Abstract
Ghrelin enhances feeding by activating the growth hormone secretagogue receptor (GHSR). In the brain, GHSRs are expressed in regions responsible for regulating food motivation including the ventral tegmental area (VTA). Endogenous cannabinoids also promote food-seeking behaviors through the cannabinoid receptor-1 (CB-1Rs) in brain regions including the VTA. It is not known, however, if ghrelin and endocannabinoids interact in the VTA to produce these effects. We therefore examined if GHSR and CB-1R interact within the VTA to enhance food motivation. Results show that GHSR and CB-1R mRNA are expressed in the VTA cells in male and female rats and mice, with the GHSR being expressed in dopamine cells and the CB-1R being expressed primarily in nondopaminergic cells with no obvious sex differences. Ghrelin directly activated and increased excitatory tone onto dopamine cells of male and female mice. Male rats lacking fully functional GHSR signaling showed disrupted gene expression of transcripts important for regulating the synthesis, release, and degradation of endocannabinoids and lowered the levels of 2-arachidonoylglycerol (2-AG) within the VTA. Moreover, pharmacological antagonism of VTA CB-1Rs attenuates the orexigenic and appetitive effects of intra-VTA ghrelin in rats and blocks the ability of ghrelin to promote excitatory drive to VTA dopamine neurons. Finally, blocking the breakdown of cannabinoids in the VTA enhances the effects of ghrelin on food motivation. Together, our data show that ghrelin stimulates VTA dopamine cells and ultimately food motivation in part through a mechanism that involves endocannabinoid signaling at the CB-1R.
Significance Statement
The current paper utilized a molecular, cellular, and behavioral set of approaches to explore the interaction between the ghrelin and endocannabinoid systems in dopamine cells in the ventral tegmental area (VTA). Our results show that ghrelin activates VTA dopamine neurons and increases food motivation in rodents in part by recruiting the endocannabinoid system. Blocking the cannabinoid receptor-1 in the VTA attenuates the effects of ghrelin on reward-seeking behaviors and excitatory tone onto VTA dopamine neurons. Overall, these data support the hypothesis that ghrelin interacts with endogenous cannabinoids to stimulate dopamine cells in the VTA.
Introduction
Ghrelin is a hormone secreted primarily by the stomach and known for its potent orexigenic effects (Kojima et al., 1999; Tschop et al., 2000; Nakazato et al., 2001). Plasma ghrelin levels are higher in fasted animals and humans and are usually entrained to peak in anticipation of meals (Cummings et al., 2001; Blum et al., 2009). Peripheral and central injections of ghrelin result in a rapid onset of feeding in laboratory animals and strong feelings of hunger in human participants (Tschop et al., 2000; Nakazato et al., 2001; Wren et al., 2001). Moreover, continuous ghrelin delivery to laboratory animals results in weight gain, primarily due to increases in adiposity, and this is related to increased carbohydrate use as an energy source while sparing, and therefore storing, fat (Tschop et al., 2000). The effects of ghrelin on feeding and energy balance are mediated by ghrelin binding the growth hormone secretagogue receptor (GHSR), a receptor found in several hypothalamic nuclei (Howard et al., 1996; Guan et al., 1997; Nakazato et al., 2001; Zigman et al., 2005). In addition to the hypothalamus, the GHSR is also expressed in other brain regions including regions important for the regulation of positive and negative valence (Guan et al., 1997; Zigman et al., 2006). One of these is the ventral tegmental area (VTA), a heterogeneous midbrain region associated with reward-seeking behaviors (Wise, 2004; Koob and Volkow, 2016). The VTA contains neurons that release several neurotransmitters including dopamine, GABA, and glutamate (Morales and Margolis, 2017). Dopamine neurons have classically been implicated in reward processes, including food-seeking and other reward-seeking behaviors (Wise, 2004; Koob and Volkow, 2016). Interestingly, 50–60% of dopamine cells in the VTA express the GHSR, and these cells increase their firing rate in response to ghrelin (Abizaid et al., 2006). Ghrelin also stimulates the release of dopamine into the NAc, an event correlated with increased dopamine release (Abizaid et al., 2006; Jerlhag et al., 2006). Importantly, ghrelin delivery into the VTA increases feeding and promotes behaviors associated with increased motivation to obtain food, especially palatable foods (Naleid et al., 2005; Abizaid et al., 2006; Egecioglu et al., 2010; Skibicka et al., 2011; St-Onge et al., 2016).
Endogenous cannabinoids produce an orexigenic effect similar to the orexigenic effects of ghrelin [for review, see Sharkey and Pittman (2005); Horvath (2006); Lau et al. (2017)], and this similarity has led some to suggest that the ghrelin and cannabinoid systems interact to increase appetite. Indeed, there is accumulating evidence suggesting that these two systems do interact (Kola and Korbonits, 2009; Edwards and Abizaid, 2016). For instance, mice lacking the GHSR fail to increase food intake in response to cannabinoids (Lim et al., 2013). Similarly, mice lacking the cannabinoid receptor-1 (CB-1R) do not increase food intake in response to peripheral injections of ghrelin (Kola et al., 2008). In the hypothalamus, both ghrelin and cannabinoids stimulate intracellular kinase signals associated with negative energy states (Kola et al., 2008; Kola and Korbonits, 2009). Mice exhibiting genetic elevations in endocannabinoid signaling exhibit enhanced sensitivity to the orexigenic effects of ghrelin as well as increased ghrelin-mediated activation of AMPK in the hypothalamus (Kola et al., 2008; Kola and Korbonits, 2009; Balsevich et al., 2023). There is also evidence suggesting that cannabinoids increase ghrelin secretion from the stomach and increasing peripheral ghrelin can stimulate dopamine release from the VTA to increase appetite (Zbucki et al., 2008; Senin et al., 2013).
Like the GHSR, the CB-1R is expressed in the VTA of rats and mice (Guan et al., 1997; Zigman et al., 2006; Parsons and Hurd, 2015; Liu et al., 2020). To date, however, little is known about the interaction of ghrelin and the endocannabinoid system within the VTA to modulate food-seeking behaviors. In this study, we investigated the contribution of endogenous cannabinoids on the effects of ghrelin on VTA dopamine cell activity and food motivation.
Materials and Methods
Animals
All procedures were approved by the Carleton University Animal Care Committee and the Animal Welfare Committee at the Institute of Experimental Medicine of the Hungarian Academy of Sciences. All closely followed the guidelines of the Canadian Council for Animal Care. Adult Long–Evans rats weighing 300–350 g were procured from Charles River Laboratories to be used in the behavioral studies. Adult FHH-Ghsrm1Mcwi rats and their wild-type (WT) fawn–hooded hypertensive (FHH) littermates were used in the qPCR and cannabinoid content studies. These rats contain an ENU-mediated mutation that produces a GHSR sequence that is truncated at Position 343 resulting in a premature stop codon 22 amino acids short of the full sequence and ultimately results in disrupted GHSR signaling (Bulbul et al., 2011; Chebani et al., 2016; MacKay et al., 2016). Rats were generated at the Medical College of Wisconsin, purchased from Transposagen Biopharmaceuticals and bred at Carleton University animal facilities. All FHH-Ghsrm1Mcwi and WT littermates used weighed between 350 and 400 g at the time of sacrifice. For the electrophysiology studies, we used Egfp-L10a mice, a line engineered with a floxed transcriptional stop cassette in front of a ribosomal protein L10-EGFP reporter construct (Jackson Laboratory; JAX ID, 024750) mated with heterozygous Thcre mice (Jackson Laboratory; JAX ID, 008601). These mice were also bred at Carleton University animal facilities. Animals were housed in pairs or groups of three until experiments began. All animals were housed under standard laboratory conditions (room temperature 22°C, humidity 45–55%) and had access to standard ad libitum rodent chow and tap water. Finally, Long–Evans rats purchased from Charles River Laboratories and TH-Egfp-L10a mice were used for the in situ hybridization/immunocytochemistry studies.
Drugs
Ghrelin and SR 141716A (rimonabant) were purchased from Tocris Bioscience. We purchased AM251 (CB-1 antagonist), MJN10 [monoacylglycerol lipase (MAGL) inhibitor], and PF-04457845 [fatty acid amide hydrolase (FAAH) inhibitor] from Cayman Chemical. Ghrelin was mixed in sterilized 0.9% isotonic saline at the desired doses. All other compounds were mixed in 10% DMSO mixed in 0.9% saline to the desired dose dilution.
Double in situ hybridization/immunohistochemistry
Adult male and female Long–Evans rats (N = 3) and Th-cre;L10-Egfp male and female mice (N = 3) were deeply anesthetized with isoflurane (5% in oxygen) and perfused with 100 ml of 0.9% saline followed by 200 ml of 4% paraformaldehyde (PFA) mixed in 0.1 M PB. Brains were collected, postfixed in 4% PFA for 24 h (4°C), and then placed in 30% sucrose until brains sank. Brains were sliced to obtain four sets of 30 μ sections using a cryostat (Thermo Shandon). VTA sections from one set from each rat were processed for in situ hybridization to detect GHSR and CB-1R mRNA and immunocytochemistry to detect tyrosine hydroxylase (TH). In short, tissue sections were washed and mounted onto polylysine-coated slides. Once these dried, slides were processed for dual fluoresce and in situ hybridization using RNAscope probes specific for the gene encoding rat ghsr (catalog #480031) and cnr1 (catalog #412501) genes (Advanced Cell Diagnostics) following the RNAscope multiplex fluorescence protocol. The ghsr probe was labeled with an Opal 520 fluorescent tag, whereas the cnr1 probe was labeled with an Opal 650 tag (Akoya Biosciences). At the end of the in situ protocol, slides were washed in buffer (0.1 M PB) and incubated in a solution containing a mouse anti-TH antibody (1/10,000; ImmunoStar) and tagged using a horse anti-mouse Alexa Fluor 488 secondary antibody. Slides were rinsed and coverslipped with aqueous mounting media containing DAPI to visualize cell nuclei. For mouse tissues, the protocol was the same with the exception of using mouse specific RNAscope probes to detect mouse ghsr (catalog #426141) and cnr1 (catalog #457341). These mice showed GFP expression in TH neurons, so these were not processed for TH immunocytochemistry. Combined in situ/immunocytochemistry-labeled sections were imaged using a Nikon C2 confocal system fitted with a Plan Apochromat 10× objective scanning.
Immunohistochemical detection for GFP and TH
Brain sections containing the VTA of TH-Egfp-L10a male and female mice were collected to demonstrate the colocalization of GFP and TH. Briefly, sections were washed several times in 0.1 M PB and incubated for 48 h in a primary antibody solution containing a chicken anti-GFP (Abcam, 1:1,000) and a mouse anti-TH (ImmunoStar, 1:10,000). Sections were washed 3–5 times and incubated in a secondary antibody solution containing an Alexa Fluor 488-conjugated donkey anti-chicken (1:250) and an Alexa Fluor 598-conjugated donkey anti-mouse (1:250) antibodies. Sections were washed and coverslipped in aqueous media containing DAPI (VectorLabs). Images were visualized using the same Nikon C2 confocal system as above to determine double labeling for GFP and TH.
Ultrastructural detection of the CB-1R
Adult, male CD1 mice were deeply anaesthetized and perfused transcardially with 10 ml 0.01 M phosphate-buffered saline (PBS), pH 7.4, followed sequentially by 10 ml of 4% PFA in Na-acetate buffer, pH 6.0, and 50 ml of 4% PFA in borax buffer, pH 8.5. The brains were rapidly removed and postfixed in 4% PFA in 0.1 M PB, pH 7.4, for 24 h at 4°C. Serial 25-μm-thick coronal sections were cut on a vibratome (Leica Microsystems). Sections from the VTA were treated with 0.5% H2O2 in PBS for 15 min; then, sections were cryoprotected in 15% sucrose in PBS for 15 min at room temperature and in 30% sucrose in PBS overnight at 4°C. After cryoprotection, the sections were quickly frozen over liquid nitrogen to improve antibody penetration into the tissue. Sections were put into goat antiserum to CB1 [directed against the C-terminal 31 amino acids (443–473) of mouse CB-1R (1:1,600); Fukudome et al., 2004] for four nights at 4°C; then they were incubated overnight in biotinylated donkey anti-sheep IgG (1:500). The immunoreaction was detected by Ni-DAB and amplified by silver-intensification using the Gallyas method (Gallyas, 1979). After detection of immunolabeling, sections were transferred into 0.1 M PB and treated in 1% osmium-tetroxide in 0.1 M PB for 30 min and then washed in 50% ethanol for 2 min and 70% ethanol for 1 min. After treatment in 2% uranyl-acetate in 70% ethanol for 30 min, sections were dehydrated with 70, 90, 96, and 100% ethanol and with acetonitrile for 2 min, embedded in Durcupan ACM epoxy resin (Fluka) and polymerized at 60°C for 2 d. Then, 60–70-nm-thick ultrasections were cut with an ultramicrotome (Leica Microsystems) and examined with a JEOL electron microscope.
Measurement of cannabinoid content from the VTA of FHH-GHSRm1/Mcwi rats
For endocannabinoid quantification experiments, rats were killed by rapid decapitation to minimize distortion of physiological endocannabinoid concentrations that are known to spike during stress (e.g., CO2 asphyxiation) and shortly after death (Buczynski and Parsons, 2010). Brains were immediately extracted from the skull following decapitation and flash frozen on dry ice. A 1 mm coronal brain slice, containing the VTA, was cut, and the VTA was microdissected out with the aid of a dissection microscope. VTA samples were transferred to Eppendorf tubes and stored at −80°C until processed for LC–MS as described previously (Qi et al., 2015). In short, preweighed frozen VTA samples from FHH-Ghsrm1Mcwi and their WT littermates were homogenized with a glass rod in borosilicate glass tubes containing 2 ml of acetonitrile, 186 pmol of [2H8]2AG (2-arachidonoylglycerol), and 84 pmol of [2H8] anandamide. Samples were sonicated for 30 min, incubated at −20°C overnight to precipitate proteins, and spun at 1,500 × g, and supernatants were transferred to new glass tubes. Samples were dried under N2, rinsed with methanol to minimize lipid loss, and dried again under N2. Twenty microliters of methanol were used to resuspend lipid extracts containing the endocannabinoids. Levels of 2AG and AEA were subsequently separated, identified, and quantified via isotope-dilution LC–MS.
Quantitative real-time polymerase chain reaction (RT-qPCR) measures
FHH-Ghsrm1Mcwi and their WT littermates were rapidly decapitated, and their brains were flash frozen and stored at −80°C until processed for qPCR. To do this, we punched the VTA from 300 μ sections collected with a cryostat using the Palkovitz technique (Palkovits, 1973). Once punched, VTA samples were homogenized in TriZol reagent (Invitrogen) to encourage release of cell contents and dissociation of nucleoprotein complexes. Chloroform was added to each sample prior to centrifugation (at 12,000 × g for 15 min) to separate the RNA aqueous phase from the DNA and protein containing organic phases. Isopropyl alcohol and linear acrylamide were added to each sample to facilitate RNA precipitation from the aqueous phase upon centrifugation (at 12,000 × g for 15 min). RNA pellets were washed in 75% ethanol and dissolved in RNase-free water. The extracted RNA was kept at −80°C until required for reverse transcription complementary DNA (cDNA) synthesis. The concentration and quality of extracted RNA samples were assessed using a Thermo Fisher Scientific NanoDrop 100 spectrophotometer (Fleige and Pfaffl, 2006). Only RNA of sufficient quality and concentration were reversed transcribed into cDNA using a SuperScript II Reverse Transcriptase (SSII RT) kit and the method provided by the manufacturer (Life Technologies). Briefly, oligo (dT) primers (Invitrogen) were mixed with diluted RNA samples (5,000 ng of RNA per sample) and then heated for 5 min at 70°C. An aliquot of master mix containing 5× first strand buffer (Invitrogen), dithiothreitol (Invitrogen), RNase inhibitor (Promega), deoxynucleotide triphosphate (Invitrogen), DEPC water, and SSII RT (Invitrogen) was added to each sample as specified by the manufacturer (Life Technologies). Samples were run in a MJ Research PTC-200 Thermal Cycler (Marshall Scientific) at 42°C for 1.5 h. A final 10 min 90°C cycle was used to inactivate the reverse transcriptase. cDNA samples were stored at −20°C until required for RT-qPCR experiments. The nucleotide sequences of all qPCR primers used are included in Table 1. All primer pairs were tested for their amplification efficiencies using a standard curve. Only primer pairs which fell between 90 and 110% efficient were utilized for RT-qPCR experiments. Primers, cDNA (diluted to a concentration within the linear dynamic range of the primer pair standard curve), Milli Q H2O, and iQ SYBR Green PCR Super Mix were combined according to the method of the manufacturer (Bio-Rad Laboratories). All samples including nontemplate and nonreverse transcription controls were run in triplicates using primers for the gene of interest as well as corresponding GAPDH and Actb housekeeping genes. PCR plates were run on a CFX Connect TM Real-Time PCR machine (Bio-Rad Laboratories) using the following program (with sight modification to the annealing temperature depending on the melting temperature of primers): 30 s 95°C step, 40 cycles of denaturing (10 s at 95°C), annealing (30 s at ∼55°C), and extension steps (20 s at 72°C) and a final 1 min 95°C step. RT-qPCR data were analyzed using the 2-ΔΔCt method (Livak and Schmittgen, 2001).
RT-qPCR primer sequences
Intra-VTA cannula surgical implantation
Male Long–Evans rats were anesthetized with an isoflurane:oxygen gas mixture of 5:2 for induction and 2.5:2 for maintenance. Metacam (5 mg/ml) was administered intraperitoneally (0.5 mg/kg) prior to the initiation of surgery to ensure adequate analgesia during and postsurgery. Scalps were shaved and cleaned with germistat, alcohol, and proviodine to maintain an aseptic canvas around the incision site, and tear gel was applied to prevent dehydration of the eyes. Rats were secured in a stereotaxic apparatus (Kopf Instruments) and adequate anesthetic depth was confirmed before a midline incision was made. Hemostats were used to retract the skin, and the anterior to posterior (AP −5.6 mm) and medial to lateral (ML +2.0 mm) coordinates of the VTA relative to the bregma were measured and delineated on the skull. A surgical drill was used to bore holes in the skull for implantation of the unilateral guide cannula and associated anchoring screws. Anchoring screws were affixed, and the guide cannula, angled toward the midline (10°) to avoid the aqueduct, was inserted to a depth just above the VTA (DV −7.8 mm). Dental cement was molded around the cannula and anchoring screws to stabilize and prevent movement of the guide cannula throughout behavioral experiments. Once dry, the incision was sutured, and a dummy cannula was inserted into the guide cannula to mitigate the chance of clogging. Rats were placed in freshly clean cages and monitored closely for at least 2 h after surgery for signs of surgical complications. Rats were treated postoperatively with subcutaneous Metacam (0.5 mg/kg) for a minimum of 2 d after surgery and were monitored twice daily for a least a week to ensure optimal recovery.
Operant training
Operant conditioning and experiments were conducted in standard operant conditioning chambers (height × width × depth = 13 × 10 × 12 in; Coulbourn Instruments) equipped with two levers (i.e., active and inactive), a pellet dispensing hopper, a house light, and a grid floor. Active and inactive lever presses as well as overall locomotor activity were recorded by the Graphic State (version 3) software (Coulbourn Instruments). For operant conditioning training and experiments, rats received a standard chow-like 45 mg Dustless Precision Pellet (Bio-Serv; 3.6 kcal/g; 5.6% fat; 59.1% carbohydrates; 18.7% protein) upon satisfying the scheduled number of active lever presses required to obtain a reward.
Upon arrival, rats were allowed to habituate to the laboratory facilities before food intake and body weight baseline measurements were recorded. They were then food-restricted to 90% of their baseline body weight to facilitate operant training. Rats were first trained to bar press on a fixed-ratio 1 (FR-1) schedule where each depression of the active lever resulted in the delivery of a single food pellet. These training sessions were 30 min in duration and were conducted during the first half of the light cycle. Rats received their respective training session at the same time for the duration of operant conditioning training. Once their responding stabilized (i.e., <15% variation in active lever presses between three consecutive training sessions), they were moved up to a FR-3 schedule, where three active lever presses were required to obtain the food rewards. Similarly, once rats were stably responding on a FR-3 schedule, they were graduated to a FR-5 schedule, where five active lever presses were required to obtain food rewards. On average, rats required 7–10 d to graduate from a FR-1 to an FR-3 and 3–5 d to reach the FR-5 schedule. After reaching FR-5 responding stability guidelines, rats were given ad libitum food and water for 3 d prior to their stereotaxic surgery (i.e., implantation of guide cannula above the VTA). Rats were given a minimum of 7 d to recover from surgery; however, they were not readmitted to operant training procedures until sufficiently recovered. Once recovered, they were again food-restricted to 90% of postsurgery recovery weight (i.e., Day 7) and given four FR-5 training sessions to ensure that the surgery did not disrupt their capacity to bar press. Rats with no surgical complications and stable FR-5 responding rates were subjected to mock microinfusions and given one practice PR training session to acclimate them both to the microinfusion procedure and to a schedule where the number of active lever presses required to obtain food rewards progressively increases. The PR schedule used (i.e., 1, 2, 4, 6, 9, 11, 15, 20, 25, 32, 40, 50, 62, 77, 95, 118, 145, 178, 219, 268, etc.) was developed by Richardson and Roberts (1996) to assess the efficacy of a reinforcer (e.g., drugs) at promoting motivated behaviors. The schedule was derived from the following equation: response ratio (rounded to nearest integer) = [5e (number of rewards obtained × 0.2)] − 5 (Richardson and Roberts, 1996). Practice and final test PR sessions ended when rats failed to obtain their next reward within 30 min of their previous reward. The amount of food rewards obtained before a rat gave up and timed out was defined as their breakpoint and was taken as an index of how motivated a rat was to work for food rewards. Rats were sorted into four groups after their practice PR session so that the average number of active lever presses across the last three FR-5 training sessions and within the practice PR session did not differ between treatment groups. After the PR practice session, rats were given ad libitum food and water for 3 d prior to their final PR test. In experiments where rats were given intra-VTA infusions of rimonabant and ghrelin, rats were assigned to one of the following treatments: rimonabant (0.5 μg)/ghrelin (1 μg), rimonabant (0.5 μg)/saline, vehicle/ghrelin (1 μg), or vehicle/saline. These treatments were given spaced 30 min apart. In the study where rimonabant was delivered peripherally, rimonabant was injected intraperitoneally at a dose of 1 mg/kg mixed in 0.9% isotonic saline 30 min before the intra-VTA ghrelin infusions at the same dose as above. Finally, we also infused a threshold dose of ghrelin (0.5 μg/μl of saline) preceded by infusions of MJN110 and PF-04457845, inhibitors of the cannabinoid-degrading enzymes MAGL and FAAH (6 μg/μl of DMSO). Immediately following the second infusion, rats were placed in the operant chamber for their final PR test session to assess treatment-induced changes in active lever pressing.
Histological analyses of cannula placement
Upon completion of behavioral experiments, rats were overdosed with urethane (1 g/kg) and killed via transcardial perfusion. Isotonic saline was circulated to clear the blood before circulation of a 4% PFA fixative solution. Following fixation, rats were quickly decapitated, and brains were extracted. Brains were incubated in 4% PFA for an additional 48 h and then transferred to a 30% sucrose solution: 0.1 M phosphate buffer cryoprotectant solution (weight/volume). Brains were left in cyroprotectant at 4°C until adequately dehydrated (2–3 d). Following dehydration, brains were frozen and sliced coronally on a CM1900 cryostat (Leica Biosystems) into 50 μM sections. The position of the tip of the cannula in each rat brain was noted and classified as either falling within or out of the delineations of the VTA. Data from rats with missed cannula were excluded from all analyses. Rats that did not recover well from surgeries (i.e., which lost a significant amount of weight, seemed lethargic, were anorectic) were killed.
Electrophysiology
TH-EgfpL10 male and female mice (4–8 weeks old) were anesthetized, and their brains were rapidly dissected out and immersed into cold (2–4°C), carbogenated (95% O2, 5% CO2) slicing solution (297–302 mOsm/L), pH 7.4, containing the following (in mM): 118 NaCl, 3 KCl, 1.3 MgSO4, 1.4 NaH2PO4, 5 MgCl2, 10 D-glucose, NaHCO3, and 0.5 CaCl2 as previously described (Spencer et al., 2024). In brief, brains were blocked and mounted on a stage submerged in slicing solution to facilitate either coronal or horizontal slicing on a vibratome (Leica VT 1000S, Leica Biosystems). Coronal sections (250 μm) containing the VTA were cut and transferred to a warmed (37°C), carbogenated artificial cerebrospinal fluid bath solution (297–302 mOsm/L), pH 7.4, containing the following (in mM): 124 NaCl, 3 KCl, 1.3 MgSO4, 1.4 NaH2PO4, 10 D-glucose, 26 NaHCO3, and 2.5 CaCl2. Slices were incubated for 10 min at 37°C and then recovered at room temperature (21–23°C) for at least 1 h before use for patch-clamp recordings.
At this point, brain slices were placed in a slice chamber of a fixed stage (Gibraltar, ThorLabs) that was continuously superfused (2–2.5 ml/min, Masterflex CL Cole Palmer pump) with carbogenated bath solution. This solution was warmed by a temperature controller (Warner Instruments) such that it entered the recording chamber between 32 and 34°C. A homemade platinum and nylon fiber harp was used to keep slices immobilized during recordings. Slices were illuminated with a Nikon power supply and visualized with an upright Eclipse E600FN microscope (Nikon) affixed with an Axio iCM1 camera (Carl Zeiss) sitting on a microscope translator (Gibraltar). Filamented borosilicate glass patch pipettes were pulled to yield pipettes with resistances between 5 and 8 MΩ when backfilled with a potassium gluconate-based internal solution (285–288 mOsm/L), pH 7.24, containing the following (in mM): 120 K-gluconate, 10 KCl, 10 HEPES, 1 MgCl2, 1 EGTA, 4 MgATP, 0.5 NaGTP, and 10 phosphocreatine. This potassium-based solution was used for all current-clamp recordings and voltage-clamp recordings of glutamatergic events held at −60 mV. Pipettes were filled with a cesium methanesulfonate-based internal solution (285–288 mOsm/L), pH 7.24, containing the following (in mM): 125 CsMs, 11 KCl, 10 HEPES, 1 CaCl2, 1 EGTA, 4 MgATP, and 0.5 NaGTP to recording GABAergic events in voltage-clamp experiments held at −5 mV.
Whole-cell patch–clamp recordings were made with pipettes connected to a head stage of an Axopatch 200B Amplifier (Axon Instruments) filtered at 2 kHz. For all experiments, a 10 min baseline period of recording was collected to ensure cell stabilization before drug treatments were applied. Similarly, all cells were minimally given a 15 min washout period after drug treatment completion to examine if drug-induced effects were reversible.
Resting membrane potential and action potential continuous trace recordings were obtained in current clamp without injection of current, except when cells were over- (action potential frequency of 4 Hz or greater) or underactive (very low resting membrane potential). In these instances, a small amount of negative or positive current was injected to prevent cell fatigue or to bring these cells closer to their firing thresholds, respectively.
The membrane potential of cells was sampled every second using Clampfit 10.7 (Axon Instruments) and averaged into 30 s bins. Resting membrane potential changes (Δ RMP) of VTA dopaminergic neurons relative to a 5 min baseline period (immediately before ghrelin application) was calculated for each current-clamp recording. A time course of the Δ RMP across cells was graphed to facilitate examination of drug-induced changes in resting membrane potential across time. In addition, representative Δ RMP means (across cells) were calculated and graphed for the baseline, rimonabant, ghrelin, rimonabant, and ghrelin and washout periods to summarize treatment-induced changes in resting membrane potential.
An action potential template was created in Clampfit 10.7 (Axon Instruments) to detect the number of action potential events that exceed 0 mV in each of our current-clamp recordings. The number of action potentials was organized into 30 s bins, and action potential frequencies were calculated. A time course of action potential firing frequencies across cells was likewise graphed across time to examine drug-induced changes in the firing rate of these VTA dopamine neurons. Once again, representative action potential frequency means were calculated and graphed for each treatment condition.
Spontaneous excitatory and inhibitory postsynaptic currents (sEPSCs and sIPSCs, respectively) appeared as downward (sEPSCs) and upward deflections (sIPSCs) in voltage-clamped continuous recordings. These deflections were recorded in separate sets of cells. Mini Analysis (Synaptosoft) was used to detect the number of sEPSCs and sIPSCs present in voltage-clamp recordings. The number of events was placed into 30 s bins, and event frequencies were calculated. sEPSC and sIPSC frequencies were averaged across cells and graphed across time to facilitate detection of treatment-induced changes in excitatory and inhibitory tone onto dopaminergic neurons of the VTA.
Data from both voltage- and current-clamp experiments were collected and integrated using a Digidata 1322A (Molecular Devices) digitizer connected to a computer running the Clampex 10.7 software (Axon Instruments). Current and voltage data were analyzed using Clampfit 10.7 (Axon Instruments) and Mini Analysis (Synaptosoft), respectively. Sample traces were created with Origin 2018 (OriginLab). Statistics and graphs were produced using Prism 7 (GraphPad). Results are represented as the mean ± SEM, with mean differences considered statistically significant at p < 0.05 unless otherwise stated.
Statistical analyses
Statistical assumptions were tested (i.e., normality and sphericity), and appropriate corrections were made when required prior to analysis. Repeated-measure ANOVAs followed by post hoc tests (Fisher LSD) were conducted to analyze differences in feeding and locomotor activity across different timepoints after drug infusion. One-way ANOVAs followed by post hoc tests (Fisher LSD) were used to analyze group differences in operant behavior studies. Independent group t tests were used to determine mean differences in VTA endocannabinoid content and gene expression between GHSR-KO and WT rats. Action potential frequency, Δ RMP, Δ sEPSC, and Δ sIPSCs frequency means were calculated for each treatment condition. For ghrelin only experiments, means were calculated right before ghrelin administration (baseline), at the peak ghrelin effect (ghrelin), and 10 min after ghrelin removal (washout). The same methodology was adopted for experiments when ghrelin was applied in the presence of tetrodotoxin (TTX) and/or CB-1R antagonists (i.e., baseline, ghrelin, and washout measurements were made in the presence of TTX and/or CB-1R antagonists). Treatment means were compared using repeated-measure one–way ANOVAs. Fisher LSD post hoc tests were run, when appropriate, to detect significant differences in these measurements between treatments.
Results
Dopamine neurons within the VTA of rats and mice do not coexpress GHSR and CB-1R mRNA
We first conducted an in situ hybridization study to determine if GHSR and/or the CB-1R mRNA were expressed in dopamine cells within the VTA. We used Th-cre;L10-Egfp mice expressing enhanced green fluorescence protein under the Th promoter; hence putative dopaminergic cells were represented by Egfp expression. We determined if cells expressing EGFP-immunoreactivity coexpressed Ghsr and Cnr1 hybridization and detected Ghsr hybridization throughout the mouse VTA, substantia nigra, and rostral linear nucleus of the raphe (RLI; Fig. 1a). Many VTA cells expressing Ghsr mRNA also expressed EGFP, but we also observed Ghsr mRNA in VTA cells that did not express EGFP. By contrast, Cnr1 mRNA in the VTA was primarily observed in non-EGFP cells, and interestingly, we did not detect Cnr1 hybridization in EGFP cells that expressed Ghsr mRNA in the VTA (Fig. 1b) or substantia nigra (data not shown). By contrast, Ghsr-positive EGFP cells in the RLI frequently expressed Cnr1 hybridization (Fig. 1c).
Sample images obtained from anterior VTA sections from male TH-Egfp-L10a mice (panel a) or Long–Evans rats (panel b) that were processed for double in situ hybridization using RNAscope probes specific for mouse or rat ghsr (red) and cnr1 (pink) transcripts in combination with immunocytochemistry (green) for GFP (mouse) or TH (rat). Higher-magnification images from these tissues show that within the VTA, ghsr expression is observed in dopamine and nondopamine cells and rarely colocalized with cnr1 in dopamine cells. In contrast, both transcripts colocalized in the RLI, a ventral midbrain region that contains dopamine neurons but that does not project to the nucleus accumbens (panels c–f). Electron microscopy analyses of immunogold-labeled CB-1R show that within the VTA, this receptor is localized predominantly (∼66%) of putative presynaptic excitatory synapses (panels g, h) and less frequently in putative inhibitory synapse (∼34%; panels i, j).
This pattern of Ghsr and Cnr1 hybridization at dopaminergic cells was consistent between the mouse and rat. In Long–Evans rats, both Ghsr and Cnr1 were similarly abundant in the VTA and surrounding midbrain regions that comprised dopaminergic cells that expressed TH immunoreactivity (Fig. 1d). As in the mouse, we found Ghsr hybridization at TH-positive and TH-negative cells and TH-labeled cells, but we did not detect Cnr1 mRNA in Ghsr TH cells in the VTA (Fig. 1e). However, the coexpression of Cnr1 and Ghsr was more frequent at TH cells within the RLI (Fig. 1f). Interestingly, both transcripts were coexpressed in non-DA cells within the VTA (Fig. 1c–f) and other midbrain regions.
To examine the ultrastructural localization of CB-1R protein within the VTA, we conducted an electron microscopy study labeling CB-1R in the mouse VTA. CB-1R immunoreactivity was observed primarily on axon varicosities in the mouse VTA (Fig. 1g–j) and was more abundant in the anterior than posterior VTA, where labeled synapses were uncommon. In the anterior VTA, CB1-R immunoreactivity was found in most cases relatively far from the presynaptic side in the varicosities. To elucidate the excitatory or inhibitory nature of the CB1-immunoreactive terminals, we followed the synapses of these terminals through serial sections to determine the synapse type. Based on our analyses of 180 synapses formed by CB1-IR varicosities, 65% of these synapses were asymmetric (Fig. 1g,h), and 35% were symmetric (Fig. 1i,j), thus suggesting that endocannabinoid release has a more predominant impact on excitatory neurotransmitter release than on inhibitory neurotransmitter release in the anterior VTA. Taken together, these data suggested that if the effects of ghrelin on dopaminergic VTA required endocannabinoid signaling, it would not occur through GHSR and CB-1R activation at the same dopamine cells. Rather, our ultrastructural analyses suggested that ghrelin may influence CB-1R activation at excitatory or inhibitory presynaptic inputs to dopamine cells.
Ghrelin delivery into the VTA increases food intake and food reward, and this effect is attenuated by CB-1 receptor blockade
Behavioral evidence suggests an interaction between GHSR activation and the cannabinoids on food motivation (Kalafateli et al., 2018), and the molecular and electrophysiological data described above suggest that these two systems interact within the VTA. To determine whether the effects of intra-VTA ghrelin on food-seeking behavior are mediated by endocannabinoid-dependent mechanisms, we implanted rats with a unilateral cannula positioned at the VTA and delivered either saline or the CB-1 receptor antagonist rimonabant (1 mg/kg) systemically followed by intra-VTA infusion of ghrelin (1 μg/0.5 μl) or saline (0.5 μl). Ghrelin infusion into the rat VTA increased food intake (F(3, 24) = 3.668; p < 0.05; Fig. 2a), but this orexigenic action of ghrelin was attenuated by pretreating the rats with rimonabant intraperitoneally. There were no treatment effects on locomotor activity (p > 0.05; Fig. 2a).
Intra-VTA ghrelin increases feeding and food motivation, and this effect is attenuated by peripheral or intra-VTA CB-1R antagonists. In Figure 1a we show that unilateral intra-VTA ghrelin infusions (1 μg/1 μl) increased food intake (but not locomotor activity) in Long–Evans rats and this effect was attenuated by peripheral injections of rimonabant (1 mg/kg). In panel b, we show that the effects of unilateral intra-VTA ghrelin (1 μg) on food intake and food seeking are attenuated by intra-VTA pretreatment with rimonabant (0.05 μg). Similarly, intra-VTA treatment with ghrelin increased the break point and the number of active lever presses of rats in the progressive ratio task and effect also attenuated by rimonabant treatment. The infusions of ghrelin and/or rimonabant in the VTA did not alter locomotor activity significantly. *p < 0.05; **p < 0.01; ***p < 0.001.
To pinpoint the extent by which local CB-1R signaling within the VTA supported ghrelin-mediated feeding, we infused rimonabant directly into the VTA via a cannula 30 min prior to ghrelin infusion. Expectedly, ghrelin gradually increased food intake over time (Fig. 2b), and this orexigenic effect of ghrelin was long-lasting for at least 4 h (interaction effect; F(9,114) = 1.97; p < 0.05; Fig. 2b). Importantly, intra-VTA infusion of rimonabant suppressed ghrelin-mediated feeding over time. Neither ghrelin nor rimonabant treatment affected locomotor activity (p > 0.05; Fig. 2b).
In rats and mice, the activity of dopamine neurons in the VTA is linked with increased efforts to obtain food reinforcers in a variety of behavioral paradigms like the progressive ratio test (Richardson and Roberts, 1996; Arnold and Roberts, 1997). In this test, rats or mice are first trained to bar press for reinforcers until they are reliably producing a fixed number of bar presses or nose pokes to obtain a food reward. Once this is accomplished, food rewards become available after an increasing number of responses until the effort required is too great for the animal to continue responding (break point; Richardson and Roberts, 1996; Arnold and Roberts, 1997). Previous reports show that ghrelin delivery into the VTA not only increases food intake but also the effort that rats are willing to produce to get additional food reinforcers in the progressive ratio task as reflected by increased number of bar presses and higher break points compared with controls (King et al., 2011; St-Onge et al., 2016). To investigate whether cannabinoids mediated this aspect of ghrelin's effects on food motivation, we first trained rats to bar press for food on a FR-4 schedule until they were reliably responding for food rewards. The rats were then implanted with an intra-VTA cannula, and after 7–10 d of a recovery period, rats were pretreated with rimonabant or vehicle and with either saline of ghrelin. Thirty minutes after the second infusion, rats were transferred to operant boxes and tested on the progressive ratio task. As shown in Figure 2c, rats infused with ghrelin increased the number of bar presses and had higher break points than rats given control infusions (F(3,23) = 3.19; p < 0.05). More importantly, intra-VTA ghrelin treatment did not increase the amount of work rats performed to get food reinforcers when rats were pretreated with rimonabant infused into VTA (p > 0.05; Fig. 2c).
The endocannabinoid system is altered in rats with GHSR signaling deficits
Ghrelin binding to the GHSR, a G-coupled protein receptor of the Gq family, stimulates intracellular pathways that increase the activity of phospholipase C, calmodulin kinase II, and the release of stored calcium (for review, see Abizaid and Hougland, 2020). GHSR activation translocates diacylglycerol (DAG) to the membrane where DAG is hydrolyzed by the enzyme diaglycerol lipase α (DAGLα) and converted into the endogenous cannabinoid 2AG (Busquets-Garcia et al., 2018). We thus tested the hypothesis that ghrelin acts through these mechanisms to increase 2AG concentrations in the VTA, which would be expected to modulate presynaptic inputs to dopamine cells and facilitate neuronal excitability. We reasoned that if GHSR activation does indeed lead to increases in 2AG concentrations, then impaired GHSR signaling should result in overall lower 2AG concentrations in the VTA.
We collected brains from FHH-Ghsrm1Mcwi (GHSR-KO) rats with impaired GHSR signaling that have been linked with deficits in reward mechanisms like impaired rebound feeding following a fast sensitization to stimulants, sex motivation, and brain stimulation reward (Clifford et al., 2012; Wellman et al., 2012; MacKay et al., 2016; Hyland et al., 2018). Analyses of VTA punches from these rats demonstrated that VTA levels of 2AG were lower in GHSR-KO rats (t(18) = 2.17; p < 0.05) than those in the VTA of WT littermates (Fig. 3a). Anandamide concentrations also tended to be lower in GHSR-KO rats, but this difference was not statistically significant (t(18) = 1.27; p > 0.05; Fig. 3b).
GHSR signaling interacts with the endocannabinoid system. Rats with a partial mutation that truncates the GHSR (FHH-GHSRm1/Mcwi rats) have significantly lower VTA concentrations of 2AG (panel a) but not AEA (panel b) compared with control WT rats. FHH-GHSRm1/Mcwi rats also show lower expression of Dagla and faah mRNA in the VTA compared with control WT littermates (panel c). Importantly, the effects of intra-VTA ghrelin infusions (0.5 μg/μl of saline) at a subthreshold dose on food motivation were enhanced by intra-VTA infusions of MJN110 and PF-04457845 (6 μg/μl of DMSO), drugs that block the degradation of endocannabinoids (panel d). *p < 0.05.
In a follow-up study, we examined differences in the expression of enzymes that regulate the synthesis and degradation of anandamide and 2AG in the VTA of WT and GHSR-KO rats. Gene expression for both the synthetic enzyme DAGLα (t(17) = 3.03; p < 0.05) and FAAH (t(17) = 3.31; p < 0.05), an enzyme important for the degradation of both anandamide and 2AG (Sharkey and Pittman, 2005; Hill et al., 2013; Parsons and Hurd, 2015), was lower in GHSR-KO rats compared with WT rats, but the gene expression for MAGL, another enzyme that degrades 2AG, did not differ between GHSR-KO and WT littermates (p > 0.05; Fig. 3c). The gene expression for N-acyl phosphatidylethanolamine-specific phospholipase D (NAPE-PLD), an enzyme that is important for the generation of anandamide, appeared to be elevated in GHSR-KO compared with WT littermates, but this effect was not statistically significant (t(17) = 1.86; p = 0.08; Fig. 3c). Together, our findings suggested that reduced 2AG levels following deficits in GHSR signaling are most likely due to dysregulated DAGLα levels and/or activity.
Inhibition of 2AG degrading enzymes in the VTA in vivo increases the effects of a low dose of ghrelin on the amount of work rats are willing to perform to obtain food
The data described thus far suggest that ghrelin has a multitude of pre- and postsynaptic effects to stimulate the mesolimbic dopaminergic system. Ghrelin appears to target VTA dopamine cells and stimulate the release of 2AG onto presynaptic CB-1R receptors to increase excitatory tone onto dopamine cells. We therefore reasoned that if this were the case, inhibition of the enzymes that are responsible for degrading 2AG would enhance the behavioral effects of a dose of ghrelin that alone was not sufficient to increase food motivation. To determine if this was the case, male Long–Evans rats were trained and equipped with cannulae aimed at the VTA as described above. Once it was determined that the surgeries did not affect the ability of the animals to respond on a progressive ratio schedule under food-restricted conditions, the animals were given ad libitum access to food and then tested on a progressive ratio paradigm as used in the previous experiments. In this case, however, animals received unilateral infusions (50 μl) of either vehicle (10% DMSO in 0.9% saline), ghrelin (150 pM), MAGL + FAAH inhibitors (MJN110, 5 μg + PF-04457845, 3 μg), or ghrelin and MAGL + FAAH inhibitors. The dose of ghrelin was chosen from previous work in our lab that had threshold effects on food motivation as reflected in performance on the progressive ratio paradigm (Naleid et al., 2005). Doses of MAGL and FAAH inhibitors were calculated based on previous research on the ability of MJN110 and PF-04457845 to sufficiently increase cortical 2AG and AEA (Parker et al., 2015; Sticht et al., 2019). The MAGL/FAAH inhibitors were administered 30 min before ghrelin or saline infusion.
Intra-VTA administration of ghrelin at the threshold dose failed to increase the breakpoint of responses and the total active lever presses until breakpoint (Fig. 3d). Similarly, intra-VTA MAGL and FAAH inhibitor cocktail administration did not significantly increase breakpoint or total active lever presses until breakpoint. When the threshold dose of ghrelin was delivered into the VTA in combination with the MAGL and FAAH inhibitors, there was a significant increase in the breakpoint and in the average active lever presses until breakpoint (breakpoint, F(1,22) = 6.661; p < 0.05; active lever presses, F(1,22) = 6.774; p < 0.05).
Locomotor activity, inactive lever presses, and nose pokes were recorded as control measures to ensure that the increases in bar pressing responses observed in the animals treated with the cannabinoid-degrading enzyme inhibitors and ghrelin were not due to hyperactivity. As shown in Figure 5d, no significant differences were observed between groups in locomotor activity, inactive lever presses, or nose pokes suggesting that the effects on breakpoints and active lever presses were specifically reflecting food motivation and general arousal on the animal (p > 0.05; Fig. 3d).
Direct excitatory effect of ghrelin at TH VTA cells is partially CB-1R dependent
Ghrelin binding to the GHSR increases action potential frequency in dopamine cells in the VTA resulting in increased dopamine release and feeding (Abizaid et al., 2006), but it is not known if ghrelin action required endocannabinoid signaling. Using Th-cre;L10-Egfp mice that expressed native EGFP fluorescence overlapping with TH immunoreactivity in the VTA (Fig. 4a), we prepared acute brain slices and performed whole-cell patch–clamp recordings at EGFP-labeled TH VTA cells (Fig. 4b). Bath application of ghrelin (500 nM) reversibly depolarized the resting membrane potential of TH VTA cells (change in membrane potential; baseline, 1.78 ± 0.3 mV; ghrelin, 3.67 ± 0.77 mV; washout, 0.71 ± 0.67 mV; n = 9; F(2,16) = 11.66; p < 0.05). To determine if the stimulatory effects of ghrelin required endocannabinoid signaling, we pretreated the VTA brain slice with the CB-1R antagonists rimonabant or AM251 (5 μM). Both antagonists had a similar effect, so we combined the data from cells recorded in the presence of rimonabant and AM251. As shown in Figure 4d, the effects of ghrelin on resting membrane potential persisted despite pre-exposure to CB-1R antagonist and remained elevated throughout the washout period (change in membrane potential; baseline, 0.26 ± 0.37 mV; ghrelin, 2.49 ± 0.98 mV; washout, 3.14 ± 1.93 mV, n = 10; significant main effect for treatment phase, F(2,34) = 17.72; p < 0.05).
Patch-clamp recordings VTA dopamine cells from TH-Egfp-L10a mice (panel a). As shown in panel b, ghrelin (500 nM) increased the frequency of action potentials and depolarized the resting membrane potential of dopamine cells and effect that disappeared after washout (panel b). These effects persisted when slices were pretreated with rimonabant (5 μM), a CB-1R antagonist (panel c). Importantly, ghrelin continued to have a depolarizing effect on the resting membrane in TTX-treated sections that were pretreated with rimonabant (panel d). *p < 0.05; **p < 0.01; ***p < 0.001.
This ghrelin-mediated depolarization also increased the frequency of action potentials elicited by TH VTA cells (baseline, 0.225 ± 0.12 mV; ghrelin, 1.27 ± 0.24 mV; washout, 0.56 ± 0.19 mV; significant main effect for treatment phase, F(2,26) = 12.24; p < 0.05; Fig. 4c). Surprisingly, while action potential frequency was overall elevated in the presence of CB-1R antagonists, this increase in frequency was not statistically significant (p > 0.05; Fig. 4c). Close inspection of the data showed that, of the cells recorded, only three responded to ghrelin with the remaining cells being unresponsive to ghrelin stimulation in the presence of the CB-1R antagonists increasing overall variability (baseline, 0.196 ± 0.11 mV; ghrelin, 1.22 ± 0.65 mV; washout, 0.74 ± 0.36 mV; n = 7). This indicated that the ability of ghrelin to depolarize the membrane is direct and independent of endocannabinoid CB-1R signaling, but the ability of ghrelin to increase action potential frequency is restricted if CB-1R signaling is reduced potentially through a presynaptic mechanism.
To further elucidate the effects of ghrelin on membrane potential, we determined if ghrelin application directly stimulated TH VTA cells in the presence of rimonabant (5 μM) and TTX (500 nM) to block activity-dependent neurotransmission. As shown in Figure 4d, the resting membrane potential was still sensitive to the depolarizing effects of ghrelin (500 nM) despite pre-exposure to rimonabant and the addition of TTX. Interestingly, and in contrast to cells that were recorded in the presence of CB-1R antagonists but not pre-exposed to TTX, the membrane potential gradually returned to the baseline during the washout (change in membrane potential RIM, 0.29 ± 0.47 mV; RIM + ghrelin, 3.67 ± 0.97 mV; washout, 1.09 ± 1.28 mV; n = 6; F(2,10) = 12.48; p < 0.05; Fig. 4d). Taken together, these data indicated that ghrelin directly excited TH VTA neurons independent of CB-1R signaling and that diminished CB-1R signaling impacts the ability of ghrelin to excite neurons via presynaptic mechanisms.
Ghrelin increased excitatory input to TH VTA cells in a CB-1R–dependent manner
As VTA cells are regulated by both glutamatergic and GABAergic inputs (Melis and Pistis, 2012), we assessed if ghrelin regulates GABAergic or glutamatergic input, measured as sIPSC or sEPSC events, arriving at EGFP-labeled cells from the VTA of Th-cre;L10-Egfp mice. Bath application of 500 nM ghrelin did not alter the frequency of sIPSC events at TH VTA cells (baseline, 0.09 ± 0.26 Hz; ghrelin, 0.003 ± 0.25 Hz; washout, 0.25 ± 0.24 Hz; F(2,32) = 0.91; p > 0.05; Fig. 5a). In contrast, ghrelin reversibly increased the frequency of sEPSC events at ∼60% (7/12) of TH VTA neurons (Fig. 5b). Indeed, a repeated-measure ANOVA showed that ghrelin significantly increased sEPSC frequency relative to the baseline, an effect that dissipated with ghrelin washout (change in sEPSC frequency baseline, 0.35 ± 0.18 Hz; ghrelin, 1.08 ± 0.34 Hz; washout, 0.44 ± 0.08 Hz; Fig. 5b,c), an effect that was prevented by pretreatment with AM251 (change in sEPSC frequency baseline, −0.17 ± 0.09 Hz; ghrelin, −3.9 ± 0.18 Hz; washout, −0.35 ± 0.18 Hz; significant interaction effect, F(2,24) = 6.125; p < 0.05; Fig. 5b,c).
Effects of ghrelin on sIPSCs and sEPSCs obtained from recordings on dopamine cells from TH-Egfp-L10a. As shown in panel a, sIPSC were not affected by ghrelin treatment (500 nM) suggesting that ghrelin is not directly affecting inhibitory tone onto dopamine cells (panel a). In contrast, ghrelin treatment increased the frequency of sEPSCs, and this effect was prevented by pre-exposure to the CB-1R receptor antagonist AM-251 (5 μM; panel b). *p < 0.05.
Discussion
The data presented above support the idea that ghrelin acts on VTA dopamine neurons through mechanisms that are dependent and independent from endocannabinoid release. Specifically, while the ability of ghrelin to activate VTA dopamine neurons appeared to be independent of endocannabinoid signaling, the increased excitatory drive onto VTA dopamine cells produced by ghrelin was dependent on endocannabinoid function. Importantly, increases in the behavioral expression of reward-seeking behaviors in response to ghrelin can be attenuated by CB-R antagonists delivered directly onto the VTA and enhanced by drugs that block the degradation of endogenous cannabinoids. Finally, disruptions in GHSR signaling result in deficits in 2AG content in the VTA and in deficits in DAGLα, the enzyme that hydrolyzes DAG into 2AG, suggesting that intact GHSR signaling is critical for optimal regulation 2AG release. Together, these data suggest that ghrelin can recruit the endocannabinoid signaling to augment the activity of VTA dopamine neurons, which in turn invigorates food-motivated behavior and consumption.
Within the hypothalamus and striatum, the GHSR and the CB-1R are coexpressed in the same cells, and these receptors seem to interact intracellularly via the phosphorylation of the metabolic enzyme adenosine monophosphate kinase to increase feeding (Kola et al., 2005, 2008). In the striatal cells, coexpression of GHSR and CB-1R reduces CB-1R signaling, but the presence of cannabinoids enhances GHSR-mediated increases in intracellular calcium signaling (Lillo et al., 2021). Thus, one could hypothesize that a similar interaction could occur in the VTA if the receptors were to be coexpressed in the same cells. This hypothesis, however, is not supported by our neuroanatomical studies. Indeed, our data confirm previous work showing that VTA cells express ghsr and cnr1 mRNA and that ∼60% of dopamine cells within the lateral portion of the anterior VTA express ghsr (Guan et al., 1997; Abizaid et al., 2006; Zigman et al., 2006). These dopamine cells, however, not express cnr1 mRNA in rats or mice. Within this region of the VTA, ultrastructural analyses of CB-1R protein predominantly localized CB-1R within putatively excitatory synapses and inhibitory synapses suggesting that CB-1R modulation of neurotransmission is likely to be complex and include the regulation of excitatory and inhibitory presynaptic inputs. Given that 66% of presynaptic inputs expressing the CB-1R protein were asymmetric and thus putatively excitatory, it is suggested that if ghrelin engaged the endocannabinoid system, it did through a modulation of excitatory tone.
Electrophysiological studies supported the excitatory actions of ghrelin in the VTA and indicated that ghrelin could act in an endocannabinoid-independent and endocannabinoid-dependent manner (Fig. 6). As previously demonstrated (Abizaid et al., 2006), ghrelin robustly depolarized and increased the firing of EGFP-labeled dopaminergic VTA cells, and this direct excitatory action of ghrelin was independent of CB-1R activation. Direct effects of ghrelin on dopamine neurons could be mediated by changes in ion channel conductance similar to those observed in ghrelin-stimulated dopamine cells in the substantia nigra, where ghrelin depolarizes the resting membrane potential of dopamine cells via a reduction of potassium channel conductance (Shi et al., 2013). Alternatively (or simultaneously), ghrelin activation of the GHSR may increase the surface density of postsynaptic AMPA receptors to enhance membrane excitability as previously demonstrated in the hippocampus (Ribeiro et al., 2014).
Potential endocannabinoid-independent (panel a) or endocannabinoid-dependent (panel b) mechanisms of ghrelin facilitation of dopamine neuron excitation. In Figure 6a, we depict cannabinoid-independent mechanisms where ghrelin activation of the GHSR enhances membrane excitability through the closing of Kv7/KNCQ potassium channels and/or the trafficking of AMPA receptors to the membrane to enhance membrane excitability. In Figure 6b, we propose that activation of the GHSR results in the release of 2AG, and this will result in astrocyte activation and/or inhibit GABA synapses to enhance the release of glutamate onto dopamine cells, effects that would be blocked by CB-1R antagonists like rimonabant or AM251. Created in BioRender. https://BioRender.com/e44s311.
Additionally, ghrelin can also act indirectly by regulating glutamatergic afferent input to the VTA. This was consistent with ghrelin-mediated release of nonclassical neurotransmitters including endogenous cannabinoids and nitric oxide, as suggested by previous work (Edwards and Abizaid, 2016; Kalafateli et al., 2018; Engel et al., 2023). We expected that ghrelin would recruit endocannabinoid release to decrease GABAergic input to VTA dopamine cells (Melis and Pistis, 2012) and in a CB-1R–dependent manner, but surprisingly, ghrelin had no overall impact on GABAergic transmission. In contrast, ghrelin increased glutamatergic input to dopaminergic VTA cells, and this stimulatory effect was blocked by pretreatment with a CB-1R antagonist. This noncanonical cannabinoid effect could be mediated through decreased inhibition of presynaptic excitatory synapses (Melis et al., 2004) or through an astrocyte-mediated stimulatory effect on excitatory synapses as has been demonstrated (Requie et al., 2022; Fig. 6b).
Regardless of the cellular mechanism of action, ghrelin delivery into the VTA increased feeding responses and the motivation to eat as determined by increases in the efforts to obtained food in the progressive ratio task, and this increase was attenuated by pretreatment with the CB-1R antagonist rimonabant. This is like the previous work showing that ghrelin delivery into the VTA of mice increases feeding, locomotor activity, and dopamine release, and at least locomotor activity is attenuated by peripheral or intra-VTA pretreatment with rimonabant (Kalafateli et al., 2018). Here, we show that in rats intra-VTA ghrelin infusions increase feeding and operant responses to obtain food in the progressive ratio task, and these behavioral effects are attenuated in animals pretreated with rimonabant. We observed no effects of drug treatment on locomotor activity. The differences between our current data and the data using mice (Kalafateli et al., 2018) may reflect species differences in the distribution of terminals stemming from the A10 dopaminergic neurons to their cortical and limbic targets (Gorelova et al., 2012; Hnasko et al., 2012) and differences in the behavioral assays used. Nevertheless, our current data confirm that ghrelin elicits behavioral responses associated with increased activity in VTA dopamine cells (Abizaid et al., 2006; Egecioglu et al., 2010) and show that these behavioral responses are in part dependent on the interaction of between the GHSR and the CB-1 receptor in the VTA. Importantly, we show that the effects of a subthreshold dose of ghrelin on food motivation were enhanced by intra-VTA treatment with inhibitors of enzymes that degrade 2AG before the ghrelin infusions. These data support the idea that ghrelin binding on GHSR in the VTA can alter endocannabinoid concentrations to increase dopamine cell activity. This idea is further supported by our results showing that rats with disrupted GHSR signaling show decreased mRNA expression of DAGL and decreased overall content of 2AG in the VTA. Overall, these data unveil a previously undocumented mechanism of action by which GHSR signaling synergizes with the cannabinoid system to increase food motivation. The direction in which ghrelin influences endocannabinoid concentrations remains to be determined. While our data would suggest that ghrelin stimulates the synthesis and release of 2AG, there is evidence that GPCR-mediated increases in intracellular calcium signaling can decrease 2AG tone to enhance the excitatory tone of gonadotropin-releasing hormone neurons (Farkas et al., 2016). Whether a similar mechanism occurs in the VTA remains to be determined.
While our electrophysiological experiments were conducted using the tissue from male and female animals and no clear sex differences were observed, we do acknowledge that conclusions from our current behavioral and biochemical data are limited to males, and these studies require replication in female subjects. There are sex differences in mesolimbic dopamine neurotransmission, and females are generally more sensitive to stimuli such as drugs of abuse (Calipari et al., 2017). Conversely, female rodents seem less sensitive to the orexigenic effects of ghrelin making it critical to determine whether ghrelin directly infused into the VTA produces similar feeding and food motivation responses to those seen in males, together with the extent of cannabinoid involvement in these responses (Clegg et al., 2007; Smith et al., 2022).
The relationship between GHSR and CB-1R signaling has been previously examined in relation to feeding and metabolism (Kola et al., 2008; Mani et al., 2020; Lillo et al., 2021). While mice lacking both the GHSR and the CB-1R do not show gross alterations in food intake and weight gain (Mani et al., 2020), GHSR-KO mice do not show orexigenic responses to cannabinoids, and mice lacking CB-1R do not show orexigenic responses to ghrelin (Kola et al., 2008). It is not known, however, if challenging the double mutant mice used by Mani et al. would result in deficits. If GHSR signaling is critical to increase CB-1R activation in the VTA, one would expect that the GHSR/CB-1R KO mice or mice lacking GHSR would work less to obtain food rewards than the mice only lacking CB-1R. This hypothesis, however, requires further testing, and one cannot ignore the potential for compensatory mechanisms influencing results from these studies.
Overall, our data suggest a novel mechanism through which ghrelin increases the activity of reward circuitry. Ghrelin first acts on dopamine cells directly to stimulate their activity, and this is followed by an increase in the release of endogenous cannabinoids that enhance presynaptic excitatory tone to facilitate burst activity of dopamine cells and ultimately increase food seeking and intake. Potentially this mechanism could also result in enhanced motivation for other reinforcing stimuli like social and sexual interactions, drugs of abuse, hypercaloric diets, and escape from aversive stimuli.
Footnotes
This manuscript was funded by a Discovery Grant awarded to A.A. (Grant Number 06248); a National Brain Research Program (NAP 3.0; NAP2022-I-10/2022) of the Hungarian Academy of Sciences awarded to C.F.; Natural Sciences and Engineering Research Council of Canada graduate scholarships awarded to A.E., C.D.S., and A.S.; and a Canadian Institutes of Health Research Grant awarded to M.N.H. We thank Dr. Barbara Woodside for her insightful comments on this manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to Alfonso Abizaid at alfonso_abizaid{at}carleton.ca.
