The CB1 receptor mediates the peripheral effects of ghrelin on AMPK activity but not on growth hormone release

Abstract

This study aimed to investigate whether the growth hormone release and metabolic effects of ghrelin on AMPK activity of peripheral tissues are mediated by cannabinoid receptor type 1 (CB1) and the central nervous system. CB1-knockout (KO) and/or wild-type mice were injected peripherally or intracerebroventricularly with ghrelin and CB1 antagonist rimonabant to study tissue AMPK activity and gene expression (transcription factors SREBP1c, transmembrane protein FAS, enzyme PEPCK, and protein HSL). Growth hormone levels were studied both in vivo and in vitro. Peripherally administered ghrelin in liver, heart, and adipose tissue AMPK activity cannot be observed in CB1-KO or CB1 antagonist-treated mice. Intracerebroventricular ghrelin treatment can influence peripheral AMPK activity. This effect is abolished in CB1-KO mice and by intracerebroventricular rimonabant treatment, suggesting that central CB1 receptors also participate in the signaling pathway that mediates the effects of ghrelin on peripheral tissues. Interestingly, in vivo or in vitro growth hormone release is intact in response to ghrelin in CB1-KO animals. Our data suggest that the metabolic effects of ghrelin on AMPK in peripheral tissues are abolished by the lack of functional CB1 receptor via direct peripheral effect and partially through the central nervous system, thus supporting the existence of a possible ghrelin–cannabinoid–CB1–AMPK pathway.—Kola, B., Wittman, G., Bodnár, I., Amin, F., Lim, C. T., Oláh, M., Christ-Crain, M., Lolli, F., van Thuijl, H., Leontiou, C. A., Füzesi, T., Dalino, P., Isidori, A. M., Harvey-White, J., Kunos, G., Nagy, G. M., Grossman, A. B., Fekete, C., Korbonits, M. The CB1 receptor mediates the peripheral effects of ghrelin on AMPK activity but not on growth hormone release.

Obesity and its related diseases contribute significantly to high rates of mortality and constitute a major health and social problem in developed countries. Ghrelin and the cannabinoids have emerged as important players in the intricate and complex neuronal and hormonal system of the regulation of food intake, body weight, and peripheral glucose and lipid metabolism. Ghrelin administration leads to increased body weight, increased adiposity, and impaired glucose tolerance (1). It has been suggested that ghrelin favors preservation of lipid stores and the catabolism of carbohydrate-derived fuel, thereby increasing the respiratory quotient (2, 3). Ghrelin exerts its known appetite-inducing and metabolic effects by binding to the G_q-coupled growth hormone secretagogue receptor (GHS-R). Cannabinoids stimulate food intake through central cannabinoid receptor type 1 (CB1), but pivotal studies have also proved the importance of the adipose and liver CB1 receptor in mediating the peripheral effects of cannabinoids or cannabinoid antagonists on metabolic regulation (4). CB1-knockout (KO) animals are hypophagic, leaner, and lighter when compared to wild-type (WT) littermates (5), resistant to high-fat diet (HFD; ref. 6), and CB1 antagonist, rimonabant, has weight-independent beneficial effects on lipid and glucose metabolism both in human and animal studies (7, 8). Liver-specific CB1-KO mice fed an HFD had less steatosis, hyperglycemia, dyslipidemia, and insulin and leptin resistance than WT mice (4).

The ghrelin and the cannabinoid signaling pathways converge on AMP-activated protein kinase (AMPK), a highly conserved energy sensor of the cell and a master enzyme of metabolic regulation (9). Once activated, AMPK switches off anabolic pathways, such as lipid, carbohydrate, and protein synthesis, and switches on catabolic pathways, such as glycolysis, glucose uptake, fatty-acid oxidation, and mitochondrial biogenesis. In the hypothalamus, AMPK activation leads to an increase in appetite through the neuropeptide Y–uncoupling protein 2 (NPY-UCP2) pathway (10–12). We have previously shown that both cannabinoids and ghrelin stimulate AMPK activity in the hypothalamus and heart, and both inhibit liver and adipose AMPK activity (13). The interaction between the two systems seems to apply to several of their effects. We have previously shown, at the hypothalamic level, that an intact endogenous cannabinoid system is required for the effects of ghrelin on hypothalamic AMPK activity, on neuronal activity in the paraventricular nucleus of the hypothalamus, and ultimately on appetite (14).

In the current study, we investigated, using the CB1 antagonist rimonabant, and CB1-KO mice, whether the metabolic effects of ghrelin on AMPK and lipidogenic enzymes in peripheral tissues are also CB1 dependent and whether the central nervous system is involved in the mediation of the effects of ghrelin on the AMPK activity of peripheral tissues.

We showed here that the AMPK-related effects of ghrelin on peripheral tissues require the presence of CB1 receptor, as the tissue-specific effects of ghrelin on AMPK activity can be blocked by rimonabant in WT mice and are absent in CB1-KO animals. Furthermore, we showed that intracerebroventricular (i.c.v.) ghrelin treatment inhibits liver and white adipose tissue (WAT) AMPK activity, in agreement with a previous study showing that central ghrelin treatment affects lipogenesis in WAT and thermogenesis in brown adipose tissue (BAT) (15–17). Interestingly, this effect of ghrelin also seems to be dependent on the presence of the CB1 receptor, as it is not present in CB1-KO mice. In particular, the central CB1 receptor seems to mediate this effect, as i.c.v. rimonabant treatment antagonized the inhibitory effect of central ghrelin on liver and adipose tissue of WT mice. These data together suggest a tight interaction of the ghrelin and the cannabinoid systems at multiple levels, and it appears that the central cannabinoid system plays a crucial role in mediating the effect of central ghrelin on liver and adipose AMPK activity and, consequently, on glucose and lipid metabolism.

MATERIALS AND METHODS

Animals

The experimental CB1-KO and WT mice were derived from a genotyped stock obtained from the Institut de Recherche et d'Innovation Biomédicale en Haute Normandie (IRIBHN; Université Libre de Bruxelles, Brussels, Belgium; ref. 18) and were bred at the Institute of Experimental Medicine (IEM; Budapest, Hungary). The parent (Belgian) stock was generated from heterozygotes bred for 14 generations on a CD1 (Charles River, L'Arbresle, France) outbred background, with selection for the mutant CB1 gene at each generation. Adult male WT (n=6–8) and age-matched CB1-KO littermates (n=6) weighing 30–35 g, were used in the in vivo experiments. The animals were housed under standard environmental conditions (light between 0600 and 1800, temperature 22±1°C, rodent chow and water available ad libitum). All experimental protocols were reviewed and approved by the Animal Welfare Committee at the IEM.

In the first set of experiments, WT and CB1-KO mice received intraperitoneal (i.p.) injections of 500 μg/kg rat ghrelin (EC₅₀ 2.1 nM, kind donation of Prof. Masayasu Kojima, Kurume University, Kurume, Japan; ref. 19) in a volume of 100 μl (20). In the second set of experiments, WT mice received i.p. injections of 3 mg/kg rimonabant (SR141716; IC₅₀ 13.6 nM, Sanofi-Aventis, Paris, France; refs. 14, 21, 22) or vehicle and 10 min later, with 500 μg/kg ghrelin or vehicle. At 1 h after the treatment, while no food was provided, the animals were decapitated. Tissue samples [heart, liver, subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), BAT, and muscle] were collected in prechilled Eppendorf tubes, and the tissue samples were immediately frozen on dry ice. All samples were stored at −80°C until assayed.

For the third and the fourth sets of experiments, WT and CB1-KO mice were implanted with 26-gauge stainless-steel guide cannula (Plastics One, Roanoke, VA, USA) into the lateral cerebral ventricle under stereotactic control (coordinates from Bregma: anteroventral −0.2; lateral 1.0; dorsoventral 2.0) through a burr hole in the skull (23). The cannula was secured to the skull with “Krazy Glue” (Electron Microscopy Sciences, Fort Washington, PA, USA) and dental cement, and temporarily occluded with a dummy cannula. Bacitracin ointment was applied to the interface of the cement and the skin. Animals were weighed daily, and those showing signs of illness or weight loss were removed from the study and euthanized. For the third experiment, 1 wk after i.c.v. cannulation, WT and CB1-KO mice were divided into 2 groups and received either artificial cerebrospinal fluid (aCSF; 140 mm NaCl; 3.35 mm KCl; 1.15 mm MgCl₂; 1.26 mm Ca Cl₂; 1.2 mm Na₂HPO₄; 0.3 mm NaH₂PO₄; and 0.05% BSA, pH 7.4) or 1 μg of i.c.v. ghrelin in 4 μl aCSF through a 33-gauge stainless-steel internal cannula with 0.5-mm projection between 0900 and 1000 in the light phase. For the fourth set of experiments, the WT i.c.v. cannulation group was divided into 2 groups, and mice received i.c.v. injections, as described above, of 2 μl vehicle (1% Tween 80 in aCSF) or 10 μg rimonabant in 2 μl vehicle. Each group was further divided and, 10 min later, received i.c.v. injections of either 1 μg of ghrelin in 2 μl aCSF or 2 μl aCSF. At 1 h after the last treatment, while no food was provided, the animals were decapitated. Tissue samples (heart, liver, SAT, VAT, BAT, and muscle) were collected in prechilled Eppendorf tubes, and the tissue samples were immediately frozen on dry ice. All samples were stored at −80°C until assayed.

For the determination of the effect of ghrelin on the growth hormone (GH) secretion of WT and CB1-KO (fifth experiment), a jugular vein was catheterized, and blood samples (50 μl) were collected into heparinized tubes containing 100 μl PBS (pH 7.4) containing 0.3% BSA before, 5 min and 15 min after intravenous (i.v.) injection of either 500 ng ghrelin in 50 μl vehicle (0.9% NaCl, 20 U/ml heparin, and 0.05% BSA) or 50 μl vehicle. Samples were centrifuged, and the supernatant was stored frozen for GH measurements.

Primary pituitary culture

The pituitary glands obtained from 12-wk-old male WT and CB1-KO mice were removed under aseptic conditions. The anterior lobe was separated from the neurointermediate lobe and placed into a Petri dish containing SMEM (Gibco-Invitrogen, Carlsbad, CA, USA) and 0.1% BSA and minced into 1-mm³ pieces. Tissue fragments of the anterior lobe were dispersed with trypsin (8 mg/10 ml SMEM), portioned into 10-mm sterile dishes (∼8×10⁵ cells/well), then cultured for 24 h in DMEM (Gibco-Invitrogen) containing 0.1% BSA and 2.5% FCS for 24 h. On the day of experiments, cells were washed with serum-free fresh medium (DMEM and 0.1% BSA), then either incubated with 1 ml test medium (DMEM with 0.1% BSA and 0.01% ascorbic acid, pH 7.4; controls), or exposed to GH-releasing hormone (GHRH; 10⁻⁷ M) and ghrelin (10⁻⁹ M) for 1 h.

AMPK activity assay

The kinase assay for AMPK activity has been previously described (13, 24). Briefly, tissues were weighed and homogenized with Precellys 24 using CK14 tubes containing ceramic beads (Stretton Scientific, Stretton, UK) at 6000 rpm for 1–3 cycles of 20 s in lysis buffer containing 50 mM Tris-HCl, 50 mM NaF, 5 mM Na pyrophosphate, 1 mM EDTA, 10% (v/v) glycerol, 1% Triton X-100, 1 mM DTT, 1 mM benzamidine, 0.1 mM phenylmethane sulfonyl fluoride, and 5 μg/ml soybean trypsin inhibitor, and the tissue protein content was determined using the BCA assay (Pierce, Rockford, IL, USA). AMPK was immunoprecipitated with an equal mixture of AMPK α1 and AMPK α2 antibodies (24), and AMPK activity was determined by the entity of phosphorylation of SAMS, a synthetic peptide substrate of AMPK.

RNA extraction, reverse transcription (RT), and conventional polymerase chain reaction (PCR)

RNA from liver tissues was extracted using the SV RNA extraction kit (Promega, Southhampton, UK). Adipose tissues (100 mg) were homogenized in 1 ml of QIAzol reagent (Qiagen, Crawley, UK), and total RNA was purified according to the manufacturer's instructions, cleaned up, and treated with DNase enzyme (Qiagen). Quality of RNA and total RNA transcription was performed as described previously (25). For conventional PCR, CB1 and CB2 primers (Table 1) were designed with Primer3 (Massachusetts Institute of Technology, Cambridge, MA, USA; http://biotools.umassmed.edu/bioapps/primer3_www.cgi) and were obtained from Sigma-Aldrich (Gillingham, UK). PCR using 2.5 μl cDNA was performed in 25-μl reaction volumes on the GeneAmp 9700 thermal cycler (Applied Biosystems, Foster City, CA, USA) using Qiagen Hotstart PCR kit (Qiagen). An initial denaturing/activating step of 94°C for 15 min for CB1 and for 5 min for CB2 was required, followed by 35 cycles of 94°C for 30 s, an annealing temperature of 60°C for 1 min for CB1, 58°C for CB2, and 72°C for 1 min. A final extension step of 72°C for 7 min was used. PCR products were run on ethidium bromide-stained 2% agarose gels.

Table 1.

Mouse-specific primer details

Primer	Sequence or assay ID	GenBank accession no.
FAS	Mm00662319_m1	NM_007988.3
SREBP1c	Mm00550338_m1	NM_011480.1
HSL	Mm00495359_m1	NM_001039507.1
PEPCK1c	Mm00440636_m1	NM_011044.1
ACTB	4352341E	NM_007393
CB1 forward	5′CTGGTTCTGATCCTGGTGGT3′	NM_007726.3
CB1 reverse	5′TGTCTCAGGTCCTTGCTCCT3′	NM_007726.3
CB2 forward	5′ATGCTGTGCCTTGTTAACTC3′	NM_009924.3
CB2 reverse	5′CTTATCCTTCAGGACCAGAG3′	NM_009924.3

Real-time PCR

Levels of gene expression were quantified using real-time PCR with the mouse-specific assay-by-design primer and probe sets by Applied Biosystems (ABI; Warrington, UK) and the ABI PRISM 7900 Sequence Detector System using mouse-specific primers for sterol regulatory element-binding protein 1c (SREBP1c), fatty acid synthase (FAS), phosphoenolpyruvate carboxykinase 1c (PEPCK-1c), and hormone-sensitive lipase (HSL) (Table 1). Control reactions for RT (containing RNA but no RT enzyme) and PCR (containing PCR mixture but no cDNA) were run together with samples. All gene expression assays have FAM reporter dye at the 5′ end of the TaqMan MGB probe (ABI) and a nonfluorescent quencher at the 3′ end of the probe. The TaqMan MGB probes and primers have been premixed (20×) to a concentration of 18 μM for each primer and 5 μM for the probe. Real-time PCR was performed in a final volume of 10 μl. All of the reactions were obtained in a duplex PCR reaction with β-actin (β-ACTB) as endogenous control (VIC MGB Probe, part. no. 4352341E; ABI) at these conditions: 5 μl TaqMan Universal Master Mix (ABI), 0.5 μl 20× assay mix primer, 0.35 μl 20× Assay Mix ACTB, and 3.5 μl TE. Reaction was run at 50°C for 2 min in the first stage, at 95°C for 10 min in the second and for 40 cycles at 95°C for 15 s and 60°C for 1 min in the third stage. Data were analyzed using the standard curve method. The relative quantities of target transcripts were calculated from duplicate samples after normalization of the data against the housekeeping gene β-actin.

Statistical analysis

Data were analyzed using the Student's t test, ANOVA followed by the Newman-Keuls test or the Kruskal-Wallis test followed by Conover-Inman comparison, or 2-way ANOVA as appropriate, using the StatsDirect program (Ian Buchan, Cambridge, UK) and SPSS 21 (IBM SPSS, Chicago, IL, USA). Significance was taken at P < 0.05. Data are expressed as means ± sem; n = 6–14 in each treatment group, except where otherwise specified.

RESULTS

AMPK activity in WT and CB1-KO mice after peripheral (i.p.) treatment

Liver

WT

In accordance with our previous rat study (13), ghrelin significantly inhibited liver AMPK activity at 77.4 ± 4.36% of control WT mice (P=0.01, Fig. 1A). In rimonabant-treated animals, ghrelin did not show a significant effect (P=0.6), and no significant interaction was seen between ghrelin and rimonabant treatment (P>0.05, 2-way ANOVA).

Figure 1.

Ghrelin and cannabinoid effect on AMPK activity 1 h after i.p. administration of ghrelin (G, 500 μg/kg), rimonabant (R, 3 mg/kg), or a combination of rimonabant and ghrelin (R+G), compared to vehicle-treated (C) animals in WT (n=6/treatment group) and CB1-KO animals (n=8/treatment group) on liver (A), visceral adipose tissue (B), subcutaneous adipose tissue (C), brown adipose tissue (D), and myocardium (E). Data are shown as means ± sem. *P < 0.05, **P < 0.01.

CB1-KO

Ghrelin (87.52±13.4% of control) had no effect on liver AMPK activity (Fig. 1A), and we found no interaction between genotype and treatment (P>0.05, 2-way ANOVA). These data are compatible with the hypothesis that the CB1 receptor is necessary for the ghrelin effect on hepatic AMPK activity.

Adipose tissue

WT

Similar to the liver tissue, AMPK activity was significantly inhibited by ghrelin treatment in VAT (58.5±5.7% of control; P=0.036; Fig. 1B) and SAT (75.4±14.1%; P=0.027; Fig. 1C), while it did not have a significant effect on BAT (120.2±15.4%; P=0.65; Fig. 1D). Rimonabant-treated animals did not show changes in response to ghrelin administration in VAT (P=0.8), SAT (P=0.8), and BAT (P=0.23) (Fig. 1B–D). There was no significant interaction between the two treatments in any of the three types of adipose tissues (P>0.05, 2-way ANOVA).

CB1-KO

Ghrelin did not have an effect on AMPK activity in any of the 3 subtypes of adipose tissue in CB1-KO mice: VAT, 81.2 ± 17.2% of control; P = 0.6 (Fig. 1B); SAT, 103 ± 29.4% of control, P = 0.93 (Fig. 1C); and BAT, 73.4 ± 8.9% of control, P = 0.17 (Fig. 1D), and we found no interaction between genotype and treatment (P>0.05, 2-way ANOVA). These data implicate CB1 in mediating the effect of ghrelin in WAT.

Gene expression in adipose tissue

AMPK activation has important beneficial effects in peripheral lipid metabolism. Many of these AMPK effects are due to its influence on expression of some of the enzymes involved in these processes. Therefore, we studied in WT and CB1-KO animals whether the changes that we observed in AMPK activity after ghrelin treatment were also reflected in gene expression changes of the lipid-metabolizing enzymes. We studied genes involved in fatty acid synthesis (FAS, SREBP1c, and HSL) and in glyceroneogenesis (PEPCK) in SAT. We found a significant effect of genotype (P=0.023), significant effect of treatment (P=0.031) and significant interaction between treatment and genotype (P=0.023), where ghrelin had a significant effect on FAS expression in WT (228.6±28.1% of control; P=0.003, Fig. 2) but not in CB1-KO animals. This would be in agreement with the finding that ghrelin decreases AMPK activity in this tissue. No significant changes were observed for SREBP1c, HSL, and PEPCK1c in the ghrelin-treated WT and CB1-KO animals (Fig. 2). In summary, the effect of ghrelin on FAS mRNA expression in adipose tissue is mediated by the CB1 receptor.

Figure 2.

FAS, SREBP1c, HSL, and PEPCK1c mRNA expression in subcutaneous adipose tissue of CB1-WT and CB1-KO after 1 h i.p. ghrelin treatment. Data are shown as means ± sem; n = 6/WT treatment group, n = 8/CB1-KO treatment group. *P < 0.05, **P < 0.01.

Heart

WT

Ghrelin increased AMPK activity in the heart of WT mice: 178 ± 21.4% of control (P<0.01; Fig. 1E). If animals were treated with rimonabant, there was no significant effect of ghrelin (P=0.17). We found no significant interaction between the ghrelin and rimonabant treatments (P>0.05, 2-way ANOVA). Using RT-PCR, we detected CB1 mRNA expression in heart tissue of WT animals, while no CB2 mRNA expression was seen (data not shown).

CB1-KO

Ghrelin had no significant stimulatory effect on myocardial AMPK activity in the CB1-KO animals (92.9±12.5% of control, P=0.27; Fig. 1E), and we found no interaction between genotype and treatment (P>0.05, 2-way ANOVA). These data suggest that ghrelin's effect on myocardial AMPK activity is mediated by CB1 receptor and is blocked by a CB1 antagonist.

AMPK activity in WT and CB1-KO mice after i.c.v. treatment

Liver

WT

I.c.v. ghrelin treatment significantly inhibited AMPK activity in liver tissue of WT mice (74.6±5.9 of control, P=0.013; Fig. 3A). In central rimonabant-treated animals, ghrelin had no significant effect (93.8±9.5%, P=0.42; Fig. 3A), and no interaction was observed between treatments.

Figure 3.

Effect of ghrelin (1 μg) and rimonabant (10 μg) on liver (A), visceral adipose tissue (B), subcutaneous adipose tissue (C), brown adipose tissue (D), and myocardial (E) AMPK activity in WT and CB1-KO mice 1 h after i.c.v. administration. Data are shown as means ± sem; n = 6/WT treatment group. n = 8/CB1-KO treatment group. *P < 0.05, **P < 0.01.

CB1-KO

I.c.v. ghrelin did not have an effect on liver AMPK activity (93.2±12.3%; P=0.79; Fig. 3A), and no interaction was observed between genotype and treatment. These data, together, with the effect of central rimonabant in WT animals, suggest that CB1 receptor is necessary for the central ghrelin effect on hepatic AMPK activity.

Adipose tissue

WT

Similar to the liver tissue, AMPK activity in VAT and SAT was significantly inhibited by central ghrelin treatment (VAT: 58.7±8.5% of control, P<0.001, Fig. 3B; SAT: 45.1±8.8%, P<0.01, Fig. 3C). Ghrelin did not have a significant effect in BAT (Fig. 3D). Central rimonabant alone did not affect AMPK activity in any of the adipose tissue subtypes analyzed (Fig. 3B–D). Coadministration of i.c.v. rimonabant antagonized the effect of i.c.v. ghrelin in both visceral and subcutaneous tissues, with AMPK activity levels in the ghrelin+rimonabant group being comparable to the control group (89.6±6.3%, P=0.9, Fig. 3B; and 80.2±9.7%, P=0.34, Fig. 3C).

CB1-KO

Ghrelin did not have an effect on AMPK activity in any of the 3 subtypes of adipose tissue (VAT: 157.3±82.7% of control, P=0.39, Fig. 3B; SAT: 127.6±37.1%, P=0.4, Fig. 3C; BAT: 103.2±53.2%, P=0.9, Fig. 3D). These data, together with the effect of central rimonabant in WT animals, indicate that CB1 is the mediator of centrally administered ghrelin's action on VAT and SAT.

Heart

WT

The effect of the central administration of ghrelin on cardiac AMPK activity of WT mice (181.6±38.4% of control) did not reach statistical significance (P=0.14; Fig. 3E). Rimonabant-treated animals given ghrelin did not change AMPK activity (127.3±56.3% of control, P=0.54; Fig. 3E), and no interaction was observed between the treatments.

CB1-KO

I.c.v. ghrelin did not show any significant stimulatory effect on myocardial AMPK activity in the CB1-KO animals (124.6±19.3% of control, P=0.8; Fig. 3E), and no interaction was observed between genotype and treatment.

Effect of ghrelin on GH release

As our current and previously published data suggest, some of the central, as well as the peripheral, effects of ghrelin require the presence of CB1 receptor (14), and we therefore studied the effect of ghrelin on GH release in WT and CB1-KO animals in vivo and in vitro. In the in vivo setting, we found significant effect of treatment on area under the curve for GH (P=0.002, 2-way ANOVA), and there was no significant effect of genotype or interaction between treatment and genotype: WT and CB1-KO animals respond similarly to i.v. ghrelin injection in terms of peak GH release or area under the curve of GH release [WT ghrelin: 40±8 ng/ml/30 min vs. vehicle (8±5), P=0.002; CB1-KO ghrelin: 83±3 vs. vehicle (14±6), P=0.04; WT ghrelin vs. CB1-KO ghrelin, P=0.35; Fig. 4A).

Figure 4.

Ghrelin stimulates growth hormone (GH) release in vivo (A; peak and area under the curve ghrelin vs. vehicle, both P < 0.01; n=6), and in vitro (B n=3). Data are shown as means ± sem. *P < 0.05.

Primary pituitary cell cultures prepared from the pituitary glands of WT and CB1-KO animals showed significant effect of the treatment on GH release (P=0.001; 2-way ANOVA) and no significant effect of genotype and no interaction between genotype and treatment: GH responses to ghrelin treatment (WT vehicle: 64±0.3 ng/ml; WT ghrelin: 123±10; CB1-KO vehicle: 110±19 vs. CB1-KO ghrelin: 164±14; P<0.05 for both), while there was no difference between WT ghrelin vs. CB1-KO ghrelin treatment, suggesting that CB1 is not involved in the GH-releasing effect of ghrelin (Fig. 4B). GHRH was used as a positive control.

DISCUSSION

The present findings suggest a role for the endogenous cannabinoid system in mediating the effects of peripherally and centrally administered ghrelin on liver, adipose, and possibly heart AMPK activity, but not on GH release. AMPK is a key regulatory enzyme of cholesterol and triglyceride synthesis, fatty acid oxidation, glycogen synthesis, and glucose output in liver and of lipogenesis and lipolysis in adipose tissue (9). Numerous elegant studies have proved that the metabolic effects of various hormones and compounds, including leptin, adiponectin, and metformin, are mediated by AMPK (26). We have previously shown that ghrelin and cannabinoid treatments inhibit AMPK in liver and adipose tissues, and we suggested that this is involved in the mechanism for their adipogenic and gluconeogenic effects (13). In the current study, we showed, using genetic deletion or pharmacological antagonism of the CB1 receptor, that the inhibitory effect of peripheral ghrelin on liver and WAT AMPK activity requires the presence of CB1 receptor, while ghrelin had no effect on brown adipose tissue AMPK activity. It is well known that CNS application of hormones can directly affect peripheral tissue metabolism (15, 16). Here, we report that central ghrelin treatment, in low, systemically negligible doses, affects peripheral AMPK activity via the involvement of central CB1, suggesting a CNS-liver/adipose tissue neural link (27). The present findings, on the other hand, do not prove or disprove the possible additional role of peripheral CB1 in the cannabinoid-ghrelin interaction when both are administered systemically.

We have found a decrease in AMPK activity in the liver of WT mice following ghrelin treatment. Rimonabant-treated or CB1-KO animals had no ghrelin response. CB1 receptor mRNA has been shown to be present in the mouse liver (28). Cannabinoid agonists induce the expression of the lipogenic transcription factor SREBP1c and its target enzymes ACC and FAS in the liver. Liver conditional CB1-KO mice are not only resistant to diet-induced steatosis, but also do not develop elevated LDL cholesterol and reduced HDL cholesterol levels (28), in agreement with results of several multicenter phase III studies, which documented substantial improvements of the plasma lipid profile of obese individuals chronically treated with rimonabant (7, 29–31). These data suggest that liver CB1 is important for the effects of a high-fat diet on liver fatty acid synthesis and fatty acid oxidation, and support the hypothesis that the inhibitory effect of peripheral rimonabant on ghrelin treatment is due to an antagonism of liver CB1 receptors, while a possible central effect could also play a role.

Ghrelin inhibited AMPK activity in all types of WAT. Rimonabant-treated animals showed no response to ghrelin, and no effect of ghrelin was seen on adipose tissue of CB1-KO mice. Inhibition of AMPK activity in adipose tissue leads concomitantly to increased lipogenesis (32), possibly through an increase in the expression of lipogenic enzymes and to increased lipolysis. Ghrelin treatment significantly stimulated FAS mRNA expression, and the effect on PEPCK suggests a 50% rise, although this did not reach significance. This could be accounted by the relatively big standard deviation. Ghrelin's effect on FAS was CB1 dependent, reinforcing the idea of a CB1-AMPK-mediated effect on adipose tissue lipid metabolism. The CB1 receptor has been shown to be present in differentiated adipocytes: CB1 stimulation leads to activation of lipoprotein lipase (5), whereas CB1 antagonism, through AMPK activation and up-regulation of e-NOS, increases mitochondrial biogenesis and oxidative metabolism (33). In accordance, chronic rimonabant treatment reduced adiposity in diet-induced obesity (DIO) mice through induction of enzymes of fatty acid β oxidation and the TCA cycle (30), and the decrease in body weight is attributable more to the increase in energy expenditure rather than to a reduction in food intake (34). The role of the endogenous cannabinoid system in the development of obesity has further been confirmed by observation of increased endocannabinoid content in adipose tissue of mice with DIO and in obese humans (35, 36). Central leptin treatment reduces anandamide levels in adipose tissue, and this seems to be important for its inhibitory effect on lipogenesis (37). Insulin treatment reduces AEA and 2-AG in cultured adipocytes and, accordingly, modifies enzymes of EC metabolism, whereas insulin-resistant adipocytes present with opposite findings (36). On the basis of our data, we suggest that the CB1 receptor either centrally or both centrally and peripherally mediates the effect of ghrelin on adipose tissue lipogenesis. These findings, together, with the above studies, suggest that targeting of the CB1 receptor may influence adipogenic processes and that this happens via AMPK-regulated processes.

Thermogenesis and energy expenditure are regulated in BAT. Cannabinoid antagonists favor thermogenesis, and this is at least part of the mechanism by which they affect energy expenditure and body weight. Treatment of DIO mice with a CB1 antagonist, AM251, increased UCP1 and UCP3 in BAT (38), whereas treatment with rimonabant for 40 d reversed the alterations in gene expression levels induced by obesity in WAT and BAT, with a decrease of both WAT and BAT mass (30). However, our current data suggest no effect of acute ghrelin treatment on AMPK activity in BAT of WT mice, but longer treatment might have different effects.

Peripheral ghrelin administration did not have an effect on heart AMPK activity in CB1-KO, while it significantly stimulated it in the wild type. Rimonabant-treated animals did not respond to ghrelin treatment, suggesting a role for CB1 receptor in mediating ghrelin's effect in the myocardium. Diastolic dysfunction associated with myocardial stunning is improved with ghrelin analog treatment (39). This effect is possibly mediated by a stimulatory effect of ghrelin on AMPK activity, as AMPK activation protects cardiac ATP levels and reduces infarct size and damage to myocytes during ischemia (40). The effects of cannabinoids on the myocardium are controversial, with previous studies suggesting possible negative effects, while more recent studies supporting the beneficial effects of CB1 antagonism (41–45). To date, clinical studies with this compound have not reported any negative effects on the heart. However, novel nonbrain-penetrant CB1 antagonists and inverse agonists (46, 47) should be studied for cardiac effects in the future. In addition, it will be interesting to look into the role of AMPK in possibly mediating the effects of cannabinoids on the myocardium.

We have identified an effect of i.c.v. ghrelin on liver and WAT AMPK activity. In i.c.v. rimonabant-treated animals, ghrelin had no effect, suggesting that brain CB1 receptors may play a crucial role in this CNS-peripheral tissue metabolism circuit. The neuronal link between the hypothalamus and the periphery is implicated in the regulation of peripheral metabolism. Elegant studies have delineated a role for hypothalamic AMPK in this complex neuronal circuit. Hypothalamic AMPK, possibly via the sympathetic nervous system, mediates the effect of leptin on skeletal muscle (48). I.c.v. injection of adiponectin results in decreased energy expenditure, possibly as a result of a reduced expression of UCP1 in the BAT (49). Central α-lipoic acid, which inhibits hypothalamic AMPK activity, increases energy expenditure and UCP1 expression in BAT, and this effect is prevented with i.c.v. treatment with AICAR, an AMPK activator (50). When α2 AMPK is specifically knocked out in hypothalamic POMC neurons, the mice, contrary to expectations, have increased body weight and reduced energy expenditure as shown by reduced UCP1 and PPARγ coactivator 1α in BAT (51).

In our study, acute central ghrelin treatments affected AMPK activity in liver and white adipose tissue but not in BAT. These data support the important role of central regulation of liver and adipose tissue metabolisms (15, 16, 17, 52). Recent data suggested that central ghrelin administration can increase adiposity independently of appetite effects (16).

Central chronic ghrelin treatment influences adipose tissue metabolism (15), as it increases glucose utilization rate of WAT and BAT and stimulates the expression of enzymes involved in lipid synthesis. These effects were independent of the ghrelin-induced hyperphagia and seemed to be mediated by the sympathetic nervous system. The effect of central ghrelin treatment on peripheral metabolism, via the sympathetic nervous system, similar to other hormones, could be mediated by the stimulatory effect of ghrelin on hypothalamic AMPK activity. Central cannabinoids could also be part of this neuronal circuit and mediate the effect of central ghrelin on sympathetic nervous system, as i.c.v. rimonabant was able to block the effect of ghrelin on the AMPK activity of liver, adipose tissue, and heart.

The lack of significant effect on BAT could be due to both short treatment time, as all the previous studies reported effects of chronic ghrelin treatments (6 d, 2 wk, or 6 wk) and to the low dose of ghrelin used in our study. Chronic central ghrelin treatment in high doses stimulated insulin-induced glucose uptake and reduced expression of UCP1 (15), suggesting an effect of the hormone on energy dissipation. Chronic peripheral ghrelin treatment has been shown to decrease UCP1 mRNA expression in BAT after 7 d of treatment (53), whereas 2 and 8 wk ghrelin treatment had opposite effects on UCP1 and β3-adrenergic receptor mRNA expression in BAT in mice subjected to gastrectomy, with no effect in sham-treated mice (54).

Central administration of ghrelin caused a mean increase in AMPK activity of 184% in the heart, but under our current conditions, and for a number of animals included in the study, this did not reach significance. Therefore, we cannot draw firm conclusions regarding the central ghrelin effects on cardiac tissue.

AMPK is a crucially important metabolic enzyme, and the relative changes in AMPK activity are relatively small in experimental settings, as shown by publications from various laboratories. Therefore, one of the limitations of our study is that it did not allow a full analysis of the interaction between the various treatments in our complex experimental settings. Although we were able to see a clear effect of ghrelin treatment in the WT animals and lack of effect in the CB1-KO or rimonabant-treated animals, we limited our comparisons and our conclusion regarding these effects. Another limitation of our study is that following previous published and preliminary studies with various drug doses, we identified the administered ghrelin and rimonabant doses to perform the full extensive treatments and analysis, and we do not have dose-response curves with either of the drugs administered; therefore, the data can only provide information regarding the experimental settings that we have utilized.

As a number of metabolic effects of ghrelin are mediated by CB1, we wondered whether the GH-releasing effect of ghrelin also involves the CB1. As ghrelin releases GH via a dual hypothalamic and direct pituitary effect (55), we studied i.v. ghrelin injection, as well as in vitro primary pituitary cell culture in CB1-KO animals. Our data suggest that ghrelin's effect on GH is independent of CB1. The effect of rimonabant was studied on ghrelin-induced GH release and a reduction of GH levels was identified (56). The discrepancy between the two studies could be explained by species differences (rats vs. mice) or that compensatory mechanisms are present in CB1-KO animals supporting a normal response to ghrelin injection, which are not operating at acute CB1 blockade elicited by rimonabant.

In summary, the effects of ghrelin on AMPK activity in peripheral tissues, similarly to the hypothalamus, are CB1 dependent, suggesting a close interaction of the two systems in the regulation of metabolic pathways.