Palatable Foods, Stress, and Energy Stores Sculpt Corticotropin-Releasing Factor, Adrenocorticotropin, and Corticosterone Concentrations after Restraint

Abstract

Previous studies have shown reduced hypothalamo-pituitary-adrenal responses to both acute and chronic restraint stressors in rats allowed to ingest highly palatable foods (32% sucrose ± lard) prior to restraint. In this study we tested the effects of prior access (7 d) to chow-only, sucrose/chow, lard/chow, or sucrose/lard/chow diets on central corticotropin-releasing factor (CRF) expression in rats studied in two experiments, 15 and 240 min after onset of restraint. Fat depot, particularly intraabdominal fat, weights were increased by prior access to palatable food, and circulating leptin concentrations were elevated in all groups. Metabolite concentrations were appropriate for values obtained after stressors. For unknown reasons, the 15-min experiment did not replicate previous results. In the 240-min experiment, ACTH and corticosterone responses were inhibited, as previously, and CRF mRNA in the hypothalamus and oval nucleus of the bed nuclei of the stria terminalis were reduced by palatable foods, suggesting strongly that both neuroendocrine and autonomic outflows are decreased by increased caloric deposition and palatable food. In the central nucleus of the amygdala, CRF was increased in the sucrose-drinking group and decreased in the sucrose/lard group, suggesting that the consequence of ingestion of sucrose uses different neural networks from the ingestion of lard. The results suggest strongly that ingestion of highly palatable foods reduces activity in the central stress response network, perhaps reducing the feeling of stressors.

Palatable food intake modifies autonomic and neuroendocrine components of the central stress corticotropin-releasing factor network, suggesting that activation of this network decreases with increased metabolic well-being.

Stressors stimulate activity in brain and thence in the hypothalamo-pituitary-adrenal (HPA) axis ultimately increasing circulating glucocorticoids secreted by the adrenal cortex. Predicting the magnitude and timing of ACTH and glucocorticoid responses to stressors has proven to be very difficult (1,2). Some animals and people secrete high concentrations of ACTH and glucocorticoids for prolonged periods, whereas others, under apparently the same conditions and stimuli, secrete much less of these hormones at a given time for a shorter duration, and it is likely that this depends on prior history of the individual (3). However, glucocorticoid secretion is very tightly controlled, and it is likely that the differential magnitudes of adrenal responses observed are also a consequence of regulatory systems that modulate the central control of the HPA axis.

In addition to fast and delayed negative feedback regulation of stress responses by adrenal glucocorticoid secretion (4), there are two regulatory inputs to the hypothalamic motor component [neurons that secrete corticotropin-releasing factor (CRF)] of the HPA axis that appear to provide very important modulation of the amplitude of responses to stressors. These are the amount of energy that is stored as fat in the body and the availability of coping, or displacement behaviors (reviewed in Refs. 5,6). We have previously shown that allowing adrenalectomized rats ad libitum high-density (30%) sucrose (but not energy free saccharin) to drink under basal conditions normalizes all measured components of the HPA axis in the absence of corticosterone (7,8,9), demonstrating that the extra sucrose calories can substitute for the steroid in the absence of a stressor. Furthermore, we found that allowing the choice of eating lard ad libitum in addition to chow decreases HPA responses to acute restraint (10). Finally, we have shown that ad lib provision of both 32% sucrose and lard as well as chow diminishes HPA responses to the repeated stressor of restraint (11). In all of those cases, the rats had access to calorie sources that were pleasurable and increased their fat stores when eaten regularly.

The present studies were performed for several reasons. We hypothesized that the availability of palatable foods (ad libitum access to 32% sucrose and to lard) during the week before a restraint stressor would serve to inhibit both basal and acute hypothalamic CRF mRNA and heteronuclear RNA (hnRNA) levels on the day of restraint and that, in parallel, both ACTH and corticosterone responses would be decreased (12). Because of previous findings (7,13), we thought that sucrose drinking might increase amygdala CRF expression after restraint but that perhaps lard eating would not affect it. Finally, we hypothesized that, because high glucocorticoids, in the presence of insulin, are known to increase both the short- and long-term intake of palatable foods (5,8,9,11,14,15), after the acute restraint-induced glucocorticoid responses, palatable food intake would increase. This hypothesis has been supported by human studies, in which stressors have been shown to acutely, as well as chronically, stimulate feeding of pleasurable calories (16,17,18,19), and stressors have been shown to stimulate the secretion of dopamine over the nucleus accumbens in proportion to cortisol responses (20,21). Together, these findings suggest strongly that with elevated glucocorticoids, pleasurable calories are more salient, or wanted, with high glucocorticoids (22).

Materials and Methods

Adult male Sprague Dawley rats (Charles River, Hollister, CA) were housed singly in hanging basket cages, with ad libitum access to chow (Purina 5008, 3.31kcal/g; Purina, St. Louis, MO) and water, in light (12 h light at 0700, 12 h dark)- and temperature (21–23 C)-controlled rooms. The rats were allowed to adapt to the new environment for 5 d, with daily body weight and chow intake measured. Experiments and procedures were approved by the University of California, San Francisco, Institutional Animal Care and Use Committee; vigorous attempts were made to diminish external sources of pain and distress as well as to limit the experiments to the shortest duration that was essential to achieve their aims.

After adaptation to facility conditions and chow diet, four experimental groups weighing 320–340 g were established (seven rats/group): chow only (chow); chow and 32% sucrose solution (provided in a bottle; 1.4 kcal/g; Safeway, Pleasanton, CA; chow/sucrose); chow and lard (provided in a metal cup; 9 kcal/g; Armour, Omaha, NE; chow/lard); and all three calorie sources (chow/sucrose/lard). Those groups that did not receive sucrose or lard had an empty bottle and/or metal cup installed in the cage. Nutrient sources and water were available ad libitum throughout. Body weight and intake of each nutrient and water were measured at 1000 h for the following 7 d.

Experiment 1

After 7 complete days of eating chow and free choice of sucrose and/or lard (starting on d 1), all rats were restrained for 30 min, at 1000 h on the eighth day. Food was removed from the cages and weighed while the rats were restrained. Blood samples were collected from an initial cut over a lateral tail vein at the time of entry into the restraint tubes (0 min sample) and again, after removal of the formed scab, at 15, 30, 60, 90, and 120 min. Approximately 300 μl (total of 1.5 ml over 2 h per 300 g body weight) blood were collected into EDTA-coated tubes and maintained at 4 C until centrifuged. Rats were restrained for 30 min, and then they were returned to their home cages; subsequent blood samples were collected by briefly wrapping the rat in a towel, collecting the sample and returning the animal to its home cage. Chow, sucrose bottles, and lard dishes were available when the rats were returned to their home cages, and the 3.5-h intake of each nutrient was measured at 240 min, when the rats were killed by decapitation. Trunk blood was collected in chilled tubes containing 0.1 ml EDTA (65 mg/ml), kept on ice, and centrifuged in the cold. The remaining food was weighed. Aliquots of plasma were stored at −80 C until assayed. Brains were collected, placed in plastic molds surrounded with Tissue-Tek, and flash frozen. The rest of the carcass was put on ice for subsequent dissection and weighing of the white adipose tissue (WAT) fat pads and adrenal and thymus glands.

Experiment 2

This was intended to repeat experiment 1, except that, after taking 0 time and 15 min tail nick blood samples in restraint, the rats were then decapitated (∼15 min), primarily for measurements of CRF mRNA and hnRNA in medial parvocellular paraventricular nucleus (mpPVN) of the hypothalamus and CRF mRNA in the central nucleus of the amygdala (CeA) and bed nucleus of the stria terminalis (BST) by semiquantitative in situ hybridization (ISH). Again, trunk blood was collected as were brains, fat depots, and adrenal and thymus glands, as outlined above.

Assays

WAT depots, adrenals, and thymuses were cleaned of connective tissue and weighed to the nearest milligram [sc WAT (IWAT), epididymal WAT, perirenal WAT, and mesenteric WAT] as reported previously (23,24). Plasma ACTH, corticosterone, insulin, and leptin concentrations were assessed by RIA (Linco Research Inc., St. Charles, MO; MP Biomedicals, Orangeburg, NY); plasma glucose, triglycerides, glycerol, and free fatty acids (FFAs) were measured colorimetrically on a plate reader using kits (Mega Diagnostics, Los Angeles, CA; Sigma-Aldrich, St. Louis, MO; Wako Chemicals, Neuss, Germany), all as previously described (24,25).

Brains

Brains were sectioned at 14 μm in the coronal plane, and sections were collected onto Superfrost Plus microscope slides (Fisher Scientific, Pittsburgh, PA) and stored at −80 C until further processing. One of the six series of sections was used for cresyl violet staining and another for CRF mRNA ISH in both experiments. In experiment 2, ISH was also performed to measure CRF hnRNA in the mpPVN on a third set of sections. A comparison of the same gene in different forms can provide insight into timing and magnitude of peptide regulation. Because the brains in experiment 2 were collected about 15 min after the onset of restraint, we assumed that the CRF mRNA would represent the ongoing state of hypothalamic CRF mRNA before the stressor of restraint, whereas the CRF hnRNA would reflect the acute response to restraint.

ISH used techniques as previously described (26,27). Sections were fixed, washed, and acetylated, rinsed with 2× standard sodium citrate (SSC), dehydrated, and air dried. Localization of CRF mRNA or CRF hnRNA used ³³P-labeled antisense cRNA probes transcribed from corresponding full-length cDNAs. CRH hnRNA cDNA was kindly provided by Dr. Robert Spencer (University of Colorado, Denver, CO). The probes were labeled in a reaction mixture, separated from unincorporated nucleotides over a Micro Bio-Spin column (Bio-Rad, Hercules, CA) and diluted in hybridization buffer. Probe (∼1 × 10⁶ cpm) was applied to slides, incubated overnight at 55 C, and washed with ribonuclease A at 37 C for 1 h. Slides were washed in descending concentrations of SSC at room temperature for 1 min, followed by a wash in 0.5× SSC for 1 h at 68 C and in 0.1× SSC (1 min at room temperature). They were dehydrated in ascending concentrations of ethanol and then air dried. Dry slides were exposed to Hyperfilm MP (Amersham Pharmacia Biotech, Buckinghamshire, UK), for 2 d (CRF mRNA in mpPVN), 6 d (CRF mRNA in BST and CeA), and 25 d (hnRNA in mpPVN). Semiquantitative densitometric analyses of the relative level CRF mRNA or hnRNA levels were performed using Spot 3.2.2 (Diagnostics Instruments, Sterling Height, MI) to measure OD. Analyses were performed blind to the identification of any group. Anterior-posterior coronal levels were the same as those described previously (28).

The brain sections containing CRF mRNA in the oval (BSTov) and fusiform (BSTfus) nuclei of the BST also contained distinct clusters of CRF-expressing cells in the medial preoptic area (mPOA), and the expression in these clusters was also semiquantified (see supplemental Fig. 1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org).

Statistical analyses

In experiment 1, blood samples were lost from one rat, and six rats were deleted because of brain damage during collection or damaged slices. In experiment 2, one rat was excluded because blood samples were lost. Thus, four to six brains/group were analyzed. Data were analyzed by one-way ANOVA. A significant (P < 0.05) effect was followed by post hoc tests of individual group differences (Fisher’s protected least significant difference). Corticosterone, ACTH, body weight, and caloric intake were analyzed by ANOVA corrected for repeated measures over time.

Results

Caloric intake and body and fat depot weights

Seven days of palatable foods available ad libitum provided the same results on caloric intake, body weight gain, fat depot weights, and adrenal and thymus weights in both experiments, and the results are shown in Fig. 1 (and in supplemental Tables 1 and 2). Caloric intake was defined by caloric content of the foods ingested. In both experiments, rats with access to sucrose and/or lard derived about 30–50% of their daily calories from sucrose, and/or lard, with the rest as chow (Fig. 1, left panel, supplemental Table 1). In all groups given palatable foods in addition to chow, total caloric intakes increased significantly [experiment 1: sucrose F(1,25) = 46.95, P < 0.0001; lard F(1,25) = 18.38, P < 0.0001; experiment 2: sucrose F(1,23) = 34.08, P < 0.0001; lard F(1,23) = 8.78, P < 0.008], although chow intake decreased significantly [experiment 1: sucrose F(1,25) = 7.09, P < 0.014; experiment 2: sucrose F(1,23) = 39.7, P < 0.001; lard F(1,23) = 11.53, P < 0.003; interaction F(1,23) = 10.16, P < 0.005]. In both experiments, body weight gain did not increase in the groups given sucrose and/or lard (Fig. 1, right panel; supplemental Table 1; P = NS).

Figure 1.

Food intake, body weight, and caloric storage in fat depots in experiment 1. Upper left panel, Mean caloric intake by food type. All groups offered palatable foods increased intake. Upper right panel, Body weight gain did not differ in the four groups over the time tested. Bottom panel, Fat depot weights. EWAT, Epidymal fat; PWAT, perirenal fat; MWAT, mesenteric fat. Bars and symbols, mean values; lines, represent ± 1 sem. Letters above the bars represent significant differences among the groups; bars with the same letters do not differ from one another.

Although body weight did not increase with sugar and fat intake over the 8-d period, WAT depot weights did increase (Fig. 1 and supplemental Table 2). Overall fat depot weights increased significantly and were affected by both sucrose and lard intake [experiment 1: sucrose F(1,25) = 9.69, P < 0.0005, Lard F(1,25) = 9.31, P < 0.006; experiment 2: sucrose F(1,23) = 4.92, P < 0.038, lard F(1,23) = 9.28, P < 0.006]. Individual depot weights (shown for experiment 1, Fig. 1) accrued fat differentially in response to the diets provided. Although there was an effect of lard on the epididymal WAT, it was slight and not significant by one-way ANOVA for the depot (P > 0.1). Perirenal WAT increased after both sucrose and lard and diet increased the depot weight significantly by one-way ANOVA (P < 0.005). There were similar effects of the diets on IWAT and mesenteric WAT weights in experiment 1 (both P < 0.001). Similar results for most of the individual fat depots were found in experiment 2, except that the increase in IWAT weight compared with chow only approached significance (not shown).

Leptin, insulin, and metabolites

Circulating leptin concentrations were increased 3-fold in all groups given the sucrose and/or lard to eat in experiment 1, but only the group eating both palatable foods differed significantly from chow control (P < 0.015). Similar results (P < 0.007) were observed for the group eating both palatable foods in experiment 2, with leptin 2.5- to 4-fold higher in the groups eating sucrose and/or lard (supplemental Table 1).

Because it was of interest to us to examine the relationship of insulin to glucose under stressful conditions, we included measurement of time 0 plasma samples from tail blood when sufficient sample was available after measurement of ACTH and corticosterone. Because neither the initial insulin nor glucose concentrations differed between the two experiments at time 0, we pooled the results into a single group and included the results from terminal samples collected at about 15 min (experiment 2) and 240 min (experiment 1) in Tables 1 and 2. In the initial sample, insulin was increased above the chow control group in the groups eating chow/lard and chow/lard/sucrose, with a marginally significant elevation in the chow/sucrose group. Plasma glucose concentrations were strongly elevated in the groups that had lard available but not in the group drinking sucrose only. At about 15 min, after the onset of restraint, mean insulin concentrations increased in all groups above 0 min, probably as a consequence of the marked, stressor-induced increase in glucose concentrations. By 240 min, both insulin and glucose concentrations were returning toward the 0 min, initial concentrations.

Table 1.

Plasma insulin concentrations at the onset (time 0) and after restraint show that choice of diet, regardless of composition, increases insulin

	Insulin (ng/ml)
	Chow control	Chow/ sucrose	Chow/ lard	Chow/ sucrose/ lard
Time after restraint (min)
0	1.2 ± 0.1a	2.3 ± 0.3b	2.5 ± 0.5b	2.5 ± 0.3b
15	2.2 ± 0.3a	3.0 ± 0.3a,ba,b	3.1 ± 0.5a,b	3.6 ± 0.6b
240	1.9 ± 0.2a	3.3 ± 0.3b	2.4 ± 0.2a	3.4 ± 0.6b

Time 0 values represent data pooled from both experiments, which did not differ; n/group insulin is 9 or greater; n/group glucose is 11 or greater; groups within rows with same superscript letter do not differ from each other. 

Table 2.

Plasma glucose concentrations at the onset (time 0) and after restraint shows that lard, but not sucrose ingestion increases glucose concentrations, suggesting that lard causes insulin resistance

	Glucose (mg/dl)
	Chow control	Chow/ sucrose	Chow/ lard	Chow/sucrose/ lard
Time after restraint (min)
0	123 ± 4a	121 ± 7a	144 ± 6b	143 ± 8b
15	188 ± 8	206 ± 8	190 ± 7	178 ± 17
240	146 ± 7	151 ± 8	150 ± 5	152 ± 7

Time 0 values represent data pooled from both experiments, which did not differ; n/group insulin is 9 or greater; n/group glucose is 11 or greater; groups within rows with same superscript letter do not differ from each other. 

Lipid-associated metabolites, glycerol, FFAs, and triglycerides differed between the two experiments at time 0; therefore, we did not pool the data, and each experiment is accompanied by its time 0 value (supplemental Table 3). There were no marked effects of diet or time in restraint on these values. However, it should be noted that the time 0 (tail nick) values for FFAs are much higher than the values collected from trunk blood (∼15 and 240 min); this also is true to a lesser extent for glycerol concentrations. This may be a consequence of either the different collection sites (venous blood from the tail, mixed central arterial, and venous blood from the trunk) or the effect of initial handling and tail-nick on autonomic outflow.

ACTH, corticosterone, and brain CRF after restraint

In experiment 1, the ACTH response to restraint was damped by the preceding provision of sucrose, lard, and the combination at 15 min during the stressor [Fig. 2A, left panel; one-way ANOVA at 15 min, F(3,25) = 32.87, P < 0.0001]. At 30 min of restraint, there were again significant main effects by two-way ANOVA, but the interaction term was not significant; one-way ANOVA was also significant [F(3,25) = 4.54, P < 0.012]. However, at 30 min, examination of the results shows that it was the group that allowed both sucrose and lard that had higher ACTH concentrations than the chow/sucrose, chow/lard, and chow control groups. Calculation of the area under the curve (AUC) for the entire 120 min (Fig. 2A, right panel) showed that ACTH in the chow/sucrose and chow/lard groups was significantly suppressed compared with chow only, whereas the response of the chow/sucrose/lard group was intermediate between chow control and the other two groups [one-way ANOVA (F(3,25) = 5.83, P < 0.004].

Figure 2.

Time course of ACTH and corticosterone in response to restraint in experiments 1 and 2. A, ACTH time course and AUC, experiment 1. B, ACTH time course, experiment 2. C, Corticosterone time course and AUC, experiment 1. D, Corticosterone time course, experiment 2. Bars and symbols, mean values; lines, represent ± 1 sem. Letters above the bars represent significant differences among the groups; bars with the same letters do not differ from one another.

The results for corticosterone responses to restraint in experiment 1 were similar to those for ACTH (Fig. 2C). By one-way ANOVA there were significant effects at time 0 [F(3,24) = 3.81, P < 0.025] and at 90 min [F(3,24) = 3.17, P < 0.046]. The AUCs for the corticosterone responses, with significant differences among groups, are shown in Fig. 2C.

In experiment 2, we only had two samples, taken at 0 and 15 min, for ACTH and corticosterone; these did not distinguish among the dietary groups as they did in experiment 1 (Fig. 2, B and D). Examination of the 0 and 15 min ACTH and corticosterone results suggests strongly that there was an unusually rapid (and thus missed) ACTH response to restraint that was strongly inhibited by the marked increase in corticosterone that occurred by 15 min.

In the mpPVN, CRF hnRNA at about 15 min and CRF mRNA at about 15 and 240 min after restraint are shown in Fig. 3. At about 15 min, there were no significant differences in either CRF hnRNA (Fig. 3, top panel) or mRNA (Fig. 3, middle panel). At 240 min after restraint, mRNA expression in mpPVN was significantly lower than chow controls for each of the sucrose and/or lard eating groups.

Figure 3.

CRF hnRNA and mRNA in the mpPVN in experiments 1 and 2. Top panel, CRF hnRNA at 15 min, experiment 2; middle panel, CRF mRNA at 15 min, experiment 2; bottom panel, CRF mRNA at 240 min, experiment 1. Bars, mean values, lines, ±1 sem. Letters above the bars represent significant differences among the groups; bars with the same letters do not differ from one another.

In the CeA, CRF mRNA was significantly elevated at about 15 min after the onset of restraint in the chow/sucrose group but not in the chow/lard and chow/sucrose/lard groups, above the chow control group [one-way ANOVA, F(3,18) = 4.84, <0.015] (Fig. 4A). At 240 min after restraint, there was no longer a significant difference among groups, although the mean value in the chow/sucrose group was still greater than in the other groups (Fig. 4B).

Figure 4.

CRF mRNA in limbic brain and mPOA in experiments 1 and 2. Results from experiment 2 and 1 are plotted on the left and right panels, respectively. A and B, CRF in the CeA. C and D, CRF in the BSTov. E and F, CRF in the BSTfus. G and H, CRF in scattered cells of the mPOA. Bars, mean values, lines, ±1 sem. Letters above the bars represent significant differences among the groups; bars with the same letters do not differ from one another.

CRF mRNA in the BSTov, BSTfus, and cell clusters of the mPOA were quantified at the same A-P levels. At about 15 min (Fig. 4C), the chow/lard group had significantly (P < 0.05; one-way ANOVA) reduced CRF in the BSTov, compared with chow control. At 240 min after the onset of restraint, there were distinct significant differences among the groups with the chow control group having the highest expression levels of CRF mRNA (chow/sucrose, P < 0.1; both groups eating lard P < 0.05) (Fig. 4D). In the BSTfus, CRF mRNA was not different among the groups at either time (Fig. 4, E and F).

There were no significant differences induced by diet in the mPOA at about 15 min (Fig. 4G); however, at 240 min, there was decreased CRF mRNA in the chow/sucrose group, compared with the chow control and chow/lard groups (P < 0.05) (Fig. 4H). The group with chow/sucrose/lard also had decreased CRF mRNA, but this was not different from any other group.

Chow, sucrose, and lard ingestion during the 3.5 h after restraint

In experiment 1, all rats were allowed 2.0 h of feeding after the 30 min of restraint. Only sucrose was ingested to an appreciable extent by the rats (Fig. 5, top panel); there was very little chow or lard intake. The total caloric intake/group over the 3.5-h period is shown in Fig. 5 (bottom panel). In the two groups that had sucrose available to them and drank it, mean plasma concentrations of insulin were higher at the end of the period (Tables 1 and 2, 240 min).

Figure 5.

Calorie intake during the 3.5 h after restraint in experiment 1. Upper panel, Calories of specific foodstuffs ingested in each group. Lower panel, Total calories ingested during the 3.5 h period in the four groups. Bars and symbols, mean values, lines, ±1 sem. Letters above the bars represent differences among the groups; bars with the same letters do not differ from one another.

Discussion

These studies showed that voluntary ingestion of pleasurable foods, either sucrose and/or lard, increases adiposity, with little short-term weight gain. All components of the HPA axis response to an acute stimulus were damped in the experiment in which chow control rats had hormonal restraint responses of normal magnitude and timing. Both autonomic and neuroendocrine components of the central CRF network are modified by the palatable food intake in a direction that suggests that activity of the central stress network is reduced in parallel with increased metabolic well-being. Sucrose, but little lard, ingestion occurred after the stressor, reinforcing the notion that coupling ingestion of pleasurable foods after stressors could become habitual.

Voluntary intake of either, or both, sucrose and lard increases fat masses over the short term without increases in body weight. This was associated with reduced subsequent HPA responses to an acute stressor in experiment 1 but not experiment 2. The results in experiment 1 showed the usual temporal ACTH and corticosterone patterns of response to restraint (10,11,15,29,30,31,32,33). ACTH peaked at 15–30 min, and peak corticosterone responses occurred between 25 and 35 μg/dl at 30–60 min. Despite normal initial values, the very rapid and unusually large (>50 μg/dl) corticosterone response to restraint at 15 min observed in experiment 2 suggests that the rats were primed at the onset of that experiment for unusually fast responses of high amplitude, perhaps as a consequence of having previously endured an unmonitored environmental stressor (34,35,36).

Experiment 1 showed that ingestion of both sucrose and lard are major determinants of the magnitude of the HPA response to restraint. Prior access to either or both determined the hypothalamic CRF mRNA concentration that, in turn, reflected (12) the ACTH and adrenal corticosteroid responses. However, the responses of limbic and anterior hypothalamic CRF-containing cell groups differed, depending on what food was ingested, suggesting that ingestion of sucrose and lard affect responsivity of the motor components of the HPA and autonomic axes through different neural inputs.

Heteronuclear and messenger RNA for CRF in the mpPVN were not different at 15 min among the four dietary groups. However, mRNA for CRF in the mpPVN at 15 min presented the same pattern as that seen 240 min after the onset of restraint, with reduced CRF expression in all groups given sucrose and/or lard compared with those that had been eating chow only. Because the large intracellular pool of CRF mRNA changes slowly after stressors (12,37), we designed experiment 2 with the assumption that the mRNA values would reflect conditions existing at the onset of restraint, whereas the small intracellular pool of hnRNA and its fast response would reflect the response of the PVN to the stimulus of restraint. The pattern of 15 min mRNA for CRF suggests that having both lard and sucrose to ingest, in addition to chow, reduced CRF expression in those groups, as we reported that they did when measured in unstressed rats after a week of eating palatable foods (11). By contrast, mean hnRNA expression was increased in the sucrose and/or lard groups, possibly reflecting the extraordinarily high corticosterone concentrations and missed ACTH response in this experiment. As expected, with such a large corticosteroid feedback rate signal (38,39,40,41), ACTH was very low and not different from time 0. Because the initial mRNA for CRF was relatively low in the rats eating sucrose and/or lard, it is logical that the initially damped CRF neurons responded with hnRNA for CRF to a greater mean extent than those in the chow control group.

CRF mRNA in the central nuclei of the amygdala was significantly elevated in the sucrose group at about 15 min and remained elevated at 240 min. Because the dynamic of changes in CRF mRNA after a stressor has not been delineated in amygdala as it has in the PVN (37), we cannot assume that it will not increase within 15 min. It is clear that amygdala CRF secretion is acutely responsive to stressors, glucocorticoids, and the fed or fasted state (42,43,44,45). We previously found that amygdala CRF mRNA is increased in adrenalectomized rats drinking sucrose (7). Moreover, studies of c-Fos protein responses to drinking sucrose have shown increased responses in both the CeA and basolateral amygdala (46,47,48), suggesting that drinking sucrose stimulates amygdalar activity. By contrast, eating lard, stimulated by infusion of insulin into the superior mesenteric vein of diabetic rats, inhibits expression of c-Fos protein in the CeA (13), suggesting that the central neural network that results in decreased CRF mRNA expression in the PVN in rats drinking sucrose differs from that in rats eating lard. This suggestion is supported by the finding that opioids in the PVN specifically stimulate fat, but not sucrose, intake, whereas systemic administration of opioid antagonists inhibit both fat and sucrose ingestion (49), showing that opioids can alter intake of both but not at the same CNS site. Moreover, mean CRF mRNA expression in the CeA of the group given both sucrose and lard was the same as that in the chow control group at both times, suggesting that when both foodstuffs are available, the inhibitory lard signal overrides the apparently stimulatory sucrose signal to CRF mRNA in the amygdala.

The BSTs are considered extensions of the medial core of the amygdala (50,51), and there is a known interaction between CRF-containing neurons in the CeA and CRF-containing neurons in the BST (52). Based on its connections, the anteromedial BST has been called a site for cerebral integration of neuroendocrine, autonomic and behavioral aspects of energy balance (see Refs. 53,54). The outputs of BSTov and BSTfus differ; the former heavily innervates the latter as well as central amygdala and premotor autonomic regions of brain, whereas the latter projects heavily to the neuroendocrine portion of the PVN as well as other medial hypothalamic sites (51). Changes in CRF expression in the BST as a function of diet/restraint occurred only in the oval, but not the fusiform, nuclei. At 240 min, the pattern of CRF mRNA observed in the BSTov bears a striking resemblance to that in the mpPVN (compare Figs. 2 and 3D). The fact that the primarily autonomic-regulating BSTov alters CRF expression in response to sucrose and/or lard in a pattern similar to that of the neuroendocrine mpPVN suggests that, as proposed (56), signals of metabolic well-being may generally reduce activity in central stress-response networks. Although we saw no interactive effects of feeding on restraint in CRF expression of the BSTfus, discrete lesions of the fusiform/dorsomedial cell groups of the BST have been shown to result in decreased HPA responses of rats to a single bout of restraint (57,58) or IL-1β (59).

The scattered CRF cells in the mPOA exhibited a distinct reduction in CRF mRNA at 240 min after restraint in the sucrose-drinking groups. These scattered cells may be an anterior extension of the CRF-expressing cells in the lateral hypothalamus studied [Ref. 60 and personal communication from A.G. Watts, University of Southern California, Los Angeles, CA] that respond to hypertonic saline and reduced feeding.

Circulating leptin concentrations were increased in all rats eating palatable foods, reflecting their increased fat stores. Exogenous leptin inhibits function in the HPA axis (61,62) and thus might be a signal that damps HPA responses to restraint in the rats eating palatable foods. We believe this is not the case; however, because comparison of leptin concentrations in rats, given the choice of eating lard and forced lard ingestion, showed equal elevation in both groups compared with control, whereas only the rats with choice of eating lard had inhibited ACTH responses to restraint (10). In that study and in this one, circulating insulin was the only hormone measured that had a pattern similar to the ACTH-response pattern (10). None of the circulating metabolites that were measured had response patterns that fit the central and peripheral inhibition of the HPA axis observed in this study, although clearly glucose, FFAs, and triglycerides can act at the hypothalamus to alter neuroendocrine outflow (63,64,65,66).

Eating behavior after the 30-min period of restraint and blood sampling revealed that only rats with highly palatable sucrose available indulged appreciably during the poststress interval. This result resembles that seen in humans after stress, particularly in humans with abdominal obesity (19,67). The rats given sucrose and/or lard were measurably fatter than the chow control rats at the time they were restrained, and it was particularly the intraabdominal fat depots that were consistently increased in weight. In rats, afferent neural inputs from abdominal fat depots are distinguished from inputs from sc depots at the level of the PVN (68), and it may be that the abdominal depot signals serve to both inhibit responses to stressors and block CRF-induced inhibition of ingestion (28,69). Moreover, CRF input to the shell of the nucleus accumbens has been shown to increase the drive to obtain sucrose (70,71); however, a CRF-mediated drive to obtain lard has not been tested.

There are long-term implications of the findings that eating preferred high-caloric foods alter central CRF expression and reduce HPA responses to acute stressors. Our studies dealt only with prior exposure to palatable foods and the acute stress response, not with the effects of these foods on responses to sustained, uncontrollable stressors. However, there is reason to believe that if a behavior is associated with a good outcome (ingestion of pleasurable foods and diminished feelings of stress), this may easily become habitual (72). Thus, learning and memory of the hedonic attributes of the intake of comfort foods may result in habit formation. A recent 20-yr prospective report from the Whitehall II studies and the effects of chronic stressors showed that those civil servants who were initially in the lowest quartile for job control and highest quartile for job strain and who also had poor social support had increased body mass index and abdominal obesity (73). Intraabdominal fat is associated with considerably increased risk of cardiovascular disease, metabolic syndrome, and type 2 diabetes and reduced life span. Understanding the neural changes involved with both acute and chronic stressors and the interactions of stressors with feeding is key to understanding at least a portion of the current epidemic of obesity (74,55).