Molecular Psychiatry (2013) 18, 916 -- 921 & 2013 Macmillan Publishers Limited All rights reserved 1359-4184/13 www.nature.com/mp

ORIGINAL ARTICLE Reduced cannabinoid CB1 receptor binding in dependence measured with positron emission tomography

J Hirvonen1, P Zanotti-Fregonara1,3, JC Umhau2,3, DT George2, D Rallis-Frutos1, CH Lyoo1, C-T Li1, CS Hines1, H Sun2, GE Terry1, C Morse1, SS Zoghbi1, VW Pike1, RB Innis1,3 and M Heilig2,3

Brain cannabinoid CB1 receptors contribute to alcohol-related behaviors in experimental animals, but their potential role in humans with alcohol dependence is poorly understood. We measured CB1 receptors in alcohol dependent patients in early and protracted abstinence, and in comparison with control subjects without alcohol use disorders, using positron emission 18 tomography and [ F]FMPEP-d2, a radioligand for CB1 receptors. We scanned 18 male in-patients with alcohol dependence twice, within 3--7 days of admission from ongoing drinking, and after 2--4 weeks of supervised abstinence. Imaging data were compared with those from 19 age-matched healthy male control subjects. Data were also analyzed for potential influence of a common functional variation (rs2023239) in the CB1 receptor gene (CNR1) that may moderate CB1 receptor density. On the first scan, CB1 receptor binding was 20--30% lower in patients with alcohol dependence than in control subjects in all brain regions and was negatively correlated with years of alcohol abuse. After 2--4 weeks of abstinence, CB1 receptor binding remained similarly reduced in these patients. Irrespective of the diagnostic status, C allele carriers at rs2023239 had higher CB1 receptor binding compared with non-carriers. Alcohol dependence is associated with a widespread reduction of cannabinoid CB1 receptor binding in the human brain and this reduction persists at least 2--4 weeks into abstinence. The correlation of reduced binding with years of alcohol abuse suggests an involvement of CB1 receptors in alcohol dependence in humans.

Molecular Psychiatry (2013) 18, 916--921; doi:10.1038/mp.2012.100; published online 10 July 2012 Keywords: alcohol dependence; cannabinoid CB1 receptor; endocannabinoid; imaging; positron emission tomography

INTRODUCTION CB1 receptors (CNR1) tags a haplotype associated with substance 19 Animal studies suggest that the brain cannabinoid system use disorders including alcohol dependence. A recent study of contributes to alcohol-related behaviors.1--3 This system consists patients with alcohol dependence linked the rs2023239 C allele of endogenous cannabinoids, such as anandamide and 2- with greater subjective reward from alcohol, greater midbrain and arachidonoylglycerol, enzymes responsible for their synthesis prefrontal cortex activation in response to alcohol cues, and and catabolism and cannabinoid receptors of the CB and CB higher density of CB1 receptors in postmortem samples of 1 2 20 subtypes. The CB1 receptor is widely distributed in the human prefrontal cortex. Whether this SNP also moderates CB1 receptor brain, with highest densities in the basal ganglia, hippocampus, density in vivo is unknown. 4,5 cingulate cortex and molecular layer of cerebellum. Most CB1 Here, we evaluated CB1 receptor binding in patients with alcohol receptors are pre-synaptic, and inhibit release of other neuro- dependence and controls without alcohol use disorders. We used transmitters, such as g-amino butyric acid and glutamate.6,7 positron emission tomography (PET) and a recently developed 18 21,22 Animal studies suggest that reinforcing properties of alcohol are radioligand for CB1 receptors, [ F]FMPEP-d2, 21 in part mediated through the endocannabinoid system. Stimula- with high affinity and selectivity in the human brain. Patients were tion of CB1 receptors increases alcohol intake across a variety of scanned twice: within 3--7 days of admission from ongoing alcohol rodent models,8 whereas the opposite is observed with genetic9--11 use, and after 2--4 weeks of abstinence on a monitored unit. Based 8,12--14 8 or pharmacological CB1 receptor blockade. Chronic alcohol on animal experiments, we hypothesized that CB1 receptor exposure increases the concentration of endogenous cannabinoids binding would be decreased in patients with alcohol dependence in most brain regions3 and decreases the density of CB immediately after ongoing alcohol use. Based on human post- 1 20 receptors,3,15--17 a change that is reversible upon abstinence.16,17 mortem data, we predicted that the C allele of rs2023239 would Accordingly, a postmortem study recently found decreased CB1 be associated with higher CB1 receptor binding. Finally, we asked receptor density in the ventral striatum of patients with alcohol whether a hypothesized decrease in CB1 receptor binding in 18 dependence. Whether cannabinoid CB1 receptors are decreased patients with alcohol dependence recovers after abstinence. in vivo in human subjects with alcohol dependence is unknown. Functional genetic variation that moderates endocannabinoid signaling in response to alcohol might moderate susceptibility for SUBJECTS AND METHODS excessive alcohol consumption. Accordingly, a common single- The NIH CNS Institutional Review Board approved the protocol and the nucleotide polymorphism (SNP), rs2023239, in the gene encoding consent forms. Written informed consent was obtained from all subjects.

1Molecular Imaging Branch, National Institute of Mental Health, NIH, Bethesda, MD, USA and 2Laboratory of Clinical and Translational Studies, National Institute on Alcohol Abuse and , NIH, Bethesda, MD, USA. Correspondence: Dr M Heilig, Laboratory of Clinical and Translational Studies, National Institute on Alcohol Abuse and Alcoholism, Building 10, Room 1E-5334, 10 Center Drive, Bethesda, MD 20892-1108, USA. E-mail: [email protected] or [email protected] 3These authors contributed equally to this work. Received 26 March 2012; revised 6 June 2012; accepted 12 June 2012; published online 10 July 2012 CB1 receptor downregulation in alcohol dependence J Hirvonen et al 917 Subjects Table 1. Demographic, clinical and radiochemical information of the Detailed eligibility criteria can be found at clinicaltrials.gov. (NCT00816439). study participants Patients (N ¼ 18) were males with alcohol dependence according to 23 24 DSM-IV, established by structured diagnostic interviews (SCID), and Alcohol Control P-value alcohol consumption within the last month (Table 1). Subjects did not have other substance use disorders except nicotine dependence, as determined Number of subjects 18 19 by the SCID. Urine samples on admission were negative for cannabinoids, Age (years) 44±10 40±8 0.251 À2 opiates, amphetamines, cocaine metabolites and . Body mass index (kg m )27±428±5 0.738 Patients were admitted to an in-patient unit at the NIH Clinical Center, Tobacco smokers/non-smokers (N) 11/7 2/17 0.001 Amount of alcohol use (drinks per 18±6 1 0.001 and assessed for severity of alcohol dependence (Alcohol Dependence o o 25 26 day, TLFB) Scale ), alcohol use in the previous 90 days (Time-Line Follow Back ), ± 27 Alcohol Dependence Severity 20 6 o1 o0.001 alcohol craving (Penn Alcohol Craving Scale ) and symptoms of (ADS) 28 depression and anxiety (Comprehensive Psychopathology Rating Scale ). Baseline craving (PACS) 12±81±2 o0.001 Withdrawal intensity was assessed with the Clinical Institute Withdrawal Baseline peak withdrawal (CIWA) 9±5 N.A. N.A. 29 18 Assessment Scale. A total of 10 (56%) patients received to Injected activity of [ F]FMPEP-d2 187±10 181±12 0.103 relieve withdrawal symptoms. The cumulative oxazepam dose (mean: (MBq) 18 ± ± 273 mg, range: 30--540 mg) was used to control for possible medication Injected amount of [ F]FMPEP-d2 1.7 0.9 1.9 0.8 0.547 effects. In all, 13 of the patients (72%) smoked cigarettes. Throughout the (nmol) Fraction of free [18F]FMPEP-d in 0.38±0.1 0.39±0.2 0.886 study, all subjects participated in a standard behavioral in-patient alcohol 2 plasma (%) rehabilitation program, but did not receive any prescription medications other than benzodiazepines as described above and vitamin B1 (thiamine) Abbreviations: ADS, Alcohol Dependence Scale; CIWA, Clinical Institute supplementation according to clinical guidelines. Withdrawal Assessment Scale; N.A., not applicable; PACS, Penn Alcohol Healthy male controls (N ¼ 19) were free of somatic and psychiatric Craving Scale; TLFB, Time-Line Follow Back. ± illness as confirmed by history, physical examination, structured diagnostic Values are number or mean s.d. interviews (SCID, full version) electrocardiogram and blood and urine tests. Control subjects did not have current or lifetime history of alcohol use (Shapiro--Wilk test). Therefore, results were confirmed with non-parametric or other substance use disorders, and had urine samples negative for Mann--Whitney U tests. Variance was homogeneous across groups cannabinoids, opiates, amphetamines, cocaine metabolites and benzodia- (Levene’s test). To test whether CB receptors were decreased in patients zepines. Control subjects completed assessments of alcohol use and severity 1 with alcohol dependence at baseline, we first applied a mixed model two- of alcohol dependence as described for the patients. None of the controls way analysis of variance, with group status (alcohol dependence vs control) were total abstainers, but 7 out of 19 (37%) reported no alcohol use in the as between-subject factor and brain region as within-subject factor. Body past 90 days. Two control subjects (11%) smoked cigarettes (Table 1). 30 mass index entered the model as a covariate as it affects VT. To assess the contribution of the CNR1 rs2023239 SNP, this factor was then entered into Genotyping the model. Correlations with clinical variables were assessed with Pearson’s 20 Genotyping for CNR1 rs202039 was done largely as described previously correlation coefficient. To test whether CB1 receptors increased after and in detail in Supplementary Methods online. abstinence, we applied a two-way analysis of variance with time point (early vs protracted abstinence) and brain region as within-subject factors. Potential effects of genotype were also evaluated with time  genotype Positron emission tomography and measurement of parent radioligand in arterial plasma interaction. P-values o0.05 were considered statistically significant. 18 22 [ F]FMPEP-d2 was prepared as described previously and in detail in our Investigational New Drug Application 105 198, submitted to the US Food RESULTS and Drug Administration (available at http://pdsp.med.unc.edu/snidd/). 18 The radioligand was obtained in high radiochemical purity (499%) and Overall, VT of [ F]FMPEP-d2 was lower in patients with alcohol had a specific radioactivity of 110±43 MBq nmolÀ1 at time of injection. dependence than in control subjects (main effect of group: F(1, 18 34) ¼ 10.1, P ¼ 0.003). A decrease was seen in all brain regions After intravenous injection of [ F]FMPEP-d2 (Table 1), images were acquired for 120 min using an Advance camera (GE Healthcare, Milwaukee, examined, but its magnitude varied significantly between regions WI, USA) as described previously.21 Arterial blood samples were drawn as (group  region interaction: F(3, 103) ¼ 4.21, P ¼ 0.007; Figure 1); previously described.30 Plasma time-activity curve was corrected for the decrease in body mass index-adjusted VT ranged from 21% in fraction of unchanged radioligand by radio-high-performance liquid hippocampus to 35% in white matter. This finding was confirmed chromatography separation,31 and the plasma-free fraction was measured by an independent statistical parametric mapping analysis of using ultrafiltration.32 voxel-wise VT values (Figure 2) and a non-parametric Mann-- PET images were analyzed by applying a template of volumes of Whitney U test (all Po0.019). The main effect of diagnostic status interest33 as implemented in PMOD, version 3.0 (PMOD Technologies, was also significant when analyzed without correcting for body Zurich, Switzerland),34 in the standard stereotactic space35 as previously mass index (F(1, 35) ¼ 9.36, P ¼ 0.004). Among patients with 30 alcohol dependence, years of alcohol use correlated negatively described. Distribution volume (VT) was estimated according to the two- 2 36 with VT in midbrain (R ¼ 0.28, P ¼ 0.023), posterior cingulate tissue compartmental model with concentration of parent radioligand in 2 21 cortex (R ¼ 0.37, P ¼ 0.008; Supplementary Figure 1) and putamen plasma as input function, using PMOD, as described previously. To assess 2 the potential confound of brain atrophy, PET images were corrected for (R ¼ 0.26, P ¼ 0.030); patients who had been drinking longer had partial volume effects using the Rousset method implemented in the lower VT than those who had been drinking for a shorter period. 37 This analysis was not confounded by age, as age did not correlate PVElab software. Statistical parametric mapping of VT values at voxel 30 level was done using SPM8 as described previously.30 with VT in control subjects. Whole-brain VT did not correlate with alcohol dependence severity (Alcohol Dependence Severity) (R2 ¼ 0.02, P ¼ 0.56), measures of craving (R2 ¼ 0.08, P ¼ 0.26), 2 Statistical analysis of VT data average drinks per day (R ¼ 0.01, P ¼ 0.97), peak withdrawal score Data were analyzed using the SPSS Statistics 17.0. for Windows (Chicago, IL, (R2 ¼ 0.16, P ¼ 0.10), dose received (R2 ¼ 0.05, 2 USA; Release 17.0.0, copyright SPSS, 1993--2007). VT values were normally P ¼ 0.38) or ratings of depression (R ¼ 0.01, P ¼ 0.92) or anxiety distributed except for 4 out of 18 brain regions in the patient group (R2 ¼ 0.01, P ¼ 0.90).

& 2013 Macmillan Publishers Limited Molecular Psychiatry (2013), 916 -- 921 CB1 receptor downregulation in alcohol dependence J Hirvonen et al 918

18 Figure 1. At baseline (within 3--7 days of ongoing alcohol use), VT of [ F]FMPEP-d2 is lower in patients with alcohol dependence (black bars, n ¼ 18) than in control subjects (gray bars, n ¼ 19). Values are estimated marginal means from the repeated measures analysis variance and adjusted to an average body mass index of 27.3 kg mÀ2. Error bars are s.e. of the mean. ACC, anterior cingulate cortex; AMY, amygdala; CAU, caudate nucleus; CER, cerebellum; HIPP, hippocampus; INS, insula; MIDBR, midbrain; OCC, occipital cortex; PAR, parietal cortex; PCC, posterior cingulate cortex; PFC, prefrontal cortex; PHIPP, parahippocampal gyrus; PUT, putamen; TEMP, lateral temporal cortex; THA, thalamus; VST, ventral striatum; WM, white matter. *Po0.05; **Po0.005, two-tailed t-test.

Figure 2. Statistical parametric mapping analysis shows lower VT in patients with alcohol dependence (n ¼ 18) than in control subjects (n ¼ 19) at baseline as a large single cluster. This cluster comprised of 81,026 voxels and had a maximum t value of 3.2 at (0, À34, 60) and a cluster-level family-wise error) corrected P-value of 0.038. Color bar represents t value in each voxel within the significant cluster.

Among control subjects and patients, 28 and 39% carried the rs2023239 C allele, respectively (w2 ¼ 0.50; P ¼ 0.48). When we included genotype as a factor in the analysis, the main effect of group remained (F(1, 31) ¼ 7.72, P ¼ 0.009), and a main genotype effect was found (F(1, 31) ¼ 4.44, P ¼ 0.043). There was no significant group genotype (F(1, 31) 1.47, P 0.24) or group- Figure 3. Variation in rs2023239 in the CNR1 gene regulates whole-  ¼ ¼ brain V of [18F]FMPEP-d : carriers of the C allele (black bars) have  genotype  region (F(3, 96) ¼ 2.00, P ¼ 0.12) interaction, sug- T 2 higher VT than non-carriers (gray bars), irrespective of the diagnostic gesting that genotype affected CB1 binding similarly in both status. groups. Carriers of the C allele had 31% higher VT than non-carriers among patients with alcohol dependence and 19% among control subjects (Figure 3). There were no statistically significant differences between carriers and non-carriers of the C allele in Finally, to determine whether decreased VT in patients with any of the demographic, or in the clinical variables among alcohol dependence is reversible, we repeated PET measurements patients with alcohol dependence. after 19±4 days of abstinence (range 14--28 days). After More patients with alcohol dependence (61%) than control protracted abstinence, VT did not significantly change in any brain subjects (11%) smoked cigarettes. However, smoking had no region (main effect of time: F(1, 17) ¼ 1.18, P ¼ 0.29; time  region effect on VT among patients with alcohol dependence (main interaction: F(3, 49) ¼ 0.71, P ¼ 0.55), and thus, remained signifi- effect: F(1, 33) ¼ 0.01, P ¼ 0.93; smoking  region interaction: F(3, cantly decreased (Figure 4). The average value of whole-brain VT 99) ¼ 0.34, P ¼ 0.80). Furthermore, lower VT in patients with alcohol increased by only þ 5%. Change after abstinence remained not dependence than in control subjects remained significant even significant after controlling for rs2023239 genotype (main effect of after smoking was included in the statistical model (main effect of time: F(1, 16) ¼ 0.83, P ¼ 0.38; time  genotype interaction: F(1, group: F(1, 33) ¼ 7.32, P ¼ 0.011). Therefore, cigarette smoking was 16) ¼ 0.34, P ¼ 0.57). Change in whole-brain VT correlated posi- unlikely to significantly confound the main finding. tively with baseline Alcohol Dependence Score (R2 ¼ 0.27, To control for a potential confound of brain atrophy in the P ¼ 0.027; Supplementary Figure 2), but not with other clinical patient group, we corrected the PET data for partial volume variables at baseline, with change in clinical variables after effects. After correction, group difference in VT remained abstinence, or with interval between two PET scans. Fraction of significant (main effect of group: F(1, 34) ¼ 8.42, P ¼ 0.006; free radioligand in arterial plasma did not change after abstinence group  region interaction: F(3, 99) ¼ 3.79, P ¼ 0.011). (P ¼ 0.75).

Molecular Psychiatry (2013), 916 -- 921 & 2013 Macmillan Publishers Limited CB1 receptor downregulation in alcohol dependence J Hirvonen et al 919

18 Figure 4. VT of [ F]FMPEP-d2 did not change after abstinence in patients with alcohol dependence. Error bars are s.d. of percent change. ACC, anterior cingulate cortex; AMY, amygdala; CAU, caudate nucleus; CER, cerebellum; HIPP, hippocampus; INS, insula; MIDBR, midbrain; OCC, occipital cortex; PAR, parietal cortex; PCC, posterior cingulate cortex; PFC, prefrontal cortex; PHIPP, parahippocampal gyrus; PUT, putamen; TEMP, lateral temporal cortex; THA, thalamus; VST, ventral striatum; WM, white matter.

DISCUSSION correlation between withdrawal intensity, benzodiazepine dose 18 When evaluated within 1 week of admission from ongoing alcohol and VT of [ F]FMPEP-d2, as well as the persistence of the findings 2--4 weeks after admission makes a confound by these factors use, patients with alcohol dependence had 20--30% lower VT of 18 unlikely. [ F]FMPEP-d2 in all brain regions compared with control subjects Finally, we considered whether differential displacement of the without substance use or other psychiatric disorders, consistent 18 with a wide-spread downregulation of CB receptors. This down- radioligand [ F]FMPEP-d2 by endogenous cannabinoids in 1 patients compared with control subjects could confound our regulation was likely related to long-term alcohol exposure, 30 because it was more pronounced in patients who had used alcohol results. As some PET radioligands can be displaced by endogenous neurotransmitters, we have previously tested the longer than in those who had done so for a shorter time, while no effects of cannabinoid agonists on the binding of [11C]MePPEP, a effect of age per se was observed in controls. V of [18F]FMPEP-d T 2 closely related analog, in rat brain.38 In that study, we used high remained similarly lowered in the alcohol dependent patients after doses of cannabinoid agonists (anandamide, methanandamide or 2--4 weeks of abstinence. This lack of reversibility is in contrast with CP 55,940) or an inhibitor of fatty acid amide hydrolase, the main the reversible downregulation of cortical CB1 receptors recently 30 enzyme that breaks down endocannabinoids. Although rimona- found in subjects who chronically abuse cannabis, and suggests a 11 bant, an inverse CB1 agonist, potently displaced [ C]MePPEP, more persistent neuroadaptive change in CB1 receptors in alcohol none of the direct agonists did so. Importantly, potentiation of dependence. We also demonstrated a significant effect of genetic endogenous cannabinoid transmission by fatty acid amide variation at rs2023239,aCNR1 SNP previously associated with risk hydrolase was also without effect. These findings suggests that 19 11 18 for substance use disorders including alcohol dependence, with [ C]MePPEP, a close structural analog of [ F]FMPEP-d2 used in brain responses to alcohol associated cues in functional imaging, this study, is not displaced by direct CB1 agonists, including the 20 38 and with cortical CB1 receptor density on postmortem analysis. endogenous ligand anandamide. Based on these observations, Consistent with the postmortem binding data, carriers of the C allele we believe it unlikely that differential displacement of our PET 18 in our study had higher VT of [ F]FMPEP-d2 than non-carriers in ligand by endocannabinoids accounts for the differences ob- vivo, providing additional support for the validity of our data. served between patients and controls in this study. Nevertheless, 18 Decreased VT of [ F]FMPEP-d2 in patients with alcohol additional studies would be required to directly determine 18 dependence was unlikely caused by methodological confounds. whether [ F]FMPEP-d2 can be displaced by endogenous (and First, potential alterations in peripheral distribution or metabolism exogenous) CB1 receptor agonists in humans. of the radioligand in patients with alcohol dependence did not Animal studies utilizing genetic inactivation as well as affect our results, because the outcome measure VT corrects for pharmacological blockade of CB1 receptors have consistently these effects. Second, only free radioligand (not bound to plasma shown that reinforcement of alcohol consumption is in part proteins) is able to enter the brain, while the fraction of free mediated by endocannabinoids acting at this receptor.3,8 Accord- radioligand was similar between the groups (Table 1). Third, about ingly, chronic alcohol exposure has been shown to result in 18 3 90% of [ F]FMPEP-d2 VT in monkey brain represents specific elevated tissue levels of endocannabinoids, while more recent 21 binding to CB1 receptors; therefore, changes in nonspecific microdialysis experiments have directly shown that self-adminis- binding (for example, to brain lipids) would be unlikely to tration of alcohol is associated with increased central levels of confound the main finding. Fourth, brain atrophy in patients with extracellular endocannabinoids, and that this increase is corre- alcohol dependence may have ‘diluted’ the PET signal via partial lated with the amount of alcohol consumed.39 Consistent with volume effects. However, the group differences remained similar increased endocannabinoid activity resulting from alcohol con- even after correction for partial volume effect based on structural sumption, most animal studies have found downregulation of CB1 images. Fifth, cigarette smoking was more prevalent among receptors in several brain regions after chronic alcohol expo- patients with alcohol dependence than among control subjects. sure15--17 (although see Gonzalez et al.40 for a negative finding). Yet, the main finding persisted after controlling for smoking One postmortem human study also found decreased CB1 receptor 18 in the statistical model, and VT was not different between smoking protein in the ventral striatum of alcoholics. Our present finding and nonsmoking patients. This is in agreement with our recent that long-term alcohol exposure in alcohol dependence is 18 observation that cigarette smoking did not influence [ F]FMPEP- associated with downregulation of CB1 receptors in vivo thus 30 d2 VT in people who chronically abuse cannabis. Sixth, a lack of represents a direct translation of the animal data. Our findings

& 2013 Macmillan Publishers Limited Molecular Psychiatry (2013), 916 -- 921 CB1 receptor downregulation in alcohol dependence J Hirvonen et al 920 diverge from the animal studies in that those found evidence analysis. Our findings suggest that CB1 receptor has a different of reversibility after 116 and 40 days.17 A likely reason for role in early vs late phases of alcohol dependence. A potential this divergence is the difference in duration of alcohol exposure, implication of our findings is that enhanced, rather than blocked which was 3--60 days in the animal studies and 5--50 years in our CB1 signaling may be beneficial in late stage, treatment seeking patients. alcohol dependent patients. Our finding of downregulated CB1 receptors in alcohol dependence is consistent with a conceptualization of this condition as an allostatic process.41,42 According to this frame- CONFLICT OF INTEREST work, alcohol is initially consumed for its pleasurable, positively The authors declare no conflict of interest. reinforcing properties, but neuroadaptions to continued heavy alcohol use gradually reduce these alcohol actions. Because of the established role of CB1 receptors in positive reinforcement from ACKNOWLEDGEMENTS alcohol, their downregulation with prolonged heavy alcohol use is We thank Kimberly Jenko, Kacey Anderson and David Clark for assisting in the a plausible candidate mechanism that might contribute to this measurements of radioligand in plasma; Maria D Ferraris Araneta, Gerald Hodges, process. Although the positively reinforcing effects of alcohol are William C Kreisl and Barbara Scepura as well as Chris Geyer and the NIAAA nursing gradually lost with time, brain stress systems become overactive staff for subject recruitment and care; the NIH PET Department for imaging; and the and alcohol use becomes pursued in large part to ameliorate PMOD Technologies for providing its image analysis and modeling software. This negative emotionality and promote stress coping.42 Animal research was supported by the Intramural Programs of NIMH and NIAAA. Jussi Hirvonen was supported by personal grants from The Academy of Finland; The studies have shown that CB1 receptors have a critical role in modulating stress responses through actions in basolateral Finnish Cultural Foundation; The Finnish Foundation for Alcohol Studies; The Finnish 43,44 Medical Foundation; The Instrumentarium Foundation; The Jalmari and Rauha amygdala and hippocampus. Downregulation of CB1 receptors in these structures in later stages of alcohol dependence can Ahokas Foundation; The Paulo Foundation; The Research Foundation of Orion Corporation; and The Yrjo¨ Jahnsson Foundation. therefore be expected to contribute to enhanced negative emotionality and impaired stress coping. This in turn would provide an incentive for resumption of alcohol intake to restore REFERENCES endocannabinoid signaling and its ability to promote stress 1 Maccioni P, Colombo G, Carai MA. Blockade of the cannabinoid CB1 receptor and coping to normal levels. In this context, we note that the CB1 receptor inverse agonist rimonabant was not effective when alcohol dependence: preclinical evidence and preliminary clinical data. CNS 45,46 Neurol Disord Drug Targets 2010; 9: 55--59. evaluated in chronic alcohol dependence. Several factors may 2 Mechoulam R, Parker L. Cannabis and alcohol -- a close friendship. Trends Pharma- have contributed to these negative results, but one possibility is col Sci 2003; 24: 266--268. that a blockade by rimonabant of positively reinforcing properties 3 Basavarajappa BS. The endocannabinoid signaling system: a potential target of alcohol was in part offset by an increased incentive for for next-generation therapeutics for alcoholism. Mini Rev Med Chem 2007; 7: consumption to restore affective homeostasis. If this interpretation 769--779. is correct, augmenting rather than inhibiting endogenous 4 Herkenham M, Lynn AB, Little MD, Johnson MR, Melvin LS, de Costa BR et al. cannabinoid function (for example, by inhibiting their metabo- Cannabinoid Receptor Localization in Brain. Proc Natl Acad Sci 1990; 87: 1932--1936. lism47) might be of interest for treatment of alcohol dependence. 5 Glass M, Dragunow M, Faull RL. Cannabinoid receptors in the human brain: a First, augmented endocannabinoid function might restore detailed anatomical and quantitative autoradiographic study in the fetal, neonatal and adult human brain. Neuroscience 1997; 77: 299--318. affective homeostasis in the absence of alcohol, thereby reducing 6 Howlett AC, Barth F, Bonner TI, Cabral G, Casellas P, Devane WA et al. International or eliminating the incentive to consume alcohol for its negatively Union of Pharmacology. XXVII. Classification of cannabinoid receptors. Pharmacol reinforcing properties. Second, and more generally, CB1 receptor Rev 2002; 54: 161--202. stimulation reduces glutamate release, and is therefore neuropro- 7 Wilson RI, Nicoll RA. Endocannabinoid signaling in the brain. Science 2002; 296: tective.7 Chronic alcohol dependence is characterized by a 678--682. 48 8 Colombo G, Serra S, Vacca G, Carai M, Gessa G. Endocannabinoid system and hyperglutamatergic state and CB1 receptor stimulation protects against glutamatergic excitotoxic cell death during alcohol alcohol addiction: pharmacological studies. Pharmacol Biochem Behav 2005; 81: withdrawal.49 369--380. In a direct in vivo correlate of postmortem ligand-binding data,20 9 Hungund BL, Szakall I, Adam A, Basavarajappa BS, Vadasz C. Cannabinoid CB1 receptor knockout mice exhibit markedly reduced voluntary alcohol consumption we found higher CB1 ligand binding in carriers of the C allele at and lack alcohol-induced dopamine release in the nucleus accumbens. the rs2023239 CNR1 SNP, irrespective of diagnostic group. Genetic J Neurochem 2003; 84: 698--704. association data suggest a small but significant contribution of 10 Wang L, Liu J, Harvey-White J, Zimmer A, Kunos G. Endocannabinoid signaling via this allele to the risk for substance use disorders, including alcohol cannabinoid receptor 1 is involved in ethanol preference and its age-dependent dependence.19 This risk increase may be mechanistically ac- decline in mice. Proc Natl Acad Sci 2003; 100: 1393--1398. counted for by the observations that carriers of the C allele exhibit 11 Racz I, Bilkei-Gorzo A, Toth ZE, Michel K, Palkovits M, Zimmer A. A critical role for greater brain responses to alcohol associated cues and greater the cannabinoid CB1 receptors in alcohol dependence and stress-stimulated ethanol drinking. J Neurosci 2003; 23: 2453--2458. alcohol-induced reward, presumably in part via higher CB1 receptor density.20 As indicated above, rewarding or positively 12 Rodriguez de Fonseca F, Roberts AJ, Bilbao A, Koob GF, Navarro M. Cannabinoid receptor antagonist SR141716A decreases operant ethanol self administration in reinforcing properties of alcohol are likely most important in early rats exposed to ethanol-vapor chambers. Zhongguo Yao Li Xue Bao 1999; 20: stages of alcohol dependence, and enhanced reward from alcohol 1109--1114. due to a gain-of-function mutation at the CNR1 locus would be 13 Arnone M, Maruani J, Chaperon F, Thiebot MH, Poncelet M, Soubrie P et al. expected to increase alcoholism risk. In later stages of alcohol Selective inhibition of sucrose and ethanol intake by SR 141716, an antagonist of dependence, alcohol reward becomes less important, stress central cannabinoid (CB1) receptors. Psychopharmacology 1997; 132: 104--106. 14 Colombo G, Agabio R, Fa M, Guano L, Lobina C, Loche A et al. Reduction of systems are activated and restored CB1 receptor function may be beneficial by counteracting negative emotionality and voluntary ethanol intake in ethanol-preferring sP rats by the cannabinoid glutamate excitotoxicity. antagonist SR-141716. Alcohol Alcohol 1998; 33: 126--130. In conclusion, we found that CB receptor binding is decreased 15 Basavarajappa BS, Cooper TB, Hungund BL. Chronic ethanol administration down- 1 regulates cannabinoid receptors in mouse brain synaptic plasma membrane. in patients with alcohol dependence and that this downregulation Brain Res 1998; 793: 212--218. persists several weeks into abstinence. We also found that the C 16 Vinod K, Yalamanchili R, Xie S, Cooper T, Hungund B. Effect of chronic ethanol allele of the rs2023239 locus is associated with higher CB1 receptor exposure and its withdrawal on the endocannabinoid system. Neurochem Int binding in vivo, similar to what has been found on postmortem 2006; 49: 619--625.

Molecular Psychiatry (2013), 916 -- 921 & 2013 Macmillan Publishers Limited CB1 receptor downregulation in alcohol dependence J Hirvonen et al 921 17 Mitrirattanakul S, Lo´pez-Valde´s HE, Liang J, Matsuka Y, Mackie K, Faull KF et al. anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002; 15: Bidirectional alterations of hippocampal cannabinoid 1 receptors and their 273--289. endogenous ligands in a rat model of alcohol withdrawal and dependence. 34 Burger C, Mikolajczyk K, Grodzki M, Rudnicki P, Szabatin M, Buck A. Java tools for Alcohol Clin Exp Res 2007; 31: 855--867. quantitative post-processing of brain PET data. J Nucl Med 1998; 39: 277p--278p. 18 Vinod KY, Kassir SA, Hungund BL, Cooper TB, Mann JJ, Arango V. Selective 35 Friston KJ, Holmes AP, Worsley KJ, Poline JP, Frith C, Frackowiak RSJ. Statistical alterations of the CB1 receptors and the fatty acid amide hydrolase in the ventral parametric maps in functional imaging: a general linear approach. Hum Brain striatum of alcoholics and suicides. J Psychiatric Res 2010; 44: 591--597. Mapp 1995; 2: 189--210. 19 Zhang PW, Ishiguro H, Ohtsuki T, Hess J, Carillo F, Walther D et al. Human 36 Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, Gunn RN et al. Consensus cannabinoid receptor 1: 50 exons, candidate regulatory regions, polymorphisms, nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood haplotypes and association with polysubstance abuse. Mol Psychiatry 2004; 9: Flow Metab 2007; 27: 1533--1539. 916--931. 37 Quarantelli M, Berkouk K, Prinster A, Landeau B, Svarer C, Balkay L et al. Integrated 20 Hutchison KE, Haughey H, Niculescu M, Schacht J, Kaiser A, Stitzel J et al. The software for the analysis of brain PET/SPECT studies with partial-volume-effect incentive salience of alcohol: translating the effects of genetic variant in CNR1. correction. J Nucl Med 2004; 45: 192--201. Arch Gen Psychiatry 2008; 65: 841--850. 38 Terry G, Liow J, Chernet E, Zoghbi S, Phebus L, Felder C et al. Positron emission 21 Terry GE, Hirvonen J, Liow JS, Zoghbi SS, Gladding R, Tauscher JT et al. Imaging tomography imaging using an inverse agonist radioligand to assess cannabinoid and quantitation of cannabinoid CB1 receptors in human and monkey brains CB1 receptors in rodents. Neuroimage 2008; 41: 690--698. using (18)F-labeled inverse agonist radioligands. J Nucl Med 2010; 51: 112--120. 39 Caille S, Alvarez-Jaimes L, Polis I, Stouffer DG, Parsons LH. Specific Alterations of 22 Donohue SR, Krushinski JH, Pike VW, Chernet E, Phebus L, Chesterfield AK et al. extracellular endocannabinoid levels in the nucleus accumbens by ethanol, Synthesis, ex vivo evaluation, and radiolabeling of potent 1,5-diphenylpyrrolidin- heroin, and cocaine self-administration. J Neurosci 2007; 27: 3695--3702. 2-one cannabinoid subtype-1 receptor ligands as candidates for in vivo imaging. 40 Gonzalez S, Fernandez-Ruiz J, Sparpaglione V, Parolaro D, Ramos JA. Chronic J Med Chem 2008; 51: 5833--5842. exposure to , cocaine or ethanol in rats produced different effects in 23 Diagnostic and Statistical Manual of Mental Disorders 4th edn., text revision. brain cannabinoid CB(1) receptor binding and mRNA levels. Drug Alcohol Depend American Psychiatric Association: Washington, DC, 2000. 2002; 66: 77--84. 24 First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSM- 41 Koob GF, Le Moal M. Addiction and the brain antireward system. Annu Rev Psychol IV Axis I Disorders, Clinician Version (SCID-CV). American Psychiatric Press: 2008; 59: 29--53. Washington, DC, 1996. 42 Koob GF. A role for brain stress systems in addiction. Neuron 2008; 59: 11--34. 25 Skinner HA, Allen BA. Alcohol dependence syndrome: measurement and 43 Hill MN, Patel S, Campolongo P, Tasker JG, Wotjak CT, Bains JS. Functional validation. J Abnorm Psychol 1982; 91: 199--209. interactions between stress and the endocannabinoid system: from synaptic 26 Sobell LC, Sobell MB. Timeline follow-back: a technique for assessing self-reported signaling to behavioral output. J Neurosci 2010; 30: 14980--14986. ethanol consumption, In: Measuring Alcohol Consumption: Psychosocial and 44 Hill MN, McEwen BS. Involvement of the endocannabinoid system in the Biochemical Methods, Allen, J, Litten, R (eds). 41--72. Humana Press, Totowa, NJ, 1992. neurobehavioural effects of stress and glucocorticoids. Prog Neuropsychopharma- 27 Flannery BA, Volpicelli JR, Pettinati HM. Psychometric properties of the Penn col Biol Psychiatry 2010; 34: 791--797. Alcohol Craving Scale. Alcohol Clin Exp Res 1999; 23: 1289--1295. 45 Soyka M, Koller G, Schmidt P, Lesch O-M, Leweke M, Fehr C et al. Cannabinoid 28 Svanborg P. State and Trait Measures in the Affective Disorders. Karolinska Institute, Receptor 1 Blocker Rimonabant (SR 141716) for Treatment of Alcohol Department of Clinical Neuroscience, Psychiatry Institute: Stockholm, 1999. Dependence. J Clin Psychopharmacol 2008; 28: 317--324. 29 Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM. Assessment of 46 George DT, Herion DW, Jones CL, Phillips MJ, Hersh J, Hill D et al. Rimonabant alcohol withdrawal: the revised clinical institute withdrawal assessment for (SR141716) has no effect on alcohol self-administration or endocrine measures in alcohol scale (CIWA-Ar). Br J Addict 1989; 84: 1353--1357. nontreatment-seeking heavy alcohol drinkers. Psychopharmacology 2009; 208: 30 Hirvonen J, Goodwin RS, Li CT, Terry GE, Zoghbi SS, Morse C et al. Reversible and 37--44. regionally selective downregulation of brain cannabinoid CB(1) receptors in 47 Cippitelli A, Cannella N, Braconi S, Duranti A, Tontini A, Bilbao A et al. Increase of chronic daily cannabis smokers. Mol Psychiatry 2012; 17: 642--649. brain endocannabinoid anandamide levels by FAAH inhibition and alcohol abuse 31 Gandelman MS, Baldwin RM, Zoghbi SS, Zea-Ponce Y, Innis RB. Evaluation of behaviours in the rat. Psychopharmacology 2008; 198: 449--460. ultrafiltration for the free-fraction determination of single photon emission 48 Umhau JC, Momenan R, Schwandt ML, Singley E, Lifshitz M, Doty L et al. Effect of computed tomography (SPECT) radiotracers: beta-CIT, IBF, and . J Pharm acamprosate on magnetic resonance spectroscopy measures of central glutamate Sci 1994; 83: 1014--1019. in detoxified alcohol-dependent individuals: a randomized controlled experi- 32 Zoghbi SS, Shetty HU, Ichise M, Fujita M, Imaizumi M, Liow JS et al. PET imaging of mental medicine study. Arch Gen Psychiatry 2010; 67: 1069--1077. the dopamine transporter with 18F-FECNT: a polar radiometabolite confounds 49 Rubio M, Villain H, Docagne F, Roussel BD, Ramos JA, Vivien D et al. brain radioligand measurements. J Nucl Med 2006; 47: 520--527. Pharmacological activation/inhibition of the cannabinoid system affects alcohol 33 Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N withdrawal-induced neuronal hypersensitivity to excitotoxic insults. PloS one et al. Automated anatomical labeling of activations in SPM using a macroscopic 2011; 6: e23690.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp)

& 2013 Macmillan Publishers Limited Molecular Psychiatry (2013), 916 -- 921