Analysis of the role of drug-predictive environmental stimuli in tolerance to the hypothermic effects of the benzodiazepine midazolam

Psychopharmacology (1986) 90:513-521 Psychopharmacology © Spfinger-Verlag 1986

Analysis of the role of drug-predictive environmental stimuli in tolerance to the hypothermic effects of the benzodiazepine midazolam* J.W. Griffiths and A.J. Goudie Department of Psychology, University of Liverpool, P.O. Box 147, Liverpool L69 3BX, UK

Abstract. The role of classical conditioning processes in the development of tolerance to the hypothermic effects of the short-acting benzodiazepine midazolam was studied in three experiments in rats. The experiments were all designed so that one set of environmental stimuli reliably predicted drug treatments whilst another set of stimuli predicted control (vehicle) treatment. According to the classical conditioning account of tolerance, the degree of tolerance observed should be greater in the presence of drug-predictive stimuli than in their absence, i.e. tolerance should show environmental (context) specificity. A preliminary study was conducted to determine the dose- and time-effect curves for midazolam-induced hypothermia. The results of this study provided essential background data for the design of all the subsequent tolerance studies. In the first tolerance study, it was found that virtually complete tolerance developed to the hypothermic effects of 4 mg/kg (IP) midazolam given on alternate days. However, the observed tolerance was clearly not environmentally specific. Since there is evidence that conditioned tolerance to some drug effects develops most readily if drugs are given at low doses with long inter-injection intervals, a second study was conducted with a lower (1.6 mg/kg IP) dose of midazolam, which was given every 5th day. Despite these procedural changes, the second study indicated that the observed tolerance again did not show context specificity, even though tolerance developed rapidly with the lower dose of a short acting drug which was given infrequently. In a final study, the experimental procedure was changed again so that the environmental stimuli which predicted drug treatment were only present during the o n s e t of drug-induced hypothermia, in contrast to the procedure adopted in the two previous studies in which the drug-predictive stimuli were present during the onset and the offset of the drug's hypothermic effect. This procedural change was introduced because it was considered possible that the presence of stimuli associated with recovery from the drug's effects might have prevented the development of conditioned tolerance in the first two studies. However, no evidence was obtained for context specific tolerance, even after this further procedural manipula- tion. These data indicate clearly that it is difficult to demon-

* An abstract report of some of the data presented in this paper has already been published (Goudie and Griffiths 1985).

Of fpr in t reques ts to : A.J. Goudie

strate context specificity of midazolam hypothermic tolerance. A number of possible reasons for these results are considered.

Key words: Benzodiazepines - Midazolam -- Tolerance - Classical conditioning - Rats - Body temperature

A number of experimental studies have provided evidence for the involvement of classical conditioning in tolerance to various actions of a range of drugs (see Siegel 1982; Siegel and MacRae 1984; Cunningham et al. 1984; Goudie and Demellweek 1986 for reviews). According to the classical conditioning account of tolerance, if any cues exist which reliably predict the presence of a drug, these cues (conditional stimuli or CSs) may be associated with the development of a conditioned compensatory response (CR) which opposes the acute drug e f fec t - the unconditioned response (UCR). Since the CR and UCR are opposite in direction, they are believed to summate algebraically. The net response is a progressively diminishing drug effect, i.e. the development of tolerance. Thus tolerance may show environmental (context) specificity/fa drug is administered in the presence of drug-predictive cues. In such circum- stances, tolerance will be absent in drug-experienced subjects when the drug is administered in the absence of the relevant cues.

The studies reported here were concerned with analysing context specific tolerance to the benzodiazepine (BZ) midazolam. When the experiments reported here were initiated, the only study reporting on the role of environmental cues in BZ tolerance was a study by File (1982) on tolerance to the sedative effect of chlordiazepoxide (CDP) in rats. File's (1982) findings provided only partial support for the idea that BZ tolerance could be context specific, since she found that long-term retention, but not the acquisition, of tolerance to the sedative effect of CDP was dependent upon drug administration in the presence of drug-predictive apparatus cues. No evidence was obtained which implicated conditioning in tolerance acquisition as opposed to retention (File 1982). In the present series of experiments we therefore investigated the development of tolerance to B Z - induced hypothermia. A number of studies in rodents with sedative-hypnotics (Le etal. 1979; Mansfield and Cun-

514

Table 1. Procedure for counterbalanced discrimination conditioning and tests for context-specific tolerance in experiment 2

Phase of experiment Treatment of four experimental groups

Midazolam home Midazolam distinct Control home Control distinct (n = 16) (n = 16) (n = 8) (n = 8)

1. Tolerance Odd days : Drug in Drug in Vehicle in Vehicle in acquisition home room distinct room home room distinct room (days 1-32) Even days : Vehicle in Vehicle in Vehicle in Vehicle in

distinct room home room distinct room home room

2. Midazolam Cued home Cued distinct tolerance test (n = 8) (n = 8) (day 34) Drug in Drug in

home room distinct room N.B. The two drug groups of phase 1 Uncued distinct Uncued home were split into (n = 8) (n = 8) two sub-groups each Drug in Drug in

distinct room home room

Control home Control distinct

Drug in Drug in home room distinct room

ningham 1980; Cappell et al. 1981; Crowell et al. 1981; Melchior and Tabakoff 1981; Hinson and Rhijnsburger 1984) have all provided evidence for context specific tolerance to drug-induced hypothermia.

The specific BZ used in these studies was midazolam, a short-acting agent which possesses most "typical" benzodiazepine actions (Pieri et al. 1981). In an initial experiment, the dose- and time-effect curves for midazolam's action on body temperature were determined. The aims of this experiment were: to validate our procedure for temperature assessment, to identify midazolam doses that could be used in subsequent tolerance studies and to determine the duration of action of midazolam to provide a rationale for the choice in tolerance studies of the temporal parameters relat- ing the drug's actions to the presence or absence of the specific environmental cues (CSs) used to signal the drug, since the temporal relationship between the CS and the UCS is generally thought to be of considerable significance in conditioning studies (cf. Sherman 1979). In an initial study on the role of environmental cues in the development of tolerance, we obtained no evidence that tolerance was context specific. However, studies with morphine have suggest that context specific tolerance may be obtained most reliably if low doses are given infrequently (Baker and Tif- fany 1985). We therefore initiated a second tolerance study, involving lower doses given less frequently, to increase the chance of demonstrating context specific tolerance. How- ever, no evidence was obtained for context specific tolerance in this study. We therefore conducted a third tolerance study in which we changed the temporal relationship between exposure to a distinctive drug-predictive environment (CS) and the drug effect, so that the CS predicted only the onset of drug-induced hypothermia and not the total drug effect (i.e. onset and offset), as had been the case in the two prior tolerance studies. Again, we failed to find evidence for context specific tolerance. Subsequent to com- pletion of the three tolerance studies reported here, Greeley and Cappell (1985) published some data suggesting that it may be possible to demonstrate context specific CDP tolerance. The relationship between Greeley and Cappell's (1985) data and our findings is considered in the discussion section of this paper.

Materials and methods

Experiment 1 : Determination of dose- and time-effect curves for midazolam-induced hypothermia. Female hooded rats (250-300 g, Olac, UK) were assigned to four groups (n = 6 per group) which received IP injections (2 ml/kg in distilled water) of 0, 2, 8 or 32 mg/kg midazolam maleate (Roche Products, Welwyn Garden City, UK) made up as the salt. Prior to, and at 7 times up to 300 min post-injection, core body temperatures were recorded with a thermocouple probe and a microprocessor thermometer (Comark Elec- tronics, UK). Rectal probes were inserted 5 cm into the colon whilst rats were lightly restrained, 15 s were allowed after probe insertion for temperature measurement to sta- bilise to the nearest 0.1 ° C. Subjects were individually caged with food and water freely available in a temperature (22.3_+1.3°C) and light (0900-1800 hours) controlled room. Temperature data in this and other studies were ana- lysed with repeated measures A N O V A and Tukey HSD multiple comparison tests.

Experiment 2: Investigation into the role of drug-predictive cues in the development of midazolam hypothermic tolerance. Forty-eight female hooded rats (240-285 g) housed as described above served as subjects in a counterbalanced discrimination design with "no-drug" controls, as recom- mended by Cunningham et al. (1984) as the optimal procedure for reliably demonstrating context specific tolerance (cf. Stewart and Eikleboom 1986). The basic experimental procedure is shown in Table 1.

In the counterbalanced discrimination procedure, two groups of subjects which received drug treatment were exposed equally often to two environments (the home room and a distinct room) during tolerance acquisition. These two groups were therefore equated for drug exposure and experience of the two environments. In the two "no drug" control groups subjects received vehicle injections in both environments in a counterbalanced fashion for the two groups. By counterbalancing drugged groups over environments and by using two "no-drug" controls, any potential differential effects of the environments, or of order of exposure to the two environments, on body temperatures could be identified.

515

After tolerance acquisition a test for environmentally specific tolerance was initiated (Table 1). Half the subjects in each drugged group received drug treatment in each of the two environments. Thus, for each group, the drug was given in the drug-predictive environment to half the subjects ("cued" groups). The other subjects received the drug in an environment never previously associated with drug treatment ("uncued" groups). The two control groups received midazolam for the first time in this phase of the study, one control group receiving drug treatment in each of the two environments. Thus the overall design of the experiment allowed an assessment, in the tolerance test phase of the study, of i) the degree of tolerance shown by drugged subjects relative to drug-naive controls, ii) the effects of drug-predictive environmental cues on the degree of tolerance observed and iii) any intrinsic effects of the environments themselves or of order of exposure to the environments on body temperature responses.

During the tolerance acquisition and tolerance test phases midazolam maleate was injected IP at 4 mg/kg in a volume of 2 ml/kg. Body temperature was recorded as described for Experiment 1. Only one measurement of body temperature was taken, at 80 min post-injection. Tempera- ture was measured at this time because Experiment 1 indicated that this was the time when 4 mg/kg midazolam would produce its peak effect. For subjects that received drug in the home cage room (midazolam home group-see Table 1), the procedure involved briefly removing subjects from their home cage room on odd-numbered days, trans- port to an adjacent room where they were weighed and then returned immediately to the home cage room. Ten minutes later core body temperatures were recorded and subjects were injected. 80 rain later the hypothermic effects of midazolam were recorded. On even-numbered days subjects in this group were weighed in the adjacent room and then taken immediately to a "distinct" room, where they were left undisturbed for 10 rain prior to injection of the vehicle solution (see Table 1), a second temperature assessment being conducted 80 rain post-injection, as in the home room. After 140 min subjects were returned to their home cage room. For the second drugged group (midazolam dis- t inct-see Table i) the conditions described above were reversed to produce the counterbalanced design. After weigh- ing and baseline temperature assessment, midazolam was injected in the "distinct room" on odd-numbered days and injections of distilled water were given in the home room on even-numbered days. Temperature assessment occured 80 rain post-injection on each day. On odd-numbered days, subjects were returned to the home room 140 min after midazolam injection in the distinct room.

The odd-numbered days of the tolerance acquisition phase thus constituted conditioning trials in which subjects were exposed to stimuli present in the home or distinct rooms which could serve as potential CSs. Exposure to these CSs preceded exposure to the drug by 10 min. Expo- sure to the CSs in the distinct room terminated 150 min after subjects were taken to the room (i.e. 60 min after temperature measurement). Thus the CSs of the distinct room were designed to reliably predict drug action, to precede it by 10 min and to overlap with both the peak drug effect and its gradual offset. Such a procedure might be expected to produce a substantial degree of conditioning because it maximises both the cueing effect of the CSs and CS-UCS temporal contiguity.

The distinct room used in this experiment differed from the home room in a number of ways. Firstly, it was satu- rated with the odour of amyl acetate (BDH Chemicals Ltd., Poole, UK), 20 ml of which was placed in an open beaker in the room. Amyl acetate was also soaked into paper towels located in the centre of the room, 10 rain before subjects were taken into the room. The subject's cages were arranged I m away from the beaker and from the towels to maximise olfactory cues. Secondly, the ambient illumination in the "distinctive room" was much lower than that of the home room (13 versus 200 Lux). The two rooms were further dif- ferentiated by their ambient noise levels. A "white" noise generator was present in the distinct room, (peak noise level= 81 dB at 500 c/s). In the home room, the noise level was lower (peak level= 66 dB at 63 c/s). The noise in the home room was attributable to the action of a continuously operative ventilation fan. Thus the two rooms differed from each other along a number of sensory dimensions.

Experiment 3 : Investigation number two into the role of drug- predictive cues in the development of midazolam hypothermic tolerance. In Experiment 2 no evidence was obtained for context specific tolerance, despite the fact that unequivocal evidence was obtained for the development of tolerance. Baker and Tiffany (1985) suggested that context specific morphine tolerance develops most reliably with low drug doses which are given infrequently (cf. Goudie and Demell- week 1986; and File 1985 for similar suggestions with other drugs). Thus in Experiment 3 we reduced the dose of midazolam from 4 to 1.6 mg/kg and gave the drug every 5th day as opposed to every other day, in an attempt to maximise the possibility of demonstrating context specific tolerance. A discriminative conditioning design with no-drug controls was used (as in Experiment 2), although it was not possible in this study to counterbalance exposure to the two environments, since drug injections were given every 5th day, vehicle injections being administered in the alternate environment on the 4 intervening days. This experiment involved the home room and the distinct room described above. However, we were concerned that our failure to find evidence in Experiment 2 for context specific tolerance could have been due to the use of a highly salient odour cue which might have been inadvertently transported between the distinct room and the home room, either via the laboratory ventilation or on the experimenter's clothes. We therefore eliminated the olfactory cue from the distinct environment in this and a subsequent study.

Subjects were female hooded rats (250-350 g) housed as described above. One group (n = 12) received midazolam (1.6 mg/kg; IP) every 5th day in the home room, the procedure being as described for Experiment 2 above. On the 4 intervening days these subjects received vehicle injections and temperature measurements in the distinct room. An- other group (n= 12) received midazolam injections every 5th day in the distinct room with vehicle injections and temperature measurements on intervening days in the home room. The remaining 12 subjects were split into two no- drug control groups (n=6 each) that were treated in the same way as in Experiment 2, i.e. one of the two groups acted as a matched control for the group that received drug in the home room and the other as a control for the group that received drug in the distinct room.

Temperature measurements were taken as described in Experiment 2 except that, in this study, the hypothermic

516

effect of midazolam was assessed 60 min post-injection, this being the time when we expected (on the basis of the results of Experiment 1) that a 1.6mg/kg dose of midazolam would have its peak effect. In this study exposure to the distinct room terminated 100 rain post-injection when we anticipated that drug-induced hypothermia would no lon- ger be present. Thus with the lower dose used in this study we recorded body temperature at a shorter time post-injection and we removed subjects from the distinct room earlier than in Experiment 2.

After five drug injections (with interspersed vehicle injections) for the two drugged groups and 25 vehicle injections for the two control groups, a test for context specific tolerance was conducted on Day 26 of the study. The two drug groups were each split into sub-groups (n = 6) which were matched for their hypothermic response on the 5th drug day. Two sub-groups received midazolam (1.6 rag/ kg IP) in their specific drug associated environments (cued groups), the other two sub-groups received uncued drug injections in the environment previously associated with vehicle injections.

Experiment 4: Investigation number three into the role of drug-predictive cues in the development of midazolam hypothermic tolerance. The results of Experiment 3 indicated clearly that no evidence had been obtained for context specific tolerance. We therefore conducted a further study in which we manipulated the temporal relationship between exposure to the drug-predictive distinct environment (the putative CS) and the drug's action in inducing hypothermia. In both the studies described above the CS was presented so as to overlap with both the onset and offset of drug- induced hypothermia. Some studies which have demonstrated context specific tolerance (e.g. Siegel 1975, 1976, 1977) have used procedures in which exposure to the CSs terminates at the time of peak drug effect. It is possible that cueing both the onset and offset of drug action may retard conditioning, since exposure to the CS during drug offset may inhibit conditioning (cf. Sherman 1979). Non- pharmacological classical conditioning studies have demonstrated such inhibitory effects in a variety of procedures (Barnes 1956; Burkhardt and Ayres 1978; Schneidermann 1966). We therefore conducted an experiment involving a fully counterbalanced conditioning procedure with a 2 rag/ kg dose of midazolam (IP). However, in this study, exposure to the distinct environment was terminated 80 rain post-injection at the time we expected (on the basis of the results of Experiment 1) midazolam's hypothermic effect at the dose used to be at a peak level.

Subjects were female rats (180-340 g) housed as described above. One group of subjects (n-- 12) received midazolam in the home room and vehicle injections in the distinct room on alternate days. For the second drugged group (n=12) this procedure was reversed. The procedure for body temperature assessment was as described above except that temperatures were recorded 80 min post-injection. Sub- jects were always returned from the distinct room to the home room immediately after temperature measurement. One vehicle-treated control group (n = 6) received the same schedule of assignments to rooms as the group that received drug treatment in the home room and the other (n= 6) was treated like the group of subjects that were drugged in the distinct room.

After eight drug injections a test for context specific

Time since in ject ion (min) Dose(mg/kg) group

5 20 40 80 120 160 200 300

0 I I I I I [ I

0

8

- 5

Fig. 1. Mean (±SE) change in temperature (° C) from baseline induced by midazolam at various doses (0, 2, 8 and 32 mg/kg) during the 300-min period after drug injection in Experiment t. The 0 mg/kg group was a control which received vehicle injection only

- 1

~-2

oE " - 3

c

- 4

tolerance was conducted. The group that had received drug in the home room was split into two sub-groups (n= 6) which were tested for tolerance in the home room ("cued") and in the distinct room ("uncued"). Similarly, two sub- groups were obtained from the subjects that had received drug in the distinct room and these sub-groups were tested in cued and uncued conditions, respectively. The two control groups received drug treatment in the home cage or the distinct environment respectively.

Results

Experiment 1: Dose- and time-effect study. The baseline core body temperatures (mean_+ SE) for the four experimental groups varied between 38.43__ 0.04 ° C and 38.31 +0.10 ° C. ANOVA demonstrated that there were no significant differences between the groups in baseline temperatures (F< 1, df= 3, 20). Figure I shows the change in temperature from baseline for each group as a function of time since injection.

A two-factor (groups, times) repeated measures AN- OVA indicated that doses, time and interaction effects were all significant (F=78.86, df=3, 20; F=62.77, df=7, 140 and F=14.93, df=21, 140, respectively, P<0.001 in all cases). At 5 rain post-injection the 8 and 32 mg/kg groups did not differ significantly from each other (Tukey test), but both groups were hypothermic relative to the control. At 20 min post-injection there was again no significant difference between the 8 and 32 mg/kg groups. At 300 min the 32 mg/kg group was hypothermic relative to all other groups, which did not differ significantly from each other. At all other times each group differed significantly from all other groups. These multiple comparisons show clearly that midazolam induced dose- and time-dependent hypothermia.

Experiment 2. The data collected on the vehicle (even-numbered) days of the tolerance acquisition phase of this study (and other studies described below) are not included in the results section of this report, since body temperature mea-

517

0

~ - 0 . 5 - c

0= - I . 0 - JQ E o

-1 .5- g c

- 2 . 0 - O

-2.5- 2 3 4 ~ 6 ~ 8 9 I'01'11'21'31~I'51'6

Injeotions of rnidazolam (4rnglkg) Tolerance

Acquisi t ion phase lest

Fig. 2. The left hand portion of the graph shows the mean temperature change from baseline (__ SE) of the control (circles) and midazolam (squares) (n=16) groups during the 16 drug days in the home room of the tolerance acquisition phase of Experiment 2. The right hand portion of the graph shows the mean hypothermic effect (--SE) of 4mg/kg midazolam in subjects of the cued (squares), uncued (triangles) and control (circles) sub-groups in the tolerance test of Experiment 2. A star indicates P<0.01 vs the control group in the tolerance test

0

- 0 . 5 " =o

e) - 1 . 0

E o

-1 .5 - g c

- 2 o - O

UU I i r [ r -2.5- ~ ~ ~ ~, g ~ ~ ~ 91'ot ' l t21a141518

Injections of midazolam (4mg/kg) Tolerance

Acquisit ion phase test

Fig. 3. The left hand portion of the graph shows the mean temperature change from baseline ( 4- SE) of the control (circles) and midazolam (triangles) groups during the 16 drug days in the distinctive room of the tolerance acquisition phase of Experiment 2. The right hand portion of the graph shows the mean hypothermic effect (_+SE) of 4 mg/kg midazolam in subjects in the cued (triangles), uncued (squares) and control (circles) sub-groups in the tolerance test of Experiment 2. A star indicates P < 0.01 vs the control group in the tolerance test

surements were only taken on these days to ensure that the temperature measurement procedure did not itself act as a reliable drug-predict ive cue.

The hypothermic effects of midazo lam during the tolerance acquisi t ion phase of this s tudy are shown in the left hand por t ions of Figs. 2 and 3.

These da ta indicate clearly that the subjects receiving midazo lam in the home room (Fig. 2) and in the distinct room (Fig. 3) showed similar initial hypothermic responses (approximate ly 2 ° C) relative to their respective control groups. Thus, the effects of midazolam were clearly not modif ied by differences between the two environments. After their 16th injection of midazo lam the two midazolam- treated groups showed approximate ly 0 .6°C of hypothermia relative to their respective controls. Equivalent levels of tolerance therefore developed in the home and distinct rooms. Compar i son between Figs. 2 and 3 also indicates clearly that the rate of tolerance acquisi t ion was similar over successive injections in the two rooms.

The right hand por t ions of Figs. 2 and 3 show the results of the tolerance tests conducted in the home (Fig. 2) and distinct (Fig. 3) rooms, respectively. The home room tolerance test (Fig. 2) produced a significant difference between groups in their hypothermic responses (F=12.56 , dr=2, 21, P<0 .001) . The drug-naive control group was hypothermic relative to both the cued and uncued drug experienced groups (Tukey's H S D tests, P < 0.01), but there was no difference in the effect of midazolam in the cued and uncued groups. There was also a significant overall groups effect for the three groups that received midazolam in the distinct room (Fig. 3) (F~-18.07, d f = 2 , 21, P<0 .001) . The drug-naive group was again significantly hypothermic com- pared to both the drug-experienced uncued and cued groups (P<0 .01 in both cases). However, the effect of midazo lam did not differ in these two groups. These results indicate clearly that tolerance was acquired by subjects that received drug in both environments, as noted during the tolerance acquisi t ion phase of the study. However, the tolerance ac-

- 0 . 8 - v iii z

w co

< -1 .6 rn

2; O 0" ri- LL

z <

O ~ - 0 . 8 -

- 1 . 6 -

HOME ROOM

/2/ -Jr

DISTINCTIVE ROOM

1 2 3 4 5 TOLERANCE DRUG DAYS TEST

Fig. 4. Mean (--SE) temperature change from baseline induced by 1.6 mg/kg midazolam given every 5th day for subjects in the home and distinctive rooms over 5 days of treatment and during the tolerance test. Circles = two control groups which received saline prior to the tolerance test and midazolam on the test. For the tolerance test in the home room squares=cued group, triangles = uncued group. For test in the distinct room, triangles = cued group and squares=uncued group. Empty star denotes P<0.05 vs control groups (filled stars) in each room

quired by both drugged groups was not dependent upon drug adminis t ra t ion in the relevant drug-predictive rooms, i.e. it was not context specific. Fur ther tolerance tests with these subjects (data not shown) confirmed these findings (i.e. no evidence for context specificity was obtained) and also indicated that cross-tolerance to ethanol (•.2 or 1.6 g/

518

o v

w Z m . J [u cO <

0 t~ u_

w fl:

I..- <

0.

t~ I -

Z

Z < 1- ¢J

- 0 . 7 -

- 1 . 4 -

ACQUISITION PHASE : TOLERANCE TEST

HOME ROOM

DISTINCTIVE ROOM

o-

- 1 . 4 -

1 2 3 4 5 6 7 MIDAZOLAM INJECTIONS

Fig. 5. Mean ( + SE) temperature change from baseline for subjects in home and distinct rooms during tolerance acquisition and test- ing. Circles=two control groups which received saline prior to the test and midazolam on the test. For the tolerance test in the home room squares=cued group, triangles=uncued group. For the test in the distinct room, triangles=cued group, squares=uncued group. Circle with an empty star = P < 0.01 and star = P < 0.05 vs control groups

kg) was absent, regardless of whether or not ethanol was given in a drug-predicfive environment. This study provides no support at all for the idea that midazolam tolerance can show context specificity.

Experiment 3. The results of this study are shown in Fig. 4. The data shown in the left hand portions of the figure

indicate clearly that subjects which received 5 days of chronic midazolam treatment in either environment acquired tolerance to drug-induced hypothermia. The lower degree of hypothermia seen in this study relative to Experi- ment 2 was due to treatment with a lower dose of midazo- lain. It resulted in the development of complete tolerance in a shorter period of chronic drug treatment (compare Figs. 2 or 3 with Fig. 4). For the three groups that received midazolam in the home room during the tolerance test (right hand portion of top panel of Fig. 4) there was a significant group effect (F=4.7, df=2, 15, P<0.05). The drug-naive control group was hypothermic relative to both drug-experienced groups (cued and uncued), but these two groups did not differ from each other (Tukey HSD tests e= 0.05). Similarly, for the groups that received the tolerance test in the distinct room (right hand portion of lower panel of Fig. 4), there was a significant groups effect (F= 6.7, df=2, 15, P<0.01). However, the drug-naive group again proved to be hypothermic relative to the two drugged groups (cued and uncued), whilst these did not differ significantly from each other (e= 0.05). Thus clear evidence was obtained for tolerance to midazolam-induced hypothermia,

but there was no evidence that the observed tolerance was context specific.

Experiment 4. The results from this study are shown in Fig. 5.

The data shown in the left hand portions of Fig. 5 indicate clearly that tolerance developed to the hypothermic effect of midazolam in both environments. In the test for context specific tolerance the data obtained in the home room (right hand portion of top panel of Fig. 5) indicated that there was a significant overall groups effect (F= 16.2, df=2, 33, P<0.01). The control group was significantly hypothermic (c~=0.01) relative to the two drugged groups (cued and uncued), which did not differ significantly from each other (0~= 0.05). For the data from the distinct room (lower panel of Fig. 5) there was also a significant overall groups effect (F=8.6, df=2, 33, P<0.01). Again, the control group was significantly hypothermic relative to the two drug groups (cued and uncued), which did not differ significantly from each other. Thus it was again clear that tolerance developed in subjects that received drug in either environment, but the observed tolerance was not context specific.

Discussion

Midazolam, like other sedative/hypnotics (Kalant and Le 1984) induced hypothermia at low ambient temperatures in agreement with previous findings with BZs (Greeley and Cappell 1985). It is not, however, at present clear whether hypothermia was due to a "pure" hypothermic drug effect or to poikilothermia (cf. Myers 1981). Midazolam-induced hypothermia reliably showed tolerance, as is typical with sedative/hypnotics (Kalant and Le 1984). However, no evidence was obtained for context specific tolerance in three separate studies in which we varied drug dose, frequency of drug administration and the temporal relationship between the drug effect and exposure to the drug-predictive environment. Negative findings were obtained with procedures which resembled those used to show context specific tolerance to hypothermic effects of ethanol, morphine and pentobarbital (Le et al. 1979; Crowell et al. 1981; Greeley et al. 1984; Shapiro et al. 1983). Many of these studies used counterbalanced discrimination designs with environments differing in terms of auditory and visual cues, as in the studies reported above. Furthermore, our data showing no effect of environmental stimuli on the acquisition of BZ tolerance resemble the findings of File (1982) with CDP; although it should be noted that File's (1982) data indicat- ing that environmental stimuli facilitated CDP tolerance retention are compatible with a conditioning account of BZ tolerance.

There are three different potential explanations for our failures to find context specific tolerance. Firstly, the environmental CSs may simply have not been sufficiently different for subjects to discriminate between environments. Sec- ondly, the environmental CSs that were designed to signal the presence or absence of the drug may not have supported conditioning because they were overshadowed by other more salient stimuli common to both environments. Final- ly, tolerance to midazolam hypothermia may genuinely not show context specificity. We consider below these different interpretations- of the data.

519

Although we explicitly attempted to set up two environments that differed in terms of auditory, olfactory and visual cues to make them maximally discriminable, it is possible that, as far as our rodent subjects were concerned, they may have simply been indistinguishable. Thus subjects may not have shown context specific tolerance since the two contexts did not differ "functionally". It is difficult to evaluate this potential explanation of our data since, despite the wide literature on context specific tolerance, very little research has been conducted into the precise nature of the "functional" CSs that control such tolerance (Corfield- Sumner and Stolerman 1978). Most studies on context specific tolerance have used complex compound CSs which, as in the studies described above, differed in terms of a number of sensory dimensions. It is not clear what, if any, specific aspects of these contextual CSs are the most salient cues for experimental subjects. A priori, it is therefore only possible to guess at the type of environmental stimuli which should support conditioning. This is clearly unsatisfactory. The type of CSs used in conditioning studies with drugs may also determine the direction and magnitude of acquired CRs (Flaherty et al. 1980; Eikelboom and Stewart 1982; Stewart and Eikelboom 1986) and hence the results of studies into contextual determinants of tolerance and sensitiza- tion (Siegel and MacRae 1984). Since the results of conditioning studies with drugs may depend critically on the nature of the CSs used, future parametric research might pro- fitably attempt to analyse the nature of the "functional" CS in so far as this determines the results of tolerance studies. At present, failures to obtain context specific tolerance, such as those described above, may always be attrib- uted to the use of the "wrong" type of CSs.

Walter and Riccio (1983) reported that overshadowing can occur in studies of context specific morphine tolerance. They therefore suggested that failures to obtain evidence for context specific tolerance may involve unknown environmental or procedural stimuli common to a number of environments which overshadow the "nominal" CSs. Walter and Riccio (1983) suggested that interoceptive stimuli generated by the drug itself may act as drug-predictive cues (cf. Greeley et al. 1984) and overshadow exteroceptive cues, leading to a general trans-situational form of conditioned tolerance. In similar fashion, Dafters and Bach (1985) argued that failures to show context specific tolerance may be due to overshadowing by highly salient injection-related CSs of exteroceptive CSs. In support of this argument Dafters and Bach (1985) reported that context specific morphine tolerance was only obtained in subjects that had been pre-exposed before conditioning to the injection procedure which did not therefore subsequently act as a reliable predictor of the drug. It is therefore possible that our failures to detect context specific tolerance were due to overshadowing of our "nominal" environmental CSs by either exteroceptive or interoceptive stimuli common to both environments. This second potential explanation for our negative findings is also difficult to evaluate, since we cannot assess the relative salience of, and degree of overshadowing between, injection procedures and drug stimuli on the one hand and the environmental CSs on the other. It is even possible that the salience of injection-related CSs may vary depending on the specific manner in which animals are injected (Dafters and Bach 1985), in which case the salience of injection-related cues may vary between ex- perimenters and between laboratories. This discussion again

highlights the issue discussed above of the importance of trying to define clearly what the "functional" CSs are in studies of context specific tolerance. We cannot, therefore, at present discount the possibility that our negative findings were attributable to overshadowing of environmental CSs by unknown cues.

A third potential explanation for our data is that it will never prove possible to demonstrate context specific tolerance to midazolam hypothermia, since such tolerance may never show environmental specificity under any conditions. Obviously, the validity of this account of our data depends on the validity of the two alternative accounts of our findings outlined above. Since we have already described the difficulties involved in evaluating these alternative explanations, it is clear that we cannot, on the basis of our data alone, make an unequivocal assessment of the notion that midazolam hypothermic tolerance will never show context specificity. This is one possible explanation for our negative findings, but the data available at present do not allow us to discriminate between this explanation and the two alternatives outlined above. However, in dis- cussing this account of our data it is clearly important to relate our findings to the studies of Greeley and Cappell (1985) on context specific CDP tolerance. These authors presented some data which provided support for the idea that it is possible to demonstrate context specific CDP tolerance. In a study on the sedative effects of CDP, Greeley and Cappell (1985) initially failed to find evidence for context specific tolerance in two experiments using a behavioural measure ("sleep time") which had previously been reported in their laboratory to produce context specific pentobarbital tolerance (Hinson et al. 1982). They did, however, find evidence for context specific tolerance to the sedative effects of CDP when sedation was measured in terms of a different behavioural index - "inactivity time". In a further study (Greeley and Cappell 1985, Experiment 2), an attempt was made to demonstrate context-specific hypothermic tolerance with CDP. However, they failed to detect any difference in hypothermic response between subjects receiving cued and uncued CDP, exactly as in all our studies with midazolam described above. When the hypothermic effect of CDP was tested in a completely novel environment, tolerance appeared to be "lost" since subjects showed a greater degree of hypothermia than that seen in both the cued and uncued environments. However, this experiment, which was interpreted by Greeley and Cappell (1985) as providing evidence for context specific hypothermic tolerance, failed to control for the possibility that some intrinsic property of the novel environment itself could have inter- acted with the hypothermic effect of CDP. This specific study does not, therefore, provide reliable evidence for context specific hypothermic tolerance, since it is well known that stress can modify the thermic effects of drugs (e.g. Stewart and Eikelboom 1979). In contrast, our studies, which have controlled for such effects, have not generated evidence for context specific tolerance to the hypothermic effects of midazolam. We therefore conclude that there is not at present reliable evidence that context specific tolerance develops to the hypothermic effects of BZs. Treit (1985) described a preliminary study which suggests that tolerance to the actions of diazepam in an animal model of anxiety is not affected by environmental cues. It would therefore seem that it is unlikely that all actions of BZs will show context specific tolerance. It is possible that other

520

measures of BZ activity may provide more robust evidence for context specific BZ tolerance, since different effects of a drug are known to be differentially susceptible to context specificity effects (Stewart and Eikelboom 1978; Domjan and Siegel 1983). Recent evidence (Pinel and Puttaswa- maiah 1985) indicates that tolerance to the anticonvulsant action of ethanol does not show environmental specificity, in contrast to the well documented (Goudie and Demell- week 1986) context specificity shown for the hypothermic, sedative and toxic actions of this drug. It is not at present clear why some effects of a drug may show context specificity whilst others may not. Nevertheless, the findings of Gree- ley and Cappell (1985) on CDP tolerance acquisition and the data reported by File (1982) on retention of CDP tolerance suggest that unequivocal evidence for context specific BZ tolerance is perhaps more likely to derive from studies of the actions of BZs on activity than on body temperature for reasons which are not at present clear, although it is possible that the difficulties we have encountered in demonstrating context specific hypothermic tolerance with a BZ drug may perhaps be related to the well-known amnestic effects of BZs which could interfere with conditioning and retard the acquisition of associative tolerance (cf. Greeley and Cappell 1985).

In summary, the data reported above indicate that in three separate studies midazolam did not produce context specific tolerance. These negative findings may be attributable either to poor "functional" discriminability between the specific environmental CSs that we used or to overshadowing; alternatively, it is possible that the hypothermic actions of BZs, in contrast to those of other sedative/hypnotics, do not show context specificity which may be more readily demonstrated with other ways of assessing BZ tolerance. Further research is clearly needed to clarify the conditions under which BZs do and do not show context specificity of tolerance.

Acknowledgements. We are indebted to Mrs. Anne Halliwell for typing the paper, to Mike Taylor for care of our animals.

References

Baker TB, Tiffany ST (1985) Morphine tolerance as habituation. Psych Rev 92:78-107

Barnes GW (1956) Conditioned stimulus intensity and temporal factors in spaced-trial classical conditioning. J Exp Psychol 51 : 192-198

Burkhardt PE, Ayres JJB (1978) CS and US duration effects in one-trial simultaneous fear conditioning as assessed by conditioned suppression of licking in rats. Anim Learn Behav 6 : 22%230

Cappell H, Roach C, Poulos CX (1981) Pavlovian control of cross- tolerance between pentobarbital and ethanol. Psychopharma- cology 74:54-57

Corfield-Sumner PK, Stolerman IP (1978) Behavioural tolerance. In: Blackman DE, Sanger DJ (eds) Contemporary Research in Behavioural Pharmacology. Plenum Press: New York, pp 391-448

Crowell CR, Hinson RE, Siegel S (1981) The role of conditional drug responses in tolerance to the hypothermic effects of ethanol. Psychopharmacology 73 : 51 54

Cunningham CL, Crabbe JC, Rigter H (1984) Pavlovian conditioning of drug-induced changes in body temperature. Pharmac Ther 23 : 365-391

Darters R, Bach L (1985) Absence of environment-specificity in morphine tolerance acquired in non-distinctive environments: Habituation or stimulus overshadowing? Psychopharmacology 87:101 106

Demellweek C, Goudie AJ (1983) Behavioural tolerance to amphetamine and other stimulants: The case for considering behavioural mechanisms. Psychopharmacology 80:287 307

Domjan M, Siegel S (1983) Attenuation of the aversive and analgesic effects of morphine by repeated administration: Different mechanisms. Physiol Psychol 1 t : 155-158

Eikelboom R, Stewart J (1982) Conditioning of drug-induced phys- iological responses. Psych Rev 89 : 507-528

File SE (1982) Development and retention of tolerance to the sedative effects of chlordiazepoxide: Role of apparatus cues. Eur J Pharmacol 81:637-643

File SE (1985) Tolerance to the behavioral actions of benzodiazepines. Neurosci Biobehav Rev 9:113-121

Flaherty CF, Uzwiak A J, Levine J, Smith M, Hall P, Schuler R (1980) Apparent hyperglycaemic and hypoglycaemic conditioned responses with exogenous insulin as the unconditioned stimulus. Anim Learn Behav 8:382-386

Goudie A J, Griffiths JW (1985) Benzodiazepine hypothermic tolerance: Analysis of the role of drug-predictive environmental stimuli and cross-tolerance to ethanol. Soc Neurosci Abstr 11:636

Goudie AJ, Demellweek C (i986) Conditioning factors in drug tolerance. In: Goldberg SR, Stolerman IP (eds) The behavioral basis of drug dependence. Academic Press, New York, pp 225 285

Greeley J, Cappell H (1985) Associative control of tolerance to the sedative and hypothermic effects of chlordiazepoxide. Psy- chopharmacology 86:487483

Greeley J, Le AD, Poulos CX, Cappell H (1984) Alcohol is an effective cue in the conditional control of tolerance to ethanol. Psychopharmacology 83 : 159-162

Hinson RE, Poulos CX, Cappell H (1982) Effects of pentobarbital and cocaine in rats expecting pentobarbital. Pharmacol Bio- chem Behav 16:661 666

Hinson RE, Rhijnsburger J (1984) Learning and cross drug effects: Thermic effects of pentobarbital and amphetamine. Life Sci 34:2663-2640

Kalant H, Le AD (1984) Effects of ethanol on thermoregulation. Pharmacol Ther 213 : 313 364

Le AD, Poulos CX, Cappell H (1979) Conditioned tolerance to the hypothermic effect of ethyl alcohol. Science 206 : 1109 1110

Mansfield JG, Cunningham CL (1980) Conditioning and extinction of tolerance to ethanol in rats. J Compr Physiol Psychol 94:962-969

Melchior CL, Tabakoff B (1981) Modification of environmentally cued tolerance to ethanol in mice. J Pharmacol Exp Ther 219:175-180

Myers RD (1981) Alcohol's effect on body temperature: Hypother- mia, hyperthermia or poikilothermia? Brain Res Bull 7 : 209-229

Pieri L etal. (1981) Pharmacology of midazolam. Drug Res 31:2180-2201

Pinel JPJ, Puttaswamaiah S (1985) Tolerance to alcohoFs anticonvulsant effect is not under Pavlovian control. Pharmacol Bio- chem Behav 23 : 959-964

Schneidermann N (1966) Interstimulus interval function of the nic- titating membrane response of the rabbit under delay versus trace conditioning. J Comp Physiol Psychol 62 : 397-407

Shapiro NR, Dudek BC, Rosellini RA (1983) The role of associative factors in tolerance to the hypothermic effects of morphine in mice. Pharmacol Biochem Behav 19:32~333

Sherman JE (1979) The effects of conditioning and novelty on the rats analgesic and pyretic responses to morphine. Learn Motiv 10 : 383-418

Siegel S (1975) Evidence from rats that morphine tolerance is a learned response. J Comp Physiol Psychol 89:498-506

Siegel S (1976) Morphine analgesic tolerance: Its situation specifici-

521

ty supports a Pavlovian conditioning model. Science 193:323-325

Siegel S (1977) Morphine tolerance acquisition as an associative process. J Exp Psychol: Anim Behav Processes 3 : 1-13

Siegel S (1982) Pavlovian conditioning and drug effects: Recent research. In: Colpaert FC, Slangen JL (eds) Drug Discrimina- tion: Applications in C.N.S. Pharmacology. Elsevier Biomedi- cal Press, Amsterdam, pp 287-304

Siegel S, MacRae J (1984) Environmental specificity of tolerance. Trends Neurosci 7:140-143

Stewart J, Eikelboom R (1978) Pre-exposure to morphine and the attenuation of conditioned taste aversion in rats. Pharmacol Biochem Behav 9 : 639 645

Stewart J, Eikelboom R (1979) Stress masks the hypothermic effect of naloxone in rats. Life Sci 25 : 1165 ] 171

Stewart J, Eikelboom R (1986) Conditioned drug effects. In: Iver- sen LL, Iversen SD, Snyder SH (eds) Handbook of Psychophar- macology. Vol 19, Academic Press, New York (in press)

Treit D (1985) Evidence that tolerance develops to the anxiolytic effect of diazepam in rats. Pharmacol Biochem Behav 22:383 387

Waiter TA, Riccio DC (1983) Overshadowing effects in the stimulus control of morphine analgesic tolerance. Behav Neurosci 97 : 658-662

Received March 13, 1985; Final version June 25, 1986

Documents

Analysis of the role of drug-predictive environmental stimuli in tolerance to the hypothermic effects of the benzodiazepine midazolam