GR89,696: A Potent κ-Opioid Agonist with Subtype Selectivity in Rhesus Monkeys

Materials and Methods

Subjects

Captive-bred, intact rhesus monkeys of either sex (Macaca mulatta; weight range 4.5–11 kg) were singly housed in rooms maintained at 20–22°C with controlled humidity, and a 12:12-h light/dark cycle (lights on at 7:00 AM). Monkeys used in serum prolactin (six females) and antinociception studies were fed standard primate chow biscuits (Purina, Richmond, VA) daily, supplemented by fruit two times per week. Monkeys in the drug discrimination experiments were fed appropriate amounts to maintain body weight at approximately 90% of free-feeding levels. Water was freely available in home cages, via an automatic waterspout. Animals used in these studies were maintained in accordance with the Institutional Animal Care and Use Committees of Rockefeller University and the University of Michigan, and Guidelines of the Committee on the Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Health Council (Department of Health, Education and Welfare, Publication ISBN 0-309-05377-3, revised 1996).

Radioligand Binding Studies

Procedure.

Rhesus monkey cortical membranes were prepared in Tris-HCl buffer according to the methods of Emmerson et al. (1994). Membranes (400 μg of protein) were incubated in Tris-HCl buffer (50 mM, pH 7.4) with appropriate radioligand in a total volume of 1 ml for 60 min at 25°C with increasing concentrations of GR89,696. μ-Opioid receptor sites were labeled with [³H]DAMGO (1.06 nM) and δ-opioid receptor sites were labeled with [³H]DPDPE (1.8 nM). κ-Opioid sites were labeled by [³H]U69,593 (1 nM) or with [³H]bremazocine (1 nM) in the presence of 1 μM DAMGO and 1 μM DPDPE to prevent binding to μ- and δ-opioid receptor sites, respectively. A standard concentration of 1 to 1.8 nM ligand was used for each assay as the level of³H-labeled ligand was taken into account in determining K_i values of competing ligands. In binding experiments, nonspecific binding was defined with naloxone (10 μM). Bound and free ligands were separated by vacuum filtration through glass fiber filters and quantified by liquid scintillation counting.

Data Analysis.

IC₅₀ values were determined with GraphPad Prism (version 2.0; GraphPad, San Diego, CA) and converted to K_i values according to the Cheng-Prusoff equation. The K_dvalues for [³H]U69,593 (0.95 nM) and [³H]DAMGO (0.57 nM) are from Emmerson et al. (1994). The K_d value for [³H]bremazocine (0.12 nM) in the presence of DAMGO and DPDPE (1 μM each; to mask μ- and δ-receptors, respectively) was determined from saturation binding analysis (data not shown).

[³⁵S]GTPγS Assay

GR89,696-induced stimulation of [³⁵S]GTPγS binding was studied in a C6 cell line expressing the rat μ-receptor and in CHO cells expressing the human κ-receptor, as described previously (Traynor and Nahorski, 1995; Zhu et al., 1997). Briefly, membranes were incubated for 1 h with [³⁵S]GTPγS (100 pM), in the presence of several concentrations of GR89,696. Control (unstimulated) binding of [³⁵S]GTPγS was studied in the absence of GR89,696. Maximal stimulation of [³⁵S]GTPγS binding was determined with 10 μM DAMGO (μ-receptors) or 10 μM U69,593 (κ-receptors). The effect of GR89,696 in this assay was expressed as a percentage of the maximal effect of the above-mentioned reference compounds, which are high-efficacy agonists in their respective cell lines. To study the antagonism of μ-receptor-mediated GTPγS binding, DAMGO, at an approximately 80% effective concentration (100 nM), was incubated as described above (100 pM GTPγS, 1 h) in the presence of increasing concentrations of GR89,696 (0.001 nM-100 μM).

Data Analysis.

ED₅₀ values for the stimulation of GTPγS were determined using nonlinear regression analysis (GraphPad Prism). The AD₅₀ value of GR89,696 (concentration of antagonist reversing the effect of DAMGO by 50%) was determined using regression analysis (GraphPad Prism), and data were compared using a Student's t test.

EKC Discrimination

The procedure has been described in detail previously (Butelman et al., 1999d).

Apparatus.

All experiments used similar operant panels consisting of two primate response levers (BRS-LVE model PRL-001; Laurel, MD). The levers required a force of 0.25 N over a distance of 2 mm for a response to be recorded. A panel of 7.5-W stimulus lights was located above, and a food receptacle located between, the levers, and both were mounted to one wall of the testing chamber. Delivery of 300-mg banana-flavored pellets (Noyes formula G/T; Lancaster, NH) was controlled by an externally mounted food dispenser (Gerbrands model G5210; Arlington, MA). A PC-compatible computer connected to a custom-made interface controlled the scheduling of events and recorded data. A lever press was counted as a response.

Procedure.

Chair-trained monkeys (n = 3) were trained to respond for pellets under a fixed ratio 30 (FR 30) schedule for food reinforcement. Subsequently, discrimination training was undertaken in which reinforcer delivery was made contingent upon the selection of the lever previously paired with a drug stimulus. For all subjects, EKC (0.0056 mg/kg s.c.) was paired with the right, and saline with the left, lever. Training sessions consisted of two to five cycles. Each cycle comprised two periods: an initial 10-min time-out period in which light stimuli were extinguished and responding had no consequence. This was followed by a 5-min response period in which light stimuli were illuminated, and monkeys could obtain up to 10 pellets (reinforcers) through completion of the FR 30 schedule of reinforcement on the drug-appropriate lever. If 10 reinforcers were obtained prior to the end of the response period, the stimulus lights were extinguished and responding had no consequences. Monkeys were trained five days/week and were not tested until greater than 80% drug-appropriate responding was attained prior to the first reinforcer delivery and across the session; this criterion was maintained across four of five consecutive training sessions. In addition, no tests were performed unless response rates were maintained within 20% across four of five sessions.

Test sessions were performed as described above, with a 15-min cycle (10-min time-out, 5-min response period), and reinforcer delivery was made contingent upon completion of the fixed ratio schedule regardless of the lever selected. For GR89,696, both single (time course) and cumulative dosing procedures were used. During single-dosing procedures, the first response period commenced 5 min after drug administration, and discrete 15-min cycles were performed thereafter, continuing for 4 h after drug administration. During cumulative dosing procedures, vehicle or drugs were administered at the beginning of consecutive 15-min cycles.

Data Analysis.

For discrimination data, lever selection was expressed as percentage of EKC responding by dividing the number of responses on the EKC lever by the total number of responses. The percentage of control response rates for individual monkeys was calculated by dividing the response rates for each cycle by the saline control response rates for that session, and are presented graphically. Generalization was considered to occur if a subject emitted at least 80% of responses on the EKC-appropriate lever.

U69,593 (U69) Discrimination

Apparatus.

Two-lever primate operant boxes were custom built by MED Associates (Georgia, VT), and connected with a MED Associates interface to a PC computer. Chaired subjects were placed in the boxes; they had within easy reach the two levers and a food hopper. A light was lit above each of the levers to signal food availability during sessions, and a house light was also lit during the period of food availability.

Procedure.

A separate group of three subjects was trained to discriminate s.c. injections of U69,593 from vehicle, in a food-reinforced FR 20 operant procedure (sessions were carried out 5 days/week). U69,593 training doses were 0.0056 mg/kg for one subject and 0.013 mg/kg for the two other subjects. Subjects were trained essentially as described above, with some modifications. Each cycle commenced with a s.c. injection of drug or vehicle and was followed by a 15-min time-out, during which responses had no consequence. This time-out was followed by a 5-min response period. Ten reinforcers were available during each response period, and these occurred after 20 responses on the injection-appropriate lever (FR 20). Before the onset of testing, subjects had to satisfy the following performance criteria: five consecutive sessions had to reach at least 90% injection-appropriate responding, with response rates above 1 response/s. After the initial training criteria were met, subjects were tested after a minimum of two consecutive sessions meeting the above-mentioned criteria. Test sessions were identical to that described above, except that 20 responses resulted in a food presentation, irrespective of the injection.

Design and Data Analysis.

Subjects were tested in time course sessions; one injection was therefore followed by response periods starting at different intervals (e.g., 5, 15, 30, 60 min after injection). A vehicle control experiment was completed in each subject. Different doses of U69,593 (0.0032, 0.01 mg/kg; 1–2 determinations at each dose) and GR89,696 (0.00001–0.00032 mg/kg; 2–3 determinations at each dose) were studied for each compound. Mean ± S.E.M. data are presented for %U69-appropriate responding and rate of responding (responses/s). Generalization was considered to occur if a subject emitted at least 90% of responses on the U69-appropriate lever.

Warm Water Tail Withdrawal Assay (Antinociception)

Apparatus and Procedure.

The procedure has been described in detail previously (Dykstra et al., 1987a). Monkeys were seated in primate restraint chairs, and the lower portion of the shaved tail (approximately 15 cm) was immersed in a polycarbonate flask containing water at either 40, 50, or 55°C. Monkeys were tested at the three water temperatures in varying order, with tests in the same monkey separated from each other by approximately 2 min. Tail withdrawal latencies were timed on a stopwatch in 0.1-s increments. To prevent tissue damage, tails were removed from the water if they remained immersed for 20 s (cutoff latency). Sessions began with control determinations at each water temperature, presented in a varied order among the monkeys.

Experimental sessions were carried out no more frequently than once per week. The antinociceptive effects of GR89,696 were determined using a cumulative dosing procedure with a 1-h interinjection interval (testing occurred 50 min after each injection). The testing parameters were based on initial pilot studies on the time course of the effects of GR89,696.

Antagonist studies of GR89,696 were conducted using three opioid antagonists. First, the antagonist effects of naltrexone were studied with various pretreatment doses (0.032–0.56 mg/kg s.c.). The antinociceptive effects of GR89,696 were redetermined 15 min after pretreatment of a single dose of naltrexone. Second, a single dose (0.1 mg/kg s.c.) of clocinnamox was administered 24 h prior to determination of GR89,696-induced antinociception. Clocinnamox is a functionally irreversible μ-opioid receptor antagonist (Zernig et al., 1994). The present dose of clocinnamox was selected based on a previous study, in which clocinnamox produced an approximately 1-week-long period of μ-opioid receptor antagonism in rhesus monkeys (Zernig et al., 1994). Finally, pretreatment with a single dose of nor-BNI (3.2 mg/kg s.c.) was used to compare the “κ₁” receptor antagonism between U50,488 and GR89,696. The antinociceptive effects of U50,488 and GR89,696 were redetermined 1 day and 3 days after nor-BNI pretreatment, respectively. This dose of nor-BNI (3.2 mg/kg) produces a long-lasting antagonism of U50,488 in rhesus monkeys (Butelman et al., 1993b).

Data Analysis.

Individual data were converted to %maximum possible effect (%MPE) by the following calculation: %MPE = [(test latency − control latency)/(cutoff latency − control latency)] × 100%. Individual ED₅₀values were calculated from individual %MPE values by linear regression, and a mean ED₅₀ (±95% confidence limits) were presented. For in vivo apparent pA₂analysis, dose ratios produced by naltrexone were analyzed in a Schild plot and individual pA₂ values were obtained (PHARM/PCS program; Microcomputer Specialists, Philadelphia, PA). The same group of four subjects was used for all the pharmacological comparisons in this assay.

Respiration

Apparatus and Procedure.

The apparatus was similar to that described previously (Howell et al., 1988; Butelman et al., 1993a). Briefly, unanesthetized monkeys were seated in restraint chairs enclosed in sound-attenuating chambers. Gas, either air or a mixture of 5% CO₂ in air (hereafter referred to as CO₂), was pumped through the helmet and removed at a rate of 8 l/min. Changes in gas flow inside the helmet were measured with a pressure transducer connected to a polygraph (model 7E; Grass Instrument Co., Quincy, MA). The data were recorded on a polygraph trace and in a microprocessor via an analog to digital converter. Minute volume (V_E) was determined by integration of changes in flow through the plethysmograph. Frequency (f) was directly determined and tidal volume (V_T) was calculated as V_T = V_E/f.

Experimental sessions contained several consecutive cycles, with each cycle including a 53-min exposure to air followed by a 7-min exposure to 5% CO₂. An experiment was started after measurement of control respiratory values in air and CO₂. The compound was then administered (i.m. in the thigh) at the beginning of each cycle using a cumulative dosing procedure with a 1-h interinjection interval. The doses of both GR89,696 and morphine were chosen from the same range used in the warm water tail withdrawal procedure. Comparison of both compounds was made in the same monkeys across conditions.

In separate experiments (n = 4), the potential μ-antagonist effects of GR89,696 were studied by administering single pretreatment doses of GR89,696 (0.000032, 0.00032, and 0.001 mg/kg) 30 min before redetermination of the morphine dose-effect curve (0.32–10 mg/kg) in this assay.

Data Analysis.

Data from the last 3 min of exposure to air and the second 3 min of exposure to CO₂ in each cycle were used for analysis of drug effects on respiratory measures. Values obtained in each cycle are expressed as percentage of the respective control parameters collected before drug administration.

Sedation/Muscle Relaxation Observational Rating

Procedure.

Monkeys (n = 5) were rated while in their home cages on two observational rating scales described previously (Butelman et al., 1999b). Rating was carried out by a nonblinded observer, familiar with the individual subjects' baseline behavior. At each sampling point, subjects were rated initially on a muscle relaxation scale (scores 0–5; based on a subject's spontaneous posture) followed by a sedation scale (scores 0–6; based operationally on the environmental stimulus that is required to elicit a behavioral response).

Design and Data Analysis.

Cumulative dose effect curves for GR89,696 (0.000032–0.001 mg/kg) and U69,593 (0.001–0.032 mg/kg) were studied with a 60-min interinjection interval. Rating occurred before the onset of dosing (preinjection control) and 50 min or 20 min after each injection for GR89,696 and U69,593, respectively. A similar design was instituted for repeated vehicle administration (four consecutive sterile water injections at 60-min interinjection intervals, with rating 50 min after each injection). The same five subjects were used for all the observational rating studies. Dose effects curves for GR89,696 and U69,593 were analyzed with Friedman's repeated measures ANOVAs (α level was set at the 0.05 level).

Diuresis

Procedure.

The procedure has been described previously (Butelman et al., 1999d). Urine volumes were collected at hourly intervals over 3-h subsequent to the administration of the test drugs (GR89,696, U69,593, and fentanyl). Tests were performed in the home cage of each monkey (n = 3), with a clean cage pan placed under the grid floor, for urine collection. Water was available via automatic spouts throughout the session. Sessions were carried out twice per week.

Data Analysis.

Mean urine volumes were analyzed with a one-factor (dose) repeated measures ANOVA. When significant, post hoc Dunnett's comparisons were performed. In tests involving antagonists, paired t tests were performed. α was 0.05, two-tailed.

Serum Prolactin Levels

Procedure.

The procedure has been described previously (Butelman et al., 1999c). Chair-trained monkeys were tested after habituation to the experimental situation. Monkeys were chaired and brought into the experimental room between 9:30 and 10:00 AM on each test day. An indwelling catheter (24 gauge, Angiocath; Becton Dickinson, Sandy, UT) was placed in a superficial leg vein and secured with tape. An injection port (Terumo, Elkton, MD) was attached to the hub of the catheter; the port and catheter were flushed (0.3 ml of 50 U/ml heparinized saline) before use, and after each blood sampling. Approximately 15 min following catheter placement, two preinjection baseline blood samples were collected, 5 min apart (defined as −10 and −5 min relative to the onset of dosing). At each sampling point, an initial 0.5-ml blood aliquot was obtained; this aliquot was not used in the present assay due to the presence of the heparin “lock” solution. This was followed by a separate 1.5-ml blood aliquot, which was placed in a plain vacutainer, and kept at room temperature until the time of spinning (5 min at 3000 rpm; 4°C) and serum separation. Serum samples were kept at −40°C until the time of analysis.

The samples were analyzed in duplicate with a standard human prolactin radioimmunoassay kit (44 samples/kit; Nichols Diagnostics Institute, San Juan Capistrano, CA). Standard calibration curves were determined for each kit with human prolactin (3–150 ng/ml). The reported cross-reactivity of this assay was of largest magnitude for human growth hormone (0.07% cross-reactivity).

Monkeys were tested either in a time course or cumulative dosing design. Time course studies were carried out by administering a single s.c. injection (after baseline sample collection). Blood samples were then obtained at intervals (5–120 min) after injection. Cumulative s.c. dose-effect curve studies were carried out with a 60-min interinjection interval, with doses increasing by 0.5 log units in each cycle. Sample collection occurred 50 min after injection (for GR89,696) and 20 min after each injection (for U69,593). These sampling times were adjusted based on the approximate peak effect times determined from time course studies for these agonists.

Design and Data Presentation.

Each experiment was carried out in four to five females in the follicular phase (days 2–12 of each cycle of 28 days, as defined by the onset of visible bleeding). Consecutive experiments in the same subject were separated by at least 48 h (72 h following naltrexone administration). Time course studies were determined for GR89,696 (0.0001 or 0.00032 mg/kg) and U69,593 (0.032 mg/kg). Based on these time course studies, cumulative dose-effect curves were designed for GR89,696 (0.000032–0.00032 mg/kg) and U69,593 (0.001–0.01 mg/kg). For control purposes, a vehicle experiment was studied with an identical design to the GR89,696 dose-effect curve experiment described above. The GR89,696 and U69,593 dose-effect curves were redetermined after 5-min pretreatment with naltrexone (0.1 or 0.32 mg/kg). Naltrexone apparent pK_B values were calculated for GR89,696 and U69,593, from individual dose ratios. Apparent pK_B values were calculated according to a modified formula (Negus et al., 1993): pK_B = −log[B/DR − 1], where B was the dose of naltrexone in moles per kilogram and DR was the dose ratio. The same pool of subjects (n = 4–5/experiment) was used for the pharmacological comparisons in this assay; the 0.05 α level was adopted for all the studies presented herein.

Chemicals and Drugs.

GR89,696 (Sigma/RBI, Natick, MA), naltrexone HCl (National Institute on Drug Abuse, Research Triangle Park, NC), and nor-binaltorphimine (provided by Dr. H. I. Mosberg, Division of Medicinal Chemistry, University of Michigan, Ann Arbor, MI) were dissolved in sterile water. U50,488 and U69,593 (Pharmacia & Upjohn, Kalamazoo, MI) were dissolved in sterile water with 1 to 2 drops of lactic acid (if necessary). Compounds were injected in volumes of 0.05 to 0.1 ml/kg. All doses are expressed as the above-mentioned forms of the compounds. DAMGO and DPDPE were purchased from Sigma (St. Louis, MO). The following radioligands were used: [³H]DAMGO (40.7 Ci/mmol; Amersham Pharmacia Biotech, Arlington Heights, IL), [³H]DPDPE (30 Ci/mmol; PerkinElmer Life Science Products, Boston, MA), [³H]bremazocine (30 Ci/mmol; PerkinElmer Life Science Products), and [³H]U69,593 (47.9 Ci/mmol; PerkinElmer Life Science Products). [³⁵S]GTPγS was purchased from PerkinElmer Life Science Products.

Results

Radioligand Binding.

In the presence of DAMGO and DPDPE masking agents, GR89,696 displaced the specific binding of [³H]U69,593 with a 5-fold lowerK_i than for [³H]bremazocine sites [K_i (±S.E.M.): 0.22 ± 0.05 and 1.15 ± 0.47 nM, respectively]. GR89,696 also displaced binding of [³H]DAMGO (μ-sites) and [³H]DPDPE (δ-sites) [K_i (±S.E.M.) were 0.65 ± 0.15 and 30.61 ± 11.8 nM, respectively].

[³⁵S]GTPγS Assay.

GR89,696 was a potent, high-efficacy “full” agonist on cloned κ-receptors expressed in CHO cells, as measured by its ability to stimulate [³⁵S]GTPγS binding, relative to U69,593 (Table 1). GR89,696 was approximately 100-fold less potent at μ-receptors in C6 cells, and also displayed a lower maximal stimulatory effect on [³⁵S]GTPγS binding, compared with DAMGO. U69,593 and DAMGO are high-efficacy “full” agonists at cloned κ- and μ-receptors, respectively, in this assay (Alt et al., 1998;Remmers et al., 1999).

A single concentration of DAMGO (100 nM) caused 77 ± 2.5% stimulation of GTPγS binding. This was reduced to a 20.7 ± 2.1% stimulation (p < 0.0001) in a concentration-dependent manner by GR89,696, affording an AD₅₀ for GR89,696 of 212 ± 40 nM. In the present experiment, GR89,696 had to compete for μ-receptors with DAMGO; thus, the AD₅₀ value depended on the amount of DAMGO used. Therefore, the GR89,696 AD₅₀ value is not directly comparable to the ED₅₀ value for GR89,696 alone in the present assay.

EKC Discrimination.

Single doses of GR89,696 (0.00001–0.0001 mg/kg s.c.) dose and time dependently increased EKC lever selection (Figs. 1 and2). At the highest dose tested (0.0001 mg/kg), greater than 80% EKC lever selection was observed 30 to 120 min after administration. Similarly, cumulative dosing with GR89,696 dose dependently engendered EKC responding, with greater than 80% EKC lever selection observed at 0.0001 mg/kg. U69,593 (0.0001–0.0032 mg/kg s.c.), but not fentanyl (0.0001–0.0032 mg/kg s.c.), was generalized in EKC-trained monkeys.

U69,593 Discrimination.

Vehicle administration (s.c. bolus, 0.1 ml/kg) produced less than 5% mean U69-appropriate responding when tested between 5 and 60 min after injection (n = 3; one or two determinations). Time course studies with single U69,593 doses (0.0032 and 0.01 mg/kg) resulted in dose-dependent generalization (Fig.3). The 0.01 mg/kg U69,593 dose resulted in generalization in all three subjects 15 min after administration. Vehicle-appropriate responding was observed in all three subjects by 120 min after 0.01 mg/kg of U69,593 (Fig. 3). GR89,696 (0.00001–0.0001 mg/kg) also occasioned dose-dependent generalization in all three subjects (Fig. 3). Two of the subjects generalized at the 0.0001-mg/kg dose, between 30 and 120 min after administration (Fig. 3). The third subject did not generalize at this dose, therefore a larger dose (0.00032 mg/kg) was probed in this subject alone (two determinations). In both these determinations, the subject generalized GR89,696 (0.00032 mg/kg) to the training drug (at 30 and 60 min).

Antinociception.

The subjects (n = 4) used in this study had a consistent profile of tail-withdrawal responses. In the absence of drug treatment, they kept their tails in 40°C water for 20 s (cutoff latency) and removed their tails from 50 and 55°C water rapidly (within 1–3 s). GR89,696 (0.0001–0.0018 mg/kg) dose dependently produced antinociception against the 50^o and 55°C water stimuli (Fig.4). ED₅₀ (95% CL) values for either temperature were 0.00052 mg/kg (0.00046–0.00058) and 0.0013 mg/kg (0.0012–0.0013), respectively. Pretreatment with naltrexone (0.032–0.56 mg/kg) dose dependently produced rightward shifts of the dose-effect curve of GR89,696-induced antinociception (Fig. 4). The naltrexone apparent pA₂ value (95% CL) for GR89,696 in 50°C water was 7.0 (6.9–7.1), the obtained slope from a Schild plot was −1.1 (1.3–1.0). The naltrexone apparent pA₂ value (95% CL) for GR89,696 in 55°C water was 6.9 (6.7–7.1) and with a Schild slope of −1.1 (1.4–0.8).

Pretreatment with the μ-selective antagonist clocinnamox (0.1 mg/kg; 24 h PT did not significantly antagonize GR89,696-induced thermal antinociception (data not shown). This clocinnamox treatment is sufficient to cause profound antagonism of μ-receptor mediated effects in this assay (Zernig et al., 1994).

In separate experiments, 3-day pretreatment with the κ₁-selective antagonist nor-BNI (3.2 mg/kg) also did not antagonize the antinociceptive effects of GR89,696 in either 50 or 55°C water (Fig. 5). In a control study, 1-day pretreatment with nor-BNI (3.2 mg/kg) produced 3.3- and 3.6-fold rightward shifts of U50,488-induced antinociception against 50 and 55°C water, respectively (Fig. 5; Butelman et al., 1993b). Nor-BNI's inactivity against GR89,696 was not due to a difference in the time of testing; the presently used nor-BNI dose was active against U50,488 for more than 10 days after administration (Butelman et al., 1993b).

Respiration.

Untreated monkeys (n = 4) exhibited a consistent pattern of respiration in the head plethysmograph apparatus. A stable baseline rate of respiration was recorded in monkeys breathing air. The mean control values (± S.E.M.) for monkeys breathing air were as follows: f = 28 ± 3 (frequency; breaths/min), V_T = 102 ± 15 (tidal volume; ml/breath), and V_E = 2882 ± 407 (minute volume; ml/min). When 5% CO₂ was introduced, a prompt increase in all respiratory parameters was observed, with individual monkeys displaying stable patterns of CO₂-induced increases across sessions. The mean control values (± S.E.M.) for monkeys breathing 5% CO₂ in air were as follows: f = 37 ± 3 (breaths/min), V_T = 167 ± 12 (ml/breath), and V_E = 5992 ± 486 (ml/min). Cumulative administration of morphine (1–18 mg/kg) dose dependently produced suppression of f, V_T, and V_E in both air and 5% CO₂. Figure 6 portrays the effects of morphine and GR89,696 in monkeys breathing 5% CO₂ (the profile of these effects is the same for monkeys breathing air; data not shown). Following the largest dose administered, morphine suppressed V_E to 25% ± 3 (S.E.M.) of control levels. In contrast, the largest dose of GR89,696 (0.0018 mg/kg) only suppressed V_E to 69% ± 8 (S.E.M.) of control levels. This difference between morphine and GR89,696 at the largest doses was also observed with V_T and f parameters (V_T: morphine (41 ± 3%) versus GR89,696 (81 ± 4%); f: morphine (60 ± 7%) versus GR89,696 (84 ± 7%).

Pretreatment with GR89,696 (0.000032, 0.00032, or 0.001 mg/kg;n = 4 each) did not affect the respiratory depressant effects of morphine (0.32–10 mg/kg) under the present conditions. In these determinations, the morphine baseline ED₅₀value for CO₂ V_E (the most sensitive parameter) was 2.7 mg/kg (95% CL = 1.0–7.6). In the presence of GR89,696, the CO₂V_E values were 2.9 mg/kg (95% CL = 1.1–7.4; 0.000032-mg/kg GR89,696 PT dose), 2.5 mg/kg (95% CL = 0.7–9.3; 0.00032-mg/kg PT dose), and 1.5 mg/kg (95% CL = 1.1–2.2; 0.001-mg/kg PT dose).

Sedation/Muscle Relaxation Rating.

Monkeys rated in the absence of pharmacological treatment (preinjection controls;n = 5) displayed no signs of sedation or muscle relaxation (i.e., scores = 0), as defined in the present rating scales. Likewise, four consecutive vehicle injections (60-min interinjection interval, with rating 50 min after each injection) produced scores of 0 for all subjects in the present scales. Both GR89,696 (0.000032–0.001 mg/kg) and U69,593 (0.001–0.032 mg/kg) caused dose-dependent elevations in sedation and muscle relaxation scores (Fig. 7). Up to the largest doses presently tested, both compounds produced similar maximal effects on either rating scale. GR89,696 was approximately 30-fold more potent than U69,593, as estimated from the plotted dose-effect curves. Significant Friedman's ANOVAs were obtained for GR89,696's effects on sedation and muscle relaxation (X²_F [5] = 18.23 and 17.27, respectively). Post hoc Dunn's tests revealed that for both scales, the largest GR89,696 dose (0.001 mg/kg) produced higher scores than preinjection control. Significant Friedman's ANOVAs were also obtained for U69,593's effects on sedation and muscle relaxation scales (X²_F [5] = 17.44 and 18.02, respectively). Dunn's tests also revealed that the largest U69,593 dose (0.032 mg/kg) caused a significant increase in sedation and muscle relaxation scores from preinjection control.

Diuresis.

Saline-injected monkeys produced approximately 30 ml of urine during the 3-h observation period. GR89,696 (0.00001–0.00032 mg/kg i.m.) dose dependently increased urine output [F(4,8) = 7.14, p = 0.01] (Fig.8), reaching significance beginning at 0.0001 mg/kg. At the highest dose tested (0.00032 mg/kg), urine volume was increased to approximately 230%. Pretreatment with quadazocine (1 mg/kg) prevented the diuretic effects of GR89,696 [0.00032 mg/kg; t(2) = 5.03, p = 0.04], returning urine volumes to vehicle control [t(2) = 1.22, N.S.], while producing no effects on its own [t(2) = 0.77, N.S.] (Fig.9).

The κ-agonist U69,593 (0.01–0.32 mg/kg i.m.) also increased urine output at all doses tested [F(4,8) = 29.97,p < 0.01]. U69,593 (0.1 mg/kg) produced a 283% increase in urine volume (Fig. 8). Pretreatment with quadazocine reversed the diuretic effects of U69,593 [0.1 mg/kg; t(2) = 6.96,p = 0.02] (Fig. 9). In contrast, the μ-agonist fentanyl (0.001–0.01 mg/kg) dose dependently decreased urine output [F(3,6) = 4.62, p = 0.05] (data not shown).

Serum Prolactin Levels.

Preinjection prolactin levels were typically 3 ng/ml or less in these female subjects. Three consecutive vehicle injections (60-min interinjection intervals, with blood sampling occurring 20 min after each injection) did not result in an elevation of prolactin levels compared with preinjection levels (n = 3; data not shown). The time course of single GR89,696 (0.0001 and 0.00032 mg/kg) and U69,593 (0.032 mg/kg) doses was studied between 5 and 120 min after s.c. administration. GR89,696 peak effects on prolactin were observed 30 to 60 min after administration, whereas U69,593 caused a peak increase in prolactin levels 15 to 30 min after administration (Fig. 10).

Both GR89,696 (0.000032–0.00032 mg/kg) and U69,593 (0.001–0.01 mg/kg) caused dose-dependent increases in prolactin levels, when tested in a cumulative dosing design (Fig. 11). GR89,696 was 23-fold more potent than U69,593 in this neuroendocrine end point (Table 2). Naltrexone (0.1–0.32 mg/kg) was administered 5 min before the redetermination of GR89,696 and U69,593 dose-effect curves (Fig. 11). The smaller naltrexone pretreatment dose (0.1 mg/kg) did not affect the GR89,696 dose-effect curve, whereas the larger naltrexone dose (0.32 mg/kg) caused a 2.6-fold rightward surmountable shift. In contrast, the smaller naltrexone pretreatment dose (0.1 mg/kg) caused an 8.6-fold rightward surmountable shift in the U69,593 dose-effect curve. Naltrexone alone (0.01–0.32 mg/kg; studied in a cumulative dose-effect curve) did not cause an elevation in serum prolactin levels (n = 4; data not shown). Apparent pK_B analysis (Table 2) confirmed that naltrexone was a significantly more potent antagonist of U69,593 than of GR89,696 in this neuroendocrine end point.

Discussion

GR89,696 produced agonist effects in assays used to characterize κ-opioids in rhesus monkeys, consistent with its high affinity for κ-opioid receptors in monkey brain, and its high efficacy at cloned κ-receptors. As indicated by a potency comparison with several benzomorphan or arylacetamide compounds, GR89,696 is one of the most potent κ-agonists to be characterized in vivo in rhesus monkeys to date (Dykstra et al., 1987a; France et al., 1994; Butelman et al., 1999a).

Radioligand Binding and Stimulation of [³⁵S]GTPγS Binding.

GR89,696 displayed low nanomolar affinity for κ-opioid sites in monkey brain. GR89,696 had approximately 5-fold selectivity for κ-sites sites labeled by [³H]U69,593 over κ-sites labeled by [³H]bremazocine (in the presence of masking agents for μ- and δ-opioid sites). This is consistent with studies in rodents, suggesting that GR89,696 does not possess a high degree of binding selectivity for either of the proposed κ-subtypes. Based on ex vivo studies in rats, it has been suggested that GR89,696 is an antagonist at the κ₁receptor subtype, whereas it is an agonist at the κ₂ receptor subtype (Caudle et al., 1997). GR89,696 had moderate affinity for sites labeled by [³H]DAMGO (μ-receptors) and lower affinity for sites labeled by [³H]DPDPE (δ-receptors) in monkey brain.

GR89,696 was a potent high efficacy “full” agonist in CHO cells transfected with κ-opioid receptors, as measured by the stimulation of [³⁵S]GTPγS binding. However, GR89,696 had approximately 100-fold lower agonist potency on μ-receptors, and was a low efficacy (i.e., partial) agonist in this μ-receptor preparation. This was confirmed by the ability of GR89,696 to antagonize the effect of the high efficacy μ-agonist DAMGO in this preparation.

The above-mentioned in vitro findings are consistent with the profile of GR89,696 as a potent high-efficacy κ-agonist in vivo. The finding that GR89,696 is a high-efficacy κ-agonist (similarly to the κ₁ ligand U69,593) suggests that this in vitro preparation may not parallel the characteristics that generate differential pharmacological effects for proposed κ₁ and κ₂ agonists, as encountered in vivo or ex vivo (Zukin et al., 1988; Caudle et al., 1997; Schoffelmeer et al., 1997). A similar situation has been encountered previously with regard to the in vitro versus in vivo effects of several κ-opioids (France et al., 1994; Remmers et al., 1999). The low efficacy (or antagonist effects) of GR89,696 at μ-receptors could be detected in vitro in the GTPγS assay, but not in vivo in an assay of respiratory depression (see below).

EKC and U69,593 Discrimination.

GR89,696 (0.000032–0.0001 mg/kg) was generalized by all monkeys trained to discriminate EKC from vehicle. EKC is not a selective ligand with respect to κ- versus μ-opioid sites, or with respect to κ₁ versus all κ-receptors (Zukin et al., 1988; Emmerson et al., 1994; Butelman et al., 1998). However, rhesus monkeys trained to discriminate EKC from vehicle generalize to κ- but not μ-agonists (Dykstra et al., 1987b; France et al., 1994). Therefore, generalization observed in EKC-trained monkeys with GR89,696 indicates that this compound shares the discriminative stimulus effects of κ-opioid agonists (see Carey and Bergman, 2001).

U69,593 is considered a prototypical “κ₁”-selective ligand (Zukin et al., 1988). In the present studies, GR89,696 was generalized by all subjects trained to discriminate U69,593 from vehicle. Under the present conditions, GR89,696 exhibited a relatively slow onset, and was approximately 30 to 100 times more potent than U69,593. Therefore GR89,696, a proposed “κ₂” agonist, shares discriminative stimulus effects (interoceptive effects) with U69,593, a ligand selective for the κ₁ site.

Thermal Antinociception.

GR89,696 dose dependently elevated tail withdrawal latencies in 50 and 55°C water. The antinociceptive effect of GR89,696 was insensitive to pretreatment with clocinnamox (0.1 mg/kg; 24 h PT). This indicates that the effects of GR89,696 were not mediated by μ-opioid receptors in this assay (Zernig et al., 1994). The effects of GR89,696 in this assay were dose dependently antagonized by naltrexone. The potency of naltrexone (as determined with apparent pA₂ values) was similar to that previously reported for non-κ₁-selective agonists (such as bremazocine), and is lower than for agonists selective for the κ₁ site (e.g., U50,488 and U69,593; Ko et al., 1998) (Table 3). Consistent with this conclusion, the antinociceptive effects of GR89,696 were also insensitive to pretreatment with nor-BNI (3.2 mg/kg). This pretreatment dose of nor-BNI is sufficient to antagonize the antinociceptive effects of selective κ₁ ligands such as U50,488 and U69,593, but not those of bremazocine (present study; Butelman et al., 1993b; Table 3). Overall, the present antagonism experiments are consistent with the notion that GR89,696 produces its antinociceptive effects under these conditions at least partially through κ₂ receptors.

Respiratory Depression.

Up to the largest dose studied presently, GR89,696 only caused moderate respiratory depression. The largest GR89,696 dose studied in the respiratory depression assay was fully effective in the assay of thermal antinociception, and produced prominent behavioral (e.g., sedative) effects in the present studies. The limited respiratory depressant effects of GR89,696 are similar to those previously reported for selective κ-agonists in rhesus monkeys (e.g., U69,593 or U50,488; Howell et al., 1988; France et al., 1994). By comparison, doses of the μ-agonist morphine, which produce antinociceptive effects (Zernig et al., 1994), caused more profound respiratory depression in the present studies. Thus, at antinociceptive doses, GR89,696 has only a limited capacity to cause respiratory depression.

Given the low-efficacy/antagonist actions of GR89,696 at μ-receptors in the GTPγS assay (see above), the potential μ-antagonist effects of GR89,696 were studied in this in vivo assay. A broad range of GR89,696 PT doses (up to doses that produced robust κ-receptor-mediated effects) did not significantly affect the respiratory depressant effects of morphine under the present conditions. Larger GR89,696 doses were not assessed to avoid possible untoward effects.

Sedation and Muscle Relaxation.

GR89,696 produced sedative and muscle relaxant effects qualitatively similar to those U69,593 in the same group of rhesus monkeys. GR89,696 was approximately 30-fold more potent than U69,593 on both these behavioral end points. Therefore, GR89,696 shared the overt behavioral effects caused by an agonist selective for the κ₁ site, U69,593.

Diuresis.

GR89,696 produced dose-dependent diuresis in rhesus monkeys, similarly to previously reported studies with κ-opioid ligands (Dykstra et al., 1987a). The diuretic effect of GR89,696 was of similar magnitude to that of U69,593, although GR89,696 was approximately 100-fold more potent than U69,593. The diuretic effects of both GR89,696 and U69,593 were sensitive to pretreatment with the opioid antagonist quadazocine. Quadazocine has been used previously to antagonize the diuretic effects of structurally diverse κ-opioids in rhesus monkeys (Dykstra et al., 1987a; Butelman et al., 1999d). The ability of GR89,696 to cause diuresis in rhesus monkeys is consistent with previous studies in human and nonhuman primates, since both κ₁-selective and non-κ₁-selective agonists produce diuresis (Tang and Collins, 1985; Dykstra et al., 1987a; Rimoy et al., 1991;Butelman et al., 1999d).

Serum Prolactin Levels.

In common with synthetic or peptidic κ-agonists, as well as μ-agonists, GR89,696 produced a dose-dependent increase in prolactin levels in intact, follicular phase, female rhesus monkeys (Butelman et al., 1999c). GR89,696 displayed a slower onset than U69,593 in this assay, and was 23-fold more potent than U69,593 in dose-effect curve determinations. Naltrexone has higher affinity for binding sites labeled by U69,593 (a κ₁-selective ligand) than by bremazocine (a nonselective κ-ligand) in rhesus monkey brain (Ko et al., 1998). It has been suggested that the agonist effects of GR89,696 are mediated by κ₂ sites (Ho et al., 1997). Consistent with this suggestion, the prolactin-releasing effects of GR89,696 were less sensitive to naltrexone antagonism than those of U69,593, in the present studies. These neuroendocrine findings are, to our knowledge, the first to show an in vivo differentiation of a proposed κ₁ versus κ₂ effect in a non-behavioral end point.

The prolactin-releasing effects of opioid agonists are thought to be mediated by a decrease in tone in the dopaminergic hypothalamic systems (Moore and Lookingland, 1995). The potency and duration of action of GR89,696 in releasing prolactin therefore suggest that this κ-agonist could produce a long-lasting decrease in hypothalamic dopaminergic tone in vivo, following systemic administration in primates (Butelman and Kreek, 2001).

Summary

GR89,696 was a high-efficacy and high-potency agonist at κ-opioid receptors, but only a low-efficacy agonist at μ-opioid receptors, in the GTPγS assay in vitro. In rhesus monkeys in vivo, GR89,696 had the profile of a potent and long-lasting κ-agonist, in that it produced thermal antinociception, sedation and κ-agonist-like discriminative stimulus effects, as well as prolactin release. Naltrexone and nor-BNI antagonism studies in antinociceptive and neuroendocrine assays are consistent with mediation of these effects of GR89,696 by non-κ₁ (i.e., possibly κ₂) receptors. Overall, the present studies demonstrate that GR89,696 shares the in vivo effects of a κ₁ ligand (U69,593; see Table 3 for a summary comparison). However, pretreatment studies with naltrexone and nor-BNI suggest that the effects of GR89,696 may be mediated by a subset of κ-opioid receptors with a distinct profile of antagonist sensitivity.