Physiology&Behavior,Vol. 52, pp. 583-590, 1992

0031-9384/92 $5.00 + .00 Copyright © 1992 Pergamon Press Ltd.

Printed in the USA.

Effect of Chronic Stress and Exogenous Glucocorticoids on Regional Fat Distribution and Metabolism M. R E B U F F I ~ - S C R I V E , *l U. A. WALSH,* B. M c E W E N ~

AND

J. R O D I N *

*Department of Psychology, Yale University, New Haven, C T and 7Department of Neuroendocrinology, Rockefeller University, New York, N Y R e c e i v e d 28 O c t o b e r 1991 REBUFFt~-SCRIVE, M., U. A. WALSH, B. McEWEN AND J. RODIN. Effect of chronic stress and exogenous glucoeorticoids on regionalfat distribution and metabolism. PHYSIOL BEHAV 52(3) 583-590, 1992.--It has been proposed that increased glucocorticoid hormones and decreased sex hormones affect regional fat metabolism and distribution. In the present work, it was hypothesized that chronic, uncontrollable stress, known to affect the pituitary-adrenal and pituitary-gonadal axes might, therefore, lead to differences in regional fat accumulation. In comparison with controls, male Sprague-Dawley rats stressed for 28 days, had significantly larger adipocytes. In addition, a tendency for a heavier fat pad and an increased lipoprotein lipase activity in the mesenteric depot was suggested. No significan t changes were seen in epididymal, retroperitoneal, and inguinal regions. In order to study if the effects observed could be attributed to increased glucocorticoids, the response to a direct administration of supraphysiological doses of corticosterone, given either in the drinking water or via subcutaneous implantation of corticosterone pellets, was studied. Increased fat accumulation was shown in all fat depots in a dose-response fashion, but was significantly more pronounced in the mesenteric region. It was concluded that mesenteric fat tissue may respond to stress in a different manner from other fat depots. Glucocorticoids seem to be partly, but not solely, responsible for the changes observed in adipose tissue metabolism and distribution following exposure to uncontrollable stress. Stress

Glucocorticoid

Internal fat

LPL activity

R E C E N T epidemiological studies have shown a statistical association between fat distribution in the abdominal region, particulafly the intraabdominal, and increased risk factors for cardioand cerebrovascular diseases as well as for diabetes mellitus [for examples (10,20,26,27)]. The causal mechanisms for this relationship are still unclear. Whether increased fat accumulation in the abdominal regions is directly causal to the diseases or a parallel p h e n o m e n o n caused by an unknown factor, possibly endocrine (1), is not known. Endocrine aberrations, particularly in steroid hormone levels, have been found in both men and w o m e n with a more abdominal distribution of fat, showing elevated glucocorticoid secretion and low sex steroid hormone secretion (2). The role of steroid hormones on the regulation of adipose tissue metabolism and distribution is partly known (28). Glucocorticoids seem to favor fat accumulation specifically in the abdominal regions. This is shown, for example, in Cushing's patients in w h o m ~/bdominal lipoprotein lipase (LPL) activity, the key enzyme for fat accumulation, is considerably higher in comparison to control or android obese women (31). It has also

been shown that these patients have five times more intraabdominal fat than controls (22). Moreover, it has been suggested that the regional differences observed in the effects of glucocorticoids on adipose tissue metabolism might be explained by differences in their receptor protein (30) or m R N A density, both of which are particularly high in the intraabdominal fat depots (34). In men, testosterone seems to inhibit fat accumulation in the abdominal regions (33), while in women, estrogen and progesterone seems to favor fat deposition in the gluteo-femoral regions (29). In rats, the effects ofglucocorticoids on facilitating lipid storage have been shown, although there are some contradictory results (6,7,18). Glucocorticoids also seem to have a permissive effect on fat mobilization in animals (8,12). Regional differences in glucocorticoid receptor density have been demonstrated, with a higher density in the internal fat regions (9). In animals, increased glucocorticoids and decreased sex hormones are well known neuroendoerine responses to certain stressors, particularly uncontrollable stress which induces submission and defeat. Here, the pituitary adrenocortical axis is

Dr. Rebuff6-Scrive is a visiting research scientist in the Department of Psychology at Yale University. Requests for reprints should be addressed to M. Rebuff6-Scrive, Department of Psychology, Yale University, P.O. Box 11A Yale Station, New Haven, CT 06520.

583

584

REBUFFI~-SCRIVE ET AL.

stimulated, leading to increased ACTH and cortisol secretion, while the LH/FSH axis is inhibited, leading to low sex hormone production (13). Such experiments have been repeated in primates (36). Finally, immobilization stress appears to elevate serum triglyceride levels (14) and increase the incidence of type I diabetes in Bio Breeding rats (21). Therefore, the association between stress and neuroendocrine disturbances leading to abnormal steroid hormone production might be a potential generator for the increased risk factors associated with abdominal obesity. This might occur through changes in fat metabolism and distribution. This is also suggested by the results from a recent prospective study, where, in women, feeling of stress, as well as other psychological maladjustments, were found to be associated with increased distribution of abdominal fat (19). In the present work, using an animal model, we studied first whether environmental factors such as a chronic, uncontrollable stress, known to stimulate the pituitary-adrenal axis, might affect fat accumulation particularly in the intraabdominal regions (Experiment I). In a second set of experiments, we studied whether the effect observed following stress could be attributed only to glucocorticoids or if other mechanisms were involved. Therefore, the direct action of exogenous glucocorticoids, given either in drinking water or via implantation of corticosterone (CORT) pellets, on fat accumulation and distribution were measured (Experiments IIA and IIB). EXPERIMENT I

Animals and Experimental Design Twenty male Sprague-Dawley rats (Charles River Laboratories, Cambridge, MA), weighing 190-210 g, were housed in individual wire-mesh cages in a room maintained at 20 _+ 2°C, with a 12-h light cycle (0700-1900 h). Rats were fed a pellet diet of known and stable composition (11.5% fat, 65% carbohydrate, 20% protein calories, plus vitamins, minerals and fiber, Research Diets, NJ). After a 1-week acclimatization period, rats with unusually low or high food or fluid intakes in comparison to the mean consumed b y all the animals, were eliminated from the study. The remaining rats (N = 16) were randomly divided into stressed (S) and control (nonstressed) groups (N = 8 i n each group). Stressed animals were subjected to an uncontrollable stress for 28 days. The stressors consisted of rotation (Plexiglas cages rotating on a platform with a motor allowing speeds from 0-100 rpm/ min), and restraint (individual L Plexiglas containers with adjustable side and front panels). The regimen of the stressors (randomly generated by computer) was alternated and increased in frequency, duration, and severity over time in an attempt to decrease the degree of habituation (21). One day before the beginning of the experiment and 28 days later, blood samples were taken for CORT determination. At the same time of day, blood was obtained from the tail of both stressed rats after having been placed in a restrainer for 1 h and controls who had remained in their cages for the same period of time [see (21) for discussion of procedure and the minimal degree of stress of the tail bleed procedure per se]. Control and stressed animals were pair fed, to control for the effect of stress on food intake and, therefore, body weight (21). Each day the controls were given the mean amount of food eaten by the stressed rats on the previous day, _+0.3 g. Rats were weighed twice weekly. After the 28th day of experimental stress, the same number of stressed and control rats were sacrificed by decapitation in the morning. Because all the rats could not be killed the same

day for technical reasons, the rats were started in the experimental stress protocol on different days, in order to maintain a constant period of 28 days until sacrifice for each rat. Blood was collected and serum was frozen for determination of insulin, glucose, total testosterone (total hormone secreted), free testosterone (physiologically active fraction of total testosterone), and triglycerides.

Method Adrenals, thymus, spleen were dissected and weighed. Four fat depots, epididymal (around the testes, epididymis, and ductus deferens), retroperitoneal (along the posterior wall of the animal, from the kidneys to the hip region), mesenteric (along the mesentery, starting from the lesser curvature of the stomach and ending at the sigmoid colon), and inguinal (subcutaneous fat between the lower part of the rib cage and the thighs) were carefully removed, weighed, and sampled for fat cell size and LPL activity. Fat cell size was measured using electronic counting of osmium-fixed cells as described by Hirsch and Gallian (15) and was expressed as fat cell weight. The triglyceride content of the cells was extracted according to Dole (5) and was determined according to Carlson (3). LPL activity was determined after elution with heparin, using a stable radioactive triolein emulsion according to Nilsson-Ehle and Schotz (25). Released free fatty acids were separated by liquid-liquid partition as described by Gammeltoft and Glieman (11). LPL activity was expressed per cell and per cell surface area (SA = 7rd2 + SD2), where d = mean fat cell diameter and SD = standard deviation). Serum triglycerides (TG) were measured according to Soloni (37). Radioimmunoassays were used to determine testosterone (Diagnostic Products Corp.) and insulin (24), Intra- and interassay coefficient of variation for testosterone and insulin were respectively (2.9% and 5.2, and 6% and 10%). Serum CORT was measured by radioimmunoassay using rabbit antiserum raised against corticosterone-21-hemisuccinate BSA (B21-42, Endocrine Sciences, Tarzana, CA). Assay sensitivity was 10 log CORT, and coefficients of variation within and between assays were 4% and 8%, respectively (38).

Statistical Analysis Students' t-tests were used for comparisons between stressed and control rats. Linear correlation coefficients were measured. All p values are two tailed.

Results After 28 days of chronic stress, body weight was not different between stressed and control animals (Table 1). A slight increase in the adrenal weight and decrease in thymus and spleen weights

TABLE 1 VALUESAT SACRIFICE

Body weight (g) Glucose (mg/dl) Insulin (gU/ml) Triglycerides(mg/dl) Total testosterone (ng/ml) Free testosterone (%)

Controls

Stressed Rats

451.1 + 12.9 104 _+6 27 _+3 197 + 29 0.96 -+ 0.32 78 + 1

451.7 _+ 10.i 108 + 5 27 _+5 252 + 58 0.74 _+0.24 74 _+ 1"

Mean -+ SE. n = 8 . * p < 0.05 comparisons with controls.

STRESS, GLUCOCORTICOIDS, AND REGIONAL FAT DISTRIBUTION were observed, but these changes did not reach statistical significance (data not shown)• Fat cell weight was significantly increased only in the mesenteric region, in the stressed animals in comparison with controls (p < 0.01) (Fig. 1B). Furthermore, it was interesting to note that, in comparison with controls, mesenteric fat pad weight tended to increase (p < 0.08) in the stressed rats (Fig. 1A). Similarly, LPL activity expressed per cell (Fig. 1C) was doubled in the mesenteric fat depot of the stressed rats in comparison with controls, although the t-value was only marginally significant (p _< 0.08). No trends were observed in the other depots. Similar results were obtained when LPL was expressed per cell surface area. As shown in Table 1, stressed and control rats did not differ in levels of insulin, glucose, or triglycerides. However, free testosterone (%) was significantly lower in the stressed animals (p < 0.05). After 1 hour in the restrainer, the stressed rats had high CORT values (48.4 _+ 3.8 /ag/dl) in comparison with normal values in the morning (2-5 /~g/dl), and these values remained elevated after 28 days of stress (42.1 + 5.8 ~g/dl). When CORT values after 28 days of stress were correlated with the weight of the various fat pads, a positive correlation was found only with the mesenteric fat depot (p < 0.05) (Fig. 2). No correlations were found between plasma CORT and fat pad weights in the control animals (not shown),

585

lo~ ~. = O ua ~ ~

8-

~

2

A

6' 4-

Controls

[]

0 MES

EPI

REGIONS 0.8

= N ~ ~ ua oI--

,-,

B

0.6 Controls

~ 0.4 "~. -~)

0.2

LI.

Discussion

This study has shown that mesenteric adipose tissue of male rats seems to respond to a chronic, uncontrollable stress in a different manner than other fat depots. The type of stress used in this experiment increased plasma CORT concentrations as shown by the values obtained after 1 hour in the restrainer. No significant changes in adrenals, thymus, and spleen weights were observed after 28 days of stress; however previous work in this laboratory has shown that when rats were submitted to the same stressors for 3 months, adrenal weights were significantly increased and thymus and spleen weights decreased (21). Furthermore, the stressed rats had high levels of CORT after 1 hour in the restrainer and these values remained elevated after 28 days of chronic stress, suggesting that the rats did not habituate to the stress. In good agreement with earlier studies performed with mice (13), the percentage of free testosterone, but not total testosterone, was decreased in the stressed animals. This suggests that no differences existed in total testosterone secretion between stressed and control rats, but that changes in the binding of the hormone to albumin might have occurred after 28 days of stress. In human studies (17,35), it has been shown that percentagefree testosterone was more strongly correlated than total testosterone to fat distribution. The adipose tissue changes observed suggest that even though the effects induced by stress were small, they were specific to the mesenteric fat region. Mesenteric fat cell weight was significantly enlarged in the stressed rats. This can be explained mainly by an increased fat accumulation or a decreased fat mobilization. The latter was not measured in this study where the stress used stimulated primarily the pituitary-cortical axis. However, fat accumulation (measured' as LPL activity) was doubled in the stressed rats in comparison with controls. LPL activity was available from only six animals, and this might explain the lack of full significance for these data. In humans, previous work has shown that glucocorticoids stimulate LPL activity both in vitro (4) and in vivo (31), with a regional specificity towards the abdominal fat regions• It was suggested that these regional differences in glucocorticoid effects might be explained by differences

Stressed

o.(} EPI

REr

k ING

[]

Stressed

•

Controls

[]

Stressed

MES

REGIONS

3

~

C 2

'

1

0 EPI

FIEF

ING

MES

REGIONS FIG. 1. Fat pad weight (A), fat cell weight (B), LPL activity (C) in epididymal (EPI), retroperitoneal (RET), inguinal (ING), and mesenteric (MES) fat tissues from control (black columns) and stressed (hatched columns) rats. Mean + SE. (*) 0.05 < p < 0.1, **p < 0.01. Comparisons between groups.

in their receptor density (30,34). In rats, differences in glucocorticoid binding have been found, showing more binding in internal fat in comparison with subcutaneous depots (9). These results, as well as the finding of a positive correlation between CORT levels after 28 days of stress and the weight of the mesenteric fat pad, suggest that the response of mesenteric adipose tissue to stress might be at least partly mediated by glucocorticolds. Therefore, in the next two experiments, the direct effects of exogenously-manipulated glucocorticoids on fat accumulation and distribution were studied. CORT was administrated either orally in drinking water for 28 days (IIA), or hormone pellets were implanted subcutaneously for 2 weeks (IIB).

586

REBUFFI~-SCRIVE ET AL. y

.045x

.596,

R-squared:

.637

2.4

2.2,

0

0

r = .80 p < .05

2,

1.8, 1.6, 1.4, .

0

m

1.2,

e-

2~

0

15

25 '

2'0

30 '

35 ' 4'0 4; s'0 Blood corticosterone (ug %)

55 '

60 '

65

FIG. 2. Correlation between CORT and mesenteric fat pad weight in stressed rats.

EXPERIMENT II

Animals and Experimental Design Sixty (IIA) and 30 (IIB) male Sprague-Dawley rats (Charles River Laboratories, Cambridge, MA), weighing 190-210 g were utilized. They were housed and fed in the same condition as the animals in Experiment I.

Statistical Analyses The data were analyzed with one-way analysis of variance followed by post hoc contrasts using Tukey's test for comparisons between control and experimental animals receiving various doses of CORT. Data were also tested for linear trend to assess dose-response relationships.

Results EXPERIMENT

IIA

Method After 2 weeks of acclimatization and exclusion of unusual eaters and drinkers (see Experiment I), 50 rats were randomly assigned to one of the five following groups and were given, for 28 days, various solutions in their drinking water, as follows: group Cont(W) was given distilled water to drink, group Cont(A1) received 4% alcohol (pure ethyl alcohol, Warner-Graham company, MD). Cont(A1) was included as a control for the fact that the corticosterone given to the three experimental groups had to be dissolved in alcohol, with concentrations that ranged to a maximum of 4% at the highest dose. The remaining three groups were given CORT (4-pregnen- 11B, 21-Diol-3, 20-Dione, Steraloids Inc., NH) at different concentrations; 100 ~g/ml (group Cort 100), 200 #g/ml (group Cort 200), 400 #g/ml (group Cort 400). CORT was dissolved in alcohol and then diluted in distilled water (15), providing 1-4% alcohol in the CORT solutions. Fresh solutions were prepared weekly. All rats were pair fed in order to control for the effects of glucocorticoids on food intake. Cont(W), Cont(A1), Cort 100 and Cort 200 rats were fed the average amount of food eaten by group Cort 400 on the previous day, +0.3 g. Rats were weighed twice weekly. As described for Experiment I, 10 rats were sacrificed every day (two from each experimental group), on 5 successive days, after having completed the 28 days experimental protocol (ingestion of either of the solutions as described above). All other methods were similar to those in Experiment I.

To determine whether animals drinking the 4% alcohol solution Cont(Al) differed from control animals Cont(W) drinking plain water, student's t-tests were used to compare groups on each variable. No differences between these two control groups (drinking water vs. drinking 4% alcohol) were found on any variable measured. Therefore, the data were pooled and the mean values were compared to the groups receiving different doses of CORT. All animals had similar body weights at the beginning of the CORT administration. After 28 days, however, there was a significant group difference, F(3, 45) = 5.18, p < 0.004. Tukey contrasts indicated that the Cort 200 and Cort 400 rats weighed significantly less than controls (p < 0.05, see Table 2). Even after adjusting for these differences in body weight, there were significant group differences for adrenal, F(3, 45) = 20.45, p < 0.0001, spleen, F(3, 45) = 6.59, p < 0.001, and thymus, F(3, 43) - 4.90, p < 0.005, weights. In all three cases, as seen on Table 2, the magnitude of the decrease in organ weight was dose related, with a significant linear trend from the controls through Cort 400 (p < 0.0001). Table 2 also presents group means for levels of plasma insulin, glucose. There was a significant group effect for insulin, F(3, 45) = 7.19, p < 0.0005, and glucose, F(3, 46) = 5.87, p < 0.002. Both showed significant dose-related increases (linear trend p < 0.0005). There was no significant group effects and no linear trend for triglycerides (not shown). The results for the measures of fat depot weight, cell size and LPL are displayed in Fig. 3. For fat pad weight, there were significant group differences in mesenteric, F(3, 46) = 3.29, p