Estradiol benzoate potentiates the effects of body-restraint in suppressing food intake and reducing body weight

Physiology&Behavior,Vol. 45, pp. 1-5. Copyright©Pergamon Press plc, 1989. Printed in the U.S.A.

0031-9384/89 $3.00 + .00

Estradiol Benzoate Potentiates the Effects of Body-Restraint in Suppressing Food Intake and Reducing Body Weight ROBIN STEVENS

Department of Psychology, Nottingham University Nottingham, NG7 2RD, United Kingdom R e c e i v e d 26 J a n u a r y 1988 STEVENS, R. Estradiol benzoate potentiates the effects of body-restraint in suppressingfood intake and reducing body weight. PHYSIOL BEHAV 45(1) 1-5, 1989.--Food intakes and body weights were recorded daily over a 15 day period for groups of ovariectomized rats. During days 6--10 rats were injected with estradiol benzoate at one of four concentrations, 0, 0.625/zg, 1.25/~g and 2.5/~g, and half the animals in each group were stressed by being restrained in small Plexiglas chambers for 30 rain per day. During the treatment period there was a dose response inhibitory effect of estradiol benzoate on intake and body weight that was greater in confined than in unrestrained animals especially at the intermediate dose of estradiol. In the posttreatment period the effect of restraint on intake persisted whereas the hormone effect did not, but the effects of estradioi were apparent on body weight in both confined and unconfined rats. This synergistic relationship between small doses of estrogen and stress in producing anorexic effects provides an animal model of anorexia nervosa; a possible neuroendocrine mechanism that could explain the effect is discussed. Anorexia

Anorexia nervosa

Body weight

Estradiol benzoate

A N animal model of the clinical disorder anorexia nervosa has been put forward by Donohoe (5) and Haslam, Stevens and Donohoe (7). This model presumes that there is an interaction between estradiol levels and stress---the features of the model will be elaborated later. Although it is widely believed that stress procedures cause increases in both food intake and body weight in rats (1) there is also evidence that some forms of stress have the opposite effect (12). Thus overeating and even obesity occurs when rats suffer from mild tail-pinching but bodily restraint causes weight loss and anorexia (6, 7, 9, 10, 12). Why these contrasting effects of different stress procedures occur is yet to be determined but possibly different stressors act on different neural mechanisms. Females in a variety of species show the same contrasting effects of overeating and anorexia depending on their exposure to estrogens. When female rats are deprived of endogenous estrogens by ovariectomy they eat more and become overweight, but during estrus, when they have high titers of estrogen in their blood, or when they are exposed to exogenous estrogen, they eat less and can loose body weight (15,16). We have proposed that the relatively mild anorexia that is found when ovariectomized female rats are exposed to small doses o f estrogen will be greatly potentiated by also confining the animals for brief periods of time. However, despite combining what we considered to be a minimal dose o f estradiol benzoate (2.5 /xg) with restraint in several experiments we have only produced an additive effect between the anorexia and weight loss produced by estradiol treatment

Food intake

Stress

and repeated brief periods of restraint [(7,8); Stevens, 1986, unpublished study]. Possibly the restraint period used, 30 minutes, is too short, so a longer time of restraint combined with a low estrogen dose would have a synergistic effect. But longer periods of restraint, 2 hours, have been used (6, 9, 10, 12) and the anorexia produced is comparable to that we obtain with 30 minute confinements. An alternative parametric manipulation would be to expose different groups of ovariectomized rats to the same fixed short period of restraint combined with different doses of estradiol benzoate. It was decided to use 3 doses o f estradiol benzoate, the largest being the 2.5/~g dose we had previously used. The prediction was that there would be a potentiation between the effects of confinement stress and one or more of the doses of estrogen and not just a simple additivity between the effects of confinement alone and the estrogen alone. METHOD

Subjects F o r t y ovariectomized Wistar rats were used; these were obtained from Charles River and were 120 days old when tested. They had been ovariectomized at 80 days of age, and were somewhat heavier (about 10%) than intact rats of this strain and species at this age. The rats were assigned to the 4 dose conditions (0, 0.625, 1.25 and 2.5 /zg estradiol benzoate); half the animals from each dose were allocated to the confinement condition. The rats were housed singly in standard plastic RB3 cages from North Kent Plastics; they

2

SFEVENS TABLE 1 FOOD INTAKESAND BODY WEIGHTS(MEAN ± S.E.M.) DURINGTHE THREE PHASESOF THE EXPERIMENT Food Intake (g) Baseline Confined Groups Oil 0.625/zg EB 1.25/xg EB 2.5/zg EB Unconfined Groups Oil 0.625/.tg EB 326.1 ± 3.45 1.25/xg EB 2.5 p.g EB

23.2 21.8 22.2 22.6

± 0.53 --- 0.78 ± 0.46 ± 0.29

Treatment

21.0 18.8 ± 17.4 ± 17.3 ±

Body Weight (g) Posttreatment

Baseline

Treatment

Posltreatment

0.56 0.57 0.82 0.41

21.0 ± 0.62 20.2 ± 0.57 19.6 ± 0.79 19.9 -+ 0.46

321.0 324.6 ± 322.8 ± 327.2 ±

4.73 4.42 3.72 3.42

321.1 ± 5.26 319.6 ± 4.75 312.9 -+ 3.83 318.9 _+ 3.14

324.9 _+ 5.70 317.2 ± 5.68 307.3 ~ 3.60 316.5 ± 2.99

22.8 ± 0.57 22.4 ± 0.39

22.3 ± 0.58 21.0 ± 0.49

22.6 ± 0.43 21.7 _+ 0.49

331.2 _+ 3.75

334.7 ± 3.65 329.4 + 3.73

340.8 ± 3.54 328.0 ± 3.65

23.5 -+ 0.55 23.3 -+ 0.50

21.5 ± 0.52 20.4 ± 0.43

21.2 ± 0.49 21.7 ± 0.46

331.0 +_ 3.68 330.9 ± 4.24

331.9 _+ 3.37 330.2 _+ 3.94

329.9 __*3.69 328.1 ± 3.92

were on a 12 hr: 12 hr day-night cycle with lights coming on at 8 a.m. They were allowed access to food and water ad lib; the food was Pilsbury diet 41B pellets.

Apparatus The confinement chambers were made of clear Plexiglas. These had four converging sides; there were five slits in the top, spaced 1 cm apart, through which a back section could be inserted to accommodate rats of different sizes. A hole at the front of the chamber ensured adequate ventilation. An Oertling model HC22 digital electronic averaging balance was used for all the weighings.

Procedure The experiment lasted 15 days, divided into three blocks of 5 days. During the first (pretreatment) and last (posttreatment) blocks only food intakes and body weights were measured. The allocation of rats to treatment conditions was made at the end of the 5 pretreatment days in a way that ensured that the groups were homogeneous in weight and intake. In the middle block the rats were also injected immediately after being weighed. Weighings were made at 1 p.m.; the food pellets remaining were weighed as well as any food crumbs that had fallen from the hoppers into the animals' cages. A weighed amount of food was then placed back in the food hopper so that the amount eaten in the next 24 hours could be calculated by subtracting weights. During the 5 treatment days the rats were injected subcutaneously with 0.1 ml of sunflower seed oil. The oil contained the appropriate amount of estradiol benzoate, obtained from Sigma Chemicals U.K. Ltd., for the treatment condition the animal was in. The rats that were to be confined were placed in the restrainers immediately after being injected and left there for 30 min. After restraint the animals were returned to their cages. Nonrestrained rats were handled in the same way as restrained rats but returned to their home cages soon after being injected. RESULTS Mean food intake and body weight data for the baseline, treatment and posttreatment periods are summarized in

Table 1. Before analyzing the intake and weight data for the treatment and posttreatment days, scores were proportionalized by dividing each value by the animal's average intake or body weight score during the pretreatment period. These transformed data were then analyzed using mixed design analyses of variance. The proportionalized data on intake are summarized in Fig. 1. An analysis carried out with restraint and hormone treatment as between subject factors and treatment/posttreatment phase and days as within subject factors showed there was an interaction between the first three factors. F(3,32)=3.57, p<0.025. Therefore. to simplify the picture two further analyses were carried out with just three factors (restraint, hormone treatment, and days) one for the treatment phase and another for the posttreatment phase. In the first of these analyses there were significant main effects of restraint, F(1,32)=23.7, p<0.001, and hormone treatment, F(3,32)=7.0, p<0.001. However, a simple main effects analysis showed that the hormone effect in the unconfined rats was not quite significant (p <0. I) whereas there was a significant effect in the restrained rats, F(3,32)=5.3. p<0.01. Comparing restrained and unrestrained animals at dose levels showed no significant difference between the oil-treated groups, nor those given 0.625 tzg of EB, but the differences were significant at the 1.25 /zg, F(1,32)=11.8, p<0.01, and 2.5 /zg, F(1,32)=7.9, p<0.01, doses. Further comparisons between pairs of means were made using the Newman-Keuls statistic: within the confined groups those given the intermediate and highest hormone doses showed significantly more suppression of eating than the oil-treated animals (p<0.05 in each easel. Comparisons between the unstressed oil-treated animals and each of the three EBtreated confined groups showed a nonsignificant difference in the case of the lowest dose group but significant differe n c e s in the case of the 1.25/~g and the 2.5 ~g dose groups (p<0.01 in both cases). In general there was a progressive reduction in eating over the 5 treatment days which accounts for the si.maiflcarrt days effect in the analysis, F(4,128)= 13.1, p <0.001; days did not interact with any other factor. There was no difference in the suppression in intake found between the first two days of treatment nor between any of the last three days of treat-

E S T R O G E N , STRESS A N D A N O R E X I A

3 Days

Days 1 I

2

3

4

5

6

7

s

9

M

s

"7

0

.~_

-10

g

-~s

~

-°

°

,\

/

/ \°

\,

o

.o

4

5

6

o

•

-20

10

0~0..,.._..0

=

0 ~

°~a -

-/

o

-1 -2 Unconfined

Groups

-4 -5

groups

Days

-6 -3~

9

0-------0

-3

Unconfined

8

°~ O~

1

,.~

-25

7

3

o

o.

oi"+

3

1o

I0 -~

2

1

Days

2

3

4

5

6

7

8

9

10

4 1

2

3

4

5

6

7

8

9

10

I0 "~

1

.=_

o -g~o.~°--._o,

0 -5

.E

._~

-15

o~O~ -20 .25

.,f

.

.

.

.

-1 -2

-10

m

N

2

\

o

~ °-~

=

'

~ -4

o

Confined

groups

"6

\.__.7 Confined

Groups

.30 FIG. 1. Mean percentage change in food intake (compared with baseline days) during treatment (days 1 to 5) and posttreatment (days 6 to 10) periods; oil treatment (i---~), 0.625 p.g EB (O--O), 1.25 p.g EB (I---il), 2.5/zg ([]---U).

FIG. 2. Mean percentage change in body weight (compared with baseline days) during treatment (days 1 to 5) and posttreatment (days 6 to 10) periods; oil treatment (O-----~),0.625 tzg EB (O---~), 1.25/~g EB ( l - - - I ) , 2.5 ~g ([3---[3).

ment, however, there was significantly more suppression of eating on each of the last three days of treatment than on either of the first two days of treatment (O<0.01 or better in all pair-wise comparisons). During the posttreatment phase the animals that were restrained continued to eat less than the unconfined rats, F(1,32)=8.59, p<0.01, but the effect from hormone treatment just failed to achieve significance (o<0.06). In a comparison of restrained and unrestrained rats across hormone treatment only the oil-treated condition was significant F(1,32)=5.9, p<0.025. Although there was a main effect of days, F(4,128)= 19.4, p<0.001, there was also an interaction between hormone treatment and days, F(12,128)=1.85, p<0.05. This is explained by the significance of the hormone treatment being progressively reduced across posttreatment days, so by day 15 it was not significant; the simple main effects of days for the intermediate, F(4,128)=5.7, p<0.001, and highest, F(4,128) = 13.3, p <0.001, hormone dose groups (combined across the restraint factor) were highly significant, whereas the days effects were small in the oil-treated, F(4,128)=3.3, p<0.05, and low hormone dose groups, F(4,128)=2.7, p <0.05. These days effects reflect the recovery in eating that occurs with cessation of treatment (see Fig. 1). The reason why the simple main effect of hormone just failed to reach significance in the unconfined animals during the treatment period may be in part because the comparison

was made between groups. A further mixed design A N O V A was therefore performed on raw intake data just for these subjects, the between subjects factor was hormone treatment, and there were two within subject factors: pretreatment versus treatment phase (so the effect of hormone treatment can be studied within groups) was one and days was the other. There was a significant effect of phase, F(1,16)=22.67, p <0.001; in a simple main effect analysis the oil-treated animals ate comparable amounts in the two periods (p>0.5), the low dose animals nearly reduced intake by a significant amount in the treatment period (O<0.1) and the intermediate, F(1,16)=7.8, p<0.025, and high dose, F(1,16)= 16.9, p<0.001, rats did eat significantly less during this period. The proportionalized body weight data are summarized in Fig. 2. There was a significant four way interaction, F(12,128)=2.27, p<0.025, when all four factors were included in an A N O V A performed on this data so two simpler A N O V A s were c a r d e d out as outlined above. Although in the treatment phase there were significant main effects of restraint, F(1,32)=48.46, p<0.001, and hormone treatment, F(3,32)=8.09, p<0.001, on a simple main effects analysis there was a significant effect of hormone on restrained animals, F(3,32)=9.0, p<0.001, but not on unconfined animals. When restrained and unconfined animals were compared at different levels o f hormone treatment, there were differences at each of the three doses of EB,

4

S[EVENS

F(1,32)=9.5, 28.7, 13.9 w i t h p < 0 . 0 1 or better in each case, but the difference was not quite significant for the oil-treated groups (p <0.1). In the restrained animals each hormone dose group lost significantly more weight than did the oil-treated restrained animals (,o<0.05 or better in the three comparisons). During the posttreatment phase both restraint, F(1,32) =45.71, p<0.001, and hormone treatment, F(3,32)= 26.23, p<0.001, continued to have an effect. But the effect of the hormone treatment was now present in confined, F(3,32)= 19.5, p<0.001, and unrestrained, F(3,32)=9.0, p<0.001, groups. When comparing between levels of restraint there were small differences between the oil-treated groups, F(1,32)=5.1, p<0.05, and the lowest hormone dose, F(1,32)=7.0, p<0.025, and large effects for both the higher hormone doses, F(1,32)= 31.2, p<0.001 and, F(1,32)=9.3, p<0.01, for the intermediate and high doses respectively. DISCUSSION

The results are suggestive of a potentiation between the effects of stress induced by bodily constraint and the effects of treatment with estradiol benzoate in this experiment. Animals that were confined and treated only with oil reduced their intake by 9.3% during treatment whereas unconfined animals also treated with oil ate 1.8% l e s s - - w e have consistently found such reductions in this baseline control condition during treatment, these may be a consequence of the injection procedure or because of experimenter effects or as a response to pheromones released by the stressed rats. Thus confinement alone produces a 7.5% reduction in eating (the difference between these two values). There was a progressive increase in the effectiveness of estradiol in unrestrained rats during the treatment phase, the smallest dose caused a 4.3% loss in intake, the middle dose produced a 6.1% loss, and the highest a 10.1% loss (over that found in the oil-treated group). If these reductions are added to that produced by confinement alone (7.5%) we can predict that on a simple additive basis there should be reductions of 11.8%, 13.6% and 17.6% for the three groups o f restrained and hormone-treated animals. However we found a 11.2%, 20.3% and 21.7% reduction in eating in these three groups. At the lowest dose there is a close correspondence between predicted and actual food intake reduction, while at the other doses there are large discrepancies--the largest being at the 1.25 ~g dose--indicative of a potentiation between the effects of the combined treatments. Looking at the changes in body weight over the treatment period gives a similar picture, but because of the small changes that occur in body weight the values are small. The predicted loss in body weight in the low dose restrained group is 1.81% (summing the losses attributable to restraint alone and that dose of hormone alone) but the acutal loss in weight is 2.63%. F o r the 1.25 t~g hormone dose restrained group the predicted loss is 1.89%, actual loss was 4.13%; and in the 2.5/~g confined rats these values are 3.31% and 3.58% respectively. Again the pattern is like that for intake, the biggest discrepancy is in the second largest hormone dose condition. Thus it seems that our past use of estradiol doses of 2.5/~g [(7,8); Stevens, 1986 unpublished] is a crucial factor in explaining why we previously found additivity between estradiol and stress and not potentiation as in this study. If the suppression of eating produced by estrogen is large, and about the same in size as caused by stress alone.--as is true

of our previous studies--then additivity may be the only possible outcome because summation of the two suppression effects may reach a 'ceiling' of effectiveness. Thus potentiation can only arise when the suppression effect of estrogen is small, as was the case with our 1.25 p~g dose (which is not a subthreshold dose according to our analysis of intake during the baseline and treatment phases). A further aspect of the data is worthy of comment. There were significant effects of repeated treatments (days effects) on both intake and body weight (p <0.0001 in both cases) that are consistent with our previous findings. Examination of the data showed that there was no adaptation to the effects of treatments, whether hormone alone or restraint alone, or these in combination during the treatment period. This is unlike the finding of Kennett et al. (9,10) who reported an adaptation to the effects of stress after 3 days. These differences between their findings and ours may arise because of the shorter period of restraint that we use (30 min versus 2 hours). Also there are procedural differences in the way the animals are restrained, they strapped their rats to a grid, and sex differences in the subjects u s e d - - t h e y used males. However, Donohoe et al. (6) used female rats in a similar procedure to Kennett et al. and also found adaptation to restraint by the third day so it is unlikely that the difference can be explained by a sex factor. Whether there would be an adaptation to the effects of restraint (alone or combined with hormone treatment) if the treatments were prolonged is unclear. The longest restraint period that has been used in this laboratory is 7 days (8) and in this instance, as here, intake suppression stabilized during the latter period of treatment. Since there is potentiation between some low doses of estrogen and stress on the suppression of eating how might this happen? The reductions in eating that occur when female rats are exposed to stress and estrogen may be produced by independent mechanisms that happen to act synergistically under certain parametric values. Alternatively, there may be a common mechanism that is activated either by stress or by estrogen and whose effect is to suppress eating, and, somewhat more slowly, body weight. Under certain circumstances it shows positive feedback effects, hence an amount of estrogen that is insufficient to cause much suppression in eating biases the mechanism's response so that a major suppression in eating is triggered by exposure to stress. In this model estrogen acts as a neuromodulator in a neural mechanism that is involved in response to stressors. This mechanism may be an aminergic midbrain-forebrain system or alternatively, and I think less plausibly, an opiate system. Serotonin (5-HT) might be the neurotransmitter used by this system since stress increases 5-HT concentrations in the brain (4), and increased levels of 5-HT are associated with eating suppression (3). Biegon and McEwen (2) showed that estradiol treatment alters 5-HT receptor density, in the hypothalamus, preoptic area and a m y g d a l a - regions that are rich in estradiol receptors. These areas have been implicated in stress and the regulation of eating and could contribute to the neural mechanism in question. Why should a synergistic effect between stress and estrogen be fundamental to a model of anorexia nervosa? After all typical anorexic girls have plasma levels of gonadal hormone that are significantly lowered in concentration [see (14)] while other aspects of their endocrine profile indicate they are under stress [see (14,17)]. But Russell (13) has argued that a hypothalamic dysfunction is involved in anorexia nervosa and Young (18) proposed that there is an incomplete

E S T R O G E N , STRESS A N D A N O R E X I A

5

hypothalamic maturation at puberty in anorexic girls. By combining these arguments with the evidence that the hypothalamus is as much as 500 times m o r e sensitive to estrogen before puberty than after it (11), based on the estrogenic suppression of luteinizing hormone, we have the basis of a plausible psychobiological model of anorexia nervosa. This disorder would occur when the ovaries start functioning at puberty but only in those individuals who have a hypothalamic disturbance and who are subjected to additional environmental stressors. The subsequent reduced plasma levels of estrogen in the continuing anorexic would be part of a compensatory response to the hypothalamic supersensitivity.

The significant aspect of this experiment is that some really low levels of estrogen combine with stress in a multiplicative and not a simple additive fashion which is consistent with the endocrine profile of the anorexic girl who is being subjected to one or several environmental stressors.

ACKNOWLEDGEMENTS I would like to thank Lynne Easom and Teresa Sharpe for their help in collecting data and for their excellent care of our animals, and David Matthews for the photography.

REFERENCES 1. Antelman, S. M.; Schetzman, H.; Chin, P.; Fisher, A. E. Tailpinch induced eating, gnawing and licking behavior in rats: dependence on the nigrostriatal dopamine system. Brain Res. 99:319-337; 1975. 2. Biegon, A.; McEwen, B. S. Modulation by estradiol of serotonin receptors in the brain. J. Neurosci. 2:199-205; 1982. 3. Blundell, J. E. Is there a role for 5-hydroxytryptamine in feeding? Int. J. Obes. 1:15-42; 1977. 4. Campbell, I.; Cohen, R. M.; Murphy, D. L.; Torda, T. Some biochemical changes associated with stress: possible relationship to the onset of depression. In: Rose, F. C., ed. Metabolic disorders of the nervous system. London: Pitman; 1981:446460. 5. Donohoe, T. P. Stress induced anorexia: implications for anorexia nervosa. Life Sci. 34:203-218; 1984. 6. Donohoe, T. P.; Kennett, G. A.; Curzon, G. Immobilization stress-induced anorexia is not due to gastric ulceration. Life Sci. 40:467-472; 1986. 7. Haslam, C.; Stevens, R.; Donohoe, T. P. The influence of cyproheptadine on immobilization and oestradiol benzoate induced anorexia in ovariectomized rats. Psychopharmacology (Berlin) 93:201-206; 1987. 8. Hayward, C. The effect of ovarian and stress hormones on feeding. Doctoral thesis, University of Nottingham; 1986. 9. Kennett, G. A.; Dickinson, S. L.; Curzon, G. Enhancement of some 5-HT-dependent behavioural responses following repeated immobilization in rats. Brain Res. 330:253-263; 1985.

10. Kennett, G. A.; Dickinson, S. L.; Curzon, G. Central serotonergic responses and behavioural adaption to repeated immobilization: the effect of corticosterone synthesis inhibitor metyrapone. Eur. J. Pharmacol. 119:143-152; 1985. 11. Kulin, H. E.; Grumbach, M. M.; Kaplan, S. L. Gonadalhypothalamic interaction in prepubertal and pubertal man: effect of clomiphene citrate on urinary follicle-stimulating hormone and plasma testosterone. Pediatr. Res. 6:162-171; 1972. 12. Perhach, J. L., Jr.; Barry, H. Stress responses of rats to acute body or neck restraint. Physiol. Behav. 5:443-448; 1970. 13. Russell, G. F. M. The present status of anorexia nervosa. Psychol. Med. 7:363-367; 1977. 14. Stevens, R. Anorexia nervosa: an enigmatic disorder. In: Stevens, R., ed. Aspects of consciousness, vol. 4. Clinical issues. London: Academic Press; 1984:167-205. 15. Wade, G. N. Gonadal hormones and behavioral regulation of body weight. Physiol. Behav. 8:523-534; 1972. 16. Wade, G. N. Sex hormones, regulatory behaviors and body weight. In: Rosenblatt, J. S.; Hinde, R. A.; Shaw, E.; Beer, C., eds. Advances in the study of behavior, vol. 6. New York: Academic Press; 1976:201-279. 17. Walsh, B. T. Endocrine disturbances in anorexia nervosa and depression. Psychosom. Med. 44:85-91; 1982. 18. Young, J. K. A possible neuroendocrine basis of two clinical syndromes: anorexia nervosa and the Kleine-Levin syndrome. Physiol. Psychol. 3:322-330; 1975.