The Journal of Clinical Investigation|August 2003| Volume 112|Number 3
A combination of central and peripheral mechanisms
modulates food intake and energy consumption to
maintain metabolism and body composition in mam-
mals. Diseases characterized by impaired energy bal-
ance, such as obesity, rank among the most immediate
health threats in industrialized countries, creating an
urgent need to generate new pharmacologic treatment
options (1, 2). The discovery of leptin has triggered
intensified research efforts in the field of energy home-
ostasis, leading to a better understanding of a complex
network of central and peripheral factors that influ-
ence both appetite and energy expenditure (3).
The discovery of cannabinoid receptor type 1
(CB1) and cannabinoid receptro type 2 (CB2), pro-
vided a molecular basis for investigating the effects
The endogenous cannabinoid
system affects energy balance
via central orexigenic drive and peripheral lipogenesis
Daniela Cota,1Giovanni Marsicano,2Matthias Tschöp,3Yvonne Grübler,1
Cornelia Flachskamm,4Mirjam Schubert,5Dorothee Auer,5Alexander Yassouridis,6
Christa Thöne-Reineke,7Sylvia Ortmann,8Federica Tomassoni,9Cristina Cervino,9
Enzo Nisoli,10Astrid C.E. Linthorst,4Renato Pasquali,9Beat Lutz,2Günter K. Stalla,1
and Uberto Pagotto9
1Clinical Neuroendocrinology Group and
2Molecular Genetics of Behavior Group, Max-Planck-Institute of Psychiatry, Munich, Germany
3Department of Pharmacology, German Institute of Human Nutrition, Bergholz-Rehbrücke, Germany
5Magnetic Resonance Imaging and Spectroscopy Group, and
6Biostatistics Group, Max-Planck-Institute of Psychiatry, Munich, Germany
7Max-Rubner-Laboratory, German Institute of Human Nutrition, Bergholz-Rehbrücke, Germany
8Institute for Zoo and Wildlife Research, Berlin, Germany
9Endocrinology Unit and Centro di Ricerca Biomedica Applicata, Sant Orsola-Malpighi Hospital, Bologna, Italy
10Center for Study and Research on Obesity, Department of Preclinical Sciences, School of Medicine, University of Milan,
Laboratori Interdisciplinari di Tecnologie Avanzate Vialba, Luigi Sacco Hospital, Milan, Italy
The cannabinoid receptor type 1 (CB1) and its endogenous ligands, the endocannabinoids, are involved
in the regulation of food intake. Here we show that the lack of CB1in mice with a disrupted CB1gene
causes hypophagia and leanness. As compared with WT (CB1+/+) littermates, mice lacking CB1 (CB1–/–)
exhibited reduced spontaneous caloric intake and, as a consequence of reduced total fat mass, decreased
body weight. In young CB1–/–mice, the lean phenotype is predominantly caused by decreased caloric
intake, whereas in adult CB1–/–mice, metabolic factors appear to contribute to the lean phenotype. No
significant differences between genotypes were detected regarding locomotor activity, body tempera-
ture, or energy expenditure. Hypothalamic CB1 mRNA was found to be coexpressed with neuropep-
tides known to modulate food intake, such as corticotropin-releasing hormone (CRH), cocaine-amphet-
amine–regulated transcript (CART), melanin-concentrating hormone (MCH), and prepro-orexin,
indicating a possible role for endocannabinoid receptors within central networks governing appetite.
CB1–/–mice showed significantly increased CRH mRNA levels in the paraventricular nucleus and
reduced CART mRNA levels in the dorsomedial and lateral hypothalamic areas. CB1was also detected
in epidydimal mouse adipocytes, and CB1-specific activation enhanced lipogenesis in primary adipocyte
cultures. Our results indicate that the cannabinoid system is an essential endogenous regulator of ener-
gy homeostasis via central orexigenic as well as peripheral lipogenic mechanisms and might therefore
represent a promising target to treat diseases characterized by impaired energy balance.
J. Clin. Invest. 112:423–431 (2003). doi:10.1172/JCI200317725.
Received for publication December 30, 2002, and accepted in revised
form April 15, 2003.
Address correspondence to: Uberto Pagotto, Department of
Internal Medicine and Gastroenterology,Endocrinology Unit and
C.R.B.A., S. Orsola-Malpighi Hospital, 40138 Bologna, Italy.
Phone: 39-051-6363009; Fax: 39-051-6363080;
Daniela Cota and Giovanni Marsicano contributed equally to this
Conflict of interest: The authors have declared that no conflict of
Nonstandard abbreviations used: cannabinoid receptor type 1
(CB1); corticotropin-releasing hormone (CRH); cocaine-amphet-
amine–regulated transcript (CART); digoxigenin (DIG); melanin-
concentrating hormone (MCH); energy expenditure (EE); area
under the curve (AUC); respiratory quotient (RQ); lean body mass
(LBM); dual-energy x-ray absorptiometry (DEXA); in situ
hybridization (ISH); lipoprotein lipase (LPL); the paraventricular
nucleus (PVN); neuropeptide Y (NPY); the ventromedial nucleus
(VMN); lateral hypothalamic area (LHA); arcuate nucleus (ARC);
dorsomedial nucleus (DMN); Agouti-related protein (AGRP).
See the related Commentary beginning on page 323.
The Journal of Clinical Investigation| August 2003|Volume 112| Number 3
of marijuana (Cannabis sativa) and its constituents.
Moreover, the characterization of endogenous ligands
for cannabinoid receptors (endocannabinoids) along
with their biosynthesis and degradation pathways
revealed the existence of the endogenous cannabinoid
system, which has recently been shown to modulate
several physiologic functions (4, 5). The appetite-stim-
ulating effect of marijuana in humans has been well
known for centuries (6). Several reports have demon-
strated that administration of cannabinoids stimulates
food intake in animal models (7, 8). Both peripheral
and central administration of anandamide, one of the
major endocannabinoids, increase food intake in
rodents (9–11). On the basis of the observation that
CB1 and endocannabinoids are present in the brain
regions controlling food intake (4), the endogenous
cannabinoid system has been proposed as a putative
modulator of feeding behavior (12, 13). This concept
has been further substantiated by the use of specific
CB1 antagonists, which provided evidence for the role
of CB1 in mediating the orexigenic effect of exogenous
or endogenous cannabinoids (14–16). Importantly, the
levels of hypothalamic endocannabinoids were shown
to be decreased after leptin administration, and the
blockade of CB1 was demonstrated to inhibit starva-
tion-induced hyperphagia in mice (17). However, little
is known about the putative cross-talks of endo-
cannabinoids and CB1 with the large number of hypo-
thalamic neuropeptides known to regulate appetite.
Very recently, additional peripheral targets of the
endogenous cannabinoid system, which are involved in
energy homeostasis, have been described (18).
Thus, CB1-deficient (CB1–/–) mice (19) were used in
the current study as a model to determine the role of
the endogenous cannabinoid system in the regulation
of energy balance. By using this animal model as a
tool, we carefully compared body weight, body com-
position, and feeding behavior in the presence and in
the absence of CB1 in vivo. In particular, we focused
on a potential role of the endogenous cannabinoid
system in centrally modulating energy homeostasis
via hypothalamic neuropeptides as well as on putative
functions of endocannabinoids in adipose tissue.
Animals. Male mice deficient for CB1(CB1–/–) and male
WT littermates (CB1+/+), originally generated and
genotyped as described earlier (19), were used. Mice
were in a mixed genetic background, with a predomi-
nant C57BL/6N contribution (seven backcrossings).
In complementary experiments, no differences in
body weight were observed between male CB1+/–and
CB1+/+littermates at any age (data not shown). There-
fore, in all experiments only homozygous mutant
mice were used. Mice were individually housed in the
animal facility of the Max-Planck-Institute of Psychi-
atry, Munich, Germany, under standard conditions
with a 12:12-hour light/dark cycle (light on, 0600
hours) at 22°C. All animal procedures complied with
the guidelines for the care and use of laboratory ani-
mals of the Governments of the States of Bavaria and
Weight curves. Fifteen CB1+/+and 15 CB1–/–litter-
mates were studied. All mice had adlibitum access to
pelleted standard mouse chow (Altromin, Lage, Ger-
many) and tap water. Body weight was measured
weekly at 0900 hours in all animals starting at 3
weeks of age until 15 weeks of age.
Glucose and hormones measurements. Blood was drawn
at 0800 hours from 15-week-old animals and cen-
trifuged at 1000 g. Plasma glucose levels were meas-
ured with the use of the Beckman Glucose Analyzer 2
(Beckman, Palo Alto, California, USA). Plasma insulin
and leptin levels were measured with RIA kits (Linco,
St. Charles, Michigan, USA).
Analysis of body composition by quantitative NMR. At the
end of the evaluation of the body weight curve, ani-
mals were killed and the carcasses were analyzed for
body composition. The measurements were per-
formed by using NMR technology (Bruker’s Minispec
MQ10, Houston, Texas, USA). The software to meas-
ure body tissue composition was developed by Echo
Systems (Houston, Texas, USA) for Eli Lilly (Indi-
anapolis, Indiana, USA). NMR imaging was per-
formed in mice by 7 Tesla Scanner (Bruker). The
imaging protocol consisted of T2-weighted images in
all three orientations.
Indirect calorimetry, locomotor activity, and body temper-
ature measurements. Energy expenditure (EE) and res-
piratory quotient (RQ) were measured for 24 hours,
in a different, additional set of six male CB1–/–and six
male CB1+/+mice, 16-weeks-old, by using an open cir-
cuitry calorimetry system. For measurements of oxy-
gen consumption VO2(MAGNOS 16, Hartmann &
Braun, Frankfurt/Main, Germany) and carbon diox-
ide production VCO2(URAS 14, Hartmann & Braun),
mice were placed in air-tight respiratory cages, which
were continuously ventilated with a flow rate of about
30 l/h. Total VO2was recorded in 6-minute intervals
for each animal and calculated as described (20). RQ
is the ratio of VCO2to VO2. Weight-specific EE was
calculated by dividing EE (kJ/d) by the lean body mass
(LBM) of the animal and expressed as (kJ/g LBM/d).
LBM was determined by using a dual-energy x-ray
absorptiometry (DEXA) method (PIXImus, Lunar,
Madison, Wisconsin, USA). In a separate set of exper-
iments, performed on a new batch of age-matched
mice, core body temperature and motor activity were
monitored with a computer-controlled automatic
biotelemetry system (Data Sciences International, St.
Paul, Minnesota, USA) as described before (21).
Briefly, a battery-powered transmitter (TA-F40) was
implanted intraperitoneally under isoflurane anes-
thesia and sterile conditions. After the surgery, mice
were housed individually. The assessment of the cir-
cadian rhythms of body temperature and locomotor
activity started 1 week after the surgery to guarantee
complete recovery of the animals. Every 2 minutes for
The Journal of Clinical Investigation|August 2003| Volume 112| Number 3
48 hours, body temperature was measured and the
cumulative activity score for the respective time inter-
val was recorded continuously.
Pair-feeding experiments. For pair-feeding experi-
ments, new batches of male mice were used. Twelve
CB1+/+and six CB1–/–littermates were maintained on
standard pelleted mouse chow (Altromin). Food
intake and body weight were monitored daily at 0900
hours. To measure food intake, the pelleted food was
weighed and then placed into the food container of
the cage; the food remaining 24 hours later was
weighed again and the difference, corrected for the
food spillage, represented the daily food intake. Six
CB1+/+and six CB1–/–were fed ad libitum, receiving a
preweighed amount of food known to be largely in
excess of daily consumption. The remaining six CB1+/+
were the pair-fed group that, on each day, received the
same amount of food as consumed by six CB1–/–lit-
termates on the previous day. The experiments were
performed with two different groups of animals:
young (3 weeks old) and adult (20 weeks old).
In situ hybridization. A new batch of seven male CB1+/+
and seven male CB1–/–mice, 16 weeks old, fed adlibi-
tum, were killed and theirbrains were removed. Hypo-
thalamic coronal sections (16 µm) were cut on a cryo-
stat (Microtome HM560, Walldorf, Germany),
mounted onto frozen SuperFrost/Plus slides (Menz-
er-Glaser, Braunschweig, Germany), dried, and stored
at –20°C, until used. CB1cDNA was produced from a
1530-bp fragment as described (22). MCHcDNA (513-
bp fragment, GenBank accession AK020723,
nucleotides 64-577, forward primer 5′-GGATG-
GCAAAGATGACTCTCT and reverse primer 5′-CGGAC-
CAACAGGTATCAAACT) was generated by RT-PCR
from total mouse brain RNA and was cloned into
pBluescript KS–(Stratagene, La Jolla, California,
USA). Mouse neuropeptide Y (NPY) cDNA was an
IMAGE clone (number 482891) purchased from
Research Genetics-Invitrogen (Karlsruhe, Germany).
Rat cocaine-amphetamine-regulated transcript (CART)
cDNA (Genbank accession AI112077) was a clone
purchased from Research Genetics-Invitrogen and
kindly provided by J. Dreyer, University of Fribourg,
Switzerland. CRH cDNA was a 530-bp fragment
(nucleotides 61-591) of mouse CRH (23) kindly pro-
vided by W. Wurst, Max-Planck-Institute of Psychia-
try, Munich, Germany. Prepro-orexin riboprobe was
generated from a clone kindly provided by T.E. Scam-
mell (Beth Israel Deaconess Medical Center, Boston,
Massachusetts, USA). The identity of all cDNAs was
checked by sequencing. Restriction enzymes (New
England Biolabs, Beverly, Massachusetts, USA) used
for linearization, and RNA polymerases (Roche,
Mannheim, Germany) used for the generation of each
riboprobe were as follows: CB1 sense, PstI, T7; CB1
antisense, BamHI, T3; MCH sense, HindIII, T7; MCH
antisense, XbaI, T3; NPY sense NotI, T7; NPY anti-
sense, EcoRI, T3; CART sense, NotI, T7; CART anti-
sense, EcoRI, T3; CRH sense, BamHI, T7; CRH anti-
sense, XbaI, Sp6; prepro-orexin sense, NotI, T3; pre-
pro-orexin antisense, SpeI, T7. According to protocol
design, riboprobes were labeled with 35S or with digox-
igenin (DIG), respectively, as described (24).
Single in situ hybridization (ISH) was carried out as
described (22). In brief, after pretreatment, hybridiza-
tion was carried out overnight at 65°C in hybridization
buffer containing 35S-labeled riboprobe (35,000–70,000
cpm/µl). Slides were then washed and exposed on Bio-
max MR films (Kodak, Stuttgart, Germany). Quantifi-
cation of mRNA expression for each neuropeptide was
performed on autoradiographic films, by using NIH
Image program (http://rsb.info.nih.gov/nih-image/).
For each area analyzed, at least 4 sections per animal
were quantified after image thresholding, and data
were expressed as percentage of mean value of control
CB1+/+mice.Double ISH was performed as described
(24). In brief, after pretreatment, hybridization was car-
ried out in buffer containing the 35S- and DIG-labeled
riboprobes at 54°C overnight. After washing, sections
were treated with the TSA Biotin System (NEN,
Boston, Massachusetts, USA) and incubated with strep-
tavidin-alkaline phosphatase (Roche, Mannheim, Ger-
many). After developing of the chromogenic reaction
with Vector Red kit (Vector Laboratories, Burlingame,
California, USA), sections were exposed on Biomax MR
Films (Kodak) and dipped in NTB-2 photographic
emulsion (Kodak) on the following day. After exposure
for 5 to 15 days at 4°C, depending on the riboprobe
used, slides were developed and counterstained in tolu-
idine blue. CB1-specific signal was evaluated as
described (24). Coexpressing cells were counted at a sin-
gle-cell resolution. Values were expressed as percentage
of CB1-expressing cells per number of cells positive for
a specific neuropeptide.
Adipose tissue cell culture. Ten 16-week-old C57BL/6N
mice (Charles River, Sulzfeld, Germany) were killed and
the epidydimal fat pads were removed and immediate-
ly collected in Krebs buffer (NaCl 123 mM, KCl 5 mM,
CaCl213 mM, glucose 5 mM, BSA 1.5%, and HEPES
100 mM). Primary adipose cells were cultured as
described (25). Briefly, cells were grown in DMEM con-
taining 10% calf serum, sodium pyruvate (1 mM), L-glu-
tamine (4 mM), and antibiotics and were maintained
in 5% CO2at 37°C. After 3 days, differentiation was
induced by adding insulin (1 µM) and triiodothyronine
(0.2 nM) to the medium. Cells were treated after 3 days
with the CB1 agonist WIN-55,212 (Tocris, Bristol, UK)
or the CB1 antagonist SR 141716A (NIMH Chemical
Synthesis and Drug Supply Program) or a combination
of both compounds (addition of SR 141716A was 45
minutes before WIN-55,212 treatment) for 48 hours.
Controls were treated with vehicle (according to the
dilution of the drugs, 0.0001% of DMSO was added to
the medium). Heparin-releasable lipoprotein-lipase
(LPL) activity was measured with a commercial kit
(CONFLUOLIP total lipase test, Progen, Heidelberg,
Germany) according to manufacturer’s instructions.
Experiments were performed in quadruplicates.
The Journal of Clinical Investigation|August 2003| Volume 112| Number 3
RNA isolation and RT-PCR. Total RNA was extracted
from epydidimal fat pads of CB1+/+and CB1–/–mice
and from primary adipocytes cell cultures derived
from C57BL/6N mice and reverse transcribed as
described earlier (26). Primers for CB1 were as fol-
lows: CB1 5′-GGTTCTGATCCTGGTGGTGTTGAT and
MM18374). Primers for β-actin were as described
previously (27). Conditions for CB1 amplification
were 35 cycles, for 1 minute each at 94°C, 55°C, and
72°C and conditions for β-actin were 35 cycles, for 1
minute each at 94°C, 60°C, and 72°C. The PCR
products were separated on a 1.5% agarose gel, and
the bands were visualized with ethidium bromide
staining. Negative controls for RT-PCR were per-
formed in the absence of reverse transcriptase.
Mouse hippocampus RNA was used as a positive
control for CB1.
Statistical analysis. All values are reported as means ±
SEM. The data were analyzed partly by one factorial
and partly by two factorial univariate or multivariate
analyses of variance ([M]ANOVA). In the two factori-
al ANOVAs, the one factor was a within-subjects fac-
tor expressing the repeated measures sequence. In
cases of significant factor effects, tests with contrasts
or post-hoc Scheffé tests followed to locate pairs of
factor levels with significant differences in the exam-
ined variables. Unpaired t test was used for the com-
parison of plasma hormone concentrations between
groups of animals and the Mann-Whitney U test was
used for the analysis of ISH data. As nominal level of
significance, α = 0.05 was accepted and corrected
accordingly (Bonferroni adjustment) for all posteriori
tests (test with contrasts, Scheffé test, etc.). P values
less than 0.05 denote statistical significance.
CB1–/–mice are lighter and leaner than CB1+/+littermates.
Male CB1+/+and CB1–/–were fed adlibitum on a stan-
dard diet from 3 to 15 weeks of age and body weight
was recorded weekly. Starting from week 3, CB1–/–
mice had a lower body weight than CB1+/+mice, and
CB1–/–mice maintained a significantly reduced body
weight as compared to CB1+/+throughout the whole
period of observation (P <0.05) (Figure 1a). The analy-
sis of body composition of the same animals using
quantitative NMR showed a markedly decreased per-
centage of fat mass and a slightly increased percent-
age of lean mass in CB1–/–mice compared with CB1+/+
(CB1–/–fat mass, 9.7% ±0.55% vs. CB1+/+12.3% ±0.51%,
P < 0.005) (CB1–/–lean mass, 73.1% ± 0.43% vs. CB1+/+
71.6% ±0.44%, P <0.05) (Figure 1, b and c). Plasma lep-
tinlevels were significantly lower in CB1–/–mice (CB1–/–
leptin, 1.1 ± 0.1 ng/ml vs. CB1+/+1.8 ± 0.3 ng/ml,
P < 0.05), whereas plasma insulin and glucose levels
were not different between CB1–/–and CB1+/+mice
(CB1–/–insulin, 0.9 ± 0.2 ng/ml vs. CB1+/+1.0 ± 0.3
ng/ml, P >0.05; CB1–/–blood glucose, 169 ±11 mg/dl
vs. CB1+/+164 ± 11 mg/dl, P >0.05) (Table 1).
To determine whether differences in thermogenesis
or spontaneous physical activity were contributing to
the lean phenotype of male CB1–/–mice, body temper-
ature and locomotor activity of both genotypes were
studied. Both CB1–/–and CB1+/+mice showed circadi-
an variations in body temperature (time: F23–184 = 23.4,
P < 0.001) and locomotor activity (time: F23–184 = 5.4,
Body weight, body composition and EE of CB1–/–mice. (a) Body
weight curves in male mice starting at 3 weeks of age. Each data
point represents mean ± SEM of 15 mice. *P < 0.05 vs. CB1+/+lit-
termates. (b) NMR imaging performed in 16-week-old mice. White
arrow, visceral fat; black arrow, subcutaneous fat; T, testis; B, blad-
der. (c) Analysis of body composition by quantitative NMR. Fat
mass and lean mass expressed as percentage of the body weight.
The columns represent the mean ± SEM of 15 CB1+/+and CB1–/–
mice, respectively. *P < 0.05 and **P < 0.005 vs. CB1+/+controls.
(d) EE in male CB1+/+and CB1–/–mice. Data are normalized to
LBM. Each data point represents mean ± SEM of six mice.
Plasma glucose and hormone levels in CB1+/+and CB1–/–mice
164 ± 11
1.0 ± 0.3
1.8 ± 0.3
169 ± 11
0.9 ± 0.2
1.1 ± 0.1A
Plasma glucose and hormone levels were measured in 15-week-old mice
(n = 15 per genotype).AP < 0.05.
The Journal of Clinical Investigation|August 2003| Volume 112| Number 3
P < 0.001), as both parameters increased during the
dark period. However, no significant differences were
observed between genotypes or in the interaction
between time and genotype (P >0.05).
Energy expenditure (EE) was evaluated by perform-
ing indirect calorimetry (simultaneous analysis of O2
consumption and CO2production), reflecting energy
combustion and the proportion of fat and carbohy-
drate oxidation. Both genotypes showed circadian
variations in EE (time: F23–227= 15.8, P <0.001). How-
ever, no significant differences were observed between
genotypes either during the light period (genotype:
F1–10= 0.1, P > 0.05; genotype × time: F11–107= 0.7,
P >0.05) (Figure 1d) or during the dark period (geno-
type: F1–10= 1.3, P >0.05; genotype ×time: F11–110= 1.7,
P >0.05) (Figure 1d).
To further elucidate the mechanisms underlying
the body fat deficiency observed in CB1–/–mice, we
determined the caloric intake of young (3 weeks of
age) and adult (20 weeks of age) male CB1–/–vs. CB1+/+
mice. A significant reduction in food intake was
observed in both young and adult CB1–/–mice com-
pared with CB1+/+littermates (caloric intake meas-
ured as area under the curve [AUC]; young mice:
CB1–/–AUC65.6 ± 1.71 vs. CB1+/+77.6 ± 1.1, P < 0.05;
adult mice: CB1–/–AUC 214.53 ± 5.85 vs. CB1+/+
260.84 ± 9.12, P <0.05) (Figure 2, a and b). Pair-feed-
ing experiments were performed to unmask the puta-
tive contribution of impaired metabolic processes
to the lean phenotype of CB1–/–mice, in addition
to their decreased caloric intake described above.
Food intake and pair-feeding studies in CB1–/–
mice. (a) Daily energy intake of young CB1+/+and
CB1–/–mice. Each data point represents mean ±
SEM of six mice for each group. *P < 0.05, AUC of
caloric intake of CB1–/–mice vs. CB1+/+littermates.
(b) Daily energy intake of adult CB1–/–and CB1+/+
mice. Each data point represents mean ±SEM of six
mice for each group. *P < 0.05, AUC of caloric
intake of CB1–/–mice vs. CB1+/+littermates. (c)
Body weight curves during pair-feeding in young
mice. Each data point represents mean ±SEM of six
mice for each group. *P < 0.05 CB1–/–mice vs.
CB1+/+controls; †P < 0.05 pair-fed CB1+/+mice vs.
CB1+/+controls. (d) Body weight curves during pair-
feeding in adult mice. Each data point represents
mean ± SEM of six mice for each group.*P < 0.005
CB1–/–mice vs. CB1+/+controls; #P < 0.005 CB1–/–
mice vs. pair-fed CB1+/+mice. Body weight did not
significantly differ between pair-fed CB1+/+mice and
CB1 transcripts are co-localized with mRNAs of hypothalamic neu-
ropeptides. Bright field micrographs. Vector Red staining, CB1; sil-
ver grains, neuropeptides. (a) Co-localization of CB1 and CRH
mRNA in PVN. (b) Co-localization of CB1 and CART mRNA in
PVN. (c) Co-localization of CB1 and prepro-orexin mRNA in LHA.
(d) Co-localization of CB1 and MCH mRNA in LHA. Filled arrow,
cell coexpressing CB1 and the respective neuropeptide; open arrow,
cell expressing only the respective neuropeptide; asterisk, cell
expressing only CB1 mRNA. Scale bars, 10 µm.
The Journal of Clinical Investigation|August 2003| Volume 112|Number 3
Under caloric restriction based on pair-feeding con-
ditions, 3-week-old CB1+/+mice showed a significant
decrease in body weight (P < 0.05). After 1 week of
pair feeding, differences in body weight between pair-
fed CB1+/+and CB1–/–mice were abolished, whereas
CB1+/+mice with ad libitum access to food main-
tained their higher body weight (P <0.05) (Figure 2c).
Notably, pair-feeding experiments of 20-week-old
CB1+/+mice did not cause a significant decrease of
body weight compared with CB1–/–littermates. Body
weight of adult CB1–/–mice after this pair-feeding
experiment was still lower compared with their CB1+/+
littermates (P <0.005) (Figure 2d).
Characterization of hypothalamic neuropeptide expression
patterns in CB1–/–mice. To investigate possible hypo-
thalamic pathways responsible for the decreased food
intake in CB1–/–mice, ISH experiments were per-
formed on hypothalamic tissue samples derived from
male CB1+/+mice to co-localize hypothalamic CB1
mRNA with several neuropeptides known to be
involved in feeding regulation.
ISH for CB1 confirmed our earlier findings that CB1
hybridization signals were detectable in the medial and
lateral preoptic nucleus, the paraventricular nucleus
(PVN), the ventromedial nucleus (VMN), and in the lat-
eral hypothalamic area (LHA) (22). As expected, CRH-
expressing cells were found in the medial preoptic area
and PVN (28) and the expression of NPY mRNA was
limited to the arcuate nucleus (ARC) (29). CART mRNA
was abundantly expressed in the hypothalamus, espe-
cially in the supraoptic nucleus, PVN, VMN, dorsome-
dial nucleus (DMN), LHA, and ARC (30), whereas the
expression of prepro-orexin and MCH mRNAs was
restricted to the neurons localized in LHA and in the
zona incerta (31, 32). In ISH experiments, correspon-
ding sense RNA probes did not give any detectable sig-
nals (data not shown).
CB1 was found to be co-localized with CRH, CART,
prepro-orexin, and MCH (Figure 3), whereas no coex-
pression was found with NPY-positive neurons in
ARC (data not shown). In PVN, 51.4% of CRH-posi-
tive and 52.4% of CART-positive neurons expressed
CB1 mRNA. In DMN and LHA, 24.6% of CART-posi-
Coexpression of CB1 and feeding controlling neuropeptides in
and CB1 (%)
CB1, cannabinoid receptor type 1; NC, number of cells counted; CRH, cor-
ticotropin-releasing hormone; PVN, paraventricular nucleus; CART, cocaine-
amphetamine-regulated transcript; DMN, dorsomedial nucleus; LHA, lat-
eral hypothalamic area; MCH, melanin-concentrating hormone.
Altered expression of hypothalamic neuropeptides transcripts in CB1–/–mice. (a) Representative autoradiographic images showing upreg-
ulation of CRH mRNA in the PVN of CB1–/–hypothalamus. (b) Densitometric quantification by image analysis of areas of CRH mRNA
expression as percentage of control by using four sections from seven animals in each group. *P < 0.05 vs. CB1+/+control. (c) Represen-
tative autoradiographic images showing downregulation of CART mRNA in the DMN, LHA, and ARC of CB1–/–hypothalamus. (d) Den-
sitometric quantification by image analysis of areas of CART mRNA expression as percentage of control by using four sections from seven
animals in each group. **P < 0.005 vs. CB1+/+control.
The Journal of Clinical Investigation| August 2003| Volume 112|Number 3
tive cells expressed CB1 mRNA, in LHA 7.5% of pre-
pro-orexin–positive neurons and 4.2% of MCH-posi-
tive cells coexpressed CB1 mRNA (Table 2).
CRH mRNA in the PVN of male CB1–/–mice was
19.5% ±7.5% higher compared with CB1+/+littermates
(P < 0.05) (Figure 4, a and b). A 14.2% ± 3.3% lower
expression of CART mRNA was observed in DMN
and LHA of CB1–/–mice compared with CB1+/+litter-
mates (P < 0.005) (Figure 4, c and d). No differences
were observed between the two genotypes regarding
the levels of expression of prepro-orexin, MCH, or
NPY (data not shown).
CB1 is expressed in fat tissue and CB1 agonists stimulate
lipogenesis in vitro. CB1 mRNA was found in ex vivo
epidydimal fat pads from male CB1+/+, but not from
CB1–/–mice (Figure 5a) and its expression was main-
tained in in vitro primary adipocyte cell cultures
derived from C57BL/6N mice (Figure 5b). Stimulation
of primary adipocyte cells (derived from C57BL/6N
epidydimal fat) with the CB1 agonist WIN-55,212
dose-dependently increased lipoprotein lipase activity,
and this effect was blocked by the preincubation with
the CB1 selective antagonist SR 141716A, demon-
strating a CB1-mediated effect (Figure 5c).
The obesity epidemic is currently developing into a
major health problem of industrialized countries.
Effective pharmacologic treatment options are need-
ed to limit its harmful consequences. Despite the dra-
matic progress in obesity research, a safe and potent
weight-reducing drug is still not available. The opti-
mism triggered by the discovery of leptin in 1994 was
soon overshadowed by clinical studies showing that
the fat-reducing effects of leptin were limited to a
handful of patients with congenital leptin deficiency
(33). Currently, other clinical candidates targeting
neuroendocrine energy balance mechanisms are
under investigation (34). In this context, the endoge-
nous cannabinoid system represents a new and prom-
ising drug target (13, 35).
Our current study substantiates this concept by
demonstrating that the germline deletion of CB1 in
mice results in a phenotype characterized by
decreased body weight, reduced fat mass, and
hypophagia. The fact that the simultaneous deletion
of the two most potent orexigenic neuropeptides
known to date, NPY and Agouti-related protein
(AGRP), failed to produce a lean phenotype, demon-
strates the redundancy of neuroendocrine factors
driving energy intake (36). Our study, however, indi-
cates that even the high number of endogenous orex-
igenic factors is unable to compensate for the lack of
endogenous cannabinoid action, reflecting its crucial
role in balancing energy homeostasis.
The lean phenotype of CB1–/–mice is likely to be pre-
dominantly caused by reduced orexigenic drive
throughout their lifetime, which would be in agreement
with the reported role of the endogenous cannabinoid
system for hyperphagic response to starvation (17).
Reduced body weight in young and adult CB1–/–
mice might be a direct result of the decreased caloric
intake, which, in turn, might be due to the alter-
ations in the central orexigenic drive as a conse-
quence of the lack of CB1 in hypothalamic circuits.
However, the differing pair-feeding results in young
and mature animals (Figure 2) point to an essential
role of orexigenic drive during early ages and suggest
that additional, food intake-independent mecha-
nisms also contribute to the lean phenotype of adult
CB1–/–mice. Although these observations might be
merely due to adaptation efforts of the body weight
control system, they may point to a previously
unknown function of the endogenous cannabinoid
system within peripheral mechanisms regulating
energy balance. Besides the obvious fact that adipose
tissue represents a major determinant of numerous
metabolic processes, adipocytes caught our attention
Functional CB1 is expressed in adipocytes. (a) RT-PCR for CB1 performed on CB1+/+and CB1–/–epididymal fat pads. Ma, marker; lane 1,
CB1+/+; lane 2, CB1–/–; lane 3, negative control for CB1+/+; lane 4, negative control for CB1–/–; lane 5, hippocampus; lane 6, water. (b)
RT-PCR for CB1 performed on primary adipocyte cell cultures from C57BL/6N mice. Ma, marker; lane 1, C57BL/6N adipocytes; lane 2,
negative control; lane 3, hippocampus; lane 4, water. (c) Effects of different doses of the CB1 agonist WIN-55,212 (WIN) and CB1 antag-
onist SR 141716A (SR) on heparin-releasable LPL activity (expressed as percentage of vehicle control) in primary adipocyte cells from
C57BL/6N mice. *P < 0.05 compared with vehicle control. †P < 0.05 compared with 10–6M WIN.
The Journal of Clinical Investigation|August 2003| Volume 112|Number 3
for their ability to store exogenous cannabinoids,
which are lipophylic substances, for up to 2 weeks
(37). Our data indicate that mRNA coding for CB1 is
expressed in epidydimal fat pads. Moreover, CB1
expressed by primary adipocytes cultured from
C57BL/6N mice appears to be functional because
lipogenesis can be induced by specific stimulation of
the receptor. These observations, together with the
presence of monoacylglycerol lipase in fat tissue, one
of the main endocannabinoid-degrading enzymes
(38, 39), strongly support adipose tissue as a novel
physiologic and pharmacologic target for cannabi-
noid action. Therefore, our results suggest that the
reduced body fat in CB1–/–mice might be determined
by both hypothalamic alterations and impaired
adipocyte function. Notably, a recent study showed
that administration of the CB1 antagonist SR
141716A leads to a marked and sustained reduction
of adiposity in diet-induced obese mice, which could
not be explained by the transient reduction in food
intake caused by this drug (40). It is also intriguing
that endocannabinoids derived from the gastroin-
testinal tract appear to be able to modulate feeding
behavior (18) and it seems likely that CB1 expressed
on adipocytes might be one of the effectors of these
and other sources of endocannabinoids.
On the other hand, the altered expression of hypo-
thalamic neuropeptides in CB1–/–mice supports a role
of the endogenous cannabinoid system in the central
regulation of food intake. CB1–/–mice show higher lev-
els of CRH mRNA, a neuropeptide known to inhibit
food intake as well as to influence energy balance via
the sympathetic nervous system and the hypothalamo-
pituitary-adrenal axis (41). Albeit, NPY fibers project
from the ARC to CRH neurons in PVN (42), it is not
yet clear whether the anorectic effect of CRH might be
inhibited by NPY signaling. Neither is CB1 expressed
in NPY neurons, nor did CB1–/–mice show altered
basal levels of NPY mRNA expression. In addition, the
CB1 antagonist SR 141716A still inhibits starvation-
induced hyperphagia in NPY–/–mice (17). Altogether,
these data indicate that the effects of hypothalamic
endocannabinoids on energy balance might partially
be mediated by CRH, but those effects are likely to be
independent from NPY.
An unexpected result of our ISH analysis was the
reduced CART expression in CB1–/–mice. Initial evi-
dence suggested an anorexigenic role for this peptide
and decreased hypothalamic CART expression was
found in animal models with an obese phenotype
(43). As leptin is a positive regulator of hypothalam-
ic CART mRNA expression (43), CB1–/–mice might
show reduced CART expression because of the
decreased circulating levels of the hormone, overall
reflecting a physiologic compensation effort in reac-
tion to decreased body fat stores. However, recent
data challenged previous observations on the
appetite-inhibiting effect of CART, showing that
intrahypothalamic injection of the peptide might
even increase food intake (44). Moreover, the admin-
istration of the CB1 antagonist SR 141716A did not
affect feeding behavior in CART–/–mice (45). There-
fore, our results seem to confirm that CART might
play a crucial role in the hypothalamic mediation of
orexigenic endocannabinoid action. The coexpres-
sion of CB1 with CRH and CART suggests a direct
influence of the endocannabinoids on the expression
or function of these neuropeptides.
The exact mechanisms underlying the effects of
the endogenous cannabinoid system on food intake
and fat storage remain to be elucidated and further
experiments are needed to clarify these mechanisms
both at central and peripheral levels. The generation
of tissue-specific mutant mice for CB1 will help to
dissect the relative influence of the endogenous
cannabinoid system on different central neuronal
circuits and on different peripheral tissues in regu-
lating energy homeostasis. Another interesting
aspect, which is still to be fully explained, is the ori-
gin of a mild gender specificity of the lean pheno-
type, which is less pronounced in female CB1–/–mice
(unpublished data). In contrast to our results, previ-
ous investigations on CB1–/–mice reported no
changes in body weight between genotypes, proba-
bly because of differences in the inbred strains used
and/or the lack of the analysis of homogenous sub-
cohorts (e.g., males vs. females) (17, 46, 47). Howev-
er, sexual differences are rather common features
observed in other genetically engineered animal
models of body weight regulation, such as MC3-R–/–
and CART–/–mice (48, 49).
Currently, clinical phase III trials (http://www.
clinicaltrials.gov) are ongoing to assess the effects of
the CB1 selective antagonist SR141716A on body
weight in obese patients with and without dislipi-
demia or type 2 diabetes mellitus. Our findings,
showing that the endogenous cannabinoid system
modulates energy homeostasis via a dual mecha-
nism, intriguingly suggests that CB1 antagonists
will be able to centrally target food intake regulation
as well as peripherally block lipogenetic processes,
paving the way for one of the most promising clini-
cal candidates to fight obesity.
This work was supported bya grant Giovani Ricerca-
tori, University of Bologna, and in part by Fondazione
Cassa di Risparmio, Bologna, Italy. Additional sup-
port was given by a grant from the German Research
Council (to B. Lutz). We thank M.L. Heiman
(Endocrine Research, Eli Lilly and Co., Indianapolis,
Indiana, USA) for the body composition analysis; C.
Pagano (University of Padova, Padova, Italy) for the
glucose levels measurements; H. Hermann, B. Wölfel,
A. Daschner, and C. Plaue for technical help; J. Drey-
er and A. Copper for the gift of CART; and T.E. Scam-
mell and W. Wurst for the gifts of prepro-orexin and
CRH clones, respectively.
The Journal of Clinical Investigation| August 2003| Volume 112| Number 3 Download full-text
1.Flegal, K.M., Carroll, M.D., Ogden, C.L., and Johnson, C.L. 2002. Preva-
lence and trends in obesity among US adults, 1999-2000. JAMA.
2.Yanovski, S.Z., and Yanovski, J.A. 2002. Obesity. N. Engl. J. Med.
3.Zhang, Y., et al. 1994. Positional cloning of the mouse obese gene and
its human homologue. Nature. 372:425–432.
4.Howlett, A.C., et al. 2002. International Union of Pharmacology.
XXVII. Classification of cannabinoid receptors. Pharmacol. Rev.
5.Lutz, B. 2002. Molecular biology of cannabinoid receptors.
Prostaglandins Leukot. Essent. Fatty Acids. 66:123–142.
6.Abel, E.L. 1975. Cannabis: effects on hunger and thirst. Behav. Biol.
7.Williams, C.M., Rogers, P.J., and Kirkham, T.C. 1998. Hyperphagia in
pre-fed rats following oral delta9-THC. Physiol Behav. 65:343–346.
8.Koch, J.E. 2001. Delta(9)-THC stimulates food intake in Lewis rats:
effects on chow, high-fat and sweet high-fat diets. Pharmacol. Biochem.
9.Williams, C.M., and Kirkham, T.C. 1999. Anandamide induces overeat-
ing: mediation by central cannabinoid (CB1) receptors. Psychopharma-
cology (Berl). 143:315–317.
10.Jamshidi, N., and Taylor, D.A. 2001. Anandamide administration into
the ventromedial hypothalamus stimulates appetite in rats. Br. J. Phar-
11.Hao, S., Avraham, Y., Mechoulam, R., and Berry, E.M. 2000. Low dose
anandamide affects food intake, cognitive function, neurotransmitter
and corticosterone levels in diet-restricted mice. Eur. J. Pharmacol.
12.Mechoulam, R., and Fride, E. 2001. A hunger for cannabinoids. Nature.
13.Cota, D., et al. 2003. The endogenous cannabinoid system as a modu-
lator of food intake. Int. J. Obes. Relat. Metab. Disord. 27:289–301.
14.Simiand, J., Keane, M., Keane, P.E., and Soubrie, P. 1998. SR 141716, a
CB1 cannabinoid receptor antagonist, selectively reduces sweet food
intake in marmoset. Behav. Pharmacol. 9:179–181.
15.Colombo, G., et al. 1998. Appetite suppression and weight loss after
the cannabinoid antagonist SR 141716. Life Sci. 63:L113–L117.
16.Freedland, C.S., Poston, J.S., and Porrino, L.J. 2000. Effects of
SR141716A, a central cannabinoid receptor antagonist, on food-main-
tained responding. Pharmacol. Biochem. Behav. 67:265–270.
17.Di Marzo, V., et al. 2001. Leptin-regulated endocannabinoids are
involved in maintaining food intake. Nature. 410:822–825.
18.Gomez, R., et al. 2002. A peripheral mechanism for CB1 cannabinoid
receptor-dependent modulation of feeding. J. Neurosci. 22:9612–9617.
19.Marsicano, G., et al. 2002. The endogenous cannabinoid system con-
trols extinction of aversive memories. Nature. 418:530–534.
20.Klaus, S., Munzberg, H., Truloff, C., and Heldmaier, G. 1998. Physiol-
ogy of transgenic mice with brown fat ablation: obesity is due to low-
ered body temperature. Am. J. Physiol. 274:R287–R293.
21.Linthorst, A.C.E., Karanth, S., Barden, N., Holsboer, F., and Reul,
J.M.H.M. 1999. Impaired glucocorticoid receptor function evolves in
aberrant physiological responses to bacterial endotoxin. Eur. J. Neurosci.
22.Marsicano, G., and Lutz, B. 1999. Expression of the cannabinoid recep-
tor CB1 in distinct neuronal subpopulations in the adult mouse fore-
brain. Eur. J. Neurosci. 11:4213–4225.
23.Seasholtz, A., Bourbonais, F.J., Harnden, C.E., and Camper, S.A. 1991.
Nucleotide sequence and expression of the mouse corticotropin-releas-
ing hormone gene. Mol. Cell. Neurosci. 2:266–273.
24.Hermann, H., Marsicano, G., and Lutz, B. 2002. Coexpression of the
cannabinoid receptor type 1 with dopamine and serotonin receptors
in distinct neuronal subpopulations of the adult mouse forebrain. Neu-
25.Ailhaud, G. 2001. Adipose tissue protocols. Humana Press. Totowa, New
Jersey, USA. 334 pp.
26.Pagotto, U., et al. 1995. Expression and localization of endothelin-1
and endothelin receptors in human meningiomas. Evidence for a role
in tumoral growth. J. Clin. Invest. 96:2017–2025.
27.Matsumoto, F., et al. 2002. Effects of nutrition on the cell survival and
gene expression of transplanted fat tissues in mice. Biochem. Biophys.
Res. Commun. 295:630–635.
28.Olschowka, J.A., O’Donohue, T.L., Mueller, G.P., and Jacobowitz, D.M.
1982. The distribution of corticotropin releasing factor-like
immunoreactive neurons in rat brain. Peptides. 3:995–1015.
29.Allen, Y.S., et al. 1983. Neuropeptide Y distribution in the rat brain.
30.Couceyro, P.R., Koylu, E.O., and Kuhar, M.J. 1997. Further studies on
the anatomical distribution of CART by in situ hybridization. J. Chem.
31.de Lecea, L., et al. 1998. The hypocretins: hypothalamus-specific pep-
tides with neuroexcitatory activity. Proc. Natl. Acad. Sci. U. S. A.
32.Bittencourt, J.C., et al. 1992. The melanin-concentrating hormone sys-
tem of the rat brain: an immuno- and hybridization histochemical
characterization. J. Comp. Neurol. 319:218–245.
33.Farooqi, I.S., et al. 1999. Effects of recombinant leptin therapy in a
child with congenital leptin deficiency. N. Engl. J. Med. 341:879–884.
34.Dhillo, W.S., and Bloom, S.R. 2001. Hypothalamic peptides as drug
targets for obesity. Curr. Opin. Pharmacol. 1:651–655.
35.Berry, E.M., and Mechoulam, R. 2002. Tetrahydrocannabinol and
endocannabinoids in feeding and appetite. Pharmacol. Ther.
36.Qian, S., et al. 2002. Neither agouti-related protein nor neuropeptide
Y is critically required for the regulation of energy homeostasis in mice.
Mol. Cell Biol. 22:5027–5035.
37.Kreuz, D.S., and Axelrod, J. 1973. Delta-9-tetrahydrocannabinol: local-
ization in body fat. Science. 179:391–393.
38.Karlsson, M., Contreras, J.A., Hellman, U., Tornqvist, H., and Holm, C.
1997. cDNA cloning, tissue distribution, and identification of the cat-
alytic triad of monoglyceride lipase. Evolutionary relationship to
esterases, lysophospholipases, and haloperoxidases. J. Biol. Chem.
39.Dinh, T.P., et al. 2002. Brain monoglyceride lipase participating in
endocannabinoid inactivation. Proc. Natl. Acad. Sci. U. S. A.
40.Ravinet, T.C., et al. 2003. Anti-obesity effect of SR141716, a CB1 recep-
tor antagonist, in diet-induced obese mice. Am. J. Physiol. Regul. Integr.
Comp. Physiol. 284:R345–R353.
41.Inui, A. 1999. Feeding and body-weight regulation by hypothalamic
neuropeptides—mediation of the actions of leptin. Trends Neurosci.
42.Li, C., Chen, P., and Smith, M.S. 2000. Corticotropin releasing hor-
mone neurons in the paraventricular nucleus are direct targets for neu-
ropeptide Y neurons in the arcuate nucleus: an anterograde tracing
study. Brain Res. 854:122–129.
43.Kristensen, P., etal. 1998. Hypothalamic CART is a new anorectic pep-
tide regulated by leptin. Nature. 393:72–76.
44.Abbott, C.R., etal. 2001. Evidence of an orexigenic role for cocaine- and
amphetamine-regulated transcript after administration into discrete
hypothalamic nuclei. Endocrinology. 142:3457–3463.
45.Kunos, G., et al. 2002. Endocannabinoids and appetite: possible inter-
actions with other neurotransmitter systems. Symposium on the
Cannabinoids. Burlington, Vermont, USA, International Cannabinoid
Research Society. (Abstr.)
46.Zimmer, A., Zimmer, A.M., Hohmann, A.G., Herkenham, M., and Bon-
ner, T.I. 1999. Increased mortality, hypoactivity, and hypoalgesia in
cannabinoid CB1 receptor knockout mice. Proc. Natl. Acad. Sci. U. S. A.
47.Ledent, C., etal. 1999. Unresponsiveness to cannabinoids and reduced
addictive effects of opiates in CB1 receptor knockout mice. Science.
48.Chen, A.S., etal. 2000. Inactivation of the mouse melanocortin-3 recep-
tor results in increased fat mass and reduced lean body mass. Nat.
49.Asnicar, M.A., et al. 2001. Absence of cocaine- and amphetamine-reg-
ulated transcript results in obesity in mice fed a high caloric diet.