Endocannabinoid receptor 1 gene variations
increase risk for obesity and modulate body
mass index in European populations
Michael Benzinou1,2, Jean-Claude Che `vre1, Kirsten J. Ward1, Ce ´cile Lecoeur1, Christian Dina1,
Stephane Lobbens1, Emmanuelle Durand1, Je ´rome Delplanque1, Fritz F. Horber3,
Barbara Heude4,5, Beverley Balkau4,5, Knut Borch-Johnsen6,7, Torben Jørgensen8,
Torben Hansen6, Oluf Pedersen6,7, David Meyre1and Philippe Froguel1,2,?
1CNRS 8090-Institute of Biology, Pasteur Institute, Lille, France,2Genomic Medicine, Hammersmith Hospital, Imperial
College, Du Cane Road, London W12 0NN, UK,3Klinik Lindberg, Schickstrasse 11, Winterthur CH-8400, Switzerland,
4Universite ´ Paris Sud, France,5INSERM U780-IFR69, Villejuif, France,6Steno Diabetes Center, Niels Steensens
Vej 1, NLC2.13, Gentofte DK-2820, Denmark,7Faculty of Health Science, University of Aarhus, Aarhus,
Denmark and8Research Centre for Prevention and Health, Glostrup University Hospital, Glostrup, Denmark
Received January 4, 2008; Revised and Accepted March 16, 2008
The therapeutic effects of cannabinoid receptor blockade on obesity-associated phenotypes underline the
importance of the endocannabinoid pathway on the energy balance. Using a staged-approach, we examined
the contribution of the endocannabinoid receptor 1 gene (CNR1) on obesity and body mass index (BMI) in the
European population. With the input of CNR1 exons and 30and 50regions sequencing and HapMap database,
we selected and genotyped 26 tagging single-nucleotide polymorphisms (SNPs) in 1932 obese cases
and 1173 non-obese controls of French European origin. Variants that showed significant associations
(P < 0.05) with obesity after correction for multiple testing were further tested in two additional European
cohorts including 2645 individuals. For the identification of the potential causal variant(s), we further geno-
typed SNPs in high linkage disequilibrium (LD) with the obesity-associated variants. Of the 25 successfully
genotyped CNR1 SNPs, 12 showed nominal evidence of association with childhood obesity, class I and II
and/or class III adult obesity (1.16 < OR < 1.40, 0.00003 < P < 0.04). Intronic SNPs rs806381 and rs2023239,
which resisted correction for multiple testing were further associated with higher BMI in both Swiss obese
subjects and Danish individuals. The genotyping of all know variants in partial LD (r2> 0.5) with these two
SNPs in the initial case–control study, identified two better associated SNPs (rs6454674 and rs10485170).
Our study of 5750 subjects shows that CNR1 variations increase the risk for obesity and modulate BMI in
our European population. As CB1 is a drug target for obesity, a pharmacogenetic analysis of the endocanna-
binoid blockade obesity treatment may be of interest to identify best responders.
The endocannabinoid pathway is mediated by the binding of
2-arachidonylglycerol (2-AG) and anandamide (AEA) with
the cannabinoid receptor (CB1) (1). Animal studies have
identified its physiological role in the regulation of energy
metabolism (2–5), as well as in food intake and addictive
behaviours (6–8). These discoveries have led to the
development of CB1 antagonists as potential therapeutic
agents for treating obesity (9–12) and metabolic compli-
cations (12,13). In humans, CB1 antagonists induce weight
loss, reduce food intake (10–13), improve glucose metabolism
(10–13) and modulate lipid levels (10–13). These findings
show the importance of CB1 in human physiology and
qualify CNR1 as a biological candidate for human obesity.
?To whom correspondence should be addressed. Tel: þ44 2083833989; Fax: þ44 2083838577; Email: firstname.lastname@example.org
# The Author 2008. Published by Oxford University Press. All rights reserved.
For Permissions, please email: email@example.com
Human Molecular Genetics, 2008, Vol. 17, No. 13
Advance Access published on March 28, 2008
by guest on May 29, 2013
Previously, a study showed modest evidence of a modulat-
ing effect of the rs12720071 CNR1 variant on body fat mass
and distribution in white European adult men (14). Other
studies have also tested the contribution of the CNR1
rs1049353 single-nucleotide polymorphism (SNP) on obesity
but found inconsistent results (15–17). To analyze whether
CNR1 variations are implicated in human obesity, and in
view of the weak linkage disequilibrium (LD) coverage of
the previous studies, we chose to investigate the possible
role of 26 tagging SNPs, representating 100% of the
common variation in a 37.4 kb region encompassing the
CNR1 locus (http://www.hapmap.org) in 1932 obese cases
and 1173 controls of French origin. We then selected two
SNPs displaying P-values that resisted permutation-based cor-
rection for multiple testing and analyzed them in two indepen-
dent European cohorts. Finally, in an attempt to identify
putative functional SNP(s), we genotyped in the initial
case–control study, all SNPs having a minor allele frequency
.5% andin LD
obesity-associated tagging SNPs.
(r2. 0.5) withthetwo
SNP identification and selection
Over 21 kb, including the four known exons described by
Zhang et al. (18), the exon–intron boundaries, the intronic
SNP-rich regions as well as 5.3 kb upstream and 5.9 kb down-
stream of CNR1, were sequenced in 24 adult controls and 48
obese children (Supplementary Material, Fig. S1). Thirty-six
SNPs with a MAF ?5% were identified and used to establish
an LD map using Haploview. In addition, we ran ‘Tagger’ of
the Haploview program with the HapMap phase II data in a
37.4 kb region defined by LD blocks and encompassing the
CNR1 locus (chr6:88 904 301–88 941 649 on NCBI Build
36.1). Combining the sequencing results to the HapMap
phase II data (http://www.hapmap.org) and for an r2. 0.8,
we selected and genotyped for the case–control association
studies, 26 tagging SNPs with a MAF ?5% that ensure com-
plete coverage of the known haplotype diversity of the locus
according to the HapMap II.
Twenty-five of the 26 tagging SNPs that best represent the
known haplotype diversity of CNR1 were successfully geno-
typed in 1932 obese cases and 1173 controls. Results of the
case–control analyses are given in Table 1 and in Supplemen-
tary Material, Table S1. All SNPs were in Hardy–Weinberg
equilibrium (P . 0.05) in the normal, glucose-tolerant non-
obese French Caucasian control collection. Twelve SNPs
showed nominal evidence of association with childhood
obesity,classI and II
obesity [odds ratios (ORs) between 1.16 and 1.40; P ¼ 0.04
to P ¼ 0.00003]. Using spectral decomposition, we estimated
the total number of efficient tests at 16 (19,20). The number of
SNPs with a P-value ,0.05 accounted for nine tests (accord-
ing to the same method). As an estimate of the concentration
of high P-values, we estimated that the probability of 10 sig-
nificant tests (P , 0.05 level) out of 16 trials was 5.91 ?
The haplotypes showing the lowest Akaike Information Cri-
terion (AICmin) were tested for association with obesity. None
of the major haplotypes (frequency .5%) showed a stronger
association with obesity than the single SNPs (Supplementary
Material, Table S2).
For the replication cohorts, we selected SNPs with P-values
that survived permutation-based correction for multiple testing
in the case–control analysis. After this test, the rs806381 SNP
(210908 A.G) was associated with childhood obesity
rs2023239 SNP (25489 T.C) with class I and II adult
obesity (eP-values ¼ 0.016) (Table 1). To further account
for multiple testing, we also performed a False discovery
rate (FDR) analysis for the 25 SNPs: the tests that displayed
q-value ,0.05 occurred for the two variants described
above (Table 1).
eP-values ¼ 0.002) and the
The potential effect of rs806381 and rs2023239 on the con-
tinuous obesity-related trait, body mass index (BMI), was
then tested in two additional independent European cohorts
(Table 2). In a Swiss cohort of 865 obese adults, both
rs806381 obesity G at-risk allele and rs2023239 obesity
T at-risk allele were associated with increased BMI. For the
Table 1. Association study of two CNR1 variants showing evidence of association with obesity after multiple testing corrections
.Odds ratio (Allelic)
rs806381 (210908 A.G) Chr6:88922620Controls
Class I and II obese adults
Class III obese adults
Class I and II obese adults
Class III obese adults
rs2023239 (25489 T.C) Chr6:88917201
Nucleotide changes are given with the most frequent allele first.
aSignificant P-values after permutation-based analyses (based on 25 SNPs and three cohorts).
bSignificant P-values after the FDR test for a q-value,0.05.
Human Molecular Genetics, 2008, Vol. 17, No. 13 1917
by guest on May 29, 2013
r806381, the GC heterozygous carriers were 1 BMI unit
heavier and the GG homozygous carriers ?3 BMI units
heavier (P ¼ 0.015) than the individuals with rs806381 CC
genotypes. A similar trend was observed for the rs2023239
variant (P ¼ 0.02). Additionally, we studied the effect of
two CNR1 SNPs on BMI in a general population of 1780
Danish adults. Data were in agreement with the Swiss
results, showing a significant additive effect of the rs806381
G at-risk allele: heterozygous carriers had a higher BMI than
the CC homozygotes and a lower BMI when compared with
the GG homozygotes (P ¼ 0.0023). For the rs2023239
variant, the obesity T at-risk allele carriers had a significantly
higher BMI in the Danish cohort (P ¼ 0.021). We further
tested for association of rs806381 and rs2023239 on BMI in
the French obese children cohort and in a pooled cohort of
French obese adults (class I, II and III obese individuals).
While no significant result of the Z-score of BMI was observed
in the children cohort, we found that the rs806381 obesity
G at-risk allele carriers had a significantly higher BMI when
compared with the non carriers in the obese adult cohort
(P ¼ 0.016).
Overall significance was assessed using Fisher’s method,
which combines P-values of independent analyses. We used
the number of effective tests (19) at each step to correct for
multiple testing while accounting for the between SNPs corre-
lations. The meta-analysis combining evidence of association
for obesity and higher BMI found for SNP rs806381 and
rs2023239, P (df ¼ 6) ¼ 1.1 ? 1026and P ¼ 4.7 ? 1024,
Fine mapping of rs2023239 and rs806381 LD blocks
In an attempt to identify the potential causal variant(s), we
increased the SNP density of rs2023239 and rs806381 LD
blocks. Using phase II genotyping data from the HapMap
project (http://www.hapmap.org) in the 37.4 kb region defin-
ing the CNR1 locus, we selected 17 SNPs in LD (r2. 0.5)
with rs806381 and rs2023239. The comparison of the fine
mapped SNP to the initial LD block representative SNP
(rs806381 and rs2023239), (for the same subjects) found one
additional variant for each block with increased significance
of association with obesity (Fig. 1 and Supplementary
Material, Table S3). In the rs2023239 block, in the class
I and II obese adult group, the rs10485170 obesity A at-risk
allele carriers are at substantially increased risk of being
obese [OR ¼ 1.85 (1.41–2.44), P ¼ 0.000008] compared
with the rs2023239 T at-risk allele carriers [OR ¼ 1.40
(1.16–1.69), P ¼ 0.0004]. In the rs806381 block, in the child-
hood obesity group, the risk was also increased for one of the
SNPs: rs6454674 OR ¼ 1.52 (1.26–1.84), P ¼ 0.00001 com-
pared with rs806381 OR ¼ 1.39 (1.19–1.63), P ¼ 0.00003
(Fig. 1 and Supplementary Material, Table S3).
To test if these two new SNPs were the best associated
SNPs in their respective blocks, we performed a logistic
regression analysis including in the model the initial LD
block representative (i.e. rs2023239) and the fine mapping
SNP (i.e. rs10485170). Interestingly, the rs10485170 was
found to be significantly associated with obesity independently
of the rs2023239 (P ¼ 0.015), whereas rs2023239 no longer
remained associated with obesity with the rs10485170 in the
model (P ¼ 0.23). In the logistic regression analysis of the
other LD block, we could not discriminate the effect of
rs806381 and rs6454674 on obesity.
Although little is known on how the endocannabinoid system
contributes to human energy homeostasis (21), data derived
from phase III trials of the CB1-receptor inhibitor rimona-
bant indicate clinical benefits of this antagonist on obesity
and associated phenotypes (10–13). Based on these findings,
we have tested the contribution of the endocannabinoid
receptor 1 gene (CNR1) frequent genetic variation on the
risk of obesity and increased BMI. An initial French case–
control analysis of 25 tagging SNPs representative of the
known haplotype diversity of the CNR1 locus, found a sig-
nificantly higher number of associated SNPs than expected
by chance (5.91 ? 10210). In an attempt to minimize false
positives that could result from the number of variants
tested, we performed a multiple testing permutation-based
correction and selected two variants that were not in LD
(r2, 0.05) for replication. The first replication sample was
a Swiss cohort with patients with severe affection status
(BMI . 35 kg/m2). With the initial case–control study, the
Swiss cohort included a large proportion of cases enriched
Table 2. BMI analysis according to CNR1 genotypes in a Swiss obese adult cohort and in a Danish cohort representative of the general population
Swiss obese adults (Class II and III)
Danish subjects representative of the general population (Inter99)
BMI (kg/m2) 25.9+4.3
43.7+6.2 41.6+5.9 0.02
Data are means+SD.?Significant P-values after simple Bonferroni correction for four tests (1 phenotypes ? 2 SNPs ? 2 cohorts).
P-values are one-sided as we test the specific hypothesis of increased frequency of rs806381 G-allele and rs2023239 T-allele in replication cohorts.
1918Human Molecular Genetics, 2008, Vol. 17, No. 13
by guest on May 29, 2013
for family history leading to an overestimation of the general
population risk. For this reason, we chose to further test this
variant in a general population-based Danish cohort. Despite
the dissimilarity of these two independent replication
cohorts, we found that both rs2023239 and the rs806381
at-risk allele carriers had a significantly higher BMI.
remain slightly over the genome-wide level of significance
(P ¼ 5.1027), the consistent associations provide encoura-
ging prospects for subsequent studies.
Recently a group found no evidence of an association of the
rs2023239 in German obese children and adolescents (17).
Interestingly, this variant that was strongly associated with
obesity and BMI in our obese adult case–control and adult
replication cohorts (overall P , 1025) was not found to be
associated with childhood obesity in our study group.
Although it is too early to conclude an age-dependent effect
of CNR1 variants, this result supports the study of our most
rs6454674) in obese children cohorts. Our work also follows
the association study of two other CNR1 variants (3813A/
obesity-related traits (14). This study that only tagged 8% of
the common variation of the CNR1 locus found an association
of rs12720071 with BMI and obesity-related phenotypes (18).
The number of variants more significantly associated with
of the three studies
obesity in our study and the low level of LD of rs12720071
with the two rs806381 and rs2023239 LD blocks tends to
exclude this variant as the causal one.
Presently, there are no features to qualify rs806381 and
rs2023239 as functional SNPs, as there are no clear genetic
mechanisms to explain how they may alter the function or
expression of CNR1. For this reason, we chose to genotype
all SNPs in partial LD with these two variants for the identifi-
cation of potential causal variants. The genotyping of 17
additional variants in our initial case–control groups identified
two variants showing increased significance of association
with obesity. The comparison of these variants to the two
LD block representatives found that the rs10485170 had a sig-
nificantly stronger effect on obesity than the rs2023239, while
we could not discriminate between the effect of the rs806381
and rs6454674 on obesity.
In conclusion, we suggest that the consecutive replications
and overall significance of associations with obesity and
BMI found for frequent CNR1 variants, in a total of 5750
white European participants, of all ages, in the most complete
analysis of the CNR1 locus with obese children and adults,
strongly support the contribution of the CNR1 gene to poly-
genic obesity in European populations. Furthermore, our
results raise the important question of whether the pharmaco-
logical effect of CNR1 antagonists could differ between
patients with different CNR1 genotypes.
Figure 1. Structure of the CNR1 gene with the location of 26 common polymorphisms and the LD map. (a) Schematic representation of human CNR1 transcript
isoforms (CB1A-E) as described by Zhang PW et al. (18); blue thin lines and bars represent introns and exons (ex1. ex1a. ex2. ex3a. ex3 and ex4) respectively;
green CDS box indicates coding region. (b) Bars show sequenced regions in the 37.4 kb studied interval (c) 17 tagging SNPs found using the pairwise method
(r2. 0.8) based on the sequencing data and 9 additional tagging SNPs (?) from HapMap II; SNPs highlighted in bold have a P-value that resisted multiple
testing corrections; positions of new markers with no rs number are given relative to first base of ATG start codon. (d) LD plot indicating r2between the
26 SNPs with a MAF ?5% found by sequencing. Shades of grey indicate 0 , r , 1. r2¼ 0 being white and r2¼ 1 being black.
Human Molecular Genetics, 2008, Vol. 17, No. 131919
by guest on May 29, 2013
MATERIALS AND METHODS
The study protocols were approved by all local ethics commit-
tees and informed consent was obtained from each subject
before participation in the studies.
Case–control study. Subjects in the case–control association
studies were all French Caucasians recruited through a multi-
media campaign run by the Centre National de la Recherche
Scientifique (CNRS), Hotel Dieu Hospital, the Pasteur Insti-
tute, Lille and the Department of Paediatric Endocrinology
of Jeanne de Flandres Hospital.
The case–control association study included three case
groups: (i) 615 obese children (BMI above the 97th percen-
tile), (ii) 625 class I and II obese adults (30 kg/m2? BMI ?
40 kg/m2) (22), (iii) 692 class III obese adults (BMI .
40 kg/m2) (22). The case groups were compared to a total of
1173 controls. The control subjects were unrelated adult non-
obese French Caucasians pooled from three separate studies: a
set of (i) 266 individuals (mean BMI ¼ 23.1+2.16 kg/m2;
mean age 42.5+4.5 years; men/women 105/161), (ii) 171
controls (mean BMI ¼ 22.9+2.3 kg/m2; mean age 60.7+
11.5 years; men/women 71/100, (iii) 736 individuals (mean
BMI ¼ 23.8+1.8 kg/m2; mean age 53.5+5.6 years; men/
women 293/443) recruited, respectively, at the CNRS Lille,
the ‘Fleurbaix-Laventie Ville Sante ´’ study (23) and the Epide-
miologic Data on the Insulin Resistance Syndrome (DESIR)
Study (24). Before being pooled, the three control sets were
compared with each other. The Woolf test was applied to
assess the genotypic homogeneity among the control studied
groups on 2 ? 2 ? k tables over strata (25,26). This test ident-
ified no genotypic heterogeneity between the different groups
(Supplementary Material, Table S4).
Replication cohorts. Associations with BMI have been tested
in 865 class II and III (BMI ? 35 kg/m2) unrelated Cauca-
sian subjects from Switzerland (27) (mean BMI ¼ 43.4+
7.1 kg/m2; mean age 42.8+10.7 years; men/women 667/
198) and in a random sample of 1780 middle-aged individuals
recruited from the Danish general population (Inter99 cohort)
(mean BMI ¼ 26+4.4 kg/m2; mean age 45.9+7.9 years;
men/women 914/866) (Supplementary Material, Table S5).
Sequencing and SNP genotyping
Sequencing was performed using the automated ABI Prism
3730 DNA sequencer in combination with the Big Dye Ter-
minator Cycle Sequencing Ready Reaction Kit (Applied Bio-
systems). Tagging SNPs were genotyped with SNplex,
Taqman (Applied Biosystems) or Lightyper (Roche Diagnos-
tics) technologies. As a standard laboratory quality control
measure, a random 10% of DNA samples were systematically
re-genotyped to ensure minimal genotyping error. Con-
cordance rate was comprised between 99 and 100% for the
Case–control analyses. Allele frequencies in cases and con-
trols were compared using the x2test integrated in Cocaphase
(28). ORs and P-values were from logistic regression, adjusted
for age and gender, in the class I and II and class III adult
obesity case–control study. For the childhood obesity case–
control, the test was adjusted for gender. To test the associa-
tion between case–control data and the combination of
CNR1 SNPs, we selected haplotypes with frequencies of at
least 5% and showing the lowest AICmin. These models
were tested using Cocaphase.
BMI analysis. Associations between BMI and the SNPs were
analyzed using SPSS software (version 14.0.2, SPSS Inc.,
Chicago, IL, USA). For the BMI analysis, we used a general
linear model taking into account gender and age.
Fisher’s test. To produce an overall significance of increased
allelic frequency in obese individuals, we combined the
P-value from the case–control study (all obese versus con-
trols) with the P-values of the replication studies using
Fisher’s method in which the twice the negative sum of the
natural log of n P-values follows x2distribution with 2n
degrees of freedom.
Multiple testing. Multiple testing issues in the case–control
comparisons were addressed using permutation analysis that
takes into consideration the inter-SNPs correlations. For the
study of 25 SNPs in three cohorts, we performed 1000 permu-
tations of case labels for each case–control analysis, obtaining
three empirical P-values per SNP. These P-values were then
corrected for the number of cohorts to obtain an eP-value
for each SNP.
FDR analysis was also performed for the case–control com-
parisons: in this study the cut-off for significance was the
P-value ¼ 0.0006 with an FDR q-value , 0.05.
For the BMI analysis, the uncorrected P-values are pre-
sented and discussed in the main text. In consideration of
the number of statistical tests carried out, a Bonferroni
correction was applied and the significant corrected P-values
(P ? 0.05) in the Table 2 are identified by an asterisk (?).
Fine mapping. To determine the best associated SNPs in each
respective block, we performed a logistic regression test
including in the model, the gender, the initial LD block repre-
sentative and the ‘fine mapping’ SNP.
Supplementary Material is available at HMG Online.
We are indebted to all families and other individuals who par-
ticipated to this study. We thank Olfert Landt at Tib-Molbiol
David-Alexandre Tre ´goue ¨t from INSERM, UMR S 525,
Paris, F-75634 France for his statistical advice and Christophe
Wachter for his advice on the in silico analysis. We would like
1920Human Molecular Genetics, 2008, Vol. 17, No. 13
by guest on May 29, 2013
to thank Natascha Potoczna and Barbara Heude for providing Download full-text
us DNA and quantitative data on the Swiss, Fleurbaix and
Danish cohorts, respectively. We would also like to thank
Philippe Boutin for his help in the management of the
project, Alex I.F. Blakemore for her help on the interpretation
of the data and Vincent Vatin for his technical support.
Conflict of Interest statement: none declared.
This work was funded by the French MINEFI—Ministe `re de
l’Industrie et des Finances—through the project ‘Integrated
Genomic approaches for the molecular dissection of Cannabi-
noid receptor 1 signalling and of CB1 antagonist effects in
Obesity and in the Metabolic Syndrome’. The Danish part of
the project was supported by grants from the Lundbeck Foun-
dation Centre of Applied Medical Genomics in Personalized
Disease Prediction, Prevention and Care, the Danish Health
Research Council and The European Union (HEPADIP,
grant no. LSHM-CT-2005-018734).
1. Hillard, C.J. (2000) Biochemistry and pharmacology of the
endocannabinoids arachidonylethanolamide and 2-arachidonylglycerol.
Prostaglandins Other Lipid Mediat., 61, 3–18.
2. Cota, D., Marsicano, G., Tschop, M., Grubler, Y., Flachskamm, C.,
Schubert, M., Auer, D., Yassouridis, A., Thone-Reineke, C., Ortmann, S.
et al. (2003) The endogenous cannabinoid system affects energy balance
via central orexigenic drive and peripheral lipogenesis. J. Clin. Invest.,
3. Di Marzo, V., Goparaju, S.K., Wang, L., Liu, J., Batkai, S., Jarai, Z.,
Fezza, F., Miura, G.I., Palmiter, R.D., Sugiura, T. et al. (2001)
Leptin-regulated endocannabinoids are involved in maintaining food
intake. Nature, 410, 822–825.
4. Ravinet Trillou, C., Arnone, M., Delgorge, C., Gonalons, N., Keane, P.,
Maffrand, J.P. and Soubrie, P. (2003) Anti-obesity effect of SR141716, a
CB1 receptor antagonist, in diet-induced obese mice. Am. J. Physiol.
Regul. Integr. Comp. Physiol., 284, R345–R353.
5. Ravinet Trillou, C., Delgorge, C., Menet, C., Arnone, M. and Soubrie, P.
(2004) CB1 cannabinoid receptor knockout in mice leads to leanness,
resistance to diet-induced obesity and enhanced leptin sensitivity.
Int. J. Obes. Relat. Metab. Disord., 28, 640–648.
6. Ledent, C., Valverde, O., Cossu, G., Petitet, F., Aubert, J.F., Beslot, F.,
Bohme, G.A., Imperato, A., Pedrazzini, T., Roques, B.P. et al. (1999)
Unresponsiveness to cannabinoids and reduced addictive effects of opiates
in CB1 receptor knockout mice. Science, 283, 401–404.
7. Bari, M., Paradisi, A., Pasquariello, N. and Maccarrone, M. (2005)
Cholesterol-dependent modulation of type 1 cannabinoid receptors in
nerve cells. J. Neurosci. Res., 81, 275–283.
8. Soria, G., Mendizabal, V., Tourino, C., Robledo, P., Ledent, C.,
Parmentier, M., Maldonado, R. and Valverde, O. (2005) Lack of CB1
cannabinoid receptor impairs cocaine self-administration.
Neuropsychopharmacology, 30, 1670–1680.
9. Pagotto, U., Vicennati, V. and Pasquali, R. (2005) The endocannabinoid
system and the treatment of obesity. Ann. Med., 37, 270–275.
10. Pi-Sunyer, F.X., Aronne, L.J., Heshmati, H.M., Devin, J. and Rosenstock,
J. (2006) Effect of rimonabant, a cannabinoid-1 receptor blocker, on
weight and cardiometabolic risk factors in overweight or obese patients:
RIO-North America: a randomized controlled trial. Jama, 295, 761–775.
11. Van Gaal, L.F., Rissanen, A.M., Scheen, A.J., Ziegler, O. and Rossner, S.
(2005) Effects of the cannabinoid-1 receptor blocker rimonabant on
weight reduction and cardiovascular risk factors in overweight patients:
1-year experience from the RIO-Europe study. Lancet, 365, 1389–1397.
12. Scheen, A.J., Finer, N., Hollander, P., Jensen, M.D. and Van Gaal, L.F.
(2006) Efficacy and tolerability of rimonabant in overweight or obese
patients with type 2 diabetes: a randomised controlled study. Lancet, 368,
13. Despres, J.P., Golay, A. and Sjostrom, L. (2005) Effects of rimonabant on
metabolic risk factors in overweight patients with dyslipidemia.
N. Engl. J. Med., 353, 2121–2134.
14. Russo, P., Strazzullo, P., Cappuccio, F.P., Tregouet, D.A., Lauria, F.,
Loguercio, M., Barba, G., Versiero, M. and Siani, A. (2007) Genetic
variations at the endocannabinoid type 1 receptor gene (CNR1) are
associated with obesity phenotypes in men. J. Clin. Endocrinol. Metab.,
15. Peeters, A., Beckers, S., Mertens, I., Van Hul, W. and Van Gaal, L. (2007)
The G1422A variant of the cannabinoid receptor gene (CNR1) is
associated with abdominal adiposity in obese men. Endocrine, 31,
16. Aberle, J., Fedderwitz, I., Klages, N., George, E. and Beil, F.U. (2007)
Genetic variation in two proteins of the endocannabinoid system and their
influence on body mass index and metabolism under low fat diet. Horm.
Metab. Res., 39, 395–397.
17. Muller, T.D., Reichwald, K., Wermter, A.K., Bronner, G., Nguyen, T.T.,
Friedel, S., Koberwitz, K., Engeli, S., Lichtner, P., Meitinger, T. et al.
(2007) No evidence for an involvement of variants in the cannabinoid
receptor gene (CNR1) in obesity in German children and adolescents.
Mol. Genet. Metab., 90, 429–434.
18. Zhang, P.W., Ishiguro, H., Ohtsuki, T., Hess, J., Carillo, F., Walther, D.,
Onaivi, E.S., Arinami, T. and Uhl, G.R. (2004) Human cannabinoid
receptor 1: 50exons, candidate regulatory regions, polymorphisms,
haplotypes and association with polysubstance abuse. Mol. Psychiatry, 9,
19. Nyholt, D.R. (2004) A simple correction for multiple testing for
single-nucleotide polymorphisms in linkage disequilibrium with each
other. Am. J. Hum. Genet., 74, 765–769.
20. Li, J. and Ji, L. (2005) Adjusting multiple testing in multilocus analyses
using the eigenvalues of a correlation matrix. Heredity, 95, 221–227.
21. Horvath, T.L. (2006) The unfolding cannabinoid story on energy
homeostasis: central or peripheral site of action? Int. J. Obes. (Lond), 30
(Suppl 1), S30–S32.
22. (2000) Obesity: preventing and managing the global epidemic. Report of a
WHO consultation. World Health Organ. Tech. Rep. Ser., 894, i–xii.
23. Lafay, L., Basdevant, A., Charles, M.A., Vray, M., Balkau, B., Borys,
J.M., Eschwege, E. and Romon, M. (1997) Determinants and nature of
dietary underreporting in a free-living population: the Fleurbaix Laventie
Ville Sante (FLVS) Study. Int. J. Obes. Relat. Metab. Disord., 21,
24. Jaziri, R., Lobbens, S., Aubert, R., Pean, F., Lahmidi, S., Vaxillaire, M.,
Porchay, I., Bellili, N., Tichet, J., Balkau, B. et al. (2006) The PPARG
Pro12Ala polymorphism is associated with a decreased risk of developing
hyperglycemia over 6 years and combines with the effect of the APM1
G-11391A single nucleotide polymorphism: the Data from an
Epidemiological Study on the Insulin Resistance Syndrome (DESIR)
study. Diabetes, 55, 1157–1162.
25. Woolf, B. (1955) On estimating the relation between blood group and
disease. Ann. Hum. Genet., 19, 251–253.
26. Agresti, A. (2001) Exact inference for categorical data: recent advances
and continuing controversies. Stat. Med., 20, 2709–2722.
27. Branson, R., Potoczna, N., Kral, J.G., Lentes, K.U., Hoehe, M.R. and
Horber, F.F. (2003) Binge eating as a major phenotype of melanocortin 4
receptor gene mutations. N. Engl. J. Med., 348, 1096–1103.
28. Dudbridge, F. (2003) Pedigree disequilibrium tests for multilocus
haplotypes. Genet. Epidemiol., 25, 115–121.
Human Molecular Genetics, 2008, Vol. 17, No. 131921
by guest on May 29, 2013