Introduction

Top Abstract Introduction Methods Results Discussion References

Zolpidem (AMBIEN®) is an imidazopyridine hypnotic drug that acts at the BZ recognition site on the GABA_A receptor/chloride channel complex. The BZ site recognized by zolpidem is distinct from the site recognized by other positive allosteric modulators of GABA function, e.g., pentobarbital (see Lüddens et al., 1995; Sanger et al., 1994, for review). Moreover, the receptor binding profile of zolpidem is distinct from the profile of classic BZ agonists in several respects. For example, binding sites for zolpidem show a different pattern of distribution in the CNS, being in greater proportion in the cerebellum than in other regions (Benavides et al., 1988, 1992, 1993; Dennis et al., 1988). This CNS distribution is consistent with zolpidem being a selective ligand for the BZ₁ subtype of the BZ receptor (see Lüddens et al., 1995; Sanger et al., 1994, for review). Consistent with this notion, molecular biological studies have shown that the affinity of zolpidem for binding to reconstituted receptors, in contrast to typical BZs, depends on expression of  subunits of the GABA_A receptor/chloride channel complex. Thus, zolpidem binds with highest affinity to receptors expressing ₁ subunits (K_i = 15-25 nM), which are thought to be associated with the BZ₁ receptor subtype (Hadingham et al., 1993; Pritchett and Seeburg, 1990). In contrast, zolpidem binds with intermediate to low affinity (K_i values ranging from 350 to >15,000 nM) to receptors expressing ₂, ₃ and ₅ subunits, the latter three associated with BZ₂ receptor subtypes (Hadingham et al., 1993; Pritchett and Seeburg, 1990). Interestingly, recent in vivo binding data suggest a less selective profile in terms of CNS distribution for zolpidem (Byrnes et al., 1992; Schmid et al., 1995). Moreover, the in vivo binding characteristics of zolpidem in baboon brain are different from in vitro binding characteristics of zolpidem in rat brain, with less evidence of heterogeneity for zolpidem binding occurring in various brain structures in baboons (Schmid et al., 1995). Resolution of these issues may occur with the characterization of as yet undiscovered BZ receptor subtypes. Nonetheless, zolpidem's neurochemical profile remains unique among the BZ ligands.

The profile of behavioral effects produced by zolpidem administration in rodents also is unique among BZ ligands. Notably, the discriminative stimulus effects of zolpidem are distinguishable from typical BZs in rats (Depoortere et al., 1986; Sanger, 1987; Sanger et al., 1987; Sanger and Zivkovic, 1986, 1987). For example, in rats trained to discriminate 5 mg/kg chlordiazepoxide from saline, a high dose of zolpidem (3 mg/kg) produced only partial chlordiazepoxide-appropriate responding (i.e., greater than 20%, but less than 80%, drug-appropriate responding) and often suppressed responding completely (Depoortere et al., 1986; Sanger et al., 1987). In contrast, both chlordiazepoxide and triazolam occasioned full chlordiazepoxide-appropriate responding (i.e., greater than 80% drug-appropriate responding) in all rats tested (Depoortere et al., 1986; Sanger et al., 1987). Conversely, when zolpidem (2.0 mg/kg) was trained as a discriminative stimulus, triazolam and chlordiazepoxide occasioned only partial drug-appropriate responding, even at doses that significantly suppressed responding (Sanger and Zivkovic, 1986). In further studies, the partial agonists CGS 9896 and ZK 91296 antagonized the discriminative stimulus effects of zolpidem, but not chlordiazepoxide (Sanger and Zivkovic, 1987), whereas the mixed agonist-antagonists Ro 16-6028 and Ro 17-1812 occasioned full drug-appropriate responding in rats trained to discriminate chlordiazepoxide, but not zolpidem (Sanger, 1987). More recently, zolpidem was shown to occasion full drug-appropriate responding at a low (0.32 mg/kg) but not high (3.2 mg/kg) dose of midazolam in rats trained in a three-choice procedure (Sannerud and Ator, 1995b). Chlordiazepoxide also occasioned full drug-appropriate responding at the low, but not high, midazolam training dose, whereas triazolam occasioned full drug-appropriate responding at both training doses (Sannerud and Ator, 1995b). Taken together, these results suggest that the profile of discriminative stimulus effects of zolpidem often is distinguishable from BZs such as triazolam and chlordiazepoxide, and further raise the possibility of a unique profile of discriminative stimulus effects for ligands with selectivity for the BZ₁ receptor subtype.

In contrast to results obtained from drug discrimination studies in rats, results from a recent study by Griffiths et al. (1992), with nonhuman primates as subjects, suggested that the discriminative stimulus effects of zolpidem were similar to typical BZs. Zolpidem occasioned 80% drug-appropriate responding in four of five baboons trained to discriminate lorazepam from a no-drug condition (Griffiths et al., 1992). In addition, zolpidem suppressed response rate in only one of the four baboons at the highest dose tested (Griffiths et al., 1992). In previous work by these authors (Ator and Griffiths, 1989), typical BZs (e.g., triazolam), with few exceptions, have been shown to occasion full drug-appropriate responding in lorazepam-trained baboons and to have relatively few effects on rate of responding. These results with baboons suggest the possibility of a species difference between rats and baboons, also corroborated with data from other behavioral paradigms (e.g., tolerance and withdrawal studies; cf. Griffiths et al., 1992; Sanger et al., 1994).

In addition to the results in lorazepam-trained baboons, zolpidem occasioned full drug-appropriate responding in baboons trained to discriminate the barbiturate pentobarbital from a no-drug condition (Griffiths et al., 1992). In a previous study (Sanger and Zivkovic, 1986), pentobarbital produced only partial drug-appropriate responding in zolpidem-trained rats. As mentioned, it has generally been found that BZs occasion pentobarbital-appropriate responding up to 100% in a dose-dependent manner (Ator and Griffiths, 1983; Herling et al., 1980; Jarbe, 1976; Nader et al., 1991; Overton, 1976; Winger and Herling, 1982; Woolverton and Nader, 1995). Because of the pharmacological selectivity of drug discrimination procedures, these results suggest a common mechanism of action of BZs and pentobarbital (Ator and Griffiths, 1989). Therefore, the partial pentobarbital-appropriate responding observed in zolpidem-trained rats further suggests a mechanism of action different from typical BZs. In addition, this finding raises the possibility of a species difference between rodents and nonhuman primates in the ability of zolpidem to occasion drug-appropriate responding in subjects trained to discriminate a classic sedative/hypnotic agent, such as pentobarbital, a drug generally attributed with having discriminative stimulus effects that are nonspecific with respect to sedative/hypnotic agents (Ator and Griffiths, 1989).

The purpose of the present study was to assess whether pentobarbital-like discriminative stimulus effects of zolpidem vary as a function of species. To do so, the discriminative stimulus effects of zolpidem were compared with a triazolobenzodiazepine, triazolam, in pentobarbital-trained rhesus monkeys and rats. This study is part of a series of experiments designed to make cross-species comparisons with human subjects trained to discriminate pentobarbital from placebo (Rush et al., 1997). Thus, zolpidem was tested in this series because, as noted above, its discriminative stimulus effects in pentobarbital-trained subjects may vary as a function of species. Triazolam was included as a positive control, because previous drug discrimination experiments conducted in nonhuman primates and rats have shown that triazolam occasions pentobarbital-appropriate responding (Ator and Griffiths, 1989). Various doses of zolpidem and triazolam were tested in rhesus monkeys and rats involved in ongoing experiments in our laboratory, with two different experimental protocols for two-lever drug discrimination: FR 1 discrete-trials shock avoidance (rhesus monkeys) and FR 10 food reinforcement (rats). Finally, the possibility that differences in the discriminative stimulus effects of zolpidem between rhesus monkeys and rats were caused by procedural differences was explored by varying parameters of the protocol used for rats. Thus, the discriminative stimulus effects of zolpidem in pentobarbital-trained rats were assessed after varying pretreatment times, after varying the schedule of reinforcement, and after varying the route of administration. These manipulations were conducted to equate various aspects of the two protocols used.

	Methods

Top Abstract Introduction Methods Results Discussion References

Rhesus Monkeys

Subjects. The subjects were four adult rhesus monkeys, three males (8236, 8106, 8814) and one female (7976), weighing between 7.0 and 9.8 kg. At the time of the study, the monkeys had been subjects in the present drug discrimination paradigm for approximately 4 years. During that time, they had been tested with various BZs and psychomotor stimulants. The monkeys were housed individually in stainless steel cages in which water was continuously available. They were fed 120 to 150 g of monkey chow after each session and were given a chewable multiple vitamin tablet, formulated for children, 3 days/week. A 12-hr light/dark cycle was maintained, and all experimental sessions were conducted during the light phase of the cycle.

Apparatus. During experimental sessions each monkey was seated in a restraining chair (Plas Labs, Lansing, MI) and placed in a wooden cubicle (175 cm high × 85 cm wide × 65 cm deep) containing two response levers mounted 100 cm above the floor with a distance between the levers of 25 cm, from center to center of the levers. A minimal downward force of 0.30 N was required to activate each lever. A 40-W white house light was mounted on the ceiling. The monkey's feet were placed into shoes, the bottoms of which were fitted with brass plates that could deliver electric shocks using a shock generator (Model SG-903, BRS/LVE, Beltsville, MD). Programming and recording of experimental events were accomplished with an Aim 65 microprocessor located in an adjacent room.

Discrimination training. The monkeys had been trained previously to discriminate pentobarbital from saline in a two-lever, discrete-trials shock avoidance procedure (FR 1). Each monkey was placed in the chair, given an intragastric infusion (via nasogastric tube) of saline or 10 mg/kg pentobarbital, then returned to the home cage. Fifty-five minutes later the monkey was seated again in the chair and placed in the chamber. After 5 min, the house lights and lever lights were illuminated (trial) and responding on one lever (correct lever) avoided electric shock (avoidance response) and extinguished the lights. Responding on the incorrect lever started a 2-s changeover delay during which correct lever responding had no consequence. If a correct lever response was not made within 5 s of onset of the lights, an electric shock (250 ms duration, 7 mA intensity) was delivered; if a correct response was made within 2 s after the first shock (escape response), the trial was terminated. Otherwise, a second shock was delivered and the trial ended automatically. Two consecutive trials in which a monkey failed to make an avoidance or escape response automatically ended the session. Trials were separated by a 30-s timeout, during which all lights in the chamber were extinguished and responding was recorded but had no programmed consequences. The session lasted 30 trials or 20 min, whichever came first. The correct lever was determined by the infusion that was administered before the session. For two monkeys, the left lever was correct after drug infusion and the right lever was correct after saline infusions. This condition was reversed for the other monkeys. The sequence of daily (5 days/week) sessions was SDDSS, DSSDD (S, saline pretreatment; D, drug pretreatment).

Testing. When the first response of the session and at least 90% of the total trials were completed on the correct lever for at least seven of eight consecutive sessions, test sessions were added to this sequence such that the first test session each week was preceded by two training sessions, one with saline and one with drug pretreatment and the second test session of the week was preceded by either saline or drug pretreatment (i.e., SDTST, DSTDT, where T denotes a test session). Test sessions were identical with training sessions except that the drug pretreatment was novel and a response on either lever avoided or escaped shock delivery. In the event that either criterion for stimulus control was not met during the training sessions, the training sequence continued without test sessions until the criteria were met for seven of eight sessions.

Drugs. Sodium pentobarbital (Sigma Chemical Co., St. Louis, MO) was diluted with 0.9% saline to a final concentration of 40 mg/ml, from a stock solution (400 mg/ml) with a vehicle of propylene glycol, 95% ethanol and water (4:1:5). Zolpidem tartrate (NIDA, Baltimore, MD) was mixed in sterile water, and triazolam (NIDA, Baltimore, MD) was prepared in a vehicle of Emulphor EL-620 (Alkamuls EL-620, Rhone Poulenc, Cranbury, NJ):95% ethanol (1:1). Test solutions/suspensions were prepared immediately before the session in which they were tested. Doses were varied by varying the volume of the standard solution. Pentobarbital and BZs were given intragastrically via nasogastric tube followed by a flush with 1 to 2 ml of saline to clear the tube.

Data analysis. The percentage of the total trials completed on the pentobarbital-appropriate lever and the mean latency/trial (mean time between the onset of a trial and a lever press, averaged across trials) were calculated for test sessions for each subject. Data are presented as individual subjects and mean percent drug-appropriate trials for each drug. If drug doses were tested twice, data were analyzed and presented as the average of the two determinations. Full pentobarbital-appropriate responding was concluded if responding on drug-appropriate trials was 80% or more of total responding. Saline-appropriate responding was concluded if responding on drug-appropriate trials was 20% or less of total responding. Between 20 and 80% was considered partial drug-appropriate responding. To assess reliable dose effects on latency, tests were planned a priori to compare the latency at each dose with the saline test session latencies. Dunnett's tests were used for these comparisons. The reported statistics are q values, which are analogous to t values, but with use of the within-subject residual error value as the error term of the ratio [degrees of freedom = (treatment level  1) × (number of subjects  1)]. The alpha level for these tests was P  .05.

Rats

Subjects. Twelve naive adult male Sprague-Dawley rats (Harlan Industries, Indianapolis, IN), weighing between 275 and 300 g before reducing their food, were used as subjects. The rats were singly housed, and acclimated to the laboratory for approximately 1 week, then food was reduced until the rats were maintained at approximately 80% of their free-feeding weights. Once training started, the rats were fed 15 g of food after an experimental session and on days when no sessions were conducted. Over the course of the experiment, the weights of the rats gradually increased such that weights toward the end of the experiments were approximately 130% of their initial free-feeding weights. Water was available continuously in the individual home cages. A 12-hr light/dark cycle was maintained, and all experimental sessions were conducted during the light phase of the cycle. The sessions were conducted at noon each day.

Apparatus. Experimental sessions were conducted in eight identical rat operant chambers (Gerbrands, Arlington, MA). In each chamber, two response levers were mounted on one wall (12.5 cm apart, center to center; 10 cm above the floor) and a food receptacle was located between them. A minimal downward force of 0.30 N was required to activate each lever. Each chamber was illuminated at the onset of the session by a single 6-W light located on the wall opposite the two levers and by a white light located above each lever. Extraneous noise was diminished by enclosing each chamber in an insulated box and by operating a ventilation fan mounted on the outside of each box. Data collection and experimental events were controlled by a Macintosh II microcomputer, with custom software and interfaces.

Discrimination training. Rats initially were trained to press a lever once (i.e., FR 1) to obtain a 40-mg food pellet (Noyes Co., Lancaster, NH) after an injection of 8.0 mg/kg i.p. pentobarbital or 1.0 ml/kg saline. For six rats, food pellets were delivered only after responding on the right lever in the presence of pentobarbital; this condition was reversed for the other six rats. Drug or saline were administered 15 min before the 15-min session. Ten minutes after the injections, the rats were placed in the chamber for the remaining 5 min of the pretreatment time, during which the lever lights were extinguished (timeout). The trial started when the lever lights were illuminated. There was no timeout after pellet delivery during the trial. Lever pressing by six rats was shaped initially by differentially reinforcing successive approximations after pentobarbital injection; lever pressing by the remaining six was shaped after saline injection. As soon as FR 1 responding was acquired, or the rat had received three consecutive daily injections of drug or saline, the rat was placed on a double alternation schedule and the FR gradually increased to FR 10. Responding on the inappropriate lever reset the response requirement on the appropriate lever. Once FR 10 responding was obtained, rats were trained 5 days a week with a daily injection schedule of SDDSS, DSSDD (S, saline; D, pentobarbital). The criteria for stimulus control were: 1) 90% of responding in a session on the appropriate lever for 7 of 8 days, and 2) 90% of responding on the appropriate lever before the first food pellet delivery for 7 of 8 days.

Testing. Once a rat reached criteria, test sessions were run in which 10 responses on either lever resulted in food pellet delivery. Tests were conducted twice a week according to the schedule SDTST, DSTDT (T, test); if responding was not on criteria during the training sessions, the rats were returned to training sessions with no test sessions until criteria were again met for seven of eight sessions. The first test session consisted of a pentobarbital training dose test (8.0 mg/kg pentobarbital i.p.), and the second test session consisted of a saline test (1 ml/kg i.p.). Next, four doses of pentobarbital (1.0, 2.0, 4.0, 16 mg/kg i.p., 15-min pretreatment) were tested in nonsystematic order. The pentobarbital dose-response function was determined first in all rats.

Zolpidem time-course study. After testing with triazolam and zolpidem as described above, the four doses of zolpidem were retested at either 5 or 45 min before the session. For the 5-min pretreatment, the rats were injected with a zolpidem dose and placed immediately into the chambers for the 5-min timeout, followed by the 15-min session. For the 45-min pretreatment, the rats were injected and returned to the home cage for 40 min, after which they were placed in the chamber for the 5-min timeout and 15-min session. Thus, the time course consisted of 5-, 15- (obtained previously) and 45-min pretreatment times. The doses for the 5-min pretreatment time were determined twice in eight rats, whereas the doses at the 45-min pretreatment time were determined once in four rats.

Zolpidem FR 1 study. Five rats, four of which were tested in the previous conditions, were tested with pentobarbital (8.0 mg/kg i.p.), saline and zolpidem (1.0, 2.0 and 4.0 mg/kg) under a FR 1 schedule. These tests were conducted with a 15-min pretreatment time (pentobarbital, saline) or a 5-min pretreatment time (zolpidem, saline) as described above, except that the FR was lowered from 10 to 1 during the test sessions only. Each zolpidem dose was determined twice.

Zolpidem i.g. study. Three doses of pentobarbital (4.0, 8.0 and 16 mg/kg) and four doses of zolpidem (1.0, 2.0, 4.0 and 6.0 mg/kg), as well as saline, were tested via the i.g. route in the four rats not tested with zolpidem or triazolam in the initial FR 10 studies (one rat was tested in the FR 1 study), as well as rats R1 to R4 that were tested with zolpidem and triazolam. The i.g. doses were administered by gavage, by use of a stainless steel gavage tube, 15 min before the session. Thus, the rats were given the i.g. dose, returned to the home cage, then placed in the chambers for the 5-min timeout followed by the 15-min session. Each dose of pentobarbital and zolpidem was determined twice.

Drugs. Pentobarbital was diluted from Nembutal® (50 mg/ml pentobarbital, 40% propylene glycol, 10% ethanol, 50% sterile water; Abbott Laboratories, Chicago, IL) in saline every 2 days, and injected at a volume of 1.0 ml/kg for both i.p. and i.g. doses. Triazolam was suspended each day of testing in a Tween 80 solution (10% Tween 80, 90% sterile, distilled water). Zolpidem tartrate was dissolved each day of testing in saline, and injected in a volume of 2.0 ml/kg (i.p.) or 3.0 ml/kg (i.g. and 6.0 mg/kg i.p).

Data analysis. For all drug doses that were tested twice, data were analyzed and presented as the average of the two determinations. Drug-appropriate responding was expressed as the percent of responding on the drug-appropriate lever out of the total number of responses. Drug lever responding was calculated in the FR 10 studies only if the rat completed a response requirement. Rate of responding was calculated by dividing the total number of responses for a session by the session time of 900 s. These data were calculated even if a rat did not complete a response requirement. Response rate was not calculated in the FR 1 study; instead total number of responses in the session were calculated (latency data were not available).

Dose (milligrams per body surface area) and Relative Dose Calculations

To compare doses across species, the doses based on body weight were converted to doses based on body surface area, according to Dews (1976). Area (A) was estimated as the 2/3 power of weight, and relative doses were calculated by dividing dose in milligrams per kilogram by dose in mg/kg^2/3, according to the formula: relative dose = dose/(wt·dose/kg^2/3), given A  kg^2/3. In addition, pentobarbital, zolpidem and triazolam doses from the companion manuscript that used humans as subjects (Rush et al., 1997) were converted to relative doses for comparison with rhesus monkeys and rats.

	Results

Top Abstract Introduction Methods Results Discussion References

Rhesus Monkeys

Pentobarbital. As can be seen in the top panel of figure 1, pentobarbital produced a dose-dependent increase in drug-appropriate responding (percent drug-appropriate trials) in the four monkeys. That is, percent drug-appropriate trials was at or near zero percent for the four monkeys when tested with saline or 3.0 mg/kg pentobarbital. The mean percent drug-appropriate trials was 50% at 5.6 mg/kg pentobarbital, reflecting 50% responding in two monkeys, 100% responding in one monkey, and 0% responding in the other monkey. The 50% responding in the two monkeys reflected an average of 0 and 100% responding in double determinations of this dose. At the training dose of 10 mg/kg pentobarbital, all monkeys responded at or near 100% drug-appropriate trials.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   The effects of pentobarbital in monkeys trained to discriminate 10 mg/kg pentobarbital from saline. (Top panel) The percentage of trials completed on the pentobarbital-appropriate lever during test sessions in which responding on either lever avoided electric shock. (Bottom panel) The average latency (seconds) to respond in test sessions. (Abscissae) Dose of pentobarbital in mg/kg given intragastrically via nasogastric tube. Each point represents data from individual monkeys identified in the symbol key. Solid lines through the points represent lines connecting the group means. Each point represents the mean of two determinations, except for monkeys 7976 and 8106 in which each point represents a single determination.

Zolpidem. As can be seen in the left top panel of figure 2, zolpidem produced a dose-dependent increase in drug-appropriate responding (percent drug-appropriate trials) in the four monkeys. As with pentobarbital, some intermediate drug-appropriate responding after zolpidem represented twice-determined points in which one determination was at or near 0% and the second was at or near 100% (3.0 mg/kg zolpidem in subject 8814, 17 mg/kg zolpidem in subject 7976). Intermediate responding observed for monkey 8814 at 10 mg/kg zolpidem reflected intermediate responding on the levers during the test sessions for both determinations, rather than the average of near 0% and near 100% drug-appropriate responding. Compared with pentobarbital, there was more variation in the dose range of zolpidem that produced dose-dependent drug-appropriate responding. Thus, in monkey 8106, zolpidem occasioned 0 and 100% drug-appropriate responding at doses of 0.30 and 1.0 mg/kg, respectively. In contrast, 80% drug-appropriate responding was observed at 17 mg/kg in monkey 8814 and only at 30 mg/kg in the remaining two monkeys. Because monkey 8106 showed this 30-fold difference in doses of zolpidem that occasioned full drug-appropriate responding, this monkey was omitted from analyses of the zolpidem data.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   The effects of zolpidem (left panels) and triazolam (right panels) in monkeys trained to discriminate 10 mg/kg pentobarbital from saline. (Top panels) The percentage of trials completed on the pentobarbital-appropriate lever during test sessions in which responding on either lever avoided electric shock. (Bottom panels) The average latency (seconds) to respond in test sessions. (Abscissae) Dose of zolpidem or triazolam in mg/kg given intragastrically via nasogastric tube. Each point represents data from individual monkeys identified in the symbol key. Solid lines through the points represent lines connecting the group means. The long-dashed line in the top and bottom left panels represents data from one monkey that were not included in the group mean calculation. The horizontal short-dashed lines represent the range of pentobarbital-appropriate responding (top panels) and latencies (bottom panels) after saline tests. Each point represents the mean of two determinations.

Triazolam. As can be seen in the right top panel of figure 2, triazolam produced a dose-dependent increase in drug-appropriate responding (percent drug-appropriate trials) in the four monkeys. As with pentobarbital and zolpidem, some intermediate responding at or near 50% drug-appropriate trials represented the average of 0% or near 0% and 100% or near 100% drug-appropriate trials after double determinations. Intermediate responding that reflected intermediate distribution of completed trials occurred for subjects 8106, 7976 and 8814 at 0.10 mg/kg of triazolam. Compared with zolpidem, there was less variation in the dose range of triazolam that produced dose-dependent pentobarbital-appropriate responding. Thus, in the monkey in which zolpidem occasioned 100% pentobarbital-appropriate responding at a 30-fold lower dose than the other monkeys (monkey 8106), triazolam occasioned 40 and 96% drug-appropriate responding at doses of 0.1 and 0.3 mg/kg, respectively. This pattern of results was similar to the other three monkeys, all of whom showed 80% drug-appropriate responding at 0.3 mg/kg triazolam only.

Rats

Pentobarbital. As can be seen in the top panel of figure 3, pentobarbital produced a dose-dependent increase in drug-appropriate responding (percent drug lever) in rats R1 to R8. Pentobarbital dose-response data in rats R9 to R12 under the FR 10/i.p. condition were not included in this study, but did not differ from the results of rats R1 to R8. For the pentobarbital dose-response function, percent drug lever responding was at or near zero percent for all rats when tested with saline, 1.0 and 2.0 mg/kg pentobarbital; but increased to 80 to 100% at 8.0 and 16 mg/kg pentobarbital. Intermediate responding was observed at 4.0 mg/kg (mean percent drug lever = 40%), largely because of five rats responding predominantly on the saline lever and three rats responding predominantly on the drug lever.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   The effects of pentobarbital in rats (n = 8) trained to discriminate 8.0 mg/kg pentobarbital from saline. (Top panel) The percentage of responses completed on the pentobarbital-appropriate lever during test sessions in which responding on either lever resulted in food pellet delivery. (Bottom panel) The average response rate (responses/second) of responding on either lever in test sessions. (Abscissae) Dose of pentobarbital in mg/kg given via the i.p. route. Each point represents data from individual rats identified in the symbol key. Solid lines through the points represent lines connecting the group means. Each point represents the mean of two determinations.

Zolpidem. As can be seen in the three top panels of figure 4, zolpidem produced inconsistent drug-appropriate responding (percent drug lever) at the 5-min (top left panel) and 15-min (top middle panel) pretreatment times, and no drug-appropriate responding at the 45-min pretreatment time (top right panel). Drug-appropriate responding generally was not dose-dependent and showed considerable variability among subjects. For example, at the 15-min pretreatment time, rat R3 (open triangles) responded exclusively on the saline lever at each dose tested, whereas rat R6 (filled squares) responded exclusively on the drug lever at each dose. Other rats demonstrated no pentobarbital-appropriate responding at some doses and 80% pentobarbital-appropriate responding at other doses; but, in general, no consistent effects were observed as a function of zolpidem dose. The mean percent drug lever values at the 5- and 15-min pretreatment times ranged from 13% to 46%, whereas drug lever responding at all doses in the four rats tested was at 0% at the 45-min pretreatment time.

View larger version (24K): [in this window] [in a new window]   Fig. 4.   The effects of zolpidem in rats trained to discriminate 8.0 mg/kg pentobarbital from saline. (Top panels) The percentage of responses completed on the pentobarbital-appropriate lever during test sessions in which responding on either lever resulted in food pellet delivery. (Bottom panels) The average response rate (responses/second) of responding on either lever in test sessions. (Left panels) Discriminative stimulus and rate effects of zolpidem given 5 min before the session. (Middle panels) Discriminative stimulus and rate effects of zolpidem given 15 min before the session. (Right panels) Discriminative stimulus and rate effects of zolpidem given 45 min before the session. (Abscissae) Dose of zolpidem in mg/kg given via the i.p. route. Each point represents data from individual rats identified in the symbol key. Solid lines through the points represent lines connecting the group means. Eight rats were tested at the 5- and 15-min pretreatment times, doses in which symbols are missing indicate that the rat did not complete a FR (all rats were included in the rate calculations). Four rats were tested at the 45-min pretreatment time. For the 5- and 15-min pretreatment times, each point represents the mean of two determinations. Each point at the 45-min pretreatment time represents a single determination.

Triazolam. As can be seen in the top panel of figure 5, triazolam produced a dose-dependent increase in drug-appropriate responding (percent drug lever) in rats R1 to R8. That is, percent drug lever responding was at or near zero percent for all rats when tested with vehicle and 0.025 mg/kg triazolam, but increased to 70 to 100% at 0.20 mg/kg triazolam, with all but one rat (R8) showing 80% pentobarbital-appropriate responding at the highest dose. Intermediate responding was observed at 0.050 mg/kg because of four rats responding predominantly on the saline lever and four rats responding predominantly on the drug lever. At 0.10 mg/kg, all rats showed full drug-appropriate responding except rat R7, which responded entirely on the saline lever.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   The effects of triazolam in rats (n = 8) trained to discriminate 8.0 mg/kg pentobarbital from saline. (Top panel) The percentage of responses completed on the pentobarbital-appropriate lever during test sessions in which responding on either lever resulted in food pellet delivery. (Bottom panel) The average response rate (responses/second) of responding on either lever in test sessions. (Abscissae) Dose of triazolam in mg/kg given via the i.p. route. Each point represents data from individual rats identified in the symbol key. Solid lines through the points represent lines connecting the group means. Each point represents the mean of two determinations.

Pentobarbital and zolpidem, FR 1. As can be seen in table 1, 8.0 mg/kg pentobarbital occasioned 100% drug-appropriate responding (percent drug lever), when tested under a FR 1 schedule of reinforcement. Saline occasioned 0% drug-appropriate responding when administered at the same pretreatment time as pentobarbital. Mean total responding was not reliably different for pentobarbital compared with saline. When zolpidem was administered 5 min before the session and tested under the FR 1 schedule, the pattern of results observed was similar to that obtained under the FR 10 schedule. Thus, zolpidem did not produce a dose-dependent increase in drug-appropriate responding over the dose range of 1.0 to 4.0 mg/kg i.p. In contrast to the FR 10 schedule, however, all rats responded and obtained reinforcers at every dose. Because of this increase in response output, a higher dose of zolpidem was tested (6.0 mg/kg i.p.). This higher dose of zolpidem occasioned a mean percent drug lever of 62%. Of the five rats tested, one rat did not respond, two rats responded exclusively on the drug lever and two rats responded on both levers (93% and 13% drug-lever responding). The total responses obtained after 4.0 and 6.0 mg/kg zolpidem were reliably lower than obtained after saline administered 5 min before the session [q(12) = 5.40, P < .05 and q(12) = 5.36, P < .05, respectively; data analyzed from n = 4 rats].

Pentobarbital and zolpidem i.g. As can be seen in table 2, pentobarbital, administered i.g. and tested under a FR 10 schedule, occasioned a dose-dependent increase in drug-appropriate responding (percent drug lever). Dunnett's tests revealed no reliable difference in rates for any pentobarbital dose compared with saline administered i.g. Similar to the i.p. route, zolpidem occasioned partial drug-appropriate responding over a dose range of 1.0 to 4.0 mg/kg i.g. In contrast to the i.p. route, Dunnett's tests revealed no reliable difference in rates for zolpidem doses of 1.0, 2.0 and 4.0 mg/kg i.g. compared with saline (because of the repeated measures design, only n = 4 rats were used in the rate analysis). Because no rate effects were observed at these doses, a higher dose of zolpidem was tested (6.0 mg/kg i.g.). This high dose of zolpidem occasioned 100% drug-appropriate responding, but only two of the seven rats tested completed a response requirement, and of the rats not completing a response requirement, only one rat responded at all (this rat made one response on the saline lever). When the rate data for the four rats tested in all conditions were compared with saline, the 6.0 mg/kg zolpidem dose was reliably lower than saline [q(12) = 4.0, P < .05].

Dose (milligrams per body surface area) and Relative Dose

For milligrams per body surface area and relative dose comparisons among rhesus monkeys, rats and humans (table 3), the body weights averaged across the experiments were used. As can be seen in table 3, the pentobarbital training doses expressed as milligrams per body surface area were similar for rats and humans (5.9 and 5.3 mg/A, respectively), whereas for monkeys, the pentobarbital training dose was approximately 4-fold higher (20 mg/A). In general, higher doses of pentobarbital, zolpidem and triazolam were required to obtain full drug-appropriate responding in monkeys, when dose was expressed as milligrams per body surface area, than for either rats or humans. To compare the relationship of dose based on weight of the subject to dose based on body surface area, relative dose values were computed (see table 3). As can be seen in the table, relative dose was constant across drugs within a species. The relative dose decreased as the weight of the subject increased, with a rank order of rat > monkey > human.

	Discussion

Top Abstract Introduction Methods Results Discussion References

As found in baboons (Ator and Griffiths, 1989; Griffiths et al., 1992), both zolpidem and triazolam occasioned at least 80% drug-appropriate responding in pentobarbital-trained rhesus monkeys. Thus, zolpidem and triazolam were similar to other BZ agonists, such as diazepam, chlordiazepoxide, etizolam and brotizolam, which also occasioned 80% or more drug-appropriate responding in rhesus monkeys trained to discriminate pentobarbital from saline (Nader et al. 1991; Winger and Herling, 1982; Woolverton and Nader, 1995). Additionally, Takada et al. (1986) have reported that the BZ agonist, midazolam, engendered greater than 80% drug-appropriate responding in rhesus monkeys trained to discriminate the barbiturate, methohexital, from saline. Thus, the present results extend those findings in rhesus monkeys to the imidazopyridine BZ agonist zolpidem, and the triazolobenzodiazepine triazolam, and support the results found in baboons (Griffiths et al., 1992). These results suggest that in primates, the behavioral effects of zolpidem are similar to pentobarbital and classic BZs. Moreover, in the companion paper in the present series (Rush et al., 1997), in which humans discriminated pentobarbital from placebo, both zolpidem and triazolam occasioned full pentobarbital-appropriate responding. In addition, the subject-rated and performance-impairing effects of zolpidem were similar to those of triazolam and other classic sedative/hypnotic agents, such as barbiturates and BZs (Rush and Griffiths, 1996; Rush et al., 1997). BZs and barbiturates have consistently been found to have similar subjective effects in humans (deWit and Griffiths, 1991). Taken together, these results suggest that the discriminative stimulus effects of zolpidem in humans and nonhuman primates are similar to classic BZs, and these results generally support the hypothesis that discriminative stimulus effects in nonhuman primates predict subjective effects in humans.

Despite the similarity of discriminative stimulus profiles between zolpidem and triazolam in pentobarbital-trained monkeys, there were some differences between the two compounds. First, in terms of potency differences among monkeys, zolpidem was more variable than triazolam. That is, zolpidem was approximately 17- to 30-fold more potent in one monkey than the other three monkeys, but potencies were similar for triazolam across monkeys. In addition, increases in latency to respond were reliably higher than saline latency at 10 mg/kg zolpidem, whereas triazolam did not reliably alter latency to respond, although a trend for an increase was evident at the higher triazolam doses. It is not clear whether these differences in interanimal variability or performance decrements represent a unique profile of effects for zolpidem compared with triazolam. Previous research in our laboratory with the same procedures typically has not revealed interanimal differences in potency such as those observed for zolpidem in monkeys tested with diazepam, chlordiazepoxide, etizolam (Woolverton and Nader, 1995) or brotizolam (Nader et al., 1991). However, considerable variability has been observed with the latency measure for these compounds, with BZ effects on latency being inconsistent across drugs and among animals (Nader et al., 1991; Woolverton and Nader, 1995).

In contrast to the results observed in rhesus monkeys, zolpidem occasioned only partial drug-appropriate responding in rats trained to discriminate pentobarbital from saline, under similar conditions in which pentobarbital was administered (15-min pretreatment i.p.). Moreover, zolpidem produced severe reductions in rate, such that many rats at the high zolpidem doses did not complete FRs or emit any responses at all. These results with zolpidem were similar to those reported previously in chlordiazepoxide-trained rats (Depoortere et al., 1986; Sanger et al., 1987). In addition, pentobarbital occasioned only partial drug-appropriate responding in rats trained to discriminate zolpidem from saline (Sanger and Zivkovic, 1986). Finally, the triazolobenzodiazepine triazolam occasioned 80% or higher drug-appropriate responding in the pentobarbital-trained rats, consistent with previous findings and similar to other classic BZs (Ator and Griffiths, 1989). Compared with the results with rhesus monkeys, baboons and humans, these data with zolpidem in pentobarbital-trained rats raise the possibility of an important species difference between rodents and primates, including humans.

One possibility for the pattern of effects observed after zolpidem administration in rats may be related to zolpidem's pharmacokinetics. Zolpidem is a relatively fast-acting drug that is rapidly eliminated (Sanger and Zivkovic, 1986; Trenque et al., 1994). To test this possibility, two other pretreatment times were assessed: 5 min before the session and 45 min before the session. At the 5-min pretreatment time, zolpidem again occasioned only partial pentobarbital-appropriate responding. However, the dose-response function for rate suppression was shifted to the left relative to the 15-min pretreatment time. Additionally, more rats did not complete FRs at the 5-min pretreatment time than at the 15-min pretreatment time, with no rats completing a FR at the highest dose tested (4.0 mg/kg). At the 45-min pretreatment time, all responding was on the saline-appropriate lever and no effects on rate of responding were observed. Thus, the peak effect of zolpidem on rate of responding was observed with the 5-min pretreatment time, whereas zolpidem occasioned primarily partial drug-appropriate responding at 5- and 15-min pretreatment times. All effects of zolpidem had dissipated at the 45-min pretreatment time. These data suggest that the lack of consistent dose-dependent pentobarbital-appropriate responding observed after zolpidem in rats was not caused by pharmacokinetic factors, such as testing the drug at too late or too early a pretreatment period for sufficient levels of zolpidem to be in the CNS during the testing period.

Although species differences may have accounted for the difference in discriminative stimulus effects for zolpidem observed between rats and rhesus monkeys, a variety of procedural factors also may have contributed to this difference. The difference between rats and rhesus monkeys in discriminating zolpidem may have been caused by the fact that shock avoidance (i.e., negative reinforcement) was used as the contingent event for rhesus monkeys, whereas food presentation (i.e., positive reinforcement) was used as the contingent event for rats. However, full drug-appropriate responding was obtained in humans (Rush et al., 1997) and baboons (Griffiths et al., 1992) by use of positive reinforcement; furthermore, the baboon study used food presentation. Nonetheless, at least two hypotheses based on procedural differences may be posited. First, because zolpidem produces rate decreases of relatively higher magnitude in rats compared with other benzodiazepines, it is possible that the discriminative stimulus effects of zolpidem were masked in the rat studies by the inability of the rat to respond. Thus, one hypothesis is that zolpidem has pentobarbital-like discriminative stimulus effects that are attainable only under conditions of low response output. Consistent with this notion, zolpidem shared discriminative stimulus effects with pentobarbital in rhesus monkeys responding under a FR 1 schedule. Second, the discriminative stimulus effects of zolpidem may vary as a function of route of administration. Consistent with this notion, all positive results with humans and nonhuman primates were obtained after i.g. or perioral administration, whereas the negative results obtained in the present study were observed after i.p. administration.

To test the hypothesis that zolpidem's rate effects interfere with measurement of its discriminative stimulus effects, zolpidem was tested in the pentobarbital-trained rats under a FR 1 schedule of reinforcement. Similar to the FR 10 schedule, partial drug-appropriate responding was observed after zolpidem tested under a FR 1 schedule. This partial drug-appropriate responding was observed even after a dose of zolpidem was tested that was higher than could have been tested under the FR 10 schedule. Thus, these results indicate that relatively low response requirements, such as the FR 1 schedule used in the rhesus monkey study, likely did not account for zolpidem showing full pentobarbital-appropriate responding. Consistent with this, zolpidem occasioned full pentobarbital-appropriate responding in baboons responding under a range of FRs from 15 to 40 (Griffiths et al., 1992).

To test whether the discriminative stimulus effects of zolpidem vary as a function of route of administration, zolpidem was administered i.g. in pentobarbital-trained rats. Interestingly, zolpidem occasioned 80% drug-appropriate responding when tested at a high dose (6.0 mg/kg) via the i.g. route, but only two of seven rats completed a FR 10 at this dose. These results suggest that full pentobarbital-appropriate responding may be observed only after oral doses of zolpidem that severely impair performance, i.e., under a condition in which only 29% of the subjects responded. Thus, it is difficult to fully attribute the difference in discriminative stimulus effects of zolpidem in pentobarbital-trained rats and primates to a difference in i.p. versus oral absorption, because no evidence of a similar relationship between performance decrements and discriminative stimulus effects was observed in either human or nonhuman primate subjects after any zolpidem dose.

Other factors which may have accounted for the differences observed between rats and monkeys should be noted. One possibility may be the result of relative differences in the training dose of pentobarbital. In general, a larger number of drugs, varying in such factors as pharmacological mechanism and intrinsic activity, often share discriminative stimulus effects at low compared with high training doses (e.g., Sannerud and Ator, 1995a,b). Thus, the training dose of pentobarbital may have been relatively low in rhesus monkeys, increasing the probability of observing drug-appropriate responding to a greater variety of drugs, compared with the training dose used in rats. To examine the issue of training dose, the doses of drugs were converted from mg/kg to mg/A, according to Dews (1976). This conversion is based on body surface area and takes into account that different species may not only vary in body weight, but in surface area as well, the latter being a more important factor when considering drug absorption (Dews, 1976). In the present study, the relative dose computation revealed a rank order of rat > rhesus monkey > human, which indicated that as the size of the subject increased, the proportion of drug dose based on weight to drug dose based on surface area decreased, as expected (Dews, 1976). Training doses of pentobarbital, expressed as mg/A, were similar for rats and humans, both of which were lower than the training dose for monkeys. However, the species with the greatest mg/A difference in the training dose of pentobarbital, monkeys and humans, demonstrated similar discriminative stimulus effects of zolpidem. Thus, the dose comparisons based on mg/A suggest that training dose of pentobarbital did not account for the observed differences in the discriminative stimulus effects of zolpidem across species. Nevertheless, the influence of pentobarbital training dose on the discriminative stimulus effects of zolpidem is an important consideration and remains to be determined conclusively.

It is important to note that it is difficult, often impossible, to precisely equate procedural details across species. Nevertheless, cross-species comparisons are essential to establish the relevance to humans of data collected with use of animal models. The results of the present series of studies in rats, rhesus monkeys and humans suggest that the discriminative stimulus effects of zolpidem may differ in rats compared with primates. Consistent with this notion, recent studies have found a different profile of in vivo binding of zolpidem in nonhuman primate brain compared with previous in vitro work in rodent brain (Schmid et al., 1995). Using positron emission tomography, monophasic displacement of [¹¹C]flumazenil by zolpidem was observed in cerebellum of baboons, whereas biphasic displacement was observed in hippocampus (Schmid et al., 1995). These results contrast autoradiographic data obtained in rat brain, in which zolpidem displacement of [³H]flumazenil binding could be fit to a three-site function in the hippocampus and a one-site function in the cerebellum (Benavides et al., 1993). Although comparisons between in vitro and in vivo binding may be problematic, possible species differences in the binding characteristics of zolpidem may somehow reflect the distinct profile of discriminative stimulus effects observed in rats, in contrast to the more typical BZ-like discriminative stimulus effects observed in rhesus monkeys, baboons and humans. Whether or not these behavioral differences reflect differences in BZ receptors across species, the in vivo and in vitro results taken together clearly distinguish zolpidem among the BZs and raise the possibility of functional heterogeneity of BZ receptor subtypes that vary as a function of species.