		G3
	38mMsaccharin		S
Binary mixture	38mMsaccharin	167 mM fructose	SF1
	38mMsaccharin	250 mM fructose	SF2
	38mMsaccharin	333 mM fructose	SF3
	38mMsaccharin	167 mM glucose	SG1
	38mMsaccharin	250 mM glucose	SG2
	38mMsaccharin	333 mM glucose	SG3

Experiment 1: Measuring Taste Responses

We used two measures of taste responsiveness. 1) We recorded responses of the CT nerve to lingual stimulation with the sweetener solutions. The CT nerve, which relays signals from taste cells in the front of the tongue to the brain, responds strongly to sweeteners (33) and plays a central role in stimulating intake of sweeteners (72). 2) We recorded initial licking responses to the sweetener solutions. Because these licking responses occurred when the stimulus first contacted the orosensory receptors (29) and because mice consumed miniscule volumes of each sweetener during each 5-s trial, the lick test minimized the contribution of postingestive (negative or positive) factors to the ingestive response (17, 64).

CT nerve recordings.

We used a previously described technique (13, 78) to record from the CT nerve as it passed through the middle ear cavity. The mouse was anesthetized with 2–4% isoflurane (Butler Schein, Albany, NY), which was delivered through 1) an induction chamber during the initial knockdown, 2) a nose cone during the tracheotomy, and 3) a tracheal cannula during the ear surgery and CT nerve recordings. To access the CT nerve, we secured the mouse in a nontraumatic head holder, excised the pinna and ear canal, and lacerated the anterior-dorsal region of the tympanic membrane. Under a dissecting microscope (at ×60 magnification), the CT nerve could be seen traversing the gap between the anterior tympanic spine and malleus head (74). All recordings were made from this section of nerve. Throughout the surgery and nerve recordings, the mouse was maintained on a thermostat-controlled circulating-water heating pad set at 37°C (HTP-1500, Adroit Medical Systems, Loudon, TN).

Because the sheath of the CT nerve was thin, we could record robust responses simply by contacting the intact nerve with a sharpened tungsten electrode. The indifferent electrode was shunted to the ground electrode. The neural response was amplified 10,000× with an optically coupled isolated bioamplifier (ISO-80, World Precision Instruments), passed through a band-pass filter (40–3,000 Hz), and then digitized (sampling rate = 2,000 samples/s), transformed (root mean square), and integrated (time constant = 1 s; Biopac Software, Goleta, CA). As the mouse breathed, its entire body moved. This movement caused the CT nerve to rub against the tungsten electrode, creating a breathing artifact in each neural response. The data analysis was not impaired, however, because we could quantify the neural signal between each breathing artifact.

The fungiform taste papillae were stimulated using a continuous-flow system (VC-6 Perfusion Valve Control System, Warner Instruments, Hamden, CT). Solutions (∼22°C) were delivered to the anterior surface of the tongue at a rate of 10 ml/min. Each lingual stimulation lasted 20 s and was preceded and followed by ≥40 s of water rinse. To provide a reference stimulus and control for time-dependent changes in responsiveness, we recorded the response to 100 mM ammonium chloride (AC) regularly during each stimulus series. For consistency with previous studies (32, 44), we used AC as the reference stimulus. Each mouse was subjected to one of two solution series. The glucose series (n = 7 males and 7 females) consisted of AC, G₁, G₂, G₃, S, AC, AC, SG₁, SG₂, SG₃, and AC; the fructose series (n = 6 males and 6 females) consisted of AC, F₁, F₂, F₃, S, AC, AC, SF₁, SF₂, SF₃, and AC.

The dependent measure was the relative integrated response of the CT nerve to each solution. The calculation of this measure involved three steps. We initially used Biopac software to determine the mean voltage (in mV) over the 20 s immediately prior to stimulation with water (i.e., baseline response) and the mean voltage over the first 20 s of stimulation with a sweetener solution (e.g., G₁; i.e., excitatory response). We excluded the breathing artifacts from the mean voltage calculations, which resulted in the loss of 20–40% of the neural response to each stimulus, depending on breathing rate. Then we measured the difference between the baseline and excitatory response (i.e., absolute response). Finally, we divided the absolute response to a sweetener solution by the mean of the absolute response to AC, yielding the relative response to each sweetener solution.

The relative responses were analyzed in four ways. 1) To test for sex differences, we ran a two-way ANOVA on the relative response to the sweetener solutions, separately for the glucose and fructose series. We treated sex as a between factor and sweetener solution as a within factor. Because there was no significant effect of sex (glucose series: F_1,72 = 1.1, P > 0.05; fructose series: F_1,60 = 0.87, P > 0.05), we collapsed across sex in all subsequent electrophysiological analyses. 2) To test for differences in relative response to sweetener solutions, we ran one-way repeated-measure ANOVAs and Tukey's post hoc tests (modified for repeated-measures data) separately for the glucose and fructose series. 3) To examine responses to the single-component sugar solutions alone, we ran a separate mixed-model ANOVA, treating concentration (167, 250, and 333 mM) as a within factor and sugar type (glucose and fructose) as a between factor. 4) To determine whether the response to each binary mixture (e.g., SG₁) was synergistic, we determined the predicted relative response, on the basis of an additive model (i.e., the sum of the relative responses to S and G₁ alone); then we used a paired t-test to compare the actual with the predicted relative responses. If the actual relative response was significantly larger, we inferred that the response was synergistic. In all statistical comparisons, we used an alpha level of 0.05.

Initial licking responses.

We used a no-choice two-bottle testing paradigm, because it provides a highly sensitive measure of the difference in oral acceptability of two solutions (27). In this paradigm, we compared the number of licks a mouse emitted for two sweetener solutions during successive 5-s trials. We measured licks with a commercially available gustometer (Davis MS160-Mouse, DiLog Instruments, Tallahassee, FL) (29). The gustometer consisted of a testing chamber, a taste stimulus delivery system, and a dedicated computer system (with software that controlled the presentation of taste stimuli and recorded the precise timing of each lick). Once a mouse was placed in the test chamber, a motorized shutter opened. This provided the mouse with access to a single sipper tube (connected to a fluid reservoir) through a slot in the back wall of the testing chamber. The gustometer automatically centered the appropriate sipper tube in the slot behind the shutter.

Prior to testing, each mouse was subjected to 3 days of training, during which the mouse was familiarized with the gustometer and trained to lick from the sipper tube to obtain fluid. The mouse was water-deprived for 22.5 h prior to each 30-min training session to motivate licking. Each training session began when the shutter opened and the mouse took its first lick. On training day 1, the shutter remained open throughout the test session, permitting the mouse to drink freely from a single stationary spout that dispensed water. Immediately after this training session, the mouse was allowed 1 h of ad libitum access to water. Then it was water-deprived for another 22.5 h. On training day 2, the mouse was allowed more limited access to two sipper tubes, each of which dispensed water. In this case, once the shutter opened, the mouse initiated each 5-s trial by taking a lick from the sipper tube. At the end of the trial, the shutter closed for 7.5 s (during which time the second sipper tube was positioned in the center of the slot) and then reopened, enabling the mouse to initiate another trial of the same duration. In this manner, the mouse could initiate up to 144 trials across the 30-min test session. On training day 3, the same procedure was repeated. All mice adapted readily to the gustometer across the 3 training days, as indicated by the fact that they took >250 licks per training session. Once training was complete, each mouse was allowed food and water ad libitum for ≥24 h.

For testing, each mouse was subjected to three 30-min test sessions. The procedure for running the test sessions was similar to that used during training sessions 2 and 3. During a test session, the mouse was offered two sipper tubes, each of which dispensed a different sweetener solution. We treated the two sweetener solutions as a block and randomized (without replacement) the order of presentation of each sweetener within a block, so that each solution was presented once before the initiation of a second block. The mouse could initiate up to 72 blocks. To motivate the mice to initiate a large number of trials during the 30-min test session, we restricted their access to food and water for 23 h before each test session: each mouse was provided with a single 1-g chow pellet (F0173, Bio-Serv, Frenchtown, NJ) and 2 ml of tap water, which amounted to ∼19 and 30% of their normal daily food and water intake, respectively (J. I. Glendinning, unpublished data). We demonstrated previously that this deprivation procedure effectively increases the number of trials initiated by B6 mice but does not alter their lick responsiveness for sugars (27). After each test session, the mouse was allowed a recovery day, during which food and water were provided ad libitum.

We limited each mouse to three test sessions, because ingestive responses to sugar solutions can increase with dietary experience (52). Given that there were 21 stimulus pairs, we offered each of 7 cohorts of mice 3 unique combinations of stimulus pairs: 1) S vs. G₁, S vs. SG₁, and G₁ vs. SG₁; 2) S vs. G₂, S vs. SG₂, and G₂ vs. SG₂; 3) S vs. G₃, S vs. SG₃, and G₃ vs. SG₃; 4) S vs. F₁, S vs. SF₁, and F₁ vs. SF₁; 5) S vs. F₂, S vs. SF₂, and F₂ vs. SF₂; 6) S vs. F₃, S vs. SF₃, and F₃ vs. SF₃; or 7) G₁ vs. F₁, G₂ vs. F₂, and G₃ vs. F₃ (see Table 1 for abbreviations). We tested 9–13 mice per cohort (sex ratio per cohort balanced), resulting in a total number of 66 mice. To avoid order effects, we counterbalanced the presentation order of the three stimulus pairs across mice within each cohort.

Once a lick test was completed, we determined the number of 5-s trials initiated and the mean number of licks emitted per trial for each of the two sweetener solutions. Then to compare taste responsiveness to each sweetener solution, we ran a paired t-test. To increase the reliability of our measure of taste responsiveness, we included only those mice that initiated ≥18 trials in a test session (i.e., 9 trials per sweetener solution). This inclusion criterion caused us to reject only one mouse. The mean number of trials initiated per test session was 50.

To test for an effect of sex on lick responsiveness, we ran mixed-model ANOVAs, separately for each of the 21 stimulus pairs. In each ANOVA, we treated sex as a between factor and solution type as a within factor. Because there was no significant main effect of sex in any of the ANOVAs (i.e., P > 0.05), we collapsed across sex in the lick analyses.

Experiment 2: Measurement of Daily Intake

We measured daily intake with a 48-h two-bottle preference test. Each mouse was tested individually in its home cage. On each testing day, a mouse was provided with laboratory chow ad libitum and two bottles. One bottle contained the control solution [i.e., water (W)] and the other the experimental solution (see below). The mice accessed the solutions through sipper tubes located in the cage lid. The sipper tubes protruded ∼1 cm into the cage, were positioned ∼5 cm apart, and had a 1.5-mm hole designed for mice (Ancare). To control for position preferences, we tested each sweetener solution over 2 consecutive days, with alternation of its left-right position each day. To determine daily fluid intake from the two bottles, we measured (to the nearest 0.01 g) their change in weight over each successive day of testing, using an electronic balance interfaced to a computer. We estimated daily spillage on the basis of weight changes of four bottles that were placed on two empty cages (2 bottles contained the control solution, and 2 contained the experimental solution). This daily spillage was subtracted from each measure of daily fluid intake. We did not measure the pattern or amount of chow ingested each day.

Because intake of sugar solutions can increase with dietary experience (52), we divided the experimental solutions across six separate cohorts of mice (n = 10 mice per cohort, total number of mice = 60); the net effect was that each cohort was run through four 48-h preference tests. In each series of tests, we kept the control solution (W) constant but varied the experimental solution (see below). During test 1, the experimental solution was W; during test 2, it was saccharin (S), and during tests 3 and 4, it was one of the other sweetener solutions. Accordingly, each cohort was tested with one of the following solution series: 1) W, S, G₁, and SG₁; 2) W, S, G₂, and SG₂; 3) W, S, G₃, and SG₃; 4) W, S, F₁, and SF₁; 5) W, S, F₂, and SF₂; or 6) W, S, F₃, and SF₃ (see Table 1 for abbreviations and caloric value of each solution). We tested 9–10 mice per cohort (sex ratio per cohort balanced), resulting in a total number of 57 mice. To avoid order effects, we counterbalanced the presentation order of experimental solutions across mice within each cohort.

For the data analysis, we focused on mean daily intake from the experimental solutions across the 48-h preference tests. The only exception was when both the experimental and control solutions were water; in this case, we present combined daily intake of both solutions. To test for sex differences, we ran a two-way ANOVA, separately for each cohort. We treated sex as a between factor and experimental solution as a within factor. Because the main effect of sex was not significant (P > 0.05) for each of the cohorts, we collapsed across sex for all subsequent analyses. For each cohort, we used a one-way repeated-measures ANOVA and Tukey's post hoc test (modified for repeated-measures data) to compare daily intakes across the different experimental solutions.

RESULTS

Experiment 1: Measuring Taste Responses

CT nerve responses.

We show typical CT nerve responses to several of the sweetener solutions in Fig. 1A, and mean responses to the sweetener solutions in Fig. 1, B and C. There was a significant main effect of sweetener solution on the responses to the G_1–3, S, and SG_1–3 solutions (F_6,97 = 39.7, P ≤ 0.05), and the relative magnitude of response to each solution was as follows: G₁ = G₂ = G₃ < S < SG₁ = SG₂ = SG₃ (Fig. 1B). Similarly, there was a significant main effect of sweetener solution on the responses to the F_1–3, S, and SF_1–3 solutions (F_6,83 = 29.6, P ≤ 0.05), and the relative magnitude of response to each solution was as follows: F₁ = F₂ = F₃ < S < SF₁ = SF₂ = SF₃ (Fig. 1C).

Fig. 1.

Responses of the chorda tympani (CT) nerve to sweetener solutions. A: typical integrated responses of the CT nerve to 0.1 M ammonium chloride (AC) and 6 other sweetener solutions. Upward and then downward deflections of each integrated response indicate application of sweetener solution to the tongue (on) and rinsing of solution from the tongue (off) with water (e.g., arrows on CT nerve response to saccharin in top row). B: relative responses to saccharin, 3 concentrations of glucose, and binary mixtures of saccharin + each glucose concentration. C: relative responses to saccharin, 3 concentrations of fructose, and binary mixtures of saccharin + each fructose concentration. See Table 1 for composition of each sweetener solution and its abbreviation. In B (n = 14 mice) and C (n = 12 mice), each bar indicates mean ± SE. Different letters (a, b, c) above bars indicate means that differ significantly from one another (Tukey's post hoc test, P ≤ 0.05).

Because CT nerve responses to the single-component sugars were so weak, subtle concentration-dependent changes in CT nerve response could have been missed when we conducted ANOVAs on the entire glucose or fructose series. As a result, we ran a separate two-way ANOVA on the CT nerve responses to G_1–3 and F_1–3. There was a significant main effect of sugar concentration (F_2,48 = 17.0, P ≤ 0.05) and sugar type (F_1,48 = 5.9, P ≤ 0.05), but no sugar concentration × sugar type interaction (F_2,48 = 1.76, P > 0.05). This analysis, together with visual inspection of Fig. 1, B and C, reveals that the CT nerve response to both sugars increased significantly with concentration and that the CT nerve responses to F_1–3 were larger than those to G_1–3.

To determine whether the CT nerve responses to the binary mixtures were larger than those predicted on the basis of an additive model (i.e., sum of responses to each sweetener alone), we compared the observed and predicted relative responses to each binary mixture (Fig. 2). The observed responses to SG₁, SG₂, and SG₃ were significantly higher than the predicted responses. In contrast, the observed responses to SF₁, SF₂, and SF₃ did not differ significantly from the predicted responses. Thus these analyses revealed synergistic interactions between saccharin and glucose and additive interactions between saccharin and fructose.

Fig. 2.

Predicted (open bars) and observed (closed bars) relative responses of CT nerve to SF₁, SF₂, and SF₃ (top; n = 12) and SG₁, SG₂, and SG₃ (bottom; n = 14). Predicted responses are based on an additive model, in which relative CT nerve responses to each component stimulus were summed. Each pair of predicted and observed responses was compared using paired t-tests: *P ≤ 0.05. A significantly higher observed CT nerve response indicates that response to binary mixture was synergistic. Each bar indicates mean ± SE.

Licking responses.

In Fig. 3, we show licking responses to 18 different pairs of sweetener solutions. These results reveal that saccharin stimulated significantly more licks per trial than did any of the single-component sugars (i.e., F₁, F₂, F₃, G₁, G₂, and G₃) and that the binary mixtures (i.e., SG₁, SG₂, SG₃, SF₁, SF₂, and SF₃) stimulated significantly more licks per trial than did saccharin or any of the single-component sugars. These licking responses corroborate the CT nerve recordings; that is, the solutions that elicited the strongest CT nerve responses (i.e., the binary mixtures) also stimulated the most licks per trial. On the other hand, the solutions that elicited the weakest CT nerve responses (i.e., the single-component sugars) stimulated the fewest licks per trial.

Fig. 3.

Licking responses to 18 different pairs of sweetener solutions. Top: licking for F vs. S, F vs. SF, and S vs. SF, with 3 different concentrations of F. Bottom: licking for G vs. S, G vs. SG, and S vs. SG, with 3 different concentrations of G. In each panel, results of 3 separate no-choice 2-bottle acceptability tests are shown. For each test, number of licks per trial between each pair of sweeteners was compared using paired t-test: *P ≤ 0.05. Each bar indicates mean ± SE. Data in each panel are derived from different mice (n = 9–10 mice per panel).

We ran a final set of tests to compare licking for G₁ vs. F₁, G₂ vs. F₂, and G₃ vs. F₃ (Table 2). However, there were no significant differences in the number of licks per trial for any of the isomolar concentrations of glucose and fructose. This finding contradicts the nerve recordings, as the CT nerve responses to the fructose solutions were greater than those to the isomolar glucose concentrations.

Table 2.

Licking for equimolar concentrations of glucose and fructose by B6 mice, as determined during no-choice two-bottle tests

Stimulus Pair	Licks per 5-s Trial	Paired t Value (df)
G1	16.7 ± 2.1	0.09(9)
F1	16.9 ± 1.3
G2	16.9 ± 1.4	1.67(9)
F2	19.4 ± 1.2
G3	20.5 ± 2.0	0.57(9)
F3	21.8 ± 1.4

Values are means ± SE; n = 5 males and 5 females per lick test. See Table 1 for composition of each sweetener solution and its abbreviation. Lick rates within each stimulus pair were compared using paired t-tests. All paired t-tests were nonsignificant (P > 0.05). df, Degrees of freedom.

Experiment 2: Measurement of Daily Intake

Daily intakes of the experimental solutions are presented in Fig. 4. For each cohort, there was a significant main effect of experimental solution on daily intake (in all cases, P ≤ 0.05). The mice invariably consumed significantly more sweetened solutions than water. However, daily intakes of the sweetened solutions varied greatly across the six cohorts.

Fig. 4.

Daily intake of sweetener solutions. Top: daily intake of water (W) compared with daily intake of F₁, S, and SF₁; F₂, S, and SF₂; and F₃, S, and SF₃. Bottom: daily intake of W compared with daily intake of G₁, S, and SG₁; G₂, S, and SG₂; and G₃, S, and SG₃. Each bar indicates mean ± SE. Different letters above bars indicate means that differ significantly from one another within a panel (Tukey's post hoc test, P ≤ 0.05). Data in each panel are derived from different mice (n = 9–10 mice per panel).

There were dramatic differences in daily intake of the fructose vs. glucose solutions. For instance, daily intake of F₁, F₂, and F₃ increased only weakly with concentration (i.e., from 5.5 to 9.0 g). In contrast, daily intake of G₁, G₂, and G₃ increased nearly threefold with concentration (i.e., from 10.6 to 28.5 g). Furthermore, daily intake of G₁, G₂, and G₃ solutions was significantly higher than daily intake of each of the isocaloric F₁, F₂, and F₃ solutions, respectively (in each unpaired t-test comparison, P ≤ 0.05).

Daily intakes also differed markedly between the binary mixtures of saccharin + fructose and saccharin + glucose (Fig. 4). For instance, daily intake of the saccharin + fructose mixtures was relatively low and did not increase with fructose concentration (i.e., from 10.5 to 11.0 g), whereas daily intake of the saccharin + glucose mixtures increased markedly with glucose concentration (i.e., from 15 to 26 g). Because the mice did not consume significantly more SF₁, SF₂, or SF₃ than of each component sweetener, it follows that saccharin and fructose (at any concentration) failed to act in an additive manner to increase daily intake. On the other hand, because the mice consumed significantly more SG₁ and SG₂ than S, G₁, or G₂, it follows that S + G₁ (and S + G₂) acted additively to increase daily intake. Although daily intake of SG₃ was higher than daily intake of S, it was equivalent to daily intake of G₃. While this latter finding may seem surprising at first glance, it is notable that daily intake of SG₃ and G₃ exceeded the body weight of the mice. Thus the absence of an additive effect of S + G₃ on daily intake probably reflects digestive processing constraints.

The central goal of this study was to assess the contribution of taste to daily intake. We predicted that if taste was a major determinant of sweetener intake, then daily intake should increase with magnitude of the CT nerve and lick responses. Our results contradicted this prediction in several ways. For example, despite eliciting some of the weakest CT nerve and licking responses, the G₃ solution stimulated the highest daily intake. Similarly, saccharin and the binary mixtures of saccharin + fructose elicited the strongest taste responses but the lowest daily intakes.

DISCUSSION

The mice exhibited widely divergent taste-mediated responses to the sweetener solutions. In the CT nerve recordings, the relative magnitude of response to the sweetener solutions was as follows: binary mixtures > saccharin > individual sugars. Similarly, the relative number of licks per trial for the sweetener solutions was as follows: binary mixtures > saccharin > individual sugars. It follows, therefore, that the mice licked most avidly for the solutions that elicited the strongest CT nerve signal and, by extension, the most intense sweet taste.

We observed one discrepancy between the CT nerve and licking responses. Despite eliciting slightly (but significantly) higher CT nerve responses, the F₂ and F₃ solutions elicited the same number of licks per trial as the G₂ and G₃ solutions. The most likely explanation for this discrepancy is that the magnitude of difference in CT nerve response between the aforementioned isomolar glucose and fructose solutions was too small to support behavioral discrimination. On the other hand, the magnitude of the difference in CT nerve response between the F₃ and S solutions (or the G₃ and S solutions) was three times larger, making them easier to discriminate.

Caveats

Before discussing the role of taste in stimulating intake of the sweetener solutions, it is necessary to address three caveats about our experimental approach. 1) In mice, ingestion of the sweetener solutions did not merely stimulate the taste system; it also stimulated the olfactory and trigeminal chemosensory systems (48, 79). Although it is difficult to assess the relative contributions of these three sensory modalities to sweetener intake, it is notable that genetic ablation of sweet taste transduction proteins (e.g., T1R3 or Trpm5) strongly attenuates licking responses for sweeteners in mice (77, 78). 2) We recorded from only one of the three taste nerves (i.e., the CT nerve) and, thus, provide an incomplete picture of sweet taste input. Little is known about the relative importance of input from the CT, glossopharyngeal, and greater superficial petrosal nerves to sweetener intake in rodents, but it is significant that bilateral ablation of the CT nerve significantly attenuates daily intake of dilute concentrations of sucrose and Polycose (a polysaccharide) in rats (72). 3) We made whole nerve recordings and, thus, could not determine the percentage of each CT nerve response that reflected activation of sweet-sensitive afferent fibers (8, 68). For instance, the free Na⁺ in the Na⁺-saccharin solutions may have increased the CT nerve responses by activating salt-sensitive afferent fibers.

To determine the extent to which free Na⁺ in 38 mM Na⁺-saccharin increased CT nerve responses, we measured the relative responses of the CT nerve to 38 mM Na⁺-saccharin and 38 mM NaCl, with and without 100 μM amiloride (n = 4 B6 mice). Amiloride blocks a majority of the peripheral taste response to Na⁺ (62). For these measurements, we used the same procedures described above. The mean relative responses were as follows: 1) 0.86 to Na⁺-saccharin, 2) 0.75 to Na⁺-saccharin + amiloride, 3) 0.11 to NaCl, and 4) 0.04 to NaCl + amiloride (J. I. Glendinning, unpublished data). These findings indicate that ∼13% of the CT nerve response to Na⁺-saccharin reflected the contribution of amiloride-sensitive salt fibers and ∼5% reflected the contribution of amiloride-insensitive salt fibers. To assess the potential contribution of both classes of salt-sensitive afferent fibers to the CT nerve responses in Fig. 1, B and C, we subtracted 18% from the relative responses to each of the solutions containing Na⁺-saccharin. This transformation had no impact on the relative magnitude of the CT nerve responses to the different sweetener solutions; they remained as follows: binary mixtures > saccharin > single-component sugars. Accordingly, it is unlikely that the free Na⁺ in solutions containing Na⁺-saccharin confounded interpretation of the CT nerve responses.

Notwithstanding the three caveats discussed above, we believe that strong agreement between the relative CT nerve responses and relative licking responses to the sweetener solutions confers on both measures a high degree of reliability and compensates for each of their respective limitations. It follows, therefore, that the CT nerve responses and licking responses provide a robust measure of the initial taste response.

Did Taste Determine Intake of the Dilute Sugar Solutions?

We found that daily intake of the sweetener solutions varied independently of taste-mediated responses to the same sweetener solutions. For instance, the G₂ and G₃ solutions caused some of the largest daily intakes but elicited some of the weakest taste responses. On the other hand, the SF_1–3, F_1–3, and S solutions caused some of the smallest daily intakes but elicited the strongest taste responses. Furthermore, the G_1–3 solutions elicited weaker CT nerve responses than (and lick rates similar to) the corresponding F_1–3 solutions but, nevertheless, stimulated greater daily intake. On the basis of these findings, we infer that the taste responses did not determine daily intake of the sweetener solutions.

If taste did not determine daily intake, then the most likely alternative mechanism is postoral nutritive stimulation. This proposition is based on two observations. 1) Intragastric (IG) administration of sugars can provide robust stimulation of feeding in B6 mice. For instance, when intake of one flavored solution (CS+) was paired with IG infusions of sucrose and intake of another flavored solution (CS−) was paired with IG infusions of water, the mice consumed significantly larger quantities of the CS+ solution and developed a strong preference for the CS+ over the CS− solution (58, 59). 2) Rodents developed strong preferences for flavored solutions when their intake was paired with IG infusions of glucose but failed to do so when intake was paired with IG infusions of fructose; accordingly, the postoral stimulation from glucose was much stronger than that from fructose (1, 56). On the basis of these findings, one would predict that 1) glucose should stimulate higher daily intake than fructose, 2) daily intake should increase more robustly with glucose than fructose concentration, and 3) binary mixtures of saccharin and glucose should stimulate greater daily intakes than binary mixtures of saccharin and fructose. This is exactly what we report in the present study.

Two lines of evidence argue against the possibility that the mice consumed less of the fructose solutions because they were more satiating (i.e., suppressed intake by activating negative-feedback mechanisms in the gut). 1) Although relevant mouse data are not available, published rat studies indicate that fructose is not more satiating than glucose. For instance, three studies reported that glucose and fructose were equally satiating (7, 36, 57), two reported that fructose was less satiating (46, 73), and only one reported that fructose was more satiating (35). 2) Published mouse studies found that daily intakes of glucose and fructose solutions increase with concentration until ∼333 mM but then decrease monotonically with successively higher concentrations (3, 4). The latter observation is thought to reflect the fact that daily intake of concentrated (>333 mM) sugar solutions is controlled by a complex interaction of positive (i.e., sweet taste and postoral nutritive stimulation) and negative (i.e., activation of postoral satiety mechanisms) inputs (17, 53, 57). To minimize the contribution of negative inputs, we tested only dilute (≤333 mM) concentrations of glucose and fructose.

Did Sweet Taste and Postoral Nutritive Stimulation Mutually Reinforce One Another?

We are not suggesting that taste had no impact on daily intake of the sweetener solutions. Rather, our findings indicate that sweet taste and postoral nutritive stimulation mutually reinforced one another, resulting in a positive-feedback loop (59). For instance, the SG₁ and SG₂ solutions stimulated greater daily intakes than did the G₁ or G₂ solutions, respectively, despite having equivalent caloric values. The most parsimonious interpretation of this finding is that the addition of saccharin to the G₁ and G₂ solution increased their sweet taste intensity, thereby causing the mice to lick more avidly; this in turn caused more glucose to be delivered to the gut, leading to greater postoral nutritive stimulation; the high postoral nutritive stimulation in turn motivated the mouse to initiate a greater number of bouts (59). Given that sucrose stimulation of the taste and postoral viscerosensory systems increases dopamine flux in the nucleus accumbens (19, 43), the positive-feedback loop could be mediated by this structure.

It is notable that, despite eliciting a weak taste response, the G₃ solution stimulated extremely high daily intakes. While this observation would appear to contradict the positive-feedback model discussed above, there is evidence that indicates otherwise. For instance, when intake of a flavored solution is associated repeatedly with IG glucose infusions, rats will develop increased lick burst sizes and positive taste reactivity for the same solution (38, 39). These findings reveal that postoral nutritive stimulation from a sugar solution can increase the hedonic value of its taste. If a similar phenomenon occurred here with the G₃ solution, then the hedonic value of its taste would have increased over the course of the 48-h intake tests. Accordingly, the enhanced taste acceptability of the G₃ solution would have reinforced its strong postoral nutritive stimulation, producing the observed high rates of daily intake. This possibility could be tested by comparison of lick burst sizes and taste reactivity across B6 mice with or without 48 h of ad libitum exposure to the G₃ solution.

We cannot reject the possibility that the fructose solutions caused weak postoral stimulation, because the F₂ and F₃ solutions stimulated daily intakes that were statistically indistinguishable from those stimulated by the S solution. If daily intakes of the F₂ and F₃ solutions were driven exclusively by taste, then the mice should have consumed less of the F₂ and F₃ solutions than of the S solution, given that the F₂ and F₃ solutions elicited significantly weaker taste responses.

Synergistic Interactions of Sweeteners in the Peripheral Taste System

It has been reported that preexposure of the tongue to saccharin enhances subsequent peripheral taste responses to certain amino acids in mice (41) and rhesus monkeys (40). Furthermore, binary mixtures of monosodium glutamate and sucrose were reported to have synergistic effects on the response of sucrose-insensitive units in the CT nerve of rats (26). Here, we provide the first evidence that binary mixtures of saccharin and glucose have synergistic effects on integrated CT nerve responses in mice. While the mechanism underlying the saccharin + glucose synergy is unclear, molecular modeling studies implicate cooperative binding at distinct sites on the T1R2-T1R3 heterodimeric sweet taste receptor (20, 37). Further work is needed to explain the absence of a saccharin + fructose synergy. This latter result may reflect weaker cooperative binding by saccharin + fructose than by saccharin + glucose. It may also reflect the fact that the CT nerve responses of mice tested with the fructose solutions were generally stronger than those of mice tested with the glucose solutions (Figs. 1 and 2). Accordingly, our ability to observe a saccharin + fructose synergy could have been constrained by a ceiling effect. This possibility could be evaluated by testing a wider range of concentrations of both sweeteners.

Binary mixtures of saccharin and glucose have been found to produce synergistic increases in daily intake in mice (12) and rats (71) and sweetness intensity in humans (49). That the mixture of saccharin + glucose tastes better than either component alone is supported by a study in which rats were offered two sipper tubes: one dispensed saccharin and the other glucose. As the rats consumed both solutions, they switched rapidly between the two sipper tubes in an apparent effort to create the more preferred blend of saccharin and glucose in their mouth (65).

We found that the SG₁ solution stimulated significantly greater intake than did the S or G₁ solutions. However, because intake of the SG₁ solution was not greater than the sum of the daily intakes of the S and G₁ solutions, it follows that the SG₁ solution had an additive (and not a synergistic) effect on daily intake. In contrast, Capretta (12) reported that daily intake of the SG₁ solution was greater than the sum of the daily intakes of S and G₁ alone, indicating that SG₁ had a synergistic effect on daily intake in B6 mice. These contradictory findings likely reflect differences between our two-bottle testing procedures. Whereas we paired each of the sweetener solutions (i.e., SG₁, S, or G₁) with water, Capretta paired the SG₁ solution with the S or G₁ solution. Daily intake of the SG₁ solution in our study and the study of Capretta was similar (i.e., ∼15 g), but daily intake of the S and G₁ solutions differed. It is likely that the lower intake of the S and G₁ solutions in the study of Capretta stems from the fact that the reward value of these solutions was depreciated, because they were each presented alongside the more preferred SG₁ solution. This phenomenon is referred to as a negative-contrast effect (24, 25, 31). In the present study, the reward value of the S and G₁ solutions would not have been depreciated by a negative-contrast effect, because they were each presented alongside a less preferred solution (i.e., water).

Perspectives and Significance

While the sweet tooth is often assumed to be activated primarily by taste, our results indicate that it reflects a complex interaction between taste and postoral nutritive stimulation (1, 53, 58, 59). At this point, however, little is known about the viscerosensory mechanism that mediates postoral nutritive stimulation. It is thought to reside within the epithelium of the duodenum and jejunum (2), respond strongly to glucose and fats (2, 59), and communicate with the central nervous system via a humoral mechanism. The latter inference is based on reports that neither abdominal vagotomy nor celiac-superior mesenteric ganglionectomy blocks nutrient-flavor conditioning in rats (54, 60). Even though many of the same transduction proteins that mediate sweet taste in the mouth are also expressed in the small intestine (69, 75), there is no evidence that any of these signaling proteins mediate postoral nutritive stimulation. A greater understanding of the viscerosensory signaling mechanism(s) would provide investigators with a new set of targets for blocking the rewarding properties of sugars.

GRANTS

This work was supported in part by a grant from the Howard Hughes Medical Institute to Barnard College.

DISCLOSURES

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