Several findings suggest the existence of a “fatty” taste, and the CD36 fatty acid translocase is a candidate taste receptor. The present study compared fat preference and acceptance in CD36 knockout (KO) and wild-type (WT) mice using nutritive (triglyceride and fatty acid) and nonnutritive (Sefa Soyate oil) emulsions. In two-bottle tests (24 h/day) naive KO mice, unlike WT mice, displayed little or no preference for dilute soybean oil, linoleic acid, or Sefa Soyate emulsions. At high concentrations (2.5–20%), KO mice developed significant soybean oil preferences, although they consumed less oil than WT mice. The postoral actions of fat likely conditioned these preferences. KO mice, like WT mice, learned to prefer a flavored solution paired with intragastric soybean oil infusions. These findings support CD36 mediation of a gustatory component to fat preference but demonstrate that it is not essential for fat-conditioned flavor preferences. The finding that oil-naive KO mice failed to prefer a nonnutritive oil, assumed to provide texture rather than taste cues, requires explanation. Finally, CD36 deletion decreased fat consumption and enhanced the ability of the mice to compensate for the calories provided by their optional fat intake.
- linoleic acid
- soybean oil
- Sefa Soyate oil
- flavor conditioning
- gastric infusions
the flavor of fat has traditionally been attributed to its texture (mouth-feel) and, secondarily, to its associated odors (7, 21). However, there is accumulating evidence for the existence of a fat “taste” mediated by a gustatory response to fatty acids. In 1997, Gilbertson et al. (10) proposed that essential fatty acids activate taste cells by inhibiting a delayed rectifying potassium channel. That same year, Fukuwatari et al. (8) provided evidence that a fatty acid translocase (CD36) is localized in lingual taste buds, and they hypothesized that this protein participates in the oral recognition of fat. More recently, Laugerette et al. (16) confirmed the localization of CD36 in circumvallate taste buds and showed that it colocalized in some taste cells with α-gustducin, a taste signaling protein. Most importantly, they reported that CD36 knockout (KO) mice, unlike wild-type (WT) mice, failed to prefer a fatty acid emulsion (2% linoleic acid) in short- and long-term two-bottle tests. In contrast, the KO and WT mice did not differ in their taste preference for and avoidance of sucrose and quinine, respectively. The fatty acid preference displayed by the WT mice is consistent with other studies suggesting the existence of a fat taste (20, 33, 36). Furthermore, the attraction of rodents to triglycerides appears to be mediated, in part, by fatty acids released in the mouth by the action of salivary lipase (15). According to this result, CD36 KO mice should also show reduced preferences for triglycerides, which are the source of fatty acids in the diet. However, this was not tested in the Laugerette et al. (16) study.
The present study further examined the role of CD36 in fat appetite. Two-bottle preference and acceptance responses of CD36 KO and WT mice to triglyceride (soybean oil), as well as to fatty acid (linoleic acid), emulsions were compared. In addition, preferences for nonnutritive oil emulsions (Sefa Soyate), nonnutritive (saccharin), and nutritive (sucrose) sweeteners were investigated to determine the specificity of the “taste” preference deficits of KO mice. Prior studies indicate that, while rodents prefer nutritive to nonnutritive oil, they nevertheless consume substantial amounts of nonnutritive emulsions (1, 22). In some cases the preference for nutritive over nonnutritive oil develops over successive test trials (1), which suggests that it is mediated by post-oral nutrient feedback. Consistent with this interpretation, rats and mice learn to prefer an arbitrary flavor paired with intragastric infusions of nutritive oil over another flavor paired with intragastric water infusions (18, 31).
This study was designed to evaluate both unconditioned and conditioned preferences for nutritive oil emulsions in CD36 KO and WT mice. In experiment 1A, preferences were initially evaluated at dilute oil concentrations, which minimize post-oral conditioning effects, and then preferences were reassessed after the animals had consumed a 2.5% soybean oil emulsion that has a post-oral reinforcing effect. Soybean oil preference and acceptance were also measured at higher concentrations (5–20%) to determine whether CD36 gene deletion alters the intake of energy-dense oil emulsions. Experiment 1B directly assessed the post-oral flavor conditioning action of soybean oil in KO and WT mice using the intragastric infusion procedure. This was of interest because CD36, in addition to being present in lingual taste buds, is localized in the stomach and upper intestinal tract, where it facilitates fatty acid absorption and may serve a signaling function (3, 5). Thus, the post-oral, as well as the oral appetitive actions of nutritive oil, may be compromised in CD36 KO mice relative to WT animals. In experiment 2, linoleic acid preference was compared in KO and WT mice before and after they had experience with soybean oil emulsions to follow up findings from experiment 1.
Experiment 1 used male CD36 KO mice backcrossed to the C57BL/6J background (n = 10) bred at Washington University (St. Louis, MO) and male C57BL/6J wild-type mice (WT; n = 10) obtained from the Jackson Laboratory (Bar Harbor, ME); the mice were 11-wk-old at the start of testing. The mice were singly housed in clear plastic test cages designed for gastric infusions (31). In experiment 2, male CD36 KO mice (n = 10) and male C57BL/6J wild-type mice (n = 9) bred in the Washington University colony were used. The animals were housed in clear plastic tub cages and were 13 wk old at the start of testing. In both experiments, the animals were housed in rooms with a 12:12-h light-dark cycle maintained at 22°C and had ad libitum access to laboratory chow (5001, PMI Nutrition International, Brentwood, MO). Experimental protocols were approved by the Institutional Animal Care and Use Committee at Brooklyn College and were performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals.
Test Fluids and Intake Measures
The test fluids used in the various two-bottle trials are listed in Table 1. The nutritive and nonnutritive oils contained soybean oil and Sefa Soyate, respectively. The nonnutritive oil Sefa Soyate (Proctor & Gamble, Cincinnati, OH) consists of sucrose esters of fatty acids made from partially hydrogenated soybean oil and contains less than 0.05% of unbound free fatty acids. Sefa Soyate rather than mineral oil was used because it is the precursor of olestra used in human food products, and preliminary work revealed that rats and mice prefer Sefa Soyate emulsions to mineral oil emulsions (K. Ackroff and A. Sclafani, unpublished findings).
In initial tests (Tests 1.2, 1.3, 1.4), Sefa Soyate oil and soybean oil (Crisco oil, J. M. Smucker, Orrville, OH) emulsions were prepared at concentrations of 0.313–2.5% wt/wt using deionized water and 0.15% Emplex (sodium stearoyl lactylate, American Products, Kansas City, MO). The mixture was homogenized (Ultra-Turrax T25, IKA-Works, Cincinnati, OH) at high speed for 5 min and further processed in a microfluidizer (HC-5000, Microfluidics, Newton, MA). The oil emulsions were not completely stable over 24 h, and analysis of the 2.5% emulsion indicated that the oil concentration at the bottom of the drinking tube (the fluid available to the animals) had declined to ∼1.8% at 24 h.
In test 1.5, linoleic acid (99% pure, Sigma Chemical, St. Louis, MO) was prepared at a 2% concentration suspended in water using 0.3% xanthan gum (Sigma) as in the prior study of CD36 KO mice (16). For comparison, 2% Sefa Soyate and 2% soybean oil were also tested as xanthan gum suspensions. In test 1.6, soybean oil was presented in the form of a stable emulsion using 20% Intralipid (Baxter, Deerfield, IL), which contains 20% soybean oil, 2.25% glycerol, and 1.2% egg yolk phospholipids. As in a previous study (30), the 20% Intralipid was used at full strength and was also diluted with deionized water to produce emulsions that contained 0.313–10% soybean oil. These diluted versions are referred to as 0.313–10% Intralipid. Depending upon the test, oil preference was compared with a vehicle solution containing 0.15% Emplex, 0.3% xanthan gum, or deionized water (Table 1). In test 1.1, saccharin (sodium saccharin; Sigma) solution was presented at a concentration of 0.2%. In test 1.7, sucrose (Domino Foods, Yonkers, NY) solutions were prepared at concentrations of 4–32%. The test solutions used in experiment 1B are described in the procedure section.
In test 2.1 and 2.4, linoleic acid (0.0025–2%) suspensions were prepared using 0.3% xanthan gum. In tests 2.2 and 2.3, soybean oil (0.0025%–16%) and Sefa Soyate oil (0.5%) emulsions were prepared using Emplex as in experiment 1, except that the microfluidizer was operated at a higher air pressure, which resulted in very stable emulsions. After 24 h, the 2.5% soybean oil emulsion was determined to be 2.4% oil at the bottom of the drinking tube.
All test fluids, except the Intralipid emulsions, were prepared on a wt/wt basis because solution intakes were recorded by weight. Because 20% Intralipid is formulated on a vol/vol basis, lower concentrations were also prepared vol/vol. However, Intralipid intakes were measured by weight, and caloric intakes were calculated accordingly (the caloric density of 20% Intralipid is 2.0 kcal/ml or 2.03 kcal/g).
Test fluids were presented in 50- or 80-ml plastic tubes and were available through a stainless-steel sipper spout with a 1.5-mm opening designed for mice (Ancare, Bellmore, NY). In experiment 1, the two drinking tubes were mounted on the front of the test cage, and the mouse licked the spouts through two slots (5 × 20 mm, 32 mm apart) (28). In experiment 2, the drinking tubes were placed on the top of the tub cage to the right of the feeder. Fluid intakes were measured to the nearest 0.1 g by weighing the drinking tubes on an electronic balance interfaced to a laptop computer; the weights were automatically entered into a spreadsheet. Spillage was estimated daily by recording the change in weight of two drinking tubes that were placed on an empty cage; one tube contained that day's test solution, the other tube contained the vehicle. The spillage from these tubes averaged 0.2–0.3 g·tube−1·day−1, and intake measures were corrected by this amount. To eliminate the possibility that the mice would acquire preferences for sipper spouts associated with specific nutrient solutions (see Ref. 6), the sipper spouts were assigned to a side rather than a fluid (28). The left-right positions of the test fluids alternated daily.
The mice were adapted to the laboratory for 2 wk and were then given a series of two-bottle preference tests with the various test fluids as outlined in Table 1. Except where noted, the test fluids were presented in order of increasing concentration, and each concentration for two consecutive days; between each test series, the mice were given two-bottle access to deionized water for 2 to 7 days. The solutions were available 23 h/day, and the bottles were weighed, washed, and refilled during the remaining hour. The mice were first given a two-bottle test (test 1.1) with 0.2% saccharin vs. water followed by a Sefa Soyate oil vs. vehicle (Emplex) at oil concentrations of 0.313–2.5% (test 1.2). In test 1.3, the mice were offered the choice of soybean oil (0.313–2.5%) vs. vehicle (Emplex). To examine the ability of the mice to discriminate between the orosensory properties of nutritive and nonnutritive oils, before concentrated soybean oil emulsions conditioned flavor preferences (31), midway into test 1.3 (after the 0.625% concentration) a Sefa Soyate vs. soybean probe test was conducted (test 1.4). The mice were first given 0.625% Sefa Soyate vs. vehicle for 2 days followed by 0.625% Sefa Soyate vs. 0.625% soybean oil for 2 days.
In test 1.5, the mice were given the choice of 2% linoleic acid vs. vehicle (gum). This was followed by 2% Sefa Soyate oil vs. vehicle, and then 2% soybean oil vs. Sefa Soyate oil. A gum vehicle was used in this test to replicate a prior study of CD36 KO mice (16). The mice were then offered Intralipid vs. water at oil concentrations of 0.313–20% (test 1.6). Chow intake was recorded during the 10% and 20% Intralipid test days and compared with the water-only days that preceded test 1.6. The chow and Intralipid data were used to determine whether the CD36 KO and WT mice differed in their caloric regulation when consuming a concentrated lipid source. In a final test (test 1.7) the mice were offered sucrose vs. water at sugar concentrations of 4% to 32%. The sucrose concentration started at 4% to match the concentration used in a prior CD36 KO study (16).
The mice from experiment 1A were anesthetized with isoflurane (2%) inhalation and fitted with a chronic intragastric catheter (31). The distal end of the catheter was routed under the skin and emerged at the back of the neck, where it was heat-sealed. One week later, each mouse was briefly (5 min) anesthetized with isoflurane, the gastric catheter was attached to tubing that passed through an infusion harness with a spring tether (CIH62; Instech Laboratories, Plymouth Meeting, PA), and the tether was fitted to the mouse. The animals were returned to the test cages, and the intragastric catheter tubing was attached via a swivel to an infusion pump. Licking was monitored by an electronic lickometer and a microcomputer, which controlled the syringe pumps. As the mouse licked, a matched volume of fluid was infused into its stomach at a rate of 0.5 ml/min. The mice were adapted to drink water paired with intragastric water infusions for several days. This was followed by a saccharin solution (0.1 or 0.05%) paired with intragastric water infusions for 3 days (see results). The mice were then given 6 one-bottle training days with flavored saccharin solutions paired with intragastric infusions of 5% Intralipid (odd days) or water (even days). For half of the mice, the conditioned stimulus flavor (CS+) paired with lipid infusions was cherry (0.05% Kool-Aid mix, Kraft Foods, White Plains, NY), and the flavor (CS−) paired with water infusions was grape; the flavor-infusion pairs were reversed for the remaining mice. At the end of training, the mice were given two-bottle preference tests with CS+ vs. CS−. For 2 days, the CS+ and CS− solutions were paired with intragastric Intralipid and water infusions, respectively (reinforced test) and for the next 2 days, both solutions were paired with water infusions (nonreinforced test). Each gastric catheter was attached to two infusion pumps using a Y-connector during the two-bottle tests. Throughout training and testing, the CS solutions were available ad libitum 22 h/day; the infusion equipment was serviced during the remaining 2 h/day. Chow was available throughout the experiment.
This experiment repeated some of the tests of experiment 1 except that the order of nutrient presentation was changed. The intent was to examine the effect of previous experience on linoleic acid preferences. The mice were given a series of two-bottle (23 h/day) tests outlined in Table 1. In the first test (test 2.1), linoleic acid vs. vehicle (gum) was presented at linoleic acid concentrations of 0.0025 to 2%. A soybean oil vs. vehicle (Emplex) test was then conducted at ascending concentrations of 0.0025–16%. Midway into the test (after the 0.5% concentration), the mice were given the choice between 0.5% Sefa Soyate oil vs. vehicle followed by a test with 0.5% soybean oil vs. 0.5% Sefa Soyate (test 2.3). The mice were next retested with linoleic acid vs. vehicle (test 2.4) as in the first test.
Daily fluid intakes were expressed as grams of intake per mouse per day. Lipid and sucrose intakes were also expressed as kilocalories per day. Taste preferences were expressed as percent intakes (solution intake/total intake × 100). Between-strain differences were assessed using repeated-measures ANOVA. Significant interaction effects were evaluated using simple main effects tests according to Winer (37). The significance of the fluid preference at each concentration was evaluated for each strain by comparing test fluid intake vs. vehicle intake using paired t-tests. To control for the multiple comparisons, the α level (0.05 prior to correction) for the t-tests was corrected with the Bonferroni procedure that yielded a critical level of statistical significance that depended upon the number of concentrations within a test series.
The CD36 KO and B6 mice had similar body weights at the start (22.9 and 23.9 g) and end of the experiment (26.3 and 25.2 g).
Test 1.1: saccharin vs. water.
To test specificity of the effect of CD36 deletion on taste perception, WT and CD36 KO mice were compared for their preference for a nonnutritive sweetener. In the choice test with 0.2% saccharin vs. water, both the KO and WT mice consumed substantially more saccharin than water [F(1,18) = 280.24, P < 0.001] and their intakes (11.1 and 13.2 g/day) and preferences (96 and 99%) did not differ significantly.
Test 1.2: Sefa Soyate oil vs. vehicle.
To examine whether CD36 deletion impacts preference for nonnutritive oils, the mice were presented with Sefa Soyate vs. vehicle. Figure 1 presents the results expressed as Sefa Soyate preference (percent intake) and acceptance (absolute intake). Overall, Sefa Soyate oil intake increased as concentration increased from 0.313 to 2.5% [F(3,54) = 25.31, P < 0.001]. This effect was most pronounced in the WT mice [group × concentration, F(3,54) = 5.75, P < 0.01]. WT mice consumed more (P < 0.01) 2.5% Sefa Soyate oil than did CD36 KO mice and showed a greater increase in Sefa Soyate preference as concentration increased [F(3,54) = 7.85, P < 0.01]. The groups differed in their preference for 2.5% Sefa Soyate: WT mice consumed more (P < 0.05) 2.5% Sefa Soyate oil than vehicle, while the KO mice did not.
Test 1.3: Soybean oil vs. vehicle.
Absence of CD36 was reported to blunt the preference for fatty acids (16), but it remained undetermined whether it also impacts the preference for triglycerides. This was tested by comparing preferences for soybean oil vs. vehicle (Fig. 1). Oil intake increased with concentration [F(3,54) = 57.51, P < 0.001] with the change being greater for WT mice [F(3,54) = 11.04, P < 0.001], which consumed more 2.5% soybean oil than did the KO mice. The soybean oil preference of the WT mice also exceeded that of the KO mice, and this difference was significant at every concentration [F(3,54) = 9.74, P < 0.001]. The WT mice consumed more soybean oil than vehicle at all concentrations, and the KO mice consumed more oil at all but the 0.625% concentration.
Test 1.4: Soybean oil vs. Sefa Soyate oil.
The ability of the WT and CD36 KO mice to discriminate between nutritive and nonnutritive oils was compared directly in test 1.4. As illustrated in Fig. 2, when retested for their preference for 0.625% Sefa Soyate midway through test 1.3, the KO mice were again indifferent to the nonnutritive oil, whereas the WT mice consumed more (P < 0.01) oil than vehicle [group × fluid interaction, F(1,18) = 31.57, P < 0.001]. The Sefa Soyate preferences of the WT and KO mice were 87% and 56%, respectively [t(18) = 6.91, P < 0.01]. In the 0.625% Sefa Soyate oil vs. soybean oil test the KO mice were again indifferent, while the WT mice consumed more (P < 0.01) soybean oil than Sefa Soyate oil [group × oil interaction, F(1,18) = 28.27, P < 0.001]. Preferences of the WT and KO groups for soybean oil over Sefa Soyate oil were 80% and 53%, respectively [t(18) = 6.95, P < 0.01].
Test 1.5: linoleic acid, Sefa Soyate oil, and soybean oil tests.
Preference for fatty acids was tested using the conditions of Laugerette et al. (16). In the 2% linoleic acid preference test (Fig. 2), both groups consumed more linoleic acid than gum vehicle [F(1,18) = 295.05, P < 0.01]. The absolute and percent linoleic acid intakes of KO mice, however, were less (P < 0.01) than those of the WT mice. In the Sefa Soyate test, the KO and WT groups consumed more of the nonnutritive oil than gum vehicle [F(1,18) = 135.91, P < 0.01], and there were no group differences in oil intake or preference. Both groups also consumed more soybean oil than Sefa Soyate oil in the final test of this series, although the WT mice consumed more than twice as much soybean oil than did the KO mice [group × oil, F(1,18) = 7.15, P < 0.01].
Test 1.6: Intralipid vs. water.
This test compared lipid intakes of CD36 KO and WT over a wide range of oil concentrations and examined caloric compensation with 10% and 20% Intralipid. Figure 3 shows that both the CD36 KO and WT groups displayed strong preferences (>90%) for Intralipid. The groups did not differ in their percent intakes, although preferences varied somewhat over concentrations [F(6,108) = 3.51, P < 0.01]. The groups did differ, however, in their absolute intake of Intralipid [F(1,18) = 5.90, P < 0.05], and this effect varied with concentration [F(6,108) = 4.59, P < 0.001]. In particular, the WT mice consumed more (P < 0.05) Intralipid emulsion at the 1.25% and 2.5% concentrations than did the KO mice; the differences were also nearly significant (P < 0.06) at the 0.625% and 5% concentrations. Overall, emulsion intakes increased and then decreased as concentration increased from 0.313 to 20% [F(6,108) = 98.45, P < 0.001], while lipid calorie intake only increased [F(6,108) = 775.74, P < 0.001] and varied as a function of group [group × concentration, F(6,108) = 8.42, P < 0.001]. In particular, the WT mice consumed more (P < 0.05) energy as Intralipid than did the KO mice at concentrations of 2.5% to 20%. Analysis of the total energy consumed also revealed group differences. As illustrated in Fig. 4, when offered 10% and 20% Intralipid, the KO and WT mice compensated for their lipid energy by decreasing chow intake relative to their chow-only baseline [F(2,36) = 1,277.05, P < 0.001]. The compensation was less precise in WT mice, which increased their total energy intake over chow-only baseline when 20% Intralipid was available compared with the KO mice (111% vs. 97% of baseline, P < 0.001).
Test 1.7: sucrose vs. water.
The CD36 KO and WT groups both displayed strong preferences (>90%) for sucrose over water, but they differed in sucrose intake as a function of concentration [F(3,54) = 13.11, P < 0.01]. In particular, the KO mice consumed less than WT mice of 4% (22.3 vs. 34.5 g/day) and 8% (26.5 vs. 32.3 g/day) sucrose than did the WT mice. The KO and WT mice did not differ in their intake of 16% (16.5 vs. 17.4 g/day) and 32% sucrose (8.6 vs. 9.1 g/day), however.
To compare the postingestive effects of lipids in CD36 KO and WT mice, in test 1.8, the animals were infused intragastrically with Intralipid. Six CD36 KO and eight WT mice survived intragastric surgery and completed the experiment. The total fluid intakes (oral water plus intragastric water) before training were similar in the KO and WT mice (7.9 vs. 8.2 g/day). However, when given 0.1% saccharin to drink, the WT mice consumed more than did the KO mice (oral saccharin + intragastric water intake, 45.2 vs. 27.8 g/day, P < 0.001). The saccharin concentration offered to the WT mice was therefore reduced to 0.05% while the KO mice continued to receive 0.1% saccharin for 2 days. At these concentrations, the KO and WT groups consumed similar amounts (23.8 vs. 23.3 g/day). Therefore, the KO and WT groups were trained and tested with flavored 0.1% and 0.05% saccharin solutions, respectively. A recent study of B6 and 129 mice also used different sweetener concentrations to equate intakes and compare intragastric nutrient conditioning in the two inbred strains (31).
Figure 5 presents the one-bottle training and two-bottle test results. Analysis of the daily training intakes revealed that CS+ intakes increased and CS− intakes decreased over the 6 training days, and the intake patterns were similar in the two groups [CS × days, F(2,24) = 11.43, P < 001]. The KO mice consumed less CS+ and CS− than did the WT mice during one-bottle training [F(1,12) = 20.50, P < 0.001]. In the two-bottle choice tests, both groups consumed substantially more CS+ than CS− [F(1,12) = 111.75, P < 0.001] and consumed more CS+ in the reinforced test than the nonreinforced test [CS × test interaction F(1,12) = 29.06, P < 0.001]. Overall, the KO mice consumed less than did the WT mice in the two-bottle tests [F(1,12) = 10.22, P < 0.001], and this difference was greater for the CS+, although the group × CS interaction was marginal [F(1,12) = 4.46, P = 0.056]. The KO and WT groups did not differ in their CS+ preference expressed as percent CS+ intakes (91% vs. 91% averaged over reinforced and nonreinforced tests), and both groups decreased their preference somewhat from the reinforced to the nonreinforced test [96% to 86% averaged over groups, F(1,12) = 8.86, P < 0.05].
The CD36 KO and WT mice had similar body weights at the start (24.5 and. 23.9 g) and end of the experiment (28.1 and 27.6 g).
Test 2.1: linoleic acid vs. vehicle.
Our findings in experiment 1, test 1.5 of a preference for linoleic acid in CD36 KO mice differ from those of Laugerette et al. (16). However, the animals in test 1.5 had prior experience with nutritive oil emulsions, which may have altered their response to linoleic acid. Test 2.1 therefore examined linoleic acid preferences in oil-naïve mice. As illustrated in Fig. 6, the WT mice displayed a stronger preference for linoleic acid than did the KO mice with the difference being significant at the 1 and 2% concentrations [group × concentration, F(5,85) = 4.02, P < 0.01]. The WT also consumed more linoleic acid than did the KO mice at the 2% concentration [group × concentration, F(5,85) = 3.16, P < 0.05]. Within-group comparisons indicated that the WT mice consumed more linoleic acid than vehicle at the 0.5–2% concentrations, but this difference was not significant due to one outlier mouse that avoided linoleic acid. With this animal excluded, the differences in linoleic acid and vehicle intake were significant. The KO mice did not consume significantly more linoleic acid than vehicle at any concentration.
Test 2.2: soybean oil vs. vehicle.
As illustrated in Fig. 7, the KO and WT mice displayed very similar preferences for soybean oil, which increased with concentration [F(8,136) = 183.65, P < 0.001]. Within-group comparisons indicated that both groups consumed more soybean oil than vehicle starting at the 0.25% concentration. Although percent intakes did not differ, absolute intakes of the soybean oil emulsion did differ: the WT mice consumed more than did the KO mice at 2–8% concentrations [group × concentration, F(8,136) = 5.64, P < 0.001]. Analysis of the caloric intake data indicated that the WT mice consumed more lipid energy than did the KO mice at the 4 to 16% concentrations [group × concentration, F(8,136) = 3.33, P < 0.01].
Test 2.3: soybean oil and Sefa Soyate oil.
When first given 0.5% Sefa Soyate oil, both groups consumed more oil than vehicle [KO mice: 5.5 vs. 1.1 g/day; WT: 5.9 vs. 1.0 g/day; F(1,17) = 96.12, P < 0.001] and they did not differ in their oil intakes. In the choice test with the two oils, overall, the mice consumed more soybean oil than Sefa Soyate oil [F(1,17) = 15.83, P < 0.01]. The group × oil interaction was not significant (P = 0.075), but post hoc tests indicated that the WT mice consumed more soybean oil than Sefa Soyate [5.5 vs. 1.8 g/day, P < 0.01]; the KO mice did not differ significantly in their oil intakes (4.0 vs. 2.7 g/day).
Test 2.4: linoleic acid vs. vehicle.
To directly examine the effect of previous experience on preference for linoleic acid, a second test similar to test 2.1 was conducted. As illustrated in Fig. 6, in the second test with linoleic acid, the KO and WT mice showed similar preferences that varied as a function of concentration [F(5,85) = 12.90, P < 0.001]. In particular, linoleic acid preference increased as concentration increased from 0.0025% to 0.25%, then remained stable until it declined at the 2% concentration. The KO and WT mice were also similar in their absolute intakes of linoleic acid, which varied as a function of concentration [F(5,85) = 16.45, P < 0.001]. The within-group analysis indicated that the KO mice consumed significantly more linoleic acid than vehicle at 0.025% to 2% concentrations; for the WT mice, the differences were significant at 0.25% to 1%. At the 2% concentration, there was one outlier in each group that strongly avoided linoleic acid. Excluding these animals, both groups consumed more linoleic acid than vehicle. Compared with the first linoleic acid test (test 2.1), the KO mice displayed higher linoleic preferences at all concentrations except 0.025% [test × concentration, F(5,45) = 2.83, P < 0.05]. For the WT mice, preferences were higher only at the 0.25% and 0.5% concentrations compared with the first linoleic test [F(5,40) = 3.70, P < 0.05].
The present study confirms the recent report (16) that naive CD36 KO mice, unlike WT mice, are indifferent to 2% linoleic acid in two-bottle tests and further reveals that KO mice fail to prefer soybean oil and Sefa Soyate oil at dilute concentrations. However, the current results also demonstrate that after experience with oil emulsions, KO mice develop preferences for linoleic acid, soybean oil, and Sefa Soyate oil at low to high concentrations. These findings together with the results of the intragastric infusion experiment indicate that fat preferences in KO mice can be “rescued” by the post-oral conditioning effects of fat. Yet KO mice continue to show a reduced fat acceptance, as indicated by their below-normal intakes of concentrated soybean oil emulsions.
In test 1.3, the CD36 KO mice were indifferent to soybean oil until the 2.5% oil concentration, which they preferred to vehicle. The WT mice, in contrast, preferred all soybean oil concentrations to vehicle. These oil results confirm the prediction that KO mice would show a reduced preference for triglyceride oil, as well as free fatty acids. This prediction was based on the findings that lingual lipase rapidly hydrolyzes triglyceride oil to free fatty acid on the rodent tongue and that blocking this hydrolysis with orlistat reduces triglyceride oil preference (15). CD36 is localized in cells lining the circumvallate papillae, which are exposed to locally released lingual lipase. Consequently, KO mice may show less preference for triglyceride oil emulsions because they do not detect the free fatty acids released near the taste cells.
KO mice, unlike WT mice, were also initially indifferent to Sefa Soyate oil (test 1.2), which is not hydrolyzed by lingual lipase. This finding was unexpected, as we predicted that KO and WT mice would show similar ingestive responses to the nonnutritive oil based on its textural properties. Sefa Soyate consists of free fatty acids attached to a sucrose core; conceivably, the free fatty acid tails may bind with CD36 on taste cells and therefore produce a fatlike taste. Sefa Soyate also contains a small amount of unbound free fatty acids (<0.05% in pure oil). Even if Sefa Soyate does have a fatlike taste, as well as mouth-feel, WT mice preferred 0.625% soybean oil to Sefa Soyate oil in test 1.3. The release of fatty acids by the action of lingual lipase on the soybean oil but not Sefa Soyate could account for this preference. The KO mice, in contrast, did not discriminate between Sefa Soyate oil and soybean oil in test 1.3, which could be due to their inability to detect the different fatty tastes of the two emulsions. These results suggest that CD36 could contribute to triglyceride perception both in the presence and absence of lingual lipase. This interpretation would be consistent with the ability of the molecule to bind not only fatty acids but a variety of other lipids, including phospholipids and lipoproteins (reviewed in Ref. 24). KO mice also show a weaker preference for dilute mineral oil emulsions compared with WT mice, and the reason for their reduced responsiveness to nonnutritive oils requires further study (unpublished observations).
While indifferent to 0.313 to 1.25% soybean oil emulsions in test 1.3, the KO mice displayed a significant preference (86%) for 2.5% soybean oil that approached the strong preference displayed by the WT mice (96%). One interpretation of this finding is that 2.5% soybean oil has orosensory properties, other than CD36-mediated fatty taste, that are inherently attractive to KO and WT mice alike. Alternatively, the post-oral reinforcing action of 2.5% soybean oil may condition an attraction to its flavor. Supporting this interpretation are the significant flavor preferences conditioned by the intragastric Intralipid infusions in test 1.8. Note that while the Intralipid infusates contained 5% soybean oil, this was diluted to 2.5% by the matched volumes of ingested CS+ solution. Thus, it is reasonable to assume that the mice drinking 2.5% soybean oil experience post-oral reinforcement that increases their evaluation of the oil's flavor.
After dietary experience with 2.5% soybean oil in test 1.3, the KO mice showed a significant preference for 2% linoleic acid that was nearly as strong as that displayed by the WT mice (90 vs. 98%; test 1.5). This finding is particularly notable because the mice had no prior experience with linoleic acid or the xanthan gum vehicle, which is thought to mask the viscous mouth-feel of oil (16, 36). The finding that KO mice readily discriminated between linoleic acid, as well as Sefa Soyate oil vs. plain gum indicates that the orosensory properties of the oils, other than CD36-mediated fatty taste, are distinctive to oil-experienced mice. Oily mouth-feel (lubricity) may be one factor that allowed the KO mice to prefer linoleic acid and Sefa Soyate to plain gum (26, 27). Yet, the KO and WT mice also preferred soybean oil to Sefa Soyate oil, which suggests that oily texture alone is not sufficient to explain the observed oil preferences. It may be that the mice discriminated between soybean oil and Sefa Soyate oil based on odor cues. Although an intact olfactory system is not essential for oil preferences in mice, animals may learn to associate olfactory stimuli with the post-oral reinforcing actions of nutritive oils (9, 35). Note that the CS flavors used in the intragastric conditioning experiment are distinguished primarily by their odor.
The linoleic acid preference displayed by the oil-experienced KO mice in test 1.5 is in contrast to the lack of preference observed with KO mice by Laugerette et al. (16). Experiment 2, though, replicated Laugerette's findings in oil-naive animals. In test 2.1 the percent intakes of 2% linoleic acid of the KO and WT mice were 44% and 76%, respectively, which are comparable to the ∼55% and ∼82% preferences reported by Laugerette et al. (16) in KO and WT mice, respectively. Yet, when retested with linoleic acid in test 2.4 after being exposed to a range of soybean oil emulsions, the KO and WT mice displayed equivalent (77 and 79%) preferences for 2% linoleic acid. The two groups also showed similar preferences for the 0.25–0.5% linoleic acid emulsions (86–94%), which were actually stronger than the preferences for 2% linoleic acid. The decline in linoleic acid preference at the 2% concentration suggests that, unlike triglycerides (12, 22, 38), the palatability of linoleic acid decreases at higher concentrations. Post-oral factors may have contributed to the reduced preferences observed for 2% linoleic acid in the 24-h tests of experiment 2. A recent study, however, obtained a similar decline in the licking response to linoleic acid in short-term tests (38). It may be that the irritant properties of fatty acids, as reported by humans (see Ref. 2), reduce the palatability at higher concentrations.
While the KO mice failed to prefer linoleic acid in test 2.1, their experience with the fatty acid appears to have enhanced their preference for soybean oil. The KO mice showed significant preferences for dilute soybean oil (0.25%-1%) in test 2.2, whereas the KO mice in the first experiment failed to prefer soybean oil over a similar concentration range (0.313–1.25%) in test 1.3. The primary difference between the two KO groups is that mice in experiment 1 were tested, prior to the soybean oil, with Sefa Soyate oil, whereas the mice in the second experiment were tested with linoleic acid. Conceivably, the post-oral actions of the linoleic acid were sufficient to condition an oil preference even though it was not expressed during the linoleic acid test. Note, though, that the KO mice in experiment 2, unlike the WT mice, did not significantly prefer 0.5% soybean oil to 0.5% Sefa Soyate oil in test 2.3. This indicates that while they discriminated oil vs. vehicle, they did not discriminate between nutritive and nonnutritive oils. After more extensive experience with soybean oil, the KO mice in the first experiment displayed a significant preference for soybean oil over Sefa Soyate oil (test 1.5). This suggests that post-oral effects were responsible for their soybean oil preference.
In both experiments, the oil-experienced CD36 KO mice and WT mice displayed similar near-total preferences for concentrated (2.5–20%) soybean oil emulsions. Yet, the KO mice consistently consumed less soybean oil compared with the WT mice. Conceivably, the KO mice may have underconsumed the oil emulsions because of their insensitivity to fatty taste. Alternatively, a blunted cephalic phase response to fatty taste in KO mice may have impaired their digestive processing of the lipids. Laugerette et al. (16) reported that oral exposure to linoleic acid increased pancreatobiliary secretions in WT mice and that this response was absent in KO mice. A third possibility is that KO mice consume less of lipid-rich foods and fluids because of altered intestinal processing of lipids due to their lack of intestinal CD36 (5, 23). The results of the intragastric infusion experiment (test 1.8) are consistent with this idea. During one-bottle training the KO mice consumed 36% less of the CS+ paired with intragastric Intralipid infusions than did the WT mice; the same mice consumed 39% less of the 2.5% Intralipid emulsion than did the WT mice in test 1.6. In the intragastric experiment, the mice consumed flavored saccharin solutions by mouth, and thus differences in fatty taste and/or fatty taste-elicited cephalic responses were not a factor in their reduced intakes of the CS+ and paired intragastric Intralipid infusions.
While their intestinal processing of lipids may be compromised, the KO mice still acquired a significant preference for and increased acceptance of the CS+ flavor relative to the CS− flavor in the intragastric conditioning experiment. These findings indicate that CD36 does not mediate the post-oral positive reinforcing actions of lipids. The KO mice also showed precise caloric compensation in test 1.6 when drinking the 10% and 20% Intralipid emulsions. They reduced their chow intake to compensate for their lipid intake, so that total energy consumption remained unchanged from the chow-only baseline period. The WT mice, in contrast, increased their total energy intake with the 20% emulsion. Together, these findings help explain why WT mice gain more weight than KO mice when fed a high-fat diet for a prolonged period (13).
Sweetener Preference and Acceptance
In their original study of CD36 KO mice, Laugerette et al. (16) reported that KO mice had a normal preference for sucrose (4%) and an aversion to quinine (0.3 mM). In the present study, KO mice were similar to WT mice in their preference for saccharin (0.2%, test 1.1) and for sucrose solutions (4–32%; test 1.7), which further indicates that KO mice do not have a general taste deficit. However, unlike the Laugerette et al. (16) study, the KO mice in test 1.7 consumed less 4% sucrose than did the WT mice. The 4% sucrose solution intake (34.2 g/day) of WT mice in test 1.7 was higher than that observed with other B6 mice studied in this laboratory (25–28 g/day) (28, 30). The reason for the elevated sucrose intakes of the B6 mice in the present study is not certain, but it may be related to their prior experience with oil emulsions of increasing palatability (Sefa Soyate, soybean oil, Intralipid). The sugar alcohol (glycerol) in Intralipid may give it a slightly sweet taste, which may be enhanced by the fatty taste of the emulsion (11). This experience may have sensitized the WT mice to the intake-stimulating actions of the dilute sucrose solutions. Furthermore, it may explain the differential saccharin intakes of the WT and KO in experiment 1B.
Fat Taste, Flavor, and Preference
A fatty taste signaling function for CD36 is suggested by its localization in lingual taste bud cells (8, 16), the lack of linoleic acid preference in KO mice (16) and the lack of a pancreatobiliary response in KO mice to oral linoleic acid (16). The present study confirmed that naive KO mice, unlike WT mice, fail to prefer linoleic acid. In addition, it showed that CD36 deletion attenuates preferences for triglycerides as well as fatty acids. The exact mechanism by which CD36 mediates oil preferences remains to be determined, but other findings support gustatory system involvement. Gustatory nerve transection impairs fatty acid detection and/or preference in rats and mice (17, 25, 34). In addition, knockout mice missing Trpm5 or P2X2/P2X3 taste signaling proteins essential for sweet and bitter taste processing are indifferent, like CD36 mice, to dilute soybean oil emulsions (29, 32). Whether CD36 interacts with these proteins remains to be determined. α-Gustducin knockout mice, in contrast, display normal preference responses to oil emulsions (29), despite the colocalization of α-gustducin and CD36 in some taste cells (16). This later finding indicates that α-gustducin does not mediate the taste response stimulated by fatty acids.
Although there is accumulating evidence that CD36 mediates, at least in part, the taste response to fatty acids, the present results demonstrate that it is not essential for the development of robust fat preferences. With experience, the KO mice displayed significant preferences for linoleic acid and soybean oil emulsions that were as strong as those observed in WT mice. These acquired preferences can be attributed to a learned association between CD36-independent oral stimuli and the post-oral reinforcing actions of fat. This is demonstrated by the flavor preferences conditioned by intragastric Intralipid infusions in both WT and KO mice. Thus, although fatty taste contributes to the palatable flavor of fat, it is not essential for the expression of profound fat preferences. The mouth-feel, aroma, and perhaps other sensations of fat apparently provide adequate stimuli for rodents to acquire preferences for even dilute oil emulsions. However, despite the experience-induced rescue of fat preferences in CD36 KO mice, their oil intake remained consistently lower than that of WT mice. The mice were also more precise in reducing their chow consumption to compensate for the increased energy from oil. This indicates that CD36 function may modulate food intake and the susceptibility to diet-induced weight gain.
CD36 deficiency results in multiple metabolic abnormalities that have been described in detail in earlier publications (14, 19) and include abnormalities of blood lipids and adipokines and altered insulin sensitivity. The specific contribution of the fat taste defect and of decreased fat acceptance to the general metabolic phenotype is not known and cannot be evaluated from the data in the present study, but this will be possible in the future with the generation of mice deficient for taste bud or intestinal CD36, specifically. CD36 deficiency and polymorphisms in the CD36 gene are relatively common in humans and there is strong evidence for their association with abnormalities in the metabolism of both lipid and glucose (4, 14, 19). It will be important to examine whether these polymorphisms impact fat acceptance and energy regulation in humans.
This research was supported by National Institutes of Health Grants DK031135 (AS), DK033301, and DK060022 (NAA).
The authors thank Martin Zartarian, Steven Zukerman, and Terri Pietka for their expert technical assistance and John I. Glendinning for his comments on this paper.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society