Effects of the 5-HT3 antagonist GR38032F (ondansetron) on benzodiazepine withdrawal in rats

Effects of the 5-HT3 antagonist GR38032F (ondansetron) on benzodiazepine withdrawal in rats

European Journal of Pharmacology, 185 (1990) 179-186 179 Elsevier EJP 51457 Effects of the 5 - H T 3 antagonist GR38032F (ondansetron) on benzodiaz...

640KB Sizes 1 Downloads 37 Views

European Journal of Pharmacology, 185 (1990) 179-186

179

Elsevier EJP 51457

Effects of the 5 - H T 3 antagonist GR38032F (ondansetron) on benzodiazepine withdrawal in rats x A n d r e w J. G o u d i e a n d M i c h a e l J. L e a t h l e y Department of Psychology, Unioersityof Liverpool, P.O. Box 147, Liverpool L69 3BX, U.K.

Received 15 February 1990, revised MS received 8 May 1990, accepted 5 June 1990

Effects of the 5-HT3 receptor antagonist GR38032F (ondansetron) on chlordiazepoxide withdrawal were assessed in rats which received ehlordiazepoxide b.i.d, for 21 days at doses up to 40 mg/kg per injection. Withdrawal signs recorded were: body weight and food intake, which fell and then recovered over 9 days. Saline or GR38032F (1.0, 0.1 or 0.01 mg/kg) were administered b.i.d during withdrawal. Clear withdrawal signs were seen in rats treated with saline after chronic chlordiazepoxide. However, GR38032F at 0.1 mg/kg reduced the severity of withdrawal. At 0.01 mg/kg GR38032F shortened withdrawal duration, but did not diminish peak withdrawal signs. At 1.0 mg/kg GR38032F, did not attenuate withdrawal signs at all. GR38032F (0.01-1.0 mg/kg) had no effect on ad lib food intake, therefore the attenuation of withdrawal was probably not simply due to stimulation of appetite. These data support recent claims that GR38032F attenuates benzodiazepine withdrawal, and they indicate that this effect shows an inverted U-shaped dose-response curve. Benzodiazepines; Dependence; Withdrawal; Ondansetron (GR38032F); 5-HT3 receptors; (Rat)

1. I n t r o d u c t i o n

Anxiolytic benzodiazepines (BZs) induce withdrawal signs after chronic administration (Woods et al., 1987). There is therefore a need for anxiolytics which do not induce dependence and for agents which can be used to treat BZ withdrawal. Research efforts have consequently been directed to developing novel anxiolytics acting on 5-HT systems (Chopin and Briley, 1987). One such agent, buspirone, is an effective anxiolytic (Goa and Ward, 1986; Simpson et al., 1989),

1 Someof the data reported here were presented to The British Pharmacological Society at the Jarluary 1990 (London) meeting and are 'in press' in Abstract form in The British Journal of Pharmacology. Correspondence to: A.J. Goudie, Department of Psychology, University of Liverpool, P.O. Box 147, Liverpool 1..69 3BX, U.K.

which is thought to act via 5-HT1A receptors. However, buspirone is of little value in the treatment of BZ dependence (Schweizer and Rickels, 1986; Costall et al., 1989b). Indeed, prior BZ experience m a y retard anxiolytic actions of buspirone and related agents (Schweizer et al., 1986; H a r t o et al., 1988; Simpson et al., 1989). Anxiolytics acting at the 5-HT1A receptor are therefore probably of limited value in treating BZ dependence or in benzodiazepine experienced subjects. Some studies suggest that the BZ antagonist flumazenil m a y be effective in the treatment of BZ dependence (Gallagher et al., 1986; Baldwin and File, 1988; 1989) although the evidence in favour of this hypothesis is not unequivocal (Sannerud et al., 1989). Selective 5 - H T 3 receptor antagonists have been reported to be anxiolytic in some (Jones et al., 1988; Tyers, 1989; Costall et al., 1989a) but not all (File and Johnston, 1989) animal studies. Since

0014-2999/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

180

there are at present no established treatments for BZ dependence; and since anxiolytics acting at the 5-HTIA receptor may be of limited use in BZ-experienced subjects, it is important to determine the actions of 5-HT3 antagonists in animals after prior BZ treatment. In the studies reported here we assessed the actions of the 5-HT3 antagonist ondansetron (GR38032F) in animals dependent upon chlordiazepoxide. Anorexia and weight loss are reliable indices of BZ withdrawal (Woods et al., 1987), we consequently examined the effect of ondansetron on BZ withdrawal and we report that at specific doses ondansetron attenuated, but did not block, these withdrawal signs. One potential explanation for these data could relate to the well-known actions of 5-HT in the modulation of food intake. Since 5-HT release inhibits food intake (Blundell, 1984), it was possible that 5-HT3 antagonists might stimulate food intake. If such an effect occurred with ondansetron, it would be expected to attenuate the weight loss and anorexia induced by BZ withdrawal, due simply to stimulation of 'appetite'. However, we subsequently determined that when ondansetron was administered with a regime which attenuated BZ withdrawal-induced anorexia and weight loss, it had no significant effects on food intake and body weight. We conclude that o n d a n s e t r o n attenuates BZ withdrawal signs and that this effect is probably not due simply to stimulation of appetite. The studies reported here were initiated after abstract reports were published suggesting that ondansetron attenuates behavioural signs of BZ withdrawal (Oakley et al., 1988; Costall et al., 1988). These data were recently published in a full paper (Costall et al. 1989b), after we completed our studies. The relationship between our data and those of Costall et al. (1989b) is considered in detail in the discussion section of this paper.

2. Materials and methods

2.1. Animals Female wistar rats (210-323g; Liverpool University Psychology Department) were housed individuaUy in a humidity and temperature controlled

room on a 10/14 h light/dark cycle (lights on at 0 8 : 0 0 h). The rats were habituated to the room for at least six days before any injections were administered. Water and food (Pilsbury Small Animal Diet. Liverpool, UK) were available ad lib.

2.2. Experimental procedures Two experiments were carried out. The first was concerned with assessing the effects of GR38032F on BZ withdrawal and the second with assessing the effects of GR38032F on food intake. The experimental design for the first study involved five groups of subjects (n = 8-10 per group). The study was carried out over 30 days in two phases. During phase I, all subjects were treated chronically (for 21 days) b.i.d, with chlordiazepoxide on an escalating dose regime. Chlordiazepoxide was initially administered at 10 m g / k g b.i.d. This dose was progressively increased by 2 m g / k g per day up to a final dose of 40 m g / k g on day 16. Between days 16-21 the dose administered b.i.d. remained constant at 40 mg/kg. All drug injections were administered at 10:00 and 16:00 h daily. In phase II of the study withdrawal was initiated in one of the experimental groups (group C-S) by substituting saline (b.i.d.) for chlordiazepoxide. In a second group (coded C-C) subjects continued to receive chlordiazepoxide at 40 m g / k g b.i.d, during the 'withdrawal' phase of the study. In the three remaining groups (coded C-G1.0, CG0.1, and C-G0.m) GR38032F at three different doses (1.0, 0.1 and 0.01 m g / k g b.i.d.) was substituted for chlordiazepoxide. The overall design of the study was such that various groups of animals withdrawn from chlordiazepoxide could be compared with a group (C-C) that did not experience withdrawal from chlordiazepoxide. It was also possible to determine whether any withdrawal signs seen in phase II of the study in group C-S were attenuated or enhanced by treatment over a number of days with GR38032F at three different doses. Throughout the study bodyweights and 24 h food intakes were recorded daily for each animal. These measurements were taken before the first daily injection at 10:00 h (i.e. 18 h after the last injection at 16 : 00 h on the previous day).

181

In the second study, which assessed effects of GR38032F on food intake and body weight, subjects were allocated to four groups matched for body weights (n = 9 per group). The groups were subsequently treated with saline (controls) or GR38032F at one of the three doses used in the previous study (0.01, 0.1 and 1.0 mg/kg). After habituation to their housing conditions, baseline body weights and 24 h food intakes were recorded for each of the experimental groups. After baseline body weights and 24 h food intakes were recorded, subjects received two injections (at 11:00 and 17:00 h). This dosing regime was based on that used in the previous study, in which G R 3 8 0 3 2 F was f o u n d to ameliorate BZ withdrawal. After drug or saline treatment, the posttreatment body weights and 24 h food intakes were recorded on the following day.

285

280 275 p3: uJ

* ..... * .....

* *

C-G C-G

0.01

• ......



C-G

1.0

• ..... • .....

• •

C-S C-C

0.1 /

/ .~.~'~__

~

-

270 265

>-

260 O

rn 255 250 245

DAY Fig. 1. Mean group body weights (g) for the five experimental groups in the first study (see text for group codes) over the experimental days that involved drug withdrawal. (S.E.s are omitted for the sake of clarity.) Withdrawal from chlordiazepoxide was initiated on day 22 for all experimental groups except group C-C.

2.3. Drugs Drugs were: chlordiazepoxide hydrochloride (Sigma Chemicals, Poole, U.K.) and ondansetron (GR38032F) (Glaxo Group Research Ltd., Ware, U.K.). Drugs were made up as salts dissolved in isotonic saline at room temperature. They were injected i.p. at a volume of 2 m l / k g of rat.

2.4. Statistical analyses In the first study body weight data were analysed with two factor (groups, days) repeated measures ANOVAs, followed by Tukey HSD multiple comparison tests where appropriate. 24 h food intake data were analysed with non-parametric Kruskal-Wallis ANOVAs followed by MannWhitney U-tests where appropriate. In the second study, body weight and food intake data were analysed with separate two factor (groups, days) ANOVAs with repeated measures over days. The days factor had two levels (data obtained on the baseline day and that obtained after drug treatment).

3. Results

The results of the two experiments are presented consecutively. Figure 1 shows the mean

weights of the groups in the first experiment over the withdrawal phase of the study. Over days 22-30 no weight loss was apparent in group C-C which actually gained weight. In contrast, group C-S lost weight on withdrawal from chlordiazepoxide, showing maximal weight loss on day 25. This was followed by a progressive increase in weight such that by day 30 animals had regained some of the lost weight, although even by day 30 they were not at the same level as animals that did not experience withdrawal (group C-C). All three groups treated with GR38032F during withdrawal also lost weight, although group C-G0. t lost less weight than group C-S, showing a maximal loss of weight on day 23. T o analyse these data statistically so as to make valid between group comparisons it was not possible simply to compare absolute weights, as there were differences between groups in their mean weights prior to withdrawal. Each individual's weight over the withdrawal period (days 22-30) was therefore expressed as a percentage of the relevant pre-withdrawal (day 21) baseline. These percentage body weight data are shown in fig. 2. An ANOVA on these data revealed a significant days effect F(8,336)--31.4, P < 0.001), and more importantly a highly significant groups × days interaction (F(32,336) -- 2.5, P < 0.01). Post h o c Tukey tests (et = 0.05) showed

182

that, relative to subjects maintained on chlordiazepoxide (Group C-C), group C-S subjects had a significantly lower body weight on days 24 through 30. Therefore significant withdrawal-induced loss of weight was observed, and this persisted throughout the study. Relative to the group of animals withdrawn from chlordiazepoxide which were treated with saline (group C-S), group C-G0.1 had a significantly higher body weight on day 25, and on days 27-30 inclusive. Therefore GR38032F at this specific intermediate dose significantly attenuated withdrawal-induced loss of weight. However, this group also had a significantly lower body weight on day 26 than animals never withdrawn from chlordiazepoxide (group C-C). Thus GR38032F at 0.1 mg/kg attenuated, but did not fully block, the effects of chlordiazepoxide withdrawal. At doses above and below 0.1 mg/kg (i.e. in groups C-G1.0 and C-G0.01), no significant differences in percentage body weights relative to group C-S were observed on any day, except in group C-G0.01 on days 29 and 30, when this group was at a significantly higher bodyweight. Thus at the lowest dose of GR38032F studied (0.01 mg/kg), ondansentron did not affect the magnitude of the peak withdrawal effect. However, it did shorten the duration of time over which withdrawal induced significant weight loss (fig. 2). In summary, GR38032F attenuated, but did not block, withdrawal signs, particularly at the intermediate dose of 0.1 mg/kg. However, evidence of GR38032F-induced attenuation of withdrawal was consistently absent in animals treated with the highest dose studied - 1.0 mg/kg. The group median 24 h food intake data recorded during the withdrawal phase of this study

105

104

* ..... * .....

*

103

• .....

• C-G 1.0

• .....



P-'r

102

;11

100

101

A. . . . .

#

C-G C-G c-s

=' C - C

0.01 0.1

&

./--~, ~.,_.....~

~v-----=~

j . ~

~ ,T

99 0 m Z ,< tJU ,~

98 97 96 95 94 93

DAY

Fig. 2. Mean group body weights for the five experimental groups in the first study (see text for group codes) over experimental days 22-30. All bodyweights were expressed as percentages of the baseline levels recorded on day 21 prior to the onset of withdrawal. (S.E.s are omitted for the sake of clarity.)

are shown in table 1. A number of animals ate virtually nothing at all on some days during withdrawal, these data were therefore analysed with nonparametric statistics, as they were not normally distributed. Separate Kruskal-Wallis ANOVAs were conducted for each day's data. Significant overall differences between groups were only seen on days 24 (H(4)= 11.6, P < 0.05) and 25 (H(4)= 13.5, P < 0.01). Between group comparisons showed that on day 24, group C-S ate less than group C-C (P < 0.05, two-tailed). Withdrawal from chlordiazepoxide therefore reduced food intake significantly in saline-treated animals. However, on this day group C-G0.z ate significantly more than group C-S (P < 0.01, two-

TABLE 1 Median mounts of food eaten (g) in the preceding 24 h periods for the five experimental groups in the first study (see text for group codes) over experimental days 22-30 during the withdrawal phase of the study. Group C-S C-C C-Gz.0 C-G0a C-G0.0o1

Day 22

23

24

25

26

27

28

29

30

18.0 14.1 17.8 18.5 18.6

11.0 13.8 7.8 12.2 12.0

9.3 a 16.3 13.1 16.4 b 12.3

14.6 16.9 18.6 20.5 b 18.0

18.4 18.4 15.4 19.0 20.3

16.5 17.7 17.9 21.4 19.7

19.6 18.3 17.8 19.0 22.6

19.0 17.7 18.2 20.6 21.3

17.0 16.9 19.1 20.6 22.7

a Significantly different from group C-C. b Significantly different from group C-S.

183

tailed). Groups C-Ga.0 and C-G0.01 did not eat significantly more on this day than group C-S. Therefore withdrawal-induced reductions in intake were significantly less marked in animals treated with GR38032F at 0.1 mg/kg, but not in animals treated with higher or lower doses. On day 25 group C-G0a ate significantly more (P < 0.01, two-tailed) than group C-S, confirming the finding that GR38032F at 0.1 mg/kg significantly attenuated the effects of chlordiazepoxide withdrawal on food intake. No other between group comparisons were significant on this day. Collectively, the food intake data directly support the conclusions drawn from the body weight data. GR38032F attenuated chlordiazepoxide withdrawal, maximal effects being seen at the intermediate dose of 0.1 mg/kg. As noted above, one fairly trivial explanation of these data is that GR38032F stimulates food intake. To attempt to address this question, in our second study we examined the effects of GR38032F on 24 h food intake and body weight in animals treated b.i.d, with saline or GR38032F at doses between 0.01 and 1.0 mg/kg b.i.d. These data are shown in table 2 which shows the baseline pre-drug body weights and the post-drug 24 h food intakes of animals treated with saline or GR38032F at three different doses. These data were analysed with repeated measures ANOVAs. An effect of GR38032F on food intake or body weight would have been apparent as a significant TABLE 2 Mean+ S.D. (in parentheses) bodyweight and 24 h food intake data for each of the four experimental groups, as recorded on the baseline and the drug day. GR38032F m g / k g 0 Bodyweights Baseline 264 (26) Drug 266 (23) 24 h food intakes Baseline 17.5 (3.0) Drug 18.2 (1.3)

0.01

0.1

1.0

266 (19) 269 (21)

262 (23) 266 (22)

266 (26) 266 (26)

22.9 (3.8) 19.3 (2.1)

15.2 (1.7) 17.0 (3.8)

17.3 (1.7) 17.0 (1.6)

groups × days interaction. However, the two ANOVAs indicated that for neither measure was there a significant interaction (F (3,32) = 1.93, NS, for body weight; F (3,32)= 1.82, NS, for food intake), demonstrating that GR38032F administered with a dosing regime which attenuated signs of chlordiazepoxide withdrawal did not significantly effect either food intake or body weight.

4. Discussion Anorexia and weight loss were observed following chronic chlordiazepoxide treatment. Ondansetron had similar effects against these withdrawal signs, attenuating both at 0.1 mg/kg. Both signs peaked at similar times after cessation of chronic BZ treatment, on day 25 some 90 h after the last injection. In unpublished studies we have found that the withdrawal signs of weight loss and anorexia are highly correlated. It therefore seems likely that they are both induced by a common process which is modified by GR38032F. The data obtained in the second experiment suggest that the attenuation by ondansetron of withdrawal-induced anorexia and weight loss was not due to stimulation of appetite. Attenuation of withdrawal was seen at doses that had no effects on body weight or food intake in animals on ad lib food. However, the study of ondansetron's effects on food intake and body weight in animals on ad lib food is not an ideal 'control' experiment in this context. Ondansetron might restore food intake and body weight which is suppressed, rather than at baseline levels, as was the case in our animals experiencing BZ withdrawal. The ideal control experiment to eliminate the possibility that the reversal of the withdrawal signs we studied was due to stimulation of appetite would involve assessment of the effects of ondansetron on suppressed food intake and body weight. However, a major problem arises in trying to define what the appropriate control experiment should be in this context. We do not know what processes are involved in withdrawal-induced anorexia and weight loss. It is consequently impossible to define the ideal control experiment. If, for example, the weight loss/anorexia was due to subjects being

184 'ill' during withdrawal, then the appropriate control might be to examine effects of ondansetron on food intake/weight loss induced by a treatment believed to induce nausea or motion sickness. If, alternatively, the weight loss/anorexia was due to subjects being anxious during withdrawal, the appropriate control might be to examine effects of ondansetron on food intake/weight loss reduced by an anxiogenic stimulus. Alternatively, if the withdrawal signs were due to subjects being hypersensitive to the palatability of food, then the appropriate control might be to examine effects of ondansetron on food intake/weight loss suppressed by adding quinine to the diet. It is not possible to define which, if any, of these experiments would be the ideal control for the current study. Thus we enter a possible caveat about the adequacy of the control experiment we have described. However, current evidence suggests that 5-HT3 antagonists actually suppress, rather than facilitate, food intake. Van der Hoek and Cooper (in press) reported that ondansetron suppressed the intake of palatable foods, and Hrupka et al. (1989) reported that ICS 205-930 suppressed intake of a low protein diet. Since the reported effects of 5-HT3 antagonists on food intake are either suppressant or nonexistent (our data), it seems unlikely that the attenuation of BZ withdrawal signs noted was simply a consequence of a stimulant action of ondansetron on food intake. Further support for this hypothesis comes from evidence (Costall et al., 1989b) that ondansetron blocks anxiogenic behavioural signs of BZ withdrawal. It seems most unlikely that the ability of ondansetron to attenuate these different signs of BZ withdrawal could be related simply to stimulation of appetite, although it will be important in the future to assess effects of ondansetron on other BZ withdrawal signs, which do not involve appetitive tasks, to clarify the validity of this conclusion. The most effective dose of ondansetron against BZ withdrawal was 0.1 mg/kg, which inhibited the severity of withdrawal (fig. 2). The lower dose (0.01 mg/kg) reduced the duration of withdrawal without significantly m o d i f y i n g the peak withdrawal effect. The higher dose (1.0 m g / k g ) failed to modify withdrawal. Thus ondansetron

attenuated withdrawal with an inverted U-shaped dose-effect curve. A number of behavioural studies have reported that the anxiolytic actions of various 5-HT3 antagonists are absent at high doses (Jones et al., 1988; Costall et al., 1989a; Glenn and Green, 1989; Papp and Przegalinski, 1989). Thus our findings with high doses of GR38032F are in accord with other data obtained with 5-HT3 antagonists. The maximally effective dose of ondansetron against BZ withdrawal is a dose reported to have anxiolytic actions in animal tests of anxiety (Costall et al., 1989a, b; Tyers, 1989; Papp and Prezgalinski, 1989). Therefore it seems reasonable to expect that ondansetron could suppress BZ withdrawal at doses which might have clinically effective anxiolytic actions. Our data essentially support the conclusions of Costall et al. (1989b), who reported that GR38032F attenuated behavioural signs of BZ withdrawal considered indicative of anxiogenesis. Costall et al. (1989b) also reported that ondansetron attenuated BZ withdrawal-induced weight loss. Since the earlier abstract reports on these studies (Oakley et al., 1988; Costall et al., 1988) did not refer specifically to attenuation of BZ withdrawal-induced weight loss, our studies were conducted without knowledge of these findings. They can therefore be considered independent replications of the findings of Costall et al. (1989b). Moreover, although our data support the main conclusions of Costall et al. (1989b), they extend these findings in a number of ways. Firstly, they demonstrate that ondansetron attenuates signs of withdrawal from chlordiazepoxide as well as from diazepam. Secondly, they demonstrate that ondansetron attenuates withdrawal-induced anorexia, a sign not studied by Costall et al. (1989b). Finally, and clearly most importantly, they demonstrate that ondansetron attenuates BZ withdrawal with an inverted U-shaped dose-effect curve. Costall et al. (1989b) did not assess the efficacy of different doses of ondansetron against any specific withdrawal sign. This finding obviously may be important for the therapeutic administration of this agent. Although our data support the basic findings of Costall et al. (1989b), they also differ from them in one potentially important way. Costall et al.

185

(1989b) reported that ondansetron blocked BZ withdrawal fully. Indeed, ondansetron did not simply attenuate reductions in social interaction induced by BZ withdrawal, it increased social interaction to above control levels, so that the drug had an anxiolytic action in subjects experiencing withdrawal. However, we only detected attenuation, not elimination, of BZ withdrawal. There are a number of explanations for this apparen( discrepancy between the findings of Costall et al. (1989b) and our data. Firstly, the treatment regime we used may not have been optimal for blocking BZ withdrawal. Secondly, the severity of withdrawal in our study was greater than that in the study of Costall et al. (1989b), in which the weight loss reported in rats after 10 days of b.i.d. diazepam treatment was ca. 2%, in comparison to the 5.6% weight loss noted in our study, after more prolonged (21 days) b.i.d. BZ treatment. The ability of ondansetron to attenuate BZ withdrawal may be reduced as withdrawal signs become more severe with prolonged dosing. Finally, Costall et al. (1989b) reported that the behavioural signs of BZ withdrawal that they observed peaked abruptly, 8 h after cessation of diazepam treatment, in contrast to our data which indicated that peak withdrawal was only seen 4 days after cessation of chlordiazepoxide treatment, suggesting that the different withdrawal signs might be caused by different mechanisms. The BZ withdrawal syndrome may be mediated by multiple neurotransmitter systems, with different systems mediating different withdrawal signs (Baldwin et al., 1989). For example, withdrawal-induced hyperactivity may be mediated by a2-adrenoreceptor systems (Kunchandy and Kulkarni, 1986), although such systems are not involved in withdrawal-induced anxiogenesis (Baldwin et al., 1989). If the BZ withdrawal syndrome is a heterogeneous phenomenon, it is possible that ondansetron may be more effective against some withdrawal signs than against others. Thus it will be important in the future to assess the effects of ondansetron against a range of withdrawal signs, to determine which aspects of the withdrawal syndrome are most effectively blocked by ondansetron. In summary, although there is emerging evidence that ondansetron can alleviate various BZ

withdrawal signs, the generality of these findings needs to be established more clearly. It is not at present known whether, if at all, ondansetron's ability to alleviate BZ withdrawal is affected by the treatment regime utilised, by the withdrawal sign(s) studied, or by the severity of withdrawal experienced. Experimental studies are also required to determine whether different types of 5-HT 3 antagonists can also attenuate BZ withdrawal signs. These important questions will only be answered by further parametric studies on the effects of 5-HT3 antagonists on BZ withdrawal.

Acknowledgements The work reported here was supported by funds from the Liverpool University Research Development Fund. We are greatly indebted to Glaxo Group Research for generous supplies of ondansetron.

References Baldwin, H.A., P.K. Hitchcott and S.E. File, 1989, Evidence that the increased anxiety in the elevated plus-maze during chlordiazepoxide withdrawal is not due to enhanced noradrenergic activity, Pharmacol. Biochem. Behav. 34, 931. Baldwin, H.A. and S.E. File, 1988, Reversal of increased anxiety during benzodiazepine withdrawal: Evidence for an anxiogenic endogenous ligand for the benzodiazepine receptor, Brain Res. Bull. 20, 603. Baldwin, H.A. and S.E. File, 1989, Flumazenil prevents the development of chlordiazepoxide withdrawal in rats tested in the social interaction test, Psychopharmacology 97, 424. Blundell, J.E., 1984, Serotonin and appetite, Nenropharmacology 23, 1537. Costall, B., A.M. Domeney, B.J. Jones, M.E. Kelly, P.A. Gerrard, R.J. Naylor and M.B. Tyers, 1988, Influence of GR38032F on the behavioural consequences of ceasing sub-chronic treatment with drugs of abuse, Br. J. Pharmacol. 95, 764P. Costall, B., A.M. Domeney, P.A. Gerrard, B.J. Jones, M.E. Kelley, N.R. Oakley and M.B. Tyers, 1989a, The anxiolytic activities of the 5-HT3 receptor antagonists GR38032F, ICS 205-930 and BRL 43964, in: Behavioural Pharmacology of 5-HT, eds. P. Bevan, A.R. Cools and T. Archer (Lawrence Erlbaum Associates, Hillsdale, New Jersey) p. 383. Costall, B., B.J. Jones, M.E. Kelly, R.J. Naylor, N.R. Oakley, E.S. Onaivi and M.B. Tyers, 1989b, The effects of ondansetron (GR38032F) in rats and mice treated subchronically with diazepam, Pharmacol. Biochem. Behav. 34, 769. Chopin, P. and M. Briley, 1987, Animals models of anxiety: The effect of compounds that modify 5-HT neurotransmission, Trends Pharmacol. Sci. 8, 383.

186 File, S.E. and A.L. Johnston, 1989, Lack of effect of 5-HT3 antagonists in the social interaction and elevated plus-maze tests of anxiety in the rat, Psychopharmacology 99, 248. Gailagher, D.W., K. Heninger and G. Heninger, 1986, Periodic benzodiazepine antagonist administration prevents benzodiazepine withdrawn symptoms in primates, European J. Pharmacol. 132, 31. Glenn, B. and S.E. Green, 1989, Anxiolytic profile of GR38032F in the potentiated startle paradigm, Behav. Pharmacol. 1, 91. Goa, K.L. and A. Ward, 1986, Buspirone: A preliminary review of its pharmacological properties and therapeutic efficacy as an anxiolytic, Drugs 32, 114. Harto, N.E., R.J. Braconnier, K.F. Spera and E.C. Dessain, 1988, Clinical profile of gepirone, a nonbenzodiazepine anxiolytic, Psychopharmacol. Bull. 24, 154. Hrupka, B.J., J.L. Beverly, D.W. Gietzen, V.A. Hammer and Q.R. Rodgers, 1989, The effect of centrally administered ICS-205-930 on food intake of rats fed an amino acid imbalanced diet, Soc. Neurosci. Abstr. 15 (Abstract 246.18), 617P. Jones, B.J., B. Costall, A.M. Domeney, M.E. Kelly, R. Naylor and M.B. Tyers, 1988, The potentiai anxiolytic activity of GR38032F, a 5-HT3 antagonist, Br. J. Pharmacol. 93, 985. Kuchandy, J. and S.K. Kulkarni, 1986, Reversal by alpha-2antagonists of diazepam withdrawal hyperactivity in rats, Psychopharmacoiogy 90, 198. Oaldey, N.R., B.J. Jones and M.B. Tyers, 1988, Tolerance and withdrawal studies with diazepam and GR38032F in the rat, Br. J. Pharmacol. 95, 905P.

Papp, M. and E. Przegalinski, 1989, The 5-HT3 receptor antagonists ICS-205-930 and GR38032F, putative anxiolytic drugs, differ from diazepam in their pharmacological profile, J. Psychopharmacol. 3, 14. Sannerud, C.A., J.M. Cook and R.R. Griffiths, 1989, Behavioral differentiation of benzodiazepine ligands after repeated administration in baboons, European J. Pharmacol. 167, 333. Schweizer, E. and K. Rickels, 1986, Failure of buspirone to manage benzodiazepine withdrawal, Am. J. Psychiat. 143, 1590. Schweizer, E., K. Rickels and I. Lucki, 1986, Resistance to the anti-anxiety effects of buspirone in patients with a history of benzodiazepine use, N. Engl. J. Med. 314, 719. Simpson, M.A., H.S. Schoeman and B.J.C. Allman, 1989, A profile of buspirone in the treatment of anxiety in general practice patients, Cuff. Ther. Res. 46, 980. Tyers, M.B., 1989, A review of the evidence supporting the anxiolytic potential of 5-HT3 receptor antagonists, in: Behavioural Pharmacology of 5-HT, eds. P. Bevan, A.R. Cools and T. Archer (Lawrence Erlbanm Associates, Hillsdale, New Yersey) p. 353. Van der Hoek, G.A. and S.J. Cooper, Effects of ondansetron on food intake in rats, Br. J. Pharmacol. (in press) (Abstract). Woods, J.H., J.L. Katz and G. Winger, 1987, Abuse liability of benzodiazepines, Pharmacol. Rev. 39, 251.