- Email: [email protected]
Learned anticipatory rise in body temperature due to handling
Physiology & Behavior, Vol. 37, pp. 64%653. Copyright©PergamonPress Ltd., 1986. Printed irt the U.S.A.
0031-9384/86$3.00 + .00
Learned Anticipatory Rise in Body Temperature Due to Handling ROELOF EIKELBOOM
D e p a r t m e n t o f Psychology, Queen's University, Kingston, Ontario, Canada K7L 3N6 R e c e i v e d l l O c t o b e r 1985 EIKELBOOM, R. Learned anticipatory rise in body temperature due to handling. PHYSIOL BEHAV 37(4) 649--653, 1986.--Hyperthermia produced by handling becomes evident at the initial daily measurement if temperature is measured at a consistent time. This hyperthermia may be a learned effect occurring in anticipation of handling. In Experiment One male Wistar rats were either unhandled or had their temperatures measured daily in the dark or the light part of the day. All animals had their temperatures measured on Day 29, in the dark. Rats usually tested in the dark were hyperthermic, 38.8°C, relative to rats previously handled only in the light, 38. I°C, and to naive rats, 37.9°C. In Experiment Two rats were handled three times daily in either the light or the dark. On Day 9 each group was divided in two, and temperatures were measured at either the usual time or at the other time. Rats tested at their usual time were hyperthermic, relative to rats normally handled in the other part of the cycle. This suggests a conditioned hyperthermia occurs in response to stimuli predictive of handling. Body temperature Conditioning Rectal temperature measurements
Handling stress Thermoregulation
Arousal
SIMPLY measuring a rat's rectal temperature elevates body temperature about 1.0°C for over an hour [3, 7, 12, 14]. This elevation has been attributed to the arousing effects of handling the rats [3,12]. The fact that the effects of drugs on body temperature may be masked, changed, or even prevented when they occur in conjunction with this handlinginduced rise in temperature suggests this is not an innocuous procedure [1, 12-15]. Ideally the solution to this problem would be to not handle the rats and use some form of telemetric recording of body temperature. The relative ease of rectal measurement, however, suggests that many studies will still use this method. Further, in most studies animals still need to be handled for injections or weighing and so would still undergo the handling-induced temperature increase. The present experiments report that the handlinginduced hyperthermia may be conditionable to general stimuli, i.e., time of day, presence of the experimenter, which are predictive of handling. In previous work [6,7] we noticed that the baseline home cage temperature of the first few animals handled was lower (37.2°C) than that of subsequent rats (over 38°C for the last animals handled). This difference between rats is similar in magnitude to the handling effect but cannot be due to the handling of the animals as it takes several minutes for a rat's body temperature to increase this amount [12]. Preliminary studies found that there was a correlation between the order in which the animals were handled and their body temperature. While group mean temperature of rats in several replications ranged between 37.7 and 37.9°C, the regression lines revealed a Y intercept of 37.1 to 37.3°C, the body temperature of unstressed animals [10,12]. This correlation was
Pavlovian conditioning
Hyperthermia
not evident initially and developed over the first few days of handling and body temperature measurement; it was evident even if animals had previously only been weighed. After several days of daily temperature measurement the correlations (Pearson r) between the order the rats were handled and their body temperature ranged, in the various replications, from .68 to .77. The fact that this correlation developed over days suggests it was not simply due to signals emanating from the first rats producing changes in the later tested rats. Since these studies involved daily temperature measurement the difference between the temperature of the first and last few rats may have been a learned effect occurring in anticipation of the actual temperature measurement. The arrival of the experimenter and the handling of the first few rats could act as a conditioned stimulus resulting in a conditioned temperature increase that mimicks the increase induced by the rectal probing. Conditioned temperature changes have been demonstrated using drugs that affect body temperature [4, 6, 8] and various stressful events like microwave-induced hyperthermia [2], electric shocks [5], and exercise [10]. If simply handling the rats can produce a conditioned temperature increase it suggests that temperature changes can be conditioned with relatively benign unconditioned stimuli. This conditioned effect might in turn be a confound in many of the other temperature conditioning studies. EXPERIMENT 1 If the correlation between order of handling and body temperature is due to an anticipatory conditioned increase in
~Preparation of this article was supported by a grant from the National Sciences and Engineering Research Council of Canada (A 2575) to R. Eikelboom who is a career scientist with the Ontario Ministry of Health. Portions of this work have been presented at the Canadian Psychological Association meeting, June, 1985.
649
650
klKELBOOM
temperature, then, by manipulating the stimuli predicting handling, it should be possible to affect body temperature. That is, the body temperature of animals measured in the presence of stimuli previously associated with handling should be higher than that of animals not normally handled in that situation. In this experiment animals were divided into three groups, two of which had distinctive temporal and lighting cues predicting temperature measurement and a third group that was not handled until the test days.
METHOD
Subjects Male Wistar rats (n=33) weighing between 225 and 250 g at the time of arrival were obtained from Charles River Canada, St. Constant, Quebec. Animals were individually housed in standard stainless steel cages with food and water available ad lib throughout these studies. The animal room was maintained at a constant temperature of 22_+ I°C with a 12/12 light/dark cycle (lights off 14:00 hr). One small light was constantly on to permit testing in the dark phase of the cycle.
Procedure Temperatures were measured using a Yellow Springs Telethermometer with a model 423 probe. Animals were removed from their cage, carried to a table in the animal room, and placed in a small three sided trough (6x22 x 2 2 x 4 cm). The rats tails were lifted so their hind paws were in the air and the lubricated probe inserted 6 cm into the rectum [11]. The animals were then let down and gently held in the trough for the approximately 45 seconds necessary to get a stable temperature measurement. Animals were weighed and returned to their home cage. The total time elapsed from cage opening to cage closing was a little over a minute. The probe was then cleaned and the next animal's cage opened. All animals were left unhandled in their cage for thirty days to habituate to their housing and lighting conditions. The animals were then randomly divided into three groups of 11 rats. The first group of animals (Light group) had their temperatures measured and were weighed daily at 9:00+0:30 hr, during the light part of the cycle. Animals in the second group (Dark group) had their temperatures measured and were weighed daily at 16:30---0:30 hr, during the dark part of the cycle. Animals in the last group (Naive group) remained in their home cages. Each day the order the animals were handled was varied. It took approximately 15 minutes to measure the body temperature and weight of 11 rats. After 14 days of this procedure, on Day 15, all animals had their temperatures measured three times at 45 minute intervals starting at 9:00 hr. Three measurements were taken to determine the consequences for body temperature of the first measurement. On this day, animals were not weighed, permitting a measurement of all 33 animals' temperatures within 30 minutes. One animal from each group was tested before the next animal in the group was handled. After the three temperature measurements, animals were left unhandied until the next day. From Day 16 to Day 28 animals were treated in their usual manner, i.e., one temperature measurement a day, except for a two day break when no animals were handled (Days 17 and 18). On Day 29, animals in the Light group were handled as usual at 9:30 hr, but at 16:30 hr in the dark, all 33 animals
had their temperatures measured three times ~f 45 minute intervals. RESUI.TS
Mean body temperature of Light group animals measured in the light part of the cycle was always lower franging from 37,3 to 37.9°C) than that of Dark group animals measured in the dark part of the cycle (37.8 to 38.7°C1. On the first day, correlations between order of handling and body temperature were not significant (r=.20 for the Light group and r=.40 for Dark group). On subsequent days the correlation rose, becoming and remaining significant (r>.60) by Day 2 for rats in the Light group and by Day 4 for rats in the Dark group. On test days (Day 15 and Day 29), the correlations for Group Naive were not significant (r=. 13 and r=.30, respectively). The body temperature of animals in each of the three groups starting at 9:00 hr on Day 15, during the light part of the cycle, and at 16:30 hr on Day 29, during the dark part of the cycle, are shown in Fig. 1. Analysis of variance (ANOVA) of the first body temperature measurement revealed no significant differences on Day 15 ( F < I ) but did find significant group differences on Day 29, F(2,30)=7.75, p<0.01. On Day 29, Group Dark rats which usually had their temperature measured at this time had a body temperature that was significantly higher than animals not usually handled at this time (Scheffe p<0.05). Temperatures of rats in Group Light and Group Naive did not differ significantly. Figure 1 also shows the second and third temperature measurement taken on each of the test days. On Day 15, while there were no significant differences in the first temperature measurement of the three groups, there was a significant difference in subsequent temperature readings. Animals never handled were warmer than animals in the other two groups for the second and third temperature measurement. A Group x Time A N O V A for the second and third measurement revealed only a significant Group effect, F(2,30)=3.95, p<0.05. A similar Group × Time A N O V A for the second and third temperature measurements was done for Day 29 and revealed a similar significant Group effect, F(2,30)=4.55, p<0.05. In this case, however, the initial temperature difference between groups makes interpretation of this subsequent difference difficult. Though the increase in temperature from the first measurement was largest in the Naive group animals, the absolute temperatures were highest in Dark group animals. DISCUSSION
It was predicted that a temperature difference would be evident between animals exposed to stimuli predictive of handling and animals for whom handling had not been conditioned to these stimuli. This difference was found on Day 29 in the dark part of the cycle: animals normally handled at this time were warmer than animals not usually handled at this time. The difference was not seen on the first test day, Day 15, when animals were tested during the light part of the cycle. This difference in results could be due either to an insufficient number of training or conditioning trials by Day 15 or to the fact that the effect is less evident during the day. The next experiment will investigate these two possibilities. The contrast between the lack of a group difference on Day 15 and the rapid development of the correlation between handling order and temperature deserves some comment, but may not be all that surprising. F o r development of the
L E A R N E D H Y P E R T H E R M I A I N D U C E D BY H A N D L I N G
~.5
D
39.C --
DAY 15
DAY 29
9:00h (light)
16: 30h (dark)
METHOD
Subjects Male Wistar rats (n=50) were obtained and housed in a manner similar to that used in Experiment 1. Lights went on and off at 4:00 hr and 16:00 hr, respectively.
/4
o Naive = Light : Dark
0
58.5
58.0--
Procedure
37.5--
37.~
651
I
I
I
I
I
I
0
45
90
0
45
90
T/ME (minutes) FIG. 1. Mean (-+SEM) body temperatures taken at 45 minute intervals of rats in the three groups tested on Day 15 during the light part of the cycle and on Day 29 during the dark part of the cycle.
Animals were weighed twice in the first 21 days after arrival, to gentle them slightly, before the start of the experiment. On Day 1, half of the animals (the Light group) had their temperatures measured and were weighed at 8:45 hr, a procedure taking approximately 35 minutes. Forty-five and ninety minutes later these animals had their temperatures measured a second and third time. At 18:30 hr, the remaining animals (the Dark group) were handled in a similar manner, 3 temperature measurements at 45 minute intervals. This procedure was followed for the subsequent seven days with starting times not varying by more than an hour. On the eighth day, animals were not handled. The next day (Day 9), the two groups were divided and half the animals in each group had their temperatures measured three times at 45 minute intervals starting at 8:00 hr, during the light part of the cycle (groups Light-Light and Dark-Light). The remaining animals from each group were tested in a similar manner starting at 18:00 hr, in the dark part of the cycle (groups Light-Dark and Dark-Dark). RESULTS
correlation between handling and body temperature there may have been any number of very salient cues (presence of the experimenter or signals from the first rats) which were predictive of handling half the time. By way of contrast these salient stimuli had to be treated differentially depending on lighting conditions for the difference between the groups to be evident in this experiment. It is therefore not surprising that this discrimination was harder to learn for these animals. As might be expected, on Day 15, naive rats responded to the initial handling by showing a dramatic rise in body temperature. (Similarly on Day 29 the largest rise in temperature was in Naive group rats.) Animals that had been handled previously, whether they were handled in the light or dark phase of the cycle, did not show as large a hyperthermia after the initial temperature measurement. This replicates previous findings that the rise in temperature seen after handling disappears over days, possibly due to habituation [3, 7, 14]. Note, however, that while on Day 15 animals previously handled showed no rise in temperature, on Day 29, after more handling, the same animals did show a hyperthermic response to handling. This suggests that there may be circadian influence operating. EXPERIMENT 2 In this experiment, rats were given extensive handling in either the light or dark part of the cycle before testing. In order to potentiate the salience of the lighting cue the animals were handled three times a day rather than once a day as in the first experiment. On the test day, half the animals were handled during the part of the cycle in which they were normally handled, while the other half were handled at the other time, producing a two by two design.
Two rats lost a few grams from Day 7 to Day 9 suggesting they were not completely healthy and were therefore excluded from the experiment. This resulted in 12 animals in each of the four groups. Figure 2 shows the temperatures of the four groups on Day 9. A Handling × Cycle × Time A N O V A revealed significant Handling, F(1,44)=12.62, p<0.01, Cycle, F(1,44)=60.75, p<0.001, and Time, F(2,88)=4.21, p<0.02, main effects. The Handling effect reflects the fact that temperatures were warmer for animals handled at the normal time, groups Light-Light and DarkDark, than for animals normally having their temperatures measured at the other time, groups Light-Dark and DarkLight. It is clear from Fig. 2 that temperatures were higher at night, that is in the dark, than during the light part of the cycle; the Cycle effect. The Time effect reveals that overall the temperatures were still being affected by the repeated handling, though in absolute terms the differences were not large, 37.95, 38.15, and 38.02°C for the first, second and third measurements, respectively. None of the interactions were significant. From Fig. 2, however, it appears that animals handled at the normal time do not show the marked increase in temperature from the first to second temperature measurement shown by animals normally handled at the other time. To test this a similar three-way A N O V A was carried out analyzing only the first and second temperature measurements. Again, all the main effects were significant but in this analysis the Handling x Time interaction was also significant, F(1,44)=4.23, p<0.05. No other interactions were significant. Simple main effects revealed that the temperature rose significantly from the first to the second measurement for the animals in groups not normally handled at this time, groups Light-Dark and Dark-Light, F(1,44)=11.11, p<0.005, but did not rise significantly for the animals in groups being handled at the normal time, groups Light-Light and Dark-Dark ( F < 1).
652
t+~! K t-,1, BOO M
Light 39.0
_
Dark I8:OOh
8:00h
~
Dork- Dark
38.5 i g h t - Dark
Lioht-Lioht 38.0
.14
Dark - Light
37.5
37.0
'"
I 0
I
I
I
4590 TIME
0
45
I 90
(minutes)
FIG. 2. Mean (_SEM) body temperatures of rats in the four groups
taken at 45 minute intervals either at 9:00 hr during the light part of the cycle or at 18:00 hr during the dark part of the cycle.
DISCUSSION
In the first experiment the conditioned handling effect was not evident during the light on Day 15. In this experiment both during the light and dark parts of the cycle animals handled at the normal time were warmer than animals handled at an unusual time. This supports the suggestion that in the first experiment there was insufficient conditioning before the first test, on Day 15. Further support that the hyperthermic effect of handlings can be learned is shown in the finding that only animals handled at an unusual time showed an increase in temperature from the first to the second temperature measurement. Animals handled at the normal time did not show this increase, suggesting it may have occurred prior to the first temperature measurement. GENERAL DISCUSSION Temperature increases can be conditioned to stimuli occurring in anticipation of events such as forced exercise [ 10], microwave radiation induced hyperthermia [2], or drug injections and rotarod measurement (a motor performance test) [17]. These studies subjected the rats to manipulations which in themselves had profound effects on body temperature. Similar conditioned temperature changes have been seen using temperature changing drugs (see [4,8]). In the present experiments similar sized conditioned changes in body temperature could be seen in animals who were simply removed from their home cages, had their temperature measured, were weighed, and then returned to their cages.
Animals handled at their usual time were hyperthermiu relative to animals not normally handled at that time. Temperature of animals measured at a novel time was uloser to that reported previously for that of unstressed animaN [10.12]. Thc procedures used here are unaviodable in animal experiments and are generally not considered particularly stressful. The results of one of the preliminary experiments, where re+ peated weighing alone resulted in a significant temperatureorder correlation, suggests it is the handling itself which is the critical variable and that rectal temperature measurement is no more stressful then weighing. A comment should be made here about the behavior of these animals. Experienced animal workers "'gentle" their animals by handling them several times prior to experiments, When rats are first handled they seem very upset and aroused by the procedure, but over a few trials they become more tranquil. Similar changes in the animals' behavior were evident here. It was much easier to measure the temperature of the animal the second time than the first. This initial reaction of naive animals to any handling often includes struggling and vocalization which could be arousing to other animals. Thus. both the initial days and any test clays which include handling of naive rats are characterized by increased amounts of vocalization. This could result in an increased general arousal, a factor which might reduce any group differences in these experiments, and may be partially responsible for the lack of group difference on Day 15 in Experiment 1, a test which involved temperature measurements of naive animals. A second concern is the nature of the relation between the conditioned hyperthermia and the gentling procedure. From these experiments it can be seen that the two phenomena are separable. In both experiments there are groups which have equivalent handling experience, and seemed undisturbed by handling, yet the animals having their temperatures measured at the normal time are hyperthermic relative to animals not usually handled at that time. The handling-induced increase in body temperature is known to decrease over days or weeks [3, 7, 16]. This could be due either to a decrease in the second temperature measurement or to an increase in the initial temperature. Data from the present experiments support both possibilities. On Day 15 of the first experiment, animals in the Naive group showed an elevated second temperature measurement relative to both their initial temperature and that of normally handled rats, which exhibited no change in temperature over the three measurements. This suggests that for the animals normally being handled the temperature increasing effect of the initial temperature measurement had habituated; and, as initial temperatures were not significantly different in these three groups, it reflected a reduction in the second temperature measurement. By way of contrast, in the second experiment, animals handled at an unusual time show an increase in temperature from the first to second measurement while animals normally handled do not (see Fig. 2). This rise. however, only reduces the initial difference between the two groups, suggesting that in animals normally handled the hyperthermia has occurred earlier. Since the habituation of the temperature rise can occur prior to the development of the conditioned hyperthermia (Day 15, Experiment l) it argues that these may be independent processes. Recent work suggests that the relation between non-noxious stressors and body temperature is a complex one and may produce either a rise or a fall in temperature depending on the particular procedure involved [161. Further work looking
L E A R N E D H Y P E R T H E R M I A I N D U C E D BY H A N D L I N G at the relation between the conditioned hyperthermic effect and effects of repeated handling is necessary. These effects of repeated handling suggest that studies of conditioned temperature changes using drugs should not only report temperature difference measurements. The use of difference scores may obscure changes in initial temperature misrepresenting any learned drug effects. As the conditioned effect evident in the present experiments was a hyperthermia, studies reporting a conditioned increase in body temperature (most drug studies, see [4,8]) should be examined very carefully to see if such conditioning could not be due to the procedure used rather than the drugs administered. Finally, the question of the nature of this conditioned hyperthermia must be addressed. It is clear from the fact that
653 the effect is evident during both phases of the light-dark cycle that it is not simply due to waking of a sleeping animal: if anything, the effect seems strongest at night when the rat is most active. In work looking at emotional arousal and temperature regulation in rabbits, Franzini, Lenzi and Cianci [9] suggest the rise in temperature due to arousal is not a coherent thermoregulatory response, that is, an elevation in a set point, but is due to changes in effectors for respiration and vasomotion which in turn result in temperature changes. The fact that minor differences in procedure in rat studies can produce profound differences in body temperature response [16] suggests that these temperature changes may also be secondary to other physiological responses. It might prove profitable therefore to do similar experiments with other physiological measures.
REFERENCES 1. Benedek, G., M. Szikszay and F. Obal. Stress-related changes of opiate sensitivity in thermoregulation. Life Sci 33: Suppl 1, 591-593, 1983. 2. Bermant, R. I., D. L. Reeves, D. M. Levinson and D. R. Justesen. Classical conditioning of microwave-induced hyperthermia in rats. Radio Sci 14: 201-207, 1979. 3. Briese, E. and M. G. Quijada. Colonic temperature of rats during handling. Acta Physiol Latinoam 20: 97-102, 1970. 4. Cunningham, C. L., J. C. Crabbe and H. Rigter. Paviovian conditioning of drug-induced changes in body temperature. Pharrnacol Ther 23: 365-391, 1984. 5. Delini-Stula, A. The development and the extinction of hyperthermia induced by conditioned avoidance behavior in rats. J Exp Anal Behav 14: 213-218, 1970. 6. Eikelboom, R. and J. Stewart. Conditioned temperature effects using morphine as the unconditioned stimulus. Psychopharrnacology (Berlin) 61: 31-38, 1979. 7. Eikelboom, R. and J. Stewart. Hypophysectomy increases the sensitivity of rats to naloxone-inducedhypothermia. Life Sci 28: 1047-1052, 1981. 8. Eikelboom, R. and J. Stewart. The conditioningof drug-induced physiological responses. Psychol Rev 89: 507-528, 1982.
9. Franzini, C., P. Lenzi and T. Cianci. Interactions between temperature regulation and emotional arousal in the rabbit. Exp Brain Res 43: 87-92, 1981. 10. Gollnick, P. D. and C. D. Ianuzzo. Colonic temperature response of rats during exercise. J Appl Physio124: 747-750, !968. 11. Lomax, P. Measurement of "core" temperature in the rat. Nature 210: 854-855, 1966. 12. Poole, S. and J. D. Stephenson. Core temperature: Some shortcomings of rectal temperature measurements. Physiol Behav 18: 203-205, 1977. 13. Schwen, R. J. and L. C. Jones. Non-stressful measurement of morphine and capsaicinoid-induced hypothermia in the rat. Pharmacologist 26: 140, 1984. (Abstract) 14. Stewart, J. and R. Eikelboom. Stress masks the hypothermic effect of naloxone in rats. Life Sci 25:1165-1172, 1979. 15. Stewart, J. and R. Eikelboom. Interaction between the effects of stress and morphine on body temperature in rats. LifSci 28: 1041-1045, 1981. 16. Vidal, C., C. Suadeau and J. Jacob. Regulation of body temperature and nociception induced by non-noxious stress in rat. Brain Res 297: 1-10, 1984. 17. York, J. L. and S. G. Regan. Conditioned and unconditioned influences on body temperature and ethanol hypothermia in laboratory rats. Pharrnacol Biochem Behav 17:11%124, 1982.
0031-9384/86$3.00 + .00
Learned Anticipatory Rise in Body Temperature Due to Handling ROELOF EIKELBOOM
D e p a r t m e n t o f Psychology, Queen's University, Kingston, Ontario, Canada K7L 3N6 R e c e i v e d l l O c t o b e r 1985 EIKELBOOM, R. Learned anticipatory rise in body temperature due to handling. PHYSIOL BEHAV 37(4) 649--653, 1986.--Hyperthermia produced by handling becomes evident at the initial daily measurement if temperature is measured at a consistent time. This hyperthermia may be a learned effect occurring in anticipation of handling. In Experiment One male Wistar rats were either unhandled or had their temperatures measured daily in the dark or the light part of the day. All animals had their temperatures measured on Day 29, in the dark. Rats usually tested in the dark were hyperthermic, 38.8°C, relative to rats previously handled only in the light, 38. I°C, and to naive rats, 37.9°C. In Experiment Two rats were handled three times daily in either the light or the dark. On Day 9 each group was divided in two, and temperatures were measured at either the usual time or at the other time. Rats tested at their usual time were hyperthermic, relative to rats normally handled in the other part of the cycle. This suggests a conditioned hyperthermia occurs in response to stimuli predictive of handling. Body temperature Conditioning Rectal temperature measurements
Handling stress Thermoregulation
Arousal
SIMPLY measuring a rat's rectal temperature elevates body temperature about 1.0°C for over an hour [3, 7, 12, 14]. This elevation has been attributed to the arousing effects of handling the rats [3,12]. The fact that the effects of drugs on body temperature may be masked, changed, or even prevented when they occur in conjunction with this handlinginduced rise in temperature suggests this is not an innocuous procedure [1, 12-15]. Ideally the solution to this problem would be to not handle the rats and use some form of telemetric recording of body temperature. The relative ease of rectal measurement, however, suggests that many studies will still use this method. Further, in most studies animals still need to be handled for injections or weighing and so would still undergo the handling-induced temperature increase. The present experiments report that the handlinginduced hyperthermia may be conditionable to general stimuli, i.e., time of day, presence of the experimenter, which are predictive of handling. In previous work [6,7] we noticed that the baseline home cage temperature of the first few animals handled was lower (37.2°C) than that of subsequent rats (over 38°C for the last animals handled). This difference between rats is similar in magnitude to the handling effect but cannot be due to the handling of the animals as it takes several minutes for a rat's body temperature to increase this amount [12]. Preliminary studies found that there was a correlation between the order in which the animals were handled and their body temperature. While group mean temperature of rats in several replications ranged between 37.7 and 37.9°C, the regression lines revealed a Y intercept of 37.1 to 37.3°C, the body temperature of unstressed animals [10,12]. This correlation was
Pavlovian conditioning
Hyperthermia
not evident initially and developed over the first few days of handling and body temperature measurement; it was evident even if animals had previously only been weighed. After several days of daily temperature measurement the correlations (Pearson r) between the order the rats were handled and their body temperature ranged, in the various replications, from .68 to .77. The fact that this correlation developed over days suggests it was not simply due to signals emanating from the first rats producing changes in the later tested rats. Since these studies involved daily temperature measurement the difference between the temperature of the first and last few rats may have been a learned effect occurring in anticipation of the actual temperature measurement. The arrival of the experimenter and the handling of the first few rats could act as a conditioned stimulus resulting in a conditioned temperature increase that mimicks the increase induced by the rectal probing. Conditioned temperature changes have been demonstrated using drugs that affect body temperature [4, 6, 8] and various stressful events like microwave-induced hyperthermia [2], electric shocks [5], and exercise [10]. If simply handling the rats can produce a conditioned temperature increase it suggests that temperature changes can be conditioned with relatively benign unconditioned stimuli. This conditioned effect might in turn be a confound in many of the other temperature conditioning studies. EXPERIMENT 1 If the correlation between order of handling and body temperature is due to an anticipatory conditioned increase in
~Preparation of this article was supported by a grant from the National Sciences and Engineering Research Council of Canada (A 2575) to R. Eikelboom who is a career scientist with the Ontario Ministry of Health. Portions of this work have been presented at the Canadian Psychological Association meeting, June, 1985.
649
650
klKELBOOM
temperature, then, by manipulating the stimuli predicting handling, it should be possible to affect body temperature. That is, the body temperature of animals measured in the presence of stimuli previously associated with handling should be higher than that of animals not normally handled in that situation. In this experiment animals were divided into three groups, two of which had distinctive temporal and lighting cues predicting temperature measurement and a third group that was not handled until the test days.
METHOD
Subjects Male Wistar rats (n=33) weighing between 225 and 250 g at the time of arrival were obtained from Charles River Canada, St. Constant, Quebec. Animals were individually housed in standard stainless steel cages with food and water available ad lib throughout these studies. The animal room was maintained at a constant temperature of 22_+ I°C with a 12/12 light/dark cycle (lights off 14:00 hr). One small light was constantly on to permit testing in the dark phase of the cycle.
Procedure Temperatures were measured using a Yellow Springs Telethermometer with a model 423 probe. Animals were removed from their cage, carried to a table in the animal room, and placed in a small three sided trough (6x22 x 2 2 x 4 cm). The rats tails were lifted so their hind paws were in the air and the lubricated probe inserted 6 cm into the rectum [11]. The animals were then let down and gently held in the trough for the approximately 45 seconds necessary to get a stable temperature measurement. Animals were weighed and returned to their home cage. The total time elapsed from cage opening to cage closing was a little over a minute. The probe was then cleaned and the next animal's cage opened. All animals were left unhandled in their cage for thirty days to habituate to their housing and lighting conditions. The animals were then randomly divided into three groups of 11 rats. The first group of animals (Light group) had their temperatures measured and were weighed daily at 9:00+0:30 hr, during the light part of the cycle. Animals in the second group (Dark group) had their temperatures measured and were weighed daily at 16:30---0:30 hr, during the dark part of the cycle. Animals in the last group (Naive group) remained in their home cages. Each day the order the animals were handled was varied. It took approximately 15 minutes to measure the body temperature and weight of 11 rats. After 14 days of this procedure, on Day 15, all animals had their temperatures measured three times at 45 minute intervals starting at 9:00 hr. Three measurements were taken to determine the consequences for body temperature of the first measurement. On this day, animals were not weighed, permitting a measurement of all 33 animals' temperatures within 30 minutes. One animal from each group was tested before the next animal in the group was handled. After the three temperature measurements, animals were left unhandied until the next day. From Day 16 to Day 28 animals were treated in their usual manner, i.e., one temperature measurement a day, except for a two day break when no animals were handled (Days 17 and 18). On Day 29, animals in the Light group were handled as usual at 9:30 hr, but at 16:30 hr in the dark, all 33 animals
had their temperatures measured three times ~f 45 minute intervals. RESUI.TS
Mean body temperature of Light group animals measured in the light part of the cycle was always lower franging from 37,3 to 37.9°C) than that of Dark group animals measured in the dark part of the cycle (37.8 to 38.7°C1. On the first day, correlations between order of handling and body temperature were not significant (r=.20 for the Light group and r=.40 for Dark group). On subsequent days the correlation rose, becoming and remaining significant (r>.60) by Day 2 for rats in the Light group and by Day 4 for rats in the Dark group. On test days (Day 15 and Day 29), the correlations for Group Naive were not significant (r=. 13 and r=.30, respectively). The body temperature of animals in each of the three groups starting at 9:00 hr on Day 15, during the light part of the cycle, and at 16:30 hr on Day 29, during the dark part of the cycle, are shown in Fig. 1. Analysis of variance (ANOVA) of the first body temperature measurement revealed no significant differences on Day 15 ( F < I ) but did find significant group differences on Day 29, F(2,30)=7.75, p<0.01. On Day 29, Group Dark rats which usually had their temperature measured at this time had a body temperature that was significantly higher than animals not usually handled at this time (Scheffe p<0.05). Temperatures of rats in Group Light and Group Naive did not differ significantly. Figure 1 also shows the second and third temperature measurement taken on each of the test days. On Day 15, while there were no significant differences in the first temperature measurement of the three groups, there was a significant difference in subsequent temperature readings. Animals never handled were warmer than animals in the other two groups for the second and third temperature measurement. A Group x Time A N O V A for the second and third measurement revealed only a significant Group effect, F(2,30)=3.95, p<0.05. A similar Group × Time A N O V A for the second and third temperature measurements was done for Day 29 and revealed a similar significant Group effect, F(2,30)=4.55, p<0.05. In this case, however, the initial temperature difference between groups makes interpretation of this subsequent difference difficult. Though the increase in temperature from the first measurement was largest in the Naive group animals, the absolute temperatures were highest in Dark group animals. DISCUSSION
It was predicted that a temperature difference would be evident between animals exposed to stimuli predictive of handling and animals for whom handling had not been conditioned to these stimuli. This difference was found on Day 29 in the dark part of the cycle: animals normally handled at this time were warmer than animals not usually handled at this time. The difference was not seen on the first test day, Day 15, when animals were tested during the light part of the cycle. This difference in results could be due either to an insufficient number of training or conditioning trials by Day 15 or to the fact that the effect is less evident during the day. The next experiment will investigate these two possibilities. The contrast between the lack of a group difference on Day 15 and the rapid development of the correlation between handling order and temperature deserves some comment, but may not be all that surprising. F o r development of the
L E A R N E D H Y P E R T H E R M I A I N D U C E D BY H A N D L I N G
~.5
D
39.C --
DAY 15
DAY 29
9:00h (light)
16: 30h (dark)
METHOD
Subjects Male Wistar rats (n=50) were obtained and housed in a manner similar to that used in Experiment 1. Lights went on and off at 4:00 hr and 16:00 hr, respectively.
/4
o Naive = Light : Dark
0
58.5
58.0--
Procedure
37.5--
37.~
651
I
I
I
I
I
I
0
45
90
0
45
90
T/ME (minutes) FIG. 1. Mean (-+SEM) body temperatures taken at 45 minute intervals of rats in the three groups tested on Day 15 during the light part of the cycle and on Day 29 during the dark part of the cycle.
Animals were weighed twice in the first 21 days after arrival, to gentle them slightly, before the start of the experiment. On Day 1, half of the animals (the Light group) had their temperatures measured and were weighed at 8:45 hr, a procedure taking approximately 35 minutes. Forty-five and ninety minutes later these animals had their temperatures measured a second and third time. At 18:30 hr, the remaining animals (the Dark group) were handled in a similar manner, 3 temperature measurements at 45 minute intervals. This procedure was followed for the subsequent seven days with starting times not varying by more than an hour. On the eighth day, animals were not handled. The next day (Day 9), the two groups were divided and half the animals in each group had their temperatures measured three times at 45 minute intervals starting at 8:00 hr, during the light part of the cycle (groups Light-Light and Dark-Light). The remaining animals from each group were tested in a similar manner starting at 18:00 hr, in the dark part of the cycle (groups Light-Dark and Dark-Dark). RESULTS
correlation between handling and body temperature there may have been any number of very salient cues (presence of the experimenter or signals from the first rats) which were predictive of handling half the time. By way of contrast these salient stimuli had to be treated differentially depending on lighting conditions for the difference between the groups to be evident in this experiment. It is therefore not surprising that this discrimination was harder to learn for these animals. As might be expected, on Day 15, naive rats responded to the initial handling by showing a dramatic rise in body temperature. (Similarly on Day 29 the largest rise in temperature was in Naive group rats.) Animals that had been handled previously, whether they were handled in the light or dark phase of the cycle, did not show as large a hyperthermia after the initial temperature measurement. This replicates previous findings that the rise in temperature seen after handling disappears over days, possibly due to habituation [3, 7, 14]. Note, however, that while on Day 15 animals previously handled showed no rise in temperature, on Day 29, after more handling, the same animals did show a hyperthermic response to handling. This suggests that there may be circadian influence operating. EXPERIMENT 2 In this experiment, rats were given extensive handling in either the light or dark part of the cycle before testing. In order to potentiate the salience of the lighting cue the animals were handled three times a day rather than once a day as in the first experiment. On the test day, half the animals were handled during the part of the cycle in which they were normally handled, while the other half were handled at the other time, producing a two by two design.
Two rats lost a few grams from Day 7 to Day 9 suggesting they were not completely healthy and were therefore excluded from the experiment. This resulted in 12 animals in each of the four groups. Figure 2 shows the temperatures of the four groups on Day 9. A Handling × Cycle × Time A N O V A revealed significant Handling, F(1,44)=12.62, p<0.01, Cycle, F(1,44)=60.75, p<0.001, and Time, F(2,88)=4.21, p<0.02, main effects. The Handling effect reflects the fact that temperatures were warmer for animals handled at the normal time, groups Light-Light and DarkDark, than for animals normally having their temperatures measured at the other time, groups Light-Dark and DarkLight. It is clear from Fig. 2 that temperatures were higher at night, that is in the dark, than during the light part of the cycle; the Cycle effect. The Time effect reveals that overall the temperatures were still being affected by the repeated handling, though in absolute terms the differences were not large, 37.95, 38.15, and 38.02°C for the first, second and third measurements, respectively. None of the interactions were significant. From Fig. 2, however, it appears that animals handled at the normal time do not show the marked increase in temperature from the first to second temperature measurement shown by animals normally handled at the other time. To test this a similar three-way A N O V A was carried out analyzing only the first and second temperature measurements. Again, all the main effects were significant but in this analysis the Handling x Time interaction was also significant, F(1,44)=4.23, p<0.05. No other interactions were significant. Simple main effects revealed that the temperature rose significantly from the first to the second measurement for the animals in groups not normally handled at this time, groups Light-Dark and Dark-Light, F(1,44)=11.11, p<0.005, but did not rise significantly for the animals in groups being handled at the normal time, groups Light-Light and Dark-Dark ( F < 1).
652
t+~! K t-,1, BOO M
Light 39.0
_
Dark I8:OOh
8:00h
~
Dork- Dark
38.5 i g h t - Dark
Lioht-Lioht 38.0
.14
Dark - Light
37.5
37.0
'"
I 0
I
I
I
4590 TIME
0
45
I 90
(minutes)
FIG. 2. Mean (_SEM) body temperatures of rats in the four groups
taken at 45 minute intervals either at 9:00 hr during the light part of the cycle or at 18:00 hr during the dark part of the cycle.
DISCUSSION
In the first experiment the conditioned handling effect was not evident during the light on Day 15. In this experiment both during the light and dark parts of the cycle animals handled at the normal time were warmer than animals handled at an unusual time. This supports the suggestion that in the first experiment there was insufficient conditioning before the first test, on Day 15. Further support that the hyperthermic effect of handlings can be learned is shown in the finding that only animals handled at an unusual time showed an increase in temperature from the first to the second temperature measurement. Animals handled at the normal time did not show this increase, suggesting it may have occurred prior to the first temperature measurement. GENERAL DISCUSSION Temperature increases can be conditioned to stimuli occurring in anticipation of events such as forced exercise [ 10], microwave radiation induced hyperthermia [2], or drug injections and rotarod measurement (a motor performance test) [17]. These studies subjected the rats to manipulations which in themselves had profound effects on body temperature. Similar conditioned temperature changes have been seen using temperature changing drugs (see [4,8]). In the present experiments similar sized conditioned changes in body temperature could be seen in animals who were simply removed from their home cages, had their temperature measured, were weighed, and then returned to their cages.
Animals handled at their usual time were hyperthermiu relative to animals not normally handled at that time. Temperature of animals measured at a novel time was uloser to that reported previously for that of unstressed animaN [10.12]. Thc procedures used here are unaviodable in animal experiments and are generally not considered particularly stressful. The results of one of the preliminary experiments, where re+ peated weighing alone resulted in a significant temperatureorder correlation, suggests it is the handling itself which is the critical variable and that rectal temperature measurement is no more stressful then weighing. A comment should be made here about the behavior of these animals. Experienced animal workers "'gentle" their animals by handling them several times prior to experiments, When rats are first handled they seem very upset and aroused by the procedure, but over a few trials they become more tranquil. Similar changes in the animals' behavior were evident here. It was much easier to measure the temperature of the animal the second time than the first. This initial reaction of naive animals to any handling often includes struggling and vocalization which could be arousing to other animals. Thus. both the initial days and any test clays which include handling of naive rats are characterized by increased amounts of vocalization. This could result in an increased general arousal, a factor which might reduce any group differences in these experiments, and may be partially responsible for the lack of group difference on Day 15 in Experiment 1, a test which involved temperature measurements of naive animals. A second concern is the nature of the relation between the conditioned hyperthermia and the gentling procedure. From these experiments it can be seen that the two phenomena are separable. In both experiments there are groups which have equivalent handling experience, and seemed undisturbed by handling, yet the animals having their temperatures measured at the normal time are hyperthermic relative to animals not usually handled at that time. The handling-induced increase in body temperature is known to decrease over days or weeks [3, 7, 16]. This could be due either to a decrease in the second temperature measurement or to an increase in the initial temperature. Data from the present experiments support both possibilities. On Day 15 of the first experiment, animals in the Naive group showed an elevated second temperature measurement relative to both their initial temperature and that of normally handled rats, which exhibited no change in temperature over the three measurements. This suggests that for the animals normally being handled the temperature increasing effect of the initial temperature measurement had habituated; and, as initial temperatures were not significantly different in these three groups, it reflected a reduction in the second temperature measurement. By way of contrast, in the second experiment, animals handled at an unusual time show an increase in temperature from the first to second measurement while animals normally handled do not (see Fig. 2). This rise. however, only reduces the initial difference between the two groups, suggesting that in animals normally handled the hyperthermia has occurred earlier. Since the habituation of the temperature rise can occur prior to the development of the conditioned hyperthermia (Day 15, Experiment l) it argues that these may be independent processes. Recent work suggests that the relation between non-noxious stressors and body temperature is a complex one and may produce either a rise or a fall in temperature depending on the particular procedure involved [161. Further work looking
L E A R N E D H Y P E R T H E R M I A I N D U C E D BY H A N D L I N G at the relation between the conditioned hyperthermic effect and effects of repeated handling is necessary. These effects of repeated handling suggest that studies of conditioned temperature changes using drugs should not only report temperature difference measurements. The use of difference scores may obscure changes in initial temperature misrepresenting any learned drug effects. As the conditioned effect evident in the present experiments was a hyperthermia, studies reporting a conditioned increase in body temperature (most drug studies, see [4,8]) should be examined very carefully to see if such conditioning could not be due to the procedure used rather than the drugs administered. Finally, the question of the nature of this conditioned hyperthermia must be addressed. It is clear from the fact that
653 the effect is evident during both phases of the light-dark cycle that it is not simply due to waking of a sleeping animal: if anything, the effect seems strongest at night when the rat is most active. In work looking at emotional arousal and temperature regulation in rabbits, Franzini, Lenzi and Cianci [9] suggest the rise in temperature due to arousal is not a coherent thermoregulatory response, that is, an elevation in a set point, but is due to changes in effectors for respiration and vasomotion which in turn result in temperature changes. The fact that minor differences in procedure in rat studies can produce profound differences in body temperature response [16] suggests that these temperature changes may also be secondary to other physiological responses. It might prove profitable therefore to do similar experiments with other physiological measures.
REFERENCES 1. Benedek, G., M. Szikszay and F. Obal. Stress-related changes of opiate sensitivity in thermoregulation. Life Sci 33: Suppl 1, 591-593, 1983. 2. Bermant, R. I., D. L. Reeves, D. M. Levinson and D. R. Justesen. Classical conditioning of microwave-induced hyperthermia in rats. Radio Sci 14: 201-207, 1979. 3. Briese, E. and M. G. Quijada. Colonic temperature of rats during handling. Acta Physiol Latinoam 20: 97-102, 1970. 4. Cunningham, C. L., J. C. Crabbe and H. Rigter. Paviovian conditioning of drug-induced changes in body temperature. Pharrnacol Ther 23: 365-391, 1984. 5. Delini-Stula, A. The development and the extinction of hyperthermia induced by conditioned avoidance behavior in rats. J Exp Anal Behav 14: 213-218, 1970. 6. Eikelboom, R. and J. Stewart. Conditioned temperature effects using morphine as the unconditioned stimulus. Psychopharrnacology (Berlin) 61: 31-38, 1979. 7. Eikelboom, R. and J. Stewart. Hypophysectomy increases the sensitivity of rats to naloxone-inducedhypothermia. Life Sci 28: 1047-1052, 1981. 8. Eikelboom, R. and J. Stewart. The conditioningof drug-induced physiological responses. Psychol Rev 89: 507-528, 1982.
9. Franzini, C., P. Lenzi and T. Cianci. Interactions between temperature regulation and emotional arousal in the rabbit. Exp Brain Res 43: 87-92, 1981. 10. Gollnick, P. D. and C. D. Ianuzzo. Colonic temperature response of rats during exercise. J Appl Physio124: 747-750, !968. 11. Lomax, P. Measurement of "core" temperature in the rat. Nature 210: 854-855, 1966. 12. Poole, S. and J. D. Stephenson. Core temperature: Some shortcomings of rectal temperature measurements. Physiol Behav 18: 203-205, 1977. 13. Schwen, R. J. and L. C. Jones. Non-stressful measurement of morphine and capsaicinoid-induced hypothermia in the rat. Pharmacologist 26: 140, 1984. (Abstract) 14. Stewart, J. and R. Eikelboom. Stress masks the hypothermic effect of naloxone in rats. Life Sci 25:1165-1172, 1979. 15. Stewart, J. and R. Eikelboom. Interaction between the effects of stress and morphine on body temperature in rats. LifSci 28: 1041-1045, 1981. 16. Vidal, C., C. Suadeau and J. Jacob. Regulation of body temperature and nociception induced by non-noxious stress in rat. Brain Res 297: 1-10, 1984. 17. York, J. L. and S. G. Regan. Conditioned and unconditioned influences on body temperature and ethanol hypothermia in laboratory rats. Pharrnacol Biochem Behav 17:11%124, 1982.