Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Benzodiazepines and barbiturates bind to distinct sites on -aminobutyric acid_A (GABA_A) receptor complexes where they facilitate the actions of GABA. Drugs that enhance GABA-stimulated chloride influx are called positive GABA_A modulators, whereas drugs that inhibit chloride influx are called negative GABA_A modulators; drugs that bind to benzodiazepine receptors and do not modify GABA-stimulated chloride influx are called neutral GABA_A modulators (Obata et al., 1988). Although positive modulators reduce anxiety and produce sedation, their profile of effects differs slightly, depending on the site on the GABA_A receptor complex with which they interact. As a result, benzodiazepines have a larger margin of safety than barbiturates; unfortunately, there are still adverse effects associated with their use, including dependence liability (Busto et al., 1986). One approach that has been used in an attempt to retain the therapeutic effectiveness of benzodiazepines while improving their margin of safety is to develop compounds that are selective for benzodiazepine receptor subtypes. At least two subtypes have been hypothesized: BDZ-1 and BDZ-2 (Sanger et al., 1994). Compounds that bind selectively to the BDZ-1 receptor subtype (e.g., zolpidem) are thought to have lower dependence liability than compounds that are not selective for particular subtypes (Sanger and Zivkovic, 1992; Richards and Martin, 1998; Weerts et al., 1998).

Quantitative (i.e., Schild) analyses can be useful for evaluating differences in selectivity among drugs (e.g., opioids). When the slope of a Schild plot does not differ from unity, a simple, competitive interaction at a single (sub)type of receptor can be inferred. Differences in pA₂ values for a given antagonist can indicate that the observed effects are mediated by different receptors and, therefore, can detect variations in selectivity among drugs. Unfortunately, few studies have applied these analyses to drugs that act at benzodiazepine receptors. One study examined the ability of five compounds to antagonize the response rate-decreasing effects of the positive GABA_A modulator midazolam in squirrel monkeys (Paronis and Bergman, 1999). Schild analyses yielded slopes that conformed to unity for four drugs, including the neutral GABA_A modulator flumazenil (slope = 0.85; pA₂ = 7.18). For the fifth compound, -carboline-3-carboxylate-t-butyl ester (-CCt), Schild analyses yielded a slope that was less than unity (slope = 0.68), suggesting that -CCt and midazolam are not interacting in a simple and competitive manner. Because -CCt appears to be selective for the BZD-1 subtype of benzodiazepine receptor (Shannon et al., 1984), one possibility is that the effects of midazolam are mediated by more than one benzodiazepine receptor subtype and -CCt does not bind to one of those subtypes. Other studies have reported slopes that differ from unity. For example, Woolverton and Nader (1995) studied flumazenil in combination with the positive GABA_A modulator diazepam in two rhesus monkeys and obtained slopes of 1.34 and 1.49, again suggesting that the interaction was not simple and competitive. In a study with rats, the antagonism of diazepam by flumazenil was orderly (Herling and Shannon, 1982); however, post hoc Schild analyses of those data yielded an apparent pA₂ value of 4.7 with a slope of 1.5 (Rowlett and Woolverton, 1996). These analyses support the notion that neutral GABA_A modulators do not consistently antagonize positive GABA_A modulators in a simple and competitive manner at a homogeneous receptor population. Furthermore, post hoc analyses (Rowlett and Woolverton, 1996) of studies with positive GABA_A modulators in combination with negative GABA_A modulators in rats (Shannon and Davis, 1984; Kunchandy and Kulkarni, 1986; Shannon and Katzman, 1986; Shannon et al., 1988) reported slopes that differed from unity. Although affinity estimates would be expected to vary depending on endogenous tone or the efficacy of a negative modulator (Kenakin, 1993), these factors alone should not affect the slope of the Schild analysis. Deviations from unity (1) would be expected if more than one (sub)type of receptor mediates the effect of interest; however, a paucity of data on this topic, particularly for GABA modulators, precludes an identification of the factors that contribute to variations in slope and the resulting failure to satisfy the assumptions of Schild analysis.

Previous studies showed that the slope of the Schild plot for flumazenil in combination with midazolam was different from unity (Lelas et al., 1999), suggesting that the discriminative stimulus effects of midazolam in rhesus monkeys might be mediated by more than one benzodiazepine receptor subtype. This study attempted to extend those observations by conducting Schild analyses for flumazenil, as well as the negative GABA_A modulators Ro 15-4513 and -CCE, in combination with the positive GABA_A modulators midazolam, triazolam, and diazepam. These studies specifically tested the hypothesis that a single benzodiazepine receptor subtype mediates the discriminative stimulus effects of benzodiazepine-positive GABA_A modulators in rhesus monkeys.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects. Two adult and one juvenile female and one adult male rhesus monkey (Macaca mulatta) were housed individually with free access to water in a colony room maintained on a 14-h light and 10-h dark schedule. Diet comprised primate chow (Harlan Teklad High Protein Monkey Diet, Madison, WI), fresh fruit, and peanuts; sufficient quantities of food were provided to maintain body weight (adults) or to allow for normal growth (juvenile). All four monkeys previously discriminated 0.1 mg/kg triazolam from vehicle and, subsequently, midazolam from vehicle (Lelas et al., 1999). The animals were maintained in accordance with the Institutional Animal Care and Use Committee, Louisiana State University Health Sciences Center, and guidelines of the Committee on Care and Use of Laboratory Animal Resources, National Research Council (Department of Health, Education and Welfare, Publication No. (NIH) 85-23, revised 1983).

Apparatus. During experimental sessions, monkeys were seated in Lexan and aluminum primate chairs that provided restraint at the neck and waist; the chairs were placed in well-ventilated, sound-attenuating chambers. A response panel contained a stimulus light located above each of the three response levers; during response periods, two of the stimulus lights were illuminated and two corresponding levers were active. For two monkeys, the inactive lever was the right lever; for the remaining two monkeys, the inactive lever was the center lever, and for those monkeys, it was retracted throughout experimental sessions. At the front of the chair, the feet of the monkeys were restrained in shoes containing brass electrodes to which brief electric shock (3 mA, 250 ms) could be delivered from an a.c. shock generator located outside the chamber. The experiments were controlled and data collected by a microprocessor (Dell computer) and commercially available interface (MedAssociates, St. Albans, VT) with MedAssociates software.

Procedure. All four monkeys were previously trained to discriminate 0.56 mg/kg midazolam from vehicle under a fixed-ratio 5 (FR5) schedule of stimulus-shock termination (SST) (Lelas et al., 1999). Training sessions consisted of two to seven 15-min cycles. The first 10 min of a cycle was a pretreatment period during which the chamber was dark and responses had no programmed consequence; the last 5 min of a cycle was a response period during which the schedule of SST was in effect. The beginning of the response period was signaled by illumination of two stimulus lights; in the presence of this visual stimulus, shock was scheduled to occur every 10 s. Five consecutive responses on the lever designated correct by the injection administered during the 1st min of the cycle extinguished the stimulus lights and postponed the shock schedule for 30 s. The selection of vehicle- and drug-appropriate levers varied among monkeys and remained the same for an individual throughout the study. At the end of the 30-s time-out period, stimulus lights were illuminated and the SST schedule was again in effect. Responding on the incorrect lever reset the response requirement on the correct lever. Response periods ended either 5 min after stimulus lights were initially illuminated for that cycle (i.e., the 30-s time-out periods were included in the 5-min response periods) or after the delivery of four shocks, whichever occurred first. The total length of cycles, as well as the interval between injections, was always 15 min; if the response period ended before 5 min had elapsed, the time remaining between the response period and the beginning of the next cycle was a time-out period.

Data Analyses. Drug discrimination data are expressed as the percentage of total responses occurring on the drug-appropriate lever averaged among monkeys and plotted as a function of dose (mean ± 1 S.E.). Drugs that produced 80% responding on the drug-appropriate lever were considered to have substituted completely for the training stimulus. When a particular test was conducted more than once, the two determinations were averaged for an individual subject before other analysis. ED₅₀ values were calculated for each subject by linear regression when more than two data points were available; otherwise by interpolation. Control response rates represent the average of the five vehicle training sessions before the test. Response rates were calculated as a percentage of control rates for individual animals, then averaged among animals and plotted as a function of dose (mean ± 1 S.E.). Discrimination data for an individual were not included in analyses when response rates were less than 20% of control; however, response rates are included in the figures. For the Schild analyses, the ED₅₀ values were calculated for individual subjects, then averaged among all animals; these averages were used to calculate dose ratios for each dose of the antagonist. Rates of responding were seldom decreased to <50% of control rates, thereby precluding determination of ED₅₀ values for rate of responding; therefore, pA₂ analyses were not performed on these data. The pA₂ analyses were carried out with the Pharm/PCS Pharmacological Calculation system (version 4.2) based on Tallarida and Murray (1987). Slopes of Schild plots were considered to conform to unity when the 95% CI included 1 and did not include 0 (Paronis and Bergman, 1999). The pA₂ analyses could not be conducted for data obtained when -CCE was studied in combination with triazolam because mean ED₅₀ values for triazolam could be determined for only two of the three doses of -CCE. Therefore, single-dose apparent affinity estimates for -CCE in combination with triazolam were calculated with a modified equation of Tallarida et al., (1979) where pK_B = log[B/dose ratio  1] with B expressed in moles per kilogram of body weight.

Drugs. The following drugs were administered in this study: midazolam hydrochloride (Roche Pharma Inc., Manati, Puerto Rico); triazolam (gift from Pharmacia and Upjohn, Kalamazoo, MI); diazepam, Ro 15-4513 (ethyl 8-azido-6-dihydro-5-methyl-6-oxo-4H-imidazo[1,5-a]-[1,4]benzodiazepine-3-carboxylate), and -CCE (ethyl -carboline-3-carboxylate; Research Biochemicals International, Natick, MA); and flumazenil (a gift from F. Hoffmann LaRoche Ltd., Basel, Switzerland). Midazolam was purchased as a commercially prepared solution in a concentration of 5 mg/ml and diluted with water. Diazepam, triazolam, and -CCE were dissolved in a vehicle comprising 20% emulphor, 10% ethanol, and 70% water. Flumazenil and Ro 15-4513 were dissolved in a vehicle comprising 40% propylene glycol, 50% saline, and 10% ethanol. Stock solutions were prepared every 2 to 3 weeks. Drugs were administered s.c., typically in a volume of 0.1 ml/kg b.wt. Doses are expressed in terms of the forms listed above.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Under control conditions, response rates (±1 S.E.) were 1.80 ± 0.33 responses/s. Triazolam, diazepam, and midazolam dose dependently increased midazolam-lever responding with doses larger than 0.032 mg/kg triazolam, 3.2 mg/kg diazepam, or 0.1 mg/kg midazolam producing >80% midazolam-lever responding (filled circles, top, Figs. 1-3). The relative potencies of the three positive GABA_A modulators (ED₅₀ ± 1 S.E.) for discriminative stimulus effects were triazolam (0.04 ± 0.01 mg/kg) > midazolam (0.16 ± 0.02 mg/kg) > diazepam (1.92 ± 0.50 mg/kg). At doses of the positive modulators that produced >80% midazolam-lever responding, mean response rates were >40% of control (filled circles, bottom, Figs. 1-3). The neutral (flumazenil) and negative (Ro 15-4513 and -CCE) GABA_A modulators neither substituted for midazolam nor decreased response rates in monkeys discriminating midazolam from saline (points above V, top and bottom, Figs. 1-6). In fact, -CCE dose dependently increased response rates to >130% of control at a dose of 1.0 mg/kg (points above V, bottom, Figs. 5 and 6). Increased response rates also were observed after administration of flumazenil or Ro 15-4513, although these effects were neither dose-related nor consistent among determinations (points above V, bottom, Figs. 1-4).

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Discriminative stimulus and rate effects of triazolam alone and in combination with several doses of flumazenil in rhesus monkeys discriminating between 0.56 mg/kg midazolam and vehicle. Abscissae: dose in milligrams per kilogram body weight; V, vehicle. Ordinates: mean (±1 S.E.) percentage of responding on the drug-appropriate lever (%DR = drug responding, top) and mean (±1 S.E.) response rates expressed as percentage of control (vehicle training days) rates [rate (% of control), bottom]. Triazolam dose-effect curves determined in the absence of other treatment, in the presence of 0.01 mg/kg flumazenil, or in the presence of 1.0 mg/kg flumazenil represent an average of two determinations in each of three monkeys and one determination in the fourth monkey. Triazolam dose-effect curves obtained in the presence of 0.032, 0.1, or 0.32 mg/kg flumazenil represent an average of two determinations in each of four monkeys. The triazolam dose-effect curve obtained in the presence of 1.78 mg/kg flumazenil represents an average of one determination in three monkeys.

Flumazenil Antagonism. Flumazenil dose dependently antagonized the discriminative stimulus and rate-decreasing effects of triazolam and diazepam. Doses of 0.032, 0.1, 0.32, and 1.0 mg/kg flumazenil shifted the triazolam dose-effect curve for the discriminative stimulus effects 1.8-, 4.0-, 7.0-, 21.3-, and 55.8-fold to the right, respectively (Fig. 1, top). After administration of 1.78 mg/kg flumazenil, only two monkeys responded >50% on the midazolam-appropriate lever up to the largest dose of triazolam that could be studied due to its solubility. Schild analysis yielded an apparent pA₂ value (95% CI) for flumazenil in combination with triazolam of 7.41 (7.24, 7.58) with a slope (95% CI) of 0.90 (1.04, 0.77; r = 0.99) for discriminative stimulus effects. All doses of flumazenil shifted the triazolam dose-effect curve for the rate-decreasing effects to the right (Fig. 1, bottom), although generally response rates were not decreased to <50% of control after administration of 0.01 mg/kg flumazenil in combination with triazolam.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Discriminative stimulus and rate effects of diazepam alone and in combination with several doses of flumazenil in rhesus monkeys discriminating midazolam and vehicle. The diazepam dose-effect curve obtained in the absence of other treatment represents an average of two determinations in each of three monkeys. Diazepam dose-effect curves obtained in the presence of 0.01, 0.032, or 0.1 mg/kg flumazenil represent an average of two determinations in two monkeys and one determination in another monkey. The diazepam dose-effect curve determined in the presence of 0.178 mg/kg flumazenil represents one determination in three monkeys. For other details, see Fig. 1.

Ro 15-4513 Antagonism. Ro 15-4513 dose dependently antagonized the discriminative stimulus and rate-decreasing effects of midazolam and triazolam. Doses of 0.01, 0.032, 0.1, and 0.32 mg/kg Ro 15-4513 shifted the midazolam dose-effect curve for the discriminative stimulus effects 1.9-, 4.5-, 5.2-, and 14.0-fold to the right, respectively (Fig. 3, top). The largest dose of Ro 15-4513 appeared to produce a further shift to the right in the midazolam dose-effect curve; however, only two of the four monkeys responded >50% on the midazolam-appropriate lever when 1.0 mg/kg Ro 15-4513 was studied in combination with midazolam. Larger doses of midazolam could not be studied because concentrations of midazolam larger than 5 mg/ml were not available. Schild analysis yielded an apparent pA₂ value (95% CI) for Ro 15-4513 in combination with midazolam of 7.53 (6.77, 8.31) with a slope (95% CI) of 0.72 (1.30, 0.14; r = 0.97) for discriminative stimulus effects. All doses of Ro 15-4513 shifted the midazolam dose-effect curve for the rate-decreasing effects to the right (Fig. 3, bottom); however, in most cases response rates were not decreased to <50% of control.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Discriminative stimulus and rate effects of midazolam alone and in combination with several doses of Ro 15-4513 in rhesus monkeys discriminating midazolam and vehicle. Midazolam dose-effect curves obtained in the absence of other treatment or in the presence of 0.032 mg/kg Ro 15-4513 represent an average of two determinations in each of four monkeys. Midazolam dose-effect curves obtained in the presence of 0.01, 0.1, or 0.32 mg/kg Ro 15-4513 represent the average of two determinations in each of three monkeys and one determination in a fourth monkey. The midazolam dose-effect curve determined in the presence of 1.0 mg/kg Ro 15-4513 represents an average of two determinations in two monkeys and one determination in two other monkeys. For other details, see Fig. 1.

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Discriminative stimulus and rate effects of triazolam alone and in combination with several doses of Ro 15-4513 in rhesus monkeys discriminating midazolam and vehicle. Triazolam dose-effect curves determined in the absence of other treatment or in the presence of 0.032 or 0.1 mg/kg Ro 15-4513 represent an average of two determinations in each of the three monkeys and one determination in the fourth monkey. Triazolam dose-effect curves determined in the presence of 0.32 and 1.0 mg/kg Ro 15-4513 represent an average of two determinations in two monkeys and one determination in the remaining two monkeys. The triazolam dose-effect curve determined in the presence of 1.78 mg/kg Ro 15-4513 represents an average of one determination in each of three monkeys. For other details, see Fig. 1.

-CCE Antagonism. -CCE dose dependently antagonized the discriminative stimulus and rate-decreasing effects of midazolam and triazolam. Doses of 0.1 and 1.0 mg/kg -CCE shifted the midazolam dose-effect curve for the discriminative stimulus effects 3.8- and 10.1-fold to the right, respectively (Fig. 5, top). Only three of the four monkeys responded >50% on the midazolam-appropriate lever when midazolam was studied in combination with 0.32 mg/kg -CCE; however, in the three monkeys in which ED₅₀ values could be determined, this dose of -CCE shifted the midazolam dose-effect curve 8.4-fold to the right. Schild analysis yielded a slope [0.52 (3.04, 2.01); r = 0.93] that deviated from unity, thereby precluding determination of an apparent pA₂ value for -CCE in combination with midazolam. All doses of -CCE shifted the midazolam dose-effect curve for the rate-decreasing effects to the right (Fig. 5, bottom), although in most cases response rates were not decreased to <50% of control.

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Discriminative stimulus and rate effects of midazolam alone and in combination with several doses of -CCE in rhesus monkeys discriminating midazolam and vehicle. The midazolam dose-effect curve obtained in the absence of other treatment represents the average of two determinations in each of four monkeys. Midazolam dose-effect curves obtained in the presence of 0.1, 0.32, or 1.0 mg/kg -CCE represent the average of two determinations in three monkeys and one determination in the remaining monkey. For other details, see Fig. 1.

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Discriminative stimulus and rate effects of triazolam alone and in combination with several doses of -CCE in rhesus monkeys discriminating midazolam and vehicle. The triazolam dose-effect curve obtained in the absence of other treatment represents an average of two determinations for each of the three monkeys and one determination in the fourth monkey. Triazolam dose-effect curves obtained in the presence of 0.1, 0.32, or 1.0 mg/kg -CCE represent the average of two determinations in two monkeys and one determination in the two remaining monkeys.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Flumazenil is a neutral modulator at the benzodiazepine receptor on the GABA_A receptor complex and it has been shown to antagonize the effects of benzodiazepines under a variety of conditions (Bonetti et al., 1982; Herling and Shannon, 1982; Paronis and Bergman, 1999). Consequently, flumazenil has proven to be a useful tool for distinguishing among positive GABA_A modulators because it antagonizes the effects of benzodiazepines and not the effects of barbiturates (Bonetti et al., 1982; Herling and Shannon, 1982). Previous studies in this and other laboratories have shown that flumazenil antagonized the discriminative stimulus and rate-decreasing effects of benzodiazepines (Rowlett and Woolverton, 1996; Lelas et al., 1999; Paronis and Bergman, 1999). In each case, the antagonism was orderly and dose-related with Schild analyses yielding high regression coefficients. However, Schild analyses yielded slopes that conformed to unity for only one set of data (Paronis and Bergman, 1999). When flumazenil was studied in combination with midazolam, under conditions identical with those used in the current experiment (Lelas et al., 1999), the analyses yielded a slope that deviated from unity [slope (95% CI) = 0.79 (0.88, 0.70)], suggesting that the discriminative stimulus effects of midazolam could be mediated by more than one benzodiazepine receptor subtype. The present experiment was designed to extend these findings to determine whether a single benzodiazepine receptor subtype mediates the discriminative stimulus effects of midazolam, triazolam, and diazepam.

To further evaluate the discriminative stimulus effects of triazolam and diazepam, each positive GABA_A modulator was studied in combination with flumazenil. Triazolam and diazepam dose dependently substituted for the midazolam discriminative stimulus in rhesus monkeys discriminating midazolam from vehicle. The dose-effect curves for the discriminative stimulus and rate-decreasing effects of triazolam and diazepam were shifted to the right by flumazenil. The shifts in the triazolam and diazepam dose-effect curves for the discriminative stimulus effects obtained with increasing doses of flumazenil were orderly and dose-dependent (triazolam, r = 0.99; diazepam, r = 0.99). The conformity of the triazolam slope of 0.90 and the diazepam slope of 0.82 to unity suggests that the discriminative stimulus effects of each of the two positive GABA_A modulators are mediated by homogeneous receptor populations, and the similarity in apparent pA₂ value for triazolam (7.41) to that of diazepam (7.69) further suggests that the discriminative stimulus effects of triazolam and diazepam are mediated by the same receptor population.

Results of Schild analyses in the current and previous (Lelas et al., 1999) studies might suggest that the discriminative stimulus effects of triazolam and diazepam are not identical with those of midazolam. Although the antagonism of midazolam by flumazenil was extremely orderly (Lelas et al., 1999), the apparent deviation in slope from unity suggested that the discriminative stimulus effects of midazolam might be mediated by more than one subtype of benzodiazepine receptor. However, the conformity of slopes of Schild plots to unity for flumazenil studied in combination with either triazolam or diazepam prompted a reevaluation of data from the previous study. In the previous study there was an unequal number of observations for each subject under each condition and these multiple determinations were treated as independent observations. In contrast, for this study multiple determinations were averaged for an individual subject first and mean ED₅₀ values were then calculated among subjects. This seemingly minor difference in the method used to analyze dose ratios generated markedly different confidence limits for slope estimates. With the method of analysis from this study used to analyze data from the previous study (Lelas et al., 1999), we obtained a slope (95% CI) of 0.76 (1.32, 0.19); although the slope of the Schild plot is not markedly different between the two methods of calculation [0.79 (Lelas et al., 1999) versus 0.76], the 95% CI for the current method of analyses suggests that the slope conforms to unity. The corresponding pA₂ value (95% CI) for flumazenil in combination with midazolam was 7.83 (7.12, 8.54). Thus, when the data are analyzed with the same method, the discriminative stimulus effects of midazolam, triazolam, and diazepam appear to be mediated by a single subtype of benzodiazepine receptor and similarities among pA₂ values suggest that the same subtype mediates these effects of the three positive GABA_A modulators. Clearly, additional studies are needed to clarify this issue and to further test the utility of these analyses to these dependent variables.

Ro 15-4513, a negative GABA_A modulator, dose dependently shifted the dose-effect curves for the discriminative stimulus and rate-decreasing effects of midazolam and triazolam to the right. As was the case with antagonism by flumazenil, the shifts were orderly and dose-dependent (midazolam, r = 0.97; triazolam, r = 0.98). Given the conformity of slopes obtained with Schild analyses to unity, a single subtype of benzodiazepine receptor appears to mediate the discriminative stimulus effects of midazolam as well as the effects of triazolam. Moreover, the apparent pA₂ values obtained for Ro 15-4513 in combination with midazolam (7.53) or triazolam (6.88) suggest that these effects of the two positive modulators are mediated by the same receptor subtype.

-CCE, a negative GABA_A modulator, dose dependently shifted the dose-effect curves for the discriminative stimulus and rate-decreasing effects of midazolam and triazolam to the right. The shifts to the right in the midazolam dose-effect curve for the discriminative stimulus effects obtained with -CCE were orderly (r = 0.94) and dose-dependent. The slope of the Schild plot for -CCE in combination with midazolam was 0.52; however, the upper and lower limits of the 95% CI were 2.01 and 3.04, respectively, indicating that the slope deviated significantly from unity. -CCE has been shown to have negative efficacy (Braestrup et al., 1982; Barrett et al., 1985; Shimada et al., 1995; Kitano et al., 1996), which might influence apparent affinity estimates but should not cause the slope of a Schild analysis to deviate from unity. In addition, the notion that negative efficacy accounts for this deviation is not supported by data obtained with -CCE studied in combination with triazolam. Only two doses of -CCE effectively antagonized triazolam, thereby precluding the use of Schild analyses for this drug combination; however, pK_B values obtained with those two doses yielded similar values (6.49 and 6.35), which is consistent with the view that -CCE interacts with triazolam in a simple, competitive manner. Thus, some factor other than the efficacy of the negative modulator must account for the deviation in slope for -CCE with midazolam. One possibility is that -CCE and midazolam interact with more than one subtype of benzodiazepine receptor. Studies with receptor-selective GABA_A modulators such as zolpidem (Arbilla et al., 1985), or possibly -CCt, will need to be conducted to determine whether the antagonism by -CCE is due to actions at multiple receptor subtypes.

In conclusion, these data demonstrate that flumazenil, Ro 15-4513, and -CCE dose dependently antagonize the discriminative stimulus and rate-decreasing effects of midazolam, triazolam, and diazepam. In the case of flumazenil in combination with midazolam, triazolam and diazepam, and Ro 15-4513 in combination with midazolam and triazolam, the slopes of the Schild plots conformed to unity, suggesting that the discriminative stimulus effects of each positive GABA_A modulator are mediated by a single benzodiazepine receptor subtype. In contrast, -CCE does not appear to interact with midazolam in a simple and competitive manner. Together with other studies in vivo, these data highlight the need for better pharmacological tools that would facilitate definitive studies on the relative importance of benzodiazepine receptor subtypes in the therapeutic and abuse-related effects of positive GABA_A modulators.