Multiple exposure to activity anorexia in rats: effects on eating, weight loss, and wheel running

Multiple exposure to activity anorexia in rats: effects on eating, weight loss, and wheel running

Behavioural Processes 61 (2003) 159–166 Multiple exposure to activity anorexia in rats: effects on eating, weight loss, and wheel running Benjamin M...

100KB Sizes 0 Downloads 27 Views

Behavioural Processes 61 (2003) 159–166

Multiple exposure to activity anorexia in rats: effects on eating, weight loss, and wheel running Benjamin M. Hampstead a,∗ , Lynda Paulson LaBounty b , Carmen Hurd b a

Department of Psychology, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104, USA Department of Psychology, Macalester College, 1600 Grand Avenue, St. Paul, MN 55105, USA


Accepted 5 December 2002

Abstract Animals were given five cycles of an activity anorexia (AA) procedure in order to determine the effect of additional experience on eating, running, and weight loss. Female Sprague–Dawley rats were given a 1 h meal and allowed access to a running wheel for the remainder of each day. Upon reaching 75% of free-feeding body weight, each animal was denied wheel access and given ad libitum food until it regained the lost weight. Then, food was again restricted and wheel access provided. Sedentary control animals were placed on the restricted feeding schedule for the median number of days experimental animals required to reach weight loss criterion. Experimental animals showed adaptation by increasing food consumption and decreasing the rate of weight loss despite an increase in running across cycles. Additionally, the distribution of running shifted gradually so that during the later cycles, much of the running occurred in the hours just before feeding. The results support the hypothesis that running interferes with adaptation to the restricted feeding schedule and also that the marked increase in anticipatory behavior during the later cycles is primarily responsible for the maintenance of AA. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Activity anorexia; Anticipatory behavior; Repeated exposure; Wheel running

1. Introduction When rats are placed on a restricted feeding schedule and given access to a running wheel, they typically become excessively active, reduce their food intake, lose weight, and may eventually starve. In contrast, control subjects placed on the same feeding schedule with no access to a running wheel will stabilize their body weights at a reduced level and survive (Pierce and Epling, 1991). This excessive wheel running and associated reduced food consumption has been var∗

Corresponding author. E-mail address: [email protected] (B.M. Hampstead).

iously termed activity anorexia (Pierce and Epling, 1991), activity-based anorexia (Epling et al., 1983), and self-starvation (Routtenberg and Kuznesof, 1967). We will use the term activity anorexia (AA). During periods of increased activity, decreased food consumption seems paradoxical, since one would expect animals to eat more to compensate for the increased energy demands. Adjusting the feeding schedule is known to mitigate weight loss during an AA procedure. Rats given two 30 min or four 15 min meals per day as opposed to the standard 1 or 1.5 h meal consumed more food and did not develop AA (Kanarek and Collier, 1983). Based on these results, Kanarek and Collier believe AA occurs because animals fail to adapt to the restricted feeding schedule.

0376-6357/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0376-6357(02)00188-2


B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166

More recent work has hypothesized that the high rates of running (Boakes and Dwyer, 1997) and, more specifically, the anticipatory running that develops in the hours prior to feeding (Dwyer and Boakes, 1997) prevents the animals from adapting to the restricted feeding schedule. Exposing animals to the restricted feeding schedule prior to the start of an AA procedure has been found to result in higher survival rates (Routtenberg, 1968) and a decreased likelihood that the rats will develop ulcers, which are a common measure of stress in this paradigm (Pare et al., 1985). Dwyer and Boakes (1997) allowed one group of animals to fully adapt to the restricted feeding schedule before beginning the AA procedure. The preadapted group consumed more food and lost less weight than the nonadapted group. Additionally, preadapted animals increased their food intake as they gained experience with the restricted feeding schedule (Boakes et al., 1999). These findings suggest that wheel running does interfere with adaptation to the feeding schedule. A rival hypothesis has emerged, stating that running serves to suppress food intake leading to the rapid weight loss (e.g. Lett et al., 2001). Although the specific cause of this decrease is still uncertain, both sickness (Lett and Grant, 1996) and addiction to endogenous opiates (Pierce and Epling, 1991) have been offered as potential causes. Lett et al. (2001) modified the procedure used by Boakes and Dwyer (1997) and found that, despite being fully adapted to the feeding schedule, the animals that were given wheel access consumed less food and lost weight more rapidly than controls that were also preadapted. The authors believe that, because animals were fully adapted to the restricted feeding schedule, running should have had no effect on food intake and AA should not have developed. Therefore, Lett et al. (2001) concluded that running does suppress food intake in some fashion. As indicated earlier, prior access to the restricted feeding schedule has been found to increase food consumption and decrease weight loss. Additionally, prior access to the running wheel alone has also resulted in increased food consumption (Boakes and Dwyer, 1997). Therefore, we hypothesized that if animals were repeatedly exposed to both the restricted feeding schedule and the running wheel, they would adapt by increasing food consumption and decreasing weight loss and thus eventually not develop AA at all. Accordingly, a repeated acquisition procedure was

devised where rats were placed on a standard AA procedure until they lost 25% of their weight and were then refed until their free-feeding weight was reached. Following this, they were once again put on the AA procedure. This was repeated for five “cycles”. Such a method also allows for the comparison of the hypotheses offered by Boakes and Dwyer (1997) and Lett et al. (2001). Both hypotheses would predict an initial decrease in food consumption and increased weight loss. Animals should increase food consumption and reduce weight loss and wheel running across cycles if activity does interfere with adaptation. However, if running does suppress food intake, we would expect lower food consumption with higher rates of running. This relationship would be unlikely to change across cycles.

2. Materials and methods 2.1. Subjects Twenty female Sprague–Dawley rats, 10 experimental and 10 control, served as subjects in this study. All animals were bred in the Macalester College animal facility, were 60–70 days old, and weighed 190–210 g at the start of the experiment. These animals had been previously briefly food deprived (2–3 days at 90% or greater body weight) and used in a demonstration of schedule-induced polydipsia. All animals were housed singly in plastic cages in a room maintained on a 12/12 h light/dark cycle (lights on at 7 a.m.) and at 21 ◦ C. Ad libitum food and water was provided until the start of the experiment. Rats were allowed free access to water throughout experimentation. 2.2. Apparatus Each experimental animal was housed in a wire mesh cage that was attached to 1.0 m circumference activity wheel. An opening, approximately 7.1 cm (h) by 6.1 cm (w), in the sidewall of each cage allowed the rat to move between the cage and the wheel. Sliding the cage backward exposed the opening to the wheel and allowed access, while moving the cage forward covered the opening thus denying access. A microswitch, attached to an arm extending from the running wheel, was connected to a Dell 486 computer

B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166

that used Med-PC software to record the number of wheel revolutions in 10-min intervals. 2.3. Procedure In order to acclimate animals to their new environment, rats were individually housed in the wire cages with attached running wheels for 4 days before the start of the experiment. During these 4 days, rats did not have access to the running wheels. On the first day of the study, food was removed at 2 p.m., and the experimental rats were given free access to the running wheels for 23 h, until 1 p.m. the next day. At this time, access to the wheels was denied, the rats were individually weighed, and given a previously weighed amount of rat chow. After 1 h, the remaining food was collected and weighed, including any sizable chunks that fell through the wire mesh, and the experimental rats were again given access to the wheels. This routine of exercise, weighing, and feeding was continued until a rat reached 75% of its prerestriction weight, whereupon it entered the refeeding phase. During the refeeding phase of the experiment, access to the wheel was denied, rats were given food ad libitum, and were weighed daily at 1 p.m. The criterion for completion of the refeeding phase was 4 consecutive days at or above the previous (100%) free-feeding weight. Upon meeting this criterion, a new free-feeding weight was calculated using the average of those 4 days. At 2 p.m. on the fourth criterion day, rats began the next cycle of the experiment. A cycle was defined as the completion of a food-restricted/activity phase and a refeeding phase. The experiment consisted of five such cycles. Data from control rats was gathered after testing of all experimental animals was completed. Four days before the start of their first cycle, animals were placed in individual wire mesh cages. At 2 p.m. on the fourth day, food was removed and control rats began their first cycle, but were given no wheel access. As with experimental animals, control rats were weighed daily at 1 p.m. and given 1 h access to food, until 2 p.m. that same day. Control rats were also required to complete five cycles of food restriction and refeeding, but the number of days they were food restricted was based on the median number of days experimental rats required to reach 75% body weight in a given cycle. As a result,


control rats were to be food restricted for 6, 7, 9, 9, and 10 days for cycles 1 through 5, respectively. Due to technical errors in data recording, data from two experimental and three control rats were unusable and were excluded from analysis. Thus, eight experimental and seven control rats completed the study. The control rats were also inadvertently removed from cycle 3, 2 days prematurely. As a result, they received 7 as opposed to the scheduled 9 days of food restriction during this cycle. 2.4. Data analysis Data were analyzed using ANOVA (Group by Cycle) with cycle as a repeated measure. Effects were considered significant if P < 0.05. SPSS 10.0 (SPSS Inc.) was used for all analyses.

3. Results The amount of food consumed by experimental and control animals across the five cycles is presented in Fig. 1. The amount consumed increased across the five cycles for both groups. However, consumption was initially lower for the experimental group and increased more rapidly than for the sedentary controls. These data were submitted to a 2 (group) × 5 (cycle) mixed design ANOVA in which group was a between-subjects variable and cycle was within-subjects. Over the five cycles, rats in the control group did not eat significantly more than those in the experimental group (ME = 7.64 versus MC = 7.71), yielding no main effect for group (F < 1). The Cycle main effect, however, was highly significant [F(4, 52) = 30.87, P < 0.001]. As is evident in Fig. 1, this main effect was qualified by a significant Group × Cycle interaction [F(4, 52) = 10.78, P < 0.001]. Indeed, a linear contrast analysis revealed that the slopes of the Experimental and Control functions differed significantly [F(1, 13) = 19.41, P = 0.001] confirming that eating in the experimental group increased more rapidly than that in the control group. Fig. 2 shows the average percent of weight lost per day for experimental and control animals for each of the five cycles. For both groups daily average weight loss decreased with successive cycles. In addition, the


B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166

Fig. 1. Mean grams of food consumed by each group during the food-restriction/activity portion of each cycle (±S.E.M.).

experimental group lost more weight than the controls in all five cycles. These data, also, were submitted to the same ANOVA test as above. The observed difference in weight loss between the experimental and control groups was significant [ME = 3.83 versus MC = 2.01, F(1, 13) = 119.00, P < 0.001]. There was also a significant Group × Cycle interaction [F(4, 52) = 10.87, P < 0.001]. A linear contrast analysis confirmed that the slopes of the Experimen-

tal and Control functions shown in Fig. 2, indeed, differed significantly [F(1, 13) = 29.23, P < 0.001]. The experimental group also required more days to reach the 75% body weight criterion with successive cycles. These animals required an average of 6, 6.75, 8.5, 9, and 9.5 days to lose the weight in cycles 1–5, respectively. These differences in days to criterion body weight were significant [F(4, 28) = 17.12, P < 0.001].

Fig. 2. Average percentage of body weight lost during the restriction/activity portion of each cycle (±S.E.M.).

B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166


Fig. 3. Average number of wheel revolutions run by experimental rats during the restriction/activity portion of each cycle (±S.E.M.).

As food consumption increased and weight loss per day decreased across the five cycles, wheel running by the experimental group increased rapidly through three cycles and then leveled off (Fig. 3). These data were submitted to a within-subjects design ANOVA in which cycle was the within-subjects variable. The observed increase in running was significant [F(4, 28) = 28.85, P < 0.001]. In addition, a linear within-subjects contrast analysis confirmed that

the slope of the function was significantly positive [F(1, 7) = 113.79, P < 0.001]. The distribution of running by the experimental group throughout the 23 h changed dramatically over the five cycles. Fig. 4 shows hour by hour running for cycles 1, 3, and 5. These three cycles are shown for ease of reading the figure. During the first cycle, animals ran more during the dark period (dark period indicated by solid horizontal line beginning at sixth hour

Fig. 4. Hourly distribution of daily running during cycles 1, 3, and 5 by experimental animals. Bold horizontal line along the top right side represents the 12 h dark cycle. Feeding occurred at time 0.


B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166

Fig. 5. Proportion of the maximum cycle mean for food consumption, wheel revolutions, and percentage of weight lost per day across the five cycles. Thus, the largest mean for each variable was assigned a value of 1.0 while the other four means for each variable were transformed into a proportion of that mean.

after feeding). As animals gained more experience with the schedule in subsequent cycles, they increased the proportion of running that occurred during the 4 h before feeding. ANOVA results showed significant differences in the percentage of running performed during these 4 h before feeding when cycles 1, 3, and 5 were analyzed [F(2, 14) = 32.68, P < 0.001]. In order to compare the changes in running, eating, and weight for the experimental animals across the five cycles, the group means for each cycle were transformed as follows. For each variable, each cycle mean was divided by the maximum mean for that variable. This transformation into proportion of maximum value made it possible to include all three resulting functions in one figure. For example, the largest weight loss occurred in the first cycle, so that data point is 1.0. In subsequent cycles, weight loss was 0.92, 0.69, 0.64, and 0.60 of the average daily weight loss in the first cycle, respectively. Fig. 5 shows the resulting functions for eating, weight loss, and running. Most striking in Fig. 5 is the close correspondence between the functions for eating and for running while weight loss rate is roughly inversely related to eating and running. For all three variables, the greatest changes occur over the first three cycles and then change relatively little thereafter.

4. Discussion With repeated exposure to the AA procedure, experimental rats ate more and lost weight more slowly even though they increased wheel running across cycles. Control rats also consumed more food and slowed weight loss with successive cycles. However, experimental animals continued to reach the 25% weight loss criterion indicating that the high levels of running were responsible for this weight loss as control rats remained above criterion throughout the experiment. The results of this study support Boakes and Dwyers’ (1997) hypothesis that running interferes with adaptation to the feeding schedule. The experimental animals increased their food intake with each cycle, actually consuming as much or more than control animals during cycles 3, 4 and 5 despite the accompanying increase in wheel running. In addition, the percentage of body weight lost per day decreased across cycles. Had running suppressed food intake as proposed by Lett et al. (2001), the 84% increase in total running from cycles 1 to 3 should have resulted in decreased food consumption across those cycles by the experimental animals and an increase in weight lost per day. Instead, animals increased food consumption by 54% and decreased weight loss by 31% from cycles 1 to 3.

B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166

Our results also support Dwyer and Boakes’ (1997) findings that anticipatory behavior contributes to the development and maintenance of AA. These authors found animals given wheel access for only the 4 h prior to feeding ran more and were more likely to develop AA when compared to animals given wheel access for the 19 h immediately following feeding, but not the 4 h before feeding (Dwyer and Boakes, 1997). Our experimental animals shifted their pattern of daily running throughout the course of the experiment. Animals decreased running in the hours after feeding, while at the same time greatly increasing running in the hours before feeding. During the first cycle, approximately 15% running occurred during the 4 h before feeding. This increased to approximately 23% during the third cycle and more than 35% of the total daily revolutions during the fifth cycle. The excessive running prior to food delivery was apparently induced by the food schedule and had the paradoxical effect of contributing to the continued weight loss. Thus, while total running interfered with adaptation to the restricted feeding schedule during the early cycles, the anticipatory running, not reduced food consumption, caused the continued weight loss during the later cycles. This finding also fails to support the hypothesis put forth by Lett et al. (2001), because as running increased across cycles, food consumption should have decreased. Many species of rodents have been found to become excessively active in times of food restriction or famine (Cornish and Mrosovsky, 1965). The observed increase in running during the hours before feeding in the present study has also been noted by other authors and labeled a form of anticipatory behavior (Dwyer and Boakes, 1997; Boakes et al., 1999; Beneke et al., 1995; Boulos et al., 1980). In the present study, light onset may have served as a cue for the forthcoming delivery of food. To test this hypothesis, rats could be placed on a similar multiple cycle procedure with either lights on continuously or randomized feeding periods in each 24 h period. Such procedures should reduce or eliminate the anticipatory running and thereby eliminate or reduce the sustained weight loss as well if light onset affected the pattern of running. The use of female rats in the present study may be cause for concern as the majority of studies to date on AA have used male rats as subjects. No differences in


rate of weight loss or in the proportional amount of food consumed were found when male and female rats of the same age and litter were compared in an AA procedure (Boakes et al., 1999). However, female rats were found to run more than males despite lower initial body weights. These authors also found that male rats ran significantly more as they lost more weight, while female rats maintained high levels of running regardless of their body weight. It is possible that the estrus cycle may also have affected the animals used in the current study. In a study of estrus cycle, body temperature, and wheel running, Kent et al. (1991) found that females displayed a 4-day cycle when given access to a running wheel. These authors also found that body temperature was affected more by activity than by hormones, thus it seems that hormones are not directly causing the animals to run. Since hormones do not appear to directly induce wheel running, and female rats were found to have equivalent levels of food intake and weight loss when compared with males (Boakes et al., 1999), the use of female animals in the present study does not appear to be a significant concern. Additionally, the potential clinical relevance of this study is improved through the use of female animals, as a greater percentage of humans with anorexia are female. It is unlikely that failure to provide the full 9 days of food restriction to the control animals in cycle 3 biased the results. Experimental animals consumed more food than controls throughout cycle 3. Although control animals ate more in cycle 4 than in cycle 3, consumption during cycle 5 returned to an amount similar to cycle 3. In addition, experimental animals continued to consume more food than controls during cycles 4 and 5. In summary, providing the animals with five cycles, and essentially five times the experience with the AA procedure resulted in gradual recovery of eating to levels that were equivalent if not greater than sedentary controls, and slowed weight loss in spite of increases in running throughout the five cycles. Additionally, the pattern of running changed as animals ran more in the hours before feeding during the later cycles. The results support the hypothesis that running interferes with adaptation to the restricted feeding schedule and also that the marked increase in anticipatory behavior during the later cycles is primarily responsible for the maintenance of activity anorexia in these animals.


B.M. Hampstead et al. / Behavioural Processes 61 (2003) 159–166

Acknowledgements This research was supported in part by the Macalester College Psychology Department, and the NIMH-IRP. The authors wish to thank Robert Hampton for all his comments on earlier versions of this article and Jason Bomberger for his assistance in conducting this research. A summary of this research was presented at the Association for Behavior Analysis Conference in May of 2000, Washington D.C. References Beneke, W., Schulte, S., Vander Tuig, J., 1995. An analysis of excessive running in the development of activity anorexia. Physiol. Behav. 58, 451–457. Boakes, R.A., Dwyer, D.M., 1997. Weight loss in rats produced by running: effects of prior experience and individual housing. Q. J. Exp. Psychol. 50B, 129–148. Boakes, R.A., Mills, K.J., Single, J.P., 1999. Sex differences in the relationship between activity and weight loss in the rat. Behav. Neurosci. 113, 1080–1089. Boulos, Z., Rosenwasser, A.M., Terman, M., 1980. Feeding schedule and the circadian organization of behavior in the rat. Behav. Brain Res. 1, 39–65. Cornish, E., Mrosovsky, N., 1965. Activity during food deprivation and satiation of six species of rodent. Anim. Behav. 13, 242– 248.

Dwyer, D.M., Boakes, R.A., 1997. Activity-Based anorexia in rats as failure to adapt to a feeding schedule. Behav. Neurosci. 111, 195–205. Epling, W.F., Pierce, W.D., Stefan, L., 1983. A theory of activity-based anorexia. Int. J. Eat. Disord. 3, 27–46. Kanarek, R., Collier, G., 1983. Self-starvation: a problem of overriding the satiety signal? Physiol. Behav. 30, 307–311. Kent, S., Hurd, M., Satinoff, E., 1991. Interactions between body temperature and wheel running over the estrus cycle in rats. Physiol. Behav. 49, 1079–1084. Lett, B.T., Grant, V.L., 1996. Wheel running induces conditioned taste aversion in rats trained while hungry and thirsty. Physiol. Behav. 59, 699–702. Lett, B.T., Grant, V.L., Smith, J.F., Koh, M.T., 2001. Preadaptation to the feeding schedule does not eliminate activitybased anorexia in rats. Q. J. Exp. Psychol. 54B (3), 193– 199. Pare, W., Vincent, G., Natelson, B., 1985. Daily feeding schedule and housing on incidence of activity-stress ulcer. Physiol. Behav. 34, 423–429. Pierce, W.D., Epling, W.F. 1991. Activity anorexia: an animal model and theory of human self-starvation. In: Boulton, A., Baker, G., Martin-Iverson, M. (Eds.), Neuromethods, Volume 18: Animal Models in Psychiatry 1. The Human Press Inc., pp. 267–310. Routtenberg, A., 1968. Self-starvation of rats living in activity wheels: adaptation effects. J. Comp. Physiol. Psychol. 66, 234– 238. Routtenberg, A., Kuznesof, A., 1967. “Self-starvation” of rats living in activity wheels on a restricted feeding schedule. J. Comp. Physiol. Psychol. 64, 414–421.