It is well established that rats control their protein intake. When offered two diets varying in protein concentration, rats maintain a protein intake above their requirement for maintenance and growth, but do not regulate their protein intake at a precise level (Harper and Peters 1989, Leathwood 1987, Peters et al. 1983, Ross et al. 1983). Under these conditions, rats select between 18 and 43% of their diet as protein (Peters et al. 1983), depending on the protein content of the diets offered (Peters et al. 1983) and the type (quality) of protein offered (Ashley and Anderson 1975, Peters and Harper 1981). Under normal conditions, protein intake is controlled by dietary choice, and only under extreme conditions does dietary protein affect the quantity of food ingested. Such conditions include presentation to the rat of a single food source containing either a low or high protein diet, amino acid imbalanced diets, diets containing excesses of a single amino acid or diets devoid of, or severely deficient in, an essential amino acid (reviewed in Rogers and Leung 1977).

Under deficient states, the animal is no longer allowed the option of maintaining adequate protein intake, and the first limiting amino acid becomes the nutrient limiting for growth. In this situation, the concentration of the dietary limiting amino acid (LAA)⁴ may be the signal used for dietary selection. Mechanisms which would prevent protein-deficient rats from selecting diets more deficient in the LAA and aiding in selection of diets containing greater quantities of the LAA have obvious evolutionary advantages. Ashley and Anderson (1975) and Peters and Harper (1981) found that rats allowed to select between 15 and 55% wheat gluten diets (deficient in lysine) will decrease the proportion of diet they consume as wheat gluten when the diets are supplemented with lysine. However, this effect is not evident when rats select between two diets containing protein sources that are not limiting in an amino acid. It seems that the dietary limiting amino acid can affect dietary selection, but only when the diets are deficient in the LAA.

The physiological mechanisms involved in recognizing deficient diets may be similar to those used to detect amino acid-imbalanced diets. Imbalanced diets are believed to simulate an amino acid deficiency because the essential amino acids added to a low protein diet transitorily stimulate visceral protein synthesis (Benevenga et al. 1968, Noda et al. 1976) and increase threonine degradation (Davis and Austic 1994), subsequently reducing the plasma concentration of the LAA as the concentrations of the other essential amino acids increase in the plasma (Leung et al. 1968, Peng et al. 1972). The excesses of other essential amino acids compete with the LAA for uptake at the blood brain barrier and thereby exacerbate the reduction in LAA concentration in the brain (Peng et al. 1972, Tews et al. 1978) and prepiriform cortex (Gietzen et al. 1986). This decreased LAA concentration in the brain may signal that the imbalanced diet is more deficient than the original low protein diet to which the rats were adapted. Because rats develop a conditioned taste aversion to an imbalanced diet and select against it when given a choice, several experiments were conducted to determine whether decreasing the concentration of the LAA in low protein diets by very small amounts would cause rats to select against these more deficient diets.

Table 1. Dietary composition of threonine-deficient diets for experiments 1, 3 and 4 [View Table]

In preliminary experiments, we adapted growing rats to a purified low protein (Basal) diet in which threonine (Thr) was the LAA (0.200 g/100 g diet). Rats given a choice between this diet and an identical diet in which the threonine concentration was decreased to 0.188% threonine, selected the 0.200% Thr-Basal diet over the 0.188% Thr-Basal diet almost exclusively by d 3, even though the only difference between diets was 0.012% threonine. The goals of the present experiments were as follows: 1) to determine whether the dietary selections observed are the result of neophobic or neophilic responses, 2) to determine how sensitive rats are to decreases in dietary threonine concentration, and 3) to determine whether plasma and brain amino acid concentrations, which are known to be affected by amino acid-imbalanced diets, are also affected by such small changes in the concentration of the LAA.

MATERIALS AND METHODS

Adaptation to the Basal diets was followed by a base-line period in which rats were offered their Basal adaptation diet in two food cups each at opposite ends of the cage, and intake was recorded to determine base-line intake and side preferences. During the experimental period, which immediately followed the base-line period, rats were offered two diets that differed only in the concentration of the LAA. Food intake was recorded daily and corrected for spillage. All food cups were washed at the beginning of the base-line and experimental periods. For each experiment, all diets were made at the same time and separated into aliquots so that fresh diet was used at the beginning of the base-line and experimental periods to prevent any contamination that might affect the sensory characteristics of the diet.

Table 2. Dietary treatments during the adaptation and choice periods of experiment 1 [View Table]

Table 3. Dietary composition of threonine-deficient diets used in Experiment 2 [View Table]

Brain. Amino acid concentrations were determined in prepiriform cortex (PPC) and anterior cingulate cortex (ACC). The PPC is involved in the initial response to amino acid imbalance, whereas the ACC is involved in the adaptive response and has been used as a control area (Gietzen et al. 1988, Leung and Rogers 1971, Meliza et al. 1983). Approximately 1-mm thick slices of frozen brain were dissected on histology slides over dry ice. PPC and ACC were cut from slices 9.6-9.8 ± 1.0 mm rostral to the interaural line according to the atlas of Pellegrino and Cushman (1967). Tissue was weighed, and sonicated in 50 × perchloric acid diluent (0.2 mol/L perchloric acid, with 0.5 g disodium EDTA and 100 mg sodium bisulfite/L). After sonication, samples were centrifuged at 15,000 × g for 10 min in an Eppendorf microcentrifuge housed in a cold room. Supernatants were filtered through microfilters (0.45-µm pore size), and aliquots were stored at 80°C for amino acid analysis or monoamine determination.

Concentrations of amino acids were determined using an automated amino acid analyzer (Beckman 7300, Beckman Instruments, Palo Alto, CA). Monoamine concentrations in the PPC were analyzed by HPLC as described by Gietzen et al. (1986), using the method of Wagner et al. (1982). Briefly, samples were injected onto a C18, reversed-phase column (dimensions 100 × 4.6 mm, Perkin-Elmer, Torrence, CA) and the eluate monitored by electrochemical detection. The mobile phase contained 37.5 mmol/L sodium hydroxide, 1.71 mmol/L 1-octanesulfonic acid, 110 mmol/L citric acid, 0.95 mmol/L EDTA and 5.0% (v/v) acetonitrile at pH 3.15. A glassy carbon electrode was used for electrochemical detection at +850 mV potential, with a 2.0-s RC filter (Bioanalytical Systems, West Lafayette, IN).

RESULTS

Table 4. Intake of the less deficient Thr-Basal diet during Experiment 11 [View Table]

Rats that were given a choice between their adaptation diet and a less deficient diet tended to select the less deficient diet on d 1, but made no choice for either diet thereafter, resulting in no significant selection effect (P > 0.15) (Table 4, treatment). It is not clear whether the selection for the diet containing the higher threonine concentration on d 1 was the result of neophilia, chance or some physiological response caused by the dietary threonine concentration.

Rats that received a choice between 2 "novel diets" (Table 4, treatment 3), one less deficient and one more deficient than their adaptation diet, chose the diet containing the higher threonine concentration (P < 0.03). Total food intake was not significantly different among treatments (P > 0.05).

Table 5. Amino acid concentration in the plasma of rats fed Thr-Basal diets containing 0.188, 0.200 and 0.212% Thr for 6 h1 [View Table]

Table 6. Amino acid concentration in the prepiriform cortex of rats fed Thr-Basal diets containing 0.188, 0.200 and 0.212% Thr for 6 h1 [View Table]

Table 7. Amino acid concentration in the anterior cingulate cortex of rats fed Thr-Basal diets containing 0.188, 0.200 and 0.212% Thr for 6 h1 [View Table]

Table 8. Amino acid concentrations in the plasma of rats fed Thr-Basal diets containing 0.188 and 0.212% Thr for 54 h1 [View Table]

Table 9. Amino acid concentrations in the prepiriform cortex of rats fed Thr-Basal diets containing 0.188 and 0.212% Thr for 54 h1 [View Table]

DISCUSSION

Booth (1985) and Rozin and Rodgers (1967) suggest that under states of nutrient deficiency (or excess) a rat will alter its behavior and become increasingly neophilic. This increases the probability that the rat will alleviate the deficiency or toxicity by finding a new food source. Rats fed a diet lacking a specific vitamin for several weeks select a novel diet containing the deficient vitamin over a diet that does not contain the vitamin (Scott and Quint 1946, Scott and Verney 1947). This may occur because the rat has acquired a nutrient-specific hunger, but Booth (1985) suggests that most acquired nutrient-specific hungers are artifacts and are expressed because the animal has developed a conditioned aversion to the available deficient diets. Rozin and Rodgers (1967) observed that thiamine-, pyridoxine- and riboflavin-deficient rats showed a strong preference for novel diets and through a series of experiments concluded that the "specific hungers" were the result of a learned aversion to the vitamin-deficient diets. A key point in Rozin and Rodgers experiments in which deficient rats preferred novel diets is that the novel diets offered in place of the deficient diets contained novel ingredients and could easily be detected as "novel."

In contrast to the classic vitamin deficiency experiments, the diets used in Experiment 1 differed only in the quantity of threonine and did not contain novel ingredients or flavors. This paradigm is similar to one of Scott and Quint (1946), who made rats thiamine deficient by feeding them a vitamin-free diet plus a supplemental pill that contained all vitamins except thiamine. When the rats were then given a choice between the vitamin-free diet that they had been prefed and the same diet containing a vitamin mix that included thiamine, it took them an average of 1.3 d before they chose the vitamin-containing diet almost exclusively. It seems reasonable that if rats were making this selection based on neophilia, they would have made the choice sooner. Scott and Verney (1949) found that rats adapted to a diet low in thiamine (0.1-1.0 µg thiamine/gm diet) would select this diet over the same diet devoid of thiamine. However, when they were adapted to diets containing adequate thiamine, they did not select between their adaptation diet and a devoid diet. It appears that rats fed a diet marginally deficient in thiamine can select on the basis of thiamine concentration rather than simply on the basis of neophilia. This type of diet selection may be due to the fact that the only difference between the diets rats were allowed to select was the thiamine concentration or a vitamin mixture, so that rats theoretically never received a "novel" tasting diet.

It is interesting that rats in Experiment 1 that were fed the 0.188% Thr-Basal diet and then given a choice between this diet and the 0.200% Thr-Basal diet did not make a consistent choice for the better diet. Mori et al. (1991) have shown that rats fed a diet limiting in lysine can select among 15 water bottles to obtain the one containing lysine. Rogers and Harper (1970) demonstrated that rats fed a histidine-imbalanced diet will select a histidine·HCl solution over water or dilute HCl solution to obtain more histidine. Under amino acid-deficient states, rats clearly select solutions containing more of the limiting amino acid; however, their sensitivity to small increases in the dietary LAA concentration may not be as great as their sensitivity to reductions in dietary LAA concentration. It is likely that the small increase from 0.188 to 0.200% threonine was not adequate for the rats to detect, because the extra threonine was probably utilized for protein synthesis by gut and/or liver and did not have an opportunity to affect amino acid concentrations of plasma or other tissues.

Experiment 2 estimated the threshold at which rats can detect decreases in the dietary limiting amino acid. Rats adapted to the 0.200% Thr-Basal diet were able to detect the difference between this diet and the 0.191% Thr-Basal diet. This is a difference of 90 mg threonine/kg diet, or 90 ppm. It took d 6 before rats stopped consuming the 0.191% Thr-Basal diet, but the rats had no flavor cues on which to base their selection. Because rats offered a choice between the 0.200 and 0.188% Thr-Basal diets selected the diet containing the higher concentration of threonine by d 3, it is probable that the relative change in amino acid concentration between diets affects the time required to detect the deficiency and respond appropriately. This has been shown consistently with amino acid-imbalanced diets; rats fed severely imbalanced diets will reduce their food intake more rapidly than rats fed only mildly imbalanced diets. Several investigators (Harper and Kumta 1959, Henderson et al. 1953, Kumta and Harper 1960, Sanahuja and Harper 1963) have produced amino acid imbalances by adding small amounts of the second limiting amino acid to a low protein diet. In their studies, addition of amino acids at levels of 0.05-0.06% of the diet can both create and correct imbalances. The experiments presented in this paper demonstrate that rats can detect even smaller differences when they are allowed to select between diets.

It is known that the concentration of the dietary limiting amino acid is reduced in both plasma (Leung et al. 1968, Peng et al. 1972) and brain (Gietzen et al. 1986, Peng et al. 1972) of rats fed amino acid-imbalanced diets. It is assumed that these changes are critical for detection of amino acid imbalance/deficiencies because infusion of the limiting amino acid into the carotid artery (Leung and Rogers 1969, Tobin and Boorman 1979) or injection directly into the PPC (Beverly et al. 1990a and b) can increase intake of imbalanced diets. Gietzen et al. (1986) found that the concentration of the dietary limiting amino acid and norepinephrine are both significantly reduced 2-3 h after the start of the dark cycle in the PPC of rats fed amino acid-imbalanced diets. This is about the time at which rats start to reduce their intake of the imbalanced diet.

We chose to sample plasma and brian amino acid concentrations 6 h after the start of the dark cycle because Morii et al. (1991) and Peng et al. (1972) observed that the drop in the limiting amino acid in plasma is greatest at about the middle of the dark cycle. Because the difference in dietary limiting amino acids is so small between these diets, it was thought that 6 h after after the start of the dark cycle would be the best time to detect changes if they were occurring. In Experiment 3, threonine concentration in plasma and ACC was not significantly affected by dietary treatment, nor were the concentrations of any other amino acids affected by dietary treatment 6 h after the start of the dark cycle on d 1. PPC threonine concentration was also not affected at the 6-h point, but did tend to reflect the small changes in dietary threonine. On d 3 (Experiment 4), when rats typically alter their dietary selection patterns in favor of the less deficient diet, amino acid concentrations in the plasma and PPC were not significantly affected by dietary treatments. Although threonine concentration in the PPC was not significantly different at 54 h, it was 33% lower in rats fed the 0.188% Thr-Basal diet than in rats fed the 0.212% Thr-Basal diet. This difference is greater than expected based on the differences in dietary threonine concentration. Norepinephrine was not affected by dietary treatment 6 h after the start of the dark cycle on d 1 (Experiment 3) or d 3 (Experiment 4) of diet presentation. It is uncertain whether dietary treatments did not affect plasma and brain amino acids and PPC norepinephrine concentrations or whether timing plays a critical role in their detection. It is possible that changes in plasma and brain amino acids were too small to detect by conventional analysis, or that they are transitory and detection requires sampling at the appropriate time. It is also possible that the rate of the drop in LAA concentration in plasma or elsewhere rather than the absolute concentration of the LAA is important for detection of a deficiency. It is not surprising that such small changes in dietary amino acid concentrations did not significantly alter plasma or brain amino acid patterns. From these results it is clear, however, that amino acids can affect feeding without causing major changes in plasma amino acid patterns.

In summary, rats adapted to threonine-limiting diets selected diets based on the limiting amino acid concentration. This selection appears to be due to learned associations between dietary limiting amino acid concentrations and postingestive consequences rather than on neophobia or neophilia. Rats adapted to a 0.200% Thr-Basal diet clearly responded to decreases as small as 0.009% in the concentration of threonine and selected against the more deficient diet when given a choice between it and their adaptation diet. The small changes in dietary threonine concentration (0.012%) that can cause rats to alter their dietary selection did not cause significant changes in plasma or brain amino acid concentrations, or prepiriform norepinephrine concentration.

ACKNOWLEDGEMENTS