Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

In drug discrimination studies, differential reinforcement procedures are used to train subjects to make one response when a particular dose of a drug has been administered and to make a different response when it has not. When good control of responding is shown under training conditions, the drug is said to serve a discriminative stimulus function for that behavior. Tests with drugs other than the training drug generally show that administration of pharmacologically related drugs results in the same lever selection as the training drug but that administration of pharmacologically disparate drugs does not (Jarbe, 1989). This outcome has made the drug discrimination procedure valuable for studying commonalities of subjective drug effects in animals and for exploring the functional relevance of particular molecular mechanisms of action.

Benzodiazepines potentiate GABAergic synaptic transmission by binding at a modulatory site on the GABA_A-receptor complex (Haefely et al., 1985). Clinically useful Bzs differ primarily in their potencies for producing the anticonvulsant, muscle relaxant, anxiolytic and hypnotic effects characteristic of this class of drugs. There generally have not been qualitative behavioral differences among Bzs across a variety of behavioral paradigms (Dantzer, 1977; Sanger and Blackman, 1981); and in drug discrimination studies, animals and people trained to discriminate a particular Bz generalize to other Bzs (Ator, 1990; Kamien et al., 1993; Rush et al., 1996).

Behavioral effects of Bzs are similar in many ways to those of the barbiturates, which they have replaced as anxiolytics and hypnotics in clinical practice (Lader, 1995). The similarity is not surprising, because barbiturates also enhance GABA, albeit through a different site on the GABA_A-receptor complex. Animals trained to discriminate a Bz typically generalize to barbiturates and vice versa (Ator, 1990; Ator and Griffiths, 1989a). In fact, training with either Bzs or barbiturates generally has resulted in most sedatives and anxiolytics occasioning the drug-paired response. Thus, such training conditions have not seemed highly selective for differentiating among these compounds (Colpaert et al., 1976; Overton, 1984).

In the first study with lorazepam as a training drug, however, generalization to pentobarbital occurred in only one of four baboons, even though pentobarbital doses were tested that clearly were behaviorally active (Ator and Griffiths, 1983a). The same study showed that pentobarbital-trained baboons did generalize to lorazepam. A subsequent study in rats, which also manipulated lorazepam and pentobarbital training doses, replicated this asymmetrical generalization profile (Ator and Griffiths, 1989b). In later studies, novel, nonBz, ligands for the Bz modulatory site did occasion lorazepam-lever responding in baboons and rats, which showed that lorazepam-trained animals would generalize to some types of nonBz GABAergic drugs (but they did not generalize to the direct GABA agonist THIP; Ator and Griffiths, 1986, 1992; Griffiths et al., 1992; Sannerud et al., 1992). Neuroactive steroids, which enhance GABA through a nonBz, non-barbiturate site on the GABA_A complex, however, did not produce reliable generalization in lorazepam-trained rats but did so in rats trained to discriminate diazepam, pentobarbital and EtOH (Ator et al., 1993). Thus, animals trained to discriminate lorazepam showed selectivity for compounds that enhance GABA through the Bz site rather than generalizing to compounds that enhance GABA per se.

To date, however, the generalization profile for a range of barbiturates and other classic sedative/anxiolytic compounds has not been characterized in lorazepam-trained animals. Such a profile is essential for fully evaluating the selectivity of lorazepam training conditions. Particularly in the context of emerging data on the heterogeneity of the GABA_A-receptor and the selectivity of its subtypes, including ones that form the putative ₁/Bz₁ and ₂/Bz₂ subtypes (Langer and Arbilla, 1988; Sieghart, 1995), a full profile of effects of classic sedative/anxiolytic, anticonvulsant compounds is a prerequisite for understanding whether apparent selectivity of the training drug condition matches its neuropharmacological selectivity. The present paper presents a generalization profile for baboons trained to discriminate lorazepam. A range of Bzs, barbiturates and other compounds that have anxiolytic, anticonvulsant, anesthetic, muscle relaxant and/or sedative-hypnotic effects were tested; drugs from other classes that share certain characteristics with sedatives also were tested, as well as compounds not believed to share any effects with sedative/anxiolytics. The use of a reliable p.o. dosing procedure in the baboon permitted study of virtually all test drugs across a wide dose range via the route by which most of the test drugs usually are used by humans. Where negative results occurred with barbiturates, route of administration and/or time of testing were manipulated to determine whether the probability of lorazepam-lever responding would increase.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Subjects

Six adult male baboons (Papio anubis except that baboon LO was Papio cynocephalus; Primate Imports, New York, NY) were housed individually, with water available at all times. All had served in different studies of i.v. drug self-administration (LE: cocaine; LO, MS, RA: cocaine, nicotine; ML: heroin; RF: cocaine, nicotine, methohexital, baclofen); and all had served in other drug discrimination studies in which tests were conducted with Bz-site ligands, pentobarbital, azapirones, pentylenetetrazole and caffeine (Ator, 1990; Ator and Griffiths, 1983a, 1985, 1986; Ator et al., 1989; Griffiths et al., 1992; Sannerud et al., 1991, 1992, 1993). Rations of laboratory monkey chow were supplemented daily with a multivitamin and fresh fruit. Feeding occurred at approximately the same time each day, which was 30 to 75 min after the regular experimental session; on days when the drug time course was evaluated, the ration of monkey chow was omitted. Ketamine hydrochloride (HCl) preceded by atropine sulfate (SO₄) was administered i.m. approximately every 2 weeks to permit weighing the baboons. The baboons were not maintained at reduced weights compared with a free-feeding weight (Ator, 1991). Weights generally increased across the 5 to 7 years in which the data were collected. The range of weights (rounded to the nearest kilogram) was: 24 to 32 for LE, 29 to 35 for LO; 30 to 36 for ML, 27 to 38 for MS and 28 to 34 for RA and RF.

Apparatus

Two systems were used at different times during the studies. At the beginning, sessions with five of the baboons were conducted when they had been adapted to restraint chairs (Findley et al., 1971) and placed in a sound-attenuating chamber equipped with an intelligence panel. Some of the data for alprazolam (LO, ML, RA), bromazepam (ML, RA) diazepam (ML), temazepam (LO, ML, RF) and phenobarbital (LO, RA) and all data (except LE's) for triazolam and i.m. pentobarbital were obtained with the chair apparatus. All other dose-effect curves were obtained with the baboons seated on a bench in front of the intelligence panel, which formed the rear wall of a standard stainless steel primate cage (see fig. 4 in Ator, 1991). The cage was enclosed in a moderately sound-attenuating chamber for all except the methohexital determinations; for those, visual isolation from the baboons on either side was provided, during sessions, by masonite panels. The intelligence panels (constructed in the laboratory) contained two jewel lights mounted over either two Lindsley levers (Gerbrands Corp., Arlington, MA) or two stainless steel levers (constructed in the laboratory), which closed microswitches when operated. The levers were approximately 15 cm apart on the lower left half of the panel and within easy reach. A drinkometer (Kandota Instruments, Sauk Centre, MN), used for p.o. drug delivery, was mounted on the left above the levers. A food hopper, illuminated coincident with pellet delivery, was located in the upper middle of the panel. An electromechanical feeder (various sources) was used to deliver 1-g banana-flavored pellets (P.J. Noyes, Lancaster, NH or BIO-Serv, Inc., Frenchtown, NJ) and was mounted above the chamber as was a tone generator, which delivered white noise and tones through a speaker mounted on the back of the intelligence panel. A 5-×-5 cm translucent white plexiglas panel, which could be transilluminated, was mounted in the upper right quadrant of the panel. Experimental conditions were programmed and data collected by use of PDP8 computers (Digital Corp., Maynard, MA) programmed in SUPERSKED (State Systems, Kalamazoo, MI). Graphic records of each session's performance were collected with cumulative recorders (Gerbrands Corp., Arlington, MA).

Procedure

Training sessions. A 60-min time-out, which coincided with the lorazepam pretreatment time, preceded each experimental session. During time-out, the translucent panel was illuminated and lever responses were counted but had no programmed consequences. White noise was turned on at the beginning of this time-out and continued until the end of the session. When the time-out ended, the panel light was extinguished, both jewel lights were illuminated, a 3-s tone sounded and a 20-min period of food pellet availability began. A response on either lever in the presence of the jewel lights produced a 0.1-s tone. In training sessions, food pellet delivery depended on a fixed number of consecutive responses on the lever appropriate to the drug or ND condition in effect. Responses on the inappropriate lever reset the response requirement. The lever paired with the drug and ND conditions was counterbalanced across baboons. Completion of the response requirement turned off the jewel lights, operated the feeder, illuminated the food tray for 1 s and initiated a 6-s time-out. The response requirement for each baboon was determined empirically to be that which best maintained criterion performance. It was 10 or 15 responses for MS, 15 or 20 responses for LO, 20 responses for LE and ML, 35 responses for RF and 40 responses for RA. Lorazepam and ND training sessions generally alternated.

Test sessions. A test session was conducted only if responding in the training sessions met the criteria that 1) 95 to 100% of the total responses had been on the correct lever and 2) before the first food delivery of the session, completion of the required number of consecutive responses on the correct lever had not been preceded by the same number of consecutive responses on the incorrect lever. Before study of the dose-effect relationship for each drug, test sessions first were conducted with the training dose of lorazepam and with vehicle to confirm the reliability of the trained discrimination. Test sessions were identical with training sessions except that the length of the presession time-out corresponded to the test drug pretreatment time and completing the usual required number of consecutive responses on either lever produced a food pellet. The order of drug and ND training sessions between test sessions was counterbalanced so that test sessions were preceded equally as often by a lorazepam as by a ND training session. At least one drug and one ND training session occurred between tests with novel drug doses. If criterion performance was not shown in a training session, training sessions alternated until criterion performance occurred in four consecutive sessions.

Design and data analysis. Data are reported on 31 drugs. A single-subject design (Sidman, 1960) was used, in which each baboon served as his own control for determining whether each test drug shared discriminative stimulus effects with lorazepam and whether response rates were affected. Not all drugs were studied in all baboons; the range was 16 (baboon LE) to 31 (baboon MS). The dose at which individual baboons showed effects different from vehicle sometimes varied as much as a 0.5 log₁₀ unit or more. Thus, this design not only conserved the amount of drug needed in these large animals but also avoided subjecting individual baboons to doses unnecessarily high for the purposes of the study. Order of testing was mixed except that i.m. pentobarbital was studied first and methohexital was studied last. Two or more observations were made at each dose for some or all of the baboons. The dose range encompassed one or more low doses that occasioned <20% drug-lever responding and one or more high doses that occasioned at least 80% drug-lever responding. For drugs that did not engender lorazepam responding, a log₁₀ range or greater was studied with almost all to encompass a range of possible effects and to try to include at least one dose that produced a decrease in the rate of lever pressing. A portion of the determinations with bromazepam, clomethiazole, diazepam, methaqualone, temazepam and triclofos were conducted blind in conjunction with the development of a sedative-stimulant screening program by the Committee (now College) on Problems of Drug Dependence.

Drugs and dosing. Doses are expressed as the form of the drug listed below. The following drugs were donated: alprazolam and triazolam, Upjohn Co., Kalamazoo, MI; chlordiazepoxide HCl, diazepam, methyprylon, midazolam maleate and nordiazepam, Hoffmann-LaRoche, Inc., Nutley, NJ; lorazepam and meprobamate, Wyeth Laboratories, Philadelphia, PA; Ketamine HCl, Warner-Lambert Co., Ann Arbor, MI and also purchased as Ketaset (100 mg/ml), Fort Dodge Laboratories, Inc., Fort Dodge, IA; temazepam, Sandoz, Inc., East Hanover, NJ; d-amphetamine SO₄, Smith, Kline, & French, Philadelphia, PA; mesocarb, PCP and ⁹-THC, National Institute on Drug Abuse, Research Triangle Park, NC; phenytoin, Parke-Davis, Ann Arbor, MI. The following drugs were purchased: chloral hydrate, haloperidol, hexobarbital, pentobarbital sodium (Na), and secobarbital Na, Sigma Chemical Co., St. Louis, MO; methaqualone, Lemmon Pharmacal Co., Sellersville, PA; amobarbital Na, and phenobarbital Na, Ganes Chemicals, Inc., Pennsville, NJ; methohexital Na (as Brevital), Eli Lilly and Co., Indianapolis, IN; morphine SO₄, Mallinckrodt, Inc., St. Louis, MO; valproic acid, Saber Laboratories, Inc., Morton Grove, IL. Bromazepam, clomethiazole HCl and triclofos were obtained exclusively through the Committee on Problems of Drug Dependence. Solutions of 10% EtOH weight/volume were prepared by adding distilled water to 95% volume/volume EtOH (both at room temperature). All drugs were given p.o. except haloperidol, ketamine and morphine, which were given i.m. only. Chlordiazepoxide, pentobarbital and methohexital were studied i.m. and p.o.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Stimulus control was well maintained at criterion level under the training conditions and was readily reestablished after periods of absence from the procedure. Each baboon's range of response rates in the lorazepam training sessions generally was the same as or only slightly higher than the range of response rates in the ND sessions. However, for one baboon, LO, response rates in the lorazepam training sessions generally were 0.5 to 1 response (r)/s higher than those in the ND sessions such that the ranges did not overlap or did so only slightly in most of the studies.

Benzodiazepines

Lorazepam. Three lorazepam dose-effect determinations were made across a 4- to 8-month period and are shown in figure 1. All baboons generalized to doses lower than the 1.8 mg/kg lorazepam training dose, but they differed in the probability that doses as low as 0.1 or 0.32 mg/kg would occasion responding on the lorazepam lever. As shown in table 1, the ED₅₀ for lorazepam ranged from 0.1 to 0.56 mg/kg across baboons; and the ED₈₀ ranged from 0.32 to 1.8 mg/kg. Only the generalization gradients for baboon MS were quantal (i.e., responding was essentially 100% on one or the other lever at all doses). The other baboons produced intermediate percentages of responding (i.e., between 10 and 90%) on the lorazepam lever in some determinations at doses lower than the training dose. Although baboons LO and RF had the highest ED₅₀ values (0.56 mg/kg), baboon LO's generalization gradient was very similar to that of the baboon with the lowest ED₅₀ (0.1 mg/kg for baboon RA). In both LO and RA, 0.1 mg/kg, which occasioned 0% responding in the other three baboons, occasioned considerable lorazepam lever responding on some tests. For all baboons, the dose higher than the training dose, 3.2 mg/kg, occasioned 100% responding on the lorazepam lever.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Percentages of responding on the lorazepam lever and overall response rates in test sessions 60 min after administration of lorazepam or its vehicle (V) to baboons trained to discriminate lorazepam (1.8 mg/kg p.o.) from the ND condition. Individual points are data from single test sessions in the first (), second () and third () determinations of the dose-response relationship across a 4-month (baboons MS, RA) or 8-month (LO, ML, RF) period; solid lines indicate the means. Response rates are given as a percentage of the mean rate in control ND sessions. Control mean response rates (r/s) for each baboon were: LO, 0.6; ML, 2.7; MS, 1.4; RA, 3.1; and RF, 2.7.

Alprazolam, bromazepam, diazepam, midazolam, temazepam, triazolam. Eight Bzs other than lorazepam were studied. Dose-effect determinations were conducted in six baboons with diazepam and in five baboons with the other Bzs; each dose was studied one to three times in each baboon. Group means for lorazepam and six other Bzs, all administered by the p.o. route and assessed at the same interval after dosing, are shown in figure 2. The triazolo-Bzs, triazolam and alprazolam, were more potent than lorazepam in occasioning lorazepam-lever responding; and the 1,4-Bzs diazepam, midazolam, bromazepam and temazepam were less potent than lorazepam. This order roughly held for individual baboons as well, as shown by the ED₅₀ and ED₈₀ values in table 1. Responding peaked at 100% on the lorazepam lever for all those drugs and for each baboon tested except for diazepam for one baboon. Baboon RF did not make >80% lorazepam lever responses at any dose of diazepam, although the other five baboons had ED₅₀ values of 0.32 to 1.0 mg/kg and showed full generalization (ED₈₀) at 1.0 or 3.2 mg/kg (table 1). The group diazepam generalization gradient in figure 2 peaks at 100% at 5.6 mg/kg because baboon RF was not tested at that dose. Diazepam dose was increased to 18 and then 32 mg/kg for this baboon (data not shown), which occasioned 25% and 65% lorazepam-lever responding, respectively. Neither time course studies nor redetermination of the p.o. diazepam curve 7 years later produced a different qualitative result in total session percentages of lorazepam-lever responding for baboon RF. Diazepam also was given to baboon RF i.m. (in the commercial injectable form, 5 mg/ml concentration); but it occasioned no lorazepam-lever responding up to 1.0 mg/kg, a dose that had occasioned 100% lorazepam-lever responding in baboons ML and RA (Ator and Griffiths, 1983a). Although baboon RF was the only baboon whose initial training was with 1.8 mg/kg lorazepam p.o., a generally lessened sensitivity to the discriminative stimulus effects of the other Bzs with which he was tested was not shown except for temazepam (see table 1). [However, this same baboon was the only one of five baboons that did not show maximal lorazepam-lever responding with the ₁/Bz₁ ligand zolpidem (Griffiths et al., 1992)].

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Mean percentages of responding on the lorazepam lever and overall response rates (as a percentage of the control ND rate) in test sessions 60 min after p.o. administration of Bzs to lorazepam-trained baboons. Each data point represents the mean of four or five baboons, except the points for the lowest dose of temazepam and for the lowest and highest doses of diazepam and bromazepam are for two baboons (see text for 32 mg/kg diazepam results, not shown). The ranges of mean control ND response rates (r/s) for individual baboons during these curves were: 1.5 to 2.6 for alprazolam, 1.3 to 2.6 for bromazepam, 0.5 to 3.1 for diazepam, 0.6 to 3.1 for lorazepam, 0.6 to 2.9 for midazolam, 1.5 to 2.4 for temazepam and 1.4 to 2.5 for triazolam. The lorazepam curve is the grand mean of the data in figure 1. In vehicle test sessions (not shown), percentages of lorazepam-lever responding were <5%; and response rates ranged between 80 and 120% of the ND control rate for individual baboons.

Nordiazepam and chlordiazepoxide. The group curves for nordiazepam and chlordiazepoxide are not shown in figure 2, because those results are not well characterized by the group means. Neither of the mean generalization gradients reach 80% lorazepam-lever responding. The peak for each drug was at 18 mg/kg: 61% for nordiazepam and 64% for chlordiazepoxide. However, both Bzs did share discriminative stimulus effects with lorazepam in most of the baboons tested (fig. 3). Nordiazepam occasioned 100% drug-lever responding in four of the five baboons tested, but the effective dose varied substantially across baboons (fig. 3, table 1). For example, 0.1 mg/kg occasioned 94% lorazepam lever responding in baboon LO; but generalization from lorazepam to nordiazepam did not occur except at 56 mg/kg for baboon RA. Baboon MS's mean curve peaks only at 67%. This baboon made 86% and 91% lorazepam-lever responses when tested at 10 and 18 mg/kg, respectively, but retests with those doses twice more resulted in lower, intermediate, percentages of lorazepam-lever responding. Retests with 10 mg/kg in baboons ML or RF, however, replicated the original result. For RF, both tests at 10 mg/kg occasioned >90% lorazepam-lever responding, but the single test at 18 mg/kg occasioned 62%. Spacing between nordiazepam tests was 7 days or more. The generalization gradient for nordiazepam was an inverted U for two of the five baboons, which was unusual (as described above, the inverted U for diazepam in fig. 2 was an artifact of one baboon's failure to generalize). Response rates again were increased above the control ND range for baboon LO but were below the control range for lorazepam training sessions during the nordiazepam study (i.e., 242-387% of ND rates).

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Percentages of responding on the lorazepam lever and overall response rates in test sessions 60 min after nordiazepam or chlordiazepoxide and their vehicles (V) in lorazepam-trained baboons. Each data point generally represents a single determination for each baboon, except: Nordiazepam: for RF, V, 3.2, 10 = two tests; for ML, V, 10 = three tests, 1.0 = two and 3.2 = four tests; for MS, V, 10, 18 = three tests, 3.2 = two. Chlordiazepoxide: for RA, 3.2 = two tests; for LE, 10 = two and 18 = four tests; for LO, 32 = two tests; for MS, V, 32 = two tests; 10, 18 = three tests. Control mean response rates (r/s) were: LE, 2.3; LO, 0.6; and MS, 1.7; they differed across curves for three baboons and were: ML, 2.2 and 2.3; RA, 3.0 and 2.7; and RF, 1.6 and 2.5 for, the left and right panels, respectively.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Percentages of responding on the lorazepam lever and overall response rates in test sessions 60 min after chlordiazepoxide or vehicle (V) p.o. or i.m. in lorazepam-trained baboons. Symbols represent each of the same baboons as in figure 3, and the solid or dashed lines represent the group mean. For the p.o. doses, the data points are the same as those in figure 3. For the i.m. doses, each data point represents a single determination.

Barbiturates and Other Sedatives/Anxiolytics

Amobarbital, hexobarbital, methohexital, methyprylon, pentobarbital, phenobarbital, secobarbital. Neither barbiturates nor methyprylon, a piperidinedione sedative-hypnotic, occasioned maximal lorazepam-lever responding in most baboons. Group generalization gradients for five barbiturates and for methyprylon, all tested 60 min after dosing, are shown in figure 5. These gradients generally were a monotonically increasing function of dose, but peaked between 20% (hexobarbital) and 50% (phenobarbital). The intermediate mean percentages of lorazepam-lever responding for drugs shown in figure 5 generally resulted from each drug's occasioning a maximum of 80 to 100% lorazepam-lever responding in one or two baboons and between 10% and 90% in one to three others. Figure 6 presents the results for baboons individually and shows that phenobarbital and methyprylon each occasioned full dose-dependent generalization in two baboons. The maximum percentage of lorazepam-lever responding after amobarbital, secobarbital and pentobarbital was at the 18 mg/kg dose and was in baboon ML for both amobarbital and pentobarbital (respectively, 62% and 55%) and in baboon LE for secobarbital (74%). (Baboon RF was not tested above 18 mg/kg amobarbital and phenobarbital nor with the other two barbiturates because he did not reliably consume those drugs.) The drugs generally did decrease response rates below the range of rates in ND control sessions at 18 mg/kg and above (see also fig. 5). As with the Bzs, baboon LO's response rates increased well outside the range of rates in control ND sessions at one or more doses except with phenobarbital (fig. 6).

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Mean percentages of responding on the lorazepam lever and overall response rates after various barbiturates and methyprylon in lorazepam-trained baboons. See figures 6, 7 and 8 for further details.

View larger version (33K): [in this window] [in a new window]   Fig. 6.   Percentages of responding on the lorazepam lever and response rates in test sessions 60 min after administration of barbiturates, methyprylon and their vehicles (V) to lorazepam-trained baboons. Data points generally represent single determinations for each baboon, except that they are means of two for some tests with methyprylon (10-32 mg/kg for LE) and phenobarbital (10 mg/kg for MS and 18 mg/kg for RA). The grand means of these data are in figure 5. The control mean response rates (r/s) for each baboon across these curves were: LE, 2.6 to 3.0 and LO, 0.5 or 0.6, except both were 1.5 during phenobarbital; ML, 2.2 to 2.8; MS, 1.6 to 2.1, except 1.1 during phenobarbital; RA, 1.7 to 3.1; and RF, 2.4 to 2.8, except 1.6 during methyprylon.

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Percentages of responding on the lorazepam lever and response rates in test sessions 30 or 60 min after administration of p.o. or i.m. pentobarbital to lorazepam-trained baboons. Each symbol represents a different baboon (see fig. 3 for key) and the solid or dashed lines represent the group means. Each data point represents a single determination, except that the points for the open squares in the right-hand panels represent the mean of two determinations (three at 5.6 mg/kg).

View larger version (17K): [in this window] [in a new window]   Fig. 8.   Percentages of responding on the lorazepam lever and response rates 30 and 60 min after administration of hexobarbital to lorazepam-trained baboons. Each symbol represents a different baboon (see fig. 6 for key) and each data point is for a single determination; the solid or dashed lines represent the group means. The group means for the tests 60 min after dosing are the same as those in figure 5. Control mean response rates (r/s) were: LO, 0.7; ML, 2.4; MS, 0.9; RA, 3.3; and RF, 2.6.

View larger version (18K): [in this window] [in a new window]   Fig. 9.   Percentages of responding on the lorazepam lever and response rates after p.o. and i.m. administration of methohexital to lorazepam-trained baboons. Test sessions began 10 min after i.m. injections and 15 min after p.o. administration. Each symbol represents a different baboon (see fig. 4 for key) and the solid or dashed lines represent the group means. Each i.m. data point represents the mean of two tests for each baboon, except that for closed squares, 3.2 = three tests; for closed hexagons, V, 1.0 = one test; for closed diamonds, V = one test. Each p.o. data point represents a single determination for each baboon, except that for open squares, 1.8, 3.2 = two tests; for open diamonds, V, 1.8, 5.6 = two tests and 3.2 = three tests. Control mean response rates (r/s) were: ML, 1.7; MS, 1.6; RA, 1.0; and RF, 2.1.

Chloral hydrate, clomethiazole, EtOH, meprobamate, methaqualone and triclofos. Table 2 presents the results of tests with six other sedative-hypnotics. Although they were tested across a wide range of doses, only methaqualone occasioned greater than 80% drug lever responding in any baboon tested (92% for baboon LE at 10 mg/kg; response rate was 144% of control). Chloral hydrate, clomethiazole and methaqualone decreased response rates outside the control range in all or all but one baboon and the median percentage of the ND control rate was between 50 and 60%. However, EtOH did so in only one of the three baboons tested; and the anxiolytic meprobamate and the sedative/hypnotic triclofos did not do so in any baboon, even though doses of 100 and 180 mg/kg were tested. Time course studies were conducted out to 4, 7, 9 and 15 h after dosing with triclofos, clomethiazole, chloral hydrate and meprobamate and methaqualone, respectively; but no later onset of lorazepam-lever responding occurred.

Drugs Not Classified as Sedatives or Anxiolytics

Table 2 also shows the results of testing nonsedative/anxiolytic drugs, some of which share some pharmacological effects with the sedatives and anxiolytics described above. Although tested at behaviorally active doses, neither the narcotic analgesic morphine, nor the dissociative anesthetics ketamine, phencyclidine, ⁹-THC, nor the anticonvulsants phenytoin and valproic acid, nor the neuroleptic haloperidol, nor the stimulants mesocarb and d-amphetamine produced >80% lorazepam-lever responding in any baboon. In fact, the median peak (i.e., maximum) percentage of lorazepam lever responding in the baboons tested was not greater than 5%. The drugs were tested up to doses that produced response rate decreases below the range of ND control rates in at least half of the baboons tested (the median percentage of the ND control rate for those baboons was less than 50% in most cases). Time course studies were conducted up to 7 h after dosing with ⁹-THC and 15 h with the anticonvulsants and no later onset of lorazepam-lever responding occurred.

First-Pellet Analysis

Certain results were at variance with those typically obtained for animals trained to discriminate Bzs. Conclusions about drugs sharing discriminative stimulus effects with lorazepam were evaluated further in terms of "initial lever selection." For all test sessions, the lever on which the baboon's responding first produced a food pellet was assessed to determine whether this evaluation would have yielded a different conclusion about whether the drug did or did not share discriminative stimulus effects with lorazepam. This analysis did not change the conclusion about those drugs that did or did not share discriminative stimulus effects with lorazepam, nor did it reveal greater consistency across baboons for curves in which there was a great deal of intersubject variability, but it did reveal that results of single tests would have been seen differently.

Lorazepam lever selection analysis of the data for the non-Bzs shown in figure 6 revealed: amobarbital-nonmonotonic lever selection for baboon LE, but not for the other baboons, phenobarbital-nonmonotonic lever selection for baboon RF rather than zero drug-lever responding, and lever selection at 32 mg/kg for baboon MS, compared with partial generalization; secobarbital-lever selection by baboon LE at 18 mg/kg, compared with partial generalization; methyprylon-nonmonotonic lever selection for baboon MS, rather than an orderly function, and lever selection at 56 mg/kg for baboon RF rather than partial generalization. For hexobarbital (fig. 8), there was lorazepam lever selection at 56 mg/kg p.o. for baboon MS 30 min after dosing, but there was no lorazepam lever selection for the one baboon (ML) that had otherwise been shown to generalize. For methohexital (fig. 9), the conclusion of partial generalization for baboon ML with i.m. dosing would change to none.

Lorazepam lever selection data also were evaluated for the Bzs. For the baboon (RF) that showed a maximum of 65% lorazepam responding in any test with diazepam, the first food pellet of the test sessions at 18 and 32 mg/kg was obtained after responding exclusively on the lorazepam lever. Thus, such an assessment results in a conclusion that diazepam shared discriminative stimulus effects with lorazepam in all six baboons. Two baboons failed to generalize completely to chlordiazepoxide (fig. 3). For one (LO), the first test with 32 mg/kg resulted in lorazepam-lever selection; but the second resulted in selection of the ND lever (overall percentages in the two tests were 56 and 59%, respectively). Thus, averaging the results of the tests in terms of lever selection also would have yielded a 50% choice. For the other baboon (LE) that failed to generalize to chlordiazepoxide, selection of the lorazepam lever occurred in two of the four tests with 18 mg/kg p.o., which mirrors the total percentage of lorazepam lever responding in the four tests (i.e., 4- 45%). Thus, conclusions about generalization to chlordiazepoxide were not altered by use of a lever selection analysis. Even for Bzs for which full generalization was shown (including lorazepam), there were tests in which total lorazepam-lever responding was high but the ND lever was selected initially.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The present study in baboons is the first to explore systematically the generalization profile for lorazepam-trained animals by testing a wide range of sedative/anxiolytic and other drugs. The lorazepam generalization gradients replicate the initial mean p.o. lorazepam gradients previously presented for those same baboons (Ator and Griffiths, 1986). The present mean gradients generally showed a 0.25 to 0.5 log₁₀ unit shift to the left compared with those earlier gradients. However, within the three dose-effect determinations conducted across a 4- to 8-month period for the present study, there was no consistent probability of a shift to the right or left with subsequent determinations. Thus, lorazepam stimulus threshold is concluded to fall within the range of doses encompassed between the one that never occasioned lorazepam responding and the one that always did (cf. Ator, 1990).

The nine Bzs tested differed in their chemical structure (i.e., classical 1,4-Bzs, such as diazepam; the 1,4 triazoloBzs, triazolam and alprazolam; the 1,4-imidazoleBz, midazolam). They differed in their usual clinical uses (e.g., antianxiety, intravenous anesthesia, hypnotic), bioavailability, routes of metabolism, and formation of active metabolites (review in Rall, 1990). All were tested via the oral route and at the same pretreatment time as the training drug, lorazepam. Despite the fact that these Bzs have been shown, in humans, to have different rates of uptake and elimination, and to differ in their formation of active metabolites (Ellinwood et al., 1985; Garzone and Kroboth, 1989; Greenblatt et al., 1989; Kaplan et al., 1976), most occasioned 100% lorazepam-lever responding in all baboons tested. The strongest exception was chlordiazepoxide, which was concluded to share discriminative stimulus effects with lorazepam in only three of five baboons, despite manipulation of the route of administration and study of the time course of the drug effect. The biotransformation of chlordiazepoxide begins shortly after administration to a large number of slowly eliminated metabolites, and desmethylchlordiazepoxide may be the major active compound as soon as 30 min after administration (Schwartz, 1973). One of the active metabolites of both chlordiazepoxide and diazepam is nordiazepam (also called desmethyldiazepam and desoxydemoxepam), which occasioned full generalization but with different potencies in all baboons.

Previous work in our laboratory produced the surprising result that baboons and rats trained to discriminate lorazepam did not generalize reliably to pentobarbital (Ator and Griffiths, 1983a, 1989b). Failure of baboons in the present study to generalize reliably to any of the six barbiturates replicates and extends the initial finding. The barbiturates tested included representatives of those traditionally classified as long-acting (phenobarbital), ultrashort-acting (hexobarbital, methohexital) and short- to intermediate-acting (amobarbital, pentobarbital, secobarbital). None of them occasioned >80% lorazepam-lever responding in more than one or two baboons per drug, regardless of manipulations of route and/or pretreatment time with most of them. Thus, a group mean dose-effect curve for each barbiturate can be characterized as showing partial generalization. The previous finding that baboons trained to discriminate lorazepam 1.0 mg/kg i.m. failed to generalize to intramuscular pentobarbital (Ator and Griffiths, 1983a) was extended to training with lorazepam 1.8 mg/kg p.o. and to testing with oral pentobarbital. Although the training dose variations in baboons were not large, another study that varied lorazepam training dose by 0.5 log₁₀ units in rats also did not find reliable generalization to pentobarbital (Ator and Griffiths, 1989b). Furthermore, tolerance to the rate-decreasing effects of pentobarbital in lorazepam-trained rats did not result in full generalization to pentobarbital (Ator and Griffiths, 1989b).

Other classic sedative-hypnotics/anxiolytics also failed to occasion substantial lorazepam-lever responding, except that methyprylon, like some of the barbiturates, occasioned >90% lorazepam-lever responding in two baboons, which yielded a maximum mean percentage of almost 50% lorazepam-lever responding. The other drugs of this class failed to occasion a maximum of even 2%. The specificity of the lorazepam training condition was confirmed further by the fact that nine drugs from other pharmacological classes (narcotic analgesic, dissociative anesthetic, anticonvulsant, neuroleptic, euphoriants) failed to occasion a maximum of >5% lorazepam-lever responding, even though some were drugs that share some pharmacological effects with Bzs (i.e., anticonvulsant activity, anesthetic effects, sleep induction, mood enhancement).

Differences across baboons in age and/or size, as well as genetic differences, may have produced differences in the pharmacokinetics of the test drugs across baboons. Such factors likely influenced the effective doses and sensitivity to changes in response rates across baboons. However, the differences in generalization profile from other Bz training conditions are not seen to be a function of the fact that the subjects were baboons. Studies in rats trained and tested with lorazepam have produced results similar to those in baboons (Ator and Griffiths, 1985, 1986, 1989a, b). A companion study to the present one in lorazepam-, diazepam- and pentobarbital-trained rats showed that the specificity of the lorazepam training condition in baboons for full Bz agonists can be replicated in rats (Ator and Griffiths, in preparation; cf. 1989a).

All the Bzs tested in the lorazepam-trained baboons were tested previously in one or more studies with rats trained to discriminate other Bzs: chlordiazepoxide (Colpaert et al., 1976; Gardner, 1989; Sanger and Benavides, 1993; Woudenberg and Slangen, 1989), diazepam (Shannon and Herling, 1983; Tang and Franklin, 1991; Wettstein and Gauthier, 1992, Young and Glennon, 1987), midazolam (Garcha et al., 1985; Rauch and Stolerman, 1987; Sannerud and Ator, 1995b; Woudenberg and Slangen, 1989), triazolam (Ator and Griffiths, 1989b). Some also were tested in midazolam- or oxazepam-trained pigeons (de la Garza et al., 1987; Evans and Johanson, 1989) and midazolam-trained squirrel monkeys (Spealman, 1985). All Bzs occasioned drug lever responding under these two-lever procedures, even when the training dose was manipulated. In a three-lever procedure, when rats were trained to discriminate among ND, 0.32 and 3.2 mg/kg midazolam conditions, responding shifted from the ND, to the low-dose and then to the high-dose lever as a function of dose for diazepam and triazolam. However, responding never shifted from the low- to the high-dose lever with chlordiazepoxide or lorazepam (Sannerud and Ator, 1995b). Thus, some evidence for differentiation among full agonist Bzs by Bz-trained animals was suggested. However, to date, only the results with chlordiazepoxide in the present study (and in lorazepam-trained rats, Ator and Griffiths, 1989a) suggest such a differentiation for a two-lever discrimination.

Most previous studies in which a Bz discrimination was trained with a drug other than lorazepam did not systematically test barbiturates and other non-Bz sedatives. However, the general finding has been that such drugs occasion the Bz-trained response, just as Bzs occasion the drug response in animals trained to discriminate a barbiturate (Ator and Griffiths, 1989a). Both pentobarbital and phenobarbital produced full generalization in chlordiazepoxide-trained rats (Ator and Griffiths, 1989a; Colpaert et al., 1976; De Vry and Slangen, 1986; Sanger et al., 1985); and pentobarbital did so in diazepam-trained rats (Ator and Griffiths, 1989b; Nierenberg and Ator, 1990; Shannon and Herling, 1983; Tang and Franklin, 1991), triazolam-trained rats (Ator and Griffiths, 1989b) and oxazepam-trained pigeons and rats (de la Garza et al., 1987; Hendry et al., 1983). Midazolam training conditions have produced mixed results: studies in rats showed full generalization to pentobarbital (Ator, 1990; Garcha et al., 1985; Rauch and Stolerman, 1987; Woudenberg and Slangen, 1989). Those in pigeons and squirrel monkeys did not find reliable generalization to pentobarbital or barbital, although pigeons did generalize to phenobarbital (Evans and Johanson, 1989; Spealman, 1985). Sannerud and Ator (1995a), with use of the three-lever procedure described above, suggested that this apparent cross-species difference may be a function of differences in effective midazolam training dose (i.e., taking route and cross-species differences in body weight into account). In the three-lever procedure, increasing doses of pentobarbital shifted rats from the ND to the low- but not to the high midazolam-dose lever.

Only a few studies with Bz training drugs other than lorazepam have tested sedatives or anxiolytics other than barbiturates, and these drugs typically have produced either full or partial generalization. The classic bis-carbamate anxiolytic meprobamate occasioned the drug response in chlordiazepoxide- or diazepam-trained rats (Nierenberg and Ator, 1990; Sanger et al., 1985) and in midazolam-trained pigeons (Evans and Johanson, 1989). Methaqualone produced full generalization in midazolam-trained pigeons (Evans and Johanson, 1989) and generalization in 40% of diazepam-trained rats (Haug and Gotestam, 1982). Partial generalization was produced by EtOH in diazepam-trained rats and pigeons (Jarbe and McMillan, 1983; Shannon and Herling, 1983). The present study found no generalization to those drugs, nor to the classic sedative-hypnotics clomethiazole or chloral hydrate nor the related compound triclofos (see Rall, 1990 for description); an earlier study found no lorazepam-lever responding in baboons or rats tested with the muscle relaxant methocarbamol (Sannerud et al., 1991). The present study did find partial generalization, similar to some barbiturates, to methyprylon, which is said to produce hypnotic effects very similar to secobarbital (Rall, 1990). As for the nonsedative/anxiolytic drugs tested, the present study and others have noted the specificity of the Bz generalization profile. Behaviorally active doses of haloperidol, morphine and phenytoin failed to occasion drug-lever responding in chlordiazepoxide-, midazolam- or diazepam-trained animals (Colpaert et al., 1976; Evans and Johanson, 1989; Sannerud and Ator, 1995a; Tang and Franklin, 1991). In rats the anticonvulsant sodium valproate did occasion 50% midazolam-lever responding (Rauch and Stolerman, 1987) and PCP occasioned partial diazepam-lever responding (Shannon and Herling, 1983). Thus, the full limits of the specificity of training with other Bzs have not been determined.

Other studies with a lorazepam training condition further suggested that it is unique in the selectivity of its generalization profile compared with other Bz training conditions. In studies with rats, Bz-receptor partial agonists (bretazenil, U-78875: Bronson, 1993; Rijnders et al., 1991; Sanger, 1987; Tang and Franklin, 1991) and neuroactive steroids (Ator et al., 1993, 1995) produced full generalization in animals trained to discriminate other full-agonist Bzs but not in lorazepam-trained rats or baboons. However, studies with some novel non-Bz compounds that do bind the Bz receptor (the ₁/Bz₁-selective ligands abecarnil, CL 218,872 and zolpidem; and the nonselective ligand zopiclone) showed full generalization in lorazepam-trained animals (Ator and Griffiths, 1986; Griffiths et al., 1992; Sannerud et al., 1992), as they have in animals trained to discriminate other Bzs (Andrews and Stephens, 1991; Gardner, 1989; Tang and Franklin, 1991; review in Sanger et al., 1994). Taken together, the results suggest that animals trained to discriminate lorazepam are most likely to generalize only to compounds that are full agonists at the Bz receptor. Further differentiation between the ₁/Bz₁ and ₂/Bz₂ receptors is not possible yet, given the fact that compounds selective for the Bz₂ receptor have not been developed. However, it is intriguing that a compound that has been suggested to show more affinity for the ₂/Bz₂ than the ₁/Bz₁ receptor, chlordiazepoxide (Sanger and Benavides, 1993), occasioned less reliable generalization across baboons than the other Bzs in the present study; and this result has occurred in studies with lorazepam-trained rats as well (Ator and Griffiths, in preparation, cf. 1989a).

In studies of lorazepam's behavioral effects, a few striking differences from other Bzs have been reported. Babbini et al. (1979) showed that in rats studied in a Geller-Seifter procedure, lorazepam was the only Bz out of 12 for which the ED₅₀ for increasing punished responding was higher than that for decreasing unpunished responding. Similarly, Steru et al. (1986) found that, alone among 11 Bzs, lorazepam's potency in increasing the punished behavior of mice in the four-plate test was so low that anxiolytic efficacy could not be concluded. Both studies found that lorazepam was either the most or second most potent in decreasing general motor activity. Although studies of human performance or subjective effects have not shown lorazepam to be different from other Bzs with which it was compared (summarized in Rush et al., 1993), lorazepam appeared to have a greater effect on certain aspects of memory than diazepam or oxazepam (Curran et al., 1987; Vidailhet et al., 1996). Clinically, lorazepam decreased rapid-eye-movement sleep more than triazolam or flurazepam (Roth et al., 1980); but in general, lorazepam is not distinguished from the other Bzs in clinical use (American Psychiatric Association, 1990; Griffiths and Weerts, in press, 1997).

Lorazepam itself has not, to date, shown characteristics markedly distinctive biochemically from other Bzs. Like oxazepam, it is a 3-hydroxy-1,4-Bz and is most similar to oxazepam in its metabolic route (i.e., single-step glucuronidation, which yields pharmacologically inactive metabolites; Greenblatt, 1981). Yet animals that have been trained to discriminate oxazepam generalized to pentobarbital (de la Garza et al., 1987; Hendry et al., 1983). Absorption of lorazepam after oral doses in humans was complete within 2.5 h, and mean absorption half-life was less than 30 min; elimination half-life after single and multiple oral doses was similar and the mean was 15 h (review in Greenblatt, 1981). Selectivity in lorazepam binding between ₁/Bz₁ and ₂/Bz₂ receptors, nor among GABA_A-receptor subtypes, has not been reported. However, lorazepam has not often been included in studies of the receptor activity of Bzs. In a binding study comparing different structures of rat central nervous system, lorazepam was most potent in the cerebellum, which is high in ₁/Bz₁ receptors, compared with all other brain regions studied (Sanger and Benavides, 1993). This result differed from the other Bzs: that is, except for triazolam, which was most potent in the spinal cord, other Bzs were not differentially potent across brain regions. As further research in Bz biochemistry occurs, it will be interesting to determine whether further distinctions of lorazepam from other Bzs emerge that might subserve the distinction profile of lorazepam as a discriminative stimulus.

Finally, it often has been of interest to determine the extent to which results in particular drug discrimination paradigms correlate with those from self-administration procedures. Most of the drugs evaluated in the present study also have been studied in a standard intravenous self-administration procedure in baboons that permits categorization of drugs in terms of low, moderate and high levels of self-injection of test drugs. Results from this and other self-administration procedures have shown whether a drug readily serves as a reinforcer, which has been predictive of whether the drug is likely to be abused (Griffiths et al., 1980; Katz and Goldberg, 1988). In the standard paradigm in