Rimonabant Redux and Strategies to Improve the Future Outlook of CB1 Receptor Neutral‐Antagonist/Inverse‐Agonist Therapies

Abstract

The endogenous cannabinoid (eCB) system plays a significant role in appetitive drive and feeding behavior. Therefore, antagonism of this receptor-mediated system was predicted to provide therapeutic benefit for the treatment of disorders associated with excess appetitive drive, such as obesity and substance abuse, as well as other disorders (1,2,3). Toward this end, cannabinoid receptor antagonists have been designed and investigated as pharmacotherapeutic agents. Rimonabant was the first-in-class of such compounds. Its biological activity is thought to be mediated by CB1 (cannabinoid receptor subtype 1) antagonist or inverse-agonist action, i.e., it acts as a functional antagonist in vivo. Rimonabant suppresses feeding in laboratory animals (4,5,6) and also shows efficacy at reducing body weight and improving cardiovascular and metabolic risk factors in nondiabetic and diabetic overweight and obese patients (7,8,9,10). However, concerns over psychiatric adverse effects such as depression, anxiety, and suicidal ideations, and other adverse effects (e.g., increased incidence of nausea, vomiting, and pruritis) led to the withdrawal of rimonabant from the market worldwide. The experience with rimonabant prompted the questioning of the CB1 antagonism approach. However, concern has now been expressed that the experience with one or two drugs might have potentially led to premature abandonment of an entire mechanism—a mechanism that might be valuable for treatment of several disorders that currently lack effective therapy (11,12,13,14). That is, the observed adverse effects might be related to the individual drug (a drug effect) rather than the mechanism of action (a class effect). Therefore, further understanding of the behavioral pharmacological, neurochemical, and cellular effects of rimonabant and of more recent novel CB1 receptor antagonists might shed some light on the therapeutic potential of such compounds. Specifically, (i) potential efficacy of peripherally restricted CB1 antagonists for obesity and type 2 diabetes and other disorders, (ii) the effects of CB1 neutral antagonists, as opposed to inverse agonists, on weight and mood disorders, (iii) the extent and mechanism(s) by which central CB1 receptor antagonism produces global effects on anhedonia or depressed mood, and (iv) whether combining lower doses of these agents with agents from other drug classes could improve or maintain efficacy while lessening adverse effects (see also Table 1 for a summary of points 1–3). The eCB system (AKA endocannabinoid) was identified in the early 1990s during investigations into the mechanism of action of Delta(9)-tetrahydrocannabinol, the major psychoactive chemical component in Cannabis sativa (15,16). The cloning of cannabinoid receptors led to the identification of endogenous molecules capable of binding and activating them, defined as “endocannabinoids” because of their affinity to Delta(9)-tetrahydrocannabinol binding sites (17). Demonstrated or putative eCBs include anandamide (arachidonoylethanolamide (18)), 2-arachidonoyl glycerol (19,20), and several others (e.g., refs. 21,22,23,24). The endocannabinoids are generated on an as-needed basis (25,26,27,28,29) and are released postsynaptically to travel retrogradely and bind to presynaptic receptors (30,31,32). Both 2-arachidonoyl glycerol and arachidonoylethanolamide have been found in blood plasma, brain, heart, gut, liver, pancreas, muscle, adipose tissue, kidney, testis, skin, spleen, uterus and skeleton, suggesting that they regulate a wide range of effects through the activation of the cannabinoid receptors present in such tissues. There are two main cannabinoid receptor types, termed CB1 and CB2, both of which are seven-transmembrane (7TM) G protein–coupled receptors. In human, CB1 is found ubiquitously but is preferentially localized in the brain and the spinal cord (33). Indeed, CB1 receptors are one of the most abundant G protein–coupled receptors located within the mammalian brain and are present in highest concentration in olfactory and cortical brain regions (neocortex and pyriform cortex), hippocampus, amygdala, basal ganglia, thalamic and hypothalamic nuclei, cerebellar cortex, and brainstem nuclei (34,35). These central CB1 receptors had previously been most widely studied, as it was presumed that they were the key target underlying rimonabant's antiobesity effects. However, a wealth of more recent evidence has clearly demonstrated the pivotal role that peripheral CB1 receptors play in energy homeostasis in sites such as the gastrointestinal tract, liver, pancreas, skeletal muscle, and adipocytes (see refs. 36,37 for review). Rimonabant (N-(Piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1 H-pyrazole-3-carboxamide (38)) was approved in the European Union for the treatment of obesity (defined as BMI ≥30 kg/m2) or for patients with BMI >27 kg/m2 plus associated risk factors (such as type 2 diabetes or dyslipidemia). It was denied approval by the US Food and Drug Administration due to adverse psychiatric side effects (39) and was subsequently withdrawn from the other markets. Shortly after the European Medicines Agency called for withdrawal of rimonabant from the market in 2008, several pharmaceutical companies (i.e., Merck (Whitehouse Station, NJ), Pfizer (New York, NY), Sanofi-aventis (Paris, France), Solvay (Brussels, Belgium) announced that they would stop further clinical research on this class of drug. However, medicinal chemistry teams within other companies and academia have continued to synthesize novel CB1 receptor antagonists, largely in an effort to generate: peripherally restricted compounds; “neutral” or “pure” antagonists; or compounds having both features. The following is a brief summary of some of the most well-characterized of these alternate CB1 antagonists. Taranabant (MK-0364), N-[(1S,2S)-3- (4-Chlorophenyl)-2-(3-cyanophenyl)-1-methylpropyl]-2-methyl-2-{(5-(trifluoromethyl)pyridin-2-yl)oxy}propanamide (40) was discovered at Merck starting from a high throughput screening lead. It is a cyclic amide analog of the lead compound for oral efficacy and minimized production of reactive metabolites. Mutagenesis and computer-simulated docking studies (41) suggest that taranabant binds to the same general area of the CB1 receptor as does rimonabant, but with some differences, notably that taranabant hydrogen bonds with S(7.39)383 but not K(3.28)192, whereas the opposite is true for rimonabant. This difference is the presumptive explanation for taranabant's higher affinity for the CB1 receptor. Taranabant acts as an inverse agonist and inhibits food intake and weight gain and reduces body weight correlated with brain CB1 receptor occupancy (42). In addition, mechanism-of-action studies have strongly suggested that taranabant's weight loss effects are at least in large part due to increases in energy expenditure and fat oxidation (43). In low-dose (44) and high-dose (45) phase 3 clinical trial studies, taranabant produced statistically significant incidences of psychiatric-related adverse experiences. Otenabant (CP-945,598), 1-[9-(4-Chlorophenyl)-8-(2-chlorophenyl)-9H- purin-6-yl]-4-ethylamino piperidine-4-carboxylic acid amide, was discovered at Pfizer as part of an effort to design conformationallyrestricted analogs of the bioactive conformation of the diarylpyrazole rimonabant (46). It was hypothesized that the low-energy conformation and hydrogen bonding of the amide carbonyl of rimonabant could be mimicked by a purine. It was also desired to achieve good central nervous system penetration and minimize production of reactive metabolites. Otenabant displays high (sub nmol/l) affinity for human CB1 receptors vs. CB2 receptors (>10,000 lower affinity) and antagonizes cannabinoid agonist-induced effects in vivo. It inhibits food intake and stimulates energy expenditure (47) and had reached phase 3 clinical trials before its withdrawal from development. Surinabant (SR147778), 5-(4-Bromophenyl)-1-(2,4-dichlorophenyl)-4-ethyl-N-(1-piperidinyl)-1H-pyrazole-3-carboxamide, from Sanofi-Synthelabo (Paris, France), is a CB1 receptor antagonist that has high (sub nmol/l) affinity for human CB1 recombinant receptors, much lower affinity (>100-fold) for CB2 receptors and is inactive at other receptor, enzyme, or ion channel targets (48). It is orally active and reduces ethanol or sucrose consumption in mice and rats and it reduces food intake in fasted and in nonfasted rats (48). Surinabant reached phase 2 clinical trials before its withdrawal from development. Ibipinabant (SLV-319), 4S-(–)-3-(4-Chlorophenyl)-N-methyl-N'-[(4-chlorophenyl)-sulfonyl]-4-phenyl-4,5-dihydro-1H-pyrazole-1-carboxamidine, was synthesized at Zydus based on structural similarity to rimonabant. The design was based on the bioisosteric replacement of rimonabant's pyrazole moiety with dihydropyrazole(49). N-aminomorpholine was the optimal side chain and the bisulfate salt was best for bioavailability. Computer modeling was consistent with docking of the ligand to the CB1 receptor. Oral efficacy was demonstrated by inhibition of sucrose intake in fasted female Zucker fa/fa rats. Bioisosteric replacement of the dihydropyrazole nucleus of SLV-319 by either imidazole or oxazole results in loss of required CB1 receptor binding conformation and pharmacologic effect as a CB1 receptor antagonist (50). Prior to its withdrawal from development, Ibipinabant reached phase 2 clinical trials. TM38837 is a first-in-class second generation CB1 antagonist derived from 7TM Pharma's discovery program targeting the development of CB1 receptor antagonists acting specifically in the periphery. Although the structure of TM38837 has not been published, it is likely similar to those described in 7TM Pharma patents based on the structure of rimonabant. Most recently, the company announced that a phase 1 clinical trial confirmed TM38837 is restricted to the periphery of the human body. The double-blind placebo controlled crossover phase 1 clinical trial assessed the ability of TM38837 to attenuate the centrally mediated side effects of Delta(9)-tetrahydrocannabinol (51). Mild abdominal discomfort such as nausea and diarrhea was reported in <10% of the subjects receiving the drug. 7TM researchers had previously reported at the 2010 International Obesity Conference in Stockholm that TM38837 decreases body weight comparable to rimonabant in rodents, but does not reach the brain in significant levels. In addition, they report that TM38837 is devoid of a range of behavioral effects indicative of activity at brain CB1 receptors (52). AM4113 (N-piperidin-1-yl-2,4-dichlorophenyl-1H-pyrazole-3-carboxamide analog) is a pyrazole analog structurally related to rimonabant, synthesized at the Center for Drug Discovery, Northeastern University (Boston, MA), in the laboratory of Alexander Makriyannis. AM4113 is a putative neutral antagonist that suppresses food intake and food-reinforced behavior, but unlike rimonabant does not induce signs of nausea in rats (53). AM6545, [5-(4-[4-cyanobut-1-ynyl]phenyl)-1-(2,4-dichlorophenyl)- 4-methyl-N-(1,1-dioxo-thiomorpholino)-1H-pyrazole-3-carboxamide], is a peripherally restricted, less lipid soluble analog of rimonabant that retains high affinity and selectivity for the CB1 receptor. In radioligand displacement assays, AM6545 has a KI of 3.3 nmol/l for CB1 receptor, which is similar to that of rimonabant, and >100-fold CB1/CB2 selectivity (54). It is also synthesized at the Center for Drug Discovery, Northeastern University, in the laboratory of Alexander Makriyannis. Following acute parenteral or oral administration, a brain to plasma concentration ratio of 0.03 was reported, as compared with a ratio of 0.8 for rimonabant. AM6545 can also be classified as a neutral antagonist as it does not reduce GTPγs binding in mouse brain. AM6545 improves the metabolic profile in mice with diet-induced obesity but was reported to be inactive in a range of behavioral paradigms that measure activity at brain CB1 receptors (54,55). However, AM6545 does reduce food-reinforced operant responding and selectively attenuates intake of high-carbohydrate and high-fat diets vs. lab chow (56), despite its restriction to the periphery. LH-21, 5-(4-Chlorophenyl)-1-(2,4-dichlorophenyl)-3-hexyl-1H-1,2,4-triazole, a putatitve peripherally restricted CB1 antagonist, is an analog of rimonabant. It has sub mmol/l affinity for the CB1 receptor and weak selectivity (approximately twofold) over the CB2 receptor (57). It suppresses food intake and body weight gain in rats (58). It has a nearly 1:1 plasma-to-brain concentration ratio in rats and it causes similar reduction in food intake and body weight gain in CB1 receptor knockout mice as in wild-type mice (57), suggesting that the compound's anorectic effect might not be mediated by the CB1 receptor. URB447, [4-amino-1-(4-Chlorobenzyl)-2-methyl-5-phenyl-1H-pyrrol-3-yl](phenyl) methanone, resulted from introduction of a phenyl ring onto a pyrrole-based scaffold to mimic the phenyl ring on the pyrazole-based scaffold of rimonabant. It lowers food intake and body weight gain in mice (59). It appears that URB447 selectively acts at peripheral CB1 receptors. It is speculated that this selectivity is pharmacokinetic, arising as a consequence of high clearance by efflux transporters or by rapid metabolism by CYP cytochromes in astrocytes. O-2050, (6aR,10aR)-3-(1-Methanesulfonylamino-4-hexyn-6-yl)-6a,7,10,10a-tetrahydro-6,6,9-trimethyl-6H-dibenzo[b,d]pyran, a putative neutral CB1 antagonist, is a sulfonamide analog of the polar cannabinoid Δ8-tetrahydrocannabinol. It reduces food consumption and body weight of nonfood-deprived rats (60). Interpretation of these results is complicated by the findings that the compound also reduced water consumption (so it might reduce motivation in general instead of only appetite suppression) and locomotor activity (but this was not solely responsible for the reduction in feeding). It is also not clear whether the compound possesses some unidentified non-cannabinoid activity. Recently, a structural series of novel thioamide derivatives having CB1 receptor antagonist activity was reported by AstraZeneca (61). The compounds are synthesized by the exchange of a thioamide linker for the carboxamide linker in 5,6-diaryl-pyrazine-2-amide derivatives, an exchange that maintains the proposed bioactive conformation. These compounds reduce the body weight of cafeteria diet-induced obese mice, an effect that remains even after a 1–2 week washout period. The anorectic effect of these compounds is believed to be due to neutral-antagonist properties (61). Although it is well established that rimonabant blocks central CB1 receptors that mediate hunger and appetite to produce reductions in food intake, its effects on peripheral tissue metabolism leading to increased energy expenditure are now also generally well accepted. Specifically, at the tissue level, fat mass reduction, liver lipid reduction, and improved insulin sensitivity have been shown. Thus rimonabant's effects could be due to (i) action on brain CB1 receptors that influence tissue metabolism through the autonomic nervous system, (ii) direct action on peripheral CB1 receptors expressed on adipocytes, hepatocytes, pancreatic islets, or skeletal muscle, or (iii) action on both brain and peripheral receptors. Initially, Cota et al. (62) showed that even when pair-fed, mice lacking the CB1 receptor (CB1 receptor “knockout” mice) exhibit significantly less fat mass and higher energy expenditure than do wild-type mice, suggesting a role for CB1 receptors in metabolism. Similarly, Ravinet-Trillou et al. (63) suggested that rimonabant's efficacy includes a metabolic component, because the body weight loss in rimonabant-treated mice was significantly higher than in vehicle-treated mice during a 24-h fasting, as well as in a paired-feeding experiment (but see ref. (64)). Further, Jbilo et al. (65) investigated changes at the tissue and cell level involved in rimonabant-induced reduction of adipose mass in diet-induced obese rats. They reported that reduction in obesity was paired with a restoration of adiposity similar to lean rats and a reversal in gene expression changes induced by obesity. In lean and obese Zucker rats, chronic rimonabant produced tolerance to its hypophagic effect while continuing to affect body weight (66). The role of the endocannabinoid system on the regulation of lipogenesis has been substantiated by the localization of CB1 receptors in white adipose tissue (62) and liver (67). Taken together, these results strongly suggest a peripheral metabolic/energy component to the weight-decreasing effect of rimonabant. Presently, the relative contribution of antagonism of central (appetite-suppressing) vs. peripheral (metabolism-stimulating) CB1 receptors in weight-decreasing effects of rimonabant has not been completely resolved. These findings did suggest that the synthesis of peripherally restricted CB1 antagonists may lead to compounds that maintain efficacy at producing weight loss while being devoid of adverse central nervous system effects. A small but growing number of studies have now examined behavioral effects of putative “peripherally restricted” CB1 receptor antagonists, but most of these compounds have either not been demonstrated to completely lack effects on central CB1 receptors (e.g., LH-21 (58)), have been suggested to include non-CB1 mechanisms (e.g., LH-21 (58), URB447 (59)), or the results regarding energy metabolism or weight loss have not been disclosed (e.g., SR140098 (68)). Two recent exceptions may be the preclinical compound AM6545 and the phase 1 clinical trial compound TM38837 (see above). AM6545 appears to have an exceptional preclinical profile of peripheral restriction, loss of body weight, and improvement of metabolic profile independent of body weight changes (54). Perhaps most surprising and promising is the additional preclinical finding that this putative peripherally restricted drug still maintains efficacy at selectively decreasing motivation for highly palatable foods in addition to its positive metabolic effects (56). 7TM Pharma had reported in press releases that its first-in-class second generation CB1 antagonist TM38837 shows efficacy in various rodent models of obesity and type 2 diabetes and shows restriction to the periphery in nonhuman primates undergoing positron emission tomography scans. In addition, 7TM most recently reported that TM38837 was ineffective at reversing well-established central nervous system-mediated effects of Delta(9)-tetrahydrocannabinol in phase 1 clinical trial with 24 subjects (51). More recent evidence also highlights that framing CB1 antagonist effects on body weight as either central regulation of appetite or peripheral regulation of energy expenditure is overly simplistic. In an attempt to anatomically localize CB1 receptors involved in energy expenditure, conditional mutant mice lacking central CB1 receptors in the forebrain and nucleus of the solitary tract while maintaining CB1 expression on nonneuronal peripheral organs were developed and characterized (69). These CB1 mutant mice displayed a lean phenotype and resistance to diet-induced obesity that appeared to result as a consequence of enhanced sympathetic nervous system activity and a decrease in energy absorption, suggesting that CB1 blockade may specifically be efficacious in treating obesity in patients with low sympathetic tone. An important next step will be to determine the extent to which peripherally restricted CB1 receptor antagonists, putatively devoid of the central nervous system adverse effects, can also impact this system to enhance sympathetic tone in a similarly effective manner. The preponderance of literature supports the notion that rimonabant functions as an inverse agonist in vitro and in vivo. However, it is less clear to what extent this action at the CB1 receptor is necessary for its weight-reducing effects or to what extent this action contributes to its adverse effects. As used here, agonism refers to the dual processes of weak chemical bonding to a receptor (receptor affinity) and consequent activation of some 2nd messenger signal transduction mechanism (intrinsic activity); inverse agonism refers to activation of a receptor that produces an intrinsic effect opposite of prototypical agonists for that receptor; and neutral antagonism refers to binding to a receptor that does not produce any activation of a transduction mechanism, therefore leading to no change in basal response. Although putative neutral CB1 receptor antagonists have been developed (e.g., LH-21, O-2050, AM4113, AM6545, VCHSR1, NESS 0327, and O-2645), the majority of CB1 receptor antagonists, including rimonabant (and taranabant, AM251, AM281, and LY320135), behave as inverse agonists, possessing intrinsic activity and producing effects that are opposite in direction from those produced by eCB agonists at these receptors. Rimonabant has much greater affinity for CB1 than for CB2 receptors (Ki in the low nmol/l range vs. >1000 nmol/l, respectively (38)) and has no significant affinity for other receptor sites, such as adenosine, adrenergic, amino acid, benzodiazepine, cholecystokinin, dopamine, histamine, neurotensin, opioid, purinergic, serotonin, sigma, or tachykinin or Cl−, Na+, Ca2+ or K+ channels (38). It penetrates the blood-brain barrier (70). Rimonabant inhibits guanosine γ-thiophosphatebinding to membranes of cell preparations in which no endocannabinoid is known to be present, suggesting intrinsic activity at CB1. Rimonabant can increase locomotor activity, enhance memory, induce emesis, increase intestinal transit, and decrease food consumption in several rodent models. All of these effects are opposite to those produced by CB1 receptor agonists. However, several seemingly inverse cannabimimetic effects could still result from true neutral antagonism at CB1 receptors that are under basal activation from endogenous endocannabinoids such as anandamide and 2-arachidonoyl glycerol. If this is the case, even neutral CB1 receptor antagonists might function as inverse agonists in vivo. It is also possible that compounds such as rimonabant decrease the number of constitutively active CB1 receptors (which are active even in the absence of endogenous agonist). Neutral antagonists, in this case, would not affect the number of constitutively active receptors. Whether rimonabant's effectiveness at decreasing appetite and body weight is a result of blocking endogenous receptor activation or by generating its own intrinsic activity (inverse agonism) is poorly understood. Previously, only a couple of studies assessed the effect of a CB1 receptor neutral antagonist on these behaviors. However, with the failures of rimonabant and the related taranabant in the clinic, studies of the behavioral pharmacologic effects of putative neutral antagonists have increased. It is important to ascertain whether these compounds retain their effects of feeding and weight loss while perhaps producing a lesser incidence of adverse events. The neutral antagonist LH-21 was reported to suppress chow consumption in food-deprived animals and reduce food intake and weight gain in free-feeding Zucker rats (58), suggesting that neutral antagonism of the CB1 receptor can attenuate feeding behavior and perhaps also increases energy expenditure. Interestingly, LH-21 did not stimulate locomotor activity or induce grooming and scratching behavior at any dose tested, effects often reported following moderate to high doses of rimonabant in rats. Moreover, LH-21 was not associated with anxiety-like behavior. However, since LH-21 has significantly lower blood-brain barrier permeation than rimonabant, the lack of these effects might be explained by its restriction to the periphery rather than to its lack of inverse-agonist activity. In another study, the putative neutral-antagonist O-2050 attenuated food consumption in nondeprived rats (60), suggesting that CB1 antagonism can suppress feeding behavior by blocking eCB signaling, rather than by intrinsic activity at CB1 receptors. Several reports have demonstrated that the CB1 neutral-antagonist AM4113 reduces food intake (53,71,72) and produces weight loss (73). Importantly, treatment with AM4113 may also be devoid of some adverse effects associated with rimonabant. For example, AM4113 neither induces signs of nausea in rats (53), nor does it enhance the retention of a conditioned fear response (74). However, AM4113 produces anxiety-like effects similar to that seen with rimonabant in an open field assay (75). More recently, a novel class of CB1 receptor neutral-antagonist thioamide derivatives have shown efficacy at reducing food intake (61), but the adverse effect profile is currently unknown. Lastly, there is increasing debate as to whether it is feasible for a ligand to behave as a true neutral antagonist in a living system (76,77), and those interested in this discussion as it pertains to CB1 receptor antagonists should see ref. (78). It can be argued that an effective pharmacotherapy for a disorder characterized by excessive drive should treat only the targeted behavior while leaving other healthy behaviors intact. For example, the treatment of uncontrollable or dysregulated eating disorders should selectively target excess eating behavior while sparing homeostatic food consumption, as well as selectively regulate the hedonic value of food without dampening other pleasurable behaviors by creating a general anhedonic state. A large amount of laboratory animal data suggest that the CB1 antagonist rimonabant not only decreases motivation for both standard and palatable lab chows, but also for several classes of abused drugs. So while both obesity and addiction researchers see promise in CB1 antagonists for management of excess food or drug intake, the extent to which these agents selectively attenuate excessive (pathological) appetitive behavior or, instead, create a global anhedonic state remains unclear. Several studies have tested whether CB1 receptor antagonism attenuates excess food consumption motivated by various experimental manipulations. Others have assessed the effect of these agents on homeostatic feeding behavior in normal-weight, nonfood-deprived animals. Overall, rimonabant and similar antagonists robustly attenuate feeding behaviors in food-deprived and/or hyperphagic animals, but can also produce anorectic effects in food-sated animals. For example, rimonabant decreases NPY- (neuropeptide Y) induced feeding in mice made hyperphagic by NPY administration (79), but also decreases motivation for palatable food in fully sated mice (80). Also, rimonabant reduces food intake in obese and lean Zucker rats, but the effect is greater in obese animals (66). In addition, while some reports suggest that rimonabant preferentially attenuates the incentive salience of highly palatable foods (4,5), other reports suggest that consumption of high carbohydrate, high fat, and normal chow is equally suppressed by a CB1 antagonist (81,82). Most recently, we have observed that rimonabant is as effective at suppressing motivation for water in water-deprived mice as it is at suppressing motivation for vanilla-flavored Ensure in food-deprived mice when the work requirement to gain access to either reinforcer is high. However, when the work requirement is low, rimonabant is ineffective at decreasing operant responding for water, but is effective at suppressing Ensure self-administration (S.J. Ward, R.G. Hamby, E.A. Walker, unpublished data). Anhedonia is a condition characterized by an inability to experience pleasure in normally pleasurable activity. In laboratory animals, anhedonia is widely identified by a decrease in seeking and/or consumption of a positive reinforcer. The fact that CB1 antagonists decrease behaviors directed at a wide range of food and drug reinforcers suggests that CB1 receptor antagonists may block the rewarding impact of all positive reinforcers. As mentioned previously, this could greatly decrease the safety and tolerability of these drugs in humans. While mounting preclinical data may support this hypothesis, other studies indicate that some stimuli retain reinforcing or rewarding properties following CB1 receptor antagonism. In the chronic mild stress model of anhedonia in mice, for example, chronic treatment with rimonabant reversed the decreased sucrose consumption induced by exposure to stressors (83), suggesting that CB1 antagonists may actually be protective against a depressive state. Others have reported that rimonabant fails to attenuate cocaine's reinforcing effects in rats (84,85), although rimonabant attenuates cocaine self-administration behavior under more demanding operant schedules (86). Following a similar pattern, rimonabant decreases heroin self-administration under higher ratio requirements, but does not attenuate heroin-maintained responding at a low ratio requirement (87,88). Also, while rimonabant attenuates conditioned place preference induced by several abused drugs (for review see ref. (89)), rimonabant fails to produce an aversion on its own, and has actually been reported to induce a place preference ((refs. 90,91), but see ref. (92)). Aside from food and drug reinforcers, the authors are unaware of any published studies regarding the effect of rimonabant treatment on other positive reinforcers, such as novel environment or social interaction. Interestingly, a recent study demonstrated that the CB1 antagonist/inverse agonist AM251 actually increased sexual motivation and receptivity in female rats, suggesting that CB1 antagonists might serve as a new treatment option for women with abnormally low libido (93). The initial excitement over rimonabant's efficacy at reducing weight was greatly tempered in 2007 by the growing concern about the safety of CB1 receptor antagonists because of increased rates of depression, anxiety, and suicidal ideations related to their use. These concerns led the European Medicines Agency in 2008 to recommend suspension of marketing of rimonabant in the European Union, where it had been available by prescription since 2006. Rimonabant was eventually voluntarily withdrawn from the European Union by the sponsor because clinical use was not following criteria established to maximize its benefit/risk ratio, i.e., community clinicians were prescribing it to patients at high risk for depression. Studies using preclinical models of anxiety or depression have yielded equivocal results. For example, depending on the study, acute CB1 antagonist treatment produces either anxiogenic-like (94) or anxiolytic-like effects (95,96,97), and these discrepant effects do not differentiate along the species or model used. CB1 receptor inverse agonism produces no effect on depressant-like behavior in rats (98,99,100), but an antidepressant-like profile in mice (101). Recently, Beyer et al. (102) reported that chronic dosing with rimonabant produced depressive-like effects in rats in the forced swim test paradigm. This study went a step further and investigated possible mechanisms responsible for production of psychiatric adverse effects seen with chronic rimonabant treatment. They found, for example, that chronically treated rats had decreased levels of serotonin in frontal cortex and marked reductions in hippocampal cell proliferation and in survival. A model for using combination therapy for obesity is provided from the treatment of hypertension, which advanced from single agents to combinations of drugs that impacted different physiologic mechanisms of blood pressure control. Combination agents appear to be a logical approach to management of obesity in order to improve efficacy while decreasing compliance- and safety-limiting adverse effects. Four drug combinations of currently or formerly (sibutramine was removed from the market October 2010) approved drugs have recently shown efficacy for reducing body weight, including orlistat and sibutramine, orlistat and metformin, sibutramine and pramlintide, and phentermine with pramlintide, and several others are at various phases of clinical testing (see ref. (103) for review). The CB1 receptor system has been demonstrated to interact with several other neurotransmitter/neuromodulator systems, including mu-opioid, vanilloid TRP (TRPV1), glutamate, and serotonin (5-HT). Most noteworthy, the 5-HT receptor system plays an established role in appetite regulation. Both serotonin selective and serotonin/norepinephrine selective reuptake inhibitors induce hypophagia, and the serotonin/norepinephrine selective reuptake inhibitor sibutramine is one of only two currently US Food and Drug Administration-approved drugs for long-term treatment of overweight and obesity (although sibutramine recently underwent suspension of its marketing authorization in the European Union due to adverse cardiovascular effects). The serotonin selective reuptake inhibitors, though not approved for long-term treatment of overweight, are prescribed for this purpose, especially in patients comorbid for depression (104). Specifically, the 5-HT2C receptor is the most widely implicated receptor subtype in the regulation of appetite and weight loss (for review see ref. (105)). The 5-HT indirect agonist dexfenfluramine, the most successful marketed antiobesity drug used for several years to treat obesity, produces significant, sustained decrease in appetite and body weight that have been attributed to its high affinity for 5-HT2C receptors (106). Newer, direct acting 5-HT2C receptor agonists are currently in various phases of clinical trials. The coadministration of a CB1 antagonist with a 5-HT agonist would therefore seem to offer potential as an antiobesity pharmacotherapy, possibly lowering the dose of each required to be effective and potentially improving the therapeutic window. To test this hypothesis preclinically, we administered rimonabant or the 5-HT2C receptor agonist m-chlorophenylpiperazine alone and in combination to mice. When administered alone, both agents decreased motivation for vanilla-flavored Ensure (as measured by operant responding under a progressive ratio schedule of reinforcement). When administered in combination, rimonabant + m-chlorophenylpiperazine produced a synergistic decrease in motivation for the palatable food reinforcer (107). This observation is supported by other findings that pharmacological manipulation of CB1 receptors can modulate serotonin efflux selectively within the mesocorticolimbic system (108) compared to hypothalamus (109). The seeming benefit of using 5-HT compounds for overweight and obesity is their concomitant antidepressant efficacy. The 5-HT receptor system plays a well-established role in emotionality, and during the last 20 years, serotonin selective reuptake inhibitors have progressively become the most commonly prescribed antidepressants. Interestingly, antidepressant effects of CB1 agonists in rats have been linked to interactions with the 5-HT system (110,111). Furthermore, as previously mentioned, chronic rimonabant administration associated with a depressive phenotype in rats was associated with decreased levels of serotonin in frontal cortex as well as reduced hippocampal neurogenesis (103). It is important to note here that reversal of reduced hippocampal neurogenesis is believed to contribute to the therapeutic effects of serotonin-acting antidepressants. Most convincingly, Takahashi et al. (112) demonstrated that subthreshold doses of rimonabant and serotonin selective reuptake inhibitors, which separately had no effect on depressant behavior in mice, showed a clear antidepressant effect when coadministered. Given the difficulties and expense of developing drugs with new mechanisms of action, it is not surprising that the failure of individual compounds prompt questioning of the overall mechanistic approach. This has been the case for functional antagonism of the CB1 receptor. However, several new developments suggest that abandonment of the CB1 approach might be premature, including advances in endocannabinoid (especially peripheral) pharmacology, design and synthesis of novel compounds with chemical diversity, and preclinical results using drug combinations from diverse classes. Based on the current state of knowledge pertaining to the behavioral effects of second generation CB1 antagonists, considerable promise is especially seen for peripherally restricted CB1 antagonists, given the recent preclinical effects of AM6545 (54,55,56) and the preclinical and clinical effects of TM38837 (51,52). This work was supported by National Institutes of Health grants F32-DA01931 (principal investigator (PI): S.J.W) and R01-DA014673 (PI: Ellen A. Walker, PhD, Department of Pharmaceutical Sciences, Temple University School of Pharmacy, Philadelphia, PA). The authors declared no conflict of interest.