Psychopharmacology (2001) 154:362–374 DOI 10.1007/s002130000667

O R I G I N A L I N V E S T I G AT I O N

Jonathan L. Katz · Gregory E. Agoston Kenneth L. Alling · Richard H. Kline Michael J. Forster · William L. Woolverton Theresa A. Kopajtic · Amy H. Newman

Dopamine transporter binding without cocaine-like behavioral effects: synthesis and evaluation of benztropine analogs alone and in combination with cocaine in rodents Received: 13 July 2000 / Accepted: 30 November 2000 / Published online: 27 February 2001 © Springer-Verlag 2001

Abstract Rationale: Previous SAR studies demonstrated that small halogen substitutions on the diphenylether system of benztropine (BZT), such as a para-Cl group, retained high affinity at the cocaine binding site on the dopamine transporter. Despite this high affinity, the compounds generally had behavioral effects different from those of cocaine. However, compounds with meta-Cl substitutions had effects more similar to those of cocaine. Objectives: A series of phenyl-ring analogs of benztropine (BZT) substituted with 3′-, 4′-, 3′,4′′- and 4′,4′′-position Cl-groups were synthesized and their pharmacology was evaluated in order to assess more fully the contributions to pharmacological activity of substituents in these positions. Methods: Compounds were synthesized and their pharmacological activity was assessed by examining radioligand binding and behavioral techniques. Results: All of the compounds displaced [3H]WIN 35,428 binding with affinities ranging from 20 to 32.5 nM. Affinities at norepinephrine ([3H]nisoxetine) and serotonin ([3H]citalopram) transporters, respectively, ranged from 259 to 5120 and 451 to 2980 nM. Each of the compounds also inhibited [3H]pirenzepine binding to J.L. Katz (✉) · K.L. Alling · T.A. Kopajtic Psychobiology Section, Medications Discovery Research Branch, NIDA Intramural Research Program, National Institutes of Health, P.O. Box 5180, Baltimore, MD 21224, USA Fax: +1-410-550-1648 G.E. Agoston · R.H. Kline · A.H. Newman Medicinal Chemistry Section, Medications Discovery Research Branch, NIDA Intramural Research Program, National Institutes of Health, P.O. Box 5180, Baltimore, MD 21224, USA K.L. Alling · W.L. Woolverton Department of Psychiatry, University of Mississippi, Jackson, Mississippi, USA M.J. Forster Department of Pharmacology, University of North Texas Health Science Center at Fort Worth, Fort Worth, Texas, USA

muscarinic M1 receptors, with affinities ranging from 0.98 to 47.9 nM. Cocaine and the BZT analogs produced dose-related increases in locomotor activity in mice. However, maximal effects of the BZT analogs were uniformly less than those produced by cocaine, and were obtained 2–3 h after injection compared to the relatively rapid onset (within 30 min) of cocaine effects. In rats trained to discriminate IP saline from 29 µmol/kg cocaine (10 mg/kg), cocaine produced a dose-related increase in responding on the cocaine lever, reaching 100% at the training dose; however, none of the BZT analogs fully substituted for cocaine, with maximum cocaine responding from 20 to 69%. Despite their reduced efficacy compared to cocaine in cocaine discrimination, none of the analogs antagonized the effects of cocaine. As has been reported previously for 4′-Cl-BZT, the cocaine discriminative-stimulus effects were shifted leftward by co-administration of the present BZT analogs. Conclusions: The present results indicate that although the BZT analogs bind with relatively high affinity and selectivity at the dopamine transporter, their behavioral profile is distinct from that of cocaine. The present results suggest that analogs of BZT may be useful as treatments for cocaine abuse in situations in which an agonist treatment is indicated. These compounds possess features such as reduced efficacy compared to cocaine and a long duration of action that may render them particularly useful leads for the development of therapeutics for cocaine abusers. Keywords Cocaine · Dopamine transporter · Stimulant effect · Benztropine analogs · Discriminative stimulus effect

Introduction Novel analogs of benztropine [3α-(diphenylmethoxy)1H,5H-tropane] have been developed that have high-

363

affinity for the dopamine transporter and are selective for the dopamine transporter over the other monoamine transporters (Newman et al. 1994, 1995). Although these compounds bind to the dopamine transporter and inhibit the uptake of dopamine in vitro, their behavioral effects are generally different from those of the typical dopamine uptake inhibitors, for which cocaine is a prototype (Katz et al. 1999). For example, 4′-Cl-BZT [4′-chloro-3(diphenylmethoxy)tropane] has a 30 nM affinity for the dopamine transporter which is comparable to the highaffinity binding of cocaine. However, 4′-Cl-BZT is only marginally efficacious as a stimulant of locomotor activity and does not produce cocaine-like discriminativestimulus effects (Katz et al. 1999; Tolliver et al. 1999). Further, this compound does not maintain rates of responding as high as those maintained by cocaine in a “self-administration” paradigm (Woolverton et al. 2000) or break points as high as those for cocaine in a self administration progressive ratio procedure (Woolverton et al. 2001). As such, this and similar compounds suggest a challenge to the dopamine transporter hypothesis of the behavioral effects of cocaine. According to that hypothesis, compounds that bind to the dopamine transporter and inhibit dopamine uptake will have behavioral effects like those of cocaine (Kuhar et al. 1991). In addition, an understanding of the differences in pharmacological mechanisms of cocaine and the benztropine analogs may provide insight into the neurobiological substrates that underlie the abuse liability of cocaine, and help understand the functioning of the dopamine transporter. We previously reported that, in contrast to many of the BZT analogs such as 4′-Cl-BZT, the 3′-Cl-analog had discriminative stimulus effects similar to those of cocaine (Kline et al. 1997). This different behavioral spectrum of action of 3′-Cl-BZT comes despite a minimal structural difference: a meta- versus a para-chloro substituent. The present experiment was designed to further investigate the pharmacology of 3′-substituted analogs of BZT, with the goal of providing a better understanding of the behavioral consequences of substitution at the 3′ position. We report the synthesis and behavioral effects of the novel synthetic entity, 3′,4′′-diCl-BZT [3′,4′′-dichloro-3α-(diphenyl-methoxy)tropane], an analog that incorporated both the 3′-Cl substituent on one phenyl ring and a 4′′-Cl substituent on the other phenyl ring (Fig. 1). Based on structure-activity relationships (SAR) derived for a large series of phenyl-ring-substituted BZT analogs (Newman et al. 1995) and a recent 3D quantitative SAR study using comparative molecular field analysis (Newman et al. 1999), it was predicted that the 3′,4′′diCl-BZT analog would have similar binding affinity at the dopamine transporter as the parent compounds. It was anticipated that an assessment of the pharmacology of the 3′,4′′-diCl analog of BZT would provide information on whether the 3′-Cl substituent is primarily responsible for the cocaine-like effects of 3′-Cl-BZT, or whether the 4′-Cl substituent interferes with cocaine-like behavioral effects. Of course, the new ligand might have a novel profile that may provide insight into mechanistic

Fig. 1 Basic structures of cocaine and benztropine analogs with substitutions in the 3′, 4′, or 4′′ positions

correlates to behavioral actions in this class of dopamine uptake inhibitors. The present studies further examined the pharmacology of 3′-Cl-BZT and compared it to that of structurally related 4′-Cl-substituted BZT analogs. The 4′-Cl-analogs have high affinity for the dopamine transporter and, among the BZT analogs, have behavioral effects least like those of cocaine (Katz et al. 1999). Because the Clsubstituted analogs exhibited a relatively high affinity for the dopamine transporter accompanied by a lack of cocaine-like discriminative-stimulus effects, interactions with cocaine were of considerable interest. Our previous study indicated that 4′-Cl-BZT potentiated the effects of cocaine (Tolliver et al. 1999). Thus in the present study we not only further examined the pharmacology of BZT analogs, we also examined their interactions with cocaine. The analogs examined included 3′-Cl-3′,4′′-diCl-, 4′,4′′-diCl-, and 4′-Cl-BZT.

Materials and methods [3H]WIN 35,428 binding assay Details of the procedures used have been published previously (Izenwasser et al. 1993). Fresh caudate putamen (from male Sprague-Dawley rats; Taconic, Germantown, N.Y., USA) was homogenized in 30 vol ice-cold modified Krebs-HEPES buffer (15 mM HEPES, 127 mM NaCl, 5 mM KCl, 1.2 mM MgSO4, 2.5 mM CaCl2, 1.3 mM NaH2PO4, 10 mM D-glucose, pH adjusted to 7.4) using a Brinkman polytron and centrifuged at 20,000×g for 10 min at 4°C. The resulting pellet was then washed two more times by resuspension and recentrifuged at 20,000×g for 10 min at 4°C and the resulting pellet was resuspended in buffer to a concentration of 5 mg/ml (original wet weight). Binding assays were conducted in modified Krebs-HEPES buffer on ice. The total volume per tube was 0.5 ml. Membrane suspensions were preincubated for 5 min in the presence or absence of the compound being tested. [3H]WIN 35,428 [New England Nuclear, Boston, Mass., USA (final concentration 1.5 nM)] was added and the incubation continued for 1 h on ice. The incubation was terminated by the addition of 3 ml of ice-cold buffer and rapid filtration through Whatman GF/B glass fiber filter paper (presoaked in 0.1% BSA in water) using a Brandel Cell Harvester (Gaithersburg, Md., USA). The filters were washed with three additional 3 ml washes and transferred to scintillation vials. Abso-

364 lute ethanol and Beckman Ready Value Scintillation Cocktail were added to the vials. Non-specific binding was defined as binding in the presence of 100 µM (–)cocaine HCl. Each compound was tested in three independent experiments, performed in triplicate. In this and in all of the studies using radioligands, a Beckman 6000 LS counter (Beckman Coulter Instruments, Fullerton, Calif., USA) was used to determine radioactivity. Displacement data were analyzed and Ki values determined by the use of the non-linear least squares curve-fitting computer program LIGAND (Munson and Rodbard 1980). Data from replicate experiments were modeled together to produce a set of parameter estimates and the associated standard errors of these estimates. In each case, the model reported fit significantly better than all others according to the F test at P<0.05.

pended in cold buffer to a concentration of 15 mg/ml (original wet weight). Ligand binding experiments were conducted in assay tubes containing 0.5 ml of buffer for 60 min at 25°C (room temperature). Each tube contained 1.4 nM [3H]citalopram (New England Nuclear) and 1.5 mg midbrain tissue. Non-specific binding was determined using 10 µM fluoxetine (RBI, Natick, Mass., USA). Incubations were terminated by rapid filtration through Whatman GF/B filters (presoaked in 0.3% polyethylenimine in water) using a Brandel Cell Harvester. The filters were washed twice with 5 ml cold buffer, transferred to scintillation vials to which Beckman Ready Safe was added. Data were analyzed using GraphPad Prism software. [3H]Pirenzepine binding assay

Dopamine uptake inhibition Fresh striatum (from male Sprague-Dawley rats, Taconic Labs) was homogenized in ice cold buffer (5 mM HEPES, 0.32 M sucrose), using ten strokes with a Teflon glass homogenizer followed by centrifugation at 1,000×g for 10 min at 4°C. The supernatant was saved and recentrifuged at 10,000×g for 20 min at 4°C. The supernatant was then discarded and the pellet was gently resuspended in ice-cold incubation buffer (127 mM NaCl, 5 mM KCl, 1.3 mM NaH2PO4, 1.2 mM MgSO4, 2.5 mM CaCl2, 1.498 mM HEPES acid, 10 mM D-glucose, 1.14 mM L-ascorbic acid, pH 7.4) and placed on ice for 15 min. The tissue preparation was incubated in buffer in test tubes at 37°C to which 10 µM pargyline and either the test drug or di H2O was added. After a 10-min incubation, [3H]dopamine (final concentration, 0.5 nM) was added to each tube. The incubation continued for 5 min and was terminated by rapid filtration through Whatman GF/B glass fiber filter paper (presoaked in 0.1% polyethylenimine in water) using a Brandel Cell Harvester. The filters were washed with two additional 3 ml washes and transferred to scintillation vials. Methanol and Ready Value scintillation fluid (Beckman-Coulter) was added. Non-specific binding was defined using 100 µM (–)cocaine HCl. Data were analyzed using the nonlinear regression analysis of GraphPad Prism version 2.0 (GraphPad Software, San Diego, Calif., USA).

Whole frozen rat brains excluding cerebellum (Taconic) were thawed in ice-cold buffer (10 mM TRIS-HCl, 320 mM sucrose, pH 7.4) and homogenized with a Brinkman polytron in a volume of 10 ml/g tissue. The homogenate was centrifuged at 1,000×g for 10 min at 4°C. The resulting supernatant was then centrifuged at 10,000×g for 20 min at 4°C. The resulting pellet was resuspended in a volume of 5 ml/g in 10 mM TRIS buffer (pH 7.4). Assays were conducted in binding buffer (10 mM TRIS-HCl, 5 mM MgCl2). The total volume in each tube was 0.5 ml, and the final concentration of membranes was approximately 200–300 mg protein/sample. [3H]Pirenzepine (New England Nuclear; final concentration 3 nM) was added to each tube. Quinuclidinyl benzilate (QNB), 100 µM final concentration was used to determine nonspecific binding. The reaction was started with the addition of the tissue and incubated for 60 min in a 37°C water bath. The incubation was terminated by the addition of 5 ml ice-cold buffer (10 mM TRIS-HCl, pH 7.4) and rapid filtration through Whatman GF/B glass fiber filter paper (presoaked in 0.5% polyethylenimine) using a Brandel Cell Harvester. The filters were washed with two additional 5 ml washes and transferred to scintillation vials. Absolute ethanol and Beckman Ready Safe was added. Data were analyzed by using GraphPad Prism software. Receptor screen

[3H]Nisoxetine binding assay Membranes from frozen frontal cortex (from male SpragueDawley rats; Taconic Labs) were homogenized in 20 vol (w/v) of 50 mM TRIS containing 120 mM NaCl and 5 mM KCl, (pH 7.4 at 25°C), using a Brinkman Polytron (setting 6 for 20 s). The tissue was centrifuged at 50,000×g for 10 min at 4°C and the resulting pellet was resuspended in buffer, recentrifuged and resuspended to a concentration of 80 mg/ml (original wet weight). Ligand binding experiments were conducted in assay tubes containing 0.5 ml buffer for 60 min at 0–4°C. Each tube contained 0.5 nM [3H]nisoxetine (New England Nuclear), and 8 mg frontal cortex tissue. Non-specific binding was determined using 1 µM desipramine. Incubations were terminated by rapid filtration through Whatman GF/B filters, presoaked in 0.05% polyethylenimine, using a Brandel Cell Harvester (Brandel Instruments). The filters were washed twice with 5 ml cold buffer, transferred to scintillation vials to which Beckman Ready Safe was added. Data were analyzed using GraphPad Prism software. [3H]Citalopram binding assay Membranes from frozen midbrain (from male Sprague-Dawley rats; Taconic Labs) were homogenized in 20 vol (w/v) of 50 mM TRIS, 120 mM NaCl, 5 mM KCl pH 7.4 at 25°C, using a Brinkman Polytron (setting 6 for 20 s). The tissue was centrifuged at 20,000×g for 10 min at 4°C and the resulting pellet was resuspended in buffer and recentrifuged. The final pellet was resus-

One compound, 4′,4′′-diCl-BZT was subjected to a characterization of its activity at various receptor sites (ProfilingScreen procured from MDS Panlabs Pharmacology Services, Bothell, Wash., USA). The screen consists of a panel of radioligand binding assays designed to profile the activity of the compound at 31 mammalian receptors. The compound was screened in duplicate in each assay at a concentration of 10 µM. If there was greater than 50% inhibition of ligand binding at that concentration, the test was repeated in duplicate at the original concentration, and at 10-, 100-, and 1,000-fold lower concentrations to get an approximation of affinity for the site. Concurrent vehicle and reference standards were conducted with each assay, and the following sites were targeted: adenosine Al, adenosine A2A, adrenergic-αl, (non-selective), adrenergic-α2, (non-selective), adrenergic-βl, adrenergic-β2, calcium channel type L (dihydropyridine site), dopamine Dl, dopamine D2L, estrogen, GABAA (agonist site), GABAA (chloride channel), glucocorticoid, glutamate (NMDA, phencyclidine site), glutamate (non-selective), glycine (strychnine-sensitive), histamine Hl (central), insulin, muscarinic M2, muscarinic M3, opiate δ, opiate κ, opiate µ, phorbol ester, potassium channel [KATP], progesterone, serotonin 5-HT1, serotonin 5-HT2, sigma, non-selective, sodium channel (site 2), and testosterone. For details of procedures, tissues, buffers, and other details, see MDS Panlabs Pharmacology Services (2000). For sites at which activity was identified, IC50 values were determined by a non-linear, least squares regression analysis using Data Analysis Toolbox (MDL Information Systems, San Leandro, Calif., USA). Where inhibition constants (Ki) are presented, the Ki values were calculated using the equation of Cheng and Prusoff

365 (1973) using the observed IC50 of the tested compound, the concentration of radioligand employed in the assay, and the MDS Panlabs historical value for the Kd of the ligand. Because IC50 were determined from four concentrations of cold ligand, the derived binding constants should be interpreted with caution. Locomotor activity Male Swiss Webster mice (Taconic Farms) were placed singly in clear acrylic chambers (40 cm3) for the assessment of horizontal locomotor activity (ambulation). The acrylic chambers were contained within monitors (Omnitech Electronics, Columbus, Ohio, USA) equipped with light sensitive detectors, spaced 2.5 cm apart along two perpendicular walls. Mounted on the opposing walls and directed at the detectors were infrared light sources. One horizontal activity count was registered each time the subject interrupted a single light beam. Mice were injected [intraperitoneal (IP) injections administered in volumes of 1 ml/100 g] and immediately placed in the apparatus for 1 h, with horizontal activity counts totaled each 10 min. Each drug dose was studied in eight mice, and mice were used only once. In some experiments the onset and duration of effects on locomotor activity were assessed. Mice were injected and immediately placed in the apparatus for 8 h and data were collected each 10 min. In other studies of time course mice were injected and placed in the chamber at 2 or 3 h (see below) after injection and studied for 1 h. All other aspects of these experiments were identical to those in which activity was assessed for 60 min. Cocaine discrimination Male Sprague-Dawley rats (Charles River, Wilmington, Mass., USA) weighing 320–350 g were individually housed with unrestricted access to water under a 12-h light/dark cycle (lights on 0700 hours). Rats were fed daily about 15 g of standard lab chow at least 30 min after testing. Rats were tested in two-lever operant-conditioning chambers (Med Associates, Model ENV 007, St Albans, Vt., USA) that were housed within light- and sound-attenuating enclosures. White noise was present throughout testing to mask extraneous sounds. Ambient illumination was by a lamp in the top center of the front panel (houselight). Levers were set 17 cm apart, with pairs of lamps (light-emitting diodes, LEDs) above each of the levers, also on the front panel. A downward force on either lever of 0.4 N through about 1 mm was defined as a response, and produced an audible click. Reinforced responses dispensed one 45mg pellet (BioServe, Frenchtown, N.J., USA) into a food tray centered between the levers on the front panel of the chamber. Online experimental control and data collection were by PC MSDOS computers with Med Associates interfacing equipment and operating software (Med Associates). Rats were initially trained to press both levers under a 20-response fixed-ratio (FR20) schedule of food reinforcement and to discriminate IP injections of 29 µmol/kg cocaine (10 mg/kg) from IP injections of saline. After cocaine injection, responses on only one lever were reinforced; after saline injection, responses on the other lever were reinforced. The assignment of cocaine- and saline-appropriate levers was counterbalanced across rats. Immediately after injection, rats were placed inside the experimental chambers. A 5-min time-out period, during which the houselight and LEDs were extinguished and responding had no scheduled consequences preceded the illumination of the houselight and the LEDs. Only responses on the appropriate lever were reinforced and responses on the inappropriate lever reset the FR response requirement. Each food presentation was followed by a 20-s timeout period during which all lamps were off, and responding had no scheduled consequences. Sessions ended after 20 food presentations or 15 min, whichever occurred first. Training sessions with cocaine (C) and saline (S) injections were conducted daily 5 days per week, and ordered in a double alternation sequence (e.g....SCCS...).

Testing was initiated when performances reached criteria of at least 85% appropriate responding overall and during the first FR20 of the session over four consecutive sessions. Tests were conducted with different doses of cocaine, doses of the novel compounds, or combinations of doses administered prior to sessions. Selected doses of the test compounds were administered at different times up to 120 min before session in order to examine the time course of the discriminative-stimulus effects. After a test session, a subject was required to meet the above performance criteria over two consecutive (cocaine and saline) training sessions in order to be tested again. Repeated test sessions were conducted, with at least two training sessions between tests, until entire dose effects were determined in each subject. Test sessions were identical to training sessions, with the exception that 20 consecutive responses on either lever were reinforced.

Analysis of data Locomotor activity in mice was assessed with counts collected during each successive 10-min epoch; counts during the first and last three epochs of the 1-h assessments were cumulated for separate analyses of the first and last 30 min. Effects of individual doses were determined significant by analysis of variance (ANOVA) and subsequent planned comparisons (Stevens 1990). In general, the effects obtained in the first and second 30-min periods were comparable though there were some quantitative differences. Therefore only the analyses of data from the 30-min period in which maximal stimulation was obtained are described. Half-maximum stimulation was calculated by adding the number of horizontal locomotor activity counts at the dose that produced the largest increase in activity to the number of counts after vehicle injection. That sum was divided by two. The dose that produced this half maximal stimulation (ED50 value) was determined by linear regression. For these analyses, points on the linear part of the ascending portion of the dose-effect curve were used. The data from the 8-h observation period were analyzed using two-way analysis of variance (ANOVA) and post-hoc testing to determine significance of effects at individual doses, time epoch, and their interaction. For analysis of the stimulant dose-effect curves we selected the 30-min time period at which the highest level of locomotor activity was obtained from among the time periods at which there was a significant stimulant effect compared to vehicle. This was the most conservative way to assess the effectiveness compared to cocaine of the BZT. For each of the rats studied in the cocaine-discrimination procedure, the overall response rate and the percentage of responses occurring on the cocaine-appropriate lever were calculated. The mean values were calculated for each measure at each drug dose tested. If less than half of the rats responded at a particular dose, no mean value was calculated for percentage of cocaine-appropriate responding at that dose. At least 20% cocaine-appropriate responding was adopted as a conservative criterion at which to assume a significant difference from saline; 80% or higher cocaineappropriate responding was taken as similar to the training dose of cocaine, and intermediate levels of cocaine-appropriate responding were considered partial substitution. Each dose-effect curve was analyzed using standard ANOVA and linear regression techniques. ED50 values and their 95% confidence limits were derived from data using the linear portions of the dose-effect curves (Snedecor and Cochran 1967). Pairs of ED50 values were considered to be significantly different if their 95% confidence limits did not overlap. In order to assess the degree of change in the cocaine dose-effect curve produced by coadministration of the BZT analogs, data were also analyzed by standard parallel-line bioassay techniques as described by Finney (1964). This analysis consists of a one-way ANOVA that determines whether the slopes of the two dose-response curves are significantly different from parallel, and fits a common slope to the two dose-response curves. It then compares the ratio of doses for a 50% effect to provide a value for relative potency as a measure of the degree of shift in the cocaine dose-effect curve. The relative

366 magnesium bromide (18 ml, 1 M solution in ether) was added dropwise to the stirring solution at 0°C. The reaction mixture was allowed to stir for 2 h at 0°C and was then quenched slowly by the addition of 3 ml H2O. The organic layer was washed with 2.8 N HCl (2×50 ml), saturated NaHCO3 (1×25 ml) and brine (1×25 ml) and dried (Na2SO4). Upon filtration of inorganics, the solution was concentrated in vacuo to give 3.68 g 3′,4′′-dichlorobenzhydrol (~100% crude yield) which was dissolved in SOCl2 (50 ml) and stirred at room temperature under argon overnight. The excess SOCl2 was removed in vacuo. Toluene (3×50 ml) was added and removed in vacuo to ensure complete removal of the SOCl2. The crude acylchloride (3.07 g) was reacted without further purification with tropine hydrate (1.78 g, 12.6 mmol) that was melted and stirring at 160°C. The reaction was allowed to stir at 160°C for 1 h. The resulting residue was allowed to cool and was dissolved in ether (30 ml), washed with 2.8 N HCl (3×20 ml) and evaporated to an oil. Purification by flash column chromatography (eluting solvent: CHCl3/MeOH/NH4OH, 95:5:1) gave 1.99 g of the desired product (53% yield) which was converted to the HCl salt in 2propanol and recrystallized from hot ethyl acetate. Melting point 176–180°C; analysis C, H, N: theoretical: C, 61.30%; H, 5.88%; N, 3.41%, actual: C, 61.09%; H, 5.94%; N, 3.30%. Synthetic method B

Fig. 2 Synthetic scheme for 3′,4′′-diCl-BZT potency value represents the dose of cocaine, in subjects co-administered one of the BZT analogs, equal to 1 µmol/kg cocaine alone (i.e. a relative potency value of 0.5 indicates a 2-fold shift to the left of the cocaine dose-effect curve in the presence of the BZT analog). A significant shift in the cocaine dose-effect curve is indicated when the 95% confidence limits for the relative potency ratio do not include the value 1.0. Drugs The drugs tested were: (–)-cocaine HCl (Sigma, St Louis, Mo., USA), and diphenylmethoxytropane analogs of BZT. The basic skeleton of the diphenylmethoxytropane analogs is shown in Fig. 1. Substitutions examined in the present study were exclusively on the 4′-, 4′′-, or 3′-positions and were: 3′-Cl-BZT, 4′-ClBZT, 4′,4′′-diCl-BZT, and 3′,4′′-diCl-BZT. All drugs were dissolved in 0.9% NaCl. The drugs were administered IP on the basis of body weight at 1 ml/kg (rats) or 1 ml/0.1 kg (mice). The synthesis of all of these BZT analogs with the exception of 3′,4′′-diClBZT have been described previously (Newman et al. 1994, 1995; Kline et al. 1997). The synthesis of 3′,4′′-diCl-BZT is described below. General analytical methods Melting points were determined on a Thomas-Hoover melting point apparatus. The 1H and 13C NMR data were recorded on a Bruker (Billerica, Mass., USA) AC-300 instrument and all spectra corresponded to the structures assigned. Infrared spectra were recorded on a Perkin Elmer 1600 series FTIR and all spectra corresponded to functional groups assigned. Microanalyses were performed by Atlantic Microlab, Inc (Norcross, Ga., USA) and agree within 0.4% of calculated values. All reagents were purchased from Aldrich. The synthetic scheme is shown in Fig. 2. Synthetic method A 3-Chlorobenzaldehyde (2.03 g, 14 mmol) was dissolved in 50 ml anhydrous diethyl ether and the Grignard reagent 4-chlorophenyl-

In an effort to improve the yield from the 3′,4′′-dichlorobenzhydrol, an alternate strategy was employed. 3′,4′′-Dichlorobenzhydrol (1.6 g, 6.3 mmol), tropine hydrate (0.452 g, 3.2 mmol) and ptoluenesulfonic acid monohydrate (1.20 g, 6.3 mmol) were dissolved in benzene (200 ml) in a 500 ml round bottom flask fitted with a Dean-Stark trap. The reaction mixture was allowed to stir at reflux for 6 h. Additional 3′,4′′-dichlorobenzhydrol (1.6 g, 6.3 mmol), p-toluene sulfonic acid monohydrate (0.156 g, 0.82 mmol) was added to the reaction flask and was stirred at reflux overnight. The benzene was removed in vacuo and the residue was dissolved in H2O, basified to pH 9 with concentrated NH4OH and extracted with CHCl3 (3×100 ml). The combined organic fraction was dried (K2CO3), filtered and concentrated to a crude oil. Purification by flash column chromatography (eluting solvent: CHCl3/MeOH/NH4OH, 97:3:0.5) gave 1.45 g (100%) product as an oil. The free base was converted to the HBr salt in HBr/MeOH and recrystallized from 2-propanol and anhydrous diethyl ether. Melting point 102–104°C; analysis C, H, N: theoretical: C, 55.16%; H, 5.29%; N, 3.06%, actual: C, 55.39%; H, 5.57%; N, 2.97%.

Results Dopamine uptake inhibition and radioligand binding assays All of the drugs displaced [3H]WIN 35,428 from caudate putamen membranes with relatively high affinity. The Ki values determined for these 3′-and 4′-substituted BZT analogs were all very similar, ranging from 20.0 to 32.5 nM (Table 1). None of the present 3α-diphenylmethoxytropane analogs produced a displacement profile that was better fit to a two-site model than a one-site model. These affinities were from approximately 4- to 6fold greater than those for the parent compound, BZT (Katz et al. 1999) and 6- to 9-fold greater than that obtained for cocaine using a single-site model, and comparable to that for cocaine high-affinity binding using a two-site model (Izenwasser et al. 1994). In keeping with the affinities for the dopamine transporter, all of the

367 Table 1 Potencies of benztropine analogs in binding to the dopamine transporter and M1 muscarinic receptors, norepinephrine transporter, and serotonin transporter Compound

DAT Ki value (nM)

DA uptake inhibition IC50 value (nM)

M1 Ki value (nM)

NET Ki value (nM)

SERT Ki value (nM)

Cocaine BZT* 4′-Cl-BZT 4′,4′′-diCl-BZT 3′-Cl-BZT 3′,4′′-diCl-BZT

187 (±18.7) 118a (±10.6) 30.0c (±3.60) 20.0c (±2.80) 21.6d (±1.51) 32.5 (±4.88)

236 (20.5) NT 23.1 (1.80) 23.4 (3.00) 12.5 (0.91) 12.3 (1.13)

NT 2.1a (±0.29) 7.89 (±0.85) 47.9 (±5.18) 0.98a (±0.01) 21.5 (±2.63)

3450 (±445) >10,000b 1470 (±180) 2980 (±182) 451 (±62.5) 1660 (±239)

286 (±37.9) >10,000b 5120 (±395) 1640 (±236) 258 (±19.1) 3870 (±303)

a This b This

value has been reported previously by Katz et al. (1999) value has been reported previously by Newman et al. (1995)

Table 2 Potencies of 4′,4′′diCl-BZT in binding to sites at which it displaced ligand to less than 50% at a concentration of 10 µM

c This d This

value has been reported previously by Kline et al. (1997) value has been reported previously by Newman et al. (1994)

Assay

Ligand

Ki value (nM)

Histamine H1, central Muscarinic M3 Muscarinic M2 Adrenergic α1, non-selective Sigma, non-selective Serotonin 5-HT2 Dopamine D1 Sodium channel, site 2 Opiate µ Adrenergic α2, non-Selective Calcium channel type L, dihydropyridine

3 nM [3H]pyrilamine 0.29 nM [3H]N-methylscopolamine 0.29 nM [3H]N-methylscopolamine 0.25 nM [3H]prazosin 0.8 nM [3H]DTG 0.5 nM [3H]ketanserin 1.4 nM [3H]SCH23390 1.5 nM [3H]batrachotoxin 0.6 nM [3H]diprenorphine 0.7 nM [3H]rauwolscine 0.1 nM [3H]nitrendipine

5.17 10 64 137 152 231 674 766 3070 4140 4990

compounds inhibited the accumulation of dopamine in synaptosomes (Table 1). The least potent of the compounds was cocaine, with the BZT analogs approximately 10- to 20-fold more potent. These potency relations closely followed the affinities for the dopamine transporter as determined by displacement of [3H]WIN 35,428. All of the BZT analogs displaced [3H]pirenzepine from whole brain membranes (Table 1) with affinities (Ki values) ranging from 0.98 for 3′-Cl-BZT (Kline et al. 1997) to 47.9 for 4′,4′′-diCl-BZT. All of the displacement data were better fit to a one-site model than a twosite model. Affinities for displacement of [3H]nisoxetine ranged from 451 to 2980 nM. Affinities for displacement of [3H]citalopram were similarly low, ranging from 258 to 5120 nM. The sites at which 10 µmol of 4′,4′′-diCl-BZT displaced ligand to less than 50% bound are shown in Table 2, along with approximate Ki values. Among the 31 sites evaluated, 4′,4′′-diCl-BZT had activity in the low nM range at central histamine H1 (Ki about 5.17 nM), muscarinic M3 (Ki about 10 nM), and muscarinic M2 (Ki about 63 nM) receptors. In addition, µM affinity was obtained at several other sites (see Table 2). Locomotor activity All of the compounds stimulated locomotor activity at some dose and some time after injection. Cocaine, as has

been demonstrated previously, increased ambulatory activity with a maximum of 531 counts per min during the first 30 min of the session at 118 µmol/kg (Table 2). Though this was the highest dose tested, the dose-effect curve appeared to reach a plateau at that dose (Fig. 3A). Each of the 3′- and 4′-Cl substituted analogs of BZT also stimulated locomotor activity (Figure 3A). The most effective of these were the 4′- and 3′-substituted compounds, which produced maxima of 380 and 355 counts per min, respectively, during the second 30 min of the session. The 3′,4′′-diCl-substituted analog was less effective, producing a maximum of 300 counts per min over 30 min. The least effective of the compounds was 4′,4′′-diCl-BZT, which only marginally increased locomotor activity (see Fig. 3A). Because these BZT analogs uniformly produced the greatest increases in locomotor activity in the second 30 min of the 1-h observation period, the time course of their effects was subsequently assessed over an 8-h period following drug administration. As was seen in the 1-h session, each of the compounds increased locomotor activity, as did cocaine. Two-way analysis of variance of the effects of cocaine revealed significant effects of time, dose and the interaction of the two [F(time)47,1680= 54.7, P<0.001; F(dose)4,1680=32.8, P<0.001; F(txd)188,1680=2.3, P<0.001]. As can be seen in Fig. 4, the locomotor stimulant effects of cocaine were relatively brief, even decreasing within the first 30 min of the 8-h observation period at all of the active doses. Significant stimulant effects of cocaine were generally obtained

368

Fig. 3A, B Dose-dependent effects of benztropine analogs on locomotor activity in mice. Ordinates: horizontal locomotor activity counts after drug administration. Abscissae: dose of drug in µmol/kg, log scale. Each point represents the average effect determined in eight mice. The data are from the 30-min period during a 60-min observation period, in which the greatest stimulant effects were obtained. A Maximal effects obtained in a 30-min period during the 1st hour after injection. Vehicle values averaged 232 (±9.47) for cocaine (first 30 min), and 118 (±21.2) counts/min for the BZT analogs (second 30 min). B Maximal effects obtained in a 30-min period during a 1-h observation period immediately after injection (cocaine), 2 h (3′-Cl-BZT, 4′-Cl-BZT), or 3 h (4′,4′′diCl-BZT, 3′,4′′-diCl-BZT) after injection. Vehicle values for this experiment averaged 208 (±19.4) counts/min. Note that cocaine is more effective than the other compounds, and that 4′,4′′-diCl-BZT was the least effective

through the first 100 min at 59 µmol/kg. The maximal increase in locomotor activity produced by cocaine occurred in the first 10 min and was greater than 600 counts per minute. The average counts per min in the first 30 min in the 8-h study was comparable to that obtained in the 1-h study (Table 3). The 8-h time course study also revealed a significant locomotor stimulant effect of 4′-Cl-BZT with maximal locomotor counts obtained between 130 and 160 min after injection (Table 2). Two-way analysis of variance revealed significant effects of time, dose and the interaction of the two [F(time)47,2016=12.5, P<0.001; F(dose)4,2016=83.4, P<0.001; F(txd)235,2016=1.6, P<0.001]. As can be seen (Fig. 4), the maximal level of activity during the 1-h observation period (approximately 380 counts/min; Fig. 3) that was produced by 26 µmol/kg, was not eclipsed during the 8-h session. At 66 µmol/kg, an initial 40-min suppression of locomotor activity was followed by a long-lasting stimulant effect (Fig. 4) that was significant in the third hour. These effects diminished over the course of the session but remained significant until the last hour of the 8-h session. The time course of the effects of the 3′-Cl-substituted analog of BZT revealed a significant and long-lasting locomotor-stimulant effect that approached, but did not reach, the maximal locomotor count obtained with cocaine (Figure 4). Maximal counts were obtained between 100 and 130 min after injection (Table 2). Two-way

Fig. 4 Time course of effects of benztropine analogs on locomotor activity in mice. Ordinates: horizontal locomotor activity counts after drug administration. Abscissae: time since injection and placement of subject in experimental chamber. Effects of several doses are shown (see symbol keys on each panel). Each point represents the average effect determined in eight mice, for successive 10-min time periods up to 480 min (8 h) after injection. Note that in contrast to cocaine the durations of action of the benztropine analogs were generally greater, with maximal stimulation occurring in the second or third hours and diminished in the 7th or 8th hours after injection

analysis of variance of the effects of 3′-Cl-BZT revealed significant effects of time, dose and the interaction of the two [F(time)47,1680=5.9, P<0.001; F(dose)4,1680= 295.7, P<0.001; F(txd)188,1680=2.4, P<0.001]. At a dose of 26 µmol/kg, which produced maximal levels of activity in the 1-h session, there was a significant increase in activity that occurred in the second 10 min and was long lasting, diminishing only in the 7th hour after injection. At a higher dose (79 µmol/kg), significant stimulant effects were obtained generally after the 3rd hour, and throughout the rest of the 8-h session (Fig. 4).

485 (±31.0)@26 µmol/kg 451 (±26.9)@243 µmol/kg≠

549 (±30.3)@59 µmol/kg 357 (±45.1)@66 µmol/kg

5.55 (3.38–9.29) NS Regr. 120

30–60

30–60

30–60

30–60

4′-Cl-BZT

4′,4′′-diCl-BZT

3′-Cl-BZT

3′,4′′-diCl-BZT

in the regression aA non-significant linear regression precluded calculation of an ED50 value bValue is not provided because there was a significant deviation from linearity

180

160–190

NS Regr. 180

130–160

100–130

366 (±28.3)@26 µmol/kg 294 (±34.1)@24 µmol/kg 409 (±38.8)@26 µmol/kg 305 (±16.8)@24 µmol/kg NS Regr.a 120

8.33 (6.35–11.0) 50.6b

25.6 (19.2–32.4) 24.8 (17.2–39.6) 0–30 Cocaine

17.9 (13.8–21.5) 9.39 (4.57–13.2) 6.98 (0.583–39.9) 8.14 (5.79–11.5) 15.0 (11.7–20.1)

531 (±23.2)@118 µmol/kg 380 (±26.3)@26 µmol/kg 183 (±15.8)@24 µmol/kg 355 (±24.6)@26 µmol/kg 300 (±28.2)@24 µmol/kg

0–30

Time period (min) Maximal stimulation (counts/min) ED50 value (µmol/kg) Pretreatment time (min) Time period

ED50 value (µmol/kg)

Maximal stimulation (counts/min)

ED50 value (µmol/kg)

8-h observation period Pretreatment study 1-h observation period Compound

Table 3 Comparisons of effects of benztropine analogs on locomotor activity in the 1-h and 8-h studies

Maximal stimulation (counts/min)

369

As with the 3′-Cl-substituted compound, 3′,4′′-diClBZT produced a significant increase in activity that was long lasting and diminished substantially only in the 7th hour after injection (Fig. 4). The maximal level of locomotor activity was obtained between 160 and 190 min after injection (Table 2). Two-way analysis of variance of the effects of 3′,4′′-diCl-BZT revealed significant effects of time, dose and the interaction of the two [F(time)47,1632=13.0, P<0.001; F(dose)4,1632=172.0, P<0.001; F(txd)188,1632=2.3, P<0.001]. With this compound the dose producing the greatest stimulant effect (243 µmol/kg) produced an initial depression of activity during the first 30 min after injection. Significant stimulant effects were obtained generally with 243 µmol/kg from 80 min after the injection, and diminished in the 8th hour. At a lower dose (73 µmol/kg), significant stimulant effects were generally obtained from 230 to 370 min after the injection. As can be seen in Fig. 4, the maximal levels of locomotor activity were obtained with several of the drugs in the 8-h study at times well after injection, when control levels of activity were much less than those occurring when cocaine had its maximal effect. In order to compare the potential locomotor stimulant effects of the BZT analogs on similar baselines they were administered, prior to a 1-h session, at the time ensuring assessment during the period at which they produced the greatest level of locomotor activity in the 8-h study. As can be seen in Fig. 3B, when compared across similar baselines, cocaine produced increases in locomotor activity greater than those produced by either 3′-Cl-BZT or 4′-Cl-BZT. Smaller effects were obtained with 3′,4′′-diCl-BZT, and 4′,4′′-diCl-BZT. Cocaine discrimination As has been shown previously, there was a dose-related increase in the percentage of cocaine-appropriate responses in subjects trained to discriminate cocaine (29 µmol/kg) from saline (Fig. 5, filled symbols). In contrast, none of the BZT analogs produced a level of drug appropriate responding that exceeded 80% (Fig. 5). The most effective of the BZT analogs was 3′,4′′-diCl-BZT, which produced a maximum of 72% cocaine-appropriate responding at the highest dose examined (14 µmol/kg; higher doses could not be tested due to pronounced effects on response rate; Fig. 5, lower panel, squares). The other 3′-Cl-substituted BZT analog also produced a level of substitution greater than that produced by saline, though it also failed to substitute fully for cocaine (Fig. 5, upper panel triangles pointed up). In contrast, 4′Cl-BZT did not produce substitution significantly greater than saline levels, consistent with a previous study (Katz et al. 1999). The present failure of 4′,4′′-diCl-BZT to substitute at a level significantly greater than that produced by saline has also been reported (Katz et al. 1999). The differences in the effectiveness of the BZT analogs in substituting for cocaine do not depend on the

370

Fig. 5 Effects of benztropine analogs in rats trained to discriminate injections of cocaine from saline. Ordinates for top panels: percentage of responses on the cocaine-appropriate key. Ordinates for bottom panels: rates at which responses were emitted (as a percentage of response rate after saline administration). Abscissae: drug dose in µmol/kg (log scale). Each point represents the effect in four to six rats. The percentage of responses emitted on the cocaine-appropriate key was considered unreliable, and not plotted, if fewer than half of the subjects responded at that dose. Note that none of the compounds fully substituted for cocaine, though the 3′-substituted compounds produced a partial substitution that was greater than levels after 4′-Cl-BZT and vehicle

measure of substitution used. Other measures that have been used to assess the substitution of the test drug for the training drug were also examined. These alternate measures were: 1) the percentage of responses on the cocaine-appropriate lever before the first reinforcer, 2) the percentage of subjects selecting the cocaine-appropriate lever (those with over-the-session percentages of cocaine-appropriate responding greater than 80%), and 3) the percentage of subjects selecting the cocaine-appropriate lever (those with greater than 80% cocaine lever responses) on the first trial of the session. There was a very close correspondence among all of these measures (data not shown) including indications of maximum effect (maximum % cocaine-appropriate responding) and potency (ED50 values). The r2 values for the correlation of the percentage of responses on the cocaine-appropriate lever over the entire session and the three alternative measures listed above were 0.9903, 0.9823, and 0.9447, respectively. All four of these measures indicated that none of the compounds had maximal effects comparable to those of cocaine, and that the 3′-Cl substituted analogs generally were more effective than the 4′-Cl-substituted analog.

Because locomotor stimulant effects of some of the BZT analogs appeared later after injection, the timecourse of effects of 3′-Cl-BZT and 3′,4′′-diCl-BZT was studied; the time-course of the discriminative effects of of 4′-Cl-BZT and 4′,4′′-diCl-BZT has been published (Katz et al. 1999). Maximal substitution for cocaine was obtained with 3′-Cl-BZT at a dose of 26 µmol/kg when it was injected 5 min before testing (Fig. 5). When the time between injection and testing was extended to 30, 60, and 90 min there were no further increases in drugappropriate responding. The maximum percent drug-appropriate responding produced by 3′-Cl-BZT in the timecourse study averaged 33.3% after 26 µmol/kg at 90 min. The maximal decreases in response rates produced by 3′Cl-BZT at either 7.9 or 26 µmol/kg were obtained when it was injected at 5 min before testing. By 60 min after injection response rates were at control levels. Maximal substitution for cocaine was obtained with 3′,4′′-diCl-BZT at doses of 1.3 and 13.6 µmol/kg when injected 5 min before testing (Fig. 5). When the time between injection and testing was extended to 30, 60, 90 and 120 min there were no further increases in drug-appropriate responding by either of these doses. The maximal decreases in response rates produced by 3′,4′′-diClBZT at either dose were obtained when it was injected at 5 min before testing. By 30 min after injection response rates were at control levels. Drug interactions In general the BZT analogs shifted the dose-effect curve for discriminative-stimulus effects of cocaine to the left in a dose-related manner. For example, 7.9 µmol/kg of 4′-Cl-BZT shifted the cocaine dose-effect curve to the left by a factor of about two. The 26 µmol/kg dose shifted the cocaine dose-effect curve approximately 17-fold (Fig. 6; Table 2). The 4′,4′′-diCl-analog of BZT also shifted the cocaine dose-effect curve to the left (Fig. 6) by a factor of 5 at 24 µmol/kg (Table 4), whereas the lower dose did not significantly change the potency of cocaine (Table 4 : 95% CL of RP estimate was inclusive of 1.0). Neither of the two lowest doses of 3′-Cl-BZT significantly altered the discriminative effects of cocaine (Fig. 6; Table 4). At the highest dose of 3′-Cl-BZT, the cocaine dose-effect curve was significantly altered, as evidenced by a shift left and a change in slope (Fig. 6). The relative potency estimate indicated that the leftward shift in the cocaine dose-effect curve was significant, however there was also a significant difference in the range of effects obtained, with that after the combination of 3′-Cl-BZT more narrow than the full range of effects displayed by the various doses of cocaine. In general, 3′,4′′-diCl-BZT altered the dose-effect curve for the discriminative-stimulus effects of cocaine in a manner similar to that produced by the other BZT analogs. At a dose of 7.3 µmol/kg, the cocaine dose-effect curve was shifted to the left by a factor of about two; how-

371

Fig. 6 Changes in the cocaine dose-effect curve for discriminative stimulus effects produced by pretreatments with the benztropine analogs. Ordinates: percentage of responses on the cocaine-appropriate key (top panels) or response rates (bottom panels). Abscissae: cocaine dose in µmol/kg (log scale). Each point represents the Table 4 Effects of interactions of benztropine analogs with cocaine

aValue

is not provided because the regression was not significant bValue is only an estimate because there was a significant effect of preparations

effect in four to six rats. The percentage of responses emitted on the cocaine-appropriate key was considered unreliable, and not plotted, if fewer than half of the subjects responded at that dose. Note that each of the benztropine analogs shifted the cocaine dose-effect curve to the left

Condition

Cocaine discrimination ED50 value (µmol/kg)

Cocaine alone Cocaine with 7.9 µmol/kg 4′-Cl-BZT Cocaine with 26 µmol/kg 4′-Cl-BZT Cocaine alone Cocaine with 2.4 µmol/kg 4′,4′′-diCl-BZT Cocaine with 7.3 µmol/kg 4′,4′′-diCl-BZT Cocaine alone Cocaine with 2.6 µmol/kg 3′-Cl-BZT Cocaine with 7.9 µmol/kg 3′-Cl-BZT Cocaine with 26 µmol/kg 3′-Cl-BZT Cocaine alone Cocaine with 2.4 µmol/kg 3′,4′′-diCl-BZT Cocaine with 7.3 µmol/kg 3′,4′′-diCl-BZT

7.32 (4.94–10.24) 3.43 (2.65–4.55) 0.40 (0.24–0.56) 7.32 (4.94–10.2) 5.66 (2.53–48.2) NS regressiona 10.4 (7.32–15.3) 11.3 (7.84–18.1) 6.62 (3.44–10.7) 0.20 (0.00–0.99) 8.76 (3.29–21.5) 4.80 (1.77–8.29) 3.16 (0.94–5.38)

ever, this effect did not achieve statistical significance, as evidenced by 95% confidence limits for the relative potency estimate that were inclusive of the value 1.0.

Discussion In the present study, the pharmacology of a selected series of phenyl-ring analogs of benztropine (BZT) was

Relative potency

0.52 (0.35–0.80) 0.06 (0.03–0.09) 0.81 (0.32–3.76) 0.22 (0.04–2.03) 1.19 (0.72–2.02) 0.71 (0.35–1.32) 0.04b (0.00–0.18) 0.70 (0.25–1.67) 0.50 (0.18–1.10)

evaluated. The present results are consistent with previous studies that have demonstrated that BZT analogs can have high affinity and selectivity for the dopamine transporter compared to the norepinephrine and serotonin transporters (Newman et al. 1995). Further, past and present results demonstrate that these compounds are potent inhibitors of dopamine transport in a functional assay. Despite this profile, the BZT analogs did not have behavioral effects that were equivalent to those of co-

372

caine, a finding that is also consistent with that reported previously for a wider series of BZT analogs (Katz et al. 1999). In that study as in the present one, several but not all of the BZT analogs produced dose-related increases in locomotor activity in mice, however, maximal effects were uniformly less than that produced by cocaine. In rats discriminating cocaine from saline, cocaine produced a dose-related increase in responding on the cocaine-appropriate lever, approximating 100% at the training dose (29 µmol/kg). In contrast, none of the BZT analogs fully substituted for cocaine, with maximum cocaine responding of from 20% for 4′-Cl-BZT to 73% for 3′,4′′-diCl-BZT. The dose-effect curves for cocaine substitution indicated some cocaine-like actions, but only at the highest doses at which response rates are substantially decreased. It is possible that the rate-decreasing effects of high doses of the benztropine analogs interfered with their effectiveness in substituting for cocaine. Because of this possibility, we analyzed all lever selection data regardless of response rate (though we excluded lever selection data when less than half of the subjects responded). In addition, we examined a number of alternative measures of lever selection to ensure the reliability of the partial substitution by the BZT analogs. Across these various measures, and with a liberal inclusion of data, the BZT analogs consistently showed efficacy less than that of cocaine. BZT has well documented antimuscarinic effects and each of its analogs inhibited [3H]pirenzepine binding, with affinities ranging from 0.98 nM for 3′-Cl-BZT to 47.9 nM for 4′,4′′-diCl-BZT. It could be suggested that the decreased effectiveness of the BZT analogs in producing locomotor stimulation or cocaine-like discriminative effects is due to events consequent to muscarinic receptor binding that interfere with the expression of cocaine-like behavioral effects. However, several previous studies have shown that antimuscarinics potentiate stimulants, rather than interfering with various behavioral effects (e.g. Scheckel and Boff 1964), and the potentiation was replicated with cocaine and atropine or scopolamine in procedures identical to the present ones (Katz et al. 1999). In the present study, 3′-Cl-BZT most closely approximated the locomotor-stimulant effects of cocaine and was the most effective in substituting in subjects discriminating cocaine from saline. However, this compound, among those presently studied, had the highest affinity for M1 receptors, which was about 22-fold higher than its affinity for the dopamine transporter. In contrast, 4′-Cl-BZT and 4′,4′′-diCl-BZT (Katz et al. 1999) each were relatively less effective in producing cocainelike stimulus effects, yet each of these compounds had lower affinity for M1 receptors and their respective ratios of dopamine transporter to M1 affinities were substantially lower than that for 3′-Cl-BZT. In fact, 4′,4′′-diClBZT had a higher affinity for the dopamine transporter than for M1 receptors and this compound was previously reported to be relatively ineffective as a stimulant of locomotor activity and to lack cocaine-like discriminative

stimulus effects (Katz et al. 1999). Thus, the present findings are consistent with the previous study suggesting that a lack of full cocaine-like behavioral effects is likely not due to an interference of those effects by M1 muscarinic antagonist effects of the BZT analogs. As indicated above, there are differences among the individual BZT analogs with regard to their effectiveness in producing cocaine-like effects, in addition to the differences among the present BZT analogs. The relatively low efficacy of 4′-Cl-BZT is consistent with our previous report on this compound (Newman et al. 1994). A subsequent study indicated that 3′-Cl-BZT fully substituted for cocaine (Kline et al. 1997). While that full substitution wasn't completely replicated in the present study, 3′-ClBZT was less effective than cocaine, but still more effective than its 4′-Cl-BZT homolog. In addition, in the present study 3′,4′′-diCl-BZT partially substituted for cocaine. Finally, Katz et al. (1999) demonstrated that 4′,4′′diF-BZT, but neither 4′-Cl- nor 4′,4′′-diCl-BZT, had full cocaine-like discriminative effects, but with a delayed onset. Thus, the present findings along with the initial study of 3′-Cl-BZT (Kline et al. 1997) indicate that the BZT analogs are not universally devoid of cocaine-like behavioral effects, but that there can be gradations in cocaine-like efficacy within this series of compounds. As mentioned above, we have previously focused on the possibility that antimuscarinic actions interfere with cocaine-like discriminative-stimulus effects, and the currently available evidence suggests otherwise (Katz et al. 1999). However, the possibility that some other action of the BZT analogs interfered with their cocaine-like effects cannot be ruled out. The presently reported screen of various binding sites accessed by 4′,4′′-diCl-BZT indicated nanomolar affinity for a number of other receptors. Effects through these receptors may mediate actions that interfere with the expression of cocaine-like activity. Obviously, it is not known at present whether the profile of binding activity for this individual compound is representative of the other analogs. High affinity for M3 and M4 muscarinic receptors was indicated for 4′,4′′-diCl-BZT. The possibility that these muscarinic actions might interfere with a cocaine-like effect seems unlikely, however, due to the previous studies with scopolamine and atropine (Katz et al. 1999) and the reported lack of selectivity of atropine among these subtypes of muscarinic receptors (Buckley et al. 1989; Moriya et al. 1999). The present screen also indicated nanomolar affinity of 4′,4′′-diCl-BZT to histamine H1 receptors, activity at other histamine receptors was not assessed. This affinity was anticipated on the basis of the histamine antagonist effects of the parent compound (Esplin 1965). A number of studies have suggested an interaction of central histamine and dopamine systems that might have relevance to the present results (e.g. Schlicker et al. 1993; Bergman and Spealman 1988). More recently, Bäckström et al. (2000) have indicated that the histamine H3-receptor antagonist, thioperamide, dose-dependently decreased the number of cocaine infusions in rats self administering

373

cocaine. In contrast, the histamine H3-receptor agonist, R-α-methylhistamine, had no effects. In locomotor activity experiments, thioperamide potentiated cocaineinduced locomotor activity. These findings suggest that if the present BZT analogs have antagonist activity at H3 as well as H1 histamine receptors, they would be expected to enhance their cocaine-like actions. Subsequent studies will examine the contribution of histaminic actions to the effects reported here. Several other sites were found at which 4′,4′′-diClBZT had micromolar affinity. It remains possible that actions at σ, α1 adrenergic and 5-HT2 receptors may play a role in the present effects. Affinity at these sites was from 7- to 12-fold lower than affinities at the dopamine transporter. Actions at the other sites at which 4′,4′′diCl-BZT had effects seem unlikely factors in that affinity at these sites were from 34 to 250 times lower affinity at this site than the dopamine transporter. Obviously, it will never be possible to rule out the interpretation that some other action exists which interferes with the expression of a cocaine-like discriminative stimulus. However, it should be noted that the reduced effectiveness of the BZT analogs in the cocaine discrimination procedure is paralleled by reduced effectiveness in stimulation of locomotor activity (present study and Katz et al. 1999) and reduced effectiveness in maintaining responding in a self-administration procedure (Woolverton et al. 2000, 2001). Any hypothesis regarding some other action interfering with the ability of the BZT analogs to express full efficacy should also parsimoniously account for their reduced efficacy for these other two behavioral effects. One general finding with the analogs of BZT is that those with para-Cl substituents have a relatively high affinity for the dopamine transporter but generally do not produce effects of a magnitude comparable to those produced by cocaine. The data from the present study along with those for 4′,4′′-diCl-BZT from a previous study (Katz et al. 1999) rank order the effectiveness of the drugs in stimulation of locomotor activity as follows: cocaine>3′-Cl-BZT≈3′,4′′-diCl-BZT>4′-Cl-BZT>4′,4′′-diClBZT. For the production of cocaine-like discriminative stimulus effects the rank order is: cocaine>3′,4′′-diClBZT>3′-Cl-BZT>4′,4′′-diCl-BZT>4′-Cl-BZT. In general, the analogs with 3-Cl-substituents are less effective than cocaine but more effective than those with 4-Clsubstituents. Thus, BZT analogs with 4-Cl-substituents were least effective, but that substitution was not sufficient to decrease the effectiveness of 3′,4′′-diCl-BZT. Therefore, it appears that the meta-Cl substituent is a critical structural feature contributing to an increased cocaine-like effectiveness of the present BZT analogs. Because of the reduced effectiveness among the BZT analogs compared to cocaine, it was of interest to examine the interactions of these compounds with cocaine. As has been reported previously for 4′-Cl-BZT (Tolliver et al. 1999), cocaine discriminative-stimulus effects were shifted leftward by co-administration of the present BZT analogs. The leftward shift in the cocaine dose-effect

curve was not necessarily expected for compounds that bind to the same site as cocaine, but do not produce effects similar to those of cocaine. One hypothesis that will be examined in a subsequent study is that the BZT analogs compete with cocaine for plasma protein binding. As a consequence, greater concentrations of free cocaine in blood would result in greater central cocaine concentrations. Alternatively, recent studies have suggested that the binding of BZT analogs to the dopamine transporter is at domains on the protein that are different from those for cocaine (Vaughan et al. 1999). It is possible that with binding to these domains by a BZT analog, the transporter protein assumes a conformational state that renders it more susceptible to uptake inhibition by cocaine. By analogy to methadone as a treatment for heroin abuse, the present results suggest that analogs of BZT may be useful as treatments for cocaine abuse in situations in which an agonist treatment is indicated. By analogy to buprenorphine, these compounds may possess a therapeutic advantage over other inhibitors of dopamine transport in that these compounds had a reduced in vivo efficacy compared to cocaine. In addition, the stimulant effects were long lasting, adding to the potential of these drugs as therapeutics. Finally, recent studies indicate that neither 3′-Cl- nor 4′-Cl-BZT have efficacy like that of cocaine in a drug self administration procedure (Woolverton et al. 2000, 2001). Thus the most recent data showing reduced stimulant efficacy, long-lasting effects, and a lack of significant abuse liability suggest that there is a potential for use of certain BZT analogs as therapeutics to treat cocaine abuse. Acknowledgements The authors would like to thank Bettye Campbell, Jian Jing Cao, Dawn French, and Rob Mitkus for technical support, Patty Ballerstadt for administrative and clerical support, and Dawn French for expert data analysis. Collection of some of the locomotor activity data were funded through a contract with the NIDA Medications Development Division (Michael J. Forster, PI). These studies were supported in part by an IntraAgency Agreement with the NIDA Division of Treatment Research and Development. We thank Dr. F. Vocci for his support. Portions of this paper were presented at the 1998 Annual Meeting of the College on Problems of Drug Dependence.

References Bäckström P, Crockford S, Hyytiä P (2000) Histamine H3-receptor antagonism enhances cocaine reinforcement. Soc Neurosci Abstr 30:816.7 Bergman J, Spealman R (1988) Behavioral effects of histamine H1 antagonists: comparison with other drugs and modification by haloperidol. J Pharmacol Exp Ther 245:471–478 Buckley NJ, Bonner TI, Buckley CM, Brann MR (1989) Antagonist binding properties of five cloned muscarinic receptors expressed in CHO-K1 cells. Mol Pharmacol 35:469–476 Cheng Y, Prusoff WH (1973) Relationship between the inhibition constant (Ki) and the concentration of inhibitor which causes 50 per cent inhibition (IC50) of an enzymatic reaction. Biochem Pharmacol 22:3099–3108 Esplin DW (1965) Centrally acting muscle relaxants; drugs for Parkinson's disease. In: Goodman LS, Gilman A (eds) The pharmacological basis of therapeutics, 3rd edn. Macmillan, Toronto, Ontario, pp 237–246

374 Finney DJ (1964) Statistical methods in biological assay, 2nd edn. Hafner, New York Izenwasser S, Rosenberger JG, Cox BM (1993) The cocaine analog WIN 35,428 binds to two sites in fresh rat caudate-putamen under the correct assay conditions. Life Sci/Pharmacol Lett 52:PL141–145 Izenwasser S, Terry P, Heller B, Witkin JM, Katz JL (1994) Differential relationships among dopamine transporter affinities and stimulant potencies of various uptake inhibitors. Eur J Pharmacol 263:277–283 Katz JL, Izenwasser S, Kline RH, Allen AC, Newman AH (1999) Novel 3α-diphenyl-methoxytropane analogs: selective dopamine uptake inhibitors with behavioral effects distinct from those of cocaine. J Pharmacol Exp Ther 288:302–315 Kline RH, Izenwasser S, Katz JL, Joseph DB, Bowen WD, Newman AH (1997) 3′-Chloro-3α-(diphenylmethoxy)tropane but not 4′-chloro-3α-(diphenylmethoxy)tropane produces a cocaine-like behavioral profile. J Med Chem 40:851–857 Kuhar MJ, Ritz MC, Boja JW (1991) The dopamine hypothesis of the reinforcing properties of cocaine. Trends Neurosci 14:299–302 Moriya H, Takagi Y, Nakanishi T, Hayashi M, Tani T, Hirotsu I (1999) Affinity profiles of various muscarinic antagonists for cloned human muscarinic acetylcholine receptors (mAChR) subtypes and mAChRs in rat heart and submandibular gland. Life Sci 64:2351–2358 Munson PJ, Rodbard D (1980) Ligand: a versatile approach for characterization of ligand-binding systems. Anal Biochem 107:220–239 Newman AH, Allen AC, Izenwasser S, Katz JL (1994) Novel 3α(diphenylmethoxy)tropane analogs: potent dopamine uptake inhibiitors without cocaine-like behavioral profiles. J Med Chem 37:2258–2261 Newman AH, Kline RH, Allen AC, Izenwasser S, George C, Katz JL (1995) Novel 4′- and 4′,4′′-substituted-3α-(diphenylmethoxy)tropane analogs are potent and selective dopamine uptake inhibitors. J Med Chem 38:3933–3940

Newman, AH Robarge M, Izenwasser S, Kline RH (1999) A comparative molecular field analysis (CoMFA) study of novel ring-substituted 3α-(diphenylmethoxy)tropane analogs at the dopamine transporter. J Med Chem 42:3502–3509 Scheckle CL, Boff E (1964) Behavioral effects of interacting imipramine and other drugs with d-amphetamine, cocaine and tetrabenazine. Psychopharmacologia 5:198–208 Schlicker E, Fink K, Detzner M, Gothert M (1993) Histamine inhibits dopamine release in the mouse striatum via presynaptic H3 receptors. J Neural Trans 93:1–10 Snedecor GW, Cochran WG (1967) Statistical methods, 6th edn. Iowa State University Press, Ames, Iowa Stevens JP (1990) Intermediate Statistics. Lawrence Erlbaum, Hillsdale, New Jersey Tolliver BK, Newman AH, Katz JL, Ho LB, Fox LM, Hsu K Jr, Berger PS (1999) Behavioral and neurochemical effects of dopamine transporter ligands alone and in combination with cocaine: characterization of 4-chlorobenztropine in vivo. J Pharmacol Exp Ther 289:110–122 Vaughan RA, Agoston GE, Lever JR, Newman AH (1999) Differential binding sites of tropane-based photoaffinity ligands on the dopamine transporter. J Neurosci 19:630–636 Woolverton WL, Rowlett JK, Wilcox KM, Paul IA, Kline RH, Newman AH, Katz JL (2000) 3′ and 4′-Chloro-substituted analogs of benztropine: intravenous self-administration and in vitro radioligand binding studies in rhesus monkeys. Psychopharmacology 147:426–435 Woolverton WL, Hecht GS, Agoston GE, Katz JL, Newman AH (2001) Further studies of the reinforcing effects of benztropine analogs in rhesus monkeys. Psychopharmacology DOI 10.1007/s002130000616